EDITED BY : Zhengqing Fu, Feng Qu, Shui Wang, Yi Li and Thomas Mitchell PUBLISHED IN : Frontiers in Plant Science and Frontiers in Microbiology

#### 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 978-2-88963-293-0 DOI 10.3389/978-2-88963-293-0

### 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

# ACTIVATION AND SUPPRESSION OF PLANT IMMUNITY

Topic Editors:

Zhengqing Fu, University of South Carolina, United States Feng Qu, The Ohio State University, United States Shui Wang, Shanghai Normal University, China Yi Li, Peking University, China Thomas Mitchell, The Ohio State University, United States

Plants constantly face many kinds of abiotic and biotic stresses. One of the major threats is from many plant fungal, oomycete, viral, bacterial and nematode pathogens. Plant diseases caused by these pathogens reduce crop yield by 10-15% worldwide every year. Throughout the human history, plant diseases are responsible for many famines including the infamous Irish Potato Famine. Besides the negative impact on the yield, the quality of the infected crop will be adversely affected and the toxins produced by plant pathogens pose threat to human health.

During the co-evolution between plants and pathogens, plants developed elegant defense system against pathogen infection and plant pathogens deploy a variety of strategies to suppress plant innate immunity. A deeper understanding the molecular mechanisms on the activation of plant defense in plants and suppression of plant defense by plant pathogens will be crucial to develop effective ways to minimize the detrimental effects from plant diseases on human beings.

This Research Topic aims to increase our understanding on the molecular interactions between plants and pathogens.

Citation: Fu, Z., Qu, F., Wang, S., Li, Y., Mitchell, T., eds. (2019). Activation and Suppression of Plant Immunity. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-293-0

# Table of Contents


Yingbo Liang, Shichun Cui, Xiaoli Tang, Yi Zhang, Dewen Qiu, Hongmei Zeng, Lihua Guo, Jingjing Yuan and Xiufen Yang

*29 The Polycistronic miR166k-166h Positively Regulates Rice Immunity via Post-transcriptional Control of* EIN2

Raquel Salvador-Guirao, Yue-ie Hsing and Blanca San Segundo


Selena Gimenez-Ibanez, Dagmar R. Hann, Jeff H. Chang, Cécile Segonzac, Thomas Boller and John P. Rathjen


You-xin Yang, Chaoqun Wu, Golam J. Ahammed, Caijun Wu, Zemao Yang, Chunpeng Wan and Jinyin Chen

*135 Two Strategies of* Pseudomonas syringae *to Avoid Recognition of the HopQ1 Effector in* Nicotiana *Species*

Patrycja Zembek, Aleksandra Danilecka, Rafał Hoser, Lennart Eschen-Lippold, Marta Benicka, Marta Grech-Baran, Wojciech Rymaszewski, Izabela Barymow-Filoniuk, Karolina Morgiewicz, Jakub Kwiatkowski, Marcin Piechocki, Jaroslaw Poznanski, Justin Lee, Jacek Hennig and Magdalena Krzymowska


Qiulan Huang, Lin Li, Minghui Zheng, Fang Chen, Hai Long, Guangbing Deng, Zhifen Pan, Junjun Liang, Qiao Li, Maoqun Yu and Haili Zhang


Yuankun Yang, Yi Zhang, Beibei Li, Xiufen Yang, Yijie Dong and Dewen Qiu

*252 Exogenous Nicotinamide Adenine Dinucleotide Induces Resistance to Citrus Canker in Citrus*

Fernando M. Alferez, Kayla M. Gerberich, Jian-Liang Li, Yanping Zhang, James H. Graham and Zhonglin Mou


Juying Long, Congfeng Song, Fang Yan, Junhui Zhou, Huanbin Zhou and Bing Yang

# *279 BAKing up to Survive a Battle: Functional Dynamics of BAK1 in Plant Programmed Cell Death*

Xiquan Gao, Xinsen Ruan, Yali Sun, Xiue Wang and Baomin Feng

*293 A Novel G16B09-Like Effector From* Heterodera avenae *Suppresses Plant Defenses and Promotes Parasitism* Shanshan Yang, Yiran Dai, Yongpan Chen, Jun Yang, Dan Yang, Qian Liu and

Heng Jian *308 Comparative Genomics Reveals the High Copy Number Variation of a Retro Transposon in Different* Magnaporthe *Isolates*

Pankaj Kumar Singh, Ajay Kumar Mahato, Priyanka Jain, Rajeev Rathour, Vinay Sharma and Tilak Raj Sharma

# Las15315 Effector Induces Extreme Starch Accumulation and Chlorosis as *Ca.* Liberibacter asiaticus Infection in *Nicotiana benthamiana*

Marco Pitino, Victoria Allen and Yongping Duan\*

*US Horticultural Research Laboratory, USDA-ARS, Fort Pierce, FL, United States*

Huanglongbing (HLB), a destructive plant bacterial disease, severely impedes worldwide citrus production. HLB is associated with a phloem-limited α-proteobacterium, *Candidatus* Liberibacter asiaticus (Las). Las infection causes yellow shoots and blotchy mottle on leaves and is associated with excessive starch accumulation. However, the mechanisms underlying the starch accumulation remain unknown. We previously showed that the Las5315mp effector induced callose deposition and cell death in *Nicotiana benthamiana*. In this study, we demonstrated that Las can experimentally infect *N. benthamiana* via dodder transmission. Furthermore, we revealed another key function of the Las5315 effector by demonstrating that transient expression of the truncated form of the effector, Las15315, induced excessive starch accumulation by 6 fold after 8 dpi in *N. benthamiana* after removal of the chloroplast transit peptide from the Las5315mp. The induction mechanisms of Las15315 in *N. benthamiana* were attributed to the up-regulation of ADP-glucose pyrophosphorylase, granule-bound starch synthase, soluble starch synthase, and starch branching enzyme for increasing starch production, and to the significant down-regulation of the starch degradation enzymes: alpha-glucosidase, alpha-amylase, and glycosyl hydrolase for decreasing starch degradation. This is the first report that Las can infect the model plant *N. benthamiana*. Using this model plant, we demonstrated that the Las15315 effector caused the most prominent HLB symptoms, starch accumulation and chlorosis as Las infection in *N. benthamiana*. Altogether the Las 5315 effector is critical for Las pathogenesis, and therefore, an important target for interference.

Keywords: citrus huanglongbing, liberibacter asiaticus, effector, starch, *nicotiana benthamiana*

# INTRODUCTION

Huanglongbing (HLB) is one of the most complex and severe diseases that affects all commercial citrus species. HLB is associated with an unculturable phloem-limited α-proteobacterium Candidatus Liberibacter (Garnier and Bové, 1983; Jagoueix et al., 1994) and is transmitted by two species of citrus psyllids, Diaphorina citri Kuwayama (Asian citrus psyllid: ACP) and Trioza erytreae (del Guercio) (African citrus psyllid) (Bove, 2006). The most noticeable symptoms of HLB are yellow shoots and blotchy mottle, a random pattern of yellowing on leaves. HLB infection eventually renders infected plants completely unproductive and causes death within 3–5 years.

#### *Edited by:*

*Zhengqing Fu, University of South Carolina, United States*

#### *Reviewed by:*

*Youfu Zhao, University of Illinois at Urbana–Champaign, United States Alberto A. Iglesias, National University of the Littoral, Argentina*

> *\*Correspondence: Yongping Duan yongping.duan@aus.usda.gov*

#### *Specialty section:*

*This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science*

*Received: 07 December 2017 Accepted: 22 January 2018 Published: 07 February 2018*

#### *Citation:*

*Pitino M, Allen V and Duan Y (2018) Las*1*5315 Effector Induces Extreme Starch Accumulation and Chlorosis as Ca. Liberibacter asiaticus Infection in Nicotiana benthamiana. Front. Plant Sci. 9:113. doi: 10.3389/fpls.2018.00113*

**6**

Another common HLB characteristic is callose deposition and extremely high starch accumulation in leaves (Yelenosky and Guy, 1977), which is believed to result in disintegration of the chloroplast thylakoid system and thus the observed chlorosis in infected tissues (Schaffer et al., 1986). There is no known cure for HLB and no commercially viable, effective treatment.

Ca. Liberibacter asiaticus (Las) is a bacterial plant pathogen with a dual host life cycle dependent on sap-feeding insects for transmission to plants. Pathogens may also manipulate the development of their psyllid hosts to improve growth, enhance colonization, and increase transmission (Sugio et al., 2011; Win et al., 2012; Petre et al., 2014). Protein secretion is a key virulence mechanism of pathogenic and symbiotic bacteria, and the presence of genes coding for a complete Sec-translocon in Las (Duan et al., 2009) suggests an important role for the secretion of proteins including virulence factors. Moreover, Las seems to possess effectors with the ability to manipulate plant physiology (Pitino et al., 2016).

We previously identified 16 Las candidate effector proteins (Pitino et al., 2016) using a fast-forward screen pipeline for Las candidate effectors containing signal peptide (SP), which is usually required to achieve final folding and localization of exported proteins outside the prokaryotic cell (Pugsley, 1993; Tuteja, 2005; Jehl et al., 2011; Palmer and Berks, 2012). We found that one particular effector (Las5315) caused cell death when transiently expressed in Nicotiana benthamiana without the SP which is normally cleaved by signal peptidases (Las5315mp). This effector induces callose deposition and targets the chloroplast. Interestingly, the excessive callose formation in the phloem seems to be linked to a plant defense reaction to Las bacterium. Las infection processes can lead to perturbation of normal carbon partitioning, excessive starch accumulation in plastids, chloroplast disruption, increase of H2O2, insufficient root development, reduced size and aberrant fruit development, and slow growth (Bove, 2006; Albrecht and Bowman, 2008; Folimonova and Achor, 2010; Gottwald, 2010; Rosales and Burns, 2011; Koh et al., 2012; Aritua et al., 2013; Martinelli et al., 2013).

Starch can be classified into two types: transitory starch, which occurs in photosynthetic organs and is also called leaf starch, and reserve starch, which occurs in storage organs (Wang et al., 2013). The starch accumulated during the day is degraded during the subsequent night, providing a continued supply of carbohydrates in the absence of photosynthesis (Gibon et al., 2009). In higher plants the production of starch is orchestrated by chloroplastlocalized biosynthetic enzymes affected by endogenous hormone levels that regulate genes involved in the starch metabolism pathway (Liu et al., 2015). Viral factors can affect chloroplast ultrastructure, symptom development and starch accumulation (Allan et al., 2001; Zhao et al., 2016); for example, TMV movement protein induces starch accumulation in transgenic plants (Olesinski et al., 1996). In addition, exudates from fungal spores induced starch accumulation in wild-type Lotus japonicus roots (Gutjahr et al., 2009). Sinorhizobium meliloti flavodoxin-overexpressing rhizobia also led to high starch accumulation in nodules (Redondo et al., 2009). Leaves exposed to volatiles emitted by Alternaria alternata revealed that starch overaccumulation was accompanied by up-regulation of starch synthesis genes (Ezquer et al., 2010); emission of Microbial volatile organic compounds (MVOCs) strongly promoted starch accumulation in leaves of both mono-and dicotyledonous plants (Kanchiswamy et al., 2015). Ca. Liberibacter asiaticus, the presumptive causal agent of HLB, is another organism that induces excessive starch accumulation (Yelenosky and Guy, 1977), yet the mechanism remains unclear.

Excessive starch accumulation is observed in the phloem sieve elements of Las infected citrus (Etxeberria et al., 2009; Folimonova and Achor, 2010), which is not common in higher plants. In fact, starch induction by Las is so extreme that the starch content of citrus leaves can be used for diagnosis of HLB (Whitaker et al., 2014). Originally, starch accumulation was thought to result from necrotic phloem blockages which disrupt the natural carbon source/sink balance of the plant tissues (Schneider, 1968), however more recently, it has been shown that granule-bound starch synthase is upregulated well before plants become symptomatic (Nwugo et al., 2013a). This apparent conflict underscores the need for further research into the host-pathogen dynamics of starch accumulation. Transcriptomic and proteomic studies have revealed that several enzymes related to starch synthesis, degradation, and transport are up or downregulated in response to Las infection (Martinelli et al., 2012; Nwugo et al., 2013a). These include large subunit ADP-glucose pyrophosphorylase (Kim et al., 2009; Martinelli et al., 2012; Aritua et al., 2013), various starch synthase isoforms, granule-bound starch synthase, glucanotransferase, maltose excess protein, betaamylases, glucoamylase, and glycosyltransferase (Fan et al., 2010; Martinelli et al., 2012; Aritua et al., 2013; Nwugo et al., 2013a; Xu et al., 2015). Additionally, increased activity of cell-wall invertase, a protein involved in carbohydrate transport and linked to starch accumulation, has been observed in Las-infected leaves (Fan et al., 2010). The molecular mechanism by which Las infection induces these changes remains unknown.

In this study, we revealed how Las15315 induced extreme starch accumulation and severe chlorosis, typical HLB symptoms as seen both in Las-infected N. benthamiana and citrus. We also confirmed N. benthamiana is an experimental host plant of Ca. Liberibacter asiaticus.

# MATERIALS AND METHODS

# Las15315

Downstream of the N-terminal SP sequence of Las5315 effector protein, we previously found a chloroplast transit peptide domain 56 a.a. in length using ChloroP 1.1 (Pitino et al., 2016). In this work we generated a new construct with deletion of the chloroplast transit peptide, referred to as Las15315. The sequence was prepared by amplification of corresponding fragments using polymerase chain reaction (PCR). Primers were tailed with attB1 and attB2 sites. The sequences were amplified using Las infected periwinkle DNA and HIFI PCR master mix (Clontech). The PCR fragments were first cloned into pDONR/Zeo by Gateway <sup>R</sup> BP Clonase <sup>R</sup> II (Invitrogen) and then subcloned into the Gateway destination vector ImpGWB405 (Nakagawa et al., 2007).

# Transient Expression of *Las*1*5315* in *N. benthamiana* and Localization

Agrobacterium tumefaciens GV3101 transformant cells carrying Las5315, Las5315mp, and Las15315 in vector ImpGWB405 (Nakagawa et al., 2007) were cultured overnight in LB medium with 50 µg ml−<sup>1</sup> of rifampicin and 100 µg ml−<sup>1</sup> spectinomycin and resuspended in 10 mM MgCl2. The culture was diluted to an optical density at 600 nm of 0.5, and acetosyringone (final concentration of 100µM) was added. For each construct, we infiltrated three leaves of three 4 week-old N. benthamiana plants with the A. tumefaciens suspension. The agro-infiltrated leaves were analyzed for protein localization at 3 dpi under a microscope (Olympus BX51-P) equipped with a UV light source. Agroinfiltrated plants were kept in a greenhouse for the duration of the experiment and observed for phenotypic changes.

# Starch Staining and Quantitation

Starch was visualized by iodine staining reagent (Sigma). Four dpi detached leaves or Las infected leaves were boiled in 95% ethanol to remove leaf pigments thoroughly and then washed twice with deionized water. Rehydrated leaves were stained for 10 min in 5% iodine solution (5% [w/v] I<sup>2</sup> and 10% [w/v] KI) and incubated in water to allow fading until a clear background was obtained. Iodine staining was repeated at least three times for each construct.

Enzymatic measurement of starch in leaves was performed using Starch Fluorometric Assay Kit (BioVision, California, US), 1 leaf disc was collected using cork borer number 6 at 2, 4, 6, and 8 dpi from tissue agro-infiltrated with the corresponding constructs. Leaf discs were pulverized using liquid nitrogen and then starch was extracted and digested following the manufacturer's instructions. After the digestion the samples were measured at Ex/Em wavelength 535/587 nm by the LUMIstar microplate reader (BMG Labtech). Three biological replicates were used for the analyses.

# Protoplast Extraction and Starch Iodine Stain

Detached leaves were rinsed with distilled water and blotted dry. Leaves were sliced with a razor blade into 1- to 2-mm strips and placed in 20 ml of filter-sterilized Enzyme Solution (0.5 M sucrose, 10 mM MES-KOH [pH 5.7], 20 mM CaCl2, 40 mM KCl, 1% Cellulase R-10, 1% Macerozyme R-10). Plant tissue was shaken at 35 rpm in the dark at room temperature for 16–18 h. Protoplasts were collected by sieving through 8 layers of cheesecloth and centrifuged for 7 min at 100 g. Supernatant was removed, and the pellet was used for iodine staining and microscope analysis.

# Immunoblot Detection of Las15315 in *N. benthamiana*

Two leaf discs were collected from the infiltrated part of N. benthamiana leaves at 4 dpi using a 6 mm cork borer; leaf discs were ground in liquid nitrogen and suspended in extraction buffer plus protease inhibitor cocktail as per the manufacturer's protocol (denaturing). Proteins were separated by SDS-PAGE 4–12% acrylamide gel and electro-blotted to a nitrocellulose membrane using an iBlot blotting system (Life Technologies). The membrane was blocked with TBS-T containing 5% skimmed milk powder for 1 h at room temperature.

For immunoblot analysis, antibodies for Las15315 were generated against a peptide sequence of Las15315 (CISRTRIDSSPPPHG) (GenScript) and then HRP conjugated using Lighting-Link HRP kit (Innova Bioscience). The membrane was first incubated with primary antibody for 1 h with mild agitation and then washed 3 × 10 min; protein band was detected and analyzed using WestenSure Chemiluminescence substrate and the digital imaging system C-DiGit <sup>R</sup> Blot Scanner (LI-COR).

# Chlorophyll Assay

Leaf tissues were inspected at 4 dpi for the development of localized chlorosis around the point of infiltration. Chlorophyll assays were carried out using leaf discs excised with a cork borer number 6 from three individual plants, two leaf discs per leaf, for a total of six leaf discs. Each sample with two leaf discs was incubated for 24 h in 2 mL 95% ethanol at 4◦C in the dark. Chlorophyll concentrations were determined using chlorophyll equations based on the formula: 13.70 · A665–5.76 · A<sup>649</sup> used − 7.60 · A<sup>665</sup> + 25.8 · A<sup>649</sup> (Ritchie, 2006).

# RNA Extraction and cDNA Synthesis

Total RNA was extracted from 3 separate agroinfiltrated leaves using 3 N. benthamiana plants at the end of the photoperiod (day) and immediately before onset of the photoperiod (night). Sample leaves were quickly frozen in liquid nitrogen and ground to a powder using an autoclaved mortar and pestle. Total RNA was extracted using TRIzol Reagent (Invitrogen). Total RNA concentration and purity were determined from the ratio of absorbance readings at 260 and 280 nm, using a Nanodrop 1000 spectrophotometer (Thermo Scientific). cDNA synthesis was performed with poly-T primers using the M-MLV reverse transcriptase system (Promega) according to the manufacturer's instructions.

# Gene Expression Analysis

Primers were purchased from IDT and used in a 15 µL reaction with 7.5 µL of 2 × FAST SYBR Green Master Mix (Quanta Bio) reagent and 2 µL of DNA template. The following standard thermal profile was used for all amplifications: 95◦C for 5 min, followed by 40 cycles of 95◦C for 3 s and 62◦C for 30 s. Primer sequences are listed in **Table S1**. Reactions were performed in triplicate and normalized to EF1 expression; the 2−11Ct method was used to calculate relative expression as previously described (Pitino et al., 2015).

# Transmission Procedure and Genomic DNA Extraction for Las Detection

Dodder (Cuscuta pentagona) germinated on Las-infected Duncan citrus plants was used to inoculate N. benthamiana with Las bacteria. Connections were established on 4 week-old N. benthamiana plants and maintained for 4 weeks, after which the dodder was removed. Control plants were connected with uninfected dodder for the same time period. Plants were kept in a greenhouse.

To determine Las bacterial titer, symptomatic and healthy leaves were collected from N. benthamiana 1 and 2 and 3 months after dodder inoculation and tested using primers HLBasf, HLBr and probe HLBp targeting the 16S rDNA of Las (Li et al., 2006) as described previously (Pitino et al., 2017). Starch staining was also carried out on controls and infected N. benthamiana immediately before onset of the photoperiod (night) as described above.

# RESULTS

# Expression of *Las*1*5315* Results in a Dramatic Visual Phenotype and Accumulates in Vescicle Structures

The Las5315 effector is composed of a SP at the N-terminal, followed by a chloroplast transit peptide and the remainder of the protein (**Figure 1**). In our previous study we observed robust phenotypes upon expressing Las5315 with the SP deleted (Las5315mp). Like the secretory SP, it is likely that the chloroplast transit peptide is also cleaved in vivo once Las5315mp has localized to the chloroplast. Transit peptide directs post-translational localization and is removed upon arrival in the organelle (Bruce, 2000; Shen et al., 2017). Therefore, in order to further characterize Las5315, we created a construct lacking both the secretory peptide and the chloroplast transit peptide (Las15315). Las15315 was transiently expressed in N. benthamiana using Agrobacterium-mediated expression. The protein was detectable by western blot 2 days post-infiltration (dpi) (**Figure S1**), yielding a single band with a molecular mass of slightly less than 37 kilodaltons (kDa). This was consistent with the expected combined molecular mass of Las15315 (8 kDa) and GFP (27 kDa). Three leaves were agroinfiltrated with the constructs (**Figure 2A**), and 4 days after agroinfiltration, the treated areas exhibited chlorosis on the adaxial side of the leaf, and a noticeably water-soaking appearance on the abaxial side (**Figure 2B**, **Figures S2A**,**S3A**). In contrast, cell death was observed in the zone of infiltration with Las5315mp, as in our previous study (Pitino et al., 2016), (**Figure 2B**, **Figure S3B**). Agroinfiltration with a control vector containing only GFP did not cause any visible phenotype (**Figure 2B**, **Figure S3C**).

In order to establish the localization pattern of Las15315, we imaged the leaf epidermal layer of the infiltrated zones and compared to zones infiltrated with the control GFP vector. While the control showed diffuse fluorescence (**Figures S4A,B**), GFP-tagged Las15315 accumulated in small and large vescicles throughout the cytosol (**Figures S4C–F**). Interestingly, these vescicles often contained one or more chloroplasts, in which fluorescence was absent. This suggests that Las15315 protein does not enter chloroplasts, but rather envelops them.

# Expression of *Las*1*5315* Causes Visible Starch Accumulation Revealed by Iodine Staining

A major pathology associated with HLB is starch accumulation. Therefore, we investigated whether transient Las15315 expression in N. benthamiana leaves induced starch accumulation by using iodine staining for starch visualization. We observed massive starch accumulation at 4 days postinfiltration (dpi) (**Figure 3A**, **Figure S3B**). Agroinfiltration with a control vector containing GFP did not increase leaf starch content. In order to better visualize starch granules, protoplasts were isolated and similarly stained. The zone of infiltration for the control strain showed green chloroplasts in the protoplast, with no to minimal starch visible (**Figure 3B**). In contrast, the zone of infiltration for the Las15315 strain showed dark coloration and visible starch granules (**Figure 3C**).

# Quantitation of Starch Accumulation Induced by *Las*1*5315* Expression

To quantify starch accumulation, a fluorometric assay was carried out at 2, 4, 6, and 8 dpi, both at the end (day; **Figure 4A**) and beginning (night; **Figure 4B**) of the photoperiod. The relative soluble starch content was greater in tissue expressing Las15315 than in tissue expressing GFP at all time points except for 2 dpi daytime. Starch concentrations increased steadily in Las15315-expressing tissue throughout the time course, but stayed relatively constant in tissue expressing GFP, resulting in a 3-fold difference during the day and 5-fold difference during the night at 8 dpi. This led us to hypothesize that in tissue expressing Las15315, starch synthesis enzymes were upregulated, starch degradation enzymes were downregulated, or some combination of the two.

# Las15315 Alters Starch-Related Gene Expression

To investigate whether the starch accumulation observed in Las15315-expressing tissue was due to overproduction of starch, failure to degrade starch, or combination of both, we quantified transcript levels of starch synthesis- and degradation-related genes. Since starch is accumulated during the photoperiod we quantified transcript levels of starch biosynthesis genes from samples taken near the end of the photoperiod. Since starch is broken down to serve as a source of glucose when photosynthesis is inactive, we quantified transcript levels of starch degradation genes from samples taken shortly before

FIGURE 1 | Sequence and expression of LasDelta5315 in *N. benthamiana* adapted from Pitino et al. (2016). Amino acid sequence of full-length Las5315. Signal peptide sequence shown in red; chloroplast transit peptide shown in green. Sequence of Las15315 shown in black.

onset of the photoperiod by qRT-PCR. Several starch synthesis enzymes, including ADP-glucose pyrophosphorylase, granulebound starch synthase, soluble starch synthase, and starch branching enzyme (**Figure 5A**) showed increased transcript levels in Las15315-expressing tissue relative to controls (4, 3, and up to 10 fold difference, respectively). In contrast, transcript levels of the starch degradation enzymes alpha-glucosidase, alpha-amylase, and glycosyl hydrolase (Chiba, 1997) were down-regulated (4, 11, and 4 fold difference, respectively) (**Figure 5B**). These data suggest that Las15315 expression causes starch accumulation both by increasing starch production and decreasing starch degradation.

# *Las*1*5315* Expression Reduces Chlorophyll Concentration

Starch over-accumulation and chlorosis are two of the most prominent symptoms of HLB. Chlorosis results from breakdown of the thylakoid membrane, which is thought to result from starch accumulation within the chloroplast (Schaffer et al., 1986). After observing massive starch accumulation induced by Las15315, we further tested whether this phenotype was correlated with a reduction in chlorophyll, as would be expected if starch granules cause breakdown of chloroplast structures. Consistent with this hypothesis, we observed reduced concentrations of both chlorophyll a (**Figure 6A**) and chlorophyll b (**Figure 6B**) at 4 dpi in tissues expressing Las15315 compared to tissues expressing a control vector (p = 0.0035 and p = 0.0089, respectively). This result suggests that starch accumulation may play a role in chloroplast breakdown, and therefore be associated with chlorosis.

# *N. benthamiana* Is a Novel Host for *Ca.* Liberibacter asiaticus

While we revealed the expression of Las15315 induced HLB-like symptoms in N. benthamiana, we confirmed that N. benthamiana plants were Las infected by qPCR, using primers specific for the

FIGURE 3 | Starch accumulation in leaves expressing Las15315 revealed by iodine staining. (A) Whole leaf infiltrated on the left expressing control vector, and on the right expressing for Las15315. The Las1-expressing zone shows dark coloration from the iodine stain, indicating the presence of starch. (B) Protoplasts extracted from leaves infiltrated with a control vector (left) or a vector expressing Las15315. Control appears green with little to no dark after iodine staining. (C) Las15315 expressing tissue is stained to a dark shade by the iodine, and starch granules are visible (white arrows). Scale bar indicates 10µM.

(black). Samples taken after 12 h of dark (nighttime). Asterisks indicate significant differences between control and Las15315; bars indicate standard error, the results were analyzed for significant difference with Student's *t*-test.

Las 16S sequence (Li et al., 2006) (**Figure 7A**) and previously identified Las effector genes (data not shown) (Pitino et al., 2016). We demonstrated that N. benthamiana can be infected by Las via dodder transmission (**Figure 7B**) 2 months after inoculation, plants showed yellowing symptoms as expected (**Figures 7C,E**). Branches developed symptoms progressively with an uneven pattern of distribution. Like transient expression of Las15315, Las infection elicited massive starch accumulation in infected tissues in N. benthamiana (**Figure 7E**).

# DISCUSSION

HLB has devastated the Florida citrus industry, and continues to threaten citrus crops worldwide. Massive starch accumulation is thought to be the main cause of HLB symptoms. However, the molecular mechanism by which the bacterium induces starch accumulation in host plants remains unknown. Ectopic expression of Las proteins provides a convenient alternative method to study host-pathogen interactions. We previously

identified Las5315mp effector, which induced cell death and callose deposition in N. benthamiana after transient expression (Pitino et al., 2016). Here we show that transient expression of Las15315 (**Figure 1**) causes massive starch accumulation in N. benthamiana leaves (**Figures 3**, **4**), similar to that seen in Las-infected citrus and N. benthamiana plants. This is the first study to show a direct causal relationship between expression of a Las effector and starch accumulation. We observed that Las15315:GFP was localized inside vesicles structures within the cytoplasm, many of which contained chloroplast organelles (**Figures S4C–H**).

Expression of Las15315 caused a striking phenotype, with chlorosis on the adaxial side and a water-soaking appearance on the abaxial side (**Figure 2B**, **Figures S2A**,**S3B**), along with reduced levels of both chlorophyll a and chlorophyll b (**Figure 6**). The chlorotic phenotype is characteristic of Las infection/HLB, but the water-soaking appearance on the abaxial side of the leaves expressing Las15315 is not similar to any known HLB symptom. However, since Las affects pathways involved in source-sink communication, including sucrose and starch metabolism (Martinelli et al., 2012), and increases potassium concentration (Nwugo et al., 2013b), it is conceivable that this Las effector could indirectly change the osmotic balance in the plant tissues. We speculate that hexoses may accumulate, leading to an increase of osmotic potential and causing water-soaking appearance of the abaxial side.

It is worth noting that rather than being a secondary result of phloem disruption, expression of this Las effector directly causes starch accumulation (**Figures 3**, **4**). After expressing Las15315, N. benthamiana tissue began to show the starch accumulation and chlorosis characteristic of HLB symptoms. These results suggest that Las5315 is a critical effector of Las, and may be directly associated with HLB symptoms. Interestingly, although starch content was consistently higher in tissue expressing Las15315 than in tissue expressing a control vector, greater differences were observed during the dark portion of the circadian cycle than during the photoperiod, indicating an altered regulation of starch degradation. Given that transitory starch is synthesized during the day and broken down at night, we hypothesize that failure to adequately degrade starch contributes to the observed starch accumulation phenotype more than increased synthesis.

HLB research has long been hindered by the inability to culture Las in vitro and by prohibitively slow citrus growth and Las transmission protocols. The use of periwinkle (Catharanthus roseus) as an artificial host for the bacterium has improved the situation, but genetic tools for use in periwinkle have not yet been developed. In contrast, a wide variety of genetic tools including viral expression vectors, agroinfiltration and extensive studies on plant disease resistance already are available for N. benthamiana (Van Ooijen et al., 2008a,b; Win et al., 2011; Du et al., 2014; El Kasmi et al., 2017). This is the first report of Las infection of the genetically tractable N. benthamiana. To confirm similar pathology of Las infection in N. benthamiana as in citrus, we analyzed starch accumulation in infected and control N. benthamiana plants following the normal night starch degradation phase. Upon iodine staining, infected leaves showed the same dark-colored phenotype as expression of the Las15315 effector alone, indicating that Las indeed prevents starch degradation during the night, causing starch accumulation. Interestingly, when a single infected N. benthamiana leaf was imaged before and after iodine staining (**Figures 7D,E**), the darker starch stained areas co-localized with the more chlorotic parts of the leaf. In contrast, the lighter, unstained areas colocalized with greener parts of the leaf. These observations are consistent with the hypothesis that starch accumulation leads to breakdown of thylakoid and chloroplast integrity, causing chlorosis, although further testing will be necessary to prove a causal relationship.

Las infected citrus plants accumulate large amounts of starch in the aerial tree parts (Etxeberria et al., 2009), vascular parenchyma and phloem elements (Folimonova and Achor, 2010). Several reports have shown that genes coding for ADP-glucose pyrophosphorylase, starch synthase, granulebound starch synthase and starch debranching enzyme were up regulated and contributed to accumulation of starch in HLB-affected leaves (Ballicora et al., 2004; Kim et al., 2009; Aritua et al., 2013; Nwugo et al., 2013a), while other genes coding for key starch degradation enzymes, such as beta-amylase, glycosyl hydrolase, alpha-glucosidase, and sucrose-phosphate synthase were down-regulated (Albrecht and Bowman, 2008; Kim et al., 2009). It is likely that increased expression of starch synthesis genes and decreased expression of starch breakdown genes work synergistically to cause starch accumulation in infected tissue. Excessive starch buildup eventually causes disintegration of the chloroplast thylakoid, contributing to chlorosis, a characteristic symptom of HLB (**Figure 8**). To further investigate whether altered expression of starch degradation and synthesis genes was responsible for the starch accumulation observed in plants infiltrated with Las15315, we analyzed starch-related genes that were affected by Las in citrus plants, both in normal diurnal and night conditions (**Figure 5**). Consistent with previous reports of altered transcription in Las-infected tissue, tissue expressing Las15315 showed increased expression of starch synthesis genes and decreased expression of starch degradation genes. Importantly, our data

staining.

suggest that Las5315 effector protein may be responsible for the dysregulation of starch-related enzyme transcripts, causing starch accumulation and the consequent chlorosis phenotype.

In conclusion, here we demonstrated that the Las15315 effector induced excessive starch accumulation in plant cells by modulating the transitory starch metabolism, which led to dysfunctional chloroplasts. Evaluation of starch content in Lasinfected N. benthamiana demonstrated a failure to breakdown starch during the night. This is the first study to show a causal relationship between Las effector expression, starch accumulation, and chlorosis, indicating a potential target for novel HLB therapies. It is also the first report of Las transmission to N. benthamiana, which is a suitable and convenient host for studying Las due to its status as a ubiquitous model plant. Our establishment of a N. benthamiana Las-infection model, as well as our identification of a protein target for HLB therapies, represent important progress toward the understanding and elimination of HLB.

# AUTHOR CONTRIBUTIONS

MP and YD designed the research. MP and VA performed data collection and analysis. MP, VA, and YD interpreted the data and prepared the manuscript.

# ACKNOWLEDGMENTS

We thank Carrie Vanderspool for the plants used in this study, Christina Latza for excellent technical assistance and Viviana MacKade for her critical review. Funding was provided by the U.S. Department of Agriculture. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018. 00113/full#supplementary-material

Figure S1 | Western blot of agroinfiltrated tissues using an antibody against Las15315. Molecular mass markers (1), leaf samples expressing Las15315::GFP (2), and leaf samples expressing GFP (3).

# REFERENCES


Figure S2 | Visual phenotypes of leaves expressing *Las*1*5315.* (A) Las15315 infiltration zone shows a chlorotic, Yellowing phenotype on the adaxial side and a water-soaking phenotype on the abaxial side. (B) Las15315 expressing tissue is stained to a dark shade by the iodine.

Figure S3 | Characteristic visible phenotypes on leaves expressing Las5315 mp and Las15315, respectively. (A) Leaves expressing Las5315 mp show cell death as evidenced by progressive fading on the adaxial side. (B) Leaves expressing Las15315 show a chlorotic, yellowing phenotype on the adaxial side and a water-soaking phenotype on the abaxial side. (C) Control leaves agroinfiltrated with a GFP vector remain healthy and green. dpi = days post-infiltration. Representative images are shown.

Figure S4 | Localization of Las15315:GFP in *Nicotiana benthamiana.* Epi-fluorescence and bright field micrograph of GFP and Las15315 proteins. (A,B) GFP only; (C,D) Localization of Las15315:GFP (green) and chloroplasts (red); (E,F) enlarged image showing in detail Las15315:GFP localization in vesicles and chloroplasts. Scale bars indicate 20µm.

Table S1 | Primer sequences used for qRT-PCR.

with 'Candidatus Liberibacter asiaticus'. Plant Pathol. 59, 1037–1043. doi: 10.1111/j.1365-3059.2010.02328.x


carbohydrate metabolism and endogenous hormone crosstalk. Biotechnol. Biofuels 8:64. doi: 10.1186/s13068-015-0245-8


**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 Pitino, Allen and Duan. 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.

# An Asparagine-Rich Protein Nbnrp1 Modulate Verticillium dahliae Protein PevD1-Induced Cell Death and Disease Resistance in Nicotiana benthamiana

Yingbo Liang, Shichun Cui, Xiaoli Tang, Yi Zhang, Dewen Qiu, Hongmei Zeng, Lihua Guo, Jingjing Yuan and Xiufen Yang\*

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

### Edited by:

Yi Li, Peking University, China

### Reviewed by:

Yule Liu, Tsinghua University, China Hui-Shan Guo, Institute of Microbiology (CAS), China

> \*Correspondence: Xiufen Yang yangxiufen@caas.cn

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 02 November 2017 Accepted: 22 February 2018 Published: 07 March 2018

#### Citation:

Liang Y, Cui S, Tang X, Zhang Y, Qiu D, Zeng H, Guo L, Yuan J and Yang X (2018) An Asparagine-Rich Protein Nbnrp1 Modulate Verticillium dahliae Protein PevD1-Induced Cell Death and Disease Resistance in Nicotiana benthamiana. Front. Plant Sci. 9:303. doi: 10.3389/fpls.2018.00303 PevD1 is a fungal protein secreted by Verticillium dahliae. Our previous researches showed that this protein could induce hypersensitive responses-like necrosis and systemic acquired resistance (SAR) in cotton and tobacco. To understand immune activation mechanisms whereby PevD1 elicits defense response, the yeast two-hybrid (Y2H) assay was performed to explore interacting protein of PevD1 in Arabidopsis thaliana, and a partner AtNRP (At5g42050) was identified. Here, AtNRP homolog in Nicotiana benthamiana was identified and designated as Nbnrp1. The Nbnrp1 could interact with PevD1 via Y2H and bimolecular fluorescence complementation (BiFC) analyses. Moreover, truncated protein binding assays demonstrated that the C-terminal 132 amino acid (development and cell death, DCD domain) of Nbnrp1 is required for PevD1-Nbnrp1 interaction. To further investigate the roles of Nbnrp1 in PevD1-induced defense response, Nbnrp1-overexpressing and Nbnrp1-silence transgenic plants were generated. The overexpression of Nbnrp1 conferred enhancement of PevD1-induced necrosis activity and disease resistance against tobacco mosaic virus (TMV), bacterial pathogen Pseudomonas syringae pv. tabaci and fungal pathogen V. dahliae. By contrast, Nbnrp1-silence lines displayed attenuated defense response compared with the wild-type. It is the first report that an asparagine-rich protein Nbnrp1 positively regulated V. dahliae secretory protein PevD1-induced cell death response and disease resistance in N. benthamiana.

Keywords: Verticillium dahliae, PevD1, Nbnrp1, protein–protein interaction, defense response, disease resistance

# INTRODUCTION

Plants have evolved a sophisticated innate immune system to detect and ward off potential dangers in the course of plant-pathogens co-evolution (Dangl and Jones, 2001; Akira et al., 2006; Chisholm et al., 2006; Boller and Felix, 2009). Plant immune system follows two major strategies. The recognition of conserved pathogen or microbe-associated molecular patterns (P/MAMPs)

known as microbe signature is the primary layer of the plant immune system, called as PAMP-triggered immunity (PTI). For infecting host plant successfully, pathogens suppress PTI by employing effectors that target or interfere with host defense signaling components. Afterward, plants generated specific recognition system to perceive such effectors, leading to effector-triggered immunity (ETI) (Dodds and Rathjen, 2010). ETI is generally characterized stronger and more sustained immune responses than PTI, whereas PTI represent more durable and broad-spectrum resistance (Katagiri and Tsuda, 2010; Tsuda and Katagiri, 2010). In facts, it is well documented that the distinction between PTI and ETI, between PAMPs and effectors, even between R protein and defense protein is imprecise (Thomma et al., 2011). PAMPs and effectors trigger similar defense responses and converge in common downstream immune signal cascade including hypersensitive response (HR)-like cell death, oxidative burst, activation of kinase signaling cascades, expression of defenserelated gene and phytoalexin accumulation, etc. which lead to systemic acquired resistance (SAR) that confers broad spectrum pathogen resistance to bacteria, fungi, and virus disease in plants.

Hypersensitive response is commonly considered as an example of programmed cell death (PCD) and a typical rapid defense response induced by microbe effectors, which alerts neighboring cells and causes rapid plant cell death that lead to restriction of pathogen further spread (Pennell and Lamb, 1997; Heath, 2000; Desaki et al., 2006; Mur et al., 2008). Plant cell death response can be triggered by different mechanisms, in which HR induced by microbe effectors is typically incidental to resistance (Mur et al., 2008; Tsuda and Katagiri, 2010; Zhang et al., 2017a). Given these studies, it's essential to enrich our understanding of how can microbe effectors activate downstream signal component in HR.

We generally accept that plant directly or indirectly sense elicitor/effectors through pattern recognition receptors (PRRs) localized in the plasma membrane or cytoplasm. Pathogenic fungi secrete hundreds of effectors/PAMPs during the infection process, which modulate the plant-fungus interaction through targeting plant molecular. Identification of elicitor target in plant is vital for understanding the elicitor-activating plant defense signaling cascade. Although many elicitors/effectors have been characterized, the described target partner in plants is limited.

PevD1, a proteinaceous elicitor secreted by Verticillium dahliae, could induce typical HR-like necrosis and apoptosisrelated events in tobacco (Wang et al., 2011) as well as improve SAR against tobacco mosaic virus (TMV) and cotton Verticillium Wilt (Bu et al., 2014). The infiltration of PevD1 elevated the expression level of SAR-related genes of PR1-a, PR1-b, NPR1 and defense-related genes of PAL, C4H1, 4CL, which involve in phenylpropanoid metabolism pathway, and also systematically elicit H2O<sup>2</sup> production, NO generation, lignin deposition, and vessel reinforcement in cotton plants (Bu et al., 2014). PevD1 transgenic Arabidopsis thaliana lines have been demonstrated to improve disease resistance against Botrytis cinerea and P.s. pv. tomato DC3000 compared to wild type (Liu et al., 2016). These data indicate that PevD1 could induce broad spectrum pathogen resistance in host plants. Nevertheless, the underlying molecular mechanism involving in PevD1-induced HR and disease resistance in plants needs further research.

To investigate how PevD1 modulates immunity response in plants, we have identified an asparagine-rich protein (AtNRP) in A. thaliana as the interacting partner of PevD1 via Y2H system (Zhou et al., 2017). NRPs (Asparagine-rich proteins or N-rich proteins) were originally characterized for their sequence with high amount of the amino acid asparagine in N-terminal. These proteins also contain a DCD domain in C-terminal, which is conserved and involved in development and cell death in plants. The AtNRP could interact with cryptochrome 2 (CRY2), leading to accumulation of CRY2 in the cytoplasmic. Further investigation suggested PevD1 indirectly activated cryptochrome 2 by antagonizing NRP functions, resulting in early flowering upon V. dahliae infection (Ludwig and Tenhaken, 2001; Zhou et al., 2017).

In this study, a Nicotiana benthamiana NRP protein designated as Nbnrp1 was homology cloned based on conserved PevD1 partner AtNRP. The interaction of Nbnrp1 and PevD1 was confirmed via Y2H and bimolecular fluorescence complementation (BIFC) assays. Further exploration revealed the C-terminal DCD domain of Nbnrp1 was required for the Nbnrp1-PevD1 interaction. Many evidences showed that asparagine-rich protein was involved in response to different stresses such as oxidative stress, salt stress, pathogen infection and transduce cell death signal through either the endoplasmic reticulum (ER) or osmotic stress (Costa et al., 2008; Reis et al., 2011). To investigate the role of Nbnrp1 in PevD1 induced cell death and disease resistance, we generated Nbnrp1 overexpressing and Nbnrp1-silence transgenic N. benthamiana plants. TMV, Pseudomonas syringae pv. tabaci and V. dahliae were used to compare disease resistance between transgenic plants and wild type plants. The results indicated that Nbnrp1-overexpressing plants displayed enhanced necrosis activity and disease resistance, whereas Nbnrp1-silence plants showed impaired necrosis activity and disease resistance. Accordingly, we proposed that Nbnrp1 is a positive regulator involved in PevD1-elicited cell death and disease resistance in N. benthamiana.

# MATERIALS AND METHODS

# Yeast, Bacterial, and Plant Culture

The prokaryotic expression vector pGEX-6P-2, the plant expression vector pBI121 and pCAMBIA2300, RNAi vector pRNAi1017 were taken from laboratory stocks. TMV-GFP was a gift from Yule Liu (Tsinghua University, Beijing, China). Y2H gold Yeast (Saccharomyces cerevisiae) was grown in PDA or SD medium at 28◦C. Escherichia coli Trans1-T1 and BL21 were purchased from TransGen Biotech (China) and were grown in LB at 37◦C. Agrobacterium tumefaciens GV3101 was grown in LB medium at 28◦C. P.s. pv. tabaci was a gift from Jun Liu (Institute of Microbiology, Chinese Academy of Sciences, Beijing, China), and grown in KB medium with Rif at 28◦C, 200 rpm in an orbital shaker and harvested at log phase of growth (OD<sup>600</sup> = 1.0).

OD600 = 0.002 of P.s. pv. tabaci (1 × 10<sup>6</sup> CFU.mL−<sup>1</sup> ) was used for syringing (Zhang et al., 2015).

Tobacco seeds (N. benthamiana) were surface sterilized for 3 min in 75% ethanol, rinsed with sterile water for five times, and then germinated in 1/2 MS medium in a growth chamber maintained at 25◦C (14 h light/10 h dark). Following germination, seedlings were transferred to plantlets with autoclaved soil consisting of 1:1 (v/v) high-nutrient soil and vermiculite in pots and then cultured in a growth chamber at 25◦C with 50% humidity (14 h light/10 h dark). The plants were watered on alternate days.

# Gene Clone and Yeast Two-Hybrid Analysis

AtNRP protein sequence was used for blast search in N. benthamiana genome database<sup>1</sup> and finally obtained raw sequence of a putative protein (SGN-U514876) that contains 57.7% protein sequence similarity to AtNRP. The coding sequence of the protein was cloned with gene-specific primer set (Supplementary Table S1) and we designated this protein as Nbnrp1.

To investigate the PevD1-Nbnrp1 interaction, PevD1 and Nbnrp1 gene were cloned into BD vector of the pGBKT7 and AD vector of pGADT7 respectively. Y2H analysis was performed according to the protocol of the manufacturer (Matchmaker Gold Yeast Two-Hybrid System). The fragment of Nbnrp11C (residues 1–199) and Nbnrp11N (residues 200 to 332) were cloned into pGADT7 vector, respectively. The pevD1- PGBKT7 and PGADT7 derivates were co-transformed into yeast competent cells. Transformants were screened for growth on SD medium containing X-α-gal but lacking Leu, His, Trp, and Ade (Jin et al., 2007).

# BiFC Assays

The pSCYNE and pSCYCE plasmids were used for BIFC assays. PevD1 was cloned into pSCYNE to generate PevD1-pSCYNE and Nbnrp1 was cloned into pSCYCE to generate Nbnrp1-pSCYCE (Walter et al., 2004). The constructed PevD1-pSCYNE and Nbnrp1-pSCYCE plasmids were transformed into A. tumefaciens GV3101, respectively, and GV3101 were then infiltrated into the leaves of 4-week-old plants at the same time. The epidermal layers of the leaves were assayed for fluorescence 2 days after infiltration (Waadt et al., 2008). The fluorescence of cyan fluorescent protein (CFP) was examined under a laser confocal microscope (Fluo View 1000, Olympus).

# Protein Preparation

PevD1 was expressed and purified according to a previously described protocol (Wang et al., 2011). Protein concentration was measured using the BCA protein assay kit (Pierce, Rockford, IL, United States). Specific primers were designed to amplify full length of Nbnrp1, DCD domain (Nbnrp11N) and N-rich domain (Nbnrp11C) fragments (Supplementary Table S1). These three fragments were digested with BamH I and Sal

<sup>1</sup>https://solgenomics.net/

I, and then were cloned into pGEX-6p-2 vector, respectively. Recombinant vectors were transformed into E. coli BL21. Cells transformed with pGEX-6p-2-Nbnrp1, pGEX-6p-2-Nbnrp11N and pGEX-6p-2-Nbnrp11C were cultured in LB medium containing ampicillin (100 mg.L−<sup>1</sup> ) at 37◦C, 220 rpm for 8 h. Isopropyl b-D-thiogalactoside (IPTG) was then added to a final concentration of 0.2 mM to induce expression at 16◦C for 8 h. A culture transformed with the empty pGEX-6P-2 vector was used as a control. The bacteria were pelleted and re-suspended in buffer I (50 mM Tris and 200 mM NaCl, pH 8.0) and the cells were disrupted via sonication. The disrupted cells were then pelleted, and the supernatant was collected. GST affinity purification technology, desalination and ion exchange chromatography were utilized to purify GST-Nbnrp1, GST-Nbnrp11N, GST-Nbnrp11C and GST. All the samples were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis (Zhang et al., 2017c).

# Pull-Down Assays

6 × His tagged PevD1 and GST, GST-Nbnrp1, GST-Nbnrp11N, GST-Nbnrp11C expressed and purified from E. coli were used for GST pull-down assay. 6 × His tagged PevD1 and GST-fused protein were mixed with the GST affinity at 4◦C for 4 h. The GST affinity tag was washed with buffer (PBS) 3–5 times and then with the elution buffer (PBS, 20 mM GSH). The eluate was separated via 15% SDS-PAGE and transferred to a polyvinylidene fluoride (PVDF) membrane (Bio-Rad) using a Trans-Blot SD semi-dry electrophoretic transfer cell (Bio-Rad). Western blot was utilized to analyze the results. Proteins were probed with primary antibody anti-His tag mouse monoclonal antibody and anti-GST tag mouse monoclonal antibody (Abbkine) followed by secondary antibody alkaline phosphatase-conjugated goat antirabbit IgG (Transgene). The membrane was visualized with a 1 mL BCIP/NBT solution (Transgene) and observed the results using Tanon 5200 chemiluminescence apparatus (Beijing Yuanpinghao Biotech) (Sambrook and Russell, 2004; Shi et al., 2017).

# Generation of Nbnrp1-Overexpressing and Nbnrp1 Silence Transgenic Tobacco Plants

The coding sequence of Nbnrp1 was amplified by PCR using specific primers PBI121-Nbnrp1-F/R (Supplementary Table S1) and inserted into pBI121 to construct the fusion vector pBI121- Nbnrp1. The Nbnrp1 fragment and the expression vector pBI121 were digested with BamH I, and then the Nbnrp1 fragment was ligated into pBI121 vector using T4 ligase. The final overexpression vector, pBI121-Nbnrp1, containing the Nos terminator, NPT-II gene, and the CaMV35S promoter, was then transformed into N. benthamiana.

The sense fragment of Nbnrp1 was amplified by PCR using primers Sense-F/R (Supplementary Table S1), digested with Bgl II and constructed into the pRNAi1017 vector to get pRNAi1017-S vector. The anti-sense fragment of Nbnrp1 was amplified by PCR using primers Anti-sense-F/R (Supplementary Table S1), and digested with Sal I and BamH I, and then

ligated into the pRNAi1017-S vector to construct pRNAi1017-SA vector. Then the pCAMBIA2300 vector and the pRNAi1017- SA vector were digested with Sal I and Pst I, and the fragment with anti-sense and sense fragment of Nbnrp1 and intron was inserted into pCAMBIA2300 to construct the fusion vector pCAMBIA2300-pRNAi1017-Nbnrp1. The plasmid was subsequently transformed into Agrobacterium GV3101. Agrobacterium GV3101 harboring fusion vector was then transformed into N. benthamiana according to the method proposed by Horsch (1985).

1/2 MS medium with Kanamycin (30 mg/L) was used to select for positive transformants and PCR analysis was performed to detect putative transgenic tobacco plants. Genomic DNA from fresh, fully expanded tobacco leaves was extracted and PCR analysis was conducted using a specific fragment of the NTP II gene as a primer. After identification of positive transgenic plants, Southern blotting was performed to determine the copy numbers of Nbnrp1 according to the protocol provided with the DIG High Prime DNA Labeling and Detection Starter Kit II (Amersham Biosciences). To confirm that the selected transgenic lines were overexpression lines or with high silence efficiency, qPCR were performed using RNA prepared from transgenic plants. Non-transgenic plants served as controls. T<sup>3</sup> homozygous plants were used in this study.

# PevD1-Mediated Cell Death and Pathogen Inoculation Assay

In cell death assay, 4-week-old N. benthamiana leaves were infiltrated with 50 µl PevD1 solution in a gradient of concentrations (0.2, 0.5, 1.0, 2, and 5 µM) and the buffer was used as control. The resultant necrotic lesions were observed at 12, 24, and 48 h after treatment (Zhang et al., 2017b). TMV-GFP is a recombinant virus in which the jellyfish GFP gene in inserted into the coat protein (CP) open reading frame (ORF) of native TMV. GFP was visualized using a 100 W long-wave UV lamp (Black Ray model B 100AP; UVP, Upland, CA, United States). Recombination did not influence infection and movement of the virus in N. benthamiana (Liu et al., 2002).

Nicotiana benthamiana (Nbnrp1-overexpressing, wild type and Nbnrp1-silence lines) were infiltrated with 30 µl of 10 µM PevD1. BSA was used as negative control. Three days later, the upper three untreated leaves were inoculated with the TMV-GFP solution. The number of TMV-GFP lesions on each leaf was counted at 3 days post-inoculation (dpi), as previously described (Wang et al., 2011). The inhibition of TMV-GFP lesions was calculated using the following formula: inhibition (%) = [(number of lesions on wild-type plants − number of lesions on Nbnrp1-overexpressing plants)/number of lesions on wild-type plants] × 100.

For bacteria bioassay, the upper three PevD1-untreated leaves were inoculated with 50 µL bacterial suspension of P. s. pv. tabaci at 3 days post PevD1 treatment. The plants were maintained at a constant humidity for 3 days. Whole leaves were detached from the host plant and placed in 70% ethanol solution for 1 min, then rinsed in sterile distilled water three times. 1.5 cm<sup>2</sup> leaf disks were excised from the sampled leaves and ground with a plastic pestle in a 1.5 mL microfuge tube with 100 µL of sterile distilled water. A serial 1:10 (100 µL: 900 µL) dilution series was generated for each sample. The samples were plated on KB medium (Rif) and kept at 28◦C for 48 h. The bacterial colonies obtained from each dilution of each sample were counted and analyzed.

Assays for the pathogenicity of V. dahliae were performed as described (Bu et al., 2014; Zhang et al., 2017a). V. dahliae was cultured on potato dextrose agar medium at 28◦C in the dark for 5 days, and then the fungus was transferred into Czapek's medium with shock culturing at 150 rpm for 2–3 days at 25◦C in the dark. The conidial suspension was adjusted to 10<sup>7</sup> order magnitude using sterile distilled water for inoculation. Four-week-old N. benthamiana were treated with 30 µl of 10 µM PevD1 on cotyledons using a 1 ml syringe without needles. BSA was used as negative control. The inoculation of V. dahliae was conducted as described below: 20 ml of the conidial suspension was poured into each pot until complete absorption. Every experiment was performed with 24 plants and replicated three times. The inoculated seedlings were then grown for 14 days at 25◦C, with a day/night period of 14/10 h. The degree of wilt disease was divided into six grades: 0—healthy plant; grade 1—yellowing cotyledons; grade 2—wilting of one third of the leaves; grade 3—wilting of two thirds of the leaves; grade 4—wilting of all leaves; grade 5—plant death. Disease index value = [6(the number of seedling of every grade × relative grade)/total seedlings × highest score (4)] × 100 (Zhang et al., 2017a).

# RT-PCR and Quantitative Real-Time PCR

To investigate the transcription of defense-related genes, Nbnrp1 overexpressing lines, Nbnrp1-silence lines and wild type plant leaves were infiltrated with 10 µM PevD1. The samples were collected from the upper leaves at the indicated times and then rapidly frozen in liquid nitrogen. Total RNA was extracted with the RNA prep pure Plant Kit (TIANGEN Biotech). Residual genomic DNA was eliminated by treatment with a gDNA Eraser. First-strand cDNA was synthesized from 100 ng of total RNA using reverse transcriptase (TIANGEN Biotech) according to the supplier's protocol. Quantitative Real-time quantitative PCR (qPCR) was performed to determine the relative expression levels of several defense-related genes using SYBR Green PCR Master Mix (TIANGEN Biotech). Specific genes primers were designed according to the coding sequences of each gene using Beacon Designer 8. PCR mixture was processed on a Bio-Rad CFX Manager (Bio-Rad). Three technical replicates were amplified for each sample, including negative controls. EF-1a was used as an internal standard. Quantification of the relative changes in gene transcript levels was performed using the 2 <sup>−</sup>11C<sup>T</sup> method (Dean et al., 2002; Schmittgen and Livak, 2008).

# Statistical Analysis

All experiments and data presented here involved at least three repeats. The data are presented as means and standard deviations. The statistical analysis was performed with Statistical Analysis System (SAS) software using Student's t-test.

some plants. Consensus represents conserved amino acid residues, black line highlighted areas show DCD domain. (B) Phylogenetic analysis of Nbnrp1 and its homologs. (C) Interaction of PevD1 and Nbnrp1 in yeast. The indicated AD and BD constructs were transformed into the Y2H gold yeast strain. Transformants were assayed for the activity of protein–protein interactions using reporter genes based on growth on QDO/A/x-α-gal selective medium (SD/-Ade/-His/-Leu/-Trp medium containing Aureobasidin A and x-α-gal). DDO (SD/-Leu/-Trp medium) was used to observe the growth of transformants on non-selective control plates. (D) BiFC visualization of the interaction between PevD1 and Nbnrp1 in tobacco leaves. Scale bar = 50 µm. CFP fluorescence and bright field images of leaf cells from Nicotiana benthamiana infiltrated with a mixture of Agrobacterium suspensions harboring constructs encoding the indicated proteins.

# RESULTS

# Identification of the PevD1 Interacting Protein Nbnrp1

Elicitor PevD1 has been shown to induce broad spectrum resistance in various plants including cotton, tobacco, and Arabidopsis. Accordingly, partner protein of PevD1 should be conserved in plants. As NRPs are widely distributed and highly conserved among plants, we cloned AtNRP homologous gene from N. benthamiana, designated as Nbnrp1.

Bioinformatics analysis showed that Nbnrp1 contains 396 amino acid residues, has a predicted molecular mass of 37 kDa, and includes 2 domains (an asparagine-rich domain and a DCD domain). Nbnrp1 was characterized as an asparagine-rich protein because of a high content of the amino acid asparagine (about 20%) in its N-terminus. Nbnrp1 belongs to subgroup of the DCD protein family according to the location of the DCD domain in the protein (Tenhaken et al., 2005). Thus far, the functions of DCD protein have been described in only four homologs, including the B2-protein from carrot (Schrader et al., 1997), GDA2 identified in pea (Li et al., 1998), N-rich protein isolated in soybean (Ludwig and Tenhaken, 2001) and AtNRP that was found from Arabidopsis previously (Zhou et al., 2017). The amino acid sequence of Nbnrp1 showed 46.27, 40.65, 55.38, and 57.7% similarity with B2, GDA2, N-rich protein, and AtNRP, respectively. Sequence alignment of Nbnrp1 and its homologs indicated the conserved DCD domain (**Figure 1A**). In an unrooted phylogenetic tree (**Figure 1B**), Nbnrp1 shares high similarity with N-rich protein and AtNRP. To investigate whether Nbnrp1 could bind with PevD1, the Y2H assay was carried out. The BD vector with the PevD1 without the signal peptide sequence and AD vector with Nbnrp1 were constructed and co-transformed into yeast cell. Co-transformed BD/AD, BD/AD- Nbnrp1 and BD-PevD1/AD were as negative controls. Y2H assays showed all of the transformants were able to grow on DDO medium (SD medium lacking His and Trp (SD/-His/- Trp), but only the BD-PevD1/AD-Nbnrp1 co-transformed yeast could grow on QDO/A/x-a-gal medium (SD/-Ade/-His/-Ieu/- Trp/AbA/x-a-gal), indicating PevD1 could interact with Nbnrp1 in yeast (**Figure 1C**). BiFC was further performed to verify the interaction between PevD1 and Nbnrp1 in tobacco cells. CFP fluorescence was detected in tobacco cells under laser confocal scanning microscopy, showing the two proteins could interact in tobacco cell (**Figure 1D**).

# The DCD Domain Is Required for Nbnrp1-PevD1 Interaction

To determine whether the DCD domain of Nbnrp1 involved in the interaction between PevD1 and Nbnrp1, we constructed two Nbnrp1 deletion mutants: Nbnrp11C (residues 1–199), which

contained four sequences of low compositional complexity, and Nbnrp11N (residues 200–332), which contained the conserved DCD domain (**Figure 2A**). The Y2H assays showed that only Nbnrp11N could interact with PevD1 in yeast cells (**Figure 2B**). To further confirm the Y2H results, GST pull-down assay was performed. The purified GST-Nbnrp1, GST-Nbnrp11N, GST-Nbnrp11C, and GST proteins were produced in E. coli (**Figure 2C**). The GST pull-down assay showed that GST-Nbnrp1 and Nbnrp11N could interact with PevD1 (**Figure 2D**). These results suggested that the DCD domain is required for the Nbnrp1-PevD1 interaction.

# Generation of Nbnrp1-Overexpressing and Nbnrp1-Silence Transgenic Tobacco Plants

To investigate the role of Nbnrp1 in PevD1-induced necrosis and disease resistance, Nbnrp1-overexpressing and Nbnrp1-silence transgenic tobacco plants are generated. The fusion vector was transformed into tobacco via the Agrobacterium-mediated method. A specific fragment of the NTPII gene (approximately 560 bp in length) was amplified to identify the transgenic lines (**Figure 3A**). Southern blotting showed that the identified transgenic plants carried a single copy according to the protocol (**Figure 3B**). The expression level of Nbnrp1 in transgenic plants was detected by qPCR, we then selected Nbnrp1-overexpressing lines with high Nbnrp1 expression level (OL3 and OL4) and Nbnrp1-silence lines with high silence efficiency (SL2 and SL6) for further study (**Figure 3C**). The T3 homozygous were used for PevD1-mediated cell death and the disease resistance assays.

# Effect of Nbnrp1-Overexpressing Lines and Nbnrp1-Silence Lines on PevD1-Induced Necrosis Response

The transgenic and wild type tobacco leaves were infiltrated with 50 µl of PevD1 protein solution in a gradient of concentrations (0, 0.2, 0.5, 1, 2, and 5 µM) to understand the role of Nbnrp1 in PevD1-mediated necrosis, the necrosis was then observed at 12, 24, and 48 h post infiltration. Nbnrp1-overexpressing lines appeared HR earlier than wild type plants, and showed more severe necrosis at the same observation period. Likewise, Nbnrp1-silence lines showed weakened effect compared with wild type plants (**Figure 4A**). Meanwhile, the transcription level of the HR marker gene HSR203J increased in the Nbnrp1 overexpressing lines and decreased in the Nbnrp1-silence line compared with the wild-type at 12 h after PevD1 infiltration (**Figure 4B**). These results indicated that Nbnrp1 mediates PevD1-induced HR responses.

# Nbnrp1-Overexpressing and Nbnrp1-Silence Lines Impact PevD1-Induced Disease Resistance

The N. benthamiana Nbnrp1-overexpressing lines, Nbnrp1-silence lines and wild type were inoculated with

∗∗0.001 < P < 0.01, ∗∗∗P < 0.001).

(

TMV at 3 days post PevD1 infiltration, respectively. The Nbnrp1 overexpressing lines showed enhanced disease resistance against TMV and the number of TMV-GFP lesions in systemic leaves was obviously less than that of wild type plants. The number of TMV lesions was reduced by approximately 41.2% at 4 dpi. But the Nbnrp1-silence lines appeared attenuated

TMV disease resistance. The number of TMV-GFP lesions in systemic leaves was obviously more than that of wild-type plants, and the number of TMV lesions increased by 71.7% at 4 dpi (**Figures 5A**, **6A**). Furthermore, the Nbnrp1-overexpressing lines also exhibited enhanced systemic resistance against P.s. pv. tabaci, the bacterial counts in the Nbnrp1-overexpressing lines were significantly reduced by 91.1% at 3 dpi. While the Nbnrp1-silence lines attenuated systemic resistance against P.s. pv. tabaci. The bacterial counts in the Nbnrp1-silence lines were increased by 67% at 3 dpi (**Figures 5B**, **6B**). Besides, the capacity of PevD1 to induce systemic disease resistance against the fungal pathogen V. dahliae in Nbnrp1 overexpressing lines was significantly higher than that in Nbnrp1-silence lines and wild type. Compared with wild type, the disease index displayed a statistically significant decrease in Nbnrp1-silence lines (**Figures 5C**, **6C**). These data indicate that Nbnrp1 positively regulates PevD1-induced disease resistance.

To further confirm that the Nbnrp1-overexpressing lines and Nbnrp1-silence lines impact PevD1-induced disease resistance, the relative expression level of two defense-related genes PR1-a and PR1-b were detected via qPCR. The relative expression level of PR1-a and PR1-b were elevated in the Nbnrp1-overexpressing lines and declined in the Nbnrp1-silence lines compared to wild type plants at 3 days post PevD1 infiltration (**Figure 6D**).

# DISCUSSION

The fungal pathogen V. dahliae causes serious yield losses worldwide in extensive plants including cotton and tobacco (Miao et al., 2010). Hundreds of proteins are secreted in

V. dahliae infection process and contribute to regulate compatible and incompatible plant-pathogen interaction (Ellis et al., 2006). However, how these fungal secretory proteins enter the plant cell is still limited. Some of these proteins have been shown to produce necrosis and induce oxidative burst and phytoalexin accumulation when applied to plants in an isolated form. For example, a 65 kDa glycoprotein from the V. dahliae could induce phytoalexin accumulation and oxidative burst (Davis et al., 1998). The fusion protein VdNEP triggered PR gene expression, gossypol and sesquiterpene phytoalexins synthesis (Wang et al., 2004).

We previously characterized a V. dahliae new secretory protein PevD1 that could trigger cell death in tobacco as well as a series of defense responses and improve disease resistance to TMV and V. dahliae (Wang et al., 2011; Bu et al., 2014). Whereas how PevD1 trigger plant immunity is unknown. The interacting partner is a key player in uncovering defense signal network. Our previous research demonstrated that PevD1 could interact with Arabidopsis NRP protein, while NRPs are conserved in plant kingdom. NRPs involved in responsing to different stresses such as salt stress, oxidative stress, mechanical perturbation and pathogen infection (Hoepflinger et al., 2011). NRPs also mediated cell death signaling in ER stress pathway, and further researches showed that DCD domain plays an important role in cell death (Tenhaken et al., 2005; Faria et al., 2011). To investigate that NRP mediated PevD1-induced immune response in N. benthamiana, in present research, we cloned homologous gene of AtNRP from N. benthamiana genome, named as Nbnrp1 that encode a putative protein of NRP family. Binding assays showed that Nbnrp1 could bind with PevD1 (**Figure 1**) and C-terminal DCD domain of Nbnrp1 was required for PevD1-Nbnrp1 interaction (**Figure 2**).

Nbnrp1 is a putative protein of NRP family in N. benthamiana genome and its functions are unclear. Our research showed that Nbnrp1-silence plant lines attenuated PevD1-induced necrotic cell death (**Figure 4A**) and the transcription of the HR marker gene HSR203J also reduced (**Figure 4B**). Furthermore, the Nbnrp1-overexpressing transgenic lines displayed accelerated cell death and elevated transcription of marker gene HSR203J. This phenomenon indicated that Nbnrp1 positively regulate PevD1-induced cell death and disease resistance, although detail mechanism is still unknown. It is first report that N. benthamiana NRP mediates fungal elicitor protein-triggered plant immunity.

Plant cell death is a conventional indicator of resistance, which based on plant disease resistance genes (R genes) recognizing the pathogen avirulence (Avr) genes, leading to activate cell death pathways (Heath, 2000). Except for R genes, plant immunity components or regulators are determinants of necrotic cell death via the directly or indirectly association with pathogen elicitors/effectors. For example, tomato papain-like cysteine protease (PLCP), C14 target pathogen effectors EPIC1 and EPIC2B secreted from Phytophthora infestans, contributed to immunity via participation in cell death signaling. C14 silencing plants showed increased susceptibility to hemibiotrophs of P. infestans (Kaschani et al., 2010). In our present research, bioassay showed the Nbnrp1-overexpressing lines and Nbnrp1 silence lines remarkable influence PevD1-induced necrosis activity and disease resistance against viruses TMV and bacterial pathogen P.s. pv. tabaci. These results indicated that Nbnrp1 mediate PevD1-induced cell death and resistance in

was used as negative control.

N. benthamiana. However, how the Nbnrp1 mediate or regulate cell death signal activated by elicitor PevD1 remains unknown. Whether Nbnrp1 mediate ER-stress cell death also require further exploration.

The structure of PevD1 resembles C2 domain-like structure with a calcium ion bound to the C-terminal acidic pocket and C-terminal 57 amino acid of PevD1 is essential to trigger HR in N. benthamiana (Liu et al., 2014; Zhou et al., 2017). Calcium ion (Ca2+) plays a crucial role in regulating cellular responses to environmental stresses (Dodds and Rathjen, 2010). Ca2<sup>+</sup> concentration increases when plant perceived stimulus and activates downstream of Ca2<sup>+</sup> signal transduction pathways. In view of the above reasons, we speculated that C-terminus of PevD1 maybe contains a key amino acid fragment that responsible to bind partner protein Nbnrp1, which activates MAPK cascade leading to plant immune response. Details of amino acid in PevD1 binding with Nbnrp1 remain need to be investigated further.

Overexpression of NRP in soybean protoplast resulted in chlorophy II reducing, leaf chlorosis and senescence marker gene induction (Costa et al., 2008; Reis et al., 2011). Whereas, there was no abnormal development and leaf chlorosis were observed in Nbnrp1-overexpressing lines and Nbnrp1-silence lines. We speculated that the function of Nbnrp1 in N. benthamiana may be different from that of NRPs in soybean, although they belong to the same family of N-rich proteins and exhibit about 60% similarity. The effect of Nbnrp1 in N. benthamiana development needs further research.

# CONCLUSION

fpls-09-00303 March 6, 2018 Time: 17:11 # 11

We first identified the PevD1 binding protein Nbnrp1 in N. benthamiana and demonstrated that Nbnrp1 mediate PevD1 induced cell death and disease resistance. DCD domain of Nbnrp1 was required for the interaction with PevD1. The Nbnrp1 in N. benthamiana transgenic lines obviously impacted PevD1 induced cell death and disease resistance to TMV, bacterial pathogen P.s. pv. tabaci and fungal pathogen V. dahliae compared with the wild type plants. Our results demonstrated that the protein Nbnrp1 positively modulates PevD1-induced immune response in N. benthamiana.

# AUTHOR CONTRIBUTIONS

XY designed the experiments. YL and SC carried out the experiments and wrote the manuscript. XT analyzed the

# REFERENCES


experimental results. YZ, DQ, HZ, LG, and JY assisted the experiments.

# FUNDING

This research was supported by the National Natural Science Foundation of China (Grant No. 31772151).

# ACKNOWLEDGMENTS

We thank Professor Jun Liu from the Chinese Academy of Sciences for carefully reviewing the manuscript.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00303/ full#supplementary-material

is a downstream component of the ER stress- and osmotic stress-induced NRPmediated cell-death signaling pathway. BMC Plant Biol. 11:129. doi: 10.1186/ 1471-2229-11-129



**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 Liang, Cui, Tang, Zhang, Qiu, Zeng, Guo, Yuan and Yang. 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 Polycistronic miR166k-166h Positively Regulates Rice Immunity via Post-transcriptional Control of EIN2

#### Raquel Salvador-Guirao<sup>1</sup> , Yue-ie Hsing<sup>2</sup> and Blanca San Segundo1,3 \*

<sup>1</sup> Centre for Research in Agricultural Genomics CSIC-IRTA-UAB-UB, Universitat Autònoma de Barcelona, Barcelona, Spain, 2 Institute of Plant and Microbial Biology, Academia Sinica, Taipei, Taiwan, <sup>3</sup> Consejo Superior de Investigaciones Científicas, Barcelona, Spain

MicroRNAs (miRNAs) are small RNAs acting as regulators of gene expression at the post-transcriptional level. In plants, most miRNAs are generated from independent transcriptional units, and only a few polycistronic miRNAs have been described. miR166 is a conserved miRNA in plants targeting the HD-ZIP III transcription factor genes. Here, we show that a polycistronic miRNA comprising two miR166 family members, miR166k and miR166h, functions as a positive regulator of rice immunity. Rice plants with activated MIR166k-166h expression showed enhanced resistance to infection by the fungal pathogens Magnaporthe oryzae and Fusarium fujikuroi, the causal agents of the rice blast and bakanae disease, respectively. Disease resistance in rice plants with activated MIR166k-166h expression was associated with a stronger expression of defense responses during pathogen infection. Stronger induction of MIR166k-166h expression occurred in resistant but not susceptible rice cultivars. Notably, the ethyleneinsensitive 2 (EIN2) gene was identified as a novel target gene for miR166k. The regulatory role of the miR166h-166k polycistron on the newly identified target gene results from the activity of the miR166k-5p specie generated from the miR166k-166h precursor. Collectively, our findings support a role for miR166k-5p in rice immunity by controlling EIN2 expression. Because rice blast is one of the most destructive diseases of cultivated rice worldwide, unraveling miR166k-166h-mediated mechanisms underlying blast resistance could ultimately help in designing appropriate strategies for rice protection.

Keywords: blast, ethylene-insensitive 2 (EIN2), miR166, Oryza sativa, Magnaporthe oryzae, rice

# INTRODUCTION

MicroRNAs (miRNAs) are endogenous, small non-coding RNAs that mediate post-transcriptional gene silencing in eukaryotes (Jones-Rhoades et al., 2006). They are transcribed as long primary transcripts (pri-miRNAs), forming an imperfect fold-back structure, and are sequentially processed by a DICER-like ribonuclease (typically DCL1) to produce a pre-miRNA and finally a doublestranded miRNA duplex, the miRNA-5p/miRNA-3p duplex (previously named miRNA/miRNA<sup>∗</sup> duplex) (Kurihara and Watanabe, 2004). The miRNA-5p/miRNA-3p duplexes are then transported

Edited by:

Yi Li, Peking University, China

#### Reviewed by:

Wen-Ming Wang, Sichuan Agricultural University, China Xiaoming Zhang, University of Chinese Academy of Sciences (UCAS), China

\*Correspondence:

Blanca San Segundo blanca.sansegundo@cragenomica.es

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 27 December 2017 Accepted: 28 February 2018 Published: 20 March 2018

#### Citation:

Salvador -Guirao R, Hsing Y-i and San Segundo B (2018) The Polycistronic miR166k-166h Positively Regulates Rice Immunity via Post-transcriptional Control of EIN2. Front. Plant Sci. 9:337. doi: 10.3389/fpls.2018.00337

**29**

to the cytoplasm, where the functional miRNA strand is incorporated into an ARGONAUTE1 (AGO1)-containing RNAinduced silencing complex (RISC) (Baumberger and Baulcombe, 2005; Jones-Rhoades et al., 2006; Rogers and Chen, 2013). miRNAs guide post-transcriptional gene silencing via sequencespecific cleavage or translational repression of target transcripts (Llave et al., 2002; Brodersen et al., 2008).

The crucial role of miRNAs in controlling plant developmental processes and response to abiotic stress is well documented (De Lima et al., 2012). Alterations in the accumulation of a substantial fraction of the miRNAome during pathogen infection is also described in different pathosystems, and for some miRNAs a role in plant immunity has been described (Shivaprasad et al., 2012; Campo et al., 2013; Boccara et al., 2014; Li et al., 2014; Baldrich and San Segundo, 2016; Soto-Suárez et al., 2017). However, our current knowledge of the biological roles of pathogen-regulated miRNAs in plant immunity is still limited, and most comes from studies in the interaction of Arabidopsis thaliana with the bacterial pathogen Pseudomonas syringae (Staiger et al., 2013; Weiberg et al., 2014; Fei et al., 2016; Kuan et al., 2016).

miRNAs are thought to have originated by duplication of pre-existing protein-coding genes with subsequent mutations (Allen et al., 2004; Rajagopalan et al., 2006). The spontaneous evolution from hairpin structures in the genome, or derivation from transposable elements, has also been proposed to explain the origin of plant miRNAs (Felippes et al., 2008; Nozawa et al., 2012). Whole-genome duplication events, and tandem or segmental duplications of MIR genes, are believed to be responsible for the expansion and diversification of miRNA gene families in plants (Maher et al., 2006; Nozawa et al., 2012). In animals, the occurrence of miRNA clusters is common, but only a few miRNA clusters have been described in plants, mainly in Arabidopsis (Boualem et al., 2008; Merchan et al., 2009; Barik et al., 2014; Baldrich et al., 2016). These clustered miRNAs can be transcribed independently or simultaneously as polycistronic transcripts. Furthermore, transcripts of polycistronic miRNAs might contain copies of members belonging to the same miRNA family (homologous polycistron), or unrelated miRNAs (nonhomologous polycistron).

The miR166 family comprises multiple members in monocotyledonous and dicotyledonous plants that are transcribed independently (monocistrons). This is a highly conserved family of miRNAs with conserved target genes, the Class III homeodomain-leucine zipper (HD-ZIP III) transcription factors. These transcription factors, such as the Arabidopsis PHABULOSA (PHB) and PHABOLUTA (PHV), are involved in diverse developmental processes (Emery et al., 2003; Itoh et al., 2008). Altered accumulation of miR166 during abiotic stress also led to the notion that miR166 might play a role in the plant response to diverse abiotic stresses. Very recently, it has been described that miR166 knockdown triggers drought resistance in rice (Zhang et al., 2018). Evidence for miR166 in adapting to pathogen infection in plants has not been reported.

Recently, we described the occurrence of a rice polycistronic miRNA, miR166k-166h, comprising two miR166 family members (miR166k and miR166h). Expression profiling revealed that mature miRNAs generated from the miR166k-166h precursor are co-expressed in rice leaves (Baldrich et al., 2016). In other studies, various miR166 species were found to differentially respond to infection by the rice blast fungus M. oryzae or to differentially accumulate in blast-resistant and blast-susceptible rice varieties (Li et al., 2014, 2016).

In this work, we present evidence supporting that MIR166k-166h plays a role in rice immunity. We show that rice plants with activated MIR166k-166h expression exhibit resistance to infection by the fungal pathogens M. oryzae and Fusarium fujikuroi, the causal agents of the rice blast and bakanae disease, respectively. Rice blast is one of the most devastating diseases of cultivated rice due to its widespread distribution and destructiveness (Wilson and Talbot, 2009). The phenotype of disease resistance is associated with a stronger induction of defense responses during pathogen infection. MIR166h-166k expression was strongly induced by M. oryzae infection in blastresistant but not in blast-susceptible rice varieties. Moreover, we identified a novel target gene for miR166k, the ethyleneinsensitive 2 (EIN2) gene (targeted by miR166k-5p in the miR166k-166h polycistron). Overall, our results support that the polycistronic miR166k-166h positively regulates rice immunity through modulation of EIN2 expression.

# MATERIALS AND METHODS

# Plant Material

Rice (Oryza sativa) plants were grown at 28°C/22°C under 16-h light/8-h dark conditions. The T-DNA insertion line for MIR166k-166h (M0110144) and wild-type genotype (O. sativa japonica cv Taining 67) were obtained from the Taiwan Rice Insertional Mutant collection (TRIM<sup>1</sup> ). Genotyping of the TRIM mutant was carried out by PCR on genomic DNA using a T-DNA-specific primer located at the left border of the T-DNA and a primer located in the vicinity of the insertion site. PCR products were confirmed by DNA sequencing. Quantitative PCR (qPCR) was used to determine the T-DNA copy number in the rice mutant with the monocopy sucrose phosphate synthase gene used as the endogenous reference (Ding et al., 2004) (primers are listed in Supplementary Table S1).

The rice cultivars Saber, TeQing, Kanto 51, Maratelli and Vialone Nano were obtained from the germplasm seed bank of the Consiglio per la Ricerca e la Sperimentazione in Agricoltura (CRA-Rice Research Unit, Vercelli, Italy).

# Infection Assays and Elicitor Treatment

The fungus M. oryzae (strain Guy-11) was grown on complete medium as described (Campos-Soriano et al., 2012). For infection assays with M. oryzae, 3-week-old plants were spray-inoculated with a spore suspension (5 × 10<sup>5</sup> spores/ml), or mockinoculated. Development of disease symptoms was followed over time. Lesion area was determined by using Assess 2.0 software (American Phytopathological Society). For infection assays with Fusarium fujikuroi, the fungus was grown on PDA (Difco, Franklin Lakes, NJ, United States). Rice seeds were

<sup>1</sup>http://trim.sinica.edu.tw/

pregerminated for 24 h on Murashige and Skoog (MS) medium and then inoculated with a suspension of F. fujikuroi spores (1 × 10<sup>6</sup> spores/ml), or sterile water. Seedlings were allowed to continue germination for 1 week. Three independent infection experiments were performed, with at least 24 plants per genotype in each experiment. Statistically significant differences were determined by one-way ANOVA. qPCR was used to quantify fungal DNA in infected leaves with specific primers for the 28S DNA gene of the corresponding fungus (Qi and Yang, 2002; Jeon et al., 2013). For this, standard curves were prepared by using M. oryzae or F. fujikuroi DNA.

For elicitor treatment, 3-week-old plants were sprayed with an elicitor suspension of M. oryzae (3 × 10<sup>2</sup> µg/ml) or mockinoculated as described (Casacuberta et al., 1992).

# 1-Aminocyclopropane-1-Carboxylic Acid (ACC) Treatment

Three-week old rice plants were treated with ACC (Merck, Darmstadt, Germany) at a concentration of 50 µM for 15 min, 1, 4, and 24 h. Control plants were mock-inoculated.

#### RT-qPCR, Stem-Loop RT-PCR and 5 <sup>0</sup> RACE-PCR

Total RNA was extracted by using TRIzol Reagent (Invitrogen). First-strand cDNA was synthesized from DNAse-treated total RNA (1 µg) with SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA, United States) and oligo-dT. RTqPCR was performed with Light Cycler 480 and SYBR Green (Roche, Basel, Switzerland). Primers were designed by using the Primer3 software<sup>2</sup> . Primers for detection of pre-miR166k-166h were designed based on the precursor sequence information from miRBase. PCR products were confirmed by DNA sequencing. The average cycle threshold (Ct) values were obtained by PCR from three independent biological replicates and normalized to the mean Ct values for cyclophilin 2 gene (Os02g02890) from the same RNA preparations, yielding the relative expression (1Ct value). The 2−11Ct method was used to determine the fold-change of gene expression (infected/elicitor-treated "versus" mock-inoculated).

Stem-loop RT-qPCR was performed as described (Varkonyi-Gasic et al., 2007). Modified 5<sup>0</sup> -RNA ligase-mediated RACE was performed as described (Llave et al., 2011). The PCR products were cloned and sequenced to determine the cleavage site in target genes. Primers used for RT-qPCR and stem-loop RT-PCR are in Supplementary Table S1.

# Agroinfiltration in Nicotiana benthamiana Leaves

For transient expression of MIR166k-166h, the genomic DNA fragment encompassing the entire miR166k-166h precursor was obtained by PCR from genomic DNA and cloned into the pCAMBIA5300 vector (pC5300)<sup>3</sup> under the control of the maize ubiquitin promoter. The OsEIN2.1 cDNA sequence (Ma et al.,

<sup>2</sup>http://www.ncbi.nlm.nih.gov

<sup>3</sup>www.cambia.org

2013) was cloned into the pCAMBIA2300 expression vector (pC2300) designed to produce C-terminal GFP-tagged fusion proteins under the control of the 35S Cauliflower Mosaic Virus promoter. Plasmid constructs were introduced into the Agrobacterium tumefaciens EHA105 strain. As a negative control, the empty vector was used. The N. benthamiana RDR6-IR line deficient in expression of RNA-dependent RNA polymerase 6 was used as a host plant (Schwach et al., 2005). Constructs harboring the miR166k-166h precursor or EIN2-GFP, alone or in combination, were agroinfiltrated in N. benthamiana leaves, and their expression was monitored at 2 days after agroinfiltration.

Northern blot analysis of the agroinfiltrated leaves involved using the small RNA fraction obtained from total RNA (200 µg). The oligonucleotide complementary to the miR166 sequence (Supplementary Table S1) was labeled with digoxigenin with the DIG oligonucleotide 3<sup>0</sup> -End Labeling kit (Roche, Basel, Switzerland). For detection of the EIN2-GFP protein, total protein extracts (50 µg) were subjected to SDS-PAGE (12.5% gels) and probed with an anti-GFP antibody (Invitrogen, Carlsbad, CA, United States).

# RESULTS

# MIR166k-166h Activation Enhances Resistance to Infection by the Rice Blast Fungus M. oryzae

The rice genome contains several loci encoding monocistronic miR166s distributed on 7 chromosomes: miR166a, miR166b, miR166c, miR166d, miR166e, miR166f, miR166g, miR166i, miR166j, miR166l and miR166m (miRBase release 21) (Supplementary Figure S1). Furthermore, a polycistronic miR166 encoding two miR166 family members, the miR166k-166h precursor, was identified on chromosome 2 (Baldrich et al., 2016). The mature miR166k and miR166h species locate in one or another hairpin of the miR166k-166h precursor structure (**Figure 1A**, left panel). Of note, loci encoding monocistronic transcripts for miR166k or miR166h have not been identified in the rice genome.

In this work, a T-DNA tagged line (M0110144) carrying the T-DNA insertion upstream of the MIR166k-166h locus was identified in the TRIM collection produced in the Tainung 67 (TN67) background (Hsing et al., 2007). Of note, mutant alleles for miRNAs are not easily found in insertional mutant collections because of the small size of MIR genes. The T-DNA contains 8 copies of the CaMV35 enhancer near the left border, and genes within 15 kb of the T-DNA left border and 5 kb of the right border might be activated by these enhancers. Knowing this, we hypothesized that this mutant might be an activation mutant for MIR166k-166h. The T-DNA insertion site was confirmed by PCR genotyping followed by DNA sequencing of the PCR products (Supplementary Figure S2A). No homozygous MIR166k-166h plants were identified in PCR genotype screens. Most importantly, heterozygous mutant plants accumulated higher levels of miR166k-166h precursor transcripts, which correlated well with an increase in the accumulation of mature

FIGURE 1 | Characterization of polycistronic miR166k-166h mutant plants. (A) Structure of the miR166k-166h precursor and location of mature miR166 sequences (left panel). The accumulation of miR166k-166h precursor transcripts and mature miR166 sequences in wild-type (TN67) and miR166k-166h-Ac mutant plants was determined by RT-qPCR and stem-loop RT-qPCR, respectively (right panels). Note that the stem-loop RT-qPCR does not discriminate among miR166k-3p and miR166h-3p sequences (indicated as miR166kh-3p). (B) Phenotype of wild-type and miR166k-166-Ac mutant plants at 7 days post-inoculation with M. oryzae spores (5 × 10<sup>5</sup> spores/ml). The percentage of leaf area affected by blast lesions was determined by image analysis (APS Assess 2.0) (right upper panel). Quantification of M. oryzae DNA was determined by qPCR with specific primers of the M. oryzae 28S gene (right lower panel). (C) Resistance to infection by F. fujikuroi in miR166k-166h-Ac mutant plants. Pictures were taken at 7 days after inoculation with fungal spores. Quantification of fungal DNA was carried out by qPCR using specific primers for F. fujikuroi (right panel). (D) Accumulation of transcripts for the defense marker genes OsPBZ1 and OsPR1a in wild-type and miR166k-166h-Ac plants in response to M. oryzae infection determined by RT-qPCR. Plants were inoculated with M. oryzae spores (5 × 10<sup>5</sup> spores/ml) or mock-inoculated (+ and –, respectively). Ct values obtained in the PCR reactions were normalized to the average Ct values for the cyclophilin 2 gene (for graphical representation, the values are multiplied by 100). Data are mean ± SD (∗∗∗P ≤ 0.001; ∗∗P ≤ 0.01, ANOVA test, M. oryzae-inoculated versus mock-inoculated).

miR166k and miR166h sequences (**Figure 1A**, right panel). Since the rice genome does not contain monocistronic miR166k and miR166h loci, the miR166k and miR166h mature sequences accumulating in rice leaves are expected to be generated from the polycistronic miR166k-166h precursor. These observations confirmed that the TRIM mutant is an activation mutant for MIR166k-166h (hereafter referred to as miR166k-166h-Ac). However, miR166 has been shown to repress the seed maturation program in Arabidopsis, and difficulties in generating transgenic lines overexpressing miR166 were previously reported (Tang et al., 2012). Presumably, high levels of miR166 expression and concomitant silencing of HD-ZIP III might compromise normal plant development. Therefore, it is not surprising that homozygous miR166k-166h-Ac mutant plants could not be identified in this study. The miR166k-166h-Ac mutant harbors a single copy of the T-DNA inserted in its genome (Supplementary Table S2).

We considered the possibility that the expression of genes other than MIR166k-166h might be activated in the miR166k-166h-Ac mutant. Two genes, OsSAUR12 (Os02g52990) and Erwinia-induced protein (Os02g53000), were identified upstream and downstream, respectively, of the T-DNA insertional site (Supplementary Figure S2A). However, we found no altered accumulation of OsSAUR12 or Erwinia-induced protein transcripts in the miR166k-166h-Ac mutant (Supplementary Figure S2B). There were no obvious phenotypic differences between miR166k-166h mutant and wild-type plants under controlled greenhouse conditions (Supplementary Figure S2C).

To investigate whether miR166k-166h miRNA plays a role in rice immunity, we performed blast disease resistance assays. Wild-type (cv TN67) and miR166k-166h-Ac plants were sprayinoculated with spores of the rice blast fungus M. oryzae, and disease symptoms were followed over time. The miR166k-166h-Ac plants consistently showed reduced disease symptoms as compared with wild-type plants (**Figure 1B**, left panel). Blast resistance was confirmed by quantification of fungal biomass and determination of lesion area in the infected leaves (**Figure 1B**, right panels).

The miR166k-166h-Ac mutant plants also showed enhanced resistance to infection by the fungus F. fujikuroi, the causal agent of bakanae in rice (Ou, 1985). The fungus infects the plant through the roots (or crowns) and grows systemically within the plant. At 7 days after inoculation, the miR166k-166h-Ac seedlings exhibited more vigorous growth of the root system compared to wild-type seedlings which also had extensive necrosis in their roots (**Figure 1C**, left panel). Quantification of fungal biomass confirmed limited fungal growth in roots of miR166k-166h-Ac seedlings (**Figure 1C**, right panel).

To obtain further insights into the mechanisms underlying disease resistance in the miR166k-166h-Ac mutant, we determined the expression pattern of the defense genes OsPBZ1 (Probenazole-inducible 1) and OsPR1a (Pathogenesis-Related 1a) in mutant and wild-type plants at different times after infection with M. oryzae (24, 48, and 72 h post-inoculation [hpi]). OsPBZ1 (a member of the PR10 family of PR genes) and OsPR1a genes are markers for the activation of the rice defense response to M. oryzae infection (Midoh and Iwata, 1996; Agrawal et al., 2001). As expected, fungal infection induced OsPR1a and OsPBZ1 expression in wild-type plants. Importantly, transcript levels of these defense genes were higher in M. oryzae-inoculated miR166k-166h-Ac than M. oryzae-inoculated wild-type plants at all times of infection (**Figure 1D**). These findings support that the miR166k-166h-Ac mutant responds to pathogen challenge with a super-induction of defense genes, which is consistent with the phenotype of disease resistance observed in these plants.

# MIR166k-166h Expression During Fungal Infection and Treatment With Elicitors

Given that activation of MIR166k-166h affects disease resistance, we sought to investigate whether MIR166k-166h expression is itself regulated during the normal host response to infection. Upon pathogen challenge, miR166k-166h precursor transcript level was increased in leaves of M. oryzae-inoculated compared with non-inoculated wild-type TN67 plants, with a parallel increase in mature miR166k and miR166h sequences (both miRNA-5p and miRNA-3p species) (**Figure 2A**). Interestingly, accumulation of precursor and mature miR166 sequences also increased in response to treatment with a crude preparation of elicitors (**Figure 2B**). Elicitor treatment resulted in faster induction of miR166k-5p and miR166h-5p species versus miR166kh species. Induction of marker genes of defense activation, OsPBZ1 and OsPR1a, confirmed that the host plant detects and responds to elicitor treatment (Supplementary Figure S3A). Finally, we examined the elicitor-responsiveness of the monocistronic miR166s, miR166a and miR166c. The accumulation of precursor transcripts for these miR166 family members (pre-miR166a and pre-miR166c) was found to be transiently, but not significantly, regulated during elicitor treatment (Supplementary Figure S3B).

From these results, we concluded that pathogen infection and also treatment with fungal elicitors upregulates MIR166k-166h expression, which suggests a role of this polycistronic miRNA in pathogen-associated molecular pattern (PAMP) triggered immunity (PTI).

The promoter region of protein-coding genes often includes cis-acting regulatory elements responsible for pathogen inducibility. Knowing that fungal infection and elicitor treatment induced MIR166k-166h expression, we scanned the MIR166k-166h promoter region for the presence of cis-regulatory elements related to biotic stress. The sequence upstream of the precursor structure for the miR166k-166h precursor was extracted from the NCBI database and the transcription start site (TSS) was identified by using the TSSP Softberry program for identifying TSS in plants<sup>4</sup> . Cis-acting elements present in the 1.6 Kb DNA region upstream of the TSS were searched in the PLACE database<sup>5</sup> . The MIR166k-166h promoter was found to contain an important number of cis-elements required for response to pathogen infection or elicitor treatment (Supplementary Figure S4 and Supplementary Table S3). We identified several W-boxes (TGAC core sequences), such as WBOXATNPR1 (TTGAC), elicitor responsive element (ERE; TTCAGG), WRKY710S

<sup>4</sup>http://softberry.com/

<sup>5</sup>http://www.dna.affrc.go.jp/PLACE

(TGAC), WBOXNTERF3 (TGACY), and ASF1 (TGACG) cis-elements (Supplementary Figure S4). These regulatory ciselements are the binding sites for salicylic acid-induced WRKY transcription factors and are also found in many pathogenand elicitor-responsive genes. The SEBF regulatory element (SEBFCONSSTPR10A, YTGTCWC), initially characterized in the promoter of the pathogen and elicitor inducible potato PR-10a gene and later in the promoter of several other PR genes, was also identified in the MIR166k-166h promoter. Other functional pathogen/elicitor-responsive elements identified were the GT1- SCAM4 (GAAAAA) and PAL-responsive (CCGTCC) elements. Finally, regulatory elements associated with defense-related hormone signaling also present in the MIR166k-166h promoter included the ethylene (ERELEE4, ethylene-responsive element; AWTTCAAA) and methyl jasmonic acid (T/G BOXPIN2, AACGTG) regulatory elements. Although pathogen/hormoneresponsive cis-elements were identified in the MIR166k-166h promoter, their functionality in controlling MIR166k-166h expression remains unknown.

# MIR166k-166h Expression in Resistant and Susceptible Rice Varieties

We examined MIR166k-166h expression in rice varieties showing a phenotype of disease resistance against the rice blast fungus: Kanto 51, Saber and TeQing (resistant varieties), and Vialone Nano and Maratelli (susceptible varieties). The resistant genotypes are characterized by the presence of the resistance (R) genes: Pik in Kanto51, and Pib in Saber and TeQing (Tacconi et al., 2010). The basal level of expression varied among the different rice varieties (**Figure 3**). At 72 hpi with M. oryzae. MIR166k-166h expression was strongly induced in the three resistant rice genotypes here assayed, whereas its expression was barely affected or was even decreased by M. oryzae infection in the susceptible cultivars Vialone Nano and Maratelli (**Figure 3**). Thus, induction of MIR166k-166h expression appears to occur in resistant but not susceptible rice cultivars.

# Prediction and Experimental Validation of a Novel Target for miR166

As previously mentioned, HD-ZIP III genes are conserved target genes for miR166 in plants. In monocistronic miR166s, the mature miR166 sequences that direct cleavage of HD-ZIP III transcripts are located at the 3<sup>0</sup> arm of the precursor structure, namely miR166h-3p and miR166k-3p. In rice, five HD-ZIP III genes have been described: Oshox9 (Os10g33960), Oshox10 (Os03g01890), Oshox29 (Os01g10320), Oshox32 (Os03g43930) and Oshox33 (Os12g41860) (Agalou et al., 2008). Degradation tags indicative of miR166-mediated cleavage of Oshox9, Oshox10, Oshox32, and Oshox33 were identified by degradome analysis, which supports that they are real targets of rice miR166s (Li et al., 2010; Baldrich et al., 2015). In addition, RT-qPCR analysis revealed reduced levels of Oshox9, Oshox10, and Oshox32 in miR166k-166h-Ac mutant versus wild-type plants (**Figure 4A**),

which confirms the functionality of mature miRNAs encoded by the polycistron. As for Oshox29 and Oshox33, these genes were found expressed at very low levels in wild-type plants, and their expression was not significantly affected in miR166k-166h-Ac mutant plants as compared with wild-type plants (**Figure 4A**).

Knowing that MIR166k-166h activation has an impact on blast resistance, we considered the possibility that this phenotype might be caused by the activity of miR166 species encoded in the miR166k-166h precursor on novel, non-conserved target genes. We performed a target prediction analysis by using the psRNATarget tool<sup>6</sup> . Similar to other species, the target search predicted HD-ZIP III as target genes of miR166s encoded in the miR166k-166h polycistron (miR166k-3p and miR166h-3p). This computational prediction identified a putative target gene for the miR166k-5p sequence, the EIN2 gene. As for miR166h-5p, a possible binding site for this miRNA in a ferredoxin-nitritereductase gene was predicted.

A function for EIN2 as mediator of ethylene-dependent defense responses in plants is well established (Iwai et al., 2006; Helliwell et al., 2013, 2016; Yang et al., 2017). Accordingly, in this work we investigated whether EIN2 is a target gene for miR166k-5p. In contrast to Arabidopsis, in which EIN2 is encoded by a single gene, the rice genome possesses four EIN2 genes: OsEIN2.1 (also named MHZ7; Os07g06130), OsEIN2.2 (Os03g49400), OsEIN2.3 (Os07g06300), and OsEIN2.4 (Os07g06190) (Ma et al., 2013; Yang et al., 2015). Based on sequence homology, OsEIN2 genes can be classified into two groups, the first comprising OsEIN2.1 and OsEIN2.2 and the second OsEIN2.3 and OsEIN2.4. The four OsEIN2 genes have the binding site for miR166k-5p (Supplementary Figure S5A). RT-qPCR analysis was used to quantify OsEIN2.1, OsEIN2.2 and OsEIN2.3/4 expression in wild-type (TN67) and mutant plants (with the high sequence homology between OsEIN2.3 and OsEIN2.4, we could not design PCR-specific primers for these genes). Location of the primers used for detection of EIN2.1, EIN2.2 and EIN2.3/2.4 is shown in Supplementary Figure S5B. This analysis revealed downregulation of OsEIN2.1 and OsEIN2.2 in miR166k-166h-Ac plants (**Figure 4B**, left panel). The observed inverse correlation between mature miR166k-5p levels and EIN2.1 and EIN2.2 transcripts in miR166k-166h-Ac plants already indicated a possible miR166k-5p-mediated downregulation of this particular OsEIN2 family members. Intriguingly, OsEIN2.3/4 transcripts accumulated to a higher level in miR166k-166h-Ac mutant than wild-type plants. The amount of uncleaved OsEIN2 transcripts was determined by using PCR primers flanking the miR166k-5p cleavage site. Although the accumulation of uncleaved EIN2 transcripts was notably reduced in the activation mutant, uncleaved transcripts still accumulated to an important level in these plants, likely due to the contribution of EIN2.3/EIN2.4 transcripts (**Figure 4B**, right panel). This observation suggests the existence of complex regulatory mechanisms governing the expression of OsEIN2 in rice in which downregulation of OsEIN2.1 and OsEIN2.2 expression is accompanied by an increase in EIN2.3/EIN2.4 transcripts.

These observations prompted us to further investigate whether EIN2 gene is a real target gene for miR166k-166h by performing RNA ligase-mediated 5<sup>0</sup> RACE (5<sup>0</sup> -RACE). Sequencing of the 5<sup>0</sup> -RACE PCR clones revealed cleavage

<sup>6</sup>http://plantgrn.noble.org/psRNATarget

fragments of EIN2.1 transcripts (**Figure 4C**, left panels). Transcripts were found cleaved at the canonical position of miRNA/target mRNA pairing (between nucleotides 10 and 11 from the 5<sup>0</sup> end of the miRNA), which supports that EIN2 is indeed a target gene for miR166 in rice. As well, cleavage products of EIN2.1 accumulated to a lower level in wild-type than miR166k-166h-Ac plants (**Figure 4C**, right panel). As a control, miR166-guided cleavage products of hox32 were also identified by 5'-RACE (**Figure 4D**). Altogether, these results demonstrated that miR166 cleaves EIN2.1 transcripts and that the miR166k-5p strand in the miR166k-166h precursor is functional.

# miR166k-166h Mediates Cleavage of EIN2 Transcripts That Reduce Levels of EIN2 Protein

Among the newly identified miR166 targets, EIN2 is worth describing specifically. This gene is a central signal transducer in the ethylene signaling pathway in plants, and ethylene signaling is known to modulate plant immune responses (Solano and Ecker, 1998; Jun et al., 2004; Denancé et al., 2013; Ma et al., 2013).

To further investigate the functional relationship between miR166h-166k activity and EIN2 expression, we performed agroinfiltration experiments in N. benthamiana leaves in which the miR166k-166h precursor and a GFP-tagged EIN2.1 gene were co-expressed. As controls, constructs bearing the empty vector, the miR166k-166h precursor alone, or the EIN2.1- GFP chimeric gene alone were also assayed in agroinfiltration experiments. RT-PCR analysis revealed the accumulation of precursor miR166k-166h transcripts in agroinfiltrated leaves (**Figure 5A**, left panel). Accumulation of mature miR166 sequences derived from this precursor was confirmed by ST-RTqPCR and Northern blot analyses (**Figure 5A** and Supplementary Figure S6). These analyses indicated that the miR166k-166h precursor is properly expressed and processed in N. benthamiana leaves when expressed alone or with EIN2.1-GFP. However, levels of miR166k-166h transcripts were higher in miR166k-166h-only agroinfiltrated leaves versus leaves in which the miR166k-166h precursor was co-expressed with EIN2.1 (**Figure 5A**, left panel, pre-miR166 and pre-miR166+EIN2), an aspect that deserves further investigation.

When examining the transcript accumulation of EIN2.1, coexpression of the miR166k-166h precursor with EIN2.1-GFP reduced the EIN2.1-GFP transcript level as compared with expression of EIN2.1-GFP alone (**Figure 5B**, left panel, EIN2 and pre-miR166+EIN2). The observed reduction in EIN2.1-GFP transcripts was accompanied by a reduced EIN2-GFP protein level, as revealed by immunoblotting of protein extracts with an anti-GFP antibody (**Figure 5B**, right panel). From these results, we conclude that miR166k-166h targets and cleaves OsEIN2.1 and that cleavage of OsEIN2.1 transcripts reduces EIN2 protein accumulation.

Finally, knowing that MIR166k-166h expression is upregulated during M. oryzae infection in wild-type (cv TN67) plants (**Figure 2A**), and that OsEIN2.1 is a target gene for miR166, we investigated the expression of OsEIN2 family members during pathogen infection. OsEIN2.1 and OsEIN2.2 were downregulated during M. oryzae infection (**Figure 5C**), which is consistent with the observed increase in miR166k-5p level in the same tissues. In contrast, OsEIN2.3/4 expression was upregulated during pathogen infection. Presumably, the increased level of miR166k-5p in M. oryzae-infected leaves would be responsible for downregulation of OsEIN2.1 during pathogen infection.

# Expression of Ethylene Signaling Components in miR166k-166h-Ac Plants

In the absence of ethylene, phosphorylation of EIN2 prevents transduction of ethylene signaling. However, in the presence of ethylene, EIN2 phosphorylation is reduced and the C-terminal fragment of EIN2 is cleaved and translocated to the nucleus where the downstream EIN3 and EIL1 transcriptional cascade is activated. In addition, EIN2 and EIN3/EIL1 are regulated by proteasomal degradation via EIN3-binding F-box protein 1 and 2 (EBF1/2). Then, EIN3 and EIL1 regulate the expression of ethylene-responsive genes including Ethylene Response Factor 1 (ERF1) which, in turn, modulates the expression of various ethylene-responsive genes such as PDF1.2 and chitinase genes (Lorenzo et al., 2003; Abiri et al., 2017). It is generally assumed that EIN2 functions as a positive regulator of ethylene signaling, as revealed by repression of ethyleneinducible defense genes in ein2 antisense rice plants (Jun et al., 2004). The construct used to obtain ein2 antisense rice plants covered a 638-bp DNA fragment of the EIN2.1 cDNA encompassing the C-terminal region of EIN2, a region with high sequence conservation among OsEIN2 family members. Thus, silencing of all four OsEIN2 genes is expected to occur in the ein2 antisense plants previously described (Jun et al., 2004).

Accumulating evidence also indicates that ethylene signaling is required in rice for basal resistance against the blast fungus M. oryzae (Singh et al., 2004; Iwai et al., 2006; Helliwell et al., 2013, 2016; Yang et al., 2017). Thus, the observed increase in miR166k-166h accumulation and concomitant downregulation of OsEIN2.1 and OsEIN2.2 expression in miR166k-166h-Ac plants (**Figures 1**, **4**, respectively) apparently contradicts OsEIN2 positively regulating ethylene signaling in the rice response to M. oryzae infection.

To address the apparent contradiction of downregulation of OsEIN2 expression in miR166h-166k-Ac plants, showing blast resistance, we investigated the expression of genes acting downstream of EIN2 in the ethylene signaling pathway in mutant plants. OsEIN3 and OsEIL1, as well as OsERF1, were upregulated in miR166k-166h-Ac plants as compared with wild-type plants, whereas OsEBF2 expression was downregulated (**Figure 6A**). Consistent with upregulation of OsERF1, the expression of ethylene-responsive defense genes, such as PDF1.2 and chitinase genes (e.g., CHIT8 and CHIT14, members of the PR3 family of PR genes; and WIP5, a PR4 family member) was also upregulated in miR166k-166h-Ac plants (**Figure 6B**). These data indicate that although miR166k-166h activation downregulates OsEIN2.1 and OsEIN2.2, components in the pathway for ethylene signal transduction downstream of OsEIN2 are induced in miR166k-166h-Ac plants, which would agree with the resistance phenotype

that is observed in miR166k-166h-Ac mutant plants. Knowing that OsEIN2.3/2.4 is activated in the miR166k-166h-Ac mutant (see **Figure 4B**), OsEIN2.3/2.4 activation is likely responsible for the observed induction of downstream components of ethylene signaling in these plants, including ethylene-regulated defense genes.

To provide additional clues for the function of miR166k-166h in rice immunity, we investigated whether MIR166k-166h expression itself is regulated by ethylene in wild-type plants. For this, wild-type plants were treated with the ethylene precursor ACC, and the accumulation of miR166k-166h precursor transcripts was determined at different times after ACC treatment (15 min, 1, 4, and 24 h). ACC treatment resulted in a clear and gradual increase in the accumulation of miR166k-166h precursor transcripts in wild-type plants (**Figure 6C**).

Finally, expression analysis were performed to determine the accumulation of EIN2 transcripts in wild-type in response to ACC treatment. Consistent with up-regulation of MIR166k-166h in response to ACC treatment, EIN2.1 and EIN2.2 were found to be down-regulated during the same period of treatment (**Figure 6D**). However, EIN2.3/2.4 transcripts accumulated at a higher level in ACC-treated plants compared to control plants (**Figure 6D**). Thus, a different trend in the regulation of EIN2 family members occurs in response to ACC treatment which correlates with differences previously observed between miR166k-166h-Ac mutant plants and wild-type plants (see **Figure 4B**).

# DISCUSSION

In this work, we provide evidence that the polycistronic miR166k-166h plays a role in rice immunity. Thus, activation of MIR166k-166h in miR166k-166h-Ac plants, and concomitant increase in mature miR166s derived from the miR116h-166k precursor, enhances resistance to infection by hemibiotrophic (M. oryzae) and necrotrophic (F. fujikuroi) fungal pathogens (Ou, 1985; Wilson and Talbot, 2009; Campos-Soriano et al., 2013). Resistance to M. oryzae infection in miR166k-166h-Ac plants is associated with a stronger induction of defense gene expression, at both the biotrophic (24–48 hpi) and necrotrophic (72 hpi) stages of the infection. In wild-type plants, miR166k-166h accumulation was increased during pathogen infection and also in response to treatment with fungal elicitors, which supports that MIR166k-166h is a component of PTI. The observation that MIR166k-166h expression is activated in resistant rice cultivars but not in susceptible varieties (72 hpi with M. oryzae spores) further supports the role of MIR166k-166h in the rice response to the rice blast fungus. A more detailed analysis is, however, needed to examine the expression kinetics of MIR166k-166h in resistant and susceptible rice varieties during the infection process.

Of note, MIR166k-166h is found in the genome of both japonica and indica subspecies of the O. sativa genus (AA genome) (Baldrich et al., 2016). The MIR166k-166h locus is also present in the genome of wild relatives of current cultivated rice varieties, namely O. rufipogon and O. nivara (wild relatives of O. sativa), and O. barthii (wild relative of O. glaberrima, or African rice). These observations support conservation of the miR166k-166h polycistron in the Oryza genus (Baldrich et al., 2016). miR166 clusters have been identified in the genome of several plant species (e.g., M. truncatula, soybean and Physcomitrella patens) but, in most cases, the polycistronic nature of these miR166 clusters has not been demonstrated (Boualem et al., 2008; Zhang et al., 2009; Barik et al., 2014; Li et al., 2017).

Our evidence supports that EIN2 is a novel target gene for miR166, this gene being targeted by miR166k-5p in the MIR166k-166h polycistron. Supporting this conclusion, we found opposite expression patterns of miR166k-5p and OsEIN2.1 in miR166k-166h-Ac mutant plants. Also, miR166k-5p and its target gene showed opposite expression patterns in response to fungal infection (upregulation and downregulation, respectively). Definitive proof of a miR166k-5p-mediated cleavage of EIN2.1 transcripts came from 5<sup>0</sup> -RACE analyses and agroinfiltration experiments in N. benthamiana leaves. The observed miR166 guided cleavage of EIN2.1 transcripts was accompanied by reduced EIN2 protein level. From these results, we conclude that EIN2.1 represents a novel target gene for miR166k-5p species encoded by the polycistronic miR166k-166h precursor.

Clearly, the existence of multiple miR166 family members might contribute to diversification and functional specialization of miR166 in plants. In line with this, miR166b has been reported to target rice RDD1 (rice Dof daily fluctuations 1), a non-HD-ZIP III transcription factor involved in nutrient uptake and accumulation (Iwamoto and Tagiri, 2016). Very recently, miR166-guided cleavage of ATHB14-LIKE transcripts encoding a homeobox-leucine zipper protein has been described in soybean (Li et al., 2017). In M. truncatula, a miR166 polycistron containing two copies of miR166a targeting HD-ZIP III transcripts was found to control root architecture and nodule development after infection by Sinorhizobium meliloti (Boualem et al., 2008). Presumably, mature miRNAs encoded by the miR166k-166h polycistron might have evolved to mediate rice defense responses to pathogen infection.

When considering the mature miR166s encoded by the miR166k-166h precursor, we noticed that miR166 species targeting OsEIN2.1 correspond to miR166-5p in monocistronic miR166s, while miR166-3p sequences target hox genes. Hence, the two strands of the miR166k duplex in the miR166k-166h precursor appear to be functional. There are other examples in which the two strands of a miRNA are functional, as for miR393 in Arabidopsis. Here, the miR393 strand guides cleavage of transcripts encoding auxin receptor genes (TIR1, AFB2, AFB3), and the miR393-3p strand cleaves MEMB12 transcripts encoding a SNARE protein involved in exocytosis of the PR1 protein (Zhang et al., 2011). Degradome analysis revealed miR166e-3p and miR166h-5p-mediated events for genes involved in the arbuscular mycorrhizal symbiosis in Medicago truncatula (e.g., Sucrose synthase, Tyr protein kinase and protein phosphatase 2C) (Devers et al., 2011). In addition to being represented by multiple copies in the rice genome, the ability of miR166 precursors to produce two mature functional strands in the same miRNA-5p/miRNA-3p duplex also represents an effective strategy to diversify miR166 function.

Our results indicate that MIR166k-166h activation enhances defense gene expression, most probably by modulating OsEIN2 expression. An intriguing aspect of this study was the finding of a different trend in the regulation of OsEIN2 expression in miR166k-166h-Ac plants depending on the family member. Whereas EIN2.1 and EIN2.2 are downregulated in the rice mutant

(the two genes being more closely related to one another than either EIN2.3 or EIN2.4), EIN2.3 and EIN2.4 are upregulated in these plants. Additional regulatory forces controlling the abundance of EIN2 transcripts must then exist. Several possibilities can be considered to explain the finding of OsEIN2.1 and OsEIN2.2 being downregulated in miR166k-166h-Ac plants and OsEIN2.3/2.4 upregulated in this mutant. They include the existence of regulatory mechanisms in which miR166k-5p and EIN2 family members regulate each other's expression, or interconnecting networks controlling the expression of OsEIN2 family members themselves (i.e., the abundance of a particular OsEIN2 gene might affect the level of another OsEIN2 family member). Previous studies in Arabidopsis demonstrated crossregulation among transcription factor family members targeted by miRNAs (i.e., regulation of GROWTH REGULATING FACTORS by miR396 species) (Hewezi and Baum, 2012). Crossregulation of auxin response factors (ARFs) has been also described, where ARF6 and ARF8 (targets of miR167) and ARF17 (targets of miR160) regulate each other's expression at both transcriptional and posttranscriptional levels by modulating miR160 and miR167 availability (Gutierrez et al., 2009). The possibility of miR166k-5p-mediated translational repression of EIN2 family members should not be ruled out. If such regulatory mechanisms operate in rice, this would represent an additional layer of regulation of OsEIN2 expression, which would help in maintaining appropriate OsEIN2 levels, rather than completely turning off OsEIN2 expression to allow optimal expression of defense responses with no negative impact on plant growth.

A working model of the role of miR166k-166h in governing expression of ethylene-regulated defense genes is in **Figure 7**. According to this model, pathogen recognition triggers ethylene biosynthesis and activation of MIR166k-166h expression, which in turn would regulate components of the ethylene signaling pathway leading to induction of ethylene-regulated defense genes (PDF1.2, chitinases). We propose an interlocking regulation mechanism governing the expression of OsEIN2 family members and downstream signaling components leading to activation of defense gene expression. Further studies are required to determine the interlocking mechanisms among OsEIN2 family members and among miR166k-miR166h and EIN2.

Basal resistance to M. oryzae has been reported to require activation of ethylene biosynthesis and signaling networks during the biotrophic phase of the infection process (Singh et al., 2004; Iwai et al., 2006; Helliwell et al., 2013, 2016; Yang et al., 2017). However, the mechanisms by which the pathogen induces ethylene biosynthesis remain unknown. Because MIR166k-166h expression is itself regulated by treatment with the ethylene precursor ACC (**Figure 6**), the M. oryzae-induced production of ethylene might induce MIR166k-166h expression. Furthermore, the M. oryzae-mediated ethylene accumulation has been found to affect JA signaling (Yang et al., 2017). Whether defense hormone networking is altered in miR166k-166k-Ac plants deserves further investigation.

It is also known that ethylene has antagonistic effects in controlling the rice defense response depending on the pathogen lifestyle and also on the type of pathogen. Whereas the accumulation of ethylene appears to be required for resistance against M. oryzae (Iwai et al., 2006), repression of ethylene signaling has been shown to enhance resistance against the necrotrophic rice brown spot fungus Cochliobolus miyabeanus (Vleesschauwer et al., 2010). A major future challenge is to determine the molecular processes by which MIR166k-166h function is integrated in the complex regulatory mechanisms involved in ethylene-regulated immune responses to M. oryzae infection and whether activation of MIR166h-166k expression confers resistance to pathogens other than M. oryzae.

Besides playing a role in plant responses to pathogen infection, ethylene is considered a phytohormone involved in regulation of plant growth and development. Because excessive ethylene production under pathogen infection might negatively affect plant development, the host plant must then maintain a tight control of ethylene homeostasis to cope with pathogenic infections with no growth penalty. In this respect, negative feedback mechanisms have been proposed to coordinate plant growth and ethylene/salinity responses (Tao et al., 2015).

Given the well-established roles of miR166 and its HD-ZIP III target genes in controlling developmental processes in a broad range of plant species, an intriguing question is why MIR166k-166h activation does not affect normal growth in the miR166k-166h mutant. A possible threshold of miR166k-166h level (and subsequent miR166-regulated Oshox transcripts) might explain this observation. Under heterozygosity, the miR166k-166h-Ac mutant plants would not accumulate sufficient levels

FIGURE 7 | Proposed model for the function of miR166k-166h in ethylene signaling during infection of rice plants by the blast fungus M. oryzae. In the absence of ethylene, active ethylene receptors negatively regulate OsEIN2 via phosphorylation, thus repressing the downstream signaling transduction. Pathogen recognition would trigger ethylene biosynthesis, which is perceived by its receptors. Upon ethylene perception, the EIN2 phosphorylation is reduced and the carboxy-terminal fragment of EIN2 is cleaved and translocated to the nucleus for activation of EIN3/ELI1 and Ethylene Response Factor 1 (ERF1), thereby activating defense gene expression. Pathogen-induced ethylene production would also induce MIR166k-166h expression, which would then regulate the expression of OsEIN2 family members (downregulation of OsEIN1.2 and OsEIN2.2; upregulation of OsEIN2.3/2.4). MIR166k-166h activation in miR166k-166h mutant plants would mimic the activation of ethylene signaling pathways induced by M. oryzae infection in the host plant. Arrows and blunt ends indicate positive and negative regulation, respectively. Arrows with broken lines indicate still unknown interlocked regulatory mechanisms among EIN2 family members.

of miR166kh species to alter normal developmental programs due to excessive downregulation of miR166 HD-ZIP III target genes. Moderate levels of mature miR166s produced by the miR166k-166h polycistron would provide a means to mount a more successful defense response without no penalty on normal development.

The functional significance of the organization of miRNAs as polycistrons is still debated. Polycistronic transcription can finetune gene expression in related or unrelated biological processes (e.g., defense responses and developmental processes). A single promoter drives the expression of polycistronic miRNAs, which allows for the expression of multiple miRNAs in a coordinated spatial and/or temporal manner.

# CONCLUSION

Our results support that miR166k-166h is a positive regulator of rice immunity via regulation of OsEIN2. A better knowledge of miR166k-166h functioning in blast resistance will help in deciphering the functional consequences of polycistronic expression of miRNAs in plants. Because blast is one of the primary causes of rice losses worldwide, unraveling miR166k-166h-mediated mechanisms underlying blast resistance could ultimately help in designing novel strategies for crop protection.

# AUTHOR CONTRIBUTIONS

RS-G performed the experiments and analyzed the data. Y-iH and BSS designed and conceived the work. All the authors contributed to the manuscript writing.

# REFERENCES


# FUNDING

This work was supported by grants from the Spanish Ministry of Economy and Competitiveness [BIO2012-32838, BIO2015-67212-R] and the CSIC/NSC (Spanish Research Council/National Science Council of Taiwan)-Cooperative Research Project-Formosa Program (2009TW0041). We also acknowledge financial support from the CERCA Program from the Generalitat de Catalunya, and MINECO through the "Severo Ochoa Program for Centers of Excellence in R&D" 2016-2019 [SEV-2015-0533]". RS-G is a recipient of a Ph.D. grant from the Spanish Ministry of Economy and Competitiveness (BES-2013- 065521).

# ACKNOWLEDGMENTS

We thank Dr. J-S Zhang (CAS, Beijing) for the EIN2 cDNA, Dr. D. Baulcombe (University of Cambridge, United Kingdom) for the N. benthamiana RDR6-IR line, and Drs. E. Lupotto and G. Valé for resistant and susceptible rice cultivars. We also thank N. Fernández for assistance with parts of this work.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00337/ full#supplementary-material


pathogen infection. Mol. Plant Pathol. 13, 579–592. doi: 10.1111/J.1364-3703. 2011.00773.X


binding site–leucine-rich repeats and other mRNAs. Plant Cell 24, 859–874. doi: 10.1105/tpc.111.095380


**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 Salvador-Guirao, Hsing and San Segundo. 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.

