# INSECT OLFACTORY PROTEINS (FROM GENE IDENTIFICATION TO FUNCTIONAL CHARACTERIZATION)

EDITED BY : Peng He, Nicolas Durand and Shuang-Lin Dong PUBLISHED IN : Frontiers in Physiology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-266-4 DOI 10.3389/978-2-88963-266-4

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# INSECT OLFACTORY PROTEINS (FROM GENE IDENTIFICATION TO FUNCTIONAL CHARACTERIZATION)

Topic Editors: Peng He, Guizhou University, China Nicolas Durand, University of Picardie Jules Verne, France Shuang-Lin Dong, Nanjing Agricultural University, China

Citation: He, P., Durand, N., Dong, S.-L., eds. (2019). Insect Olfactory Proteins (From Gene Identification to Functional Characterization). Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-266-4

# Table of Contents

*07 Editorial: Insect Olfactory Proteins (From Gene Identification to Functional Characterization)*

Peng He, Nicolas Durand and Shuang-Lin Dong

#### CHEMOSENSORY RECEPTORS


Huan Liu, Zheng-Shi Chen, Dong-Ju Zhang and Yong-Yue Lu

*64 Proceeding From* in vivo *Functions of Pheromone Receptors: Peripheral-Coding Perception of Pheromones From Three Closely Related Species,* Helicoverpa armigera, H. assulta*, and* Heliothis virescens

Bing Wang, Yang Liu and Gui-Rong Wang

*76 Functional Studies of Sex Pheromone Receptors in Asian Corn Borer*  Ostrinia furnacalis

Wei Liu, Xing-chuan Jiang, Song Cao, Bin Yang and Gui-rong Wang

## ODORANT BINDING PROTEINS AND CHEMOSENSORY PROTEINS

*85 Various Bee Pheromones Binding Affinity, Exclusive Chemosensillar Localization, and Key Amino Acid Sites Reveal the Distinctive Characteristics of Odorant-Binding Protein 11 in the Eastern Honey Bee,*  Apis cerana

Xin-Mi Song, Lin-Ya Zhang, Xiao-Bin Fu, Fan Wu, Jing Tan and Hong-Liang Li


Xing Ge, Tofael Ahmed, Tiantao Zhang, Zhenying Wang, Kanglai He and Shuxiong Bai

*123 Silencing the Odorant Binding Protein* RferOBP1768 *Reduces the Strong Preference of Palm Weevil for the Major Aggregation Pheromone Compound Ferrugineol*

Binu Antony, Jibin Johny and Saleh A. Aldosari

*140 Distinct Subfamilies of Odorant Binding Proteins in Locust (Orthoptera, Acrididae): Molecular Evolution, Structural Variation, and Sensilla-Specific Expression*

Xingcong Jiang, Jürgen Krieger, Heinz Breer and Pablo Pregitzer

*155 Identification and Characterization of Odorant Binding Proteins in the Forelegs of* Adelphocoris lineolatus *(Goeze)*

Liang Sun, Qian Wang, Qi Wang, Kun Dong, Yong Xiao and Yong-Jun Zhang

*167 Two Odorant-Binding Proteins of the Dark Black Chafer (*Holotrichia parallela*) Display Preferential Binding to Biologically Active Host Plant Volatiles*

Qian Ju, Xiao Li, Xiao-Qiang Guo, Long Du, Chen-Ren Shi and Ming-Jing Qu

*181 Identification of Odorant-Binding Proteins (OBPs) and Functional Analysis of Phase-Related OBPs in the Migratory Locust*

Wei Guo, Dani Ren, Lianfeng Zhao, Feng Jiang, Juan Song, Xianhui Wang and Le Kang


Daniele S. Oliveira, Nathália F. Brito, Thiago A. Franco, Monica F. Moreira, Walter S. Leal and Ana C. A. Melo

*223 Molecular and Functional Characterization of Odorant Binding Protein 7 From the Oriental Fruit Moth* Grapholita molesta *(Busck) (Lepidoptera: Tortricidae)*

Xiu-Lin Chen, Guang-Wei Li, Xiang-Li Xu and Jun-Xiang Wu


Immacolata Iovinella, Federico Cappa, Alessandro Cini, Iacopo Petrocelli, Rita Cervo, Stefano Turillazzi and Francesca R. Dani

*265 Sex- and Tissue-Specific Expression Profiles of Odorant Binding Protein and Chemosensory Protein Genes in* Bradysia odoriphaga *(Diptera: Sciaridae)*

Yunhe Zhao, Jinfeng Ding, Zhengqun Zhang, Feng Liu, Chenggang Zhou and Wei Mu

*281 iTRAQ-Based Comparative Proteomic Analysis Reveals Molecular Mechanisms Underlying Wing Dimorphism of the Pea Aphid,*  Acyrthosiphon pisum

Limei Song, Yuhao Gao, Jindong Li and Liping Ban


Zhumei Li, Lulu Dai, Honglong Chu, Danyang Fu, Yaya Sun and Hui Chen

*319 Silencing of Chemosensory Protein Gene NlugCSP8 by RNAi Induces Declining Behavioral Responses of* Nilaparvata lugens Muhammad I. Waris, Aneela Younas, Muhammad T. ul Qamar, Liu Hao, Asif Ameen, Saqib Ali, Hazem Elewa Abdelnabby, Fang-Fang Zeng and Man-Qun Wang

#### ODORANT DEGRADING ENZYMES


#### MISCELLANEOUS

*360 Chemosensory Gene Families in* Ectropis grisescens *and Candidates for Detection of Type-II Sex Pheromones*

Zhao-Qun Li, Zong-Xiu Luo, Xiao-Ming Cai, Lei Bian, Zhao-Jun Xin, Yan Liu, Bo Chu and Zong-Mao Chen

*374 Differential Expression Analysis of Olfactory Genes Based on a Combination of Sequencing Platforms and Behavioral Investigations in*  Aphidius gifuensis

Jia Fan, Qian Zhang, Qingxuan Xu, Wenxin Xue, Zongli Han, Jingrui Sun and Julian Chen

*384 Antennal Transcriptome Analysis of the Chemosensory Gene Families From Trichoptera and Basal Lepidoptera*

Jothi Kumar Yuvaraj, Martin N. Andersson, Dan-Dan Zhang and Christer Löfstedt

*400 Identification and Comparison of Chemosensory Genes in the Antennal Transcriptomes of* Eucryptorrhynchus scrobiculatus *and* E. brandti *Fed on*  Ailanthus altissima

Xiaojian Wen, Qian Wang, Peng Gao and Junbao Wen


Su-fang Zhang, Zhen Zhang, Xiang-bo Kong, Hong-bin Wang and Fu Liu

*446 Chemoreception of Mouthparts: Sensilla Morphology and Discovery of Chemosensory Genes in Proboscis and Labial Palps of Adult* Helicoverpa armigera *(Lepidoptera: Noctuidae)*

Mengbo Guo, Qiuyan Chen, Yang Liu, Guirong Wang and Zhaojun Han

*461 A Synergistic Transcriptional Regulation of Olfactory Genes Drives Blood-Feeding Associated Complex Behavioral Responses in the Mosquito* Anopheles culicifacies

Tanwee Das De, Tina Thomas, Sonia Verma, Deepak Singla, Charu Chauhan, Vartika Srivastava, Punita Sharma, Seena Kumari, Sanjay Tevatiya, Jyoti Rani, Yasha Hasija, Kailash C. Pandey and Rajnikant Dixit


#### REVIEWS


# Editorial: Insect Olfactory Proteins (From Gene Identification to Functional Characterization)

Peng He<sup>1</sup> \*, Nicolas Durand<sup>2</sup> \* and Shuang-Lin Dong<sup>3</sup> \*

<sup>1</sup> State Key Laboratory Breeding Base of Green Pesticide and Agricultural Bioengineering, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Guizhou University, Guiyang, China, <sup>2</sup> FRE CNRS 3498, Ecologie et Dynamique des Systèmes Anthropisés, Université de Picardie Jules Verne, Amiens, France, <sup>3</sup> Education Ministry Key Laboratory of Integrated Management of Crop Disease and Pests, College of Plant Protection, Nanjing Agricultural University, Nanjing, China

Keywords: odorant-binding proteins, chemosensory proteins, odorant receptors, ionotropic receptors, odorantdegrading enzymes

#### **Editorial on the Research Topic**

#### **Insect Olfactory Proteins (From Gene Identification to Functional Characterization)**

Olfaction is essential for the survival and reproduction of many insects. Their extremely sophisticated and sensitive olfactory system helps them accomplish key behaviors, such as seeking food resources, avoiding predators, locating mate partners, and selecting egg-laying sites. The involved molecular actors, the olfactory proteins, play crucial roles in the responses triggered by external chemical stimuli. Classification includes receptor proteins, the odorant receptors (ORs) and the ionotropic receptors (IRs), and perireceptor proteins, the odorant binding proteins (OBPs), chemosensory proteins (CSPs), sensory neuron membrane proteins (SNMPs) and odorant degrading enzymes (ODEs). Olfactory proteins are expressed in olfactory sensilla, hair-like structures predominantly located on insect antennae. Sensilla house the olfactory receptor neurons (ORNs), whose dendrites bath in a lymph. Recent advances in bioinformatics and biological techniques have enabled the identification of numerous olfactory-related genes and unveiled their functions. However, our knowledge of the molecular mechanisms of the insect olfactory system is still very limited. This Research Topic is presenting the identification and the molecular and functional characterization of insect olfactory proteins in diverse insect species. One third of articles are dedicated to the moths (Lepidoptera), reflecting the outstanding interest in the chemical ecology of these diverse crop or forest pests. The other articles are treating olfactory proteins of representatives of hemipteran, coleopteran, dipteran, and orthopteran pests or disease vectors, respectively. In addition, three articles focus on parasitoid or pollinating hymenopterans and two articles present comparative studies between species from at least two insect orders.

#### MODERN SEQUENCING TECHNOLOGIES FACILITATE THE IDENTIFICATION AND EXPRESSION ANALYSIS OF NOVEL OLFACTORY GENES AND PROTEINS

In the last decade, high throughput sequencing of transcriptomics and proteomics has become widely used to identify the large repertoire of olfactory genes in numerous insect species, and to investigate their expression between various physiological states and tissues. Here, Jin et al. analyzed antennal transcriptomes from the oriental fruit fly Bactrocera dorsalis, corresponding to different maturity and mating status, and identified

#### Edited and reviewed by:

Sylvia Anton, Institut National de la Recherche Agronomique (INRA), France

#### \*Correspondence:

Peng He phe1@gzu.edu.cn Nicolas Durand nfdurand@gmail.com Shuang-Lin Dong sldong@njau.edu.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 07 September 2019 Accepted: 30 September 2019 Published: 18 October 2019

#### Citation:

He P, Durand N and Dong S-L (2019) Editorial: Insect Olfactory Proteins (From Gene Identification to Functional Characterization). Front. Physiol. 10:1313. doi: 10.3389/fphys.2019.01313

**7**

43 ORs and 21 IRs. Many of them are strongly regulated by mating or egg-laying, reflecting that these olfactory genes could have close relationships with these crucial behaviors (Jin et al.). Similarly, in the chive gnat Bradysia odoriphaga, analysis of antennal and body adult transcriptomes led to the identification of 49 OBPs and 5 CSPs. 22 OBPs and 3 CSPs are preferentially expressed in the antennae, including 9 male enriched OBPs which are relevant candidates for sex pheromone component binding and transport (Zhao et al.). In the beet armyworm Spodoptera exigua, Zhang et al. analyzed transcripts expressed in the chemosensory organs of adults and identified 64 ORs, 22 IRs, 24 OBPs and 19 CSPs. In another study, 53 ORs and 4 IRs have been identified in the antennae of the citrus long-horned beetle Anoplophora chinensis by transcriptomic analysis (Sun et al.). In the vetch aphid Megoura viciae, 10 OBPs were identified in the antennal transcriptome (Bruno et al.). A great diversity of putative ODEs was also unveiled in the antennal transcriptomes of two moths, 33 glutathione-Stransferases (GSTs) in Spodoptera littoralis (Durand et al.) and 35 carboxylesterases (CXEs) in Ectropis obliqua (Sun et al.). To study the evolution of insect chemoreception, Yuvaraj et al. analyzed the antennal expressed genes and compared the olfactory repertoires of two basal Lepidoptera species and one species of its sister group Trichoptera. Combined phylogenetic analysis suggests that the pheromone receptors (PRs) and the pheromone binding proteins (PBPs) have evolved in parallel with the transition of sex pheromone types in Lepidoptera, while other chemoreceptor subfamilies show a broader taxonomic occurrence than hitherto acknowledged (Yuvaraj et al.). Wen et al. sequenced the antennal transcriptomes of two closely related weevils Eucryptorrhynchus scrobiculatus and E. brandti, reared on the same host plants although on different parts. Total numbers of olfactory genes identified in the two species were similar, 111 (49 ORs, 17 IRs, 31 OBPs, 11 CSPs, and 3 SNMPs) in E. scrobiculatus, and 112 (45 ORs, 25 IRs, 28 OBPs 11 CSPs, and 3 SNMPs) in E. brandti, however species-specific olfactory genes were highlighted, with a possible role in the recognition of specific volatiles from different plant parts. Olfactory genes were identified in the sibling moth species, Ectropis grisescens and E. obliqua, including 59 ORs, 24 IRs, 40 OBPs and 30 CSPs, from E. grisescens male antennae (Li et al.) and 52 ORs and 36 OBPs from E. obliqua (Li et al.). Fan et al. paired two sequencing techniques, next generation sequencing [NGS and single-molecule real-time sequencing (SMRT)] to obtain full-length olfactory genes in the parasitoid Aphidius gifuensis, including 66 ORs, 25 IRs, 16 OBPs, and 12 CSPs. Among them, 25 proteins could be potentially involved in aphid alarm pheromone E-β-farnesene detection. Zhang et al. compared the antennal transcriptomes between virgin and mated female adults of the moth Dendrolimus punctatus, and identified new olfactory-related genes including 8 ORs and 5 IRs. In addition, a subset of olfactory proteins is up-regulated after mating, indicating a putative association with oviposition site seeking behavior (Zhang et al.). Das De et al. conducted a comparative analysis of prior and post-blood meal groups in the mosquito Anopheles culicifacies, and unraveled several ORs and OBPs that could drive blood feeding associated behaviors in this vector.

Olfactory genes are mainly identified in insect antennae but they could also be highlighted in other chemosensory organs, such as palps and proboscis. Here, Guo et al. analyzed second generation sequencing data of the labial palps and proboscis of the moth Helicoverpa armigera, and unveiled a vast number of olfactory genes including 4 ORs, 6 IRs, 39 OBPs, 26 CSPs, and 2 SNMPs. Similarly, 11 ORs and 16 OBPs have been identified from antennae, maxillary and labial palps transcriptome of the locust Locusta migratoria, including four ORs (OR12, OR13, OR14, and OR18) and OBP8, which are expressed in higher levels in palps than in antennae (Li et al.). In addition, a limited number of olfactory genes have been identified in the leg transcriptome of the mirid bug Adelphocoris lineolatus (Sun et al.), with only 8 OBPs. Proteomics also lead to the identification and the expression analysis at the protein level. Iovinella et al. adopted a proteomic approach to investigate the expression levels of olfactory proteins (OBPs, CSPs, and ODEs) in honeybees among different castes, tasks and ages, highlighting major differences between queens and workers. In addition, based on isobaric tags for relative and absolute quantitation (ITRAQ) comparative proteomic analysis, Song et al. unveiled that two OBPs and three CSPs are differently expressed between alate and apterous morphs of the pea aphid Acyrthosiphon pisum.

## MOLECULAR AND FUNCTIONAL CHARACTERIZATION OF ODORANT RECEPTORS

ORs are seven-transmembrane receptors located on the dendrites of ORNs and activated by OBP/CSP-odorant complexes or odorants only. ORs can be divided into OR co-receptors (Orcos) and ligand-specific ORs (ORx), which interact with each other to form an ORx-ORco complex to generate a ligand-gated cation channel. Orcos are well-conserved among insect species and Orco knock-out insects are anosmic, indicating Orcos as promising pest management targets. In this Research Topic, Wang et al. compared and analyzed Orco genes from five sibling mirid bug species: Apolygus lucorum, Lygus pratensis, A. lineolatus, A. suturalis, and A. fasciaticollis. The five Orco genes shared high deduced amino acid identities and are wellconserved. However, at genome level, these five Orco genes present significantly different exon-intron structures, especially on the insertion sites and length of introns, suggesting variation in their evolution rates (Wang et al.). Pheromone receptors (PRs) are an ORx subtype generally located in trichoid sensilla and activated by sex pheromone components. PRs have been deorphanized in many species, especially in Lepidoptera. In this Research Topic, Liu et al. verified the correspondence between PRs and a series of sex pheromone components of the Asian corn borer Ostrinia furnacalis. The results obtained in vitro with the Xenopus oocyte expression system are in line with the in vivo electrophysiological analysis of four types of trichoid sensilla (Liu et al.). The functional characterization methods for PRs or other ORxs are multiple and could produce variability, which may result in somewhat different conclusions. Wang et al. compared two functional assays, in vivo transgenic fly system and in vitro Xenopus oocyte expression system, using three PR clades (HarmOR6, OR13, and OR16) from three phylogenetically close noctuid moths, H. armigera, H. assulta, and Heliothis virescens. Results from the two methods seem consistent, with nevertheless an increased sensitivity of the in vitro system. However, the active sex pheromone components of OR6 obtained by the two methods are different in H. armigera and H. assulta, and therefore, other techniques such as genome editing combined with behavior tests are needed to further verify the precise functions of olfactory receptors in the future (Wang et al.).

Compared to PRs, broadly tuned ORxs are complex to deorphanize, because each ORx is tuned to a variety of odorants, while one odorant could bound to multiple ORxs. Furthermore, non-PR ORxs in general are highly divergent in amino acid sequence among insect species. In this Research Topic, Liu et al. showcased a comprehensive method to deorphanize one OR, BdorOR88a, expressed in adult males of B. dorsalis. First, a comparison of gene expression in the antennal transcriptomes of methyl eugenol (ME)-exposed and control insects highlighted that two OR genes, BdorOR63a-1 and BdorOR88a, are upregulated after ME exposure. Then a complementary in vitro functional study demonstrated that only BdorOR88a/Orco robustly responded to ME. Finally, behavioral experiments with BdorOR88a knock-down male flies revealed a reduced attraction of these insects to ME (Liu et al.). In another study, Li et al. first determined the active odorants that produced increased electrophysiological responses with maxillary and labial palps as compared with antennae in L. migratoria. Then, palp transcriptomes were analyzed and four palp-enriched ORs (LmigOR12, OR13, OR14, and OR18) were identified. Finally, RNA interference (RNAi) combined to electroantennogram recordings indicated that OR12 was responsible for detection of three aldehyde odorants (E,E)-2,4-heptadienal, hexanal, and E-2-hexenal (Li et al.). These approaches could be used to deorphanize more broadly tuned ORxs. In addition, the cellbased expression system combined with Ca2<sup>+</sup> level investigation could be used to set up a high throughput and rapid screening of vast potential odorant candidates.

#### ODORANT CARRYING PROTEINS POTENTIALLY INVOLVED IN PHEROMONE AND PLANT VOLATILE DETECTION

OBPs and CSPs abundantly expressed in insect sensillar lymph are considered as putative carriers of odorants to the chemosensory receptors anchored in the dendritic membrane of ORNs. Pheromone binding proteins (PBPs) constitute a subgroup of OBPs participating in sex pheromone detection. Here, two PBPs, CpunPBP2 and CpunPBP5, enriched in the antennae of the adult male of the yellow peach moth Conogethes punctiferalis have been functionally characterized. Recombinant proteins presented in vitro extremely high binding abilities with the two sex pheromone components compared to a pheromone analog and a panel of plant volatiles. Moreover, activity of mutated proteins indicate that several amino acid residues are potentially involved in sex pheromone binding (Ge et al.). In the Chagas disease vector Rhodnius prolixus, Oliveira et al. focused on two male antennae-specific PBPs, RproOBP26 and RproOBP27 and performed RNAi experiments suggesting that RproOBP27 is involved in sex pheromone detection. In the eastern honey bee Apis cerana, a PBP(AcerOBP11) displays in vitro strong affinities with the main queen mandibular pheromone compounds, the alarm pheromone and worker pheromone components. Experiments with mutated PBPs demonstrated that two residues (Ile97 and Ile140) play crucial roles in binding with various bee pheromones (Song et al.).

In this Research Topic, the important role of OBPs for aggregation pheromone recognition has also been demonstrated. In adults of the red palm weevil Rhynchophorus ferrugineus, Antony et al. first identified four antennal-enriched or antennalspecific OBPs, and then determined that RferOBP1768 plays a major role in detection of the aggregation pheromone by RNAi experiments. Similarly, Guo et al. uncovered that OBPs could be involved in aggregation pheromone detection. Seventeen OBPs were identified in the genome of L. migratoria, two of which, LmigOBP2 and LmigOBP4, are up-regulated during gregarization and down-regulated during solitarization. Subsequently, they performed RNAi and behavioral experiments, and proposed that LmigOBP4 is the sole OBP that potentially participates in aggregation pheromone detection in L. migratoria (Guo et al.).

Additional studies in this Research Topic indicated that certain OBPs could present broad ligand affinity, suggesting their multiple roles in chemosensation. Ma et al. identified EoblOBP6 in E. obliqua, as an OBP dominantly expressed in the antennae and legs of adult insects, and presenting strong affinities with plant volatiles (such as benzaldehyde, nerolidol, α-farnesene) and with the aversive bitter alkaloid berberine, suggesting roles in both olfaction and gustation. In another study, GmolOBP7, an OBP from the oriental fruit moth Grapholita molesta is highly expressed in antennae and wings of both sexes. In vitro ligand binding assays and in vivo RNAi experiments suggested a dual role of GmolOBP7 in detection of both sex pheromone components and host plant volatiles (Chen et al.).

Sensilla expression patterns of OBPs in the insect antennae can provide valuable information about their putative roles and likely interplays among OBP partners within a sensillum. In the desert locust Schistocerca gregaria, Jiang et al. first demonstrated that the OBP repertoire is divided into four major phylogenetic clades, then characterized the specific sensilla expression patterns of representatives from each OBP clade. OBPs of subclade I-A are expressed in both trichoid and basiconic sensilla, while members of subclade II-A are restricted to coeloconic sensilla. OBPs of III-A, III-B, and I-B are exclusively found in chaetic sensilla, with a specific OBP of I-B being also expressed in coeloconic sensilla. The atypical OBP subtype from subclade IV-A is expressed in a subpopulation of coeloconic sensilla, and lastly, the plus-C type-B OBP subtypes from subclade IV-B are expressed in all four antennal sensillum types (trichoid, basiconic, chaetic, and coeloconic). Furthermore, a subset of OBPs, such as SgreOBP1, 2, 5, 6, 10, and 14, are co-localized in the same sensilla and could interact as partners to be active (Jiang et al.; Jiang et al.). Some OBPs are also located in other organs besides antennae, including non-chemosensory organs, indicating their multiple roles apart from olfaction. In the vetch aphid M. viciae, Bruno et al. first described the distribution of three types of sensilla (trichoid, coeloconic, and placoid) on antennae, mouthparts, legs and cauda. Then functional hypotheses were determined based on the distribution profiles of 5 OBPs in these sensilla by immunolocalization (Bruno et al.).

CSP display greater sequence conservation but wider tissue expression patterns compared to OBPs, implying multiple putative functions. In this Research Topic, a few CSPs are associated with insect olfaction or gustation. Li et al. first successfully identified a nymph antennae-enriched CSP from the sycamore lace bug Corythucha ciliata, CcilCSP2, by screening the tissue expression pattern of 15 CSPs. Then recombinant CcilCSP2 was functionally characterized in vitro. CcilCSP2 binds with high affinity to the alarm pheromone geraniol and to the repellent phenyl benzoate (Li et al.). Another two studies further characterized the olfactory function of CSPs using RNAi experiments followed with electrophysiological or behavioral tests. In CSP knock-down insects, electrophysiological responses toward plant volatiles are diminished, and host plant location is altered. Li et al. identified a CSP, DarmCSP2, preferentially expressed in the antennae of the Chinese white pine beetle Dendroctonus armandi. The in vitro ligand-binding assays showed that the sex pheromone components and terpene host plant volatiles are potential active ligands of DarmCSP2. Further electrophysiological tests showed that the response of DarmCSP2 knock-down individuals to (+)-αpinene, (+)-3-carene, (+)-β-pinene, myrcene, (–)-β-pinene, and (+)-camphene declined steeply compared to control insects (Li et al.). In the brown planthopper Nilaparvata lugens, Waris et al. showed that NlugCSP8 displayed wide binding abilities with rice plant volatiles such as nerolidol and hexanal in ligand binding experiments. Furthermore, the knock-down insects displayed the significant loss of attraction to the set of plant odorants (Waris et al.).

Functional studies of odorant carrying proteins as well as other olfactory proteins, could be of great importance for the improvement of a variety of pest control methods. One such application, known as "reverse chemical ecology," utilize the target olfactory proteins to screen for high affinity chemicals, which are in turn used to develop more effective agents for environmentally friendly pest management. In this Research Topic, two antennal-enriched OBPs of the dark black chafer beetle Holotrichia parallela, HparOBP20, and HparOBP49, bind with high affinity to the green leaf volatile 3-hexenyl acetate, which triggers high electrophysiological and behavioral activity for adult insects. Introduction of this compound into the field traps significantly increases male catches when combined with the sex pheromone (Ju et al.).

#### IDENTIFICATION OF PUTATIVE ODORANT-DEGRADING ENZYMES

ODEs include multiple enzyme families expressed in the sensillar lymph and likely involved in the fast inactivation of odorants to keep the olfactory system sensitive. In this Research Topic, Sun et al. suggested ODEs could convert ester odorants to inactive corresponding alcohols in E. obliqua. Expression analysis revealed that 12 CXEs are enriched in insect antennae. Complementary localization experiments determined that the signals of EbolCXE7 and EbolCXE13 engulfed not only in trichoid and basiconic olfactory sensilla but also in putative gustatory styloconic sensilla, indicating a putative dual role in both olfaction and gustation (Sun et al.). A second study determined the tissue expression profiles of 33 GSTs identified in S. littoralis antennae by transcriptomic approach, and highlighted four SlitGSTs (SlitGSTd2, e9, e15, and MGST1-3) dominantly expressed in olfactory organs and potentially acting as ODEs (Durand et al.). However, the enzymatic activities toward odorants for these ODE candidates need to be assessed. As a general mechanism, the "rapid" inactivation of odorants by ODEs needs more functional studies with different ODE families and insect species, as there are only few evidences that support this mechanism. As reflected in this Research Topic, ODEs draw much less attention from researchers than odorant receptors (ORs) and odorant carriers (OBPs and CSPs). However, major advances on the understanding of the function of these antennal enzymes could be expected in the future.

#### PROSPECTS, CHALLENGES, AND POTENTIAL APPLICATIONS ON PEST MANAGEMENT

Although numerous olfactory-related genes have been identified and functionally characterized, downstream applications in pest management are very limited. Nevertheless, achievements obtained in gene identification and functional clarification are critical for application aspects of pest and vector control. In this context, mosquitoes are major targets since they transmit various deadly viruses and other diseases to human populations around the globe. These vectors rely mainly on human emanations to seek and distinguish their hosts. In this Research Topic, Sparks et al. proposed feasible approaches for mosquito management by focusing on chemosensory receptors (especially ORs and IRs), including CRISPR-CAS9-mediated alterations of these receptors. Co-receptors like Orco and Irco (IR8a, IR25a, and IR76b) should be the primary candidate genes for mosquito management because they are well-conserved among all mosquito species (Sparks et al.). However, there are still several barriers on pest management by using chemosensory genes. Venthur and Zhou reviewed the progress on chemosensory gene characterization (ORs and OBPs) and compared the feasibility on pest management between ORs and OBPs by highlighting their advantages and drawbacks. OBPs lack ligand specificities but could be easily produced in vitro and crystallized due to their small size and solubility, allowing a ligand screening of potential odorant disruptants, in combination with molecular docking analysis and dynamic stimulations. In contrast, ORs present complex crystallization procedures, however, they generally are more specific to odorants, which indicates that disrupting ORs could interfere with specific behaviors of insects (Venthur and Zhou). With a deeper understanding of the molecular mechanisms of insect olfaction and technological advances (especially on protein 3-D modeling and ligand docking, gene editing and manipulation), we will witness breakthrough achievements on the application of olfaction-based techniques in control of pests and vector insects in the near future.

## AUTHOR CONTRIBUTIONS

PH, ND, and S-LD wrote, edited, and finalized this manuscript.

## FUNDING

This report was supported by the National Natural Science Foundation of China (Grant nos. 31860617, 31360528, and 31401750 to PH; 31372264 and 31672350 to S-LD), the Natural Science Foundation of Guizhou Province of China (Grant nos. QKH-J [2019]1109 and QKH-J [2014]2062 to PH), and the talented people of Guizhou QKH platform ([2017]5788 to PH).

## ACKNOWLEDGMENTS

We truly appreciate all authors' contributions to this Research Topic, which make it possible to illustrate the diversity of studies currently center on insect olfactory proteins. We also thank all reviewers and editors who assisted us and provided thorough comments and invaluable suggestions, as well as the Frontiers editorial team for its support on the Research Topic management.

**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 He, Durand and Dong. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identification and Expression Patterns of Anoplophora chinensis (Forster) Chemosensory Receptor Genes from the Antennal Transcriptome

#### Edited by:

*Peng He, Guizhou University, China*

#### Reviewed by:

*Min Lu, Institute of Zoology (CAS), China Robert Maxwell Collignon, United States Department of Agriculture, United States Jacob Wickham, Chinese Academy of Sciences, Institute of Chemistry, China*

#### \*Correspondence:

*Long-Wa Zhang zhanglw@ahau.edu.cn*

*† These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

Received: *27 November 2017* Accepted: *26 January 2018* Published: *13 February 2018*

#### Citation:

*Sun L, Zhang Y-N, Qian J-L, Kang K, Zhang X-Q, Deng J-D, Tang Y-P, Chen C, Hansen L, Xu T, Zhang Q-H and Zhang L-W (2018) Identification and Expression Patterns of Anoplophora chinensis (Forster) Chemosensory Receptor Genes from the Antennal Transcriptome. Front. Physiol. 9:90. doi: 10.3389/fphys.2018.00090* Long Sun1†, Ya-Nan Zhang2†, Jia-Li Qian1†, Ke Kang1,3†, Xiao-Qing Zhang<sup>1</sup> , Jun-Dan Deng<sup>1</sup> , Yan-Ping Tang<sup>1</sup> , Cheng Chen<sup>1</sup> , Laura Hansen<sup>4</sup> , Tian Xu<sup>4</sup> , Qing-He Zhang<sup>5</sup> and Long-Wa Zhang<sup>1</sup> \*

*<sup>1</sup> Anhui Provincial Key Laboratory of Microbial Control, School of Forestry & Landscape Architecture, Anhui Agricultural University, Hefei, China, <sup>2</sup> College of Life Sciences, Huaibei Normal University, Huaibei, China, <sup>3</sup> Forest Diseases and Insect Pests Control and Quarantine Station of Chaohu City, Chaohu, China, <sup>4</sup> College of Environmental Science and Forestry, State University of New York, Syracuse, NY, United States, <sup>5</sup> Sterling International, Inc., Spokane, WA, United States*

The citrus long-horned beetle (CLB), *Anoplophora chinensis* (Forster) is a destructive native pest in China. Chemosensory receptors including odorant receptors (ORs), gustatory receptors (GRs), and ionotropic receptors (IRs) function to interface the insect with its chemical environment. In the current study, we assembled the antennal transcriptome of *A. chinensis* by next-generation sequencing. We assembled 44,938 unigenes from 64,787,784 clean reads and annotated their putative gene functions based on gene ontology (GO) and Clusters of Orthologous Groups of proteins (COG). Overall, 74 putative receptor genes from chemosensory receptor gene families, including 53 ORs, 17 GRs, and 4 IRs were identified. Expression patterns of these receptors on the antennae, maxillary and labial palps, and remaining body segments of both male and female *A. chinensis* were performed using quantitative real time-PCR (RT-qPCR). The results revealed that 23 ORs, 6 GRs, and 1 IR showed male-biased expression profiles, suggesting that they may play a significant role in sensing female-produced sex pheromones; whereas 8 ORs, 5 GRs, and 1 IR showed female-biased expression profiles, indicating that these receptors may be involved in some female-specific behaviors such as oviposition site seeking. These results lay a solid foundation for deeply understanding CLB olfactory processing mechanisms. Moreover, by comparing our results with those from chemosensory receptor studies in other cerambycid species, several highly probable pheromone receptor candidates were highlighted, which may facilitate the identification of additional pheromone and/or host attractants in CLB.

Keywords: antennal transcriptome, expression pattern, odorant receptor, gustatory receptor, ionotropic receptor, Anoplophora chinensis

## INTRODUCTION

The citrus long-horned beetle (CLB), Anoplophora chinensis (Forster) (Coleoptera: Cerambycidae) is a polyphagous woodboring beetle native to China, Japan, and the Korean peninsula (Haack et al., 2010). This beetle has spread to Europe through international shipments of wood-packing materials and live plants from Asia and is a quarantine pest species on the European Union (EU) and European and Mediterranean Plant Protection Organization (EPPO) A1 list (Rizzi et al., 2013; Ge et al., 2014). It has a very broad range of host plants (>100 species from 19 families), of which 48 species are distributed in China (Ge et al., 2014). Larval infestation damages the vascular system and woody tissues of host plants, ultimately causing severe damage to ornamental and forest trees that may lead to mortality (Haack et al., 2010). As in most insects, CLB utilizes olfaction to recognize volatile cues that regulate a series of behaviors including mating, foraging, oviposition, and host-seeking. Recently, Yasui and Fujiwara-Tsujii (2016) discovered the sesquiterpene β-elemene can function as a femaleacquired repellant pheromone against males from a different host plant population in Anoplophora malasiaca, a synonym of A. chinensis, while Hansen et al. (2015) identified a maleproduced pheromone component, 4-(n-heptyloxy)butan-1-ol for A. chinensis.

Peripheral olfactory proteins include odorant binding proteins (OBPs), chemosensory proteins (CSPs), odorant receptors (ORs), ionotropic receptors (IRs), gustatory receptors (GRs), and sensory neuron membrane proteins (SNMPs) (Leal, 2013). ORs, GRs, and IRs are membrane-bound chemosensory receptors localized to sensillum chemosensory dendrites, bridge the gap between the extracellular odorant signal and the intracellular neurological response, and are critical for the olfactory response (Xu et al., 2015). These receptors are particularly attractive molecular targets for the development of new pest control strategies. ORs are seven transmembrane domain proteins with an inverted membrane topology (Ha and Smith, 2009; Leal, 2013). A heterometeric ligand-gated ion channel between an olfactory receptor co-receptor (Orco) and a more specialized OR is required in order to transduce odor-evoked signals (Gu et al., 2015). Orco acts as an ion channel and is highly conserved across insect orders and widely expressed in the majority of ORNs (Leal, 2013; Cattaneo et al., 2017). More specialized ORs may be tuned to a pheromone, certain plant volatiles, or other compounds (Ha and Smith, 2009; Liu et al., 2013; Cattaneo et al., 2017). Insect GRs are mainly expressed in gustatory receptor neurons (GRNs) of the gustatory organs (Ebbs and Amrein, 2007; Crava et al., 2016), but are also found in ORNs (Scott et al., 2001). Most insect gustatory organs are distributed on body surfaces such as proboscises, legs, wings, female genitals, and labial palps (Scott et al., 2001; Vosshall and Stocker, 2007). These GRs generally detect soluble compounds acquired from contact with a substrate, including sugars, amino acids, salts, and bitter compounds, but can also respond to carbon dioxide or pheromone signals (Ebbs and Amrein, 2007; Kwon et al., 2007; Sánchez-Gracia et al., 2009; Zhang et al., 2013). Insect IRs are a novel family of chemosensory receptors that are related to ionotropic glutamate receptors (iGluRs) (Benton et al., 2009; Croset et al., 2010), and act as ligand-based ion channels (Croset et al., 2010; Abuin et al., 2011). IRs are a more ancestral and conserved group of receptors than ORs and have been identified throughout protostomes, including nematodes, arthropods, mollusks, and annelids (Croset et al., 2010; Gu et al., 2015; Wang et al., 2015). Insect IRs are generally divided into two subfamilies, "antennal IRs," expressed in insect antennal ORNs, and species-specific "divergent IRs," mainly expressed in the gustatory organs and involved in the detection of tastants (Croset et al., 2010). Two well-conserved antennal IRs, IR8a, and IR25a, have a similar function to Orco and are diffusely expressed in insect ORNs (Croset et al., 2010; Kaupp, 2010; Abuin et al., 2011). IRs are essential for odor-evoked neuronal responses and for detecting environmental volatile chemicals and tastes (Croset et al., 2010; Ai et al., 2013; Rytz et al., 2013).

The objectives of our study were to (1) identify the chemosensory receptors (ORs, GRs and IRs) of A. chinensis via the antennal transcriptome sequencing, (2) examine the expression profiles of these receptors in multiple tissues of both sexes using quantitative real time PCR (RT-qPCR), (3) conduct a thorough comparison to the ORs identified in other cerambycid species including Megacyllene caryae and Anoplophora glabripennis, which may contribute to the identification of additional pheromone and host attractants in CLB, and (4) compare and contrast A. chinensis ORs identified in our study to those recently identified by Wang et al. (2017). Although there is some overlap, the strong disparities in research priorities [olfactory binding-protein genes families (OBPs and CSPs) vs. chemosensory receptor superfamilies (ORs, GRs, and IRs)], insect samples (sample size, collection sites and host plants) and total number of receptor genes identified between these two studies (see discussion for a detailed comparison) make both works complementary and valuable, and cross-validate each other.

#### MATERIALS AND METHODS

#### Insects and Tissue Collections

Live adult CLBs were collected from Acer rubrum stands in Hefei, Anhui Province, China in June, 2017. Forest Pest Control Station of Anhui Province issued the permit for the field collection (by the director, Jun Fu). Beetles were sexed and reared separately on fresh shoots of A. rubrum in clean, well-ventilated plastic cages (17.0 × 12.0 × 6.8 cm) at 25◦C and 75% RH. Excised female and male antennal tissues were immediately frozen in liquid nitrogen, and then stored at−80◦C for subsequent RNA-seq sequencing.

#### RNA Extraction, cDNA Library Construction and Illumina Sequencing

The antennae of both sexes were blended for total RNA extraction using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. RNA degradation and contamination were monitored on 1% agarose gel, RNA concentration was measured using Qubit <sup>R</sup> RNA Assay Kit with a Qubit <sup>R</sup> 2.0 Fluorometer (Life Technologies, CA, USA), and RNA purity was evaluated with a NanoPhotometer <sup>R</sup>

spectrophotometer (Implen, CA, USA). Illumina sequencing of the samples was performed at Novogene Co., Ltd., Beijing, China. Sequencing libraries were generated using NEBNext <sup>R</sup> UltraTM RNA Library Prep Kit for Illumina <sup>R</sup> (NEB, Ipswich, MA, USA) according to manufacturer's recommendations, and index codes were added to attribute sequences to each sample. Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature in 5X NEBNext First Strand Synthesis Reaction Buffer.

First strand cDNA was synthesized using random hexamer primer and M-MuLV Reverse Transcriptase (RNase H). Second strand cDNA synthesis was then performed using DNA Polymerase I and RNase H. Remaining overhangs were converted into blunt ends via exonuclease/polymerase activities. After adenylation of DNA fragment 3' ends, NEBNext Adaptor with hairpin loop structure was ligated to prepare for hybridization. The adaptor-ligated cDNA was incubated at 37◦C for 15 min followed by 5 min at 95◦C prior to PCR with Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. PCR products were purified (AMPure XP system) and library quality was assessed on the Agilent Bioanalyzer 2,100 system. Finally, library preparations were sequenced on an Illumina HiseqTM 4,000 platform and paired-end reads were generated.

#### Assembly and Functional Annotation

Clean reads were obtained from raw data by removing low quality reads and reads containing adapter or poly-N. A transcriptome was assembled based on clean reads using Trinity (Grabherr et al., 2011) to generate transcripts.

Unigenes were obtained from transcriptome assembly by choosing the longest transcript of each gene. BLASTx searches were used to align unigenes and compare them to the NCBI non-redundant (nr) protein database using an E-value threshold of 1 × 10−<sup>5</sup> . Unigenes were also annotated using other protein databases including Nt, Pfam, KOG/COG, Swiss-Prot, KO, and GO. ORFs of each unigenes were then predicted with ORF finder (http://www.ncbi.nlm.nih.gov/gorf/gorf.html) and the transmembrane domains of putative olfactory genes were determined using TMHMM Server v. 2.0 (http://www.cbs.dtu. dk/services/TMHMM/).

#### Phylogenetic Analysis

OR, GR, and IR amino acid sequences from A. chinensis and other insect species were aligned using ClustalX2.0. The OR data set contained identified sequences from A. chinensis (53), Tribolium castaneum (47), A. glabripennis (25), M. caryae (30), Dendroctonus ponderosae (12) (Andersson et al., 2013) and Ips typographus (19) (Andersson et al., 2013), along with 4 Orco genes from Phyllotreta striolata (Wu et al., 2016), Anomala corpulenta (Li et al., 2015), Monochamus alternatus (Wang et al., 2014) and Tenebrio molitor (Liu et al., 2015). The GR data set included 67 protein sequences reported from Drosophila melanogaster (7), and Bombyx mori (4), and the six coleopterans: A. chinensis (17), P. striolata (16), A. glabripennis (7), T. castaneum (11), D. ponderosae (2) and I. typographus (3) (Scott et al., 2001; Robertson et al., 2003; Wanner and Robertson, 2008; Guo et al., 2017). The IR data set contained sequences from A. chinensis (4), A. glabripennis (1), P. striolata (15), M. alternatus (6), A. corpulenta (5), D. ponderosae (10), I. typographus (4), T. molitor (6), and D. melanogaster (10). OR, GR, and IR unrooted phylogenetic trees were constructed using the MEGA6 neighbor-joining method (Tamura et al., 2013). Node support was assessed by bootstrap method with 1,000 bootstrap replicates.

## RT-qPCR Validation of ORs, GRs, and IRs

Expression profiles of putative chemosensory receptor unigenes in different body sections of both sexes were analyzed with RT-qPCR using an ABI 7300 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Total RNA was isolated from 20 antennae, 100 maxillary palps, 100 labial palps, and 20 bodies without antennae, maxillary palps, and labial palps from each sex, using the methods described above. Isolated RNA was reverse transcribed into cDNA using PrimeScript1 RT reagent Kit with gDNA Eraser (Perfect Real Time, Takara, Beijing, China). 2.5 ng cDNA was used as the RT-qPCR template. RT-qPCR target and reference gene primers were designed using Beacon Designer 7.9 software (PREMIER Biosoft International, Palo Alto, CA, USA) with CLB GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and actin reference genes (**Table S1**). The RT-qPCR reaction mixtures were composed of 20 µL 2×SYBR Green qPCR Master Mix-R (YIFEIXUE BIO TECH, Nanjing, China), 0.4 µL of both forward and reverse primer (10µM), 1 µL sample cDNA, and 8.2 µL sterilized H2O. RT-qPCR cycling parameters were set at 95◦C for 10 min, followed by 40 cycles of 95◦C for 15 s and 60◦C for 1 min. The Q-gene method (Simon, 2003) was used to calculate the expression levels of these genes in the four tissues from each sex. RT-qPCR data were analyzed and plotted using Graphpad Prism 5.0 (GraphPad Software, CA, U.S.A).The statistical classification of each target gene was calculated in each tissue with SPSS 22.0 (SPSS Inc., Chicago, IL, USA) using a oneway nested analysis of variance (ANOVA) followed by Duncan's new multiple range test (α = 0.05).

## RESULTS

## Transcriptome Sequencing and Unigenes Assembly

In total, 66,908,284 raw reads and 64,787,784 clean reads with a Q20 percentage of 97.02% were obtained from the CLB antennal transcriptome. From these, 44,938 unigenes were screened from 89,311 transcripts. Unigene and transcript mean lengths were 1392 and 842 bp, respectively, while N50 lengths were 2143 and 1718 bp, respectively. Length distribution analysis indicated that 33,989 unigenes, or 75.63% of all unigenes, were longer than 500 bp and the longest unigene was 26,202 bp (**Figure S1**). 26,701 unigenes (59.41%) were compared to proteins in the NCBI nonredundant (nr) protein database using the BLASTX algorithm (cut-off E-value of 10−<sup>5</sup> ). Homology analysis with other insect species revealed that T. castaneum was the best match (55.6%), followed by D. ponderosae (14.4%) and Lasius niger (1.8%) (**Figure S2**).

## Gene Ontology (GO) Annotation and KEGG Analysis

GO annotation was used to classify unigenes into different functional categories. Overall, Blast2GO (Götz et al., 2008) assigned 48.52% (21,808) of unigenes to three functional categories: cellular components (37,292), biological processes (60,089), and molecular function (27,255) (**Figure 1A**). In cellular components, cell part (7,260), cell (7,260), and organelle cellular component (5,113) were the most represented subcategories, in biological processes, cellular process (12,624), metabolic process (11,484) and single-organism process (10,031) were the most represented, and in molecular function binding (13,122) and catalytic activity (9,265) were most represented. KO annotation was used to classify 12,777 unigenes into five branches of the KEGG pathway (**Figure 1B**), including cellular processes (A), environmental information processing (B), genetic information processing (C), metabolism (D), and organismal systems (E).

## Identification of Putative Odorant Receptors

Antennal transcriptome analysis of CLB samples identified 53 putative ORs (**File S1**). Among these, 11 sequences contained a full-length ORF, and five genes (AchiOR1, AchiOR24, AchiOR32, AchiOR43, and AchiOR44) contained seven-transmembrane domains (**Table S2**). We identified an OR gene (AchiOR1) with a high sequence homology with the conserved Orco gene family of other insect species and have designated it as AchiOrco. Phylogenetic analysis in previous studies has divided coleopteran species ORs apart from the Orco gene subfamily (which includes AchiOrco, MaltOrco, McarOrco, PstrOrco, TmolOrco, and AcorOrco), into multiple subgroups numbered 1–7 (Engsontia et al., 2008; Andersson et al., 2013, **Figure 2**). 52 putative OR sequences were classified into four subgroups (group 1–3 and 7), with 19 sequences assigned to group 1, 18 sequences assigned to group 2, 10 sequences assigned to group 3, and five sequences assigned to group 7, respectively. Group 7 was further divided into two subsets: group 7a and group 7b. The remaining three subgroups 4–6 contained only T. castaneum sequences. Furthermore, 6 sequences (AchiOR22, AchiOR23, AchiOR26, AchiOR32, AchiOR34, and AchiOR44) were clustered with high orthology to pheromone receptors from M. caryae.

## Identification of Putative Gustatory Receptors

Bioinformatic analysis identified 17 putative GRs in the CLB antennal transcriptome (**File S1**); four of which were fulllength genes (**Table S2**). GR protein sequences from A. chinensis and seven additional insect species were used to construct a phylogenetic tree (**Figure 3**). In this tree, genes were classified into "sugar," "fructose," "bitter," and "CO2" GR functions. AchiGR1 was highly homologous to known sugar receptors (Chyb et al., 2003; Dahanukar et al., 2007; Kent and Robertson, 2009), AchiGR9 was highly homologous with a novel fructose sugar receptor (Sato et al., 2011; Miyamoto and Amrein, 2014), AchiGR6 and AchiGR15 were highly homologous to known bitter receptors (Wanner and Robertson, 2008), and AchiGR7 was highly homologous to known carbon dioxide receptors (Kwon et al., 2007; Robertson and Kent, 2009).

## Identification of Putative Ionotropic Receptors

Four putative IRs were identified in the combined antennal transcriptome (**File S1**). Among them, IR genes AchiIR2 and AchiIR3 had full-length ORFs, and the IR gene AchiIR4 was the only one without a transmembrane domain (**Table S2**). According to the phylogenetic analysis of IRs from eight species of coleopterans and D. melanogaster (**Figure 4**), IR genes can be classified into different subgroups. AchiIR2 clustered with DponIR76b and DmelIR76b at high percent identity, suggesting it belongs to the IR76b group. In addition, the phylogenetic tree classified AchiIR3 into the IR25a coreceptor subfamily.

#### Tissue- and Sex-Specific Expressions of Putative Chemosensory Receptors

Expression patterns of chemosensory receptors (53 ORs, 17 GRs, 4 IRs) in CLB antennae, maxillary palps, labial palps, and the remaining insect bodies of both sexes were determined using RT-qPCR. 41 putative OR genes were significantly expressed in the beetle antennae (**Figure 5**), of which antennal expression of 8 OR sequences (AchiOR2, AchiOR5, AchiOR10-11, AchiOR15, AchiOR25, AchiOR39, and AchiOR51) was significantly femalebiased, antennal expression of 23 OR sequences (AchiOR1, AchiOR3-4, AchiOR6, AchiOR12-14, AchiOR16-17, AchiOR19, AchiOR21, AchiOR27, AchiOR33-34, AchiOR36, AchiOR38, AchiOR42-43, AchiOR45-46, AchiOR48, and AchiOR52-53) was significantly male-biased, and the remaining 10 OR sequences (AchiOR7, AchiOR9, AchiOR18, AchiOR22, AchiOR26, AchiOR35, AchiOR37, AchiOR40, AchiOR44, and AchiOR50) were expressed at the same or similar levels in both female and male antennae. In addition, AchiOR49 was highly expressed in the maxillary palps. AchiOR20, AchiOR28, AchiOR30, and AchiOR47 were expressed at a significantly higher level in female bodies. Finally, AchiOR41 was highly expressed in the labial palps of both sexes.

11 of the 17 GR genes showed significantly higher expression in beetle antennae (**Figure 6**). Antennal expression of 5 GRs (AchiGR5-6, AchiGR9, and AchiGR14-15) was significantly female-biased, while antennal expression of the remaining 6 GRs (AchiGR3, AchiGR7-8, AchiGR10, AchiGR12, and AchiGR17) was significantly male-biased. AchiGR2 expression in female labial palps was significantly higher than in any other tissues, while AchiGR13 showed the highest expression in the female bodies. AchiGR11 was highly expressed in male labial palps and female bodies. Among the four IRs identified, AchiIR2 showed the highest expression in female antennae, whereas AchiIR4 was mainly expressed in male antennae. In addition, AchiIR1 and AchiIR3 showed similar expression levels among all tested tissues (**Figure 6**).

## DISCUSSION

Although Coleoptera is the largest insect order, the olfactory mechanisms of coleopterans at the molecular level are largely unknown. Furthermore, olfactory genes from Cerambycidae, an economically important coleopteran family, have only been partially identified in M. alternatus (Wang et al., 2014), Batocera horsfieldi (Li et al., 2014), M. caryae (Mitchell et al., 2012), A. glabripennis (Hu et al., 2016; Mitchell et al., 2017) and A. chinensis (Wang et al., 2017; and this paper).

In the transcriptome sets, a total of 44,938 unigenes were assembled from 89,331 transcripts, and 75.63% of these unigenes were longer than 500 bp, indicating the high depth and quality of the transcriptome sequences. The BLASTX homology analysis showed the best match with T. castaneum (55.6%), partly because a number of genes, including olfactory genes, were identified from genome data. GO and KO annotation exhibited some of the most represented subcategories: binding was the most abundant subcategory in the molecular function category, while signal transduction was the most abundant term in the environmental information processing pathway. The above unigenes may play vital roles in odorant binding and transduction activities in antennal chemosensory processes. CLB genes from the three multigene families of chemosensory receptors, including 53 ORs, 17 GRs, and 4 IRs, along with their expression patterns in different tissues of both sexes have now been identified through transcriptome analysis and RTqPCR.

Tcas: *T. castaneum*; Agla: *A. glabripennis*; Dpon: *D. ponderosae*; Ityp: *I. typographus*; Pstr: *P. striolata*; Acor: *A. corpulenta*; Malt: *M. alternates*; Tmol: *T. molitor*.

The 53 ORs identified in CLB were less than those identified in T. castaneum adult heads (111) (Engsontia et al., 2008) or P. striolata antennae and terminal abdomens (73) (Wu et al., 2016), but more than in A. glabripennis (37) (Hu et al., 2016), A. planipennis (2) (Mamidala et al., 2013), A. corpulenta (43) (Li et al., 2015), M. alternatus (9) (Wang et al., 2014), Brontispa longissima (48) (Bin et al., 2017), or Rhyzopertha dominica (6) (Diakite et al., 2016). According to the constructed OR phylogenetic tree (**Figure 2**), 52 putative OR sequences were distributed into four subgroups belonging to seven known coleopteran specific subgroups. In the present study, AchiOR1 was identified as AchiOrco due to the high level homology with the conserved Orco gene family, and clustered with other Orcos from M. alternatus, M. caryae, P. striolata, T. molitor, and A. corpulenta, probably attributed to the conserved nature of the chaperone OR. Interestingly, in the OR phylogenetic tree, six AchiOR genes, AchiOR22, AchiOR23, AchiOR26, AchiOR32, AchiOR34, and AchiOR44, were highly similar to three functionally characterized pheromone receptors (PRs), McarOR3, McarOR5, and McarOR20, from the cerambycid beetle M. caryae. Among them, AchiOR44 was orthologous to McarOR3, a receptor sensitive to the cerambycid pheromone (S)-2-methyl-1-butanol. AchiOR23, AchiOR26, and AchiOR34 formed a small clade around McarOR5, which is known to be sensitive to 2-phenylethanol, while AchiOR22 and AchiOR32 were clustered with McarOR20, a receptor of (2S,3R)-2,3 hexanediol and 3-hydroxyhexan-2-one. Mitchell et al. (2017) recently noted that the discovery of attractive volatile compounds could be expedited through further research on the expression of olfactory receptors. Due to their high level of sequence

similarity to the three PRs, McarOR3, McarOR5, and McarOR20, these AchiORs may be associated with the detection of the above pheromones or other behaviorally active compounds. The discovery of new attractive substances for CLB is necessary for pest management as the currently known attractants have yet to be developed into a commercially viable attractive lure.

Previous research has revealed that most insect OR expression is localized in the antennae (Vosshall et al., 1999; Wang et al., 2015). In the current study, 41 ORs showed an antenna-specific expression profile. Of these, the 23 ORs with male-biased expression may play a significant role in sensing female-produced sex pheromones and female-acquired host-derived sexual attractants, while the 8 ORs with femalebiased expression may be involved in some female specific behaviors such as oviposition site seeking, and 10 ORs, which were not biased toward either sex, may be associated with an aggregation pheromone or the detection of plant volatiles. Notably, AchiOR34 showed a clear male-biased expression profile and was clustered with pheromone receptors of M. caryae on the phylogenetic tree (**Figure 2**), strongly suggesting that it might be a pheromone receptor for sensing a female-produced sex pheromone in CLB. The ORs with high maxillary or labial palp expression may be involved in host selection for both sexes and oviposition site selection for females. A few ORs highly

indicate significant differences (*P* < 0.05).

expressed in non-olfactory tissues is consistent with what has been reported in other insects (Li et al., 2015; Zhang et al., 2016; Zhao et al., 2016).

Of the 17 putative GRs belonging to four function groups, AchiGR1 is a probable sugar receptor, AchiGR9 shared a high similarity with the fructose receptor family members that respond to D-fructose such as DmelGR43a and BmorGR9 (Sato et al., 2011; Miyamoto and Amrein, 2014), AchiGR6 and AchiGR15 showed a high degree of similarity to the bitter receptor family, and AchiGR7 may be involved in detecting CO2. Similarly to ORs, most GRs were prominently expressed in antennae, likely because all GRs were identified from antennal transcriptome rather than the complete genome. Several other GRs that were highly expressed in labial palps or other gustatory organs are likely involved in detecting soluble stimulants and feeding behaviors. Only 4 IRs were identified in CLB, less than that in P. striolata (49) (Wu et al., 2016) or B. longissima (19) (Bin et al., 2017), but similar to the number identified from long-horned beetles, A. glabripennis (4) (Hu et al., 2016) or M. alternates (7) (Wang et al., 2014). In the IR phylogenetic tree, AchiIR2 clustered with IR76b orthologs, while AchiIR3 clustered with coreceptor IR25a orthologs. Compared to ORs and GRs, IRs are involved in regulating sensory transduction of olfaction and gustation, and are expressed in both olfactory and gustatory organs (Croset et al., 2010; van Giesen and Garrity, 2017). Two of the four identified IRs showed markedly antennae-biased expression while the remaining two IRs were widely expressed in all the tested tissues.

As we were finalizing our present manuscript for submission, an independent and complementary work on CLB was published online by Wang et al. (2017) that focused on olfactory-bindingprotein gene families (OBPs and CSPs) rather than chemosensory receptor superfamilies. A total of 44 ORs, 19 GRs, and 23 IRs were identified by Wang et al. (2017), while 53 ORs, 17 GRs, and 4 IRs were identified in our current study. Five of our 17 AchiGRs had 100% identity with a counterpart and one AchiIR out of our four AchiIRs matched 100% with a corresponding IR reported in Wang et al. (2017). Only nine AchiORs presented 100% identity with a corresponding AchiORs. Notably, our AchiOR1 in our study, which clustered well with MaltOR01 and McarOR01, showed 100% identity with AchiOR35 in Wang et al. (2017), and both were defined as the conserved Orco gene. Our phylogenetic trees included sequences from M. caryae and A. glabripennis rather than those from Bombyx mori (Wang et al., 2017), and showed six of our AchiORs (AchiOR22, AchiOR23, AchiOR26, AchiOR32, AchiOR34, and AchiOR44) clustered well with three of the PR genes in M. caryae. An overview comparison of the receptor gene sequences identified in our study and Wang et al. (2017) using NCBI protein-protein BLASTP 2.6.0+ indicated that only 40 of our 74 identified receptors matched their recently published receptor genes with >90% amino acid identity (**Table S3**). We attribute our different identifications to our greater sample size, differences in collection sites and host plants, or other unseen reasons.

In the present study, ORs genes from the congener A. glabripennis were used to generate the neighbor-joining phylogenetic trees, surprisingly, these genes were not included in Wang et al. (2017) paper. An comparison of the 53 AchiORs identified in our study and 37 AglaORs in Hu et al. (2016) using NCBI protein-protein BLASTP 2.6.0+ indicated that at least 15 of our putative AchiORs showed high amino acid identities (up to 80%) with AglaORs (Hu et al., 2016, **Table S4**). Among them, AchiOR43 and AchiOR53 both had 100% identity with their corresponding genes AglaOR29 and AglaOR31 and the other four AchiORs (AchiOR8, AchiOR14, AchiOR19, and AchiOR49) had at least 95% identity with corresponding AglaORs. Closely related cerambycid species often share pheromones or pheromone motifs (Millar and Hanks, 2017). As the congeners CLB and A. glabripennis are previously known to both use 4-(n-heptyloxy)butanol as part of their pheromone systems (Hansen et al., 2015), the high level of homology between the AchiORs and AglaORs suggests that one or several of these may be pheromone receptor(s) tuned to 4-(n-heptyloxy)butanol. Further research on the functional characteristics of these receptors is surely needed.

Additionally, we conducted a further similarity analysis on the ORs to compare our 53 AchiORs with the 132 AglaORs sequences from the genome of A. glabripennis reported by Mitchell et al. (2017) (**Table S5**). Interestingly, all of the 53 AchiORs (except two ORs, AchiOR18, and AchiOR27) shared at least a 73% amino acid identity with AglaORs from the A. glabripennis genome, and 28 AchiORs had at least 95% identity with corresponding AglaORs. More importantly, our AchiOR1 (AchiOrco) was incredibly similar to AglaOR1/Orco, with 99.6% identity. This result further verified the attribute of AchiOR1 as a conserved Orco gene.

## CONCLUSION

In order to better understand the olfactory system molecular mechanisms of CLB, a polyphagous long-horned beetle that infests a wide range of broadleaved trees across many countries, we generated its antennal transcriptome. We then identified 74 putative receptor genes from the chemosensory receptor gene families, including 53 ORs, 17 GRs, and 4 IRs through bioinformatic analysis. RT-qPCR generated expression profiles of these chemosensory receptors demonstrated that most were prominently expressed in antennae, especially in male antennae, indicating that they may play a critical role in sensing sex pheromones. Functional characterization of putative pheromone receptors such as AchiOR34 in order to explore their binding capacity to known ceramycid pheromones, particularly pheromones of both A. glabripennis and A. chinensis, is a highly attractive future research objective. Our discovery of these chemosensory receptors may lead to a new perspective for controlling these economically important pest insects.

## AUTHOR CONTRIBUTIONS

L-WZ, LS, and Y-NZ: Conceived and designed the experiments; LS, J-DD, J-LQ, KK, and Y-NZ: Performed the experiments; LS, L-WZ, X-QZ, KK, Y-NZ, CC, and Y-PT: Analyzed the data; L-WZ and Y-NZ: Contributed reagents, materials, analysis tools; LS, L-WZ, KK, Y-NZ, LH, TX, and Q-HZ: Wrote the paper.

## FUNDING

This project was funded by the National Key Research and Development Program (2017YFD0600101), Natural Science Foundation of Anhui province, China (1508085SMC216), National Natural Science Foundation of China (31000304, 31170616, and31501647), Research Innovation Program for College Graduates of Anhui Agricultural University (Grant No. 2017YJS12), National Research Innovation Program for undergraduates Graduates (Grant No. 201710364019).

## ACKNOWLEDGMENTS

We thank Dr Tao Jing (Beijing Forestry University, China) for providing olfactory gene sequences of A. glabripennis and A. chinensis, and Bachelor students Chen-Guang Zhao, Cheng Zhan (Anhui Agricultural University, China) for help with insect collection. We also thank Dr Le-tian Xu (Hubei University, China), for his assistant on data processing, and Master students Xiao-Xue Xu (Anhui University, China) for RT-qPCR experimental guidance.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00090/full#supplementary-material

Figure S1 | Distribution of unigenes and transcript lengths in the *Anoplophora chinensis* transcriptome assembly.

Figure S2 | Species distribution of homology search with the *Anoplophora chinensis* unigenes against the Nr database.

Table S1 | Primes used for quantitative real time-PCR.

Table S2 | Blastx matches for the putative chemosensory receptor genes of *Anoplophora chinensis*.

Table S3 | Comparative overview of *Anoplophora chinensis* chemosensory receptor genes idenetified in this study and in Wang et al. (2017) study.

Table S4 | Comparative overview of *Anoplophora chinensis* chemosensory receptor genes identified in this study and *Anoplophora glabripennis* chemosensory receptor genes reported in Hu et al. (2016) study.

#### REFERENCES


Table S5 | Comparative overview of *Anoplophora chinensis* odorant receptor genes identified in this study and *Anoplophora glabripennis* OR genes reported in Mitchell et al. (2017) study.

File S1 | The amino acid sequences of *Anoplophora chinensis* putative chemosensory receptor genes.

Turpentine Beetle (RTB), Dendroctonus valens. PLoS ONE 10:e0125159. doi: 10.1371/journal.pone.0125159


of the cerambycid beetle Megacyllene caryae. Insect Biochem. Mol. Biol. 42, 499–505. doi: 10.1016/j.ibmb.2012.03.007


**Conflict of Interest Statement:** Author Qing-He Zhang is an employee at Sterling International, Inc. (SII), but not a SII shareholder or officer, and thus has no financial conflict of interest as it related to this work.

The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Sun, Zhang, Qian, Kang, Zhang, Deng, Tang, Chen, Hansen, Xu, Zhang and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Olfactory Plasticity: Variation in the Expression of Chemosensory Receptors in *Bactrocera dorsalis* in Different Physiological States

Sha Jin1, 2, Xiaofan Zhou<sup>3</sup> , Feng Gu1, 2, Guohua Zhong1, 2 \* and Xin Yi 1, 2 \*

<sup>1</sup> Key Laboratory of Crop Integrated Pest Management in South China, Ministry of Agriculture, South China Agricultural University, Guangzhou, China, <sup>2</sup> Key Laboratory of Natural Pesticide and Chemical Biology, Ministry of Education, South China Agricultural University, Guangzhou, China, <sup>3</sup> Guangdong Province Key Laboratory of Microbial Signals and Disease Control, Integrative Microbiology Research Centre, College of Agriculture, Guangzhou, China

#### *Edited by:*

Monique Gauthier, Université Toulouse III Paul Sabatier, France

#### *Reviewed by:*

Mauro Mandrioli, University of Modena and Reggio Emilia, Italy Nicolas Durand, University of Orléans, France

#### *\*Correspondence:*

Xin Yi yixin423@126.com Guohua Zhong guohuazhong@scau.edu.cn

#### *Specialty section:*

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

*Received:* 19 May 2017 *Accepted:* 23 August 2017 *Published:* 14 September 2017

#### *Citation:*

Jin S, Zhou X, Gu F, Zhong G and Yi X (2017) Olfactory Plasticity: Variation in the Expression of Chemosensory Receptors in Bactrocera dorsalis in Different Physiological States. Front. Physiol. 8:672. doi: 10.3389/fphys.2017.00672 Changes in physiological conditions could influence the perception of external odors, which is important for the reproduction and survival of insect. With the alteration of physiological conditions, such as, age, feeding state, circadian rhythm, and mating status, insect can modulate their olfactory systems accordingly. Ionotropic, gustatory, and odorant receptors (IR, GR, and ORs) are important elements of the insect chemosensory system, which enable insects to detect various external stimuli. In this study, we investigated the changes in these receptors at the mRNA level in Bactrocera dorsalis in different physiological states. We performed transcriptome analysis to identify chemosensory receptors: 21 IRs, 12 GRs, and 43 ORs were identified from B. dorsalis antennae, including almost all previously known chemoreceptors in B. dorsalis and a few more. Quantitative real-time polymerase chain reaction analysis revealed the effects of feeding state, mating status and time of day on the expression of IR, GR, and OR genes. The results showed that expression of chemosensory receptors changed in response to different physiological states, and these changes were completely different for different types of receptors and between male and female flies. Our study suggests that the expressions of chemosensory receptors change to adapt to different physiological states, which may indicate the significant role of these receptors in such physiological processes.

Keywords: olfactory plasticity, odorant receptor, ionotropic receptor, gustatory receptor, *Bactrocera dorsalis*

## INTRODUCTION

Insects need to respond to environmental cues in line with their own physiological states. Olfactory plasticity enables them to modify their response to odors according to age, feeding state, circadian rhythm, or mating status (Gadenne et al., 2016). For example, a sense of satiety partly determines food ingestion behaviors (Croll and Chase, 1980). Starvation can increase the response of Drosophila melanogaster to food signals, the effect escalates over the starvation time (Edgecomb et al., 1994). In mosquitoes, a blood meal can promote ovarian development and inhibit feeding behaviors (Klowden and Lea, 1979a,b). After ingesting sufficient amounts of blood, mosquitoes devote more energy to find suitable oviposition sites and become more attracted to relevant signals (Gadenne et al., 2016). The regulation of olfactory system also occurred upon mating at all levels. Generally, after mating, insects become less sensitive to sexual signals, whereas are inclined to be attracted by oviposition-site cues or food odors. The antennal neurons become less sensitive to sex pheromones, while the responses to host plant odors are unchanged after mating in Spodoptera littoralis males (Kromann et al., 2015). Moreover, like many other activities, olfactory behaviors vary by the time of day (Gadenne et al., 2016). Studies showed that olfactory sensory neurons and antennal sensitivity of many insect species tend to be controlled by the biological clock or body rhythm (Krishnan et al., 1999; Page and Koelling, 2003).

Three molecular components of the insect chemosensory system have been identified to be significant for the perception and recognition of odorant stimuli: the odorant receptors (ORs), the ionotropic receptors (IRs), and gustatory receptors (GRs) (Benton et al., 2009; Agnihotri et al., 2016; Shen, 2017). They are indispensable for detecting a wide range of environmental stimuli: bitterness, sweetness, odor, pheromones, humidity, carbon dioxide, and carbonated water (Bargmann, 2006; Vosshall and Stocker, 2007). To cope with various external stimuli, olfactory plasticity may regulate insect olfactory system by changing the expressions of these receptors when the insect enters specific physiological states (Su and Wang, 2014).

The oriental fruit fly, Bactrocera dorsalis (Hendel), is one of the main fruit pests in the Asia-Pacific region. They inhabit broad range of host species, have wide climate tolerance, display high dispersal capacity, and significant fecundity (Hsu et al., 2016). During development, adult oriental fruit fly lay eggs inside the fruits of different types of host plants for feeding and oviposition (Zheng et al., 2012). This makes the fly a highly invasive polyphagous species. Due to the direct damage to crops and negative effects on export markets, they are considered a quarantine pest. Our group has conducted a series of studies concerning the chemosensory system of oriental fruit fly (Yi et al., 2013, 2014). In B. dorsalis, chemosensory perception plays a key role in behavior regulation such as, host-seeking, mating and oviposition. It also exhibits remarkable developmental phases in olfactory behaviors (Wu et al., 2016). In this study, we examined the de novo transcriptome of B. dorsalis and identified 21 IR, 12 GR, and 43 OR genes. To test whether insect can modulate their olfactory system via the changes in the expression of chemosensory receptors according to their physiological states, we performed quantitative real time PCR to examine expression patterns of receptor genes to different physiological states, including feeding states, time of day and mating status. Our results illuminate the potential role of the chemosensory receptors in physiological state-dependent forms of olfactory plasticity.

#### MATERIALS AND METHODS

#### Insect Rearing

B. dorsalis were obtained from a laboratory-reared stock colony (Key Laboratory of Pesticide and Chemical Biology, South China Agricultural University, Guangzhou, China) and maintained at 28◦C in 70% relative humidity, with a 14:10 (light: dark) photoperiod. Adult flies were reared on artificial diets consisting of yeast extract, sugar, honey, and agar (Wu et al., 2015).

#### RNA Isolation

Total RNA was isolated from antennae from female and male adult flies (female/male ratio = 1:1) at different stages, including 2 d after eclosion, sexual immaturity (8 d after eclosion), sexual maturity but unmated (12 d after eclosion, females and males in two separate cages before maturity), and mated (15 d after eclosion). Antennae isolated from flies of different stages were mixed together, and the amount of the antennae sample was more than 5 mg. We extracted total RNA by the RNA isolation kit (Omega, USA) according to the manufacturer's instructions. We measured the concentration of isolated RNA using Nanodrop (Thermo Fisher Scientific, USA).

## Construction of the cDNA Library and Illumina Sequencing

We constructed the cDNA library using an Illumina kit following the manufacturer's recommendations. mRNA was split into small fragments after purification with oligo (dT) magnetic beads. Using mRNA as a template, we synthesized the first strand of cDNA using a random hexamer primer. Then, to obtain doublestrand cDNA, we added buffer for reverse transcriptase, dNTPs, RNase H, and DNA polymerase I. For end repair and poly (A) addition, the double-stranded cDNA was purified. Finally, the 5′ and 3′ ends of the fragments were ligated. Suitable fragments, as judged by agarose gel electrophoresis, were selected for use as templates for PCR amplification to create a cDNA library. The cDNA library was sequenced on an Illumina sequencing platform (HiSeqTM 2000) and 100 bp paired-end reads were generated (Zhang et al., 2015).

#### Identification of Olfactory Genes

Trinity was used to perform de novo transcriptome assembly on the filtered reads. We further processed the initial assembly generated by Trinity by CD-HIT to remove redundant transcripts and by COREST to cluster together transcripts that shared a high number of reads. We identified potential gene coding regions (hereafter "genes") from the final transcriptome assembly using TransDecoder. Then we analyzed translated amino acid sequences of all identified genes using InterProScan v5.23 and the sequences that contain the characteristic domains of insect chemosensory receptors (IR: IPR001320; GR: IPR009318 and IPR013604 and OR: IPR004117) were identified as B. dorsalis IR, GR, and OR proteins, respectively.

For each of the IR, GR, and OR gene families, we aligned amino acid sequences of genes in B. dorsalis (identified in this study) and D. melanogaster (collected from previous studies) using MAFFT v7.310 with the high accuracy option "E-INS-I." Columns with high proportion of gaps were filtered from the resultant multiple sequence alignments using TrimAL v1.4 with the "-gappyout" option. Evolutionary models best fitting the data were selected using the "ModelFinder" feature of IQ-TREE v1.5.4, the models "LG+F+R5" and "LG+F+R7" were selected for OR/GR and IR, respectively. Phylogenetic reconstructions were then conducted on the filtered alignments using IQ-TREE v1.5.4 with the "-nstop" parameter (maximum number of continuous unsuccessful iterations) set to 500. The reliabilities of estimated phylogenetic trees were assessed using the ultra-fast bootstrap (1,000 replicates) implemented in IQ-TREE v1.5.4. The phylogenetic trees were visualized using the interactive tree of life (iTOL) server v3.

## Collection of Sample in Different Physiological States

For the assessment of feeding state, flies (15 d after eclosion) were starved for 4 h, or 8 h, respectively; un-starved flies were kept in cages with the usual food supply. There were 100 flies (female or male) in each experimental group, and each treatment was performed with three replicates. After starvation, antennae from female and male flies were isolated separately. The antennae were collected at the same time and stored at −80◦C before RNA extraction.

For the assessment of mating states, sexually mature female and male flies (14 d after eclosion) were separated into two different cages to avoid mating, but were raised under the same conditions. Female and male flies in the mated group were raised together in the same cage. There were 100 flies (female or male) in each experimental group, and each treatment was performed with three replicates. The antennae of all groups were collected at the same time and stored at −80◦C before RNA extraction. Female and male flies were examined separately.

Gravid female flies were collected at different times of day (9 a.m., 4 p.m., and 10 p.m.) to test olfactory gene expression levels at different time points. Only female flies were included in each group with three replicates. The antennae were collected at the same time and were stored at −80◦C before RNA extraction.

#### Expression Level Examination

We investigated expression patterns of all identified olfactory genes by qRT-PCR. qRT-PCR was performed as previously reported (Yi et al., 2014). We reverse-transcribed 1 µg isolated RNA to first-strand cDNA by using M-MLV reverse transcriptase (TaKaRa, China) and oligo(dT)<sup>18</sup> as primer at 42◦C for 60 min. The reaction was terminated by heating at 95◦C for 5 min, and the products were stored at −20◦C. We performed qRT-PCR using the iCycler iQ Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) with SYBR green dye (Taraka, China) binding to double-strand DNA at the end of each elongation cycle. Amplification was performed by the primers listed in Supplementary Table 1. All amplifications were performed with three biological replicates. We analyzed relative gene expression data using the 2-11CT method as described by Livak and Schmittgen (2001).

## Heat Mapping

Phylogenetic trees were made with the maximum likelihood method with multiple alignments of amino acid sequences of identified BdorIRs, BdorGR, and BdorORs, respectively. Bootstrapping supports were indicated beside the branches at 1,000 simulations. Changes at expression levels were calculated as the ratio of different experimental groups. Log<sup>2</sup> scale of fold changes of each treatment group relative to the control group were shown in the heat map. Orange color indicated that the expression level was significantly increased, while blue color indicated that the expression level was significantly decreased. White color indicated that the expression levels between two different treatments were not significant changed.

## Statistical Analysis

All results from experimental replicates were expressed as means (±S.E.M) and analyzed with one-way analysis of variance or t-tests using SPSS 17.0 for Windows (SPSS Inc., Chicago, IL, USA).

## RESULTS

#### Identification of *B. dorsalis* Chemosensory Receptors by *De novo* Transcriptome Assembly

In total, 25,232,772 clean reads were obtained from the antennal transcriptome of B. dorsalis. These reads were assembled into 56,899 unigenes, with an average length of 1449.02 bp and an N50 of 2131. 93.4 % clean reads aligned on the assembly. The data were deposited at National Center for Biotechnology Information (NCBI) under SAR database, with the accession number SRR5801940. We then performed the de novo assembly of the antennal transcriptome of B. dorsalis and identified 21 IRs, 12 GRs genes, and 43 ORs (Supplementary Datasheet 1). Then we investigated the evolutionary relationships among the IR, GR, and OR genes in B. dorsalis and D. melanogaster. Our results showed that most B. dorsalis chemoreceptor genes (19/21 IRs, 9/12 GRs, and 30/43 ORs) had well-supported (co-)orthologs in D. melanogaster (**Figures 1**–**3**). As shown in **Figures 1**, **2**, the IR and GR genes in B. dorsalis were named after their counterparts in D. melanogaster when our phylogenetic analyses indicated clear orthologous relationships between them. For OR genes, however, evolutionary relationships between B. dorsalis and D. melanogaster genes were much more complicated: most B. dorsalis OR genes had either no or multiple orthologous genes in D. melanogaster. Therefore, for this gene family, we chose to rename B. dorsalis genes in numerical order (**Figure 3**). Moreover, many B. dorsalis chemoreceptor genes exhibited oneto-one relationship with their counterparts in D. melanogaster, including the highly conserved Orco (Or co-receptor) and "antennal" IR genes. However, a few species-specific gene duplications were also observed. Note that there was a B. dorsalisspecific clade in the OR family consisting of 10 genes, while the ortholog in D. melanogaster (DmelOr7a) contain single-copy (**Figure 3**), which suggested that B. dorsalis expanded rapidly after the divergence of the ancestors of the two species. We also compared the sequences identified in this study with those of previous studies (Supplementary Table 2). Our study covered almost all receptors reported in previous papers: 10/11 IRs, 6/6 GRs, and 22/23 ORs in study of Wu et al. (2015) and 10/12 IRs, 32/35 ORs in study of Liu et al. (2016), indicating a very robust analysis of our data.

## Variations in Chemosensory Receptor Expression before and after Mating

To reveal how mating affects the expression of chemosensory receptors in B. dorsalis, we measured expression levels of the

receptors before and after mating by qRT-PCR. In total, 13 IRs in females and 7 IRs in males showed significant changes after mating. However, the pattern of change was different between male and female. Expression of most IRs (16/21) in female flies was down-regulated after mating. In contrast, expression of most IRs (13/21) in male flies was increased after mating (**Figure 4A**). Most GRs showed a negligible change in expression after mating. However, a few GRs showed similar changes in males and females (the expression of GR2 and GR28b increased in both male and female; and the expression of GR21a, GR39a, GR59, GR63a, GR64f decreased in both male and female). The largest changes occurred in GR21a of female flies and GR2 of male flies, which decreased 1.83-fold and increased 3.20 fold, respectively, after mating (**Figure 4B**), which indicates their potential role in mating or after mating. The expression levels of ORs in B. dorsalis varied significantly after mating. Almost all identified ORs (39/43) were down-regulated by mating in female flies (**Figure 4C**). However, the pattern of changes in male flies had distinct differences from that in female (10 ORs were upregulated and 14 ORs were down-regulated by mating). This indicates that ORs may function differentially in sensing male and female pheromone during mating processes. Specifically, the expression of OR10 and OR2f in mated males was 12.41-fold and 17.11-fold higher than that in unmated male flies, implying an essential requirement of these two ORs in sexual pheromone sensing in male flies (**Figure 4C**).

## Variation in Chemosensory Receptor Expression at Different Points during Egg Laying

Next we studied expression at different times of the day. Because the egg—laying peak of B. dorsalis occurred at 4 p.m., we collected flies at 9 a.m., 4 p.m. and 10 p.m. to analyze the expression levels of the chemosensory receptors before and after the egg laying peak. For all IRs, expression at 9 a.m. and 4 p.m. tended to be stable, and the highest or lowest expression often occurred at 10 p.m. Six IRs (IR40a, IR64a, IR76b, GluR2, GluR3 and GluR4) showed the highest expression at 10 p.m., while 11 IRs showed the lowest expression at this time. The expression of five IRs (IR8a, IR21a, GluR2, GluR4, and GluR5) displayed sharp fluctuations over these three time points (**Figure 5A**). Except for GR1, all other GRs showed downtrend of expression along

the three time points, reaching their lowest point at 10 p.m. (**Figure 5B**). Similar to IRs, many ORs (17/43) exhibited the highest expression at 10 p.m., whereas 19 ORs showed the lowest expression at this time (**Figure 5C**). In addition, 15 ORs showed the highest expression at 9 a.m., whereas 11 ORs showed the highest expression levels at 4 p.m., which indicated that these ORs may be involved in oviposition behaviors. The expression levels of 15 ORs showed tendency to descend along with the timeline, while the expression levels of 12 ORs were on the decrease along with the time, which indicated that the expression changes of these ORs may follow an internal clock to maintain a certain rhythm.

#### Variation in Chemosensory Receptor Expression in Different Feeding Conditions

Feeding state also affects the chemosensory recognition behaviors of insects. Flies were starved for 4 or 8 h (the survival rate was 100% after starvation treatment), and expression of the receptors was measured with un-starved flies serving as the control. Starvation induced decreased expressions of most IRs in both males and females, with only a few exceptions that showed dramatic increase after 4 h starvation (IR21, IR68a, IR93a in female flies, and GluR3, IR25 in male flies; **Figure 6A**). For GRs, 4 h of starvation induced rapid increase of expression of seven GRs in female flies, followed by decrease with the prolongation of starvation time. In male flies, starvation brought about a sustained decrease in the expression of most GRs (**Figure 6B**). Expression of most ORs decreased after starvation in both female and male flies. Some exceptions occurred at 4 h starvation time point in female flies: some ORs showed increased expression after 4 h starvation. This may indicate that female flies become more sensitive to food odors after 4 h of starvation, and those ORs may play a role in detecting food chemicals. Three ORs in male flies were up-regulated by starvation. The increased expression of these three genes may be involved in food-searching behavior in starved male flies (**Figure 6C**).

#### Heat Mapping

Mating behavior decreased expression of most ORs in female flies, and increased expression of most ORs in male flies. Expression of GRs and IRs did not show significant change after mating in either female or male flies. Starvation for 4 h increased expressions of many ORs and GRs genes in female flies, although expression decreased when the female flies were starved for 8 h.

However, the scenario was different for male flies: starvation for 4 and 8 h generally decreased expression of many chemosensory receptors. The expression of one cluster of ORs and most IRs showed an upward trend over three time points during the day, whereas the expression of all identified GRs showed a downward trend (**Figure 7**).

#### DISCUSSION

In this study, we focused on three crucial families of chemosensory receptors: IRs, GRs, and ORs. Using RNAsequencing and de novo transcriptome analysis, we identified 21 IR, 12 GR, and 43 OR genes in the antennae of B. dorsalis. Previously, Wu et al. performed transcriptome sequencing on mixed tissues at different developmental stages (egg, larva, pupa, and adult) of B. dorsalis and reported 23 ORs, 6 GRs, and 11 IRs (Wu et al., 2015). Liu et al identified 12 IRs and 35 ORs in the transcriptome of the antennae of male and female oriental fruit flies (Liu et al., 2016). In comparison, our analysis of the antennae not only covered nearly all previously reported receptor genes (Supplementary Table 2), but also identified a number of new chemoreceptor genes. To our knowledge, many chemosensory receptors were annotated and reported for the first time in this species of fly, however, some receptors may be difficult to be identified by transcritome analysis. This may be partly due to the limit of quality of transcriptome sequencing annotation, and the fact that many chemosensory receptors are expressed in the tissues other than antennae (Rinker et al., 2013).

Behavioral responses and sensitivity to plant odors, sex pheromones, or oviposition cues may be influenced by mating status, which indicates that the expression of chemosensory receptors may change before and after copulation. Our results showed that expression of most IRs and ORs decreased after mating in female flies, but not in male flies. This indicates the potential role of these chemosensory receptors in recognition of sexual signals in female flies, as olfactory responses to sexual signals are switched off very rapidly after mating. In a previous paper, OR19 was identified as highly expressed in male antennae (Liu et al., 2016), which is consistent with our results that BdorOR17 (same gene as OR19) is up regulated in the antennae of starved males compared to females. This result

in B. dorsalis before and after mating. The data represent the mean ± S.E.M of three replicates (\*\*\*p < 0.001, \*\*p < 0.01, \*p < 0.05, t-test).

FIGURE 6 | Variation in chemosensory receptors at different feeding states in Bactrocera dorsalis. Variation in (A) BdorIRs, (B) BdorGRs, and (C) BdorORs expression at different feeding states of B. dorsalis. The data represent the mean ± S.E.M of three replicates. Different letters indicate significant differences in expression (p < 0.05, one-way analysis of variance).

suggested its important role in male flies. The mating-related changes of receptor expression were more obvious in female flies, which is consistent with the observation that the female flies have decreasing sexual receptivity and increasing urge to find appropriate place for egg production after copulation (Krupp and Levine, 2014; Hussain et al., 2016). Many mated insects are more attracted to food odors compared to unmated ones (Hussain et al., 2016). The observed increase of expression levels of many receptors in male flies may account for the increased attraction to food signals after mating behavior. In this study, expression of almost all GRs was up-regulated by mating, which indicates the important role of GRs in the process of courtship. In previous studies, GR32a was suggested to function as a pheromone receptor of a male inhibitory pheromone (Miyamoto and Amrein, 2008), and GR39a plays a role in sustaining courtship behavior in males (Watanabe et al., 2011). Our results also showed that the variation pattern of IRs is different in male and female flies, indicating different involvements of these receptors during mating between female and male flies. For example, IR52c and IR52d played a role in male mating behavior and sexually dimorphic expression in neurons of the male foreleg to contact female during courtship (Koh et al., 2014). Further study needs to illustrate functional difference of IRs in male and female during courtship and mating.

Olfactory behaviors vary at different times of the day, which helps the insect respond well over the time (Gadenne et al., 2016). The biological clock is a cell-autonomous system that coordinates physiology and metabolism to align behavioral processes with the day/night cycle (Bass and Takahashi, 2010). The rhythm is important for reproduction through its effect on ovulation (Zhang et al., 2017). We observed that the egg-laying peak occurred at 4 p.m. for the female B. dorsalis, consistent with a previous report (Yang et al., 1994). Drosophila suzukii has a peak in oviposition activity at 8 p.m., whereas this peak pattern occurs from 4 p.m. to 4 a.m. in D. melanogaster (Lin et al., 2014). The peak in expression of chemosensory proteins may correspond to the time of increasing chemosensory activity to odor signals (Rund et al., 2013). Expression of one cluster of ORs and some IRs showed clear up-regulation at 10 p.m. and 4 p.m. compared to 9 a.m., which suggests that these ORs may be required for the recognition of oviposition cues. One oviposition-related chemical in Anopheles gambiae was culicine water, which could negatively influence the peak oviposition time (Sumba et al., 2004). The relatively higher expression of some chemosensory receptors at 10 p.m. may be the result of the retention of mature oocytes at the beginning of the night (Allemand and David, 1984). The highest expression of most GRs in gravid female oriental fruit flies occurred at 9 a.m., before peak oviposition, which suggests that GRs may play a role in the recognition of stimulants and subsequent signal transduction inducing oviposition behavior (Ozaki et al., 2011). The roles of some GRs in oviposition process were documented previously. For instance, GR5a can detect trehalose (Chyb et al., 2003), and function as a receptor for caffeine (Moon et al., 2006), GR21a and GR63a can mediate CO<sup>2</sup> detection (Jones et al., 2007), and GR68a acts in pheromone reception in courtship behavior (Bray and Amrein, 2003).

The expression of chemosensory receptors was significantly higher (orange) or lower (blue) at different physiological states; non-differentially expressed chemosensory receptors are denoted as zeros (white). All measurements were made with the log2 fold change scale.

Besides the conventional role of GRs in insect chemoreception, GR28b also acts as a thermoreceptor to mediate warmthsensing (Barbagallo and Garrity, 2015). Therefore, it is also very important to investigate the changes of GRs expression in response to thermal variations. Our current study indicates that the biological clock plays an important role in the regulation of insect chemosensory receptors expression levels, especially for ORs and GRs.

The nutritional state of the insect can also influence olfactory behaviors. A blood meal could regulate sensitivity to many kinds of odor in blood-feeding insects, including odors from food—resource and oviposition cues (Gadenne et al., 2016). Starvation affected sensitivities of all ORs and IRs in all sensillum types (Farhan et al., 2013), resulting in increased behavioral and physiological sensitivity to food blend (Root et al., 2011). In this study, we found that starvation induced decreased expression of almost all candidate chemosensory receptors in male flies. In contrast, the expression levels of most chemosensory receptors increased after 4 h starvation treatment and decreased with the longer starvation time in female flies. Previous studies showed controversy over the starvation effect on physiological responses. Root et al. suggested that starvation could increase the activity of olfactory sensory neurons (Root et al., 2011), whereas Farhadian did not find any effect of starvation on olfactory sensory neurons that expressed OR47a (Farhadian et al., 2012). In one study, starved flies were more sensitive to odors, a response regulated by short neuropeptide F receptor (sNPF) (Root et al., 2011). Another study showed that starvation increased the sensitivity of olfactory receptors cells to odors, whether they expressed sNPF or not (Farhan et al., 2013). More evidence is needed on the effect of starvation on recognition of chemicals. Although changes in mRNA expression might not necessarily be sufficient to explain behavioral changes, the observation of differential effect of starvation on the expression of chemosensory receptors between male and female flies suggests gender-specific difference of olfactory recognition in response to starvation.

#### REFERENCES


In conclusions, we investigated variations in the expression of chemosensory receptors in different physiological conditions. The results in this study provided us an overview of expression patterns of chemosensory receptors in response to different physiological changes in oriental fruit flies. Our study could lay a foundation for further functional validation of specific chemosensory receptors in different physiological processes of insects.

#### AUTHOR CONTRIBUTIONS

SJ, XZ, and FG performed the experiments: XZ analyzed the data, and XY and GZ wrote and revised the manuscript.

#### ACKNOWLEDGMENTS

This study was funded by Science and Technology Planning Project of Guangdong Province (2016A020210090) and Youth Science and technology training project of South China Agricultural University.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00672/full#supplementary-material


(Bactrocera dorsalis). PLoS ONE 11:e0147783. doi: 10.1371/journal.pone. 0147783


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Jin, Zhou, Gu, Zhong and Yi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Characterization and Comparative Analysis of Olfactory Receptor Co-Receptor Orco Orthologs Among Five Mirid Bug Species

Qi Wang<sup>1</sup> , Qian Wang1,2, Yan-Le Zhou1,3, Shuang Shan<sup>1</sup> , Huan-Huan Cui <sup>1</sup> , Yong Xiao<sup>1</sup> , Kun Dong<sup>1</sup> , Adel Khashaveh<sup>1</sup> , Liang Sun<sup>4</sup> \* and Yong-Jun Zhang<sup>1</sup> \*

#### Edited by:

*Peng He, Guizhou University, China*

#### Reviewed by:

*Ya-Nan Zhang, Huaibei Normal University, China Yihan Xia, Lund University, Sweden Joe Hull, Agricultural Research Service (USDA), United States*

#### \*Correspondence:

*Yong-Jun Zhang yjzhang@ippcaas.cn Liang Sun liangsun@tricaas.com*

#### Specialty section:

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

Received: *14 November 2017* Accepted: *16 February 2018* Published: *05 March 2018*

#### Citation:

*Wang Q, Wang Q, Zhou Y-L, Shan S, Cui H-H, Xiao Y, Dong K, Khashaveh A, Sun L and Zhang Y-J (2018) Characterization and Comparative Analysis of Olfactory Receptor Co-Receptor Orco Orthologs Among Five Mirid Bug Species. Front. Physiol. 9:158. doi: 10.3389/fphys.2018.00158* *<sup>1</sup> State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China, <sup>2</sup> Department of Plant Protection, College of Horticulture and Plant Protection, Yangzhou University, Yangzhou, China, <sup>3</sup> DanDong Entry-Exit Inspection and Quarantine Bureau, Dandong, China, <sup>4</sup> Key Laboratory of Tea Quality and Safety Control, Ministry of Agriculture, Tea Research Institute, Chinese Academy of Agricultural Sciences, Hangzhou, China*

The phytophagous mirid bugs of *Apolygus lucorum*, *Lygus pratensis* as well as three *Adelphocoris* spp., including *Adelphocoris lineolatus*, *A. suturalis,* and *A. fasciaticollis* are major pests of multiple agricultural crops in China, which have distinct geographical distribution and occurrence ranges. Like many insect species, these bugs heavily rely on olfactory cues to search preferred host plants, thereby investigation on functional co-evolution and divergence of olfactory genes seems to be necessary and is of great interest. In the odorant detection pathway, olfactory receptor co-receptor (Orco) plays critical role in the perception of odors. In this study, we identified the full-length cDNA sequences encoding three putative Orcos (*AsutOrco*, *AfasOrco,* and *LpraOrco*) in bug species of *A. suturalis*, *A. fasciaticollis,* and *L. pratensis* based on homology cloning method. Next, sequence alignment, membrane topology and gene structure analysis showed that these three Orco orthologs together with previously reported AlinOrco and AlucOrco shared high amino acid identities and similar topology structure, but had different gene structure especially at the length and insertion sites of introns. Furthermore, the evolutional estimation on the ratios of non-synonymous to synonymous (Ka/Ks) revealed that Orco genes were under strong purifying selection, but the degrees of variation were significant different between genera. The results of quantitative real-time PCR experiments showed that these five Orco genes had a similar antennae-biased tissue expression pattern. Taking these data together, it is thought that Orco genes in the mirid species could share conserved olfaction roles but had different evolution rates. These findings would lay a foundation to further investigate the molecular mechanisms of evolutionary interactions between mirid bugs and their host plants, which might in turn contribute to the development of pest management strategy for mirid bugs.

Keywords: olfactory receptor co-receptor, mirid bugs, gene structure, sequence analysis, evolution analysis

**37**

## INTRODUCTION

Due to long-term adoption of transgenic Bt (Bacillus thuringiensis) cotton and the associated reduction in broadspectrum insecticide used for controlling Helicoverpa spp. (Wu et al., 2008), several species of the mirid bugs (Hemiptera: Miridae) including Apolygus lucorum, Lygus pratensis as well as three Adelphocoris spp., including Adelphocoris lineolatus, A. suturalis and A. fasciaticollis have become most important pest species in cotton fields of northern China (Lu et al., 2010). Besides cotton, these polyphagous mirid species cause severe destructions to many other important crops including vegetables, fruits trees and tea plants (Lu and Wu, 2008). It was reported that these five mirid species are significantly different in geographic distribution and seasonal abundance in China (Lu et al., 2008a). The A. lucorum is widely distributed in whole China, while three Adelphocoris species and L. pratensis occur mainly in Yangtze River region and the northern parts of Yellow River region, and in the colder region of northwest China, respectively (Lu and Wu, 2008). The screening of overwintering and early season host plant ranges suggested that mirid bugs from different regions employed distinct host plant ranges to survive winter and early spring, and these differences are significantly linked to their reliance on local plants (Lu et al., 2011). Consequently, the interactions between mirid species and local host plants should play crucial roles in determining ecological landscape-level especially their different geographic distribution and seasonal abundance. A better understanding of the underlying species-preferential host plants tracking would help to define co-evolution between different mirid species and their host plants, and ultimately facilitate the development of regional forecasting and pest management strategies.

Insect olfaction plays important roles in locating host plant. Several classes of molecules including odorant binding proteins (OBPs), chemosensory proteins (CSPs), odorant receptors (ORs), sensory neuron membrane proteins (SNMPs) and odorant degradation enzymes (ODEs) play important roles in odorant signal transduction pathway (Leal, 2013). ORs located in the dendrite membrane of olfactory sensory neurons (OSNs) and are considered to play a central role in identifying the distinct odorants and activating the OSNs (Clyne et al., 1999; Hallem et al., 2004). Compared with mammal ORs, insect ORs have seven transmembrane domains (TMDs) but employ a "reversed" topology with their N-terminus inside the cell and the C-terminus exposed to the external environment (Benton et al., 2006; Lundin et al., 2007; Hull et al., 2012). To detect the odorants, ORs could interact with a conserved olfactory receptor co-receptor (Orco) and then form ligand-gated ion channels (Sato et al., 2008; Wicher et al., 2008).

Orco is previously referred to as OR83b in Drosophila melanogaster, OR2 in Bombyx mori, and OR7 in mosquito species (Vosshall and Hansson, 2011). Conventional ORs demonstrate low sequence identity, whereas Orco is strikingly well conserved across insect species. It was reported that Orco has no direct relation with odor binding or discrimination (Nichols and Luetje, 2010; Nichols et al., 2011), but is essential for ion channel formation and olfactory cues transduction. In fact, Orco could interact with conventional ORs to form heterodimeric complexes, whereas conventional ORs were responsible for specifically binding to structurally diverse odorants (Larsson et al., 2004; Benton et al., 2006). Also, Orco was confirmed to be activated by VUAA1 as a functional ion channel in homomeric complex, even in the absence of conventional olfactory receptors (Jones et al., 2011). However, VU0183254, one of the analogs of VUAA1, showed the ability to "lock" hemomeric and homomeric ion channels in a non-competitive way due to its affinity to Orco (Jones et al., 2012). Coincidentally, these functional hemomeric and homomeric channels can be also blocked by amiloride derivatives when they were activated (Pask et al., 2013). Disruption in the transcript expression of Orco could significantly impair olfactory behavior responses in all the tested insect species, including D. melanogaster (Larsson et al., 2004), Acyrthosiphon pisum (Zhang et al., 2017), Locust amigratoria (Li et al., 2016), Spodoptera litura (Dong et al., 2013), Lymantria dispar (Lin et al., 2015), Aedes aegypti (DeGennaro et al., 2013), Microplitis mediator (Li et al., 2012), A. lucorum (Zhou et al., 2014), and Bactrocera dorsalis (Zheng et al., 2012). Due to the crucial role in olfactory perception, Orco is known as an excellent target for investigating co-evolution across sibling insect species (Lu et al., 2009).

The plant mirid species of Lygus spp., Adelphocoris spp., and other species strongly rely on olfactory cues to regulate their chemical perception behaviors. Series of studies on chemoreception of plant mirids were reported such as antennal morphological and electrophysiological characteristic (Chinta et al., 1997; Sun et al., 2014a), putative odorants (Koczor et al., 2012; Sun et al., 2013), physiological functions of OBPs (Gu et al., 2011; Hull et al., 2014; Sun et al., 2014b) and conventional ORs (Yan et al., 2015; An et al., 2016; Xiao et al., 2016; Zhang et al., 2016). In the current study, we focused on the evolutionary divergence of Orco orthologs among plant bug species from distinct geographic regions of China. Three Orco genes from A. suturalis, A. fasciaticollis and L. pratensis are were newly identified. Gene structures, substitution rates and tissuesbiased expression of Orco orthologs from five bug species were investigated to further figure out the evolutionary divergence in different mirid bugs.

## MATERIALS AND METHODS

#### Insect Collection and Rearing

Five mirid bug species including A. lucorum, L. pratensis, A. lineolatus, A. suturalis and A. fasciaticollis were collected from cotton fields at Langfang (Latitude 39.53◦N, Longitude 116.70◦E) or Kuerle (Latitude 41.45◦N, Longitude 85.48◦E) experimental station of the Chinese Academy of Agricultural Sciences. The laboratory colony was kept in 20 × 10 × 6 cm rearing containers and was reared on green beans (Phaseolus vulgaris L.) and a 10% sucrose solution (Lu et al., 2008b). Green beans also served as the oviposition substrate and were changed every other day. Beans containing eggs were subsequently placed in rearing containers lined with filter paper. After the emergence of the nymphs, the individuals were transferred to identical containers that were covered with nylon organdy mesh to allow air circulation. The

nymphs were provided with fresh food every 2 d until the emergence of adults. Each container housed approximately 100 nymphs or 60 adults. The laboratory colony was maintained at 29 ± 1 ◦C, 60 ± 5% relative humidity (RH), and 14 h:10 h light: dark (L: D) photoperiod.

#### RNA Extraction and cDNA Synthesis

Antennae from newly eclosion adults were excised and immediately frozen in liquid nitrogen, then stored at −80◦C until use. Total RNA was isolated by Trizol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer's instructions. The RNA quantity and integrity were checked using 1.2% agarose gel electrophoresis and a NanoDrop 2000 spectrophotometer (NanoDrop, Wilmington, DE, USA). Total RNA was treated with RQ1 RNase-Free DNase (Promega, Madison, USA) at 37◦C for 30 min to remove residual DNA. The cDNAs were synthesized using the Superscript III Reverse Transcriptase system (Invitrogen, Carlsbad, CA).

#### Gene Cloning and Sequence Analysis

AsutOrco, AfasOrco, and LpraOrco genes were cloned using degenerate primers (**Table S1**). Each reaction contained 300 ng antennal cDNA and 0.5 units of Ex Taq DNA Polymerase (TaKaRa, Dalian, China). The cycling parameters were: 95◦C for 2 min followed by 35 cycles at 94◦C for 30 s, 55◦C for 30 s, 72◦C for 60 s, and final extension at 72◦C for 10 min. The PCR product was gel-purified and sub-cloned into the pEASY-T3 vector (TransGen, Beijing, China) and then sequencing validation was performed. The 5′ and 3′ regions of Orco genes were amplified using SMARTerTM RACE cDNA amplification kit (Clontech, Mountain View, CA, USA) using gene-specific primers (GSP) (**Table S1**). Touchdown PCR was performed as follows: 95◦C for 2 min followed by 5 cycles at 94◦C for 30 s, 72◦C for 2 min; 5 cycles at 94◦C for 30 s, 70◦C for 30 s, and 72◦C for 90 s, 30 cycles at 94◦C for 30 s, 68◦C for 30 s, and 72◦C for 90 s; and a final 10 min incubation at 72◦C. The RACE PCR products were sub-cloned into the pEASY-T3 vector (Transgene, Beijing, China) and then sequenced. The full-length Orco genes were confirmed with LA Taq DNA polymerase (Takara, Dalian, China) by PCR using gene-specific primers (**Table S1**).

The full length Orco sequences were aligned by ClustalX 2.1 and edited by GeneDoc 2.7.0 software. TOPCONS (http:// topcons.cbr.su.se/) (Tsirigos et al., 2015) was used to identify the number and location of predicted transmembrane domains. The topology diagrams were constructed using TOPO2 Transmembrane Protein Display by the server at http://www. sacs.ucsf.edu/TOPO2/ (SJ)<sup>1</sup> .

## Gene Structure and Selective Pressure Analysis

Genomic DNAs from antennae were extracted using TIANamp genomic DNA kit (TIANGEN, Beijing, China) followed the manufacturer's instruction. Introns of Orco genes were amplified using specific primers (**Tables S2**–**S4**).The neighbor joining tree of Orco gene from various insect species were constructed using MEGA7.0 program with a p-distance model and a pairwise deletion of gaps. Bootstrapping was performed by the re-sampling amino acid positions of 1000 replicates, the synonymous and non-synonymous divergence was analyzed using modified Nei-Gojobori (Jukes-Cantor) (assumed transition/transversion bias = 1.21) method in MEGA 7.0 (Jukes and Cantor, 1969; Zhang et al., 1998; Kumar et al., 2016).

<sup>1</sup> S.J., J., TOPO2, Transmembrane protein display software. http://www.sacs.ucsf. edu/cgi-bin/open-topo2.py.

## Quantitative Real-Time PCR (qPCR) Measurement

The expressions profiles of Orco gene in different tissues of both genders were evaluated by using qPCR measurement on an ABI Prism 7,500 Fast Detection System (Applied Biosystems, Carlsbad, CA, USA).The reference genes β-actin (GenBank accession number: GQ477013, KU230353, KF921006, KU188517, and MG397129, separately) were used as the endogenous control to normalize the target gene expression and correct for any sample-to-sample variation. The primers (**Table S5**) of the target and reference genes were designed by BEACON DESIGNER 7 (PREMIER Biosoft International). The specificity of each primer set was validated by melt-curve analysis, and the efficiency was calculated by analyzing standard curves with a five-fold cDNA dilution series. Each qPCR reaction was conducted in 20 µL mixture containing 10 µL of 2 × Super-Real PreMix Plus (TIANGEN, Beijing, China), 0.6µL of each primer (10µM), 0.4 µL of 50 × Rox Reference Dye, 1 µL of sample cDNA and 7.4 µL of sterilized H2O. The qPCR cycling parameters consisted of 95◦C for 15 min, followed by 40 cycles of 95◦C for 10 s and 62◦C for 30 s, and melt curve stages at 95◦C for 15 s, 60◦C for 1 min, and 95◦C for 15 s. The experiments for the test samples, endogenous control and negative control were performed in triplicate to ensure reproducibility. The comparative 2−11CT method was used to calculate the relative transcript levels in each tissue samples (Livak and Schmittgen, 2001). All of the data were normalized to endogenous β-actin levels from the same tissue samples.

## RESULTS AND DISCUSSION

#### Cloning and Sequence Analysis of Orcos

Among the five plant bug species, two Orcos, AlinOrco from A. lineolatus and AlucOrco from A. lucorum were identified in our previous work (Zhou et al., 2014; Xiao et al., 2016). Here, we focused on other three Orco genes from A. suturalis, A. fasciaticollis, and L. pratensis. The rest three Orco genes were obtained by homology-based cloning (Hull et al. 2012) using degenerate primers (**Table S1**). A 400 bp fragment encoding putative Orco was amplified from A. fasciaticollis, A. suturalis,

and L. pratensis, respectively. The remaining 5′ and 3′ end sequences were further obtained using RACE PCR using gene specific primers. Finally, three full length sequences encoded AfasOrco, AsutOrco, and LpraOrco were assembled and deposited in GenBank with the accession numbers MF153393, MF153394, and MF153395, separately. The open reading frames (ORFs) of AsutOrco, AfasOrco, and LpraOrco were 1416, 1416, and 1422 bp, respectively, which resembled the full length of previously reported Orco genes (Hull et al., 2012; An et al., 2016; Xiao et al., 2016).

Results of sequence alignment indicated that all five Orcos including AfasOrco, AsutOrco, LpraOrco, AlinOrco and AlucOrco were rather conserved across the species (**Figure 1**). The amino acid identity among species of genus Adelphocoris and even across the genera of Adelphocoris, Lygus, and Apolygus was up to 99.6 and 96.8 %, respectively (**Table S6**). Unlike highly conventional ORs (Clyne et al., 1999; Gao and Chess, 1999), alignment of 200 Orco amino acid sequences (**Table S7**) from 8 orders showed a 62.6% identity (data not shown). These findings coincide with the previous point of view that Orco is highly conserved (Krieger et al., 2003; Melo et al., 2004; Briguad et al., 2009; Zhao et al., 2013).

Generally, different regions in the gene may play different roles. A predicted algorithm based on TOPCONS revealed these five Orco shared a similar atypical seven trans-membrane topology with their N-terminus inside the cell and the C-terminus exposed to the external environment (**Figure 1** and **Figure S1**). Consequently, the full Orco sequences can be divided into 15 regions, including the intracellular N terminal region, the seven transmembrane regions, the three intracellular loops, the three extracellular loops, and the C terminal region. These data were also consistent with the previous reports (Carraher et al., 2012; Missbach et al., 2014). The amino acid variation among different regions was significantly different with the highest variable level observed at transmembrane regions TM3 and intracellular loop 2 (IL2) that could be involved in ligands binding (Chao et al., 1999; Capendeguy et al., 2006). While no variation was found at intracellular loop 3 (IL3), TM7 and C terminus (**Figures S1**, **S2**). It was reported that IL3 participates in the channel activation interaction between conventional ORs and Orco in D. melanogaster and (Benton et al., 2006; Turner et al., 2014). As a key residue, the conserved aspartic acid in TM7 could influence the responses of Orco hemomeric and homomeric ion channels to agonist VUAA1 and odors (Kumar et al., 2013).

#### Gene Structures of Orcos From Five Bug Species

Introns in Orco genes from different bug species were distinct and sequence identity of at the same position across five bug


species was extremely low (about 41 %) in comparison to rather conserved amino acids (data not shown) (**Figure 2**). Orco within genus Adelphocoris shared similar seven exons, six introns and their insertion loci suggesting the most closely relationships among the three bug species. AlucOrco also had seven exons and six introns, but the length of each intron was significantly larger than that of corresponding introns from genus Adelphocoris. Moreover, the insertion sites of third and fourth introns were also different from Orco in genus Adelphocoris (**Figures 2A,B**). Notably, only six exons and five

Leu245. Generally, the more intron number and larger intron length indicate a higher phylogenetic level (Nixon et al., 2002; Koonin, 2006; Wu et al., 2013; Park et al., 2014). Adult A. lucorum displays the most extensive distribution in China, whereas L. pratensis mainly occurred in Xinjiang Uygur Autonomous Region (Jiang et al., 2015). The host range is consistence with phylogenetic level among the five mirid bug species; A. lucorum has the widest host range including 54 families, however, L. pratensis merely owns 21 families (Jiang et al., 2015). Additionally, adult A. lucorum prefers to track better host plant food during different seasons than that of other four bug species (Pan et al., 2015; Wang et al., 2017). Likewise, olfaction especially the OR family is believed to play essential roles in the host selection for mirid bugs plant (Yan et al., 2015; Zhang et al., 2016). Therefore, our analyses indicate there might be a potential association between Orco evolution rate and the ecological adaption among these five mirid species, which could contribute to clarify the molecular mechanisms of evolutionary interactions between mirid bugs and their host plants. However, this speculation still needs to be proved by more evidences.

introns were found in LpraOrco gene, the last intron of which was lost and the third intron was located between Glu244 and

#### Evolution Analysis of Orco Orthologs

There was a clear conserved orthologous relationship among AfasOrco, AsutOrco, LpraOrco, and other four bug Orcos (AlucOrco, AlinOrco, LlinOrco, LhesOrco). Phylogenetic relationship was largely consistent with the species tree constructed from the alignment of species-specific cytochrome oxidase subunit I (COI) (**Figures 3A,B**). So, we suggested that Orco was significantly conserved and could function as a molecular marker of evolution across bug species. Also relatedness analysis of these seven Orcos to the other 193 Orco sequences from eight insect orders indicated that Orco was highly conserved within insect order. Orco sequences of the same order were strictly clustered together with strong bootstrapping support (**Figure 3C**), indicating this phylogenetic clade was highly conserved and may fulfill conserved function.

The ratios of non-synonymous to synonymous substitutions estimated for 14 Orco genes from 5 orders were listed in **Table 1**. All the ratios were far less than 1.0 indicating that Orco genes are under strong purifying selection pressure. The strong purifying selection pressure suggested a functional conservation, which had been proven by substantial documents. The lack of Orco leading to a similar reduction of olfaction

indicates the consistent roles in odor perceptions, suggesting the interspecific conservation of Orco indirectly (Zhou et al., 2014; Liu et al., 2017; Trible et al., 2017). Furthermore, the interspecific functional conservation has been confirmed directly by transgenic rescue experiment. The defects of olfaction in DmelOrco mutant could be rescued by transgenic expression of DmelOrco, CcapOrco, AgamOrco and HzeaOrco, respectively (Jones et al., 2005). It was also demonstrated that Orco, as an obligatory part of ligand-gated ion channel, played conservative functions in ligand binding, and was activated by the agonist VUAA1 dutifully (Benton et al., 2006; Sato et al., 2008; Jones et al., 2011).

Orcos are under strongly purifying selection pressure and exhibit potential conserved olfaction roles. However, our estimation on the ratios of non-synonymous to synonymous substitutions (Ka/Ks) revealed that their levels of the purifying selection pressure significantly varied in the genera and species. Generally, the values of Ka/Ks were similar among species within same genus, but were different from species across genera. As shown in **Table 1**, when used DmelOrco, AgamOrco or Orco genes from other model species as outgroup, the range of Ka/Ks values of Orco genes from three Adelphocoris species were evaluated as (0.288–0.454), which was significantly different to that of AlucOrco (0.365–0.496) from Apolygus genus, or LpraOrco, LlinOrco and LhesOrco (0.361–0.497) from Lygus genus. These findings indicted there might be a strong constraint on functional variation within Orco from same genus, as illustrated above. In addition, these results were faultlessly correlated to the phylogenetic analyses (**Figure 3**). Three Orco genes from Adelphocoris species fall into the same clade, AlucOrco and three Orco from Lygus species clustered in another clade. Because of the evolutionary synchronization between Orco genes and their mirid species (**Figures 3A,B**), we proposed that the degrees of variation (suggested by Ka/Ks values) on Orco protein coding regions could reflected the phylogenetic levels of mirid bug species, and our data would lay a foundation on the further studies on the molecular mechanisms of speciation of mirid bugs.

#### Expression Profiles of Five Orcos

In general, target gene with different tissue expressions would play different physiological function. To figure out the potential roles of Orco in mirid bugs species, qPCR measurement was conducted to assess their tissue-specific expressions (**Figure 4**). The results demonstrated that these five Orco genes share similar antennae-biased expression profiles, which were similar to that in L. hesperus (Hull et al., 2012). So, we suspected that Orco in different mirid bugs could be associated with clear olfactory roles. It was reported that silencing in A. lucorum of the olfactory co-receptor Orco gene by RNA interference could induce EAG response declining to two putative semiochemicals (Zhou et al., 2014). However, some Orco could be also expressed in non-olfactory organs such as proboscis and legs, suggesting that Orco might be involved in the contact chemosensory perception and could help to search hosts in close distance and perceive the status of hosts (Lu et al., 2009; Hull et al., 2012). In this study, faint transcript levels of these five Orcos were detected in stylets, legs, head and other non-olfactory organs (**Figure 4**) suggesting the potential roles of Orco in taste recognition of bugs. Besides in mirid bug species, Orco of B. dorsalis could fulfill a role involved in the perception

of Rhodojaponin-III, a non-volatile compound (Yi et al., 2014).

#### AUTHOR CONTRIBUTIONS

Y-JZ and LS conceived and designed the experimental plan. QiW, Y-LZ, and H-HC performed the experiments. QiW, QianW, SS, YX, KD, and AK analyzed the data. QiW and QianW drafted the manuscript. LS and Y-JZ refined and approved the final manuscript.

#### ACKNOWLEDGMENTS

Dr. Xianhui Wang (Institute of Zoology, Chinese Academy of Sciences, Beijing, China) and Dr. Bo Zhang (Institute of Plant Protection, Chinese Academy of Agricultural Sciences) kindly helped analyse the data on evolution. This work was supported by the China National Basic Research Program (2012CB114104), the National Natural Science Foundation of China (31471778, 31501652,31621064 and 31772176), Central public-interest Scientific Institution Basal Research Fund (No. 1610212016015), Research Foundation of State Key Laboratory for Biology of Plant Diseases and Insect Pests (SKLOF201514 and SKLOF201719), and The Science and Technology Innovation

#### REFERENCES


Project of the Chinese Academy of Agricultural Sciences (CAAS-ASTIP-2015-TRICAAS).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00158/full#supplementary-material

Figure S1 | Predicted transmembrane topologies of Orco with variable sites highlighted in red. The double line indicates the membrane region with extracellular and cytoplasmic sides labeled.

Figure S2 | Ratio of the relative amino acid differences per domain averaged for Orco.

Table S1 | Primers used in identification of Orco genes from mirid bug *A. suturalis*, *A. fasciaticollis,* and *L. pratensis*.

Table S2 | Primers used in amplification of AlucOrco gene introns.

Table S3 | Primers used in amplification of AlinOrco, AfasOrco and AsutOrco gene introns.

Table S4 | Primers used in amplification of LpraOrco gene introns.

Table S5 | Primers used in qPCR measurement.

Table S6 | Identities of amino acid sequences among Orco genes from five mirid bugs.

Table S7 | Orcos used in phyologenetic construction and sequence analysis.

for humans and are not repelled by volatile DEET. Nature 498, 487–491. doi: 10.1038/nature12206


highly conserved and expressed in olfactory and gustatory organs. Chem. Senses 29, 403–410. doi: 10.1093/chemse/bjh041


to some plant odors. Int. J. Mol. Sci. 17:1165. doi: 10.3390/ijms170 81165


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wang, Wang, Zhou, Shan, Cui, Xiao, Dong, Khashaveh, Sun and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# BdorOR88a Modulates the Responsiveness to Methyl Eugenol in Mature Males of Bactrocera dorsalis (Hendel)

Huan Liu, Zheng-Shi Chen, Dong-Ju Zhang and Yong-Yue Lu\*

Department of Entomology, South China Agricultural University, Guangzhou, China

#### Edited by:

Peng He, Guizhou University, China

#### Reviewed by:

William Benjamin Walker III, Swedish University of Agricultural Sciences, Sweden Feng Liu, Michigan State University, United States Dong Wei, Southwest University, China Ivan Manzini, Justus-Liebig-Universität Gießen, Germany

#### \*Correspondence:

Yong-Yue Lu luyongyue@scau.edu.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 28 February 2018 Accepted: 05 July 2018 Published: 26 July 2018

#### Citation:

Liu H, Chen Z-S, Zhang D-J and Lu Y-Y (2018) BdorOR88a Modulates the Responsiveness to Methyl Eugenol in Mature Males of Bactrocera dorsalis (Hendel). Front. Physiol. 9:987. doi: 10.3389/fphys.2018.00987 Insect attractants are important prevention tools for managing populations of the Oriental fruit fly, Bactrocera dorsalis (Hendel), which is a highly destructive agricultural pest with health implications in tropical and subtropical countries. Methyl eugenol (ME) is still considered the gold standard of B. dorsalis attractants. Mature male flies use their olfactory system to detect ME, but the molecular mechanism underlying their olfactory detection of ME largely remains unclear. Here, we showed that ME activates the odorant receptors OR63a-1 and OR88a in mature B. dorsalis males antennae by RNA-Seq and qRT-PCR analysis. Interestingly, ME only elicited robust responses in the BdorOR88a/BdorOrco-expressing Xenopus oocytes, thus suggesting that BdorOR88a is necessary for ME reception and tropism in B. dorsalis. Next, our indoor behavioral assays demonstrated that BdorOR63a-1 knockdown had no significant effects on ME detection and tropism. By contrast, reducing the BdorOR88a transcript levels led to a significant decrease in the males' responsiveness to ME. Taken together, our results gave novel insight in the understanding of the olfactory background to the Oriental fruit fly's attraction toward ME.

Keywords: Bactrocera dorsalis, methyl eugenol, transcriptomic analysis, olfactory, odorant receptor, Xenopus oocytes

## INTRODUCTION

Insects rely primarily on sophisticated olfactory reception systems to detect and discriminate many exogenous chemical signals, and odorant receptors (ORs) are at the core of odorant detection (Hallem and Carlson, 2006; Missbach et al., 2014; Zhang et al., 2016; Fleischer et al., 2017; Wang et al., 2017). Apparently, ORs show a high degree of sensitive and play critical roles in detecting long-range volatile odorants and triggering the transduction of chemical signals into electric signals (Mitsuno et al., 2008; Leal, 2013). For example, an odorant receptor isolated from Plutella xylostella, PxylOR1, was narrowly tuned to the main component of the sex pheromone, (11Z) hexadecenal (Z11-16Ald) (Sakurai et al., 2011). Interestingly, the direct activation of CquiOR136, is necessary for DEET (N,N-diethyl-3-methylbenzamide) reception and repellency effects in Culex quinquefasciatus (Xu et al., 2014). Additionally, knockdown of AjapOR35 in Anastatus japonicus reduced its antennal response to two oviposition attractants, β-caryophyllene and (E)-α-farnesene (Wang et al., 2017). In this respect, using a "computational reverse chemical ecology" strategy will likely generate new insights for the rapid screening of potentially effective semiochemicals that modify the behavioral patterns of insects (Siderhurst and Jang, 2006; Wu et al., 2015).

The oriental fruit fly, Bactrocera dorsalis (Hendel), is among the most destructive fruit/vegetable-eating agricultural pests in the world (Zheng et al., 2013; Liu et al., 2016). Due to polyphagia in the larval stages and high fecundity of the adults, B. dorsalis can cause serious damage to more than 250 species of commercially grown vegetables and fruits (Clarke et al., 2005; Stephens et al., 2007; Shen et al., 2012). Plant damage caused by B. dorsalis consists of oviposition stings to host fruit tissues by adult females and the subsequent larval feeding and decaying in the fruit pulp. Controlling the male adult fly's behavior is the main method to reduce the damage caused by this pest (Clarke et al., 2005). Methyl eugenol (ME), a highly potent phytochemical lure, has been exploited in the male annihilation technique (MAT) systems for detecting, monitoring, and luring B. dorsalis male individuals (Vargas and Prokopy, 2006; Pagadala et al., 2012; Shelly, 2017). Particularly noteworthy is the fact that ME has its risks, however: it is carcinogenic to humans and its attractiveness is limited to mature males (Smith et al., 2002; Khrimian et al., 2009; Zheng et al., 2012). Although novel attractants of B. dorsalis are diligently being developed, their progress toward better and more affordable attractants has nonetheless been slow (Khrimian et al., 1994, 2006, 2009; Jang et al., 2011). Design of attractants to target specific ORs may promote the development of new baits for pest management (Di et al., 2017; Mitchell et al., 2017). However, ME receptors ORs in B. dorsalis are hitherto unknown. Therefore, exploring the molecular mechanism underlying olfactory detection of ME in B. dorsalis is of great importance for developing sustainable pest control strategies based on manipulating insect chemosensory communication.

In trying to identify the chemosensory genes responsible for detecting ME, recent studies have identified and confirmed that BdorOrco, BdorOBP83a-2, and BdorOBP2 actively participate in the process of ME detection by B. dorsalis male adults (Zheng et al., 2012; Wu et al., 2016; Liu H. et al., 2017). Nevertheless, the specific ORs and their molecular functional involvement in the mature male fly's reception of ME remain a mystery. Here, we used an RNA-Seq approach coupled to RNA interference silencing, supplemented by in vivo expression in Xenopus laevis oocytes, to investigate the functional roles of BdorOR63a-1 and BdorOR88a; both genes are abundantly expressed in MEtreated male antennae and were hypothesized to encode possible receptors for ME. Based on the results presented here, we infer that BdorOR88a likely plays an essential role in the molecular mechanism underlying B. dorsalis olfactory detection of ME.

## MATERIALS AND METHODS

#### Ethics Statement

No specific permits were required in our studies of this widespread agriculture pest. We confirm that the study locations were not privately owned or protected. This work did not involve endangered or protected species. To avoid chemical hazards, always observe safety laboratory practice when operating the ME.

#### Insect Rearing

The B. dorsalis genetic sexing strain (GSS) used in this study was reared in a laboratory for more than 30 generations at the South China Agricultural University, Guangzhou, China. The male pupae are brown and the female pupae are white. Insects were reared under a photoperiod cycle of 14 h light/10 h dark at 27 ± 1 ◦C, 75 ± 1% relative humidity (RH). Larvae were reared on an artificial diet that included yeast (15.06%), sugar (8.99%), nipagen (0.15%), sodium benzoate (0.15%), citric acid (1.70%), wheat germ oil (0.15%), and water (73.81%) (Chang et al., 2006; Liu H. et al., 2017). Adult flies were maintained in 35 cm × 35 cm × 35 cm wooden cages and fed a diet consisting of sugar: yeast extract at 1:1 (w/w) (Liu H. et al., 2017).

#### Treatments and Samplings

The ME solution used consisted of 1:1 dilution with mineral oil (MO) (Energy Chemical Company, Shanghai, China), of which 0.5-ml amounts were used to coat the inner wall of 500-ml conical flasks. Next, sample of 200 mature males (15 day-old) were randomly selected and placed in each flask. Flies in the control group were likewise handled but put into flasks containing an equal volume of MO only. After being treated for 1 h, all the antennae were dissected and flash frozen in liquid nitrogen, and were then immediately transferred to a −80◦C freezer pending RNA extraction. Three independent biological replicates were performed for use in the transcriptome analyses. Experiments were conducted between 9 and 11 a.m. During these experiments, RH and temperature in the laboratory were maintained at 75 ± 1% RH and 27 ± 1 ◦C, respectively.

#### RNA Preparation, Library Construction, and Transcriptome Sequencing

Under an RNA-free environment, antennal total RNAs were extracted by using a RNA extraction kit (Takara Biotechnology Co., Ltd., Japan), following the manufacturers protocol. Each sample consisted of 200 males' antennae. The purity of all RNA samples was assessed at absorbance ratios of OD260/<sup>230</sup> and OD260/280, while the integrity of RNA was verified through 1%-RNase-free agarose gel electrophoresis. The concentration of RNA was quantified by measuring their absorbance at 260 nm in a spectrophotometer (Thermo Nano DropTM 2000c; Santa Clara, CA, United States) and qualified using an Agilent 2,100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, United States). High-quality RNA from each replication of the ME treatment and control groups was used for the next suite of steps: mRNA was first isolated from total RNA using magnetic beads with oligo (dT) and sheared into short fragments in a fragmentation buffer. Then complementary DNA (cDNA) was synthesized from the mRNA fragments, by using SuperScript III Reverse Transcriptase (Invitrogen). A cDNA library was constructed for each sample. The library preparations were sequenced on an Illumina HiSeq 4000TM platform (Illumina Inc., San Diego, CA, United States) with 125-nucleotide (nt) paired-end reads; this final step carried out by the Gene Denovo Technology Company (in Guangzhou, China).

#### Sequence de Novo Assembly and Functional Annotation of Unigenes

For quality control, the raw sequencing data in the FASTQ format were processed by in-house Perl scripts to obtain clean data reads. Before assembly, any adapter sequences were removed from the raw reads. Short or low-quality reads – those reads containing an adaptor, reads containing >5% unknown nt "N," and reads with >20% quality value ≤10 – were removed from raw data to obtain more reliable results. The Q20, Q30, and GC contents of the cleaned data were calculated. Next, the clean reads were de novo-assembled by using Trinity software (version v2013- 02-25) (Trinity Software, Inc., Plymouth, MA, United States) and clustered with TGICL Clustering tools (Version 2.1) (The Institute for Genomic Research, Rockville, MD, United States) (Pertea et al., 2003; Grabherr et al., 2011). Functional annotation of these assembled unigenes was performed with BLASTx<sup>1</sup> , an integrated Gene Ontology (GO) annotation and data mining tool that assigns GO based on four publically available databases: the National Center for Biotechnology Information (NCBI) nonredundant protein database (Nr), the Kyoto Encyclopedia of Genes and Genomes (KEGG<sup>2</sup> ), the Eukaryotic Ortholog Groups (KOG<sup>3</sup> ), and the Swiss-Prot protein database<sup>4</sup> . All searches were performed with an E-value < 10−<sup>5</sup> . We used the Blast2GO<sup>5</sup> program to do the GO functional classification for all unigenes, according to their molecular function, biological process, and cellular component (Ashburner et al., 2000; Conesa et al., 2005).

#### Differential Expression Analysis of Genes

The sequenced reads for each sample were remapped onto the reference sequences with RSEM software v1.2.12 (Li et al., 2011; Zhang et al., 2016). Gene expression levels were estimated using the Fragments Per Kilobase of transcript per Million fragments (FPKM) method that is based on the number of uniquely mapped reads (Trapnell et al., 2010). Genes differentially expressed between the male antennae from ME treatment and MO groups were identified based on their FPKM values (Mortazavi et al., 2008; Anders and Huber, 2010). The false discovery rate (FDR) adjustment was made to correct the P-value threshold in these multiple tests and analyses (Benjamini and Hochberg, 1995). An FDR-adjusted P-value < 0.05 and an absolute value of the log<sup>2</sup> ratio > 1 were set a priori as the significance threshold for gene differential expression in this study. For convenience, the differential expression genes showing higher expression levels in the ME than in the MO group were designated as "up-regulated," whereas those displaying lower expression levels were designated as "down-regulated."

#### Functional Analysis of Differentially Expressed Genes (DEGs)

Differentially expressed genes were also annotated using the GO database, and the numbers of DEGs in each GO term were calculated. To determine, which GO terms were significantly enriched in the DEGs, we performed a GO enrichment analysis that used a hypergeometric test to map all differentially expressed proteins to the GO terms in the database. This test used the following equation (Blüthgen et al., 2005; Tu et al., 2015):

$$P = 1 - \sum\_{i=0}^{\mathfrak{m}-1} \frac{\binom{\mathfrak{M}}{i} \binom{N-M}{n-i}}{\binom{N}{n}}$$

where, N is the number of all genes with a GO annotation; n is the number of DEGs in N; M is the number of all genes annotated to specific GO terms; and m is the number of DEGs in M (M– m ≥ 0). The calculated P-value was first subjected to a Bonferroni correction, taking a corrected P-value of 0.05 as a threshold for statistical significance. GO terms fulfilling this condition were defined as significantly enriched GO terms in the DEGs.

All identified genes were mapped to pathways in the KEGG database by using the KEGG Automatic Annotation Server software. To identify significantly enriched metabolic pathways or signal transduction pathways in DEGs, we used the same formula calculation as in the GO analysis. Here, N represented the number of all genes with a KEGG annotation, n is the number of DEGs in N, M is the number of all genes annotated to specific pathways, and m is the number of DEGs in M.

#### Gene Expression Validation by Quantitative Real-Time PCR (qRT-PCR)

To verify our RNA-Seq results, 16 genes related to insect olfactory transport process that showed different expression levels, as revealed via RNA sequencing, were randomly selected for validation in a qRT-PCR analysis. In addition, other independent sampling experiments were conducted in order to obtain new biological replicates by ME and MO treatment. Those experiments were performed as described above. A RNA extraction kit (Takara Biotechnology Co., Ltd., Japan) was used to extract antennal total RNA from the ME treatment and control groups of male flies according to the manufacturer's protocol, and a gDNA eliminator spin column removed genomic DNA. Approximately 1 µg of total RNA from each sample was used to synthesize cDNA, by using a PrimeScriptTM RT Reagent Kit (Takara Biotechnology Co., Ltd., Japan), which then served as a template for qRT-PCR. The gene-specific primers were designed according to the gene sequences in Primer v5.0 software (Premier, Canada) were listed in Supplementary Table S1. RT-PCR was performed to test whether all primers could amplify the correct products. Amplification efficiencies of all primers were validated before the gene expression analysis.

To perform the qRT-PCR reactions, a SYBR Premix ExTaq Kit (Tiangen, Guangzhou, China) was used following the manufacturer's instructions, and run on a Stratagene Mx3000P thermal cycler (Agilent Technologies, Wilmington, Germany). qRT-PCR was carried out according to the protocol reported in our previous study (Liu H. et al., 2017). The α-tubulin gene of B. dorsalis was amplified to serve as the internal control (GenBank accession number: XM\_011212814) (Shen et al., 2010; Yang et al., 2013; Liu et al., 2015; Gui et al., 2016). Dissociation

<sup>1</sup>http://www.geneontology.org

<sup>2</sup>http://www.genome.jp/kegg/

<sup>3</sup>http://www.ncbi.nlm.nih.gov/COG/

<sup>4</sup>http://www.ebi.ac.uk/uniprot/

<sup>5</sup>http://www.blast2go.com/

curve analyses were performed to ensure amplification specificity. Three independent biological and three technical replicates were used and performed for each gene, respectively. The relative gene expression levels were calculated by using the 2−11C<sup>T</sup> method as described previously (refer to Livak and Schmittgen, 2001).

#### Sequence Analysis of BdorOR63a-1 and BdorOR88a

To identify the characteristics of B. dorsalis OR genes, particularly the significant differently expression ORs (BdorOR63a-1, GenBank accession number: KP743726 and BdorOR88a, GenBank accession number: KP743732), and their relationship to other Dipteran insects. To do this, a maximum likelihood tree for the ORs was constructed using the amino acid sequences derived from B. dorsalis and the published sequences of two Dipteran species: Ceratitis capitata (Wiedemann) and Drosophila melanogaster (Meigen). All the information on the amino acid sequences of ORs was obtained from the NCBI database. Alignments of OR amino acid sequences were performed using the program ClustalW2. The maximum likelihood tree was constructed in MEGA v7.0 software (Molecular Evolutionary Genetics Analysis, v4.0, Sudhir Kumar, United States) and with the Interactive Tree Of Life (iTOL) web tool<sup>6</sup> .

#### Effects of Age and Daily-Rhythm on the Male Fly Responsiveness to ME and the BdorOR63a-1, BdorOR88a Expression Levels

Bioassays were performed in the laboratory following a method similar to that of Karunaratne and Karunaratne (2012) and Liu H. et al. (2017). For the assessment of effects of age on the male responses to ME, each sample of 2-day- and 10-day-old males (200 of each) were randomly selected as the test subjects and released into a screened cage (1.0 m × 1.0 m × 1.0 m) without a trap. Approximately 30 min later, a fly trap containing 1.0 mL of pure ME was placed inside the screened cage. For the control, a trap was likewise placed inside the cage but without any ME. After trapping for 2 h, we removed the traps and counted the number of attracted flies. This bioassay was conducted between 9 and 12 a.m. under daylight conditions. Three independent biological replicates were performed.

To determine whether the mature male response to ME varied throughout the day, the responses of 10-day-old male B. dorsalis were assayed at 9 a.m., 1 p.m., and 5 p.m., as described above. For this, three biological replicates of flies were used per ME treatment and per control group for each time point. Additionally, from the untested individuals, the antennae from the 2-day- and 10-day-old males, and the mature males at 9 a.m. (morning), 1 p.m. (early afternoon), and 5 p.m. (near dusk), were dissected and immediately frozen in liquid nitrogen. Then, their total RNAs were extracted and reverse transcribed into single-chain cDNAs. Next, the expression of BdorOR63a-1 and BdorOR88a was evaluated by qRT-PCR quantitative technique. Each treatment was replicated three times.

#### Expression of BdorOR88a and BdorOR63a-1 in Xenopus laevis Oocytes and Two-Electrode Voltage-Clamp Electrophysiological Recordings

BdorOR88a, BdorOR63a-1, and BdorOrco (GenBank accession number: EU621792) were amplified using specific primers (Supplementary Table S2). The purified PCR products were ligated into the pCS2<sup>+</sup> vector; then, linearized modified pCS2<sup>+</sup> vectors were used to synthesize cRNAs by using the mMESSAGE mMACHINE SP6 Kit (Ambion, Austin, TX, United States) following the manufacturer's instructions. The purified cRNAs were re-suspended in nuclease-free water at 2000 ng/µL.

The oocyte microinjection and two-electrode voltage clamp recording were performed following published protocols (see Xu et al., 2014; Li et al., 2017; Liu F. et al., 2017). Briefly, mature healthy X. laevis oocytes (stage V–VII) were treated with 2 mg/ml of collagenase I (GIBCO, Carlsbad, CA, United States) in a washing buffer (96 mM NaCl, 5 mM MgCl2, 2 mM KCl, and 5 mM HEPES [pH = 7.6]) for ca. 1 h at room temperature. Next, the oocytes were microinjected with 27.6 ng of BdorOR cRNAs and 27.6 ng of BdorOrco cRNA, and then incubated at 18◦C for 3–8 days in 1 × Ringer's solution (96 mM NaCl, 5 mM MgCl2, 5 mM HEPES, 2 mM KCl, and 0.8 mM CaCl2 [pH = 7.6]). Stock solutions (1 M) of ME and MO were prepared in DMSO and they were subsequently diluted to the indicated concentrations with 1 × Ringer's buffer. The two-electrode voltage-clamp (TEVC) technique measured the odorant-induced currents in Xenopus oocytes. Whole-cell currents were recorded and amplified by an OC-725C amplifier (Warner Instruments, Hamden, CT, United States) at a holding potential of −80 mV, low-passfiltered at 50 Hz, and digitized at 1 kHz. Oocytes with nuclease-free water injection served as the negative control. Data acquisition and analysis were conducted with Digidata 1440A and pCLAMP10 software (Axon Instruments Inc., Union City, CA, United States).

#### RNA Interference Bioassays

An RNA interference experiment was performed to demonstrate the roles of BdorOR88a and BdorOR63a-1 in ME detection by male flies. To prepare the double-stranded RNA (dsRNA), we used a template cDNA generated by PCR-targeting fragments. Those primers with T7 promoter sequences used to synthesize dsRNA are listed in Supplementary Table S3. The GFP gene served as the control dsRNA (dsGFP) (GenBank accession number: AHE38523). Based on the manufacturer's protocol, the dsRNAs were synthesized and purified by the MEGAscript <sup>R</sup> RNAi Kit (Thermo Fisher Scientific, United States); then, their concentrations were quantified on a Nanodrop 1,000 (Thermo Fisher Scientific, United States) and their integrity was determined by 2%-agarose gel electrophoresis.

Subsequently, expression of the dsRNAs and the injection procedure for the male flies were carried out following established techniques (Liu et al., 2015; Dong et al., 2016; Hou et al., 2017; Liu H. et al., 2017). Each sample of 50 mature males (15-day-old) was randomly selected and placed into a 35 cm × 35 cm × 35 cm

<sup>6</sup>http://itol.embl.de

cage. Needles were prepared with a puller at 60◦C (PC-10, Narishige, Tokyo Japan). Microinjection was performed using an Eppendorf Microinjector (Eppendorf Ltd., Germany). The injection condition was set to an injection pressure (Pi) of 600 hPa and a timing setting (Ti) of 0.6 s. For each treated male fly, 400 nL of dsBdorOR88a or dsBdorOR63a-1 (2,000 ng/µL) was injected into its body cavity between the second and third abdominal segments. Males were injected with an equal volume of dsGFP served as negative control groups. The blank control group consisted of male flies that were fed normally. Injected and non-injected flies were reared on the artificial diet in the cages as described above. The respective numbers of dead files were counted after treatment for 24 h and 48 h. For the bioassay, males from the dsBdorOR88a or dsBdorOR63a-1 treatment groups, the dsGFP treatment group, and the blank control group, were placed separately inside a 1.0 m × 1.0 m × 1.0 m screen cage equipped with ME as a trap. The lured males were counted after 2 h, and the silencing efficiency of BdorOR88a or BdorOR63a-1 was detected by qRT-PCR. Each bioassay experiment was replicated five times.

#### Statistical Analyses

All analyses were carried out by SAS v9.20 software (SAS Institute Inc. Cary, NC, United States). Results from the experimental replicates were expressed as the mean ± SE. The responses of immature and mature male flies to ME, diurnal pattern of mature male responsiveness to ME, and the expression pattern of ORs were analyzed by independent Student's t-test (P = 0.05). Cases of P-values < 0.05 were considered to be statistically significant. Datasets of the attractiveness of ME to mature male, adult mortality, and gene silencing efficiencies were checked for normality of distribution and homogeneity of variances with Shapiro-Wilk's and Levene's tests, respectively. If data were normally distributed and had similar variances, then the means were compared by one-way analysis of variance (ANOVA). Following a significant ANOVA result, multiple comparisons among five groups were assessed by Duncan's multiple range test (DMRT, P = 0.05). Non-normally distributed data were analyzed with the non-parametric Kruskal–Wallis test to compare medians (P = 0.05), followed by a Mann–Whitney test for follow-up pairwise comparisons. All results were plotted with Origin v9.0 software.

## RESULTS

#### Transcriptome Sequencing and Analysis

The RNA taken from the ME-treated and control (MO) males was used for RNA-Seq – with three experimental replicates per treatment – generated 23,552,399,343 raw reads. In general, all the libraries were of good quality, with average Q20 and Q30 percentages of over 95.60 and 89.41%, respectively. After removing the low-quality reads and trimming the adapter sequences, 157,514,454 clean reads were obtained for sequencing from the six samples (**Tables 1**, **2**). These clean reads were ultimately assembled into 36,215 unigenes that had a mean length of 1,147 bp, an N50 of 2,362 bp, and a GC content of 41.61% (**Table 2**). The transcriptome data were deposited into the NCBI Short Reads Archive (SRA) database under the accession number SRP124917.

#### Identification of Differentially Expressed Antennal Genes Between the ME and MO Treatment Male Flies

To determine the effects of ME exposure on antennal gene expression in male B. dorsalis adults, their DEGs were identified using the FPKM method. A total of 4,433 DEGs were detected from the ME treatment and MO (control) groups (|log2FC| > 1, P-value < 0.05; FDR < 0.05). Of these, 3,813 (86.01%) DEGs were up-regulated and 620 (13.99%) were down-regulated in the ME-treated male antennae (**Figure 1A**). The global expression changes of all genes with RNA-Seq ratios are shown in **Figure 1B**; the green and red circles are genes having a differential expression pattern in ME-treated males compared with the control (MO group). **Figure 1C** shows the hierarchical clustering analysis of their 4,433 DEGs.

#### GO and KEGG Pathway Enrichment Analysis of DEGs

With the GO annotation results in hand, the DEGs were classified and categorized into 83 functional groups (Supplementary Figure S1). Specifically, the GO enrichment analysis (Pvalue < 0.05) revealed that the most enriched biological process terms were cellular, followed by metabolic, singleorganism, biological regulation, and response to stimulus;

TABLE 1 | Statistical raw sequencing data of the RNA-Seq reads for the examined samples.


CK1, CK2, and CK3 are repeats of the mineral oil (control) group; T1, T2, and T3 are repeats of the methyl eugenol treatment. Q20: percentage of bases for which the Phred value is >20; Q30: percentage of bases for which the Phred value is >30.

TABLE 2 | Statistics for the resulting assembled sequences.


Unigene N50 is the length at which the accumulated length value is greater than 50% of the total lengths after ranking the unigenes from shortest to longest.

while binding and catalytic activity were the most enriched terms for the molecular function category; finally, under the cellular component category, cell and cell part were the most enriched terms (Supplementary Figure S2A). To investigate the biological pathways actively involved in ME detection by male flies, the DEGs were assigned to reference canonical pathways in KEGG: this revealed 20 significantly enriched pathways (Supplementary Figure S2B). The annotations presented here provide a valuable information for investigating the specific processes, pathways, and functions involved in the ME detection process in B. dorsalis, and perhaps other Diptera, too.

#### Identification of Key DEGs Potentially Involved in Olfactory Function

Many DEGs associated with response to stimulus, catalytic activity, binding, biological adhesion, molecular transducer activity, and transporter activity may contribute to ME detection in adult B. dorsalis males. Based on the literature and our GO/KEGG enrichment analyses, we identified putative DEGs encoding proteins involved in insect olfactory transport, viz: three DEGs for odorant binding proteins (BdorOBP57c, BdorOBP5, and BdorOBP2), two for ORs (BdorOR88a and BdorOR63a-1), one encoding an ionotropic receptor (BdorIR92a), and one encoding a sensory neuron membrane protein (BdorSNMP1-1) (**Figure 2**). Notably, relative to the control (MO) group, both BdorOR88a and BdorOR63a-1 were up-regulated by 6.33- and 2.06-fold, respectively, in the antennae of males exposed to ME. Additionally, the expression levels of carboxylesterase, esterase B1, cytochrome P450-6a14,- 313a, and UDP-glucuronosyltransferase 2A3, 3A1 – genes that encode candidate odorant-degrading enzymes (ODEs) – were significantly down-regulated at the transcriptional level.

FIGURE 1 | Statistical analysis of the differential expression of genes in male antennae of Bactrocera dorsalis flies from the MO (mineral oil) and ME (methyl eugenol) treatment groups. (A) Classification for the differential abundance of genes. (B) Volcano plots of differentially expressed genes (DEGs) between the ME and MO treatments. Y axis represents -log10 significance. X axis represents logFC (fold change). Red points indicate up-regulated expression of DEGs; green points indicate down-regulated expression of DEGs; and black points show non-differentially expressed genes. Genes with a P-value < 0.05 and fold change ≥1.0 were considered as DEGs in this study. (C) Hierarchical clustering graph of differential gene expression profiles in the ME and MO treatments. Clustering was done using RNA-Seq data derived from the six samples based on log10FPKM values. The column and row indicate the sample and gene, respectively. CK and T represent MO- and ME-treatment, respectively. Red and blue bands indicate, respectively, those genes that were significantly up-regulated and down-regulated in the ME-treated males.

respectively. Three biological replicates were conducted for each treatment.

## Validation of DEGs by qRT-PCR

We used qRT-PCR analysis to validate the results of differential gene expression obtained from the RNA-sequencing data (**Figure 3**). Two genes, BdorOBP57c and BdorOBP5 expressed no significant difference in qRT-PCR analysis, which were inconsistent with RNA-Seq results. However, of the 16 selected genes, 14 agreed with our RNA-Seq results. For example, BdorOR88a, BdorOR63a-1, BdorIR92a, and BdorSNMP1-1 were all significantly up-regulated in the ME-treated male antennae, as found in the RNA-Seq analysis, and multiple cytochrome P450 and carboxylesterase encoding genes were down-regulated in the ME-treated males in both the RNA-Seq and RT-qPCR analyses, with a similar fold change detected. For the other genes tested – BdorOR43a-1, BdorOR43b, BdorOR7a-2, BdorOR7a-3, BdorOR7a-5, BdorOR67c, BdorOR59a, and BdorOR69a – they expressed no significant differences according to the qRT-PCR test, not unlike the RNA-Seq results. Hence, the qRT-PCR analysis revealed up- or down-regulated gene expression profiles that were consistent with the RNA-Seq data, confirming that our comparative transcriptome analysis was robust and reliable.

#### Phylogenetic Analysis of the BdorORs

The full-length BdorOR88a and BdorOR63a-1 cDNA segments consisted of 1,245 and 1,248 nt, encoding a polypeptide of 414 and 415 amino acids, respectively. To determine the phylogenetic relationship between BdorORs and the other ORs reported in C. capitata and D. melanogaster, a maximum likelihood tree was constructed. The BdorORs clustered together with the orthologous ORs from two Dipteran species with the best BLASTP hit. The ligand binding ORs from B. dorsalis shared phylogenetic relationships with the OR homologs of both Dipteran species. Notably, BdorOR88a clustered in a branch with DmelOR88a, but clustered in a different group than BdorOR63a-1 (**Figure 4**).

#### Behavioral Activities of Male Flies in Response to ME and BdorOR63a-1, BdorOR88a Expression Level

Under laboratory conditions, the male flies' taxis to ME had a profile similar to their sexual development. As **Figure 5A** shows, fly responsiveness increased with age, with newly emerged males (2-day-old) presenting the lowest taxis, but by the time they were 10-day-old the males had become highly attracted to ME. The 10-day-old male responders numbered 173.67 ± 3.84, which was about 17 times higher than the abundance of 2 day-old males (11.67 ± 1.00) (t = 32.40; df = 2; P = 0.001). Furthermore, in contrast to the low expression observed in the

2-day-old male flies' antennae, BdorOR63a-1 and BdorOR88a were highly expressed in 10-day-old males, corresponding to a 2.40-fold (t = 5.20; df = 2; P = 0.035) and 4.57-fold (t = 13.01; df = 2; P = 0.0059) increase, respectively (**Figures 5B,C**).

The diurnal pattern of male responsiveness to ME is illustrated in **Figure 5D**. The male fly response to ME was highest during the morning (173.67 ± 3.84 males), declining to a lower level in the early afternoon (102.33 ± 3.28), but dropped markedly near dusk (53.00 ± 5.51) (F = 312.49; df = 2; P < 0.0001). Furthermore, the expression levels of BdorOR63a-1 and BdorOR88a were not uniform throughout the day. Specifically, the BdorOR88a expression level was significantly reduced in the afternoon (p.m.) male adult flies compared with their morning (a.m.) counterparts (t = 18.57; df = 2; P < 0.0001) (**Figure 5F**). Interestingly, the transcript levels of BdorOR63a-1 and BdorOR88a at dusk were dramatically higher than those in the morning and early afternoon males (**Figures 5E,F**).

## Functional Characterization of BdorOR88a and BdorOR63a-1 in the Xenopus laevis Oocytes Expression System

To verify whether ME olfactory detection in the B. dorsalis flies is mediated by BdorOR63a-1 or BdorOR88a, we expressed these putative receptors along with co-receptor BdorOrco, by using the Xenopus oocytes and two-electrode voltage clamping recording system. We found that H2O injected oocytes did not generate detectable currents when challenged with either MO or ME (**Figure 6A**). BdorOR63a-1/BdorOrco-expressed oocytes did not respond to MO and only weakly generated a ∼30 nA current in response to ME, even at concentrations as high as 1 × 10−<sup>3</sup> M (**Figure 6B**). By contrast, the BdorOR88a/BdorOrcoexpressing oocytes were clearly activated by ME, with no response to MO. Additionally, ME elicited dose-dependent currents from

the BdorOR88a/BdorOrco-expressing oocytes. This analysis also indicated that the lowest measurable response was observed at a concentration of 1 × 10−<sup>8</sup> M (**Figure 6C**) and the EC<sup>50</sup> value – half the maximal effective concentration refers to the concentration of odorant, which induces a response halfway between the baseline and maximum – was 2.83 × 10−<sup>5</sup> M (**Figure 6D**). Therefore, it is conceivable that BdorOR88a is likely to play a role in the reception of ME by B. dorsalis males.

#### BdorOR88a Mediates the Responsiveness of Mature Male Flies to ME

Microinjection had a clear and negative influence on B. dorsalis mature male flies' survival. As **Figures 7A,B** show, the average mortalities of flies in the dsBdorOR63a-1, dsBdorOR88a, and dsGFP treatment groups at 24 h were 9.20 ± 0.49, 7.60 ± 0.75, and 8.40 ± 0.75%, respectively, and this mortality increased to 15.20 ± 1.02, 12.80 ± 1.36, and 13.60 ± 0.75% at 48 h (**Figures 7A,B**). However, the mortalities of the blank controls were comparatively much lower: 2.40 ± 0.75% at 24 h (F = 16.83; df = 3; P = 0.0001) and 3.60 ± 0.75% at 48 h (F = 24.56; df = 3; P < 0.0001). Notably, mortality of the dsBdorOR63a-1- and dsBdorOR88a-treated male flies was not significantly different from that of the dsGFP-treated male flies.

Consequently, transcript levels of BdorOR63a-1 and BdorOR88a in the adult males were significantly reduced in the dsRNAs-injected flies compared with the two control groups (**Figures 7C,D**). As **Figure 7C** shows, BdorOR63a-1

significant, <sup>∗</sup>P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 (t-tests).

expression was down-regulated in the dsBdorOR63a-1-treated males, to a level approximately 1.49-fold lower than that of the blank control and dsGFP treatment groups at 24 h (F = 55.62; df = 2; P < 0.0001), a significant difference that reached c. 6.25-fold at 48 h (F = 154.08; df = 2; P < 0.0001). Similarly, after treatment for 24 h and 48 h, there was a pronounced decrease in BdorOR88a expression of approximately 1.69-fold (F = 31.72; df = 2; P = 0.0002) and 3.29-fold (F = 46.69; df = 2; P < 0.0001) when compared with that of the dsGFP and blank control groups, respectively (**Figure 7D**).

We then analyzed the effects of RNAi of the BdorOR63a-1 and BdorOR88a gene transcript on male responsiveness to ME. Only the dsBdorOR88a-treated males were much less trapped by ME than were the dsGFP-treated and blank control flies. After 24 h of the treatment, the proportion of males trapped in the dsGFPtreated (89.38 ± 2.83%) and blank control (92.97 ± 3.06%) groups were both significantly higher than recorded in the dsBdorOR88a treatment (73.60 ± 4.16%; F = 53.37; df = 2; P < 0.0001). Furthermore, this difference was greatest at 48 h, when the proportion of ME-trapped males was 58.80 ± 4.25% in the dsBdorOR88a treatment group, significantly less than in the dsGFP-treated (88.63 ± 3.13%) and control (90.71 ± 2.82%) groups (F = 114.81; df = 2; P < 0.0001) (**Figure 7F**). By contrast, males were similarly trapped in the dsBdorOR63a-1 and dsGFP treatment groups and the blank control at 24 h (F = 2.83; df = 2; P = 0.1180) and 48 h (F = 2.87; df = 2; P = 0.1151) (**Figure 7E**).

## Hypothesized Modal Analysis of ME Detection and Transportation Process in Bactrocera dorsalis Mature Males

#### Antennae

Here, we propose a model describing how chemosensory proteins might be used by a mature male to distinguish the ME odorant from a noisy environment (**Figure 8**). Once the ME odorant penetrates the pore tubules of the antennae, the protein products of BdorOBP83a-2 and BdorOBP2 are posited to assist in transporting the ME odorant across the aqueous sensillum lymph. After its release from these proteins, the ME odorant is then transferred to the protein encoded by BdorOR88a. With the ME odorant now bound, the BdorOR88a/BdorOrco complexes are activated to trigger the signals that lead to neural spikes generated in the fly's brain, evoking its characteristic response behavior to ME. After the activation of the odorant receptor, ME odorant is deactivated and degraded by the ODEs.

## DISCUSSION

Because ME is a potent attractant of B. dorsalis mature males, this parapheromone has been widely used for decades as a sexual lure in the detection and control of male B. dorsalis field populations worldwide (Smith et al., 2002; Shelly et al., 2010; Pagadala et al., 2012). However, until now, we do not know how or why it works. Based on this study's RNA-Seq and qRT-PCR results,

two odorant receptor genes (BdorOR63a-1 and BdorOR88a) were found abundantly expressed in mature males' antennae after ME stimulation. Next, we functionally characterized these ORs in a heterologous expression system. In contrast to BdorOR63a-1, BdorOR88a co-expressed with BdorOrco in the Xenopus oocytes elicited dose-dependent inward currents upon application of ME. Additionally, silencing BdorOR63a-1 via the injection of dsRNA had no significant effect on the males' attraction to ME, whereas silencing BdorOR88a significantly reduced the number of males attracted to ME. As such, we conclude that BdorOR88a is necessary for explaining the observed attraction of mature males toward ME. Hence, our present results further improve the current knowledge of the molecular mechanism underpinning ME detection by B. dorsalis mature male flies.

Male fly responsiveness to ME was clearly age-dependent. Males were strongly attracted to ME when 10-day-old, while immature male flies (2-day-old) were not attracted to ME. This result is not unlike that found in prior studies (Karunaratne and Karunaratne, 2012; Liu H. et al., 2017). Accordingly, in contrast to BdorOR63a-1, BdorOR88a was extreme abundantly expressed in the mature male antennae. ME functions as a precursor for B. dorsalis male synthesis of sex pheromone components (Shelly et al., 2008; McInnis et al., 2011; Liu H. et al., 2017). ME-acquired males produced a more attractive sexual pheromonal signal and enjoyed a higher mating success than ME-deprived males (Nishida et al., 1997; Shelly et al., 2000). Decades of publications have suggested that ingestion of ME enhances male mating competitiveness and thus demonstrated that this effect underlies the strong attraction of males to ME (Shelly, 2010). Prior studies had already revealed that male responsiveness to ME was not uniform throughout the day: it peaks in the morning, declines in the afternoon, and drops markedly at dusk (Ibrahim and Hashim, 1980; Tan et al., 1986; Karunaratne and Karunaratne, 2012). Our present study agrees rather well with these observations. The daily fluctuation in male flies responsiveness to ME displays a negative correlation with the daily rhythms of their courting and mating behavior (Karunaratne and Karunaratne, 2012; Liu H. et al., 2017). The attractiveness of a volatile depends on both the chemical properties of volatile and the physiological status of insect (Anton et al., 2007; Gadenne et al., 2016). Generally, responses to sex pheromones are switched on during courting and mating, whereas responses to oviposition-site cues or food odors are switched off. B. dorsalis mature males were very active at dusk, but extremely low numbers were attracted to the ME source (Karunaratne and Karunaratne, 2012). During the courting and mating at dusk, B. dorsalis males may transiently stop responding to the ME until the next morning and begin responding to sex pheromone in search of a mating partner.

Olfactory plasticity is a powerful evolutionary strategy that optimizes critical resources for insect survival

(Gadenne et al., 2016). Recently, study on olfactory plasticity in insects, such as Agrotis ipsilon, Spodoptera littoralis, D. melanogaster, Anopheles gambiae, C. capitata, and B. dorsalis, is becoming one of the hot topics (Jang, 1995; Zhou et al., 2009; Barrozo et al., 2011; Deisig et al., 2012; Rund et al., 2013a,b; Kromann et al., 2015; Jin et al., 2017). Furthermore, olfactory plasticity enables insects to modify their response to semiochemical according to their physiological conditions, such as, feeding state, circadian rhythm, age, and mating status (Jin et al., 2017). Interestingly, olfactory-guided behaviors vary according to the time of day, which help insects respond to environmental chemical stimulus at the right moment (Gadenne et al., 2016). Considerable literature indicated that antennal and olfactory sensory neurons (OSNs) sensitivity of many insects tends to be regulated by the body endogenous rhythm or biological clock (Krishnan et al., 1999; Page and Koelling, 2003; Tanoue et al., 2004; Jin et al., 2017). Clock genes in the Drosophila antennae allow autonomous rhythmicity of environmental cues detection (Krishnan et al., 1999; Tanoue et al., 2004). In Anopheles gambiae mosquitoes, OBPs gene expression rhythms are driven in part by the endogenous circadian clock (Rund et al., 2013a,b). The chemosensory receptors gene expression levels of B. dorsalis gravid female flies fluctuate rhythmically at different times of the day (Jin et al., 2017). In our study, the expression level of BdorOR63a-1 increased gradually from morning to dusk in the mature male antennae. Nonetheless, BdorOR88a was abundantly expressed during the morning, but much less so in the afternoon. Yet more remarkably, BdorOR88a expression level increased considerably at the dusk period. B. dorsalis adults mating activity is restricted to dusk. The accurate detection and recognition of a potential mating partner is the key step in insect courtship and subsequently mating (Sayin et al., 2018). The peak in expression of BdorOR63a-1 and BdorOR88a at dusk may correspond to the time of increasing male chemosensory activity to female-emitted sex pheromone blend (containing several components).

Comparative phylogenetic analyses of the OR repertoire of insects can provide useful information on the evolutionary origin of OR families and their expansion in insect lineages (Missbach et al., 2014; Koenig et al., 2015). More strikingly, in our phylogenetic analysis, BdorOR88a was distributed within a distinct cluster of ORs with DmelOR88a, suggesting that they probably perform the same molecular function in the neuronal circuitry. Intriguingly, in Drosophila, the olfactory receptor OR88a attuned to semiochemicals has been identified. OSNs expressing the olfactory receptor OR88a housed in trichoid sensilla of Drosophila antennae can mediate responses to unidentified odors in the fly's male and female body wash extracts (van Naters and Carlson, 2007). More recently, D. melanogaster OR88a has been characterized as a receptor of the fly-produced odorants methyl myristate, methyl palmitate, and methyl laurate that mediated copulation and attraction (Dweck et al., 2015). Analogous to other fly OR types, the role of DmelOR88a has been considered that of a pheromone receptor (PR) (Dweck et al., 2015; Fleischer et al., 2017). Therefore, we strongly suspect that B. dorsalis mature males' possible use BdorOR88a to detect female sex-pheromones and locate their mates at dusk, although further functional experiments are needed to confirm this hypothesis.

degraded and inactivated by odorant-degrading enzymes.

From chemical stimulus to behavioral response, the olfactory process involves the capture, binding, transport, and inactivation of odors. The initial steps in odor detection involve the binding of an odor to the ORs positioned on the dendritic membrane of the OSNs within antennae. OBPs are able to bind various hydrophobic odorant molecules in the environment (Sato et al., 2008; Taylor et al., 2008; Silbering et al., 2011; Siciliano et al., 2014). Once semiochemicals are bound, the OR/Orco heteromeric complexes are activated to trigger signals leading to characteristic spikes generated in the brain, thereby producing a behavioral response in the insect (Neuhaus et al., 2005; Benton et al., 2006; Hallem and Carlson, 2006; Sato et al., 2008; Croset et al., 2010; Silbering et al., 2011; Di et al., 2017; Fleischer et al., 2017). Gene silencing, via the ingestion or microinjection of dsRNA, recently confirmed that BdorOrco, BdorOBP83a-2, and BdorOBP2 actively had critical roles in mediating the taxis of B. dorsalis males to ME (Zheng et al., 2012; Wu et al., 2016; Liu H. et al., 2017). Considering our results alongside those of prior research to date, we preliminarily posit that the ME odor molecules bind to either BdorOBP2 or BdorOBP83a-2, after which they are transferred to OSNs, where BdorOR88a coordinates with BdorOrco to bind to the ME odor molecules – only then is the olfactory signal transduction pathway finally activated. Our findings from this study thus represent an important breakthrough in our mechanistic understanding of how B. dorsalis mature male flies are attracted to ME. Hopefully, this will advance how attractants are designed to target this specific olfactory pathway, which should promote the development of better and more affordable attractants for B. dorsalis management in the field.

Combinatorial coding, that is individual OR can be activated by multiple odorants and a specific odor ligand can be detected by multiple ORs, is the primary coding mode of the insect olfaction system (Malnic et al., 1999; Hallem et al., 2004; Suh et al., 2014; Andersson et al., 2015; Fleischer et al., 2017). In D. melanogaster, olfactory system sensitivity seems to be enhanced by a combinatorial coding, with specific groups of ORs detecting the same chemical cues (Leal, 2013). Yet more remarkably, knock-down BdorOR88a gene did not lead to the complete disappearance of the mature male flies' responsiveness to ME. Therefore, presumably, in addition to BdorOR88a, there are further ORs and other receptors, which may also

contribute to detect ME. However, due to the lack of B. dorsalis genomic information, many genes cannot yet be annotated and their functions remain unknown, especially for those olfactoryrelated genes. In their detection of odorant signals, insects use several families of chemosensory receptors, including the ORs, ionotropic receptors (IRs), gustatory receptors (GRs), and sensory neuron membrane proteins (SNMPs) (Benton et al., 2007, 2009; Jin et al., 2008; Suh et al., 2014). The identification of a new family of IRs, complementary to the ORs family yet expressed in different olfactory neurons, has provided new insight into the molecular mechanisms of odor detection in insect (Benton et al., 2009; Silbering et al., 2011). IR92a mediates attraction behavior to ammonia and volatile amines in D. melanogaster (Min et al., 2013). SNMPs are highly conserved in multiple insect species and involved in pheromone-based chemical communication (Nichols and Vogt, 2008). One subfamily in particular, SNMP1, when co-expressed with PRs, is believed to contribute to the sensitivity of pheromone detection in insects (Rogers et al., 1997; Benton et al., 2007; Vogt et al., 2009; Liu et al., 2014; Pregitzer et al., 2014). Study has demonstrated that SNMP1 played a vital role in detecting the sex pheromone Z11-18OAc in D. melanogaster (Benton et al., 2007; Jin et al., 2008). Accordingly, in moths, SNMP1 is considered indicative of sex pheromone-responsive neurons in antennae (Forstner et al., 2008; Thode et al., 2008). In our present study, the consistency between the RNA-Seq results and the mRNA expression from the qRT-PCR analysis implies that BdorIR92a and BdorSNMP1-1 possibly participate in the processing of ME detection by mature B. dorsalis male flies. Nonetheless, knowledge of the precise functional relevance of BdorIR92a and BdorSNMP1-1 for ME signaling remains elusive. To conclude, we deduce that these genes are involved in how B. dorsalis males detect ME, but this awaits further investigation and testing.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

HL and Y-YL conceived and designed the experiments and wrote the manuscript. HL, Z-SC, and D-JZ performed the experiments. HL analyzed the data. Z-SC and D-JZ provided material support. All authors discussed the results and reviewed the final manuscript.

#### FUNDING

This research was supported by Guangdong Province College High-Quality Professional Foundation (No. 246), National Key Research and Development Program (No. 2016YFC1201200), and Guangzhou Science and Technology Project, China (No. 201601010179). The funders have no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

#### ACKNOWLEDGMENTS

We sincerely thank the Charlesworth Group for correction of the manuscript language and the Guangzhou Gene Denovo Co., Ltd., for technology support and helpful advice. We are grateful to the members of our laboratory for their cooperation in oriental fruit fly rear and treatment.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00987/full#supplementary-material


olfactory pathway of a male moth. PLoS One 7:e33159. doi: 10.1371/journal. pone.0033159


dorsalis (Hendel). J. Chem. Ecol. 35, 209–218. doi: 10.1016/S0278-6915(02) 00012-1



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Liu, Chen, Zhang and Lu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proceeding From in vivo Functions of Pheromone Receptors: Peripheral-Coding Perception of Pheromones From Three Closely Related Species, Helicoverpa armigera, H. assulta, and Heliothis virescens

#### Edited by:

Peng He, Guizhou University, China

#### Reviewed by:

Haonan Zhang, University of California, Riverside, United States Feng Liu, Michigan State University, United States Longwa Zhang, Anhui Agricultural University, China

#### \*Correspondence:

Yang Liu yangliu@ippcaas.cn Gui-Rong Wang wangguirong@caas.cn; grwang@ippcaas.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 31 May 2018 Accepted: 07 August 2018 Published: 30 August 2018

#### Citation:

Wang B, Liu Y and Wang G-R (2018) Proceeding From in vivo Functions of Pheromone Receptors: Peripheral-Coding Perception of Pheromones From Three Closely Related Species, Helicoverpa armigera, H. assulta, and Heliothis virescens. Front. Physiol. 9:1188. doi: 10.3389/fphys.2018.01188 Bing Wang, Yang Liu\* and Gui-Rong Wang\*

State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China

Three closely related species, Helicoverpa armigera, H. assulta, and Heliothis virescens from Lepidoptera Noctuidae, are used as a model system for exploring sexual communication and species isolation. Pheromone receptors (PRs) previously discovered in model moth species include seven in H. armigera, six in H. assulta, and six in H. virescens. PRs named OR6, OR13, and OR16 among these species were found to be functional, characterized by an in vitro Xenopus oocytes system. Using an in vivo transgenic fly system, functional assays of OR6, OR13, and OR16 clades from three closely related Noctuidae species showed that OR13 function was highly conserved, whereas OR6 and OR16 exhibited functional divergence. Similar results were produced from assays in the Xenopus oocytes system. Combined with earlier behavioral results and electrophysiological recordings, we found corresponding relationships among pheromones, PRs, and neurons at the periphery sensory system of each species. Our results provide vital information at the neuronal and molecular level, shedding insight into the sexual communication of closely related species in Lepidoptera.

Keywords: sex pheromones, Helicoverpa armigera, Helicoverpa assulta, Heliothis virescens, pheromone receptors, transgenic fly

## INTRODUCTION

Moth sex pheromones are released by female moths to attract conspecific males, allowing for longdistance mating attraction. Reception of moth sex pheromones among closely related species is complicated by diverse releasing and receiving pheromone signals, as well as varying components, quantities, and ratios of moth sex pheromones (Berg et al., 1995; Hansson et al., 1995; Baker et al., 2004; Wang et al., 2010; Vasquez et al., 2011; Zhang and Löfstedt, 2015). Sexual communication of closely related moth species in Lepidoptera Noctuidae has been studied over a few decades as a model system for exploring sex pheromone recognition and species isolation mechanisms

**64**

(Kehat and Dunkelblum, 1990; Almaas and Mustaparta, 1991; Berg et al., 1998; Baker et al., 2004; Krieger et al., 2004; Groot et al., 2006; Gould et al., 2010; Wang et al., 2010; Wu et al., 2015; Chang et al., 2016). However, there is not sufficient evidence to explain how intra- and interspecific sexual communication signals of closely related species are discriminated (Vasquez et al., 2011; Zhang et al., 2016).

Three Lepidoptera species across two genera, Helicoverpa armigera and H. assulta in Helicoverpa and Heliothis virescens in Heliothis, are phylogenetically closely related and have been thoroughly studied. H. armigera and H. assulta are sympatrically occurring species found throughout different regions of China, and Heliothis virescens is found in America and other countries (Wang et al., 2005; Cho et al., 2008). Sex pheromone blends found in females of these three species overlap in several sex pheromone components. The major component is (Z)-11 hexadecenal (Z11-16:Ald) in H. armigera and H. virescens, and (Z)-9-hexadecenal (Z9-16:Ald) in H. assulta (Vetter and Baker, 1983; Cork et al., 1992; Baker et al., 2004), occurring in different ratios with other minor components (Nesbitt et al., 1979; Cork et al., 1992; Chang et al., 2016). Five additional compounds were identified in gland extracts of H. armigera females: hexadecanal (16: Ald), hexadecanol (16: OH), (Z)-11-hexadecenol (Z11- 16:OH), (Z)-7-hexadecenal (Z7-16:Ald), and (Z)-9-tetradecenal (Z9-14:Ald) (Nesbitt et al., 1979; Dunkelblum et al., 1980; Kehat and Dunkelblum, 1990). Similarly, seven compounds were identified from gland extracts of H. assulta females: 16:Ald, (Z)- 9-hexadecenyl acetate (Z9-16:OAc), (Z)-11-hexadecenyl acetate (Z11-16:OAc), hexadecanyl acetate (16:OAc), (Z)-9-hexadecenol (Z9-16:OH), Z11-16:OH, and hexadecanol (16:OH) (Cork et al., 1992; Berg and Mustaparta, 1995). However, the H. virescens female glands only produce six aldehydes and alcohols rather than acetates, including tetradecanal (14:Ald), Z9-14:Ald, Z7- 16:Ald, Z9-16:Ald, Z11-16:OH, 16:Ald (Tumlinson et al., 1975; Klun et al., 1980; Vetter and Baker, 1983; Ramaswamy et al., 1985; Teal et al., 1986; Groot et al., 2006, 2009, 2013).

Field tests and behavior experiments have shown that binary pheromone blends of Z11-16:Ald and Z9-16:Ald effectively attract H. armigera males (Kehat et al., 1980; Kehat and Dunkelblum, 1990). Z11-16:OH significantly reduced catches but 16: Ald acted in opposite function when mixed with the sex pheromone principal of H. armigera (Wu et al., 1997). In addition, the pheromone component Z9-14:Ald (found in H. armigera but not H. assulta) mixed with binary pheromone blends of Z11-16:Ald and Z9-16:Ald caught more H. armigera males at lower concentrations compared to H. assulta, whereas it significantly inhibited the attraction behavior of H. armigera at higher concentrations (Gothilf et al., 1978; Kehat and Dunkelblum, 1990; Zhang et al., 2012; Wu et al., 2015). In H. assulta, addition of Z9-14:Ald or Z9-16:OH to the principal pheromone blend in certain amounts significantly reduced trap catch of male H. assulta in both field and laboratory experiments (Cork et al., 1992; Park et al., 1994; Boo et al., 1995). However, when Z9-16:OAc and Z11-16:OAc were added to binary pheromone blends of Z9-16:Ald and Z11-16:Ald at a certain ratio the male H. assulta would show attractive and mating behavior (Cork et al., 1992; Park et al., 1994). In H. virescens, males use Z11-16:Ald and Z9-14:Ald as the principal pheromone blend for upwind flight behavior (Vetter and Baker, 1983; Ramaswamy et al., 1985). When 16:Ald was added to pheromone blends of Z11-16:Ald and Z9-14:Ald, close-range sexual behaviors of male moths usually increased (Vetter and Baker, 1983). However, H. virescens does not produce acetates compared to H. armigera and H. assulta (Tumlinson et al., 1975; Klun et al., 1980; Vetter and Baker, 1983; Ramaswamy et al., 1985; Teal et al., 1986; Groot et al., 2006).

In previous studies, electrophysiological responses of sex pheromone have been recorded from a single cell within trichoid sensillum of male antennae in H. armigera, H. assulta, and H. virescens, showing specific neuron responses activated by sex pheromones (Baker et al., 2004; Gould et al., 2010; Wu et al., 2015; Chang et al., 2016; Xu et al., 2016). Genes encoding pheromone receptors (PRs), expressed on the dendritic membrane of specific olfactory receptor neurons (ORNs) in trichoid sensilla of adult male antennae, are vital to the reception of conspecific sex pheromones (Baker, 2009; Wang et al., 2010; Zhang and Löfstedt, 2015). PRs have been identified and characterized by species from genomic databases, cDNA-library screenings, and the antennal transcriptome sequencing, with seven PRs in H. armigera, six in H. assulta, and six in H. virescens (Krieger et al., 2004; Liu et al., 2012; Zhang et al., 2015a). The function and localization of PRs were demonstrated by electrophysiology methods and in situ hybridization studies (Krieger et al., 2004, 2009; Grosse-Wilde et al., 2007; Baker, 2009; Wang et al., 2010, 2016; Liu et al., 2013a; Jiang et al., 2014; Chang et al., 2016; Xu et al., 2016).

To date, several strategies for deorphanizing Lepidoptera PRs have been developed both in vitro and in vivo systems (**Supplementary Table S1**). The most common method to study insect ORs is in vitro heterologous expression in Xenopus oocytes (de Fouchier et al., 2014; Zhang and Löfstedt, 2015; Cui et al., 2018). Another transgenic fly lines have been used to assay OR function since 2003. The earliest system for studying OR functions was the Drosophila "empty neuron" system (Dobritsa et al., 2003). The advantage of this system is that the target OR gene is expressed in the Drosophila "empty neuron," offering an actual cellular environment and allowing heterologous OR coupling with endogenous Orco. At the same time, the odorants can be delivered in gaseous form and combined with the Drosophila OBPs, in vivo (Hallem et al., 2004; Carey et al., 2010). However, the "empty neuron" system has some limitations for testing other ORs, such as lepidopteran pheromone receptors (Syed et al., 2010). These limitations likely arise due to some essential factors, for instance, sensory neuron membrane proteins (important for pheromone-evoked neuronal activity) are lacking in the ab3A neuron (Benton et al., 2007). However, some studies have proven that the Or67dGAL4 knock-in system is better for detecting the function of moth pheromone receptors in terms of structural, biochemical, and/or biophysical features of the at1 trichoid sensilla (Syed et al., 2010; Vasquez et al., 2013; Wang et al., 2016).

In this study, we constructed a phylogenetic tree from seven identified Lepidopteran species, and revealed orthology with closely related Noctuidae PRs. According to their evolutionary relationships and functions, we selected three sets of homologous

genes, OR6, OR13, and OR16, from H. armigera, H. assulta, and H. virescens, respectively, and predicted highly conserved sequences motifs. Then, we constructed nine transgenic fly lines using the Or67dGAL4 knock-in system for further functional characterization. Specifically, we compare PR functions between the Xenopus oocytes system and the Or67dGAL4 knock-in system, as well as the relationships between PRs and neurons in the peripheral nervous system. Our results summarize the correlations among pheromones, pheromone receptors, and neurons at the periphery of the sensory system from three closely related species in Lepidoptera, as well as provide information to further detect evolutionary relationships of sex pheromones.

#### MATERIALS AND METHODS

#### Insect Rearing

Drosophila stocks were fed cornmeal-agar-molasses medium and maintained under a 12 h light: 12 h dark cycle at 25◦C and 60% relative humidity. The medium was changed after 10 days. Three to ten-days adults were used to test.

#### Fly Strains

Transgenic lines were generated according to standard procedures as described below. The open reading frame encoding OR6/OR13/OR16 genes was cloned into the pVALIUM20 vector (Ni et al., 2011). Independent homozygous UAS-OR lines (with transgene insertions into chromosome II) were generated at the Tsinghua Fly Center (Beijing, China). Driver mutant allele Or67dGAL4 stock was provided by Dr. Barry J. Dickson (Kurtovic et al., 2007). The balancer w-/w-; sp/CyO; TM3/TM6B was used to cross with homozygous driver lines. The driver line in the Or67dGAL4 mutant background was then crossed with the UAS-OR balancer line to establish a final homozygous stock w+/w+; UAS-OR/UAS-OR; Or67dGAL4/ Or67dGAL4 which expressed OR6/OR13/OR16 genes in at1 sensilla neurons. Each OR6/OR13/OR16 insertion was confirmed by sequencing genomic DNA prepared from mutant lines. The final stock was used for electrophysiological experiments.

#### Sequence Analysis and Phylogenetic Tree Construction

The amino acid sequences of OR6, OR13, and OR16 from H. armigera, H. assulta, and H. virescens, respectively, were aligned using ClustalX software (Version 2.1, European Bioinformatics Institute). Dendrograms were labeled by FigTree software<sup>1</sup> . The transmembrane domains of PR6, PR13, and PR16 were predicted using TMHMM Server Version 2.0<sup>2</sup> . The phylogenetic tree of PRs genes in different Lepidoptera species was constructed by RaxML version 8 with Jones-Taylor-Thornton amino acid substitution model (JTT) (Stamatakis, 2014). Node support was assessed using a bootstrap method based on 1000 replicates. The PR and Odorant receptor coreceptor (Orco) data set contained 38 PR and seven Orco sequences identified in Lepidoptera [eight from H. armigera (Liu et al., 2012), seven from H. assulta (Zhang et al., 2015a), seven from H. virescens (Wang et al., 2010), eight from B. mori (Nakagawa et al., 2005; Wanner et al., 2007), five from S. exigua (Liu et al., 2013a), five from S. litura (Zhang et al., 2015b), and five from S. littoralis (Montagné et al., 2012; de Fouchier et al., 2015)]. The phylogeny of the seven moth species above was constructed on the basis of cytochrome oxidase subunit I (COI) genes.

#### Motif-Pattern Analysis

The motif-pattern analysis of proteins was performed broadly using the MEME online server (MEME Suite Version 4.11.2)<sup>3</sup> . A total of nine PRs from H. armigera, H. assulta, and H. virescens were selected to predict the conserved motif pattern. The parameter settings were as follow: maximum number of motifs was eight, minimum motif width was six, maximum motif width was 15, and Expectation maximization (EM) improvement threshold was 10−<sup>5</sup> .

#### Single Sensillum Recordings

Using a transgenic in vivo system, the OR6, OR13, and OR16 genes across three Heliothis/Helicoverpa species were respectively expressed in at1 neurons of Drosophila, and the resulting UAS-OR flies were crossed with a mutant knock-in allele Or67dGAL4 driver line. Extracellular electrophysiological recordings were performed on single at1 sensilla of one to 10 day old flies. The antenna was fixed using standard procedures (de Bruyne et al., 2001; Syed et al., 2006). The reference electrode was placed in the fly eye, under a microscope (LEICA Z16 APO, Germany) at 920 × magnification. Action potentials were recorded by inserting a tungsten wire electrode in the base or in the shaft of a sensillum of the fly antenna. Signals were amplified 10× by a high impedance pre-amplifier (IDAC-4 USB System, Syntech, Kirchzarten, Germany), sent to a PC via an analog-digital converter, and analyzed off-line with AUTOSPIKE v. 3.9 software (Syntech, Kirchzarten, Germany). The filter was set with a 500 Hz low cutoff and a three kHz high cutoff. AC signals were recorded for 10 s, starting 1 s before stimulation. Responses were calculated by counting the number of action potentials 1 s after stimulation (with a delay of 200 ms to allow the odorant to travel down the airstream), and subtracting the number counted in the second before stimulation. Three dimensional bar charts were created in SigmaPlot Version 12.5 (SYSTAT, San Jose, CA, United States). Heatmaps of different PR functions activated by sex pheromone components and analog were generated by Heml 1.0 software (Deng et al., 2014).

#### Odor Stimulation

In total, nine sex pheromone components and analogs, Z9- 14:OAc, Z9-16:OAc, Z11-16:OAc, Z9-14:Ald, Z9-16:Ald, Z11- 16:Ald, Z9-14:OH, Z9-16:OH, and Z11-16:OH, were used to screen in vivo functions of all three types ORs across three Heliothis/Helicoverpa species with paraffin oil as a control.

<sup>1</sup>http://tree.bio.ed.ac.uk/software/figtree/

<sup>2</sup>http:// www.cbs.dtu.dk/services/TMHMM/

<sup>3</sup>http://meme-suite.org/tools/meme

Aliquots of sex pheromone components were dissolved in paraffin oil (v/v), and 10 µL of each solution were loaded onto a 5 × 40 mm Whatman filter paper strip, which was placed inside a Pasteur pipette. Paraffin oil alone was tested as a negative control. For dose-response relationships, serial dilutions were made in increasing doses of 0.001, 0.01, 0.1, 1, 10, and 100 µg/µL and loaded on separate filter paper strips. Each preparation was held in a humidified continuous air flow delivered by the Syntech Stimulus controller (CS-55 model, Syntech) at 1.4 L/min. Stimulus pulses were added for 300 ms. During stimulation, the compensatory flow was switched off.

## RESULTS AND DISCUSSION

## Phylogenetic Analysis Reveals Orthology With Closely Related Noctuidae PRs

In some Lepidoptera species (especially in the superfamily Noctuidae), the number of PRs revealed, identified, and characterized by species were four in Spodoptera exigua (OR6, 11, 13, 16), four in Spodoptera litura (OR6, 11, 13, 16), four in Spodoptera littoralis (OR6, 11, 13, 16), seven in H. armigera (OR6, 11, 13, 14, 14b, 15, 16), six PRs in H. assulta (OR6, 11, 13, 14, 14b, 16), and six in H. virescens (OR6, 11, 13, 14, 15, 16). In addition to Noctuidae species, seven PRs (OR1, 3, 4, 5, 6, 7, 9) were identified and characterized in Bombyx mori, belonging to Bombycidae (Nakagawa et al., 2005; Wanner et al., 2007; Wang et al., 2010, 2016; Liu et al., 2012; Montagné et al., 2012; Liu et al., 2013a,b; Jiang et al., 2014; de Fouchier et al., 2015; Zhang et al., 2015a,b; Chang et al., 2016) (**Figure 1A** and **Supplementary Table S1**).

Full-length amino acid sequences of candidate PRs genes were used to construct a phylogenetic tree from seven identified lepidopteran species including B. mori, H. armigera, H. assulta, H. virescens, S. exigua, S. litura, and S. littoralis (**Figure 1B**). Orthologous genes of the highly conserved co-receptor Orco, were clustered together as Clade I. As expected, sequence identity among them was very high. Another five orthologous clades were shown as noctuids species in Clade II-VI, representing clades OR6, OR11, OR13, OR14/14b/15, and OR16 (**Figure 1B**). The amino acid sequences of PRs across various noctuids species in OR13 clade are quite conserved, showing functional conservation; the sequences of OR6 or OR16 clade are relatively less conserved, exhibiting functional differentiation (**Figure 2**) (Wang et al., 2010; Liu et al., 2013b; Jiang et al., 2014; de Fouchier et al., 2015).