# Mutation in Rice Abscisic Acid2 Results in Cell Death, Enhanced Disease-Resistance, Altered Seed Dormancy and Development

Yongxiang Liao<sup>1</sup>† , Que Bai<sup>1</sup>† , Peizhou Xu<sup>1</sup>† , Tingkai Wu<sup>1</sup> , Daiming Guo<sup>1</sup> , Yongbin Peng<sup>1</sup> , Hongyu Zhang<sup>1</sup> , Xiaoshu Deng<sup>1</sup> , Xiaoqiong Chen<sup>1</sup> , Ming Luo<sup>2</sup> , Asif Ali<sup>1</sup> , Wenming Wang<sup>1</sup> \* and Xianjun Wu<sup>1</sup> \*

<sup>1</sup> Rice Research Institute, Sichuan Agricultural University, Sichuan, China, <sup>2</sup> Agriculture and Food, Commonwealth Scientific and Industrial Research Organization (CSIRO), Canberra, ACT, Australia

#### Edited by:

Zhengqing Fu, University of South Carolina, United States

#### Reviewed by:

Aardra Kachroo, University of Kentucky, United States Marina Gavilanes-Ruiz, Universidad Nacional Autónoma de México, Mexico

#### \*Correspondence:

Wenming Wang j316wenmingwang@163.com Xianjun Wu wuxj@sicau.edu.cn

†These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 09 November 2017 Accepted: 14 March 2018 Published: 28 March 2018

#### Citation:

Liao Y, Bai Q, Xu P, Wu T, Guo D, Peng Y, Zhang H, Deng X, Chen X, Luo M, Ali A, Wang W and Wu X (2018) Mutation in Rice Abscisic Acid2 Results in Cell Death, Enhanced Disease-Resistance, Altered Seed Dormancy and Development. Front. Plant Sci. 9:405. doi: 10.3389/fpls.2018.00405 Lesion mimic mutants display spontaneous cell death, and thus are valuable for understanding the molecular mechanism of cell death and disease resistance. Although a lot of such mutants have been characterized in rice, the relationship between lesion formation and abscisic acid (ABA) synthesis pathway is not reported. In the present study, we identified a rice mutant, lesion mimic mutant 9150 (lmm9150), exhibiting spontaneous cell death, pre-harvest sprouting, enhanced growth, and resistance to rice bacterial and blast diseases. Cell death in the mutant was accompanied with excessive accumulation of H2O2. Enhanced disease resistance was associated with cell death and upregulation of defense-related genes. Map-based cloning identified a G-to-A point mutation resulting in a D-to-N substitution at the amino acid position 110 of OsABA2 (LOC\_Os03g59610) in lmm9150. Knock-out of OsABA2 through CRISPR/Cas9 led to phenotypes similar to those of lmm9150. Consistent with the function of OsABA2 in ABA biosynthesis, ABA level in the lmm9150 mutant was significantly reduced. Moreover, exogenous application of ABA could rescue all the mutant phenotypes of lmm9150. Taken together, our data linked ABA deficiency to cell death and provided insight into the role of ABA in rice disease resistance.

Keywords: abscisic acid, xanthoxin dehydrogenase, gibberellin, ABA/GA ratio, lesion mimic mutant, pre-harvest sprouting, defense responses, Oryza sativa L.

# INTRODUCTION

Abscisic acid (ABA) is one of the multi-functional phytohormones that is involved in many essential physiological processes during growth and development in plants, such as seed maturation, seed desiccation, seed dormancy, germination, and stress-induced responses (Rock and Quatrano, 1995; Leung and Giraudat, 1998; Maia et al., 2014). In rice, ABA modulates seed dormancy mainly through the balance of ABA and GA (ABA/GA ratio) (Liu et al., 2010, 2014; Shu et al., 2013). In different plant–pathogen interaction systems, it is well-established that ABA negatively regulates plant defense. On one hand, ABA-deficient mutants, such as abscisic acid2-1 (aba2-1) and aba3-1 in Arabidopsis, sitiens (sit) in tomato, display enhanced resistance to Golovinomyces cichoracearum and Erwinia chrysanthemi, respectively (Asselbergh et al., 2008;

**45**

Xiao et al., 2017). Conversely, ABA-increased mutants in Arabidopsis become more susceptible (Gao et al., 2016). On the other hand, exogenous application of ABA increases susceptibility to different pathogens in rice, Arabidopsis and tomato (Asselbergh et al., 2008; Xu et al., 2013; Xiao et al., 2017). ABA-treatment on rice plants leads to enhanced disease symptoms on both susceptible and resistant accessions (Koga et al., 2004; Jiang et al., 2010). However, to our knowledge there is no report of an ABA-deficient mutant associated with disease resistance in rice.

The biosynthesis of ABA has two pathways: (1) the direct pathway via mevalonate pathway in Fungi (Hirai et al., 2000; Inomata et al., 2004) and (2) the indirect pathway via carotenoid pathway that is the main biosynthesis pathway in higher plants. The indirect pathway contains three stages (Milborrow, 2001; Schwartz et al., 2003). First, zeaxanthin is synthesized by zeaxanthin epoxidase (ZEP) in plastids (Marin et al., 1996; Agrawal et al., 2001). Second, zeaxanthin is converted into 9-cisviolaxanthin and 9-cis-neoxanthin by neoxanthin synthase that is encoded by ATABA4 (North et al., 2007). They are dissociated to xanthoxin (Xan) by 9-cis-epoxycarotenoid dioxygenase (NCED) (Iuchi et al., 2001; Riahi et al., 2013; Zhang et al., 2014). Third, Xan is transferred to the cytosol and converted to abscisyl aldehyde (ABAld) by xanthoxin dehydrogenase (XanDH) (Miguel et al., 2002; Endo et al., 2014). Then, ABAld is catalyzed into ABA by Abscisic Aldehyde Oxidase (AAO) (Seo et al., 2000).

XanDH belongs to a short-chain dehydrogenase/reductase (SDR) superfamily that is encoded by ABA2. AtABA2 is a single copy gene in Arabidopsis thaliana and is constitutively expressed. Expression of AtABA2 activates the glucose signal, antagonizes the ethylene signal and promotes the synthesis of ABA (Cheng et al., 2002; Miguel et al., 2002). ZmABA2 is the homolog of AtABA2 in maize. ZmABA2 interacts with ZmMPK5 and coordinately regulates the ABA level in maize (Ma et al., 2016). OsABA2, a rice ABA2 homolog, could restore the Arabidopsis ataba2 mutant phenotypes (Endo et al., 2014), indicating the conservation of ABA biosynthesis function of OsABA2. However, the phenotypic changes in rice resulted from mutation in OsABA2 has not been described.

Previous studies have described mutants that have disrupted functions of genes in ABA biosynthetic pathway in different plants, such as sitiens (sit) and notalilis (not) in tomato (Burbidge et al., 1999; Okamoto et al., 2002), aba1, aba2, and aba3 in Arabidopsis (Léon-Kloosterziel et al., 1996; Schwartz et al., 1997a; Xiong et al., 2002) and pre-harvest sprouting (phs) and viviparous (vp) mutants (vp10, vp13, and vp14) in maize (Tan and Mccarty, 1997; Schwartz et al., 1997b, 2003). The grains of those phs mutants easily germinate in the ear or panicle before harvest under wet conditions. In rice, the osaba1 mutant has a mutation in the gene encoding zeaxanthin epoxidase that involves in ABA synthesis in rice (Agrawal et al., 2001). The genes of phytoene desaturase (OsPSD), ζ-carotene desaturase (OsZDS), carotenoid isomerase (OsCRTISO), and lycopene β-cyclase (β-OsLCY) encode essential enzymes in different steps of the carotenoid synthetic pathway and the disruption of carotenoid biosynthesis leads to PHS trait (Fang et al., 2008). How other genes involve in ABA biosynthesis modulate pre-harvest sprouting remains to be tested.

The lesion mimic mutants (LMMs) display cell death, similar to hypersensitive response (HR), and are highly valuable in the investigation of the molecular mechanism of programmed cell death and defense responses (Greenberg, 1997; Moeder and Yoshioka, 2008; Tamiru et al., 2016). A number of LMMs have been isolated and characterized in Arabidopsis (Cao et al., 1994), maize (Hu et al., 1998), barley (Persson et al., 2009), and rice (Kiyosawa, 1970; Liu et al., 2017). In rice, more than 15 LMM-related genes have been isolated and characterized, which encode proteins of distinct functions, such as OsCUL3a that is the prominent component of cullin 3-based RING E3 ubiquitin ligases (Liu et al., 2017), an AAA-type ATPase (Fekih et al., 2015; Zhu et al., 2016), an eEF1A-like protein (Wang et al., 2017), a mitogen-activated protein kinase kinase kinase (Wang S. et al., 2015), a heat stress transcription factor protein (Utako et al., 2002), a cytochrome P450 monooxygenase (Tadashi et al., 2010), an U-box/armadillo repeat protein (Zeng et al., 2004), a clathrin-associated adaptor protein comlex1 (Qiao et al., 2009), a splicing factor 3b subunit 3 protein (Chen et al., 2012), and a thylakoid-bound protein (Li et al., 2010). Therefore, the molecular mechanism of lesion mimic formation in plant is regulated by a complicated regulatory network. Although a variety of LMM-related genes have been identified, whether ABA synthesis pathway is involved in lesion formation and how ABA modulates disease response in rice remains to be answered.

Here, a new rice lesion mimic mutant 9150, named lmm9150, was isolated and characterized. The lmm9150 mutant exhibited spontaneous cell death on leaves, pre-harvest sprouting under field condition, and enhanced growth of leaves and stems. The lesions in lmm9150 were associated with typical defense responses, such as accumulation of H2O<sup>2</sup> and enhanced resistance to bacterial blight and rice blast diseases. Using a map-based cloning approach, we detected a point mutation in LOC\_Os03g59610, which encodes OsABA2, a short-chain alcohol dehydrogenase. Knocking-out of OsABA2 with CRISPR/Cas9 led to lesion mimic spots, enhanced disease resistance and growth, and PHS traits. Consistence with the function of OsABA2 in ABA synthesis, the lesion mimic spots and other phenotypes of lmm9150 could be rescued by exogenous application of ABA. Therefore, our results provide further insights into the molecular function of OsABA2 in lesion mimic formation, disease-resistance, seed dormancy, and development in rice.

# MATERIALS AND METHODS

# Plant Materials and Growth Conditions

The lmm9150 mutant was generated by ethyl methane sulfonate (EMS) treatment of a Chinese indica cultivar Yixiang1B, which is one of elite backbone parents in hybrid rice breeding programs in China. The wild type (WT) Yixiang1B and lmm9150 were grown, respectively, in the paddy field in Chengdu city (N30.67◦E104.06◦ ), Sichuan Province and in Lingshui county (N18.47◦E110.04◦ ), Hainan Province, China.

## Histochemical Analysis

fpls-09-00405 March 26, 2018 Time: 15:31 # 3

Leaves from the lmm9150 mutant, with obvious lesion mimic spots, at the seedling stage and the WT at the same growth stage were collected for histochemical analysis. Trypan blue staining was performed to detect cell death as previously described (Zhu et al., 2016). In brief, samples were submerged in lactic acidphenol-trypan blue solution (2.5 mg/ml trypan blue, 25% (w/v) lactic acid, 23% water-saturated phenol, and 25% glycerol in H2O) and kept at room temperature for 6–12 h, followed by destaining with solution containing 30% (w/v) chloral hydrate for 5 days. Then, the samples were equilibrated with 50% glycerol for 1 day for taking photos. For detection of hydrogen peroxide (H2O2) accumulation, the 3,3<sup>0</sup> -diaminobenzidine (DAB) staining was performed as described previously (Zhu et al., 2016).

# Chlorophyll Content Measurement

Leaves were soaked in 20 ml of 95% alcohol for 48 h in dark, until the leaves became colorless. The absorbance values were examined by a spectrophotometer at 665 and 649 nm. Then, the content of chlorophylls was calculated as follows:

Chlorophyll (Chl) a absorbance value = 13.95 × OD665−6.88 × OD<sup>649</sup>

Chlorophyll (Chl) b absorbance value = 24.96 × OD649−7.32 × OD<sup>665</sup>

Chlorophyll content (mg/g) = (C × V × D)/1000 × W

C: chl a or chl b absorbance value; V: volume of extracting solution; D: dilution index; W: weight of sample.

Chlorophyll content was measured in three biological repeats. Statistical analysis was performed using Student's t-test.

# Inoculation of Pathogens and Disease Resistance Assay

The resistance assay to bacterial blight disease was performed as described previously. In brief, 70-days-old (at tillering stage) plants of the mutant lmm9150 and the WT were used for inoculation of the bacterial pathogen Xanthomonas oryzae pv. oryzae (Xoo). The Xoo strains, P6, P3, 8248, and X004, which are compatible with the WT, were used for inoculation. Xoo bacterial suspensions with 0.5 of OD<sup>600</sup> were used to inoculate by using the scissors-dip method (Zuo et al., 2014). Disease lesion lengths were measured at 15 days post-inoculation (dpi).

For resistance assay to rice blast disease, 60-days-old (at tillering stage) plants of the lmm9150 mutant and the WT were used for inoculation with the fungal pathogen Magnaporthe oryzae (M. oryzae). The M. oryzae strains, ZhongI, and Tetep, which are compatible with the WT, were used for inoculation following a previous report (Park et al., 2012). The disease lesions were observed at 5 dpi.

# DNA Extraction and PCR, RNA Extraction, and qRT-PCR

Genomic DNA was extracted from leaves using the cetyltrimethylammonium bromide (CTAB) method (Murray et al., 1980). The PCR mixture was mixed with 2 µl DNA (10–50 ng/µl), 2 µl primers (10 µmol/µl), 0.3 µl dNTPs (10 mmol/L), 0.2 µl Taq (5 U/µl), and 13.5 µl H2O. The running procedure of PCR was performed as the following: pre-denaturation at 95◦C for 5 min followed by 30 cycles of denaturation at 95◦C for 30 s, annealing at 56◦C for 30 s, extension at 72◦C for 1 min, with a final extension at 72◦C for 10 min. The PCR products were separated by electrophoresis in a 3% agarose gels, stained with ethidium bromide (EB) and photographed.

Total RNA was extracted using Trizol (Invitrogen, Carlsbad, CA, United States) following the procedures of the manufacturer. The mRNA was digested with DNase I according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, United States) and was subjected to reverse transcription to synthesize first-stand cDNA. Oligo (dT) was used as primer and SuperScript II (Invitrogen, Carlsbad, CA, United States) was used as reverse transcription enzyme.

The qRT-PCR was performed using a Bio-Rad CFX96 Real-Time System coupled to a C1000 Thermal Cycler (Bio-Rad, Hercules, CA, United States). The housekeeping gene Ubiquitin5 (Ubq5) was used as the internal control. The sequences of the primers were listed in **Supplementary Table S1**.

# Genetic Analysis and Map-Based Cloning

Two F<sup>1</sup> and two F<sup>2</sup> populations derived from the crosses of Yixiang1B × lmm9150 and lmm9150 × Yixiang1B were used for genetic analysis. The F<sup>2</sup> population derived from the cross of 02428 × lmm9150 was used for mapping of the mutant gene. For the bulk segregation analysis (BSA) (Michelmore et al., 1991), equal amount of leaf blades from 10 F<sup>2</sup> plants with the lesion mimic phenotype and 10 F<sup>2</sup> plants with the WT phenotype were collected for DNA extraction to construct the mutant and the WT DNA pools, respectively. The physical linkage map was then constructed using molecular markers nearby the lmm9150 gene.

The SSR primers were synthesized according to the information of Gramene database<sup>1</sup> . InDel markers were developed based on the alignment results of the reference 93–11, an indica rice<sup>2</sup> and the Nipponbare, a japonica rice<sup>2</sup> genome sequence at the candidate region. Primers were designed using Primer3 web version4.0.0<sup>3</sup> . The specificity of each primer in the rice genome was confirmed by BLAST<sup>4</sup> and PCR analysis. The sequences of SSR and InDel markers were listed in **Supplementary Table S2**.

For analysis of the PCR products, amplified products were separated by electrophoresis at 3.0% agarose gel in 0.5 × TBE buffer, and visualized and photographed under UV light.

For whole-genome re-sequencing, the lmm9150 was backcrossed with the WT and self-crossed to generate BC1F<sup>2</sup> population. The equal total DNA of 20 BC1F<sup>2</sup> plants with lesion mimic spots, was mixed and re-sequenced at Novogene

<sup>2</sup>http://rise2.genomics.org.cn/page/rice/index.jsp

<sup>1</sup>http://www.gramene.org/microsat

<sup>3</sup>http://primer3.ut.ee/

<sup>4</sup>http://ensembl.gramene.org/Tools/Blast?db=core

Corporation (Beijing, China). At the same time, the WT genome DNA was also sequenced as control.

# Sequence Analysis

fpls-09-00405 March 26, 2018 Time: 15:31 # 4

Gene prediction was performed using the Rice Genome Annotation Project database<sup>5</sup> . Sequence alignments were performed using the software DNAMAN Version 6.0. Alignments of amino acid were performed using the software Clustax2.

# Construction of Knockout Lines by CRISPR/Cas9

Vector construction was performed to knockout the candidate gene as previously described (Ma et al., 2015). In brief, two targets (**Supplementary Table S3**), named lmm9150-Y and lmm9150-B, were designed in the coding sequence of conservative region. The two target sequences were amplified by PCR. Using endonuclease, Eco311 and T4 ligase, the lmm9150-Y was inserted into the pBWA (V) H-cas9i2 and the lmm9150-B was inserted into the pBWD (LB) DNAi. Finally, Pbwa (V) H-cas9i2-lmm9150-Y and pBWD (LB) DNAilmm9150-B were assembled into the final vector pBWA (V) H-cas9i2-lmm9150 using endonuclease SapI and T4 ligase. The constructs were verified by sequencing and then introduced into the WT by Agrobacterium-mediated transformation as described previously (Jeon et al., 2000). The sequences near the editing region were verified by extracting genomic DNA from T1 transgenic plants (sequencing primer in **Supplementary Table S3**).

# ABA, GA, and Water Loss Assay

Determination of ABA and GA content was carried out using enzyme-linked immunosorbent assay (ELISA) method as previously described (Teng et al., 2006). To test the water-loss rate, seedlings were grown in a paddy field and leaves (300 mg from 10 seedlings) of 40-days-old seedlings were kept on dish at room temperature (28◦C). The weight of samples was measured at 20 min interval until 120 min.

# Germination Assay

Seeds from the lmm9150 mutant and the WT at 30 days after flowering were used for germination test. Seeds were cultured in a growth chamber at 30◦C,14 h light and 24◦C,10 h dark cycle for 8 days.

# RESULTS

# Isolation and Characterization of the lmm9150 Mutant

To explore the molecular connection of lesion mimic and disease resistance in rice, we carried out an extensive forward genetic screening for lesion mimic mutants from an EMS-mutagenized population of indica cultivar Yixiang1B. Subsequently, more than 20 mutants were identified. One of these mutants, named lmm9150, was chosen for further characterization because of its multiple phenotypes in development. The 3-week-old seedlings of the mutant lmm9150 begun to develop brown lesions on the top part of the third leaf in the paddy field in both Chengdu (N: 30.67◦ ) and Hainan (N: 18.47◦ ) (**Figure 1A**). At tillering stage, the apical part of older leaves also displayed lesion mimic spots (**Supplementary Figure S1A**). At mature stage, the spots

<sup>5</sup>http://rice.plantbiology.msu.edu/

emerged to the top part of flag leaf (**Supplementary Figure S1B**). In addition, lmm9150 mutant displayed obvious pre-harvest sprouting at 30 days after flowering under natural condition (**Figures 1B,C**).

However, apart from plant height and the flag leaf of lmm9150 that were significantly larger than WT (**Figures 1D,E**), the yield-related agronomic traits, including the number of tillers, seed-setting rate, grain weight per plant, and 1000-grains weight, had no significant difference from those of WT (**Supplementary Figures S1C–F**).

To characterize the lesions in lmm9150, we performed a trypan blue staining assay to examine cell death. Compared with WT, a lot of blue spots were observed on the leaf of lmm9150 after staining (**Figure 2B**), indicating existence of dead cells. Because cell death is often associated with the production of reactive oxygen species (ROS) such as H2O2, we further examined the accumulation of H2O<sup>2</sup> in lmm9150 by DAB staining assay. Our data indicated that DAB-stained was observed in the mutant but not WT leaf blade, whereas there were rarely observed in the WT (**Figure 2C**).

Next, we examined the content of chlorophylls. Small pieces from cognate positions of flag leaves of WT and lmm9150 at flowering stage were collected for measurement of chlorophylls. The results showed that the content of both chlorophyll a and chlorophyll b in lmm9150 was significantly lower than that of WT (**Figures 3A,B**), implying that the formation of lesion mimic spots might attributable to the decrease of photosynthesis pigments in the lmm9150 mutant.

In summary, this mutant began to exhibit lesions as red brown spots from the top part of the third leaf at seedling stage, which eventually extended to the flag leaf at mature stage. Cell death was detected in the mutant, which was accompanied with excessive accumulation of H2O2. In addition, the mutant also exhibited higher plant height, longer flag leaf, and pre-harvest sprouting of seeds at mature stage.

# Enhanced Resistance to Bacterial Blight and Rice Blast Pathogens by lmm9150

Many lesion mimic mutants exhibit enhanced defense responses in plants (Wang J. et al., 2015; Zhu et al., 2016). We speculated that the lmm9150 mutant may also become resistant to diseases in rice. To confirm this speculation, we first analyzed the expression of Pathogenesis-Related (PR) genes using quantitative reversetranscription PCR (qRT-PCR), including PR1a, PR1b, PR10 (Nahar et al., 2011; Sang et al., 2011). The results demonstrated that all the examined PR genes were significantly upregulated in lmm9150 (**Figure 4A**), implying that lmm9150 was a potential auto-immune mutant. Then, we tested disease resistance of lmm9150 by inoculating Xanthomonas oryzae pv. oryzae (Xoo) and Magnaporthe oryzae at tillering stage, respectively. The results showed that the inoculated lmm9150 leaves exhibited disease lesions obviously shorter than that of WT leaves at 15 days after inoculation of Xoo (**Figures 4B–E** and **Supplementary Figures S2A–C**), indicating enhanced resistance. Quantification of the lesion lengths on leaves revealed that the average disease lesion length in lmm9150 was significantly shorter than that of WT. Intriguingly, the lesions on flag leaves were also

FIGURE 2 | Observation of cell death in wild type (WT) and lmm9150 mutant. (A–C) Representative leaf sections of WT and lmm9150 mutant (MT) show lesions (A), dead cells revealed by trypan blue staining (B), and H2O<sup>2</sup> accumulation revealed by 3,3<sup>0</sup> -diaminobenzidine (DAB) staining (C) in the mutant, respectively. The lmm9150 mutant and WT leaf samples were collected at 30 days after sowing. Scale bar: 1 cm.

FIGURE 3 | Comparison of the chlorophylls content between wild type (WT) and lmm9150 mutant. (A,B) The content of Chlorophyll a (chl a) and chlorophyll b (chl b) shows reduction in lmm9150 in comparison with WT. Mean and SD were obtained from three measurements. Statistical analysis was performed using Student's t-test, <sup>∗</sup> and ∗∗ represents P < 0.05 and P < 0.01, respectively.

significantly longer than those on the second leaves in lmm9150 (**Figure 4C** and **Supplementary Figure S2D**). In addition, lesions caused by Magnaporthe oryzae on the lmm9150 leaves were remarkably smaller than those of WT leaves (**Figure 4D** and **Supplementary Figure S2E**), indicating enhanced resistance to rice blast disease.

To explore the possible hormone pathway of defense responses activated in lmm9150, we examined the expression of jasmonic acid (JA) and salicylic acid (SA)-related genes. The lipoxygenase2.1 (OsLOX2.1) and allene oxide synthase2 (OsAOS2) are key genes of JA synthesis pathway (Liu et al., 2012). The

phytoalexin deficient4 (OsPAD4) and lipase1 (OsEDS1) are key genes of SA signaling pathway (Pieterse et al., 2012). The results indicated that the expression of OsLOX2.1 and OsAOS2 was significantly higher in lmm9150 than that in WT (**Figure 4E**), indicating increase of JA synthesis. In contrast, the expression of OsPAD4 and OsEDS1 had marginal difference between lmm9150 and WT (**Supplementary Figure S2F**), implying that SA signaling pathway may not be activated in lmm9150. To address why SA signaling was not activated and confirm JA synthesis was increased, we examined SA level and JA level in leaf blades of WT and lmm9150 at seedling stage and tillering stage. The results showed that JA level of lmm9150 in leaf blades

was significantly higher than that of WT at both stages, whereas SA level of lmm9150 in leaf blades had no difference from that of WT (**Figure 4F** and **Supplementary Figure S2G**). These data indicated that JA biosynthesis is increased and thus JA-signaling pathway may be associated with the defense responses activated in lmm9150.

# Identification of lmm9150 as a Mutant of OsABA2

The two F<sup>1</sup> and two F<sup>2</sup> populations derived from the crosses of Yixiang1B (WT) × lmm9150 and lmm9150 × Yixiang1B were used for the genetic analysis. We found that all F<sup>1</sup> plants were not observed lesion mimic spots on leaf blades and the segregation between WT and lmm9150 in F<sup>2</sup> population fitted 3:1 (**Supplementary Table S4**). Thus, lmm9150 lesion mimic phenotype was controlled by a single recessive nuclear gene.

To isolate the candidate gene that was responsible for the phenotypes of lmm9150, a mapping population was constructed by crossing lmm9150 with a japonica cultivar 02428. Bulked segregation analysis (BSA) found that eight molecular markers on the end of chromosome 3 co-segregated with the mutant phenotypes of lmm9150. Linkage analysis showed that the mutant gene was mapped to a 488-kb interval between the InDel marker I403.3 and the SSR marker RM3684, co-segregated with I403.2 (**Figures 5A,B**). Next, the whole-genome was re-sequenced by using the DNA sample bulked from 20 BC1F<sup>2</sup> individuals with the lesion mimic phenotype. The genomic DNA of the WT was also sequenced as a control. A single base mutation, which SNP index was one, was found by comparing sequences between the bulked mutants and WT in this interval. Because this SNP localized in the second exon of LOC\_Os03g59610, we sequenced the gene in lmm9150 and the WT. As expected, a G-to-A single base substitution was detected at the 1487th base in lmm9150, which resulted in a change from Aspartic acid (D) to Asparagine (N) at the 110th amino acid (**Figure 5C** and **Supplementary Figure S3**). These results indicated that lmm9150 was likely arisen from the single base substitution in the LOC\_Os03g59610. This gene encodes XanDH, which is orthologous of ABA2 in Arabidopsis (Cheng et al., 2002; Miguel et al., 2002) and had been named OsABA2 in rice (Endo et al., 2014).

Sequence alignment of LOC\_Os03g59610 between Nipponbare and Yixiang1B revealed a 9-bp deletion in the first exon resulting in deletion of 3 amino acid residue in Yixiang1 B (GenBank accession number: MG334011), which is the WT of lmm9150 (**Supplementary Figure S4**).

To confirm that the phenotypes of lmm9150 were specifically associated with a mutation in OsABA2, we performed a knockout experiment in the WT background by using CRISPR/Cas9 editing. Five transgenic lines with either deletions or insertions in OsABA2 were obtained (**Figure 6A** and **Supplementary Figure S5**). All of these transgenic lines displayed lesion mimic spots on leaves (**Figure 6B**). Furthermore, gene-edited lines showed higher plant heights (**Figure 6C**), longer flag leaves (**Figure 6D**), pre-harvest sprouting phenotype (**Figures 6E,F**), and enhanced resistance to bacterial blight and rice blast diseases (**Figures 6G–I** and **Supplementary Figures S6A–G**). Taken together, these data confirmed that the loss-of-function of OsABA2 led to the lmm9150 phenotypes.

# The Function of OsABA2 in the Formation of Lesion Mimic Spots, Plant Development, and Seed Dormancy

OsABA2 encodes XanDH, an enzyme involves in ABA biosynthesis pathway (Endo et al., 2014). We hypothesized that mutation in this gene should influence the activity of XanDH and ABA level. To this end, we examined the activity of XanDH in leaf blades at three developmental stages. The results indicated

to the change of Aspartic acid (Asp) to Asparagine (Asn) at the 110th amino acid.

knockout lines are deletions. The mutation of #2 knockout line is insertion. (B) 5 knockout lines displays lesion mimic spots on leaves at tillering stage. (C) Comparison of plant height between WT and 5 knockout lines. (D) Comparison of flag leaf length between WT and 5 knockout lines. (E) The panicles for 30 days after flowering were collected and cultivated 3 days, at 28◦C. All knockout lines display pre-harvest spouting. (F) Comparison of the pre-harvest sprouting rates between WT and 5 knockout lines 30 days after flowering under natural condition. (G,H) The disease lesion of leaves of WT and #3 knockout line were inoculated with Xanthomonas oryzae pv. oryzae (Xoo) strain, P6, at 15 days post-inoculation. KL: knockout lines; FL: the flag leaf; SL: the second leaf. Data were obtained from 10 plants (C), 10 flag leaves of main panicles (D), 15 main panicles (F), and 20 leaves of main panicles (H). (I) The disease lesions of leaves of WT and #3 knockout line was inoculated 5 days post-inoculation with Magnaporthe oryzae (M. oryzae) strain, ZhongI. The experiments were repeated twice with similar results. Scale bar: 15 cm in (A), 5 cm in (E), 1 cm in (I). Statistical analysis was performed using Student's t-test, ∗∗ indicates P < 0.01.

that XanDH activity in lmm9150 was significantly lower than that in WT at all the tested stages (**Figure 7A**). Next, we examined the ABA level in leaf blades, seeds, and stems of lmm9150. Consistent with the lower activity of XanDH in lmm9150, ABA content in leaves, seeds, and stems were all significantly lower than that in the WT at all the tested developmental stages (**Figures 7B–D**). By contrast, the GA content in leaf blades, seeds and stems were all significantly higher than that of WT (**Supplementary Figures S7A–C**). Consistently, ABA/GA ratio was dramatically decreased in leaf blades, seeds and stems, respectively, which may explain why the seeds of lmm9150 showed lack of dormancy (**Figures 7E,F** and **Supplementary Figures S7D,E**). Taken together, we speculated that ABA deficiency likely resulted in the phenotypes of lmm9150.

Next, we applied ABA onto the seedlings of lmm9150 and WT. Such ABA-treatment led to three interesting changes of phenotypes in lmm9150. (1) A total of 20 days after treatment, the lmm9150 leaves in mock treatment displayed red brown spots from the older leaves (**Figures 8A,B**). In contrast, the lmm9150 leaves in ABA-treatment grew as healthy as that of the WT leaves (**Figures 8C,D**). (2) A total of 20 days after treatment, the plant height of lmm9150 was significantly higher than that of WT in mock treatment (**Figures 8E,F**). In contrast, the plant height of lmm9150 after ABA-treatment was similar to that of WT at seedling stage (**Figures 8G,H**). (3) The water loss rate of lmm9150 was obviously higher than that of WT plants at seedling stage (**Supplementary Figure S8A**). However, after ABA-treatment, the water loss rate of lmm9150 was similar to that of WT (**Supplementary Figure S8B**).

In addition, we tested the effect of different concentration of ABA on germination of seeds in ABA-containing media. The germination rate of lmm9150 seeds decreased when the concentration of ABA increased in a certain range, reflecting exogenous ABA could effectively inhibit seed germination (**Figure 8I** and **Supplementary Figures S8C,D**). Both the seeds of lmm9150 and WT could not germination in media containing 100 µmol/L ABA (**Supplementary Figure S8E**), indicating that this concentration may be the lethal dosage.

FIGURE 7 | Comparison of xanthoxin dehydrogenase (XanDH) activity and abscisic acid (ABA) contents between wild type (WT) and lmm9150 mutant. (A) Changes of XanDH activity in the leaf of WT and the lmm9150 mutant. SS: seedling stage; TS: tillering stage; MS: mature stage. (B–D) Comparison of ABA contents in the leaves (B), seeds (C), and stems (D) between WT and the lmm9150 mutant at different developmental stages. Hormone content was tested using enzyme-linked immunosorbent assay (ELISA) method. MRS: milk ripe stage; DS: dough stage; YMS: yellow mature stage. (E) The ratio of ABA/GA content in seed of WT and the lmm9150 mutant 30 days after flowering. (F) The germination rate of lmm9150 mutant and WT 30 days after flowering. Mean and SD were obtained from three replicates. Statistical analysis was performed using Student's t-test, <sup>∗</sup> and ∗∗ indicates P < 0.05 and P < 0.01, respectively.

FIGURE 8 | Analysis of relationship between abscisic acid (ABA) and plant development in lmm9150 mutant. (A–D) Phenotypic changes in the leaf blades of wild type (WT) and lmm9150 after ABA-treatment. The lmm9150 shows lesion mimic spots without ABA-treatment (A,B) and phenotype of the lmm9150 mutant is similar to WT when lmm9150 was treated with 100 µmol/L ABA for 15 days (C,D). The numbers 1–4 represent the first top leaf, the second top leaf, the third top leaf, and the fourth top leaf, respectively. (E–H) Change of plant height of lmm9150 mutant and WT 20 days after ABA-treatment at seedling stage. Images show that the plant height of lmm9150 and WT after water treatment (E,F) and ABA-treatment (G,H), respectively. The experiments were repeated twice with similar results (A–D,E,G). Data were obtained from 10 seedlings (F,H). (I) Comparison of seed germination rate in WT and lmm9150 mutant with 10 µmol/L ABA. Mean was obtained from three replicates. Statistical analysis was performed using Student's t-test, ∗∗ indicate P < 0.01.

Therefore, all the examined phenotypes of lmm9150, such as lesion mimic spots, plant height, water loss rate, and seed germination, could be rescued through the exogenous application of ABA.

# DISCUSSION

Lesion mimic mutants are an ideal tool to investigate the association between PCD and defense responses in plants. In the present study, we isolated a rice lesion mimic mutant, lmm9150, from EMS-mutagenized indica cultivar Yixiang1B. The lmm9150 mutant showed ABA deficiency which likely resulted in some novel and some expected phenotypes, such as spontaneous cell death, pre-harvest sprouting and enhanced growth of stem and leaf (**Figures 1**, **2**, **4**). Map-based cloning and CRISPR/Cas9-aided knocking-out identified a point mutation at the LOC\_Os03g59610 (OsABA2) locus encoding the xanthoxin dehydrogenase (XanDH) that catalyzes Xan into ABAld in the ABA biosynthesis pathway (Endo et al., 2014). Previously, OsABA2 is reported to be able to restore the ataba2 mutant in Arabidopsis (Endo et al., 2014), indicating that the function of ABA2 in ABA biosynthesis pathway may be conserved in monocot and dicot plants.

The lmm9150 mutant displayed spontaneous cell death on the third leaves starting from the seedling stage to the flag leaf at the yellow mature stage (**Figure 1A** and **Supplementary Figures S1A,B**), which was accompanied with excessive accumulation of H2O<sup>2</sup> in the mutant (**Figure 2C**), similar to other rice lesion mimic mutants (Zhu et al., 2016; Liu

et al., 2017; Wang et al., 2017). H2O<sup>2</sup> is a major by-product of β-oxidation and acts as a signal molecule in the promotion of cell death (Ren et al., 2002; Zhang et al., 2003; Gechev et al., 2005). Thus, the elevated H2O<sup>2</sup> in lmm9150 may have contributed to the formation of lesions. It has been reported that ABA suppresses PCD through increasing activities of ROS scavenging enzymes in barley (Bethke et al., 1999; Fath et al., 2001). In maize, loss of ABA synthesis, such as the ABA-deficient mutant viviparous9 (vp9), leads to the early onset of endosperm cell death (Young and Gallie, 2000). Here, we identified that lmm9150 was an ABA-deficient mutant (**Figures 7B–D**) and exhibited cell death lesions on leaves (**Figure 1A** and **Supplementary Figures S1A,B**). Exogenous application of ABA could inhibit the formation of lesions in lmm9150 (**Figures 8A–D**), confirming that deficiency in ABA biosynthesis can lead to formation of lesion mimics in rice. Nevertheless, the other rice ABA-deficient mutants, osaba1, phs1, phs2, phs3, and phs4, did not exhibit lesion mimic spots. H2O<sup>2</sup> level of these mutants was not detected and the superoxide accumulation (O<sup>2</sup> <sup>−</sup>) significantly increased in phs3 mutant. It is unknown why these ABA-deficient mutants do not exhibit lesion mimic phenotypes in rice (Agrawal et al., 2001; Fang et al., 2008).

During the lesion mimic formation, disease resistance responses are often auto-activated in LMMs, leading to enhanced resistance to pathogens in rice (Zhu et al., 2016; Liu et al., 2017; Wang et al., 2017). Consistent with this, lmm9150 displayed enhanced resistance to both bacterial blight and rice blast diseases (**Figure 4** and **Supplementary Figure S2**), similar to the enhanced disease resistance in the other ABA-deficient mutants, such as aba2-1, aba3-1 in Arabidopsis and sit in tomato (Xiao et al., 2017). Intriguingly, we noticed that the lesions of bacterial disease on flag leaves were longer than those on the second leaves in lmm9150 (**Figure 4C** and **Supplementary Figure S2D**). This could be due to the earlier formation of lesions on the second leaf than that on the flag leaf, as the formation of lesions features the activation of defense against pathogens. Consistent with the enhanced resistance of lmm9150 to pathogens, the expression of three PR genes, PR1a, PR1b, and PR10 were significantly upregulated in lmm9150 (**Figure 4A**). Moreover, the expression of two marker genes of JA-biosynthesis pathway was upregulated in the mutant, whereas the expression of two marker genes of SA signaling pathway was not obviously changed (**Figure 4E** and **Supplementary Figure S2F**). Consistent with this, JA level in lmm9150 was significantly increased, but SA level was the same as the WT (**Figure 4F** and **Supplementary Figure S2G**). The hormones, including ABA, JA, and SA, are secondary signal molecules involved in defense responses (Zeevaart and Creelman, 1988; Creelman and Mullet, 1997; Dempsey et al., 1997). It has been reported that the antagonistic interactions between ABA and JA–ET signaling pathway regulate disease resistance in Arabidopsis (Anderson et al., 2004). However, JA signaling and SA signaling can be mutually antagonistic or synergistic in defense response (Takahashi et al., 2004; Bari and Jones, 2009; Francisco et al., 2016). Therefore, the cross-talk between JA and SA is very complicated (Proietti et al., 2013). In addition, SA signaling is usually associated with the upregulation of PR genes (Vallad and Goodman, 2004; Zarate et al., 2007; Jiang et al., 2009). We also detected significantly increased expression of PR genes in lmm9150, although SA level had no obvious change (**Figure 4A** and **Supplementary Figure S2G**), which could be due to the activation of defense responses by JA signaling, because JA could also activate the expression of PR genes (Zarate et al., 2007). Collectively, our results indicated that the mutation of OsABA2 may alter ABA biosynthesis, which in turn, promote JA signaling pathway to activate defense against different pathogens. However, the exact mechanisms of the decreased level of ABA in lmm9150 that enhances PR gene expression, JA level, and defense against pathogens are yet to be determined in the future.

Seed dormancy and germination are mainly governed by ABA and GA (Shu et al., 2016). Generally, ABA maintains seed dormancy, whereas GA releases dormancy and promotes germination (Bewley, 1997; Shu et al., 2013; Liu et al., 2014), suggesting that the ratio of ABA/GA content is a major regulator for dormancy and germination (Fang et al., 2008; Finkelstein et al., 2008). Consistent with the previous literatures, our results showed that the ABA content was decreased and the GA content was increased, which resulted in a decreased ratio of ABA/GA and pre-harvest sprouting in the seeds of lmm9150 (**Figures 7B,E** and **Supplementary Figure S7B**). These findings were consistent with the observations in the other ABA-deficient mutants in rice that are loss of seed dormancy and exhibit pre-harvest sprouting, including osaba1, phs1, phs2, phs3, and phs4 (Agrawal et al., 2001; Fang et al., 2008). On the other hand, GA is a key regulator of release dormancy and promotion germination in plants (Peng and Harberd, 2002; Ye et al., 2015; Xiong et al., 2017). The balance of ABA/GA determines seed dormancy and germination (White et al., 2000; Fang et al., 2008; Finkelstein et al., 2008). It was possible that the dysfunction of ABA biosynthesis pathway can enhance GA biosynthesis indirectly through a feedback mechanism, as geranylgeranyl pyrophosphate (GGPP) is the common precursor for ABA and GA biosynthesis (Fray et al., 1995; Rodríguezconcepción et al., 2001) and the ABA deficiency may lead to more supply of GGPP for GA production in the mutant. Therefore, it is explainable that the lower ABA/GA ratio in lmm9150 contributed to its preharvest sprouting phenotype (**Figures 7C,E** and **Supplementary Figure S7B**).

In Arabidopsis, ataba2 displays dwarfism and small size leaf (Cheng et al., 2002; Miguel et al., 2002). In contrast, we found that the plant height of lmm9150 was higher than that of WT and the flag leaf of lmm9150 was longer than that of WT (**Figures 1D,E**), suggesting that the function of ABA2 in regulating growth in rice is different from that in Arabidopsis, although both genes are involved in ABA biosynthesis. We showed that mutation in OsABA2 led to decreased level of ABA content (**Figures 7B–D**), but increased level of GA content (**Supplementary Figures S7A–C**), and exogenous application of ABA could rescue the plant height phenotype of the lmm9150 mutant (**Figures 8E–H**), suggesting that OsABA2 negatively regulates plant height and leaf length through modulation of ABA and GA levels.

Together, our data demonstrated that novel functions of OsABA2 in modulating plant growth, cell death, and disease resistance, in addition to the expected function in seed dormancy and add new information of the roles of ABA in different biological processes in plant.

# AUTHOR CONTRIBUTIONS

fpls-09-00405 March 26, 2018 Time: 15:31 # 12

XW designed and performed the project. WW and ML directed the research. YL, QB, PX, and TW performed the analysis of agronomic traits, histochemical staining, shading treatment, and assessed rice disease resistance. YL and QB carried out genetic analysis, map-based cloning, and germination assay. DG and YP performed the mutation analysis and sequence analysis. XD, HZ, XC, and AA assisted in the data analysis. YL, XW, and WW wrote the manuscript. All authors read and approved the final manuscript.

# FUNDING

This work was supported by grants from the National Key Research and Development Program of China (Grant No. 2016YFD0100406), the Special Scientific Research Project of Agricultural Public Welfare of MOA, China (Grant No. 201403002-3), and the National Natural Science Foundation of China (Grant No. 31430072).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00405/ full#supplementary-material

FIGURE S1 | Phenotypic characterizations of lmm9150 mutant. (A,B) Comparison of four leaves between wild type (WT) and lmm9150 mutant exhibits the lesion mimic spots of lmm9150 at tillering stage (A) and mature stage (B). The numbers 1, 2, 3 and 4 represent the first top leaf, the second top leaf, the third top leaf and the fourth top leaf, respectively. (C–F) Quantification analysis of tiller numbers (C), grains setting rates (D), grains weight per plant (E), 1000-grains weight (F), respectively. Data were obtained from 10 plants of wild type and mutant (C–E). Mean of 1000-grains weight obtained from 10 replicates. Statistical analysis was performed using Student's t-test, <sup>∗</sup> and ∗∗ indicates P < 0.05 and P < 0.01, respectively. Scale bar: 2 cm in (A,B).

FIGURE S2 | Determination of wild type (WT) and lmm9150 plants resistance to bacterial blight and rice blast diseases. (A–D) The disease lesion leaves of the wild type (WT) and lmm9150 mutant were inoculated with Xanthomonas oryzae pv. oryzae (Xoo) strains, X004, P6 and P3, at 15 days post-inoculation (dip). The image (D) shows that the disease lesions of WT are longer than that of lmm9150 and the flag leaf disease lesions of lmm9150 are longer than that of the second leaf of mutant. FL: the flag leaf; SL: the second leaf. Data were obtained from 20 leaves of main panicles. (E) The disease lesions of leaves of the wild type (WT) and lmm9150 mutant were inoculated 5 days with Magnaporthe oryzae (M. oryzae) strain, Tetep. The experiments were repeated twice times with similar results.

# REFERENCES

Agrawal, G. K., Yamazaki, M., Kobayashi, M., Hirochika, R., Miyao, A., and Hirochika, H. (2001). Screening of the rice viviparous mutants generated by endogenous retrotransposon Tos17 insertion. Tagging of a zeaxanthin epoxidase gene and a novel OsTATC gene. Plant Physiol. 125, 1248–1257. doi: 10.1104/pp.125.3.1248

(F) Quantitative reverse-transcription PCR (qRT-PCR) data show the relative expression levels of salicylic acid (SA) signaling-related genes in lmm9150 and wild type (WT). (G) Comparison of SA content in leaf of wild type (WT) and lmm9150. SS: seedling stage; TS: tillering stage. The content of hormones was tested by enzyme-linked immunosorbent assay (ELISA) method. Mean and standard deviation were obtained from three measurements (F,G). The housekeeping gene Ubiquitin5 (Ubq5) was used as control. Statistical analysis was performed using Student's t-test, <sup>∗</sup>and∗∗ indicate P < 0.05 and P < 0.01, respectively. Scale bar: 1 cm in (E).

FIGURE S3 | Comparison of the amino acid sequences of xanthoxin dehydrogenases. Box I indicates cofactor binding site and box II indicates catalytic activity site. The arrow indicates the mutation site of OsABA2 in lmm9150 mutant. BRADI\_1g04320 in Brachypodium, GRMZM2G332976 in Zea mays, LOC8057083 in Sorghum bicolor, AT1G52340.1 in Arabidopsis, LOC102577811 in Solanum tuberosum. These symbols describe conversation degree, "<sup>∗</sup> , : and. " indicates strong, middle and weak respectively.

FIGURE S4 | Comparison of candidate gene sequence. LOC\_Os03g59610 is the reference sequence and comes from a japonica rice Nipponbare. Yixiang1B-CDS sequence was cloned from an indica cv. Yixiang1B, which is the wild type of lmm9150 mutant. The arrow indicates that 9-bp deletion is found in Yixiang1B.

FIGURE S5 | Mutational types of 5 knockout lines. Mutational types of #1, #3, #4 and #5 knockout lines are deletion. Mutational type of #2 knockout line is insertion. The black triangles indicate the mutational sites.

FIGURE S6 | Determination of wild type (WT) and a knockout line plants resistance to bacterial blight and rice blast diseases. (A–F) The disease lesion of leaves of the wild type (WT) and one of knockout lines were inoculated with Xanthomonas oryzae pv. oryzae (Xoo) strains, 8428, P3 and X004, at 15 days post-inoculation. KL: knockout lines; FL: the flag leaf; SL, the second leaf. Data were obtained from 20 leaves of main panicles. (G) The disease lesions of leaves of the wild type (WT) and one of knockout lines was inoculated 5 days with Magnaporthe oryzae (M. oryzae) strain, Tetep. The experiments were repeated twice times with similar results. Scale bar: 1cm in (G). Statistical analysis was performed using Student's t-test, ∗∗ indicates P < 0.01.

FIGURE S7 | Comparison of gibberellin (GA) and ABA/GA content ratio in the wild type (WT) and lmm9150 mutant. (A–C) Comparison of gibberellin (GA) content in leaves (A), seeds (B) and stems (C) between the wild type (WT) and lmm9150 mutant at different developmental stages. SS: seedling stage; TS: tillering stage; MS: mature stage; MRS: milk ripe stage; DS: dough stage; YMS: yellow mature stage. GA content was tested by enzyme-linked immunosorbent assay (ELISA) method. (D,E) The ratio of ABA/GA content in leaves (D) and stems (E) of the wild type (WT) and the lmm9150 mutant. Mean and standard deviation were obtained from three measurements. Statistical analysis was performed using Student's t-test, <sup>∗</sup> and ∗∗ indicates P < 0.05 and P < 0.01, respectively.

FIGURE S8 | Effects of ABA on water loss rate and germination in lmm9150 mutant. (A,B) Water loss assays for the leaves of the wild type (WT) and lmm9150 mutant were examined at seedling stage. (C–E) Investigation of germination rate in lmm9150 mutant and wild type seeds with different concentration of ABA (0.1, 1 and 100 µmol/L). Mean were obtained from three measurements.

TABLE S1 | List of primers for qRT-PCR.

TABLE S2 | List of polymorphic molecular markers for mapping.

TABLE S3 | List of targets sequences for PCR.

TABLE S4 | Segregation ratio of F2 populations.


with Erwinia chrysanthemi. Mol. Plant Pathol. 9, 11–24. doi: 10.1111/J.1364- 3703.2007.00437.X


pathway in fungi. Biosci. Biotechnol. Biochem. 64, 1448–1458. doi: 10.1271/bbb. 64.1448


fpls-09-00405 March 26, 2018 Time: 15:31 # 13

corresponding to the ABA locus of Arabidopsis thaliana. EMBO J. 15, 2331–2342.


fpls-09-00405 March 26, 2018 Time: 15:31 # 14


heterologous expression in tobacco. Biol. Plant 58, 89–98. doi: 10.1007/s10535- 013-0373-6


**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 Liao, Bai, Xu, Wu, Guo, Peng, Zhang, Deng, Chen, Luo, Ali, Wang 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 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.

fpls-09-00405 March 26, 2018 Time: 15:31 # 15

# Hsp90 Interacts With Tm-2<sup>2</sup> and Is Essential for Tm-2<sup>2</sup> -Mediated Resistance to Tobacco mosaic virus

Lichao Qian<sup>1</sup>† , Jinping Zhao1,2 \* † , Yumei Du<sup>1</sup> , Xijuan Zhao<sup>1</sup> , Meng Han<sup>1</sup> and Yule Liu<sup>1</sup> \*

<sup>1</sup> MOE Key Laboratory of Bioinformatics, Center for Plant Biology, Tsinghua-Peking Joint Center for Life Sciences, School of Life Sciences, Tsinghua University, Beijing, China, <sup>2</sup> Texas A&M AgriLife Research and Extension Center at Dallas, Texas A&M University, Dallas, TX, United States

The tomato resistance gene Tm-2<sup>2</sup> encodes a coiled coil-nucleotide binding site-leucine rich repeat type resistance protein and confers effective immune response against tobamoviruses by detecting the presence of viral movement proteins (MPs). In this study, we show that the Nicotiana benthamiana Heat shock protein 90-kD (Hsp90) interacts with Tm-2<sup>2</sup> . Silencing of Hsp90 reduced Tm-2<sup>2</sup> -mediated resistance to Tobacco mosaic virus (TMV) and the steady-state levels of Tm-2<sup>2</sup> protein. Further, Hsp90 associates with SGT1 in yeast and in plant cells. These results suggest that Hsp90-SGT1 complex takes part in Tm-2<sup>2</sup> -mediated TMV resistance by functioning as chaperone to regulate Tm-2<sup>2</sup> stability.

#### Edited by:

Yi Li, Peking University, China

#### Reviewed by:

Aiming Wang, Agriculture and Agri-Food Canada (AAFC), Canada Tao Zhou, China Agricultural University, China

#### \*Correspondence:

Jinping Zhao jinpingzhao@sina.cn Yule Liu yuleliu@mail.tsinghua.edu.cn

†These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Virology, a section of the journal Frontiers in Plant Science

Received: 01 February 2018 Accepted: 14 March 2018 Published: 10 April 2018

#### Citation:

Qian L, Zhao J, Du Y, Zhao X, Han M and Liu Y (2018) Hsp90 Interacts With Tm-2<sup>2</sup> and Is Essential for Tm-2<sup>2</sup> -Mediated Resistance to Tobacco mosaic virus. Front. Plant Sci. 9:411. doi: 10.3389/fpls.2018.00411 Keywords: Tobacco mosaic virus, plant-virus interaction, Hsp90, Tm-2<sup>2</sup> , SGT1, NBS-LRRs, Nicotiana benthamiana, immunity

# INTRODUCTION

In the natural environment, pathogen microbes, such as viruses, bacteria, fungi, oomycetes, and nematodes, can cause disease in host plants. To counteract the pathogen attack, plants have evolved multilevel and sophisticated mechanisms to protect them from potential pathogen invasion. One of them is resistance (R) gene-mediated immunity (Dangl and Jones, 2001). R gene or its product recognizes the cognate pathogen avirulence protein directly or indirectly and activates powerful, specific resistance response. The activation of resistance pathways often culminate in a rapid hypersensitive response (HR) cell death at the pathogen infection sites. However, some R genes can induce resistance response without visible HR (Heath, 2000; Jones and Dangl, 2006). A well-known example is the Rx gene from potato, which mediates extreme resistance against Potato virus X (PVX) without any visible cell death at the initial infection sites (Bendahmane et al., 1999). Besides, tomato resistance gene Tm-2<sup>2</sup> can mediate extreme resistance against tobamoviruses (Zhang et al., 2013). Nevertheless, they still have the potential to induce local cell death response in some conditions (Hall, 1980; Bendahmane et al., 1999; Du et al., 2013; Zhang et al., 2013).

Three tomato genes Tm-1, Tm-2, and Tm-2<sup>2</sup> mediate resistance against tobamoviruses including Tomato mosaic virus (ToMV) and Tobacco mosaic virus (TMV). In contrast to Tm-1 and Tm-2, Tm-2<sup>2</sup> mediates much more durable resistance and has been applied in crop cultivation for several decades (Lanfermeijer et al., 2003). Tm-2<sup>2</sup> contains a coiled-coil (CC) domain, a nucleotide binding site (NBS) domain and a leu-rich repeat (LRR) domain. Tm-2<sup>2</sup> detects the presence of tobamovirus MPs (Weber and Pfitzner, 1998; Lanfermeijer et al., 2004) and functions on the plasma membrane (Chen et al., 2017). The Tyr-767 in Tm-2<sup>2</sup> LRR domain is essential for the recognition

of the MP of ToMV strain B7 (Kobayashi et al., 2011), suggesting that Tm-2<sup>2</sup> recognizes viral MP through the LRR domain. In addition, Tm-2<sup>2</sup> requires all domains for its activity and PM localization (Chen et al., 2017). The N-terminus of ToMV MP is important for Tm-2<sup>2</sup> recognition (Weber et al., 2004; Chen et al., 2017). The expression of N-terminus (1-187aa) of viral MP is able to trigger Tm-2<sup>2</sup> -dependent cell death (Weber et al., 2004; Chen et al., 2017), although the two amino acid substitutions (S238R and K244E) in the C-terminus of ToMV MP lead to the overcoming of Tm-2<sup>2</sup> -mediated resistance (Weber et al., 1993). RuBisCO small subunit positively involves in Tm-2<sup>2</sup> -mediated extreme resistance (Zhao et al., 2013). Type I J-domain Hsp40 proteins (called NbMIP1s) and co-chaperone SGT1 are also indispensable for Tm-2<sup>2</sup> -mediated extreme resistance (Du et al., 2013; Zhao et al., 2013). Nevertheless, the molecular mechanism of Tm-2<sup>2</sup> -mediated virus resistance is largely unknown.

Heat shock protein 90-kD (Hsp90) is a molecular chaperone required for the stability and activity of many proteins during a variety of cellular processes, such as protein maturation, complex assembly, signal transduction and genetic buffering. For examples, plant Hsp90 can facilitate the folding of mammalian glucocorticoid receptors in vitro (Stancato et al., 1996). Hsp90 associates with the 26S proteasome and is critical for ATP-dependent assembly and maintenance of the 26S proteasome (Imai et al., 2003). Hsp90 is also essential for plant disease resistance. Hsp90 modulates RPS2- and RPM1-mediated resistance in Arabidopsis (Hubert et al., 2003; Takahashi et al., 2003). Silencing Hsp90 using viral-induced gene silencing (VIGS) suppressed the plant resistance conferred by several R genes including N, Rx and Pto in N. benthamiana (Kanzaki et al., 2003; Lu et al., 2003; Liu et al., 2004). Suppression of TaHsp90.2 or TaHsp90.3 compromised the resistance against stripe rust fungus in common wheat (Wang et al., 2011). Knock down of Hsp90 compromised I-2 mediated cell death completely, suggesting that Hsp90 is essential for the tomato I-2-mediated resistance (de la Fuente van Bentem et al., 2005). In addition, Hsp90 is also involved in Mi-1-mediated pest immune response (Bhattarai et al., 2007). SGT1 interacts with Hsp90, and functions as a co-chaperone of Hsp90 to modulate the immune response by regulating R protein stability (Lu et al., 2003; Liu et al., 2004; Zhang et al., 2004).

In this study, we report that N. benthamiana Hsp90 associates with Tm-2<sup>2</sup> in vitro and in vivo, and plays an essential role in Tm-2 2 -mediated TMV resistance by regulating its protein stability.

# MATERIALS AND METHODS

# Plant Materials, Plasmids and Pathogens

Wild type N. benthamiana and transgenic Tm-2<sup>2</sup> N. benthamiana plants were described (Zhang et al., 2013). All N. benthamiana plants were grown in greenhouse at 23–25◦C under a 16 h light/8 h dark cycle with 40–60% relative humidity and 40 umol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> white light illumination.

DNA fragments of Tm-2<sup>2</sup> -nLUC, cLUC-NbHsp90, Tm-2<sup>2</sup> -4 × myc and 3 × HA-NbHsp90 were generated by overlapping PCR, and then cloned into T-DNA vector pJG045, a pCAMBIA1300-based T-DNA vector (Zhao et al., 2013). PVX-based vector PVX-LIC was described (Zhao et al., 2016). The coding sequences of NbHsp90 (AY368904: nt1859-2103) was RT-PCR amplified and cloned into PVX-LIC vector for VIGS. All constructs were verified by DNA sequencing.

GFP-tagged TMV (TMV-GFP) was described (Liu et al., 2002a).

# Yeast Two-Hybrid Assays

The full-length Tm-2<sup>2</sup> , Tm-2<sup>2</sup> -LRR were PCR amplified and cloned into the LexA DNA binding domain (BD) containing yeast vector pYL302 (Liu et al., 2002b) to generate the bait vectors BD-Tm-2<sup>2</sup> and BD-Tm-2<sup>2</sup> -LRR. The full-length NbHsp90 cDNA was amplified by RT-PCR and cloned into the B42 activation domain (AD)-containing yeast vector pJG4-5 NbHsp90 to generate AD-NbHsp90. The yeast two-hybrid prey library containing tomato cDNAs (Liu et al., 2002b) was used to screen Tm-2<sup>2</sup> -LRR-binding proteins. The yeast two-hybrid screening and interaction assay were performed as described (Liu et al., 2002b).

# Luciferase Complementation Imaging (LCI) Assays

Luciferase complementation imaging assay was conducted as described (Chen et al., 2008; Du et al., 2013). All tested combinations were agroinfiltrated into leaves of 4 weeks old N. benthamiana. The leaves were collected 48 h post infiltration (hpi) and sprayed with luciferin (1 mM) followed by capturing the LUC image using a cooled CCD imaging apparatus (iXon, Andor Technology).

# Protein Analysis and Co-Immunoprecipitation (Co-IP)

We used Agrobacterium-mediated transient expression approach for protein expression. The GV2260 strains containing the relevant expression vector were infiltrated into leaves of N. benthamiana. The leaves were collected at 48 hpi for protein extraction. Protein samples were extracted with Laemmli buffer (Laemmli, 1970) and subjected to electrophoresis on SDS–PAGE gel followed by western blot assays using anti-myc (Abmart) or anti-Hsp90 (Santa Cruz Biotechnology) primary antibodies and were detected using Pierce ECL western blotting substrate (Pierce).

For Co-IP assays, HA-NbHsp90 was co-expressed with Tm2<sup>2</sup> -myc or cLUC-myc control in N. benthamiana. The infiltrated leaf tissues were collected 48 hpi and total protein extracts were subjected to IP procedure using anti-HA beads under agitation at 4◦C for 2 h, then the beads were washed four times with ice-cold extraction buffer at 4◦C (Du et al., 2013). The immunoprecipitates and input were extracted with Laemmli buffer and subjected to electrophoresis on SDS–PAGE gel followed by western blot assays using anti-myc or anti-HA antibody (Cell Signaling Technology) and detected using Pierce ECL western blotting substrate (Pierce).

# Gene Expression Assays

fpls-09-00411 April 6, 2018 Time: 16:24 # 3

RT-PCR and quantitative RT-PCR were conducted, respectively, as described (Liu et al., 2002a; Wang et al., 2013). NbActin mRNA was served as an internal control for normalization. Primers were designed with Primer3web<sup>1</sup> .

# VIGS, Virus Inoculation and GFP Imaging

For VIGS assays, PVX: NbHsp90 or control plasmid was transformed into Agrobacterium tumefaciens strains GV2260 and then infiltrated into the leaves of 4 weeks old N. benthamiana plants. For TMV infection, TMV-GFP was agroinfiltrated into the plant leaves (Liu et al., 2002a). Each silencing experiment was repeated using at least five independent plants at least four times Pictures were photographed under white and UV light using a Canon 650D camera.

# RESULTS

#### Identification of NbHsp90 as Tm-2<sup>2</sup> -Interacting Partner

Tm-2<sup>2</sup> LRR domain is reported to be involved in virus recognition (Kobayashi et al., 2011). To understand Tm-2<sup>2</sup> action, we conducted a yeast two-hybrid screen of a tomato cDNA library using Tm-2<sup>2</sup> -LRR (aa: 444-961) as bait, and identified several host proteins interacted with Tm-2<sup>2</sup> (Liu et al., 2004; Du et al., 2013). In this screen, we identified SGT1 and NbMIP1s as partners interacting with Tm-2<sup>2</sup> (Liu et al., 2002b; Du et al., 2013). In addition, Hsp90 (AY368906) (Liu et al., 2004) was also identified to interact with Tm-2<sup>2</sup> . Further, two N. benthamiana Hsp90 homologs (AY368904, AY368905) (Wang et al., 2011) were identified to share high identity with tomato Hsp90 (AY368906). It should be noted that two NbHsp90 homologs are almost identical to one another. Because N. benthamiana is an allotetraploid, we believe that these two NbHsp90 homologs are two alleles of same gene.

#### NbHsp90 Interacts With Tm-2<sup>2</sup> in Yeast

Further, we verified the interaction of NbHsp90 with Tm-2<sup>2</sup> using LexA based yeast two-hybrid system (Du et al., 2013). Both BD- and AD- fusion genes were driven by a galactose-inducible promoter. Yeasts transformed AD-NbHsp90 and BD-Tm-2<sup>2</sup> or BD-Tm-2<sup>2</sup> -LRR grew on galactose medium lacking leucine, and became blue on medium containing X-gal and galactose/raffinose but not glucose (**Figure 1**). In contrast, control yeasts containing AD or BD alone did not grow on the medium lacking leucine or turn blue on X-gal medium (**Figure 1**). Therefore, both Tm-2<sup>2</sup> and Tm-2<sup>2</sup> -LRR interact with NbHsp90 in yeast.

#### NbHsp90 Interacts With Tm-2<sup>2</sup> in Plant Cells

To examine whether NbHsp90 interacts with Tm-2<sup>2</sup> in plant cells, we conducted Co-IP assay. The HA-tagged NbHsp90 (HA-NbHsp90) was co-expressed with myc-tagged Tm-2<sup>2</sup> (Tm-2<sup>2</sup> myc) or cLUC-myc (as a negative control) in N. benthamiana

<sup>1</sup>http://primer3.ut.ee/

FIGURE 1 | NbHsp90 Interacts with Tm-2<sup>2</sup> in Yeast. Yeast cells containing NLS-LexA BD-Tm-2<sup>2</sup> or BD-Tm-2<sup>2</sup> -LRR baits transformed with AD-NbHsp90 grew on Leucine deficient medium (Leu−) and turned blue on X-gal medium containing galactose (Gal) and raffinose (Raf) but not on medium containing glucose (Glu) at 28◦C for 4 days. Yeast cells transformed with either AD or BD empty vector alone were used as negative control.

leaves. Leaf tissues were detached 48 hpi. Total protein was extracted and immunoprecipitated using anti-HA agarose, followed by western blot assays with anti-HA and anti-myc antibodies. We found that NbHsp90 co-immunoprecipitated with Tm-2<sup>2</sup> , but not with cLUC-myc (**Figure 2A**).

We further validated the in vivo interaction of NbHsp90 with Tm-2<sup>2</sup> via LCI assay (Chen et al., 2008). N-terminus (nLUC) and C-terminus (cLUC) of the firefly luciferase can reconstitute active enzyme when they are fused, respectively, with two interacting proteins. To this end, we generated Tm-2<sup>2</sup> -nLUC and cLUC-NbHsp90 and co-expressed them in N. benthamiana leaves. Positive signals were observed

for the combination of cLUC-NbHsp90 with Tm-2<sup>2</sup> -nLUC (**Figure 2B**). However, no signals were observed for the control combinations (cLUC-NbHsp90 plus nLUC, cLUC plus Tm-2<sup>2</sup> -nLUC) (**Figure 2B**). These results, along with our Co-IP data, suggest that NbHsp90 interacts with Tm-2<sup>2</sup> in plant cells.

#### NbHsp90 Is Essential for Tm-2<sup>2</sup> -Mediated TMV Resistance

To determine the role of NbHsp90 in N. benthamiana plants, we cloned a partial fragment of NbHsp90 (nt: 1859-2103) into PVX VIGS vector PVX-LIC (Zhao et al., 2016) to generate PVX-NbHsp90, and the PVX vector alone was used as negative control. Silencing of NbHsp90 induced developmental abnormalities including stopping growing and severely stunted (Supplementary Figure S1) (Liu et al., 2004), and quantitative RT-PCR data showed that the NbHsp90 mRNA level was greatly reduced in the PVX-NbHsp90 plants compared to the PVX vector plants (**Figure 3D**). Further, western blot assays using Hsp90-specific antibody showed that silencing of NbHsp90 greatly reduced Hsp90 protein level (**Figures 3E,F**).

Then we investigated the role of NbHsp90 in Tm-2<sup>2</sup> -mediated TMV resistance. To this end, we performed this experiment in transgenic Tm-2<sup>2</sup> N. benthamiana plants (thereafter called Tm-2<sup>2</sup> plants) that show effective resistance against TMV-GFP (Zhang et al., 2013). We agroinfiltrated the NbHsp90-silenced and PVX control non-silenced Tm-2<sup>2</sup> plants with Agrobacterium containing TMV-GFP plasmid (Liu et al., 2002a) and observed virus infection foci in inoculated leaves at 3 dpi (**Figure 3A**, left). Compared to the non-silenced Tm-2<sup>2</sup> plants, NbHsp90 silenced Tm-2<sup>2</sup> plants developed more TMV-GFP foci and subsequently developed obvious necrosis lesions at 7 dpi in inoculated leaves (**Figure 3A**, right). Furthermore, at 14 dpi TMV-GFP spread into the systemic leaves of NbHsp90 silenced Tm-2<sup>2</sup> plants but not that of non-silenced control Tm-2<sup>2</sup> plants (**Figure 3B**). RT-PCR showed that TMV RNA was readily detected in the systemic leaves of NbHsp90 silenced Tm-2<sup>2</sup> plants but not in the systemic leaves of non-silenced Tm-2<sup>2</sup> plants (**Figure 3C**). Taken together, these findings suggest that Tm-2<sup>2</sup> -mediated TMV resistance requires NbHsp90.

(right) but not the PVX control plants (left). (D) Quantitative RT-PCR assays to confirm the reduction in NbHsp90 mRNA (means ± SEM, n = 3). <sup>∗</sup>P < 0.05, Student's t-test. NbActin mRNA levels were used as the internal control. (E,F) Western blot assays to confirm the reduction in NbHsp90 protein level (means ± SEM, n = 4). <sup>∗</sup>P < 0.05, Student's t-test. Equal loading of protein samples was validated by Ponceau Red staining of Rubisco subunit.

# NbHsp90 Is Essential for Stability of Tm-2<sup>2</sup> Protein

NbHsp90 is essential for Rx-mediated PVX resistance by regulating the protein level of Rx-4 × HA in N. benthamiana (Lu et al., 2003). RPM1 level is also reduced in Arabidopsis hsp90.2 mutant (Hubert et al., 2003). To investigate how NbHsp90 regulates Tm-2<sup>2</sup> -mediated TMV resistance, we expressed Tm-2<sup>2</sup> -myc in NbHsp90-silenced and non-silenced N. benthamiana plants to investigate the effect of NbHsp90 silencing on Tm-2<sup>2</sup> protein accumulation (Du et al., 2013). Western blot assays showed that NbHsp90-silenced plants accumulated less Tm-2<sup>2</sup> protein compared with non-silenced control plants (**Figure 4A**). However, quantitative RT-PCR assay indicated that NbHsp90 silencing had no significant effect on Tm-2<sup>2</sup> mRNA level (**Figure 4B**). Taken together, these findings indicate that NbHsp90 is indispensable for Tm-2<sup>2</sup> protein stability.

# NbHsp90 Interacts With NbSGT1 in Yeast and in Plant Cells

We had reported that NbSGT1 interacts with Tm-2<sup>2</sup> and is essential for Tm-2<sup>2</sup> -mediated TMV resistance by regulating Tm-2<sup>2</sup> protein stability (Du et al., 2013). To investigate whether Hsp90 regulates Tm-2<sup>2</sup> protein stability through NbHsp90- NbSGT1 chaperone complex, we first tested the interaction between NbHsp90 and NbSGT1 using yeast two-hybrid system. Yeast cells harboring both AD-NbHsp90 and BD-SGT1 grew on medium lacking leucine, and became blue on medium containing X-gal and galactose/raffinose but not glucose (**Figure 5A**). However, control yeasts containing AD or BD vector alone neither grew on the medium lacking leucine nor turned blue on X-gal medium (**Figure 5A**). NbHsp90 therefore interacts with

NbSGT1 in yeast. Further, we performed LCI assays to investigate whether NbHsp90 interacts with NbSGT1 in plant cells. We found that cLUC-NbHsp90 interacts with NbSGT1-nLUC, but not with empty cLUC control (**Figure 5B**). These experiments show that NbHsp90 interacts with NbSGT1 in both yeast and plant.

# DISCUSSION

# The Role of Hsp90 in Plant R Gene-Mediated Resistance Against Viruses

Hsp90 is a highly abundant and conserved cellular chaperone known to regulate various biological processes, and is reported to play crucial roles in plant disease resistance (Hubert et al., 2003; Lu et al., 2003; Takahashi et al., 2003; Liu et al., 2004). Using high throughput VIGS assay, Hsp90 was characterized to be a cofactor of Rx protein to stabilize its protein level (Lu et al., 2003). Association of Hsp90 with NBS-LRR proteins has been

AD-NbHsp90 grew on Leu- deficient medium and turned blue on X-gal medium containing galactose (Gal) and raffinose (Raf) but not on medium containing glucose (Glu) at 28◦C for 4 days. Yeast cells transformed with either AD or BD empty vector alone were used as negative control. (B) Firefly LCI assays for the in vivo interaction between NbHsp90 with SGT1. cLUC-NbHsp90 was transiently co-expressed with SGT1-nLUC or nLUC in N. benthamiana leaves followed by LCI assay.

reported (Hubert et al., 2003; Liu et al., 2004; Zhu et al., 2017). Hsp90 associates with N protein directly to modulate the immune response to TMV (Liu et al., 2004). Hsp90, SGT1 and Rar1 form a complex and act as co-chaperones during virus disease resistance (Picard, 2002). In fact, the structurally conserved Hsp90-SGT1 complex (Seo et al., 2008; Shirasu, 2009) are functionally required for different NBS-LRR proteins' function as immune modulator against various pathogens including bacteria (Takahashi et al., 2003; Zhang et al., 2004), fungi (Bieri et al., 2004; Thao et al., 2007), oomycetes (Michael Weaver et al., 2006; Bhaskar et al., 2008; Oh et al., 2014), nematodes (Bhattarai et al., 2007; Zhu et al., 2017). Here we reported that Hsp90 directly interacts with Tm-2<sup>2</sup> , a CC-NBS-LRR type of resistance protein, confers robust immune response against tobamoviruses. Besides, we found that Hsp90 interacts with SGT1 in yeast and in plant cells. This finding is consistent with our earlier report that SGT1 participates in Tm-2<sup>2</sup> -mediated resistance against TMV by regulating protein stability of Tm-2<sup>2</sup> through its interaction with Tm-2<sup>2</sup> (Du et al., 2013). Hsp90 and its co-chaperone SGT1 may facilitate the folding and maturation of R proteins. The misfolded R proteins can be eliminated by protein quality control machine. In such a scenario, knock down of Hsp90 or SGT1 decreases the amount of R protein and compromises R protein function (Lu et al., 2003; Liu et al., 2004; Zhang et al., 2004). Accordingly, silencing of Hsp90 suppressed Tm-2<sup>2</sup> -mediated TMV resistance and reduced the stability of Tm-2<sup>2</sup> protein. Taken together, our findings further support that Hsp90-SGT1 chaperone mediates the stabilization and maturation of R proteins.

# The Role of Hsp90-Related Co-chaperones in Plant-Virus Interaction

The DnaJ/Hsp40 works as a co-chaperone in Hsp90-Hsp70- Hsp40 complex, and can also form complex with Hsp90 during protein folding process (Verchot, 2012). DnaJ/Hsp40 proteins play dual roles in plant virus infection and host resistance. Via directly interaction with virus effectors, varied DnaJ/Hsp40 type proteins positively or negatively affect the replication and/or movement of several plant viruses including PVX, PVY, TSWV, and TMV (Soellick et al., 2000; Hofius et al., 2007; Cho et al., 2012; Shimizu et al., 2009). In addition, type I DnaJ/Hsp40 protein NbMIP1s also interact with Tm-2<sup>2</sup> and SGT1 and are required for Tm-2<sup>2</sup> -mediated resistance by sustaining the protein stability (Du et al., 2013). Type-III DnaJ/Hsp40 plays a positive role in plant defense against Soybean mosaic virus in soybean (Liu and Whitham, 2013).

# REFERENCES


In this study, we found that Hsp90, like NbMIP1, is required for Tm-2<sup>2</sup> -mediated resistance against TMV. However, no interaction between TMV MP and Hsp90 is detected (data not shown). In addition, NbHsp90 expression (at mRNA and protein levels) was not induced by TMV infection and Tm-2<sup>2</sup> -mediate resistance (Supplementary Figure S2). Considered that NbMIP1 interacts with TMV MP, and NbMIP1s is induced by TMV infection and Tm-2<sup>2</sup> -mediate resistance (Du et al., 2013), Hsp90 and NbMIP1s may exist in different cellular protein complexes during plant-virus interaction (Cintron and Toft, 2006). Indeed, Hsp90 and DnaJ/HsP40 proteins are not necessarily linked in their role as chaperones to facilitate the folding of diverse client proteins during different biological processes such as virus infection and plant resistance (Li et al., 2012; Verchot, 2012), and the chaperone machinery Hsp90-Sgt1 and Hsp90-Hsp40 is of different partnership for client recruitment and folding (Park and Seo, 2015).

# AUTHOR CONTRIBUTIONS

YL, JZ, LQ, and YD designed the experiments, analyzed the data, and prepared the manuscript. JZ, LQ, YD, XZ, and MH carried out the experiments. All authors contributed the revision of manuscript through the discussion.

# FUNDING

This work was supported by the National Natural Science Foundation of China (Grant Nos. 31530059, 31421001, and 31470254).

# ACKNOWLEDGMENTS

We are grateful to Jianmin Zhou (Institute of Genetics and Developmental Biology, Chinese Academy of Sciences) for providing LCI vectors.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00411/ full#supplementary-material


capsid protein affects viral replication and movement. Biochem. Biophys. Res. Commun. 417, 451–456. doi: 10.1016/j.bbrc.2011.11.137



Zhu, X., Xiao, K., Cui, H., and Hu, J. (2017). Overexpression of the Prunus sogdiana NBS-LRR subgroup gene PsoRPM2 promotes resistance to the rootknot nematode Meloidogyne incognita in Tobacco. Front. Microbiol. 8:2113. doi: 10.3389/fmicb.2017.02113

**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 Qian, Zhao, Du, Zhao, Han 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) 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.

# Differential Suppression of Nicotiana benthamiana Innate Immune Responses by Transiently Expressed Pseudomonas syringae Type III Effectors

Selena Gimenez-Ibanez1,2 \* † , Dagmar R. Hann1,3,4† , Jeff H. Chang5,6, Cécile Segonzac1,7 , Thomas Boller<sup>3</sup> and John P. Rathjen1,8

#### Edited by:

Zhengqing Fu, University of South Carolina, United States

#### Reviewed by:

Brian H. Kvitko, University of Georgia, United States Frederik Börnke, Leibniz-Institut für Gemüse-und Zierpflanzenbau (IGZ), Germany

\*Correspondence:

Selena Gimenez-Ibanez selena.gimenez@cnb.csic.es

†These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 25 January 2018 Accepted: 04 May 2018 Published: 23 May 2018

#### Citation:

Gimenez-Ibanez S, Hann DR, Chang JH, Segonzac C, Boller T and Rathjen JP (2018) Differential Suppression of Nicotiana benthamiana Innate Immune Responses by Transiently Expressed Pseudomonas syringae Type III Effectors. Front. Plant Sci. 9:688. doi: 10.3389/fpls.2018.00688 <sup>1</sup> The Sainsbury Laboratory, Norwich, United Kingdom, <sup>2</sup> Plant Molecular Genetics Department, Centro Nacional de Biotecnología-Consejo Superior de Investigaciones Científicas, Madrid, Spain, <sup>3</sup> Department of Environmental Sciences, Botanical Institute, University of Basel, Basel, Switzerland, <sup>4</sup> Institute of Genetics, Ludwig-Maximilians-Universität München, Munich, Germany, <sup>5</sup> Department of Botany and Plant Pathology, Oregon State University, Corvallis, OR, United States, <sup>6</sup> Center for Genome Research and Biocomputing, Oregon State University, Corvallis, OR, United States, <sup>7</sup> Department of Plant Science, Plant Genomics and Breeding Institute and Research Institute of Agriculture and Life Sciences, College of Agriculture and Life Sciences, Seoul National University, Seoul, South Korea, <sup>8</sup> Research School of Biology, Australian National University, Acton, ACT, Australia

The plant pathogen Pseudomonas syringae injects about 30 different virulence proteins, so-called effectors, via a type III secretion system into plant cells to promote disease. Although some of these effectors are known to suppress either pattern-triggered immunity (PTI) or effector-triggered immunity (ETI), the mode of action of most of them remains unknown. Here, we used transient expression in Nicotiana benthamiana, to test the abilities of type III effectors of Pseudomonas syringae pv. tomato (Pto) DC3000 and Pseudomonas syringae pv. tabaci (Pta) 11528 to interfere with plant immunity. We monitored the sequential and rapid bursts of cytoplasmic Ca2<sup>+</sup> and reactive oxygen species (ROS), the subsequent induction of defense gene expression, and promotion of cell death. We found that several effector proteins caused cell death, but independently of the known plant immune regulator NbSGT1, a gene essential for ETI. Furthermore, many effectors delayed or blocked the cell death-promoting activity of other effectors, thereby potentially contributing to pathogenesis. Secondly, a large number of effectors were able to suppress PAMP-induced defense responses. In the majority of cases, this resulted in suppression of all studied PAMP responses, suggesting that these effectors target common elements of PTI. However, effectors also targeted different steps within defense pathways and could be divided into three major groups based on their suppressive activities. Finally, the abilities of effectors of both Pto DC3000 and Pta 11528 to suppress plant immunity was conserved in most but not all cases. Overall, our data present a comprehensive picture of the mode of action of these effectors and indicate that most of them suppress plant defenses in various ways.

Keywords: Pseudomonas, effector, PTI, cell death, suppression, Nicotiana benthamiana

# INTRODUCTION

fpls-09-00688 May 18, 2018 Time: 16:56 # 2

Plants respond to infections via an innate immune system that employs external and internal receptors (Jones and Dangl, 2006). Primary perception is based on the ability to discriminate between self and non-self. Plants rely on surface-localized pattern recognition receptors (PRRs) to detect conserved molecules called pathogen-associated molecular patterns (PAMPs), or sense wound- and injury-related molecules called danger-associated molecular patterns (DAMPs) to infer the presence of microbes (Boller and Felix, 2009). This recognition leads to PAMPtriggered immunity (PTI), which is considered to be the first level of plant defense, and restricts pathogen infection in most plant species. To overcome such immunity, successful bacterial pathogens use type III secretion systems (TTSSs) to deliver virulence molecules called effector proteins into eukaryotic cells (Collmer et al., 2002). The second branch of the plant innate immune system recognizes type III effectors inside the plant cell via nucleotide-binding leucine-rich repeat (NB-LRR) resistance (R) proteins and is called effector-triggered immunity (ETI) (Chisholm et al., 2006). This specific recognition leads to strong activation of a similar range of plant responses as PTI and is characteristically associated with programmed cell death known as the hypersensitive response (HR) at the infection site.

Pattern recognition receptors are generally membrane-bound receptor kinases (RKs) or receptor proteins (RPs). Ligand perception at the cell surface causes the immediate formation of stable heterodimers comprised of the specific PRR and certain co-receptors. This constitutes an active receptor complex that initiates intracellular signaling (Chinchilla et al., 2007; Heese et al., 2007; Boutrot et al., 2010; Liu et al., 2012; Cao et al., 2014). In Arabidopsis, the paradigm for PAMP sensing is the perception of the bacterial flagellin protein by the leucine-rich repeat (LRR) RK FLS2 (FLAGELLIN-SENSING 2), which triggers formation of an active receptor complex with the LRR-RK co-receptor BAK1 (BRI1-ASSOCIATED KINASE 1) (Chinchilla et al., 2007; Heese et al., 2007). Similarly, chitin oligosaccharides released from fungal cell walls are perceived by the Arabidopsis lysin motif (LysM) RKs LYK5 (LYSIN MOTIF-CONTAINING RECEPTOR-LIKE KINASE 3) and CERK1 (CHITIN ELICITOR RECEPTOR KINASE 1) that form a receptor complex in which both bind chitin (Liu et al., 2012; Cao et al., 2014). Upon PAMP sensing, the activation of different PRR complexes leads to the selective phosphorylation and activation of specific cytoplasmic receptor-like cytoplasmic kinases (RLCKs). These include BIK1 (BOTRYTIS-INDUCED KINASE 1) downstream of flagellin recognition, and PBL27 downstream of chitin perception (Lu et al., 2010; Kadota et al., 2014; Shinya et al., 2014; Yamada et al., 2016).

Activation of PRRs induces a variety of host responses, including evolution of a burst of Ca2<sup>+</sup> and reactive oxygen species (ROS), post-translational activation of mitogen-activated protein kinases (MAPK), induction of defense genes, callose deposition into plant cell walls and functional immunity (Kunze et al., 2004; Zipfel et al., 2004; Chinchilla et al., 2006). Upon PAMP stimulation, one of the earliest responses is changes in ion fluxes across the plasma membrane. This includes an influx of Ca2<sup>+</sup> from the apoplast within seconds (Chandra et al., 1997; Blume et al., 2000; Lecourieux et al., 2002; Davies et al., 2006; Segonzac et al., 2011). Downstream of the Ca2<sup>+</sup> influx, two distinct branches of signaling occur, one leading to ROS production by NADPH oxidases and the other to the activation of MAPKs and transcriptional changes (Segonzac et al., 2011). The plasma membrane Arabidopsis NADPH oxidase RBOHD (RESPIRATORY BURST OXIDASE HOMOLOG D) is part of the PRR complex and is directly regulated through phosphorylation by the RLCK BIK1 in a calcium-independent manner. This leads to ROS production and antibacterial immunity (Kadota et al., 2014; Li et al., 2014). In contrast, the RLCK PBL27 directly connects the PRR CERK1 with the MAPK kinase kinase MAPKKK5, establishing a link between PRRs and activation of intracellular MAPK cascades (Yamada et al., 2016). However, the fact that the influx of Ca2<sup>+</sup> across the plasma membrane is nevertheless required for the PAMP-induced ROS burst and MAPK activation (Segonzac et al., 2011; Kadota et al., 2014), suggests that calcium-based regulation is also required for the ultimate activation of RBOHD and MAPKKK5. Successful initiation of signaling leads to characteristic transcriptional reprogramming. Microbial perception alters the expression of approximately 10% of the whole plant transcriptome (Navarro et al., 2004; Moore et al., 2011). This results in the production of antimicrobial compounds and cell wall-reinforcing materials around the infection site (Hahlbrock et al., 2003), which is supposed to be critical for plant immunity.

Type III effectors contribute collectively to bacterial pathogenesis by targeting host defense pathways (Katagiri et al., 2002). Effectors hijack multiple cellular processes including phytohormone signaling, proteasome-dependent protein degradation, cytoskeleton formation, manipulation of stomatal openings, establishment of intercellular apoplastic living spaces, and vesicle transport (Buttner, 2016; Toruno et al., 2016). Among all these functions, suppression of PTI has emerged as a primary role of bacterial effectors to ensure pathogenesis. Multiple effectors directly target PRRs, their co-receptors, RLCKs and MAPKs in a redundant manner and by multiple strategies to ensure suppression (Macho and Zipfel, 2015; Buttner, 2016). Some examples include the effectors AvrPto, AvrPtoB, HopAO1, and HopF2, which act as direct suppressors of PRRs such as FLS2 and/or their co-receptors BAK1 and CERK1; HopAR1 that targets BIK1 and other RLCKs; and HopAI1 and HopF2 that suppress MAPK signaling by directly modifying these proteins (Macho and Zipfel, 2015; Buttner, 2016; Toruno et al., 2016). Despite the success in identifying effectors using functional and genetic analysis, the modes of action of most of them are yet unknown.

The Gram-negative bacterial species Pseudomonas syringae (P. syringae) comprises at least 50 pathovars that can be distinguished by their host ranges (Sawada et al., 1999). For example, P. syringae pv. tomato (Pto) DC3000 infects tomato and can cause disease on Arabidopsis whereas P. syringae pv. tabaci (Pta) 11528 is infectious on Nicotiana tabacum and can grow to high levels on the model species Nicotiana benthamiana (N. benthamiana). The Arabidopsis-Pto DC3000 system represents the primary model for plant–bacteria interactions

(Katagiri et al., 2002). In contrast, the Pta 11528-N. benthamiana system represents an interaction that offers complementary benefits to Arabidopsis, including high amenability to Agrobacterium tumefaciens-mediated transient expression of foreign genes and virus-induced gene silencing (VIGS) for gene knockdowns (Goodin et al., 2008). Pto DC3000 encodes nearly 30 effectors whereas Pta 11528 carries approximately 20 genes with homology to previously described effectors (Chang et al., 2005; Vinatzer et al., 2005; Ferreira et al., 2006; Lindeberg et al., 2006; Schechter et al., 2006; Studholme et al., 2009). Several of these Pto DC3000 effectors are conserved in Pta 11528 but it is currently unknown whether they function similarly to their Pto DC3000 homologs. We currently lack an integrated view of which defense pathways are suppressed by the effector repertoire, how many effectors act redundantly, and the conservation of effector function among homologs in different Pseudomonas strains. Also, although most of the current work has focused on defense responses elicited by flagellin, Pseudomonas contains many more PAMPs which may stimulate alternate pathways. It is not clear whether different PRRs feed into a limited number of signal transduction pathways, or if effectors target certain pathways specifically or act more broadly. Thus, we require a more integrative view of global molecular activities and the range of suppression by bacterial effectors to understand how host processes normally prevent successful infection.

In this study, we transiently expressed type III effector genes in leaves of N. benthamiana to screen and identify those with the ability to suppress defense responses. We analyzed nearly all Pto DC3000 effectors and most Pta 11528 secreted proteins for the ability to interfere with effector induced cell death, a typical hallmark of effector recognition by plant R genes. Secondly, we analyzed their capacities to abolish a broad range of PAMP responses including changes in cytoplasmic Ca2<sup>+</sup> levels, production of ROS and induction of defense genes. Strikingly, among the 32 effectors tested, 29 showed suppressive activity in at least one of our assays. This observation is consistent with the hypothesis that one of the most important roles of effector proteins is to suppress host immunity. This work helps to identify P. syringae effectors that are capable of suppressing plant defenses, and to define the spectrum of their activities.

# MATERIALS AND METHODS

# Elicitors

Crab shell chitin and flg22 peptide (CKANSFREDRNEDREV) were purchased from Sigma (United Kingdom) and Peptron (South Korea), respectively. Pto DC3000, Pta 11528, and A. tumefaciens (strain C58C1) were grown in L-medium at 28◦C on a rotary shaker. Crude A. tumefaciens, Pto DC3000 and Pta 11528 bacterial suspensions were prepared by centrifuging overnight cultures and resuspending bacterial pellets in water.

# Statistical Methods

Statistical significance based on t-test analysis was developed by the GraphPad Prism program. Eight independent samples were used to analyze the significance of ROS and Ca2<sup>+</sup> generation assays. For quantitative gene expression analysis, statistical significance was calculated from three independent samples.

# Bacterial Strains

Agrobacterium tumefaciens strain C58C1 was used for transient assays transformed with pGWB14 (Pto DC3000 and Pta 11528 effector libraries) (Nakagawa et al., 2007). Both the Pto DC3000 and Pta 11528 effector libraries were described previously (Chang et al., 2005; Gimenez-Ibanez et al., 2014). The Pta 11528 effector library contains individual effector genes under control of the 35S promoter fused in-frame to a sequence encoding three C-terminal hemagglutinin (HA) epitope tags (Gimenez-Ibanez et al., 2014). In contrast, the full-length Pto DC3000 effectors used in this study were recombined into pDONR207 (Invitrogen) (Chang et al., 2005). These plasmids, a Gateway BP II kit (Invitrogen), and the pGWB14 (Nakagawa et al., 2007) destination vector were used to generate expression constructs in which each Pto DC3000 effector gene is under the control of the 35S promoter and fused to a sequence encoding three HA epitope tags. These constructs were verified by DNA sequencing and then transferred to A. tumefaciens strain C58C1. Additional bacterial strains used in this study were Pseudomonas syringae pv. tomato (Pto) DC3000, Pto DC3000 hrcC, Pto DC30001avrPto, Pto DC30001avrPtoB, and Pto DC30001avrPto1avrPtoB (Lin and Martin, 2005).

# A. tumefaciens-Mediated Transient Expression Assays

For transient gene expression, A. tumefaciens C58C1 was syringe infiltrated in N. benthamiana leaves at OD<sup>600</sup> = 0.3–0.5 in 10 mM MgCl<sup>2</sup> and 10 mM MES. Samples were collected 2 days post-inoculation for ROS, Ca2+, MAPK and gene expression assays. For co-infiltration assays, each of the two A. tumefaciens strains containing the respective effector gene were prepared at OD<sup>600</sup> = 0.5 in 10 mM MgCl<sup>2</sup> and 10 mM MES and then mixed in equal volumes. This solution was infiltrated in N. benthamiana leaves and phenotypes were scored between 1 to 7 days postinoculation as indicated in the text.

# Virus-Induced Gene Silencing (VIGS)

Virus-induced gene silencing was performed using a tobacco rattle virus vector as previously described (Peart et al., 2002).

# Measurement of Reactive Oxygen Species (ROS) Generation

Leaf disks (0.38 cm<sup>2</sup> ) were floated on water overnight and ROS released by the leaf tissue were measured using a chemiluminescent assay (Keppler et al., 1989). The water was replaced with 200 µl of a solution containing 20 µM luminol (Sigma, St. Louis, MO, United States) and 1 µg of horseradish peroxidase (Fluka, Buchs, Switzerland). ROS was elicited with 100 nM flg22 or 100 µg/ml of chitin in all experiments. Elicitation in the absence of any PAMP (water treatment) was included as a negative control. Luminescence was measured over a time period of 30 min using the Photek camera system (East Sussex, United Kingdom), and data were recorded as total counts.

# Measurement of Ca2<sup>+</sup> Burst Generation

Transgenic N. benthamiana plants expressing the Ca2<sup>+</sup> sensor 35S:Aequorin (N. benthamiana SLJR15) were used to measure intracellular Ca2<sup>+</sup> concentrations (Segonzac et al., 2011). Leaf disks (0.38 cm<sup>2</sup> ) were floated overnight on an aqueous 2.5 µM coelenterazine solution in the dark at room temperature. The Ca2<sup>+</sup> influx was elicited with water, flg22 (100 nM) or chitin (100 µg/ml), and luminescence was measured using the Photek camera system as total counts over a time period of 30 min.

# Quantitative RT-PCR

Nicotiana benthamiana leaf disks expressing each individual effector were collected 2 days post-infiltration (dpi), and then floated overnight in water. Leaf disks were subsequently elicited with water (Mock EV control), 100 nM flg22 or 100 µg/ml chitin for 60 min and frozen in liquid nitrogen. Total RNA was extracted by using TRIzol-Reagent (Sigma), and the absence of genomic DNA was checked by PCR amplification of the housekeeping NbEF1α gene by using 2.5 µg of RNA. For analysis of gene expression, first-strand cDNA was synthesized from 2.5 µg of RNA using SuperScript RNA H-Reverse Transcriptase (Invitrogen, United Kingdom) and an oligo (dT) primer, according to the manufacturer's instructions. For quantitative PCR, 2 µl of cDNA was combined with SYBR master mix. PCRs were performed in triplicate with a PTC-200 Peltier Thermal Cycler (MJ Research, Waltham, MA, United States), and the data were collected and analyzed with Chromo 4 Continuous Fluorescence detection system. The NbEF1α RNA was analyzed as an internal control and used to normalize the values for transcript abundance. All samples were related to the Mock EV negative control. Primers for genes used here are as follows: NbCyp71D20, 5<sup>0</sup> -AAGGTCCACCGCACCATGTCCTTAGAG-3 0 and 5<sup>0</sup> -AAGAATTCCTTGCCCCTTGAGTACTTGC-3<sup>0</sup> ; NbACRE132, 5<sup>0</sup> -AAGGTCCAGCGAAGTCTCTGAGGGTGA-3<sup>0</sup> and 5<sup>0</sup> -AAGAATTC-CAATCCTAGCTCTGGCTCCTG-3<sup>0</sup> ; and NbEF1α gene 5<sup>0</sup> -AAGGTCCAGTATGCCTGGGTGCTTGAC-3<sup>0</sup> and 5<sup>0</sup> -AAGAATTCACAGGGACAGTTCCAATACCA-3<sup>0</sup> .

# Cell Death Assays on N. benthamiana

For elicitation of the HR by different P. syringae pathovars, overnight bacterial cultures were pelleted, resuspended in sterile 10 mM MgCl<sup>2</sup> and infiltrated into the leaves of transgenic N. benthamiana plants at high densities (5 × 10<sup>7</sup> cfu/ml). The development of cell death was scored between 1 to 4 dpi.

# RESULTS

# Construction of Effector Libraries for Pto DC3000 and Pta 11528

Pto DC3000 and Pta 11528 encode nearly 30 and 20 effectors, respectively (Chang et al., 2005; Vinatzer et al., 2005; Ferreira et al., 2006; Lindeberg et al., 2006; Schechter et al., 2006; Studholme et al., 2009). To investigate the functions of most of these, we analyzed two libraries containing 22 Pto DC3000 and 10 Pta 11528 effector genes, respectively. Both libraries were described previously (Chang et al., 2005; Gimenez-Ibanez et al., 2014) and express each effector from a T-DNA under control of the strong 35S promoter, fused in frame to a sequence encoding three HA epitope tags. There were seven pairs of homologous effectors shared between the Pto DC3000 and Pta 11528 libraries sampled here, with varying levels of amino acid conservation. Based on protein sequence identity to their Pto DC3000 homologs, some Pta 11528 effectors were almost identical, namely HopO1-1Pta<sup>11528</sup> (99%, Supplementary Figure S1) and HopT1-1Pta<sup>11528</sup> (99%, Supplementary Figure S2); others showed high levels of identity such as AvrPtoBPta<sup>11528</sup> (70%, Supplementary Figure S3) and HopX1Pta<sup>11528</sup> (72%, Supplementary Figure S4); whereas another group including AvrPtoPta<sup>11528</sup> (42%, Supplementary Figure S5), HopI1Pta<sup>11528</sup> (55%, Supplementary Figure S6) and HopF1Pta<sup>11528</sup> (49%, Supplementary Figure S7) showed significantly less homology with their respective Pto DC3000 homologs. A complete list of the effector genes used in this study is in **Table 1**.

# Several Pto DC3000 and Pta 11528 Effectors Elicit Cell Death in N. benthamiana

Mutations affecting individual effector genes typically have no or only subtle effects on bacterial pathogenicity, due to redundancy among effectors (Kvitko et al., 2009). To overcome this, we used Agrobacterium tumefaciens to transiently express individual effector genes in N. benthamiana (Hann and Rathjen, 2007). Western blot analysis showed that most effector proteins accumulated to detectable levels in N. benthamiana (**Table 1** and Supplementary Figure S8). We detected 25 out of 32 proteins at 2 dpi. An additional seven effectors that could not be detected by western blots suppressed defense responses in at least one assay performed in this work, indicating that these proteins were expressed at undetectable levels in the plant cell. Overall, there is evidence that all Pto DC3000 and Pta 11528 effector proteins accumulated within the plant cell after transient expression.

Four Pto DC3000 and three Pta 11528 effectors elicited various forms of necrosis when transiently expressed in N. benthamiana leaves, whereas no phenotype was observed in control tissue expressing an EV construct (**Figure 1A**). Leaves when transiently expressing hopAD1PtoDC3000, avrE1Pta11528, hopW1-1Pta11528, or hopT1-1Pta<sup>11528</sup> displayed strong cell death within two or three dpi. However, expression of avrE1Pta<sup>11528</sup> induced a stronger and faster necrosis compared to expression of hopAD1PtoDC3000, hopW1-1Pta11528, and hopT1-1Pta11528. In contrast, expression of hopM1PtoDC3000 induced only a patchy necrosis within 3 to 4 dpi, whereas hopAA1-1PtoDC3000 led to appearance of shiny areas, obvious only on the abaxial leaf surface that never developed into full necrosis. Leaves expressing hopQ1-1PtoDC3000 displayed only mild chlorosis within 4 days. In our experiments, such chlorotic areas did not normally progress further into necrosis. Finally, we observed mild chlorotic areas associated with avrPtoBPta<sup>11528</sup> and hopAR1Pta<sup>11528</sup> expression that appeared consistently at 7 dpi (data not shown). Effector protein accumulation did


List of Pto DC3000 and Pta 11528 effector proteins used in this study. + indicates positive detection of protein accumulation by western blotting (WB) when transiently delivered by A. tumefaciens in N. benthamiana leaves; A indicates detected effector activity in at least one assay in this study although no protein accumulation was detected.

not correlate with the intensity of necrosis. For example, neither AvrE1Pta<sup>11528</sup> nor HopAD1PtoDC3000 proteins could be detected by western blots, but they both induced fast and strong necrosis. Overall, 7 of 32 effector genes caused cell death when expressed transiently in N. benthamiana leaves.

Plant cell death is associated with effector recognition and subsequent HR in resistant plants as well as with the formation of lesions in susceptible plants (Alfano and Collmer, 2004). To investigate the nature of the cell death phenotype caused by individual expression of several Pto DC3000 and Pta 11528 effectors, we used VIGS to knock down the expression of NbSgt1 in N. benthamiana. NbSgt1 is required for NB-LRR proteins to trigger the defense-associated HR in response to effector recognition (Peart et al., 2002). Three weeks post-silencing, we transiently expressed hopM1PtoDC3000, hopAD1PtoDC3000, hopQ1-1PtoDC3000, avrE1Pta11528, hopT1-1Pta11528, and hopW1- 1Pta<sup>11528</sup> in EV- or NbSgt1-silenced plants. Because of the difficulty in scoring the mild phenotype of hopAA1-1PtoDC3000 expression, we excluded this gene from the assay. In addition, we co-expressed avrPtoPtoDC3000 and the protein kinase gene Pto as a control. Pto in complex with the NB-LRR protein Prf recognizes AvrPtoPtoDC3000 in vivo, leading to induction of a NbSgt1-dependent HR (Scofield et al., 1996; Tang et al., 1996; Peart et al., 2002; Mucyn et al., 2006). As expected, silencing of NbSgt1 compromised the cell death phenotype developed by the AvrPtoPtoDC3000/Pto control in N. benthamiana, and also by hopQ1-1PtoDC3000, which was reported previously to be NbSgt1-dependent (Wei et al., 2007). In contrast, cell death induced by any of the Pto DC3000 or Pta 11528 effectors was unaffected in both EV- and NbSgt1 silenced plants (**Figures 1B,C**). Thus, the necrosis induced by hopM1PtoDC3000, hopAD1PtoDC3000, avrE1Pta11528, hopT1- 1Pta11528, and hopW1-1Pta<sup>11528</sup> was not due to recognition by plant R genes. Taken together, these results suggest that NbSgt1 does not play a role in the cell death induced by these effectors.

# Pto DC3000 and Pta 11528 Effectors Redundantly Suppress NbSgt1-Independent Effector-Induced Cell Death

Some effector proteins can mask the ability of other effectors to trigger cell death (Jackson et al., 1999; Jamir et al., 2004; Guo et al., 2009). We investigated whether the cell death phenotypes induced by certain Pto DC3000 or Pta 11528 effectors could be suppressed by others encoded by these pathovars. We transiently co-delivered the necrosis-inducing effector hopAD1PtoDC3000 with each of the remaining effectors of the Pto DC3000 library. Among the 21 genes screened, 8 effectors completely blocked hopAD1PtoDC3000 induced cell death, whereas 7 additional genes significantly delayed the timing of necrosis appearance (**Figure 2A**). Effector genes that completely blocked cell death included avrPtoBPtoDC3000, hopT1-1PtoDC3000, hopF2PtoDC3000, hopK1PtoDC3000, hopQ1-1PtoDC3000, hopV1PtoDC3000, hopY1PtoDC3000, and hopD1PtoDC3000. In addition, effector genes that delayed the appearance of cell death were avrPtoPtoDC3000, hopO1-1PtoDC3000, hopA1PtoDC3000, hopI1PtoDC3000, hopG1PtoDC3000, hopN1PtoDC3000, and hopX1PtoDC3000. These effectors interfered not only with hopAD1PtoDC3000 but also with hopM1PtoDC3000 induced necrosis, with the exception of hopX1PtoDC3000 (Supplementary Figure S9). To test whether a similar activity is present in the Pta 11528 effector inventory, we screened these genes for the capacity to suppress hopT1- 1Pta<sup>11528</sup> induced cell death. Similar to previous results, most Pta 11528 effectors interfered with necrosis induced by this protein (**Figure 2B**). Expression of avrPtoBPta11528, hopF1Pta11528, hopX1Pta<sup>11528</sup> completely suppressed hopT1-1Pta<sup>11528</sup> induced cell death, whereas avrPtoPta11528, hopO1-1Pta11528, hopI1Pta11528, and hopAR1Pta<sup>11528</sup> delayed it. We next investigated whether the cell death phenotypes induced by the Pto DC3000 effectors hopAD1 and hopM1 could be suppressed by an effector encoded within the Pta 11528 repertoire, such as avrPtoB. The effector avrPtoBPta<sup>11528</sup> also interfered with the cell death phenotypes induced by both hopAD1PtoDC3000 and hopM1PtoDC3000 (Supplementary Figure S10). Therefore, our data indicate that the vast majority of Pto DC3000 and Pta 11528 effectors can interfere with cell death induced by certain effectors within the same bacterial pathovar and in the case of avrPtoBPta11528, also across bacterial pathovars.

hopW1-1Pta<sup>11528</sup> effector genes. Co-infiltrated avrPtoPtoDC3000 and Pto were included on the same leaf as a positive control for NbSgt1 silencing. Pictures were

We next analyzed the ability of Pto DC3000 effectors to antagonize specific R gene-dependent HR events in N. benthamiana. In tomato, the unrelated effectors AvrPtoPtoDC<sup>3000</sup> and AvrPtoBPtoDC<sup>3000</sup> trigger disease resistance in plants carrying Pto and Prf (Scofield et al., 1996; Tang et al., 1996; Kim et al., 2002; Mucyn et al., 2006). Thus, we next assessed the ability of transgenic N. benthamiana plants expressing the tomato Pto and Prf genes (N. benthamiana R411A) (Balmuth and Rathjen, 2007) for the ability to trigger HR upon infiltration of high bacterial densities of Pto DC3000, or isogenic strains lacking avrPtoPtoDC<sup>3000</sup> (Pto DC30001avrPto), avrPtoBPtoDC<sup>3000</sup> (Pto DC30001avrPtoB) or both (Pto DC30001avrPto1avrPtoB). We additionally included a Pto DC3000 strain lacking a functional TTSS required for effector secretion as a negative control (Pto DC3000 hrcC). Three days post-inoculation, Pto DC3000, Pto DC30001avrPto, and Pto DC30001avrPtoB elicited strong HRs (Supplementary Figure S11), indicating that the effectors secreted by Pto DC3000 do not interfere with the

taken 4 days after inoculation. Similar results were obtained in three independent experiments.

AvrPtoPtoDC<sup>3000</sup> and/or AvrPtoBPtoDC3000-dependent recognition by Pto/Prf.

# Multiple Pto DC3000 and Pta 11528 Effectors Target PAMP(s) Signaling Pathways at Different Steps

To identify suppressive effectors, we screened our collection for the ability to suppress the PAMP-dependent Ca2<sup>+</sup> and ROS bursts, and activation of defensive marker genes in N. benthamiana. For these experiments, we used two PAMPs; the flg22 peptide derived from bacterial flagellin (Chinchilla et al., 2007), and the oligosaccharide chitin which is a component of fungal cell walls. These two PAMPs are structurally unrelated and perceived by different receptor complexes (Boutrot and Zipfel, 2017). Each effector was expressed individually in N. benthamiana for 2 days and the harvested tissue subjected to treatments with either flg22 or chitin followed by one of

the three assays. The first was a ROS assay done in wildtype (WT) N. benthamiana leaves. The second assayed for an increase in cytosolic Ca2<sup>+</sup> concentrations using transgenic N. benthamiana plants expressing the 35S:Aequorin reporter (N. benthamiana SLJR15) (Segonzac et al., 2011). Finally, we assayed induction of the well-characterized defense genes NbCyp71D20 and NbACRE132 by quantitative RT-PCR (Navarro et al., 2004; Segonzac et al., 2011). The majority of Pto DC3000 and Pta 11528 effectors interfered with PAMP-induced responses and could be divided into three major groups based on their range of suppressive activities. Firstly, broad-range suppressors of all tested PAMP-induced early responses (Group A). Secondly, suppressors of the PAMP-induced ROS burst, without affecting Ca2<sup>+</sup> influx (Group B). And finally, effectors that suppressed PAMP-induced transcriptional activation of defense genes without affecting other outputs (Group C).

### Group A: Broad-Range Suppressors of PAMP-Induced Early Responses

As expected, leaf tissue expressing an EV control and treated with either flg22 or chitin led to an increase in cytosolic Ca2<sup>+</sup> levels (**Figure 3A**), the production of ROS (**Figure 3B**) and the strong induction of both NbCyp71D20 and NbACRE132 defensive marker genes (**Figures 3C,D**). We found that six Pto DC3000 and four Pta 11528 effectors compromised the activation of all tested responses to both PAMPs simultaneously (**Figure 3**). These effectors included the Pto DC3000 effectors AvrPtoPtoDC3000, AvrPtoBPtoDC3000, HopM1PtoDC3000, HopAD1PtoDC3000, HopAA1-1PtoDC3000 and HopQ1-1PtoDC3000, and the Pta 11528 effectors AvrPtoPta11528, AvrPtoBPta11528, HopT1-1Pta11528, and HopW1-1Pta11528. AvrEPta<sup>11528</sup> was omitted from these experiments because of the strong cell death phenotype caused by its expression. Interestingly, in contrast to the other effectors of this group, AvrPtoPtoDC3000 and AvrPtoPta<sup>11528</sup> only mildly suppressed or had no effect on chitin-induced defense responses, while they strongly compromised all flg22-induced outputs. Overall, 10 effectors from Pto DC3000 and Pta 11528 acted as early broad-range suppressors of PAMP-induced responses. These results suggest that multiple specificities and strategies may converge for simultaneous suppression of early PAMP-triggered immunity elicited by multiple microbial elicitors.

### Group B: Suppressors of PAMP-Induced ROS Burst

Apoplastic generation of ROS upon microbial perception plays a key role in the activation of disease resistance mechanisms in plants (Yoshioka et al., 2008). We found three effectors, HopT1- 1PtoDC3000, HopX1Pta11528, and HopAR1Pta<sup>11528</sup> that suppressed PAMP-induced ROS production but not the Ca2<sup>+</sup> influx (**Figures 4A,B**). HopT1-1PtoDC3000 did not distinguish between flg22 and chitin, whereas HopX1Pta<sup>11528</sup> and HopAR1Pta<sup>11528</sup>

FIGURE 3 | Group A effectors: broad-range suppressors of early PAMP-induced responses. (A) Ca2<sup>+</sup> burst in N. benthamiana SLJR15 transgenic plants transiently expressing an EV or each Group A effector as indicated after treatment with 100 nM flg22 or 100 µg/ml chitin. Data are presented relative (%) to control EV (PAMP-treated). Error bars represent the standard error of the mean (SEM; n = 8). Statistical significance compared to N. benthamiana tissue transiently expressing EV is indicated by asterisks (Student's t-test, <sup>∗</sup>p ≤ 0.01). (B) ROS burst in N. benthamiana plants transiently expressing an EV or each Group A effector from Pto DC3000 or Pta 11528 after treatment with 100 nM flg22 or 100 µg/ml chitin. Data are presented relative (%) to control EV (PAMP-treated). Error bars represent the standard error of the mean (SEM; n = 8). Statistical significance compared to N. benthamiana tissue transiently expressing EV is indicated by asterisks (Student's t-test, <sup>∗</sup>p ≤ 0.01). (C) Quantitative RT-PCR analysis of NbCyp71D20 and NbACRE132 gene expression 60 min after treatment with 100 nM flg22 in N. benthamiana leaf tissue transiently expressing an EV control or each Group A effector protein as indicated. A mock induction treatment in N. benthamiana leaf tissue transiently expressing EV (Mock EV) is included as a negative control. All samples were normalized against the housekeeping gene NbEF1α and the measurements represent the ratio of expression levels (%) compared to the flg22-induced EV sample. Error bars represent (SEM; n = 3). Statistical significance compared to PAMP-induced N. benthamiana tissue expressing EV is indicated by asterisks (Student's t-test, <sup>∗</sup>p ≤ 0.01). (D) Quantitative RT-PCR analysis of NbCyp71D20 and NbACRE132 defense gene expression 60 min after treatment with 100 µg/ml chitin in N. benthamiana leave tissue transiently expressing an EV control or each Group A effector protein as indicated. A mock induction treatment in N. benthamiana leaf tissue transiently expressing EV (Mock EV) is included as a negative control. All samples were normalized against the housekeeping gene NbEF1α and the measurements represent the ratio of expression levels (%) compared to the chitin-induced EV sample. Error bars represent (SEM; n = 3). Statistical significance compared to PAMP-induced N. benthamiana tissue expressing EV is indicated by asterisks (Student's t-test, <sup>∗</sup>p ≤ 0.01). The results shown in (A–D) are representative of three independent experiments. All effector genes were tested but those that didn't show any response of suppression are not represented.

suppressed chitin-induced ROS more efficiently than flg22 ROS. Only HopT1-1PtoDC3000 suppressed induction of the defense markers NbCyp71D20 and NbACRE132 by both elicitors, whereas HopX1Pta<sup>11528</sup> and HopAR1Pta<sup>11528</sup> did not suppress defense gene induction (**Figures 4C,D**). We designated these Group B effectors, based on their ability to suppress the ROS burst upon PAMP perception without interfering with the Ca2<sup>+</sup> influx. The specificity of HopX1Pta<sup>11528</sup> and HopAR1Pta<sup>11528</sup> for suppression of chitin-induced ROS suggests that the chitin and flg22 perception systems may differ in some key components required for ROS production. It is remarkable that HopT1-1PtoDC3000 and HopX1Pta<sup>11528</sup> behaved differently from their respective Pta 11528 and Pto DC3000 homologs. Compared to HopT1-1PtoDC3000, HopT1-1Pta<sup>11528</sup> acted as

a broad-range suppressor of early PAMP-induced responses (Group A), whereas HopX1PtoDC3000 did not compromise the chitin-triggered ROS burst that was typically suppressed in the presence of HopX1Pta11528. The capacity of several effectors to suppress ROS production upon PAMP sensing supports the importance of apoplastic ROS production in immunity to bacterial pathogens.

## Group C: Suppressors of PAMP-Induced Transcriptional Activation of Defense Genes

Recognition of microbial elicitors orchestrates an extensive defense-oriented transcriptional reprogramming of the affected cell (Navarro et al., 2004; Wan et al., 2008). In this study, we identified six effectors which downregulated flg22- and chitin-induced transcriptional activation of NbCyp71D20 and NbACRE132 genes. These included the Pto DC3000 effectors HopF2PtoDC3000, HopAF1PtoDC3000, HopI1PtoDC3000, and HopH1PtoDC3000, and the Pta 11528 effectors HopF1Pta<sup>11528</sup> and HopI1Pta11528. These effectors reduced rather than abolished gene induction, to a range of about 40–70% of the EV controls. Each effector caused very similar reductions in the expression of each marker gene. None of them compromised the Ca2<sup>+</sup> and ROS bursts produced after flg22 or chitin perception (**Figures 5A,B**). Among them, HopH1PtoDC3000 showed the strongest ability to compromised flg22- and chitin-induced expression of both NbCyp71D20 and NbACRE132 genes (**Figures 5C,D**), whereas the remaining effectors showed reproducibly intermediate levels of suppression. HopF1Pta<sup>11528</sup> and HopI1Pta<sup>11528</sup> consistently reduced flg22- and chitininduced gene expression to a similar extent as their Pto DC3000 homologs. The fact that effector proteins from both Pto DC3000 and Pta 11528 compromised PAMP-induced gene expression shows that this function is conserved between strains. In all cases, these effectors compromised transcriptional responses to both flg22 and chitin, suggesting that these pathways converge at some point downstream of each PRR complex.

# HopAD1 and HopM1 Compromise PAMP-Dependent ROS Production in the Presence of the Cell Death Blocker HopY1

Some Pto D3000 effectors both suppress PAMP responses and elicit cell death when expressed transiently in N. benthamiana leaves. In order to determine whether cell death caused the absence of PAMP-induced defense responses, we suppressed the hopAD1PtoDC3000 and hopM1PtoDC3000 induced necroses by co-expressing them with the previously described effector hopY1PtoDC3000 in N. benthamiana. This effector blocked the necroses triggered by hopAD1PtoDC3000 and hopM1PtoDC3000, but did not compromise PAMP responses in any of our assays. Both hopAD1PtoDC3000 and hopM1PtoDC3000 suppressed the ROS burst upon flg22 and chitin treatment to the same extent independent of the presence of hopY1PtoDC3000 (**Figure 6**), suggesting that ROS suppression upon PAMP treatment by these effectors is likely the outcome of their activities within the plant cell rather than due to the death of the infiltrated tissue.

# DISCUSSION

Bacterial pathogens use a TTSS to deliver effector proteins into eukaryotic cells. This mechanism enables the bacterium to grow to high levels and produce disease symptoms (Collmer et al., 2002). We found that 20 of the 22 Pto DC3000 effectors and 9 of 10 Pta 11528 effectors tested in this study showed suppressive activity in at least one of defense assay when overexpressed transiently using Agrobacterium-mediated transformation. This is consistent with the notion that a major role of effector proteins is to suppress plant innate immunity, and that they do so at different times and points in the signaling pathways. A comprehensive picture of the activity of effector molecules inside plant cells is summarized in **Figures 7**, **8**.

# Evolutionary Aspects of Type III Effectors Conserved in Both Pto DC3000 and Pta 11528

Comparing the activities of the effector repertoires of Pto DC3000 and Pta 11528 is a useful first step toward an understanding how these bacteria cause disease on different host species. We generated a library of 10 Pta 11528 effector genes, which constitute about half of the effectors present in the genome (Studholme et al., 2009). Seven of these effectors have homologs in the Pto DC3000 effector library analyzed in this work. However, the level of conservation varies among them. Some effectors show high levels of identity, including HopO1-1Pta<sup>11528</sup> and HopT1-1Pta11528, which are 99 and 98% identical at the amino acid level with their respective Pto DC3000 homologs (Supplementary Figures S1, S2). Putative catalytic functions identified by prediction programs are present in both the Pta 11528 and Pto DC3000 effector homologs. Therefore, it seems likely that they are conserved functionally. Despite this, the transgenic expression phenotypes of the Pta 11528 and Pto DC3000 HopT1-1 homologs differed significantly. While HopT1-1Pta<sup>11528</sup> elicited strong necrosis in N. benthamiana, HopT1-1PtoDC3000 did not. Although the effectors share high identity at the amino acid level, they differ in five residues that likely confer the basis for this behavior. Likewise, the cysteine protease HopX1Pta<sup>11528</sup> is highly similar to its Pto DC3000 homolog and catalytic residues required for putative enzymatic activity are conserved (Supplementary Figure S4). However, only the Pta 11528 version of HopX1 interfered with the chitin-induced ROS burst. It was previously reported that HopX1Pta<sup>11528</sup> but not HopX1PtoDC3000 targets conserved JAZ transcriptional repressors to activate jasmonate (JA) hormone signaling (Gimenez-Ibanez et al., 2014). In contrast to Pto DC3000, the Pta 11528 strain does not produce coronatine (COR), a phytotoxin that mimics the active form of JA, and therefore, exploits an alternative evolutionary strategy to activate the pathway through the HopX1Pta<sup>11528</sup> effector. HopX1 and COR biosynthetic genes generally do not co-exist in a single strain, but in the few cases where both occur, the HopX1 alleles contain mutations in functionally essential residues (Yang et al., 2017), suggesting that redundancy between COR and HopX1 might have inactivated and/or allowed the effector to evolve

plants transiently expressing an EV or each Group C effector as indicated after treatment with 100 nM flg22 or 100 µg/ml chitin. Data are presented relative (%) to control EV (PAMP-treated). Error bars represent the standard error of the mean (SEM; n = 8). Statistical significance compared to N. benthamiana tissue transiently expressing EV is indicated by asterisks (Student's t-test, <sup>∗</sup>p ≤ 0.01). (B) ROS burst in N. benthamiana plants transiently expressing an EV or each Group C effector as indicated after treatment with 100 nM flg22 or 100 µg/ml chitin. Data are presented relative (%) to control EV (PAMP-treated). Error bars represent the standard error of the mean (SEM; n = 8). Statistical significance compared to N. benthamiana tissue transiently expressing EV is indicated by asterisks (Student's t-test, <sup>∗</sup>p ≤ 0.01). (C) Quantitative RT-PCR analysis of NbCyp71D20 and NbACRE132 defense gene expression 60 min after treatment with 100 nM flg22 in

(Continued)

#### FIGURE 5 | Continued

N. benthamiana leaf tissue transiently expressing an EV control or each Group C effector as indicated. A mock induction treatment in N. benthamiana leaf tissue transiently expressing EV (Mock EV) is included as a negative control. All samples were normalized against the housekeeping gene NbEF1α and the measurements represent the ratio of expression levels (%) compared to the flg22-induced EV sample. Error bars represent (SEM; n = 3). Statistical significance compared to PAMP-induced N. benthamiana tissue expressing EV is indicated by asterisks (Student's t-test, <sup>∗</sup>p ≤ 0.01). Arrows indicate consistent reduction observed in independent experiments. (D) Quantitative RT-PCR analysis of NbCyp71D20 and NbACRE132 defense gene expression 60 min after treatment with 100 µg/ml chitin in N. benthamiana leaf tissue transiently expressing an EV control or each Group C effector proteins as indicated. A mock induction treatment in N. benthamiana leaf tissue transiently expressing EV (Mock EV) is included as a negative control. All samples were normalized against the housekeeping gene NbEF1α and the measurements represent the ratio of expression levels (%) compared to the chitin-induced EV sample. Error bars represent (SEM; n = 3). Statistical significance compared to PAMP-induced N. benthamiana tissue expressing EV is indicated by asterisks (Student's t-test, <sup>∗</sup>p ≤ 0.01). Arrows indicate consistent reduction observed in independent experiments. The results shown in (A–D) are representative of three independent experiments. All effector genes were tested but those that didn't show any response of suppression are not represented.

toward other functions. In contrast, the Pto DC3000 E3-ligase effector AvrPtoB also shares high level of amino acid identity (70%) with AvrPtoBPta<sup>11528</sup> (Supplementary Figure S3) and the amino acids critical for E3 ligase activity (F479A, T450, F525, and P533) are present in AvrPtoBPta11528. This suggests that AvrPtoBPta<sup>11528</sup> is also an active E3-ligase in plant cells and congruently, both Pto DC3000 and Pta 11528 homologs showed similar abilities to interfere with PAMP-triggered immunity (Janjusevic et al., 2006; Ntoukakis et al., 2009). Interestingly, although AvrPtoPta11528, HopI1Pta11528, and HopF1Pta<sup>11528</sup> show significantly less homology with their respective Pto DC3000 homologs, they showed functional conservation in PAMP-induced defense suppression assays (Supplementary Figures S5–S7). The putative homolog AvrPtoPta<sup>11528</sup> shares just 42% sequence identity at the amino acid level. The N- and C-terminal regions are highly conserved but there is significant diversity in the central region required for both the virulence and avirulence functions of this effector in tomato (Chang et al., 2001; Wulf et al., 2004; Xing et al., 2007). Similarly, HopI1Pta<sup>11528</sup> shares 55% identity with HopI1PtoDC3000 (Supplementary Figure S6), and both the N- and C-terminal regions are highly conserved maintaining a chloroplast targeting signal and the J domain required for virulence (Jelenska et al., 2007). However, 152 central amino acids containing the proline- and glutamine- (P/Q)-rich repeat region of unknown function are completely absent in HopI1Pta11528. Despite these differences, both the AvrPto and HopI1 homologs in Pto DC3000 and Pta 11528 strains showed similar abilities to suppress PAMP-induced defense responses, and thus the non-conserved regions may not be required for virulence functions in N. benthamiana. The differences between the Pta 11528 effectors and the Pto DC3000 homologs might indicate functional diversification as a consequence of divergent evolution, as well as diversification to escape recognition.