#### Three Sets of Homologous PR Genes Selected and Cloned From Closely Related Species

Evolutionarily, H. armigera, H. assulta, and H. virescens are highly related compared with other Lepidopteran species (Wang et al., 2005; Cho et al., 2008). PRs of these three species could respond to overlapping sex pheromone components (Wang et al., 2010; Liu et al., 2013b; Jiang et al., 2014). Thus, studying evolutionary relationships among PRs in H. armigera and related species will provide valuable information on reproductive isolation.

Based on previous studies of PRs across Heliothis/Helicoverpa species, several pheromone components were used to determine response profiles of all PRs across Heliothis/Helicoverpa species, mainly using an in vitro two-electrode voltage-clamp system (Wang et al., 2010, 2016; Liu et al., 2013b; Jiang et al., 2014; Chang et al., 2016; Xu et al., 2016). We found that none of the OR11 and OR15 PRs across three Heliothis/Helicoverpa species were activated by any pheromone component tested. However, only HvirOR14 of all OR14 PRs across three Heliothis/Helicoverpa species showed response, and was activated by Z11-16:OAc and Z9-14:Ald. Similarly, OR14b from H. virescens was not identified (**Supplementary Table S2**). Therefore, we selected homologous OR6, OR13, and OR16, which play an important role in mating, for comparing the functions across three species. Three sets of homologous PR genes (total of nine PRs) were cloned from cDNA sequences according to the genomic database and antennal transcriptome sequence (Krieger et al., 2004; Liu et al., 2012; Zhang et al., 2015a). Subsequently, all genes were subcloned into the expression vector of the transgenic fly for further functional screening.

## Sequence Analysis of Noctuidae PRs Genes

According to amino acid sequences of orthologous PR genes in the closely related species, H. armigera, H. assulta, and H. virescens, three multiple sequence alignments (OR6, OR13, and OR16) revealed relatively conserved characteristics among orthologous PRs. Each alignment contained seven transmembrane domains (**Figures 2C–E**), with sequence identities of 89.95, 95.54, and 94.08% corresponding to OR6, OR13, and OR16 alignments, respectively.

Nine PR sequences were used to predict highly conserved motifs. A total of eight motifs composed the most common pattern of sequence "7-6-8-5-2-3-1-4," which represented traits with three types of ORs in H. armigera, H. assulta, and H. virescens, respectively (**Figure 2B**). The most typical conserved sequence patterns were located in the conserved C-terminal region as (A/G)-V-Y-(G/L/S)-(V/L)-P-W-(E/D) -(C/Y)-M-D-(T/V)-K-N-R in motif 1, F-H-Q-(A/Y/T)-S-G-C- (L/I)-L-L-L-(E/G)-C-S-Q in motif 2, Q-Q-L-I-Q-(L/I)-S-V-I -F-E-L-(V/L)-G-(S/T) in motif 3, and G-V-(T/Q)-(T/S)-M- (A/T)-(A/S)-I-L-K-T-S-(M/E)-S-Y in motif 4 (**Figure 2A**). The functions of these motifs were thought to be important in protein-protein interactions (Miller and Tu, 2008), especially in the formation of the OR/Orco heteromeric complex (Benton et al., 2006; Vasquez et al., 2013). In addition, another four motifs, motif 5 (H/N)-(W/C/V)-(I/F/V)-(I/L)-S-Y-(L/T)-C- (S/T/A)-(T/S/C)-(W/Y)-F-C-(M/Y)-(F/Y), motif 6 L-F-N- (L/M/I)-(I/T)-P-(M/F)-Y-(S/N)-(N/C)-(Y/L)-(A/S) -(A/R)-G- (R/M/K), motif 7 K-(I/T)-H-L-F-(Y/H)-(Y/H)-(K/R)-(D/H/ N/E)-(R/K)-S-(K/E/D)-(Y/H/Q/A)-A-(M/Y), and motif 8 N-(S/A/T/R)-T-(F/Y)-(E/D)-H-(S/A)-(L/V/M)-(Y/F)-Y-(S/L/P)- (Y/V)-P-F-(D/N), had lower conservation and exhibited more sequence variation. It is possible that some amino acid residues were highly variable, resulting in functional differentiation.

species. Six clades (I to VI) are shown in this tree representing Orco, OR13, OR11, OR14/15, OR16, and OR6 clades, respectively.

FIGURE 2 | Motif analysis of pheromone receptors (PRs) identified from three closely related Lepidoptera species, and the alignment of amino acid sequence of three set of PRs. (A) The eight motif-pattern discovered in nine PRs from Helicoverpa armigera, H. assulta, and Heliothis virescens. (B) The locations of each motif-pattern on the predicted protein sequence from N-terminal to C-terminal. Smaller numbers indicate higher conservation. (C) The alignment of amino acid sequence of clade OR6 from H. armigera, H. assulta, and H. virescens. TM1-TM7 indicates seven transmembrane domains. Harm: H. armigera; Hass: H. assulta; Hvir: H. virescens. (D) The alignment of amino acid sequence of clade OR13 from H. armigera, H. assulta, and H. virescens. (E) The alignment of amino acid sequence of clade OR16 from H. armigera, H. assulta, and H. virescens.

However, the reason for evolutionary differences of PRs presented in closely related species remains unclear.

#### In vivo Functional Assays of Closely Related Noctuidae PRs

In H. armigera and H. assulta, the OR6-expressing neurons in at1 sensilla mainly responded to the sex pheromone component analogs Z9-14:OH and Z9-16:OH, at a dose of 1 mg loaded in the stimulus cartridge, whereas HvirOR6-expressing neurons responded to Z9-14:Ald and analog Z9-14:OH (**Figures 3A,D** and **Supplementary Figure S1**). In a dose–response experiment, neurons in at1 sensilla started firing at doses as low as 10 ng, with Z9-14:OH and Z9-16:OH EC<sup>50</sup> values of 3.85 × 10−<sup>5</sup> and 5.84 × 10−<sup>5</sup> g in H. armigera, 9.66 × 10−<sup>5</sup> and 5.99 × 10−<sup>5</sup> g in H. assulta, and Z9-14:Ald and Z9-14:OH EC<sup>50</sup> values of 2.75 × 10−<sup>5</sup> and 1.26 × 10−<sup>4</sup> g in H. virescens (**Figure 4A** and **Supplementary Figure S1**).

The function of the OR13 gene was highly conserved (**Figure 3D** and **Supplementary Figure S2**). We found that OR13-expressing neurons in at1 sensilla responded specifically to the sex pheromone component Z11-16:Ald at a dose of 1 mg across three Heliothis/Helicoverpa species (**Figure 3B**). Dose– response results showed neurons in at1 sensilla started to respond to Z11-16:Ald at a threshold of 10 ng, and continued to receive stimulation in a concentration gradient up to 1 mg (**Figure 4B** and **Supplementary Figure S2**). The EC<sup>50</sup> values of Z11-16:Ald were 2.13 × 10−<sup>4</sup> , 2.42 × 10−<sup>4</sup> , and 2.16 × 10−<sup>4</sup> g in H. armigera, H. assulta, and H. virescens, respectively.

By comparison, the OR16 gene exhibited functional divergence (**Figure 3D** and **Supplementary Figure S3**). In H. armigera, the HarmOR16-expressing neurons in at1 sensilla responded to the sex pheromone components Z9-14:Ald, Z11-16:OH, and Z11-16:OAc (**Figure 3C** and **Supplementary Figure S3**). In a dose–response experiment, neurons in at1 sensilla started firing at doses as low as 10 ng, with a Z9-14:Ald EC<sup>50</sup> value of 1.26 × 10−<sup>3</sup> g and a Z11-16:OH EC<sup>50</sup> value of 7.94 × 10−<sup>5</sup> g (**Figure 4C** and **Supplementary Figure S3**). In H. assulta, the HassOR16-expressing neurons in at1 sensilla responded to the sex pheromone components Z9-16:Ald, Z9- 14:Ald, and Z9-16:OH (**Figure 3C** and **Supplementary Figure S3**). In addition, neurons in at1 sensilla showed a dose–response, with a Z9-16:Ald EC<sup>50</sup> value of 8.56 × 10−<sup>5</sup> g (**Figure 4C** and **Supplementary Figure S3**). In H. virescens, the sex pheromone components Z11-16:OH, Z9-16:Ald, Z11-16:OAc, and Z9-14:OH activated the HvirOR16-expressing neurons in at1 sensilla (**Figure 3C** and **Supplementary Figure S3**). The dose–response experiment showed Z11-16:OH and Z11-16:OAc EC<sup>50</sup> values of 3.89 × 10−<sup>5</sup> and 8.88 × 10−<sup>5</sup> g (**Figure 4C** and **Supplementary Figure S3**), respectively.

## PR Functional Comparison Test Between Xenopus Oocytes and Or67dGAL4 Knock-In Systems

According to the previous functional identifications of PRs using the Xenopus oocytes system (Wang et al., 2010; Liu et al., 2013b; Jiang et al., 2014; Chang et al., 2016), we summarized

transgenic flies. (B) Responses of OR13-expressing neurons in at1 sensilla of transgenic flies. (C) Responses of OR16-expressing neurons in at1 sensilla of transgenic flies. (D) Heatmap of response spectra of PR-expressing neurons in at1 sensilla of transgenic flies.

the sex pheromone response profiles of PRs across H. armigera, H. assulta, and H. virescens, and the functions of these PRs using the Or67dGAL4 knock-in system (**Table 1**). Through a comparative analysis of different methods on functional identification, we found that ligand-binding traits of PRs detected by the Xenopus oocytes system are essentially consistent with that of the Or67dGAL4 knock-in system. This was especially true for functionally conserved PR, OR13, where the best binding-ligand of three orthologous OR13s across Heliothis/Helicoverpa species was the sex pheromone component Z11-16:Ald, regardless of which methods we used. In general, OR/Orco expressed in the Xenopus oocytes system was more sensitive to the sex pheromone components. However, the in vivo Or67dGAL4 knock-in system has generally proven to be more accurate and specific (Wang et al., 2016).

By comparison, the function of OR6 was relatively divergent. HvirOR6 was mainly tuned to Z9-14:Ald in both in vivo and in vitro systems. However, HarmOR6/Orco and HassOR6/Orco were all tuned to Z9-14:Ald, Z9-16:Ald, Z9-16:OH, and Z9- 14:OH using the Xenopus oocytes system, whereas only Z9- 16:OH and Z9-14:OH activated HarmOR6/HassOR6 expressing at1 neurons (**Table 1**). These results may be explained by additional factors; for instance, the suitability of ligand concentrations, or whether some OR genes were able to work properly in the Or67dGAL4 knock-in system. It is pointed out that Z9-14:OH is not a sex pheromone component in any of these closely related species (Nesbitt et al., 1979; Klun et al., 1980;Cork et al., 1992), but instead activates HarmOR6/ HassOR6/ HvirOR6 expressing at1 neurons. This phenomenon requires further investigation.

The function of OR16 was highly divergent and widely tuned to more than three sex pheromone components or analogs, including Z11-16: OH, Z11-16:OAc, and Z9-14:Ald. The major ligands from HarmOR16, HassOR16, and HvirOR16 using both in vivo and in vitro methods were essentially identical (**Table 1**).

#### The Relationship Between PRs and Neurons in the Peripheral Nervous System

Three closely related species use their sensitive olfactory system to specially recognize interspecific-overlapping sex pheromone components. Using previous results from the Xenopus oocytes system (Wang et al., 2010, 2016; Liu et al., 2013a; Jiang et al., 2014; Chang et al., 2016) and our results from the Or67dGAL4 knock-in system combined with in situ hybridization and electrophysiological recordings, functional characterization between neurons and odorant receptors were predicted (**Table 1**).

In previous studies, electrophysiological responses of peripheral sex pheromone recognition were recorded from a single sensilla within trichoid sensillum of male antennae in H. armigera, H. assulta, and H. virescens (Baker et al., 2004; Gould et al., 2010; Wu et al., 2015; Chang et al., 2016; Xu et al., 2016). A total of three trichoid sensilla subtypes have been identified to perceive sex pheromone components, A-type, B-type (missing in H. assulta) and C-type, each housing two ORNs.

Combined with behavioral results, there may be a correlation between some electrophysiological responses and the functional identification of pheromone receptors. For instance, in H. armigera, Z9-14:Ald was previously found to effectively enhance attractions at lower concentrations, and significantly inhibit attraction behavior at higher concentrations


TABLE 1 | The comparison of functional characterizations between neurons and odorant receptors.

"?" Represents uncertainty between neurons and odorant receptors. Italic with underline means the possible alternative. The highlighted with bold font represents the result in this study. Refs: <sup>a</sup> (Baker et al., 2004); <sup>b</sup> (Wang et al., 2010); <sup>c</sup> (Chang et al., 2016); <sup>d</sup> (Liu et al., 2013b); <sup>e</sup> (Jiang et al., 2014); <sup>f</sup> (Yang et al., 2017). It is all predicted the relation between neurons and odorant receptors according to references in this table.

(Gothilf et al., 1978; Kehat and Dunkelblum, 1990; Zhang et al., 2012; Wu et al., 2015), whereas Z11-16: OH was found to be a behavioral inhibitor (Wu et al., 1997). Single sensillum recordings showed an "a-spike" ORN (HarmOR6 or HarmOR14b. The predictions of expressed neurons are given for each ORN) in C-type sensillum was tuned to two behavioral agonists, Z9-14:Ald and Z9-16:Ald, while a "b-spike" ORN (HarmOR16) in C-type sensillum was tuned to three behavioral antagonists, Z9-14:Ald, Z11-16:OH, and Z11-16:OAc (Chang et al., 2016, 2017; Yang et al., 2017) (**Table 1**). In H. assulta, an "a-spike" ORN (HassOR6 or HassOR14b) in C-type sensillum was tuned to Z9-16:Ald and Z9-14:Ald, while a "b-spike" ORN (HassOR16) in C-type sensillum was tuned to the behavioral antagonist, Z9-14:Ald, and analogs Z9-14:OH and Z9-16:OH (Chang et al., 2016, 2017; Yang et al., 2017). In H. virescens, an "a-spike" ORN (HvirOR14) in C-type sensillum was tuned to Z11-16:OAc, while a "bspike" ORN (HvirOR16) in C-type sensillum was tuned to Z11-16:OH (interspecific inhibitor) and Z9-14:Ald (Almaas and Mustaparta, 1991; Baker et al., 2004; Wang et al., 2010) (**Table 1**).

In general, electrophysiological responses showed an "aspike" ORN (predicting OR13-expressing neuron) in A-type sensillum across all three species was activated by the sex pheromone component Z11-16:Ald, but another "b-spike" ORN (OR11) in A-type sensillum is still uncharacterized (**Table 1**). In A-type sensillum, the functions of expressed ORs are relatively conserved. In addition, the number of A-type sensilla confers a larger proportion of all trichoid sensilla in H. armigera than in H. assulta, in accordance with the understanding that Z11-16: Ald is major sex pheromone component in H. armigera (Chang et al., 2016).

One "a-spike" ORN (OR14b or OR6) in B-type trichoid sensillum is known to be mainly tuned to the sex pheromone component Z9-14: Ald, whereas none of ligands activate a "bspike" ORN (OR15) in B-type sensillum.

Overall, we summarized the relationships among sensilla, neurons, and PRs involving sex pheromone recognition in the peripheral-coding olfactory system of three Heliothis/ Helicoverpa species (**Table 1**). It is evident that neuron function in type-A trichoid sensilla completely matched the function of PRs (OR13 and OR11). However, relationships between neurons in type-B or -C trichoid sensilla and PRs did not fully clarified. The Or67dGAL4 knock-in system used to detect the function of moth pheromone receptor is nearly identical to the Xenopus oocytes system (Wang et al., 2016). A few functional differences are observed between PRs and endogenous neurons in moths which may be driven by many factors such as the cell environment, gene expression, lack of accessories, and category and concentration of ligand. In addition, the functions of OR14b or OR6 in H. armigera and H. assulta still exist differences in previous studies (**Table 1**) (Jiang et al., 2014; Chang et al., 2016; Yang et al., 2017). Therefore, the CRISPR/Cas9 genome editing technique combined with electrophysiological response assays are needed for functional characterization of OR14b (OR6). It is better for elucidation of the molecular and neuronal mechanisms of sex pheromone identification.

#### The Evolution of Lepidoptera PRs Selectivity

Three Heliothis/Helicoverpa male species can perceive respective sex pheromone components released from their female pheromone blends. A few hypotheses have been proposed on how variation is generated during pheromone evolution of closely related species, such as the "asymmetric tracking" hypothesis and the gene duplication hypothesis (Phelan, 1992; Gould et al., 2010; Heckel, 2010). However, it is still elusive how subtle variations of sex pheromone components are precisely distinguished by males of different species. Certain moth PRs of closely related species are evolutionarily conserved under strong selective pressure, whereas PRs are more functionally divergent if relaxed from evolutionary constraint (Zhang and Löfstedt, 2013, 2015). The latter is broadly tuned to the behavioral antagonists and agonist, which efficiently increased the specificity and selectivity of interspecific pheromone detection (Zhang and Löfstedt, 2015). This is consistent with our finding that OR16 from three closely related species exhibits largely functional divergences. The function of HarmOR16 from H. armigera has been confirmed to be activated by the pheromone antagonist Z11-16:OH, which regulates optimal mating time and influences fecundity (Chang et al., 2017). One study revealed that single mutations in PRs across Asian and European corn borers selectively altered pheromone recognition

#### REFERENCES


(Leary et al., 2012). Another study showed that two site mutations of HassOR14b changed ligand selectivity (Yang et al., 2017). Thus, the evolutionary relationship of structure and function of PRs in closely related Lepidoptera species will help reveal the mechanisms underlying reproductive isolation and speciation.

#### AUTHOR CONTRIBUTIONS

BW, YL, and G-RW designed the experiments. BW performed the experiments and analyzed the data. YL and G-RW contributed reagents, materials, and gene identification. BW and G-RW wrote and revised the paper.

## FUNDING

This work was funded by the National Natural Science Foundation of China (31725023, 31621064, 31402023, and 31230062) and China Postdoctoral Science Foundation (2014M550905).

#### ACKNOWLEDGMENTS

We thank Dr. A. Ray for kindly providing the Or67dGAL4 mutant knock-in fly line with Dr. B. Dickson's permission; Dr. J. Q. Ni for kindly providing the pVALIUM20 vector and for the transformation service; and Dr. J. Shen for kindly providing the balancer w-; sp/CyO; TM3/TM6B strain.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01188/full#supplementary-material

FIGURE S1 | SSR traces from OR6-expressing neurons in at1 sensilla of Drosophila in response to pheromone compounds.

FIGURE S2 | SSR traces from OR13-expressing neurons in at1 sensilla of Drosophila in response to pheromone compounds.

FIGURE S3 | SSR traces from OR16-expressing neurons in at1 sensilla of Drosophila in response to pheromone compounds.

TABLE S1 | Functional characterizations of PR genes in some Lepidopteran species.

TABLE S2 | Functional characterizations of PRs in three Heliothis/Helicoverpa species in vitro.




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wang, Liu and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Functional Studies of Sex Pheromone Receptors in Asian Corn Borer Ostrinia furnacalis

Wei Liu<sup>1</sup> , Xing-chuan Jiang<sup>2</sup> , Song Cao<sup>1</sup> , Bin Yang<sup>1</sup> \* and Gui-rong Wang<sup>1</sup> \*

<sup>1</sup> State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China, <sup>2</sup> School of Plant Protection, Anhui Agricultural University, Hefei, China

Lepidopteran insects use sex pheromones for sexual communication. Pheromone receptors expressed on peripheral olfactory receptor neurons (ORNs) are critical part to detect the sex pheromones. In genus Ostrinia, several pheromone receptors were functional analyzed in O. nubilalis and O. scapulalis but the knowledge in O. furnacalis was rare. In this study, seven pheromone receptors were deorphanized by heterologous expression system of Xenopus oocytes. Functional types of sensilla trichoidea were classified by single sensillum recordings to interpret the response pattern of olfactory sensory neurons to Ostrinia pheromone components. OfurOR4 and OfurOR6 responded to the major sex pheromone Z/E12-14:OAc. OfurOR4 is the main receptor for both Z/E12-14:OAc and OfurOR6 mainly responded to E12- 14:OAc. Functional differentiation of gene duplication were found between OfurOR5a and OfurOR5b. OfurOR5b showed a broad response to most of the pheromone components in O. furnacalis, whereas OfurOR5a was found without ligands. OfurOR7 showed a specific response to Z9-14:OAc and OfurOR8 mainly responded to Z11- 14:OAc and E11-14:OAc. OfurOR3 did not respond to any pheromone components. Our results improved the current knowledge of pheromone reception in Ostrinia species which may contribute to speciation.

Keywords: odorant receptors, ligands, single sensillum recordings, olfactory, Xenopus oocytes

#### INTRODUCTION

Sex pheromone has been used by organisms for sexual communication, this remarkable trait is representative in insects especially for Lepidopterans (Symond et al., 2011). Male could detect and respond to female pheromone over long distance, e.g., 11 km for emperor moth Pavonia pavonia (Regnier and Law, 1968). Moth percept the sex pheromone via the pheromone sensitive trichoid sensilla distributed on their antennae. The entire olfactory system is heavily dependent on the types of receptors expressed in peripheral olfactory receptor neurons (ORNs; also called olfactory sensory neurons) which housed in the olfactory sensilla (Leal, 2013). This has been proved unambiguously by expressing an allospecific pheromone receptor PxylOR1 from the diamondback moth in the ORN that houses the bombykol receptor BmorOR1 in the silkworm moth, Bombyx mori. Electrophysiological and behavioral experiments showed that PxylOR1 expressing male silkworm moths responded equally to bombykol (E10Z12-16:OH) and Z11-16:Ald (Sakurai et al., 2011).

#### Edited by:

Shuang-Lin Dong, Nanjing Agricultural University, China

#### Reviewed by:

Xianhui Wang, Institute of Zoology (CAS), China Wei Xu, Murdoch University, Australia Dan-Dan Zhang, Lund University, Sweden Alex Liu, Auburn University, United States

#### \*Correspondence:

Bin Yang byang@ippcaas.cn Gui-rong Wang wangguirong@caas.cn; grwang@ippcaas.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 27 February 2018 Accepted: 02 May 2018 Published: 23 May 2018

#### Citation:

Liu W, Jiang X-c, Cao S, Yang B and Wang G-r (2018) Functional Studies of Sex Pheromone Receptors in Asian Corn Borer Ostrinia furnacalis. Front. Physiol. 9:591. doi: 10.3389/fphys.2018.00591

**76**

The genus Ostrinia (Lepidoptera: Crambidae) consists of 21 species worldwide and served as the model system for research of pheromone communication. Several species in this genus are important agricultural pests such as O. nubilalis and O. furnacalis (Mutuura and Munroe, 1970; Ohno, 2003). The species in this genus use relatively simple components (Z9-14:OAc, E11- 14:OAc, Z11-14:OAc, E12-14:OAc, Z12-14:OAc and E11-14:OH) for the recognition among individuals (Roelofs et al., 1985; Huang et al., 1998a,b,c; Ishikawa et al., 1999a,b; Takanashi et al., 2000). Eight pheromone receptors and odorant receptor co-receptor have been successfully functionally characterized either in vivo or in vitro among O. furnacalis, O. nubilalis, O scapulalis, and O. latipennis (Miura et al., 2009, 2010; Wanner et al., 2010; Leary et al., 2012; Yang et al., 2016).

Asian corn borer, O. furnacalis, is a grievous pest in China and causing serious damage on economic crop maize for 10–30% yield lost (Wang et al., 2000). In addition, this species fed on various host (over 27 species) belonging to nine families (Yuan et al., 2015). Females of O. furnacalis use Z12-14:OAc and E12-14:OAc with the ratio of 1:1 as their major sex pheromone components to attract males (Cheng et al., 1981; Huang et al., 1998b). Although the pheromone receptors were functionally characterized in the sibling species such as O. nubilalis and O scapulalis, only one pheromone receptor (OfurOR4) and an odorant receptor co-receptor (OfurOR2) has been deorphanized in O. furnacalis (Leary et al., 2012; Yang et al., 2015, 2016; Zhang et al., 2015). The functional types of the sensilla have been described by Takanashi et al. (2006) and Domingue et al. (2007). In this study, all the pheromone receptors in O. furnacalis were functionally characterized using Xenopus oocytes system. In addition, single sensillum recordings were carried out to confirm the ORNs response for detecting the pheromones.

## MATERIALS AND METHODS

#### Insects

Ostrinia furnacalis was maintained under laboratory conditions with artificial diet at 28◦C, 14:10 (L:D), 60% relative humidity. Pupae were placed in tube individually for eclosion. Two-dayold adults were used in the present study. Male antennae were removed and frozen in liquid nitrogen immediately, then stored under −80◦C until use.

#### Pheromone Components

The pheromone components including (Z)-9-tetradecenyl acetate (Z9-14:OAc), (Z)-11-tetradecenyl acetate (Z11-14:OAc), (E)-11-tetradecenyl acetate (E11-14:OAc), (E)-11-tetradecen-1-ol (E11-14:OH), (Z)-12-tetradecenyl acetate (Z12-14:OAc), E-12-tetradecenyl acetate (E12-14:OAc) (95% minimum purity) were purchased from Nimrod Inc. (Changzhou, China). For Xenopus oocyte system, chemicals were prepared in dimethyl sulfoxide (DMSO) to form the stock solutions (1 M) and stored at −20◦C. The stock solution was diluted in 1× Ringer's buffer (96 mM NaCl, 2 mM KCl, 5 mM MgCl2, 0.8 mM CaCl2, and 5 mM HEPES pH 7.6) before experiments. 1× Ringer's buffer was used as a negative control. For single sensillum recording, each pheromone compound was prepared as 1 µg/µl in hexane solution and stored at −20◦C. The hexane was used as a negative control.

## RNA Extraction and cDNA Synthesis

Total RNA was isolated from male antennae with TriZol reagent (Invitrogen, Carlsbad, CA, United States) according to the manufacturer's instruction. The cDNA was synthesized using RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific) after a DNase I (Thermo Scientific) treatment. The quality of RNA was verified by Nanodrop ND-1000 spectrophotometer (NanoDrop Products, Wilmington, DE, United States) and gel electrophoresis.

#### Cloning of Pheromone Receptors in O. furnacalis

Full length of ORF encoding odorant receptors of O. furnacalis was obtained from antennal transcriptomic analysis and amplified by PCR using primeSTAR HS DNA polymerase following the manual (Takara, Dalian, China) (Yang et al., 2015). Primers used in this study were listed in **Supplementary Table S1**. Transmembrane domains were predicted by TMHMM Server Version 2.0<sup>1</sup> and multiple sequence alignment and identity calculation were done by the DNAMAN 6.0 (Lynnon Biosoft, United States).

#### Electrophysiological Recordings Using Xenopus Oocyte System

Each receptor was first cloned into a blunt-vector (TransGen Biotech, China), subsequently subcloned into a PT7TS vector, and then took for cRNA synthesis using mMESSAGE mMACHINETM T7 Kit (Thermo Fisher Scientific). Mature healthy Xenopus oocytes (stage V-VII) were prepared according the description from Liu et al. (2013). Briefly, the oocytes were separated and then treated with 2 mg/ml collagenase I in washing buffer (96 mM NaCl, 2 mM KCl, 5 mM MgCl<sup>2</sup> and 5 mM HEPES, pH 7.6) for 1–2 h at room temperature. The 1:1 mixture of pheromone receptor and OfurOrco (OfurOR2) cRNA (27.6 ng each) were microinjected into the oocytes. After an incubation for 4–7 days at 18◦C in incubation medium (1× Ringer's buffer, 5% dialysed horse serum, 50 mg/ml tetracycline, 100 mg/ml streptomycin, and 550 mg/ml sodium pyruvate), oocytes were recorded with a two-electrode voltage clamp. Currents induced by pheromone components (100 µM) were recorded using an OC-725C oocyte clamp (Warner Instruments, Hamden, CT, United States) at a holding potential of −80 mV. The data were acquired and analyzed with Digidata 1440A and pCLAMP 10.0 software (Axon Instruments Inc., Union City, CA, United States).

## Single Sensillum Recordings

Sensilla trichoidea from 2-day-old male adults were used for the recordings. Individuals were restrained in a remodeled 1 ml plastic pipette tip with an exposed head fixed by dental wax, and antenna from one side was attached to a coverslip with doubleface tape. Two tungsten wire electrodes were used with one

<sup>1</sup>http://www.cbs.dtu.dk/services/TMHMM/

inserting into an compound eye and another into the sensilla. Ten individuals were recorded at basal (4), middle (3), and proximal (3) part of the antennae and ten sensilla were recorded for each individuals. Ten micrograms pheromone components (dissolved in hexane) were performed for each trial. Air flow was set at 1.4 L/min with a 300 ms stimulus air pulse controlled by Syntech Stimulus controller (CS-55, Syntech, Kirchzarten, Germany). AC signals were recorded (10 s, starting 1 s before stimulation) using a data acquisition controller (IDAC-4, Syntech, Kirchzarten, Germany) and analyzed with AUTOSPIKE v. 3.9 software (Syntech, Kirchzarten, Germany). The filter setting was 500 Hz at low cutoff and 3 kHz at high cutoff. Responses were calculated by counting the number of action potentials 1 s after stimulation.

#### Phylogenetic Analysis

fphys-09-00591 May 18, 2018 Time: 16:55 # 3

Sequences of O. furnacalis were based on the transcriptome data (Yang et al., 2015). Sequences from other Ostrinia species were from the reported references (Miura et al., 2009, 2010; Yasukochi et al., 2011) and downloaded through NCBI. The amino acid sequences of pheromone receptors were aligned by MAFFT<sup>2</sup> . Phylogenetic tree was constructed and analyzed by bootstrap test with 1000-resampling through RAxML version 8 with the Jones-Taylor-Thornton amino acid substitution model (JTT) (Stamatakis, 2014).

#### Statistical Analysis

Data in the present study were normalized by log(X+1) and represented as mean ± SEM. The differences of responses to each pheromone components were analyzed by One-Way ANOVA and followed Duncan test (P < 0.05) by SPSS 20.0 (IBM, Endicott, NY, United States).

## RESULTS

#### Gene Cloning and Sequence Analysis of Pheromone Receptors in O. furnacalis

All the pheromone receptor names in this study were followed Yang et al. (2015). The naming system of pheromone receptors among O. furnacalis, O. nubilalis, and O. scapulalis were shown in **Table 1**. Full length of amino acid sequences of the pheromone receptors (ranged from 421 to 474aa) and the predicted seven transmembrane domains were shown in **Figure 1**. The identity between all pheromone receptors was 58.66%. Among all the pheromone receptors, OfurOR5a and OfurOR5b shared high similarity and their identity was 88.21%. Identities among other receptors were significantly lower (e.g., OfurOR8/Ofur5a, 71.30%; OfurOR4/OfurOR6, 64.71%; OfurOR1/OfurOR3, 64.08% etc.). OfurOR1 was not cloned from the strain we used.

## OR4 and OR6 Are Main Receptors for Z/E12-14:OAc

OfurOR4 mainly responded to the main sex pheromones of O. furnacalis, Z12-14:OAc and E12-14:OAc, with the current

<sup>2</sup>https://www.ebi.ac.uk/Tools/msa/mafft

values of 1876.8 ± 165 and 727.9 ± 120.4 nA, respectively. Both of the responses are significantly higher than that to other components (358.2 ± 156.6 nA < currents < 526.2 ± 110.1 nA, F = 31.821, P < 0.001, N = 5) (**Figure 2A**). OfurOR6 showed a much lower response to E12-14:OAc compared to OfurOR4, with the current value of 140.7 ± 6.0 nA, but the response was still significantly higher than that to other components (17.9 ± 2.2 nA < currents < 50.6 ± 10.0 nA, F = 33.490, P = 0.000, N = 7) (**Figure 2B**). Considering the effect for the applying order of the components, we used different order for OfurOR4, which E12-14:OAc was firstly applied to the oocytes, the response to E12-14:OAc became extremely strong (current > 3404.5 nA, N = 2) and even inhibited the response of Z9-14:OAc, Z11-14:OAc, and E11-14:OAc (**Supplementary Figure S1**).

#### OR5b, OR7, and OR8 Broadly or Narrowly Tuned to Other Pheromones

OfurOR7 showed a specific response to one pheromone component Z9-14:OAc, with the current value of 212.2 ± 46.3 nA (F = 21.053, P = 0.000, N = 6) (**Figure 2C**). OfurOR8 significantly responded to Z/E11-14:OAc (F = 45.2210, P = 0.000, N = 5), with the current values of 76.8 ± 14.1 nA(Z) and 144.6 ± 28.3 nA(E), respectively (**Figure 2D**). Besides, a weak response to Z9-14:OAc (8.5 ± 5.2 nA) was also found in OfurOR8. Interestingly, OfurOR5a and OfurOR5b shared high sequence similarity, but only OfurOR5b responded to the pheromone components. OfurOR5b broadly tuned to Z9-14:OAc, E11-14:OAc, Z11- 14:OAc, E12-14:OAc, and Z12-14:OAc. The responses to Z9- 14:OAc, Z11-14:OAc were significantly higher than to E11- 14:OAc, E12-14:OAc, and Z12-14:OAc (F = 4.155, P = 0.000, N = 4) (**Figure 2E**). OfurOR5a and OfurOR3 did not respond to any pheromone compounds supplied in this study (**Figure 2F**).

## Electrophysiological Analysis of the Male s. trichoidea

The single sensillum recordings were performed on the s. trichoidea of male antennae. In total 95 s. trichoidea were

TABLE 1 | Name system of functionally characterized pheromone receptors in genus Ostrinia between different research articles.



successfully recorded, among them, 82 sensilla responded to the provided pheromone components. Spontaneous activity often indicated more than one class of spike amplitudes that suggested that spikes from more than one neuron were recorded. But it was difficult to discriminate how many neurons in one sensillum or which neuron was responsible for the stimuli because the boundary between spikes was unclear. Four types (A–D) of sensilla were observed in which most of them were Type A (79.2%, 76/96) and they responded to all the

provided pheromones except E11-14:OH. The mean responses to Z/E12-14:OAc were relatively higher than other components but no significant difference between them in Type A sensilla (**Figure 3**). Other types were also observed but the abundance was very low, with the number of 2 (Type B), 3(Type C), and 1(Type D). Type C sensilla responded to three components: E11-14:OAc, Z/E12-14:OAc. Type B and Type D showed specific response to Z/E12-14:OAc and Z9-14:OAc, respectively (**Figure 3**).

## DISCUSSION

The genus Ostrinia has been treated as the model system to study sex pheromone communication because sex pheromone components have been identified in nine species and many species use same pheromone components with different proportion. We cloned seven sex pheromone receptors based on the previous transcriptomic study (Yang et al., 2015) and reviewed the names of the deorphanized pheromone receptor system in different Ostrinia research articles (**Table 1**). Unlike Bombyx mori (Sakurai et al., 2004; Nakagawa et al., 2005), in which the main pheromone receptors were narrowly tuned, most of pheromone receptors in O. furnacalis were broadly tuned to the pheromone components in Xenopus oocyte system. The result was basically consistent with the previous studies (Miura et al., 2009; Wanner et al., 2010; Leary et al., 2012). Among all pheromone receptors, OfurOR4 had significantly stronger response than the other tested receptors. The possible reason might be the system we used was heterologous expression system. When the pheromone receptor expressed in vivo there are other factors which affect the odor perception like OBPs, SNMP, etc. It was reported that the PBPs could increase the sensitivity of PRs to pheromones (Chang et al., 2015). Other receptors we tested might need OBP or SNMP to achieve higher sensitivity. It was also possible that other receptors need to expressed together to form a channel to achieve higher sensitivity coordinately. In O. nubilalis, different ORs could be observed in one neuron by in situ hybridization (Koutroumpa et al., 2014).

OfurOR4 has been identified as the receptor which could equally response to main components Z12-14:OAc and E12- 14:OAc (Leary et al., 2012). Our results were basically consistent with the previous study. Z/E12-14:OAc might share same binding sites and could interfere with each other thus stimulate order could affect the results of the recording. That might cause the difference in response of the different stimuli order for OfurOR4. We found the additional receptor (OfurOR6) for main component E12-14:OAc. It seems that O. furnacalis need OfurOR4 and OfurOR6 to perceive its pheromone components coordinately, but the mechanism need to be further studied. O. furnacals use Z12-14:OAc and E12-14:OAc with ratio of 1:1 (Cheng et al., 1981; Huang et al., 1998b). In the field test, any trap lure loaded with a ratio other than 1:1 of Z/E12-14:OAc (more Z12 or E12) will cause the reduced captures (Cheng et al., 1982). Thus OfurOR4 might receive specific ratio of 1:1 Z/E components to initiate mating behavior. If the ratio deviates from 1:1 like more E12, OfurOR6 might have specific response to this redundant part of E12 and initiate antagonistic behavior together with OfurOR4.

It is interesting that the phenomenon of gene duplication for pheromone receptors in Ostrinia is very common. Various duplicates for pheromone receptors could be observed in each OR group (Yasukochi et al., 2011). In O. furnacalis, functional

differentiation of gene duplication was found in OfurOR5. Similar phenomenon was found in other Lepidopterans. In Helicoverpa armigera, HarmOR14 and HarmOR14b shared high degree of identity but with different function in vitro. HarmOR14b responded to Z9-14:Ald whereas HarmOR14 did not response to any of H. armigera pheromone components (Liu et al., 2013; Chang et al., 2016). In Agrotis segetum, AsegOR1, AsegOR6-10 share high levels of amino acid sequence identity with each other (>70%), whereas their function were dramatically different (Zhang and Löfstedt, 2013).

OfurOR7 showed a specific response to Z9-14:OAc which is the sex pheromone component of O. zaguliaevi and O. zealis (Huang et al., 1998a; Ishikawa et al., 1999a) and a behavioral antagonist (Takanashi et al., 2006). OfurOR8 mainly responded to Z11-14:OAc and E11-14:OAc which were the sex pheromone components of O. nubilalis sex pheromone (Roelofs et al., 1985). Thus, OfurOR7 and OfurOR8 might be involved with interspecific recognition in Ostrinia species. Besides, OfurOR7 is the only one of pheromone receptors which highly expressed in male and female simultaneously (Yang et al., 2015), indicate that Z9-14:OAc might be an important pheromone clue for both sexes. Those receptors might contribute to reproductive isolation between Ostrinia species.

Phylogenetic relationship showed that each OR group (OR1, OR3-8) in Ostrinia formed a clade and shared high degrees of identity (81.35–97.44%) (**Figure 4**). But most of the response pattern, especially for receptor responsible to the main pheromone components, was quite different among those closely related Ostrinia species when compare with previous studies. In genus Ostrinia, the ratio of the Z/E main pheromone components was usually considered to regulate the mating behavior. Those different response patterns make that mechanism more complex and need to be solved case by case. O. nubilalis and O. scapulalis used same pheromone components with same ratio (Z/E11-14:OAc, 97:3-Ztype, and 1:99-Etype) (Glover et al., 1987; Ishikawa et al., 1999b), OnubOR4 mainly responded to E11-14:OAc and OnubOR6 responded to Z11- 14:OAc, the response values were equal in this two main receptors (Wanner et al., 2010). OscaOR4 showed a similar response compared to OnubOR4, but no response of OscaOR6 to any pheromones (Miura et al., 2010). O. furnacalis used Z/E12- 14:OAc with 1:1 ratio which was quite different from O. nubilalis and O. scapulalis, OfurOR4 equally responded to Z12-14:OAc and E12-14:OAc, OfurOR6 mainly responded to E12-14:OAc different from OnubOR6 which responded to Z components. Comparisons of other pheromone receptors in Ostrinia were listed in **Figure 4** in detail.

The results of single sensillum recordings were basically similar to the previous studies (Takanashi et al., 2006; Domingue et al., 2007), most of the sensilla (Type A) responded to five pheromone components but we failed to distinguish the exact neurons. Corresponding to the results from Xenopus oocyte

system, it seems that multiple the pheromone receptors were expressed on the neurons in Type A sensilla. We found three other types in which Type B sensilla only responded to Z/E12- 14:OAc, indicated that the neuron in these sensilla might specifically express OfurOR4 and OfurOR6. Similarly, Type D of which specifically responded to Z9-14:OAc are possibly associated with the expression of OfurOR7. Type C responded to E11-14:OAc, Z/E12-14:OAc, which similar to Type A but difficult to speculate the expressed receptors in these sensilla according the results from Xenopus oocyte system. Possibly because the pheromone receptors co-expressed in O. furnacalis, that pattern has been reported in its closely related specie O. nubilalis (Koutroumpa et al., 2014). We did not find any neuron that responded E12-14:OH. In Ostrinia, OscaOR1, and OlatOR1 has been reported for responding E12-14:OH (Miura

et al., 2009). Thus OfurOR1 might has same response profile. OfurOR1 could not be cloned in our strain and also not exist in the transcriptome (Yang et al., 2015) might indicated the degeneration of OfurOR1 in the colony we used. And it can also be that expression level of OfurOR1 is too low. Utilization of in situ hybridization and CRISPR-Cas9 might further elucidate the neuron distribution and receptor expression pattern in single sensillum.

#### AUTHOR CONTRIBUTIONS

WL, BY, and G-rW designed the research, analyzed the data, and wrote the paper. WL and SC performed the research. X-cJ provided biological samples.

#### FUNDING

This work was funded by the National Natural Science Foundation of China (31701859, 31725023, and 31621064).

#### ACKNOWLEDGMENTS

fphys-09-00591 May 18, 2018 Time: 16:55 # 8

We thank Sai Zhang and Yi Lin for rearing the experimental moth for the study.

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00591/full#supplementary-material

FIGURE S1 | Current traces of OfurOR4/OfurOR2 in response to pheromone compounds (100 µM) with different order.

TABLE S1 | Primers of pheromone receptors used for PCR.



antennae of Asian Corn Borer, Ostrinia furnacalis (Guenee) (Lepidoptera: Crambidae). PLoS One 10:e0128550. doi: 10.1371/journal.pone.012 8550

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Liu, Jiang, Cao, Yang and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Various Bee Pheromones Binding Affinity, Exclusive Chemosensillar Localization, and Key Amino Acid Sites Reveal the Distinctive Characteristics of Odorant-Binding Protein 11 in the Eastern Honey Bee, Apis cerana

Xin-Mi Song1†, Lin-Ya Zhang1,2†, Xiao-Bin Fu1†, Fan Wu<sup>1</sup> , Jing Tan<sup>1</sup> and Hong-Liang Li <sup>1</sup> \*

*<sup>1</sup> Zhejiang Provincial Key Laboratory of Biometrology and Inspection and Quarantine, College of Life Sciences, China Jiliang University, Hangzhou, China, <sup>2</sup> College of Life Science, Shangrao Normal University, Shangrao, China*

#### Edited by:

*Peng He, Guizhou University, China*

Reviewed by: *Ke Yang, Institute of Zoology (CAS), China Liang Sun, Tea Research Institute (CAAS), China*

\*Correspondence:

*Hong-Liang Li hlli@cjlu.edu.cn*

*†These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

Received: *07 February 2018* Accepted: *04 April 2018* Published: *23 April 2018*

#### Citation:

*Song X-M, Zhang L-Y, Fu X-B, Wu F, Tan J and Li H-L (2018) Various Bee Pheromones Binding Affinity, Exclusive Chemosensillar Localization, and Key Amino Acid Sites Reveal the Distinctive Characteristics of Odorant-Binding Protein 11 in the Eastern Honey Bee, Apis cerana. Front. Physiol. 9:422. doi: 10.3389/fphys.2018.00422* Odorant-binding proteins (OBPs) are the critical elements responsible for binding and transporting odors and pheromones in the sensitive olfactory system in insects. Honey bees are representative social insects that have complex odorants and pheromone communication systems relative to solitary insects. Here, we first cloned and characterized OBP11 (*AcerOBP11*), from the worker bees antennae of Eastern honey bee, *Apis cerana*. Based on sequence and phylogenetic analysis, most sequences homologous to AcerOBP11 belong to the typical OBPs family. The transcriptional expression profiles showed that AcerOBP11 was expressed throughout the developmental stages and highly specifically expressed in adult antennae. Using immunofluorescence localization, AcerOBP11 in worker bee's antennae was only localized in the sensilla basiconica (SB) near the fringe of each segment. Fluorescence ligand-binding assay showed that AcerOBP11 protein had strong binding affinity with the tested various bee pheromones components, including the main queen mandibular pheromones (QMPs), methyl p-hydroxybenzoate (HOB), and (*E*)-9-oxo-2-decanoic acid (9-ODA), alarm pheromone (n-hexanol), and worker pheromone components. AcerOBP11 also had strong binding affinity to plant volatiles, such as 4-Allylveratrole. Based on the docking and site-directed mutagenesis, two key amino acid residues (Ile97 and Ile140) were involved in the binding of AcerOBP11 to various bee pheromones. Taken together, we identified that AcerOBP11 was localized in a single type of antennal chemosensilla and had complex ligand-binding properties, which confer the dual-role with the primary characteristics of sensing various bee pheromones and secondary characteristics of sensing general odorants. This study not only prompts the theoretical basis of OBPs-mediated bee pheromones recognition of honey bee, but also extends the understanding of differences in pheromone communication between social and solitary insects.

Keywords: Apis cerana, odorant-binding protein, transcriptional expression profile, immunofluorescence localization, fluorescence binding assay, site-directed mutagenesis

## INTRODUCTION

Different with solitary insects, honey bees are typical social insects and bee colony generally has three types of bees (one queen, numerous workers, and several drones) (Plowes, 2010). As the core of the colony, the queen is a unique female bee that has the ability to breed offspring (after mating with drones) and assemble the whole of bee colony. The workers are in charge of rearing brood larvae, defending hives and foraging pollen and nectar (Pirk et al., 2011). Young adult worker bees always act as nurse bees for rearing brood larvae in the hive, and can change as foragers when older gradually (Weng et al., 2013). Bee colony always has quite complex pheromone cognitive system, which includes sex pheromones between virgin queen and drones, worker pheromones between worker bees, brood pheromones released from brood larvae, and alarm pheromones instantly released from guard worker bees when endangered (Pirk et al., 2011) and so on. Due to the hugeness of the numbers of bee colony, bee members have to utilize the bee pheromones to communicate each other in the hive. Therefore, bee pheromones and the corresponding sensing systems play a crucial role involved in regulating the complex social behavior of bee colonies.

In general, insects recognize odors or pheromones through their olfactory system. Odor molecules in the external environment are first carried by the odorant-binding proteins (OBPs) across the chemosensillar lymph, and then interact with the olfactory receptors (ORs) on the dendritic membrane of olfactory neurons, eventually resulting in electrical signals toward the central nervous system (Shanbhag et al., 1999; Brito et al., 2016). OBPs are low molecular weight, water-soluble globulins that transport odorant molecules across the lymph (Pelosi, 1996; Pelosi et al., 2017). In insects, OBPs can be divided into three subfamilies: pheromone-binding proteins (PBPs), general odorant-binding proteins (GOBPs), and antennal specific proteins (ASPs) or antennal-binding protein (ABPx) (Zhou, 2010).

Up to now, research on OBPs have focused on in solitary insects (Pelosi, 1996; Pelosi et al., 2017), such as Lepidoptera (Wang et al., 2004; Yang et al., 2016; Dong et al., 2017), Hemiptera (Sun et al., 2017; Wang et al., 2017), Blattodea (He et al., 2017), Diptera (Kim et al., 1998), and Coleoptera (Leal et al., 1998) etc. For typical social Hymenopteran, such as Apis melliera, its chemoreceptive system are obviously complex for its social behavior and life cycle. Based on the whole of A. mellifera genome (Honeybee Genome Sequencing Consortium, 2006), 21 OBPs were found and 9 OBPs of them were primarily expressed in antennae (Forêt and Maleszka, 2006). There are 171 olfactory receptors in the genome (Robertson and Wanner, 2006), and AmOr11 is the receptor for the major queen substance component 9-ODA (Wanner et al., 2007). As the functional studies, the OBP1 (ASP1) has been characterized as the queen pheromone-binding protein (Danty et al., 1999; Birlirakis et al., 2001; Pesenti et al., 2008). The OBP2 (ASP2) belong to GOBPs family (Danty et al., 1997; Briand et al., 2001), and OBP14 is a Cminus OBPs (having lost two conserved cysteines) (Zhou, 2010; Schwaighofer et al., 2014) etc. Recently, OBP11 in A. mellifera, was identified to be expressed in rare antennal sensilla basiconica in female bees, both workers and queens (Kucharski et al., 2016), while the physiological function of OBP11 associated with odor binding profiles is still unclear.

As the similar bee species of A. mellifera, Apis cerana is unique to China and capable of searching for sporadic nectar sources, and plays an important role in pollination of plants in mountainous areas (Radloff et al., 2010). Up to now, 17 OBPs have been found in A. cerana (Zhao et al., 2016). So far, three typical OBPs of them has been reported in-depth, ASP2 (OBP2) was specially distributed in worker bee antennae (Li et al., 2008), and bind the floral volatile with the dynamic binding mode (Li et al., 2013). ASP1 (OBP1) was expressed abundantly on the sensilla placoidea in drone antennae (Zhao et al., 2013b), and it can bind queen pheromone component with the static binding mode (Weng et al., 2015). OBP11 have the highest expression in the stage of foragers, which display the highest olfactory sensitity in the A. cerana (Zhao et al., 2013a). In order to protect unique domestic bee resources in China, it is necessary to further study the physiological mechanisms of olfactory recognition system related to social behavior.

In this study, we successfully cloned AcerOBP11 from the antennae of A. cerana worker bees. The expression profiles of AcerOBP11 in different developmental stages and tissues were determined by qRT-PCR, and the chemosensillar localization was observed in worker bee antennae. Moreover, we generated recombinant and mutant AcerOBP11 proteins, and identified that AcerOBP11 can bind to bee pheromones and related plant (floral) volatiles using a competitive fluorescence assay. We then predicted amino acids of AcerOBP11 that bind candidate ligands, and confirmed their role in ligand binding by molecular docking and site-directed mutagenesis. Our functional analysis of AcerOBP11 is of great significance to complement the characteristics of OBPs family of olfactory systems that are associated with A. cerana's unique social behavior.

## MATEIALS AND METHODS

#### Insects and Tissue Preparation

A. cerana colonies were maintained in Langstroth hives in Hangzhou city, Zhejiang province, China. The developmental stages of workers were classified following Michelette (Michelette and Soares, 1993). Antennae of 1,000 adult worker bees were pooled for RNA extraction of transcriptional sequencing. To analyze gene expression pattern during development, 100 worker eggs, 3 larvae, and 3 pupae were used for RNA extraction. To analyze gene expression in various adult tissues in different castes, antennae, head, thorax, abdomen, legs and wings from 1-day old workers, nurse workers (with feeding behavior of larvae, usually 6–18 days old) and forager workers (with carrying powder and pollination behavior, usually after 18 days of age) were used for RNA extraction, where each tissue/caste combination contained tissue from 50 worker bees.

#### Plant Volatiles and Bee Pheromones

All enzymes, kits and vectors, unless specified otherwise, were bought from TaKaRa (JP). ProteinIso <sup>R</sup> Ni-NTA Resin and fast mutagenesis system kit were purchased from Transgen Biotech Co. Ltd (Beijing, CN). Primers were synthesized from Sangon biotech Co. Ltd (Shanghai, CN), immunofluorescence related reagents were purchased from Beyotime Biotechnology (Shanghai, CN), plant volatiles and bee pheromones (purity > 97%) were purchased from J&K and TCI Technology Co., Ltd (Tokyo, JP). The rest of the reagents were domestic analytical reagents.

## Total RNA Isolation, cDNA Synthesis, and Cloning of Full-Length AcerOBP11 cDNA

Total RNA was extracted from each tissue using Trizol (Invitrogen, US) according to the manufacturer's protocol. First-strand cDNA was synthesized using PrimeScriptTM 1st Strand cDNA Synthesis Kit (TaKaRa, JP). Based on the OBP11 homologous sequence of A. mellifera (GenBank accession: DQ435328.1), the full-length primer of AcerOBP11 was designed and BamH I and Xho I restriction enzyme sites were introduced into the upstream and downstream primers. The upstream primer sequence was 5′ -CCGGATCCATGAAAGCAGCAGAAA T-3′ and the downstream primer sequence was 5′ -TTCTCGAGT CACGGAGCAATAAACGC-3′ . The purified PCR products were subcloned into the pMD18-T vector (TaKaRa, JP) using a 1:3 (vector: PCR products) molar ratio by incubating the mixture with T4-DNA ligase at 4◦C for 16 h. After transforming the ligation product into trans5α competent E. coli cells, the positive colonies were selected by white/blue screening and PCR with gene specific primers. Products were then submitted for sequencing company (Sangon, CN).

## Sequencing Analysis and Phylogenetic Tree Construction

The putative N-terminal signal peptides and cleavage site were predicted using SignalP V4.0 (http://www.cbs.dtu.dk/services/ SignalP/) (Petersen et al., 2011). OBPs protein alignments were made using ClustalX V1.83 (Thompson et al., 1997) with default gap penalty parameters of gap opening 10 and extension 0.2, and were edited using ESPript (http://espript.ibcp.fr/ESPript/ ESPript/) (Robert and Gouet, 2014). Phylogenetic tree was constructed by the neighbor joining method using MEGA V6.0 (http://www.megasoftware.net/) (Tamura et al., 2013) with bootstrap support of tree branches assessed by re-sampling amino acid positions 1,000 times.

## Quantitative Real-Time PCR (qRT-PCR)

Quantitative RT-PCR was performed using the iCycler iQ Real-Time PCR Detection System (Bio-Rad, US) with SYBR green dye (TaKaRa, JP). Experimental primers were qOBP11-F 5′ -CT ACGGAATACGGAGAA-3′ and qOBP11-R 5′ -AATAAACGCT ATGGGAT-3′ , and control primers to amplify β-Actin was Be-Actin-F 5′ -TCCTGCTATGTATGTCGC-3′ and Be-Actin-R was 5 ′ -AGTTGCCATTTCCTGTTC-3′ . The relative gene expression data were analyzed using the 2−11CT method by Livak (Livak and Schmittgen, 2001). Statistical analysis data (mean ± SE) from various samples (The developmental and tissues stage were analyzed, respectively) were subjected to one-way analysis of variance (ANOVA) followed by a least significant difference (LSD) test for mean comparisons. The significant differences were determined by p-values. Each experiment was performed in triplicates.

#### Expression, Purification, and Confirmation of Recombinant AcerOBP11 Protein

AcerOBP11 was subcloned into the pET-32a (+) prokaryotic expression vector, and expressed in E. coli at high yields (>16 mg/L) through inducing by IPTG (the final concentration is 1 mmol/L). The AcerOBP11 recombinant protein was first expressed in the supernatant, then the denatured protein was purified using Ni2<sup>+</sup> affinity chromatography for two rounds. After purification, the N-terminal tag was removed by enterokinase. The digested protein products were dialyzed 6–7 times with urea-free PBS dialysate (pH = 7.4) to obtain stable proteins with high purity. All purified AcerOBP11 recombinant protein was detected using standard SDS-PAGE method. The gel band containing the aim proteins was first cut out, digested by trypsin, and the detailed peptide sequences of the target proteins were identified using an LC-MS/MS mass spectrometry (Easy-nLC 1000 LTQ Obitrap ETD, Thermo Fisher, US). The secondary structure of purified AcerOBP11 recombinant protein was analyzed using circular dichroism (CD) spectrometry (815 type, Jasco, JP). Bradford assay was used to determine the concentration of AcerOBP11 and protein samples were stored in −20◦C to generate polyclonal antisera and conduct the binding assays.

## Scanning Electron Microscopy (SEM)

For scanning electron microscopy (SEM), antennae of A. cerana worker bees were cleaned in 0.01 mol/L PBS (pH = 7.4) 3 times for 1 h. After treatment with 70% ethanol for 30 min, the samples were air-dried. The preparations were mounted on holders and examined using a SEM of XL30-ESEM (Philip, NL) after gold coating using a K500X sputter coater (Emitech, UK). Different sensilla types were classified following previously published criteria (Dietz and Humphreys, 1971).

## Fluorescence Immunocytochemical Localization

Female Bal B/C mice were repeatedly injected with AcerOBP11 recombinant protein emulsified in Freund's adjuvant, and antisera were obtained after 6–8 weeks and used without further purification. A. cerana worker foragers were collected from the hive, and their antennae were cut and embedded in OCT-Freeze medium. Antenna samples were sectioned using the Lecia-CM 1900 freezing microtome (Leica, DE). For the fluorescence immunocytochemical analysis, antennal sections were incubated with blocking buffer [1% BSA in TBS (contain 20% Tween-20, v/v)] for 1 h at RT, and then incubated with anti-AcerOBP11 antibodies in TBST at a 1:500 dilution for 1 h. After three washes with TBST, the sections were incubated with goat anti-mouse IgG conjugated with DyLight549 red dye (Beyotime, CN) at a 1:1,000 dilution in TBST for 1 h. The secondary antibodies of AcerOBP11 were used in the experiment as the negative control. After three washes with TBST, the sections were mounted in antifade mounting medium (Beyotime, CN) and observed with an Axio Observer Z1 microscope equipped with a LSM710 confocal laser scanning microscope (CarlZeiss, DE).

#### Fluorescence Competitive Binding Experiments

Fluorescence experiments on AcerOBP11 with N-phenyl-1 naphthylamine (1-NPN) were carried out on a Shimadzu RF-5301 spectrofluorimeter using a quartz cuvette in a right-angle configuration. The interactions were monitored by recording 1-NPN fluorescence upon addition of 1-NPN aliquots with excitation wavelength of 337 nm, and emission wavelength of 350–450 nm, where the slit was 5 nm. Titrations were carried out at 25◦C with 1 µmol/L recombinant protein in PBS buffer. The fluorescence intensities at the maximum of emission wavelength (400∼410 nm) were recorded to calculate Scatchard plots. The dissociation constant of the protein/1-NPN complex (K1−NPN) was calculated from Scatchard plots and applied in the equation below to estimate ligand binding affinity. All 23 ligands (1 mM) used in competition experiments were dissolved in spectrally pure grade methanol. Three independent measurements were taken for binding data. The concentrations of competitors that resulted in a reduction of fluorescence to half-maximal intensity (IC<sup>50</sup> values), were taken as a measure of binding affinity constants calculated from the corresponding IC<sup>50</sup> values using the following formula (Ban et al., 2003): K<sup>D</sup> = [IC50]/(1+[1 – NPN]/K1−NPN), where [1-NPN] is the free concentration of 1- NPN and K1−NPN is the dissociation constant of the complex AcerOBP11/1-NPN, which were calculated from the binding curve using the Origin 8.5 (OriginLab Inc.).

#### Molecular Docking

A 3D structure (Kiefer et al., 2009) of AcerOBP11 was predicted from A. mellifera odor binding protein 5 (AmOBP5) crystal structure (PDB entry code 3r72.1) using SWISS-MODEL online (https://www.swissmodel.expasy.org/). The 3D structures of all candidate pheromones and plant volatiles were obtained from NCBI PubChem online (https://pubchem.ncbi.nlm.nih.gov/). The 3D structure of the strongest binding ligand was docked with the predicted crystal structure of OBP11 via the Molegro Virtual Docker (MVD) 4.2 (free trial). The MolDock Optimizer and MolDock Score was used as the search criteria and grading standards, respectively (René and Christensen, 2006). The best docking model was selected for the pose display of OBP11 binding with candidate ligand. Residue distribution and hydrogen bond around AcerOBP11 were obtained when bound with a ligand to the key amino acid sites. Docking models were visualized with the UCSF Chimera package (Pettersen et al., 2004). Based on the docking analysis, the detailed energy values and hydrogen bonds involved in the binding of AcerOBP11 with ligands were calculated, and then displayed as a heat-map. The energy intensity was indicated as the depth of color, and the predicted hydrogen bonds were labeled using black frames.

#### Site-Directed Mutagenesis and Confirmation of Key Sites

In order to verify whether the predicted interaction sites played a role in the binding of AcerOBP11 protein with ligands, site-directed mutagenesis of the corresponding amino acid was carried out. The mutant primers were designed by a partial overlap method before synthesis. The plasmids of pET-32a/AcerOBP11 wild-type were mutated by using fast mutagenesis system kit (Transgen, CN) and then transformed into BL21 (DE3) competent cells. The mutant plasmids of target sites were confirmed by sequencing, and then the mutant proteins were obtained by the induction and purification in the same method as AcerOBP11 wild-type above. The secondary structure of purified AcerOBP11 recombinant protein mutants was also analyzed using circular dichroism (CD) spectrometry (815 type, Jasco, JP). The mutant proteins were used for competitive fluorescence experiments with the six candidate ligands chosen from the previous section. The target amino acids and binding mode of AcerOBP11 binding with candidate ligands were acquired by comparing the dissociation constant K<sup>D</sup> between wild-type and mutant AcerOBP11. Statistical analysis data (mean ± SE) of K<sup>D</sup> values for the same mutated amino acid site were also used as one-way analysis of variance (ANOVA) followed by a mean LSD test.

## RESULTS

#### Coding and Amino Acid Sequences of AcerOBP11

We cloned the coding sequence of AcerOBP11 from A. cerena. The peptide did not contain predicted signal peptides by SignalP 4.1 Server (http://www.cbs.dtu.dk/services/SignalP/). We aligned the protein sequence with other homologous sequences and its predicted secondary structure, and found that AcerOBP11 contained amino acid sequence characteristics of OBPs, such as three pairs of disulfide bonds composed of six conserved Cysteines (**Figure 1A**). The AcerOBP11 full-length ORF is 429 bp and the protein's molecular weight is approximately 15 kDa with an isoelectric point of 5.21, the GenBank accession of AcerOBP11 was obtained as KC818631.1. The phylogenic tree (**Figure 1B**) showed that AcerOBP11 shared relative with some homologous OBPs from diverse Hymenopteran species. The amino acid sequence of AcerOBP11 shared highly similar to other PBPs, such as A. mellifera OBP11 (91%), Apis dorsata PBP3 (90%), Eufriesea mexicana PBP2 (79%), Melipona quadrifasciata PBP1 (66%), and Bombus terrestris GOBP 99a (72%) etc. It suggests that AcerOBP11 belongs to OBPs family in A. cerana, and has partly sequence characteristics of insect's PBPs and GOBPs.

#### Transcriptional Profiling of AcerOBP11 in Various Tissues

We characterized the expression profiles of AcerOBP11 in different tissues and developmental stages of A. cerena using real-time PCR. During the stage of development, AcerOBP11 expression was higher in pupae than larvae and eggs (p < 0.01, ANOVA LSD; **Figure 2A**). In the adult workers, AcerOBP11 was highly expressed in the antennae of newborn, nurse and forager bees (p < 0.01, ANOVA LSD). In addition, AcerOBP11 showed high expression in the wing of newborn bees, low expression in legs of all three adult stages and low expression in the wings

represents the amino acids sequences that do not contain a signal peptide. Red box represents conserved amino acids domains including six highly cysteines (labeled by green numbers below). The predicted secondary structures (e.g., α-helix) are shown above the corresponding sequences. (B) The phylogenetic tree of AcerOBP11 with other homologous proteins based on the method of Neighbor-Joining (Bootstrap = 1,000 times) using MEGA 6.0 software.

of the forager (**Figure 2B**). We found highest expression levels in the antennae, suggesting that AcerOBP11 expression may be important in antennal physiological activity of worker bees. For different adult stages, we observed highest expression of AcerOBP11 in newborn worker bees, followed by forager bees and nurse bees, indicating that AcerOBP11 expression is dynamic in the antennae of adult worker bees.

## Preparation and Confirmation of Recombinant AcerOBP11 Protein

For the preparation of antibodies and the functional characterization of AcerOBP11, we induced and expressed AcerOBP11 in E. coli and purified recombinant AcerOBP11 (**Figure 3**). The recombinant AcerOBP11 proteins without His-Tag were then purified through Ni2<sup>+</sup> affinity chromatography column to obtain for subsequent experiments (**Figure 3A**, lane 4). The peptide sequence of the AcerOBP11 protein was identified using LC/MS-MS. As shown in **Figure 3C**, Figure S1, the identified peptide containing 37 peptides with high scores only belonged to the same AcerOBP11 protein group with a total coverage of 67.17%. It indicated that the purified AcerOBP11 recombinant protein should be integrated and errorless. Purified proteins were then used to generate mouse antibodies against AcerOBP11 and the fluorescence binding assay.

#### Immunocytochemical Localization

OBP proteins are generally expressed in the chemosensilla, and we found that A. cerana chemosensilla are primarily distributed on the antennal flagellum by SEM (**Figure 4A**). Using the newly generated AcerOBP11 antibodies, we conducted fluorescentlabeled immunocytochemical staining of the A. cerana worker bee antennae, and found high expression in the sensilla basiconica, but not in the sensilla trichoid and sensilla placodea on the antennae. AcerOBP11 expressing sensilla basiconica were mainly localized to the tip of antennae (**Figures 4B,D**), as well as restricted areas close to the interval between two segments on the antennal flagellum (**Figures 4C,E**). These results suggest that AcerOBP11 is specifically expressed in the antennal sensilla basiconica near the fringe of each segment in A. cerana worker bee.

## Ligand-Binding Assay of AcerOBP11

Using the 1-NPN fluorescence reporter, we tested the binding affinity of AcerOBP11 to candidate plant volatiles and bee pheromones (**Figure 5A**, **Table 1**). The fluorescence competitor assay curve for each compound is shown in **Figures 5B,C**. All the values of dissociation constants were calculated and listed in **Table 1**. Among the 23 ligands in the assay, 22 candidate ligands except for methyl oleate reduced the relative fluorescence of 1-NPN to below 50% of AcerOBP11, indicating that AcerOBP11 bound to these compounds. The K<sup>D</sup> values of QMP component (HOB), plant volatiles (4-hydroxyveratrole), and alarm pheromone (n-hexanol) were 1.35, 2.67, and 2.79 µmol/L, respectively. The three compounds had lower K<sup>D</sup> values (<3 µmol/L), suggesting that they have stronger affinity to bind AcerOBP11. Moreover, the other bee alarm pheromones (isoamyl acetate), worker pheromone (farnesol), and brood pheromone (ethyl palmitate) were the strongest competitive ligands for 1- NPN in each group of components (**Table 1**).This is the first study about the function of OBPs in A. cerana.

## Predicting Key Sites Through Analysis of Docking and Energy

Molecular docking can predict the interaction between proteins and small molecules. We generated a heat-map with detailed energy and hydrogen bond of the amino acids, and identified

including pET32a-AcerOBP11 plasmid without and with induction of 1 mmol·L −1 IPTG, respectively. Lane 3 and 4 represent purified recombinant AcerOBP11 proteins before and after digestion with enterokinase, respectively. (B) The N-terminal tag of the mutant recombinant proteins are removed by enterokinase. M is the protein molecular weight marker. Lane 1–3 contains AcerOBP11m-Ile140, m-Phe101, and m-Ile97 mutant proteins after enterokinase digestion. The two rounds purified AcerOBP11 protein is labeled by red and blue arrow on the right, respectively. (C) The purified AcerOBP11 recombinant protein was identified by LC-MS/MS, and the green letters represent those amino acid sequences that have a total coverage of 67.17% with AcerOBP11 protein.

three amino acids, Ile97, Ile140, and Phe101, that are likely to play important roles in the binding of the AcerOBP11 to six ligands that had high affinity to AcerOBP11 (**Figure 6**, Table S1). In particular, Ile97 contributed a hydrogen bond for AcerOBP11 to bind to 4-hydroxyveratrole, HOB, isoamyl acetate and ethyl palmitate. Ile140 contributed a hydrogen bond for AcerOBP11 to bind to n-hexanol, HOB, and farnesol. Considering the energy contributions, we predicted that Ile97, Ile140, and Phe101 might be key amino acids in AcerOBP11 for its binding to ligands.

For the assessment of AcerOBP11 mutant with test ligand, a binding example of AcerOBP11 with n-hexanol was described. As displayed in **Figure 7A**, in AcerOBP11 wild-type, n-hexanol was located in a binding cavity composed of four hydrophobic amino acids of Ile97, Val96, Lys95, and Ile140. When Ile97 was mutated as glycine, the acting amino acids changed as the three hydrophobic amino acids, Ile140, Phe139, and Met131 close to the C-terminal (**Figure 7B**, Table S2). Compared with the AcerOBP11 wild-type, the hydrogen bond was changed from Ile140 to Met131. The number of key amino acids decreased and the location also changed. It indicates that Ile97 may play a key role in the binding of AcerOBP11 with n-hexanol.

## Confirming Ligand Binding Sites Through Mutagenesis

Using the fast mutagenesis system kit, we generated mutant AcerOBP11 by replacing amino acids Ile97, Ile140, and Phe101 with glycine (the corresponding primers listed in Table S3). All three AcerOBP11 mutant proteins were induced, purified, and confirmed by SDS-PAGE (**Figure 3B**, Figure S2). For the secondary structures, all three AcerOBP11 mutant proteins and wild-type showed the different degrees of protein characteristics by the confirmation of CD spectra (Figure S3). We performed competitive binding assays of the three mutant proteins with six candidate ligands to confirm the predicted role of these amino acids in binding to ligands. Compared with AcerOBP11-wt, the dissociation constant K<sup>D</sup> of the AcerOBP11m-Ile97 mutation significantly increased for n-hexanol, 4-hydroxyveratrol, isoamyl acetate, and farnesol (**Figure 8**, the detailed data can be seen from Table S4). Especially with n-hexanol, the K<sup>D</sup> of AcerOBP11m-Ile97 showed a significant 3.60-fold increase (p < 0.01, ANOVA). The K<sup>D</sup> of AcerOBP11m-Ile140 mutant bound to the six chosen ligands also increased, and the largest increase of K<sup>D</sup> was with isoamyl acetate (K<sup>D</sup> increased 1.8-fold, p < 0.01, ANOVA, **Figure 8**). However, we did not observe significant increases in of AcerOBP11m-Phe101 when bound to 4-hydroxyveratrol,

Frontiers in Physiology | www.frontiersin.org

## DISCUSSION

Social insects possess complex pheromone-driven behaviors that are regulated by chemical communication systems, regulating the social activities of the whole colony (Pankiw et al., 2004). Here, we cloned and functionally characterized a OBPs gene, AcerOBP11, from the antennae of A. cerana. It did not contain a signal peptide, and showed high similarity with homologous proteins in other insect OBPs (**Figure 1**). According to the sequence alignments and phylogenic tree analysis, it suggests that AcerOBP11 belongs to a typical odorant-binding protein family in A. cerana.

In A. mellifera, OBP11 is highly expressed in the antennae of forager workers and queens, and is not expressed in the egg, larval, and pupal stages (Forêt and Maleszka, 2006). AcerOBP11 showed high expression in pupae compared with eggs and larvae (**Figure 2A**, p < 0.01, t-test). Ligand-binding assay showed that AcerOBP11 could bind with some brood pheromone components (**Table 1**) that is released from larvae and sensed by nurse bees in the hive. It suggests that AcerOBP11 may be relevant to the synthesis and transportation of brood pheromones in these two stages. AcerOBP11 was also abundantly TABLE 1 | Fluorescence competitive assay of candidate ligands binding with recombinant AcerOBP11.


*(Continued)*

#### TABLE 1 | Continued


expressed in the antennae and wings of newborn workers, suggesting that AcerOBP11 may play a role in newly eclosed A. cerana.

In particular, AcerOBP11 was highly expressed in the antennae of worker bee at various ages (**Figure 2B**, p < 0.01, t-test), strongly indicating that it is involved in the olfactory behavior of worker bees. AcerOBP11 expression was higher in forager bees than nurse bees (**Figure 2B**), and this may be related to the behavioral activity for foraging honey and pollen. This expression results was almost consistent with the previous reports (Zhao et al., 2013a, 2016). Therefore, AcerOBP11 may be involved in the eclosion of worker bee and olfactory sensing functions during the forager stage.

AmelOBP11 was distributed only in sensilla basiconica at the top of antenna 3–10 segments (Kucharski et al., 2016). Our results are consistent with this finding that AcerOBP11 was localized to the antennal sensilla basiconica near the top of each segment in A. cerana (**Figures 4D,E**). Based on the external morphology of antennal sensilla of Apoidea, sensillar basiconica in bee antennae is likely involved in olfactory functions (Galvani et al., 2012). In the same Hymenoptera social insect, Camponotus japonicus, sensilla basiconica can recognize cuticular hydrocarbon (CH) pheromones to determine nest-mates and non-nest-mates (Ozaki et al., 2005). Furthermore, the GOBPs protein ASP2 in A. cerana is mainly expressed in sensilla placodea and plays a typical olfactory role in sensing general odors (Li et al., 2008). Overall, considering that AcerOBP11 was specially expressed in the sensilla basiconica (rather than sensilla placodea), it is likely that AcerOBP11 tends to the primary characteristics of bee

pheromones sensing and the secondary characteristics of insect GOBPs.

In all 23 candidate chemical pheromones and plant volatiles that were tested for AcerOBP11 binding in this study, we found that 12 bee pheromones bound to AcerOBP11 (except for methyl oleate). The queen mandibular pheromone (QMP) components HOB, 9-ODA, and HVA showed high affinity to AcerOBP11 with K<sup>D</sup> values < 10 µmol/L. HOB showed highest binding affinity to AcerOBP11 (K<sup>D</sup> = 1.35 µmol/L). 9-ODA is the typical bee sex pheromone that drones perceive, and 9-ODA is released by virgin queens to induce courtship and mating in males (Brockmann et al., 2006; Villar et al., 2015). HVA is a unique QMPs component in the western honey bee (Plettner et al., 1997), and also bound strongly to OBP11 here. The QMPs component can inhibit and regulate the ovary development of worker bees (Hoover et al., 2003; Peso et al., 2013), regulate programmed cell death in worker bee ovaries (Ronai et al., 2016), and affect activation of the worker bee ovary and ovarian duct (Ken et al., 2015). Therefore, the high affinity of AcerOBP11 with QMPs components suggests that AcerOBP11 may play an important role in the process of worker bees sensing the QMPs released by queen, and then affect the regulation of bee colony.

In addition, AcerOBP11 also bound strongly to brood pheromone components methyl palmitate and ethyl palmitate, rather than to another component methyl oleate (**Table 1**). Brood pheromones can inhibit ovarian development in worker bees (Arnold et al., 1994), stimulate and regulate pollen foraging activity of bees (Pankiw et al., 1998; Pankiw, 2004), and induce release of pheromones by the queen (Mohammedi et al., 1996). Moreover, AcerOBP11 strongly bound to alarm pheromones and worker pheromones (**Table 1**), suggesting that AcerOBP11 may play a role in worker bee behavior to maintain and defend the colony. Considering the high expression level of AcerOBP11 in nurse and forager bees, and the high affinity of AcerOBP11 with a variety of bee pheromones suggests that AcerOBP11 is an odorant-binding protein that can sense and regulate bee pheromones that are important to the A. cerana colony.

In addition to sensing bee pheromones, in this study, we found that AcerOBP11 can strongly bind with 11 plant volatiles (**Figure 5C**, **Table 1**). For example, β-ionone is a volatile produced in flowering plants (Li et al., 2013), and had higher binding affinity with AcerOBP11 (K<sup>D</sup> =3.78 µmol/L) than ASP1 (K<sup>D</sup> = 14.69 µmol/L) (Weng et al., 2013) and ASP2 (K<sup>D</sup> = 5.14 µmol/L) (Li et al., 2013). This suggests that AcerOBP11 may be involved in olfactory orientation for searching nectar sources, consistent with the results that AcerOBP11 is highly expressed in forager bee antennae (**Figure 2**). In conclusion, based on the integrated immunolocalization and functional studies of binding with bee pheromone and plant volatiles, AcerOBP11 was identified to play a dual-role that it had the primary characteristics of sensing various bee pheromones and secondary characteristics of sensing general odorants.

Molecular docking and site-directed mutagenesis can reveal amino acids that mediate ligand binding (Pelosi et al., 2014; Lu et al., 2015; Zhu et al., 2016). We identified Ile97, Ile140, and Phe101 are potential regulators of ligand binding, where Ile97 and Ile140 contributed hydrogen bonds to the ligand (**Figure 6**). Using AcerOBP11 mutant proteins, we found that binding affinity significantly decreased when Ile97 and Ile140 were mutated (**Figure 8**), suggesting that Ile97 and Ile140 may mediate AcerOBP11 binding with some ligands. In addition, we noticed that in the multiple sequence alignments the same positions of No. 97 and 140 of AcerOBP11 were always shown as Leu97/140 or Phe140 in other sequences, instead of Ile97/140 (**Figure 1A**). When we manually substituted Ile97/140 to Leu97/140 or Phe140, respectively, the Moldock scores and hydrogen bonds energies were shown by the analysis of docking (Figure S4). It was evidently that the energies of all AcerOBP11 wild-type were always the lowest, the mutants of m-Ile97/140Gly were the highest, whereas the other predicted mutants of m-Ile97Leu/Ile140Leu(Phe) had slightly higher energies than AcerOBP11 wild-type. It suggests that the alkane chain in isoleucines (Ile97/140) or leucines (Leu97/140) might play significant role in the interactions between AcerOBP11 and ligands.

Hydrogen bond is usually one of the most common forces that bind proteins with small molecules (Jiang et al., 2009; Zhuang et al., 2014; Li et al., 2016). In AcerOBP11, Ile140 contributes to the unique hydrogen with n-hexanol (**Figures 6**, **7**), while the hydrophobic amino acid Ile97 plays a major role rather than Ile140 according to the results of mutagenesis (**Figure 8**). These results suggest that hydrophobic interactions between AcerOBP11 and its ligands are critical for binding, especially for interactions between AcerOBP11 and n-hexanol, similar to findings from ASP2 and imidacloprid in A. cerana (Li et al., 2015), implying that complex interactions take place between olfactory proteins and compounds involved in the cognitive system of social insects.

## AUTHOR CONTRIBUTIONS

X-MS and H-LL: conceived and designed the experiments; X-MS and L-YZ: performed the experiments; X-BF, JT, and FW: analyzed the data; X-MS and H-LL: wrote the manuscript.

## ACKNOWLEDGMENTS

We are very grateful to three reviewers for giving us suggestive comments on our studies. We also thank Dr. Ping Wen from Xishuangbanna Botanical Garden, Chinese Academy of Sciences, for his help with 9-ODA and Dr. Bing-Hua Xie from Hangzhou Normal University, for her help with the fluorescence immunocytochemical localization. The authors acknowledge financial supports from the National Natural Science Foundation of China (No. 31772544, 31372254).

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00422/full#supplementary-material

## REFERENCES


Helicoverpa armigera (Hubner). Arch. Insect Biochem. Physiol. 57, 15–27. doi: 10.1002/arch.20009


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Song, Zhang, Fu, Wu, Tan and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Odorant Binding Protein 6 Expressed in Sensilla Chaetica Displays Preferential Binding Affinity to Host Plants Volatiles in Ectropis obliqua

Long Ma<sup>1</sup>† , Zhaoqun Li<sup>2</sup>† , Wanna Zhang<sup>3</sup> , Xiaoming Cai<sup>2</sup> , Zongxiu Luo<sup>2</sup> , Yongjun Zhang<sup>4</sup> \* and Zongmao Chen<sup>2</sup> \*

#### Edited by:

Peng He, Guizhou University, China

#### Reviewed by:

Hongliang Li, China Jiliang University, China Tiantao Zhang, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, China Patrizia Falabella, University of Basilicata, Italy

#### \*Correspondence:

Yongjun Zhang yjzhang@ippcaas.cn Zongmao Chen zmchen2006@163.com †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 10 February 2018 Accepted: 24 April 2018 Published: 17 May 2018

#### Citation:

Ma L, Li Z, Zhang W, Cai X, Luo Z, Zhang Y and Chen Z (2018) The Odorant Binding Protein 6 Expressed in Sensilla Chaetica Displays Preferential Binding Affinity to Host Plants Volatiles in Ectropis obliqua. Front. Physiol. 9:534. doi: 10.3389/fphys.2018.00534 <sup>1</sup> Jiangxi Key Laboratory of Bioprocess Engineering, College of Life Sciences, Jiangxi Science & Technology Normal University, Nanchang, China, <sup>2</sup> Tea Research Institute, Chinese Academy of Agricultural Sciences, Hangzhou, China, 3 Institute of Entomology, Jiangxi Agricultural University, Nanchang, China, <sup>4</sup> State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China

The monophagous tea geometrid Ectropis obliqua selectively feed on tea plants, requiring the specialized chemosensory system to forage for certain host. A deep insight into the molecular basis would accelerate the design of insect-behaviormodifying stimuli. In the present study, we focused on the odorant-binding protein 6 (EoblOBP6) with the high abundance in legs transcriptome of E. obliqua moths. qRT-PCR coupled with western blot analyses revealed the dual expression pattern of EoblOBP6 in antennae and legs. Cellular immunolocalization indicated that EoblOBP6 was predominantly labeled in the outer sensillum lymph of uniporous sensilla chaetica, which is not innervated by sensory neurons. No specific staining was observed in other sensillum types. The fluorescence competition assay showed a relatively narrow binding spectrum of recombinant EoblOBP6. EoblOBP6 could not only bind with intact tea plant volatiles benzaldehyde but also display high binding ability to nerolidol and α-farnesene which are tea plant volatiles dramatically induced by herbivore infestation. Besides, EoblOBP6 tightly bound to the aversive bitter alkaloid berberine. Taken together, EoblOBP6 displayed an unusual expression in sensilla chaetica, exhibited the potential involvement in olfaction and gustation, and may play a functional role in host location of female E. obliqua moths.

Keywords: Ectropis obliqua, odorant-binding protein, immunolocalization, sensilla chaetica, fluorescence competition assay

## INTRODUCTION

Moths have evolved a sophisticated olfactory system to detect various semiochemicals, guiding their feeding, mating, predator avoidance and oviposition behaviors. The hydrophobic odorant and taste molecules diffuse through pores in the sensillum surface and enter the sensillum lymph (Steinbrecht et al., 1995), after which they are delivered by carrier proteins to receptors

**100**

located within the dendritic membrane of sensory neurons (Pelosi, 1996). During this process, the high sensitive and selective insect olfaction depend heavily on two types of proteins, the carrier proteins and the olfactory receptors (ORs) (Große-Wilde et al., 2006; Benton et al., 2007; Forstner et al., 2009). Insect odorant binding proteins (OBPs) are small soluble carrier proteins (∼15 kDa) and are characterized by a specific domain that constitutes six α-helices joined by two-four disulphide bridges (Leal et al., 1999; Tegoni et al., 2004; Pelosi et al., 2014). Studies by in situ hybridization and immunolocalization have confirmed that OBPs are synthesized by the auxiliary cells surrounding neurons and are subsequently secreted into the sensillum lymph in a high concentration (Michael, 2000; de Santis et al., 2006). Involved in the initial steps of odorant reception, insect OBPs are presumed to bind, solubilize and transport the hydrophobic odorants through an aqueous lymph, and eventually reach sensory dendrites, where they activate the membrane-bound ORs (Pelosi et al., 2006).

Since the first identification of insect OBPs in the silkmoth Antheraea polyphemus where they bind with sex pheromones (Vogt and Riddiford, 1981), numerous OBPs have been investigated for their indispensable roles and potential involvement in olfaction (Berg and Ziegelberger, 1991; Xu et al., 2005; Biessmann et al., 2010; Wang et al., 2013). In Acyrthosiphon pisum, the repellent behavior to the alarm pheromone (E)-β-farnesene (EBF) was significantly impaired after dual knockdown of ApisOBP3 and ApisOBP7 which were known to bind EBF (Sun Y.F. et al., 2012; Zhang et al., 2017). Similarly, in Helicoverpa armigera and Chilo suppressalis, the CRISPR/Cas9 mediated pheromone binding proteins (PBPs) mutagenesis resulted in the severely impaired responses to sex pheromone components in male adults (Dong et al., 2017; Ye et al., 2017). Moreover, the behavioral assays in aphids (Qiao et al., 2009; Sun Y.F. et al., 2012) and Drosophila mutants (Matsuo et al., 2007; Swarup et al., 2011) also revealed that OBPs are truly engaged in the semiochemical perception. However, till now, the mode of action of these proteins remains incomplete. The Drosophila mutants lacking LUSH (OBP76a) were insensitive to their sex pheromone 11-cis-vaccenyl acetate (cVA), proving an indispensable role of LUSH in pheromone signal transduction (Xu et al., 2005). Likewise, LUSH is proved to be required for response to VA when VA receptors are expressed in non-T1 neurons (Ha and Smith, 2006). Laughlin et al. (2008) further concluded that LUSH bound to cVA forms an OBP-odorant complex that activates the pheromone-sensitive neuron. But later research showed that high concentration of pheromone can per se induce neuronal activity when devoid of LUSH, indicating that pheromone molecules alone directly activate its neuronal receptors (Gomez-Diaz et al., 2013). Besides, studies involving the combinations of PBP and pheromone receptors (PRs) from Chilo suppressalis and Bombyx mori indicate that PRs sensitivity to pheromones is greatly enhanced when co-expressed with PBPs (Syed et al., 2010; Chang et al., 2015). A reasonable explanation is that the presence of OBPs can increase the sensitivity of olfactory receptors to odorants (Große-Wilde et al., 2006; Sun et al., 2013; Chang et al., 2015).

Most insect OBPs are exclusively or dominantly expressed in antennae, and many studies have documented the different arrangement of OBPs in certain types of antennal sensilla (Steinbrecht et al., 1995; Shanbhag et al., 2001; Gu et al., 2013a). PBPs are positioned in lymph of sensilla trichodea and have high binding affinities with sex pheromone (Steinbrecht et al., 1995; Forstner et al., 2006; Große-Wilde et al., 2007; Forstner et al., 2009; Gu et al., 2013a), while general OBPs binding to plant volatiles are found in either sensilla basiconica or sensilla trichodea, or both (Hekmat-Scafe et al., 1997; Zhang et al., 2001; Wang et al., 2003; Sun et al., 2014). Moreover, some OBPs are expressed in leg, larval antenna, maxillary palp, mouthpart and proboscis (Bohbot and Vogt, 2005; Sengul and Tu, 2010; Sun et al., 2016; Zhu et al., 2016), even in the non-chemosensory organs, such as reproductive organs of male (Sun Y.L. et al., 2012; Ban et al., 2013) and female (Gu et al., 2013b; Zhang Y.N. et al., 2015). The diverse expression pattern indicates that the function of OBPs is more complicated than previously imagined, beyond the chemosensation. In Drosophila, OBP49a is expressed in the thecogen cells of the major taste organ labellum, interacts with bitter chemical, and is required for avoiding bitter-tasting compounds (Jeong et al., 2013); besides, OBP10 from Helicoverpa species, able to bind an insect repellent and highly enriched in seminal fluid, is delivered to females during mating and is finally located on shell of fertilized eggs (Sun Y.L. et al., 2012). Recent study shows that the mouthparts enriched OBP11 of the alfalfa plant bug exhibits a strong binding ability to non-volatile plant secondary metabolites, suggesting an involvement in feeding behavior (Sun et al., 2016). Members of OBPs that are found in non- olfactory organs are becoming an interesting aspect of function research.

The tea geometrid Ectropis obliqua Prout is one destructive defoliator of tea bushes in China, resulting in considerable economic losses. Given the healthy and environmental risks of chemical control against E. obliqua, safer alternatives based on insect-behavior-modifying stimuli are developed to manage this pest, such as synthetic pheromone lures and "push-pull" habitat management (Zhang Z. et al., 2015; Yang et al., 2016). Undoubtedly, deep insights into insect chemical communication could contribute to the design of pest repellents or attractants. For instance, in tortricid moth Epiphyas postvittana, the monoterpene citral recognized by OR3 elicits the notable repellent activity against the ovipositing female moths (Jordan et al., 2009). In our previous work, the ultrastructure of antennal and tarsal sensilla in E. obliqua moths was observed (Ma et al., 2016a). Subsequently, 24 OBP transcripts were identified from legs transcriptome of E. obliqua moths, of which EoblOBP6 showed the highest expression based on RPKM metric (Ma et al., 2016b; Zhang et al., 2018). Previously, many studies have documented the unusual distribution of insect OBPs in non-olfactory organs, but their physiological roles remain largely unknown. Here we focus on EoblOBP6, particularly for its high abundance and the dual expression pattern in antennae and legs. In this work, the specific sensillum location of EoblOBP6 is investigated by cellular immunolocalization, and the ligand-binding specificity of EoblOBP6 to host volatiles, non-host plant volatiles, herbivoreinduced volatiles, plant secondary metabolites and tastants are further measured.

## MATERIALS AND METHODS

fphys-09-00534 May 17, 2018 Time: 12:26 # 3

#### Insect Rearing and Tissue Collection

Adult E. obliqua were originally collected from the experimental tea plantation of the Tea Research Institute, Chinese Academy of Agricultural Sciences (Hangzhou, Zhejiang, China). The laboratory colony was reared on fresh tea shoots in enclosed nylon mesh cages and maintained in controlled environment of 25 ± 1 ◦C and 70 ± 5% relative humidity under a photoperiod of 14-h light: 10-h dark. After pupation, female and male individuals were kept separately until eclosion. After emergence, moths were supplied with 10% honey solution. Different tissues from E. obliqua adults of both sexes including antennae, stylets, heads (without antennae), thoraxes, abdomens, legs and wings were sampled for both RT-PCR and western blot analysis. Three biological pools were prepared, and all samples were frozen immediately and stored in −80◦C.

#### RNA Extraction and cDNA Synthesis

Total RNA of each sample was extracted by using Trizol reagent (Invitrogen, Carlsbad, CA, United States). The integrity and purity of extracted total RNA was examined with 1.0% agarose electrophoresis, and RNA quantity was determined using a spectrophotometer NanoDropTM (NanoDrop Inc., Wilmington, DE, United States). A FastQuant RT-kit with gDNA Eraser (TianGen, Beijing, China) was employed to synthesize the first-strand cDNA using 2 µg RNA.

#### qRT-PCR Analysis and RT-PCR Verification

The expression profiles of EoblOBP6 (Accession No. ALS03854.1) in different tissues were determined by RT-PCR (Supplementary Table S1). Each PCR reaction contained 200 ng cDNA template, performed by Taq Master Mix (CWBIO, Beijing, China) under a general 3 step amplification by 33 cycles of 94◦C for 20 s, 58◦C for 20 s, 72◦C for 40 s. PCR products were checked by electrophoresis and further confirmed by sequencing. The β-actin gene (Accession No. KT860051) was served as an endogenous control. Each reaction was performed three times with different biological samples.

The relative expressions of EoblOBP6 among tissues were measured by qRT-PCR on a Bio-Rad CFX96 touch real-time PCR detection system. Two reference gene, β-actin and GAPDH (Accession No. KT991373), were employed to normalize the target EoblOBP6 expression and to rectify the sample-to-sample discrepancy. To determine the amplification efficiencies of the reference and target genes, the efficiency of each primer pair was measured by constructing a standard curve with serial template dilutions. The standard curves created regression line with slopes ranging from −3.4 to −3.3, and the amplification efficiency of target gene was approximate to that of the reference genes. The qRT-PCR reaction using SuperReal PreMix Plus (TianGen, Beijing, China) was performed as previously reported (Ma et al., 2016b). The relative transcript level was calculated by the comparative 2−11CT method.

#### Recombinant Protein Expression and Purification

The open reading frame of EoblOBP6 was amplified for the construction of recombinant expression vector. PCR reaction was performed as follows: initial denaturation at 95◦C for 2 min, followed by 35 cycles of 94◦C for 20 s, 58◦C for 30 s and 72◦C for 30 s, and a final elongation step at 72◦C for 10 min. The correct product confirmed by sequencing was subcloned into the bacterial expression vector pET32a(+). The recombinant plasmid was then transformed into Escherichia coli BL21 (DE3) cells. The recombinant protein was induced at 37◦C for 6 h with 1 mM isopropyl ß-D-1-thiogalactopyranoside (IPTG). The protein was purified by two rounds of Ni2<sup>+</sup> ion affinity chromatography with gradient concentration imidazole washing, and the His-tag was excised using recombinant enterokinase (Novagen, Madison, WI, United States). The purified protein was desalted through extensive dialysis, and the size and purity of the recombinant protein were confirmed by 15% SDS-PAGE analysis.

#### Western Blot Analysis

The polyclonal antibody against EoblOBP6 was produced by injecting adult rabbits subcutaneously with the purified recombinant protein. The immunized rabbits were reared individually in comfortable cages, and all procedures were operated conforming to the ethical guidelines to minimize pain and discomfort to the animals. The serum was purified using a MAb trap kit (GE Healthcare).

The tissue was homogenized in lysis buffer (8 M urea, 4% chaps, 40 mM Tris–HCl, 5 mM EDTA, 1 mM PMSF and 10 mM DTT, pH 8.0) containing a mixture of protease inhibitors (Roche, Switzerland). Crude protein extracts from adult tissues, including female legs, male legs, female antennae and male antennae, were quantified by a Bio-Rad protein assay with bovine serum albumin (BSA) as standard, and then diluted to obtain equal amounts of the total proteins. After separation by 15% (w/v) SDS-PAGE, samples were transferred onto nitrocellulose (NC) membrane blotting filters at 100 V for 1 h at 4◦C. Membranes were then blocked with 5% (w/v) skimmed milk in PBST at 4◦C overnight. After washing thrice with PBST, the blocked the membrane was incubated with β-actin antibody (1:2000 dilution) and EoblOBP6 antibody (1:4000 dilution) for 1 h at room temperature, separately. After three washes with PBST, the membrane was incubated for 1 h at room temperature with goat anti-rabbit IgG HRP-linked secondary antibody (Sigma, St. Louis, MO, United States) at 1:10,000 dilution with PBST. The immunoreactivity was visualized using an enhanced electrochemiluminescence detection kit (TransGen, Beijing, China) and photographed by Image Quant LAS4000 mini (GE-Healthcare, Germany). Additionally, western blot analysis was also performed to examine the specificity of the antibody using the purified EoblOBP6 protein.

#### Immunocytochemical Localization

The foreleg tarsus of female adult, and intact antennae detached from male and female adults were prefixed in a mixture of paraformaldehyde (4%) and glutaraldehyde (2%) in 0.1 M PBS (pH 7.4) for 24 h at room temperature, dehydrated in an ethanol series, and then embedded in Luria-Bertani white resin (Taab, Aldermaston, United Kingdom) for polymerization at 60◦C. Ultrathin sections (60 nm) were cut by a diamond knife on a Reichert Ultracut ultramicrotome (Reichert Co., Vienna, Austria). For immunostaining, the grids were floated in droplets of PBS (containing 50 mM glycine), followed by PBGT (PBS containing 0.2% gelatine, 1% bovine serum albumin, and 0.02% Tween-20). The grids were then incubated with EoblOBP6 antiserum (diluted at 1:3000) at 4◦C overnight. After rinsing six times in PBGT, the grids were incubated with secondary antibody (anti-rabbit IgG) coupled with 10 nm colloidal gold granules (Sigma) (diluted at 1:20) for 90 min at room temperature. The grids were then transferred to silver intensification and stained with 2% uranyl acetate to increase the contrast. Finally, sections were observed with HITACHI H-7500 TEM (Hitachi Ltd). The serum supernatant from an uninjected rabbit was used as the negative control.

#### Fluorescence Competitive Binding Assays

For the ligand binding assays, the tested compounds, including terpenoids, tea volatiles, herbivore-induced plant volatiles and non-volatile tastants, were selected according to the previously reported isolation from the E. obliqua host plant and non-host plant (Sun X.L. et al., 2014; Zhang Z. et al., 2015). Fluorescence binding assays were performed on a fluorescence spectrophotometer F-380 (Tianjin, China) with a 1 cm light path quartz cuvette and 10 nm slits for excitation and emission. The excitation wavelength was set at 337 nm, and the emission spectrum was recorded between 390 and 500 nm. Both the fluorescent probe N-phenyl-1-naphthylamine (1-NPN) and the tested chemicals were dissolved in methanol in preparation for 1 mM stock solution. To determine the dissociation constant of EoblOBP6 with 1-NPN, 2 µM protein solution in 50 mM Tris-HCl (pH 7.4) was titrated with aliquots of 1 mM 1-NPN solution to final concentrations ranging from 1 to 16 µM. Then the affinities of ligands were tested by competitive binding assays through titrating the chemical competitor from 2 to 30 µM into the 1-NPN and EoblOBP6 mixed solution (both at 2 µM). The fluorescence intensities at the maximum fluorescence emission were plotted against the free ligand concentration to determine the binding constants. The bound ligand was evaluated from the fluorescence intensity in the assumption of the protein was 100% dynamic, with a stoichiometry of 1:1 (protein: ligand) at saturation. The binding curves were linearized using Scatchard Plot. The dissociation constants of competitors were calculated from the corresponding IC<sup>50</sup> values following the equation: Ki = (IC50)/(1+(1-NPN)/K1−NPN), where (1-NPN) is the free concentration of 1-NPN and K1−NPN is the dissociation constant of the protein/1-NPN complex.

## Homology Modeling and Phylogenetic Analysis

The SWISS-MODEL workspace<sup>1</sup> (Biasini et al., 2014) was employed to search for the most suitable template to build the 3D structure. Because of the high global quality estimation score (GMQE) with EoblOBP6, the template structure of Bombyx Mori GOBP2 was selected for the homology modeling by means of automatic mode. A Ramachandran plot was employed to evaluate the rationality of the established model. The secondary structure was predicted by ESPript 3.0 program (Robert and Gouet, 2014) based on the constructed 3D model and the aligned sequences. The 158 OBP sequences from Lepidoptera species were selected for elucidating the evolutionary history (Supplementary Table S3). The phylogenetic tree was constructed by MEGA 6.0 using the Neighbor-joining mode with a p-distance model and a pairwise deletion of gaps. Bootstrap support was assessed by a boot strap procedure based on 1000 replicates.

## RESULTS

## Tissue Expression Pattern of EoblOBP6

The RT-PCR results indicated that EoblOBP6 was clearly detected in both antennae and legs of adults in both sexes, whereas a plain band was also observed in stylets, abdomen and wings in both sexes (**Figure 1A**). The relative expression was further confirmed by qRT-PCR measurement. The results revealed that EoblOBP6 transcripts were abundantly transcribed in tissues of antennae and legs, followed by stylets. EoblOBP6 was weakly expressed in wings, abdomen and heads. Besides, higher transcripts abundance was detected in female antennae than that in male antennae (**Figure 1B**). Meanwhile, western blot analysis confirmed EoblOBP6 protein was distributed in adult antennae and legs (**Figure 1C**).

## Expression and Purification of EoblOBP6

The recombinant protein of EoblOBP6 was successfully expressed in a bacterial expression system and purified twice using Ni2<sup>+</sup> ion affinity chromatography, followed by excision of the His-tag with enterokinase. The SDS-PAGE analysis showed the highly purified protein as a single band with the molecular weight of approximately 14 kDa (**Figure 2**), consistent with the predicted molecular mass.

## Specific Localization of EoblOBP6 in Sensilla Chaetica

The polyclonal antiserum against recombinant EoblOBP6 protein was prepared to investigate the cellular immunolocalization in distinct sensilla of adult antennae and foreleg tarsomere according to the previous elucidation of sensillum ultrastructures (Ma et al., 2016a). First, the specificity of antiserum was confirmed by western blot analysis, and EoblOBP6 antibody could reacted specifically with EoblOBP6 protein (**Figure 2**). The immunostaining of EoblOBP6 in

<sup>1</sup>https://swissmodel.expasy.org/

EoblOBP6 cleaved His-tag by rTEV protease; Lane 7, western blot analysis of the purified EoblOBP6 using polyclonal rabbit antiserum.

antennal sensilla indicated that EoblOBP6 was predominantly labeled in the large outer sensillum lymph of sensilla chaetica, which is not innervated by sensory neurons. Although the crescent-shaped outer sensillum lumen was heavily labeled, the inner dendritic cytoplasm and the cuticle of the hair wall showed more than few unspecific gold spots (**Figures 3F–I**). Both crosswise and longitudinal sections indicated that the sensillum lymph of sensilla chaetica was intensely stained by the anti-EoblOBP6 antiserum, and the fierce immunolabeling was detected in the top sections. No obvious staining was observed in either sensilla trichodea or sensilla basiconica, neither in sensilla auricillica (**Figures 3A–E**).

Moreover, the microscopy of E. obliqua moths revealed the distribution of setae and sensilla chaetica in the ventral side of foreleg fifth tarsomere (**Figure 4A**). The seta had a thick sensillum wall with no pores. Results of the immunostaining showed that anti-EoblOBP6 antibody specifically labeled the outer sensillum lymph of sensilla chaetica, which housed the receptor cell dendrites. And the fierce immunolabeling was observed to encircle the inner sensillum lumen. However, no obvious staining was detected in the inner sensillum lumen where several neuronal dendrites reside (**Figure 4**).

## Ligand Binding Assays of EoblOBP6

In preparation for the ligand binding assay, the binding affinity of the fluorescent probe 1-NPN with the purified EoblOBP6 was first measured (**Figure 5**). Results revealed that EoblOBP6 was capable of binding 1-NPN with binding affinity of 2.70 ± 0.24 µM. Subsequently, the binding properties of EoblOBP6 to the selected host compounds from different functional groups were measured, and the results indicated that EoblOBP6 displayed a relatively narrow binding spectrum

few unspecific gold grains (cross sections F,G,H and longitudinal sections I). w, sensillum wall; p, pores.

FIGURE 4 | Immunolabeling of EoblOBP6 in types of sensilla present on E. obliqua moth fifth tarsomere. Black spots (arrow indicates the location) represent the immunostained EoblOBP6 protein. The sensilla chaetica (Sch) and the mechano-sensitive setae were observed on foreleg fifth tarsomere (A). EoblOBP6 was not stained in either wall or lumen in seta (B,C). Longitudinal sections of sensilla chaetica revealed the strong labeling of EoblOBP6 in sensilla cavity beneath the cuticle (D). Basic section of sensilla chaetica indicated the staining of EoblOBP6 in sensilla lumen but not the sensilla wall (E). Heavy labeling of anti-EoblOBP6 antibody (black spots) was specifically present in the crescent-shaped outer sensillum lymph (osl) which are devoid of the receptor-cell dendrites; the innervated inner sensillum lumen (isl) showed few unspecific gold grains (F,G). w, sensillum wall.

FIGURE 5 | Fluorescence competitive binding assay of E. obliqua odorant-binding protein 6 (EoblOBP6). (A) Binding curve and relative Scatchard plot for 1-NPN and EoblOBP6. (B) Competitive binding curves of the active volatiles and tastants to EoblOBP6.

fphys-09-00534 May 17, 2018 Time: 12:26 # 6



(**Figure 5** and **Table 1**). Of the 52 tested compounds, only five odorants and one tastant exhibited strong binding abilities to EoblOBP6 (Supplementary Table S2). For the non-host volatiles, two terpenoids, α-caryophyllene and α-terpinene, showed binding affinity to EoblOBP6, with dissociation constants of 15.55 and 18.31 µM, respectively. Besides, the majority of host volatiles, including (Z)-3-hexenol, decanal, 1-hexanol and hexyl acetate, could hardly bind to the recombinant protein, except for benzaldehyde (Ki = 15.08 µM). Interestingly, nerolidol and α-farnesene, volatiles dramatically induced by the herbivore infestation (Sun X.L. et al., 2014), exhibited high binding affinities with EoblOBP6 of 10.87 and 11.02 µM, respectively. For the non-volatile tastants, EoblOBP6 could only bind strongly to berberine (**Table 2**).


#### Homology Modeling

The SWISS-MODEL workspace was employed to search for the structural template. The GOBP1 from Bombyx Mori (template library identity: 2wc5.1) shared 31% homology with EoblOBP6 and gained global quality estimation score (GMQE) of 0.59, and thus was chosen as the template for homology modeling. The result of Ramachandran plot showed that 94.3% of the residues were in preferred regions, 4.9% of the residues were in the allowed region and 1 residue was identified as an outlier (Supplementary Figure S2), suggesting that the predicted model is generally reliable. The predicted 3D structure of EoblOBP6 was composed of six α-helices between residues Glu4-Leu15 (α1), Ala19-His25 (α2), Ile44-Lys54 (α3), Pro67-His74 (α4), Ala81-Ser96 (α5) and Gly108-Ile125 (α6), forming an α-helix-enriched globular protein. Three pairs of disulphide bridges connecting Cys21 in α2 and Cys51 in α3, Cys47 in α3 and Cys109 in α6, Cys94 in α5 and Cys118 in α6 contributed to the stability of the tertiary structure and the formation of α-helixes (**Figure 6**).

To deduce the evolutionary relationships and underlying functions, 158 lepidopteran OBPs from six species were chosen for phylogenetic tree construction (Supplementary Figure S1). The results revealed a divergent OBP repertoire, and EoblOBP6, EoblOBP22, HarmOBP4 and SexiOBP3 clustered into a same clade. Multiple alignment showed EoblOBP6 shared 46, 40, 32% identity to EoblOBP22, HarmOBP4 and SexiOBP3, separately (Supplementary Figure S1). Overall, these results indicated a specific evolutionary status of EoblOBP6 different from the other lepidopteran OBPs.

#### DISCUSSION

In present study, we reveal that EoblOBP6 possesses a dual expression pattern in adult antennae and legs in both sexes, and it is predominantly expressed in the outer sensillum lymph of the uniporous sensilla chaetica. This unique distribution pattern arouses great interest owing to that sensillum chaetica is generally considered as the typical mechano-sensitive sensillum.

Several studies have documented the expression of OBPs beyond the olfactory organs, and their physiological functions would be more complicated. Herein we intend to investigate the potential involvement of EoblOBP6 in gustatory and olfactory sensation.

The tea geometrid, Ectropis obliqua, is one lepidopteran pest feeding exclusively on tea leaves and tender buds. The female moths possess a remarkable capability to locate suitable host plants which is fundamental to the survival of their offspring, because the young larvae cannot easily forage for alternative hosts (Ryuda et al., 2013). Host plant selection by herbivorous insects involves searching, landing, contact evaluation, and a final decision for acceptance or rejection (Schoonhoven et al., 2005). Contact chemosensilla play a dominant part in detecting phytochemical compounds after landing on plant, which allow insects to perceive the compounds on/in the surface of leaves and flowers (Chapman, 2003; Calas et al., 2007; Newland and Yates, 2008; Zhang et al., 2010). Typically, insect contact chemoreceptors are derived from mechanosensory bristles and are mainly scattered on tarsi, ovipositor, mouthparts, and antennae (Chapman, 2003). Previous study in E. obliqua

has documented that the arrangement of uniporous sensilla chaetica comprised the majority of chemosensilla in tarsi, and are presumed to be responsible for gustatory cognition (Ma et al., 2016b). Actually, many studies have documented that the arrangement of lepidopteran tarsal chemosensilla are responsible for tastant recognition. In butterflies, including Papilio xuthus, Papilio polytes, and Heliconius melpomene, female butterflies recognize the oviposition stimulants by the contact chemosensilla distributed on the ventral side of their foreleg tarsus (Nakayama et al., 2003; Briscoe et al., 2013); while in moths of H. armigera, Mnesampela privata, and Lobesia botrana, chemosensilla sensitive to sugars or amino acids are situated on the ventral surface of the fifth tarsomere (Calas et al., 2006, 2009; Zhang et al., 2010). In general, contact chemoreceptors respond to chemicals of low-volatility, and have a single pore at the distal tip through which chemicals gain access to several sensory neurons.

Our cellular immunolocalization reveals that EoblOBP6 is strongly labeled in the outer sensillum lumen of the contact sensilla chaetica in fifth tarsomere. This remarkable localization pattern suggests EoblOBP6 may function as a carrier to enable the hydrophobic molecules from the outer sensillum-lymph cavity to reach the dendritic membranes in the inner cavity. Moreover, this cellular localization pattern is consistent with the report of the putative OBP PBPRP2 in Drosophila that PBPRP2 is expressed in the outer sensilla lumen of taste sensilla, rather than the lumen where the dendrites of the gustatory neurons reside (Shanbhag et al., 2001). It is commonly accepted that sensilla trichodea is sensitive to pheromone, and the non-labeling of EoblOBP6 in sensilla trichodea may due to the absence of EoblOBP6 in pheromone detection. Recent studies have proposed that OBPs expressed in gustatory organs get involved in gustatory coding. In Drosophila, OBP49a enriched in labella is indispensable for perceiving the bitter substances, and OBP49a specifically interacted with bitter chemicals, including berberine, denatonium and quinine (Jeong et al., 2013); two OBPs encoded by OBP57e and OBP57d expressed in chemosensory hairs of tarsus are implicated in taste perception as well as the host–plant preference (Matsuo et al., 2007). Actually, the non-volatile plant metabolites are comparable to odors in the way that they are both small hydrophobic molecules, therefore, it is reasonable to conclude that OBPs act as carrier of such type of poorly water-soluble molecule to gustatory receptors, similar to their performance in olfaction. Our results from the ligand binding assay reveal that EoblOBP6 specifically binds to the alkaloid berberine, which is an aversive bitter stimuli to insect (Pontes et al., 2014). In Drosophila, OBP28a abundant in proboscis functions as a transporter of bitter tastants to gustatory receptors, modulating the sugar intake in response to bitter tastants (Swarup et al., 2014). Yet unfortunately, there is no direct evidence supporting that the sensilla chaetica of E. obliqua respond to bitter substances. Generally, the presence of bitter compounds, an indication of toxicity, is reported by taste organs. Evaluation of these tastants informs the decision as to whether to accept a host plant as food source or oviposition site, and we presume the underlying participation of EoblOBP6 in this process. In such a scenario, EoblOBP6 present in sensilla chaetica may act as a carrier for hydrophobic bitter compounds.

The tissue-biased distributions of OBPs in insect are indicative of biological function. Results from both qRT-PCR and western blot analysis indicate that EoblOBP6 possesses a dual expression pattern in adult antennae and legs from both sexes. In general, an antenna-abundant expression correlates tightly with olfactory sensation, while the abundance in gustatory organs indicates an involvement in taste detection. The fluorescence competition assay provides further insight into the physiological roles of EoblOBP6. The results show that EoblOBP6 displays a strong binding to nerolidol and α-farnesene, both of which are tea plant volatiles dramatically induced by herbivore infestation. These herbivore-associated plant volatiles are closely associated with the host-search behavior of herbivores. Actually, female E. obliqua moths are more attracted by the infested tea plants and preferentially oviposit on these plants, in order to reduce the predation by the natural enemies (Sun X.L. et al., 2014). Besides, benzaldehyde emitted from the intact tea leaves has a relative high binding affinity with EoblOBP6 (Maeda et al., 2006); α-terpinene, a type of terpenoid which is mainly emitted from aromatic plants and elicits strong electrophysiological responses from the antennae of E. obliqua (Zhang Z. et al., 2015), shows binding affinity to EoblOBP6. Overall, our results propose that EoblOBP6 is a general OBP that selectively binds to odors of host plant source and may play an important part in host location of female E. obliqua moths.

Taken together, this study reports the identification of EoblOBP6 expressed in sensilla chaetica of both antennae and tarsus, and EoblOBP6 preferentially binds to the herbivore-induced plant volatiles, host plant volatiles and plant secondary compound. These results indicate the potential involvement of EoblOBP6 in olfactory and gustatory coding, playing a functional role in host location. Given the great economic impact of E. obliqua, a deep insight into their chemosensory system would accelerate the development of insect-behavior-modifying stimuli. Further investigations by RNAi or CRISPR/Cas9 editing to establish the EoblOBP6 targeted mutagenesis would be performed in functional study.

#### AUTHOR CONTRIBUTIONS

The experimental plan conceived and designed by LM and WZ. The experiments performed by LM, WZ, and ZL. The data processed and analyzed by LM, WZ, and ZL. Wrote and edited the manuscripts by LM, WZ, XC, ZL, YZ, and ZC.

#### FUNDING

This work was supported by Open Foundation of State Key Laboratory for Biology of Plant Diseases and Insect Pests (SKLOF201517), Science and Technology Program of Department of Education of Jiangxi Province (GJJ170660), Natural Science Foundation of Jiangxi Province (20171BAB214028, 20171BAB214004), National Natural Science Foundation of China (31601892), and Zhejiang Provincial Natural Science Foundation of China (LY16C140003).

#### REFERENCES

fphys-09-00534 May 17, 2018 Time: 12:26 # 10


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00534/full#supplementary-material


Pelosi, P. (1996). Perireceptor events in olfaction. J. Neurobiol. 30, 3–19.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer TZ declared a shared affiliation, with no collaboration, with one of the authors, YZ, to the handling Editor.

Copyright © 2018 Ma, Li, Zhang, Cai, Luo, Zhang and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Binding Specificity of Two PBPs in the Yellow Peach Moth Conogethes punctiferalis (Guenée)

Xing Ge1,2, Tofael Ahmed<sup>3</sup> , Tiantao Zhang<sup>1</sup> \*, Zhenying Wang<sup>1</sup> \*, Kanglai He<sup>1</sup> and Shuxiong Bai <sup>1</sup>

<sup>1</sup> State Key Laboratory for Biology of Plant Disease and Insect Pest, Institute of Plant Protection, Chinese Academy of Agricultural Science, Beijing, China, <sup>2</sup> Department of Plant Protection, Henan Institute of Science and Technology, Xinxiang, China, <sup>3</sup> Bangladesh Sugarcrop Research Institute, Pabna, Bangladesh

#### Edited by:

Nicolas Durand, Université Pierre et Marie Curie, France

#### Reviewed by:

Guan-Heng Zhu, University of Kentucky, United States Loic Briand, UMR6265 Centre des Sciences du Goût et de l'Alimentation (CSGA), France Herbert Venthur, Universidad de La Frontera, Chile

#### \*Correspondence:

Tiantao Zhang zhtiantao@163.com Zhenying Wang zywang@ippcaas.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 23 October 2017 Accepted: 14 March 2018 Published: 03 April 2018

#### Citation:

Ge X, Ahmed T, Zhang T, Wang Z, He K and Bai S (2018) Binding Specificity of Two PBPs in the Yellow Peach Moth Conogethes punctiferalis (Guenée). Front. Physiol. 9:308. doi: 10.3389/fphys.2018.00308 Pheromone binding proteins (PBPs) play an important role in olfaction of insects by transporting sex pheromones across the sensillum lymph to odorant receptors. To obtain a better understanding of the molecular basis between PBPs and semiochemicals, we have cloned, expressed, and purified two PBPs (CpunPBP2 and CpunPBP5) from the antennae of Conogethes punctiferalis. Fluorescence competitive binding assays were used to investigate binding affinities of CpunPBP2 and CpunPBP5 to sex pheromone and volatiles. Results indicate both CpunPBP2 and CpunPBP5 bind sex pheromones E10-16:Ald, Z10-16:Ald and hexadecanal with higher affinities. In addition, CpunPBP2 and CpunPBP5 also could bind some odorants, such as 1-tetradecanol, trans-caryopyllene, farnesene, and β-farnesene. Homology modeling to predict 3D structure and molecular docking to predict key binding sites were used, to better understand interactions of CpunPBP2 and CpunPBP5 with sex pheromones E10-16:Ald and Z10-16:Ald. According to the results, Phe9, Phe33, Ser53, and Phe115 were key binding sites predicted for CpunPBP2, as were Ser9, Phe12, Val115, and Arg120 for CpunPBP5. Binding affinities of four mutants of CpunPBP2 and four mutants of CpunPBP5 with the two sex pheromones were investigated by fluorescence competitive binding assays. Results indicate that single nucleotides mutation may affect interactions between PBPs and sex pheromones. Expression levels of CpunPBP2 and CpunPBP5 in different tissues were evaluated using qPCR. Results show that CpunPBP2 and CpunPBP5 were largely amplified in the antennae, with low expression levels in other tissues. CpunPBP2 was expressed mainly in male antennae, whereas CpunPBP5 was expressed mainly in female antennae. These results provide new insights into understanding the recognition between PBPs and ligands.

Keywords: pheromone binding proteins, Conogethes punctiferalis, fluorescence competitive binding assays, molecular docking, qPCR

## INTRODUCTION

Insects depend on a well-developed olfactory system to distinguish odorants and sex pheromones. Odorant binding proteins (OBPs), chemosensory proteins (CSPs), odorant receptors (ORs) and odorant degrading enzymes (ODEs) are involved in the selectivity and sensitivity of olfaction (Leal, 2005, 2013; Fan et al., 2011; Ahmed et al., 2014, 2017). OBPs are small, water-soluble proteins

**111**

identified from the chemosensory organs, that are carriers between the external environment and chemoreceptors (Ishida et al., 2002; Leal, 2013). As a multi-genes family, OBPs usually are divided into PBPs, general odorant binding proteins (GOBPs) and antennal binding proteins (ABPs) in lepidopteran insects, based on their binding affinity with sex pheromone and odorant molecules (Vogt et al., 1991; Krieger et al., 1996). Actually, GOBPs and ABPs in many insect species also play roles in pheromone detection, because some of them were found to be expressed in long trichoid sensilla, which are known as pheromone-sensitive sensilla, and most of the main contributors to the ligand binding pocket are conserved (Feng and Prestwich, 1997; Maibeche-Coisne et al., 1998; Zhou et al., 2009; He et al., 2010; Liu et al., 2012). Surprisingly, some GOBP have higher binding affinities with sex pheromone than PBP (Zhou et al., 2009; Liu et al., 2012). PBPs are thought to bind and transport hydrophobic sex pheromone molecules across the aqueous sensillum-lymph to specific pheromone receptors on the dendritic membrane of olfactory neurons (Vogt and Riddiford, 1981; Leal et al., 2005; Forstner et al., 2006; Pelosi et al., 2006). In the earlier studies, PBPs are considered mostly male-specific, while other OBPs are expressed in both males and females (Pelosi et al., 2006). As the first step of pheromone recognition, when PBPs bind to different components of sex pheromones, they can lead to species specificity (Willett and Harrison, 1999).

So far, the 3D structure of PBPs in Bombyx mori (Sandler et al., 2000; Horst et al., 2001), Antheraea polyphemus (Mohanty et al., 2004), Leucophaea maderae (Lartigue et al., 2003), Amyelois transitella (Xu et al., 2010; di Luccio et al., 2013), Apis mellifera (Lartigue et al., 2004) have been elucidated both alone and in combination with various ligands. The structure of B. mori PBP (BmorPBP) with bombykol was the first to be studied by X-ray diffraction spectroscopy and nuclear magnetic resonance (NMR) techniques (Sandler et al., 2000; Horst et al., 2001). The binding pocket of BmorPBP was formed by four antiparallel helices (α1, α4, α5, and α6; Sandler et al., 2000), and the conformational transition in solution displayed pH-dependence (Horst et al., 2001). Stability of protein and ligands are maintained by amino acid residues. Some of these residues are critical for binding ligands (Sandler et al., 2000; Mohanty et al., 2004; Thode et al., 2008; Jiang et al., 2009; Yin et al., 2015; Tian and Zhang, 2016; Zhu et al., 2016; Ahmed et al., 2017; Zhang et al., 2017). Of the residues in BmorPBP, Met5, Phe12, Phe36, Trp37, Ile52, Ser56, Phe76, Val94, Glu98, Ala115, and Phe118 are more conserved and involved in binding to bombykol, which suggests they are interacting with ligands (Sandler et al., 2000; Klusák et al., 2003). Thr57, Ser52 and Thr48 in Drosophila melanogaster LUSH are involved in the binding of short-chain n-alcohols. Thr57 mutants had a significant decrease in ability to bind alcohol compounds compared with wild type, which indicates Thr57 is the key site of LUSH binding to small alcohol molecules (Kruse et al., 2003; Thode et al., 2008).

Insect pheromones play an important role in intra-species communication, sexual attraction, mating aggregation and oviposition host-marking. In many moth species, sex pheromones are usually blends of chemical compounds. Airborne pheromones of moths often consist of two or three chemical components, each of which is perceived by specific olfactory receptor neurons (Abraham et al., 2005).

The yellow peach moth, Conogethes punctiferalis (Guenée; Lepidoptera: Crambidae), is an important agricultural pest of peach, apple, chestnut, maize, and sorghum (Luo and Honda, 2015; Ge et al., 2016). The main sex pheromone compounds of yellow peach moth are (E)-10-hexadecenal (E10-16:Ald), along with the two minor components (Z)-10-hexadecenal (Z10-16:Ald) and hexadecenal (16:Ald; Konno et al., 1982; Liu et al., 1994; Kyungsaeng and Park, 2005). Field trials indicate that Z10-16:Ald and 16:Ald alone do not attract males. A blend of these compounds (two or three) was more attractive (Liu et al., 1994). A better understanding of the molecular mechanisms of sex pheromone perception would improve the use of pheromones to control this pest. In this study, two PBP genes, CpunPBP2 and CpunPBP5, which were identified as pheromone binding proteins, are cloned in the antennae of C. punctiferalis and successfully expressed in Escherichia coli. In order to better understand the function of these PBPs, fluorescence displacement binding assays of CpunPBP2 and CpunPBP5 and their mutants are carried out with sex pheromone components.

#### MATERIALS AND METHODS

#### Insects Rearing

C. punctiferalis larvae were collected from the sunflower Helianthus annuus at Langfang Experimental Station of Chinese Academy of Agricultural Sciences, Hebei Province, China, and reared on fresh maize in an environmentally controlled room. Rearing conditions were 27 ± 1 ◦C, 70–80% relative humidity (RH) and 16:8 light: dark (L:D). Adults were provided with 10% honey solution. After eclosion, the antennae from males and females (80 pairs of each sex) were immediately cut and processed for RNA extraction.

## RNA Extraction and Reverse Transcription

Total RNA was isolated from the antennae using Trizol Reagent (Invitrogen, Carlsbad, CA, USA) following manufacturer's recommendations. The integrity of total RNA was assessed with 1.2% agarose gel electrophoresis and the concentration was determined on a NanoDrop 2000 spectrophotometer (Thermo, USA). One µg RNA was added for reverse transcription to cDNA according to product kit instructions (TransGen, Beijing, China).

#### Cloning and Sequencing

CpunPBP2 (GenBank accession number: GEDO010000019.1; Jia et al., 2016) and CpunPBP5 (GenBank accession number KP985227) of C. punctiferalis were obtained from the antennal cDNA library. The primers were designed to clone the coding region of CpunPBP2 and CpunPBP5 (**Table S1**; Underlined bases show restriction enzyme sites for forward and reverse primers, respectively). PCR products were separated by electrophoresis on 1% agarose gels in 1 × TAE buffer. Then the specific fragments were cut and purified by DNA gel extraction kit (Axygen, Hangzhou, China) following the manufacturer's protocol. The purified products were cloned into pGEM-T easy vector

(TransGen, Beijing, China) and then transformed to TransT1 E. coli competent cells (TransGen, Beijing, China). Positive clones were selected by PCR using M13 primers and then sequenced.

#### Sequencing Analysis

Sequences obtained for alignment and phylogenetic tree construction were downloaded from NCBI database (https:// www.ncbi.nlm.nih.gov/), the putative signal peptides were predicted with SignalP 4.1 server (http://www.cbs.dtu.dk/ services/SignalP/). Sequence alignments were produced with DNAMAN software. The phylogenetic tree was constructed using the neighbor-joining method with the MEGA 5.2 program (bootstrapping with 1,000 replications; Tamura et al., 2011). Evolutionary distances were computed using the Poisson correction method.

#### Recombinant Protein Expression and Purification

Prokaryotic expression system (Gu et al., 2012) was used to express CpunPBP2 and CpunPBP5. First, the pGEM plasmid containing the positive clones were digested by Bam HI and Hind III enzymes (NEB, Beijing, China). The expected band was purified and cloned into the bacterial expression vector pET 30a(+) digested with the same enzymes. The pET 30a(+)- CpunPBP2 and pET 30a(+)-CpunPBP5 were transformed into the TransT1 competent cells and grown on LB solid medium with 10 µL kanamycin (10 mg/mL). Positive colonies were selected by PCR using T7 primers and transformed into BL21 (DE3) competent cells (TransGen, Beijing, China). The verified single colony was cultured overnight in 5 mL LB broth including 50µg/mL kanamycin. LB broth (0.5 L) was inoculated with 5 mL overnight culture at 37◦C for 3 h until the absorbance at OD<sup>600</sup> reached to 0.6. Then the protein was induced with isopropylβ-d-thiogalactoside (IPTG) in a final concentration of 1 mM at 37◦C for 6 h (Prestwich, 1993). The induced bacterial cells were centrifuged at 4◦C for 10 min (10,000 rpm) and resuspended in the PBS buffer (NaCl 137 mmol/L, KCl 2.7 mmol/L, Na2HPO<sup>4</sup> 10 mmol/L, KH2PO<sup>4</sup> 2 mmol/L, pH 7.4), agitated by ultrasonic waves (an interval of 5 s, 10 min) and centrifuged again (15,000 rpm, 20 min, 4◦C). The supernatant and pellet were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis

(SDS-PAGE), which showed that CpunPBP2 and CpunPBP5 were expressed mainly in the precipitate. Precipitate was resolved in 8M carbamide and purified by 6 × His-Tagged Purification Kit (CWbio, Beijing, China). Refolded proteins were dialyzed within PBS buffer overnight at 4◦C and then concentrated using Amicom 10 kDa cutoff concentrators (Millipore Billerica, MA, USA). The purity and size were checked by SDS-PAGE. The concentration was determined by the Bradford method using bovine serum albumin (BSA) as standard protein.

## Fluorescence Displacement Binding Assay

Fluorescence binding assay was used to measure the affinity of the CpunPBP2 and CpunPBP5 to 3 sex pheromone and 21 volatile compounds (Konno et al., 1982; Kyungsaeng and Park, 2005). The fluorescence intensity was recorded on a FluoroMax-4 spectrophotometer (Horiba Scientific, USA) at room temperature using a 1 cm light path fluorimeter quartz cuvette. The fluorescent probe N-phenyl-1-naphthylamine (1- NPN) and all the tested chemicals were dissolved in HPLC purity methanol. The final concentration was prepared 1 mM. To measure the affinity of florescent ligand 1-NPN to each



The Int represents the ration of fluorescence intensity values at the pheromone concentration of 6 mM to the initial fluorescence intensity without the pheromone. The farnesene is a mixture of α-farnesene and β-farnesene.

protein, a 2µM solution of the protein in 50 mM Tris-HCl, pH 7.4, was titrated with aliquots of 1 mM ligand in methanol to final concentrations of 1–8µM. The fluorescence of 1-NPN was excited at 337 nm and emission spectra were recorded between 300 and 450 nm. The affinity of other ligands was measured in competitive binding assays, using 1-NPN as the fluorescent reporter at 2µM concentration and different concentrations of each ligands. The GraphPad Prism 5 (GraphPad Software, Inc.) was used to estimate the K1−NPN (K<sup>D</sup> of complex protein /1-NPN) values by nonlinear regression for a unique site of binding. It was assumed that the proteins were 100% active, with a stoichiometry of 1:1 (protein:ligand) at saturation. For other competitor ligands, the dissociation constants were calculated from the corresponding IC<sup>50</sup> (concentrations of ligands halving the initial fluorescence value of 1-NPN) values using Microsoft Office Excel 2010, with the formula: K<sup>D</sup> = [IC50]/(1+[1-NPN]/K1−NPN). In the equation, [1-NPN] is the free concentration of 1-NPN, and K1−NPN is the dissociation constant of the complex protein /1-NPN.

#### Molecular Docking

Sequences of CpunPBP2 and CpunPBP5 were submitted to the SWISS-MODEL server (http://swissmodel.expasy.org/) for structural modeling with all known proteins to obtain template sequences. Then target and template sequences were aligned with ClustalW program. Finally, three dimensional models of CpunPBP2 and CpunPBP5 were generating using I-TASSER Protein Structure and Function Prediction web server (http://zhanglab.ccmb.med.umich.edu/I-TASSER/; Zhang, 2008; Yang et al., 2015). The 3D structure of E10-16:Ald and Z10-16:Ald were obtained from ChemOffice (http://www. cambridgesoft.com/Ensemble\_for\_Chemistry/ChemOffice/ ChemOfficeProfessional/) and was further refined by the CHARMm force field (http://www.charmm.org/). The model was rendered in PyMol (http://www.pymol.org/). The energy minimization was used to refine the ligand poses. Based on the established homology model, the docking program CDOCKER was used to dock the sex pheromone compounds (E10-16:Ald and Z10-16:Ald) with CpunPBP2 and CpunPBP5 models,

respectively. The binding energy included van der Waals energy (Evdw), electrostatic interaction energy (Eeie) and total interaction energy (Etotal). The energy required for interactions among sex pheromone and CpunPBP2 and CpunPBP5 were calculated to select key residues.

#### Preparation of Site-Directed Mutants

Four mutants of CpunPBP2 and four mutants of CpunPBP5 were developed using the QuikChange Lightning Site-Directed Mutagenesis Kit (Stratagene, USA). The mutational primers were designed manually. Mutation sites are underlined in **Table S2**. The CpunPBP2/pGEM-T Easy construct was used as a template. The PCR conditions were 95◦C for 5 min, followed by 30 cycles of 95◦C for 30 s, 58◦C for 30 s and 68◦C for 1 min, and final extension at 72◦C for 10 min. The correct insertion of mutation was subcloned into pGEM-T Easy vector (TransGen, Beijing, China). The expression system and fluorescence binding assay were conducted as mentioned for wild type proteins.

#### Relative Expression Pattern of CpunPBP2 and CpunPBP5

Antennae, proboscises, maxillary palps, thoraxes, legs, abdomens, heads (without antennae, proboscises, and maxillary palps), and wings (50 pairs of each sex) were collected for total RNA extraction using Trizol reagent (Invitrogen, Carlsbad, CA, USA). The first strand cDNA template was synthetized with One-Step gDNA removal and cDNA Synthesis kit (TransGen, Beijing, China) including oligo dt-primer according to product manual recommendations. The primers of CpunPBP2, CpunPBP5 and reference gene (β-actin, accession number JX119014) for realquantitative PCR (qPCR) were designed using Primer premier 5.0 program (Premier Biosoft International, Palo Alto, CA, USA; **Table S1**). qPCR were conducted on ABI 7500 fast real-time PCR system (Applied Biosysterm, USA). Each amplification reaction was performed with 20 µL volume using SYBR Premix Ex Taq II (Tli RNaseH Plus) master mix (Takara-Bio, Shiga, Japan) under the following conditions: 95◦C for 30 s, followed by 40 cycles of 95◦C for 3 s and 60◦C for 30 s. To check reproducibility, each test sample was done in triplicate technical replicates and three biological replicates. Relative quantification was analyzed using the comparative 2−11CT method (Livak and Schmittgen, 2001). The relative expression levels in different tissues were calculated with the transcript level of the female antennae used as the calibrator.

## RESULTS

#### Sequence Analysis of CpunPBP2 and CpunPBP5

Coding regions of CpunPBP2 and CpunPBP5 were obtained from the antennal cDNA library. Sequence analysis shows that the full-length cDNA encoding CpunPBP2 consists of 513 nucleotides that encode 170 amino acid residues. SignalP predicts that the signal peptide cleavage sites have 25 amino acids. On the other hand, CpunPBP5 contains 507 nucleotides for a polypeptide of 168 amino acids with 25 amino acids as signal peptide. The alignment of amino acid sequences shows that CpunPBP2 and CpunPBP5 have six conserved cysteines, which are typical of classic OBPs (**Figure S1**; Pelosi et al., 2006). Additionally, a few amino acids also are conserved in the aligned sequences. Compared with the other 81 Lepidopteran PBPs, the phylogenetic tree based on the amino acid sequences shows that CpunPBP2 and CpunPBP5 share closer ancestry with PBPs in Crambidae, Lepidoptera (**Figure S2**).

## Recombinant Protein Expression and Fluorescence Displacement Binding Assay

Recombinant CpunPBP2 and CpunPBP5 (wild type) proteins expressed in E. coli occurred in inclusion bodies and were high yield. The precipitate was resuspended and purified by affinity chromatography (**Figure 1**) to produce ∼1 mg/mL protein, which was used in the fluorescence displacement binding assay.

Fluorescence of CpunPBP2/1-NPN and CpunPBP5/1-NPN complexes were excited at 337 nm, and the fluorescence peak was 390–410 nm. The dissociation constants (KD) of CpunPBP2/1- NPN and CpunPBP5/1-NPN complexes are 0.76 ± 0.10µM and 0.58 ± 0.04µM as measured by Scatchard plots (**Figure 2**). The IC<sup>50</sup> values and the calculated K<sup>D</sup> of 21 volatiles and 3 sex pheromone analogs to CpunPBP2 and CpunPBP5 are shown in **Table 1**.

Fluorescence intensity of CpunPBP2 and CpunPBP5 gradually declined with the increased concentrations of volatiles and sex pheromone (**Figure 3**). The results show that CpunPBP2 and CpunPBP5 have the highest binding ability to sex pheromones E10-16:Ald, Z10-16:Ald compared with hexadecanal and other volatiles. Compared with CpunPBP5, CpunPBP2 has a higher binding affinity to the sex pheromone E10-16:Ald and hexadecanal. CpunPBP2 also has a similar binding affinity between E10-16:Ald and Z10-16:Ald. This result indicates there is a definite apparent interaction between the sex pheromones and the two PBPs. Among the volatiles, the binding results indicate that 1-tetradecanol had the highest binding affinity with CpunPBP2 and CpunPBP5, followed by transcaryopyllene, farnesene, β-farnesene. Interestingly, results also indicate that CpunPBP2 and CpunPBP5 could discriminate the chiral structure of chemical molecules. The two PBPs could bind to α-ionone better than its isomer β-ionone, while is counter to the isomer of pinene. Hexenal, cis-3-hexen-1-ol, α-pinene, βpinene, trans-2-nonenal and linalool had the minimum binding affinities to CpunPBP2. The vanillin, heptanal, cis-3-hexen-1-ol, linalool had the minimum binding abilities to CpunPBP5.

#### Molecular Docking

To predict the 3D structure of CpunPBP2 and CpunPBP5, sequences from other similar proteins were compared. The search suggests BmorPBP (PDB id: 1ls8) and AtraPBP1 (PDB id: 2 kph) were used to construct the 3D structure of CpunPBP2 and CpunPBP5 with high similarity (54.0 and 45.8%), respectively (**Figure 4**). The predicted 3D structure of CpunPBP2 and CpunPBP5 consists of seven α-helices, and the antiparallel helices converge to form the hydrophobic binding pocket (**Figure 4**). To further study the binding site of sex pheromone to CpunPBP2 and CpunPBP5, E10-16:Ald and Z10-16:Ald were docked with the predicted CpunPBP2 and CpunPBP5 models (**Figure 5**). The interaction energies between key residues and the ligands are predicted and listed in **Tables 2**, **3**. Based on the interaction energy of docking models, several residues including Phe9, Phe33, Ser53, and Phe115 in CpunPBP2 and Ser9, Phe12, Val115, and Arg120 in CpunPBP5 seem to play crucial roles in the binding to E10-16:Ald and Z10-16:Ald.

#### Fluorescence Displacement Binding Assay of Mutants

The recombinant mutant proteins were expressed and purified as described for wild type and analyzed by SDS-PAGE (**Figure S3**). The emission wave lengths of mutants with 1-NPN were 400–410 nm. The binding curve (**Figure 6**) of CpunPBP2 and CpunPBP5 mutants with 1-NPN complexes were made. The binding affinities of mutant between proteins and sex pheromones are listed in **Table 4**. The results showed that, compared with CpunPBP2, the mutant Cpun2-m4 likely lost the binding ability to the two sex pheromones (**Figure 6**). The binding abilities of the three remaining mutants show no significant differences with wild CpunPBP2. Compared with CpunPBP5, the binding ability of all CpunPBP5 mutants to sex pheromones are reduced by varying degrees (**Figure 6**). The binding affinity of mutant of CpunPBP5-m3 to E10-16:Ald decreased the most, and the binding capacity of CpunPBP5-m4 to Z10-16:Ald also decreased considerably.

#### Tissues-Specific Expression Pattern of CpunPBP2 and CpunPBP5

The expression levels of CpunPBP2 and CpunPBP5 in different tissues (male and female antennae, proboscises, maxillary palps, thoraxes, legs, abdomens, heads, and wings) were evaluated using qPCR. The target product was largely amplified in antennae, with low expression level in other tissues (**Figure 7**). CpunPBP5 is mainly expressed in the female antennae, which contrasts

TABLE 3 | Interaction energies (kcal/mol) between the key residues of CpunPBP5 and pheromone compounds.

CpunPBP5 E10-16:Ald Z10-16:Ald

TABLE 2 | Interaction energies (kcal/mol) between the key residues of CpunPBP2 and pheromone compounds.


Etotal Evdw Eeie Etotal Evdw Eeie MET5 −0.764 −1.058 0.294 −0.997 −1.144 0.146 MET8 −2.308 −1.781 −0.527 −0.715 −0.850 0.135 SER9 −10.204 −1.951 −8.253 −7.416 −2.154 −5.262 PHE12 −5.351 −4.785 −0.566 −3.593 −3.777 0.185 PHE13 −0.862 −0.553 −0.309 – – – LEU33 −0.273 −0.219 −0.054 – – – PHE36 −1.285 −1.374 0.088 −1.224 −1.142 −0.082 TRP37 – – – −0.878 −0.448 −0.430 ILE52 −1.349 −1.553 0.204 −1.880 −1.700 −0.180 ALA56 −0.383 −0.569 0.185 −0.967 −1.108 0.141 GLN59 −1.109 −0.981 −0.129 – – – LEU61 −1.481 −1.659 0.177 −1.897 −1.906 0.009 VAL62 – – – −1.666 −1.738 0.072 TYR67 – – – −2.049 −1.855 −0.194 ARG68 −0.833 −0.741 −0.092 MET69 −1.262 −1.337 0.075 PHE77 −1.404 −1.783 0.379 −0.061 −0.412 0.473 ILE91 −1.350 −1.260 −0.090 −0.295 −0.264 −0.031 ILE95 −3.348 −3.445 0.097 −2.878 −2.964 0.086 GLU99 – – – −2.575 −2.008 −0.567 ARG111 – – – −1.269 −1.490 0.221 VAL112 −0.515 −0.730 0.215 −1.873 −2.000 0.127 VAL115 −2.391 −2.418 0.027 −3.168 −3.170 0.002 SER116 −2.254 −2.360 0.106 −1.693 −1.941 0.247 PHE119 −3.560 −3.639 0.078 −1.987 −1.940 −0.047 ARG120 −10.096 −2.088 −8.007 – – – LEU135 −0.498 −0.366 −0.132 −0.445 −0.339 −0.106

Etotal, total interaction energy; Evdw, Van der Waals energy; Eeie, electrostatic interaction energy.

Etotal, total interaction energy; Evdw, Van der Waals energy; Eeie, electrostatic interaction energy.

TABLE 4 | IC50 values (µM) and calculated dissociation constants (KD) (µM) of CpunPBP2 and CpunPBP5 with their mutants to two pheromones.


For each pheromone compound, different letters within a column of each protein indicate significant differences (LSD test, P < 0.05).

with CpunPBP2 and its male-specific expression. In general, expression levels of CpunPBP2 and CpunPBP5 in proboscises, maxillary palps, thoraxes, legs, abdomens, heads, and wings were very low or null.

#### DISCUSSION

Odorant binding proteins are essential for insect olfactory perception because they are transporters between the external environment and insect chemoreceptors (Sun Y. L. et al., 2013). Fluorescence binding affinity has emerged as an important method to demonstrate binding capacity with ligands and help elucidate mechanisms of OBPs (Campanacci et al., 2001; Fan et al., 2011). Jia et al. (2015) cloned a PBP from C. punctiferalis and named as CpunPBP1 (GenBank accession number: KP027286), which is similar to CpunPBP2 we obtained. But in 2016 (Jia et al., 2016), they got the same sequence by transcriptome analysis and named as CpunPBP2 (GenBank accession number: GEDO010000019.1). In order to eliminate the confusion, we use the second name in our study. In this study, CpunPBP2 and CpunPBP5 had strong binding abilities with two sex pheromone compounds, indicating that the two PBPs may play important roles in transporting sex pheromones within the sensillar lymph. Furthermore, CpunPBP2 and CpunPBP5 also bind volatiles: 1-tetrodecanol, trans-caryopyllene, farnesene, and β-farnesene, which suggest CpunPBP2 and CpunPBP5, may share similar amino acid binding sites with GOBPs associated with the volatiles (Mao et al., 2016). Interestingly, CpunPBP2 and CpunPBP5 discriminate the chiral structure of chemical molecules, similar to AlinOBP5 results in Adelphocoris lineolatus (Wang et al., 2013). We speculate that the chiral structure of ligands may affect the binding constants and need to be further investigated.

Protein structure plays crucial roles in recognition and binding of ligand molecules. Studies of the interactions between proteins and ligands are necessary to better understand the binding mechanism. Structures of OBP and PBP in other lepidopteran insects, such as B. mori (Sandler et al., 2000; Horst et al., 2001), A. polyphemus (Mohanty et al., 2004) and A. transitella (Xu et al., 2010; di Luccio et al., 2013), were used to provide insights into our PBPs. In this study, the key residues

were evaluated based on the energy values. After site-directed mutagenesis, four mutants of CpunPBP2 and four mutants of CpunPBP5 protein were purified and used to analyze the binding mechanism. Compared with CpunPBP2, the binding ability of CpunPBP2 mutants were not significantly reduced, expect for CpunPBP2-m4. We speculate that the amino acid substitution of the three mutants of CpunPBP2 had a slight effect of relaxing the compact structure of the binding site, similar to the loss of high specificity with Plutella xylostella mutants (Zhu et al., 2016). Because Phe115 in CpunPBP2 had a stronger hydrophobic interaction than other amino acids (**Table 2**) and the binding affinity between CpunPBP2-m4 and sex pheromone compounds sharply decreased, we speculate that Phe115 in CpunPBP2 are involved in sex pheromone recognition. The binding abilities of CpunPBP5 mutants with sex pheromones varied, which suggests that the small protein modifications may have affected the hydrogen bond between protein and sex pheromones. These results may be due to the change of hydrocarbon interactions or the stabilization of the hydrophobic binding pocket. This suggests that the conformation of PBP was influenced by the transformation of the single amino acid (Zhang et al., 2017). Further research using NMR or x-ray to analyze the protein structure may be necessary to better understand these changes.

The expression levels measured by qPCR showed that CpunPBP2 and CpunPBP5 were mainly expressed in antennae, with low expression in the other tissues. These results suggest that CpunPBP2 and CpunPBP5 play a crucial role in odorant chemoreception (including sex pheromone). CpunPBP2 gene was more abundantly expressed in male antennae than in female antennae, which is similar to results found in other insects, including Spodoptera exigua, P. xylostella, Agrotis ipsilon, Helicoverpa armigera, and Maruca vitrata (Xiu and Dong, 2007; Zhang et al., 2011; Gu et al., 2013; Sun M. J. et al., 2013; Mao et al., 2016). High expression of CpunPBP2 in male antennae may indicate that CpunPBP2 is involved in male-female recognition. Expression level of the CpunPBP5 gene in male antennae was lower than that of female antennae, which is similar to results found with M. vitrata and Sesamia inferens (Jin et al., 2014; Mao et al., 2016). Thus, these results suggest CpunPBP2 may be involved in the detection of conspecific sex pheromone and autodetection of sex pheromone compounds (Yang et al., 2009; Holdcraft et al., 2016; Mao et al., 2016).

In conclusion, our study provides key information about CpunPBP2 and CpunPBP5 in C. punctiferalis, which may be useful for developing effective pest management strategies for this pest.

## AUTHOR CONTRIBUTIONS

TZ and ZW: Conceived and designed the experimental plan; XG: Preformed the experiments; XG, TZ, and SB: Analyzed the sequence and data; ZW and KH: Provided all the materials and lab facilities necessary for this work; TA and ZW: Revised the manuscript. All authors read and approved the final manuscript.

#### ACKNOWLEDGMENTS

This work was supported by Special Fund for Agro-scientific Research in the Public Interest (201303026) and China Agriculture Research System (CARS-02). The authors gratefully acknowledge Dr. Richard Hellmich for comments and suggestions on this manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00308/full#supplementary-material

Figure S1 | Multiple sequence alignment of CpunPBP2 and CpunPBP5 with other Lepidopteran insect PBPs.

Figure S2 | Phylogenetic tree of CpunPBPs amino acid sequences with other 81 PBPs from different insect species.

Figure S3 | SDS-PAGE analyses of purified CpunPBP2 and CpunPBP5 with their mutants. M: marker protein. (A)1:CpunPBP2 original protein. (A)2: CpunPBP2-Phe9 mutant (CpunPBP2-m1). (A)3: CpunPBP2-Phe33 mutant (CpunPBP2-m2). (A)4: CpunPBP2-Ser53 mutant (CpunPBP2-m3). (A)5: CpunPBP2-Phe115 mutant (CpunPBP2-m4). (B)1:CpunPBP5 original protein. (B)2: CpunPBP5-Ser9 mutant (CpunPBP5-m1). (B)3: CpunPBP5-Phe12 mutant (CpunPBP5-m2). (B)4: CpunPBP5-Val115 mutant (CpunPBP5-m3). (B)5: CpunPBP5-Arg120 mutant (CpunPBP5-m4).

Table S1 | Primers for expression and qPCR.

Table S2 | Primers for mutants of CpunPBP2 and CpunPBP5.

## REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ge, Ahmed, Zhang, Wang, He and Bai. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Silencing the Odorant Binding Protein *RferOBP1768* Reduces the Strong Preference of Palm Weevil for the Major Aggregation Pheromone Compound Ferrugineol

#### Binu Antony\* † , Jibin Johny † and Saleh A. Aldosari

Chair of Date Palm Research, Department of Plant Protection, College of Food and Agricultural Sciences, King Saud University, Riyadh, Saudi Arabia

#### *Edited by:*

Shigehiro Namiki, Research Center for Advanced Science and Technology, The University of Tokyo, Japan

#### *Reviewed by:*

Bingzhong Ren, Northeast Normal University, China Nicolas Durand, Université Pierre et Marie Curie, France

> *\*Correspondence:* Binu Antony bantony@ksu.edu.sa

†These authors have contributed equally to this work.

#### *Specialty section:*

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

*Received:* 21 November 2017 *Accepted:* 06 March 2018 *Published:* 21 March 2018

#### *Citation:*

Antony B, Johny J and Aldosari SA (2018) Silencing the Odorant Binding Protein RferOBP1768 Reduces the Strong Preference of Palm Weevil for the Major Aggregation Pheromone Compound Ferrugineol. Front. Physiol. 9:252. doi: 10.3389/fphys.2018.00252 In insects, perception of the environment—food, mates, and prey—is mainly guided by chemical signals. The dynamic process of signal perception involves transport to odorant receptors (ORs) by soluble secretory proteins, odorant binding proteins (OBPs), which form the first stage in the process of olfactory recognition and are analogous to lipocalin family proteins in vertebrates. Although OBPs involved in the transport of pheromones to ORs have been functionally identified in insects, there is to date no report for Coleoptera. Furthermore, there is a lack of information on olfactory perception and the molecular mechanism by which OBPs participate in the transport of aggregation pheromones. We focus on the red palm weevil (RPW) Rhynchophorus ferrugineus, the most devastating quarantine pest of palm trees worldwide. In this work, we constructed libraries of all OBPs and selected antenna-specific and highly expressed OBPs for silencing through RNA interference. Aggregation pheromone compounds, 4-methyl-5-nonanol (ferrugineol) and 4-methyl-5-nonanone (ferruginone), and a kairomone, ethyl acetate, were then sequentially presented to individual RPWs. The results showed that antenna-specific RferOBP1768 aids in the capture and transport of ferrugineol to ORs. Silencing of RferOBP1768, which is responsible for pheromone binding, significantly disrupted pheromone communication. Study of odorant perception in palm weevil is important because the availability of literature regarding the nature and role of olfactory signaling in this insect may reveal likely candidates representative of animal olfaction and, more generally, of molecular recognition. Knowledge of OBPs recognizing the specific pheromone ferrugineol will allow for designing biosensors for the detection of this key compound in weevil monitoring in date palm fields.

Keywords: red palm weevil, pheromone-binding protein, aggregation pheromone, RNAi, EAG, olfactometer

## INTRODUCTION

Perception of odorants and chemical sensing are essential processes for the survival of all animals. Research of olfaction and the olfactory system has experienced a quantum leap in recent decades mainly because of patented applications in fields such as biosensors, behavior-based robots, perfumes, and the chemical industry (Du et al., 2013; Yeon et al., 2015; Brito et al., 2016

**123**

Ando and Kanzaki, 2017; Garm et al., 2017; Hadagali and Suan, 2017; Leal, 2017; Lutz et al., 2017). Some aspects of human olfaction are difficult to study; conversely, such systems are more readily investigated in insects, organisms that rely strongly on olfaction. Although some differences between olfaction in mammals and insects exist, they are similar in many important ways. In this study, we examined the olfactory system of the red palm weevil (RPW) Rhynchophorus ferrugineus, the most invasive and globally important quarantine pest of palm trees. R. ferrugineus was introduced to Saudi Arabia from Southeast Asia during the 1980s; it subsequently spread to all Middle Eastern countries and has since migrated into Spain and Southern France (Barranco et al., 1996; Martín et al., 2000; Dembilio and Jaques, 2015; Al-Dosary et al., 2016). The regional and global spread of palm weevil was primarily facilitated by humans via the transport of infested offshoots and young or mature date palm trees from weevil-outbreak areas into uninfected areas (Faleiro, 2006; Al-Dosary et al., 2016). When RPWs attack a palm tree, the male weevils release an aggregation pheromone (4-methyl-5-nonanol and 4-methyl-5-nonanone); other RPWs within the vicinity are attracted to the signal, which often leads to a coordinated mass attack and eventually results in the death of the palm tree (Soroker et al., 2005; Faleiro, 2006). Palm weevil aggregation pheromones function in various processes, including defense against predators, overcoming host resistance by mass attack and mate selection. Because of the economic and ecological impacts of this pest, we selected it for study to obtain more extensive knowledge regarding its olfactory communication.

Insect pheromone reception is a complex process in which odorants reach the aqueous environment of the sensillar lymph through multiple pores present on the surface of sensilla. Pheromone-binding proteins (PBPs), odorant receptors (ORs), ionotropic receptors (IRs), sensory neuron membrane proteins (SNMPs), chemosensory proteins (CSPs), and odorantdegrading enzymes (ODEs) are the main proteins of the peripheral olfactory system involved in odorant perception (Hansson and Stensmyr, 2011; Leal, 2013; Zhang et al., 2014; Andersson et al., 2015). The different olfactory protein families involved in insect olfaction have been identified (Hansson and Stensmyr, 2011; Vosshall and Hansson, 2011; Leal, 2013; Missbach et al., 2014; Fleischer et al., 2018), and we selected one set of these genes that code for proteins involved in the first stage of the olfactory process: odorant-binding proteins (OBPs). OBPs interact with particular molecules in the chemical cues of individual odorants and transport them to receptors (Vogt and Riddiford, 1981; Pelosi and Maida, 1995; Pelosi et al., 2006, 2014). Insect OBPs comprise approximately 130–140 amino acids, are abundantly distributed in chemosensilla, consist of four to six αhelical domains and are characterized by four to six conserved cysteines paired into two to three interlocked disulfide bridges (Angeli et al., 1999; Leal et al., 1999; Sandler et al., 2000; Tegoni et al., 2004; Vieira and Rozas, 2011; Pelosi et al., 2014). OBPs are present at high concentrations in the lymph between the dendritic membrane and the cuticular wall (Pelosi et al., 2006, 2014).

Although pheromone detection involving PBPs in insects has been extensively studied, most of the research to date has been performed in moths, mosquitoes and Drosophila, whereas there are only a few reports for Coleoptera (Brito et al., 2016; Pelosi et al., 2017). Furthermore, there is a lack of information on olfactory perception and the molecular mechanism by which OBPs participate in the transport of aggregation pheromones in Coleoptera. We selected palm weevil because it is a global pest of palm trees that mainly uses aggregation pheromones to coordinate mass attacks on palm trees, with both host searching and reproductive activity relying strongly on male-produced pheromones. We aimed to identify and characterize a specific subclass of pheromone-specific OBPs by selectively silencing key OBPs using RNA interference and assessing changes in weevil behavior using behavioral trials and electrophysiological recordings. As R. ferrugineus is among the world's most invasive pest species of palm trees and this pest has wreaked havoc in the date palm industry in Middle Eastern countries, our current research findings on R. ferrugineus OBPs may be applicable in the development of biosensors for pheromonebased monitoring or might be used to screen behaviorally active compounds (attractants or repellents) in an approach similar to "reverse chemical ecology" (Leal et al., 2008; Leal, 2017).

## MATERIALS AND METHODS

#### Insect Collection and Rearing

RPW collections were performed with the direct consent of a cooperating land owner [Saudi Arabia, Al-Kharj region (24.1500◦N, 47.3000◦E)] in the year 2009. The collected RPWs were maintained in our laboratory on sugarcane stems at 28–30◦C with a photoperiod of 18 h:6 h (light: dark), as described previously (Antony et al., 2016, 2017).

## Identification and Phylogenetic Analysis of RferOBPs

Red palm weevil antennal transcriptome data (Antony et al., 2016) were screened and annotated for candidate OBP genes. Both Blast2GO and manual annotations were performed for the nomenclature, and for convenience, we added a prefix, Rfer (R. ferrugineus) for OBP transcripts, followed by the identification number. Reads per kilobase per million (RPKM) values were calculated according to a published formula (Mortazavi et al., 2008). The identified candidates were further annotated and checked for duplications and open reading frame (ORF) identification using the NCBI BLASTx homology search and ORF Finder (https://www.ncbi.nlm.nih. gov/orffinder/). The ORF amino acid sequences were used for phylogenetic tree construction along with selected OBP protein sequences retrieved from NCBI and Protein Data Bank. Multiple sequence alignment was performed using MUSCLE (Edgar, 2004), and a neighbor-joining (NJ) analysis based phylogenetic tree was reconstructed using MEGA v6 (Kumar et al., 2016), with the tree branches supported by 1,000 bootstrap replications.

## Selection of Candidate OBPs for Gene Silencing

#### Tissue-Specific Expression Analysis

For tissue-specificity and qRT-PCR studies, the antennae, snout, legs, thorax, abdomen, and wings were excised from 20-day-old adult insects. Total RNA was extracted from 30 mg of tissue for each sample using PureLink RNA Mini Kit (Ambion, USA), and first-strand cDNA was synthesized using SuperScript IV Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. The quality and quantity of the RNA and cDNA were examined using a NanoDrop spectrophotometer (Thermo, Delaware, USA). Primers were designed using Primer3 software (Untergasser et al., 2012) with the following parameters: Tm, 56–60◦C; GC content, 40–50%; and product size, 190–200 bp (Table S1). Touchdown polymerase chain reaction (PCR) [95◦C for 5 min, 35 cycles of 95◦C for 1 min, 60◦C (touchdown to 54◦C) for 30 s and 72◦C for 30 s; and one cycle at 72◦C for 10 min] was carried out using GoTaq Green PCR Master Mix (Promega, USA), and the PCR products were evaluated by 2.5% agarose gel electrophoresis alongside a 100-bp DNA ladder (Solis BioDyne, Tartu, Estonia) as a marker and visualized using ethidium bromide (Promega, USA) staining.

#### Relative Expression Analysis by qRT-PCR

cDNAs were prepared from RNA extracted from the antennae of 20-day-old insects, as mentioned above. qRT-PCR was carried out using SYBR Green PCR Master Mix (Life Technologies, USA) with three biological and three technical replicates according to the manufacturer's instructions. The oligonucleotide primers used were the same as those used in the tissue-specific studies, and tubulin (Table S1) was employed to normalize gene expression. The relative RferOBP expression levels were measured by the 2−11<sup>C</sup> <sup>T</sup> method (Schmittgen and Livak, 2008). The following thermal programme was used to perform the PCR amplification: holding stage at 50◦C; 95◦C for 2 or 5 min; 40 cycles of 95◦C for 15 s; and 60◦C for 32 s; and a continuous melting curve stage of 95◦C for 15 s, 60◦C for 1 min, 95◦C for 30 s, and 60◦C for 15 s. The qRT-PCR products were examined by 3% agarose gel electrophoresis and visualized via ethidium bromide staining.

#### Rapid Amplification of cDNA Ends (RACE) and Generation of the Full-Length Sequence

The SMARTer rapid amplification of cDNA ends technique (SMARTer RACE Kit, Clontech, CA, USA) was used to obtain the full-length sequences of candidate OBPs by amplifying both cDNA ends (5′ and 3′ ends). The 5′ and 3′ RACE cDNAs were prepared from total RNA of adult R. ferrugineus antennae, as described (Soffan et al., 2016). Gene-specific primers (GSPs) for 5 ′ - and 3′ -RACE were designed based on partial RferOBP23, RferOBP107, RferOBP1768, and RferOBPu1 nucleotide sequences (Table S1). The amplification reactions were carried out as follows: 95◦C for 5 min; 30 cycles of 95◦C for 1 min, 65◦C (touchdown to 60◦C) for 30 s and 72◦C for 2 min; and one cycle at 72◦C for 10 min. The amplified PCR products were purified using Wizard SV Gel Purification Kit (Promega, USA) and cloned into the pGEM-T vector (Promega, USA) followed by transformation into JM109 competent cells (Promega, USA). The plasmids were isolated from bacteria, sequenced in both directions (ABI 3500, Life Technologies, MD, USA), aligned and annotated using a BLASTx homology search.

#### Structural and Functional Analyses

Amino acid similarity and identity were calculated using the SIAS tool (http://imed.med.ucm.es/Tools/sias.html). Sequence logos of the aligned R. ferrugineus OBP orthologs were created using WebLogo 3.1 (Crooks et al., 2004). The DISULFIND web server tool (http://disulfind.dsi.unifi.it) was used to predict the distribution of disulfide bonds. Compute pI/Mw (http://web. expasy.org/compute\_pi/) was used to predict the theoretical pI (isoelectric point) and Mw (molecular weight). The SignalP 4.0 Server program (http://www.cbs.dtu.dk/services/SignalP) and Euk-mPLoc 2.0 (http://www.csbio.sjtu.edu.cn/bioinf/euk-multi-2/) were applied to predict signal peptides and subcellular localization, respectively. The 3DLigandSite tool (http://www. sbg.bio.ic.ac.uk/3dligandsite) was utilized to predict ligandbinding sites in the proteins. The Phyre2 tool (http://www.sbg. bio.ic.ac.uk/\$\sim\$phyre2/) was employed to predict secondary structures, and PyMol (https://pymol.org/2/) was used to visualize simulated three-dimensional structures.

#### RferOBP Silencing by RNA Interference (RNAi)

We used plasmids containing the full-length OBP ORF as template DNA to synthesize double-stranded RNA (dsRNA). ORF reverse primers with a T7 overhang and T7 forward primers (Table S1) were used to amplify and linearize ORFs, which were rechecked by direct sequencing (ABI 3500, Life Technologies, USA). dsRNA synthesis was performed using MEGAscript RNAi Kit (Life Technologies, USA) according to the manufacturer's instructions, and the results were quantified using a NanoDrop 2,000 (Thermo Scientific, DE, USA). dsRNA was examined by 1% agarose gel electrophoresis to evaluate the integrity and efficiency of duplex formation. We selected 10-day-old R. ferrugineus pupae for RNAi experiments, and 40 ng/µL dsRNA (in 20 µL) was injected at a depth of 0.5 cm into the first dorsal segment of the abdomen, close to the thorax, using a 0.5-mL BD Micro-FineTM PLUS syringe (Becton, Dickinson Co., NJ, USA). dsRNA-injected RPW pupae were maintained as previously described (Soffan et al., 2016). As two separate controls, RPW pupae were injected with nuclease-free water (hereafter referred to as "NFW") or not injected (hereafter referred to as "NI"). The adults emerging at 21 days were further subjected to quantification of gene silencing (qRT-PCR), behavioral assays using an olfactometer, and electrophysiological recording using an electroantennogram (EAG), as described below.

#### Gene Silencing Validation by qRT-PCR

cDNAs were prepared from RNA extracted from the antennae of each individual insect in the experimental (dsRNA injected) and control (NFW and NI) groups and used as template for qRT-PCR. Reactions were carried out using SYBR Green PCR Master Mix (Life Technologies, USA) according to the manufacturer's instructions, with six biological and six technical replicates. Tubulin and β-actin primers were used to normalize gene expression (Table S1). The relative expression levels of OBPs in the silenced vs. control groups were measured by the 2−11<sup>C</sup> T method (Schmittgen and Livak, 2008). PCR amplification, data analysis, statistical analysis, and gel evaluation were performed as described above.

#### Behavioral and Electrophysiological Assays

#### Olfactometer Assay

The olfactometer assay was used to evaluate the responses to stimuli by the dsRNA-injected, NI and NFW groups of RPW adults. We used a customized olfactometer unit (Volatile Collection System Co, Gainesville, FL) consisting of a Ytube (main-tube length: 47 cm; arm length: 68 cm; diameter: 5 cm; with 40-cm-long/2-cm-diameter plastic tubes in each arm connected to the source of the stimulus), an air-delivery system (humidified air and carbon filter), and a stimulus container (diameter: 8 cm, length: 10 cm). A commercial aggregation pheromone contained 4-methyl-5-nonanol (ferrugineol) and 4 methyl-5-nonanone (ferruginone) at the approximate ratio of 9:1 (ChemTica Int., Costa Rica) and ethyl acetate (Sigma Aldrich, USA) were used in one arm of the instrument, and charcoalfiltered air was applied in the other arm. We used ethyl acetate because several studies have reported that it enhances the efficacy of weevil catch (Soroker et al., 2005; Shagagh et al., 2008; Al-Saoud, 2013; Vacas et al., 2013, 2017). The unit was operated at a pressure of 15 psi and a zero air inlet flow of 1.2 L per minute. Adult insects were starved overnight, and the response to stimuli was recorded three times for each insect. Failure to move within 5 min in the olfactometer Y-tube was classified as "no response." As our preliminary study showed that NI and NFW adult RPWs exhibit similar responses to the stimulus, further assays were carried out with the NI and dsRNA-injected groups only; each group comprised 16 adult RPWs of similar age (ratio of 1:1, male: female). The numbers of times (three times on different experimental days: n) each RPW chose "air," "stimulus," or "no response" were recorded, and the results are expressed as percentages of the total.

#### Electroantennography (EAG)

To validate the effect of gene silencing using RNAi, insects with positive results in the olfactometer assay were subjected to electroantennography. Six adult RPWs were tested per group (dsRNA injected, NFW injected and NI) at the age of 21 days. After demobilization using CO<sup>2</sup> for 1–2 min, the antennae of each insect were excised from the base. Each antenna was then attached to the electrode holders of an EAG system (Syntech, Hilversum, Netherlands) using SPECTRA 360 electrode gel (Parker Lab, Inc. Fairfield, NJ, USA) and subjected to a constant flow of humidified air. Each insect from the experimental groups was exposed to three different stimuli, (4RS,5RS)-4-methylnonan-5-ol, (Phe1) (>92% purity, ChemTica Int., Costa Rica), 4(RS)-methylnonan-5-one (Phe2) (>92% purity, ChemTica Int., Costa Rica), and ethyl acetate (Sigma Aldrich, USA), at concentrations of 0.02 mg/mL (diluted in n-hexane).

A glass Pasteur pipette with a filter paper strip inside (with 4 µL of the stimulus compound) was used to deliver the stimulus via an air-stimulus controller (Model CS-55 Ver.2.7, Syntech, Hilversum, The Netherlands) fitted with a charcoal filter. Odor stimulation puffs were applied twice at 0.1-s intervals and with 20–30-s intervals between each odor compound. The antennal response to each stimulus was recorded using a Syntech Acquisition IDAC-2 controller connected to a computer and processed using GC-EAD 2012 v1.2.4 (Syntech, Kirchzarten, Germany).

#### RferOBP1768 Expression Analysis in Male and Female R. ferrugineus

Differences in RferOBP1768 expression in adult male and female R. ferrugineus weevils were compared by qRT-PCR. Antennae from 21-day-old male and female adult insects were excised, and total RNA extraction and cDNA synthesis were performed as described above. Three biological and three technical replicates were used for male and female RPWs; RferOBP1768 expression was normalized to that of tubulin and β-actin (Table S1) and calculated using the 2−11<sup>C</sup> <sup>T</sup> method (Schmittgen and Livak, 2008).

#### Statistical Analysis

The mean fold change, 2−11<sup>C</sup> <sup>T</sup> values (Livak and Schmittgen, 2001), were calculated using MS Excel (Microsoft corporation, USA). Three experimental groups, consisting of dsRNA RferOBP-injected (dsRNA), not-injected (NI), and NFWinjected groups, were established with triplicate biological and technical replicates. Significant differences among the experimental groups for qRT-PCR, the olfactometer assay and EAG were assessed using one-way analysis of variance (ANOVA), followed by multiple-comparison testing with the least significant difference (LSD) test (P < 0.05) (for the olfactometer assay and qRT-PCR) or with Tukey's HSD test (for EAG analysis) (Stelinski and Tiwari, 2013) using SPSS program v24. Homogeneous subsets in both the olfactometer and EAG assays were identified by Waller-Duncan statistics (α = 0.05) using SPSS program v24 (IBM SPSS statistics, NY, USA).

#### RESULTS

#### Identification and Selection of Candidate OBPs for Gene Silencing

A comprehensive search of the RPW antennal transcriptome identified 38 OBPs, and we confirmed these transcripts by checking for duplication based on BLASTx hits and concluded that 36 OBPs are present in RPW (Table S2). Sequence homology and characterization of the RPW OBPs were performed. RPKM values calculated for the assembled OBP transcripts are presented in **Figure 1**; this analysis revealed highly abundant transcripts of three OBPs (RferOBPu1, RferOBP23, and RferOBP107) in the RPW antennal transcriptome (RPKM > 7,000).


FIGURE 1 | Relative tissue-specific expression analysis of 36 OBPs identified from Rhynchophorus ferrugineus. Tissues used are indicated as AM (male antennae), AF (female antennae), Sn (male snout), Lg (male legs), Thx (male thorax), Ab (male abdomen), and Wg (male wings). tubulin was used to normalize gene expression. Expression of all RferOBPs in the antenna was quantified by qRT-PCR, and the mean fold changes in gene expression compared to tubulin are provided under qRT-RQ. The color gradient indicates the relative level of expression from higher (blue) to lower (red). Primer details and PCR product sizes are provided in Table S1. The original gel image (with DNA ladder) is provided in Figure S1.

## Tissue-Specific Expression Analysis Demonstrates Antenna-Specific RferOBPs

We aimed to investigate OBP(s) involved in the first stage of detecting and transporting aggregation pheromones of R. ferrugineus. For an initial clue regarding their function, we first mapped expression of 36 OBPs in the antennae and other body parts of R. ferrugineus. Among the 36 OBPs, only one candidate OBP (RferOBP1768) was found to be exclusively antenna specific (**Figure 1**). RferOBPu1 exhibited antenna-enriched expression but low expression in the snout (**Figure 1**). Similarly, RferOBP3213 and RferOBP29 showed antenna-enriched expression but with low expression in the leg and abdomen (**Figure 1**). We identified four candidates with reduced expression in the antennae than in other body parts (RferOBP8586, RferOBP7073, RferOBP12010, and RferOBP14511), and the remaining OBPs displayed ubiquitous expression patterns (**Figure 1**). Among the highly expressed candidate OBPs, RferOBP23 was expressed in all tissues studied except the thorax and wings, whereas RferOBP107 was ubiquitously expressed in all tissues (**Figure 1**). Interestingly, two OBPs (RferOBP77 and RferOBP28119) exhibited no expression in the antenna of females but were expressed in all other tissues. For antenna-enriched OBPs, expression of RferOBP12511 and RferOBP12481 was high in RPW females compared to males, with low expression in the snout and abdomen (**Figure 1**).

#### Relative Expression Analysis Reveals Key OBPs in the *R. ferrugineus* Antenna

Expression of all OBPs in the R. ferrugineus antenna was quantitatively measured, and the RQ values are provided in **Figure 1**. Based on qRT-PCR data, RferOBP23, RferOBP77, RferOBP382, RferOBP3199, and RferOBP446 are the highly expressed OBPs in R. ferrugineus. Compared to other OBPs, RferOBPu23 and RferOBP107 were found to be highly expressed in the antenna (**Figure 1**). Other candidate genes showing high expression in the antenna were RferOBP77, RferOBP382, RferOBP3199, and RferOBP446. Conversely, RferOBPu1 expression was lower than that of the highly expressed OBPs (**Figure 1**). The antenna-specific candidate gene RferOBP1768 also displayed moderate expression in the antenna (mean 0.76-fold change normalized by tubulin gene expression), as shown in **Figure 1**. The antenna-enriched OBPs RferOBP12511, RferOBP19755, and RferOBP12481 all showed very low expression (**Figure 1**).

## Structural and Functional Analyses

#### Molecular Cloning, Full-Length Sequencing, and Phylogenetic Analysis

We selected RferOBP23, RferOBP107, RferOBP1768, and RferOBPu1 for full-length cloning and analysis because the first two were found to be highly expressed and the last two were found to be antenna specific and antenna enriched, respectively. Full-length OBP sequences were obtained for RferOBPu1, RferOBP23, RferOBP107, and RferOBP1768 using the SMARTer RACE technique, assisted by a primer walking sequencing strategy. The RferOBPu1, RferOBP23, RferOBP107, and RferOBP1768 genes were confirmed to have full lengths of 612, 643, 703, and 636 bp, respectively, with ORFs of 396, 402, 429, and 399 bp, corresponding to 131, 133, 142, and 132 amino acids (**Figure 2**). The theoretical pI (isoelectric point)/Mw

PBP30, and PBP31. Highly conserved cysteine residues are marked by dark arrowheads. Signal peptides are boxed. Residues highlighted in bright-green have high (>90%) consensus values. Conserved residues are shown with a green background. Because the four OBPs (B. mori) are from different insect orders, homologies are low. Sequence logos of the aligned R. ferrugineus RferOBP1768 and RferOBP23 orthologs are shown in Figure S3.

(molecular weight) of the proteins encoded by RferOBPu1, RferOBP23, RferOBP107, and RferOBP1768 are 4.41/15.16, 4.44/14.98, 4.72/15.88, and 5.08/14.92, respectively. We identified OBP extracellular localization, a typical characteristic of OBP proteins, using the Euk-mPLoc 2.0 server. The full-length amino acid sequences of RferOBPu1 shows 25.11, 16.66, and 48.09% identity with RferOBP23, RferOBP107, and RferOBP1768, respectively. RferOBP23 exhibits 25 and 32.33% identity with RferOBP107 and RferOBP1768, respectively, and RferOBP107 exhibits 16.66% identity with RferOBP1768. Using SignalP-4.1 euk predictions, we identified a highly divergent signal peptide at the N-terminal region, as shown in **Figure 2**.

A NJ rooted tree of various different annotated OBPs and Bombyx mori OBPs (Gong et al., 2009) was used as a reference to classify RferOBPs. We focused on RferOBP23, RferOBP107, RferOBP1768, and RferOBPu1 based on the results obtained in tissue-specificity studies and relative OBP expression analysis. We identified RferOBP23 and RferOBP107 as belonging to ABP II subfamilies, and RferOBP1768 and RferOBPu1 were classified as Minus-C subfamilies (Figure S2). The Minus-C RferOBP1768 clade also includes other R. ferrugineus OBPs, namely, RferOBP12511, RferOBPu1, RferOBP1689, and RferOBP19755. RferOBP1768 shows 65.1 and 66.9% amino acid identity with RferOBP1689 and RferOBP19755 (Figure S3). In our tree, the RferOBP1768 clade, with 91% bootstrap support, forms a clade with TcasOBPC06 and TcasOBPC09 (Figure S2). The tree also revealed that RferOBP23 belongs to an orthologous sequence group containing TcasOBP6, TcasOBP8, and BmoriOBP20, with 56, 54, and 44% bootstrap support, respectively (Figure S2), and that RferOBP107 belongs to an orthologous sequence group containing BmoriOBP21, with 51% bootstrap support (Figure S2). The RferOBP23 clade contains RferOBP3213 and RferOBP107, BmoriOBP20 and OBPs from scarab beetles (Anomala osakana, AosaOBP; Anomala octiescostata, AoctOBP; Anomala cuprea, AcupOBP) and the Japanese beetle Popillia japonica (PjapOBP) (Wojtasek et al., 1998; Nikonov et al., 2002).

OBP NJ tree was constructed based on amino acid sequences using R. ferrugineus and Rhynchophorus palmarum; RpalOBP2 and RpalOBP4 (Nagnan-Le Meillour et al., 2004) and 10 other coleopterans [Tomicus yunnanensis (Liu et al., 2018); Holotrichia oblita (Li K. et al., 2017); Cyrtotrachelus buqueti (Yang et al., 2017a) Colaphellus bowringi (Li X. et al., 2015); Galeruca daurica (Li K. et al., 2017); Tenebrio molitor (Liu et al., 2015); Tribolium castaneum (Dippel et al., 2014), Anomala corpulenta (Li X. et al., 2015)], scarab beetles, and the Japanese beetle (Wojtasek et al., 1998; Nikonov et al., 2002) (**Figure 3**). Members of the ABP II clade show diversity in sequence and function. Gene expansion was identified within this clade, particularly in the cluster of RferOBP23 and RferOBP107 (**Figure 3**). The phylogenetic tree shows that RferOBP23 is similar to the American palm weevil (APW), R. palmarum OBP4 (RpalOBP4) (Nagnan-Le Meillour et al., 2004), with sound bootstrap support (86%); and also found related to T. yunnanensis, TyunOBP7; T. castaneum, TcasOBP7; RferOBP3213, T. molitor, TmolOBP19; TcasOBP8, H. oblita, HoblOBP1, and OBPs from scarab beetles and the Japanese beetle (**Figure 3**). Similarly, the phylogenetic analysis identified ortholog of RferOBP1768 from other coleopteran insects, which include TyunOBP1 and C. bowringi; CbowOBP19 and CbowOBP5 (**Figure 3**).

#### RNAi-Based Gene Silencing of RferOBPs

We selected RferOBP23, RferOBP107, RferOBPu1, and RferOBP1768 for the RNAi experiments because the first two OBPs were highly expressed and the remaining two were antenna enriched and antenna specific, respectively. Regarding RferOBP1768 silencing, qRT-PCR gene expression data with normalization using multiple control genes (tubulin and β-actin) showed 99.44 and 92.77% silencing (in 21-day-old adult weevils) compared to NFW and NI RPWs, respectively (**Figure 4**). For RferOBP107, RferOBP23, and RferOBPu1, we achieved 85.52, 93.48, and 85.21% silencing, respectively, compared to the NI samples (**Figure 4**, P < 0.001), and we achieved 73.29, 98.25, and 85.44% silencing for RferOBP107, RferOBP23 and RferOBPu1, respectively, compared to the NFW experimental group (**Figure 4**, P < 0.001).

#### Behavioral and Electrophysiological Assays

#### Olfactometer Assay

The silencing of RferOBP1768 and RferOBP23 also resulted in behavioral changes in R. ferrugineus in response to commercial aggregation pheromone in the olfactometer assay. Among RferOBP1768-silenced insects, 31% showed no response, 17% recognized the pheromone, and the remaining 52% moved away from the pheromone, toward the filtered-air arm of the Y-tube olfactometer (F = 43.8, df = 2, and P = 0.0002). In insects with RferOBP23 silencing, a similar pattern of olfactometer response was observed, but with only 40% moving toward the air; 35% showed no response, and 25% responded to the pheromone (F = 19.5, df = 2 and P = 0.002). ANOVA of the percentages of OBP-silenced RPW adults that moved away from the pheromone compared with the control indicated significantly more efficient RferOBP1768 silencing compared to the other RferOBPs (**Figure 5**). The olfactometer assay response was calculated for each group tested, and the results are presented as a percentage of the total number of insects in **Table 1**. Only 17% of RferOBP1768-silenced RPWs were able to detect the aggregation pheromone, which was significantly different from the results for all other experimental groups (F = 81.27; df = 4; P < 0.0001) (**Table 1**, Table S3). Nevertheless, 52% of RferOBP1768-silenced RPWs moved away from the pheromone, which was also significantly different from all other experimental groups (F = 9.66; df = 4; P < 0.0001) (**Table 1**). In the case of RferOBP23-silenced insects, only 25% responded to the commercial aggregation pheromone, also a significant reduction compared to the control (**Figure 5**, Table S3). In contrast, more than 40% of RferOBPu1- and RferOBP107 silenced RPW adults responded to the aggregation pheromone. We selected RferOBP1768-, RferOBPu1-, RferOBP107-, and RferOBP23-silenced RPW adults for EAG studies.

using 1,000 replications and the bootstrap values are indicated at the nodes. The branch containing Drosophila OBP LUSH (DmelLush PDB: 2GTE) was used as an outgroup to root the tree. The RferOBP1768 and RferOBP23 (ABP II) clades are highlighted in green and yellow, respectively. The OBPs from different species were marked with different colors. Phylogenetic tree was visualized with the software FigTree (http://tree.bio.ed.ac.uk/software/figtree/) and branch appearance was colored based on the bootstrap values. Scale: 0.4 amino acid substitutions per site.

#### Electroantennography (EAG)

To validate the altered behavior observed for RferOBP dsRNAinjected R. ferrugineus adults in the olfactometer assay, RPW antennae were excised and exposed to three stimuli (Phe1, Phe2, and EA) in EAG analysis. The antennal response to the different stimuli for each experimental group was recorded and compared to that of the NI group (**Figure 6**). RferOBP1768-silenced RPWs showed significantly reduced responses to Phe1 compared to NI control RPWs (F = 7.52; df = 4, P = 0.005) (**Figure 6**, Table S4). We also noted that RferOBP1768-silenced insects exhibited a comparatively attenuated response to ethyl acetate than the respective controls and other experimental groups (F = 4.45; df = 4, P = 0.025) (**Figure 6**, Table S4). Nevertheless, the response of RferOBP1768-silenced insects to Phe1 was significantly lower

dsRNA-injected groups. The first row shows expression of RferOBPs in the different experimental groups, and the second and third rows show tubulin and β-actin expression (refer Table S1) in the different experimental groups.

than that of RferOBP23-, RefOBP107-, and RferOBPu1-silenced RPW adults, whereas RferOBP107 and RferOBPu1 responded to Phe1 normally (Table S4). Moreover, all RPW experimental groups responded normally to Phe2 (**Figure 6**).

Table S4 presents the data of a comparison of RPW responses to Phe1, Phe2, and ethyl acetate among the different experimental groups and the NI control. In RferOBP1768-silenced insects, the difference in response to Phe1 was significant, with a P-value of 0.015. Interestingly, we observed a moderate reduction in the response to ethyl acetate in RferOBP1768 silenced RPWs compared to the control (P = 0.015). In contrast, RferOBP23-, RferOBP107-, and RferOBPu1-silenced RPWs responded to ethyl acetate normally (**Figure 6**, Table S5). It is worth mentioning that a moderate difference in response to Phe1 was observed for RferOBP23-silenced RPWs; however, based on Tukey's HSD, it was insignificant compared to the response of RferOBP1768-silenced RPWs to Phe1 (**Figure 6**, Table S5).

provided in Table S3.

TABLE 1 | Olfactometer preferences of not-injected, NFW-injected and dsRNA-injected insects toward pheromone, air and no response, as expressed as percentages of the total.


The SEM is provided in parentheses. Significance was measured by using one-way ANOVA followed by LSD analysis, with a significance level of P < 0.05. Homogeneous subsets were identified by Waller-Duncan statistics (α = 0.05), and the results are represented as a, b, c, and d.

#### Structure Modeling of RferOBP1768 and RferOBP23

Based on behavioral trials and electrophysiological recordings, we selected RferOBP1768 for further study and built a structure model using the PYRE 2 web model (Kelley et al., 2015) based on the crystal structure (86% of residues modeled at >90% confidence) of Locusta migratoria odorant binding protein 2 (Zheng et al., 2015)(PDB: 4PT1). As we noted a moderate affinity of RferOBP23 for Phe1, we also created a model based on insect pheromone/odorant binding proteins (OBPs) (88% of residues modeled at > 90% confidence) (Lartigue et al., 2004) (PDB: 3BJH). The modeled RferOBP1768 3-D structure is typical of insect OBPs, comprising six α-helices folded into a very compact and stable globular structure. The predicted binding site of RferOBP1768 corresponds to H82, which is likely involved in pheromone binding (**Figure 7**). The RferOBP1768 protein contains 6 cysteine residues, which can form three disulfide bonds [(C2, C124), (C14, C67), and (C36, C107)]. The predicted binding site of RferOBP23 corresponds to Ile79 and Asp80, which are likely involved in pheromone binding (Figure S4). The RferOBP23 protein contains 6 cysteine residues, which can form three disulfide bonds [(C18, C34), (C61, C65), and (C103, C112)] (Figure S3).

#### RferOBP1768 Expression Analysis in Male and Female R. ferrugineus

The relative expression of RferOBP1768 was low in R. ferrugineus males compared to that in females (expression was normalized using multiple house-keeping genes: tubulin and β-actin). We observed a slight difference in expression patterns between males and females, with mean fold change values of 0.0472 and 0.069, respectively, for male and females (**Figure 9**). However, the values were not significantly different (P = 0.764) and thus did not define a sex-specific variation in RferOBP1768 expression, which supported our tissue-specific expression analysis (**Figure 1**).