# Cell Death Promotion and Suppression by Pto DC3000 and Pta 11528 Effectors

Several Pto DC3000 and Pta 11528 effectors elicited cell death in N. benthamiana. Five effectors including hopM1PtoDC3000, hopAD1PtoDC3000, hopT1-1Pta11528, avrE1Pta11528, and hopW1- 1Pta<sup>11528</sup> strongly elicited necrosis whereas two effectors, hopAA1-1PtoDC3000 and hopQ1-1PtoDC3000 displayed mild chlorosis when expressed transiently in N. benthamiana leaves. This supports previous observations indicating that multiple P. syringae effectors trigger cell death in N. benthamiana

(Vinatzer et al., 2006; Wei et al., 2007; Wroblewski et al., 2009; Choi et al., 2017). For example, hopM1PtoDC3000, hopAA1- 1PtoDC3000, hopQ1-1PtoDC3000, avrE1PtoDC3000, hopT1-1 PtoDC3000, and hopAD1PtoDC3000 effectors elicit cell death in N. benthamiana when delivered by Pseudomonas fluorescens heterologously expressing a P. syringae TTSS (Wei et al., 2007) or though Pto DC3000 secretion effector polymutants (Wei et al., 2007, 2015). Similar to our experiments, transient expression of a number of effectors from Pto DC3000, P. syringae pv. actinidiae (Psa), and P. syringae pv. syringae (Psy) B728a, showed the ability of hopQ1-1, avrE1, hopT1-1, hopM1, and hopAA1-1 to induce cell death in N. benthamiana (Vinatzer et al., 2006; Wroblewski et al., 2009; Choi et al., 2017). It is noteworthy that in our assays, hopT1-1PtoDC<sup>3000</sup> did not trigger cell death in N. benthamiana, although this has been previously reported (Wei et al., 2007; Wroblewski et al., 2009).

Cell death is also associated with effector recognition by R proteins. This has been reported for HopQ1-1PtoDC3000, and to a much weaker extent for HopAD1PtoDC3000 and the Pma ES4326 effector HopW1-1 in N. benthamiana (Wei et al., 2007, 2015; Lee et al., 2008). HopQ1-1PtoDC3000 and HopAD1PtoDC3000 are recognized in N. benthamiana and consequently, Pto DC3000 lacking the hopQ1-1 or hopAD1 genes is capable of causing disease symptoms comparable to the virulent Pta 11528 (Wei et al., 2007, 2015). The R protein Roq1 mediates recognition of Xanthomonas and Pseudomonas effector proteins XopQ and HopQ1 in the Nicotiana genus (Schultink et al., 2017), whereas in Arabidopsis, a genetic interaction between Qpm3.1 and hopW1-1 determines resistance to bacterial infection (Luo et al., 2017). However, several lines of evidence suggest that host R proteins do not recognize all effectors that elicited cell death here. Firstly, necrosis induced by hopM1PtoDC3000, hopAD1PtoDC3000, hopT1-1Pta11528, avrE1Pta11528, and hopW1- 1Pta<sup>11528</sup> effector genes was not dependent on NbSgt1, which is typically required for R gene function (McDowell and Dangl, 2000). The NbSgt1-independence of avrE1 induced cell death has been reported previously (Choi et al., 2017). However, cell death induced by the Psy B728a effector hopM1 and the Psa effectors hopT1-1 and hopW1-1 were reported to be at least partially dependent on NbSgt1 (Vinatzer et al., 2006; Choi et al., 2017). Therefore, further analyses are required to determine to what extent the cell death caused by these effectors is dependent on NbSgt1 in N. benthamiana. On the other hand, NbSgt1 may not be required for all R gene-mediated plant defenses and thus, correlation of NbSgt1-dependency with effector recognition by R proteins should be undertaken cautiously. Secondly, Pta 11528 is virulent on N. benthamiana despite possessing a HopW1-1 homolog. It is possible that some of these effectors would not elicit cell death when expressed under native conditions, i.e., in bacteria and delivered to the plant cell via the TTSS. And thirdly,

death in N. benthamiana and the activation of PAMP-dependent defense responses including the Ca2<sup>+</sup> influx, generation of ROS and defense gene expression. Dark gray indicates complete suppression. Light gray indicates significant reduction compared to positive control. White indicates no suppression. ND, not done. The cell death suppression box indicates those effectors for which cell death was compromised, namely hopAD1PtoDC3000 or hopM1PtoDC3000 for Pto DC3000, and hopT1-1Pta<sup>11528</sup> for Pta 11528.

there is precedence for host cell death to be associated with effector virulence, as demonstrated for members of the AvrE1 and HopM1 effector families (Badel et al., 2006; Boureau et al., 2006; Ham et al., 2006). Both avrE1 and hopM1 are functionally redundant and reside in the conserved effector locus (CEL) of the P. syringae Hrp pathogenicity island (DebRoy et al., 2004; Badel et al., 2006; Kvitko et al., 2009). HopM1 and AvrE1 are required to establish a water-soaked aqueous living space in plants that is crucial for virulence (Xin et al., 2016). Whether HopAA1-1, another effector encoded in the CEL locus, or the remaining cell death-inducing effectors of Pto DC3000 and Pta 11528 function similarly to HopM1/AvrE in altering apoplastic environmental conditions is unknown. The conservation of necrosis-inducing effectors in various bacteria suggests this function may be an important virulence strategy for bacterial infections on plants.

A number of Pto DC3000 and Pta 11528 effectors masked the activity of cell death-inducing effectors when co-expressed transiently in N. benthamiana. Similar data exist for suppression of effector-triggered cell death phenotypes by Pto DC3000 and Psa effectors (Jamir et al., 2004; Guo et al., 2009; Wei et al., 2015; Choi et al., 2017). Indeed, most Pto DC3000 effectors that acted as cell death suppressors in this work where previously identified as suppressors of the HR response induced by HopA1 in tobacco when delivered by Pseudomonas fluorescens heterologously expressing a P. syringae TTSS (Guo et al., 2009). Notably, it has been reported recently that HopQ1 mediated cell death suppression in N. benthamiana is due to attenuation of Agrobacterium-mediated protein expression rather than a genuine virulence activity. Thus, it is possible that additional effectors may act in a similar fashion (Adlung and Bonas, 2017). We speculate that necrosis induction by effectors must be tightly regulated to establish and maintain an optimal intracellular niche. Although it is possible that some effectors would not interfere with cell death when expressed at native levels, the potent activity of some of them suggests that cell death suppression may occur during infection to at least some extent. On the other hand, the ability of some effectors to interfere with the outcomes of others identifies which effectors perturbate common defensive pathways. HopAD1PtoDC3000-, HopM1PtoDC3000-, and HopT1-1Pta11528-dependent cell death could be efficiently suppressed by multiple effectors from both Pto DC3000 and Pta 11528 repertoires, which include both AvrPtoB homologs. This supports recent work by Wei et al. (2015) indicating interplay between HopAD1 and AvrPtoB in regulating HopAD1-dependent cell death. Despite this, Pto DC3000 effectors could not collectively suppress the HR events activated upon recognition of AvrPtoB and/or AvrPto in transgenic N. benthamiana plants expressing the tomato Pto and Prf resistance genes (Balmuth and Rathjen, 2007). In addition, neither the effector complements of Pto DC3000 or Pta 11528 were able to collectively suppress HopQ1-1 or HopAD1 induced defenses in N. benthamiana (Wei et al., 2007, 2015). Interplay between effectors within the bacterial repertoires is emerging as an important but poorly understood phenomenon controlling defensive phenotypic outcomes (Wei et al., 2015; Huett, 2017). More studies are needed in this area to understand the complex network of interactions between the effectors among the repertoires and how this translates into an effective immune response or a virulence strategy to promote disease.

# Pseudomonas syringae Effectors Differentially Suppress Innate Immune Responses

In recent years, many effectors have been shown to suppress PAMP-induced defenses (Macho and Zipfel, 2015; Buttner, 2016; Toruno et al., 2016). Here, we analyzed the ability of effectors to interfere with PAMP-triggered defenses activated between seconds to minutes to hours after elicitation, which has allowed us to group effectors into three major categories based on their range of suppressive activities.

Group A effectors acted as broad-range suppressors of all early PAMP-induced responses tested. AvrPtoPtoDC3000 and AvrPtoBPtoDC3000 which fall into this group, are well known suppressors of flg22-induced defenses by targeting PRR complexes directly at the plasma membrane (Chinchilla et al., 2007; Heese et al., 2007; Gohre et al., 2008; Xiang et al., 2008). The additional Pto DC3000 and Pta 11528 effectors within Group A may also target PRR complexes due to their comprehensive suppression activities. Indeed, direct inactivation or destabilization of PAMP receptor complexes at the plasma membrane is a common strategy of multiple effectors (Macho and Zipfel, 2015; Buttner, 2016; Toruno et al., 2016). HopM1PtoDC3000 suppresses vesicle trafficking to overcome host innate immunity (Nomura et al., 2006). Our data indicates additional roles for this effector in early PAMP signaling suppression. Both HopM1PtoDC3000 and HopQ1-1PtoDC3000 associate with multiple 14-3-3 proteins (Nomura et al., 2006; Giska et al., 2013; Li et al., 2013; Lozano-Duran et al., 2014). 14-3-3 proteins are highly conserved eukaryotic regulatory adapters whose interaction with client proteins can regulate protein activity and remarkably, chemical inhibition of 14-3-3s results in suppression of the PAMP-triggered ROS burst (Lozano-Duran et al., 2014). Some 14-3-3 proteins associate with several defense-related proteins in planta, including the FLS2 co-receptor BAK1 (Chang et al., 2009). Whether HopM1PtoDC3000 and HopQ1-1PtoDC3000 act through interaction with 14-3-3s remains to be explored.

Group B effectors suppressed the PAMP-induced ROS burst not the Ca2<sup>+</sup> influx. These included HopT1-1PtoDC3000, HopX1Pta11528, and HopAR1Pta11528. These effectors abolished the PAMP-induced ROS burst with differential impact on activation of defense gene markers. While HopT1-1PtoDC3000 compromised flg22- and chitin-dependent induction of defense genes, HopX1Pta<sup>11528</sup> and HopAR1Pta<sup>11528</sup> did not. Furthermore, their ROS-suppressive action was restricted to chitin-induced responses. The P. syringae pv. phaseolicola effector HopAR1 suppresses PTI by directly targeting BIK1 and other RLCKs (Zhang et al., 2010), which contrast with our results. However, the hopAR1Pta<sup>11528</sup> and hopAR1Pphrace<sup>3</sup> versions of this effector differ significantly in their protein sequences. HopAR1Pta<sup>11528</sup> contains an extra 44 N-terminal amino acids (in a relatively small effector of 311 amino acids), whereas the common C-terminal domain shares a moderate 78% identity, which may account for the observed differences. In an alternative model, HopX1Pta<sup>11528</sup>

targets conserved JAZ transcriptional repressors to activate JA signaling and promote infection in Arabidopsis (Gimenez-Ibanez et al., 2014), and therefore, additional targets would be expected for this effector. This is unsurprising as increasing evidence suggests that effector proteins target multiple host proteins simultaneously in a "death by a thousand cuts" strategy to abolish defense responses (Navarro et al., 2008; Shan et al., 2008; Gimenez-Ibanez et al., 2009).

Group C effectors are suppressors of PAMP-induced gene induction. This group only compromised PAMP-induced gene expression but not the other outputs. They include the Pto DC3000 effectors HopF2PtoDC3000, HopAF1PtoDC3000, HopI1PtoDC3000, and HopH1PtoDC3000, and the Pta 11528 effectors HopF1Pta<sup>11528</sup> and HopI1Pta11528. Because this group suppressed gene induction but not plasma membrane-related events, we suggest that they act later in signal transduction perhaps proximal to or within the nucleus. In this group falls HopF2PtoDC3000, which was previously shown to suppress Arabidopsis immunity by targeting the co-receptor BAK1 (Zhou et al., 2014). This finding is difficult to reconcile with our data as this effector did not suppress early flg22-induced Ca2<sup>+</sup> and ROS bursts in our experiments to any extent. This is similar to HopB1PtoDC3000, one of the few effectors that did not suppress any of the defensive outputs tested in this work, despite the fact that it acts as a protease that targets BAK1 (Li et al., 2016). On the other hand, HopAF1PtoDC3000 suppresses ethylene production (Washington et al., 2016) whereas HopI1 targets plant heat shock chaperone protein Hsp70 (Jelenska et al., 2010). None of these functions explains the ability of these effectors to suppress PAMP-induced gene expression. Thus, we expect alternative targets for these effectors to disrupt transcriptional activation. Effectors that interfere with the transcriptional machinery are not yet known from Pto DC3000, but are well described in other pathogens such as Xanthomonas which produces transcription activator-like effectors (TALEs) (Boch et al., 2009; Moscou and Bogdanove, 2009). Whether Pto DC3000 also injects direct modulators of host gene expression remains an open question for the future.

# Most Effectors Interfere With Responses Elicited by Different PAMPs

Most effectors interfered with both flg22- and chitin-induced defense responses simultaneously, indicating that a considerable overlap exists between these pathways. One exception was the AvrPto homologs of both Pto DC3000 and Pta 11528 which suppressed flagellin signaling but not chitin responses. This was surprising because AvrPtoPtoDC3000 was previously reported to block chitin-induced defense gene expression in transgenic Arabidopsis protoplasts (Shan et al., 2008). It is possible that the chitin-suppressing effect of AvrPtoPtoDC3000 was simply due to overexpression in Arabidopsis. Alternatively, the kinase domain of the Arabidopsis chitin receptor/receptor complex might be simply a better target for AvrPtoPtoDC3000 than its N. benthamiana homolog (Shan et al., 2008; Xiang et al., 2008). Therefore, it remains to be determined how chitininduced signaling in N. benthamiana evades AvrPto-mediated suppression. Conversely, HopX1Pta<sup>11528</sup> and HopAR1Pta<sup>11528</sup> suppressed chitin-induced ROS generation, but not that generated by flg22 treatment. This might be simply a consequence of the different strength of ROS levels induced by both elicitors. While flg22 induces a very strong response, the chitin response is relatively mild. Therefore, smaller effects might be more easily revealed using chitin as an elicitor. Overall, much of our data suggest that chitin- and flg22-induced signaling converge soon after PAMP perception, since most effectors suppressed defense responses induced by both elicitors.

# Final Considerations

Nicotiana benthamiana has emerged as an important and widely used experimental system to study plant biology. Its advantages include its large leaves, its high amenability for A. tumefaciensmediated transient expression, the possibility to rapidly test the involvement of host factors using VIGS, and the reproducibility of the data (77, 78). Agrobacterium-mediated transient expression has already been used to study some effectors that suppress plant defense responses at the molecular level (24, 79). Thus, the efficiency and versatility of the agroinfiltration technique in N. benthamiana prompted us to investigate the ability of a large number of P. syringae type III effectors to suppress a collection of defense responses activated upon PAMP treatment. However, it should be taken into account that overexpression of effector proteins can led to misleading results by causing protein imbalances, promiscuous interactions, and regulation of pathways that are associated with the degree of overexpression rather than the function of the protein. On the other hand, overexpression of effector proteins in plant cells, combined with loss-of-function analyses, is a common and powerful approach to understanding effector function. Identification of effector targets is the next step to unraveling how pathogens can overcome plant immunity and promote pathogenesis. This will lead ultimately to new strategies for crop protection in the field.

# AUTHOR CONTRIBUTIONS

DH, JR, and SG-I designed the experiments. DH, CS, and SG-I conducted the experiments. DH, JR, and SG-I analyzed the data. JR, JC, TB, and SG-I wrote the manuscript. All authors edited and approved the final manuscript.

# FUNDING

This work was funded by the Gatsby Charitable Foundation and the Spanish Ministry of Science and Innovation Grant BIO2014- 55884-JIN to SG-I.

# ACKNOWLEDGMENTS

We thank Professor Dr. Jeffery L. Dangl for providing fulllength Pto DC3000 effectors recombined into pDONR207 (Chang et al., 2005) and Dr. Gregory Martin for the Pto

DC30001avrPtoB1avrPto strain. In addition we thank Dr. Birgit Schulze for careful revision of the manuscript and Anna Egger for technical support. We acknowledge support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI).

# REFERENCES


# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00688/ full#supplementary-material


subsp. stewartii, causes disease-associated cell death in corn and requires a chaperone protein for stability. Mol. Plant Microbe Interact. 19, 1092–1102. doi: 10.1094/MPMI-19-1092


plant immunity. Cell Host Microbe 15, 329–338. doi: 10.1016/j.chom.2014. 02.009



effector polymutants reveal an interplay between HopAD1 and AvrPtoB. Cell Host Microbe 17, 752–762. doi: 10.1016/j.chom.2015.05.007


**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 BK declared a past co-authorship with one of the authors JC to the handling Editor.

Copyright © 2018 Gimenez-Ibanez, Hann, Chang, Segonzac, Boller and Rathjen. 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.

# Glycans as Modulators of Plant Defense Against Filamentous Pathogens

#### Chayanika Chaliha<sup>1</sup> , Michael D. Rugen<sup>2</sup> , Robert A. Field<sup>2</sup> and Eeshan Kalita1,2 \*

<sup>1</sup> Department of Molecular Biology and Biotechnology, Tezpur University, Tezpur, India, <sup>2</sup> Department of Biological Chemistry, John Innes Centre, Norwich Research Park, Norwich, United Kingdom

Plants and microbes utilize glycoconjugates as structural entities, energy reserves for cellular processes, and components of cellular recognition or binding events. The structural heterogeneity of carbohydrates in such systems is a result of the ability of the carbohydrate biosynthetic enzymes to reorient sugar monomers in a variety of forms, generating highly complex, linear, branched, or hierarchical structures. During the interaction between plants and their microbial pathogens, the microbial cell surface glycans, cell wall derived glycans, and glycoproteins stimulate the signaling cascades of plant immune responses, through a series of specific or broad spectrum recognition events. The microbial glycan-induced plant immune responses and the downstream modifications observed in host-plant glycan structures that combat the microbial attack have garnered immense interest among scientists in recent times. This has been enabled by technological advancements in the field of glycobiology, making it possible to study the ongoing co-evolution of the microbial and the corresponding host glycan structures, in greater detail. The new glycan analogs emerging in this evolutionary arms race brings about a fresh perspective to our understanding of plant–pathogen interactions. This review discusses the role of diverse classes of glycans and their derivatives including simple sugars, oligosaccharides, glycoproteins, and glycolipids in relation to the activation of classical Pattern-Triggered Immunity (PTI) and Effector-Triggered Immunity (ETI) defense responses in plants. While primarily encompassing the biological roles of glycans in modulating plant defense responses, this review categorizes glycans based on their structure, thereby enabling parallels to be drawn to other areas of glycobiology. Further, we examine how these molecules are currently being used to develop new bio-active molecules, potent as priming agents to stimulate plant defense response and as templates for designing environmentally friendly foliar sprays for plant protection.

Keywords: carbohydrates, glycans, elicitors, priming, plant defense, fungi, oomycete

# INTRODUCTION

Plant cell walls are complex configurations of highly recalcitrant interlocking polysaccharides which insulate against microbial invasion and abiotic stress. Filamentous plant pathogens, which mainly comprise fungi and oomycetes, breach the plant cell wall by releasing enzymes that deconstruct polysaccharides, proteins, and lignin based polymers (Zhao et al., 2013). The resulting breakdown products that accumulate in apoplastic fluids represent the first molecular interaction

#### Edited by:

Thomas Mitchell, The Ohio State University, United States

#### Reviewed by:

William Underwood, Agricultural Research Service (USDA), United States Wei Zeng, The University of Melbourne, Australia

> \*Correspondence: Eeshan Kalita ekalita@tezu.ernet.in

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 13 February 2018 Accepted: 11 June 2018 Published: 04 July 2018

### Citation:

Chaliha C, Rugen MD, Field RA and Kalita E (2018) Glycans as Modulators of Plant Defense Against Filamentous Pathogens. Front. Plant Sci. 9:928. doi: 10.3389/fpls.2018.00928

between microbes and the plant. As such, the apoplast also represents the zone where the plants need to distinguish between symbionts and pathogens, based on their molecular signatures (Okmen and Doehlemann, 2016).

The activation of the complex array of plant innate defense mechanisms relies on the recognition of pathogen signatures by the host transmembrane pattern recognition receptors (PRRs) (Zipfel, 2014). The signatures perceived by PRRs are conserved molecular patterns called PAMPs/MAMPs (Pathogen or Microbe Associated Molecular Patterns) which are either the breakdown products resulting from microbial enzyme action or pathogenic effectors secreted by microbes to promote infection (**Figures 1**, **2** and **Table 1**; Trouvelot et al., 2014). However, pathogens also secrete effectors to mask the PTI response resulting from PAMP recognition by the host. The secretion of these effectors mediate the remodeling of the microbial cell wall due to which the pathogen is able to either escape the host defense or trigger the ETI responses (Hopke et al., 2018). This has been seen in mammalian systems, yet its importance in carbohydrate mediated interactions between plants and microbes remains to be determined.

Although PAMPs represent a broad range of molecules, which includes carbohydrates, lipids, proteins, peptides, lipopolysaccharides, glycolipids, and glycoproteins (Boller and Felix, 2009; Trouvelot et al., 2014), it is only during recent times that advances in functional glycomics have encouraged researchers to analyze the role of glycans in plant–pathogen interactions. The threat to food security from filamentous pathogens, which rely on a host of glycans to mediate pathogenesis, accounting for an estimated loss of 10–40% of total crop production worldwide (Sobhy et al., 2014; Anderson et al., 2016) has driven intense research in this area. In turn, this had led to concepts like "sweet immunity" and "sugar enhanced defense" that explore the multi-faceted and systemic role of carbohydrates as modulators of plant immunity (Trouvelot et al., 2014; Rovenich et al., 2016).

In the context of glycan-mediated plant immunity, identification of microbial cell-surface glycans, building blocks of fungal/oomycetes cell walls (e.g., chitin and β-glucans) and bacterial glycoconjugates (e.g., lipopolysaccharide, glycoproteins, and lectins) that act as regulators of plant defense signaling, presents new perspectives to analyze and understand host– microbe interactions (Vidal et al., 1998; Boudart et al., 2003; Trouvelot et al., 2014). Within this review, we look at how the different classes of glycans become a part of the strategic interactions during plant–microbe interaction and their future potential as defense priming agents for plant protection. The review covers a broad range of topics, providing a brief insight that can act as a primer to an audience unfamiliar with this topic alongside those studying plant–pathogen interaction.

# SIMPLE SUGARS: MONO AND DI-SACCHARIDES

Simple sugars including mono and di-saccharides are central to plant–microbe interactions, serving both as energy sources to drive the PTI and ETI responses as well as themselves acting as signaling molecules that drive signal fluxes leading to localized or systemic defense responses, when challenged with filamentous pathogens (Trouvelot et al., 2014).

# Glucose, Sucrose, and Associated Metabolites

The role of the glucose sensor Hexokinase (HXK), which is responsible for the conversion of glucose to glucose 6-phosphate, has been most investigated. Among the several isoforms of HXK1, the mitochondria-associated HXK1 is central to the control of programmed cell death (PCD) during microbial pathogenesis and is also responsible for regulation of several pathogenesis related (PR) genes, during HR mediated cell death (Kim et al., 2006). Glucose also activates the expression of several PR genes, of which some may require the HXK to be catalytically active. For instance, glucose mediated the induction of PR-1 and PR-5 in Arabidopsis in an AtHXK1 dependent manner and also Arabidopsis lines overexpressing mitochondrial HXK have higher basal transcript levels of PR genes showing enhanced resistance to the necrotrophic fungal pathogen Alternaria brassicicola (Xiao et al., 2000; Rojas et al., 2014).

Sucrose has emerged as an important molecule in plant sugar signaling networks owing to recent evidences of its involvement in the modulation of innate immunity and defense responses during microbial attack (Gomez-Ariza et al., 2007; Bolouri Moghaddam and Van den Ende, 2012). It is suggested that cellwall localized invertases hydrolyze sucrose to generate glucose, which in turn act as signal fluxes that are sensed by HXKs to activate downstream defense signaling (Moore et al., 2003; Cho et al., 2009; Tauzin and Giardina, 2014). Additionally, sucrose has been seen to drive the expression of secondary metabolite synthesis pathways including anthocyanin and isoflavonoid production, as a defense response against Fusarium oxysporum in Lupinus angustifolius and in embryo axes of Lupinus luteus L. cv. Juno (Morkunas et al., 2005; Formela et al., 2014).

Trehalose and trehalose-6-phosphate (T-6-P) are considered important sugar signals, modulating defense responses through complex sugar sensing pathways (**Figure 3**). Trehalose induces the activation of the defense genes Phenylalanine Ammonia-Lyase (PAL) and Peroxidase (POX) during wheat challenge with Blumeria graminis, the causal agent of powdery mildew disease (Reignault et al., 2001; Muchembled et al., 2006). Plant cell derived T-6-P acts as a molecular switch that negatively regulates the activity of the sucrose non-fermenting related protein kinase 1 (SnRK1). SnRK1 is a master regulator that controls sugar metabolism during biotic stresses (Hulsmans et al., 2016). KIN 10 and KIN 11 the Arabidopsis analogs of SnRK1, which have been seen to be functional under both types of stress, establish a link between sugar metabolism and the metabolic disruptions seen under pathogen attack. Rice cultivars sensitive to Magnaporthe grisea, for instance, were seen to have fewer metabolites involved in sugar metabolism, compared to the resistant varieties (Jones et al., 2011). It was later shown by Baena-Gonzalez (2010) that SnRK1 is the primary modulator repressing the energy intensive synthetic pathways and activating

are recognized as PAMPs by host pattern recognition receptors (PRRs) triggering innate immune response. (C) The pathogen secretes effector molecules that (D) suppress PTI defense responses by (E) converting immunogenic chitin to the less immunogenic chitosan (F) sequestering or masking the PAMPs released, to evade detection. (G) Remodeling of the cell wall components by the pathogen (e.g., accumulation of α-1,3-glucan) mitigates the effect of host-hydrolytic enzymes on the microbial cell wall glycans, thereby preventing hydrolysis. (H) Lastly, the pathogens may secrete certain effectors that directly inhibit host hydrolytic enzymes. (Adapted from Rovenich et al., 2016 and reproduced with permission from John Wiley and Sons).

catabolic pathways under stressed conditions, to restore cellular homeostasis . Interestingly, the effect of SnRK1 activation, under abiotic stresses could be reversed by the exogenous application of sucrose. Inhibition was also seen upon the external application of glucose and glucose-6-phosphate (G-6-P) (Zhang et al., 2009; Nunes et al., 2013). This repression was attributed to the fact that drops in levels of sucrose, glucose, G-6-P and T-6-P were acting as starvation signals which in turn induced SnRK1. In Arabidopsis seedlings, the increased T-6-P levels act as a "feast signal" suppressing SnRK1 activity (Morkunas and Ratajczak, 2014).

SnRK1 has also been shown to drive the expression of stress inducible genes in plants aided by the heterodimerization of S-group and C-group bZIP transcription factors (for basic region/Leu zipper motifs) leading to protection against pathogens via PCD. The response of the transcription factors to sugar levels is, however, variable as seen for Arabidopsis AtbZIP11, which is sugar inducible, while AtbZIP1, AtbZIP2, and AtbZIP53 are sugar repressible. As new insights are drawn regarding the role of sugars in modulating the metabolic state and defense responses under pathogen attack, gaps in the understanding of signaling pathways continue to emerge. With respect to sucrose, glucose, G-6-P, and T-6-P there is sufficient evidence to suggest that they function as signals reflecting the general metabolic health of the plants and also behave as systemic defense modulators in certain cases. However, various questions remain unanswered regarding

the site and mechanism of action for these glycans, as well as their holistic roles in plant defense.

# Galactinol, Raffinose, and Related Oligosaccharides

Galactinol in plants is present as a precursor for the synthesis of the raffinose family oligosaccharides (**Figure 4**), which serve as an osmoprotectant of plants, as a transporter of sugar in phloem sap, and as storage sugars (Sengupta et al., 2015). Soybean plants under drought stress conditions, when pretreated with H2O2, were seen to have increased levels of galactinol which is known to function as an ROS scavenger (Ishibashi et al., 2011; Savvides et al., 2016). A similar phenomenon was also observed in Arabidopsis where oxidative stress induced the expression of high levels of galactinol synthase (GolS) (Nishizawa et al., 2008; Petrov et al., 2015). This enzyme catalyzes the first step toward the biosynthesis of galactinol which was first detected in a crude extract of maturing pea seeds (Zhou et al., 2017). The induction of the GolS gene leading to the biosynthesis of galactinol has mostly been studied in relation to osmotic stresses across various plants such as Brassica napus L., Coffea arabica L., Salvia miltiorrhiza, grapevine, Medicago falcata, chestnut, and Cicer arietinum L. The first evidence of galactinol being involved in inducing systemic resistance during biotic stress was observed during the interaction between Pseudomonas chlororaphis and cucumber or tobacco plants. Galactinol enhanced the accumulation of defenserelated genes PR1a, PR1b, and NtACS1 (Nicotiana tabacum 1-aminocyclopropane-1-carboxylic acid synthase 1), which in turn enhanced resistance against Botrytis cinerea and Erwinia carotovora in pathogen challenged cucumber and tobacco plants (Kim et al., 2008). Preliminary investigations on Camellia sinensis GolS gene (CSGolS) have shown that CsGolS1 regulation was mainly related to abiotic stresses such as water deficiency, low temperature, and ABA treatment. On the other hand, significant regulation of CsGolS2 and CsGolS3 was seen in relation to


biotic stresses such as E. oblique (moth) attack, SA, and MeJA treatment (Zhou et al., 2017). Although the GolS gene has been extensively studied under abiotic stress, the regulation of the galactinol levels by this gene during biotic stress (pest attack and microbial invasion) needs to be further investigated to understand the underlying mechanisms. Furthermore, the roles played by galactinol derivatives in response to stresses including pathogen attack remain to be elucidated (Sengupta et al., 2015).

# Polyols

Polyols, often referred to as sugar alcohols, are the reduced form of aldose and ketose sugars (Noiraud et al., 2001). Mannitol is a soluble polyol which is structurally related to the aldohexose mannose. It is widely present in bacteria, algae, fungi, and more than 100 species of higher plants including many crops such as celery, olive, and carrot (Patel and Williamson, 2016). Mannitol is an osmoprotectant, contributing to salt tolerance in plants and it has been demonstrated to be an in vitro quencher of reactive oxygen species that limits cell damage (Jennings et al., 2002). Interestingly, mannitol is also a common metabolite in most filamentous plant pathogens including ascomycetes, basidiomycetes, deuteromycetes, and zygomycetes (Jennings, 1985; Solomon et al., 2007). The fungal pathogens Alternaria alternata, A. brassicicola, and Cladosporium fulvum are reported to secrete mannitol when they encounter plant tissues or extracts (Williamson et al., 2013). Mannitol accumulation was first documented in the apoplast of tomato leaves upon infection with a virulent strain of C. fulvum, but was absent in leaves infected with the avirulent stain (Joosten et al., 1990). The role of mannitol as a pathogenicity factor was further consolidated by reports stating greatly reduced virulence of A. alternate mannitol synthesis knockout mutants (1mtdh/1m1pdh) (Vélëz et al., 2007). In light of such reports, mannitol is speculated to be involved in quenching ROS-mediated plant defenses by virtue of its antioxidant properties (Calmes et al., 2013). The current opinion in this regard suggests that mannitol secretion by filamentous pathogens serves as a self-defense strategy against plant defense responses and may also be a means of obstructing the plant defense signaling pathways by blocking plant mediated ROS signals (Juchaux-Cachau et al., 2007). This aspect has been recently reviewed by Patel and Williamson (2016) to present a mechanistic hypothesis (**Figure 5**) regarding the interplay between mannitol and mannitol dehydrogenase during pathogenesis of the mannitol-secreting pathogens and how it may be balanced between detection of pathogens by plants and evasion of defense response by the pathogen.

# OLIGOSACCHARIDES

Most oligosaccharides implicated in plant–pathogen interactions are generated by the enzymatic degradation of polysaccharides from the structural constituents of fungal cell wall or pathogen virulence factors (Zhang S. et al., 2015). The activity of these oligosaccharides is highly dependent on their degree of polymerization (DP) (Trouvelot et al., 2014).

# β-1,3-/ β-1,6-glucans

β-1,3-/ β-1,6-Glucans based oligosaccharides have been extensively explored in view of their involvement in plant– pathogen interactions for several decades (Fesel and Zuccaro, 2016). β-glucans were first isolated from crude or enriched fractions of Phytophthora sojae mycelial cell wall hydrolysate, which mediated plant defense by the induction of PAL activity in plants (Ayers et al., 1976; Ebel et al., 1976). In tobacco and Arabidopsis, sulfated β-1,3-glucans were able to induce a SA mediated defense reactions conferring resistance to tobacco

against pathogens through post translational modification of key metabolic enzymes, activating PCD and bZIP (basic leucine zipper) mediated transcriptional reprogramming. Trehalose on the other hand can also sense pathogen stress and respond by activating defense genes and blocking the energy intensive starch biosynthesis. Sucrose regulates defense signaling both positively and negatively by activating secondary metabolite production under low sugar conditions on one hand and inhibiting SnRK1 on the other, during normal conditions. Additionally, the mitochondrial HXKs are also implicated in activating PCD and defense gene activation during pathogen attack. However, little is known if Glucose mediates the process. Thus, the components of signalling pathway of sugars (like sucrose, glucose and trehalose) that maintains the balance between stress conditions and homeostatsis are yet to be discovered and are indicated by "?" in the figure.

mosaic viruses (Menard et al., 2004). Laminarin, a β-1,3 glucan, derived from the brown algae Laminaria digitata was seen to elicit a variety of defense reactions in tobacco plants. Treatment of tobacco with laminarin as an elicitor led to the enhanced activity of PAL, caffeic acid, O-methyl transferase, and lipoxygenase, accumulation of SA, and transcriptional activation

with permission from Elsevier).

of PR proteins. These events resulted in activation of induced resistance against the soft rot pathogen E. carotovora (Klarzynski et al., 2000). Using laminarin elicitors on grapevine induced early defense responses including calcium influx, oxidative burst, extracellular alkalization of the culture medium, and activation of mitogen-activated protein kinases (MAPK) (Aziz et al., 2003). Laminarin in later stages of host defense induced the expression of defense genes associated with the octadecanoid and phenylpropanoid pathways leading to a significant protection of grapevine leaves against B. cinerea and Phomopsis viticola. However, HR mediated cell death was totally absent in these plants (Aziz et al., 2003). In most cases β-1,3-glucans and laminarin have DPs between 10 and 16 which are often referred to as optimal for induction of plant defense responses (Navazio et al., 2002; Galletti et al., 2008; Vorholter et al., 2012). However, Fu et al. (2011) have demonstrated using tobacco cell suspensions that β-1,3-glucans having DPs as low as 2–10, provide a higher protection against tobacco mosaic virus compared to those with higher DP (25–40) (Fu et al., 2011). Similarly, in Oryza sativa cell suspensions β-1,3-glucan oligomers with a DP ≥4 were shown to stimulate chitinase activity while those with DP 6 acted as inducers of PAL activity (Inui et al., 1997).

Plants have been seen to respond differently to structurally distinct forms of β-glucans, as seen in soybean and rice which are able to recognize only branched β-glucans (Cheong and Hahn, 1991; Yamaguchi et al., 2013) whereas tobacco recognizes the linear β-1,3-glucans. With respect to the response to chemically modified glucans as elicitors, acetylated oligoglucuronans having DP ≥14 could induce the transient production of H2O<sup>2</sup> and defense gene expression (PAL, Chitinase and Polygalacturonase inhibiting protein). This strategy was effective in reduction of B. cinerea infection of grapevine leaves upon treatment with these acetylated oligoglucuronans (Caillot et al., 2012). In the case of natural laminarin, chemically sulfated analogs with DP >5 were seen to be effective in stimulating the SA signaling pathway leading to protection against pathogens in tobacco and Arabidopsis plants (Menard et al., 2004). A heptaβ-glucan isolated from culture medium during germination of Phytophthora megasperma f. sp. glycinea was found to elicit the synthesis of isoflavonoid phytoalexins in soybean cotyledons (Sharp et al., 1984a,b). This interaction was seen to stimulate the induction of localized HR mediated resistance against P. megasperma f. sp. glycinea in several other legumes namely soybean, alfalfa (Medicago sativa), bean (Vicia faba), lupin

(Lupinus albus), pea, Medicago truncatula, and Lotus japonicus (Cosio et al., 1996; Cote et al., 2000).

The recognition of fungal β-glucans by animal cells is known to induce the production of inflammatory chemokines/cytokines such as TNFα, IL-1b, IL-10, IL-6, IL-23, CCL2, CCL3, etc., when perceived by the homo-dimerization and phosphorylation of Dectin-1 (Brown and Gordon, 2003; Drummond and Brown, 2011). Cell signaling in case of Dectin-1 can proceed through both SYK-dependent (Spleen Tyrosine Kinase) and independent pathways (RAF1-dependent), bringing about a horde of effects ranging from internalization of pathogen for antigen-presentation to cytokine/caspase activation. Dectin-1 consists of an extracellular C-type lectin domain that is connected to the plasma membrane by a stalk region which protrudes in to the extracellular space. The cytoplasmic C-terminus contains an immunoreceptor tyrosine-based activation motif (ITAM). Dectin-1 is primarily localized on the surface of macrophages and to a lesser extent on dendritic cells that binds to β-1,3-glucans and mixed β-1,3/1,6-glucans like laminarin, zymosan, and complete yeast cells. The intracellular ITAM motif is phosphorylated by SRC, a tyrosine-protein kinase, upon recognition of β-1,3 glucans which activates the downstream Syk signaling cascade leading to immunogenic responses (Levitz, 2010). However, the phosphorylation of ITAM is preceded by the dimerization of two Dectin-1 receptors, following β-1,3-glucan binding (Brown et al., 2007). The interaction between β-glucan and Dectin-1 is mediated through the Trp221 and His223 amino acids on the C-type lectin domain of Dectin-1 (Adachi et al., 2004) which shares a 30% homology with A. thaliana and soybean, C-type lectin (Greeff et al., 2012; Singh and Zimmerli, 2013). However, the structure of the β-glucan binding groove in the corresponding plant C-type lectin domain is not known and it is speculated that the binding of a more complex β-glucan could be accommodated by a different set of amino acids in the plant Dectin-1 orthologs (Brown et al., 2007). Dectin-2 and Dectin-3 hetero-dimers were reported to recognize α-mannans as PAMPS, during the Candida albicans infection of mice. The recognition of α-mannans by Dectins 2 and 3 is mediated by the ITAM-containing cytoplasmic adaptor protein FcRγ, as they themselves are devoid of the ITAM motifs on the cytoplasmic end. The activation and phosphorylation of the Dectin 2 and 3 was reported to induce the production of NF-κB and pro-inflammatory cytokines such as TNF-α, IL10, IL12, and IL-6 (Zhu et al., 2013). However, no plant analogs for Dectin 2/3 have been reported so far.

Although various findings have shown a prominent role of β-glucan in priming plant defense, the putative sequence and domain structure for the plant β-glucan receptor is not known. One of the reasons could be that such studies have mostly been restricted to the Col-0 ecotype of A. thaliana, which is not known to elicit a strong immune response against β-glucan treatment (Fesel and Zuccaro, 2016). Furthermore, the work done on β-glucan mediated immunity during fungus– animal interactions appears to provide leads toward the better understanding of β-glucan perception in plants, wherein plant Dectin-1 homologs are believed to harbor a mammalian SYK-like domain, instead of the ITAM domain (Fesel and Zuccaro, 2016). Based on the information available, a schematic depicting the β-glucan and α-mannan mediated immunity in animals and the hypothetical model of its plant counterpart is depicted in **Figure 6**. A genome wide survey of the available Arabidopsis ecotype resources or investigation of the β-glucan and α-mannan perception in other model plants could possibly unearth the missing links to complete the picture.

# Cellodextrins

Cellodextrins are a group of oligosaccharides that consist of a linear β-1,4- linked glucose backbone, which are the endproducts of cellulose degradation in plants. It has been reported that cellodextrin treatment of plant cells leads to the activation of PAL genes, which enhances synthesis of lignin, SA, and phytoalexin as an innate immune response. They in turn activate acidic (Chit3, Chit4c) and basic (Chit1b) chitinases and β-1,3 glucanase which are responsible for hydrolysis of chitin and β-1,3-glucans in fungal cell walls (Farmer et al., 1991; De Lorenzo and Ferrari, 2002). Exogenous application of cellodextrin as an elicitor was also seen to activate protection against B. cinerea by triggering induction of oxidative burst, defense-related genes, elevation of cytosolic Ca2+, and stimulation of chitinase and β -1,3-glucanase activity (Aziz et al., 2007). In grapevine, the generation of H2O<sup>2</sup> and subsequent oxidative burst was highest in response to heptasaccharide cellodextrins (Aziz et al., 2007). Also, in Arabidopsis, treatment with cellobiose induced the innate immune responses such as increased cytosolic calcium and activation of MAPK pathway in response to Pseudomonas syringae infection (De Azevedo Souza et al., 2017). However, the fundamentals of cellodextrin signaling still remain unclear and the correlation between the degree of polymerization and intensity of defense response needs further investigation.

# Chitin and Chitin Derivatives

Chitin, a polymer of N-acetylglucosamine, is a major constituent of fungal cell walls and is the most well characterized filamentous PAMP (Shibuya and Minami, 2001). It has been seen to stimulate phytoalexins and trigger the MAPK based defense signaling in response to Magnaporthe oryzae challenge of rice, thereby providing a systemic resistance (Yamada et al., 1993; Yamaguchi et al., 2013). The first PRR for chitin, Chitin Elicitor Binding Protein (OsCEBiP), was identified as a plasma membrane localized receptor in rice cells, which was responsible for defense signal transduction (Miya et al., 2007; Kishimoto et al., 2010). In barley, the basal resistance against M. oryzae was seen to develop upon the recognition of chitin by the OsCEBiP homolog, HvCEBiP (Tanaka et al., 2010). The chitin PRR in Arabidopsis, AtCERK1 (also known as RLK1/LYK1) contains additional extracellular carbohydrate binding lysine motifs (LysMs), as well as an intracellular kinase domain essential for chitin based defense signaling (Egusa et al., 2015). Recent discoveries on Arabidopsis PRR have shown that AtCERK receptor also mediates recognition of 1,3-β-D-(Glc)6, a hexasaccharide from the necrotrophic fungus Plectosphaerella cucumerina which induces innate defense responses such as increased levels of cytosolic Ca2+, activation of MAPKs cascades and elevated expression of PR proteins (Melida et al., 2018). Chitin oligosaccharides when used as elicitors were reported

the filamentous pathogens of plants. The dotted lines represent the signaling pathways involved.

to induce the expression of pathogenesis-related protein, PR-10 and generation of ROS in both suspension-cultured rice cells and Arabidopsis seedlings (Akimoto-Tomiyama et al., 2003). In a recent study, the interaction between Phytophthora palmivora and L. japonicus led to the rapid transcript accumulation of LYS12, a LysM receptor protein of the NFR5-type, which could control the progression of disease by regulating defenses genes like POX, Germin-like protein, and Chitinase. Further, LYS12 mutants responded to treatments with glycan elicitors such as chitin oligomers and β-1,3- and β-1,6-glucan. It is speculated that there could be other functional components in the perception and signaling of these elicitors as LYS12 lacks the ATP binding loop and the typical Asp-Phe-Gly (DFG) motif in subdomain VII, essential for signaling. Also, several LysM receptors have been found to recognize and distinguish pathogenic/mutualistic carbohydrate signatures emerging from GlcNAc derived and microbial exopolysaccharide sources. It is therefore plausible that LYS12 is involved in monitoring carbohydrate PAMPs produced by P. palmivora as well as the damage associated signatures emerging from the host plant, thereby regulating disease progression (Fuechtbauer et al., 2018).

Chitosan, β-1,4-linked glucosamine is a deacetylated derivative of the chitin that forms a major component of many fungal cell walls, including those of fungal pathogens. Chitosan or its fragments have been shown to induce defense responses in dicotyledonous plant species like soybean and parsley mediated by the synthesis of callose, a β-1,3-glucan polymer (Kohle et al., 1985; Conrath et al., 1989). In melon plants, chitosan oligomers were shown to stimulate chitinase activity, while in wheat, treatment with chitosan oligosaccharides induced lignin deposition and increased phenolic acids level, in leaves (Burkhanova et al., 2007).

Chitin oligosaccharide derivatives bearing lipid modifications are known as lipochitooligosaccharides and act as microbial

signals to initiate symbiotic association of arbuscular mycorrhizal symbiosis (Myc factors) and root-nodule symbiosis (Nod factors) (Zipfel and Oldroyd, 2017). On the other hand, chitin oligosaccharide tetramers and heptamers have been reported to induce oscillations of Ca2<sup>+</sup> levels in the cell nucleus during symbiotic association in various legumes pea, M. truncatula and L. japonicus and rice (Sun et al., 2015). Thus chitin and its oligosaccharide derivatives acting either as defense elicitors or symbiosis activating signals may begin to answer how plants distinguish between friendly and pathogenic interactions in the apoplastic space. However, the role of the glycans in the interplay between the defense and symbiotic signaling networks and how plant signaling processes discern and deliver specific outcomes, remains to be unraveled.

# Alginate and Fucans

Alginate oligomers are formed by depolymerization of alginates obtained from sea weed that are made up of alternating blocks of L-glucuronic and D-mannuronic residues (Gonzalez et al., 2013). Alginate oligomers are gaining increasing attention as new elicitor materials for inducing plant defense machinery by stimulating the accumulation of phytoalexin and inducing PAL activity in soybean cotyledons (An et al., 2009). Furthermore, alginate oligosaccharides rich in polymannuronate are reported to activate defense responses in O. sativa against M. oryzae by inducing increased level of PAL, phytoalexin production, and POX activity (Zhang S. et al., 2015).

Sulfated fucan oligosaccharides made up of mono- and disulfated fucose units alternatively bound by α-1, 4- and α-1,3 glycosidic linkages have been shown to be strong inducers of PAL, in tobacco cell suspension cultures (Kombrink and Somssich, 1995; Klarzynski et al., 2003). Inoculation of tobacco leaves with these fucan oligosaccharides resulted in the accumulation of several PR proteins and the development of jasmonate mediated systemic resistance toward TMV (Klarzynski et al., 2003).

# GLYCOPROTEIN AND LECTINS

Alongside secreted protein effectors that destroy plant cell integrity filamentous pathogens are known to secrete small proteinaceous molecules that induce necrosis, shrinkage of cytoplasm, silencing of defense genes, electrolyte leakage, and generation of ROS in their hosts. These secreted molecules not only contribute to invasion and absorption of nutrients, but also assist in establishing necrotrophic/hemibiotrophic lifestyles by inducing HR-mediated cell death (Gonzalez et al., 2017). The first report of glycoproteins being involved as an elicitor that induced the defense related phytoalexin production was identified in P. megasperma and parsley interactions (Nurnberger et al., 1994). Much later the glycoprotein, BcGs1 produced from the culture filtrate of the necrotrophic fungus B. cinerea was seen to act as a necrosis-inducing elicitor that activated the typical HR as well as components of the SA, JA, and ET defense pathways, in tomato. BcGs1 treatment of tomato plants was found to induce the SA- dependent defense marker PR-1a, Prosystemin, a known elicitor of JA defense signaling and the tomato protein kinase 1 (TPK1b), which is an ET-mediated shared defense signal for protection against necrotrophic fungi and herbivorous insects. (Zhang Y. et al., 2015). However, the interaction between the SA, JA, and ET defense pathways, for rendering defense remains to be investigated. Earlier studies on the B. cinerea protein secretome have also shown the presence of an abundantly secreted glycoprotein, BcIEB1, with unknown function (Espino et al., 2010). Work on the structural analysis of the BcIEB1 glycoprotein has shown two serine/threoninerich residues glycosylated with α-1,2-/ α-1,3-linked mannose (Gonzalez et al., 2012, 2014). Recently, the elicitor function of BcIEB1 glycoprotein was discovered by Frias et al. (2016), wherein BcIEB1 expressed in yeast when assayed as a purified elicitor on tobacco, tomato, onion, and Arabidopsis, was able to induce defense responses against B. cinerea. The elicitor activity of the BcIEB1 protein induced ROS burst, electrolyte leakage, seedling growth inhibition, cytoplasm shrinkage and cell autofluorescence (Frias et al., 2016). Further studies on the BcIEB1 elicitor by Gonzalez et al. (2017) led to the discovery of the tobacco plant osmotin, a stress response protein, a member of the PR5 family, as the PPR of BcIEB1 glycoprotein. In a separate study, osmotin also was found to accumulate in plants as a response to invasion of fungal pathogens, leading to activation of PCD in fungi, lysis of fungal membranes and promoting the hyperaccumulation of osmolyte proline, which quenches ROS. Hence, the roles and diversity of effectors and their corresponding plant receptors extend well beyond the conventional effectors of a proteinaceous nature to involve many diverse molecules and pathways.

Glycan-binding lectin proteins are reported to regulate many of the defense signaling pathways during host–microbe interactions and immune responses (Silipo et al., 2010). Lectins mediate the recognition of plant pathogens upon perception of characteristic epitopes or damage-associated patterns, using protein–protein interactions as well as proteinglycan interaction (Lannoo et al., 2014). Lectin domain containing receptor-like kinases (LecRLKs) involved in pathogen recognition including LecRK-I.9 from Arabidopsis and Pi-d2 from rice (O. sativa) are reported to act against the oomycete pathogen Phytophthora brassicae and the ascomycete M. oryzae, respectively (Weidenbach et al., 2016). In LecRK, the extracellular lectin domain of LecRK is composed of a conserved hydrophobic groove that helps in the recognition of glycans (Andre et al., 2005; Wang and Bouwmeester, 2017). LecRK-I.9 domain in Arabidopsis interacts with oligosaccharides via two tripeptide Arg-Gly-Asp (RGD) in the lectin domain (Gouget et al., 2006). Also, extracellular ATP released by plants upon pathogen invasion acts as a ligand perceived by LecRk-I.9 domain (Choi et al., 2014). However, insights into the mechanism regulating both of these processes and how pathogens adapt to LecRK-mediated defense remains to be seen.

# GLYCOLIPIDS

Lipopolysaccharide (LPS) is a vital structural component of the outer layer of Gram negative bacteria with three distinct

functional domains which include: the lipophilic lipid A (LA) the di-glucosamine moiety which carries 4–7 fatty acids, an oligosaccharide core region, and the O-antigen region consisting of a variable number of oligosaccharide repeats (Alexander and Rietschel, 2001). The oligosaccharides in LPS act as a signaling moiety during induced immunity while the LA domain is recognized, even at picomolar concentrations, as a PAMP through different extra- and intra-cellular LPS sensors (Whitfield and Trent, 2014). The LPS sensing receptor in plants is RLK LORE (Lipooligosaccharide-specific Reduced Elicitation), which belongs to the plant-specific class of bulb-type lectin S-domain-1 kinases (SD-RLKs). RLK LORE can sense Pseudomonas LPS as PAMP in Arabidopsis and other crucifers triggering typical PTI responses (Ranf et al., 2015). The significance of RLK LORE in LPS perception was demonstrated by restoration of LPS sensitivity in LPS-insensitive tobacco plants through the expression of RLK LORE.

LPS mediated defense signaling is well characterized in the mammalian system where it binds to the LPS receptor, TLR4/MD-2 (toll-like receptor 4/myeloid differentiation factor-2), to trigger activation of phagocytosis, production of proinflammatory cytokines, and interferons and antimicrobial peptides (**Figure 7**; Beutler et al., 2001; Tan and Kagan, 2014). The inflammatory cascade can be activated directly by the recognition of the bacterial LPS by the mammalian glycoprotein CD14, in a TLR independent fashion. Alternatively, CD14 mediates the transfer of the bacterial LA by the mammalian serum protein LPS-binding protein (LBP), in TLR dependent fashion. Taking

the serum protein LBP and transferred to CD14, which occurs as a soluble (sCD14) and membrane-linked (mCD14) version. CD14 can transfer LPS to the membrane-resident TLR4/MD-2 receptor complex. Depending on the cellular localization (at the plasma membrane or in endosomes upon CD14-dependent endocytosis), TLR4/MD-2/LPS complexes activate production of either interferons or cytokines through distinct signaling adapters (TIRAP/MyD88 or TRIF/TRAM). In some cell types CD14 can directly trigger immune responses. In plants, the bulb-type lectin S-domain-1 RLK LORE (Lipooligosaccharide-specific Reduced Elicitation) was identified as the first LPS receptor component in plants and mediates sensitive perception of LA domain of Pseudomonas. However, it is yet unknown whether LA directly binds directly to LORE or to an LPS-binding co-receptor to activate the receptor complex and downstream signaling [Adapted from Ranf, 2016 and reproduced with permission under Creative Commons Attribution (CC BY) license].

inference from mammalians LPS signaling, it is currently believed that LORE undergoes dimerization during LPS perception. However, it is still not clear if any other plant defense modulators facilitate the transfer of the LA or facilitate LPS recognition and signaling in a LORE-independent manner.

Galactolipids, which are primarily composed of monogalactosyldiacylglycerol (MGDG) and digalactosyldiacylglycerol (DGDG), are important plant glycerolipids, known for their role in mediating in systemic acquired resistance (SAR) based plant defenses. It was seen that Arabidopsis mgd1 mutant which is deficient in MGDG synthesis and the dgd1 mutant, deficient in DGDG synthesis could not mount a SAR mediated defense response, when challenged with a secondary infection of the virulent pathogen P. syringae pv. maculicola after they were primed with the avirulent pathogen P. syringae pv. Tomato (Chaturvedi et al., 2008; Gao et al., 2014). This provides a strong evidence regarding the role of MGDG and DGDG in establishing SAR during pathogen attack; however, the complete pathway remains to be elucidated.

Phosphoinositol sphingolipids and glucosylceramides (GlcCer) are another group of glycosylated lipids found in fungi that feature a characteristic C-9 methyl group present on the long chain fatty acid base (Warnecke and Heinz, 2003). These fungal GlcCer behave as elicitors that induce plant defense responses like phytoalexin production and PR protein synthesis, when sprayed on to rice (Umemura et al., 2000).

Natural rhamnolipids produced by Pseudomonas aeruginosa have been shown to induce defense responses in grapevines, tobacco, and Arabidopsis cells via oxidative burst, ROS production, and inducing systemic acquired resistance (Varnier et al., 2009; Sanchez et al., 2012). In this context, natural rhamnose based glycolipids were synthesized enzymatically to use as elicitors of defense response and were found to induce extracellular ROS production when tested on tobacco cells (Nasir et al., 2017).

# CONCLUSION

Our current understanding of the roles that carbohydrates play in PAMP based defense priming in plants remains fragmented. Although the potential role of glycans as elicitors of plant defense signaling cascades has been identified as an important aspect, there remains a huge information gap when compared to what we currently understand about proteinaceous receptors of these glycan elicitors. The likely reason for this is the complexity of carbohydrate structures and the fact that the techniques used to analyze glycans lag behind those for protein and gene analysis. Improvements in oligosaccharide extraction, purification and identification techniques are beginning to enable studies using

# REFERENCES

more specific glycan species rather than crude extracts containing polydisperse carbohydrate species. The understanding of the roles played by carbohydrates in plant–pathogen interactions falls behind knowledge in human–pathogen interactions, which has attracted attention due to the implications in medicine development and disease treatment. It is only recently that we have begun to realize the importance carbohydrates have in plant protection strategies and appreciate the potential that carbohydrates have as specific intervention technologies.

In their natural habitat, plants are exposed to a wide variety of symbiotic and pathogenic microflora at any given time, which paves the way for a novel and extensive area of research to investigate the controlled and diversified signaling cascades generated within plants in response to simultaneous attack from different microbes. Plant genomes encode a large number of RLKs, RLPs, and lectins; biochemical and structural studies on these receptors will lead to better knowledge of their workings and allow identification of the most suitable oligosaccharide candidates.

Coming to the concept of sustainable agriculture, there is potential to use oligosaccharides for crop protection, to induce pathogen resistance and prime plants for microbial attack. The use of biodegradable, environmentally friendly glycans and glycan mimics will help to replace the conventionally used chemicals with safer alternatives. This will enable more sustainable agriculture and help avoid major crop losses, allowing better food security, given mankind's growing demand for food quality and quantity.

# AUTHOR CONTRIBUTIONS

CC, EK, and MR conceived the idea and prepared the draft of the manuscript. RF supervised the development of topics for discussion and edited the draft manuscript.

# FUNDING

This work was supported by a Biotechnology and Biological Sciences Research Council (BBSRC, United Kingdom) Institute Strategic Program Grant (MFN)(BB/P012523/1) to the John Innes Centre. EK acknowledges DBT, Government of India for the Overseas Associateship fellowship which was availed at the John Innes Centre, at RF's laboratory (BT/20/NE/2011). CC would like to acknowledge DST, Government of India for her DST INSPIRE Junior Research Fellowship (IF150964). MR is supported by BBSRC-ERA-CAPS grant (BB/N010272/1) to RF. EK and CC wish to acknowledge DBT, Government of India, for the Twinning Research Grant (Grant No. BT/427/NE/TBP/2011).

Akimoto-Tomiyama, C., Sakata, K., Yazaki, J., Nakamura, K., Fujii, F., Shimbo, K., et al. (2003). Rice gene expression in response to N-acetylchitooligosaccharide elicitor: comprehensive analysis by DNA microarray with randomly selected ESTs. Plant Mol. Biol. 52, 537–551. doi: 10.1023/A:10248906 01888

Adachi, Y., Ishii, T., Ikeda, Y., Hoshino, A., Tamura, H., Aketagawa, J., et al. (2004). Characterization of β-glucan recognition site on C-type lectin, dectin 1. Infect. Immun. 72, 4159–4171. doi: 10.1128/IAI.72.7.4159



membranes triggers multiple defense responses. Cell 78, 449–460. doi: 10.1016/ 0092-8674(94)90423-5


to basal resistance of barley to Magnaporthe oryzae. BMC Plant Biol. 10:288. doi: 10.1186/1471-2229-10-288


disease resistance in host plants. Biochem. Biophys. Res. Commun. 457, 627–634. doi: 10.1016/j.bbrc.2015.01.038


Zipfel, C., and Oldroyd, G. E. (2017). Plant signalling in symbiosis and immunity. Nature 543, 328–336. doi: 10.1038/nature 22009

**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 Chaliha, Rugen, Field and Kalita. 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.

# Overexpression of SSBXoc, a Single-Stranded DNA-Binding Protein From Xanthomonas oryzae pv. oryzicola, Enhances Plant Growth and Disease and Salt Stress Tolerance in Transgenic Nicotiana benthamiana

Yanyan Cao<sup>1</sup> , Mingtao Yang<sup>2</sup> , Wenxiu Ma<sup>1</sup> , Yujing Sun<sup>1</sup> and Gongyou Chen<sup>1</sup> \*

#### Edited by:

Zhengqing Fu, University of South Carolina, United States

#### Reviewed by:

Chang-Sik Oh, Kyung Hee University, South Korea Hai-Lei Wei, Cornell University, United States

> \*Correspondence: Gongyou Chen gyouchen@sjtu.edu.cn

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 20 March 2018 Accepted: 13 June 2018 Published: 05 July 2018

#### Citation:

Cao Y, Yang M, Ma W, Sun Y and Chen G (2018) Overexpression of SSBXoc, a Single-Stranded DNA-Binding Protein From Xanthomonas oryzae pv. oryzicola, Enhances Plant Growth and Disease and Salt Stress Tolerance in Transgenic Nicotiana benthamiana. Front. Plant Sci. 9:953. doi: 10.3389/fpls.2018.00953 <sup>1</sup> School of Agriculture and Biology, State Key Laboratory of Microbial Metabolism, Shanghai Jiao Tong University, Shanghai, China, <sup>2</sup> State Key Laboratory of Crop Biology, College of Life Sciences, Shandong Agricultural University, Tai'an, China

We previously reported that SSBXoc, a highly conserved single-stranded DNA-binding protein from Xanthomonas spp., was secreted through the type III secretion system (T3SS) and functioned as a harpin-like protein to elicit the hypersensitive response (HR) in the non-host plant, tobacco. In this study, we cloned SsbXoc gene from X. oryzae pv. oryzicola (Xoc), the causal agent of bacterial leaf streak in rice, and transferred it into Nicotiana benthamiana via Agrobacterium-mediated transformation. The expression of SsbXoc in transgenic N. benthamiana enhanced growth of both seedling and adult plants. When inoculated with the harpin Hpa1 or the pathogen Pseudomonas syringae pv. tomato DC3000 (Pst DC3000), the accumulation of reactive oxygen species (ROS) was increased more in SsbXoc transgenic lines than that in wild-type (WT) plants. The expression of pathogenesis-related protein genes (PR1a and SGT1), HR marker genes (HIN1 and HSR203J) and the mitogen-activated protein kinase pathway gene, MPK3, was significantly higher in transgenic lines than in WT after inoculation with Pst DC3000. In addition, SsbXoc transgenic lines showed the enhanced resistance to the pathogenic bacteria P. s. tabaci and the improved tolerance to salt stress, accompanied by the elevated transcription levels of the defense- and stress-related genes. Taken together, these results indicate that overexpression of the SsbXoc gene in N. benthamiana significantly enhanced plant growth and increased tolerance to disease and salt stress via modulating the expression of the related genes, thus providing an alternative approach for development of plants with improved tolerance against biotic and abiotic stresses.

Keywords: transgenic N. benthamiana, SsbXoc, plant growth, hypersensitive response, pathogen resistance, stress tolerance

# INTRODUCTION

fpls-09-00953 July 3, 2018 Time: 18:26 # 2

Plants are exposed to diverse stress conditions throughout their life cycle, including biotic and abiotic stresses. To cope with biotic stress, plants employ innate immune systems to overcome the microbial invasion (Jones and Dangl, 2006; Thomma et al., 2011). The first line of defense is induced by pathogen-associated molecular patterns (PAMPs), which includes a diverse group of molecules such as flagellin (Felix et al., 1999), EF-Tu (Kunze et al., 2004), chitin and harpins (He et al., 1993; Zou et al., 2006). Harpins are glycine-rich, heat-stable and protease-sensitive proteins that are secreted through the type III secretion system (T3SS) (Wei et al., 1992). Previous researches have demonstrated that plants are highly sensitive to harpin elicitors. The harpins stimulate the hypersensitive cell death, the oxidative burst and the expression of defense-related genes (He et al., 1993; Andi et al., 2001; Ichinose et al., 2001), and activate the mitogen-activated protein kinase (MAPK)-dependent signaling pathway (Desikan et al., 1999, 2001; Lee et al., 2001), which finally induce the defence response in plants.

Previous studies have shown that treatment with harpins induces plant growth (e.g., stimulates the elongation of roots) and enhances resistance to aphids in Arabidopsis (Dong et al., 2004; Lü et al., 2011, 2013). Up to now, multiple harpins have been expressed in plants, including Arabidopsis, rice, wheat, tobacco, cotton, and soybean, and the transgenic plants exhibited enhanced plant growth and improved resistance to pathogens (Jang et al., 2006; Shao et al., 2008; Miao et al., 2010; Choi et al., 2012; Wang D. et al., 2014; Du et al., 2018). For example, the transformation of cotton with hpa1 enhanced the defense response against Verticillium dahliae (Miao and Wang, 2013; Zhang et al., 2016). Furthermore, the heterologous expression of a functional fragment of the harpin protein Hpa1Xoo induced phloem-based defense against the English grain aphid in wheat (Fu et al., 2014). In addition, the expression of harpins also improves tolerance to abiotic stress. Previous studies demonstrate that HrpN increased drought tolerance by activating abscisic acid (ABA) signaling in Arabidopsis, and the harpin-encoding gene, hrf1, increased tolerance to drought stress in rice (Dong et al., 2005; Zhang et al., 2011). Recent studies indicate that overexpression of the harpin-encoding gene, popW, enhances plant growth and resistance to R. solanacearum, and also increases drought tolerance in transgenic tobacco (Wang C. et al., 2014; Wang et al., 2016; Liu et al., 2016). Increasing evidence shows that the multiple effects of harpins can be attributed to cross-talk of distinct signaling pathways to regulate development and defense in plants (Chen et al., 2008).

SSBs are highly conserved single-stranded DNA-binding proteins that protect ssDNA from nucleolytic digestion (Fedorov et al., 2006). We recently demonstrated that the SSB protein from Xanthomonas oryzae pv. oryzicola (Xoc) was shown to function as a harpin in tobacco (e.g., elicited an HR). Furthermore, treatment with SSBXoc improved plant growth and resistance to the fungal pathogen Alternaria alternate in Nicotiana tabacum cv. Xanthi (Li et al., 2013). In this study, the gene encoding Ssb in X. oryzae pv. oryzicola was transformed into N. benthamiana. Our research displays that SsbXoc transgenic plants exhibit enhanced plant growth, improved pathogen resistance, and increased tolerance to salt stress. To our knowledge, up to now there are no prior reports showing that the overexpression of harpins can enhance salt tolerance.

# MATERIALS AND METHODS

# Generation of SsbXoc Transgenic N. benthamiana Plants

Full-length SsbXoc gene (552 bp) was amplified by PCR using the specific primers (**Table 1**). The amplified product was cloned into pMD18-T Simple Vector (TaKaRa, Dalian, China) and then subcloned into the binary vector pCAMBIA2300 at XbaI and BamHI sites, which were placed downstream of the constitutive cauliflower mosaic virus 35S promoter (CaMV35S) and upstream of the polyadenylation signal of the nopaline synthase terminator (NOS) (**Figure 1A**). The recombinant clone, pCAMBIA2300-SsbXoc, was then transferred into Agrobacterium tumefaciens EHA105 for transformation of N. benthamiana. The SsbXoc transgenic plants were determined by PCR amplification with the specific primers of SsbXoc till T<sup>2</sup> generation.

# Plants and Growth Conditions

Seeds of WT and SsbXoc transgenic lines OE-1 and OE-9 (T2 generations) were surface-sterilized with 75% ethanol and 10% sodium hypochlorite for 0.5 and 5 min, respectively. They were then separately transferred to Murashige and Skoog (MS) medium without or with 100 mg L−<sup>1</sup> kanamycin and cultivated in a light-controlled incubator at 25◦C. Fifteen days later, the seedlings were transplanted to pots and grown in a greenhouse with a 16-h light/8-h dark photoperiod with 50% relative humidity at 25◦C.

# Plant Growth Analysis

The root lengths of transgenic lines (T2 generations) and WT plants grown in MS medium were measured after 15 days. Three independent experiments were performed and at least 20 seedlings were analyzed in each experiment. The phenotypes of plants were determined after the seedlings were transplanted to pots and cultivated for 4 weeks.

# Bacterial Strains and Growth Conditions

Bacterial strains used in this study were Pst DC3000 and P. s. tabaci. Both of them were grown at 28◦C on King's medium B (KMB) with or without rifampicin, respectively. They were resuspended and diluted to the appropriate concentration with 10 mM MgCl<sup>2</sup> for subsequent research.

# Determination of ROS Levels

Fully developed leaves of 2-month-old WT and T2 SsbXoc transgenic plants were separately injected with 100 µl Hpa1 protein (10 µg ml−<sup>1</sup> )and Pst DC3000 (OD<sup>600</sup> = 0.01) using 1-mL needleless syringes. After 6 h, treated leaves were collected and

#### TABLE 1 | Primers designed and used for PCR.

fpls-09-00953 July 3, 2018 Time: 18:26 # 3


incubated in diaminobenzidine (DAB) for 8 h at 25◦C and then were immersed in boiling ethanol (95%) for 10 min to remove the dye (Thomas and Lemmer, 2005). After further incubation in 60% ethanol for 4 h, photographs were taken for visualization of reactive oxygen species (ROS). To quantify ROS accumulation, treated samples were collected separately at 0 and 6 hpi for detection of H2O<sup>2</sup> contents as described previously (Bernt and Bergmeyer, 1974; Cao et al., 2015).

# Bacterial Growth Analysis

The fully expanded leaves of 2-month-old WT and T2 SsbXoc transgenic lines were inoculated with P. s. tabaci (OD = 0.01), and

the phenotypes were photographed at 36 hpi. In order to quantify the bacterial growth, the plants were inoculated with 10<sup>5</sup> CFU/ml P. s. tabaci as described previously (Klement, 1963; Thilmony et al., 1995). Briefly, a P. s. tabaci strain was grown overnight in KMB, washed twice, and resuspended at the appropriate concentration in 10 mM MgCl2. And bacterial suspensions were then infiltrated into fully developed leaves using 1-mL needleless syringes. To determine bacterial growth in plants, 1 cm<sup>2</sup> leaf disks were excised from the inoculated tissue of each treatment at 0, 1, and 2 dpi. The bacterial populations in the leaves were determined by plating serial dilutions on KMB.

# RNA Isolation and Gene Expression Analysis

Total RNA was isolated from leaves of WT and SsbXoc transgenic plants (T1 and T2 generation) using TRIzol reagent (TaKaRa, Japan) as recommended by the manufacturer. RT-PCR with genespecific primer pairs was performed to evaluate the expression of SsbXoc in WT and transgenic plants. The expression of SsbXoc and genes related to the defense response, oxidative stress, and salt stress was measured using quantitative real-time PCR (qRT-PCR), and all of the primers used in these experiences were listed in **Table 1**. EF1α was used as an internal standard in these experiments.

# Southern Blot Analysis

Genomic DNA was extracted from WT and T1 SsbXoc transgenic lines using CTAB as described previously (Murray and Thompson, 1980). The recombinant plasmid pMD18-SsbXoc and genomic DNA were digested with BamHI and XbaI enzymes, and fragments were separated by electrophoresis in a 1.3% agarose gel at 80 V for 12 h. DNA was transferred to nylon membranes and hybridized with the SsbXoc PCR product, which was labeled with digoxigenin as recommended by the manufacturer (Dig-Labeling Kit, Roche). Conditions for hybridization and detection were followed as described by Aviv et al. (2011). The primers used for amplifying the SsbXoc probe were listed in **Table 1**.

# Salt Stress Tolerance Assays

To examine germination rates during salt stress, seeds of T2 SsbXoc transgenic lines and WT plants were surface-sterilized and sown on MS medium supplemented with 100 mM NaCl cultivated in a light-controlled incubator with a 14-h light/10-h dark photoperiod at 25◦C. Germination rates were assayed after 5 days. For analysis of chlorophyll content, leaf disks (1 cm diameter) were excised from fully expanded leaves and floated separately on solutions containing 0, 200, and 400 mM NaCl for 4 d in the incubator. Chlorophyll contents were measured as described by Porra (2002), Kanneganti and Gupta (2008). Leaves were sampled for the measurements of malondialdehyde (MDA) and proline using previously described methods (Bates et al., 1973; Cao et al., 2014) after treatment with salt for 4 days.

# Statistical Analysis

All experiments were repeated three times. Data were presented as the mean ± SD and analyzed using Excel and SPSS. Tukey's test (P < 0.05) was used to determine significant differences.

# RESULTS

# Generation of SsbXoc Transgenic N. benthamiana

To quickly determine whether SsbXoc gene was present in transformed N. benthamiana, potential transgenic plants (T<sup>0</sup> generation) were initially screened by PCR using the SsbXocspecific primers. Nine lines were obtained that existed a prominent 552-bp fragment in the genomic DNA, which was the predicted size of SsbXoc gene (**Figure 1B**). Two transgenic lines designated OE1 and OE9 were randomly selected for further characterization. Genomic DNA was extracted from OE1 and OE9 and analyzed by Southern blot hybridization. Both lines contained a 0.55-kb hybridizing fragment that corresponded with the predicted size of SsbXoc gene, and this signal was not detected in WT plants (**Figure 1C**). Thus, both PCR and Southern blot analyses indicated that SsbXoc gene had been incorporated into the genome of OE1 and OE9 transgenic plants. To determine whether SsbXoc was expressed in the transgenic lines, the accumulation of SsbXoc mRNA was evaluated by RT-PCR using EF1α as an internal standard. A 552-bp product was amplified from the transgenic lines OE1 and OE9, but not from WT (**Figure 1D**), indicating that SsbXoc gene was successfully expressed in transgenic lines. In addition, to quantify the expression level of SsbXoc gene in transgenic lines, the qRT-PCR experiment was performed using SsbXoc specific primers (**Table 1**). The result showed that the expression level of SsbXoc in OE9 line was higher than that in OE1 line (Supplementary Figure S1).

# Expression of SsbXoc in Transgenic N. benthamiana Enhances Plant Growth

To evaluate whether the growth of SsbXoc transgenic plants was enhanced, root lengths were measured after cultivation in MS medium for 15 days. The transgenic lines OE1 and OE9 exhibited increased root lengths as compared with the WT (**Figures 2A,B**), and the difference was significant (P < 0.05). Four weeks after transplantation to pots, the transgenic lines still exhibited enhanced plant growth (**Figure 2C**). Previously, Goh et al. (2012) reported that genes in the expansin family, e.g., AtEXPA1, AtEXPA5 and AtEXPA10, were required for leaf growth, furthermore, the suppression of AtEXPA decreased foliar growth in Arabidopsis. EIN2 is demonstrate as an essential positive regulator in the ethylene signaling pathway, which is involved in many aspects of the plant life cycle (Johnson and Ecker, 1998; Wang et al., 2002). Thus, we measured the expression levels of expansin-encoding gene, EXPA1, and EIN2, to investigate whether the transcription of the two genes was enhanced in SsbXoc transgenic plants. As shown in **Figure 2D**, the transgenic lines exhibited higher expression of EXPA1 and EIN2 in comparison with WT, which further confirmed the enhanced growth evident in transgenic plants (**Figure 2D**).

# SSBXoc Improves Defense Responses to Hpa1 and Pst DC3000 in Transgenic N. benthamiana

The Hpa1 protein and the pathogen of Pst DC3000 were individually inoculated to WT and SsbXoc transgenic plants to examine defense responses. DAB staining results indicated that ROS levels were significantly enhanced in SsbXoc transgenic lines as compared with WT (**Figure 3A**). H2O<sup>2</sup> contents were then evaluated to quantify ROS levels in treated leaves. As shown in **Figure 3B**, transgenic lines exhibited higher levels of H2O<sup>2</sup> accumulation than WT plants after inoculation with Hpa1 (**Figure 3B**, upper panel) and Pst DC3000 (**Figure 3B**, lower panel).

The accumulation of ROS in response to harpins and incompatible pathogens is generally accompanied by the HR (Zurbriggen et al., 2010). Therefore, WT and SsbXoc transgenic lines were evaluated visually for the HR at 24 hpi. The results showed that, after inoculated with Hpa1 and Pst DC3000 for 24 h, WT plants started to appear the HR, while transgenic lines reacted earlier and formed a more prominent HR at the inoculation site (**Figure 3C**), indicating that SsbXoc transgenic plants activated defense response earlier than WT, and this promoted the pathogen resistance.

# SSBXoc Enhances the Expression of Defense Related-Genes in Transgenic N. benthamiana

The expression of many defense genes can be activated during pathogen invasion in plants, including the pathogenesis-related (PR) genes, which play an important role in plant defense response (Maurhofer et al., 1994; Van Loon, 1997). To further investigate the mechanism underlying the increased pathogen resistance of SsbXoc transgenic plants, the expression levels of the PR genes, PR1a and SGT1, HR marker genes, HIN1 and HSR203J, and a gene involved in the MAPK-dependent signaling pathway, MPK3, were examined during infection by Pst DC3000. The results showed that at the time of inoculation with Pst DC3000 (0 hpi), the expression of defense-related genes was higher in transgenic lines as compared to WT; at 6 hpi, the expression levels of the five genes were all upregulated in all of the plants, while they were increased more in transgenic lines (**Figure 4** and Supplementary Figure S2), further indicating that SsbXoc transgenic lines could respond more quickly to the invasion of Pst DC3000.

# Overexpression of SsbXoc Improves Resistance to P. s. tabaci

In order to investigate whether SsbXoc transgenic plants could improve bacterial disease resistance, one pathogenic bacteria, P. s. tabaci, was used. As shown in **Figure 5A**, SsbXoc transgenic lines displayed less disease symptoms than WT plants at 36 h after inoculation with P. s. tabaci (**Figure 5A**). Correspondingly, the growth of P. s. tabaci was significantly lower in transgenic lines than that in WT plants at 1 and 2dpi, respectively (**Figure 5B**), being consistent with the necrosis symptoms in plants. In addition, the expression of defense genes was assayed in WT and SsbXoc transgenic plants after inoculation with P. s. tabaci. The results displayed that SsbXoc transgenic lines showed a higher expression of the pathogenesis-related genes, PR1a, PR2, PR4 and SGT1, than that of WT at the time of inoculation (0 hpi), and at 6 hpi, the expression levels of all the four genes were upregulated, however, they were more higher in transgenic lines than in WT (**Figure 6** and Supplementary Figure S2). All of the above results indicated that SsbXoc transgenic plants had an improvement in resistance to the pathogenic bacterium, P. s. tabaci.

FIGURE 3 | The oxidative burst assay in WT and SsbXoc transgenic N. benthamiana plants inoculated with Hpa1 and Pst DC3000. WT and transgenic plants were injected with Hpa1 protein (10 µg ml−<sup>1</sup> ) and Pst DC3000 pathogen (OD = 0.01), and at 6 hpi, treated leaves were collected. (A) Visualization of H2O<sup>2</sup> accumulation by DAB staining in leaves inoculated with Hpa1 and Pst DC3000. (B) Evaluation of H2O<sup>2</sup> levels in leaves. Upper panel shows H2O<sup>2</sup> levels in WT and transgenic lines inoculated with empty vector preparation (EVP; negative control) and Hpa1; Lower panel shows H2O<sup>2</sup> levels in WT and transgenic lines inoculated with 10 mM MgCl<sup>2</sup> (negative control) and Pst DC3000. (C) Phenotypes of WT and transgenic lines inoculated with Hpa1 and Pst DC3000 after 24 h. Inoculation sites are indicated with open circles. Error bars represent SD, and values with different letters are significant at P < 0.05.

CFU/ml). Error bars represent SD, and values with different letters are significantly different at P < 0.05.

# SSBXoc Enhances Seed Germination and Chlorophyll Retention During Salt Stress

The potential role of SSBXoc in improving salt stress tolerance was initially investigated by measuring the germination of the seeds after treatment with 100 mM NaCl. As shown in **Figure 7A**, the percentages of seed germination of the two SsbXoc transgenic lines were 54.5 and 71%, respectively, which were significantly higher than WT (38.8%). This result showed that the improved germination rate was most pronounced in OE9 transgenic line (**Figure 7A**).

Chlorophyll retention is used as a physiological indicator of salt tolerance in plants (Sui et al., 2017). In the present study, a chlorophyll retention assay was performed to evaluate the salt tolerance in WT and SsbXoc transgenic plants when they were exposed to 0, 200, and 400 mM NaCl. The results showed that when exposed to 200 mM NaCl, the chlorophyll contents of WT, OE1 and OE9 were 54.3, 65.9, and 67%, respectively, and they were further reduced to 23.3, 39.1, 49% during treatment with 400 mM NaCl, respectively (**Figures 7B,C**). Thus, chlorophyll retention was higher in SsbXoc transgenic lines than in WT, suggesting that overexpression of SsbXoc improved salt tolerance in transgenic N. benthamiana.

# SSBXoc Decreases MDA Level and Increases Proline Content During Salt Stress

Malondialdehyde level has been used as a biological marker for the end-point of lipid peroxidation (Yoshimura et al., 2004; Wang et al., 2017), thus, we measured the MDA levels in WT and SsbXoc transgenic plants under the salt stress. No differences were observed in MDA contents between WT and SsbXoc transgenic lines when exposed to 0 mM NaCl, however, MDA level was significantly higher in WT than in transgenic lines after treatment with 200 mM NaCl (**Figure 8A**), indicating that lipid peroxidation, and hence membrane damage, was lower in transgenic N. benthamiana.

The accumulation of proline in plant cells is indicative of enhanced salt stress tolerance (Vinocur and Altman, 2005; Miller et al., 2010; Wang et al., 2015). Therefore, we evaluated the proline contents of leaves in WT and transgenic lines when they were exposed to salt stress. As shown in **Figure 8B**, no obvious differences were observed in proline contents between WT and SsbXoc transgenic lines without NaCl treatment, however, in transgenic lines, proline contents significantly increased more than in WT after treatment with 200 mM NaCl (**Figure 8B**). Thus, the increased proline contents implies the improved salt tolerance in SsbXoc transgenic plants.

# SSBXoc Improves the Expression of Stress-Related Genes During Salt Stress

More and more results demonstrated that plants modulate the expression of many stress-related genes as an adaptation to environmental stress (Umezawa et al., 2006; Chinnusamy et al., 2007; Hirayama and Shinozaki, 2010; Bharti et al., 2016). To better understand the mechanistic basis of salt tolerance in SsbXoc transgenic lines, we measured the expression levels of three stress-related genes, APX, GPX and CAT1, which separately encode ascorbate peroxidase, glutathione peroxidase, and catalase. As shown in the **Figure 9**, SsbXoc transgenic plants displayed a higher basal expression level of the three genes as compared to WT without salt stress; under 200 mM NaCl treatment, the expression levels of these three genes were all significantly enhanced in WT and SsbXoc transgenic lines, while they were increased more in the latter (**Figure 9**). These results indicated that SsbXoc transgenic N. benthamiana plants improved salt tolerance through up-regulating the expression of stressrelated genes.

# DISCUSSION

# SSBXoc Improves Plant Growth in Transgenic N. benthamiana

We previously demonstrated that the exogenous application of SSBXoc enhanced growth of tobacco and Arabidopsis (Li et al., 2013). In this study, we cloned SsbXoc gene from X. oryzae pv. oryzicola and transferred it into N. benthamiana via Agrobacterium-mediated transformation. Two SsbXoc transgenic lines (OE1 and OE9) were characterized, and both of them

FIGURE 6 | Expression analysis of defense genes in WT and SsbXoc transgenic N. benthamiana plants inoculated with P. s. tabaci. Two-month-old seedlings were inoculated with P. s. tabaci (OD = 0.01). At 0 and 6 hpi, the leaves were sampled to extract the total RNA to synthesize cDNA, and the expression levels of PR1a, PR2, PR4, and SGT1 genes were analyzed by qRT-PCR. Error bars represent SD, and values with different letters are significantly different at P < 0.05.

showed improved root elongation and enhanced foliar growth as compared to WT plants (**Figures 2A–C**). Previous study reported that the expansin family genes were required for leaf growth (Goh et al., 2012) and EIN2 participated in the process of plant development and positively regulated the ethylene signaling pathway (Johnson and Ecker, 1998; Wang et al., 2002). Thus, we measured the expression levels of one expansin-encoding gene, EXPA1 and EIN2 to investigate whether or not they were changed in SsbXoc transgenic plants. As shown in **Figure 2D**, the transgenic lines exhibited higher expression of the two genes in comparison with WT, which further confirmed the growth phenotypes of SsbXoc transgenic plants (**Figure 2C**).

# SSBXoc Transgenic Plants Exhibit Potentiated Defense Responses

Many studies have demonstrated that the activation of MAPKdependent signaling cascades (Nakagami et al., 2005), ROS, and defense gene expression (Nürnberger, 1999; Gómez-Gómez and Boller, 2000) occurs in most plant-pathogen interactions, which leads to an improved defense resistance. During this process, the activities of defense enzymes are usually triggered initially in the plant-pathogen interactions (Ramamoorthy et al., 2002), and the speed of these defense responses is faster in incompatible interactions (Kombrink and Somssich, 1995). Corresponded to these conclusions, the expression of PR genes was increased more

FIGURE 8 | Analysis of physiological indicators of lipid peroxidation (MDA) and proline in WT and SsbXoc transgenic N. benthamiana plants under salt stress. (A) MDA levels and (B) Proline contents in WT and transgenic plants after treatment with 0 and 200 mM NaCl for 4 d. Error bars represent SD, and values with different letters are significant at P < 0.05.

in WT and SsbXoc transgenic N. benthamiana when the tested plants were inoculated with Pst DC3000 rather than P. s. tabaci, excepting for the expression of SGT1 in WT plants.

Reactive oxygen species, e.g., H2O<sup>2</sup> and O<sup>2</sup> <sup>−</sup>, are primarily produced at the site of attempted pathogen invasion in plant cells (Nanda et al., 2010; Jwa and Hwang, 2017) and are indicative of pathogen recognition and activation of plant defense responses (Lamb and Dixon, 1997; Torres, 2010). Up to now, more and more researches demonstrate that exogenous harpins, including Hpa1, induce ROS accumulation in tobacco and Arabidopsis cell cultures (Desikan et al., 1998; Andi et al., 2001; Samuel et al., 2005; Zou et al., 2006; Li et al., 2013; Choi et al., 2013). In the current study, the ROS level was higher in SsbXoc transgenic plants after inoculation with Hpa1 protein and the incompatible pathogen, Pst DC3000 (**Figure 3B**), which finally led to an earlier HR (**Figure 3C**). In addition, the expression of PR genes, HR marker genes, and MPK3 gene was also higher in transgenic lines than that in WT after inoculation with Pst DC3000 for 6 h (**Figure 4**). In a word, the higher levels of ROS and the improved expression of defense-related genes in SsbXoc transgenic plants were consistent with the rapid elicitation of the HR. Previous studies have shown that the HR generally appears within 24 h after inoculation with an incompatible pathogen or harpin (Wei et al., 1992; He et al., 1993). In this study, we inoculated N. benthamiana plants with reduced levels of Hpa1 (10 µg ml−<sup>1</sup> ) and Pst DC3000 (OD<sup>600</sup> = 0.01). Using this approach, we discovered that SsbXoc transgenic plants were more sensitive to the two eliciting agents accompanied with the increased expression of SsbXoc gene in transgenic plants, finally leading to producing a stronger HR at 24 hpi than WT plants (**Figure 3C** and Supplementary Figure S1).

Previously, Nicotiana tabacum cv. Xanthi plants infiltrated with SSBXoc displayed an improved resistance to the tobacco pathogen, Alternaria alternata (Li et al., 2013). In the current study, another pathogenic bacterium, P. s. tabaci, was used to inoculate WT and SsbXoc transgenic N. benthamiana plants. The results showed that SsbXoc transgenic lines had the higher basal transcription levels of PR1a, PR2, PR4, and SGT1 as compared to WT plants. After inoculation with P. s. tabaci for 6 h, expression of PR genes was significantly increased more in SsbXoc transgenic lines, and this was accompanied by a slight reduction in pathogen growth than WT plants (**Figures 5**, **6**), suggesting the enhanced bacterial resistance in SsbXoc transgenic N. benthamiana.

# SsbXoc Transgenic Plants Show Improved Salt Tolerance

Salt stress has many deleterious effects on plant growth and development, and inhibits seed germination, chlorophyll retention, root length, and fructification (Zhang et al., 2006; Sui et al., 2017; Liang et al., 2018). We initially used percent seed germination and chlorophyll retention to evaluate salt tolerance

fpls-09-00953 July 3, 2018 Time: 18:26 # 9

and discovered that SsbXoc transgenic N. benthamiana plants displayed higher levels of germination rates and chlorophyll contents when exposed to different concentrations of NaCl (**Figure 7**), indicating the enhanced salt tolerance of transgenic plants.

We next used MDA and proline as bioindicators to investigate the salt stress tolerance of SsbXoc transgenic N. benthamiana in the present study. MDA is the main product of membrane lipid peroxidation when plants are under salt stress (Liang et al., 2018), and MDA content has been used as a biological marker for the degree of membrane damage (Yoshimura et al., 2004; Wang et al., 2017). In our current study, we noted lower MDA levels in SsbXoc transgenic lines than in WT under salt stress condition (**Figure 8A**), suggesting that the degree of lipid peroxidation was lower in transgenic lines. Proline is an important osmotic adjustment compound in plant cells and plays a crucial role in protecting macromolecules and cellular membranes (Singh et al., 2000; Miller et al., 2010; Liang et al., 2018). The elevated accumulation of proline in plant cells is indicative of enhanced salt stress tolerance (Vinocur and Altman, 2005; Miller et al., 2010). In our research, we also observed a significant increase of proline contents in transgenic lines as compared to WT plants (**Figure 8B**), implying the enhanced salt tolerance in SsbXoc transgenic N. benthamiana.

During salt stress, the concentration of ROS increases to a potentially toxic level. To overcome H2O2-related cellular damage, organisms produce various antioxidant enzymes, including ascorbate peroxidase (APX), glutathione peroxidase (GPX), and catalase (CAT) (Ozyigit et al., 2016). The improved expression of APX, GPX, and CAT was correlated with the increased salt tolerance in both WT and transgenic plants (Mishra and Tanna, 2017). In the current study, the expression of APX, GPX and CAT1 was higher in SsbXoc transgenic lines than in WT both under normal and salt stress conditions, particularly in OE9 line, which had the higher expression level of SsbXoc gene (**Figure 9** and Supplementary Figure S1). Thus, in addition to the elevated proline levels, the activities of ROS-scavenging enzymes were also increased in transgenic lines, finally leading to the enhanced tolerance to salt stress in SsbXoc transgenic plants.

However, little is known about the mechanisms how harpins and SSB protein trigger many similar beneficial effects on plants, though both harpins (including Hpa1) and SSB protein have some common features as mentioned elsewhere in this report. We hypothesize that, SSBXoc, like Hpa1, is translocated through the T3SS into plant cells, and possibly also perceived in plant apoplast, where it is recognized by unknown receptor(s) that recruit other proteins to activate downstream signal transduction cascades for HR induction, leading to expression of Ethdependent genes for plant growth and SA- or JA-dependent genes

# REFERENCES

Andi, S., Taguchi, F., Toyoda, K., Shiraishi, T., and Ichinose, Y. (2001). Effect of methyl jasmonate on harpin-induced hypersensitive cell death, generation of hydrogen peroxide and expression of PAL mRNA in tobacco suspensioncultured BY-2 cells. Plant Cell Physiol. 42, 446–449. doi: 10.1093/pcp/ pce056

for plant defense. Nevertheless, the discovery of harpin or SSB receptors in plants is the key to understand this point.

# CONCLUSION

Our previous research displays that SSB from X. oryzae pv. oryzicola shares many features in common with the harpin Hpa1. Similar to Hpa1, SSBXoc is an acidic glycine-rich, heat-stable protein that lacks cysteine residues, which can also stimulate an HR in tobacco (Li et al., 2013). Thus, in many aspects, SSBXoc functions in a similar manner to harpins. The present studies have shown that SSB proteins in Escherichia coli are found to bind to ssDNA in a sequence-independent manner, and protect ssDNA from forming secondary structures and subsequent degradation by nucleases (Shereda et al., 2008; Bianco, 2017). Although SSBXoc clearly functions as a harpin, it may also have additional functions that are similar to SSB in E. coli. Thus, it is tempting to speculate that SSBXoc may impart increased resistance to ROS in transgenic plants via the protective roles, such as the increased repair ability of single-stranded breaks due to oxidative stress. In a word, regardless of the precision mechanisms in the current study, SSBXoc has the potentials in improving plant growth, imparting enhanced disease resistance and improving salt tolerance in N. benthamiana.

# AUTHOR CONTRIBUTIONS

YC and GC designed the experiments. YC performed most of the experiments and analyzed most of the data. MY detected the H2O<sup>2</sup> contents and analyzed part of the data. WM provided some experimental methods. YS constructed the plasmid of pCAMBRIA2300-SsbXoc. GC and YC wrote the manuscript.

# FUNDING

This study was supported by the National Major Project for Developing New GM Crops (2016ZX08001-002) and the National Natural Science Foundation of China (31471742).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00953/ full#supplementary-material


leaf development. Plant Physiol. 159, 1759–1770. doi: 10.1104/pp.112. 200881


fpls-09-00953 July 3, 2018 Time: 18:26 # 11


**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 Cao, Yang, Ma, 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.

fpls-09-00953 July 3, 2018 Time: 18:26 # 12

# Biogenesis and Function of Multivesicular Bodies in Plant Immunity

Xifeng Li<sup>1</sup>† , Hexigeduleng Bao<sup>2</sup>† , Zhe Wang<sup>3</sup> , Mengxue Wang<sup>2</sup> , Baofang Fan<sup>3</sup> , Cheng Zhu<sup>2</sup> \* and Zhixiang Chen1,2,3 \*

<sup>1</sup> Department of Horticulture, Zhejiang University, Hangzhou, China, <sup>2</sup> College of Life Sciences, China Jiliang University, Hangzhou, China, <sup>3</sup> Department of Botany and Plant Pathology, Center for Plant Biology, Purdue University, West Lafayette, IN, United States

#### Edited by:

Thomas Mitchell, The Ohio State University, United States

#### Reviewed by:

Roger W. Innes, Indiana University Bloomington, United States Xiaohong Zhuang, The Chinese University of Hong Kong, Hong Kong

#### \*Correspondence:

Cheng Zhu pzhch@cjlu.edu.cn; 10a0901067@cjlu.edu.cn Zhixiang Chen zhixiang@purdue.edu †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 02 March 2018 Accepted: 15 June 2018 Published: 09 July 2018

#### Citation:

Li X, Bao H, Wang Z, Wang M, Fan B, Zhu C and Chen Z (2018) Biogenesis and Function of Multivesicular Bodies in Plant Immunity. Front. Plant Sci. 9:979. doi: 10.3389/fpls.2018.00979 Multivesicular bodies (MVBs) are specialized endosomes that contain intraluminal vesicles generated from invagination and budding of the limiting membrane. In the endocytic pathway, MVBs are late endosomes whose content can be degraded through fusion with lysosomes/vacuoles or released into the extracellular space after fusion with the plasma membrane (PM). The proteins retained on the limiting membrane of MVBs are translocated to the membrane of lysosomes/vacuoles or delivered back to the PM. It has been long suspected that MVBs might fuse with the PM to form paramural bodies in plant cells, possibly leading to release of building blocks for deposition of papillae and antimicrobial molecules against invading pathogens. Over the past decade or so, major progress has been made in establishing the critical roles of MVBs and associated membrane trafficking in pathogen recognition, defense signaling, and deployment of defense-related molecules during plant immune responses. Regulatory proteins and signaling pathways associated with induced biogenesis and trafficking of MVBs during plant immune responses have also been identified and characterized. Recent successful isolation of plant extracellular vesicles and proteomic profiling of their content have provided additional support for the roles of MVBs in plant–pathogen interactions. In this review, we summarize the important progress and discuss how MVBs, particularly through routing of cellular components to different destinations, contribute to the complex network of plant immune system.

#### Keywords: MVBs, plant immunity, LIP5, endocytosis, endosomal trafficking

# INTRODUCTION

Plants are constantly exposed to a wide spectrum of microbial pathogens and have evolved both constitutive and inducible defense mechanisms. Constitutive defenses include physical barriers such as waxy epidermal cuticles and cell wall on the plant surface to prevent pathogen penetration (Underwood, 2012). Inducible defenses are activated upon recognition of potential pathogenic microbes by sophisticated surveillance mechanisms in plants (Jones and Dangl, 2006). First, plasma membrane (PM)-localized pattern recognition receptors (PRRs) recognize microbe-associated molecular patterns such as bacterial flagellin and initiate signaling to activate pattern-triggered immunity (PTI), a set of responses including burst of reactive oxygen species (ROS), activation of mitogen-activated protein kinases (MAPKs), and defense gene expression (Jones and Dangl, 2006).

**115**

Pathogens can secret effectors to plant cells to suppress PTI but some of the effectors may be recognized by plant resistance (R) proteins and activate effector-triggered immunity, a strong defense often manifested as hypersensitive response (HR) (Jones and Dangl, 2006).

Vesicle trafficking controls the structures of intracellular compartments and communication between cells and environment (Reyes et al., 2011; Paez Valencia et al., 2016). In the endocytic pathway, molecules are delivered from the PM in endocytic vesicles to early endosomes (EEs), which mature into late endosomes (LEs). Multivesicular bodies (MVBs) are LEs that contain intraluminal vesicles (ILVs), which are generated from the invaginations and budding of the limiting membrane into the lumen. Cargo-containing ILVs can be degraded upon fusion with lysosomes/vacuoles or released as exosomes upon fusion with the PM (Hanson and Cashikar, 2012; **Figure 1A**). In plants, MVBs also play important roles in trafficking proteins to the vacuoles in the secretory pathway and are also referred to as prevacuolar compartments (PVCs; Cui et al., 2016; **Figure 1A**). Studies over the past decades have also uncovered critical roles of MVBs in pathogen-triggered trafficking processes important for both the signaling and execution of plant defense. In this review, we discuss the progress in our understanding of the role, action, and regulation of MVB biogenesis and trafficking in plant immunity.

# PATHOGEN-INDUCED MVB BIOGENESIS

Multivesicular bodies mature from trans-Golgi network (TGN)/EEs (Cui et al., 2016; **Figure 1A**). The characteristic ILVs of MVBs originate from the vesicle budding from the limiting membrane into the lumen through the action of protein complexes named ESCRT-0, I, II, and III (endosomal sorting complex required for transport; Wollert and Hurley, 2010; **Figure 1B**). Plants have no canonical ESCRT-0 subunits but contain TOM1-like (TOL) proteins that bind ubiquitin and are required for the endocytosis and vacuolar sorting of the auxin efflux carrier component PIN2 (Korbei et al., 2013; Gao et al., 2017). ESCRT-1 components further cluster ubiquitinated proteins and also recruit ESCRT-II. The Vps25 subunit of ESCRT-II interacts with the Vps20/CHMP6 subunit of ESCRT-III to activates assembly of ESCRT-III on the endosomal membrane for cargo sorting, concentration, and vesicle formation (Wollert and Hurley, 2010). ESCRT-III assembles transiently on the endosome membrane and is then disassembled from the membrane into the cytoplasm after each cycle (Piper and Katzmann, 2007; **Figure 1B**). ESCRT-III disassembly is catalyzed by the Vps4p/SKD1 AAA ATPase, which is activated by Vta1/LIP5 (**Figure 1B**). Both Vps4p/SKD1 and Vta1/LIP5 are critical for MVB biogenesis in yeast and mammalian cells (Yeo et al., 2003; Shiflett et al., 2004; Ward et al., 2005; Azmi et al., 2006). In Arabidopsis, disruption of SKD1 is lethal, but lip5 mutants are normal in growth and development (Haas et al., 2007; Wang et al., 2014).

Multivesicular body biogenesis and associated endocytosis are pathogen-inducible in plants (An et al., 2006a,b; Wang et al., 2014). After flagellin binding, activated plant PRR FLS2 undergoes rapid endocytosis and accumulates in MVBs before degradation (Robatzek et al., 2006; Beck et al., 2012; Spallek et al., 2013). In the penetration resistance of cereal plants against the powdery mildew fungus, increased formation and relocalization of MVBs were associated with local cell wall appositions (papillae) surrounding the pathogen's penetration points (An et al., 2006a,b, 2007). Similar relocalization of defenserelated molecules such as the PEN3 ATP binding cassette (ABC) transporter possibly through MVBs for cell surface defense has also been found in Arabidopsis (Underwood and Somerville, 2013). Using both the FM1-43 endocytosis marker and the GFP-labeled ARA6 GTPase MVB marker, we have observed that both endocytosis and MVB biogenesis increase in response to Pseudomonas syringae in Arabidopsis (Wang et al., 2014). Importantly, pathogen-induced endocytosis and MVB biogenesis was not observed in the lip5 mutants (Wang et al., 2014). Thus, LIP5 is required for pathogen-induced MVB biogenesis and endocytosis. This role of LIP5 in pathogen-induced MVB biogenesis is dependent on its enhanced protein stability upon phosphorylation by pathogen-responsive MAPK3 and 6 in Arabidopsis (Wang et al., 2014; **Figure 2**).

# MVBs IN DEFENSE SIGNALING

In Arabidopsis, plant receptor FLS2 mediates immunity against bacterial infection through recognition of bacterial flagellin (Gomez-Gomez and Boller, 2000). Following flagellin binding, activated FLS2 undergoes endocytosis and accumulates in MVBs before being degraded in the vacuole (Robatzek et al., 2006; Choi et al., 2013). Several lines of evidence suggest that endocytosis of FLS2 functions as a molecular mechanism not only for its attenuation but also for proper immune responses (Salomon and Robatzek, 2006). First, mutant FLS2 proteins impaired in endocytosis are compromised in conferring flagellintriggered responses. FLS2T867V, which carries a substitution of the potentially phosphorylated residue threonine-867, had impaired endocytosis and could not complement the enhanced disease susceptibility in the fls2 mutants (Robatzek et al., 2006). Endocytosis of FLS2P1076A, which contains a mutation in a PEST motif, was abolished, and the mutant FLS2 protein failed to confer flagellin-induced ROS production in the fls2 mutants (Robatzek et al., 2006). Second, treatment with vesicular trafficking inhibitors such as wortmannin-reduced FLS2 endocytosis and impaired flagellin-induced production of ROS (Smith et al., 2014). Third, endocytosis and trafficking of FLS2 and other PRRs require clathrin and clathrin-dependent endocytosis is involved in FLS2-mediated stomatal closure and callose deposition (Mbengue et al., 2016). In addition, endocytosed FLS2 associates and co-localizes with the VPS37-1 and VPS28-2 components of ESCRT-I (Spallek et al., 2013). In Arabidopsis vps37-1 and vps28-2 mutants, FLS2 endocytosis was reduced and MVB sorting was compromised (Spallek et al., 2013). When surface-inoculated with virulent P. syringae, vps37-1, and vps28-2 also supported increased bacterial growth due to compromised stomatal closure, even though ROS production and

MAPK activation were normal in the mutants (Spallek et al., 2013). Finally, flagellin peptide flg22 could move to distal organs through vascular tissues (Jelenska et al., 2017). Delivery of flg22 to vascular tissues and its long-distance transport were dependent on its endocytosis together with the FLS2 receptor (Jelenska et al., 2017). Thus, FLS2 endocytosis is also required for long-distance movement of flg22, possibly for systemic immune responses.

Multivesicular bodies also play a role in the functioning of NB-LRR R proteins. Potato R protein R3a can mount an immune response upon recognition of the Avr3aK80I103 (Avr3aKI) effector from Phytophthora infestans, but not the Avr3aE80M103 (Avr3aEM) effector, which differs from Avr3aEM in only two amino acids (Engelhardt et al., 2012). R3a is normally in the cytoplasm but relocalized to MVBs when co-expressed with its cognate effector Avr3aKI, but not with Avr3aEM (Engelhardt et al., 2012). Co-expressed Avr3aKI, but not Avr3aEM, is also relocalized to MVBs in a R3a-dependent manner prior to R3a-mediated HR. Relocalization of R3a to MVBs is inhibited by vesicle trafficking inhibitors brefeldin and wortmannin, which also blocked R3a/Avr3aKI-mediated HR, indicating that relocalization of R3a is required for the immune response (Engelhardt et al., 2012). Thus, MVBs act as subcellular compartments from which signaling can be initiated from internalized PRRs and internal R proteins to activate immune responses. Studies in other organisms also indicate MVBs can function in signaling or even as signaling organelles. During Wnt signal transduction in animals, Wnt signaling can trigger sequestration of glycogen synthase kinase into MVBs, allowing the activation of many cytosolic proteins (Dobrowolski and De Robertis, 2011). In addition, MVBs containing nerve growth factor (NGF) and

its receptor TrkA in mouse sympathetic neurons can evade lysosomal fusion and instead evolve into membrane vesicles that are signaling components (Ye et al., 2018).

# MVBs IN CELL SURFACE DEFENSE

Trafficking defense-related molecules through MVBs plays a critical role for defense against invading pathogens at plant cell surface. In the interactions between plants and filamentous pathogens, the spores of the pathogens germinate on the leaf surface and develop infection pegs to invade the epidermal cells, which can induce a range of plant defense responses including the formation of local cell wall appositions (papilla) at the sites of pathogen attack (Collins et al., 2003; Assaad et al., 2004; Schulze-Lefert, 2004). Studies have showed accumulation of MVBs and cell wall-associated paramural bodies (PMBs) in the vicinity of pathogen-induced papillae (An et al., 2006a,b, 2007). PMBs, which are situated between the cell wall and the PM, likely result from the fusion of MVBs with the PM (Marchant and Robards, 1968). Plant MVBs and PMBs have been observed near papillae in plant cells infected by pathogenic fungal, bacteria, and nematodes for delivery of defense-related molecules including phytoalexins, callose, and ROS to papillae (An et al., 2006a,b; Meyer et al., 2009; Bohlenius et al., 2010; Nielsen et al., 2012). In Arabidopsis, two pathways have been identified in the penetration resistance to the non-adapted powdery mildew fungus Blumeria graminis f. sp hordei (Bgh) defined by the PM syntaxin, PEN1, and the β-glucoside hydrolase PEN2. The PEN1-mediated pathway delivers building materials for the papilla, while PEN2 functions with the PEN3 ABC transporter in mediating the synthesis and export of antimicrobial metabolites to the attack sites. Relocalization of defense-related molecules such as PEN1 and PEN3 for cell surface defense in response to conserved pathogen elicitors has also been observed in Arabidopsis (Underwood and Somerville, 2013). Given the diverse roles in recycling, degradation, and relocalization of defense-related molecules and different structural appearances of MVBs and associated PMBs, it would be of interest to determine whether there exist multiple pathways for MVB biogenesis and trafficking and whether ESCRT complexes are involved in the formation of PMBs.

Upon successful penetration, filamentous pathogens can develop a special feeding structure into plant cells called haustoria. Each haustorium is surrounded by the PM of the plant cell termed extrahaustorial membrane (EHM), which is likely synthesized de novo (Berkey et al., 2017). In tobacco (Nicotiana benthamiana) cells invaded by oomycete pathogen P. infestans, the Rab7 GTPase RabG3c MVB marker protein, but not a tonoplast-localized sucrose transporter, is recruited to the EHM (Bozkurt et al., 2015). In Arabidopsis, the Rab5 GTPase, also an MVB marker, accumulates in the EHM after infection with a powdery mildew fungus (Nielsen et al., 2017). Thus, specific rerouting of MVBs from the vacuole to the host–pathogen interface may participate in the formation or modulation of the EHM. Importantly, cell surface immune receptors such as FLS2 and RPW8 resistance proteins are also recruited to the EHM upon pathogen infection, probably as a host border control mechanism at the plant–pathogen interface (Eckardt, 2009; Kim et al., 2014; Bozkurt et al., 2015; Berkey et al., 2017).

Further evidence for a role of plant MVBs in plant immune responses has been provided from plant extracellular vesicles (EVs), which are likely generated from the fusion of MVBs with the PM. EVs from Arabidopsis rosettes contain PEN1, RIN4 (RPM1-INTERACTING PROTEIN4), RIN4-interacting proteins and proteins involved in metabolism and transport of antimicrobial compounds (Rutter and Innes, 2017). EV secretion is enhanced in Arabidopsis after P. syringae infection (Rutter and Innes, 2017). Comparison of EV proteins with other published proteomes indicates that they are most similar to those from the TGN/MVB compartments (Rutter and Innes, 2017), supporting that EVs are derived from the ILVs of MVBs. EVs isolated from the extracellular fluids of sunflower seedlings are also enriched in defense proteins and could be taken up by the phytopathogenic fungus Sclerotinia sclerotiorum to cause growth inhibition, morphological changes, and cell death (Regente et al., 2017). These studies provide strong evidence for the existence of plant exosomes and their antimicrobial roles in plant immune responses.

# GENETIC EVIDENCE FOR THE ROLE OF MVBs IN PLANT IMMUNITY

In spite of the extensive microscopic data, genetic analysis of the role of MVBs in plant immune system is relatively limited, in part, because of the lethal phenotype of mutants for key components of MVB biogenesis such as SKD1 (Haas et al., 2007; Spitzer et al., 2009). However, some components required for

MVB biogenesis and trafficking are encoded by gene families and disruption of individual family members can allow for genetic analysis of their specific roles in plant defense. For example, genetic analysis of the VPS37-1 and VPS28-2 components of the ESCRT-I complex has revealed their role in FLS2 endocytosis and stomata-mediated defense (Spallek et al., 2013). The barley ARF GTPase ARFA1b/1c has been localized to MVBs and shown to be important for callose-deposition in papillae and penetration resistance of barley (Bohlenius et al., 2010). Furthermore, an ARF-GTP exchange factor, GNOM, is involved in papilla formation, callose deposition, and penetration resistance (Nielsen et al., 2012). However, the MVB localization of the ARF1 factor was later disputed, and evidence was presented for its localization to the Golgi and TGN (Robinson et al., 2011). More recently, it has been reported that endosome-associated VPS9a from Arabidopsis, the conserved guanine-nucleotide exchange factor activating Rab5 GTPases required for MVB maturation and fusion with other membranes, plays a critical role in both preinvasive and, to a greater extent, post-invasive immunity against a non-adapted powdery mildew fungus, as well as defense against an adapted powdery mildew fungus (Nielsen et al., 2017).

To analyze LIP5-mediated MVB biogenesis in plant immune responses, we have characterized two independent lip5 mutants. Both lip5 mutants are normal in growth and development but are highly susceptible to P. syringae (Wang et al., 2014). The critical role of LIP5 in plant immunity is dependent on its interaction with SKD1, indicating that the role of LIP5 in plant immune system is mediated by its action in MVB biogenesis. Basal levels of endocytosis and MVB biogenesis under normal conditions were normal in the lip5 mutants, consistent with its normal growth and development. After pathogen infection, both the endocytosis and MVB biogenesis were induced in a LIP5-dependent manner (Wang et al., 2014). Pathogen infection resulted in increased numbers of MVBs and PMBs in wild-type plants but not in the lip5 mutants (Wang et al., 2014). Therefore, pathogen-responsive MVB biogenesis and associated trafficking are activated by pathogen-responsive MAPK3/6 through LIP5 phosphorylation and plays an important positive role in plant immune system, most likely through trafficking and deployment of defense-related molecules at the cell surface to counter invading pathogens (**Figure 2**).

Intriguingly, a recent study on a rice AAA ATPase, LRD6-6, has indicated a negative role of MVB-mediated vesicular trafficking in plant immunity (Zhu et al., 2016). Rice LRD6-6 was identified from a rice lesion resembling disease (lrd) mutant resulting from insertion of a 534-nt DNA fragment in the proteincoding sequence of LRD6-6, which would disrupt its function (Zhu et al., 2016). The lrd6-6 mutant displays spontaneous lesions, enhanced disease resistance, and increased accumulation of antimicrobial compounds (Zhu et al., 2016). LRD6-6 is structurally related to SKD1, interacted with components of ESCRT-III, and associated with MVBs (Zhu et al., 2016). In addition, the lrd6-6 mutant was altered in expression of genes associated with MVB-mediated trafficking and defective in the trafficking of the Arabidopsis soluble vacuolar carboxypeptidase Y (AtCPY) from the ER to the vacuole, which is mediated by the secretory pathway through MVBs/PVCs (Zhu et al., 2016). However, sequence analysis reveals that Arabidopsis SKD1 protein is mostly closely related to three other proteins encoded by LOC\_Os01g04814, LOC\_Os02g06490, LOC\_Os01g04840 in rice (unpublished data). On the other hand, rice LRD6-6 is most similar to the spastin protein, an AAA ATPase widely found in eukaryotes with a microtubule-severing activity associated with membrane trafficking, assembly of nuclear protein complexes, cytokinesis, and secretion (Connell et al., 2009; Lumb et al., 2012). Furthermore, the genes with altered expression in lrd6-6 are mostly associated with early stages of the secretory pathway encoding membrane coat, clathrin coat of coated pit, ER membrane, and clathrin coat of TGN vesicle (Zhu et al., 2016), which could explain its defect in the transport of the Arabidopsis AtCPY from the ER to the vacuole (Zhu et al., 2016). Plant MVBs, also known as PVCs, play important roles in mediating protein traffic in both the endocytic and secretory pathways (Cui et al., 2016). Disruption of the secretory pathway could lead to impaired traffic of vacuolar cargo, defects in vacuolar biogenesis, or even cell death as observed in the ldr6-6 mutant.

# CONCLUSION AND FUTURE PROSPECTS

Important progress has been made in identifying components important for plant immunity by participating in pathogenresponsive MVB biogenesis and associated trafficking for relocalization of cellular components important for effector recognition, modulation of PRR activity, delivery of defenserelated proteins, and antimicrobial metabolites, building of encasement surrounding invading pathogens and regulation of cell growth and death. Over the next few years, it is expected that additional components in pathogen-responsive MVB and associated trafficking will be identified and their roles analyzed. It has become increasingly apparent that the timing and destinations of endosomal trafficking through MVBs are highly cargo-specific, and it is important to elucidate the underlying mechanisms. Further analysis will also be necessary to comprehensively establish the specific activities of the exosome content including possible regulatory activities in communication among plant cells or between plant and pathogen cells (Rutter and Innes, 2018). These studies can lead to a better understanding of plant endosomal trafficking and plant immune responses.

# AUTHOR CONTRIBUTIONS

CZ and ZC conceived the idea. XL, HB, and ZC wrote the manuscript. XL, HB, ZW, MW, BF, CZ, and ZC evaluated the manuscript. All authors read and approved the manuscript.

# FUNDING

This work was supported by the U.S. National Science Foundation (Grant Nos. IOS-0958066 and IOS1456300).

# REFERENCES

fpls-09-00979 July 5, 2018 Time: 21:45 # 6


Saccharomyces cerevisiae. J. Biol. Chem. 279, 10982–10990. doi: 10.1074/jbc. M312669200


**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, Bao, Wang, Wang, Fan, Zhu 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.

fpls-09-00979 July 5, 2018 Time: 21:45 # 7

# Red Light-Induced Systemic Resistance Against Root-Knot Nematode Is Mediated by a Coordinated Regulation of Salicylic Acid, Jasmonic Acid and Redox Signaling in Watermelon

You-xin Yang<sup>1</sup> , Chaoqun Wu<sup>1</sup> , Golam J. Ahammed<sup>2</sup> , Caijun Wu<sup>1</sup> , Zemao Yang<sup>3</sup> , Chunpeng Wan<sup>1</sup> and Jinyin Chen1,4 \*

 Jiangxi Key Laboratory of Crop Physiology, Ecology and Genetic Breeding, College of Agronomy, Jiangxi Agricultural University, Nanchang, China, <sup>2</sup> College of Forestry, Henan University of Science and Technology, Luoyang, China, Germplasm Lab, Institute of Bast Fiber Crops, Chinese Academy of Agricultural Sciences, Changsha, China, Pingxiang University, Pingxiang, China

### Edited by:

Zhengqing Fu, University of South Carolina, United States

### Reviewed by:

Richard Bostock, University of California, Davis, United States Yule Liu, Tsinghua University, China

> \*Correspondence: Jinyin Chen jinyinchen@126.com

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 26 February 2018 Accepted: 07 June 2018 Published: 10 July 2018

#### Citation:

Yang Y-X, Wu C, Ahammed GJ, Wu C, Yang Z, Wan C and Chen J (2018) Red Light-Induced Systemic Resistance Against Root-Knot Nematode Is Mediated by a Coordinated Regulation of Salicylic Acid, Jasmonic Acid and Redox Signaling in Watermelon. Front. Plant Sci. 9:899. doi: 10.3389/fpls.2018.00899 Red light (RL) can stimulate plant defense against foliar diseases; however, its role in activation of systemic defense against root diseases remains unclear. Here, the effect of RL on root knot nematode Meloidogyne incognita (RKN) infestation was investigated in watermelon plants (Citrullus lanatus L.). Plants were exposed to 200 µmol m−<sup>2</sup> s −1 photosynthetic photon flux density RL at the canopy level for 21 days using light-emitting photodiodes. The results showed that RL significantly suppressed gall formation and nematode development, which was closely associated with the RL-induced attenuation of oxidative stress in roots. Gene expression analysis showed that RL caused a transient upregulation of PR1 and WRKY70 transcripts at 7 days post inoculation in RKN-infected plants. Further investigation revealed that RL-induced systemic defense against RKN was attributed to increased jasmonic acid (JA) and salicylic acid (SA) content, and transcript levels of their biosynthetic genes in roots. Interestingly, while malondialdehyde content decreased, H2O<sup>2</sup> accumulation increased in RL-treated RKNplants, indicating a potential signaling role of H2O<sup>2</sup> in mediating RL-induced systemic defense. Furthermore, analysis of enzymatic and non-enzymatic antidoxidants revealed that RL-induced enhanced defense agaist RKN was also attributed to increased activities of antioxidant enzymes as well as redox homeostasis. Taken together, these findings suggest that RL could enhance systemic resistance against RKN, which is mediated by a coordinated regulation of JA- and SA-dependent signaling, antioxidants, and redox homeostasis in watermelon plants.

Keywords: Citrullus lanatus, hormones, red light, root knot nematodes, systemic resistance

# INTRODUCTION

The root-knot nematode Meloidogyne incognita is a soil-dwelling, microscopic nematode that feeds exclusively on the cytoplasm of living plant cells. Disease symptoms on infected plants include the presence of galls on roots, which may increase susceptibility to other pathogenic diseases such as Fusarium wilt (Kyndt et al., 2013). M. incognita invades the most economically important fruits

and vegetable crops, including watermelon, tomato and cucumber and causes substantial losses around the world (Jones et al., 2013). Until now, a great deal of soil fumigants and nematicides are being used to control nematodes, which raise issues regarding food safety, environmental pollution and human health, and also threaten sustainable agricultural development. Therefore, it is indispensable to develop environmental friendly strategies of nematode control to assure food safety and sustainable crop production.

Plants have evolved various defense strategies employing intricate mechanisms that induce immune responses through complex signaling networks and molecules, including reactive oxygen species (ROS), phytohormones and defense related genes (Martìnez-Medina et al., 2017; Song et al., 2017). For instance, in response to nematode invasion, ROS were quickly accumulated in the invaded cells and cell walls, especially in the cells that are close to hypersensitive response (Melillo et al., 2006). In Arabidopsis thaliana, the NADPH oxidases, RbohD and RbohF, produce ROS when infected by parasitic nematodes, and thus restricting infected plant cell death and promoting nurse cell formation (Siddique et al., 2014). Plant hormones play essential role in mediating plant defense. In particular, salicylic acid (SA), jasmonic acid (JA), ethylene (ET), brassinosteroids (BRs), and abscisic acid (ABA) are key components in plant defense against nematode infection (Nahar et al., 2011; Kammerhofer et al., 2015; Kyndt et al., 2017; Song et al., 2017). Notably, JA pathway plays a pivotal role in systemic defense induction against root knot nematodes (RKN) in rice plants (Nahar et al., 2011). In tomato, JA defense-dominated genotype (35S::Prosystemin) shows stronger resistance to nematodes than that of the wild-type Castlemart and JA-deficient mutant spr2 (Sun et al., 2010). Recent studies revealed that JA-induced enhanced defense against RKN is associated with alterations in antioxidative defense and photosynthetic processes (Bali et al., 2017). Furthermore, exogenous SA added as a soil drench is able to restrict J2s of nematodes in tomato roots by triggering a systemic acquired resistance (SAR)-like response (Molinari et al., 2014). A recent study showed that Trichoderma could reduce parasitic nematodes by triggering host defense, which was attributed firstly to SAprimed defense that limited root invasion of nematodes, and secondly to enhanced JA-regulated defense that antagonized the nematodes-induced deregulation of JA-dependent immunity (Martìnez-Medina et al., 2017). Interestingly, ABA interacts antagonistically with JA in rice defense against Meloidogyne graminicola, and thus, ABA application on rice plants aggravates nematode-caused disease symptoms (Kyndt et al., 2017). However, BRs suppress tomato defense against root-knot nematodes, without involving the classical defense pathways, such as SA, JA/ET or ABA signaling, rather triggering the apoplastic RESPIRATORY BURST OXIDASE HOMOLOGdependent MPK1/2/3 activation in tomato plants (Song et al., 2017).

Light is an essential environmental signal for plant growth and development, and also plays critical role in plant defense responses to pathogens through multiple hormonal pathways and antioxidant systems (Griebel and Zeier, 2008; Roden and Ingle, 2009; Cheng et al., 2016). Light regulates plant defense mainly by modifying SA and/or JA signaling and homeostasis (Cerrudo et al., 2012; de Wit et al., 2013). For instance, plant defense against Pseudomonas syringae depends on the light length, while longer light length results in an elevated SA accumulation, increased transcripts of pathogenesisrelated genes, and a more pronounced hypersensitive response (Griebel and Zeier, 2008). Light could suppress P. syringae pv. tabaci population in tobacco leaves through the accumulation of H2O<sup>2</sup> during infection (Cheng et al., 2016). In addition, light quality influences plant defense against diseases by modulating response of different photoreceptor to light wavelengths. Low red/far-red ratios (R:FR) reduce Arabidopsis resistance to necrotrophic pathogen Botrytis cinerea and JA responses (Cerrudo et al., 2012). Similarly, reduced R:FR ratios inhibit both SA-dependent and JAdependent disease resistance in Arabidopsis (de Wit et al., 2013).

Red light (RL) has more profound effect on activation of plant defense as compared to other monochromatic light (Wang et al., 2010; Yang et al., 2015b). For example, RL is more effective than other light qualities in inducing the defense response to powdery mildew disease through activation of SA-dependent signaling pathway, H2O<sup>2</sup> accumulation, and associated metabolism in cucumber plants (Wang et al., 2010). This is consistent with our recent study that supplemental RL (at night) shows a more profound effect than other light quality on nematode resistance (Yang et al., 2014). However, most studies related to RL-induced resistance against pathogens were focused on aboveground plant parts, while a few studies reported RL-affected plant resistance against root-feeding nematodes. Moreover, the role of RL in SA- and JA- mediated interactions, especially in belowground plant defense responses to root-knot nematodes is largely unknown.

Watermelon (Citrullus lanatus) is one of the most economically important global fruits. However, the rootknot nematode M. incognita causes tremendous economic losses in watermelon production throughout the world (Davis, 2007). Although light quality-induced resistance to RKN has been reported in some model plants, the role of RL in the defense response of watermelon plants to nematodes remains largely unknown. It is to be noted that the defense response of plants depends on light quality, light intensity, stress types, and plant species (Zarate et al., 2007; Hogewoning et al., 2010). Therefore, in the current study, we investigated the effect of RL on defense response to nematode infection in watermelon plants. Disease symptoms, expression of SA and JA biosynthesis-related genes, the content of SA, JA, ABA, and indole-3-acetic acid (IAA), enzymatic and non-enzymatic antioxidants were assayed. The results showed that RL exposed onto watermelon leaves could potentially activate systemic defense against RKN by modulating hormone pathways, antioxidant systems, and redox homeostasis. This study shed some lights on underlying mechanism of RLinduced systemic resistance against M. incognita in watermelon and may have potential implication in protected vegetable production and resistance breeding.

# MATERIALS AND METHODS

# Plant Material and Growth Condition

Seeds of watermelon (C. lanatus cv. Xinong No. 8) were sown in trays filled with a mixture of peat and vermiculite (2:1, v/v) and placed in growth chambers at a temperature of 25/19◦C (day/night), and a photoperiod of 12 hr day/night (8:00 a.m. to 8:00 p.m.), a photosynthetic photon flux density (PPFD) of 200 µmol m−<sup>2</sup> s −1 supplied from fluorescent tubes and a relative humidity (RH) of 70%. Seedlings at the three-leaf stage were transplanted into pots filled with steam-sterilized sands and watered with Hoagland's nutrient solution once in a week. At the four-leaf stage, seedlings were exposed to RL from 8:00 a.m. to 8:00 p.m. with a maximum wavelength at 660 nm provided by light-emitting photodiodes (LEDs, 10 W, Huizhou Kedao Technology Co. Ltd., Huizhou, China). Control plants were exposed to white LED light at the same time. The intensity of light was set at 200 µmol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> PPFD at the level of the canopy. Simultaneously, watermelon plants were subject to M. incognita infection. Mock plants that were treated with the same amount of water served as control.

The M. incognita (race 1) was provided by Prof. Deliang Peng, Chinese Academy of Agricultural Sciences, Beijing, China. A pure nematode culture was maintained on watermelon cultivar Xinong 8 grown in a greenhouse. Nematodes were extracted from 3-month-old infected plants using the Baermann method (Luc et al., 2005). The nematode suspension was collected after 48 h. Watermelon plants at four-leaf stage was inoculated with approximately 1000 second-stage juveniles (j2) of M. incognita per plant or mock treated with water pouring over the surface of the sand around the roots (Schaff et al., 2007).

Thus the study comprised four treatments, such as control (mock, white light and water solution), RL (RL treatment and water solution), RKN (white light and root knot nematode M. incognita infection), and RL+RKN (RL treatment and root knot nematode M. incognita infection). Six plants served as a replicate and there were four replicates for each treatment.

Seven-days after RL and RKN treatment, leaf and root samples were harvested for biochemical and gene expression analyses. However, for defense gene expression analysis, samples are also collected at 3 dpi and 14 dpi. Immediately after harvesting, samples were frozen in liquid nitrogen and stored at −80◦C. After 21 days exposure to RL and RKN, six plants from each treatment were randomly sampled for the evaluation of infection level. The roots were washed with running tap water, and the galls were counted with the aid of a stereomicroscope. Nematode susceptibility of the plants was evaluated by calculating the number of galls in the roots per plant. To visualize the galls, roots were boiled with a mixture of 0.8% acetic acid and 0.013% fuchsin for 3 min and washed with running tap water. Then the roots were destained in acid glycerol to visualize the galls.

# Determination of Malondialdehyde (MDA) and Electrolyte Leakage

The level of lipid peroxidation was estimated by quantifying the content of MDA in the roots. Root extracts were mixed with 20% trichloroacetic acid (TCA) containing 0.65% (W/V) 2-thiobarbituric acid (TBA) and incubated in boiling water for 25 min, and the reaction was stopped by immediately placing the samples in an ice bath as described previously (Hodges et al., 1999). MDA equivalents were calculated according to Hodges et al. (1999). The relative electrolyte leakage from root tissues was measured and calculated as previously described elsewhere (Cao et al., 2007).

# RNA Extraction and Quantitative Real Time PCR (qRT-PCR) Assay

Total RNA was extracted from 0.1 g of leaf and root tissues using the total RNA Miniprep Kit (Axygen Biosciences, Union City, CA, United States) according to the manufacturer's protocol. Genomic DNA was removed with the RNeasy Mini Kit (Qiagen, Hilden, Germany). Total RNA (1 µg) was reverse-transcribed for the synthesis of cDNA using the ReverTra Ace qPCR-RT Kit (Toyobo, Japan) according to the manufacturer's instructions. qRT-PCR was performed using the iCycler iQTM Real-time PCR Detection System (Bio-Rad, Hercules, CA, United States). The specific primers used for qRT-PCR are shown in Supplementary Table S1. PCR was performed using the SYBR Green PCR Master Mix (Takara, Tokyo, Japan). The PCR condition consisted of denaturation at 95◦C for 3 min followed by 40 cycles of denaturation at 95◦C for 30 s, annealing at 58◦C for 30 s and extension at 72◦C for 30 s. A dissociation curve was generated at the end of each PCR cycle to verify that a single product was amplified using software provided with the iCycler iQTM Real-time PCR Detection System. The software provided with the PCR system was used to calculate the threshold cycle values and to quantify the mRNA expression levels based on Livak and Schmittgen (2001). A set of multiple reference genes, clathrin adaptor complex subunit (ClCAC) and α-tubulin (ClTUA), was used as internal controls (Kong et al., 2014).

# Determination of Plant Hormones

Phytohormone extraction and quantification from watermelon leaves and roots were performed following previously described procedures with some modification (Song et al., 2017). Briefly, 100 mg of frozen leaf or root material was homogenized in 1 mL of ethyl acetate spiked with D5-JA, D5-SA, D6- IAA, and D6-ABA (OlChemIm) as internal standards to a final concentration of 100 ng mL−<sup>1</sup> . Tubes were centrifuged at 18,000 g for 10 min at 4◦C. The pellet was re-extracted with 1 mL of ethyl acetate. Both supernatants were combined and evaporated to dryness under N<sup>2</sup> gas. The residue was re-suspended in 0.5 mL of 70% methanol (v/v), centrifuged, and recentrifuged at 18,000 g for 2 min at 4◦C. The supernatants were analyzed in a liquid chromatography tandem mass spectrometry system (Varian 320-MS LC/MS, Agilent Technologies, Amstelveen, Netherlands). The parent ions, daughter ions, and collision energies used in these analyses are listed in Supplementary Table S2. Phytohormone concentration was expressed as nanogram per gram of fresh mass leaf and root material.

# Determination of H2O<sup>2</sup> Content

Content of H2O<sup>2</sup> was determined in leaves and roots by a peroxidase (POD) assay according to Willekens et al. (1997). Plant tissues (0.3 g) were homogenized in 3 ml of HClO<sup>4</sup> (1.0 M) using pre-chilled mortar and pestle. The homogenates were then transferred to 10 ml plastic tubes and centrifuged at 6000 g for 5 min at 4◦C. Resulting supernatant's pH was adjusted to 7.0 with 4 M KOH and centrifuged at 6000 g for 1 min at 4◦C. The supernatant was passed through an AG1.8 prepacked column and H2O<sup>2</sup> was eluted with double-distilled H2O. Equal recovery from the different samples was checked by analyzing duplicate samples. The reaction system consisted of the sample (900 µl) and 900 µl of reaction buffer containing 1 mM 2,2<sup>0</sup> -azino-di (3-ethylbenzthiazoline-6-sulfonic acid) in 100 mM potassium acetate at pH 4.4. Finally, reaction was initiated through addition of 3 µl (0.5 U) horseradish peroxidase. The absorption at 412 nm was recorded for the measurement of H2O2.

# Determination of Antioxidant Enzyme Activity, Glutathione Content, and Ascorbic Acid Content

Antioxidant enzyme activities were assayed spectrophotometrically in leaves and roots. For extraction of enzymes, frozen leaf or root sample (0.3 g) was ground with 2 mL ice-cold 50 mM PBS (pH 7.8) containing 0.2 mM EDTA, 2 mM AsA,

M. incognita infection.

and 2% PVP. Homogenates were centrifuged at 4◦C at 12,000 g for 20 min, then the resulting supernatants were used for the determination of enzymatic activity. The activities of superoxide dismutase (SOD), POD, and catalase (CAT) were measured following the previously described protocols (Li et al., 2017). Ascorbate peroxidase (APX) activity was analyzed by measuring the decrease in A290 according to the method of Nakano and Asada (1981). The glutathione (GSH) content was determined according to Sgherri and Navari-Izzo (1995) by an enzymatic recycling method. Ascorbic acid (AsA) content was measured using α-α 0 -bipyridyl-based colorimetric assay (Gillespie and Ainsworth, 2007).

# Statistical Analysis

All data were analyzed using the statistical software SAS 8.1 (SAS Institute Inc., Cary, North Carolina, United States). The data were subjected to analysis of variance (ANOVA) and means were compared with Duncan's multiple range test (p < 0.05).

# RESULTS

# Red Light Suppresses Nematode Incidence

The effects of RL on the resistance of watermelon plants against nematodes were investigated by determining the gall number in the roots. The results showed that control plants had 121 galls/plant root fresh weight, in sharp contrast, RL decreased the whole gall number per plants by 19% compared with the control (**Figure 1A**). As shown in **Figure 1B**, acid fuchsin staining revealed that the galls on the roots and J2 nematodes in the roots were obviously less in RL than white light after nematodes infection. Compared with mock treatment, RKN infection significantly increased electrolyte leakage and MDA, while RL+RKN treatment remarkably reduced electrolyte leakage and MDA compared with RKN treatment (**Figures 1C,D**).

# Red Light Regulates Defense-Related Gene Expression in Leaves and Roots

To explore the effect of RL on plant defense against nematodes, total RNA was isolated from roots and leaves at 3-, 7-, and 14 days- post inoculation (dpi) for the qRT-PCR analysis. The expression analysis of defense-related genes showed that transcript levels of PATHOGENESIS-RELATED 1 (PR1) and WRKY70 were slightly increased by RKN inoculation, followed by RL treatment, particularly at 7 dpi as compared with that in control plants. However, combined treatment of RL and RKN resulted in a drastic upregulation in the transcript levels of those genes, which peaked at 7 dpi (except for WRKY70 in roots). For instance, at 7 dpi transcript levels of PR1 gene were increased by 8-fold and 7-fold, respectively in leaves of RL+RKN plants compared with that of control and RKN only treatment. While transcripts of the PR1 genes in roots increased gradually until 14 dpi, transcript levels of both PR1 and WRKY70 declined after 7 dpi in both roots and leaves of RL+RKN plants. Nonetheless, transcript levels of PR1 and WRKY70 in roots of RL+RKN plants

still remained elevated at 14 dpi, while those transcripts in leaves declined to the levels of RKN alone in leaves of RL+RKN plants (**Figure 2**).

As we noticed a transient peak at 7 dpi for defense gene expression, we then assayed expression of defense hormonesrelated genes such as SA biosynthetic gene ISOCHORISMATE SYNTHASE (ICS), and JA biosynthetic gene ALLENE OXIDE SYNTHASE (AOS) and LIPOXYGENASE (LOX) at 7 dpi (**Figure 3**). Both RL and RKN single treatment significantly increased transcript levels of those genes over control, except for ICS in leaf, which was suppressed by RL or RKN. Consistently, transcripts analysis also showed that combined treatment of RL and RKN resulted in the highest transcript abundance of ICS, AOS, and LOX at 7 dpi (except for ICS in leaf). However, in all tissues tested, RL+RKN increased transcript levels of ICS, AOS, and LOX at 7 dpi as compared to that in RKN alone (**Figure 3**).

# Red Light Alters Endogenous Hormone Levels in Leaves and Roots

To further dissect RL and/or RKN-induced changes in endogenous hormone levels, we quantified concentrations of JA, SA, ABA, and IAA in roots and leaves of healthy and infected watermelon plants under RL and white light at 7 dpi. The results showed that contents of IAA, JA, and SA were much higher in roots than that in leaves in all treatments. However, the ABA contents in leaves were approximately 9-fold lower than that of root concentration. As shown in **Figure 4**, RL enhanced JA, ABA, and IAA contents in leaves regardless of RKN treatment. In leaves, while JA content declined, SA content enhanced in response to RKN inoculation. However, RL treatment with RKN inoculation increased JA content but decreased SA content in leaves as compared to that of RKN treatment. Meanwhile, the responses of endogenous hormones to RL or RKN were different in different tissues. In roots, both JA and SA contents were induced by RL+RKN treatment compared to that in RKN alone. ABA content in roots followed the same trend of SA content in leaves, which was not affected by RL in absence of RKN inoculation, but was significantly suppressed by RL in presence of RKN infection. While RKN treatment increased IAA content as compared with mock, RL had no significant effect on IAA content either in mock or RKN treatment (**Figure 4**).

# H2O<sup>2</sup> Is Involved in RL-Mediated Defense Response to RKN

H2O<sup>2</sup> is an important molecule that plays dual roles in plant stress response. Therefore, we measured H2O<sup>2</sup> contents in leaves and roots of healthy and infected watermelon plants grown under WL and RL (**Figure 5**). We found that RL alone did not affect the content of H2O<sup>2</sup> both in leaves and roots in absence of RKN infection. However, RKN infection significantly induced H2O<sup>2</sup> levels. More importantly, H2O<sup>2</sup> contents in leaves and roots of RL+RKN plants were significantly higher than that of only RKN plants, indicating a specific signaling role of H2O<sup>2</sup> in RL-promoted defense against RKN.

To further understand the oxidative status as influenced by RL and/or RKN in plants, we analyzed activities of some key antioxidant enzymes that are responsible for rapid scavenging of ROS. Consistent with the membrane damage as shown in **Figure 2**, the activities of SOD, POD, CAT, and APX were not affected by RL in absence of RKN inoculation. However, RKN infection resulted in a significant increase in the activities of those enzymes both in leaves and roots (**Figure 6**). Interesting, RL+RKN treatment increased all above antioxidant enzyme activities in leaves and roots, compared with that in RKN alonestressed plants (**Figure 6**).

# Red Light Enhances Defense Against RKN by Stimulating Redox Homeostasis

Ascorbate and GSH are the key components of AsA-GSH pools that control cellular redox state and play important role in defense against nematode stress. While total ascorbate remained virtually unchanged in response to RL, RL treatment significantly reduced total GSH content, particularly in leaves, in absence of RKN inoculation (**Figure 7**). However, RKN treatment without RL promoted total AsA and reduced GSH in leaves/roots by 29.73%/83.39%, and 16.02%/31.66%, respectively, compared with mock. Interesting, compared with RKN only treatment, RL treatment on RKN-inoculated plants further increased total AsA and GSH. RL and RKN-induced changes in AsA-GSH pools differentially modulated redox state in leaves and roots. For instance, RL treatment increased AsA:DHA in leaves but not in roots regardless of RKN infection. In line with the trend of total GSH and total ascorbate, RL+RKN significantly increased the

ratio of reduced/oxidized glutathione (GSH/GSSG) both in leaves and roots as compared to that in only RKN treatment, indicating a specific effect of RL on redox homeostasis in response to RKN inoculation (**Figure 7**).

# DISCUSSION

Red light stimulates plant defense against both biotrophic and necrotrophic pathogens in a range of plant species (Mutar and Fattah, 2013; Yang et al., 2015b). However, the mechanisms of light quality-mediated defense responses vary from species to species and thus the mechanisms of RL-induced watermelon defense against RKN remain largely unknown. In addition, the roles of diurnal RL exposure of above-ground foliar parts in plant defense against below-ground pathogens remain elusive, particularly in watermelon plants. In this study, we found that the incidence of RKN in watermelon plants was significantly suppressed by diurnal RL compared with that under white light environment. RL-induced enhancement in defense against RKN was closely associated with increased transcript levels of PRI, WRKY70, ICS, AOS, and LOX, contents of SA, JA, and H2O2, antioxidant enzyme activity and redox homeostasis in watermelon. This study also suggests that RL improves systemic resistance against RKN through selective regulation of SA and JA biosynthesis in different tissues of watermelon plants.

FIGURE 5 | H2O<sup>2</sup> content as affected by red light and/or root knot nematode in leaves and roots of watermelon plants. H2O<sup>2</sup> content was assayed at 7 days post inoculation (dpi) with M. incognita. Plants were kept in white light (WL, open column) or red light (RL, gray column) without (Mock) or with the inoculation of M. incognita (RKN). Data are the means ± SD of four replicates. Different letters indicate statistically significant differences (Duncan's multiple range test with p < 0.05).

In the current study, it is highly likely that RL exposure onto watermelon leaves activated systemic resistance against root disease. RL-induced inhibition of nematode development,

replicates. Different letters indicate statistically significant differences (Duncan's multiple range test with p < 0.05).

lipid peroxidation, and electrolyte leakage provide convincing evidence that RL could minimize RKN-induced oxidative stress and damage to watermelon plant. The suppression of RKN disease incidence by RL in the current study is consistent with previous studies in Arabidopsis and tomato, where treatment with continuous and nightly RL treatment induced systemic disease resistance against root-knot nematode M. javanica and M. incognita, respectively (Islam et al., 2008; Yang et al., 2014).

Phytohormones such as SA, JA, ET, IAA, ABA, and BR function as signaling molecules and mediate defense response to phytopathogens (Yang et al., 2015a; Song et al., 2017). In particular, JA and SA play critical roles in plant defense against nematodes including M. incognita (Branch et al., 2004; Kammerhofer et al., 2015). In our study, plants inoculated with M. incognita accumulated a higher amount of SA, but a lower amount JA in leaves (**Figure 4**). Activation of the SA pathway and suppression of JA pathway is believed to be a strategy by which nematodes antagonizes the host plant immune response for a successful invasion (Ithal et al., 2007; Martìnez-Medina et al., 2017). Thus, our results are consistent with the observations that SA and JA signaling often display inverse patterns of expression in above-ground plant parts (Pieterse et al., 2012). However, JA levels increased in roots upon RKN infection, suggesting that JA might be transported through vascular tissues from above-ground to the below-ground parts and triggered plant defense against nematodes, which resulted in a higher JA content in roots than that in leaves following the RKN infection (Zhang and Baldwin, 1997). We also noticed that RL+RKN treated plants showed an increased JA accumulation both in leaves and roots compared with that of RKN treatment alone (**Figure 4**), which was consistent with the expression of JA biosynthetic genes in respective tissues. Accumulating evidence from previous studies suggests that JA signaling is required for the rhizo-bacteria-induced systemic resistance (Fujimoto et al., 2011). Thus it is highly likely that RL-induced JA biosynthesis perhaps triggers induced systemic resistance (ISR) against RKN in watermelon (Zhang and Baldwin, 1997). Our results are in agreement with the previous reports on tomato plants, which showed that foliar application of methyl jasmonate (MeJA) significantly reduces the infection of RKN (Fujimoto et al., 2011), and the systemic defense signals associated with the JA pathway are transported between above-ground and below-ground plants parts (van Dam et al., 2003). However, SA content in leaves decreased in RL+RKN plants compared with that in RKN alone due to

an antagonistic effect of increased JA accumulation in leaves (**Figure 4**).

Unlike leaves, exposure to RL resulted in an increased transcript levels of SA biosynthetic gene ICS and elevated SA content in roots after M. incognita infection (**Figure 3**), indicating that SA signaling pathway was activated by RL to induce watermelon resistance against nematodes in roots (Nandi et al., 2003; Branch et al., 2004). Notably, SA is also involved in Mi1 mediated defense response to root-knot nematode M. incognita in tomato (Branch et al., 2004). Meanwhile, JA biosynthetic genes LOX and AOS as well as JA content were remarkably induced in roots by RL as compared with that in white light condition in response to RKN infection (**Figure 3**), which may indicate a potential synergistic local interaction between SA and JA to combat RKN in roots (van Wees et al., 2000). These results are well in agreement with early findings that RL could stimulate SA and JA biosynthesis and signaling pathway genes to activate plant defense against RKN (van Wees et al., 2000; Yang et al., 2014). This implies that RL-mediated plant defense is attributed to a proper integration of SAR and ISR, which is an interesting mechanism for enhancing plant immunity.

Transcript factor WRKY70 plays a critical role in resistance (R) gene Mi-1-mediated resistance against RKN in plants (Atamian et al., 2012). Additionally, WRKY72 is transcriptionally up-regulated in the roots of resistant tomato genotype and mediates Mi-1.2-induced effectortriggered immunity against RKN (Bhattarai et al., 2010).

A recent study shows that class-A heat shock factor (HsfA1a) and whitefly induced 1 (Wfi1)-dependent apoplastic H2O<sup>2</sup> accumulation is required for Mi-1.2-mediated resistance against RKN in tomato plants (Zhou et al., 2018). In the current study, we found that RL could induce WRKY70 transcript and H2O<sup>2</sup> accumulation both in leaves and roots of watermelon plants in presence of RKN inoculation (**Figure 2**), suggesting that WRKY70 and ROS signaling are potentially involved in RL-induced plant defense against RKN in watermelon plants. In addition, tissue specific induction of WRKY70 may suggest a node of convergence between JAmediated and SA-mediated signals in plant defense (Li et al., 2004).

A close look into the hormone profiles showed that RL can increase IAA content significantly in the leaf tissues (**Figure 4**), which is consistent with a former research report that RL could trigger phyB-mediated auxin synthesis and increase lower hypocotyl elongation (Guo et al., 2016). Notably, nematode infection also increases root elongation irrespective of white light and RL treatment (Grunewald et al., 2009) as nematode invasion enhances nutrient demand for cell division and differentiation, resulting in giant cell formation at the early stage of nematode infection (Shukla et al., 2017). However, RL treatment on RKNinoculated plants did not improve IAA content compared with white light treatment on RKN-inoculated plants, indicating that IAA was not involved in RL-induced defense against RKN. In plants, ABA generally plays a negative role in defense against RKN. In the present study, only RKN treatment did not affect ABA content both in leaves and roots, however, RL treatment with RKN inoculation enhanced ABA content in leaves (**Figure 4**). Meanwhile, RL+RKN treatment slightly decreased ABA content compared with that of RKN alone, implying that increased JA biosynthesis might suppress ABA accumulation in roots (Nahar et al., 2012; Kammerhofer et al., 2015). All these results led us to propose that RL-induced watermelon defense against RKN is principally mediated by JA-dependent ISR and SA-dependent SAR, whereas auxin and ABA pathways possibly play minor role in defense against RKN infection.

Hydrogen peroxide (H2O2) functions as a signaling molecule and mediates plant responses to abiotic and biotic stresses (Neill et al., 2002). In our study, H2O<sup>2</sup> was induced by RL specifically in presence of RKN in watermelon plants (**Figure 5**). This observation supports the assumption that RL could stimulate H2O<sup>2</sup> for defense against pathogens (Wang et al., 2010). Moreover, SA-induced SAR is likely to be mediated by elevated amount of H2O<sup>2</sup> (Chen et al., 1993), indicating a signaling role of H2O<sup>2</sup> in SAR (Alvarez et al., 1998). Therefore, we speculate that RL-induced H2O<sup>2</sup> accumulation in RKN-infected roots might induce H2O<sup>2</sup> in leaves, thus up-regulating the JA signaling pathway genes and JA accumulation (Orozco-Cárdenas et al., 2001).

Both enzymatic and non-enzymatic anti-oxidant systems play an important role in plant tolerance to root-knot nematode M. incognita-induced oxidative stress (Oliveira et al., 2012). Here, we also found that the activities of key antioxidant enzymes such as SOD, CAT, and POD and APX were highly induced by RKN (**Figure 6**). Moreover, RL further stimulated antioxidant enzyme and enhanced plant resistance against RKN stress, which is evidenced by reduced MDA content and electrolyte leakage from roots under RL+RKN treatment compared with that in only RKN treatment in watermelon plants. Furthermore, light quality signals, particularly R/FR ratios, are important regulators of antioxidant synthesis and accumulation (Bartoli et al., 2009). R/FR signaling has a major control over the extent of AsA accumulation in leaves over a single photoperiod and JA signaling triggers ascorbate metabolism (Shan and Liang, 2010). In line with this, we noticed a significantly increased AsA content following exposure of watermelon plants to RL+RKN treatment, indicating the important role of RL-induced AsA in enhancing plant defense against RKN. GSH is a key regulator of redox signaling and its buffering activates defense genes (Foyer and Noctor, 2011). Similar to AsA, GSH content was increased by RL treatment in presence of RKN infection, which eventually increased redox homeostasis and enhanced tolerance to RKN. Our findings are in agreement with previous reports that light quality signals regulate the GSH pool (Bartoli et al., 2009) and JA signaling can promote GSH metabolism (Shan and Liang, 2010). Previous reports have also indicated that light quality, especially RL could influence plant tolerance to abiotic and biotic stressors via photoreceptors and phytohormones and photoreceptors such as phytochrome as well (Cerrudo et al., 2012; Yang et al., 2015b; Wang et al., 2016). For instance, PhyA and PhyB regulate plant tolerance to low temperature stress via ABA-dependent JA signaling in tomato (Wang et al., 2016). Phytochromes also play a critical role in resistance against Magnaporthe grisea by regulating SA- and JA-dependent defense pathways in rice plants (Xie et al., 2011). Inactivation of Phy B results in low levels of constitutive defenses and down-regulation of MeJA-induced defenses against herbivore in tomato plants (Cortes et al., 2016). Therefore, in addition to phytohormones, phytochromes may have a role in RLmediated defense against RKN in watermelon. However, a limitation with a crop like watermelon, versus other model plants, is the lack of targeted mutants in the pathways of interest to establish cause and effect. Therefore, it will be interesting to further explore the role of phytohormones and phytochromes in RL-mediated watermelon defense against RKN by establishing advanced genetic tools for targeted gene suppression in watermelon.

In the current study, we found that RL plays a vital role in plant defense against root knot nematode in watermelon plants. RL acts as a positive signal that triggers tissue specific accumulation of phytohormones and ROS. In addition, RL improves redox homeostasis by modulating the activities of antioxidant enzymes as well as GSH and ASA contents. Our circumstantial evidence suggests that RL exposure on watermelon leaves can enhance systemic defense against RKN infection through a coordinated regulation of SA, JA and redox signaling in watermelon plants. This study provides new insights into the underlying mechanism of RL-induced systemic defense against M. incognita in watermelon, which may have potential implication in protected vegetable production.

# AUTHOR CONTRIBUTIONS

fpls-09-00899 July 7, 2018 Time: 16:52 # 11

Y-XY, ZY, and JC designed the research. Y-XY and CQW executed the experiments. Y-XY, CJW, CPW, and GA analyzed and discussed the data. Y-XY, GA, and JC wrote the manuscript.

# FUNDING

This work was jointly supported by the National Natural Science Foundation of China (31560572), Jiangxi Province Postdoctoral

# REFERENCES


Science Foundation, China (2016KY06), Natural Science Foundation of Jiangxi Province, China (20171BAB214030), and the Henan University of Science and Technology (HAUST) Research Start-up Fund for New Faculty (13480058).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00899/ full#supplementary-material


Annu. Rev. Cell Dev. Biol. 28, 489–521. doi: 10.1146/annurev-cellbio-092910- 154055


fpls-09-00899 July 7, 2018 Time: 16:52 # 12

Meloidogyne incognita. Plant Growth Regul. 76, 167–175. doi: 10.1007/s10725- 014-9986-9


Wfi1 transcription and H2O2 production. Plant Physiol. 176:01281.2017. doi: 10.1104/pp.17.01281

**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 Yang, Wu, Ahammed, Wu, Yang, Wan 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.

# Two Strategies of Pseudomonas syringae to Avoid Recognition of the HopQ1 Effector in Nicotiana Species

Patrycja Zembek<sup>1</sup>† , Aleksandra Danilecka<sup>1</sup>† , Rafał Hoser<sup>1</sup>† , Lennart Eschen-Lippold<sup>2</sup>† , Marta Benicka<sup>1</sup> , Marta Grech-Baran<sup>1</sup> , Wojciech Rymaszewski<sup>1</sup> , Izabela Barymow-Filoniuk<sup>1</sup> , Karolina Morgiewicz<sup>1</sup> , Jakub Kwiatkowski<sup>1</sup> , Marcin Piechocki<sup>1</sup> , Jaroslaw Poznanski<sup>1</sup> , Justin Lee<sup>2</sup> , Jacek Hennig<sup>1</sup> and Magdalena Krzymowska<sup>1</sup> \*

1 Institute of Biochemistry and Biophysics (PAS), Warsaw, Poland, <sup>2</sup> Leibniz Institute of Plant Biochemistry, Halle, Germany

#### Edited by:

Zhengqing Fu, University of South Carolina, United States

#### Reviewed by:

Brian H. Kvitko, University of Georgia, United States Hai-Lei Wei, Cornell University, United States

#### \*Correspondence:

Magdalena Krzymowska krzyma@ibb.waw.pl †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 28 February 2018 Accepted: 15 June 2018 Published: 10 July 2018

#### Citation:

Zembek P, Danilecka A, Hoser R, Eschen-Lippold L, Benicka M, Grech-Baran M, Rymaszewski W, Barymow-Filoniuk I, Morgiewicz K, Kwiatkowski J, Piechocki M, Poznanski J, Lee J, Hennig J and Krzymowska M (2018) Two Strategies of Pseudomonas syringae to Avoid Recognition of the HopQ1 Effector in Nicotiana Species. Front. Plant Sci. 9:978. doi: 10.3389/fpls.2018.00978 Pseudomonas syringae employs a battery of type three secretion effectors to subvert plant immune responses. In turn, plants have developed receptors that recognize some of the bacterial effectors. Two strain-specific HopQ1 effector variants (for Hrp outer protein Q) from the pathovars phaseolicola 1448A (Pph) and tomato DC3000 (Pto) showed considerable differences in their ability to evoke disease symptoms in Nicotiana benthamiana. Surprisingly, the variants differ by only six amino acids located mostly in the N-terminal disordered region of HopQ1. We found that the presence of serine 87 and leucine 91 renders PtoHopQ1 susceptible to N-terminal processing by plant proteases. Substitutions at these two positions did not strongly affect PtoHopQ1 virulence properties in a susceptible host but they reduced bacterial growth and accelerated onset of cell death in a resistant host, suggesting that N-terminal mutations rendered PtoHopQ1 susceptible to processing in planta and, thus, represent a mechanism of recognition avoidance. Furthermore, we found that co-expression of HopR1, another effector encoded within the same gene cluster masks HopQ1 recognition in a strain-dependent manner. Together, these data suggest that HopQ1 is under high host-pathogen co-evolutionary selection pressure and P. syringae may have evolved differential effector processing or masking as two independent strategies to evade HopQ1 recognition, thus revealing another level of complexity in plant – microbe interactions.

Keywords: TTSS effectors, HopQ1, HopR1, virulence, Pseudomonas syringae

# INTRODUCTION

Pseudomonas syringae is a widespread bacterium that can infect almost 200 plant species. Its various pathovars cause diseases in several agriculturally important plants – halo blight in bean, bacterial speck in tomato, bacterial blight in soybean or broccoli, angular leaf spot in cucumber or wildfire in tobacco. Like many other gram-negative pathogenic bacteria, P. syringae secretes type III effectors into host cells to facilitate colonization of plants. The effectors play multiple roles during the infection process. They are primarily used to subvert the host cellular machinery, but they are also involved in nutrient acquisition or control of microbial community

(Snelders et al., 2018). Nearly 100 effector families have so far been identified in P. syringae (Büttner, 2016), however, the effector repertoire (effectome) of a particular strain does not usually exceed 30 proteins (Baltrus et al., 2011). Even a single effector may define the host range by promoting bacterial multiplication in one plant while in other species presence of this same effector may trigger plant defense response leading to cessation of bacterial growth. Thereby, the composition of the effectome contributes to host specificity of a given bacterial strain. Various forces shape the effectome but the most significant is the need to avoid plant recognition (Koebnik and Lindeberg, 2011). Several mechanisms to overcome selection pressure have been described for individual effectors including their loss, mutagenesis or acquisition of novel domains but the mechanisms that tailor the whole effectomes remain largely unknown (Koebnik and Lindeberg, 2011). Recent reports show (Wei et al., 2015, 2018) that interplay between effectors contributes to several aspects of the infection process including bacterial growth rate in plant tissues, symptom development but also suppression of host defense. The fact that one effector is able to suppress response triggered by the second effector from the cooperating pair suggests that adaptation to the partner may be another factor that drives evolution of effectors.

HopQ1 (for Hrp outer protein Q) is an effector hypothesized to be acquired recently by P. syringae (Rohmer et al., 2004). It promotes disease development in bean, tomato, and Arabidopsis plants (Ferrante et al., 2009; Li et al., 2013b). In contrast, HopQ1 is recognized by Nicotiana spp., which have evolved systems to sense its presence and initiate defense responses (Wei et al., 2007). This response is mediated by Roq1 (for Recognition of XopQ 1), a receptor that directly interacts with HopQ1 and XopQ, a close homolog from Xanthomonas spp. Therefore, to avoid perception, strains of P. syringae pv. tabaci evolutionarily eliminated the sequence encoding HopQ1 from their genomes (Ferrante et al., 2009). Here, we report two mechanisms employed by P. syringae to remain undetected in Nicotiana spp. despite expressing HopQ1.

# MATERIALS AND METHODS

# P. syringae Strains and Inoculation

Sequences encoding HopQ1 from P. syringae pv. tomato DC3000 (PtoHopQ1), HopR1 from P. syringae pv. phaseolicola 1448A (PphHopR1) or from tomato DC3000 (PtoHopR1) were PCR amplified (see Supplementary Tables 1, 2 for the list of the strains and primers used in this study) and cloned into the pENTR/D-TOPO vector. hopQ1 variants were made by sitedirected mutagenesis, as described previously (Giska et al., 2013). All the sequences were PCR amplified to add appropriate restriction sites and cloned into pJET 1.2. Next, the sequences were cut with the restriction enzymes and cloned under the control of Tac promoter in pBBR1-MCS2-pTac, the modified broad-host-range vector pBBR1MCS-2 (Giska et al., 2013).

To prepare pseudo-operons that co-express HopQ1 and HopR1, hopR1 variants were PCR amplified with primers adding a ribosome binding site and a FLAG-epitope encoding sequence to the 5<sup>0</sup> and 3<sup>0</sup> ends of the products, respectively, and KpnI restriction sites to both ends. PCR products were cloned into pJET 1.2 and re-cloned into pBBR1MCS-2-pTac derivatives carrying appropriate hopQ1 sequences. All the constructs were electroporated into P. syringae pv. syringae B728a and PtoDC3000D28E P. syringae strains. The bacteria were prepared for inoculation as described previously (Krzymowska et al., 2007). Following centrifugation at 3,500 × g for 10 min, the pellet was washed once and resuspended in sterile 10 mM MgCl2. The bacterial suspension was adjusted to OD<sup>600</sup> = 0.2 (that corresponds to approximately 10<sup>8</sup> colony forming units [cfu]/ml) and further diluted, as indicated. Bacterial titers were checked by plating.

To assay bacterial growth in Nicotiana benthamiana, whole plants were dip-inoculated with Pss (culture density 10<sup>6</sup> cfu/ml) expressing HopQ1 variants or the pseudo-operons. At the indicated time points, three 1 cm-diameter leaf disks were punched out, surface-sterilized with 70% ethanol for 1 min, rinsed with sterile water for 1 min and ground in 300 µl 10 mM MgCl2. Serial dilutions were plated on LB agar plates for bacteria enumeration. To assess the impact of HopQ1 variants on Pss growth in Nicotiana tabacum plants, the bacterial suspensions expressing the indicated variants were infiltrated into leaves and the bacteria were isolated at the indicated time points.

For assessment of hypersensitive response (HR) development in tobacco, the leaves were infiltrated using a needleless syringe with bacterial suspensions adjusted to approximately 10<sup>8</sup> cfu/ml. PtoDC3000D28E (50 µl) was applied locally and to measure loss of cell membrane integrity whole tobacco leaves were infiltrated with Pss suspension.

# Transient Expression in Protoplasts

To express C-terminally HA-tagged HopQ1 variants in Arabidopsis protoplasts, the sequences encoding the effector variants were recombined into pUGW14 vector (Nakagawa et al., 2007). Protoplast isolation, transformation and elicitation with flg22 was performed as described previously (Yoo et al., 2007; Ranf et al., 2011). Activation of MAP kinases was assayed with antibodies directed against the phosphorylated activation loop (anti-pTEpY; #9101 Cell Signaling, Tech.). Protein amounts detected by immunoassay were calculated as described by Imkampe et al. (2017). Luciferase reporter activity (pNHL10- LUC) was measured and normalized as described previously (Pecher et al., 2014).

# Ion Conductivity

At the indicated time points, eight leaf disks (1 cm diameter) were cut from infiltrated zones and floated abaxial side up on 5 ml milliQ water for 10 min at 18◦C with gyratory agitation (50 rpm). The conductivity of the water was measured with a WTW InoLab Multi 9310 IDSCDM83 benchtop meter and expressed in µScm−<sup>1</sup> .

# Confocal Laser Scanning Microscopy

To generate a construct expressing HopR1-eYFP, the entry clones carrying HopR1 variants were LR recombined with the Gateway

pGWB441 destination vector. The resulting constructs were electroporated into Agrobacterium tumefaciens (GV3101) cells. Subsequently, A. tumefaciens cultures containing the constructs were infiltrated into N. benthamiana leaves, and tissues were analyzed using an FV1000 confocal system (Olympus, Tokyo, Japan) equipped with a 60x/1.2 water immersion objective lens. eYFP was excited with the 515 nm line from an argon ion laser and fluorescence signals were recorded using diffraction grate based spectral detector with 530–640 nm detection window. Chlorophyll autofluorescence was excited with 440 nm laser diode and detected using 750/50 emission filter (Chroma).

# Accession Numbers

Sequence data from this article can be found in the GenBank data libraries under accession numbers PphHopQ1 (AAZ37975.1), PtoHopQ1 (also known as HopQ1-1, NP\_790716.1), PphHopR1 (AAZ37024.1), PtoHopR1 (NP\_790722.1).

# RESULTS AND DISCUSSION

Despite a very high level of amino acid (aa) identity between two HopQ1 variants derived from P. syringae pv. phaseolicola 1448a (Pph) and P. syringae pv. tomato DC3000 (Pto), their expression in a virulent P. syringae strain resulted in different disease outcomes in dip-inoculated N. benthamiana plants (**Figure 1A**). Consistent with our previous experiments (Giska et al., 2013), PphhopQ1 rendered P. syringae pv. syringae B728a (Pss) avirulent toward N. benthamiana. Introduction of PtohopQ1 to Pss also reduced disease severity of Pss but compared to bacteria expressing PphhopQ1, the bacteria multiplied more rapidly at the early stages of the infection and evoked severe disease symptoms.

Since PtoHopQ1 and PphHopQ1 proteins differ only in six aa (**Figure 1B**), we aimed at identification of those residues that affect the effector properties. To this end, we generated variants by site-directed mutagenesis. To reduce the number of possible

FIGURE 1 | PtoHopQ1 and PphHopQ1 differ in their avirulence properties. (A) Nicotiana benthamiana plants were dip-inoculated with Pseudomonas syringae pv. syringae B728A expressing either one of the effectors. Bacterial titers were determined at 0/1/2/4 days post inoculation (upper panel). Note that due to severe tissue collapse of Pss infected leaves, the collection of samples was not possible at 4 dpi. Lower panel shows the plants 7 days after inoculation. The experiment was performed three times with similar results. Data were analyzed using repeated measures analysis of variance (ANOVA), followed by Tukey HSD post hoc test performed for each time point separately. Statistically distinct groups are marked with different letters above each column. (B) Representative model of PtoHopQ1 generated by I-TASSER (Zhang, 2008; Roy et al., 2010) and visualized using Yasara View (Krieger and Vriend, 2014). The flexible N-terminal part of the protein, which varies in particular models, is shown in gray. The residues that differ in PtoHopQ1 compared to PphHopQ1 are denoted in ball and stick representation. (C) Comparison of HopQ1 variants from the selected strains.

#### FIGURE 2 | Continued

protein (CFP) and AvrPto served as negative and positive controls, respectively. Fourteen hours after transformation, protoplasts were treated with 100 nM flg22 for 10 min. MAPK activation was monitored by immunoblot analysis with anti-pTEpY antibodies and the expression level of HopQ1 variants was checked with anti-HA antibodies. Amido black staining of the membranes was used to demonstrate equal loading. The numbers correspond to ImageJ-based quantification of the protein band intensities (MAPK activation strength is the sum of all three MAPK bands). (B) Arabidopsis protoplasts were co-transformed with constructs expressing HopQ1 variants, pNHL10-LUC (luciferase) as a reporter and pUBQ10-GUS (β-glucuronidase). Luciferase activity was recorded for 3 h, following flg22 treatment, and depicted as LUC/GUS ratios. Data for each protein variant were analyzed using repeated measures ANOVA, yielding significant effects of variant, time and their interaction (p < 0.001). Differences between H2O-treated samples (green traces) and flg22 treatments were tested with Student's t-test. Statistical significant differences in the flg22-treated samples as compared to the H2O-treated samples are highlighted by the color-coded p-values adjusted using Benjamini–Hochberg procedure. (C) Area under the curve (AUC) values were calculated for the graphs. One-way ANOVA was performed separately for both treatments and was followed by Tukey HSD post hoc test. Letters correspond to statistically homogenous groups (p < 0.05). Inlet: τ parameter values obtained after curve fitting to the fold changes for each protein variant (see Supplementary Figure 1). Bars correspond to standard errors in parameter estimation. The experiment was performed three times with similar results.

variants, we focused on aa combinations that naturally occur in HopQ1 effectors in other P. syringae pathovars (**Figure 1C**), namely pv. savastanoi NCPPB3335, pv. actinidiae MAFF302091, pv. mori and pv. oryzae 1\_6. Based on this sequence comparison, we prepared constructs encoding four PphHopQ1 mutants (L19S; V31A; L19S\_S72A; L19S\_G154E) and one PtoHopQ1 mutant (S87L\_L91R).

PtoHopQ1 has been reported previously to suppress flg22 induced activation of MAP kinases in Arabidopsis (Hann et al., 2014). Therefore, to assess the properties of the HopQ1 variants, we transiently expressed them in Arabidopsis mesophyll protoplasts. Both PtoHopQ1 variants equally suppressed flg22 mediated activation of MPK3, MPK6, and MPK4/11 in protoplasts (**Figure 2A**). For the PphHopQ1 wild type and V31A mutant versions, also the same levels of suppression were observed. However, suppression of MAPK activation was less pronounced by the L19S single mutant and both double mutants, L19S-S72A and L19S-G154E, completely lost suppressive activity (**Figure 2A**).

To quantify the impact of HopQ1 variants on flg22-induced plant responses, we used a previously described luciferase reporter system that monitors expression of the firefly luciferase (LUC) gene under the control of the flg22-inducible A. thaliana NHL10 (NDR1/HIN1-LIKE 10) promoter (Boudsocq et al., 2010; Pecher et al., 2014). In this assay, PphHopQ1-WT strongly suppressed basal and flg22-induced NHL10 promoter activity over a 3 h measurement period (**Figure 2B**). To facilitate comparison between the various HopQ1 variants and between treatments, total promoter activities were further calculated as "area under the curve" (AUC; **Figure 2C** and Supplementary Figure 1). All PphHopQ1 mutant variants tested suppressed both basal and flg22-induced promoter activities, but were

significantly less active than the WT version. Expression of both PtoHopQ1 variants enhanced basal promoter activity compared to the CFP transfection control (**Figure 2B**), leading to significantly higher total signal under control conditions (H2O; **Figure 2B**). Importantly, upon flg22 elicitation, PtoHopQ1 expressing samples did not reach the level of the CFP-expressing controls (**Figure 2B**) and total signal calculations revealed a significant reduction, indicative of a strong suppressive capability on elicitor-induced activity.

Interestingly, we noticed that PtoHopQ1, in contrast to PphHopQ1, was reproducibly detected in two forms, presumably the full-length and a truncated version. Since the HA-tag was located at the C-terminus of PtoHopQ1, we could conclude that the truncated form of the effector was N-terminally cleaved. The presence of two forms of PtoHopQ1 was previously observed in transgenic Arabidopsis and tomato plants (Li et al., 2013a), as well as in N. benthamiana leaves transiently expressing PtoHopQ1 (Li et al., 2013b), suggesting that PtoHopQ1 is prone to N-terminal processing. Consistent with this notion, the six aa that are different in PtoHopQ1 compared to PphHopQ1 lie within its N-terminus (**Figure 1B**). In contrast to the parental PtoHopQ1, the PtoHopQ1\_S87L\_L91<sup>R</sup> mutant was detected mainly in the presumed full-length form indicating that the presence of S87 and L91 renders PtoHopQ1 susceptible to proteolytic cleavage. These two aa are located in the predicted hinge region (loop) linking the N-terminal and the central nucleoside hydrolase domains, putatively an exposed area susceptible for cleavage (**Figure 1B**). ELM (Eukaryotic Linear Motif) analysis (Dinkel et al., 2016) revealed a region, located between aa 89 and 93, as a putative subtilisin cleavage site (Supplementary Figure 2) and subtilisin-like proteases (subtilases) are implicated in plant defense (Figueiredo et al., 2014). Although our data indicate that S87 and L91 are involved in N-terminal processing, they are not absolutely required since the S87L-L91R mutant still accumulates the truncated form but to a reduced extent. Consistently, reciprocal substitutions within PphHopQ1 (PphHopQ1\_L87S\_R91L) lead to the partial cleavage of the effector indicating that although presence of these two aa renders HopQ1 susceptible to the cleavage it is not sufficient for the effective processing (Supplementary Figure 3). Importantly, the PtoHopQ1\_S87L\_L91<sup>R</sup> mutant showed a similar behavior like the wild type version in the luciferase reporter assay, thus, the cleavability of PtoHopQ1 can be uncoupled from the enhanced basal promoter activity as well as the suppression of flg22 induced promoter activity mediated by the effector. To further analyze the importance of in planta PtoHopQ1 processing for its virulence function, we concentrated our efforts on this aspect using the PtoHopQ1\_S87L\_L91<sup>R</sup> mutant as a tool.

bacterial growth rate in a resistant host. (A) Ion leakage assay. Nicotiana tabacum leaves were infiltrated with Pss bacteria (culture density ca. 10<sup>8</sup> cfu/ml) expressing PphHopQ1, PtoHopQ1, or PtoHopQ1\_S87L\_L91R. At selected time points, cellular ion leakage to the apoplast was measured after floating leaf disks on the milliQ water. The photographs show leaf tissue at the time of visible symptoms development and/or at the maximal conductivity level. (B) Pss growth in planta. The bacterial suspensions (ca. 10<sup>3</sup> cfu/ml) expressing the indicated variants were infiltrated into N. tabacum leaves and at the indicated time points bacteria were isolated and serial dilutions were plated for enumeration. Data were analyzed using repeated measures ANOVA, followed by Tukey HSD post hoc test performed for each time point separately. Letters correspond to statistically homogenous groups (p < 0.05). The experiment was performed twice with similar results.

Compared to patchy and non-homogenous necroses obtained with N. benthamiana, N. tabacum is a better model, for investigating HopQ1-triggered HR. It was previously shown that full-length HopQ1 triggers the HR in N. tabacum (Giska et al., 2013; Li et al., 2013a). Since expression of HopQ1 lacking the first 89 aa did not lead to visible tissue collapse (Li et al., 2013a), we hypothesized that the recognition of HopQ1 that undergoes cleavage may be compromised in tobacco plants. To test this, we introduced plasmids expressing PphHopQ1, PtoHopQ1, and PtoHopQ1\_S87L\_L91<sup>R</sup> into Pss and measured bacteria-induced ion leakage that reflects loss of plasma membrane integrity in the course of the hypersensitive cell death (Krzymowska et al., 2007). As shown in **Figure 3A**, 10 h post infiltration (hpi), that is the time when first macroscopic signs of tissue collapse became visible, the conductivity reached the maximum level in response to Pss expressing PphHopQ1. This effect was delayed in response to PtoHopQ1-expressing bacteria. In this case, the first symptoms – vitrification when viewed from the abaxial leaf surfaces – were only visible 12 hpi and the maximum increase in conductivity was reached at 14 hpi. Interestingly, upon infiltration of Pss expressing PtoHopQ1\_S87L\_L91R, maximum conductivity was recorded already at 10 hpi but in contrast to PphHopQ1, this level remained elevated till the last time point (16 hpi). These data suggest that the aa substitutions that reduce HopQ1 susceptibility to proteolytic cleavage may restore early recognition of the effector in tobacco plants. We asked further, whether the changed HopQ1 perception affects virulence of Pss in this host plant. Therefore, we monitored multiplication of the Pss strains expressing HopQ1 variants. As shown in **Figure 3B**, expression of all HopQ1, variants reduced the ability of Pss to grow in tobacco leaves compared to the mCherry-expressing control strain. This effect, however, was less pronounced with Pss expressing PtoHopQ1. This suggests that the delayed HR onset leads to an enhanced bacterial growth at the early stages of the infection and, thus, synthesis of the cleavable form of HopQ1 seems to be beneficial in the resistant host. Consistent with this model, bacteria expressing PtoHopQ1\_S87L\_L91<sup>R</sup> that is less prone to processing multiplied to intermediate levels.

Besides individual functions, effector proteins may act in concert within host cells and this is particularly likely for effectors that are sequentially delivered by the type III secretion system (Büttner, 2016). HopQ1 is grouped together with HopR1 in the same gene cluster in Pto (Kvitko et al., 2009). String database analysis<sup>1</sup> revealed a significant co-occurrence of hopQ1 with hopR1. These findings suggested that these two effectors might act co-operatively in plant cells. To test this hypothesis, we prepared vectors carrying pseudo-operons of PtohopQ1 or PphhopQ1 with PtohopR1 or PphhopR1 under control of a constitutive version of Tac promoter (**Figure 4A**). We introduced pseudooperons carrying the sequences coding for the effector pairs into a Pss strain virulent on N. benthamiana. Subsequently, we scored disease symptoms and determined bacterial growth upon dip-inoculation of N. benthamiana plants (10<sup>6</sup> cfu/ml; **Figure 4B**). Compared to Pss expressing PphHopQ1 alone (**Figure 1A**), Pss expressing both PphHopQ1 and PphHopR1

FIGURE 4 | HopR1 masks HopQ1-mediated recognition of P. syringae. (A) A schematic representation of pseudo-operons that co-express HopQ1 and HopR1 from P. syringae pv. phaseolicola 1448A or tomato DC3000. 'H' and 'F' stands for His and FLAG tag, respectively. (B) N. benthamiana plants were dip-inoculated with Pss expressing the pseudo-operons. At the indicated time points bacteria were isolated from leaf tissue and serial dilutions were plated on LB agar plates. Data were analyzed using repeated measures ANOVA, followed by Tukey HSD post hoc test performed for each time point separately. Letters correspond to statistically homogenous groups (p < 0.05). The photographs were taken 7 days after inoculation. (C) PtoDC3000D28E strain expressing the indicated combinations of HopQ1 or HopR1 were locally infiltrated into N. tabacum leaves. Necrosis development was observed already 24 h later and the photographs were taken 5 days after infiltration. The experiment was performed twice with similar results.

<sup>1</sup>http://string-db.org/

with Agrobacterium tumefaciens strains carrying constructs encoding the HopR1 variants fused to eYFP. The images were recorded by confocal microscopy 72 h

induced strong disease symptoms and multiplied to high levels (**Figure 4B**). The operon with PphhopQ1 and PtohopR1 rendered Pss less virulent, similar to bacteria expressing PphHopQ1 alone (**Figure 1A**). Inoculation with Pss expressing PtoHopQ1 led to blight disease symptom development no matter which HopR1 variant was co-expressed. In both PtoHopQ1 combinations, however, the macroscopic symptoms were less pronounced than upon infection with bacteria expressing PphHopQ1 along with PphHopR1. As bacterial titers did not perfectly reflect the disease symptoms induced by the different effector combinations, additional mechanisms are involved. Nevertheless, the findings are indicative of interplay between HopQ1 and HopR1 when delivered into plant cells by Pss.

after agroinfiltration. DIC, differential interference contrast; bars = 10 µm.

To reduce additional effects of other bacterial effectors present in Pss, we used PtoDC3000D28E, a mutant strain of P. syringae pv. tomato DC3000 with 28 effector genes deleted (Cunnac et al., 2011), to specifically deliver various combinations of HopQ1 and HopR1. Importantly, this strain expresses HopAD1, that along with HopQ1 is required to trigger HR and as a consequence of single deletion of HopQ1 or HopAD1 PtoDC3000 gains virulence toward N. benthamiana (Wei et al., 2015). The transformed PtoDC3000D28E strains were infiltrated into leaves of N. tabacum plants to address whether they differ in their ability to induce hypersensitive cell death. None of the HopR1 variants triggered HR (**Figure 4C**), suggesting that HopR1 is not recognized in tobacco. In contrast, both HopQ1 variants expressed separately induced HR whereas the presence of HopR1 from the same P. syringae strain completely abolished this response. Interestingly, both combinations of HopQ1 and HopR1 derived from two different strains elicited HR. Furthermore, this response was stronger than triggered by HopQ1 variants alone. The fact that HopR1 expressed by bacteria infiltrated at very high inoculum (6 × 10<sup>8</sup> CFU/ml) into N. benthamiana leaves triggers HR (Wei et al., 2018) suggests a possibility that HopR1 evokes cell death response also in N. tabacum and, thus, observed enhancement of tissue collapse (**Figure 4C**) would be due to synergistic/additive effect of HopQ1 and HopR1 action. The similarity between PphHopR1 and PtoHopR1 is 86%, whereas it is 98% between PphHopQ1 and PtoHopQ1. Collectively, these data suggest that the effector pairs co-evolved within a single strain and due to evolutionary diversification fail to co-operate when transferred individually from one strain to the other. This resembles a phenomenon described by Wei et al. (2015) when members of HopAB family displayed various abilities to suppress HopAD1 dependent cell death.

This strain-specificity of HopR1 in blocking HR mediated by HopQ1 suggests that both effectors directly interact rather that HopR1 interfering with the signaling pathway initiated upon HopQ1 recognition. This model is, unfortunately, not consistent with the previous reports that HopQ1 is predominantly cytoplasmic (Giska et al., 2013; Li et al., 2013b), whereas HopR1 was shown to be imported into isolated chloroplasts (de Torres Zabala et al., 2015). However, HopR1 was detected both in chloroplasts and the cytoplasm when transiently expressed in N. benthamiana (**Figure 5**) and, thus, their association in the cytoplasm might still occur. The fact that in the native PtoDC3000 strain HopR1 is not able to block HopQ1-triggered

cell death seems to be contradictory to our results. However, a similar case has been described for HopQ1 and HopI1 (Wei et al., 2018). Here, HopI1 was shown to block HopQ1 recognition in N. benthamiana. Although, PtoDC3000 that secretes both effectors is avirulent in this plant. In general, the (genetic) interactions between effectors are still poorly understood and, in this context, the mechanism of how HopR1 interferes with HopQ1 signaling requires further investigation.

Our data demonstrate that specific amino acid residues of PphHopQ1 and PtoHopQ1 determine the disease outcome in N. benthamiana and N. tabacum. Sequence comparison showed that only six aa differ in the HopQ1 homologs studied. Two of these aa substantially affected HopQ1 properties. The presence of serine at position 87 and leucine at position 91 correlated with the susceptibility of the effector to the proteolytic cleavage within plant cells and debilitated effector recognition. Considering co-evolutionary adaptations, P. syringae would directly profit from HopQ1 cleavage, since HopQ1 recognition is avoided, even if it partially reduces its virulence properties in a susceptible host (**Figure 1**). The reduced virulence of the truncated form most likely results from the loss of interaction with 14-3-3 proteins in the host cell, since the HopQ1 N-terminus carries a canonical 14-3-3 binding site (RSXpSXP; pS indicates phosphoserine) that is important for proper effector localization and stability (Giska et al., 2013; Li et al., 2013b). From the "plant's perspective," cleavage would block the function of a single effector but would reduce sensing of the bacteria and, thereby, lead to disease development. Thus, we hypothesize that simultaneous maintenance of HopQ1 in the intact and truncated forms reflects a "calculated risk strategy" of P. syringae. In a susceptible plant, the virulence properties of the intact form sustain disease and the slightly reduced virulence properties of the truncated form still support bacterial proliferation. In a resistant plant, the N-terminally truncated form avoids recognition (Li et al., 2013a) and mediates suppression of a proper defense response. However, we cannot exclude that the cleavage of HopQ1 had been primarily a manifestation of the plant response that was later "corrupted" by Pseudomonas.

It was previously inferred from multilocus sequence typing that P. syringae pv. tabaci eliminated the sequence encoding HopQ1 from its genome to avoid detection (Ferrante et al., 2009). Our data suggest that besides this known mechanism, P. syringae may have evolved other strategies to prevent recognition. HopQ1 from P. syringae pv. phaseolicola (PphHopQ1) co-evolved

# REFERENCES


with HopR1 that masks its presence, pointing to tight coadaptation in P. syringae pv. phaseolicola. In contrast, HopR1 from pathovar tomato was not able to block HR triggered by PphHopQ1. Interestingly, N. benthamiana plants inoculated with Pss expressing PtoHopQ1 displayed a different phenotype than those plants inoculated with the strain expressing PphHopQ1 (**Figure 1**). While introduction of PphhopQ1 rendered bacteria less virulent, PtohopQ1 compromised virulence of Pss to a lesser extent (**Figure 1**) which we hypothesize is linked to cleavage of PtoHopQ1 (**Figure 3**). Collectively, previous reports and our current results suggest that in order to avoid recognition of HopQ1, P. syringae evolved three different strategies that rely on (i) loss of the effector encoding sequences from its genome (Ferrante et al., 2009), (ii) partial susceptibility of the effector variants to proteolytic cleavage, and (iii) masking of the effector recognition by the co-adapted HopR1.

# AUTHOR CONTRIBUTIONS

RH, LE-L, JL, JH, and MK conceived and designed the experiments. PZ, AD, RH, LE-L, MB, MG-B, IB-F, KM, JK, and MP performed the experiments. PZ, AD, RH, LE-L, MG-B, WR, JP, JL, JH, and MK analyzed the data. MK, LE-L, and JL wrote the paper.

# FUNDING

This work was supported by grant no. 2013/11/B/NZ9/01970 from the National Science Centre (to MK). RH was supported by a Short Term Scientific Mission grant from the COST FA 1208 program (http://www.cost-sustain.org). LE-L was currently supported by the German Research Foundation grant LE2321/3-1 to JL. The equipment used was sponsored in part by the Centre for Preclinical Research and Technology (CePT), a project co-sponsored by European Regional Development Fund and Innovative Economy, The National Cohesion Strategy of Poland.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00978/ full#supplementary-material

Cunnac, S., Chakravarthy, S., Kvitko, B. H., Russell, A. B., Martin, G. B., and Collmer, A. (2011). Genetic disassembly and combinatorial reassembly identify a minimal functional repertoire of type III effectors in Pseudomonas syringae. Proc. Natl. Acad. Sci. U.S.A. 108, 2975–2980. doi: 10.1073/pnas.1013031108



**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 Zembek, Danilecka, Hoser, Eschen-Lippold, Benicka, Grech-Baran, Rymaszewski, Barymow-Filoniuk, Morgiewicz, Kwiatkowski, Piechocki, Poznanski, Lee, Hennig and Krzymowska. 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.

# The Arabidopsis Elongator Subunit ELP3 and ELP4 Confer Resistance to Bacterial Speck in Tomato

Juliana A. Pereira<sup>1</sup> , Fahong Yu<sup>2</sup> , Yanping Zhang<sup>2</sup> , Jeffrey B. Jones<sup>1</sup> \* and Zhonglin Mou<sup>3</sup> \*

<sup>1</sup> Department of Plant Pathology, University of Florida, Gainesville, FL, United States, <sup>2</sup> Interdisciplinary Center for Biotechnology Research, University of Florida, Gainesville, FL, United States, <sup>3</sup> Department of Microbiology and Cell Science, University of Florida, Gainesville, FL, United States

Although production of tomato (Solanum lycopersicum) is threatened by a number of major diseases worldwide, it has been difficult to identify effective and durable management measures against these diseases. In this study, we attempted to improve tomato disease resistance by transgenic overexpression of genes encoding the Arabidopsis thaliana Elongator (AtELP) complex subunits AtELP3 and AtELP4. We show that overexpression of AtELP3 and AtELP4 significantly enhanced resistance to tomato bacterial speck caused by the Pseudomonas syringae pv. tomato strain J4 (Pst J4) without clear detrimental effects on plant growth and development. Interestingly, the transgenic plants exhibited resistance to Pst J4 only when inoculated through foliar sprays but not through infiltration into the leaf apoplast. Although this result suggested possible involvement of stomatal immunity, we found that Pst J4 inoculation did not induce stomatal closure and there were no differences in stomatal apertures and conductance between the transgenic and control plants. Further RNA sequencing and real-time quantitative PCR analyses revealed a group of defense-related genes to be induced to higher levels after infection in the AtELP4 transgenic tomato plants than in the control, suggesting that the enhanced disease resistance of the transgenic plants may be attributed to elevated induction of defense responses. Additionally, we show that the tomato genome contains single-copy genes encoding all six Elongator subunits (SlELPs), which share high identities with the AtELP proteins, and that SlELP3 and SlELP4 complemented the Arabidopsis Atelp3 and Atelp4 mutants, respectively, indicating that the function of tomato Elongator is probably conserved. Taken together, our results not only shed new light on the tomato Elongator complex, but also revealed potential candidate genes for engineering disease resistance in tomato.

Keywords: tomato, the Elongator complex, AtELP4, transgenic overexpression, disease resistance, Pseudomonas syringae

# INTRODUCTION

Tomato (Solanum lycopersicum) fruit was once thought to be poisonous, but since being integrated as part of the human diet, its popularity and consumption have increased over the years. Tomato production has an economic impact worldwide, but it is also a costly crop to produce. It is a labor-intensive crop that requires significant amount of chemical inputs to be protected from a

#### Edited by:

Zhengqing Fu, University of South Carolina, United States

#### Reviewed by:

Yuxin Hu, Institute of Botany (CAS), China Wen-Ming Wang, Sichuan Agricultural University, China

> \*Correspondence: Jeffrey B. Jones jbjones@ufl.edu Zhonglin Mou zhlmou@ufl.edu

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 09 May 2018 Accepted: 29 June 2018 Published: 24 July 2018

#### Citation:

Pereira JA, Yu F, Zhang Y, Jones JB and Mou Z (2018) The Arabidopsis Elongator Subunit ELP3 and ELP4 Confer Resistance to Bacterial Speck in Tomato. Front. Plant Sci. 9:1066. doi: 10.3389/fpls.2018.01066

wide variety of pests and diseases. There are a number of diseases that affect tomatoes, including bacterial speck, which is caused by the bacterial pathogen Pseudomonas syringae pv. tomato. Bacterial speck can cause up to 75% losses in yield, if present early in the production cycle (Yunis et al., 1980). Control of the disease is primarily based on application of bactericides and sanitary measures. Pathogen-free seeds and resistant varieties carrying the resistance (R) gene Pto have been implemented to control the disease (Monroe and Sasser, 1980; Pedley and Martin, 2003). However, P. syringae pv. tomato strains have evolved to overcome the R gene-mediated resistance in tomato (Thapa and Coaker, 2016).

Since tomatoes are susceptible to many diseases, studies involving identification of disease resistance-related genes in model plants have increased dramatically (Piquerez et al., 2014). Currently, one strategy that is being pursued is to utilize resistance-related genes identified in Arabidopsis and their orthologs in other plant species (Jones et al., 2014). Arabidopsis is a well-established model system, with the complete genome sequenced.<sup>1</sup> Furthermore, multiple Arabidopsis genes have been cloned, characterized, and reported to confer resistance to diseases when overexpressed in diverse crop species (Lin et al., 2004; Chan et al., 2005; Lacombe et al., 2010; Schwessinger et al., 2015; Silva et al., 2017), making Arabidopsis a suitable source of defense-related genes for engineering resistance in tomato.

The Elongator protein (ELP) complex is a highly conserved multitasking protein complex in eukaryotes (Otero et al., 1999; Wittschieben et al., 1999; Hawkes et al., 2002; Nelissen et al., 2010; Woloszynska et al., 2016). It consists of six subunits, including ELP1 and ELP2 (scaffolds for complex assembly), ELP3 (catalytic subunit), and an accessory complex formed by ELP4– ELP6 (Svejstrup, 2007). Elongator has been shown to be involved in several distinct cellular processes, such as exocytosis, histone modification, tRNA modification, α-tubulin acetylation, zygotic paternal DNA demethylation, and miRNA biogenesis (Hawkes et al., 2002; Huang et al., 2005; Rahl et al., 2005; Creppe et al., 2009; Okada et al., 2010; Ding and Mou, 2015; Fang et al., 2015). It has been clearly demonstrated that Elongator functions in both the nucleus and the cytoplasm (Versées et al., 2010). In the nucleus, Elongator regulates histone acetylation and DNA methylation/demethylation (Winkler et al., 2002; Lin et al., 2012; Wang et al., 2013), thus being involved in gene transcription. In the cytoplasm, it is responsible for tRNA modification, which consequently regulates protein translation (Huang et al., 2005; Esberg et al., 2006; Glatt et al., 2012).

It has been well documented that the A. thaliana Elongator protein (AtELP) complex plays an important role in plant immunity, likely by regulating the transcription of defense genes (Ding and Mou, 2015; Wang et al., 2015). However, whether Elongator has a similar role in plant species other than Arabidopsis remains to be determined. Although it has been reported that silencing of a tomato AtELP2-like gene, SlELP2L, resulted in pleiotropic phenotypes similar to those of the Atelp mutants, defense phenotypes of the SlELP2L-RNAi lines were not tested (Zhu et al., 2015). Furthermore, some of the phenotypes

<sup>1</sup>www.arabidopsis.org

displayed by the SlELP2L-RNAi lines appear to be different from those of the Atelp mutants. For instance, while ethylene signaling and auxin levels are elevated in the Atelp mutants, both are reduced in the SlELP2L-RNAi lines (Nelissen et al., 2010; Zhu et al., 2015). These differences suggest that the function of Elongator in tomato might not be exactly the same as that in Arabidopsis. Further characterization of genes encoding the Elongator subunits in tomato will not only help in understanding the function of Elongator in plants, but may also identify new strategies for improving disease resistance in tomato.

In this study, we characterized transgenic tomato plants expressing the Arabidopsis AtELP3 and AtELP4 genes. We show that overexpression of AtELP3 and AtELP4 significantly enhanced resistance to tomato bacterial speck caused by the P. syringae pv. tomato strain J4 (Pst J4) without clear detrimental effects on plant growth and development. Interestingly, the enhanced resistance was detected only when plants were inoculated via foliar sprays of bacterial suspensions but not infiltration into the apoplast, suggesting possible involvement of stomatal immunity. However, Pst J4 inoculation did not induce stomatal closure and there were no differences in stomatal apertures and conductance between the transgenic and control plants, indicating that a defense mechanism other than stomatal immunity was activated in the transgenic plants. Indeed, further RNA sequencing (RNA-seq) revealed a group of defense-related genes that were confirmed by real-time quantitative PCR (qPCR) analysis to be induced to higher levels after infection in the AtELP4 transgenic tomato plants than in the control, suggesting that the enhanced disease resistance of the transgenic plants may be attributed to elevated induction of defense responses. Additionally, we show that the tomato genome encodes all six Elongator subunits (SlELPs) and that the tomato SlELP3 and SLELP4 genes complemented the Arabidopsis Atelp3 and Atelp4 mutants, respectively. Thus, the tomato Elongator is most likely functional and AtELP3, AtELP4 as well as their tomato orthologs SlELP3 and SlELP4 could potentially be employed in the improvement of disease resistance in tomato plants.

# MATERIALS AND METHODS

# Plasmid Construction and Plant Transformation

The T-DNA plasmids (pK7WG2D, 1-AtELP3 and pK7WG2D, 1-AtELP4) reported previously (Silva et al., 2017) were used to transform the tomato cultivar "Moneymaker" following an Agrobacterium tumefaciens-mediated genetic transformation protocol (Lin et al., 2004). The tomato genetic transformation experiment was conducted by the UNL Plant Transformation Facility<sup>2</sup> . For complementation of the Arabidopsis Atelp3 and Atelp4 mutants, the coding regions of the tomato orthologs (SlELP3 and SlELP4) were amplified from "Moneymaker" cDNAs by PCR using gene specific primers (Supplementary Table S1) and cloned into the binary vector pBI1.4T (Mindrinos et al., 1994). The resulting plasmids were introduced into the A. tumefaciens

<sup>2</sup>https://biotech.unl.edu/plant-transformation

strain GV3101(pMP90) by electroporation (Shen and Forde, 1989). The Arabidopsis Atelp3-10 and Atelp4/elo1-1 mutant alleles (Nelissen et al., 2005; Defraia et al., 2010), which are in the Columbia (Col-0) and Landsberg erecta ecotype backgrounds, respectively, were used for A. tumefaciens-mediated genetic transformation following the floral dip method (Clough and Bent, 1998).

# Identification of Single T-DNA Insertion Homozygous Transgenic Lines

The T<sup>1</sup> transgenic tomato plants obtained from the UNL Plant Transformation Facility were allowed to set seeds. The T<sup>2</sup> plants were subjected to PCR analysis using gene-specific primers (Supplementary Table S1) to analyze T-DNA insertion copy numbers based on the expected ratio of 3:1 for a single T-DNA insertion. The transgenic lines that showed the expected ratio for a single T-DNA insertion were kept and seeds from the individual T<sup>2</sup> plants were collected separately. The T<sup>3</sup> progeny plants from each individual T<sup>2</sup> plants were subjected to PCR analysis to identify homozygous plants for each transgenic line. Seeds from the homozygous T<sup>3</sup> plants were pooled for further analysis. For Arabidopsis transgenic lines, T<sup>2</sup> seeds from individual T<sup>1</sup> plants were plated on Murashige and Skoog (MS) medium with 50 µg/mL kanamycin to identify single T-DNA insertion lines based on the segregation ratio of the neomycin phosphotransferase II (nptII) gene. Homozygous plants were similarly identified in the T<sup>3</sup> generation.

# Pathogen Infection and Bacterial Population Assay

To evaluate the resistance of the transgenic tomato plants to bacterial speck, a bacterial suspension of Pst J4, adjusted to 1 × 10<sup>8</sup> colony-forming units (cfu)/mL, was sprayed on 4 week-old tomato plants in pots with a diameter of 10 cm. This inoculum is able to induce consistent levels of disease severity on tomato plants (Kozik and Sobiczewski, 2000). The plants were then immediately covered with bags and a rubber band was placed around the base of the pot to seal the bag in order to maintain high humidity for 40 h. A total of six plants per line were tested and non-transformed "Moneymaker" was included as the control. Inoculated plants were incubated in the growth chamber and maintained at 22◦C under a regimen of 12 h dark and 12 h light. The disease symptoms were evaluated 6 days postinoculation. The disease assessment consisted of the following disease scores: 0 indicates no symptom development; 1 indicates few slightly visible lesions; 2 indicates a significant number of discernible lesions; 3 indicates a higher amount of discernible necrotic and chlorotic lesions; and 4 indicates extensive necrotic and chlorotic lesions and extensive dead tissue.

For quantifying bacterial populations of Pst J4 in the inoculated tomato plants, leaf tissues were sampled every 6 days. Three leaf disks with an area of 1 cm<sup>2</sup> were obtained from each transgenic line using a cork borer. The leaf disks were placed into glass tubes and ground in 1 mL of sterile water. The resulting suspensions were diluted by making five 10-fold dilutions. The dilutions were plated on nutrient agar medium and then the plates were incubated at 28◦C for 2 days. Colonies typical of P. syringae pv. tomato were counted and the bacterial number per cm<sup>2</sup> of leaf tissue was calculated.

For testing growth of the bacterial pathogen Psm ES4326 in Arabidopsis, leaves were infiltrated with a suspension of Psm ES4326 (OD<sup>600</sup> = 0.0001) using a 1 mL needleless syringe as described previously (Defraia et al., 2010). Leaf disks were collected from eight leaves 3 days post-inoculation using a cork borer. Each leaf disk was placed into a tube containing 500 µL of 10 mM MgCl<sup>2</sup> and ground with a sterile pellet pestle. The resulting suspensions were serially diluted 20-fold four times. The dilutions were then plated on Trypticase Soy Agar medium supplemented with 25 µg/mL streptomycin and incubated at 28◦C for 2 days. Colonies that grew on the plates were counted and the bacterial number per leaf disk tissue was calculated.

# Stomatal Conductance Measurement

Tomato plants were sprayed with a Pst J4 bacterial suspension (10<sup>8</sup> cfu/mL). Stomatal conductance was measured before the inoculation (time 0) and every 30 min post-inoculation using a portable photosynthesis system (LI-6800, LI-COR Biosciences<sup>3</sup> ). The principle of the measurement is that the time required to force a certain volume of air through the plant leaf is inversely proportional to leaf stomatal conductance (Rebetzke et al., 2000). Ten fully expanded leaves per plant were used for the measurement at each time point, and the readings from the abaxial side of the leaves were recorded.

# RNA Sequencing and Real-Time Quantitative PCR Analysis

Tomato plants were sprayed with a Pst J4 suspension (10<sup>8</sup> cfu/mL). Three replicates of leaf tissues from six plants per genotype were collected at 0, 8, and 24 h post-inoculation. Total RNA was extracted from the collected leaf tissues using the RNeasy plant mini kit following the manufacturer's protocol (Qiagen<sup>4</sup> ). RNA concentration and quality were determined using a Qubit 2.0 Fluorometer (ThermoFisher<sup>5</sup> ) and an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.<sup>6</sup> ), respectively. Total RNA samples with 28S/18S >1 and RNA integrity number ≥7 were used for RNA-seq analysis. The RNA samples from the three biological replicates were pooled and equal amounts of RNA from the pooled samples were used for RNA-seq library preparation. Briefly, 1 µg of total RNA together with 2 µL of 1:200 diluted ERCC (External RNA Controls Consortium) RNA spike-in mix was used for mRNA extraction with 15 µL of NEBNext Magnetic Oligo d(T)<sup>25</sup> and fragmented in NEBNext First Strand Synthesis Buffer by heating at 94◦C for 8 min, then followed by first strand cDNA synthesis using reverse transcriptase and random primers. Synthesis of double-stranded cDNA was done using the second strand master mix provided in the kit. The resulting double-stranded cDNA was subjected to end-repair and dA-tailing and then ligated with NEBNext

<sup>3</sup>https://www.licor.com

<sup>4</sup>https://www.qiagen.com/

<sup>5</sup>https://www.thermofisher.com/

<sup>6</sup>https://www.agilent.com/

adaptors. Finally, the library was enriched by PCR amplification and purified by Agencourt AMPure beads (Beckman Coulter<sup>7</sup> ). Barcoded libraries were sized and quantitated. qPCR was used to validate the library's functionality, using the KAPA library quantification kit (Kapa Biosystems<sup>8</sup> ). The six individual samples were pooled equimolarly for one lane of HiSeq 3000 2 × 100 cycles run. Sequencing was performed on the Illumina HiSeq 3000 instrument at the University of Florida Interdisciplinary Center for Biotechnology (UF ICBR) NextGen DNA Sequencing core. The reads that passed Illumina quality control filtering were cleaned up with the Cutadapt program (Martin, 2011) to trim off sequencing adaptors and low quality bases with a quality Phred-like score <20. Reads <40 bases were excluded from RNA-seq analysis. The transcripts of tomato from National Center for Biotechnology Information (NCBI) were used as reference sequences for RNA-seq analysis. The cleaned reads of each sample were mapped independently to the reference sequences using the mapper of bowtie2 with a maximum of three mismatches for each read. The mapping results were processed with the samtools and scripts developed in house at the UF ICBR to remove potential PCR duplicates and select uniquely mapping reads for gene expression estimation. The number of mapped reads for each individual gene was counted. Comparison was made between the AtELP4 transgenic line 61-5 and the control samples collected at the same time point.

For qPCR analysis, total RNA was extracted from tomato plants using Trizol reagent (Thermo Fisher Scientific) and treated with RNAse-free DNAse I (Thermo Fisher Scientific). First strand complementary DNA was synthetized using 10 µg of total RNA with oligo (dT) primer and Moloney murine leukemia virus reverse transcriptase (Thermo Fisher Scientific). Gene-specific primers used for qPCR analysis were listed in Supplementary Table S1. qPCR was performed using ABsolute SYBR Green PCR master mix (Thermo Fisher Scientific) using the SYBR Green protocol (Applied Biosystems<sup>9</sup> ). Reactions were run and analyzed on a MX3000P qPCR system (Agilent<sup>10</sup>). The relative mRNA levels of the target genes were expressed relative to the tomato Actin gene (Zhou et al., 2014a,b), and calculated using the 1C<sup>T</sup> method (Wittwer et al., 2001).

# Statistical Analysis

Statistical analyses were performed using one-way ANOVA followed by a Tukey's multiple comparisons test in Prism 7 (GraphPad Software<sup>11</sup>).

# Accession Number

Sequence data from this article can be found in the Arabidopsis Genome Initiative, the Tomato Genome Sequencing Project, or GenBank/EMBL databases under the following accession numbers: AtELP3 (At5g50320); AtELP4 (At3g11220); PR1b1 (Y08804.1); PR-5x (AY093595); DES (AF317515); ER1 (J04099.1); SlELP1 (Solyc05g054630); SlELP2 (Solyc06g008310); SlELP3 (Solyc03g110910); SlELP4 (Solyc11g010950); SlELP5 (Solyc02g086100); SlELP6 (Solyc12g009500); and NCBI Gene Expression Omnibus Series number GSE97697 (RNA-seq data).

# RESULTS

# Generation and Morphological Characterization of Transgenic Tomato Lines Overexpressing AtELP3 and AtELP4

Based on PCR amplification of the cDNA of AtELP3 or AtELP4 using gene specific primers (Supplementary Table S1), out of 80 T<sup>1</sup> putative transgenic plants produced by the University of Nebraska–Lincoln (UNL) Plant Transformation Facility, 36 carried the AtELP3 transgene and 35 the AtELP4 gene. Note that the PCR reactions did not amplify any products from the non-transformed "Moneymaker" plants, which were used as the negative control in the experiment, demonstrating the specificity of the primers. The transgenic plants with T-DNA insertion were kept for seed production. Single T-DNA insertion lines were identified in the T<sup>2</sup> generation based on the 3:1 segregation ratio expected for a single T-DNA insertion, and homozygosity of the single insertion lines was determined in the T<sup>3</sup> generation. In total, five and four single insertion homozygous lines were identified for AtELP3 and AtELP4, respectively. Expression levels of AtELP3 and AtELP4 in the single insertion homozygous lines were determined in the T<sup>4</sup> generation by qPCR using genespecific primers (Supplementary Table S1). While AtELP3 and AtELP4 transcripts were barely detectable in the control plants, the transgenes were expressed at varied levels in the different lines (**Figures 1A,B**). For AtELP3, expression levels of the transgene were significantly higher in lines 56-9, 60-5, and 51-9 than in lines 51-2, 44-2, and the control plants (**Figure 1A**). For AtELP4, the transgenic lines could clearly be classified into three groups: one high expresser (line 37-3), two medium expressers (lines 23-1 and 61-5), and one low expresser (line 28-6) (**Figure 1B**).

The overall morphology and development of the transgenic tomato plants were very similar to those of the control plants under standard greenhouse conditions (**Figure 1C**). There were no significant differences in plant height between the transgenic lines and the control (**Figure 1D**). Furthermore, all of the transgenic lines formed flowers and fruits. The fruit weight of all the transgenic lines except 44-2 was not significantly different from that of the control (**Figure 1E**). The fruit from the transgenic line 44-2 was very small and similar to that produced by cherry tomato varieties. The small-fruit phenotype of line 44-2 was unlikely caused by overexpression of AtELP3, since other lines (56-9, 60-5, and 51-9) that expressed higher levels of AtELP3 than line 44-2 did not show such a phenotype. It might be possible that the small-fruit phenotype was caused by a T-DNA insertion mutation. Alternatively, there might be seed contamination during the development of transgenic plants. Nevertheless, these results indicate that transgenic overexpression of AtELP3 and AtELP4 does not affect tomato plant growth and development.

<sup>7</sup>https://www.beckman.com

<sup>8</sup>https://www.kapabiosystems.com/

<sup>9</sup>http://www.appliedbiosystems.com/

<sup>10</sup>https://www.genomics.agilent.com/

<sup>11</sup>https://graphpad.com

control. Photos were taken 30 days after germination. (D) Plant height of the transgenic lines and the control. Data represent the average of six plants with SD. Different letters above the bars indicate significant differences (Tukey's test, P < 0.05). (E) Fruit weight of the transgenic lines and the control. Data represent the average weight of fruit from six plants with SD. Different letters above the bars indicate significant differences (Tukey's test, P < 0.05).

# Disease Resistance of the Transgenic Tomato Lines

To test whether transgenic overexpression of AtELP3 or AtELP4 in tomato improves disease resistance, we inoculated the single insertion homozygous transgenic lines with the bacterial pathogen Pst J4, which causes bacterial speck on tomato plants. Both leaf infiltration and foliar sprays were employed in the experiment, since the transgenic plants might respond differently to these two commonly used inoculation methods. The bacterial speck disease symptoms, characterized by small, black, or brown necrotic lesions surrounded by a chlorotic halo, appeared 3 days post-inoculation on the transgenic plants and the control for both inoculation methods. When the plants were inoculated using the leaf infiltration method, no significant differences were observed between the bacterial speck disease symptoms developed on any of the transgenic lines and the control, and Pst J4 grew to similar levels in all the tested plants (Supplementary Figure S1). In contrast, when foliar sprays were used, the disease symptoms

on different transgenic lines and the control differed drastically. To quantify the disease symptoms, different disease scores were assigned to the transgenic lines based on the disease severity on the leaves (**Figures 2A,B**). The disease symptoms on the AtELP3 transgenic lines 44-2 and 60-5 were similar to those on the control, the disease symptoms on the AtELP3 transgenic lines 51-2, 56-9, and 51-9 as well as the AtELP4 transgenic lines 28- 6 and 37-3 were slightly less severe than those on the control, and the disease symptoms on the AtELP4 transgenic lines 23- 1 and 61-5 were much less severe than those on the control (**Figure 2B**). Interestingly, there was no clear correlation between the disease severity and the expression levels of the transgenes (**Figures 1A,B**, **2B**), which is not without precedent (Luhua et al., 2008). The transgenic line 61-5, which was a medium expresser of AtELP4 (**Figure 1B**), exhibited the strongest resistance to Pst J4 (**Figure 2**). The bacterial speck disease progression on the transgenic line 61-5 was markedly slower than that on the control plants (**Figure 2C**). At 3 days post-inoculation, leaves on the control had already wilted, whereas those on the transgenic line 61-5 still stayed uncurled (**Figure 2D**). We

also determined bacterial titers in the AtELP4 transgenic lines, since, based on disease symptoms, two independent AtELP4 transgenic lines (23-1 and 61-5) displayed clear resistance to Pst J4. Consistent with the observed disease symptoms, the bacterial titers in the transgenic lines 61-5 and 23-1 were the lowest and the second lowest, respectively, among the AtELP4 transgenic lines as well as the control (**Figure 2E**). These results indicate that overexpression of the AtELP4 gene in tomato is able to significantly improve resistance to Pst J4-caused bacterial speck disease.

# Stomatal Conductance of the Transgenic Tomato Line With Increased Disease Resistance

The different responses of the transgenic lines to leaf infiltration and foliar sprays suggested a possible involvement of stomatal immunity (Melotto et al., 2006). To test this possibility, we assessed stomatal morphology and movement after foliar sprays of Pst J4 under optical microscope. Interestingly, inoculation of tomato plants with Pst J4 did not induce stomatal closure and no appreciable differences in stomatal apertures between the most resistant transgenic line 65-1 and the control were observed (Supplementary Figure S2). We further measured stomatal conductance using a portable photosynthesis system. Overall, there were no clear differences in stomatal conductance between the transgenic line 61-5 and the control in a period of 4.5 h following foliar sprays of Pst J4 (**Figure 3**). Taken together, these results suggest that alteration of stomatal immunity may not be a predominant factor for the enhanced disease resistance observed in the AtELP4 transgenic plants.

# Induction of Defense Genes in the Transgenic Tomato Line With Increased Disease Resistance

To uncover potential mechanisms underlying the enhanced disease resistance of the transgenic tomato plants overexpressing AtELP4, we performed an RNA-seq experiment to compare Pst J4-induced transcriptome changes in the transgenic line 61-5 and the control, and then examined genes that were potentially induced to higher levels in the transgenic line 61-5 than in the control (NCBI Gene Expression Omnibus Series number GSE97697). Interestingly, a group of defense-related genes, including PATHOGENESIS-RELATED (PR) gene PR1b1, PR-5 family member PR-5x, DIVINYL ETHER SYNTHASE (DES), and ETHYLENE-RESPONSIVE (ER) PROTEASE INHIBITOR 1 (ER1) (GenBank accession numbers: Y08804.1, AY093595, AF317515, and J04099.1, respectively), which have previously been shown to be associated with disease resistance in tomato (Pautot et al., 1991; Ishihara et al., 2012), were potentially induced to higher levels in the transgenic line 61-5 than in the control. Since the RNA-seq experiment did not include biological replicates, statistical significance could not be evaluated. To confirm the RNA-seq results for the selected genes, we used qPCR to monitor the induction of PR1b1, PR-5x, DES, and ER1 in the transgenic line 61-5 and the control after Pst J4 infection. Consistent with the RNA-seq results, the induction of the four selected genes in the transgenic line 61-5 was significantly higher than that in the control plants (**Figure 4**). These results suggest that the enhanced disease resistance of the transgenic plants overexpressing AtELP4 may be due to increased induction of defense genes.

# Tomato Orthologs Encoding the Elongator Subunits

It is generally believed that the Elongator complex is highly conserved in eukaryotes. In agreement with this belief, Zhu et al. (2015) have shown that silencing of a tomato AtELP2-like gene, SlELP2L, resulted in pleiotropic phenotypes similar to those of Atelp mutants. Thus, the tomato genome should encode all SlELPs. To test this, BLAST searches were conducted on the tomato (S. lycopersicum) genome<sup>12</sup> using AtELP protein sequences as the queries. The results showed that each subunit of the Elongator complex is encoded by a single-copy gene in the tomato genome. Amino acid sequence alignments indicated that the SlELP proteins all share high identities (>53%) and similarities (>70%) with the AtELP proteins (Supplementary Figure S3). Particularly, SlELP3 has very high amino acid sequence identities (93%) and similarities (96%) with AtELP3, indicating that the Elongator complex catalytic subunits are highly conserved in tomato and Arabidopsis. Therefore, tomato probably has a functional Elongator complex.

# Complementation of the Atelp3 and Atelp4 Mutants With the Tomato Orthologs

To assess the functionality of the SlELP3 and SlELP4 proteins, the SlELP3 and SlELP4 open reading frames driven by the 35S promoter were introduced into the Atelp3 and Atelp4 mutants, respectively, via A. tumefaciens-mediated genetic transformation. Multiple single insertion homozygous lines were obtained for both SlELP3 and SlELP4. Morphologically, the Atelp3 and Atelp4

<sup>12</sup>https://solgenomics.net/organism/Solanum\_lycopersicum/genome

mutant plants have narrow leaves, long petioles, and shortened siliques (Nelissen et al., 2005). These morphological phenotypes all were completely restored to wild type in the transgenic Atelp3 and Atelp4 plants expressing SlELP3 and SlELP4, respectively (**Figures 5**, **6**). Thus, the functions of SlELP3 and SlELP4 are conserved.

To further confirm the morphological complementation results, we inoculated the wild-type Col-0, Atelp3, and two independent Atelp3 complementation lines, Com-1 and Com-2, with the bacterial pathogen P. syringae pv. maculicola (Psm) ES4326 to test whether the enhanced disease susceptibility phenotype of Atelp3 was also complemented by SlELP3 (Defraia et al., 2013). As shown in **Figure 7**, while the Atelp3 mutant was significantly more susceptible than the wild type to Psm ES4326, the growth of Psm ES4326 in the complementation lines was comparable to that in the wild type. This result indicates that the SlELP3 gene can also complement the defense phenotype of the Atelp3 mutant.

# DISCUSSION

Bacterial speck, caused by Pst, is an important disease that concerns tomato growers worldwide (Bashan et al., 1978; Devash et al., 1980; Smitley and McCarter, 1982). Because of the lack of an effective control for the disease (Smitley and McCarter, 1982; Ramos et al., 1989), we investigated the potential of Arabidopsis defense-related genes for improvement of disease resistance against Pst J4 in tomato. The Arabidopsis genes encoding the Elongator subunits were chosen as the candidates for tomato transformation, as their effectiveness in mediating resistance to bacterial diseases has previously been demonstrated in Arabidopsis (Defraia et al., 2010, 2013; Wang et al., 2015; An et al., 2017) and strawberry (Silva et al., 2017). Additionally, given that Elongator functions at the chromatin level and is not directly involved in specific recognition of pathogens (Wang et al., 2013), the possibility of the pathogen's ability to overcome the resistance is remote, which is critical for generating durable resistance in crop plants.

The aim of this work was to investigate the disease resistance of single insertion homozygous AtELP3 and AtELP4 transgenic tomato plants under growth chamber conditions. In total, we identified nine single insertion homozygous transgenic lines (**Figures 1A,B**). Although Arabidopsis Elongator mutants display striking morphological phenotypes (Nelissen et al., 2005), overexpression of AtELP3 and AtELP4 in tomato did not cause any abnormality. All transgenic lines except line 44-2 displayed morphological and developmental traits similar to those of the control (**Figures 1C–E**). The observed small-fruit phenotype of line 44-2 is not associated with the expression of the transgene AtELP3 (**Figure 1A**), and may thus be caused by a T-DNA insertion mutation or seed contamination during transgenic plant development.

FIGURE 5 | Complementation of the Arabidopsis Atelp3 mutant by SlELP3. Morphological phenotypes of Arabidopsis wild type, Atelp3, and two independent complementation lines (35S::SlELP3 Atelp3): Atelp3 Com-1 and Atelp3 Com-2. (Top) 3-weak-old plants; (Middle) 6-week-old plants; (Bottom) siliques.

FIGURE 6 | Complementation of the Arabidopsis Atelp4 mutant by SlELP4. Morphological phenotypes of Arabidopsis wild type, Atelp4, and two independent complementation lines (35S::SlELP4 Atelp4): Atelp4 Com-1 and Atelp4 Com-2. (Top) 3-weak-old plants; (Middle) 6-week-old plants; (Bottom) siliques.

significant differences (Tukey's test, P < 0.05).

Overexpression of AtELP3 and AtELP4 both improved tomato resistance to bacterial speck, which is in agreement with the result reported in strawberry (Silva et al., 2017). However, the improvement conferred by AtELP3 was marginal (**Figure 2B**). This result suggests that AtELP3 may not be effective in improving disease resistance in tomato or that the AtELP3 protein levels accumulated in the five AtELP3 transgenic lines used in this study may not be sufficient for activation of strong disease resistance. On the other hand, two AtELP4 transgenic lines (23-1 and 61-5) displayed considerable enhanced resistance to bacterial speck under growth chamber conditions. The bacterial speck disease symptoms on both lines were dramatically alleviated (**Figures 2B–D**). Consistently, the bacterial populations in these transgenic lines were significantly lower over an 18-day period than those in the control (**Figure 2E**). We noticed that the enhanced disease resistance was not tightly correlated with the expression levels of the transgene, which is a common phenomenon in transgenic studies (Luhua et al., 2008). Unfortunately, no anti-AtELP4 antibody is available to determine the protein levels in the transgenic lines. Nevertheless, our results indicate that overexpression of AtELP4 in tomato is able to significantly enhance resistance to Pst J4.

Interestingly, increased disease resistance was not observed when the bacterial suspension was infiltrated into the apoplast of the transgenic tomato plants (Supplementary Figure S1). This suggested that AtELP3 and AtELP4 might improve stomatal immunity in tomato. We therefore investigated if stomatal morphology and conductance were affected in the most resistant transgenic line. Surprisingly, inoculation of tomato plants with Pst J4 did not cause stomatal closure (Supplementary Figure S2), suggesting that Pst J4 has some mechanisms to keep stomata open (Cai et al., 2011). Moreover, no clear differences in stomatal morphology and conductance were detected between the transgenic line and the control (**Figure 3**; Supplementary Figure. S2). Further investigation is thus needed to understand the mechanisms underlying the observed resistance in the AtELP3 and AtELP4 transgenic lines, which is effective only when a spray inoculation method is used.

By using RNA-seq and qPCR, we identified a group of genes (PR1b1, PR-5x, DES, and ER1) that were induced to higher levels after Pst J4 infection in the most resistant AtELP4 transgenic line than in the control (**Figure 4**). Note that these genes were not constitutively expressed in the transgenic tomato plants, which is different from the constitutive defense gene expression seen in the transgenic strawberry plants (Silva et al., 2017). These tomato genes have been reported to be involved in defense responses to pathogen infections. For instance, PR1b1, PR-5x, and DES have been shown to be associated with resistance to Ralstonia solanacearum in tomato (Ishihara et al., 2012). PR1b1 and PR-5x proteins were also found to accumulate in tomato xylem upon infection by Fusarium oxysporum (Rep et al., 2002), and divinyl ethers, the products of DES, have been reported to accumulate more rapidly in potato cultivars with increased resistance to late blight, a disease caused by Phytophthora infestans (Weber et al., 1999). Furthermore, The ER1 gene has been shown to be associated with bacterial speck disease in tomato (Pautot et al., 1991). Therefore, the enhanced disease resistance in the AtELP4 transgenic tomato plants is likely attributed to elevated induction of defense-related genes. Although it has been well demonstrated that Elongator regulates gene transcription by modifying chromatin structure (histone acetylation and/or DNA methylation) (Wang et al., 2013, 2015), whether overexpression of AtELP4 in tomato alters the chromatin structure of the identified defense-related genes requires further investigation.

It is interesting that overexpression of a single Elongator subunit can dramatically improve tomato disease resistance. This phenomenon has also been seen in strawberry where overexpression of AtELP3 or AtELP4 significantly increased resistance to several bacterial and fungal pathogens (Silva et al., 2017). Such results appear to be in conflict with the notion that the Elongator complex functions as a whole and mutations in any of the subunits compromise the activity of the complex (Versées et al., 2010; An et al., 2016). However, it has been shown that overexpression of ELP3, but not ELP4, in human 293 T cells suppressed cell growth and enhanced transcription, and that overexpression of ELP4 and ELP3 together exhibited a synergistic effect on transcription activation (Gu et al., 2009). Moreover, elevating ELP3 expression in yeast suppressed the anaphase-promoting complex 5 mutant defects (Turner et al., 2010). These results together strongly suggest that individual Elongator subunits may either be able to increase the Elongator complex activity or have some Elongator complex-independent functions. Further investigation is clearly warranted to pinpoint the underlying molecular mechanisms.

Tomato appears to have a functionally conserved Elongator complex. The tomato genome contains single-copy genes encoding all SlELPs and the SlELP proteins share high identities and similarities with their corresponding AtELP proteins (Supplementary Figure S3). Furthermore, tomato orthologs of the Arabidopsis AtELP3 and AtELP4 genes, when transformed

into the Arabidopsis Atelp3 and Atelp4 mutants, were able to restore wild-type morphology to the mutants (**Figures 5**, **6**). Resistance to Psm ES4326 was also completely restored in the Atelp3 mutant plants expressing the SlELP3 gene (**Figure 7**). These results indicate that the functions of SlELP3 and SlELP4 are conserved and suggest that the function of the Elongator complex may be conserved in tomato. Indeed, silencing of SlELP2 in tomato plants resulted in pleiotropic phenotypes similar to those of the Atelp mutants (Zhu et al., 2015). These results taken together indicate that the tomato Elongator complex not only is essential for plant fitness, but may also play an important roles in immunity. Although it has been shown that Elongator modulates the transcription of thousands of genes in Arabidopsis and Saccharomyces cerevisiae (Krogan and Greenblatt, 2001; Wang et al., 2013), how Elongator functions in tomato remains to be elucidated.

This study, together with our previous study (Silva et al., 2017), revealed several dramatic differences between the transgenic tomato and transgenic strawberry AtELP3 and AtELP4 plants. First, overexpression of AtELP3 and AtELP4 did not clearly impact tomato plant growth and development, which is in sharp contrast to the collateral effects observed in strawberry. Second, overexpression of AtELP3 and AtELP4 conferred resistance in tomato to bacterial speck caused by Pst J4 only when inoculated through foliar sprays but not through infiltration into the leaf apoplast, but in strawberry it provided resistance to the angular leaf spot-causing bacterial pathogen Xanthomonas fragariae when the pathogen was infiltrated into the apoplast. And finally, the elevated resistance in tomato is likely attributed to a stronger induction of defense responses in the transgenic plants than in the control, whereas in the transgenic strawberry plants resistance was associated with constitutive expression of defense genes. These results suggest that different plant species may respond differently to overexpression of genes encoding the subunits of the evolutionarily conserved Elongator complex. Further

# REFERENCES


investigations are required to fully understand this interesting phenomenon.

# AUTHOR CONTRIBUTIONS

JP, JJ, and ZM designed the research and wrote the manuscript. JP characterized transgenic plants and conducted qPCR assay. YZ performed RNA-seq analysis. FY analyzed RNA-seq data.

# FUNDING

This work was supported by a doctoral fellowship to JP from CNPq (Brazilian National Council for Scientific and Technological Development – Grant Procs. 245345/2012-4) and grants from the Florida Tomato Committee to JJ and ZM.

# ACKNOWLEDGMENTS

We are grateful to the University of Nebraska–Lincoln Plant Transformation Facility for generation of transgenic tomato plants, Mr. Gerald Minsavage for assistance with greenhouse work and bacterial population assays, Dr. Chenggang Wang for assistance with qPCR analysis, Ms. Xudong Zhang for assistance with T-DNA vector construction, and Dr. Sixue Chen (University of Florida) for access to the LI-6800 Portable Photosynthesis System.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01066/ full#supplementary-material


transcription and exocytosis. Mol. Cell 24, 139–148. doi: 10.1016/j.molcel.2006. 07.031


for normal histone acetylation levels in vivo. Proc. Natl. Acad. Sci. U.S.A. 99, 3517–3522. doi: 10.1073/pnas.022042899


important role in acclimation-induced cross-tolerance in tomato. J. Exp. Bot. 65, 595–607. doi: 10.1093/jxb/ert404

Zhu, M., Li, Y., Chen, G., Ren, L., Xie, Q., Zhao, Z., et al. (2015). Silencing SlELP2L, a tomato Elongator complex protein 2-like gene, inhibits leaf growth, accelerates leaf, sepal senescence, and produces dark-green fruit. Sci. Rep. 5:7693. doi: 10.1038/srep07693

**Conflict of Interest Statement:** ZM is a co-inventor on a patent application titled "Use of Elongator genes to improve plant disease resistance."

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

Copyright © 2018 Pereira, Yu, Zhang, Jones and Mou. 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 Analysis of PsAvr3c Effector Family From Phytophthora Provides Probes to Dissect SKRP Mediated Plant Susceptibility

Ying Zhang<sup>1</sup> , Jie Huang<sup>1</sup> , Sylvans O. Ochola<sup>1</sup> and Suomeng Dong1,2 \*

<sup>1</sup> Department of Plant Pathology, Nanjing Agricultural University, Nanjing, China, <sup>2</sup> Key Laboratory of Integrated Management of Crop Diseases and Pests, Ministry of Education, Nanjing Agricultural University, Nanjing, China

PsAvr3c is an effector identified from oomycete plant pathogen Phytophthora sojae that causes soybean root and stem rot disease. Earlier studies have demonstrated that PsAvr3c binds to a novel soybean spliceosomal complex protein, GmSKRP, to reprogram the splicing of hundreds of pre-mRNAs and consequently subvert host immunity. PsAvr3c family genes are present in some other Phytophthora species, but their function remains unknown. Here, we characterized the functions of PsAvh27b (PsAvr3c paralog from P. sojae), ProbiAvh89 and PparvAvh214 (orthologs from P. cinnamomi var. robiniae and Phytophthora parvispora, respectively). The study reveals that both PsAvh27b and ProbiAvh89 interact with GmSKRPs in vitro, and stabilize GmSKRP1 in vivo. However, PparvAvh214 cannot interact with GmSKRPs proteins. The qRT-PCR result illustrates that the alternative splicing of pre-mRNAs of several soybean defense-related genes are altered in PsAvh27b and ProbiAvh89 when over-expressed on soybean hairy roots. Moreover, PsAvr3c family members display differences in promoting Phytophthora infection in a SKRP-dependent manner. Overall, this study highlights that the effector-mediated host pre-mRNA alternative splicing occurs in other pathosystems, thus providing new probes to further dissect SKRP-mediated plant susceptibility.

### Edited by:

Shui Wang, Shanghai Normal University, China

### Reviewed by:

Felix Mauch, University of Fribourg, Switzerland Philipp Franken, Leibniz-Institut für Gemüse- und Zierpflanzenbau (IGZ), Germany

> \*Correspondence: Suomeng Dong smdong@njau.edu.cn

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 28 March 2018 Accepted: 09 July 2018 Published: 25 July 2018

#### Citation:

Zhang Y, Huang J, Ochola SO and Dong S (2018) Functional Analysis of PsAvr3c Effector Family From Phytophthora Provides Probes to Dissect SKRP Mediated Plant Susceptibility. Front. Plant Sci. 9:1105. doi: 10.3389/fpls.2018.01105 Keywords: PsAvr3c, Phytophthora, effector family, virulence, SKRP, alternative splicing

# INTRODUCTION

Phytophthora, a genus of plant pathogen oomycetes, cause many destructive crop diseases and result in considerable losses in agriculture and economy (Kroon et al., 2012). These pathogens are notoriously difficult to manage due to their capability to evolve fast to escape field resistance (Tyler, 2007). A well-characterized species is Phytophthora infestans, the causal agent of late blight disease, a major global threat to potato (Solanum tuberosum) and tomato (Solanum lycopersicum) production (Fry et al., 2015; Kamoun et al., 2015). Phytophthora sojae, another model research pathogen, induces soybean root and stem rot, resulting in significant economic losses every year around the world (Tyler, 2001). Understanding the pathogenesis of these pathogens is critical to developing durable plant resistance.

One of the important virulence mechanisms of Phytophthora species is their ability to deliver a range of secreted proteins (effector) into the host cells to subvert plant immunity. The best studied Phytophthora effector proteins belong to a so-called RxLR effector family. The RxLR effectors have a signal peptide followed by a N-terminal conserved RxLR (Arg-any residue-Leu-Arg) motif (Morgan and Kamoun, 2007). Many studies have demonstrated that RxLR effectors play roles in suppressing plant immunity through distinct ways. One way is to block host gene expression machinery. For example, P. sojae effector PsAvh23 enhances plant susceptibility through reducing the levels of GCN5-mediated H3K9 acetylation to suppress plant defense gene expression (Kong et al., 2017). Signaling transduction suppression is another common strategy, P. infestans effector PexRD2 suppress plant immunity through disruption of the MAPKKKε dependent signaling pathways (King et al., 2014). Moreover, effector could act in a plant hormone manipulation manner. For instance, P. infestans effector PexRD24 enhances plant susceptibility by attenuating jasmonic acid and salicylic acid-mediated transcriptional responses of the host plant (Boevink et al., 2016). These results demonstrate efficient but diversed virulence functions of Phytophthora RxLR effectors.

Previously, we identified P. sojae effector PsAvr3c that could be recognized by soybean cultivars that carry the cognate resistant gene Rps3c (Dong et al., 2009). Genome investigation indicated that besides the two copies of PsAvr3c, another PsAvr3c homologous gene, PsAvh27b, is also present in the vicinity of PsAvr3c loci. However, PsAvr3c rather than PsAvh27b triggers cell death on plants carrying Rps3c (Dong et al., 2009). Moreover, knockout of PsAvr3c in P. sojae strain P6497 results in a gain of virulence toward Rps3c soybean. However, these knockout mutants exhibited reduced virulence on susceptible soybeans, suggesting that PsAvr3c is required for full virulence of P. sojae. In addition, PsAvr3c carries a functional nuclear localization signal (NLS) peptide which accumulates PsAvr3c in the nucleus thus enhancing host susceptibility. We previously demonstrated that PsAvr3c binds to and stabilize soybean serine, lysine and argininerich protein (GmSKRPs), a novel plant spliceosome component that is involved in the alternative splicing of plant pre-mRNA (Huang et al., 2017). GmSKRPs is a negative regulator of plant immunity, and ectopic expression of GFP-GmSKRP1 in N. benthamiana also promoted P. capsici colonization, PsAvr3c and GmSKRP1 work through same genetic pathway (Huang et al., 2017). Interestingly, more than four hundred genes are differentially spliced in both GmSKRPs and PsAvr3c expressing lines, including soybean resistance related genes such as NAC and WRKY transcription factors (Huang et al., 2017). Intriguingly, PsAvr3c homologous genes are also present in other Phytophthora species. However, whether PsAvr3c family members from other Phytophthora species share similar virulence function is entirely unknown.

In this study, the functional analysis of PsAvr3c homologs from several other Phytophthora species revealed the conservation of effector proteins that act in PsAvr3c-similar manner. This study suggests that effector-mediated host pre-mRNA splicing change may act as a conserved virulence function in a range of Phytophthora species. Furthermore, the study also highlights the functional diversity of PsAvr3c thus providing us with valuable tool to further dissect the mechanisms of SKRP-mediated premRNA alternative splicing in plant immunity.

# MATERIALS AND METHODS

# Plant and Microbe Cultivation

Nicotiana benthamiana plants were grown in a greenhouse for 5– 6 weeks under a 16 h day at 25◦C and 8 h night at 22◦C. Etiolated soybean seedlings were grown at 25◦C without light for 5–6 days before inoculation. P. sojae (P6497) and P. capsici (Pc35) strains were routinely maintained on 10% vegetable (V8) juice medium at 25◦C in the dark.

# GenBank Accession Numbers

ProbiAvh89 (MH450044); P. pistAvh226 (MH450043); P. parvAvh214 (MH450045); P. niedAvh208 (MH450046); P. cijaAvh190 (MH450047); P. vignAvh281 (MH450048). We harvested the full sequences of PsAvr3c and PsAvh27b from the publication paper (Dong et al., 2009).

# Plasmid Construction

The PsAvh27b gene was cloned using cDNA from P. sojae, and ProbiAvh89, PparvAvh214 genes were artificially synthesized through the given amino acid sequence (Supplementary Table 1). All of these genes without a signal peptide were amplified using a combination of primers (Supplementary Table 2), and then ligated into pBINGFP2 (a plasmid containing green fluorescent protein (GFP) with the In-Fusion HD Cloning Kit (Clontech, Mountain View, CA, United States). The resulting recombinant plasmids were transformed into Agrobacterium rhizogenes K599 or A. tumefaciens GV3101 using the freeze-thaw method. The validated transformants were then used for transient expression of the corresponding effector genes into N. benthamiana or soybean using previously described protocols (Kereszt et al., 2007). PsAvr3c and PsAvh27b (without signal peptide and RXLRdEER motif), and, ProbiAvh89 and PparvAvh214 (without signal peptide) genes were amplified using the combination of primers (Supplementary Table 2), and ligated into the pET32a vector (containing His tag) for pull-down assays, GmSKRPs was inserted into the pGEX-4T-2 vector (containing GST tag) (GE Healthcare Life Science). Individual colonies for each construct were tested for inserts by PCR, and selected clones verified via sequencing.

# In Vitro GST Pull-Down Assays

pET32a empty vector, His-PsAvr3c, His-PsAvh27b, His-ProbiAvh89, His-PparvAvh214, GST empty vector and GST-GmSKRPs were expressed in E. coli strain Rosetta2 respectively. The pull-down assay was performed using ProFound pull-down GST protein-protein interaction kit (Pierce) according to the manufacturer's instructions. The soluble total GST-fusion proteins were incubated with 25 µL glutathione-agarose beads (Invitrogen) for 5 h at 4◦C. The beads were washed three times and then incubated with 1 mL of

bacterial lysates containing His proteins for another 4 h at 4◦C. The beads were then rewashed three times, and the presence of His proteins was detected by western blot using anti-His antibody.

# Transformation of Soybean

Soybean cotyledons were inoculated with A. rhizogenes K599 carrying pBINGFP2, pBINGFP2-PsAvr3c, pBINGFP2-PsAvh27b, pBINGFP2-ProbiAvh89, and pBINGFP2-PparvAvh214. Individual cotyledons were collected from 6-day-old soybean seedlings. Detached cotyledons were surface-sterilized with 70% ethanol before a small, roughly circular (diameter = ∼0.4 cm) cut was made in each cotyledon ∼0.3 cm from the petiole end. The wounded cotyledons were then transferred to a sterile Petri plate containing 0.8% agar. A. rhizogenes cells grown in LB medium supplemented with kanamycin were washed and resuspended in 10 mM MgCl<sup>2</sup> to a final concentration of OD<sup>600</sup> = 0.6. Twenty microliters (20 µL) of the cell suspension was directly inoculated onto the wound site of each cotyledon. The inoculated cotyledons were incubated in a growth chamber at 25◦C under high humidity with a 16h/8h light/dark regime. GFP-tagged PsAvh27b, ProbiAvh89 and PparvAvh214 or GFP were expressed under the control of the CaMV 35S promoter. Fluorescence microscopy was used to select GFP, PsAvh27b, ProbiAvh89, and PparvAvh214 expressing roots. Green fluorescence was detected in hairy roots using a fluorescence stereomicroscope (Leica MZ FLIII, Wetzlar, Germany) with a GFP2 filter (excitation 480/40 nm, emission 510 nm). The expression of PsAvh27b, ProbiAvh89, and PparvAvh214 proteins in hair roots was confirmed by western blotting.

# qRT-PCR Analyses and Measurement of Splicing Efficiency Ratio

The RNA was quantified using a spectrophotometer (ND-1000; NanoDrop, Wilmington, DE, United States). Spectrophotometric analysis was used to determine the yield and purity of total RNA and make sure the ratio of OD260/280 is between 1.9 and 2.0. Agarose gel electrophoresis was used to test whether there existed RNA degeneration. To eliminate contaminating genomic DNA in the RNA samples, 3 µg of total RNA was treated with two units of RNase-free DNase I (Takara Bio Inc., Otsu, Japan) at 37◦C for 30 min. First-strand cDNA was synthesized using Moloney murine leukemia virus reverse transcriptase (RNase-free) and an oligo(dT) 18 primer (Invitrogen, Carlsbad, CA, United States). qPCR was performed in 20 µl reactions containing 20 ng of cDNA, 0.2 mM gene-specific primer or the reference actin gene, 10 µl of SYBR Premix ExTaq (TaKaRa) and 6.8 µl of deionized water. PCR was performed on an ABI Prism 7500 Fast Real-Time PCR System (Applied Biosystems Inc., Foster City, CA, United States) under the following conditions: 95◦C for 30 s and 40 cycles at 95◦C for 5 s and 60◦C for 34 s, followed by a dissociation step, that is, 95◦C for 15 s, 60◦C for 1 min and 95◦C for 15 s.

The splicing efficiency ratio was calculated by determining the level of spliced RNA normalized to the level of unspliced RNA. The spliced primers were designed to measure the intronspliced isoform expression, that crossed exon–exon junctions, while the unspliced primers were designed to span the intron– exon junction in order to measure the intron-retained isoform expression. Data were shown as average fold-change over the splicing efficiency ratio from three biological repeats.

# Plant Inoculation

For assays of P. capsici infection on N. benthamiana leaves, coexpression of GFP empty vector, GFP-PsAvr3c, GFP-PsAvr3CM4 , GFP-PsAvh27b, GFP-ProbiAvh89, and GFP-PparvAvh214 with GFP-GmSKRP1 proteins were confirmed by western blotting using an anti-GFP antibody, with pBinGFP2 empty vector and GFP-PsAvr3CM4 as the negative control and GFP-PsAvr3c as the positive control, respectively. P. capsici isolate LT263 was used for infection of N. benthamiana leaves. The same size mycelial plugs were obtained using a 6 mm cork borer and used to inoculate the abaxial surface of detached N. benthamiana leaves, placed on moist tissue paper and incubated in sealed boxes. The lesion areas (cm<sup>2</sup> ) were measured at 36 h under UV light after inoculation by mycelium plug. The primers (Supplementary Table 2) specific for P. capsici and N. benthamiana actin gene were used to quantify the relative biomass of pathogen by quantitative PCR. The experiments were conducted in triplicates and were repeated at least three times with a minimum of 18 infected leaves.

# Agro-Infiltration Assay

The pBINGFP2-PsAvr3c, pBINGFP2-PsAvh27b, pBINGFP2- ProbiAvh89, pBINGFP2-PparvAvh214 and RFP-GmSKRP1, pBINGFP2-GmSKRP1 fusion vectors were transformed into A. tumefaciens strain GV3101 and then grown overnight at 28◦C in Luria-Bertani culture medium with kanamycin and rifampicin antibiotics. The cells were harvested by centrifugation at 5000 rcf and resuspended in Agro-infiltration buffer (10 mM 2-[N-morpholino] ethanesulfonic acid [MES], 10 mM MgCl<sup>2</sup> and 150 mM acetosyringone). The OD600 was adjusted to 0.05– 0.5, depending on the experiments, before syringe infiltration into the leaves of 3 to 4-week-old N. benthamiana plants. For co-expression of multiple constructs, suspensions carrying each construct were thoroughly mixed before infiltration. The agrobacterial suspension was left at room temperature for 2 h before infiltration.

# Confocal Microscopy

After 2 dpi, patches of agro-infiltrated N. benthamiana leaves were cut and mounted in distilled water and analyzed using an LSM 710 laser scanning microscope (Carl Zeiss, Germany) with a 20×, 40×, or 60× objective lens. The green and red fluorescence were observed at excitations of 488 nm or 561 nm, respectively.

# RESULTS

# Sequence Analyses of PsAvr3c Effector Family Proteins in Phytophthora Species

The previous study demonstrated that PsAvr3c is a virulence effector which can reprogram host pre-mRNA splicing to

promote disease (Huang et al., 2017). Recently we found that potential homologous proteins of PsAvr3c exist in several Phytophthora species. To further validate the presence of PsAvr3c gene family and make functional analysis, we conducted a BLAST search of PsAvr3c homologs in both NCBI database and unpublished Phytophthora genome initiative database. PsAvr3c homologs were identified from seven Phytophthora spp. (P. sojae, P. cinnamomi var. robiniae, P. pistaciae, P. parvispora, P. niederhauserii, P. cijani, and P. vignae) with an E-value cutoff at 1e-15 (Supplementary Table 1). Among these sequences, only PsAvh27b is a paralog from P. sojae, while all the others are regarded as PsAvr3c orthologs. Multiple sequence alignment analysis revealed that all of these effectors had a signal peptide and RXLR motif, and also contained two conserved W-motifs within the effector domain. However, no other functional domains or NLSs were apparent in the predicted sequences of these effectors (**Figure 1A**).

To investigate evolutionary relationships among PsAvr3c and its family members, we ran sequence alignment and conducted an unrooted phylogenetic tree using the neighborjoining (NJ) algorithm (**Figure 1B**). The results revealed that PsAvr3c family members are classed into three main clades, PsAvr3c and PsAvh27b belong to one clade, ProbiAvh89 and P. pistAvh226 form one clade, the others including P. parvAvh214, P. niedAvh208, P. cijaAvh190, and P. vignAvh281 form another clade. In addition, we also concluded that both PsAvh27b and ProbiAvh89 are the closest genes from PsAvr3c in the phylogenetic tree, and may share the most similar biological functions. PparvAvh214 together with other homologous are quite distant from PsAvr3c, indicating potentially diversified functions from PsAvr3c. On this basis, we focused our research on PsAvh27b, ProbiAvh89, and PparvAvh214, with the aim to support the hypothesis that the effector family members that were most closely related to PsAvr3c from the phylogenetic tree may share the fundamental virulence function, and the gene which was farther related to PsAvr3c differentiated into other virulence functions during evolution.

# PsAvr3c Family Proteins Bind to GmSKRPs in Vitro and Promote GmSKRPs Stability in Planta

PsAvr3c physically binds to and stabilize GmSKRPs, a negative immune regulator rich in residues of serine, lysine, and arginine (Huang et al., 2017). GmSKRP1 and GmSKRP2 differ in DNA sequences length, coding 558 bp and 552 bp respectively, but still share 94% similarity at the amino acid sequence level (Huang et al., 2017). To test whether PsAvr3c homologous proteins interact with GmSKRPs, we expressed recombinant His-PsAvh27b, His-ProbiAvh89, His-PparvAvh214, and His-PsAvr3c (as positive control) proteins together with GST-GmSRKP1/2 or GST (as negative control) proteins in Escherichia coli, and performed GST pull-down assays. The results (**Figure 2A**) show that both PsAvh27b and ProbiAvh89 were detected in GmSRKP1 pull-down products in vitro; however, no band could be detected between PparvAvh214 and GmSRKP1. Surprisingly, ProbiAvh89 was the only one that comes down with GmSKRP2. Neither PsAvh27b nor PparvAvh214 could be detected in the GST-GmSKRP2 pull-down assay (**Figure 2B**). This observation suggests that ProbiAvh89 directly binds to both GmSKRPs as PsAvr3c, while PsAvh27b only binds to GmSKRP1 but not GmSKRP2. However, no direct interaction between PparvAvh214 and GmSKRPs were observed in our assay.

To test whether selected PsAvr3c family effectors stabilize GmSRKP1 in vivo, we expressed FLAG-GmSRKP1 together with individual GFP tagged PsAvr3c family members in N. benthamiana and harvested protein samples for Coimmunoprecipitation (Co-IP) precisely as we had done in our earlier study (Huang et al., 2017). Immuno-blot data indicated that significant amounts of GmSRKP1 were accumulated in the presence of PsAvh27b or ProbiAvh89 compared to PparvAvh214 and GFP (**Figure 2C**). These results together with our in vitro pull-down data suggested that PsAvr3c family proteins interact with GmSKRPs with different binding activities, and provides excellent natural variants to study SKRP mediated plant susceptibility.

# The Alternatively Splicing Pattern of Several Soybean Pre-mRNA Are Changed in the Presence of PsAvr3c Family Effectors

To further investigate whether soybean pre-mRNAs are alternatively spliced by other family members, we expressed PsAvr3c family effectors in individual-soybean hairy roots, and five soybean hairy roots were gathered to extract RNA for measuring the splicing ratio of selected marker genes by qRT-PCR. The marker genes we selected for this study were a NAC transcription factor (Glyma.02G222300), a WRKY transcription factor (Glyma.03G220800), a histidine-containing phosphotransfer factor (Glyma.02G150800) and a control gene called COP9 signalosome complex subunit (Glyma.03G016800), which has the intron splicing that is not affected by PsAvr3c and GmSKRP1 as we previously examined in our earlier study (Huang et al., 2017).

Significant changes in splicing ratio of splicing variants from WRKY transcription factor and histidine-containing phosphotransfer factor were detected by qRT-PCR in ProbiAvh89 and PsAvh27b expressed hairy roots. However, no change was detectable in PparvAvh214 over-expressed lines (**Figures 3A,B**). Unexpectedly, the splicing ratio of the NAC transcription factor gene was significantly altered in all the effector expressing hairy roots (**Figure 3C**). As a control, intron splicing of COP9 signalosome complex subunit gene was not affected in all the lines (**Figure 3D**). Meanwhile, the leftover of the same samples was also used to extract proteins for effector gene expression testing. Immuno-blot data demonstrated that these effector proteins accumulated at a similar level in soybean hairy roots (**Figure 3E**), suggesting that the changes of splicing ratio are not likely due to the gene expression difference. These results demonstrate that ectopic expression of either ProbiAvh89 or PsAvh27b in soybean

database and unpublished Phytophthora genome initiative database; alignment was generated by BioEdit. Identical residues are boxed in black. (B) Phylogenetic relationships of PsAvr3c homologous genes from various Phytophthora spp. Neighbor-Joining analysis of Kimura's distances calculated based on the nucleotide sequences with the MEGA5 program, using full-length amino acid sequences, numbers indicate bootstrap value from 1,000 replicates. The sequence of PsAvr3b was used as an out-group.

hairy roots affects the alternative splicing of selected genes in a PsAvr3c-similar manner.