## DISCUSSION

As a first step to understanding the function of the large repertoire of OBPs involved in pheromone communication in the highly invasive quarantine pest R. ferrugineus, we first identified antenna-specific RferOBP1768. We then demonstrated that dsRNA injection caused a significant reduction in the

FIGURE 6 | Comparison of EAG responses from treated and non-treated samples to three different stimuli, as represented as three groups. The values represent the amplitudes of signals in mV, and the error bars represent the SEM. The detailed statistical analysis is provided in Table S4.

electrophysiological recording of the response to a major aggregation pheromone compound, (4RS,5RS)-4-methylnonan-5-ol (ferrugineol), leading to altered behavior that ultimately resulted in the failure to sense the pheromone in a behavioral assay. The results of the behavioral assay regarding the response to ferrugineol supported the physiological role of RferOBP1768 as the ferrugineol-binding protein that aids in the capture and transport of aggregation pheromones to receptors in the palm weevil R. ferrugineus. In contrast, no significant differences in electrophysiological response to ferrugineol and a minor pheromone compound, 4-methyl-5-nonanone (ferruginone), or to a kairomone, ethyl acetate, were reported for other highly expressed orthologous OBPs. With 92–94% OBP silencing achieved with dsRNA-injected R. ferrugineus, our study demonstrates that pheromone communication disruption can occur through RferOBP1768 silencing. Our study results have an application in the field of OBP-based biosensors, and RferOBP1768 is the most promising candidate for fabricating biosensors to detect ferrugineol in "reverse chemical ecology" approaches (Leal et al., 2008; Leal, 2017). RNAi and electrophysiological approaches are widely used and well-accepted methods for the characterization of OBPs, especially PBPs, in insects (Xu et al., 2005; Laughlin et al., 2008; Biessmann et al., 2010; Pelletier et al., 2010). In addition, the use of RNAi and electrophysiological approaches in characterizing OBPs is well documented in mosquitoes (Biessmann et al., 2010; Pelletier et al., 2010), Drosophila (Xu et al., 2005; Laughlin et al., 2008); Aphis gossypii (Rebijith et al., 2016); Adelphocoris lineolatus (Zhang et al., 2017); and Helicoverpa armigera (Dong et al., 2017). Such attempts have confirmed the role of OBPs in olfaction, as carriers of hydrophobic odorants and pheromones through the aqueous environment of the sensillum lymph to ORs (Leal, 2013). Regardless, no RNAi studies to date related to the role of OBPs have been reported in beetles, though several studies have been performed to characterize odorant co-receptors by RNAi and electrophysiological approaches (Soffan et al., 2016; Zhang et al., 2016). To the best of our knowledge, the current study is the first attempt to specifically characterize aggregation pheromone-specific OBPs using a gene silencing approach in a beetle.

We previously identified 38 OBPs and grouped R. ferrugineus OBPs into different OBP-subfamilies (Antony et al., 2016) to provide a basis for evolutionary and functional analyses of OBPs in palm weevil. In previous degenerate PCR approaches, two OBPs were identified from a species related to RPW, the APW R. palmarum (Nagnan-Le Meillour et al., 2004), and more recently, tissue-specific expression profiling was reported for 11 OBPs from R. ferrugineus (Yan et al., 2016). OBPs have been reported from a wide range of insect species, and the number of OBPs in some species with sequenced genomes ranges from a 12 in ant species, at least 35 putative OBPs in Drosophila, and 44 in silkworms to more than 100 in certain mosquitoes (Hekmat-Scafe et al., 2002; Gong et al., 2009; Smith et al., 2011; Vieira and Rozas, 2011; Manoharan et al., 2013). We used qRT-PCR and tissue-specific expression patterns to select candidate OBPs for RNAi. The tissue-specific expression analysis revealed only one antenna-specific OBP gene, RferOBP1768, with all other OBPs showing expression in other specific tissues or in all tissues. The fundamental role of OBPs in olfaction is supported by several studies demonstrating that OBPs involved in pheromone transport are specifically expressed in the antenna (Shanbhag et al., 2001; Pelosi et al., 2006). This approach has been applied to several lepidopteran insects to identify the PBPs that are uniquely expressed in antennae (Vogt and Riddiford, 1981; Vogt et al., 1991; Nikonov et al., 2002; Pelosi et al., 2006; Zhou, 2010; Sun et al., 2013; Jiao et al., 2016).

Expression of OBPs in different tissues may be related to their roles in other physiological functions in that tissue; however, it has also been proposed that the type of sensillum where an OBP is expressed, rather than the organ, might define the role of the protein in taste or olfaction (Pelosi et al., 2006; Zhou, 2010). We also selected two highly expressed OBPs for characterization because the majority of insect OBPs studied to date are highly expressed in chemosensory structures, including antennae (Brito et al., 2016), and we thus initially presumed that these OBPs might be involved in pheromone detection in R. ferrugineus. We did not select the second-most highly expressed R. ferrugineus OBP, RferOBP77, because this candidate was found not to be expressed in the female antenna (**Figure 1**) and an earlier report indicated that male R. ferrugineus-produced aggregation pheromone can attract both female and male RPWs (Hallett et al., 1993; Oehlschlager et al., 1995). Nevertheless, preliminary studies on the scarab beetle A. octiescostata showed expression of PBPs in both sexes (Nikonov et al., 2002). Phylogenetic analysis of R. ferrugineus OBPs revealed that the RferOBP1768 clade also contains other R. ferrugineus OBPs, such as RferOBP19755, RferOBP12511, and RferOBP1689; however, we did not select these candidates for further study because the first two showed very low expression and the last showed ubiquitous expression (**Figure 1**). Although we ranked RferOBP1689 as the seventhmost highly expressed candidate, we eliminated it from further study because we observed expression in wings (**Figure 1**). We included RferOBPu1 in silencing experiments due to its antennaenriched expression and because this candidate shows high identity to RferOBP1768 (48% amino acid sequence identity and 99% bootstrap support), though the predicted protein structures and binding sites (H82 for RferOBP1768 and H81 for RferOBPu1) are surprisingly similar. Regardless, our results showed that RferOBPu1-silenced RPWs respond to Phe1, Phe2, and EA normally (**Figure 8**). Based on the observed sequence identity and similar binding sites, we assume that RferOBP1768 would function as a ferrugineol-specific OBP and be able to activate pheromone-sensitive neurons, whereas RferOBPu1 would act as an antagonist-binding protein and be able to activate different neurons or bind to non-pheromone ligands for other functions.

The olfactometer assay showed significantly altered behavior in RferOBP1768-silenced R. ferrugineus, and EAG recordings indicated that RferOBP1768 silencing in palm weevils decreases the insect's strong preference for the aggregation pheromone ferrugineol. We observed a perfect correlation between the reduction in RferOBP1768 transcript levels and modest antennal responses to the pheromone, and the simplest explanation is that RferOBP1768 may be involved in the detection of ferrugineol. Moreover, we observed slight differences in expression of RferOBP1768 in both sexes (**Figure 9**), and the higher expression level in R. ferrugineus females indicates different roles in pheromone perception for males and females. A similar observation of differential expression patterns of key OBPs in male and female insects has been reported previously (Maida et al., 2005; Campanini et al., 2017). The differential detection of ferrugineol in males and females associated with distinct sexual behaviors might be because more RferOBP1768 is required in females, leading to differentiation in expression level. Another possibility is that females may be able to recognize ferrugineol as a species-specific pheromone to elicit important ecological and behavioral consequences, and hence a different form of olfactory perception occurs in female R. ferrugineus. There is no femaleproduced sex pheromone reported thus far in R. ferrugineus, and studies have shown that male-produced aggregation pheromone can attract females to the vicinity and ultimately facilitate mating (Hallett et al., 1993; Oehlschlager et al., 1995; Kaakeh, 1998; Abdel-Azim et al., 2012; Inghilesi et al., 2015).

The results of our study also indicate that RferOBP23 can detect the R. ferrugineus aggregation pheromone; however, based on EAG recordings, this result was not significant (P = 0.153) compared to that of RferOBP1768. The results of behavioral trials and EAG clearly proved a significantly higher discriminatory affinity for RferOBP1768 compared to RferOBP23 toward ferrugineol (**Figure 8**, Table S5). However, considering the ability of RferOBP23 to detect ferrugineol, we assume that this OBP can accommodate ferrugineol in addition to other unknown ligands, which need to be determined. As RferOBP23 is a highly expressed OBP in R. ferrugineus, its broad binding abilities indicate that it may act as a general odorant binding protein (GOBP) to carry out a variety of functions. In addition, both OBPs may be associated with the

RferOBPs represented as waveforms. Measurements were performed at 10 mV and 15-s intervals. 4-Methyl-5-nonanol was Phe1, and 4-methyl-5-nanone was Phe2. Both pheromone compounds were dissolved in hexane. EA indicates ethyl acetate (kairomone).

detection of ferrugineol, as reported in the case of H. armigera, in which both HarmPBP1 and HarmPBP2 are responsible for the detection of the major sex pheromone component, Z11–16:Ald (Dong et al., 2017). Studies have also shown that OBPs undergo specific conformational changes upon binding to their ligand molecules, and only in selected cases do such changes enable the OBP to interact with the OR and generate a physiological response (Laughlin et al., 2008). Thus, GOBPs that do not undergo suitable conformational changes may not be able to trigger the subsequent physiological response. Several previous studies have reported the phenomenon of OBPs exhibiting a broad spectrum of binding (Maida et al., 2000; Plettner et al., 2000; Campanacci et al., 2001; Leal et al., 2005a,b; Zhou, 2010). Regardless, there are studies, mostly in lepidopteran insects, suggesting that PBPs can selectively bind to sex pheromone components produced by females; the pheromone (E,Z)-10,12 hexadecadienol (bombykol) is the specific ligand for B. mori PBP (BmorPBP1), and the pheromone component (E,Z)-6,11 hexadecadienal is the specific ligand for Antheraea polyphemus PBP (ApolPBP1) (Sandler et al., 2000). However, studies have also demonstrated that OBPs can also bind to a wide range of odorant chemicals (Honson et al., 2003, 2005; Zhou, 2010; Zhou et al., 2010; Venthur et al., 2014) and that different PBPs can bind to the same sex pheromone component (Campanacci et al., 2001; Guo et al., 2012; Gu et al., 2013). Despite studies to support selective binding, a full understanding of the discriminative ability of OBPs remains elusive (Pelosi et al., 2014, 2017; Brito et al., 2016). Nevertheless, all the OBP functional studies mentioned above are based solely on lepidopteran and dipteran insects (Zhou, 2010; Leal, 2013; Pelosi et al., 2014, 2017; Brito et al., 2016), and hence, the results may not hold in the case of coleopteran insects. In R. ferrugineus, the male-produced aggregation pheromone ferrugineol can equally attract both female and male adult weevils (Hallett et al., 1993; Oehlschlager et al., 1995); hence, GOBP/PBP may not be specifically involved in both sexes. Our results are consistent with the idea that RferOBP1768 is antenna specific, and our phylogenetic analysis and structural analyses classified RferOBP1768 in the Minus-C category (Hekmat-Scafe et al., 2002; Gong et al., 2009). In addition to ferrugineol, RferOBP1768 also exhibits affinity toward a kairomone, ethyl acetate, as we observed in the EAG recording of dsRNA-injected weevils (**Figure 8**). These broad binding abilities indicate that RferOBP1768 may not act as a GOBP/PBP for specific pheromone binding; however, there is no report on the functional identification of OBPs in coleopteran insects involved in aggregation pheromone detection for comparison with the results in R. ferrugineus. Although the T. castaneum genome is available and TcasOBP6 and TcasOBP9 were found to be similar to RferOBP1768 (Figure S2), no specific role of OBPs has yet been proposed; thus, it is difficult to suggest a common function of these clustered OBPs in the family. Similarly, C. bowringi; CbowOBP5 and CbowOBP19 (Li X. et al., 2015) and T. yunnanensis; TyunOBP1 (Liu et al., 2018) were found to be similar to RferOBP1768 (**Figure 3**), no functional role of these OBPs has yet been proposed. However, it is worth noting that the T. yunnanensis transcriptome data revealed the presence of 45 OBPs, from which TyunOBP1 was more antennal-specific and significantly expressed in the antennae (Liu et al., 2018).

In the current study, we focused on RferOBP23, RferOBP107, RferOBP1768, and RferOBPu1 based on results obtained in tissue-specificity studies and relative OBP expression analysis. Phylogenetic analysis revealed that the ABPII subfamily of OBPs

7.708).

containing RferOBP23, RferOBP107, and RferOBP3213 from R. ferrugineus form a clade with PBPs from scarab beetle and Japanese beetle (Wojtasek et al., 1998; Nikonov et al., 2002) (**Figure 3**), with RferOBP23 and RferOBP3213 showing more than 50% amino acid identity with PBPs from these beetles (Figure S4). Scarab beetle and Japanese beetle PBPs are reported to be involved in detecting the sex pheromone enantiomers (S) japonilure and (R)-japonilure, respectively, based on the singleneuron technique and are the only PBPs identified thus far from Coleoptera (Wojtasek et al., 1998). As per phylogenetic analysis, RferOBP3213 is a promising candidate for testing in silencing experiments; however, based on its low expression in the snout, leg and abdomen in tissue-specific expression analysis and its low expression (RQ-value 0.39) in qRT-PCR analysis, we did not include further evaluate this candidate. It is interesting to note that the RferOBP3199, which is ubiquitously expressed in R. ferrugineus (**Figure 1**); we identified a putative ortholog in another curculionid, C. buqueti; CbuqOBP1 (Yang et al., 2017a) (97% bootstrap support, **Figure 3**) and this putative PR reported to be related to the recognition of dibutyl phthalate, a sex pheromone analog in C. buqueti (Yang et al., 2017b). Based on our phylogenetic analysis, we speculate that in Curculionidae such genes may have the same ancestral gene, and the possibility is that the OBP expansions facilitated the adaptive evolution of a variety of specialized functions among different species.

Rhynchophorus ferrugineus has recently received greater attention due to its invasiveness and quarantine pest status. Conventional methods have proven ineffective for the management of palm weevil, leading to proposals of synthetic biology approaches intended at disrupting pheromone communication, given that olfaction interference has the potential to interrupt critical behaviors such as host and mate selection, ultimately disrupting reproductive success and causing weevil population decline (Antony et al., 2016; Soffan et al., 2016). We previously reported RferOrco silencing, and together with OBP silencing in R. ferrugineus via dsRNA injection, this approach is promising for the disruption of pheromone communication in R. ferrugineus (Soffan et al., 2016). To enable use of the RPW RNAi technique, RferOBP1768 and RferOrco dsRNA delivery via feeding or effective delivery systems such as synthetic nanoparticle and engineered microorganisms (Baum et al., 2007; Kolliopoulou et al., 2017), the generation of transgenic bacteria that express dsRNA (Tian et al., 2009), the chemical synthesis of siRNA (Gong et al., 2011) or the application of dsRNA in a spray form to facilitate its spread might offer excellent future prospects for controlling this invasive pest. Another promising area is the development of OBP-based biosensors for the detection of odorants. Such a biotechnological application of OBPs against R. ferrugineus is yet to be explored, and thus our identification of a ferrugineol-specific OBP from RPW holds great promise for the development of insect behavioral attractants or repellents or artificial biosensors. Considering that pheromone communication is an important aspect of R. ferrugineus attack of palm trees, where individual insects use male aggregation pheromone to find trees and coordinate a group attack that eventually leads to palm tree death, understanding the key OBP involved in this mechanism is a significant achievement for the date palm industry. Although substantial antennal transcriptome data are available for coleopteran insects (Dippel et al., 2014; Liu et al., 2015, 2018; Li X. et al., 2015; Li X.M. et al., 2015; Li K. et al., 2017; Li L. et al., 2017), PBPs from scarab beetle and Japanese beetle are the only coleopteran OBPs identified thus far (Wojtasek et al., 1998; Nikonov et al., 2002). However, their functional characterization has not been reported, and hence there is much work needed in exploring the olfactory mechanism in beetles and the pattern of OBP relatedness between beetles. Further aspects of the identified candidate OBPs, such as structure and ligand-binding capability, also need to be explored.

#### DATA AVAILABILITY STATEMENT

All relevant data are within the paper and its Supporting Information files. The OBP nucleotide sequences can be obtained from the Transcriptome Shotgun Assembly project DDBJ/EMBL/GenBank under the accession number GDKA00000000. The OBP contig names are mentioned in abbreviated form; for example, the RferOBP23 GenBank acc. no. is GDKA01000023, and the RferOBP1768 GenBank acc. no. is GDKA01001762. The full length sequences reported in this paper have been deposited in the GenBank database (accession nos.: RferOBP1768, MH026102; RferOBPu1, MH026103; RferOBP23, MH026104; RferOBP107, MH026105, and RferOBP3213, MH026106).

## AUTHOR CONTRIBUTIONS

BA conceived the study and acquired the grant, also participated in its design, coordination and supervision; SA paid the expenses for the weevil collection and culture; JJ and BA carried out the laboratory experiments and analyzed the data; BA wrote the paper with contributions from JJ, and all authors read and approved the final manuscript.

## FUNDING

Funding for this research (awarded to BA) (Grant no. KACST-NSTIP 12-AGR2854-02) was provided by the National Plan for Science, Technology and Innovation (MAARIFAH) of King Abdul Aziz City for Science and Technology (KACST), Kingdom of Saudi Arabia.

## ACKNOWLEDGMENTS

We thank the KSU Deanship of Scientific Research, Research Chair Program, Saudi Arabia. We are grateful to Mehmoud Abdelazim and Samy M. Mustafa of CDPR for sharing the RPW culture. JJ is grateful for the researcher stipend provided by the 12-AGR2854-02 project.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00252/full#supplementary-material

## REFERENCES


binding proteins of Helicoverpa armigera Involved in the perception of the main sex pheromone component Z11–16: Ald. J. Chem. Ecol. 43, 207–214. doi: 10.1007/s10886-016-0816-6


male and female silk moths, Bombyx mori. J. Neurocytol. 34, 149–163. doi: 10.1007/s11068-005-5054-8


ant (Linepithema humile). Proc. Natl. Acad. Sci. U.S.A. 108, 5673–5678. doi: 10.1073/pnas.1008617108


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Antony, Johny and Aldosari. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Distinct Subfamilies of Odorant Binding Proteins in Locust (Orthoptera, Acrididae): Molecular Evolution, Structural Variation, and Sensilla-Specific Expression

#### Xingcong Jiang<sup>1</sup> , Jürgen Krieger <sup>2</sup> , Heinz Breer <sup>1</sup> and Pablo Pregitzer <sup>1</sup> \*

*1 Institute of Physiology, University of Hohenheim, Stuttgart, Germany, <sup>2</sup> Department of Animal Physiology, Institute of Biology/Zoology, Martin Luther University Halle-Wittenberg, Halle, Germany*

#### *Edited by:*

*Shuang-Lin Dong, Nanjing Agricultural University, China*

#### *Reviewed by:*

*Paolo Pelosi, University of Pisa, Italy Dan-Dan Zhang, Lund University, Sweden Long Zhang, China Agricultural University, China*

*\*Correspondence: Pablo Pregitzer p\_pregitzer@uni-hohenheim.de*

#### *Specialty section:*

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

*Received: 20 July 2017 Accepted: 11 September 2017 Published: 26 September 2017*

#### *Citation:*

*Jiang X, Krieger J, Breer H and Pregitzer P (2017) Distinct Subfamilies of Odorant Binding Proteins in Locust (Orthoptera, Acrididae): Molecular Evolution, Structural Variation, and Sensilla-Specific Expression. Front. Physiol. 8:734. doi: 10.3389/fphys.2017.00734* Odorant binding proteins (OBPs) play an important role in insect olfaction, facilitating transportation of odorant molecules in the sensillum lymph. While most of the researches are concentrated on Lepidopteran and Dipteran species, our knowledge about Orthopteran species is still very limited. In this study, we have investigated OBPs of the desert locust *Schistocerca gregaria*, a representative Orthopteran species. We have identified 14 transcripts from a *S. gregaria* antennal transcriptome encoding SgreOBPs, and recapitulated the phylogenetic relationship of SgreOBPs together with OBPs from three other locust species. Two conserved subfamilies of classic OBPs have been identified, named I-A and II-A, exhibiting both common and subfamily-specific amino acid motifs. Distinct evolutionary features were observed for subfamily I-A and II-A OBPs. Surface topology and interior cavity were elucidated for OBP members from the two subfamilies. Antennal topographic expression revealed distinct sensilla- and cellular- specific expression patterns for SgreOBPs from subfamily I-A and II-A. These findings give first insight into the repertoire of locust OBPs with respect to their molecular and evolutionary features as well as their expression in the antenna, which may serve as an initial step to unravel specific roles of distinct OBP subfamilies in locust olfaction.

Keywords: locust, *Schistocerca gregaria*, odorant binding protein, evolution, structure, sensilla

#### INTRODUCTION

In insects, the process of olfactory signal processing begins in hair-like cuticle appendages, called sensilla, located mainly on the antennae and palps (Steinbrecht, 1996; Hansson and Stensmyr, 2011; Suh et al., 2014). Olfactory sensory neurons (OSNs) project their dendrites into the lumen of the sensillar hairs, which is filled with sensillum lymph (Hansson and Stensmyr, 2011; Suh et al., 2014). The hydrophobic odorant molecules enter the sensillum via the porous cuticle and have to pass the aqueous lymph till reaching the chemosensory membrane of the sensory neurons (Vogt et al., 1999; Leal, 2013; Suh et al., 2014). This passage is thought to be mediated by small soluble proteins enriched in the sensilla lymph, the so called odorant binding proteins (OBPs), which are produced and secreted by accessory cells (Pelosi et al., 2006, 2017). OBPs are polypeptides comprised of ∼110–200 amino acids; usually they exhibit a considerable degree of sequence divergence. Based on the number of conserved cysteine (C)-residues, several subtypes are discriminated. Whereas, the pattern of six conserved C-residues represents a hallmark of classic OBPs (Pelosi et al., 2006), OBPs with more or with less C-residues are designated as plus-C and minus-C OBPs (Zhou et al., 2004; Foret and Maleszka, 2006). In addition, atypical OBPs have been classified which may originate from a fusion of two classic OBPs (Xu et al., 2003; Vieira and Rozas, 2011). Typically, the tertiary structure of insect OBPs consists of six α-helices forming an interior binding cavity. This structure is maintained and stabilized by disulfide bridges formed by conserved C-residues (Leal et al., 1999; Scaloni et al., 1999; Sandler et al., 2000). However, OBP structures with more than six helices have been reported (Horst et al., 2001; Lagarde et al., 2011). It is also proposed that the C-terminal domain that is variable in length can spatially interfere with the interior binding cavity and thus may affect the ligand binding mechanism (Damberger et al., 2000; Horst et al., 2001; Tegoni et al., 2004; Pelosi et al., 2017).

Most of our current knowledge of insect OBPs is based on studies of species from the taxa Lepidoptera and Diptera (Hekmat-Scafe et al., 2002; Vogt et al., 2002; Leal, 2013; Pelosi et al., 2017). The desert locust, Schistocerca gregaria is a representative of the taxa Orthoptera, which is quite distant from the orders Lepidoptera and Diptera on the phylogenetic scale (Wheeler et al., 2001; Vogt et al., 2015) and as hemimetabolous insects their developmental process differ significantly from that of holometabolous insects. Very little is known about OBPs in Orthoptera; only a limited number of sequences have recently been reported for a few locust species: Locusta migratoria (Ban et al., 2003; Xu et al., 2009; Yu et al., 2009), Oedaleus asiaticus (Zhang et al., 2015), and Ceracris kiangsu. Information about the expression of locust OBPs in the olfactory sensilla is limited to LmigOBP1, which was found to be expressed in sensilla trichodea and sensilla basiconica (Jin et al., 2005). Concerning another olfactory sensillum type, the sensilla coeloconica, a possible expression of OBPs has rarely been documented even in holometabolous insect species (Larter et al., 2016). Incidentally, the crystal structure of locust OBPs has only been resolved for LmigOBP1, which establishes a unique seven-α-helices structure (Zheng et al., 2015). The possibility of structural differences between locust OBPs is still an open question.

In order to extend our knowledge about OBPs in Orthopteran locust species, in the current study we have performed a systematic characterization of locust OBPs with respect to the molecular evolution, structural variation and sensilla-specific expression. Based on the OBP sequences of S. gregaria newly identified from an antennal transcriptome and the documented OBP sequences from other locust species, we conducted a phylogenetic analysis of the current locust OBP repertoire. The emerging two subfamilies of classic OBPs were compared for sequence divergence, selection pressure and variation of the predicted tertiary structure in detail. Analysis of the topographic expression pattern revealed that the molecular and phylogenetic distinctness between the two subfamilies are accompanied by a sensilla-specific expression pattern.

## MATERIALS AND METHODS

## Identification of *S. gregaria* OBP Transcripts

A S. gregaria antennal transcriptome database was generated comprising a total of 55,060 contigs with an N50 of 2,223 bp. The strategy of homology-mining was adopted to identify the candidate OBP transcripts. We retrieved documented OBPs from different insect species including Anopheles gambiae (AgamOBPs, Diptera), Apis mellifera (AmelOBPs, Hemiptera), Drosophila melanogaster (DmelOBPs, Diptera), Tribolium castaneum (TcasOBPs, Coleoptera), Acyrthosyphon pisum (ApisOBPs, Hemiptera), Bombyx mori (BmorOBPs, Lepidoptera) (Vieira and Rozas, 2011), Blattella germanica (BgerOBPs, Blattaria) (Niu et al., 2016), and Zootermopsis nevadensis (ZnevOBPs, Isoptera) (Terrapon et al., 2014), as well as from three other locust species, including L. migratoria (LmigOBPs) (Ban et al., 2003; Yu et al., 2009), O. asiaticus (OasiOBPs) (Zhang et al., 2015), and C. kiangsu (CkiaOBPs). Using the collected sequences as queries, we conducted a local tBLASTx search on BioEdit 7.2.5 against the transcriptome database with an E-value < 10−<sup>5</sup> . Annotation of the screened contigs was inspected by performing tBLASTx and BLASTp search against non-redundant (nr) protein database in NCBI (Bethesda, MD, USA). The extracted contigs which putatively encode OBPs were in turn used as new queries to identify additional candidates using tBLASTx and BLASTp methods. Open reading frames in the identified OBP transcripts were inspected by Genamics Expression (Hamilton, New Zealand). Accession numbers for the newly identified SgreOBPs and other locust OBPs are deposited in the Supplementary Material.

## Characterization of Consensus Amino Acid Motifs

Signatures of sequence divergence underlying locust subfamily I-A and II-A OBPs were addressed by identifying consensus amino acid motifs. Toward that goal, the online MEME SUITE v. 4.11.2 (http://meme-suite.org/tools/meme) was used (Bailey et al., 2009), with the default setting (motif width: 6–50 amino acids; motif distribution: zero or one occurrence per sequence). The output comprised six consensus motifs which was ascertained to be sufficient to recapitulate the sequence information of subfamily I-A and II-A. The identified six motifs were also utilized to target sequences of the locust OBP repertoire to obtain the motif match degree (match E-value) using MAST module (Motif Alignment and Search Tool) implemented in MEME SUITE. The motif match E-value assesses statistical significance of the consensus motif toward a targeted sequence based on its log likelihood level and the occurrence frequencies of background amino acids. The default statistical significant threshold setting was e−<sup>5</sup> .

#### Phylogenetic Analysis

The OBP amino acid sequences from four hitherto documented locust species were utilized to recapitulate the phylogenetic relationship: 16 from L. migratoria, 15 from O. asiaticus, 7 from C. kiangsu and the currently identified 14 candidates from S. gregaria. Amino acid sequences of OBPs from the four locust species are deposited in the Supplementary Material. The predicted signal peptide (SP) on the N-terminal domain was deleted before the sequences being further investigated due to two reasons: (1) SP is cut off in post-translational modification when the protein is secreted into the sensillum lymph; (2) SP exhibits a certain degree of sequence divergence but may contain limited bio-information (Vieira et al., 2007). Prediction of SP was based on SignalP 4.1 (http://www.cbs.dtu.dk/services/ SignalP/) (Petersen et al., 2011). Multiple sequence alignments were conducted by MAFFT v. 7 (http://mafft.cbrc.jp/alignment/ server/) using the algorithm E-INS-I, which is accuracy favored and is suitable for sequences with multiple conserved domains (Katoh and Standley, 2013). After the alignment, Gblocks v. 0.91b (http://molevol.cmima.csic.es/castresana/Gblocks\_server. html) was used to inspect the poorly aligned sites and divergent regions (Castresana, 2000). To search an optimal amino acid substitution model, we chose the Find Best Protein Model implemented in MEGA 6.0 which performs a comprehensive parametric assessment (e.g., BIC scores, AICc value, lnL value) (Tamura et al., 2013). The Whelan and Goldman model (WAF), discrete GAMMA distribution (G) and an assumed fraction of evolutionary invariable sites (I) was considered to describe the substitution best. RAxML v. 8.2.9 implemented in the CIPRES Science Gateway (https://www.phylo.org/) was used for the locust OBP phylogeny construction (Miller et al., 2012; Stamatakis, 2014). A search of best scoring maximum likelihood tree (-f a) was launched, supported by 1,000 rapid bootstrap iterations (autoMRE based bootstopping criterion). The generated maximum likelihood tree was graphically edited by FigTree v. 1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/). A similar strategy was applied to analyze the phylogenetic relationship between locust OBPs and OBPs from eight other insect species. In brief, SignalP, MAFFT, and Gblocks were used to prepare the multiple sequence alignment; RAxML was responsible for building the maximum likelihood tree (-f a, 1,000 iteration) using the proposed best fitting substitution model (WAG+G+I) by MEGA.

#### Selection Constraint on Locust OBP Repertoire

The nucleotide coding sequences underlying the locust OBP repertoire (see Supplementary Material) were aligned in accordance with the multiple sequence alignment from the above mentioned phylogenetic analysis using TranslatorX (http://translatorx.co.uk/). The sequence order of alignment was guided by the constructed phylogenetic tree mentioned above. The signatures of selection regime acting on sequences of the locust OBP phylogeny were estimated by resolving three principle concepts: the non-synonymous substitution rate (dN), synonymous substitution rate (dS) and the ω rate (dN/dS). Toward that, HyPhy batch program was utilized which implements maximum likelihood estimate and post-likelihood ratio test (Kosakovsky Pond et al., 2005). A local fit model (MG94xREV\_3x4 substitution model) was adopted (Kosakovsky Pond et al., 2009), and each single branch in the locust OBP phylogeny was assigned with a unique set of dN and dS values, assuming the branch-to-branch variant ω rates. To support the local fit model, we additionally conducted a coarse estimate of the ω rate using the alternative global fit model, assuming invariable ω rate shared by different phylogenetic branches. A likelihood ratio test compared the results obtained from two distinct models, and strongly favored the local fit model (P = 10−<sup>3</sup> ). Normality distribution of dN, dS, and the ω rates was assessed by D'Agostino-Pearson test, and the statistical difference was evaluated by non-parametric Mann-Whitney U-test. GraphPad Prism 5.0 was used to analyze the data and generate the diagrams (San Diego, CA, USA).

## Synthesis of Riboprobes For *in Situ* Hybridization

The coding sequences of six SgreOBPs from locust OBP subfamily I-A and II-A were amplified, sequenced and then cloned into the pGEM-T vectors (Invitrogen) for subsequent transcription. Linearized pGEM-T vectors carrying SgreOBPs coding sequences were utilized to synthesize digoxigenin (Dig) and biotin (Bio) labeled anti-sense and sense RNA probes using the T7/SP6 RNA transcription system (Roche, Germany). The sense (s) and antisense (as) primers used for amplication of the SgreOBP sequences were:

SgreOBP1 s, ctgggacgtcaacatgaaact; SgreOBP1 as, aatgcacgaactaccaggctg; SgreOBP5 s, ggccgcgccgtcttctcataagga; SgreOBP5 as, cggccctggcgcagcacctgcatt; SgreOBP6 s, acagcacaccaccgtcacac; SgreOBP6 as, ggtgcttgcttgaagaggcac; SgreOBP10 s, gcgtatcacccggctgtgta; SgreOBP10 as, agtctcacctctgccagcga; SgreOBP11 s, tggaccgcacgacaacaaca; SgreOBP11 as, cgatagcgtatgccctttcac; SgreOBP14 s, ctgttgggtgcagtcctgtt; SgreOBP14 as, gtcgtgacagctcctccactg

#### *In Situ* Hybridization

Antennae of adult S. gregaria were dissected and embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek Europe, The Netherlands). Cryosections at 12 µm were thaw mounted on SuperFrost Plus slides (Menzel-Gläser, Braunschweig, Germany) at −21◦C (Jung CM300 cryostat). RNA in situ hybridization **(**ISH) was conducted as previously reported (Yang et al., 2012; Guo et al., 2013; Jiang et al., 2016). Section were fixed (4% paraformaldehyde in 0.1 M NaHCO3, pH 9.5) at 4◦C for 22 min. The following consecutive steps were conducted at room temperature: a wash for 1 min in PBS (phosphate buffered saline = 0.85% NaCl, 1.4 mM KH2PO4, 8 mM Na2HPO4, pH 7.1), an incubation for 10 min in 0.2 M HCl, another wash for 1 min in PBS, an incubation for 10 min in acetylation solution (0.25% acetic anhydride freshly added in 0.1 M triethanolamine) and washes for three times in PBS (3 min each). Sections were prehybridized for 1 h at 60◦C in hybridization buffer (50% formamide, 5× SSC, 50 µg/ml heparin, and 0.1% Tween-20). 100 µl hybridization solution containing the labeled RNA in hybridization buffer was placed onto the tissue section. A coverslip was placed on top and slides were incubated in a moister box at 60◦C overnight (18–20 h). After hybridization, slides were washed twice for 30 min in 0.1× SSC at 60◦C, then each slide was treated with 1 ml 1% blocking reagent (Roche) for 40 min at room temperature.

Visualization of Dig-labeled probe hybridizations was achieved by using an anti-Dig alkaline phosphatase (AP) conjugated antibody (1:500, Roche) and NBT/BCIP substrate. Antennal sections were analyzed on a Zeiss Axioskope2 microscope (Zeiss, Oberkochen, Germany) equipped with Axiovision software. For two-color FISH visualization of hybridized probes was performed by using an anti-Dig APconjugated antibody in combination with HNPP/Fast Red (Roche) for Dig-labeled probes and an anti-biotin streptavidin horse radish peroxidase-conjugate together with fluoresceintyramides as substrate (TSA kit, Perkin Elmer, MA, USA) for Bio-labeled probes. Sections from FISH experiments were analyzed with a Zeiss LSM510 Meta laser scanning microscope (Zeiss, Oberkochen, Germany). Confocal images stacks were processed by ZEN 2009 software. The pictures shown represent projections of optical planes selected from confocal image stacks. For clear data presentation, images were only adjusted in brightness and contrast. Antennal sections of both male and female antennae were analyzed using each generated probe. No obvious difference between sexes regarding the labeling intensity and labeling pattern was observed. Thus, only the images of male antenna were adopted in this study.

#### Structure Modeling and Electrostatic Potential

In silico simulation of OBP tertiary structure was performed by I-TASSER server (http://zhanglab.ccmb.med.umich.edu/I-TASSER/) (Roy et al., 2010), which implements the iterative template threading refinement making full use of established homologous protein structures. PyMol was used to visualize the simulated protein tertiary structures (DeLano, 2002). The molecular surface was solvent excluded and the solvent radius was set 1.4 as default. APBS plug (Unni et al., 2011) implemented in PyMol was employed to calculate the surface electrostatic potentials in the range of −6 to 6 kT/e, and was presented as blue-red hue gradient.

## RESULTS

## Identification, C-Skeleton Pattern and Phylogenetic Relationship of Locust OBPs

Toward an identification of OBPs from S. gregaria and a comprehensive characterization of OBPs in locust species, we have performed a homology-based data mining of an antennal transcriptome which resulted in 14 transcripts putatively encoding SgreOBPs. Subsequently, a multiple sequence alignment was conducted addressing the amino acid sequences of the newly identified SgreOBPs together with hitherto documented OBPs from three other locust species: 16 from L. migratoria, 15 from O. asiaticus and 7 from C. kiangsu. Several OBP subtypes could be categorized based on the number of conserved C-residues (**Figure 1A**). First, 33 OBPs were classified as classic OBPs comprising six conserved C-residues, the hallmark of classic OBPs. Second, 15 OBPs were categorized in two types of plus-C OBPs harboring more than six conserved C-residues. Finally, only one minus-C OBP with less than six conserved C-residue and three atypical OBPs with extraordinary long stretches between conserved C1 and C2 were identified.

As a next step, we analyzed the phylogenetic relationship of the locust OBP repertoire by constructing a phylogenetic tree utilizing the maximum likelihood algorithm and bootstrap iterations. The emerging picture indicated that the repertoire of locust OBPs can be divided into four major families (I–IV), which apparently split at the internal nodes (**Figure 1B**). We further classified three additional subfamilies (I-A, II-A, and III-A), based on the presence of higher bootstrap support (above 80%) on the divergent nodes. It is noteworthy that subfamily I-A and II-A both represent classic OBPs and each subfamily apparently comprise three distinct groups with 3–4 orthologous OBPs from different locust species (**Figure 1B**). Within each subfamily, the sequence identity between OBPs from different groups ranged from 28 to 35%; OBP members within each ortholog group exhibit generally above 80% sequence identity. Incidentally, plus-C OBPs type-A converged onto a subfamily III-A and segregated from their counterparts plus-C OBPs type-B and classic OBPs. Together, the data indicate a considerable degree of orthology in the OBP repertoires across the four analyzed locust species and no marked species-specific expansion within the OBP phylogeny.

#### Elucidation of Subfamily-Specific Consensus Amino Acid Motifs

To better elucidate the clustering regime of individual subfamilies, we analyzed the consensus amino acid motifs characteristics underlying subfamily I-A and II-A OBPs. The local consensus motifs were calculated by recapitulating repeatedly occurring sequence patterns along OBP sequences. Six consensus motifs with various widths were identified and localized at distinct positions (**Figure 2**). The motif 1 and motif 2 appeared as common motifs in all OBPs of both subfamilies, whereas the other four motifs specifically fit either the repertoire of subfamily I-A OBPs (motif combination 4 and 6) or the repertoire of subfamily II-A OBPs (motif combination 3 and 5). Therefore, two less divergent sequence domains were unraveled by the presence of motif combination 1 and 2, spanning the domains of C2–C3 and C4–C6. In contrast, the sequence domains close to the N-terminus (42 amino acids, motif 3 and motif 4) and ahead of C4 (11–15 amino acids, motif 5 and motif 6) appeared to be more divergent.

Utilizing the six identified consensus motifs in **Figure 2** we have quantified the sequence divergence for the locust OBP repertoire at a local motif scale (Figure S1). Apart from subfamilies I-A and II-A, the common motif 1 and motif 2, especially the latter, recapitulate sequence information present in many of the other locust OBPs analyzed (E-value below 10−<sup>5</sup> ) indicating particular phylogenetic conservation of these regions. Not surprisingly, the subfamily-specific motifs 3–6 failed to match OBP members (E-value above 10−<sup>5</sup> ) other than subfamily

I-A and II-A OBPs, despite a small number of OBPs in family I and family II (Figure S1). Taken together, the motif analysis unraveled the presence of both stabilized and diversified domains residing on the global sequences.

#### Selection Pressure and Orthology Evolution of Locust Subfamily I-A and II-A

The appearance of two distinct conserved subfamilies in the locust OBP phylogeny, coupled with the clustering pattern of different ortholog groups is presumably a consequence of particular selection regimes. To prove this notion, we have tried to quantify the strength of selection pressure acting on genes encoding the locust OBP repertoire. We analyzed three principal concepts which reflect the selection pressure, namely, the non-synonymous substitution rates (dN), the synonymous substitution rates (dS) and the ω rates (dN /dS) (**Figure 3**). We found a significantly reduced median dN level for both subfamily I-A (dN = 0.030, U = 60, p = 0.016, Mann-Whiteny U-test) and subfamily II-A (dN =0.028, U = 60, p = 0.016, Mann-Whiteny U-test), in comparison with that of other OBP members (dN = 0.085, **Figure 3A**). However, the median dS level appeared to be quite similar among subfamily I-A (dS = 0.12, p = 0.154,

U = 88.5, Mann-Whiteny U-test), subfamily II-A (dS = 0.16, U = 86, p = 0.131, Mann-Whiteny U-test) and the other OBP members (dS = 0.31, **Figure 3B**). For the ω rates, the values ranged from 0 to 0.7 for nearly 90% of locust OBPs (**Figure 3C**), which is indicative of purifying selection acting on locust OBP repertoire in general. For a few exceptions, ω rates larger than one were found which may indicate a positive selection. Notably, median ω rates for OBPs of subfamily I-A (ω = 0.18, U = 63, p = 0.021, Mann-Whiteny U-test) and subfamily II-A (ω = 0.22, U = 69, p = 0.036, Mann-Whiteny U-test) were significantly reduced in comparison with other OBP members in the phylogeny (ω = 0.35, **Figure 3C**).

Exposed to a similar selection regime, we wondered if orthologous OBPs in other species would undergo similar divergent events in relation to the two locust OBP subfamilies. To address the issue, we made a phylogenetic analysis of the two locust OBP subfamilies and the reference OBPs derived from 8 other insect species which gradually emerged in the course of insect evolution. The analysis revealed that locust subfamily II-A OBPs remained on an intact clade without intermingling with reference OBP genes on the newly constructed phylogenetic tree (Figure S3). A different result was obtained for the subfamily I-A: the original clustering relationship of ortholog groups in locust phylogeny was disrupted and altered with a complex reclustering pattern integrating reference OBPs. The orthologous relationship (Theißen, 2002) of OBPs between the two locust subfamilies and other species was also inferred. It is found that the number of locust subfamily I-A orthologous OBPs in the inspected insect species expanded considerably, and exhibited a many-to-many orthologous relationship with locust subfamily I-A (**Figure 3D**), with A. pisum as apparent exception likely due to a smaller OBP gene repertoire (Zhou et al., 2010). In contrast, the number of locust subfamily II-A orthologous OBPs in other species apparently decreased, and displayed a 1-to-many or 0-to-many orthologous relationship with locust subfamily II-A (**Figure 3D**). Moreover, it was found that locust subfamily II-A OBPs and their orthologous OBPs may share a common ancestor verified by the convergence of a mono phylogenetic clade with the bootstrap support above 70% at the basal divergent node (Figure S3). However, the common ancestral status for locust subfamily I-A OBPs and their orthologous OBPs appeared ambiguous because of the absence of evident bootstrap support (Figure S3). In sum, our results provide evidence that locust subfamily I-A and II-A OBPs are subject to mutually similar strengthened

analysis are given in the supplementary material. (D) Orthologs of locust subfamily I-A and II-A OBPs in seven other insect species. It is noted that the complete genome has been sequenced for the seven inspected species, namely, *Anopheles gambiae* (Agam), *Apis mellifera* (Amel), *Drosophila melanogaster* (Dmel), *Tribolium castaneum* (Tcas), *Acyrthosyphon pisum* (Apis), *Bombyx mori* (Bmor), and *Zootermopsis nevadensis* (Znev). Orthology assignment was obtained by using EggNOG 4.5.1 which performed a hierarchical orthologous annotation (Huerta-Cepas et al., 2016). The criteria *E*-value for assessing orthologous relationship of locust subfamily I-A is set to e−20, while e−<sup>10</sup> for subfamily II-A. Short bar denotes that there are no appropriate hints that could be assigned as orthologous OBPs. Nomenclature of OBPs for the seven inspected insect species conforms to Vieira and Rozas (2011) and Terrapon et al. (2014).

purifying selection, whereas distinct divergent events occur during evolution of their orthologous OBPs in other species.

#### Prediction of Tertiary Structures for OBPs in Subfamily I-A and II-A

The intriguing sequence and evolutionary characteristics underlying locust subfamily I-A and II-A OBPs inspired us to explore the possible concurrent variation of their tertiary structures. Therefore, we have simulated the tertiary structures for OBP members from both two subfamilies covering different ortholog groups and locust species. Parametric estimates toward the accuracy and reliability of the structure prediction was scrutinized, which permitted to investigate structural variation as an exploratory trial. To unravel structural variation between the two subfamilies, we superimposed the backbone structures of those simulated OBPs to LmigOBP1, the hitherto only established crystal structure for the locust OBP repertoire (Zheng et al., 2015). The averaged RMSD score obtained by imposing subfamily II-A OBPs to LmigOBP1 (2.8) doubled that of imposing subfamily I-A OBPs to LmigOBP1 (1.39 in average, Figure S4), indicating an enhanced structural similarity within one subfamily.

Multiple sequence alignment of subfamily I-A OBPs revealed a striking variation on the C-terminal domain (Figure S4). It is known that LmigOBP1 has a prolonged C-terminus with ∼17 amino acids to form a seventh α-helix (Zheng et al., 2015). In contrast, the C-terminus in OasiOBP3 and SgreOBP6 is shortened to a 7 amino acids motif and most likely constitute a coiled-coil strand instead of a seventh α-helix (**Figure 4**); a groove emerged on the collapsed surface due to the shortened C-terminus. The electrostatic potential pattern varies greatly at a global surface scale as well as on the local C-terminal surface scale (cyan dash line, **Figures 4A,C,E**). Another striking structural difference is the enlarged cavity of LmigOBP1 bordered by the prolonged C-terminus, whereas the cavity for the other two counterparts, representative of different ortholog groups shrinks to some extent (white dash line, **Figures 4B,D,F**). Unlike subfamily I-A, the multiple sequence alignment of subfamily II-A OBPs exhibited an aligned C-terminus but an unaligned Nterminus, namely, an extra extension of a 9–10 amino acids motif in the LmigOBP10 ortholog group (Figure S4). Correspondingly, this alteration was predicted to result in a coiled-coil structure on the N-terminal domain for LmigOBP10; at the same surface position, an opening structure was observed on its two counterparts, the OasiOBP11 and SgreOBP11 (**Figures 5A,C,E**). Apart from that, the surface electrostatic potential profile seems to vary slightly, both at the global surface scale and at the local N-terminal surface scale (cyan dash line, **Figures 5C,E**), regardless of the extra N-terminal coil present on LmigOBP10. However, the interior cavity could be enriched with negative potentials (LmigOBP10 and SgreOBP11,**Figures 5B,F)**, or with positive potentials (OasiOBP11, **Figure 5C**).

## Topographic Expression Patterns of SgreOBPs from Subfamily I-A and II-A

To approach this question, whether locust subfamily I-A and II-A OBPs may be expressed in different sensillum types and different cells, we set out to unravel the expression patterns of SgreOBPs from the two locust subfamilies in sensilla on the antenna, the major olfactory organ. By adopting RNA in

outlined with a white dash line. Electrostatic potential was calculated in the range of −6 to 6 kT/e and was presented as blue-red hue gradient. Blue, negative potential; red, positive potential; k, Boltzmann's constant; T, temperature; e, charge of an electron.

situ hybridization (ISH) on antennal sections using specific OBP probes, we acquired a strikingly sensilla-specific expression pattern for SgreOBPs in the two subfamilies. For SgreOBP1, SgreOBP5 and SgreOBP6, the members of subfamily I-A, we found alike expression in the cells of both sensilla basiconica and sensilla trichodea (**Figure 6**). In contrast, none of the subfamily I-A SgreOBPs was expressed in sensilla coeloconica or sensilla chaetica. Conversely, for members of subfamily II-A SgreOBPs, namely, SgreOBP10, OBP11, and OBP14, the expression was found to be restricted to the cells of sensilla coeloconica; there was no evidence for an expression in cells of any other sensillum type (**Figure 6**). The notion that a similar expression pattern is conserved for orthologous OBPs from other locust species is supported by the finding that LmigOBP1 is specifically expressed in sensilla basiconica and sensilla trichodea of L. migratoria (Jin et al., 2005), alike its ortholog in S. gregaria, the SgreOBP1.

Thus, an apparent sensilla-specific expression pattern for each locust OBP subfamily emerged. To extend and specify this aspect, the expression of OBP subtypes was compared with the expression of sensilla-specific receptor types. The odorant receptor co-receptor Orco and the ionotropic receptor (IR) type IR8a are ubiquitous co-receptors expressed in insect OSNs, either together with ligand-specific ORs or with IRs, and are considered as general markers for sensilla basiconica/sensilla trichodea and sensilla coeloconica, respectively (Yang et al., 2012; Guo et al., 2013). As a marker specific for distinct sensilla trichodea, the expression of the sensilla-specific receptor type OR3 in S. gregaria was monitored (Pregitzer et al., 2017). We designed riboprobes labeled by either Dig or Bio, which specifically targeted the distinct sensory neuron markers and SgreOBPs of the two subfamilies. Subsequently, two-color fluorescent in situ hybridization (FISH) experiments were performed to visualize the expressing cells (Figure S5). The results indicated that SgreOBPs of subfamily I-A are expressed in cells located in sensilla basiconica; these cells extended cytoplasmic processes and enclosed clusters of Orco expressing neurons. Similarly, SgreOBPs of subfamily I-A were found to be expressed in cells located in sensilla trichodea, as characterized by their close association with OR3 expressing OSNs. In the sensilla coeloconica, characterized by the IR8a-positive neurons, the neurons were found to be engulfed by cells which express OBPs of the subfamily II-A.

Although, our data demonstrated that SgreOBPs from different ortholog groups in each subfamily are expressed in the same sensillum type, it remained unclear to what extent they are expressed in the same set of sensilla and whether they are coexpressed in the same cells within a distinct sensillum. To resolve this question, we performed two-color FISH on sections through the antenna of S. gregaria using riboprobes targeting SgreOBPs from different ortholog groups. The results for SgreOBPs in

FIGURE 5 | The surface topologies and interior cavities of subfamily II-A OBPs. (A,C,E) Comparison of the backbone structures, surface topologies, and surface potentials of LmigOBP10, OasiOBP11, and SgreOBP11 which represent the three different ortholog groups in subfamily II-A. Left: the prolonged N-terminus in LmigOBP10 (see Figure S4B) was predicted to form a short coiled-coil shown on the backbone structure (highlighted in purple, A), but was absent from OasiOBP11 (C) and SgreOBP11 (E). Middle: the N-terminal domain of LmigOBP10 was plotted on the surface and sketched by a cyan dash line (A). The N-terminal domain of LmigOBP10 was labeled on the same surface position for OasiOBP11 (C) and SgreOBP11 (E). The visible opening structure is denoted by a black circle for OasiOBP11 (C) and SgreOBP11 (E). Right: a map of electrostatic potential on the molecular surface. Generally similar electrostatic potential pattern is observed among LmigOBP10 (A), OasiOBP11 (C) and SgreOBP11 (E). (B,D,F) A symmetric presentation of the interior cavity with the electrostatic potential. Electrostatic potential was calculated in the range of −6 to 6 kT/e and was presented as blue-red hue gradient. Blue, negative potential; red, positive potential; k, Boltzmann's constant; T, temperature; e, charge of an electron.

subfamily I-A indicate that SgreOBP1 was expressed in a cell population present in almost all basiconic and trichoid sensilla, whereas SgreOBP5 and SgreOBP6 were expressed only in a much smaller subset of cells than SgreOBP1 in the same sensillum (**Figure 7**). These differences became apparent in both horizontal sections giving a view onto superficial cellular layer (no cytoplasmic process expected, **Figure 7A**) as well as in longitudinal sections which allowed a view into deeper layers (cytoplasmic process expected, **Figure 7B**) of the antenna. Unlike SgreOBP1-positive cells which could be visualized both at the superficial and the deeper cellular layer, most of SgreOBP5 and SgreOBP6-positive cells appeared to be restricted to the superficial cellular layer close to the cuticle; slim cytoplasmic processes stretched to deeper cellular layers. Incidentally, there was evidence that SgreOBP5 and SgreOBP6 were expressed in the same set of cells of a sensillum (**Figures 7E,F**).

In contrast to the subfamily I-A, for subfamily II-A we did not find any evidence for an OBP subtype that was ubiquitously expressed in coeloconic sensilla (Figure S5). This result has led to the notion that particular OBP members of subfamily II-A may be specifically expressed in subsets of coeloconic sensilla. In fact, we frequently observed that expression of SgreOBP10 and SgreOBP14 were restricted to different cells in sensilla coeloconica (**Figures 8A,B**). For the subtypes SgreOBP11 and SgreOBP14 a co-expression in the same cells or expression in different cells were observed at a similar rate (**Figures 8C,D**). For the subtypes SgreOBP10 and SgreOBP11 it was frequently observed that they were co-expressed in the same cells (**Figure 8E**), indeed, more often than an expression in different cells (**Figure 8F**). Moreover, we verified the spatially separated expression of SgreOBPs from subfamily I-A and II-A (Figure S6), consistent with the results in **Figures 6**, **7**. Taken together, the results unravel a characteristic subfamily-dependent cellular expression pattern for different OBP subtypes.

#### DISCUSSION

The complex behavior of locust species, including the unique switch between a solitarious phase and a gregarious phase, is strongly based on a sophisticated chemical communication system (Pener and Yerushalmi, 1998; Hassanali et al., 2005; Wang and Kang, 2014). Great efforts have been made to unravel the chemical cues and underlying chemosensory mechanisms in mediating locust enigmatic behavior (Heifetz et al., 1996; Anton et al., 2007). Out of these efforts, a variety of olfactory genes, including gene families encoding odorant receptors and candidate pheromone receptors have recently been identified (Guo et al., 2011; Wang et al., 2015; Pregitzer et al., 2017). Since much less was known about their counterparts which deliver the olfactory signal molecules to the receptors, the OBPs, this study was concentrating on a systematic analysis of locust OBPs with respect to their molecular evolution as well as on an evaluation of predicted protein structures for OBP subtypes and their expression pattern in stinct sensillum types.

The in-depth analysis of locust OBP sequences uncovered the presence of both common and specific amino acid

FIGURE 6 | Sensilla-specific expression of subfamily I-A and II-A OBPs in the antenna of *S. gregaria*. Antisense riboprobes which specifically target the SgreOBPs were used to visualize the appropriate structures by means of chromogenic *in situ* hybridization (ISH). SgreOBP1, SgreOBP5, and SgreOBP6 are representing three different ortholog groups of subfamily I-A, whereas SgreOBP10, SgreOBP11, and SgreOBP14 are representing three different ortholog groups of subfamily II-A. Labeling obtained with probes for subfamily I-A SgreOBPs was restricted to sensilla basiconica (ba) and sensilla trichodea (tr), but was absent in sensilla coeloconica (co) and sensilla chaetica (ch). Labeling obtained with probes for subfamily II-A SgreOBPs was detected only in sensilla coeloconica (co), but was absent in the other three sensillum types. Black arrows indicate the visible OBP labeling while black circles denote the absence of OBP labeling.

motifs (**Figure 2**). The common motifs adequately recapitulate sequence information in most of the locust OBPs, while specific motifs selectively represent locust OBP subfamilies which may contribute to the clustering of sequences on the phylogenetic tree (**Figure 1**). The mixed common and specific motif profile is reminiscent of the findings that selection regimes may vary among different sequence domains (Policy and Conway, 2001; Sawyer et al., 2005). The subfamily specific motifs define sequence domains that apparently withstand diversifying selection constraints, presumably shaped by the sensilla environment, including their likely interplay-partner, the endogenous receptor types (Figure S5). In contrast, the common motifs define sequence domains that appear to share similar stabilizing selection constraints, presumably required for the maintenance of the common globular structures of the proteins (Pelosi et al., 2017), or for retaining the conserved ligand binding sites (Yu et al., 2009).

The four locust species tackled in this study differ significantly in their geographic distribution. While S. gregaria (the desert locust) occurs in Africa, the Middle East and Asia and L. migratoria (the migratory locust) in Africa and Asia, but also in Australia and New Zealand, the locusts O. asiaticus and C. kiangsu (the yellow-spined bamboo locust) appear to live locally in North China and South China. Nevertheless, a molecular and evolutionary stabilized status can be assigned to locust OBP subfamily I-A and II-A that appear to be subject to purifying selection pressure (**Figure 3C**), indicative for conserved chemosensory roles. In addition, the chemosensory adaptation to different habitats supposedly implies positive selection constraints (Cicconardi et al., 2017), and several of the locust OBPs appear to reflect such a selection regime (**Figure 3C**).

For the locust OBP subfamily I-A, the selective expression in two distinct sensillum types, sensilla basiconica, and sensilla trichodea, appears to be a characteristic hallmark (**Figure 6** and Figure S5). This feature is also found for OBPs from other species, which are orthologous of locust OBPs subfamily I-A (**Figure 3D**). For example, in Drosophila melanogaster, most of the subfamily I-A orthologous OBPs are associated with sensilla basiconica and sensilla trichodea, similar to their locust counterparts. It was found that DmelOBP83a and DmelOBP83b were associated with sensilla basiconica and sensilla trichodea, while DmelOBP69a and DmelOBP76a seemed to be restricted to sensilla trichodea

(Larter et al., 2016). However, for a few orthologous OBPs such as DmelOBP56d an extra sensillar expression has been reported (Larter et al., 2016). The concept of a sensilla-specific expression pattern for orthologous OBPs of locust subfamily I-A is also supported by the finding in the moth Manduca sexta, where two orthologous OBPs of locust subfamily I-A, named MsexABP2 and MsexABPx, are specifically expressed in sensilla basiconica (Nardi et al., 2003). Since the Orthopteran locust species emerged at a much earlier stage than the moth and fly species during the insect species divergence (Vieira and Rozas, 2011; Vogt et al., 2015), it is conceivable that a dual expression of subfamily I-A OBPs in both sensilla basiconica and sensilla trichodea may represent an ancestral status. In insect species like moths and flies, which emerged later in evolution, some OBP subtypes may have evolved towards a more specific function and expression in either sensilla basiconica or sensilla trichodea (Maida et al., 2005; Larter et al., 2016).

Our analysis suggests that the locust OBPs of subfamily II-A and their orthologous OBPs in other species have originated from a common ancestor (Figure S3), and may share a sensilla coeloconica specific expression pattern (**Figure 6**, Figure S5). In Drosophila melanogaster, DmelOBP84a, the only orthologous OBP of locust subfamily II-A is actually among the few OBPs that have been reported to be specifically expressed in sensilla coeloconica (Larter et al., 2016). Interestingly, the gene encoding OBP84a is retained in most, if not all, Drosophila species genomes (Cicconardi et al., 2017). Moreover, the OBP84a ortholog group in Drosophila species withstands apparent purifying selection pressure (Vieira et al., 2007) and converges onto a segregated phylogenetic clade (Cicconardi et al., 2017), which is very similar to the locust OBP subfamily II-A. These molecular and phylogenetic commonalities may point to some similarities with regard to their functional roles. In this regard, it is interesting to note that single sensillum recordings from sensilla coeloconica

of locust, flies and moths have revealed a response spectrum confined to certain ecologically important odorants, including organic acid, amines and plant derived odorants (Pophof, 1997; Ochieng and Hansson, 1999; Yao, 2005). Thus, it will be of particular interest to unravel a potential role of locust subfamily II-A OBPs and their orthologs in other species for the detection of cognate odorants in sensilla coeloconica. While concentrating on OBPs of subfamily II-A, we are aware that sensilla coeloconica may also comprise OBPs of other phylogenetic clades.

Unlike DmelOBP84a, which is broadly expressed in almost all sensilla coeloconica (Larter et al., 2016), the OBPs of the locust subfamily II-A are expressed in sensilla coeloconica in a combinatorial mode (**Figure 8**). This is in line with the previous finding that different subsets of sensilla coeloconica in S. gregaria showed individual response spectra to a repertoire of odorants (Ochieng and Hansson, 1999), suggesting a sensillaspecific response spectrum and sensilla-specific repertoire of odorant sensing proteins. Thus, it is conceivable that a distinct combination of OBPs in a sensillum coeloconicum (**Figure 8**) may correlate with particular endogenous IR types.

Although amino acid sequences of OBPs can be highly divergent, the folding of proteins forming a hydrophobic pocket is well conserved across insect species; in fact to date the structures of more than 20 OBPs have been solved by Xray crystallography and/or nuclear magnetic resonance (NMR) spectroscopy (Pelosi et al., 2017). The results of these studies revealed that the C-terminal domain, especially the length of the C-terminus has important implications on the mechanism of ligand-binding (Tegoni et al., 2004). Long C-terminus apparently enter the binding pocket and determine the shape of the cavity (Sandler et al., 2000), medium-length C-terminus act as a lid covering the entrance to the binding pocket (Lartigue et al., 2004). In view of these findings, simulation of the putative tertiary structures of locust OBPs revealed some interesting features. The three ortholog groups of subfamily I-A significantly differ in their C-terminal domain. LmigOBP1 and its orthologs have a long (17 aa) C-terminus, long enough to form an extra α-helix and thus affecting the shape of the cavity (**Figure 4**, Figure S4); other two ortholog groups have both a medium size C-terminus (7 aa), however, significantly different in the amino acid sequence. These observations may suggest significant differences in the mechanisms of OBP/ligand interaction among the three ortholog groups in subfamily I-A.

The results of this study indicate that in a considerable number of sensilla at least two OBP subtypes are co-expressed (**Figures 7**, **8**). This is of particular interest, since heteroand homo-dimerization of OBPs have been reported in vitro (Andronopoulou et al., 2006), which is accompanied by a set of conformational changes (Wogulis et al., 2006; Mao et al., 2010). Although the underlying mechanisms are still elusive, there is evidence that electrostatic interaction at short range forming the salt bridges may contribute to specific protein-protein interaction (Sheinerman et al., 2000; Kumar and Nussinov, 2002). In locusts, the patch of charged residues buried on the OBP-interface (**Figures 4**, **5**) is likely to provide hot spots for protein-protein interactions. In addition, changes of the OBP tertiary structure has been demonstrated as a consequence of pH changes in the environment (Zubkov et al., 2005; Pesenti et al., 2008). This notion may also fit for locust OBPs since an intermingled distribution of both negative and positive charged residues was observed by elucidating a map of electrostatic potentials (**Figures 4**, **5**). The presence of multiple OBP subtypes and their possible interaction may have functional implications for the binding capacity of the olfactory system. In fact, recent binding assays have shown that in the presence of two OBPs the binding affinity to cognate ligands altered considerably compared to the binding characteristics of a single OBP type (Qiao et al., 2011; Sun et al., 2016). This notion may be particular relevant with respect to sensilla basiconica of locusts, which house up to 50 sensory neurons responding to a variety of different odorants (Ochieng et al., 1998; Ochieng and Hansson, 1999), and the fact that the number of OBP genes is much smaller than the size of the OR gene family in locusts, encoding more than 140 ORs in L. migratoria (Wang et al., 2014) and at least 120 ORs in S. gregaria (Pregitzer et al., 2017). The selective sensilla expression pattern implies that a small number of OBP

#### REFERENCES


subtypes are present in the sensillum lymph (**Figure 7**, Figure S6). Assuming that each OBP subtype has distinct ligand specificity, the mixture may provide a much broader binding spectrum. A possible combinatorial mode of OBP participation in locust olfaction is an interesting aspect for future studies.

#### AUTHOR CONTRIBUTIONS

HB, JK, XJ, and PP conceived the study. XJ conducted the experiments. HB, JK, XJ, and PP interpreted the results. XJ and PP drafted the preliminary manuscript. HB and JK refined and approved the final manuscript.

#### FUNDING

The author XJ is supported by a grant from Chinese Scholarship Council (CSC) with the award number 201406350032.

#### ACKNOWLEDGMENTS

We are grateful to Dr. Ewald Grosse-Wilde (Max Planck Institute for Chemical Ecology) for the help with generating the antennal transcriptome sequence data base and initial bioinformatics analysis. We thank Heidrun Froß for her excellent technical assistance, Rosolino Bumbalo for his suggestions for the data analysis and Christian Heidel for suggestions and comments on the structure modeling.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00734/full#supplementary-material

two folded solution conformations of the pheromone-binding protein from Bombyx mori. Protein Sci. 9, 1038–1041. doi: 10.1110/ps.9.5.1038


Blattella germanica (Blattaria: Blattidae). Comp. Biochem. Physiol. D Genomics Proteomics 18, 30–43. doi: 10.1016/j.cbd.2016.03.002


evidence for a common ligand release mechanism. Biochem. Biophys. Res. Commun. 339, 157–164. doi: 10.1016/j.bbrc.2005.10.191


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Jiang, Krieger, Breer and Pregitzer. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identification and Characterization of Odorant Binding Proteins in the Forelegs of *Adelphocoris lineolatus* (Goeze)

Liang Sun1, 2, 3 \* † , Qian Wang2, 4†, Qi Wang<sup>2</sup> , Kun Dong<sup>2</sup> , Yong Xiao<sup>2</sup> and Yong-Jun Zhang<sup>2</sup> \*

*<sup>1</sup> Key Laboratory of Tea Quality and Safety Control, Ministry of Agriculture, Tea Research Institute, Chinese Academy of Agricultural Sciences, Hangzhou, China, <sup>2</sup> State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China, <sup>3</sup> Key Laboratory of Integrated Pest Management on Crops in East China, Ministry of Agriculture, Key Laboratory of Integrated Management of Crop Diseases and Pests, Ministry of Education, College of Plant Protection, Nanjing Agricultural University, Nanjing, China, <sup>4</sup> College of Horticulture and Plant Protection, Yangzhou University, Yangzhou, China*

#### *Edited by:*

*Peng He, Guizhou University, China*

#### *Reviewed by:*

*Su Liu, Anhui Agricultural University, China Ya-Nan Zhang, Huaibei Normal University, China Joe Hull, Agricultural Research Service (USDA), United States*

#### *\*Correspondence:*

*Liang Sun liangsun@tricaas.com Yong-Jun Zhang yjzhang@ippcaas.cn*

*† These authors have contributed equally to this work and should be considered co-first authors.*

#### *Specialty section:*

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

*Received: 30 July 2017 Accepted: 11 September 2017 Published: 26 September 2017*

#### *Citation:*

*Sun L, Wang Q, Wang Q, Dong K, Xiao Y and Zhang Y-J (2017) Identification and Characterization of Odorant Binding Proteins in the Forelegs of Adelphocoris lineolatus (Goeze). Front. Physiol. 8:735. doi: 10.3389/fphys.2017.00735* The chemosensory system is essential for insects to detect exogenous compounds, and odorant binding proteins (OBPs) play crucial roles in odorant binding and transduction. In the alfalfa plant bug *Adelphocoris lineolatus*, an important pest of multiple crops, our understanding of the physiological roles of antenna-biased OBPs has increased dramatically, whereas OBPs related to gustation have remained mostly unexplored. In this study, we employed RNA sequencing and RACE PCR methods to identify putative OBPs from the adult forelegs of both sexes. Eight candidate OBPs were identified, and three OBPs (AlinOBP15, 16, and 17) were novel. Full-length sequence alignment and phylogenetic analyses suggested that these three candidate OBPs had characteristics typical of the insect OBP family. AlinOBP16 and 17 displayed six highly conserved cysteines, placing them in the classic OBP subfamily, whereas AlinOBP15 resembled AlinOBP14 and clustered with the Plus-C clade. Quantitative real-time PCR (qRT-PCR) revealed distinct and significant tissue- and sex-biased expression patterns. *AlinOBP15* was highly expressed in female heads, and *AlinOBP16* and *17* were strongly expressed in female antennae. In particular, *AlinOBP11*, the most abundant OBP gene in our foreleg transcriptome dataset, was predominately expressed in adult legs. Furthermore, four types of sensilla hairs were observed on the forelegs of adult *A. lineolatus*, including sensilla trichodea, setae, and two types of sensilla chaetica (Sch1 and Sch2). Anti-AlinOBP11 antiserum strongly labeled the outer sensillum lymph of Sch2, implying that it has important gustatory functions in *A. lineolatus*. Our current findings provide evidence that OBPs can be functionally expressed in the tarsal gustatory sensilla of hemipteran mirid species, broadening our understanding of OBP chemosensory function in insects and facilitating the discovery of new functional targets for the regulation of insect host-searching behaviors.

Keywords: *Adelphocoris lineolatus*, odorant binding protein, expression profiles, phylogenetic analyses, cellular immunolocalization, gustation

## INTRODUCTION

Host plant location is essential for phytophagous species survival and drives the rapid evolution of insect-plant interactions. Insect species encounter a wide range of environments that eventually result in different life styles and host plant adaptions (Peccoud et al., 2010). Insect foraging behaviors primarily rely on chemical sensing (Visser, 1986). During the initial step of insect host orientation, plant volatiles and the insect olfactory system play crucial roles (Takken, 1991; Li and Liberles, 2015). However, after landing on a plant, another important chemosensory repertoire, namely, gustation on tarsi and labella plays a more important role. This system enables insects to locate favorable oviposition sites, avoid plant toxins and determine whether a plant is suitable for habitation (Romani et al., 2005).

Specialized insect antennal chemosensilla, such as sensilla basiconica, house general olfactory sensory neurons (OSNs) and are responsible for recognizing host plant volatiles (Park et al., 2013; Yuvaraj et al., 2013). By contrast, gustatory chemosensilla, such as contact sensilla chaetica on tarsi, labella and wing margins, possess gustatory sensory neurons (GSNs), and express gustatory receptors (GRs), enabling insect perception of taste substances on host plant surfaces (Ave et al., 1978; Anderson and Hallberg, 1990; Isidoro et al., 2001; Leopold et al., 2003; Sun et al., 2014a). In general, chemical cues for insect host plant location, either the volatile odorants or non-volatile tastants, have poor hydrophilic characteristics, and it is often difficult for them to pass through the hydrophilic chemosensillum lymph barrier to activate odorant receptors (ORs) or GRs for chemical signal transduction. Numerous reports indicate that carrier proteins, particularly odorant binding proteins (OBPs), are highly expressed in the sensillum lymph and function as adaptor molecules between chemical cues and their receptors (Leal, 2013; Pelosi et al., 2014, 2017).

Insect OBPs are small, acidic, water-soluble proteins and were first identified in the Lepidopteran moth antennal sensillum (Vogt and Riddiford, 1981). Their homologous genes have been explored in a wide range of insect species, including moths (Gong et al., 2009; Zhang T. et al., 2011; Glaser et al., 2013; Zhang et al., 2013; Walker et al., 2016; Sun et al., 2017a), flies (Graham and Davies, 2002; Hekmat-Scafe et al., 2002; Meunier et al., 2003; Leitch et al., 2015), mosquitoes (Xu et al., 2003; Zhou et al., 2008; Pelletier and Leal, 2011; He et al., 2016), aphids (Zhou et al., 2010; Gu et al., 2013), planthopper (He and He, 2014), and bugs (Gu et al., 2011a; Ji et al., 2013; Hull et al., 2014; Yuan et al., 2015; Paula et al., 2016). Six highly conserved cysteines that form three disulfide bridges help insect OBPs fold into a large pocket for molecular uptake (Leal et al., 1999; Pelosi et al., 2013), and it is clear that OBPs in the olfactory repertoire contribute to odorant recognition (Leal, 2013; Brito et al., 2016). For instance, one subfamily of OBPs known as pheromone binding proteins (PBPs) are specifically synthesized and expressed by non-neuronal auxiliary cells (trichogen and tormogen cells) in pheromone-sensitive long trichoid sensilla. These proteins show strong binding affinities to insect sex pheromones and enhance the sensitivity and specificity of olfactory receptors to such pheromones (Wang et al., 2004; Große-Wilde et al., 2006; Sun M. et al., 2013; Chang et al., 2015; Liu et al., 2015). Suppression of PBP transcript levels can seriously disrupt the responses of male insects to female-produced sex pheromones (Dong et al., 2017). The other subfamilies of OBPs, such as general odorant binding proteins (GOBPs), have been shown to be necessary for both general odorant and insect pheromone perception (He et al., 2010; Yin et al., 2012).

The physiological functions of insect OBPs might be more complicated. In addition to the odorant detection in the olfactory system, they were also reportedly expressed in gustatory organs, including taste sensilla in labellum, tarsi, and wings and were supposed to be involved in recognition of taste compounds (Ozaki et al., 1995; Galindo and Smith, 2001; Shanbhag et al., 2001; Hull et al., 2014; Sparks et al., 2014; He et al., 2017). The study of electrophysiological responses of contactchemoreceptor sensilla on the labellum of the blowfly, Phormia regina suggested that a unique type of OBP known as CRLBP could functions as a carrier for monoterpenes (Ozaki et al., 2003). Direct evidences supporting this hypothesis were reported in Drosophila species. For instance, two OBP genes, Obp57d and Obp57e, were co-expressed in the leg taste sensilla of Drosophila species and contributed to the sensation of octanoic acid and the evolution of taste perception and host-plant preference (Matsuo et al., 2007; Yasukawa et al., 2010). Suppression of Drosophila melanogaster feeding behavior on sweet substances by bitter compounds required OBP49a (Jeong et al., 2013). Subsequent RNAi interference assay demonstrated that OBP functions in a combinatorial and sexually dimorphic manner in the gustatory system of D. melanogaster (Swarup et al., 2014).

Transgenic Bacillus thuringiensis (Bt) cotton is commonly cultivated in China, and outbreaks of the alfalfa plant bug, Adelphocoris lineolatus (Goeze), and other mirid species are frequent in cotton fields (Lu et al., 2010). Furthermore, substantial evidence indicates that A. lineolatus can destroy many other important crops, including alfalfa (Medicago sativa L.), green bean (Phaseolus vulgaris), and tea plants (Camellia sinensis; Lu and Wu, 2008). Due to the polyphagous host-feeding behavior and strong migration among different host plants (Wang et al., 2017), it is very difficult to prevent and control rapidly growing populations of mirid bugs using traditional pest management strategies. Studies of the physiological and molecular basis of insect host plant selection and adaptability could yield effective complimentary measures, particularly for species that rely heavily on chemosensing for preferential host plant searching (Koczor et al., 2012).

The molecular mechanisms of A. lineolatus olfaction, in particular OBP identification and their binding repertoires to plant volatiles have been extensively studied (Gu et al., 2011b; Sun L. et al., 2013; Sun et al., 2014b). Interestingly, we found that antennae-enriched or mouthpart-biased OBPs potentially bind to non-volatile plant secondary metabolites (Sun et al., 2016, 2017b). Mirid species reportedly contact the host plant surface via foreleg tarsi, and therefore, it is reasonable to hypothesize that OBPs expressed on tarsi help mirid bugs to respond to contact substances on host plant surfaces. To test this hypothesis, we first identified putative OBP genes from adult forelegs using transcriptome analysis; we then assessed tissue- and sex-biased expression patterns, with a particular focus on immunolocalization in gustatory tarsi sensilla. Screening for highly expressed OBPs in gustatory organs strongly indicates the potential for physiology functions and provides a better understanding of the molecular basis of A. lineolatus gustation.

## MATERIALS AND METHODS

#### Insect Rearing and Tissue Collection

Adult A. lineolatus were collected from alfalfa fields at the Langfang Experimental Station of the Chinese Academy of Agricultural Sciences, Hebei Province, China. The laboratory colony was established in plastic containers (20 × 13 × 8 cm), which were maintained at 29 ± 1 ◦C, with 60 ± 5% relative humidity, under a 14 h light: 10 h dark cycle. Adults and newly emerged nymphs were reared on green beans and 10% honey.

For transcriptome sequencing, 300 forelegs were collected from eclosion-stage bugs of both sexes (6-d old). Various tissues from A. lineolatus adults of both sexes, including antennae, heads without antennae, thoraxes, abdomens, legs, and wings were collected for quantitative Real-Time PCR (qRT-PCR). Samples for each tissue were collected from three biological pools, and all specimens were immediately stored at −80◦C for future use.

#### cDNA Library Construction, Transcriptome Assembly, and Functional Annotation

Total RNA was extracted from male and female antennae using a Trizol reagent (Invitrogen, Carlsbad, CA, USA). The quantity of RNA samples was checked by using 1.1% agarose gel electrophoresis and a NanoDropTM spectrophotometer (Thermo Scientific, Wilmington, DE, USA). The messenger RNA were further isolated from the total RNA using a PolyA (+)-tract mRNA isolation System III (Promega, Madison, WI, USA), and ∼2.5µg messenger RNA was further purified from 250µg total RNA. The mRNAs were then sheared into ∼800 nucleotides via RNA Fragmentation Solution (Autolab, Beijing, China) at 70◦C for 30 s, then cleaned and condensed using an RNeasy MinElute Cleanup Kit (Qiagen, Valencia, CA, USA).

The cDNA library was generated from aforesaid obtained mRNA using the SMART cDNA Amplification Kit (Clontech, Mountain View, CA, USA) and the Ion XpressTM Plus gDNA Fragment Library Kit (Life Technologies, Thermo Scientific, Wilmington, DE, USA), following the manufacturer's protocols. The cDNAs (300–400 bp) were purified using the Min Elute Gel Recovery Kit (Qiagen, Valencia, CA, USA) and sequenced using the Proton I chip of Ion ProtonTM System (Life Technology, Thermo Scientific, Wilmington, DE, USA). Using the TagDust, LUCY, and SeqClean software programs with default parameters, short or low-quality sequences and adaptor sequences were removed (Li and Chou, 2004; Chen et al., 2007; Lassmann et al., 2009). Male and female reads were assembled separately, and all reads were assembled using the MIRA3.0 (Chevreux et al., 2004) and CAP3 software programs (Huang and Madan, 1999) with default parameters. Two steps were performed to assemble the clean reads. First, the sequence assembler Mimicking Intelligent Read Assembly MIRA3 was used with the assembly settings of a minimum sequence overlap of 30 bp and a minimum percentage overlap identity of 80%. Then, Contig Assembly Program CAP3 was used with the assembly parameters of an overlap length cutoff >30 and an overlap percent identity cutoff >90%. The resulting contigs and singletons that were more than 100 bases were retained as unigenes. BLASTX and BLASTN programs were used to perform a homology search against the GenBank nonredundant protein (nr) and nucleotide sequence (nt) databases on NCBI with an E-value cut-off of 1.0E-5. Gene Ontology terms were obtained from the best hits obtained from BLASTX against the nr database using the Blast2GO program (Conesa et al., 2005).

## Identification and Phylogenetic Analyses of Putative OBPs

In addition to keyword searching, a FASTA file of non-redundant contigs was created from a local nucleotide database file using the BioEdit Sequence Alignment Editor program version 7.1.3.0, and the local TBLASTN program was performed using available bug OBPs (**Table S1**) as the queries (Gu et al., 2011a; Hull et al., 2014; Yuan et al., 2015). Candidate unigenes encoding putative OBPs were manually checked using the BLASTX online program at the NCBI and confirmed according to the conserved cysteine pattern feature C1-X25−30-C2-X3-C3-X36−42-C4-X8−14- C5-X8-C<sup>6</sup> (Xu et al., 2009; Zhou et al., 2010).

The 5′ and 3′ regions of OBP genes were amplified using SMARTerTM RACE cDNA amplification kit (Clontech, Mountain View, CA, USA) with gene-specific primers (GSP) (**Table S2**). Touchdown PCR was performed as follows: 95◦C for 2 min followed by 5 cycles at 94◦C for 30 s, 72◦C for 2 min, then 5 cycles at 94◦C for 30 s, 70◦C for 30 s, and 72◦C for 90 s, then 30 cycles at 94◦C for 30 s, 68◦C for 30 s, and 72◦C for 90 s, and a final 10 min incubation at 72◦C. The RACE PCR products were subcloned into the pEASY-T3 vector (Transgene, Beijing, China) and sequenced. The full-length OBP genes were confirmed with LA Taq DNA polymerase (Takara, Dalian, China) by PCR using gene-specific primers (**Table S2**).

The full-length OBP amino acid sequence alignments were performed using the program ClustalX 2.1 with default gap penalty parameters of gap opening 10 and extension 0.2 (Thompson et al., 1997). They were then edited using the GeneDoc 2.7.0 software. The neighbor-joining tree was constructed using the program MEGA 6.0 with a p-distance model and pairwise deletion of gaps (Tamura et al., 2013). The bootstrap support for the tree branches was assessed by re-sampling amino acid positions 1,000 times.

#### qRT-PCR

Total RNA for each sample was isolated using Trizol reagent (Invitrogen, Carlsbad, CA, USA). The integrity of the total RNA was examined using 1.2% agarose electrophoresis, and the purity was assessed using a NanoDropTM instrument (Wilmington, DE, USA). First-strand cDNA was synthesized from 2µg RNA using a FastQuant RT kit with gDNA Eraser (TianGen, Beijing, China), according to the manufacturer's instructions.

For the subsequent qRT-PCR reaction, the cDNA was diluted to a concentration of 200 ng/µL. experiments were performed using an ABI 7500 Real-Time PCR System (Applied Biosystems, Carlsbad, CA). Primers were designed using the Beacon Designer Sun et al. Characterizing of OBPs from the Forelegs of *A. lineolatus*

7.90 program (PREMIER Biosoft International) and are shown in **Table S2**. Each reaction was performed in a total reaction volume of 25µL, consisting of 12.5µL of SuperReal PreMix Plus (TianGen, Beijing, China), 0.75µL each primer (10 mM), 0.5µL Rox Reference Dye, 1 µL sample cDNA, and 9.5µL sterilized H2O. The reaction cycling parameters were as follows: 95◦C for 15 min, followed by 40 cycles of 95◦C for 10 s and 60◦C for 32 s. The PCR products were heated to 95◦C for 15 s, cooled to 60◦C for 1 min, heated to 95◦C for 30 s, and cooled to 60◦C for 15 s to measure the dissociation curves. A. lineolatus ß-actin was identified as a stable reference gene between different tissue samples and was used to normalize target gene expression and correct for sample-to-sample variation (Gu et al., 2011a). For data reproducibility, the qRT-PCR reactions for each sample were performed using three technical replicates and three biological replicates. The amplification efficiencies of the target and reference gene were assessed using gradient dilution templates to examine the variation of 1C<sup>T</sup> (CT, Target gene − CT, Reference gene) with template dilution. The absolute values of the slopes of all lines from template dilution plots (log cDNA dilution vs. 1CT) were close to zero, indicating that the amplification efficiency between target genes and the reference gene was similar and the comparative 2−11CT method was used to calculate relative levels between tissues (Livak and Schmittgen, 2001). Comparative analyses of target gene expression among the various tissues were performed using one-way nested analysis of variance (ANOVA), followed by Tukey's honestly significance difference (HSD) tests, using the SPSS Statistics 18.0 software program (SPSS Inc., Chicago, IL, USA).

#### Scanning and Transmission Electron Microscopy and Immunocytochemical Labeling

To confirm that OBPs play a role in gustatory function in the tarsi, the structures and distributions of tarsi sensilla were observed using scanning and transmission electron microscopy (SEM, TEM), and immunolocalization of AlinOBP11 on different types of tarsi sensilla were performed.

Three female and male forelegs were removed from adult A. lineolatus, fixed in 70% ethanol for 3 h, cleaned in an ultrasonic bath (250 W) for 10 s and finally subjected to gradient elution in an ethanol series (70, 80, 90, 95, and 100%). The samples were dried in an oven thermostat at 25◦C for 10 h. After coating with gold-palladium and mounting on holders, the samples were observed using a Hitachi S570 SEM (Hitachi Ltd., Tokyo, Japan).

For TEM observation and immunocytochemical labeling, newly cut forelegs were fixed separately in a mixture of 4% paraformaldehyde and 2% glutaraldehyde in 0.1 M PBS (pH 7.4) at room temperature for at least 24 h, dehydrated in an ethanol series (30, 50, 70, 80, 90, 95, and 100%), and embedded in LR white resin (Taab, Aldermaston, Berks, UK) for polymerization at 60◦C. Ultrathin sections (60–80 nm) were made using the diamond knife on a Reichert Ultracut ultramicrotome (Reichert Company, Vienna, Austria). Double-staining was performed with uranyl acetate and lead citrate, and sections were observed using a Hitachi H-7500 TEM (Hitachi Ltd., Tokyo, Japan).

The localization of AlinOBP11 on different tarsi sensilla was determined using an immunocytochemical labeling assay. A polyclonal antiserum against AlinOBP11 was produced, and its specificity was confirmed by western blotting analysis in our previous study (Sun et al., 2016). The immunocytochemical labeling assay was performed according to previously reported methods (Sun et al., 2014b). Briefly, grids that contained the ultrathin bug tarsi sections were floated in 25-µL droplets of PBSG (PBS containing 50 mM glycine) followed by droplets of PBGT (PBS containing 0.2% gelatin, 1% bovine serum albumin, and 0.02% Tween-20), and then incubated with purified rabbit anti-AlinOBP11 antiserum (dilution 1:2,000) at 4◦C overnight. After washing six times with PBGT, the sections were incubated with secondary antibody (anti-rabbit IgG) coupled with 10 nm colloidal gold granules (Sigma, St. Louis, MO, USA) at a 1:20 dilution at room temperature for 90 min. Sections were subjected to optional silver intensification for 15 min, stained with 2% uranyl acetate to increase the contrast, and observed using a HITACHI H7500 TEM (Hitachi Ltd., Tokyo, Japan). Immunocytochemical assays were conducted on three biological replicates. Serum supernatant from an uninfected healthy rabbit at the same dilution rate was used as the negative control.

#### RESULTS

#### RNA Sequencing and *De novo* Assembly

We performed RNA sequencing on male and female A. lineolatus adult forelegs to identify gustatory organ-biased OBPs. We obtained 7,348,393 clean reads with an average length of 127 bp for males, and 6,728,599 clean reads with an average length of 119 bp for females. High-quality fragments were assembled into 48,127 (mean length 477 bp) and 50,149 unigenes (mean length 477 bp), respectively. Subsequently, both male and female clean reads were assembled together to generate 50,801 unigenes with an average length of 469 bp **(Table 1** and **Figure 1A)**.

#### Homology Searching and Functional Annotation

The BLASTX program was used to annotate the acquired unigenes against an NCBI nr protein database with a cutoff E-value of 10−<sup>5</sup> . The results showed that 12,425 (24%) unigenes had BLASTX hits. The best match percentage was 14.31% for Tribolium castaneum sequences, followed by 13.66% for Acyrthosiphon pisum, 9.27% for Pediculus



humanus corporis, 7.80% for Nasonia vitripennis, and 6.31% for Camponotus floridanus (**Figure 1B**). Based on the Gene Ontology (GO) annotations, 5,682 unigenes could be assigned to the following three functional categories: molecular function, cellular components and biological processes. Individual unigenes could be assigned to more than one biological process, and no significant differences were observed between sexes for each GO category. For the molecular function GO category, catalytic activity (2,482 male unigenes and 2,472 female unigenes) and binding (2,230 male unigenes and 2,249 female unigenes) were the two most abundant subcategories. For the cellular components and biological processes categories, cell (3,159 male unigenes and 3,167 female unigenes) and cellular processes (2,852 male unigenes and 2,860 female unigenes) were the most common subcategories, respectively (**Figure 1C**).

#### Identification and Full-Length Sequence Alignments of Putative OBPs

Eight candidate OBPs were identified from the A. lineolatus adult foreleg cDNA library by homology analysis. Five transcriptencoded OBPs, AlinOBP1, 2, 7, 11, and 14, were previously reported in A. lineolatus (Gu et al., 2011a). Three OBPs, which we named AlinOBP15–17, were novel, and their sequences were deposited in GenBank (accession numbers KT596720– KT596722; **Table 2)**.

Among the 8 identified OBPs, only one transcript, AlinOBP11, had a full-length sequence of 453 bp. As the full-length open reading frames (ORFs) of AlinOBP1, 2, and 7 were previously reported, here we report the cloned full-length sequences for the four other identified OBPs (AlinOBP14, 15, 16, and 17) based on a 5′ and 3′ RACE-PCR strategy. Full-length sequence verification showed that AlinOBP14–17 were encoded on ORFs of 615, 666, 444, and 432 bp, respectively. As shown in **Figure 2**, these four newly cloned OBPs can be divided into two subfamilies. AlinOBP16 and 17 have the typical six cysteine signature (C1-X25−30-C2-X3-C3-X36−42-C4-X8−14-C5-X8-C6) and belong to the classic OBP subfamily. In contrast, AlinOBP14 and 15 possess three extra conserved cysteines (C4a, C6a, and C6b), as well as a conserved proline (P) immediately after the sixth cysteine (C1-X20−41-C2-X3-C3-X41−46-C4-X19−29-C4a-X9- C5-X8-C6-P-X9−10-C6a-X9−10), which are typical characteristics of the insect Plus-C OBP subgroup.

#### Phylogenetic Analyses of OBPs

To deduce the evolutionary relationships and potential functional differences between the OBPs, 95 Hemipteran OBP sequences (**Table S3**) from five bug species were selected to construct phylogenetic tree **(Figure 3)**. The phylogenetic analyses revealed that OBP within species were significantly divergent, with the amino acid identity in A. lineolatus only reaching 23.36%. In contrast, homologous OBPs across species shared very high similarities and clustered into the same clade with high bootstrap support, suggesting that they originated from the same ancestors and have conserved


functions. Neither "minus-C" nor "dimer" OBP subfamily members were found. Only two types of motifs, referred to as "Plus-C" and "classic" OBP subgroups, were observed across the mirid bug species, and the 8 identified OBPs from A. lineolatus tarsi fell with these two categories. AlinOBP7, AlinOBP14, and AlinOBP15 clustered into the insect Plus-C OBP subfamily, and AlinOBP16, 17, 1, 2, and 11 and OBPs in the other bug species were assigned to the classic OBP clade.

#### Tissue- and Sex-Biased Expression Patterns of Candidate OBPs

The tissue- and sex-biased expression profiles of the three novel OBP genes, AlinOBP15, 16, and 17, were determined by qRT-PCR. AlinOBP11 was selected as a target gene to determine PCR reaction rate and reproducibility, because the RPKM value analysis revealed that AlinOBP11 was the most abundant transcript in both the male and female foreleg transcriptomes (**Figure S1**). This OBP gene was reported to be highly expressed in A. lineolatus gustatory organs legs and mouthparts (Gu et al., 2011a; Sun et al., 2016). As expected, the results of our qRT-PCR showed that AlinOBP11 was strongly expressed in the adult legs of A. lineolatus, and no significant difference in expression levels was found between the sexes (**Figure 4**). The three novel OBP genes AlinOBP15, 16, and 17 shared a similar female-biased expression patterns. In particular, AlinOBP16 and AlinOBP17 were highly expressed in female antennae, whereas AlinOBP15 was strongly detected in female heads **(Figure 4)**.

#### Types of Sensilla on *A. lineolatus* Forelegs and Immunolabeling of AlinOBP11

Three tarsi were found on the forelegs and four different types of sensilla hairs were present on the tarsi and tibia of adult A. lineolatus forelegs, including sensilla trichodea (Str), setae and two types of sensilla chaetica (Sch1 and Sch2; **Figures 5A–F)**. Sensilla trichodea (Str) were primarily distributed on the 3rd tarsus, whereas setae were present only on the tibia. Sch1 could be found in both foreleg tarsi and tibia, and Sch2 was absent on tibia but present on all the three tarsi. Furthermore, TEM revealed that these four sensilla had distinct ultrastructures. Str had well-pore structures and one sensillum lumen. By contrast, the seta had a thick wall and no pores on the sensilla wall. SCh1 and Sch2 showed significantly different ultrastructures. Sch1 have one sensillum lumen, whereas Sch2 have two chambers and clear sensilla dendrites were found on

the inner sensillum lumen rather than the outer sensillum cavity (**Figures 5G–J**).

We further investigated the cellular immunolocalization of AlinOBP11 because, compared with the other antennae- and head- enriched OBPs, this protein was most strongly expressed in the gustatory leg organs. Results of the immunolabeling assay showed that the anti-AlinOBP11 antibody predominately labeled the outer sensillum of Sch2, and no obvious staining was observed in either the inner sensillum lumen or the other sensilla types **(Figures 5K–N)**.

#### DISCUSSION

In this study, we identified putative OBPs from the foreleg, an important taste organ in hemipteran insect species, and then we characterized different types of gustatory sensilla present on foreleg tarsi, where one bug OBP was predominately localized. These results provide direct morphological and molecular evidence that the foreleg tarsi of A. lineolatus harbor contact sensilla and that AlinOBP11, a putative carrier of bitter compounds, such as catechin and quercetin (Sun et al., 2016), plays a functional role in the tarsal gustatory repertoire.

Many reports have proposed that OBPs are expressed in gustatory organs and are involved in insect perception of hydrophobic substances to determine the host-seeking behaviors (Galindo and Smith, 2001; Matsuo et al., 2007; Jeong et al., 2013; Swarup et al., 2014). However, compared with the wellcharacterized process of olfactory perception, the physiological functions of OBPs associated with insect taste detection are far less clear. To date, direct evidence that insect OBPs contribute to gustation are confined to OBP28a (Swarup et al., 2014) and OBP49a (Jeong et al., 2013) as well as OBP57d/57e in D. sechellia (Matsuo et al., 2007). For mirid bugs, non-volatile host substances such as gossypol, catechin, and quercetin are crucial for determining whether plant species are suitable for feeding, and foreleg tarsi, which contain multiple taste sensilla, allow bugs to sensitively detect these biologically important substances. Therefore, we hypothesized that OBPs expressed on foreleg tarsi would be associated with the recognition of these contact substances on host plant surfaces. Eight candidate OBPs were identified through RNA sequencing and transcriptomic data analysis. This number was less than that previously reported for A. lineolatus antennae (Gu et al., 2011a) and lower than that identified in tarsi of the mosquito Aedes aegypti (Sparks et al., 2014). However, eight OBPs were comparable to the

number found in the proboscis taste organ in the sibling species Apolygus lucorum (Hua et al., 2012) and the number identified in the foreleg tarsi of the swallowtail butterfly Papilio xuthus (Ozaki et al., 2008). Furthermore, it is likely that chemosensory genes, particularly those encode sensilla lymph-biased OBPs are differentially expressed in distinct insect tissues during specific developmental/physiology life stages and can even be induced by chemical cues (Sun et al., 2014b; Wan et al., 2015).

Insect OBPs are grouped into different subfamilies, including classic, Plus-C, Minus-C, dimer, and atypical OBPs, according to sequence variations, and these structural differences likely enable OBPs to bind to different ligands with diverse sizes and shapes (Xu et al., 2003; Zhou et al., 2004; Zhou, 2010). Among the eight candidate OBPs identified from A. lineolatus foreleg tarsi, five OBPs (AlinOBP1, 2, 11, 16, and 17) belong to the classic subgroup, and three OBPs (AlinOBP7, 14, and 15) have features typical of the Plus-C OBP subfamily. Phylogenetic analysis of these eight OBPs and homologous OBPs from five mirid bug species revealed that mirid OBPs can be divided into two subgroups, classic and Plus-C, and that none were related to the minus-C or other subfamily groups. Furthermore, the OBPs were generally divergent within the same species, and each bug OBP clustered with at least one OBP protein from another species; species-specific clades were not observed.

The distinct tissue-biased distributions of OBP genes in insects are strongly indicative of biological function (Hull et al., 2014). Generally, an antenna-enriched expression profile is correlated with a role in olfactory perception, whereas genes that are strongly expressed in gustatory organs, such as the proboscis, tarsi and ovipositor, could be involved in taste detection (Pelosi et al., 2014; Brito et al., 2016). Our qRT-PCR results, in combination with previous reports (Gu et al., 2011a; Sun et al., 2016), indicate that these eight OBP genes have four distinct tissue expression patterns related to distinct physiological functions. For example, AlinOBP1, 2, 16, and 17 were enriched in the antennae, and AlinOBP 1 and 2 were demonstrated to be physiologically important for the detection of odorants such as female bug-produced butyrate sex pheromones and host plant terpenoids (Gu et al., 2011b). The two genes that encode AlinOBP14 and 15 (two Plus-C OBPs) were strongly expressed in the head, the non-chemosensory organ and their putative ligands have not been identified. The transcript-encoded protein AlinOBP11 was highly expressed in the gustatory organs,

(G–J) transmission electron microscopy (TEM), and (K–N) immunolocalization of AlinOBP11. Str, sensilla trichodea; Sch, sensilla chaetica; sl, sensillum lymph; isl, inner sensillum lymph; osl, outer sensillum lymph; w, sensillum wall; p, sensillum pore; d, dendrites; s, socket.

legs, and mouthparts (Sun et al., 2016) and is therefore a good candidate for the detection of non-volatile substances.

Insect foretarsi possess gustatory receptor neurons (GRNs) that are linked to the detection of specific sweet and bitter tastants (Sanchez et al., 2014). Our cellular immunolocalization labeling indicated that the taste organ-biased AlinOBP11 is strongly expressed in the outer sensillum lymph of the contact sensilla Sch2 (**Figures 5K–N**). This type of sensilla is the most abundant sensilla hair present on the foretarsi of adult A. lineolatus (**Figures 5A–F**), and its ultrastructure resembles the tarsal gustatory sensilla of the honey bee Apis mellifera (Sanchez et al., 2014), D. melanogaster (Nayak and Singh, 1983), and Helicoverpa spp. (Zhang et al., 2010; Zhang Y. F. et al., 2011), which have been demonstrated to account for the perception of sucrose and bitter substances. Furthermore, this cellular immunolocalization is consistent with previous reports of AlinOBP11 ligand-binding, which suggests that AlinOBP11 can tightly bind the bitter substances catechin and quercetin isolated from bug host plants (Sun et al., 2016). Hence, AlinOBP11 represents an attractive target for understanding the molecular basis of gustatory coding in A. lineolatus foretarsi, although there is currently no direct evidence supporting that Sch2 in A. lineolatus responds to bitter substances such as catechin and quercetin.

To date, two OBPs in A. lineolatus have been implicated in the perception of bitter substances, such as catechin and quercetin. One is the antennal contact sensilla-expressed AlinOBP6 (Sun et al., 2017b), and the other is AlinOBP11, which is expressed highly in mouthparts (Sun et al., 2016) and the tarsal gustatory sensillum lymph of Sch2 **(Figures 5K–N)**. These results indicate that mirid bug species, at least for A. lineolatus have evolved a complex gustatory repertoire to perceive important taste substances for host plant-seeking behavior. Such sophisticated taste recognition likely requires the activation of GRNs in taste sensilla located on antennae, mouthparts, and foretarsi and involves the cooperation of different OBPs. A combinatorial mechanism for the physiological function of OBPs in the gustatory system has been proposed in D. melanogaster (Swarup et al., 2014), however, this conclusion still requires in vivo evidence in A. lineolatus. In the future, gene expression modification by either RNA interference (He et al., 2011) or CRISPR/Cas9 editing (Zhu et al., 2016) should be used to clarify these issues.

#### AUTHOR CONTRIBUTIONS

LS and YZ conceived and designed the experimental plan. LS, QianW, and YX preformed the experiments. LS, KD, and QiW analyzed the data. LS and QianW drafted the manuscript.

#### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (31501652, 31471778, 31272048), China National Basic Research Program (2012CB114104), Central public-interest Scientific Institution Basal Research Fund (No. 1610212016015), Zhejiang Provincial Natural Science Foundation of China (LQ16C140003), Research Foundation of State Key Laboratory for Biology of Plant Diseases and Insect Pests (SKLOF201514 and SKLOF201719), Open Fund of Key Laboratory of Integrated Management of Crop Diseases and Pests, Ministry of Education/Key Laboratory of Integrated Pest Management on Crops in East China, Ministry of Agriculture, College of Plant Protection, Nanjing Agricultural University, and The Science and Technology Innovation Project of the Chinese Academy of Agricultural Sciences (CAAS-ASTIP-2015-TRICAAS).

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00735/full#supplementary-material

Figure S1 | Expression levels of putative OBP transcripts identified from the *A. lineolatus* foreleg transcriptomes assessed by RPKM values.

Table S1 | The "query" sequences used in candidate OBPs identification.

Table S2 | Primers used in this study.

Table S3 | OBP sequences (with signal peptides removed) used in the phylogenetic tree construction.

lucerne plant bug Adelphocoris lineolatus (Goeze). Insect Biochem. Mol. Biol. 41, 254–263. doi: 10.1016/j.ibmb.2011.01.002


of the blowfly, Phormia regina: putative role of an odorant-binding protein. Chem. Senses 28, 349–359. doi: 10.1093/chemse/28.4.349


Plutella xyllotella. J. Insect Physiol. 59, 46–55. doi: 10.1016/j.jinsphys.2012. 10.020


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Sun, Wang, Wang, Dong, Xiao and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Two Odorant-Binding Proteins of the Dark Black Chafer (Holotrichia parallela) Display Preferential Binding to Biologically Active Host Plant Volatiles

Qian Ju† , Xiao Li† , Xiao-Qiang Guo, Long Du, Chen-Ren Shi and Ming-Jing Qu\*

Shandong Peanut Research Institute, Qingdao, China

#### Edited by:

Peng He, Guizhou University, China

#### Reviewed by:

Jin Zhang, Max-Planck-Institut für chemische Ökologie, Germany Ya-Nan Zhang, Huaibei Normal University, China

> \*Correspondence: Ming-Jing Qu 13455277580@163.com

†These authors have contributed equally to this work and should be considered first co-authors.

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 05 December 2017 Accepted: 01 June 2018 Published: 18 July 2018

#### Citation:

Ju Q, Li X, Guo X-Q, Du L, Shi C-R and Qu M-J (2018) Two Odorant-Binding Proteins of the Dark Black Chafer (Holotrichia parallela) Display Preferential Binding to Biologically Active Host Plant Volatiles. Front. Physiol. 9:769. doi: 10.3389/fphys.2018.00769 The dark black chafer (DBC), Holotrichia parallela, is an important pest of multiple crops. Insect host-searching behaviors are regulated by host plant volatiles. Therefore, a better understanding of the mechanism linking the chemosensory system to plant volatiles at the molecular level will benefit DBC control strategies. Based on antenna transcriptome data, two highly expressed antenna-specific odorant-binding proteins (HparOBP20 and 49) were selected to identify novel DBC attractants using reverse chemical ecology methods. We expressed these proteins, mapped their binding specificity, and tested the activity of the plant volatiles in the field. The ligands used in the binding specificity assays included 31 host-plant-associated volatiles and two sex pheromone components. The results showed that (1) HparOBP20 and 49 are involved in odor recognition; (2) these proteins bind attractive plant volatiles strongly and can therefore be employed to develop environmentally friendly DBC management strategies; and (3) the green-leaf volatile (Z)- 3-hexenyl acetate shows a high binding affinity to HparOBP20 (Ki = 18.51 µM) and HparOBP49 (Ki = 39.65 µM) and is highly attractive to DBC adults, especially females. In the field test, a (Z)-3-hexenyl acetate trap caught an average of 13 ± 1.202 females per day, which was significantly greater than the corresponding male catch (F2,6 = 74.18, P < 0.0001). (Z)-3-Hexenyl acetate may represent a useful supplement to the known sex pheromone for DBC attraction. In the present study, the binding characteristics of two HparOBPs with host plant volatiles were screened, providing behaviourally active compounds that might be useful for DBC control, based on reverse chemical ecology.

Keywords: Holotrichia parallela, odorant-binding proteins, host plant volatiles, reverse chemical ecology, (Z)-3-hexenyl acetate

## INTRODUCTION

The dark black chafer (DBC), Holotrichia parallela Motschulsky (Coleoptera: Scarabaeidae), is an important pest in agriculture and forestry. DBC larvae, often referred to as grubs, live in soil and can cause significant damage to peanut, sweet potato, soybean, corn, and various other vegetable crops as well as to turf and ornamental species (Ju et al., 2012; Shan et al., 2014). Due to its cryptic and subterranean nature, this beetle is difficult to control. The main tactic employed for DBC

**167**

management is chemical control, which has environmentally detrimental consequences, such as residual toxicity, environmental contamination, and insecticide resistance. Mass trapping using sex pheromone-based attractants is an environmentally friendly control tactic and has become well established. However, this tactic has several shortcomings, including a male response bias to the sex pheromone traps and a short duration of residual activity (Reddy and Guerrero, 2004; Said et al., 2005). Similar to insect pheromones, plant volatiles are important signaling compounds that regulate insect behavior and exhibit potential as natural pesticides, lures, or antifeedants (Hanks et al., 2012; Hanks and Millar, 2013; Jung et al., 2013; Collignon et al., 2016; Wang F. et al., 2016; Wang Y.L. et al., 2016). Therefore, studies addressing the physiological and molecular basis of host plant selection could serve as an important basis for developing novel control tactics for the DBC (Koczor et al., 2012).

The interaction between plant volatiles and the insect olfactory system plays a critical role in the initial step of insect host orientation (Liu et al., 2015; Brito et al., 2016; Sun et al., 2017a). Plant volatiles consist of various classes of chemicals, such as green-leaf volatiles, general odorants and terpenoids (Aartsma et al., 2017). Due to the great diversity of plant volatiles, behavioral response methods for selecting active host plant volatiles require a great deal of time and effort. In this context, the reverse chemical ecology approach is gaining importance (Mao et al., 2010; Jayanthi et al., 2014), as it narrows down the number of odorant candidate compounds based on their binding affinity to olfactory proteins, saving time and reducing research costs compared with conventional trial-and-error screening performed in the field (Leal, 2017). Odorant-binding proteins (OBPs) are one of the major types of peripheral olfactory proteins involved in the reception of odorants in insects (Vogt et al., 1985; Klein, 1987; Leal, 2013). The physiological functions of insect OBPs have been described based on biochemical, biophysical, structural biology and kinetic studies (Sandler et al., 2000; Horst et al., 2001; Leal et al., 2005; Zhu et al., 2017), and it is clear that OBPs are important for transporting odorants through the sensillar lymph and increase the sensitivity of the olfactory system (Pelosi et al., 2014; Leal, 2017). The role of OBPs in the transport of molecules in insect antennae was described for the first time in Lepidoptera using male Antheraea polyphemus antennae (Vogt and Riddiford, 1981). Knockdown studies have demonstrated that DmelOBP76a (LUSH) is necessary for the olfactory process in Drosophila melanogaster (Xu et al., 2005; Laughlin et al., 2008). Furthermore, behavioural assays in Drosophila mutants (Matsuo et al., 2007; Swarup et al., 2011) and aphids (Qiao et al., 2009; Sun et al., 2012) have indicated that OBPs are involved in semiochemical detection. Previous studies have shown that a blend of volatiles derived from host plants can bind to OBPs and be used as a luring agent. A good example is provided by Loxostege sticticalis OBP2, which has been shown to exhibit a high affinity to host plant volatiles (Yin et al., 2012). OBP1 of Grapholita molesta exhibits dual functions in the recognition of host plant volatiles (Li et al., 2016). Two Spodoptera exigua OBPs share a common odorantresponse spectrum, with a considerable binding affinity to host odorants (Liu et al., 2017). Binding assays of two OBPs from H. oblita with various compounds showed that benzoates (leaf volatiles from host plants) fit inside the OBPs (Deng et al., 2012).

However, little is known about the molecular mechanisms underlying the interactions between DBCs and the odorous environment of their host plants. To date, only one report has described the binding functions of two OBPs in the DBC (Ju et al., 2012). Using a rapid amplification of cDNA ends (RACE) approach, the HparOBP1 and HparOBP2 genes were identified, and their ligand-binding properties were examined. Due to recent transcriptome projects, a large number of insect OBP sequences are available. Additionally, 25 OBP genes were obtained from the DBC whole-body transcriptome (Ju et al., 2014). However, the OBPs predicted from insect whole-body genomes are all unlikely to represent true olfactory proteins. In D. melanogaster, for instance, the OBP gene family comprises as many as 51 putative OBPs, but only seven of them have been demonstrated to be expressed specifically in adult olfactory organs (Galindo and Smith, 2001). At present, investigation of the antennal transcriptome is an effective way to find functional OBPs binding to plant volatiles. In this study, we identified the OBP genes expressed in DBC antennae using the transcriptome sequencing approach. Two HparOBPs were selected based on their specific phylogenetic position and antenna-specific expression pattern to determine their ligand-binding properties. Furthermore, the attractive properties of ligands binding to the two HparOBPs were verified in behavioral responses tests and field evaluations. Taken together, our results extend the knowledge of OBP genes in the DBC and pave the way for the development of novel environmentally friendly control tactics for DBC management.

## MATERIALS AND METHODS

#### Insects and Insect Maintenance

Adults DBCs were collected from the field of the experimental station at Shandong Peanut Research Institute, Qingdao, China. The beetles were separated into males and females and were reared with fresh elm tree (Ulmus pumila L.) leaves in a rotating chamber with aerating meshes. The relative humidity in the rearing chamber was maintained at 18–20%. Fresh antennae were obtained from both males and females for experimentation.

## Transcriptome Sequencing

Total RNA was isolated from adult antennae (∼200 antennae from both males and females). The RNA was quantified using a NanoDrop spectrophotometer (Thermo, Franklin, TN, United States). The mRNA was subsequently used for cDNA synthesis as described by Rice et al. (2000). cDNA synthesis, library construction and sequencing, gene annotation and prediction, and OBP identification and confirmation were conducted as described in previous articles (Ju et al., 2014). Briefly, the double-stranded cDNAs were fragmented into segments of 300–500 bp via sonication, and the sonicated mixture was purified using Agencourt-AMPure beads (Beckman, Schaumburg, IL, United States). A cDNA library was then generated using the TruSeqTM RNA Sample Prep Kit (Illumina, San Diego, CA, United States). The cDNA library was

subsequently sequenced on the Illumina HiSeq 4000 sequencing platform. Raw read quality was assessed using FastQC<sup>1</sup> prior to assembly, and Trimmomatic was used to filter adaptor sequences and trim reads bases with a PHRED quality score below 20. After adaptor filtering, the resulting reads were de novo assembled into contigs using the Trinity program. The 'align and estimate abundance' script in the Trinity package was used to align the reads and perform transcript abundance estimation using the RSEM method. The assembled contigs were further clustered using the TGI Clustering Tool (Pertea et al., 2003).

#### RNA Isolation, CDNA Synthesis, and PCR Cloning

The cloning primers were designed using Primer Express 3.0 and are listed in Supplementary Table S1. PCR was carried out in a total volume of 50 µl containing 200 ng of cDNA template, 5 µl of 10× buffer, 4 mmol/L MgCl2, 0.8 µmol/L of each forward and reverse primer, 1 mmol/L dNTPs, and 2.5 U of Taq polymerase. The PCR program started at 95◦C for 5 min for denaturation, followed by 25 cycles of 30 s at 95◦C, 30 s at 60◦C, and 30 s at 72◦C, with a final extension at 72◦C for 5 min.

#### Phylogenetic Analysis

Both the novel OBP genes identified in this study and the reported OBP gene sequences retrieved from previous studies were included in the phylogenetic analysis (Ju et al., 2014; Li X. et al., 2015). Multiple alignments of OBP genes were generated using MAFFT alignment software version 7.215 (Katoh et al., 2009). Based on the capability for parallelizing computation, the IQ-TREE program version 1.5 was employed to construct a phylogenetic tree using the protein sequences of these OBP genes according to the maximum likelihood principle (Lam-Tung et al., 2015). The best protein substitution model was selected by the built-in model-selection function of the IQ-TREE program, and bootstrap support values from 1000 replicates were assessed with ultrafast bootstrap approximation.

#### Fluorescence Competitive Binding Assay

Recombinant protein expression and purification were performed according to our previously reported protocols (Ju et al., 2012). Briefly, plasmid constructs containing the HparOBP genes were generated and transformed into Rosetta (DE3) competent cells for recombinant protein expression, and the resulting proteins were highly induced with 1 mM isopropyl ß-D-1-thiogalactopyranoside (IPTG) for 3–6 h at 37◦C. Purification was performed via Ni ion affinity chromatography (GE Healthcare, Beijing, China), and the His-tag was removed using enterokinase for HparOBP20 or tobacco etch virus (TEV) protease for HparOBP49. Renaturation and extensive dialysis were performed as previously reported (Ju et al., 2012), and the size and purity of the recombinant proteins were verified through SDS-PAGE.

For the ligand-binding assays, 33 compounds (>95% purity, Sigma-Aldrich, Shanghai, China) were selected based on their previously reported isolation from DBC host plants (Cheng et al., 2010; Lee et al., 2010; Shi et al., 2011; Ju et al., 2012; Tang et al., 2012; Iqbal et al., 2014). We used an F-380 fluorescence spectrophotometer (Tianjin, China) to determine the results of the perform the binding assay at room temperature (25◦C). The excitation wavelength was 337 nm, and the emission spectrum was recorded between 390 and 460 nm. N-phenyl-1-naphthylamine (1-NPN) is an effective fluorescent probe for insect OBP binding studies. First, we measured the constant emission of HparOBPs with 1-NPN, and titrated 2 µM proteins in 50 mM Tris-HCl (pH 7.4) with 1 mM 1-NPN in methanol to final concentrations ranging from 1 to 24 µM. Then, the affinities of other ligands were tested in competitive binding assays using 1-NPN as a fluorescent reporter at a concentration of 2 µM, while the concentration of each competitor ranged from 2 to 30 µM. We evaluated each bound chemical based on its fluorescence intensity, with the assumption that the protein was 100% active of 1:1 (protein/ligand) saturation. The binding curves were linearized using a Scatchard plot, and the dissociation constants of the competitors were calculated from the Scatchard plot of the binding data and the corresponding IC50 values based on the following equation: Ki = [IC50]/(1+[1- NPN]/K1−NPN), where [1-NPN] is the free concentration of 1-NPN, and K1−NPN is the dissociation constant of the complex protein/1-NPN.

#### Electroantennogram (EAG) and Olfactory Response Assays

The biologically attractive effects of chemicals with an ability to bind HparOBPs strongly were tested.

The EAG responses of virgin male/female antennae were measured after removing the tips of the three antennal lamellae (at approximately 1 mm) and separating them from each other. The chemicals used for the EAG and behavior assays were diluted in methanol (HPLC grade) to varying concentrations (0.1, 1, and 10 µg/µl), and methanol was used as a control. A 10 µl aliquot of each concentration was applied to a filter paper (25 × 8 mm). EAG responses were recorded for 5 s, with a stimulation interval of 30 s and a flow rate of 4 ml/s for both the stimulant and purge airflow. Each chemical was tested against six antennae, and each antenna was tested with three repeated stimulations. The EAG apparatus consisted of a signal acquisition system (IDAC-4), a micromanipulator assembly (INR-5), a stimulus controller (CS-05), and a system for outputting the EAG results (Syntech Company, Holland).

The behavioral responses of female and male adults to the putative ligands were tested using a Y-tube olfactometer in a dark room at 27 ± 1 ◦C. A filter paper (25 × 8 mm) with 10 µl of the test compound was placed at the end of one arm of the Y-tube, with 10 µl of methanol at the end of the other arm (control tube). The airflow was 500 ml/min. Six replicates were performed for each stimulant with 10 healthy virgin adults in the main stem of the Y-tube, and 10 min was allowed for their distribution. The response rate was calculated according to the following equations: response rate = (T+C)/SUM and selective response rate = T/(T+C), where T represents the number of

<sup>1</sup>http://www.bioinformatics.babraham.ac.uk/projects/fastqc

FIGURE 1 | Phylogenetic tree of OBP genes in the DBC and A. corpulenta. The tree was constructed using IQ-TREE version 1.5.

beetles in the treatment tube; C indicates the number of beetles in the control tube; and SUM is the number of beetles tested.

#### Field Evaluation

According to the laboratory evaluation, in addition to the main sex pheromone component, L-leucine methyl ester, (Z)- 3-hexenyl acetate was considered as a candidate compound for attracting DBC adults in the field. The tested chemicals were individually dissolved with methanol to 360 mg/ml. A dispenser was constructed using oil-free cotton wool with 360 mg of the tested chemical and stored in a freezer before use. Methanol was employed as a control. The treatments were as follows:


All experiments were performed at the experimental station of the Shandong Peanut Research Institute, Lai Xi Wang Cheng, Qingdao, China. Traps were placed in middle of the field, and each trap was located 60 m from any other trap, so that the individual treatments were 60 m apart. To avoid crosscontamination, only one compound was tested at each sub-site at any time, and each compound was tested at only one sub-site (Isberg et al., 2017). Each trap was set to operate from 1 h before sunset until 1 h after sunset. Three traps (replicates) were selected for each treatment. The test period was June 1–20, 2017. As a rule, the traps were checked every day, and the individuals that were caught in all experiments were sexed.

FIGURE 3 | Saturation binding curves and relative Scatchard plots of the affinity of 1-NPN to HparOBPs. The dissociation constants of 1-NPN with the HparOBPs were 7.439 ± 1.45 (HparOBP20) and 14.67 ± 2.96 (HparOBP49), respectively.

## RESULTS

#### Characterization of Antenna OBP-Encoding Genes

A total of 30,338,129 paired-end reads were produced with a read length equal to 150 bp (Data Availability Statement: All the illumina sequencing data are available from the SRA database, accession number SRP148674). After low-quality filtering and adaptor cleaning, 30,297,575 filtered reads (representing 99.87% of total raw reads) were used for de novo assembly, resulting in a total of 106,562 contigs with an N50 length of 1,351 bp. The metrics of the DBC transcriptome assemblies were compared with those of the pine shoot beetle transcriptome (Zhu et al., 2012). The quality of these two transcriptome assemblies was comparable, indicating that the DBC assembly was suitable for downstream transcriptome analyses.

#### Phylogenetic Analysis

A total of 48 HparOBP-encoding transcripts (containing 113– 223 amino acids) were identified through BLAST searches. The

OBPs of Anomala corpulenta and the DBC were employed to construct a phylogenetic tree (**Figure 1**). The phylogenetic tree showed that HparOBP20 (KR733566.1) and HparOBP49 (KR733548.1) were clustered with AcorOBP7 and AcorOBP8. Analysis of expression levels indicated that compared with other HparOBPs, HparOBP20 and HparOBP49 showed higher


(Continued)

#### TABLE 1 | Continued

fphys-09-00769 July 18, 2018 Time: 16:26 # 8


Each value was obtained from three independent experiments. IC50 values labeled ">50" indicate that the binding affinity could not be calculated directly with the tested ligand concentrations. Therefore, the Ki values of these ligands are designated "−".

transcriptional activity. Therefore, HparOBP20 and HparOBP49 were selected for further analysis due to their tissue-specific expression pattern and high transcriptional activity in antennae (Ju et al., 2014).

#### In Vitro Expression, Purification of Recombinant HparOBPs and Fluorescence Binding Assays of HparOBPs

Recombinant HparOBPs were expressed in bacterial expression systems and purified using Ni ion affinity chromatography. SDS-PAGE analysis of the recombinant proteins showed that their molecular weights were 14–18 kDa, consistent with their predicted molecular masses (**Figure 2**).

NPN can be used as a probe in fluorescence binding assays of insect OBPs, and the binding properties of 1-NPN to OBPs have been well characterized (Sun et al., 2013; Zhuang et al., 2014; Li D.Z. et al., 2015). Therefore, 1-NPN was employed to establish saturation binding curves and Scatchard plots (**Figure 3**). The dissociation constants of 1-NPN with the HparOBPs, calculated using Scatchard plots, were 7.439 ± 1.45 (HparOBP20) and 14.67 ± 2.96 (HparOBP49), respectively.

A total of 33 semiochemicals, including 31 host plantassociated volatiles and two sex pheromone components, were selected for fluorescence binding assays (**Figure 4** and **Table 1**). Among the 17 general odorants, HparOBP20 showed broad binding activity from Ki = 13.84 µM (pentadecane) to 40.60 µM (dodecane); HparOBP49 specifically bound to hexanoic acid with a Ki of 42.20 µM. Among the eight green-leaf volatiles (GLVs), (Z)-3-hexenyl acetate showed a high binding affinity to HparOBP20 and HparOBP49, with Ki values of 18.51 and 39.65 µM, respectively. In addition, (E)-2-hexenyl acetate, (Z)-3 hexen-1-ol and (E)-3-hexen-1-ol showed high binding affinities to HparOBP20, with Ki values of 23.25, 25.21, and 25.37 µM, respectively. Among the six terpenoids, HparOBP20 bound to α-pinene and (R)-(+)-limonene with Ki values of 22.41 and 23.99 µM. None of the tested terpenoids could displace 1-NPN bound to HparOBP49.

#### EAG and Olfactory Responses to Host-Associated Volatiles and Sex Pheromone Components

Based on the results of the fluorescence binding assays, five putative ligands of the recombinant HparOBPs were selected as candidates for EAG testing in both male and female antennae (**Figure 5**). In males, the highest responses were observed for L-leucine methyl ester and (Z)-3-hexenyl acetate at 1 µg/µl, with EAG responses of 5.68 and 4.02 mV, respectively. The highest response for females was observed for (Z)-3-hexenyl acetate at 1 µg/µl, with an EAG response of 4.84 mV. The dose-dependent EAG responses to (Z)-3-hexenyl acetate were similar in the two sexes. Significantly different EAG responses

FIGURE 6 | Behavioral responses of male and female DBCs to five putative HparOBP ligands at 1 µg/µl in a Y-tube olfactometer trial. A high response rate (greater than 80%) was observed for all DBC individuals. The indices were calculated using the following formulas: response = T/SUM or C/SUM, response rate = (T+C)/SUM and selective response rate = T/(T+C), where T represents the number of beetles in the treatment tube; C indicates the number of beetles in the control tube; and SUM is the number of beetles tested. Mean ± SE (N = 6). Asterisks indicate statistically significant differences between females and males (by Student's t-test): <sup>∗</sup>P < 0.05, ∗∗P < 0.01.

between the sexes were found for L-leucine methyl ester, with male antennae being more responsive than female antennae (t = 12.062, P < 0.01 for L-leucine methyl ester at 0.1 µg/µl; t = 11.635, P < 0.01 for L-leucine methyl ester at 1 µg/µl; and t = 19.231, P < 0.01 for L-leucine methyl ester at 10 µg/µl). At the concentration of 1 µg/µl, β-caryophyllene elicited a significantly higher response in female antennae than in male antennae (t = 5.350, P < 0.01).

**Figure 6** summarizes the olfactory responses of DBC adults to the tested volatiles at 1 µg/µl. A good response rate (>80%) suggested that the tests were valid. Similar to the EAG responses, the highest selective response rate of females to (Z)-3-hexenyl acetate was 98%. A significantly higher selective response rate in males (96%) than in females was observed for L-leucine methyl ester. Significant differences in behavioral responses were observed between the controls and treatments for α-phellandrene and L-leucine methyl ester, with the treatment being more attractive than the control (t = 13.738, P < 0.01 for α-phellandrene; t = 13.538, P < 0.01 for L-leucine methyl ester). Females exhibited upwind movement into the volatiles containing pentadecane and (Z)-3-hexenyl acetate (t = 6.010, P < 0.01 for pentadecane, t = 23.756, P < 0.01 for (Z)-3-hexenyl acetate). L-Leucine methyl ester, an established sex pheromone component, attracted few female adults (t = −6.188, P < 0.01).

#### Field Evaluation

fphys-09-00769 July 18, 2018 Time: 16:26 # 11

Based on the EAG and olfactory responses, L-leucine methyl ester and (Z)-3-hexenyl acetate were selected for field evaluation (**Figure 7**). The results showed that all of the tested lures attracted more males than females. The sex pheromone resulted in significantly higher male catches than (Z)-3-hexenyl acetate (F2,<sup>6</sup> = 272.1, P < 0.0001). For males, (Z)-3-hexenyl acetate yielded 83 ± 4.933 DBCs, and the sex pheromone yielded 258 ± 12.860. The average number of females per trap per day was 13 ± 1.202 using (Z)-3-hexenyl acetate (F2,<sup>6</sup> = 74.18, P < 0.0001).

#### DISCUSSION

In this study, we focused on OBPs, which are relatively accessible targets for research, because they are small, soluble, stable and relatively easy to manipulate and modify (Brito et al., 2016; Leal, 2017; Zhu et al., 2017). A. corpulenta Motschulsky (Coleoptera: Scarabaeidae: Rutelinae) and DBC larvae, which are the main pests in many crop fields, exhibit overlapping active times, and adults of these species also overlap on some host plant species. Therefore, these pests may exhibit similar olfactory proteins in their olfactory systems, which could be the functional proteins interacting with plant volatiles. In A. corpulenta, AcroOBP7 and AcroOBP8 display antenna-specific expression (Li X. et al., 2015), and HparOBP20 and HparOBP49 exhibit antenna-specific expression in H. parallela (Ju et al., 2014). We have revised the nomenclature system of the Holotrichia parallela OBP genes in this paper. The OBP2 gene from the previous study (Ju et al., 2014) has been renamed HparOBP49. We hypothesize that these proteins are responsible for chemical communication, and the phylogenetic tree of A. corpulenta and H. parallela showed that HparOBP20 and HparOBP49 clustered with AcorOBP7 and AcorOBP8. Furthermore, an analysis of expression levels indicated that HparOBP20 and HparOBP49 showed higher transcriptional activity than that of other HparOBPs. Therefore, HparOBP20 and HparOBP49 were selected for further study. Their binding specificity may pave the way for the identification of active host plant volatiles.

To confirm the functions suggested by the phylogenetic tree, along with the tissue expression profiles and quantification analysis, the binding affinity of the two HparOBPs to 33 volatiles was determined using fluorescent binding assays. All the volatile compounds tested in this study were isolated from DBC host plants and may be biologically significant for the DBC. We found that HparOBP20 showed a broad spectrum of binding activity, and HparOBP49 specifically bound to general odorants and GLVs. Overall, HparOBP20 exhibited a high binding affinity to three volatiles (Ki < 20 µM): pentadecane, (Z)-3-Hexenyl acetate and α-phellandrene. However, all the volatiles tested in this study showed a relatively weak binding affinity (Ki > 20 µM) to HparOBP49. Compensation effects may exist between HparOBP20 and HparOBP49, as observed for Cnaphalocrocis medinalis OBP2 and OBP3 (Sun et al., 2016), and Chrysopa pallens OBP3, −6 and −10 (Li et al., 2017).

Among the three compounds that displayed a high binding affinity (Ki < 20 µM) to HparOBP20, (Z)-3-hexenyl acetate showed a higher affinity to both HparOBP20 and HparOBP49. (Z)-3-Hexenyl acetate is a GLV metabolized from one of the most abundant GLVs, (Z)-3-hexenal (Deng et al., 2004; Matsui, 2006; D'Auria et al., 2007; Allmann et al., 2013). (Z)-3-Hexenyl acetate is a common plant volatile released in large amounts after damage and plays important roles in insect-plant interactions (Arimura et al., 2008; Mumm and Dicke, 2010; Szendrei et al., 2011; von Arx et al., 2012). For example, a mixture of plant volatiles including (Z)-3-hexenyl acetate attracts the Colorado potato beetle, Leptinotarsa decemlineata Say (Visser, 1986), and the scarab beetle Anomala octiescostata Burmeister (Leal et al., 1994). Here, we tested the behavioral response and field attraction of the DBC to (Z)-3-hexenyl acetate, and a clear behavioral influence was observed in the EAG, Y-tube and field evaluations. The results were consistent with those of previous studies. Furthermore, (Z)-3-hexenyl acetate has been confirmed to activate olfactory sensory neurons (OSNs) expressing different sets of odorant receptor types on Manduca sexta female antennae (Allmann et al., 2013) and to enhance the responses of some insect species to sex pheromones (Deng et al., 2004; Varela et al., 2011; Ju et al., 2017). In the field, (Z)-3-hexenyl acetate mixed with sex pheromone in a 1:1 ratio increased the number of trapcaught females by 6- to 7-fold and the number of males by 20–30% compared with traps baited with sex pheromone alone (Reddy and Guerrero, 2000). Therefore, the synergistic effect between (Z)-3-hexenyl acetate and the sex pheromone requires further study. However, it is worth noting that, while (E)-2 hexenyl acetate displayed a lower binding affinity (Ki = 23.25) to HparOBP20, when it was employed in a trap along with the DBC sex pheromone, many DBCs were caught (Ju et al., 2017).

The general odorants pentadecane and α-phellandrene showed a higher binding affinity (Ki < 20 µM) to HparOBP20 and exerted a clear influence on behavior in the EAG and Y-tube assays but exhibited a low attractant ability in traps. Fluorescence binding assays often provide candidate compounds, but not all of the screened compounds exhibit biological activity in insects (Yi et al., 2018). AfunOBP1 from Anopheles funestus binds to 1-octen-3-ol, but when 1-octen-3-ol was used in a trap, only a few mosquito species were caught (Xu et al., 2010). However, pentadecane has been reported to bind to a Locusta migratoria OBP (Jiang et al., 2009), and the molecular docking results for α-phellandrene showed that it could tightly bind to the Adelphocoris lineolatus OBP6 pocket (Sun et al., 2017b). In the future, we may focus more research effort on these two odorants to obtain a greater number of DBC attractants.

#### AUTHOR CONTRIBUTIONS

QJ and XL analyzed and interpreted the data and performed the molecular examination. X-QG and LD performed the EAG and olfactory response examination.

C-RS performed the field examination. M-JQ was a major contributor in writing the manuscript. All authors read and approved the final manuscript.

#### FUNDING

This study was funded by research grants from the National Key R&D Program of China (2017YFD0200400), China Agriculture

#### REFERENCES


Research System (CARS-14), and the Youth Fund of Shandong Academy of Agriculture Sciences (2016YQN15).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00769/full#supplementary-material


nonvolatile host compounds in Adelphocoris lineolatus (Goeze). Insect Mol. Biol. 26, 601–615. doi: 10.1111/imb.12322


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ju, Li, Guo, Du, Shi and Qu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fphys-09-00769 July 18, 2018 Time: 16:26 # 14

# Identification of Odorant-Binding Proteins (OBPs) and Functional Analysis of Phase-Related OBPs in the Migratory Locust

Wei Guo<sup>1</sup> , Dani Ren<sup>1</sup> , Lianfeng Zhao<sup>2</sup> , Feng Jiang<sup>2</sup> , Juan Song<sup>1</sup> , Xianhui Wang<sup>1</sup> \* and Le Kang1,2 \*

<sup>1</sup> State Key Laboratory of Integrated Management of Pest Insects and Rodents, Institute of Zoology, Chinese Academy of Sciences, Beijing, China, <sup>2</sup> Beijing Institute of Life Science, Chinese Academy of Sciences, Beijing, China

#### Edited by:

Shuang-Lin Dong, Nanjing Agricultural University, China

#### Reviewed by:

Xin-Cheng Zhao, Henan Agricultural University, China Guirong Wang, Institute of Plant Protection (CAAS), China Tom Matheson, University of Leicester, United Kingdom

#### \*Correspondence:

Xianhui Wang wangxh@ioz.ac.cn Le Kang lkang@ioz.ac.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

> Received: 25 April 2018 Accepted: 04 July 2018 Published: 20 July 2018

#### Citation:

Guo W, Ren D, Zhao L, Jiang F, Song J, Wang X and Kang L (2018) Identification of Odorant-Binding Proteins (OBPs) and Functional Analysis of Phase-Related OBPs in the Migratory Locust. Front. Physiol. 9:984. doi: 10.3389/fphys.2018.00984 Olfactory plasticity, which is one of the major characteristics of density-dependent phase polyphenism, plays critical roles in the large-scale aggregation formation of Locusta migratoria. It is still unknown whether odorant-binding proteins (OBPs) are involved in phase-related olfactory plasticity of locusts, despite the confirmed involvement of several types of olfactory perception genes. In this study, we performed a large-scale search for OBPs and verified their expression patterns in the migratory locust. We identified 17 OBPs in the L. migratoria genome, of which 10 were novel, and we found their scattering distribution characteristics by mapping the genomic loci. Next, we revealed that these OBPs with close phylogenic relationships displayed similar tissue-specific expression profiles by a combined analysis of qRT-PCR and phylogenetic tree reconstruction. In all identified locust OBPs, seven OBPs showed differential mRNA expression levels in antenna tissue between gregarious and solitarious nymphs. Six of these seven OBPs displayed higher mRNA expression in the antennae of gregarious nymphs. The mRNA expression of LmigOBP2 and LmigOBP4 increased during gregarization and decreased during solitarization. RNAi experiments confirmed that only LmigOBP4 regulates the behavioral traits to affect gregarious behavior. These results demonstrated that OBPs also play important roles in the regulation of phase-related behavior of the locusts.

Keywords: odorant-binding proteins (OBPs), phase-related behavior, expression profile, locust aggregation, RNAi

## INTRODUCTION

The olfactory sense plays a critical role in behaviors related to food selection, host seeking, courtship, aggregation, and avoidance in insects when receiving external chemical cues (Leal, 2005; Pelosi et al., 2006; Benton, 2007). Despite the diversity of antennal morphs, sensillum types and olfactory gene repertoires among insect species, a general olfactory pathway has been proposed, extending from the reception of odorants to their transmission to odorant receptors (ORs) and activation of an olfactory sensory neuron (OSN) to the projection in the glomerulus in the antennal lobe and coding in higher brain centers (Pelosi et al., 2006; Carey and Carlson, 2011; Leal, 2013).

**181**

However, the reception of odorants has long been a question because of the complex mixture of numerous and high concentration of olfactory proteins around the dendrites of the OSNs in the sensillum lymph (Pelosi et al., 2006).

As one of the most important chemoreception proteins in insects, odorant-binding proteins (OBPs) have been suggested to play important roles in the reception of odorants. OBPs belong to a large gene family with low protein conservation among family members (Vieira and Rozas, 2011). Generally, these genes are abundantly expressed in chemosensory sensilla, especially in the antennae and labial/maxillary palps. Recent studies supposed that OBPs mainly act as transporters to deliver volatiles or non-volatile chemicals to the ORs and mediate the first step of olfactory signal transmission (Fan et al., 2011). OBPs contribute to insect olfactory perception at various levels. Depending on the types of ligands, OBPs transmit chemical signals to ORs to give rise to corresponding behavioral responses among conspecific insects and across species (Fan et al., 2011; Leal, 2013). OBPs have been reported to be involved in the reception of some oviposition attractants and the determination of reproductive sites by regulating the sensitivity of the insect's olfactory system (Harada et al., 2008; Pelletier et al., 2010). In addition, OBPs can modulate feeding behavior by regulating the perception to host plant odorants or by affecting sucrose intake in response to bitter compounds (Swarup et al., 2014; Li et al., 2016).

Locusts are one of the most important agricultural pests in the world because of the plague outbreaks resulting from swarm formation and large-scale migration. They display density-dependent behavioral plasticity in transitioning from the disconsolate "solitarious" to the manic "gregarious" phase (Pener and Simpson, 2009). Our recent studies indicated that olfactory regulation related to phase change is a complex process when integrated with the environmental input, gene interaction network, and phenotypic output (Kang et al., 2004; Chen et al., 2010; Guo et al., 2011; Ma et al., 2011). Olfactory perception displays significant differences between solitarious and gregarious locusts in the peripheral and central olfactory nervous systems (Guo et al., 2011; Wang and Kang, 2014; Wang et al., 2015). In the peripheral olfactory perception system, we have found that the olfactory genes, CSP and takeout, initiate behavioral aggregation by balancing attraction and repulsion responses to conspecific other individuals (Guo et al., 2011). An OR-based signaling pathway mediates the attraction of locusts to aggregation pheromones (Wang et al., 2015). Recent studies have identified 14 OBPs in Schistocerca gregaria and determined their distinct sensilla-specific expression patterns (Jiang et al., 2017, 2018). However, it is still unknown whether OBPs are involved in the regulation of phase-related behavioral plasticity in locusts.

In this study, we performed a large-scale search for OBPs in the Locusta migratoria genome and analyzed their phylogenetic relationships. A qRT-PCR technique was used to investigate the temporal-spatial expression of OBP genes. RNAi and behavioral assays were used to elucidate the potential function of OBP genes on the behavioral plasticity. We found that OBPs might also be involved in the regulation of the locust phase-related behavior of locusts.

## MATERIALS AND METHODS

#### Insects

The locusts were from the gregarious and solitarious colonies in the Institute of Zoology, CAS, China. Gregarious cultures were reared in large, well-ventilated cages (25 cm × 25 cm × 25 cm) at densities of 200 to 300 insects per cage. Reared solitarious insects were kept in physical, visual, and olfactory isolation that was achieved by ventilating each cage (10 cm × 10 cm × 25 cm) with charcoal-filtered compressed air. Rearing conditions of both colonies were under a 14 h/10 h light/dark photo regime at 30 ± 2 ◦C on a diet of fresh greenhouse-grown wheat seedlings and wheat bran. Fourth-instar gregarious and solitarious nymphs were used in all of the following experiments.

## Experimental Samples

To investigate the tissue-specific expression profiles of OBPs, tissues of antennae, labial palps, brains, wings, and hind legs were collected from gregarious and solitarious nymphs. To investigate the expression profiles of OBPs during phase changes, all the insects were sampled at the same time point (9:00 am) and antenna tissues were collected after 0, 4, 8, and 16 h of solitarization or gregarization. Six individuals were dissected and pooled into one biological replicate and four biological replicates were sampled for each experiment. The sexual ratio of all samples was 1:1. All these samples were stored in liquid nitrogen for further use.

#### Identification of OBP Genes and Molecular Cloning of Novel OBPs

We first searched the genes annotated as putative OBPs in the gene set of locust genome (Wang et al., 2014) and the locust transcriptome (Chen et al., 2010). Then, we aligned these sequences and assembled the sequences with high similarity to acquire as long as cDNA sequences by using Geneious Pro 4.8.6 (Biomatters Ltd.). Finally, we confirmed the identified OBPs sequences by Sanger sequencing. According to the assembled sequences, we designed the gene-specific PCR primers for 10 novel OBPs (**Supplementary Table S1**). PCR amplifications were conducted using an ABI veriti thermal cycler and initiated with a 2-min incubation at 94◦C, followed by 35 cycles of 94◦C, 20 s; 56◦C, 20 s; and 72◦C, 40 s. PCR products were cloned into T-easy vector (Promega) and sequenced.

#### qRT-PCR Analysis

Total RNA was extracted using the RNeasy Mini Kit (QIAGEN) according to the manufacturer's protocol. The cDNA was reverse-transcribed from 2 µg (0.5 µg from the labial palps) DNase-treated total RNA using MMLV reverse transcriptase (Promega). The mRNA expression level was measured using a SuperReal PreMix Plus (SYBR Green) Kit (Tiangen Biotech, Beijing) and normalized to ribosome protein 49. PCR cycling conditions were based on the manufacturer's recommendations. PCR amplification was conducted using a Roche Light cycler 480. A melting curve analysis was performed to confirm the specificity of amplification. All samples from IG and CS treatments to test for one OBP gene were run on one individual plate. The qRT-PCR primers of 17 OBP genes are listed in **Supplementary Table S1**.

#### Phylogenetic Analysis

fphys-09-00984 July 18, 2018 Time: 16:18 # 3

MAFFT online version<sup>1</sup> was used for multiple sequence alignment with the G-INS-1 method and BLOSUM 62 scoring matrix. MEGA 7.0 software was used for the phylogenetic analysis (Kumar et al., 2016). The evolutionary history was inferred using the neighbor-joining method, and the bootstrap consensus tree was inferred from 500 replicates. Branches corresponding to partitions reproduced in less than 50% bootstrap replicates were collapsed. The evolutionary distances were computed using the Poisson correction method, and the units are the number of amino acid substitutions per site.

#### RNA Interference

Double-stranded RNA (dsRNA) of green fluorescent protein (GFP), LmigOBP2 and LmigOBP4 was prepared using the T7 RiboMAX Express RNAi system (Promega) following the manufacturer's instructions. Fourth-instar gregarious or solitarious nymphs were injected with 18 µg (6 µg/µl) of dsGFP, dsLmigOBP2, or dsLmigOBP4 in the second ventral segment of the abdomen ∼12 h after molting. The injected gregarious locusts were later marked and placed back into gregarious-rearing cages. Three days later, the effects of RNAi on the mRNA relative expression levels were investigated by qRT-PCR, and the behavior was examined as described below.

#### Behavioral Assay

The arena assay experiment was performed in a rectangular Perspex arena (40 cm long × 30 cm wide × 10 cm high) with opaque walls and a clear top. One of the separated chambers (7.5 cm × 30 cm × 10 cm) contained 15 fourth-instar gregarious locusts as a stimulus group, and the other chamber was left empty. Before measurement, the locusts were restricted in a Perspex cylinder for 2 min. Then the locusts were released into the arena and monitored for 300 s. An EthoVision video tracking system (Netherlands, Noldus Information Technology) was used to automatically record individual behavior. A binary logistic model, Pgreg = e<sup>η</sup> /(1+e η ), η = −2.110 + 0.005 × attraction index + 0.012 × total distance moved + 0.015 × total duration of movement, was used to measure the behavioral phase state of individual locusts (Guo et al., 2011). Pgreg = 1 means fully gregarious behavior and Pgreg = 0 means fully solitarious behavior.

#### Data Analysis

Data were analyzed using the IBM SPSS Statistics v.19 software (SPSS Inc., Chicago, IL, United States). Differences between treatments were compared either by Student's t-test or by one-way analysis of variance (ANOVA) followed by a Tukey's test for multiple comparisons. Behavior-related data were analyzed by Mann–Whitney U test because of its nonnormal distribution characteristics. Differences were considered significant at p < 0.05. Values are reported as means ± SE.

## RESULTS

#### Identification, Sequence Alignment, and Genomic Loci of the OBPs

Based on the L. migratoria genome assembly v.2.4, we have identified 17 genes encoding putative OBPs. Of them, seven members that were already known from previous studies: LmigOBP1, LmigOBP2, LmigOBP3, LmigOBP4, LmigOBP5, LmigOBP15, and LmigOBP16 (**Table 1**). Next, we confirmed the cDNA sequences of 10 novel OBPs by PCR cloning and sequencing based on the transcriptomic and genomic sequences (Chen et al., 2010; Wang et al., 2014). The deduced protein lengths of these OBPs ranged from 124 to 271 amino acids, and 10 of the 17 OBPs have predicted signal peptides. Sequence alignment and analysis indicated that these deduced proteins belong to four subtypes: classic, plus-C type-A, plus-C type-B, and atypical OBPs (Jiang et al., 2017). All identified OBPs had six conserved cysteine residues (C1–C6), a three amino acid interval between C2 and C3, and an eight amino acid interval between C5 and C6 (**Figure 1**). Eleven of the 17 OBPs are classic OBPs. LmigOBP3, LmigOBP4, LmigOBP7, and LmigOBP11 had an additional three cysteine residues (C3<sup>0</sup> , C4<sup>0</sup> , and C5<sup>0</sup> ) and belong to the plus-C type-A subtype. LmigOBP6, which belongs to the plus-C type-B subtype, had two additional cysteine residues with one in front of the C1 residue and one behind the C6 residue. LmigOBP16 is a long OBP with 271 amino acids and is classified as an atypical OBP.

These 17 OBPs genes were scattered on 16 scaffolds of the whole genome sequences (**Table 2**). Only LmigOBP5 and LmigOBP8 were on the same scaffold with an ∼32 kb intergenic region. The length of the OBP genes ranged from 10374 bp (LmigOBP2) to 79617 bp (LmigOBP9), in which LmigOBP1, LmigOBP6, LmigOBP8, LmigOBP9, and LmigOBP16 had seven exons and the other 12 OBPs all had six exons. The length of the exons and introns ranged from 20 bp to 210 bp and 83 bp to 58448 bp, respectively. Like the other coding genes of L. migratoria (Wang et al., 2014), the OBP genes also had many long introns and the lengths of ∼55% of the introns is more than 5 kb (**Table 2**).

#### Tissue-Specific Expression Profiles and Phylogenetic Analysis of the OBPs

We determined the expression levels of locust OBP genes in five tissues including the antenna, labial palp, brain, wing, and hind leg. Their expression profiles can be divided into five patterns: (a) Eight OBPs are antenna-rich expressions, including LmigOBP1, LmigOBP2, LmigOBP4, LmigOBP5, LmigOBP9, LmigOBP10, LmigOBP13, and LmigOBP14; (b) Five OBPs are labial palprich expressions, including LmigOBP7, LmigOBP11, LmigOBP12, LmigOBP15, and LmigOBP17; (c) Two OBPs are antenna and labial palp-rich expressions, including LmigOBP3 and

<sup>1</sup>https://mafft.cbrc.jp/alignment/server/

Frontiers in Physiology | www.frontiersin.org


TABLE 1 | Detailed information of 17 locust OBP genes.

fphys-09-00984 July 18, 2018 Time: 16:18 # 4

OBPs identified in previous studies are marked with an<sup>∗</sup> .

LmigOBP16; (d) One OBP is a brain-rich expression, including LmigOBP8; and (e) One OBP is a multi-tissue expression, including LmigOBP6 (**Figure 2**).

We further compared the expression levels of these OBPs between gregarious and solitarious locusts. Nine of 17 OBP genes were differentially expressed in a range of tissues between the gregarious and solitarious locusts (**Figure 2**; shading in yellow). Seven OBPs, LmigOBP1, LmigOBP2, LmigOBP4, LmigOBP5, LmigOBP9, LmigOBP14, and LmigOBP16, were differentially expressed in the antennal tissue (Student's t-test, t = 3.311, 4.973, 7.747, 3.510, 3.689, 3.079, 3.411; p = 0.002, 0.004, 0.001, 0.017, 0.014, 0.027, 0.041, respectively). Except for LmigOBP16, the other six OBPs were highly expressed in the antennal tissue of gregarious locusts. Three OBPs, LmiOBP2, LmigOBP15, and LmigOBP16, were highly expressed in the labial palp tissue of gregarious locusts (Student's t-test, t = 8.869, 3.704, 4.493; p = 0.002, 0.033, 0.020, respectively). Two OBPs, LmigOBP8 and LmigOBP16, were highly expressed in the brain tissue of gregarious locusts (Student's t-test, t = 2.639, 4.074, p = 0.039, 0.007, respectively).

The 17 OBPs were classified into four clades according to their phylogenetic relationships. Except for LmigOBP2 in clade 3, the classic OBPs are all in clades 1 and 2. Four plus-C type-A OBPs are all in clade 3. LmigOBP6 (plus-C type-B OBP) and LmigOBP16 (atypical OBP) are both in clade 4 (**Figure 2**). In clade 1, seven OBPs are divided into three branches, which represent antenna-, labial palp-, and brain-rich expression OBPs. OBPs with close phylogenetic relationships had similar gene expression patterns, such as the gene expression among LmigOBP5, LmigOBP10, and LmigOBP13. In clade 2, three OBPs had a similar tissue expression pattern and were all highly expressed in gregarious antenna tissue. In clade 3, four OBPs displayed diverse antenna-labial palp expression patterns. The LmigOBP2 gene was expressed significantly higher in both antenna and labial palp tissues of gregarious nymphs (**Figure 2**). In clade 4, LmigOBP16 was significantly highly expressed in the solitarious antenna, gregarious labial palp, and brain. The expression level of LmigOBP6 displayed no significant differences in different tissues or phase individuals (**Figure 2**).

#### Phylogenetic Analysis of Related Insect OBPs

We used 31 OBPs from holometabolous Drosophila melanogaster and 14, 17, and 16 OBPs from hemimetabolous S. gregaria, L. migratoria, and Acyrthosiphon pisum, respectively, for phylogenetic analysis (**Figure 3** and **Supplementary Data Sheet S1**). The phylogenetic relationships indicated that L. migratoria OBPs is distributed in four families together with the OBPs of other insect species. Family I includes most OBPs from D. melanogaster and many OBPs are specific to D. melanogaster. SgreOBP2 has no homolog in L. migratoria, and AcpiOBP10 and AcpiOBP15 are A. pisum-specific OBPs. In family II, most OBPs are from L. migratoria and no A. pisum OBPs distribute in this family. LmigOBP12 and LmigOBP15, which are highly expressed in labial palp tissue, have no homologs in other insect species. In family III and IV, most OBPs are from hemimetabolous insect species and 13 of 16 A. pisum OBPs distribute in these two families.

#### Time-Course Gene Expression Profiles During Locust Phase Change

To determine the relationships between OBPs and phase change in the locusts, we investigated the time-course expression profiles of seven differentially expressed OBPs in antenna tissue (**Figure 4**). LmigOBP2 or LmigOBP4 displayed a reverse expression pattern after IG (isolation of gregarious locusts) or CS (crowding of solitarious locusts). The expression level of


cysteine residues are indicated by capital bold letters (C1–C6 and C30–C5<sup>0</sup>

).

fphys-09-00984 July 18, 2018 Time: 16:18 # 5


TABLE 2 | Genomic loci of OBP genes in the L. migratoria genome.

fphys-09-00984 July 18, 2018 Time: 16:18 # 6

LmigOBP2 did not change until 16 h during the IG or CS process (ANOVA, F3,<sup>12</sup> = 6.726, 53.345; p = 0.007, 0.000, respectively). The expression of LmigOBP4 decreased significantly after IG for 16 h, increased rapidly from 0 to 4 h, and then stayed at a stable level during the CS process (ANOVA, F3,<sup>12</sup> = 7.092, 21.549; p = 0.005, 0.000, respectively). Although there are small changes of LmigOBP5 and LmigOBP9 expression because of smaller variation in those trials during CS time-course, the expression of LmigOBP1, LmigOBP5, LmigOBP9, LmigOBP14, and LmigOBP16 might be considered to have no changes during IG and CS timecourses because most of the expression changes of OBPs were not significant (IG, ANOVA, F3,<sup>12</sup> = 2.585, 1.590, 0.719, 0.225, 1.215; p = 0.102, 0.243, 0.561, 0.877, 0.350, respectively. CS, ANOVA, F3,<sup>12</sup> = 2.680, 6.083, 7.414, 3.017, 1.463; p = 0.094, 0.009, 0.005, 0.053, 0.278, respectively). So, we categorized these five OBPs into one pattern that no response to IG and CS treatments. Therefore, LmigOBP2 and LmigOBP4 would be related to the phase changes of the locusts because their expression responds to the time-course treatments (IG and CS).

## Effects of LmigOBP2 and LmigOBP4 RNAi on Phase-Related Behavior

To investigate the potential functional significance of LmigOBP2 and LmigOBP4, RNAi and behavioral assays were performed to identify their functions in vivo. We injected dsRNAs to knock down their expression levels in gregarious and solitarious nymphs, respectively. In gregarious locusts, compared with the double-stranded GFP-injected (dsGFP) control, the expressions of both genes, LmigOBP2 and LmigOBP4, decreased significantly after the injections of double-stranded LmigOBP2 or LmigOBP4 (dsLmigOBP2 or dsLmigOBP4) (Student's t-test, t = 9.043, 8.295; p = 0.001, 0.014, respectively; **Figure 5A**).

FIGURE 4 | Time-course expression profiles of OBPs showing differential expressions in antennae between gregarious and solitarious fourth-instar nymphs. IG, isolation of gregarious nymphs; CS, crowding of solitarious nymphs. Shading in yellow, differentially expressed in both IG and CS processes; shading in gray, no different expression in either the IG or CS process. One-way analysis of variance (ANOVA) followed by Tukey's test for multiple comparisons was used. Means labeled with the same letter (capital letters for IG process and lower-case letters for CS process) within each treatment are not significantly different. Differences were considered significant at p < 0.05. Values are reported as means ± SE.

Because of the low identity (18.78%) between dsLmigOBP2 and dsLmigOBP4 fragments, it is quite low for the possibility that LmigOBP2 is knocked down by dsLmigOBP4 injection, and LmigOBP4 is knocked down by dsLmigOPB2 injection (**Supplementary Figure S1**). A behavioral assay indicated that the behavioral traits were significantly altered after LmigOBP4 gene knockdown (Mann–Whitney U = 471, p = 0.023) and median Pgreg changed from 0.995 to 0.785 (**Figure 5B**).

Correspondingly, the attraction index, total distance moved and total duration of movement in dsLmigOBP4-injected locusts were significantly reduced to 28.7, 60.3, and 70.9% of the dsGFP-injected locusts (Mann–Whitney U = 486, 461, 492.5; p = 0.035, 0.018, 0.042, respectively; **Figure 5C**). However, LmigOBP2 knockdown didn't change the behavioral phase state (Mann–Whitney U = 564, p = 0.149; **Figure 5B**) and median ˆPgreg changed a little from 0.995 to 0.980 (**Figure 5B**). The attraction index, total distance moved and total duration of movement in dsLmigOBP2-injected locusts were not altered at all (Mann–Whitney U = 644.5, 589, 541.5; p = 0.556, 0.239, 0.092, respectively; **Figure 5C**). In solitarious locusts, compared with the double-stranded GFP-injected (dsGFP) control, the expressions of both genes, LmigOBP2 and LmigOBP4, decreased significantly after the injections dsLmigOBP2 or dsLmigOBP4 (Student's t-test, t = 14.548, 6.837; p = 0.000, 0.021, respectively; **Supplementary Figure S2A**). The behavioral phase state was not affected after LmigOBP2 or LmigOBP4 knockdown (Mann–Whitney U = 305, 312, p = 0.706, 0.992, respectively; **Supplementary Figure S2B**). The attraction index, total distance moved, and total duration of movement were not changed at all after LmigOBP2 knockdown (Mann–Whitney U = 319, 316, 315.5; p = 0.904, 0.865, 0.815, respectively; **Supplementary Figure S2C**).or LmigOBP4 knockdown (Mann– Whitney U = 306.5, 283, 278; p = 0.900, 0.567, 0.318, respectively. **Supplementary Figure S2C**).

## DISCUSSION

This study describes the identification, temporal-spatial expression, and effects on phase-related behavior of OBPs in the migratory locust. Ten OBPs were identified as novel member of OBPs in the locusts. OBPs with close phylogenetic relationships displayed similar tissue-specific expression patterns. Through filtering genes related to the time-course of phase changes in antenna tissue, we suggested that LmigOBP4 might be involved in the regulation of the behavioral transition of in locusts.

In this study, we identified 17 OBPs in the L. migratoria genome. The number of L. migratoria OBPs is more than that of the other three orthopteran species, S. gregaria (14 OBPs), Oedaleus asiaticus (15 OBPs), and Ceracris kiangsu (7 OBPs). Most L.migratoria OBPs have high homology with those of S. gregaria. We did not find the orthologs of SgreOBP2, SgreOBP7, and SgreOBP13 in L. migratoria. Whereas several L. migratoria OBPs including LmigOBP4, LmigOBP8, LmigOBP12, LmigOBP15, and LmigOBP17 have no orthologs in S. gregaria (Jiang et al., 2017). The numbers of OBPs differed markedly among insect species and ranged from 4 (Pediculus humanus) to 81 (Anopheles gambiae) in the genome (Vieira and Rozas, 2011). Compared to species of Diptera, Lepidoptera, and Coleoptera, orthopteran species have less expansion of the OBP family, which is similar to several representative hemipteran and hymenopteran insects (Fan et al., 2011; Vieira and Rozas, 2011; He and He, 2014; Jiang et al., 2017). However, L. migratoria has a large expansion of the OR family (142 ORs) (Wang et al., 2015). Considering the transport of odorant molecules from OBPs to ORs, it is probable that one OBP might transport multiple odors to variant ORs with different binding capabilities (Leal, 2013).

We revealed that most locust OBPs with a close phylogenetic relationship have similar tissue-specific expression profiles (**Figure 2**). In general, genes with similar tissue-specific expression patterns could be regulated simultaneously to perform related functions, especially for members of a gene family (Stevens et al., 2008; Huang et al., 2015; Gu, 2016). This phenomenon also suggests that phylogenetically correlated OBPs probably constitute a functional cluster to separate and discriminate odors in a complex environmental context. However, the knowledge of gene expression patterns per se is insufficient to infer gene function (Yanai et al., 2006). Functional confirmation of these OBPs needs further analysis of their protein distributions on a range of sensilla, protein structures, ligand-binding properties, and behavioral phenotypes (Harada et al., 2008; Yu et al., 2009; Li et al., 2016).

Most OBPs were highly expressed in antennal or labial palp tissues, indicating that locust OBPs might mainly be involved in olfactory or gustatory functions, as are those of other insect species (Fan et al., 2011; Swarup et al., 2011). Increasing evidence indicates that OBPs are also extensively expressed in variant tissues, such as the brain, maxillary galeae, mandibular glands, and legs (Forêt and Maleszka, 2006; Gu et al., 2011; Iovinella et al., 2011; Yoshizawa et al., 2011). Interestingly, LmigOBP8 is highly expressed in brain tissues of both gregarious and solitarious nymphs. In honey bees, the OBP10 gene begins to express in the pupae and increases to the highest level in the brain of newly emerged bees (Forêt and Maleszka, 2006). As carriers of small molecules, OBPs might also transport some ligands for neural development or signal transmission. In the locusts, some ORs and IRs are also detected in the brain tissue (Wang et al., 2015). Whether LmigOBP8 is involved in ligand transport to the ORs and IRs needs further investigation. The significantly different expression of LmigOBP8 between the two phases suggested its potential function in the regulation of phenotypic plasticity in the central nervous system. In addition, in Bombyx mori, the expression of OBPs and CSPs in the female pheromone glands suggested their function in the solubilization and delivery of pheromonal components (Dani et al., 2011). In D. melanogaster, Obp57d and Obp57e were co-expressed in the taste sensilla on the legs to sense host plant toxins (Harada et al., 2008; Yasukawa et al., 2010). The evidences indicated that OBPs could be involved in multiple physiological processes besides olfactory perceptions.

Olfaction plays critical roles in tuning behavior to the rapid adaptation to environmental change in the locust, especially changes of population density (Guo et al., 2011; Wang et al., 2015). Here, we identified that an OBP, LmigOBP4, was involved in the phase-related behavior of locusts. The orthologs of several OBP members of L. migratoria in family III, which LmigOBP4 belongs to, have recently been reported to distribute in the sensilla chaetica of

the antennae in S. gregaria (Jiang et al., 2017, 2018). So, we inferred that LmigOBP4 might have similar sensilla distribution in the antenna. The sensilla chaetica can not only perceive stimulation resulting from contact chemical molecules (Isidoro et al., 1998), but also volatiles (Ma et al., 2018). Several body volatiles or cuticular hydrocarbons, probably acting as pheromones, were involved in the induction of phaserelated behavior (Heifetz et al., 1997; Wei et al., 2017). Therefore, LmigOBP4 might bind with these volatiles and transmit the chemical cues to activate OSNs. Differential expression of LmigOBP4 between two phases might contribute to differential olfactory sensitivity and inspire different behavioral responses to conspecific volatiles. The knockdown of LmigOBP2, which also displayed differential expression during phase changes, did not change the behavioral traits in our arena assay. The possible reason is that LmigOBP2 might contribute to differential sensitivity in response to contact chemical compounds or plant volatiles. Our previous studies indicated that two olfactory genes, LmigCSP3 and LmigTO1, can regulate the attractive/repulsive responses to conspecifics during locust phase changes (Guo et al., 2011). So, these olfactory proteins might play different roles in chemical perception for a rapid adjustment or long-term adaptation during aggregation.

OBPs bridge the interaction between odorants and ORs (Xu et al., 2005). The identification of tissue-specific OBPs expression patterns provides cues for research about their functions in variant tissues of the locusts. The functional confirmation of LmigOBP4 in locust phase-related behaviors will benefit further studies of the interactions between odorants and ORs. These findings provide further insights into olfactory plasticity in related insect species.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

LK, WG, and XW conceived the study. WG, DR, LZ, and JS conducted the experiments. WG, LK, FJ, and XW interpreted the results. WG drafted the preliminary manuscript. LK and XW refined and approved the final manuscript.

## FUNDING

This work was supported by the National Natural Science Foundation of China (Grant Nos. 31572333, 31201748, and 31772531) and the Youth Innovation Promotion Association, CAS (Grant No. 2016080).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00984/full#supplementary-material

FIGURE S1 | Sequence alignment of cDNA fragments that were used for dsRNA synthesis of LmigOBP2 and LmigOBP4.

FIGURE S2 | Effects of RNAi knockdown of LmigOBP2 and LmigOBP4 genes in their expression levels and behavioral phenotypes in solitarious locusts. (A) Relative mRNA expressions of LmigOBP2 and LmigOBP4 in antennal tissue after dsLmigOBP2 or dsLmigOBP4 injection. ∗∗p < 0.01. (B) Effect of dsGFP, dsLmigOBP2 or dsLmigOBP4 injection on the behavioral phase state in fourth-instar nymphs. Arrows indicate median Pgreg values. n = number of individuals. (C) Effects of dsLmigOBP2 or dsLmigOBP4 injection on attraction index, total distance moved and total duration of movement. n.s., not significant.

TABLE S1 | Primers for OBPs sequence cloning, qRT-PCR and RNAi.



Drosophila melanogaster. Chem. Senses 39, 125–132. doi: 10.1093/chemse/ bjt061


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Guo, Ren, Zhao, Jiang, Song, Wang and Kang. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Sensilla Morphology and Complex Expression Pattern of Odorant Binding Proteins in the Vetch Aphid Megoura viciae (Hemiptera: Aphididae)

Daniele Bruno<sup>1</sup>† , Gerarda Grossi<sup>2</sup>† , Rosanna Salvia<sup>2</sup>† , Andrea Scala<sup>2</sup>† , Donatella Farina<sup>2</sup> , Annalisa Grimaldi<sup>1</sup> , Jing-Jiang Zhou<sup>3</sup> , Sabino A. Bufo<sup>2</sup> , Heiko Vogel<sup>4</sup> , Ewald Grosse-Wilde<sup>5</sup> \*, Bill S. Hansson<sup>5</sup> and Patrizia Falabella<sup>2</sup> \*

#### Edited by:

Nicolas Durand, Université Pierre et Marie Curie, France

#### Reviewed by:

Andrew Dacks, West Virginia University, United States Julian Chen, Institute of Plant Protection (CAS), China

#### \*Correspondence:

Ewald Grosse-Wilde grosse-wilde@ice.mpg.de Patrizia Falabella patrizia.falabella@unibas.it †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 19 December 2017 Accepted: 04 June 2018 Published: 25 June 2018

#### Citation:

Bruno D, Grossi G, Salvia R, Scala A, Farina D, Grimaldi A, Zhou J-J, Bufo SA, Vogel H, Grosse-Wilde E, Hansson BS and Falabella P (2018) Sensilla Morphology and Complex Expression Pattern of Odorant Binding Proteins in the Vetch Aphid Megoura viciae (Hemiptera: Aphididae). Front. Physiol. 9:777. doi: 10.3389/fphys.2018.00777 <sup>1</sup> Department of Biotechnology and Life Sciences, University of Insubria, Varese, Italy, <sup>2</sup> Department of Sciences, University of Basilicata, Potenza, Italy, <sup>3</sup> Department of Biological Chemistry and Crop Protection, Rothamsted Research, Harpenden, United Kingdom, <sup>4</sup> Department of Entomology, Max Planck Institute for Chemical Ecology, Jena, Germany, <sup>5</sup> Department of Evolutionary Neuroethology, Max Planck Institute for Chemical Ecology, Jena, Germany

Chemoreception in insects is mediated by several components interacting at different levels and including odorant-binding proteins (OBPs). Although recent studies demonstrate that the function of OBPs cannot be restricted to an exclusively olfactory role, and that OBPs have been found also in organs generally not related to chemoreception, their feature of binding molecules remains undisputed. Studying the vetch aphid Megoura viciae (Buckton), we used a transcriptomic approach to identify ten OBPs in the antennae and we examined the ultrastructural morphology of sensilla and their distribution on the antennae, legs, mouthparts and cauda of wingless and winged adults by scanning electron microscopy (SEM). Three types of sensilla, trichoid, coeloconic and placoid, differently localized and distributed on antennae, mouthparts, legs and cauda, were described. The expression analysis of the ten OBPs was performed by RT-qPCR in the antennae and other body parts of the wingless adults and at different developmental stages and morphs. Five of the ten OBPs (MvicOBP1, MvicOBP3, MvicOBP6, MvicOBP7, and MvicOBP8), whose antibodies were already available, were selected for experiments of whole-mount immunolocalization on antennae, mouthparts, cornicles and cauda of adult aphids. Most of the ten OBPs were more expressed in antennae than in other body parts; MvicOBP1, MvicOBP3, MvicOBP6, MvicOBP7 were also immunolocalized in the sensilla on the antennae, suggesting a possible involvement of these proteins in chemoreception. MvicOBP6, MvicOBP7, MvicOBP8, MvicOBP9 were highly expressed in the heads and three of them (MvicOBP6, MvicOBP7, MvicOBP8) were immunolocalized in the sensilla on the mouthparts, supporting the hypothesis that also mouthparts may be involved in chemoreception. MvicOBP2, MvicOBP3, MvicOBP5, MvicOBP8 were highly expressed in the cornicles-cauda and two of them (MvicOBP3, MvicOBP8) were immunolocalized in cornicles and in cauda, suggesting a possible new function not related to chemoreception. Moreover, the response of M. viciae to different components

of the alarm pheromone was assessed by behavioral assays on wingless adult morph; (−)-α-pinene and (+)-limonene were found to be the components mainly eliciting an alarm response. Taken together, our results represent a road map for subsequent indepth analyses of the OBPs involved in several physiological functions in M. viciae, including chemoreception.

Keywords: vetch aphid, chemoreception, odorant-binding proteins, RT-qPCR, immunolocalization, behavioral assays

#### INTRODUCTION

Chemical perception in insects is known to be mediated by molecules belonging to the classes of olfactory, gustatory and ionotropic receptors and to the classes of soluble olfactory proteins, odorant-binding proteins (OBPs) and chemosensory proteins (CSPs); however, what these proteins do and how they interact is still not completely clear (Fan et al., 2010; Leal, 2013; Pelosi et al., 2017). In particular, OBPs have long been thought to act exclusively as carriers of chemicals that, once solubilized, were transported to the olfactory receptors (Pelosi et al., 2006; Brito et al., 2016). The generally hydrophobic odorants need to reach the specific receptors bound to the plasma membrane of sensory neuron dendrites, overcoming the hydrophilic barrier that is the sensillar lymph (Pelosi, 1996; Jeong et al., 2013). Several studies performed in vivo indicate that OBPs play a key role in chemoreception. RNAi was used to reduce the expression of OBPs in Anopheles gambiae and Culex quinquefasciatus (Biessmann et al., 2010; Pelletier et al., 2010), in Drosophila melanogaster (Swarup et al., 2011) and in Acyrthosiphon pisum (Zhang et al., 2017). Results demonstrated that OBPs play a specific role in olfactory perception, suggesting there is a direct correlation between the expression level of OBPs and the ability of insects to perceive odors. Previous studies found that Drosophila OBP76a (LUSH) mutants, played an essential role in binding and mediating the recognition of the sex pheromone 11-cis-vaccenyl acetate (c-VA) (Xu et al., 2005; Ha and Smith, 2006; Laughlin et al., 2008). These preliminary results should be partially reconsidered in light of more recent research demonstrating that, at sufficiently high concentrations, c-VA is able to activate neuronal stimuli without LUSH (Gomez-Diaz et al., 2013; Li et al., 2014). However, LUSH is still considered a protein that can increase the sensitivity of the c-VA receptor, also protecting pheromone molecules from degradation by ODEs (Gomez-Diaz et al., 2013). Moreover, the capability of LUSH to bind c-VA has been further demonstrated by in vitro experiments (Kruse et al., 2003).

It has been demonstrated that deleting OBP28a in Drosophila melanogaster basiconic sensilla did not reduce the insect's ability to respond to olfactory stimuli (Larter et al., 2016), suggesting that OBP28a is not required for odorant transport. Larter and colleagues hypothesize a novel role for OBP, namely, that it modulates odor perception by mitigating the effect of rapid changes in the level of environmental odors. In their model, odorants are transported from the sensillum pore to the sensory neuron through hydrophobic tunnels called pore tubules (Steinbrecht, 1997). However, since in Drosophila melanogaster only basiconic sensilla contain pore tubules (Shanbhag et al., 2000), the authors do not exclude that OBP28a expressed in other sensilla type may play different roles including the classical function of odorants carrier (Larter et al., 2016).

Alternatively, different studies suggest that a sensible reduction in olfactory function is related to the reduced levels of certain OBPs (Xu et al., 2005; Biessmann et al., 2010; Pelletier et al., 2010; Swarup et al., 2011). Within the processes relying on chemoreception, it has been proposed that OBPs also play a role in removing chemicals, both those bound to the ORs and those located in the sensory lymph, in order to speed up nervous stimulus termination (Vogt and Riddiford, 1981; Ziegelberger, 1995). That the role of OBPs is related to their binding task is apparent from their multifunctional features, which are not confined to chemical perception (Smartt and Erickson, 2009; Sun Y.L. et al., 2012; Ishida et al., 2013; Pelosi et al., 2017). Indeed, OBPs are expressed in organs that are not connected to chemoreception. In some cases, the same OBP is expressed in chemoreceptive and non-chemoreceptive tissues, suggesting that one type may have multiple roles (Calvello et al., 2003; Li et al., 2008; Sirot et al., 2008; Vogel et al., 2010; Dani et al., 2011; Iovinella et al., 2011; Sun Y.L. et al., 2012; De Biasio et al., 2015). For example, since the same OBPs are expressed in antennae and reproductive organs (Sun Y.L. et al., 2012; Ban et al., 2013), or in antennae and in pheromone glands (Jacquin-Joly et al., 2001; Strandh et al., 2009; Gu et al., 2013; Zhang et al., 2013, 2015; Xia et al., 2015), they may both mediate the recognition of and assisting with the release of the same chemical message. In both cases, the role of OBPs is to solubilize hydrophobic pheromones, binding them in a hydrophilic environment where OBPs are present in high concentration (Nagnan-Le Meillour et al., 2000; Jacquin-Joly et al., 2001; Pelosi et al., 2017).

The different functions imputed to OBPs are in any case linked to the ability of these proteins to bind small hydrophobic molecules, signals of different types originating from different sources. However, the expression of soluble olfactory proteins in chemosensory structures (mainly antennae and mouthparts) indicates that they play a role in chemoreception (Pelosi et al., 2017).

Chemoreception is just one of the roles that OBPs play in aphids (Hemiptera: Aphididae), a group of insects that includes major crop pest in the world. Aphids cause damage directly and indirectly, by feeding and transmitting plant viruses (Nault, 1997; Hogenhout et al., 2008; Webster, 2012). Aphids use their olfactory system and semiochemicals, such as plant volatiles and pheromones, for many purposes: to locate their host plants, select a partner, and escape from danger

(van Emden and Harrington, 2007). In aphids, as in other insects, OBPs have the capability to transport semiochemicals across the sensillar lymph toward the ORs located on the sensory neuron membrane (Qiao et al., 2009; Vandermoten et al., 2011; Sun Y.F. et al., 2012). Even if the mode of action of OBPs is not completely understood, the chemical message is known to be transduced into a neuronal impulse that starts at the dendrite of the olfactory sensory neuron (Leal, 2013); next, the signal reaches the antennal lobe in the brain, where it is processed and leads to a behavioral response (Distler and Boeckh, 1996; Fan et al., 2010).

In the present work, we adopted a multidisciplinary approach to study chemoreception in the vetch aphid Megoura viciae (Buckton), which feeds exclusively on members of Leguminosae (Nuessly et al., 2004).

After constructing and analyzing the M. viciae antennal transcriptome, we identified the OBPs expressed in antennae and determined their expression using the reads per kilobase per million mapped reads (RPKM) method. The expression profile of all the identified OBPs at different developmental stages and in different body parts was also analyzed by RT-qPCR. Moreover, whole mount immunolocalization of five identified OBPs was performed using available antibodies. In addition, scanning electron microscopy (SEM) was carried out on antennae, legs, mouthparts and cauda of both wingless and winged adult morphs to scrutinize the morphology of sensilla expressing the analyzed OBPs at the ultrastructural level. Furthermore, we performed a behavioral assay using the different components of M. viciae's alarm pheromone.

Although our study focuses on the typical chemoreceptive organ, the antennae, and investigates how the expression of OBPs supports the putative role in olfactory and gustatory perception, our results suggest that these soluble proteins play other roles in addition to chemoreception.

## MATERIALS AND METHODS

#### Insect Rearing and Sample Collection

Megoura viciae was reared on potted broad bean plants (Vicia faba L.) at 24 ± 1 ◦C, 75% ± 5% RH and 16 h light – 8 h dark photoperiod. Aphid cultures were started with insects originally collected from broad bean plants in southern Italy near Salerno (40◦ 37<sup>0</sup> N; 15◦ 3 <sup>0</sup> E). In order to synchronize aphid nymphal instars, parthenogenetic females were placed on potted broad bean plants; newborn aphids were separated as soon as they appeared, and adults were removed from plants. Newborn aphids were maintained on plants for 6 days and collected at different developmental stages, from first nymphal instar to adults, both wingless (apterous) and winged (alatae) morphs. Samples were frozen using liquid nitrogen and stored at −80◦C until the RNA extraction used for RT-qPCR experiments. Antennae, de-antennaed heads, legs, cornicles, cauda and remaining body parts of wingless adult aphids were dissected under the microscope, fixed and prepared for SEM, immunolocalization experiments, or homogenized in TRI Reagent (Sigma, St. Louis, MO, United States) and stored at −80◦C until the RNA extraction used for RT-qPCR experiments. Wingless adults were used in behavior experiments. Some specimens deriving from the described original strain were sent to the Department of Biological Chemistry and Crop Protection, Rothamsted Research, Harpenden, United Kingdom, where aphids were reared in the same conditions described above (24 ± 1 ◦C, 75% ± 5% RH and 16 h light – 8 h dark photoperiod). Antennae cut from wingless adults were used for RNA extraction and sequencing at the Beijing Genomics Institute (BGI).

## Scanning Electron Microscopy (SEM)

Adult aphids (6 in the wingless morph and 2 in the winged morph) were prepared as described by Sun et al. (2013). Briefly, they were fixed in 70% ethanol for 2 h and cleaned in an ultrasonic bath for 1 min in the same solution. Finally, samples were dehydrated in 100% ethanol for 30 min, airdried, coated in gold by K250 sputter coater (Emitech, Ashford, Kent, United Kingdom) and examined with SEM-FEG XL-30 microscope (Philips, Eindhoven, The Netherlands).

#### Total RNA Extraction and cDNA Synthesis

Total RNA, collected from 800 antennae, 80 de-antennaed heads, 500 legs, 500 cornicles-cauda and 40 remaining body parts of wingless adult aphids and from 30 aphids of each different nymphal instar (I, II, III, IV) and each adult morph, was extracted using TRI Reagent (Sigma, St. Louis, MO, United States), following the manufacturer's instructions. The concentration of total RNA was measured spectrophotometrically at 260 nm, using a NanoDrop ND-1000 instrument (Nanodrop Technologies, Wilmington, DE, United States). The purity of RNA was estimated at absorbance ratios OD260/280 and OD260/230, and the integrity was verified on 0.8% agarose gel electrophoresis. In order to efficiently remove genomic DNA contamination, the samples were treated with 1U of DNase I (Deoxyribonuclease I, Amplification Grade, Invitrogen-Life Technologies, Carlsbad, CA, United States) per microgram of RNA for 15 min at room temperature, following the manufacturer's guidelines. cDNA was synthesized using the SuperScript III First-Strand Synthesis System for RT-qPCR (Invitrogen-Life Technologies), according to the manufacturer's protocol, using 5 µg of total RNA per sample. The cDNA synthesis reaction was diluted with nucleasefree water to a final volume of 100 µl and immediately used for RT-qPCR studies or stored at −20◦C.

## RNA-Seq Data Generation and de novo Transcriptome Assembly

Antennal transcriptome sequencing was performed with poly(A)enriched mRNA fragmented to an average of 150 nucleotides. Sequencing was carried out by the BGI using paired-end sequencing on an Illumina HiSeq2000 sequencer.

After transformation to raw data, low quality (reads with unknown sequences 'N') adaptor sequences were removed; reads with certain lengths of overlap were combined to form longer fragments, called contigs. These contigs were subjected to further processing of sequence clustering to form longer sequences without N. Such sequences were defined as unigenes.

Reads were trimmed of adapters using Cutadapt (Martin, 2011), and of bad quality regions using Sickle (Joshi and Fass, 2011). Subsequently, reads were assembled using Trinity 2.2 with default parameters (Grabherr et al., 2011).

#### Annotation of OBP Coding Transcripts

The base of the annotation was a hand-curated database of OBP proteins which, among others, contained known aphid candidate protein sequences. The assembled sequences were compared with the references dataset using blastx. All sequences that generated a hit were further scrutinized by blastx comparison against the NCBI non-redundant database (nr), removing any sequences with evidence for an identity that differs from OBP. Finally, the remaining candidates were translated and aligned with the references using MAFFT (Katoh and Standley, 2013), removing candidates that did not align well with known OBP protein sequences. During this step, candidates were also scrutinized for the presence of the conserved OBP cysteine-pattern.

#### Quantitative Real Time PCR (RT-qPCR)

RT-qPCR experiments were carried out in a 7500 Fast Real-Time PCR System (Applied Biosystems- Life Technologies, Carlsbad, CA, United States), on cDNA samples prepared from 5 different nymphal instars, including winged and wingless morphs, and from different body parts (antennae, de-antennaed heads, legs, cornicles and cauda and remaining body parts) of wingless adults. Ribosomal protein S9 (RPS9) and ribosomal protein L32 (RPL32), whose use was validated in a previous work (Cristiano et al., 2016), were chosen as reference genes for the normalization of data obtained from aphids of different nymphal instars and aphids' different body parts RT-qPCR, following the guidelines reported in minimum information required for publication of quantitative real-time PCR experiments (MIQE) (Bustin et al., 2009) and minimum information necessary for quantitative realtime PCR experiments (Johnson et al., 2014). Specific primers were designed for each M. viciae OBP gene and for the reference genes, using Primer Express v3.0 software (ABI, Foster City, CA, United States). Primers of about 20 bp, with approximately 50% G/C content, were selected (**Table 1**). PCR amplifications were performed using GoTaq qPCR Master Mix (Promega, Madison, WI, United States). The reactions were carried out in a 20 µl final volume containing 5 µl of diluted first-strand cDNA (60 ng/µl) and 0.3 µmol/L primer final concentration. Cycling conditions for all genes were: 2 min at 95◦C, 40 cycles of 15 s at 95◦C and 1 min at 60◦C. At the end of each run, a melting curve analysis was performed in order to confirm the specificity of PCR products. All amplification reactions were run in triplicate (technical replicates) and included negative controls (no template reactions, replacing cDNA with H2O). All the experiments were performed for a set of 3 biological replicates. In order to evaluate gene expression levels, relative quantification was performed using the equations described by Liu and Saint (2002), based on PCR amplification efficiencies of reference and target genes. Amplification efficiency of each target gene and of RPS9 and RPL32 was determined according to the equation E = 10−1/<sup>S</sup> −1 (Lee et al., 2006), where S is the slope of the standard curve generated from 4 serial 10-fold dilutions of cDNA. TABLE 1 | Primers used for RT-qPCR.


F, forward primer; R, reverse primer; RPS9, RPL32, reference genes.

All data (mean ± SD) were compared by one-way analysis of variance (ANOVA) and Tukey's HSD multiple comparisons test using GraphPad Prism 6.00 software for Windows (GraphPad Software, La Jolla, CA, United States<sup>1</sup> ). Significant differences were expressed in terms of p-value (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

## Whole-Mount Immunolocalization Experiments

For this assay, only the wingless morphs, collected at the first day of the adult stage, were considered. In particular, 6 antennae, 6 mouthparts, 6 cornicles and 6 caudae from wingless specimens were dissected under the microscope and washed twice with PBS, pH 7.4. Given that winged aphids are rare and difficult to recover and maintain in breeding they were not considered for this analysis. After the washing step, samples were fixed in 4% paraformaldehyde in PBS for 2 h and then washed twice with the same buffer. Samples were then incubated for 30 min with PBS containing 2% BSA (to reduce non-specific binding) and 0.1% of the detergent Tween 20 (Sigma) to permeabilize tissues favoring the entrance of antibodies. Samples were then incubated for 1 h at room temperature with antisera raised in rabbit, diluted 1:200. Whole mount immunolocalization experiments were carried out on five among the ten identified OBPs because only five antibodies were already available. We used antisera against OBPs

<sup>1</sup>www.graphpad.com

1, 3, 6, 7, and 8 of A. pisum since they are ortholog genes of M. viciae OBPs (Zhou et al., 2010). Antibodies, kindly provided by Prof. Paolo Pelosi (University of Pisa), were produced against the entire amino acid OBP sequences and they were not affinity purified. Since recombinant OBPs were not available for preadsorption controls against OBP antibodies, we validated their specificity by western blot using protein extract from the whole M. viciae body (**Supplementary Figure S1**). Briefly, we used 20 µg of proteins (each lane), separated by a 12% polyacrylamide gel electrophoresis and transferred on a Whatman nitrocellulose membrane. Anti-OBP antibodies were diluted 1:1000 in trisbuffered saline and 0.1% Tween 20 (TBS-T) with 5% bovine serum albumin (BSA). Goat anti-rabbit antibodies conjugated to horseradish peroxidase, diluted 1:15000 in TBS-T, was used as a secondary antibody after a pre-absorption using an extra lane loaded with protein extracted from aphid whole body. For detection, enhanced chemo luminescence (ECL) was used and signals were measured with ChemidocTM MP System.

These antibodies have been previously used in experiments on the pea aphid A. pisum OBPs (De Biasio et al., 2015) and on the peach aphid Myzus persicae OBPs (Sun et al., 2013), that are orthologs of A. pisum OBPs (Zhou et al., 2010). We confirmed the high similarity level among A. pisum and M. viciae OBPs by amino acid alignment reported in **Supplementary Figure S2**.

Samples were washed with PBS and incubated for 1h in a dark chamber with the secondary goat anti-rabbit tetramethylrhodamine (TRITC)-conjugated antibody diluted 1:200 (Jackson, Immuno Research Laboratories Inc., West Grove, PA, United States) in blocking solution containing 0.1% Tween 20. In all controls, primary polyclonal anti-OBPs antibodies were omitted or substituted with rabbit pre-immune serum (1:200), and sections were treated with blocking solution containing 0.1% Tween 20 (Sigma) and incubated only with the secondary antibody. Coverslips were mounted with City fluor (City fluor Ltd., London, United Kingdom), and immunofluorescence was analyzed using an inverted laser-scanning confocal microscope (TCS SP5, Leica Microsystems, Wetzlar, Germany) equipped with a HCX PL APO lambda blue 63.0 × 1.40 NA OIL UV objective. Images were acquired using the Leica TCS software (emission windows fixed in the 551–626 range) without saturating any pixel. Z-stack sections acquisition was carried out by selecting the optimized acquisition parameters. The displayed bright field and fluorescent images represent Z-stack projections of sections obtained with the open source image software Fiji (average intensity) (Schindelin et al., 2012). Fluorescence and bright field images were combined with Adobe Photoshop (Adobe Systems Incorporated, San Jose, CA, United States).