# PsAvr3c Family Effector Enhanced SKRP Mediated Plant Susceptibility With Different Activities

Ectopic expression of GmSKRP1 in N. benthamiana promotes Phytophthora capsici infection, and co-expression of PsAvr3c has a synergistic effect on GmSKRP1 mediated susceptibility (Huang et al., 2017). To investigate whether other family members enhanced GmSKRP1 mediated susceptibility, we co-expressed the individual effector together with GmSKRP1 in plant cells prior to pathogen challenge. As shown in data, only co-expression of ProbiAvh89 with GmSKRP1 resulted in increased plant susceptibility to P. capsici (Supplementary Figure 1). However, this synergistic effect exhibited no clear differences when the PsAvh27b and PparvAvh214 were co-expressed with GmSKRP1 respectively (**Figures 4A,B**). Here, PsAvr3c and its GmSKRP binding deficient mutant PsAvr3cM4 were used as a positive and negative control, respectively. Proteins of the expected size were detected (**Figure 4C**). Our data indicated that only effector ProbiAvh89 functions in a similar manner to PsAvr3c and promoted plant susceptibility in a GmSKRP1-dependent manner.

# Translocation of GmSKRP From the Nucleoplasm to Nucleolus Is Not Associated With Effector-Mediated Susceptibility and Pre-mRNA Alternative Splicing

To further investigate the relationship of effector genes with their subcellular localization, we generated the RFP-GmSKRP1 fusion, and GFP tagged effector fusion constructs and introduced them into N. benthamiana for confocal

observations. When we co-expressed GmSKRP1 with GFP in N. benthamiana, GmSKRP1 was diffusely distributed in the nucleoplasm, nuclear speckles and a mixed pattern including nucleoplasm and nucleolus (**Figure 5A**). GmSKRP1 proteins predominantly (>50%) localized in the nucleoplasm of a total of 150 randomly selected and analyzed cells from N. benthamiana plant leaves. In other cases, GmSKRP localized in either nuclear speckles or in non-specific pattern (**Figure 5B**). We then co-expressed other family members with GmSKRP1 and checked their subcellular localization. We observed that ProbiAvh89 localized in the nucleolus like PsAvr3c, but both PsAvh27b and PparvAvh214 were excluded from the nucleolus and mainly localized in the nucleoplasm (**Figure 5C**). Interestingly, all of these effectors resulted in the accumulation of RFP-GmSKRP1 protein from the nucleoplasm to the nucleolus (Supplementary Figure 2), a phenomenon that was not observed in GFP control tests (**Figure 5C**). Expression of these fusion proteins were verified by western blot (**Figure 5D**). These findings combined with our previous data, indicate that translocation of SKRP from the nucleoplasm to the nucleolus is not associated with virulence and splicing.

# DISCUSSION

Alternative splicing has been found to be an important regulatory process in global gene expression (Yang et al., 2014). Previously, we reported that Phytophthora sojae core effector, PsAvr3c, targets and stabilizes GmSKRP proteins. GmSKRPs are soybean RNA spliceosome-associated proteins that participate in premRNA splicing process, that affect hundreds of soybean premRNA splicing (Huang et al., 2017). SKRP homologous proteins exist in soybean, tomato, Arabidopsis, N. benthamiana, Zea

different isoform transcript level of WRKY transcription factor (A), histidine-containing phosphotransfer factor (B), NAC transcription factor (C), and control gene COP9 signalosome complex subunit (D). The left panel shows the schematic gene model; the blue arrows indicate the primers designed to span the intron–exon junction that were used to measure the alternatively spliced isoform. The spliced primers crossing exon–exon junctions are shown with black arrows over the intron. Splicing efficiency ratio was calculated by determining the level of spliced RNA normalized to the level of alternatively spliced RNA. The soybean actin gene CYP2 was used as an internal control gene. Experiments were repeated three times with similar results. Means and standard errors from three replicates are shown ( <sup>∗</sup>P < 0.05; ∗∗P < 0.01; one-way ANOVA). (E) Immuno-blot analyses of the GFP-PsAvr3c family effectors or GFP proteins in fluorescent hairy roots of soybean cv. Williams. Transformed hairy roots were selected based on the green fluorescence. The total proteins were extracted from the hairy roots for analyses. Recombinant protein expression was confirmed by immuno-blotting using anti-GFP antibody. Asterisks indicate protein bands.

mays, Oryza sativa (Huang et al., 2017). However, whether effector-SKRP associations are prevalent in other Phytophthora species, and manipulate host plant RNA splicing machinery remains unknown. In the present study, we made the sequences alignment of PsAvr3c homologous, and found that although PsAvr3c and its homologous effector are not highly conserved, important functional motifs are apparent, including the signal peptide, the RxLR and EER motifs in the host-targeting region, and core residues in the two W domains. Although our analysis included just seven PsAvr3c homologous from other Phytophthora species, other effectors may also play roles in the manipulation of the pre-mRNA splicing machinery. For instance in P. capsici, an Oomycete, with a broad host range, and a robust model for investigations (Lamour et al., 2012). Considering that NbSKRP amino acid sequence shares high similarity with the GmSKRPs, silencing the NbSKRP gene in N. benthamiana resulted in significant resistance to P. capsici than the control plants, suggesting that SKRP in N. benthamiana and soybean confers susceptibility to other Phytophthora (Huang et al., 2017). SKRP-effector interactions seem to occur in other Phytophthora pathosystems. Therefore, we hypothesized that the pathogen effector interfering with host RNA splicing machinery to reprogram the splicing of plant pre-mRNA is a general pathogenic mechanism. Further experiments are required to validate this possibility.

In our study, we identified both ProbiAvh89 and PsAvh27b that target and stabilize GmSKRP1. In line with these data, our experiments also demonstrate that splicing ratio of WRKY, NAC transcriptional factor and histidine-containing phosphotransfer factor were lower in both ProbiAvh89 and PsAvh27b expression line than control. The ability of the conserved homologous effector to share similar function is not surprising, given that homologous oomycete effectors that employ similar functions have been previously reported. For instance, Hpa RXLR effector HaRxL96 and its homologous effector from P. sojae PsAvh163 share conserved functional

PsAvr3c family effectors. Confocal images of N. benthamiana leaf epidermal cell nuclei transiently expressing the RFP-GmSKRP1 with GFP-PsAvr3c homologous proteins demonstrate that GmSKRP1 protein were relocated from the nucleoplasm to the nucleolus when they are co-expressed with GFP-PsAvr3c family proteins but not with GFP control. Scale bar represents 10 µm. (D) Immunoblot analyses of GFP, GFP-PsAvr3c, GFP-PsAvh27b, GFP-ProbiAvh89, and GFP-PparvAvh214 proteins. Total proteins were extracted at 48 hpi. Protein expression was confirmed by immunoblotting using an anti-GFP antibody. Protein bands are indicated by asterisks while protein loading is indicated by Ponceau stain.

domains and target a central component of the resistance network that contributes to plant resistance, suggesting that conserved effectors manipulate the same or similar targets in the pathogens' respective hosts (Anderson et al., 2012). Previous studies have demonstrated that expression of Phytophthora sojae suppressor of RNA silencing (PSR2) significantly enhanced the susceptibility of soybean to P. sojae. The homolog of P. sojae PSR2 (PsPSR2) in P. infestans (PiPSR2), is capable of suppressing RNA silencing activity when expressed in N. benthamiana and enhance the susceptibility of N. benthamiana to Phytophthora infection (Xiong et al., 2014). These discoveries together with our current findings present a compelling case that this mechanism that effector target SKRP protein to interfere with host RNA metabolism is not an isolated phenomenon in Phytophthora.

Interestingly, our assay does not detect clear enhancement of the susceptibility when PsAvh27b was co-expressed with GmSKRP1 in N. benthamiana, suggesting that the roles of PsAvh27b and PsAvr3c may share similarity but are not entirely redundant. Several pieces of evidence support this hypothesis. Although PsAvh27b target GmSKRP1 in a PsAvr3c similar pattern, there was no apparent binding signal when PsAvh27b was co-incubated with GmSKRP2 in E. coli, suggesting that PsAvh27b only binds to GmSKRP1 whereas PsAvr3c and PsAvh89 bind to both GmSKRP1 and GmSKRP2. Hence, the results indicate that both GmSKRP proteins must be affected to obtain a phenotypic effect on disease. Moreover, we noticed that the subcellular localization of PsAvh27b and PsAvr3c are not precisely the same. PsAvh27b predominantly localized at nucleoplasm, while PsAvr3c mainly accumulated in the

nucleolus. Therefore, it is possible that although PsAvh27b maintains some similarity with PsAvr3c, it possesses diverse differentiated function. Taken together, that may explain why when only PsAvr3c gene was knocked-out in P. sojae strain P6497, the virulence of the mutants declined significantly in susceptible soybeans (Huang et al., 2017). The conclusion fits and complements existing theories of neofunctionalization driven by positive selection. Previous studies supports this hypothesis. For example, Phytophthora sojae crinkling and necrosis (CRN) gene PsCRN172–2 triggers programmed cell death (PCD) whereas its duplicated copy PsCRN172–1, which only differs by seven amino acids does not (Shen et al., 2013).

In contrast to PsAvh27b and ProbiAvh89, PparvAvh214 showed the opposite effect on the splicing pattern in WRKY transcriptional and histidine-containing phosphotransfer factor pre-mRNA (**Figure 3**), even though it shares 37% identity to PsAvr3c. This suggests that PsAvr3c family effector PparvAvh214 may have differentiated during Phytophthora evolution. Indeed, the effector gene family members differentiation into distinct functions have been previously reported. For example, a previous study demonstrated that the evolution of bacterial type III secreted effectors (T3SEs) is strongly influenced by a non-homologous recombination process, that generates novel T3SEs with new virulence functions (Stavrinides et al., 2006). Recent studies have revealed that Xanthomonas campestris (Xcv) effector XopD, has a longer N-terminal domain that determines functional specificity in tomato. While XopD family effectors share EAR motifs and SUMO protease activity, they play a specific role within the bacterial-host interaction, and do not complement Xcv 1xopD mutant phenotypes (Kim et al., 2011).

Interestingly we also found alterations in soybean NAC splicing patterns in the presence of PparvAvh214. One of the possibilities is that PparvAvh214 influences the host plant premRNA splicing through other pathways in a SKRP independent manner. It is still possible that although PparvAvh214 cannot interact with GmSKRPs, it associates with other alternative splicing factors such as SR proteins, thus influencing the splicing ratio of NAC transcription factor. Meanwhile, we did not observe whether PparvAvh214 enhance GmSKRPs induced susceptibility. Another possibility is that SKPRs have natural

# REFERENCES


variations in different plants, thus it is possible that SKRP in the host of P. parvispora, such as A. unedo (Scanu et al., 2013), is quite different from soybean SKRP. The interaction of PparvAvh214 with its natural host will be important for an in-depth physiological dissection and analysis of its function. Nevertheless, the functional analysis of PsAvr3c family effector illustrates that the mechanism of effector association with SKRPlike proteins is prevalent in plants and provides probes to dissect plant immunity.

# AUTHOR CONTRIBUTIONS

YZ and JH analyzed the data and contributed constructs. YZ, JH, and SD designed the experiments. YZ, SD, and SO wrote the manuscript. All authors commented on the article before submission.

# FUNDING

This work was supported by Fundamental Research Funds for Central Universities (KJYQ201501), the Fundamental Research Funds for the Central Universities (KYTZ201403), and the Priority Academic Program Development of Jiangsu Higher Education Institution.

# ACKNOWLEDGMENTS

We thank Prof. Brett Tyler (Oregon State University) and Dr. Danyu Shen for providing sequence of PsAvr3c-like proteins, Dr. Meixiang Zhang (Nanjing Agricultural University) for providing P. capsici strains PC35.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01105/ 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 Zhang, Huang, Ochola 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.

# A Plant Extract Acts Both as a Resistance Inducer and an Oomycide Against Grapevine Downy Mildew

Yuko Krzyzaniak<sup>1</sup> , Sophie Trouvelot<sup>1</sup> , Jonathan Negrel<sup>1</sup> , Stéphanie Cluzet<sup>2</sup> , Josep Valls<sup>2</sup> , Tristan Richard<sup>2</sup> , Ambrine Bougaud<sup>1</sup> , Lucile Jacquens<sup>1</sup> , Agnès Klinguer<sup>1</sup> , Annick Chiltz<sup>1</sup> , Marielle Adrian<sup>1</sup> and Marie-Claire Héloir<sup>1</sup> \*

<sup>1</sup> UMR 1347 Agroécologie, AgroSup Dijon, Centre National de la Recherche Scientifique, Institut National de la Recherche Agronomique, Université Bourgogne Franche-Comté, Dijon, France, <sup>2</sup> Université de Bordeaux, Institut des Sciences de la Vigne et du Vin, EA 4577, Institut National de la Recherche Agronomique, USC 1366, Unité de Recherche Œnologie, Villenave d'Ornon, France

Protecting vineyards from cryptogamic diseases such as downy mildew, caused by Plasmopara viticola, generally requires a massive use of phytochemicals. However, the issues on unintentional secondary effects on environment and human health, and the occurrence of P. viticola resistant strains, are leading to the development of alternative strategies, such as the use of biocontrol products. In this paper, we evidenced the ability of a plant extract to protect grapevine from P. viticola. Further experiments carried out both on cell suspensions and on plants revealed that plant extract activates typical defense-related responses such as the production of H2O2, the up-regulation of genes encoding pathogenesis-related proteins and stilbene synthase, as well as the accumulation of resveratrol or its derivative piceid. We also brought to light a strong direct effect of PE on the release and motility of P. viticola zoospores. Furthermore, we found out that PE application left dried residues on leaf surface, impairing zoospores to reach stomata. Altogether, our results highlight the different modes of action of a new biocontrol product able to protect grapevine against downy mildew.

### Edited by:

Yi Li, Peking University, China

### Reviewed by:

Stéphane Compant, Austrian Institute of Technology, Austria Mostafa Abdelwahed Abdelrahman, Tottori University, Japan

#### \*Correspondence:

Marie-Claire Héloir marie-claire.heloir@u-bourgogne.fr

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 30 March 2018 Accepted: 05 July 2018 Published: 25 July 2018

#### Citation:

Krzyzaniak Y, Trouvelot S, Negrel J, Cluzet S, Valls J, Richard T, Bougaud A, Jacquens L, Klinguer A, Chiltz A, Adrian M and Héloir M-C (2018) A Plant Extract Acts Both as a Resistance Inducer and an Oomycide Against Grapevine Downy Mildew. Front. Plant Sci. 9:1085. doi: 10.3389/fpls.2018.01085 Keywords: biocontrol, Vitis vinifera, Plasmopara viticola, induced resistance, plant protection, oomycide, elicitor

# INTRODUCTION

Although wine industry has been meeting an economic, social and cultural wealth, it is currently facing multiple difficulties. Indeed, adapting to climate changes and keeping on protecting vineyards from major diseases are a worldwide challenging concern. Among cryptogamic diseases, downy mildew can cause severe reductions of the yields, especially in vineyards with rainy springs (Gessler et al., 2011). This disease is caused by the biotrophic oomycete Plasmopara viticola [(Berk. and Curt.) Berl. and De Toni], which infects all green parts of the plant. During P. viticola's asexual cycle, motile zoospores released from sporangia reach stomata where they encyst. The pathogen then develops an intercellular mycelium with haustoria for nutrient uptake from mesophyll cells. It next emits sporangiophores from other stomata to sporulate, which constitute a source of inoculum for next infections. Generally, to ensure the quantity and the quality of the harvest, the management of downy mildew and cryptogamic diseases requires a massive use of fungicides. However, some of them cause unintentional secondary effects on environment and human health (Gill and Garg, 2014). Therefore, several actors, including researchers and farmers, now focus on using alternative or complementary strategies, such as the use of biocontrol products (Gessler et al., 2011).

Among biocontrol products, crude extracts of plants are widely studied, since they can contain a rich cocktail of active compounds. They can contain alkaloids, phenolic compounds or essential oils, which can exhibit direct toxicity toward plant pathogens. Abundant literature is available about antifungal activity of plant extracts (PEs; for review, see Sokovic et al., ´ 2013; Shuping and Eloff, 2017). For example, weed extracts showed strong antifungal activity against Fusarium oxysporum in vitro (Sharma and Kumar, 2009). Leaf extracts and essential oils of Cymbopogon citratus or Eucalyptus citriodora inhibited the mycelial growth of Didymella bryoniae, the causal agent of gummy stem blight on Cucurbitaceae (Fiori et al., 2000). Considering grapevine protection, extracts from Vitis canes (Schnee et al., 2013), Inula viscosa oily paste extracts (Cohen et al., 2006) or neem oil extract (Achimu and Schlösser, 1992) exhibited direct toxicity against P. viticola. Crude extracts of plants can also contain other compounds such as polysaccharides, acting as elicitors, able to activate defense mechanisms in host plants (Ebel and Cosio, 1994).

Plant perception of elicitors leads to a cascade of signaling events including: ions fluxes, plasma membrane depolarization (Lecourieux et al., 2002; Vandelle et al., 2006), nitric oxide and reactive oxygen species (ROS, among them H2O2) production, acting as secondary messengers (Garcia-Brugger et al., 2006). Moreover, a phosphorylation cascade of MAPKs is activated. All this complex cascade of signaling events induces the expression of defense genes. This leads to the synthesis of (i) pathogenesisrelated (PR) proteins, including hydrolytic enzymes (e.g., β-1,3 glucanases and/or chitinases), which degrade pathogen cell wall (van Loon et al., 2006); (ii) antimicrobial compounds, such as phytoalexins (Adrian et al., 2012); (iii) compounds involved in the cell wall reinforcement (Underwood, 2012), like callose deposits or hydroxyproline rich glycoproteins (HRGPs). These defense responses are regulated by phytohormones, such as salicylic acid (SA), jasmonic acid (JA) and ethylene (Pieterse et al., 2009).

In grapevine, purified molecules or natural extracts able to induce defense reactions have been widely reported (for review Delaunois et al., 2014). Nonetheless, few of them lead to an actual resistance against pathogens, especially in vineyard conditions (Walters et al., 2013; Delaunois et al., 2014; Trouvelot et al., 2014). In this context, biocontrol products are of interest as they can activate defenses and/or exhibit direct toxicity against pathogen. The aim of this study is to evaluate the efficacy of a novel formulated PE, and to understand its mode(s) of action in grapevine/P. viticola interaction. Firstly, PE's efficacy against downy mildew was evaluated on plants. Secondly, eliciting properties of PE were investigated by analyzing defense responses on grapevine cell suspensions and plants. Finally, direct toxic effects of PE on P. viticola were assessed on the release and motility of zoospores, and the early infection steps.

# MATERIALS AND METHODS

For all experiments, three independent biological repetitions were performed, unless otherwise specified.

# Compound Preparation for Grapevine Treatment

A PE, was supplied by Arysta Lifesciences-Laboratoires Goëmar (Saint-Malo, France). The origin and the chemical specifications of PE are confidential. The stock solution (80 g L−<sup>1</sup> of PE) was diluted for experiments in MilliQ <sup>R</sup> water. PE is formulated in a mixture of two conserving agents (cons. agents) of confidential nature. As control, this mixture was also tested alone for all experiments using a mixture stock solution prepared according to the proportion present in PE stock solution. Equal volume of water was also used as control.

For in vitro experiments, we tested concentrations ranging from 0.5 to 3 g L−<sup>1</sup> PE to find a non-lethal dose for grapevine cells at 24 h post-treatment (hpt), as described in the section "Viability Assessment." Then, this PE concentration was used for all analyses. For in planta experiments, the concentration used was 5 g L−<sup>1</sup> PE, as recommended by the providing company.

# Cell Suspension Culture, Sampling, and Viability Assessment

# Cell Culture and Preparation

Grapevine (Vitis vinifera L. cv. Gamay) cell suspensions were cultivated in Nitsch–Nitsch medium (Nitsch and Nitsch, 1969) on a rotary shaker (125 rpm) at 25◦C and under continuous light. They were subcultured every 7 days by transferring 10 mL of cell suspensions into 90 mL of new culture medium. Seven-day-old cultures were diluted twice in Nitsch–Nitsch medium 24 h prior to use for all experiments.

For early signaling events (ROS and MAPK), cells were washed twice with the equilibration buffer M10 (10 mM MES, 175 mM mannitol, 0.5 mM K2SO4, 0.5 mM CaCl2; pH 5.3), then re-suspended at 0.1 g fresh weight of cells (FWC) per mL in M10, and equilibrated for 2 h under light (25◦C, 125 rpm), before treatments. For gene expression and resveratrol analyses, cells were adjusted at 0.1 g FWC mL−<sup>1</sup> in Nitsch–Nitsch medium before treatments.

## Sampling

For MAPK and gene expression analyses, aliquots of treated cell suspensions (1.5 and 2 mL, respectively) were separated from culture medium through vacuum filtration on GF/A WhatmanTM filters. Cells were collected into 2 mL-microtubes containing a 6 mm glass bead, then frozen in liquid nitrogen and kept at −80◦C before analysis. Samples were ground with a mixer mill twice for 30 s at 30 Hz (MM 200, Retsch) before extraction. For resveratrol analyses, samples were collected simultaneously as those intended for gene expression, with a 5 mL-microtube placed beneath the filter, to recover 2 mL of corresponding culture medium, then kept at −20◦C before analysis.

## Viability Assessment

Fluorescein diacetate (FDA), a cell permeant esterase substrate, was used to assess cell viability. Twenty microliters FDA (500 µg mL−<sup>1</sup> ) were added to 1 mL of cell suspension at 24 hpt. To visualize living cells, observations were immediately realized using an epifluorescence microscope [Leica, λexc 450–490 nm,

λem 515 nm (filter LP), magnification ×400] equipped with a digital camera. Ten pictures were acquired per condition, and percentage of viability was assessed extemporaneously by counting living and dead cells on each field of observation.

# Plant Production, Treatment, and Inoculation

Vitis vinifera L. cv. Marselan plants, susceptible to P. viticola, were obtained from herbaceous cuttings placed in individual pots (8 cm × 8 cm × 8 cm) containing a mixture of peat and perlite (3/2, v/v). They were grown until 5-leaf stage at 23◦C/15◦C (day/night), under a 16-h light photoperiod in a greenhouse. Plants were sub-irrigated with a nutritive solution (N/P/K 10-10- 10, Plantin, France).

PE, cons. agents or water (as controls) were sprayed onto upper and lower leaf faces of the second and third youngest fully expanded leaves, until run-off point (Steimetz et al., 2012). Unless otherwise specified, 48 hpt-treated leaves were inoculated with P. viticola by spraying a freshly prepared sporangia suspension at 2.10<sup>4</sup> sporangia mL−<sup>1</sup> (Kim Khiook et al., 2013).

# Protection Assays Against P. viticola

PE efficacy against P. viticola was assessed as described by Kim Khiook et al. (2013). Briefly, 6 days after inoculation, leaf disks were punched out and placed with the abaxial side uppermost, on a moist Whatman paper, in a closed plastic box. This system was left overnight in darkness and saturated relative humidity to trigger sporulation. Disease intensity was assessed by measuring the leaf area covered by the pathogen sporulation using a "macro" developed for the image analysis Visilog 6.9 software (Noesis, France; Kim Khiook et al., 2013). Forty leaf disks per condition were used and four independent biological repetitions were performed.

# Defense-Related Responses Assessment

# H2O<sup>2</sup> Production

H2O<sup>2</sup> production was measured in cell suspension as described by Gauthier et al. (2014). Measurements were carried out by using a luminol chemiluminescence assay with a luminometer (Lumat LB 9507, Berthold, Evry, France). Two hundred and fifty microliters of cell suspension were added to 300 µL of H50 medium (50 mM HEPES, 175 mM mannitol, 5 mM CaCl2, 0.5 mM K2SO4; pH 8.5), and 50 µL of 0.3 mM luminol. Relative luminescence was recorded within a 10 s period and was converted in nmol of H2O<sup>2</sup> per gram of FWC, after the establishment of a H2O<sup>2</sup> reference range with untreated cell suspension.

## Detection of Phosphorylated MAPKs by Western Blot Analyses

Protein extraction and western blot analyses were performed as described by Poinssot et al. (2003). Briefly, aliquots of treated cells were collected, as described above, at 0, 15, 30, 60 min post-treatment (pt). After protein extraction, aliquots containing 15 µg of protein per sample were solubilized in Laemmli buffer (Laemmli, 1970) and submitted to 10% SDS-PAGE, before transfer to nitrocellulose membrane (Hybond ECL, Amersham biosciences, Munchen, Germany) for western blotting. Phosphorylated MAPKs were detected with an antibody raised against a synthetic phosphor-thr202/tyr204 peptide of human phosphorylated extracellular regulated protein kinase 1/2 (α-perk1/2, Cell Signaling, Danvers, MA, United States). Probing and detection were carried out by an ECL western detection kit (Amersham biosciences, Little Chalfont, United Kingdom).

## Gene Expression Analyses by qRT-PCR **In cell suspension**

Aliquots of treated cells were collected as described above at 0, 1, 3, 6, 9, 12, and 24 hpt. Total RNA isolation was carried out with Trizol (Ambion Life technologies, Saint Aubin, France) according to the manufacturer's instructions. The RNA yield and purity were determined by Nanodrop 2000 (Thermo Scientific, Waltham, MA, United States), then checked on 1% agarose gel. Total RNA (1 µg) was used to synthesize cDNA using Superscript III reverse transcriptase kit (Life technologies, Saint Aubin, France). qRT-PCR experiments were performed using the AbsoluteTM qPCR Sybr Green ROX mix (Thermo Scientific, Waltham, MA, United States) as previously described by Gamm et al. (2011).

### **In foliar tissues**

Treated leaves were collected [0, 24, 48 hpt and at 24, 48 h post-inoculation (hpi)] and immediately frozen in liquid nitrogen. Total RNA was extracted from 100 mg of fine ground leaves with PurelinkTM Plant RNA Reagent (12322- 012, Invitrogen, Winooski, VT, United States) according to the manufacturer's instructions, with an extra step with chloroform to obtain clear aqueous phase. DNA contaminations were removed with the DNA-freeTM DNA removal kit (AM1906, Ambion Life Technologies, Saint Aubin, France) according to the manufacturers' specifications. Concentration and purity of RNA, reverse transcription and qRT-PCR steps were performed as for cell suspension study.

For all experiments, the relative change of defense gene expression was determined with the 2−11CT method (Livak and Schmittgen, 2001). Reference genes, EF1α or EF1γ, for cell and plant assays, respectively, were used as internal control (Gamm et al., 2011; Dufour et al., 2013). Sequences of the primer pairs used are reported in **Supplementary Table S1**.

## Stilbene Analyses

## **Quantification of resveratrol in cell suspensions**

Aliquots of culture medium were collected as described above at 0, 1, 3, 6, 9, 12, and 24 hpt. Trans-resveratrol was analyzed by RP-HPLC by using a Beckman System Gold chromatography system equipped with a diode array detector Model 168 and a Beckman 507 sample injector equipped with a 20 µL sample loop, as described by Krzyzaniak et al. (2018).

## **Quantification of stilbenes in leaf tissues**

Samples collected for defense-related gene expression analysis were also used for stilbene analyses. One milliliter of methanol was added to freeze-dried leaf powder (80 mg) and put into an

ultrasonic water bath at 25◦C for 15 min. After centrifugation (20,000 × g, 5 min), the supernatant was removed and conserved in a new tube. The powder was extracted again four more times as previously described. All the supernatants were evaporated to a 500 µL final volume to which 4.5 mL of water were added into a volumetric flask. The diluted phenolic extracts were centrifuged at 4500 × g for 5 min.

Analysis of stilbenes was performed by a 1260 Infinity UPLC (Agilent Technologies, Courtaboeuf, France) coupled to a 6430 triple quadrupole mass spectrometer (Agilent Technologies, Courtaboeuf, France), equipped with a Gerstel MPS2 autosampler. Five microliters of leaf extract were injected into an Agilent Zorbax SB-C18 (100 mm × 2.1 mm, 1.8 µm) thermostated at 40◦C, and separation of the compounds was performed at a flow rate of 0.4 mL min−<sup>1</sup> with a mobile phase composed of solvent A (distilled water, 0.1% formic acid) and solvent B (acetonitrile, 0.1% formic acid). The run was as follows: 0 to 3.5 min, 18% B; 3.5 to 6.5 min from 18% B to 33% B; 6.5 to 12 min from 33% B to 40% B; 12 to 13 min 40% B to 95% B; 13 to 16 min, 95% B; 16 to 16.5 min, from 95% B to 18% B. Total ion chromatograms were obtained using negative mode. The parameters were: capillary voltage, +3 kV; nebulizer pressure, 15 psi; dry gas, 11 L min−<sup>1</sup> ; dry temperature, 350◦C. Specific MRM transitions for each stilbene were used for identification and quantification with the Agilent Data Analysis software (Agilent Technologies, Courtaboeuf, France). Stilbenes (trans-resveratrol, piceid, piceatannol, astringin, α-viniferin, and miyabenol) were determined from calibration curves of pure standards (injected concentrations ranging from 0.004 to 10 µg mL−<sup>1</sup> ) and concentrations were expressed in µg g−<sup>1</sup> of pure phenolic compound. The linearity of the response of the standard molecules was checked by plotting the peak area versus the concentration of the compounds. All the standard stilbenes were produced and purified in laboratory conditions (UR Oenology, Villenave d'Ornon, France).

# Evaluation of PE Toxicity Toward P. viticola

### In Vitro Experiments

To assess the effect of treatments on the ability of P. viticola sporangia to release zoospores, 5 mL of sporangia suspension (10<sup>5</sup> sporangia mL−<sup>1</sup> ) were incubated with PE (0.5 to 2.5 g L −1 ), cons. agents or water in glass erlenmeyer for 2 h at room temperature, and then mounted in Malassez hemacytometer. The number of swimming zoospores crossing a defined unit square for 1 min were counted under light microscope (magnification ×100, Leica DME).

To assess the effect of treatments on the motility of released zoospores, untreated sporangia suspension was let for 2 h, time necessary for zoospore release. After checking their motility, zoospores were treated by PE (0.5 g L−<sup>1</sup> ), cons. agents or water, and swimming zoospores were counted 2 min later as described above.

### In Planta Experiments

In order to evaluate the direct effect of PE on P. viticola development in planta, plants were inoculated (10<sup>5</sup> sporangia mL−<sup>1</sup> ) after only 2 hpt, instead of 48 hpt, a time period supposed to be insufficient for the plant to activate its full defense responses (Trouvelot et al., 2008). At 24 and 48 hpi, leaf disks were collected, clarified, observed and analyzed as described by Kim Khiook et al. (2013). Pathogen presence was detected after aniline blue staining, and observation by epifluorescence microscopy [magnification ×200, Leica, λexc 340–380 nm, λem 425 nm (LP filter)]. In that context, pathogen presence appeared in light blue fluorescence. For 24 hpi samples, the number of stomatal infection sites per field of observation was counted and for 48 hpi samples, internal colonization by P. viticola was evaluated by image analysis using Visilog software. Three pictures per disk were acquired from 10 disks per biological repetition per condition.

## Observation of Leaf Surface by Scanning Electron Microscopy (SEM)

At 24 hpi, 0.5 cm<sup>2</sup> leaf squares were excised from treated and inoculated (10<sup>5</sup> sporangia mL−<sup>1</sup> ) leaves. The leaf surface was then characterized by cryo-scanning electron micrographs (cryo-SEMs) taken using a Hitachi (SU 8230) scanning electron microscope equipped with Quroum PP3000 t cryo attachment. Prior to imaging, leaf samples were rapidly frozen in liquid nitrogen in order to fix them. Thereafter they were sublimated at −90◦C for 5 min and coated with a thin platinum layer by sputtering at 5 mA for 10 s.

# RESULTS

# PE Protects Grapevine Leaves Against P. viticola

In order to check the ability of PE to protect grapevine against P. viticola, plants were sprayed with PE solution at 5 g L−<sup>1</sup> as recommended by the providing company, then inoculated with P. viticola 48 h later.

PE induced a good level of protection since a significant reduction of P. viticola sporulation, by 67% or 74% compared to water or cons. agents, respectively, was observed at 6 dpi (**Figure 1**). Compared to water control, cons. agents alone had no significant effect on disease severity.

# PE Activates Defense Responses in Grapevine Cell Suspensions

We first looked into PE ability to trigger some defense markers in cell suspensions. To determine the concentration to use on cell suspension, viability test after FDA staining was performed at 24 hpt according to a PE concentration range. PE reduced the percentage of cell viability at all assessed concentrations (by 20% and 40% from 1 to 3 g L−<sup>1</sup> ), except at 0.5 g L−<sup>1</sup> (**Supplementary Figure S1**), which was therefore chosen for the following experiments. Cons. agents, applied at the equal volume of the PE highest concentration, did not induce cell death.

mean ± confidence interval of four independent biological repetitions. Significant differences (p ≤ 0.05) were identified with Kruskal–Wallis coupled with Dunn's multiple comparison test. Conditions with different letters are significantly different.

# PE Induces H2O<sup>2</sup> Production and MAPK Phosphorylation

Production of H2O<sup>2</sup> was determined by chemiluminescence assay (**Figure 2A**). In response to PE treatment, cells generated H2O<sup>2</sup> from 5 min pt, and the maximum (about 47 nmol g−<sup>1</sup> FWC) was reached 20 min after treatment. This level remained steady during 30 min, and decreased from 50 min. Equivalent volume of cons. agents or water did not induce any H2O<sup>2</sup> production.

Phosphorylation of MAPK was investigated by western blotting, with polyclonal antibodies binding specifically to the phosphorylated form of plant ERK-related activated MAPKs (**Figure 2B**). Bands corresponding to two phosphorylated MAPK of 49 and 45 kDa were detected in PE-treated cells. The activation of these MAPK peaked at 15 min pt, was still important at 30 min and decreased to a basal level within 60 min. Only a slight signal was detected in cons. agents-treated conditions especially for 49 kDa MAPK, compared to water control.

## PE Induces Defense Gene Expression

qRT-PCR was carried out under a tight time course, in order to follow the expression of six selected defense-related genes (**Figure 3**) encoding a phenylalanine ammonia-lyase (PAL), a stilbene synthase (STS), a pathogenesis-related protein 1 (PR1), two lipoxygenases (9-LOX, 13-LOX) and a chitinase 4c (PR3). Globally, expression of all genes was induced in response to PE treatment with differences in kinetics and intensities, except for 13-LOX, which was repressed. PAL (**Figure 3A**) and STS (**Figure 3B**) were early upregulated, peaking from 1 hpt with 70 fold and 30-fold relative expression to water, respectively, before decreasing. Interestingly, 9-LOX was upregulated 50-fold at 3 hpt (**Figure 3D**), while 13-LOX was downregulated (0.19-fold) at the same time point (**Figure 3E**). PR3 was the gene the most upregulated by PE treatment, with an accumulation of transcripts peaking (800-fold) at 6 hpt, before decreasing slowly thereafter (**Figure 3C**). However, at 24 hpt, it still remained quite high at 200-fold. Finally, a progressive upregulation of PR1 from 6 to 24 hpt was detected with a maximum of ninefold (**Figure 3F**). Results also showed that, cons. agents did not seem to affect the expression of the studied genes.

## PE Increases Resveratrol Production

The grapevine phytoalexin resveratrol was quantified in the culture medium of grapevine treated cells (**Figure 4**). A basal amount was detected in water and cons. agents-treated conditions along the time course of the experiment, ranging from 2 to 8 µg resveratrol g−<sup>1</sup> FWC. No significant difference was observed between these two controls. Curiously, in response to PE-treatment, no resveratrol could be detected at 1 and 3 hpt (or under detection threshold). However, its level considerably increased from 6 hpt, peaked at 12 hpt (around 160 µg g−<sup>1</sup> FWC), and declined back to basal level at 24 hpt.

# PE Also Activates Defense Responses in Planta

## PE Induces Defense Gene Expression

In order to complete the results obtained with cell suspensions, qRT-PCR analyses of the same defense-related genes were conducted in planta in response to treatments before and after inoculation. In addition, the expression of PR2 encoding a β-1,3 glucanase, a lytic enzyme toward oomycete cell walls, was also followed. For samples taken before inoculation, results showed that, conversely to cons. agents, PE globally induced all studied gene expressions (**Figure 5**). The upregulation of PR3, PAL, and 9-LOX (**Figures 5C,D,F**), was the most intense at 24 hpt whereas the expression of PR1, PR2, STS, and 13-LOX (**Figures 5A,B,E,G**) peaked at 48 hpt. For inoculated samples, a strong variability was observed among biological repetitions, and results were too variable to be interpreted (data not shown).

## Among Stilbene Compounds, PE Induces Only Piceid Production

Stilbene compounds (trans-resveratrol, piceid, piceatannol, astringin, α-viniferin and miyabenol) were quantified by UPLC. Among the six analyzed compounds, only piceid was significantly detected whatever the conditions.

Without pathogen challenge (**Figure 6**, unstriped bars), in the water control, a basal level of about 2 µg piceid g−<sup>1</sup> DW was found through the kinetic. In cons. agents treated samples, piceid tended to accumulate but it was not statistically different from the water control. Conversely, PE induced a progressive and significant increase of piceid production, mainly after 48 hpt as observed in mock-inoculated samples (Pv−). Indeed, at 24 hpi (i.e., 72 hpt) or 48 hpi (i.e., 96 hpt) mock-inoculated, piceid

content was twofold higher in PE condition than in controls (10.5 and 3 µg g−<sup>1</sup> DW, respectively).

Results also showed that P. viticola inoculation itself increased piceid production (**Figure 6**, striped bars). Indeed, piceid content in water control was four times higher at 48 hpi than before pathogen challenge at 48 hpt. Though, a great variability between biological repetitions was observed after inoculation, and no significant difference between treatments was found.

# PE Inhibits P. viticola Zoospores Release, Motility, and Infection Process

## Inhibition of Zoospore Release and Motility in Vitro

At 2 hpt, about 30 mobile zoospores µL <sup>−</sup><sup>1</sup> min−<sup>1</sup> were counted in water or cons. agents-treated sporangia suspension (**Figure 7A**), and no irregular swimming pattern was noticed. Conversely, a stunning failure for sporangia to release their zoospores was observed in response to PE, whatever the concentration tested (0.5, 1, or 2.5 g L−<sup>1</sup> ).

We also studied the effect of PE on released zoospores using the lowest dose (0.5 g L−<sup>1</sup> ; **Figure 7B**). PE completely stopped the motility of zoospores. Cons. agents seemed to reduce slightly the motility, but there was no significant difference compared to water.

## Inhibition of Early Infection Steps of P. viticola in Planta

Since in vitro studies cannot completely illustrate the interaction between a biotrophic pathogen and its host plant, experiments were also performed in planta. PE significantly reduced the

treated with PE (0.5 g L−<sup>1</sup> ), or equal volumes of cons. agents or water. Six genes encoding a (A) phenylalanine ammonia-lyase (PAL); (B) stilbene synthase (STS); (C) chitinase 4c (PR3); (D) lipoxygenase 9 (9-LOX); (E) lipoxygenase 13 (13-LOX); and (F) pathogenesis-related protein 1 (PR1) were investigated by qRT-PCR. Results represent relative fold expression calculated with the 2−11Ct method, compared to the housekeeping gene EF1α and to water control for each respective time point. Data correspond to the mean ± confidence interval of three independent biological repetitions with three technical replicates each. Significant differences were identified with Student t-test compared to cons. agents condition (∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001).

number of both infection sites per field of observation at 24 hpi (**Figure 8A**), and the hyphal colonization by 10 times at 48 hpi (**Figure 8B**), as compared to water or cons. agents-treated leaves.

### PE Droplet Residues May Trouble Zoospore Swimming

Scanning electron microscopy (SEM) observation showed that spray application of cons. agents left no trace on leaf surface (**Figure 9A**, global view). We could note the pathogen encystment in stomata (**Figure 9A**, middle view, see focus on stomata), demonstrating that the infection succeeded. Conversely, PE-treated leaves showed the presence of dried deposits of different sizes on leaf surface, ranging from 50 to 500 µm of diameter (**Figure 9B**, global view). The leaf surface was not completely covered by these deposits, and two cases could be observed. In case 1, presence of PE residues was detected, and several 'footprints' looking like P. viticola structures, were spotted on the deposit (**Figure 9B**, case 1, middle view, see focus on stomata). In case 2, there was no PE residue, and P. viticola structures were observed on the leaf surface (**Figure 9B**, case 2, middle view). Indeed, according to their size, these structures were identified as zoospores, although they seemed to have lost their flagella (**Figure 9B**, case 2, focus on stomata). Most of them were far from stomata, and the closest were not encysted.

# DISCUSSION

This study aimed to value the use of a novel formulated PE as a means to protect grapevine from downy mildew, within the current biocontrol strategy context. Leaf disks sporulation assays enabled us to reveal its great efficacy against P. viticola. This encouraged us to better understand its modes of action, by bringing insights to the following questions: does PE induce resistance against P. viticola? Does it directly affect the pathogen as well and how? To this purpose, in vitro and in planta experiments were carried out to answer these questions.

# PE Is an Elicitor of Grapevine Defenses

We first focused on the potential elicitor capacity of PE on grapevine cell suspension and plants, by following a panel of

time point. Data represent mean from three technical replicates of one representative biological repetition.

defense markers. As a first step, the cell suspension model allowed us to better guarantee the bioavailability of PE toward the target cells. The effects of PE were investigated on some early signaling defense events. MAPK phosphorylation was assessed as this event plays a key role in response to abiotic or biotic stresses, and elicitors (Meng and Zhang, 2013). Our results showed that PE induced the phosphorylation of two MAPK of 45 and 49 kDa. In a comparable way, a transient phosphorylation of these MAPK was previously reported for grapevine cells in response to various elicitors such as soybean hydrolysates, oligogalacturonides, flagellin or laminarin (Aziz et al., 2003;

Dubreuil-Maurizi et al., 2010; Lachhab et al., 2014; Trdá et al., 2014). PE also induced H2O<sup>2</sup> production, which maximum level remained stable for 30 min. This result differs from previous works for which elicitor treatment rather induced a transient peak of H2O<sup>2</sup> (Ménard et al., 2004; Aziz et al., 2007; Gauthier et al., 2014). The rapid decrease of H2O<sup>2</sup> amount is mainly attributed to detoxification by peroxidases or catalases (Anjum et al., 2016), which use H2O<sup>2</sup> as substrate, in order to limit oxidative burst that can lead to cell death (Baker et al., 1995). This could suggest that PE affects this detoxification process. However, we showed that cell viability was not impacted at the PE concentration used.

2 min post treatment. Sporangia were collected from a sporulating leaf and concentrated at 10<sup>5</sup> sporangia mL−<sup>1</sup> , then PE was applied at different concentrations. As controls, cons. agents or water were added (volumes corresponding to the highest concentration of PE). Motile zoospores crossing one unit of Malassez hemacytometer were counted during 1 min. Results correspond to the mean ± confidence interval of three independent biological repetitions, with three technical repetitions each. Significant differences (p ≤ 0.05) were identified with Kruskal–Wallis coupled with Dunn's multiple comparison test. Conditions with different letters are significantly different.

with Kruskal–Wallis coupled with Dunn's multiple comparison test. Conditions with different letters are significantly different. Pictures presented hereby are the ones with the closest values to the mean. Scale bars represent 100 µm.

PE changed the expression of all studied defense-related genes, which is in common with other documented elicitors (Aziz et al., 2007; Lachhab et al., 2014; Dufour et al., 2016). Interestingly, using cell suspensions, we showed that PE simultaneously upregulated the expression of 9-LOX and repressed 13-LOX one. Lipoxygenase gene family encodes enzymes involved in the octadecanoid pathway, resulting in the biosynthesis of oxylipins, which constitute potent signaling or antimicrobial compounds (Blée, 2002; Prost et al., 2005; Shah, 2005). Oxylipins are produced upon pathogen challenge or elicitor applications (Weber, 2002; Randoux et al., 2010; Boubakri et al., 2013). Both 9-LOX and 13-LOX use the same substrate α-linolenic acid but they lead to different products 13-LOX is involved in jasmonate biosynthesis, while 9-LOX leads to other oxylipin synthesis (Porta et al., 2002; Wasternack, 2007). The activation of one biosynthesis pathway is therefore in detriment to the other one. Our results reflect this counterbalance. Curiously, both LOX genes were upregulated in planta. It would be interesting to further investigate these differences between cell and plant responses.

Furthermore, the effect of PE on phytoalexin production was investigated. Phytoalexins are low molecular weight compounds produced and accumulated in response to stresses (Kuc, 1990 ´ ). In grapevine, phytoalexins are produced by the phenylpropanoid pathway. The enzyme PAL acts upstream and converts phenylalanine into cinnamic acid, and STS downstream synthesizes resveratrol from p-coumaroyl-CoA and 3 malonyl-CoA. Resveratrol is the precursor of other stilbenes after chemical modifications such as glycosylation, methoxylation, and di- or

polymerization (Bavaresco and Fregoni, 2001; Chong et al., 2009; Jeandet et al., 2010). In cell suspension, it has been shown that resveratrol can be excreted in the culture medium (Adrian et al., 2012). In our study, we showed that PE upregulated PAL and STS expressions. Moreover, trans-resveratrol transiently accumulated in the culture medium just as Lachhab et al. (2014) observed in response to soybean hydrolysates, within the same order of magnitude. Interestingly, no resveratrol was detected within the first 3 h in response to PE while some was detected in controls. As PE also induced an H2O<sup>2</sup> production, resveratrol may have crosslinked to the cell walls via peroxidases as previously reported (Adrian et al., 2012).

In leaf samples, only piceid accumulated. The absence of resveratrol was quite unexpected, as it was often reported in grapevine that foliar application of some elicitors, such as oligosaccharides from B. cinerea culture extracts (Saigne-Soulard et al., 2015), chitosan (Aziz et al., 2006) or MeJA (Belhadj et al., 2006) were able to induce piceid, and also resveratrol production. In our conditions, neither resveratrol was detected in untreated-infected samples, whereas authors usually report its accumulation in response to P. viticola infection (Schmidlin et al., 2008). Vrhovsek et al. (2012) detected some resveratrol at 6 dpi, whereas it was undetectable at 2 dpi. In our case, samples were collected 1 or 2 days after inoculation, probably prior to resveratrol accumulation, so it could be interesting to assess its production at later time points. Additionally, it has been shown that plants susceptible to P. viticola infection can accumulate strong amounts of piceid, a glycosylated compound that is less toxic than the aglycon resveratrol; whereas resistant plants rather accumulate resveratrol or even viniferins, which are more active products in terms of antimicrobial capacity (Pezet et al., 2004).

Taken together, our results demonstrate that PE is an elicitor of defense reactions. However, these latter seemed to be more pronounced and less variable in cell suspension than in plants. It could be explained by PE bioavailability, less guaranteed in plants, because penetration of active molecules through leaf cuticle can be held back if not adequately formulated for instance (Paris et al., 2016).

# PE Has Also an Oomycide Activity Against P. viticola

Besides its ability to trigger defense responses, PE showed a direct antimicrobial effect toward P. viticola. It prevented zoospore release from sporangia, and inhibited their swimming in in vitro tests. Similar studies pointing out toxic effect of PEs against P. viticola, as for Juncus effusus (Thuerig et al., 2016) or pine extracts (Gabaston et al., 2017). Oomycides can have different modes of action, such as the alteration of zoospore energy production. For example, macrotetrolides antibiotics of Streptomyces species displayed similar motility impairing activities against P. viticola, Phytophthora capsici,

and Aphanomyces cochlioides. They are suspected to enhance zoospore mitochondrial ATPase activity, leading to ATP depletion and impairment in zoospore motility (Islam et al., 2016).

Microscopic observations of P. viticola in planta or in vitro were consistent since PE exerted a direct effect against the oomycete in each case. PE prevented stomatal encystment, thus confirming that zoospores were rendered unable to reach stomata at the surface of PE-treated leaves for most of them. As Tröster et al. (2017) stated accurately, there is a very short phase in the life cycle of an oomycete, when the pathogen is highly vulnerable between hatching from the sporangium and the encystment at stomata. This short phase was described like the Achilles' heel of the pathogen. SEM observations showed that many patches of PE residues were still present on the leaf surface at 72 hpt. Even the additional input of water brought by inoculation at 48 hpt did not cleanse or totally remove PE residues, suggesting that PE has notable persistence at leaf surface. After inoculation, some P. viticola structures are therefore directly in contact with PE, but conversely to in vitro experiments, SEM observations showed that sporangia were able to liberate zoospores at the leaf surface. Indeed, this contact could either kill zoospores directly, impair their swimming or physically block access to stomata. Chitosan application was reviewed to form films, which act as physical barriers around the sites of a pathogen attack (El Hadrami et al., 2010; El Guilli et al., 2016), although no supporting pictures with SEM were done. Using SEM to observe phytosanitary products residues (herbicide or insecticide) on leaves is not recent (Hart, 1979; Pedibhotla et al., 1999). However, this is the first time to our knowledge, that residues of a biocontrol product such as PE are observed in situ, especially, in presence of the pathogen.

# Implications for Research Studies Disclosing the Modes of Action of Biocontrol Products

Some authors only study the induction of defense responses to explain the resulting protection against a pathogen (Dutsadee and Nunta, 2008; Jaulneau et al., 2011; Ali et al., 2016). For foliar applications, a case in which the pathogen may directly encounter the product, we believe that information concerning direct toxicity should be required. Chitosan is a well-studied elicitor, which can also have a strong antimicrobial effect (El Hadrami et al., 2010; El Guilli et al., 2016). However, to demonstrate these two modes of action, the concentration, the molecular weight and the degree of acetylation of chitosan used were not always the same. Other authors assessed both induction of resistance and antimicrobial effects, but they showed that the protection was only correlated to eliciting activity (Trouvelot et al., 2008; Boubakri et al., 2013; Nesler et al., 2015; Clinckemaillie et al., 2017). Therefore, the double mode of action of compounds or extracts, such as PE's, is scarcely published. For example, Rheum palmatum root extract and Frangula alnus bark extract were able to display both direct toxic effect and induction of resistance on grapevine against downy mildew (Godard et al., 2009).

Finally, it is worth underlining the difficulty of quantifying precisely the part of the protection level due to each mode of action (antimicrobial/defense activation), especially against biotrophic pathogens. It would be possible by using mutant plants deprived from their ability to activate their defense response, for example. Apart from that aspect, the two modes of action of PE make it worth developing as a potential biocontrol product, since if one mode of action is ineffective (e.g., application on young organs with low-responsiveness to induced resistance), the other one may take the lead and keep on protecting the host.

# CONCLUSION

The efficacy of PE against grapevine downy mildew was described thanks to the complementarity of analyses conducted in vitro and in planta. It clearly revealed that PE acted both as a resistance inducer and an oomycide against P. viticola. Further experiments should be performed to widen PE's efficacy spectrum to other pathogens and host plants, with the aim of integrating this novel product into biocontrol-friendly crop management strategies.

# AUTHOR CONTRIBUTIONS

MA, M-CH, and ST led the project. YK, JN, AB, M-CH, AK, and AC shared in vitro experiments. YK, ST, LJ, JV, TR, and SC performed analyses in planta. YK and M-CH wrote the manuscript. MA, SC, and ST revised the manuscript and contributed to interpret data. All authors reviewed and approved the manuscript.

# FUNDING

This work was supported by Région Bourgogne Franche Comté, Feder, BPI France and by the Bordeaux Metabolome Facility and MetaboHUB (ANR-11-INBS-0010 project).

# ACKNOWLEDGMENTS

We thank DImaCell Imaging Center (INRA, Université Bourgogne Franche-Comté) and especially Aline Bonnotte for conducting the experiments on CRYO-SEM. We also thank Estelle Moreau and Grégory Lecollinet from Arysta Life Sciences-Goëmar Laboratories for providing the products.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01085/ full#supplementary-material

FIGURE S1 | Viability of cells after PE treatment. Grapevine cells were treated with PE (0.5–3 g L−<sup>1</sup> ), or with cons. agents or water (volumes corresponding to the

highest concentration of PE). At 24 hpt, the viability of grapevine cells was assessed after fluorescein diacetate (FDA) staining. Observation of living cells was realized using an epifluorescence microscope [Leica, λexc 450–490 nm, λem 515 nm (filter LP), magnification ×400] equipped with a digital camera. Ten pictures were acquired per condition, and percentage of viability was assessed extemporaneously by counting living and dead cells (about 300 cells per

# REFERENCES


repetition). Results correspond to the mean ± standard deviation of three independent biological repetitions. Significant differences (p ≤ 0.05) were identified with Kruskal–Wallis coupled with Dunn's multiple comparison test. Conditions with different letters are significantly different.

TABLE S1 | Sequences of the primers used for qRT-PCR analyses.

knowledge on their mode of action from controlled conditions to vineyard. Environ. Sci. Pollut. Res. Int. 21, 4837–4846. doi: 10.1007/s11356-013-1841-4




**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 Krzyzaniak, Trouvelot, Negrel, Cluzet, Valls, Richard, Bougaud, Jacquens, Klinguer, Chiltz, Adrian and Héloir. 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.

# Multifunctional Roles of Plant Cuticle During Plant-Pathogen Interactions

Carmit Ziv<sup>1</sup> , Zhenzhen Zhao<sup>2</sup> , Yu G. Gao3,4 and Ye Xia<sup>2</sup> \*

<sup>1</sup> Department of Postharvest Science of Fresh Produce, Agricultural Research Organization – the Volcani Center, Rishon LeZion, Israel, <sup>2</sup> Department of Plant Pathology, The Ohio State University, Columbus, OH, United States, <sup>3</sup> The Ohio State University South Centers, Piketon, OH, United States, <sup>4</sup> Department of Horticulture and Crop Science, The Ohio State University, Columbus, OH, United States

In land plants the cuticle is the outermost layer interacting with the environment. This lipophilic layer comprises the polyester cutin embedded in cuticular wax; and it forms a physical barrier to protect plants from desiccation as well as from diverse biotic and abiotic stresses. However, the cuticle is not merely a passive, mechanical shield. The increasing research on plant leaves has addressed the active roles of the plant cuticle in both local and systemic resistance against a variety of plant pathogens. Moreover, the fruit cuticle also serves as an important determinant of fruit defense and quality. It shares features with those of vegetative organs, but also exhibits specific characteristics, the functions of which gain increasing attention in recent years. This review describes multiple roles of plant cuticle during plant-pathogen interactions and its responses to both leaf and fruit pathogens. These include the dynamic changes of plant cuticle during pathogen infection; the crosstalk of cuticle with plant cell wall and diverse hormone signaling pathways for plant disease resistance; and the major biochemical, molecular, and cellular mechanisms that underlie the roles of cuticle during plant-pathogen interactions. Although research developments in the field have greatly advanced our understanding of the roles of plant cuticle in plant defense, there still remain large gaps in our knowledge. Therefore, the challenges thus presented, and future directions of research also are discussed in this review.

Keywords: plant cuticle, cutin and wax, plant-pathogen interaction, plant defense, cuticle-cell wall continuum, hormone signaling

# INTRODUCTION

Plant cuticle is the outermost layer of plants, which covers leaves, fruits, flowers, and non-woody stems of higher plants. It protects plants against drought, extreme temperatures, UV radiation, chemical attack, mechanical injuries, and pathogen/pest infection. It also provides mechanical support and serves as a barrier against organ fusion (Yeats and Rose, 2013; Kim et al., 2017).

Plant cuticle mainly comprises a matrix of cutin (an insoluble polyester) and embedded wax (soluble lipids) (**Figure 1**) (Kunst and Samuels, 2009). The wax and cutin compositions of plant cuticle can vary widely among plant species and various organs (Yeats and Rose, 2013). The cuticle of each organ has specific characteristics, for example, fruit cuticle is generally thicker than leaf cuticle and lacks stomata. Because fruit cuticle is a critical modulator of postharvest fruit quality, such as its effects on fruit water retention (Kosma et al., 2010), responses to physical and biological stresses (Kosma et al., 2009), and firmness (Matas et al., 2009; Lara et al., 2014; Vallarino et al., 2017), it is attracting increasing research attention.

### Edited by:

Zhengqing Fu, University of South Carolina, United States

#### Reviewed by:

Lirong Zeng, University of Nebraska System, United States Dylan Kosma, University of Nevada, Reno, United States

### \*Correspondence:

Ye Xia xia.374@osu.edu

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 30 April 2018 Accepted: 05 July 2018 Published: 25 July 2018

#### Citation:

Ziv C, Zhao Z, Gao YG and Xia Y (2018) Multifunctional Roles of Plant Cuticle During Plant-Pathogen Interactions. Front. Plant Sci. 9:1088. doi: 10.3389/fpls.2018.01088

**183**

For most plant species, cutin polymers mainly contain linked C16 and C18 esterified and oxygenated fatty acids (FAs), small amounts of glycerol, phenyl-propanoids, etc., (Heredia, 2003). The cuticular waxes are complex mixtures, which mainly consist of various primary and secondary alkanes, alcohols, aldehydes, ketones, and esters derived from very-long-chain FAs (C20–C34) (Raffaele et al., 2009; Malinovsky et al., 2014). Thus, FAs are the main precursors for biosynthesis of both cutin and wax, which occur mainly in plant chloroplasts. The resulting FAs are exported to endoplasmic reticulum (ER), across plasma membrane and cell wall of epidermal cells, and are deposited at the nascent cuticular membrane, where they form cutins, waxes, and suberins (complex polyesters) (Kunst and Samuels, 2009; Yeats and Rose, 2013).

There has been significant progress in the past 20 years on identification and characterization of the genes involved in plant cutin and wax biosynthesis (Suh et al., 2005; Lee and Suh, 2013; Fich et al., 2016). Also, factors involved in transportation of precursors of cutins and waxes, such as acyl-CoA-binding proteins (ACBPs) (Xia et al., 2012; Xue et al., 2014), lipid transfer protein (LTP) (Deeken et al., 2016), and ABC transporter (Luo et al., 2007), were characterized. Furthermore, several transcription factors, such as AP2, MYB94, MYB96, MYB16, and zinc-finger NFXL2 were found to play critical roles in regulating biosynthesis of plant cutin and wax (Fich et al., 2016; Lee et al., 2016). There is increasing evidence that cuticle is not merely a physical layer that protects plants; it appears to be actively involved in plant defense and signaling pathways for growth and development (Raffaele et al., 2009; Javelle et al., 2011; Aragón et al., 2017). Various studies showed that plant cuticle could function in the first layer of plant defense pattern-triggered immunity [PTI, including MAMP (microbe-), PAMP (pathogen-), and DAMP (damage-) associated molecular patterns] and the second, stronger layer of plant defense effector-triggered immunity (ETI). Thus, it serves to activate local and systemic acquired resistance against diverse pathogens (**Figure 1**) (Heredia, 2003; Xia et al., 2009; Aragón et al., 2017). The plant genes and transcription factors involved in cuticle biosynthesis/signaling and associated plant-microbe interactions were well summarized in several informative reviews (Chassot and Métraux, 2005; Muller and Riederer, 2005; Reina-Pinto and Yephremov, 2009; Lee and Suh, 2013; Serrano et al., 2014; Fich et al., 2016; Aragón et al., 2017). Studies of the plant microbial community associated with cuticular surface were well reviewed by Aragón et al. (2017). These studies, although relevant, will not be further discussed in this mini-review because of space limitation.

This review briefly summarizes the multifunctional roles of plant cuticle during plant-pathogen interactions, with emphasis on: dynamic changes of plant cuticle and their regulation; the crosstalk of cuticle with cell wall and diverse hormone signaling pathways for plant defense; and major biochemical, molecular, and cellular mechanisms that underlie the roles of cuticle during plant-pathogen interactions. The present review also discusses the challenges and future directions for related study, such as novel genes and mechanisms involved in transportation, regulation, assembly, and deposition of cuticle precursors/signals, and the associated plant-pathogen interactions.

# PLANT CUTICLE AND CELL WALL FORM A CONTINUUM AT THE PLANT SURFACE, WHICH IS DYNAMIC AND RESPONSIVE TO DIVERSE PATHOGEN INFECTIONS

Plant cell wall spans between cuticle and epidermis' cell membrane and forms a continuum with cuticle (**Figure 1**; Nawrath et al., 2013). The main components of plant cell wall are cellulose, hemicellulose, pectin, and lignin. They are involved in maintaining cell shape, supporting plant growth and development, and protecting plants from biotic and abiotic stresses (Keegstra, 2010). These cell wall polysaccharides can be incorporated into cutin matrix and thereby determine the elasticity and stiffness of the whole cuticle (Lopez-Casado et al., 2007). Cuticular wax forms a barrier to transpiration (Schonherr, 1976) and cutin matrix contributes to its mechanical strength (Kolattukudy, 1980).

Cuticle and cell wall play overlapping roles in plants; in addition to their roles as passive physical and chemical barriers, they actively function together in regulating the movement of molecules into and out of plants. They also play critical roles in relaying signals inside and outside plant cells in response to diverse stimulations for plant growth and development, and resistance to biotic and abiotic stresses (Segado et al., 2016). Both cell wall and cuticle can expand and change their compositions during various plant growth and development stages and in response to varying environmental conditions (Bargel and Neinhuis, 2005; Underwood, 2012). During plant-pathogen interactions, plant cuticle and cell wall compositions might be affected by pathogens, and conversely, pathogens can sense plantsurface components and adjust their pathogenesis and virulence accordingly. At an early stage of infection, phytopathogenic fungi can synthesize hydrolytic enzymes, such as cutinases, esterases, and lipases, which directly target cuticle and thereby play key roles in pathogenic infection (Berto et al., 1999; Garrido et al., 2012; Leroch et al., 2013; Wang et al., 2017). For instance, the fungal pathogen Fusarium oxysporum secretes cutinases, which degrade plant leaf cuticle and produce basal levels of cutin monomers to facilitate pathogen adhesion to hosts at an early stage of infection. Once a pathogen senses the resulting plant cuticle monomers, it can expand its cutinase activity to facilitate further penetration and infection in the plant cuticle layer (Woloshuk and Kolattukudy, 1986). Plant leaf cutin components, such as hexadecanediol in rice, could induce the germination and appressorium differentiation of the rice blast fungus Magnaporthe grisea (Gilbert et al., 1996) and spore germination and cutinase expression of the gray mold fungus Botrytis cinerea (Leroch et al., 2013). Furthermore, plant leaf wax components, such as very-long-chain C26 aldehydes of maize (Zea mays) could affect spore germination and penetration of barley powdery mildew Blumeria graminis f.sp. hordei (Hansjakob et al., 2011).

Interestingly, many fungi, such as Botrytis (Castillo et al., 2017), Phytophthora (Blackman et al., 2014), and M. oryzaea (Quoc and Chau, 2017) also secrete cell wall-degrading enzymes (CAZymes) even before they penetrate the cuticle, which is further evidence that the cuticle/cell wall continuum is a significant factor in plant-pathogen interactions.

Conversely, plants can recognize the attachment of pathogens and react very quickly to the elicitors (MAMPs/PAMPs) they produce. DAMPs, the pathogen-infection generated plantdegradation products, such as cutin monomers and cell wall oligosaccharides, also serve as signals that activate plant defenses against pathogen (Underwood, 2012; Malinovsky et al., 2014). For instance, tomato fruit cuticle was remodeled in response to infection of the fungal pathogens Colletotrichum gloeosporioides, and fruit cuticle biosynthesis was up-regulated during appressorium formation even before penetration (Alkan et al., 2015). Another example, during infection of citrus petals by Colletotrichum acutatum, the epidermal cells responded to the pathogen by increasing lipid synthesis and deposition of cuticleand cell wall-associated compounds, and this eventually altered the cuticle structures (Marques et al., 2016). In Arabidopsis, instead of cuticle, the crown galls caused by bacterium Agrobacterium tumefaciens infection are covered with suberin, which needs transportation of FAs to support its synthesis (Deeken et al., 2016).

In addition to wax and cutin, plant cuticle contains terpenoids and flavonoids, which have antifungal activities (Arif et al., 2009; Zacchino et al., 2017). Biosynthesis of these phenylpropanoids and flavonoids follows the formation of cuticular lipids (Mintz-Oron et al., 2008), which could be induced in response to environmental signals, such as C. gloeosporioides infection, thereby activating plant defenses in tomato and mango fruits (Alkan et al., 2015; Sivankalyani et al., 2016).

# CUTICLE PERMEABILITY, COMPOSITION, AND MULTIPLE ROLES DURING PLANT-PATHOGEN INTERACTION

Plant cuticle also plays critical roles in plant defense against diverse bacterial and fungal pathogens, most of which use natural openings, such as stomata and hydathodes in leaves, or lenticels in fruits to enter plants without directly penetrating the cuticle layer (Buxdorf et al., 2014). The integrity and permeability of cuticle are very important for its function during plant-pathogen interactions; for instance, a more permeable plant cuticle could lead to either resistance or susceptibility to pathogen infections.

Previous studies on cuticle-defective mutants of Arabidopsis and tomato, such as CYP86A2 (cytochrome P450-dependent oxidases) (Xiao et al., 2004), LACS2 (long-chain Acyl-CoA synthetases) (Bessire et al., 2007; Tang et al., 2007), PER57 (overexpressed peroxidase 57) (Survila et al., 2016), BODYGUARD (alpha-beta hydrolase) (Céline et al., 2007), and DEWAX transcription factor (Suh and Go, 2014; Ju et al., 2017) were shown to increase leaf cuticle permeability. Interestingly, these mutants improved plant resistance against the fungal pathogen B. cinerea but increased susceptibility to the bacterial pathogen Pseudomonas syringae. However, not all cuticle-related mutants showed resistance to B. cinerea: for example, our previous study found that Arabidopsis ACP4 (acyl carrier protein 4) and GL1 (GLABROUS 1) mutants had decreased levels of cutin and wax components, enhanced the permeability of leaf cuticle, and increased susceptibility to both pathogens (Xia et al., 2009, 2010). In other studies, mutations in SHINE transcription factors resulted in altered cuticle, which led to plant susceptibility to B. cinerea infection (Sela et al., 2013; Buxdorf et al., 2014).

Several mechanisms may account for the increased leaf resistance to B. cinerea and other pathogens when plant cuticle permeability increases:


A hypothetical model of the related plant disease resistance mechanism is depicted in **Figure 1A**.

Several mechanisms were suggested by which leaf cuticle permeability could raise plant susceptibility to P. syringae and other pathogens:


(5) Changes to the stomata and the cuticle-cell wall continuum that provide an easier way for pathogens to enter plants and release their virulence effectors inside.

The hypothetical model for explaining the related mechanism in plant disease susceptibility is illustrated in **Figure 1B**.

Since B. cinerea is a necrotrophic pathogen and P. syringae is a hemi-biotrophic pathogen, the cuticle may have specific modes of action to interact with pathogens having differing life styles. The specific functions of various plant leaf cuticle components may play diverse roles in plant-pathogen interactions; however, this requires further investigation.

The role of fruit cuticle in postharvest protection from pathogens has been extensively studied. Recent findings suggest that cuticle composition, rather than its mere thickness, determines fruit response to postharvest pathogens. For instance, during fruit ripening, tomato fruit susceptibility to necrotrophic fungal pathogen infection increased (Segado et al., 2016), whereas grape berries acquired resistance to biotrophic fungal pathogen powdery mildew (Uncinula necator) as they grew and developed (Fich et al., 2016). Thus, the multiple roles of plant cuticle during plant-pathogen interactions can be affected by cuticle thickness, permeability, or specific cuticular components in different tissues; they also vary with differing growth stages and environmental conditions.

# HORMONES ARE INVOLVED IN CUTICLE FORMATION AND RELATED SIGNALING DURING PLANT-PATHOGEN INTERACTIONS

Hormones regulate plant growth throughout the entire life cycle, controlling cell division, elongation and differentiation, tissue pattern formation and development, and responses to the biotic and abiotic stresses (Robert-Seilaniantz et al., 2007). Significant progress has been made in studying the biosynthesis and regulation of various hormones, such as salicylic acid (SA), jasmonic acid (JA), gibberellins (GA), abscisic acid (ABA), and ethylene (ET) (Kimbara et al., 2013; De Smet et al., 2015). And the critical roles of these hormones in plant-microbe interactions were well documented (Denancé et al., 2013). The study of crosstalk between plant hormones and cuticle for its biosynthesis and related functions during stress conditions, such as pathogen infection, also has been investigated, but there remain large gaps in our knowledge.

Several plant hormones were shown to influence plant cuticle formation and stress tolerance. For instance, we found that GA4 and GA7-treated Arabidopsis plants showed increased levels of cuticular wax and cutin components, which were associated with the improved plant immunity against bacterial pathogen P. syringae infection (Xia et al., 2010).

Jasmonic acid, an important hormone in plant defense, is derived from the 18:3 FA (linolenic acid). Methyl-JA treatment of Vicia sativa seedlings was found to induce production of ω-hydroxy FAs, which were involved in cutin formation (Pinot et al., 1998). Moreover, the ω-hydroxy FAs could induce plant resistance against pathogen infection by functioning as endogenous signaling molecules; for example, they could play critical roles in barley resistance against the fungal pathogen Erysiphe graminis f.sp. hordei (Schweizer et al., 1996).

The Arabidopsis RST1 (RESURRECTION1) gene functions importantly both in biosynthesis of cutin and wax, and in plant defense. The rst1 mutant plants showed resistance to B. cinerea that was associated with the up-regulated expression levels of JA and the related defense gene PDF1.2. However, the rst1 mutant plants exhibited down-regulated levels of SA and PR proteins, such as PR-1, which were correlated with their susceptibility to the biotrophic pathogen, Erysiphe cichoracearum (Mang et al., 2009).

Arabidopsis transcription factor SHN1 mutant plants exhibited defective leaf cuticle compositions and enhanced susceptibility to B. cinerea. The shn1-1D plants accumulated high levels of H2O2, and up-regulated a large set of genes associated with senescence, oxidative stress, and defense (Sela et al., 2013). For example, the ROS-associated genes PROPEP3 (elicitor peptide 3 precursor) and AOX1d (alternative oxidase gene) were highly up-regulated. PROPEP3 had been predicted to be the amplifier for ET/JA and SA pathways, and expression of AOX1d was also associated with these pathways (Lin and Wu, 2004; Huffaker and Ryan, 2007). The B. cinerea susceptibility of shn1-1D plants could be excessively accelerated the generation of ROSs, such as H2O2, leading to uncontrolled, excessive defense reactions, and consequent plant sensitivity and death (Sela et al., 2013).

In addition, exogenous application of ABA can specifically stimulate the formation of cuticular components in Arabidopsis, Lepidium sativum, and tomato plants, and this helped to decrease plant water loss during drought (Kosma et al., 2009; Macková et al., 2013; Cui et al., 2016; Martin et al., 2017). Furthermore, a tomato ABA-deficient sitiens (sit) mutant with reduced ABA levels and increased cuticle permeability exhibited increased resistance against B. cinerea. The related disease resistance was associated not only with changes in cuticle permeability, but also with changes in cell wall compositions. For example, after pathogen infection levels of pectin methyl-esterification and various oligosaccharides were higher in mutant than in wild-type plants (Curvers et al., 2010).

The mechanisms of crosstalk between cuticle and plant hormone pathways during pathogen interactions with fruits have started to be elucidated only recently. Indeed both ABA and ET signaling play important roles in regulation of fruit cuticle biosynthesis and function (Alkan and Fortes, 2015; Leida et al., 2016; Wang et al., 2017).

The possibilities that other hormone pathways crosstalk with cuticle biosynthesis and signaling pathways are largely unknown. The interactions between plant hormones and plant cuticle in relation to response to pathogen infections need to be further investigated.

fpls-09-01088 July 23, 2018 Time: 15:52 # 5

# CONCLUSION AND PERSPECTIVES

Altogether, we have briefly summarized recent advances in our knowledge of multiple roles of plant cuticle during interactions with diverse pathogens. Research in related fields has yielded evidence that plant cuticle plays critical roles during plantpathogen interactions. However, we are still far from fully understanding the relevant mechanisms and from developing efficient strategies to utilize the plant cuticle for plant defense. Furthermore, studies of several aspects will strengthen our understanding of the related mechanisms, which include the specific roles of the various components of cutin and wax as important factors and signaling molecules that promote either resistance or susceptibility; transmission and perception of the related factors and signals; and the crosstalk between cuticle-cell wall and hormone signaling pathways, etc. Studies of all these will provide us with more detailed knowledge to develop breeding and biotechnological approaches for enhancing cuticle function and thereby improving plant health and yield.

# REFERENCES


# AUTHOR CONTRIBUTIONS

YX and CZ organized and wrote the whole manuscript, ZZ and YG contributed to part of the writing and improvement.

# FUNDING

This project was supported by the Hatch Project from USDA-NIFA-OHO01392; the Ohio Agricultural Research and Development Center (OARDC) seed grant OHOA1591& OHOA1615; startup fund from OARDC and Ohio State University to YX.

# ACKNOWLEDGMENTS

We would like to thank Yonatan Ziv for his help with the figures.


antagonistic interaction in plant defense to biotrophs and necrotrophs. Plant Physiol. 151, 290–305. doi: 10.1104/pp.109.142158


fpls-09-01088 July 23, 2018 Time: 15:52 # 7


**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 Ziv, Zhao, Gao and Xia. 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.

# Salicylic Acid: A Double-Edged Sword for Programed Cell Death in Plants

#### Ana Radojici ˇ c´ 1 , Xin Li1,2 and Yuelin Zhang<sup>1</sup> \*

<sup>1</sup> Department of Botany, The University of British Columbia, Vancouver, BC, Canada, <sup>2</sup> The Michael Smith Laboratories, The University of British Columbia, Vancouver, BC, Canada

In plants, salicylic acid (SA) plays important roles in regulating immunity and programed cell death. Early studies revealed that increased SA accumulation is associated with the onset of hypersensitive reaction during resistance gene-mediated defense responses. SA was also found to accumulate to high levels in lesion-mimic mutants and in some cases the accumulation of SA is required for the spontaneous cell death phenotype. Meanwhile, high levels of SA have been shown to negatively regulate plant cell death during effector-triggered immunity, suggesting that SA has dual functions in cell death control. The molecular mechanisms of how SA regulates cell death in plants are discussed.

#### Keywords: salicylic acid, hypersensitive reaction, programed cell death, effector-triggered immunity, plant immunity

#### Edited by:

Shui Wang, Shanghai Normal University, China

#### Reviewed by:

Zhonglin Mou, University of Florida, United States Kenichi Tsuda, Max-Planck-Institut für Pflanzenzüchtungsforschung, Germany

> \*Correspondence: Yuelin Zhang yuelin.zhang@ubc.ca

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 13 May 2018 Accepted: 13 July 2018 Published: 07 August 2018

#### Citation:

Radojici ˇ c A, Li X and Zhang Y (2018) ´ Salicylic Acid: A Double-Edged Sword for Programed Cell Death in Plants. Front. Plant Sci. 9:1133. doi: 10.3389/fpls.2018.01133 Salicylic acid (SA) is a plant hormone that plays key roles in defense signaling (Vlot et al., 2009). Pathogen infection induces SA biosynthesis and accumulation. Two groups of Arabidopsis mutants, salicylic acid induction deficient2 (sid2) and enhanced disease susceptibility5 (eds5), are deficient in pathogen-induced SA accumulation and exhibit increased susceptibility to biotrophic pathogens (Nawrath and Metraux, 1999; Dewdney et al., 2000). sid2 mutants carry mutations in the isochorismate synthase ICS1, suggesting that SA is synthesized from chorismate following pathogen infection via ICS1 (Wildermuth et al., 2001). EDS5 encodes a multi-antimicrobial extrusion protein (MATE) transporter (Nawrath et al., 2002). The exact role of EDS5 in SA metabolism is unclear. It is likely to be involved in exporting SA or a precursor of SA out of plastids (Serrano et al., 2013).

SA is perceived by two groups of receptors, NONEXPRESSOR OF PATHOGENESIS-RELATED GENES1 (NPR1) and NPR3/NPR4, all of which display high affinity with SA (Fu et al., 2012; Wu et al., 2012; Manohar et al., 2015; Ding et al., 2018). However, they have opposite roles in transcriptional regulation of defense gene expression (Ding et al., 2018). NPR1 functions as a transcriptional activator that promotes SA-induced defense gene expression and pathogen resistance (Fan and Dong, 2002). Loss of NPR1 results in reduced SA-induced PR gene expression and increased susceptibility to pathogens (Cao et al., 1994; Delaney et al., 1995). On the other hand, NPR3 and NPR4 serve as redundant transcriptional co-repressors that prevent activation of defense gene expression when the SA level is low (Ding et al., 2018). When SA levels are high, SA inhibits the transcriptional repression activity of NPR3/NPR4 to activate the expression of SA-responsive genes. The NPR4-4D mutant protein that is unable to bind SA constitutively represses defense gene expression and blocks SA-induced immunity, rendering the mutant plants with enhanced disease susceptibility (Ding et al., 2018). Regulation of defense genes by NPR1 and NPR3/NPR4 is directly facilitated by a group of redundant bZIP transcription factors, including TGA2, TGA5, and TGA6, which interact with both NPR1 and NPR3/NPR4 (Zhang et al., 1999, 2003, 2006; Despres et al., 2000; Zhou et al., 2000).

Increased SA accumulation is associated with hypersensitive response (HR), a form of programed cell death often induced by effector-triggered immunity (ETI), as well as spontaneous cell death in lesion-mimic mutants. Early studies showed that activation of N gene-mediated defense responses by tobacco mosaic virus led to about 20-fold increase in endogenous SA levels in the infected tobacco leaves (Malamy et al., 1990). Activation of ETI by Pseudomonas effectors AvrRpm1 and AvrRpt2 in Arabidopsis also results in dramatic increases in local SA levels in a SID2 and EDS5-dependent manner (Nawrath and Metraux, 1999). Meanwhile, in mutants with spontaneous cell death, SA accumulates at much higher levels than in wild type (Bruggeman et al., 2015). However, in autoimmune mutants with no spontaneous lesion formation, such as suppressor of npr1-1, constitutive1 (snc1) and defense, no death1 (dnd1), SA levels are still dramatically increased (Yu et al., 1998; Li et al., 2001), suggesting that cell death is not required for the activation of

SA biosynthesis and high levels of SA alone are not sufficient to activate cell death.

Salicylic acid has been shown to be required for spontaneous cell death in several lesion-mimic mutants (**Table 1**). Treatment with low levels of SA activates runaway cell death in lesion simulating disease 1 (lsd1) (Dietrich et al., 1994). Blocking SA accumulation by expressing the SA hydroxylase encoded by the bacterial NahG gene suppresses lesion formation in lsd6, lsd7, accelerated cell death 6 (acd6), and acd11 mutants (Weymann et al., 1995; Rate et al., 1999; Brodersen et al., 2005). In the syntaxin of plants 121 (syp121) syp122 double mutant, spontaneous cell death is also attenuated when SA biosynthesis or SA perception is blocked (Zhang et al., 2007). However, not all lesion-mimic mutants require SA accumulation for activation of spontaneous cell death. For example, expression of NahG does not affect lesion formation in lsd2 and lsd4 mutants (Dietrich et al., 1994; Hunt et al., 1997).


<sup>∗</sup>ND, not determined; WT, wild type.

Interestingly, pre-treatment of Arabidopsis Col-0 plants with SA blocks HR activated by Pseudomonas syringae pv maculicola (P.s.m.) ES4326 carrying avrRpm1 (Devadas and Raina, 2002). In transgenic plants overexpressing NPR1, activation of cell death by the bacteria is also attenuated (Rate and Greenberg, 2001). In addition, increased ion leakage was observed in eds5-3 compared to wild type following treatment with Pseudomonas syringae pv tomato (P.s.t.) DC3000 with avrRpt2 (**Figure 1A**), indicating that AvrRpt2-induced cell death is enhanced in eds5-3. These findings suggest that activation of SA signaling plays an important role in negative regulation of cell death during ETI.

Consistent with the role of pathogen-induced SA in negative regulation of cell death in ETI, enhanced cell death was observed in the npr1-1 mutant compared to wild type following treatment with P.s.m. ES4326 carrying avrRpm1 (Rate and Greenberg, 2001), suggesting that perception of SA by NPR1 is critical for the attenuation of AvrRpm1-induced cell death. When npr1-1, npr4-4D, and the npr1-1 npr4-4D double mutant plants were challenged with P.s.t. DC3000 carrying avrRpt2, cell death in the npr1-1 and npr4-4D single mutants was similar to that in wild type, whereas npr1-1 npr4-4D exhibited enhanced cell death (**Figure 1B**), suggesting that npr1-1 and npr4-4D have additive effect on AvrRpt2-induced cell death. These data also suggest that SA signaling mediated by both NPR1 and NPR3/NPR4 plays critical roles in dampening cell death during ETI.

Consistent with the effects of pathogen-induced SA accumulation on inhibition of HR, avirulent pathogeninduced cell death in several autoimmune mutants with high SA levels was found to be greatly reduced. For example, cell death induced by P.s.m. ES4326 strains carrying avrRpt2 or avrRpm1 is dramatically reduced in aberrant growth and death2 (agd2) plants (Rate and Greenberg, 2001). The reduced cell death can be restored back to wild type level by introducing NahG or npr1-1 into agd2, suggesting that the high SA level in agd2 is responsible for the suppression of cell death activated during ETI. In the hypersensitive response like lesions1 (hrl1) mutant, cell death induced by AvrRpt2 and AvrRpm1 is also greatly reduced (Devadas and Raina, 2002). Similarly, introducing NahG or npr1- 1 into hrl1 leads to restoration of RPM1-mediated cell death. In another class of autoimmune mutants, including dnd1 and dnd2, gene-for-gene resistance is normal, but there is almost no HR following infection by avirulent bacterial pathogens (Yu et al., 1998; Jurkowski et al., 2004). Both dnd1 and dnd2 accumulate high levels of SA in the absence of pathogen infection, which is likely responsible for the lack of ETI-induced HR in these mutants.

Arabidopsis NPR3 and NPR4 function redundantly in negative regulation of defense gene expression. npr3 npr4 double mutants accumulate similar levels of SA as wild type plants, but constitutively express PR genes and exhibit enhanced resistance to virulent pathogens (Zhang et al., 2006). Interestingly, HR activated by AvrRpt2 is almost completely blocked in npr3 npr4 double mutant plants (Fu et al., 2012). AvrRpt2-induced HR is restored in the npr3 np4 npr1 triple mutant [9], suggesting that constitutive activation of SA response in npr3 npr4 mutants is responsible for the suppression of cell death activated by

FIGURE 1 | Analysis of ion leakage in eds5-3, npr1-1, npr4-4D, and npr1-1 npr4-4D plants after treatment with P.s.t. DC3000 avrRpt2. Leaves of 4-week-old plants of the indicated genotypes grown under 12 h/12 h light/dark photoperiod at 23◦C were infiltrated with mock (10 mM MgCl2) or P.s.t. DC3000 avrRpt2 (OD<sup>600</sup> = 0.02). For each plant, two leaves were infiltrated and one leaf disk was cut from each leaf immediately after infiltration. The leaf disks were subsequently washed twice in distilled water. Six leaf disks from three plants, representing one biological replicate, were transferred into a 50-ml plastic tube containing 20 ml of distilled water and electrical conductivity was measured at different time points after infiltration using a VWR EC meter (Model 2052). Each data point on the graph represents the mean ± SD of three biological replicates. In (A), Two-tailed t-test was performed for each time point between wild type (Col-0) and eds5-3 plants treated with P.s.t. DC3000 avrRpt2 ( ∗∗p < 0.01). In (B), one way ANOVA with post hoc Tukey HSD test was performed for each time point among the different genotypes. Different letters (a,b) indicate statistically significant differences between the samples (p < 0.01).

AvrRpt2. This is consistent with reduced ETI-induced cell death in autoimmune mutants with high SA levels.

In conclusion, SA plays dual roles in the regulation of programed cell death in plants. The exact mechanism of how SA regulates cell death is currently still unclear. Analysis of early SA-responsive genes by RNA-sequencing revealed that a large number of positive regulators of defense signaling are strongly

up-regulated 1 h after SA treatment (Ding et al., 2018). Induction of these defense regulators may play critical roles in potentiating defense signaling leading to activation of cell death. Meanwhile, many known negative regulators of plant immunity are also rapidly induced after SA treatment. Induction of such negative immune regulators could lead to negative feedback regulation of defense responses and cell death, which is critical in controlling the magnitude of cell death and preventing the spread of cell death beyond the infection site. The key regulatory components downstream of the SA receptors that are involved in SA-mediated inhibition of ETI-induced cell death remain to be determined in the future.

# REFERENCES


# AUTHOR CONTRIBUTIONS

YZ designed the experiments. AR performed the experiments. All authors wrote the manuscript.

# ACKNOWLEDGMENTS

We thank Yuli Ding, Tongjun Sun, and Di Wu (The University of British Columbia) for assistance with the ion leakage analysis and Natural Sciences and Engineering Research Council (NSERC) of Canada and Canada Foundation for Innovation (CFI) for the financial support.


required for salicylic acid induction of the PR-1 gene. Proc. Natl. Acad. Sci. U.S.A. 96, 6523–6528. doi: 10.1073/pnas.96.11.6523


acid. Mol. Plant Microbe Interact. 13, 191–202. doi: 10.1094/MPMI.2000.13.2. 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 © 2018 Radojiˇci´c, Li 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(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.

# TCP Transcription Factors Interact With NPR1 and Contribute Redundantly to Systemic Acquired Resistance

Min Li 1,2, Huan Chen1,2, Jian Chen1,2, Ming Chang1,2, Ian A. Palmer <sup>1</sup> , Walter Gassmann<sup>3</sup> , Fengquan Liu2,4 \* and Zheng Qing Fu<sup>1</sup> \*

*<sup>1</sup> Department of Biological Sciences, University of South Carolina, Columbia, SC, United States, <sup>2</sup> Institute of Plant Protection, Jiangsu Academy of Agricultural Sciences, Nanjing, China, <sup>3</sup> Division of Plant Sciences, C.S. Bond Life Sciences Center and Interdisciplinary Plant Group, University of Missouri, Columbia, MO, United States, <sup>4</sup> Jiangsu Key Laboratory for Food Quality and Safety-State Key Laboratory Cultivation Base of Ministry of Science and Technology, Nanjing, China*

#### Edited by:

*Hua Lu, University of Maryland, Baltimore County, United States*

#### Reviewed by:

*Zhilong Bao, Shandong Agricultural University, China Yasuomi Tada, Nagoya University, Japan*

#### \*Correspondence:

*Fengquan Liu fqliu20011@sina.com Zheng Qing Fu zfu@mailbox.sc.edu*

#### Specialty section:

*This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science*

Received: *14 May 2018* Accepted: *19 July 2018* Published: *14 August 2018*

#### Citation:

*Li M, Chen H, Chen J, Chang M, Palmer IA, Gassmann W, Liu F and Fu ZQ (2018) TCP Transcription Factors Interact With NPR1 and Contribute Redundantly to Systemic Acquired Resistance. Front. Plant Sci. 9:1153. doi: 10.3389/fpls.2018.01153* In *Arabidopsis*, TEOSINTE BRANCHED 1, CYCLOIDEA, PCF1 (TCP) transcription factors (TF) play critical functions in developmental processes. Recent studies suggest they also function in plant immunity, but whether they play an important role in systemic acquired resistance (SAR) is still unknown. NON-EXPRESSER OF PR GENES 1 (NPR1), as an essential transcriptional regulatory node in SAR, exerts its regulatory role in downstream genes expression through interaction with TFs. In this work, we provide biochemical and genetic evidence that TCP8, TCP14, and TCP15 are involved in the SAR signaling pathway. TCP8, TCP14, and TCP15 physically interacted with NPR1 in yeast two-hybrid assays, and these interactions were further confirmed *in vivo*. SAR against the infection of virulent strain *Pseudomonas syringae* pv. *maculicola* (*Psm*) ES4326 in the triple T-DNA insertion mutant *tcp8-1 tcp14-5 tcp15-3* was partially compromised compared with Columbia 0 (Col-0) wild type plants. The induction of SAR marker genes *PR1*, *PR2*, and *PR5* in local and systemic leaves was dramatically decreased in the *tcp8-1 tcp14-5 tcp15-3* mutant compared with that in Col-0 after local treatment with *Psm* ES4326 carrying *avrRpt2*. Results from yeast one-hybrid and chromatin immunoprecipitation (ChIP) assays demonstrated that TCP15 can bind to a conserved TCP binding motif, GCGGGAC, within the promoter of *PR5*, and this binding was enhanced by NPR1. Results from RT-qPCR assays showed that TCP15 promotes the expression of *PR5* in response to salicylic acid induction. Taken together, these data reveal that TCP8, TCP14, and TCP15 physically interact with NPR1 and function redundantly to establish SAR, that TCP15 promotes the expression of *PR5* through directly binding a TCP binding site within the promoter of *PR5*, and that this binding is enhanced by NPR1.

Keywords: plant immunity, systemic acquired resistance, transcriptional regulation, NON-EXPRESSER OF PR GENES 1, TCP transcription factors, PATHOGENESIS-RELATED genes

# INTRODUCTION

TCP proteins, as plant specific transcription factors (TFs), are named after the first characterized members, TEOSINTE BRANCHED1 (TB1) in maize (Zea mays), CYCLOIDEA (CYC) in snapdragon (Antirrhinum majus) and PCF in rice (Oryza sativa) (Nicolas and Cubas, 2016). Based on the sequence of the TCP specific helix-loop-helix DNA binding domain, 24 Arabidopsis TCP members are divided into class I and class II groups (Cubas et al., 1999; Martín-Trillo and Cubas, 2010). Class I proteins prefer to bind the consensus element KHGGGVC (Davière et al., 2014), whereas class II proteins bind the GTGGNCCC consensus DNA sequence (Aggarwal et al., 2010). TCP proteins govern essential functions in developmental processes, including endo-reduplication, seed germination, internode length, leaf shape, and flower development (Kieffer et al., 2011; Uberti-Manassero et al., 2012; Resentini et al., 2015; Lucero et al., 2017). In addition to regulating developmental processes, accumulating experimental evidence also implies that TCP TFs play key functions in plant immunity. Being convergently targeted by effectors from multiple pathogens suggested TCP TFs function essentially in plant immunity (Mukhtar et al., 2011; Weßling et al., 2014). TCP TFs were subsequently found to interact with SUPPRESSOR OF rps4- RLD1 (SRFR1), a negative immune regulator, and contributed redundantly to effector-triggered immunity (ETI) (Kim et al., 2014). TCP TFs can also mediate the activity of phytohormone. For instance, TCP8, TCP9, and other TCP proteins were verified to coordinately regulate the expression of ISOCHORISMATE SYNTHASE 1 (ICS1) which is responsible for pathogen-induced salicylic acid (SA) biosynthesis (Wang et al., 2015). Furthermore, Type III effector HopBB1 promotes disease susceptibility via targeting and degrading TCP14, which functions as a negative regulator of the jasmonic acid signaling pathway (Yang et al., 2017). In addition to separate functions in plant immunity or developmental processes, TCP TFs can also serve as a bridge to connect both responses. A recent study provided evidence that TCP15 connects the plant immune response with cell cycle progression by interacting with MODIFIER OF snc1-1 (MOS1) (Zhang et al., 2018).

Systemic acquired resistance (SAR) is an induced plant immunity which can be activated by pathogen infection or SA application (Fu and Dong, 2013). Pathogen infection induces the accumulation of SA which functions as an endogenous immune signal (Fu et al., 2012). SA is required for SAR, because blocking SA accumulation suppresses SAR induction (Gaffney et al., 1993). In Arabidopsis, the expression of the PATHOGENESIS-RELATED (PR) genes PR1, PR2 (encoding β-1,3-glucanase), and PR5 (encoding a thaumatin-like protein) are used as hallmarks for SAR because they maintain high expression levels during SAR (Ward et al., 1991; Uknes et al., 1992). After synthesis on the rough endoplasmic reticulum, small PR proteins (5–75 kDa) are secreted and targeted to vacuoles, or to the apoplast where bacterial pathogens are found (Dong, 2004; Edreva, 2005). PR proteins are associated with disease resistance because they exhibit anti-microbial functions both in vitro and in vivo (Ryals et al., 1996; Edreva, 2005; Breen et al., 2017).

Arabidopsis NPR1 is required for SAR and SA induced expression of PR1, PR2, and PR5 (Cao et al., 1994, 1997). As a critical transcriptional regulatory node in SAR, NPR1 regulates the expression of 2,248 out of 2,280 SA responsive genes (Wang et al., 2006). NPR1 lacks a DNA binding domain but contains two protein-protein interaction domains, suggesting that NPR1 functions as a cofactor by interacting with TFs to regulate downstream gene expression (Fan and Dong, 2002; Rochon et al., 2006; Boyle et al., 2009). Indeed, it was found that NPR1 interacts with the TGA subclass of basic Leu zipper (bZIP) family TFs to regulate the expression of PR1 (Fan and Dong, 2002). Induced SA upon pathogen infection results in cellular redox potential change, which triggers the reduction of cytosolic oligomeric NPR1 into monomeric NPR1 (Mou et al., 2003). Monomeric NPR1 proteins then enter the nucleus and interact with TGAs to facilitate the expression of PR1 (Fan and Dong, 2002; Rochon et al., 2006).

Since PR1, PR2, and PR5 are co-induced by SAR inducers, TGA TFs are thought to co-regulate their expression. However, there are no TGA binding sites (TGACGt/g, ACGTCA) (Jakoby et al., 2002) within the promoter of PR5. In addition to chemical evidence, genetic evidence also suggests separate regulators exist that regulate the expression of PR1, PR2, and PR5. First, PR2 and PR5, but not PR1, were found constitutively expressed in WRKY70 overexpression transgenic plants (Li et al., 2004). Second, only PR1 mRNA levels were found to be reduced in enhanced disease susceptibility 5-1 mutants, while the mRNA levels of PR2 and PR5 showed no apparent change (Rogers and Ausubel, 1997). All these data suggest that additional NPR1 interacting TFs are required to explain how NPR1 regulates the expression of PR5, and prompted us to screen for new NPR1 interacting TFs. Here, we show that TCP8, TCP14, and TCP15 interact with NPR1, and that they contribute redundantly to SAR establishment. TCP15 is shown to bind to the TCP binding site within the promoter of PR5 and promote its expression. In addition, the binding ability of TCP15 to the PR5 promoter was enhanced by NPR1.

# MATERIALS AND METHODS

# Plant Materials

All mutants and transgenic lines were derived from Arabidopsis [(Arabidopsis thaliana (L.) Heynh.)] ecotype Columbia-0 (Col-0). Single T-DNA insertion mutants tcp8-1 (CS875709), tcp14- 5 (CS458588), tcp15-3 (CS68533) (Kim et al., 2014), and tcp15-1 (CS875923) (Kieffer et al., 2011) were purchased from Arabidopsis Biological Resource Center (ABRC). Double mutants tcp8-1 tcp14-5, tcp8-1 tcp15-3, tcp14-5 tcp15-3 and triple mutant tcp8-1 tcp14-5 tcp15-3 were described before (Kim et al., 2014). The mutant npr1-2 was described before (Cao et al., 1997). Agrobacterium tumefaciens (strain GV3101) mediated transformation was used to construct transgenic plants through the floral dipping method. Transgenic lines pTA:TCP15-EYFP and 35S:GFP (Wang et al., 2015) were provided by Dr. Ai-wu Dong. The Dex:Flag-TCP15 constructs were transformed into the Col-0 and npr1-2 background to generate Dex:Flag-TCP15/Col-0 and Dex:Flag-TCP15/npr1- 2 transgenic plants. T3 homozygous transgenic lines were screened on 1/2 Murashige, and Skoog (MS) medium with 10µM hygromycin B. Inducible FLAG-TCP15 protein expression level was verified by immunoblot. The point mutation in the npr1-2 mutant was confirmed by restriction enzyme digestion (Cao et al., 1997).

# Growth Condition and Chemical Treatments

Seeds were vernalized at 4◦C for 3 days before growth. Soilgrown plants were placed in a growth chamber at 22◦C with 60% humidity under 12 h light. For in vitro growth, surface sterilized seeds were grown on MS plates at 22◦C with 50% humidity under 16 h light. The bacterial strains of Pseudomonas syringae pv. maculicola (Psm) ES4326 and Psm ES4326 carrying avrRpt2 were grown on King's B (KB) medium under streptomycin and both streptomycin and tetracycline selection, respectively at 28◦C. SA solutions were diluted from a 100 mM sodium salicylate (Sigma Aldrich) stock solution. Dexamethasone (DEX) solutions were diluted from a 30 mM stock solution dissolved in ethanol.

# Plasmid Construction

Primers used to amplify gene-specific sequences are listed in **Supplementary Table 1**. Fragments used in all constructs were validated by DNA sequencing. The Arabidopsis transcription factor library was purchased form ABRC (CD4-89). The entire coding regions of NPR1, TCP8, TCP14, and TCP15 were amplified by PCR with Phusion <sup>R</sup> DNA Polymerase (NEB). PCR products were subsequently introduced into the Gateway <sup>R</sup> (GW) entry vector pDONR207 (Clontech) using GW BP Clonase II Enzyme Mix (Invitrogen). Resulting entry clones were cloned into GW destination vector pDEST22 or pDEST32 using LR Clonase II Enzyme Mix (Clontech) to generate pDEST22-TCP8, pDEST22-TCP14, pDEST22-TCP15, and pDEST32-NPR1. The promoter sequences of PR1 (2,380 bp), PR2 (1,513 bp), PR5 (1,000 and 500 bp), and NPR1 (1,000 bp) were amplified by PCR and also introduced into pDONR207 by BP reactions. Resulting entry clones were remobilized into GW destination vector pLacZi (Pruneda-Paz et al., 2014) using the LR reaction to generate pPR1:lacZ, pPR2:lacZ, pPR5:lacZ, and pNPR1:lacZ constructs. To introduce site-specific mutations within the promoter of PR5, the pPR5: lacZ plasmid DNA was amplified with a pair of complementary primers with GCGGGAC to ATAAACT mutations. After digestion with Dpnl enzyme (NEB), the resulting PCR products were transformed into Escherichia coli (E. coli) strain TOP10 by electroporation. TCP15 from pDONR207-TCP15 was introduced into the GW compatible vector pTA7002\_Flag-GW (Chen et al., 2017) to generate Dex:Flag-TCP15 constructs. TCP14 and TCP15 from pDONR207-TCP14 and pDONR207-TCP15, respectively, were cloned into the pEarlyGate201 destination vector to make transient expression constructs 35S:HA-TCP14 and 35S:HA-TCP15 by LR reaction. TCP8 from pDONR207-TCP8 was cloned into the pEarlyGate202 destination vector to make a transient expression construct 35S:Flag-TCP8. The 35S:NPR1- GFP construct generated with pCB302 binary vector was described before (Mou et al., 2003).

# Yeast Two-Hybrid (Y2H) Assays

Bait plasmid pDEST32-NPR1 was transformed into the Saccharomyces cerevisiae strain AH109 (MATa). Prey plasmids pDEST22-TFs were transformed into the S. cerevisiae strain Y187 (MATα). Y2H library screening was described before Ou et al., 2011). The pDEST22 and pDEST32 empty vectors were served as negative controls. After mating, healthy diploid yeast cells growing on the double dropout medium without leucine and tryptophan (control plates) were selected. The triple dropout medium lacking leucine, tryptophan, and histidine with 1 mM 3-amino-1,2,4-triazole (3-AT) was used as selective plates. Aliquots of 10 µl diploid yeast cell suspension were spotted on control and selective plates at a concentration of OD<sup>600</sup> = 1, 0.1, and 0.01. The yeast transformation, mating, plasmid isolation and interaction test processes described in the Yeast Protocols Handbook (Clontech) were followed.

# Yeast One-Hybrid Assays

The promoter DNA:lacZ constructs were first digested with restriction enzyme NcoI (NEB), and resulting linearized constructs were subsequently integrated into the chromosome of S. cerevisiae strain YM4271 (MATa). The constructs of pDEST22-TCP8, pDEST22-TCP14, and pDEST22-TCP15 were transformed into the yeast strain YU (MATα) (Pruneda-Paz et al., 2014). The pDEST22 empty vector was also included to serve as a negative control. After mating, healthy diploid yeast cells growing on the double dropout medium lacking tryptophan and uracil were selected. The binding ability of the prey transcription factors to the bait promoter was calculated with the β-galactosidase activity assay described previously (Zheng et al., 2015).

# Agrobacterium-Mediated Transient Expression Assay

The constructs 35S:HA-TCP14, 35S:HA-TCP15, 35S:HA-EV, 35S:Flag-TCP8, 35S:Flag-EV, 35S:NPR1-GFP, 35S:EV-GFP and p19 were transformed into the A. tumefaciens strain GV3101. Resulting strains were grown in YEB culture with gentamicin, rifampicin, and kanamycin A at 28◦C overnight. Bacterial cells were collected by centrifugation at 5,000 g for 5 min. Precipitated cells were washed twice and resuspended in induction buffer (10 mM MES, pH 5.7, 10 mM MgCl2, and 200µM acetosyringone). A final concentration of resuspended cells at OD<sup>600</sup> = 0.5 was used to infiltrate into young leaves of 4-week-old Nicotiana benthamiana. Infiltrated leaves were harvested at 48 h after infiltration and stored at −80◦C for subsequent analysis.

# Plant Protein Extraction and Immunoblotting

Aliquots of 0.15 g plant sample were homogenized with 150 µl protein extraction buffer [PEB, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.1% TritonTM X-100 (Sigma-Aldrich), 0.2% IGEPAL CA-630 (Sigma-Aldrich), 50µM MG115 (Signa-Aldrich), 1 mM PMSF, 10 mM DTT, and 1 × protease inhibitor cocktail (Sigma-Aldrich)] using 2010 Geno/Grinder (SPEX). Total protein extracts were obtained by centrifuging homogenized culture at 15,000 g for 15 min twice at 4◦C. Protein concentration was quantified with Bradford reagent (Bio-Rad) using a spectrophotometer (Eppendorf). Protein samples in 5 × sample buffer (250 mM Tris-HCl, pH 6.8, 500 mM DTT, 6% SDS, 0.08% bromophenol blue, and 30% glycerol) were denatured at 70◦C for 15 min. After separation on a precast ExpressTM PAGE gel (GeneScript) by electrophoresis, denatured protein samples were transferred onto a nitrocellulose membrane (GE Healthcare). Total protein was stained with Ponceau S solution (0.1% Ponceau S (Abcam) and 5% acetic acid) to verify equal protein loading. The membrane was first incubated with a primary antibody [anti-GFP (Clontech), anti-HA-peroxidase (3F10, Roche), or anti-FLAG M2 (Sigma-Aldrich)], then incubated with a secondary antibody [goat-antirabbit lgG-HRP (Agrisera) or goat-anti-mouse lgG-HRP (Santa Cruz Biotech)]. After incubation with chemiluminescent agent SuperSignalTM West Pico or Dura substrate (ThermoFisher), targeted proteins were visualized on photographic film using an SRX-101A Medical Film Processor (Konica).

# Co-immunoprecipitation (Co-IP)

NPR1-GFP with FLAG-TCP8, HA-TCP14, or HA-TCP15 were transiently co-expressed in N. benthamiana. After 48 h, 1.5 g N. benthamiana leaves were homogenized with PEB. Homogenized samples were centrifuged at 20,000 g for 30 min twice at 4◦C. For the interaction of NPR1 with TCP14 or TCP15, protein extracts were incubated with 20 µl GFP-Trap <sup>R</sup> \_ MA Beads (Chromotek) with gentle rotation at 4◦C overnight. The conjugated beads were separated by a magnetic stand (Promega) and subsequently washed three times with cold wash buffer (10 mM Tris/HCl pH 7.5, 150 mM NaCl, and 0.5 mM EDTA). Beads were resuspended in 2 × Laemmli Sample Buffer (Bio-Rad). Immunoprecipitated proteins were eluted by boiling beads for 10 min. The bound HA-TCP14 and HA-TCP15 were detected by immunoblots with the anti-HA antibody. For the interaction between NPR1 and TCP8, protein extracts were incubated with 20 µl anti-FLAG M2 Magnetic Beads (Sigma-Aldrich) with gentle rotation at 4◦C overnight. The bound NPR1-GFP proteins were detected by immunoblot with the anti-GFP antibody. No DTT was added to any buffer to prevent disrupting disulfide linkages in the anti-FLAG beads.

# Chromatin Immunoprecipitation

Two grams of 12-day-old seedlings were used in each experiment. After treatment with 30µM DEX for 24 h, samples were harvested at 24 h after treatment with 0.5 mM SA. The ChIP assays were performed according to a previously described protocol (Komar et al., 2016). For pTA:TCP15-EYFP and 35S:GFP transgenic lines, chromatin was immunoprecipitated by PierceTM Protein A/G Magnetic Beads (ThermoFisher) bound with the anti-GFP antibody. For Dex:Flag-TCP15/Col-0 and Dex:Flag-TCP15/npr1-2 transgenic lines, chromatin was immunoprecipitated by beads bound with the anti-FLAG M2 antibody. Primers used in the ChIP assays are listed in **Supplementary Table 1**.

# RNA Extraction and Quantitative PCR

RNA was extracted with TRIzolTM (Invitrogen) according to its protocol. cDNA was synthesized using reverse transcriptase (Quanta). Real-time quantitative polymerase chain reaction (RT-qPCR) was performed with SYBR <sup>R</sup> Green (Quanta). The expression level of UBIQUITIN 5 (UBQ5) was used as an internal control. Three biological replicates were assayed. Primers for RT-qPCR are listed in **Supplementary Table 1**.

# SAR Assay

Two lower leaves of 3-week-old plants were hand-infiltrated with 10 mM MgSO<sup>4</sup> or avirulent pathogen Psm ES4326 carrying avrRpt2 (OD<sup>600</sup> = 0.02). Three days later, three upper leaves were hand-infiltrated with virulent pathogen Psm ES4326 (OD<sup>600</sup> = 0.001). Leaf samples for bacterial growth were collected at 3 days after the secondary infection. Six plants were used in each treatment.

# RESULTS

# TCP8, TCP14, and TCP15 Interact With NPR1

To test our hypothesis that NPR1 can interact with TFs other than TGAs, we performed Y2H screens (Ou et al., 2011) using Arabidopsis NPR1 as a bait and an Arabidopsis transcription factor library (Pruneda-Paz et al., 2014) as prey. Fifteen interactors were identified. To eliminate false positive interactions, we performed Y2H assays using individual candidate TFs as preys. Interactions were tested by the growth of diploid yeast cells on triple dropout medium lacking leucine, tryptophan, and histidine with 1 mM 3-AT. Supporting our initial assays, diploids containing NPR1 fused with a GAL4 DNA binding domain (BD-NPR1) and empty vector with a GAL4 activation domain (AD-EV) did not grow on selective plates (**Figure 1A**). Yeast diploids with BD-EV and AD-TFs were included as negative controls to exclude self-activation activity of TFs (**Figure 1A**). For this study, we focused on transcription factors in the TCP family. Yeast diploids containing BD-NPR1 and AD-TCP15 grew on selective plates (**Figure 1A**), indicating TCP15 was a true positive interactor of NPR1. We also found that yeast diploids containing BD-NPR1 with AD-TCP8 and AD-TCP14 grew on selective plates (**Figure 1A**), indicating NPR1 interacts with TCP8 and TCP14 in Y2H assays.

To test the association of NPR1 with TCP8, TCP14, and TCP15 in planta, we performed co-immunoprecipitation (Co-IP) assays in N. benthamiana. TCP8 fused with an N-terminal FLAG tag (FLAG-TCP8) and NPR1 fused with a C-terminal GFP tag (NPR1-GFP) were transiently co-expressed in N. benthamiana by agroinfiltration. NPR1-GFP proteins were coimmunoprecipitated with FLAG-TCP8 bound to the anti-FLAG magnetic beads (**Figure 1B**), indicating NPR1 associates with TCP8 in planta. NPR1-GFP and TCP14 fused with an Nterminal HA tag (HA-TCP14) (**Figure 1C**) or NPR1-GFP and HA-TCP15 (**Figure 1D**) were also transiently co-expressed in

proteins were analyzed by immunoblot using anti-HA and anti-GFP antibodies.

N. benthamiana by agroinfiltration. HA-TCP14 (**Figure 1C**) and HA-TCP15 (**Figure 1D**) proteins were co-immunoprecipitated with NPR1-GFP bound to the anti-GFP beads, indicating TCP14 and TCP15 are associated with NPR1 in planta. Together, results from Y2H and Co-IP assays indicate that TCP8, TCP14, and TCP15 physically interact with NPR1.

To investigate whether NPR1 affects the transcription of TCP8, TCP14, and TCP15, their mRNA levels in wild-type Col-0 and the npr1-2 mutant after 24 h with or without SA application were examined by RT-qPCR. The SA-induced increase of TCP8 mRNA levels was abolished in npr1-2, while the mRNA levels of TCP14 and TCP15 did not show a significant difference between npr1-2 and Col-0 with or without SA for 24 h (**Supplementary Figure 1**). Our results indicate that NPR1 does not affect the transcription of TCP14 and TCP15 at this later time point, and an increase in SA-induced TCP8 mRNA levels is dependent on NPR1.

# TCP8, TCP14, and TCP15 Contribute Redundantly to SAR Establishment

To investigate whether NPR1 interactors TCP8, TCP14, and TCP15 also function in SAR, we carried out SAR tests in the tcp8-1, tcp14-5, and tcp15-3 single, corresponding double, and triple mutants (Kim et al., 2014). Two lower leaves were infected with Psm ES4326 carrying avrRpt2 to induce SAR. Three days later, two upper leaves were infiltrated with the virulent pathogen Psm ES4326. The bacterial growth of Psm ES4326 in the tcp8-1, tcp14-5, and tcp15-3 single and corresponding double mutants decreased 8.6- to 12.8-fold which was similar to that reduction (11.7-fold) in Col-0 after SAR induction (**Figure 2A**). These results demonstrated that the deletion of one or two of TCP8, TCP14, or TCP15 genes is not enough to disrupt SAR. However, the Psm ES4326 population in tcp8- 1 tcp14-5 tcp15-3 (tcp8/14/15) only decreased 2.5-fold after

FIGURE 2 | TCP8, TCP14, and TCP15 contribute redundantly to SAR establishment. (A) Bacterial population in Col-0, *npr1-2*, and the *tcp* mutants. Two lower leaves of 3-week-old plants were infiltrated with 10 mM MgCl2 (-SAR) or *Pseudomonas syringae* pv. *maculicola* (*Psm*) ES4326 carrying *avrRpt2* at OD600 = 0.02 (+SAR). Three days later, two upper leaves were infiltrated with *Psm* ES4326 at OD600 = 0.001. Bacterial growth in infected systemic leaves was calculated 3 days after the second infection. Error bars represent SD of six biological repeats. Statistical analysis was studied by *t*-test (\**p* < 0.05, \*\*\**p* < 0.001) using Excel 2016. (B,C) The expression of *PR1*, *PR2*, and *PR5* in Col-0, *npr1-2*, and *tcp8/14/15*. Two local leaves were inoculated with 10 mM MgCl2 (Mock) or *Psm* ES4326 with *avrRpt2* at OD600 = 0.02 (AvrRpt2). The mRNA levels of *PR1*, *PR2*, and *PR5* in local leaves at 12 h post inoculation (hpi) (B) and in systemic leaves at 48 hpi (C) were analyzed by RT-qPCR. Values were normalized to the *UBQ5* mRNA levels. Error bars represent SD of three biological repeats. Statistical analysis was studied by two-way ANOVA following multiple comparisons with turkey test (95% confidence interval) using GraphPad Prism 7. Different small letters above the bars mean significant differences.

SAR induction (**Figure 2A**), indicating that SAR induction was partially compromised in tcp8/14/15. Consistent with bacterial growth results, Psm ES4326 infected systemic leaves of the single and double mutants showed less chlorosis compared with that of tcp8/14/15 after SAR induction (**Supplementary Figure 2**).

To examine whether the expression of SAR marker genes was affected, the mRNA levels of PR1, PR2, and PR5 in local and systemic leaves of Col-0, npr1-2, and tcp8/14/15 were examined by RT-qPCR assays. After locally treated with 10 mM MgCl2, the mRNA levels of PR1, PR2, and PR5 in both local and systemic leaves of Col-0, npr1-2, and tcp8/14/15 did not show a significant difference (**Figure 2B**), indicating TCP8, TCP14, and TCP15 did not affect the basal expression of PR1, PR2, and PR5. Whereas, after locally infected with Psm ES4326 with avrRpt2, the local and systemic induction of PR1, PR2, and PR5 were all obviously decreased in tcp8/14/15 compared with Col-0 (**Figures 2B,C**), consistent with the increased local susceptibility of tcp8/14/15 to avrRpt2-expressing DC3000 (Kim et al., 2014). Together, these results reveal that TCP8, TCP14, and TCP15 contribute redundantly to establish SAR either directly or indirectly.

# TCP15 Binds to the Promoter of PR5 at the TCP Binding Site

Since the reduction of mRNA levels of PR1, PR2, and PR5 in tcp8/14/15 may be a direct or indirect effect, we subsequently investigated whether TCP8, TCP14, and TCP15 can directly bind to the promoters of PR1, PR2, and PR5 through yeast one-hybrid (Y1H) assays. Constructs of pDEST22-TCPs were transformed into the yeast strain YU, and PR promoters constructed in the pLacZi vector were integrated into the yeast strain YM4271 (Pruneda-Paz et al., 2014). Healthy yeast diploids growing on double dropout medium lacking tryptophan and uracil were selected. The binding affinity of TFs to promoters was quantified by β-galactosidase activity shown as fold change over empty vector control. In addition, 3-fold induction was set as the cut-off (Zheng et al., 2015). Expression of any TCP8, TCP14, or TCP15 with PR1 and PR2 promoters fused with the LacZ reporter gene did not activate more than a 3-fold change of β-galactosidase activity (**Figure 3A**), demonstrating that these TCP TFs did not bind to the promoter of PR1 and PR2; however, expression of AD-TCP15 with the LacZ reporter gene fused PR5 promoter resulted in a 4.13**-**fold change of β-galactosidase activity compared with vector control (**Figure 3A**), demonstrating that TCP15 physically bound to the PR5 promoter and functioned as a transcriptional activator in yeast.

As a member of the class I TCP TFs, TCP15 was shown to bind to the consensus element KHGGGVC (Davière et al., 2014). Such a motif, GCGGGAC, was found in the promoter of PR5 at −788 bp from the transcription start site (TSS) (**Figure 3B**). To investigate whether TCP15 targets the PR5 promoter specifically at the TCP binding site (TBS), site mutations (GCGGGAC to ATAAACT) were introduced into the TBS (TBSm). The presence of TCP15 did not activate the expression of the LacZ reporter gene fused after the PR5 promoter with TBSm (**Figure 3C**). To further confirm the binding specificity, a shorter PR5 promoter (500 bp) without TBS was used. The presence of TCP15 did not activate reporter gene expression either (**Figure 3C**). These results indicate that the TBS within the promoter of PR5 is required for the binding of TCP15 in yeast.

To further confirm this specific binding in planta, we performed chromatin immunoprecipitation (ChIP) assays using DEX inducible TCP15 transgenic seedlings pTA:AtCP15-EYFP (Li et al., 2012). The relative enrichment of selected PR5 promoter regions was tested by RT-PCR. Primers were designed to amplify three PR5 promoter fragments a, b, and c, in which the b region contained the TBS (**Figure 3B**). Our data show that fragments a and c did not show obvious enrichment, while fragment b showed significant enrichment in the immunoprecipitated samples from pTA:AtCP15-EYFP compared with 35S:GFP (**Figure 3D**), demonstrating that TCP15 binds in vivo to the TBS within the promoter of PR5. Our results from Y1H assays and ChIP assays indicate that TCP15 activates the transcription of PR5 via binding to the TBS within its promoter.

# TCP15 Regulates the Expression of PR5

To investigate whether TCP15 affects the expression of PR5, the mRNA levels of PR5 in Col-0, npr1-2, and the two TCP15 T-DNA insertion lines tcp15-1 (Kieffer et al., 2011) and tcp15-3 were analyzed after SA treatment using RT-qPCR assays. The SAinduced PR5 transcript level was significantly compromised in tcp15-1 and tcp15-3 compared with Col-0 (**Figure 4A**), indicating that TCP15 is required for SA-induced PR5 expression.

To further confirm that TCP15 promotes the expression of PR5, the mRNA levels of PR5 in Col-0, npr1-2, and two DEX inducible TCP15 transgenic lines were analyzed after SA treatment using RT-qPCR assays. Since most constitutive overexpression TCP15 transgenic lines showed growth arrest (Li et al., 2012), Arabidopsis transgenic lines expressing FLAG-TCP15 under the control of a DEX-inducible promoter in the Col-0 background (DEX:Flag-TCP15/Col-0) were generated. Two T3 homozygous transgenic lines, #6 and #10, which exhibited different FLAG-TCP15 protein levels after DEX treatment, were used (**Figure 4B**). Before SA treatment, the presence of TCP15 in both transgenic lines significantly increased the mRNA levels of PR5 (**Figure 4C**). With SA treatment, the presence of TCP15 increased the PR5 mRNA levels to a dramatically higher level (**Figure 4C**), indicating that the upregulation of PR5 by TCP15 was promoted by SA. Taken together, these results indicate that TCP15 positively regulates the expression of PR5.

# NPR1 Facilitates TCP15 Binding to the PR5 Promoter

Without a DNA binding domain, NPR1 functions as a transcriptional co-activator to facilitate TGA TFs binding to the promoter of PR1 by interacting with TGA TFs (Zhang et al., 1999; Fan and Dong, 2002; Johnson et al., 2003). To test whether NPR1 can enhance TCP15 binding to the PR5 promoter, we performed ChIP assays using Arabidopsis transgenic lines expressing FLAG-TCP15 under the control of a DEX-inducible promoter in the Col-0 background and npr1-2 background (DEX:FLAG-TCP15/npr1-2). The DEX:FLAG-TCP15/Col-0

TCP binding site with mutations. (C) Interactions between TCP15 with the *PR5* promoter containing a TCPm and the *PR5* promoter (500 bp) in Y1H assays. Error bars represent SD of three repeats. (D) Association between TCP15 and the *PR5* promoter was studied by chromatin immunoprecipitation (ChIP) assays combined with RT-PCR analysis. Twelve-day-old seedlings of *pTA:AtTCP15-EYFP* and *35S:GFP* transgenic plants were treated with 30µM DEX for 24 h. Samples were collected at another 24 h after 0.5 mM SA induction. The enrichment of *PR5* promoter DNA in immunoprecipitated samples was shown as % input. Error bars represent SD of three technical repeats. Statistical analysis was studied by *t*-test (\**p* < 0.05) using Excel 2016. Another independent repeat showed similar results.

transgenic lines #6 and #10 and DEX:FLAG-TCP15/npr1-2 transgenic lines #13 and #17, which expressed similar Flag-TCP15 protein levels after DEX application (**Figure 5A**), were used. FLAG-TCP15 proteins and their cross-linked DNA were immuno-precipitated from chromatin extracts using beads bound with the anti-FLAG antibody. Immuno-precipitated DNA was analyzed by RT-PCR using primers amplifying fragment b described in **Figure 3B**. Fragment b showed significantly higher enrichment in DEX:FLAG-TCP15/Col-0 compared with DEX:FLAG-TCP15/npr1-2 (**Figure 5B**), demonstrating that NPR1 facilitates TCP15 binding to the promoter of PR5 in Arabidopsis.

# DISCUSSION

In this work, we provide clear evidence that TCP8, TCP14, and TCP15 physically interact with NPR1 and coordinately contribute to SAR establishment, that TCP15 positively regulates the expression of PR5 by directly binding to the TBS within the PR5 promoter, and that

*PR5* in Col-0, *npr1-2*, *tcp15-1*, and *tcp15-3*. Twelve-day-old seedlings were treated with water (Mock) or 0.5 mM SA (SA) for 24 h. The mRNA levels of *PR5* were analyzed by RT-qPCR. Values were normalized to the *UBQ5* mRNA levels, then to the value of Col-0 (mock), which was arbitrarily set at 1. Statistical analysis was studied by two-way ANOVA following multiple comparisons with turkey test (95% confidence interval) using GraphPad Prism 7. (B) FLAG-TCP15 protein levels in DEX inducible transgenic lines. Two T3 homozygous lines expressing *FLAG-TCP15* under control of a DEX-inducible promoter in the Col-0 background (*DEX:FLAG-TCP15*/Col-0) were used. Twelve-day-old seedlings were treated with either 0.01% ethanol (–DEX) or 30µM DEX (+DEX) for 24 h. Protein levels were measured by immunoblot with the anti-FLAG antibody. Rubisco large subunits stained with *(Continued)* FIGURE 4 | Ponceau S were used to show equal total protein loading. (C) The expression of *PR5* in Col-0 and two *DEX:FLAG-TCP15*/Col-0 lines. Twelve-day-old seedlings described in (B) were treated with either 0.5 mM SA (+SA) or water (–SA) for 24 h after being treated with 0.01% alcohol (-DEX) or 30µM DEX (+DEX) for 24 h. The *PR5* mRNA levels were examined by RT-qPCR. Values were normalized to the *UBQ5* mRNA levels, then to the value of Col-0 (–DEX –SA) which was arbitrarily set at 1. Error bars represent SD of three biological replicates. Statistical analysis was studied by two-way ANOVA following multiple comparisons with turkey test (95% confidence interval) using GraphPad Prism 7. Different small letters above the bars mean significant differences.