#### Behavioral Assays

The behavioral response of M. viciae to the components of the alarm pheromone was investigated under the conditions reported in Sun Y.F. et al. (2012) for A. pisum, using a Y-tube. Briefly, an airflow of 0.5 L/min was introduced into each arm of the glass Y-tube olfactometer through a glass stimulus chamber (odor source adapter) attached to each of the two arms. In each test, 1 µl of hexane solution of each chemical compound, concentration 0.5%, was placed in the glass stimulus chamber of the "treatment" arm. As a control, 1 µl of hexane was placed in the glass stimulus chamber of the "control" arm of the olfactometer. Groups of twenty wingless adult aphids were introduced at the bottom of the Y-shaped copper wire and allowed to walk to either arm at the Y-junction. After 15 min, the number of aphids in the treatment and control sides of the olfactometer were counted. Six replications with each compound were performed. Tested compounds were (E)-β-farnesene (Bedoukian Research, Danbury, CT, United States), (±)-α-pinene, β-pinene, (−) α-pinene, (+)-limonene, hexane (Sigma-Aldrich-Fluka) and a mixture comprising (E)-β-farnesene 14.2%, (−)-α-pinene 11.8% and β-pinene 74% (Francis et al., 2005). The behavioral responses to all the analyzed compounds and mixture were compared by one-way analysis of variance (ANOVA) and Tukey's HSD multiple comparisons test using GraphPad Prism 6.00 software for Windows (GraphPad Software) (∗p < 0.05, ∗∗p < 0.01, ∗∗∗ p < 0.001).

## RESULTS

#### Scanning Electron Microscopy of Sensilla

Scanning electron microscopy observations of M. viciae highlighted differences of legs and antennae both in the morphology and in the distribution of sensilla (**Supplementary Figures S3A–D** and **Figures 1A–N**). In legs, numerous trichoid sensilla, uniform in size, shape and distribution, were visible. In the vetch aphid, sensilla showed a peak with a rounded shape, without pores (**Supplementary Figures S3A,B,D**). SEM images show the insertion of the sensillum basal portion in a cuticular extension on the leg (**Supplementary Figure S3C**). On the antennae of both wingless (**Figures 1A–N**) and winged morph (**Supplementary Figures S4A–H**), different types of sensilla were recognizable, depending on the segment. Type II trichoid sensilla were located on the antenna tip of the 6th segment and along the processus terminalis on the same segment. Type II trichoid sensilla located on the antenna tip appeared as short hairs with a blunt tip showing fissure-like structures and grooves (**Figures 1A,B** and **Supplementary Figures S4A,A'**). Type II trichoid sensilla on the processus terminalis (**Figures 1C,C**' and **Supplementary Figures S4B,B'**), and type I trichoid sensilla, visible from the base of the antenna to the 6th segment, were characterized by a grooved surface and a swollen tip with fissure-like and porous structures (**Figures 1D–F,I,J,L,M** and **Supplementary Figures S4C,F,F',G,H**). Primary rhinaria were clearly observable on the 5th and 6th antennal segments (**Figure 1D** and **Supplementary Figures S4C,F**). In particular, a placoid sensillum was located in the distal end of the 5th segment (**Figures 1D,I** and **Supplementary Figure S4F**), while on the 6th segment 1 large placoid sensillum, 2 smaller ones, 2 type I and 2 type II coeloconic sensilla were distinguishable and surrounded by cuticular fringes (**Figures 1D,E,G,H** and **Supplementary Figures S4C–E**). The placoid sensilla appeared as circular plates showing porous structures on their flat surface (**Figures 1E,I,K** and **Supplementary Figures S4E,F**). On the 3rd antennal segment, secondary rhinaria were constituted by about

30 placoid sensilla in the wingless aphids (**Figure 1L**) and of about 60 placoid sensilla in the winged morph (**Supplementary Figures S4G,H**), both showing a smooth ridge not surrounded by cuticular fringes and small pores on their flat surface (**Figure 1N**). Moreover, we found that in the winged aphids the 3rd segment was longer than in wingless morph (1040 µm instead of 743 µm). Both the wingless and winged vetch aphid presented sensilla associated with mouthparts and caudal region. Since no differences between the two morphs were found, only data of winged morph were shown (**Figures 2A–I**). In the mouthparts, these sensilla showed different morphologies: they had pre-lobed apical extensions (**Figure 2B**) or branched tips (**Figure 2C**). Numerous short sensilla, arranged symmetrically, were evident on the labium end part (**Figure 2A**). SEM observations of the cauda (**Figures 2D–F**) showed the presence of long sensory hair-like structures with small pores (**Figure 2E**) or a fissure-like structure (**Figure 2F**). The entire caudal surface was covered by numerous finger-like projections arranged in groups (**Figure 2D**). Similar structures were also found on the surface of cornicles (**Figures 2G,H**). In addition, the terminal region of cornicles was characterized by the presence of cuticular fingers among which holes were visible (**Figure 2I**).

Scanning electron microscopy observations of M. viciae legs and antennae highlighted differences both in the morphology and in the distribution of sensilla. In legs, numerous trichoid sensilla were visible. On the antennae of both wingless and winged morph type II trichoid sensilla, type I trichoid sensilla, primary rhinaria (5th and 6th segments) and secondary rhinaria (3rd segment) were found. Moreover, the vetch aphid presented sensilla associated with mouthparts and caudal region.

#### Identification of OBP Candidates

First, putative OBP coding sequences needed to be identified. To this end, RNA sequencing of M. viciae antennae was performed. Sequencing data were assembled using the Trinity assembler, resulting in 43,251 predicted transcripts from 36,239 'genes'. The N50 of the assembled transcripts was 2,063 bp, with a corresponding median contig length of 571 bp, average of 1,115 bp and 48,243,578 total nucleotides in the assembly. The assembled data were used in the identification and annotation of ten candidate OBP genes, named MvicOBP1, MvicOBP2, MvicOBP3, MvicOBP4, MvicOBP5, MvicOBP6, MvicOBP7, MvicOBP8, MvicOBP9, and MvicOBP10. The nucleotide sequences were deposited in GenBank under the accession numbers listed in **Table 2**. OBPs expression level in antennae was estimated as reads per kilobase per million mapped reads (RPKM).

Among the ten identified candidate OBP genes, MvicOBP1, MvicOBP3, MvicOBP6, MvicOBP7 and MvicOBP8 were selected for immunolocalization analysis because antibodies were already available. Antibodies against A. pisum OBPs were used because of the high sequence similarity among the selected M. viciae OBPs and the same A. pisum OBPs (**Supplementary Figure S2**). The alignment of the 10 identified antennal M. viciae OBPs is shown in **Supplementary Figure S5**.

RNA sequencing and assembly of M. viciae antennae allowed the identification and the annotation of ten candidate OBP genes. MvicOBP1, MvicOBP3, MvicOBP6, MvicOBP7, and MvicOBP8 were selected for immunolocalization analysis because specific antibodies were already available.

#### OBP Expression Patterns in Different Body Parts and Nymphal Instars of M. viciae

In order to evaluate the expression level in different body parts of all the ten identified M. viciae OBPs, RT-qPCR experiments were carried out using gene-specific primers and using RPS9 and RPL32 as reference genes. We validated the use of these reference genes in RT-qPCR experiments on different developmental stages of M. viciae, in a previous work (Cristiano et al., 2016) and we repeated the validation step on the analyzed different body parts observing that the expression levels of RPS9 and RPL32 remained the same (**Supplementary Figure S6**). **Supplementary Figure S7** shows the OBPs relative expression calibrated on RPS9 and RPL32, respectively. RT-qPCR results showed that MvicOBP1 and MvicOBP10 transcripts were significantly more expressed in M. viciae antennae than in the other body parts (∗∗∗p < 0.001). Transcripts coding for MvicOBP2 were more expressed in antennae, cauda and bodies than in heads and legs (∗∗p < 0.01), while transcripts for MvicOBP3 were significantly more expressed in antennae (∗p < 0.05) and in cauda (∗∗p < 0.01). For MvicOBP4 the statistically highest transcript levels were observed in antennae and bodies (∗∗p < 0.01), while the expression levels of MvicOBP5 were statistically the same in antennae, cauda, bodies and legs (∗p < 0.05). For MvicOBP6 and MvicOBP7, the statistically highest transcript expression levels were observed in antennae (∗∗p < 0.01) and in heads (∗∗p < 0.01 for MvicOBP6 and <sup>∗</sup>p < 0.05 for MvicOBP7). Moreover, we found that the gene encoding for MvicOBP8 was statistically mainly expressed in the cauda and in heads (∗∗p < 0.01), while MvicOBP9 transcripts were more expressed in antennae (∗∗p < 0.01) and heads (∗p < 0.05) (**Figure 3**).

RT-qPCR experiments were confirmed by whole-mount immunolocalization experiments carried out on five OBPs for which antibodies were available (**Figure 4**). In particular, MvicOBP1, MvicOBP3, MvicOBP6, and MvicOBP7 were immunolocalized in type II trichoid sensilla (**Figures 4A–I**) and in the primary rhinaria located on the 5th and 6th segments of antenna (**Figures 4K–S**). MvicOBP1 was expressed mainly in the lymph of type I trichoid sensilla located on the 6th segment (**Figures 4F–K**). Moreover, MvicOBP1 was expressed on placoid sensilla located on the 3rd, 5th, and 6th antennal segments (**Figures 4K,P,U**). MvicOBP3 was expressed in the lymph of type II trichoid sensilla located on the distal region of the antenna (**Figures 4B,G**) and in the large placoid sensilla on the 6th segment (**Figures 4L**). Moreover, MvicOBP3 was expressed in placoid sensilla on the 5th and 3rd segments (**Figures 4Q,V**). In contrast, the small placoid sensilla and the coeloconic sensilla on the 6th segment were not labeled by the antiserum against MvicOBP3 (**Figure 4L**). MvicOBP6 was immunolocalized in the lymph of all sensilla located on the 3rd, 5th, and 6th antennal segments, except in type I trichoid sensilla, and in the 6th segment coeloconic sensilla

(**Figures 4C,H,M,R,W**). Finally, placoid and trichoid sensilla on the 3rd and 5th segments and the lymph of type II trichoid sensilla, placoid and coeloconic sensilla on the 6th segment were

(C), 5 µm; bars in (C',J,N), 500 nm; bars in (D,L), 50 µm.

labeled specifically by the antibody against MvicOBP7, while type I trichoid sensilla on the 6th segment were not stained by this antibody (**Figures 4D,I,N,S,X**). In none of sensilla described

(arrowhead) and trichoid sensilla (arrow). (F,J,M) Details of type I trichoid sensilla showing a groove surface and porous structures on the tip. (N) Detail of a placoid sensillum with a smooth surface not surrounded by cuticular fringes and small pores on the flat surface. Bars in (A,E,I,M), 10 µm; bars in (B,F–H,K), 1 µm; bar in

above, we found the expression of MvicOBP8 (**Supplementary Figure S8A**). The expression profile of OBPs in the mouthparts (**Figure 4A'**) and in the terminal body part (**Figure 4B**') is shown in **Figures 4C'–E',G'–L'**. In the mouthparts, MvicOBP6, MvicOBP7 and MvicOBP8 were expressed in the inner lymph of hair-like sensilla (**Figures 4C'–E'**). In contrast, no signal was detected for MvicOBP1 and MvicOBP3 (**Supplementary Figures S8C,F**). MvicOBP3 and MvicOBP8 were detected in



the hair-and finger-like structures of the terminal region of the body and in the cornicles (**Figures 4G'–L'**), while in both these regions no signals were found for MvicOBP1, MvicOBP6, MvicOBP7 (**Supplementary Figures S8D,E,G–J**). No signal was detected in control experiments in which the primary antibodies were substituted with the rabbit pre-immune serum (**Figures 4E,J,O,T,Y,F',M',N'**) or omitted (**Supplementary Figure S8B**).

**Table 3** summarizes the localization of the five analyzed MvicOBPs in different sensilla types in the wingless morph.

RT-qPCR was also used to investigate on the OBPs expression levels in different nymphal instars. Results showed that MvicOBP1 transcripts were significantly more expressed in the IV nymphal instar (∗∗∗p < 0.001), in the winged adults ( ∗∗p < 0.01) and both in the wingless adults and III nymphal instar (∗p < 0.05). MvicOBP2 transcripts were significantly more expressed in the winged morph (∗∗p < 0.01). Transcripts encoding for MvicOBP3 showed high expression levels in the IV nymphal instar and in the wingless adults (∗∗p < 0.01), which agrees with the lower levels of expression observed in the early nymphal instars (∗p < 0.05) and in the winged adults. MvicOBP4 transcripts were more expressed in the II and IV nymphal instar ( ∗∗p < 0.01), while expression of MvicOBP5 was statistically higher only in the IV nymphal instar (∗∗p < 0.01). MvicOBP6 transcripts were found to be more expressed in the early nymphal instars (I, II, III) (∗p > 0.05). Equally, the levels of transcription of the gene encoding for MvicOBP8 were statistically higher in the first two pre-productive stages (I and II) and in the winged adult morph (∗p > 0.05). The expression of the gene encoding for MvicOBP7 was higher both in the II and IV nymphal instar and in the wingless adult stage (∗p > 0.05), but lower in the other immature stages (I, III) and in winged. Equally, transcripts encoding for MvicOBP9 were more expressed in the IV instar ( ∗∗p < 0.01) and in the II and wingless morph (∗p < 0.05). The expression of the gene encoding for MvicOBP10 was higher both in the IV nymphal instar (∗p < 0.05) and in winged adult ( ∗∗p < 0.01) (**Figure 5**).

All the ten identified MvicOBPs were analyzed by RT-qPCR in different body parts and in all the developmental stages. MvicOBP1, MvicOBP3, MvicOBP6, MvicOBP7 and MvicOBP8 were selected for further analysis of immunolocalization showing a complex immunolocalization pattern in all the analyzed body parts (antennae, mouthparts, cornicles and cauda).

#### Behavioral Experiments

Behavioral experiments on M. viciae wingless adults were performed with the main compounds identified in a cornicle droplet ((E)-β-farnesene, β-pinene, (−)-α -pinene and (+) limonene). For the experiments, a Y-tube olfactometer was used, and aphids that did not choose either of the two arms of the olfactometer (chemical or solvent) were not included in the analysis. The repellency (R) of each compound was calculated by the formula R = (C−T)/(C+T), where T indicates the number of aphids in the arm with the compound to be tested, and C indicates the number of aphids in the control arm. A value of R = 1 indicates that all the insects that have chosen were found in the control arm, while R = 0 indicates that as the aphids were distributed equally between the two arms, the tested substance clearly had no effect. Results are shown in **Figure 6**. The aphids were repelled significantly by (−)-α-pinene, (+) limonene and the mixture containing (E)-β-farnesene 14.2%, (−)-α-pinene 11.8%, β-pinene 74% (Francis et al., 2005), with the R-values of 0.40, 0.28 and 0.48, respectively. In contrast, ( ± )-α pinene, β-pinene and (E)-β-farnesene alone were not repellent for M. viciae, with the R-values of 0.07, −0.05 and 0.02, respectively (**Figure 6**).

Behavioral experiments on M. viciae unwinged adults were performed with the main compounds identified in a cornicle droplet. Aphids were repelled significantly by (−)-α-pinene, (+)-limonene and the mixture containing (E)-β-farnesene, (−) α-pinene and β-pinene.

#### DISCUSSION

Odorant-binding proteins are classically defined as olfactory soluble proteins (Vogt et al., 1991; Pelosi, 1994). Since OBPs are expressed in organs devoted to chemoreception, such as antennae and mouthparts, they likely play a role related to chemoreception. The fact that OBPs are expressed in sensilla whose cuticular surface allows the entry of molecules able to stimulate the olfactory and gustatory receptors located on the sensory neurons strengthens this likelihood (Diehl et al., 2003; De Biasio et al., 2015; Pelosi et al., 2017). Considering that OBPs are also expressed in several organs not related to olfactory and gustatory perception, they can conceivably perform different functions (Nomura et al., 1992; Kitabayashi et al., 1998).

In addition, the same OBP can perform different roles when expressed in different organs and tissues which is related to the general ability of OBPs to bind and transport a range of small molecules, not only those deriving from the external environment (Jacquin-Joly et al., 2001; Zhou et al., 2004; Smartt and Erickson, 2009; Strandh et al., 2009; Sun Y.L. et al., 2012; Gu et al., 2013; Ishida et al., 2013; Zhang et al., 2013, 2015; Xia et al., 2015; Pelosi et al., 2017).

Although it is now generally recognized that OBPs are involved in cellular processes other than chemoreception, the important role of OBPs in chemoreception is confirmed. These soluble proteins, by binding small hydrophobic molecules, allow their solubilization in the sensory lymph (carrier role) and at


the same time the protection against degradation performed by odorant degrading enzymes (ODEs) and the increase of sensitivity toward the receptors (Gomez-Diaz et al., 2013; Chertemps et al., 2015). In this work, we focused on the ten OBPs identified as transcripts in the aphid Megoura viciae antennae. Since the sensilla type and morphology provides an indication about the attribution of a hypothetical functional role of the OBPs expressed therein, an integrated and multidisciplinary approach has been adopted, starting from the analysis of the antennal ultrastructure in both wingless and winged adult morphs and on the different types of sensilla, through SEM.

Two types of trichoid sensilla (I and II) have been described in M. viciae adults (wingless and winged) as in other aphid species (Bromley et al., 1980; Sun et al., 2013; De Biasio et al., 2015). Four type II trichoid sensilla, with a blunt tip characterized by the presence of fissure-like structures are located on the aphid antenna distal part on the 6th segment, both in wingless and winged morphs. These fissure-like structures described for the first time on the type II trichoid sensilla at the end of the processus terminalis would appear similar to those found in the pea aphid A. pisum on the long hair tip of the mouthparts (De Biasio et al., 2015). In A. pisum, the inner lymph of fissured hair like sensilla on the mouthparts was immunostained by the antibody against an ApisOBP (ApisOBP8). Similarly, in M. viciae lymph of fissured type II trichoid sensilla on the antenna tip is immunostained by antibodies against MvicOBP3, MvicOBP6, MvicOBP7. The immunolocalization of all these OBPs and the simultaneous presence of fissure-like structures suggest that fissures on these types of sensilla might be involved in the entry of chemical molecules.

Otherwise, in M. viciae, type II trichoid sensilla located along the processus terminalis and type I trichoid sensilla visible along the whole length of the antennae are characterized by the presence of apical and longitudinal grooves similar to those observed in other insect species (Diehl et al., 2003; Palma et al., 2013; Missbach et al., 2014) where these sensilla were described as olfactory sensilla. They are morphologically different from the same category of sensilla observed in the two aphid species, A. pisum (De Biasio et al., 2015) and M. persicae (Sun et al., 2013), where a smooth surface and a rounded tip have been described. It is interesting to observe that type I trichoid sensilla in M. viciae are stained by antibodies against MvicOBP1, MvicOBP3, MvicOBP6, MvicOBP7, which is in contrast to A. pisum and M. persicae in which type I trichoid sensilla were not stained by any anti-OBP antibody, and for which a mechanoreceptive function was hypothesized (Shambaugh et al., 1978; Bromley et al., 1979; Sun et al., 2013; De Biasio et al., 2015). A possible role of M. viciae type I and type II trichoid sensilla in chemical perception could be hypothesized on the basis of immunolocalization signals and on the basis of the observed morphology that at the ultrastructural level highlights the presence of grooves.

Moreover, SEM observations show the presence of a single large placoid sensillum, two smaller placoid sensilla and four coeloconic sensilla (type I and II) located on the 6th segment, and a single large placoid sensillum on the 5th segment of both wingless and winged adults, as already described for A. pisum and for other species of aphids (Shambaugh et al., 1978; Bromley et al., 1979; Sun et al., 2013; De Biasio et al., 2015). Already available data describing the ultrastructure of aphid placoid sensilla show the localization of pore structures on these sensilla surface (Bromley et al., 1979; Sun et al., 2013). Pore like structures have been observed also in Megoura viciae placoid sensilla indicating that they could be typical chemosensilla as demonstrated in other aphids (Wohlers and Tjallingii, 1983; Park and Hardie, 2004). In primary rhinaria (5th and 6th segments) differences between wingless and winged adults concerning shape, distribution and number of placoid sensilla have not been observed. Secondary rhinaria on the 3rd antennal segment in M. viciae are constituted by placoid sensilla too, similar in the general morphology to those found in the 5th and the 6th segments, suggesting a shared chemosensory function. In winged M. viciae morph, about 60

placoid sensilla on the 3rd segment have been counted, whereas about 30 placoid sensilla have been counted in wingless insects on the same segment. In addition, the length of the 3rd segment increases by about 40% in winged adults. These differences suggest a potential involvement of these sensilla in the location of new host plants. Indeed, aphids acquire wings only when they need to change host plant or mate; therefore, these sensilla could be involved in the detection of plant volatiles (Pickett et al., 1992; Sun et al., 2013).

MvicOBP1, MvicOBP3, MvicOBP6, MvicOBP7 have been immunolocalized in the lymph of placoid sensilla on the 3rd and 5th segments and in large placoid sensilla on the 6th aphid antennal segment. RT-qPCR data confirm the immunolocalization signals of MvicOBP1 showing that the relative expression of this OBP is significantly higher in the antennae. The immunolocalization pattern of MvicOBP6 follows what had been already described in A. pisum in which OBP6 was immunolocalized in placoid sensilla (large and small) on the 6th segment, in placoid sensilla on the 5th segment and in secondary rhinaria (De Biasio et al., 2015). RT-qPCR data confirm the immunolocalization signals, showing that the relative expression of MvicOBP6 is significantly higher in the antennae. EAG experiments performed on different aphid species demonstrate that primary rhinaria (both proximal and distal) are able to perceive a range of plant volatiles. More specifically, the distal primary rhinaria (DPR) are significantly more responsive to tested alcohols than aldehydes in comparison to the proximal primary rhinaria (PPR) and vice versa, indicating a difference in the perception of plant volatiles between the two primary rhinaria (Pickett et al., 1992; van Giessen et al., 1994). Behavioral and electrophysiological studies demonstrated that secondary rhinaria in M. viciae and in other aphids are responsive to sex

pheromone (Pettersson, 1971; Marsh, 1975; Dawson et al., 1987, 1988; Campbell et al., 1990). The immunolocalization signals of MvicOBP1 and MvicOBP6 both in primary and secondary rhinaria and the significantly high relative expression level of these OBPs in the antennae suggest a possible involvement of MvicOBP1 and MvicOBP6 in the perception of host plant chemical volatiles and sex pheromones.

RT-qPCR data also confirm the immunolocalization signals of MvicOBP3 and MvicOBP7 showing that the relative expression of these OBPs is higher in the antennae. A. pisum and M. persicae OBP3 and OBP7, orthologs of M. viciae (Zhou et al., 2010), have high binding affinity to the (E)-ß-farnesene (EBF) which is the only component of the alarm pheromone in these two aphid species. The alarm pheromone triggers physiological and behavioral responses in the aphid colony, to stimulate conspecifics to leave the host plant immediately (Sun Y.F. et al., 2012). In M. persicae and A. pisum, OBP7 was immunolocalized in the primary rhinaria of the 5th and the 6th segments (PPR and DPR), but only in M. persicae OBP7 was also localized in the secondary rhinaria of the 3rd segment. M. persicae OBP3, on the other hand, was immunolocalized in the PPR and only low signals were detected in the other placoid sensilla (Sun et al., 2013) whereas ApisOBP3 was exclusively expressed in the DPR (De Biasio et al., 2015). It had been demonstrated that in A. pisum the perception of EBF involves only primary rhinaria and more specifically DPR, totally excluding secondary rhinaria. Similarly, in the vetch aphid Megoura viciae, EBF is exclusively perceived by DPR (Wohlers and Tjallingii, 1983). Nevertheless, MvicOBP3 and MvicOBP7 have been immunolocalized both in primary (distal and proximal) and secondary rhinaria, unlike ApisOBP3 and ApisOBP7. This may seem surprising but it is conceivable that the involvement of at least MvicOBP3 in the perception of the other components of the alarm pheromone, as previously demonstrated (Northey et al., 2016), may take place in sensilla different from primary rhinaria. Indeed, it has been demonstrated that different OBPs can bind the same molecules in a single organism (Sun Y.F. et al., 2012). Likewise, orthologous OBPs can bind the same molecules in different organisms (Sun Y.F. et al., 2012; Northey et al., 2016) but also different molecules in different organisms (Northey et al., 2016).

Immunolocalization experiments localize MvicOBP3 also in cornicles and cauda, which is confirmed at the mRNA level by RT-qPCR results. This finding does not represent an absolute novelty, since OBP3 expression in A. pisum, evaluated by RTqPCR and immunolocalization, was also observed in cornicles and cauda (De Biasio et al., 2015). The authors hypothesized that ApisOBP3 could be involved in the transport of the alarm pheromone EBF to the environment. Indeed, aphid cornicles are involved in the release of liquid substances in response to dangerous situations such as the presence of predators or parasitoids (Capinera, 2008). The fluid is composed of the alarm pheromone and of other lipid compounds, such as triglycerides, with sticky properties able to trap natural enemies (Strong, 1967; Callow et al., 1973; Greenway and Griffiths, 1973; Butler and O'Neil, 2006; van Emden and Harrington, 2007; De Biasio et al., 2015). Since it has been demonstrated that MvicOBP3 binds EBF and other components of the alarm pheromone mixture (Northey et al., 2016), it is reasonable to suppose that MvicOBP3, expressed in the cornicles, on which hole-like structures are evident, could be involved in the transport of the alarm pheromone mixture to the environment, suggesting also in this species that OBPs could perform roles other than chemoreception.

The alarm pheromone covers an important physiological role in aphids and its use has been proposed in the development of potential strategies for aphid population control (Sun et al., 2011). The identification of OBPs able to bind this pheromone with high affinity is therefore particularly relevant. Although in most aphid species, including A. pisum, the major component of alarm pheromone is the EBF, in M. viciae the alarm pheromone is composed by a mixture of different compounds, including EBF (Bowers et al., 1972; Edwards et al., 1973; Pickett and Griffiths, 1980; Francis et al., 2005). It was demonstrated that ApisOBP3, ApisOBP7 and ortholog proteins have high binding affinity for EBF (Sun Y.F. et al., 2012; Zhang et al., 2017). MvicOBP3 binds EBF with high affinity but it was not able to bind the other components of the alarm pheromone ((−)-α-pinene, β-pinene, (+)-limonene) with the same affinity (Northey et al., 2016). The evaluation of the contribution of each component and the mixture to aphids repulsion behavior is required to address the identification of MvicOBPs binding these components. As expected, the mix of (E)-β–farnesene, (−)-α-pinene, β-pinene and (+)-limonene is significantly more repellent in comparison to the effect of the single components. Surprisingly, (E)-β–farnesene alone, as well as β-pinene alone and the racemic mixture ( ± )-α-pinene, is not active against M. viciae. The most active single components are (−)-α-pinene and (+)-limonene. The behavioral assay represents the basis to address the identification and functional characterization of MvicOBPs directly involved in mediating M. viciae dispersion behavior.

MvicOBP8 is expressed in cornicles and in cauda long sensilla, where pores and fissure like structures have been observed, as well as in finger–like extensions that cover the entire cauda surface,

different to what has been described for A. pisum (De Biasio et al., 2015). RT-qPCR data confirm the immunolocalization of MvicOBP8, showing that this OBP is significantly expressed in cornicles and cauda. It is interesting to note that, similarly to A. pisum OBPs, also M. viciae OBPs, such as MvicOBP8 in this case, are expressed in organs apparently not related to chemoreception, such as the finger–like extensions on the cauda, suggesting a possible new function that needs to be further investigated.

In insects in general and in aphids in particular, other organs besides the antennae are related to chemoreception. SEM revealed that both the wingless and the winged vetch aphid present sensilla associated with mouthparts. Immunolocalization experiments performed on the mouthparts show that the lymph of these sensilla are stained with MvicOBP6, MvicOBP7 and MvicOBP8 antibodies. RT-qPCR data confirm the immunolocalization signals of these OBPs, showing also that the relative expression levels are significantly higher in heads. In accordance with what had been already observed in A. pisum, whose OBP8 was immunolocalized in the sensilla on the mouthparts (De Biasio et al., 2015), MvicOBP8 is immunolocalized in the long sensilla located on the lateral part of the labium. However, unlike what had been observed in A. pisum, OBP6 and OBP7 in Megoura viciae are found in the long hair sensilla. The observed expression patterns suggest that the three OBPs could cover a task in gustatory perception. Indeed, plant volatiles and non-volatiles (such as alkaloids and terpenoids) are moderately soluble in water and the three OBPs may be involved also in the interaction with hydrophobic non-volatile molecules (Galindo and Smith, 2001; Jeong et al., 2013; Swarup et al., 2014), suggesting a greater complexity in the mechanisms of chemoreception also involving M. viciae mouthparts.

Numerous trichoid sensilla have been found on the whole surface of the leg. These types of sensilla are uniform in size, shape and distribution and are similar to those already described in A. pisum (De Biasio et al., 2015). RT-qPCR shows a very low expression level for all the analyzed OBPs, with the exception of MvicOBP5, and no signal in the immunolocalization experiments.

All the results obtained by RT-qPCR experiments on the OBPs whose antibodies were already available are consistent with the results obtained by immunolocalization. We have thus carried out RT-qPCR experiments also on the other OBPs identified in the transcriptome (MvicOBP2, MvicOBP4, MvicOBP5, MvicOBP9, MvicOBP10), for which immunolocalization experiments have not been possible since no specific antibodies were available. All OBPs show significantly higher relative expression levels in the antennae compared to the other organs tested, allowing to hypothesize a possible role in chemoreception for these OBPs. MvicOBP2 and MvicOBP5 show a similar expression pattern, except for the higher relative expression level of MvicOBP5 in legs. MvicOBP5 is the only OBP among those identified in the transcriptome that is significantly expressed in the legs. Since in this aphid species the sex pheromone is produced and released at numerous plaques localized on the hind tarsi, a potential role for MvicOBP5 in sex pheromone release and/or interaction can be speculated. Different roles were attributed to these organs on hind tarsi and it was suggested that they produce a sex pheromone able to attract male aphids (Flogejl, 1905; Weber, 1935; Smith, 1936; Bodenheimer and Swirski, 1957; Stroyan, 1958; Pettersson, 1971; Marsh, 1972, 1975). MvicOBP9 show a relative expression pattern similar to MvicOBP6 and MvicOBP7. Although it was not possible to evaluate the immunolocalization for this OBP, the similar expression profile suggests an analogous function. Similarly, we hypothesize that MvicOBP10 may be involved in a task analogous to that covered by MvicOBP1 in the light of the very similar expression pattern.

MvicOBP3, MvicOBP5, MvicOBP7 and MvicOBP9 are most highly expressed in IV nymphal instar and wingless morph. The observed higher expression levels of these two OBPs could relate to a higher necessity of these later developmental stages to perceive certain compounds (Roitberg and Meyers, 1978) when compared to lower transcript levels in the early stages. MvicOBP1 displays the highest expression levels in IV nymphal instar and winged adults while MvicOBP2, MvicOBP8 and MvicOBP10 are primarily expressed in the winged morph. Moreover, MvicOBP6 is mostly expressed in the first nymphal instars while MvicOBP4 is expressed in the first nymphal instars and in the more mature instars (including the winged morph). The marked heterogeneity of our M. viciae OBPs expression level analysis at different developmental stages could be explained with the complexity of the molecular mechanisms that drive the behavioral response of the different aphids' nymphal instars to the chemical molecules. Indeed, different plant chemicals are able to trigger different behavioral responses that are also dependent on aphid morph and developmental stage; moreover, different morphs of the same aphid species show different behaviors in response to the same volatiles (Lilley and Hardie, 1996; Quiroz and Niemeyer, 1998; Powell and Hardie, 2001; Webster, 2012). Within the same morph, the response to volatile compounds can vary widely in relation to the stage of development (Glinwood and Pettersson, 2000a,b).

#### CONCLUSION

In this work we have verified which of the identified OBPs were expressed in sensilla that, for their position in typical chemoreceptive organs and for the presence of morphological features such as pores, grooves and fissure-like structures, could potentially cover chemoreceptive functions. Considering the traditional role attributed to OBPs, the gained information would have led us to assign automatically a specific role of odorants carrier toward the olfactory receptors to the identified OBPs. In the light of recent works (e.g., Larter et al., 2016) the OBPs expressed in chemosensilla are certainly involved in chemoreception but their roles can be multiple, although the specific feature of binding proteins remains unaltered (Pelosi et al., 2017). Our data on the ultrastructure of sensilla as well as on OBP expression profiles in different developmental stages and various body parts allow to state that OBPs in Megoura viciae show a very complex expression pattern. The increasing

knowledge about the different tasks performed by OBPs in insects leads us to hypothesize that the described level of complexity of Megoura viciae OBPs pattern can be ascribed to the different functions of these proteins in physiological pathways of the vetch aphid. The knowledge acquired with this work could represent the road map for guiding future studies aimed to the detailed clarification of the role of each M. viciae OBP.

#### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

#### AUTHOR CONTRIBUTIONS

PF designed the experiments, and wrote and critically revised the paper. EG-W, HV, BH, J-JZ, AS, GG, RS, and SB contributed to the data interpretation and critically revised the paper. AG and DB performed the SEM experiments and immunolocalization experiments. GG, DF, and RS performed the samples collection, RT-qPCR, and antibodies validation. GG and AS performed the behavioral assays. J-JZ performed the antennal transcriptome sequencing. GG, HV, and EG-W performed the transcriptome analysis. All authors read and approved the manuscript.

#### FUNDING

This work was supported by the Max Planck Society and by University of Basilicata (RIL funds).

#### ACKNOWLEDGMENTS

DB is a student of the Ph.D. program in Biotechnology, Biosciences and Surgical Technologies, School in Biological and Medical Sciences, University of Insubria. We would like to thank Emily Wheeler for editorial assistance, Silvia Sacchi and Marcella Reguzzoni for their technical assistance for SEM and confocal image analysis and Vincenzo Trotta for the assistance in statistical analysis.

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00777/full#supplementary-material

FIGURE S1 | Western blot performed with antisera against OBPs 1, 3, 6, 7, and 8 of A. pisum on protein extract from the whole M. viciae body (20 µg of proteins per each lane).

FIGURE S2 | Alignment of amino acid sequences of Megoura viciae and Acyrthosiphon pisum OBPs

FIGURE S3 | SEM images showing the distribution of sensilla on M. viciae legs. Trichoid sensilla present a typical hair shape and are covered by a thin cuticle (arrowheads in (A–C)). These sensilla show a peak with a rounded shape, without pores. Bar in (A), 100 µm; bar in (B), 25 µm; bar in (C), 100 µm; bar in (D), 2 µm

FIGURE S4 | SEM images showing the distribution and morphology of different sensilla on winged M. viciae antennae. (A,B) Type II trichoid sensilla located on the terminal part of the antenna (arrowhead in (A)) and on processus terminalis (B) with grooves on tip surface (arrowhead in (A',B')). (C) Global view of primary rhinaria on 6th segment (arrowhead) and type II trichoid sensilla (arrow). (D) Details of small placoid sensilla (SP), and type I (CI) and type II (CII) coeloconic sensilla in the 6th segment surrounded by cuticular fringes (arrowheads). (E) Detail of porous structure on the surface of the large placoid sensillum (arrowheads). (F) Details of placoid sensillum of 5th segment and type I trichoid sensilla (arrow) with grooved surface (arrow in (F')). Porous structures were visible on the flat surface on the placoid sensillum of this segment (arrowhead). (G,H) Placoid sensilla (secondary rhinaria) on the 3rd segment (white arrowhead in (G,H)) and trichoid sensilla type I (arrow in (G,H)). Bars in (A,C,F), 10 µm; bars in (A',B',F'), 500 nm; bar in (B), 2 µm; bars in (D,E), 2 µm; bar in (G), 100 µm; bar in (H), 20 µm.

FIGURE S5 | Alignment of amino acid sequences of candidate OBPs from Megoura viciae.

FIGURE S6 | RPS9 and RPL32 constant expression level in M. viciae body parts.

FIGURE S7 | Relative expression level of M. viciae OBPs in different body parts (A,B) and in different nymphal instars (C,D) calibrated on RPL32 and RPS9, respectively. OBP expression levels were quantified by RT-qPCR. Bars represent the standard deviation of the mean for 3 independent experiments. Significant differences are denoted by asterisks (Tukey's test, (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001)). (A,B) Lg, legs; Cd, cornicles-cauda; Hd, head; Bd, body; An, antennae. Calibrator sample: antennae. (C,D) I, 1st nymphal instar; II, 2nd nymphal instar; III, 3rd nymphal instar; IV, 4th nymphal instar; Ap, apterous adults; Al, alata adults. Calibrator sample: 1st nymphal instar.

FIGURE S8 | Whole-mount immunolocalization experiments showing the absence of signal for MvicOBP8 in antenna (A), MvicOBP1 and MvicOBP3 in the mouthparts (C,F), MvicOBP1, MvicOBP6, MvicOBP7 in the cauda (D,G,I), and in cornicles (E,H,J). (B) Negative control in which the primary antibodies were omitted. Bars in (A,B), 30 µm; bars in (C,F), 10 µm; bars in (D,G,I), 50 µm; bars in (E,H,J), 20 µm.




primary rhinaria of Acyrthosiphon pisum (Harris) to C4–C8 primary alcohols and aldehydes. J. Chem. Ecol. 20, 909–927. doi: 10.1007/BF02059587


tissues of putative chemosensory genes identified by transcriptome analysis of insect pest the purple stem borer Sesamia inferens (Walker). PLoS One 8:e69715. doi: 10.1371/journal.pone.0069715


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Bruno, Grossi, Salvia, Scala, Farina, Grimaldi, Zhou, Bufo, Vogel, Grosse-Wilde, Hansson and Falabella. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Functional Characterization of Odorant Binding Protein 27 (RproOBP27) From Rhodnius prolixus Antennae

Daniele S. Oliveira<sup>1</sup> , Nathália F. Brito<sup>1</sup> , Thiago A. Franco<sup>1</sup> , Monica F. Moreira1,2 , Walter S. Leal<sup>3</sup> and Ana C. A. Melo1,2 \*

<sup>1</sup> Laboratório de Bioquímica e Biologia Molecular de Vetores, Instituto de Química, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, <sup>2</sup> Instituto Nacional de Ciência e Tecnologia em Entomologia Molecular-CNPq, Rio de Janeiro, Brazil, <sup>3</sup> Department of Molecular and Cellular Biology, University of California, Davis, Davis, CA, United States

#### Edited by:

Peng He, Guizhou University, China

#### Reviewed by:

Guan-Heng Zhu, University of Kentucky, United States Ya-Nan Zhang, Huaibei Normal University, China Joe Hull, Agricultural Research Service (USDA), United States George F. Obiero, Max-Planck-Institut für Chemische Ökologie, Germany

> \*Correspondence: Ana C. A. Melo anamelo@iq.ufrj.br

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 28 April 2018 Accepted: 06 August 2018 Published: 23 August 2018

#### Citation:

Oliveira DS, Brito NF, Franco TA, Moreira MF, Leal WS and Melo ACA (2018) Functional Characterization of Odorant Binding Protein 27 (RproOBP27) From Rhodnius prolixus Antennae. Front. Physiol. 9:1175. doi: 10.3389/fphys.2018.01175 Olfactory proteins mediate a wide range of essential behaviors for insect survival. Odorant binding proteins (OBPs) are small soluble olfactory proteins involved in the transport of odor molecules (=odorants) through the sensillum lymph to odorant receptors, which are housed on the dendritic membrane of olfactory sensory neurons also known as olfactory receptor neurons. Thus, a better understanding of the role(s) of OBPs from Rhodnius prolixus, one of the main vectors of Chagas disease, may ultimately lead to new strategies for vector management. Here we aimed at functionally characterize OBPs from R. prolixus. Genes of interest were selected using conventional bioinformatics approaches and subsequent quantification by qPCR. We screened and estimated expression in different tissues of 17 OBPs from R. prolixus adults. These analyses showed that 11 OBPs were expressed in all tissues, whereas six OBP genes were specific to antennae. Two OBP genes, RproOBP6 and RproOBP13, were expressed in both male and female antennae thus suggesting that they might be involved in the recognition of semiochemicals mediating behaviors common to both sexes, such host finding (for a blood meal). Transcripts for RproOBP17 and RproOBP21 were enriched in female antennae and possibly involved in the detection of oviposition attractants or other semiochemicals mediating female-specific behaviors. By contrast, RproOBP26 and RproOBP27 might be involved in the reception of sex pheromones given that their transcripts were highly expressed in male antennae. To test this hypothesis, we silenced RproOBP27 using RNAi and examined the sexual behavior of the phenotype. Indeed, adult males treated with dsOBP27 spent significantly less time close to females as compared to controls. Additionally, docking analysis suggested that RproOBP27 binds to putative sex pheromones. We therefore concluded that RproOBP27 might be a pheromone-binding protein.

Keywords: odorant binding proteins, Rhodnius prolixus, RNAi, sexual behavior, olfaction

## INTRODUCTION

fphys-09-01175 August 22, 2018 Time: 9:41 # 2

Chemical signals are essential to promote specific behaviors in different species (Gaillard et al., 2004). Insects, in particular, depend on the correct identification of volatile compounds (semiochemicals) for survival and reproduction (Cruz-Lopez et al., 2001; Syed and Leal, 2009; Pitts et al., 2014). Rhodnius prolixus is one of the main vectors of the protozoan Trypanosoma cruzi, the etiological agent of Chagas disease. According to estimates based on 2010 WHO data, 5,742,167 people in 21 Latin American countries are infected (WHO, 2015). New cases due to vector transmission were estimated to 29,925/year (WHO, 2015). Several proteins participate in insect chemosensation, including odorant binding proteins (OBPs), which transports odor molecules through the sensillum lymph to odorant receptors (ORs) (Fan et al., 2011; Leal, 2012; Brito et al., 2016; Pelosi et al., 2018); ORs located in the membrane of olfactory sensorial neurons (Benton, 2006), which recognize volatile odorant molecules (de Bruyne and Baker, 2008); and ionotropic receptors (IRs), which detect diverse chemical ligands from the environment (Benton et al., 2009). OBPs represent the first contact between semiochemicals from the environment and the olfactory sensory system since they are responsible for transporting hydrophobic ligands to their specific ORs (Wojtasek and Leal, 1999; Fan et al., 2011). OBPs are small soluble proteins secreted by accessory cells into the antenna sensillar lymph surrounding the olfactory sensory neurons (Brito et al., 2016; Pelosi et al., 2018). Initially, OBPs were identified and characterized at molecular level in Drosophila melanogaster (Brito et al., 2016). After that, other studies reported that OBPs were identified in different insect species, including the disease vectors Anopheles gambiae (Vogt, 2002; Mastrobuoni et al., 2013), Aedes aegypti (Zhou et al., 2008), Culex quinquefasciatus (Pelletier and Leal, 2009), and Glossina morsitans morsitans (Liu et al., 2010). In hemipterans, the first characterized OBP was Lygus antennal protein (LAP) from the phytophagous insect Lygus lineolaris (Dickens et al., 1998). LAP expression was shown to be adult-specific, initiating development in antennae during the transitional period that precedes adult molt (Vogt et al., 1999). Subsequently, it was reported that in the alfalfa plant bug Adelphocoris lineolatus, some OBP genes exhibited high differential expression in male and female antennae (Gu et al., 2011a). More recently, the genome of the hematophagous hemipteran R. prolixus was released and has been predicted to encode 27 putative OBP genes (Mesquita et al., 2015). However, only 17 OBPs were actually identified in the antenna proteome (Oliveira et al., 2017), suggesting that these proteins could be associated with odor detection. Hemipteran insects have many intricate behaviors such as male aggregation (Vitta et al., 2009; Pontes and Lorenzo, 2012), oviposition aggregation (Rolandi and Schilman, 2017), food ingestion (Diaz-Albiter et al., 2016; Franco et al., 2016), and avoidance behavior (Zermoglio et al., 2015). Despite the importance of R. prolixus as a vector of Chagas disease, the role(s) of OBPs in odor recognition has not yet been investigated, even though there is strong evidence that these insects use chemical signals to mediate sexual communication. It is already known that males can be oriented toward air currents carrying volatiles produced by female metasternal glands (MGs) (Pontes et al., 2008, 2014). Recently, several studies have used the RNA interference (RNAi) technique to identify OBP functions in insects (Biessmann et al., 2010; Pelletier et al., 2010; Zhang et al., 2016). Moreover, it might be possible to link behavior to OBP(s) by gene silencing (He et al., 2011; Swarup et al., 2011; Deng et al., 2013; Li et al., 2016; Shorter et al., 2016; Lin et al., 2017). In fact, RNAi based studies have shown that OBPs are involved in the detection of oviposition attractants (Biessmann et al., 2010; Pelletier et al., 2010), plant volatiles (He et al., 2011; Li et al., 2016; Zhang et al., 2016), host molecules (Deng et al., 2013), in the survival of insects (He et al., 2011; Lin et al., 2017), and regulates mating behavior (Shorter et al., 2016). Therefore, the purpose of this study was to investigate the role of the 17 OBPs previously identified in antenna proteome (Oliveira et al., 2017), in R. prolixus chemical communication. Results revealed that 11 OBPs were expressed in all tissues, whereas six OBPs were shown to be antennae-specific. RproOBP6 and RproOBP13 were expressed in both male and female antennae. RproOBP17 and RproOBP21 were enriched in female antennae. In contrast, RproOBP26 and RproOBP27 were significantly expressed in male antennae, which suggests these proteins could play a role in male specific behaviors. Interestingly, RproOBP26 was also reported overexpressed in the insect gut (Ribeiro et al., 2014), suggesting that RproOBP26 might be involve in multiple roles. The potential role of RproOBP27 in the detection of odorants was further investigated by RNAi because this protein is male antennae-specific and thus a putative pheromonebinding protein. Additionally, docking analysis suggested that RproOBP27 favorably binds the most abundant chemicals (putative sex pheromones) identified in female MGs (Pontes et al., 2008), which indicates this OBP could be involved in the detection of female-derived semiochemicals. In a behavioral assay, males injected with dsOBP27 spent significantly less time close to females when compared to controls, strongly suggesting RproOBP27 plays a role in the reception of female-derived semiochemicals.

#### MATERIALS AND METHODS

#### Insect Rearing

Rhodnius prolixus were taken from a colony at Insect Biochemistry Laboratory/Federal University of Rio de Janeiro/Brazil. Insects were maintained at 28◦C and 80– 90% relative humidity under a photoperiod of 12 h of light/12 h dark. Insects used in this work were unmated males fed on rabbit blood at 3-week intervals. Male R. prolixus injected with dsRNA were kept on cages maintained under the same conditions. In dsRNA experiments, unfed male nymphs (5th instar, N5) were injected with 1 µg of dsRNA (dsOBP27 or dsβ-gal) diluted in 1 µL of RNase-free water into the metathoracic cavity using a 10 µL Hamilton syringe. Nymphs were fed on rabbit blood 7 days after dsRNA treatment.

#### Ethics Statement

fphys-09-01175 August 22, 2018 Time: 9:41 # 3

All animal care and experimental protocols were conducted following the guidelines of the institutional care and use committee (Committee for Evaluation of Animal Use for Research from Federal University of Rio de Janeiro), which are based on the National Institute of Health Guide and Use of Laboratory Animals (ISBNo-309-05377-3). The protocols were approved by the Committee for Evaluation of Animal Use for Research (CAUAP) from the Federal University of Rio de Janeiro, under register number CEAU-UFRJ#1200.001568/2013- 87, 155/13. Technicians dedicated to the animal facility at Federal University of Rio de Janeiro carried out all aspects related to rabbit husbandry under strict guidelines to ensure careful and consistent handling of the animals.

#### Tissue Isolation, RNA Extraction, and cDNA Synthesis

Antennae, proboscis, legs, and heads (without antennae and proboscis) from 30 blood-fed male and 30 blood-fed female were dissected using forceps. Tissues were transferred to polypropylene tube separately, frozen in liquid nitrogen and triturated with plastic pestle. Total RNA was extracted from different tissues using TRIzol (Invitrogen, Carlsbad, CA, United States) according to the manufacturer's instructions. RNA concentrations were determined at 260 nm on a UV-1800 spectrophotometer (Shimadzu, Inc., Kyoto, Japan). RNA integrity was evaluated in 1% agarose gel. RNAs were treated with RNase-free DNAse I (Fermentas International, Inc., Burlington, ON, Canada), 1 µg of RNA was used for cDNA synthesis with High-Capacity cDNA Reverse Transcription kit and random primers (Applied Biosystems, Foster City, CA, United States).

#### Spatial Transcript Quantification

Gene sequences of 17 RproOBPs were downloaded from R. prolixus genome database<sup>1</sup> for primer design using OligoPerfectTM Designer – Thermo Fisher Scientific tool. All primer sequences are listed in **Supplementary Table S1**. PCR studies were performed using GoTaq <sup>R</sup> Green Master Mix kit (Promega, Madison, WI, United States). R. prolixus' ribosomal gene 18S (RproR18S) was used as the reference gene (Majerowicz et al., 2011). PCRs were performed on Veriti <sup>R</sup> Thermal Cycler-96 well thermocycler (Applied Biosystems, Foster City, CA, United States), consisting of 35 cycles for RproOBPs and 25 cycles for RproR18S under the following conditions, 94◦C for 3 min, followed by denaturation steps at 94◦C for 30 s, annealing temperature was set according to each primer pair (**Supplementary Table S1**) for 30 s and the extension step at 72◦C for 1 min and 30 s, finally followed by 72◦C for 10 min. cDNA from antennae, proboscis, legs, and heads (without antennae and proboscis) obtained from adults were used as templates for PCR. PCR products were analyzed on a 1% agarose gel stained with GelRedTM (Biotium, Hayward, CA, United States) in TAE buffer pH 8 (40 mM Tris-acetate, 1 mM EDTA). Gels were digitalized on DNR MiniBIS Pro Bio-Imaging Systems (BioAmerica Inc., Miami, FL, United States). qPCRs were performed on a StepOneTM Real-Time PCR System (Applied Biosystems) thermocycler using Power SYBR <sup>R</sup> Green PCR Master Kit (Applied Biosystems). cDNA from adult antennae, proboscis, legs and heads (without antennae and proboscis) were used as templates for qPCRs. Oligonucleotide concentrations consisted of 400 nM for RproR18S and 600 nM for RproOBPs. Reactions were carried out in three biological replicates and three technical replicates for each sample, in a 48-well optical plate with the following initial cycle, 50◦C for 2 min; 95◦C for 10 min; followed by denaturation steps at 94◦C for 15 s then 60◦C for 15 s and extension at 72◦C for 1 min for 40 cycles; dissociation curves were obtained under standard conditions of the instrument. RproR18S gene was used as reference gene for the normalization of Ct (threshold cycle) values. The relative gene expression of the RproOBPs was determined by 2−11Ct method (Livak and Schmittgen, 2001). Data were presented as mean ± standard error of three independent experiments in biological triplicates.

#### dsRNA Synthesis and Gene Silencing Assays

Fragments of PCR product encoding RproOBP27, size 146 bp, were amplified by PCR using cDNA from blood-fed male adults antennae produced as described above. The following conditions were used for amplification: one cycle for 3 min at 94◦C, following by 35 cycles of 30 s at 94◦C for denaturation, 30 s at 59◦C for annealing and the extension step at 72◦C for 1 min and 30 s, followed by 72◦C for 10 min. The primers used for amplification of templates for dsRNA synthesis are listed at **Supplementary Table S1**. These primers contained a T7 polymerase binding sequence required for dsRNA synthesis. These products were used as the template for the transcription reactions using the enzyme T7 RNA polymerase with MEGAscriptRNAi kit (Ambion, Austin, TX, United States), according to manufacturer's protocol. The β-galactosidase protein (β-gal) gene from Culex quinquefasciatus (Xu et al., 2014) cloned into pGEM-T (Promega) was amplified by PCR using T7 minimal promoter primers under the following conditions: one cycle for 3 min at 94◦C, following by 35 cycles of 30 s at 94◦C for denaturation, 30 s at 56◦C for annealing and the extension step at 72◦C for 1 min and 30 s, followed by 72◦C for 10 min. The PCR product generated was used as the template for β-gal dsRNA synthesis used as a control in the silencing assay. Following in vitro synthesis, all dsRNAs were purified using phenol-chloroform (1:1), quantified using a spectrophotometer at 260 nm and analyzed by 1% agarose gel electrophoresis stained with GelRedTM. RNAi experiments were performed as described by Franco et al., 2016. Briefly, 1 µL of dsRNA (1 µg/µL RNase-free water) was injected into the metathoracic cavity of starved N5 males (N = 20 for each dsRNA treatment), using a 10 µL Hamilton syringe, after 7 days insects were blood fed and monitored during 21 days until ecdysis. The resulting dsRNAtreated adults were fed on rabbit blood. In bioassays, insects from the different groups were tested individually.

Starved N5 males treated with dsRNA as described above (N = 20 for dsOBP27 and N = 20 for dsβ-gal) were kept under

<sup>1</sup>https://www.vectorbase.org/organisms/rhodnius-prolixushttp

controlled temperature and humidity conditions. Mortality was monitored from the 3rd to the 20th day after dsRNA injection. The number of survival N5 in this period was registered. The effect of dsRNA injection on blood feeding was performed as described by Franco et al., 2016. The dsOBP27- and dsβ-gal-treated N5s were weighed 2 h before and 2 h after feeding. The ingested mass (mg) was calculated by the weight difference after and before feeding.

#### Female Recognition Bioassay

The ability of adult males treated with dsRNA to recognize females was accessed using a bioassay adapted from Zermoglio et al. (2015) (**Supplementary Figure S5**). Adult males injected with dsRNA in the N5 stage were used in the bioassay. dsRNAtreated N5 males were blood-fed (N = 20 dsOBP27; N = 20 dsβ-gal) 7 days after injection. Males were then blood-fed 7 days after molt. Bioassays were conducted 1 week after blood meals. A polystyrene tube (falcon tube) with approximately 10 cm long and 2 cm in diameter was used (**Supplementary Figure S5**). This tube was divided into three zones: female zone (FZ), intermediate zone (IZ), and male release zone (MZ). A gate divides the MZ from IZ. A protective mesh was used to separate MZ and IZ from FZ. An adult female was placed in front of the protection mesh attached by a tape on the tube. Then a male was placed in the MZ and the gate was opened after 5 min of acclimation. The time spent by males to move across the tube toward the female was measured using a digital chronometer and estimated in a maximum period of 300 s. When the insect reached the FZ, the chronometer was reset and started again to record the interval of time that male stayed near the female. The bioassay was repeated 3 times for each insect in each group (dsOBP27 and dsβ-gal).

#### Docking Studies

Since 3D structures have not yet been characterized for Rhodnius OBPs, the primary sequence of mature RproOBP27 was used to construct a 3D model for in silico docking studies. Threedimensional modeling was developed using the online protein threading program PHYRE2 (Kelley et al., 2015). Stereochemical quality and accuracy of the predicted model were evaluated using the software PROCHECK (Laskowski et al., 1996) and Verify3D (Eisenberg et al., 1997). The most abundant compounds identified as volatiles emitted by MGs of females and reported as being able to modulate male orientation (Pontes et al., 2008, 2014) were selected for docking studies: 2-methyl-3-buten-2 ol, (2S)-pentanol, (3E)-2-methyl-3-penten-2-ol, and (2R/2S)-4 methyl-3-penten-2-ol. We used thermodynamic principle that ligands tightly bind the active site of a protein when the free binding energy of the process is low (Du et al., 2016). Therefore, such parameter was used to estimate binding affinities of the MGs ligands to RproOBP27. Three-dimensional structures of compounds were obtained from PubChem<sup>2</sup> (Kim et al., 2016). Molecular docking with RproOBP27 and each of the selected ligands was carried out 100 times using Docking Server (Bikadi and Hazai, 2009) and the free bindingenergy scoring function was considered to estimate binding affinity.

## Statistical Analysis

Statistical analysis of qPCRs and bioassays were performed using t-test followed by the Mann–Whitney test (GraphPad PRISM 6.00 software, San Diego, CA, United States). qPCRs analyses were done by using three biological and three technique replicates for each gene. Bioassays were carried out independently in three technique replicates. Bars represent the standard error of three replicate, asterisks indicate statistically significant differences (P < 0.05).

## RESULTS

#### Spatial Expression of OBPs

Previous results showed that 17 OBPs were expressed in the adult antennae (Oliveira et al., 2017), which suggests that at least 17 genes predicted as OBPs in the genome actually encodes antennal functional proteins. In order to investigate which of those 17 OBP genes were antennae-specific, different tissues of adult insects [antennae, proboscis, legs, and heads (without antennae and proboscis)] were screened by PCR. Spatial expression showed that 11 OBPs (RproOBP1, RproOBP7, RproOBP11, RproOBP12, RproOBP14, RproOBP18, RproOBP20, RproOBP22, RproOBP23, RproOBP24, and RproOBP29) were expressed in multiple tissues (**Figure 1**; all original gels appear in **Supplementary Figures**), which strongly suggests that proteins produced by these genes are not specifically related to odorant transport. In contrast, four OBPs (RproOBP6, RproOBP13, RproOBP17, and RproOBP21) were detected specifically in adult antennae, although minor bands for RproOBP13 and RproOBP21 were detected in other tissues (**Figure 2**). Two OBPs were highly expressed in the male antennae, RproOBP26 and RproOBP27 (**Figure 2**), with minor RproOBP27 bands being observed in male proboscis and male and female legs (**Figure 2**; see also the original gels in the **Supplementary Figures S1**–**S4**). To further investigate these qualitative profiles, OBPs that were found to be enriched in the antenna were quantified by qPCR. Proboscis, legs, and heads (without antenna and proboscis) were also analyzed by qPCR. Considering that we did not identify any transcripts in proboscis, heads, and legs, the above described bands in these tissues (detected by conventional PCR) were probably not specific bands for the tested genes (**Figure 2**). Quantitative results confirmed that RproOBP6 and RproOBP13 were expressed exclusively in male and female adult antennae and did not exhibit transcripts in other tissues (**Figures 3A,B**). In addition, RproOBP17 and RproOBP21 were enriched in female antennae (P < 0.05) (**Figures 3C,D**). On the other hand, RproOBP26 and RproOBP27 were shown to have high and specific expression in male antenna (P < 0.05) (**Figures 3E,F**).

#### Role of RproOBP27 on Male Behavior Silencing of RproOBP27

Next, we reduced the expression of RproOBP27 using RNAi and evaluated the behavior of the male phenotype. Transcript levels of RproOBP27 were compared to control dsβ-gal. RproR18S was utilized as a reference gene to calculate relative expression. dsOBP27 injected-group exhibited a significant reduction in

<sup>2</sup>https://pubchem.ncbi.nlm.nih.gov

FIGURE 1 | Expression profile of RproOBP1, RproOBP7, RproOBP11, RproOBP12, RproOBP14, RproOBP18, RproOBP20, RproOBP22, RproOBP23, RproOBP24, and RproOBP29 in different Rhodnius prolixus tissues evaluated by conventional PCR. N, negative control; FA, female antennae; MA, male antennae; FP, female proboscis; MP, male proboscis; FH, female head; MH, male head; FL, female legs; ML, male legs. RproR18S was used as an endogenous control. The amplicon size (bp) is indicated on the right. Heads were used without antennae and proboscis.

RproOBP27 expression (8x) when compared to control groups (**Figure 4A**). In fact, the dsβ-gal fold change mean was 1.06, while dsOBP27 was 0.13, which indicates an 88% expression decrease.

#### Effects of Reduction in the Expression of RproOBP27 on Male Physiology

Insect survival was monitored from 3 to 20 days before molting. The survival index ranged from 70 to 95%, which showed that the injection of dsβ-gal and dsOBP27 did not affect the insect's lifespan (**Figure 4B**). Another important aspect of male physiology, which was not affected by dsRNA treatment, was blood feeding. The reduction of RproOBP27 expression did not affect the ability of male adults to take a blood meal (**Figure 4C**). There was no significant difference (P = 0.4206) in blood intake by dsOBP27-treated (249.1 ± 16.85 mg/blood, N = 5) and dsβ-gal-treated (215.8 ± 23.14 mg/blood, N = 5) insects.

#### Behavioral Response of dsOBP27-Treated Male

The time spent by the male to move across the tube and reach next to the female was recorded for a period of 300 s. dsOBP27-treated insects accessed the FZ (female zone) with a speed of 0.96 ± 0.12 mm/s, whereas dsβ-gal-treated insects responded significantly faster (1.63 ± 0.19 mm/s, N = 14, P = 0.0065) (**Figure 4D**). dsOBP27 insect-groups stayed close to females for a significantly (P = 0.002) shorter period of time (126.1 ± 17.6 s) than dsβ-gal-treated insects (205.9 ± 15.2 s)

FIGURE 2 | Expression profile of RproOBP6, RproOBP13, RproOBP17, RproOBP21, RproOBP26, and RproOBP27 in different R. prolixus tissues evaluated by conventional PCR. N, negative control; FA, female antennae; MA, male antennae; FP, female proboscis; MP, male proboscis; FH, female head; MH, male head; FL, female legs; ML, male legs. RproR18S was used as an endogenous control. The amplicon size (bp) is indicated on the right. Heads were used without antennae and proboscis.

(**Figure 4E**). Additionally, we observed that, as opposed to treated insects, control males attempted to copulate with females through the mesh separating them in the arena.

#### 3D Model Prediction and in silico Forecasting of RproOBP27 Function

Using Phyre2, 12 3D models were obtained, including Antheraea polyphemus PBP1 [PDB#2JPO; confidence (C) = 99.4; % i.d. = 18]; Amylois transitella PBP1 (PDB#4INW; C = 99.4; % i.d. = 18); Bombyx mori PBP1 (PDB#1DQE; C = 99.3; % i.d. = 15), OBP2 (PDB#2WCL; C = 99.3; % i.d. = 22), Leucophaea maderae PBP (PDB#1OW4; C = 99.1; % i.d. = 10); Apis melifera OBP5 (PDB#3R72; C = 99.1; % i.d. = 12); An. gambiae OBP4 (PDB#3Q8I; C = 99.0; % i.d. = 17); A. melifera OBP14 (PDB#3S0B; C = 98.9; % i.d. = 10), OBP (PDB#1R5R; C = 98.9; % i.d. = 14), Phormia regina OBP56a (PDB#5DIC; C = 98.9; % i.d. = 14); An. gambiae OBP20 (PDB#3BV1; C = 98.8; % i.d. = 16), and Locusta migratoria OBP1 (PDB#4PT1; C = 98.8; % i.d. = 16). Then, PROCHECK and Verify 3D were used to find a model for RproOBP27. The best model for RproOBP27 (**Figure 5A**) was obtained using the crystal structure of OBP20 from An. gambiae (AgamOBP20, PDB#3VB1) as template and used in docking studies. This model was the one which best satisfied the criteria required by PROCHECK and Verify 3D in order to validate as a good model (**Supplementary Figures S6**, **S7**). Binding affinities of RproOBP27 were tested against 2-methyl-3-buten-2-ol, (2S)-pentanol, (3E)-2-methyl-3-penten-2-ol and (2R/2S)-4 methyl-3-penten-2-ol. Thermodynamically, ligands tightly bind the active site of a protein when the free binding energy of the process is low. Therefore, such parameter was used to estimate binding affinities of 2-methyl-3-buten-2-ol, (2S)-pentanol, (3E)- 2-methyl-3-penten-2-ol and (2R/2S)-4-methyl-3-penten-2-ol to RproOBP27. Negative values suggested favorable interactions with all tested ligands (**Figure 5B**). However, since using a cut-off value of −4.00 results still indicate that RproOBP27 is able to bind

standard deviation (SD) of the means of three biological replicates. Statistical analysis was performed using t-test followed by the Mann–Whitney test. RproR18S was used as an endogenous control. Asterisks represent a significant difference between males and females (P < 0.05). FA, female antennae; MA, male antennae.

(3E)-2-methyl-3-penten-2-ol and (2R/2S)-4-methyl-3-penten-2 ol, both compounds already described as being involved in flight orientation modulated by female-emitted volatiles, a malespecific behavior. The 3D model of RproOBP27 docked with MG volatiles (3E)-2-methyl-3-penten-2-ol and (2R/2S)-4-methyl-3 penten-2-ol appears in **Supplementary Figure S8**. The amino acid sequence of RproOBP27 is displayed in **Supplementary Figure S9**.

#### DISCUSSION

Chemical communication is one of the oldest forms of communication used from worms to mammals (Wyatt, 2014; Tomberlin et al., 2016). Insects have a refined olfactory system for the detection of chemical signals from the environment. Chemical signals evoke specific behaviors which allows insects to obtain food, find mates and shelter, and run away from predators. The first contact between the external environment and the internal olfactory machinery occurs when odorants penetrate through the sensillum pores in antennae and reach soluble OBPs found in the sensillar lymph (Brito et al., 2016). Subsequently ORs, IRs, and odorant degrading enzymes are involved. The processing of these semiochemicals ultimately leads to a behavioral response which is essential for insect survival (Leal, 2012). Thus, blocking the first step of the process could be a key factor for controlling insect populations. Research regarding olfactory mechanisms of R. prolixus for such purpose only became possible after genome release, when many genes related to chemosensory detection were identified

males to arrive close to a caged female was recorded during for up to 300 s. (E) Time spent by dsRNA treated male close to female. After molt, the adult injected with dsOBP27 and dsβ-gal was blood-fed. After 7 days fed insects were individually tested using a polystyrene tube. Time spent by males close to female was recorded for up to 300 s. Error bars represent standard deviation of the means of three biological and technical replicate. Statistical analysis was performed using t-test followed by the Mann–Whitney test. Asterisks represent a significant difference (P < 0.05).

(Mesquita et al., 2015). Here, we present a compilation of data that strongly suggests the role of RproOBP27 in R. prolixus behavior.

## Profile of Odorant Binding Protein Genes

Although the genome predicts 27 genes which encode OBPs (Mesquita et al., 2015), only 17 OBPs were found to be expressed in adult antennae (Oliveira et al., 2017), which suggests that several genes belonging to the OBP family may not be directly involved in odor transport, as observed in other insects (Pelosi et al., 2018). Moreover, amongst the 11 OBP transcripts identified in antenna, leg, proboscis, and head from adults (**Figure 1**), four had already been described in the midgut transcriptome: RproOBP1, RproOBP11, RproOBP14, and RproOBP24 (Ribeiro et al., 2014). Such evidence favors the assumption that these proteins might be involved in transporting general molecules, not necessarily related to odorant reception. In fact, RproOBP11, known as Rhodnius heme-binding protein (RHBP), is responsible for the transport of heme radicals generated from blood digestion, shielding cells from oxidative stress (Dansa-Petretski et al., 1995). Some OBPs, for instance, are important in nutrition as lipids solubilizers and other components of the insect diet (Sanchez-Gracia et al., 2009). Therefore, it was not entirely surprising to find transcripts for OBPs distributed in nonolfactory tissues.

Using conventional PCR, six OBPs transcripts were specifically expressed in antennae: RproOBP6, RproOBP13, RproOBP17, RproOBP21, RproOBP26, and RproOBP27 (**Figure 2**), which suggests these OBPs may, in fact, be associated with odorant transport as it has been reported for other insects (Leal, 2012; Schultze et al., 2012; Sun et al., 2014). Of note, no clear differences were observed in transcript levels of antenna specific OBPs between male and female using conventional PCR (**Figure 2**).

Given that qPCR data showed RproOBP6 and RproOBP13 were expressed in male and female antennae (**Figures 3A,B**), it is conceivable that these OBPs are involved in the detection of odorants eliciting common adult behaviors (e.g., host finding). R. prolixus belongs to the Reduviidae family, where adults are hematophagous (Guerenstein and Lazzari, 2009; Sant'Anna et al., 2017), therefore, adults need to accurately detect host specific volatiles to acquire their blood meal (Otalora-Luna et al., 2004). Thus, we propose that RproOBP6 and RproOBP13 might transport host emanations.

Transcripts for RproOBP17 and RproOBP21 were enriched in female antennae (**Figures 3C,D**), indicating these proteins might be involved in female-specific behaviors. This hypothesis is supported by the finding that in the mosquito Culex quinquefasciatus, another hematophagous insect, some OBPs expressed in the female antenna are specifically related to the detection of oviposition odorants. OBP2 is postulated to carry the oviposition attractant skatole, whereas OBP1 and OBP5 were implicated in the transport of a mosquito oviposition pheromone (MOP), which induced oviposition behavior in females (Pelletier et al., 2010; Yin et al., 2015).

Lastly, transcripts for RproOBP26 and RproOBP27 were found to be significantly expressed in the male antenna (**Figures 3E,F**). These results strongly suggest that these proteins could play a role in male-specific behaviors, such as sex pheromone detection. In the mosquito Aedes aegypti, OBP10 is enriched in antennae and wings of adult male and it expression pattern has been suggested to correspond to proteins that may play a role on male chemosensory behavior such as pheromone detection (Bohbot and Vogt, 2005). Although RproOBP26 was highly expressed in antennae (**Figure 3E**), it was also reported to be overexpressed in the midgut of R. prolixus (called RP-3726) (Ribeiro et al., 2014). Here we showed that transcripts for RproOBP26 were significantly more expressed in male than female antennae (**Figure 3E**). However, proteome studies have found soluble RproOBP26 in both male and the female antennae (Oliveira et al., 2017). Thus, we cannot rule out the possibility that RproOBP26 might be involved in the transport of non-sensorial molecules in the gut, as well as semiochemicals in antennae. Although, only one gene for RproOBP26 has been annotated in the genome (Mesquita et al., 2015), we could not exclude the possibility of RproOBP26 has alternative splicing, as previously observed for other insect species (Forêt and Maleszka, 2006; Hull et al., 2014).

#### Role of OBP27 in R. prolixus Behavior

Previously, we have demonstrated a direct correlation between an olfactory protein (Orco) and R. prolixus behavior by RNAi (Franco et al., 2016). We then surmised that silencing OBPs might lead to behavioral changes in the phenotype. After all, gene silencing has already been successfully applied to investigate functions of OBPs in other insects (Chen et al., 2008; Pelletier et al., 2010; Rebijith et al., 2016). Of the two OBPs specific to male antennae, we selected RproOBP27 for these studies. We envisioned that this protein might generate a clearer picture than RproOBP26 given the possible dual role (or multiple roles) played by RproOBP26 in R. prolixus physiology.

Adult males treated with dsOBP27 had a reduction of 88% in RproOBP27 expression (**Figure 4A**), representing a drastic decrease in the amount of protein circulating in antennae. However, this reduction in gene expression did not interfere with survival or blood-intake, since both groups (control- and dsOBP27-insects) ingested almost the same amount of blood (**Figures 4B,C**). Differently, a reduction in expression of odorant coreceptor Orco in R. prolixus antenna affected directly the ability of insect to take a blood meal (Franco et al., 2016). Thus, we can suggest that RproOBP27 is not involved in the host-seeking or blood-intake behavior. Next, we tested whether RNAi treatment would affect male ability to detect females. Insects injected with a control gene were able to detected females and run in their direction faster than dsOBP27-treated males (**Figure 4D**). Further, dsOBP27-males spent almost 40% less time nearby the female when compared to control insects (**Figure 4E**). In addition, while males from control groups tried to stay close to females, dsOBP27-treated insects kept running around the tube, indicating they were not able to detect a female. Based on this clear behavioral difference, we hypothesize that RproOBP27 may be involved in the reception of semiochemicals related to mating finding. This hypothesis is further supported by in silico analysis.

Volatile compounds emitted by R. prolixus female MGs are known to modulate male orientation and to increase copulation attempts (Pontes et al., 2014). Of the 12 compounds identified in MGs, four are considered putative sex pheromones, namely, 2-methyl-3-buten-2-ol, (2S)-pentanol, (3E)-2-methyl-3-penten-2-ol, and (2R/2S)-4-methyl-3-penten-2-ol (Pontes et al., 2008). Docking results (**Figure 5**) indicate favorable interactions with all four tested ligands due to negative values calculated for free binding energy. Even when a more restricted analysis, based on previous studies for predicting behaviorally active compounds (Jayanthi et al., 2014), is used to estimate binding potential, 2 methyl-3-penten-2-ol and (2R/2S)-4-methyl-3-penten-2-ol still meet the criteria for high binding affinity to RproOBP27. These results further support our hypothesis that RproOBP27 is a carrier of female-derived semiochemicals.

In the Lucerne plant bug, Adelphocoris lineolatus, expression of OBP1 is 1.91 times higher in male than in female antennae and this protein was shown to exhibit high binding affinity with two putative pheromone components (Gu et al., 2011b). Recent study suggested that OBP expression could be regulated by nutritional state. In A. lineolatusstarvation significantly increased expression of AlinOBP13 in male and female antenna (Sun et al., 2014). Likewise, starved R. prolixus males did not express RproOBP27 (**Supplementary Figure S4B**), which was found only in the antennae of fed males. This dataset is consistent with the findings that unfed males from this species do not respond to sexual signals (Baldwin et al., 1971). Taking together, the evidence presented here strongly suggests that RporoOBP27 is likely involved in the reception of sex pheromone(s).

#### AUTHOR CONTRIBUTIONS

fphys-09-01175 August 22, 2018 Time: 9:41 # 9

AM and WL designed the project and experiments. DO, NB, and TF performed the experiments. DO, NB, TF, MM, WL, and AM analyzed the data. DO, NB, WL, and AM wrote the paper. DO, NB, TF, MM, WL, and AM revised the paper.

#### FUNDING

This work was supported by grants from Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ), Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Instituto Nacional de Ciência e Tecnologia em Entomologia Molecular (Brazil) (INCT-EM/CNPq), and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES). Work at the University of California, Davis was supported in part by National Institute of Allergy and Infectious Diseases of the National Institutes of Health (NIH) (Grant No. R01AI095514).

#### ACKNOWLEDGMENTS

The authors are grateful to Lauriene D. Severino, Yasmin de Paule Gutierrez Simão, and Desenir Adriano Pedro for their excellent technical assistance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01175/full#supplementary-material

#### REFERENCES


FIGURE S1 | Expression profile of (A) RproOBP1 and RproOBP24; (B) RproOBP26; (C) RproOBP11 and RproOBP13; (D) RproOBP12 and RproOBP7 in different R. prolixus tissues evaluated by conventional PCR. Original 1% agarose gel stained with GelRedTM. M, molecular weight; N, negative control; FA, female antennae; MA, male antennae; FP, female proboscis; MP, male proboscis; FH, female head; MH, male head; FL, female legs; ML, male legs.

FIGURE S2 | Expression profile of (A) RproOBP14 and RproOBP6; (B) RproOBP17; (C) RproOBP18 and RproOBP20; (D) RproOBP21 in different R. prolixus tissues evaluated by conventional PCR. Original 1% agarose gel stained with GelRedTM. M, molecular weight; N, negative control; FA, female antennae; MA, male antennae; FP, female proboscis; MP, male proboscis; FH, female head; MH, male head; FL, female legs; ML, male legs.

FIGURE S3 | Expression profile of (A) RproOBP22; (B) RproOBP22 and RproOBP23; (C) RproOBP22 and RproOBP26; (D) RproOBP29 and RproOBP27 in different R. prolixus tissues evaluated by conventional PCR. Original 1% agarose gel stained with GelRedTM. M, molecular weight; N, negative control; FA, female antennae; MA, male antennae; FP, female proboscis; MP, male proboscis; FH, female head; MH, male head; FL, female legs; ML, male legs.

FIGURE S4 | Expression profile of (A) RproR18S; (B) RproOBP27, RproOBP26, and RproOBP21 in different R. prolixus tissues evaluated by conventional PCR. Original 1% agarose gel stained with GelRedTM. M, molecular weight; N, negative control; FA, female antennae; MA, male antennae; FP, female proboscis; MP, male proboscis; FH, female head; MH, male head; FL, female legs; ML, male legs.

FIGURE S5 | Device used in female recognition bioassay. A polystyrene tube (10 × 2 cm) divided into three zones: female zone (FZ), intermediate zone (IZ), and male release zone (MZ). A gate divides the MZ from IZ. A protective mesh was used to separate MZ and IZ from FZ. An adult female was placed in front of the protection mesh attached by a tape on the tube. Then a male was placed in the MZ and the gate was opened. Adapted from Zermoglio et al. (2015).

FIGURE S6 | PROCHECK results from predicted 3D model of RproOBP27.

FIGURE S7 | Verify3D results from predicted 3D model of RproOBP27.

FIGURE S8 | 3D model of RproOBP27 docked with metasternal gland volatile compounds (putative sex pheromones). (A) RproOBP27 docked with (3E)-2-methyl-3-penten-2-ol. (B) RproOBP27 docked with (2R/2S)-4-methyl-3-penten-2-ol.

FIGURE S9 | RproOBP27 sequence. The signal peptide is highlighted in red.

TABLE S1 | Oligonucleotides used in the PCR, qPCR and dsRNA synthesis reactions.




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Oliveira, Brito, Franco, Moreira, Leal and Melo. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Molecular and Functional Characterization of Odorant Binding Protein 7 From the Oriental Fruit Moth *Grapholita molesta* (Busck) (Lepidoptera: Tortricidae)

Xiu-Lin Chen1,2†, Guang-Wei Li 2†, Xiang-Li Xu<sup>1</sup> \* and Jun-Xiang Wu<sup>1</sup> \*

*<sup>1</sup> Key Laboratory of Plant Protection Resources and Pest Management (Northwest A&F University), Ministry of Education, Yangling, China, <sup>2</sup> Shaanxi Province Key Laboratory of Jujube, College of Life Science, Yan' an University, Yan'an, China*

#### *Edited by:*

*Peng He, Guizhou University, China*

#### *Reviewed by:*

*Guo Mengbo, Nanjing Agricultural University, China Haonan Zhang, University of California, Riverside, United States*

#### *\*Correspondence:*

*Xiang-Li Xu xuxiangli@nwsuaf.edu.cn Jun-Xiang Wu Junxw@nwsuaf.edu.cn*

*†These authors have contributed equally to this work*

#### *Specialty section:*

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

*Received: 11 July 2018 Accepted: 22 November 2018 Published: 10 December 2018*

#### *Citation:*

*Chen X-L, Li G-W, Xu X-L and Wu J-X (2018) Molecular and Functional Characterization of Odorant Binding Protein 7 From the Oriental Fruit Moth Grapholita molesta (Busck) (Lepidoptera: Tortricidae). Front. Physiol. 9:1762. doi: 10.3389/fphys.2018.01762* Odorant-binding proteins (OBPs) are widely and abundantly distributed in the insect sensillar lymph and are essential for insect olfactory processes. The OBPs can capture and transfer odor molecules across the sensillum lymph to odorant receptors and trigger the signal transduction pathway. In this study, a putative OBP gene, *GmolOBP*7, was cloned using specific-primers, based on the annotated unigene which forms the antennal transcriptome of *Grapholita molesta*. Real-time PCR (qRT-PCR) analysis revealed that *GmolOBP*7 was highly expressed in the wings of males and the antennae of both male and female adult moths, while low levels were expressed in other tissues. The recombinant GmolOBP7 (rGmolOBP7) was successfully expressed and purified via Ni-ion affinity chromatography. The results of binding assays revealed that rGmolOBP7 exhibited a high binding affinity to the minor sex pheromone 1-dodecanol containing *K<sup>i</sup>* of 7.48µM and had high binding capacities to the host-plant volatiles, such as pear ester, lauraldehyde and α-ocimene. RNA-interference experiments were performed to further assess the function of GmolOBP7. qRT-PCR showed that the levels of mRNA transcripts significantly declined in 1 and 2 day old male and female moths, treated with *GmolOBP*7 dsRNA, compared with non-injection controls. The EAG responses of dsRNA-injected males and females to pear ester, as well as the EAG responses of dsRNA-injected males to 1-dodecanol, were significantly reduced compared to the GFP-dsRNA-injected and non-injected controls. We therefore infer that GmolOBP7 has a dual function in the perception and recognition of the host-plant volatiles and sex pheromones.

Keywords: *Grapholita molesta*, odorant binding protein, olfaction, fluorescence binding assay, tissue expression

## INTRODUCTION

The sophisticated olfactory system plays an essential role in an insect's survival and reproduction. Adult insects greatly depend on olfactory cues to locate mates and optimal host plants and avoiding predators (Takken and Knols, 1999; Leal, 2013; Suh et al., 2015). In the early events of olfactory processing, airborne chemical signals must pass through the aqueous barrier of the sensillum lymph, surrounding the dendrites of the olfactory receptor neuron (ORNs) cells (Li et al., 2015).

**223**

Odorant-binding proteins (OBPs), a kind of transport protein, can selectively bind and carry hydrophobic odorants across the sensillum lymph to odorant receptors (ORs) and trigger the signal transduction pathway (Pelosi et al., 2005). The converted electrophysiological signals are then sequentially processed in the antennal lobes, mushroom bodies and central nervous area, to induce a behavioral response in specific semiochemicals of insects (Feng and Prestwich, 1997; Helfrich-Förster, 2000; Hallem et al., 2006; Pelosi et al., 2006; Leal, 2013; Yi et al., 2014). OBPs are responsible for the connection between the external environment and ORNs in the odorant-molecule recognition process. OBPs also mediates the first stage of the physiological process involved in the sensing of the external environment, by insects (Willett and Harrison, 1999; Laughlin et al., 2008; Pelosi et al., 2014; Leal and Leal, 2015).

OBPs belong to a class of small water-soluble proteins that are impregnated in the sensillum lymph at extremely high concentrations (up to 10 mM; Vogt and Riddiford, 1981; Klein, 1987; Steinbrecht et al., 1992). The first insect OBP was identified in the antennae of male Antheraea ployphemus. By using a radiolabeled photo-affinity analog, this protein was designated as a pheromone binding protein (PBP) as it specifically bound to the female sex pheromone E6, Z11-hexadecadienyl acetate (Vogt and Riddiford, 1981). Since then, OBPs have been discovered in various insect orders (Hansson and Stensmyr, 2011; Antony et al., 2018; Fleischer et al., 2018). Lepidopteran OBPs are usually subdivided into three subfamilies including PBPs, general OBPs, and antennal binding proteins (ABPX), on the basis of aminoacid sequence homologies (Hekmat-Scafe et al., 2002). The PBPs are located in the sensilla trichodea and exhibit specific binding to female sex pheromones (Bette et al., 2002; Lautenschlager et al., 2007). GOBPs (further classified as GOBP1 and GOBP2) are primarily distributed in the sensilla basiconica and their function is mainly involved in the detection of general odorants (e.g., host plant volatiles; Vogt et al., 2002; Nardi et al., 2003; Maida et al., 2005; Liu et al., 2015). In some Lepidopteran species, GOBP2 also showed high-binding affinities to sex pheromones in addition to general odorants (Liu et al., 2010, 2012; Li et al., 2016a). ABPX are more divergent among insects and its functions may play a similar role than PBPs or GBOPs in the discrimination and transportation of semiochemicals (Tian et al., 2018).

The binding affinities of insect OBPs to odorant molecules have been measured via fluorescence competitive binding assays with N-phenyl-1-naphthylamine (1-NPN) as a probe (Pelosi et al., 2006; Zhou, 2010). For example, Helicoverpa armigera HarmOBP17 and HarmOBP18 have strong binding capacities to β-ionone (Li et al., 2013). Locusta migratoria LmigOBP1 exhibited specific-binding affinities to pentadecanol and 2 pentadecanone, where Asn74 formed the key binding site in these two ligands (Jiang et al., 2009). Grapholita molesta GmolGOBP2 had specific binding ability to the minor sex pheromone component 1-dodecanol (Li et al., 2016a). The fluorescence competitive binding assay is only in vitro and the binding functions of OBPs still need to be verified further by experiments in vivo. RNAi experiments demonstrated that OBPs are indispensable in the olfactory communication of insects. For example, the electroantennogram (EAG) responses of female Adelpocoris lineolatus to tridecanal and 1-hexanol were drastically reduced after the double-stranded RNA (dsRNA) of AlinOBP4 was injected into both female and male adult insects (Zhang et al., 2017). The EAG values of AgosOBP2-dsRNAtreated Aphis gossypii to cotton-derived volatiles were remarkably lower than those of non-injected controls (Rebijith et al., 2016). By silencing the RferOBP1768 gene of Rhynchophorus ferrugineus, the adults apparently lose the ability to recognize the aggregation pheromone compounds 4-methyl-5-nonanol and 4 methyl-5-nonanone (Antony et al., 2018).

The oriental fruit moth Grapholita molesta, is a destructive fruit pest species that causes considerable economic losses in fruit yields on a global scale (Rothschild and Vickers, 1991). The first three moth generations mainly infest peach shoots in the early growing season, whereas the third generation begins to shift and attack pear and apple orchards in the late growing season. The migration of the adults is predominantly guided by the change in volatile components emitted by these host plants (Myers et al., 2007). At present, monitoring the G. molesta mainly depends on the pheromone trapping of male moths. However, the females have multiple mating abilities and their flight capabilities are three to six times greater than that of males and the females also have higher mating rates in the pheromone trapping orchards (Hughes and Dorn, 2002; Il'ichev et al., 2007; Zhang et al., 2012). Therefore, a strategy to monitor both female and male moths, based on olfactory cues emitted from host plants, is desirable. For example, a three-compound mixture of (Z)-3-hexen-1-ol, (Z)- 3-hexen-1-yl acetate, and benzaldehyde in proportion 1:4:1 can attract female G.molesta just as well as the natural blend from peach shoots can (Natale et al., 2003).

In this study, GmolOBP7 was cloned using specific-primers based on the annotated unigene from the antennal transcriptome of G. molesta. qRT-PCR was performed to determine the expression patterns of GmolOBP7 in different tissues, genders, and developmental stages of the G. molesta. The binding affinities of the rGmolOBP7 with sex pheromone components and the host plants' volatiles, were measured via fluorescence binding assays. Furthermore, the ligand-binding functions of GmolOBP7 were further verified in vivo by knocking down the GmolOBP7 gene. The olfactory mechanism of the oriental fruit moth was further explicated to provide a theoretical basis for the design and implementation of control strategies against this fruit pest.

## MATERIALS AND METHODS

#### Insect Samples

G. molesta individuals were obtained from the College of Plant Protection, Northwest A&F University, Yangling, Shaanxi, China. The laboratory colony has been maintained for more than 90 generations. The larvae were reared on an artificial diet at 25 ± 1 ◦C, 70% ± 5% RH under a day/night cycle of 15:9, until pupation (Du et al., 2010). After pupation, male, and female pupae were placed in separate glass tubes and maintained under the same conditions described above. The adults were fed 5% honey solution daily. To detect the tissue distribution of GmolOBP7 in adult moths, various tissues (including antennae, heads without antennae, thoraces, abdomens, legs, and wings) were collected from 3-day-old males and females and immediately transferred to 1.5 mL Eppendorf tubes immersed in liquid nitrogen. All samples were stored at −80◦C prior to use. In order to determine the transcript level of GmolOBP7 in different developmental stages of the G. molesta, samples of eggs, larvae (including 1st, 2nd, 3rd, 4th, and 5th instars), pupae (including prepupae and later-pupae) and adults (including 1-d-old, 3-d-old, and 5-d-old adults) were collected and stored at −80◦C prior to use.

## RNA Extraction, OBP Cloning, and Sequencing

Total RNA of all samples was extracted using a RNAiso Plus reagent (TaKaRa, Daian, China) according to the manufactures' instructions. The residual genomic DNA in the total RNA was removed using DNase I (Thermo Scientific, USA), and the first-strand cDNA was synthesized in accordance with the recommended protocols of the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, USA). The products were stored at −80◦C.

The predicted coding region of GmolOBP7 was cloned using specific-primers based on the annotated unigene from the antennal transcriptome of G. molesta (**Table 1**). The predicted results showed that GmolOBP7 had no signal peptide at the N-terminus of the amino acid sequence. In order to confirm whether we acquired the complete coding sequence of GmolOBP7, gene-specific primers were synthesized and used for 5' RACE (rapid amplification of cDNA ends; **Table 1**) Refers to a procedure in a previous study (Luo et al., 2011). A first 41-cycle touchdown PCR was performed using 5' RACE outer primers (named outer5F and outer5R; **Table 1**). A 25 µL PCR reaction system contained 12.5 µL of 2 × Super Pfx MasterMix (CWBIO, Beijing, China), 0.8 µL of each primer (10µM), 1 µL of sample cDNA, and 9.9 µL of nuclease free water. The thermocycling program included denaturation at 95◦C for 5 min, followed by 16 cycles of 30 s at 95◦C, 1 min at 65 ◦C, and 2 min at 72 ◦C, and the annealing temperature was decreased 1◦C each four cycles. The remaining 25 cycles consisted of 30 s at 95◦C, 1 min at 61 ◦C, and 2 min at 72 ◦C, and a final extension step of 72◦C for 10 min. The PCR products were diluted 80 times with sterilized ddH2O. The second 41-cycle touchdown PCR was conducted using 5' RACE inner primers (named inner5F and inter5R; **Table 1**) and the template with diluted PCR products. The reaction system and procedure was the same as the first round of PCR. The amplified product was purified with an Universal DAN Purification Kit (TianGen, Beijing, China), and cloned into the pMD <sup>R</sup> 19-T cloning vector (TaKaRa, Dalian, China) and then transformed into DH5α Escherichia coli competent cells (TianGen, Beijing, China). Five positive clones were randomly selected for sequencing at the Aoke Biotech Company (Aoke, Xi'an, China).

#### Sequence and Phylogenetic Analyses

The online programs of ORF Finder (http://www.ncbi.nlm. nih.gov/gorf/gorf.html), SignalP 4.0 (http://www.cbs.dtu. dk/services/SignalP/), and ExPASy server (https://web. expasy.org/compute\_pi/) were used to predict the ORFs, signal peptides and the molecular weight and isoelectric point of mature protein of GmolOBP7, respectively. The amino-acid sequences were aligned using ClustalX 1.83 software. A phylogenetic tree was established by the MEGA6.0 software using the neighbor-joining method (NJ) with 1,000 bootstrap replications, and the tree was drawn using Adobe Photoshop CS5.

#### Expression Analysis Using qRT-PCR

The expression levels of GmolOBP7, in different tissues and developmental stages of G. molesta, were measured via qRT-PCR. All qRT-PCR experiments were performed according to the MIQE Guidelines (Bustin et al., 2009). Specific primers were designed using the program Primer3-blast (https://www. ncbi.nlm.nih.gov/tools/primer-blast/), available online (**Table 1**). The elongation factor 1-alpha gene (EF1-α) (GenBank No: KT363835.1) and the β-actin gene (GenBank No: KF022227.1) were used as reference genes. The reactions were performed on a CFX96 Real-Time PCR Detection System (Bio-Rad, USA). Each amplification reaction was conducted using a 20 µL reaction system containing 10 µL of 2×SYBR <sup>R</sup> Premix Ex TaqTM II mixture (TaKaRa, Dalian, China), 0.8 µL of each primer (10µM), 1 µL of sample cDNA, and 7.4 µL of nuclease-free H2O. Samples without a template cDNA served as negative controls. To check reproducibility, test samples and negative controls were performed in triplicates. qRT-PCR was performed via initial denaturation at 95◦C for 30 s, followed by 40 cycles of 95◦C for 5 s, 60◦C for 30 s and 72 ◦C for 30 s. The melting curves were used to examine primer specificity, and the standard curves were used to determinate the amplification efficiencies of target and reference genes. The expression levels of GmolOBP7 in different adult tissues and development stages were performed based on previous methods (Livak and Schmittgen, 2001; Liu et al., 2016). The expression level of all samples of the GmolOBP7 was calculated using the value of the amplification efficiency (E) and the value of the cycle threshold (Ct) (Equation 1). The normalized expression level of the tested samples was calculated by the geometric means of the expression level of the reference genes (β-actin and EF-1α) (Equation 2) (Vandesompele et al., 2002). The significant differences in different tissues and developmental stages were analyzed by the Tukey's HSD tests with a critical level of α = 0.05. The paired t-test was used to measure the impacts of the expression of GmolOBP7 between male and female moths. All the data were analyzed using SPSS 18.0 software (SPSS Inc., Chicago, IL, USA).

$$\text{Express level} = \left(1 + E\right)^{-Ct} \tag{1}$$

$$\text{Normalized expression level of target gene}$$

$$= \frac{\left(1 + E\_{\text{target}}\right)^{-Ct\_{\text{target}}}}{\sqrt{\left(1 + E\_{\text{action}}\right)^{-Ct\_{\text{active}}} \times \left(1 + E\_{EF-1\alpha}\right)^{-Ct\_{EF-1\alpha}}}} \tag{2}$$

#### Expression Vector Construction

Specific primers with restriction enzyme sites were designed to clone the coding region of the GmolOBP7 (**Table 1**), and the TABLE 1 | List of primers used in the current research.


*The restriction endonucleases are in parentheses after each primer, and the restriction sites are underlined. "–" means the amplified DNA fragment is an unknown in size.*

PCR product was then cloned into the pMD <sup>R</sup> 19-T cloning vector (TaKaRa, Dalian, China) and sequenced. The recombinant plasmid pMD <sup>R</sup> 19-T/GmolOBP7 and the expression vector pET32a(+) (Novagen, Madison, WI, USA) were digested with the same restriction endonucleases, and the released DNA fragment was cloned into pET-32a(+) and then transformed into BL21 E. coli competent cells (Tiangen, Beijing, China). A positive clone containing pET32a(+)/GmolOBP7 was further confirmed by sequencing.

#### Protein Expression and Purification

The overnight bacterial solution was diluted with 750 mL of LB medium (with 100 mg/mL ampicillin) and cultured at 37◦C until its cell density reached a value of OD<sup>60</sup> = 0.6. The cultures were induced by adding isopropyl-β-Dthiogalactoside (IPTG) at a final concentration of 0.5 mM for an additional 5 h at 37◦C, 220 rpm. The bacterial cells were harvested by centrifugation (10 min at 8,000 rpm, 4◦C), and the pellets were then sonicated in a lysis buffer (1 mM phenylmethanesulfonyl fluoride, 250 mM NaCl, and 20 mM Tris-HCl pH 7.4) and centrifuged again (13,000 g, 30 min, 4◦C). A sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis revealed that the rGmolOBP7 was mainly present in supernatants. The supernatant of rGmolOBP7 was enriched by a Ni-NTA His·Bind Resin column (7 sea Pharmatech Co., Shanghai, China) in accordance with the manufacturer's instructions. To avoid the effects of His-tag on subsequent experiments, the His-tag was cleaved by a recombinant enterokinase (NEB, Beijing, China) and removed using the column mentioned above. The target protein was purified using affinity chromatography, and the concentration was determined by the BCA protein kit (Beyotine, Shanghai, China).

## Fluorescence Binding Assays

Fluorescence intensity was detected on a spectrophotofluorometer (F-4500, Hitachi, Japan) at room temperature using a quartz cuvette with a 1 cm light path. The silt width of excitation and emissions were all 10 nm. The fluorescence probe 1-NPN was excited at 337 nm and the emission spectra were recorded between 370 and 550 nm. Four sex pheromone components and 31 volatiles, derived from peach shoots and pear fruits, were selected for binding assays (**Table 2**). The probe 1-NPN and tested ligands were all dissolved in spectrophotometric-grade methanol to obtain a 1 mM stock solution. The binding affinity of rGmolOBP7 with 1-NPN was measured by adding aliquots of 1-NPN to a 2µM protein solution (diluted with 20 mM Tris-HCl pH 7.4) to final concentrations of 0 to 18µM.

To test the binding affinities of GmolOBP7 to various ligands, 2µM solutions of rGmolOBP7 and 1-NPN were titrated with the 1 mM solution of each ligand to a final concentration of 0– 14µM for sex pheromones and 0–35µM for host volatiles. The corresponding florescence intensity values were collected as three TABLE 2 | Binding affinities of GmolOBP7 to various ligands were measured via competitive binding assays using 1-NPN as a fluorescent probe.


*More than* >*35*µ*M indicates that the IC*<sup>50</sup> *and K<sup>i</sup> values are above the concentration ranges tested.*

independent measurements. The binding constant (K1−NPN) of 1-NPN to rGmolOBP7 was calculated using GraphPad Prism 5 software (GraphPad Software, Inc.) via nonlinear regression for a unique site of binding. The dissociation constant (Ki) of each ligands competitive binding to rGmolOBP7, were calculated from the corresponding IC50, by using the equation Ki = [IC50]/(1+[1-NPN]/K1−NPN), where [1-NPN] is the free concentration of 1-NPN, and K1−NPN is the dissociation constant of the complex protein/1-NPN.

#### dsGmolOBP7 and dsGFP Synthesis

The specific-primers, including T7 RNA polymerase promoter, were designed to clone DNA fragments of GmolOBP7 for 317 bp and green fluorescent protein (GFP) for 315 bp (**Table 1**). The purified PCR products were used as a template for dsRNA (dsGmolOBP7 and dsGFP) synthesis using the T7 RiboMAXTM Express RNAi System kit (Promega, USA) according to the manufacturer's instructions. The purified dsRNA was quantified via spectrophotometry (SimpliNano, GE, USA), and the dsRNA integrity was monitored by electrophoresis on 1.5% agarose gel. Each dsRNA sample was dissolved in nuclease-free water to a final concentration of 3,500 ng/µL.

#### dsRNA Microinjection

Based on the expression patterns observed at different insect stages, 5-day-old G. molesta pupae (later-pupae) were selected to receive a dsRNA microinjection. The conjunctivum between the prothorax and mesothorax, the conjunctivum between the mesothorax and metathorax and the conjunctivum between the thorax and abdomen were initially selected as putative injection sites. Then, 39, 69, and 138 nL of RNase-free H2O were injected into different conjunctiva, respectively. A total of 69 nL (approximately 241.5 ng dsRNA) of dsGmolOBP7 or dsGFP was injected into the appropriate injection sites of each 5-dayold pupa, by using a PL1-100 Pico-Injector (Harvard Apparatus, Holliston, MA, USA) operated by an MP-255 Micromanipulator (Sutter, Novato, CA, USA). Each type of dsRNA was injected into 600 male and 600 female moths. The heads (with antenna) of 1-, 2-, 3-, and 4 day-old adults were dissected and immediately stored at −80◦C prior to use. The total RNA and first-strand cDNA were obtained in accordance with previous methods. The specimens were used for qRT-PCR analysis to test the reduction in GmolOBP7 transcription. Experiments were performed with three biological replicates and three technical replicates.

#### EAG Assays

EAG responses of dsRNA-injected (including dsGmolOBP7 and dsGFP) moths and non-injected controls, to sex pheromones and host plant volatiles were detected using Electroantennography. All stimulants were diluted with liquid paraffin to the final concentration of 10 mg/mL. Liquid paraffin and cis-3-hexenyl acetate were used as the blank and reference control, respectively. Both ends of adult antennae were cut and blocked with a drop of Spectra <sup>R</sup> 360 Electrode Gel (Parker Laboratories, Fairfield, USA). The basal section was connected to the reference electrode while the distal end was linked to the recording electrode. A filtered humidified air stream was delivered by a Syntech stimulus controller (CS55 model, Syntech, Germany) at a constant flow rate of 50 cm/s, and the time of stimuli flow was 0.5 s. Filter paper strips (0.6 cm × 4.5 cm) were dripped with 15 µL of chemical solution as a stimulus source and inserted into a 1.5 mL micropipet tip. A set of stimulants consisted of four sex pheromones and five host-plant volatiles which can be strongly bound with rGmolOBP7. Each antenna measured the group of randomly arranged stimulants described above. The antennae were stimulated one time with liquid paraffin and cis-3-hexenyl acetate and dissolved in solvent, before and after each group stimulation, in order to ensure the tested antenna were activated and the connecting pipe was not contaminated by stimulants. Recordings per stimulant were taken and the antennal responses were recorded. Eight male and female antennae were measured for each stimulant. The paired t-test was used to determine whether the difference in EAG values were significant between the dsRNA injected moths (dsGmolOBP7 and dsGFP) and the non-injected control. All the data were analyzed using SPSS 18.0 software (SPSS Inc., Chicago, IL, USA).

## RESULTS

#### Identification of *GmolOBP*7 in *G. molesta*

We obtained the ORF of GmolOBP7 (GeneBank No. MF066359) using ordinary PCR and 5' RACE PCR based on the annotated unigene from the antennal transcriptome of G. molesta. The ORF of GmolOBP7 is 504 bp in length and encodes 167 amino acids (**Figure S1**). GmolOBP7 possesses a common characteristic of known classical-OBPs with six-conserved cysteine motifs (**Figure 1**). The mature protein has a predicted molecular weight of 21.47 kDa and a theoretical pI of 6.85. The SignalP 4.1 server prediction indicated that GmolOBP7 did not have signal peptides at the N-terminus of the amino acid sequence. GmolOBP7 shares the highest identities with Ectropis obliqua EoblOBP1 and Spodoptera exigua SexiOBP11, with an identity of 67 and 64%, respectively. Phylogenetic analysis showed that GmolOBP7 were clustered into a small branch close to ABPXs from Ectropis obliqua and Cnaphalocrocis medinalis (**Figure 2**).

## Expression Profiles of *GmolOBP*7 in *G. molesta*

The amplification efficiency and melting curve of target and reference genes showed that the two specific-primers were similar in amplification speed and no nonspecific products were produced (**Figure S2**), therefore, the primers can be used for relative quantification. GmolOBP7 was detected in all tested tissues of both female and male moths (**Figure 3A**), but the expression quantity was higher in the wings of males and antennae of both sexes, than that in other tested tissues. The expressed quantity of GmolOBP7 was different between males and females, with the quantity in male wings and legs being significantly higher than in female wings and legs (with about 13.31 and 2.99 times differences, respectively), and was higher in female thoraces than in male thoraces (about 2.43-fold higher). The expression levels of GmolOBP7 in different developmental stages were performed by qRT-PCR (**Figure 3B**), the highest expression quantity was found in adults, followed by the eggs, 1st instar larvae, and later pupae (5-day-old pupae), the expression levels of GmolOBP7 were extremely low in second- to fourthinstar larvae.

#### Expression and Purification of *GmolOBP*7

Recombinant GmolOBP7 (rGmolOBP7) was successfully expressed in E. coli as a soluble protein (**Figure 4A**). After being purified, about 38 kDa recombinant protein with His-tag was obtained. To avoid the effect by the His-tag on subsequent binding assays, the His-tag of pET32(a+)/GmolOBP7 was cleaved by enterokinase and then removed by Ni-NTA His·Bind Resin and SDS-PAGE analysis showed that rGmolOBP7 had a higher purity after being purified for a second time via affinity chromatography (**Figure 4B**). The purified recombinant protein was then tested for the binding affinity with various ligands.

#### Fluorescent Binding Assay of GmolOBP7

The binding curve and the derived Scatchard plot showed that the dissociation constant for rGmolOBP7 with the fluorescence probe 1-NPN was 2.30µM (**Figure 5**). This result suggests the

existence of a single binding site and the absence of an allosteric effect between the recombinant protein and the fluorescence probe.

rGmolOBP7 showed broad binding properties with 35 putative ligands, 21 out of 35 ligands succeeded in displacing 1-NPN from the GmolOBP7/1-NPN complex by half, at concentrations up to 35µM. The IC50 values and the calculated binding constants (Ki) are shown in **Table 2**. rGmolOBP7 exhibited high binding affinity to the minor sex pheromone component 1-dodecanol (12:OH) with a K<sup>i</sup> value of 7.48µM. However, rGmolOBP7 did not bind to the major sex pheromone components (Z)-8-dodecenyl acetate (Z8-12:Ac), (E)-8-dodecenyl acetate (E8-12:Ac) and (Z)-8-dodecenyl alcohol (Z8-12:OH) (**Figure 6A**). rGmolOBP7 had the strongest binding capacity to pear ester [Ethyl (E, Z)-2,4-decadienate] (with K<sup>i</sup> value of 2.52µM) in various volatiles emitted from peach shoots and pear fruits (**Figure 6D**). Additionally, rGmolOBP7 showed pronounced binding affinities with lauraldehyde and α-Ocimene with K<sup>i</sup> values of 4.31 and 7.75µM, respectively, (**Figures 6B,E**). rGmolOBP7 displayed intermediate binding affinities to some other aldehydes, alcohols, esters, and terpenes and nitriles with K<sup>i</sup> values of 9.81 to 19.91µM (**Figure 6**).

## Effect of RNAi Treatment on the Expression Level of *GmolOBP*7

Pilot experiments showed low eclosion rates when the pupae were injected with 138 nL of water at all putative conjunctiva (<30%). The emergence rates of the pupae injected with 39 and 69 nL of water at different candidate conjunctiva ranged from 78.6 to 84.2%. Thus, the conjunctivum between the prothorax and mesothorax was selected as the appropriate injection site, and 69 nL was selected as the appropriate dosage.

qRT-PCR analysis revealed that the transcription levels of dsGFP-injected moths had no significant differences compared to the non-injected male and female moths. The transcript levels of GmolOBP7 decreased to 52.57% (with 1-d eclosion) and 67.68% (with 2-d eclosion) in GmolOBP7-dsRNA-injected males compared to that in the dsGFP-treated and non-treated controls (**Figure 7A**), and decreased to 59.50% (with 1-d eclosion) and 77.17% (with 2-d eclosion) in GmolOBP7-dsRNA-injected females compared to that in the controls (**Figure 7B**). The transcript levels of GmolOBP7 in dsRNA-treated moths were increased to normal values after 3-days of eclosion. Thus, 1-dayold adult moths were selected for subsequent EAG assays.

## Electrophysiological Experiments

The EAG response values of dsGFP-treated moths to nine tested stimulants had no significant differences to the non-injected male and female moths (**Figure 8**). The t-tests showed that the responses of both female and male moths, to pear ester were significantly reduced (P < 0.05) after injection with GmolOBP7 dsRNA, and the response value of male moths to 12:OH was also significantly decreased. However, the response to (Z)- 8-dodecenyl acetate, (E)-8-dodecenyl acetate, (Z)-8-dodecenyl alcohol, (Z)-3-hexenyl acetate, lauraldehyde, α-pinene, and αocimene was not significantly different between dsRNA-treated moths and the non-injected control.

## DISCUSSION

The OBP family genes are composed of many highly differentiated subfamily genes, the olfactory function of those OBPs highly-expressed in the antennae, such as GOBPs and PBPs, has been studied extensively (Zhou et al., 2009; Yin et al., 2012; Khuhro et al., 2017). However, the ligand-binding capacities of OBPs which had no antenna-specific expression or were lowly-expressed in antennae, remains poorly understood. We identified 26 OBPs from the antennal transcriptome of G. molesta (Li et al., 2015), the antenna-highly-expressed OBPs (PBP1-3, GOBP1-2, OBP8, OBP11, and OBP15) all with their preferred odorant ligands, such as GmolPBP2, which can bind specifically to the major sex pheromone components Z8-12:Ac and E8-12:Ac (Song et al., 2014), GmolGOBP1 have strong binding affinities to the major sex pheromone component Z8- 12:OH and plant volatile decane (Li et al., 2016a), while hexanal was the preferred ligand of GmolOBP15 (Li et al., 2016b). We speculated that the relatively low-expression antennal OBPs may play a role in capturing and transporting the specific compounds of the host plant volatiles. The RPKM (reads per kilobase per million mapped reads) value of GmolOBP7 ranked 18th in 28 OBPs (Li et al., 2015), and can be expressed in soluble forms in a prokaryotic system. Therefore, we selected GmolOBP7

to evaluate its role in perceiving and recognizing the trace components emitted from peach shoots and pear fruits.

The expression profiles of olfactory-related genes in different tissues and sexes can provide clues to understand their physiological function (Ju et al., 2014). Numerous experiments have revealed that the antennae-enriched OBPs play an important role in detecting sex pheromones and host plant compounds (Sun et al., 2014; Yang et al., 2016; Khuhro et al., 2017). GmolOBP7 was expressed at relatively high levels in the antennae compared to other tissues, and might have potential functions in olfactory chemoreception. G. molesta reached the peak of mating after emergence (2- to 3-days) after which the flourishing period of oviposition of 3- to 5-day-old female adults occurred. The transcript levels of GmolOBP7 were slightly higher in 3-day-old male adults than in females of the same age, and the expressed levels were enhanced slightly in 3-dayold female adults. These expression characteristics implied the GmolOBP7 may be involved in the detection of sex pheromones and host-plant volatiles. In addition to antennae, GmolOBP7 was also abundantly expressed in the male wings of G. molesta, while similar expression profiles were found in BodoOBP17 from Bradysia odoriphaga (Zhao et al., 2018), MsepOBP19 from

FIGURE 3 | Expression profiles of *GmolOBP*7 in different tissues (A) and developmental stages (B) of male and female moths. An, antennae; He, heads; Th, thoraces; Ab, abdomens; Le, Legs; Wi, Wings; 1st, first-instar larvae; 2nd, second-instar larvae; 3rd, third-instar larvae; 4th, fourth-instar larvae; 5th, fifth-instar larvae; Pup, prepupae; Later Pup, 5-d-old pupae; 1♂, 1-d-old adult males; 1♀, 1-d-old adult females; 3♂,3-d-old adult males; 3♀, 3-d-old adult females; 5♂, 5-d-old adult males; 5♀, 5-d-old adult females. Different lowercase and capital letters indicate significantly different expression levels among different tissues of female and male, respectively (Tukey's test, α = 0.05). Asterisks indicate significant different expression levels of GmolOBP7 between two sexes in the same tissue (Independent *t*-test, α = 0.05).

Mythimna separata (Chang et al., 2017), AmalOBP8 from Agrilus mali (Cui et al., 2018), and AlucOBP6 from Apolygus lucorum (Hua et al., 2012). The chemoreception sensilla have been found on the wings of A. mali, as well as the taste organ and taste bristles and were also located on the wings of Drosophila melanogaster (Galindo and Smith, 2014). We speculated that the GmolOBP7 may play an important role in olfactory or gustatory perception, and further studies with non-volatile secondary metabolites of host plants are needed to verify this.

G. molesta thrives mainly on plants of the rosaceae family, and the peach and pear are considered the optimal host plants (Rice et al., 1972; Rajapakse et al., 2006). Plant volatiles serve as olfactory cues for G. molesta orientation, and guide the adults to switch from peach orchards to pear orchards during the growing season (Zhao et al., 1989; Najar-Rodriguez et al., 2013). We selected four sex pheromone components and 31 potential host-plant volatiles or its analogs, to determine the binding characteristics of rGmolOBP7. The sex pheromone components have been identified and widely used in the sexual trapping of male G. molesta (Cardé et al., 1975, 1979; Reinke et al., 2014). The tested volatiles are known to be emitted from peach shoots and pear fruits. EAG studies on G. molesta have shown

that some saturated and unsaturated volatile components of aldehydes, alcohols, acetate esters, terpenes and benzonitriles can effectively elicit responses from the antennal lobes of adult moths (Natale et al., 2003; Piñero and Dorn, 2009; Lu et al., 2015). Behavior response assays also indicated that the individual volatile component or mixture of several volatile compounds caused obvious attraction in adult moths (Natale et al., 2004; Piñero et al., 2008; Il'ichev et al., 2009; Yu et al., 2015).

The binding assays showed that GmolOBP7 has broad binding activities to various ligands including aldehydes, alcohols, esters, terpenoids and nitriles compounds. Pear ester, lauraldehyde, and dodecanol were the first three strongest ligands that bound to rGmolOBP7. Previous reports confirmed that the minor sex pheromone component 12:OH only elicited a weak EAG response to male antennae of G.moletsa. Its main function is a synergist attractant, that increases the frequency of male landing and is a stimulus that induces mating behavior when the male and female are close to each other, or when the male is close to pheromone lures (Cardé et al., 1975, 1979). EAG responses of GmolOBP7-dsRNA-treated males to 12:OH were significantly reduced compared with GFP-dsRNA-injected and non-injected controls. The simplest explanation is that GmolOBP7 may be involved in the perception of the sex pheromone 12:OH, and a behavioral response test of GmolOBP7 daRNA-treated to 12:OH is required to confirm this in future studies. Pear ester (Ethyl (E,Z)-2,4-decadienoate) belongs to a volatile derived from pear fruits and is widely applied in trapping female codling moth, Cydia pomonella, which is a closely related species of G. molesta (Vanessa et al., 2008). Pear ester exhibited the strongest binding affinity with GmolOBP7, and the EAG response values of dsRNA-treated males and females to pear ester were significantly decreased. GmolOBP7 may play the same role in perception of pear ester in male and female moths. rGmolOBP7 showed strong binding ability to lauraldehyde, but the EAG responses of dsRNA-treated male and female moths, to this compound, were not significantly different compared to non-injected controls. OBPs have a binding pocket formed by a six-α-helix fold, and usually have similar binding affinities to the ligand with the same structure and size. For example, Locusta migratoria LmigOBP1 binds to pentadecanol (C15), 2-pentadecanone (C15) and ethyl tridecanoate (C15) (Jiang et al., 2009), Bombyx mori BmorGOBP2 binds to (10E,12Z) hexadecadien-1-ol (bombykol) and (10E,12Z)-hexadecadienal (bombykal) (Zhou et al., 2009), Loxostege sticticalis LstiGOBP2 binds to 1-hexanol and 1-hexanal (Yin et al., 2012). We speculated that GmolOBP7 bound to lauraldehyde because of its size. Similar to pear ester and 12:OH, the lauraldehyde is also a derivative of a linear aliphatic hydrocarbon with 12 carbon atoms in the main chain. The molecular size of these three compounds are similar.

The binding assays were performed as recombinant OBPs expressed in vitro and the binding of OBPs with the ligands are affected by the shape and amino acid residues of the binding pocket of proteins, as well as the carbon-chain lengths, functional groups, isomers, and C = C bonds of ligands (Sandler et al., 2000; Mohanty et al., 2004; Wogulis et al., 2006; Li et al., 2008; Christina et al., 2017). OBPs may bind to many tested ligands

FIGURE 6 | Binding curves of recombinant GmolOBP7 to a series of tested ligands. (A) sex pheromones; (B) aldehydes; (C) alcohols; (D) esters; (E) terpenes; (F) nitriles. The protein was diluted to a fixed concentration of 2µM and then titrated with 1 mM of each competing ligand to a concentration of 0–14µM for sex pheromones and 0–35µM for host-plant volatiles. Fluorescence intensities are displayed as the percentage of the initial fluorescence. The calculated dissociation constants for all the ligands are listed in Table 2.

of GmolOBP7 between dsRNA-treated moths and non-injected moths (independent *t*-test, α =0.05).

with similar structures or sizes. Whether the odorants with strong binding activity to OBPs play a role in chemoreception such as mating and host selecting in insects, still needs to be verified by electrophysiological and behavioral assays. The methods of the RNAi combined with an EAG assay is an effective way to verify whether the binding-active odorants can be recognized by insects (Zhang et al., 2017). We found that GmolOBP7 exhibited binding activities in 21 of 35 tested ligands. The EAG assays preliminary revealed that GmolOBP7 may be involved in the detection of 12:OH and pear ester, however,

whether GmolOBP7 participates in the perception of other remaining binding-active odorants, requires further functional verification.

#### AUTHOR CONTRIBUTIONS

X-LC, G-WL, X-LX, and J-XW conceived and designed the experimental plan. X-LC and G-WL performed the experiments. X-LC, J-XW, and X-LX analyzed and processed the data. X-LC wrote the paper. All authors read and agreed to publish this paper.

## ACKNOWLEDGMENTS

This study was supported and funded by the National Natural Science Foundation of China (Grant No. 31272043; 31860506), National Key Research and Development Program of China

#### REFERENCES


(Grant No. 2018YFD0201400), and the Project of Industry-University-Research Collaboration of Yan'an University in 2017 (Grant No. 2017cxy04). We appreciate Prof. Deguang Liu (Key Laboratory of Applied Entomology, Northwest A&F University, China) for their suggestions on the previous versions of our manuscript.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01762/full#supplementary-material

Figure S1 | Nucleotide sequence and deduced aa sequence of GmolOBP7 in *Grapholita molesta*. The initiation and termination codons are indicated in boxes. The six conserved cysteines are marked by a circle with a blue background.

Figure S2 | Standard curves and melting curves of reference and target genes in qRT-PCR. (A,C,E) represents the standard curves of *Gmol*β*-actin*, *GmolEF1-*α, and *GmolOBP*7, respectively. (B,D,F) were the melting curves of *Gmol*β*-actin*, and *GmolEF1-*α, and *GmolOBP*7, respectively.


fruit moth, Cydia molesta, in the laboratory. J. Appl. Entomol. 128, 22–27. doi: 10.1046/j.1439-0418.2003.00802.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Chen, Li, Xu and Wu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Odorant Binding Proteins of the Desert Locust Schistocerca gregaria (Orthoptera, Acrididae): Topographic Expression Patterns in the Antennae

Xingcong Jiang<sup>1</sup> \*, Miriam Ryl<sup>1</sup> , Jürgen Krieger<sup>2</sup> , Heinz Breer<sup>1</sup> and Pablo Pregitzer<sup>1</sup> \*

1 Institute of Physiology, University of Hohenheim, Stuttgart, Germany, <sup>2</sup> Department of Animal Physiology, Institute of Biology/Zoology, Martin Luther University Halle-Wittenberg, Halle, Germany

#### Edited by:

Shuang-Lin Dong, Nanjing Agricultural University, China

#### Reviewed by:

Paolo Pelosi, Università degli Studi di Pisa, Italy Jin Zhang, Max Planck Institute for Chemical Ecology (MPG), Germany

#### \*Correspondence:

Xingcong Jiang jiangxingcong@126.com Pablo Pregitzer p\_pregitzer@uni-hohenheim.de

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 28 February 2018 Accepted: 04 April 2018 Published: 17 April 2018

#### Citation:

Jiang X, Ryl M, Krieger J, Breer H and Pregitzer P (2018) Odorant Binding Proteins of the Desert Locust Schistocerca gregaria (Orthoptera, Acrididae): Topographic Expression Patterns in the Antennae. Front. Physiol. 9:417. doi: 10.3389/fphys.2018.00417 Odorant binding proteins (OBPs) enriched in the sensillum lymph are instrumental in facilitating the transfer of odorous molecules to the responsive receptors. In Orthopteran locust species, an in-depth understanding of this important soluble protein family is still elusive. In a previous study, we have demonstrated that the repertoire of locust OBPs can be divided into four major clades (I–IV) on the phylogenetic scale and for representatives of subfamily I-A and II-A a distinct sensilla-specific expression pattern was determined. In this study, by focusing on a representative locust species, the desert locust Schistocerca gregaria, we have explored the antennal topographic expression for representative OBPs of other subfamilies. First, subtypes of subfamily III-A and III-B were exclusively found in sensilla chaetica. Then, a similar expression pattern in this sensillum type was observed for subfamily I-B subtypes, but with a distinct OBP that was expressed in sensilla coeloconica additionally. Moreover, the atypical OBP subtype from subfamily IV-A was expressed in a subpopulation of sensilla coeloconica. Last, the plus-C type-B OBP subtype from subfamily IV-B seems to be associated with all four antennal sensillum types. These results profile diversified sensilla-specific expression patterns of the desert locust OBPs from different subfamilies and complex co-localization phenotypes of distinct OBP subtypes in defined sensilla, which provide informative clues concerning their possible functional mode as well as a potential interplay among OBP partners within a sensillum.

Keywords: locust, Schistocerca gregaria, odorant binding protein, sensilla, topographic expression

## INTRODUCTION

Insects utilize hair-like cuticle appendages, so called sensilla, to receive environmental olfactory signals (Steinbrecht, 1996; Hansson and Stensmyr, 2011; Suh et al., 2014). Hydrophobic odorous molecules have to travel through the aqueous sensillum lymph before reaching the receptors residing in the chemosensory membrane of olfactory neurons in the antennae (Vogt et al., 1999; Leal, 2013; Suh et al., 2014). This passage is supposed to be facilitated by odorant binding proteins (OBPs) in the sensillum lymph, an important soluble protein family that is capable to accommodate and transfer odorant molecules (Vogt and Riddiford, 1981; Pelosi et al., 2006, 2014;

**237**

Vieira and Rozas, 2011). OBPs are short polypeptides of approximately 110–200 amino acids that fold into a globular shape forming an interior binding cavity, where the interaction with odorous molecules takes place (Sandler et al., 2000; Tegoni et al., 2004). The sequence of classic OBPs is characterized by six conserved cysteine (C) residues, a hall mark of classic OBPs; plus-C or minus-C OBPs are categorized with more or less than six conserved C-residues (Xu et al., 2003; Zhou et al., 2004; Foret and Maleszka, 2006; Vieira and Rozas, 2011). OBPs are produced by auxiliary cells which envelope the sensory neurons by their extended processes. The enrichment of OBPs in the sensillum types that respond to olfactory cues has been reported for many insect species (Pelosi et al., 2014, 2017). Beyond the olfactory sensilla, OBP expression has also been found in the sensilla that are seemingly dedicated to gustatory cues (Galindo and Smith, 2001; Jeong et al., 2013). Incidentally, besides the sensilla-specific expression in the chemosensory organs, like the antennae, OBPs are also expressed in other tissues of which the functional connotations seem to be less associated with chemical communication (Pelosi et al., 2017).

Schistocerca gregaria, the desert locust, represents a model organism of the Orthopteran order, which emerged much earlier than the Lepidopteran and Dipteran orders on the evolutionary scale (Wheeler et al., 2001; Vogt et al., 2015). Locusts are characterized by a hemimetabolous life circle and a population density dependent behavioral plasticity, which involves the perception of behavioral relevant semiochemicals (Pener and Yerushalmi, 1998; Hassanali et al., 2005; Guo et al., 2011; Wang and Kang, 2014). For locust species an in-depth understanding of the OBP family from either molecular or cellular perspective is still elusive (Ban et al., 2003; Jin et al., 2005; Jiang et al., 2009; Xu et al., 2009; Yu et al., 2009). Previously, we have conducted a comprehensive sequence analysis of the OBP families from Schistocerca gregaria and three other locust species which classifies locust OBPs into several categories, e.g., classic, plus-C type-A, plus-C type-B, minus-C and atypical OBPs. Based on the phylogenetic relationship locust OBPs reside within four major phylogenetic clades. Concentrating on the two OBP subfamilies I-A and II-A, which comprise the classic OBP subtypes, we have found a characteristic sensilla-specific expression pattern for the desert locust OBP representatives in the antennae (Jiang et al., 2017). In the present study, we set out to explore the antennal topographic expression of desert locust OBPs from the remaining subfamilies on the phylogenetic tree.

## MATERIALS AND METHODS

#### Animals and Tissue Collection

The desert locust Schistocerca gregaria reared on the gregarious phase were purchased from Bugs-International GmbH (Irsingen/Unterfeld, Germany). Antennae of adult male and adult female were dissected using autoclaved surgical scissors and were immediately frozen in liquid nitrogen. Tissues were stored at −70◦C before subsequent RNA extraction.

## RNA Extraction and Reverse Transcription PCR (RT-PCR)

Total RNA was extracted from the frozen tissues using TRIzol reagent (Invitrogen) following the protocol recommended by the manufacturer. The poly (A)<sup>+</sup> RNA was purified from 100 µg of total RNA using oligo (dT)<sup>25</sup> magnetic dynabeads (Invitrogen) conforming to the recommendation of the supplier. The generated mRNA was reverse transcribed to cDNA in a total volume of 20 µl employing SuperScriptTM III Reverse Transcriptase (Invitrogen). PCR conditions used in RT-PCR experiments were: 94◦C for 1 min 40 s, then 20 cycles with 94◦C for 30 s, 60◦C for 30 s and 72◦C for 2 min, with a reduction in the annealing temperature by 0.5◦C per cycle, which was followed by a further cycles (20 times) on the condition of the last cycling step (annealing temperature was 50◦C) and a final extension step for 7 min at 72◦C. The sense (s) and antisense (as) primer pairs used for amplification of the desert locust OBP coding sequences were:

OBP2 s, atggccagccattgccacgccacc OBP2 as, ttctccggatttcctaaactccgc OBP3 s, atgctgctggcagcccccgcaaagg OBP3 as, ctttttcctgatcaagcatccacc OBP4 s, cctgtggcgacacttggtggccg OBP4 as, gcctttagccatcatcccctt OBP7 s, cgatgtgcttcgtcggtgggtgat OBP7 as, acgtcgttctcgtcggactctgga OBP8 s, agactcgccaacccgccaca OBP8 as, ttctgacggggcgtgtggga OBP9 s, gccacagtccggtgcagcat OBP9 as, aatctggtcgctgacgcact OBP12 s, acaactcttgcagccatgaagtgg OBP12 as, tccacttcttgttcccatactggt OBP13 s, gagctgaggtaatgaagagggtca OBP13 as, cctgcacattcagatccaagcagc

The primer pairs against other desert locust OBP subtypes were given in (Jiang et al., 2017).

## Synthesis of Riboprobes for in Situ Hybridization

PCR products of the desert locust OBP coding sequences were sequenced and then cloned into pGEM-T vectors (Invitrogen) for the subsequent in vitro transcription. The linearized pGEM-T vectors consisting of desert locust OBP coding sequences were utilized to synthesize both sense and antisense riboprobes labeled with digoxigenin (Dig) or biotin (Bio) using the T7/SP6 RNA transcription system (Roche, Germany). The synthesis procedure stringently followed the protocol provided by the manufacturer.

## In Situ Hybridization

Antennae of adult Schistocerca gregaria were dissected and embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek Europe, Netherlands). Cryosections with a 12 µm-thickness were thaw mounted on SuperFrost Plus slides (Menzel-Gläser, Braunschweig, Germany) at −21◦C (Jung CM300 cryostat). RNA In situ hybridization was performed as previously reported (Yang et al., 2012; Guo et al., 2013; Jiang et al., 2016, 2017). In brief,

the cryosections were firstly fixed (4% paraformaldehyde in 0.1 M NaHCO3, pH 9.5) at 4◦C for 22 min, followed by a series of treatments at room temperature: a wash for 1 min in PBS (phosphate buffered saline = 0.85% NaCl, 1.4 mM KH2PO4, 8 mM Na2HPO4, pH 7.1), an incubation for 10 min in 0.2 M HCl, another wash for 1 min in PBS, an incubation for 10 min in acetylation solution (0.25% acetic anhydride freshly added in 0.1 M triethanolamine) and washes for three times in PBS (3 min each). Afterward, the sections were pre-hybridized for 1 h at 60◦C bathed in hybridization buffer (50% formamide, 5x SSC, 50 µg/ml heparin, and 0.1% Tween-20). A volume of 150 µl hybridization solution containing experiment riboprobes in hybridization buffer was evenly applied onto the tissue section. A coverslip was placed on top and slides were incubated in a moister box at 60◦C overnight (18–20 h). After hybridization, slides were washed twice for 30 min in 0.1x SSC at 60◦C, then each slide was treated with 1 ml 1% blocking reagent (Roche) for 35 min at room temperature.

Visualization of Dig-labeled riboprobe hybridizations was achieved by using an anti-Dig alkaline phosphatase (AP) conjugated antibody (1:500, Roche) and NBT/BCIP as substrates. Antennal sections were analyzed on a Zeiss Axioskope2 microscope (Zeiss, Oberkochen, Germany) equipped with Axiovision software. For two-color fluorescent in situ hybridization visualization of hybridized riboprobes was performed by using an anti-Dig AP-conjugated antibody in combination with HNPP/Fast Red (Roche) for Dig-labeled probes and an streptavidin horse radish peroxidase-conjugate together with fluorescein-tyramides as substrate (TSA kit, Perkin Elmer, Waltham, MA, United States) for biotin-labeled probes. Tissue sections in two-color FISH experiments were analyzed with a Zeiss LSM510 Meta laser scanning microscope (Zeiss, Oberkochen, Germany), and the acquired confocal images stacks were processed by ZEN 2009 software. The images presented in this paper integrate the projections of a series of optical planes selected from continuous confocal image stacks. For clear data presentation, images were only adjusted in brightness and contrast. It is noted that the images obtained via the two-color FISH approach always contained the cuticle unspecifically stained, most likely due to the intrinsic fluorescence. To clarify

FIGURE 1 | Sensilla chaetica express OBP subtypes of two phylogenetic clades. The schematic diagram of the phylogenetic tree (left in A,B) was adapted from Jiang et al. (2017) where OBP families of four locust species have been analyzed. The specific S. gregaria OBP subtypes studied in this analysis were indicated. A detail classification of different subfamilies is illustrated in Supplementary Figure S1. Topographic expression of OBPs was visualized by using antisense riboprobes specifically targeting distinct OBP subtypes in conjunction with chromogenic in situ hybridization (ISH). (A,B) Visualization of the labeled cells expressing distinct OBP subtypes of subfamily III-A, III-B, I-A, and I-B in four morphological types of antennal sensilla. Ba, sensilla basiconica; Tr, sensilla trichodea; Ch, sensilla chaetica; Co, sensilla coeloconica. The visible labeled structures are denoted by black arrows. (C) Visualization of the cells expressing distinct OBP subtypes from different subfamilies on the tip of the antennae. Notably, sensilla chaetica are exclusively enriched on the antennal tip (Ochieng et al., 1998). The area of the antennal tip is indicated by a box with a dashed line. The visible cell clusters are denoted by black arrows, and in some images the interface between the cuticle and cellular layer is depicted as a white dashed line. The subfamily to which a distinct OBP subtype belongs is annotated below the images. Scale bars, 20 µm.

the specific fluorescent labeling, a dashed line was added to indicate the interface between the cuticle and the cellular layers. Antennal sections of both male and female were analyzed under the same experimental conditions and were tested with each generated riboprobes. There were no discernible gender dependent differences regarding to the labeling intensity as well as the labeling pattern. Therefore, only the images acquired from male antenna sections were presented in this paper.

#### RESULTS

#### Topographic Expression Patterns of OBP Subtypes From Clade I and III

A previously performed phylogenetic analysis of OBPs from four locust species revealed that the locust OBP family can be divided into four major clades consisting of three conserved subfamilies. For the two subfamilies I-A and II-A, which both comprise classic OBP subtypes, we found that the representative I-A subtypes are expressed in sensilla basiconica and sensilla trichodea, whereas the representative II-A subtypes are expressed in sensilla coeloconica (Jiang et al., 2017). In this study, we concentrated on the conserved subfamily III-A, which includes the plus-C type-A OBP subtypes that share only low sequence identities with the classic OBP subtypes. In order to explore their sensilla-specific expression pattern, we adopted the strategy of mRNA in situ hybridization and assessed the expression of OBP4, a representative subtype of subfamily III-A, in the four morphologically distinguishable types of antennal sensilla. The results of these approaches revealed a discernible labeling of OBP4 expressing cells in sensilla chaetica; no labeling was visible in any of the other three sensillum types (**Figure 1A**). Apart from the subfamily III-A, clade III also comprises subfamily III-B, which includes the classic OBP subtype OBP8 and its orthologs. Analysis of the expression pattern revealed that OBP8 positive cells were also exclusively enriched in sensilla chaetica, thus resembling the plus-C type-A subtype OBP4 (**Figure 1A**). Together, these results imply that OBP subtypes of the clade III are specifically expressed in sensilla chaetica and thus deviate from the distribution of OBP subtypes from subfamilies I-A and II-A (Jiang et al., 2017).

In view of a clade-specific spatial expression pattern as seen for clade III (see above) it is interesting to note that clade I comprises, besides the conserved subfamily I-A, the more divergent subfamily I-B (**Supplementary Figure S1**). Since representatives of subfamily I-A were found to be restricted to

sensilla basiconica and trichodea (**Figure 1B**) (Jiang et al., 2017), the question arises, whether OBPs of subfamily I-B may also be expressed in the same sensillum types. To scrutinize this notion, we have analyzed OBP2 and OBP7, the two subtypes in subfamily I-B. The results are depicted in **Figure 1B** and indicate that labeling for OBP2 and OBP7 was neither found in sensilla basiconica nor in sensilla trichodea; however, the labeling was present in sensilla chaetica and for OBP2 the labeled cells were concomitantly visible in sensilla coeloconica (**Figure 1B**). These data indicate that the topographic distribution of subfamily I-B OBPs clearly deviate from that of their counterparts of subfamily I-A and demonstrate that there is no clade-specific spatial expression pattern for members of clade I.

Previous anatomical studies have shown that sensilla chaetica are highly enriched at the tip of the antennae, a region with relatively few of the other three sensillum types (Ochieng et al., 1998). This spatial segregation of sensilla chaetica allows a more detailed analysis of the four identified OBP subtypes in this sensillum type. As shown in **Figure 1C**, numerous labeled cells were visualized using the probes for OBP4 (subfamily III-A), OBP8 (subfamily III-B) as well as OBP2 and OBP7 (subfamily I-B). In contrast, with the riboprobes for OBP subtypes that are specifically expressed in other sensillum types, such as OBP5 (subfamily I-A) and OBP11 (subfamily II-A), no discernible labeling was found at the antennal tip (**Figure 1C**).

#### Co-localization of OBP Subtypes From Different Subfamilies in Sensilla Chaetica

Since the four OBP subtypes reside in two different phylogenetic clades, we ask whether the different OBP subtypes are present in the same set of cells or in distinct cell populations of sensilla chaetica. To approach this question, we have generated either DIG- or BIO-labeled riboprobes for each OBP subtype and by means of two-color FISH analysis we have visualized the relative topographic localization of the labeled cells (**Figure 2**). In a first step, we have analyzed the subtypes from the same phylogenetic clade. For the two subtypes from clade III, OBP4 and OBP8, a widely overlapped labeling was found indicating that they were co-localized in the same set of cells in many, if not all, inspected sensilla chaetica (**Figure 2A**). Analysis for the two subtypes from subfamily I-B, OBP2 and OBP7, also revealed a largely overlapped labeling (**Figure 2A**). These results suggest that within clade III and subfamily I-B OBP subtypes are generally expressed in the same set of cells in sensilla chaetica. In a next step, we explored whether OBP subtypes from different clades may either be expressed in the same or a different set of cells. For the member of subfamily III-A (OBP4) and the members of subfamily I-B (OBP2 and OBP7) a largely overlapping labeling was observed (**Figure 2B**). However, for the member of subfamily III-B (OBP8) and the members of subfamily I-B (OBP2 and OBP7) no labeling overlap was found (**Figure 2C**). While labeling for OBP2 and OBP8 was found in different sets of cells of the same sensillum chaeticum, interestingly, OBP7 seemed to be present in the cells of distinct sensilla chaetica which differ from sensilla with OBP8-positive cells (**Figure 2C**). These results emphasize the complex co-localization relationship among OBP2, OBP4, and OBP8. The notion that OBP4 and OBP8 may be separately expressed in a subset of sensilla chaetica was confirmed upon a comprehensive inspection of the labeling for OBP4 and OBP8 (**Supplementary Figure S2**), indicating a broader expression scope for OBP4 in certain sensilla chaetica. In sum, the results indicate that sensilla chaetica express OBP subtypes from more than one phylogenetic clade, and co-localization of the OBP subtypes in distinct sensilla subtypes occurs in a combinatorial mode.

#### OBP2, Member of Subfamily I-B, Is Expressed in Sensilla Coeloconica and Chaetica

The results depicted in **Figure 1** indicate that OBP2, a subtype of subfamily I-B, may not only be expressed in sensilla chaetica (see above) but also in sensilla coeloconica. To substantiate the observation that OBP2 is in fact expressed in sensilla coeloconica, we utilized IR8a, the co-receptor of divergent IRs (Abuin et al., 2011; Guo et al., 2013), as a specific marker of sensory neurons housed in sensilla coeloconica. The results of double labeling experiments indicate that labeled OBP2 cells are tightly surrounding IR8a-positive cells in sensilla coeloconica (**Figure 3**).

FIGURE 3 | OBP2 from subfamily I-B is expressed in sensilla coeloconica and sensilla chaetica. The relative localization of OBP2 and the marker genes indicating expression in sensilla coeloconica (co) was analyzed by utilizing antisense riboprobes targeting specific molecular elements in conjunction with two-color FISH. (Upper) OBP2 expressing cells surround a sensory neuron positive for IR8a, a specific molecular marker for sensilla coeloconica. (Middle and lower ) OBP10 and OBP14 from the subfamily II-A are specifically expressed in sensilla coeloconica and are employed to mark two different sets of auxiliary cells in this sensillum type (Jiang et al., 2017). The interface between the cuticle and the cellular layer is denoted by a white dashed line. Distinct cell clusters positive for the DIG-labeled OBP2 probe (red) are encircled by white dashed lines. The position of these cell clusters is also indicated on the images showing the merged red and green fluorescence channels. Scale bars, 20 µm.

Given that in sensilla coeloconica OBP subtypes of subfamily II-A are specifically expressed, the question arises as to whether OBP2, a member of subfamily I-B, may be co-expressed with OBP subtypes of subfamily II-A. As representatives for subfamily II-A OBP10 and OBP14 were investigated. The results depicted in **Figure 3** indicate that the labeling for OBP2 indeed overlapped with that for the subfamily II-A representatives, indicating that in a set of sensilla coeloconica OBP subtypes from subfamily I-B and subfamily II-A coexist. Furthermore, the results confirm that OBP2 is in fact present in the two types of sensilla, sensilla coeloconica and sensilla chaetica.

#### Topographic Expression Pattern of an Atypical OBP Subtype From Subfamily IV-A

The atypical OBP subtypes converge onto the subfamily IV-A (**Supplementary Figure S1**) and are characterized by an extraordinary long span between C1 and C2 in comparison to the classic OBP subtypes (Jiang et al., 2017). This unique feature has raised the question whether atypical OBP subtypes may be expressed in specific sensillum types and/or in distinct cell populations. To approach this question, we have analyzed the expression pattern of OBP12, a subtype of subfamily IV-A. The results of labeling experiments are depicted in **Figure 4A** and indicate that OBP12 expressing cells were exclusively located in sensilla coeloconica. The sensilla specificity was subsequently confirmed by demonstrating the co-localization of OBP12 expressing cells and IR8a-positive cells in one sensillum coeloconicum (**Figure 4A**). Since OBPs of subfamily II-A are specifically expressed in sensilla coeloconica, we explored whether OBP12 may be co-localized with OBPs of subfamily II-A. Intriguingly, we found that the labeling for OBP12 cells did not overlap with the cells positive for OBP10 or OBP14 (**Figure 4B**), suggesting that OBP12 is expressed in a distinct subset of sensilla coeloconica.

It is yet unclear how many IR8a-positive neurons are surrounded by the auxiliary cells that express OBPs of subfamily

FIGURE 4 | An atypical OBP subtype pronounces a segregated subpopulation of sensilla coeloconica. (A) OBP12, an atypical OBP subtype residing in subfamily IV-A, is exclusively expressed in sensilla coeloconica (co). Upper panel: OBP12 expressing cells were analyzed in four morphological types of antennal sensilla using specific riboprobe by means of ISH. Labeled OBP12 cells were detected only in sensilla coeloconica and are indicated by a black arrow. Ba, sensilla basiconica; Tr, sensilla trichodea; Ch, sensilla chaetica; Co, sensilla coeloconica. Lower panel: A co-localization of OBP12 expressing cells and an IR8a-positive neuron in sensilla coeloconica was visualized by means of two-color FISH. (B) The labeling of OBP12-positive cells does not overlap with the labeling of cells expressing OBP10 and OBP14 from subfamily II-A. The interface between the cuticle and the cellular layer is depicted by a white dashed line. (C) Three OBP subtypes of subfamily II-A label the major population of auxiliary cells in sensilla coeloconica. The presented optical view was adopted from a distal antennal segment and presumably illustrates the typical association between IR8a neurons and subfamily II-A OBP cells. The utilized DIG-labeled probes representing the three ortholog groups comprised in subfamily II-A (Supplementary Figure S1) were generated by mixing the riboprobes against OBP10, OBP11, and OBP14, respectively, at a ratio of 1:1:1. Areas encircled by white dashed lines indicate IR8a neurons that are co-localized with auxiliary cells expressing the subfamily II-A OBPs in the same coeloconic sensillum. White arrows indicate those IR8a neurons that are presumably not associated with auxiliary cells expressing subfamily II-A OBPs. Scale bars, 20 µm.

II-A. To scrutinize this notion, double labeling experiments were performed with a probe for IR8a and a mix of riboprobes for OBP10, OBP11 and OBP14, which represent the three ortholog groups in subfamily II-A (**Supplementary Figure S1**). The results depicted in **Figure 4C** indicate that a considerable portion of IR8a-positive cells are engulfed by cells expressing OBPs of subfamily II-A (ovals in dash line). The remaining fraction of IR8a neurons seems to express non-II-A OBP subtypes, possibly OBP12. Together the results indicate that the atypical OBP subtype OBP12 is expressed in a segregated population of sensilla coeloconica.

#### Topographic Expression and Sensillum-Association of a Plus-C Type-B OBP Subtype

We have previously distinguished two categories of the plus-C OBPs based on the distinct conserved-C-patterns (Jiang et al., 2017). While the type-A OBP subtypes are grouped into the subfamily III-A, the type-B OBP subtypes are grouped into the subfamily IV-B (**Supplementary Figure S1**). Whereas type-A OBPs are expressed in sensilla chaetica (**Figure 1**), the expression pattern of type-B OBP subtypes is unclear. It is possible that the type-B OBPs share the sensilla specificity either with their close relatives in subfamily IV-A, e.g., OBP12, or with their type-A counterparts in subfamily III-A, e.g., OBP4. To approach this question, we have used a specific riboprobe for OBP9, a representative plus-C type-B subtype and assessed series of horizontal sections through the antennae. Upon an inspection of a deep anatomical plane close to the antennal nerve bundle, we found labeled structures for OBP9 which seemed to be less associated with a specific sensillum type, as typically found for the other OBP subtypes (**Figures 1**, **3**, **4**). Nevertheless, labeled cell bodies seemed to extend cytoplasmic processes which enclosed sensory neurons (**Figure 5A**). Interestingly, when we inspected an anatomical plane located closer to the cuticle, a more intense labeling was observed and a distinct nest-like labeling pattern for OBP9 emerged (**Figure 5B**).

The notion that OBP9 labeling seems to be associated with multiple sensillum types was scrutinized by analyzing a possible co-localization of OBP9 labeling with markers for distinct neuron types. In a first approach, Orco, the obligate co-receptor of ORs, was used to label the multiple sensory neurons in sensilla basiconica (Ochieng et al., 1998). It was found that OBP9 cells tightly surrounded the Orcopositive neuron clusters (**Figure 6**). Similarly, OR3 was used as a marker for sensilla trichodea and IR8a was used as a marker for sensilla coeloconica; it was observed that OBP9 labeling engulfed OR3- and IR8a- expressing neurons (**Figure 6**). OBP8 is considered to be specific for sensilla chaetica (**Figure 1**) and the results of double labeling experiments with OBP9 and OBP8 clearly indicated a co-localization (**Figure 6**). Together, these results indicate an association of the plus-C type-B OBP9 with all four antennal sensillum types.

FIGURE 5 | Topographic expression of the plus-C type-B OBP9 in the antennae. The topographic expression of OBP9 was analyzed by using a specific antisense riboprobe in conjunction with ISH. (A,B) Labeling of OBP9 expressing cells in two different anatomical planes of the antennae. OBP9 represents the plus-C type-B OBPs that are grouped into subfamily IV-B (diagrams, left lane). Two different horizontal planes are shown to visualize the OBP9 expression pattern: the first deep plane (A, middle lane, red dashed frame) penetrates into the central nerve bundle; the second superficial plane (B, middle lane, red dashed frame) is located between the cuticle and central nerve bundle. For each plane a selected area (magenta box, middle lane) of the analyzed section is shown at a higher magnification on the right. Black arrows indicate the visible cell bodies as well as their extended processes. The border between the cellular layer and the nerve bundle is depicted by a black dashed line. Tr, sensilla trichodea; Co, sensilla coeloconica; Ba, sensilla basiconica. Scale bars, 20 µm.

## DISCUSSION

Insects have evolved sensilla that are diversified in the external morphology as well as in the repertoire of molecular elements to act as versatile communication channels for environmental chemical signals (Hansson and Stensmyr, 2011; Leal, 2013; Suh et al., 2014). OBPs are considered to play an important role toward this task due to their capacity to accommodate and transfer odorous molecules. The present study, in conjunction with our previous work (Jiang et al., 2017), has concentrated on this important class of soluble proteins in the locust species Schistocerca gregaria, trying to decipher the principles how the multiple OBP subtypes are allocated among and within different sensillum types present on the locust antennae. The findings of this study revealed that subtypes of the desert locust OBP family display a diversified sensilla-specific expression profile and a

marker for auxiliary cells of sensilla chaetica (see Figure 1). Scale bars, 20 µm.

complex co-localization phenotype in defined sensilla (**Figure 7**). Uncovering the sensillar and cellular organization pattern of distinct locust OBP subtypes may allow a first glimpse on their putative functional role as well as their potential interplay with distinct co-partners.

Our results indicate that several OBP subtypes from two phylogenetic clades are expressed in sensilla chaetica (**Figure 1**). A plus-C type-A subtype together with three classic subtypes were found to be co-expressed in a set of sensilla chaetica (**Figure 2**); this scenario is reminiscent of what was previously reported for sensilla trichodea of Anopheles gambiae (Schultze et al., 2013). Sensilla chaetica are characterized by distinct structural features, such as a thick and poreless cuticle wall, an apical pore and relatively few dendrites (Ochieng et al., 1998; Zhou et al., 2009); consequently, sensilla chaetica are considered as relevant for the reception of gustatory tastants rather than odorants. For the

fruit fly this view was supported by extracellular recordings, calcium imaging and behavioral assays (Montell, 2009; Chen and Amrein, 2017; Scott, 2018). This view may also hold true for sensilla chaetica in locusts which are enriched on the tip of the antennae and palps (Blaney and Chapman, 1969; Ochieng et al., 1998) and are proposed with a receptive role of contact stimuli (Blaney, 1974, 1975; Saini et al., 1995). Thus, the presence of four OBP subtypes in sensilla chaetica on the tip of the antennae (**Figure 1**) suggests that these OBPs may be tuned to mediate the reception of gustatory stimuli. This view would be analogous to the finding for Drosophila melanogaster where OBP subtypes expressed in gustatory sensilla are involved in the reception of tastants (Jeong et al., 2013). This is further supported by a recent study demonstrating that knock-down of a sensilla chaeticaspecific OBP subtype in Locusta migratoria caused a reduced neuronal response to chemical stimuli (Zhang et al., 2017). This finding further supports the notion that OBPs are intimately involved in detecting chemical compounds via sensilla chaetica. Intriguingly, it has been reported that the sensilla chaetica of locust, as well as contact sensilla of other insect species, have a sensillum lymph cavity which is separated into an inner and outer compartment (Ochieng et al., 1998; Shanbhag et al., 2001; Zhou et al., 2009). In a recent study, the labeling for an OBP subtype in Locusta migratoria was mainly observed in the non-innervated outer lumen, but not in the inner sensillum lymph which baths the chemosensory dendrites (Yu et al., 2009); this observation has led to speculations of how the cognitive ligands may reach the chemosensory dendrites. The discovery that four distinct OBP subtypes are expressed in this sensillum type (**Figures 1**, **2**) opens the door for revisiting this aspect in more detail.

Distinct OBP subtypes from three phylogenetic clades were found to be expressed in sensilla coeloconica (**Figures 1**, **3**, **4**) (Jiang et al., 2017). Whereas OBP representatives from subfamily II-A (**Figure 4**) together with OBP2 (**Supplementary Figure S3**) were found in the majority of this sensillum type, the atypical OBP subtype OBP12 from subfamily IV-A was present in a subpopulation of sensilla coeloconica. This observation seems to coincide with a previous finding that apart from a receptive role for leaf odors and organic acids (Ochieng and Hansson, 1999), a subset of sensilla coeloconica in locusts appears to be responsive to hygro- or thermo- stimuli (Altner et al., 1981). Such a functional versatility of this sensillum type may be based on distinct sets of cells equipped with specific receptors in combination with appropriate co-partners, e.g., OBP12. Remarkably, the atypical OBP subtype OBP12 belongs to the OBP gene family OBP59a, which is conserved in many insect species, except in Hymenoptera (Vieira and Rozas, 2011). For Drosophila melanogaster it has recently been shown that OBP59a is specifically expressed in sensilla coeloconica (Larter et al., 2016), similar to its counterpart in the desert locust (**Figure 4**).

An unexpected finding of this study is the expression of OBP2 in two types of sensilla, sensilla coeloconica and sensilla chaetica (**Figures 1**, **3**). The two types of sensilla differ markedly in their external morphology and their functional implications

(Montell, 2009; Rytz et al., 2013; Joseph and Carlson, 2015; Scott, 2018). On the other hand, in both sensillum types some common chemosensory genes are expressed, most notably the ionotropic receptor type IR25a, one of the co-receptors of divergent IRs (Abuin et al., 2011; Guo et al., 2013). Exploring the functional mode of IR25a in Drosophila melanogaster has recently uncovered a multidimensional role for this receptor type (Rimal and Lee, 2018) and it is conceivable that such a versatile function may also be assigned to the OBPs. In fact, it has been proposed that OBPs may be involved in quite different functions (Pelosi et al., 2006, 2014, 2017). In this regard, the observation that OBP2 is always accompanied by a set of other OBP subtypes in a sensillum (**Figures 2**, **3**) may indicate that OBP2 operates in concert with other OBPs to fulfill the distinct functions conferred to the two types of sensilla.

One of the novel finding of this study was the discovery that the plus-C type-B subtype OBP9 is associated with the four antennal sensillum types. Although the functional implication of such a broad sensillum-association is unknown, one could imagine that OBP9, as an ubiquitous OBP, may contribute a general component for the interplay of co-localized OBP partners. Indeed, an interaction of OBP subtypes has been documented in mosquito species and the OBP complex showed a broader ligand spectrum (Qiao et al., 2011). This aspect may be of particular interest in view of the finding that in locust sensilla basiconica, with a large set of OR subtypes (Wang et al., 2015; Pregitzer et al., 2017), only a small set of OBPs is expressed (**Figure 7**). However, it can also not be excluded that OBP9 may be involved in quite different functions. In this context, it is interesting to note that in cockroach and honeybee, the chemosensory proteins, another important class of small soluble proteins, are involved in regulating tissue regeneration and embryonic development (Nomura et al., 1992; Maleszka et al., 2007; Cheng et al., 2015). Given such a broad sensillumassociation, OBP9 may be involved in some general processes, such as development and/or survival of the auxiliary cells.

#### AUTHOR CONTRIBUTIONS

HB, JK, XJ, and PP: current study conception. XJ and MR: experiments conduction and the data acquisition. HB, JK, XJ, and PP: results interpretation. XJ and PP: preliminary manuscript

#### REFERENCES


composition. HB and JK: refinement and approval of final manuscript.

#### FUNDING

XJ was funded by a grant from China Scholarship Council (CSC) with the grant number 201406350032.

#### ACKNOWLEDGMENTS

We thank Heidrun Froß for her excellent technical assistance. We also thank Prof. Jörg Strotmann and Dr. Patricia Widmayer for their constructive suggestions to improve the manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00417/full#supplementary-material

FIGURE S1 | Classification of different subfamilies of locust OBPs. The phylogenetic tree shown was adapted from a previous study analyzing phylogenetic relationship of OBP families from four locust species (Jiang et al., 2017). The branches colored in red, green, blue, and magenta represent the clade I, II, III, and IV, respectively. The classification of the subfamily I-A, II-A, and III-A was based on emergence of a higher bootstrap values on the inner divergent nodes, while other subfamilies were categorized by the emerging topologies. The subtypes belonging to desert locust OBPs were colored and denoted accordingly.

FIGURE S2 | A subset of sensilla chaetica selectively express OBP4 but not OBP8. Cells expressing the respective genes were visualized by using antisense riboprobes specifically targeting OBP4 and OBP8 and by means of two-color FISH. The position of cell clusters visualized by the DIG-labeled OBP4 probe (red) was delineated by dashed lines and is indicated in the images showing the OBP8 labeling and the merge of red and green fluorescence channels, respectively. Notably, no OBP8 labeling was detected. The interface between the cuticle and cellular layer is depicted by a white dashed line. Ch, sensilla chaetica; Ba, sensilla basiconica. Scale bar, 20 µm.

FIGURE S3 | OBP2 and OBP12 are expressed in different cells in sensilla coeloconica (co). Specific antisense riboprobes against OBP2 and OBP12 were used to visualize the expressing cells by means of two-color FISH. The interface between the cuticle and the cellular layer is depicted by a white dashed line. Scale bar, 20 µm.



mosquito, Anopheles gambiae. Insect Mol. Biol. 12, 549–560. doi: 10.1046/j. 1365-2583.2003.00440.x


mosquito Anopheles gambiae. Gene 327, 117–129. doi: 10.1016/j.gene.2003. 11.007

Zhou, S. H., Zhang, S. G., and Zhang, L. (2009). The chemosensilla on tarsi of locusta migratoria (Orthoptera: Acrididae): distribution, ultrastructure, expression of chemosensory proteins. J. Morphol. 270, 1356–1363. doi: 10.1002/ jmor.10763

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Jiang, Ryl, Krieger, Breer and Pregitzer. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antennal Protein Profile in Honeybees: Caste and Task Matter More Than Age

Immacolata Iovinella<sup>1</sup> , Federico Cappa<sup>1</sup> , Alessandro Cini1,2, Iacopo Petrocelli<sup>1</sup> , Rita Cervo<sup>1</sup> , Stefano Turillazzi<sup>1</sup> and Francesca R. Dani1,3 \*

<sup>1</sup> Department of Biology, Università degli Studi di Firenze, Florence, Italy, <sup>2</sup> Centre for Biodiversity and Environment Research, University College London, London, United Kingdom, <sup>3</sup> Mass Spectrometry Centre, Centro di Servizi di Spettrometria di Massa, Università degli Studi di Firenze, Florence, Italy

Reproductive and task partitioning in large colonies of social insects suggest that colony members belonging to different castes or performing different tasks during their life (polyethism) may produce specific semiochemicals and be differently sensitive to the variety of pheromones involved in intraspecific chemical communication. The main peripheral olfactory organs are the antennal chemosensilla, where the early olfactory processes take place. At this stage, members of two different families of soluble chemosensory proteins [odorant-binding proteins (OBPs) and chemosensory proteins (CSPs)] show a remarkable affinity for different odorants and act as carriers while a further family, the Niemann-Pick type C2 proteins (NPC2) may have a similar function, although this has not been fully demonstrated. Sensillar lymph also contains Odorant degrading enzymes (ODEs) which are involved in inactivation through degradation of the chemical signals, once the message is conveyed. Despite their importance in chemical communication, little is known about how proteins involved in peripheral olfaction and, more generally antennal proteins, differ in honeybees of different caste, task and age. Here, we investigate for the first time, using a shotgun proteomic approach, the antennal profile of honeybees of different castes (queens and workers) and workers performing different tasks (nurses, guards, and foragers) by controlling for the potential confounding effect of age. Regarding olfactory proteins, major differences were observed between queens and workers, some of which were found to be more abundant in queens (OBP3, OBP18, and NPC2-1) and others to be more abundant in workers (OBP15, OBP21, CSP1, and CSP3); while between workers performing different tasks, OBP14 was more abundant in nurses with respect to guards and foragers. Apart from proteins involved in olfaction, we have found that the antennal proteomes are mainly characterized by castes and tasks, while age has no effect on antennal protein profile. Among the main differences, the strong decrease in vitellogenins found in guards and foragers is not associated with age.

Keywords: Apis mellifera, nurses, guards, foragers, queens, olfaction, odorant-binding proteins, chemosensory proteins

#### Edited by:

Shuang-Lin Dong, Nanjing Agricultural University, China

#### Reviewed by:

Stephanie Biergans, University of Toronto, Canada Xianhui Wang, Institute of Zoology (CAS), China

\*Correspondence: Francesca R. Dani francescaromana.dani@unifi.it

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 18 April 2018 Accepted: 28 May 2018 Published: 20 June 2018

#### Citation:

Iovinella I, Cappa F, Cini A, Petrocelli I, Cervo R, Turillazzi S and Dani FR (2018) Antennal Protein Profile in Honeybees: Caste and Task Matter More Than Age. Front. Physiol. 9:748. doi: 10.3389/fphys.2018.00748

## INTRODUCTION

fphys-09-00748 June 19, 2018 Time: 16:7 # 2

Colony organization and task partitioning in social insects largely depends on chemical communication, particularly in large communities. Pheromones regulate several aspects of social life (Leonhardt et al., 2016), such as hierarchy, reproduction control, recruitment to foraging sites, brood care, colony defense, nestmate recognition, and mate search. Moreover, sensitivity to food source odors, such as floral volatiles in bees, is fundamental for efficient foraging (Raguso, 2008).

Reproductive and sterile females perform different tasks: queens or queen-like individuals hardly leave the nest, while a large number of workers perform their tasks outside the nest, foraging being the main one. Moreover, an additional specialization can occur within the worker caste, with individuals performing different tasks during their life, as in honeybees, or being in some species both behaviorally and morphologically specialized. In most eusocial insects, caste and task differentiation may lead females to work for large parts of their life in different environments where specific sensory abilities are required. Moreover, individuals may interact with nestmates of different castes and ages (for instance with reproductive individuals or immature brood), thus being exposed to semiochemicals of different chemical nature.

The large repertoire of compounds secreted by pheromonal glands in social insects, together with the variety of volatiles present in the environment need to be analyzed by an efficient olfactory system (Wittwer et al., 2017). Differences in the perception of environmental and conspecific odorants between castes or during the life cycle has so far received limited attention compared to other aspects of phenotype plasticity (Kelber et al., 2010; Zhang et al., 2016; Nie et al., 2018).

Among social hymenopterans, the European honeybee, Apis mellifera, was the first species in which olfaction was studied at the molecular and neurophysiological levels (Deisig et al., 2006; Forêt and Maleszka, 2006; Robertson and Wanner, 2006; Forêt et al., 2007). This fact, together with the good knowledge on the chemical nature of pheromones involved both in colony communication and sexual behavior (Bortolotti and Costa, 2014) make the honeybee a model organism for the study of chemical communication and olfaction in insects.

Antennal chemosensilla are the main peripheral olfactory organs, where uptake, binding, transport, signal transduction, and signal inactivation occur. Odorants enter through cuticular pores, cross the sensillar lymph and reach the membrane of olfactory neurons (ONs), where two classes of receptors, ORs (olfactory receptors) and IRs (ionotropic receptors) are expressed.

Within the chemosensilla, the dendrites of ONs are bathed in the sensillar lymph containing high concentrations of small soluble proteins, carriers for odorants and pheromones (Pelosi et al., 2006, 2014, 2018; Leal, 2013). Three classes of these proteins have been described so far, but also in other organs producing pheromones. In fact, dual roles have been demonstrated for several members of these proteins, in detecting and releasing semiochemicals (Pelosi et al., 2018). Odorant-binding proteins (OBPs) were the first to be discovered (Vogt and Riddiford, 1981) and currently are the best studied group of olfactory carrier proteins both at the structural level, with more than 20 threedimensional structures solved (Tegoni et al., 2004), four of which in the honey bee (OBP1: Pesenti et al., 2008; OBP2: Lescop et al., 2001; OBP5: unpublished, PDB: 3R72; OBP14: Spinelli et al., 2012), and at functional level (Pelosi et al., 2006). OBPs are 120– 150 amino acid long, present a compact structure made of six α-helical domains and reversibly bind odorants and pheromones with micromolar dissociation constants (Pelosi et al., 2006). Several pieces of evidence have shown that their presence is important for a correct detection of chemical stimuli (Xu et al., 2005; Grosse-Wilde et al., 2006; Forstner et al., 2009; Swarup et al., 2011; Sun et al., 2012; Shiao et al., 2015; Zhang et al., 2017).

Chemosensory proteins (CSPs) is the second class of carrier proteins, smaller than OBPs (110–130 amino acids), also made in α-helical segments, but folded in structures different from those of OBPs (Pelosi et al., 2006). Three CSP structures have been solved (Lartigue et al., 2002; Campanacci et al., 2003; Tomaselli et al., 2006; Jansen et al., 2007) but none belong to honeybee. Like OBPs, several CSPs have been studied at the functional level and show to bind both general odorants and pheromones (Pelosi et al., 2014, 2018).

The third class of insect carrier proteins, NPC2 (Niemann-Pick type C2 protein) has been studied only recently. Although NPC2 proteins have been known for a long time in vertebrates as cholesterol carriers (Storch and Xu, 2009), it was only in the last few years that these proteins were proposed as semiochemical carriers in arthropods, mainly based on their large duplication and differentiation in this phylum (Pelosi et al., 2014). Their localization in chemosensilla and their affinity to small volatile molecules provided further support to this hypothesis (Ishida et al., 2014; Iovinella et al., 2016; Zhu et al., 2018). NPC2 proteins present a folding similar to lipocalins (Flower et al., 2000), with eight β-sheets assembled in a sort of compact β-barrel (Xu et al., 2007).

The genome of the honeybee contains 21 genes encoding OBPs, 6 encoding CSPs, and 5 encoding NPC2. Proteomic studies have identified 13 OBPs, 2 CSPs, and 2 NPC2 in the antennae of workers (Dani et al., 2010; Chan et al., 2013). Some of these proteins and others of the same families are also abundantly expressed in mandibular glands, where they likely assist release of pheromones, with expression patterns related to caste and age (Iovinella et al., 2011).

Previous work has demonstrated that whole-body, haemolymph, and brain protein profiles differ between honeybee queens and workers as well as between hive workers (i.e., workers performing activities inside the nest) and foragers (Engels and Fahrenhorst, 1974; Hummon et al., 2006; Corona et al., 2007; Wolschin and Amdam, 2007a,b; Garcia et al., 2009; Hernández et al., 2012). It has been proposed that the proteomic divergence might reflect the different life history of the two castes and, within workers, be partially explained by a shift in physiological and metabolic requirements as individuals approach different tasks (Corona et al., 2007; Wolschin and Amdam, 2007b; Garcia et al., 2009).

Here, we provide a comprehensive characterization of the antennal proteome of Apis mellifera in a functional perspective,

through a shotgun proteomic approach. We address the question of how protein expression, both in general and with particular reference to soluble olfactory proteins, is related to castes, to different tasks of workers and to ages (**Figure 1**). We show that antennal protein profile, besides changing according to castes, also differs between workers performing different tasks, while it does not appear to be shaped by age.

## MATERIALS AND METHODS

The overall study protocol is shown in **Figure 1**.

#### Apis mellifera Rearing and Sampling

All specimens of Apis mellifera ligustica originated from hives housed at the Department of Biology of the University of Florence (Florence, Central Italy).

Queens of three physiological stages (virgin, newly mated, and established) and workers (i.e., nurses and foragers) originated from three different hives.

First and second instar larvae from three different colonies were reared into queens by transferring them into plastic queen cell cups which were inserted into orphanised colonies maintained within Apidea mating hives. Queens aged 2–4 days were collected either before (virgin, n = 3) or after mating flights (newly mated, n = 3). Fertile queens aged about 1 year (established queens, n = 3) were removed from the same colonies from which also nurses (n = 3) and foragers (n = 3) were collected. Nurses were identified by inspecting brood combs of each hive and searching for bees repeatedly attending brood cells, i.e., bees inserting their head and thorax in a cell containing a larva for at least 5 s (Withers et al., 1993; Crailsheim et al., 1996), while foragers were collected among bees returning from the foraging flights that gathered at the entrance of each hive after blocking it with a grid. All specimens were introduced into plastic tubes, transferred to the lab and soon killed by freezing.

Worker bees performing three different tasks (nurses, guards, and foragers) and of three different ages (1, 2, and 3 weeks) to be used as control, were collected from the same three hives.

Specific worker tasks might require a sensory specialization. Nurses inside the hive should be able to perceive queen and brood-specific semiochemicals emitted by the queen and larvae to respond to their requests (Bortolotti and Costa, 2014); guard bees at the hive entrance may specialize to recognize the difference in the chemical profile of conspecific approaching the colony in order to discriminate nestmates from potential intruders (Breed, 1998) as well as health from diseased individuals (Baracchi et al., 2012; Cappa et al., 2016), while foragers should be equipped to detect different floral odors identifying the flowering plants which provide the richest rewards.

Nurses and foragers were identified as described above. Bees were identified as guards if they patrolled the entrance board with their wings held open, chasing landing bees and inspecting or attacking other bees (Butler and Free, 1952; Downs and Ratnieks, 2000; Cappa et al., 2014, 2016).

Bees of different known ages were obtained by marking newly emerged workers for 5 weeks and collecting them at intervals of 7 days, so to obtain individuals aged 1, 2, and 3 weeks.

Combs with sealed brood, freed from adult individuals with a bee brush, were transferred to the nearby laboratory where workers emerging during the following 2 h were marked (using Uni Posca <sup>R</sup> paints). Combs and marked bees were then reinserted into their hives. Starting from the second up to the eighth week, marked workers of 1, 2, and 3 weeks were collected from the hives. Workers aged 1, 2, and 3 weeks were considered as control for, respectively, nurses, guards, and foragers (Moore et al., 1987; Breed et al., 1990, 2004; Withers et al., 1993; Crailsheim et al., 1996).

#### Preparation of Proteins Samples and Analysis

Dissections were performed immediately before protein extractions and the following samples were prepared: antennae from single queens (virgin, mated, and established) and from single workers (nurses and foragers); pools of antennae from 9 workers (3 from each hive) performing different tasks (nurses, guards, and foragers) and of different age (1, 2, and 3 week-old,). Three biological replicates for each sample were prepared.

The extracts from collected samples were prepared by crushing the tissue in a mortar under liquid nitrogen and the proteins extracted with 6M Urea/2M Thiourea in Tris-Cl 50 mM pH 7.4. The protein extracts were centrifuged at 14.000 rpm for 40 min at 4◦C and the supernatants were collected for the analysis. The total amount of protein in each sample was assessed by the Bradford colorimetric assay (Bradford, 1976), with the "Bio-Rad Protein Assay" kit using serial dilutions of bovine serum albumin to generate a standard curve. Protein sample concentration was measured by Infinite PRO 200 reader (TECAN).

Protein extract were prepared, processed and analyzed on a nanoLC-ESI-LTQ-Orbitrap mass spectrometer as described in Iovinella et al. (2015).

#### Reagents

Ammonium bicarbonate, DTT, iodoacetamide, sodium chloride, formic acid, acetonitrile, trifluoroacetic acid, acetic acid, and thiourea were from Sigma-Aldrich (Milan, Italy), while Tris and urea from Euroclone. Trypsin was purchased from Promega (Sequencing Grade Modified Trypsin) and Lys-C from Thermo Scientific (MS grade). The hand-made desalting/purification STAGE column were prepared using three C18 Empore Extraction Disks (3M).

#### Protein Identification and Quantification

The identification of proteins was performed using MaxQuant software (version 1.5.2.6) (Cox and Mann, 2008). The derived peak list was searched with Andromeda search engine (Cox et al., 2011). We used as database all the proteins of Apis mellifera from Uniprot merged with a set of commonly observed contaminants, such as human keratins, bovine serum proteins, and proteases. Additional variable modifications were set for sequences of

antimicrobial peptides (sequences downloaded from Uniprot<sup>1</sup> ) in 'Group-specific parameters.' In the parameter section, we set as enzyme Trypsin and Lys-C, allowing up to two missed cleavages. The minimum required peptide length was seven amino acids. Carbamidomethylation of cysteine and oxidation of methionine were set as variable modifications. As no labeling was performed, multiplicity was set to 1. During the main search, parent masses were allowed an initial mass deviation of 4.5 ppm and fragment ions were allowed a mass deviation of 0.5 Da. PSM (peptide spectrum match) and protein identifications were filtered using a target-decoy approach at a false discovery rate (FDR) of 1%.

Relative, label-free quantification (LFQ) of proteins was done using the MaxLFQ algorithm integrated into MaxQuant. The match between runs option was enabled with a match time window of 2 min and an alignment time window of 20 min. For protein quantification we used 1 as minimum ratio count, "Unique+Razor" peptides (i.e., those exclusively shared by the proteins of the same group), peptides with variable modifications, and selected "discard unmodified counterpart peptide."

#### Data Analysis

The data relative to identification and quantification are contained in the MaxQuant output files named proteinGroups.txt and are reported in **Supplementary Table S1** for the queens and control workers, and **Supplementary Table S2** for workers of different age and task. Acquisition methods, databases used, and raw files are available through ProteomeXchange<sup>2</sup> (accession: PXD009062).

Further analysis of the MaxQuant-processed data was performed using Perseus software (version 1.5.1.6). Annotations according to gene ontology (GO) categories, Protein family (Pfam) and InterPro were downloaded from the link available in Perseus software<sup>3</sup> and each protein identifier was associated with those categories if available. The data were filtered to eliminate hits to the reverse database, contaminants and proteins only identified with modified peptides.

Differences in single protein levels were first evaluated between queens and workers. A Venn diagram was drawn between queens (virgin, mated, and established) and workers (nurse and foragers), considering "Unique+Razor" peptides identified in at least 3 replicates, out of 9 for queens, and 2 replicates, out of 6, for workers. Differences in single protein levels were evaluated between the two castes, independently from age and/or physiological stage, considering only proteins with at least 5 observations (out of 15), through a t-test on log<sup>2</sup> transformed LFQ intensity values, with a FDR = 0.05 (permutation based false discovery rate), number of randomization set to 1000 and S0 set to 0.1. This latter value is an artificial within groups variance which controls both the relative importance of t-test p-value and difference between means (Tusher et al., 2001). Differential expression analysis between queens and workers of different ages and/or physiological

<sup>1</sup>http://www.uniprot.org/uniprot/?query=antimicrobial+peptides+apis+ mellifera&sort=score

<sup>2</sup>www.proteomexchange.org

<sup>3</sup>http://141.61.102.106:8080/share.cgi?ssid=0qF9uFn

stages was performed using ANOVA, where p-values were Benjamini Hochberg corrected at 5% FDR. A post hoc twosample t-test, with the same correction, was applied to determine differences in single protein levels between antennae of workers and queens, compared according age, and between queens at different physiological stages. Hierarchical clustering analyses were performed using average Euclidean distance and the default parameters of Perseus (300 clusters, maximum 10 iterations).

The same approach was used to evaluate differences between workers of different tasks and ages. Differential expression analysis was performed using ANOVA, where p-values were Benjamini Hochberg corrected at 5% FDR, considering only proteins with at least 6 observations (out of 18). A post hoc two-sample t-test, with the same correction, was applied to determine differences in single protein levels between antennae of workers performing different tasks, as well as comparing them with the respective age control samples. Hierarchical clustering analyses were performed using average Euclidean distance and the default parameters of Perseus (300 clusters, maximum 10 iterations).

Differential expression of olfactory proteins (OBPs, CSPs, NPC2, and ORs) and odorant degrading enzymes (ODEs) was further analyzed by considering reduced datasets containing only data of these proteins. Missing LFQ values were imputed (width = 0.3, downshift = 1.8), and 0 was manually substituted when values were missing in all replicates of one caste/task/age category. T-test (Benjamini Hochberg corrected at 5% FDR) was calculated on these data.

#### RESULTS AND DISCUSSION

Aim of this work was a proteomic analysis of antennae of honeybees belonging to different castes (queens and workers) and of workers performing different tasks; for these latter bees of known ages were used as control, in order to understand if age influences protein expression profile.

#### Differences Between Castes

Search of LC-MS data acquired for antennal extracts from single individuals (queens and control workers) identified 395 proteins. Data regarding the identification of all proteins, together with other information (accessions, scores, percent coverage, missed cleavages, etc.) are reported in **Supplementary Table S1**.

Firstly, we compared the global expression of proteins between the two castes, regardless of age and/or physiological stage (nurses and foragers as workers vs. young virgin, young mated, and established queens).

We obtained a comparable distribution of protein families, with the PBP/GOBP family as the most represented in both castes. Thirteen proteins were exclusively found in queens; none of these proteins have been reported to have a role in olfaction or be linked to caste differentiation (**Table 1**), except for Major royal jelly protein 1 (acc. O18330), that is the most abundant protein found in the royal jelly, the food of the queen honey bee larva that determines the development of the young larvae and is responsible for the high reproductive ability of honeybee queens (Buttstedt et al., 2014).

Abundance (log<sup>2</sup> transformed LFQ values) of proteins quantified in at least 5 out of the 15 samples, was compared through a t-test (FDR = 0.05) and graphically represented by a volcano plot (**Figure 2**); 20 and 31 proteins were more expressed in workers and queens, respectively (**Supplementary Table S3**). In workers two OBPs (OBP2 and OBP15) and two CSPs (CSP1 and CSP3) were significantly more expressed, together with several enzymes possibly involved in degradation of odors and/or pheromones, a couple of structural proteins (calreticulin and tubulin) and enzymes involved in various biological processes, such as metabolism and transport.

Among proteins significantly more abundant in queens there are two OBPs (OBP3 and OBP18) and the NPC2-1, two cuticle proteins and several lipid transport proteins, among which we found two apolipophorins and three vitellogenins. Differences of olfactory proteins ranged from 2 (OBP2, CSP3, and NPC2-1) to around 4 times.


FIGURE 3 | Heatmap representation of the expression of proteins significantly different (one-way ANOVA, Benjamini Hochberg-corrected FDR = 5%) between groups of both castes. The map has been built making an unsupervised hierarchical clustering (300 clusters, maximum 10 iterations) based on LFQ (label-free quantification). Uniprot accession numbers are reported in brackets. Color scale reports Z-score log2 transformed LFQ intensity values. Missing data are reported in gray. Groups belonging to the two castes are clearly separated, as displayed in the cluster grouping biological replicates.

The data regarding olfactory proteins are in good agreement with those reported by Chan et al. (2013), where a proteomic study of different organs of Apis mellifera belonging to different castes was conducted; quantitative differences between queens and workers were comparable, apart from OBP15, which was not found in their work.

Quantitative differences in protein expression between the single groups (nurse, foragers, virgin queens, mated queens, and established queens) belonging to the different castes were evaluated through one-way ANOVA (Benjamini Hochbergcorrected FDR = 5%). The heatmap reported in **Figure 3** shows the 13 proteins differentially expressed between castes (**Table 2**). Most of them are 'uncharacterized proteins'; about one half present a higher expression in queens, including the OBP3 and two storage proteins (a vitellogenin and a hexamerin). This latter finding reflects their physiological role. In fact, vitellogenin has been reported to act as an antioxidant to promote longevity in queen bees (Corona et al., 2007). The presence of the hemolymph protein hexamerin 70 in the antennae has been reported in young queens (4 days old) and it has been suggested that it could be used in the building up of antennal cuticle structures and it could be related to modifications of the external structure of the sensilla placodea (Danty et al., 1998).

Protein profiles of queens of different age/stage were compared among them and toward those of the corresponding group of workers through a t-test (Benjamini Hochbergcorrected FDR = 2%).

No proteins differed between virgin and mated queens, therefore, we pooled the two groups together as young queens and we compared them with nurses (workers of comparable age) and established queens (older queens in a different physiological stage). Since several differences were observed with respect to established queens, we can deduce that a stable reproductive status also affects antennal protein expression pattern, while mating has little or no effect on it.

The highest number of differences in protein abundance was obtained comparing nurses with virgin and mated queens, showing that NPC2-1, OBP3, Vitellogenin (acc. A0A088ADL8), and Hexamerin (acc. A6YLP7) are typical

TABLE 2 | Proteins differentially expressed in single groups of the two castes, according to one-way ANOVA (Benjamini Hochberg-corrected FDR = 5%) and to post hoc two-sample Student's t-tests.


of young queens, while the circadian clock-controlled protein-like, a Haemolymph juvenile hormone (JH) binding protein, characterizes both nurses and foragers compared to young queens and established queens, respectively. The t-test statistics concerning each comparison are reported in **Table 2**.

Besides the global expression pattern of antennal proteins, our primary aim was to analyze how castes influence the profile of olfactory proteins. We identified 12 of the 21 predicted OBPs, 2 of the 6 predicted CSPs, and 1 out of the 5 NPC2, in our proteomic analysis and we have analyzed their expression comparing queens and workers of the same age (**Figure 4A**). The t-tests (Benjamini Hochberg-corrected FDR = 5%) performed considering only these proteins showed that, in the comparison between nurse and virgin queens, OBP2 is significantly more abundant in nurse, while OBP3 and NPC2-1 are more expressed in queens. Moreover, OBP3 is more abundant in mated compared to established queens, confirming that this protein characterizes young queens. In the comparison between foragers and established queens the proteins OBP14 and OBP18 were more abundant in established queens. OBP2 has been found to have a good affinity for components (2-heptanone, isoamyl acetate) of alarm pheromone (Briand et al., 2001), while OBP3 binds benzoate (Dani et al., 2010), although information is not available for methyl p-hydroxybenzoate, one of the major components of the queen mandibular gland. OBP18, together with OBP16, has been reported to be more expressed in workers with higher hygienicity and bind long chain fatty acids and their ethyl and methyl esters (Guarna et al., 2015), some of which are constituents of the brood pheromone.

In addition to OBPs, CSPs, and NPC2, other protein families are involved in peripherical processes of odor perception in insects, in particular the ODEs, involved in inactivation through degradation of the chemical signals, once the message is conveyed. Among the Pfams containing proteins that have been reported to be involved in this process (Yu et al., 2009; Durand et al., 2012; Leal, 2013), we selected those significantly enriched (Fisher exact test; Benjamini Hochbergcorrected FDR = 0.02) and we evaluated differences between single groups of both castes (**Figure 4B**). Only in the comparison between nurses and young queens three proteins were statistically significant (t-test): the delta-1-pyrroline-5 carboxylate dehydrogenase, mitochondrial (acc. A0A088A1I8), and the esterase FE4-like (acc. A0A088AW01) more expressed

FIGURE 4 | (A) Bar chart reporting the log<sup>2</sup> transformed and imputed LFQ intensity values of the olfactory proteins, averaged for biological replicates (±SE). Proteins marked with a symbol are significant to t-test (Benjamini Hochberg-corrected FDR = 5%) for the comparison nurse-virgin queen (asterisk), nurse-mated queen (hash), mated-established queen (filled circle), and foragers-established queen (circle). (B) Bar chart reporting the log<sup>2</sup> transformed and imputed LFQ intensity values of the odorant degrading enzymes (ODEs), indicated with Uniprot accession number, averaged for biological replicates (±SE). Protein marked with an asterisk are significant to t-test (Benjamini Hochberg-corrected FDR = 5%) between nurse and young queens (virgin and mated).

in nurses, while the esterase E4-like (acc. A0A088AQ81) is significantly more abundant in young queens. The t-test statistics for olfactory proteins, concerning each comparison, are reported in **Supplementary Table S4**.

#### Differences Between Tasks

To understand which factor, different tasks and/or age, could influence protein expression in workers we analyzed antennae from pools of nurses, guards, and foragers (different tasks) and of workers of comparable age (1, 2, or 3 weeks, respectively), but for which specific task was not assessed. Search of LC-MS data acquired for pools of antennae of workers identified 530 proteins. Data regarding the identification of all proteins, together with other information (accessions, scores, percent coverage, missed cleavages, etc.) are reported in **Supplementary Table S2**.

Considering separately each group of workers of different tasks and ages, we did not find proteins exclusively expressed in one group. Differences in protein expression between groups of different tasks (nurse, guards, and foragers) and ages (first, second, and third week) were evaluated through one-way ANOVA on log<sup>2</sup> transformed LFQ values (Benjamini Hochbergcorrected FDR = 5%). The heatmap reported in **Figure 5** shows the 39 proteins differentially expressed between castes (**Table 3**). Most of them (29 proteins) are enzymes and present higher expression in guards and foragers, with respect to nurse, which are closer to workers of know age for which task was not assessed.

Protein profiles of workers carrying out different tasks were compared to those of the corresponding coetaneous workers through a t-test (Benjamini Hochberg-corrected FDR = 5%). No differences were obtained comparing honeybees with defined ages (1st week versus 2nd and 3rd week, 2nd week versus 3rd week), as well as between guards compared to foragers and nurse compared to honeybees of 1st week. Major differences were obtained between guards compared to honeybees of 2nd week. Thus, the observed differences appear to be linked to the specific task performed by workers rather than by the different age. As already reported by previous studies, task specialization is often followed by biochemical and physiological specialization of bee workers (Robinson, 1987; Huang et al., 1994; Pearce et al., 2001; Amdam et al., 2003; Münch et al., 2008). Young nurses performing inside hive duties present high titers of vitellogenin (Huang et al., 1994; Amdam et al., 2003; Münch et al., 2008), whereas middle-aged guard bees and older foragers show very

FIGURE 5 | Heatmap representation of the expression of proteins significantly different (one-way ANOVA, Benjamini Hochberg-corrected FDR = 5%) between groups of workers of different tasks and ages. The map has been built making an unsupervised hierarchical clustering (300 clusters, maximum 10 iterations) based on LFQ (Label-free quantification). Uniprot accession numbers are reported in brackets. Color scale reports Z-score log2 transformed LFQ intensity values. Missing data are reported in gray. Major differences are between old workers (guards and foragers) and nurse, which are in the same cluster of bees with undetermined task.