FIGURE 5 | NPR1 enhances TCP15 binding to the *PR5* promoter. (A) FLAG-TCP15 protein levels in DEX inducible transgenic lines. Twelve-day-old seedlings of T3 homozygous lines expressing *FLAG-TCP15* under the control of a DEX-inducible promoter in the Col-0 and *npr1-2* background (*DEX:FLAG-TCP15*/*npr1-2*) were treated with 30µM DEX for 24 h. Protein levels were measured by immunoblot with the anti-FLAG antibody. (B) The binding ability of TCP15 to the *PR5* promoter in transgenic plants described in (A) were examined with ChIP assays combined with RT-PCR analysis. Twelve-day-old seedlings were treated with 30µM DEX, then treated with 0.5 mM SA after 24 h. Samples were collected 24 h after SA treatment. Primers amplifying fragment b described in Figure 3B were used in RT-PCR. The enrichment of *PR5* promoter DNA in immunoprecipitated samples was shown as % input. Error bars represent SD of three technical replicates. Statistical analysis was studied by two-way ANOVA following multiple comparisons with turkey test (95% confidence interval) using GraphPad Prism 7. Different small letters above the bars mean significant differences. Another independent replication demonstrated similar results.

NPR1 can enhance this binding. Our results suggest that in addition to acting as mediators of SA biosynthesis (Wang et al., 2015), TCP TFs are also essential players in the SA signaling pathway, and that TCP proteins not only contribute to ETI (Kim et al., 2014), but also function in SAR.

Taking advantage of the comprehensive Arabidopsis transcription factor library (Pruneda-Paz et al., 2014), TCP15 was identified as a novel NPR1 interactor through Y2H screens. Interactions between NPR1 and all the 24 Arabidopsis TCP proteins were also tested using Y2H assays, and more interactors were identified (unpublished data). As a critical transcriptional regulatory node, NPR1 directly regulates the expression of 2,248 SA-responsive genes (Wang et al., 2006). The absence of a DNA binding domain and the presence of two protein-protein interaction domains suggest NPR1 regulates the expression of downstream genes through interactions with TFs (Fu and Dong, 2013). Apparently, the known interactions between NPR1 and TGAs are not sufficient to explain the expression of all the genes regulated by NPR1. TCP proteins, as novel NPR1 interactors, will be excellent candidates to fill these missing links.

TCP proteins were reported to mediate SA biosynthesis by coordinately regulating the expression of ICS1, which encodes an enzyme responsible for pathogen-induced SA biosynthesis (Wang et al., 2015). This suggests that TCP proteins can affect the activity of NPR1. Although one conserved TCP binding motif was located at −169 bp from the TSS within the promoter of NPR1, no direct interaction was found between the NPR1 promoter and TCP proteins in yeasts (**Supplementary Figure 3**), suggesting that TCP proteins could affect NPR1 activity indirectly. Because TCP8, TCP14, and TCP15 can interact with each other, and one characteristic of TCP TFs is their functional redundancy (Kim et al., 2014), it is not surprising that SAR could still be induced in the tcp8-1, tcp14-5, and tcp15-3 single or corresponding double mutants (**Figure 2A**). SAR was abolished in the npr1-2 mutants, while it was partially compromised in the tcp8/14/15 triple mutants (**Figure 2A**), suggesting that other TFs are required in NPR1-mediated SAR signaling pathway. These TFs could be TGA proteins or other NPR1-interacting TCP proteins.

In addition, failed SAR induction could be caused by decreased initial immunity signal production and/or blocked mobile signal transduction. The expression of SAR marker genes (PR1, PR2, and PR5) in both local and systemic leaves were all significantly decreased after local infection with an avirulent pathogen (**Figures 2B,C**), consistent with a reduced initial immune signal production in tcp8/14/15 plants (Kim et al., 2014). Our finding that TCP15 directly promotes the expression of PR5 (**Figures 4A,B**) supports this explanation. As TCP8 and TCP14 did not bind the promoter of PR5 in yeast, they may facilitate TCP15 binding to the promoter of PR5 through a complex formation. Since TCP8, TCP14, and TCP15 did not show direct interactions with the promoter of PR1 and PR2 in Y1H assays (**Figure 3A**), the decrease of PR1 and PR2 may be indirectly regulated by these TCP TFs; however, whether TCP8, TCP14, and TCP15 are involved in SAR by mediating the activity of a mobile signal is unknown and will be an exciting topic to study in the future. Because PR1 and PR2 promoters both have TGA binding sites, it is possible that either TCP8, TCP14, and TCP15 interact with TGA transcription factors or the interactions between NPR1 and TCP8/14/15 promote the interactions between NPR1 and TGAs, to facilitate the expression of PR1 and PR2 to establish SAR.

How TCP15 responses to SA signaling and bind to the promoter of PR5 is unclear. Based on the activity of TCP15 is dependent on redox modulation (Viola et al., 2013, 2016), the molecular mechanism of how SA induces the TCP15-dependent PR5 expression is proposed. Under oxidizing conditions, the DNA binding activity of TCP15 is inhibited by dimers formed with disulfide bonds, while under reducing conditions induced by SA accumulation, the inhibition is reversed, resulting in extensively upregulated expression of PR5.

Although NPR1 could enhance TCP15 binding to the promoter of PR5 (**Figure 5B**), the exact molecular mechanism through which NPR1 facilitates this association is unclear and will be an interesting topic to investigate. It is perhaps significant in this context that NPR1 activity is also regulated by SUMOylation-induced complex formation (Saleh et al., 2015), and TCP8, TCP14, and TCP15 were recently shown to associate with the nuclear SUMOylation machinery and to be SUMOylated as well (Mazur et al., 2017).

# AUTHOR CONTRIBUTIONS

ML, HC, JC, MC, WG, FL, and ZF contributed conception and design of the study. ML collected the data. ML, HC, WG, JC, and ZF organized the data. ML performed the statistical analyses. ML, IP, WG, and ZF wrote the manuscript. All authors contributed to manuscript revision, and have read and approved the submitted version.

# FUNDING

This work is supported by a SPARC Graduate Research Grant (2017-2018) sponsored by the University of South Carolina (to ML) and the National Science Foundation (grant IOS-1758994 to ZF and grant IOS-1456181 to WG).

# ACKNOWLEDGMENTS

We thank Ai-wu Dong for providing the seeds of pTA:TCP15- EYFP and 35:GFP transgenic lines; Li-jia Qu for providing the pDEST22 empty vector; and Steve A. Kay for providing the pLacZi plasmid, yeast strain YM4271 and YU. We thank Johannes Stratmann, Beth Krizek, Hexin Chen, and Kim E. Creek for working as the committee members of ML, and for providing suggestions about the experimental designs used in this study. We thank the Chinese Scholarship Council for providing a fellowship to ML.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018. 01153/full#supplementary-material

# REFERENCES


required for salicylic acid induction of the PR-1 gene. Proc. Natl. Acad. Sci. U.S.A. 96, 6523–6528. doi: 10.1073/pnas.96.11.6523

Zheng, X. Y., Zhou, M., Yoo, H., Pruneda-Paz, J. L., Spivey, N. W., Kay, S. A., et al. (2015). Spatial and temporal regulation of biosynthesis of the plant immune signal salicylic acid. Proc. Natl. Acad. Sci. USA. 112, 9166–9173. doi: 10.1073/pnas.1511182112

**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, Chen, Chen, Chang, Palmer, Gassmann, Liu and Fu. 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.

# The Tryptophan decarboxylase 1 Gene From Aegilops variabilis No.1 Regulate the Resistance Against Cereal Cyst Nematode by Altering the Downstream Secondary Metabolite Contents Rather Than Auxin Synthesis

#### Edited by:

Yi Li, Peking University, China

### Reviewed by:

Laurence Veronique Bindschedler, Royal Holloway, University of London, United Kingdom Yule Liu, Tsinghua University, China

#### \*Correspondence:

Maoqun Yu yumaoqun@cib.ac.cn; yumq@cib.ac.cn Haili Zhang zhanghl@cib.ac.cn

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 13 June 2018 Accepted: 17 August 2018 Published: 04 September 2018

#### Citation:

Huang Q, Li L, Zheng M, Chen F, Long H, Deng G, Pan Z, Liang J, Li Q, Yu M and Zhang H (2018) The Tryptophan decarboxylase 1 Gene From Aegilops variabilis No.1 Regulate the Resistance Against Cereal Cyst Nematode by Altering the Downstream Secondary Metabolite Contents Rather Than Auxin Synthesis. Front. Plant Sci. 9:1297. doi: 10.3389/fpls.2018.01297 Qiulan Huang1,2,3, Lin Li<sup>4</sup> , Minghui Zheng<sup>4</sup> , Fang Chen<sup>2</sup> , Hai Long<sup>1</sup> , Guangbing Deng<sup>1</sup> , Zhifen Pan<sup>1</sup> , Junjun Liang<sup>1</sup> , Qiao Li<sup>1</sup> , Maoqun Yu<sup>1</sup> \* and Haili Zhang<sup>1</sup> \*

<sup>1</sup> Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu, China, <sup>2</sup> College of Life Sciences, Sichuan University, Chengdu, China, <sup>3</sup> University of the Chinese Academy of Sciences, Beijing, China, <sup>4</sup> School of Basic Medical Sciences, Zunyi Medical University, Zunyi, China

Cereal cyst nematode (CCN, Heterodera avenae) is a most important pathogen of wheat and causes tremendous yield loss annually over the world. Since the lack of resistance materials among wheat cultivars, identification and characterization of the resistance-related genes from the relatives of wheat is a necessary and efficient way. As a close relative of wheat with high resistance against CCN, Aegilops variabilis No.1 is believed to be a valuable source for wheat breeding against this devastating disease. However so far, very few resistance-associated genes have been characterized from this species. In this study, we present that the tryptophan decarboxylase genes from Ae. variabilis No.1 (AeVTDC1 and AeVTDC2) were both induced by CCN juveniles at the early stage of resistance response (30 h post-inoculation), with AeVTDC1 more sensitive to CCN infection than AeVTDC2. Silencing of AeVTDC1 led to compromised immunity to CCN with more CCN intrusion into roots; while overexpression AeVTDC1 in Nicotiana tabacum dramatically enhanced the resistance of plants by reducing the knots formed on roots. Metabolism analysis showed that the contents of secondary metabolites with activity of resistance to varied pathogens correlated with the expression level of AeVTDC1 in both Ae. variabilis No.1 and the transgenic tobacco plants. In addition, the content of IAA was not affected by either silencing or overexpressing of AeVTDC1. Hence, our research provided AeVTDC1 a valuable target that mediates resistance to CCN and root knot nematode (RKN, Meloidogyne naasi) without influencing the auxin biosynthesis.

Keywords: cereal cyst nematode, Aegilops variabilis No.1, tryptophan decarboxylase, secondary metabolite, indole acetic acid

# INTRODUCTION

fpls-09-01297 October 10, 2019 Time: 18:22 # 2

The cereal cyst nematode (CCN, Heterodera avenae) is a vital pathogen of graminaceous crops, such as wheat and barley. CCN is widely distributed and causes great production losses (Williamson and Kumar, 2006; Hajihasani et al., 2010; Zheng et al., 2015; Li et al., 2016). Many efforts have been made to identify CCN resistance (Cre) genes. However, gene resource resistant to CCN is scarce in wheat but abundant in its relatives (Montes et al., 2008). Rha genes were mapped in barley. Through changing the transcript abundance and composition of cell wall, Rha2-mediated CCN resistance drives rapid deterioration of CCN feeding sites (Aditya et al., 2015). CreX and CreY were identified in Aegilops variabilis (Barloy et al., 2006). Some resistance lines were bred through hybridization. Although several loci related to CCN resistance have been reported, few of them has been cloned and their biological functions hadn't been clarified (de Majnik et al., 2003; Safari et al., 2005; Zhang et al., 2016). Ae. variabilis No.1 (2n = 4x = 28, UUS<sup>v</sup> S v ), belonging to the genus Aegilops of the Triticeae tribe, is known as a well-resistant material, which confers strong resistance against CCN and root knot nematode (RKN, Meloidogyne naasi) (Barloy et al., 2006; Coriton et al., 2009; Xu et al., 2012; Zheng et al., 2015; Wu et al., 2016).

Plant tryptophan decarboxylase (TDC, 4.1.1.28) catalyzes the formation of tryptamine from tryptophan (Trp) (Giebel and Jackowiak, 1976; Hallard et al., 1997; Canel et al., 1998). Tryptamine is a precursor for the biosynthesis of serotonin, indole alkaloids, and indole acetic acid (IAA) (Bartel, 1997; Valletta et al., 2010; Liu et al., 2012; Dubouzet et al., 2013). Transformation from tryptamine to serotonin is catalyzed by Tryptamine-5-hydroxylase (T5H) (Kang and Back, 2009). TDCs are key enzymes in the biosynthetic pathway of terpenoid indole alkaloids (TIAs), since they link primary to secondary metabolism by converting Trp into tryptamine. Conversion of Trp into tryptamine is a common backbone for many secondary metabolites, which have divergent biological activities regulated by developmental and environmental factors (Mehrotra et al., 2013; Verma et al., 2015). Plant TDC cDNA was firstly isolated from Catharanthus roseus by screening a cDNA expression library (Hallard et al., 1997; Geerlings et al., 1999). Gradually, a few TDC genes had been cloned and characterized from other species, such as Nicotiana tabacum, Mitragyna speciosa, and Withania coagulans (Berlin et al., 1993; Goddijn et al., 1994; Charoonratana et al., 2013; Jadaun et al., 2017). The biological functions of TDCs have been reported in several plant–pathogen interactions. It was reported that TDC gene played a role in resistance against Malacosoma disstria Hub and Manduca sexta L. through tryptamine, which had adverse effects on their feeding behaviors and physiology (Gill and Ellis, 2006). Serotonin defended against Magnaporthe oryzae infection in rice leaves (Gill et al., 2003; Hayashi et al., 2016). Ectopic expression of TDC1 significantly suppressed the growth of insect pests by sufficient tryptamine accumulation in poplar and tobacco leaf tissue (Gill et al., 2003). The inhibition of TDC enzyme activity with S-αFMT resulted in susceptibility of Ae. variabilis No.1 to CCN and RKN, which indicated AeVTDCs play important roles in resistance to nematodes (Li et al., 2016). However, it remains to be clarified which AeVTDC involved in resistance to CCN and RKN and its mechanism of function.

Previous RNA-Seq analysis indicated that the TDC genes of Ae. variabilis No.1 (AeVTDCs) showed different expression pattern between control and CCN invaded roots at different time points (Zheng et al., 2015). AeVTDC1 gene was cloned and its protein had the ability of catalyzing the formation of tryptamine from tryptophan. In this study, we reported AeVTDC1 played a positive role at the early stage of plant resistance to CCN infection and overexpression of AeVTDC1 in tobacco led to reduced susceptibility to RKN. Silencing or overexpression of AeVTDC1 didn't affect accumulation of IAA, but changed the downstream secondary metabolites of AeVTDC1.

# MATERIALS AND METHODS

# Plant Materials and Culture Condition

Seeds of Ae. variabilis No.1 and wheat (Fielder) were surfacecleaned by the sterilized water and kept at 4◦C for 24 h. Then the seeds were germinated in Petri dishes (5-cm diameter) on wet paper at 20◦C under a 16-h light/8-h dark photoperiod. After 2 days, these little seedlings plants were cultured with water or sterilized soil for later use.

AeVTDC1 transgenic tobacco and wild type (WT) were cultured in the sterilized soil in greenhouse under 25◦C and 60% humidity.

# Nematode Hatching, Inoculation, and Staining

Second stage juveniles (J2s) of CCN (H. avenae) were hatched with cereal cysts as described previously (Li et al., 2016). For resistance assay and expression inducing assay, about 300 J2s of CCN per pot were inoculated around the root-tips under 19◦C. According to the Li's description, the eggs and second stage juveniles (J2s) of RKN were obtained (Li et al., 2016). For root nematode staining, entire root systems were collected and washed cleanly at 3 days after inoculation (DAI), and immersed in the solution (5% NaClO) for 5 min. Then roots were soaked in tap water for 15 min to remove residual NaClO. The roots were rinsed in boiling stain solution (acid fuchsin, 0.5 g/L) for 30 s, washed with tap water, then placed in 30 mL glycerin acidified with few drops of 5 mol/L HCl for water bath heating 30 s. The dyed roots were kept in glycerin for storage at 4◦C. The number of visible pink-stained nematodes present within the roots was counted under a light microscope (Leica, DM3000 LED). At least nine replicate samples (individual plants) were counted for each treatment (Aditya et al., 2015).

# RNA Extraction, Reverse Transcription, and Quantitative Real-Time PCR (QPCR)

To detect gene expression after CCN infection, two-leaf stage seedlings were cultured in sterilized soil (four seedlings each pot),

and inoculated with nematodes (300 J2s/pot). The root tissues were respectively collected at 0, 30 h and 3, 9 days. All these samples were stored at −80◦C for RNA extraction.

Total RNA was extracted using the TRIzol-A<sup>+</sup> reagent (Tiangen, Beijing, China) according to the manufacturer's instructions. The cDNA synthesis was carried out using the ReverTra Ace qPCR RT Kit (TOYOBO). QPCR was conducted as described (Wang et al., 2013). Gene-specific primers were designed to amplify PCR products about 100–200 bp, and listed in **Supplementary Table 1**. Elongation factor1-α (EF1α) mRNA was employed as an internal control for normalization (Livak and Schmittgen, 2001). Each sample or treatment was tested in three biological repeats and experiment was performed for three times. The differences were analyzed by t-test and data were presented by software Origin 8.6.

# Barley Stripe Mosaic Virus (BSMV)-Mediated Gene Silencing

The plasmids used for gene silencing were constructed as previously described (Holzberg et al., 2002). 751∼951 bp of AeVTDC1 coding sequence (ORF) was cloned and ligated to BSMV γ plasmid through NheI to construct plasmid for silencing AeVTDC1. Primers used were listed in **Supplementary Table 1**.

Infectious BSMV RNA was prepared from each linearized plasmid (α and γ digested with MluI, β digested with SpeI) by in vitro transcription using a Large Scale RNA Production System (T7 RiboMAXTM Express Large Scale RNA Production System). The BSMV inoculum was made by combining an equimolar ratio of α, β, and γ transcripts with excess inoculation buffer containing a wounding agent (GKP buffer: 50 mM glycine, 30 mM dipotassium phosphate, pH 9.2, 1% bentonite, 1% celite) as previously described (Holzberg et al., 2002). The second leaves of two-leaf seedlings were inoculated with BSMV inoculum. BSMV γ empty vector were used as negative controls.

Barley stripe mosaic virus-treated plants were kept in a cultivation chamber at 25◦C with 60% humidity. When the virus phenotype was observed (about 10 days after BSMV inoculation), new roots of these plants were sampled and used for RNA isolation. The silencing efficiency for the target gene and expression levels of other related genes compared with control were examined by QPCR. The primers for QPCR were listed in **Supplementary Table 1**.

# Overexpression of AeVTDC1 in Tobacco and RKN Resistance Assay

PCAMBIA1300-based T-DNA vector was chosen as the skeleton and hygromycin was replaced with bar gene. CaMV35S promoter and NOS terminator were amplified using pJG045 as a template to drive and terminate gene expression (Zhao et al., 2013). For generation of the overexpression construct, the ORF of AeVTDC1 was PCR amplified using template with primers (**Supplementary Table 1**) (Li et al., 2016). AeVTDC1 ORF was fused to N terminal of yellow fluorescent protein (YFP) sequence and together inserted into the modified binary vector to express AeVTDC1-YFP fusion protein. The insertion sequences were confirmed by nucleotide sequencing. Overexpression construct was introduced into Agrobacterium tumefaciens strain EHA105 for tobacco transformation (Horsch et al., 1985). Cultivar tobacco (Mammoth Gold) was used for transformation. Positive transformants and their offsprings were screened by PCR with specific primers of AeVTDC1. Stable lines (L120 and L133) were selected for QPCR analysis of expression of genes and RKN resistance assay.

In resistance assay, L120, L133 and WT seeds were simultaneously germinated and cultured with sterilized soil under 25◦C. Six-leaf seedlings were transplanted into soil containing RKN. The root knots were calculated and analyzed after 8 weeks. No less than 15 individuals each line were used for counting. The photographs were taken by camera (Canon).

# IAA Quantitative Analysis

The IAA in the roots were measured by UPLC–MS/MS as previously described (Fu et al., 2012).

# Isolation of Secondary Metabolites and Analysis by UPLC–MS/MS

All fresh roots were collected and freeze-dried for metabolites profiling. Three replicates were used in each treatment. The method was slightly modified according to the description (Chen et al., 2013). Each 0.1 g sample was powdered in liquid nitrogen using a mortar and pestle before 1 mL of extraction solution (70% aqueous methanol) was added, and the mixture was stored overnight in the dark at 4◦C. The mixture was then centrifuged at 4◦C at 10,000 g for 10 min, and each supernatant was filtered through a 0.22-µm Millipore filter before HPLC–MS/MS analysis.

### HPLC Conditions

The sample extracts were analyzed using an LC–ESI–MS/MS system (HPLC, Shim-pack UFLC SHIMADZU CBM30A system<sup>1</sup> ; MS, Applied Biosystems 6500 Q TRAP<sup>2</sup> ). The analytical conditions were as follows, HPLC: column, Waters ACQUITY UPLC HSS T3 C18 (1.8 µm, 2.1 mm × 100 mm); solvent system, water (0.04% acetic acid): acetonitrile (0.04% acetic acid); gradient program, 100:0 V/V at 0 min, 5:95 V/V at 11.0 min, 5:95 V/V at 12.0 min, 95:5 V/V at 12.1 min, 95:5 V/V at 15.0 min; flow rate, 0.4 ml /min; temperature, 40◦C; injection volume: 2 µl. The effluent was alternatively connected to an ESI-triple quadrupole-linear ion trap (Q TRAP)–MS.

## ESI–Q TRAP–MS/MS

LIT and triple quadrupole (QQQ) scans were acquired on a triple quadrupole-linear ion trap mass spectrometer (Q TRAP), API 6500 Q TRAP LC/MS/MS System, equipped with an ESI Turbo Ion-Spray interface, operating in a positive ion mode and controlled by Analyst 1.6 software (AB Sciex). The ESI source operation parameters were as follows: ion source, turbo spray; source temperature 500◦C; ion spray voltage (IS) 5500 V; ion source gas I (GSI), gas II (GSII), curtain gas (CUR) were set at 55, 60, and 25.0 psi, respectively; the collision gas (CAD) was

<sup>1</sup>www.shimadzu.com.cn/

<sup>2</sup>www.appliedbiosystems.com.cn/

high. Instrument tuning and mass calibration were performed with 10 and 100 µmol/L polypropylene glycol solutions in QQQ and LIT modes, respectively. QQQ scans were acquired as MRM experiments with collision gas (nitrogen) set to 5 psi. DP and CE for individual MRM transitions were done with further DP and CE optimization. A specific set of MRM transitions were monitored for each period according to the metabolites eluted within this period.

# RESULTS

# Expression of AeVTDC1 Was Strongly Induced by CCN Infection

Previous RNA-seq results indicated transcripts of several AeVTDCs were much higher in CCN-treated roots than that in non-treated roots. And primary results revealed inhibition of AeVTDC enzyme activity suppressed resistance to CCN (Li et al., 2016). To determine which member of AeVTDC family mainly contribute to CCN resistance, expression pattern of AeVTDC1 and AeVTDC2 were tested and compared at 0 hour (h), 30 h, 3 day (d), 9 day post-inoculation (dpi) of CCN. At 0 hpi, the expression level of the AeVTDC1 and AeVTDC2 were similar in the CCN-treated and control roots. At 30 hpi, the expression of AeVTDC1 was strikingly induced in CCN-treated roots, which is almost sixfold of that in control sample. Expression of AeVTDC2 was also induced and about twofold to that in control root. At 3 and 9 dpi, expressions of AeVTDC1 remained induced and were nearly threefold of that in control sample. And expressions of AeVTDC2 were still higher but less than two times of that in control sample (**Figure 1**). These results indicated that expression of AeVTDC1 was much more sensitive to CCN infection than AeVTDC2. Hence, AeVTDC1 was chosen as the candidate gene for the resistance assay.

# Silencing of AeVTDC1 Compromised Resistance to CCN in Ae. variabilis No.1

Virus induced gene silencing is an efficient and fast technology in gene function analysis (Holzberg et al., 2002; Liu et al., 2002). BSMV based silencing system has been widely applied in monocots (Pacak et al., 2010). In our study, we found the expression of AeVTDC1 was strikingly induced at 30 hpi, which is almost sixfold of that in control sample. To directly investigate whether AeVTDC1 gene participates in resistance to CCN in Ae. variabilis No.1, we utilized BSMV-mediated gene silencing to knock down expression of AeVTDC1 in roots and observed CCN infection. Expression of AeVTDC1 in roots of AeVTDC1-silenced plants was about 30% of that in vector control (**Figure 2C**), which indicated VIGS worked well in the roots. Furthermore, the expression of AeVTDC2 in roots of AeVTDC1-silenced plants was similar with that in vector control (**Supplementary Figure 1**). The number of dyed CCN J2 in the roots of AeVTDC1-silenced plants increased more than 45% compared with that of vector control plants at 3 dpi (**Figure 2D**). The data was collected from at least nine replicates each time, and the experiment was operated three times. This result showed silencing AeVTDC1 in Ae. variabilis No.1 plants compromised resistance to CCN at the early stage and indicated AeVTDC1 regulates early immune responses to CCN.

# Silencing of AeVTDC1 Changed the Contents of Secondary Metabolites and Expression of Related Genes

To analyze in more details of how AeVTDC1 regulated resistance to CCN in Ae. variabilis No.1, the changes of secondary metabolite and expression of related genes were detected in the roots of AeVTDC1-silenced plants.

Specific primers were designed to analyze the expression of genes downstream of TDC. Tryptamine-5-hydroxylase (T5H) catalyzes tryptamine transform into serotonin. Acetylserotonin O-methyltransferase (ASMT) catalyzes the synthesis of melatonin (Back et al., 2016). Methyl easterase (MES) is a key gene in the synthesis pathway of indole alkaloids. In the AeVTDC1 silenced plant roots, the expression of T5H, ASMT, and MES gene markedly declined comparing with vector control (**Figure 3A**). However, the content of tryptamine and melatonin had no significant changes in the AeVTDC1-silenced plant roots, and the content of serotonin might be too low to be detected both in AeVTDC1-silenced and vector control plants (**Figure 3B**). Indole and 3-Indolebutyric acid content slightly reduced. Caffeic acid 3-O-methyltransferase (COMT) methylates caffeic acid and 5-hydroxyferulic acid respectively to form ferulic acid and sinapic acid (Doorsselaere et al., 2010; Back et al., 2016). QPCR results displayed that the expression of COMT gene was unchanged (**Figure 3A**), but the content of caffeic acid, ferulic acid, and sinapic acid obviously decreased in the roots of AeVTDC1-silenced plants compared with vector control plants. Furthermore, the derivatives of serotonin, derivatives of tryptamine, and ferulic acid derivatives sharply declined in AeVTDC1-silenced plants. N-Feruloylserotonin went down 17 times. N-Feruloyltryptamine, N-Feruloylputrescine, and 3-O-Feruloylquinic acid also had significant reduction (**Figure 3B**). The results showed that silencing of AeVTDC1 gene altered the profile of downstream metabolin. The changes of metabolite (tryptamine derivatives, serotonin derivatives, ferulic acid and its derivatives) might be an important aspect for AeVTDC1 to regulate resistance to CCN in Ae. variabilis No.1.

# Accumulation of IAA, Transcripts of Its Biosynthetic and Signaling Genes Were Unaffected by Silencing of AeVTDC1

TDC catalyzes conversion from tryptophan to tryptamine, and tryptamine is a precursor for the biosynthesis of IAA (Dubouzet et al., 2013). Except for regulation of plant development, IAA also involves in plant defense to biotic stress and abiotic stress (Mathesius, 2010; Navarro, 2016).

To understand whether IAA biosynthesis and signaling pathway were altered, we analyzed the transcripts changes of biosynthetic genes and signaling genes of IAA when AeVTDC1 was silenced. QPCR results showed there were no

significant differences in the expression levels of biosynthesis genes (indole-3-acetaldehyde oxidase, AAO2; indole-3 pyruvate monooxygenase, YUCCA; nitrilase 2, NIT2; aldehyde dehydrogenase, ALDH2B) and signaling genes (small auxinupregulated RNA, SAUR15) in AeVTDC1-silenced roots and control vector roots (**Figures 4A,B**). Liquid chromatography– mass spectrometry (LC–MS) data further attested the level of IAA in the roots of AeVTDC1-silenced plants was similar with that in vector control plants (**Figure 4C**). These results demonstrated silencing of AeVTDC1 had no effect on expression of biosynthetic and signaling genes of IAA and IAA accumulation in Ae. variabilis plants. There might be other factors related to the resistance reduction in the AeVTDC1-silencing plants.

# Overexpression of AeVTDC1 in Tobacco Improved Resistance to RKN

To test whether AeVTDC1 involved resistance to other pathogen, AeVTDC1 was overexpressed in tobacco. Stable AeVTDC1 transgenic tobacco plants (L120 and L133) were obtained to monitor the susceptibility to RKN. AeVTDC1 transgenic plants and WT plants with similar growth state were planted in the homogenous soil containing RKN. Two months later, all root tissues of plants were taken out to count the number of root knots and photographed (**Figure 5A**). The knots formed on transgenic plants were much smaller than that on WT. Statistical data revealed that the number of knots formed on the transgenic plants was much less than that on WT (**Figure 5B**). The results demonstrated ectopic expression of AeVTDC1 in tobacco enhanced defense to RKN and led to reduction of root knots.

To further analyze which downstream products of TDC improved resistance to RKN in transgenic tobacco, the roots of the most resistant line (L120) and WT were respectively collected for basic metabolite profile and related gene expression analysis. We found that the expression of NtTDC was similar between L120 and WT (**Figure 5C**). And downstream genes and substances in tryptophan metabolism were also detected. The content of tryptamine showed no difference in the transgenic tobacco comparative with WT (**Figure 5D**). The tryptamine may be easily changed to tryptamine derivatives (Nhydroxytryptamine, N-Acetyltryptamine, and 5-Methoxy-N,Ndimethyltryptamine) in the tobacco, which showed obvious increase in L120. Serotonin accumulation strikingly increased in the roots of L120, almost eightfold greater than that of WT plants (**Figure 5D**), and expression of T5H was also obviously upregulated in the roots of L120 (**Figure 5C**). The content of IAA and melatonin had no obvious difference (**Figure 5D**), and the biosynthesis gene ASMT of melatonin similarly had no change, while IAA biosynthesis gene YUCCA was downregulated compared to the WT (**Figure 5C**). Moreover, we discovered that the expression of COMT and the content of ferulic acid and ferulic acid derivatives (1-O-Feruloylquinic acid, 3-O-Feruloylquinic acid glucoside, 5-O-Feruloylquinic acid glucoside) were higher in transgenic tobacco than in WT (**Figures 5C,D**). Furthermore, we also found the level of strictosidine synthase (SS) expression was improved in AeVTDC1 in tobacco. The contents of caffeic acid, chlorogenic acid and some quinine (such as 1-O-Caffeoylquinic acid, 4-O-Caffeoylquinic acid) increased by overexpression AeVTDC1 in tobacco (**Figure 5D**). These results revealed that the content changes of metabolite (serotonin, tryptamine derivatives, ferulic acid and its derivatives) in transgenic tobacco might contribute to the enhanced resistance to RKN in tobacco. Moreover, the related genes expressions were also detected in L133. Expression of AeVTDC1 was higher in L120 than that in L133. Expression of NtTDC, COMT, ASMT, and YUCCA was similar in the roots of L133 and L120. However, Expression of T5H and SS showed no obvious difference in the roots

FIGURE 2 | Silencing of AeVTDC1 gene compromised the resistance to CCN infection in Ae. variabilis No.1. (A,B) Roots were stained with acid fuchsin to dye nematodes. The visible pink nematodes were counted and photographed. (A) One vision of vector control root under microscope; (B) one vision of the AeVTDC1-silenced root. (C) The relative expression of AeVTDC1 in roots of vector control and AeVTDC1-silenced plants. VIGS inoculation was operated at the two-leaf stage of plant. About 2 weeks after inoculation, the roots of plants displayed BSMV infection symptoms were individually collected to affirm the VIGS effect by QPCR. The results were normalized with the AeVEF1α. (D) Number of CCN in AeVTDC1-silenced roots and vector control plants. The successful silencing plants were used for CCN J2 inoculation. Roots were collected and dyed with acid fuchsin 3 days after CCN inoculation. The number of visible pink-stained nematodes present within the roots was counted under a light microscope. The data were means ± SE. No less than 15 plants were used and calculated in each treatment. The asterisk represented significant differences (P < 0.05).

of L133 and WT (**Figure 5C**). The expression differences might account for the different resistances by AeVTDC1 overexpression.

# DISCUSSION

# The Role of AeVTDC1 in Resistance Against CCN and RKN

CCN (H. avenae) is soil-borne and invades plants from roots. Production losses caused by CCN gradually become bigger in recent years. However, it is difficult to control CCN disease because CCN infection is not easy to be observed. Bringing resistance-related genes from relatives into wheat has been known as an efficient strategy to enhance wheat resistance to CCN. However, few resistance-related genes were identified. High-throughput methods like RNA-seq and microarray analysis are widely used to find differential expression genes, which provide a mass of candidate genes for following work.

In our previous work, RNA-sequencing was operated to find out differential expression genes before and after CCN infection in Ae. variabilis No.1 (Xu et al., 2012). There were a large number of differential expression genes at 30 hpi, when the early response was conferred to CCN infection. Whereas there were fewer differential expression genes at 3 and 9 dpi, when CCN J2 had migrated in vascular tissues and developed into J2∼J3 stage. It was reported that most of differential expression genes gathered at 3 and 8 dpi in RNA-sequencing analysis of incompatible wheat and a compatible control cultivar infected with H. avenae at 24, 3, and 8 dpi (Kong et al., 2015). The different responses might indicate a different resistance mechanism in Ae. variabilis No.1.

Previous analysis showed transcripts of several genes in tryptophan metabolism were induced by CCN, and AeVTDCs as the key gene were also induced at the 30 hpi early response. In this study, AeVTDC1 and AeVTDC2 were further verified to be induced by CCN at 30 hpi, which was accordant with the transcriptome data (Xu et al., 2012).

FIGURE 3 | Silencing of AeVTDC1 affected expression of downstream genes and contents of relative secondary metabolites in Ae. variabilis No.1 plants. (A) Relative expressions of AeVTDC1 downstream genes (T5H, ASMT, COMT, and MES) were detected in AeVTDC1-silenced and vector control roots. (B) Fold changes of secondary metabolites in AeVTDC1-silenced compared to that in vector control roots. The content of secondary metabolite was detected by UPLC-ESI-MS/MS. Data represent the mean ± standard deviation of three replicate samples. Asterisk above the bars indicate values that were significantly different (P < 0.05).

In the wheat genome, there are more than 15 copies of TaTDCs which showed various tissue expression patterns (Choulet et al., 2014). Expression patterns of two TaTDCs (TaTDC1 and TaTDC2), highly expressed in roots and with highest homology to AeVTDC1, were tested after CCN inoculation. Expression of TaTDC1 was induced after CCN inoculation, while expression of TaTDC2 showed no obvious changes after CCN infection (**Supplementary Figure 2**). The different expression pattern of the various isoforms might indicate that not all TaTDCs isoforms respond to CCN. Biological functions of several plant TDCs were gradually disclosed. Overexpression the Catharanthus roseus TDC gene in plants (tobacco, poplar, canola, Petunia hybrida) gave rise to tryptamine accumulation in transgenic plants, which improved tobacco resistance to Manduca sexta and

content of IAA was detected by UPLC–MS. Data represent the mean ± standard deviation of three replicate samples.

poplar resistance to Malacosoma disstria (Thomas et al., 1999; Leech et al., 2000; Gill et al., 2003). In Ae. variabilis No.1 genomes, there were at least three TDC genes that may have different functions in downstream metabolism process from tryptamine (Facchini et al., 2000; Byeon et al., 2014). However, which AeVTDC participated in CCN resistance remained unclear. In this study, expression of AeVTDC1 showed greater changes than AeVTDC2 after CCN infection (**Figures 1A,B**). Here, the role of AeVTDC1 in CCN resistance was directly studied. Silencing of AeVTDC1 weakened resistance to CCN and led more CCN invading into roots (**Figure 2**). This indicated the positive regulation of AeVTDC1 in resistance to CCN. It needs further study whether there was function redundancy among AeVTDC1 and its homologs. Moreover, the role of AeVTDC1

FIGURE 5 | Number of root knots formed on roots of AeVTDC1 overexpressing tobacco plants and non-transgenic plants, and contents of downstream metabolites in roots were respectively analyzed. (A) Phenotypes of the root tissues of non-transgenic and AeVTDC1 transgenic tobacco. L120 and L133 were two independent lines of AeVTDC1 transgenic tobacco. The first row showed the overall view of root tissues, and the second row displayed the partial tissue circled with red oval in the first row. Photos were taken 2 months after planting in soil containing RKN. (B) Statistic data of root knots numbers formed on AeVTDC1 overexpressing tobacco and non-transgenic tobaccos. The mean number of knots was calculated and presented. More than 15 replicates were used for counting. (C) Relative expressions of AeVTDC1, NtTDC, TDC downstream genes T5H, ASMT, COMT, SS in tobacco, and IAA biosynthetic related genes YUCCA were detected in root tissues of L120, L133, and WT. (D) Fold change of secondary metabolite in AeVTDC1 overexpressing tobacco plants (L120) and WT roots. The content of secondary metabolite was detected by UPLC-ESI-MS/MS. The error bar represented the standard error. <sup>∗</sup>P-value < 0.05.

in RKN resistance was also tested by overexpression in tobacco. AeVTDC1 overexpression reduced knots on roots and it revealed AeVTDC1 plays a positive role in RKN resistance as in CCN resistance (**Figures 5A,B**). The different resistance on L120 and L132 might be related to the expression level of AeVTDC1 in tobacco (Zhang et al., 2013). Hence, we suggested the broad spectrum resistance of AeVTDC1 to crucial nematodes.

# Relationships Between Downstream Secondary Metabolites of AeVTDC1 and Nematode Defense

In the plant kingdom, diverse metabolites derived from Trp are found and play an important role in the plant immunity and rice (Bohlmann et al., 1995; Zhao and Last, 1996; Matsukawa et al., 2002; Kang and Back, 2009; Ueno et al., 2010; Ishihara et al., 2011; Dharmawardhana, 2013; Dubouzet et al., 2013; Hayashi et al., 2016; Lu et al., 2018). Metabolites, liking tryptamine, serotonin or their derivatives, have strong antioxidant activities and regulate resistance to avoid damage from pathogen attacks (Kang and Back, 2009; Dubouzet et al., 2013). The dramatic increased serotonin suppress leaf damage outside the halo, block expansion of the browning area and attenuate symptom of plant growth inhibition (Hayashi et al., 2016). Ectopic expression of Camptotheca acuminata TDC1 gene allowed sufficient tryptamine to accumulate in poplar and tobacco leaf tissue to significantly suppress the growth of insect pests (Gill et al., 2003). Melatonin-rich rice plants exhibit resistance to

herbicide-induced oxidative stress (Park et al., 2013). Recently, it was reported that CYP71A1 mutants with less serotonin content were more susceptible to rice blast Magnaporthe grisea, but more resistant to rice brown spot disease Bipolaris oryzae1, rice brown planthopper and striped stem borer (Lu et al., 2018).

In this study, serotonin concentration in AeVTDC1 transgenic tobacco roots was eightfold greater than in roots from WT plants, while tryptamine and melatonin remained unchanged (**Figure 5D**). No change of tryptamine was reasonable since it might be easily transformed to its derivatives (N-hydroxytryptamine, N-Acetyltryptamine, and 5-Methoxy-N,N-dimethyltryptamine), which increased in the transgenic tobacco. Our results were similar to that in overexpression of Catharanthus TDC in cell cultures of Peganum harmala (Leech et al., 2000). Serotonin and tryptamine derivatives, caffeic acid, chlorogenic acid and several quinines remarkably increased in transgenic tobacco (**Figure 5D**). Except for function in plant immunity, serotonin, feruloylserotonin, and 4-coumaroylserotonin were reported contributed to delay senescence of rice (Kang and Back, 2009). High level of serotonin accumulation in rice caused stunt phenotype (Kanjanaphachoat et al., 2012). Even though elevated level of serotonin accumulated in AeVTDC1 transgenic tobacco, the plants didn't show stunted growth (not shown) as overexpression of TDC in rice (Kanjanaphachoat et al., 2012). In the VIGS assay, the serotonin wasn't either detected in AeVTDC1-silencing root or in the control roots. That's probably because of too low level of serotonin in roots of Ae. variabilis No.1. Contents of N-Feruloylserotonin and N-Feruloyltryptamine markedly decreased in the AeVTDC1 silencing plants comparing with the control (**Figure 4B**). Ferulic acid is a well-known phytoalexins, which inhibited the growth of necrotrophic bacteria Dickeya dadantii (Pérez-Bueno et al., 2016). Furthermore, the decrease of soluble free and soluble conjugated phenolic acids, such as soluble hydrolyzable ferulic acid and sinapic acid, reduced the attraction of Diabrotica virgifera virgifera to the root of maize (Erb et al., 2015). In our study, ferulic acid significantly increased in the AeVTDC1 transgenic tobacco, and decreased in AeVTDC1-silencing plants (**Figures 4B**, **5D**). Here, we found altered expression of AeVTDC1 changed secondary metabolites and nematode resistance. It still needs further studies of direct effects of serotonin, ferulic acid, indole alkaloids, quinines and their derivatives on nematodes.

# Regulation of IAA Biosynthesis by AeVTDC1 and the Relationship Between IAA and CCN Resistance

Plant hormone IAA biosynthesis has several main pathways from Trp, respectively catalyzed by TDC, YUCCA, and NIT (Sugawara et al., 2009; Zhao, 2010). It is well-known that auxin always regulates plant defense as a negative regulator in plant immune system (Mathesius, 2010; Navarro, 2016). A series of evidence has demonstrated that auxin plays roles in balancing plant defense responses and growth in plants. Lionel Navarro reported the repression of auxin signaling made for the improving of bacterial resistance in Arabidopsis (Navarro et al., 2006). Overexpression of OsGH3.1 and OsGH3.8 in rice reduced the IAA content, influence cell growth, and enhanced disease resistance to both fungal and bacterial pathogens (Domingo et al., 2009). In this work, the content of IAA was not changed by silencing AeVTDC1 in Ae. variabilis No.1 (**Figure 3C**). There was also no change about the content of IAA in the AeVTDC1 transgenic tobacco comparing with WT (**Figure 5D**). Silencing or overexpression of AeVTDC1 gene also had no effects on the expression of IAA biosynthesis and signaling genes (**Figures 3A,B**, **5C**). The results indicate AeVTDC1 might not function to regulate IAA biosynthesis. Moreover, we used two concentrations (100 and 200 µM) of 2, 4-D to pretreat the roots of Ae. variabilis No.1 to test CCN resistance. Statistical results revealed that there was no difference of CCN number in the roots between IAA pretreat and control (**Supplementary Figure 3**). These results indicate that IAA has no impact on the interaction between CCN and Ae. variabilis No.1. In addition, IAA is likely subjected to strict monitoring during the interactions of plants and nematodes.

# AUTHOR CONTRIBUTIONS

HZ and MY designed the experiments. HZ and QH carried out sample collection, expression analysis, gene silencing, disease resistance assessment, data analysis, and wrote the manuscript. LL transformed transgenic tobacco. MZ analyzed transcriptome data. HL, GD, ZP, JL, QL, and FC assisted the experiments. All authors approved the final manuscript.

# FUNDING

This work was supported by National Natural Science Foundation of China (Grant Nos. 31501614 and 31470097), Major transgenic special project (Grant No. 2016ZX08009003- 001), and West Light Foundation of Chinese Academy of Sciences (Grant No. 2015XBZG\_XBQNXZ\_B\_010).

# ACKNOWLEDGMENTS

We are grateful of Professor Feng Liu (Shan Dong Agricultural University) for providing soil containing cereal cysts. We are grateful of Professor Deliang Peng and Shujie Luo (Institute of Plant Protection, Chinese Academy of Agricultural Sciences) for providing methods of dying CCN. We thank the expertise of Miss. Shuang Fang and Dr. Jinfang Chu [National Centre for Plant Gene Research (Beijing), Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, China] in determining the IAA contents of roots.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01297/ full#supplementary-material

# REFERENCES

fpls-09-01297 October 10, 2019 Time: 18:22 # 10



**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 Huang, Li, Zheng, Chen, Long, Deng, Pan, Liang, Li, Yu 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(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.

# Corrigendum: The *Tryptophan decarboxylase 1* Gene From *Aegilops variabilis No.1* Regulate the Resistance Against Cereal Cyst Nematode by Altering the Downstream Secondary Metabolite Contents Rather Than Auxin Synthesis

#### *Approved by:*

*Frontiers Editorial Office, Frontiers Media SA, Switzerland*

#### *\*Correspondence:*

*Maoqun Yu yumq@cib.ac.cn Haili Zhang zhanghl@cib.ac.cn*

#### *Specialty section:*

*This article was submitted to Plant Microbe Interactions a section of the journal Frontiers in Plant Science*

*Received: 10 September 2019 Accepted: 11 September 2019 Published: 10 October 2019*

#### *Citation:*

*Huang Q, Li L, Zheng M, Chen F, Long H, Deng G, Pan Z, Liang J, Li Q, Yu M and Zhang H (2019) Corrigendum: The Tryptophan decarboxylase 1 Gene From Aegilops variabilis No.1 Regulate the Resistance Against Cereal Cyst Nematode by Altering the Downstream Secondary Metabolite Contents Rather Than Auxin Synthesis. Front. Plant Sci. 10:1271. doi: 10.3389/fpls.2019.01271*

*Qiulan Huang1,2,3, Lin Li4, Minghui Zheng4, Fang Chen2, Hai Long1, Guangbing Deng1, Zhifen Pan1, Junjun Liang1, Qiao Li1, Maoqun Yu1\* and Haili Zhang1\**

 *1 Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu, China, 2 College of Life Sciences, Sichuan University, Chengdu, China, 3 University of the Chinese Academy of Sciences, Beijing, China, 4 School of Basic Medical Sciences, Zunyi Medical University, Zunyi, China*

Keywords: cereal cyst nematode, Aegilops variabilis No.1, Tryptophan decarboxylase, secondary metabolite, indole acetic acid

### **A Corrigendum on**

**The** *Tryptophan decarboxylase 1* **Gene From** *Aegilops variabilis No.1* **Regulate the Resistance Against Cereal Cyst Nematode by Altering the Downstream Secondary Metabolite Contents Rather Than Auxin Synthesis**

*by Huang Q, Li L, Zheng M, Chen F, Long H, Deng G, Pan Z, Liang J, Li Q, Yu M and Zhang H (2018). Front. Plant Sci. 9:1297. doi: 10.3389/fpls.2018.01297*

In the published article, there was an error in affiliations 2 and 3. Instead of "University of the Chinese Academy of Sciences, Beijing, China" and "College of Life Sciences, Sichuan University, Chengdu, China," it should be "College of Life Sciences, Sichuan University, Chengdu, China" and "University of the Chinese Academy of Sciences, Beijing, China."

The authors apologize for this error and state that this does not change the scientific conclusions of the article in any way. The original article has been updated.

*Copyright © 2019 Huang, Li, Zheng, Chen, Long, Deng, Pan, Liang, Li, Yu 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(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 Two-Headed Monster to Avert Disaster: HBS1/SKI7 Is Alternatively Spliced to Build Eukaryotic RNA Surveillance Complexes

Jacob O. Brunkard1,2 \* and Barbara Baker1,2

<sup>1</sup> Department of Plant and Microbial Biology, University of California, Berkeley, Berkeley, CA, United States, <sup>2</sup> Plant Gene Expression Center, USDA Agricultural Research Service, Albany, CA, United States

The cytosolic RNA exosome, a 30→5 <sup>0</sup> exoribonuclease complex, contributes to mRNA degradation in eukaryotes, limiting the accumulation of poorly-translated, improperly translated, or aberrant mRNA species. Disruption of cytosolic RNA exosome activity allows aberrant RNA species to accumulate, which can then be detected by host antiviral immune systems as a signature of pathogen infection, activating antiviral defenses. SKI7 is a critical component of the cytosolic RNA exosome in yeast, bridging the catalytic exoribonuclease core with the SKI2/SKI3/SKI8 adaptor complex that guides aberrant RNA substrates into the exosome. The ortholog of SKI7 was only recently identified in humans as an alternative splice form of the HBS1 gene, which encodes a decoding factor translational GTPase that rescues stalled ribosomes. Here, we identify the plant orthologs of HBS1/SKI7. We found that HBS1 and SKI7 are typically encoded by alternative splice forms of a single locus, although some plant lineages have evolved subfunctionalized genes that apparently encode only HBS1 or only SKI7. In all plant lineages examined, the SKI7 gene is subject to regulation by alternative splicing that can yield unproductive transcripts, either by removing deeply conserved SKI7 coding sequences, or by introducing premature stop codons that render SKI7 susceptible to nonsense-mediated decay. Taking a comparative, evolutionary approach, we define crucial features of the SKI7 protein shared by all eukaryotes, and use these deeply conserved features to identify SKI7 proteins in invertebrate lineages. We conclude that SKI7 is a conserved cytosolic RNA exosome subunit across eukaryotic lineages, and that SKI7 is consistently regulated by alternative splicing, suggesting broad coordination of nuclear and cytosolic RNA metabolism.

Keywords: RNA exosome, innate immunity, antiviral defense, RNA interference, alternative splicing, HBS1, SKI7

# INTRODUCTION

Viruses dominate the biosphere, massively outnumbering cellular organisms (Koonin, 2017). Unlike cellular organisms, however, all viruses are obligate parasites that depend on organismal ribosomes for translation and replication (Walsh et al., 2013; Wang, 2015). Viral hosts are under strong selective pressure to recognize and limit viral parasitism, and in parallel, viruses are under strong selective pressure to evade host surveillance mechanisms and coopt the host translation

#### Edited by:

Feng Qu, The Ohio State University, United States

#### Reviewed by:

Aardra Kachroo, University of Kentucky, United States Matthew R. Willmann, Cornell University, United States

> \*Correspondence: Jacob O. Brunkard brunkard@berkeley.edu

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 14 April 2018 Accepted: 24 August 2018 Published: 12 September 2018

#### Citation:

Brunkard JO and Baker B (2018) A Two-Headed Monster to Avert Disaster: HBS1/SKI7 Is Alternatively Spliced to Build Eukaryotic RNA Surveillance Complexes. Front. Plant Sci. 9:1333. doi: 10.3389/fpls.2018.01333

**220**

machinery (Walsh et al., 2013; Molleston and Cherry, 2017). Host organisms have evolved various defense and immune systems to protect against viruses, including several approaches to detect non-host (viral) nucleic acids and then trigger defenses and guided destruction of viral genomes (Barrangou et al., 2007; Narayanan and Makino, 2013; Szittya and Burgyán, 2013; Abernathy and Glaunsinger, 2015; Li et al., 2015; Molleston and Cherry, 2017). A famous example is the evolution of bacterial "CRISPR-Cas" systems that incorporate viral dsDNA into short palindromic repeats in the host genome, and then transcribe these incorporated DNA sequences into guides that target DNases to specifically cleave complementary viral DNA (Brouns et al., 2008). Eukaryotes have evolved two major approaches to detect viral nucleic acids and mount immune responses. One of these, the interferon system, evolved in the Gnathostomata lineage (vertebrates with jaws) (Zou et al., 2009). In these organisms, pattern recognition receptors detect aberrant nucleic acids (such as dsRNA) and then rapidly activate transcription of a class of cytokines, called interferons, that are secreted, detected by cell surface receptors (of both infected and uninfected cells), and ultimately trigger transcriptional reprogramming that limits viral replication and spread (Li et al., 2015; Rigby and Rehwinkel, 2015). Outside the Gnathostomata lineage, most eukaryotes employ RNA interference (RNAi) systems to combat viral infections (Shabalina and Koonin, 2008).

RNA interference relies on endoribonucleases of the RNase III family that recognize and cleave double-stranded RNA (dsRNA) as basal defense against viral infection (Seo et al., 2013; Nicholson, 2014). RNase III may recognize dsRNA synthesized by a viral replicase, or the host may synthesize dsRNA from viral RNA templates using an endogenous RNA-dependent RNA polymerase; in either case, the dsRNA is processed by RNase III (Blevins et al., 2006; Szittya and Burgyán, 2013). In plants, two RNase III enzymes, called DICER-LIKE 4 (DCL4) and DCL2, are primarily responsible for defense against RNA viruses (Szittya and Burgyán, 2013; Andika et al., 2015). Cleavage of viral dsRNA by DCL4/DCL2 generates viral short interfering RNAs (vsiRNAs). vsiRNAs are loaded into ARGONAUTE (AGO) proteins in RNAi complexes (RISCs) that search for further, complementary viral RNAs using the 21- or 22-nt siRNA sequence. Any RNA recognized by the RISC, presumably a viral genome, is then cleaved and degraded. In a poorly understood process, RDR6 can be recruited to the cleaved viral RNA to synthesize another dsRNA template for processing by DCL4/DCL2 (Qu et al., 2005, 2008; Qi et al., 2009). This RDR6 dependent process can amplify the number of antiviral siRNAs available for the immune system (Garcia-Ruiz et al., 2010).

The DCL4/DCL2/RDR6 surveillance system has since been coopted in the plant lineage to regulate endogenous gene expression (Peragine et al., 2004; Allen et al., 2005; Cuperus et al., 2010). After cleavage by a 22-nt miRNA (or, in exceptional cases, some 21-nt miRNAs), endogenous transcripts can become templates for copying by RDR6 and cleavage by DCL4/DCL2 into siRNAs (Allen et al., 2005). These siRNAs can then amplify silencing of the miRNA target by guiding RISCs to multiple sites in transcripts with the original miRNA target, or may act in trans by guiding the RISC to other transcripts with complementary (or nearly complementary) ∼21 nt sequences. The DCL4/DCL2/RDR6 system is used by plants to regulate diverse biological processes, including auxin-mediated developmental patterning (Peragine et al., 2004; Vazquez et al., 2004; Fahlgren et al., 2006) and suppression of disease resistance R genes (Li et al., 2012; Deng et al., 2018), whose overexpression can cause autoimmune syndromes (Yi and Richards, 2009).

After cleavage by RISC endonucleases, viral RNA must be degraded by host RNA exoribonucleases (Abernathy and Glaunsinger, 2015). Two major exoribonuclease mechanisms are conserved across eukaryotes: 50→3 <sup>0</sup> RNA degradation by the EXORIBONUCLEASEs (XRNs) (Nagarajan et al., 2013) and 30→5 <sup>0</sup> RNA degradation by the RNA exosome (Zinder and Lima, 2017). These enzymes are involved in housekeeping degradation of host mRNAs in the cytosol and in processing RNA transcripts in the nucleus. The XRNs have evolved two distinct gene families to handle these processes: the XRN1 family encodes a cytosolic enzyme, and the XRN2 family encodes a nuclear enzyme. In an ancestor of plants, an XRN2 paralog evolved cytosolic localization and the XRN1 gene lineage was lost; the plant cytosolic XRN is therefore called XRN4, but is functionally equivalent to eukaryotic XRN1 (Nagarajan et al., 2013). The RNA exosome forms a large, multiprotein complex, and localizes to both the cytosol and nucleus. Adaptor complexes then guide RNA substrates to the RNA exosome; for example, the nuclear TRAMP complex guides ribosomal RNA (rRNA) and small nucleolar RNA (snoRNA) to the RNA exosome for processing (Tollervey, 2015). In the cytosol, the RNA exosome is chaperoned to substrates by the SKI2/SKI3/SKI8 complex (Schmidt et al., 2016), which facilitates degradation of cleaved RNA (including deadenylated mRNAs) by guiding these RNAs into the RNA exosome catalytic core.

During RNAi, the SKI2/SKI3/SKI8 complex is recruited to siRNA-cleaved mRNAs by stalled ribosomes that reach the cleavage site (Orban and Izaurralde, 2005; Branscheid et al., 2015; Zhang et al., 2015; Szadeczky-Kardoss et al., 2018). Stalled ribosomes at the cleavage site are recognized by PELOTA/DOM34 (a translational decoding factor) and HBS1 (a translational GTPase), which recruit SKI2/SKI3/SKI8 and the RNA exosome to degrade the 5<sup>0</sup> fragment of the cleaved mRNA (Orban and Izaurralde, 2005). Supporting this hypothesis, loss of PELOTA, HBS1, or SKI2 stabilizes the 5 0 fragment of transcripts cleaved in their open reading frames by miRNA-guided RISCs, in both Drosophila and plants (Orban and Izaurralde, 2005; Szittya and Burgyán, 2013; Branscheid et al., 2015; Hashimoto et al., 2017; Szadeczky-Kardoss et al., 2018). Extending this model, we propose that PELOTA, HBS1, SKI2/SKI3/SKI8, and SKI7 are likely necessary for degradation of messenger viral RNAs after cleavage by vsiRNAs. Moreover, PELOTA, HBS1, SKI2/SKI3/SKI8, and SKI7 are all proposed to contribute to degradation of transcripts with premature termination codons via nonsense-mediated decay (NMD) (Mitchell and Tollervey, 2003; Takahashi et al., 2003; Arribere and Fire, 2018). Structural features of viral RNA can be recognized as nonsense transcripts by eukaryotic cells, leading to viral RNA degradation by NMD. Thus, PELOTA, HBS1, and SKI7 could have antiviral roles by participating in

NMD. In tomato, a natural loss-of-function variant of pelota, called ty-5, confers resistance to Tomato yellow curly leaf virus (TYCLV), highlighting the importance of the PELOTA/HBS1 RNA degradation machinery in plant-virus interactions (Lapidot et al., 2015).

In humans and in plants, the SKI2 has been implicated in preventing endogenous RNA from triggering immune responses. Depletion of the human SKI2 ortholog, HsSKIV2L (Kalisiak et al., 2017), allows accumulation of aberrant RNA species that are sensed by nucleic acid pattern recognition receptors, which in turn activate type I interferon expression and trigger autoimmune/autoinflammatory responses (Eckard et al., 2014; Rigby and Rehwinkel, 2015). Loss-of-function mutations in SKIV2L have been genetically linked to autoimmune syndromes (Crow et al., 2006), including systemic lupus erythematosus (Crow et al., 2006), and this association with autoimmune syndromes may be related to its role in limiting autoinflammatory responses to endogenous RNAs (Eckard et al., 2014). In Arabidopsis thaliana, disruption of the SKI2 gene has a similar effect: deadenylated RNA species accumulate in the cytosol (e.g., transcripts cleaved by miRNAs), and become available as templates for RDR6 to generate dsRNA (Branscheid et al., 2015), a process that is comparable to RDR6 copying of cleaved viral RNA. These RDR6-dependent dsRNA molecules are subsequently processed by DCL4/DCL2, generating siRNAs that silence host gene expression. Thus, in both humans and plants, SKI2 is required to limit the accumulation of aberrant RNA species that are otherwise detected by the cell as potential viruses, triggering antiviral immune responses in the absence of pathogen attack.

Recently, structural studies of the cytoplasmic RNA exosome in yeast and humans have revealed the crucial importance of SKI7 in bridging the RNase exosome complex with the SKI2/3/8 adaptor complex that feeds cytosolic RNA substrates into the exosome (Kalisiak et al., 2017). SKI7 was first identified in the same genetic screen as the other cytoplasmic RNA exosome components, but unlike SKI2, SKI3, and SKI8, orthologs of SKI7 were not readily identifiable in other eukaryotic genomes. A genomic investigation of Lachancea kluyveri, a fungus closely related to S. cerevisiae, revealed that SKI7 is encoded by an alternative splice form of the HBS1 locus in that species (Marshall et al., 2013). In S. cerevisiae, HBS1 and SKI7 are functionally distinct homeologs that derive from a whole-genome duplication in a recent ancestor of S. cerevisiae; L. kluyveri diverged from this lineage shortly before the whole-genome duplication. The authors of this study noted briefly that the HBS1 locus is potentially alternatively spliced in other eukaryotic lineages, but did not systematically identify HBS1/SKI7 orthologs in metazoans or plants. Subsequent studies revealed that the vertebrate HBS1 locus is also alternatively spliced, and that one of these splice forms, HBS1Lv3, encodes a protein that serves the same function as SKI7 in S. cerevisiae (Kalisiak et al., 2017).

The discovery that SKI7 and HBS1 are encoded by the same locus in many fungi and vertebrates is perhaps surprising because of their apparently unrelated functions. HBS1 is a translational GTPase that is required for the release of stalled ribosomes from mRNA, along with its interacting partner, PELOTA/DOM34 (**Figure 1**; Shao et al., 2016). SKI7 is instead a bridge between the RNA exosome and the SKI2 complex, and while the S. cerevisiae SKI7 has a C-terminal HBS1-like GTPase domain, this terminus is dispensable for its functions (Horikawa et al., 2016). Highlighting their distinct functions, the two protein isoforms encoded by the HBS1/SKI7 locus in Lachancea kluyveri can complement only one of the S. cerevisiae 1hbs1 or 1ski7 mutant strains: the long, SKI7-like isoform only complements 1ski7, and the shorter, HBS1-like isoform only complements 1hbs1 (Marshall et al., 2013). In humans, the isoform encoding the functional ortholog of SKI7, HBS1Lv3, loses exons that encode the entire C-terminal HBS1 GTPase domain (which is essential for HBS1 functions), and instead gains an exon that encodes an RNA exosome-interacting surface (**Figure 2A**). It remains unclear why vertebrate and most fungal genomes would retain a single locus to encode both HBS1 and SKI7, whereas S. cerevisiae has successfully evolved two distinct loci to separate these functions. Here, we take advantage of these recent insights into the gene and protein structures of HBS1/SKI7 in other eukaryotes to identify and characterize HBS1/SKI7 orthologs in the plant lineage.

# RESULTS

# Identification of HBS1/SKI7 in A. thaliana

Transcripts are decoded during translation by duplexes composed of a translational GTPase (trGTPase) and either an aminoacyl-tRNA or a ribosome release factor (Dever and Green, 2012; Shao et al., 2016). There are three major classes of these decoding trGTPases: eEF1α (eukaryotic Elongation Factor 1 alpha), which mediates delivery of aminoacyl-tRNAs to the 80S ribosome; eRF3 (eukaryotic Release Factor 3), which mediates delivery of eRF1 (eukaryotic Release Factor 1) to stop codons to terminate translation and facilitate ribosome dissociation from transcripts; and HBS1 (Hsp80 subfamily B Suppressor 1), which mediates delivery of PELOTA (a.k.a. Dom34 in yeast, Duplication Of Multilocus region) to stalled ribosomes to terminate translation and facilitate ribosome dissociation (**Figures 1A,B**) (Carr-Schmid et al., 2002; Becker et al., 2012; Shao et al., 2016; Hashimoto et al., 2017). Since all decoding trGTPases are similar to each other, we began by identifying orthologs of each of the three decoding factor trGTPases (eEF1α, eRF3, and HBS1) in Arabidopsis thaliana in order to confidently distinguish plant HBS1 from the other trGTPases (**Figure 1C** and **Supplementary Data File 1**). Separate loci encode orthologs of these proteins that localize to mitochondria and/or plastids; these were removed from our analysis to focus only on cytoplasmic proteins. Arabidopsis encodes one copy of eRF3 (At1g18070), four copies of eEF1α (At5g60390 and three tandem paralogs, At1g07920, At1g07930, and At1g07940), and one copy of HBS1 (At5g10630). The three tandem eEF1α paralogs evolved recently (this complex locus is not conserved across Brassicaceae). All of the decoding trGTPases are expressed throughout Arabidopsis development, although it should be noted that eRF3 transcripts are about one order of magnitude

degradation of the mRNA and the nascent polypeptide. Ribosome may stall on mRNAs with complex secondary structures or after endonucleolytic cleavage by RNAi silencing complexes, among other possible causes. (C) The three decoding factor translational GTPases are conserved across the plant, fungal, and metazoan

lineages, as represented in this phylogeny by protein sequences from Arabidopsis thaliana, Saccharomyces cerevisiae, and Homo sapiens, respectively.

RNA-Seq reads are shown in black (top panel), and select individual aligned reads are shown in green (bottom panel), with spliced sequences indicated by a black line. Note that reads for the fourth exon are ∼20% of the level of reads for the other coding sequence exons. (C) At5g10630 forms three major splice forms. A short splice form (top) skips exon 4 (yellow), yielding a transcript that encodes HBS1 (A). A long splice form (middle) includes exon 4 (yellow), yielding a transcript that encodes SKI7 (A). Rarely, an alternative acceptor site is selected for exon 4, adding five amino acids with no apparently functional consequence. A nonsense splice form (bottom) retains intron 4, which includes two codons, yielding a transcript that is likely subject to NMD. Exons are colored to match protein models in subsequent figures; UTRs are indicated with narrow, white bars.

more abundant than HBS1 transcripts, and eEF1α transcripts are at least two orders of magnitude more abundant than HBS1 transcripts, consistent with their distinct roles in translation (Cheng et al., 2017).

The Arabidopsis At5g10630 (HBS1) locus is annotated with several different possible transcripts, but there are only two major protein isoforms predicted to be encoded by these transcripts: a long splice form (**Figure 2C**) encodes a protein that is 738 amino acids long (**Figures 3**, **4A**), and a short splice form (**Figure 2C**) skips an exon to encode a shorter protein that is 668 amino acids long (**Figures 3**, **4A**). High-throughput sequencing of RNA from light-grown seedlings shows that the

alternative cassette exon is included in approximately 20% of At5g10630 transcripts (**Figure 2B**) (Cheng et al., 2017). A very minor splice form uses a weak 3<sup>0</sup> splice site that adds 15nt to the 5<sup>0</sup> end of the cassette exon; this is included in at most 4% of At5g10630 transcripts in light-grown seedlings. Because this minor splice form is rare and does not cause significant changes in the protein sequence (it neither induces a frame-shift nor includes a stop codon, and only adds 5 amino acids to the protein), it was not investigated any further. Another very minor splice form (also at most 4% of At5g10630 transcripts in light-grown seedlings, **Figure 2B**) retains the intron at the 3<sup>0</sup> end of the cassette exon (**Figures 2C**, **3**), which introduces two in-frame stop codons (this retained intron can thus be considered a "poison intron"). Since these stop codons are upstream of several exon-exon junctions, this splice form is presumably subject to NMD, and is any case unproductive. We confirmed the RNA sequences of these four splice forms of At5g10630 by RT-PCR using primers surrounding


FIGURE 4 | Diversity of HBS1/SKI7 isoforms among plant lineages. (A) The typical plant HBS1/SKI7 locus (here exemplified by At5g10630, but also found in P. patens, A. trichocarpa, T. cacao, M. truncata, N. benthamiana, and S. lycopersicum, among other species) encodes two protein isoforms, HBS1 (top) and SKI7 (bottom), defined by the inclusion or exclusion of the SKI7-like motif (yellow). (B) The Solanaceae have two distinct HBS1/SKI7-like loci: one is similar to the typical plant HBS1/SKI7 locus in panel a (exemplified by Solyc03g119290), and the second encodes only an HBS1-like protein, and is alternatively spliced to include or exclude two Zn finger domains (exemplified by Solyc12g036410). (C) In the PACMAD grasses, one HBS1/SKI7-like locus encodes an HBS1-like protein lacking the SKI7-like motif and with a poorly-aligned Patch 4-like motif (exemplified by Zm0001d026213). A second HBS1/SKI7-like locus (exemplified by Zm0001d001827) can encode two proteins: a long SKI7 isoform and a short isoform that excludes the Patch 4-like motif, and thus may function as an HBS1-like protein or be a loss-of-function isoform of SKI7. (-)N, blue, negatively charged N-terminus; ZnF, green, RAN2-type Zinc finger domain; SKI7 motif, yellow, the SKI7-like motif often encoded by a cassette exon; P4, red, the Patch 4-like motif; ∼4, orange, a poorly-aligned Patch 4-like motif unique to grass HBS1 isoforms; GTPase, gray, the HBS1/SKI7 decoding factor translational GTPase domain.

the alternatively spliced exons, followed by TOPO cloning and Sanger sequencing.

# Structural Features of the Arabidopsis HBS1/SKI7 Protein Isoforms

Comparative analysis of the two proteins predicted to be encoded by Arabidopsis HBS1/SKI7 with the functionally characterized orthologs of HBS1 (human HBS1Lv1, baker's yeast Hbs1, and budding yeast Lachancea kluyveri HBS1) and SKI7 (human HBS1Lv3, baker's yeast Ski7, and budding yeast L. kluyveri SKI7) allowed us to identify several conserved regions in the Arabidopsis proteins (**Figures 4A**, **5A**). The N-terminus (aa 1–50) begins with a stretch of mostly negatively charged amino acids (40% of the first 50 amino acids are aspartic acid or glutamic acid, **Figure 3**). This is followed by a single Zinc finger domain of the RanBP2 superfamily (aa 51–75, **Figure 5A**), putatively involved in protein-protein interactions. If the transcript is alternatively spliced to include a cassette exon, the next region encodes an amino acid sequence with some similarity to the polypeptide encoded by the L. kluyveri

alternative exon that determines SKI7 functionality (Marshall et al., 2013). This region includes a motif that corresponds to the HsSKI7 RxxxFxxxL motif required for recruiting HsSKI7 to the RNA exosome (Kalisiak et al., 2017). We have named this the "SKI7-like motif " (**Figure 5A**). Immediately after the sequence encoded by the cassette exon is the "Patch 4-like" motif (**Figure 5A**), named after the homologous yeast sequence, which was dubbed the "patch 4" motif (Kowalinski et al., 2016). In the human SKI7 protein, this is called the "PFDFxxxSPDDIVKxNQ motif " (Kalisiak et al., 2017). The Patch 4-like motif is found in all SKI7 proteins, but is not conserved in HBS1 proteins, and is proposed to mediate interactions between SKI7 and the RNA exosome subunit Csl4. Finally, the remaining C-terminus of the At5g10630 protein is an HBS1-like translational GTPase (**Figure 4A**). Thus, we defined the archetypical plant HBS1/SKI7 protein with five regions: a negatively charged N-terminus, a Znfinger domain, a SKI7-like motif, a Patch 4-like motif, and the C-terminal GTPase (**Figure 4A**).

examined, no single gene in rice can putatively encode both HBS1 and SKI7.

# Evolution of the HBS1/SKI7 Locus in Land Plants

We next used the Arabidopsis HBS1/SKI7 locus to search for orthologs in the genomes of land plants. We included the following species (**Figure 5B**): Sphagnum fallax and Physcomitrella patens (two distantly related species representing the moss lineage, which diverged early during land plant evolution), Selaginella moellendorffi (representing the lycophyte lineage, which diverged early during tracheophyte evolution), Amborella trichopoda (an ancient lineage of earliest-diverging flowering plants), Solanum lycopersicum and Nicotiana benthamiana (asterid eudicots in the Solanaceae family), Medicago truncatula (a rosid eudicot in the fabid order), Theobroma cacao (a close relative of Arabidopsis in the malvid order of rosid eudicots), Oryza sativa (a species of the BOP clade of grasses), and Zea mays and Setaria viridis (panicoid grasses). We focused on this limited set of taxa because of their excellent genome and transcriptome sequences, which allowed

SKI7-like motif includes a widely conserved proline insertion (at position 3 in the consensus sequence). Alignments of the human, Neurospora crassa (fungus), and Arabidopsis SKI7-like motif and Patch 4-like motifs, as well as surrounding residues, are shown above the consensus sequences as examples. Amino acids are colored following Clustal standards, as above (Figure 5A).

us to confidently identify potential alternative splice forms of the HBS1/SKI7 locus.

In the genomes of S. fallax, P. patens, A. trichopoda, M. truncatula, and T. cacao, there are single-copy HBS1/SKI7 orthologs that are each very similar to At5g10630 (**Figures 2C**, **4A**, **5A**). Like in Arabidopsis, alternative splicing of transcripts from these loci can include or exclude a cassette exon that encodes the SKI7-like region (**Figures 2C**, **4A**). We therefore tentatively propose that this is the ancestral form of HBS1/SKI7 in land plants, although as more deeply annotated transcriptomes of early-diverging land plant species become available, this proposal may require revision. S. moellendorffi encodes two HBS1/SKI7 orthologs that generate several transcript permutations by alternative splicing (**Figure 6A**). Again, like Arabidopsis, alternative splicing of transcripts from these loci can include or exclude the SKI7-like motif. LOC9640201 can also be alternatively spliced to remove both the SKI7-like and the Patch 4-like motifs. LOC9652039 is even more complex: it can be alternatively spliced to remove only the SKI7-like motif (variant 4), only the Patch 4-like motif (variant 3), or both the SKI7-like and Patch 4-like motifs (variant 5) (**Figure 6A**). All of these alternatively spliced variants of S. moellendorffi HBS1/SKI7 genes are predicted to encode proteins that function as HBS1, but not as SKI7.

In the Solanaceae, an ancestral gene duplication resulted in two distinct sets of orthologs (**Figure 4B**). In S. lycopersicum, one gene, Solyc03g119290, is similar to the typical land plant HBS1/SKI7 orthologs, and is alternatively spliced to include or exclude the SKI7-like region (**Figure 4B**, upper panel). A second gene, Solyc12g036410, encodes regions that are highly homologous to HBS1, including the negatively charged N-terminus and C-terminal GTPase (**Figure 4B**, lower panel). The Zn-finger domain has tandemly duplicated, and both Znfinger domains are on a cassette exon that is alternatively spliced (**Figure 4B**, lower panel; **Figure 5A**). The highly conserved Patch 4-like motif is absent in this gene (**Figure 4B**, lower panel; **Figure 5A**). A similar situation is found in a distantly related species of Solanaceae, N. benthamiana, which has two homeologous copies of the At5g10630-/Solyc03g119290-like gene, and one ortholog of the Solyc12g036410 gene that has

acids of Arabidopsis SKI7. The negatively charged N-terminus (blue, (-)N) and RAN2B-type Zinc Finger (green, ZnF) are predicted to interact with the SKI2/SKI3/SKI8 cytosolic exosome adaptor complex, and the SKI7-like motif (yellow) and Patch 4-like motif (red, P4) are predicted to interact with the surface of the RNA exosome catalytic core. The negatively charged N-terminus includes moderate-confidence α-helices that may promote interaction with SKI2/SKI3/SKI8. The ZnF domain includes two β-sheets (β1 and β2) and an α-helix that coordinate with the Zn ion. The SKI7-like and Patch 4-like motif include four α-helices (α1 through α4), which is structurally comparable to the yeast SKI7 protein. (B) Phyre2 prediction of the structure of the N-terminal 183 amino acids of Arabidopsis SKI7, based on homology modeling against several resolved protein structures (as described in the methods). Residues are colored by position, from blue (N-terminus) to red (C-terminus), closely matching the colors used in panel a and other figures. The defined regions of the protein are labeled, including α1, α2, α3, and α4, which are predicted to mediate interactions with the RNA exosome core. The C-terminus of this structure is predicted to be highly disordered, forming a flexible linker to the C-terminal trGTPase of Arabidopsis SKI7.

a duplicated Zn-finger domain and has lost the sequences to encode the SKI7-like and Patch 4 motifs (**Figure 5A**). Thus, duplication of the HBS1/SKI7 locus in an ancestor of Solanaceae led to evolution of two distinct genes: one can encode HBS1 or SKI7-like orthologs, while the other has subfunctionalized to encode only an HBS1-like protein, and not SKI7 (**Figure 4B**).

The HBS1/SKI7 orthologs are more diverse in the grasses, which have evolved distinct, subfunctionalized HBS1 and SKI7 loci. The grasses are divided into two major lineages: the PACMAD clade, which includes Z. mays (corn) and S. viridis (millet), and the BOP clade, which includes O. sativa (rice). Rice has three HBS1/SKI7 orthologs: Os04g50870, Os04g58140, and Os01g02720 (**Figures 5A**, **6B** and **Supplementary Data Sheet 2**). Os04g50870 and Os01g02720 are nearly identical paralogs (they encode proteins with 97% amino acid identity) that lack the SKI7-like and Patch 4-like motifs (**Figure 5**, "Oryza HBS1a" and "Oryza HBS1b"; **Supplementary Data Sheet 2**). Os04g50870 and Os01g02720 therefore most likely encode functional HBS1 proteins, but not functional SKI7 proteins. Os04g58140 has two splice forms (**Figure 6B**). Os04g58140.1 encodes a protein that includes the SKI7-like and Patch 4-like motifs (**Figures 5A**, **6B**). Os04g58140.2 retains an intron that includes a premature stop codon and a downstream alternative start codon (**Figure 6B** and **Supplementary Data Sheet 2**). If the alternative start codon is selected (either by skipping the upstream open reading frame (uORF) or by reinitiating translation after the uORF), the Os04g58140.2 protein contains only the Patch 4-like motif and the HBS1-like translational GTPase. The Os04g58140.2 transcript may also be subject to NMD, if the first start codon is selected and translation does not reinitiate at the alternative start codon. In either case, retention of the poison intron yields an unproductive transcript of SKI7.

In the PACMAD clade of grasses, S. viridis has two HBS1/SKI7 orthologs, which have apparently subfunctionalized: Sevir3g016200 encodes a complete SKI7-like protein, with the SKI7-like and Patch 4-like motifs, and Sevir9g199100 encodes a protein with only the negatively charged N-terminus, the Zn-finger domain, a poorly-aligned/non-consensus Patch 4-like motif, and the C-terminal HBS1-like GTPase (**Figures 4C**, **5A** and **Supplementary Data Sheet 2**). RNA-Seq analysis suggests that neither of these transcripts is alternatively spliced, although there are relatively limited data for S. viridis compared to well-established model systems, like tomato and Arabidopsis. Z. mays has orthologs of both S. viridis genes (Zm00001d026213 is orthologous to Sevir9g199100, and Zm00001d001827 is orthologous to Sevir3g016200, **Figure 4C**), but its SKI7-like gene makes many distinct transcripts. Alternative 5<sup>0</sup> and 3 0 splice sites near the beginning of the coding sequence

select between two different start codons, but only alter the N-terminus by eight amino acids. More importantly, some of these transcripts (such as Zm00001d001827.2) encode an entire SKI7-like protein, while others (such as Zm00001d001827.9) skip an exon that encodes the patch 4 motif (**Figure 4C**, lower panel), similar to S. moellendorffi LOC9652039 variant 3 (**Figure 6A**).

# Defining Conserved SKI7 Protein Features by Homology

To assemble all reliable SKI7-like amino acid sequences, we queried the NCBI RefSeq\_protein database<sup>1</sup> for proteins with SKI7-like sequences in land plants and metazoans (**Supplementary Data Sheets 2**, **3**, respectively). Using this approach, we found a number of previously unidentified SKI7-like proteins in divergent invertebrate lineages, including orthologs in cnidarians, echinoderms, cephalochordates, mollusks, brachiopods, and priapulids (**Supplementary Data Sheet 3**). These findings suggest that SKI7-like orthologs are probably ubiquitous in eukaryotes, although deeper sequencing of transcriptomes from diverse phylogenetic clades will be needed to fully support this hypothesis, as well as to determine when the metazoan SKI7-like proteins lost the C-terminal HBS1-like GTPase and how the

<sup>1</sup>https://www.ncbi.nlm.nih.gov/refseq/

HBS1/SKI7 gene/transcript structures evolved in the metazoans (**Figure 2A**).

We aligned SKI7-like sequences from metazoans and land plants to identify any conserved regions that could illuminate how sequences specific to SKI7, but not HBS1, determine its distinct functions (**Figure 7B** and **Supplementary Data Sheets 2**, **3**). There are two clearly conserved motifs across all SKI7 orthologs, which were previously named RxxxFxxxL and PFDFxxxSPDDIVKxNQ, based on the human SKI7-like protein sequence, and which we have named the SKI7-like motif and the Patch 4-like motif, respectively (**Figure 7**). Homology modeling of HsSKI7 onto the well-studied structure of the Rrp6/Rrp43 interaction revealed that the SKI7-like motif likely docks HsSKI7 with Rrp43, a core RNA exosome subunit (Kalisiak et al., 2017). The critical yeast SKI7-like motif residues, RxxxFxxxL, are not conserved across all eukaryotes, however; the consensus sequence at this site in metazoans is A+PShFAahL (where + is a positively charged residue, h is a hydrophobic residue, and a is an aliphatic residue; **Figure 7B**). In plants, this consensus sequence is slightly different: A+SLFaaa[P/L] (where + is a positively charged residue, a is an aliphatic residue, and [P/L] is usually P or L; **Figure 7B**). The Patch 4-like motif is fairly similar across all metazoans and plants. In metazoans, we found that the consensus Patch 4-like sequence is IaPF[D/R]F[K/D][S/T]aSPDDIV+A (**Figure 7B**). In plants,

the consensus Patch 4-like sequence is IaPFKFDaPSPDDhVxx (**Figure 7B**). According to the resolved cryo-EM structure of yeast SKI7 in complex with the exosome (Kowalinski et al., 2016), the Patch 4-like motif mediates interactions with the Csl4 RNA exosome subunit. It should be noted that co-immunoprecipitation experiments in humans suggest that the Patch 4-like motif is neither necessary nor sufficient to recruit HsSKI7 to the RNA exosome, but the SKI7-like motif (A+PShFAahL) and neighboring residues were necessary and sufficient to recruit HsSKI7 or GFP to the RNA exosome (Kalisiak et al., 2017). Nonetheless, loss of the Patch 4-like motif did apparently weaken the interaction between SKI7 and the RNA exosome.

We used homology modeling to predict the structure of the N-terminus of the Arabidopsis SKI7 protein (excluding the C-terminal trGTPase, **Figure 8**). The negatively charged N-terminus is predicted to form several α helices, which may mediate interactions with the SKI2/SKI3/SKI8 adaptor complex, and the ZnF is predicted to form two β-sheetlike structures (β1 and β2, **Figure 5A**) followed by an α-helix, as expected for a ZnF domain. The exons encoding the SKI7-like motif and Patch 4-like motif fold into four α-helices (α1 through α4, **Figure 5A**), very comparable to the resolved yeast SKI7 structure (Kowalinski et al., 2016). α1 overlaps with the deeply conserved SKI7 like motif (in Arabidopsis, AKSLFGSVP, **Figure 8A**). α3 is highly similar between humans and Arabidopsis (**Figure 8A**), including residues DLYKAF (Arabidopsis) or DLYKTF (human), and this α-helix has been labeled in the predicted Arabidopsis SKI7 structure (**Figure 8B**). The Patch 4-like motif is predicted to be highly structured, and forms α4. Immediately after α4, the last amino acids of this N-terminal region of Arabidopsis SKI7 are highly disordered (this pattern continues into the N-terminal residues of the trGTPase region of the protein), allowing the trGTPase to adopt a flexible position relative to the highly structured N-terminus that interacts with the RNA exosome (**Figure 8B**).

# DISCUSSION

Here, we have shown that HBS1/SKI7 is a well-conserved locus in eukaryotes that encodes two proteins with distinct molecular functions. In plants and fungi, HBS1 and SKI7 are nearly identical proteins, with an N-terminus that interacts with the cytosolic RNA exosome SKI2/SKI3/SKI8 complex and a C-terminal translational GTPase. The SKI7 isoform differs from HBS1 by as few as ∼25 amino acids that we propose promote its interaction with the RNA exosome instead of with the ribosome decoding factor, PELOTA (**Figure 9**).

Although we have shown that many plant genomes encode both HBS1 and SKI7 from a single locus by alternative splicing of an exon encoding the SKI7-like motif, other lineages have evolved distinct HBS1/SKI7 gene structures. In the earlydiverging tracheophyte S. moellendorffi, HBS1/SKI7 can be alternatively spliced to exclude or include exons encoding either the SKI7-like motif or the Patch 4-like motif, or to exclude or include both of these exons. In the Solanaceae, a second HBS1 locus has lost the SKI7-like and Patch 4 like motifs, and thus encodes only an HBS1-like protein. In rice, HBS1 and SKI7 are each encoded by their own locus, reminiscent of the subfunctionalization of Hbs1 and Ski7 loci in S. cerevisiae. Panicoid grasses (maize and millet), like rice, have two distinct HBS1 and SKI7 loci: one encodes only HBS1, and the second can encode SKI7. The panicoid grass SKI7 locus is alternatively spliced to include or exclude the Patch 4-like motif, however, which may impact its function. The Patch 4-like motif is universally conserved in all known SKI7 orthologs, but is not conserved in HBS1 orthologs that have lost other SKI7-like features (e.g., the HBS1-specific loci in Solanaceae, rice, and yeast), suggesting that the Patch 4 like motif is critical for SKI7's functions. In humans, however, Patch 4-like is neither necessary nor sufficient to recruit SKI7 to the RNA exosome (Kalisiak et al., 2017). Thus, while it seems most likely that the panicoid grass SKI7-like protein loses functionality when the exon encoding the Patch 4-like motif is excluded by alternative splicing, this will need to be determined experimentally.

Consideration of the evolutionary history of HBS1/SKI7 loci in eukaryotes reveals an important distinction: subfunctionalized HBS1 paralogs that cannot encode SKI7 have evolved repeatedly, and are often no longer regulated by alternative splicing, but in almost all eukaryotes (except for S. cerevisiae and a handful of other fungi), SKI7 orthologs are alternatively spliced (Marshall et al., 2013; Lambert et al., 2015). Moreover, alternative splicing of SKI7 consistently has strong effects, either yielding an unproductive splice form that is likely degraded by NMD, a splice form that encodes a lossof-function protein, or a transcript that instead encodes HBS1. This consistent regulation of SKI7 levels by alternative splicing suggests that the activity of the cytosolic RNA exosome is tightly coordinated with nuclear RNA processing, especially conditions that shift alternative splicing dynamics, such as oxidative stress (Staiger and Brown, 2013; Berner et al., 2017). Further characterization of the developmental or physiological conditions that influence alternative splicing of SKI7, as well as assays to determine whether SKI7 protein levels are limiting factors in cytosolic RNA exosome activity, will be needed to unravel how this mechanism influences cytosolic RNA exosome activity. Recently, the Pelota/HBS1 decoding factors and the cytosolic RNA exosome have been implicated in promoting NMD (Arribere and Fire, 2018), and NMD is known to regulate expression of the splicing and translation machinery (Lareau and Brenner, 2015). Our finding that alternative splicing of SKI7 potentially regulates SKI7 levels to limit assembly of the SKI2/SKI3/SKI8-RNA exosome complex (by either excluding an alternative exon of SKI7 to encode HBS1, or by generating a nonsense SKI7 transcript) invites speculation that SKI7 regulates its own splicing and transcript stability via its role in NMD and NMD-mediated regulation of splicing machinery gene expression.

PELOTA and the cytoplasmic RNA exosome are emerging as crucial components of plant immune systems, although their necessary interacting partners, HBS1 and SKI7, have not been comprehensively defined in plants until now. Loss of PELOTA, the decoding factor that recruits HBS1 to stalled ribosomes, confers resistance to TYCLV infection in tomato; with our identification of tomato HBS1 and HBS1/SKI7 genes, it is now possible to test whether loss of HBS1, SKI7, and/or SKI2/SKI3/SKI8 also confer resistance to TYCLV, and how this RNA degradation machinery interacts with other viruses. In rice, a recessive pelota mutant triggers a salicylic acid-associated autoimmune response, including spontaneous lesions and dwarfism, through unclear mechanisms (Ding et al., 2018; Qin et al., 2018; Zhang et al., 2018). The rice pelota defects could be related to hyperaccumulation of aberrant RNA species, similar to the tricohepatoenteric autoimmune syndrome in skiv2l human cells (Eckard et al., 2014), or due to specific dysregulation of transcripts in rice that regulate immunity. For instance, the expression and activity of the disease resistance Toll- and Interleukin-like Receptor (TIR) family of Nucleotide-binding, Leucine-rich repeat Receptors (TIR-NLRs or TLRs) is regulated by NMD in some instances; loss of PELOTA, which contributes to NMD, may therefore deregulate TLR expression, triggering autoimmune defects (Dinesh-Kumar and Baker, 2000; Riehs-Kearnan et al., 2012; Gloggnitzer et al., 2014). Whether pelota mutants can trigger autoimmune defects in other plant species remains to be determined. More broadly, NMD is proposed as a general antiviral mechanism, and so HBS1 and SKI7 may contribute to broad-spectrum antiviral defense via their roles in NMD (Balistreri et al., 2014; Garcia et al., 2014; Rigby and Rehwinkel, 2015; Hamid and Makeyev, 2016).

# CONCLUSION

We have identified the plant orthologs of HBS1 and SKI7, key regulators of RNA metabolism in eukaryotic cells. As a component of the cytosolic RNA exosome, SKI7 not only participates in co-translational RNA surveillance, but is also presumably required to clear 5<sup>0</sup> fragments of mRNAs cleaved by RISCs. RNA exosomal degradation of these 5<sup>0</sup> fragments prevents copying of host transcripts by RDR6, which can otherwise trigger post-transcriptional silencing of endogenous genes. In diverse eukaryotic lineages, SKI7 levels are controlled by alternative splicing of transcripts; alternative splice forms can either encode the functionally distinct HBS1 protein, or can be unproductive, either by removing critical residues for SKI7 function, or by introducing premature stop codons that likely subject the splice form to NMD. Co-translational RNA decay mechanisms, including HBS1/SKI7-dependent RNA degradation, are becoming more prominent to investigations of eukaryotic immune systems and defenses against viral infection. Our discovery of the alternative splicing of HBS1/SKI7 expression across anciently diverging eukaryotic lineages, including plants and invertebrate clades, implies that co-translational RNA decay mechanisms are under complex regulation to coordinate host gene expression with environmental cues, stress responses, and antiviral defense.

# MATERIALS AND METHODS

# Plant Materials

The Landsberg erecta (Ler) ecotype of Arabidopsis was grown under 16 h light (100 µmol photons m−<sup>2</sup> s <sup>−</sup><sup>1</sup> of photosynthetically active radiation)/8 h dark cycles. Shoots were harvested 4 weeks after germination.

# RT-PCR and TOPO Cloning

RNA was isolated from Ler Arabidopsis plants with the Spectrum Plant Total RNA (Sigma-Aldrich) kit with on-column DNase I digestion (New England Biolabs). cDNA was synthesized from RNA using oligo (dT)18 primers and SuperScript III reverse transcriptase (Fisher Scientific). Splice forms were amplified with Phusion DNA polymerase (New England Biolabs), adding a CACC 5<sup>0</sup> overhang to facilitate pENTR/D-TOPO cloning. RT-PCR amplified DNA was gel purified in a 1% agarose gel and extracted using a gel extraction kit (Bioneer). Purified DNA was used for TOPO reactions with pENTR/D-TOPO (Thermo Fisher), transformed into XL1-Blue E. coli chemically competent cells, and screened for resistance to kanamycin on LB agar. Plasmid was purified from positive colonies using a miniprep kit (Bioneer) and sequenced using Sanger technology with the M13F primer.

Oligonucleotides used for cloning were: JB1058: 5<sup>0</sup> -CACC ATG CCT CGT AAA GGA TTA TCC AAT TTC G-3<sup>0</sup> , JB1061: 5 0 -CACC ACA GTT GAG AGCAG ATG CAA AGA AC-3 0 , and JB1063: 5<sup>0</sup> -GCC TTT TGG ACC AGT TTT TGAGG ATG-3<sup>0</sup> . JB1058 + JB1063 surround the alternative exon, and amplified three majors products: the short splice form, the long splice form, and a small amount of the longer, minor splice form. JB1061 is specific to the 5<sup>0</sup> end of the alternative exon, and in combination with JB1063, amplified both the long splice form and the longer, minor splice form.

# Computational Analysis

Decoding trGTPases were identified using human protein sequences as queries for a BLASTp search against the Arabidopsis thaliana refseq protein database. Putative trGTPases were filtered to include only cytosol-localized proteins, based on proteomic data and consensus predictions curated by the Subcellular Localization Database for Arabidopsis Proteins 3<sup>2</sup> .

SKI7 orthologs in the NCBI protein refseq databases were identified using a tBLASTx search with the Arabidopsis

<sup>2</sup> suba3.plantenergy.uwa.edu.au

SKI7-like and patch 4-like motifs as a query for land plants, and the human SKI7-like and patch 4-like motifs as a query for metazoans. BLASTp results were then filtered to remove identical protein sequences. Protein sequences were aligned using MAFFT via JalView. Uncommon insertions were trimmed from the final alignments for clarity. Transcript structures were obtained for the select plant species described in the text from relevant databases (TAIR10, from arabidopsis.org and araport.org; MaizeGDB.org; Phytozome.jgi.doe.gov; SolGenomics.net; and CosMoss.org), and then confirmed with RNA-Seq evidence from the same databases (or by direct cloning of alternative splice forms, as described above). Consensus sequence logos were generated with WebLogo<sup>3</sup> .

The N-terminus of SKI7 (through the patch 4-like motif) structure was modeled by Phyre2<sup>4</sup> , which used structures of YY1 associated factor 2 (PDB 2D9G), HBV-associated factor (PDB 2CRC), Rubredoxin B (PDB 2KN9), NEMO CoZi (PDB 4OWF), and TAB3-NZF (PDB 3A9K). The resulting model was visualized by NGL<sup>5</sup> .

3 http://WebLogo.berkeley.edu

4 http://www.sbg.bio.ic.ac.uk/∼phyre2

5 http://proteinformatics.charite.de/ngl-tools/ngl/html/ngl.html

## REFERENCES


# AUTHOR CONTRIBUTIONS

JB designed the project, conducted the experiments, and drafted the manuscript. BB contributed to the experimental design and manuscript.

# FUNDING

This project was supported by Innovative Genomics Institute 2017 Award to BB and JB, USDA CRIS 2030- 22000-009-00D to BB, and NIH grant 5-DP5-OD023072-02 to JB.

# ACKNOWLEDGMENTS

We thank Dr. Feng Li for stimulating discussions.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01333/ 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 Brunkard and Baker. 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 Verticillium dahliae Pectate Lyase Induces Plant Immune Responses and Contributes to Virulence

### Yuankun Yang, Yi Zhang, Beibei Li, Xiufen Yang, Yijie Dong\* and Dewen Qiu\*

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

Verticillium dahliae is a wide-host-range fungal pathogen that causes soil-borne disease in hundreds of dicotyledonous hosts. In search of V. dahliae Vd991 cell death-inducing proteins, we identified a pectate lyase (VdPEL1) that exhibited pectin hydrolytic activity, which could induce strong cell death in several plants. Purified VdPEL1 triggered defense responses and conferred resistance to Botrytis cinerea and V. dahliae in tobacco and cotton plants. Our results demonstrated that the mutant VdPEL1rec lacking the enzymatic activity lacked functions to induce both cell death and plant resistance, implying that the enzymatic activity was necessary. In addition, VdPEL1 was strongly induced in V. dahliae infected Nicotiana benthamiana and cotton roots, and VdPEL1 deletion strains severely compromised the virulence of V. dahliae. Our data suggested that VdPEL1 contributed to V. dahliae virulence and induced plant defense responses. These findings provide a new insight for the function of pectate lyase in the host–pathogen interaction.

Keywords: Verticillium dahliae, pectate lyase, elicitor, plant immunity, virulence

# INTRODUCTION

Plants are exposed to a multitude of pathogens that cause massive yield and quality losses annually. To ward off invading microorganisms, plants have evolved elaborate systems to provide better immunity against pathogens. Plants utilize pattern recognition receptors (PRRs) to perceive pathogen-associated molecular patterns (PAMPs) or damage-associated molecular patterns (DAMPs) and rapidly activate PAMP-triggered immunity (PTI). This recognition initiates a cascade of signaling events resulting in induction of a battery of immune responses, including Ca2<sup>+</sup> influx, the burst of reactive oxygen species (ROS), the accumulation of defense hormones, the expression of defense-related genes and callose deposition (Boller and Felix, 2009; Couto and Zipfel, 2016). In turn, during the coevolution of hosts and microbes, pathogens also employ numerous effectors to interfere with PTI and establish successful infection, which is regarded as effector-triggered susceptibility (ETS) (Chisholm et al., 2006; Jones and Dangl, 2006; Saijo et al., 2017). As an adaption to ETS, the effectors are recognized by the resistance (R) proteins, and subsequently lead to robust immunity termed effector-triggered immunity (ETI) (Houterman et al., 2008; Stergiopoulos and de Wit, 2009). Generally, ETI is associated with stronger immune responses, such as the hypersensitive response (HR). Plant immune responses initiate from local tissues located at the site of the infection and subsequently extend to other non-infected tissues by long-distance intercellular communications, generating a systemic acquired resistance (SAR) that is effective against a broad spectrum of pathogens in the whole plant (Kulye et al., 2012).

### Edited by:

Thomas Mitchell, The Ohio State University, United States

#### Reviewed by:

Raffaella Balestrini, Consiglio Nazionale delle Ricerche (CNR), Italy Philipp Franken, Leibniz-Institut für Gemüse- und Zierpflanzenbau (IGZ), Germany

#### \*Correspondence:

Yijie Dong nkdongyijie@163.com Dewen Qiu qiudewen@caas.cn

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 29 April 2018 Accepted: 14 August 2018 Published: 13 September 2018

#### Citation:

Yang Y, Zhang Y, Li B, Yang X, Dong Y and Qiu D (2018) A Verticillium dahliae Pectate Lyase Induces Plant Immune Responses and Contributes to Virulence. Front. Plant Sci. 9:1271. doi: 10.3389/fpls.2018.01271

The plant cell wall is the first barrier, which provides mechanical strength and rigidity to protect against pathogenic infection. Correspondingly, pathogens secrete numerous of cell wall-degrading enzymes (CWDEs) to permit them to invade plant tissue and supply themselves with nutrients (Kikot et al., 2009; Klöckner et al., 2016). On the one hand, CWDEs serve as virulence factor in pathogens and play an essential role in infection process. Previous studies suggested that the deletion of the Xylanase Xyn11A gene caused a marked effect on the ability of Botrytis cinerea to infect tomato leaves and grape berries (Brito et al., 2006). And an endoxylanase named xynB contributed to virulence in Xanthomonas oryzae pv. oryzae (Pandey and Sonti, 2010). Alternatively, plants can also sense CWDEs or self-molecules (released from plant cell wall polysaccharides) as inducers of immune responses (Misas-Villamil and van der Hoorn, 2008). Recently, it was widely reported that glycoside hydrolase 12 (GH12) proteins, VdEG1 and VdEG3 from Verticillium dahliae, XEG1 from Phytophthora sojae and BcXYG1, a secreted xyloglucanase from B. cinerea contributed to virulence and simultaneously triggered plant immunity as PAMPs (Gui Y.-J. et al., 2017; Zhu et al., 2017a,b). Oligogalacturonides (OGs) were regarded as DAMPs to activate the plant immune responses (De Lorenzo et al., 2011). The enzymatic activity of VdCUT11(a cutinase from V. dahliae) was required for activating immune responses in Nicotiana benthamiana, implying that the cutinase degradation products might induce the plant resistance (Gui Y. et al., 2017).

Pectin exists widely in the plant cell walls and cell lining and maintains wall integrity and cell–cell cohesion. Due to the complexity of this highly branched polysaccharide, the degradation of pectin requires a variety of enzymes, such as pectin lyases, pectate disaccharide-lyases, and pectate lyases. Among these, pectate lyases have received the most attention. Pectate lyases randomly cleave α-1,4-polygalacturonic acid via a β-elimination reaction. They can also macerate and disassemble the plant cell wall and tissues in a manner similar to that occurring in fungal diseases (Collmer, 1986). For instance, the deletion of the pectate lyase gene CcpelA and PelB in Colletotrichum coccodes, resulted in attenuated virulence on green tomato fruit and reduced susceptibility on avocado (Persea americana) fruit, respectively (Yakoby et al., 2001; Ben-Daniel et al., 2011). The deletion of the pectate lyase genes BcPG1 and BcPG2 also reduced virulence in B. cinerea (Have et al., 1998; Kars et al., 2005). However, pectic enzymes can also elicit plant defense responses through direct or indirect ways. For instance, a pectate lyase, from Erwinia carotovora bacteria, induced defense responses against Erwinia soft rot in potato plants (Wegener, 2002).T4BcPG1, an endopolygalacturonase from B. cinerea, activated grapevine defense reactions (Poinssot et al., 2003).

Verticillium dahliae is a wide-host-range pathogen that can infect a large number of dicotyledonous hosts, including ecologically important plants and many high-value crops worldwide such as cotton, tobacco, potato, and tomato (Fradin and Thomma, 2006; Klosterman et al., 2009; Inderbitzin et al., 2014). It infects plants primarily through the formation of hyphae, which can puncture the plant root surface and colonize in xylem vessels (Zhao et al., 2014). It was demonstrated that V. dahliae secrets a large amount of CWDEs, including 13 pectate lyases. However, the specific roles of the pectate lyases in V. dahliae remain largely unknown.

The main objectives of the current study were to: (1) isolate and investigate cell death-inducing proteins in V. dahliae Vd991; (2) determine whether VdPEL1 is secreted into the apoplast in order to induce cell death; (3) examine the relationship between the enzymatic activity and cell death-inducing activity; and (4) investigate the function of VdPEL1 in immune responses and virulence.

# MATERIALS AND METHODS

# Fungal Cultures, Plants Grown

The V. dahliae strains, including Vd991 wild type strains, targeted deletion strains and complementary transformants, were maintained and cultured on PDA media or in liquid Czapek media for 7 days at 25◦C. B. cinerea strain BC-98 and Agrobacterium tumefaciens AGL-1 were grown on PDA media at 25◦C and LB (Kan and Rif) medium at 28◦C, respectively. Cotton (Gossypium hirsutum cv. Junmian 1) and N. benthamiana were grown at 23 and 27◦C, respectively, in a greenhouse with a day/night period of 14/10 h.

# Separation, Purification, and Characterization of Proteins Secreted From V. dahliae

To produce large amounts of secreted proteins, Vd991 was grown in Czapek media at 25◦C for 14 days, shaken at 150 rpm (Bailey, 1996), and then filtered through two layers of filter paper. The supernatants were collected and precipitated with 70% (NH)4SO<sup>4</sup> overnight at 4◦C. The precipitate was collected by centrifugation at 15,000 × g for 10 min at 4◦C, and then resuspended in 10 mM Tris-HCl (pH 7.5) and 1 mM EDTA (TE). Crude protein was further purified by ion-exchange chromatography and eluted with a linear sodium chloride gradient from 0.0 to 1.0 M in TE. Fractions corresponding to absorbance peaks were desalted and concentrated using a 10-kDa ultrafiltration device and tested their ability to induce cell death activity. The fraction with cell death activity was excised from the SDS-PAGE gel and identified using mass spectrometry (MS) analysis (Beijing Protein Innovation, Beijing, China) as described (Heese et al., 2007). The tandem MS + MS/MS data were automatically analyzed using the Mascot search engine (Matrix Science, London, United Kingdom).

# Cloning, Expression, and Purification of Recombinant Protein

VdPEL1 fragment (amplified with primers VdPEL1F/VdPEL1R; **Supplementary Table S2**) without the predicted signal peptides and stop codons was inserted into the pPICZαA plasmid at the BamHI and EcoRI sites (Ma et al., 2015). The recombinant plasmid pPICZαA-VdPEL1 was linearized with PmeI and transformed into Pichia pastoris KM71H for expression. The transformed yeasts were grown and induced in BMGY (buffered glycerol-complex medium) and BMMY (buffered methanolcomplex medium), respectively (Easy Select Pichia Expression Kit; Invitrogen). The supernatant was collected (3,000 × g for 10 min at 4◦C) and then purified using nickel affinity chromatography (Ma et al., 2015; Zhang et al., 2017). The purified VdPEL1 or VdPEL1rec were kept in protein buffer (20 mM Tris, pH 8.0) and further detected via SDS-PAGE and Western blotting. The concentration of the purified protein was measured using an Easy II Protein Quantitative Kit (BCA) and the protein was then stored at −80◦C.

# Site-Directed Mutagenesis

fpls-09-01271 September 11, 2018 Time: 18:47 # 3

To examine the relationship between the enzymatic activity and cell death-inducing activity of VdPEL1, we constructed a VdPEL1rec mutant with loss of the enzymatic activity. According to multiple sequence alignment, we found that Asp-125 and Asp-147 residues might be the critical catalytic sites of VdPEL1. Two Asp residues of VdPEL1 were replaced by Ala residues using the Quick ChangeTM Site-Directed Mutagenesis Kit (Stratagene, United States) to cause the loss of pectate lyase activity of VdPEL1. Primers used in these constructions are listed in **Supplementary Table S2**. The VdPEL1rec mutant was confirmed by sequence alignment and the enzyme assays.

# Enzyme Activity Assays

Hydrolase activity was measured using the 2-cyanoacetamide spectrophotometric method (Bach and Schollmeyer, 1992). The purified VdPEL1 or VdPEL1rec (500 ng) and substrate (2.5% polygalacturonic acid) were co-incubated at 30◦C in a medium buffered with MES or Tris 10 mM at indicated pH (total volume: 350 µl). Fifteen minutes later, all samples were incubated for 10 min at 100◦C to terminate the assays, and the reduced sugars released by VdPEL1 or VdPEL1rec were measured at 274 nm using a spectrophotometer. The reduced sugars were quantified using a standard calibration curve obtained with polygalacturonic acid. The experiment was repeated three times.

# Immunoblot Analysis

To confirm whether VdPEL1 was secreted into the apoplast, transient expression in N. benthamiana was performed. Three sequences, encoding VdPEL1 protein with the putative signal peptide, VdPEL1 protein without the putative signal peptide and PR1 SP-VdPEL121−<sup>255</sup> protein replaced the signal peptide from pathogenesis-related protein 1 (PR1), were amplified with primers VdPEL1-T-F/VdPEL1-T-R, VdPEL121- 255-F/VdPEL121-255-R, and PR1 SP-VdPEL121-255-F/PR1 SP-VdPEL121-255-R, respectively (**Supplementary Table S2**). Three sequences were cloned into the pYBA1132 vector which contained a C-terminal GFP tag at the XbaI and BamHI sites (Yan et al., 2012). To confirm whether the enzymatic activity of VdPEL1 was related to the cell death-inducing activity, the sequences encoding the mature VdPEL1 protein and site-directed mutagenized VdPEL1rec with the putative signal peptide (amplified with primers VdPEL1F/VdPEL1R and VdPEL1rec F/VdPEL1rec R, **Supplementary Table S2**) were cloned into pYBA1132 vector at the XbaI and BamHI sites, and then transformed into the A. tumefaciens strain GV3101. Agroinfiltration assays were performed on N. benthamiana plants. To determine whether fusion proteins were expressed, plant total protein extractions, and immunoblots were assessed as previously described (Yu et al., 2012). All the proteins were analyzed via immunoblots using anti-GFP-tag primary monoclonal antibody. The blots were visualized using the Odyssey <sup>R</sup> LI-COR Imaging System. Rubisco was used to confirm the equal protein loading.

# Elicitor Activity and Systemic Resistance Assays

To detect the induction of cell-death, 300 nM purified VdPEL1 or VdPEL1rec, PEVC and BSA were injected into leaves of 4 week-old N. benthamiana plants and 2-week-old plants of cotton, tomato, and soybean with an injector. These plants were kept in a glasshouse with a day/night period of 14/10 h, and celldeath responses were investigated after 48 h of treatment with the recombinant proteins. The accumulation of ROS in the plant leaves was stained by using 3<sup>0</sup> 3-diaminobenzidine (DAB) solution as described previously (Bindschedler et al., 2006). To visualize callose deposition, 4-week-old N. benthamiana leaves were treated with 300 nM recombinant proteins and stained with aniline blue at 24 h post-treatment, as described previously (Chen et al., 2012). To assay electrolyte leakage, 4 week-old N. benthamiana leaves were infiltrated with 300 nM purified VdPEL1 or VdPEL1rec and PEVC. The corresponding N. benthamiana leaves at different time points were harvested and submerged in sterile water at 4◦C. Ion conductivity was measured using a conductivity meter.

The purified VdPEL1 or VdPEL1rec and PEVC were individually syringe-infiltrated into 4-week-old N. benthamiana leaves. A total of 5 µl of 2 × 10<sup>6</sup> conidia/ml B. cinerea or 1 × 10<sup>6</sup> conidia/ml V. dahliae were placed on the infiltrated area or inoculated by the root-dip method, respectively. The inoculated plants were placed in a greenhouse with a day/night period of 14/10 h. The lesion development of B. cinerea on the N. benthamiana leaves was evaluated at 2 days post-inoculation by determining the average lesion diameter. In addition, disease symptoms were observed at 12 days V. dahliae post-inoculation on N. benthamiana. All the experiments were performed three times.

# Generation of VdPEL1 Deletion and Complementary Mutants

The wild-type VdPEL1 gene and 500 bp flanking sequences of the target gene were amplified from the V. dahliae Vd991 genomic DNA (gDNA). Two flanking sequences of the target gene and a hygromycin resistance cassette were constructed into a fusion fragment using a nested PCR reaction, which was subsequently introduced into the binary vector pGKO2-Gateway. To generated complementary transformants, the donor vector pCT-HN containing VdPEL1 gene was integrated into the mutant transformants using an Agrobacterium-mediated transformation method (Liu et al., 2013). All the mutants were identified using PCR with the corresponding primers.

# RNA Extraction and qRT-PCR

fpls-09-01271 September 11, 2018 Time: 18:47 # 4

To measure the expression of VdPEL1 gene during infection, 4 week-old N. benthamiana plants or 2-week-old cotton seedlings were inoculated with 1 × 10<sup>6</sup> V. dahliae conidia/ml or 5 × 10<sup>6</sup> conidia/ml by the root-dip method, respectively. We selected 10 indicated time points during different stages of postinoculation to determine expression patterns of VdPEL1 using quantitative PCR (qPCR). All the samples were obtained at the time points indicated and stored at –80◦C. Total RNA from V. dahliae infected plants was extracted with the E.Z.N.A. <sup>R</sup> Total Fungal RNA Kit I. To measure the expression of defenserelated genes, the leaves of 4-week-old N. benthamiana plants were treated with 300 nM purified VdPEL1 or VdPEL1rec and PEVC. The leaves were obtained at the time points indicated, immediately frozen in liquid nitrogen, and stored at −80◦C. A EasyPure Plant RNA Kit (TransGen Biotech) was used to extract the total plant RNA. The gDNA was digested by the DNase I (TransGen Biotech). And the gDNA remover was added when the cDNA was synthesized. To further investigate whether or not the gDNA was absolutely removed, the gene encoding the actin (GenBank: X63603.1) in tobacco and β-tubulin (VDAG\_10074) in V. dahliae were detected by PCR, respectively.

qRT-PCR was performed using a TransStart Green qPCR SuperMix UDG according to the manufacturer's instructions (TransGen Biotech). qRT-PCR was performed under the following conditions: an initial 95◦C denaturation step for 10 min followed by 40 cycles of 95◦C for 15 s and 60◦C for 1 min. The cotton 18S gene, N. benthamiana EF-1a and β-tubulin (VDAG\_10074) of V. dahliae were used as endogenous plant controls and used to quantify fungal colonization, respectively. All primers are listed in **Supplementary Table S2**. The relative transcript levels among various samples were determined using the 2−11CT method with three independent determinations (Livak and Schmittgen, 2001).

# Pathogenicity Assays

To test whether VdPEL1 functioned as a virulence factor of V. dahliae, the wild-type strain and derived mutants, including the VdPEL1 deletion and complementary mutants were used in this study. Four-week-old N. benthamiana plants or 2-weekold cotton seedlings were inoculated with 1 × 10<sup>6</sup> conidia/ml or 5 × 10<sup>6</sup> conidia/ml by the root-dip method, respectively (Zhou et al., 2013). After 21 days post-inoculation on cotton or 14 days post-inoculation on N. benthamiana, disease symptom and fungal biomass was determined as previously described (Santhanam et al., 2013). Discoloration in vascular tissues was observed by cutting root longitudinal sections at 3 weeks after inoculation. Real-time qPCR was performed using a qPCR SYBR premix Ex Taq II kit (TaKaRa, Kyoto, Japan). t-tests were performed to determine statistical significance at p-values <0.05 between two treatments groups.

# Statistical Analysis

All the experiments and data presented here were performed at least three repeats. The data are presented as the mean and standard deviations. Statistical Analysis System (SAS) software was used to perform the statistical analysis via Student's t-test.

# RESULTS

# Identification, Purification, and Characterization of VdPEL1

To identify the defense response proteins from V. dahliae, the induction of cell death in N. benthamiana leaves was used as an index to fractionate proteins from culture filtrates via anion exchange chromatography. Due to different affinities for chromatographic column, fractionation of a 70% ammonium sulfate precipitate generated three primary peaks (**Figure 1A**). Fractions corresponding to peak 1 could induce cell death in N. benthamiana leaves (**Figure 1B**). Further fractionate and purify, SDS-PAGE analysis showed a single obvious band between 25 and 35 kDa (**Figure 1C**). To identify amino acid sequence of the peak 1 band, tryptic digestion was performed, and the peptides generated were analyzed by liquid chromatography-MS. Many peptides with high credibility were co-matched a protein encoded by the gene (Gene ID: 20707618) (**Supplementary Figure S1**). The bioinformatics analysis found that the gene encodes a pectate lyase in V. dahliae Vd991, and we designated this protein as VdPEL1. Transient expression of VdPEL1 in N. benthamiana leaves showed that the protein triggers cell death 5 days after infiltration (**Figure 1D**). Immunoblot analysis using a-GFP antibody detected the expression of VdPEL1 in the leaves (**Figure 1E**).

# Nucleotide Sequence Analysis of VdPEL1

The open reading frame of VdPEL1 is 765 bp encoding a 255 amino acid protein. The first 20 N-terminal amino acids encode a signal peptide (Signal IP 4.1 server), and no transmembrane helices of VdPEL1 were found, indicating that the protein was likely to be secreted to the extracellular space. VdPEL1 has a conserved domain (39–229 amino acids) and belongs to the pectate lyase super family similar to pectate lyase A. BLAST results showed VdPEL1 homologues were present in a large number of pectate lyases of necrotrophic and hemibiotrophic plant pathogens. Phylogenetic analysis showed that all V. dahliae pectate lyase members were sorted into four distinct groups comprising five, two, one, and five pectate lyases each, respectively (**Figure 2**). The reference pectate lyases known to be involved in virulence and triggering defense responses in fungi and bacterial were also distributed into four branches. These results suggested that the function of pectate lyases members is significantly diverse in V. dahliae.

# VdPEL1 Induces Cell Death in Several Plants

To examine the necrosis-inducing activity of VdPEL1 in N. benthamiana, VdPEL1 was expressed in the P. pastoris using the pPICZaA vector (pPICZαA: VdPEL1), which could secrete proteins into the culture media. Purified VdPEL1, with a size of 29 kDa (**Supplementary Figure S4**), was infiltrated into

the mesophyll of N. benthamiana leaves using a syringe with different concentrations from 100 nM to 1 µM. The necrosis area occurred and increased with increasing concentrations of VdPEL1 after infiltration for 2 days compared with bovine serum albumin (BSA) or PEVC (P. pastoris culture supernatant from an empty vector control strain, purified in the same manner as VDPEL1), which had no HR response at 1 µM and even at 10 µM (**Figure 3A**). In parallel, the HSR203J gene and HIN1 genes, which are described for HR-marker genes in tobacco plants (Takahashi et al., 2004), were significantly induced expression in VdPEL1 treated leaves (**Figure 3B**), suggesting that VdPEL1 can trigger a severe HR. To examine the host specificity of

VdPEL1, we infiltrated VdPEL1 (300 nM) into the leaves of various plant species. The results demonstrated that VdPEL1 induced localized cell death in tomato, soybean (Glycine max), and cotton (Gossypium hirsutum) plants (**Figure 3C**). Therefore, VdPEL1 had a necrosis-inducing activity in diverse plant species.

# VdPEL1 Is Secreted Into the Apoplast in Order to Induce Cell Death in N. benthamiana

Bioinformatic analysis showed that VdPEL1 has a signal peptide with 20 amino acids and no transmembrane helices, implying

inoculation with Agrobacterium strains. (E) Immunoblot analysis of proteins from N. benthamiana leaves transiently expressing GFP and VdPEL1.

that VdPEL1 was likely to be an extracellular protein. To test whether VdPEL1 was localized to the plant apoplast to induce cell death, we constructed three A. tumefaciens strains: VdPEL121−<sup>255</sup> (deleted the N-terminal signal peptide), VdPEL1 (complete protein with the N-terminal signal peptide), and PR1 SP-VdPEL121−<sup>255</sup> (replaced the signal peptide from PR1) (**Figure 4A**). All the strains were infiltrated into N. benthamiana leaves. As expected, VdPEL1 and PR1 SP-VdPEL121−<sup>255</sup> were secreted from the plant cells to the apoplast and developed the typical necrosis, whereas expression of VdPEL121−<sup>255</sup> (remaining inside the cell) and green fluorescent protein (GFP as a negative control) didn't appear to cause a cell death response (**Figure 4B**). Western blot assays showed that the accumulation of all the examined proteins was similar in N. benthamiana 5 days post-infiltration (**Figure 4C**). These results indicated that VdPEL1 must be secreted into the apoplast to trigger cell death.

# The Enzymatic Activity of VdPEL1 Is Required for Cell Death-Inducing Activity in N. benthamiana

Previously, pectate lyases from fungi were shown to degrade polygalacturonic acid via a β-elimination reaction (Kashyap et al., 2001). We assessed the hydrolase activity of VdPEL1 by determining the level of reducing sugar using polygalacturonic acid as substrates. We found that purified VdPEL1 had a hydrolase activity, which was affected by the temperature, Ca2<sup>+</sup> concentration and pH (**Supplementary Figure S2**). Calcium (Ca2+) is known to be an essential co-factor for pectate lyases activity. The region around Ca2<sup>+</sup> is believed to be the catalytic center site, especially aspartic acids, which are present in diverse members of the pectate lyase family (Heffron et al., 1995). As shown in **Supplementary Figure S3**, VdPEL1 contained two corresponding catalytic residues (D<sup>125</sup> and D147) by sequence alignment. To examine the relationship between the enzymatic activity and cell death-inducing activity of VdPEL1, we replaced D<sup>125</sup> and D<sup>147</sup> with alanine (Ala) residues using site-directed mutagenesis and expressed the mutant proteins (VdPEL1rec) in P. pastoris (**Figure 5A** and **Supplementary Figure S4**). A hydrolase assay showed that pectate lyase activity of VdPEL1rec was abolished (**Supplementary Table S1**). In addition, purified VdPEL1 induced visible cell death in N. benthamiana, at a much higher level than the cell death symptoms after infiltration with VdPEL1rec (**Figure 5B**). We transiently expressed the wild type (VdPEL1) and mutant (VdPEL1rec) proteins in N. benthamiana. As expected, VdPEL1 triggered the cell death symptoms, while VdPEL1rec and GFP lacked the ability to produce necrosis at 5 days after agroinfiltration (**Figure 5B**). Two proteins in N. benthamiana were confirmed using immunoblot analysis (**Figure 5C**). These results suggested that the enzymatic activity of VdPEL1 is required for cell death-inducing activity in N. benthamiana.

# VdPEL1 Triggers Defense Responses and Systemic Resistance in N. benthamiana and Cotton

Plant immune system recognizes many cell death inducing proteins and activates host PTI, brings a series of typical characteristics such as accumulation of ROS, leakage of ion electrolytes, expression of defense genes, and callose deposition (Brutus et al., 2010; Zhang et al., 2014). To probe whether VdPEL1 induced immunity response, we first detected the accumulation of ROS. We observed intense staining in tobacco leaves treated with VdPEL1 (300 nM), whereas no DAB signal was detected in VdPEL1rec and PEVC treated leaves

leaves with purified 300 nM VdPEL1 and 300 nM BSA. Two days after post-infiltration, different plants leaves were photographed.

(**Figure 6A**). Meanwhile, VdPEL1 also induced electrolyte leakage and displayed an increase in conductivity over time, while VdPEL1rec or PEVC exhibited barely any change at the same concentration (**Figure 6B**). In addition, we examined the transcriptional induction of defense-responsive genes PR-1a and PR-5, which are involved in the SA-dependent defense pathway, PAL (phenylalanine ammonia lyase), NPR1 (the nonexpressor of PR1), and COI1 (CORONATINE INSENSITIVE 1), which is JA-responsive. As expected, the transcript levels of these defense genes were significantly up-regulated in N. benthamiana 24 h after treatment with VdPEL1 (**Figure 6C**). And VdPEL1rec induced a slight up-regulation of defense genes (NPR1 and PR-5) expression. Callose deposition was detected by aniline blue staining 24 h after VdPEL1, VdPEL1rec, PEVC, or flg22 treatment. N. benthamiana leaves inoculated with VdPEL1 or flg22 exhibited strong callose deposition compared with those inoculated with VdPEL1rec or PEVC, all of which exhibited low or undetectable levels of callose deposition (**Figure 6D**).

Furthermore, to explore whether VdPEL1 conferred plants disease resistance, N. benthamiana leaves were pretreated with 300 nM recombinant protein VdPEL1, VdPEL1rec or PEVC, and 24 h later, they were inoculated with B. cinerea spore suspension. Leaves pretreatment with VdPEL1 restricted the development of B. cinerea infection, but leaves pretreatment with VdPEL1rec or PEVC displayed obvious lesion (**Figure 7A**). In addition, a histogram showed determination of B. cinerea lesion diameter (**Figure 7B**). In parallel, tobacco and cotton plants pretreatment with VdPEL1 had more resistance to V. dahliae, and significantly fewer verticillium wilt symptoms and fungal biomass compared with the plants treated with VdPEL1rec or PEVC controls (**Figures 7C–F**). These results collectively suggested that VdPEL1 had the capacity to trigger plant defense responses and confer disease resistance in N. benthamiana and cotton.

# VdPEL1 Contributes to the Pathogenicity of V. dahliae

Pectate lyases are generally involved in the pathogenicity of fungi (Collmer, 1986). To determine the possible contribution of VdPEL1 to V. dahliae virulence, we assessed the expression patterns of VdPEL1 during different stages of post-inoculation. VdPEL1 was strongly expressed in tobacco and cotton at

vector.

fpls-09-01271 September 11, 2018 Time: 18:47 # 8

1.5–3 days after inoculation, and then sharply declined from 3 days onward (**Supplementary Figure S5**). Next, we generated two independent VdPEL1 deletion lines (1VdPEL1-1 and 1VdPEL1-2) and the complementary transformants (EC-1 and EC-2) by reintroducing the VdPEL1 gene (**Supplementary Figure S6**). All of the strains examined showed normal development, and there was no influence on radial growth and colony morphology (**Supplementary Figure S7**). To investigate the contribution of VdPEL1 to V. dahliae virulence, the wild type V. dahliae and VdPEL1 mutant strains were inoculated onto N. benthamiana plants and cotton plants. Interestingly, VdPEL1 deletion strains displayed significantly reduced virulence on tobacco plants compared with the wild-type V. dahliae. EC-1 and EC-2 recovered the high virulence phenotypes (**Figure 8A**). Similar results were observed in cotton plants 21 days after inoculation with all transformants. 1VdPEL1-1 and 1VdPEL1-2 decreased disease susceptibility, whereas EC-1 and EC-2 and wild strains caused more symptoms of necrosis, wilting, and vascular discoloration (**Figure 8C**). These results indicated that VdPEL1 played a positive role in V. dahliae virulence. This conclusion was

further supported by the observation that the fungal biomass of VdPEL1 deletion lines was significantly lower than the biomass of the wild-type and complementary transformants in inoculated N. benthamiana and cotton plants (**Figures 8B,D**). These results confirmed that VdPEL1 contributed to virulence to V. dahliae.

# DISCUSSION

The plant exocyst has recently emerged as an important battleground in plant–pathogen interactions (Žárský et al., 2013; Fujisaki et al., 2015). The plant cell wall serves as a natural barrier to limit the invasion of pathogens. To penetrate and colonize plants, phytopathogenic fungi produce a diverse group of plant CWDEs, which are involved in the generation of plant diseases and pathogenesis (King et al., 2011; Glass et al., 2013; Kubicek et al., 2014). Among the CWDEs, the pectate lyases are examined more closely because of their crucial roles in degrading plant pectin, which exists widely in plant cell walls and cell linings to maintain cell wall integrity. In this study, we

identified a secreted pectate lyase VdPEL1 from V. dahliae culture supernatant, which has the ability to trigger immunity plant responses and contributes to V. dahliae virulence. In addition, our study also found that the enzymatic activity of VdPEL1 was necessary for induced cell death and PTI responses. Our data provide a new avenue to advance the understanding of host–pathogen interactions.

Verticillium dahliae, a soil-borne hemibiotrophic pathogen, attacks the plant roots and spreads to the leaves through the xylem vessels resulting in verticillium wilt of cotton and diseases in over 400 different plant species (Larsen et al., 2007; Vallad and Subbarao, 2008). Recent reports demonstrated that V. dahliae secreted a large amount of pectate lyases to catalyze the degradation of the pectin and facilitate penetration during its infection processes (Bartnicki-Garcia, 1968; Durrands and Cooper, 1988a,b; Pietro et al., 2009; Tzima et al., 2010). Although pectate lyases are particularly abundant and evolutionary preserved, phylogenetic analysis showed that the pectate lyases of V. dahliae are divided into four groups. According to the difference of virulence and defense responses in fungi and bacteria, pectate lyases were also distributed into four branches (**Figure 2**). We hypothesized that the pectate lyases play diverse functions in pathogenesis and confer the ability of V. dahliae to cause disease on such a broad host range.

The HR, a form of plant cell death in the tissues surrounding the lesion, is regarded as a plant defense response to block pathogen infection (Glazebrook, 2005; Jones and Dangl, 2006; Yamada et al., 2016). The ability to recognize a few nanograms of purified VdPEL1 resulting in rapid leaf tissue necrosis was observed in soybean, tomato, cotton, and N. benthamiana (**Figure 3C**). In addition, the range of plant species responding to VdPEL1 may be larger than we detected. We confirmed that VdPEL1 triggered defense responses, including the accumulation of ROS, leakage of ion electrolytes, deposition of callose, and expression of defense genes (**Figure 6**).

As we know, due to the diversity of the host and the inability of fungicides to affect the pathogen once in the plant vascular system, verticillium wilt diseases are difficult to control. The most sustainable manner to control these diseases is the use of resistant cultivars. Thus, it is relevant to identify the new PAMPs or DAMPs, which can provide materials for disease-resistant breeding. VdPEL1 could induce plant immunity and has the potential to be used in plant breeding and as a biological pesticide.

Previous studies showed that many fungal CWDEs, including xyloglucanases, glucanases, and cellulases, can trigger celldeath responses independent of their enzymatic activity (Ma et al., 2014, 2015; Gui Y.-J. et al., 2017; Zhu et al., 2017a). Endopolygalacturonase 1 (EG1) has two biological activities (enzymatic activity and elicitor activity) that are independent of each other in B. cinerea (Poinssot et al., 2003). However, an extracellular cutinase isolated from V. dahliae triggered plant defense responses required the enzymatic activity in N. benthamiana (Gui Y. et al., 2017). To test whether the enzymatic activity of VdPEL1 was required to trigger the plant immune responses, the site-directed mutagenesis of two residues in a conserved motif of VdPEL1 resulted in the loss of enzymatic

response genes were measured in N. benthamiana leaves 24 h after infiltration of 300 nM purified VdPEL1, VdPEL1rec, and PEVC. qRT-PCR was performed. Error bars represent standard deviation of three independent replicates. Student's t-test was performed to determine the significant differences between VdPEL1 and PEVC. Asterisks " ∗∗" indicate statistically significant differences at a p-value < 0.01. (D) Callose deposition in N. benthamiana leaves was detected 2 days after infiltration of 300 nM flg22, purified VdPEL1, VdPEL1rec, and PEVC; the treated leaves were stained with aniline blue.

activity. In contrast to VdPEL1, we surprisingly found that VdPEL1rec resulted in the loss of the cell death-inducing activity and the function of a series of plant defense responses and systemic resistance (**Figures 6**, **7**). These results indicated that the enzymatic activity of VdPEL1 was necessary to trigger defense responses in N. benthamiana and cotton plants.

The pectate lyases have been implicated in pathogenicity and virulence in several plant pathogens. For example, in C. coccodes, the pectate lyase gene CcpelA contributes virulence on tomato, and in C. gleosporoides, deletion of the pectate lyase gene PelB resulted in a substantial loss of virulence on avocado (Persea americana) fruit (Yakoby et al., 2001; Ben-Daniel et al., 2011). Not all fungal pectate lyases have been conclusively shown to be involved in pathogenicity and virulence. For example, PelA (a pectate lyase gene form Fusarium graminearum) knockout strains did not show attenuated virulence during the infection of wheat coleoptiles (Blum, 2017). The expression of VdPEL1 was most significantly up-regulated during the infection stage (1.5–3 days) before dropping sharply to the initial level (**Supplementary Figure S5**). The very strong and early expression of VdPEL1 may help the pathogen to extract nutrition, in addition to the more obvious role of physically facilitating invasion of the host tissue. Meanwhile, we observed that targeted VdPEL1 deletion resulted in significantly compromised virulence of Vd991 on tobacco and cotton plants (**Figure 8**).

Unlike effectors, which interferes with the plant defense response leading to ETS, VdPEL1 appeared to be a major virulence factor due to its enzymatic function, similar to VdCUT11 (Gui Y.-J. et al., 2017). It may indicate that VdPEL1 executes the pectate lyase activity to degrade pectin in the roots, which results in the invasion of the pathogens and the release of plant cell wall fragments (DAMPs). We speculated that the immunity triggered by VdPEL1 was likely to be mediated by the degradation of plant cell wall polymers that release pectin hydrolysis products (DAMPs), which in turn, trigger defense responses in plants.

The plant cell wall may be damaged by abiotic assaults or biotic assaults, resulting in the tissue or cellular damage, which is perceived as danger signals that function as DAMPs (Bianchi, 2006; Choi and Klessig, 2016). Generally, DAMPs

FIGURE 7 | VdPEL1 confers disease resistance in N. benthamiana and cotton plants. (A) N. benthamiana leaves were pre-treated with 300 nM purified VdPEL1, VdPEL1rec, and PEVC. The treated leaves were inoculated with 5 µl of 2 × 10<sup>6</sup> conidia/ml Botrytis cinerea. Lesions symptoms were observed and photographed at 2 days post-inoculation. (B) Lesion diameter of B. cinerea on N. benthamiana leaves was measured after 2 days post-inoculation. The average lesion diameter on six leaves from six plants each was determined. Error bars represent standard deviation of three independent replicates. Student's t-test was performed to determine the significant differences between VdPEL1 and PEVC. Asterisks " ∗∗" indicate statistically significant differences at a p-value < 0.01. (C) N. benthamiana leaves were pre-treated with 300 nM purified VdPEL1, VdPEL1rec, and PEVC and inoculated 24 h later with 1 × 10<sup>6</sup> conidia/ml V. dahliae. The phenotypes were observed and photographed at 12 days post-inoculation. (D) The fungal biomasses on the N. benthamiana plants were determined using qRT-PCR. Error bars represent standard deviation of three independent replicates. Student's t-test was performed to determine the significant differences between VdPEL1 and PEVC. Asterisks " ∗∗" indicate statistically significant differences at a p-value <0.01. (E) Cotton leaves were pre-treated with 300 nM purified VdPEL1, VdPEL1rec, and PEVC and inoculated 24 h later with 1 × 10<sup>6</sup> conidia/ml V. dahliae. The phenotypes were observed and photographed at 21 days post-inoculation. (F) The fungal biomasses on cotton plants were determined using qRT-PCR. Error bars represent standard deviation of three independent replicates. Student's t-test was performed to determine the significant differences between VdPEL1 and PEVC. Asterisks " ∗∗" indicate statistically significant differences at a p-value <0.01.

appear in the apoplast and induce innate immune responses. For instance, cutin monomers and plant elicitor peptides (Peps), which are produced by pathogens depolymerization, can act as DAMPs (Chassot and Métraux, 2005; Huffaker et al., 2006; Yamaguchi et al., 2006). Similarly, OGs, fragments of the pectic polysaccharide homogalacturonan, can be released by pathogenencoded hydrolytic enzymes to induce innate immune responses, including MAPK activation, callose deposition, ROS production

and defense gene up-regulation (Decreux and Messiaen, 2005; Denoux et al., 2008).

Successful pathogens deliver effectors to surmount the host PTI response and establish infection (Jones and Dangl, 2006; Gimenez-Ibanez et al., 2009). For example, a RXLR effector suppressed XEG1-triggered immunity in oomycetes (Ma et al., 2015). In V. dahliae, carbohydrate-binding modules (CBMs) act as effectors, suppressing the GH12 protein and VdCUT11 triggered immunity and thus, facilitating host colonization (Gui Y. et al., 2017; Gui Y.-J. et al., 2017). Whether effectors mediate the suppression of VdPEL1 merits further investigation.

# AUTHOR CONTRIBUTIONS

fpls-09-01271 September 11, 2018 Time: 18:47 # 13

YD and DQ designed the experiments. YY performed most of the experiments and wrote the paper. YZ, XY, and BL participated in some part of the study.

# FUNDING

This study was supported by the National Natural Science Foundation of China (Nos. 31701782 and 31371984).

# ACKNOWLEDGMENTS

We are grateful to X. F. Dai from the Institute of Food Science and Technology, Chinese Academy of Agricultural Sciences, for the generous gift of the vectors for gene knockout and complementation. we also thank L. Yao from Beijing Academy of Agriculture and Forestry for providing the pYBA1132 plasmid and pPICZαA plasmid.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01271/ full#supplementary-material

# REFERENCES


FIGURE S1 | The MS/MS sequencing information. Proteins corresponding to Peak A were digested, and the peptides generated were analyzed using mass spectrometry. The detected peptides were covered with yellow color and matched against a protein from V. dahliae.

FIGURE S2 | The analysis of the hydrolase activity of VdPEL1 with different temperatures, Ca2<sup>+</sup> concentrations or pH. The reduced sugars were quantified using a standard calibration curve obtained with polygalacturonic acid. All the experiments were replicated three times. Standard errors from three biological replicates are shown.

FIGURE S3 | Sequence alignment of the V. dahliae pectate lyase family proteins. Sequence alignment of all the V. dahliae pectate lyase family proteins and known cutinases from other fungi. The accession numbers of known pectate lyases from other fungi and bacteria are: 1EE6 (a pectate lyase from Bacillus sp. strain Ksm-P15), Pel A (pectate lyase A from Erwinia chrysanthemi). Two red triangles indicated possible catalytic residues of VdPEL1 (D<sup>125</sup> and D147).

FIGURE S4 | SDS-PAGE of VdPEL1 and VdPEL1rec recombinant proteins. VdPEL1 is the native protein; VdPEL1rec is a site-directed mutagenized protein, in which D<sup>125</sup> and D<sup>147</sup> were substituted with Ala. Two recombinant proteins were stained with Coomassie blue.

FIGURE S5 | VdPEL1 expression analysis during infection of N. benthamiana and cotton roots. (A) The expression analysis of VdPEL1 in cotton roots. (B) The expression analysis of VdPEL1 in tobacco roots. The control (C) was mixed with non-inoculated conidia and cotton or tobacco root tissue. The housekeeping gene β-tubulin (VDAG\_10074) was used as an endogenous control. Error bars represent standard errors.

FIGURE S6 | Schematic view of the targeted deletion of VdPEL1 in V. dahliae. Two flanking sequences of the target gene and hygromycin resistance cassette were constructed into a fusion fragment. The fusion amplicon was integrated into the pGKO2-gateway vector using a homologous recombination method. The vectors were transferred into the Agrobacterium tumefaciens AGL-1 strain for fungal transformation with the wild-type (WT) strain Vd991. Two test primers were used to identify the fusion amplicon and positive targeted gene-deletion stains using PCR, respectively.

FIGURE S7 | The targeted deletion of VdPEL1 does not affect radial growth and colony morphology. (A) The radial growth and colony morphology of the wild-type V. dahliae (WT), two VdPEL1 deletion strains (1VdPEL1-1 and 1VdPEL1-2), and two ectopic transformants (EC-1 and EC-2) were determined after 11 days of incubation on PDA media at 25◦C. (B) Colony diameters were determined at the time points indicated. Values shown are the average of three colony diameters. Error bars represent standard deviations.

TABLE S1 | Hydrolysis activity test.

TABLE S2 | Primers used in this study.


Chassot, C., and Métraux, J.-P. (2005). The cuticle as source of signals for plant defense. Plant Biosyst. 139, 28–31. doi: 10.1080/11263500500056344



induce plant cell death. New Phytol. 196, 247–260. doi: 10.1111/j.1469-8137. 2012.04241.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 Yang, Zhang, Li, Yang, Dong and Qiu. 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.

# Exogenous Nicotinamide Adenine Dinucleotide Induces Resistance to Citrus Canker in Citrus

Fernando M. Alferez1,2, Kayla M. Gerberich<sup>1</sup> , Jian-Liang Li<sup>3</sup> , Yanping Zhang<sup>4</sup> , James H. Graham<sup>1</sup> and Zhonglin Mou<sup>5</sup> \*

<sup>1</sup> Citrus Research and Education Center, University of Florida, Lake Alfred, FL, United States, <sup>2</sup> Southwest Florida Research and Education Center, University of Florida, Immokalee, FL, United States, <sup>3</sup> National Institute of Environmental Health Sciences, National Institutes of Health, Durham, NC, United States, <sup>4</sup> Interdisciplinary Center for Biotechnology Research, University of Florida, Gainesville, FL, United States, <sup>5</sup> Department of Microbiology and Cell Science, University of Florida, Gainesville, FL, United States

Nicotinamide adenine dinucleotide (NAD) is a universal electron carrier that participates in important intracellular metabolic reactions and signaling events. Interestingly, emerging evidence in animals indicates that cellular NAD can be actively or passively released into the extracellular space, where it is processed or perceived by ectoenzymes or cell-surface receptors. We have recently shown in Arabidopsis thaliana that exogenous NAD induces defense responses, that pathogen infection leads to release of NAD into the extracellular space at concentrations sufficient for defense activation, and that depletion of extracellular NAD (eNAD) by transgenic expression of the human NAD-hydrolyzing ectoenzyme CD38 inhibits plant immunity. We therefore hypothesize that, during plant–microbe interactions, NAD is released from dead or dying cells into the extracellular space where it interacts with adjacent naïve cells' surface receptors, which in turn activate downstream immune signaling. However, it is currently unknown whether eNAD signaling is unique to Arabidopsis or the Brassicaceae family. In this study, we treated citrus plants with exogenous NAD<sup>+</sup> and tested NAD+-induced transcriptional changes and disease resistance. Our results show that NAD<sup>+</sup> induces profound transcriptome changes and strong resistance to citrus canker, a serious citrus disease caused by the bacterial pathogen Xanthomonas citri subsp. citri (Xcc). Furthermore, NAD+-induced resistance persists in new flushes emerging after removal of the tissues previously treated with NAD+. Finally, NAD<sup>+</sup> treatment primes citrus tissues, resulting in a faster and stronger induction of multiple salicylic acid pathway genes upon subsequent Xcc infection. Taken together, these results indicate that exogenous NAD<sup>+</sup> is able to induce immune responses in citrus and suggest that eNAD may also be an elicitor in this woody plant species.

Keywords: extracellular NAD, disease resistance, transcriptional changes, citrus canker, Xanthomonas citri subsp. citri, defense gene, Citrus sinensis

# INTRODUCTION

Nicotinamide adenine dinucleotide (NAD) is a ubiquitous electron carrier that functions in both metabolic reactions and signaling events inside the cell (Berger et al., 2004; Noctor et al., 2006). Upon environmental stimuli, cellular NAD can also be actively or passively released into the extracellular space, where it is processed or perceived by ectoenzymes or cell-surface receptors,

### Edited by:

Shui Wang, Shanghai Normal University, China

#### Reviewed by:

Pierre Pétriacq, Université de Bordeaux, France Alberto A. Iglesias, National University of the Littoral, Argentina

> \*Correspondence: Zhonglin Mou zhlmou@ufl.edu

### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 28 May 2018 Accepted: 20 September 2018 Published: 09 October 2018

#### Citation:

Alferez FM, Gerberich KM, Li J-L, Zhang Y, Graham JH and Mou Z (2018) Exogenous Nicotinamide Adenine Dinucleotide Induces Resistance to Citrus Canker in Citrus. Front. Plant Sci. 9:1472. doi: 10.3389/fpls.2018.01472

**252**

resulting in transmembrane signaling (Bruzzone et al., 2001; Contreras et al., 2003; Seman et al., 2003; Adriouch et al., 2007). In recent years, accumulating evidence indicates that extracellular NAD (eNAD) plays a unique signaling function in diverse physiological and pathological processes (Billington et al., 2006; Haag et al., 2007; Adriouch et al., 2012).

In animal cells, two groups of eNAD processing ectoenzymes, CD38/CD157 and mono(ADP-ribosyl)transferases (ARTs), have been well characterized (Billington et al., 2006). CD38 is a multifunctional ectoenzyme that utilizes NAD as the substrate to produce cyclic ADP-ribose, a second messenger triggering calcium release from intracellular stores (Ceni et al., 2003; De Flora et al., 2004; Krebs et al., 2005; Malavasi et al., 2006; Morabito et al., 2006; Partida-Sanchez et al., 2007). ARTs are another group of ectoenzymes, which use NAD as the substrate to ADP-ribosylate lipid raft-associated proteins (Nemoto et al., 1996; Han et al., 2000; Seman et al., 2003; Bannas et al., 2005; Zolkiewska, 2005). eNAD seems also to be perceived by cellsurface receptors (Moreschi et al., 2006; Mutafova-Yambolieva et al., 2007; Grahnert et al., 2009; Klein et al., 2009). However, a bona fide eNAD-binding receptor has not been identified in animal cells.

In plants, NAD has also been documented to function in response to environmental stresses including pathogen infections (Noctor et al., 2006; Mahalingam et al., 2007; Hashida et al., 2009; Pétriacq et al., 2013; Gakière et al., 2018). For instance, quinolinate-induced elevation of intracellular NAD in Arabidopsis thaliana expressing the nadC gene from Escherichia coli, which encodes the NAD biosynthesis enzyme quinolinate phosphoribosyltransferase, increases defense gene expression and resistance to multiple bacterial and fungal pathogens (Pétriacq et al., 2012, 2016). Conversely, mutations in the NAD biosynthesis gene FLAGELLIN-INSENSITIVE4 inhibit stomatal immunity in Arabidopsis. Moreover, overexpressing the A. thaliana Nudix hydrolase homolog6 (AtNUDT6), which encodes an ADP-ribose/NADH pyrophosphohydrolase, and knocking out AtNUDT6, AtNUDT7, or AtNUDT8 all lead to changes in intracellular NADH levels and salicylic acid (SA) mediated immune signaling (Bartsch et al., 2006; Ge et al., 2007; Ishikawa et al., 2010; Fonseca and Dong, 2014).

We have recently shown in the model plant Arabidopsis that exogenous NAD induces SA-dependent and -independent expression of PATHOGENESIS-RELATED (PR) genes and resistance to the bacterial pathogen Pseudomonas syringae (Zhang and Mou, 2009, Mou, 2012). Moreover, we discovered that P. syringae-induced cell death leads to release of NAD into the extracellular space at concentrations sufficient for induction of PR gene expression and P. syringae resistance (Zhang and Mou, 2009). In addition, we found that expression of the human NADmetabolizing ectoenzyme CD38 inhibits biological induction of systemic acquired resistance (SAR), a long-lasting form of plant immunity (Zhang and Mou, 2012). These results together suggest that eNAD may be a damage-associated molecular pattern (DAMP) in plants (Mou, 2017).

We subsequently used both forward and reverse genetic approaches to establish the eNAD signaling pathway in Arabidopsis. In the forward genetic screen, we revealed that the mediator complex subunits MED14/STRUWWELPETER and MED16/SENSITIVE TO FREEZING6/INSENSITIVE TO EXOGENOUS NAD1 as well as the elongator complex are downstream signaling components in the eNAD signaling pathway (Zhang et al., 2012, 2013; An et al., 2016). In the reverse genetic screen, we demonstrated that the lectin receptor kinase (LecRK), LecRK-I.8, is an eNAD-binding receptor and functions in plant basal immunity (Wang et al., 2017). These findings support that eNAD is a DAMP in Arabidopsis. However, it is not clear whether eNAD signaling is specific for Arabidopsis or the Brassicaceae family (Mou, 2017). In this study, we tested if NAD could activate immune responses in sweet orange (Citrus sinensis L. Osbeck), one of the most economically important tree fruit crops in the world. Our results indicate that treatment of citrus plants with exogenous NAD<sup>+</sup> induces profound transcriptional changes and strong resistance to citrus canker, a serious citrus disease caused by the bacterial pathogen Xanthomonas citri subsp. citri (Xcc) (Graham et al., 2004). These results suggest that eNAD signaling may be conserved in citrus.

# MATERIALS AND METHODS

# Plant Materials and Growth Conditions

Pineapple sweet orange seedlings were maintained in standard greenhouses at the Citrus Research and Education Center of the University of Florida. Citrus seedlings were cut back to produce expanding leaves 6 weeks prior to NAD<sup>+</sup> treatments.

# Chemical Treatment

NAD<sup>+</sup> sodium salt (Sigma-Aldrich, St. Louis, MO, United States) was dissolved in water and pH was adjusted to 6.0 using 0.1 M NaOH. For soil drenches, 7 days in advance of Xcc inoculation when seedlings reached 50 to 75% expanded flush, 250 mL NAD<sup>+</sup> solution (1, 5, or 10 mM) per potted seedling were applied. For infiltration, 1 day in advance of Xcc inoculation leaves were infiltrated with various concentrations of NAD<sup>+</sup> (0, 0.25, 0.5, 0.75, 1, 5, and 10 mM) using a needleless tuberculin syringe (1.0 mL) as previously described (Graham and Leite, 2004). For the positive control, Actigard (Syngenta, Greensboro, NC, United States) was dissolved in water (2 g/L) and 250 mL of the resulting solution was applied to each potted citrus seedling by soil drenching.

# Pathogen Infection

Five plants per treatment were inoculated by injection-infiltration with a suspension of 10<sup>4</sup> cfu/mL of Xcc strain X2002-0014 in PBS by pressing the needleless syringe tip against the leaf surface to produce a zone of water-soaked tissue in three areas on each side of the midrib to produce six distinct inoculation sites per leaf on three to four leaves per replicate plant for a total of 15 leaves per treatment. Inoculated shoots were covered with plastic bags for 1 day to maintain high humidity conducive for bacterial infection of leaves. At 14 day post-inoculation, lesions were counted at each inoculation site and summed as total lesions per leaf.

To determine the persistence of the treatment effect over time (previously called residual activity) (Francis et al., 2009), plants

FIGURE 1 | Exogenous NAD+-induced resistance to citrus canker in citrus. (A) Phenotypes of the citrus canker lesions on leaves of citrus plants treated with soil drenches of water, 1 mM NAD+, or Actigard. Photos were taken 14 days after Xcc inoculation. (B) Phenotypes of the citrus canker lesions on leaves pre-infiltrated with water or 1 mM NAD+. Photos were taken 14 days after Xcc inoculation. (C) Numbers of the citrus canker lesions on leaves of citrus plants treated with soil drenches of water, Actigard, or different concentrations (1, 5, and 10 mM) of NAD+. Data represent means of five biological replicates with standard error (SE). Different letters above the bars indicate significant differences (Newman–Keuls test, p < 0.05). (D) Numbers of the citrus canker lesions on leaves pre-infiltrated with water or various concentrations (0.25, 0.5, 0.75, 1, 5, and 10 mM) of NAD+. Data represent means of five biological replicates with SE. Different letters above the bars indicate significant differences (Newman–Keuls test, p < 0.05).

were pruned below the Xcc inoculation point (6 weeks after NAD application). When new single shoots produced four to six young leaves, a new set of Xcc inoculations were performed (16 weeks after NAD applications).

# RNA Analysis

Total RNA extraction, reverse transcription, and real-time quantitative PCR (qPCR) were performed as previously described (Llorens et al., 2015) with some modifications. Briefly, six disks of 6-mm diameter (one per inoculation site) from three leaves per plant from at least three plants per biological replicates were collected per each time point and frozen in liquid nitrogen, and tissues were ground twice at 30 rps in a Tissuelyzer II (QIAGEN, Hilden, Germany). RNA was extracted using an RNeasy Mini kit (QIAGEN) according to the manufacturer's instructions and RNA concentration was determined using a ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, United States). One microgram of total RNA was used for cDNA synthesis with a QuantiTect Reverse Transcription kit (QIAGEN) according to the manufacturer's instructions. qPCR was performed on a 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, United States) using a QuantiTect SYBR Green PCR kit (QIAGEN) in 20 µL reactions following the manufacturer's instructions. Samples were run in triplicate. The occurrence of non-specific amplification products was ruled out after performing a melting curve analysis. The relative gene expression analysis was performed using the 1Ct method. Time 0 expression was arbitrarily set as 1 and then data of each time point were normalized against time 0. Results were presented as mean ± standard error of three biological samples run in triplicate. Primers used in this study were listed in **Supplementary Table S2**.

## Microarray Analysis

Microarray analysis using the Affymetrix microarray platform was performed at the University of Florida Interdisciplinary Center for Biotechnology Research. RNA quality was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, United States). All microarray sample preparation used the GeneChip <sup>R</sup> 3 0 IVT Plus Express kit (Affymetrix, Inc., Santa Clara, CA, United States), and reactions were done following the manufacturer's protocols. Briefly, cDNA was synthesized from 200 ng of total RNA and template for in vitro transcription during which a biotinmodified nucleotide was incorporated. The biotin-labeled aRNA was then purified and fragmented. Samples were hybridized with rotation at 45◦C for 16 h to the Affymetrix GeneChip Citrus Genome Array. The arrays were washed and stained with the reagents supplied in GeneChip <sup>R</sup> Hybridization Wash and Stain kit (Affymetrix, Inc.) on an Affymetrix Fluidics Station 450, and scanned with a GeneChip <sup>R</sup> 7G Scanner (Affymetrix, Inc.).

The microarray data were pre-processed and normalized using the affy package. The Robust Multichip Analysis approach was applied for the normalization. After normalization, the empirical Bayes moderated t-statistics, which is implemented in the limma Bioconductor package (Smyth, 2004), was performed for differential expression detection. In each comparison, a p-value and fold change were computed for each gene locus. The gene expression fold changes were

TABLE 1 | Possible defense-related genes that were induced by NAD<sup>+</sup> treatment.


computed based on the normalized log-transformed signal intensity data. To control false discovery rate and correct multiple hypothesis testing, a q-value was calculated and used to assess the significance of each test using Benjamini and Hochberg's approach (Benjamini and Hochberg, 1995). Genes with an absolute fold change ≥ 2 and a q-value ≤ 0.05 were considered as significantly differentially expressed.

## Statistical Methods

Statistical analyses were performed by one-way ANOVA using SAS (SAS Institute Inc., Cary, NC, United States) followed by a Student–Newman–Keuls test. All experiments were repeated at least three independent times with similar trends. Results from a representative experiment are presented.

# RESULTS

# Exogenous NAD<sup>+</sup> Induces Resistance to Citrus Canker in a Concentration-Dependent Manner in Citrus

To test whether exogenous NAD<sup>+</sup> induces immune responses in citrus, citrus plants were treated with different concentrations of NAD<sup>+</sup> by soil drenches or leaf infiltration. It has previously been shown that soaking plus foliar sprays of SA and soil drenches

significant differences (Newman–Keuls test, p < 0.05). (B) Numbers of the citrus canker lesions on leaves of new shoots emerged on the citrus plants with leaves previously infiltrated with water or various concentrations (0.25, 0.5, 0.75, 1, 5, and 10 mM) of NAD<sup>+</sup> and inoculated with Xcc. The plants were pruned below the inoculation point 6 weeks after NAD<sup>+</sup> treatment. Data represent means of five biological replicates with SE. Different letters above the bars indicate significant differences (Newman–Keuls test, p < 0.05).

FIGURE 3 | NAD+-mediated priming of several SA-related defense genes. Expression of CsCM2, CsCM1, CsICS, CsPAL, CsNPR1, and CsPR5 at the indicated time points in citrus leaf tissues first infiltrated with 5 mM NAD<sup>+</sup> and then inoculated with Xcc. The arrows indicate the time point when the leaf tissues were inoculated with Xcc after NAD<sup>+</sup> treatment. Expression of the target genes was normalized against the constitutively expressed Cs18S. The y-axis values represent the relative expression levels of the indicated genes, which were calculated using the formula 2[Ct(Cs18S)−Ct(target gene)] . Data represent means of three biological replicates with SE.

of Actigard, a product containing the SA analog acibenzolar-Smethyl, induce strong and persistent resistance to citrus canker (Francis et al., 2009; Graham and Myers, 2009; Wang and Liu, 2012). We included Actigard in the soil drench experiments as the positive control. After the chemical treatment, leaves on the treated plants or the infiltrated leaves were inoculated with Xcc. Symptoms were characterized on day 14 post-inoculation. As shown in **Figure 1A**, leaves on water-treated plants produced callus-like lesions, whereas those on NAD+- or Actigard-treated plants developed fewer necrotic lesions that were smaller in size. Similarly, leaves pre-infiltrated with water generated calluslike lesions, whereas those pre-infiltrated with NAD<sup>+</sup> formed fewer and smaller necrotic lesions (**Figure 1B**). The disease symptoms on the plants drenched with 1mM NAD<sup>+</sup> and the leaves pre-infiltrated with 1 mM NAD<sup>+</sup> were comparable to those developed on the plants drenched with Actigard (2 µg/mL) (**Figures 1A,B**).

The resistance induced by NAD<sup>+</sup> was also reflected by the numbers of lesions developed at the inoculation sites. Overall, the lesion numbers on the plants or leaves treated with Actigard or NAD<sup>+</sup> were significantly lower than those on the watertreated controls. For soil drenches, the lesion numbers on the plants treated with 1 and 5 mM NAD<sup>+</sup> were comparable to those on the plants treated with Actigard, whereas the lesion numbers on the plants treated with 10 mM were significantly lower than those on the Actigard-treated plants (**Figure 1C**). For infiltration, the lesion numbers on the leaves treated with 0.25, 0.5, 0.75, and 1 mM NAD<sup>+</sup> were not significantly different from each other, whereas the lesion numbers on leaves treated with 5 and 10 mM NAD<sup>+</sup> were significantly lower than those on the leaves treated with lower concentrations of NAD<sup>+</sup> (**Figure 1D**). Thus, exogenous NAD<sup>+</sup> induces disease resistance in citrus in a concentration-dependent manner.

# Exogenous NAD<sup>+</sup> Triggers Profound Transcriptome Changes in Citrus

To understand the molecular events underlying NAD+-induced disease resistance in citrus, a microarray experiment was carried out to examine NAD+-triggered transcriptional changes in citrus (National Center for Biotechnology Information [NCBI] Gene Expression Omnibus series number GSE113735). Independent samples in triplicate were assayed, and the results were analyzed to identify genes that displayed a twofold or higher induction or suppression with a low q-value (≤0.05) in the NAD+-treated samples compared with the water-treated control samples. A total of 660 and 574 genes were up- and down-regulated, respectively, at 4 h after the NAD<sup>+</sup> treatment (**Supplementary Table S1**). The numbers of genes that were up- and down-regulated in citrus are much smaller than those (2,155 and 2,014 genes upand down-regulated, respectively) in Arabidopsis under similar conditions (Wang et al., 2017). These results suggest that citrus may respond more slowly or be less sensitive to exogenous NAD<sup>+</sup> treatment than Arabidopsis. Nevertheless, a large number of defense-related genes, such as ENHANCED DISEASE SUSCEPTIBILITY1, NIM1-INTERACTING PROTEIN2, CHORISMATE MUTASE2 (CM2), PATHOGENESIS-RELATED (PR) GENES TRANSCRIPTIONAL FACTOR PTI5, PR4A, PR4B, GLUTATHIONE S-TRANSFERASE6, just to name a few, were up-regulated by the NAD<sup>+</sup> treatment (**Table 1** and highlighted

in yellow in **Supplementary Table S1**). Thus, as in Arabidopsis (Wang et al., 2017), exogenous NAD<sup>+</sup> treatment activates defense signaling pathways in citrus.

# Exogenous NAD<sup>+</sup> Activates Long-Lasting Resistance in Citrus

Previous work has shown that the effectiveness of soil-applied Actigard can last for more than 16 weeks (Francis et al., 2009). To determine the effectiveness of soil-applied NAD<sup>+</sup> over time, plants were pruned below the Xcc inoculation point (6 weeks after NAD<sup>+</sup> treatment). When new single shoots produced four to six young leaves, the young leaves were inoculated with Xcc (16 weeks after NAD<sup>+</sup> treatment). On day 14 post-inoculation, lesions developed on the inoculated leaves were counted. As shown in **Figure 2A**, the effectiveness of soil drenches of 1 or 5 mM NAD<sup>+</sup> was comparable to that of Actigard, whereas 10 mM NAD<sup>+</sup> had a stronger effect at this time point. We also tested the effectiveness of leaf infiltration of NAD<sup>+</sup> over time. Intriguingly, leaf infiltration of NAD was also able to provide protection to the newly emerged tissues against citrus canker, and the effectiveness was comparable among the applied six different concentrations (0.25, 0.5, 0.75, 1, 5, and 10 mM) (**Figure 2B**).

# Exogenous NAD<sup>+</sup> Primes Citrus Leaf Tissues

The long-lasting resistance provided by NAD<sup>+</sup> treatment might be due to NAD's priming effects. To test this hypothesis, we infiltrated citrus leaves with NAD<sup>+</sup> and half of the infiltrated leaves were collected at 4 and 24 h, and the other half of the NAD+-infiltrated leaves were inoculated with Xcc at 24 h after NAD<sup>+</sup> treatment. Expression of six SA pathway genes, CsCM2, CsCM1, CsICS (ISOCHORISMATE SYNTHASE), CsPAL (PHENYLALANINE AMMONIA LYASE), CsNPR1 (NON-EXPRESSOR OF PR GENES), and CsPR5, was analyzed by qPCR. At 4 h after NAD<sup>+</sup> treatment, CsCM2 and CsCM1 were dramatically induced, CsPAL, CsNPR1, and CsPR5 were slightly induced, and CsICS was barely induced (**Figure 3**). Surprisingly, after Xcc inoculation, all six genes were induced to much higher levels in the leaf tissues pre-treated with NAD<sup>+</sup> than in those pretreated with water (**Figure 3**). These results indicate exogenous NAD<sup>+</sup> treatment primes these SA pathway genes.

# DISCUSSION

Although eNAD is being established as a novel DAMP in Arabidopsis, its role in other plant species has not been investigated (Mou, 2017). In this study, we tested exogenous NAD+-triggered immune responses in citrus. Our results show that exogenous NAD<sup>+</sup> induced strong resistance to citrus canker (**Figures 1A–D**), a serious leaf and fruit disease damaging multiple economically important citrus cultivars including grapefruit and certain sweet oranges (Graham et al., 2004). Furthermore, NAD<sup>+</sup> treatment triggered profound transcriptome changes in citrus leaves, with about 1,200 genes being up-regulated or down-regulated by twofold or more (**Table 1** and **Supplementary Table S1**). These results indicate that citrus leaf tissues are highly responsive to exogenous NAD<sup>+</sup> treatment. Thus, as the herbaceous plant Arabidopsis (Zhang and Mou, 2009; Wang et al., 2017), the woody plant citrus may also possess sensitive eNAD perception mechanisms.

Exogenous NAD<sup>+</sup> provides long-lasting protection against citrus canker in citrus. Previous work showed that soil drenches of the SAR-inducing product Actigard was able to protect against citrus canker for 16–24 weeks (Francis et al., 2009). The effect of soil drenches of NAD<sup>+</sup> was comparable to that of Actigard, which in our experiments lasted for at least 16 weeks (**Figure 2A**). This long-lasting protection may be attributed to the priming effect of NAD<sup>+</sup> on defense genes such as the SA pathway genes tested in **Figure 3**. Our previous work in Arabidopsis has also shown that although low concentrations (0.2 and 0.4 mM) of NAD<sup>+</sup> did not drastically induce PR gene expression, they significantly enhanced resistance to the bacterial pathogen P. syringae pv. maculicola ES4326 (Zhang and Mou, 2009). It is thus likely that low concentrations of NAD<sup>+</sup> are also sufficient for priming defense genes in citrus (**Figure 2B**).

Surprisingly, resistance induced by leaf infiltration of NAD<sup>+</sup> was also able to persist in newly emerged shoots even after removal of the NAD+-treated leaves together with the shoots (**Figure 2B**). It is possible that NAD+, its derivatives, or NAD+ induced signaling molecules are able to move throughout the citrus plants, leading to resistance in new flushes. In agreement with this observation, we have shown in Arabidopsis that local infiltration of 5 mM NAD<sup>+</sup> or NADH, a concentration higher than the physiological levels, was able to induce resistance in systemic leaves (Zhang and Mou, 2012). Although these results suggest that NAD may induce SAR, further investigations are required to fully understand if and how endogenous eNAD activates immunity systemically in plants.

# AUTHOR CONTRIBUTIONS

ZM and JG conceived and designed the experiments. FA and KG performed the experiments. FA, J-LL, and YZ analyzed the data. ZM and FA wrote the paper. All the authors carefully checked and approved this version of the manuscript.

# FUNDING

This work was supported by grants from the Citrus Research and Development Foundation (Grant Nos. 13-030-754 and 15-020).

# ACKNOWLEDGMENTS

We thank Tony McIntosh for taking care of the greenhouse citrus plants.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01472/ full#supplementary-material

# REFERENCES

fpls-09-01472 October 5, 2018 Time: 14:4 # 7


by ART2 activates the cytolytic P2X7 purinoceptor. Immunity 19, 571–582. doi: 10.1016/S1074-7613(03)00266-8


acquired resistance and jasmonate/ethylene-induced defense pathways. Plant Cell 24, 4294–4309. doi: 10.1105/tpc.112.103317


**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 Alferez, Gerberich, Li, Zhang, Graham and Mou. 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.

# Recent Advances in Synthetic Chemical Inducers of Plant Immunity

#### Mian Zhou<sup>1</sup> \* and Wei Wang2,3 \*

<sup>1</sup> Department of Plant Pathology and Microbiology, Iowa State University, Ames, IA, United States, <sup>2</sup> School of Life Sciences, Peking University, Beijing, China, <sup>3</sup> Peking University – Tsinghua University Joint Center for Life Sciences, Beijing, China

Different from the conventional biocidal agrochemicals, synthetic chemical inducers of plant immunity activate, bolster, or prime plant defense machineries rather than directly acting on the pathogens. Advances in combinatorial synthesis and high-throughput screening methods have led to the discovery of various synthetic plant immune activators as well as priming agents. The availability of their structures and recent progress in the mechanistic understanding of plant immune responses have opened up the possibility of identifying new or more potent chemical inducers through rational design. In this review, we first summarize the chemical inducers identified through large-scale screening and then discuss the emerging trends in the identification and development of novel plant immune inducers including natural elicitor based chemical derivation, bifunctional combination, and computer-aided design.

#### Edited by:

Shui Wang, Shanghai Normal University, China

#### Reviewed by:

Wenli Chen, South China Normal University, China Susheng Song, Capital Normal University, China

#### \*Correspondence:

Mian Zhou zhoumian1986@aliyun.com Wei Wang oneway1985@pku.edu.cn

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 15 September 2018 Accepted: 17 October 2018 Published: 06 November 2018

#### Citation:

Zhou M and Wang W (2018) Recent Advances in Synthetic Chemical Inducers of Plant Immunity. Front. Plant Sci. 9:1613. doi: 10.3389/fpls.2018.01613 Keywords: plant immunity, plant immune inducers, chemical derivation, ionic liquids, diversity-oriented synthesis, computer-aided design

# INTRODUCTION

While plants are important nutritional source of humans, they are also consumed by various heterotrophic microorganisms, which cause diverse plant diseases and considerable economic loss to agriculture. To reduce the yield loss, conventional chemical pesticides have been developed. They exert their functions through direct biocidal effects on the pathogens. However, besides the toxicity on the pathogens, conventional pesticides may also have negative impacts on the crops, beneficial microorganisms and the health of farmers and consumers. Furthermore, continuous application of conventional pesticides can result in the selection of pesticide-resistant pathogen strains and eventually voids the use of the specific pesticide (Burketova et al., 2015). Synthetic chemical inducers of plant immunity are attractive and promising alternatives. They stimulate or prime the endogenous immunity of plants to combat pathogenic invasions rather than kill the pathogens directly.

Unlike animals that have evolved specific immune cells, nearly each cell in plants is able to act as an "immune cell" to fight against pathogen attacks. Plants can perceive the presence of pathogens through recognition of microbe-associated molecular patterns (MAMPs) or damageassociated molecular patterns (DAMPs) by pattern recognition receptors (PRRs). MAMPs are highly conserved molecular signature within different classes of microbes, for instance, flagellin and elongation factor Tu (EF-Tu) from bacteria, chitin and xylanase from fungi and heptaglucan from oomycetes. DAMPs are plant endogenous immune elicitors released by the pathogen-triggered mechanical stress or enzymatic activities controlled by pathogens, e.g., oligogalacturonides (Schwessinger and Ronald, 2012). The downstream defense activation events following PRR activation include changes of ion fluxes across the plasma membrane, the oxidative burst, activation

of mitogen activated protein kinase (MAPK) cascades, gene activation and callose deposition. This MAMPs/DAMPstriggered immunity (MTI) is the first layer of plant immune system (Jones and Dangl, 2006; Zipfel, 2009). Some pathogens have evolved effectors to interfere with MTI (Dangl et al., 2013). Through co-evolution, plants have developed intracellular immune receptors, Resistance (R) proteins, to recognize the presence of effectors and activate effector-triggered immunity (ETI), which is the second layer of plant immune system (Spoel and Dong, 2012). These two layers of immunity are usually referred to as plant innate immunity (Schwessinger and Ronald, 2012).

The activation of plant innate immunity in local tissue (the infected part) leads to transportation of the mobile defense signals to systemic (uninfected) tissue, resulting in a longlasting resistance to a broad spectrum of pathogens. This acquired immunity is known as systemic acquired resistance (SAR). The induction of SAR usually confers by ETI, however, it has been reported that MTI can also trigger SAR under some circumstances (Mishina and Zeier, 2007). In addition to pathogens, SAR can be induced by exogenous application of chemical inducers, including salicylic acid (SA), its analogs 2, 6 dichloroisonicotinic acid (INA) and benzothiadiazole S-methyl ester (BTH), its derivatives acetylsalicylic acid (aspirin) and methyl SA (MeSA) (White, 1979; Uknes et al., 1992; Cao et al., 1994; Lawton et al., 1996; Durrant and Dong, 2004; Park et al., 2007), nitric oxide (NO), reactive oxygen species (ROS) (Wang et al., 2014), dicarboxylic acid azelaic acid (AzA) (Jung et al., 2009), the phosphorylated sugar glycerol-3 phosphate (G3P) (Chanda et al., 2011), the abietane diterpenoid dehydroabietinal (DA) (Chaturvedi et al., 2012), the aminoacid derivative pipeolic acid (Pip) (Navarova et al., 2012), and N-hydroxypipecolic acid (NHP) (Chen et al., 2018; Hartmann et al., 2018).

To communicate with the systemic tissue, mobile signals are generated in local tissue and then transported to systemic tissue through phloem. Although, it is well-known that SAR is associated with the accumulation of SA in both local and systemic tissues, grafting studies demonstrated that SA is not the mobile SAR signal (Vernooij et al., 1994). Several chemical candidates for this long-distance signal have been proposed, including MeSA (Park et al., 2007), AzA (Jung et al., 2009), glycerol-3-phosphate (G3P) (Chanda et al., 2011), DA (Chaturvedi et al., 2012), Pip (Navarova et al., 2012), and more recently, its derivative, NHP (Hartmann et al., 2018). Key protein players involved have also been identified including Defective in Induced Resistance 1 (DIR1) (Maldonado et al., 2002; Carella et al., 2017), AzA Insensitive 1 (AZI1) (Jung et al., 2009), and Lipid Transfer Protein 2 (LTP2). Plasmodesmata (PD) is considered to be the transportation route of these signals (Lim et al., 2016). These putative SAR signals might function coordinately to achieve longdistance signal transduction (Dempsey and Klessig, 2012; Shah et al., 2014; Wang et al., 2014).

Once SAR signals are perceived, systemic tissues generate SA to activate a key immune regulator, NON-EXPRESSER OF PR1 (NPR1) to trigger massive transcriptional reprogramming, including the induction of Pathogenesis-related (PR) genes and endoplasmic reticulum (ER)-resident genes, which aid secretion of PR proteins (Wang et al., 2005, 2006; Spoel and Dong, 2012; Fu and Dong, 2013). Continuous efforts have been made to study the mechanism of how NPR1 responds to SA and regulates downstream defense genes. SA or pathogen infection could cause changes in cellular redox status (Mou et al., 2003). As a result of the cellular redox changes, the cysteine residues of NPR1 (C82 and C216) are reduced by thioredoxins, leading to an oligomer-to-monomer switch in NPR1 conformation and nuclear translocation of the monomer NPR1 (Tada et al., 2008). Nuclear NPR1 monomer then undergoes phosphorylation to promote its transcriptional activity in SAR and its turnover (Spoel et al., 2009). As a transcription co-factor, nuclear NPR1 interacts with TGAs and NIMI-interacting (NIMIN) TFs to regulate the expression of downstream defense genes (Despres, 2003; Kesarwani et al., 2007). TGAs mainly activate NPR1 mediated genes; while NIMIN represses the expression of defense genes (Zhou et al., 2000; Johnson et al., 2003). After fulfillment of its function, ubiquitination of "exhausted" NPR1 leads to its degradation by the proteasomes, allowing "fresh" NPR1 to reinitiate the transcription cycle (Spoel et al., 2009). Recently NPR1 and its paralogs, NPR3 and NPR4, have been found to directly bind SA and serve as its receptors to mediate transcriptional reprogramming (Fu et al., 2012; Wu et al., 2012; Ding et al., 2018). Besides SA, indolic compounds, jasmonic acid (JA), monoterpenes, NO, ROS and intact cuticle also contribute to the establishment of SAR (Truman et al., 2007, 2010; Xia et al., 2009; Navarova et al., 2012; Wendehenne et al., 2014; Riedlmeier et al., 2017).

Induced systemic resistance (ISR) is another form of systemic immunity, which is triggered by non-pathogenic beneficial microbes (Pieterse et al., 2014). Although ISR and SAR are both systemic defense mechanism, they differ in several ways. First, the triggers of ISR and SAR are fundamentally different. SAR is triggered by either compatible or incompatible pathogenic interactions while ISR is initiated by non-pathogenic microbes. Second, although ISR and SAR are both broad-spectrum, their effective spectrum only partially overlaps (Ton et al., 2002). Third, SA is critical to SAR but ISR is less dependent on SA and mainly regulated by JA and ethylene (ET) (Pieterse et al., 1998; Pieterse et al., 2014). Fourth, SAR is accompanied with induction of PR genes and proteins while SA-independent ISR is not (Hoffland et al., 1995). Instead of direct induction of defense machineries, ISR-conditioned plants can elicit faster and/or stronger defenses upon subsequent pathogenic interactions. This sensitization mechanism is called priming (Conrath et al., 2006). It has been shown that priming can reduce the fitness cost associated with constitutive activation of defenses (van Hulten et al., 2006; Walters et al., 2008; Vos et al., 2013). Despite these distinctions between ISR and SAR, SA-independent ISR also depends on NPR1, the key component of SA signaling pathway (Pieterse et al., 1998; Iavicoli et al., 2003; Ryu et al., 2003; Ahn et al., 2007; Hossain et al., 2008; Stein et al., 2008; Segarra et al., 2009; Weller et al., 2012). Cumulating studies suggest that ISR may mainly rely on the cytosolic function of NPR1 while SAR more depends on the nuclear role of NPR1 (Pieterse et al., 2014).

# SYNTHETIC CHEMICAL INDUCERS OF PLANT IMMUNITY

Synthetic chemical inducers of plant immunity are structurally different from the natural plant defense elicitors. They may activate or prime plant immunity by simply mimicking the structures of natural immune inducers. Alternatively, they can also be structurally unrelated to natural elicitors and target a subset of defense signaling components. In general, they do not have in vitro antimicrobial activity. In this section, we mainly focus on the legacy inducers related to the recently discovered ones, which will be discussed in the "Emerging trends" section.

# SA Derivatives

As a major plant immune hormone, SA plays a pivotal role in the establishment of plant immunity. SA is among the first plant endogenous chemicals reported to induce SAR, which is accompanied by accumulation of PR proteins and resistance to TMV in tomato (White, 1979). In the same study, the famous synthetic SA derivative, Aspirin, was also shown to induce SAR (White, 1979). Later mono- and dichloro substituted SA derivatives including 4-chloro-SA, 5 chloro-SA and 3, 5-chloro-SA were found to induce PR proteins accumulation and resistance against TMV infection in tobacco (Conrath et al., 1995). More comprehensive investigations of mono- and multiple-substituted SA suggest that 3- and 5 position substitutions are more active than 4- and 6-position substitution. Electron-withdrawing substituents are important to the enhanced activity. Except for 6-fluoro-SA, all fluoro- and chloro-SA tested induced more resistance against TMV than SA (Silverman et al., 2005). Aside from the simple substituted SA, a new class of salicyl glycoconjugates containing hydrazide and hydrazone moieties were synthesized and studied on their in vitro and in vivo antifungal activity using cucumber (Cui et al., 2014). While the SA hydrazine derivative showed little in vitro antifungal activity, significant in vivo antifungal activity against Colletotrichum orbiculare, Fusarium oxysporum, Rhizoctonia solani, and Phytophthora capsici was demonstrated. Intriguingly, while the SA hydrazine derivative is structurally derived from SA, it did not induce the expression of SA marker genes but rather induce JA marker genes. This suggests that the SA hydrazine derivative may not be an SA agonist and function through targeting of other immune signaling components.

# Isonicotinic Acid Derivatives

INA was first identified by Ciba-Geigy, the predecessor of Syngenta, through large-scale screening to identify chemicals that can induce resistance in cucumber against the fungal pathogen Colletotrichum lagenarium (Metraux et al., 1991). INA has been shown to induce pathogen resistance in various plants including Arabidopsis, tobacco, pear, pepper, rice, cucumber, and beans (Kuc, 1982; Metraux et al., 1991; Ward et al., 1991; Uknes et al., 1992). INA can trigger similar immune responses as SA but independent of SA accumulation as it can still induces SAR in transgenic plants expressing SA hydrolase (NahG) in which SA accumulation is compromised (Delaney et al., 1994; Vernooij, 1995). Therefore it functions downstream of the SA accumulation. Recent identifications of SA receptors, NPR3 and NPR4 suggest that INA is likely to be a genuine SA agonist. Similar to SA, INA can also promote the interactions between NPR1 and NPR3. Furthermore, in a competition binding assay, INA was shown to compete with SA to bind its receptors, NPR3 and NPR4 (Fu et al., 2012). Besides the interaction with NPR3 and NPR4, interactions between INA and other SA-binding proteins may also contribute to its role in elicitation of immunity (Durner and Klessig, 1995). However, due to its phytotoxicity effects, INA or its derivatives have not been commercialized for agricultural use.

N-cyanomethyl-2-chloro isonicotinic acid (NCI) is another potent plant immune inducer, which belongs to the isonicotinic acid derivative family. It was identified in a screen of 2 chloroisonicotinamide derivatives for control of rice blast (Yoshida et al., 1990a,b). NCI did not show biocidal effects on rice blast in vitro even when a high dose was used. However, its in vivo antifungal activity against rice blast can last 30 days after a single application. In tobacco, NCI induces expression of PR genes even in nahG plants (Nakashita, 2002). This suggests that the immune inducing effect of NCI does not rely on SA accumulation. In Arabidopsis, NCI-induced immunity is independent of SA accumulation but depends on NPR1 (Yasuda et al., 2003; Yasuda, 2007). Therefore NCI appears to interact with the signaling steps between SA and NPR1.

# Thiadiazole and Isothiazole Derivatives

BTH is another potent synthetic SAR inducer identified by Ciba-Geigy through a large-scale screening of thiadiazole derivatives (Schurter et al., 1993; Kunz et al., 1997; Oostendorp et al., 2001). BTH does not exhibit antimicrobial activity in vitro. However, it can trigger disease resistance against a diverse spectrum of pathogens in various plant species. BTH has been tested in more than 120 pathosystems including resistance in apple and pear against fire blight, tomato against bacterial canker, grapefruit against canker, canola against blackleg disease, cowpea against anthracnose, etc. (Latunde-Dada and Lucas, 2001; Brisset et al., 2002; Soylu et al., 2003; Potlakayala et al., 2007; Graham and Myers, 2011). BTH induces the expression of PR genes and BTHtriggered SAR in Arabidopsis is dependent on NPR1 (Lawton et al., 1996). In rice, however, BTH-induced defense responses against rice blast does not require rice ortholog of Arabidopsis NPR1 but rather involves WRKY family transcription factor, OsWRKY45 (Shimono et al., 2007). Similar to INA, BTH is also able to induce SAR and expression of PR genes in nahG plants (Molina et al., 1998). BTH can be converted by methyl SA esterase to acibenzolar. This conversion is required for BTH-induced PR protein expression as BTH failed to induce PR1 in the methyl SA esterase silenced tobacco seedlings (Tripathi et al., 2010). Besides direct induction of plant defense responses, low doses of BTH can prime plant immunity. In Arabidopsis, this priming effect is dependent on NPR1 (Kohler et al., 2002; Goellner and Conrath, 2008). Induction of MAPKs and histone modifications have also been found to associate with and may explain this priming effect (Beckers et al., 2009; Jaskiewicz et al., 2011). Different from INA, BTH has been commercialized as an effective agrochemical.

The isothiazole-based synthetic plant immune inducer, Isotianil, was identified by Bayer AG and Sumitomo Chemical Co., Ltd., through comprehensive search for this type of compounds as protectant against both rice blast and rice blight. Besides rice, Isotianil has also been shown to protect wheat against powdery mildew, cucumber against anthracnose and bacterial leaf spot, Chinese cabbage against Alternaria leaf spot, pumpkin against powdery mildew, strawberry against anthracnose and peach against bacterial shot hole (Ogawa et al., 2011; Krämer et al., 2012). Isotianil does not have antimicrobial activity in vitro but relies on its strong immune inducing power to protect rice against rice blast. An exceptionally low dosage is enough to assure its in vivo antimicrobial effect (Ogawa et al., 2011). Its effective dose is lower than any other existing plant defense activators (Ogawa et al., 2011). Transcriptome profiling revealed that Isotianil induces the expression of defenserelated genes in rice including NPR1, NPR3, and WRKY family transcription factors as well as gene involved in SA catabolism (Krämer et al., 2012). Up till now, more in-depth molecular basis of how Isotianil achieves its immune eliciting activity has not been reported (Maienfisch and Edmunds, 2017).

## JA Analog

While SA regulates defense against biotrophic pathogens, JA and methyl-JA (MeJA) mainly control the immunity against necrotrophic pathogens and herbivores (Santino et al., 2013). JA can be metabolized to MeJA and JA-isoleucine (JA-Ile) which is a biologically active form (Svoboda and Boland, 2010; Pieterse et al., 2012). JA signal is transduced to transcription through JA-Ile triggered degradation of Jasmonate ZIM-domain (JAZ) type transcriptional repressors by the JA receptor, Coronatine Insensitive 1 (COI1) (Yan et al., 2013, 2018). With the removal of these repressors, JA-responsive genes are de-repressed and JA-dependent defense responses are activated (Browse, 2009; Pieterse et al., 2012; Monte et al., 2014). The phytotoxin, coronatine, is a natural structural and functional mimic of JA-Ile (Weiler et al., 1994; Fonseca et al., 2009). Coronatine can elicit similar responses as JA. In an effort to identify more potent mimics of coronatine, the synthetic JA mimic coronalon was synthesized (Schuler et al., 2001). Coronalon was later shown to mediate stress responses in various plants species (Schuler et al., 2004). It can induce known MeJA-activated defense products as well as MeJA-responsive genes (Pluskota et al., 2007). Besides coronalon, several synthetic JA mimics have been studied and shown to induce JA signaling and defense responses in lima bean, soybean and coyote tobacco (Krumm et al., 1995; Fliegmann et al., 2003; Pluskota et al., 2007). However, whether these JA mimics bind COI1 has not been investigated. Based on the co-receptor structure, a coronatine derivative, coronatine-O-methyloxime (COR-MO), was synthesized through direct chemical derivation and identified as a potent competitive antagonist of jasmonate perception (Monte et al., 2014).

# β-Aminobutyric Acid (BABA)

BABA is a non-protein amino acid that has been known to induce plant resistance since 1963 (Papavizas and Davey, 1963). It has been shown to protect about 40 different plant species against a diverse range of pathogen and pests including virus, bacteria, oomycetes, fungi, nematode, and arthropods (Cohen et al., 2016). BABA primes multiple defense mechanisms regulated by SAdependent and SA-independent pathways (Zimmerli et al., 2000; Ton et al., 2005). The priming effects elicited by BABA can be maintained to the next generation, making BABA the first plant immune inducer with transgenerational efficacy (Slaughter et al., 2012). BABA is sensed by an aspartyl-tRNA synthetase, IBI1 (Luna et al., 2014). Binding of BABA to IBI1 primes it for alternative defense activity. However, the inhibition of BABA on the aspartyl-tRNA synthetase activity leads to toxicity in plants, which makes BABA unsuitable for agricultural use. While BABA has long been considered as a synthetic plant immune priming agent, a recent study has unequivocally identified BABA as an endogenously metabolite synthesized by various plant species including Arabidopsis, Chinese cabbage, maize, teosinte, and wheat (Thevenet et al., 2017).

# EMERGING TRENDS

Large-scale screens performed by the private sector identified the first-generation synthetic elicitors including INA and BTH. Over the last 15 years, advances in combinatorial chemistry and development of high-throughput screening systems have equipped the scientists outside the private sector with the ability to carry out comprehensive screens for synthetic plant immune inducers. This has led to the discovery of a rich arsenal of the second-generation synthetic elicitors (Bektas and Eulgem, 2015). While systematic screens will continue to help us unveil new and better synthetic elicitors, approaches based on the knowledge of known synthetic and/or natural elicitors are emerging.

# Chemical Derivation

Simple chemical derivation of known immune inducers has been and continues to be a shortcut to the identification of more potent immune elicitors. Recently, a new class of SA derivative, benzoylsalicylic acid (BzSA) was identified from seed coats of Givotia rottleriformis, a soft-wood tree species (Kamatham et al., 2016). BzSA induces SAR-related gene expression more effectively than SA. It also induced more local and systemic resistance against TMV in tobacco than SA. Through relatively simple chemical derivation, Kamatham et al. (2017) synthesized 14 BzSA derivatives and tested their bioefficacy using the tobacco-TMV pathosystem. When low dosage was tested, all 14 derivatives caused more reduction of the lesion size than both SA and BzSA. The immune-inducing effects of BzSA derivatives are not dependent on SA accumulation as they can still induce resistance in nahG plants.

With the availability of a diverse collection of known synthetic and nature plant immune inducers, comparison between known elicitors may help identify specific moiety critical to the immune inducing ability. The 3-methylfuran-containing natural products like menthofuran, furanoeremophilane, caclol, and tanshinone are plant secondary metabolites involved in plant defense (Hägele and Rowell-Rahier, 2000; Maffei et al., 2012; Liu et al., 2013). Based on the prediction that 3-methylfuran moiety may be

important to the antimicrobial activity of these secondary metabolites, He et al. (2017) used diversity-oriented synthesis to generate a small natural-products-like library containing the 3 methylfuran scaffold. Five 3-methylfuran derivatives were found to significantly induce the resistance in rice against brown planthopper, supporting the initial speculation on the critical role of 3-methylfuran (He et al., 2017).

Besides specific functional moiety, the pattern of known immune elicitors can also be useful information for the design of new ones. Rhamnolipids and lipopeptides have been found as a new class of MAMPs (Jourdan et al., 2009; Sanchez et al., 2012; Farace et al., 2015). Both rhamnolipids and lipopeptides are amphiphilic compounds. Due to the biocompatibility and biodegradability, rhamnoside-based bolaamphiphiles surfactants have been increasingly recognized and investigated (Gatard et al., 2013; Akong and Sandrine, 2015). The bolaamphiphiles surfactants contains a long hydrophobic spacer connecting two hydrophilic moieties. Luzuriaga-Loaiza et al. (2018) synthesized rhamnolipid bolaforms (SRBs) and tested their immune induction activity. Depending on the acyl chain length, SRBs differentially induce defense responses and confer local resistance in Arabidopsis against the hemibiotrophic bacteria Pseudomonas syringae but not the necrotrophic fungal pathogen Botrytis cinerea.

Chemical derivation based on known natural immune inducers has great expedited the invention of better synthetic immune inducers. However, the lack of the mechanistic understanding of the interactions between the new synthetic immune inducers and their cognate targets in plants has limited our ability to improve the efficacy or lower the phytotoxicity in a more rational manner. More comprehensive biochemical studies using the new synthetic immune inducers will provide a promising guide.

# Bifunctional Combination

Bifunctional combination approaches combine a known synthetic plant immune inducer with another compound, which brings other functions to the final product. Strobilurins are a class of broad spectrum fungicides. Widespread use of strobilurins have caused pathogen resistance (Gisi et al., 2002; Leiminger et al., 2014). 3,4-dichloroisothiazole derivatives have diverse biological activities including immune-inducing activity. For example, as mentioned in Section 2.3, Isotianil, a 3,4 dichloroisothiazole derivative, is a very potent immune elicitor. In an effort to identify new strobilurins for future market, Chen et al. (2017) combined 3,4-dichloroisothiazoles with strobilurins. Through the incorporation of 3,4-dichloroisothiazole, new strobilurins with good in vivo and in vitro fungicide activities were identified.

JA-Ile is a natural conjugation of JA and isoleucine and was previously identified as the sole endogenous bioactive JA molecule. In an effort to identify additional endogenous bioactive jasmonates, Yan et al. (2016) coupled 20 natural amino acids with coronafacic acid (CFA) which is a part of the phytotoxic natural JA-Ile mimic, coronatine, and identified 5 non-polar amino acid conjugates of CFA including CFA-Ile, CFA-Leu, CFA-Val, CFA-Met, and CFA-Ala as new synthetic JA signaling pathway elicitors. Following these findings, JA-Leu, JA-Val, JA-Met, and JA-Ala were further discovered as new endogenous bioactive JA molecules. Through integration of the structural information of all these bioactive JA molecules, general rules of bioactive JA conjugates were proposed. Based on these rules, two additional JA signaling pathway elicitors, CFA-N-Leu and CFA-Ch-Gly were identified (Yan et al., 2016).

Besides covalent combination, ionic pairing is another attractive method, since one can choose ions independently. The same plant immune inducer can be paired with surfactanttype cation for better wetting or tetrabutylammonium cation for faster dissolution. Using this strategy, 15 immunity inducers including SA, BTH, INA, BABA, etc., were paired with the cholinium cation to form ionic liquids (Kukawka et al., 2018). Their abilities to induce SAR were tested using the tobacco-TMV pathosystem. Cholinium is an essential nutrient. Pairing with cholinium reduced phytotoxicity of these immune inducers while only mild perturbation to the immune-inducing ability was recorded.

While bifunctional combination approaches have shown the potential to either improve the efficacy or reduce phytotoxicity, the introduction of the second chemical moiety has also brought more complications. For example, Pip, a SAR mobile signal candidate, showed significantly reduced SAR-inducing activity when paired with cholinium (Kukawka et al., 2018). On the other hand, while isonicotinate did not induce SAR, its cholinium ionic liquid was shown to induce SAR (Kukawka et al., 2018). Therefore bifunctional combination is not merely the addition of the biological activities of the two chemical moieties but rather results in potentially complicated interactions between the signaling pathways induced by the two moieties. Careful characterization is thus essential to understand the full spectrum of the biological activities of the new synthetic immune inducers identified through bifunctional combination approaches.

# Computer-Aided Design

Manual inspection can only process a handful of immune elicitors for recognition of potentially critical bioactive substructures and patterns of known immune inducers (He et al., 2017; Luzuriaga-Loaiza et al., 2018). Advances in high-performance computing have made it possible to screen tens of thousands of leadlike molecules computationally. This computer-aided design (CAD) drug design strategy has been increasingly recognized and utilized in pesticide discovery and property analysis (Xia et al., 2014; Veselinovi et al., 2015; Burden et al., 2016). Using SA, MeSA, BTH, and Tiadinil, the four known immune inducers as query templates, Chang et al. (2017) performed virtual screening against the 5,3000 hit-like and lead-like compounds in the Maybridge database and identified three benzotriazole scaffolds as promising leading compounds. One of them, L1 shows high 3D structure similarity to BTH despite their differences in 2D topology. Furthermore, L1 also shares similar pharmacophore features to BTH. In vivo screening of L1 derivatives identified new immune inducers with comparable or improved efficacy against Mycosphaerella melonis, Corynespora cassiicola, P. syringae, B. cinerea, and F. oxysporum in cucumber, Phytophthora infestans in tomato and R. solani in rice.

Besides the knowledge of the small lead-like compounds, structural understanding of plant receptors can also lend power to the virtual screening of new leading compounds. Using the high quality structural model of JA receptor, COI1, 767 JA analogs were analyzed in terms of their ability to bind COI1 (Pathak et al., 2017). Two such analogs ZINC27640214 and ZINC43772052 showed higher binding affinity compared to JA. ZINC27640214 appears to have efficient, stable and good cell permeability properties, making it a good candidate for experimental validation. Buswell et al. (2018) combined the knowledge of the structural information on the BABA receptor, IBI1 and small-scale screening of β-amino acids using the ibi1 mutant to search for BABA analogs, which induce plant immunity without severe growth inhibition. Out of the seven resistanceinducing compounds, five of them showed no inhibition on growth. Among these five, (R)-β-homoserine (RBH) showed the strongest resistance-inducing activity without affecting vegetative growth or global plant metabolism. Interestingly, RBH appears to elicit partially different signaling pathways from those affected by BABA, making it a promising new crop protectant. Through in silico docking and subsequent molecular dynamics simulation, the keto group of a stereoisomer of coronatine showed the potential to control the binding selectivity between its derivatives and different subtypes of JAZ (Takaoka et al., 2018). An oxime derivative of this coronatine stereoisomer was then developed as a synthetic JAZ subtypeselective agonist, specifically targeting JAZ9 and JAZ10. This selectivity in JAZ enabled induction of pathogen resistance without a cost on growth. It is noteworthy that small-scale targeted characterization of synthetic agonist candidates rather than large-scale screening was realized in this study owing to the integration of the structural information on both the ligands and the receptor.

As an emerging trend, application of CAD in the discoveries of new synthetic immune inducers awaits further exploitation. While lead-like compound databases have provided a critical foundation for virtual chemical screening, they also restrain the chemical diversity and may potentially hinder the discovery of completely novel scaffolds. On the other side, CAD based on the structural information of plant defense signaling components does not set a limit on the chemical diversity. However, synthetic immune activators identified through this route may be only effective in the specific plant species studied due to the sequence variation among different plant species. Integration of evolutional conservation information may help alleviate this issue.

# REFERENCES


# CONCLUSION AND PERSPECTIVES

In this review, we provided a focused overview on the discovery and functional properties of synthetic plant immune inducers and emerging trends in the search for new and improved synthetic inducers. A rich knowledge of the structural, chemical and pharmacological properties of the known inducers has opened up some shortcuts to expedite the discovery procedure. Instead of in vivo screening tens of thousands small molecules, small-scale screening involves only a few dozens or even a handful of compounds is able to identify new inducer derivatives or even completely new scaffolds through integration of prior knowledge. While the availability of the structures of small compounds is the major drive for this advancement, we anticipate that integration of more prior information will greatly facilitate the discovery of novel and better plant immune elicitors. This includes the structural information, biological function and evolutional conservation of key plant immune-related signaling components, physical and biochemical features of the small compounds as well as the structural basis and evolutional conservation of the molecular interactions between small compounds and their cognate plant immune signaling components.

The great expansion of synthetic immune inducers has also provided opportunities to dissect the signaling networks of plant immune system that is not accessible to genetic screens due to the lethality and gene redundancy. With the discovery of the hidden drug-able targets in plant immune system, new synthetic immune inducers may be developed to target these hidden points. Then in turn, these new inducers can again enhance our ability to dissect plant immune system and keep this discovery cycle going on.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# FUNDING

This work was supported by the funds from School of Life Sciences, Peking University – Tsinghua University Joint Center for Life to WW.

priming of stress responses in Arabidopsis thaliana. Plant Cell 21, 944–953. doi: 10.1105/tpc.108.062158

Bektas, Y., and Eulgem, T. (2015). Synthetic plant defense elicitors. Front. Plant Sci. 5:804. doi: 10.3389/fpls.2014.00804

Brisset, M. N., Faize, M., Heintz, C., Cesbron, S., Chartier, R., Tharaud, M., et al. (2002). Induced resistance to Erwinia amylovora in apple and pear. Acta Hortic. 590, 335–338. doi: 10.17660/ActaHortic.2002.590.49

Browse, J. (2009). Jasmonate passes muster: a receptor and targets for the defense hormone. Annu. Rev. Plant Biol. 60, 183–205. doi: 10.1146/annurev.arplant. 043008.092007


rottleriformis induces systemic acquired resistance in tobacco and Arabidopsis. Phytochemistry 126, 11–22. doi: 10.1016/j.phytochem.2016.03.002


resistance signalling in Arabidopsis. Nature 419, 399–403. doi: 10.1038/nature 00962




resistance in Arabidopsis without salicylic acid accumulation. J. Agric. Chem. Soc. Japan 67, 322–328.


**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 Zhou 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.

# Non-TAL Effectors From Xanthomonas oryzae pv. oryzae Suppress Peptidoglycan-Triggered MAPK Activation in Rice

Juying Long1,2† , Congfeng Song1,2† , Fang Yan<sup>3</sup>† , Junhui Zhou<sup>2</sup> , Huanbin Zhou<sup>3</sup> \* and Bing Yang2,4,5 \*

#### Edited by:

Zhengqing Fu, University of South Carolina, United States

### Reviewed by:

Seiji Tsuge, Kyoto Prefectural University, Japan Kaijun Zhao, Institute of Crop Sciences (CAAS), China

#### \*Correspondence:

Huanbin Zhou hbzhou@ippcaas.cn Bing Yang yangbi@missouri.edu †These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 05 September 2018 Accepted: 30 November 2018 Published: 12 December 2018

#### Citation:

Long J, Song C, Yan F, Zhou J, Zhou H and Yang B (2018) Non-TAL Effectors From Xanthomonas oryzae pv. oryzae Suppress Peptidoglycan-Triggered MAPK Activation in Rice. Front. Plant Sci. 9:1857. doi: 10.3389/fpls.2018.01857 <sup>1</sup> Key Laboratory of Monitoring and Management of Plant Diseases and Insects, Ministry of Education, Nanjing Agricultural University, Nanjing, China, <sup>2</sup> Department of Genetics, Development, and Cell Biology, Iowa State University, Ames, IA, United States, <sup>3</sup> Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China, <sup>4</sup> Division of Plant Sciences, University of Missouri, Columbia, MO, United States, <sup>5</sup> Donald Danforth Plant Science Center, St. Louis, MO, United States

Xanthomonas oryzae pv. oryzae, the causal pathogen of bacterial blight of rice, depends on its type III secretion system and associated effector proteins to grow and colonize the vascular tissues of rice plants. The type III effectors include a family of closely related transcription activator-like (TAL) effectors and the rest of diverse effectors, so-called non-TAL effectors. Our understanding of non-TAL effectors for pathogenesis in rice blight is still limited. Here we report a feasible method to rapidly detect the activation of mitogenactivated protein kinase pathway in rice mesophyll protoplasts by the X. oryzae pv. oryzae derived peptidoglycan and screen for virulent effectors that can suppress the pathogen-associated molecular pattern triggered immunity (PTI) response. Amongst 17 non-TAL effectors transiently expressed in rice cells, we found that three effectors (XopZ, XopN, and XopV) were able to suppress the peptidoglycan-triggered MAPK activation. The triple mutant of the X. oryzae pv. oryzae strain PXO99<sup>A</sup> lacking XopZ, XopN, and XopV showed additively reduced virulence. Adding back either of genes restored the virulence of the triple mutant. Our results demonstrate the collective and redundant ability of defense suppression by non-TAL effectors in causing bacterial blight of rice.

Keywords: Xanthomonas, type III effector, TAL effector, non-TAL effector, MAPK, immunity, rice

# INTRODUCTION

Plants ward off the infection of microbial pathogens through two major layers of defense, namely pathogen-associated molecular pattern (PAMP) triggered immunity (PTI) and effector triggered immunity (ETI) (Boller and Felix, 2009). PTI is initiated upon perception of PAMPs by the pattern recognition receptors (PRRs) at the cell surface and subsequently activation of a cascade of gene expression and complex signaling pathways, leading to a basal disease resistance in plants. The PAMPs include lipopolysaccharide (LPS), peptidoglycan (PGN), flagellin, chitin, elongation factor Tu, glucan and even DNA fragments of bacterial and fungal origins. Bacterial and fungal

pathogens can also use a suite of effector proteins to suppress the PTI to enable microbial growth and disease symptom development. The effectors are virulence factors per se. However, they could be recognized by host resistance genes or gene products, leading to stronger burst of host defense responses including the hallmark character of hypersensitive reaction (HR), a process of rapid, localized cell death, to limit the microbial proliferation, the second layer of host immunity (ETI) (Jones and Dangl, 2006).

Bacterial blight of rice, inflicted by the pathogenic Xanthomonas oryzae pv. oryzae (Xoo), causes great losses of rice production in south Asian and west African countries (Mew et al., 1993). Xoo cells enter rice leaves through hydathodes and wounds, followed by their colonization and spread in vascular tissues of leaves and sheaths, and eventually cause leaf blight characteristic of gray to white opaque necrotic lesion in leaf and stem of rice plants (Niño-Liu et al., 2006). Disease control based on genetic resistance remains the only effective way in the field (Mew, 1987).

At the molecular level, the blight is the outcome of interactions between host rice and pathogen Xoo. For example, the apoplastdwelling pathogen uses its type III secretion system (T3SS) to translocate the bacterial effector proteins into host rice cells to manipulate the host transcriptional and physiological processes, rendering host more susceptible to bacterial growth and symptom progression. In this case, each strain of Xoo contains about 35 type III effectors that can be divided into two groups: TALEs (transcription activator-like effectors) and non-TALEs. The interaction of TALEs and host genes can be characterized as protein/DNA interaction, i.e., TALEs, once internalized into nuclei of host cells, recognize the promoters and transcriptionally activate the disease susceptibility genes, inducing a state of disease. In addition, a group of TALE variants (e.g., so-called iTALEs or truncTALEs, due to C-terminal truncations of typical TALEs) involve ETI suppression of XA1 mediated resistance against most, if not any, full-length TALEs, which could be characterized as protein/protein interaction (Ji et al., 2016). On the other hand, the non-TALEs include a suite (ca.18-23) of structurally diverse type III effectors. The functions of some non-TALEs have been characterized as virulence factors by suppressing PTI in bacterial blight of rice (Song and Yang, 2010; Akimoto-Tomiyama et al., 2012; Yamaguchi et al., 2013a,b; Zhao et al., 2013; Ishikawa et al., 2014; Wang et al., 2016; Qin et al., 2018). Overall, the studies for the molecular interaction have been largely focused on the TALE biology and major breakthrough have been made in the last decades. However, our understanding of type III effectors beyond TALEs (so called non-TALEs) is limited.

In present study, we established a protoplast system to study the MAPK activation as one of PTI processes in rice and screen type III effectors for PTI suppressors. The PTI process involves the MAPK (mitogen-activated protein kinase) activation in response to peptidoglycan (PGN), a common PAMP extracted from Xoo. We identified three Xop (Xanthomonas outer protein) effectors that were able to suppress the MAPK activation and revealed their redundant role in virulence in Xoo infection process and lesion formation.

# MATERIALS AND METHODS

# Plant, Bacterial Strains, and DNA Manipulations

The wild-type Arabidopsis thaliana Col-0 was used for isolation of leaf mesophyll protoplasts from 3–4-week-old plants. The japonica rice (Oryza sativa) cultivar Kitaake was used for isolation of protoplasts from young seedlings and 5–6 weeks old plants for disease assay. Strains of Escherichia coli and Xoo used in this study are listed in **Table 1**. Bacterial culture and DNA manipulation were performed with standard techniques (Ausubel et al., 1998). Xoo was grown in either nutrient broth (NA) (BD, Difco) or tryptone sucrose medium (TS) (tryptone, 10 g; sucrose, 10 g; glutamic acid, 1 g; Difco Bacto agar, if solid, 15 g per liter) at 28◦C. Plasmids were transferred into E. coli or Xoo through electroporation. Antibiotics used in this study were ampicillin (100 µg/ml), cephalexin (10 µg/ml), kanamycin (50 µg/ml), and spectinomycin (100 µg/ml).

# Plasmid Construction

The plasmids for expression of rice MPK1 (Os06g06090) and MPK5 (Os03g17700) were constructed by cloning their cDNA-derived PCR (polymerase chain reaction) amplicons into pUC:35S. pUC:35S was a pUC19 derived vector containing the 35S promoter, Nos terminator and multiple cloning sites (MCS) between. The PCR fragment of each MPK was cloned into pUC:35S between the BamHI and HindIII restriction sites of MCS and in frame at its 3<sup>0</sup> end with a FLAG-coding sequence. pHBT-DMEKK1-FLAG expressing the Arabidopsis 1MEKK1 was kindly provided by Ping He. All non-TAL effector genes were PCR-amplified with gene-specific primers and genomic DNA of PXO99<sup>A</sup> using the Phusion high-fidelity DNA polymerase (New England Biolabs). The amplicons were individually cloned at the appropriate restriction sites into an expression vector a modified pUC:35S containing the 35S promoter and in fused with a sequence encoding an HA epitope-tag at the C terminus. Construct to express the Xoo flagellin was made through PCR-amplification of fliC coding region with genomic DNA of PXO99<sup>A</sup> and cloning into the E. coli expression vector pHTb at BamHI and HindIII. The recombinant protein flagellin N-terminally fused with 6xHIS tag was produced in BL21 Star (DE3)pLysS (Thermo Fisher Scientific) and purified with Ni-NTA Agarose (QIAGEN) following the manuals of both manufacturers. The primer sequences for PCR-amplification are provided in **Supplementary Table S1**.

# Protoplast Preparation and PEG-Mediated Transfection

The Arabidopsis mesophyll protoplasts were isolated from rosette leaves as previously described (Yoo et al., 2007). Rice

protoplast preparation and transfection were carried out as reported before (Jiang et al., 2013). Briefly, rice seed was surfacesterilized with 50% bleach, germinated and grown on 1/2 MS medium + B5 Vitamins containing 1.5% sucrose and 0.6% agar in ice cream cone cups in growth chamber at 30◦C and with 12 h lighting. Leaves and leaf stems of 8–10 days old seedlings were used for protoplast isolation and PEG-mediated transfection. The protoplasts were harvested at the time points post-transfection as specified in test for further analysis. For immunoblot detection of endogenous MAPK activation and pull-down assay, an aliquot of 0.8 ml of protoplasts at a density of 2 × 10<sup>6</sup> /ml was transfected with 30 µg of each

TABLE 1 | Bacterial strains and plasmids used in this study.


plasmid DNA (OsMAPK1, OsMAPK5, OsMAPK6, 1MEKK1 and/or Xop genes). Each experiment was repeated at least three times.

# Peptidoglycan (PGN) Extraction and Elicitor Treatment

Isolation and purification of PGN were carried out from the Xoo strain PXO99<sup>A</sup> as described previously (Barnard and Holt, 1985). Briefly, the bacterial cells were boiled in 100 volumes of 5% SDS for 30 min. After cooling down to room temperature, the SDS-insoluble material was collected by centrifuging at 30,000 g at 4◦C for 30 min. The pellet was treated with trypsin digestion to remove protein contamination and dialyzed extensively against distilled water.

Arabidopsis and rice protoplasts were treated with flg22 (Sigma-Aldrich) at a final concentration of 1 µM. Chitin (Sigma-Aldrich) was applied to rice protoplasts at a final concentration of 10 µM. Before flg22 or chitin treatment, rice and Arabidopsis protoplasts were isolated and incubated for about 12 h in W5 solution under dark, then incubated under light for 2 h in fresh W5 solution. After treatment (e.g., 10 min), protoplasts were harvested for immunoblot analysis. For PGN treatment, protoplasts were collected after 12 h posttransfection of effector gene constructs and resuspended in 100 µl of W5 solution. After recovered under light for 2 h in the 30◦C growth chamber, protoplasts were treated with 5 mg/ml PGN for 15–60 min and then harvested for immunoblot analysis.

# Pull-Down Assay

Total proteins were extracted from protoplasts with 0.5 ml of lysis buffer that is composed of 50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10% glycerol, 0.1% NP-40, 0.5 mM DTT, 1 mM PMSF and protease inhibitor cocktail (Sigma-Aldrich). Cell debris was pelleted at 20,000 g for 15 min at 4◦C. The supernatants were incubated with 30 µl of Anti-FLAG M2 Magnetic Bead (Sigma-Aldrich) for at least 1 h at 4◦C with gentle shaking, the beads were then washed three times with lysis buffer. An aliquot of 2X SDS loading buffer (25 µl) was added to the beads, and then heated at 95◦C for 5 min. 10 µl of each sample was loaded on the protein gel for immunoblot analysis as described below.

# Immunoblot Analysis

Proteins were separated through electrophoresis in 12% SDS-PAGE gel and electro-transferred to Hybond-P PVDF membranes (Amersham Biosciences) for blotting. Immunodetection was carried out using standard methods. The anti-HA, anti-FLAG (Santa Cruz Biotechnology) and anti-Phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) antibodies (Cell Signaling Technology) were used at a dilution of 1:1,000. HRP-conjugated secondary antibody (Sigma-Aldrich) was used at a dilution of 1:20,000. Blots were visualized with SuperSignal West Pico Chemiluminescent Substrates (Thermo Scientific) following the manufacturer's instructions.

# Generation of Individual and Triple Mutant for XopN, XopV, and XopZ

fpls-09-01857 December 10, 2018 Time: 14:0 # 4

The suicide vector pKMS1 was used to individually and sequentially delete the three Xop genes (XopN, XopV, and XopZ) for a triple mutant using a method as described (Zou et al., 2011). Specifically, two pairs of gene-specific primers (sequences provided in **Supplementary Table S1**) were designed to match two separate regions of each gene based on the PXO99<sup>A</sup> genome sequence (NCBI accession, CP000967).

The PCR was performed to amplify the upstream and downstream regions of each locus by using the PXO99<sup>A</sup> genomic DNA as template. The two PCR amplicons for each gene deletion were fused by overlap PCR and cloned into the MCS (multiple cloning site) of pKMS1 and confirmed by sequencing for accuracy of sequence. The mutagenesis was performed on PXO99<sup>A</sup> targeting individual Xop genes. Plasmid DNA was electroporated into competent cells of PXO99<sup>A</sup> and transformants were plated on the kanamycin NA lacking sucrose. Single colonies were transferred to nutrient broth medium lacking sucrose and incubated with shaking for 12 h at 28◦C. Bacterial cells were then plated on NA containing sucrose. Sucrose tolerant colonies were duplicated on NA and kanamycin-containing NA plates. The kanamycin-sensitive colonies were screened for gene deletions by PCR with gene deletion-specific primers. The mutants were used for sequential deletion through the second and third rounds of mutagenesis for triple mutant using the similar approach.

# Disease Assay

The rice plants were grown in growth chamber containment with temperature of 28◦C and relative humility of 75% and photoperiod of 12 h in light/dark. Fully expanded leaves of 2-month-old rice plants (n = 5–10) were used for leaf tipclipping inoculation with the bacterial concentration at 0.5 OD<sup>600</sup> (approximately 5.0 × 10<sup>7</sup> ). Lesion lengths were measured 12 days post inoculation (DPI) from leaf tip to the diseased edge of lesions. One-way analysis of variance statistical analyses was performed on all measurements. The Tukey's honest significant difference test was used for post analysis of variance pair-wise tests for significance, set at 5% (p < 0.05).

# RESULTS

# Establishment of a Method to Rapidly Study PAMP-Triggered Immunity in Rice Mesophyll Protoplasts

A number of non-TAL effectors of Xanthomonas genus were identified in suppressing flg22-induced signaling pathway and inhibiting early defense gene expression in the dicot model plant A. thaliana (Popov et al., 2016; Wang et al., 2016). However, little is known about its real functions in the natural host species, like rice. Therefore, we sought to set up a simple and quick system for PAMP signaling studies and screening for virulent

FIGURE 1 | Pathogen-associated molecular pattern (PAMPs) activate MAPK cascade in rice protoplasts. (A) Rice protoplasts do not respond to flg22 treatment. Rice protoplasts were treated with 1 µM flg22, MAPK activation (top) and loading control (bottom) are shown. (B) flg22 activates endogenous MAPK cascade in Arabidopsis protoplasts. Arabidopsis protoplasts were treated with 1 µM flg22, MAPK activation (top) and loading control (bottom) are shown. (C) Endogenous MAPK activation is induced by PGN under the lights. Rice protoplasts were treated with or without 5 mg/ml PGN, MAPK activation (top) and loading control (bottom) are shown. (D) PGN has weak effect on endogenous MAPK activation in the dark. Rice protoplasts were treated with or without 5 mg/ml PGN, MAPK activation (top) and loading control (bottom) are shown. (E) Endogenous MAPK cascade is activated at different temperature under the lights. Rice protoplasts were treated with or without 5 mg/ml PGN under the lights for 10 min, MAPK activation (top) and loading control (bottom) are shown. (F) Chitin activates endogenous MAPK cascade in rice protoplasts. Rice protoplasts were treated with 10 µM chitin, MAPK activation (top) and loading control (bottom) are shown. The data shown here are one representative of three independent experiments.

effectors in the rice mesophyll protoplasts by immunoblotting with p44/42 MAP kinase antibody, which specifically recognizes phosphorylated forms of MAPK kinases. When the widely used flg22 peptide was first applied in the rice protoplasts which were recovered in the dark at room temperature, no elevated OsMAPK activation was observed at the early time points (**Figure 1A**). By contrast, flg22 perception triggered strong MAPK activation in Arabidopsis protoplasts within 10 min under the same conditions (**Figure 1B**). Furthermore, when flagellin of Xoo was ectopically expressed in E. coli, purified and used to treat rice protoplasts, no OsMAPK activation was detected either, which is consistent with the notion that Xoo evades the flagellin detection system in rice (Wang et al., 2015). Therefore, we assume that the Xoo flagellin and the Pseudomonas syringae flg22 epitope may not be the conserved PAMPs to induce PTI responses in rice.

In this regard, peptidoglycan (PGN), another type of the wellknown PAMPs, was extracted from the PXO99<sup>A</sup> and further tested for its capability of triggering OsMAPK activation with and without lights at room temperature. As shown in **Figure 1**, PGN induced OsMAPK phosphorylation more obviously under lights (**Figure 1C**) compared to that in the dark (**Figure 1D**); the enhanced level was maintained at 30 min and restored to the normal level at 60 min with PGN treatment. Next, we tested if high temperature would affect the PGN-induced OsMAPK phosphorylation level given that the optimal growth temperature for rice is near 30◦C. The rice protoplasts were recovered and treated with PGN at different temperatures (20 and 30◦C), we found that rapid induction of OsMAPK phosphorylation occurred within 10 min at both temperatures and it was more consistent at 30◦C than 20◦C (**Figure 1E**). Thus, we carried out all downstream experiments with the rice protoplast system under lights and at 30◦C. In addition, chitin, a typical fungal PAMP that trigger various defense responses in both monocots and dicots, was tested in our system as well. The results revealed that it triggered rapid activation of OsMAPK cascade as PGN did (**Figure 1F**). Taken together, these data indicate an establishment of a fast and sensitive system for test of PTI in rice protoplasts, which is suitable for studies on PTI-related molecules derived from both bacterial and fungal pathogens.

# PGN-Triggered PTI Pathway in Rice

Chitin-induced MAPK activation is regulated by common homologous elements in both rice and Arabidopsis, which involves pattern-recognition receptors (PRRs), receptor-like cytoplasmic kinases (RLCKs) and three sequentially activated mitogen-activated protein kinases (MAPKs) (Kawasaki et al., 2017; Yamada et al., 2017). Early studies show that OsLYP4/6 and OsRLCKs play important roles in PGN signaling, but little is known about OsMAPK functioning downstream of PAMP perception (Liu et al., 2012; Ao et al., 2014; Li et al., 2017). To address this question, we cloned three rice homologous genes (OsMAPK1, OsMAPK5, and OsMAPK6) of Arabidopsis MPK3, MPK4, and MPK6 which are important in the flg22 dependent phosphorylation pathway related to PTI defense response, expressed individually in rice protoplasts followed by PGN treatment, immuno-precipitated with anti-FLAG M2 magnetic beads, and tested for activation by immunoblotting with p44/42 MAP kinase antibody. The results indicated that both OsMAPK1 and OsMAPK5 were activated after PGN application (**Figure 2A**). The involvement of OsMAPK6 cannot be determined, however, resulting from its poor expression.

Meanwhile, we examined whether PGN extracted from the Xoo was able to activate MAPKs in Arabidopsis protoplasts as well. Treatment of protoplasts with PGN induced rapid and

FIGURE 2 | OsMPK1 and OsMPK5 mediate PGN-induced MAPK activation in rice protoplasts. (A) OsMPK1 and OsMPK5 are activated by PGN in rice protoplasts. OsMPK1 or OsMPK5 were expressed in rice protoplasts overnight and immuno-precipitated with anti-FLAG-beads after 5 mg/ml PGN treatment for 10 min. MAPK activation (top) and MAPK expression (bottom) are shown. (B) PGN extracted from Xoo activates endogenous MAPK cascade in Arabidopsis protoplasts. Arabidopsis protoplasts were treated with 5 mg/ml PGN. MAPK activity (top) and loading control (bottom) are shown. (C) Constitutively active AtMEKK1 activates endogenous MAPK cascade in rice protoplasts. Rice protoplasts were transfected with construct expressing constitutively active MAPKKK of Arabidopsis. MAPK activation (top), MAPKKK expression (middle) and loading control (bottom) are shown. (D) Constitutively active AtMEKK1 activates OsMPK1 and OsMPK5 in rice protoplasts. OsMPK1 or OsMPK5 were co-expressed with constitutively active AtMEKK1 in rice protoplasts overnight, and then immuno-precipitated with anti-FLAG-beads. MAPK activation (top) and MAPK expression (bottom) are shown. The data shown here are representative of five independent experiments.

strong MAPK activation within 10 min (**Figure 2B**), suggesting PGN is a conserved PAMP that could be recognized in both rice and Arabidopsis. Therefore, 1MEKK1, a constitutively active catalytic domain (326–592) of MEKK1 that is known to activate MPK3 and MPK6 in Arabidopsis (Asai et al., 2002), in plasmid pHBT-DMEKK1-FLAG, was tested for activating OsMAPKs in rice protoplasts. The results showed that transient expression of the Arabidopsis-derived1MEKK1 resulted in OsMAPK activation to an extent as PGN did (**Figure 2C**). The FLAG-tagged OsMAPKs were further co-expressed individually with1MEKK1 in rice protoplasts, immunoprecipitated and tested for activation. As shown in **Figure 2D**, both OsMAPK1 and OsMAPK5 were highly phosphorylated in presence of1MEKK1. In light of these findings, we assume that PGN-triggered activation of MAPK cascades is conserved in rice and Arabidopsis.

# Identification of the Virulent Non-TAL Effectors of Xoo Related to PGN-Triggered PTI

Type III secreted effectors interfere with plant cellular pathways to benefit the pathogen and promote bacterial multiplication during infection. Of which, targeting MAPK cascades has been found to be a common strategy for a wide range of bacterial pathogens (Bi and Zhou, 2017). However, compared to Hop effectors from Pseudomonas genus that have been intensively studied, most of non-TAL effectors from Xoo have not been well characterized. Therefore, we sought to identify the virulent effectors of Xoo that have potential to suppress OsMAPK cascades in rice.

There are 18 non-TAL effector-encoding genes, including 2 identical copies of XopZ, annotated in the genome of the Xoo strain PXO99<sup>A</sup> (Song and Yang, 2010). Therefore, we cloned a total of 17 Xop genes and expressed the HA-tagged fusion proteins individually in rice protoplasts followed by PGN treatment. Western blot analysis with p44/42 MAP kinase antibody clearly showed that XopN and XopV inhibited the PGN-induced phosphorylation of OsMAPKs (**Figure 3A**). XopZ, a virulent effector we identified previously through mutagenesis of the strain PXO99A, was also capable of inhibiting the activation (**Figure 3B**). Expression of all Xop effectors were verified by immunoblotting with anti-HA antibody (**Figures 3A,B**, the lower panels). Thus, expression of three Xop effectors from PXO99<sup>A</sup> in rice protoplasts results in compromised OsMAPK activation induced by PGN, highlighting their putative virulence functions during pathogenesis.

# The Virulence Contribution of XopZ, XopN, and XopV

Since three Xop effectors in PXO99<sup>A</sup> are individually capable of suppressing MAPK activation induced by PGN in rice protoplasts and MAPK signaling is known to play an important role in PTI in rice, we resonate that those Xop effectors are important virulence factors in blight disease. We next tested the individual and collective role of virulence played by the three Xop effectors in pathogenesis with PXO99A. Single, double

and triple gene mutants were generated by deleting, individually and in combination, XopN, XopV and two copies of XopZ in PXO99A. All the mutants grew in nutrient medium normally without obvious growth defect. The disease assay through leaf tip-clipping inoculation in japonica rice variety Kitaake revealed no obvious virulence reduction by mutants of individual XopN and XopV (**Supplementary Figure S1**), like in IR24, an indica rice variety in a prior study (Song and Yang, 2010). The double mutant of XopN and XopV also caused lesion lengths similar to the wild type (**Figure 4A**). In contrast to the prior disease assay in IR24, XopZ single mutant lacking two XopZ loci did not show obvious different lesion length from PXO99<sup>A</sup> in Kitaake (**Figure 4A**). However, a significant decrease in lesion length with the triple mutation was observed (**Figure 4A**). The reduction in virulence in the triple mutant was restored by introduction of XopZ (**Figure 4B**), XopN and XopV (**Supplementary Figure S1**) in term of lesion lengths in Kitaake. The results indicate that XopZ, XopN, and XopV collectively contribute strain virulence in rice Kitaake.

## DISCUSSION

In our study, we established a rice protoplast system and used it to detect the early and quick activation of MAPK after treatment with PGN derived from Xoo. The system enabled us to identify individual XopN, XopV, and XopZ that were capable of interfering MAPK signaling pathway through transiently expressing each protein from a complement of 17 non-TALEs from PXO99A. Plant species including rice respond to microbial infection by activating multiple defense mechanisms. MAPK signaling pathway represents such an important component in coordinating the early detection and quick defense response to invading plant pathogens (Pedley and Martin, 2005). MAPK cascades are used by eukaryotic organisms to transduce signals upon perception of extracellular molecules derived from bacterial and fungal pathogens (Bi and Zhou, 2017). The protoplast system can be used as a simple and reliable system to study some aspects of host cellular response to pathogen stimuli. MAPK activation and its mediated host defense responses against bacterial and fungal diseases have been demonstrated in rice (Reyna and Yang, 2006; Yang et al., 2015; Jalmi and Sinha, 2016; Ma et al., 2017). The use of protoplasts enables to synchronize the whole collection of cells to respond to the stimulus (PGN in our study), resulting in quick and uniform activation of MAPK activation. In contrast, it is challenging to achieve this in planta where tissues contain multiple layers of cells and a waxy structure on surface. Furthermore, expression of individual effector protein from a suite of type III effectors in protoplasts enables us effectively to tease out the consequence of other effectors, while in a natural phytopathogen systems, including bacterial blight of rice, pathogen secretes a repertoire of effector proteins into host cells to collectively benefit the infection and disease development (Galán et al., 2014; Büttner, 2016). It is worthy to note that the protoplast system may have a limitation in discovery of effectors that act on the host preinvasive immunity, such as stomatal defense, which is the first layer of barrier that pathogen should overcome to get entry of the host extracellular spaces (Melotto et al., 2006, 2017).

Bacterial blight of rice is one of the most important crop diseases and a model for studying host/microbe interaction (Zhang and Wang, 2013). The interaction is bridged by a T3SS and exerted by factors from both pathogen Xoo and its host rice. The pathogenesis determinants associated with T3SS include a large family of TALEs and about 20 non-TALE proteins in individual Xoo strains<sup>1</sup> . Several non-TAL type III effectors (e.g., XopK, XopP, XopR, XopY, XopZ, and XopAA) have been found to suppress PTI in rice or the heterologous systems (Song and Yang, 2010; Akimoto-Tomiyama et al., 2012; Yamaguchi et al., 2013a,b; Ishikawa et al., 2014; Wang et al., 2016; Qin et al., 2018). For example, XopR from Xoo, when conditionally expressed in Arabidopsis, was capable of enhancing bacterial growth of a T3SS defective hrcC mutant of X. campestris pv. campestris, and also capable of suppressing the

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

induction of defense genes by this mutant and callose deposition induced by the flg22 peptide of flagella (Akimoto-Tomiyama et al., 2012). In another study, XopR has been found to physically interact with the rice BIK1 protein, a receptor-like cytoplasmic kinase (RLCK) and component of immune receptor complex, and it again, when expressed in Arabidopsis, suppresses PAMPtriggered stomatal closure (Wang et al., 2016). It has also been found that some non-TAL effectors suppress the upstream components of MAPK pathway, i.e., OsRLCK185 by XopY (or Xoo1488) and OsBAK1 by XopAA (or Xop875), and OsSERK1 by XopK (Yamaguchi et al., 2013a,b; Qin et al., 2018). Finally, XopP has been found to interact with the rice E3 ubiquitin ligase OsPUB44 and inhibit its ligase activity, thus suppressing immunity to Xoo in rice (Ishikawa et al., 2014). However, the molecular or biochemical mechanism in interfering with host immunity by other effectors remains unknown. By using MAPK activation in response to PGN as a readout coupled with expression of individual 17 non-TALEs, we found that three of them could suppress the MAPK activation, suggesting MAPK signal pathway as the target of those three effectors for virulence in blight disease. However, what exactly the interference in the cascade of MAPK (e.g., MAP kinases or upstream MAPKK or MAPKKK) by each of three Xop effectors remains to be determined.

In the model Xoo strain PXO99A, 18 type III effectors were identified as non-TALEs with XopZ as a virulence factor through systematic mutagenesis. XopZ was found to suppress the callose deposit in Nicotiana benthamiana when ectopically expressed. The XopZ knockout mutant of PXO99<sup>A</sup> showed reduced virulence when it was inoculated in the indica rice IR24 (Song and Yang, 2010). However, when the same mutant was assessed in the japonica rice Kitaake, no obvious virulence reduction was observed. This might be due to the host genetic context or genetic interaction of both host and pathogen. Similar phenomenon has been observed in XopK from PXO99<sup>A</sup> (Kitaake vs. IR24) (Qin et al., 2018) and XopN of KXO85 (Cheong et al., 2013). Similarly, single gene knockouts for other two non-TAL effectors XopN and XopV did not show obvious virulence reduction either, which is consistent with that in the prior study (Song and Yang, 2010). Indeed, the triple mutant of XopN, XopV,

# REFERENCES


and XopZ showed significant virulence reduction compared to their single knockout and parental strains, and the phenotype could be restored by introduction of single Xop gene, suggesting an involvement of MAPK activation in immunity and functional redundancy of those three Xop effectors in MAPK suppression. Better understanding of how those virulence determinants in Xoo to manipulate host MAPK signaling pathways will enhance host disease resistance by delicately engineering MAPK cascade given the fact that MAPK involves multiple processes of biotic and abiotic stress tolerance (Meng and Zhang, 2013; Xu and Zhang, 2015).

# AUTHOR CONTRIBUTIONS

JL, CS, FY, and JZ performed the experiments and analyzed the data. HZ and BY conceived, designed, and coordinated the research. All authors read and approved the final manuscript.

# FUNDING

This research was supported by the USDA NIFA research grant (Award No. 2017-67013-26521 to BY), the National Key Research and Development Program of China (Award No. 2017YFD0200900 to HZ), and the National Natural Science Foundation of China (Award No. 31471744 to CS).

# ACKNOWLEDGMENTS

The authors thank Dr. Ping He for providing the Arabidopsis 1MEKK1 expression construct.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01857/ 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 Long, Song, Yan, Zhou, Zhou and Yang. 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.

# BAKing up to Survive a Battle: Functional Dynamics of BAK1 in Plant Programmed Cell Death

Xiquan Gao1,2 \*, Xinsen Ruan1,2, Yali Sun1,2, Xiue Wang1,2 and Baomin Feng<sup>3</sup> \*

<sup>1</sup> State Key Laboratory of Crop Genetics and Germplasm Enhancement, Nanjing Agricultural University, Nanjing, China, <sup>2</sup> Jiangsu Collaborative Innovation Center for Modern Crop Production, Nanjing Agricultural University, Nanjing, China, <sup>3</sup> Haixia Institute of Science and Technology, Fujian Agricultural and Forestry University, Fuzhou, China

In plants, programmed cell death (PCD) has diverse, essential roles in vegetative and reproductive development, and in the responses to abiotic and biotic stresses. Despite the rapid progress in understanding the occurrence and functions of the diverse forms of PCD in plants, the signaling components and molecular mechanisms underlying the core PCD machinery remain a mystery. The roles of BAK1 (BRASSINOSTEROID INSENSITIVE 1-associated receptor kinase 1), an essential co-receptor of multiple receptor complexes, in the regulation of immunity and development- and defenserelated PCD have been well characterized. However, the ways in which BAK1 functions in mediating PCD need to be further explored. In this review, different forms of PCD in both plants and mammals are discussed. Moreover, we mainly summarize recent advances in elucidating the functions and possible mechanisms of BAK1 in controlling diverse forms of PCD. We also highlight the involvement of post-translational modifications (PTMs) of multiple signaling component proteins in BAK1-mediated PCD.

### Edited by:

Yi Li, Peking University, China

# Reviewed by:

Irene Serrano, Georg-August-Universität Göttingen, Germany Yusuke Saijo, Nara Institute of Science and Technology (NAIST), Japan

#### \*Correspondence:

Xiquan Gao xgao@njau.edu.cn Baomin Feng baomin2006@126.com

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 04 May 2018 Accepted: 10 December 2018 Published: 08 January 2019

#### Citation:

Gao X, Ruan X, Sun Y, Wang X and Feng B (2019) BAKing up to Survive a Battle: Functional Dynamics of BAK1 in Plant Programmed Cell Death. Front. Plant Sci. 9:1913. doi: 10.3389/fpls.2018.01913 Keywords: BAK1, development, co-receptor, programmed cell death, immunity

# INTRODUCTION

Plants have evolved surveillance systems and cellular responses to sustain their growth while protecting themselves against various environmental stresses, often through deploying programmed cell death (PCD) to balance the survival signaling with proper development patterns and abiotic stresses or microbial infections (Van Hautegem et al., 2015; Kabbage et al., 2017). In response to microbial invasions, plants reply on cell surface receptor proteins to detect extracellular molecules produced by pathogens, collectively called pathogen-associated molecular patterns (PAMPs) and activate the PAMP-triggered immunity (PTI) pathway (Zipfel, 2014; Bigeard et al., 2015). Typical PTI responses include callose deposition, ROS production, and the expression of specific marker genes (Jones and Dangl, 2006). The receptor proteins located on plant cell surface that perceive PAMPs are known as pattern-recognition receptors (PRRs), which include receptor kinases (RKs) and receptor-like proteins (RLPs; Song et al., 1995; Gomez-Gomez and Boller, 2000; Chinchilla et al., 2006; Zipfel et al., 2006; Dardick et al., 2012).

In addition to PTI, a second layer of plant defense detects the presence of effector proteins from pathogens by intracellular immune receptor proteins, most of which are nucleotide-binding site leucine-rich repeat (NB-LRR or NLR) proteins and trigger robust immune responses, including ROS production, activation of specific effector-triggered immunity (ETI) marker genes, and rapid

collapse of living tissues named as hypersensitive response (HR), a type of plant-specific PCD. PCD may limit the spread of pathogens; under other stresses, PCD may allow the recycling of nutrients to sustain growth (Jones and Dangl, 2006; Caplan et al., 2008; Eitas and Dangl, 2010; Feng and Zhou, 2012). In plant–microbe interactions, PCD has been recognized as a hallmark of ETI, and also important responses in certain PTI processes (Jones and Dangl, 2006; Caplan et al., 2008; Eitas and Dangl, 2010; Coll et al., 2011; Feng and Zhou, 2012). Other works revealed that PCD is a common and fundamental process that occurs in most eukaryotic organisms (Danon et al., 2000; Ameisen, 2002). PCD is also an intrinsic and indispensable process for plant vegetative and reproductive development. For example, differentiation and maturation of tracheary elements, and abscission of floral organs and tapetum degeneration all involve PCD. PCD was first reported in animals, where it serves as a mechanism to remove unwanted or damaged cells through development-related cell suicide and disease-related cell death (Zakeri et al., 1995; Raff, 1998).

Interestingly, plant PCD associated with development and immunity seems to be often connected to a cell-surface localized receptor kinase named BRI1-associated receptor kinase 1 (BAK1) (Chinchilla et al., 2007; Heese et al., 2007; Chinchilla et al., 2009; Ma et al., 2016). BAK1 was originally discovered as a key component of brassinosteroid (BR) signaling. Research in past decades demonstrated that BAK1 functions as a coreceptor in multifaceted receptor complexes to regulate a variety of processes, including BR-dependent development involving the receptor BRASSINOSTEROID INSENSITIVE 1-(BRI1), and Flagellin-Sensitive 2 (FLS2)-dependent PTI responses (Chinchilla et al., 2009; Ma et al., 2016). Readers who are interested in the topics about the role of BAK1 in innate immunity are strongly suggested to refer to several recent excellent reviews and herein (Ma et al., 2016; Yamada et al., 2016; Yasuda et al., 2017).

There is growing evidence to suggest that BAK1 plays an essential role in regulating various types of PCD (He et al., 2007; Kemmerling et al., 2007; Gao et al., 2009; Schwessinger et al., 2011; Gao et al., 2013b; de Oliveira et al., 2016; Du et al., 2016). The current review will focus primarily on the recent progress made in elucidating the functions of BAK1 in both development- and immunity-associated cell death. The possible mechanisms underlying BAK1-mediated cell death via its cooperation with multiple signaling components and its diverse regulatory mechanisms, including post-translational modification (PTM), will also be discussed.

# PCD IN PLANTS

# Unique Features of Plant Cell Death

Based on the cell morphology, cell death in mammalian cells was classified into three types: apoptosis, autophagy, and necrosis (Zakeri et al., 1995; Kroemer et al., 2009; van Doorn, 2011). These three different types of cell death have different causes and different regulatory mechanisms. Apoptosis is characterized by early collapse and condensation of the nucleus, fragmentation of chromatin, generation of nucleosomal ladders, nuclear blebbing, and cytoplasmic condensation (Zakeri et al., 1995; Yu et al., 2002). The typical biochemical features of apoptosis include DNA cleavage, degradation of the DNA repair enzyme poly (ADP ribose) polymerase, and activation of caspases, a group of cysteine protease (Yu et al., 2002). Autophagy is normally characterized by an increase in the abundance of lytic vesicles termed autophagosomes (Berry and Baehrecke, 2008; Xu et al., 2015), and is regulated by a number of autophagy-related genes (ATGs) (Zhang and Tang, 2005; Patel et al., 2006; Berry and Baehrecke, 2008; Xu et al., 2015). Necrosis is typically accompanied by cell and organelle swelling, rupture of organelles and the plasma membrane, increases in ROS and calcium in cytoplasm, and decreases in cellular ATP (Farber, 1994; Postel et al., 2010; Heller et al., 2018). PCD is also involved in numerous biological processes in plants (Greenberg, 1996; Beers, 1997; Pennell and Lamb, 1997; Lam et al., 1999; Van Hautegem et al., 2015), such as cell differentiation and regulation of cell number, embryogenesis (Bozhkov et al., 2005), abiotic stress responses (Gepstein and Glick, 2013), and plant–pathogen interactions (Beers, 1997; Greenberg, 1997; Coll et al., 2011). Compared to animals, plants are sessile and plant cell surface is coated with much larger number of receptor proteins (∼600 RLK RLPs in Arabidopsis genome) (Tang et al., 2017), therefore, plant is thought to deploy differential, yet complexity strategies of PCD to adapt to diverse and harsh environmental stresses. To better understand the similarities and differences of PCD between plant and animal kingdoms, readers are suggested to refer to several excellent reviews (Kroemer et al., 2009; van Doorn, 2011; Rantong and Gunawardena, 2015; Dickman et al., 2017; Kabbage et al., 2017).

There are several commonalities of PCD features between plants and animals, including the existence of apoptosis-like morphological features and caspase-like proteases, as well as ATG genes in plants (Dickman et al., 2017; Kabbage et al., 2017). The hallmarks of PCD in mammalian systems, including the formation of apoptotic bodies and DNA cleavage (Mittler et al., 1995; Levine et al., 1996; Wang et al., 1996), also exist in plants (Greenberg, 1996). For example, in Nicotiana benthamiana, BECLIN 1, an ortholog of yeast ATG6/VPS30 and mammalian beclin 1 has been identified and characterized as a key regulator in both developmental and HR-type PCD. BECLIN 1 deficient N. benthamiana plants showed accelerated leaf senescence, while the HR-type cell death was suppressed in BECLIN1-silenced plants (Liu et al., 2005). However, many of the features and precise mechanisms of plant PCD are not fully understood.

Several features of PCD are distinct in plant cells, involving different types of caspases and phagocytosis systems compared with those in mammalian cells; the mechanisms involved in plant PCD differ from those in animal PCD, probably due to the genetic and functional redundancy of PCD components as well as plantspecific cellular features such as rigid cell walls, totipotency, and the presence of chloroplasts (Williams and Dickman, 2008).

The nature of apoptosis in plants is controversial. For instance, the viable aleurone cells of mature barley seeds undergo PCD when the seed starts to germinate, and the cells become highly vacuolated; however, aleurone cell death at this stage does

not display the hallmarks of mammalian apoptosis-like PCD, probably due to the absence of major animal apoptosis regulators in plants (Fath et al., 2000; Dickman et al., 2017; Kabbage et al., 2017). Intriguingly, treatments with the Fumonisin B1 mycotoxin or abiotic stresses can trigger the formation of apoptosis-like bodies in plants (Li and Dickman, 2004; Li W. et al., 2010); therefore, caution should be paid when the presence of apoptosis-like cell death in plants is proposed (Dickman et al., 2017). Moreover, during senescence and the differentiation of the tracheary elements, cell death-associated physiological changes often involve vacuole collapse, providing evidence for the essential role of the vacuole in plant PCD (Jones, 2001). Besides vacuole, other organelles, including mitochondria and chloroplasts, have also been suggested to function in plant PCD (Lam et al., 2001).

Numerous environmental factors can trigger PCD, including salt (Li et al., 2007), drought (Duan et al., 2010; Hameed et al., 2013), ozone (Overmyer et al., 2005), and heat (Zuppini et al., 2007; Li Z. et al., 2012). When Arabidopsis roots were subjected to water deficit stress, for example, the typical features of PCD, including increased vacuole size, organelle degradation, and the collapse of tonoplast and plasma membrane, were observed in the apical meristem of the Arabidopsis primary root (Duan et al., 2010). Prolonged salt stress for 24 h caused by treatment with either NaCl or KCl resulted in the significant degradation of organelles in the green algae Micrasterias denticulate (Affenzeller et al., 2009). All together, these studies point to the existence and complexity of different forms of PCD in plants, in response to differential types of stresses. It has to be noted that, however, it remains enigmatic whether a common core machinery is shared for different types of PCD upon perception of either developmental or various abiotic or biotic stress cues (Huysmans et al., 2017).

## Classification of PCD in Plants

The classification of PCD in plants is contradictory, depending on the criteria. It was divided previously into two classes based on morphological features: vacuolar cell death and necrosis (van Doorn et al., 2011) but later updated to autolytic and non-autolytic cell death (van Doorn, 2011). Based on the triggers of PCD in plants, however, several previous studies have suggested that PCD could be classified into development-related PCD (dPCD), environment-related PCD (ePCD), and pathogentriggered PCD (pPCD) (Daneva et al., 2016; Huysmans et al., 2017). dPCD is morphologically characterized by senescence, vacuolar collapse, nuclear degeneration or fragmentation, and cell elimination, which facilitates the successful establishment of reproductive organ identity and structural determination (Daneva et al., 2016). dPCD also occurs during vegetative development in plants, such as xylogenesis, as well as in organ abscission and dehiscence, where it is characterized by tonoplast rupture, vacuolar content release, mitochondrial degradation, and cytoplasmic clearance (Kuriyama, 1999; Yu et al., 2002). ePCD is thought to arise as a response to stress caused by diverse environmental conditions, including abiotic and biotic factors (Wu et al., 2014; Petrov et al., 2015). To better conceptually delineate the mechanisms involved in PCD, we propose that PCD is classified into dPCD, aPCD (abiotic stress-related PCD), and bPCD (biotic stress-related PCD). Currently, there is very limited information for the direct function of BAK1 in controlling aPCD; therefore, we will only focus on the involvement of BAK1 in the regulation of dPCD and bPCD in the following sections.

# BAK1 IS INVOLVED IN THE REGULATION OF DIVERSE FORMS OF PCD

BAK1 belongs to the SERK (somatic embryogenesis-related kinase) family, which are a small group of membrane-localized RLKs that can perceive diverse extracellular ligand stimuli and relay these signals, normally via a phosphorylation cascade (Li, 2010). The first plant SERK identified, DcSERK, was detected in carrot (Daucus carota) hypocotyl cell suspension cultures (Schmidt et al., 1997) during a search for marker genes to enable the monitoring of the transition from somatic cells into embryogenic cells. Most SERKs contain a small extracellular LRR-domain with five repeats, a single transmembrane domain, and a cytoplasmic kinase domain (Li, 2010). The Arabidopsis thaliana genome encodes five SERKs, AtSERK1, AtSERK2, AtSERK3, AtSERK4, and AtSERK5, which arose through gene duplication (Aan den Toorn et al., 2015). Baudino et al. (2001) isolated and identified maize (Zea mays) ZmSERK1 and ZmSERK2 using degenerate primers based on DcSERK and AtSERK1, and ZmSERK3/BAK1 was later characterized for its function in embryogenesis (Zhang et al., 2011). Identification of SERK homologs in sequenced genomes of both higher plants and lower plants, such as moss (Physcomitrella patens), suggests an evolutionarily conserved significance of SERKs (Aan den Toorn et al., 2015).

To date, numerous genetic and biochemical studies have demonstrated that BAK1 functions as a master player at the convergence of multiple physiological processes, including the regulation of development, and responses to biotic stresses (Heese et al., 2007; Chinchilla et al., 2009; Schwessinger et al., 2011; Shen et al., 2011; Meng et al., 2015). For instance, BAK1 participates in BR signaling, vascular differentiation, stem elongation, flowering, floral abscission, fertility, and senescence (Li et al., 2002; Nam and Li, 2002; Postel et al., 2010; Meng et al., 2016). It was also reported to function in PHYTOSULFOKINE alpha (PSK)-regulated root growth (Ladwig et al., 2015), ERECTA (ER) and EPIDERMAL PATTERNING FACTORS (EPFs) dependent cell fate specification in stomatal patterning (Meng et al., 2015).

How can BAK1 as a single RLK participate in so many different signaling pathways? One important reason is that BAK1 could function as a co-receptor or signaling regulator of multiple receptor kinases and RLPs; for example, BAK1 forms complexes with BRI1 to activate BR signaling, and with FLS2 to regulate a PTI pathway (Nam and Li, 2002; Chinchilla et al., 2007; Kemmerling et al., 2007; Ma et al., 2016). Additionally, BAK1 cooperates with multiple immune-related RLKs or RLPs at either the plasma membrane TABLE 1 | Different forms of BAK1-mediated PCD in plants.


or in the cytoplasm, modulating distinct PCD processes (**Table 1**).

# BAK1 in Controlling dPCD

fpls-09-01913 December 27, 2018 Time: 17:39 # 5

Numerous studies have revealed that BAK1 plays crucial roles in regulating dPCD; for example, silencing GhBAK1 in cotton (Gossypium hirsutum) triggers high levels of cell death accompanied by increased ROS production, suggesting that the regulation of cell death by BAK1 is conserved in diverse plant species (Gao et al., 2013b). Interestingly, the BAK1 homolog, SERK5, does not regulate cell death in the Arabidopsis ecotype Col-0, whereas in the ecotype Landsberg erecta it has a regulatory role in cell death (Wu et al., 2015). The serine/threonine protein kinase BOTRYTIS-INDUCED KINASE 1 (BIK1) functions with BAK1; in the bak1bik1 double mutant, a constitutive immune response and spontaneous cell death causes severe growth defects and a dwarf phenotype, accompanied with enhanced expression levels of immune genes, including PR1, PR5, PAD4, WRKY45, and ERF1 (Liu et al., 2017). Additionally, BAK1 interacts with BIR1 (BAK1-interacting receptor-like kinase 1), and the bir1 mutant displays a constitutive cell death phenotype (Gao et al., 2009). Both BAK1 and BIR1 interact in vitro and in vivo with BONZAI1 (BON1), a calcium-dependent phospholipid-binding protein; bon1 mutants genetically interacted with bir1 to produce temperature-dependent growth defects and cell death in Arabidopsis (Wang et al., 2011).