TABLE 3 | Proteins differentially expressed in single tasks/ages groups of workers, according to one-way ANOVA (Benjamini Hochberg-corrected FDR = 5%) and to post hoc two-sample Student's t-tests.


(Continued)

#### TABLE 3 | Continued

fphys-09-00748 June 19, 2018 Time: 16:7 # 11


high levels of JH promoting, respectively, aggressive behavior in guards (Breed, 1983; Pearce et al., 2001) and the onset of foraging in older bees (Robinson, 1985, 1987; Sullivan et al., 2000; Elekonich et al., 2001).

In our samples, two vitellogenins, one hexamerin, and the OBP14 are more abundant in nurses and workers of 2nd and 3rd week with respect to guards and foragers, while a JH binding protein (acc. A0A087ZV30) is significantly more expressed in guards and foragers with respect to their age-control workers. The higher expression of a JH-binding protein may be linked to the higher titers of such hormone in these specific groups of workers (Breed, 1983; Robinson, 1985, 1987; Sullivan et al., 2000; Elekonich et al., 2001; Pearce et al., 2001). Among the enzymes, there are a 'farnesol dehydrogenase-like' protein (acc. A0A087ZPU1) and a 'Cytochrome c' (acc. P00038), whose function could be probably related to the inactivation of chemical signals, which are more expressed in guards and foragers, respectively. The t-test statistics concerning each comparison are reported in **Table 3**. The absence of conspicuous differences among workers of the different age groups (1st, 2nd, and 3rd week) compared to the ones observed in task-specific groups, may be due to the fact that each age groups is likely to

**259**

include workers involved in different tasks. Indeed, variation in task performance among similarly aged workers is common in honeybee colonies (Winston, 1991; Huang et al., 1994) and the presence of bees performing different tasks in our age groups could mask the differences observed among task-specific groups.

A more detailed analysis was conducted on the expression patterns of soluble olfactory proteins, in order to understand if their profile could characterize workers of different tasks and ages.

We identified 11 of the 21 predicted OBPs, 3 of the 6 predicted CSPs, 2 out of 5 predicted NPC2, and one odorant receptor in our proteomic analysis (**Figure 6A**). We can observe that in this case, with respect to the samples from single individuals, we identified more proteins, and this is certainly due to the use of pools of antennae (from 9 bees). In fact, comparing nurses and foragers, for whom we have both single specimens and pooled samples, we observed an increase of 10% in the number of identified proteins. However, among OBPs, OBP18 was identified with only 1 peptide and the protein was included in the same protein group with OBP21 (**Figure 5**), while this was not the case in the single specimen samples. Moreover, in this analysis we found CSP4 and NPC2-2 that were not found in antennal extract from single individuals. Surprisingly we have also identified the odorant receptor 67a-like isoform X1 that being a transmembrane protein is not easy to solubilize given our mild protein extraction. In general, proteomic studies are more suitable to target soluble proteins than membrane proteins and our results are consistent with other similar analyses on insect chemosensory organs.

Unexpectedly, we did not find OBP1, that was instead identified in a 2D-gel spot of foragers antennae in our previous work, together with OBP16 (Dani et al., 2010). The expression of the OBP1 encoding gene was found to be limited to antennae and comparable between drones, queens, and foragers (Forêt and Maleszka, 2006); however, the protein is around 5 times more abundant in drones with respect to workers and queens (Chan et al., 2013), and this could explain why we don't identify the protein in our sample. Apart from these differences, the abundances of all the OBPs and the CSPs identified in nurses and foragers of both datasets are strongly consistent.

The t-tests (Benjamini Hochberg-corrected FDR = 5%) performed only on olfactory proteins showed that, in the comparison between nurse and old workers (guards and foragers considered together) only the OBP14 was significantly more expressed (more than 2 times) in nurse, while none of the considered proteins was differentially expressed comparing workers with defined task and their age-controlled sample. The differences in OBP14 suggest that this protein can be involved in pheromonal communication within the hive rather than to perception of floral odors. This finds a biological correlation with the affinity for compounds reported for aggregation (farnesol, geraniol, and citral) or alarm pheromones (2-heptanone and isoamyl acetate) but not with the very strong affinity reported for eugenol (Iovinella et al., 2011).

Even in this case we selected Pfams significantly enriched (Fisher exact test; Benjamini Hochberg-corrected FDR = 0.02) containing proteins that have been reported as ODEs (**Figure 6B**). Differences have been detected only between guards and workers of 2nd week for a Glutathione S-transferase (acc. A0A088ABV3) and a 'peroxisomal multifunctional enzyme type 2-like' (acc. A0A088AVD8), which are both more expressed in guards. The t-test statistics for olfactory proteins are reported in **Supplementary Table S5**.

A similar approach to that used in the present work has been adopted for a comparative transcriptome analysis conducted on Apis mellifera antennae of workers performing different tasks by Nie et al. (2018). None of the proteins encoded by the genes reported as associated with nursing and foraging behavior were found to be differentially expressed in our samples. With regards to OBPs and CSPs, similarly to results by Nie et al. (2018), we also observed a decrease of OBP17 level from nursing to foraging task, although the difference in abundance was not statistically significant.

#### CONCLUSION

This work presents for the first time a detailed proteomic investigation of Apis mellifera antennae where bees belonging to different castes, at different physiological stages, and workers performing different tasks have been compared. To control for age-related changes workers were also compared with bees of different ages but of unassessed task.

Expression analysis has highlighted differences between the two castes, including several proteins involved in olfaction. Among these, the NPC2-1 and the OBP3 characterize young and still not egg-laying queens, together with storage proteins well known for their role in caste determination (two vitellogenins and one hexamerin).

Major differences have been found between groups of workers performing different tasks and groups of defined age, while antennal protein profiles of honeybees at 1st, 2nd, and 3rd week do not show differences. Among the soluble olfactory proteins, we found that OBP14 is typical of nurse bees with respect to guards and foragers.

The data here reported are in good, although not complete, agreement with the results at the RNA level reported by Forêt et al. (2007) and the proteomic analysis of antennae between castes (Chan et al., 2013), while they have limited correspondence with the comparative transcriptomic work by Nie et al. (2018), where antennae of workers of different tasks were studied.

Our data suggest that caste, physiological stage and performed task shape the antennal profile of honeybees and that two OBPs and one NPC2 are differentially expressed. Since the binding properties have been defined only for a few honeybee soluble olfactory proteins, studies aimed at understanding how expression of these proteins associates with castes and with task transitions may suggest which semiochemicals should be targeted to clarify their physiological role.

## ETHICS STATEMENT

Honeybees used in this work were reared in semi-natural conditions. They were treated as well as possible given the constraints of the experimental design. This study was carried out in accordance with the Italian guidelines on animal wellness.

## AUTHOR CONTRIBUTIONS

FD, RC, and ST designed the research. II, FC, AC, and IP performed the experiments. II and FD performed the statistical analysis and drafted the manuscript. All authors discussed the results during the progress of the work, participated in revising the article critically, helped finalizing the manuscript, and gave final approval for publication.

#### FUNDING

Financial support was provided by the project Progetti di Rilevante Interesse Nazionale (PRIN) 2012 (Prot. No. 2012RCEZWH\_001) to ST.

## ACKNOWLEDGMENTS

We are very grateful to Michele Giovannini and Anna Marta Lazzeri for helping in honeybee collection and marking, to Elena Michelucci for technical assistance during LC-MS/MS analysis, and to Paolo Pelosi for revising our manuscript. The

mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaíno et al., 2016) partner repository with the dataset identifier PXD009062.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00748/full#supplementary-material

TABLE S1 | Complete list of proteins identified in proteomic analysis of antennae from single individuals of nurses (N), foragers (F), virgin queens (RV), mated queens (RF), and established queens (Rold). The protein groups table contains

#### REFERENCES


information on the proteins identified in all processed raw-files. Each single row contains the group of proteins that could be reconstructed from a set of peptides.

TABLE S2 | Complete list of proteins identified in proteomic analysis of antennae from pool of 9 individuals of nurses (N), guards (G), foragers (F) and workers aged 1-week (A), 2-week (B) and 3-week (C). The protein groups table contains information on the proteins identified in all processed raw-files. Each single row contains the group of proteins that could be reconstructed from a set of peptides.

TABLE S3 | Proteins significantly different between castes (t-test, FDR = 0.05), graphically represented in volcano plot of Figure 2.

TABLE S4 | Soluble olfactory proteins differentially expressed in single groups of the two castes (Student t-test Benjamini Hochberg-corrected FDR = 5%).

TABLE S5 | Soluble olfactory proteins differentially expressed in groups of workers (Student t-test Benjamini Hochberg-corrected FDR = 5%).



with their social relatives. PNAS. 114, 6569–6574. doi: 10.1073/pnas.16207 80114


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Iovinella, Cappa, Cini, Petrocelli, Cervo, Turillazzi and Dani. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Sex- and Tissue-Specific Expression Profiles of Odorant Binding Protein and Chemosensory Protein Genes in *Bradysia odoriphaga* (Diptera: Sciaridae)

#### Yunhe Zhao<sup>1</sup> , Jinfeng Ding<sup>1</sup> , Zhengqun Zhang<sup>2</sup> , Feng Liu<sup>1</sup> , Chenggang Zhou<sup>3</sup> and Wei Mu<sup>1</sup> \*

*<sup>1</sup> College of Plant Protection, Shandong Provincial Key Laboratory for Biology of Vegetable Diseases and Insect Pests, Shandong Agricultural University, Tai'an, China, <sup>2</sup> College of Horticultural Science and Engineering, Shandong Agricultural University, Tai'an, China, <sup>3</sup> College of Plant Protection, Shandong Agricultural University, Tai'an, China*

#### *Edited by:*

*Bin Tang, Hangzhou Normal University, China*

#### *Reviewed by:*

*Pablo Pregitzer, University of Hohenheim, Germany Peng He, Guizhou University, China*

> *\*Correspondence: Wei Mu muwei@sdau.edu.cn*

#### *Specialty section:*

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

*Received: 30 November 2017 Accepted: 02 February 2018 Published: 03 April 2018*

#### *Citation:*

*Zhao Y, Ding J, Zhang Z, Liu F, Zhou C and Mu W (2018) Sex- and Tissue-Specific Expression Profiles of Odorant Binding Protein and Chemosensory Protein Genes in Bradysia odoriphaga (Diptera: Sciaridae). Front. Physiol. 9:107. doi: 10.3389/fphys.2018.00107* *Bradysia odoriphaga* is an agricultural pest insect affecting the production of Chinese chive and other liliaceous vegetables in China, and it is significantly attracted by sex pheromones and the volatiles derived from host plants. Despite verification of this chemosensory behavior, however, it is still unknown how *B. odoriphaga* recognizes these volatile compounds on the molecular level. Many of odorant binding proteins (OBPs) and chemosensory proteins (CSPs) play crucial roles in olfactory perception. Here, we identified 49 OBP and 5 CSP genes from the antennae and body transcriptomes of female and male adults of *B. odoriphaga*, respectively. Sequence alignment and phylogenetic analysis among Dipteran OBPs and CSPs were analyzed. The sex- and tissue-specific expression profiles of 54 putative chemosensory genes among different tissues were investigated by quantitative real-time PCR (qRT-PCR). qRT-PCR analysis results suggested that 22 OBP and 3 CSP genes were enriched in the antennae, indicating they might be essential for detection of general odorants and pheromones. Among these antennae-enriched genes, nine OBPs (*BodoOBP2/4/6/8/12/13/20/28/33*) were enriched in the male antennae and may play crucial roles in the detection of sex pheromones. Moreover, some OBP and CSP genes were enriched in non-antennae tissues, such as in the legs (*BodoOBP3/9/19/21/34/35/38/39/45* and *BodoCSP1*), wings (*BodoOBP17/30/32/37/44*), abdomens and thoraxes (*BodoOBP29/36*), and heads (*BodoOBP14/23/31* and *BodoCSP2*), suggesting that these genes might be involved in olfactory, gustatory, or other physiological processes. Our findings provide a starting point to facilitate functional research of these chemosensory genes in *B. odoriphaga* at the molecular level.

Keywords: *Bradysia odoriphaga*, odorant binding protein, chemosensory protein, expression profiles analysis, transcriptomes

## INTRODUCTION

The Chinese chive maggot, Bradysia odoriphaga (Diptera: Sciaridae), is the major destructive pest of Chinese chive and other liliaceous vegetables in China (Zhang et al., 2015; Chen et al., 2017). The larvae of this pest feed on the underground roots, bulbs, and immature stems of Chinese chive and cause yield losses of more than 50% in the absence of insecticide protection (Ma et al., 2013). Thus far, the application of chemical insecticides remains the primary measure for controlling B. odoriphaga, and it has led to many adverse impacts, such as widespread insecticide resistance and toxic residues in chives, threatening consumer health (Zhang P. et al., 2016; Chen et al., 2017). Hence, a new ecofriendly pest management strategy is needed to control this pest. Previous studies have shown that B. odoriphaga was significantly attracted by sex pheromones, the volatiles derived from host plants and microbial secondary metabolites (Li et al., 2007; Chen et al., 2014; Uddin, 2016; Zhang Z. J. et al., 2016), and that it was repelled by green leaf volatile compounds (Chen C. Y. et al., 2015). Moreover, B. odoriphaga exhibited a strong electroantennogram (EAG) response to trans-2-hexenal and benzothiazole (Chen C. Y. et al., 2015). The evidence from these behavioral responses contribute to control of this pest using push-pull strategies (Cook et al., 2007). Despite these reports on chemosensory behavior, however, the mechanism by which B. odoriphaga recognizes these volatile compounds on the molecular level is still unknown.

Olfaction is the primary sensory modality in insects and plays a crucial role in various physiological behaviors, such as locating sexual partners, food sources, oviposition sites, and avoiding predators (Visser, 1986; Leal, 2013). The antennae are the principal olfactory organs for insect olfaction, and the olfactory perception process generally includes two main steps. First, odorant molecular penetrate into the sensillar lymph through pores, and they are bound by small, amphipathic proteins [odorant binding proteins (OBPs) or chemosensory proteins (CSPs); (Pelosi et al., 2006; Zhou, 2010; He et al., 2017)]. Second, the OBPs or CSPs will transfer the odorant molecule through the sensillar lymph to the olfactory receptors (ORs), activate the olfactory receptor neurons (ORNs) and convert chemical signals into electrical signals that are sent to the insect brain (Vogt et al., 1999; Leal, 2013; Pelosi et al., 2018). Hence, OBPs and CSPs are very important because they mediate the first step of odor perception (Li et al., 2015; Brito et al., 2016).

The first step toward understanding the molecular mechanism of olfactory perception process is to investigate olfaction-related genes, which encode the proteins that function in odorant molecular detection. Since OBPs and CSPs were identified and characterized in the model insect, Drosophila melanogaster (Robertson et al., 2003), a large number of OBP and CSP genes have been identified from diverse families of Diptera insects, including sanitary pests (Pelletier and Leal, 2011; Manoharan et al., 2013; Rinker et al., 2013; Scott et al., 2014; Chen X. G. et al., 2015; Leitch et al., 2015; He X. et al., 2016), agricultural pests (Andersson et al., 2014; Gong et al., 2014; Ohta et al., 2014, 2015; Elfekih et al., 2016; Liu et al., 2016), and predators (Wang et al., 2017). Furthermore, the functions of some OBP and CSP genes in the olfactory perception process of insects have been predicted and verified (Swarup et al., 2011; Siciliano et al., 2014; Wu et al., 2016; Zhu et al., 2016). However, thus far, only two OBP genes and one CSP gene have been identified in B. odoriphaga from Sciaridae, and the number, classification, expression characteristics and functions of OBP and CSP genes in B. odoriphaga are still unknown.


Frontiers in Physiology | www.frontiersin.org

TABLE

1


*Bradysia odoriphaga*.


In the present study, we performed transcriptome analysis of the antennae and body of female and male adult of B. odoriphaga, respectively, and identified 54 putative chemosensory genes comprising 49 OBPs and 5 CSPs. Then, sequence alignment and phylogenetic analysis were undertaken among Dipteran OBPs and CSPs. The transcript expression levels of 54 putative chemosensory genes among different tissues (female antennae, male antennae, legs, wings, abdomens and thoraxes, and heads) were investigated by quantitative real-time PCR (qRT-PCR) (**Graphical Abstract**). This work provides a starting point to facilitate functional studies of these OBP and CSP genes in B. odoriphaga at the molecular level.

#### MATERIALS AND METHODS

#### Insect Culture and Tissue Collection

A laboratory colony of B. odoriphaga was collected from a Chinese chive field in Liaocheng, Shandong Province, China (36◦ 02′N, 115◦ 30′E) in April 2013. The insects were reared on fresh chive rhizomes and placed in Petri dishes, which were maintained at 25 ± 1 ◦C, 70 ± 5% RH with a photoperiod of 14:10 h (L:D) in a climate-controlled chamber. The antennae and the remaining body parts (mixture of heads, thoraxes, abdomens, legs and wings) of female and male adults were separated quickly and then stored in liquid nitrogen until RNA extraction (female antennae: FA; male antennae: MA; female body: FB; male body: MB). Approximately 1,000 antennae and 30 bodies of females and males were collected for RNA extraction, and three biological replicates were performed.

#### RNA Isolation, cDNA Library Construction, and Illumina Sequencing

Total RNA was isolated from antennae and bodies using Trizol reagent (Invitrogen, Carlsbad, CA, USA), according to the manufacturer's instructions. Then, all the RNA samples were treated with DNase I (Invitrogen, Carlsbad, CA, USA) to eliminate the genomic DNA. The concentration of isolated RNA was measured with a NanoDrop ND-2000 spectrophotometer (NanoDrop products, Wilmington, DE, USA), and the integrity of RNA extractions were determined by agarose gel electrophoresis. cDNA library construction was performed using a TruseqTM RNA sample prep Kit (Illumina, San Diego, CA, USA) and was sequenced on an Illumina HiSeq 4000 (Illumina, San Diego, CA, USA). After removing the low quality and adapter sequences, clean short reads were mapped to contigs, and contigs were assembled to unigenes by the short-read assembly program Trinity (Grabherr et al., 2011). Then, unigenes were annotated using different databases, including the non-redundant protein (Nr), nucleotide sequence (Nt), Swiss-Prot, Clusters of Orthologous Groups (COG), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Ontology (GO) databases (E-value < 10−<sup>5</sup> ).

#### Identification and Comparison of Transcript Abundance of OBP and CSP Genes

The tBLASTn program was used to identify candidate unigenes that encode putative OBPs and CSPs from the antennae,

 *genes* 

 *a* 

 *or* 

body transcriptomes and fourth instar larval transcriptome of B. odoriphaga (unpublished data). All putative OBP and CSP genes were confirmed by the BLASTx program at the National Center for Biotechnology Information (NCBI, http://blast.ncbi. nlm.nih.gov/Blast.cgi). The open reading frames (ORFs) of OBP and CSP genes were predicted by the ORF Finder (https:// www.ncbi.nlm.nih.gov/orffinder/). The conserved domains of these candidate OBPs and CSPs were predicted utilizing SMART (http://smart.embl.de; Letunic and Bork, 2017).

To compare the expression levels of the candidate OBP and CSP genes in the antennae and body transcriptomes (FA, MA, FB, and MB) of B. odoriphaga, the FPKM (fragments per kilobase of exon per million fragments mapped) values were used for calculating transcript abundance (Andersson et al., 2014). Heatmaps of gene expression for different OBPs among FA, MA, FB and MB were generated by R version 3.4.1 (R Development Core Team, The R Foundation for Statistical Computing, Vienna, Austria).

## Verification of the OBP and CSP Sequences by Cloning and Sequencing

All the putative OBP and CSP nucleotide sequences obtained from the B. odoriphaga transcriptomes were confirmed by gene cloning and sequencing. Gene-specific primers were designed to amplify the complete or partial ORF sequences of each OBP and CSP gene (Table S1). The cDNA template was synthesized by the TransScript <sup>R</sup> All-in-One First-Strand cDNA Synthesis SuperMix for PCR Kit (TransGen Biotech, Beijing, China). PCR amplification was performed in a 25 µl volume containing 2.0 µl of cDNA (300 ng), 0.5 µl of TransScript <sup>R</sup> KD Plus DNA polymerase (TransGen Biotech, Beijing, China), 5 µl of 5×TransScript <sup>R</sup> KD Plus Buffer, 2 µl of dNTPs (2.5 mM), 0.5 µl each of the forward and reverse primers (10µM), and 14.5 µl of nuclease free H2O. The cycling conditions were an initial denaturation at 94◦C for 3 min, followed by 35 cycles of 94◦C for 30 s, 56◦C for 30 s, 68◦C for 45 s, and a final extension at 68◦C for 10 min. Then, the PCR products were purified by agarose gel electrophoresis and an EasyPure <sup>R</sup> Quick Gel Extraction Kit (TransGen Biotech, Beijing, China), and subcloned into the pEASY <sup>R</sup> -Blunt cloning vector (TransGen Biotech, Beijing, China) and sequenced.

#### Sequence and Phylogenetic Analysis

The putative N-terminal signal peptides of BodoOBPs and BodoCSPs were predicted by the SignalP V 4.1 program (http:// www.cbs.dtu.dk/services/SignalP/; Nielsen, 2017). Multiple alignments and identity calculation were conducted by Clustal X 2.0 software (Larkin et al., 2007). A total of 280 OBP protein sequences from four Diptera species were used to construct the phylogenetic tree, including 49 OBPs from B. odoriphaga identified in this study, 51 OBPs of D. melanogaster, 69 OBPs of Anopheles gambiae, and 111 OBPs of Aedes aegypti (Sequences are listed in Table S2). In addition, 97 CSP protein sequences from seven Diptera species were used for the phylogenetic analysis, including 5 CSPs of B. odoriphaga identified in the present study, 4 CSPs of D. melanogaster, 8 CSPs of A. gambiae, 8 CSPs of Anopheles sinensis, 43 CSPs of A. aegypti, 27 CSPs of Culex quinquefasciatus, and 2 CSPs of D. antiqua (sequences are listed in Table S3). All the phylogenetic trees were constructed by MEGA 6.0 software with the neighborjoining method using default settings and 1,000 bootstrap replications (Tamura et al., 2013). The final phylogenetic tree was visualized by an online tool, EvolView (He Z. L. et al., 2016).

## Motif Analysis

A total of 318 OBPs (from 6 Diptera species) and 138 CSPs (from 18 Diptera species) were used for comparing the motif pattern between Diptera OBPs and CSPs. All OBP and CSP sequences (Table S4) with intact ORFs were used for motif discovery and pattern analysis. The protein motifs analysis was performed using the MEME (version 4.12.0) online server (http://meme-suite.org; Bailey et al., 2015). The parameters used for motif discovery were: minimum width = 6, maximum width = 10, and the maximum number of motifs to find = 8.


#### Tissue Expression Profile Analysis

The expression profiles for different tissues of these 49 OBPs and 5 CSPs were evaluated by qRT-PCR. The female antennae (FA), male antennae (MA), legs (L), wings (W), abdomens and thoraxes (AT), and heads (H) were collected from adult B. odoriphaga after eclosion without mating. Total RNA was isolated from different tissues using Trizol reagent (Invitrogen, Carlsbad, CA, USA), according to the manufacturer's instructions. The cDNA template was synthesized by the TransScript <sup>R</sup> All-in-One First-Strand cDNA Synthesis SuperMix for qPCR (One-Step gDNA Removal) Kit (TransGen Biotech, Beijing, China). Specific primers used for qRT-PCR were designed by the software Beacon Designer 7.90 (PREMIER Biosoft International) and are listed in Table S5. Two reference genes, RPS15 (ribosomal protein S15) and RPL18 (ribosomal protein L18) were used for normalizing target gene expression and to correct for sample-to-sample variation (Shi et al., 2016). The experiment was conducted using the LightCycler <sup>R</sup> 96 System (Roche Molecular Biochemicals, Lewes, United Kingdom) and each reaction was conducted in a 20 µl reaction mixture containing 1.0 µl of sample cDNA (150 ng), 10 µl of Mix (2×TransScript <sup>R</sup> Tip Green qPCR SuperMix) (TransGen Biotech, Beijing, China), 1.0 µl of forward primer (10µM), 1.0 µl of reverse primer (10µM), and 7 µl of nuclease free H2O. The reaction programs were as follows: 95◦C for 10 min, followed by 45 cycles of amplification (95◦C for 10 s and 60◦C for 30 s). Then, a melting curve was analyzed for PCR products to detect a single gene-specific peak and to check for the absence of primer dimer peaks. Negative controls were nontemplate reactions (replacing cDNA with H2O). Three technical replicates and three biological replicates were conducted for all experiments.

The results were analyzed using the LightCycler <sup>R</sup> 96 software. Relative quantification of different tissues was calculated by the comparative 2−11Ct method (Livak and Schmittgen, 2001). Comparative analyses of each target gene among different tissues were determined using one-way ANOVA tests followed by Tukey's HSD method using SPSS statistical software (version 18.0, SPSS Inc., Chicago, IL, USA) (P < 0.05). When applicable, the values are shown as the mean ± SE.

## RESULTS

#### Overview of the Transcriptome of *B. odoriphaga*

A total of 42.6 GB of clean data was obtained from the antennae and body transcriptomes of B. odoriphaga. After assembling all samples together, we identified 55,867 unigenes with an N50 length of 2,806 bp (Table S6). For the annotation, 32,492, 17,867, 26,930, 26,289, 15,633, 26,541, and 11,578 unigenes were annotated to Nr, Nt, SwissProt, InterPro, KEGG, COG, and GO databases, respectively, which covered 35,013 (62.67%) of the total unigenes (Table S7).

Gene Ontology (GO) annotation analysis was used to categorize these unigenes into different categories. In the molecular function category, the genes associated with binding, catalytic, and transporter activities were the most abundant groups. In the biological process category, most genes were involved in cellular process, metabolic process, and singleorganism process. Cell, cell part, and membrane were the most prevalent in the cellular component category (Figure S1).

#### Identification and Analysis of OBP Genes

A total of 46 putative OBP genes (BodoOBP1-46) were identified in the antennae and body transcriptome of adult B. odoriphaga (**Table 1**). Moreover, we also discovered three other putative OBP genes (BodoOBP47-49) from the fourth instar larval transcriptome of B. odoriphaga (unpublished data). Forty-eight of the 49 OBP genes (except for BodoOBP32) have intact open reading frames (ORFs) with lengths ranging from 378 to 759 bp (**Table 1**). Nearly all full-length OBPs had a predicted signal peptide (a signature of secretory proteins) at the Nterminal region except for BodoOBP22/25. All 49 OBPs had the predicted domains of pheromone/general odorant binding protein (PhBP or PBP\_GOBP) (InterPro: IPR006170) (Table S8). Based on the number and location of the conserved cysteines, all BodoOBPs could be divided into the following three groups: Minus-C OBPs group (BodoOBP14/23/26/31/33/41/42/43/44), Plus-C OBPs group (BodoOBP19/34), and the remaining OBPs belong to Classic OBPs group (Figure S2).

Gene expression levels of all 46 OBPs identified from antennae and body transcriptomes were assessed using FPKM-values, represented in a heatmap (**Figure 1**). The three repetitions of each tissue (FA, MA, FB, and MB) were clustered together, indicating that the results are stable and repeatable. Based on the expression levels in different tissues, all 46 OBP genes were clustered into 4 groups. Cluster analysis revealed that 20 OBP genes (Cluster 1) have similar expression patterns and were relatively high in the female and male antennae (FA and MA). Four and fourteen OBPs were more highly expressed in the FB (Cluster 3) and MB (Cluster 4), respectively. Moreover, the remaining eight OBPs were relatively highly expressed in not only the FA and MA but also the MB (Cluster 2) (**Figure 1**).

#### Identification and Analysis of CSP Genes

We have identified five putative CSP genes (BodoCSP1-5) from the antennae, body and larval transcriptome of B. odoriphaga. All the CSP genes have intact ORFs with lengths ranging from 327 to 708 bp, and with predicted signal peptide sequences at the Nterminus (**Table 2**). All BodoCSPs had typical structural features of insect CSPs with four conserved cysteines (Figure S3) and a conserved OS-D domain (olfactory system of D. melanogaster) (InterPro: IPR005055) (Table S9).

Gene expression levels of all five CSPs in different tissues were assessed by FPKM-values. BodoCSP3 and BodoCSP5 were significantly higher expressed in the female and male antennae (FA and MA), BodoCSP1 and BodoCSP2 were relatively highly expressed in the MB, and BodoCSP4 exhibited similar expression levels in different tissues (**Table 2**).

#### Phylogenetic Analysis of *B. odoriphaga* OBP and CSP Genes

A phylogenetic tree of 280 OBPs from 4 Diptera species (B. odoriphaga, D. melanogaster, A. gambiae, and A. aegypti) was constructed using the protein sequences to reveal the diverging relationships among them (**Figure 2**). Some pairs of BodoOBPs are paralogous genes, such as BodoOBP26/33, BodoOBP4/20, BodoOBP1/2, BodoOBP18/46, BodoOBP22/25, BodoOBP10/12, BodoOBP3/45, BodoOBP16/24, BodoOBP17/47, BodoOBP23/43, and BodoOBP31/41. All of these paralogous genes showed very high bootstrap values, which may indicate that these genes are the result of a recent gene duplication event within the B. odoriphaga genome. Moreover, 2 putative Plus-C OBPs (BodoOBP19 and 34) were clustered into the Plus-C OBP group with the 50 Plus-C OBPs from the other Diptera insect, and 7 putative Minus-C OBPs (BodoOBP14/23/31/41/42/43/44) were clustered into the Minus-C OBP group with 5 Minus-C OBPs from D. melanogaster, suggesting their different evolutionary relationships compared to the classic OBPs (**Figure 2**). In addition, BodoOBP13/22/25 were

clustered with the DmelOBP76a (LUSH, an OBP with binding affinity to the pheromone), and BodoOBP1/2/4/8/20/26/28/33 were clustered with DmelOBP83a/83b (OS-E/OS-F, an OBP group co-expressed with LUSH and associated with pheromone detection) (**Figure 2**), indicating that they might have a similar function in the detection of candidate pheromones in B. odoriphaga.

The neighbor-joining tree of CSPs was conducted using 5 putative BodoCSPs and 92 CSPs from 6 other Diptera species (D. melanogaster, A. gambiae, A. sinensis, A. aegypti, C. quinquefasciatus, and D. antiqua) (**Figure 3**). Five putative BodoCSPs were scattered into five subgroups (Groups 1–5), where each group included one BodoCSP. Moreover, four DmelCSPs were scattered into four subgroups (Groups 1–4), with one DmelCSP in each group (**Figure 3**). Almost every group included one or more CSPs from each Dipteran species, suggesting that the CSP gene has been highly conserved among different Dipteran insects.

## Motif Pattern Analysis of OBPs and CSPs

The motif pattern analysis results showed that 68 different motif patterns were observed in the 318 OBPs, and 195 OBPs (61.32%) had the most common five motif-patterns. Eighty-six of them had the most common motif-pattern 4-1-2, fifty-three OBPs only had motif 1, and thirty-six OBPs had the motif-pattern 1-2 (**Figure 4**). The motif pattern analysis results of 138 CSPs of Diptera insects showed that 8 different motif patterns were found, suggesting that CSPs were more conserved than the OBPs. In the 8 different motif patterns, 123 CSPs (89.13%) had the most common three motif patterns: 93 CSPs had motif pattern 8-5-6-1-3-2-4-7, 16 CSPs had motif pattern 6-1-3-2-4, and 14 CSPs had motif pattern 5-6-1- 3-2-4-7 (Figure S4). The remaining 15 CSPs shared the 5 other different motif patterns.

## Transcript Expression Levels of *B. odoriphaga* OBPs

The transcript expression levels of 49 BodoOBP genes in female antennae (FA), male antennae (MA), legs (L), wings (W), heads (without antennae, H), and abdomens and thoraxes (AT) were analyzed by qRT-PCR. The results suggested that 22 OBP genes (BodoOBP1/2/4/5/6/7/8/10/11/12/13/15/18/20/22/24/26 /28/33/41/43/46) were significantly higher expressed in the antennae (FA or MA) (**Figures 5A,B**), and 9 of the 22 antennae-biased OBP genes (BodoOBP2/4/6/8/12/13/20/28/33) were predominantly expressed in the male antennae (MA) (**Figure 5A**). Moreover, nine BodoOBP genes (BodoOBP3/9/19/21/34/35/38/39/45) were intensively expressed in the legs (L) than in other tissues (**Figure 5C**), whereas five BodoOBP genes (BodoOBP17/30/19/21/34) were mainly detected in the wings (W) (**Figure 5D**). Three BodoOBP genes (BodoOBP14/23/31) were significantly higher expressed in the heads (H), and two BodoOBP genes (BodoOBP29/36) showed higher expression levels in the abdomens and thoraxes (AT) (**Figure 5E**). In addition, the remaining eight BodoOBP genes (BodoOBP16/25/27/40/42/47/48/49) were expressed in more than three tissues, or they showed no significant differences among different tissues (**Figure 6**).

## Transcript Expression Levels of *B. odoriphaga* CSPs

The quantitative expression levels of five BodoCSP genes in different tissues were characterized using qRT-PCR. The results showed that BodoCSP1 had higher expression levels in the legs (L) than in other tissues (**Figure 7**), BodoCSP2 was significantly higher expressed in the heads (H), and BodoCSP3 and BodoCSP5 were mainly expressed in antennae (FA and MA). Moreover, BodoCSP4 showed predominantly expression in the male antennae (MA) and higher expression in the female antennae (FA) and heads (H) (**Figure 7**).

FIGURE 4 | Motif analysis of Diptera OBPs. Parameters used for motif discovery were as follows: minimum width = 6, maximum width = 10, maximum number of motif to find = 8. The upper parts list the eight motifs discovered in the Diptera OBPs. The numbers in the boxes correspond to the numbered motifs in the upper part of the figure, where a small number indicates high conservation. The numbers on the bottom show the approximate locations of each motif on the protein sequence, starting from the N-terminus. The protein names and sequences of the 318 OBPs from different Diptera species are listed in Table S4.

FIGURE 5 | Transcript levels of tissue-specific OBP genes in different tissues of *B. odoriphaga*. FA, female antennae; MA, male antennae; L, leg; W, wing; H, head (without antennae); AT, abdomen and thorax. (A) MA-specific, (B) antennae-specific, (C) L-specific, (D) W-specific, (E) H- and AT-specific. Two reference genes, RPS15 (ribosomal protein S15) and RPL18 (ribosomal protein L18) were used for normalizing OBP gene expression and to correct for sample-to-sample variation. Transcript levels were normalized to those of AT. The standard error is represented by the error bar, and the different lower cases above each bar indicate significant differences (*P* < 0.05).

## DISCUSSION

In the present study, we sequenced and analyzed the transcriptomes of antennae and bodies of adult B. odoriphaga (female and male), and searched for OBP and CSP genes from the transcriptomes of adults and larvae (our unpublished data). In total, we identified 49 OBP and 5 CSP genes in B. odoriphaga. The number of OBPs in B. odoriphaga was similar to the number in D. melanogaster (52), D. simulans (52), Episyrphus balteatus (49), and Eupeodes corollae (44) (Vieira and Rozas, 2011; Wang et al., 2017). Meanwhile, the number of OBPs in B. odoriphaga was greater than in some other Dipteran agricultural pests. For example, 15 OBPs were found in Delia antiqua, 20 in Delia platura, 20 in Bactrocera dorsalis, 32 in Mayetiola destructor Say, and 26 in Sitodiplosis mosellana (Andersson et al., 2014; Gong et al., 2014; Ohta et al., 2014, 2015; Liu et al., 2016) (**Figure 8**). There are likely multiple reasons responsible for identifying so many OBP genes in our study. First, this pest has a wide range of host plants (such as chive, shallot, garlic, cabbage, and mushrooms) (Ma et al., 2013), which might result in an increase in the number of OBP genes for detecting various odor molecules in a complex environment. Second, OBP genes were identified not only from the adult antennae transcriptome but also from the adult body and larval transcriptomes. If we solely identified OBP genes from the antennae transcriptome, the "Cluster 3" and "Cluster 4" genes (18 OBP genes) (**Figure 2**) and 3 larval transcriptome OBP genes may not have been identified. Additionally, previous studies have shown that the sequencing depth of different sequencing platforms will influence the number of identified OBP genes (Gu et al., 2015; Cui et al., 2017). The FPKM-values of 13 OBP genes were lower than 25 in the antennae and body transcriptomes of B. odoriphaga, which suggests that the sequencing depth of the Hiseq 4000 sequencing platform was superior, and this may be another reason for the identification of so many OBP genes in the present study. In addition, we identified five CSP genes in B. odoriphaga, and this number is very close to the number of CSP genes in D. melanogaster (4), D. simulans (4), B. dorsalis (5), and E. balteatus (6) (Vieira and Rozas, 2011; Liu et al., 2016; Wang et al., 2017). Compared with the OBP genes (mean value: 53.65), only a small number of CSP genes (mean value: 10.25) were detected in 17 species of Diptera insects (**Figure 8**), which is due to the evolutionary pattern in the CSP gene family and is less dynamic than in the OBP gene family (Vieira and Rozas, 2011). In addition, previous studies demonstrated that the C-patterns of OBPs and CSPs are similar among different insect Orders, whereas the motif-patterns are different (Zhou, 2010; Gu et al., 2015; He et al., 2017). For example, the motif-patterns between Dipteran and Lepidopteran GOBPs are different (Xu et al., 2009). Our present study also found that the motif-patterns among Dipteran OBPs were different, this is because the C-patterns of OBPs determines their crucial conserved structure, and motif-patterns fine-tune their specific functions (Xu et al., 2009).

The tissue expression profiles of chemosensory genes may be indicative of their biological functions, and they contribute to our understanding of the molecular mechanism of insect olfaction (He et al., 2011; Gu et al., 2015; Yuan et al., 2015). Various investigations have suggested that a high percentage of OBP genes are expressed in the antennae of insects, and antennae-enriched OBPs play crucial roles in detecting sex pheromones and host volatile compounds (Gong et al., 2014; Brito et al., 2016). In the current study, 22 of 49 BodoOBPs were uniquely or primarily expressed in the antennae compared to other tissues (**Figures 5A,B**). Among the 22 antennaeenriched OBPs, 9 were specifically expressed in male antennae (BodoOBP2/4/6/8/12/13/20/28/33) and might have potential functions in sex pheromone detection. Moreover, a phylogenetic analysis of OBPs suggested that BodoOBP13 clustered with the 11-cis-vaccenyl acetate binding PBP DmelOBP76a (LUSH) (Ha and Smith, 2006), and BodoOBP2/4/8/20/28/33 clustered together with DmelOBP83a/83b, an OBP group associated with the detection of volatile pheromones in D. melanogaster (Shanbhag et al., 2001a; Siciliano et al., 2014) (**Figure 2**). Hence, our results suggest that these

proteins (BodoOBP2/4/8/13/20/28/33) may be involved in the detection of sex pheromones in B. odoriphaga. In addition, 13 other OBPs that were highly expressed in the antennae (BodoOBP1/5/7/10/11/15/18/22/24/26/41/43/46) might be associated with functions in general host odorant perception.

Although the majority of OBPs are specifically expressed in antennae, it has become clear that many OBPs are enriched in non-antennal tissues and play key roles in olfactory or gustatory perception (Yasukawa et al., 2010; Sparks et al., 2014; Sun et al., 2017). For instance, two OBP genes (OBP57d and OBP57e) in Drosophila species were co-expressed in the taste sensilla of the leg, and these contribute to the perception of octanoic acid and the location of host plants (Yasukawa et al., 2010). Previously it was demonstrated that AlinOBP11 is predominately expressed in adult legs of Adelphocoris lineolatus and has a crucial role for detection of non-volatile secondary metabolites of host plants (Sun et al., 2016, 2017). In the present study, qRT-PCR results show that nine BodoOBPs (BodoOBP3/9/19/21/34/35/38/39/45) were significantly higher expressed in the legs (**Figure 5C**), and the transcript abundance (FPKM-value) of these genes in transcriptomes suggested that four of nine leg-specific OBPs (BodoOBP9/35/38/39) were male body (MB) enriched (**Figure 1**), implying that these four OBPs might also function in the recognition of sex pheromone compounds. The remaining five leg-specific OBPs may probably have a function to bind host plant volatile or non-volatile compounds. Previous studies have suggested that OBPs were also more highly expressed in gustatory organs, such as the heads and wings (Galindo and Smith, 2001; Shanbhag et al., 2001b; Jeong et al., 2013). In the present study, five OBP genes (BodoOBP17/30/32/37/44) were abundantly expressed in the wings, and three OBP genes (BodoOBP14/23/31) were enriched in the heads, suggesting that these genes might also participate in taste functions (Amrein and Thorne, 2005). In addition, two OBP genes (BodoOBP29/36) were significantly more highly expressed in the abdomens and thoraxes (AT), and heatmap results show that BodoOBP29/36 were specifically expressed in the female body (FB), indicating that these two genes might be involved in the synthesis and release of sex pheromones, or in the detection of egg-laying substrates (Zheng et al., 2013; Yuan et al., 2015).

CSPs belong to another type of small soluble proteins identified in multiple insect species (Brito et al., 2016; Pelosi et al., 2018). Compared with OBPs, CSPs are more conserved, often exhibiting 40–50% identical amino acid residues between orthologs from different species (Pelosi et al., 2006, 2018). In the present study, the results of MEME motif analysis showed that 123 CSPs (89.13%) had the three most common motifpatterns, whereas this number was only 55.03% in the OBPs. Moreover, the CSP-gene phylogeny suggested that most CSPs were scattered into five subgroups. Nearly every group included one or more CSPs from each Diptera species, which also suggests that CSPs are highly conserved among different Diptera insects. In olfactory perception, CSPs have similar functions to OBP. The hydrophobic pocket of CSPs can also recognize and transport chemical signals to chemoreceptors (Sun et al., 2014; Wang et al., 2016). Our results show that BodoCSP3/5 were antennaeenriched and might be involved in the chemosensory process. Moreover, previous studies have demonstrated that CSPs are not only associated with chemoreception but also participate in multiple physiological processes, such as limb regeneration of cockroaches, embryo maturation of honeybees, and larvae ecdysis of fire ants (Kitabayashi et al., 1998; Maleszka et al., 2007; Cheng et al., 2015; Pelosi et al., 2018). BodoCSP1 and BodoCSP2 were significantly more highly expressed in the legs and heads, respectively, and BodoCSP4 was more highly expressed in both the antennae and heads. We speculate that these CSPs might have other crucial physiological functions and require further functional verification.

## REFERENCES


In conclusion, we identified 49 putative OBP and 5 putative CSP genes in the adult (antennae and body) and larval transcriptomes of B. odoriphaga, and further tissue expression profiles and phylogenetic tree analyses indicated that some of these genes were antennae- or non-antennae-enriched and may play crucial roles in identifying hosts, locating mates and oviposition sites, avoiding natural enemies, and other important physiological processes. Based on the results of this work, future research will focus on the binding function of antennaeenriched OBPs with identified sex pheromones and host volatile components. The results of the present study provide a starting point to facilitate functional studies of these chemosensory genes in B. odoriphaga at the molecular level.

#### AUTHOR CONTRIBUTIONS

YZ, CZ, and WM designed the experiments. YZ, JD, and ZZ carried out the experiments. YZ, ZZ, FL, and WM analyzed the data. YZ, ZZ, FL, and WM drafted the manuscript. All authors approved the final version of the manuscript.

#### ACKNOWLEDGMENTS

This study was supported by grants from the Natural Science Foundation of Shandong Province (ZR2018MC019), the National Natural Science Foundation of China (Grant No. 31501651), and the Special Fund for Agro-scientific Research in the Public Interest from the Ministry of Agriculture of China (201303027). We would like to thank Dr. Weiguang Zhang (Shandong Agricultural University, Tai'an, China) for his guidance and assistance in photographing the insects.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00107/full#supplementary-material


transcriptome of Adelphocoris suturalis Jakovlev. Comp. Biochem. Phys. D. 24, 139–145. doi: 10.1016/j.cbd.2016.03.001


(Bactrocera dorsalis). PLoS ONE 11:e0147783. doi: 10.1371/journal.pone. 0147783


Holotrichia oblita Faldermann (Coleoptera: Scarabaeida). PLoS ONE 9:e107059. doi: 10.1371/journal.pone.0107059


proteins from expressed sequence tags in insects. BMC Genomics 10:632. doi: 10.1186/1471-2164-10-632


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zhao, Ding, Zhang, Liu, Zhou and Mu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# iTRAQ-Based Comparative Proteomic Analysis Reveals Molecular Mechanisms Underlying Wing Dimorphism of the Pea Aphid, *Acyrthosiphon pisum*

#### *Edited by:*

*Peng He, Guizhou University, China*

#### *Reviewed by:*

*Hao Guo, Institute of Zoology (CAS), China Patrizia Falabella, University of Basilicata, Italy Haonan Zhang, University of California, Riverside, United States Julian Chen, Institute of Plant Protection (CAS), United States*

> *\*Correspondence: Liping Ban liping\_ban@163.com*

#### *Specialty section:*

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

*Received: 28 April 2018 Accepted: 09 July 2018 Published: 07 August 2018*

#### *Citation:*

*Song L, Gao Y, Li J and Ban L (2018) iTRAQ-Based Comparative Proteomic Analysis Reveals Molecular Mechanisms Underlying Wing Dimorphism of the Pea Aphid, Acyrthosiphon pisum. Front. Physiol. 9:1016. doi: 10.3389/fphys.2018.01016* Limei Song1,2, Yuhao Gao<sup>3</sup> , Jindong Li 1,2 and Liping Ban1,2 \*

*<sup>1</sup> State Key Laboratory of Animal Nutrition, College of Animal Science and Technology, China Agricultural University, Beijing, China, <sup>2</sup> Department of Grassland Science, College of Animal Science and Technology, China Agricultural University, Beijing, China, <sup>3</sup> Affiliated High School of Peking University, Beijing, China*

Wing dimorphism is a widespread phenomenon in insects with an associated trade-off between flight capability and fecundity. Despite the molecular underpinnings of phenotypic plasticity that has already been elucidated, it is still not fully understood. In this study, we focused on the differential proteomics profiles between alate and apterous morphs of the pea aphid, *Acyrthosiphon pisum* at the fourth instar nymph and adult stages, using isobaric tags for relative and absolute quantitation (iTRAQ) in a proteomicbased approach. A total of 5,116 protein groups were identified and quantified in the three biological replicates, of which 836 were differentially expressed between alate and apterous morphs. A bioinformatics analysis of differentially expressed protein groups (DEPGs) was performed based on gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG). KEGG enrichment analysis showed that DEPGs mainly participated in energy metabolism, amino acid biosynthesis and metabolism, and signal sensing and transduction. To verify the reliability of proteomics data, the transcriptional expression of 29 candidates of differentially expressed proteins were analyzed by quantitative real-time PCR (qRT-PCR), showing that 26 genes were consistent with those at proteomic levels. In addition, differentially expressed proteins between winged and wingless morphs that were linked to olfactory sense were investigated. Quantitative real-time PCR revealed the tissue- and morph-biased expression profiles. These results suggested that olfactory sense plays a key role in wing dimorphism of aphids. The comparative proteomic analysis between alate and apterous morphs of the pea aphid provides a novel insight into wing development and dimorphism in aphids and will help facilitate our understanding of these concepts at molecular levels.

Keywords: wing dimorphism, Acyrthosiphon pisum, migration, Homoptera, iTRAQ, proteomics, olfactory sense

## INTRODUCTION

Phenotypic plasticity is a life history strategy of organisms, allowing them to adapt to various environmental conditions (Hall, 1999; West-Eberhard, 2003). Polyphenism is an extreme phenomenon of phenotypic plasticity in which multiple discrete phenotypes are produced by the same genotype in developing organisms in response to extrinsic factors (Nijhout, 1999, 2003). Wing polyphenism of insects has been considered to contribute to their diversity and evolutionary success (Roff, 1990; Dudley, 2002) and has evolved in numerous insect taxa, including those from the orders Coleoptera, Diptera, Heteroptera, Homoptera, Hymenoptera Lepidoptera, Orthoptera, Psocoptera, and Thysanoptera (Zera et al., 1997; Whitman and Ananthakrishnan, 2009). Wing polymorphic insects exhibit common dispersal and non-dispersal morphs (Roff, 1986; Braendle et al., 2006). Dispersal morphs of insects, with long wings and wing musculature, are capable of longdistance flight and migration to new habitats with fresh resources from deteriorated environments (Harrison, 1980; Roff, 1990), whereas short-wing or wingless morphs, without flight muscles, produce offsprings earlier and have greater reproductive output relative to dispersal morphs (Harrison, 1980; Zera et al., 1997; Simpson et al., 2011). In other words, wing dimorphism involves trade-offs between flight capability and other traits (Zera et al., 1997; Simpson et al., 2011). Wing dimorphism has been studied across a wide range of wing-polymorphic insect species, such as short-/long-winged morphs in crickets (Zhao and Zera, 2002), migratory locusts (Simpson and Sword, 2009; Tanaka and Nishide, 2012), and planthoppers (Denno et al., 1989; Xue et al., 2010), as well as wingless (apterous)/winged (alate) morphs in aphids (Brisson et al., 2007; Xu et al., 2011; Yang et al., 2014; Shang et al., 2016; Vellichirammal et al., 2016).

Wing dimorphism in aphids is associated with their complex life cycle (Brisson, 2010). Several aphid species exhibit clear differences between alate and apterous morphs. In addition to the presence or lack of wings and wing musculature, the differences can also be found in morphological, physiological, behavioral, and life cycle aspects. Alate morphs not only possess wings and flight muscles but also have more extensive sclerotization of heavier sclerotized head and thorax, more developed compound eyes, ocelli, larger numbers of secondary rhinaria on their antennae, and some species also have larger siphunculi and cauda (Kring, 1977; Miyazaki, 1987; Tsuji and Kawada, 1987a; Ishikawa and Miura, 2007) compared with apterous morphs. In addition, winged morphs are also more resistant to starvation, have a longer life, and have a more elaborate sensory system for flight navigation and for detecting host plants (Tsuji and Kawada, 1987b; Hazell et al., 2005). Olfaction plays a key role in the perception of chemical signals in insects.

The pea aphid, Acyrthosiphon pisum (Harris) (Homoptera: Aphididae), is an prominent sap-sucking pest in several species of legumes (Fabaceae) worldwide, including pea, clover, alfalfa, and broad bean (Blackman and Eastop, 2000), causing damage to the host plant by feeding on their phloem tissue directly as well as transmitting many viruses indirectly (Van Emden and Harrington, 2017). The pea aphid is a good study model organism with alate and apterous morphs that reflect the tradeoff of dispersal and fecundity. The transition from the fourth instar winged-nymph to alate adult is the key period in aphid wing development (Brisson et al., 2010; Shang et al., 2016). The publication of the whole genome sequence of A. pisum provides a platform for better understanding the wing dimorphism of aphids at the molecular level (The International Aphid Genomics Consortium, 2010).

Wing dimorphism of aphids has been studied extensively to elucidate the molecular mechanism, including the analysis of the gene expression between winged and wingless adults (Brisson et al., 2007; Yang et al., 2014; Shang et al., 2016; Vellichirammal et al., 2016). However, these studies have mainly been performed at genomic and transcriptomic levels and focused on the adult stage. Proteomic analyses of wing dimorphism in insects are lacking. In recent years, advances in mass-spectrometry (MS)-based approaches for proteomics have enabled us to investigate the mechanisms of the wing dimorphism of insects at proteomic levels. Isobaric tags for relative and absolute quantitation (iTRAQ) is an isobaric labeling approach combined with liquid chromatography and tandem mass spectrometry to identify and quantify proteins (Cha et al., 2012; Ren et al., 2016) and has been increasingly used in the past few years (Brewis and Brennan, 2010; Unwin, 2010). The objectives of this study are (1) to investigate the differential protein expression profiles between alate and apterous morphs of the pea aphid at different developmental stages and (2) to investigate the potential functions of chemoreception genes in wing dimorphism of aphids. Our study provides a novel insight into the molecular mechanisms of wing dimorphism in aphids.

## MATERIALS AND METHODS

#### Insect Rearing and Sample Collection

The pea aphid A. pisum used in this study was established in 2014 from a single female adult aphid collected from an alfalfa field at China Agricultural University, Beijing, China. Aphids had been established in the laboratory for more than a year before subsequent experiments. Stock colonies were maintained on vetch seedlings (Vicia faba Linnaeus, 1753) in a climatecontrolled environment at 20 ± 1 ◦C with 70–75% relative humidity and a photoperiod of 16: 8h (Light:Dark). Winged morphs were induced under high-density conditions after being transferred to new host plants (Sutherland, 1969; Ishikawa et al., 2008). The impact of rearing conditions lasts over two or three generations (MacKay and Wellington, 1977). The specimens including wingless adults (AWL), wingless fourth instar nymphs (N4WL), winged adults (AW), and winged fourth instar nymphs (N4W) were collected with three replicates (200 aphids for each sample) and frozen in liquid nitrogen immediately and then stored at −80◦C for future use. Tissues (antennae, heads without antennae, legs, thoraxes, abdomens, and wings) from alate adults were dissected under the microscope and individuals from each development stage of aphid (first instar nymphs, second instar nymphs, wingless third instar nymphs, wingless fourth instar nymphs, wingless adults, winged third instar nymphs, winged fourth instar nymphs, and winged adults) were collected. Samples were stored at −80◦C until needed.

#### Protein Preparation and iTRAQ Labeling

Proteins were obtained by grinding samples in liquid nitrogen, dissolved in moderate lysis buffer (7 M carbamide, 2 M thiocarbamide, 0.1% CHAPS), suspended for several seconds, followed by ultrasonication (0.5 s on, 2 s off), and then incubation at room temperature for 30 min before being centrifuged at 15,000 × g for 20 min at 4◦C. The supernatant was collected and a Bradford protein assay (Sigma) was used to determine total protein concentrations (Bradford, 1976). The supernatant proteins were lyophilized and then kept at −80◦C for further analysis.

Protein digestion was conducted according to the filter-aided sample preparation (FASP) procedure described in a previous study (Wi´sniewski et al., 2009). In brief, for each sample, 100 µg of proteins were solubilized in 10 µl reducing reagent at 37◦C for 60 min. Then 2 µl cysteine-blocking reagent was added at room temperature for 30 min followed by centrifugation at 12,000 × g for 20 min. The filters were washed with 100 µl of dissolution buffer (Applied Biosystems, Foster City, CA, USA) and centrifuged at 12,000 × g for 20 min, which was repeated three times. Proteins were then in-solution digested with trypsin (Promega) according to the protein/trypsin ratio of 50:1 at 37◦C overnight. Then, the filter unit was transferred to a new tube and centrifuged at 12,000 × g for 15 min. The filtrate was collected, and the peptide concentration was estimated by ultraviolet (UV) light spectral density at 280 nm (Wi´sniewski et al., 2009).

According to the manufacturer's protocol, iTRAQ labeling of the peptide samples was performed using iTRAQ reagent 4-plex kits (AB Sciex Inc., MA, USA). For every development stage of the pea aphid, three biological replicates were iTRAQ labeled. The iTRAQ tags for each sample were 114, 115, 116, and 117 (AB Sciex, Foster City, USA) (**Figure 1**).

## Reverse-Phase (RP) HPLC

The pooled iTRAQ mixtures were resuspended in buffer A (2% acetonitrile, 98% water with ammonia at pH 10), loaded onto a 4.6 × 250 mm Durashell-C18 column (150 Å, 5µm particles, Agela), and fractionated by high performance liquid chromatography (HPLC) (RIGOL, China). Peptides were eluted using an increasing acetonitrile gradient with buffer A and buffer B (98% acetonitrile, 2% water with ammonia). The flow rate was 0.7 ml/min and the elution gradient was as follows: 5–8% buffer B for 0–5 min, 8–18% buffer B for 5–35 min, 18–32% buffer B for 35–62 min, 32–95% buffer B for 62–64 min, 95% buffer B for 64–78 min and changed to 5% buffer B within 4 min. Fractions were collected every 1 min, pooled into 12 fractions by intervals, and dried by vacuum centrifugation. All samples were stored at −80◦C.

## LC-MS/MS Analysis

The peptide fragments from each sample were redissolved in 2% methyl alcohol and 0.1% formic acid and then centrifuged at 13,000 rpm for 10 min. LC-MS/MS was carried out using an EasynLC nanoflow HPLC system connected to a Q Exactive mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). The mass spectrometer was operated in the data-dependent mode with positive polarity at electrospray voltage of 2.0 kV. MS spectra (full scan) were acquired over a range of 350–1,800 m/z and resolving powers of the MS scan and MS/MS scan at 100 m/z were set as 70,000 and 17,500, respectively. In addition, MS automatic gain control (AGC) target was 3e6, and maximum injection time was 80 ms; MS2 AGC target was 2e4, and maximum injection time was 19 ms. Normalized collision energy (NCE) was 30% and dynamic exclusion was set to 18 s. Each sample was loaded onto Acclaim PepMap 100 C18 (2 cm × 100µm, 5µm C18) using an autosampler and then the sequential separation of peptides on Thermo Scientific EASY column (EASY-Spray column, 12 cm × 75µm, C18, 3µm) was accomplished with a gradient of buffer B (100% acetonitrile and 0.1% formic acid) at a flow rate of 350 nl/min with the following conditions: 6–9% buffer B for 0–8 min, 9–14% buffer B for 8–24 min, 14–30% buffer B for 24–60 min, 30– 40% buffer B for 60–75 min, 40–95% buffer B for 75–78 min, 95% buffer B for 78–85 min, and then changed to 6% buffer B within 1 min and equilibrated for 4 min.

#### Protein Identification and Quantification

The raw data were analyzed using Proteome Discoverer 1.4 software (Thermo Fisher Scientific) with Mascot search engine (Matrix Science, London, UK; version 2.2) to identify the proteins in a search of the protein database of A. pisum downloaded from the National Center for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov/) (download date March 7, 2016). For protein identification, search parameters were as follows: precursor ion mass tolerance ±15ppm; MS/MS tolerance ±20 mmu; two missed cleavages were allowed with the enzyme trypsin; carbamidomethylation (C) was set as fixed modification, and oxidation (M) and iTRAQ labeling (K, Y, and N-term) were set as dynamic modifications; peptides with peptide score ≥ 10 and false discovery rate (FDR) < 0.01. The protein identification contained at least one unique peptide. The proteome data was uploaded to the public repository iProX (ID: IPX0001238000).

#### Bioinformatics Analysis

GO annotation analysis, including molecular function, cellular component, and biological process of the differentially expressed proteins, was performed using Uniprot (http://www.uniprot. org/). The KEGG pathway database (http://www.genome.jp/ kegg/) was used to classify and group these differentially expressed proteins (Kanehisa et al., 2007). KEGG pathway and GO enrichment analysis of the differentially expressed proteins were performed, and the formula used was:

$$P = 1 - \sum\_{i=0}^{m-1} \frac{\binom{M}{i} \binom{N-m}{n-i}}{\binom{N}{n}}$$

where N represents the number of all identified proteins with a GO or a KEGG pathway annotation; n is the number of differential proteins in N; M is the number of proteins that are annotated to the specific GO term or pathway; and m is the

number of differential proteins in M. If p-value is below 0.05, the GO term or pathway was defined as a significant enrichment of differential proteins. The false discovery rate (FDR) was controlled by the Bonferroni step-down test to correct the p-value.

#### Real-Time Quantitative PCR

Quantitative real-time PCR primer pairs were designed using Primer 5.0 and primer sequences are listed in **Table S1**. Total RNA was extracted using a TRIzol kit (Invitrogen, USA) according to the manufacturer's instructions. The quantity of total RNA was measured using NanoDrop 2000 (Thermo Fisher Scientific), and the quality was assessed using 1.0% denaturing agarose gel electrophoresis. Complementary DNA was synthesized from 1000 ng RNA using PrimeScript RT reagent Kit with gDNA Eraser (Takara, Dalian, China). Quantitative real-time PCR was performed using the SYBR Premix Ex Taq kit (Tli RNaseH Plus) (Takara, Dalian, China) according to the manufacturer's instructions with the ABI 7500 Real Time PCR thermal cycler (Applied Biosystems) using the following cycle conditions: 95◦C for 30 s, then 40 cycles at 95◦C for 5 s and 60◦C for 34 s, a final cycle of 95◦C for 15 s, 60◦C for 60 s, and 95◦C for 15 s. Housekeeping genes, β-actin and 16S rRNA, were selected as reference genes to normalize the expression level of target genes using 2−11CT method (Livak and Schmittgen, 2001). All experiments were carried out using three biological replications and three technical replications. Differences in transcript expression in different tissues and developmental stages were analyzed with a one-way analysis of variance (ANOVA) using SPSS software (version 19.0; IBM, Armonk, NY, USA) followed by the least-significant difference (LSD) test. Differences were considered statistically significant at a p-value less than 0.05.

## RESULTS

## Identification and Quantification of Differentially Expressed Proteins Between Alate and Apterous Morphs of *A. pisum*

To investigate the differentially expressed proteins in alate and apterous morphs of A. pisum, quantitative iTRAQ labeling-based proteomic analysis was performed. A schematic representation of the experimental workflow is shown in **Figure 1**. Based on 4-plex iTRAQ proteomic labeling and LC-MS/MS analysis, a total of 5,116 protein groups were identified quantified in all experiments at the two developmental stages and three replicates (**Table S2**). Based on previous studies (Yang et al., 2013; Chen et al., 2016), DEPGs are defined based on a 1.2–1.5-fold change threshold. Among those proteins, only one protein in two or three replicates with fold changes ≥1.2 was defined up-regulated, or ≤0.83 was down-regulated. Following this criterion, a total of 563 DEPGs in alate fourth instar nymphs and 494 DEPGs in alate adults were detected, respectively. A total of 538 DEPGs were up-regulated under at least one developmental stage, including 168 DEPGs upregulated at two stages (N4W/N4WL and AW/AWL), and 200 and 168 DEPGs up-regulated at N4W/N4WL and AW/AWL, respectively. Among 309 down-regulated DEPGs under at least one development stage, 44 DEPGs were shared by two stages (AW/AWL and N4W/N4WL), whereas 149 and 114 DEPGs were down-regulated at N4W/N4WL and AW/AWL, respectively (**Figure 2**).

In general, three clearly different expression profiles under different wing morphs at fourth instar nymph and adult stages were generalized among 836 DEPGs and the summarized data of DEPGs are listed in **Tables S3**–**S5**: 168 DEPGs (20.1%) at two stages were up-regulated; 44 DEPGs (5.3%) at two stages were down-regulated; 9 DEPGs (1.1%) were up-regulated in N4W then down-regulated in AW (**Table S3**); 151 DEPGs (18.1%) were down-regulated and 191 DEPGs (22.8%) were up-regulated only in the N4W sample (**Table S4**), whereas 105 DEPGs (12.6%) were down-regulated and 168 DEPGs (20.1%) were up-regulated only in the AW sample (**Table S5**).

## Functional Enrichment Analysis of Differentially Expressed Proteins

To analyze the differentially abundant protein groups between wingless and winged aphids, all DEPGs were submitted to Uniprot for functional annotation and a GO category enrichment analysis was conducted. The GO annotation of proteins included biological process, molecular function, and cellular component and the detailed information is shown in **Figure 3**. For molecular function, the differentially abundant proteins of N4W/N4WL and AW/AWL were mainly enriched in structural constituents of ribosome and adenosine triphosphate (ATP)-binding proteins. In addition,

the DEPGs of AW/AWL were found enriched in odorantbinding. According to biological process, the differentially abundant protein groups between alate and apterous morphs were mainly assigned to translation and tricarboxylic acid cycle. The cellular component of DEPGs was categorized as the integral components of membrane, ribosome, nucleus, and mitochondrion.

To investigate the enrichment pathways of the DEPGs between winged and wingless morphs of aphids, KEGG analysis was performed. According to KEGG analysis, 33 pathways were enriched (p-value ≤ 0.05) in N4W/N4WL and the main KEGG functional classifications were oxidative phosphorylation, ribosome, biosynthesis of antibiotics, tricarboxylic acid (TCA) cycle, cardiac muscle contraction, fatty acid degradation, fatty acid metabolism, peroxisome proliferator-activated receptor (PPAR) signaling pathway, pyruvate metabolism, and glycolysis/gluconeogenesis (**Figure 4A**). In AW/AWL, 38 pathways were enriched (pvalue ≤ 0.05) and the main KEGG functional classifications of the DEPGs were oxidative phosphorylation, biosynthesis of antibiotics, ribosome, citrate cycle (TCA cycle), cardiac muscle contraction, biosynthesis of amino acids, fatty acid metabolism, fatty acid degradation, PPAR signaling pathway, glyoxylate and dicarboxylate metabolism, and pyruvate metabolism (**Figure 4B**). DEPGs between alate and apterous morphs involved in the PPAR signal pathway in KEGG is shown in **Figure 5** and DEPGs in this pathway were listed in **Table S6**.

#### Transcriptional Expression Analysis of Selected Proteins as Revealed by qRT-PCR

To evaluate the proteomic data and provide further information of the correlation between protein abundance and their mRNA expression patterns, qRT-PCR was performed to quantify the mRNA transcript level for 29 proteins including top 19 up-regulations in winged adult and top 10 up-regulations in wingless adult. The result showed that expression profiles of 26 genes were consistent with the proteomic changes and the mRNA levels (**Table 1**).

According to GO analysis, the DEPGs of AW/AWL were enriched in odorant-binding. The description and fold changes

of four chemosensory proteins (CSPs) and four odorant-binding proteins (OBPs) that were differentially expressed between alate and apterous morphs are summarized in **Table 2**. qRT-PCR was performed to quantify the mRNA transcript level for them. In addition, the expression patterns of OBP3, OBP6 to OBP13 between alate and apterous A. pisum adults were investigated at mRNA levels (**Figure S1**). The results showed that OBP6 and OBP10 were significantly up-regulated in alate adults. Based on combined analyses of proteomics and mRNA expression profiles, up-regulated genes in alate morph aphids including three CSPs and two OBPs were further investigated in different body parts, instars, and wing morphs (**Figure 6**). Compared with apterous adults, transcriptional expression levels of three CSPs were significantly higher in alate adults. The results showed similar expression patterns as indicated by the proteomics analysis. CSPORF1 genes were sharply increased from the third instar nymphs to adults of alate morphs. An expression peak was present in the alate adults and apterous third instar nymphs, for CSPORF2 and CSPORF5 genes, respectively. The transcriptional expression profiles of OBP6 and OBP10 in alate aphids shared the similar patterns with CSPORF1. Transcription profiles for these genes were also determined in different body parts. For OBP6, OBP10, and CSPORF2, the highest transcript levels were observed in the antennae. The expression levels of CSPORF1 and CSPORF5 were both significantly higher in the legs, followed by high expression in the wings.

#### DISCUSSION

Flight capability is a vital feature in insects and plays an important role in their evolutionary success. Flight benefits dispersal capacity balanced against potential metabolic, reproductive, and survival costs (Langellotto et al., 2000; Castañeda et al., 2010). Here, iTRAQ-coupled 2D LCMS/MS was used to analyze the molecular mechanisms of wing development and dimorphism in A. pisum. By analyzing the differential expression of proteins between alate and apterous aphids at the fourth instar nymph and adult stages, a total of 836 DEPGs were obtained between alate and apterous morph aphids, of which 538 DEPGs were up-regulated under at least one developmental stage, whereas 309 were down-regulated. In both stages, most of the differentially expressed proteins showed higher levels in alate morphs than in apterous morphs. Those genes associated with flight capability are more than those associated with reproduction and this discrepancy is a reflection of trade-offs between dispersal capability and reproductive structures. The number of DEPGs between winged fourth instar nymphs and winged adults were similar.

#### Differentially Expressed Proteins Involved in Energy Metabolism

The trade-offs between flight capability and other traits result in energy allocation discrepancy between alate and apterous morphs (Roff and Fairbairn, 1991; Zera et al., 1998). In this study, DEPGs of different morphs mainly participated in energy metabolism pathways including oxidative phosphorylation (path: ko00190), citrate cycle (TCA cycle) (path: ko00020), fatty acid metabolism (path: ko01212), fatty acid degradation (path: ko00071), glycolysis/gluconeogenesis (path: ko00010), pyruvate metabolism (path: ko00620), and propanoate metabolism (path: ko00640). These results reflected active and complex protein abundance change patterns between alate and apterous aphids at the molecular level. Differential energy allocation is a crucial part of trade-offs between dispersal capability and reproduction, with alate morphs investing energy into building wings and flight muscles to maintain energetically costly flight performance rather than developing quickly and maintaining high levels of offspring production as in apterous morphs (Brisson, 2010). Most DEPGs were related to energy metabolism and showed significantly up-regulated expression levels in alates (**Figure 4**), which suggested that winged morphs require more energy and have higher metabolic costs than wingless morphs. In flying, organisms incur two costs: developing a flight apparatus and fueling for flight (Dixon and Kindlmann, 1999). The results showed that up-regulated proteins involved in energy production at fourth instar nymph and adult stages were different, and these results agree with the metabolic requirements for the construction of wings and muscles as fourth instars and the capability of flight and the maintenance of flight muscles as adults (Zera and Denno, 1997).

Our research suggested that lipids provide resources for wing development and dispersion of alate aphids. Lipids are mainly stored in the fat body of insects as triacylglycerol and are used as fuel in flight muscles (Chino and Downer, 1982). The content of triacylglycerol is higher in alate morphs of aphids (Dixon et al., 1993; Itoyama et al., 2000; Xu et al., 2011), which is the same situation as in long-winged planthoppers (Itoyama et al., 1999) and crickets (Zera et al., 1994; Zera and Larsen, 2001). Correspondingly, well-developed flight muscles are reported for winged aphids (Ishikawa and Miura, 2007), long-winged crickets (Mole and Zera, 1993; Tanaka, 1993; Zera et al., 1997), and firebugs (Socha, 2006). In our study, the three proteins that showed significantly higher expression levels in alate adults




Song et al. Wing Dimorphism in Aphids

associated with fatty acid metabolism and degradation were short-chain specific acyl-CoA dehydrogenase (SCAD), mediumchain specific acyl-CoA dehydrogenase (MCAD), and isocitrate dehydrogenase (NAD) subunit alpha (IDH3A) (**Table S5**). MCAD catalyzes the initial step of fatty acid beta-oxidation, and SCAD is a key enzyme of fatty acid β-oxidation. This result is consistent with the previous observation that the content of total lipid, triglyceride, and free fatty acid was dramatically higher in winged adults (Beenakkers et al., 1985; Itoyama et al., 2000; Shi et al., 2010). In apterous adults, only fatty acid desaturase-like (FADS) and stearoyl-CoA desaturase-like (SCD) related to fatty acid metabolism were up-regulated.

Glycogen also provides resources for alate aphids besides lipids, which is consistent with the report in alate brown citrus aphid, Toxoptera citricida (Shang et al., 2016). Early reports showed that both lipid and glycogen are consumed during tethered flight of insects such as migratory locust (Locusta migratoria) (Worm and Beenakkers, 1980), planthopper (Nilaparuara lugens) (Padgham, 1983), and Aphis fabae Scop (Cockbain, 1961). Glycogen is used during early flight, and fat is the principal fuel after the first hour (Cockbain, 1961). Pyruvate is a key intersection in the network of metabolic pathways and is known as the "hub" of carbohydrate, fatty acids, and proteins (Tatusov et al., 2003; Simpson et al., 2011). Pyruvate can be made from glucose through glycolysis (Simpson et al., 2011). In this study, DEPGs that were involved in glycolysis/gluconeogenesis and pyruvate metabolism were almost up-regulated in alate morphs aphids. However, these genes were different between winged fourth instars and winged adults. Pyruvate dehydrogenase E1 component subunit alpha (PDHA1) gene, with higher expressions in winged fourth instars and adults (**Table S3**), was involved in pyruvate metabolism, whereas acylphosphatase gene only had higher expression in alate adults. Our results suggested glycogen and lipid not only provide energy resources for flight in adults but also for development of wing and muscle of fourth instars (Shang et al., 2016).

#### Differentially Expressed Proteins Involved in Amino Acid Biosynthesis and Metabolism

In this study, DEPGs mainly participated in amino acid biosynthesis and metabolism, including ribosome (path: ko03010); biosynthesis of amino acids (path: ko01230); valine, leucine, and isoleucine degradation (path: ko00280); tryptophan metabolism (path: ko00380); spliceosome (path: ko03040); and RNA transport (path: ko03013). The ribosome is required for protein synthesis and plays vital roles in the growth and development of organisms (Zhu et al., 2017). Early reports showed gene products that are components of ribosomes were over-represented in A. pisum (Brisson et al., 2007). Our study found that more ribosomal protein was detected in the alates, which implied that ribosomal proteins play important roles in development and dispersal of winged aphids. Many proteins in the ribosome were differentially expressed between alates and apterous aphids, including neurofilament heavy polypeptide

of *A. pisum*. N1, first instar nymphs; N2, second instar nymphs; N3WL, wingless third instar nymphs; N4WL, wingless fourth instar nymphs; AWL, wingless adults; N3W, winged third instar nymphs; N4W, winged fourth instar nymphs; AW, winged adults; An, antennae; L, legs; H, heads; T, thoraxes; Ab, abdomens; W, wings. Lowercase letter above each bar indicates a significant difference (*P* < 0.05) in mean transcript levels which were compared using one-way ANOVA, followed by the least-significant difference (LSD) method.

and ribosomal protein small and large subunits. Ribosomal protein L19e-like, peptidyl-prolyl cis-trans isomerase-like, and 40S ribosomal protein S21-like, which were highly expressed in wingless fourth instars and adults (**Tables S3**, **S4**) might be also important for the development and reproduction of wingless morphs (Xue et al., 2010). The functions and mechanisms of ribosomal proteins between the two morphs remain largely unknown, and our data of ribosome proteins expression profiles in the pea aphid can assist in understanding them in future.

#### Differentially Expressed Proteins Involved in Signal Sensing and Transduction

In this study, we found that PPAR signaling pathway (path: ko03320) (**Figure 5**), which is thought to participate in lipid metabolism (Schoonjans et al., 1996), was the co-enriched pathway at alate fourth instar nymphs and adults and might play a critical role in the wing development and dispersal of aphids. Early reports found that PPAR signaling pathway was significantly up-regulated in dispersing morphs (Xue et al., 2010; Shang et al., 2016). In the current study, most DEPGs in this pathway were up-regulated in alates (**Table S6**) and are involved in facilitating lipid metabolism and gluconeogenesis to increase metabolism of energy sources. Fatty acid binding protein (FABP), which is a small cytosolic protein abundantly found in the muscle and transports lipophilic molecules from the outer cell membrane to certain intracellular receptors, exhibits significantly higher expression in alate morphs (Tan et al., 2002). Fatty acid transport proteins (FATPs) are a family of six integral membrane proteins with an extracellular/luminal N-terminal and C-terminal domain with fatty acyl-CoA synthetase activity. In the future, to reveal the mechanism of wing development of aphids, more genes and proteins including those in the PPARrelated metabolic pathways require further study.

For flight navigation and detecting new habitats, alate aphids have a more detailed sensory system (Tsuji and Kawada, 1987b; Hazell et al., 2005). In this study, proteins involved in chemoreception are also significantly different between winged and wingless morphs. Early reports suggested that alarm pheromone, a volatile compound released from aphid colonies' cornicles due to high-density triggers or predator attacks, could induce aphids to produce winged dispersal morphs (Kunert et al., 2005; Verheggen et al., 2009; Hatano et al., 2010). Studies showed that an unidentified "spacing pheromone" released from crowded aphids could change their behaviors (Pettersson et al., 1995). In some aphid species, antennae act as a pivotal part in the perception of tactile signals (Johnson, 1965; Lees, 1967; Sutherland, 1969). OBPs are small, water-soluble proteins abundant in sensillar lymph of insect antennae and other non-sensory organs that transport hydrophobic semiochemicals (pheromones and plant volatiles) through the sensillar lymph and finally reach sensory dendrites, where released chemicals activate membrane-bound odorant receptors (Brito et al., 2016). Like OBPs, CSPs belong to another family of small, soluble proteins (Brito et al., 2016). The abundance and diverse expression patterns of different CSPs suggest that they are involved in multiple functions in insects such as recognition of sex pheromones (Jacquin-Joly et al., 2001) and general odorants (Liu et al., 2012), development (Maleszka et al., 2007), and feeding (Liu et al., 2014). Based on combined analysis of transcriptome and proteome, higher expression levels of two OBP and three CSP genes in alate morphs were investigated. OBP6 and OBP10 were of higher expression in alate aphids, suggesting a possible role in wing development and migration. OBP6 had significantly higher expression in antennae in alate morphs, suggesting an olfactory role for this protein in discrimination (E)-β-farnesene, "spacing pheromone" or mediating the perception of molecules related to new host-plant location (Vogt et al., 1999; Sun et al., 2012; De Biasio et al., 2015; Xue et al., 2016). The expression pattern of OBP10 is similar to OBP6, which might have similar function. In addition, OBP6 was also abundantly expressed in heads (without antennae) of alate adults, suggesting a possible role in host-plant selection during migration (De Biasio et al., 2015). Besides OBPs, abundant expression of three CSP genes was also detected in alate morphs. CSPORF1 and CSPORF5, which were abundantly expressed in legs, followed by high expression in the wings of winged aphids, might be involved in contact with the plant, leaf surface characteristics, or the process of volatile reception and be indicative of mechanoreceptor or chemoreception sensilla on the legs and wings (Pettersson et al., 2007; Zhou et al., 2008; Yasukawa et al., 2010; Harada et al., 2012). CSPORF2, which is specifically expressed in antennae and only increased in winged adults, might be involved in chemoreception during migration (Ghanim et al., 2006; González et al., 2009; Xue et al., 2016). RNA interference (RNAi) and fluorescence competition assays should be used in the future to investigate the function of these genes.

#### CONCLUSIONS

In conclusion, this study is the first report to investigate protein expression profiles between winged and wingless aphids using a new proteomic profiling method iTRAQ-coupled 2D LC-MS/MS. A total of 836 differentially expressed protein groups were detected, in which 563 and 494 DEPGs were identified in alate aphids at fourth instar nymph and adult stages, respectively. Based on the GO and KEGG enrichment analysis, we concluded that olfactory senses have an important function in alate aphids and winged aphids using lipids and glycogen as fuel resources for wing development and migration. In addition, protein groups involved in the PPAR signaling pathway of aphids were found to play a crucial role in winged morphs. Although our report provides knowledge of some proteins associated with development and dispersion, gene function analysis is needed to further understand the roles of these proteins. Our findings may provide new clues for elucidating the molecular mechanisms underlying wing dimorphism in aphids.

## AUTHOR CONTRIBUTIONS

LS, YG, and LB designed the experiments. LS, YG, and JL preformed the experiments. LS and LB analyzed data and drafted the manuscript. LS, JL, and LB revised the manuscript. All authors read and approved the manuscript for final submission.

#### ACKNOWLEDGMENTS

This work was supported by Beijing Agriculture Innovation Consortium (BAIC09-2018) and the Natural Science Foundation of China (31372364).

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01016/full#supplementary-material

Figure S1 | Expression profiles of OBPs in adult of *Acyrthosiphon pisum* between alate and apterous morphs. Lowercase letter above each bar indicates a significant difference (*P* < 0.05) in mean transcript levels of alate *vs* apterous which were compared using one-way ANOVA, followed by the least-significant difference (LSD) method.

Table S1 | Primers of quantitative RT-PCR for the selected genes.

Table S2 | The proteins were identified and quantified.

Table S3 | The DEPs were identified and quantified both in winged fourth instar-nymphs and winged adults.

#### REFERENCES


Table S4 | List of up- or down-regulated proteins under only winged morphs at fourth instar-nymph stage.

Table S5 | List of up- or down-regulated proteins under only winged morphs at adult stage.

Table S6 | DEPs involved in PPAR signal pathway at fourth instar-nymph and adult stage.


Hall, B. K. (1999). Evolutionary Developmental Biology. Dordrecht: Kluwer.


morsitans is related to female host-seeking behaviour. Insect Mol. Biol. 21, 41–48. doi: 10.1111/j.1365-2583.2011.01114.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Song, Gao, Li and Ban. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identification of an Alarm Pheromone-Binding Chemosensory Protein From the Invasive Sycamore Lace Bug Corythucha ciliata (Say)

Fengqi Li <sup>1</sup> , Ningning Fu<sup>1</sup> , Du Li <sup>1</sup> , Hetang Chang<sup>2</sup> , Cheng Qu<sup>1</sup> , Ran Wang<sup>1</sup> , Yihua Xu<sup>1</sup> \* and Chen Luo<sup>1</sup> \*

1 Institute of Plant and Environment Protection, Beijing Academy of Agriculture and Forestry Sciences, Beijing, China, <sup>2</sup> Stowers Institute for Medical Research, Kansas City, MO, United States

#### Edited by:

Nicolas Durand, Université Pierre et Marie Curie, France

#### Reviewed by:

Herbert Venthur, Universidad de La Frontera, Chile Patrizia Falabella, University of Basilicata, Italy

#### \*Correspondence:

Yihua Xu xuyihua@baafs.net.cn Chen Luo luochen1010@126.com

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 18 December 2017 Accepted: 20 March 2018 Published: 06 April 2018

#### Citation:

Li F, Fu N, Li D, Chang H, Qu C, Wang R, Xu Y and Luo C (2018) Identification of an Alarm Pheromone-Binding Chemosensory Protein From the Invasive Sycamore Lace Bug Corythucha ciliata (Say). Front. Physiol. 9:354. doi: 10.3389/fphys.2018.00354 The spread of the exotic insect pest sycamore lace bug Corythucha ciliata (Say) is increasing worldwide. The identification of behaviorally active compounds is crucial for reducing the current distribution of this pest. In this study, we identified and documented the expression profiles of genes encoding chemosensory proteins (CSPs) in the sycamore lace bug to identify CSPs that bind to the alarm pheromone geraniol. One CSP (CcilCSP2) that was highly expressed in nymph antennae was found to bind geraniol with high affinity. This finding was confirmed by fluorescence competitive binding assays. We further discovered one candidate chemical, phenyl benzoate, that bound to CcilCSP2 with even higher affinity than geraniol. Behavioral assays revealed that phenyl benzoate, similar to geraniol, significantly repelled sycamore lace bug nymphs but had no activity toward adults. This study has revealed a novel repellent compound involved in behavioral regulation. And, our findings will be beneficial for understanding the olfactory recognition mechanism of sycamore lace bug and developing a push-pull system to manage this pest in the future.

Keywords: alarm pheromone, sycamore lace bug, chemosensory proteins, behavioral study, geraniol

#### INTRODUCTION

Sycamore lace bug, Corythucha ciliata (Say) (Heteroptera: Tingidae), is an important exotic invasive pest. This insect has spread to many countries and seriously affected city landscapes and disturbed people's lives (Halbert and Meeker, 1983; Wang et al., 2008; Ju et al., 2015). Currently, the management of this pest is dependent on insecticides, such as pyridines (Ju et al., 2015). The discovery of behaviorally active compounds is an efficient approach to developing pest management strategies (Foster and Harris, 1997). Examples of these types of compounds include alarm pheromones, which affect insect behavior, development and oviposition and are important for insects to defense against dangers such as natural enemies (Ono et al., 2003; Kunert et al., 2005; Dewhirst et al., 2010). In addition, alarm pheromones and other repellent compounds can enhance the efficiency of insecticides (Kuwahara et al., 2011). Thus, these types of compounds are important targets in pest management. The alarm pheromone of sycamore lace bug has been identified as geraniol, E-3,7-dimethyl-2,6 octadien-1-ol (Kuwahara et al., 2011). Geraniol is only detected in the nymphs of this pest, indicating that it does not act on adults (Kuwahara et al., 2011). In nymphs, geraniol acts a repellent. For example, sycamore lace bug nymphs crowd together on the leaves of platanus trees and show evasive behavior when an individual in the center of the crowd is squashed and emits geraniol (Kuwahara et al., 2011). Interestingly, geraniol is also the alarm pheromone of the chrysanthemum lace bug, Corythucha marmorata, which is a new exotic invasive pest in Asia (Watanabe and Shimizu, 2015), and one of the alarm pheromones in the hawthorn lace bug, Corythucha cydoniae, and eggplant lace bug, Gargaphia solani (Aldrich et al., 1991). Thus, geraniol may be common to insect species in Tingidae. Although geraniol may potentially be used to control sycamore lace bugs and multiple other insect species, it is unstable in the environment, and this limits its large-scale application.

Chemosensory proteins (CSPs) are a class of soluble carrier proteins that are thought to be involved in insect chemoreception (Kulmuni and Havukainen, 2013), and each Hemipteran insect has approximately 10 CSP genes (Zhou et al., 2010; Gu et al., 2013; Wu et al., 2016; Wang et al., 2017). Based on threedimensional (3D) structures, CSPs consist of 6 alpha helices stabilized by α-α loops. There are four highly conserved and structurally important cysteines that are connected by two pairs of non-interlocked disulphide bridges (Wanner et al., 2004).

CSPs are thought to be associated with chemoreceptive sense to chemical information in insect. Research on CSPs can promote the development of insect behaviorally active compounds. However, at present, we lack of understanding of the CSPs in this pest. In the current study, we combined bioinformatics analysis, expression profiling, molecular docking, fluorescence competitive binding assays and behavioral studies to identify the sycamore lace bug CSPs that recognize geraniol. We also performed a virtual screen for additional compounds that potentially bound to CSPs and then determined which of these compounds could significantly regulate the behavior of this pest. The novel insect behavior control compounds that we have identified in this study will improve our understanding of the olfactory mechanisms of sycamore lace bug and facilitate the development of strategies to control the behavior of this important pest.

## MATERIALS AND METHODS

#### Insects

Sycamore lace bugs were obtained from Platanus × acerifolia trees grown at the Chinese Academy of Agricultural Science, Haidian District, Beijing. A population of sycamore lace bugs was reared on P. × acerifolia leaves in a greenhouse under the following conditions: temperature 25 ± 2 ◦C, humidity 50– 70%, and a 16L−8D photoperiod. The nymphs and adults were collected from the leaves for expression analysis and behavioral studies.

#### Identification and Expression Pattern of CSPs

Our previously published transcriptome datasets were used in this study (Li et al., 2016, 2017). Briefly, total RNA was extracted from sycamore lace bugs at different developmental stages (nymphs, female adults, and male adults) and physiological stages (dormant and non-dormant) using the RNAqueous-Micro kit (Life Technologies, Carlsbad, CA, USA). From these RNA samples, cDNA libraries were constructed and sequenced using the Illumina HiSeq2500 sequencer at Novogene Company (Beijing, China). Sequencing data were deposited into the NCBI Sequence Read Archive (SRA) under accession numbers SRR3170921, SRR3170922, SRR3170923, SRR3883369, and SRR3883370. To identify CSPs, known hemipteran CSP amino acid sequences were used as queries in tBLASTn searches against the transcriptome database. Then, the putative CSPs were confirmed by BLASTx searches against the nr database in NCBI (E < 1.0E-5).

RNA for expression analysis of individual genes was extracted from sycamore lace bugs using the same method as for the transcriptome analysis. The quantity of the total RNA was determined using DS-11 Spectrophotometer (DeNovix, USA). The cDNA was synthesized from 1 µg of total RNA using PrimeScript RT Master Mix (Perfect Real Time) (TaKaRa, Dalian, China). Quantitative Real-Time PCR (qRT-PCR) was carried out to determine the expression patterns of sycamore lace bug CSPs in the antennae of nymphs, adult males and adult females and the transcript levels of CcilCSP2 in different nymph tissues. All gene-specific primers were designed using the Integrated DNA Technologies web site (http://sg.idtdna.com). Primer sequences are listed in **Table S1**. GAPDH (Genbank number: MG948453) and 18S rRNA(Genbank number: MG948452) were used as internal controls. The 2×GoTaq <sup>R</sup> qPCR Master Mix (Promega, Madison, WI, USA) was used for qRT-PCR performed on an ABI Prism <sup>R</sup> 7500 (Applied Biosystems, Carlsbad, CA, USA). The PCR-cycling conditions were as follows: 95◦C for 2 min, 40 cycles of 95◦C for 30 s, 60◦C for 1 min, and a final melting cycle at 95◦C for 15 s, 60◦C for 15 s, 95◦C for 15 s. For each sycamore lace bug CSP, qRT-PCR was repeated three times on three independent biological replicates. Relative expression levels of genes were calculated using the 2−11Ct method (Pfaffl and, 2001).

#### Cloning of CSP Genes

To clone the CSP genes, gene-specific primers were designed using Primer version 5.0 (Lalitha, 2004). PCR amplification was performed using the PrimeSTAR HS DNA Polymerase (Takara, Dalian, China) with the following conditions: 95◦C for 1 min; 30 cycles of 98◦C for 10 s, 58◦C for 15 s, and 72◦C for 1 min, and a final extension at 72◦C for 10 min. The PCR products were purified with the EasyPure Quick Gel Extraction Kit (TransGen, Beijing, China). Then, the fragments were ligated into the pEASY-Blunt cloning vector (TransGen, Beijing, China) and sequenced by TsingKe (Beijing, China). The protein coding regions were predicted with ORF Finder (http://www.ncbi. nlm.nih.gov/projects/gorf/). The phylogenetic relationships were determined using the neighbor-joining algorithm with 1,000 bootstrap replicates.

#### Molecular Docking

We searched for potential templates for CcilCSP2 in the NCBI Protein Data Bank (PDB) database using the BLAST server. The homology modeling of CcilCSP2 was performed using the SWISS-MODEL function in Swiss Pdb viewer. The model was further refined by molecular dynamics simulations.

The final 3D model was assessed using several methods on the online Structure Analysis and Verification Server (http:// services.mbi.ucla.edu/SAVES/), including Procheck, Verify\_3D and ERRAT. The evaluation of PDF total energy was carried out on the ProSA-web server (Wiederstein and Sippl, 2007). The best models for CcilCSP2 were confirmed using the evaluation of PDF total energy, verify score and Ramachandran plots.

A total of 101 compounds were downloaded from the ZINC database. These compounds included commercially available host volatile substances, insect pheromones or their analogs (**Table S2**). The virtual screen was performed using AutoDock Vina and AutoDock. First, the AutoDock Vina program was used to perform automated computational docking to quickly obtain docking scores for the binding of these compounds with CSPs. Then, the binding modes of compounds with docking scores <-7 were further estimated using AutoDock version 4.2. The Lamarckian genetic algorithm was used for molecular docking; 100 Lamarkian genetic algorithm runs were performed with 25 × 10<sup>6</sup> evaluations.

#### Fluorescence Competitive Binding Assays

CcilCSP2 was expressed and purified following our previously published protocols (Chang et al., 2015). Briefly, the sequence of CcilCSP2 encoding the mature CSP protein was expressed in Escherichia coli BL21 (DE3) competent cells (Transgen). Cells were grown at 37◦C until OD600 ≈ 0.60, when expression was induced with 1 mM IPTG. After incubating an additional 8 h at 28◦C, cells were harvested by centrifugation. The purification of CcilCSP2 was performed using HisTrap affinity columns (GE Healthcare Biosciences, Uppsala, Sweden). To measure the binding affinity of CcilCSP2 to the fluorescent probe N-Phenyl-1-naphthylamine (1-NPN), a 2µM protein solution in Tris-HCL buffer (pH 7.4, 50 mM) was titrated with 1 mM 1-NPN in methanol to a final concentration of 2–16µM. The binding of each ligand was tested in competitive binding assays using 1-NPN as the fluorescent reporter and final concentrations of 0.2–1.6µM for each ligand. Binding constants of competitors were calculated from the corresponding IC50 values using the following equation: K<sup>D</sup> = [IC50]/1+[1-NPN]/K1−NPN), with [1-NPN] being the free concentration of 1-NPN and K1−NPN being the binding constant of the protein complex/1-NPN.

#### Behavioral Study of Sycamore Lace Bug Adults

Compounds that potentially bind with high affinity to CSPs were purchased from Sigma (St Louis, MO, USA) to test their effects on sycamore lace bug behavior. The behavioral response of sycamore lace bug adults (1:1 sex ratio) was assessed using a Y-shaped olfactometer. The length of the arms and the diameter of the tube were 10 and 5 cm, respectively. The authentic standards were dissolved in hexane at concentrations of 1 and 0.1 µg/µl. Then 10 µl of solution was applied to a 1 cm<sup>2</sup> filter paper. The filter paper was placed in one arm of the Y-shaped olfactometer, and 10 µl hexane was placed in the other arm as a control. The amount of airflow was set at 250 ml/min. In each experiment, a single sycamore lace bug adult was released at the base of the olfactometer stem and observed for 10 min. During this time, insects not making any choice were recorded as having "no response". Insects entering more than halfway into the olfactometer arm and staying for at least 10 s were recorded as having a "response". After 10 insects were tested, the orientation of the Y-shaped olfactometer was reversed to avoid environmental effects, and the inner wall was cleaned with degreased cotton containing acetone, washed with ethanol, and then washed with distilled water and dried in an oven. The experiment was conducted in a behavioral observation chamber at 26◦C. The experiment was repeated seven times, and each time the behavior of 10 insects was tested. The choice of insects between chemicals and hexane was compared using chi-square analysis.

#### Behavioral Study of Sycamore Lace Bug Nymphs

The Y-shaped olfactometer strategy is not suitable for studying the behavior of sycamore lace bug nymphs due to their small body size. Therefore, two alternative methods were used for tests of nymph behavior.

#### Petri Dish Test

One test was conducted in a petri dish (15 cm diameter) at 22–26◦C. Two filter papers (3 cm diameter) were placed symmetrically on both sides of the petri dish. 20 µl of the compound to be tested (1, 0.1 or 0.01 µg/µl) was added to one filter paper. 20 µl hexane was added as a control to the other filter paper. The filter papers were placed at a distance of 5 cm from the start position at the center of the petri dish where 40 sycamore lace bug nymphs were released. The number of nymphs on each filter paper was recorded after 5 min. Each assay was replicated four times with a new petri dish for each replicate. The choice between hexane and a blank was also tested. The choices between those compounds that were found to affect nymph behavior and geraniol were also tested. The numbers of insect on each filter paper were compared using chi-square analysis.

#### Plant Leaf Test

To further confirm the behavioral effects of each active substance, the behavioral response of sycamore lace bug nymphs was tested

TABLE 1 | The behavioral response of sycamore lace bug nymphs to geraniol and phenyl benzoate in petri dish tests.


<sup>a</sup>Chi-square value. \*P < 0.05, \*\*P < 0.01.

by directly applying the compounds to a P. × acerifolia leaf. One microliter tested compound (1 or 0.1 µg/µl) was applied at a distance of 0.5 mm from a squashed colony. Photographs were then taken at 0, 2, and 4 min to assess the repellency rate, which was calculated using the following formula: Repellency rate = M/T<sup>∗</sup> 100, where T = the total number of sycamore lace bug nymphs in the tested colony at 0 min, and M = the number of sycamore lace bug nymphs that had moved at 2 or 4 min. Experimental conditions were the same as for the petri dish experiments. Hexane was used as the mock treatment. The repellency rates of the nymphs in each treatment were compared using a Kruskal Wallis H-test followed by Mann-Whitney U-tests.

#### RESULTS

#### The Behavioral Response of Sycamore Lace Bug To Geraniol

To determine if adults and nymphs respond differently to geraniol, we performed behavioral assays. Behavior was assayed in a Y-shaped olfactometer for adults and in petri dishes and on leaves for nymphs. Sycamore lace bug adults and nymphs had different behavioral responses to geraniol. The nymphs were significantly repelled by geraniol in petri dishes (**Table 1**) and on plant leaves (**Table 5**). However, the behavior of sycamore lace bug adults was not significantly affected by geraniol (Y-tube, 1 µg/µl geraniol: χ <sup>2</sup> = 0.069, P > 0.05; 0.1 µg/µl geraniol: χ <sup>2</sup> = 1.52, P > 0.05).

## The CSPs Involved in Binding Geraniol

We hypothesized that the CSPs involved in the perception of geraniol would be more highly expressed in the antennae of nymphs than in the antennae of adult males and females. To test this hypothesis, a total of 15 CSPs were identified

18S rRNA genes. Transcript levels are shown relative to those in the leg. Data are presented as the mean (± SD), and different letters indicate significant differences in transcript levels (p < 0.05, LSD test).

FIGURE 1 | Relative expression levels of candidate CSP genes in the antennae of sycamore lace bugs. AF, adult female; AM, adult male. Transcription levels of the CcilCSP2 gene were normalized by GAPDH and the 18S rRNA gene. Data are presented as the mean (± SD), and different letters indicate significant differences in transcript levels (p < 0.05, LSD test).

Li et al. Corythucha Ciliata Pheromone-Binding

from the sycamore lace bug RNA-seq datasets and their expression patterns were compared. These 15 CSPs include one previously reported CcilCSP1 (Fu et al., 2017) and 14 new CSP genes. Only one CSP (c32563\_g2) was expressed at a higher level in the antennae of nymphs than in the antennae of adult males and females (**Figure 1**). Moreover, this CSP was significantly more highly expressed in antennae than other nymph tissues (**Figure 2**). Thus, the expression levels of c32563\_g2 are associated with the differing responses of sycamore lace bug nymphs and adults to geraniol, and this CSP was selected as a candidate geraniol-binding protein. The sequence of c32563\_g2 was confirmed by molecular cloning and sequencing, and phylogenetic analysis with known CSPs in other insect species revealed that c32563\_g2 clustered with the Adelphocoris suturalis Jakovlev CSP2 protein (82% identity at the amino acid level). Based on the nr annotation (**Table 2**) and phylogenetic relationships (**Figure 3**), we named c32563\_g2 as CcilCSP2.

#### Docking Analysis of CcilCSP2

Using the CcilCSP2 amino acid sequence as a query in a blastp search against the PDB database, three CSP templates, including a CSP from the desert locust Schistocerca gregaria (PDB: 2GVS, 29% identity), antennal CSP A6 from Mamestra brassicae (PDB: 1KX8, 26% identity), and CSP 1 from Bombyx mori (PDB: 2JNT, 26% identity) were selected for homology modeling. After the evaluation of model qualities, the G-factor values (**Table 3**) were all greater than−0.5, which indicated that the distribution of torsion angles and covalent geometries of the model proteins were reasonable. Greater than 82.42% of the residues had an average 3D-1D score > 0.2 in VERIFY 3D, the overall quality factor was > 60.241 in ERRAT, and the Z-score for CcilCSP2 was−4.96. These parameters indicated that the protein model obtained by homology modeling was reliable.

#### TABLE 2 | Candidate sycamore lace bug chemosensory protein unigenes.

In addition to geraniol, four compounds, including naphthalene, thymol, p-Cymene, and phenyl benzoate were identified as potential ligands of CcilCSP2 using AutoDock Vina and then verified by AutoDock4.2. The binding affinities of these five compounds for CcilCSP2 are shown in **Table 4**. Naphthalene, thymol, p-Cymene, phenyl benzoate and geraniol) exhibited low binding energies (< −5.33) and had inhibition constants (Ki) <123µM. Notably, phenyl benzoate had the lowest binding energies (−6.42) and K<sup>i</sup> (19.75µM) (**Table 4**).

Docking analysis showed that CcilCSP2 binds phenyl benzoate and geraniol in the same region, which includes amino acid residues Leu-90, Val-89, Gln-87, Ile-86, Ieu-63, Ala-67, Ala-109, Leu-105, and Trp-101 (**Figure 4**). Furthermore, CcilCSP2 amino acid residues Ile-86 forms a hydrogen bond with both phenyl benzoate and geraniol (**Figure 4**).

#### CcilCSP2 Ligand-Binding Properties

The dissociation constant (KD) for CcilCSP2 bound to 1-NPN was 2.081µM (**Figure 5A**), and 1-NPN was used as a fluorescent reporter to test the binding affinities of CcilCSP2 to different ligands. Based on the IC50 and K<sup>D</sup> values calculated from the ligand binding curves (**Figure 5B**), CcilCSP2 has high binding affinity for geraniol and phenyl benzoate (Displacement of more than 50% of 1- NPN, K<sup>D</sup> = 15.16 and 13.93µM, respectively). Other ligands were not able to displace more than 50% of 1- NPN from CcilCSP2. We also tested another CSP (c18915\_g1) that is expressed more highly in the antennae of nymphs than in the antennae of adult females. However, this CSP did not bind to any of these five ligands (**Figure S1**).

#### The Behavioral Response of Sycamore Lace Bug to the Four Candidate Compounds

The behavioral responses of sycamore lace bug nymphs to phenyl benzoate, naphthalene, p-Cymene and thymol were tested using


the petri dish test. There was no significant difference in the choice of nymphs between hexane and the blank (**Table 1**), indicating that hexane does not affecting nymph behavior. Of the four chemicals tested, only phenyl benzoate significantly repelled sycamore lace bug nymphs (**Table 1**). There was no significant difference in the choice of nymphs between phenyl benzoate and geraniol (**Table 1**). The repellent effect of phenyl benzoate was further confirmed by plant leaf tests (**Table 5**), and there was no significant difference between the nymph repellency by phenyl benzoate and geraniol at doses of 1 and 0.1 µg/µl (P > 0.05). The behavioral response of adult sycamore lace bugs to phenyl benzoate was also not significant (Y-tube test, 1 µg/µl phenyl benzoate: χ <sup>2</sup> = 0.176, P > 0.05; 0.1 µg/µl phenyl benzoate: χ <sup>2</sup> = 0.083, P > 0.05).

## DISCUSSION

Identification and expression profiling of chemosensory genes are vital for exploring their roles in insect behavior (Calvello et al., 2005; Du et al., 2016; Zhu et al., 2016a,b). To date, there have been

#### TABLE 3 | CcilCSP2 model quality estimations.


TABLE 4 | Binding affinities of six compounds for CcilCSP2 estimated by AutoDock 4.2.


no reports on the olfactory mechanisms of sycamore lace bugs. In our study, we found that geraniol only affected the behavioral responses of sycamore lace bug nymphs. This is consistent with the previous detection of geraniol exclusively in nymphs (Kuwahara et al., 2011). This characteristic also provided us a clue for identifying candidate key olfactory genes in this study. We previously identified 15 CSPs by sequencing and analyzing the sycamore lace bug transcriptome data (Li et al., 2016), which is similar to the number of CSPs identified in other Hemipteran species (Zhou et al., 2010, 2014). Of these CSPs, we identified CcilCSP2 as a putative geraniol-binding protein based on a combination of behavioral studies, expression pattern analysis, and fluorescence competitive binding assays. The characteristics of CcilCSP2 reported here will provide benefit for functionalizing the putative homologous gene CSP2 in A. suturalis (**Table 2** and **Figure 3**; Cui et al., 2016). Moreover, such expression characteristics allow us to use this model to further investigate the odorant binding proteins and olfactory receptors in C. ciliata to identify the alarm pheromone and new repellents.

In addition to geraniol, phenyl benzoate was identified as a ligand of CcilCSP2 and was shown to repel sycamore lace bug nymphs and to disrupt aggregations of nymphs on leaves. Phenyl benzoate is widely used as a starting chemical in the production of polyesters (Rosenfeld, 1987) and has many properties that

TABLE 5 | The repellency rate of sycamore lace bug nymphs to geraniol and phenyl benzoate in plant leaf tests.


\*P < 0.05.

make it more stable in open-field environments than geraniol, such as resistance to heat and UV irradiation (Gooch, 2011). Phenyl benzoate could thus serve as a behavioral regulation compound in push-pull systems in the future. Although the structure of geraniol and phenyl benzoate are very different, these

two chemicals both interact with Ile-86 through hydrogen bonds. The detailed 3D structures should be studied in future.

Interestingly, like geraniol, phenyl benzoate only repels sycamore lace bug nymphs, indicating that this compound may share the same or similar mechanisms as geraniol. Therefore, phenyl benzoate is also a candidate for the behavioral regulation of other Tingidae species whichgeraniol is the alarm pheromone, including chrysanthemum lace bug (C. marmorata), hawthorn lace bug (C. cydoniae) and eggplant lace bug (B. solani) (Aldrich et al., 1991; Watanabe and Shimizu, 2015). The behavioral response of these species to phenyl benzoate needs to be tested in the future.

Molecular docking is an important method to discover new chemicals based on the structure of proteins with known functions. This method has played an important role in drug discovery (Ruan et al., 2013), and some attractants and repellent chemicals for some insect species, including mosquitoes and aphids, have been also identified by this method (Dhivya and Manimegalai, 2014; Qin et al., 2016; Wang et al., 2016). In our previous paper (Fu et al., 2017), we investigated the interaction between CcilCSP1 and host-plant volatiles using molecular docking, and our findings suggested that this method is useful for searching the pheromone substance of sycamore lace bug. In the current study, our identification of a novel repellent compound demonstrates the potential power of molecular docking and subsequent in vitro/in vivo evaluation for developing chemicals that regulate insect behavior, and discovering olfactory protein-interacting molecules. Furthermore, this study provides guidelines for screening new repellents for sycamore lace bug nymphs. Specifically, potential repellents could be tested by molecular docking and binding experiments with CcilCSP2, and then only chemicals binding CcilCSP2 could be further tested for effects on behavior. Similar methods could also be used to identify repellent compounds in other Tingidae insect species.

The current study increases our understanding of the functions of CcilCSP2 in sycamore lace bug and the compounds that attract or repel this important pest. In addition, CcilCSP2 could be targeted by CRISPR/Cas9 or RNAi as a way to control this pest. Furthermore, detailed study of the three-dimensional structure of CcilCSP2 and its interaction with geraniol and phenyl benzoate could help in the design of new compounds that mimic alarm pheromones and affect sycamore lace bug nymph behaviors in the field.

#### AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: FL, YX, and CL; Performed the experiments: FL, NF, DL, and HC; Analyzed the data: FL, NF, DL and HC; Contributed reagents, materials, analysis tools: CQ and RW; Wrote the paper: FL and CL.

#### ACKNOWLEDGMENTS

This work was funded by the National Natural Science Foundation of China (31701789) and the Project of Regional Collaborative Innovation of Beijing Academy of Agriculture and Forestry Sciences (grant No. KJCX20170709). We thank Sharon Rose Hill (Swedish University of Agricultural Sciences) for useful suggestions.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00354/full#supplementary-material

Figure S1 | Binding of N-phenyl-1-naphthylamine (1-NPN) and selected ligands to c18915\_g1. (A) Affinity of c18915\_g1 for 1-NPN. (B) Competitive binding assays.

Table S1 | Primers used in this study.

Table S2 | Chemicals used in molecular docking.

#### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Li, Fu, Li, Chang, Qu, Wang, Xu and Luo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identification, Expression Patterns, and Functional Characterization of Chemosensory Proteins in *Dendroctonus armandi* (Coleoptera: Curculionidae: Scolytinae)

Zhumei Li <sup>1</sup> , Lulu Dai <sup>1</sup> , Honglong Chu1,3, Danyang Fu<sup>1</sup> , Yaya Sun<sup>1</sup> and Hui Chen1,2 \*

*<sup>1</sup> College of Forestry, Northwest A&F University, Yangling, China, <sup>2</sup> College of Forestry and Landscape Architecture, South China Agricultural University, Guangzhou, China, <sup>3</sup> Center for Yunnan Plateau Biological Resources Protection and Utilization, College of Biological Resource and Food Engineering, Qujing Normal University, Qujing, China*

#### *Edited by:*

*Shuang-Lin Dong, Nanjing Agricultural University, China*

#### *Reviewed by:*

*Tiantao Zhang, Intitute of Plant Protection (CAAS), China William Benjamin Walker III, Swedish University of Agricultural Sciences, Sweden Liping Ban, China Agricultural University, China*

*\*Correspondence: Hui Chen chenhui@ nwsuaf.edu.cn*

#### *Specialty section:*

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

*Received: 23 December 2017 Accepted: 12 March 2018 Published: 27 March 2018*

#### *Citation:*

*Li Z, Dai L, Chu H, Fu D, Sun Y and Chen H (2018) Identification, Expression Patterns, and Functional Characterization of Chemosensory Proteins in Dendroctonus armandi (Coleoptera: Curculionidae: Scolytinae). Front. Physiol. 9:291. doi: 10.3389/fphys.2018.00291* The Chinese white pine beetle, *Dendroctonus armandi* Tsai and Li (Coleoptera: Curculionidae: Scolytinae), is a serious pest of coniferous forests in China. Thus, there is considerable interest in developing eco-friendly pest-control methods, with the use of semiochemicals as a distinct possibility. Olfaction is extremely important for fitness of *D. armandi* because it is the primary mechanism through which the insect locates hosts and mates. Thus, here we characterized nine full-length genes encoding chemosensory proteins (CSPs) from *D. armandi*. The genes were ubiquitously and multiply expressed across different developmental stages and adult tissues, indicating various roles in developmental metamorphosis, olfaction, and gustation. Ligand-binding assays implied that DarmCSP2 may be the carrier of *D. armandi* pheromones and various plant host volatiles. These volatiles were identified through RNA interference of *DarmCSP2* as: (+)-α-pinene, (+)-β-pinene, (−)-β-pinene, (+)-camphene, (+)-3-carene, and myrcene. The systematic chemosensory functional analysis of DarmCSP2 in this study clarified the molecular mechanisms underlying *D. armandi* olfaction and provided a theoretical foundation for eco-friendly pest control.

Keywords: chemosensory proteins, *Dendroctonus armandi*, olfaction, semiochemicals, fluorescence binding assays, RNAi, EAG

## INTRODUCTION

Chemoreception (olfaction and gustation) is an indispensable biological process for many insect species (Sánchez-Gracia et al., 2009), playing a vital role in detecting the specific semiochemicals emitted by host plants or conspecifics (Yoshizawa et al., 2011). To accurately perceive such semiochemicals, insects have evolved a sophisticated, sensitive, and specific chemosensory system (Karg and Suckling, 1999; Field et al., 2000). Numerous olfactory protein groups have been identified in the insect chemosensory system, with wide-ranging functions that include locating food sources, recognizing conspecifics and predators, as well as identifying oviposition sites; these include odorant-binding proteins (OBPs), chemosensory proteins (CSPs), olfactory receptors (ORs), gustatory receptor (GRs), and odorant degrading enzymes (ODEs) (Sánchez-Gracia et al., 2009; Leal, 2013). While CSPs and OBPs have similar function, they share no sequence similarity (Pelosi et al., 2005; Gong et al., 2007). The special tertiary structure of CSPs with hydrophilic surface and hydrophobic cavity allow them to distinguish, capture, and bind hydrophobic chemicals from external environments to ORs or GRs (Pelosi et al., 2005, 2017; Gong et al., 2007; Sánchez-Gracia et al., 2009; Liu et al., 2012; Leal, 2013).

Unsurprisingly, given their critical functions, chemosensory proteins are widespread and have been isolated from multiple insect orders (McKenna et al., 1994; Angeli et al., 1999; Robertson et al., 1999; Marchese et al., 2000; Forêt et al., 2007; Andersson et al., 2013; Li et al., 2013; Yang et al., 2014; He et al., 2017). In insects of both sexes, CSPs are broadly expressed throughout development (Stathopoulos et al., 2002; Wanner et al., 2005; Qiao et al., 2013; Yang et al., 2014; Li et al., 2016) and across tissue types, including antennae, heads, thoraxes, abdomens, proboscis, eyes, legs, wings, pheromone glands, and reproductive organs (Nomura et al., 1992; Field et al., 2000; Nagnan-Le Meillour et al., 2000; Ban et al., 2003; Gu et al., 2013; Li et al., 2013; Zhou et al., 2013; Zhu et al., 2016; Wang et al., 2017). Fluorescence competitive binding assays have indicated that CSPs bind to a wide range of compounds, such as plant volatiles, insect pheromones (Briand et al., 2002; Li et al., 2015), cuticular hydrocarbons and lipids (Ozaki et al., 2005; González et al., 2009), as well as visual pigments (Zhu et al., 2016). These sophisticated expression profiles and binding ability suggest that the role of CSPs is complex, spanning from chemoreception to other functions in development, vision, nutrition, reproduction, and regeneration (Nomura et al., 1992; Briand et al., 2002; Wanner et al., 2005; Li et al., 2015; Zhu et al., 2016; Pelosi et al., 2017).

Clarifying the mechanisms underlying CSP function not only improves our understanding of insect biology but also has strong practical value for developing eco-friendly pest control. Because many insect pests are so dependent on olfaction to find hosts and mates, damage to olfactory systems or targeted release of host volatiles or pheromones to alter insect behavior should be effective control methods that do not negatively impact the surrounding ecosystem. For example, the Chinese white pine beetle, Dendroctonus armandi Tsai and Li (Coleoptera: Curculionidae: Scolytinae), uses aggregation pheromones to coordinate mass attacks on host trees, whereas odorants from host and non-host trees modulate pheromone response (Zhang and Schlyter, 2004; Erbilgin et al., 2007; Andersson et al., 2010, 2013). The beetle responds to volatiles emitted from both host and non-host plants, as well as insect pheromones (Zhang et al., 2010; Xie and Lv, 2012; Chen et al., 2015; Zhao et al., 2017a,b). This serious pest of coniferous forests in China's Qinling and Bashan Mountains primarily attacks healthy Chinese white pine (Pinus armandi Fr.), residing in the phloem across all life stages except for a brief dispersal period to mate and find new hosts (Ren and Dang, 1959; Cai, 1980; Chen and Tang, 2007). In particular, D. armandi infestation has damaged large swathes of P. armandi forests, incurring heavy economic losses and serious ecological destruction (Chen and Tang, 2007; Xie and Lv, 2012). There is an urgent need to develop effective and eco-friendly D. armandi control, with olfaction-related methods being an attractive option. However, we currently know very little about the molecular mechanisms underlying olfactory perception in this species.

Therefore, in this study, we combined molecular and physiological methods to investigate the relationship between CSP and olfactory behavior in D. armandi. We identified CSP genes from D. armandi (DarmCSPs), and assessed their tissue and developmental expression profiles. Selected DarmCSPs were expressed and their binding affinity to semiochemicals were tested. Finally, we examined how DarmCSP affected D. armandi olfaction and ascertained the specific semiochemicals that bind these proteins in adult beetles.

#### MATERIALS AND METHODS

#### Insect Collection

Larvae and pupae of D. armandi were collected from the bark of infested P. armandi trees at the Huoditang Experimental Forest Station of Northwest A&F University, located on the southern slope of the Qinling Mountains (33◦ 18′N, 108◦ 21′E) in Shaanxi, China. Logging slash of infested P. armandi was moved from the sample plot to a greenhouse, where adult beetles were collected as they emerged and then kept at 4◦C on moist paper. Adults were sexed based on external genitalia and male-specific auditory cues (Dai et al., 2014; Zhao et al., 2017a).

#### Reagents

Contech Enterprises (Delta, BC, Canada) provided (±)-exobrevicomin and (±)-frontalin. Bedoukian Research (Danbury, CT, USA) provided (–)-trans-verbenol. Finally, (1S)-(–) verbenone, HPLC-grade hexane, 1-hexanol, and methanol, as well as 10 host volatiles of D. armandi were purchased from Sigma-Aldrich.

#### Identification of *D. armandi CSP* Genes RNA Isolation and cDNA Synthesis

Total RNA for RT-PCR was isolated from larvae, pupae, and adults of both sexes using the UNIQ-10 Column TRIzol Total RNA Isolation Kit (Sangong, Shanghai, China), following manufacturer protocol. RNA integrity was verified with 1.0% agarose gels electrophoresis and quantified with spectrophotometry in a NanoDrop 2000 (Thermo, Pennsylvania, USA). Total RNA from the three developmental stages were mixed for cDNA synthesis with the PrimeScriptTM RT Reagent Kit with gDNA Eraser (TaKaRa, Dalian, China), following manufacturer protocol. Single-stranded 5′ and 3′ RACE-ready cDNA was synthesized from mixed RNA (1 µg) using a SMARTer RACE cDNA Amplification Kit (Clontech, CA, USA), following manufacturer protocol, then stored at −20◦C until use.

#### Gene Amplification and Cloning

Synthesized cDNA was used as a template in PCR reactions. Degenerate and specific primers (**Table S1**) were designed in Primer Premier 5.0, based on CSP sequences of other insects from NCBI (http://www.ncbi.nlm.nih.gov/). PCR amplifications were performed in a C1000 thermocycler (Bio-Rad, CA, USA), under the following conditions: initial denaturation for 3 min at 95◦C; followed by 30 cycles of 30 s at 95◦C, 30 s at 50–60◦C, 1 min at 72◦C; and then a final extension for 10 min at 72◦C. The 20 µL reaction mixture contained 1 µL cDNA (1:10 dilution), 0.25µM of each primer, and 2 × EcoTaq PCR SuperMix (TransGen, Beijing, China). PCR products were visualized on 1% agarose gels using 1× 4S Red Plus Nucleic Acid Stain (Sangong, Shanghai, China) and compared with a 2 K plus DNA marker (TransGene, Beijing, China). Amplified fragments were purified using the Gel Purification Kit (Omega, GA, USA), ligated into pMDTM 18-T Vector (TaKaRa, Dalian, China), and transformed into DH5α chemically competent Escherichia coli cells (TransGen, Beijing, China). Transformants were selected on Amp/LB/Xgal/IPTG plates, and positive clones were PCR-analyzed using vectorspecific primers (M13-47, M13-48). Lastly, bacterial solutions of positive clones were sequenced by a local biotechnology company (Augct, Beijing, China). Three independent clones were submitted to minimize potential PCR mutations. Sequences were manually edited with EditSeq of DNASTAR (https://www. dnastar.com/) to obtain inserts, which were then BLASTed against the NCBI database.

#### 5 ′ and 3′ RACE

Gene-specific inner and outer primers for 5′ and 3′ RACE (**Table S1**) were designed based on obtained sequence fragments. Touchdown PCR (annealing temperatures: 65–55◦C) was performed to improve amplification specificity of the 5′ -UTR and 3′ -UTR sequences. The amplified products were visualized, purified, cloned, sequenced, and blasted as described in the previous section ("Gene amplification and cloning").

#### Analysis of Full-Length cDNA Sequences

Full-length cDNA sequences were assembled in DNAMAN 6.0 (http://www.lynnon.com/), using sequence fragments and RACE results. To avoid chimera sequences, specific primers (**Table S1**) from initiation to terminator codon were designed based on complete sequences. High-fidelity PCR was performed using Phanta HS Super-Fidelity DNA Polymerase (Vazyme, Nanjing, China). Amplicons were cloned into pMD18-T and detected through sequencing and BLASTp search. Putative gene sequences were deposited in GenBank, and Accession Numbers were listed in **Table 1**. Open reading frames (ORFs) of full-length cDNA were obtained via ORF Finder (https://www.ncbi.nlm.nih. gov/orffinder/), and cDNA was then translated to amino acid sequences using the ExPASy Translate Tool (http://www.expasy. org/tools/dna.html), aligned in ClustalX 2.0.10 (Thompson et al., 1997), and colored in DNAMAN6.0. Molecular mass (kDa) and isoelectric points were determined in PROTPARAM (Gasteiger et al., 2005). DarmCSP homologs were identified with the NCBI-BlastP network server (https://blast.ncbi.nlm. nih.gov/Blast.cgi). Amino acid identity was analyzed through the construction of a homology tree in DNAMAN6.0. A neighbor-joining phylogenetic tree was built in MEGA 6.0 (Tamura et al., 2011), employing ClustalW with default parameters, p-distance model, pairwise gap deletion, and 1000 bootstrap replicates. The putative N-terminal signal peptide was predicted in Signal P 4.1 Server (http://www.cbs.dtu.dk/services/ SignalP/).

## Expression Patterns of *CSP* Genes Across Different Life Stages and Tissues

D. armandi larvae were separated into two sub-stages: larvae and mature larvae (when they stop feeding). Pupae were similarly separated into two sub-stages: early pupae (newly metamorphosed from larvae) and late pupae (close to becoming adults). Adults were separated into three sub-stages: teneral (body color still light), emerged, and feeding (invading a new host) (Dai et al., 2014). Antennae, mouthparts, heads (without antennae and mouthparts), forewings, underwings, legs, thoraxes, abdomens (without pheromone glands), and pheromone glands of male and female emerged adults were dissected. Samples were collected in triplicate, frozen in liquid nitrogen immediately, and stored at −80◦C until use. RNA isolation and cDNA synthesis followed previous descriptions ("RNA isolation and DNA synthesis").

The CFX96TM Real-Time PCR Detection System (Bio-Rad, California, USA) was used for qRT-PCR, with D. armandi β-actin (accession number: KJ507199.1) and α-tubulin (accession number KJ507202.1) as reference genes. Specific qRT-PCR primers were designed in Beacon Designer 7.7, based on nucleotide sequences (**Table S1**), and their amplification efficiencies were calculated using relative standard curves with a five-fold cDNA dilution series; the efficiency values for the primers were 100 ± 5%. The sizes of the amplicons were 231 bp (β-actin), 218 bp (α-Tubulin), 193 bp (DarmCSP1), 95 bp (DarmCSP2), 208 bp (DarmCSP3), 229 bp (DarmCSP4), 229 bp (DarmCSP5), 120 bp (DarmCSP6), 183 bp (DarmCSP7), 132 bp (DarmCSP8), and 250 bp (DarmCSP9). Amplicons were confirmed to be of the correct size after the qRT-PCR assay via gel electrophoresis, and then sequenced by a biotechnology company (Augct, Beijing, China) to make sure that the correct amplification products were obtained. The reaction mixture (20 µL) contained 10 µL of SYBR <sup>R</sup> Premix Ex TaqTM II (Tli RNaseH Plus) (TaKaRa, Dalian, China), 2 µL of cDNA (diluted 10 times), 0.6 µL of each primer, and 6.8 µL of nuclease-free water. Template-free negative controls were included in every reaction. Thermocycling conditions were as follows: 95◦C for 10 s, followed by 40 cycles of 95◦C for 5 s, and 60◦C for 30 s. At the end of each reaction, a melting curve analysis was performed to detect single gene-specific peaks and check for primer dimers. Three technical and three biological replicates were performed to verify reproducibility. DarmCSPs expression data were generated from normalizing data to the geometric average of the internal control genes (Vandesompele et al., 2002). The comparative 2 <sup>−</sup>11Ct method was used to calculate relative mRNA levels of DarmCSPs (Schmittgen and Livak, 2008); resultant values were log2-transformed for analysis of variance and plotting. Expression was normalized based on the lowest expression level.

#### Binding Characteristics of DarmCSPs E. coli Expression and Purification of DarmCSPs

To better characterize DarmCSP function, three antennaepreferential genes (DarmCSP 1–3) were chosen for expression in bacteria. Signal peptides were removed to generate properly folded proteins. PCR products encoding mature proteins


*ORF, open reading frame; pI, isoelectric point; MW, molecular weight; <sup>a</sup>As predicted by ORF Finder (https://www.ncbi.nlm.nih.gov/orffinder/). <sup>b</sup>As predicted using SignalP 4.1 Server (http://www.cbs.dtu.dk/services/SignalP/). <sup>c</sup>As predicted by Protparam program. <sup>d</sup>As predicted by BLAST (http://www.ncbi.nlm.nih.gov).*

were amplified using gene-specific primers (**Table S1**), cloned into pGEM-T easy vectors (Promega, Madison, USA), then excised and cloned into the bacterial expression vector pET32a(+) (Novagen, Madison, WI), between BamHI and XhoI restriction sites. Successful cloning was verified through PCR and sequencing. Plasmids containing the correct insert were extracted and transformed into E. coli BL21 (DE3) competent cells. Positive clones were incubated at 37◦C until absorbance = 0.6 at OD 600, and protein expression was induced with IPTG (isopropyl-β-D-1-thiogalactopyranoside) treatment (28◦C for 6 h) to a final concentration of 0.5 mM. Cells were harvested via centrifugation at 12,000 × g and 4◦C for 5 min, then cleaned using PBS buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, pH = 7.4). After resuspension in the lysis buffer (50 mM Tris-HCl, 300 mM NaCl, 25 mM Na2HPO4, pH = 8.0), cell solution was sonicated on ice for 10 min (sonication for 3 s with an interval of 5 s), then centrifuged again at 12,000 × g and 4◦C for 30 min.

Recombinant proteins were purified with N-termini His tagged from the supernatant using a Ni-NTA-Sefinose Column (Sangon, Shanghai, China), and placed in a buffer (25 mM Tris-HCl, 50 mM NaCl, and 2 mM CaCl2, pH 7.6) for dialysis. To avoid confounding effects in fluorescence binding assays, Histags were excised using Recombinant Enterokinase with His-tag (rEK) (Sangong, Shanghai, China), and the resultant complex was cleared through a Ni-NTA-Sefinose Column. NaCl and CaCl<sup>2</sup> were removed from DarmCSPs via dialysis in 50 mM Tris-HCl buffer (pH = 7.4). Purified proteins were stored at −80◦C until use. The size and purity of DarmCSPs were checked using 12% SDS-PAGE, whereas their concentration was measured with the BCA Assay Kit (Sangong, Shanghai, China).

#### Fluorescence Binding Assays

A Hitachi F-4500 fluorescence spectrophotometer was used to measure emission fluorescence spectra, in a right-angle configuration with a 1 cm light-path quartz cuvette. The slit width was 5 mm for both excitation and emission. DarmCSPs were dissolved to 2µM in 50 mM Tris-HCl buffer (pH = 7.4), whereas fluorescent probe N-phenyl-1-naphthylamine (1- NPN) and all semiochemicals were dissolved in methanol with a 1 mM stock solution. To measure DarmCSP affinity with 1- NPN, 2 mL of 2µM DarmCSP solution was titrated with 1 mM 1-NPN to a final concentration of 2–16µM. Excitation of 1-NPN occurred at 337 nm, with the emission spectra recorded from 360 to 500 nm. Corresponding fluorescence intensity values were plotted against free 1-NPN concentration to determine DarmCSP binding constants. Bound 1-NPN concentrations were assessed based on fluorescence intensity, assuming DarmCSPs were 100% active and protein: ligand = 1:1 at saturation. The dissociationconstants curve was linearized with Scatchard plots in Prism 6.0 (GraphPad Software, CA, USA).

Ligand affinity was measured with competitive binding assays. Fourteen compounds were selected based on previous reports (Zhang et al., 2010; Xie and Lv, 2012; Chen et al., 2015; Zhao et al., 2017a,b), including 10 host volatiles and four D. armandi pheromones (**Table 2**). A mixture of 2µM DarmCSP and 2µM 1-NPN was titrated with each ligand to final concentrations of 2–16µM. Corresponding fluorescence intensities were recorded from three independent measurements. Dissociation constants of competitive ligands were calculated according to IC50 values, using the equation: KD = [IC50]/(1+[1-NPN]/K1−NPN), where IC50 is competitive-ligand concentration at half the initial fluorescence of 1-NPN, 1-NPN is the concentration of free 1- NPN, and K1−NPN is the dissociation constant of DarmCSP with 1-NPN.

#### Structural Model of DarmCSP2

The predicted 3D structure of DarmCSP2 was generated via homology modeling in SWISS-MODEL (https://swissmodel. expasy.org/) with default parameters (Guex et al., 2009), with the solution structure of Schistocerca gregaria CSP4 (Tomaselli et al., 2006) as a template (identity: 44.33%). The model was rendered in PyMol (http://www.pymol.org/). A multiple protein sequence alignment was created with ClustalX 2.0.10 (Thompson et al., 1997) and colored using ESPript (http://espript.ibcp.fr/ESPript/ cgi-bin/ESPript.cgi) (Robert and Gouet, 2014).



#### RNA Interference of *DarmCSP2* Insect Treatment and qRT-PCR

As further verification of DarmCSP2 biological function, D. armandi adults were injected with gene-specific doublestranded RNA (dsRNA) for RNA interference (RNAi). Two pairs of special primers (T7DarmCSP2F/DarmCSP2R and DarmCSP2F/T7DarmCSP2R) were designed for dsRNA synthesis through the addition of T7 polymerase recognition region (5′ -taatacgactcactatagg-3′ ) at the 5′ ends (**Table S1**). The verified pMD18-T plasmid containing DarmCSP2 acted as a template for two high-fidelity PCRs using Phanta HS Super-Fidelity DNA Polymerase (Vazyme, Nanjing, China). Resultant cDNA, flanked by T7 polymerase promoter sequences, were electrophoresed on a 1% agarose gel and purified with the Gel Purification Kit (Omega, GA, USA). Purified amplicons were used as templates to synthesize dsRNA with T7 RiboMAX Express RNAi (Promega, USA), following manufacturer protocol. DarmCSP2-dsRNA (hereafter dsCSP2) integrity was checked via 1% agarose gel electrophoresis. Finally, dsCSP2 was then quantified in NanoDrop 2000 (Thermo, Pennsylvania, USA), diluted to 1000 ng/µL in DEPC water, and stored at −80◦C.

Freshly and synchronously emerged adults were anesthetized on glass petri dishes placed for 30 min in an ice bath, before injection with 0.2 µL dsCSP2 into the hemocoel at the suture under the hindleg. Injections used Hamilton MicroliterTM syringes (700 series, RN) with 32G sharp-point needles (Hamilton, Switzerland). Controls were either injected with 0.2 µL DEPC-treated water or not injected. Subjects were then transferred onto wet filter paper placed in clear glass petri dishes for continuous culture (at 20 ± 1 ◦C, 50% humidity). Three males and three females were removed at different time intervals (12, 24, and 48 h) from each treatment group for storage at −80◦C until qRT-PCR analysis. Three replicates were performed per treatment group (non-injected, water-injected, dsCSP2-injected). RNA isolation, cDNA synthesis, and qRT-PCR procedure followed methods described above ("RNA isolation and cDNA synthesis" and "Expression patterns of CSP genes across different life stages and tissues").

#### Electroantennogram Analysis

Electroantennograms (EAG) were used to detect RNAi efficiency of dsCSP2 and determine DarmCSP2 function in binding semiochemicals. Methods were modified from a previous study (Zhang et al., 2010). Subject beetles were chosen based on post-RNAi qRT-PCR and anesthetized in an ice bath. Antennae were carefully excised at their base with a scalpel and immediately connected between two electrode holders using Spectra 360 electrode gel.

Semiochemicals were selected based on the results of fluorescence binding assays and dissolved to 10 µg/µL in hexane. Hexane alone acted as a blank control and 1 µg/µL 1-hexanol was used as a standard to normalize all EAG recordings (Zhang et al., 2010). Semiochemical solutions (20 µL) were loaded onto filter paper strips (5 × 30 mm) and then transferred into a Pasteur pipette. The pipette tip was then inserted into a small hole in the wall of a steel tube (15 mm diameter × 15 cm length). The tube was connected to an air stimulus controller (CS-05b Syntech, the Netherlands) for constant humidified airflow delivery at a rate of 40 mL/min. The open end of the tube was positioned 1 cm before an antenna affixed between two electrode holders.

To stimulate the antenna, semiochemical-containing air was introduced through the pipette into the main air flow at a rate of 20 mL/min for 0.2 s. Each stimulus was separated by at least a 1 min interval to ensure complete antenna recovery. Signals were recorded using an IDAC-2 unit plus amplifier and Syntech EAG 2000 (Syntech, Netherlands). The control and standard were tested before and after every semiochemical solution. Antennae from12 individuals (six males and six females) were tested, with three replicates per antenna. To calculate EAG values, mean responses of the solvent control before and after exposure were subtracted from mean sample responses, then converted to a percentage of the accompanying standard (Zhang et al., 2010).

#### Statistical Analysis

Data from qRT-PCR and EAG were analyzed in SPSS Statistics 19.0 (IBM, Chicago, USA). Significant between-treatment differences in DarmCSP mRNA levels and EAG groups were derived through ANOVA (P < 0.05), then adjusted with a Duncan multiple-comparison test. All two-sample analyses were performed using Student's t-tests. Graphs were plotted in Prism 6.0 (GraphPad Software, CA, USA).

## RESULTS

#### Sequence Characteristics and Homology Analysis of *DarmCSP*s

Nine full-length putative CSP genes were cloned from D. armandi. Generally, DarmCSP ORFs contained ∼400 nucleotides, encoding ∼130 amino acids; the exception was DarmCSP5 with 786 nucleotides encoding 255 amino acids. Predicted molecular weights of DarmCSPs were 13.02–16.41 kDa, apart from DarmCSP5 at 28.04 kDa. Isoelectric points of DarmCSPs ranged from 4.92 to 9.48, with DarmCSP3 and DarmCSP7 being <7.00 and the remainder >7.00 (**Table 1**). Nine DarmCSPs contained a putative signal peptide at the N-terminus (**Table 1**, **Figure 1**).

Full-length BLASTp searches indicated high amino acid sequence identity between DarmCSPs and CSPs of other bark beetle species. DarmCSP5 showed 79% identity with Dendroctonus valens CSP4, whereas other DarmCSPs shared 86–97% identity with Dendroctonus ponderosae CSPs (**Table 1**). DarmCSP amino acid sequence alignment revealed a typical four-cysteine motif at conserved positions. In addition, nine DarmCSPs shared four conserved amino acids: one arginine before the first "C," as well as one glycine, one leucine, and one proline between the second and third "C." DarmCSP5 contained an exceptionally long C-terminus (**Figure 1**).

Phylogenetic analysis indicated that DarmCSPs were clustered together with CSPs of other bark beetles (D. ponderosae, Ips typographus, and D. valens). However, DarmCSPs were divergent in both the phylogenetic and homology trees, with only 25–56% amino acid identity within species (**Figure 2**, **Figure S1**).

## Distribution of *DarmCSPs* Across Development and Tissues

#### Expression Patterns Across Development

DarmCSPs were broadly expressed across development of D. armandi, but with different profiles. Interestingly, DarmCSP1, DarmCSP3, DarmCSP7, and DarmCSP8 were highly expressed in adults, but had significantly lower expression in larvae and pupae. DarmCSP1, DarmCSP7, and DarmCSP8 were highly expressed in emerged adults, whereas DarmCSP3 was highly expressed in feeding adults. In contrast, DarmCSP4, DarmCSP5, and DarmCSP6 were highly expressed in mature larvae and pupae, but lowly expressed in adults, especially at the emerged sub-stage. DarmCSP2 and DarmCSP9 were more highly expressed during the late pupae stage than in other stages, but their expression was also relatively high in adults. DarmCSP2 and DarmCSP3 had relatively high expression in larvae only (**Figure 3**).

#### Expression Patterns Across Tissues

Nine DarmCSPs were expressed at varying levels and with occasional sex differences across multiple tissues. DarmCSP1, DarmCSP2, DarmCSP3, and DarmCSP7 were highly expressed in antennae of both sexes. DarmCSP3 expression was predominantly in this tissue, but the remaining three were also ubiquitous in other tissues at relatively high levels. Specifically, DarmCSP2 was highly expressed in mouthparts, abdomens, thoraxes, and legs, with a significantly higher expression in females than in males among the latter two tissues. DarmCSP7 was more highly expressed in male than in female forewings. DarmCSP4, DarmCSP5, and DarmCSP8 had significantly higher expression in both male and female mouthparts, whereas DarmCSP9 expression was significantly higher in female mouthparts. DarmCSP9 was also more highly expressed in female than in male heads. However, its expression was significantly higher in male pheromone glands. Apart from its high expression in mouthparts, DarmCSP8 was also present in other tissues at relatively high levels. Finally, DarmCSP6 was ubiquitous in most tissues, with notably high expression in abdomens and thoraxes but low expression in antennae (**Figure 4**).

#### Binding Characteristics of DarmCSPs Bacterial Expression and Purification of DarmCSPs

Three pET32a(+)/DarmCSPs were successfully induced and expressed in BL21(DE3) PlysS cells. DarmCSP1 and DarmCSP2 exhibited good yield (more than 20 mg/L), whereas DarmCSP3 had lower expression. These three proteins were located in the supernatant after sonication. The results of 12% SDS-PAGE indicated that recombinant and pure proteins without Histags were respectively present as single bands at 32.0 and 14.0 kDa (without signal peptide) (**Figure S2**). This outcome accords with deduced molecular weights of the predicted amino acid sequences.

#### Fluorescence Binding Assays of DarmCSPs

DarmCSP2 interacted strongly with 1-NPN, exhibiting dissociation constants of 1.84 ± 0.04µM. In contrast, DarmCSP1 and DarmCSP3 had no obvious affinity to 1-NPN. Saturation results and linear Scatchard plots revealed only a single binding site for 1-NPN in DarmCSP2, with no allosteric effects, indicating that 1-NPN was suitable as the fluorescence probe (**Figure 5A**).

Fluorescence competitive binding assays revealed high binding affinity (Ki < 10µM) of DarmCSP2 to all tested host volatiles, especially (−)-α-pinene and (+)-3-carene (Ki = 1.64 ± 0.08µM and Ki = 1.97 ± 0.46µM, respectively) (**Figure 5B**, **Table 2**). Notably, DarmCSP2 showed high (Ki < 10µM) and moderate affinity (Ki < 20µM) to four pheromones (two in each category), with especially strong bonds to (−)-trans-verbenol (Ki = 2.80 ± 0.07µM) (**Figure 5C**, **Table 2**).

#### Structural Model of DarmCSP2

The 3D-structural model of DarmCSP2 revealed six α-helices, plus a very short one near the carboxyl terminus, all connected with loops to form a binding pocket. This structure is typical of CSPs. Active sites I73 and W80 in DarmCSP2 corresponded to I76 and W83 residues in S. gregaria CSP4 (**Figures 6A,B**).

#### Efficiency Analysis of RNAi on *DarmCSP2* Effect of dsRNA Treatment on DarmCSP2 Transcript Level

Injection of dsCSP2 significantly decreased target gene expression level, according to qRT-PCR results. The dsCSP2 injected group did not differ from controls (non-injected and water-injected) in DarmCSP2 mRNA levels 12 h post-injection, a significant difference emerged after 24 h, followed by a continuous decrease from control levels after 48 h (**Figure S3**).

#### Effect of dsRNA Treatment on Electrophysiological Responses to Host Volatiles and Pheromones

At 48 h post-injection, dsCSP2-injected antennae did not exhibit significant decreases in response to four test pheromones, compared with controls. However, dsCSP2 injection

the alignment.

significantly reduced antennae EAG activity in response to six test host volatiles, including: (+)-α-pinene, (+)-β-pinene, (−)-β-pinene, (+)-camphene, (+)-3-carene, and myrcene (**Figure 7**).

#### DISCUSSION

In this study, nine full-length DarmCSP genes were cloned and identified. This number is close to the amount found in several other bark beetle species: 11 in D ponderosae (Andersson et al., 2013), six in D. valens (Gu et al., 2015), and six in I. typographus (Andersson et al., 2013). DarmCSPs were classical CSP genes based on a variety of hallmarks (Vieira and Rozas, 2011). First, their deduced amino acid sequence revealed a typical four-cysteine motif at conserved positions, conforming to the CSP model of C1-X6–8-C2-X16–21-C3-X2-C4 (X represents any amino acid) (Pelosi et al., 2006). Furthermore, at the N-terminus, DarmCSPs contained a putative signal peptide of 16–25 amino acids in length (**Figure 1**).

DarmCSPs were closely related to CSPs in other bark beetles. Exhibiting high amino acid sequence identity with D. ponderosae and D. valens CSPs (**Table 1**), DarmCSPs were also clustered together with CSPs of other bark beetles in the phylogenetic analysis (**Figure 2**). Previous reports have indicated that D. ponderosae, I. typographus, and D. valens CSP genes are orthologous (Andersson et al., 2013; Gu et al., 2015). Together, these results indicated that barkbeetle CSP genes may have similar expression profiles and function.

Among the nine DarmCSPs, amino acid sequences exhibited considerable variation in identity similarity (25– 56%) (**Figure S1**). This variation was similar to sequence identity percentages in Nilaparvata lugens (10–77%) (Yang et al., 2014), Bombyx mori (10–50%) (Qiao et al., 2013), and

Independent-Samples *T*-Test). All values are mean ± sd, *n* = 3.

Papilio xuthus (20–70%) (Ozaki et al., 2008). Reflecting the sequence variation, all nine DarmCSPs were distributed in different branches of the phylogenetic tree, a pattern also found in other bark beetles (Andersson et al., 2013; Gu et al., 2015). The diversification in DarmCSP amino acid sequences suggested multiple functions.

Supporting that idea is the observation of broad variety in expression patterns among DarmCSPs (**Figures 3**, **4**). DarmCSP4, DarmCSP5, and DarmCSP6 were all highly expressed in mature larvae and pupae, stages when insects stop feeding and experience enormous morphological changes. We also observed a sudden upregulation of DarmCSP2 and DarmCSP9 before emergence. Therefore, these five DarmCSP genes may be involved in D. armandi metamorphosis. Findings in other insects support this conclusion. Specifically, regulation of CSP expression in Choristoneura fumiferana, B. mori, and Nilaparvata lugens varies with hormonal changes during metamorphosis (larvae and pupae or nymphs) (Wanner et al., 2005; Gong et al., 2007; Yang et al., 2014; Hou et al., 2016).

DarmCSP1, DarmCSP2, DarmCSP3, and DarmCSP7 were highly expressed in antennae, with CSP3 almost exclusively found there. Furthermore, DarmCSP2, DarmCSP4, DarmCSP5, DarmCSP8, and DarmCSP9 genes were enriched in mouthparts. Antennae and mouthparts are the primary chemosensory organs of insects, each covering a different function. The antennapreferential genes are probably involved in recognizing sex pheromones and plant volatiles (Tomaselli et al., 2006; Qiao et al., 2013; Yang et al., 2014; Li et al., 2015, 2016), whereas mouthpart-preferential genes likely play roles in gustation, recognizing non-volatile food sources or detecting closerange odors (Nagnan-Le Meillour et al., 2000; Jin et al.,

2006; de la Paz Celorio-Mancera et al., 2012; Hua et al., 2012).

We also found that DarmCSP1, DarmCSP2, DarmCSP7, DarmCSP8, and DarmCSP9 were ubiquitous in other tissues at relatively high levels, suggesting involvement in other adult physiological processes (Nomura et al., 1992; Gong et al., 2012; Gu et al., 2013; Zhou et al., 2013). In particular, DarmCSP6 was ubiquitous and highly expressed in most tissues, especially the abdomens and thoraxes. Coupled with its relatively low expression in antennae, these results suggest that DarmCSP6 mainly affects physiological processes, but not excluding chemoreception. In sum, tissue and developmental expression profiles indicate that DarmCSPs serve numerous functions in metamorphosis, olfaction, and gustation.

Because the primary mechanism of insect CSPs is to recognize and bind exogenous hydrophobic chemicals to receptors through the sensillum lymph of chemosensory organs (Pelosi et al., 2005; Liu et al., 2012; Leal, 2013), we examined the binding affinity of DarmCSPs. Our binding assays revealed that DarmCSP2, but not DarmCSP1 or 3, has high affinity for 1- NPN, partially corresponding to B. mori data showing that BmorCSP1 and 2 bound well to 1-NPN, whereas BmorCSP9 and 12 do not (Qiao et al., 2013). Furthermore, when we examined the competitive ligand binding properties of DarmCSP2 specifically, we found that the protein bound strongly to all tested host volatiles [especially (−)-α-pinene and (+)-3-carene] and various pheromones [especially to (−)-trans-verbenol]. In previous studies, the tested volatiles effectively elicited different degrees of EAG responses in D. armandi antennae, and some of them were an important constituent of attractants of D. armandi (Zhang et al., 2010; Xie and Lv, 2012; Chen et al., 2015; Zhao et al., 2017a,b).

The 3D model of DarmCSP2 revealed an internal hydrophobic binding cavity formed from six α-helices, corresponding to existing studies on CSP structure (Campanacci et al., 2003; Mosbah et al., 2003; Tomaselli et al., 2006; Kulmuni and Havukainen, 2013). Additionally, active sites I73 and W80 in DarmCSP2 corresponded to I76 and W83 residues in SgerCSP4, confirmed to bind oleamide (Tomaselli et al., 2006). Thus, these active sites are likely involved in binding to pheromones. Combined with the high expression of DarmCSP2 in antennae and mouthparts, these data suggest that DarmCSP2 may be a major carrier of the tested ten host volatiles and four pheromones of D. armandi. Data on CSPs in diverse insects also support this binding function: the proteins bind pheromone components in Schistocerca gregaria (Li et al., 2015), host plant volatiles and non-volatile secondary metabolites in Apolygus lucorum (Hua et al., 2012), as well as host plant volatiles and sex pheromones in Sesamia inferens and Microplitis mediator (Zhang et al., 2014; Peng et al., 2017).

The importance of DarmCSP2 in binding to major volatiles was further confirmed by our RNAi experiment. The injection of dsCSP2 significantly decreased DarmCSP2 expression, and antennae subjected to RNAi experienced significantly reduced EAG activity in response to six tested host volatiles [(+) α-pinene, (+)-β-pinene, (−)-β-pinene, (+)-camphene, (+)-3 carene, and myrcene], but not in response to pheromones. This list corresponded well with the list of volatiles found to be bound by DarmCSP2 in fluorescence binding assays. Our results corroborate recent RNAi studies that demonstrated how the silencing of genes encoding OBPs or CSPs abolished or modified electrophysiological responses, influenced odor preferences, disrupted behavior, and altered morphology in insects (Maleszka et al., 2007; Gong et al., 2012; Yi et al., 2013; Wu et al., 2016; Dong et al., 2017; Zhang et al., 2017). Together, our results and previous work suggest that DarmCSP2

collaborates with multiple binding proteins (including other CSPs and OBPs) to transport numerous compounds. For instance, in Anopheles gambiae, OBP1 and OBP4 were coexpressed in some antennal sensilla, forming heterodimers in the sensillum lymph that differed in binding characteristics from the individual proteins (Qiao et al., 2011). In Adelphocoris lineolatus, a mixture of AlinCSP5 and AlinCSP6 increased binding affinities to terpenoids that did not bind with individual AlinCSP (Sun et al., 2015). In Helicoverpa armigera, HarmPBP1 and HarmPBP2 were associated with the recognition of the major sex pheromone component, Z11-16:Ald (Dong et al., 2017). Indeed, this phenomenon of olfaction-related binding proteins forming complexes may be universal across insects, given the clear advantages in increasing binding capacity and accuracy, thus expanding their chemical communication potential.

In this study, we combined molecular and physiological methods to clarify DarmCSPs characteristics and functions. We hypothesized that they are involved in developmental metamorphosis, as well as olfaction and gustation in the adult chemosensory system. Their role in olfaction was particularly notable; CSP2 was abundant in antennae and carried host

volatiles that regulated D. armandi foraging behaviors. These data clarified the molecular mechanisms of olfactory perception in D. armandi, providing a theoretical foundation for eco-friendly pest control.

## AUTHOR CONTRIBUTIONS

ZL, LD, and HuC designed the experiments. ZL, HoC, DF, and YS preformed the experiments; ZL analyzed data and drafted manuscript. ZL, LD, HoC, and HuC revised the manuscript. All authors read and approved manuscript for final submission.

#### ACKNOWLEDGMENTS

We acknowledge the financial support of the National Key Research and Development Program of China (2017YFD0600104) and The Natural Science Basic Research Plan in Shaanxi Province of China (2017ZDJC-03).

#### REFERENCES

Andersson, M. N., Grosse-Wilde, E., Keeling, C. I., Bengtsson, J. M., Yuen, M. M., Li, M., et al. (2013). Antennal transcriptome analysis of the chemosensory gene families in the tree killing bark beetles, Ips typographus and Dendroctonus

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00291/full#supplementary-material

Figure S1 | Homology Tree of amino acid of DarmCSPs. The tree was constructed with DANMAN. The percentages were the identity of DarmCSPs.

Figure S2 | Prokaryotic expression and purification of three CSPs of *D. armandi* analyzed on SDS-PAGE (12%). Lane M, protein molecular weight marker (top to bottom: 200, 66.4, 44.3, 29.0, 20.1, 14.3, 6.5 kDa); Lane 1, total protein extracted from BL21 bacteria cells with pET32a/DarmCSP vector after induced by IPTG; Lane 2, protein in precipitate after sonication; Lane 3, protein in supernatant after sonication; Lane 4, purified fusion protein pET32a(+)/CSP; Lane 5, purified DarmCSP after His-tag cleavage by rEK.

Figure S3 | Relative expression level of *DarmCSP2* at 12, 24 and 48 h after dsCSP2 injection. Non-injected and water-injected were as control. The significant differences among different treatments in every time point were marked with letters (*P* < 0.05, one-way ANOVA, all values are mean ± sd, *n* = 3).

Table S1 | Primers used for gene isolation, RT-qPCR, Prokaryotic expression, and dsRNA synthesis.

ponderosae (Coleoptera: Curculionidae: Scolytinae). BMC Genomics 14:198. doi: 10.1186/1471-2164-14-198

Andersson, M. N., Larsson, M. C., BlaŽenec, M., Jakuš, R., Zhang, Q.-H., and Schlyter, F. (2010). Peripheral modulation of pheromone response by inhibitory host compound in a beetle. J. Exp. Biol. 213, 3332–3339. doi: 10.1242/jeb.044396


and evolutionary history of the chemosensory system. Genome Biol. Evol. 3, 476–490. doi: 10.1093/gbe/evr033


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Li, Dai, Chu, Fu, Sun and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Silencing of Chemosensory Protein Gene NlugCSP8 by RNAi Induces Declining Behavioral Responses of Nilaparvata lugens

Muhammad I. Waris <sup>1</sup> , Aneela Younas <sup>1</sup> , Muhammad T. ul Qamar <sup>2</sup> , Liu Hao<sup>1</sup> , Asif Ameen<sup>3</sup> , Saqib Ali <sup>1</sup> , Hazem Elewa Abdelnabby 1,4, Fang-Fang Zeng<sup>1</sup> and Man-Qun Wang<sup>1</sup> \*

*<sup>1</sup> Hubei Insect Resources Utilization and Sustainable Pest Management Key Laboratory, College of Plant Science and Technology, Huazhong Agricultural University, Wuhan, China, <sup>2</sup> College of Informatics, Huazhong Agricultural University, Wuhan, China, <sup>3</sup> College of Agronomy and Biotechnology, China Agricultural University, Beijing, China, <sup>4</sup> Department of Plant Protection, Faculty of Agriculture, Benha University, Banha, Egypt*

#### Edited by:

*Shuang-Lin Dong, Nanjing Agricultural University, China*

#### Reviewed by:

*Pin-Jun Wan, China National Rice Research Institute (CAAS), China Joe Hull, Agricultural Research Service (USDA), United States Bin Tang, Hangzhou Normal University, China Shuo Li, Jiangsu Academy of Agricultural Sciences (JAAS), China*

> \*Correspondence: *Man-Qun Wang mqwang@mail.hzau.edu.cn*

#### Specialty section:

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

Received: *10 January 2018* Accepted: *27 March 2018* Published: *12 April 2018*

#### Citation:

*Waris MI, Younas A, ul Qamar MT, Hao L, Ameen A, Ali S, Abdelnabby HE, Zeng F-F and Wang M-Q (2018) Silencing of Chemosensory Protein Gene NlugCSP8 by RNAi Induces Declining Behavioral Responses of Nilaparvata lugens. Front. Physiol. 9:379. doi: 10.3389/fphys.2018.00379* Chemosensory proteins (CSPs) play imperative functions in chemical and biochemical signaling of insects, as they distinguish and transfer ecological chemical indications to a sensory system in order to initiate behavioral responses. The brown planthopper (BPH), *Nilaparvata lugens* Stål (Hemiptera: Delphacidae), has emerged as the most destructive pest, causing serious damage to rice in extensive areas throughout Asia. Biotic characteristics like monophagy, dual wing forms, and annual long-distance migration imply a critical role of chemoreception in *N. lugens*. In this study, we cloned the full-length CSP8 gene from *N*. *lugens*. Protein sequence analysis indicated that NlugCSP8 shared high sequence resemblance with the CSPs of other insect family members and had the typical four-cysteine signature. Analysis of gene expression indicated that NlugCSP8 mRNA was specifically expressed in the wings of mated 3-day brachypterous females with a 175-fold difference compare to unmated 3-day brachypterous females. The NlugCSP8 mRNA was also highly expressed in the abdomen of unmated 5-day brachypterous males and correlated to the age, gender, adult wing form, and mating status. A competitive ligand-binding assay demonstrated that ligands with long chain carbon atoms, nerolidol, hexanal, and trans-2-hexenal were able to bind to NlugCSP8 in declining order of affinity. By using bioinformatics techniques, three-dimensional protein structure modeling and molecular docking, the binding sites of NlugCSP8 to the volatiles which had high binding affinity were predicted. In addition, behavioral experiments using the compounds displaying the high binding affinity for the NlugCSP8, revealed four compounds able to elicit significant behavioral responses from *N*. *lugens*. The *in vivo* functions of NlugCSP8 were further confirmed through the testing of RNAi and post-RNAi behavioral experiments. The results revealed that reduction in NlugCSP8 transcript abundance caused a decrease in behavioral response to representative attractants. An enhanced understanding of the NlugCSP8 is expected to contribute in the improvement of more effective and eco-friendly control strategies of BPH.

Keywords: Nilaparvata lugens, chemosensory protein, expression patterns, competitive binding assay, behavioral trial, RNA interference, molecular docking

#### INTRODUCTION

A considerable amount of literature has been published on insect olfactory systems. These studies recognized that olfactory systems are particularly sensitive and complex (Forêt et al., 2007; Yoshizawa et al., 2011; Gu et al., 2012; Sun L. et al., 2016). The olfactory system has immense importance in the insects because it can detect and identify a variety of chemicals from the environment (Li et al., 2017a). Investigational studies on its functions have elucidated some of the molecular components and pathways that insects utilize in identifying conspecifics, detect enemies, find mates, locate oviposition site, and to avoid natural enemies (Field et al., 2000; Bruyne and Baker, 2008; Qiao et al., 2013; Li et al., 2017b). The high specificity and sensitivity of the insect olfactory system mostly rely on the interaction between semiochemicals and different types of protein expressed in the olfactory sensilla of insects, such as sensory neuron membrane proteins (SNMPs), membrane-bound olfactory receptors (ORs) and two types of carrier protein: chemosensory proteins (CSPs) and the odorant binding proteins (OBPs) (Pelosi et al., 2006, 2017; Leal, 2013; He and He, 2014; He et al., 2014). Chemosensory proteins encompass a family of acidic, low-molecular-mass and soluble proteins in the lymph of insect olfactory receptors and probably play significant roles in insect chemoreception, such as differentiating, binding, and transporting hydrophobic chemicals from the surroundings to olfactory sensilla (Kaissling, 2001; Pelosi et al., 2005; Gong et al., 2007; Jin et al., 2017). CSPs were originally identified in the antennae of Drosophila melanogaster by McKenna et al. (1994). CSPs are around 100–120 residues long and present a conservative model of four cysteines forming two independent loops (Angeli et al., 1999; He et al., 2017). CSPs also have αhelical segments but accumulated in a folding different from that of insect OBPs (Jansen et al., 2007; Northey et al., 2016). Through the expressed sequence tag (EST) and transcriptome databases in addition with the development of genome comprehensive surveys, more and more CSP families and their biochemical functionality/expressions have been described in many insect species (Zhou et al., 2006). Various CSPs are known to be ubiquitously expressed in insects and shown to be interrelated with larval development, detection of carbon dioxide, and regeneration of tissues (Pelosi et al., 2006; Li et al., 2016; Iovinella et al., 2017). However, data has revealed that CSPs or CSP-like genes are expressed not only in the antennae, the main olfactory organ (Zhang et al., 2009), but also in the wings (Zhou et al., 2008), legs (Picimbon et al., 2001), pheromone glands (Dani et al., 2010), proboscis (Liu et al., 2014), as well as in all other components of insect body (Gong et al., 2007), and involved in odor recognition (Sánchez-Gracia et al., 2009). This comprehensive and varied expression pattern proposes that CSPs may play several functions, beyond chemosensation (Tegoni et al., 2004). CSPs highly enriched in antennae have proposed chemosensory functions in Lepidoptera (Qiao et al., 2013). Antennae-enriched CSP1 from Microplitis mediator play important functions in chemoreception and used as a potential target to regulate the olfactory behavior in M. mediator (Peng Y. et al., 2017). Other CSPs highly expressed in antennae have been concerned with serving functions in the behavioral phase change in Locusta migratoria (Guo et al., 2011). In Spodoptera exigua, SexiCSP3 has been associated with egg hatching and ovipositions (Gong et al., 2012), while PameCSP10 in Periplaneta americana appears to be the main extracellular matrix protein during limb regeneration (Kitabayashi et al., 1998). The chemosensory protein, Si-CSP1 involved in regulating the necrophoric behavior of workers in Solenopsis invicta (Qiu and Cheng, 2017). Numerous studies designated that CSPs may be involved in immune response, circadian cycles or developmental process (Oduol et al., 2000; McDonald and Rosbash, 2001; Sabatier et al., 2003). CSPs are, therefore, expected to perform many miscellaneous tasks from behavior to several physiological and biological processes (Pelosi et al., 2017). Ligands from different sources, such as plant volatiles (Fujii et al., 2010), cuticular lipids (González et al., 2009), cuticular hydrocarbon (Ozaki et al., 2005), and brood pheromones (Briand et al., 2002), are usually used in the fluorescence binding assays to characterize the binding affinity between CSPs and various odorants. Multiple functions proposed or documented that CSPs have the capability to bind and interact with small molecules, from nutrients to semiochemicals, toxic compounds or hormones (Pelosi et al., 2017). These extraordinarily complex binding functionality and expression profiles proposed that CSPs might play an important role in the insect chemosensory systems, while their exact physiological functions and mechanisms still remains unclear (Sánchez-Gracia et al., 2009).

The brown planthopper (BPH), Nilaparvata lugens (Stål) (Hemiptera: Delphacidae), is a major insect pest of rice in extensive area throughout Asia and could cause enormous economic losses (Dong et al., 2011; Bottrell and Schoenly, 2012; Peng L. et al., 2017). BPH is a monophagous herbivore that mainly feeds on cultivated rice and its associated wild rice, and therefore the strategies being used to find rice plants would be vital in BPH (Sogawa et al., 1982). In rice plants, BPH decreases the photosynthetic rate, chlorophyll content, nitrogen concentrations of stem and leaf, and organic dry weight, thereby intensively decreasing yield (Ye et al., 2017). In the adult stage, BPH shows two wing forms, short (brachypterous) and long (macropterous) ones. The long wing adults exhibit the capability to migrate across long distances, while the short wing adults expound strong reproductive abilities (Bottrell and Schoenly, 2012; Cheng et al., 2013). These biotic characteristics imply the critical role of chemoreception in BPH. However so far, limited olfactory-interrelated proteins have been categorized in N. lugens. Total of 10 genes encoding OBPs (NlugOBP1-10) and 11 genes encoding CSPs (NlugCSP1-11) are predicted from the genome in previous studies (Xu et al., 2009; He et al., 2011; Yang et al., 2014; Zhou et al., 2014). Of these predicted genes, only one CSP gene (NlugCSP7) has been cloned from the antennae of N. lugens and subsequently identified as volatile organic compound binding capabilities (Yang et al., 2014). However, previous ligand-binding analysis of NlugCSP7 revealed that it may possess physiological functions other than the chemosensation (Yang et al., 2014). The functions of other chemosensory related proteins are still unknown in N. lugens. To date, very little attention has been paid to the functions of N. lugens chemosensory related proteins. Previous studies also demonstrated that NlugCSP8 may play roles in perception of rice plant volatiles after the N. lugens dispersion (Yang et al., 2014). To confirm their specific functional roles, we conducted a more thorough study of NlugCSP8 expression and functionality. The main objective of this paper is to recognize the functions of NlugCSP8 during development. We performed qRT-PCR to monitor the expression of NlugCSP8 during different development stages of unmated and mated adults in terms of wing forms, tissues, and genders. Binding properties of NlugCSP8 were also tested using a number of ligands in fluorescence binding assay. In addition, molecular docking analyses followed by targeted gene silencing using RNAi combined with behavior bioassay were conducted.

## MATERIALS AND METHODS

#### Insect Rearing and Tissue Collection

Successive generations of BPH were reared on susceptible rice variety Taichung Native 1 (TN1) in a climatic chamber under constant conditions of 26 ± 2 ◦C, 75 ± 5% relative humidity and 16-h light: 8-h dark photoperiod. For the expression pattern analysis, unmated and mated 3- and 5-days short/long wing adults of both sexes were collected. To obtain mated male and female adults, newly emerged BPH males and females were paired in glass tubes and allowed to mate. Before sample collection, the age of adults was checked and confirmed according to previous literature (Wipfler et al., 2016). The tissues were dissected from antennae, head (without olfactory appendages), abdomen, legs, and wings of unmated and mated short and long wing adults of both sexes and collected for qRT-PCR. All samples with three replicates (50 individuals per replication) were kept at −80◦C and further arranged according to age, mating status, and sex.

## RNA Extraction and cDNA Synthesis

Total RNA was extracted from individual samples by using the TRIzol reagent (Invitrogen, CA). Then quality and quantities were examined by using 1.0% agarose gel electrophoresis and ultraviolet spectrophotometer (Eppendorf Bio Photometer Plus, Germany). The first-strand cDNA for RT-PCR and qRT-PCR were synthesized from 1 µg of total RNA using MBI RecertAid First Strand cDNA kit (MBI Fermentas, Glen Burnie, MD, USA) and PrimerScript RT Reagent kits with gDNA Eraser (Perfect Real Time; Takara) respectively, according to manufacturer's instructions. The synthesized cDNA was stored at −20◦C for future use.

## NlugCSP8 Sequence Analysis

NlugCSP8 was identified with a complete coding sequence from our previous cDNA library (Zhou et al., 2014). The open reading frame (ORF) was recognized using the ORF finder software (http://ncbi.nlm.nih.gov/gorf/gorf.html). The molecular weight was calculated using the SWISS-PROT (ExPASy server) program "Compute pI/Mw." The signal peptides were predicted using SignalP V3.0 (http://www.cbs. dtu.dk/services/SignalP/). NlugCSP8 similarity search to identify homologous genes from other insect species were performed using the NCBI-BLAST (http://blast.ncbi.nlm.nih.gov/) and sequences were further aligned by using ClustalX 1.83 and GeneDoc 2.7 computer programs (Thompson et al., 1997). Multiple sequence alignment has been performed and the evolutionary tree was constructed using the neighbor-joining method with MEGA 6.0 (Tamura et al., 2013).

## Quantitative RT-PCR

A quantitative RT-PCR (qRT-PCR) was used to study the spatiotemporal expression profiles of NlugCSP8 in mRNA level in unmated and mated, 3 and 5 days old, short and long wing adults of both sexes. We generated cDNA from the selected tissues of the short and long wings of both N. lugens sexes in different mating stages and age groups. β-actin (GenBank accession number: EU179846) was used as an internal control (Liu et al., 2008). Primer sequences were designed using the Primer 5.0 program (Premier Biosoft International, Palo Alto, CA, USA). A 10-fold dilution series was used to construct a standard curve in order to determine the qRT-PCR efficiencies and to quantifying the amount of target mRNA. In all experiments, all primers achieved amplification efficiencies of 95–100%. The qRT-PCR samples contained 10 µl of 2× Syber Green PCR Master Mix, 0.5 µl of each primer (10µM), 1 µl of cDNA and 8 µl sterilized ultrapure water. Thermal cycling was performed using an initial denaturation step at 95◦C for 3 min, followed by 40 cycles of 95◦C for 10 s and 55◦C for 30 s. The qRT-PCR was performed in triplicate using three biological samples and the relative Ct-values were quantified using the 2−11CT method (Livak and Schmittgen, 2001; Li et al., 2014).

#### NlugCSP8 Expression Vector System Construction

For expression of NlugCSP8 (NlugCSP8; accession no. ACJ64054.1), the sequence encoding NlugCSP8 was amplified by PCR with a forward primer containing an EcoRI-restriction site and a reverse primer containing an XhoI-restriction site (**Table S1**). The PCR product was ligated into a pMD-18T vector and sequenced. The pMD-18T plasmid containing target sequence flanked by the two restriction sites was digested with EcoRI and XhoI restriction enzymes and ligated into the expression vector pET-30a, which was earlier linearized with the same restriction enzymes. The obtained plasmids were sequenced and shown to encode the mature protein.

## Expression and Purification of Recombinant NlugCSP8

The recombinant pET-30a/CSP8 expression plasmid was transformed into Escherichia coli BL21 (DE3) competent cells. After DNA sequencing, a single positive clone was grown in 10 mL Luria-Bertani (LB) medium containing kanamycin (50µg/mL) with shaking overnight at 220 rpm and 37◦C. The culture was diluted to 2 L LB medium (supplemented with 50µg/mL kanamycin) and grown at 37◦C with shaking at 220 rpm until the culture reached the optical density value of ∼0.6–0.7 at 600 nm. The recombinant protein expression was induced by the addition of 2 mM IPTG (Isopropyl β-D-1 thiogalactopyranoside), followed by culturing for 4 h at 37◦C. The bacterial cells were harvested by centrifugation (10,000 rpm, 10 min) and sonicated. The expressed protein presented in the supernatant as a soluble form. Then, NlugCSP8 purification was performed using a Ni-ion affinity chromatography column (GE Healthcare, Uppsala, Sweden). His-tag was removed from the recombinant protein with the addition of recombinant bovine enterokinase (EK) in the eluted fractions of protein, followed by 16 h incubation at 25◦C. After running the digested protein back through the Ni-ion affinity chromatography column, the tag-free protein was obtained in the flow through fraction. Protein expression and purification steps were assessed by 15% SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis). Finally, purified protein was dialyzed in Tris buffer (pH 5.0) and (pH 7.4). The concentration of purified protein was determined prior to perform ligand-binding specificities of NlugCSP8 with 25 selected ligands at pH 5.0 and pH 7.4.

#### Fluorescence Ligand Binding Assays

Fluorescence-based ligand binding assays were performed based on the method described by Sun X. et al. (2016). According to previous studies about the rice-specific volatiles (Fujii et al., 2010; He et al., 2011; Yang et al., 2014; Zhang et al., 2014), 25 potential ligands were selected for the fluorescence binding assays (**Table 1**). All the ligands used in this study were purchased from Sigma-Aldrich (St. Louis, MO, USA) and stored according to manufacturer instructions. The ligand binding affinity for various ligands was determined by using the 1-NPN (N-phenyl-1-naphthylamine) as a fluorescent probe. RF-5301PC fluorimeter (Shimadzu, Kyoto, Japan) was used for fluorescence binding assay at 25◦C with 10 nm slit width and 1 cm light path quartz cuvette for emission and excitation. The 1-NPN/NlugCSP8 mixture was excited using an excitation wavelength of 337 nm, and the fluorescence intensity was recorded between 350 and 600 nm following an established protocol (Ban et al., 2002). The 1-NPN and all the potential ligands were prepared in spectrophotometric-grade methanol. The binding constant for 1-NPN was measured by adding aliquots of 1 mM 1-NPN into a 2µM solution of protein in 30 mM Tris-HCL at room temperature. To measure the binding affinity of various potential ligands, the 2µM solution of protein was titrated with 1 mM 1- NPN with the final concentration of each ligand between 0 and 20µM. For each test, fluorescence measurement was conducted after the reaction was incubated for 2 min at room temperature (Liu et al., 2015). Three independent measurements were used to obtain the binding data. 1-NPN/NlugCSP8 dissociation constants (Kd) were calculated from Scatchard plots of the binding data using the Prism 5 software (GraphPad, La Jolla, CA, USA). The curves were linearized using the Scatchard plot program (Campanacci et al., 2001). The dissociation constants of the competitors were determined by using the corresponding IC50-values according to the equation Ki = [IC50]/ (1+[1- NPN]/K1−NPN), where IC<sup>50</sup> represents the concentration of ligand which decreases the fluorescence intensity of [1-NPN], [1-NPN] is the free concentration of 1-NPN and K1−NPN is the dissociation constant of the NlugCSP8/1-NPN complex (Ban et al., 2003; Tian and Zhang, 2016). For the reader's convenience, data were recalculated as 1/ki × 1,000, for which a larger value designates a stronger ligand binding affinity.

## Double-Stranded RNA Synthesis

The full coding sequence of NlugCSP8 and green fluorescent protein (GFP) were cloned into pMD-18T vector and used as templates for the target sequences amplification. The target sequences of NlugCSP8 and GFP were amplified by RT-PCR using specific gene primers conjugated with 19 bases of the T7 RNA polymerase promoter (**Table S1**). dsRNA was synthesized from PCR products as templates by using the T7 Ribomax Express RNAi System Kit (Promega, Madison, WI, USA). After synthesis, the dsRNA was precipitated by adding isopropanol and resuspended in nuclease-free water. The purified dsRNAs were quantified spectrophotometrically at 260/280 nm and integrity was examined by agarose gel electrophoresis.

#### dsRNA Injection and Analysis of Gene Silencing

Microinjector (World Precision Instruments Inc., Sarasota, FL, USA) fitted with a glass capillary needle was used for dsRNA injection assays. BPH was anesthetized using the CO<sup>2</sup> for 30 s and placed on agarose plate. Prior to injection with dsRNAs, BPH was placed in the groove using a pointed brush. Each individual was nanoinjected with 30 nL of 5 ng/nL dsRNAs into the conjunctive between prothorax and mesothorax under a microscope. For dsCSP8 and dsGFP, 100–150 3rd instar nymphs were injected in every replication and three biological replicates were used. Injected nymphs were placed on fresh rice seedlings to recover, and reared at 26 ± 2 ◦C, humidity 75 ± 5% and 8/16 h dark/light for 1–7 days. The mortality was recorded every day following injection. Six synchronous nymphs were selected randomly at 1st, 2nd, 3rd, 5th, and 7th days after injection for subsequent RNA extraction. The relative mRNA expression levels were determined in the injected group, while others were normalized to one in the non-injection group. All the data were expressed as the mean ± SE of three separate measurements.

## Olfactory Behavioral Assays

BPH behavior responses to different ligands were tested in an H-tube olfactometer similar to which previously used by Yi et al. (2018) in our laboratory. The H-tube olfactometer mainly consists of two glass tubes (arms) with gauze at its top end. These two glass tubes were connected by another tube (5 cm in diameter, 20 cm long with a hole of 1 cm in the middle for releasing BPH). Twenty macropterous (10 from each sex) BPH adults were introduced into the H-tube and the number of BPH was counted at 30 min after their introduction. Liquid paraffin was used as in control arm. Rubber septa were absorbed in the liquid paraffin and solutions of the odor molecules to be tested (liquid paraffin+ different concentration of tested volatile) and placed at room temperature. After 24 h, one rubber septa from each control and tested volatile was put in each glass arm. After one replication, rubber septa were changed and three treatments (1, 10, and 100 µl/mL) of tested volatiles against macropterous adults were tested in eight TABLE 1 | Binding affinities of different ligands (long chain and without long chain) to NlugCSP8 evaluated via competitive ligand binding assays by using the fluorescent probe, 1-NPN.


replications. After four replications, the H-tube olfactometer was washed with 75% alcohol and the liquid paraffin rubber septa were placed in another arm to complete the other four replications. The impact of NlugCSP8-dsRNA on the preference of N. lugens was also tested by H-tube olfactometer assays. Corresponding control experiments without dsRNA injection were performed to investigate whether the preference of N. lugens was affected by volatile concentration change. Three treatments of BPH (NlugCSP8-dsRNA injected, GFP-dsRNA injected, and without injection) were tested in four replications. As in case of dsRNA injected insects, the concentration of volatiles that have highly significant attractive results on non-injected insects used for dsCSP8 and dsGFP injected insects. In order to evaluate the best RNAi effect, mRNA levels of NlugCSP8 dsRNA injected insects were determined and compared with the GFP-dsRNA-injected and non-injected insects, prior to Htube olfactometer bioassay. Based on the findings of previous step, we re-inject the BPH and the individuals with the best post-injection RNAi effect after 7-days, were used in the H-tube olfactometer bioassay. Bioassays were performed under controlled conditions at 26 ± 2 ◦C and 75 ± 5% relative humidity.

## Molecular Modeling and Ligand Docking

Delta-BLAST was performed (NCBI: http://blast.ncbi.nlm. nih.gov/Blast.cgi) with the NlugCSP8 sequence, against the protein data bank (PDB: http://www.rcsb.org) by using the SWISS-MODEL server (SWISS-MODEL: http://swissmodel. expasy.org/). After BLAST resulted sequences having identities > 40% were selected for subsequent analysis and Clustal W (http://embnet.vital-it.ch/software/ClustalW.html) was used for multiple sequence alignment. The top hit protein sequence was selected on the basis of sequence homology, query coverage, phylogeny and the number of Cys (cysteine) residues, and the template of CSPsg4 from Schistocerca gregaria (PDB ID: 2GVS\_A) was further used to build a 3D model of NlugCSP8 (Tomaselli et al., 2006). Regarding molecular docking studies, a number of docking programs are available; here we used Docking protocol implemented in MOE (MOE, version 2012.10) designed by Chemical Computing Group (Vilar et al., 2008), in order to predict the binding sites of NlugCSP8. The ligands [Nerolidol, Hexanal, Trans-2-hexenal, 2-Heptanol, and (−)-terpinen-4-ol] were chosen to dock into the binding pocket of the 3D structure of NlugCSP8 because these ligands exhibited strong binding affinities with NlugCSP8 in experimental analysis. The default parameters have been used to calculate the interaction of ligand molecules and score against respective ligands (Rescoring 1: London dG, Refinement: Forcefield, Rescoring 2: GBVI/WSA dG, Placement: Triangle Matcher). The most suitable docked ligand-protein structure was designated on the basis of RMSD (Root Mean Square Deviation) values and minimum S-score. The S-score is the value calculated by built-in scoring functions of MOE on the basis of ligand-binding affinity with receptor protein after docking. While, RMSD value is generally used to compare the docked conformation with the reference conformation or with other docked conformation (Wadood et al., 2014; Qamar et al., 2016).

#### Statistical Analysis

The SPSS (Statistical package for the social sciences) computer software version 22.0 was used for data analysis (SPSS Inc., Chicago, IL). All qRT-PCR data were statistically analyzed using ANOVA (one-way analysis of variance) followed by Tukey's Honestly Significant Difference (HSD) test. P < 0.05 was considered statistically significant. The chi-squared test was used to determine significant differences in the number of insects choosing a particular odor.

#### RESULTS

#### Characterization and Homology Analysis of the NlugCSP8

The full-length cDNA encoding NlugCSP8 was cloned and verified by sequencing. It showed 100% amino acid identity with the previously deposited sequence of NlugCSP8 (GenBank accession number: ACJ64054.1) (Xu et al., 2009). NlugCSP8 sequence analysis revealed a full-length Open Reading Frame (ORF) of 390 nucleotides encoding 129 amino acids residues, with an isoelectric point of 6.34 and a molecular weight of 14.6 kDa. At their N-terminus, NlugCSP8 contain signal peptide of 19 residues suggesting the solubility of NlugCSP8 (**Figure S1**). The sequence alignment of NlugCSP8 and the corresponding CSPs obtained from other hemipteran species was performed (**Figure 1**). The alignment analysis showed that four conserved cysteines obviously presented in all CSPs. The NlugCSP8 shares the highest identity (50–71%) with other hemipteran CSPs. The highest scoring identities, based on the morphological characters of their phylogenetic interactions, were 71% with Laodelphax striatella (LstrCSP12) and 62% with Sogatella furcifera (SfurCSP1). The phylogenetic relationship showed that the NlugCSP8 had closer ancestor from the same order of insects. We searched NlugCSP8 for homologs in other insect species using tblastn with an e-value cut off 10e-30. The search result revealed that NlugCSP8 possessed sequence homologous to 144 insect CSPs (**Figure S2**). Among them, there are 52 Hemipteran, 9 Dipteran, 25 Lepidopteran, 55 Coleopteran, 2 Hymenopteran, and 1 Neuropteran CSPs.

#### Expression Patterns of NlugCSP8

The qRT-PCR dataset (**Figure 2**) was based on different tissue samples (At, antennae; H, head; Ab, abdomen; L, leg; W, wing) from unmated and mated BPH at 3 and 5 days old, as well as the whole-body of mated (3, 5, 7, 9, 12, 15 days old) and unmated (3 and 5 day old) insects. The resulted dataset was employed to characterize the pattern of developmental expression of the NlugCSP8 gene in different developmental stages. Transcript levels were also tested in the two different wing forms of males and females. The qRT-PCR results showed that NlugCSP8 was highly expressed in mated brachypterous female antennae with low expression level in unmated brachypterous female antennae (**Figure 2A**). Significant differences of expression levels were also observed in the head between male and female at 5-day-old mated adults (**Figure 2B**). Pursuing this further, NlugCSP8 expression level was also higher in the unmated 5 day-old brachypterous male abdomen when compared to the mated 5-day-old brachypterous male abdomen (**Figure 2C**). On the other hand, NlugCSP8 expression in mated 3 and 5 days-old brachypterous male leg was significantly higher than unmated 3 and 5-days-old brachypterous male leg (**Figure 2D**). The qRT-PCR results displayed that the levels of NlugCSP8 mRNA were correlated with age and mating status and the gene was highly expressed in mated 3-day brachypterous female wing (**Figure 2E**). No significant differences were observed between macropterous and brachypterous BPHs of both sexes, except for unmated 3-day-old and mated 5, 9, and 15 daysold BPH (**Figure 2F**). However, significant differences between male and female expression levels were observed for NlugCSP8 in 15 days-old BPH (for mated adults). For instance, the expression levels of NlugCSP8 in mated 3, 5, and 15 days-old macropterous BPHs were higher than those in brachypterous adults at the same stage (**Figure 2F**). Closer inspection of the 3-day unmated insects showed that the expression level of NlugCSP8 was significantly higher in brachypterous male than macropterous male. Interestingly, the relative expression in macropterous females was significantly affected by mating status. The expression levels of NlugCSP8 in mated macropterous females were significantly (P < 0.05) higher than in unmated macropterous females at 5-day-old BPH (**Figure 2F**). However, significant differences were also observed between mated and unmated 5-day-old macropterous males (P < 0.05). Overall, NlugCSP8 was more highly expressed in mated males and females than in unmated individuals.

## Fluorescence Binding Assay

NlugCSP8 was successfully expressed using a bacterial system with high recombinant protein yield (about 20 mg/L) as a soluble protein. The recombinant protein was then purified by passing it through a Ni-ion affinity chromatography column. The His-tag was cleaved off with recombinant bovine enterokinase (**Figure S3**). The expression and purification of the recombinant protein were assessed by 15% SDS-PAGE (**Figure 3**). The fluorescence binding assays were performed using the fluorescent probe N-phenyl-1-naphthylamine (1-NPN) as a reporter. First, NlugCSP8 titration with increasing concentration of 1-NPN, saturated and linear Scatchard plots were observed at pH 7.4 and pH 5.0, with a dissociation constant of 4.99 and 6.80µM, respectively (**Figure 4A**). A fluorescence competitive binding assay of NlugCSP8 with long chain and without long chain

compounds using 1-NPN as a fluorescent probe was performed (**Table 1**). Considering the different mechanisms of ligandbinding and release in CSPs/OBPs, we used pH 7.4 and pH 5.0 in order to simulate the pH environment and dynamic changes in the body in vitro. **Figure 4B** compares the binding values of ligands at both pH-values. The comparison indicated that the ligands displayed higher binding affinities at pH 5.0 (**Figure 4B**). The most striking results to emerge from data is the broad binding properties of NlugCSP8 toward most of host plantderived volatiles emitted from rice. These results demonstrated that NlugCSP8 achieved the highest binding affinities with nerolidol, hexanal, trans-2-hexenal, and 2-heptanol (Ki < 10) at pH 5.0 (**Figures 4C,D**) and pH 7.4 (**Figures 4F,G**). In the same vein, the NlugCSP8 displayed high binding affinities with (−)-terpinen-4-ol (Ki < 10) at pH 5.0 (**Figure 4E**), and with R-(+)-Limonene at pH 7.4 (**Figure 4H**). However, NlugCSP8 exhibited weak binding affinity to cyclohexanol and farnesene (Ki

> 40µM) at pH 7.4 and pH 5.0. Taken together, these results also suggest that there is a relationship between the binding affinity of NlugCSP8 and carbon chain length of ligands. In particular, long chain ligands exhibited a higher binding affinity as compared with shorter chain ligands. For example, nerolidol with a backbone of 12 carbon atoms exhibited the strongest binding affinity to NlugCSP8 at pH 5.0, followed by hexanal, trans-2-hexenal, and 2-heptanol with backbones of 6, 7, and 7 carbon atoms, respectively.

#### Behavioral Trials

The behavioral responses to the 5 compounds that exhibited high binding affinities (Ki < 10µM) for the NlugCSP8 were tested in an H-tube olfactometer. Four compounds out of five were able to elicit behavioral responses in N. lugens (**Figure 5**). Contrasting responses were also observed in chemical compounds that modulate behavior due to concentration-dependent effect. BPHs

displayed repellency when the concentration of hexanal was 1 µl/mL, while it strongly behaved as attractant at 100 µl/mL. Such attraction became weakened at 10 µl/mL. Nerolidol showed a significant attraction to BPHs at a concentration of 10 µl/mL. However, the BPHs showed significant aversion to 2-heptanol and trans-2-hexenal, while (−)-terpinen-4-ol was attractive at concentrations of 1 µl/mL and 10 µl/mL with no significant effect on insect's behavior.

## Behavioral Analysis After NlugCSP8 mRNA Expression Profile Silencing by dsRNA

To determine the function of NlugCSP8 in vivo, dsRNA against N. lugens (NlugCSP8) were injected into 1-day-old third-instar nymphs. At the seventh day, the average mortality of the nymphs injected with the dsCSP8 and dsGFP increased to 55.85 and 20%, respectively (**Figure 6A**). The durations of three nymphal instars (N3-N5) were not affected by dsRNA-NlugCSP8 injection (**Figure 6B**). In addition, no significant differences were observed in the mRNA levels of the target gene between non-injected and dsGFP injected groups. NlugCSP8 expression was significantly reduced by 25.5% in 1 day after injection with 150 ng dsCSP8 (**Figure 6C**). Compared with the control group that received dsCSP8 against green fluorescent protein (dsGFP), the maximum reduction of 86±1.01% occurred at the 7th day (**Figure 6C**).

To explore the possible impact of NlugCSP8 knockdown, we conducted initial behavior screening to identify chemical compounds that may elicit behavior response in BPH. We identified two selected compounds that elicited the strongest attractive responses from BPH (**Figure 5**). For hexanal, behavior response was reduced significantly in RNAi treated insects, as compared with controls (**Figure 6D**). On the other hand, the behavioral activity of nerolidol was sharply reduced in the knockdown BPHs and the attraction activity was completely lost in the insects injected with dsCSP8. However, the ratio of "no response" BPHs in dsCSP8 group was also significantly increased compared to dsGFP and non-injection control group (**Figure 6E**).

## Three-Dimensional Structure Modeling and Molecular Docking

The NlugCSP8 sequence was compared to all known proteins in the Protein Data Bank (PDB) and the results revealed that chemosensory protein sg4 from S. gregaria (CSPsg4) (PDB ID: 2GVS\_A) achieved the highest sequence similarity (54%) with NlugCSP8 and it was selected as a template to model the 3D structure of the NlugCSP8 (**Figures 7A,B**). From the results of homology modeling, the best model (**Figure 7C**) was selected on the basis of RMSD-value (0.34Å) and its quality was further checked by Ramachandran Plot on the basis of ϕ and ψ-values constrained in specific areas (**Figure S4**). Ninety-one of residues were found in the favored region which highlights the quality of a predicted model and plot also showed a larger number of residues found in α-helices region (**Figure S4**). The results of the predicted

3D structure showed that NlugCSP8 is an α-helix-rich globular protein that consists of six α-helices: α1 (residues Leu34–Ser39), α2 (residues Gln41–Met52), α3 (residues Pro58–Ala72), α4 (residues Glu80–Lys96), α5 (residues Pro98–Tyr108), and α6 (residues Arg115–Ala122) and contains multiple hydrophobic cavities, which could be involved in ligand binding. Evaluation of structure and superimposition of selected model with the template also exhibited that it consists of six α-helices with a very low RMSD-value of 0.34Å. The RMSD-value 0.34Å indicates that both template and NlugCSP8 protein have similar folds. It also further supports the idea that the complete confirmation of the modeling target was very similar to that of the template (**Figure 7D**).

To confirm the results of our ligand binding assay and provide insight into the mechanism of NlugCSP8 interaction with host compounds, molecular docking of five selected compounds [Nerolidol, Hexanal, Trans-2-hexenal, 2-Heptanol, and (−) terpinen-4-ol] was performed. The protein binding sites and functional residues interacting between the NlugCSP8 and ligands are presented in **Table 2**. The residues identified by current docking simulations, including Lys83, Thr86, and Glu87 were the main participants for NlugCSP8, whereas residues including Ala70, Leu71, Aal74, Cys75, Met90, Lys91, Tyr114, and Tyr118 had a close relationship with NlugCSP8. **Figure 8A** shows the interaction model of the NlugCSP8 and different compounds with some potential residues. As far as NlugCSP8 is concerned, there are 5 amino acid residues (Ala70, Leu71, Ala74, Cys75, and Tyr114) that may interact with nerolidol. Glu87 and Lys83 could form a hydrogen bond (H-bond) with the nerolidol. Similarly, hexanal, 2-Heptanol, and (−)-terpinen-4-ol, which also showed strong binding to NlugCSP8, formed H-bond with NlugCSP8. The docking results displayed that the selected compounds could tightly bind toward the center of the NlugCSP8 pocket and influence its activity. In the same vein, the docking result of selected ligands presented a tunnel formed in the NlugCSP8 core and all five ligands docked at the same binding site, where all interactions between the ligands and protein involved residues from helices α3, α4, α5, and α6 (**Figure 8B**).

#### DISCUSSION

CSPs are pervasive and play pivotal roles in the survival and reproduction of arthropods (Pelosi et al., 2014). CSPs are responsible for capturing outside odorants and transport them to the olfactory receptors which are crucial for the development of an olfactory system of insects (Leal, 2013; Li et al., 2015). In insects, the number of CSPs genes ranges from 4 in D. melanogaster to almost 70 in L. migratoria indicating the number of CSPs genes variability in insect species (Zhou et al., 2013). In this study, we cloned a chemosensory protein (NlugCSP8) from the BPH, and NlugCSP8 has four conserved cysteine C1-C4, which is typical of CSPs and is shared by many other species (Cao et al., 2014; Wang et al., 2016; Xue et al., 2016). NlugCSP8 shares highest identity with CSPs from other insects, possesses the CSP common signature of low molecular mass, an isoelectric point between 5 and 6 and four conserved cysteine residues that conform to the CSP common cysteine sequence spacing pattern (Picimbon et al., 2000). We also identified other amino acid residues that are completely conserved between the examined sequences and NlugCSP8 with four conserved cysteines. The alignment of NlugCSP8 with these CSPs may support the hypothesis that CSPs are highly conserved as they share sequence identity even between CSPs from different insect species (Wanner et al., 2004), and infer important functions that might play role in insect physiology (Gu et al., 2012). In accordance with early research that CSPs had closer ancestry from the identical species, showing CSPs diversification within an order may have curtailed from duplications inside that order (Kulmuni and Havukainen, 2013).

The analysis of relative expression level in different tissues showed that NlugCSP8 is expressed in all of the tissues, indicating that NlugCSP8 has a broad tissue expression profile in N. lugens. These results also support the hypothesis that CSPs genes are expressed not only in the antennae as the main olfactory organ, but also in various parts of the insect body, such as the legs, head, thorax, proboscis, pheromone gland, and wings (Wanner et al., 2004; Zhang and Lei, 2015). In particular, some CSPlike genes have been reported to be precisely expressed in the antennae (Calvello et al., 2005), while the NlugCSP8 expression is mainly enriched in wing, abdomen and leg tissues. So far, many CSPs were expressed in different parts of insect body, and some were even expressed in non-chemosensory organs (Jacquin-Joly et al., 2001). For instance, BmorCSP10 from Bombyx mori are

at pH 5.0 and pH 7.4. (C,D,F,G) competitive binding curves of long chain ligands to NlugCSP8 at pH 5.0 and pH 7.4. (E,H) competitive binding curves of without long chain ligands to NlugCSP8 at pH 5.0 and pH 7.4. A mixture of the recombinant NlugCSP8 and 1-NPN in 30 mM Tris-HCL (pH 5.0 and pH 7.4) was titrated with 1 mM solution of each competing ligand to the final concentration of 0–20µM.

proposed to be involved in contact chemoreception. Expression of BmorCSP10 is more highly in contact organs (antennae, wings, and legs) than in noncontact organs (head, thorax, and abdomen) (Gong et al., 2007). However, the NlugCSP8 expression is also detected in contact organs (antennae, wings, and legs) and availability of olfactory sensilla on these contact organs, it is anticipated that NlugCSP8 may take part in contact chemoreception, recognizing, and transporting semiochemicals. The CSP from L. migratoria (LmigCSP-II) was highly expressed in the sensilla chaetica of the wings and assumed to be involved in contact chemoreception (Zhou et al., 2008). In the western flower thrips, Frankliniella occidentalis, the FoccCSP was mainly expressed in the antennae and leg tissues and reported to be involved in transporting semiochemicals or some hydrophobic molecules from the lymph to chemosensory receptors (Zhang and Lei, 2015). In this study, NlugCSP8 was highly expressed in the male abdomen and very weakly expressed in abdomen of females, which strongly suggests that this CSP is associated with the reproduction events in N. lugens males. Similarly, NlugCSP1 expression in non-olfactory male abdomen also suggested that it might be involved in reproduction process of N. lugens (Yang et al., 2014). Additionally, BPH CSPs such as NlugCSP11 are highly expressed in wings and abdomen. A possible explanation for this finding might be that these proteins are involved in gustatory functions, BPH metamorphosis and determination of oviposition and feeding sites (Zhou et al., 2014). In insects, adult female normally does not automatically oviposit at spawning sites, but first examines the appearance of spawning sites through the tarsal sensilla (Thompson, 1988). The high expression of NlugCSP8 in the wing, leg, and abdomen infers that it might be involved in the attraction activity of BPHs adult toward the potential host, which allows the insect to determine the feeding or oviposition site based on the evaluation of the leaf surface using their abdomens or legs (Higashiura, 1989). Another relationship

was found between the levels of NlugCSP8 mRNA and age or mating status. It is commonly been assumed that the peak mating is on day 3, and peak laying is on day 5 (Thompson, 1988). Based on these findings, the transcripts of NlugCSP8 were tested from peak mating and peak laying stage of adults. The high level of NlugCSP8 expression in antennae and wings of mated brachypterous females on day 3 might reflect the role of NlugCSP8 in mate seeking behavior and it may also have something to do with gustatory functions because insects wings play somewhat gustatory roles (Xu et al., 2009). The NlugCSP8 expression levels in antennae were more closely related to mating status as compared to age. The observed increase of NlugCSP8 expression level from M3D to M5D in the macropterous female antennae and wings provides further support that this gene might be involved in finding oviposition sites, because day 5 belongs to the peak laying in BPH (Thompson, 1988). A positive correlation was also found between mating behavior and CSP expression level in N. lugens (Zhou et al., 2014). In the same way, NlugCSP8 was more highly expressed in mated males and females than in unmated individuals. This high expression after mating may provide evidence that NlugCSP8 plays an important role in the chemoreception of N. lugens. Therefore, we focused on the binding characteristics of NlugCSP8 and their relationship with volatiles.

In order to study the functions of NlugCSP8, a total of 25 compounds, mainly rice plant volatiles (Lou et al., 2005; Yang et al., 2009; Fujii et al., 2010), were selected for the fluorescence binding assay at pH 5.0 and pH 7.4. There is some evidence to suggest that nerolidol is a well-known component of rice plant volatile (Hernandez et al., 1989; Yan et al., 2010). In our study, nerolidol showed high binding affinity with NlugCSP8 with Kivalues of 10.01 and 8.38µM at pH 7.4 and pH 5.0, respectively. The high binding affinity between NlugCSP8 and the plant volatile nerolidol supports the hypothesis that NlugCSP8 may play olfactory roles through binding and transporting the plant volatiles. On the other hand, green leaf volatile hexanal was the most abundant volatile of rice and produced high Electroantennogram response in BPH and some other insects from Hemiptera (Hernandez et al., 1989; Youn, 2002). As expected, NlugCSP8 could bind hexanal, although the Ki was 9.56µM at pH 5.0. Similarly, 2-tridecanone volatile, also isolated from rice plants, was able to attract BPH (Obata et al., 1983). In our experiments, 2-tridecanone also showed relatively high binding affinities to NlugCSP8, which produced Ki-values of 12.51 and 10.43µM at pH 7.4 and pH 5.0, respectively. This outcome is contrary to that of Yang et al. (2014) who found that 2-tridecanone possessed relatively weak binding affinity with NlugCSP7. However, to date, functional research of CSPs protein levels in Delphacidae is rare, except for the previous report of Yang et al. (2014) on CSP7 in N. lugens. In this report, nerolidol and hexanal also exhibited weak affinities to NlugCSP7, while both strongly bound and showed attraction activity for


BPH in case of NlugCSP8 in our study. An interesting finding is that the binding activity of NlugCSP8 also depends upon chain length of ligands. Ligands with long chain exhibited a higher binding affinity as compared with the ligands without chain. Most of the volatiles with relative higher binding ability are compounds with 6–12 carbon atoms. Therefore, carbon chain length appears to affect the binding of NlugCSP8 with ligands. These results match those observed on ligand bindings of SinfCSP19 in earlier studies (Zhang et al., 2014). Nerolidol, with 12 carbon atoms, displayed the highest binding affinity which was in agreement with findings of Zheng et al. (2016) on BhorOBPm2. To support the achievement of these results, molecular modeling and ligand docking were performed. The available 3D structure of NlugCSP8 indicated that it displayed conserved structural features, such as the presence of six α-helices and an internal cavity (Lartigue et al., 2002). The constructed 3D structure of NlugCSP8 is very similar to other previously known insect CSP structures. Like the CSPsg4 of the S. gregaria and the CSPMbraA6 of the Mamestra brassicae, the CSP8 from N. lugens also featured a hydrophobic binding pocket, and the ligand binding differences may be due to some specific amino acids located in the hydrophobic region (Tomaselli et al., 2006). For example, in the CSPsg4, the Trp83 and Ile76 are involved in the binding of oleamide (Tomaselli et al., 2006), while in the CSPMbraA6, the Tyr26 plays an important role in 12 bromo-dodecanol binding (BrC12OH) (Campanacci et al., 2003). Hence, the molecular docking analysis in our study identified several residues, including Lys83, Thr86, Glu87, Ala70, Leu71, Aal74, Cys75, Met90, Lys91, Tyr114, and Tyr118 that may be essential in the binding of volatile compounds by NlugCSP8. These amino acid residues, located in the putative binding pocket of NlugCSP8, may be involved in the recognition and binding of hydrophobic ligands. Pursuing this further, modeling suggested that NlugCSP8 interacts with nerolidol, hexanal, 2 heptanol, and (−)-terpinen-4-ol in order to form H-bonds. Based on these results, we propose that some key residues may be crucial in the interaction of NlugCSP8 with these compounds. Despite these promising results, questions remain on site-directed mutagenesis to assess the function of these residues.

To further support the results of the binding assays, the behavior responses were measured. Four out of five compounds tested elicited a significant behavioral response from N. lugens. The compound with high binding affinity to NlugCSP8 did not elicit significant behavior response, signifying that high binding ability in vitro doesn't mean high behavioral activity in vivo. These behavioral outcomes could be contributed in understanding the sensitivity of insects olfaction related to plant volatiles and may provide strategies for the control of insect pest through identification of semiochemicals responsible for repulsion or attraction of specific insect (Das et al., 2013).

As mentioned earlier, NlugCSP8 probably has different functions related to the finding of oviposition sites, locating suitable mates in addition to olfaction. Thus, RNAi injection experiments against NlugCSP8 were conducted. In the previous study, RNAi technology has been effectively used in BPH, through injection (Liu et al., 2010). Hexanal and nerolidol were identified as strong attractants prior to dsRNA treatment. H-tube olfactometer bioassays of dsRNA-treated BPH revealed that two-choice behavior of BPH was significantly inhibited in hexanal and the attraction activity of nerolidol were lost in insects after silencing NlugCSP8 expression. Based on these findings, we concluded that NlugCSP8 is the pivotal recognition protein for hexanal and nerolidol. Latest studies also recognized that the participation of genes in olfactory functions could be eventually addressed by silencing single genes encoding CSPs or OBPs to influence odor preferences and weaken olfactory performance (Pelletier et al., 2010). However, the ratio of no response BPH also increased in NlugCSP8-dsRNA injected insects as compared to the noninjected control group. These facts support the assumption that NlugCSP8 is involved in behavioral responses, which are the main steps of olfactory reception. Further functional and molecular analysis of other CSPs will provide an exciting opportunity to advance our understanding of olfaction against this monophagous insect and contribute to the development of more efficient and eco-friendly BPH control strategies.

In conclusion, we cloned CSP8 gene from N. lugens. The findings from this study make several contributions to the literature. First, the NlugCSP8 might be involved in finding oviposition sites and locating suitable mates. Second, NlugCSP8 may contribute in binding, transporting, and recognizing plant volatiles. Third, hydrophobic interaction and hydrogen bond play significant roles in the ligand-binding specificity of NlugCSP8 and provide a detailed and reliable olfactory map of chemosensory-protein interaction. Fourth, the reduction in NlugCSP8 transcript abundance leads to a decrease in the behavioral responses to representative attractants. Taken together, these consequences suggest that NlugCSP8 is likely to contribute as a mediator for the responses of N. lugens adults to plant volatile attractants.

## AUTHOR CONTRIBUTIONS

The experimental plan conceived and designed by MW, LH, and M-QW. The experiments performed by MW. The data processed and analyzed by MW, AY, MuQ, SA, AA, LH, HA, and M-QW. Writing and editing manuscript MW, F-FZ, AA, AY, HA, MuQ, and M-QW.

#### ACKNOWLEDGMENTS

This study was supported and funded by the Special Fund for Agro-scientific Research in the Public Interest of China (201403030), the National High Technology Research and Development Program of China (863 Program) (2014AA10A605), and the Technical Innovation Programs of Hubei Province (2017ABA146).

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00379/full#supplementary-material

Table S1 | The primers used in the qRT-PCR, dsRNA synthesis, and protein expression.

Figure S1 | Nucleotide and deduced amino acid sequence analysis of NlugCSP8. The predicted putative signal peptides are underlined and denoted by red color. The four conserved cysteine residues are showed in red boxes. The stop codon is indicated with an asterisk.

Figure S2 | Molecular phylogenetic analysis of amino acid sequences by neighbor-joining (NJ) method. The tree was constructed using the neighbor-joining method with bootstrap support values (%) based on 1,000 replicates. NlugCSP8 are marked with a solid red circle and all other CSPs from Hemipteran are marked with solid Yellow circles. CSPs from Coleoptera, Lepidoptera, Diptera, Hymenoptera, and Neuroptera are marked with Navy, Lime, Aqua, Blue and Silver circles, respectively. All sequences are available from the NCBI database. Species abbreviations are included for taxon identifications. Dpon (*Dendroctonus ponderosae*), Tcas (*Tribolium castaneum*), Bhor (*Batocera horsfieldi*), Cbow (*Colaphellus bowringi*), Tmol (*Tenebrio molitor*), Malt (*Monochamus alternatus*), Paen (*Pyrrhalta aenescens*), Pmac (*Pyrrhalta maculicollis*), Dhel (*Dastarcus helophoroides*), Agos (*Aphis gossypii*), Mper (*Myzus persicae*), Rdom (*Rhyzopertha dominica*), Cqui (*Culex quinquefasciatus*), Csty (*Calliphora stygia*), Dant *(Delia antiqua*), Apis (*Apis mellifera*), Sinf (*Sesamia inferens*), Harm (*Helicoverpa armigera*), Sexi (*Spodoptera exigua*), Bmor (*Bombyx mori*), Cmed (*Cnaphalocrocis medinalis*), Ofur (*Ostrinia furnacalis*), Cfum (*Choristoneura fumiferana*), Nlug (*Nilaparvata lugens*), Psol (*Phenacoccus solenopsis*), Btab (*Bemisia tabaci*), Adis (*Athetis dissimilis*), Slit (*Spodoptera litura*), Cpal (*Chrysopa pallens*), Lory (*Lissorhoptrus oryzophilus*), Bdor (*Bactrocera dorsalis*), Sfur (*Sogatella furcifera*), Lstr (*Laodelphax striatella*), Alin (*Adelphocoris lineolatus*), Asut (*Adelphocoris suturalis*), Aluc (*Apolygus lucorum*), Eher (*Euschistus heros*), Lhes (*Lygus hesperus*), Tbra (*Triatoma brasiliensis*), Acor (*Anomala corpulenta*), Dcor (*Drosicha corpulenta*), and Hpar (*Holotrichia parallela*).

Figure S3 | SDS-PAGE analyses showing the expression and cleavage of recombinant NlugCSP8. Lane M: Molecular marker, Lane 1: Non-induced BL21 (DE3) bacteria with pET-30a, Lane 2–4: different IPTG concentrations used to induce recombinant protein (2, 4, 6 mM from lanes 2 to 4), Lane 5: Eluted protein before cleavage, Lane 6: cleaved protein by the recombinant enterokinase.

Figure S4 | The Ramachandran map for the model of NlugCSP8.

REFERENCES

doi: 10.1046/j.1432-1327.1999.00438.x

Ban, L., Scaloni, A., D'Ambrosio, C., Zhang, L., Yan, Y., and Pelosi, P. (2003). Biochemical characterization and bacterial expression of an odorantbinding protein from Locusta migratoria. Cell Mol. Life Sci. 60, 390–400. doi: 10.1007/s000180300032

Angeli, S., Ceron, F., Scaloni, A., Monti, M., Monteforti, G., Minnocci, A., et al. (1999). Purification, structural characterization, cloning and immunocytochemical localization of chemoreception proteins from Schistocerca gregaria. Eur. J. Biochem. 262, 745–754.


expression pattern of putative odorant binding proteins. J. Insect Sci. 14, 1–15. doi: 10.1093/jis/14.1.57


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Waris, Younas, ul Qamar, Hao, Ameen, Ali, Abdelnabby, Zeng and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identification and Expression Patterns of Putative Diversified Carboxylesterases in the Tea Geometrid Ectropis obliqua Prout

Liang Sun1, 2 \* † , Qian Wang2, 3†, Qi Wang<sup>2</sup> , Yuxing Zhang1, 4, Meijun Tang<sup>1</sup> , Huawei Guo<sup>1</sup> , Jianyu Fu<sup>1</sup> , Qiang Xiao<sup>1</sup> , Yanan Zhang<sup>5</sup> \* and Yongjun Zhang<sup>2</sup> \*

<sup>1</sup> Key Laboratory of Tea Quality and Safety Control, Ministry of Agriculture, Tea Research Institute, Chinese Academy of Agricultural Sciences, Hangzhou, China, <sup>2</sup> State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China, <sup>3</sup> College of Horticulture and Plant Protection, Yangzhou University, Yangzhou, China, <sup>4</sup> College of Plant Protection, Anhui Agricultural University, Hefei, China, <sup>5</sup> College of Life Sciences, Huaibei Normal University, Huaibei, China

#### Edited by:

Peng He, Guizhou University, China

#### Reviewed by:

Pei Liang, China Agricultural University, China Thomas Chertemps, Université Pierre et Marie Curie, France

#### \*Correspondence:

Liang Sun liangsun@tricaas.com Yanan Zhang ynzhang\_insect@163.com Yongjun Zhang yjzhang@ippcaas.cn

† These authors have contributed equally to this work and co-first authors.

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 18 August 2017 Accepted: 08 December 2017 Published: 18 December 2017

#### Citation:

Sun L, Wang Q, Wang Q, Zhang Y, Tang M, Guo H, Fu J, Xiao Q, Zhang Y and Zhang Y (2017) Identification and Expression Patterns of Putative Diversified Carboxylesterases in the Tea Geometrid Ectropis obliqua Prout. Front. Physiol. 8:1085. doi: 10.3389/fphys.2017.01085 Carboxylesterases (CXEs) belong to a family of metabolic enzymes. Some CXEs act as odorant-degrading enzymes (ODEs), which are reportedly highly expressed in insect olfactory organs and participate in the rapid deactivation of ester pheromone components and plant volatiles. The tea geometrid Ectropis obliqua Prout produces sex pheromones consisting of non-ester functional compounds but relies heavily on acetic ester plant volatiles to search for host plants and locate oviposition sites. However, studies characterizing putative candidate ODEs in this important tea plant pest are still relatively scarce. In the present study, we identified 35 candidate EoblCXE genes from E. obliqua chemosensory organs based on previously obtained transcriptomic data. The deduced amino acid sequences possessed the typical characteristics of the insect CXE family, including oxyanion hole residues, the Ser-Glu-His catalytic triad, and the Ser active included in the conserved pentapeptide characteristic of esterases, Gly-X-Ser-X-Gly. Phylogenetic analyses revealed that the EoblCXEs were diverse, belonging to several different insect esterase clades. Tissue- and sex-related expression patterns were studied via reverse-transcription and quantitative real-time polymerase chain reaction analyses (RT- and qRT-PCR). The results showed that 35 EoblCXE genes presented a diversified expression profile; among these, 12 EoblCXEs appeared to be antenna-biased, two EoblCXEs were non-chemosensory organ-biased, 12 EoblCXEs were ubiquitous, and nine EoblCXEs showed heterogeneous expression levels among different tissues. Intriguingly, two EoblCXE genes, EoblCXE7 and EoblCXE13, were not only strongly localized to antennal sensilla tuned to odorants, such as the sensilla trichodea (Str I and II) and sensilla basiconica (Sba), but were also expressed in the putative gustatory sensilla styloconica (Sst), indicating that these two CXEs might play multiple physiological roles in the E. obliqua chemosensory processing system. This study provides the first elucidation of CXEs in the chemosensory system of a geometrid moth species and will enable a more comprehensive understanding of the functions of insect CXEs across lepidopteran species.

Keywords: Ectropis obliqua, carboxylesterases (CXEs), odorant-degrading enzymes, phylogenetic analyses, expression patterns, fluorescence in situ hybridization

## INTRODUCTION

The sophisticated olfactory system, particularly the peripheral chemical signal coding, is essential for insects to find mates, locate food, and avoid predators (Dweck et al., 2013; Tauxe et al., 2013; Strauch et al., 2014; Li and Liberles, 2015). Biologically important odorants are generally perceived sensitively and specifically in the multiporous sensilla hairs on insect antennae (Meijerink and van Loon, 1999; Pophof et al., 2005; Park et al., 2013; Sun L. et al., 2014). It is well established that at least three major classes of molecules are involved in this process: odorantbinding proteins (OBPs), odorant receptors (ORs), and odorantdegrading enzymes (ODEs). In brief, airborne odorants enter the hydrosoluble sensillum lymph through the sensilla pores, bind to OBPs, activate ORs and trigger signal transduction cascades and olfactory coding; odorants are then rapidly removed from the vicinity of the ORs by ODEs to restore the sensitivity of the sensory neuron (Rützler and Zwiebel, 2005; Vogt, 2005; Pelosi et al., 2006; Leal, 2013).

The highly sensitive odorant signal transduction pathway of insects represents an excellent model that researchers can use to develop new environmentally friendly pest-management strategies through targeting key molecules and screening biologically active compounds for behavioral control. Previous functional reports regarding OBPs and ORs indeed led to the rapid discovery of high-efficiency pest repellents and attractants. For example, compounds that are behaviorally active in the mirid bug Adelphocoris lineolatus were successfully screened via studies on the interaction between antenna-enriched AlinOBP10 and its putative ligands (Sun et al., 2013). In the aphid alarm pheromone EBF perception pathway, ApisOBP3 and ApisOBP7 as well as ApisOR5 were proven to be potentially crucial targets for aphid repellent screening (Sun et al., 2012; Zhang R. et al., 2017). However, compared with OBPs and ORs, similar reports on ODEs appear to be rare. Given that the rapid degradation of redundant odorants can rescue the sensitivity of odorant sensory neurons, putative genes encoding insect ODEs that are highly expressed in the chemosensory system should be identified, and their potential roles in odorant degradation deserve thorough exploration.

Carboxylesterases (CXEs) belong to the α/β-fold hydrolase superfamily and are widely distributed in insects and other organisms. CXEs commonly include a conserved catalytic triad (Ser-His-Glu) and specifically catalyze the hydrolysis of ester bonds in various substrates (Oakeshott et al., 1999, 2005). Because most insect species, including hemipteran bugs and lepidopteran moths, utilize aliphatic esters as intraspecific sex pheromones and ovipositional stimulants (Ando et al., 2004; Millar, 2005; Pan et al., 2015), many antennae-biased CXEs have been identified, and their activities associated with sex pheromone and odorant degradation have been assessed (Vogt, 2005; Jacquin-Joly and Maïbèche-Coisne, 2009). The first CXE subfamily of ODEs, known as Apol-SE (Vogt and Riddiford, 1981), or ApolPDE (Ishida and Leal, 2005), was isolated from the giant silk moth, Antheraea polyphemus. Subsequently, genes encoding putative antennal esterases were cloned and described across insect species using a polymerase chain reaction (PCR) strategy. These genes included two other CXEs, ApolODE and Apol-IE, in A. polyphemus (Ishida and Leal, 2002); Mbra-EST from the cabbage armyworm, Mamestra brassicae (Maïbèche-Coisne et al., 2004); D-AP1, a honeybee homolog of CXE in Apis mellifera L. (Kamikouchi et al., 2004); Slit-EST and Snon-EST from the Egyptian armyworm, Spodoptera littoralis, and the Mediterranean corn borer, Sesamia nonagrioides (Merlin et al., 2007); and PjapPDE, cloned from the Japanese beetle, Popillia japonica (Ishida and Leal, 2008). Furthermore, through expressed sequence tag (EST) and RNA-Seq analyses, diverse CXE genes have been identified from various insect species, such as Epiphyas postvittana (Jordan et al., 2008), Spodoptera littoralis (Durand et al., 2010b), Agrotis ipsilon (Gu et al., 2013), Sesamia inferens (Zhang Y. N. et al., 2014), Chilo suppressalis (Liu et al., 2015; Xia et al., 2015), and Spodoptera litura (Zhang et al., 2016).

Convincing evidence obtained through biochemical characterization and enzyme kinetic activity analyses showed that Apol-SE/ApolPDE displays expression specific to male antennal sensilla and exhibits rapid catalytic activity toward the acetate sex pheromone component E6Z11-16:OAc (Vogt et al., 1985; Prestwich et al., 1986; Klein, 1987; Ishida and Leal, 2005). In vitro functional analyses and potential hydrolyzed substrates of CXEs have also been documented in other insect species, particularly lepidopteran moths, whose main sex pheromone components are acetate esters (Ishida and Leal, 2008; Durand et al., 2010a, 2011; He et al., 2014a,b,c, 2015). Additionally, an extracellular carboxylesterase esterase-6 (EST-6) of Drosophila melanogaster has been demonstrated to be a potential ODE for both the sex pheromone ester cis-vaccenyl acetate (CVA) and other bioactive volatile esters, such as pentyl acetate (Chertemps et al., 2012, 2015). All of the available data support potential roles of CXEs in degrading either sex pheromones or host plant odorants containing ester functional groups.

The tea geometrid Ectropis obliqua Prout is a common pest of the tea plant, Camellia sinensis (L.), and causes serious economic damage to tea production (Ye et al., 2014; Zhang G. H. et al., 2014). Multiple electrophysiological and behavioral studies show that E. obliqua relies heavily on chemical cues to locate host plants, oviposition sites and conspecific mates. Furthermore, larval infection of tea plants strongly induces the release of several kinds of host volatiles with ester functional groups, and these ester compounds can in turn regulate the ovipositional preference of E. obliqua adult females (Sun X. L. et al., 2014). Hence, studies on the molecular mechanism of ester odorant degradation are particularly important for the identification of potential target genes mediating oviposition signal inactivation and the development of ODE-based strategies in geometrid pest management.

In this study, we identified putative genes encoding CXEs by analyzing the BLASTX annotations of transcriptomic data. The phylogenetic relationships between the candidate CXEs and homologs in other Lepidoptera species were further analyzed. Finally, the tissue expression patterns of the identified CXEs were investigated in olfactory organs (particularly in the different antennal sensilla) and non-olfactory organs, and potential functional differentiation was discussed.

## MATERIALS AND METHODS

#### Insect Rearing and Tissue Collection

The tea geometrid E. obliqua was collected from the Yuhang tea plantation in Zhejiang Province, China. Phylogenetic identity analysis and laboratory colony construction were performed according to Zhang G. H. et al. (2014). The pupae were sexed, and male and female individuals were raised separately until eclosion. Adult moths of different sexes were maintained in different cages and fed a 10% honey solution on water-soaked cotton.

For the tissue-specific expression profile analysis of E. obliqua adults, approximately 500 antennae, three abdomens, and 300 legs of both male and female adults 1–3 days after emergence were dissected and collected. Two biological replicates were prepared for RT-PCR, and two additional biological replicates were prepared for qRT-PCR. All of the specimens were immediately stored at −80◦C until use.

#### RNA Extraction and cDNA Synthesis

Total RNA from each specimen was extracted with the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. The integrity of the total RNA was examined through 1.2% agarose electrophoresis, and the purity was assessed using a NanoDropTM instrument (Wilmington, DE, USA). First-strand cDNA was synthesized from 2µg of RNA using a FastQuant RT kit with gDNA Eraser (TianGen, Beijing, China) according to the manufacturer's instructions.

## Identification of Candidate EoblCXEs and Sequence Analysis

Candidate EoblCXEs were identified through keyword screening of the BLASTX annotations of transcriptomic data from adult E. obliqua chemosensory organs, including the antennae, legs, wings and proboscises. The TBLASTN program was also applied using the previously identified S. littoralis CXEs (Durand et al., 2010b) as the query. The open reading frames (ORFs) of genes were predicted using ORF finder (http://www.ncbi.nlm.nih.gov/ gorf/gorf.html). The theoretical isoelectric points and molecular weights of the deduced proteins were calculated using the ExPASy tool (http://web.expasy.org/compute\_pi/). Homology searches were performed with BLAST (http://blast.ncbi.nlm.nih. gov/). Catalytic residues were predicted by searching the NCBI Conserved Domain Database (http://www.ncbi.nlm.nih.gov/ structure/cdd/cdd.shtml). Putative N-terminal signal peptides were predicted using the SignalP 4.0 program (http://www.cbs. dtu.dk/services/SignalP/) (Brunak et al., 2010).

## Phylogenetic Analysis

The amino acid sequences of EoblCXEs and CXEs from other species were aligned using ClustalX 2.0 (Larkin et al., 2007). A neighbor-joining tree was constructed using the program MEGA 6.0 with the Jones–Taylor–Thornton (JTT) amino acid substitution model (Tamura et al., 2013). Node support was assessed using a bootstrapping procedure with 1,000 replicates, uniform rates, and pairwise deletion of data gaps. The protein names and accession numbers corresponding to the genes used for construction of the phylogenetic tree are listed in **Table S1**.

#### Reverse-Transcription PCR

The tissue-specific expression of EoblCXEs was determined via reverse-transcription PCR (RT-PCR) using ExTaq DNA polymerase (TaKaRa, Dalian, China). The E. obliqua glyceraldehyde-3-phosphate dehydrogenase (EoblGAPDH, GenBank accession no. KT991373) reference gene was employed as an internal control to normalize target gene expression in order to correct for sample-to-sample variation. The specific primers used for amplification are listed in **Table S2**.

The experiment was performed according to a previous report (Sun et al., 2017b): each reaction of 50µL contained 1µL of 200 ng/µL (200 ng) single-stranded cDNA, 5µL of 10× ExTaq buffer, 4µL of deoxyribonucleoside triphosphates (dNTPs), 2µL of each primer and 0.25 U of ExTaq DNA polymerase. The PCR conditions were as follows: initial denaturation at 94◦C for 4 min followed by 40 cycles of 94◦C for 30 s, 55–65◦C for 30 s, and 72◦C for 30 s and a final elongation step at 72◦C for 10 min. After PCR, the products were analyzed in 1.5% agarose gels. To check reproducibility, each RT-PCR run for each sample was performed with two biological replicates and three technical replicates. The relative expression levels of the EoblCXE genes in different tissues were calculated using the ratio of RT-PCR band intensity between the target gene and the internal reference gene, EoblGAPDH, using Bio-Rad Quantity One 4.6.2 software (Zhang et al., 2013).

## Quantitative Real-Time PCR

Based on the RT-PCR results, 18 EoblCXEs were randomly selected to conduct quantitative real-time PCR (qRT-PCR). The experiment was performed using an ABI 7500 Real-Time PCR System (Applied Biosystems, Carlsbad, CA, USA), and each reaction was conducted in a 20-µL reaction mixture containing 10µL of 2× SYBR Green PCR Master Mix (TaKaRa, Dalian, Liaoning, China), 0.8µL of each primer (10µM), 0.4µL of ROX Reference Dye II, 2µL of sample cDNA (200 ng), and 6.0µL of sterilized H2O. The qPCR cycling parameters were as follows: 95◦C for 30 s followed by 40 cycles of 95◦C for 5 s and 60◦C for 31 s. Subsequently, the fluorescence was measured using a 55– 95◦C melting curve to detect a single gene-specific peak and to confirm the absence of primer dimer peaks; single, discrete peaks were detected for all primers tested.

The primers employed for qPCR (**Table S3**) were designed using the Beacon Designer 7.90 program (PREMIER Biosoft International). The reference gene EoblGAPDH was found to be expressed at a similar level in different tissues and was used as an internal control to normalize target gene expression in order to correct for sample-to-sample variation (Sun et al., 2017a). The amplification efficiency for the target and reference genes was assessed using gradient dilution templates to examine the variation of 1C<sup>T</sup> (CT, Target gene − CT, reference gene) with template dilution (Livak and Schmittgen, 2001). The absolute values of the slopes of all lines obtained from template dilution plots (log cDNA dilution vs. 1CT) were close to zero, indicating that the efficiency for EoblCXEs was similar to that for EoblGAPDH. Nontemplate reactions (replacing cDNA with sterilized H2O) were performed as negative controls. To check the reproducibility of the qPCR assays, each reaction for each sample was performed with three technical replicates and two biological replicates.

Comparative analyses of target gene expression between different tissues were performed using one-way nested analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) test. The relative mRNA expression levels between males and females of 10 antennae-biased EoblCXEs were compared with Student's t-test. All analyses were performed using SPSS Statistics 18.0 software (SPSS Inc., Chicago, IL, USA).

#### Fluorescence in Situ Hybridization

Based on the observed tissue expression patterns and the results of phylogenetic analyses, two EoblCXE genes, EoblCXE7, and EoblCXE13, were selected for fluorescence in situ hybridization assays. Biotin-labeled antisense or sense RNA probes were transcribed from the linearized recombinant pGEM-T vector using a biotin RNA Labeling Mix (SP6/T7) (Roche, Mannheim, Germany) following the recommended protocols. RNA probes were subsequently fragmented to an average length of approximately 400 bp via incubation in carbonate buffer (80 mM NaHCO3, 120 mM Na2CO3, pH 10.2).

The experiment was performed following a reported protocol (Wang et al., 2017). The antennae of both male and female 1–3-day-old moths were dissected, embedded with Tissue-Tek optimal cutting temperature (O.C.T.) compound (Sakura Finetek, Torrance, CA, USA) and rapidly frozen at −60◦C. Sections (12µm) were prepared using a Cryostar NX50 cryostat (Thermo Scientific, San Jose, CA, USA) at −20◦C, thaw-mounted on SuperFrost Plus microscope slides (Fisher

TABLE 1 | BLASTX hits for candidate CXEs identified in the chemosensory organs of E. obliqua adults.


Scientific, Pittsburgh, PA, USA), and air-dried at room temperature for 15 min. After a series of fixing and washing procedures, the tissue sections were covered with 100µL of hybridization solution containing biotin-labeled antisense RNA probes and incubated at 60◦C for at least 16 h. After hybridization, the slides were washed twice for 20 min in 0.2 × saline-sodium citrate (SSC) at 60◦C and treated with 1% blocking reagent (Roche, Basel, Switzerland) in TBST for 30 min at room temperature. Biotin-labeled probes were detected via incubation with streptavidin-horseradish peroxidase (HRP) (Perkin Elmer, Boston, MA, USA) diluted 1:100 in TBS with 0.03% Triton X-100 and 1% blocking reagent at 37◦C for 1 h. After three 5 min washes in TBS with 0.05% Tween 20 (Sigma, Louis, MO, USA), the biotin-labeled probes were detected with the TSA Plus Fluorescein System (Perkin Elmer). Images were captured via laser scanning microscopy (LSM) using a Zeiss LSM880 confocal microscope (Zeiss, Oberkochen, Germany). Photoshop CS5 (Adobe Systems, San Jose, CA, USA) was employed to adjust the brightness or contrast of the figures.

#### RESULTS

#### Identification and Sequence Characteristics of Candidate EoblCXEs

Thirty-five candidate EoblCXEs were identified in the chemosensory organs of E. obliqua. Of these 35 candidates, 28 EoblCXEs possessed full-length open reading frames (ORFs), and seven lacked either the 5′ or 3′ region (**Table 1**). The candidate sequences were designated EoblCXE1-35 according to their presumptive orthologs in other insects, particularly S. littoralis, S. exigua and S. inferens, and were deposited in the GenBank database under sequential accession numbers from KX015843 to KX015877 (**Table 1**).

The amino acid identity among the 35 candidate EoblCXEs ranged from 13 to 98% (**Table S4**). The 28 full-length EoblCXEs exhibited an average coding region length of 1650 bp and encoded 519 to 706 amino acids. Their predicted theoretical isoelectric points ranged from 4.95 to 8.80, and their calculated molecular masses ranged from 58.10 to 74.89. Putative Nterminal signal peptide prediction showed that 16 of the 28 sequences displayed typical sequence cleavage sites. Multiple sequence alignments revealed that all 28 full-length EoblCXEs displayed a conserved sequence motif including the oxyanion hole residues, the catalytic triad (Ser-Glu-His), and the Ser active site in the conserved pentapeptide Gly-X-Ser-X-Gly, characteristic of esterases (**Table 2**).

#### Phylogenetic Analyses

To define the putative functions of the candidate EoblCXEs, phylogenetic analyses were performed (**Figure 1**). The results revealed that the insect esterases could be divided into 10 major clades: mitochondrial and cytosolic esterases, dipteran microsomal α-esterases, cuticular and antennal esterases, ß-esterases and pheromone esterases, lepidopteran juvenile hormone esterases (JHE), non-lepidopteran JHE, moth antennal



esterases, neuroligins, neuroreceptors, and gliotactins (Oakeshott et al., 2005; Durand et al., 2010b).

EoblCXEs were generally distributed in eight different clades: EoblCXE13 and 33, along with the pheromone-degrading enzymes Apol-PDE and Pjap-PDE, clustered with the ßesterase and pheromone esterase group; EoblCXE5 and 16 were distributed within a clade of cuticular and antennal esterases; and EoblCXE19, EoblCXE32, two EoblCXEs (EoblCXE15 and 23), four EoblCXEs (EoblCXE9, 11, 34, and 35), and 10 EoblCXEs (EoblCXE3, 10, 24–31) were assigned to neuroreceptors, neuroligins, lepidopteran JHE, dipteran microsomal α-esterases and mitochondrial and cytosolic esterases, respectively. The moth antennal esterases exhibited the greatest abundance of EoblCXEs (13 EoblCXEs, including EoblCXE1, 2, 4, 6–8, 12, 14, 17, 18, and 20–22), whereas no EoblCXEs clustered into the non-lepidopteran JHE or gliotactin clades (**Figure 1**).

## Tissue- and Sex-Related Expression Patterns of Candidate EoblCXE Genes

To clarify whether candidate EoblCXEs could function in chemosensory organs with physiological roles in odorant degradation, the tissue- and sex-related expression profiles of the

35 EoblCXE genes were determined via RT-PCR. As shown in **Figure 2**, the EoblCXE genes displayed four general expression patterns: 12 EoblCXE genes (EoblCXE2, 5, 7, 10, 13, 15–17, 19, 20, 22, and 24) were strongly expressed in olfactory organ antennae; two EoblCXEs (EoblCXE14 and 23) were non-chemosensory organ biased; 12 EoblCXEs (EoblCXE1, 3, 4, 6, 9, 11, 12, 18, 27, 29, 30, and 34) were ubiquitous, and their expression levels were comparable among the tested tissues; and nine EoblCXEs (EoblCXE8, 21, 26, 35, 25, 28, and 31–33) exhibited heterogeneous expression profiles.

To confirm the RT-PCR results, 18 candidate EoblCXE genes randomly selected from all expression patterns were quantified through qRT-PCR assays. The qRT-PCR results are shown in **Figure 3**. Similar to the RT-PCR results, EoblCXE2, 5, 7, 10, 13, 15, 20, 22, and 24 were strongly expressed in moth antennae, whereas EoblCXE14 and 23 were primarily expressed in the abdomen (a non-chemosensory organ), and EoblCXE6 was ubiquitously expressed within different tissues. However, the qRT-PCR and RT-PCR results were somewhat contradictory; the qRT-PCR results showed that EoblCXE3 was highly expressed in the abdomens of both sexes, whereas EoblCXE12 and 26 were highly expressed in the antennae and legs, respectively.

To dissect the putatively different roles of EoblCXEs in olfaction between male and female moths, sex-biased expression profiles were determined for the antennae-enriched EoblCXE genes. Eight of the 10 tested EoblCXE genes (EoblCXE2, 5, 7, 10, 12, 13, 15, 20, 22, and 24) exhibited comparable expression levels between the sexes. In contrast, EoblCXE5 and 10 were highly expressed in female and male antennae, respectively.

## Cellular Localization of EoblCXE7 and EoblCXE13 within Different Antennal Sensilla

EoblCXE7 and EoblCXE13 were strongly labeled on the sensilla sides of both male and female antennae but differed from each other in their cellular localization in the different types of sensilla found in each sex (**Figures 4**, **5**). EoblCXE7 exhibited a


FIGURE 2 | Tissue-related expression patterns of EoblCXE genes, as revealed via RT-PCR. The EoblGAPDH gene was used as a positive control, and NC (no cDNA template) was used as a negative control. MA, male antennae; FA, female antennae; ML, male legs; FL, female legs; MAB, male abdomen; FAB, female abdomen. The original image of this is shown in Figure S2.

similar localization in the antennae of male and female moths (**Figures 4A–F**). The antisense EoblCXE7 probe clearly labeled the base of the sensilla trichodea (Str I for male and Str II for female moths), sensilla basiconica (Sba) and sensilla styloconica (Sst) (**Figures 4A–C**). The scale sides of the antennae were not labeled in either sex. Compared with EoblCXE7, EoblCXE13 showed different expression patterns between male and female antennal sensilla (**Figures 5A–I**). The antisense EoblCXE13 probe was restricted to the base of Str I rather than Sba or Sst in male moths (**Figures 5A–D**). In contrast, strong labeling was detected not only in Str II but also in Sba and Sst in female moths (**Figures 5E–I**). The sense probes of EoblCXE7 and EoblCXE13 produced no positive signals (**Figure S1**).

#### DISCUSSION

In the present study, we identified and characterized 35 candidate genes encoding EoblCXEs from the chemosensory organs of the moth E. obliqua through transcriptomic analysis, including 28 full-length sequences. Phylogenetic analyses and tissue- and sex-related expression profiling showed that EoblCXEs exhibited diverse sequence structures, multiple subfamily clades and distinct expression patterns, suggesting potential differentiation of physiological functions among EoblCXEs. As expected, we demonstrated that 12 EoblCXE genes were highly expressed in E. obliqua antennae, particularly EoblCXE7 and EoblCXE13, which were strongly localized to the olfactory sensilla of both sexes. These results were consistent with the previously reported proposition that CXEs function specifically in insect olfaction and are involved in olfactory signal termination and maintenance of the sensitivity of the olfactory sensilla (Vogt and Riddiford, 1981; Vogt et al., 1985; Ishida and Leal, 2005; Vogt, 2005; Durand et al., 2010a,b, 2011; Chertemps et al., 2012, 2015).

Gene sequence identification represents the first step in elucidating the potential physiological functions of insect CXEs. Due to the lack of genomic information for E. obliqua, we identified putative CXEs through a transcriptomic approach. The number of putative CXE genes (35) identified in E. obliqua (**Table 1**) was comparable to those found in insect species with available genome data, including D. melanogaster (35 genes), A. gambiae (51 genes), A. aegypti (49 genes) and A. mellifera (24 genes). However, this number was significantly greater than the number of CXE genes identified in moth species that utilize ester compounds as intraspecific sex pheromones, including 20 from S. littoralis (Merlin et al., 2007; Durand et al., 2010b), 24 from S. litura (Zhang et al., 2016), 20 from S. inferens (Zhang Y. N. et al., 2014), 19 from C. suppressalis (Liu et al., 2015), 17 from A. ipsilon (Gu et al., 2013), and 30 from Cnaphalocrocis medinalis (Zhang Y. X. et al., 2017).

Insect CXEs expressed in the olfactory system are mainly related to ester odorant degradation, particularly that of lepidopteran moth sex pheromones (Vogt and Riddiford, 1981; Vogt et al., 1985; He et al., 2015); however, E. obliqua females produce sex pheromones containing unsaturated hydrocarbons and enantiomers of epoxy hydrocarbons rather than acetate

FIGURE 3 | Relative mRNA levels of EoblCXE genes in different tissues of male and female E. obliqua adults, as revealed via qRT-PCR. The reference gene EoblGAPDH was found to be expressed at a similar level in different tissues and was employed as an internal control to normalize target gene expression in order to correct for sample-to-sample variation (Sun et al., 2017a). The amplification efficiency for the target and reference genes were assessed using gradient dilution templates to examine the variation of 1CT (CT, Target gene − CT, reference gene) with template dilution (Livak and Schmittgen, 2001). The absolute values of the slopes of all lines from the template dilution plots (log cDNA dilution vs. 1CT) were close to zero, indicating that the efficiency for EoblCXEs was similar to that for EoblGAPDH. The fold changes are relative to the transcript levels in the male abdomen. The error bars represent the standard error, and different letters (a, b and c for male; α, ß, and χ for females) above each bar denote significant differences (P < 0.05). The t and p-values in Student's t-test are shown in Table S5 and the asterisk indicates significantly different relative expression levels between male and female antennae.

(D–F) female antennae; Str I, sensilla trichodea I; Str II, sensilla trichodea II; Sba, sensilla basiconica; Sst, sensilla styloconica.

esters (Yang et al., 2016a). Thus, it is particularly interesting that the number of CXE genes from E. obliqua was significantly higher than that in species that utilize ester compounds as intraspecific sex pheromones. We speculate that this situation might be attributed to either the use of a different sequencing strategy or the fact that E. obliqua adults depend on the detection of multiple odorants with ester functional groups to find host plants and egg-laying sites. Our use of a high-throughput RNA-sequencing approach in the chemosensory organs (adult antennae of both sexes, legs, wings and proboscises) enabled us to identify as many CXE genes from E. obliqua chemosensory organs as possible. The tea plant, which is the most preferred host of E. obliqua, releases large quantities of ester compounds, particularly under attack by E. obliqua caterpillars; these ester compounds, such as (Z)-3-hexenyl hexanoate and (Z)-3-hexenyl acetate, can in turn regulate the host searching and ovipositional preferences of E. obliqua female adults (Wang, 2010; Sun X. L. et al., 2014). Hence, we propose that some EoblCXEs might act as candidate ODEs and potentially exhibit crucial physiological functions in the degradation of tea plant volatiles with ester functional groups rather than the degradation of sex pheromone components produced by E. obliqua females. This inference corresponds well to the tissue- and sex-related expression patterns found in this study, in which only EoblCXE10 of the antennae-enriched EoblCXE genes displayed a male-specific expression pattern; the other EoblCXEs were either highly abundant in female antennae or were not sex-biased (**Figure 3**); however, even if then, this inference remains to be supported by the in vitro biochemical enzymatic experiment.

It should be noted that insect CXE genes belong to a multigene family that encodes sequence-divergent and functionally diverse proteins (Oakeshott et al., 1999; Tsubota and Shiotsuki, 2010). The 35 candidate EoblCXEs, which show average amino acid identities lower than 30%, fall into at least 10 different subclades; seven of these clades, which contain 33 EoblCXEs, possess clearly conserved functional characteristic features of the α/βhydrolase structure, such as the Ser-Glu-His catalytic triad and the nucleophilic elbow surrounding the active-site serine residue

(E–I) female antennae; Str I, sensilla trichodea I; Str II, sensilla trichodea II; Sba, sensilla basiconica; Sst, sensilla styloconica.

(Gly-X-Ser-X-Gly) (**Figure 1**, **Table 2**). These features suggest that most EoblCXEs are catalytically active and participate in the degradation of diverse biologically important compounds. Other CXEs fall into the neuroligin, neuroreceptor, or gliotactin clades, including EoblCXE19 in the neuroreceptor clade and EoblCXE32 among the neuroligins, and lack the crucial residue Ser responsible for catalytic activity (**Table 2**); thus, these CXEs are considered to be catalytically inactive and are mainly involved in neurological and developmental functions related to sensory processing (Biswas et al., 2008; Durand et al., 2017).

Ten EoblCXEs belonging to the dipteran microsomal αesterase or mitochondrial and cytosolic esterase clade lack a predicted signal peptide, indicating that they are intracellular esterases. These clades (particularly the α-esterases) are wellknown for their involvement in the detoxification of insecticides and xenobiotics and the digestion of dietary esters (Newcomb et al., 1997; Campbell et al., 2003; Liang et al., 2007; Tang et al., 2011; Yang et al., 2016b; Gong et al., 2017). An orthologous gene of CXE10 has been functionally studied in two closely related Spodoptera species, S. littoralis (SlittCXE10) and S. exigua (SexiCXE10) (Durand et al., 2010a; He et al., 2015). Both SlittCXE10 and SexiCXE10 are reportedly expressed in the olfactory sensilla and preferentially degrade an ester plant volatile, (Z)-3-hexenyl acetate. EoblCXE10 shares high amino acid identity with SlittCXE10 and SexiCXE10 and is therefore considered an orthologous gene in E. obliqua. Its tissue expression was found to be restricted to the antennae of adults, particularly adult males, and resembled the distribution of its orthologs in Spodoptera species (**Figure 3**), implying a similar role in the degradation of host plant volatile compounds.

Approximately half of the candidate EoblCXEs clustered in the moth antennal esterase or ß-esterase and pheromone esterase clade and presented both a catalytically active Ser residue and a predicted signal peptide (**Table 2**). These CXEs represented typical secreted or extracellular esterases that could be secreted into the sensillum lymph surrounding the sensory neurons or into the hemolymph filling the antennal lumen, indicating potential roles in the degradation of odorants and the maintenance of OSN sensitivity. Indeed, functional reports regarding several members of these clades, such as ApolPDE (Vogt and Riddiford, 1981; Ishida and Leal, 2005), PjapPDE (Ishida and Leal, 2008), SexiCXE13 (He et al., 2014a), SexiCXE14 (He et al., 2014c), and SlittCXE7 (Durand et al., 2011), indicate that they play crucial roles in the degradation of insect sex pheromones and biologically important plant volatiles.

To gain further insight into the physiological roles of the EoblCXEs involved in hydrolyzing tea plant volatiles in E. obliqua olfaction, we selected EoblCXE7 and EoblCXE13 for a fluorescence in situ hybridization assay, not only because EoblCXE7 and EoblCXE13 are comparably expressed in the antennae of both E. obliqua sexes but also because EoblCXE13 clusters with ApolPDE, PjapPDE and SexiCXE13, while EoblCXE7 falls into the same clade as SlittCXE7 and SexiCXE14; therefore, these genes provide a useful model for comparative analyses of the functional evolution of CXE orthologs across lepidopteran species. In contrast to SlittCXE7 (Durand et al., 2011), the localization of EoblCXE7 and EoblCXE13 at the sensillum level is more complex. EoblCXE7 exhibits a similar cellular localization at Str and Sba between the sexes, whereas EoblCXE13 is extensively expressed at multiple sensilla of female E. obliqua but is mainly restricted to Str I in males. Interestingly, (Z)-3-hexenyl acetate, a substrate of both CXE13 and CXE7 in Spodoptera species (Durand et al., 2011; He et al., 2014a), can attract both virgin males and mated females in E. obliqua. These results suggested that EolbCXE7 and EoblCXE13 might function in the E. obliqua olfactory system; however, whether the functions of EolbCXE7 and EoblCXE13 resemble those of their orthologs SlittCXE7 and SexiCXE13 as candidate ODEs in the degradation of (Z)-3-hexenyl acetate remains to be further confirmed. Additionally, EoblCXE7 and EoblCXE13 were highly expressed at Sst, a putative gustatory sensillum in lepidopteran species (Zenker et al., 2011; Tang et al., 2015), and we therefore cannot rule out the possibility that both of these EoblCXEs might function in the gustatory system.

In summary, the present study provides the first identification and characterization of the expression patterns of candidate CXE genes in E. obliqua, a common lepidopteran insect pest of the Geometridae, which will aid the development of new pest management techniques using CXE as potential targets for the disruption of insect foraging behavior.

## AUTHOR CONTRIBUTIONS

LS, QianW, and YoZ conceived and designed the experimental plan. LS, QianW, and QiW preformed the experiments. LS, QianW, YuZ, MT, HG, JF, QX, and YaZ analyzed the data. LS and QianW drafted the manuscript.

#### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (31501652), Zhejiang Provincial Natural Science Foundation of China (LQ16C140003), Central public-interest Scientific Institution Basal Research Fund (1610212016015), China National Basic Research Program (2012CB114104), Research Foundation of State Key Laboratory for Biology of Plant Diseases and Insect Pests (SKLOF201514 and SKLOF201719) and The Science and Technology Innovation Project of the Chinese Academy of Agricultural Sciences (CAAS-ASTIP-2015-TRICAAS). We thank master student Tengfei Mao for his help in the qPCR, and undergraduate students Mengyao Ou, Xuanxuan Yue, and Hongyue Li for their help in insect rearing and tissue collection.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2017.01085/full#supplementary-material

Figure S1 | Sense probe control for in situ hybridization with biotin-labeled probes.

Figure S2 | Original image of Figure 2.

Table S1 | Accession numbers and full names of the insect CXEs used in the phylogenetic analysis.

Table S2 | Primers employed in the RT-PCR assay.

Table S3 | Primers used in the qPCR assay.

Table S4 | Amino acid identities of EoblCXEs.

Table S5 | The t and p values of Student's t-test.

## REFERENCES


Ishida, Y., and Leal, W. S. (2005). Rapid inactivation of a moth pheromone. Proc. Natl. Acad. Sci. U.S.A. 102, 14075–14079. doi: 10.1073/pnas.0505340102


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Sun, Wang, Wang, Zhang, Tang, Guo, Fu, Xiao, Zhang and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Glutathione-S-Transferases in the Olfactory Organ of the Noctuid Moth *Spodoptera littoralis*, Diversity and Conservation of Chemosensory Clades

Nicolas Durand, Marie-Anne Pottier, David Siaussat, Françoise Bozzolan, Martine Maïbèche and Thomas Chertemps\*

Sorbonne Université, INRA, CNRS, UPEC, IRD, Univ. P7, Institute of Ecology and Environmental Sciences of Paris, Paris, France

#### *Edited by:*

Monique Gauthier, Université Toulouse III Paul Sabatier, France

#### *Reviewed by:*

Shannon Bryn Olsson, National Centre for Biological Sciences, India Liang Sun, Tea Research Institute (CAAS), China

#### *\*Correspondence:*

Thomas Chertemps thomas.chertemps@ sorbonne-universite.fr

#### *Specialty section:*

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

*Received:* 29 June 2018 *Accepted:* 27 August 2018 *Published:* 27 September 2018

#### *Citation:*

Durand N, Pottier M-A, Siaussat D, Bozzolan F, Maïbèche M and Chertemps T (2018) Glutathione-S-Transferases in the Olfactory Organ of the Noctuid Moth Spodoptera littoralis, Diversity and Conservation of Chemosensory Clades. Front. Physiol. 9:1283. doi: 10.3389/fphys.2018.01283 Glutathione-S-transferases (GSTs) are conjugating enzymes involved in the detoxification of a wide range of xenobiotic compounds. The expression of GSTs as well as their activities have been also highlighted in the olfactory organs of several species, including insects, where they could play a role in the signal termination and in odorant clearance. Using a transcriptomic approach, we identified 33 putative GSTs expressed in the antennae of the cotton leafworm Spodoptera littoralis. We established their expression patterns and revealed four olfactory-enriched genes in adults. In order to investigate the evolution of antennal GST repertoires in moths, we re-annotated antennal transcripts corresponding to GSTs in two moth and one coleopteran species. We performed a large phylogenetic analysis that revealed an unsuspected structural—and potentially functional—diversity of GSTs within the olfactory organ of insects. This led us to identify a conserved clade containing most of the already identified antennal-specific and antennalenriched GSTs from moths. In addition, for all the sequences from this clade, we were able to identify a signal peptide, which is an unusual structural feature for GSTs. Taken together, these data highlight the diversity and evolution of GSTs in the olfactory organ of a pest species and more generally in the olfactory system of moths, and also the conservation of putative extracellular members across multiple insect orders.

Keywords: GST (glutathione S transferase), *Spodoptera littoralis*, olfaction, detoxification, odorant degrading enzyme

#### INTRODUCTION

Detoxification is a common process present in nearly all living organisms, from prokaryotes to eukaryotes, allowing the elimination of toxic substances of both exogenous or endogenous origin by sequential enzymatic reactions. The first step of detoxification consists in the introduction of functional groups into lipophilic xenobiotics, mainly by oxido-reduction and hydrolysis reactions performed by phase I enzymes, such as cytochromes P450s (CYPs) and carboxylesterases (CCEs). These phase I metabolites are then conjugated to small hydrophilic molecules by phase II enzymes, a large group of broad-specificity transferases, which in combination can metabolize almost any

**349**

hydrophobic compound that contains nucleophilic or electrophilic groups (Bock, 2010). Two of the most important classes of this group are the Glutathione-S-transferases (GSTs, EC 2.5.1.18), and the Uridine diphosphate-glycosyltransferases (UGTs, EC 2.4.1.17) (Jakoby and Ziegler, 1990). These two steps of biotransformation led to more hydrophilic metabolites, facilitating their final excretion by various efflux transporters (phase III) such as multidrug resistance transporters (Dermauw and Van Leeuwen, 2014).

Among phase II enzymes, GSTs are highly diversified and play various functions in the detoxification of a wide range of xenobiotic compounds but also in normal cellular metabolism or in oxidative stress response (Li et al., 2007). They catalyze the conjugation of tri-peptide glutathione (GSH) to endogenous electrophilic molecules or to products from the phase I. GSTs are hetero- or homo-dimeric proteins of approximately 25 kDa in size. Each monomer has two domains joined by a variable linker region. The amino terminal domain is highly conserved and provides most of the GSH binding site (G-site) while the carboxyl terminal domain interacts with the hydrophobic substrate (Hsite).

In insects, GSTs can be sorted based on their cellular localizations, i.e., mainly cytosolic or microsomal, on their substrate specificities and phylogenetic relationships (Enayati et al., 2005). The cytosolic GSTs are divided in six classes (Delta, Epsilon, Omega, Sigma, Theta, and Zeta) (Sheehan et al., 2001). Despite that they catalyze similar reactions than cytosolic GSTs, microsomal GSTs, now referred as membraneassociated proteins in eicosanoid and glutathione (MAPEG), are very different in origin and structure, as they are mainly trimeric transmembrane proteins (Toba and Aigaki, 2000). Insect GSTs are intensely studied for their role in insecticide resistance (Enayati et al., 2005), in particular, members of the delta and epsilon classes, which are specific to arthropods, have been implicated in resistance to various pesticides. MAPEGs also play a role in xenobiotic detoxification (Zhou et al., 2013) and seem in addition involved in aging process (Toba and Aigaki, 2000). Omega, theta, zeta and microsomal sub-groups appear to be involved in various cellular processes, including protection against oxidative stress (Tu and Akgül, 2005).

Insect GSTs are known to be specifically or preferentially expressed into major detoxification organs such as fat body, midgut but also in epidermis and in Malpighian tubules (Huang et al., 2011). However, the expression of GSTs as well as their activities have been also highlighted in the olfactory organs of several insect species. Antennal expressed GSTs have been indeed identified in various moth species, such as Manduca sexta (Rogers et al., 1999), Helicoverpa armigera (Wang et al., 2004), Amyelois transitella (Leal et al., 2009), Bombyx mori (Tan et al., 2014), Chilo suppressalis (Liu et al., 2015a), Epiphyas postvittana (Corcoran et al., 2015), Cnaphalocris medinalis (Liu et al., 2015b), and Cydia pomonella (Huang et al., 2017), but also in the fruit fly Drosophila melanogaster (Younus et al., 2014) or in the beetles Agrilus planipennis (Mamidala et al., 2013), Dendroctonus valens (Gu et al., 2015), and Phyllotreta striolata (Wu et al., 2016). This particular localization led to the hypothesis of a possible dual function of GSTs in antennae where, besides their original implication in xenobiotic metabolism, they could play a role in the signal termination and in odorant clearance, as Odorant-Degrading Enzymes (ODEs, Vogt and Riddiford, 1981; Chertemps, 2017). Moreover, the biochemical characterization of an antennal-restricted GST in M. sexta (msexGST-msolf) confirmed the activity of such enzymes in odorant conjugation and highlighted the possible contribution of GSTs to the detoxification of compounds that might interfere with odorant detection (Rogers et al., 1999).

The cotton leaf worm Spodoptera littoralis, is a highly polyphagous crop pest in a broad area of distribution around the Mediterranean basin (Salama et al., 1971). In this pest species, transcriptomic approaches (Legeai et al., 2011; Jacquin-Joly et al., 2012) have already led to the identification of various olfactory gene repertoires, such as Odorant Receptors and Odorant-Binding Proteins, but also of several phase I and II enzymes, such as CCEs (Durand et al., 2010b), CYPs (Pottier et al., 2012), or UGTs (Bozzolan et al., 2014). Here, using more recent and complete transcriptomic data (Poivet et al., 2013), we identified 33 putative GSTs expressed in the antennae of this species. Moreover, their expression patterns were studied using both qualitative and quantitative PCR and revealed four olfactory-enriched genes in S. littoralis adults. In order to investigate the evolution and phylogenetic relationships of antennal GST repertoires in moths and the relative conservation of such diversity, we built a phylogenetic analysis based on available sequences from 18 insect species, including 13 lepidopteran species. We first re-annotated antennal transcripts corresponding to GSTs in two other moth species belonging to basal lepidopteran taxa, i.e., the peach fruit moth, Carposina sasakii from the Carposinidae family and the purplish birch-miner moth Eriocrania semipurpurella, as a member of the Eriocraniidae. We then annotated GSTs from an antennal transcriptome of a coleopteran species, the european spruce bark beetle Ips typographus, to compare with non-lepidopteran species. This large phylogenetic analysis revealed an unsuspected structural—and potentially functional—diversity of GSTs within the olfactory organ. Surprisingly, inside the delta class, we identified a conserved clade containing most of the already identified antennal-specific and antennal-enriched GSTs from moths. In addition, for all the sequences from this clade, we were able to identify a signal peptide, which is an unusual structural feature for GSTs. Taken together, these data highlight the diversity and evolution of GSTs in the olfactory organ of a pest species and more generally in the olfactory system of moths, with in particular the finding of some conserved putative extracellular members across multiple insect orders.

## MATERIALS AND METHODS

## Insects and Tissue Collection

Insects were reared on semi-artificial diet at 23◦C, 60– 70% relative humidity, and under a 16:8 h light: dark (LD) photoperiod. Adults were kept under an inverted LD regime and provided with a 10% sucrose solution. Male and female antennae and various tissues (proboscis, brain, legs, thorax and abdomen) from 2 day-old males as well as from 7th instar larvae (heads, guts and carcasses) were dissected and stored at −80◦C until RNA extraction.

#### RNA Isolation and cDNA Synthesis

Total RNAs were extracted with TRIzol<sup>r</sup> Reagent (Invitrogen, Carlsbad, CA, United States) and were quantified by spectrophotometry at 260 nm. Single-stranded cDNAs were synthesized from total RNAs (5 µg) from the various tissues using Superscript II reverse transcriptase (Gibco BRL, Invitrogen) and an oligo(dT)18 primer and they were treated with DNase I (Roche, Basel, Switzerland).

## RT-PCR and qRT-PCR

Tissue distribution of S. littoralis GSTs was first investigated by RT-PCR. The ubiquitous ribosomal SlitRpl13 gene, which presents a constant expression in all tissues tested (Durand et al., 2010a), was used as control gene. Primer pairs and PCR conditions are indicated in **Table S1**. PCR products were loaded on 1% agarose gels and visualized using Gel Red (VWR, Radnor, PE, United States) (**Figure S2**). Amplification by qPCR of 8 S. littoralis GSTs, namely SlitGSTd2, SlitGSTe6, SlitGSTe8, SlitGSTe9, SlitGSTe15, SlitGSTs6, SlitGSTo3, SlitMGST1-3, and the reference gene SlitRpl13 was performed as described in detail in Durand et al. (2010a) using the LightCycler<sup>r</sup> 480 real-time PCR system (Roche). Data were analyzed with LightCycler 480<sup>r</sup> software (Roche). The crossing point values (Cp-values) were first determined for the reference genes with a run formed by the 5-fold dilution series, the measuring points, and three negative controls. The normalized S. littoralis GST expressions were calculated with Q-Gene software (Simon, 2003) using SlitRpl13 as reference. This gene has been already demonstrated as the best reference gene in these conditions (Durand et al., 2010b). Each reaction was run in triplicate (technical replicate) with three independent biological replicates. Statistical analyses have been performed with GraphPad Prismr5 software (ANOVA, post-test Tukey's multiple comparison).

## Identification of Antennal GSTs

Putative partial GST cDNAs were identified from a de novo transcriptome of S. littoralis (Poivet et al., 2013) by tBLASTn against a dataset of 47 Spodoptera litura GST sequences (Zhang et al., 2016) and using known GST genes from insect non-redundant sequence databases (National Center for Biotechnology Information, NCBI). Sequences were completed with the de novo transcriptome of S. littoralis composed of tissues from various origin, including larvae (Poivet et al., 2013). Antennal enriched sequences where then deduced from antennal specific transcriptomes (Legeai et al., 2011), and confirmed with PCR methods (see above). We named all S. littoralis sequences according to the corresponding S. litura and S. frugiperda sequences (Zhang et al., 2016; Gouin et al., 2017). Subsequently the 7 S. littoralis GSTs that were previously published in (Lalouette et al., 2016), namely SlGSTe1, e2, e3, e4, d1, d2, and d3 were respectively renamed SlitGSTe15, e8, e14, e12, d2, d3, and SlitMGST1-3. All sequences have been deposited in Genbank with reference accession number from MH177577 to 177613. GST sequences from other antennal transcriptomes were retrieve from high quality datasets (regarding length and quality of transcripts and/or number of overall sequences) and selected in various clades of holometabolous insects for comparison. Ips typographus, E. semipurpurella, and C. sasakii were manually annotated from published antennal transcriptomes [PRJEB3262, PRJNA377940, (Yuvaraj et al., 2017), and PRJNA383289 respectively] using the same protocol as S. littoralissequences and named according our phylogenetic analysis (**Figure S3**).

## Multiple Sequence Alignments and Phylogenetic Analysis

Amino acid sequences were aligned using MAFFT (using L-INS-i option) (Katoh and Standley, 2013) implemented in the Geneious software (http://www.geneious.com, Kearse et al., 2012). Phylogenetic trees were constructed using PhyML (Guindon et al., 2010) based on the LG+G+I+F substitution model as determined by the SMS server (Lefort et al., 2017), using Nearest Neighbor Interchange (NNI). Branch supports were estimated by a Bayesian-like transformation of aLRT (aBayes) (Anisimova et al., 2011). A dendrogram was created and colored using FigTree software (http://tree.bio.ed.ac.uk/ software/figtree/). Our final dataset included 415 sequences, including 271 sequences from 13 lepidopteran species for the cytosolic GST sequences, and 15 sequences from 6 lepidopteran species for the MAPEGs.

#### Identification of Predicted Signal Peptides in GST Sequences

SignalP4.1 software (Petersen et al., 2011) was used with default D-cutoff value to predict the presence and location of signal peptide (SP) cleavage sites in the GST amino acid sequences. We performed a tBlastn analysis on the nucleotide collection and transcriptome shotgun assembly databases available on the NCBI website. To provide a clear representation of the diversity of GSTs with SP, we aligned a selection of sequences corresponding to the different organism families using MUSCLE (Edgar, 2004) (**Figure S4**).

## RESULTS

## Identification of *S. littoralis* Antennal GSTs

A total of 37 full-length sequences encoding putative GST proteins from S. littoralis (SlitGSTs) were identified in our transcriptomic analysis, including 30 cytosolic GSTs and 3 MAPEGs expressed in antennae (**Table 1** and **Figure 1**). Their molecular characteristics such as peptide lengths, estimated molecular masses as well as isoelectric points are indicated in **Table S2**. As a comparison, we analyzed several available transcriptomes from various holometabolous insects. As shown by the comparison with antennal sequences from other species listed in **Table 1**, this is the highest number of putative GSTs identified in an insect antenna. An overlook of GSTs in other holometabolous insects, such as Diptera and Coleoptera, confirmed this high number with only 19 sequences in P. striolata (Wu et al., 2016), 17 in I. typographus (this study) and 31 in D. melanogaster (Younus et al., 2014). The high number of antennal GSTs in S. littoralis is mainly due to an expansion of the epsilon clade, with 15 GSTe sequences for only one to six in the other moth species (**Table 1**), a phenomenon also observed in D. melanogaster with 12 GSTe (Younus et al., 2014).

#### Phylogeny of Lepidopteran Antennal GSTs

As shown by phylogenetic analysis (**Figure 1**), all the identified sequences from S. littoralis were assigned to the 6 already known insect cytosolic GST clades, except 2 sequences assigned to two distinct unclassified clades (Un.1 and Un.2; **Figure 1** and **Table S2**). In addition, we identified 3 microsomal sequences, that represent the highest number of MAPEGs in antennal transcriptome described so far. In E. semipurpurella and C. sasakii we also identified respectively 1 and 2 antennal MAPEGs (**Table 1**). Absence of MAPEG report in the other moth species could probably be due to previous incomplete annotations restricted to cytosolic GSTs.

According to our phylogenetic tree, epsilon, sigma and delta classes represent the major part of the identified sequences in S. littoralis, distributed into 5 well supported sub-clades for epsilon GSTs and 2 sub-clades for sigma and delta GSTs. As shown by the short branch length in the phylogenetic tree, antennal GSTz1 sequences are much conserved in moths, even between nondytrisian and ditrysian species (90.45% of identity between E. semipurpurella and S. littoralis corresponding sequences). The ratios of non-synonymous to synonymous substitutions were estimated for 10 GSTz1 (**Table S3**). Their values far <1.0 indicate that these genes are under strong purifying selection pressure, suggesting a functional conservation. The selection pressure for the GSTz2 genes seems not as important with only 67.1% amino acid identity between BmorGSTz2 and SlitGSTz2 (**Figure 1**).

Amongst SlitGSTs from the delta class, SlitGSTd2 is of particular interest as it falls into a basal clade containing one or two sequences from every species. Moreover, all the delta GSTs from moths known to have an antennal specific expression, such as MsexGST-msolf from M. sexta (Robertson et al., 1999; Rogers et al., 1999), BmGSTd4 from B. mori (Tan et al., 2014), GST-haolf (Wang et al., 2004) from H. armigera and CpomGSTd2 (Huang et al., 2017) from C. pomonella fall within this clade.

#### Tissue-Related Expression of *S. littoralis* GSTs

Seven adult and larval tissues were tested in order to precise the expression patterns of SlitGSTs by RT-PCR and qPCR. Most of them are expressed in all the tissues examined (**Figure 2A**). SlitGSTe15 is preferentially expressed in adults whereas SlitGSTs2 and SlitGSTs6 are preferentially expressed in larval tissues (**Figures 2A,B**). SlitGSTd2, e9, e6, and SlitMGST1-3 seemed preferentially expressed in adult antennae, as confirmed by qPCR on the same tissues but including male legs (**Figure 2B**). These genes are also expressed at low level in other chemosensory or nervous adult tissues, such as in the proboscis and in the legs for SlitGSTd2, in the legs for SlitGSTe9 and in the brain for SlitMGST1-3. All three genes seem more expressed in male antennae than in female antennae but they are not male enriched.


littoralis, S. frugiperda, H. armigera (Noctuidae family, branches colored in red), M. sexta, B. mori (Bombycoidea, dark blue), C. pomonella, E. postvittana (Tortricoidea; green), C. medinalis, C. suppressalis, A. transitella (Pyraloidea, purple), C. sasakii (Carposinidae, orange), P. xylostella (Plutellidae, black), and E. semipurpurella (Eriocraniidae, light blue). The MAPEG clade was used as an outgroup. Circles represent nodes highly supported by the likelihood-ratio test (small dots: aLRT > 0.9, middle dots: aLRT > 0.95, big dots: aLRT = 1). The black stars indicate antennal-enriched or antennal-specific GSTs. The scale bar represents 0.7 expected amino-acid substitutions per site.

## Identification of Predicted Signal Peptide in Insect GST Sequences

As seen previously, SlitGSTd2 clusters in a conserved subgroup which includes BmGSTd4 and CpomGSTd2 for which predicted signal peptide (SP) sequences have been identified (Tan et al., 2014; Huang et al., 2017). SlitGSTd2 possesses also a predicted 22 amino acid SP at the N-terminus of the protein sequence, and its presence was confirmed by using RT-PCR with specific primers (**Figure S1**). Screening of the 36 remaining S. littoralis GST sequences confirmed that SlitGSTd2 was the only sequence presenting this feature.

To test whether this characteristic was shared with the other members of the subclade, we first searched for SP in the sequence of AtraGST (Leal et al., 2009), MsexGST-msolf (Robertson et al., 1999; Rogers et al., 1999), CsupGSTd1 (Liu et al., 2015a) and CmedGSTd1 (Liu et al., 2015b) and we indeed identified putative SPs for all them. We then completed by bioinformatics the N-terminus sequences for the 5 other genes from this clade (SfruGSTd2, HarmGSTd02, BmGSTd1, EposGST11, and PxylGSTd1) and we also identified predicted SP for each of them. All the members of the SlitGSTd2 cluster thus contain a predicted SP.

We then searched the public databases and the literature for insect GSTs possessing this unusual feature and identified predicted SP in 67 full-length GST sequences (**Table S4**). Putative SP is present in the sequences of GSTs from the delta clade in 38 lepidopteran species, including 6 non-dytrisian species, in 6 hemipteran and 9 phasmopteran species. SP was also isolated in a sigma GST from a Coleoptera, the mountain pine beetle Dendroctonus ponderosae. Finally, we found 12 sigma and omega GSTs with SP in nematodes, another group of Ecdysozoa. Alignment of SP sequences from some insect and nematode GSTs (**Figure 3**) revealed that these signal peptides are 15–26 amino acid long and generally end with one to three alanine residues as amino acids with hydrophobic side chain. In Lepidoptera, the amino acids around the cleavage site are well conserved with a N-A-A/X-A followed by a RSK motif.

## DISCUSSION

#### Diversity of GSTs in *S. littoralis* Antenna

In the cotton leafworm S. littoralis, an unsuspected number of CCEs, UGTs, and CYPs had been previously described in antenna (Durand et al., 2010b; Pottier et al., 2012; Bozzolan et al., 2014), leading to the hypothesis that antennal enzymes could participate in signal inactivation and odorant clearance as Odorant-Degrading Enzymes, but also in detoxification processes. Indeed, various airborne compounds, such as toxic allelochemicals emitted by plants or anthropic xenobiotics could enter the olfactory sensilla and reach the olfactory receptor neurons (ORNs) and potentially harm them (Siaussat et al., 2014). In this study, we demonstrate an unsuspected diversity of GSTs in S. littoralis antenna. With 33 described GSTs, the number of antennal-expressed genes is more abundant than any other Lepidopteran described so far.

Without any genome-scale information, the total number of GSTs from S. littoralis cannot be predicted. However, with the recent completion of the genome of S. litura, a sister species (47 GSTs annotated, Cheng et al., 2017), and from a closely related species S. frugiperda (46 GSTs, Gouin et al., 2017), we can speculate that at least two third of total GSTs could be present in S. littoralis antennae. This massive expansion, drived by multiple gene duplications and polymorphism, is suspected to be part of S. littoralis ability to detect and detoxify many plant compounds,

peptides. Each sequence is representative of one family. Slit, Spodoptera littoralis; Bsup, Biston suppressaria; Msex, Manduca sexta; Atra, Amyelois transitella; Bany, Bicyclus anynana; Ehip, Eogystia hippophaecolus; Zfau, Zygaena fausta; Cpom, Cydia pomonella; Pxyl, Plutella xylostella; Tbis, Tineola bisselliella; Tque, Tischeria quercitella; Alam, Andesiana lamellata; Psp, Ptyssoptera sp.; Esem, Eriocrania semipurpurella; Llin, Lygus lineolaris; Meup, Macrosiphum euphorbiae; Ssip, Sipyloidea sipylus; Pphi, Phyllium philippinicum; Aasp, Aretaon asperrimus; Psch, Peruphasma schultei; Dpon, Dendroctonus ponderosae; Ovol, Onchocerca volvulus; Tcan, Toxocara canis; Agal, Ascaridia galli; Bxyl, Bursaphelenchus xylophilus. BxylGSTs full name is BUX.s000647.112 (Espada et al., 2016a). Identifiers in bold indicate sequences linked to representative publications. Pful, Phragmatobia fuliginosa; Bm, Bombyx mori; Sric, Samia ricini; Lsti, Loxostege sticticalis; Ppol, Papilio polytes; Acon, Argyresthia conjugella; Etia, Extatosoma tiaratum.

enabling this species to cope with numerous host plants and may contribute to insecticide resistance (Elghar et al., 2005).

Phylogenetic analysis based on available antennal transcriptomes and genomes from representative lepidopterans, including non-ditrysian basal orders showed that all SlitGST sequences fall into specific clades, with notable expansions in Noctuid family. This great number of orthologous GST groups in moths suggests recent radiation/expansion events in these species.

#### Microsomal Antennal GSTs

In insects, the MAPEG clade contains in general few gene duplicates but seems involved in various metabolic pathways, including aging and pesticide detoxification. Indeed, a mutation in D. melanogaster MGST reduced lifespan, and in Nilaparvata lugens, GSTm2 expression level is induced by several pesticides (Toba and Aigaki, 2000; Zhou et al., 2013). If S. littoralis MAPEGs are very conserved, their expression pattern is remarkably different. SlitMGST1-1 and SlitMGST1-2 were indeed broadly expressed in all tissues and development stages tested whereas SlitMGST1-3 was restricted to the antenna suggesting that these three genes could have evolved different functions after a noctuid-specific duplication event. A previous study showed that SlitMGST1-3 expression is induced after deltamethrin exposure (Lalouette et al., 2016), thus suggesting a possible specialization of SlitMGST1-3 toward toxic molecules. In the beetle P. striolata two MGSTs are also preferentially expressed in the antennae of adult insects (Wu et al., 2016) meaning that members of this protein family could have conserved important function linked to olfaction.

#### Sigma GSTs

We identified 4 sigma GSTs in S. littoralis antenna, which correspond to the second major expansion described in our phylogenetic analysis. Their expression pattern is similar in any tested tissue, they are highly expressed both in larval and adult tissues with the exception of SlitGSTs6 whose expression is almost restricted in the larval midgut and carcass. Moreover, SlitGSTs6 clustered in a conserved group with other larval specific enzymes, like PxylGSTs1 (You et al., 2015). These results are in agreement with the already described expression patterns of other sigma GSTs whose function is often associated with xenobiotic detoxification in insect larval midgut (Huang et al., 2011; Qin et al., 2014). In particular, their expression level is affected after pesticide exposure suggesting a role in response toward toxic compounds and subsequent generated oxidative stress.

Overall, our results suggest that sigma GSTs could be separate in two clades: one with a broad expression, including antenna, associated with basal conjugation functions and one larvalspecific, which activity could be linked with oxidative stress response.

#### Omega GSTs

Omega GSTs define a particular clade as their catalytic properties differ from other GSTs. Indeed, GSTo have a specific dehydroascorbate reductase and thiol transferase activities conferring an oxidative stress protection (Yamamoto et al., 2009a). This conserved function is associated with an apparent 1:1 orthology relationship of the three GSTo groups identified in our phylogenetic tree, with exception of B. mori (4 GSTo) and P. xylostella (5 GSTo). Moreover, expression analysis confirms a consistent pattern throughout any conditions. A GSTo in the silkworm is highly expressed in an insecticide-resistant strain and shows high affinity with organophosphate insecticides, indicating that it may contribute to insecticide resistance and oxidative stress responses, a potential conserved role across lepidopteran GSTs (Yamamoto et al., 2011).

## Zeta and Theta GSTs

The role of the zeta class GSTs had been first linked with phenylalanine and tyrosine catabolism, as maleylacetoacetate isomerases, suggesting a constitutive expression during every life stages (Board et al., 1997). However, BmGSTz had been associated with permethrin resistance with a predominant expression in fat body (Yamamoto et al., 2009b). This two opposite results illustrate the potential role of the two GSTz identified in Lepidoptera, which defined two highly supported clades with a strict 1:1 orthology between species. Thus, we can speculate about their function, with extremely conserved zeta1 GSTs as maleylacetoacetate isomerases (under strong purifying selection pressure) and zeta2 GSTs involved in insecticide resistance. We observed a ubiquitous expression of SlitGSTz1 and z2, in agreement with their proposed functions.

As expected, we also identified a single theta GST in S. littoralis, widely distributed in various tested tissues. In B. mori, GSTt1 has been shown to possess a role in defense mechanisms against oxidative stress and in the metabolism of lipid peroxidation products (Yamamoto et al., 2005).

## Unclassified GSTs

According to our phylogeny, unclassified GSTs segregate in two paralogs groups, each composed of 1:1 orthologs from each lepidopteran species. Functional information regarding those clades is scarce. L. migratoria GSTu1 is expressed in Malpighian tubules and its downregulation using RNAi leads to a higher sensitivity to carbaryl and chlorpyrifos insecticides (Qin et al., 2013). In silkworm, BmGSTu2 is induced in a resistant strain and is able to conjugate glutathione to the organophosphate insecticide diazinion (Yamamoto and Yamada, 2016). SlitGSTu1 and u2 are found to be expressed both at larval and adult stages, in all tissues tested, with a predominant expression of SlitGSTu1. It is likely that those genes could share similar functions than the one observed in other insects.

## Epsilon and Delta GSTs

Epsilon and delta clades are the most common GSTs in insects. They are widely recognized to have specific detoxification functions related to resistance to various insecticides (Enayati et al., 2005). In S. littoralis, epsilon clade accounts for almost half of the described sequences, with various expression patterns ranging from ubiquitous to antennal enriched genes. SlitGSTe clustered with lepidopteran GSTe functionally involved in insecticide conjugation and protection against oxidative stress in B. mori, H. armigera, S. litura, and S. exigua (Huang et al., 2011; Yamamoto et al., 2013; Liu et al., 2015b; Xu et al., 2015; You et al., 2015; Zhou et al., 2015; Wan et al., 2016; Zhang et al., 2016; Hirowatari et al., 2017; Labade et al., 2018). Moreover, three P. xylostella GSTe are also preferentially expressed in the antennae (He et al., 2017), suggesting a potential role in olfaction. SlitGSTe9 and SlitGSTe6 were predominantly expressed in the adult antenna, legs and larval midgut; in addition, SlitGSTe15 was restricted to adult life stages and restricted to chemosensory tissues. This expression pattern suggests a possible role in olfaction and gustation aside from classical detoxification processes encountered in larval midgut.

Of particular interest is a clade containing single sequences from each lepidopteran species where SlitGSTe16 appeared as a putative ortholog of B. mori noperra-bo (BmorGSTe7), a GST with cholesterol transporter activity involved in ecdysteroid biosynthesis (Enya et al., 2015). As SlitGSTe16 is expressed in all tissues, a potential role in endocrine plasticity could be mediated by this enzyme, including in antennae, as ecdysteroids have been shown to modulate S. littoralis olfactory response (Bigot et al., 2012).

Delta GSTs sit in two conserved clades; the first one with SlitGSTd1/d3 includes enzymes capable to metabolize pesticides in B. mori and H. armigera (Yamamoto et al., 2012; Labade et al., 2018) and exhibiting ubiquitous expression pattern. However, as far as expression information is available, the second one with SlitGSTd2 is only composed of antennal-specific enzymes, such as B. mori, M. sexta, and A. transitella and C. pomonella GSTs (Rogers et al., 1999; Leal et al., 2009; Tan et al., 2014; Huang et al., 2017). MsexGST-msolf has been shown to degrade odorants in vitro and may have a role of ODE, especially toward aldehyde odorants, whereas CpomGSTd2 is active toward insecticides. SlitGSTd2 and the other GSTs from this clade may have evolved a function in odorant degradation and/or in protection of the ORNs toward toxic molecules in moth antennae. According to our analysis, SlitGSTd2 is overexpressed in antenna and has the highest expression level, compared to any other genes tested here. Such an expression pattern in consistent with an ODE function as olfaction associated proteins, like OBPs, use to be highly expressed in antenna and to have a rapid turnover in this tissue (Leal, 2013).

In addition, our bioinformatics analysis revealed that all the delta-2-like GST sequences from this clade possess a signal peptide signature, suggesting that they may be secreted proteins. More globally, our extensive analysis of signal peptide presence in GSTs from various insect orders revealed that this structural feature is more widely spread than suspected and not restricted to delta GSTs, as we found also a SP in an antennal GST sigma from a coleopteran species. Extracellular GSTs have been previously characterized in several Nematodes species (Sommer et al., 2001; Liebau et al., 2008; Espada et al., 2016b), and in the pine wood nematode Bursaphelenchus xylophilus; one of these enzyme metabolize various terpenoid compounds (Espada et al., 2016a), also known as common odorant molecules for insects. These antennal GSTs could thus be secreted in the sensillar lymph surrounding the sensory neurons, where they would directly interact with their relative substrate. Several mode of action are likely to occur in this aqueous environment: GSTs can conjugate their substrate with glutathione, as demonstrated in vertebrates where both GSTs and GSH are found in the olfactory mucus (Krishna et al., 1992; Debat et al., 2007); alternatively the binding properties of GSTs (as ligandins, Gonzalez et al., 2018) could act as a scavenger of odorants and harmful compounds.

We have revealed in this present work the occurrence of a high diversity of GST genes expressed in the olfactory organ of a pest moth. Phylogenetic analysis showed that these genes were distributed amongst the well-defined insect GSTs clades, in agreement with different cellular localization. The SlitGST structural diversity together with their different relative

#### REFERENCES


spatial and developmental expression probably reflects their functional divergence and substrate specificities. Amongst this large repertoire, antennal GSTs could play a dual function in this tissue; first as detoxifying enzymes, where they could protect this delicate organ toward harmful compounds, but they could also play a role in the dynamic of olfactory signal: the conjugation of odorants (or their relative metabolites) could induce the termination of olfactory signaling. Moreover, the conjugation of such molecules may play a crucial role in odorant clearance, with the removal of any olfactory-active compounds in the sensillar lymph. Of particular interest is SlitGSTd2, as it is antennal specific, probably secreted and likely to be involved in dual mechanisms in olfactory together with detoxification functions. Further studies using in vitro biochemical assays will reveal SliGSTd2 function and substrate specificity, and will decipher if this enzyme is more related to toxic compounds, odorants or both. Overall future characterization, in particular biochemically but also physiologically, will allow to understand the precise function of all this enzymatic diversity in such a specialized organ, and to unravel their precise role in insect's biology as xenobiotic metabolizing enzymes and/or odorant degrading enzymes.

#### AUTHOR CONTRIBUTIONS

ND, FB, and TC performed analysis. M-AP and DS provide material. ND, MM, and TC wrote the manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01283/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Durand, Pottier, Siaussat, Bozzolan, Maïbèche and Chertemps. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Chemosensory Gene Families in *Ectropis grisescens* and Candidates for Detection of Type-II Sex Pheromones

Zhao-Qun Li, Zong-Xiu Luo, Xiao-Ming Cai, Lei Bian, Zhao-Jun Xin, Yan Liu, Bo Chu and Zong-Mao Chen\*

*Key Laboratory of Tea Biology and Resource Utilization, Ministry of Agriculture, Tea Research Institute, Chinese Academy of Agricultural Science, Hangzhou, China*

Tea grey geometrid (*Ectropis grisescens*), a devastating chewing pest in tea plantations throughout China, produces Type-II pheromone components. Little is known about the genes encoding proteins involved in the perception of Type-II sex pheromone components. To investigate the olfaction genes involved in *E*. *grisescens* sex pheromones and plant volatiles perception, we sequenced female and male antennae transcriptomes of *E*. *grisescens*. After assembly and annotation, we identified 153 candidate chemoreception genes in *E. grisescens*, including 40 odorant-binding proteins (OBPs), 30 chemosensory proteins (CSPs), 59 odorant receptors (ORs), and 24 ionotropic receptors (IRs). The results of phylogenetic, qPCR, and mRNA abundance analyses suggested that three candidate pheromone-binding proteins (EgriOBP2, 3, and 25), two candidate general odorant-binding proteins (EgriOBP1 and 29), six pheromone receptors (EgriOR24, 25, 28, 31, 37, and 44), and EgriCSP8 may be involved in the detection of Type-II sex pheromone components. Functional investigation by heterologous expression in *Xenopus* oocytes revealed that EgriOR31 was robustly tuned to the *E*. *grisescens* sex pheromone component (Z,Z,Z)-3,6,9-octadecatriene and weakly to the other sex pheromone component (Z,Z)-3,9-6,7-epoxyoctadecadiene. Our results represent a systematic functional analysis of the molecular mechanism of olfaction perception in *E*. *grisescens* with an emphasis on gene encoding proteins involved in perception of Type-II sex pheromones, and provide information that will be relevant to other Lepidoptera species.

Keywords: transcriptomic analysis, chemoreception gene, sex pheromone perception, digital gene expression profiling, *Ectropis grisescens*, Type-II sex pheromone compounds, *Xenopus* oocytes

## BACKGROUND

In insects, chemical cues are regarded as language and play significant roles in regulating feeding, mating, and ovipositing (Zhou, 2010). Insect antennae, which are covered with several different types of chemosensory sensilla, are the principal chemosensory organs. Olfactory signal transduction starts with the recognition of odor molecules by olfactory receptors, such as odorant receptors (ORs) and ionotropic receptors (IRs) bound to olfactory receptor neuron (ORN)

#### *Edited by:*

*Peng He, Guizhou University, China*

#### *Reviewed by:*

*Nai-Yong Liu, Southwest Forestry University, China Joe Hull, Agricultural Research Service (USDA), United States*

> *\*Correspondence: Zong-Mao Chen zmchen2006@163.com*

#### *Specialty section:*

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

*Received: 28 August 2017 Accepted: 08 November 2017 Published: 21 November 2017*

#### *Citation:*

*Li Z-Q, Luo Z-X, Cai X-M, Bian L, Xin Z-J, Liu Y, Chu B and Chen Z-M (2017) Chemosensory Gene Families in Ectropis grisescens and Candidates for Detection of Type-II Sex Pheromones. Front. Physiol. 8:953. doi: 10.3389/fphys.2017.00953*

**360**

Li et al. Antennal transcriptome of *E. grisescens*

dendrites. However, the ORNs that are located within chemosensory sensilla are surrounded by aqueous lymphatic fluid. Thus, water-soluble carriers are required to transport lipophilic compounds through the sensilla lymph. Odorantbinding proteins (OBPs) and chemosensory proteins (CSPs) enhance the solubility of odors and deliver them to the olfactory receptors.

The OBPs of insects comprise ∼150 amino acids and belong to the lipocalins superfamily (Flower, 1996), which comprises carrier proteins folded in the typical β-barrel structure (Tegoni et al., 2004). Their most striking feature is six highly conserved cysteines paired into three interlocked disulfide bridges (Pelosi et al., 2014). In the Lepidoptera (Gong et al., 2009), OBPs can be classified into pheromone-binding proteins/general odorantbinding proteins (PBPs/GOBPs), antennal binding protein I (ABPI), ABP II, chemical-sense-related lipophilic-ligand-binding protein (CRLBP), Minus-C, and Plus-C OBPs. The PBPs are thought to be involved in pheromone reception processes (Sun et al., 2013; Jin et al., 2014). The GOBPs are encoded by two paralogous genes (GOBP1 and GOBP2) and are thought to be involved in the detection of plant volatiles and sex pheromones (Liu et al., 2015). The CSPs are soluble binding proteins that consist of 100–120 amino acid residues and have four conserved cysteines forming two independent loops (Angeli et al., 1999). Insects CSPs serve varied functions, including chemosensation (González et al., 2009) and development (Maleszka et al., 2007), as well as other processes (Kulmuni and Havukainen, 2013). For example, Sesamia inferens CSP19 and Helicoverpa armigera HarmCSP6 were reported to show high binding affinities for their respective sex pheromone components (Zhang et al., 2014a; Li et al., 2015).

Insect ORs play key roles in detecting odorants and triggering the transduction of chemical signals into electric signals (Spletter and Luo, 2009; Liu C. et al., 2013). Odorant receptor coreceptor (ORco) is one of the most highly conserved OR genes among various insect species (Nakagawa et al., 2012). It interacts with other ligand-specific ORs to form an OR–ORco complex, which functions as a ligand-gated cation channel (Leal, 2012). Pheromone receptors (PRs), a subfamily of ORs, are specifically activated by sex pheromone components and have been widely studied in Lepidopteran insects (Jiang et al., 2014; Zhang et al., 2014b; Chang et al., 2015). The IRs are an important and ancient repertoire of chemosensory receptors involved in olfaction (Benton et al., 2009) and gustation (Zhang et al., 2013). Previous studies have revealed that IRs also need a coreceptor (Benton et al., 2009). The proteins IR8a and IR25a are antennal IR coreceptors that are expressed at higher levels than other IRs in Drosophila and Chilo suppressalis (Rytz et al., 2013; Cao et al., 2014).

Tea grey geometrid, Ectropis grisescens, is a devastating chewing pest distributed in tea plantations throughout China. The sex pheromone components of E. grisescens have been characterized as (Z,Z,Z)-3,6,9-octadecatriene (Z3Z6Z9-18:Hy) and (Z,Z)-3,9-6,7-epoxyoctadecadiene (Z3Z9-6,7-epo-18:Hy) (Ma et al., 2016). Moth sex pheromone components can be divided into three types according to their structure: Type-I, Type-II, and miscellaneous type with proportions of 75, 15, and 10%, respectively (Ando et al., 2004). Type-I sex pheromone components comprise C10-C<sup>18</sup> straight chain unsaturated alcohols, aldehydes, or acetate esters; and Type-II sex pheromone components consist of C17-C<sup>23</sup> straight chains with 1–3 cis double bonds separated by methylene groups (Millar, 2000; Ando et al., 2004). Therefore, E. grisescens produces Type-II sex pheromone components. Most studies on the sex pheromone perception mechanism in Lepidopteran insects have focused on Type-I pheromone components (Jiang et al., 2014; Zhang et al., 2014b; Chang et al., 2015). Comparatively, little is known about Type-II pheromone components (Zhang et al., 2016). It is acknowledged and accepted that olfaction perception plays crucial roles in the chemical detection of E. grisescens (Sun et al., 2014; Ma et al., 2016). Thus, analysis of its olfactory molecular mechanism may identify targets for pest control. However, little is known about the molecular mechanisms regulating E. grisescens olfaction because of the paucity of sequence data for olfaction genes from E. grisescens. Therefore, to obtain such data as the primary step for exploring the olfaction mechanism, we constructed cDNA libraries of female and male antennae in E. grisescens and conducted several analyses to identify olfactory-related genes.

#### RESULTS

#### Overview of Antennae Transcriptomes

The transcriptomes of female antennae (FA) and male antennae (MA) of E. grisescens were sequenced with two independent biological replicates. About 45.72 (FA-1), 46.51 (FA-2), 42.15 (MA-1), and 49.97 (MA-2) million raw reads were obtained for each transcriptome. The datasets of transcriptomes during the current study are available in the NCBI SRA database (http://trace.ncbi.nlm.nih.gov/Traces/sra/, accession numbers: SRR6004297–SRR6004301). After filtering and assembling, 114,595 transcripts were generated with an N<sup>50</sup> length of 1,715 nt (**Table 1**). For annotation, we combined the female- and maleantennal transcriptomes of E. grisescens and searched against the non-redundant (NR) database by BLASTX with a cut-off e-value of 10−<sup>5</sup> . The best match percentage (40.2%) was with Bombyx mori sequences, followed by sequences from Plutella xylostella (16.7%), Danaus plexippus (16.6%), Acyrthosiphon pisum (1.5%), and Papilio xuthus (1.4%) (**Figure 1A**). Gene ontology (GO) annotation was used to classify the transcripts into functional categories.

Among the distinct transcripts, 32,710 (28.54%) corresponded to at least one GO term.

Within the molecular function category, most transcripts were involved in binding (e.g., nucleotide-, ion-, and odorant-binding) and catalytic activity (e.g., hydrolase and oxidoreductase) (**Figure 1B**).

#### Identification of *E. grisescens* OBP/CSP/OR/IR Genes and Sequence Analyses

Sequence annotation led to the identification of 40 candidate EgriOBPs in the E. grisescens antennae transcriptome (File S1). TABLE 1 | *Ectropis grisescens* antennal transcriptome assembly and annotation.


Sequence analyses showed that 34 of them were full-length EgriOBP genes, and 31 had a predicted signal peptide (**Figure 2A**). EgriOBP13, 14, 15, 32, 37, and 39 contained four conserved cysteines but lacked the conserved cysteines C2 and C5. EgriOBP4, 5, 6, 7, 28, 35, and 40 contained more than six conserved cysteines. EgriOBP8 contained five conserved cysteines but lacked the conserved cysteine C2. All of the E. grisescens classic OBPs present the C-pattern common to Lepidoptera, C1-X25−30-C2-X3-C3-X36−42-C4-X8−14-C5-X8- C6. A total of 30 CSP genes were identified in E. grisescens antennae, 26 of which contained a full-length open reading frame (ORF), a signal peptide, and four conserved cysteine residues (**Figure 2B**). All of the E. grisescens CSPs have the C-pattern of Lepidoptera, C1-X6−8-C2-X18-C3-X2-C4. By homology analysis,

sequences of EgriCSPs. Boxes indicate predicted signal peptides, blue highlight indicates conserved cysteines.

we identified 59 candidate EgriORs in E. grisescens antennae. Sequence analyses revealed that 45 (EgriOR1–44 and EgriORco) of the 59 sequences had an intact ORF with characteristics typical of insect OR genes (full-length ORFs of about 1,200 bp). In total, we identified 24 EgriIRs from E. grisescens antennal transcriptomes, and 16 of them contained intact ORFs.

## Phylogenetic Analyses

To further investigate the function of E. grisescens OBP/CSP/OR/ IR genes, phylogenetic trees were constructed using sequences of typical OBP/CSP/OR/IRs from other insect species for which the whole genome or transcriptome data were available. In the resulting phylogenetic tree, we observed four well-supported groups; PBP, GOBP, Plus-C OBP, and Minus-C OBP (**Figure 3**). Three EgriOBPs (EgriOBP2, 3, and 25) were grouped in the PBP clade with another Lepidoptera PBP. The orthologs EgriOBP1 and 29 were in the GOBP group. EgriOBP4, 5, 6, 7, 28, 35, and 40 were distributed in the Plus-C OBP group, and EgriOBP8, 13, 14, 15, 32, 37, and 39 were distributed in the Minus-C OBP group. In the CSP phylogenetic tree, EgriCSP8 was grouped into the same clade as HarmCSP6 (**Figure 4**).

The EgriORs were distributed among the orthologous groups in the OR phylogenetic tree (**Figure 5**). EgriORco was strongly associated with ObruORco, HvirORco, and BmorOR2 with high bootstrap support. EgriOR25, 28, and ObruOR1 were grouped with B. mori, H. armigera, Helicoverpa assulta, and Heliothis virescens PRs, which are known to be receptors for Type-I pheromones. EgriOR24, 31, 37, and 44 were independently grouped without any orthologs of other Lepidoptera insects. In the IR phylogenetic tree (**Figure 6**), 11 EgriIRs (Egri3, 6, 8, 11, 12, 13, 14, 16, 18, and 24) were clustered with the presumed "antennal" orthologs IR64a, IR21a, IR31a, IR68a, IR75d, IR76b, IR93a, IR60a, and IR40a. Two EgriIRs, EgriIR10, and EgriIR21, were respectively distributed in the IR8a and IR25a groups, which are co-receptors. EgriIR1 was grouped with NMDA iGluRs (N-methyl-d-aspartate ionotropic receptors).

## Tissue Expression Profile and mRNA Abundance of *E. grisescens* OBP/CSP/OR/IR Genes

We further characterized the tissue expression pattern and abundance of E. grisescens OBP/CSP/OR/IR genes in the antennae by qPCR and by evaluating the RPKM (reads per kilobase per million mapped reads) values. The qPCR results indicated that 24 EgriOBPs were uniquely or more strongly expressed in female and male antennae, except for EgriOBP4, 7, 8, 15, 21, 22, 23, 24, 27, 28, 35, 37, 39, and 40 (**Figure 7A**). Of these 24 EgriOBPs showing antenna-biased expression, EgriOBP2, 3, 9, 12, and 25 showed significantly higher transcript levels in male antennae than in female antennae (9.8-, 10.3-, 9.0-, 7.8-, and 12.8-fold higher RPKM values, respectively, in male antennae than in female ones). Of the five male antenna-biased EgriOBPs, EgriOBP2 and 3 were much more abundant in the antennae transcriptome. EgriOBP7, 13, 21, and 33 transcripts were abundant mainly in female and male proboscises. Compared with EgriOBPs, EoblCSPs showed wider and more diverse expression patterns (**Figure 7B**). The transcript levels of EgriCSP5, 8, 13, 14, 15, 16, and 17 were markedly higher in the E. grisescens antennae transcriptomes than in the transcriptomes of other tissues. Of these seven EgriCSPs that were abundant in the antennae, EgriCSP8 was expressed at higher levels in male antennae than in female ones. The abundance of EgriCSP21 and 25 transcripts was markedly higher in female antennae than in male ones. The abundance levels of EgriCSP5, 15, and 17 were also higher in female antennae than in male ones, but the differences were not significant.

Analyses of the expression profile of EgriORs showed that these EgriOR genes were uniquely or more strongly expressed in antennae than in other tissues (**Figure 8A**). Among the ORs, including EgriORco, EgriOR28 and 37 showed the highest expression levels in antennae. Five EgriORs (EgriOR24, 28, 37, 44) were predominantly expressed in male antennae, with RPKM values in male antennae being 51.6-, 29.6-, 20.9-, and 72.7 fold that of their respective counterparts in female antennae. EgriOR31 was uniquely expressed in male antennae. Analyses of the expression patterns of EgriIR genes indicated that EgriIR8, 10, 11, 12, 16, 21, and 24 were highly expressed in the antennae (**Figure 8B**). Of these eight EgriIRs, EgriIR10, 21, and 24 showed relatively high RPKM values in female and male transcriptomes.

## Functional Study of EgriOR31

The Xenopus oocytes and two-electrode voltage clamping recording system were used to characterize the function of the EgriOR1 and 31, by co-expressing with the corresponding receptor EgriORco. The results showed that oocytes coexpressing EgriOR31 and EgriORco robustly responded to the triene Z3Z9-6,7-epo-18:Hy, but less so to Z3Z6Z9-18:Hy. The oocytes co-expressing EgriOR1 and EgriORco showed no responses (**Figure 9**).

## DISCUSSION

Chemical cues, including sex pheromones and host plant volatiles, drive several aspects of insect behavior, such as mating, feeding, and selection of oviposition sites. Sex pheromoneinduced behaviors play crucial roles in insect reproduction. The sex pheromone components of E. grisescens are Z3Z6Z9- 18:Hy and Z3Z9-6,7-epo-18:Hy (Ma et al., 2016), both of which are Type-II compounds. However, little is known about the molecular mechanisms regulating E. grisescens olfaction. Therefore, we analyzed the antennal transcriptomes of female and male E. grisescens to identify genes involved in the perception of sex pheromones and host plant volatiles. In our study, we sequenced E. grisescens female and male antennal transcriptomes, with two independent biological replicates. A total of 26.48 Gb of antennae transcriptome data was obtained. Sequence assembly yielded 114,595 transcripts from E. grisescens antennal transcriptomes. After annotation, we identified 153 candidate chemosensory genes (40 EgriOBPs, 30 EgriCSPs, 59 EgriORs, and 24 EgriIRs) in E. grisescens (File S1).

Insect PBPs represent a sub-class of OBPs that are specifically dedicated to the detection of sex pheromones (Zhou, 2010).

EgriOBP2 and 3 were the two most abundant EgriOBPs in the antennal transcriptome with ∼10-fold higher RPKM values in male antennae than in female ones. EgriOBP25 showed a relatively high RPKM value in male antennae. The phylogenetic tree showed that EgriOBP2, 3, and 25 were distributed in the PBP group with PBPs from S. inferens (Jin et al., 2014), Spodoptera exigua (Liu et al., 2014), Spodoptera litura (Liu N. Y. et al., 2013), H. armigera (Dong et al., 2017), and Hlyphantria cunea (Sanes and Plettner, 2016). That is, EgriOBP2, 3, and 25 were expressed at higher levels in male antennae than in female ones, they were more abundant than other OBPs in male antennae, and they showed homology to other insect PBPs that are known to function in sex pheromone binding. Therefore, EgriOBP2, 3 and 25 may encode PBPs for Type-II pheromone components. Aside from EgriOBP2, 3, and 25, EgriOBP9 and 12 also showed significantly higher expression in male antennae than in female ones, with relatively high RPKM values in male antennae. However, they were not homologous to PBPs of other Lepidoptera insects that produce Type-I pheromone components. Further research is required to clarify the roles of EgriOBP9 and 12 in sex pheromone perception in E. grisescens.

Another key sub-class of OBPs, the GOBPs, are known to be sensitive to both sex pheromones and plant volatiles (Liu et al., 2015). EgriOBP1 and 29 were distributed in GOBP clade. Of them, EgriOBP1 was grouped in the GOBP2 sub-classed with S. litura (Liu et al., 2015), B. mori (Zhou et al., 2009), and S. exigua (Liu et al., 2014) GOBP2 which could strongly bind sex pheromones. Consequently, we speculate that EgriOBP1 may be involved in the binding of sex pheromone components in E. grisescens.

Based on their cysteine motifs, inset OBPs can be classified into several groups: classical OBPs (with six conserved cysteines), Minus-C (with four conserved cysteines) (Forêt and Maleszka, 2006), Plus-C OBPs (with more than six conserved cysteines) (Hekmat-Scafe et al., 2002; Zhou et al., 2004), and dimer OBPs (which contain two amino acid sequence domains) (Hekmat-Scafe et al., 2002). Several Minus-C OBPs (H. armigera HarmOBP17 and 18 and Apis mellifera OBP 14) show high binding affinities to plant volatiles (Spinelli et al., 2012;

Li et al., 2013). In our study, EgriOBP14 and EgriOBP13 were distributed in the same cluster with HarmOBP17 and HarmOBP18, respectively. Thus, these two EgriOBPs might have similar functions to HarmOBP17 and 18, and play roles in plant volatiles perception. EgriOBP22 was associated with H. armigera HarmOBP10, which is expressed at high levels in both antennae and seminal fluid and may function as a carrier of oviposition deterrents (Sun et al., 2012).

Insect PRs, a key sub-class of ORs, are responsible for detecting sex pheromone components in the Lepidoptera (Jiang et al., 2014; Zhang et al., 2014b; Chang et al., 2015). Due to the lack of data for PRs for Type-II pheromone components, we constructed the phylogenetic tree using PRs for Type-I pheromone components. Out of 59 EgriORs, two (EgriOR25 and 28) were grouped in the PR clade with PRs for Type-I sex pheromone components. Among them, EgriOR28 was predominantly expressed in male antennae, with an RPKM value 29.6-fold higher in male antennae than in female antennae. EgriOR28 was the second most abundant OR in antennae. Therefore, these two EgriORs, particularly EgriOR28, may encode PRs for Type-II sex pheromone components. Generally, Lepidoptera insects have about 5∼6 PRs; for example, six PRs were identified in both H. armigera (Liu et al., 2012) and C. suppressalis (Cao et al., 2014). On the other hand, Geometroidea species that produce Type-II sex pheromones, including E. grisescens, are more highly evolved than species that produce Type-I sex pheromones. Therefore, we speculate that another EgriORs involved in the detection of sex pheromones in E. grisescens might be distributed in a group separate from that containing PRs for Type-I sex pheromone components.

In fact, four EgriORs (EgriOR24, 31, 37, and 44) formed an independent group in the phylogenetic analysis. In addition, all four of these EgriORs showed higher abundance in male antennae than in female ones (RPKM values in the male antennae being 51.6-, 29.6-, 20.9-, and 72.7-fold that of their respective counterparts in female antennae). Among these four male antenna-biased EgriORs, EgriOR37 was the most abundant EgriOR in antennal transcriptomes, and EgriOR24 and 31 showed relatively high RPKM values in male antennae.

Consequently, it is conceivable that EgriOR24, 31, 37, and 44 might be potential PRs for Type-II sex pheromone components. To test this above hypothesis, we respectively co-expressed EgriOR31 and EgriOR1 (an EgriOR that sorted to a different phylogenetic clade with EgriOR24, 31, 37, and 44) with the corresponding co-receptor EgriORco, and tested it against the sex pheromone of E. grisescens. The results showed that EgriOR31 and EgriORco robustly responded to Z3Z9-6,7-epo-18:Hy and weakly to Z3Z6Z9-18:Hy. However, EgriOR1 and EgriORco showed no responses. This result indicated that these four male antennae abundant EgriORs (EgriOR24, 31, 37, and 44) which formed an independent group in the phylogenetic analysis might also be potential PRs for Type-II sex pheromone components.

The CSPs and IRs are known to be involved in insect odorant reception. In the CSP phylogenetic tree, EgriCSP8 grouped in the same clade as HarmCSP6, which is responsible for the perception of sex pheromones (Li et al., 2015). In addition, EgriCSP8 showed a male antenna-biased expression pattern with a relatively high RPKM value in male antennae, suggesting that EgriCSP8 plays a role in E. grisescens sex pheromone detection. Consistent with the roles of IRs in olfaction, most of the EgriIRs were clustered with "antennal" orthologs and displayed high expression levels in olfactory tissues. The IRs are known to detect acids, amines, and other odorants that are not sensed by ORs (Benton et al., 2009; Ai et al., 2010, 2013). Of the 24 EgriIRs, EgriIR10, 21, and 24, which was grouped in the antennal IR group, showed relatively high RPKM values in female and male transcriptomes. These results indicated that it might play key roles in olfaction, especially in male E. grisescens.

In conclusion, we sequenced the female and male antennae transcriptomes of E. grisescens to identify the genes involved in chemoreception, with an emphasis on genes encoding proteins involved in the perception of Type-II sex pheromone components. The results of phylogenetic, gene expression, and transcript abundance analyses indicate that a number of EgriOBPs, EgriORs, and EgriCSPs with male antenna-biased expression could be involved in sex pheromone detection. In

particular, EgriOR24, 31, 37, and 44 might be potential PRs for Type-II sex pheromone components. Functional investigation revealed that EgriOR31 was tuned to the E. grisescens sex pheromone components.

## MATERIALS AND METHODS

#### Insect Rearing and Tissue Collection

Individuals of E. grisescens were originally collected from the Experimental Tea Plantation of the Tea Research Institute, Chinese Academy of Agricultural Sciences (Hangzhou, Zhejiang, China). Experimental insects were reared on fresh tea shoots in enclosed nylon mesh cages (70 × 70 × 70 cm) and kept in a climate-controlled room at 25 ± 1 ◦C and 70 ± 5% relative humidity under a 14-h light:10-h dark photoperiod. After pupation, male and female pupae were kept separately in cages for eclosion. After emergence, adult moths were supplied with 10% honey solution. For transcriptome sequencing, 100 female and 100 male antennae were collected from 2-day-old virgin insects, with two replicates. Female and male antennae, heads, thoraxes, thoraxes, legs, wings, proboscises, and pheromone glands were collected from 3-day-old virgin insects for qRT-PCR analyses. These tissues were immediately frozen and stored at −80◦C until RNA isolation.

## cDNA Library Preparation and Illumina Sequencing of Transcriptomes

Total RNA was extracted from female and male antennae using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Quality of the RNA were assessed by agarose gel electrophoresis, Nanodrop (Thermo), and Agilent 2100. The sampling quality satisfy the requirements of cDNA libraries construction. The cDNA library construction and Illumina sequencing were performed at Novogene Bioinformatics Technology Co., Ltd. (Beijing, China). Briefly, mRNAs were isolated from 5 µg pooled total RNA using oligo (dT) magnetic beads and fragmented into short fragments in the presence of divalent cations in fragmentation buffer at 94◦C for 5 min. Using these short fragments as templates, random hexamer primers were used to synthesize first-strand cDNA. Second-strand cDNA was generated using RNase H, and DNA polymerase I. After end repair and adaptor ligation, short sequences were amplified by PCR and purified with a QIAquick <sup>R</sup> PCR purification kit (Qiagen, Venlo, The Netherlands), and sequenced on the HiSeqTM 2500 platform (San Diego, CA, USA).

*De Novo* Assembly of Short Reads and Functional Annotation

pheromone gland. \**P*-value < 0.05.

Transcriptome de novo assembly was carried out with the shortread assembly program Trinity (r20140413p1) (Grabherr et al., 2011) based on the paired-end reads with default settings. Transcripts longer than 150 bp were first aligned by BLASTX to protein databases (NR, Swiss-Prot, KEGG, and COG; e-value < 10−<sup>5</sup> ) to retrieve proteins with the highest sequence similarity to the unigenes along with their protein functional annotations. We then used the Blast2GO (Conesa et al., 2005) for GO annotation of the transcripts and WEGO software (Ye et al., 2006) to plot the GO annotation results.

## Analysis of Transcript Expression in the Antennal Transcriptomes

Transcript abundance was calculated by the RPKM (reads per kilobase per million mapped reads) method (Mortazavi et al., 2008), which can eliminate the influence of different transcript lengths and sequencing discrepancies when calculating abundance (Mortazavi et al., 2008). We used the following equation:

$$RPKM(A) = \frac{C \times 10^6}{\frac{N \times L}{10^3}}$$

where RPKM (A) is the RPKM value of transcript A; C is the number of reads uniquely aligned to transcript A; N is the total number of fragments uniquely aligned to all transcripts; and L is the number of bases in transcript A.

#### Differential Expression Analysis

Genes showing differential expression between two conditions/groups were detected using the DESeq R package (1.10.1), which provides statistical routines to determine differential expression from digital gene expression data using a model based on negative binomial distribution. The resulting P values were adjusted using Benjamini and Hochberg's approach

gland; MAb, male abdomen; FL, female legs; ML, male legs; FW, female wings; MW, male wings; FPr, female proboscis; MPr, male proboscis; Pg, pheromone gland. \**P* < 0.05.

to control the false discovery rate. Genes with an adjusted P < 0.05 found by DESeq were designated as being differentially expressed.

#### Sequence Alignment and Phylogenetic Analysis

Amino acid sequence alignments of the candidate 40 EgriOBPs and 30 EgriCSPs were performed using ClustalX 2.0 (Larkin et al., 2007), and visualized using Jalview 2.4.0 b2 (Waterhouse et al., 2009). The signal peptides of EgriOBPs and EgriCSps were predicted by SignalP 4.1 (http://www.cbs.dtu.dk/services/ SignalP/). To investigate the phylogenetic relationships of OBPs, CSPs, ORs, and IRs between E. grisescens and other typical insect species, we compared them using MAFFT (E-INS-I parameter) (Katoh and Standley, 2013). Phylogenetic trees were constructed using PhyML 3.1 with LG substitution model was used to construct a maximum likelihood phylogenetic tree using Bayesian analysis (Guindon et al., 2010).

#### Quantitative Real-Time PCR Validation

The tissue expression patterns of 40 OBPs, 30 CSPs, 59 ORs, and 24 IRs in different tissues were measured by a qPCR method conducted according to the minimum information for publication of quantitative Real-Time PCR Experiments (Bustin et al., 2009). Total RNA was isolated using the SV Total Isolation System (Promega, Madison, WI, USA) according to the manufacturer's instructions. Quality of the RNA were assessed by agarose gel electrophoresis, Nanodrop (Thermo). Single-stranded cDNA templates were synthesized using the Reverse Transcription System (Promega) as per the manufacturer's instructions. The qRT-PCRs were performed on a Bio-Rad CFX96 touch real-time PCR detection system (Bio-Rad, Hercules, CA, USA). The primers were designed using Beacon Designer 7.7 based on the E. grisescens nucleotide sequences obtained from the transcriptome data (**File S2**). Templates diluted into five-fold series were used to construct a relative standard curve to determine the PCR efficiencies and for quantification analysis. Each reaction was run in triplicate (technical repeat). Two reference genes, guanine nucleotidebinding protein G(q) subunit alpha and glyceraldehyde-3 phosphate dehydrogenase (sequences were provided in **File S3**), were selected in qPCR analysis. A blank control without template cDNA (replacing cDNA with H2O) served as the negative control. Each reaction included three independent biological replicates and was repeated three times (technical replicates). Relative transcript levels were calculated using the comparative 2−11Cq method. The level of each tested mRNA was determined using SYBR <sup>R</sup> Premix Ex TaqTM II (TaKaRa, Dalian, China) according to the manufacturer's instructions.

#### Functional Study of EgriOR31

A Xenopus oocytes expression system was used to express the EgriOR1 and 31. EgriOR1 and 31 were amplified using specific primers (**File S2**). The purified PCR products were ligated into pT7Ts vector using an In-Fusion <sup>R</sup> HD Cloning Kit (Clontech, USA) following manufacturer's instructions. The cRNAs of EgriORs were synthesized using mMESSAGE Mmachine T7 kit (Ambion, Austin, TX). Electrophysiological recording was performed according to previously reported protocols (Wang et al., 2010). Mature healthy oocytes (stage V–VII) were treated with 2 mg/ml collagenase I(GIBCO, Carlsbad, CA) in washing buffer [96 mM NaCl, 2 mM KCl, 5 mM MgCl2, and 5 mM HEPES (pH = 7.6)] for about 1.5 h at room temperature. And then, oocytes were microinjected with 27.6 ng EgriOR cRNAs and 27.6 ng EgriORco. After injection, oocytes were incubated for 4–7 days at 18◦C in 1 × Ringer's solution [96 mM NaCl, 2 mM KCl, 5 mM MgCl2, 0.8 mM CaCl2, and 5 mM HEPES (pH = 7.6)] supplemented with 5% dialyzed horse serum, 50 mg/ml tetracycline, 100 mg/ml streptomycin, and 550 mg/ml sodium pyruvate. Whole-cell currents were recorded from the injected Xenopus oocytes with a two-electrode voltage clamp and recorded with an OC-725C oocyte clamp (Warner Instruments, Hamden, CT, USA). Stock solutions of tested compounds were prepared using ethyl alcohol, which were diluted to the indicated concentrations by 1 × Ringer's buffer before use. Oocytes were exposed to1 × 10−<sup>5</sup> M of sex pheromone compounds and ethyl alcohol. Oocytes without cRNA injection were set as negative control. Data acquisition and analyses were carried out with Digidata 1440A and pCLAMP 10.2 software (Axon Instruments Inc., Union City, CA).

## AUTHOR CONTRIBUTIONS

Z-QL and Z-MC conceived and designed the experiments; Z-QL performed the experiments; Z-QL, Z-XL, X-MC, LB, Z-JX, YL, and BC analyzed the data; and Z-QL wrote the manuscript. All authors reviewed the final manuscript.

## ACKNOWLEDGMENTS

This study was funded by the National Natural Science Foundation of China (31701795), the Central Public-interest Scientific Institution Basal Research Fund (1610212017006), and the National Key Research & Development (R&D) Plan (2016YFD0200900).

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2017.00953/full#supplementary-material

File S2 | Primers used in qPCR and Vector construction.

File S3 | Sequences of guanine nucleotide-binding protein G(q) subunit alpha and glyceraldehyde-3-phosphate dehydrogenase.

#### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Li, Luo, Cai, Bian, Xin, Liu, Chu and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Differential Expression Analysis of Olfactory Genes Based on a Combination of Sequencing Platforms and Behavioral Investigations in Aphidius gifuensis

Jia Fan, Qian Zhang, Qingxuan Xu, Wenxin Xue, Zongli Han, Jingrui Sun and Julian Chen\*

The State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China

Aphidius gifuensis Ashmead is a dominant endoparasitoid of aphids, such as Myzus persicae and Sitobion avenae, and plays an important role in controlling aphids in various habitats, including tobacco plants and wheat in China. A. gifuensis has been successfully applied for the biological control of aphids, especially M. persicae, in green houses and fields in China. The corresponding parasites, as well as its mate-searching behaviors, are subjects of considerable interest. Previous A. gifuensis transcriptome studies have relied on short-read next-generation sequencing (NGS), and the vast majority of the resulting isotigs do not represent full-length cDNA. Here, we employed a combination of NGS and single-molecule real-time (SMRT) sequencing of virgin females (VFs), mated females (MFs), virgin males (VMs), and mated males (MMs) to comprehensively study the A. gifuensis transcriptome. Behavioral responses to the aphid alarm pheromone (E-β-farnesene, EBF) as well as to A. gifuensis of the opposite sex were also studied. VMs were found to be attracted by female wasps and MFs were repelled by male wasps, whereas MMs and VFs did not respond to the opposite sex. In addition, VFs, MFs, and MMs were attracted by EBF, while VMs did not respond. According to these results, we performed a personalized differential gene expression analysis of olfactory gene sets (66 odorant receptors, 25 inotropic receptors, 16 odorant-binding proteins, and 12 chemosensory proteins) in virgin and mated A. gifuensis of both sexes, and identified 13 candidate genes whose expression levels were highly consistent with behavioral test results, suggesting potential functions for these genes in pheromone perception.

Keywords: Aphidius gifuensis, full-length transcriptome, pheromone perception, ORs, IRs, OBPs, CSPs

#### INTRODUCTION

Aphidius gifuensis Ashmead is a dominant endoparasitoid of aphids such as Myzus persicae and Sitobion avenae (Ohta and Honda, 2010) and is best known for its use in the control of tobacco aphids in China. Due to interest in the biocontrol properties of this species, the ecology and biology of A. gifuensis have been extensively studied. Generally, A. gifuensis can start mating 30 min after

#### Edited by:

Shuang-Lin Dong, Nanjing Agricultural University, China

#### Reviewed by:

Hao Guo, Chinese Academy of Sciences, China Ya-Nan Zhang, Huaibei Normal University, China

> \*Correspondence: Julian Chen jlchen@ippcaas.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 25 May 2018 Accepted: 08 November 2018 Published: 27 November 2018

#### Citation:

Fan J, Zhang Q, Xu Q, Xue W, Han Z, Sun J and Chen J (2018) Differential Expression Analysis of Olfactory Genes Based on a Combination of Sequencing Platforms and Behavioral Investigations in Aphidius gifuensis. Front. Physiol. 9:1679. doi: 10.3389/fphys.2018.01679

**374**

emergence. Females mate only once in a lifetime, whereas males mate repeatedly (Bi and Ji, 1994). During mating, female-borne cues are found to be responsible for eliciting courtship behaviors from male wasps (e.g., Bi and Ji, 1994). Olfactory cues are also critical for parasite searching behavior. For example, Eβ-farnesene (EBF), a common active component of the alarm pheromone in aphids, can be tracked by A. gifuensis as a kairomone to locate potential target aphids (e.g., Tan and Liu, 2014).

Various odor-related proteins, such as odorant receptors (ORs), inotropic receptors (IRs), odorant-binding proteins (OBPs) and chemosensory proteins (CSPs), are responsible for specific odor selection and peripheral signal transduction in insects. OBPs and CSPs are concentrated (as high as 10 mM) in the sensillum lymph of insect antennae (Vogt and Riddiford, 1981; Pelosi et al., 2006) and are capable of carrying the semiochemical through the lymph to the ORs or IRs. Most animals, including nematodes, employ G-proteincoupled receptors (GPCR) as transmembrane ORs. The insect OR protein family was first described in Drosophila (Clyne et al., 1999) and was thought to comprise GPCRs as well. However, an opposite transmembrane mode compared with GPCR was later identified (Benton et al., 2006), and the insect ORs were ultimately reclassified as a novel OR protein family. Moreover, insect IRs were recently shown to play roles during insect chemical sensation in Drosophila (Benton et al., 2009), indicating that insect olfactory perception operates through a unique mechanism compared with that in other animals. The dual filtration by soluble proteins (OBPs and CSPs) and transmembrane receptors (ORs and IRs) therefore ensure the high sensitivity of the insect to certain odors, such as pheromones and host odors.

Previous work identified CSPs through next-generation sequencing (NGS) analysis of the antennae transcriptome (Kang et al., 2017) and represents the only molecular biological study of this species. However, little is known about the association between the behavioral responses of this wasp to chemicals and the corresponding functional genes. The molecular mechanism of chemical sensation, including olfactory perception, remains completely unknown.

The reported average lengths of the isotigs from NGS were generally <200 bp, which prevented the assembly of full-length transcripts. Single-molecule real-time (SMRT) sequencing, a third-generation sequencing platform constructed based on PacBio RS (Pacific Biosciences of California, Inc.<sup>1</sup> ), provides long reads that are more than 4 kb in length for both genome sequencing and full-length transcriptome sequencing. Combination of SMRT sequencing with NGS reads has been shown to be ideal for accessing complete transcriptome data (Au et al., 2013).

Olfaction plays a key role in the lifecycle of A. gifuensis, and related ecological and physiological studies have been thorough. However, further study is needed to investigate the following hypotheses:


In the present study, we combined NGS and SMRT sequencing to investigate VF, MF, VM, and MM A. gifuensis wasps to generate a comprehensive full-length A. gifuensis transcriptome. Moreover, the behavioral responses of this species to the aphid alarm pheromone and to wasps of the opposite sex were investigated in detail, enabling precise correlation of the coexpression data from the resulting transcriptional data to males (virgin or mated), which are attracted by females, and to females, which are attracted by the alarm pheromone from aphids and are parasitoids of the aphids. Accordingly, this study provides insights and is a valuable resource for further studies of olfactory mechanisms in A. gifuensis.

#### MATERIALS AND METHODS

#### Insects

Aphidius gifuensis was originally collected from M. persicae mummies in August 2011 in Kunming, Yunnan province, China, and cultured in an air-conditioned insectary [25 ± 2 ◦C60 ± 10% RH, and a photoperiod of 16:8 (L: D) h]. The mummies were collected and placed separately in petri dishes (3.5 cm in diameter). Newly emerged (within 0–12 h) male and female parasitoids were placed in petri dishes (13 cm in diameter and 2 cm in height) for 24 h (24–36-h), either separately for the virgin condition (VF or VM) or together, to allow mating, for the mated condition (MF or MM). For the mated condition, each treated Aphidius (MM or MF) was exposed to 10 virgin A. gifuensis wasps of the opposite sex to ensure that mating occurred during their stay in the petri dishes. The 24-to 36-h-old parasitoids were collected for further studies, such as transcriptome sequencing, behavioral investigation and molecular analyses. Cotton balls filled with 25% defined sugar water were constantly supplied as the diet for adult wasps.

#### Behavioral Responses to EBF and Wasps of the Opposite Sex

Responses of A. gifuensis to EBF and wasps of the opposite sex were investigated in a Y-tube olfactometer. The olfactometer consisted of a Y-shaped glass tube with a 3-cm diameter, a 10-cm trunk length, and a 15-cm branch length. The airflow (0.1 L/min) was dried and purified using activated granular carbon and washed in distilled water before passing through a chamber where the odor source flowed into each arm (branch) of the Y-tube.

<sup>1</sup>http://www.pacificbiosciences.com

Assays were performed as described previously (Mondor et al., 2000; Fan et al., 2015). Briefly, for each treatment, one arm was randomly selected as the treatment arm to introduce either 5 µl of freshly prepared EBF solution (400 ng/µl) or 10 wasps of the opposite sex into the odor chamber connected to the arm, while the other arm was defined as the control arm and was either used to introduce 5 µl of paraffin oil (the solvent used for EBF) or was kept empty, depending on the treatment. EBF was purchased from Wako, Japan. Mineral oil was purchased from Sigma-Aldrich, United States.

The tested insects were visually and physically separated from the odor chamber throughout testing. To prevent the wasps from escaping, bunches of fluffy and ventilated cotton were placed into both sides of the chamber as well as at the exits of both arms of the Y-tube olfactometer. To avoid visual disturbance, a piece of white card paper was placed between the odor chamber and the test area. The tests were conducted using 24- to 36-h postemerged A. gifuensis VMs, MMs, VFs and MFs in a controlled environment at 25 ± 2 ◦C with 60 ± 10% RH, and a 16:8 (L: D) photoperiod. One Aphidius was released into the observed area of the olfactometer and allowed to move either until reaching one-third of the way up one of the arms or for 5 min (300 s). Two series of experiments were performed. In the first series, 10 wasps of the opposite sex of the tested Aphidius were loaded and allowed to move freely in the odor chamber connected to the treatment arm as the odor source. In the second series, 2000 ng (400 ng/µl, 5 µl) of EBF dissolved in mineral oil was employed (dropped onto a piece of 1<sup>∗</sup> 1 cm<sup>2</sup> filter paper) as the odor source. The EBF loaded into the treatment arm was renewed after each test, and 10 wasps of the opposite sex were kept in the odor chamber throughout the test, unless any accidental death occurred, in which case wasp replacement was necessary. Each experiment comprised 300 replications for each treatment (VF, MF, VM, MM) in each series.

#### Statistical Analysis

Differences in the behavioral responses of A. gifuensis to the odors and blank control were determined using χ 2 tests (SAS software 2002, SAS Institute Cary, NC, United States). Insects with no response were not included in the statistical analysis but were counted and are listed in **Table 1**.

#### RNA Sample Preparation

Total RNA was extracted separately from A. gifuensis VFs, MFs, VMs, or MMs (three replicates each) using TRIzol reagent (Invitrogen, United States) according to the manufacturer's instructions. Three-microgram RNA samples with standard quality ratios were purified using poly-T oligo-attached magnetic beads after testing the quality with an Agilent 2100 bioanalyzer.

#### NGS

Divalent cations under elevated temperature in a NEB Next first-strand synthesis reaction buffer (5×) were used for fragmentation. Single-stranded (ss) cDNA was synthesized using a random hexamer primer using M-MuLV reverse transcriptase, DNA polymerase I and RNase H (NEB, United States). After adenylation of the 3<sup>0</sup> ends of the DNA fragments, NEBNext adaptors with a hairpin loop structure were ligated to the fragments for hybridization. The library fragments were purified using the AMPure XP system (Beckman Coulter, United States) to select cDNA fragments that were 150–200 bp long. Then, 3 µl of USER enzyme (NEB, United States) were used with size-selected, adaptor-ligated cDNA at 37◦C for 15 min followed by 5 min at 95◦C before PCR. PCR was then performed using Phusion high-fidelity DNA polymerase, universal PCR primers and an index (X) primer. The products were purified (AMPure XP system), and library quality was assessed using the Agilent Bioanalyzer 2100 system (Agilent Technologies, United States). Clustering of the index-coded samples was performed on a cBot cluster generation system using the TruSeq PE Cluster Kit v3-cBot-HS (Illumina, China) according to the manufacturer's instructions. The library preparations were sequenced on an Illumina HiSeq 2500 platform, and paired-end reads (the sequencing strategy was PE125) were generated after cluster generation. After sequencing, the raw reads were processed to remove low quality and adaptor sequences by NGS QC and then assembled into unigenes using Trinity r20140413p1 with min\_kmer\_cov:2 and the other parameters set to default values.

#### SMRT Sequencing

First-strand cDNA was synthesized using the SMARTer PCR cDNA Synthesis Kit (Clontech<sup>2</sup> ) using SMARTScribe reverse transcriptase, CDS primer IIA [5<sup>0</sup> -AAGCAGTGGTA TCAACGCAGAGTACT30N−1N-3<sup>0</sup> ] and SMARTer IIA oligonucleotide (5<sup>0</sup> -AAGCAGTGGTATCAACGCAGAGT ACXXXXX-3<sup>0</sup> ) for 14 cycles. The purified cDNA was normalized using the Trimmer-2 cDNA Normalization Kit (Evrogen<sup>3</sup> ). Then, second-strand cDNA synthesis was performed using PrimerSTAR GXL DNA polymerase (Clontech<sup>2</sup> ) with 5<sup>0</sup> PCR primer IIA (5<sup>0</sup> -AAGCAGTGGTATCAACGCAGAGTAC-3<sup>0</sup> ) for 18 cycles. The PCR products were purified using 0.4 × AMPure beads (Beckman<sup>4</sup> ). Then, SMRT cell libraries were constructed using a DNA Template Prep Kit (3–10 kb, part; Pacific Biosciences of California, Inc.<sup>1</sup> ). The templates were bound to SA-DNA polymerase and V2 primers. The complexes of the templates and polymerase were bound to magnetic beads and transferred to a 96-well PCR plate at 50 pM on-plate concentrations to reach 50% P1 for processing on a Pacific Biosciences RSII sequencing instrument using C2 sequencing reagents. The 1–2 k library was subjected to SMRT sequencing using 3 SMRT cells, the 2–3 k library was subjected to SMRT sequencing using 3 SMRT cells and the 3–6 k library was subjected to SMRT sequencing using 2 SMRT cells. Subreads were filtered and subjected to circular consensus sequencing (CCS) using the SMRT Analysis Server 2.2.0 (Pacific Biosciences of California, Inc.<sup>1</sup> ).

#### Data Processing and Annotation

The short reads generated with HiSeq 2500 were filtered using the NGS QC Toolkit. Meanwhile, the software proovread

<sup>2</sup>http://www.clontech.com

<sup>3</sup>http://www.evrogen.com

<sup>4</sup>http://www.beckmancoulter.com

(Hackl et al., 2014) was used to correct consensus reads of the full-length transcripts by alignment with filtered NGS short reads. Redundant reads of the error-corrected consensus reads were filtered using CD-HIT-EST. Consensus reads with similarity thresholds of 0.99 were clustered, and redundant sequences were then removed. A total of 81,636 filtered nonredundant sequences were used as input data to perform the annotation. Transcriptome sequences were annotated using seven databases, namely the nonredundant protein sequence (Nr, e-value = 1e−<sup>5</sup> ), non-redundant nucleotide (Nt, e-value = 1e−<sup>5</sup> ), Pfam (evalue = 0.01), Clusters of Orthologous Groups (KOG/COG, e-value = 1e−<sup>3</sup> ), Swiss-Prot (e-value = 1e−<sup>5</sup> ), Kyoto Encyclopedia of Genes and Genomes (KEGG, e-value = 1e−10) and Gene Ontology (GO, e-value = 1e−<sup>6</sup> ) databases.

#### Quantification of Gene Expression

Gene expression levels were estimated by RSEM (Li and Dewey, 2011) for each sample: (I) Clean data were mapped back onto the transcript sequence, and (II) the read count for each gene and isoform was obtained from the mapping results.

#### Differential Expression Analysis

The reads for the Aphidius transcriptomes from four different treatments (VF, MF, VM, and MM), with three replications for each treatment, were produced based on a combination of NGS and SMRT sequencing in this study. Expression analysis of the reads obtained from different treatments was performed using tophat and cufflinks (Trapnell et al., 2012).

Based on the results of the behavioral investigation, differential expression analyses comparing each treatment to VF and VM were separately performed using the DESeq R package (1.10.1). DESeq provide statistical routines for determining differential expression in digital gene expression data using a model based on the negative binomial distribution. The resulting p-values were adjusted using Benjamini and Hochberg's approach for controlling the false discovery rate. Genes found to have an adjusted p-value < 0.05 by DESeq were denoted as differentially expressed genes. The log2(fold change) values and p-values are shown as a volcano plot.

Eight olfactory genes (2 ORs, 2 IRs, 2 OBPs, and 2 CSPs) were randomly selected from each family for qPCR verification of the results from the statistical analysis of the transcriptome sequencing. RT-qPCR was performed on an ABI Prism 7900 Sequence Detection System (Applied Biosystems, Warrington, United Kingdom). SYBR Green Real-Time PCR Master Mixes (Takara, Japan) were used for each PCR in a 20 µl reaction volume containing 1 µl of each primer (5 mM) and 4 µl of first strand cDNA. The primers used for RT-qPCR are listed in **Supplementary Table 2**. Actin served as an internal reference (internal control). Relative expression was calculated using the comparative Ct method 2−11CT and the Ct values of different treatments were normalized to the Ct values of MFs which were defined as the external reference (external control). The results are expressed as the mean ± SD. The qPCR data of the blank control, negative control and RNAi treatment were analyzed by one-way ANOVA followed by Tukey's test.

#### Functional Annotation Enrichment Analysis

According to the results of the behavioral investigation, Venn diagrams of differentially expressed olfaction genes in group1 (VM/VF, MM/VF and MF/VF) and group2 (VF/VM, MF/VM and MM/VM) were constructed using Venny2.1<sup>5</sup> . The mean RPKM values for each gene in the different treatments (VF, MF, VM and MM) were then log-transformed using "log<sup>2</sup> (RPKM + 1)" and subjected to hierarchical clustering using the minimum spanning tree; a heatmap was generated using Heml1.0 (Deng et al., 2014).

## RESULTS

#### Olfactometer Bioassay

We separately compared the taxis of VF, MF, VM, and MM to wasps of the opposite sex (10 wasps) and to the aphid alarm pheromone (2000 ng of EBF) with the taxis to the blank control. Three hundred wasps were tested for each treatment (**Table 1**).

For reciprocal attraction assay between sexes (**Figure 1**), VMs were significantly attracted to the treatment arm (10 females, χ <sup>2</sup> = 4.206; df = 1; P < 0.05) when compared with the control

<sup>5</sup>http://bioinfogp.cnb.csic.es/tools/venny/index.html


TABLE 1 | Response of A. gifuensis wasps (a) to 10 wasps of the opposite sex or a blank control and (b) to the aphid alarm pheromone EBF or a blank control.

VF, virgin female; MF, mated female; VM, virgin male; MM, mated male. All Aphidius wasps emerged during the 24- to 36-h period. N(A), number of individuals who chose the treatment arm; N(B), number of individuals who chose the control arm; N(O), number of individuals who did not select either arm within 5 min. ∗∗P < 0.01; <sup>∗</sup>P < 0.05; NS, not significant. Chi-square (χ 2 ) test comparing test arm and control arm.

arm (pure air). However, more MFs chose the control arm over the treatment arm (10 males, χ <sup>2</sup> = 6.061; df = 1; P < 0.05). Meanwhile, MMs were indifferent to both arms (χ <sup>2</sup> = 0.004; df = 1; P > 0.05), and VFs did not show any attraction to male wasps (χ <sup>2</sup> = 0.195; df = 1; P > 0.05).

For the alarm pheromone (**Figure 1**), both VFs (χ <sup>2</sup> = 4.909; df = 1; P < 0.05) and MFs (χ <sup>2</sup> = 12.368; df = 1; P < 0.01) as well as MMs (χ <sup>2</sup> = 26.828; df = 1; P < 0.01) exhibited preferences for the air from the treatment arm (2000 ng of EBF). Meanwhile, VMs exhibited weaker responses to EBF than to the control; however, the difference was not statistically significant (χ <sup>2</sup> = 1.219; df = 1; P > 0.05).

#### Combined Sequencing Approach for A. gifuensis

To identify and differentiate the transcriptomes of virgin and mated A. gifuensis of both sexes, two sequencing strategies were undertaken, using both NGS and SMRT sequencing platforms (Illumina and PacBio, respectively). First, 12 mRNA samples from four different treatments (VMs, MMs, VFs, and MFs that had emerged within the previous 24–36h; each in triplicate) were subjected to 2 × 125 paired-end sequencing using the HiSeq 2500 platform, yielding 649,863,050 reads. A total of 174178 unigenes were obtained from Illumina sequencing. Second, full-length cDNAs from 12 pooled poly(A) RNA samples were normalized and subjected to SMRT sequencing using the PacBio RS platform, Yielding a total of 518,955 raw reads. After filtering using RS\_Subreads.1 of PacBio RS, 216,385 subreads were obtained. Finally, to resolve the high error rates, all subreads were corrected using the approximately 650 million NGS reads as input data. After removal of the redundant sequences for all the SMRT subreads using CD-HIT-EST (c = 0.85), 81,636 nonredundant reads were produced, with a mean read length of 1970 bases. Of the unigenes from NGS, 56.8% were between 200–500 bp in length, and 21.47% were more than 1 kb. However, the percentage of transcripts from SMRT between 200 and 500 bp in length was only 0.11 and that of transcripts that were more than 1 kb in length was 79.17 (**Table 2**).

#### Annotation of Olfaction-Related Genes in A. gifuensis

Sixty-six ORs, 25 IRs, 16 OBPs, and 12 CSPs were identified (GenBank accession numbers are MK048947- MK049012, MK049025- MK049049, MK049050- MK049065, MK049013- MK049024, respectively) using the NCBI BLASTX program. Gene expression analysis showed that, compared with VMs, MMs

TABLE 2 | Comparison between SMRT sequencing transcripts and Illumina sequencing unigenes.


had 4060 genes that were significantly differentially expressed (MM/VM, 2515 upregulated and 1545 downregulated). The value for A. gifuensis females (MF/VF) was 556 (219 upregulated and 337 downregulated), the VF/VM value was 12608 (7253 upregulated and 5355 downregulated), MF/VM was 19185 (7000 upregulated and 12185 downregulated), VM/VF was 12608 (5355 upregulated and 7253 downregulated), and MM/VF was 13151 (5669 upregulated and 7482 downregulated) (**Figure 2**).

#### Differential Expression Analysis of Olfaction Genes

Based on the behavioral test results (see details in the behavioral investigation section), differentially expressed olfaction genes between treatments were compared with VFs or VMs (**Supplementary Table 1**) and analyzed using Venn diagrams (**Figure 3**).

Neither VFs nor MMs exhibited a behavioral response to A. gifuensis wasps of the opposite sex. However, VMs were attracted by females, and MFs were repelled by males. We firstly chose common olfactory genes with comparable expression levels (no statistically significant differences) between VFs and MMs and denoted these genes as "MM/VF false" (P > 0.05). Then, the differentially expressed olfactory genes (both up- and downregulated) in MF/VF as well as VM/VF were separately compared with "MM/VF false." The final Venn diagram showed seven common genes between "VM/VF true" and "MM/VF false," six of which (4 ORs: c55179\_g2, c53716\_g5, c53086\_g3, c34269\_g1; 1 IR: c46617\_g3, and 1 OBP: c55239\_g5) were present exclusively in two gene sets, namely, "VM/VF true UP" and "MM/VF false" (**Figure 3A**), and the other gene (**Figure 3B**, 1 IR: c56684\_g4) was present exclusively in "VM/VF true DOWN" and "MM/VF false"; "MF/VF true" and "MM/VF false" shared six common genes, two of which (2 ORs: c53272\_g1 and c51725\_g3) were present exclusively in "MF/VF true UP" and "MM/VF false," and four of which (2 ORs: c58301\_g1 and c57979\_g1; 1 IR: c50331\_g1; and 1 CSP: c55251\_g3) were present exclusively in sets "MF/VF true DOWN" and "MM/VF false."

MFs, MMs, and VFs were strongly attracted by EBF. However, VMs were indifferent to EBF. Therefore, we selected olfaction genes that were differentially expressed in MFs, MMs, and VFs compared separately with the expression levels in VM ("MF/VM true UP/DOWN," "MM/VM true UP/DOWN," and "VF/VM true UP/DOWN," P < 0.05; **Figures 3C,D**). The intersection of the Venn diagram showed 1 common gene in the 3 "UP" gene sets ("MF/VM true UP," "MM/VM true UP," and "VF/VM true UP"), namely, c56684\_g4 (IR), and three common genes in the 3 "DOWN" gene sets ("MF/VM true DOWN," "MM/VM true DOWN," and "VF/VM true DOWN"), namely c34269\_g1, c46617\_g3 and c55239\_g5 (1 OR, 1 IR, and 1 OBP, respectively). In summary, the following observations were made based on our behavioral test results: (I) Both MFs and VMs exhibited behavioral responses to wasps of the opposite sex but the responses were opposite. A. gifuensis MF exhibited a lower

Frontiers in Physiology | www.frontiersin.org

genes (downregulated), respectively, which represents seven candidate genes that could be involved in the recognition of the opposite sex by MFs. (C,D) According to the results of the behavioral test for the response to EBF, the number of common up- and downregulated olfactory genes was 1 and 3, respectively, which represent four candidate genes that may be involved in EBF perception.

preference for A. gifuensis males than for the control arms, indicating that MFs are repelled by the males. VMs were attracted by females. Meanwhile, MMs and VFs showed no preference for A. gifuensis wasps of the opposite sex (χ 2 test, P > 0.05, **Figure 1**). The common olfactory genes expressed in both MMs and VFs at comparable levels (P > 0.05) were pooled as the "MM/VF false" set. The sets of differentially expressed genes in VMs and MFs compared with VFs were screened and named "VM/VF true" and "MF/VF true," respectively. Seven common genes in "VM/VF true" and "MM/VF false," but not in "MF/VF true," were found to

be candidate genes involved in the positive behavioral response of VMs to female wasps (4 ORs: c55179\_g2, c53716\_g5, c53086\_g3, c34269\_g1; 2 IR: c46617\_g3, c56684\_g4; and 1 OBP: c55239\_g5), and six common genes in "MF/VF true" and "MM/VF false," but not in "VM/VF true," were found to be candidate genes involved in the negative behavioral response of MFs to male wasps (4 ORs: c53272\_g1, c51725\_g3, c58301\_g1 and c57979\_g1; 1 IR: c50331\_g1; and 1 CSP: c55251\_g3). (II) MFs, VFs, and MMs exhibited chemotaxis toward the aphid alarm pheromone EBF, whereas VMs did not respond. Genes that were differentially expressed in MFs, VFs, and MMs when compared with VMs are shown in a Venn diagram. Four olfaction genes were screened as candidate genes involved in EBF perception (1 OR: c34269\_g1; 2 IRs: c46617\_g3, c56684\_g4, and 1 OBP: c55239\_g5). (III) Four genes, namely, c34269\_g1, c46617\_g3, c56684\_g4 and c55239\_g5 (1 OR, 2 IRs and 1 OBP,), were screened out simultaneously based on the two strategies above; these genes are candidate genes involved in the perception of both the aphid alarm pheromone and A. gifuensis sex pheromone.

The results of differential expression analysis were then verified by qPCR (two randomly selected olfactory genes form each family, see details in **Supplementary Figure 1**). For example, qPCR data showed that CSP c55251 was more highly expressed in both VFs and VMs than MFs and MMs. With no significant difference between VFs and VMs. These results are consistent with the above results from the differential expression analysis of olfaction genes based on the transcriptome sequencing data.

## Functional Annotation Enrichment Analysis

The gene dendrograms showed many clusters of olfactory genes (**Figure 4**), which was consistent with the various functions of these genes in the detection and transmission of olfactory signals from the environment.

The heat map (**Figure 4**) showed that all 4 candidate genes for EBF perception (in VFs, MFs, and MMs, c34269\_g1, c46617\_g3, c56684\_g4 and c55239\_g5) were shared with female volatile perception, which implied that these four genes could participate in either or both physiological processes. The remaining three of the seven candidate genes for female volatile perception (in MMs) were mainly clustered together. In contrast, the six candidate genes for male smell perception (in MFs) were distributed widely across the heatmap. Compared with ORs, OBPs, and CSPs did not show much clustering, which may indicate their generalist nature in ligand binding.

## DISCUSSION

In this study, we carried out, for the first time, two olfactory behavioral investigations, one examining the response to wasps of the opposite sex and the other examining the response to EBF, the alarm pheromone from aphids. Furthermore, wasps were distinguishing by mating status (virgin or mated) rather than simply according sex (male and female) to reveal additional details of A. gifuensis mating and predatory behaviors.

For the reciprocal attraction assay between sexes, VMs were attracted by females, which demonstrated the secretion of a volatile sex pheromone by females. However, MMs did not respond to female volatiles. In addition to olfaction, vision is also believed to be very important to insects (Reeves, 2011). Additionally, the learning ability of Aphidius has been widely reported (e.g., Takemoto et al., 2012). Therefore, multiple sensory behaviors in males, such as olfaction, gustation and vision, likely participate in the recognition of females. With increasing experience, MMs may eventually employ other modes of sensory perception, most likely vision and/or taste, rather than depending solely on olfaction.

All treatment groups except VMs were significantly attracted by EBF. As a common aphid alarm pheromone, EBF signals a high risk to aphids' survival and generally repels aphids. EBF can be used by organisms at high trophic levels as a kairomone to detect and locate aphids (e.g., Micha and Wyss, 1996). A. gifuensis

is known to be an egg parasitoid of aphids, and the ability to track EBF from aphids may enhance the parasite searching behavior of females. This ability can also be helpful when searching for a female mate. However, the results showed that, in contrast to MMs, VMs do not respond to EBF. This finding indicates that after their first mating activity, A. gifuensis males likely exploit strategies other than simply responding to female smells (e.g., sex pheromone attraction). Considering that males do not prey on aphids and that the main role of the males is mating, the positive taxis of MMs to EBF may be an evolutional adaptation for locating aphids, further increasing the chances of encountering a potential mate. However, the exact reason that males stop responding to female smells after the first mating remains unclear.

This is the first study to document a repellent response of mated A. gifuensis females to males. The mechanism may be quite complicated. Females mate only once during their life cycle, whereas males continuously attempt courtship and to mate with any female, even those that have already mated. Therefore, vigilance against males from a distance is more effective than detecting males upon touch.

We also described, for the first time, four cDNA libraries from VM, MM, VF, and MF A. gifuensis wasps using transcriptomes obtained via a combination of NGS and SMRT sequencing. A total of 66 ORs, 25 IRs, 16 OBPs, and 12 CSPs were annotated, and some genes with potentially important functions were further pooled based on the olfactory behavioral investigations mentioned above.

The superiority of SMRT sequencing, which can produce full-length transcripts, compared with short-read sequencing methods has been demonstrated in various species, including humans (Sharon et al., 2013). In the present study, most transcripts (79.17%) were longer than 1 kb, and only 0.11% of the transcripts were between 0–500 bp in length. In contrast, most unigenes (78.52%) from NGS were shorter than 1 kb, and the percentage of unigenes between 0–500 bp was up to 56.84%.

A total of six candidate genes (**Figure 4**, genes are shown in both green and red) were found to be involved in the perception of wasps of the opposite sex by MFs, including four ORs, which were distributed separately on the heatmap. This result suggested that multiple infochemicals help mated females to avoid physical contact with males.

Notably, all four candidate genes associated with EBF perception were present in VMs for female smell perception. This finding may imply a closer evolutionary relationship between genes for perceiving pheromones than between those for normal odor detection.

#### REFERENCES


Substantial progress has been made in studies of insect olfaction mechanisms since ApolOBP, the first functional insect olfactory protein, was identified in Antheraea polyphemus (Vogt and Riddiford, 1981). However, the functional analysis based on 2nd + 3rd generation sequencing and behavioral investigation reported here, particularly the behavioral investigation of VF, MF, VM, MM A. gifuensis wasps, is novel. Our results identified differences in both olfactory responses to certain volatiles and the expression of the corresponding olfactory genes between "before" and "after" mating in males or females; thirteen candidate genes that are potentially involve in EBF and sex pheromone perception were identified from 119 olfactory genes (66 ORs, 25 IRs, 16 OBPs, and 12 CSPs). This approach provides reliable transcript information including coding sequences and expression levels, which have been verified by gene cloning (Au et al., 2013) and qPCR (in the present paper), respectively. Our study definitively provides valuable information for understanding olfaction in A. gifuensis at the molecular level, which will help to strengthen and even take better advantage of A. gifuensis as a powerful and natural biocontrol strategy.

#### AUTHOR CONTRIBUTIONS

JF conceived and designed the study, helped to perform the experiments, analyzed the data, and wrote the paper. QZ helped with RNA extraction as well as data analysis, and revised the manuscript. QX, WX, and ZH performed the behavioral tests and statistical analysis. JS discussed the data and revised the manuscript. JC organized and directed the project.

#### FUNDING

Our study was financially supported by the National Key R & D Plan of China (Nos. 2017YFD0200900, 2016YFD0300700, and 2017YFD0201700), the National Natural Science Foundation of China (No. 31401740), the China Scholarship Council Fund (201703250048), and State Modern Agricultural Industry Technology System (CARS-22-G-18).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01679/full#supplementary-material


odorant receptors in Drosophila. Neuron 22, 327–338. doi: 10.1016/S0896- 6273(00)81093-4


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Fan, Zhang, Xu, Xue, Han, Sun and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fphys-09-01679 November 24, 2018 Time: 16:20 # 10

# Antennal Transcriptome Analysis of the Chemosensory Gene Families From Trichoptera and Basal Lepidoptera

Jothi Kumar Yuvaraj† , Martin N. Andersson\* † , Dan-Dan Zhang and Christer Löfstedt

Department of Biology, Lund University, Lund, Sweden

Edited by:

Peng He, Guizhou University, China

## Reviewed by:

Guan-Heng Zhu, University of Kentucky, United States Hao Guo, Chinese Academy of Sciences, China Joe Hull, Agricultural Research Service (USDA), United States

\*Correspondence: Martin N. Andersson martin\_n.andersson@biol.lu.se †These authors share first authorship

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 09 July 2018 Accepted: 10 September 2018 Published: 27 September 2018

#### Citation:

Yuvaraj JK, Andersson MN, Zhang D-D and Löfstedt C (2018) Antennal Transcriptome Analysis of the Chemosensory Gene Families From Trichoptera and Basal Lepidoptera. Front. Physiol. 9:1365. doi: 10.3389/fphys.2018.01365 The chemosensory gene families of insects encode proteins that are crucial for host location, mate finding, oviposition, and avoidance behaviors. The insect peripheral chemosensory system comprises odorant receptors (ORs), gustatory receptors (GRs), ionotropic receptors (IRs), odorant binding proteins (OBPs), chemosensory proteins (CSPs), and sensory neuron membrane proteins (SNMPs). These protein families have been identified from a large number of insect species, however, they still remain unidentified from several taxa that could provide important clues to their evolution. These taxa include older lepidopteran lineages and the sister order of Lepidoptera, Trichoptera (caddisflies). Studies of these insects should improve evolutionary analyses of insect chemoreception, and in particular shed light on the origin of certain lepidopteran protein subfamilies. These include the pheromone receptors (PRs) in the "PR clade", the pheromone binding proteins (PBPs), general odorant binding proteins (GOBPs), and certain presumably Lepidoptera-specific IR subfamilies. Hence, we analyzed antennal transcriptomes from Rhyacophila nubila (Trichoptera), Eriocrania semipurpurella, and Lampronia capitella (representing two old lepidopteran lineages). We report 37 ORs, 17 IRs, 9 GRs, 30 OBPs, 7 CSPs, and 2 SNMPs in R. nubila; 37 ORs, 17 IRs, 3 GRs, 23 OBPs, 14 CSPs, and 2 SNMPs in E. semipurpurella; and 53 ORs, 20 IRs, 5 GRs, 29 OBPs, 17 CSPs, and 3 SNMPs in L. capitella. We identified IR members of the "Lepidoptera-specific" subfamilies IR1 and IR87a also in R. nubila, demonstrating that these IRs also occur in Trichoptera. Members of the GOBP subfamily were only found in the two lepidopterans. ORs grouping within the PR clade, as well as PBPs, were only found in L. capitella, a species that in contrast to R. nubila and E. semipurpurella uses a so-called Type I pheromone similar to the pheromones of most species of the derived Lepidoptera (Ditrysia). Thus, in addition to providing increased coverage for evolutionary analyses of chemoreception in insects, our findings suggest that certain subfamilies of chemosensory genes have evolved in parallel with the transition of sex pheromone types in Lepidoptera. In addition, other chemoreceptor subfamilies show a broader taxonomic occurrence than hitherto acknowledged.

Keywords: odorant receptor, gustatory receptor, ionotropic receptor, odorant binding protein, chemosensory protein, pheromone, sensory neuron membrane protein

## INTRODUCTION

fphys-09-01365 September 25, 2018 Time: 18:2 # 2

Chemoreception is of utmost importance for the survival and reproduction of insects. The insect antenna is the main olfactory appendage, and it is covered with numerous sensory structures, called sensilla (Keil, 1999; Yuvaraj et al., 2018a). The olfactory sensilla contain the dendrites of olfactory sensory neurons (OSNs), which house chemoreceptors from three divergent multi-gene families, i.e., the odorant receptor (OR) (Clyne et al., 1999; Gao and Chess, 1999; Vosshall et al., 1999), gustatory receptor (GR) (Kwon et al., 2007), and ionotropic receptor (IR) (Benton et al., 2009) families. The receptors and additional families of non-receptor proteins involved in chemosensation (Leal, 2013) have now been identified in many species (Montagné et al., 2015), providing a basis for subsequent evolutionary, structural, and functional studies of these molecular cornerstones of olfaction and taste. However, the chemosensory gene families have, to our knowledge, not yet been identified from certain taxa which could provide important clues to their evolution and origin, including the Trichoptera (caddisflies, the sister group of Lepidoptera) and older lineages of Lepidoptera comprising the so-called non-ditrysian moths (Löfstedt et al., 2016).

Insect ORs are seven-transmembrane proteins with an intracellular N-terminus and extracellular C-terminus, which is opposite to the topology of the G protein-coupled ORs of vertebrates. No homology is shared between these two groups of ORs (Clyne et al., 1999; Benton et al., 2006; Wistrand et al., 2006). In insects, each ligand-binding OR forms a heterotetrameric complex (Butterwick et al., 2018) with an evolutionary conserved odorant receptor co-receptor (Orco) (Vosshall and Hansson, 2011; Stengl and Funk, 2013; Corcoran et al., 2018). Orco is necessary for odor responses (Sato et al., 2008; Wicher et al., 2008), and also important for the ORs to localize in the cell membrane of OSN dendrites (Larsson et al., 2004; Benton et al., 2006; German et al., 2013). With few exceptions, (e.g., Dobritsa et al., 2003; Koutroumpa et al., 2014; Karner et al., 2015) each OSN expresses only one odorant-binding OR, which to a large extent determines the response profile of the OSN. The OR repertoire is highly divergent across insects, both in terms of sequence variation and the total number of ORs encoded by different genomes (Andersson et al., 2015; Montagné et al., 2015). This diversity is thought to at least partly depend on the specialization to different ecological niches (Nei et al., 2008; Hansson and Stensmyr, 2011; Andersson et al., 2015). The GRs belong to the same superfamily as the ORs (Robertson et al., 2003). The GRs are mainly expressed in taste organs and are involved in contact chemoreception (Vosshall and Stocker, 2007), but this gene family also encodes conserved receptors for carbon dioxide, often expressed in the insect antennae (Kwon et al., 2007; Robertson and Kent, 2009).

IRs are related to ionotropic glutamate receptors (iGluRs) that represent a conserved family of ligand-gated ion channels that mediate neuronal communication at synapses in both vertebrates and invertebrates. However, the IRs have atypical binding domains and are involved in sensing the external environment (Benton et al., 2009). The class of 'antennal' IRs comprises receptors involved in olfaction (Benton et al., 2009; Croset et al., 2010; Rytz et al., 2013), humidity (Enjin et al., 2016; Frank et al., 2017), temperature (Chen et al., 2015), and salt sensing (Zhang et al., 2013). These IRs are conserved across insect orders (Croset et al., 2010; Rytz et al., 2013). On the other hand, the 'divergent' IRs are highly variable across species, and members of this class have been assigned a putative role in taste (Croset et al., 2010). A third, phylogenetically distinct, group of IRs occurs in moths and butterflies, and was recently proposed to be Lepidopteraspecific (Liu et al., 2018). In contrast to the ORs, IRs are expressed in a combinatorial fashion in OSNs, and they are in Drosophila tuned to different sets of odorants, notably acids, aromatics, and nitrogen-containing compounds (Abuin et al., 2011). The IRs represent a more ancestral receptor family than the OR family as evidenced by their presence throughout protostome lineages (Croset et al., 2010; Eyun et al., 2017).

Proteins encoded by additional gene families are also important for olfaction. The sensory neuron membrane proteins (SNMPs) are expressed in certain OR-expressing OSNs. SNMPs are integral membrane proteins, related to scavenger proteins, and belonging to the CD36 family. Two representatives of SNMPs (SNMP1 and SNMP2) are found in insects (Nichols and Vogt, 2008), although some insect genomes encode multiple numbers of each group with six putative SNMP1 members being the highest number reported so far (Andersson et al., 2014, 2016). SNMP1 has an important role in pheromone detection in Drosophila and moths (Benton et al., 2007; Li et al., 2014; Pregitzer et al., 2014; Gomez-Diaz et al., 2016). In addition, the protein-rich lymph inside sensilla contains odorant binding proteins (OBPs) involved in the binding and solubilization of odorants in the lymph (Große-Wilde et al., 2006; Leal, 2013). OBPs are small (typically 135–220 amino acids) hydrophilic proteins with conserved cysteine residues forming disulfide bridges (Vogt, 2003; Sánchez-Gracia et al., 2009). Two specialized monophyletic subfamilies of OBPs, the pheromone binding proteins (PBPs) and general odorant binding proteins (GOBPs), appear to be conserved in most species of the higher Lepidoptera (Ditrysia) (Vogt et al., 2015). However, it remains unknown if they are present also in older moth lineages or in caddisflies that use a different chemical type of sex pheromone (Löfstedt et al., 2016). Similar to the OBPs, the insect chemosensory proteins (CSPs) are thought to bind hydrophobic molecules. These proteins are also small (generally around 130 amino acids), and characterized by four conserved cysteine residues forming two disulfide bridges. However, CSPs are also expressed in a variety of non-sensory tissues, and their importance in olfaction and taste remain unclear (Pelosi et al., 2006; Sánchez-Gracia et al., 2009).

The order Trichoptera contains species with aquatic immature stages, and together with the Lepidoptera, they form the suborder Amphiesmenoptera (Kjer et al., 2001; Kjer et al., 2002). Among the Lepidoptera, the early-diverging families Eriocraniidae and Prodoxidae belong to the non-ditrysian group of moths (**Figure 1**). The leaf miner moth, Eriocrania semipurpurella (Eriocraniidae) uses Type 0 sex pheromone compounds (short chain secondary alcohols or ketones) similar to the pheromones used by species in the sister group Trichoptera (**Figure 1**; Löfstedt et al., 2016). The currant shoot borer, Lampronia capitella (Prodoxidae) uses Type I pheromone compounds (C10–C<sup>18</sup>

acetates, alcohols and aldehydes) for sex communication, which is the most common type of sex pheromone in ditrysian moths (Löfstedt et al., 2016). Based on currently available pheromone data within the Lepidoptera, Adeloidea to which Prodoxidae belongs, is the earliest diverging branch in the phylogeny with species using Type I pheromone compounds (**Figure 1**; Löfstedt et al., 2016).

During the last decades, the chemosensory gene families have been extensively studied in ditrysian Lepidoptera and many other groups of insects (e.g., Krieger et al., 2002; Grosse-Wilde et al., 2011; Bengtsson et al., 2012; Andersson et al., 2013, 2014; Corcoran et al., 2015; Walker et al., 2016). On the other hand, the early-diverging lineages of Lepidoptera and its sister group Trichoptera are poorly studied in terms of their chemosensory genes, probably due to the fact that most of these species are not known to be pests of agricultural crops. Among these taxa, there has been a transition in pheromone types from Type 0 to Type I compounds, seemingly representing a major evolutionary transition in terms of chemical communication (Löfstedt et al., 2016). The ORs of E. semipurpurella and L. capitella were reported in our previous functional characterization studies (Yuvaraj et al., 2017, 2018b), but not the other chemosensory gene families. We aim to find out whether the evolution of different pheromone types is associated with complementary changes in the periphery of the pheromone detection system. For example, changes in chemosensory genes may be associated with the transition to Type I pheromones. Hence, we analyzed antennal transcriptomes of L. capitella that belongs to the earliest diverging moth lineage that uses Type I sex pheromones, E. semipurpurella belonging to an even older moth lineage using Type 0 sex pheromones, and Rhyacophila nubila that belongs to the Trichoptera, which also use Type 0 sex pheromones. We report the antennally expressed ORs, IRs, GRs, OBPs, CSPs, and SNMPs in these three species. We also estimate their expression levels based on RNAseq data, and analyze their taxonomic occurrence and evolutionary relationships with the corresponding proteins in moths of the Ditrysia. We reveal that certain subfamilies of chemosensory genes only appear in antennal transcriptomes of moths using Type I sex pheromones, whereas other subfamilies occur more broadly than previously realized. Hence, our results contribute to a better understanding of the evolution of semiochemical communication systems within the superorder Amphiesmenoptera.

## MATERIALS AND METHODS

#### Insects

Pupae of R. nubila were collected from a river stream outside Sjöbo in southernmost Sweden (55◦ 410 13.200N 13◦ 210 24.600E,

88.06 m alt.), and kept at 14–16◦C and 16:8 h light:dark cycle with external aeration in an aquarium for adults to emerge. Adult males of E. semipupurella were collected from a birch forest near Skrylle in southernmost Sweden (55◦ 380 51.000N 13◦ 410 28.100E, 39.53 m alt.) using traps baited with sex pheromone (Kozlov et al., 1996; Yuvaraj et al., 2017). Adults of male and female L. capitella were collected by hand from a black currant plantation near Roskilde, Denmark (55◦ 360 26.800N 11◦ 580 35.200E, 14.54 m alt.).

#### RNA Extraction and First-Strand cDNA Synthesis

Antennae were removed from the males and females under a stereo microscope (Olympus SZ Series Zoom, Olympus, Tokyo, Japan) and stored at −80◦C until RNA isolation. Total RNA from pools of antennae (50 pairs from each sex of R. nubila, 100 pairs from male E. semipurpurella, and 100 pairs from each sex of L. capitella) was extracted using the Trizol protocol (Thermo Fisher Scientific, Carlsbad, CA, United States), and subsequently purified using an RNA Purification Kit (Invitrogen, Carlsbad, CA, United States) according to the manufacturer's instructions. The concentration and quality of the RNA were analyzed using a Nanodrop2000 (Thermo Scientific, Saveen Werner, Malmö, Sweden). First-strand cDNA was synthesized from 1 µg of DNAse-treated total RNA using the ThermoScript RT-PCR system for First-Strand cDNA Synthesis (Thermo Fisher Scientific) following the manufacturer's instructions, except that both oligo-dT primers and random hexamers (1 µL of each) were used in the 20 µL reaction mixtures. The first-strand cDNA library was then used for cloning of chemosensory genes (see section "PCR Confirmation and RACE-PCR Amplification").

#### Sequencing, Assembly and Annotation

The cDNA libraries were sequenced paired-end (100 bp) using an Illumina HiSeq 2000 platform (Illumina, San Diego, CA, United States) at the Beijing Genomics Institute (BGI Hong Kong Co. Ltd.,) and RNAseq libraries were constructed using Illumina's standard protocols. Adaptor sequences were removed and low quality reads trimmed using Trimmomatic<sup>1</sup> . De novo transcriptome assembly of the cleaned data was then performed using the short reads assembly program Trinity (default parameters, version 20121005, Grabherr et al., 2011), and the assembled reads were clustered by TGICL (Pertea et al., 2003). Male and female derived reads were assembled both separately and combined. Primarily the combined assemblies were used for the annotation of chemosensory genes. However, on a few occasions the open reading frames of certain chemosensory genes were longer in the sex-specific assemblies, and were in these cases included in the final dataset. The completeness of each of the assembled transcriptomes (sexes combined for R. nubila and L. capitella) was assessed using the Benchmarking Universal Single-Copy Orthologs (BUSCOv3) tool performed against the Insecta odb9 dataset that includes 1658 reference genes<sup>2</sup> (Waterhouse et al., 2017).

Annotations of chemosensory genes were performed as previously described (Yuvaraj et al., 2017). Briefly, all assembled transcripts were initially included in blast searches against the pooled database of non-redundant (nr) proteins at NCBI (National Center for Biotechnology Information), using an e-value cut-off at 1e−5. Transcripts with hits for putative chemosensory genes were identified from this blast search, open reading frames (ORFs) identified, and then verified by additional BLASTp searches against the nr database. Also, the identified chemosensory genes from the three species were used as queries in additional tBLASTn searches (e-value cut-off < 1e−1) against the transcriptomes of all three species to ensure that all chemosensory genes were identified. Searches against the Pfam web-tool from EMBL-EBI<sup>3</sup> and transmembrane predictions using TMHMM server version 2.0<sup>4</sup> , were undertaken to further support the annotations. Apart from the functionally tested ORs in E. semipurpurella and L. capitella (Yuvaraj et al., 2017, 2018b), the sequences of ORs and OBPs were numbered in the order they were found in each transcriptome. No OR was given the number 2, to avoid confusion with previously reported lepidopteran Orco proteins. Two pairs of OBPs in R. nubila showed >75% identity, and were therefore given the same number (OBP18 or 26), but with an "a" or "b" suffix. SNMPs, IRs, and the OBP subfamilies PBPs and GOBPs were named according to sequence homology with other previously identified lepidopteran proteins. Similarly, putative GRs for carbon dioxide were named GR1-3 (Robertson and Kent, 2009), and putative sugar receptors were named according to sequence homology with such receptors in other moth species. GRs that were not annotated as putative carbon dioxide or sugar receptors were labeled consecutively from number 11. Finally, the CSPs were numbered consecutively based on their tree groupings. Transcript sequences encoding putative chemosensory genes with >99% amino acid identity were regarded as alleles or assembly isoforms and only one copy was included. We use the prefix Rnub for the chemosensory genes of R. nubila, Esem for E. semipurpurella, and Lcap for L. capitella.

The expression levels of transcripts were estimated using the FPKM method (fragments per kb transcript per million mapped reads). The expression of chemosensory genes was regarded as sex-biased if the FPKM values differed by >3-fold between the sexes. This more stringent cut-off compared to the standard twofold change was used due to lack of biological replication. Only genes that had FPKM values above 2 in at least one of the sexes were included in the analysis. The sequences, length details, and FPKM values of all identified chemosensory genes and proteins are presented in **Supplementary Data Sheets S1– S3**, for R. nubila, E. semipurpurella, and L. capitella, respectively.

## PCR Confirmation and RACE-PCR Amplification

To confirm the sequence of some transcripts encoding R. nubila ORs and L. capitella IRs (**Supplementary Data Sheet S3**), PCR amplification from cDNA, followed by cloning and Sanger sequencing were performed. Full length or partial genes

<sup>4</sup>http://www.cbs.dtu.dk/services/TMHMM-2.0/

<sup>1</sup>http://www.usadellab.org/cms/?page=trimmomatic

<sup>2</sup>https://busco.ezlab.org/

<sup>3</sup>http://pfam.xfam.org/search/sequence

were amplified using gene specific primers (oligonucleotide primer sequences are reported in **Supplementary Table S1**) and Platinum <sup>R</sup> Pfu polymerase (Thermo Fisher Scientific), and adenosine residues were added to the ends of the PCR products using GoTaq <sup>R</sup> Green Master mix (Thermo Fisher Scientific). The PCR products were resolved on 0.7% TAE agarose gels and bands of predicted length were cut and purified using the Wizard <sup>R</sup> SV Gel and PCR clean-up system (Promega). The purified PCR products were transformed into the pTZ57R/T vector and colonies were tested for successful transformation. Positive colonies were grown in LB media (containing ampicillin) overnight, and plasmids were extracted using the GeneJET plasmid miniprep kit (Thermo Fisher Scientific). Sequencing PCR was performed using vector-specific primers and BigDye <sup>R</sup> Terminator v1.1 Cycle Sequencing Kit (Thermo Fisher Scientific) following the manufacturer's protocol. The plasmids were then Sanger sequenced using a capillary 3130xL Genetic Analyzer (Thermo Fisher Scientific) at the Department of Biology sequencing facility (Lund University, Lund, Sweden).

Assembled transcripts did not always encode full-length proteins of chemosensory genes, causing miss-alignments that prevented proper phylogenetic analyses of the ORs in particular. Hence, 5<sup>0</sup> and 3<sup>0</sup> RACE-PCR (50 µl reactions) was carried out for some of the short OR transcripts in R. nubila (**Supplementary Data Sheet S1** and **Supplementary Table S1**) to obtain full length sequences, using the SMARTer RACE cDNA Amplification Kit (Clontech, Mountain View, CA, United States) according to the manufacturer's instructions. The following program was used: 5 cycles of 94◦C for 30 s, 72◦C for 3 min; 5 cycles of 30 s at 94◦C, 30 s at 70◦C, 3 min at 72◦C; 20 cycles of 30 s at 94◦C, 30 s at 68◦C, 3 min at 72◦C; and a final extension of 7 min at 72◦C. Further cloning and sequencing was performed as described above.

#### Phylogenetic Analyses

The amino acid sequences of predicted ORs, IRs, SNMPs, OBPs, and CSPs from R. nubila, E. semipurpurella, and L. capitella were aligned together with proteins from Manduca sexta, Plutella xylostella and Epiphyas postvittana using the MAFFT sequence alignment plugin in Geneious R7 software (Biomatters<sup>5</sup> ). To improve the robustness of the phylogenetic analysis, missaligned sequences, OR sequences below 200 amino acids, and IR sequences below 100 amino acids were not included (except the 96-amino acid fragment of IR60a from L. capitella, which aligned well). The OR tree was rooted with the lineage of conserved Orco proteins, and the IR tree with the IR8a and IR25a subfamilies. To ensure correct naming of IRs, Drosophila melanogaster IR sequences were also included in the IR tree. A non-SNMP member of CD36 family (Croquemort, Dmelcrq-A) was used to root the SNMP tree. Maximum-likelihood phylogenetic trees were constructed with RAxML8 (Stamatakis, 2014), and branch support was obtained using 500 bootstrap replicates. The trees were visualized and color coded in FigTree V 1.4.2<sup>6</sup> .

#### RESULTS

#### Assemblies

The Illumina sequencing of the R. nubila antennal samples yielded a total of 110 million reads from each sex. The reads from both sexes combined were assembled into 53,067 transcripts, with a mean length of 1005 bp and an N<sup>50</sup> value of 1991 bp. In total, 65 million reads from the male E. semipurpurella antennal sample were assembled into 68,151 transcripts with a mean length of 818 bp and N50-value of 1,761. The antennal samples of L. capitella yielded 110 million reads from each sex. The reads from both sexes combined were assembled into 60,437 transcripts, with a mean length of 1022 bp and an N<sup>50</sup> value of 2069 bp. The raw sequenced reads have been deposited in the Sequence Read Archive (SRA) database at NCBI under the Bioproject accession numbers: SRR7459244 (R. nubila), SRR5328787 (E. semipurpurella), and SRR6679363 (L. capitella). The transcriptome assemblies have been deposited in the Transcriptome Shotgun Assembly database at DDBJ/EMBL/GenBank under the accessions: GGRG00000000 (R. nubila), GFQP00000000 (E. semipurpurella), and GGRH00000000 (L. capitella). The versions described in this paper are the first versions: GGRG01000000 (R. nubila), GFQP01000000 (E. semipurpurella), and GGRH01000000 (L. capitella). BUSCO analysis using the Insecta odb9 dataset with 1658 reference genes revealed that the completeness of the transcriptomes was high, i.e., 91, 86, and 95%, for R. nubila (sexes combined), E. semipurpurella (male only), and L. capitella (sexes combined), respectively (for additional details, see **Supplementary Table S2**).

#### Receptor Gene Families Odorant Receptors

In previous studies reporting the functional characterization of sex pheromone receptors, we identified 37 ORs in E. semipurpurella (Yuvaraj et al., 2017) and 53 ORs in L. capitella (Yuvaraj et al., 2018b), including the co-receptor Orco. Here, we report 37 ORs from R. nubila, including Orco (**Table 1** and **Supplementary Data Sheet S1**). For R. nubila, two partial transcripts encoding ORs were extended to full-length using RACE-PCR (RnubOR5 and 8). The full-length sequences of nine additional RnubOR transcripts were confirmed from cDNA. Sequences of the cloned and RACE-PCR extended OR

TABLE 1 | Number of genes identified for each chemosensory gene family in Rhyacophila nubila, Eriocrania semipurpurella, and Lampronia capitella.


OR, odorant receptor; GR, gustatory receptor; IR, ionotropic receptor; OBP, odorant binding protein; CSP, chemosensory protein; SNMP, sensory neuron membrane protein.

<sup>5</sup>http://www.geneious.com

<sup>6</sup>http://tree.bio.ed.ac.uk/software/figtree/

genes from the three studied species have been deposited in GenBank (see **Supplementary Data Sheet S4** for accession numbers).

In total, 25 of the transcripts encoding RnubORs were regarded as full-length with more than 400 amino acids (**Supplementary Data Sheet S1**). Two of the longer partial OR fragments (OR24, and OR29) contained between 300 and 400 amino acids, but lacked the N- or C-terminus. Length-details of the OR-encoding transcripts in E. semipurpurella and L. capitella have been reported previously (Yuvaraj et al., 2017, 2018b), but in brief, 24 ORs are full length proteins in E. semipurpurella, and 37 ORs in L. capitella (**Supplementary Data Sheets S2, S3**).

Phylogenetic analysis of the R. nubila, E. semipurpurella, and L. capitella OR sequences was performed together with OR datasets from M. sexta, E. postvittana, and P. xylostella. As expected, the conserved Orco proteins from all species clustered together in a clade that was used to root the tree (**Figure 2**). No ORs from R. nubila or E. semipurpurella grouped within the recently extended lepidopteran pheromone receptor (PR) clade (**Figure 2**; Koenig et al., 2015; Yuvaraj et al., 2018b). In contrast, L. capitella has seven ORs that form two subfamilies within the PR clade (LcapORs 1, 4, 6, 8, and LcapORs 3, 5, 7, respectively), of which LcapORs 6–8 respond to Type I pheromone compounds (Yuvaraj et al., 2018b). Additionally, our phylogenetic analysis suggests that one RnubOR (RnubOR1), two EsemORs (EsemOR1 and 6), one LcapOR (LcapOR15) and one PxylOR (PxylOR3) share a common ancestor with the PR clade, although the position of LcapOR15 and PxylOR3 had low bootstrap support (<20), and is inconsistent with our previous analysis (Yuvaraj et al., 2018b). Based on the specific response of EsemOR1 to the plant volatile β-caryophyllene (Yuvaraj et al., 2017; indicated in **Figure 2**), there is currently no evidence to suggest that these ORs should be included in the PR clade.

As previously reported (Engsontia et al., 2014; Koenig et al., 2015), ORs from P. xylostella showed relatively large species-specific lineage expansions both within and outside the PR clade. In contrast, no major species-specific OR lineage expansions were evident among the three studied species, although a few minor expansions of 4–5 ORs could be observed (**Figure 2**). The remaining ORs from R. nubila, E. semipurpurella, and L. capitella were generally clustered basally or sister to subfamilies containing ORs from M. sexta, E. postvittana, and P. xylostella, across the tree. Several simple one-to-one orthologous relationships with bootstrap support >70% were evident between ORs in R. nubila, E. semipurpurella, and L. capitella: RnubOR3/LcapOR36, RnubOR15/EsemOR24, RnubOR29/EsemOR26, RnubOR31/LcapOR32, EsemOR7/Lcap OR23, and EsemOR9/LcapOR22 (all indicated in **Figure 2**).

The estimated expression levels (FPKM values) of the EsemORs and LcapORs were reported previously (Yuvaraj et al., 2017, 2018b). In terms of sex-biased expression, L. capitella has 7 ORs with estimated male-biased expression of which LcapOR6, 7, and 8 are located within the PR clade (**Figure 2**). Three LcapORs have female-biased FPKM values of which LcapOR3 is within the PR clade (**Table 2**). In R. nubila, 6 ORs have male-biased expression, and 2 ORs female-biased expression (**Table 2**). For E. semipurpurella, we did not have access to a female antennal transcriptome (**Supplementary Data Sheets S1–S3**).

#### Gustatory Receptors

We identified 9 GRs (6 full-length) in R. nubila, 3 GRs (1 full-length) in E. semipurpurella, and 5 GRs (2 full-length) in L. capitella (**Table 1** and **Supplementary Data Sheets S1– S3**). Among these GRs, orthologs of the three carbon dioxide receptors were identified in L. capitella (LcapGR1-3) based on sequence homology; two of them were found in R. nubila (RnubGR1 and RnubGR2), but none of them was found in E. semipurpurella. Two putative non-fructose sugar receptors were identified in R. nubila (RnubGR4 and RnubGR6) as well as in E. semipurpurella (EsemGR4 and EsemGR6), whereas one was found in L. capitella (LcapGR4). One putative fructose receptor was found in each of R. nubila and L. capitella (RnubGR9 and LcapGR9). The remaining GRs (RnubGR11-14 and EsemGR11) were regarded as putative bitter taste receptors. In general, the GRs had low FPKM values and none of them was sexbiased (**Supplementary Data Sheets S1–S3**). Due to the small number of GRs identified, which is expected for antennal transcriptomes, we do not report a phylogenetic analysis for this gene family.

#### Ionotropic Receptors

In total, 17 IRs were identified in R. nubila, 17 in E. semipurpurella, and 20 in L. capitella (**Table 1** and **Supplementary Data Sheets S1–S3**). The conserved antennal IRs (Croset et al., 2010) and IRs belonging to the so-called 'Lepidoptera-specific' IR subfamilies (Liu et al., 2018) were named based on their orthologous relationships with members in other species. Collectively in the three species, we found orthologs for the 'Lepidoptera-specific' receptors IR1 and IR87a, and the antennal receptors IR8a, IR21a, IR25a, IR40a, IR41a, IR60a, IR68a, IR76b, IR93a, and several members of the IR75 group, including IR75d, IR75p, and IR75q (**Figure 3**; Croset et al., 2010). The IR75p and IR75q proteins from L. capitella were further classified based on their phylogenetic positions within the subfamilies of IR75p.1, p.2 and q.2 proteins from other lepidopterans (no IR75q.1 ortholog was found in L. capitella). However, the two IR75p relatives from E. semipurpurella were positioned sister to the entire subfamily of IR75p.1 and p.2 proteins, and were hence named EsemIR75p.0.1 and p.0.2. Similarly, two IRs from E. semipurpurella and three IRs from R. nubila, all related to IR75q, could not be assigned to the specific subfamilies IR75q.1 or q.2. Hence, they were named EsemIR75q.0.1 EsemIR75q.0.2, and RnubIR75q.0.1-q.0.3. We found two members of IR41a in L. capitella (LcapIR41a.1 and LcapIR41a.2), and two members of IR60a in R. nubila (RnubIR60a.1 and RnubIR60a.2). Of the above-mentioned orthologs, we did not find all of them in each of the three species. Specifically, IR64a, IR75d, and putative IR75p members were identified in E. semipurpurella and L. capitella, but not in R. nubila. In addition, an ortholog to one of the divergent IR subfamilies of Lepidoptera, IR7d, was found in R. nubila and L. capitella, but not in E. semipurpurella. The occurrence of an ortholog

of the 'Lepidoptera-specific' IR87a and a member of the IR1 group also in R. nubila suggest that these IRs also occur in Trichoptera. An ortholog of the IR143 group was found only in L. capitella.

Ten IRs from R. nubila, 12 IRs from E. semipurpurella and 12 IRs from L. capitella were putatively full-length, whereas the rest of them are represented as partial genes (**Supplementary Data Sheets S1–S3**). The putative IR co-receptors, IR8a and IR25a, were among the most highly expressed IR transcripts in the three species. IR25a was estimated to be expressed 2–5 times higher than IR8a in R. nubila and L. capitella (**Supplementary Data Sheets S1–S3**). However, in E. semipurpurella the expression of IR25a was low compared to that of IR8a. In addition, RnubIR75q.0.2 and LcapIR76b showed particularly high antennal expression in these species. RnubIR75q.0.2 and RnubIR8a showed male-biased expression (**Table 2**).

TABLE 2 | Chemosensory genes from R. nubila and L. capitella with estimated sex-biased expression (>3-fold difference), presented as FPKM (Fragments Per Kilobase of transcript per Million mapped reads) values.


## Non-receptor Chemosensory Gene Families

#### Odorant Binding Proteins

We identified 30 transcripts encoding OBPs in R. nubila (23 full-length), 23 in E. semipurpurella (20 full-length) and 29 in L. capitella (24 full-length) (**Table 1** and **Supplementary Data Sheets S1–S3**). OBPs are classified into different sub-groups according to the patterns of conserved cysteine residues, and in Lepidoptera also based on phylogenetic position and putative function (**Figure 4**). Classic OBPs have six conserved cysteines, whereas the Plus-C class has 12 cysteines and one characteristic proline residue (Hekmat-Scafe et al., 2002; Sánchez-Gracia et al., 2009; Fan et al., 2011). The Minus-C class (generally) lacks two of the six conserved cysteines, i.e., those at positions two and five. We found 5 RnubOBPs, 1 EsemOBPs, and 4 LcapOBPs that belong to the Plus-C class, and 5 RnubOBPs, 3 EsemOBPs, and 4 LcapOBPs that belong to the Minus-C class (**Figure 4**).

In Lepidoptera, the PBPs (Pheromone Binding Proteins) and GOBPs (General Odorant Binding Proteins) form two monophyletic subfamilies, together sharing a common ancestor, and they appear conserved in ditrysian moths (Vogt et al., 2015). In both E. semipurpurella and L. capitella, we found three members that grouped in the GOBP clade (**Figure 4**). The three EsemGOPBs were all most closely related to members of the GOBP2 subfamily. In contrast, two of the LcapGOBPs were related to the GOBP1 clade, and one was related to the GOBP2 clade. No OBPs from R. nubila grouped within the GOBP clade. Also, we did not find any OBP member that could be classified as a PBP in R. nubila or E. semipurpurella (both using Type 0 pheromones), but we found two PBP members in L. capitella (using a Type I pheromone). LcapPBP1 fell at a position sister to the rest of the PBP clade, but with low bootstrap support. LcapPBP2 grouped together with MsexPBP4, which has been suggested to belong to the PBP-B sub-family (Vogt et al., 2015). The estimated expression levels of the OBPs were in general high. In addition, the FPKM values of a few OBPs indicated malebiased expression, i.e., RnubOBP8, 18a, 18b, 27, and LcapPBP2, OBP1, 8, and 20 (**Table 2**).

#### Chemosensory Proteins

We identified seven transcripts encoding CSPs from R. nubila (all full-length), 14 from E. semipurpurella (9 full-length) and 17 from L. capitella (15 full-length) (**Table 1** and **Supplementary Data Sheets S1–S3**). Some of the CSPs were indicated as highly expressed in the three species, but none of them as abundant as the most highly expressed OBPs or PBPs (**Supplementary Data Sheets S1–S3**). The CSPs from all three species were scattered across the phylogenetic tree, clustering together with CSPs from the other species (**Figure 5**). One CSP had estimated male-biased expression in L. capitella (**Table 2**).

#### Sensory Neuron Membrane Proteins

We identified one member of each of SNMP1 and SNMP2 in R. nubila and E. semipurpurella. In L. capitella, we found two orthologs of SNMP1 (labeled SNMP1a and SNMP1b) and one ortholog of SNMP2 (**Figure 6A** and **Table 1**). Comparing sequence identity, LcapSNMP1a appeared more conserved than LcapSNMP1b with the former sharing 50–75% identity with SNMP1 members from the other moth species included in this analysis. In contrast, LcapSNMP1b, with male-biased expression (**Table 2**), shared about 40% sequence identity with the other SNMP1 orthologs (**Figure 6B**). The shared sequence identity of SNMP2 orthologous was lower, ranging between 30 and

65% across species. All the SNMP transcripts from R. nubila, E. semipurpurella, and L. capitella represent full-length genes.

## DISCUSSION

Prior to this study, the chemosensory gene families had been identified from many species that belong to more recent lineages of Lepidoptera (Ditrysia). This is the first study reporting the identification and evolutionary analyses of the chemosensory gene families from the early-diverging lineages of the Lepidoptera, as well as its sister order Trichoptera. As such, our study enhances the compendium of chemosensory genes in these taxa, providing a foundation for improved evolutionary analyses and functional characterization.

The numbers of putative OR transcripts identified in R. nubila and E. semipurpurella (37 in both species) were lower than the number (53) identified in L. capitella. This suggests that fewer ORs are expressed in the adult antennae of trichopterans as well as in the oldest lepidopteran lineages, as compared to more recent lepidopteran lineages and many other groups of insects (Grosse-Wilde et al., 2011; Zhan et al., 2011; Bengtsson et al., 2012; Andersson et al., 2013, 2014; Engsontia et al., 2014; Corcoran et al., 2015 Dippel et al., 2016; Walker et al., 2016). Indeed, different OR subfamilies have expanded to various degrees in different insect taxa, which possibly reflects differences in ecological specialization (Nei et al., 2008; Hansson and Stensmyr,

2011; Missbach et al., 2014; Andersson et al., 2015; Benton, 2015). However, for E. semipurpurella we could only analyze the male antennal transcriptome, and therefore, ORs with female-specific expression might have been missed. In addition, our BUSCO analysis indicated that the completeness of E. semipurpurella assembly was lower than that for L. capitella (86% vs. 95%), which could partly contribute to the difference in OR numbers observed between these two species. Whether the older lepidopteran lineages and trichopterans in general express fewer antennal ORs than most species of moths should be confirmed by analysis of additional species. As expected, larger numbers of ORs have been identified in the genomes of several moth species with total counts ranging from 64 to 79 (International Silkworm Genome Consortium, 2008; Zhan et al., 2011; Heliconius Genome Consortium, 2012; Engsontia et al., 2014; Koenig et al., 2015). The numbers of ORs encoded by the genomes of the three analyzed species are likely to exceed those reported from the antennal transcriptomes.

Lampronia capitella has seven ORs that group within the lepidopteran PR clade. Three of these ORs responded to Type I pheromone compounds (Yuvaraj et al., 2018b; **Figure 2**). In contrast, no ORs from R. nubila or E. semipurpurella group within the PR clade, and the functionally characterized PRs for Type 0 pheromones in E. semipurpurella have an independent evolutionary origin (Yuvaraj et al., 2017). However, one OR from R. nubila and two ORs from E. semipurpurella are positioned sister to the PR clade and thus appear to share a common ancestor with the PRs of species using Type I pheromones (**Figure 2**). Among these ORs, EsemOR1 responded only to the plant volatile β-caryophyllene. This result led to the hypothesis that the PRs within the PR clade might have evolved their role as sex pheromone detectors from ORs that detect plant volatiles (Yuvaraj et al., 2017). The functional studies of PRs in non-ditrysian lepidopterans suggest that receptors within the PR clade gained a novel function as pheromone detectors in association with the transition from Type 0 to Type I pheromones early in the radiation of the Lepidoptera (Yuvaraj et al., 2018b). However, within the PR clade, there are many receptors with unknown ligands (Engsontia et al., 2014; Koenig et al., 2015; Yuvaraj et al., 2017). In order to improve our understanding of the function and evolution of the receptors within the PR clade, functional studies of ORs

from additional lepidopteran lineages, particularly older ones, are necessary.

Several putative one-to-one orthologous relationships were found between ORs from the three studied species (**Figure 2**), suggesting that some olfactory functions might be conserved among the older lepidopteran lineages and even with Trichoptera. In contrast, very few simple orthologous relationships were evident among the ORs in these moths and those from species in ditrysian families. Instead, the Rnub, Esem, and LcapORs were regularly positioned basally in major lepidopteran OR subfamilies. These patterns of OR relationships are consistent with the species phylogeny, and suggest a phylogenetic signal in the evolution of the OR gene family. R. nubila and L. capitella contain six and seven ORs with male-biased FPKM values, respectively, with the male-biased RnubOR21 grouping close to the Type 0 PRs in E. semipurpurella (**Figure 2**; Yuvaraj et al., 2017). It is possible that the male-biased RnubORs are involved in the detection of the female produced

sex pheromone, but this hypothesis remains to be tested. In addition, quantitative RT-PCR should be performed to verify the sex-biased expression indicated by FPKM values in this study.

The interplay between PBPs and PRs probably facilitates pheromone detection and specificity in moths (Große-Wilde et al., 2006; Forstner et al., 2009; Leal, 2013; Sun et al., 2013). The GOBPs and PBPs form two subfamilies within a Lepidoptera-specific clade, but they had previously only been identified in ditrysian species (Hekmat-Scafe et al., 2002; Vogt et al., 2002; Pelosi et al., 2006; Vieira and Rozas, 2011; Vogt et al., 2015). We did not find any binding proteins that were related to GOBPs or PBPs in R. nubila. Also E. semipurpurella appears to lack antennally expressed PBPs, however, in this species we identified three GOBPs. In L. capitella, we identified both GOBPs and PBPs, representing the first identification of PBPs in a non-ditrysian moth. It has been suggested that PBPs and GOBPs may be mostly associated with pheromonedetecting sensilla trichodea and plant volatile-sensitive sensilla basiconica, respectively (Vogt et al., 1991; Maida et al., 2005; Forstner et al., 2009; Vogt et al., 2015; but see Vogt et al., 2002; Nardi et al., 2003). The presence or absence of PBPs in the antenna may be related to the type of pheromone compounds used. For instance, R. nubila and E. semipurpurella produce Type 0 pheromone compounds whereas L. capitella uses a Type I pheromone (Löfstedt et al., 2016). As mentioned previously, the PRs for Type 0 pheromones in E. semipurpurella have probably evolved from plant odor-detecting ORs, and the characterized EsemPRs also responded secondarily to plant volatiles (Yuvaraj et al., 2017). Thus, due to the structural similarity between Type 0 pheromones and common plant volatiles, it is possible that GOBPs are associated with the detection of Type 0 pheromone compounds in E. semipurpurella. If so, it is surprising that no GOBPs were found in R. nubila, which also uses a Type 0 pheromone. Functional characterization of OBPs from these and additional species from the older Lepidoptera is necessary to test this hypothesis. Nevertheless, the current data suggest that GOBPs are found throughout the Lepidoptera, whereas PBPs appear to be associated only with species using Type I pheromones, at least when considering antennal expression.

Most of the conserved antennal IRs that are found across insects (e.g., Croset et al., 2010; Koenig et al., 2015; Zhao et al., 2015; Dippel et al., 2016; van Schooten et al., 2016; Schoville et al., 2018) were identified in this study. However, a few of the orthologs were not found in all species, which could be due to low antennal expression of some of these IRs. In addition, we found very few IRs of the divergent class (Croset et al., 2010), which was expected because these IRs are primarily expressed in gustatory tissues (Rytz et al., 2013; Koh et al., 2014; van Schooten et al., 2016). Interestingly, we identified several IRs not previously reported outside ditrysian Lepidoptera (Koenig et al., 2015; van Schooten et al., 2016; Liu et al., 2018). Specifically, we identified the first IR143a ortholog in a non-ditrysian moth (L. capitella), IR7 members in both L. capitella and the trichopteran R. nubila, as well as IR87a and IR1 members in both non-ditrysian Lepidoptera and in Trichoptera. Hence, the evolutionary radiation of several IR subfamilies appears to have started prior to the split of the two sister orders Trichoptera and Lepidoptera.

In D. melanogaster and moths, SNMP1 is important for the responses of some pheromone receptors (Benton et al., 2007; Li et al., 2014; Pregitzer et al., 2014; Gomez-Diaz et al., 2016). The SNMPs are conserved across insects (Nichols and Vogt, 2008; Vogt et al., 2009), and we identified them also in our study species. Several species have multiple members of SNMP1 (Nichols and Vogt, 2008; Andersson et al., 2013, 2014), and L. capitella has two members expressed in the antennae. While the sequence of LcapSNMP1a is similar to those of SNMP1 members in other moths and in R. nubila, LcapSNMP1b is more divergent, also in comparison to LcapSNMP1a (**Figure 6B**). Similarly, the six putative SNMP1 members in the Hessian fly, Mayetiola destructor (Diptera, Cecidomyiidae), share only 29–45% sequence identity (Andersson et al., 2014, 2016). The evolutionary forces driving divergence among multiple SNMP1 members within a species remain unknown, but relaxed purifying selection following duplication events might play a role, similar to what has been proposed for OR evolution (Zhang and Löfstedt, 2013; Andersson et al., 2015; Benton, 2015; Zhang and Löfstedt, 2015). In addition, the function of multiple SNMP1 members within a species remains to be unraveled, whether olfactory or not. In the Hessian fly, the responses of MdesOR115 to minor pheromone components were not affected by co-expression of the different SNMP1 members when tested in vitro (Andersson et al., 2016). However, this result does not rule out an important role for any of the different SNMP1s in vivo.

#### CONCLUSION

Our transcriptome analysis provides the first set of chemosensory genes from the older Lepidoptera and a species of Trichoptera, facilitating the evolutionary analysis of these gene families in these two diverse orders of Insecta. In addition to showing that several subfamilies of chemosensory genes are shared between these orders, our results suggest that the conserved PR clade of Lepidoptera and the PBPs have emerged in parallel with the evolution of Type I sex pheromones, although this hypothesis should be tested by genome analysis. Future studies should aim to characterize the function of these olfactory proteins to further our understanding of the relationship between species ecology, pheromone communication, and the evolution of olfactory proteins in relation to species diversification.

## AUTHOR CONTRIBUTIONS

JY, MA, and CL conceived and designed the study. JY collected biological material. JY performed molecular work with assistance from D-DZ. JY and MA performed transcriptome data analysis and constructed the phylogenetic trees. JY and MA wrote the manuscript together with contributions from D-DZ and CL. All authors read and approved the final version of the manuscript.

## FUNDING

This work was supported by the Swedish Royal Physiographic Society in Lund (to JY), the Swedish Foundation for International Cooperation in Research and Higher Education (Grant Numbers IB2013-5256 and IG2013-5483), and the Swedish Research Councils VR (Grant Numbers VR-621-2013-4355 and VR-2017-03804 to CL) and FORMAS (Grant Number 217-2014-689 to MA).

#### ACKNOWLEDGMENTS

fphys-09-01365 September 25, 2018 Time: 18:2 # 14

We would like to thank Erling Jirle and Hong-Lei Wang for assistance with the collection of biological material. We would like to thank Richard Newcomb, Olle Anderbrant, Ewald Grosse-Wilde, and William Walker for their constructive comments on previous drafts of this manuscript.

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01365/full#supplementary-material

TABLE S1 | Oligonucleotide primer sequences of cloned (full length) and RACE-PCR-extended odorant receptor (OR) and ionotropic receptor (IR) genes.

TABLE S2 | Assessment of transcriptome assembly completeness using the Benchmarking Universal Single-Copy Orthologs (BUSCOv3) tool performed against the Insecta odb9 dataset (https://busco.ezlab.org/).

DATA SHEET S1 | Sequences (DNA and protein), length details, and expression levels of the chemosensory genes identified in Rhyacophila nubila.

DATA SHEET S2 | Sequences (DNA and protein), length details, and expression levels of the chemosensory genes identified in Eriocrania semipurpurella.

DATA SHEET S3 | Sequences (DNA and protein), length details, and expression levels of the chemosensory genes identified in Lampronia capitella.

DATA SHEET S4 | Accession numbers and/or source references for the protein sequences included in the trees presented in Figures 2–6.



genome of the Colorado potato beetle, Leptinotarsa decemlineata (Coleoptera: Chrysomelidae). Sci. Rep. 8:1931. doi: 10.1038/s41598-018-20154-1


to gene prediction and phylogenomics. Mol. Biol. Evol. 35, 543–548. doi: 10. 1093/molbev/msx319


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Yuvaraj, Andersson, Zhang and Löfstedt. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identification and Comparison of Chemosensory Genes in the Antennal Transcriptomes of Eucryptorrhynchus scrobiculatus and E. brandti Fed on Ailanthus altissima

#### Xiaojian Wen, Qian Wang, Peng Gao and Junbao Wen\*

Beijing Key Laboratory for Forest Pests Control, College of Forestry, Beijing Forestry University, Beijing, China

#### Edited by:

Peng He, Guizhou University, China

#### Reviewed by:

Hao Guo, Chinese Academy of Sciences, China Da-Song Chen, Guangdong Institute of Applied Biological Resources, China Hetan Chang, Stowers Institute for Medical Research, United States

> \*Correspondence: Junbao Wen wenjb@bjfu.edu.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 30 July 2018 Accepted: 02 November 2018 Published: 20 November 2018

#### Citation:

Wen X, Wang Q, Gao P and Wen J (2018) Identification and Comparison of Chemosensory Genes in the Antennal Transcriptomes of Eucryptorrhynchus scrobiculatus and E. brandti Fed on Ailanthus altissima. Front. Physiol. 9:1652. doi: 10.3389/fphys.2018.01652 The key to the coexistence of two or more species on the same host is ecological niche separation. Adult Eucryptorrhynchus scrobiculatus and E. brandti both feed on the tree of heaven, Ailanthus altissima, but on different sections of the plant. Olfaction plays a vital role in foraging for food resources. Chemosensory genes on the antennae, the main organ for insect olfaction, might explain their feeding differentiation. In the present study, we identified 130 and 129 putative chemosensory genes in E. scrobiculatus and E. brandti, respectively, by antennal transcriptome sequencing, including 31 odorant-binding proteins (OBPs), 11 chemosensory proteins (CSPs), 49 odorant receptors (ORs), 17 ionotropic receptors (IRs), 19 gustatory receptors (GRs), and three sensory neuron membrane proteins (SNMPs) in E. scrobiculatus and 28 OBPs, 11 CSPs, 45 ORs, 25 IRs, 17 GRs, and three SNMPs in E. brandti. We inferred that EscrOBP8 (EscrPBP1), EscrOBP24 (EscrPBP2) and EbraOBP8 (EbraPBP1), EbraOBP24 (EbraPBP2) were putative PBPs by the phylogenetic analysis. We identified species-specific OR transcripts (10 EscrORs and 8 EbraORs) with potential roles in the recognition of specific volatiles of A. altissima. In addition to conserved "antennal IRs," we also found several "divergent IRs" orthologues in E. scrobiculatus and E. brandti, such as EscrIR16, EbraIR19, and EbraIR20. Compared with other chemosensory genes, GRs between E. scrobiculatus and E. brandti shared lower amino acid identities, which could explain the different feeding habits of the species. We examined OBP expression patterns in various tissues and sexes. Although amino acid sequence similarities were high between EscrOBPs and EbraOBPs, the homologous OBPs showed different tissue expression pattern between two weevils. Our systematic comparison of chemosensory genes in E. scrobiculatus and E. brandti provides a foundation for studies of olfaction and olfactory differentiation in the two weevils as well as a theoretical basis for studying species differentiation.

Keywords: coexistence, Eucryptorrhynchus scrobiculatus, Eucryptorrhynchus brandti, Ailanthus altissima, chemosensory genes, olfactory differentiation

## INTRODUCTION

fphys-09-01652 November 17, 2018 Time: 18:24 # 2

Over a long period of evolution, phytophagous insects and their hosts have formed a complete system of co-evolution. For two or more species living on the same host plant with a similar niche, competition over food resources is inevitable; only differentiation in time, space, or nutrition can reduce interspecific competition and enable coexistence. Ecologists generally believe that niche separation usually occurs in order to achieve coexistence for species with similar niche, and niche separation is the key to species coexistence (Caldwell and Vitt, 1999; Sedio and Ostling, 2013). Eucryptorrhynchus scrobiculatus Motschulsky and E. brandti (Harold) (Coleoptera: Curculionidae) are sympatric, closely related species native to China and are highly hostspecific, feeding on the tree of heaven, Ailanthus altissima (Mill.) Swingle, and its variant A. altissima var. Qiantouchun (Alonsozarazaga and Lyal, 1999; Yang et al., 2008; Herrick et al., 2012; Chao and Chen, 2015). The mixed cooccurrence of E. scrobiculatus and E. brandti results in extensive A. altissima deaths in the Ningxia Hui autonomous region (Hu et al., 2012; Yu et al., 2012). The coexistence of E. scrobiculatus and E. brandti can be explained by significant differentiation of trophic niches; E. scrobiculatus adults feed on 1-year-old branches, perennial branches, and petioles, while E. brandti adults feed on the stem of A. altissima (Ji et al., 2017). The role of olfaction in their feeding differentiation is unknown.

Olfaction plays a vital role in insect foraging for food resources (Leal, 2013). Antennae are the main olfactory organs in insects. Chemosensory genes in the antennae, such as genes encoding odorant-binding proteins (OBPs), chemosensory proteins (CSPs), odorant receptors (ORs), ionotropic receptors (IRs), gustatory receptors (GRs), sensory neuron membrane proteins (SNMPs), and odorant-degrading enzymes (ODEs), participate in the recognition of odor molecules (Vogt and Riddiford, 1981; Raming et al., 1993). A large number of experiments have shown that chemosensory genes in the antennae are involved in the chemical communication between insects and plants. For example, Swarup et al. (2011) found that responses to a specific odorant are frequently affected by the suppression of the expression of multiple OBPs by RNAi in Drosophila melanogaster. Wu et al. (2016) found that a reduction in Bdor83a-2 transcript abundance leads to a decrease in neuronal and behavioral responses to selected attractants, suggesting that Bdor83a-2 mediates behavioral responses to attractant semiochemicals. Previous studies have shown that gustation plays an important role in host selection (Hanson and Dethier, 1973; Boer and Hanson, 1987). GRs might be involved in gustatory (Jiang et al., 2015) and olfactory processes (Agnihotri et al., 2016).

Few studies have examined the mechanism of olfactory differentiation in closely related species; however, olfactionrelated gene expression divergence is linked to differences in host preference between closely related species. Based on different blood feeding behaviors, Yan (2014) identified ORs with significant differences in expression between Culex pipiens quinquefasciatus and Cx. molestus, and found that CquiOR5 is involved in the blood feeding behavior. Ramasamy et al. (2016) found that the evolution of olfactory genes is correlated with adaptation to new ecological niches by Drosophila suzukii and its close relative Drosophila biarmipes. They found that D. suzukii had a loss of function of ORs with affinities for volatiles produced during fermentation. They quantified the evolution of olfactory genes in Drosophila and revealed an array of genomic events that could be associated with the ecological adaptations of D. suzukii. Emeline et al. (2017) studied the differential expression of the OBP gene family in two closely related species of South American fruit flies, Anastrepha fraterculus and A. obliqua. They found eight OBP genes with differential expression between A. fraterculus and A. obliqua, suggesting that these genes have important roles in olfactory perception differences and accordingly are potentially related to species differentiation. Athrey et al. (2017) identified chemosensory gene families in olfactory organs of Anopheles coluzzii and A. quadriannulatus and inferred that divergence in OBP expression between the two species may be involved in differences in host preference.

Closely related species tend to feed on different host plants and to be polyphagous insects, with the exception of the closely related species Eucryptorrhynchus scrobiculatus and E. brandti. E. scrobiculatus and E. brandti feed on the same host but different parts of the host, and it is not clear whether chemosensory genes in the antennae affect their feeding differentiation. The role of olfaction in feeding differentiation is unknown. Yu et al. (2013) compared the antennal sensilla of both species to better understand their host-finding mechanism. In this study, we identified transcripts of OBPs, CSPs, ORs, IRs, GRs, and SNMPs in E. scrobiculatus and E. brandti antennae by high-throughput sequencing and investigated the expression patterns of OBP genes. Our results will be fundamental for studying the molecular mechanism of olfactory differentiation and provide a basis for understanding whether the differentiation of olfactory genes is related to feeding differences between the two species.

## MATERIALS AND METHODS

#### Ethics Statement

All of our experimental materials and methods are not contrary to ethics.

#### Insect Rearing and Antennae Collection

Adults of E. scrobiculatus and E. brandti were collected in Xiaoxingdun village, Pingluo County, Ningxia Hui Autonomous Region (38◦ 510 2400N, 106◦ 310 3800E) in May 2017. E. scrobiculatus were reared in nylon mesh bags (80 × 40 cm) with 1-year-old branches and perennial branches of A. altissima. E. brandti were reared in mesh bags (80 × 40 cm) with the stems (d = 4 cm) of A. altissima. The mesh bags containing E. scrobiculatus and E. brandti were placed in separate breathable cartons, and the cartons were immediately taken to the laboratory in Beijing. In the laboratory, the adults were immediately frozen in liquid nitrogen and stored at −80◦C until anatomical studies. The dissection was carried out on ice. The external genitalia were dissected to distinguish between males and females. The antennae were immediately cut from the bases of the heads of adults and

added to a 2 mL centrifuge tube. RNA extraction was performed immediately after the dissection.

#### Total RNA Extraction, cDNA Library Construction, and Illumina Sequencing

Fifteen and forty pairs of antennae of E. scrobiculatus and E. brandti were excised separately, and total RNA of female antennae (FA) and male antennae (MA) were extracted using the RNApure Total RNA Kit (Aidlab, Beijing, China). For each species and sex, data were obtained for three independent biological replicates, for a total of 12 samples. RNA was quantified using a NanoDrop 8000 (Thermo, Waltham, MA, United States). cDNA library construction and Illumina sequencing were performed at Bionova Biotechnology Co., Ltd. (Beijing, China). RNA quality was assessed by 1% agarose gel electrophoresis and analyzed using the 2100 Bioanalyzer (Agilent, Santa Clara, CA, United States). RNA was digested by DNase I to remove the DNA, and mRNA was enriched using oligo d(T). The mRNA was fragmented at a high temperature and reversetranscribed. The resultant cDNA was subjected to purification, end repair, A-tailing, adapter ligation, and PCR amplification. The quality and quantity of the library were then evaluated using the Bioanalyzer 2100 and ABI StepOnePlus Real-Time PCR system (Applied Biosystems, Forester City, CA, United States), respectively. The qualified library was then used for highthroughput sequencing. These libraries were pair-end sequenced using the PE150 strategy on the Illumina HiSeq X Ten platform.

## Assembly, Functional Annotation, and Quantitative Expression Analysis

Raw reads were pre-processed by filtered adapters. Low-quality reads, including reads containing > 10% N (uncertain bases) and those with a median quality value (Q) ≤ 25, were removed to generate clean reads for subsequent analyses. Transcriptome assembly of each clean-read dataset for FA and MA was accomplished using Trinity (version: v2012-10-05) (Grabherr et al., 2011), with min\_kmer\_cov = 2. The Trinity outputs were clustered by TGICL (TIGR Gene Indices clustering tools) (Pertea et al., 2003). Consequently, six transcript levels were obtained, including those for FA and MA of E. scrobiculatus and E. brandti and the final transcript datasets for E. scrobiculatus and E. brandti.

Transcripts were annotated using the Trinotate pipeline<sup>1</sup> . All putative genes were searched using BLASTX and BLASTP against databases, including the Swissprot-Uniprot database, KOG (euKaryotic Ortholog Groups), GO (Gene Ontology), eggNOG (evolutionary genealogy of genes: Non-supervised Orthologous Groups), and KEGG (Kyoto Encyclopedia of Genes and Genomes) (E-value cut-off, 1e-5). According to the annotation results obtained using the KOG database, the unigenes were classified into 26 groups. Then, GO classification was performed using Blast2GO (Conesa et al., 2005).

Transcript abundances from RNA-seq data were quantified using RSEM (Li and Dewey, 2011). Transcript abundances were

<sup>1</sup>https://trinotate.github.io/

calculated as the FPKM (fragments per kilobase per million mapped fragments) (Trapnell et al., 2010).

#### Identification of Chemosensory Genes

Candidate unigenes encoding OBPs, ORs, IRs, GRs, and SNMPs were found by keyword searches based on functional annotation results. Candidate unigenes encoding CSPs and pheromone-binding proteins (PBPs) were searched using BLASTX and tBLASTn according to downloaded sequences of Dendroctonus ponderosae CSPs (Andersson et al., 2013; Gu et al., 2015), Colaphellus bowringi CSPs (Li et al., 2015), and coleopteran PBPs against the local transcriptomes. All putative unigenes were confirmed by BLASTX searches against the NCBI non-redundant protein sequences database. Open reading frames (ORFs) of candidate genes were identified using ORF Finder and verified using tBLASTn in NCBI. The putative N-terminal signal peptides of candidate OBPs and CSPs were predicted using SignalP 4.1 server version with default parameters (Nielsen, 2017). The transmembrane domains of candidate ORs, IRs, GRs, and SNMPs were predicted using TMHMM server version 2.0 (Krogh et al., 2001).

#### Sequence and Phylogenetic Analysis

Amino acid sequences of candidate OBPs, CSPs, ORs, IRs, GRs, and SNMPs were aligned using ClustalX and further edited using GeneDoc. The phylogenetic trees of E. scrobiculatus and E. brandti chemosensory genes were constructed using the neighbor-joining method in MEGA 6.0 with default settings and 1000 bootstrap replicates. The dendrograms were color-coded and arranged using FigTree v1.4.3. The sequence identities of these chemosensory genes between E. scrobiculatus and E. brandti were determined using ClustalX and BLASTP.

#### Tissue- and Sex-Specific Expression of Candidate OBP Genes

The expression patterns of OBP genes in both female and male tissues (antennae, rostrum, leg, and head without the antennae and rostrum) of E. scrobiculatus and E. brandti were analyzed by RT-qPCR using a Bio-Rad CFX Connect PCR system (Bio-Rad, Hercules, CA, USA). Female and male antennae, rostrum, legs, and heads (without the antennae and rostrum) (female:male = 1:1) were collected from adult E. scrobiculatus and E. brandti. Total RNA was extracted using the RNApure Total RNA Kit (Aidlab, Beijing, China). The cDNA was synthesized from total RNA using the TRUEscript 1st Strand cDNA Synthesis Kit (Aidlab). Gene-specific primers were designed using Primer3Plus v2.4.2 (**Supplementary Table S9**). Each reaction was run in triplicate with three biological duplications, and PCRs with no template were used as controls. α-Tubulin and ribosomal protein (RPS11) were used as reference genes for E. scrobiculatus and E. brandti, respectively. Each RT-qPCR contained 10 µL of TB Green Premix Ex Taq II (Takara, Beijing, China), 1 µL of each primer, 2 µL of sample cDNA, and 6 µL of sterilized H2O. RT-qPCR

cycling parameters were 95◦C for 2 min, followed by 40 cycles of 95◦C for 5 s and 60◦C for 30 s. The melting curve was analyzed to evaluate the specificity of primers after each reaction, and the 2−11ct method was used to calculate the relative expression levels of OBP genes. The specificity of primers for each target gene was validated. RT-qPCR data were analyzed and plotted using GraphPad Prism 5.0 (GraphPad Software, La Jolla, CA, United States). The differences in the relative expression levels of OBPs were calculated using SPSS19.0 (SPSS Inc., Chicago, IL, United States) by a one-way nested analysis of variance (ANOVA), followed by Duncan's new multiple range test (α = 0.05).

## RESULTS

#### Antennal Transcriptome Sequencing and Sequence Assembly

We sequenced the transcriptomes of FA and MA of E. scrobiculatus and E. brandti with three independent biological replicates. We obtained approximately 48.78 (FA-1), 48.44 (FA-2), 48.44 (FA-3), 49.86 (MA-1), 48.63 (MA-2), and 49.50 (MA-3) million clean reads from E. scrobiculatus. These were assembled into 31757 unigenes for females and 32923 for males. A final transcript dataset (ES) with 46380 unigenes was obtained by TGICL, with a mean length of 3134 bp and N50 of 4850 bp (**Supplementary Table S1**). Similarly, we obtained approximately 41.02 (FA-1), 49.09 (FA-2), 49.33 (FA-3), 48.47 (MA-1), 49.38 (MA-2), and 49.22 (MA-3) million clean reads from E. brandti. These were assembled into 44720 unigenes for females and 37712 for males. A final transcript dataset (EB) with 56084 unigenes was obtained by TGICL, with a mean length of 2130 bp and N50 of 3396 bp (**Supplementary Table S1**). The datasets of transcriptomes during the current study have been uploaded to the NCBI SRA database (accession number: SRP155112).

## Functional Annotation

To acquire more comprehensive sequence information, all six transcript sets were annotated. In total, 29631 (63.9%) and 26487 (47.2%) unigenes from E. scrobiculatus and E. brandti were annotated by BLASTX, respectively. For both E. scrobiculatus and E. brandti, the number of annotated unigenes in the final transcript dataset (ES and EB) was significantly higher than that of female antennal transcripts (ESF and EBF) and male antennal transcripts (ESM and EBM) (**Supplementary Table S2**).

We classified 28725 (61.9%) unigenes in E. scrobiculatus and 22839 (40.7%) in E. brandti into 26 KOG protein groups. The four largest groups were general function prediction only, signal transduction mechanisms, function unknown, and posttranslational modification/protein turnover/chaperones. Very few unigenes were assigned to the nuclear structure and unnamed protein group (**Supplementary Figure S1**).

Gene ontology annotation was used to classify unigenes into different functional categories. Of the 46380 unigenes in E. scrobiculatus, 30714 (66.2%) could be annotated based on sequence similarity. We assigned 27666 of the 56084 E. brandti antennal unigenes (49.3%) to specific GO terms. The distributions of GO terms in the three major categories were similar in the two species. In the biological processes category, "cellular process" and "metabolic process" were most highly represented. In the cellular components category, the most abundant GO terms were "cell," "cell part," and "organelle." In the molecular function category, "binding" and "catalytic activity" were the most highly represented (**Supplementary Figure S2**).

## Candidate OBPs in E. scrobiculatus and E. brandti

Based on our analysis of the antennal transcriptomes in the two species, we identified 31 and 28 candidate OBP genes in the female antennal, male antennal, and combined female and male datasets for E. scrobiculatus and E. brandti, respectively. All but one transcript (EscrOBP31) had complete ORFs, and 27

transcripts of EscrOBPs and 23 transcripts of EbraOBPs included predicted signal peptide sequences. Detailed information was reported in **Supplementary Table S3**.

Odorant-binding proteins with FPKM values of ≥ 500 were defined as highly expressed genes, those with values 100–500 were defined as moderately expressed genes, and those with values of ≤ 100 were defined as weakly expressed genes. Most OBPs were weakly expressed, but a few OBPs were highly expressed in antennae, and more genes were highly expressed in E. scrobiculatus than in E. brandti (**Figure 1A** and **Supplementary Table S3**).

A phylogenetic tree was built using the newly obtained sequences and those from Diptera and Coleoptera. Among EscrOBPs, 13 showed the classic motif of six conserved cysteines, 15 were Minus-C, and 3 were undefined owing to less conserved cysteines. For EbraOBPs, we found 13 classic and 13 Minus-C OBPs; two were undefined (**Figure 2** and **Supplementary Figure S3**). Remarkably, EscrOBP8, EscrOBP24 and EbraOBP8, EbraOBP24 formed a cluster with other coleopteran PBPs, with sequence similarities of 99 and 86%, respectively. We inferred that EscrOBP8 (EscrPBP1), EscrOBP24 (EscrPBP2) and EbraOBP8 (EbraPBP1), EbraOBP24 (EbraPBP2) were putative PBPs from E. scrobiculatus and E. brandti, respectively. Most OBPs clustered with other coleopteran OBPs, except for EscrOBP19 and EscrOBP29. In the phylogenetic tree, we detected 20 OBP pairs in E. scrobiculatus and E. brandti with high homology (**Figure 2**). We evaluated the sequence identity between EscrOBPs and EbraOBPs by ClustalX and BLASTP and found that 18 OBP orthologous pairs shared amino acid identities of ≥ 90% between E. scrobiculatus and E. brandti (**Supplementary Table S8**).

## Candidate CSPs in E. scrobiculatus and E. brandti

We identified 11 different transcripts encoding candidate CSPs in E. scrobiculatus and E. brandti. All but one transcript (EbraCSP7) included full-length ORFs, and 9 EscrCSPs and 10 EbraCSPs

had predicted signal peptide sequences (**Supplementary Table S4**). All of the identified amino acid sequences possessed the highly conserved four-cysteine profile (**Supplementary Figure S4**). Most CSPs exhibited low expression levels; two EscrCSPs and three EbraCSPs were highly expressed in antennae (**Figure 1B** and **Supplementary Table S4**).

A phylogenetic tree was built using all of these CSPs and those of Lepidoptera and Coleoptera. EscrCSPs and EbraCSPs clustered with other coleopteran CSPs, and no specific CSP lineages were evident (**Supplementary Figure S5**). We detected 10 CSP orthologous pairs sharing amino acid similarities of ≥ 90% between E. scrobiculatus and E. brandti (**Supplementary Table S8**).

## Candidate ORs in E. scrobiculatus and E. brandti

We identified 49 transcripts for putative ORs in E. scrobiculatus and 45 in E. brandti. Of these, 42 EscrORs and 36 EbraORs contained complete ORFs encoding proteins of more than 300 amino acids, with 1–8 transmembrane domains. Furthermore, 16 EscrORs and 10 EbraORs encoded seventransmembrane-domain proteins. In comparisons with OR genes from other insect species by BLASTX, we found that all putative EscrORs shared identities of 26 and 92% with other ORs, with almost identical values (27–92%) for EbraORs. Detailed information is reported in **Supplementary Table S5**. Expression levels of ORs in E. scrobiculatus and E. brandti were similar; most EscrORs and EbraORs were weakly expressed (FPKM ≤ 10) in antennae. Note that EbraOR24 (EbraOrco) was highly expressed while EscrOR24 (EscrOrco) was weakly expressed (**Figure 1C** and **Supplementary Table S5**).

We performed a phylogenetic analysis using our candidate ORs and those from Lepidoptera, Diptera, and Coleoptera. EscrOR24 and EbraOR24 clustered with DmelOR83b, the highly conserved co-receptor Orco (Larsson et al., 2004; Hallem et al., 2006). Sequence identity between EscrOR24 and

EbraOR24 was very high (98%). No candidate ORs clustered with DmelOR67d, the pheromone receptors (PRs) from D. melanogaster, and other PRs. Within these OR sequences, we found a species-specific clade including 10 EscrORs (EscrOR13, 14, 15, 27, 28, 29, 33, 45, 46, and 47) and 8 EbraORs (EbraOR13, 14, 15, 27, 28, 29, 33, and 43) sharing low homology with other coleopteran ORs (**Figure 3**). These genes may be related to the detection of the characteristic volatile of A. altissima.

We compared the amino acid sequences of EscrORs and EbraORs. Sequence similarities of the 39 pairs of homologous ORs were greater than 70%. In addition, 25 OR orthologous pairs shared amino acid identities of ≥ 90% between E. scrobiculatus and E. brandti. The higher sequence similarity in homologous ORs indicated they may be involved in olfactory recognition of A. altissima (**Supplementary Table S8**).

## Candidate IRs in E. scrobiculatus and E. brandti

We identified 17 candidate IRs in E. scrobiculatus and 25 in E. brandti. All but three transcripts (EscrIR2, 9 and EbraIR2) contained full-length ORFs with one to five transmembrane domains (**Supplementary Table S6**). Expression levels of IRs were the same as the levels of ORs in E. scrobiculatus and E. brandti. Two EscrIRs and three EbraIRs were moderately expressed, EbraIR6 was highly expressed (**Figure 1D** and **Supplementary Table S6**).

To further infer the function of IR genes, a phylogenetic tree was constructed using these sequences and homologous sequences in Lepidoptera, Diptera, and Coleoptera. For EscrIRs, seven EscrIRs (EscrIR1, 2, 3, 15, 4, 7, and 10) clustered with the presumed "antennal" orthologues IR93a, 40a, 64a, 21a, 41a, and 31a. EscrIR5 and EscrIR6 were distributed in the IR8a and IR25a groups, which are coreceptors (Abuin et al., 2011). EscrIR8, 9, and 14 clustered with the non-NMDA iGluRs group (Croset et al., 2010). For EbraIRs, eight EbraIRs (EbraIR1, 2, 3, 15, 4, 7, 10, 18) clustered with the presumed "antennal" orthologues IR93a, 40a, 64a, 21a, 41a, 31a, and 68a. EbraIR8, 9, 14, 21, 22, and 24 formed a group with non-NMDA iGluRs. Notably, the conserved "antennal" orthologues IR60a and IR76b were lacking in the E. scrobiculatus and

E. brandti transcriptomes, while IR68a was only absent from E. scrobiculatus. Notably, EscrIR16, EbraIR19, and EbraIR20 were divergent compared with other DmelIRs sharing low homology (**Figure 4**). Fifteen IR orthologous pairs shared amino acid similarities of > 85% between E. scrobiculatus and E. brandti. The similarity of nine pairs of IRs exceeded 95%. These IRs may play a key role in host recognition (**Supplementary Table S8**).

## Candidate GRs in E. scrobiculatus and E. brandti

We identified 19 candidate GRs from transcript datasets of E. scrobiculatus, and 18 transcripts contained complete ORFs. Similarly, we identified 17 candidate GRs in E. brandti, and 11 transcripts contained complete ORFs (**Supplementary Table S7**). Expression levels of GRs were generally lower than those of other chemosensory genes. GRs with FPKM values ≤ 1 were defined as having low expression, those with values of 1–10 were defined as moderately expressed, and those with values ≥ 10 were defined as highly expressed. The majority of GR genes were weakly and moderately expressed in E. scrobiculatus and E. brandti; only one EscrGR (EscrGR4) was highly expressed (**Figure 1E** and **Supplementary Table S7**).

All of these protein sequences and the sequences from five additional insect species were used to construct a phylogenetic tree. We found that EscrGR8, Escr13, EbraGR13, and EbraGR16 were members of the sugar-receptor subfamily, and EscrGR1, EscrGR7, and EbraGR1 were assigned to the CO2-receptor subfamily, indicating that these GRs might be related to the detection of CO<sup>2</sup> (Jones et al., 2007; Kwon et al., 2007) and sugar (Dahanukar et al., 2007; Jiao et al., 2008; Sato et al., 2011). Furthermore, we found a species-specific clade including four GRs from E. scrobiculatus (EscrGR3, 4, 5, and 12) and five from E. brandti (EbraGR3, 4, 5, 8, and 12) that shared low homology with other coleopteran GRs (**Figure 5**). The amino acid sequences of E. scrobiculatus and E. brandti for five pairs of homologous GRs had similarities of more than 90% (**Supplementary Table S8**).

## Candidate SNMPs in E. scrobiculatus and E. brandti

We identified three transcripts encoding candidate SNMPs in E. scrobiculatus and E. brandti (**Supplementary Table S7**).

rostrum cut off). The bar represents standard error and the different small letters (a–e) above each bar indicate significant differences (P < 0.05). N/A indicates that the transcript level is too low to measure.

Two EscrSNMPs were weakly expressed (FPKM ≤ 10), one EbraSNMP was moderately expressed (FPKM, 10–100), and one EscrSNMP and two EbraSNMPs were highly expressed (FPKM ≥ 100) (**Figure 1F** and **Supplementary Table S7**).

Based on a phylogenetic analysis, EscrSNMP1, EscrSNMP3, EbraSNMP1, and EbraSNMP3 were very similar to DmelSNMP1, which encodes a protein required for correct pheromone detection (Rogers et al., 2001; Benton et al., 2007; Jin et al., 2008). EscrSNMP2 and EbraSNMP2 were similar to DmelSNMP2, which is expressed in supporting cells (Nichols and Vogt, 2008; Forstner et al., 2008; Gu et al., 2013) (**Supplementary Figure S6**). Three SNMP orthologous pairs shared amino acid identities of > 90% between E. scrobiculatus and E. brandti (**Supplementary Table S8**).

#### Tissue- and Sex-Specific Expression of Candidate E. scrobiculatus and E. brandti OBP Genes

Expression patterns of 25 OBPs in female and male antennae, rostrum, legs, and heads (excluding the antennae and rostrum) from E. scrobiculatus and E. brandti were determined by RTqPCR. All OBPs were expressed in antennae in both species. In E. scrobiculatus, we detected high expression of 11 putative OBP genes (EscrOBP2, 4, 5, 6, 7, 10, 15, 19, 21, 28, and 30) in the antennae. EscrOBP5 and EscrOBP30 were significantly femalebiased, and the antennal expression of eight OBPs (EscrOBP2, 4, 6, 7, 10, 15, 21, and 28) was significantly male-biased. Furthermore, we detected significantly higher expression of four OBPs (EscrOBP3, 11, 20, and 22) in the rostrum than in other tissues, and we detected significantly greater expression of four OBPs (EscrOBP8, 14, 17, and 24) in the head. Interestingly, EscrPBP1 (EscrOBP8) and EscrPBP2 (EscrOBP24) were more highly expressed in the head than in the antennae. In addition, we observed higher levels of four OBPs (EscrOBP1, 16, 18, and 23) in the antennae and rostrum than in other tissues, EscrOBP13 was highly expressed in all tissues except for the leg, and EscrOBP26 showed higher expression in the leg than in other tissues (**Figure 6**). In E. brandti, we detected significantly higher expression of 14 putative OBPs (EbraOBP2, 4, 5, 8, 9, 10, 15, 18, 19, 23, 24, 25, 26, and 28) in the antennae than in other tissues. Antennal expression levels of EbraOBP5, EbraOBP23 and

the transcript level is too low to measure.

EbraOBP25 were female-biased, and the expression of nine OBPs (EbraOBP2, 4, 9, 15, 18, 19, 24, 26, and 28) was male-biased. We observed significantly higher levels of five OBPs (EbraOBP3, 6, 12, 16, and 21) in the rostrum than in other tissues and significantly higher expression of four OBPs (EbraOBP13, 14, 17, and 22) in the head. Four OBPs (EbraOBP1, 9, 21, and 26) showed higher expression in the antennae and rostrum than in other tissues (**Figure 7**).

#### DISCUSSION

The weevils E. scrobiculatus and E. brandti are sympatric and closely related, feeding on the same host (A. altissima) but different parts. It is not clear whether olfaction plays an important role in feeding differentiation. We analyzed the antennal transcriptomes of E. scrobiculatus and E. brandti and searched for chemosensory genes to evaluate interspecific differences in olfactory genes.

In this study, we sequenced female and male antennal transcriptomes of E. scrobiculatus and E. brandti using the Illumina HiSeq X Ten platform, assembled reads using Trinity, and performed a clustering analysis using TGICL. We acquired and annotated female and male antennal transcripts and final transcripts of E. scrobiculatus and E. brandti. We detected more unigenes in E. brandti (56084) than in E. scrobiculatus (46380), but the mean length and N50 of unigenes in E. brandti were lower than those in E. scrobiculatus. Additionally, the number of annotated unigenes in E. brandti (47.2%) was much less than that in E. scrobiculatus (63.9%). These results suggest that the E. brandti genome contains more species-specific genes than the E. scrobiculatus genome. We searched six annotated databases and identified 130 putative chemosensory genes (31 OBPs, 11 CSPs, 49 ORs, 17 IRs, 19 GRs, and 3 SNMPs) in E. scrobiculatus and 129 (28 OBPs, 11 CSPs, 45 ORs, 25 IRs, 17 GRs, and 3 SNMPs) in E. brandti, fewer than the number of chemosensory genes identified in Rhynchophorus ferrugineus (Antony et al., 2016) and more than those in Dendroctonus valens (Gu et al., 2015), Ips typographus and Dendroctonus ponderosae (Andersson et al., 2013), and Tomicus yunnanensis (Liu et al., 2018) (**Table 1**).


TABLE 1 | The number of odorant binding protein (OBP), chemosensory protein (CSP), odorant receptor (OR), ionotropic receptor (IR), gustatory receptor (GR), and sensory neuron membrane protein (SNMP) in Curculionoidea.

We detected more candidate OBPs in the two weevils (31 in E. scrobiculatus and 28 in E. brandti) than previously reported in D. valens (21 OBPs) (Gu et al., 2015), I. typographus (15 OBPs) (Andersson et al., 2013), and L. oryzophilus (10 OBPs) (Yuan et al., 2016), but fewer OBPs than in R. ferrugineus (38 OBPs) (Antony et al., 2016) and Tribolium castaneum (49 OBPs) (Dippel et al., 2014). Weevils and D. ponderosae had similar numbers of OBPs (31 OBPs) (Andersson et al., 2013). Similar results were obtained for CSPs (**Table 1**). Chemosensory receptors play a critical role in the reception of chemicals from the environment and the regulation of insect behaviors. However, there are few known receptors in Curculionoidea, including ORs, IRs, and GRs, especially IRs and GRs. In this study, we identified 49 transcripts for putative ORs in E. scrobiculatus and 45 in E. brandti, compared with 22 ORs in D. valens (Gu et al., 2015), 43 in I. typographus (43 ORs), and 77 in R. ferrugineus (Antony et al., 2016). There were substantially more IRs and GRs in weevils than in D. valens (3 IRs and 4 GRs) (Gu et al., 2015), R. ferrugineus (10 IRs and 15 GRs) (Antony et al., 2016), I. typographus (7 IRs and 6 GRs), and D. ponderosae (15 IRs and 2 GRs) (Andersson et al., 2013). These differences could be due to differences in sample preparation and sequencing methods or the evolution of divergent physiological behaviors in distinct environments (Goldman-Huertas et al., 2015).

Despite increasing research on insect olfaction mechanisms, little is known about chemoreception in coleopterans, especially Curculionidae, compared with Lepidoptera and Diptera. Owing to the limited functional information for coleopterans, we inferred the physiological functions of chemosensory genes of E. scrobiculatus and E. brandti using a phylogenetic approach. Most OBPs in E. scrobiculatus and E. brandti clustered with those of other coleopterans with high homology, except for EscrOBP19 and EscrOBP29, which need to be further studied. All candidate EscrCSPs and EbraCSPs shared high sequence identity with CSPs of other coleopteran insects, and no species-specific CSP was found. We identified speciesspecific OR transcripts in E. scrobiculatus and E. brandti, and these may play important roles in recognizing specific volatiles of A. altissima. We found "antennal IR" orthologues in E. scrobiculatus and E. brandti, such as IR21a, IR31a, IR40a, IR41a, IR93a, IR64a, and co-receptor IR8a/IR25a. Moreover, we detected "divergent IR" orthologues in E. brandti, EbraIR19, and EbraIR20, which may act as GRs in distinct taste organs and stages of E. brandti (Croset et al., 2010). We detected more non-NMDA iGluRs in E. brandti than in E. scrobiculatus. Neither species had NMDA iGluRs, which are related to fast excitatory synaptic transmission in vertebrates and invertebrates (Littleton and Ganetzky, 2000; Tikhonov and Magazanik, 2009). The effects of non-NMDA iGluRs on E. scrobiculatus and E. brandti need to be further studied. Despite similar numbers of candidate GRs in E. scrobiculatus and E. brandti, their amino acid identities were lower than those for other chemosensory genes, and this could explain their different feeding habits. In the future, we intend to explore the expression pattern and function of these GRs in the two weevils, which will be helpful to study their feeding differentiation.

Based on the expression levels of chemosensory genes in E. scrobiculatus and E. brandti and their phylogenetic analysis, we found some differential expressed genes. Orco is highly conserved in insects, while the expression level of putative Orco in E. scrobiculatus and E. brandti (EscrOR24 and EbraOR24) was quite different. EbraOR24 (EbraOrco) was highly expressed while EscrOR24 (EscrOrco) was weakly expressed in antennae. The number of putative IRs in E. brandti was more than that in E. scrobiculatus, so is the highly-expressed genes. EscrIR1 and EscrIR6 were moderately expressed, while EbraIR5, EbraIR17, and EbraIR25 were moderately expressed, and EbraIR6 was highly expressed in antennae. EscrIR6, EbraIR5, and EbraIR6 clustered with the IR8a/IR25a group. EscrGR4 and EbraGR4 were species-specific GRs in phylogenetic analysis, EscrGR4 was highly expressed in antennae, while EbraGR4 was weakly expressed. These differential expressed genes may play the role in olfactory differentiation of two weevils, their functions need to be further studied in the future.

We investigated the expression profile of OBPs in the two weevils by RT-qPCR. Some OBPs with high amino acid sequence similarities exhibited similar expression in various tissues of E. scrobiculatus and E. brandti, including EscrOBP1, EscrOBP2, and EscrOBP3 and EbraOBP1, EbraOBP2, and EbraOBP3, indicating that these OBPs may be involved in the detection of the same host odors. Some OBPs with high amino acid sequence similarity exhibited expression differences in various tissues between E. scrobiculatus and E. brandti. For example, EscrOBP13 was highly expressed in the antennae, rostrum, and head, while EbraOBP13 was highly expressed in the head. Notably, EscrOBP8 and EscrOBP24 were highly expressed in the head and EscrOBP24 was female-biased, while EbraOBP8 and EbraOBP24 were highly expressed in the antennae and EbraOBP24 was male-biased, suggesting that these genes could have different binding affinities for pheromone compounds (Plettner et al., 2000).

Growing evidence indicates that chemosensory genes play key roles in host specialization in insects (Visser, 1986; Whiteman and Pierce, 2008; Schymura et al., 2010; Eyres et al., 2017). Of these genes, OBPs and CSPs are small, highly conserved families, mainly involved in ligand binding to receptors. By contrast, ORs and GRs are large and rapidly evolving gene families. Many studies have emphasized the roles of chemoreceptors in differences between host-associated species (Hallem et al., 2006; McBride, 2007; Smadja et al., 2012; McBride et al., 2014). For example, Eyres et al. (2017) confirmed that differences in chemosensory genes were important for the divergence of pea aphid races, especially GRs and ORs. Interestingly, most candidate OBPs and CSPs in E. scrobiculatus and E. brandti clustered with other coleopteran genes, and no species-specific clade was found, indicating that these genes were conserved. We identified species-specific clades of ORs and GRs in E. scrobiculatus and E. brandti, which might correspond to the odor of the specific host, providing evidence for olfactory differentiation in weevils. In future research, we plan to further study the function of ORs and GRs in E. scrobiculatus and E. brandti to explore the reason for their feeding differentiation.

In this study, we identified and compared putative chemosensory genes in antennae of E. scrobiculatus and

#### REFERENCES


E. brandti. Sequence identity between E. scrobiculatus and E. brandti was > 90% for more than half of the genes. These genes were likely to be related to the specific feeding on A. altissima. We also found species-specific genes in E. scrobiculatus and E. brandti; these genes might play a critical role in olfactory and feeding differentiation. These data provide a foundation for studying the molecular mechanism of olfaction and olfactory differentiation in weevils as well as a theoretical basis for the differentiation of the closely related species.

#### AUTHOR CONTRIBUTIONS

XW and JW conceived and designed the experiments. XW, QW, and PG performed the experiments. XW and QW analyzed the data. XW wrote the manuscript. All authors reviewed the final manuscript and approved the submitted version.

## FUNDING

This work was supported by National Natural Sciences Foundation of China (Grant No. 31770691), National Key R&D Program of China (2018YFC1200400), and Beijing's Science and Technology Planning Project (Z171100001417005).

#### ACKNOWLEDGMENTS

We thank Kailang Yang and Shibei Tan (Beijing Forestry University, China) for help with insect collection. We also thank Bing Guo and Ningning Fu (Beijing Forestry University, China) for RT-qPCR experimental guidance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01652/full#supplementary-material




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wen, Wang, Gao and Wen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identification of Chemosensory Genes Based on the Transcriptomic Analysis of Six Different Chemosensory Organs in Spodoptera exigua

#### Edited by:

*Peng He, Guizhou University, China*

#### Reviewed by:

*Dingze Mang, Tokyo University of Agriculture and Technology, Japan Yihan Xia, Lund University, Sweden Ramesh Kumar Dhandapani, University of Kentucky, United States*

#### \*Correspondence:

*Ya-Nan Zhang ynzhang\_insect@163.com Tao Xue xuetao\_26@163.com Liang Sun liangsun@tricaas.com*

#### Specialty section:

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

Received: *27 January 2018* Accepted: *06 April 2018* Published: *24 April 2018*

#### Citation:

*Zhang Y-N, Qian J-L, Xu J-W, Zhu X-Y, Li M-Y, Xu X-X, Liu C-X, Xue T and Sun L (2018) Identification of Chemosensory Genes Based on the Transcriptomic Analysis of Six Different Chemosensory Organs in Spodoptera exigua. Front. Physiol. 9:432. doi: 10.3389/fphys.2018.00432* Ya-Nan Zhang<sup>1</sup> \*, Jia-Li Qian<sup>1</sup> , Ji-Wei Xu<sup>1</sup> , Xiu-Yun Zhu<sup>1</sup> , Meng-Ya Li <sup>1</sup> , Xiao-Xue Xu<sup>1</sup> , Chun-Xiang Liu<sup>1</sup> , Tao Xue<sup>1</sup> \* and Liang Sun<sup>2</sup> \*

*<sup>1</sup> Department of Biological Sciences, College of Life Sciences, Huaibei Normal University, Huaibei, China, <sup>2</sup> Key Laboratory of Tea Quality and Safety Control, Ministry of Agriculture, Tea Research Institute, Chinese Academy of Agricultural Sciences, Hangzhou, China*

Insects have a complex chemosensory system that accurately perceives external chemicals and plays a pivotal role in many insect life activities. Thus, the study of the chemosensory mechanism has become an important research topic in entomology. *Spodoptera exigua* Hübner (Lepidoptera: Noctuidae) is a major agricultural polyphagous pest that causes significant agricultural economic losses worldwide. However, except for a few genes that have been discovered, its olfactory and gustatory mechanisms remain uncertain. In the present study, we acquired 144,479 unigenes of *S. exigua* by assembling 65.81 giga base reads from 6 chemosensory organs (female and male antennae, female and male proboscises, and female and male labial palps), and identified many differentially expressed genes in the gustatory and olfactory organs. Analysis of the transcriptome data obtained 159 putative chemosensory genes, including 24 odorant binding proteins (OBPs; 3 were new), 19 chemosensory proteins (4 were new), 64 odorant receptors (57 were new), 22 ionotropic receptors (16 were new), and 30 new gustatory receptors. Phylogenetic analyses of all genes and SexiGRs expression patterns using quantitative real-time polymerase chain reactions were investigated. Our results found that several of these genes had differential expression features in the olfactory organs compared to the gustatory organs that might play crucial roles in the chemosensory system of *S. exigua*, and could be utilized as targets for future functional studies to assist in the interpretation of the molecular mechanism of the system. They could also be used for developing novel behavioral disturbance agents to control the population of the moths in the future.

Keywords: Spodoptera exigua, olfactory organ, gustatory organ, transcriptome analysis, chemosensory gene

## INTRODUCTION

Over the evolutionary process, insects have developed a complex chemosensory system that can accurately perceive external chemicals. The system plays a pivotal role in many insect life activities, such as feeding, mating, host finding, searching for oviposition sites, avoiding predators, and migration (Field et al., 2000; Zhan et al., 2011; Suh et al., 2014; Sun et al., 2014; Zhang et al., 2015a). Numerous studies based on morphological and molecular biology have revealed that the antenna, proboscis, and labial palp are the main olfactory and gustatory organs in this system (Jacquin-Joly and Merlin, 2004; Briscoe et al., 2013; Sun et al., 2017).

The insect chemosensory system involves several different types of genes, including (1) soluble olfactory proteins in the lymph of chemosensilla, e.g., odorant binding proteins (OBPs) (Vogt, 2003; Xu et al., 2009; Zhou, 2010; Pelosi et al., 2018) and chemosensory proteins (CSPs) (Pelosi et al., 2005, 2006; Iovinella et al., 2013) that transfer chemicals via the chemosensilla lymph to corresponding chemosensory receptors, and (2) chemosensory membrane proteins, e.g., olfactory receptors (ORs) (Crasto, 2013; Leal, 2013; Zhang et al., 2015a, 2017), ionotropic receptors (IRs) (Vogt, 2003; Benton et al., 2009; Rytz et al., 2013), and gustatory receptors (GRs) (Clyne et al., 2000; Zhang et al., 2011; Briscoe et al., 2013; Ni et al., 2013) that are located on the dendrites of neurons in the chemosensilla and transform chemical signals into electrical signals to stimulate the corresponding behavioral responses of insects (Leal, 2013).

The acquisition, bioinformatics analysis, and expression pattern of putative chemosensory genes are the crucial steps to explore the exact roles of several key genes in the insect chemosensory process. The development of modern molecular biology techniques and experimental equipment, such as high-throughput sequencing, has created more efficient, inexpensive, and higher accuracy technologies than what has been traditionally utilized (McKenna et al., 1994; Picimbon and Gadenne, 2002; Xiu et al., 2008; Liu et al., 2012). These have been successfully applied in the identification of insect chemosensory genes, including many moth species, such as Spodoptera littoralis (Legeai et al., 2011), Sesamia inferens (Zhang et al., 2013), Helicoverpa armigera (Liu et al., 2014b), Plutella xylostella (Yang et al., 2017), and Ectropis grisescens (Li et al., 2017).

The beet armyworm, Spodoptera exigua Hübner (Lepidoptera: Noctuidae), is a major agricultural polyphagous pest that causes significant economic losses to many crops worldwide (Xiu and Dong, 2007; Acín et al., 2010; Lai and Su, 2011). To date, only partial chemosensory genes of S. exigua have been identified, including several OBPs (Xiu and Dong, 2007; Zhu et al., 2013; Liu et al., 2015b), CSPs (Liu et al., 2015b) and a few chemosensory receptor genes (Liu et al., 2013, 2014a, 2015b). This is much lower than other moth species from which chemosensory genes have been obtained from transcriptomic data of chemosensory organs. These limited gene resources impede our interpretation of the chemosensory molecular mechanism of S. exigua. To obtain greater olfactory and gustatory gene resources, we utilized the six major olfactory and gustatory organs (female antennae: FA, male antennae: MA, female proboscises: FPr, male proboscises: MPr, female labial palps: FLP, and male labial palps: MLP) of S. exigua adults in the present study. We first built a genetic database of genes that were expressed in the six chemosensory organs of S. exigua using an Illumina HiSeqTM 4000 sequencing platform and completely identified 159 genes (110 genes were newly obtained) as being potentially involved in the chemosensory system. To postulate the functions of these identified genes, we performed phylogenetic analyses of all genes and investigated SexiGRs expression patterns using quantitative real-time polymerase chain reaction (qPCR). Our results showed that several of the genes had differential expression in olfactory organs compared to gustatory organs that might play different and crucial roles in the chemosensory system of S. exigua, and could be utilized as targets for future functional studies (using the heterologous expression system of Xenopus oocytes or Escherichia coli in vitro and with genetic modification by the CRISPR/Cas9 editing system in vivo) to assist in the interpretation of the molecular mechanism of the system.

## MATERIALS AND METHODS

#### Insects Rearing and Tissue Collection

S. exigua larvae were purchased from Keyun Biology Company in Henan province, China. As we previous studies (Zhang et al., 2017a), we used same rearing conditions and methods to rear the insect. For transcriptome sequencing, 200 female antennae (FA), 200 male antennae (MA), 300 female proboscises (FPr), 300 male proboscises (MPr), 300 female labial palps (FLP), 300 male labial palps (MLP), 30 female abdomen (FAb), and 30 male abdomen (MAb) were collected from 3-day-old unmated adults. For the tissue distribution analysis, 100 FA, 100 MA, 200 FLP, 200 MLP, 200 FP, and 200 MP for each replicate experiment were collected under the same conditions. All these organs were immediately frozen in liquid nitrogen and stored at −80◦C until use.

#### cDNA Library Preparation, Clustering, and Sequencing

Sample total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). cDNA library preparation and Illumina sequencing were carried out by Novogene Bioinformatics Technology Co., Ltd. (Beijing, China). The 1.5 µg total RNA per sample was used as input material for the RNA sample preparations, and sequencing libraries were generated using NEBNext <sup>R</sup> UltraTM RNA Library Prep Kit for Illumina <sup>R</sup> (NEB, USA) following manufacturer's recommendations and index codes were added to attribute sequences to each sample. Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature in NEBNext First Strand Synthesis Reaction Buffer (5X). First strand cDNA was synthesized using random hexamer primer and M-MuLV Reverse Transcriptase (RNase H-) (NEB, USA). Second strand cDNA synthesis was subsequently performed using DNA Polymerase I (NEB, USA) and RNase H (NEB, USA). Remaining overhangs were converted into blunt ends via exonuclease/polymerase activities. After adenylation of 3′ ends of DNA fragments, NEBNext Adaptor with hairpin loop structure were ligated to prepare for hybridization. In order to select cDNA fragments of preferentially 150∼200 bp in length, the library fragments were purified with AMPure XP system (Beckman Coulter, Beverly, USA). Then 3 µL USER Enzyme (NEB, USA) was used with size-selected, adaptor-ligated cDNA at 37◦C for 15 min followed by 5 min at 95◦C before PCR. Then PCR was performed with Phusion High-Fidelity DNA polymerase, Universal PCR primers, and Index (X) Primer. At last, PCR products were purified (AMPure XP system) and library quality was assessed on the Agilent Bioanalyzer 2100 system.

The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumina, San Diego, CA, USA) according to the manufacturer's instructions. After cluster generation, the library preparations were sequenced on an Illumina HiseqTM 4000 platform and paired-end reads were generated.

#### Transcriptome Assembly and Gene Functional Annotation

Transcriptome assembly was accomplished based on the reads using Trinity (r20140413p1) (Li et al., 2010; Grabherr et al., 2011) with min\_kmer\_cov set to 2 by default and all other parameters set default. The assembly sequences of Trinity were deemed to be unigenes. Unigene function was annotated based on the following databases: Nr (NCBI non-redundant protein sequences) (https://www.ncbi.nlm.nih.gov/genbank/ and https://www.ncbi.nlm.nih.gov/protein/), Pfam (Protein family) (https://pfam.sanger.ac.uk/), KOG/COG (Clusters of Orthologous Groups of proteins) (https://www.ncbi.nlm.nih. gov/COG/), Swiss-Prot (A manually annotated and reviewed protein sequence database) (http://www.ebi.ac.uk/uniprot/), KO (KEGG Ortholog database) (http://www.genome.jp/kegg/) and GO (Gene Ontology) (http://www.geneontology.org/).

#### Differential Expression Analysis

Firstly, the read counts were adjusted by edgeR 3.0.8 program package through one scaling normalized factor for each sequenced library. Then, the differential expression analysis of two samples was performed using the DEGseq 1.12.0 R package (Wang et al., 2010). P-value was adjusted using q-value (Storey, 2003). q < 0.005 & |log2(foldchange)|>1 was set as the threshold for significantly differential expression.

#### RNA Isolation and cDNA Synthesis

Total RNA was extracted using the MiniBEST Universal RNA Extraction Kit (TaKaRa, Dalian, China), following the manufacturer's instructions, in which we used DNase I to digest sample DNase to avoid genomic DNA contamination. The RNA quality was assessed spectrophotometrically (Biofuture MD2000D, UK). Single-stranded cDNA templates were synthesized from 1 µg total RNA obtained from various tissue samples using the PrimeScriptTM RT Master Mix (TaKaRa, Dalian, China) according to the manufacturers' instructions.

#### Sequence and Phylogenetic Analysis

The ORFs of the chemosensory genes were predicted by using ORF Finder (http://www.ncbi.nlm.nih.gov/gorf/gorf.html), and the similarity searches of genes were performed by using the NCBI-BLAST Server (http://blast.ncbi.nlm.nih.gov/). Putative N-terminal signal peptides (SP) of SexiOBPs and SexiCSPs were predicted by SignalP 4.1 (http://www.cbs.dtu.dk/services/ SignalP/) (Petersen et al., 2011). Transmembrane domains (TMD) of SexiORs, SexiGRs, and SexiIRs were predicted by TMHMM Server Version 2.0 (Krogh et al., 2001) (http://www. cbs.dtu.dk/services/TMHMM).

Phylogenetic trees were constructed for the analysis of five family chemosensory genes of S. exigua, based on gene sequences of S. exigua and those of other insects. The OBP data set contained 24 sequences from S. exigua (Table S1), and 90 from other species, including B. mori (Gong et al., 2009), M. sexta (Grosse-Wilde et al., 2011), and A. lepigone (Zhang et al., 2017b). The CSP data set contained 19 sequences from S. exigua (Table S1), and 55 from other species, including B. mori (Gong et al., 2007), M. sexta (Grosse-Wilde et al., 2011), and A. lepigone (Zhang et al., 2017b). The OR data set contained 64 sequences from S. exigua (Table S1), and 91 from other species (Tanaka et al., 2009; Zhan et al., 2011; Zhang et al., 2015b). The IR data set contained 22 sequences from S. exigua (Table S1), and 131 from other species (Croset et al., 2010; Olivier et al., 2011; Rimal and Lee, 2018). The GR data set contained 30 sequences from S. exigua (Table S1), and 126 from other species (Zhan et al., 2011; Liu et al., 2014b; Guo et al., 2017). Then, we used ClustalX 1.83 (Larkin et al., 2007) to align amino acid sequences from the same family gene, and used PhyML 3.1 (Guindon et al., 2010) based on the LG substitution model (Le and Gascuel, 2008) with Nearest Neighbor Interchange (NNI) to construct the phylogenetic trees, and the branch support of tree estimated by a Bayesian-like transformation of the aLRT (aBayes) method (Anisimova et al., 2011). Lastly, we created and edited the different trees by using the FigTree 1.4.2 software (http://tree.bio. ed.ac.uk/software/figtree/).

## Quantitative Real-Time PCR (qPCR) Analysis

According to the minimum information for publication of qPCR experiments (Bustin et al., 2009) and our previous studies (Zhang et al., 2017a), we performed the qPCR assay of tissue distribution of SexiGRs in ABI 7300 (Applied Biosystems, Foster City, CA, USA) by using 2×SYBR Green PCR Master Mix (YIFEIXUE BIO TECH, Nanjing, China) as the manufacturer's instructions. Briefly, the reaction programs were 10 min at 95◦C, 40 cycles of 95◦C for 15 s and 60◦C for 1 min. The qPCR primers (Table S2) were designed using Beacon Designer 7.9 (PREMIER Biosoft International, CA, USA). Then, the relative expression levels of SexiGRs mRNA were calculated based on the Ct-values of target gene and two reference genes SexiGAPDH (glyceraldehyde-3-phosphate dehydrogenase) and SexiEF (elongation factor-1 alpha) by using the Q-Gene method in Microsoft Excel-based software of Visual Basic (Muller et al., 2002; Simon, 2003), the qPCR data are listed in Table S3. To ensure the reliability of the results, we carried out three biological replications for each sample and three technical replications for each biological replication.

#### Statistical Analysis

Data (mean ± SE) from various samples were subjected to oneway nested analysis of variance (ANOVA), followed by the least significant difference test (LSD) for comparison of means using SPSS Statistics 22.0 (SPSS Inc., Chicago, IL, USA).


#### RESULTS AND DISCUSSION

#### Overview of Transcriptomes From the Six Organs

We used next-generation sequencing to sequence the six cDNA libraries constructed from the chemosensory organs (FA, MA, FPr, MPr, FLP, and MLP) of S. exigua adults based on the Illumina HiSeqTM 4000 platform and acquired 65.81 (from 10.60 to 11.90) giga base reads. After clustering and redundancy filtering, we finally obtained 144,479 unigenes and 266,645 transcripts with

a N50 length of 2,177 base pair (bp) and 1,552 bp, respectively (**Table 1**). Statistics showed that 59.22% of the 144,479 unigenes were greater than 500 bp in length (**Figure 1**). The number of reads, unigenes, and transcripts were higher than most other insects based on transcriptome studies.

In total, 60,373 unigenes were matched to entries in the National Center for Biotechnology Information (NCBI) non-redundant (NR) protein database (http://www.ncbi. nlm.nih.gov/protein) by a BLASTX homology search with a cut-off e-value of 10−<sup>5</sup> . The highest match percentage (37.40%) was identified with sequences of Bombyx mori followed by sequences of Danaus plexippus (15.60%), P. xylostella (13.20%), Homo sapiens (4.30%), and H. armigera (1.40%; **Figure 2**).

#### TABLE 2 | The Blastx match of *S. exigua* putative OBP and CSP genes.


#### TABLE 3 | The Blastx Match of *S. exigua* putative OR, IR and GR genes.


*(Continued)*

#### TABLE 3 | Continued


*(Continued)*

#### TABLE 3 | Continued


*TMD, transmembrane domain.*

Based on methodology described in our previous studies (Zhang et al., 2013; Li et al., 2015), we applied Blast2GO to classify the functional groups of all unigenes. The results showed that only 29.29% (42,331) of the 144,479 unigenes could be annotated based on the sequence homology, with this proportion similar to that found in other insects (Gu et al., 2013; Zhang et al., 2013; He et al., 2017). One possible reason for this might be that a great amount of S. exigua unigenes belong to non-coding or homologous genes without a gene ontology (GO) term. In addition, the GO annotation of S. exigua unigenes displayed similar classification to the unigenes of chemosensory organs from other moth species (Grosse-Wilde et al., 2011; Zhang et al., 2013; Cao et al., 2014; Xia et al., 2015). For example, unigenes of S. exigua during biological processes were predicted to be mostly enriched in three sub-categories: cellular, metabolic, and singleorganism processes. There was also expected to be similarity in the cellular components (e.g., cell, cell part, and organelle) and molecular function categories (binding, catalytic, and transporter activity; **Figure 3**), indicating that some unigenes in these sub-categories might play important roles in the chemosensory behavior of moths.

#### Differentially Expressed Genes (DEGs)

To investigate the DEGs among different organs, we compared each organ pair-wise within each sex against all other organs (**Figure 4**). Gene expression dynamics can be reflected by upor down-regulation among the six different organs by pairwise comparisons. The results showed that there were a number of DEGs between different organs and different sexes, and the number of DEGs was highest in FPr vs. FLP (6,029 genes in total: 4,050 up-regulated genes and 1,979 downregulated genes), followed by MA vs. FPr (5,127 genes in total: 1,928 up-regulated genes and 3,199 down-regulated genes), and MPr vs. FLP (4,033 genes in total: 2,513 up-regulated genes and 1,520 down-regulated genes). This indicates that these DEGs, especially in the gustatory vs. olfactory organs, provide substantial genetic sources that are important for studying the differential mechanism of gustatory vs. olfactory

organs in S. exigua. Additionally, they provide some important target genes to analyse the functions of expressed sex-specific genes to reveal sex differences in chemosensory mechanisms in the future.

#### Identification of Putative Chemosensory Genes

Based on sequence similarity analyses and characteristics of insect chemosensory genes from previous studies (Xu et al., 2009; Croset et al., 2010; Zhou, 2010; Zhang et al., 2011; Ray et al., 2014), such as the conserved C-pattern of OBPs and CSPs, and the conserved transmembrane structure and motifs of chemosensory receptors (ORs, IRs, and GRs), we totally identified 159 putative genes from the transcriptomic data of S. exigua chemosensory organs that belonged to five insect chemosensory gene families. These included 24 OBPs, 19 CSPs, 64 ORs, 22 IRs, and 30 GRs (**Tables 2**, **3**). The number of putative chemosensory genes of S. exigua identified in the present study was higher than that in other moth species where the same family genes had been identified by analysis of the transcriptome of specific organs. This included H. armigera (143 genes: 34 OBPs, 18 CSPs, 60 ORs, 21 IRs, and 10 GRs) (Liu et al., 2014b), H. assulta (147 genes: 29 OBPs, 17 CSPs, 64 ORs, 19 IRs, and 18 GRs) (Xu et al., 2015), and P. xyllostella (116 genes: 24 OBPs, 15 CSPs, 54 ORs, 16 IRs, and 7 GRs) (Yang et al., 2017). We found that the amount of transcriptomic data of these three different moth species was less than that of S. exigua in the present study, which suggests that the large amount of transcriptomic data could help us obtain more insect chemosensory genes.

## OBPs

We obtained a complete set of 24 different unigenes encoding putative OBPs in S. exigua (**Table 2**), of which 3 were newly identified. Sequence analysis revealed that 23 sequences were predicted to have full-length open reading frames (ORFs) and encoded 118–239 amino acids, but only 3 of the 23 SexiOBPs did not have signal peptide sequences (**Table 2**). The phylogenetic analysis showed that all 24 SexiOBPs were clustered in an OBP tree with Manduca sexta, B. mori, and Athetis lepigone

(**Figure 5**), including 5 SexiOBPs (SexiPBP1-3, SexiGOBP1-2) clustered into the PBP/GOBP subfamily. The results suggest that these SexiOBPs belonged to the insect OBP family and should have the corresponding functions of the insect OBP (Poivet et al., 2012; Jeong et al., 2013; Pelosi et al., 2014; Liu et al., 2015a). The two new SexiOBPs (SexiOBP-N1 and SexiOBP-N3) encoded protein with high identities (97 and 99%) to OBPs in Spodoptera litura, respectively, indicating that SexiOBP-N1 and SexiOBP-N3 might have conserved functions in the two closely related species, such as recognizing the same host plant volatiles (Li et al., 2013; Gu et al., 2015). Therefore, they can be considered as target genes to simultaneously prevent and control these two pests (S. exigua and S. litura) in the future.

## CSPs

Nineteen putative genes encoding CSPs were acquired in S. exigua based on the analysis results from the transcriptomes of the six chemosensory organs, of which four were newly attained (**Table 2**). Among the 19 SexiCSPs, 18 had full length ORFs with 4 conserved cysteines in the corresponding position and a predicted signal peptide at the N-terminus. The constructed insect CSP tree using protein sequences from S. exigua, M. sexta, B. mori, and A. lepigone (**Figure 6**) indicated that all 19 SexiCSPs were distributed along various branches and each clustered with at least 1 other moth ortholog. Thus, we inferred that these SexiCSPs should have a similar chemosensory function in insects, especially moths (Lartigue et al., 2002; Campanacci et al., 2003; Zhang et al., 2014). Similar to SexiOBPs, we also found three of the four new SexiCSPs (SexiCSP-N2, SexiCSP-N3, and SexiCSP-N4) encoded proteins with high identities (99 and 100%) to CSPs in S. litura. This showed that they were very similar, maybe even the same CSPs, and might play the same role as OBPs in the two moths. In future studies, we intend to use the combination of in vitro (Jin et al., 2014; Zhang et al., 2014) and in vivo (Zhu et al., 2016; Dong et al., 2017; Ye et al., 2017) methods to explore the exact function of these conserved OBPs and CSPs in the two closely related species. In addition, we plan to study the exact functions of all the unknown functional OBPs and CSPs of S. exigua, which will help us define the odorant binding spectrum of each gene. This will provide potential behavioral disturbance agents to control the moths by using reverse chemical ecology methods (Zhu et al., 2017).

## ORs

Sixty-four different unigenes encoding putative ORs were identified by analyzing the transcriptome data of S. exigua, of which 57 were newly obtained (**Table 3**). A total of 28 out of 64 SexiORs contained full-length ORFs that encoded 351 to 473 amino acids with various transmembrane domains (TMD). The phylogenetic analysis showed that all 64 SexiORs were clustered in an OR tree with B. mori, D. plexippus, and H. armigera, with each clustering having at least one other moth ortholog (**Figure 7**). In accordance with previous studies (Liu et al., 2013), we also identified a chaperone and higher conserved insect OR—SexiOrco (Krieger et al., 2005; Nakagawa et al., 2005; Xu and Leal, 2013; Missbach et al., 2014) and four pheromone receptors (SexiOR6, 11, 13, and 16) (**Table 3**, **Figure 7**), which suggests that our sequencing and analysis methods were reliable. The results of the phylogenetic and sequence homology analyses showed that we were able to obtain the fifth PR gene of S. exigua, SexiOR59. Liu's research (Liu et al., 2013) found that only two PRs (SexiOR13 and SexiOR16) showed higher electrophysiological responses to the three sex pheromone components (Z9, E12-14:OAc, Z9-14:OAc, and Z9-14:OH) of S. exigua; however, no PRs displayed specific or higher response to the fourth pheromone component Z9, E12-14:OH. Therefore, further studies are required to determine whether SexiOR59 can respond highly or not to Z9, E12-14:OH or other pheromone components. Additionally, other researchers have found that several non-PR ORs could respond to host plant volatiles, such as SlitOR12 of S. litura (Zhang et al., 2015c), EpstOR1, and three from Epiphyas postvittana (Jordan et al., 2009). Therefore, some ORs of the 58 non-PR ORs in S. exigua might play a similar role in the chemosensation of the volatiles in host plants.

#### IRs

A total of 22 putative IR genes in S. exigua were identified, of which 16 were newly obtained (**Table 3**), and the SexiIRs number was similar to several other insects (Croset et al., 2010; Liu et al., 2014b; Xu et al., 2015). Only 7 of these genes had a fulllength ORF (SexiIR2, 4, 5, 8, 9, 11, and 15) that encoded 542 to 918 amino acids with 3 or 4 TMD. We then constructed an insect IR tree using protein sequences from S. exigua, Drosophila

melanogaster, B. mori, and Anopheles gambiae, which indicated that all 22 SexiIRs were clustered into 3 subfamilies of insect IR: 14 antennal IRs (SexiIR2, 4-6, 9, 11, 13, 14, 16-20, and 22), 6 divergent IRs (SexiIR1, 7, 8, 10, 12, and 21), and 2 IR25a/IR8a (SexiIR15 and 3), but no SexiIRs belonged to non-NMDA IGluRs subfamilies (**Figure 8**). This is similar to the conserved co-receptor Orco, where IR25a and IR8a of the insect were also co-receptors and could be co-expressed along with other IRs to ensure that insects could accurately detect external odorants via chemosensory organs (Abuin et al., 2011). Therefore, the co-receptors SexiIR15 (25a) and SexiIR3 (8a) might play the role of molecular chaperone to help with other SexiIRs functions.

#### GRs

We first identified 30 different unigenes encoding putative SexiGRs in the present study (**Table 3**). Sequence analysis revealed that 12 sequences were predicted to have full-length ORFs that encoded 339–503 amino acids with 3–8 TMD. This number of SexiGRs is higher than that of other species based on the transcriptome analysis, such as H. armigera (10 GRs) (Liu et al., 2014b), H. assulta (18 GRs) (Xu et al., 2015) and Hyphantria cunea (9 GRs) (Zhang et al., 2016), but lower than that of 3 species whose genomes have been sequenced, B. mori (69 GRs) (Wanner and Robertson, 2008; Sato et al., 2011), D. plexippus (58 GRs) (Zhan et al., 2011; Briscoe et al., 2013), and Heliconius melpomene (73 GRs) (Briscoe et al., 2013). This suggests that there is a high chance of identifying more SexiGR genes when the genome of S. exigua is successfully sequenced in the future.

An insect GR tree using protein sequences from S. exigua, B. mori, D. plexippus, and H. armigera was then constructed, and the tree showed that 3 SexiGRs (Sexi10, 13, and 25) were clustered in the CO<sup>2</sup> Receptors subfamily, 6 SexiGRs (SexiGR4, 8, 12, 16, 27, and 30) were clustered in the Sugar Receptor subfamily, and 2 SexiGRs (SexiGR13 and 29) were clustered in the Fructose Receptor subfamily (**Figure 9**), indicating that these SexiGRs might be involved in the detection of CO<sup>2</sup> (Jones et al., 2007; Kwon et al., 2007), sugar (Sato et al., 2011), and fructose (Jiang et al., 2015; Mang et al., 2016). Other SexiGRs, which do not belong to the three subfamilies, might be involved in other taste perception processes.

To better infer the potential functions of these SexiGRs, we applied the qPCR method to investigate the expression profiles of all SexiGRs in six chemosensory organs (FA, MA, FPr, MPr, FLP, and MLP) and two non-chemosensory organs (Female abdomen, FAb and Male abdomen, MAb) (**Figure 10**).

The results showed that the organ with the highest SexiGRs expression was FPr (28 genes), followed by MPr (27 genes), FLP (25 genes), and MLP (22 genes), indicating that SexiGRs mainly exist within the gustatory organs, not the olfactory or non-chemosensory organs. This explains why the numbers of GR based on the antennae or non-gustatory organs transcriptome of other insects (Liu et al., 2014b; Xu et al., 2015) are lower than the SexiGRs in the present study. Additionally, we found 4, 16, 11, and 1 SexiGR genes that were highly expressed in the antennae, proboscises, labial palps, and abdomen of S. exigua, respectively, and some genes also showed differences in sex expression, which suggests that SexiGRs not only plays a pivotal role in gustatory processes (Jiang et al., 2015; Poudel et al., 2015), but might also be involved in olfactory (Agnihotri et al., 2016; Poudel et al., 2017) and other physiology processes (Xu et al., 2012; Ni et al., 2013). These results indicate that the proboscises and labial palps play more important roles in the taste perception process of than the olfactory organs do, which provides an important reference for future study of the taste perception mechanism in S. exigua as well as in other moths.

In conclusion, 159 genes encoding putative chemosensory genes were obtained by analyzing six chemosensory organs of S. exigua. Our approach proved to be highly effective for the identification of chemosensory genes in S. exigua, for which genomic data are currently unavailable. As the first step toward understanding gene functions, we conducted a comprehensive phylogenetic analysis of these genes and investigated all SexiGRs expression patterns, most of which were highly expressed in gustatory organs. The present study greatly improves the gene inventory for S. exigua and provides a foundation for future functional analyses of these crucial genes.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

Y-NZ conceived and designed the experimental plan. Y-NZ, J-LQ, M-YL, and X-XX performed the experiment. Y-NZ, X-YZ, TX, and LS processed and analyzed the experiment data. J-WX and C-XL provided important suggestions to help modify the manuscript. Y-NZ wrote the manuscript.

#### FUNDING

This work was supported by National Natural Science Foundation of China (31501647), Natural Science Fund of Education Department of Anhui province, China (KJ2017A387, KJ2017B019 and KJ2017A384), Central Public-Interest Scientific Institution Basal Research Fund (1610212016015, 1610212018010), Key Laboratory of Biology, Genetics and Breeding of Special Economic Animals and Plants, Ministry of Agriculture, China (Y2018PT14), Key Project of International Science and Technology Cooperation, National Key Research and Development Program of China (2017YFE0107500), The Science and Technology Innovation Project of the Chinese Academy of Agricultural Sciences (CAAS-ASTIP-2015-TRICAAS).

#### ACKNOWLEDGMENTS

We thank Bachelor students Cai-Yun Yin (Huaibei Normal University, China) for their help in collecting insects.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00432/full#supplementary-material


of insect pest the purple stem borer Sesamia inferens (Walker). PLoS ONE 8:e69715. doi: 10.1371/journal.pone.0069715


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zhang, Qian, Xu, Zhu, Li, Xu, Liu, Xue and Sun. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Dynamic Changes in Chemosensory Gene Expression during the Dendrolimus punctatus Mating Process

#### Su-fang Zhang, Zhen Zhang\*, Xiang-bo Kong, Hong-bin Wang and Fu Liu

*Key Laboratory of Forest Protection, Research Institute of Forest Ecology, Environment and Protection, Chinese Academy of Forestry, State Forestry Administration, Beijing, China*

The insect chemosensory system is pivotal for interactions with their environments, and moths have especially sensitive olfaction. Exploration of the connection between the plasticity of olfactory-guided and molecular level pathways in insects is important for understanding the olfactory recognition mechanisms of insects. The pine caterpillar moth, *Dendrolimus punctatus* Walker, is a dominant conifer defoliator in China, and mating is the priority for adults of this species, during which sex pheromone recognition and oviposition site location are the main activities; these activities are all closely related to chemosensory genes. Thus, we aimed to identify chemosensory related genes and monitor the spectrum of their dynamic expression during the entire mating process in *D. punctatus*. In this study, we generated transcriptome data from male and female adult *D. punctatus* specimens at four mating stages: eclosion, calling, copulation, and post-coitum. These data were analyzed using bioinformatics tools to identify the major olfactory-related gene families and determine their expression patterns during mating. Levels of odorant binding proteins (OBPs), chemosensory proteins (CSPs), and odorant receptors (ORs) were closely correlated with mating behavior. Comparison with ORs from other *Dendrolimus* and Lepidoptera species led to the discovery of a group of ORs specific to *Dendrolimus*. Furthermore, we identified several genes encoding OBPs and ORs that were upregulated after mating in females; these genes may mediate the location of host plants for oviposition via plant-emitted volatiles. This work will facilitate functional research into *D. punctatus* chemosensory genes, provide information about the relationship between chemosensory genes and important physiological activities, and promote research into the mechanisms underlying insect olfactory recognition.

Keywords: chemosensory gene, mating, expression dynamic, pheromone receptor, insect olfaction, masson pine moth

#### INTRODUCTION

Masson pine (Pinus massoniana L.) is a dominant and native forest plant species in southern China. As it grows readily in poor soils, huge forests of this species were planted in southern China; however, these vast areas of P. massoniana monoculture forest present problems, including frequent damage by forest insects. One of the most serious pests of coniferous forests in southern

#### Edited by:

*Shuang-Lin Dong, Nanjing Agricultural University, China*

#### Reviewed by:

*Liang Sun, Tea Research Institute (CAAS), China Wei Xu, Murdoch University, Australia Guan-Heng Zhu, University of Kentucky, United States Francesca Romana Dani, University of Florence, Italy*

> \*Correspondence: *Zhen Zhang zhangzhen@caf.ac.cn*

#### Specialty section:

*This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology*

Received: *11 September 2017* Accepted: *20 December 2017* Published: *10 January 2018*

#### Citation:

*Zhang S, Zhang Z, Kong X, Wang H and Liu F (2018) Dynamic Changes in Chemosensory Gene Expression during the Dendrolimus punctatus Mating Process. Front. Physiol. 8:1127. doi: 10.3389/fphys.2017.01127*

**432**

China is the pine caterpillar moth, Dendrolimus punctatus Walker (Lepidoptera: Lasiocampidae) (Xiao, 1992). During outbreaks, high population densities of D. punctatus larvae feed intensively on pine needles, causing substantial damage to trees, and huge economic losses (Zhao et al., 1993). In the past, chemical insecticides were used to treat outbreaks, causing severe negative effects on the biodiversity of the ecosystem (Kong et al., 2007). Thus, the control of the pine caterpillar moth has been of long-term interest to forest insect researchers in China, and new methods are imperative to control this pest. Based on its potential in population outbreak monitoring and pest controlling of olfactory communication system, it attracted the interesting of many scientists (Gao et al., 2001; Kong et al., 2006; Li et al., 2015); however, only fragmentary data is available regarding the molecular mechanisms of odor detection in Dendrolimus species (Zhang S.-F. et al., 2014).

Lepidopteran species have highly specific and sensitive olfactory systems (Zhang et al., 2015). Several groups of olfactory-related genes play critical roles in the transformation of chemical signals (such as sex pheromones or plant volatiles) to electrical nervous impulses, including three receptor families, two binding protein families, and the sensory neuron membrane proteins (SNMPs) (Vogt et al., 2009; Zhang et al., 2014a). The three receptor families, odorant receptors (OR), ionotropic receptors (IR), and gustatory receptors (GR), are transmembrane molecules expressed in the sensillar neurons of insect antennae (Kwon et al., 2007; Benton et al., 2009; Robertson and Kent, 2009; Touhara and Vosshall, 2009; Kaupp, 2010). The two binding protein families include odorant binding proteins (OBPs) and chemosensory proteins (CSPs), which are small soluble proteins expressed in the lymph of antennae (Vogt, 2003; Pelosi et al., 2006; Sanchez-Gracia et al., 2009). These two classes of protein also have other functions, as recently reviewed by Pelosi et al. (2017). Classic OBPs contain six conserved cysteine residues, and there are two other type of OBPs, plus-C OBPs, which contain 4– 6 additional cysteines, and minus-C OBPs, which contain fewer cysteine residues (generally C2 and C5 are absent) (Hekmat-Scafe et al., 2002; Sanchez-Gracia et al., 2009). Most of the genes encoding these proteins exhibit considerable sequence diversity (Krieger et al., 2004; Robertson and Wanner, 2006; Engsontia et al., 2008; Tanaka et al., 2009), and their identification has primarily been based on genomic data (Zhou et al., 2006, 2008; Gong et al., 2009), or antennal transcriptomes (Grosse-Wilde et al., 2011; Legeai et al., 2011; Bengtsson et al., 2012; Khan et al., 2013; Zhang et al., 2014a; Zhou et al., 2015).

Many moths exhibit olfactory-guided behavioral plasticity, depending on the physiological status of the individual (Anton et al., 2007). In particular, mating can dramatically influence the olfactory behavior of moths. For example, virgin Vitacea polistiformis males exhibited four-fold higher electroantennogram responses to pheromones than mated males (Pearson and Schal, 1999). Moreover, only mated Amyelois transitella (Walker) (Phelan and Baker, 1987), Lobesia botrana (Masante-Roca et al., 2007), and Manduca sexta (Mechaber et al., 2002) females, but not virgins, were attracted by plant volatiles. The response of mated Plutella xylostella females to some green leaf volatiles was stronger than those of males or unmated females (Reddy and Guerrero, 2000). Clearly, mating can influence the behavioral responses of insects to volatiles, although to differing extents among species. Moths can also adjust the level of chemosensory gene expression depending on their physiological status or development stage. For example, changes in the expression levels of pheromone binding protein 1 correlated with the mating status of P. xylostella (Zhang et al., 2009). Another study showed that mating did not affect the expression of minus-C OBPs in male Batocera horsfieldi beetles; however, it could affect that of females. Nevertheless, to date, studies attempting to correlate physiological status with dynamic olfactory gene expression remain rare, and this topic warrants further attention.

The majority of D. punctatus insects of both sexes only mate once in their lives, while a few mate twice, and mating lasts ∼18 h (Zhou, 2013). Furthermore, D. punctatus adults do not eat and die soon after oviposition. Thus, mating is the priority for adult D. punctatus, and sex pheromone recognition and oviposition site location are their main activities. Notably, these activities are both closely related to olfaction. Thus, the dynamics of chemosensory gene expression during the mating process deserves further study. In general, the numbers of chemosensory genes (such as OBPs and ORs) are huge in insects; for example, there are 44 OBPs and 72 ORs in Bombyx mori (Khan et al., 2013), and the specific function of each gene remains unclear. The expression patterns of these genes provide important clues about their functions, and olfactory genes with expression levels closely related to mating and oviposition activities may perform important functions during these behaviors. In this study, we focused on two aims: first, based on our previous work (Zhang et al., 2017), identification of chemosensory genes in D. punctatus; second, monitoring the dynamic expression spectrum of chemosensory genes during the whole mating process, with the aim of inferring the functions of different genes. This work will not only facilitate follow-up functional investigation of chemosensory genes, which has potential to identify novel targets for pest control, but also determine the relationship between the spectrum of chemosensory genes and important physiological and behavioral activities, and promote research into the mechanisms underlying insect olfactory recognition.

#### MATERIALS AND METHODS

#### Insects

In 2015, we collected about 200 D. punctatus pupae in Quanzhou, Guilin City, Guangxi province, China, and reared them in our laboratory at 26 ± 2 ◦C, 50 ± 10% relative humidity, and a 16 h light: 8 h dark photoperiod. Male and female insects representing four different physiological conditions were prepared for transcriptome sequencing as follows: newly emerged (within 5 h after emergence, unmated; eclosion), calling females and corresponding males, mating status (copulating), and after mating status (post-coitum). Male and female insects were kept in two different insect rearing cages placed in close proximity to each other and separated by only two layers of screen cloth, so that males could sense the female sex pheromones. Antennae from 15 female and male D. punctatus specimens at each stage were cut off and immediately frozen in liquid nitrogen. Insect antennae from each group were divided into three equal parts, as three biological replicates. Thus, in total, we constructed 24 libraries for RNA-seq (four conditions for male and female insects respectively, and three replications for each status).

#### RNA-seq Library Preparation

As previously described (Zhang et al., 2014a, 2017), total RNA samples were extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and treated with RNase-free DNase I (TaKaRa, Dalian, Liaoning, China). Subsequently, RNA purity, concentration, and integrity were checked using the NanoPhotometer <sup>R</sup> spectrophotometer (IMPLEN, CA, USA), a Qubit <sup>R</sup> RNA Assay Kit and a Qubit <sup>R</sup> 2.0 Flurometer (Life Technologies, CA, USA), and the RNA Nano 6000 Assay Kit on the Bioanalyzer 2100 system (Agilent Technologies, CA, USA), respectively.

Duplex-specific-nuclease normalized cDNA was synthetized using 3 µg total RNA samples (Zhulidov et al., 2004; Bogdanova et al., 2008). The RIN values of all samples were > 8. We prepared sequencing libraries using an Illumina TruSeqTM RNA Sample Preparation Kit (Illumina, San Diego, CA, USA), and added four index codes to identify sequences from each sample. To preferentially select cDNA fragments of 200 bp length, we purified the libraries using the AMPure XP system (Beckman Coulter, Beverly, MA, USA). PCR (10 cycles) was performed to enrich for the two-end adaptor ligated DNA fragments. Finally, the products were purified using an AMPure XP system and quantified on an Agilent Bioanalyzer 2100.

#### Clustering and Sequencing

Index-coded samples were clustered using the TruSeq PE Cluster Kit v3-cBot-HS (Illumina) on a cBot Cluster Generation System, then sequencing performed on an Illumina Hiseq 2500 platform, according to the manufacturer's instructions.

## De Novo Assembly

Raw sequencing data were filtered to remove reads containing adapter sequence, reads with > 10% N (uncertain bases), and sequences with error rates > 1% for more than 50%, using self-written Perl scripts, to obtain clean data. Then we calculated the Q20, Q30, GC-content, and sequence duplication level of the clean data, which were subsequently used for downstream analyses. Clean data sequences were compared with the NT database to determine whether they were polluted. Trinity (vesion:trinityrnaseq\_r20131110) was used to perform transcriptome assembly (Grabherr et al., 2011). TGICL software was used to reduce redundancy (Pertea et al., 2003). The raw data from our experiments have been deposited in the NCBI SRA database under the accession number SRP102206 (Bioproject accession number PRJNA374901). We assessed the transcriptome assembly using benchmarking universal singlecopy orthologs (BUSCO) based on the percentage of sequences aligned with highly conserved protein sequences, (Simão et al., 2015).

#### Annotation

First, transcript sequences were searched using BLAST against the NR, SWISSPROT, KEGG, and KOG databases, with a cut-off value of 1e-5, and the highest sequence similarity targets selected for functional annotation of the transcripts. Next, Blast2GO was used to perform GO annotation of the transcripts (Conesa et al., 2005; Götz et al., 2008). Finally, the molecular function, biological process, and cellular component of the genes were assigned (Ashburner et al., 2000; Krieger et al., 2004).

Based on our previous research (Zhang et al., 2017), we further identified the chemosensory genes in D. punctatus. Previously identified chemosensory genes were confirmed in our new transcriptome database using tBLASTx searches, and the complete sequences of some previously identified partial genes obtained. We further identified some new chemosensory genes by contig tBLASTx searches. The open reading frames (ORFs) of possible genes were verified by additional BLAST searches (http://blast.ncbi.nlm.nih.gov/ Blast.cgi). Newly identified olfactory genes were submitted to GenBank; the updated accession numbers are listed in **Table S1**. Maximum likelihood (ML) and neighbor-joining (NJ) phylogenetic trees of chemosensory genes were constructed using MEGA5 with 1,000 bootstrap replications (Tamura et al., 2011). MEGA's model test was used to select the best model for ML tree construction. Dendrograms were colored in Adobe illustrator (Adobe Systems). Motif analysis of the predicted intact ORFs of chemosensory genes was performed using the MEME online server (Version 4.12.0.) (Bailey et al., 2009) http://meme-suite.org/tools/meme. For OBP and CSP, the motif discovery parameters were: minimum width = 6, maximum = 10, maximum motifs to find = 8; for ORs they were: minimum width = 15, maximum = 50, maximum motifs to find = 8.

## Gene Expression Quantification

To measure the gene expression levels in transcriptomes, we used the FPKM (fragments per kilobase of exon per million fragments mapped) criteria (Trapnell et al., 2010). Three biological replicates were sequenced for each D. punctatus status, and the mean FPKM value and standard error obtained from the three replicates. Differentially expressed genes (DEGs) between different mating status insects were calculated using DESeq (http://bioconductor.org/packages/release/bioc/html/DESeq.

html) (Anders and Huber, 2010), based on the reads of each unigene. Unigene expression levels and DEGs were normalized following the compatible-hits-norm model (Bullard et al., 2010). DEGs were screened to identify those generating q-values ≤ 0.05 using the false discovery rate (FDR) method (Noble, 2009).

## GO Enrichment and Expression Trend Analysis of DEGs

GO Enrichment analysis of DEGs was carried out using GOstat (Beißbarth and Speed, 2004), with p-values approximated using Chi-square tests, with all annotated genes used as the background. Short Time-series Expression Miner (STEM, vision 1.3.11) was used to analyze the expression trends of some DEGs (Ernst and Bar-Joseph, 2006). FPKM values were log2 transformed and imported into the software.

#### Quantitative Real-Time PCR (qPCR)

qPCR was carried out to validate the RNA-Seq data, similar to our previous report (Zhang et al., 2014a,b, 2017). qPCR primers (**Table S2**) were designed based on cDNA sequences. RT-PCR was performed to test whether qPCR primers could amplify the correct products. Beta-actin was used as the housekeeping gene. T-easy clones containing the tested genes were constructed as reference genes to construct qPCR standard curves. Amplification efficiencies of all primers tested were 90–100%. Real-time PCR was carried out in a Roche LightCycler 480 (Stratagene, La Jolla, CA, USA). The PCR cycles were as follows: 2 min at 95◦C, 40 cycles of 20 s at 95◦C, 20 s at 58◦C, and 20 s at 72◦C; finally, melting curve analysis (58 to 95◦C) was performed to evaluate the specificity of the PCR products. Ct values were calculated using the Roche qPCR software (version 1.5.1) with the second derivative method. Three independent biological reactions were completed for each insect status, along with three technical replicates for each reaction. Gene expression levels tested by qPCR in female and male D. punctatus antennae (relative to that of the actin gene) were compared with the transcriptome expression data (FPKM), as illustrated in **Figure S1**.

## RESULTS

#### Assembly

Transcriptomic sequence data were generated from antenna cDNA libraries from D. punctatus adults at different mating stages using Illumina HiSeqTM2500 technology. We acquired 204.30 Gbp of clean sequence data in 1,634,361,960 clean reads. After assembly, 110,760 unigenes were obtained, with an N50 of 2,380 bp. Approximately 80% of unigenes were >500 bp, with a maximum length of 54,680 bp (**Figure S2**). We evaluated the completeness and accuracy of transcriptome assembly using BUSCOs, and the results demonstrated 98.1% complete BUSCOs (C: 98.1 [S: 83.8%, D: 14.3%], F: 1.0%, M: 0.9%).

#### Annotation, GO Enrichment, and STEM Analyses

BLAST analysis indicated that D. punctatus transcriptome unigenes were most similar to amino acid sequences from three other Lepidoptera species: B. mori (16,769 hits with E-values <1e-5), Danaus plexippus (7826 hits with E-values <1e-5), and P. xylostella (7034 hits with E-values <1e-5) (**Figure 1**). These three species accounted for ∼75% of hits.

Function distribution, determined by Gene Ontology (GO) analysis, indicated that D. punctatus genes were primarily enriched for binding or catalytic activity (**Figure S3A**), and followed by transporter and structural molecule activity. KOG classification indicated that genes involved in signal transduction occupied an important position (**Figure S3B**).

GO enrichment analysis of DEGs revealed that waves of gene expression changes occurred in the antennae of D. punctatus during the mating process (**Figure S4**). Overall, olfactory detection genes, particularly olfactory receptors and odorant binding related genes, exhibited dramatic differences between sexes and also during the mating process of male and female

insects. To further elucidate the characteristics of gene expression in insects of different mating status, we analyzed the expression trend of DEGs with short time-series expression miner (STEM). Thirteen significant profiles were obtained, four of which were related to chemosensory genes (**Figure 2A**). GO enrichment of the profiles indicated that profiles 41 and 26 included the most chemosensory associated genes (**Figure 2B**). Of the four chemosensory related profiles, profile 7 continually declined and included genes expressed more highly in female antennae; profile 41 continually rose, and included genes expressed at higher levels in male antennae; profiles 26 and 27 fluctuated in the eight D. punctatus groups, and included genes with expression levels that oscillated during the mating process of this insect.

#### Identification and Expression Dynamics of Chemosensory Genes

GO enrichment and STEM analysis indicated that chemosensory genes may be very important in the mating process. Thus, detailed analyses were performed to determine the characteristics of the olfactory-related gene families identified from the transcriptomes of D. punctatus in different mating states.

In our previous work, we identified a considerable number of D. punctatus chemosensory genes (Zhang et al., 2017). Here, after further effort, the complete sequences of many of the partial gene sequences identified previously were acquired, including six OBPs (NCBI accession numbers KY225481–KY225486), one CSP (KY225487), 23 ORs (KY225488–KY225510), one GR (KY225519), and eight IRs (KY225529–KY225536). Some new genes were also identified, including eight ORs

Items marked with asterisks are associated with insect chemo-sensation.

(KY225511–KY225518), nine GRs (KY225520–KY225528), and five IRs (KY225537–KY225541).

The correlation between the expression levels of chemosensory genes and mating status was examined in detail, and the expression levels of OBPs (**Figure S5**), CSPs (**Figure S6**), ORs (**Figure S7**), GRs (**Figure S8**), and IRs (**Figure S9**) determined. Further analysis indicated that the chemosensory genes exhibited different expression levels in insects in different mating states, with three different patterns identified (**Figure 3**). First (Type I), some genes were more strongly expressed in male than female antennae. In general, these genes were upregulated during calling or mating, and downregulated after mating. There were nine OBPs (**Figure 3**, **Figure S5A**); ten CSPs (**Figure S6A**); and nine ORs (**Figure S7A**) in this category. Second (Type II) were genes expressed at higher levels in female than male antennae. These genes were generally upregulated during calling or mating, and downregulated after mating (**Figure 3**). This category consisted of four OBPs (indicated as red in **Figure S5B**), two CSPs (**Figure S6B**), and six ORs (**Figure S7B**). Third (Type III) were genes expressed at higher levels in female than male antennae, and continually upregulated over time (**Figure S3**). This third category contained three OBPs (green in **Figure S5B**) and nine ORs (green in **Figure S7B**). To further confirm the expression level of olfactory genes, the ORs (**Figure S1A**) and OBPs (**Figure S1B**) that were belong to type I were selected for qPCR verification of the transcriptome expression data, and the results indicated that the transcriptome expression data were credible (**Figure S1**).

FIGURE 4 | in less than 40% bootstrap replicates were collapsed. Proteins expressed at higher levels in male antennae are indicated by solid circles; those expressed at higher levels in female antennae are indicated by filled triangles (higher expression during calling or mating, and downregulated after mating), hollow triangles (continually upregulated over time in female antennae), and squares (other genes expressed at higher levels in female than male antennae).

Next, phylogenetic analyses of chemosensory genes identified from D. punctatus were performed. A phylogenetic tree of the identified OBPs was constructed (**Figure 4A**). Unsurprisingly, PBP1, PBP2, and two GOBPs were grouped together, and three of these four genes were expressed at higher levels in male than female antennae. Interestingly, other OBPs more strongly expressed in male antennae were all grouped with an OBP that was preferentially expressed in female antennae. Phylogenetic analysis indicated that CSPs expressed at higher levels in male antennae (indicated by solid circles) were dispersed into two subclasses in the tree, while CSPs expressed more strongly in female antennae (filled triangles) were grouped separately (**Figure 4B**). Interestingly, the male biased ORs were almost all clustered in a single branch (**Figure 4C**).

To further analyze the characteristics of D. punctatus ORs, we performed phylogenetic analysis including ORs from two sister species of D. punctatus, Dendrolimus houi, and Dendrolimus kikuchii (Zhang et al., 2014a), and four other Lepidopteran species, including B. mori, D. plexippus, M. sexta (Grosse-Wilde et al., 2011), and Cydia pomonella (Bengtsson et al., 2012). The results permitted several observations (**Figure 5**): first, the coreceptor Orco was identified in D. punctatus and was conserved among these moths; second, the ORs from D. punctatus generally formed small subgroups together with those of D. houi and D. kikuchii, and sometimes with B. mori and M. sexta; third, the sex pheromone receptors from B. mori, M. sexta, D. plexippus, and C. pomonella formed clade in the tree (labeled "sex pheromone receptors" in **Figure 5**); however, none of the ORs from the three Dendrolimus species were clustered in this group; finally, a group of ORs from D. punctatus, D. houi, and D. kikuchii (labeled "Dendrolimus Specific Odorant Receptors" in **Figure 5**) formed a subgroup that included no receptors from the other moths, this is unusual in the Lepidoptera, and the specific functions of these ORs require further investigation.

#### Motif-Pattern Analysis of OBPs, CSPs, and ORs

To further understand the sequence characteristics of the chemosensory genes in D. punctatus, we performed motif-pattern analysis of OBPs, CSPs, and ORs using the MEME server. OBP motif analysis revealed eight groups (**Figure 6**). Motif 1 was contained in all OBPs. All GOBPs and PBPs shared four motifs 1–4, while the eighth motif was exclusive to PBPs, and the seventh motif was only found in GOBPs. OBPs in the third and fourth groups that contain the fifth motif were all minus-C OBPs, indicating that Motif 5 is a characteristic of minus-C OBPs. The eighth group included two plus-C OBPs.

Motif analysis of CSP sequences indicated that they were relatively conserved, with the majority containing the same motif pattern, with some exceptions (**Figure 7**). Interestingly, two CSPs (CSP2 and CSP15) that contained different motif patterns with respect to the others were those expressed at higher levels in female antennae.

OR motif analysis indicated that the majority of sequences could be separated into nine groups (**Figure 8**), depending on their motif patterns. We designated the first five groups as class 1, and the sixth to ninth groups as class 2, as the conserved motifs of class 1 were concentrated at the 5 ′ end of the genes, while those of class 2 were at the 3′ end. Comparative analyses indicated that male biased ORs (**Figure 3**, Type I) all belong to class 1, except for OR51, while female bias ORs (**Figure 3**, Types II and III) all belong to class 2, other than OR62. Overall, the motif patterns and expression biases of these genes indicated their functional differentiation.

## DISCUSSION

Deciphering the functions of the multiple olfactory-related genes of insects is critical to understanding the olfactory recognition mechanisms of these animals. As important activities of adult insects, mating behaviors rely heavily on sensory systems (Ziegler et al., 2013; Zhang et al., 2015). The chemosensory genes involved in these processes represent a logical starting point for functional analysis of these numerous chemosensory genes. Here, we provide a relatively comprehensive account of the dynamic expression spectrum of the chemosensory genes of D. punctatus antennal transcriptomes from different mating conditions. We analyzed the expression patterns of different olfactory genes during the mating process, and discuss the relationship between different genes and mating behaviors. The results have the potential to improve our understanding of the correlations between olfactory gene expression and mating behavior.

The numbers of chemosensory genes identified in this study were much higher than those previously identified in D. houi and D. kikuchii (Zhang et al., 2014a). For example, we identified 42 OBPs and 58 ORs in D. punctatus, while the numbers were 23 and 33 in D. houi and 27 and 33 in D. kikuchii, respectively. The reason for this discrepancy may be that we constructed more than one transcriptome from D. punctatus antennae in different mating states, resulting in greatly improved detection of olfactory genes. Possibly for the same reason, we also identified some new genes with respect to our previous work, which also focused on D. punctatus (Zhang et al., 2017). Comparisons with the olfactory gene numbers in other Lepidoptera species, including B. mori (44 OBPs, 72 ORs), M. sexta (47 ORs), Spodoptera littoralis (47 ORs), and Sesamia inferens (39 ORs) (Khan et al., 2013), indicate that we obtained a relatively intact chemosensory gene pool for D. punctatus. Surprisingly, we only identified two PBPs in D. punctatus, consistent with reports for D. houi and D. kikuchii (Zhang et al., 2014a), which is unlike many other Lepidoptera

species which typically have three PBPs (Maida et al., 2000; Abraham et al., 2005; Legeai et al., 2011; Guo et al., 2012; Khan et al., 2013). Simultaneously, no male-specific pheromone receptors were detected through phylogenetic analysis, similar to D. houi and D. kikuchii (Zhang et al., 2014a). This provides further evidence that the pheromone recognition genes of Dendrolimus exhibit characteristic features. Further studies, including PBP and OR ligand binding tests, are urgently required to explore the pheromone recognition mechanisms of Dendrolimus.

The identified expression patterns of olfactory genes during the mating process are interesting, and several genes showed

different patterns of expression. Several of these olfactory genes were expressed at higher levels in male than female antennae, and were generally upregulated when calling or mating, and downregulated after mating (Type I). The expression levels of genes in this category appear to be correlated closely with mating activity, and it includes a considerable number of OBP, CSP, and OR genes belong to this category. We deduced that these genes can bind or recognize pheromones, or other odors, that are crucial during mating behaviors, leading to physiological responses of insects during mating. Similar results have been reported for Anopheles gambiae females, in which one odorant receptor is downregulated after insects have taken a blood meal (Fox et al., 2001). Moreover, behavioral and physiological influences on gene expression levels have been identified in Drosophila melanogaster and Caenorhabditis elegans (Peckol et al., 2001; Zhou et al., 2009). Interestingly, on phylogenetic analysis, the ORs belonging to this category in D. punctatus were almost all clustered into a single branch (**Figure 4C**); this D. punctatus male-antenna biased OR branch in **Figure 4C** corresponded to the ORs marked "Dendrolimus Specific Odorant Receptors" in **Figure 5**. Although these genes were not clustered into pheromone receptor branches (**Figure 5**), we strongly suspect that these ORs may be responsible for recognition of D. punctatus sex pheromones, although further functional experiments are needed to confirm this hypothesis. Furthermore, we deduced that the sex pheromone receptors of Dendrolimus were characteristic of this genus and different from those of other moths.

The genes expressed at higher levels in female than male antennae were also important categories (Type II and Type III). Interestingly, nine ORs and three OBPs (**Figure 3**) were continually upregulated over time, with peak expression after mating (Type III). We deduced that these genes were correlated with activity after mating, which for D. punctatus is oviposition, since adults of this species do not eat or drink, and mating and oviposition are the two primary behaviors of females. During the process of insect oviposition, finding a suitable location is highly dependent on olfaction (de Bruyne and Baker, 2008; Afify and Galizia, 2015). Thus, genes in this category may recognize host plant volatiles, enabling insects to identify suitable locations for oviposition. For example, behavioral evidence from other insects indicate that mated P. xylostella females respond

more sensitively to green leaf volatiles (Reddy and Guerrero, 2000) and that mated A. transitella and M. sexta females were attracted by plant volatiles (Phelan and Baker, 1987; Mechaber et al., 2002). Thus, our future functional gene investigations of the ligands of these olfactory genes may focus on host plant volatiles.

Another gene expression pattern, those highly expressed in female antennae during calling or mating, and downregulated after mating (Type II) attracted our interest. Some studies have demonstrated that male moths, such as Anticarsia gemmatalis, can also release pheromones (Heath et al., 1983, 1988). To date, the recognition mechanisms of these male released pheromones by female moths are unclear. Since the expression pattern of Type II olfactory genes was observed to be closely correlated with mating activities, these molecules may be important for recognition of pheromones released from males; however, male pheromones released by D. punctatus have yet to be identified, hence the function of Type II olfactory genes requires further investigation.

The expression dynamics of CSPs, GRs, and IRs during mating behavior was complex. Several CSPs were correlated with the mating process, including CSP5, 14, 26. Mating related functions of CSPs have been identified by other studies (Zhang Y. N. et al., 2014); however, it seems the identified GRs and IRs only fluctuate mildly during the mating process. These results may coincide with the functions of the encoded proteins. For example, in Drosophila, antennal IRs mainly respond to acids, aromatics, and nitrogen-containing compounds (Abuin et al., 2011), while GRs primarily respond to sugars, detergents, salts, and CO2, among other substances (Agnihotri et al., 2016), and such chemicals are unlikely to be crucial in mating behavior.

To summarize, we performed a comprehensive analysis of the expression of the olfactory-related genes during the D. punctatus mating process and annotated olfactory-related proteins relatively comprehensively. Considerable numbers of OBP, CSP, and OR genes with expression patterns correlated with mating behaviors were identified, including a group of Dendrolimus male specific ORs, which are candidate pheromone receptors. Furthermore, we identified several OBP and OR genes that were upregulated after mating in females, which may be those responsible for host location via plant volatiles. These results represent the first step toward comprehensive understanding of the olfactory mechanisms of Dendrolimus species, and the foundation for population control of this pest insect.

proteins. (Below) The approximate location of each motif in the protein sequences. The numbers in the boxes correspond to the numbered motifs in the upper part of the figure.

## AUTHOR CONTRIBUTIONS

SZ designed and carried out the laboratory experiments, sequence assemblies, and drafted the manuscript. HW and XK collected the insects in the field. FL performed part of the data analysis. ZZ designed the experiments and modified the manuscript.

## FUNDING

The National Nature Science Foundation of China (31670657) and the Central Public-interest Scientific Institution Basal Research Fund (CAFRIFEEP201406 and CAFYBB2017QB003) supported this work.

#### ACKNOWLEDGMENTS

We especially thank Zhongwu Yang for assistance with insect collection. We acknowledge the reviewers for their support and help in revising the manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2017.01127/full#supplementary-material

Figure S1 | Quantitative real time PCR (qPCR) validation of transcriptome data. Expression levels were determined based on fragments per kb per million reads (FPKM), qPCR data are presented as means ± *SD*. (A) OR genes; (B) OBP genes.

Figure S2 | Distribution of unigene sizes in the transcriptome assembly of *D. punctatus*.

Figure S3 | Annotation distribution of transcriptome contigs from *D. punctatus* antennae transcriptomes. (A) GO analysis; (B) KOG classification.

Figure S4 | GO enrichment of differentially expressed unigenes (DEGs) from *D. punctatus* with different mating status.

#### REFERENCES


Figure S5 | Expression pattern of *D. punctatus* odorant binding proteins (OBPs) in insects with different mating status. (A) OBPs expressed at higher levels in male than female antennae. (B) OBPs expressed at higher levels in female than male antennae. (C) OBPs exhibiting relatively high expression levels in insects with different mating status, but without sexual bias.

Figure S6 | Expression pattern of *D. punctatus* chemosensory proteins (CSPs) in insects with different mating status. (A) CSPs expressed at higher levels in male than female antennae. (B) CSPs expressed at higher levels in female than male antennae. (C) CSPs exhibiting relatively high expression level in insects with different mating status, but without sexual bias.

Figure S7 | Expression pattern of *D. punctatus* odorant receptors (ORs) in insects with different mating status. (A) ORs expressed at higher levels in male than female antennae. (B) ORs expressed at higher levels in female than male antennae. (C) ORs exhibiting relatively high expression levels in insects with different mating status, but without sexual bias.

Figure S8 | Expression patterns of candidate *D. punctatus* gustatory receptors (GR) in insects with different mating status.

Figure S9 | Expression patterns of candidate *D. punctatus* ionotropic receptors (IR) in insects with different mating status.

Table S1 | The NCBI Accession number of updated and newly Identified sensory genes.

Table S2 | Primers used for Real-time PCR of selected genes.


(Hübner) (Lepidoptera; Noctuidae) attractive to conspecific males. J. Chem. Ecol. 14, 1121–1130. doi: 10.1007/BF01019340


Xiao, G. (1992). Forest Insects of China. Beijing: China Forestry Publishing House.


moth, Plutella xylostella. Entomol. Exp. Appl. 133, 136–145. doi: 10.1111/j.1570-7458.2009.00917.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zhang, Zhang, Kong, Wang and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Chemoreception of Mouthparts: Sensilla Morphology and Discovery of Chemosensory Genes in Proboscis and Labial Palps of Adult Helicoverpa armigera (Lepidoptera: Noctuidae)

#### Mengbo Guo1,2, Qiuyan Chen<sup>2</sup> , Yang Liu<sup>2</sup> , Guirong Wang<sup>2</sup> \* and Zhaojun Han<sup>1</sup> \*

<sup>1</sup> Education Ministry Key Laboratory of Integrated Management of Crop Diseases and Pests, College of Plant Protection, Nanjing Agricultural University, Nanjing, China, <sup>2</sup> State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China

#### Edited by:

Peng He, Guizhou University, China

#### Reviewed by:

Haonan Zhang, University of California, Riverside, United States Guan-Heng Zhu, University of Kentucky, United States Ya-Nan Zhang, Huaibei Normal University, China

> \*Correspondence: Guirong Wang

grwang@ippcaas.cn Zhaojun Han zjhan@njau.edu.cn

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 11 April 2018 Accepted: 02 July 2018 Published: 07 August 2018

#### Citation:

Guo M, Chen Q, Liu Y, Wang G and Han Z (2018) Chemoreception of Mouthparts: Sensilla Morphology and Discovery of Chemosensory Genes in Proboscis and Labial Palps of Adult Helicoverpa armigera (Lepidoptera: Noctuidae). Front. Physiol. 9:970. doi: 10.3389/fphys.2018.00970 Siphoning mouthparts, consisting of proboscis and labial palps, are the exclusive feeding organs and important chemosensory organs in most adult Lepidoptera. In this study, the general morphology of the mouthpart organs and precision architecture of the proboscis was described in adult Helicoverpa armigera. Three major sensilla types with nine subtypes including three novel subtypes were identified. The novel sensilla styloconica subtype 2 was the only one having a multiporous structure, which may play olfactory roles. For further understanding of the chemosensory functions of mouthpart organs, we conducted transcriptome analysis on labial palps and proboscises. A total of 84 chemosensory genes belonging to six different families including 4 odorant receptors (ORs), 6 ionotropic receptors (IRs), 7 gustatory receptors (GRs), 39 odorant binding proteins (OBPs), 26 chemosensory proteins (CSPs), and 2 sensory neuron membrane proteins (SNMPs) were identified. Furthermore, eight OBPs and six CSPs were identified as the novel genes. The expression level of candidate chemosensory genes in the proboscis and labial palps was evaluated by the differentially expressed gene (DEG) analysis, and the expression of candidate chemosensory receptor genes in different tissues was further investigated by quantitative real-time PCR (qRT-PCR). All the candidate receptors were detected by DEG analysis and qRT-PCR, but only a small part of the OR or IR genes was specifically or partially expressed in proboscis or labial palps, such as HarmOR58 and HarmIR75p.1, however, most of the GRs were abundantly expressed in proboscis or labial palps. The reported CO<sup>2</sup> receptors such as HarmGR1, GR2, and GR3 were mainly expressed in labial palps. HarmGR5, GR6, and GR8, belonging to the "sugar receptor" clade, were mainly expressed in proboscis or antenna and were therefore suggested to perceive saccharide. The results suggest that the mouthparts are mutually cooperative but functionally concentrated system. These works contribute to the understanding of chemical signal recognition in mouthpart organs and provide the foundation for further functional studies.

Keywords: Helicoverpa armigera, mouthparts, sensilla, transcriptome, chemosensory genes

## INTRODUCTION

fphys-09-00970 August 7, 2018 Time: 14:22 # 2

As the foremost center for sensing and food ingestion, the head of most insects possesses several sophisticated organs, including antenna and mouthpart appendages. These organs play crucial roles in almost all activities conducted by insects, including detecting host plants, feeding, recognizing mates, or locating oviposition sites. Antennae are considered to be the most important multimodal sensory organs, and they contain a huge number of sensilla for perceiving not only odorants but also flavors, carbon dioxide, and mechanical stimulation (Keil, 1999). The mouthparts act as the exclusive organ for feeding, and they also have functions in chemoreception.

Morphology and evolutionary biology of the mouthparts have been well studied previously (Kristensen, 1984; Krenn et al., 2005; Nielsen and Kristensen, 2007; Lehnert et al., 2016). The majority of adults in Lepidoptera suborder Glossata possess typical siphoning mouthparts: a proboscis adapted to their feeding properties and a pair of labial palps, together with vestigial maxillary palps. As a feeding device, the proboscis consists of the pair of maxillae galeae, which are equipped with various sensilla. The capillary construction is generated by joining the two galeae together, which can then be used for sucking liquids. Various types of sensilla have been found on the proboscis, and there are significant differences among species (Krenn et al., 2001; Xue and Hua, 2014; Lehnert et al., 2016; Xue et al., 2016). The labial palps are located on each side of the proboscis and typically possess two or three segments. The role of labial palps in CO<sup>2</sup> sensing has been demonstrated in several moth species such as Pieris rapae, Manduca sexta, Bombyx mori, Mythimna separata, and Helicoverpa armigera (Lee et al., 1985; Kent et al., 1986; Stange, 1992; Zhao et al., 2013; Dong et al., 2014).

Reception of chemical signals is mediated by three families of chemoreceptors (OR, IR, and GR) with the assistance of OBP, CSP, or SNMP in the sensilla (Benton et al., 2007; Jin et al., 2008; Leal, 2013; Fleischer et al., 2017; Pelosi et al., 2017). The peripheral perception of chemosensory stimulants was mediated by several families of olfactory proteins including odorant-binding proteins (OBPs), chemosensory proteins (CSPs), odorant receptors (ORs), gustatory receptors (GRs), ionotropic receptors (IRs), and sensory neuron membrane proteins (SNMPs). The stimulants diffuse into the cavity of sensilla through micropores on the cuticular surface and then are captured by two major families of small soluble proteins such as OBPs and CSPs (Vogt et al., 1991; Pelosi and Maida, 1995; Angeli et al., 1999; Pelosi et al., 2006, 2017). Then they are moved to the dendrite membrane of chemo-sensing neurons, where several families of the transmembrane receptors (ORs, GRs, and IRs) are expressed (Benton et al., 2009; Wang et al., 2010; Ai et al., 2013; Liu et al., 2013; Cao et al., 2016; Ning et al., 2016; Xu et al., 2016). The neurons are activated by stimulants, and then the olfactory signal is transmitted by action potentials to the primary olfactory processing center, that is, the antennal lobes (ALs) (Hansson and Christensen, 1999). The signals are further processed across multiple levels of downstream neural pathways, finally provoking a corresponding behavioral response (Hansson, 1995; Leal, 2013; Riffell and Hildebrand, 2016; Fleischer et al., 2017).

Helicoverpa armigera is one of the most damaging and highly polyphagous pests in China and many other regions all over the world; the larvae populate more than 120 plant species such as cotton, tomatoes, and maize and have caused serious economic losses (Firempong and Zalucki, 1989; Wu and Guo, 1997). To date, much progress has been made in morphological studies and in identifying chemosensory genes in antennae of H. armigera (Liu et al., 2012; Liu N.Y. et al., 2014; Zhang et al., 2015; Chang et al., 2016). For the mouthpart organs, the fine structure of labial palps has been studied carefully by Zhao et al. (2013). Each of the labial palps consists of three segments that are covered with scales. The third segment of labial palp possesses an invaginated bottle-shaped structure called the labial-palp pit organ (LPO). Almost 1,200 sensilla have been found in each LPO. Hair-shaped and club-shaped sensilla were found on the upper and lower half of the pit, respectively. Although the general structure of the proboscis in H. armigera has been reported previously, only a few sensilla types were described, perhaps due to the small number of sensilla or the resolution ratio of images.

Our previous studies have identified 66 ORs, 21 IRs, 33 OBPs, 24 CSPs, and 2 SNMPs mainly in antenna through transcriptome sequencing, and Xu et al. (2016) reported 197 GRs based on the genome and transcriptome sequencing (Liu et al., 2012; Liu N.Y. et al., 2014; Li et al., 2015; Zhang et al., 2015; Xu et al., 2016; Chang et al., 2017). Abundant chemosensory genes have been identified in the antennal transcriptome of numerous insects (Gong et al., 2007; Grosse-Wilde et al., 2011; Bengtsson et al., 2012; Zhang et al., 2015, 2016; Wang et al., 2017), but systematic gene excavation in mouthpart organs has not been done. Therefore, we were interested in determining how many types of chemosensory sensilla are in mouthpart organs and whether the mouthpart organs express abundant chemosensory proteins as in the antennae.

For a better understanding on the morphology of mouthparts, the microstructure was determined using an electron microscope scan experiment in this study. Further, we systematically investigated the chemosensory protein families in the labial palps and proboscis by transcriptome sequencing. The differentially expressed gene (DEG) analysis of all the candidate chemosensory genes and qRT-PCR analysis of candidate chemosensory receptor genes were performed to investigate the gene expression levels. This work contributes to the morphological and molecular studies on the mouthpart organs of H. armigera.

#### MATERIALS AND METHODS

#### Insect Rearing and Tissue Collection

The larvae of H. armigera were fed with an artificial diet and kept in the conditions of 16:8 h (light:dark) photoperiod, 27◦C ± 1 ◦C and 50–60% RH at the Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China. Male pupae were kept separately from females. The moths were fed on 10%

honey water after emergence. For transcriptome sequencing, the proboscis and labial palp were collected separately from the 1- to 3-day-old moths and then stored in liquid nitrogen immediately until they were used for experiments.

## Scanning Electron Microscopy and Sensillum Characterization

The proboscises from 1-day-old moth of eight females and eight males were tweezered from the base carefully and then were dehydrated in a series of ethanol (70, 80, 95, and 100%). After drying in a critical point drier (LEICA EM CPD), antennae were sprayed with gold (EIKO IB-3). The samples were then glued onto SEM stubs using a double graphite adhesive tape. Scanning was performed on a Hitachi SU8010 scanning electron microscope (Hitachi, Tokyo, Japan) at 10 kV. Sensillum types were characterized based on the description in the review about the proboscis sensillum types of the Lepidoptera by Faucheux (2013). The images were adjusted using Adobe Photoshop CS6 (Adobe Systems), but only the brightness and contrast. All figures were assembled in Adobe Illustrator CS5 (Adobe Systems).

## RNA Extraction and Transcriptome Sequencing

The total RNA of proboscis and labial palps was extracted following the manufacturer's instructions using TriZol reagent (Invitrogen, Carlsbad, CA, United States). The quality of RNA was measured using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States) and a NanoDrop ND-2000 spectrophotometer (NanoDrop products, Wilmington, DE, United States). One microgram of total RNA of each tissue (male and female mixtures) was used for generating a cDNA library, respectively. Construction of the cDNA library and Illumina HiSeq 2000 (Illumina, San Diego, CA, United States) sequencing was performed at the Beijing Genomics Institute (BGI, Shenzhen, China). The insert sequence length was around 200 bp, and these libraries were paired-end sequenced using PE100 strategy.

#### Assembly and Functional Annotation

After filtering low-quality reads, trimming low-quality nucleotides at each end, and removing 3<sup>0</sup> adaptors and poly-A/T tails from the raw reads, de novo assembly was conducted using Trinity. The clean reads of the proboscis and labial palps were fed to Trinity. The Trinity outputs were clustered by TGICL (Pertea et al., 2003). Unigene annotation was performed by NCBI BlastX against the database of non-redundant (nr) and SwissProt protein database with the E-value < 1e−5.

#### Identification of Chemosensory Genes

Putative chemosensory genes of six families (ORs, IRs, GRs, OBPs, CSPs, and SNMPs) were screened with a series of strategies. Sequences were extracted using chemosensory gene keywords by running Perl scripts against assembly and annotation files of transcriptomes on the server. After removing redundant sequences, the genes were further confirmed by BlastX against a local non-redundant database under the E-value < 1e−5. The ORFs of all genes were predicted using the ExPASy server<sup>1</sup> based on the BlastX best hit result (Gasteiger et al., 2003). Putative N-terminal signal peptides of OBPs and CSPs were predicted using the SignalP 4.0 server<sup>2</sup> with default parameters (Petersen et al., 2011).

## Sequence and Phylogenetic Analysis

Alignments of amino acid sequences were performed in MAFFT<sup>3</sup> . The phylogenetic trees of chemosensory genes were constructed using RAxML version 8 with the Jones–Taylor–Thornton amino acid substitution model (JTT) (Stamatakis, 2014), and 1000 bootstrap replicates were run to assess the node support. The OBP phylogenetic tree was constructed using a total of 134 OBPs of four Lepidoptera species: 45 from H. armigera including 39 identified in our dataset, 26 from Spodoptera littoralis, 30 from H. assulta, and 33 from Bombyx mori (Gong et al., 2009; Jacquin-Joly et al., 2012; Liu et al., 2012; Zhang et al., 2015; Chang et al., 2017). For CSPs, 82 sequences were used including 30 from H. armigera (including 26 from our data), 15 from H. assulta, 21 from S. littoralis, and 16 from B. mori (Gong et al., 2007; Jacquin-Joly et al., 2012; Liu et al., 2012; Li et al., 2015; Zhang et al., 2015; Chang et al., 2017). The phylogenetic tree of SNMPs was constructed using 21 sequences of 11 species from Diptera and Lepidoptera. Sequences of novel HarmOBPs and HarmCSPs are shown as **Supplementary File S1**.

## DEG Analysis

Differentially expressed gene analysis between proboscis and labial palps was conducted using a mapping-based expression profiling analysis according to the strategies described by Wang et al. (2017). The expression levels of chemosensory genes (ORs, IRs, GRs, OBPs, CSPs, and SNMPs) were estimated by fragments per kilobase million (FPKM) values (Trapnell et al., 2010). The heat map of differential gene expression between male antennae and female antennae in both species was generated by iTOL software<sup>4</sup> .

#### qRT-PCR Analysis and Statistical Analysis

The total RNA of four tissues including the antenna, proboscis, labial palps, and legs was extracted using TRIzol reagent (Invitrogen, CA, United States) according to the manufacturer's protocol. The cDNA of each tissue reverse transcribed from 1 µg total RNA using revert aid first-strand cDNA synthesis kit (Thermo Scientific, Waltham, MA, United States). The mRNA expression level of each gene (ORs, IRs, and GRs) was examined by qRT-PCR using GoTaq <sup>R</sup> qPCR Master Mix (Promega, WI, United States) and normalized by a reference gene HarmActin. PCR amplification was conducted using a ABI 7500 Real-Time PCR System (ABI, Vernon, CA, United States). The total volume

<sup>1</sup>http://web.expasy.org/translate/

<sup>2</sup>http://www.cbs.dtu.dk/services/SignalP/

<sup>3</sup>https://www.ebi.ac.uk/Tools/msa/mafft/

<sup>4</sup>http://itol.embl.de/

of each reaction was 20 µL, which contains 10 µL of GoTaq qPCR Master Mix, 1 µL of each gene specific primer (10 µM), 2 µL of cDNA, and 6 µL of RNase-free water. The PCR cycling condition was set based on the manufacturer's recommendations as follows: 95◦C for 2 min, 40 cycles of 95◦C for 15 s, and 60◦C for 50 s. A melting curve analysis was performed to confirm the amplification efficiency of each pair of primers. The primers were listed in **Supplementary Table S2**. The expression level of each was quantified using the comparative CrmT method (Schmittgen and Livak, 2008). The 1C<sup>T</sup> was obtained by subtracting the C<sup>T</sup> of HarmActin in a same tissue from that of a specific gene. The relative expression of each gene was evaluated by the values of 2−11CT, and the 11C<sup>T</sup> was normalized by the mean 1C<sup>T</sup> of at least three repetitions in one tissue, which has the smallest 1CT. The column diagram of each gene was constructed by GraphPad Prism 6 (GraphPad software Inc., La Jolla, CA, United States). The differences of expression among tissues and sexes were analyzed by one-way ANOVA and followed by Duncan's test (P < 0.05) using SPSS 22 (SPSS Inc., Chicago, IL, United States).

## RESULTS

## Morphological Structure of the Mouthpart Organs

Adults of H. armigera possess a typical siphoning mouthpart, which consists of two main organs: proboscis and labial palps (**Figure 1A**). The proboscis is coiled up completely and attached by a pair of labial palps on each side in the resting state (**Figure 1B**). When feeding or detecting, the proboscis is stretched out as a long tube and the labial palps are twisted around. The fine structure of the labial palps, which are prominent structures in the front of the head, has been studied by Zhao et al. (2013) in detail. The function of labial palps was considered to be closely related to its structure. We performed electron microscope scan on the proboscis to observe the morphology and structure.

#### The Overall Structure of the Proboscis

The proboscis is a tubular structure, which consists of the two elongated galeae (ge) (**Figures 1C,D**). The dorsal (dl) and ventral ligulae (vl) on each galea (**Figure 1E**) are joined together, which create the capillary construction for sucking liquids. The distal region (dt) was covered with abundant peg-shaped sensilla and appeared rough (**Figures 1C,D**). This area was equipped with all the three major types of sensilla: the two subtypes of sensilla styloconica (ss1 and ss2), one subtype of sensilla basiconica (esb2), and one subtype of sensilla chaetica (sch2) (**Figure 1F**). The proximal (px) and middle (md) sections of the proboscis possessed a smooth exocuticle, with numerous triangular cuticular processes (cp) (**Figure 1H**) together with two major types of sensilla: basiconica (esb1) and one sensilla chaetica (sch1) (**Figure 1G**).

#### Proboscis Sensilla in Adult H. armigera

A total of three major types of sensilla including nine subtypes were observed on the proboscis of male and female moths: sensilla styloconica (ss1 and ss2), sensilla chaetica (sch1 and sch2), and sensilla basiconica (esb1, esb2, esb3, isb1, and isb2).

#### **Sensilla styloconica**

A large number of sensilla styloconica were present (about 60 on each galea) on the proboscis, and they were arranged only in the distal region and were nearly perpendicular to the cuticula of the proboscis. Each sensillum was composed of a large peg-shaped protrusion with a large cavity inside and six ridges outside on the longitudinal direction. A lotus-shaped pedestal was present at the base of each sensillum (**Figure 2C**). A roof-shaped bulge was observed above each peg. Two subtypes have been identified according to the composition of each bulge. Sensilla styloconica type 1 (ss1) has a uniporous smooth cone (**Figure 2A**), the top of which possesses a pore of about 0.2 µm in diameter. The largest number of ss1 was observed on the distal region. The other subtype of sensilla styloconica, ss2 (**Figure 2B**), has a sphere on the tip. The surface of a single sphere was covered by a longitudinal groove, containing numerous micropores. Ss2 was the only multiporous sensilla type what we found on the proboscis, and a low number were interspersed among the ss1.

#### **Sensilla chaetica**

Sensilla chaetica was a cuspidal bristle-shaped structure with longitudinal lines on the surface. Two subtypes of sensilla chaetica were classified according to the features at their base. Both of the two subtypes were uniporous on the top. The base of sensilla chaetica type 1 (sch1) was aporous (**Figures 2D,E**). Each of them was inserted into a cupped socket and was located only in the proximal and middle regions. The length of these sensilla varied greatly from about 10 to 70 µm. The longer sch1 only existed in the proximal part of the proboscis. The shorter sch1 was scattered in the proximal and middle sections. Sensillum chaetica type 2 (sch 2) (**Figure 2F**) inserted its base into a roof-shaped bulge and had a similar pyramid appearance with the shorter type 1. This subtype of sensilla was only located on the distal part of the proboscis.

#### **Sensilla basiconica**

Sensilla basiconica was typically composed of a blunt, short peg-shaped cone with a terminal pore. Three subtypes were found on the external surface of the proboscis. Each external sensilla basiconica type 1 (esb1) (**Figure 2G**) inserted its base into cupped sockets and were present only on the proximal and middle parts. External sensilla basiconica type 2 (esb2) (**Figure 2H**) was located on a roof-shaped bulge and was only present on the distal section. External sensilla basiconica type 3 (esb3) had a uniporous peak and a curving cone (**Figure 2I**). This subtype only existed in the ventral side of the proximal galeae and has not been described in any adult noctuidae. We named them sensilla basiconica because they are similar to some previously reported basiconica-type sensilla (Xue et al., 2016). Furthermore, the two subtypes of sensilla basiconica

FIGURE 1 | General morphology of the mouthpart organs and ultrastructure of the proboscis of adult Helicoverpa armigera. (A) Frontal view of the head shows the major siphoning mouthpart organs: proboscis (pr) and the pair of labial palps (lp). (B) The proboscis (pr) coiled up under resting state; one labial palp attached on the side (the other one was removed). (C–H) Scanning electron micrographs of proboscis. (C) Overall structure of proboscis: the rough distal section (dt) and the smooth proximal (px) and middle (md) sections were shown on the two elongated galeae (ge). (D) The distal section (dt) of two galeae shows many peg-shaped sensilla. (E) Dorsal (dl) and ventral ligulae (vl) on each galea. (F) Two major sensilla on the distal section: sensilla styloconica (ss), external sensilla basiconica subtype 2 (esb2) and sensilla chaetica subtype 2 (sch2). (G) Two types of sensilla on the proximal and middle sections: external sensilla basiconica subtype 1 (esb1) and sensilla chaetica subtype 1 (sch1). Plenty of cuticular processes (cp) arranged on the surface. (H) Triangular structure of cuticular processes.

were identified on the internal surface of the proboscis; both were low in number. Internal sensilla basiconica type 1 (isb1) (**Figure 2J**) possessed a similar cone with esb3 but had a cylindrical depression at the base. The morphology of internal sensilla basiconica type 2 (isb2) (**Figure 2K**) was the same as esb1.

fphys-09-00970 August 7, 2018 Time: 14:22 # 5

FIGURE 2 | Scanning electron micrographs of sensilla on the proboscis of adult H. armigera. (A) Sensilla styloconica subtype 1 (ss1) possessing a uniporous cone. (B) Sensilla styloconica subtype 2 (ss2) with a multiparous sphere. (C) The lotus-shaped pedestal of sensilla styloconica. (D) Long and short sensilla chaetica subtype 1 (sch1) on the proximal section. The cupped socket at the base of sch1 (white box). (E) Short sensilla chaetica subtype 1 (sch1) on the middle section. (F) Sensillum chaetica type 2 (sch 2) on the distal part of the proboscis. (G) External sensilla basiconica subtype 1 (esb1) with a basal socket and a top pore. (H) External sensilla basiconica subtype 2 (esb2) on a roof-shaped bulge and with a pore on the tip. (I) External sensilla basiconica type 3 (esb3) with an uniporous peak and a curving cone. (J) Internal sensilla basiconica type 1 (isb1) on the internal surface of the proboscis tube. (K) Internal sensilla basiconica type 2 (isb2).

## Identification of Chemosensory Genes in the Mouthpart Organs

A great number of sensilla with various morphologies have been described in the mouthpart organs earlier. Subsequently, further research on chemosensory genes was conducted by transcriptomics.

#### Sequencing and Assembly

fphys-09-00970 August 7, 2018 Time: 14:22 # 7

The mouthpart transcriptome of adult H. armigera was obtained through Illumina Hiseq2000. A total of 99,606,218 and 108,678,674 raw reads were obtained from the proboscis and labial palp transcriptomes, respectively. Then, 97,650,394 and 106,569,066 clean reads with a Q20 percentage of 98.45 and 98.38%, respectively, were assembled into 88,983 and 116,096 contigs, respectively, using Trinity assembler. Finally, 43,405 unigenes were assembled by combining the data of proboscis with labial palp. This dataset consists of 43,405 unigenes including 14,297 distinct clusters and 29,108 distinct singletons with a mean length of 1,256 nt and N50 of 2,578 nt. A blastx algorithm against the NCBI non-redundant protein database revealed that 23,563 unigenes shared sequence similarities with known proteins using (cutoff E-value of 10−<sup>5</sup> ). Homology analysis with other insect species indicated that the dataset shared the best match with B. mori (26.5%), followed by Danaus plexippus (15.50%), and Papilio xuthus (1.46%).

#### Identification of Candidate Chemosensory Genes **Chemosensory receptors**

Four candidate ORs, based on a series of strategies, were identified through transcriptome analysis. All of these genes turned out to be previously reported ORs by Blast homology analysis. The reported co-receptor HarmOrco, performing function by co-expressing with specific OR, was identified with a complete open reading frame (ORF). Partial sequences of HarmOR24, HarmOR30, and HarmOR58 were obtained (**Supplementary Table S1**). HarmOR58, which was detected only in larval antenna by previous reports, was also found here. A total of six transcripts of candidate IRs were identified in the mouthparts. Blast homology analysis indicated that they belong to the previously reported 21 IRs. Complete ORFs of three IRs (HarmIR25a, 76b, and 41a) were obtained, and the sequences of the other three IRs (HarmIR75d, IR75p, and IR75p.1) were partial (**Supplementary Table S1**). Seven candidate GRs were screened in our dataset including four long sequences, two of which had complete ORFs. All of them were identified as the known GRs with identities from 98 to 100% according to the Blastx homology analysis (**Supplementary Table S1**). A phylogenetic tree of the seven GRs was generated (**Figure 3C**). HarmGR1- GR3 belonged to the reported CO<sup>2</sup> receptor clade; HarmGR5, GR6, and GR8 were part of the "sugar" receptor group; and HarmGR180 was part of the "bitter" receptor subfamily, which was suggested to be the most extended subfamily. The transcript levels of each receptor gene were initially estimated based on the FPKM values. HarmORco was expressed in the proboscis with the highest level followed by OR30, OR24, and OR58. In the labial palp, unexpectedly, HarmOR30 and OR58 had the most abundant expression (**Figure 3**: A-heat map). The heat map of the six IRs revealed that HarmIR75p had the highest expression level in the proboscis followed by HarmIR76b. In the labial palps, HarmIR25a was expressed at a higher level than the other five IRs (**Figure 3**: B-heat map). For the GRs, their expression in proboscis and labial palps exhibited three patterns. HarmGR1, GR2, and GR3 were mainly expressed in labial palps, whereas HarmGR5, GR5, and GR8 were mainly expressed in proboscis. HarmGR180 has similar expression in both two tissues (**Figure 3**: C-heat map).

To confirm the DEG results of the three families of receptor genes, we performed qRT-PCR in four major tissues including the antenna, proboscis, labial palps, and legs of both sexes. All the ORs were detected in all the four tissues. Most of the ORs were expressed in the antenna with significant higher level than the other tissues (P < 0.05) except HarmOR58, which was expressed in labial palps with a greater abundance than in other tissues but no significant difference (P > 0.05) (**Figure 3**: A-histogram). For the IRs, the expression of HarmIR25a, 76b. 41a in the antenna was significantly higher than that in others tissues (P < 0.05). The expression patterns of HarmIR75d, 75p, 75p.1 turned out to be diverse. HarmIR75d was mainly expressed in the antenna and proboscis; HarmIR75p was expressed in all the tissues with no significant difference; in particular, HarmIR75p.1 was mainly expressed in the labial palps and legs (**Figure 3**: B-histogram). Most of GRs were abundantly expressed in proboscis or labial palps. The expression of HarmGR1, GR2, and GR3 in labial palps was significantly higher than that of other tissues (P < 0.05), and that of HarmGR5 in proboscis was significantly higher than other tissues (P < 0.05). HarmGR6 and GR8 were mainly expressed in antenna and proboscis. HarmGR180 was mainly expressed in the antenna, and its expression level in other tissues was similar to each other (**Figure 3**: C-histogram).

#### **Abundant expression of soluble proteins**

We identified 39 OBP and 26 CSP transcripts of two small soluble protein groups. Eight novel OBPs were found, together with 31 previously reported genes (**Supplementary Table S1**). A total of 26 of 39 OBPs were identified as full-length sequences with complete ORFs and 34 amino acid sequences with signal peptides. A phylogenetic tree was constructed using 134 OBPs from four Lepidoptera species including the 39 transcripts identified in mouthpart organs. These OBPs were generally clustered into three subfamilies (**Figure 4A**). The "classic" OBP group contained the most members including general odorant-binding protein (GOBP) and pheromone-binding protein (PBP) with six conserved cysteines. Members of the "minus-C" group had only four cysteines, whereas the "plus-C" group had more than six cysteines (Zhou et al., 2004; Gong et al., 2009; Gu et al., 2015; Chang et al., 2017; Wang et al., 2017). The four novel OBPs (HarmOBP39, 43, 44, and 45) together with 22 reported OBPs were part of the "classic" OBP group, HarmOBP38 belonged to the "minus C," and the three novel OBPs (HarmOBP40, 41, and 42) belonged to the "plus C" groups. Sequence alignment (**Supplementary Figure S1A**) showed the same pattern as the phylogenetic tree except for HarmOBP9, which was clustered into the "classic" OBP clade, although it has only five conserved cysteines. Transcript levels of the identified OBPs were initially

estimated based on the FPKM values. The majority of OBPs were expressed in the proboscis or labial palps at a high level (**Figure 4B**). More OBPs including HarmOBP5, OBP9, OBP1, and OBP24 were found in the proboscis with a higher expression level than in labial palps. HarmOBP5 had the highest FPKM value in labial palps, followed by HarmOBP9. The expression level of the eight novel genes was lower except for HarmOBP40.

Six novel CSPs with the addition of 20 reported genes were identified in the transcriptome of the proboscis and labial palps. A total of 20 of 26 CSPs had complete ORFs. Further analysis showed that 24 CSPs covering the six novel genes had signal peptides on the N-terminal end of their amino acid sequences (**Supplementary Table S1**). A phylogenetic tree of 82 CSPs in H. armigera, H. assulta, Spodoptera littoralis, and B. mori was inferred to investigate the homology among sequences. It was shown that the six novel CSPs in our study were orthologous with those in other species (**Figure 5A**). Sequence alignment suggested that all 26 CSPs contained four conserved cysteine residues except HarmCSP25, for which the ORF was partial (**Supplementary Figure S1B**). The expression level of each CSP was visualized by the heat map based on the FPKM values (**Figure 5B**). The expression level of several CSPs (HarmCSP4, 27, 2, 6, 7, 9, 1, 5, 15, and 25) was extremely high in both proboscis and labial palps. The expression levels of all six novel genes were lower. HarmCSP4 was expressed in the proboscis at an especially high level, and the FPKM value was 109,757. In labial palps, the most abundantly expressed gene was HarmCSP2.

#### **Sensory neuron membrane proteins**

The two reported SNMPs (SNMP1 and SNMP2), which were first identified in the antenna, were identified in our dataset with complete ORFs (**Supplementary Table S1**). A phylogenetic tree of 21 reported SNMPS in 11 species revealed two separated

clades of SNMP1 and SNMP2 (**Figure 6A**). The transcript level of the two SNMPs based on the FPKM values in different tissues suggested that the expression level of SNMP2 was very high in both proboscis and labial palps, whereas the expression level of SNMP1 was very low (**Figure 6B**).

## DISCUSSION

In the last few decades, many studies on the host recognition of insects have been performed using molecular biology methods, and much attention has been focused on the antenna, which is regarded as the primary olfactory organ. The mouthparts, however, also play crucial roles in biological activity such as finding host plants or feeding. The general morphology of the proboscis in H. armigera has been described in previous studies (Blaney and Simmonds, 1988; Wang et al., 2012). Here, we investigated the fine structure of the proboscis of adult H. armigera in detail. A total of nine subtypes belonging to the three major types of sensilla (sensilla styloconica, sensilla basiconica, and sensilla chaetica) were identified on the proboscis, and three subtypes (ss2, esb3, and isb1) were identified for the first time.

The most abundant sensilla were found at the terminal section, where fluid can be sucked up. Two subtypes of sensilla styloconica were identified according to the characters of their tips. Subtype 1 (ss1) possessed a cuspidal cone on top with a terminal uniporous. This type of sensilla was considered to be one of the most common types among most lepidopterans (Faucheux, 2013). In three noctuidae moths including H. armigera, the function of this type was previously identified as contact chemoreception by Blaney and Simmonds (1988). The sensilla responded to several substances such as "sugars" (glucose, fructose, sucrose, and others), nicotine, and amino acids (gamma-aminobutyric acid) (Blaney and Simmonds, 1988). The subtype 2 (ss2) was multiporous on the wall of the tip sphere. They were located among the top uniporous subtype 1 at a low number. This type of sensilla was first found in H. armigera and was rarely described in other noctuidae species, which could be due to the small number of sensilla or the resolution ratio of images. The subtype 2 was similar to the uniporous-multiporous sensilla styloconica (UP-MP ss) and possessed a terminal pore and wall pores at the terminal structure probably as the combination of gustatory and olfactory sensilla (Faucheux, 2013). The subtype 2 on the proboscis of H. armigera was wall-multiporous but without the top uni-pore. We theorized that this type of sensilla functioned as olfactory chemoreceptors, which may sense plant volatiles before finally sucking food. Sensilla chaetica subtype 1 (sch1) was also described by a previous study but was wrongly classified as "trichodea" due to their long hair-like outlines (Wang et al., 2012). The typical characteristic of sensilla chaetica in most Lepidoptera, however, is the longitudinal ridge surface and the basal socket (Faucheux, 2013; Xue and Hua, 2014). Sensilla basiconica esb3 and isb1 were described for the first time in H. armigera.

After identifying various sensilla types on the proboscis and labial palps, which suggest a comprehensive chemosensory system in the mouthpart organs of H. armigera, we then systematically mined the candidate chemosensory genes in the proboscis and labial palps. We obtained data of 4 ORs, 6 IRs, 7

GRs, 39 OBPs, 26 CSPs, and 2 SNMPs. We rarely detected ORs in the mouthpart transcriptome of adult H. armigera. Four of the 66 reported ORs were identified. HarmOrco, which was considered an atypical co-receptor, had the highest expression level in the proboscis and labial palps. HarmOR58, which has been identified as a larval antennal specific gene by previous work (Liu N.Y. et al., 2014), was also detected in all the four tissues but with the low expression level. These ORs in mouthpart organs might play roles beyond food finding.

As another class of chemosensory receptor, IRs were suggested to mediate detection of certain chemical stimuli, predominantly to acids, aldehydes, and amines (Benton et al., 2009). Studies on Drosophila melanogaster revealed that IR64a is co-expressed with IR8a to form a functional ligand-gated ion channel for acid sensing in vivo (Ai et al., 2013). IR84a-expressing neurons in D. melanogaster were activated by phenylacetic acid and phenylacetaldehyde, which were regarded as the signal of food sources and oviposition sites and contributed to courtship (Grosjean et al., 2011). Apart from olfaction sensing, some IRs were suggested to play versatile roles in taste perception (salt, amino acids, etc.) and temperature sensing (Rimal and Lee, 2018). Six IRs were identified in the mouthparts based on our dataset. HarmIR25a, which belongs to the most conserved clade of the IR family among species and acts as a co-receptor (Croset et al., 2010), exhibited the most abundant expression in labial palps. In contrast, research on Drosophila suggested that IR25a was involved in temperature sensing in the chordotonal organ (Chen et al., 2015). It can be speculated that HarmIR25a might play roles in temperature perception as the highly conserved properties of IR sequences among species. Analogously, HarmIR76b might be the receptor for sensing amino acids or salt based on the studies in Drosophila (Zhang et al., 2013; Ganguly et al., 2017). HarmIR75p exhibited the most abundant expression in proboscis based on the DEG analysis; however, the qRT-PCR results suggested the lower expression level than HarmIR76b and IR25a.

Within the GR family, there are two well-studied subsets: CO<sup>2</sup> receptors and "sugar" receptors. Seven GRs were identified in the mouthparts. HarmGR1, GR2, and GR3 have been reported as CO<sup>2</sup> receptors that are mainly expressed in labial palps. HarmGR1 and HarmGR3 have been reported to respond robustly to NaHCO<sup>3</sup> when they are co-expressed (Ning et al., 2016). HarmGR5, GR6, and GR8, which were mainly expressed in the proboscis, were part of the "sugar" GR clade. As mentioned earlier, electrophysiological experiments on the proboscis of H. armigera have demonstrated that many sensilla styloconica subtype 1 (ss1) respond to sugars, nicotine, and some amino acids. Previous studies have identified HarmGR4 and HarmGR9 as the receptors of several sugars (Xu et al., 2012; Jiang et al., 2015). These two GRs were not identified in the mouthpart organs based on our dataset, but HarmGR5, GR6, and GR8 belong to the same clade as HarmGR4. It could be that one of the three GRs we found, or their combination, is used for sensing sugar. HarmGR180 was part of the "bitter" receptor subfamily, which is the largest clade in the GR family (Xu et al., 2016). The only "bitter" GR might be the receptor of some alkaloids such as nicotine or some amino acids. Further, these sugar and bitter GRs are probably expressed in styloconica subtype 1.

We sequenced many small soluble proteins (39 OBPs and 26 CSPs), among which eight OBPs and six CSPs were identified for the first time. After the first OBP and CSP were discovered in the giant moth Antheraea polyphemus and D. melanogaster, respectively (Vogt and Riddiford, 1981; Mckenna et al., 1994), a large number of OBPs and CSPs have been identified in many insects. Certain OBPs and CSPs have been reported to move volatile molecules (Zhang et al., 2012; Li et al., 2013) to the membrane of chemosensory neurons, where transmembrane receptors (ORs, GRs, or IRs) are expressed. However, the reason for the large number of OBPs and CSPs in the mouthpart organs where a minority of receptor genes were found is unknown. The most likely explanation is that non-sensory functions were endowed to certain OBPs and CSPs beyond chemo-signal detection. It has been reported that OBP22 of mosquito Aedes aegypti was produced in the sperm and transferred to females (Li et al., 2008). Certain OBPs/CSPs were described in many activities including development, anti-inflammation, carrying visual pigments, insecticide resistance, and so on (see review of Pelosi et al., 2017).

The two subfamilies of insect SNMPs (SNMP1 and SNMP2), two transmembrane domain receptor proteins homologous to the mammalian CD36 receptor (a family of proteins whose members frequently interact with proteinaceous ligands) (Rogers et al., 2001; Jiang et al., 2016), were identified in the dataset with complete ORFs. Studies have shown that the SNMP1 subtype is co-expressed with PRs in pheromone sensory neurons and contributes to the sensitivity of pheromone sensing in several insect species. In contrast, SNMP2 was localized in the supporting cells of neurons (Benton et al., 2007; Forstner et al., 2008; Jin et al., 2008; Liu C. et al., 2014; Pregitzer et al., 2014; Jiang et al., 2016). The function of SNMP2 has not yet been identified. Based on our data, the transcript level was very high in both proboscis and labial palps, which suggests a role of SNMP beyond pheromone sensing.

In summary, these results suggest that the mouthparts are a mutually cooperative but functionally concentrated system. Our results contribute to the understanding of chemical signal recognition in mouthpart organs. Further functional studies about certain chemosensory proteins such as receptors, which were identified in proboscis and labial palps, need to be conducted. On one hand, these would help to investigate the physiological activities of moths when they are feeding. On the other hand, more target genes could be used in the pest management.

#### AUTHOR CONTRIBUTIONS

GW and ZH conceived the study. GW and YL acquired the grant, also participated in its design, coordination, and supervision. MG carried out the laboratory experiments with contributions from QC. MG analyzed the data and wrote the paper. All authors read the paper and gave final approval for publication.

#### FUNDING

This work was funded by the National Natural Science Foundation of China (31621064 and 31725023 to GW and 31672095 to YL) and Beijing Nova Program (Z161100004916119 to YL). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

#### ACKNOWLEDGMENTS

fphys-09-00970 August 7, 2018 Time: 14:22 # 13

We thank Ms. Liyan Yang (Institute of Plant Protection, Chinese Academy of Agricultural Sciences) for rearing insects. We also thank M. S. Liting Pan (Institute of Plant Protection, Chinese Academy of Agricultural Sciences) for taking micro photography. We are grateful to M. S. Xiangzhi Liang and Jinmeng Guo for technical assistance.

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00970/full#supplementary-material

FIGURE S1 | Alignments of amino acid sequences of OBPs and CSPs identified in the proboscis and labial palps of H. armigera. (A) Conserved cysteines of HarmOBPs were shown by C1–C6. Eight novel OBPs were marked by orange circles. (B) Conserved cysteines of HarmCSPs were shown by C1–C4. Six novel CSPs were marked by orange circles.

TABLE S1 | Sequence analysis of candidate chemosensory genes (4 ORs, 6 IRs, 7 GRs, 39 OBPs, 26 CSPs, 2 SNMPs) identified in the proboscis and labial palps of H. armigera.

TABLE S2 | Primers used in qRT-PCR.

FILE S1 | Unigenes of eight novel odorant binding proteins (OBPs) and six novel chemosensory proteins (CSPs) identified in the proboscis and labial palp.

and olfaction. PLoS Genet. 6:e1001064. doi: 10.1371/journal.pgen. 1001064



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Guo, Chen, Liu, Wang and Han. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Synergistic Transcriptional Regulation of Olfactory Genes Drives Blood-Feeding Associated Complex Behavioral Responses in the Mosquito *Anopheles culicifacies*

Tanwee Das De1,2, Tina Thomas <sup>1</sup> , Sonia Verma<sup>1</sup> , Deepak Singla<sup>1</sup> , Charu Chauhan<sup>1</sup> , Vartika Srivastava<sup>1</sup> , Punita Sharma<sup>1</sup> , Seena Kumari <sup>1</sup> , Sanjay Tevatiya<sup>1</sup> , Jyoti Rani <sup>1</sup> , Yasha Hasija<sup>2</sup> , Kailash C. Pandey 1,3 and Rajnikant Dixit <sup>1</sup> \*

<sup>1</sup> Laboratory of Host-Parasite Interaction Studies, National Institute of Malaria Research, Dwarka, India, <sup>2</sup> Department of Biotechnology, Delhi Technological University, Rohini, India, <sup>3</sup> Department of Biochemistry, National Institute for Research in Environmental Health, Indian Council of Medical Research, Bhopal, India

#### *Edited by:*

Fernando Ariel Genta, Fundação Oswaldo Cruz (Fiocruz), Brazil

#### *Reviewed by:*

Jeffrey A. Riffell, University of Washington, United States Mauro Mandrioli, Università degli Studi di Modena e Reggio Emilia, Italy Gustavo Bueno Rivas, University of Florida, United States

> *\*Correspondence:* Rajnikant Dixit dixit2k@yahoo.com

#### *Specialty section:*

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

*Received:* 20 December 2017 *Accepted:* 01 May 2018 *Published:* 23 May 2018

#### *Citation:*

Das De T, Thomas T, Verma S, Singla D, Chauhan C, Srivastava V, Sharma P, Kumari S, Tevatiya S, Rani J, Hasija Y, Pandey KC and Dixit R (2018) A Synergistic Transcriptional Regulation of Olfactory Genes Drives Blood-Feeding Associated Complex Behavioral Responses in the Mosquito Anopheles culicifacies. Front. Physiol. 9:577. doi: 10.3389/fphys.2018.00577 Decoding the molecular basis of host seeking and blood feeding behavioral evolution/adaptation in the adult female mosquitoes may provide an opportunity to design new molecular strategy to disrupt human-mosquito interactions. Although there is a great progress in the field of mosquito olfaction and chemo-detection, little is known about the sex-specific evolution of the specialized olfactory system of adult female mosquitoes that enables them to drive and manage the complex blood-feeding associated behavioral responses. A comprehensive RNA-Seq analysis of prior and post blood meal olfactory system of An. culicifacies mosquito revealed a minor but unique change in the nature and regulation of key olfactory genes that may play a pivotal role in managing diverse behavioral responses. Based on age-dependent transcriptional profiling, we further demonstrated that adult female mosquito's chemosensory system gradually learned and matured to drive the host-seeking and blood feeding behavior at the age of 5–6 days. A time scale expression analysis of Odorant Binding Proteins (OBPs) unravels unique association with a late evening to midnight peak biting time. Blood meal-induced switching of unique sets of OBP genes and Odorant Receptors (Ors) expression coincides with the change in the innate physiological status of the mosquitoes. Blood meal follows up experiments further provide enough evidence that how a synergistic and concurrent action of OBPs-Ors may drive "prior and post blood meal" associated complex behavioral events. A dominant expression of two sensory appendages proteins (SAP-1 & SAP2) in the legs of An. culicifacies suggests that this mosquito species may draw an extra advantage of having more sensitive appendages than An. stephensi, an urban malarial vector in the Indian subcontinents. Finally, our molecular modeling analysis predicts crucial amino acid residues for future functional characterization of the sensory appendages proteins which may play a central role in regulating multiple behaviors of An. culicifacies mosquito.

#### SIGNIFICANCE

Evolution and adaptation of blood feeding behavior not only favored the reproductive success of adult female mosquitoes but also make them important disease-transmitting vectors. An environmental exposure after emergence may favor the broadly tuned olfactory system of mosquitoes to drive complex behavioral responses. But, how these olfactory derived genetic factors manage female specific "pre and post" blood meal associated complex behavioral responses are not well known. Our findings suggest that a synergistic action of olfactory factors may govern an innate to prime learning strategy to facilitate rapid blood meal acquisition and downstream behavioral activities. A species-specific transcriptional profiling and an in-silico analysis predict that "sensory appendages protein" may be a unique target to design disorientation strategy against the mosquito Anopheles culicifacies.

Keywords: mosquito, host-seeking, blood feeding, behavior, olfaction

## INTRODUCTION

Mosquitoes are one of the deadliest living animals, transmitting a variety of infectious diseases such as malaria, dengue fever, chikungunya and zika fever worldwide. According to WHO report, malaria is one of the major vector-borne diseases that causes 212 million morbidity cases and more than 4 million mortalities (World Health Organization, 2015). WHO recognized that in India, malaria situation is more complex and puts an estimated socio-economic burden of \$1.94 billion annually (World Health Organization, 2015). Current tools to control and manage malaria face challenges due to the emergence of drug resistance in parasite and insecticide resistance in mosquitoes (Stein et al., 2009; Petersen et al., 2011; Winzeler and Manary, 2014; Cui et al., 2015; Liu, 2015; Sahu et al., 2015). Thus, alternative molecular tools are required to rule out the expanding vector population as well as parasite development.

One of the key molecular strategies under not-to-bite approach relies on the designing of a new class of molecular tools that are able to disorient/alter the adult female mosquitoes host-seeking behavior (Potter, 2014). Therefore, defining the molecular basis of host-seeking behavioral evolution and adaption to blood feeding by the adult female mosquitoes remains central to our understanding. This may probably be due to the complex interaction of genetic and non-genetic factors, driving mosquito navigation (Takken and Verhulst, 2013). In nature mosquitoes encounter many challenges to sustain in daily life viz. they rely immensely on their sense of smell (olfaction) for the majority of their lifecycle stages (Potter, 2014). The well-developed nasal system of mosquitoes is able to detect and discriminate thousands of different odor molecules and thus play an essential role in the facilitation of olfactory guided behavior. These complex behavioral events are largely mediated by the diverse chemosensory genes encoding odorant binding proteins (OBPs), odorant degrading enzymes (ODEs), odorant receptors (Ors) and other accessory proteins including sensory neuron membrane protein (SNMP) (Takken and Knols, 2010). Odorant binding proteins (OBPs), which are bathed within the sensillum lymph, are low molecular weight soluble proteins that mediate the first interaction of the olfactory system with the external world (Takken and Knols, 2010; Carey and Carlson, 2011; Brito et al., 2016). These globular protein molecules showed significant diversity within the same family and are believed to bind with a wide range of hydrophobic odorant molecules. After binding with the odor molecules, OBPs transport it to their respective olfactory receptors present on the olfactory receptor neurons (ORNs) (Takken and Knols, 2010; Fan et al., 2011; Martin et al., 2011). Olfactory receptors (OrX) of insects' are associated with the obligate receptor co-receptor (Orco) on the dendritic membrane of ORN for proper functioning (Takken and Knols, 2010). Orco is not only essential for dendritic trafficking and presentation of the OrX in the membrane but also facilitate the formation of odorant gated ion channels by structural alteration that is opened upon odorant binding (Zwiebel and Takken, 2004; Takken and Knols, 2010).

The genome sequencing of several Anopheline sp. facilitates the identification of different olfactory genes including OBPs and Ors from different mosquito species. Functional characterization of few Anopheline mosquitoes OBP genes (OBP1, OBP20, OBP7, OBP2, OBP48) highlights their role in host-seeking behavioral activities (Biessmann et al., 2005, 2010; Li et al., 2005; Sengul and Tu, 2008, 2010; Hoffman et al, 2012; Tsitsanou et al., 2013; Ziemba et al., 2013). Consequently, de-orphanization of several odorant receptors (AgOr1, AgOr2, AgOr8, AgOr5, AgOr65) from An. gambiae also showed their specificity to humanspecific odorant molecules (Hallem et al., 2004; Carey et al., 2010). After binding of the odorant molecules with their cognate receptors, the actual signal transduction cascade is initiated which involves either the activation of ligand-gated ion channels or stimulation of the secondary messenger pathway (Takken and Knols, 2010). In insects, including mosquitoes, a combinatorial coding mechanism of the olfactory system is believed to increase the sensitivity of the odorant reception, which enables them to respond to specific odorants (Martin et al., 2011; Andersson et al., 2015). Thus, it is plausible to hypothesize that prior blood meal, key interactions of odorants and their cognate receptors may have a significant influence on food choice decision and blood meal uptake process.

For a successful blood feeding event, an adult female mosquito needs to manage multiple behavioral coordinates including searching, locating, landing over a suitable host, followed by tracing the proper site to pierce and suck the blood within 2 min (Zwiebel and Takken, 2004; Benoit et al., 2011; Sim et al., 2012; McMeniman et al., 2014; Cardé, 2015; Van Breugel et al., 2015; Won Jung et al., 2015). Just after the piercing organ (proboscis), it is the salivary gland which mediates the immediate biochemical

**462**

interaction with the vertebrate blood and facilitate rapid blood meal uptake. Our recent study suggested that adult female mosquito's salivary glands are evolved with the unique ability of gene expression switching to manage meal specific (sugar vs. blood) responses (Sharma et al., 2015b), but the molecular nature of the olfactory and neuro-system in regulating the salivary gland function is yet to unravel.

Mosquitoes after taking a blood meal, need to enter into a new habitat favoring successful oviposition (Rinker et al., 2013; Day, 2016). In fact, after blood meal acquisition, mosquitoes undergo two major behavioral switching events; (i) searching for suitable site(s) for temporary resting and completion of blood meal digestion (∼30 h) which is necessary for egg maturation (48– 72 h); and (ii) finding a proper oviposition site for successful egg laying (Taparia et al., 2017). After completion of egg laying event, the adult female mosquitoes regain their host-seeking activity for a second blood meal to complete the next gonotrophic cycle (Takken et al., 2001; Rinker et al., 2013). Notably, "prior and post" blood meal associated habitats may have a significant difference in their physical, chemical and biological characteristics (Day, 2016), but the molecular basis that how olfactory-driven factors manage these complex events is still not well understood (Chen et al., 2017).

Immediately after mosquito emergence, an exposure to diverse environmental/chemical cues facilitate the maturation and learning of the olfactory machinery components (sensory appendages, maxillary palps and proboscis) to govern common innate behavioral activities such as nectar sugar feeding and mating in both the sexes (Takken and Verhulst, 2013; Lutz et al., 2017). However, it is yet not clear whether the mating events have any direct impact on the initiation of host seeking and blood feeding behavioral responses. Our recent finding suggested that quick-to-court protein may have a crucial role to meet the conflicting demand of sexual mate partner finding and/or a suitable vertebrate host finding by regulating the expression of unknown olfactory genes in adult An. culicifacies mosquito (De et al., 2017). In fact, the organization of the olfactory components is morphologically similar in both the sexes but carries unique structural differences which are responsible for discrete temporal peaks of activities to sense swarm and identify sex partner for a successful mating event (Pitts et al., 2011). However, in case of adult female mosquitoes, we opined that the evolutionary forces might have driven an extra specialization of the olfactory components such as proboscis, enabling rapid host seeking and blood feeding behavioral adaptation. In other words, we termed this highly sex-specific extra specialization as an "evolutionary speciality" which not only evolve adult female mosquitoes as a fast blood feeder but make them a potent vector for many disease pathogens. Once, a mosquito takes first blood meal it needs to manage major physiological activities linked to blood meal digestion and egg maturation. These physiological changes possibly may have another level of impact on olfactory perception to guide oviposition site finding behavior. We further hypothesize that first blood meal exposure must have a priming effect on the olfactory responses expediting the consecutive host seeking and blood feeding behavioral activities more rapidly than previous one.

To test and decode this evolutionary speciality, we performed RNA-Seq analysis of the complete olfactory system of adult female An. culicifacies mosquito, a dominant Indian malarial vector. A comprehensive molecular and functional annotation of RNA-Seq data unraveled a limited but remarkable change in the nature and regulation of unique sets of olfactory gene repertoire in response to distinct feeding status of the mosquitoes. Extensive transcriptional profiling of the selected transcripts showed biphasic and synergistic regulation under the distinct innate physiological status of the mosquitoes, possibly to facilitate and manage the complex host-seeking behavioral events. Finally, our structural bioinformatic analysis predicts the key residues of the selected sensory appendages proteins for future functional validation and characterization as a unique target to design disorientation strategy against the mosquito An. culicifacies, responsible for more than 65% malaria cases in India (Sharma and Dev, 2015).

## MATERIALS AND METHODS

**Figure 1** represents a technical overview of the current investigation.

#### Mosquito Rearing and Maintenance

A cyclic colony of the mosquito An. culicifacies, sibling species A and An. stephensi were reared and maintained at 28 ± 2 ◦C, RH = 80% in the central insectary facility as mentioned previously (Thomas et al., 2014; Sharma et al., 2015b). All protocols for rearing and maintenance of the mosquito culture were approved by ethical committee of the institute.

## RNA Isolation and Transcriptome Sequencing Analysis

Complete olfactory tissue which includes antennae, maxillary palp, proboscis and labium, were dissected from 0 to 1 day of age, 30 min post blood fed and 30 h post blood fed An. culicifacies mosquito and collected in Trizol Reagent. Total RNA isolated from the collected olfactory tissues of approximately 30 mosquitoes was pooled to form one single sample and a double-stranded cDNA library for each set of naïve, 30 min and 30 h post blood meal, was prepared by a well-established PCR-based protocol described previously (Dixit et al., 2011; Sharma et al., 2015b). Whole transcriptome sequencing of the olfactory tissue was performed using the Illumina MiSeq 2 X 150 paired-end library preparation protocol. The sequencing data analysis pipeline is shown in **Figure 1**. Briefly, raw reads from each set were processed for removing the adaptors and low-quality bases (<20). A denovo clustering using CLC Genomics Workbench (V6.2) (Zhu et al., 2014) was used to build final contigs/transcripts dataset with default parameters (contig length ≥ 200, Automatic word size: Yes, Perform Scaffolding: Yes, Mismatch cost: 2, Inserstion cost: 3, Deletion cost: 3, length fraction: 0.5, Similarity fraction: 0.8). Finally, assembled transcriptome was used for CDS prediction and annotation using transdecoder and BLASTX at e-value 1e−<sup>6</sup> , respectively.

For a comprehensive differential gene expression (DGE) analysis we used DESeq R Package as described earlier (Chen

et al., 2017). Briefly, the high quality reads for each sample were mapped on their respective set of CDS/transcripts and FPKM (Fragments Per Kilobase of Exon Per Million Fragments Mapped) values were calculated using following formula i.e., FPKM = 10∧9 x C / (N x L), where C is the number of reads mapped onto the CDS; N the total number of mapped reads in the experiment; and L is the number of base pairs in the CDS. The common hit accessions based on BLAST against NR database were identified for differential gene expression analysis. CDS were further classified as up and down-regulated based on their log fold change (FC) value, which was calculated by the using the formula: FC = Log<sup>2</sup> (Treated/Control). Because, DESeq calculates raw p-values using a negative binomial distribution accounting technical and biological variables, and later p-values are corrected for multiple testing using the Benjamini-Hochberg statistical procedure which controls false discovery rate (FDR). Transcripts pairs whose read numbers displayed a greater than two-fold difference with P < 0.05 was listed as differentially expressed genes.

#### Identification and Molecular Cataloging of Olfactory Genes in *An. culicifacies*

An initial BLAST search analysis predicted a total of 93 transcripts encoding putative OBP homologs from the olfactory transcriptome data of An. culicifacies mosquito. To predict additional OBPs, a merged OBPs database of mosquito and Drosophila was re-queried against An. culicifacies draft genome/predicted transcripts databases available at www.vectorbase.org and build up the final OBP catalog for phylogenetic analysis as detailed in the Figure S1. A PDB database homology search analysis and GO annotation was used to identify and catalog other putative olfactory receptor genes manually.

#### PCR Based Gene Expression Analysis

The head tissue containing the olfactory appendages of female An. culicifacies mosquito was dissected at different zeitgeber time point. The 24 h time scale of the LD cycle is represented as different Zeitgeber time (ZT) where ZT0 indicate the end of dawn transition, ZT11 is defined as the start of the dusk transition and ZT12 is defined as the time of lights off (Rund et al., 2013). At the same time other tissues such as. head (male, female), legs (male, female), brain, olfactory tissue (OLF), female reproductive organ (FRO) and male reproductive organ (MRO) of both An. culicifacies and An. stephensi mosquitoes were also dissected and collected in Trizol followed by total RNA extraction and cDNA preparation. Differential gene expression analysis was performed using the normal RT-PCR and agarose gel electrophoresis protocol. For relative gene expression analysis, SYBR green qPCR (Thermo Scientific) master mix and Illumina Eco Real-Time PCR machine were used. PCR cycle parameters involved an initial denaturation at 95◦C for 5 min, 40 cycles of 10 s at 95◦C, 15 s at 52◦C, and 22 s at 72◦C. Fluorescence readings were taken at 72◦C after each cycle. The final steps of PCR at 95◦C for 15 s followed by 55◦C for 15 s and again 95◦C for 15 s were completed before deriving a melting curve. Each experiment was performed in three independent biological replicates to better evaluate the relative expression. Actin or S7 gene was used as internal control in all the experiment and the relative quantification was analyzed by 2−11Ct method (Livak and Schmittgen, 2001). Differential gene expression was statistically analyzed using student t-test.

#### Blood Meal Time Series Follow Up

Figure S9 represents a technical overview of the blood meal follow up experimental protocol. Briefly, the olfactory tissues were collected from 25 to 30 adult female mosquitoes for both naïve sugar-fed and blood fed mosquitoes at different time points. Olfactory tissues collections were initiated from 0 to 1 day of naïve sugar-fed mosquitoes and proceed up to 6–7 days on every alternative day. After the 6th day, the adult female mosquitoes were offered first blood meal by offering a live animal (rabbit) and immediately collected olfactory tissues for 30 min time point. The full blood-fed mosquitoes were separated and kept in a proper insectary condition for further experiment. After collection of olfactory tissues at 30 h and 72 h post blood fed the gravid females were kept for oviposition and again dissected OLF tissues after 24 h of the egg laying event. Second blood meal was provided to the egg laid mosquitoes and final collection of OLF tissues was done after 30 h of 2nd blood meal. Initially, relative expression data were interpreted to evaluate a general response using one way analysis of variance (ANOVA) for multiple comparison, however, wherever required "test" sample data was compared with "control" data set and statistically analyzed using Student's t-test.

#### Structural Modeling of SAP1 and SAP2

The structure prediction analysis of SAP1 and SAP2 proteins from An. culicifacies was carried out through searching of a template for each query proteins against PDB database using BLASTP algorithm. Based on highest query coverage, identity and e-value, two best templates were selected for each used query sequence and thereafter, modeller9 v.13 was used for the building of 50 models for each query sequence using multiple templates. The best model was selected using DOPE (Discrete Optimized Protein Energy) score, which is favored by the lowest cumulative energy score for the whole structured model. The selected model was further validated by Ramachandran plot using PROCHECK software which estimates the stereo-chemical quality of the residues in allowed, disallowed and favorable regions. Finally, the selected models were used for binding site prediction using COACH software.

#### RESULTS

#### Blood Meal Causes Modest but Unique Changes to Olfactory Responses

To decode and establish the possible molecular relationship managing "prior and post" blood meal behavioral events we developed a working hypothesis (**Figure 2**), a plausible mechanism which may have a significant influence on mosquito feeding and survival in diverse ecologies. To test this hypothesis, first we generated and analyzed a total of ∼122 million RNA-Seq reads of the olfactory tissues collected from 1 to 2 day old


naive (Nv), 5- 6-day old immediate blood fed (30 m-2 h PBM) and 30 h post blood fed (30 h PBM) mosquitoes (**Table 1**). We chose 30 h PBM as a critical time when completion of blood meal digestion occurs in the midgut, which may have a direct influence on the reactivation of the olfactory system (Figure S2) (Gonalves et al., 2009; Rinker et al., 2013; Taparia et al., 2017). For molecular and functional annotation, we assembled each transcriptomic database into contigs/transcripts and compared against multiple molecular databases as described earlier (Sharma et al., 2015b). Supplementary Table 1 represents details of the annotation kinetics of mosquito olfactory databases.

To test whether blood meal alters the global expression pattern of the olfactory transcriptome, we performed a differential gene expression analysis. Initial attempt of mapping cleaned reads to the available draft reference genome failed to yield quality results, probably due to poor annotation (Figure S3). Alternatively, we mapped all the high quality reads against denovo assembled reference map, as described earlier (Sharma et al., 2015b). Blood meal causes a modest shift in the transcriptome expression (**Figure 3A**), supporting the previous report that first blood meal enhances odorant receptor transcripts abundance modestly, but causes general reduction of mosquito antennal chemosensory gene repertoire in An. gambiae (Rinker et al., 2013).

We observed that at least 85% transcriptome remains unaltered, while only ∼6% transcripts are up-regulated and ∼8.7% transcripts downregulated in 30 min post blood fed samples (Supplementary Table 2 and Dataset S1). As expected, ∼10% transcripts expression was further reduced in 30h post blood fed olfactory tissue samples while only 2% transcripts were up-regulated when compared to naive sugarfed mosquitoes (Supplementary Table 2). Interestingly, a comprehensive annotation analysis also predicted that basic composition of the mosquito olfactory tissue does not alter significantly (**Figures 3B–D**). This observation allowed us to further hypothesize that blood-feeding may not directly cause a major shift in transcript abundance but may alter the functional nature/regulation of the unique transcripts controlling key biological processes such as response to stimulus, circadian rhythm and signaling in the blood fed adult female mosquitoes (**Figures 3B–D**). To clarify this complexity, we manually shortlisted the olfactory transcripts either based on their FPKM abundance and/or predicted coding nature and analyzed a set of unique genes likely to influence mosquito host-seeking and blood-feeding behavior. To trace the possible molecular link, we extensively profiled their transcriptional regulation under distinct feeding status (see below).

#### Daily Rhythm and Expression Change of Odorant Binding Proteins (OBPs) May Influence Olfactory Responses

To negotiate and manage the navigation trajectory toward the vertebrate host, olfactory encoded odorant binding proteins (OBPs) play a crucial role to bind and deliver the odorants/chemicals to their cognate odorant receptors, an event guiding behavioral decisions. To explore the possible role of OBPs in the regulation of the olfactory behavior we identified and cataloged a total of sixty-three OBP genes by homology search analysis from the mosquito An. culicifacies (**Table 1A**). Domain prediction analysis classified the OBPs as Classic OBPs, Plus-C OBPs, Two-domain OBPs and other Chemosensory protein family (**Table 1B**; details in Supplementary Table 3), as described earlier for the mosquito An. gambiae (Manoharan et al., 2013).

A comprehensive phylogenomic analysis of the Classic putative OBPs of An. culicifacies highlights the conserved sequence relationship with An. gambiae and other mosquito/insect species (Figure S4A). Whereas, Plus-C OBPs and more dominantly Atypical OBPs seem to be unique to the mosquitoes suggesting their possible involvement in the evolution and adaptation of blood feeding behavior of adult female mosquitoes (Figures S4B,C).

Interestingly, differential gene expression (DGE) data indicated that blood meal restricted the expression of common OBP transcripts (Figure S5). However, first blood meal causes the appearance of unique OBP transcripts (**Table 1**), a crucial event in modulating the behavioral activities in response to change in the feeding status i.e., naive sugar to blood feeding. To further validate and unravel this unique relationship of OBPs regulation, we examined the RT-PCR based expression of at least 11 putative OBP transcripts under distinct feeding status of the mosquitoes. In this analysis, we also included two chemosensory proteins (CSPs) named sensory appendage protein (SAP1 & SAP2) having a dominant expression in the naive mosquito olfactory tissue (Supplementary Table 3).

Our Zeitgeber time scale expression showed that out of tested nine OBPs transcripts, at least 6 OBP transcripts showed a >2-fold modulation in their expression during late evening to midnight, in the 6-day old naïve mosquitoes (**Figure 4A**). These data also corroborate with the previous observation that the natural active biting behavior of An. culicifacies mosquito occurs in the mid-night (Singh et al., 1995; Basseri et al., 2012). Surprisingly, sensory appendage proteins (Ac-SAP1 & Ac-SAP2) showed unequivocally an enriched (16-fold for SAP1, p ≤ 0.001 and 6-fold SAP2, p ≤ 0.0001) expression than other tested OBPs. Apparently, we also observed a transient suppression (30 min)

and rapid recovery of OBPs expression just after a first blood meal (**Figure 4B**). However, surprisingly, OBP7 showed a unique pattern of a consistent up-regulation till the 6th day when compared to a gradual enrichment of other tested OBP and SAP after 3-day post-emergence in the naive adult female mosquitoes. However, it is yet to be clarified whether an early enrichment of OBP7 has any important role in aging mosquitoes' olfactory responses.

#### Innate Physiological Status May Influence Olfactory Receptor Responses to Manage Behavioral Switching Events

A transient modulation of OBPs expression in response to blood meal further prompted us to decode and establish its correlation with the olfactory receptors. To unravel this relationship, initially we retrieved, pooled and cataloged a total of 603 unique transcripts linked to response to stimulus and signaling (RTSS) categories (**Figures 3B–D**), encoding diverse nature of proteins such as anion binding, nucleic acid binding, receptor activity, hydrolases and transferase activity (**Figure 5A**). A comparative GO score distribution analysis predicted lower score hits for the blood-fed cohorts than naive mosquitoes (**Figure 5A**). Surprisingly, out of 603 transcripts, we noticed only 110 transcripts were common to all, while >100 transcripts remain uniquely associated with individual physiological conditions compared in the study (**Figure 5B**).

Olfactory receptors play a central role to receive and communicate the initial chemical message to the higher brain center through ORNs for decision-making events. Thus, we made a catalog of 50 different chemosensory receptors (**Table 2**), comprising odorant receptors (Ors); gustatory receptors (Grs) and variant ionotropic receptors (Irs), which appeared predominantly in the naïve and blood fed cohorts of the RTSS category (Supplementary Table 4). Interestingly, a cluster of 19 different olfactory receptor genes was found to be expressed abundantly and exclusively in the naïve mosquito (Supplementary Table 4). At the same time, we also observed that a distinct repertoire of chemosensory receptor genes uniquely appeared in the blood fed cohorts, but their number is much lower than the naïve mosquito (Supplementary Table 4). Observation of the constitutive expression of Orco and few other Ors and Grs (totaling 10 transcripts) in all the experimental conditions highlighted the importance of Orco for the presentation of other receptors in the olfactory system.

Unlike OBPs, poor modulation of olfactory receptor gene expression under circadian rhythm (**Figure 5C**) suggested their minimal role in the initialization of host-seeking behavioral activities. Alternatively, we also interpreted that Ors may not have direct biphasic regulation, but may influence a successful blood

FIGURE 4 | Transcriptional profiling of the odorant-binding protein genes (OBPs) under different circumstances. (A) Rhythmic expression of OBP genes in the adult female's olfactory tissues (OLF) according to different zeitgeber time (ZT) scale, where ZT0 indicate the end of dawn transition, ZT11 is defined as the start of the dusk transition and ZT12 is defined as the time of lights off. (B) Relative expression profiling of OBP genes in pre and post blood fed olfactory tissues. Olfactory tissues (OLF) were collected from 1, 3, and 6 day old sugar-fed mosquitoes which were then provided blood meal and then the olfactory tissues were collected after 30 min of post blood fed and 30 and 72 h of post blood-fed mosquitoes. The significance of suppression of OBP genes expression after 30 h of post blood meal are as follows: SAP ≤ 0.004; SAP2 ≤ 0.039; OBP7 ≤ 0.007; OBP20 ≤ 0.0004; OBP10 ≤ 0.003.

feeding event. To further corroborate with the above propositions and uncover the functional correlations of olfactory receptor responses, we monitored the transcriptional regulation of the selected Or transcripts in response to two consecutive blood meal series follow-up experiment. An age-dependent enrichment of Or transcripts till 6th day of maturation in the sugar-fed mosquitoes suggested that naïve mosquitoes may express and attain a full spectrum of chemosensory genes expression to meet all the needs of their life cycle requirements i.e., host-seeking and mate-finding behavioral response (**Figure 5D**).

First blood meal to the 6th-day old naïve mosquitoes initiates the suppression of almost all the olfactory receptor transcripts within 30 min of blood feeding, whose expression almost ceased to a basal level at 30 h post blood meal, except the slight upregulation of two transcripts named Or42 and Or62 (**Figure 5D**). Apparently after 30 h PBM, we observed a significant modulation of the receptor gene expression which started enriching till 72 h of post first blood meal, a time window coincides with the successful completion of the oviposition event. However, we did not observe any significant change in the expression of the receptor transcripts in response to second blood meal (**Figure 5D**).

#### Blood Meal Response to Other Olfactory Proteins

Encouragingly, the above data prompted us to test transcriptional profiling of few uncharacterized chemosensory class of olfactory proteins, identified from the transcriptomic data. Transcripts encoding orphan receptor R21, scavenger receptor class B (SRCB), an uncharacterized Protein (XP\_001959820) and Sensory neuron membrane protein (SNMP) showed a similar pattern of regulation, suggesting that a combination of all the receptor type represented in the olfactory tissue of An. culicifacies mosquito function concurrently in nature's aroma world and changed significantly prior and after the first blood meal as compared to the consecutive second blood meal (**Figure 6A**). Though, the involvement of G-proteins and related metabotropic signaling mechanism in the olfactory signal transduction of insects remain controversial, however, our observation of a rapid and consistent induction of adenylyl cyclase gene after 30 m PBM (**Figure 6B**), supported the previous hypothesis that the synthesis of the secondary messenger, cAMP by adenylate cyclase, facilitates odorant mediated signal transduction process which further influence downstream behavioral responses (Takken and Knols, 2010). Surprisingly, finding of <1% of transcripts encoding putative immune proteins suggested that the maintenance of a basal level of sterility is essential for proper olfactory functions (Figure S6).

#### Sensory Appendages Proteins as a Unique Target to *Anopheles culicifacies*

To test whether any species-specific olfactory derived genetic factors have any differential regulation likely to influence the behavioral responses, we compared the expression of at least 6 OBPs transcripts between two laboratories reared mosquito species An. stephensi and An. culicifacies. Though, both are dominant malaria vectors in urban and rural India, respectively, but display a significant difference in their behavioral properties such as feeding, mating, biting preferences etc., (personal observation/ST-S5). In this analysis, we also included two SAP proteins, which showed a high induction than other OBPs in the olfactory system of An. culicifacies mosquito at midnight (**Figure 4A**). Surprisingly, a sex and tissue-specific comparative transcriptional profiling of selected OBPs revealed a dominant expression of SAP1 (p ≤ 0.0003)/SAP2 (p ≤ 0.0007) in the legs of An. culicifacies mosquito (**Figures 7A,B**). Together these data indicated that An. culicifacies may draw an extra advantage of having more sensitive appendages, possibly to favor more active late night foraging behavior, than An. stephensi. Though,

FIGURE 5 | Blood meal modulates odorant receptors expression. (A) A comparative GO score distribution analysis of the response to stimulus and signaling transcripts of naïve and blood-fed mosquitoes. (B) Venn diagram showing common and unique transcripts of response to stimulus and signaling GO category of naïve and blood-fed mosquitoes. (C) Rhythmic expression of olfactory receptor genes (Ors) of An. culicifacies in the olfactory tissues of female mosquitoes, where ZT0 indicate the end of dawn transition, ZT11 is defined as the start of the dusk transition and ZT12 is defined as the time of lights off. (D) Transcriptional response of olfactory receptor genes according to blood meal time series experiment. Olfactory tissues (OLF) were collected from naïve sugar fed adult female mosquitoes till 6th day (OLF-1D, OLF-3D, OLF-6D). Then mosquitoes were provided blood meal and again olfactory tissues were collected at a different time point after blood feeding, viz. OLF-30 M: 30 min post blood fed (PBM); 30 h-PBM: 30 h of PBM; 72 h-PBM: 72 h of post blood meal; then the mosquitoes were kept for oviposition (egg laying), and again the olfactory tissues were collected 24 h of post oviposition (24 h-POT). Finally, the 2nd blood meals were provided to the egg laid mosquitoes and collected olfactory tissue 30 h of second blood meal (30 h-PBM2). A multiple comparison analysis (ANOVA) revealed a significant modulation in the expression of each gene (mean significant p < 0.05). Further, the significance of suppression of OR genes expression after 30 h of post blood meal are as follows: Putative Or ≤ 0.001; Or42 ≤ 0.05; GR ≤ 0.003; Or44 ≤ 0.0002; IR41c ≤ 3.5E−05; Or62 ≤ 0.06; Or39 ≤ 0.02; Or9 NS.

defining the basis of host preference remains uncertain, because Anopheline mosquitoes have opportunistic feeding behavior which is largely influenced by nature of the host availability (Thiemann et al., 2011). A close association of Ac-SAP proteins with the anthropophilic Anopheline mosquitoes (**Figure 7C**, Figure S7A), strongly suggested that sensory appendages proteins may have a crucial role to meet and manage the high host seeking behavioral activities, restricted to An. culicifacies.

The above findings further prompted us to carry out a 3D structure modeling analysis of Ac-SAP1 and Ac-SAP2, to predict the best possible conserved binding pockets for specific chemicals. In the absence of any available solved X-ray structure of the reference SAP protein, we applied a template based comparative molecular modeling approach. An initial BLAST analysis identified two best templates in PDB database code for chemosensory protein 2GVS and 1KX8 with identity 47–56% and coverage >80%, favoring their suitability for structure prediction. TABLE 2 | Number of Odorant Receptors retrieved from the olfactory tissue of naive and blood fed An. culicifacies mosquito.


Out of the 50 modeled 3D structures for each protein, DOPE score analysis resulted in the selection of model-49 and model-27 with score −11689.73, and −10989.75 for SAP1 and SAP2, respectively (**Figure 7D**, Figure S7B).

We validated the best-selected model using Procheck server for Ramachandran plot, showing a more than 95% allowable region, with no residue falling in the disallowed region of the plot (**Figure 7E**, Figure S7C). Based on the consensus, a best-fit ligand binding site prediction analysis within the selected models was scored by COACH server, which engages at least five different algorithms TM-SITE, S-SITE, COFACTOR, FIND-SITE, and ConCavity. Binding pocket for SAP1 and SAP2 identified eight consensus residues namely D36, E39, L40, K49, C52, Q59, Y91, and Y95 along with BDD (12-bromo-1-dodecanol) as a predicted ligand. Minimization of the steric clashes from the complex structures was done using Chimera software (**Figure 7F**, Figure S7D). Furthermore, selection of amino acid residues within 3 Å region of ligand molecule are I43 and Y95 of which Y95 is involved in H-bonding with BDD ligand. Similarly, in case of SAP2 protein, residue selection resulted in the identification of I43, D51, Q59, T63, Y95 residues of which D51 form H-bond with BDD ligand (**Figure 7G**, Figure S7E).

In our analysis, we observed the presence of at least two conserved cysteines (CYS52 and CYS55) residues in the loop region of SAP1 and SAP2 proteins, which may likely have involved in di-sulfide bond formation and stabilization of protein structure. Our analysis also showed that binding pocket forms a tunnel-like structure which is preferred by long aliphatic molecules. Presence of negatively charged aspartic and glutamic acid at both ends showed the preference for charged residue near the vicinity of ligand molecule. Moreover, the presence of conserved negatively charged aspartic acid and polar tyrosine (TYR91 in SAP1 and TYR95 in SAP2) at one end of binding pocket suggested their role in ligand binding.

#### DISCUSSION

It is well known that a circadian dependent modulation of olfactory responses significantly influences the complex behavioral responses in both sexes of Anopheline mosquito species (Rund et al., 2013). However, the evolution of the more specialized olfactory system of adult female mosquitoes favored their unique adaptation to blood feeding behavior. The female olfactory system comprises of three olfactory appendages i.e., (a) the antennae, (b) the maxillary palps and (c) female specific proboscis, which may encode a variable number of factors responsible for maintaining female specific daily olfaction rhythms such as host seeking, blood feeding, and oviposition behavior. Since, peripheral antennal appendages harbor more diverse OBPs, Ors, and other factors, it acts as the principle chemosensory organ that respond to wide range of volatile odors. Therefore, major electrophysiological and molecular studies have been focussed on this olfactory component. A few recent studies examining global profile change in response to daily rhythms and blood feeding highlighted the important role of the antennal transcripts/proteins in the modulation of distinct behavioral responses of Anopheline mosquitoes (Rinker et al., 2013; Rund et al., 2013; Chen et al., 2017). While the maxillary palps encode unique receptor proteins such as Or8, Or28, and Orco, which respond to carbon dioxide to enable mosquitoes for a successful navigation toward vertebrate hosts (Pitts et al., 2011). In the close vicinity of the targeted hosts, female mosquito's proboscis encoded factors rapidly engaged to complete the blood meal uptake process in less than 2 min. The molecular basis that how olfactory appendages encoded factors interlinked with each other

to drive highly sex-specific pre-and post-blood meal behavioral events are not well understood yet.

We have recently demonstrated that adult female mosquitoes are evolved with the unique ability of salivary gland gene expression switching to manage meal specific "prior and post" blood meal responses (Sharma et al., 2015b). Here, we further extended this idea to decode and trace the possible molecular link that how the olfactory factors of adult female An. culicifacies mosquitoes drive sex-specific host-seeking, blood-feeding and oviposition behavior. To establish the plausible mechanism of the olfactory system, we developed a working hypothesis (**Figure 2**) and compared the transcriptional response of the olfactory derived transcripts, modulating in responses to changes in the

potent OBP gene. (A,B) Sex and tissue-specific relative expression profiling of OBP genes in An. culicifacies (A) and An. stephensi (B). FOLF, female olfactory tissue (OLF); MOLF, Male OLF; FRO, Female reproductive organ; MRO, Male reproductive organ; FLeg, Female legs; MLeg, Male legs. OBP gene details: SAP, Sensory appendages protein 1; SAP2, Sensory appendages protein 2. (C) Phylogenetic analysis of An. culicifacies SAP1 (Ac-SAP1) gene. Color-coded circle represents the nature of the mosquitoes host preferences e.g., Red circle, Strongly Anthropophilic; Blue circle, Strongly Zoophilic; Yellow circle, Opportunistic Anthropohilic and/or Zoophilic; Black circle, moderate Anthropophilic; Gray circle, Moderate zoophilic; Blank Circle, Unknown; Green circle, non-blood feeder. (D) DOPE score analysis for SAP1. (E) Ramachandran Plot of SAP1 protein. (F) 3-dimensional protein structure of the Ac-SAP1 protein. (G) The binding site of SAP1 protein showed in space fill with nearby residues in stick form. \*\*\*p ≤ 0.0001.

feeding status. Surprisingly, an observation of a limited change in the global response of the olfactory system of An. culicifacies mosquito partly corroborates with the similar changes in the limited pool of antennal chemosensory genes in An. gambiae (Rinker et al., 2013). Taking in account of the nature of tissues i.e., the selected peripheral sensory appendages investigated in previous studies, we hypothesize that blood-feeding may not directly cause a major shift in transcript abundance, but may alter the functional nature/regulation of the unique transcripts controlling key biological processes. To unravel the molecular nature and function of the olfactory factors, we annotated, cataloged and selectively profiled the expression of OBPs, Ors and other members of chemosensory genes.

An initial comparison of the annotated transcripts revealed that first blood meal not only delimits the transcripts numbers but also enriches the expression of many unique transcripts having similar functions. Once reached to its saturation level, the expression of selected olfactory transcripts did not alter significantly, when offered an un-interrupted sugar meal to the aging mosquitoes (Figures S8A,B). Together, these data suggested that an abundant expression of olfactory receptors in naïve mosquitoes may be essential to encounter and manage different conflicting behavioral demands when changing from naïve sugar fed to blood fed status. Furthermore, a zeitgeber time scale experiments suggested that midnight hyper activities of OBPs, especially sensory appendages proteins (SAP-1 and SAP-2), are able to drive female specific host-seeking behavioral activities of naive adult female An. culicifacies mosquitoes, supporting the previous finding in other Anopheline mosquito species (Biessmann et al., 2005; Iovinella et al., 2013).

Our observation of a transient change in the expression of selected OBP transcripts, in response to first blood meal further raises a question that how mosquitoes manage blood feeding associated complex behavioral responses such as egg maturation, oviposition etc. After a successful blood meal, the gut physiology of the naive adult female mosquito undergoes a complex modulation to digest the blood meal and maturation of the eggs. Once the blood meal digestion completed, the mosquitoes may re-switch their olfactory responses for oviposition site finding behavior (Wong et al., 2011; Phasomkusolsil et al., 2013; Rinker et al., 2013; Lindh et al., 2015). Current literature suggested that a combinatorial coding mechanism of the olfactory receptors enables insects to recognize thousands of diverse chemical cues for selective neuro-actions to meet specific behavioral demands (Carey and Carlson, 2011; Martin et al., 2011; Andersson et al., 2015). Though previous studies suggested that first blood meal causes the alteration of OBP/Ors mediated odor sensitivity (Rinker et al., 2013), how olfactory receptors superintend and co-ordinate between innate and primed/adaptive odor responses remains largely unknown (Lutz et al., 2017). We hypothesize that a harmonious action of OBPs and Ors, which are involved in downstream odorant signal transduction cascade, may have significant influence on behavioral switching events.

To test this hypothesis, we profiled the expression of selected Ors transcripts in response to two consecutive blood meal follow up experiment, which included at least one gonotrophic cycle completion. Supporting the previous reports, we also observed that a first blood meal initiated a gradual suppression of all the olfactory receptor transcripts within 30 min of blood feeding, which was further ceased to the basal level at 30 h post blood meal. However, surprisingly, we observed a two-fold up regulation of all the receptor transcripts in response to second blood meal, when compared the expression after 30 h of first blood meal. Together, these data strongly suggested that first blood meal exposure to odorant receptors may have priming effect over host-seeking behavioral activities, enabling mosquito for rapid blood meal uptake for consecutive gonotrophic cycles.

Two most potent Indian malarial vector species An. culicifacies and An. stephensi have been reported to show predominantly zoophilic and anthropophilic behavior, respectively (Joshi et al., 1988; World Health Organization, 2007; Sharma and Dev, 2015), but the molecular basis of such biological variation is yet to unravel. Emerging evidence suggested that a significant genetic difference exists among various Anopheline mosquito species, including An. stephensi and An. culicifacies (Dash et al., 2007; Sharma et al., 2015a). Under laboratory investigation, we frequently observed that biological rhythm may have a significant influence on the biting and blood feeding behavior of An. culicifacies. Previously, several odorant binding proteins such as OBP20/OBP1/OBP7, SAP have been identified and characterized as a key molecular target in many Anopheline mosquitoes involved in host-seeking behavior (Biessmann et al., 2010; Sengul and Tu, 2010; Ziemba et al., 2013), but remains poorly understood in Indian vectors.

Therefore, to test whether species-specific olfactory derived genetic factors have any differential regulation, we compared the tissue expression of selected OBP transcripts between two laboratories reared mosquito species i.e., An. stephensi and An. culicifacies. Surprisingly, a higher expression of SAP-1 and SAP-2 in the legs of mosquito An. culicifacies indicated that this mosquito species may have drawn an extra advantage of having more sensitive appendages, possibly to favor more active late night foraging behavior than An. stephensi. A 3D molecular modeling analysis not only predicted the presence of at least two conserved cysteines (CYS52 and CYS55) residues in the loop region of SAP1 and SAP2 proteins but also suggested that binding pocket may form a tunnel-like structure, preferred by long aliphatic molecules. While the presence of conserved negatively charged aspartic acid and polar tyrosine at one end of binding pocket suggested their role in ligand binding. Though previously SAP has also been identified from other Anopheline mosquito species but their role in host-seeking and blood feeding behavior remains poorly understood (Biessmann et al., 2005; Iovinella et al., 2013). Encouraged by the above observation, we selected SAP as a unique target that may be crucial to design an effective disorientation strategy against An. culicifacies mosquito, an important malaria vector in rural India.

## CONCLUSION

Decoding the genetic relationship of sense of smell is central to design new molecular tools to disrupt mosquito-human interaction. We demonstrated that a synergistic and harmonious action of olfactory encoded unique factors govern the

FIGURE 8 | How smart actions of olfactory system manages blood feeding associated odor response: an evolutionary speciality of adult female mosquitoes. After emergence from pupae adult mosquitoes are exposed to the overwhelmed odor world, where odorants chemicals act as a language of communication with the external world. The sophisticated innate olfactory system of mosquitoes enables them to recognize and differentiate this wide variety of odorants which are crucial for their every life cycle stages. Inner physiological motivation, as well as the age and exposure of mosquitoes toward the external world, promote them for host seeking and blood feeding event. After taking blood meal mosquitoes initiate next level of physiological cum behavioral events i.e., oviposition. Apart from that, first exposure to vertebrates facilitates learning and second blood feeding events. These whole odors mediated response is tactfully managed by the synergistic actions of Odorant binding proteins (OBPs) and olfactory receptors (Ors). The overlapping circadian rhythm dependent functions of OBPs and Ors govern the pre-blood meal events of host fetching events. As soon as the mosquitoes take blood meal the functions of OBPs and Ors ceased for some period, but the recovery of OBPs actions occurs early as compared to Ors to perform the next level of behaviors. Mosquitoes, then take advantage/adapted from priming and learning of the first blood meal exposure for the more rapid consecutive blood feeding.

successful "prior and post" blood feeding associated behavioral complexities. A comprehensive RNA-Seq and extensive transcriptional profiling data, further strengthen the hypothesis that a quick recovery of the actions of odorant binding proteins immediately after blood feeding, and delayed re-activation of olfactory receptor proteins after blood meal digestion completion are unique to manage diverse behavioral responses. However, an extended blood meal follows up experimental data analysis further hypothesize that first blood meal exposure is enough for prime learning, satisfying the motivational search of mosquitoes for the completion of their gonotrophic cycles. Thus, it is plausible to propose that apart from the innate odor responses, adult female mosquitoes might take an advantage of prior odor (vertebrate) exposure, which leads an exclusive evolutionary specialty, allowing them to learn, experience and adapt as a fast blood feeder in nature (**Figure 8**).

In summary, we decoded and established a possible functional correlation that how coherent and smart actions of olfactory encoded factors enabled adult female mosquitoes to meet and manage the blood feeding associated complex behavioral activities (**Figure 8**). Furthermore, targeting species-specific unique genes such as sensory appendages proteins may be crucial to design disorientation strategy against mosquito An. culicifacies, an important malarial vector in rural India.

#### DATA DEPOSITION

The sequencing data were deposited to National Center for Biotechnology Information (NCBI) Sequence Reads Archive (SRA) system (BioProject accessions: PRJNA414162; BioSample accessions: SAMN07981002, SAMN07972755, and SAMN07775994).

#### AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: TD, RD; Performed the experiments: TD, TT, SV, DS, VS, PS, CC, SK, ST, JR; Analyzed the data: TD, RD; YH; Contributed reagents, materials, analysis tools: YH, RD, KP; Wrote the paper: TD, RD, YH, KP.

#### REFERENCES


#### FUNDING

Laboratory work was supported by Indian Council of Medical Research (ICMR), Government of India (No.3/1/3/ICRMR-VFS/HRD/2/2016) and Tata Education and Development Trust (Health-NIMR-2017-01-03/AP/db). TD is the recipient of UGC Research Fellowship (CSIR-UGC-JRF/20-06/2010/(i) EU-IV.

#### ACKNOWLEDGMENTS

We thank insectary staff members for mosquito rearing. We also thank Kunwarjeet Singh for technical assistance in laboratory. Finally, we are thankful to Xceleris Genomics, Ahmedabad, India for generating NGS sequencing data and NxGenBio Lifesciences, Delhi for bioinformatics support.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00577/full#supplementary-material


that modulate infection and blood-feeding behavior. PLoS Pathog. 8:e1002631. doi: 10.1371/journal.ppat.1002631


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Das De, Thomas, Verma, Singla, Chauhan, Srivastava, Sharma, Kumari, Tevatiya, Rani, Hasija, Pandey and Dixit. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Expressions of Olfactory Proteins in Locust Olfactory Organs and a Palp Odorant Receptor Involved in Plant Aldehydes Detection

The main chemosensory organs of locusts consisted of the antennae and the

#### Hongwei Li† , Peng Wang† , Liwei Zhang, Xiao Xu, Zewen Cao and Long Zhang\*

Department of Entomology, China Agricultural University, Beijing, China

#### Edited by:

Shuang-Lin Dong, Nanjing Agricultural University, China

#### Reviewed by:

Chen-Zhu Wang, Institute of Zoology (CAS), China Zhao-Qun Li, Tea Research Institute (CAAS), China Sofía Lavista Llanos, Max-Planck-Institut für Chemische Ökologie, Germany

\*Correspondence:

Long Zhang locust@cau.edu.cn †These authors have contributed

#### Specialty section:

equally to this work.

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 26 December 2017 Accepted: 14 May 2018 Published: 04 June 2018

#### Citation:

Li H, Wang P, Zhang L, Xu X, Cao Z and Zhang L (2018) Expressions of Olfactory Proteins in Locust Olfactory Organs and a Palp Odorant Receptor Involved in Plant Aldehydes Detection. Front. Physiol. 9:663. doi: 10.3389/fphys.2018.00663 mouthparts (maxillary and labial palps), which are suggested to perform different functions. However, very few are known about the differences of these two organs at molecular level. To understand the differences of locust antennae and palps in olfaction, the electrophysiological response and olfactory gene expression of these two organs were conducted. Our electrophysiological experiments with Locusta migratoria showed that the responses of mouthpart palps and antennae to odorants are quite different. Only a few odorants, such as (E,E)-2,4-hexadienal and (E,E)-2,4-heptadienal, elicited stronger electrophysiological responses of both maxillary and labial palps in comparison to the antennae. Additionally, we obtained 114 and 11 putative odorant receptor (OR) gene segments from the antennal and palp transcriptomes, respectively. Two novel odorantbinding proteins (OBPs; OBP15 and OBP16) and one novel OR (OR142) were identified for the first time. Out of the 16 OBP genes tested in RT-PCR and qPCR analyses, OBP8 was highly expressed in the nymphal palps. OBP4, OBP10, and OBP16 were only detected in the antennae. The other 11 OBP genes were jointly expressed in both antennae and palps. The relative expression level of OBP6 in male palps was much higher than that of female palps. Furthermore, for the 11 OR genes identified in palp transcriptome, the expression levels of OR12, OR13, OR14, and OR18 in the palps were significantly higher than those in the antennae. The OR12 in palps was demonstrated to be involved in detection of hexanal and E-2-hexenal, as well as (E,E)-2,4-heptadienal. Our results provide information on the different olfactory roles of locust antennae and palps at the molecular level.

Keywords: olfactory organs, electrophysiological response, odorant binding protein, odorant receptors, Locusta migratoria

#### INTRODUCTION

Mammals and insects have evolved sophisticated olfactory organs to receive a wide range of chemical stimuli. This distinguished ability enables them to detect and discriminate thousands of odor molecules. Many evidences propose that the different olfactory organs of a species play different roles (Smith, 2007; Su et al., 2009). For many insects, the antennae and mouthpart palps are important olfactory organs (de Bruyne et al., 1999, 2001). Both of them are covered with

**476**

a variety of chemosensory hairs that house the specialized olfactory sensory neurons (OSNs). The initial step in olfaction involves the binding of hydrophobic odorous molecules to odorant receptors (ORs) located at the ciliated dendrite endings of OSNs (Touhara and Vosshall, 2009). In this process, the high concentration of odorant-binding proteins (OBPs), which liaise the external environment and the ORs, is regarded as the important components in odor transmissions (Pelosi et al., 2006). Therefore, exploration the expressional patterns of olfactory genes in insects antennae and mouthpart palps is the basis for understanding the different roles of these olfactory organs.

Since the first insect OR and OBP were identified in Drosophila (Clyne et al., 1999; Gao and Chess, 1999; Vosshall et al., 1999) and in Antheraea polyphemus (Vogt and Riddiford, 1981), respectively, several of them have been identified in other insect species (Clyne et al., 1999; Vosshall et al., 1999; Xu et al., 2003; Pelosi et al., 2006; Zhou et al., 2010; Leal, 2013). In the genomes characterized to date, 79 OR genes have been found in mosquito (Hill et al., 2002), 162 in honey bee (Robertson and Wanner, 2006), and 341 in red floor beetle (Engsontia et al., 2008).

Locust (Locusta migratoria) is a model animal of hemimetabolous insects, and is also a notorious pest that damage worldwide agricultural productions (Hassanali et al., 2005). The feeding behavior of locust is probably mediated by chemoreception. Currently, 142 OR genes and 14 OBP genes have been identified in its genome and transcriptome<sup>1</sup> (Ban et al., 2003; Jin et al., 2005; Yu et al., 2009; Wang et al., 2015).

The relatively simple structure of the mouthpart palp represents an attractive model for investigating the neuromolecular networks which underlie chemosensation of an insect (Bohbot et al., 2014). However, the research on the olfactory genes expressed at the palps is limited (de Bruyne et al., 1999; Lu et al., 2007; Sparks et al., 2014; Dweck et al., 2016). For locust, only one study has announced that its antennae and mouthpart palps are responsible for different olfactory functions (Zhang et al., 2017). Here, we used L. migratoria as a model to investigate the electrophysiological responses of the palps and antennae as well as the different expression patterns of OBPs and ORs between these two olfactory organs. The aims of this study are to explore the different physiological functions and molecular bases in olfaction between locust antennae and palps.

#### MATERIALS AND METHODS

#### Ethics Statement

All of our experimental materials and methods are not contrary to ethics.

#### Insects and Tissues

Locusta migratoria individuals were obtained from the Department of Entomology, China Agricultural University. Detailed rearing procedures and tissue extraction were described in Xu et al. (2013).

#### Electrophysiological Studies

All electrophysiological experiments were conducted with a 10× universal AC/DC amplifier (Syntech, Netherlands), and the signals were recorded in an Intelligent Data Acquisition Controller (IDAC-4, Syntech, the Netherlands). EagPro software was used to record the absolute amplitudes after stimulation. The experimental chemicals were originally selected from leaf volatiles of maize, wheat, cotton, and soybean (Buttery and Ling, 1984; Buttery et al., 1985; Zeringue and McCormick, 1989; Njagi and Torto, 1996; Shibamoto et al., 2007; Pan et al., 2010; Michereff et al., 2011; Fu et al., 2014; Leppik and Frérot, 2014). Totally, 47 compounds (odorants) with the highest grade available (90–99.9%; Sigma-Aldrich, Shanghai, China) were used in the experiment (**Supplementary Table S1**).

Electrophysiological technique and protocols were followed by the practical introductions of Syntech (2009). For recording electroantennograms (EAGs), the antennae of fifth-instar nymphs were removed from the head and the distal tips of the antennae were immediately cut off. Each antenna was placed between the reference electrode (basal tip) and the recording electrode (distal tip) which connected by Spectra 360 electrode gel. The recordings were proceed since the signal input was stable. Diluted volatile compounds (each 10 µl) were applied to filter paper strips (length 2 cm, width 0.5 cm) which inserted into Pasteur tubes. Each Pasteur tube was only used for testing a specific compound. Paraffin oil was used as a blank control. The tube carried a constant airflow (150 ml/min), and its opening was positioned 1 cm from the antenna. The odor airflow was controlled by a stimulus air controller (CS-55, Syntech, Netherlands) and directed to the surface of the antenna. In this way, the stimuli were provided as 1 s at 20 ml/min generated by the stimulus air controller (CS-55, Syntech, Netherlands). There was an interval of 2 min between two stimulations to enable the recovery of antenna activity. The test was in the following order: paraffin oil (blank control), 20% (v/v) hexanal (positive control), 1% (v/v) chemical (test odorant), and paraffin oil (blank control). Each chemical compound was tested at least three times with different antennae. For electropalpograms (EPGs) recording, the abdomen of the locust was covered with a half-dissected centrifuge tube (0.5 ml), then fixed laterally on a glass slide using sticky tape. The dome of the fixed maxillary or labial palp (with dental wax) was directly oriented to the stimulus-supplying air tube. The reference electrode was inserted into the neck, and the recording electrode was inserted into the basal part of the dome by using an MN-151 micromanipulator (Narishige, Japan). The method was referred to electroantennograms on Drosophila (Ayer and Carlson, 1991) and flesh fly, Neobellieria bullata (Diptera: Sarcophagidae) (Wasserman and Itagaki, 2003). Each odorant was tested on at least four palps.

The mean value of the EAG, maxillary electropalpogram (EPG-M), or labial electropalpogram (EPG-L) was calculated with the following equation according to Shi et al. (2003):

$$\text{RV}\_{\text{EGG}}(\text{RV}\_{\text{EPG}}) \;= \frac{\text{Vs} - \text{Vb}}{\text{Vp} - \text{Vb}}$$

<sup>1</sup>https://www.ncbi.nlm.nih.gov/unigene?term=LOCUST+OBP&cmd= DetailsSearch

where RVEAG, RVEPG−M, or RVEPG−<sup>L</sup> represents the relative value of the response of the relevant receptor, Vs represents the recorded value of the response of the receptor to odorant, Vp represents the recorded value of response of the receptor to the positive control, and Vb represents the recorded value of response of the receptor to the blank control.

#### Transcriptome Sequencing

fphys-09-00663 May 31, 2018 Time: 17:23 # 3

To understand the molecular basis of olfaction in L. migratoria antennae and palps, transcriptome sequencing of each organ was performed as previously described by Zhang et al. (2015). In brief, the antennae or a mix of maxillary and labial palps from 30 fifth-instar locust nymphs (aged 3–5 days) were collected and their total RNA was extracted with TRIzol <sup>R</sup> Reagent (Life Technologies, United States) based on standard protocols. The RNA sample was purified, tested for purity and integrity, and finally introduced into the Illumina HiSeq <sup>R</sup> 2500 platform (Illumina, San Diego, CA, United States) for sequencing.

The method of de novo assembly was originally described by Zhang et al. (2015). In brief, de novo assembly of the short reads was performed using SOAPdenovo (Xie et al., 2014) at default parameters. The generated unigenes were analyzed by searching the non-redundant (NR). Unigene analyses were performed on a high-performance server, using a method similar to that originally described by Zhang et al. (2015). In brief, unigenes were annotated and aligned with protein databases from the National Center for Biotechnology Information (NCBI) and Swiss-Prot<sup>2</sup> . The targeted putative OR and OBP genes were then identified. A customized gene identification procedure was undertaken as follows: a local BLAST program, BioEdit (Vision 7.0.4.1) (Hall, 1999) Sequence Alignment Editor, was employed to search for more olfactory genes within the assembled and annotated unigenes library by querying for each of the annotated olfactory unigenes. Parameters were set as follows: minimum identity >95%, length >200 bp and E-value < 10−<sup>10</sup> . Finally, all repeatedly aligned olfactory unigenes were removed until only one remained. All single olfactory unigenes were subjected to BLAST alignment in the NCBI online server, and both ends of each unigene open reading frame structure were predicted.

Next, we screened the unigene sequences against protein databases Swiss-prot<sup>2</sup> , COG<sup>3</sup> , and KEGG<sup>4</sup> with blastx. We used "OR" and "OBP" as keywords to screen the annotated sequences. In order to promote identification of putative target genes, we used the known OBP and OR sequences of L. migratoria as "queries" to screen the transcriptome databases with tblastn. The putative OBP and OR genes were then confirmed using blastx. The TMHMM program (v. 2.0)<sup>5</sup> was used to predict the transmembrane domains of the OR genes.

## Tissue Expression Analysis of OBP and OR Genes

The assay included identification of gene expression in antennae, mouthparts, and guts of fifth-instar nymphs, and antennae and palps of adults of both gender. The tissue expressions of the candidate OBP and OR genes (accession numbers and gene names are listed in **Supplementary Tables S2**, **S3**) were analyzed with a method similar to that described by Zhang et al. (2015). In brief, total RNA was extracted from the above tissues with TRIzol (Invitrogen, CA, United States). Then, first-strand cDNA was synthesized using the cDNA FastQuant RT Kit (with gDNase) (Tiangen Biotech Co. Ltd., Beijing, China). The PCR product was sequenced to verify the specificity of primers used in RT-PCR. These gene-specific primers of OBP and OR were designed with Primer-BLAST (**Supplementary Tables S2**, **S3**). qPCR assays were performed in the StepOnePlusTM Real-time PCR System (Applied Biosystems, United States), with a KAPA SYBR <sup>R</sup> FAST qPCR Kit Master Mix (2X) (KAPA Biosystems, Boston, MA, United States). A qRT-PCR assay is similar to Zhang et al. (2015). In brief, the assay was carried out in a 20 µl reaction mixture in the ABI 7900 system (Applied Biosystems, Carlsbad, CA, United States). PCR was performed under the following program: 95◦C for 3 min, 40 cycles at 94◦C for 15 s, 60◦C for 20 s, and extension at 72◦C for 15 s. The melting curve was analyzed to assure specificity of the primers after each reaction and the 2−11CT method (Livak and Schmittgen, 2001) was used to calculate the expression level of each OBP and OR gene. Each sample type was replicated three times. The differences between relative expression levels of OBP or OR genes were analyzed with t-tests. The β-actin was used as a reference gene for internal standardization. PCR efficiency and specificity of primers to the target genes were validated in the experiment.

#### Phylogenetic Analysis of OBPs of L. migratoria and Other Insects

We constructed a phylogenetic tree using the 16 candidate OBPs of L. migratoria and selected OBPs of other insects, including Oedaleus asiaticus, Drosophila melanogaster, Bombyx mori; Tribolium castaneum; Adelphocoris lineolatus, Apis mellifera (the OBP amino acid sequences of all OBPs in this experiment are listed in **Supplementary Table S5**). We renamed LmigOBP13 (OBP4, GenBank: AEX33160.1,) and LmigOBP14 (OBP5, GenBank: AEX33161.1), on account LmigOBP4 and LmigOBP5 have been registered previously with the number AEV45802.1 and AFL03411.1 in NCBI GenBank by our lab. The phylogenetic tree was constructed by the neighborjoining method with Poisson-modified distance with MEGA6 software.

#### RNA Interference

Double stranded RNA (dsRNA) was synthesized based on manufacturer manual. In brief, PCR products were amplified with T7 promoter conjugated primer (primer pairs see **Supplementary Table S6**), and then purified with Wizard <sup>R</sup> SV Gel and PCR Clean-Up System (Promega, United States) as

<sup>2</sup>http://www.uniprot.org/

<sup>3</sup>http://www.ncbi.nlm.nih.gov/COG/

<sup>4</sup>http://www.genome.jp/kegg/

<sup>5</sup>http://www.cbs.dtu.dk/services/TMHMM-2.0/

templates for in vitro transcription. dsRNA was synthesized with T7 RiboMAXTM Express RNAi System (Promega, United States) and diluted into 1000 ng/µl with ddH2O and stored at −20◦C. Target dsRNA (5 µg) was delivered into each locust dorsal vessel through inter-segmental membrane (1st day of 5th instar nymph) by IM-9B microinjector (Narishige, Japan) equipped with glass capillary. dsGFP was microinjected as control group. The treated locusts were normally raised as wild individuals. RNA silencing was checked between 3th and 5th day postinjection. All RNAi-treated locusts used in EAG or EPG were checked by PCR after electrophysiological experiment to confirm the results of silencing. EAG or EPG methods are similar to those described above. The response value of the EAG or maxillary electropalpogram (EPG-M) was calculated with the following equation: RVEAG (RVEPG) = Vs − Vb. Where RVEAG or RVEPG represents the value of the response of the relevant receptor, Vs represents the recorded value of the response of the receptor to odorant, and Vb represents the recorded value of response of the receptor to the blank control. Each chemical compound was tested on at least seven different antennae or maxillary palps.

#### Statistical Analysis

Electroantennograms and EPG results were compared with one-way ANOVA with post hoc t-tests. All data was analyzed using GraphPad Prism 7 (Graphpad software, San Diego, CA, United States).

## RESULTS

#### Different Electrophysiological Responses of Locust Antennae and Palps to the Odors

Locust antennae, maxillary palps and labial palps showed responses to most of the 47 tested odorants at 1% v/v. However, the relative electrophysiological responses of the antennae were stronger than those of the palps to 43 out of 47 odorants. Of the 43 odorants, 19 of them elicited responses only in the antennae. Contrasted to the antennae could be elicited strong responses by a plenty of odorants, only two odorants, (E,E)-2,4-hexadienal and (E,E)-2,4-heptadienal, induced stronger electrophysiological

responses to both the maxillary and labial palps than to the antennae (**Figure 1**). Therefore, locust antennae and palps perceive odorants differently to some extent.

#### Different Expression of OBP in Locust Antennae and Palps

From our analysis of the transcriptomes of locust antennae and palps, two novel OBPs, named as LmigOBP15 and LmigOBP16 were identified. Together with the previously annotated 14 OBPs<sup>6</sup> (Ban et al., 2003; Jin et al., 2005; Yu et al., 2009; ), a total of 16 OBPs were obtained in transcriptomes. All of them were closest to the OBPs from another locust, O. asiaticus (Zhang et al., 2015), in the phylogenetic tree (**Supplementary Figure S1**). Among the 16 OBPs, the longest amino acid sequence was OBP16, with 271

<sup>6</sup>https://www.ncbi.nlm.nih.gov/unigene?term=LOCUST+OBP&cmd= DetailsSearch

TABLE 1 | Consensus alignment (%) of 16 odorant binding protein (OBP) amino acid sequences of L. migratoria.


Values in red boxes indicate the highest and the lowest identity of the OBPs.

fphys-09-00663 May 31, 2018 Time: 17:23 # 5

amino acids; while the shortest was OBP7, with only 133 amino acids. OBP3, OBP7, OBP11, and OBP13 were "Plus-C" OBPs (Zhou et al., 2004) (**Figure 2** and **Supplementary Table S5**). Sequence identities of the 16 OBPs ranged from 9.2 to 60.0% (**Table 1**).

RT-PCR analyses for OBPs showed that OBP4, OBP10, and OBP16 were only expressed in the antennae of nymphs and female and male adults. Expression level of OBP8 was higher in larval palps than that in adult palps and other tested organs in both adult and nymph. Additionally, OBP1, OBP2, OBP3, OBP5, OBP6, OBP11, OBP12, OBP13, and OBP14 were expressed in the antennae, palps, and mid gut (**Supplementary Figure S2** and **Supplementary Presentation S1**).

Our qPCR results revealed the relative expression levels of 15 OBP genes in the chemosensory organs, except for OBP8, which was too difficult to be detected in adult antennae and palps (**Figure 3** and **Supplementary Table S4**). The expression levels of 11 OBP genes, including OBP1, OBP4, OBP5, OBP7, OBP9, OBP10, OBP11, OBP12, OBP13, OBP14, and OBP16, in the antennae were significantly higher than those in the palps of the same sexual individuals. In contrast, OBP2, OBP3, OBP6, and OBP15 were markedly up-regulated in the palps than those in the antennae of both genders. Interestingly, the relative expression level of OBP3 in female palps was much higher than that in male palps yet in the antennae of both sexes.

Expression levels of 10 OBP genes in female antennae, including OBP1, OBP2, OBP3, OBP4, OBP5, OBP9, OBP10,

OBP12, OBP15, and OBP16, were significantly higher than those in male antennae. On the other hand, expression levels of OBP6, OBP7, and OBP14 in male antennae were significantly higher than those in female antennae. The expressions of OBP1 and OBP3 in female palps were higher than those in male palps, whereas OBP11 and OBP12 were expressed at similar levels in the palps of both sexes. OBP2, OBP5, OBP6, OBP9, and OBP15 in male palps were highly expressed than in female palps (**Figure 3** and **Supplementary Table S4**).

#### Different Expression of Odorant Receptors in the Antennae and Palps

We identified 114 putative OR gene segments (35 putative OR genes with more than 300 amino acids) from the transcriptome of locust antennae. However, only 11 putative OR gene segments were identified from the transcriptome of palps. Notably, OR142 from the antennal transcriptome was identified for the first time. It has 408 amino acid residues with 7 predicted transmembrane domains (**Figure 4A**). RT-PCR also showed that this gene was only expressed in the antennae (**Figure 4B**).

We checked the expressions of 11 putative OR genes identified from the palps using RT-PCR (**Figure 5**). Interestingly, only the OR12 was not detected in the antennae. OR16 was only detected in the antennae. OR13, OR15, OR18, and OR21 were widely expressed in the antennae, palps, and gut of nymphs and adults of both sexes. OR14, OR17, OR19, OR20, and OR22 were jointly expressed in the antennae and palps of nymphs and adults of both gender.

In the fifth-instar nymph, the relative expression levels of OR12, OR13, OR14, and OR18 genes in the palps were significantly higher than those in the antennae. In contrast, expression levels of OR15, OR16, OR17, OR19, OR21, and OR22 in the antennae were significantly higher than those in the palps (**Figure 6**). Expression levels of OR20 did not show significant differences between the antennae and the palps (**Supplementary Table S4**).

Actin, actin gene as positive control.

#### An Odorant Receptor Specifically Expressed in Palps Was Involved in Detection of Three Aldehydes

Our electrophysiological experiments showed that the palps responded remarkably stronger to (E,E)-2,4-heptadienal and (E,E)-2,4-hexadienal than antennae (**Figure 1**). Besides, we also found that another two odorants, hexanal and E-2 hexenal, elicited stronger absolute values in EPG than in EAG (**Figures 7A,B**). We speculated that there would be some specific ORs expressed in palps, which are responsible for the detection of these chemicals. Meanwhile, the RT-PCR analysis indicated that OR12 was highly expressed in palps. Thus we presumed that OR12 might be involved in detection to the aldehydes. We found that the responses of EPG of locust nymphs injected with dsRNA of OR12 to hexanal and E-2-hexenal were significantly reduced in comparison with locust injected with dsRNA of GFP (**Figures 7C,D**). Interestingly, the response of EPG of locust nymphs injected with dsRNA of OR12 to (E,E)-2,4-heptadienal was significantly lower than that of animals injected with dsRNA of GFP (**Figure 7E**). In turn, no changes of EPGs were detected to (E,E)-2,4-hexadienal between the two dsRNA experimental animals (**Figure 7F**). In contrast, there was no significant difference in EAG responses to hexanal and E-2-hexenal between the OR12 and GFP dsRNA injected locusts (**Figures 7G,H**). Moreover, the expression level of OR12 in palps was indeed depressed by injection of dsRNA of OR12 in comparison with individuals injected with dsRNA of GFP, or wild type (**Figure 7I**).

#### DISCUSSION

Locusts antennae has many olfactory sensilla of the basiconic, trichoid, and coeloconic type, while only few basiconic sensilla are present on the dome of each palp (Ochieng and Hansson, 1999; Jin et al., 2006). In the present study, the electrophysiological responses of antennae were stronger than those of palps to most tested odorants. We speculated that the abundant neurons and chemoreception proteins in the antennae, such as ORs and OBPs, induced this result. Since the varieties of odorants tested in this study were limited, we did not screen any odorant which only elicit response to palps. However, four odorants, (E,E)-2,4 hexadienal, (E,E)-2,4-heptadienal, hexanal and E-2-hexenal elicited much stronger responses to palps in comparison to the antennae. This implies that sensilla on the palps may house neurons with special olfactory receptors sensitive to these odorants.

It has been demonstrated that OBPs increase the sensitivity of odor discrimination for insects (Laughlin et al., 2008). The numbers of OBPs vary among insect species (Pelosi et al., 2006). In the present study, we identified two novel OBPs. As a result, there are a total of 16 OBPs found in L. migratoria<sup>7</sup> (Ban et al., 2003; Jin et al., 2005; Yu et al., 2009). Similarly, 15 and 14 OBPs were identified in the antennal transcriptomes of O. asiaticus (Zhang et al., 2015) and Schistocerca gregaria (Jiang et al., 2017), respectively. Orthopteran insects possess a significantly smaller number of OBPs compared to Dipteran insects, such as Drosophila and mosquitoes contain 51 and 79 OBPs, respectively (Galindo and Smith, 2001; Biessmann et al., 2002; Hekmat-Scafe et al., 2002; Xu et al., 2003; Zhou et al., 2004; Hansson and Stensmyr, 2011). This may reflect the specific evolutionary level of locust chemosensory system (Vogt, 2002; Pelosi et al., 2006; Xu et al., 2013).

PCR experiment demonstrated that a greater number of OBPs are expressed in locust antennae than in the palps. This may suggest that the olfactory functions of antennae are different from the palps. However, the relative expression levels of OBP6 are much higher in male palps than in female palps, indicating that it might be involved in detecting odors from the female. In addition, an extremely high level of OBP8 expressed in the palps of locust nymphs, suggesting that this protein may be involved in detecting specific odors that are important during nymphal stages. Moreover, the relative lower amounts of olfactory genes in palps may explain why the maxillary palps respond to a narrow range of odors.

Although more than 100 putative OR genes have been identified in the antennae of locust (this study; Wang et al., 2014, 2015), we only identified 11 OR genes in the locust palps. The different OR repertoires imply that the antennae are more versatile in olfaction than the palps. This is similar to the results in Anopheles gambiae, where there are more than 60 ORs found in the antennae, but only 13 were found in their palps (Latrou and Biessmann, 2008). Interestingly, our result showed that OR12 (named OR6 in Wang et al., 2015) was highly expressed in the palps than antennae of fifth-instar

<sup>7</sup>https://www.ncbi.nlm.nih.gov/unigene?term=LOCUST+OBP&cmd= DetailsSearch

standard errors of the mean for three independent experiments.

nymphs; but the expressional level of OR12 in nymphal palps was much lower than that in adult palps. A previous study showed a similar result for this gene in palps of fourth-instar nymphs (Wang et al., 2015). OR12 may have an important function in the palps at nymphal and adult stages. On the other hand, we found that the OR14 (named OR50 in Wang et al., 2015) was weakly expressed in the antennae and palps of both adults and nymphs. Additionally, it was proposed that OR13 (named OR133 in Wang et al., 2015) was only expressed in locust antennae, but it was detected in both antennae and palps in the present study. Similarly, OR17 (named OR5 in Wang et al., 2015) has previously been detected only in the adult antennae (Wang et al., 2015). However, in our study we detected OR17 in the antennae of both adults and fifth-instar nymph.

The qPCR data show that the expression of the OR12, OR13, OR14, and OR18 in the palps was significantly higher than in the antennae of fifth-instar nymphs. Contrarily, the expression of OR15, OR16, OR17, OR19, OR21, and OR22 was much higher in the antennae than in the palps. Similar results for OBPs expressed in antennae and palps further suggested that these two chemosensory organs might have different roles in chemoperception. In mosquitoes, the expression level of AsteOBP1 in antennae was ∼900-fold higher than that in maxillary palps (Sengul and Tu, 2010a,b). Therefore, the presence or absence of OBPs/ORs in the antennae and palps may reflect a natural selection of olfactory traits during the evolution of insect lineages (de Bruyne et al., 1999; Yasukawa et al., 2010).

Our results of RNAi demonstrated that OR12 in maxillary palps was responsible for detection of hexanal and E-2-hexenal, as well as (E,E)-2,4-heptadienal. This information partially provides a molecular basis for the antenna and palp in different olfactory functions. In Drosophila, although the antennae and palps respond to a similar spectrum of odorants, the palps display fewer high-sensitivity responses to specific odorants (Dweck et al., 2016), which also indicates the different roles of their antennae and palps in chemoperception. However, our experiments did not demonstrate that EPG of locust changed to (E,E)-2,4 hexadienal after depression of OR12. This odorant might be detected by other ORs, such as OR13, OR14, or OR18, which

dsRNA of GFP injected. The response of EPG or EAG was calculated from the response value of maxillary palp or antenna treated with chemicals minus the response value of maxillary palp or antenna treated with mineral oil as control. Error bar indicates SEM. <sup>∗</sup>p < 0. 05, one-way ANOVA with post hoc t-tests. (B) Comparison of response level to E-2-hexenal in different organs and different genotypes by EPG or EAG. Abbreviations are referred to (A). (C) Comparison of response level to hexanal in different genotypes by EPG. Abbreviations are referred to (A). ds-OR12, dsRNA of OR12 injected. (D) Comparison of response level to E-2-hexenal in different genotypes by EPG. Abbreviations are referred to (A). ds-OR12, dsRNA of OR12 injected. (E) Comparison of response level to (E,E)-2,4-heptadienal in different genotypes by EPG. (F) Comparison of response level to (E,E)-2,4-hexadienal in different genotypes by EPG. (G,H) Comparison of response level to hexanal or E-2-hexenal in different genotypes by EAG. (I) RNA silencing is checked after electrophysiological experiment with semi-quantitative RT-PCR. Actin was used to check template quality.

were demonstrated to be highly expressed in palps (**Figure 6**). The novel expression of olfactory receptors in the maxillary palps could generate a subpopulation of insects using new food source. In turn, the utilization of new resource, combined with a segregation event, may lead to the emergence of a new species.

In sum, our results show that (E,E)-2,4-hexadienal, (E,E)- 2,4-heptadienal, hexanal and E-2-hexenal elicits much stronger responses in palps than in the antennae. We found that OBP8, OR12, OR13, OR14, and OR18 were much higher expressed in the nymphal palps, suggesting that those proteins may be involved in detecting specific odors during feeding process. On the other hand, OR12 shows specific expression in palps and we showed that it was involved in the detection of three aldehydes produced by the host plant (Buttery and Ling, 1984; Buttery et al., 1985). Consequently, the palps could play an important role in speciation through food selection. The palps, therefore, would be a fruitful area for investigating the specific roles in insect chemoperception in the future.

## AUTHOR CONTRIBUTIONS

LZ designed the experiments and wrote the manuscript. HL and PW did the electrophysiological experiment. LZ and HL analyzed the transcriptomes and identified the OBP and OR genes. HL, XX, and ZC conducted the PCR experiments. HL, PW, and LZ analyzed the data. All authors contributed to the revisions.

#### FUNDING

This work was supported by the Chinese Universities Scientific Fund (2015NX001) and the National Natural Science Foundation of China (31472037).

#### REFERENCES


#### ACKNOWLEDGMENTS

We thank Dr. Ke Wei, Yinwei You, Xuewei Yin, and three reviewers for their helpful comments and suggestions for the manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00663/full#supplementary-material

FIGURE S1 | Phylogenetic analysis of OBPs of L. migratoria and other insects.

FIGURE S2 | Tissue expression analysis of OBP genes.

TABLE S1 | Details of odorants used in EAG and EPG tests.

TABLE S2 | Details of OBPs in Locusta migratoria and the primers used for qPCR.

TABLE S3 | Details of ORs in L. migratoria and the primers used for qPCR.

TABLE S4 | The OBP and OR genes result of qPCR.

TABLE S5 | List of OBPs of L. migratoria and 6 other orders of insects used for the phylogenetic tree.

TABLE S6 | Sequence of primers used in RNAi experiments. T7 promoter is shown in red.

mediate short- and long-range attraction. eLife 5:e14925. doi: 10.7554/eLife. 14925


fphys-09-00663 May 31, 2018 Time: 17:23 # 11

proteins in locusts. Cell. Mol. Life Sci. 62, 1156–1166. doi: 10.1007/s00018-005- 5014-6


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Li, Wang, Zhang, Xu, Cao and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fphys-09-00663 May 31, 2018 Time: 17:23 # 12

# Comparison of Olfactory Genes in Two Ectropis Species: Emphasis on Candidates Involved in the Detection of Type-II Sex Pheromones

Zhao-Qun Li\*, Xiao-Ming Cai, Zong-Xiu Luo, Lei Bian, Zhao-Jun Xin, Bo Chu, Yan Liu and Zong-Mao Chen\*

#### Edited by:

Nicolas Durand, Sorbonne Université, France

#### Reviewed by:

Hao Guo, Chinese Academy of Sciences, China Bing Wang, Institute of Plant Protection (CAAS), China Nicolas Montagné, Sorbonne Universités, France

#### \*Correspondence:

Zhao-Qun Li zqli@tricaas.com Zong-Mao Chen zmchen2006@163.com

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 26 May 2018 Accepted: 25 October 2018 Published: 14 November 2018

#### Citation:

Li Z-Q, Cai X-M, Luo Z-X, Bian L, Xin Z-J, Chu B, Liu Y and Chen Z-M (2018) Comparison of Olfactory Genes in Two Ectropis Species: Emphasis on Candidates Involved in the Detection of Type-II Sex Pheromones. Front. Physiol. 9:1602. doi: 10.3389/fphys.2018.01602 Key Laboratory of Tea Biology and Resource Utilization, Ministry of Agriculture, Tea Research Institute, Chinese Academy of Agricultural Sciences, Hangzhou, China

The sibling species Ectropis grisescens and E. obliqua are the major chewing tea pests in China. A difference in sex pheromone components plays a central role in premating isolation in these two species. To investigate the mechanism of premating isolation in these two Ectropis species, we sequenced the transcriptomes of the antennae of female and male E. obliqua individuals and performed phylogenetic analyses, abundance analyses, and tissue expression profile analyses to compare the olfactory genes involved in the detection of sex pheromones. A total of 36 odorant-binding proteins (OBPs) and 52 olfactory receptors (ORs) were identified in E. obliqua. Phylogenetic analyses showed that EoblOBP2, 3, and 25 were grouped in the pheromone-binding protein clade with EgriOBP2, 3, 25, and another lepidopteran PBP. EoblOR25 and 28 were grouped with EgriOR25, 28, and pheromone receptors for the detection of Type-I sex pheromone components. EoblOR24, 31, 37, and 44 were grouped with EgriOR24, 31, 37, and 44. All of these 4 EoblORs and 4 EgriORs showed higher abundance in male antennae than in female ones. Therefore, OBP2, 3, 25 and OR24, 31, 37, 44 of E. grisescens and E. obliqua might be responsible for sex pheromone component detection. However, the sequences of these genes in E. grisescens and E. obliqua were more than 90% identical. This indicates that these orthologous genes might play similar roles in the detection of sex pheromones. In contrast, the observed OBPs and ORs differed in abundance between the antennae of the two Ectropis species. Therefore, we speculate that these two Ectropis species use the different transcript levels of PRs to differentiate sex pheromone components. The results of the present study might contribute in deciphering the mechanism for premating isolation in these species and may be of use in devising strategies for their management.

Keywords: transcriptomic analysis, olfaction gene, sex pheromone perception, Ectropis grisescens, Ectropis obliqua

**488**

## INTRODUCTION

fphys-09-01602 November 13, 2018 Time: 16:14 # 2

The tea geometrid, Ectropis obliqua, is a notorious chewing pest in the tea plantations of China (Ma et al., 2016b), and use of the E. obliqua nucleopolyhedrosis virus (EoNPV) preparation is an effective management strategy for its control. The susceptibility of tea geometrids collected from Quzhou, Zhejiang province to EoNPV is reported to be about 724.5-fold higher than in those collected from Yixing, Jiangsu province (Xi et al., 2011). Based on the assessment of external morphology, molecular biology, and interspecific hybridization, these two geographical populations were shown to be two different Ectropis species, namely E. grisescens and E. obliqua (Lepidoptera: Geometridae) (Jiang N. et al., 2014; Zhang G.H. et al., 2016). Because of morphological similarities, these two Ectropis species were mischaracterized as a single tea geometrid species, E. obliqua.

In moths, courtship and mating behaviors are regulated by sex pheromones, which play crucial roles in reproduction and are associated with reproductive isolation (Wyatt, 2009). The sex pheromones in both E. grisescens and E. obliqua were reported to be (Z,Z,Z)-3,6,9-octadecatriene (Z3,Z6,Z9- 18:H) and (Z,Z)-3,9-cis-6,7-epoxy-octadecadiene (Z3,epo6,Z9- 18:H) and were present at similar ratios (Ma et al., 2016c; Yang et al., 2016), which is unusual for sibling species occurring in the same region. The female sex pheromones of the two Ectropis species were reexamined in order to clarify how these two geometrids maintain premating isolation (Luo et al., 2017). The results showed that Z3,Z6,Z9-18:H and Z3,epo6,Z9- 18:H were the sex pheromones of E. grisescens, whereas Z3,Z6,Z9-18:H, Z3,epo6,Z9-18:H, and (Z,Z)-3,9-cis-6,7-epoxynonadecadiene (Z3,epo6,Z9-19:H) were the sex pheromones of E. obliqua. Thus, the presence or absence of Z3,epo6,Z9-19:H may be the major determinant for premating isolation of these two Ectropis species. Moth sex pheromone components are classified into three groups: Type-I, Type-II, and miscellaneous type (Ando et al., 2004). Type-I sex pheromone components are composed of unsaturated compounds with C10–C<sup>18</sup> straight chain unsaturated alcohols, aldehydes, or acetate esters. Type-II sex pheromone comprise unsaturated hydrocarbons and epoxy derivatives with a C17–C<sup>23</sup> straight chain (Millar, 2000; Ando et al., 2004). Both E. grisescens and E. obliqua thus produce Type-II sex pheromone components.

Previous studies have shown that both soluble binding proteins and membrane-bound receptors are used in the detection of sex pheromones in moth (Leal, 2012).The odorantbinding proteins (OBPs) and water-soluble carriers are thought to aid in the capture and transport of odorants and pheromones to their receptors (Pelosi et al., 2014), and the pheromone-binding proteins (PBPs), a sub-class of OBPs, are thought to enhance the solubility of lipophilic Type-I sex pheromone components and deliver them to the membrane-bound receptors (Zhou, 2010; Sun et al., 2013; Jin et al., 2014). Sex pheromone receptors (PRs), a subfamily of odorant receptors (ORs), are specifically activated by Type-I sex pheromone components and have been widely studied in lepidopteran insects (Jiang X.J. et al., 2014; Zhang et al., 2014; Chang et al., 2015). In addition to PBPs and PRs, GOBP2 is also thought to be involved in the detection of Type-I sex pheromones (Liu et al., 2015). The analysis of the molecular mechanisms for the olfactory detection of sex pheromones in E. grisescens and E. obliqua might contribute to decipher the strategy for premating isolation in these two Ectropis species.

Less is known about the perception mechanism of Type-II pheromone components (Zhang D.D. et al., 2016). In our previous studies, we sequenced the antennae transcriptomes of E. grisescens, and identified 40 OBPs and 59 ORs, including an OR attuned to E. grisescens sex pheromone (Li et al., 2017). Although 24 OBPs and 4 ORs were identified from the leg transcriptome of E. obliqua (Ma et al., 2016a), the gene numbers were far different from E. grisescens. We therefore sequenced the transcriptomes of the antennae, the principal olfactory organs, of female and male E. obliqua individuals, and performed analyses of phylogeny, abundance, and tissue expression profile to compare the olfactory genes involved in sex pheromone detection in the two species.

#### MATERIALS AND METHODS

#### Insect Rearing and Tissue Collection

Individuals of E. grisescens and E. obliqua were originally collected from the Experimental Tea Plantation of the Tea Research Institute, Chinese Academy of Agricultural Sciences (Hangzhou, Zhejiang, China). The larvae of the two species were accurately identified by comparing the cytochrome c oxidase I gene sequences and were reared on fresh tea shoots in different climate-controlled rooms under the same conditions (25 ± 1 ◦C and 70 ± 5% relative humidity with a 14-h light:10-h dark photoperiod), enclosed in nylon mesh cages (70 cm × 70 cm × 70 cm). After pupation, the male and female pupae were separated based on their morphological characters and kept separately in cages for eclosion. After emergence, the adult moths were fed on a 10% honey solution. For transcriptome sequencing, antennae from 100 female and 100 male 2-day-old virgin E. obliqua individuals were collected in duplicates. For qRT-PCR analyses, a different sample of twenty 2-day-old virgin female and male E. grisescens and E. obliqua adults were used to collect antennae, heads without antennae, thoraxes, abdomen without the pheromone gland, legs, wings, proboscises, and pheromone glands. These tissues were immediately frozen and stored at −80◦C until RNA isolation. Total RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, United States). The quality of the RNA samples was assessed by agarose gel electrophoresis, NanoDrop (Thermo, Wilmington, DE, United States), and Agilent 2100 Bioanalyzer.

#### cDNA Library Preparation, Illumina Sequencing, and de novo Assembly

The cDNA library construction and Illumina sequencing of the samples were performed at Novogene Bioinformatics Technology Co., Ltd. (Beijing, China). Poly adenylated mRNAs were isolated from 5 µg total RNA using oligo (dT) magnetic beads and were fragmented into short fragments in the presence of divalent cations in fragmentation buffer at 94◦C for 5 min. Using these short fragments as templates, random hexamer primers were used to synthesize first-strand cDNAs.

Second-strand cDNAs were generated using RNase H and DNA polymerase I. After end repair and adaptor ligation, the short sequences were amplified by PCR, purified with a QIAquick <sup>R</sup> PCR Purification Kit (Qiagen, Venlo, Netherlands), and sequenced on the HiSeqTM 2500 platform (San Diego, CA, United States). The de novo transcriptome assembly was carried out using the short-read assembly program, Trinity (r20140413p1) (Grabherr et al., 2011) based on the pairedend reads with default settings. The transcriptomic data were deposited in the NCBI/SRA database (SRR7757597 and SRR7757596). Transcripts longer than 150 bp were first aligned by BLASTX against protein databases (NR, Swiss-Prot, KEGG, and COG; E-value < 10−<sup>5</sup> ) to retrieve the proteins with the highest sequence similarity to the unigenes along with their functional annotations. We then used Blast2GO (Conesa et al., 2005) for gene ontology (GO) annotation of the transcripts and WEGO software (Ye et al., 2006) to plot the results of the GO annotation.

## Expression of Transcripts and Differential Expression Analysis

Transcript abundance was calculated as reads per kilobase per million mapped reads (RPKM) method, which can eliminate the influence of different transcript lengths and sequencing discrepancies when calculating the abundance (Mortazavi et al., 2008). We used the following equation:

$$RPKM\left(A\right) = \frac{C \times 10^6}{\frac{N \times L}{10^3}}$$

where RPKM (A) represents the RPKM value of the transcript A, C is the number of reads uniquely aligned to the transcript A, N is the total number of fragments uniquely aligned to all the transcripts, and L is the number of bases in the transcript A.

Genes showing differential expression between the two conditions/groups were detected using the DESeq R package (1.10.1) (Anders and Huber, 2010), which provides statistical routines to determine differential expression from digital gene expression data using a model based on negative binomial distribution. The resulting P-values were adjusted using Benjamini and Hochberg's approach to control the false discovery rate. Genes with an adjusted P-value < 0.05 found using DESeq were considered to be differentially expressed.

## Identification of E. obliqua OBP and OR Genes and Sequence Analyses

Sequenced transcriptomes were annotated by combining the transcriptomes of the antennae from females and males, then searching against the non-redundant (NR) database using BLASTX with a cut-off e-value of 10−<sup>5</sup> . EoblOBPs and EoblORs were named according to sequence similarity with EgriOBPs and EgriORs.

## Sequence Alignment and Phylogenetic Analysis

The amino acid sequence alignments of EgriOBPs, EoblOBPs, EgriORs, and EoblORs were performed using ClustalX 2.0 (Larkin et al., 2007). To investigate the phylogenetic relationships of the OBPs and ORs between E. grisescens, E. obliqua and other insect species, we aligned them using MAFFT (E-INS-I parameter) (Katoh and Standley, 2013). The phylogenetic trees were constructed using PhyML 3.1 with LG substitution model to generate a maximum likelihood phylogenetic tree (Guindon et al., 2010). Finally, the trees were viewed and group edited with FigTree v1.4.2<sup>1</sup> .

#### Quantitative Real-Time PCR Validation

The tissue expression patterns of EgriOBPs, EgriORs, EoblOBPs, and EoblORs in different tissues were measured using a qPCR method performed according to the minimum information for publication of quantitative real-time PCR experiments (Bustin et al., 2009). Total RNA was isolated using the SV Total Isolation System (Promega, Madison, WI, United States) according to manufacturer's instructions. The quality and quantity of the RNA samples was assessed using agarose gel electrophoresis and NanoDrop (Thermo). Single-stranded cDNA templates were synthesized using the Reverse Transcription System (Promega) following manufacturer's instructions. The qRT-PCRs were performed on a Bio-Rad CFX96 touch real-time PCR detection system (Bio-Rad, Hercules, CA, United States). The primers were designed using Beacon Designer 7.7 based on the E. grisescens and E. obliqua nucleotide sequences obtained from the transcriptome data (**Supplementary Table S1**). The reaction was performed as follows: 30 s at 95◦C, followed by 40 cycles at 95◦C for 5 s and 60◦C for 34 s. Templates were diluted in a five-fold series of samples and were used to determine the amplification of primers. Each reaction was run in triplicate (technical repeats). Theguanine nucleotide-binding protein G(q) subunit alpha and glyceraldehyde-3-phosphate dehydrogenase genes of both E. grisescens and E. obliqua were selected as reference genes for the qPCR analysis. A blank control without template cDNA (replacing cDNA with H2O) served as the negative control. Each reaction included three independent biological replicates and was repeated three times (technical replicates). The relative transcript levels were calculated using the comparative 2−11Cq method (Livak and Schmittgen, 2001).

## RESULTS

## Identification of E. obliqua OBP and OR Genes

A total of 36 EoblOBPs were identified in the E. obliqua antennae transcriptome. Sequence analyses showed that 31 of the 36 EoblOBPs were full-length genes (**Figure 1**). Of the 36 EoblOBPs, EoblOBP8, 13, 14, 15, 32, C-15995, and C-6102 contained only four conserved cysteine residues. EgriOBP4, 5, and 7 contained more than six conserved cysteine residues. The other EoblOBPs contained six conserved cysteine residues. A total of 52 EoblORs were identified in the E. obliqua antennae, 37 of which contained a full-length open reading frame, and had a full-length of about 1200 bp. EgriOBP2, 3, 25 and EgriOR24, 28, 31, 27 44

<sup>1</sup>http://tree.bio.ed.ac.uk/software/figtree/

were identified as candidate genes involved in detecting sex pheromones. The identities between the sequences of EoblOBP2 and EgriOBP2, EoblOBP3 and EgriOBP3, and EoblOBP25 and EgriOBP25 were 98.77, 98.84, and 97.65%, respectively, whereas those between the sequences of EoblOR24 and EgriOR24, EoblOR25 and EgriOR25, EoblOR28 and EgriOR28, EoblOR31 and EgriOR31, EoblOR37 and EgriOR37, and EoblOR44 and EgriOR44, were 96.04, 96.08, 90.68, 92.72, 99.29, and 90.68%.

#### Phylogenetic Analyses of E. obliqua and E. grisescens OBP and OR Genes

In the phylogenetic tree based on the OBP sequences, four EoblOBPs (EoblOBP2, 3, 25, and C-39094) were grouped in the PBP clade with EgriOBP2, 3, 25, and other lepidopteran PBP (**Figure 2**). The orthologs EoblOBP1 and 29 were in the GOBP group with their orthologs EgriOBP1 and 29 (**Figure 2**). The phylogenetic tree generated using the OR sequences showed that EoblORco was well clustered with EgriORco, ObruORco(ORco of Operophtera brumata), HvirORco(ORco of Heliothis virescens), and BmorOR2(ORco of Bombyx mori) with high bootstrap support (**Figure 3**). EoblOR25 and 28 were grouped with EgriOR25, 28, ObruOR1 (a PR for Type-II sex pheromone), and PRs for Type-I sex pheromone, including B. mori, Helicoverpa armigera, H. assulta, and H. virescens (**Figure 3**). EoblOR24, 31, 37, and 44 were grouped with EgriOR24, 31, 37, and 44 (**Figure 3**).

## Abundance of E. obliqua and E. grisescens OBP/OR mRNAs

To compare the abundance of E. obliqua and E. grisescens OBP/OR in the antennae, we characterized their abundance by evaluating their RPKM values and combined these results with those of the phylogenetic analyses (**Figures 4–6**). We observed that most of the orthologs of E. obliqua and E. grisescens OBP/OR had similar expression levels in the antennae (**Figures 4**, **5**). However, several orthologous genes that were more abundant in the antennae of male individuals showed different abundance in the antennae of the two Ectropis sibling species (**Figure 6**). E. obliqua and E. grisescens OBP2, 3, 9, 12, and 25 showed higher transcription levels in the antennae of the male individuals than in those of the females (**Figure 6**). Among these 10 male antennabiased OBPs, EgriOBP2 and 3 were the two most abundant

OBPs in the E. grisescens antennae, followed by EgriOBP12. However, EoblOBP12 showed the highest transcript level in the antennae, which was more than four-fold higher than the EoblOBP2 and 3 levels. Comparing the orthologous OBP2, 3, 9, 12, and 25 of E. obliqua and E. grisescens, the RPKM values of EoblOBP3, 12 and 25 were respectively 1.8-, 17.3-, and 7.9 fold higher than their orthologs, EgriOBP3, 12 and 25,. The transcript levels of OBP2 and 9 were similar in E. obliqua and E. grisescens.

The analyses of the abundance of EoblOR and EgriOR mRNAs showed that ORco and OR37 showed the highest expression levels in the antennae of E. obliqua and E. grisescens (**Figure 6**). The E. obliqua and E. grisescens OR24, 28, 31, 37, and 44 were predominantly expressed in the antennae of the male individuals. However, the orthologs of these ORs showed different transcript levels in the two Ectropis sibling species. The RPKM values of EoblOR24 and 28 were 4.2- and 8.0-fold higher than those of their orthologous genes, EgriOR24 and 28, respectively. On the other hand, OR31, 37, and 44 showed higher expression levels in E. grisescens with RPKM values in the antennae of male E. grisescens being 3.2-, 2.6-, and 17.0-fold of the levels of their orthologs in the antennae of male E. obliqua.

#### Tissue Expression Profile of E. obliqua and E. grisescens OBP/OR Genes

Expression patterns of E. obliqua and E. grisescens OBP/OR genes in different adult tissues detected by qPCR showed that the orthologous genes of EoblOBPs and EgriOBPs had similar expression patterns (**Figure 7**). Most of the EoblOBPs and EgriOBPs were highly expressed in the antennae of both the female and male individuals. Among those, E. obliqua and E. grisescens OBP2, 3, 9, 12, and 25 were expressed at higher levels in the antennae of males than females. E. obliqua and E. grisescens OBP7, 21, and 33 were expressed in both female and male proboscises at relatively high levels. EoblOBP20 and EgriOBP20 were predominantly expressed in the legs of both sexes, and EoblOBP24, EgriOBP24, and EgriOBP35 were mainly expressed in the wings of both sexes. EoblOBP15 and 27, and EgriOBP15, 27, and 40 were highly expressed in male abdomens.

**492**

The expression patterns of the other OBP genes were ubiquitous in most tested tissues, at relatively high levels.

The analyses of the expression profile of ORs showed that the orthologs of EoblORs and EgriORs also had similar profiles, as observed for EoblOBPs and EgriOBPs (**Figure 8**). Most of the EoblORs and EgriORs were mainly expressed in the antennae. Among the ORs showing an expression bias for the antenna, E. obliqua and E. grisescens OR24, 28, 31, 37, and 44 were more highly expressed in the antennae of males than in those of females. EoblOR7, 32, 41, and EgriOR7, 32, 41, 57, 58 were expressed in female and male heads at relatively high levels. EgriOR54 was predominantly expressed in female wings, and EgriOR55 was highly expressed in male abdomens.

## DISCUSSION

Sex pheromone-induced behavior plays crucial roles in insect reproduction. The difference in the sex pheromone components of Z3,epo6,Z9-19:H may be the major determinant for premating isolation between these two Ectropis sibling species (Ma et al., 2016c; Luo et al., 2017). This difference in the sex pheromone components indicates that these two Ectropis sibling species might differ in the detection of sex pheromones, leading to premating isolation. We identified the candidate genes for detection of E. grisescens sex pheromones and analyzed the transcriptomes of the antennae of female and male individuals of E. obliqua to identify olfactory genes potentially involved in the perception of sex pheromones, for comparison with E. grisescens.

Insect PBPs are responsible for detecting the sex pheromone components in lepidopterans (Leal, 2012). In our study, we identified 36 EoblOBPs in E. obliqua. Generally, olfactory genes involved in detecting sex pheromones were expressed at higher levels in the male antennae than in female antennae. The abundance in antennae and tissue expression profiles showed that E. obliqua and E. grisescens OBP2, 3, 9, 12, and 25 were predominantly expressed in the male antennae, with relatively high RPKM values. Among these, EoblOBP2, 3, 25

and EgriOBP2, 3, 25 were grouped in the PBP clade with another lepidopteran PBP. Therefore, OBP2, 3, and 25 from E. obliqua and E. grisescens might encode the PBPs for Type-II pheromone components. However, the abundance of OBP3 and 25 in antennae differed between E. obliqua and E. grisescens. The RPKM values of OBP3 and 25 in E. obliqua were higher than

fphys-09-01602 November 13, 2018 Time: 16:14 # 7

in E. grisescens. On the other hand, the amino acid sequences of OBP2, 3, and 25 in E. obliqua and E. grisescens were more than 97% identical, indicating that these three orthologous genes might have similar functions in binding and transporting sex pheromone components. Therefore, the difference in the transcript levels might be used to detect the difference in sex pheromone components of these two Ectropis sibling species. Unlike OBP2, 3, and 25, OBP12 was not grouped in the PBP clade, but it was the most abundant EoblOBP in the antennal transcriptome of male E. obliqua individuals. The RPKM value of EoblOBP12 was 17.3-fold higher than that of EgriOBP12, indicating that OBP12 might relate to sex pheromone perception in E. obliqua and E. grisescens. The GOBP2, another sub-class of OBPs, is reported to strongly bind sex pheromones in S. litura (Liu et al., 2015), B. mori (Zhou et al., 2009), and S. exigua (Liu et al., 2014). EoblOBP1 and EgriOBP1 were grouped in the GOBP2 sub-class. Consequently, we speculate that these two OBPs might be involved in the binding of sex pheromone components.

Insect PRs, a key sub-class of ORs, are specifically dedicated to the detection of sex pheromone components in the Lepidoptera (Jiang X.J. et al., 2014; Zhang et al., 2014; Chang et al., 2015). In our previous study, we found that EgriOR25 and 28 were grouped in the PR clade with PRs for Type-I sex pheromone

of female individuals; MH, head without antennae of male individuals; FT, thorax of female individuals; MT, thorax of male individuals; FAb, abdomen without pheromone gland of female individuals; MAb, abdomen of male individuals; FL, legs of female individuals; ML, legs of male individuals; FW, wings of female individuals; MW, wings of male individuals; FPr, proboscis of female individuals; MPr, proboscis of male individuals; Pg, pheromone gland.

components and four male antenna-biased EgriORs (EgriOR24, 31, 37, and 44) formed an independent group in the phylogenetic analysis (Li et al., 2017). Of these four male antenna-biased EgriORs, EgriOR31 responded robustly to Z3,epo6,Z9-18:H but weakly to Z3,Z6,Z9-18:H (Li et al., 2017). Because PRs are a conserved sub-class of OR, EgriOR24, 25, 28, 31, 37, and 44 might be potential PRs of E. grisescens. In present study, EoblOR25 and 28 were grouped with EgriOR25, 28, ObruOR1, and PRs of B. mori, H. armigera, H. assulta, and H. virescens, and EoblOR24, 31, 37, and 44 were grouped with EgriOR24, 31, 37, and 44. The results of tissue expression profiles indicate that EoblOR24, 25, 28, 31, 37, and 44 had similar expression patterns with EgriOR24, 25, 28, 31, 37, and 44, with these more highly expressed in the male antennae than in other tissues. Therefore, the number of potential PRs that we identified in E. obliqua was the same as in E. grisescens. Moreover, the sequence identity matches of OR24, 25, 28, 31, 37, and 44 between E. obliqua and E. grisescens were greater than 90%. Among these, the identity between EoblOR37 and EgriOR37 was as high as 99.29%. The high degree of similarity in sequence identities and tissue expression patterns indicate that these orthologous genes might play similar roles in the detection of sex pheromones. The equal number and high identity of PRs between these two Ectropis sibling species implies that the differences in the detection of sex pheromones might occur at the transcription state.

We determined the RPKM values of OR24, 28, 31, 37, and 44 to characterize their abundance in the antennae of E. obliqua and E. grisescens. These values showed that the abundance of these ORs in the antennae differed between E. obliqua and E. grisescens. EoblOR24 and 28 showed much higher transcript levels than EgriOR24 and 28, while, the transcript levels of EoblOR31, 37, and 44 were much lower than those of EgriOR31, 37, and 44. The differences in the transcript levels of OR24, 28, 31, 37, and 44 between E. obliqua and E. grisescens might relate to the recognition of different sex pheromones. Because E. obliqua and E. grisescens (Jiang N. et al., 2014; Zhang G.H. et al., 2016) underwent sympatric speciation, the males of these two Ectropis sibling species must recognize females of their own species by correctly differentiating the sex pheromones of

fphys-09-01602 November 13, 2018 Time: 16:14 # 10

E. obliqua and E. grisescens. This means that the E. grisescens male can detect Z3,epo6,Z9-19:H, the sex pheromone of E. obliqua. Therefore, it is understandable that these two Ectropis sibling species have an equal number of PRs with high identities, and regulation of the transcript levels of PRs might be selected by the two species to differentiate the difference in their sex pheromone components. Further research about the functions of E. obliqua and E. grisescens OBP1, 2, 3, 9, 12, and 25 and OR24, 25, 28, 31, 37, and 44 in the detection of sex pheromones is required to understand how these two Ectropis sibling species recognize the differences of their sex pheromone components.

#### AUTHOR CONTRIBUTIONS

fphys-09-01602 November 13, 2018 Time: 16:14 # 12

Z-QL and Z-MC conceived and designed the experiments. Z-QL performed the experiments and wrote the manuscript. Z-QL, X-MC, Z-XL, LB, Z-JX, BC, and YL analyzed the data. All authors reviewed the final manuscript.

#### FUNDING

This study was funded by the National Key Research & Development (R&D) Plan (2016YFD0200900), the National

#### REFERENCES


Natural Science Foundation of China (31701795), the Modern Agricultural Industry Technology System (CARS-23), and the Central Public-Interest Scientific Institution Basal Research Fund (1610212018011).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01602/full#supplementary-material


the winter moth Operophtera brumata. Sci. Rep. 6:18576. doi: 10.1038/ srep18576


that a general odorant-binding protein discriminates between sex pheromone components. J. Mol. Biol. 389, 529–545. doi: 10.1016/j.jmb.2009.04.015

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Li, Cai, Luo, Bian, Xin, Chu, Liu and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Membrane Proteins Mediating Reception and Transduction in Chemosensory Neurons in Mosquitoes

Jackson T. Sparks<sup>1</sup> \*, Gina Botsko<sup>1</sup> , Daniel R. Swale<sup>2</sup> , Linda M. Boland<sup>3</sup> , Shriraj S. Patel<sup>3</sup> and Joseph C. Dickens<sup>3</sup> \*

<sup>1</sup> Biology Department, High Point University, High Point, NC, United States, <sup>2</sup> Department of Entomology, Louisiana State University AgCenter, Baton Rouge, LA, United States, <sup>3</sup> Department of Biology, University of Richmond, Richmond, VA, United States

#### Edited by:

Nicolas Durand, Université Pierre et Marie Curie, France

#### Reviewed by:

Rajnikant Dixit, National Institute of Malaria Research, India Clement Vinauger, Virginia Tech, United States

#### \*Correspondence:

Jackson T. Sparks jsparks@highpoint.edu Joseph C. Dickens joseph.dickens@richmond.edu

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 02 July 2018 Accepted: 30 August 2018 Published: 20 September 2018

#### Citation:

Sparks JT, Botsko G, Swale DR, Boland LM, Patel SS and Dickens JC (2018) Membrane Proteins Mediating Reception and Transduction in Chemosensory Neurons in Mosquitoes. Front. Physiol. 9:1309. doi: 10.3389/fphys.2018.01309 Mosquitoes use chemical cues to modulate important behaviors such as feeding, mating, and egg laying. The primary chemosensory organs comprising the paired antennae, maxillary palps and labial palps are adorned with porous sensilla that house primary sensory neurons. Dendrites of these neurons provide an interface between the chemical environment and higher order neuronal processing. Diverse proteins located on outer membranes interact with chemicals, ions, and soluble proteins outside the cell and within the lumen of sensilla. Here, we review the repertoire of chemosensory receptors and other membrane proteins involved in transduction and discuss the outlook for their functional characterization. We also provide a brief overview of select ion channels, their role in mammalian taste, and potential involvement in mosquito taste. These chemosensory proteins represent targets for the disruption of harmful biting behavior and disease transmission by mosquito vectors.

Keywords: olfaction, taste, gustation, mosquito, insect, vertebrate, membrane proteins, ion channels

#### INTRODUCTION

Mosquitoes are able to sense and track hosts over long distances, sometimes up to 70 m away (Chaisson and Hallem, 2012), a feat of chemosensitivity that increases the likelihood of encountering a host and transmitting disease. Mosquito chemosensation in adults includes two modalities – olfaction (smell) and gustation (taste), each crucial for host seeking, foraging, mating, and oviposition (Clements, 1992). Proper discrimination of chemical cues ensures a nutritive diet, suitable mates, and safe passage of genetic material to subsequent generations. Host-seeking behavior over a distance is informed by the olfactory system, while contact discrimination relies on the gustatory system. Overlap between olfactory and gustatory modalities is evident in some instances at both the anatomical and molecular levels (Melo et al., 2004; Kwon et al., 2006; Riabinina et al., 2016).

To locate human hosts, mosquitoes sense carbon dioxide (CO2), along with human skin and sweat odorants such as ammonia, lactic acid and other carboxylic acids (Chaisson and Hallem, 2012). Upon landing, skin emanations and blood are evaluated before full feeding behavior is initiated and completed. Disease agents are transmitted when saliva containing these agents passes

**501**

into the circulatory system or epithelial tissue of the host via specialized mouthparts of the mosquito (Clements, 1992).

Transmission of malaria by mosquitoes led to an estimated 445,000 deaths world wide in 2016 (World Health Organization [WHO], 2017; Report 2017). Mosquito borne dengue virus is responsible for at least 22,000 deaths per year (Weaver and Reisen, 2010) and represents a worsening threat according to the World Health Organization. Cases of malaria and other mosquito borne diseases number in the hundreds of millions each year, representing one of the largest healthcare burdens in the world. Pathogen transfer between mosquitoes and humans is facilitated by highly efficient chemosensory neurons in the mosquitoes that guide them to their animal hosts. A full understanding of the chemosensory receptors and other membrane proteins that transduce the chemical signals responsible for guiding behavior is important to the development of strategies to disrupt hostseeking and biting by mosquito vectors.

Here we detail the repertoire of peripheral ligand binding membrane proteins, ancillary membrane proteins, and signal transduction proteins, and discuss the outlook for their functional characterization. These chemosensory proteins are the primary molecular detectors of ecologically relevant chemicals and as such represent targets for disruption of mosquito behavior for prevention of dangerous contacts by mosquito vectors with their hosts. We restrict our review to the adult stage, but the gene families highlighted also express in mosquito larvae and may be involved with behavior of aquatic life stages (Bohbot et al., 2007; Xia et al., 2008). We focus on information gathered from mosquito species; however, the broader context of these genes requires consideration of functional data from other insect families and model organisms.

#### CHEMOSENSORY ANATOMY

Chemosensory organs of mosquitoes include external paired antennae, maxillary palps, labial palps, internal surfaces of mouthparts, distal leg segments and wing margins that are adorned with hair-like or dome shaped structures called sensilla (Slifer, 1962). The morphology of individual sensilla varies by species and cuticular location. Sensilla have one or multiple pores that allow external molecules to traverse an aqueous lumen that is innervated by the dendrites of one or more olfactory receptor neurons (ORNs) or gustatory receptor neurons (GRNs) (**Figure 1**; Clements, 1992). These dendrites contain membranebound chemosensory receptor proteins that respond with sensitivity and selectivity to chemicals that pass into the lumen of the sensillum (Hallem et al., 2004). The interactions between molecules and receptor proteins initiate signal transduction leading to an action potential. This conversion of chemical information to electrical signals allows mosquitoes to detect individual components of complex blends providing the basis for higher neural processing in the antennal lobes, mushroom bodies and elsewhere in the brain. These chemical signals may be used to locate and identify food sources, oviposition substrates, conspecifics, and potential threats (Brown et al., 1951; Davis, 1984).

Water-soluble accessory proteins, including odorant binding proteins (OBPs), of the lumen originate from support cells near the cell body of ORNs and GRNs (**Figure 1**; Vogt and Riddiford, 1981; Vogt et al., 1999). These proteins have various functions in insects including transport of odorants and tastants to dendritic interfaces and general maintenance of the biochemical content of the sensillum (Vogt and Riddiford, 1981; Leal, 2013; Jeong et al., 2013). The binding profiles and exact roles of individual OBPs associated with ORN/GRN activity of mosquitoes remains mostly unexplored. Fan et al. (2011), Brito et al. (2016), and Pelosi et al. (2018) provide comprehensive reviews of insect OBPs. Mosquitoes present two or three support cells per sensillum; these cells sheath and maintain the proper function of sensory neurons (McIver, 1982). Axons of ORNs and GRNs project as nerve bundles to organized neuropil in the brain. In general, ORN termini are more distinctly organized than termini of GRNs in insects. In the malaria mosquito Anopheles gambiae, ORN termini group by subtype in 60–70 visible glomeruli in the antennal lobe and four to six less defined glomeruli in the subesophageal ganglion (Riabinina et al., 2016). GRNs of the yellow fever mosquito Aedes aegypti terminate into seven irregular zones in the subesophageal ganglion and tritocerebrum (Ignell and Hansson, 2005); these divisions may represent different classes of molecules stimulating each grouping of GRNs, e.g., sugars or human sweat components, as observed in the vinegar fly Drosophila melanogaster (Isono and Morita, 2010).

#### PRIMARY RECEPTOR FAMILIES

The three main chemosensory receptor families expressing in mosquito appendages containing ORNs/GRNs are the odorant receptors (ORs), ionotropic receptors (IRs), and gustatory receptors (GRs) (Pitts et al., 2004, 2011; Bohbot et al., 2007; Sparks et al., 2014; Lombardo et al., 2017). The expression of these gene families has been demonstrated in more than ten mosquito species belonging to the three most important disease spreading genera: Aedes, Anopheles, and Culex. GRs represent the most ancient insect chemosensory receptor protein family (Eyun et al., 2017), dating back to the most recent common ancestor of hexapods and placozoans, multicellular animals with the simplest known cellular structure. GR genes are present in diverse aquatic animals from anemones to copepods (Robertson, 2015; Eyun et al., 2017), perhaps mediating reception of watersoluble molecules. GRs and the more recently evolved, hexapodspecific ORs are related protein families (Eyun et al., 2017), each family with characteristic seven-transmembrane structure and atypical membrane topologies (**Figure 2**; Robertson et al., 2003). OR genes are present in wingless hexapod ancestors of insects belonging to Archaeognatha and Zygentoma but are absent in more ancient hexapod lineages (Brand et al., 2018). OR gene families expanded around the time of the first winged insects, perhaps as an adaptation to navigating larger areas with more diverse and informative odorants (Missbach et al., 2014). After adapting to life on land, but before the evolution of flight, IRs and the first ORs likely mediated reception of volatile odorants in early insect ancestors. As they predate ORs, IRs are present

secreted by support cells selectively shuttle molecules to dendritic processes of odorant receptor neurons (ORNs). Odorant receptors (OR), ionotropic receptors (IR), and carbon dioxide-sensitive gustatory receptors (GR) on the membrane of ORNs selectively bind molecules initiating signal transduction leading to ORN activation. In general, ORs, IRs, and GRs do not co-express in the same ORN, but are shown together here to illustrate protein richness of ORN/lymph interface. Molecules with low or zero vapor pressures (black triangles), acids and amines contact aqueous sensillar lymph via a terminal cuticular pore in gustatory sensilla of the labella, tarsi, and wing margins. Soluble proteins secreted by support cells selectively shuttle molecules to dendritic processes of gustatory receptor neurons (GRN). GRs, IRs, and some ORs on the membrane of GRNs selectively bind molecules initiating signal transduction and GRN activation. ORN and GRN axons terminate in the antennal lobes and subesophageal ganglion (SOG). Local interneurons mediate primary processing of chemosensory information, and signals project via second order projection neurons to higher brain regions associated with the mushroom bodies and lateral horn (LH) (Siju et al., 2008) where sensory information integrates subsequently informing important behaviors and shaping memory (Zars, 2000).

in multiple phyla of protostomes including molluscs, nematodes, and arthropods (Eyun et al., 2017).

The number of receptor genes in mosquitoes varies depending on species and gene family (**Table 1**), likely reflecting the unique requirements of each species' ecological niche. Insect ORs are sensitive to compounds like esters, alcohols, and ketones, while IRs respond to various amines and acids (Suh et al., 2014). Comparative studies of receptor function are limited, but evidence suggests that relatively high sequence homology between a few mosquito ORs indicates conservation of an ancient and indispensable olfactory sensitivity to indole (Bohbot et al., 2010) and octenol, important cues for oviposition and host orientation (Dekel et al., 2016). The mechanism by which new receptor genes evolve may be primarily through

a birth-and-death model wherein duplications lead to subtle fitness cost-free shifts in receptor shape/function (Robertson et al., 2003; Hill et al., 2015). There is also evidence that ORs lacking close sequence homology between two distantly related mosquito species each respond to the same human skin odor sulcatone (Bohbot et al., 2007; Carey et al., 2010), suggesting that shared host preferences of distant relatives may be the product of independent evolution of similarly sensitive ORs. The most highly conserved receptor genes across mosquito lineages retaining functions and/or expression profiles are the CO2− (Erdelyan et al., 2012; McMeniman et al., 2014) and

sugar-sensitive GRs (Freeman et al., 2014), OR co-receptor (Jones et al., 2005), and the presumptive IR co-receptors (Rytz et al., 2013). As the majority of mosquito chemosensory receptors are highly divergent, functional characterization will require speciesspecific gene disruption or heterologous expression studies.

Expression levels of individual receptors offer some insights into function. Changes in transcript levels of receptor genes in A. gambiae are correlated with moderate chemosensory neuron sensitivity shifts following a blood meal (Rinker et al., 2013a). In addition to expression shifts due to feeding state changes, there may be natural fluctuations in chemosensory



Estimations are from genomic analyses except where indicated. Variations in gene number likely reflect unique selective pressures on each species over time, but the possibility of stochastic processes affecting gene number should be considered. <sup>a</sup>Kent et al., 2008; <sup>b</sup>Bohbot et al., 2007; <sup>c</sup>Hill et al., 2002; <sup>d</sup>Arensburger et al., 2010; <sup>e</sup>Lombardo et al., 2017; <sup>f</sup>Chen et al., 2017; <sup>g</sup>Chen et al., 2015; <sup>h</sup>Croset et al., 2010; ∗ transcriptomic data only.

protein abundance based on time of day (Rund et al., 2013; De Das et al., 2018). Mosquito feeding often peaks at dawn or dusk (Clements, 1992); thus, there may exist a relationship between functional demands for chemosensory proteins and temporal regulation of gene expression in peripheral neurons. A. aegypti display concurrent increases in ORN sensitivity to CO<sup>2</sup> and octenol, and expression levels of corresponding OR and GR transcripts throughout their first 10 days of adulthood (Bohbot et al., 2013). Differential vectorial capacity between two closely related anopheline species may be defined in part by differential expression of olfactory receptor genes (Rinker et al., 2013b), and host preference differences between two A. aegypti subspecies are directly linked to expression differences of a single OR (McBride et al., 2014). Moreover, viral infection alters expression levels of ORs and GRs in antennae of A. aegypti (Sim et al., 2012), raising the possibility that infectious agents may have evolved the ability to promote host-seeking behavior in infected vectors by targeting transcriptional activation factors for chemosensory genes in mosquito cells.

## Gustatory Receptors (GRs)

The architecture of the insect gustatory system has been widely studied from the molecular to the organismal level. GRs are primarily expressed in proboscis, legs (Hill et al., 2002; Sparks et al., 2013; Matthews et al., 2016) and maxillary palps (Erdelyan et al., 2012; Bohbot et al., 2014). Though only a single GR gene knockout/knockdown study has been published for mosquitoes, GRs likely mediate gustatory reception in GRNs based on: (1) the requirement of GRs for normal responses to a variety of tastants in D. melanogaster (reviewed in Isono and Morita, 2010) and (2) their enriched expression in mosquito tissues containing the greatest number of GRNs (Sparks et al., 2013; Matthews et al., 2016; Lombardo et al., 2017). GRNs respond to salt, feeding stimulants (e.g., sugar), water, host blood components and feeding deterrents (e.g., quinine and DEET) (Pappas and Larsen, 1978; Kessler et al., 2013; Sanford et al., 2013; Sparks and Dickens, 2016). Functional studies of mosquito GRs are unavailable, with the exception of RNAiand ZFN-based confirmations that two to three atypical GRs expressing in ORNs are required for the detection of CO<sup>2</sup> in A. aegypti (Erdelyan et al., 2012; McMeniman et al., 2014). Direct investigation of specific insect GRs using heterologous systems has been reported for a single sugar sensitive receptor (Sato et al., 2011). Other attempts to express functional non-sugar-sensitive GR assemblages have been unsuccessful, thus the generation of GR mutant strains via CRISPR-mediated alterations will likely be the next step toward GR deorphanization. Several mosquito GRs show clear homology with D. melanogaster GRs of known function (Kent et al., 2008), namely those involved in the reception of sugars or antifeedants like quinine. Whether or not mosquito GRs play a role in the reception of host cues with low vapor pressures remains an intriguing possibility.

## Odorant Receptors (ORs)

Odorant receptors are expressed in the main olfactory appendages (Qiu et al., 2004): antennae, maxillary palps, and proboscis (Fox et al., 2001; Kwon et al., 2006; Lu et al., 2007). The best characterized chemosensory gene family in mosquitoes, ORs are required for normal host discrimination (DeGennaro et al., 2013) and the reception of important host cues (McBride et al., 2014). Components of human sweat (Bernier et al., 2000) activating A. gambiae ORNs include L-lactic-acid, l−octen−3−ol and 4−methylphenol (Cork and Park, 1996). Other host odorants known to stimulate mosquito ORNs include ammonia, indole, geranyl acetone, 3-methyl-1 butanol, 6-methyl-5-hepten-2-one, 1-dodecanol, hexanedioic acid (Meijerink et al., 2001; Bohbot et al., 2010; Pelletier et al., 2010), and skatole (Hughes et al., 2010). ORs are amenable to heterologous expression and subsequent chemical screening. The odorant tuning range of individual ORs varies greatly from narrow to broad (Carey et al., 2010; Wang et al., 2010). Some ORs are only activated by compounds within a single chemical class, e.g., A. gambiae OR2 is tuned to a small set of aromatics containing a benzene ring, while others respond to chemicals from multiple classes from terpenes to heterocyclic compounds (Carey et al., 2010; Wang et al., 2010).

DeGennaro et al. (2013) examined the relative contributions of ORs, GRs, and IRs to host-seeking behavior in A. aegypti by genomic deletion of the gene coding the OR-coreceptor (ORco, necessary for all OR-mediated ORN activation). In the absence of CO2, ORco mutants did not respond to human-scented materials as is the case for wildtype controls (DeGennaro et al., 2013), indicating the ORco-independent IRs are likely not involved in the detection of host skin emanations. However, in combination with CO2, which activates a unique set of GRs (Erdelyan et al., 2012), human skin odorants do indeed elicit behavioral responses from ORco mutants suggesting the existence of redundant ORindependent pathways for detecting blends of host breath and skin emanations.

McBride et al. (2014) compared antennal transcriptomes of human-preferring, domestic forms of A. aegypti with guinea pig-preferring forest forms thereby identifying the enriched transcript OR4 among 13 other genes as significantly upregulated in domestic forms and human-preferring hybrids. Not only does increased OR4 expression appear to drive human host preference in wild populations, but specific non-synonymous variants also show strong correlation to preference and demonstrate linear

functional variance. OR4 is sensitive to sulcatone, a chemical found in uniquely high levels in human emanations as compared to other animals (McBride et al., 2014). Interestingly, levels of sulcatone exceeding those naturally emanating from human skin may elicit avoidance responses from A. aegypti (Logan et al., 2008). McBride et al. (2014) note that other odors besides sulcatone likely contribute to human preference by domestic forms of A. aegypti and other up- or down-regulated genes identified in their survey likely contribute to host preference.

ORs (Liu et al., 2010; DeGennaro et al., 2013; Xu et al., 2014) and GRs (Lee et al., 2010; Sanford et al., 2013) are involved with the reception of repellents like DEET. Xu et al. (2014) identified in the southern house mosquito Culex quinquefasciatus OR136 that, in combination with ORco, mediates responses to synthetic and natural repellents in Xenopus oocytes. Knockdown of C. quinquefasciatus OR136 transcripts reduced ORN responses to DEET (Xu et al., 2014). Repellents like DEET may also alter feeding and host seeking behaviors via interactions with many receptors at once, modulate ORco function directly or function primarily in coordination with other behaviorally relevant compounds like those emanating from host skin or breath (Dickens and Bohbot, 2013; DeGennaro, 2015).

#### Ionotropic Receptors (IRs)

Ionotropic receptor expression in olfactory and gustatory organs in D. melanogaster is well characterized, and these receptors are tuned to carboxylic acids, aldehydes, and amines (Benton et al., 2009; Abuin et al., 2011). Acids and amines are important hostseeking signals for mosquitoes (Van der Goes van Naters and Carlson, 2006). Ligands of IR-expressing ORNs were originally identified through extracellular recordings of electrical activity of sensory neurons housed within target sensilla (Yao et al., 2005). IR-expressing neurons housed within grooved-peg sensilla of the antenna (Pitts et al., 2004) are much less sensitive and slower to respond than OR-expressing neurons in insects. In further contrast, IR and OR-expressing neurons detect different classes of odorants; the strongest IR ligands only weakly activate, if at all, ORs, and the strongest OR ligands (ester, alcohols, and ketones) do not stimulate IR-expressing neurons (De Bruyne et al., 2001).

Tuning profiles of individual IRs in mosquitoes is limited to two studies and a handful of individual genes (Liu et al., 2010; Pitts et al., 2017). A study implementing knock-down of IR76b in A. gambiae larvae demonstrated its function in mediating behavioral responses to butylamine (Liu et al., 2010). Pitts et al. (2017) is the only study to date examining individual IR gene function in adult mosquitoes and is consistent with foundational work in Drosophila (Benton et al., 2009; Croset et al., 2010; Abuin et al., 2011). Different combinations of A. gambiae IRs were expressed heterologously in Xenopus oocytes and more than 400 chemicals were used to screen for IR-dependent currents (Pitts et al., 2017). Three IR "complexes" were discovered: IR41a/IR25a/IR76b (most sensitive to nitrogenous compounds 2 methyl-2-thiazoline and pyrrolidine), IR41c/IR25a/IR76b (most sensitive to pyrrolidine and 3-pyrroline), and IR75k/IR8a (most sensitive to carboxylic acids of eight or nine carbons). Many other IR genes are expressed in mosquito chemosensory tissues (**Table 1**); thus, IR deorphanization represents a crucial step toward exploring all potential receptors as targets for altering harmful host-seeking and feeding behaviors.

#### SIGNAL TRANSDUCTION

Ion channels serve as the molecular basis for membrane excitability by allowing inward or outward flow of ions across a cell membrane to enable signal transduction and the alteration of other cellular processes. Ligand-gated ion channels represent the primary ion channel type in the insect chemosensory system functioning as a "receptor." ORs, GRs, and IRs act through synaptic signaling on electrically excitable cells by converting chemical signals (i.e., tastants or odorants) to an electrical signal. Upon binding of the signal molecule(s) several actions may allow flow of cations and/or anions that stimulate neuronal transmission, downstream signaling, and other physiological processes: the ion channel protein itself may open due to a conformational shift, associated ion channels may be activated in conjunction with ligand-binding receptor activation or intracellular modulators of channel activity may initiate transmembrane ion flow indirectly.

Stimulus-specific ORs (ssORs) in insects are trafficked to dendritic membranes by co-receptor ORco (Larsson et al., 2004), a membrane protein highly conserved in sequence and function across diverse insect lineages (Jones et al., 2005) and required for fast ssOR activation (Larsson et al., 2004). ORco and ssORs form heteromeric complexes acting as ligand-gated ion channels with evidence pointing toward a pore region shared between subunits (**Figure 2**; Sato et al., 2008; Wicher et al., 2008; Nichols et al., 2011; Nakagawa et al., 2012). Evidence based on Dipteran (D. melanogaster and A. gambiae) and Lepidopteran (Bombyx mori) OR complexes expressed in Xenopus oocytes and cultured human cells suggests G-protein-coupled pathways are dispensable for OR activation in insects (Sato et al., 2008). However, other experiments probing D. melanogaster OR complexes showed that G-protein modulating compounds or genetic disruption of G-proteins significantly affected OR activation dynamics in cultured human cells (Wicher et al., 2008; Deng et al., 2011) and in vivo (Deng et al., 2011). A more recent study found no evidence for ionotropic mechanisms in lepidopteran OR complexes sensitive to pheromone (Nolte et al., 2016). Thus, OR-mediated signal transduction in mosquitoes may involve ionotropic and/or metabotropic pathways. For indepth reviews of olfactory transduction mechanisms in insects, see Fleischer et al. (2018) and Wicher (2015).

Co-receptor IRs (IRcos), such as D. melanogaster and A. gambiae IR25a and IR8a, form heteromeric relationships with stimulus-specific IRs (ssIRs) (Abuin et al., 2011; Pitts et al., 2017). IRcos are more conserved between insect species than ssIRs and possess an amino-terminal domain (ATD) that is usually absent in ssIRs (**Figure 2**). Evidence suggests IRs assemble as heterotetramers comprising two ssIR and two IRco subunits (Abuin et al., 2011; Pitts et al., 2017). Additional IR assemblies may exist as some combination of three IRs, including a second type of receptor (reviewed in Rytz et al., 2013). It is unknown whether these functional relationships apply to all mosquito IRs.

Gustatory receptor-mediated signal transduction remains poorly understood. Heterologous expression of gustatory GRs has not been as successful as similar experiments using ORs and IRs, perhaps meaning that either many GRs are required simultaneously to produce single cell responses or that other unknown factors present in GRNs are required for ligand-gated activation. Orthologs D. melanogaster GR43b and B. mori GR9 act as fructose-sensitive non-selective ionotropic channels when expressed in Xenopus oocytes or cultured human cells (Sato et al., 2011). This activity was independent of G-protein-coupled pathways, though several other reports provide evidence that G-protein-coupled pathways are involved in GR-mediated signal transduction (Ishimoto et al., 2005; Ueno et al., 2006; Ueno and Kidokoro, 2008).

#### ANCILLARY MEMBRANE PROTEINS

As reviewed above, the majority of recent research concerning the molecular biology of mosquito chemoreception has focused on the function of three receptor families (ORs, GRs, and IRs), their ligands, phylogenetic analyses, and modulation of these receptors to obtain a desirable phenotype (i.e., avoidance). Recently, studies have defined the properties of GRNs and have elegantly shown functionally different classes of GRNs expressing unique combinations of receptor genes (Freeman and Dahanukar, 2015). Specifically, ion channels are expressed within various GRNs where they are required for relevant taste modalities, such as salt taste and bitter detection (Liu et al., 2003; Al-Anzi et al., 2006), as well as neural propagation of the signal.

#### Transient Receptor Potential (TRP) Channels

Transient Receptor Potential (TRP) channels belong to the group of non-voltage gated, cation-permeable ion channels (Nilius, 2003) and are highly conserved proteins that are present in all species from yeast to mammals. Mammalian TRP channels are composed of six-transmembrane domains with a pore region between TM5 and TM6 (**Figure 2**) and can be divided into six subfamilies based on their sequence homology: TRPC (canonical), TRPV (vanilloid), TRPM (melastatin), TRPP (polycystin), TRPML (mucolipin), and TRPA (ankyrin) (Vannier et al., 1998). TRP channels respond to a wide range of stimuli and have an astonishing diversity of cation selectivity, which enables them to function as a conserved unit for integration of varied sensory information.

TRPA channels are a conserved subfamily of cation channels that are expressed in vertebrates and invertebrates, and appear to perform similar physiological functions. In vertebrates, the calcium permeable cation channel, TRPA1, is expressed in nociceptive neurons and functions to detect noxious or pungent chemicals, such as environmental irritants (Bautista et al., 2005). The majority of information regarding the physiological importance of insect TRP channels is focused on their role in the mechanisms of thermosensation and mechanosensation (reviewed in Fowler and Montell, 2013), yet over the past decade, TRPA channels have been of significant interest to insect physiologists for their role in gustation (Du et al., 2015; Freeman and Dahanukar, 2015) and repellency (Bautista et al., 2005; Kim et al., 2010; Kwon et al., 2010).

TRPA channels in D. melanogaster are encoded by painless, the fly homolog to mammalian TRPA1/ANKTM1 ion channel protein. Like GRs, painless is expressed in GRNs of the labellum, pharynx, legs and wings, and are specifically involved in the rejection of allyl and benzyl isothiocyanate, the pungent taste and insecticidal component of wasabi (Al-Anzi et al., 2006). TRPA1 is expressed in a subset of aversive GRNs and is required for avoiding aristolochic acid in food-choice assays (Kim et al., 2010), but avoidance of other bitter or aversive compounds, such as caffeine or quinine, were independent of TRPA1 function. This lack of broad activity to all bitter molecules suggests TRPA1 likely functions in tandem with additional transport proteins or receptors that may be differentially expressed. Indeed, a subset of labellar and leg GRNs coexpress the caffeine receptors (Gr66a, Gr32a, and Gr47a) and painless. The functional dependency of these two genes is not fully understood, but we speculate that the co-expression of these two receptors, and potentially others, enables a multimodal response neuron that can detect and integrate taste modalities that result in different behavior (Van Giesen et al., 2016).

Expression of A. gambiae TRPA1 (AgTRPA1) in Xenopus oocytes indicated that the channel transduces temperature sensation, and channel expression is on the distal antennal sensory structures (Wang et al., 2009). These structural and functional roles seem to be conserved in A. aegypti and the common house mosquito Culex pipiens. Together, these findings support the notion that AgTRPA1 functions as a peripheral thermoreceptor in mosquito antenna. More recent work has uncovered an additional chemosensory role for AgTRPA1 (Survery et al., 2016). Patch clamp recordings on heterologously expressed and purified, full-length AgTRPA1 and truncated 11–776 AgTRPA1 (lacking the N-terminal ARD) demonstrated that both proteins are functional, as each responded to the electrophilic compounds, allyl isothiocyanate and cinnamaldehyde, as well as heat. Their similar intrinsic fluorescence properties and related quenching of tryptophan, when activated by allyl isothiocyanate or heat, led the researchers to conclude that conformational change in the lipid bilayer occurs independently and outside of the N-terminal domain (Survery et al., 2016). As such, AgTRPA1 is both a thermo- and chemoreceptor, and while the N-terminal domain's function is unknown, it is hypothesized to play a role in tuning the channel's response (Survery et al., 2016).

Kang et al. (2010) examined the response of D. melanogaster to reactive electrophiles, including allyl isothiocyanate (AITC), N-methyl maleimide (NMM), and cinnamaldehyde (CA), and found that addition of these chemicals to food dramatically inhibited the natural proboscis extension response (PER); the inhibitory effect was considered gustatory, not olfactory, because avoidance of these non-volatiles required ingestion. This study found the responses to reactive electrophiles depend on the cation channel TRPA1 as TRPA1 mutants showed no reduction in PER when offered food containing AITC, NMM,

or CA. Promoter-knockdown experiments established peripheral sensory neurons as the site of action for dTRPA1 in gustation. TEVC recordings on A. gambiae demonstrate that reactive electrophiles activate mammalian TRPA1s; mutations in TRPA1 decreased electrophile sensitivity.

Citronellal, a plant-based acyclic monoterpene with a distinctive lemony scent is used in lotions, candles, and sprays to repel mosquitoes and other pests such as ticks and fleas. In contrast to D. melanogaster for which citronellal activated a GPCR coupled to TRPA1 channels, A. gambiae TRPA1 was directly activated by citronellal (Kwon et al., 2010). These results invite further study to confirm the potential of repellents like citronellal that activate gustatory signaling in mosquitoes and secondarily deter feeding. A. gambiae isoform TRPA1(B) did not respond robustly to citronellal (Du et al., 2015). These results encourage a comparative structure-function approach using TRPA1(A) and TRPA1(B) to probe the structural basis for citronellal actions on the cation channel activity.

TRPA1 channels are not present in hymenoptera (although they do have other TRPA channels; Matsuura et al., 2009). This suggests that TRPA1 could be targeted for mosquito control without negatively impacting important pollinators like honeybees. Further, the gustatory system of D. melanogaster employs TRPA1 for detection and subsequent avoidance of bacterial endotoxins lipopolysaccharides (LPS) (Soldano et al., 2016). Together, these data support the notion that TRPA channels are a highly conserved ion channel and have similar physiological roles in the sensory systems of mammals and invertebrates, regardless of the sensory modality involved (Rosenzweig et al., 2005).

In addition to TRPA channels, other TRP channels play key roles in gustatory avoidance. For instance, TRPL, a member of the TRPC family (Venkatachalam and Montell, 2007), is activated in vitro by the bitter tastant camphor and is expressed in the dendrites of D. melanogaster GRNs (Zhang et al., 2013). Wildtype adult and larval D. melanogaster avoid camphor, whereas TRPL302 mutants displayed a deficit in camphor avoidance while showing normal avoidance in response to other aversive tastants (Zhang et al., 2013). Interestingly, TRPL expression was reduced during prolonged exposure to camphor, and a corresponding reduction in avoidance behavior was observed (Zhang et al., 2013). These data suggest that changes in taste preference are dependent upon the concentration of receptors, and modifications of synaptic connections or receptor concentration may result in plasticity of taste interpretation; thus, this pathway may represent a novel target for antifeedants for arthropod control.

#### Epithelial Sodium Channels (ENaC)

Epithelial Sodium Channels (ENaC), a member of the degenerin (DEG)/ENaC superfamily of ion channels encoded by the pickpocket (ppk) gene family, functionally assemble as a heterotrimeric or homotrimeric proteins (Benson et al., 2002). ENaC channels have evolved different physiological functions throughout the Kingdom Animalia, but are conserved as ionotropic receptors that respond to extracellular stimuli to pass sodium ions.

Although insects and mammals have independently evolved distinct molecular pathways for gustation, there are clear parallels in the molecular organization that allows for comparison between the two systems (Yarmolinsky et al., 2009). In mammals, ENaC is involved in transepithelial sodium transport in many tissues (e.g., kidney, lung) and is critical in many epithelial tissues that require sodium transport, including taste epithelial cells (Lindemann et al., 1998; Kretz et al., 1999). Genetic knockout of ENaC in rat taste cells resulted in loss of salt attraction and sodium taste response, which validated previous pharmacological studies suggesting ENaC as the principal pathway for mediating sodium taste in mammals (Lindemann, 1997; Chandrashekar et al., 2010). Interestingly, significant overlap in behavior exists between insects and mammals when exposed to varying concentrations of salts. Considering the conservation of ENaC function and similar behavioral tendencies, it was hypothesized that DEG/ENaC proteins are responsible for salt detection in D. melanogaster. Indeed, two genes encoding ENaC, termed Pickpocket11 (ppk11) and Pickpocket19 (ppk19), are expressed in the taste-sensing terminal organ of larvae and in the taste bristles of the labella, legs, and wing margins of adult flies (Zelle et al., 2013). Importantly, knockdown of ppk11 and ppk19 resulted in loss of behavioral and electrophysiological responses to low salt concentrations. Similarly, disrupting ppk11 or ppk19 in adults negatively affected the response to high salt concentrations by eliminating avoidance behavior (Liu et al., 2003). The authors concluded that the DEG/ENaC channels encoded by ppk11 and ppk19 are critical to the detection of Na<sup>+</sup> and K<sup>+</sup> salts and contribute to the behavioral responses to various salt concentrations.

Analysis of the A. gambiae genome revealed that the ppk gene family members were reduced when compared to D. melanogaster with the mosquito consisting of 18 family members, of which 17 had homologs in the D. melanogaster genome which contains 31 total ppk genes (Zelle et al., 2013). Importantly, subfamily III of the Drosophila ENaC gene family (containing ppk19) was absent in the A. gambiae genome which suggests mosquitoes may use a different ppk gene product to detect salt.

In addition to salt detection, DEG/ENaC channels are responsible for mediating activity of water-sensitive GRNs in insects. The D. melanogaster gene ppk28 encodes a DEG/ENaC channel that is osmo-sensitive and is expressed in the taste bristles, but not in taste pegs, which was correlated back to a water-sensing neuron through imaging of an enhancertrap Gal4 line (Cameron et al., 2010). To test the functional role of ppk28, the authors generated a ppk28 null mutant and performed extracellular bristle recordings of l-type labellar sensilla. Recordings showed the mutant cells were completely insensitive to water, but were equally sensitive to sucrose when compared to controls (Cameron et al., 2010). The localization and functional data suggest ppk28 encodes an ENaC channel that responds to low osmolarity to mediate both GRN and behavioral responses to water. While it is evident that ppk gene products are part of the physiological cascade to detect water, it is likely that ENaC is functionally coupled to a series of transporters. For instance, water transport

channels, such as aquaporins, are expressed at the apical and basolateral membranes of rat taste cells and are critical for the gustatory response to water in mammals (Watson et al., 2007).

Gene products from the ppk family appear to be functionally conserved from mammals to insects and are responsible for detection of Na<sup>+</sup> and K<sup>+</sup> salts. Less is known regarding the functional conservation of their role as a sensor of osmolarity between insects and mammals. Similarly, the role of additional membrane transport pathways, such as aquaporins, and the interaction of these proteins with ppk gene products for tasting salt and sensing water remains to be determined. In summary, like TRP channel genes, ppk genes are potential gateways to activate avoidance behavior as transcripts of many conserved ppks are abundant in the taste organs of A. aegypti (Sparks et al., 2014).

#### UNDEREXPLORED ION CHANNELS IN INSECT GUSTATORY SIGNALING

Taste cells are excitable cells that use a vast array of receptors and ion channels during their activity (Bigiani et al., 2002). In particular, taste cells are known to express a variety of voltagesensitive ion channels, such as voltage gated (vg) sodium and potassium ion channels, that mediate the generation and/or propagation of action potentials (Herness and Sun, 1995; Chen et al., 1996; Ohmoto et al., 2006). The gustatory cells of mammals are known to have polarized epithelia with clear functional separation of apical and basolateral membranes (Purves et al., 2001). It is well established that proper function of any polarized epithelial tissue requires strict regulation and maintenance of the membrane potential and membrane resistance to enable an intracellular current that drives ion transport. Thus, it is reasonable to speculate that ion channels serve as gustatory receptors while also serving as critical components of the machinery responsible for proper polarization and conductance of GRNs.

The presence of these ion channels in mammalian taste cells combined with the conserved physiology of gustatory cells across organisms raises the intriguing possibility that these channels may also be functionally important for insect gustation. We provide a brief overview of select ion channels and their role in mammalian and insect taste systems below.

## Potassium (K+) Ion Channels

K <sup>+</sup> ion channels are diverse, widespread, and have been detected in almost every eukaryotic cell type examined (Latorre and Miller, 1983; Rudy, 1988). These channels represent a fundamental component of animal physiology by establishing and maintaining the membrane potential of cells, which is required for nearly all cellular functions (Urrego et al., 2014). They also play critical roles in signal integration and some function to link metabolism or cell signaling to electrical activity. Yet, despite the functional relevance of K<sup>+</sup> ion channels, the role of these channels in insect gustation and their potential utility are unexplored and ripe for discovery.

## Voltage-Gated K<sup>+</sup> (vg-K+) Ion Channels

In mammalian taste cells, two vg-K<sup>+</sup> channels, KCNQ1 and KCNH2, are expressed and involved in the repolarization of taste receptor cells. Interestingly, in one study the channels showed no specific taste modality (Ohmoto et al., 2006), which indicates these channels are likely involved in regulating the action potential. However, a study of rat fungiform taste receptors provided significant evidence that a vg-K<sup>+</sup> channel was involved in the detection and preference of polyunsaturated fatty acid (PUFA) molecules (Gilbertson et al., 1997). A delayed outwardly rectifying potassium current was reversibly inhibited by extracellular application of arachidonic acid (C20:4) or linoleic acid (C18:2) in whole cell patch clamp recordings from taste cells. Further, the same study showed that PUFAs activated inwardly rectifying potassium (Kir) currents. vg-K<sup>+</sup> channels regulate action potential firing and may be a target for taste stimuli (Kinnamon, 1992; Gilbertson, 1993), and Kir channels are important for establishing resting membrane potential and shunting current from the apical to basolateral membrane. Thus, a bimodal effect of PUFA on two distinct K<sup>+</sup> channel types with opposing conductance directions suggests PUFA may prolong the stimulus-induced depolarization to amplify the signal and ensure neurotransmitter release from the basolateral region of the cell (Gilbertson et al., 1997). Considering this and because vg-K<sup>+</sup> ion channels are exploitable insecticide targets (Bloomquist et al., 2014), the role of these channels should be studied in gustatory reception and signaling to enable a more holistic understanding of insect gustatory pathways and to test the deterrent nature of these channels for mosquito management.

## Inwardly Rectifying Potassium (Kir) Channels

Kir channels characterized from taste cells of rats are weak to moderate inward rectifiers (Sun and Herness, 1996) and contribute to both the resting and active states of the membrane potential (Hibino et al., 2010). In glial cells, Kir channels function as a route of K<sup>+</sup> clearance in the central nervous system of mammals (Kofuji and Newman, 2004; Neusch et al., 2006). Similar to the central nervous system, repetitive firing of taste cells will result in elevation of extracellular potassium ions that requires homeostatic mechanisms to clear K<sup>+</sup> ions from the extracellular space and distribute them back to areas of low intracellular K<sup>+</sup> concentration gradient. Indeed, Kir1.1, or ROMK, is localized at the apical tip of rat taste cells above the apical tight junctions, and was speculated to function as a route for buffering K<sup>+</sup> gradients during taste cell activity (Dvoryanchikov et al., 2009). Previous reports indicated that Kir channels are responsible for buffering K<sup>+</sup> ion gradients during neural activity of D. melanogaster (Chen and Swale, 2018) in a near identical manner as for mammals. This conserved role of mammalian and D. melanogaster neural Kir channels in buffering K <sup>+</sup> gradients in mammalian taste cells indicates these channels may serve a similar function in mosquito gustatory systems.

In addition to establishing K<sup>+</sup> gradients during cell function, Kir channels mediate transduction for sour and sweet taste (Yee et al., 2011; Ye et al., 2016). The mechanism of sweet taste

in mammals is not completely understood because knockout of the gene encoding the determinant of saccharin and sugar preference (T1r3) (Fuller, 1974) does eliminate the response to glucose or other sugars (Damak et al., 2003). Another type of Kir channel, ATP-gated Kir (KATP) channels, which serve as metabolic sensors in a variety of mammalian cell types, were co-expressed in taste cells with sugar transporters and glucose sensor proteins (Yee et al., 2011). Based on electrophysiological studies that confirmed KATP channel current to be functional, it was concluded that these channels regulate taste sensitivity to sweet molecules according to metabolic needs (Yee et al., 2011). KATP channels are underexplored in insects when compared to mammals, even though these channels are critical for a variety of physiological functions in taxonomically diverse arthropods, such as innate antiviral immunity (Eleftherianos et al., 2011; O'Neal et al., 2017a), honeybee heart function (O'Neal et al., 2017b), salivary gland function and feeding (Swale et al., 2017). Future work should investigate the physiological role and toxicological relevance of Kir/KATP channels in mosquito gustation.

## Voltage-Gated Chloride (Cl−) Channels

Scant information exists regarding the expression patterns or physiological role of chloride (Cl−) channels in arthropod gustatory systems. However, previous work suggests that mammalian taste cells possess several types of Cl<sup>−</sup> channels that play a key role in signal transduction of taste cells (Miyamoto et al., 1998; Herness and Sun, 1999). It was suggested that vg-Cl<sup>−</sup> channels contribute to the membrane potential and electrical excitability but are not involved in the initiation of action potentials (Huang et al., 2005). Immunohistochemical studies show that ClC-4 and ClC-4A are expressed on the plasma membrane as well as intracellular membranes of taste cells (Zhang et al., 2013), suggesting a possible role in neurotransmitter uptake and regulation of synaptic activity. ClC-4A may also be a candidate Cl<sup>−</sup> channel for acid transduction in sour taste by contributing to acidification of intracellular organelles (Huang et al., 2005). Also, ClC-3 is expressed in the synaptic vesicles of taste neurons and dissipates the membrane potential generated by the inevitable buildup of H<sup>+</sup> by serving as an electrical shunt for vesicular acidification (Huang et al., 2005). While Cl<sup>−</sup> channels are likely a component of sour taste, Kir2.1 may also function in tandem with a proton pump for sour taste transduction in mammals (Ye et al., 2016). Together, the results suggest that detection of sour taste is a complex process that remains to be fully elucidated. In addition to the studies on mammalian taste cells, mutation of a gene encoding a glutamategated chloride channel in the nematode Caenorhabditis elegans results in reduced gustatory plasticity. We suggest a need to explore this functional role of Cl<sup>−</sup> channels in insect gustation.

#### REFERENCES

Abuin, L., Bargeton, B., Ulbrich, M. H., Isacoff, E. Y., Kellenberger, S., and Benton, R. (2011). Functional architecture of olfactory ionotropic glutamate receptors. Neuron 69, 44–60. doi: 10.1016/j.neuron.2010.11.042

## CONCLUDING REMARKS

A full understanding of how individual odorants, tastants, or blends are detected, converted to neural signals and processed will depend on determining the functional relationships between all chemosensory gene families and the cells expressing them. Characterizing the response profiles of receptor complexes and comparing these responses among diverse mosquito species will further our understanding of how these successful animals have filled so many ecological niches and rapidly adapted to host availability.

Recent advances in our ability to quickly and reliably edit target genes in the germline of mosquitoes should help uncover unknown roles of key molecular components of olfactory and gustatory tissues. These studies are tedious and costly, but clarity of function will help define convenient targets for the development of novel repellents and antifeedants. In addition to high-throughput chemical screens, sophisticated modeling and simulation software can be developed and used to discover the most likely compounds capable of selectively activating or blocking neural pathways associated with harmful mosquito behaviors. Central to the creation of new vector control strategies is achieving greater resolution of ORN/GRN/IR function and the interactions between receptor complexes, ion channels and host-derived ligands.

## AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

#### FUNDING

JS and GB were supported by funds from High Point University (HPU) David R. Hayworth's School of Arts and Sciences and the Undergraduate Research and Creative Works program. LB and SP were supported by an NIH grant R15-GM096142-2 (to LB) and the University of Richmond School of Arts and Sciences.

## ACKNOWLEDGMENTS

JD thanks Dr. Krista Stenger, Chair, Department of Biology, University of Richmond for office space, and Boatwright Memorial Library, University of Richmond for reference services that facilitated his contributions to this study. JS gives thanks to HPU student Amanda Smith for her contributions in manuscript preparation.

Al-Anzi, B., Tracey, W. D. Jr., and Benzer, S. (2006). Response of Drosophila to wasabi is mediated by painless, the fly homolog of mammalian TRPA1/ANKTM1. Curr. Biol. 16, 1034–1040. doi: 10.1016/j.cub.2006.04.002

Arensburger, P., Megy, K., Waterhouse, R. M., Abrudan, J., Amedeo, P., Antelo, B., et al. (2010). Sequencing of Culex quinquefasciatus establishes a platform for

mosquito comparative genomics. Science 330, 86–88. doi: 10.1126/science. 1191864


biology, genetics, and evolution. Proc. Natl. Acad. Sci. U.S.A. 112, E5907–E5915. doi: 10.1073/pnas.1516410112





**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Sparks, Botsko, Swale, Boland, Patel and Dickens. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Odorant Receptors and Odorant-Binding Proteins as Insect Pest Control Targets: A Comparative Analysis

#### Herbert Venthur 1,2 and Jing-Jiang Zhou3,4 \*

<sup>1</sup> Laboratorio de Química Ecológica, Departamento de Ciencias Químicas y Recursos Naturales, Universidad de La Frontera, Temuco, Chile, <sup>2</sup> Center of Excellence in Biotechnology Research Applied to the Environment (CIBAMA), Universidad de La Frontera, Temuco, Chile, <sup>3</sup> Department of Biological Chemistry and Crop Protection, Rothamsted Research, Harpenden, United Kingdom, <sup>4</sup> Jilin Provincial Key Laboratory of Animal Resource Conservation and Utilization, Northeast Normal University, Changchun, China

#### Edited by:

Shuang-Lin Dong, Nanjing Agricultural University, China

#### Reviewed by:

Man-Qun Wang, Huazhong Agricultural University, China Jalal Jalali Sendi, University of Gilan, Iran Claudio C. Ramirez, University of Talca, Chile

> \*Correspondence: Jing-Jiang Zhou jjzhouchina@163.com

#### Specialty section:

This article was submitted to Invertebrate Physiology, a section of the journal Frontiers in Physiology

Received: 05 April 2018 Accepted: 03 August 2018 Published: 24 August 2018

#### Citation:

Venthur H and Zhou J-J (2018) Odorant Receptors and Odorant-Binding Proteins as Insect Pest Control Targets: A Comparative Analysis. Front. Physiol. 9:1163. doi: 10.3389/fphys.2018.01163 Recently, two alternative targets in insect periphery nerve system have been explored for environmentally-friendly approaches in insect pest management, namely odorant-binding proteins (OBPs) and odorant receptors (ORs). Located in insect antennae, OBPs are thought to be involved in the transport of odorants to ORs for the specific signal transduction of behaviorally active odorants. There is rich information on OBP binding affinity and molecular docking to bioactive compounds as well as ample 3D crystal structures due to feasible production of recombinant proteins. Although these provide excellent opportunities for them to be considered as pest control targets and a tool to design pest control agents, the debates on their binding specificity represent an obstacle. On the other hand, ORs have recently been functionally characterized with increasing evidence for their specificity, sensitivity and functional roles in pest behaviors. However, a major barrier to use ORs for semiochemical discovery is the lack of 3D crystal structures. Thus, OBPs and ORs have not been analyzed comparatively together so far for their feasibility as pest control targets. Here, we summarize the state of OBPs and ORs research in terms of its application in insect pest management. We discuss the suitability of both proteins as pest control targets and their selection toward the discovery of new potent semiochemicals. We argue that both proteins represent promising targets for pest control and can be used to identify new super-ligands likely present in nature and with reduced risk of resistance development than insect pesticides currently used in agriculture. We discuss that with the massive identification of OBPs through RNA-seq and improved binding affinity measurements, these proteins could be reconsidered as suitable targets for semiochemical discovery.

Keywords: insect olfaction, modulators, antagonists, agonists, pest management, odorant binding, chemosensory receptors

## INTRODUCTION

The human population has increased dramatically and is predicted to reach 9 billion in 2050. Food crops must be cultivated and managed to meet their demand and to increase their resistance against the damage by insect pests and crop diseases. Such situations have been managed mainly by artificial chemicals, such as insecticides and fungicides, whose persistence has caused food contamination, environmental, and health concerns, and called for alternative and integrated pest management strategies to reduce the use of these chemicals. Current specific examples are the concerns on the insecticide neonicotinoids for their role in decreasing honey bee populations (Godfray et al., 2014), the resistance to a wide range of insecticides that the peach potato aphid, Myzus persicae, has acquired because of intensive insecticide applications (Bass et al., 2014) and the spreading through globalization of several deadly diseases that are transmitted by mosquitoes, such as malaria, yellow fever, dengue and Zika (Jones et al., 2008). Therefore, the use of environmentally friendly approaches has become an attractive strategy to manage insect pests through the identification of behaviorally active chemicals (i.e., semiochemicals) to target insect olfaction systems and to either manipulate insect pest behaviors away from food crops or interrupt their sexual behaviors (Zhou, 2010; Pickett, 2014). Recently, two alternative targets in insect periphery nerve system have been explored for environmentally-friendly approaches in insect pest management, namely odorant-binding proteins (OBPs) and odorant receptors (ORs). Although chemosensory proteins (CSPs) have been also identified and reported to bind odorants (Iovinella et al., 2013; Li H. L. et al., 2016; Peng et al., 2017), their diverse tissue expression and attributed function as well as limited structural studies (e.g., only 5 crystal/NMR structures solved), have made them less attractive as targets.

Insect OBPs have been shown to increase the sensitivity of ORs to odorants using the Xenopus oocyte heterologous expression systems and voltage-clamp technique (Syed et al., 2006; Sun et al., 2013b; Zhang Q. H. et al., 2017) and HEK293 cell expression system and Ca-imaging (Grosse-Wilde et al., 2006). So far, a repertoire of OBPs have been identified in small hair-like structures (i.e., sensilla) projected at the surface of antennae from a wide range of insect species and considered as carriers to interact with semiochemicals during peripheral signal transduction (Vogt and Riddiford, 1981; Klein, 1987; Maida et al., 1993; Zhou, 2010; Pelosi et al., 2014). The functional roles of OBP57d and OBP57e of Drosophila sechellia and Drosophila melanogaster in the host selection have been demonstrated (Matsuo et al., 2007). Alternative to heterologous expression and binding assays for OBPs, the authors knocked out the expression of OBP57d and OBP57e in D. melanogaster and demonstrated that both proteins are important for the unique food preference of D. sechellia. Furthermore, when OBP57d/e genes were introduced from D. sechellia to D. melanogaster, the oviposition behavior of D. melanogaster shifted to the D. sechellia's host plant Morinda citrifolia. Therefore, OBPs have been explored as targets for semiochemical discovery. For instance, the OBP1 of the mosquito Culex quinquefasciatus CquiOBP1 was used to identify an attractive blend comprising trimethylamine (TMA) and nonanal by using gas chromatography-electroantennographic detection (GC-EAD) and in vitro binding assays along with field bioassays (Leal et al., 2008). Likewise, the identification of a potent attractant (methyl eugenol) of the fruit fly Bactrocera dorsalis was performed based on a general odorantbinding protein (GOBP) BdorGOBP (Jayanthi et al., 2014). In this study, the authors identified methyl eugenol by protein structure prediction, molecular docking and dynamics along with tryptophan fluorescence quenching assay followed by behavioral bioassays of 25 chemicals. More recently, the OBP7 of the parasitoid wasp Sclerodermus sp. (SspOBP7) was used to screen behaviorally active chemicals (Yi et al., 2018). From a group of 19 chemicals, only 6 were found to bind to SspOBP7 in the fluorescence quenching binding assays. Subsequent behavioral olfactometry bioassays confirmed that Sclerodermus sp. showed significant preference to only 2 compounds, (+)-α-longipinene and terpinolene that had a good binding affinity with SspOBP7. This so called reverse chemical ecology approach has accelerated the understanding of olfactory mechanisms and the discovery of active chemicals that could be used to manipulate insect behaviors for pest management (Leal, 2017). Since then, the discovery of ORs have further provided more sensitive targets for such reverse chemical ecology (Wetzel et al., 2001; Sakurai et al., 2004; Corcoran et al., 2014; Wang et al., 2016). These ORs act as the secondary filter for olfactory information and molecular recognition in insect antennae, converting chemical signals to electrical impulses that provoke behavioral responses (Kaissling, 2013; Bohbot and Pitts, 2015). The number of ORs varies across insect species from around 40 candidates in Lepidopterans, such as the codling moth Cydia pomonella (Bengtsson et al., 2012), the oriental leafworm moth Spodoptera litura (Feng et al., 2015) and the tobacco hawk moth Manduca sexta (Grosse-Wilde et al., 2011), to more than 70 candidates in the African malaria mosquito Anopheles gambiae (Rinker et al., 2013) and the pea aphid Acyrthosiphon pisum (Richards et al., 2010), and 170 candidates annotated in the honey bee Apis mellifera genome (Robertson and Wanner, 2006). These insect receptors function only with a highly conserved and co-expressed co-receptor (ORco) as heteromeric transmembrane complexes heterologous expression systems. This is completely different from those of other animal G-protein coupled receptors (Civelli et al., 2013), which provides unique opportunities for the development of insect pest specific control agents.

Although functional and structural characteristics as well as biotechnological applications of insect OBPs have been widely reviewed (Zhou, 2010; Leal, 2013; Pelosi et al., 2014), a comparative analysis of ORs and OBPs in terms of their use as targets for semiochemical discovery has not been made so far. On one hand, OBPs highlight as extracellular soluble proteins with significant experimental 3D structure information available and straightforward protocols for semiochemical screening by means of fluorescence binding characterizations. However, their broad specificity, wide distribution in non-olfactory tissues and secondary functions (e.g., scavengers, solubilizers, and regeneration) make their selection as targets a difficult task for semiochemical discovery. Similarly, ORs are transmembrane

proteins with no experimental 3D structure information available so far and more sophisticated to be expressed and purified in heterologous systems. However, most of the studied ORs show high sensitivity to very specific chemical groups. Some insect ORs have been successfully used as targets to identify new semiochemicals, such as C. quinquefasciatus OR36 (Choo et al., 2018). It is also possible to use them in the identification of antagonists that could serve as a new approach to disrupt the behavior of a given insect pest (Chen and Luetje, 2012, 2013, 2014). Therefore, the objective of this review is to summarize the suitability and application of both ORs and OBPs in view of the discovery of pest control agents and discuss their further perspectives in insect pest management strategies.

#### INSECT ODORANT-BINDING PROTEINS: FUNCTIONAL AND STRUCTURAL FEATURES TOWARD PEST CONTROL AGENT DISCOVERY

The first insect OBP was identified more than 30 years ago by Vogt and Riddiford (1981). Currently, a large number of OBPs, particularly pheromone-binding proteins (PBPs) and general odorant-binding protein (GOBP) in Lepidopterans, has been identified across insect species with more than 2000 amino acid sequences of insect OBPs deposited so far in NCBI database (https://www.ncbi.nlm.nih.gov) and classified into subgroups based on the number of highly conserved cysteine residues: classic, minus-C, plus-C, and atypical (Zhou et al., 2004; Venthur et al., 2014) after initial classification of PBP, GOBP, and antennal specific protein (ASP). Moreover, recent analyses of insect antennal transcriptomes have shown that insects express several OBPs highly in their antennae (**Table 1**). Despite the identification of ample insect OBPs, most of the functional studies have been relied on the OBPs expression profiles in insect antennae as well as their activity-structure binding relationships determined by means of fluorescent competitive binding and molecular docking.

## Specificity of Insect OBPs

Recent debate around the ligand binding specificity of OBPs has caused concerns for their suitability as targets for semiochemical discovery, with some authors reporting a broad binding capacity to several volatiles (Campanacci et al., 2001; Zhou et al., 2004; Zhou, 2010; Pelosi et al., 2014), while others supporting the remarkable specificity of some OBPs (Qiao et al., 2009; Damberger et al., 2013; Li et al., 2013, 2017). Traditionally, these studies have been performed by the competitive binding assays based on fluorescence displacement (Campanacci et al., 2001). In this competitive displacement binding assay, a fluorescent probe, commonly N-phenyl-1-napthylamine (1-NPN), is used for initial binding with OBPs, which is then displaced by the ligands of interest. Therefore, ligands with a high affinity are those with the strong ability to displace 1-NPN from OBP binding pockets at low concentrations of dissociation constants (KD) and inhibitory concentrations (IC50) assuming the protein is 100% active and the binding stoichiometry is 1:1.

These studies propose some OBPs as specific for chemical properties of compounds. Particularly, the sub-classes OBPs, such as PBPs and GOBPs of Lepidopteran, have shown high specificity for volatile compounds with either particular hydrocarbon lengths or specific functional groups, such as aldehydes, alcohols or esters (Zhou, 2010). For example, it was found that among 16 tested compounds, three compounds with 12 carbon (C12) atoms [codlemone, 1-dodecanol and (E,E)- 2,4-dodecadienal] showed higher affinities to the PBP1 of the codling moth C. pomonella (CpomPBP1) with binding affinity constants K<sup>D</sup> between 2.73 and 5.90µM (Tian and Zhang, 2016). The GOBP2 of Bombyx mori had higher affinity to the sex pheromone bombykol than to its isomer bombykal (Zhou et al., 2009). Likewise, the GOBP1 and GOBP2 of S. litura had stronger binding to C14-C16 alcohol-pheromone analogs, such as (Z)-9 tetradecanol, (Z)-9-hexadecanol, (Z)-11-hexadecanol, and (E)- 11-hexadecanol in fluorescence binding assays and molecular modeling (Liu N. Y. et al., 2015). On the other hand, the presence of phenolic groups in chemicals such as eugenol, isoeugenol, and 4-vinylguaiacol showed to play a key role for the high affinity of the OBP14 in the honeybee A. mellifera (Schwaighofer et al., 2014).

OBPs have also been used to screen a large number of chemicals. For instance, nonanal, acetophenone, 6-methyl-5 hepten-2-one and some terpenoids from 41 host odorants showed high binding affinities (11–16µM of KD) to the OBP6 of the alfalfa plant bug Adelphocoris lineolatus (AlinOBP6) through fluorescence competitive binding assays (Sun et al., 2017). Interestingly, AlinOBP6 exhibited a good binding affinity to nonvolatile compounds, such as quercetin, gossypol, rutin hydrated, and (–)-catechin, suggesting a broad specificity of AlinOBP6 and likely a role in mechanisms to respond to volatile and non-volatile compounds. Similarly, the binding of 45 volatile organic compounds to the OBP8, OBP9 and OBP10 of the endoparasitoid Microplitis mediator was tested by the competitive binding assays (Li et al., 2014). Their findings suggested that nonane, nonanal, farnesol, β-ionone, nerolidol, acetic ether, and farnesene have a high binding affinity to the OBPs in the µM range. Later in behavioral bioassays, β-ionone, nonanal, and farnesene showed attractant activity while nonane and farnesol showed repellent activity. This study supports the role of insect OBPs as targets to discover new semiochemicals that can act as either attractants or repellents, and to screen for super-ligands whether with their native or chemically optimized chemical structure (Hooper et al., 2009). More recently, the screening of host odorants and/or sex pheromones using the fluorescencebased binding assay has been reported for other OBPs with much better binding affinities, though the best binding affinity of the ligands are still in µM ranges. For example, multiples OBPs of the oriental fruit moth Grapholita molesta have been studied, showing high affinity of (E)-8-dodecadienyl acetate (K<sup>D</sup> of 2.18µM) to OBP8, 11 and 15, (Z)-8-dodecenyl acetate (K<sup>D</sup> of 1.09µM) to PBP1 and PBP2, and dodecanol (K<sup>D</sup> of 5.10µM) to OBP4, 5 and 10 (Li G. W. et al., 2016; Chen et al., 2018; Zhang et al., 2018). Similarly, the OBP1 of the scarab beetle Hylamorpha elegans suggested β-ionone as the best ligand with a K<sup>D</sup> value of 6.9µM among 29 tested host odorants (Venthur et al., 2016). The TABLE 1 | Number of insect OBPs and ORs identified from antennal or head transcriptome studies based on RNAseq data.


OBP13 of Japanese pine sawyer Monochamus alternatus showed butylated hydroxytoluene as the best ligand with K<sup>D</sup> of 0.77µM in 20 tested host odorants (Li et al., 2017), and farnesene was highlighted as the best ligand with a remarkable K<sup>D</sup> of 0.86µM for the OBPm2 of the white-striped longhorn beetle Batocera horsfieldi among 58 host odorants (Zheng et al., 2016).

Thus, comprehensive studies on insect OBPs using the competitive binding assays have reported specific groups of high affinity ligands in the µM range from a broad list of candidates. The OBP binding studies so far face a major challenge to measure the binding affinity from µM to nM range, which is normally regarded as high affinity binding in other biokinetic studies. However, it has been well-established for semiochemical discovery by these binding studies using OBPs that: (1) Reproducible protocols are available to clone, express, purify and test binding specificity of insect OBPs and (2) Fluorescence-based binding assays provide a robust technique to perform the rapid experimental screening for a relatively large number of chemicals. However, recent findings for the OBP1 of Aenasius bambawalei (AbamOBP1) report the binding of the protein with lower K<sup>D</sup> at acid pHs, inconsistent with the better binding at basic pHs in previous studies. The authors report that the binding stoichiometry between AbamOBP1 and tested ligands was not 1:1, which is likely caused by the presence of dimers or even trimers of OBPs and, therefore, a 100% active protein could not be assumed, suggesting false positives from the competitive binding assays (Li et al., 2018). It appears that a combined methodology such as fluorescence intrinsic quenching assays (Bette and Breer, 2002) could be in better accordance with behavioral assays. An example of dimeric forms of OBPs have been reported by Wang et al. (2013), where mixtures of recombinant OBP1 and 2 of the scarab beetle Holotrichia oblita as well as OBP2 and 4 of the same insect, showed higher binding affinities to odorants, such as β-ionone and retinol, than the OBPs alone. Later, the authors revealed that such OBP pairs were actually co-localized each in the same sensilla by immunocytochemical analyses.

#### Structural Features of Insect OBPs

The heterologous expression of insect OBPs in bacteria and the subsequent three-dimensional (3D) structure determination by either X-ray crystallography or NMR or the prediction by homology modeling provide substantial information and have attracted a great interest recently for OBPs as a suitable target for pest management strategies. The research of insect OBPs has started to focus on structural characterizations since Sandler et al. (2000) reported the first crystal structure of the PBP1 of B. mori and its interactions with the sex pheromone, bombykol [(E,Z)- 10,12-hexadecadien-1-ol]. There are about 70 X-ray crystal and 5 NMR protein structures solved and deposited in Protein Data Bank database (https://www.rcsb.org/pdb/home/home.do) to date due to their small molecular weight and ease to be expressed and purified. Most of the structures are related to OBPs of A. gambiae, D. melanogaster, B. mori, and A. mellifera with 24, 12, 10, and 9 structures, respectively. The Classic OBPs are characterized by 6 α-helices connected by 3 disulfide bridges in a specific motif pattern C1-X25−30-C2-X3-C3-X36−42- C4-X8−14-C5-X8-C6 (Xu et al., 2003), being the most studied and reviewed OBP subfamily so far (Zhou et al., 2004; Pelosi et al., 2006, 2014; Venthur et al., 2014; Brito et al., 2016). However, the identification of OBPs from other non-Lepidopteran insects relieved that such sequence motif patterns can vary and have been further grouped as minus-C OBP subfamily with 4 cysteines residues (Hekmat-Scafe et al., 2002; Weinstock et al., 2006), plus-C OBP subfamily with 3 extra cysteines and a conserved proline (Zhou et al., 2004) and atypical OBP subfamily with more cysteines in C-terminal section (Xu et al., 2003). On the other hand, the diversity and non-homologous feature of OBPs among insect genera could serve as advantages in the development of semiochemicals or even insecticides for specific insect species.

Indeed, molecular modeling approaches, such as homology modeling, have allowed, in most cases, an extensive study of their structural characteristics in complement with in vitro binding assays (Venthur et al., 2014). This computer-based method (i.e., in silico) is an approach of using experimental 3D structures as templates to predict the 3D structure of a target protein based only on its amino acid sequence (Leach, 2001; Schmidt et al., 2014). Early structural studies of insect OBPs were limited by the availability of a few crystal structures and the low percentage of sequence identity to known OBP structures (e.g., <30%). Thus, probable 3D arrangements of OBPs with no further refinement were reported (Campanacci et al., 2001; Ban et al., 2003; Tsuchihara et al., 2005; Paramasivan et al., 2007). These studies have served as a good starting point. However, with more X-ray crystal and NMR structures available for different insect orders, this in silico complementary protein/ligand interaction research has become more comprehensive and routine with methods such as dynamics simulations and molecular docking. For example, the ligand-binding mechanisms of minus-C OBP21 of Dastarcus helophoroides (DhelOBP21) were studied first by homology modeling and molecular docking and then supported by fluorescence binding assays (Li D. Z. et al., 2015). The authors proposed that hydrophobic interactions between ligands and DhelOBP21 are more crucial for binding than hydrogen bonds, and molecules with a size of 100-125 Å<sup>3</sup> are the most suitable. More recently, the structural approaches based on extensive dynamics simulations (110 ns) of DhelOBP21-ligand complexes [(+)-β-pinene, camphor and β-caryophyllene] have shown the remarkable conformational stability of DhelOBP21/(+)-βpinene complex which strongly supports the behavioral activity of D. helophoroides (Yang et al., 2017). Similarly, Tian et al. (2016) explored the structural features of the PBP2 of C. pomonella and demonstrated that hydrophobic and hydrogen bond interactions as well as chain length of C12 atoms and the unsaturation of compounds are key features during ligand binding. Likewise, the 3D structure prediction for the OBP of B. horsfieldi (BhorOBPm2) helped to demonstrate that long chain (C14) compounds had higher affinities than those with shorter chains due to the flexibility of its binding pocket (Zheng et al., 2016). The conformational flexibility of OBPs for odorant binding has also been reported for the minus-C OBP14 of A. mellifera (Schwaighofer et al., 2014). Interestingly, the authors compared the wild-type OBP14 with a mutant version of the OBP14 in which a third disulfide bridge was added and evaluated their thermal stability when they bound to volatiles. Their findings showed that a constricted flexibility in the mutant OBP14 resulted in its lower binding affinities than the wildtype OBP to some volatiles, such as eugenol, methyl eugenol, isoeugenol, and other phenolic-based compounds.

The structural studies in OBPs have allowed more specific research into the mechanisms of odorant binding and release in order to predict the OBP/ligand interactions in the olfactory system of insects. This could further advance in using OBPs as targets and a tool to design pest control agents. It has been reported that insect OBPs display an outstanding pH-dependent mechanism of odorant binding and release, which certainly contributes to the specific properties of these proteins. This process supports the idea that these proteins are able to bind odorants at a basic pH (6.5), transport and release them at an acid pH (4.5) (Lautenschlager et al., 2005; Xu et al., 2011; di Luccio et al., 2013). It is proposed from the structure studies that the odorant molecule is ejected to ORs because the long C-terminal section displaces it from the OBP binding pocket when the C-terminal section shifts at acid pHs from an extended structure to a helical form and inserts inside the binding site. The pH-dependent approach helped to elucidate the selective role of a PBP in B. mori (BmorPBP). Thus, Damberger et al. (2013), through the study of pH-dependent polymorphism of BmorPBP by NMR, reported that this protein is able to eject bombykol near the OR at an acid pH, whereas ligands with low binding affinity are released before they reach the vicinity of receptors. The pH-independent structures are also observed in the OBP1 of Locusta migratoria (LmigOBP1) (Zheng et al., 2015). However, the studies on some OBPs have proposed different mechanisms. For instance, although the long C-terminal tail in the GOBP2 of B. mori also forms an α-helix, it is located across the N-terminal helix and not buried into the binding site as in BmorPBP1 (Zhou et al., 2009). Likewise, the OBP13 of M. alternatus (MaltOBP13) exhibits high binding capacity at acid pH (5.0) than at basic pH (7.4) for several ligands, especially α-terpinolene, with a K<sup>D</sup> of 56.93µM at pH 7.4 and 7.20µM at pH 5.0 (Li et al., 2017).

The identification of semiochemicals, 3D structure prediction of OBPs, their binding mechanisms and the characterization of specificity determinants have provided an outstanding opportunity to use insect OBPs as targets in pest control management. Furthermore, the introduction of other structurebased methodologies, such as quantitative structure-activity relationship (QSAR) (Oliferenko et al., 2013), will enhance OBP's roles as targets for semiochemical discovery and optimization.

## INSECT ODORANT RECEPTORS AS PROMISING PEST CONTROL TARGETS

Insect ORs are another important component of the periphery nerve system and a key player in the signaling transduction pathway in the antennae for insect behaviors, which begins with OBP binding to ligands, transporting to ORs, and terminating by degrading enzymes which are thought to remove the ligands away from the neuron dendrite of ORs (Leal, 2013). Insect ORs are G protein-coupled receptors (GPCRs) with seven transmembrane domains. Hopf et al. (2015) suggested the 3D structure of D. melanogaster OR85b and ORco using a coevolution homology modeling approach. Their findings indicate a structural arrangement based on seven TMHs and a C-terminal faced to the extracellular section and N-terminal to the intracellular section (**Figure 1**) unlike the GPCRs of other animals (Tsitoura et al., 2010), being a different topology from those of animal GPCRs (Katritch et al., 2013). This unique feature of insect ORs from animal GPCRs places them as ideal insect specific targets to be deployed for pest control management.

Currently, a range of 40–80 ORs have being identified based on antennal transcriptome data from insect species, such as moths, beetles, flies, and mosquitoes (**Table 1**) with even more than 500 ORs based on genome sequencing in ants (**Table 2**). Uniquely, despite the divergence of insect ORs, there is a highly conserved OR (Larsson et al., 2004; Jones et al., 2005) across insect species, originally identified as OR83b and later renamed as ORco (Vosshall and Hansson, 2011). It forms a functional heteromeric complex with other ORs (ORx/ORco) (Neuhaus et al., 2005). This gives another dimension for insect OR to be consider as pest control targets in addition to their unique topology. However, the complex of insect ORs with ORco in vivo in the dendrite of olfactory receptor neurons (ORNs) are less reported. The ORco subunit has received special attention due to its high conservation across insect species, from structural features such as several motifs in its C-terminal section (Ray et al., 2014) to its ancestry presence before the appearance of other ORx (Missbach et al., 2014). Particularly, the TMH6 of ORco has been proposed as a pore domain that plays a role in the regulation of cation flow (Wicher et al., 2008; Carraher et al., 2015). In the same direction, the function of ORco has been elucidated by the use of RNA-interference (RNAi) that knocks down the expression of ORco genes. For instance, when the ORco expression was knockdown in the gypsy moth Lymantria dispar, its antennal electrophysiological response (electroantennographic response or EAG) to the sex pheromone (i.e., disparlure) was significantly decreased from 1.472 to 0.636 mV (Lin et al., 2015). Similarly, RNAi was used to reduce the expression of ORco gene in the true bug Apolygus lucorum, resulting in a decrease of EAG responses to two semiochemicals, (E)-2-hexenal and (E)-2-hexenyl butyrate (Zhou Y. L. et al., 2014). More recently, ORco knockdown by RNAi negatively affected the oviposition and blood ingestion in the Chagas disease vector Rhodnius prolixus, which was later confirmed through a series of bioassays (Franco et al., 2016). These scientific evidence is consistent with the proposal that ORco forms metabotropic gated cation channels which controls threshold responses to odorants (Stengl and Funk, 2013). Likewise, RNAi approach supports the role of either a specific OR/ORco complex or ORco alone in the recognition of agonists. Hence, making them suitable targets for identifying or optimizing novel molecules with semiochemical activity in insect pest management (Taylor et al., 2012; Tsitoura and Iatrou, 2016).

## Functional Role of Insect ORs

The early elegant studies of expressing B. mori OR1 (BmorOR1) together with ORco (OR83b) in Xenopus laevis oocytes (Sakurai

acetylcholine; nAChR, nicotinic acetylcholine receptor.

et al., 2004) and HEK293T cells (Grosse-Wilde et al., 2006; Wicher et al., 2008) demonstrated that these receptors function as ligand-gated cation channels that can unleash the influx of extracellular Ca2<sup>+</sup> in ORNs (Sato et al., 2008). It has been proposed that ORs have an intracellular binding site for calmodulin (Carraher et al., 2015; Bahk and Jones, 2016), a ubiquitous protein in eukaryotes that modulates the function of target proteins via intracellular Ca2<sup>+</sup> signaling. Some structural domains sensitive to odorants, such as the extracellular loop 2 (ECL2) and the transmembrane helices 4 (TMH4), have been demonstrated through the mutations of several amino acids, particularly alanine 195 (Ala195) (Hughes et al., 2014). For instance, it was shown that the sensitivity of A. gambiae OR15 (AgamOR15) to acetophenone was significantly decreased when Ala195 was mutated to isoleucine. Rahman and Luetje (2017) confirmed that the key role of Ala195 in AgamOR15 is to function as a part of an inhibitor interaction site. Another study (Leary et al., 2012) showed that Ala148 in the OR3 of the moth Ostrinia nubilalis (OnubOR3) alters the response to a specific pheromone when it was mutated to threonine, decreasing ∼14-fold the sensitivity to (E)-11-tetradecenyl acetate, hence, selectively narrowing the specificity of OnubOR3.

More recently, the functional role of 17 ORs in Spodoptera littoralis toward 51 chemicals emitted by flowering plants was deorphanized using a high-throughput approach based on cloning and expression of the ORs in Drosophila fly embryos followed by single-sensillum recordings (SSRs) (De Fouchier et al., 2017). The authors propose that some receptors that recognize aromatic compounds have emerged first and are more conserved, whereas receptors tuned to terpenes and aliphatic



compounds (e.g., sex pheromones) have emerged more recently and evolved faster. In the same context, a specific expansion of ORs for floral odorants has been reported for generalist honey bees (e.g., A. mellifera and A. cerana), which is not present in specialist bees, such as Dufourea novaeanglicae and Habropoda laboriosa (Karpe et al., 2017). Thus, it seems that the range of host plants for a given insect leads to the divergence of ORs, showing the olfactory process as a constantly evolving system and specific. This would be a useful point for the design of pest control agents for a given insect species, and the evidence of a reduced possibility of developing resistance to these agents by the insect pests.

Besides the structural and functional characteristics of ORs, its divergence represents a putative guide of the different odor sources for what the insect olfactory system is tuned for. Supporting the above, a remarkable difference in ORs between the fruit fly D. melanogaster and the mosquito A. gambiae has been reported. An important specie-specific OR divergence has been detected so that more than 20 A. gambiae OR-related genes have no homologous partners in D. melanogaster, and around 18 D. melanogaster ORs have no corresponding genes in A. gambiae (Hill et al., 2002). Similarly, a specific expansion of 175 OR genes was identified through genome and phylogenetic analyses in the honey bee A. mellifera, which is potentially explained by their broad olfactory perception to chemicals, such as diverse pheromone blends and floral odors (Robertson and Wanner, 2006). On the contrary, less expansions are reported in the bedbug Cimex lectularius with only 48 genes encoding ORs, which is likely explained by its limited host range as blood feeder (Benoit et al., 2016). This dynamic evolutionary process for ORs provides other aspects in semiochemical identification, especially for those insect pests with a wide range of hosts.

#### Agonist Identification Using ORx Subunit

The functional study of ORs along with the conserved ORco has provided exquisite evidence of their role in perceiving chemicals regarded as agonists. It has been suggested that ORs could act as generalist or specialist for semiochemicals (Bohbot and Dickens, 2012). Thus, non-pheromonal compounds could be recognized by generalist ORs and pheromones are only recognized by specialist ORs (Hughes et al., 2010). For example, ORs from the noctuid moths S. littoralis and S. litura have been extensively studied in terms of their sensitivity to volatile agonists. Thus, the OR6 of S. littoralis (SlitOR6) was expressed in Drosophila olfactory neurons and found to specifically recognize a pheromone component of S. littoralis, (Z,E)-9,12-tetradecenyl acetate (Montagné et al., 2012). Similarly, OR13 and OR16 of Spodoptera exigua (SexiOR13 and SexiOR16) were functionally characterized against pheromone components. It was shown that SexiOR13 was highly sensitive to (Z,E)-9,12-tetradecenyl acetate and (Z)-9-tetradecenyl acetate, whereas SexiOR16 was even more sensitive to (Z)-9-tetradecenyl acetate only (Liu et al., 2013). More research has been published with the functional study of candidate pheromone receptors (PRs), such as the OR1 of B. mori (Syed et al., 2006) or Plutella xylostella, Mythimna separate, and Diaphania indica (Mitsuno et al., 2008; Liu Y. et al., 2018); the OR6, OR13, OR14, OR15 and OR16 of Heliothis virescens (Wang et al., 2011); the OR1, OR3, and OR6a of C. pomonella (Cattaneo et al., 2017) and the OR1 of M. sexta (Wicher et al., 2017). All these studies are carried out by selecting candidate PRs based on phylogenetic analysis and/or male specific expression. In the meantime, alternative approaches for precise PR selection might seem difficult due to a large number of ORs and variable expressions. For instance, are ORs differentially expressed when insects are faced to certain conditions such as the exposure to sex pheromones or virgin/mated? This has recently been probed for the OR3, OR6, and OR11 of S. exigua, where these genes were differentially expressed when the insects were exposed to synthetic pheromone (Wan et al., 2015). The authors reported more than 1- to 3-fold increase in the relative expression after the exposure. On the other hand, age and mating status seem not to affect the expression of ORs. It was found that the OR13 and OR15 of both H. virescens and H. subflexa are mainly expressed in males, and stable in terms of their relative expression for virgin males of 2 h, 1, 2, 4, and 8 d old as well as 4 day old mated males (Soques et al., 2010). Finally, it is worth mentioning that apart from PRs of Lepidopterans, the functional roles of non-pheromone receptors (non-PRs) have also been addressed. The specificity of BmorOR1 to bombykol was probed by Sakurai et al. (2004) through the heterologous expression of the OR1 in Xenopus oocytes and demonstrated the corresponding ORco as the essential unit for the function of the OR1. This allowed the functional study of other receptors, not only of Lepidopterans, but also of insect species from different orders, such as aphids, mosquitoes and beetles. An example is the OR12 of S. litura (SlitOR12), which was expressed in Xenopus oocytes and its sensitivity to odorants was tested by electrophysiology. Their results indicate that SlitOR12 is highly sensitive to (Z)-3 hexenyl acetate, a common green leaf volatile (GLV), suggesting a key role during oviposition and/or host location by females (Zhang et al., 2013). An OR from aphid A. pisum (ApisOR4) was functionally characterized through expression in Xenopus oocytes and electrophysiology (Zhang R. B. et al., 2017). Their findings suggest a specificity of ApisOR4 to 8 volatiles that belong to aromatic and terpenoid class. Similarly, the high sensitivity of A. pisum OR5 (ApisOR5) to the alarm pheromone, (E)-βfarnesene, was elucidated by Zhang R. et al. (2017) and further corroborated when the ApisOR5 was knocked down by RNAi treatments, resulting in A. pisum individuals not repelled by (E)-β-farnesene. The OR7, OR10, and OR88 of the mosquito Aedes albopictus were tested in terms of odor recognition (Liu et al., 2016). The authors report the OR10 and OR88 are highly sensitive to human-derived odorants, such as indole and 1 octen-3-ol. Contrary to what was expected, the mosquitoes that were treated with RNAi to significantly depress the OR10 and OR88 expressions were still able to respond to indole and 1 octen-3-ol. This may imply that other generalist ORs likely complement the lack of specialist receptors for host seeking behavior. More recently, the reverse chemical ecology approach has been reported based on the responses to 230 odorants by the OR36 of C. quinquefasciatus expressing in Xenopus oocytes, resulting in acetaldehyde as not only the strongest agonist, but also behaviorally active as oviposition attractant in bioassays (Choo et al., 2018). The specificity of 17 ORs from S. littoralis to low concentrations of ligands (pM range) (De Fouchier et al., 2017) has been demonstrated. Interestingly, some SlitORs (OR14, OR24, OR15, OR27, and OR29) even seemed to be sensitive at less than 1 pmol of ligand flux when SSRs were performed.

Although volatile compounds with agonist activity have been screened against ORs, a specific chemical with strong agonist effect on mosquitoes, 2-(4-Ethyl-5-(pyridin-3-yl)-4H-1,2,4 triazol-3-ylthio)-N-(4-ethylphenyl)acetamide (VUAA1), has opened the field of research (Jones et al., 2011; Taylor et al., 2012). Later studies by Taylor et al. (2012) provided evidence of VUAA1-derived chemicals, such as VUAA4, able to increase its agonist effect by 10-fold on ORco from A. gambiae, H. virescens, and Harpegnathos saltator. Interestingly, the authors reported that any change on amide substituents will cause a complete loss of agonist activity. This yields helpful insights into the structural requirements of agonists and the structure-activity relationship between VUAA analogs and ORs. Finally, despite the enhanced agonist activity of VUAA chemicals, its relatively high molecular weight (367.47 g mol−<sup>1</sup> for VUAA1) vs. volatile agonists, such as bombykol (238.42 g mol−<sup>1</sup> ), makes a direct volatile delivery of VUAA something not feasible. With that in mind, the searching for smaller structural analogs represents an interesting focus of research.

#### Antagonism Onto ORco Subunit

Along with the study of VUAA-related analogs that can act as strong agonists, the blockage of ORco by antagonists has also emerged to guide semiochemicals and pesticide design. Thus, a structural analog of VUAA1, VU0183254 (2-(4-Ethyl-5-furan-2-yl-4H-[1,2,4]triazol-3-ylsulfanyl)-1-

phenothiazin-10-yl-ethanone), was reported to inhibit ORco response, acting as allosteric modulator in A. gambiae and disrupting the recognition of agonists such as eugenol by the complex OR65/ORco (Jones et al., 2012). Other VUAA-structural analogs have also been reported as antagonists. An example is the N-,2-substituted triazolothioacetamide compounds OLC3 and OLC12 that disrupts the ORco response in a similar fashion in C. quinquefasciatus, A. gambiae, D. melanogaster, and O. nubilalis, suggesting a conserved binding site in ORco (Chen and Luetje, 2012). Considering the inhibition of ORco as a promising strategy to disrupt behaviors of insects, it seems that subsequent efforts should aim at the compounds with lower molecular weight than VUAA-derived antagonists. For example, OX1a (232 g mol−<sup>1</sup> ), tryptamine (160.22 g mol−<sup>1</sup> ) and isopropyl cinnamate (190.24 g mol−<sup>1</sup> ) were reported to have antagonist effect on ORco (Chen and Luetje, 2013, 2014; Tsitoura et al., 2015) with roughly half or less molecular weight than VUAA1. Nevertheless, future use of these antagonists should be studied carefully, since the blockage of the conserved ORco can affect not only harmful insects, but also beneficial ones.

Besides the antagonist effect probed in vitro, the evidence at behavioral level supports the idea that structural analogs of pheromones can function as antagonists. For example, Sellanes et al. (2010) reported the inhibition of sexual response in the honeydew moth Cryptoblabes gnidiella when the structural analogs, (Z)-9-tetradecenyl formate and (Z)-11-hexadecenyl formate, were added to synthetic sex pheromone, (Z)-11 hexadecenal and (Z)-13-octadecenal, in wind tunnel tests. This pheromone antagonist effect was later corroborated in field assays, where the trapping of C. gnidiella males decreased in a dose-dependent pattern. The pheromone antagonism has also been reported for B. mandarina, an ancestor of B. mori (Daimon et al., 2012). Their findings corroborate bombykol as the sex pheromone, and bombykal [(E,Z)-10,12-hexadecadienal] and bombykyl acetate [(E,Z)- 10,12-hexadecadienyl acetate] as antagonists, which strongly inhibited the attraction of males in field to the sex pheromone bombykol. More recently, evidence of pheromone antagonism was reported for the snout moth Herpetogramma submarginale. When (Z)-13-hexadecenol was added to its sex pheromone, (Z)-13-hexadecenyl acetate, significantly decreased the number of males captured in field (Yan et al., 2015). The pheromone antagonism seems based on the differences in chemical functional group such as alcohols, aldehydes and esters depending on the insect species. Nevertheless, the antagonist effect of these structural analogs might not be due to ORco inhibition but the specificity of ORx to antagonists. A recent study suggests that the OR16 of Helicoverpa armigera is able to specifically recognize the pheromone antagonist, (Z)-11 hexadecenol (Chang H. et al., 2017). The authors supported the specific role of OR16 considering that H. armigera females emit the antagonist compound along with its sex pheromone ((Z)-11-hexadecenal and (Z)-9-hexadecenal) as a strategy to avoid non-optimal mating with immature males.

Outstandingly, when the OR16 was knocked down by the genome editing technique CRISPR/Cas9 and H. armigera males were tested by electrophysiology and behavioral assays, no EAG response was recorded and males tried to mate with immature females.

## ODORANT RECEPTORS VS. BINDING PROTEINS: PROS AND CONS FOR INSECT PEST MANAGEMENT

For the case of OBPs, the ligand specificity and mechanisms of OBPs represent controversial aspects, which seems strongly dependent on the methods used for the measurement of ligand affinity. For instance, it has been reported that PBPs, such as those from the moths' P. xylostella and Eogystia hippophaecolus, can bind both sex pheromone components and analogs (Sun et al., 2013a; Hu et al., 2018). This suggests that downstream players such as ORs could enhance specificity and sensitivity of odorant reception. Recent evidence supports that the co-expression of PBPs and PRs can increase the sensitivity toward pheromones. For example, multiple combinations from PR1-4 and PBP1-4 were used to test their response to sex pheromone components of the moth Chilo suppressalis (Chang et al., 2015). The authors found a significant increase in sensitivity of response toward (Z)-11-hexadecenal when PR4 and PR6 were co-expressed with PBP4. Although the interaction of these proteins could arise a new level of research as pest control targets, the different pairing of PRs and PBPs shed lights on the complexity of the olfactory system in insects, making the approach a difficult task for a large set of compounds and proteins to test. Despite the above, insect OBPs are of small molecular size with easy production of recombinant proteins, which makes them favorite targets for structural studies and rapid binding screening. For example, ligand screening with OBPs could allow the identification of chemical properties for better binding, such as chain length, molecular volume, functional groups, and bond unsaturation. These, combined with new protein structure prediction methods as used in the design of medical drugs and antibodies, such as homology modeling, dynamics simulations, and molecular docking, could place insect OBPs in a favorite position over ORs as targets for the development of control agents in pest management.

Insect ORs seem more specifically tuned to odorants than OBPs. The higher specificity shown by ORs and the chance of activation/inhibition of specific receptors for a given behavior make these proteins as attractive targets to manipulate pest behaviors. The feasibility of the inhibition of either ORx/ORco complex or ORco by antagonists comprises a promising strategy to disrupt insect specific behavior, such as mating via sex pheromone receptors. However, the lack of structural information is the bottleneck in using insect ORs as targets for semiochemical activity predictions. **Tables 1**, **2** summarize the number of OBPs and ORs that have been identified in insect species by transcriptome (i.e., RNA-seq) and genome sequencing. Most of insects studied so far have at least twice ORs than OBPs according to genome studies. Moreover, there is an extensive expansion of ORs in social insects from the Hymenopteran order such as the honey bee A. mellifera with 170 ORs (Weinstock et al., 2006), and the ants Solenopsis invicta and Cerapachys biroi with 400 and 506 ORs, respectively (Wurm et al., 2011; Oxley et al., 2014). Similarly, the OR expansion is also evident in some agricultural pests, such as the red flour beetle T. castaneum with 265 ORs compared to 47 OBPs (Richards et al., 2008). This makes a demanding task for the target OR selection together with the difficulty for the functional expression of transmembrane proteins such as ORs in order to screen a large number of ligands. An approximation of


important properties in both OBPs and ORs are summarized in **Table 3**.

#### FURTHER PERSPECTIVES

The functional characterization of insect ORs as well as their proven roles in insect olfaction have shed lights on the sensitivity and specificity of these insect-specific proteins. These advances will further enhance their feasibility as pest control targets by the understanding of molecular recognition mechanisms and combinatory interactions with OBPs. On the other hand, the current massive effort in the identification and binding characterization of OBPs in several agricultural important insect species will continue and provide more information on their functions in insect physiology. Thus, this review proposes as main advantage for OBPs over ORs, the availability of 3D crystal and NMR structures, which with downstream approaches, such as homology modeling (when necessary), molecular docking and molecular dynamics, would refine the search of bioactive chemicals. This last in complement with ligand affinity measurement will accelerate the study of insect OBPs to be reconsidered as the targets for semiochemical discovery and the tools to design super-ligands in pest control management.

The appearance and development of insecticide resistance in insect pests have led to the intensive research on insect olfaction and the mechanisms that are involved for neural processing. It is well-established that a number of receptors and enzymes in insect CNS are the targets for insecticide resistance development (**Figure 1**). It has been demonstrated

#### REFERENCES


that acetylcholinesterase (AChE) in soluble form provides the resistance to organo-phosphorus and carbamate insecticides, acting as bioscavengers (Lee et al., 2015). Similarly, multiple insecticide resistance mechanisms have been demonstrated in the aphid M. persicae, involving carboxylesterases, sodium channels, γ-aminobutyric acid (GABA) and nAChR (Bass et al., 2014). As important components in insect periphery nerve system and key players in insect behaviors, both insect OBPs and ORs represent alternative targets for the identification of compounds with semiochemical activity (or agonist effect) and tools to design strong antagonists to enhance desired behavioral responses of insect pests and reduce the use of insecticides and subsequent resistance.

#### AUTHOR CONTRIBUTIONS

HV wrote sections about OBP's structure and ORs, developed tables and figure. J-JZ conceived the idea for the review article, wrote the main section, such as introduction, OBPs function and structure as well as OR-OBP comparison.

#### ACKNOWLEDGMENTS

The authors would like to thank FONDECYT 3170433. J-JZ is grateful for the financial support from Northeast Normal University and Jilin University, China for his sabbatical leave to study in China. Rothamsted Research receives a grant with added funding from Biotechnology and Biological Sciences (BBSRC), UK.


Cydia pomonella respond to pheromones and kairomones. Sci. Rep. 7:41105. doi: 10.1038/srep41105


decemlineata by antennal transcriptome analysis. Front. Ecol. Evol. 3:60. doi: 10.3389/fevo.2015.00060


in Tropidothorax elegans Distant (Hemiptera : Lygaeidae). Sci. Rep. 8:7803. doi: 10.1038/s41598-018-26137-6


from Herpetogramma submarginale (Lepidoptera: Crambidae). J. Chem. Ecol. 41, 441–445. doi: 10.1007/s10886-015-0576-8


Asian corn borer, Ostrinia furnacalis (Guenée) (Lepidoptera: Crambidae). PLoS ONE 10:e0128550. doi: 10.1371/journal.pone.0128550


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Venthur and Zhou. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

digital media

of impactful research

article's readership