It has been noted that BAK1, together with other SERK family members, could function as a co-receptor of PXY (phloem intercalated with xylem) (Zhang et al., 2016). PXY is a LRR-RLK receptor of tracheary element differentiation inhibitory factor (TDIF), also known as CLAVATA3/EMBRYO SURROUNDING REGION-RELATED (CLE), which regulates vascular development in Arabidopsis, revealing the involvement of these essential components in dPCD (Ma et al., 2016; Zhang et al., 2016). It has been demonstrated that tracheary elements (TEs) typically undergo an autophagic type of PCD during differentiation in Zinnia elegans and Arabidopsis (Fukuda, 1997; Fukuda, 2000; Turner et al., 2007; Williams and Dickman, 2008). This type of PCD is characterized by a clearing process for the removal of dead protoplasts, normally achieved by multiple proteases, including xylem cysteine proteases 1 and 2 (XCP1 and XCP2), bifunctional nuclease 1/endonuclease 1 (BFN1/ENDO1) and metacaspase 9 (MC9) (Avci et al., 2008; Bollhoner et al., 2013; Xu et al., 2018). Some transcriptional factors, such as VASCULAR-RELATED NAC-DOMAIN6/7 (VND6/7), appear to also be responsible for the PCD process activated by TDIF signaling (Ito and Fukuda, 2002; Pyo et al., 2007; Zhong et al., 2010; Heo et al., 2017). However, evidence supporting that BAK1 directly regulates dPCD is still missing.

Intriguingly, BR signaling was also found to be involved in xylem differentiation, as evidenced by the findings that treatment using the BR signaling inhibitor brassinazole, and the BR biosynthesis inhibitor uniconazole, resulted in aberrant vascular patterning and PCD (Yamamoto et al., 1997; Asami et al., 2000), while the BR deficient mutant, cpd (photomorphogenic dwarf) showed defective xylem biogenesis (Szekeres et al., 1996), and bri1 single mutants and bri1 brl1 brl3 triple mutants all displayed severe vascular defects (Cano-Delgado et al., 2004). Furthermore, upon perception of TDIF ligands CLV3 and CLE41, the TDIF receptor (TDR) interacts with BIN2 (Brassinosteroid-Insensitive 2), a member of GSK3 (Glycogen Synthase Kinase 3), to suppress procambial cell differentiation into xylem, which also involves the suppression of BES1 (BRI1-EMS Suppressor 1) downstream of TDR-GSK3 (Kondo et al., 2014; Heo et al., 2017). Given that BAK1 forms a signaling complex with BRI1 and PXY, respectively, it is possible that BAK1 acts as a convergent component shared by the PXY/TDR and BRI1 signaling pathways in dPCD regulation, yet this hypothesis remains to be further investigated and approved.

The BAK1 homologs SERK1 and SERK2 might also play a role in regulating dPCD in anther, in which they interact with the receptor-like kinase EMS1 to perceive the signal of a peptide ligand, TPD1, and control the differentiation of tapetum (Li Z. et al., 2017). Degeneration of tapetum through PCD and the release of its content to nurture maturation of pollen is essential for the success of male reproductive development. In serk1, serk2, or ems1 mutants, tapetum differentiation and PCD were not properly initiated (Li Z. et al., 2017). This is a scenario similar to that in tracheary element differentiation. Again, a direct link between SERK1/2 and tapetal PCD needs solid experimental support. In the abscission zone, an IDA-HAS/HSL2 signaling pathway also relies on SERK members (SERK1/2/3) to transduce signals for abscission, where PCD is essential (Meng et al., 2016). Whether it is a common theme that BAK1 and other SERKs function upstream of certain dPCDs awaits future studies.

# BAK1 in Controlling bPCD

## Functional Perturbation of BAK1 and Its Partners Triggers bPCD

BAK1 functions as a co-receptor of multiple RLKs and is involved in diverse signaling pathways. BAK1's function seems essential and is under tight surveillance so that PCD is triggered once its function is perturbed. Previously, a genetic investigation on the single mutation of BAK1 itself and double mutation of BAK1 with its closest homolog BKK1 (SERK4) revealed that while bak1 showed strong premature senescence (Kemmerling et al., 2007; Jeong et al., 2010), cell death in bak1 bkk1 double mutants occurs post-embryogenesis, suggesting that BAK1 and BKK1 function redundantly to negatively control cell death (He et al., 2007; de Oliveira et al., 2016). bak1 single mutants developed a type of uncontained PCDs upon infection with virulent necrotrophic pathogens, which differs from both necrotizing elicitor- and SAinducible PCD (Kemmerling et al., 2007). Similarly, silencing BAK1 in N. benthamiana also leads to enhanced PCD upon infection with Hyaloperonospora parasitica (Heese et al., 2007).

Interestingly, over-expression of BAK1 or its ectodomain also elicited spontaneous PCD, accumulation of SA and expression of multiple PCD-related genes, including BON1, BIRs, and SOBIR1 (Kim et al., 2017). In line with this finding, constitutive expression of BAK1 or its ectodomain or excess of BAK1 could trigger strong dwarfism and premature death phenotype, as well as autoimmunity without microbe attacks (Domnguez-Ferreras et al., 2015). Therefore, the abundance of BAK1 seems important

and needs to be kept in check. It is hypothesized that overexpression of BAK1 might sequestrate BIR1 to trigger PCD (Ma et al., 2016). At present, it remains unclear whether bak1/bkk1 cell death is due to the loss of negative regulation of PCD by BAK1 or is caused by an unknown mechanism that monitors developmental defects in bak1.

Multiple signaling components distinct from the BRI1 pathway are also engaged by BAK1 to trigger PCD upon pathogen infection. For instance, a BAK1-interacting RLK, BIR1, was identified by a reverse genetics approach; the bir1-1 mutant displayed extensive cell death and constitutive immunity (Gao et al., 2009). Intriguingly, further suppressor screening using bir1-1 led to the identification of suppressor of bir1-1 (sobir1- 1), which strongly suppressed the cell death observed in bir1-1 (Gao et al., 2009). Moreover, over-expression of SOBIR1 results in elevated cell death, indicating that SOBIR1 functions as a positive regulator of cell death (Gao et al., 2009; Liebrand et al., 2014). Using Co-IP coupled with liquid chromatographyelectrospray ionization-tandem mass spectrometry (LC/ESI-MS/MS), two close BIR1 homologs, BIR2 and BIR3, were identified and demonstrated to constitutively interact with, and be phosphorylated by, BAK1, which in turn prevents the formation of the BAK1–FLS2 receptor kinase complex, thus negatively regulating PTI signaling (Halter et al., 2014; Imkampe et al., 2017). Examination of an allelic series of bak1 mutation showed that BIR1 and BAK1 interact genetically to regulate BR signaling, cell death and immune response (Wierzba and Tax, 2016).

bir1 has enhanced SA-dependent PCD (Liu et al., 2016). Moreover, upon infection with a necrotrophic pathogen, Alternaria brassicicola, bir2 mutants had enhanced cell death and susceptibility to this pathogen, whereas BIR2 interacts with BAK1 and suppresses the autoimmune cell death response in the absence of PAMPs (Halter et al., 2014). Furthermore, it was reported that while BIR3 interferes with BRI1-dependent growth by interacting with and stabilizing BAK1, it also negatively affects the formation of the BAK1–FLS2 complex to suppress cell death and immunity, exemplified by the enhanced spontaneous cell death in the bak1 bir3 mutant (Imkampe et al., 2017).

### BAK1 Is Involved in Certain PAMP-Triggered PCD

As a co-receptor of multiple RLKs that perceive PAMPs, BAK1 is required for certain PAMP-triggered PCD; for example, in N. benthamiana, triggering of ROS accumulation and HR by the INF1 protein secreted from the oomycete pathogen Phytophthora infestans was prevented by a mutation in BAK1 (Heese et al., 2007; Chaparro-Garcia et al., 2011). A recent study showed that silencing of SOBIR1 in N. benthamiana attenuated INF1 triggered cell death and resistance to P. infestans, while SOBIR1 was found to form a receptor complex with ELR (Elicitin Response) protein isolated from Solanum microdontum, which is a RLP perceiving INF1 (Domazakis et al., 2018). Moreover, BAK1 is recruited to the ELR/SOBIR1 signaling complex to activate downstream defense response, suggesting that both SOBIR1 and BAK1 are required for INF1-regulated PCD and immunity. Similarly, BcXYG1, a xyloglucanase protein secreted from Botrytis cinerea, interacts with BAK1 and SOBIR1 to trigger cell death and the immune response (Zhu et al., 2017). A small apoplast-targeted cysteine-rich protein, PstSCR1, secreted from the wheat rust pathogen Puccinia striiformis f. sp. tritici, triggers PCD and immunity in N. benthamiana via a pathway that appears to be dependent on the BAK1 pathway (Dagvadorj et al., 2017).

Moreover, BAK1 cooperates with SOBIR1 and RLP30 (Receptor-Like Protein 30) to control PCD caused by necrotrophic pathogens. RLP30 is responsible for the sensitivity to SCLEROTINIA CULTURE FILTRATE ELICITOR1 (SCFE1) containing fraction, which contains a proteinaceous elicitor, produced by S. sclerotiorum, and rlp30 mutants showed increased cell death and susceptibility to infection with necrotrophic pathogens S. sclerotiorum and B. cinerea (Zhang et al., 2013). In addition, BAK1 was found to interact with one of the LRR-RLP receptors in N. benthamiana, LeEix1, to attenuate the LeEix2 mediated Eix (Ethylene-inducing xylanase) response in tobacco (Nicotiana tabacum) and tomato (Solanum lycopersicum) in order to trigger the typical HR response (Bar et al., 2010).

Using proteomic approaches several RLKs, cysteine-rich receptor-like kinases (CRKs) enriched at the plasma membrane, were identified while the expression levels of these CRKs were activated upon the ligand elicitation of flagellin in Arabidopsis (Yadeta et al., 2017). Among those CRKs, the induction of CRK28 activity was highly correlated with enhanced resistance to the wheat rust pathogen P. striiformis f. sp. tritici and increased ROS production; moreover, the kinase active site of CRK28 (K377) is required for triggering cell death. CRK28 associates with the FLS2/BAK1 immune complex in a flg22-dependent manner, and CRK28-induced cell death was abolished in NbSerk3-silenced N. benthamiana plants, suggesting that BAK1 is required for CRK28-mediated cell death (Yadeta et al., 2017).

However, in some cases, loss of BAK1 function enhances the cell death triggered by PAMPs or DAMPs. A recent study showed that when PTI signaling is compromised by BAK1 disruption, danger peptide receptor PEPRs (Pep Receptors) signaling could be activated to ensure basal resistance. Depletion of BAK1 sensitized PEPRs signaling toward cell death upon ligand elicitation, as the ligand Pep2 was found to induce extensive cell death in bak1-4 mutants, which is dependent on PEPRs (Yamada et al., 2016). It is believed that such an enhanced PCD phenotype in bak1 mutant upon pathogen infection is caused by the subsequent dysfunction of BAK1/BON1 suppressed cell death, which in turn activates the PEPR signaling pathway to reversely trigger cell death and to retain immunity to biotrophic pathogens (Yamada et al., 2016).

## BAK1 and Effector Triggered PCD

Lines of evidence suggest that PCD triggered by perturbation of BAK1 functions shows similarity to R-protein-mediated PCD. SALICYLIC ACID INDUCTION-DEFICIENT (SID2) and ENHANCED DISEASE SUSCEPTIBILITY5 (EDS5), two chloroplast-localized components of the salicylic acid (SA) mediated ETI pathway, were also proposed to contribute to cell death by BAK1/BKK1 mutations, thereby regulating PCD; the sid2 and eds5 mutations suppress cell death in bak1-3bkk1- 1 mutants, and this cell death is dependent on light and

SA (Gao et al., 2017). In line with this finding, the overexpression of BAK1 resulted in the accumulation of SA and hydrogen peroxide, as well as the enhanced expression of BON1, BIRs, and SOBIR, the processes strongly associated with spontaneous cell death (Kim et al., 2017). BON1, functioning as a negative regulator of R-mediated resistance, interacts with both BAK1 and BIR1 to interfere with the immune response and PCD (Wang et al., 2011). Furthermore, enhanced cell death in bir1 plants was found to be partially dependent on PHYTOALEXIN DEFICIENT4 (PAD4) and EDS1, which is required for TIR NLR signaling. This suggests that BIR1 might be guarded by plant resistance (R) protein signaling (Gao et al., 2009).

BAK1 seems also to participate in other PCDs during ETI. Two effectors from the tomato pathogen Cladosporium fulvum, Avr4 and Avr9, are recognized by the R proteins Cf-4 and Cf-9, respectively, which in turn recruit BAK1 and subsequently trigger the HR and immunity against C. fulvum (Postma et al., 2016). The BAK1/SOBIR1-dependent pathway was also shown to mediate the interaction between the tomato resistance gene I, a LRR-RLP involved in the immunity to Fusarium oxysporum f. sp. lycopersici (Fol), and the FolAvr1 effector, triggering necrosis in N. benthamiana (Catanzariti et al., 2017). Moreover, BAK1 positively regulates the NLR protein Mi-1 in resistance and cell death upon potato aphid infection in the tomato (Peng and Kaloshian, 2014; Peng et al., 2016).

# POSSIBLE MECHANISMS UNDERLYING CONTRADICTORY FUNCTION OF BAK1 IN PCD

As reviewed above, BAK1 is involved in different forms of PCDs. However, it is intriguing that BAK1 could serve as both positive and negative regulators of PCDs. The question about how BAK1, as a single RLK, is capable of controlling different PCD processes oppositely remains open to answer. Here, we would like to summarize the possible molecular mechanisms that might explain the complicated function of BAK1 in PCD.

It is possible that the output specificity of BAK1 functions in different PCD processes is determined by the ligand specificity of corresponding RLKs. Therefore, the effects of disrupted BAK1 functioning are likely dependent on the role of BAK1 in that specific complex. BAK1 also interacts with other RLKs including BIR1, BIR2,and SOBIR1. The balance between different BAK1-incorporated complexes is obviously influenced by the abundance of BAK1. These interacting RLKs might keep each other under tight control to ensure proper activation of PCDs. It has been shown that both BIR1 and BIR2 appear to be essential in structurally keeping BAK1-regulated PCD under tight control in the case of no ligand binding, thus interfering the unwanted interaction between BAK1 and different PRRs (Ma et al., 2017). Moreover, a specific ligand stimulates the release of BIR2-sequestered BAK1, which subsequently enhances the interaction complex formation between BAK1 and PRR, FLS2, EFR, BRI1, PEPR1, etc., (Halter et al., 2014). Interestingly, BIR1, BIR3, and BIR4 all formed a stable heterodimeric complex with BAK1 at pH 6.0 through their ecto-domains, and flg22 bound FLS2 outcompeted BIR1LRR for binding to BAK1LRR (Ma et al., 2017). On the other hand, upon pathogen infection, bak1bir3 showed increased pathogen-inducible PCD (Imkampe et al., 2017); upon ligand perception, overexpression of BIR2 suppressed the BAK1/FLS2 (PRR) complex formation (Halter et al., 2014). Moreover, BAK1 overexpression results in runaway cell death, and simultaneous overexpression of BRI1 and BAK1 suppresses this cell death, suggesting that activation of the BAK1– PRR signaling complex upon ligand binding is essential to suppress this type of auto-immune PCD (Belkhadir et al., 2012; Halter et al., 2014).

The downstream signaling events upon activation of PRR– BAK1 complexes also show specificity to ligands. A very recent study suggests that the specificity might be determined by the phosphosite code in BAK1. They identified multiple BAK1 phosphosites specific to the signaling processes, e.g., immunity or growth, but not others (Perraki et al., 2018). In this case, the mutation of multiple key phosphorylation sites in BAK1, including S602D/T603D/S604D or S612D, resulted in impaired PTI responses, including flg22-induced MAPK signaling and immunity to Pst DC3000, but dispensable for BR-signaling. Moreover, the mutation of Y403 in the BAK1 C-terminal attenuated the phosphorylation and BAK/EFR signaling complex formation upon elf18 elicitation (Perraki et al., 2018). Other studies also support that the intracellular domains have separable functions in mediating different signaling processes. In a bir1 suppressor screen using bir1-1pad4-1 mutant, sobir7-1, a bak1 allele with a nonsense mutation within the carboxyl-terminal tail (CT) of BAK1, was identified (Liu et al., 2016; Wu et al., 2018). A series of genetic evidence proved that the CT domain of BAK1 was essential for its kinase activity to trigger PTI response, but dispensable for controlling cell death and BR signaling (Liu et al., 2016; Wu et al., 2018). It is worth investigating how the differential regulation of PCD involving BAK1 might also attribute to the diverse phosphosite codes activated.

As discussed above, the PCD triggered by disruption of BAK1 might be mediated by certain R proteins, which is a well-accepted theme that host protein is guarded by cognate R proteins. If this is true, the cell death caused by BAK1/BKK1 loss-of-function does not imply they are negative regulators of PCD, but instead that they are such important signaling components that their disruption is monitored by R-protein. Some cases whereby R proteins guard important host signaling components have been reported. For example, RIN4 is guarded by two NLRs, RPM1 and RPS2, and disruption of the MAP kinase cascade, MEKK1/MKK5/MPK4, will trigger the activation of SUMM2 R-protein and cell death (Zhang et al., 2012). Moreover, PAD4-dependent immunity is activated in bir1-1, raising the possibility that BIR1 is guarded by R gene, too (Liu et al., 2016). Indeed, it has been proposed that BAK1/BIR1 is probably guarded by two or more R proteins in the absence of pathogens. Mutation of BAK1 or BIR1 results in the activation of those guard R proteins, which subsequently trigger different signaling pathways, e.g., disease resistance and/or PCD, mediated by PAD4 and SOBIR1, respectively, (Gao et al., 2009). In

Gao et al. The Roles of BAK1 in PCD

line with this finding, SRF3, an LRR-RLK structurally similar to BIR2, has been reported to regulate hybrid incompatibility along with the R gene RPP1, during which necrotic PCD, enhanced SA levels and immune response were found. This illustrates a model that SRF3 is guarded by RPP1 to control incompatibility in the absence of pathogens (Alcazar et al., 2010).

Nevertheless, as a center component involved in various signaling pathways, it appears that an optimal amount of BAK1 should be strictly maintained to optimize the fitness of growth/development and disease resistance/PCD. While being tightly guarded by and released from R proteins, and responding to appropriate ligands to interact selectively with different PRRs, BAK1 has evolved multiple strategies to cooperate with diverse signaling components to fine-tune its function in regulating different types of PCD.

# REGULATION OF BAK1-MEDIATED PCD

Several studies have suggested that the perception of PAMPs by the PRRs and the subsequent activation of downstream signaling are associated with multiple regulation events, especially PTMs (such as phosphorylation, ubiquitination, glycosylation), and protein endocytosis (Lu et al., 2011; Kadota et al., 2014; Lin et al., 2014). The importance of some of these processes has also been demonstrated to be associated with the functional dynamics of BAK1 mediated PCD (Bender et al., 2015; de Oliveira et al., 2016; **Table 1**).

As discussed above, BAK1 signaling specificity might be determined by the code of phosphosites (Perraki et al., 2018). The phosphorylation of non-RD plasma membrane-localized LRR-RKs, including FLS2 and EFR, by BAK1 is essential for inducing PTI upon the perception of PAMPs, which is different from that of RD-kinase BRI1 phosphorylation by BAK1 (Schulze et al., 2010; Schwessinger et al., 2011). In this case, the bak1-5 mutant, a novel mutant BAK1 allele carrying a single amino acid substitution, C408Y, in the BAK1 cytoplasmic kinase domain, is impaired in PTI signaling but not in cell death regulation. Upon ligand elicitation, however, BAK1-5 kinase activity is required for the formation of BAK1/FLS2 or BAK1/EFR PRR complexes. Moreover, the bak1-5 line becomes insensitive to SCFE1, and did not show enhanced PCD. This was different from bak1- 3 and bak1-4 mutants, which are susceptible to B. cinerea and A. brassicicola (Kemmerling et al., 2007). These findings strongly support the mechanistically uncoupled and phosphorylationdependent activation function for BAK1 in regulating distinct signaling pathways, e.g., BR-associated development, PCD and PTI responses (Schwessinger et al., 2011; Monaghan and Zipfel, 2012).

BAK1 is also capable of phosphorylating BIK1 at tyrosine and serine/threonine sites, as evidenced by the requirement for kinase activity and the presence of three tyrosine residues (Y150, Y243, and Y250) in BIK1 for its function in immunity (Lin et al., 2014). Furthermore, upon PAMP elicitation, BIK1 could directly interact with and phosphorylate RBOHD to control the ROS burst and promote resistance to bacterial pathogens (Kadota et al., 2014).

In addition to phosphorylation, other PTM processes participate in the regulation of BAK1 activity. For instance, the PAMP flg22 triggers the recruitment of a pair of plant U-box E3 ligases, PUB12 and PUB13, to FLS2, which is subsequently degraded by ubiquitination. Interestingly, FLS2 ubiquitination by PUB12/13 requires the BAK1-mediated phosphorylation of FLS2 (Lu et al., 2011). Since it has been shown that disruption of PUB13 caused a spontaneous cell death phenotype, which was also enhanced under high humidity conditions (Li W. et al., 2012), it is reasonable to speculate that an ubiquitination event might also be involved in BAK1-mediated PCD.

Using the BAK1 cytoplasmic domain as bait to screen a yeast two-hybrid library, a glutaredoxin (GRX) C2 (AtGRXC2) protein was characterized to be a BAK1-interacting component. AtGRXC2 can S-glutathionylate and form a heterodimer with the BAK1 cytoplasmic domain in vitro in the presence of either glutathione disulfide or glutathione plus H2O2, thus inhibiting BAK1 kinase activity (Bender et al., 2015). BAK1 kinase activity was enhanced by the mutation of an AtGRXC2-targeted essential glutathionylation site, Cys408 to tyrosine, in a bak1-5 background (Bender et al., 2015). Although it remains to be determined whether BAK1 glutathionylation is directly associated with its role in the regulation of PCD and immunity, this finding reveals a novel regulatory mechanism of BAK1 signaling by redox status and glutathionylation.

The process of endoplasmic reticulum (ER)-mediated protein quality control (ERQC) and glycosylation were shown to be important for BAK1-mediated PCD as well. To identify the suppressor of cell death that silences BAK1/SERK4 (BKK1) and BIR1, a virus-induced gene silencing (VIGS)-based genetic screen was carried out in the bak1-4serk4-1 and bir1 mutant backgrounds, leading to the identification of a mutant with a defective STAUROSPORIN AND TEMPERATURE SENSITIVE3 (STT3a) protein (de Oliveira et al., 2016). The stt3a-2 mutation significantly suppresses cell death, H2O<sup>2</sup> accumulation, and PR1 and PR2 expression in the bak1-4serk4-1 and bir1 backgrounds, providing strong genetic evidence for the positive role of STT3a in triggering PCD and the immune response (de Oliveira et al., 2016). Interestingly, several specific ERQC components, such as ERdj3b and SDF2, seem to be involved in triggering cell death in the bak1-4/serk4-1 mutants (Sun et al., 2014). On the other hand, using an RNA-seq analysis, one of the most highly activated gene families in bak1-4/serk4-1 was found to be the CRKs, including CRK4 and CRK5, which strongly elicit cell death when transiently overexpressed in N. benthamiana (de Oliveira et al., 2016). Moreover, a biochemical analysis showed that CRK4 and CRK5 are likely the targets of glycosylation, as revealed by an obvious migration shift under electrophoresis. Given the known function of STT3 as a catalytic subunit of oligosaccharyltransferase in protein N-glycosylation, these findings suggest that STT3amediated N-glycosylation and ERQC are essential for CRK4 mediated PCD. It remains to be determined whether CRK4 and CRK5 are necessary and sufficient for the cell death in bak1- 4 serk4-1 mutant plants (de Oliveira et al., 2016). This result is further supported by the finding that CRK28 is also a glycosylated

transmembrane protein found in a PRR–RLK complex (Yadeta et al., 2017). Intriguingly, it is noted that a subset of LRR-RK-type PRRs, including EFR and Xa21, specifically require this ERQC pathway in their proper folding and maturation (Saijo, 2010; Beck et al., 2012), implying the possible recruitment of a similar LRR-RK to mediate cell death in the absence of BAK1.

In addition to the aforementioned ERQC pathway and glycosylation, nucleocytoplasmic trafficking is also essential to BAK1-mediated PCD. sbb1-1, another suppressor of cell death in bak1-4 serk4-1, was identified in a genetic screen (Du et al., 2016). SBB1 encodes a nucleoporin (NUP) 85-like protein that is a member of the NUP107-160 sub-complex, the largest subcomplex known to be highly conserved in vertebrates and plants. Knocking out individual NUP members including SBB1 (NUP85), SEH1, NUP160, or NUP96 fully suppresses the cell death phenotype of the bak1-4 and serk4-1 mutants. The sbb1 mutation reduced endogenous SA levels and the sbb1 mutant suppressed cell death in bak1-4 and serk4-1, and expression of SBB1 driven by its own promoter in bak1-3 bkk1-1 sbb1-2 can recapitulate cell death phenotype, suggesting that SBB1 mediated cell death in bak1-4serk4-1 is SA-dependent (Du et al., 2016). Interestingly, co-immunoprecipitation coupled with LC-MS/MS analyses identified numerous SBB1-interacting proteins, including DEAD-box RNA helicase 1 (DRH1), which was found to directly associate with SBB1. Genetic data demonstrated that SBB1-DRH1 is required for cell death in bak1-4 and serk4-1. Consistent with the observation that DRH1 is localized at the nucleus and that SBB1 functions in mRNA export, the SBB1–DRH1 complex-mediated nucleocytoplasmic trafficking process likely contributes to BAK1/SERK4-controlled cell death, which might be exerted through its interference in the export of SA-related mRNAs (Du et al., 2016).

Endocytosis is also involved in BAK1-mediated PCD. For example, interaction of BAK1 with LeEix1 results in the endocytosis of LeEix2, whereas LeEix1 interferes with the LeEix2 triggered immune response and HR, which was impaired in the BAK1-silenced plants (Bar et al., 2010). This finding reveals the key role of BAK1 in regulating Eix-induced PCD and the PTM of the LRR-RLP receptor upon pathogen infection.

# CONCLUSION AND PERSPECTIVES

Cell death is an essential process for both mammals and plants. Despite the remarkable progress made in the elucidation of the occurrence and features of PCD, more research is needed to determine how host plants perceive and transduce external signals to activate PCD. Similar to animals, PCD is deployed by plants to facilitate cell differentiation during development or to promote survival by enabling plants to adapt to environmental stresses and defend against pathogens. Accumulating evidence points to the central role of BAK1 as a co-receptor or signaling regulator of multiple receptor kinases and RLPs in different types of PCD. BAK1 likely exerts its function via deploying its specific phosphorylation sites to phosphorylate PCD-related RLKs or RLPs (Perraki et al., 2018), differentiating responses to the elicitation by various ligand bindings, and to distinctly modulate PCD.

BAK1 has been intensively investigated for its function as the key component of the BR-mediated signaling pathway and the RLK-mediated PTI signaling pathway; thus, it is reasonable to theorize that dPCD (meristem cell death) and bPCD converge at BAK1. It has been proposed that calcium signaling (for example, via the CDPKs) (Boursiac et al., 2010; Gao et al., 2013a) and ROS production are involved in the regulation of both dPCD and bPCD (Gechev and Hille, 2005; Boursiac et al., 2010; Petrov et al., 2015; Serrano et al., 2015). Given the pivotal role of BAK1 as a co-receptor of the PTI signaling complex in triggering PCD, it is tempting to speculate that BAK1 may function upstream of ROS and/or calcium signaling to regulate the diverse types of PCD, probably through the activation of a MAPK signaling cascade (He et al., 2007; Kemmerling et al., 2007; Jeworutzki et al., 2010; Gao et al., 2013b).

PCD is considered to be a hallmark of the ETI response. Upon ETI activation, PCD can also be triggered via CDPK-mediated signaling, likely through the phosphorylation of specific WRKY transcription factors (Gao et al., 2013a); however, it is unclear whether BAK1 exerts its negative role in controlling PCD as either a shared core regulator of ETI-triggered PCD or through distinct mechanisms. There is accumulating evidence to suggest that BAK1 could be targeted by multiple effectors (Macho and Zipfel, 2014); for instance, by AvrPtoB (Shan et al., 2008) and HopF2 (Zhou et al., 2014) from Pseudomonas syringae, and Avr3a from P. infestans (Chaparro-Garcia et al., 2011), resulting in disruption of PTI signaling as well as that of PCD suppressed by BAK1. Furthermore, SOBIR1 associates with or interacts with several R proteins (Qi et al., 2011; Ma and Borhan, 2015); thus, it is possible that activation of ETI-associated PCD might be attributed, at least to some extent, to the negative regulation by BAK1 and BAK1/SOBIR1 receptor complexes. One should be cautious, however, that BAK1 targeting by different effectors leading to enhanced cell death may not be pertinent to all forms of BAK1. This is because BAK1-5 is a hypoactive kinase and the bak1-5 mutant is not impaired in cell death control (Schwessinger et al., 2011), whereas the kinase domain of BAK1 seems to be essential for AvrPtoB targeting structurally (Cheng et al., 2011).

PCD functions at essential steps in development and aging as well as in abiotic and biotic stresses. It seems that plants may deploy the BAK1 signaling complex to coordinate different types of PCD and thus control the trade-off between development and immunity, possibly via subverting hormone signaling, interacting with R proteins, and integrating distinct PTM processes. Future research should center on exploring how host plants control PCD by orchestrating BAK1 homeostasis, which may also provide practical implications for crop improvement. Moreover, despite the fact that animal apoptosis-like cell death has not been fully addressed in plants, recent evidence has suggested the presence and functional importance of animal-type apoptosis in plant PCD (Dickman et al., 2017). Another research direction will therefore involve clarification of the relationship between BAK1 regulated PCD and apoptosis-like cell death, as well as other cell death processes, such as autophagy, under different stresses or environmental stimuli.

# AUTHOR CONTRIBUTIONS

fpls-09-01913 December 27, 2018 Time: 17:39 # 10

XG wrote the manuscript. XG and BF revised the manuscript with input from XR, YS and XW.

# FUNDING

We are grateful for financial support in the form of grants from the National Key Research and Development Program of China (No. 2016YFD0101002), the NSFC (Nos. 31471508 and 31671702), the Technology Foundation for Selected

## REFERENCES


Overseas Chinese Scholar, Ministry of Personnel of China (No. G0101500090), and from the Jiangsu Collaborative Innovation Center for Modern Crop Production (JCIC-MCP) and the Innovation Team Program for Jiangsu Universities (2014).

# ACKNOWLEDGMENTS

We apologize to the many authors whose significant works are not cited here due to space limitations. We acknowledge Dr. Michael V. Kolomiets and Dr. Dongping Lu for their critical review of this manuscript and their insightful discussion.



immune receptors in Arabidopsis. Mol. Plant Pathol. 12, 702–708. doi: 10.1111/ j.1364-3703.2010.00704.x


and plant immune responses. Plant Physiol. 175, 438–456. doi: 10.1104/pp.17. 00375


**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 © 2019 Gao, Ruan, Sun, Wang and Feng. 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 Novel G16B09-Like Effector From Heterodera avenae Suppresses Plant Defenses and Promotes Parasitism

Shanshan Yang, Yiran Dai, Yongpan Chen, Jun Yang, Dan Yang, Qian Liu\* and Heng Jian

Department of Plant Pathology and MOA Key Laboratory of Pest Monitoring and Green Management, China Agricultural University, Beijing, China

Plant parasitic nematodes secrete effectors into host plant tissues to facilitate parasitism. In this study, we identified a G16B09-like effector protein family from the transcriptome of Heterodera avenae, and then verified that most of the members could suppress programmed cell death triggered by BAX in Nicotiana benthamiana. Ha18764, the most homologous to G16B09, was further characterized for its function. Our experimental evidence suggested that Ha18764 was specifically expressed in the dorsal gland and was dramatically upregulated in the J4 stage of nematode development. A Magnaporthe oryzae secretion system in barley showed that the signal peptide of Ha18764 had secretion activity to deliver mCherry into plant cells. Arabidopsis thaliana overexpressing Ha18764 or Hs18764 was more susceptible to Heterodera schachtii. In contrast, BSMV-based host-induced gene silencing (HIGS) targeting Ha18764 attenuated H. avenae parasitism and its reproduction in wheat plants. Transient expression of Ha18764 suppressed PsojNIP, Avr3a/R3a, RBP-1/Gpa2, and MAPK kinases (MKK1 and NPK1Nt)-related cell death in Nicotiana benthamiana. Co-expression assays indicated that Ha18764 also suppressed cell death triggered by four H. avenae putative cell-death-inducing effectors. Moreover, Ha18764 was also shown strong PTI suppression such as reducing the expression of plant defenserelated genes, the burst of reactive oxygen species, and the deposition of cell wall callose. Together, our results indicate that Ha18764 promotes parasitism, probably by suppressing plant PTI and ETI signaling in the parasitic stages of H. avenae.

### Edited by:

Zhengqing Fu, University of South Carolina, United States

### Reviewed by:

Hai-Lei Wei, Chinese Academy of Agricultural Sciences, China Justin Lee, Leibniz-Institut für Pflanzenbiochemie (IPB), Germany

> \*Correspondence: Qian Liu liuqian@cau.edu.cn

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Plant Science

Received: 13 September 2018 Accepted: 16 January 2019 Published: 08 February 2019

#### Citation:

Yang S, Dai Y, Chen Y, Yang J, Yang D, Liu Q and Jian H (2019) A Novel G16B09-Like Effector From Heterodera avenae Suppresses Plant Defenses and Promotes Parasitism. Front. Plant Sci. 10:66. doi: 10.3389/fpls.2019.00066 Keywords: Heterodera avenae, effector, G16B09 family, suppressed plant defense, PTI, ETI

# INTRODUCTION

Heterodera avenae is an important cereal cyst nematode (CCN) that infects wheat, barley, and oat crops in the cereal-growing regions worldwide. The wheat yield losses caused by H. avenae can range widely, from 10% up to 100% in some infected fields (Bonfil et al., 2004; Smiley et al., 2017). The infective second-stage juvenile (J2) of this cyst nematode penetrates the root tips of its host plant and migrates intracellularly toward the vascular cylinder, then it inserts its stylet into a cell and induces the infusion of surrounding cells, resulting in the formation of a multinucleate syncytium (Jones, 1981). Because cyst nematodes are obligate sedentary endoparasites that feed from the syncytia until their reproduction is complete, they closely interact with their host plants. The same with other plant pathogens, nematodes can secrete effector proteins to regulate host

plant cellular processes to promote their parasitism. Most of these effectors are produced in the esophageal glands and are delivered into plant cells via the nematode's hollow stylet.

Plants have developed the immune system to protect them from pathogen attacks. A comprehensive overview of the multifaceted co-evolutionary plant–pathogen interactions is conveyed in the "zigzag" model (Jones and Dangl, 2006). In it, plants respond to pathogen infection by using a two-branched immune system. The first branch recognizes microbe/pathogenassociated molecular patterns (MAMPs/PAMPs) to trigger MAMP/PAMP-triggered immunity (MTI/PTI) responses, such as callose deposition, the burst of reactive oxygen species (ROS), and the induction of defense-related gene expression (Luna et al., 2011; Mendoza, 2011). To enhance their survival, infecting pathogens deliver effectors that interfere with PTI for successful parasitism. The second plant immune branch recognizes one effector by a resistance protein and activates an effector-triggered immunity (ETI) response, usually resulting in hypersensitive cell death at the infection site (Cui et al., 2015). Pathogen isolates might gain new effectors to suppress ETI, and this relationship illustrates the dynamic co-evolution between plants and their pathogens (Tsuda and Katagiri, 2010).

Nematode effectors play a wide variety of roles in root penetration, suppression of host defenses, and the formation and maintenance of feeding sites (Haegeman et al., 2012). In recent years, those effectors capable of suppressing plant immunity have garnered increasing attention (Chronis et al., 2013; Jaouannet et al., 2013; Diaz-Granados et al., 2016). Recently, a number of effectors were found to be capable of suppressing plant defense responses in sedentary endoparasitic nematodes (Favery et al., 2016). These include the root-knot nematode-secreted effectors Mi-CRT, MiMSP40, MiSGCR1, MiISE6, Mj-TTL5, Mh265, MeTCTP, MgGPP, and MgMO237, as well the cyst nematode-secreted effectors Hs10A06, GrSPRYSEC19, GrCEP12, GrVAP1, Ha-ANNEXIN, and HgGLAND18 (Hewezi et al., 2010; Postma et al., 2012; Chronis et al., 2013; Jaouannet et al., 2013; Lozano-Torres et al., 2014; Chen et al., 2015, 2017; Lin et al., 2016; Niu et al., 2016; Noon et al., 2016; Gleason et al., 2017; Zhuo et al., 2017; Chen J. et al., 2018; Nguyen et al., 2018; Shi et al., 2018). Moreover, the increasing availability of genome sequences for plant parasitic nematodes now promotes to identify more and more effectors (Abad et al., 2008; Opperman et al., 2008; Kikuchi et al., 2011; Thorpe et al., 2014; Eves-van Den Akker et al., 2016). In particular, many effectors are actually related proteins encoded by gene families (Cotton et al., 2014). One notable family is that of the SPRY domain gene (approximately 300 sequences) in Globodera pallida, which has several secreted protein members that function as selective suppressors of defense-related cell death in plants (Mei et al., 2015; Diaz-Granados et al., 2016). The HYP effectors comprise a large gene family with continual expression and they play an important role in plant–nematode interactions (Eves-van Den Akker et al., 2014). The diversity in the effector family may be due to selection pressures to evade recognition by the host.

The effectors G16B09, 4D06, and related proteins (here referred to as the "G16B09 family") were first identified in a gland-cell cDNA library of H. glycines, which had been built by micro-aspirating the cytoplasm from esophageal gland cells of parasitic nematode stages (Gao et al., 2003). Since then, two new G16B09-family members were likewise identified from H. glycines (Noon et al., 2015). Therefore, 11 distinct member proteins of the G16B09 family are currently known in the nematode H. glycines. In G. pallida, the G16B09 family is considered among the largest of its gene families, for which 39 members have been identified so far (Thorpe et al., 2014). The mRNAs of all these members are expressed specifically within the dorsal gland cell of parasitic stages of H. glycines or G. pallida, indicating their likely contribution to syncytium induction and formation. Nevertheless, all these members are also novel transcripts with no homology to any reported genes in public databases, rendering them the "pioneers" designation. Nor were functional domains detected in any of them using computational tools. Characterizing the functions of this complex effector family would provide crucial information for better understanding nematode–plant interactions.

In this study, we identified a G16B09 family from H. avenae. Then we characterized one G16B09-like effector protein (here named "Ha18764" after its transcriptome identification number) with a significant virulence function in nematode– plant interactions. We determined that Ha18764 provides this nematode with a significant virulence function. Furthermore, we present several lines of ancillary evidence showing this novel effector most likely works by suppressing PTI and/or ETI responses in host plants, to facilitate H. avenae parasitism. This study provided an experimental clue for further investigating the functions of G16B09-like effector proteins.

# MATERIALS AND METHODS

# Nematodes and Plants

H. avenae was propagated on wheat (Triticum aestivum cv. Aikang 58) using second-stage juveniles (J2s) hatched from cysts, previously collected from a wheat field in Qingdao, China. The pre-parasitic (pre-J2s) were collected by hatching the cysts at 15◦C after at least 4 weeks incubation at 4◦C. To obtain the parasitic nematodes, infected wheat roots were obtained at 5, 20, and 30 days post inoculation (dpi), cut into sections, and digested at 28◦C by shaking at 160 rpm in a 6%-cellulose water solution overnight (Chen et al., 2015). The parasiticstage juveniles (par-J2) were obtained directly from 5 dpi. The third-stage (J3) and fourth-stage (J4) juveniles were, respectively, obtained from 20 dpi and 30 dpi. Females were collected by hand picking them from the wheat root surfaces. In our laboratory, H. schachtii nematodes were propagated on the beet Beta vulgaris L. heir pre-J2s were collected by hatching the cysts at 25◦C.

Wheat and barley (Hordeum vulgare cv. E9) were grown in a greenhouse at 22◦C under a 16-h light/8-h dark cycle. Nicotiana benthamiana was grown in a growth chamber at 25◦C under a 14-h light/10-h dark cycle. The Arabidopsis thaliana plants were grown on solidified Murashige and Skoog (MS) medium with 2% sucrose under sterile conditions, or grown in potting soil in a growth chamber at 23◦C (16-h light/8-h dark cycle).

# Gene Amplification and Sequence Analysis

Genomic DNA and total RNA were prepared from freshly hatched pre-J2s using, respectively, the TIANamp Micro DNA Kit and RNAprep Pure Micro Kit (Tiangen, Beijing, China). The cDNA was synthesized from total RNA by using the SMART <sup>R</sup> MMLV Reverse Transcriptase (Takara, Tokyo, Japan) according to the manufacturer's instructions. Based on our H. avenae transcriptome data (Yang et al., 2017), the DNA sequence of Ha18764 and the cDNA of all the genes were cloned by PCR amplification using their specific primers (**Supplementary Table S1**). To search for homologies, the Ha18764 cDNA sequence was BLASTed against the GenBank database or the published genomic database of potato cyst nematodes (Cotton et al., 2014; Eves-van Den Akker et al., 2016). The H. schachtii Ha18764-like sequence was obtained by PCR using the primers HgG16B09cds-F/R and cDNA template. All the primers used in this study are listed in **Supplementary Table S1**.

To identify the effector gene homologs, a local, command line BLAST was carried out against the H. avenae transcriptome sequence, using an E-value threshold of 10−<sup>5</sup> and with the low complexity filtering turned off (Thorpe et al., 2014). The sequence homology of the predicted proteins was then analyzed using DNAMAN, Clustal X v2.0, and BoxShade software tools. We searched for the conserved domain and secretory signal peptide (SP) with NCBI CD-Search<sup>1</sup> and SignalP v4.1<sup>2</sup> , respectively. Prediction of putative transmembrane domains were obtained according to TMHMM<sup>3</sup> . Finally, the in planta subcellular localization was predicted using PSORT<sup>4</sup> .

# Developmental Expression Analysis and in situ Hybridization

Total RNA was extracted and cDNA synthesized from different stages nematodes as described above. Using the primer pairs Ha18764qPCR-F/Ha18764-qPCR-R and HaGAPDH-1- F/HaGAPDH-1-R, respectively, qRT-PCR amplified the Ha18764 gene and the endogenous reference gene HaGAPDH-1, with the reagent SYBR Premix Ex Taq II (Tli RNaseH Plus; Takara, Tokyo, Japan) on a ABI PRISM 7500 system (Applied Biosystems, United States). Triplicate PCR reactions for each cDNA sample were carried out, and the assay itself consisted of three technical replicates. The obtained data were analyzed following the 2 <sup>−</sup>11Ct method.

For the in situ hybridization, H. avenae J2s were hatched in leachates of wheat root and collected. The primers insitu-Ha18764-F/in-situ-Ha18764-R (**Supplementary Table S1**) were used to synthesize the DIG-labeled antisense and sense (negative control) cDNA probes (Roche, United States) by an asymmetric PCR. Hybridization was conducted as described previously (de Boer et al., 1998), and examined under a BX51 microscope (Olympus, Japan). Three independent experiments were performed.

# Subcellular Localization in N. benthamiana

The Ha18764 gene without its SP-encoding region was amplified, by using the primer pairs Ha18764dsp-F/Ha18764dsp-R that, respectively, contained Sal I and Xma I restriction enzyme sites (**Supplementary Table S1**). The ensuing amplified fragments were cloned into the corresponding sites in the p35SeGFP vector to express the eGFP fusion protein. The empty vector served as the control. The construct was confirmed by sequencing, after which it was transformed into Agrobacterium tumefaciens strain EHA105. The recombinant A. tumefaciens carrying p35SeGFP-Ha18764 or p35SeGFP was infiltrated into N. benthamiana leaves, as described by Chen et al. (2015). After 48 h, infiltrated leaves were visualized under laser confocal fluorescence microscope (Nikon Eclipse TE300, Tokyo, Japan) at an excitation wavelength of 488 nm. Three independent experiments were performed.

# Validation of the Predicted Signal Peptide in Barley

To assess whether the SP of Ha18764 is secretory, a live-cell imaging approach with slight modifications (Park et al., 2012) was developed to localize Ha18764 in barley. First, the Ha18764 gene with its SP-encoding region was amplified using the primers Ha18764VaF/R containing the Hind III and Bam HI restriction enzyme sites (**Supplementary Table S1**), respectively. These amplified fragments were cloned into the respective sites in the pRP27-mcherryNLS vector to express the mCherry fusion protein. The empty vector was used as a negative control. The ensuing constructs were transformed into the protoplast of the fungal Magnaporthe oryzae strain p131. This recombinant p131 carrying the constructs was cultured in an OTA medium (oatmeal tomato agar medium) at 26◦C for 10 days (Yang et al., 2010). Next, the spores were suspended in 5% Tween 20, to an appropriate concentration of 10–15 spores/100 µL, then inoculated to in vitro leaves of 10-day-old barley. After inoculation, the barley leaves were incubated at 26◦C for about 27 h under wet and dark conditions. Nuclei were stained with DAPI (4<sup>0</sup> ,6-diamidino-2-phenylindole; Katsuhara and Kawasaki, 1996) and visualized under a BX61 microscope (Olympus, Japan). Three independent experiments were performed.

# Silencing of Ha18764 by BSMV-HIGS and the H. avenae Infection Assay

The specificity of selected gene fragments of Ha18764 was confirmed by a BLAST search with NCBI data and our H. avenae transcriptome data. The specific Ha18764RNAi fragment was amplified by PCR using the primer pairs Ha18764RNAi-F/Ha18764RNAi-R (**Supplementary Table S1**). Barley stripe mosaic virus-medicated host-induced gene silencing (BSMV-HIGS) and nematode infection assay were conducted as previously described (Yuan et al., 2011; Chen et al., 2015). For the infection assay, approximately 300 J2s of H. avenae were inoculated to wheat plants (n = 16), then the number of

<sup>1</sup>http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi

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

<sup>3</sup>http://www.cbs.dtu.dk/services/TMHMM/

<sup>4</sup>https://www.genscript.com/psort.html

nematodes in the roots were counted at 7 dpi, and females at 50 dpi were also counted. Meanwhile, the expression level of Ha18764 in nematodes from wheat inoculated by BSMV:Ha18764 relative to that of the blank negative control (BSMV:00) and the negative control (BSMV:eGFP) were determined by qPCR. This experiment was independently repeated three times. Independent-samples t-tests or one-way ANOVA (Duncan's test for pairwise means), conducted in SPSS v13.0, were used to analyze the differences between the treatment groups.

# Suppression of Immune-Associated Cell Death

This assay were conducted as previously described (Chen C. et al., 2018). Coding sequences (without SPs) of each gene of H. avenae were constructed into the PVX vector pGR107 (Jones et al., 1999), fused with the 3× flag-tag at the N-terminus by an In-Fusion HD Cloning Kit (Clontech, United States). Using the same kit, the necrosis elicitor gene psojNIP (Qutob et al., 2002) was constructed into the pGR107 vector with an HA-tag fused at the C-terminus. Four candidate effector genes of H. avenae namely isotig16511, isotig16978, isotig19390, and isotig12969 capable of triggering cell death in N. benthamiana leaves were constructed into the vector pND108 with a HA tag. These constructs were confirmed by sequencing and then transformed into the A. tumefaciens strain GV3101 for infiltration. The pGR107:GFP with a flag-tag, and both pGR107:NbMKK1 and pGR107:NbNPK1Nt with an HA tag had been generated in our prior work (Chen et al., 2015). Other researchers kindly provided us with the construct pGR107-Bax, the vectors expressing Avr3a, R3a, Gpa2, or Rbp-1, as well as the empty vector PMD1 (see "Acknowledgments" section).

Suppression of cell death as mediated by different elicitors in N. benthamiana leaves was assayed as already described elsewhere (Wang et al., 2011). The A. tumefaciens cells carrying Ha18764, or other family genes, were infiltrated into the leaves of N. benthamiana. After 24 h, A. tumefaciens cells carrying the elicitor genes were infiltrated into the same site, while the A. tumefaciens strain carrying the Ha18764 or GFP gene and a buffer was infiltrated alone as the controls. These assays were independently repeated at least three times, with three to six N. benthamiana plant replicates inoculated each time (to three leaves per plant). Photographs of the infiltrated leaves of N. benthamiana were obtained ca. 5 days since the last infiltration was made or after decolorizing their leaves (by boiling in alcohol for 20 min). To verify gene expression, a Western blotting protocol was followed as described previously (Chen et al., 2015).

# Generation of Transgenic Arabidopsis thaliana Plants and the Heterodera schachtii Infection Assay

The Ha18764- or Hs18764-coding cDNA sequence (without SP) was amplified by the primer pairs 1300-Ha18764- F/1300-Ha18764-R or 1300-Hs18764-F/1300-Hs18764-R (**Supplementary Table S1**) and these generated sequences were inserted into the Hind III or Kpn I respective sites of the vector pSuper1300, respectively. Then the ensuring constructs were transformed into A. tumefaciens GV3101, which was used to transform the A. thaliana ecotype Col-0 by the floral dip method. Seeds of the transformants were collected and stored at 4◦C for 7 days, then selected by Hygromycin B in an MS solidified medium containing 2% sucrose for ca. 14 days, then transplanted into soil. Homozygous T3 seeds collected from the T2 lines were used.

For the infection assay, 14-day-old A. thaliana plants (col-0, col-0 containing Ha18764 or Hs18764) were inoculated with 300 pre-J2s of H. schachtii. For each host plant, their respective number of infected nematodes in the roots was counted at 14 dpi, (n = 20, respectively). This experiment was independently repeated three times. Independent-samples t-tests or one-way ANOVA (Duncan's test) were used to analyze the differences in infection between the treatment groups by SPSS software.

# Defense-Related Gene Expression in Transgenic Arabidopsis

To determine the expression levels of defense-related genes, 14-day-old A. thaliana seedlings were soaked in sterile water containing 10 µM of flg22. Total RNA were isolated from 50 mg of Arabidopsisseedlings after 4 h using the TRIzol RNA extraction reagent (Invitrogen, United States). Transcript abundances of WRKY70, WRKY29, PR-1, and CYP81F2 were determined by RT-qPCR. Arabidopsis actin served as an internal control to normalize the gene expression levels. Each sample reaction was run in triplicate. Independent-samples t-tests or one-way ANOVA (Duncan's test) were used to analyze the differences in transcript abundances.

# ROS Generation Analysis

Detection of the flg22-mediated oxidative burst was performed using a luminol-HRP-based chemiluminescence assay. In this assay, the Ha18764-coding cDNA sequence (without SP) was amplified by primer pairs 1132-Ha18764-F/1132-Ha18764-R (**Supplementary Table S1**), then subcloned into the BamH I/Sal I restriction sites of the vector pYBA1132. The ensuing construct pYBA1132:Ha18764 was introduced into A. tumefaciens GV3101 (freeze-thaw method). Then the GV3101 containing either pYBA1132 or pYBA1132:Ha18764 was infiltrated into N. benthamiana leaves (4- to 5-week-old plants). At 36-h postinfiltration, leaf discs (4 mm diam.) were collected and incubated overnight in 100 µL of H2O, in a 96-sample microplate, and substituted by 100 µL elicitor master mix (100 µM luminol, 20 µg/ml horseradish peroxidase, 100 nM flg22). The plate was immediately put into the microplate reader, with ROS production monitored for 40 min (Sang and Macho, 2017). This assays were performed three times, with triplicate reaction for each sample.

# Callose Staining

This assay were conducted as previously described (Tran et al., 2017). Following treatment with 1 µM of flg22, 8-day-old Arabidopsis seedlings were cultivated on the <sup>1</sup>/2-MS basal agar medium for 72 h. Their roots were fixed overnight in a solution containing 95% ethanol and acetic acid (3:1), followed by their rehydration in 70% ethanol for 1 h, 50% ethanol for 1 h, and

distilled water for 1 h, and then treatment with 10% NaOH for 1.5 h at 37◦C to make their root tissues transparent. Finally, the roots were incubated in a staining solution (0.01 % aniline blue, 150 mM K2HPO4, pH 9.5) for at least 1 h; root tips ca. 1–2 cm length were excised and mounted onto slides for callose observation under a Leica TCS SP8 microscope with UV light (excitation, 390 nm; emission, 460 nm). Images were photographed in the field of approximately 2 mm<sup>2</sup> that captures the root area containing the root elongation zone. Callose deposits in 12 roots per treatment from three independent experiments were counted using ImageJ software<sup>5</sup> .

# RESULTS

# Most G16B09 Family Effectors From H. avenae Suppress BAX-Triggered Programmed Cell Death (BT-PCD) in N. benthamiana

A BLAST search of our transcriptome data of H. avenae (Yang et al., 2017) examined 12 homologs of this nemtaode's G16B09 effector family (**Supplementary Figures S1**, **S7**), using an E-value threshold of 10−<sup>5</sup> (Thorpe et al., 2014). No domains, motifs, or features could be predicted from the sequences, which are identical to those in H. glycines and G. pallida (Cotton et al., 2014). Eight of the 12 homologs were successfully cloned their GenBank accession numbers are listed in **Supplementary Figure S1**—and investigated for their role in suppressing plant defenses. Infiltration with the Agrobacterium carrying Bax alone triggered a typical PCD reaction, whereas infiltration with GFP did not suppress BT-PCD (**Supplementary Figure S2A**). Of the eight H. avenae candidate effector genes evaluated, seven could suppress BT-PCD (**Supplementary Figures S2B–H**), while one gene had negligible effects on suppressing BT-PCD or triggering cell death (**Supplementary Figure S2I**). Western blotting confirmed the expression of H. avenae candidate effector protein, eGFP, and BAX. This finding suggested that the G16B09 like family of H. avenae contribute to suppressing the host plant immune system.

To further characterize the role of the G16B09 family from H. avenae in parasitism and plant defense suppression, the isotig18764 (named Ha18764 after its transcriptome identification number) with greatest amino acid similarity (61.9%) to G16B09 of H. glycines was selected for detailed study. An 844-bp genomic fragment of Ha18764 was obtained, consisting of an open reading frame (ORF) of 567 bp separated by three introns of 75 bp, 98 bp, and 104 bp (**Figure 1A**), each having conserved 5<sup>0</sup> -GT-AG-3<sup>0</sup> splice sites. Ha18764 encodes 188 amino acids with a SP of 29 amino acids at its N-terminus (as predicted by the SignalP 4.1 server). This protein has no putative transmembrane domain (based on TMHMM Server v2.0) with a predicted molecular size of 19.94 kDa. According to the NCBI CD-search, Ha18764 has no conserved domain, motifs, or features. No nuclear localization signals were predicted for Ha18764 (according to the PSORT analysis).

For the alignment analysis, the homologs with highest similarity to HgG16B09 from G. pallida and G. rostochiensis were obtained by a BLAST search against the public genome database (Cotton et al., 2014; Eves-van Den Akker et al., 2016). The homologous sequence from H. schachtii, here designated as Hs18764 (GenBank Accession MH794364), was generated by PCR using the primers HgG16B09cdsF/R (**Supplementary Table S1**). An alignment of deduced amino acid sequences of G16B09 like proteins from different nematode species is presented in **Figure 1B**. A consensus phylogeny tree based on the analyzed protein sequences divided them into three clades. Ha18764 showed 43–63% shared amino acid identity with other nematode homologs (**Figure 1C**).

# Ha18764 Is Expressed in the Dorsal Gland and Is Dramatically Up-Regulated in par-J4 of H. avenae

In situ hybridization was performed to investigate the tissue localization of Ha18764 in H. avenae. No signals were detected in H. avenae pre-J2s. However, when the pre-J2s were pre-treated with the leachates of wheat roots, signals were observed in the dorsal gland cells after the hybridization with the DIG-labeled antisense probe (**Figure 2A**). In the negative control (sense probe), no signals were detected in pre-J2s.

The expression level of Ha18764 was determined by qPCR analysis for six developmental stages: egg, pre-J2, par-J2, J3, J4, and adult female. The expression level of Ha18764 transcripts at the egg stage was set at a value of one, to serve as the baseline for examining the relative fold changes in later stages. Ha18764 transcripts increased dramatically in the parasitic stages, reaching a maximum in the J4 stage that represented a 144-fold increase in expression (**Figure 2B**). These findings suggest Ha18764 may be secreted from dorsal gland cells and that it participates in parasitic stages, particularly in the later ones of H. avenae parasitism.

# Functional Validation of the Predicted Signal Peptide (SP) of Ha18764

We have employed the yeast secretion system to verify the activity of the predicted SP of H. avenae effectors (Chen C. et al., 2018). However, the SP of Ha18764 was found to lack secretion activity in yeast cells (data not shown). Considering that secretion activity may differ between yeast and pathogens, a Magnaporthe oryzae secretion system was exploited to experimentally verify the secretion of Ha18764 (Park et al., 2012; Zhang et al., 2018). The Ha18764 ORF (including the SP encoding sequence) was fused in-frame with the mCherry gene and a nuclear localization signal (NLS). The fusion constructs were transformed into M. oryzae strain p131, which was then used to inoculate the in vitro leaves of barley. In this assay, a functional SP could guide secretion of the mCherry fusion protein into barley cells during the infection of M. oryzae, with the fusion protein imported into the cell nucleus, targeted by the NLS; this should facilitate visualization of the translocated fluorescent protein by concentrating the fusion protein in the barley nucleus. Our results revealed that red fluorescence was observed only in the nucleus of barley inoculated with M. oryzae carrying the Ha18764-mCherry-NLS

<sup>5</sup>http://imagej.nih.gov/ij/

fusion construct (**Figure 3**). By contrast, no red fluorescence could be found in the barley nucleus inoculated with M. oryzae carrying the empty vector. The result indicated the secretion signal of Ha18764 guided mCherry to secrete into barley cells.

# Ha18764 Is Localized in the Whole Plant Cell

To investigate subcellular localization of Ha18764 in plant cells, a transient expression assay was performed with N. benthamiana leaves. The Ha18764 coding sequence without its SP was fused in-frame with the enhanced green fluorescent protein (eGFP) and transformed into Agrobacterium for infiltration into N. benthamiana leaves. The transient expression of the fusion protein and eGFP alone showed the same accumulation of the GFP signal in both cytoplasm and nuclear (**Figure 4**). The nuclear accumulation of eGFP and Ha18764-eGFP may be due to its small size and passive diffusion of the fusion protein.

FIGURE 2 | Spatial and developmental expression of Ha-18764. (A) Localization of Ha18764 mRNA in the dorsal gland of pre-parasitic second stage juveniles of Heterodera avenae by in situ hybridization. The dorsal gland (DG), metacorpus (M), and stylet (S) are indicated with arrows. (B) Developmental expression pattern of Ha18764. The relative expression level of Ha18764 was quantified using qPCR for six different H. avenae stages. The fold-change values were calculated using the 2 <sup>−</sup>11Ct method and presented as the change in mRNA level in various nematode developmental stages relative to that of egg. Each column represents the mean (±SD) of three samples. This experiment was independently repeated three times, with consistent results. pre-J2, pre-parasitic second-stage juvenile; par-J2, J3, J4, parasitic second-, third- and fourth-stage juveniles, respectively.

# Expression of Ha18764 in Arabidopsis Improved Susceptibility to H. schachtii

Overexpressing an effector protein in the host plant is typically employed to investigate its involvement in parasitism. However, the efficacy of genetically transforming wheat is low and this process is slow, and H. avenae has a narrow host range. Therefore, to aid protein functional characterization, the model plant Arabidopsis was used because it is a host for H. schachtii, a close relative of H. avenae. Specifically, Hs18764 was cloned and used as a homolog control; it shared a 71.4% sequence identity with Ha18764. Two independent homozygous T3 lines expressing either Ha18764 or Hs18764 transcripts were generated. Then H. schachtii was inoculated to determine the susceptibility of these transgenic Arabidopsis lines to nematode infection. The results showed that, at 14 dpi, either Ha18764 or Hs18764 transgenic lines were significantly more susceptible to H. schachtii infection than the wild-type Arabidopsis (P < 0.05), displaying average increases in numbers of nematodes per root that ranged from 96.0 to 178.1% and 68.0% to 145.1%, respectively, over the wild-type (**Figure 5A**). These results indicated that Ha18764 plays an important role in nematode parasitism.

# BSMV-HIGS of Ha18764 Impairs Nematode Parasitism

Barley stripe mosaic virus (BSMV) vectors are efficient vehicles for virus-induced gene silencing (VIGS) in wheat

infiltrated into N. benthamiana leaves with pCamv35S: GFP used as the control. Scale bar = 20 µm.

(Yuan et al., 2011). A novel approach, called "host-induced gene silencing" (HIGS), can silence a pathogen's genes with BSMV-VIGS to interfere with its effective infection of wheat (Nowara et al., 2010; Yin et al., 2011). Recently, this system was successfully employed to silence nematode gene during parasitism of wheat (Chen et al., 2015). Here, the expression of Ha18764 in nematodes recovered from wheat inoculated by BSMV:Ha18764 showed a significant reduction compared with that from the controls BSMV:00 and BSMV:eGFP (P < 0.05) (**Figure 5B**). Accordingly, H. avenae infection of wheat inoculated by BSMV:Ha18764 showed significant resistant to nematodes, displaying a 66.2% or 67.0% reduction in juveniles abundance per plant at 7 dpi (**Figure 5C**), and a 46.7% or 55.2% reduction in female abundance per plant at 50 dpi (**Figure 5D**), when compared, respectively, to the negative control BSMV:00 or to BSMV:eGFP (P < 0.05). These results provided further evidence for the important involvement of Ha18764 in nematode parasitism.

# Ha18764 Suppresses Immune-Associated Cell Death in N. benthamiana

To further explore the possible role of Ha1874 in plant defense suppression, we tested for the suppression of cell death triggered by various elicitors. PsojNIP (Qutob et al., 2002), Avr3a/R3a (Abramovitch et al., 2003), RBP-1/Gpa2 (Sacco et al., 2009), and MAPK cascade-associated protein kinases (MKK1 and NPK1Nt) (Jin et al., 2002; Gao et al., 2008) were all used to investigate

consistent results. Columns for the same time point or treatment marked with different letters are significantly different (P < 0.05) from each other.

the ability of Ha18764 to influence cell death suppression during parasitism. At least three repeated experiments showed that Ha18764 suppressed cell death induced by any one of PsojNIP (**Figure 6A**), Avr3a/R3a or RBP-1/Gpa2 (**Figure 6B**), or MKK1 and NPK1Nt (**Figure 6C**). This suggested Ha18764 is a potent suppressor of plant immune-associated cell death.

# Ha18764 Suppresses Cell Death Induced by Other H. avenae Putative Effectors in N. benthamiana

In the screening of the transient expression assay for N. benthamiana leaves, several H. avenae putative effectors could themselves trigger cell death (Chen C. et al., 2018). Hence, it was pertinent to test whether Ha18764 might also suppress cell death triggered by these cell-death-inducing effector candidates. Thus, four cell-death-inducing genes, isotig16511, isotig16978, isotig19390 and isotig12969, were selected as cell death inducers and used for agroinfiltration tests in N. benthamiana. As expected, Ha18764 suppressed cell death induced by all the four H. avenae inducers (**Figure 7**). This indicated the cooperation of Ha18764 with other H. avenae effectors in regulating plant defenses.

# Ha18764 Suppresses Plant PTI Responses

To test whether or not Ha18764 is capable of suppressing a plant's PTI responses, we measured the ROS production, deposition of cell wall callose and expression levels of defense-related genes after inducing PTI responses by flg22. When challenged with flg22, ROS strongly decreased in the infiltrated N. benthamiana leaves expressing Ha18764 when compared with the empty vector control, which had an obvious induction of ROS production (**Figure 8A**). Similarly, callose deposition was reduced considerably in the roots of transgenic Arabidopsis plants expressing Ha18764 or Hs18764 compared with wild-type plants after their treatment with flg22 (**Figures 8B,C**).

To confirm the PTI-suppression ability of Ha18764, mRNA expression levels of four defense-related genes (CYP81F2, WRKY70, WRKY29, PR-1) were quantified by qPCR in transgenic and wild-type A. thaliana plants treated with flg22. As expected, flg22 strongly augmented the expression of these four defense marker genes in wild-type plants, which were 2.5- to 5.8-fold higher than that of the untreated control, whereas the induction levels of these defense genes were significantly repressed in transgenic Arabidopsis lines expressing Ha18764 or Hs18764 (**Figure 8D**). This provided direct evidence that Ha18764 can also function as a suppressor of the PTI response in plants.

# DISCUSSION

Some plant-parasitic nematode effectors are present in large multi-gene families, and the G16B09 family is considered among the largest of effector families (Cotton et al., 2014). To date, 11 members from H. glycines (Gao et al., 2003; Noon et al., 2015) and 39 members from G. pallida (Thorpe et al., 2014) have

been reported. Yet, no conserved domains, motifs, or features are predicted from this family, and relatively little is known about their functional roles in nematode–plant interactions. In this study, we demonstrated that several identified members of a G16B09-like family from H. avenae suppressed BT-PCD. To our knowledge, the most intensively studied effector family regulating plant defenses is SPRYSEC (secreted proteins containing a SPRY domain), considered to be one of the largest effector families in G. pallida (Thorpe et al., 2014). However, SPRYSEC effectors have been implicated in both the suppression and activation of plant immunity (Diaz-Granados et al., 2016). Effector families have also been reported from other plant pathogens; for example, effector sequences are more likely to be found in repeat-rich, gene sparse regions of the genome in Phytophthora infestans (Haas et al., 2009). The expanded members in such gene families may reflect the outcome of selection pressure to avoid detection or to maintain key functions within a host.

In H. avenae's G16B09 family, Ha18764 that is most alike to the H. glycines G16B09 effector, was selected for further detailed studies. Using the M. oryzae secretion system, we found that the SP of Ha18764 was active in guiding the protein product into the cells of the barley leaves. As reported for H. glycines and G. pallida, the in situ hybridization of Ha18764 revealed dorsal gland localization, additional evidence in support of the secretory ability of Ha18764. Usually, dorsal gland cells are more active in the sedentary parasitic stages, when they secrete effectors involved in feeding site formation and maintenance, while subventral gland cells primarily function in secretion during migratory stages, producing proteins required for root invasion and nematode movement within the host plant (Mitchum et al., 2013). Recently, G16B09 family members were found expressed exclusively within the dorsal gland cells in the parasitic stages of H. glycines and G. pallida (Gao et al., 2003; Thorpe et al., 2014; Noon et al., 2015), thus indicating this gene family's expression is restricted to the feeding stages. Our results agree with previous reporting on the G16B09 family. Firstly, the hybridization signal of Ha18764 was detectable only when the J2s were pretreated with host leachates. Secondly, developmental expression pattern analysis showed greater expression of Ha18764 during the parasitic stages that peak in the J4 stage. So, it is reasonable to presume that Ha18764 functions mainly in the nematode's parasitic stages.

carrying the elicitors. Photographs of infiltrated leaves were taken ca. 4 days after the last infiltration. Four circles marked with Roman numerals represented regions injected with different A. tumefaciens, which is indicated under the photographs. The numbers in parentheses are the proportion of infiltrated sites showing cell-death-suppressing symptoms. Results for the verification of gene expression by Western blotting are shown below. The original Western blotting images for the verification of gene expression were provided in Supplementary Figures S5, S6.

Due to the narrow host-plant range of H. avenae and wheat's complicated genetic manipulation, both the Arabidopsis– H. schachtii infection system and the wheat BSMV-HIGS system were used to investigate Ha18764 functioning during parasitism. Importantly, as a reference, H. schachtii infection of the Arabidopsis model was employed to verify the role of H. glycines effectors in parasitism (Pogorelko et al., 2016; Barnes et al., 2018). The BSMV-HIGS system has been successfully utilized to silence genes in H. avenae, by delivering dsRNA from wheat to the nematodes (Chen et al., 2015). In our present study, both Ha18764- and Hs18764-transgenic Arabidopsis lines were more susceptible to H. schachtii infection than wild-type plants. Meanwhile, silencing of Ha18764 in vivo using the BSMV-HIGS system significantly impeded nematode infection of wheat. These results further confirmed that Ha18764 has a role to play in parasitism.

Cyst nematodes are considered as biotrophic pathogens, because they feed from the syncytia until their reproduction is complete. Therefore, H. avenae needs to suppress plant defenses during the entire parasitic process. For survival, H. avenae must possess the ability to suppress the host defenses including PTI and ETI. Our results showed that Ha18764 could suppress both PTI and ETI. The PTI assay showed that Ha18764 could indeed suppress the deposition of callose and the production of ROS. Since Ha18764 was able to suppress programmed cell death induced by the R-protein/cognate effector pairs (i.e., R3a/Avr3a and RBP-1/Gpa2), this suggested Ha18764 suppresses plant ETI responses. Several nematode effectors are reported to be capable

of suppressing both ETI and PTI. For example, The CEP12 peptide from Globodera rostochiensis suppresses both resistancegene-mediated cell death and flg22-mediated ROS production (Chen et al., 2013; Chronis et al., 2013). The M. incognita putative secretory protein MiMsp40 suppresses cell death triggered by the ETI elicitors R3a/Avr3a, and overexpression of MiMsp40 in plants suppresses the deposition of callose and the expression of PTI marker genes (Niu et al., 2016). Moreover, Ha18764 also suppressed two MAPK kinases genes (MKK1 and NPK1) triggering cell death in N. benthamiana. Plant MAPK cascade pathways play remarkably important roles in plant defense signaling (Dodds and Rathjen, 2010; Tena et al., 2011), and they are considered crucial for PTI and ETI responses (Meng and Zhang, 2013). Our results indicate Ha18764 specifically targets a point downstream of MKK1 and NPK1 in the signaling pathway. Furthermore, Ha18764 could suppress the expression of a suite of key plant defense-related genes that are mainly involved in the SA signaling pathway. The Arabidopsis CYP81F2 gene encodes a P450 monooxygenase that is essential for antimicrobial defense (Bednarek et al., 2009). WRKY70 functions as an activator of salicylic acid (SA) induced genes and as a repressor of jasmonic acid (JA)-responsive genes (Li et al., 2004). The flg22-triggered transcription of WRKY29 was recently shown to depend on SA signaling (Yi et al., 2014). PR-1 is commonly used as molecular marker for SA-dependent systemic acquired resistance signaling (Bowling et al., 1994). Accordingly, we hypothesize Ha18764 interferes with the SA signaling pathway to suppress host immune responses. SA-dependent signaling is considered to be crucial for resistance against biotrophic pathogens (Delaney et al., 1994). For example, the nematode H. schachtii reportedly elicits SAdependent plant resistance in both roots and leaves of infected Arabidopsis (Hamamouch et al., 2011). Because our results are consistent with these previous reports, we propose that Ha18764 interferes with SA-dependent plant resistance to promote nematode parasitism.

Consistent with our recent work (Chen C. et al., 2018), evidence for interaction between Ha18764 and H. avenae putative effectors for cell death induction was found. Ha18764 also suppressed cell death triggered by four H. avenae cell-deathinducing effectors. The similar interaction between effectors was also found in SPRYSECs effectors of G. pallida and RxLR effectors of P. sojae (Wang et al., 2011; Mei et al., 2015). Furthermore, in both P. sojae and P. parasitica, effector XEG1 is bound by host-secreted GmGIP1 which blocked its contribution to virulence; however, these pathogens secrete a paralogous XLP1 that binds to GmGIP1 more tightly than does PsXEG1, thus freeing XEG1 to support

the infection process (Ma et al., 2017). It is possible the same mechanism exists in H. avenae. Diversity in the effector family may have arisen from selection pressure to escape recognition by hosts. Through its interacting effectors, H. avenae effectively avoided inducing host defenses against its biotrophic parasitism.

In summary, we have identified a pioneer effector family from the nematode H. avenae and verified that the majority of the G16B09-like family members could suppress BT-PCD. One novel family member in particular, Ha18764, suppresses various defense responses as a secreted effector during interaction with plants. Our study suggests Ha18764 may indeed benefit parasitism by suppressing host-plant immunity encompassing diverse PTI and ETI pathways. Further study of these family members and their interactions with receptors in host cells may reveal the molecular mechanism underlying plant defense suppression.

# AUTHOR CONTRIBUTIONS

HJ and QL conceived the idea for this study, acquired its funding, and designed the experiments. SY, YD, and YC performed the experiments. DY carried out the bioinformatics analyses. JY provided the pRP27-mCherryNLS system. QL, SY, and HJ wrote and revised the manuscript. All authors read and approved the final version of the manuscript for publication.

# FUNDING

This research was financially supported by the Special Fund for Agro-Scientific Research in the Public Interest, China (Grant No. 201503114), the National Key Research and Development Program of China (Grant No. 2017YFD0200601), and the National Natural Science Foundation of China (Grant No. 31871940).

# ACKNOWLEDGMENTS

We thank the following Professors: Xiaohong Wang (Cornell University), Yuanchao Wang and Daolong Dou (Nanjing

# REFERENCES


Agricultural University), Dawei Li and Chengui Han (China Agricultural University), and Bingyan Xie (Chinese Academy of Agricultural Sciences), for providing the R3a/Avr3a, Gpa2/RBP-1, pGR107-bax, and psojNIP expression vectors, the pMD1 and pGR107 vectors, the BSMV-HIGS system and N. benthamiana seeds. We also thank Dr. Na Jiang for use of the confocal microscope.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00066/ full#supplementary-material

FIGURE S1 | Phylogenetic tree of G16B09 family members from different nematodes. For G. pallida, the genomic gene numbers are listed. For H. glycines and H. avenae, the GenBank accession numbers are listed after the gene names. Hg, H. glycines; Ha, H. avenae; Gp, G. pallida.

FIGURE S2 | Suppression of BT-PCD by Heterodera avenae G16B09-like effector family in Nicotiana benthamiana. The N. benthamiana leaves were infiltrated with a buffer, or A. tumefaciens cells carrying the candidate gene, or the negative control eGFP gene, either alone or 24 h prior to infiltration with A. tumefaciens cells carrying Bax. Photographs of infiltrated leaves were taken ca. 4 days after the last infiltration. Numbers in parentheses indicate the proportion of infiltrated sites showing cell-death-suppressing symptoms. Shown below is the verified protein expression of H. avenae proteins, eGFP, and Bax, by Western blotting.

FIGURE S3 | The original Western blotting images embedded into Figure 6. The expression of GFP or Ha18764 were verified using anti-flag. All bands are labeled with a black line.

FIGURE S4 | The original Western blotting images embedded into Figure 6. The expression of MKK1, NPK1Nt or NIP were verified using anti-HA. All bands are labeled with a black line.

FIGURE S5 | The original Western blotting images embedded into Figure 7. The expression of GFP or Ha18764 were verified using anti-flag. All bands are labeled with a black line.

FIGURE S6 | The original Western blotting images embedded into Figure 7. The expression of Ha16511, Ha19390, Ha12969 or Ha16978 were verified using anti-HA. All bands are declared with a black line.

FIGURE S7 | Sequence analysis of Heterodera avenae G16B09-like effector family. (A) Alignment of H. avenae G16B09-like effector family. (B) Phylogenetic tree of G16B09-like family members from H. avenae.

TABLE S1 | List of all the primer sequences used in this study.



protein of the cyst nematode Heterodera schachtii. Plant Physiol. 152, 968–984. doi: 10.1104/pp.109.150557



**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 © 2019 Yang, Dai, Chen, Yang, Yang, Liu and Jian. 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.

# Comparative Genomics Reveals the High Copy Number Variation of a Retro Transposon in Different Magnaporthe Isolates

Pankaj Kumar Singh1,2, Ajay Kumar Mahato<sup>1</sup> , Priyanka Jain1,2, Rajeev Rathour<sup>3</sup> , Vinay Sharma<sup>2</sup> and Tilak Raj Sharma1,4 \*

1 Indian Council of Agricultural Research (ICAR)-National Research Centre on Plant Biotechnology, New Delhi, India, <sup>2</sup> Department of Bioscience and Biotechnology, Banasthali University, Tonk, India, <sup>3</sup> Department of Agricultural Biotechnology, Chaudhary Sarwan Kumar Himachal Pradesh Krishi Vishvavidyalaya (CSK HPKV), Palampur, India, <sup>4</sup> National Agri-Food Biotechnology Institute, Mohali, India

### Edited by:

Thomas Mitchell, The Ohio State University, United States

### Reviewed by:

Li-Jun Ma, University of Massachusetts Amherst, United States Vinicius Abreu, University of São Paulo, Brazil

> \*Correspondence: Tilak Raj Sharma trsharma@nabi.res.in; trsharma@nrcpb.org

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Microbiology

Received: 14 September 2018 Accepted: 16 April 2019 Published: 06 May 2019

#### Citation:

Singh PK, Mahato AK, Jain P, Rathour R, Sharma V and Sharma TR (2019) Comparative Genomics Reveals the High Copy Number Variation of a Retro Transposon in Different Magnaporthe Isolates. Front. Microbiol. 10:966. doi: 10.3389/fmicb.2019.00966 Magnaporthe oryzae is one of the fungal pathogens of rice which results in heavy yield losses worldwide. Understanding the genomic structure of M. oryzae is essential for appropriate deployment of the blast resistance in rice crop improvement programs. In this study we sequenced two M. oryzae isolates, RML-29 (avirulent) and RP-2421 (highly virulent) and performed comparative study along with three publically available genomes of 70-15, P131, and Y34. We identified several candidate effectors (>600) and isolate specific sequences from RML-29 and RP-2421, while a core set of 10013 single copy orthologs were found among the isolates. Pan-genome analysis showed extensive presence and absence variations (PAVs). We identified isolate-specific genes across 12 isolates using the pan-genome information. Repeat analysis was separately performed for each of the 15 isolates. This analysis revealed ∼25 times higher copy number of short interspersed nuclear elements (SINE) in virulent than avirulent isolate. We conclude that the extensive PAVs and occurrence of SINE throughout the genome could be one of the major mechanisms by which pathogenic variability is emerging in M. oryzae isolates. The knowledge gained in this comparative genome study can provide understandings about the fungal genome variations in different hosts and environmental conditions, and it will provide resources to effectively manage this important disease of rice.

#### Keywords: Magnaporthe, genome, SINE, effector, repeat, RP-2421, RML-29

# INTRODUCTION

Rice is one of the most important food security crops of developing countries. It has been reported that more than one-third of the world's population depend on it for their major calorific intake (Goff, 1999). The most devastating and economically significant disease that affects rice crops is a rice blast disease, produced by the fungal pathogen Magnaporthe oryzae. This disease has been a major problem in the most rice-producing areas and causes substantially huge yield losses of up to 30% in all parts of the world, under favorable environmental conditions (Talbot, 2003; Skamnioti and Gurr, 2009). This disease can be effectively managed by the use of fungicides, however, use of

**Abbreviations:** contig, contiguous; InDel, insertions-deletions; SINE, short interspersed nuclear element; SNP, single nucleotide polymorphism.

resistant varieties is the best option in the current scenario, which does not have any harmful effect on the environment (Singh et al., 2018a). Nonetheless, newly developed resistant rice cultivars are often defeated against the fungus after only a few years in the field due to the highly variable nature of the pathogen (Sharma et al., 2012). Blast disease resistance in rice is contingent on the existence of a related cognate avirulence (Avr) gene in M. oryzae and follows a typical gene-for-gene hypothesis (Flor, 1971; Valent, 1990).

Identification of new Avr genes and effector molecules from M. oryzae can be useful to understand the molecular mechanism involved in the fast evolution of different races of this fungus. It may help to reduce the occurrence of quick breakdown of blast resistance by the identification of potential R genes and further its application in rice resistance breeding program (Singh et al., 2018b). The emergence of advanced virulent strains of the plant pathogen is governed by Avr gene alternations through frame-shift mutations (Ridout et al., 2006), point mutation (Joosten et al., 1994), deletions (Dodds et al., 2006), and insertion of transposons (Kang, 2001; Kang et al., 2001). Instability of Avr genes via these genetic changes is a common process of evolution toward virulence as revealed by the extensive surveys of large natural populations of fungal pathogens (Van de Wouw et al., 2010). So far, over 40 Avr genes have been identified from M. oryzae (Valent and Khang, 2010), of which 11 genes, PWL1 (Kang et al., 1995), PWL2 (Sweigard et al., 1995), AvrPita (Orbach et al., 2000) ACE1 (Fudal et al., 2005), Avr1CO39 (Leong, 2008), AvrPiz-t (Li W. et al., 2009), AvrPia, AvrPii, AvrPik/km/kp (Yoshida et al., 2009), AvrPi9 (Wu et al., 2015), AvrPib (Zhang et al., 2015), and AvrPi54 (Ray et al., 2016) have been cloned and characterized. It is very essential to identify the effectors and Avr genes in the field isolates of M. oryzae in different geographical regions of the world, which may help in deployment of R genes with durable resistance against M. oryzae.

Many fungal genomes, including those from pathogenic fungi, have been decoded and made available in the public domain (Galagan et al., 2005). These provides ample opportunity for future investigations into the processes of host-pathogen interactions at the molecular level. Such studies advance our understanding about the evolution of fungal virulence genes and to recognize the gene subsets that are responsible for pathogenicity in the fungi (Soanes et al., 2002; Veneault-Fourrey and Talbot, 2005). Among the plant pathogenic fungi, M. oryzae (70-15 strain) was the first genome completely decoded in 2005 by Sanger sequencing method (Dean et al., 2005). Subsequently, several genomes of M. oryzae have been re-sequenced from various isolates of the fungus using next generation sequencing (NGS) approaches. Field isolates from China (Y34) and Japan (Ina168 and P131) were sequenced applying 454 Roche platform (Yoshida et al., 2009; Xue et al., 2012). Various field isolates from China and India, FJ81278, HN19311, B157, and MG01 have been sequenced using Illumina NGS technology (Chen et al., 2013; Gowda et al., 2015). Interestingly, the whole genome decoding of various isolates has disclosed exceeding a megabase pair isolate specific sequences, which contain hundreds of isolate specific genes. These genes determine the specificity of a particular isolate and this specificity might be due to the racial evolution after a long period of time and host range specificity, chromosomal variation and variability in repetitive elements (Dean et al., 2005; Yoshida et al., 2009; Chen et al., 2013). Several lineages of Magnaporthe are known to infect a range of host plants throughout the world (Gowda et al., 2015) and obviously they have their own specificity at the genome level. Additional comparative genome study with previously sequenced genomes help us to identify novel avirulence and effector molecules and the mechanism of host-pathogen interaction.

Several reports have been published that the transposable repeat elements act a key role in the evolution of M. oryzae genome and subsequently affect the virulence spectrum of the fungus (Xue et al., 2012; Chen et al., 2013; Gowda et al., 2015). A transposable element (TE) Pot3 is shown linkage with the virulence spectrum of M. oryzae strains (Kang et al., 2001; Li W. et al., 2009; Dai et al., 2010; Singh et al., 2014). Although, mechanisms underlying the variability of Avr determinants in M. oryzae is largely unknown. Insertion of TE (Pot3) within the promoter region of the Avr gene (AvrPita and AvrPiz-t) becomes a major causal event for the gain of virulence in the mutant strains of M. oryzae (Kang et al., 2001; Li W. et al., 2009; Singh et al., 2014). It is compelling that transposable element (TE) was always found to be connected with the malfunction or rearrangement of M. oryzae Avr genes, AvrPita, AvrPiz-t, Avr1CO39, and ACE1 (Kang et al., 2001; Farman et al., 2002; Bohnert et al., 2004; Li W. et al., 2009; Singh et al., 2014). Short interspersed nuclear element (SINE) is a retro transposable repeat element and is reported for to be actively involved in insertional inactivation of genes (Wallace et al., 1991; Goldberg et al., 1993) and in the formation of chimeric sequences (Daniels and Deininger, 1991). Such insertional inactivation of genes are operated by the mechanisms of mRNA truncation, altered polyadenylation, and modified protein structure (Oliviero and Monaci, 1988; Vidal et al., 1993; White, 1994). These repeat elements can also contribute to genomic flux and may even result in genetic disorders through the homologous and nonhomologous recombination events (Lehrman et al., 1987). Keeping in perspective the significance of R-Avr gene interaction in rice-M. oryzae pathosystem and the importance of repeat elements in M. oryzae genome, the present investigation was planned with the objectives of (i) decoding the genomes of most virulent and a least virulent strain of M. oryzae, (ii) identification of structural features in these two genomes, and (iii) comparative analysis of different M. oryzae genomes.

# MATERIALS AND METHODS

# Magnaporthe genome Sequencing and Assembly

Magnaporthe oryzae isolate Mo-nwi-31 (RP-2421) was selected for whole genome sequencing. The whole genome sequence of another isolate Mo-nwi-55 (RML-29) was generated by using Roche 454 FLX Titanium (Roche Diagnostics Corporation, Rotkreuz, Switzerland) sequencing machine. Although this genome was previously assembled and a part of data has been published by Dr. T. R. Sharma's group (Accession no. AZSW00000000; Ray et al., 2016). Both RML-29 and RP-2421

cultures were collected from the North West Himalayan region of India. The former is found to be avirulent on most of the differential rice blast resistance monogenic lines while the latter showed virulent symptoms of rice blast disease on most of the lines. These differential lines were developed and provided by the International Rice Research Institute (IRRI), Philippines. RP-2421 genome sequencing was performed by Illumina HiSeq-1000 machine at the National Research Centre on Plant Biotechnology, New Delhi. Read quality and trimming of adapter sequences from the reads were performed by three softwares, CLC Genomics Workbench 6.5 (CLC Bio, Aarhus, Denmark), FastQC V0.10.1<sup>1</sup> and NGSToolKit (Patel and Jain, 2012). The filtered reads of both data generated by 454 Roche and HiSeq-1000 were utilized for de novo assembly using CLC Genomics Workbench 6.5 (CLC Bio, Aarhus, Denmark) on Fijutsu work station with a capacity of 192 Gb physical memory. The contigs obtained after the de novo assembly was subjected to reference based assembly using reference genome of M. oryzae isolate 70- 15 (version 8, Broad Institute, Harvard). All the unmapped contigs (or unique contigs) were used as query in matching with unplaced raw reads generated during the reference genome sequencing<sup>2</sup> in the BLASTN programme of ncbi-blast-2.2.28+ package<sup>3</sup> for checking their magnitude of uniqueness exiting in the RML-29 and RP-2421 genomes. Two additional genomes of P131 (Accession no. AHZT00000000) and Y34 (Accession no. AHZS00000000) were retrieved from NCBI genome databank<sup>4</sup> and assembled them by mapping with the reference sequences of 70-15. The contigs of RP-2421 genome generated in de novo assembly were submitted at the NCBI genome databank. The assembled chromosome specific pseudo molecules of both genomes RML-29 and RP-2421 were used for further structural and functional annotations, and comparative genome analysis.

# PCR Amplification From Isolate Specific Regions

Isolate specific primers were designed using template of isolate specific sequences from each RML-29 and RP-2421 strains (**Supplementary Table 1**). Primer3 software was used for the primer designing (Untergasser et al., 2012). PCR (Gstorm thermo cycler, Somerton Biotechnology Centre, Somerton, United Kingdom) amplification was performed to confirm the isolate specific sequences present in the both strains, RML-29 and RP-2421 using standard PCR protocol of TaKaRa TaqTM DNA polymerase (DSS Takara Bio India Pvt. Ltd.) with a fixed 55◦C annealing temperature for all the primer pairs.

# Pan-Genome Analysis

Total 10 M. oryzae assembled genome sequences and their raw sequencing reads were retrieved from the NCBI genome database on the basis of N50 value and genome coverage criteria. Genome accession numbers and SRA accessions are given in **Supplementary Table 2**. The SRA data of each genome was downloaded using prefetch module of NCBI SRAToolkit<sup>5</sup> and converted into fastq format by fastq-dump module of the toolkit. The pan – genome was constructed with the help of ppsPCP software (Qamar et al., 2019). The presence and absence variation (PAV) analysis was also executed with this software with default parameters. Before this analysis, total genes were predicted from each isolate by Augustus (Stanke et al., 2006) using Magnaporthe trained dataset. For further analysis, raw reads were aligned against the pan-genome sequences using GraphMap (Sovic et al., 2016) and aligned sam files were converted and sorted into bam format using samtools (Li H. et al., 2009).

# Repeat Identification in Magnaporthe Genomes

Assembled sequences of both the genomes RML-29 and RP-2421 were passed through RepeatModeler 1.0.8<sup>6</sup> for the de novo repeat group discovery. In this de novo analysis, Repeat Scout, a module of RepeatModeler software was used to identify a set of repeat elements that was further utilized by Recon, another module of the same software to generate a classified consensus repeat library. The consensus library was then merged further with custom library prepared by using some major repeat classes viz., Grasshopper, Maggy, MGL, Mg-MINE, SINE, Pot2, MGRL-3, Mg-SINE, Ch-SINE, Occan, Pot3, Pyret, RETRO 5, RETRO 6, RETRO 7, and Cluster 1-9. The merged repeat library was subjected to RepeatMasker 4.0<sup>7</sup> for homology based masking of the repeat regions in targeted Magnaporthe oryzae genome sequences.

# Analysis of Structural Variants

Structural variants such as single nucleotide polymorphic sites (SNPs), Insertion-Deletion (InDel), and genome duplications were performed by MUMmer 3.23 (Kurtz et al., 2004), Samtools<sup>8</sup> and CLC Genomics Workbench 6.5 (CLC Bio, Aarhus, Denmark). Default parameters were kept for all these analyses.

# Gene Prediction and Functional Genome Annotation

Gene was predicted for each genome separately using the masked resulted sequences in the RepeatMasker analysis. The gene prediction was carried out using FGENESH 2.6 module of Molquest<sup>9</sup> . The prediction was based on the trained dataset of Magnaporthe available with the FGENESH module. The predicted genes were used for gene ontology and assigned their respective functional categories. Transfer RNA-coding regions were searched using tRNAscan-SE (Lowe and Eddy, 1997).

## Secretome Analysis

Secretory proteins were identified by a pipeline developed in this study. All the proteins of a genome were initially subjected to

<sup>1</sup>http://www.bioinformatics.babraham.ac.uk/projects/fastqc/

<sup>2</sup> ftp://ftp.ncbi.nlm.nih.gov/pub/TraceDB/magnaporthe\_grisea

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

<sup>4</sup>www.ncbi.nlm.nih.gov/genome/

<sup>5</sup>https://www.ncbi.nlm.nih.gov/Traces/sra/?view=software

<sup>6</sup>http://repeatmasker.org/RepeatModeler.html/

<sup>7</sup>http://www.repeatmasker.org/RMDownload.html

<sup>8</sup>http://samtools.sourceforge.net/

<sup>9</sup>http://www.molquest.com/

SignalP 4.1 (Petersen et al., 2011) and TargetP 1.1 (Emanuelsson et al., 2007) analysis. Results of both TargetP (Loc = S) and SignalP (D-score = Y) analyses were combined and then passed to TMHMM 2.0<sup>10</sup> for scanning transmembrane spanning regions in these proteins. All proteins with 0 TMs or 1 TM predicted through TMHMM if located in the first 40 amino acids were kept for further analysis. All GPI-anchor proteins were identified by the big-PI<sup>11</sup>. At this step, all proteins resulted from TMHMM and big-PI were also used to predict their localization within the cell by ProtComp 9.0 using the LocDB and PotLocDB databases<sup>12</sup>. Finally, refined secreted proteins were analyzed by running "runWolfPsortSummary fungi" of WoLF PSORT 0.2 (Horton et al., 2007). After getting refined secreted proteins, the functional annotation of these refined proteins was performed via Interproscan 5-RC-7 (Zdobnov and Apweiler, 2001). All the analyses were performed with default parameters.

# Gene Ontology and Gene Function Categorization

Gene ontology and its functional categorization were performed by using a set of softwares, BLASTX (Altschul et al., 1990), Interproscan 5-RC-7 (Zdobnov and Apweiler, 2001), and Blast2Go (Conesa et al., 2005). Top 10 hits of BLASTX results and Interproscan results in xml format were supplied for Blast2Go analysis. The obtained results of Blast2GO analysis were considered only for biological process category and further summarized them into 11 broader biological sub-processes.

# Comparative Analysis of Five M. oryzae Genomes

Five M. oryzae genomes were considered for comparative genome analysis. The whole genome sequences of two strains RML-29 and RP-2421 were generated in the current study and genome sequences of two strains P131 and Y34 was downloaded from NCBI<sup>13</sup> along with the reference genome sequence of the strain 70-15 selected for comparative genome analyses.

# Orthologs Identification and Genome Phylogeny

Orthologous identification in all the five genomes was done using OrthoMCl 1.4 (Li et al., 2003). The single copy ortholog was chosen for phylogenetic analysis. The distance matrix was calculated using YN00 method of PAML 4.8 (Yang, 2007). The tree was generated by the Neighbor Joining method of Phylip 3.695 package (Felsenstein, 2005). The synonymous and nonsynonymous calculation was done with KaKs Calculator (Zhang et al., 2006) and CodelML of PAML 4.8 (Yang, 2007).

The results obtained in various analyses of all the five Magnaporthe genomes were visualized using Circos 0.68 (Krzywinski et al., 2009), and SyMap 4.2 (Soderlund et al., 2006).

# RESULTS

# Genome Sequencing, Quality Check, and Assembly

Two strains of M. oryzae, RML-29, and RP-2421 were selected for whole genome sequencing project. This selection was based on the virulence spectrum of these isolates on differential blast resistance monogenic rice lines (**Supplementary Table 3**). RML-29 was found to be avirulent on 19 of total 25 rice blast resistance monogenic lines (**Supplementary Table 4**). While only 5 of these lines showed resistance reactions to the isolate RP-2421 (**Supplementary Table 4**). Rice lines Lijiang Xintuan Heigu (LTH) and Taipei-309 were used as susceptible controls in the pathotyping experiments. Based on this study these were characterized as avirulent (RML-29) and virulent (RP-2421) isolates. Whole genome sequencing was performed by using NGS technology. Total 1,429,817 and 452,970,034 raw reads of RML-29 and RP-2421, respectively were subjected for quality check and trimming of bad quality sequences (<20 Phred value). In this quality check, over 92% of high quality (HQ) reads and total 39,430,338,647 HQ bases present in HQ reads were obtained from the RP-2421 genome (**Supplementary Table 5**). Whereas, 99.99% nucleotides were passed as HQ bases in RML-29.

High quality combined sequences of paired reads for RML-29 and RP-2421 were 542.28 Mb (454 FLX, Roche) and 39.78 Gb (Illumina), representing ∼13.2- and ∼969.7-fold genome coverage, respectively and used for the de novo assembly (**Table 1**). Initially 1,429,684 and 416,663,802 total reads of RML-29 and RP-2421, respectively were taken for the assembly (**Table 1** and **Supplementary Figures 1A, 2A**). In this assembly, 6934 and 13347 total contigs were generated from Roche (RML-29) and Illumina (RP-2421) data, respectively (**Table 1** and **Supplementary Figures 1B,C, 2B,C**). These contigs were mapped back on the reference genome sequence of M. oryzae isolate 70-15 (version 8, Magnaporthe Comparative Sequencing Project, Broad Institute of Harvard and MIT). Total 5363 (86.92% of total bases) and 5280 (90.69% of total bases) contigs of

TABLE 1 | Summary of RML-29 and RP-2421 genomes' assemblies.


<sup>10</sup>http://www.cbs.dtu.dk/services/TMHMM/

<sup>11</sup>http://mendel.imp.ac.at/gpi/cgi-bin/gpi\_pred\_fungi.cgi

<sup>12</sup>http://www.softberry.com

<sup>13</sup>www.ncbi.nlm.nih.gov/genome/

RML-29 and RP-2421, respectively were mapped to the reference sequences (**Table 1** and **Supplementary Table 6**). The N50 and largest contig's size of RML-29 were 10.40 and 86.48 kb, respectively, while these values were 35.35 and 239.05 kb obtained in RP-2421 (**Table 1**). Average contig's size was higher in RML-29 than that of RP-2421 (**Table 1**). However, 2.03 Mb (13.08% of total bases) and 4.46 Mb (9.31% of total bases) regions of RML-29 and RP-2421, respectively were not mapped to the reference genome (**Table 1** and **Supplementary Table 6**). These un-mapped regions were called as unique or isolate specific sequences of corresponding M. oryzae isolate. Assembled sequences of RP-2421 and RML-29 genomes were submitted in the genome databank of NCBI<sup>14</sup>. These M. oryzae isolates were also deposited at Microbial Type Culture Collection and Gene Bank (MTCC), Institute of Microbial Technology (IMTECH), Chandigarh (**Supplementary Table 7**).

# Pan-Genome Construction and Presence and Absence Variations (PAVs) Analysis

A total 54 Mb sized pan-genome was constructed using whole genome sequences of 12 different Magnaporthe isolates. This Magnaporthe pan-genome contained 15287 genes. The PAV events analysis in all the isolates showed that ASM80585v1 and FJ81278 contained maximum and minimum PAVs, respectively (**Table 2**). BR32 contributed 4.44 Mb consensus sequences to the pan-genome in the form of 970 PAVs, which was the longest isolate-specific region among the isolates. This isolate also had the largest number of isolate-specific genes (**Table 2**). Only three isolates, RP-2421, BTGP6F, and ASM80585v1 have more than 200 isolate-specific genes, while the average number of isolatedspecific genes were 167.33. Average length of PAVs ranged from 0.84 kb (in RP-2421) to 4.65 kb (in BTMP13\_1). Overall average length of PAVs was 2.68 kb. The maximum number of isolatespecific genes were obtained in BR32, while the least number was found in US71. The isolate-specific gene density (genes per Mb) was maximum (1041.67) in RP-2421 which was very high

TABLE 2 | List of presence and absence variations (PAVs) in different Magnaporthe isolates.

<sup>14</sup>https://submit.ncbi.nlm.nih.gov/subs/genome/

compared to the isolates, where as other three isolates, FJ81278, RML-24, and US71 had over 200 gene density. Hence, BTJP4\_1 had only 90.94 isolate specific gene density that was the lowest among the isolates.

# Synteny Analysis of Virulent and Avirulent Strains

Comparison between the mapped regions of RML-29 and RP-2421 was performed to find the order of short nucleotide sequences and level of similarity between them. A dot plot of all the 8 chromosomes was compared between RML-29 and RP-2421 genomes (**Supplementary Figure 3**). 1129 total hits were obtained in this synteny analysis, and Pearson correlation coefficient (Pearson R) values for all the hits were found greater than 0.99. Pearson R measures linear association between two variables ranges from +1 to −1. Out of 1129, 249 hits were matched to the chromosome 1 and two short regions 138.75 kb of the chromosomes from RP-2421 and RML-29, respectively were separately aligned to see a clear view of the regions with matching DNA fragments of the first genome to the second genome along with the matched DNA fragment orders positioned in the aligned region (**Figure 1**). However, the rest of the chromosome 2, 3, 4, 5, 6, 7, and 8 generated 209, 175, 158, 104, 109, 81, and 44 hits, respectively between the genomes.

# SNPs and InDels Analyses of M. oryzae Genomes

Structural variations were determined in both RML-29 and RP-2421 genomes by comparing their sequences with the reference genome separately. In both the genomes 29776 and 35332 single nucleotide variations (SNVs) were identified from RML-29 and RP-2421, respectively (**Supplementary Table 8** and **Figure 2**). Insertions-deletions (InDels) and replacement events were found 4.4 and 1.2 times higher in RML-29 than RP-2421, respectively (**Supplementary Table 8**). The multi nucleotide variations (MNVs) were obtained more in


<sup>∗</sup>PAVs identified in respective isolate against the Magnaporthe pan-genome.

RP-2421 than RML-29. SNVs were found more in all the chromosomes of RP-2421 than RML-29, except chromosome no. 1 (**Supplementary Table 8**).

# Repeat Element Analysis

Repeat elements were identified from both RML-29 and RP-2421 genomes. Total 75 and 94 classified consensus repeat sequences obtained in the de novo analysis were used for custom library preparation of RML-29 and RP-2421, respectively. The custom libraries were further exploited to search homology based repeat elements from these strains. Cluster 1–9 repeat elements were considered to include in this analysis by following a previous study (Xue et al., 2012). The nucleotide sequences 1579 (total size = 43,037,792 bp) and 8075 (total size = 45,510,614 bp) of both RML-29 and RP-2421 isolates, respectively were subjected for homology based repeat identification. Similarly, all other isolates included in the pan-genome analysis were also subjected for the repeat masking in the genome sequences (**Table 3**). The highest level of repeats were masked in BTGP1b genome (18.50% of the genome), whereas the least level were detected in P131 with 3.15% of the genome. The maximum copy number of SINE was found in RP-2421 (2029), which was very high in all the 15 isolates (**Table 3**). Total nucleotide bases excluding "N" were counted as 42,186,663 and 45,510,614 bp for RML-29 and RP-2421, respectively. Total 11.78% and 12.28% sequences were masked in this analysis for RML-29 and RP-2421, respectively. A total of 6995 interspersed repeats was identified in RP-2421, while a lower number of the repeats 4656 were obtained in RML-29 (**Table 4**). These repeats covered 10.72 and 11.03% of total sequences of RML-29 and RP-2421 genomes, respectively (**Table 4**). A sum of 2029 short interspersed repeat elements (SINEs), including 265 unclassified, was detected in RP-2421 strain that was a significantly greater number than that of 79 SINEs recognized from RML-29 (**Tables 4**, **5**). Pot2/Pot4 transposable element showed same frequency in both the genomes, however, a combined copy number of all three RETRO elements were found 1.29 times less in RP-2421 than RML-29 (**Table 5**). Except Occan and SINE, all other DNA transposon, LTR retrotransposon and LINE were existed in higher copy number in RML-29 than RP-2421 (**Table 5**). The total frequency of Cluster 1–9 repeats were also found higher in RML-29 than RP-2421 (**Table 5**). Distribution of repeat elements in both the genomes was following a similar trend across the entire chromosome, except the unique regions of the genomes (**Figure 2**). Over 6 times more repeat elements were identified from RP-2421 than RML-29 in the unique regions (**Supplementary Figure 4**).

Transfer RNA (tRNA) prediction analysis was also conducted in both RML-29 and RP-2421 genomes. A total of 254 tRNA, excluding one undefined was recognized from RML-29, whereas a lower number (203) of this RNA was detected from RP-2421. In addition, 141 pseudo tRNAs were also identified from RP-2421, but none was identified from RML-29.

red, pathogenesis; gray, physiological; yellow, protein synthesis; very light red, repeat elements; purple, response to stress; very light green, transportation, and signaling; black, unknown. Repeats and SNPs denote in percentage and in 100 kb windows, respectively. SDs means the segmental duplications present intra- and inter-genome level.

# Gene Prediction and Identification of Isolate Specific Genes

All the mapped (Chromosome I–VIII) as well as un-mapped (Unique) sequences of both the strains RML-29 and RP-2421 were employed for gene prediction using M. oryzae (trained dataset) as reference available in FGENESH software. Gene models 13297 and 13623 were predicted from RML-29 and RP-2421 genomes, respectively. Genes that encoded amino acid (aa) below 50 were excluded from both the genomes. Hence, 12746 and 12957 genes of RML-29 and RP-2421, respectively encoded amino acid equal or above 50 in length were kept for further analyses (**Supplementary Table 9**). Chromosome wise gene density distributions for RP-2421 genome is given in **Figure 2**. Average length of protein deduced from RML-29



<sup>∗</sup>Percentage values of respective genome; SINE, Short interspersed nuclear element.

and RP-2421 were 500.90 and 510.97 aa, respectively. The 6562 and 6586 aa encoding genes were found largest gene of RML-29 and RP-2421, respectively. Isolate specific genes were also predicted from unique sequences of RML-29 and RP-2421 strains (**Supplementary Table 9**). RP-2421 contained unique sequences over 2 times of RML-29 that yielded isolate specific genes 1.38 times of RML-29 (**Supplementary Table 9**).

Functional annotations of RML-29 and RP-2421 genomes were performed to assign the function of each gene present in the genomes using a nonredundant (nr) data of NCBI-BLASTX search. Top ten hits of every individual gene were selected for categorization into 11 functional groups based on biological process, namely, growth and development (GD), hypothetical (HY), miscellaneous (MS), nucleic acid metabolism (NM), pathogenesis related (PA), physiological traits (PH), protein synthesis (PS), repeat element related (RE), response to stress (RS), transportation and signaling (TS), and unknown function or no hit found (UN). Approximately 65% of total RML-29 genes were functionally annotated, whereas around 35% of the genes were belonged to HY and UN categories (**Supplementary Table 10**). However, only 36.65% of isolate specific genes of RML-29 were functionally characterized and rest of 63.35% of the genes is still needed to be characterized and they were fallen into HY and UN groups (**Supplementary Table 10**). Total 66.42% of genes that predicted from mapped regions of the RML-29 genome, excluding unique genes or isolate specific genes were functionally grouped into all the categories, except HY and UN (**Supplementary Table 10**). Similarly, 66.49% of total genes, excluding RP-2421 isolate specific genes were found to disperse into all the functional groups, except HY and UN in functional annotation of RP-2421 genome (**Supplementary Table 11**). Unique genes of RP-2421 were also functionally annotated only 45.26% of the genes and remaining 54.74% were still required to be characterized initially at in silico level (**Supplementary Table 11**).

Automated functional annotation was also accomplished for both RML-29 and RP-2421 genomes using Blast2GO, a web based package. In this process, 6439, 6378, and 3659 genes of RML-29 were functionally annotated into three major gene ontology (GO) groups, molecular function (MF), biological process (BP), and cellular component (CC), respectively (**Supplementary Table 12**). These major groups, BP, MF, and CC were further descended into 12, 9, and 6 sub-groups, respectively in functional annotation of RML-29 genome (**Supplementary Figure 5**). Thus, functional annotation based on biological process (BP) accounted 50.04% of total RML-29 genes and maximum contribution in this annotation was of cellular process and followed by a metabolic process (**Supplementary Figure 6**). Similarly, three main categories of GO, BP, MF, and CC were also divided into 16, 12, and 7 sub-groups, respectively in RP-2421 genome (**Supplementary Table 12** and **Supplementary Figure 7**). With 55.58% of total RP-2421 genes belong to biological process category. Number of genes and their sub-groups categorized on


LTR, Long terminal repeat; LINE, Long interspersed nuclear element; SINE, Short interspersed nuclear element.

TABLE 5 | Copy number of some major repeat elements identified from M. oryzae strains.


LTR, Long terminal repeat; LINE, Long interspersed nuclear element; SINE, short interspersed nuclear element.

the basis of biological process (BP) present in RP-2421 are given in **Supplementary Figure 8**.

# Confirmation of Isolate Specific Regions

PCR amplification was performed using total 30 isolate specific primer pairs, which were specific to RML-29 and RP-2421 (**Supplementary Table 1**). The specificity of isolate sequences were validated with these primer sets using genomic DNA of both the strains (**Figure 3**). All the primer sets were worked perfectly as per the expectation since R1-15 primer sets showed bands only with RML-29 genomic DNA and not with RP-2421. Similarly, P1–15 primer sets, except P9 and P11 amplified PCR products only from RP-2421 genomic DNA and not with RML-29. These P9 and P11 primer sets did not show amplification from any of the strains. Approximately 93% of primer sets (28 of 30) were working fine and validated with amplification from genomic DNA of RML-29 and RP-2421.

# Identification of Candidate Effectors From Magnaporthe oryzae

For the identification of candidate effectors, secretome analysis was performed to discover putative candidate effector molecules from both RML-29 and RP-2421 genomes. To conduct this analysis, a pipeline was developed and used for the identification of secreted proteins from these two genomes (**Figure 4**). Proteins obtained in each of the different steps of this analysis were almost alike in number in both the genomes (**Supplementary Table 13** and **Figure 5**). This similarity was also followed at the level of functionally categorized genes (**Figure 6**). Thirty-five categories were made to group all the refined secreted proteins using BLASTP analysis. Based on the examinations of the gene content, 513, and 707 genes, respectively, were distinctive to RML-29 and RP-2421. Overall, 3.70% and 5.52% of these isolate specific genes encoded secreted proteins, and 57.89% and 48.72% of them had no significant homology in the genbank.

# Comparative Analysis of Five M. oryzae Genomes

For comparative study, three additional genomes of M. oryzae strains, 70-15, P131, and Y34 were included along with RML-29 and RP-2421. Assembled sequences of five genomes were compared with Mauve software. Comparisons of all the five genomes at the chromosome level were performed by Symap software. More clear view of the alignments were exported using Mauve software and a representative of these alignments, chromosome VIII (pseudo-chromosome of M. oryzae) is given in **Figure 7**. We found conserved blocks (25040) in this alignment with more than 90% similarity. A progressive alignment guided tree of all the five strains was generated by this analysis (**Supplementary Figure 9**). RML-29 strain was more closely associated with 70-15 than RP-2421, this grouped into a single cluster; while P131 and Y34 formed a separate group. In the phylogenetic tree, P131 was most divergent among the five strains of M. oryzae and 70-15 was least divergent.

The segmental duplication (SD), repeat, SNPs and gene prediction analyses were also performed for all these five genomes (**Supplementary Table 14**). Density per 100 kb windows of SD, SNPs and repeats were plotted in **Figure 2**. All the SD containing at least 1000 bp genomic fragment and having more than 90% similarity with different fragment within the genome are given in **Supplementary Table 14**. Maximum number of SD was obtained from 70-15 (218), followed by RP-2421 (181), RML-29 (153), Y34 (119), and P131 (72). In comparative repeat analysis, occurrence of SINE was estimated for the five isolates and its frequencies were calculated in all the genomes (**Supplementary Table 15**). None of the isolates, except RP-2421 contained over 150 copies of this retro transposon in their genomes.

Gene families were identified from these isolates using OrthoMCL program. A sum of combined 61598 proteins, 12746, 12957, 12433, 11672, and 11790 predicted from all the five strains, RML-29, RP-2421, 70-15, P131, and Y34, respectively were subjected to clustering analysis of orthologous groups. In this analysis, 11883 orthologs groups or clusters were formed by using 61598 proteins (**Supplementary Table 16**). In the five genomes 10291 clusters obtained which were divided into various groups (**Supplementary Table 16**). Distribution of orthologous clusters was also plotted in Venn diagram and 5, 13, 2, and 3 isolate specific clusters were obtained from RML-29, RP-2421, 70- 15, and Y34, respectively (**Figure 8**). There was no isolate specific

FIGURE 3 | Validation of isolate specific regions in M. oryzae. PCR amplification from (A) RML-29 and (B) RP-2421 using 15 primer sets derived from each strain specific sequences represent strain specificity by presence and absence of bands in the respective strains. M- Molecular weight marker, R1-15 primer sets specific to RML-29, and P1-15 primer sets specific to RP-2421.

cluster in P131 genome (**Figure 8**). From all the five genomes 10013 orthologous groups formed consisting of five genes in each group (**Supplementary Table 16**).

# Genome-Wide Evolutionary Analysis

Genome-wide evolutionary study was performed to identify single copy conserved orthologs by analyzing nonsynonymous (Ka) and synonymous (Ks) nucleotide substitutions. In total, 10,013 genes from each of the five genomes were used to identify nucleotide substitutions. All the five genomes RML-29, RP-2421, 70-15, P131, and Y34 were subjected to multiple alignment to make sequence length of same size for further evolutionary study. Estimation of Ka and Ks was performed by YN00 method of PAML package. Total 10 pairwise combinations were made

by using all the five genomes (**Table 6**). RML-29 and RP-2421 genome pair was found to result maximum (10 times to others) value 1.02324 for the Ka/Ks ratio. Maximum (2.18 Million) and minimum (0.51 Million) substitutions, including both Ks and Ka were obtained in 70-15 vs. RML-29 and RML-29 vs. RP-2421 genome pairs, respectively (**Table 6**). This trend was consistent for divergence time estimation, and the earliest (172,952 years back or 0.173 million years back, MYB) and the latest (0.037 MYB) divergence times among the genome pairs were estimated for the genome pairs of 70-15 vs. RML-29 and RML-29 vs. RP-2421 (**Table 6**).

Phylogenetic relationship based on 10013 conserved orthologs genes was carried out using Phylip software. As expected, RML-29, RP-2421, and 70-15 M. oryzae strains formed a separate group and a different cluster was obtained for P131 and Y34 Chinese strains (**Figure 9**). The strain RP-2421 was more closely related to 70-15 reference strain than RML-29.

# DISCUSSION

In various eukaryotic organisms, comparative study of many genomes of a particular species has been employed to refine the genome assembly, identification and annotation of the genes, and detection of structural variations like repeats, SNPs, InDels, haplotypes, etc. in the genome (Kellis et al., 2003, 2004; Novo et al., 2009; Andersen et al., 2011). M. oryzae is well-known for its frequent natural genetic variant occurrence in the field conditions, and the consequence of this is the emergence of new races and disease outbreaks in the rice growing areas of the

world (Valent and Chumley, 1991; Talbot, 2003). Comparative genomics analysis of M. oryzae also help to understand molecular aspects of the emergence of new virulence mechanism of the pathogen, for proper blast disease control (Farman, 2002; Talbot, 2003). In the present study, two field isolates of M. oryzae; RP-2421, and RML-29 from India were compared at the genome

level, which were virulent and avirulent, respectively against most of the characterized rice blast resistance genes. Genome analysis of five different M. oryzae isolates indicated that the two Indian field isolates (RP-2421 and RML-29) are more closely associated with 70-15 than to the Chinese field isolates (P131 and Y34). However, 70-15 is a laboratory made strain developed from the backcrosses of rice infecting isolate Guy11 with a sibling of a cross related to a nonrice infecting isolate that infects weeping love grass. The overall genome content and composition are similar among these five isolates, but the genome sizes of RP-2421 and RML-29 with only ATGC contents without N's were slightly higher than that of 70-15. Previously, B157, MG01, P131, and Y34 isolates of M. oryzae were also reported to have slightly larger genome size than 70-15 (Xue et al., 2012; Gowda et al., 2015).

The present study revealed that both RML-29 and RP-2421 isolates had some unique isolate specific genomic DNA sequences. The isolates RML-29 and RP-2421 had 513 and 707 unique genes, respectively. While, the pan-genome analysis showed a great reduction in the isolate specific genes in these isolates, because the pan-genome was constructed from 13 different isolates' information, including reference genome 70- 15. RML-29 had only 44 isolates specific genes, however 375 such genes were detected in RP-2421. Interestingly, RP-2421 possessed the highest isolate-specific gene density (1041.67 genes/Mb region, 735 genes in 0.36 Mb region) across the analyzed 12 different Magnaporthe isolates. Overall, more than 3% of these unique or isolate specific genes encoded secretory proteins and over 48% of them had no substantial homology in the genbank. These unique genes play diverse roles in the adaptation process of the individual isolate, even so, some of which might perhaps attribute to the specificity of distinct isolates (Xue et al., 2012). Several isolate specific genes in the rice blast fungus


TABLE 6 | Estimation of divergence time by calculating the synonymous (Ks), nonsynonymous (Ka) and their ratio (Ka/Ks) from five M. oryzae strains.

<sup>∗</sup>YN00 method of PAML software; ∗∗P-value (Fisher); M, Million or Mega; Ks, Synonymous; Ka, nonsynonymous; Subs, Substitutions; D-T, Divergence-Time; MYB, Million years back.

have been reported in the previous studies (Yoshida et al., 2009; Xue et al., 2012; Chen et al., 2013; Dong et al., 2015; Gowda et al., 2015), suggesting that the isolate specific genes might be lost or gained during the course of evolution (Xue et al., 2012). Some of the unique genes identified in the present investigation show no hits with the NCBI nonredundant database. Similar results were also found by Xue et al. (2012), and they reported that P131 and Y34 isolate specific genes had no homology with the genes present in the NCBI database. Many authors also claimed that these unique genes might have crucial roles in plant infection process. Moreover, gain or loss of the effector gene is often associated with the unstable telomeric regions of the chromosome (Yoshida et al., 2009). It is thus hypothesized that the presence of isolate specific regions at the chromosomal ends act as a source of new effectors to augment genome

evolution of M. oryzae (Dong et al., 2015). The presence or absence of avirulence genes may be directly related to the adaptation process of M. oryzae (Chuma et al., 2011) and such hypotheses suggest that additional genome decoding of various M. oryzae isolates is necessary for characterizing rice infecting strains, as crucial information can be revealed only by surveying outside the "core" genomes (Yoshida et al., 2009). The extensive chromosomal shuffling in asexual reproduction of fungus Verticillium dahlia is a general mechanism to make lineage specific regions that supply new effectors for the fungal adaptation (de Jonge et al., 2013). Although sexual reproduction in M. oryzae is very rare event reported in the field conditions, but it can be facilitated under laboratory conditions as well (Dong et al., 2015). Such asexually reproducing organisms are often considered to be less flexible in the adaptation process than sexual organisms (Burt, 2000; McDonald and Linde, 2002). The expansion of the genome, possibly signifies a paradigm of evolutionary exchanges, as the price of upholding the additional DNA fragment is compensated by the functional benefits it provides (Raffaele and Kamoun, 2012). However, Dong et al. (2015) hypothesized that horizontal gene transfer (HGT) might be another event that provides the isolate specific genes to operate the evolutionary process in M. oryzae. It is known that HGT has a role in the gaining of new genes and functions (Raffaele and Kamoun, 2012).

In the present investigation, the comparative study of five M. oryzae isolates; 70-15, RML-29, RP-2421, P131, and Y34, over 100 kb regions scattered in all the 8 chromosomes were found conserved as SD. Duplication is one of the key mechanisms for evolutionary process in the organisms. Xue et al. (2012) also found duplicated sequences in a comparative study of three isolates of M. oryzae, 70-15, P131, and Y34 and these sequences seemed to be enriched in the telomeric regions of all the 7 chromosomes. In this study, both inter- and intra-chromosomal duplications were examined, however, higher inter-chromosomal duplications was evident, and only a small part of duplication events were sustained in all the three isolates. They also reported that the duplicated sequences also code for many genes, involved

in interactions with the host plant. In the present study a total of 10291 clusters from all the five genomes, 70-15, RML-29, RP-2421, P131, and Y34 were found and these clusters were considered to contain a core gene set for M. oryzae genome. The core set of genes obtained from clustering analysis of 61598 genes, found 10013 homologous groups having single copy gene and they are highly conserved across all the five genomes of M. oryzae. The cumulative length of all these core genes from all the genomes has potential to estimate the evolutionary forces acted throughout the genome and make them enable to compare with each-other. It is considered that genes having role in the virulence of phytopathogen may be associated with a quick array of evolution for acclimatization to different environments or hosts (Chen et al., 2013). To evaluate gene evolution rate, the nonsynonymous and synonymous (Ka/Ks) ratios of nucleotide substitution in 10013 single copy orthologous genes analysis was performed, in the present study from all the five strains were estimated and found that all the 10 strain pairs, except two are under positive selection pressure. The exceptional pairs 70- 15 -RP-2421 and RP-2421 – P131 sustain against the action of resultant purifying force, because the Ka/Ks ratios were less than 1 and very close to 1 in both the pairs, i.e., 0.998060 for 70- 15 – RP-2421 and 0.998993 for RP-2421 – P131. The positive evolution pressure was also reported on many genes of several field isolates, FJ81278, HN19311, MG01, and B157 (Chen et al., 2013; Gowda et al., 2015).

Effectors are considered as basic pathogenicity determinants that alter plant innate immunity and assist in disease development during plant-pathogen interactions (Kamoun, 2007). Many Avr genes have been earlier cloned and characterized by genetic association analysis, map-based cloning, or loss-offunction approaches (Kang et al., 1995; Sweigard et al., 1995; Orbach et al., 2000; Ahn et al., 2004; Kim et al., 2005; Yoshida et al., 2009). Most of these effectors are small proteins and are secretory in nature. They do not show much homology to known proteins (Ellis et al., 2009; Stergiopoulos and de Wit, 2009; Valent and Khang, 2010). In secretome analyses of the isolates RML-29 and RP-2421 genomes, we obtained 669 and 675 refined candidate effectors, respectively. Of these candidate effectors, approximately 48% of them in both the genomes were not functionally defined and show hypothetical or unknown functions. Diverse nature of characterized effectors is well-known and their least homology with previously identified proteins indicate that further characterization of individual candidate effectors is very essential for molecular study of the host-pathogen interaction.

Since sexual reproduction in M. oryzae is a very rare event in natural conditions and genetic variation through asexual reproduction is not possible, still isolate specific genomic sequences are reported in many studies (Dean et al., 2005; Yoshida et al., 2009; Xue et al., 2012; Chen et al., 2013; Dong et al., 2015; Gowda et al., 2015). Translocations of the repeat elements may possibly be one of the key factors for genome variation as well as rapid adaptation of the pathogen to different hosts and environments (Xue et al., 2012). A gain of virulence is often related to the translocation of repetitive elements that generally inactivate or delete PAMP- (pathogen associated molecular pattern) encoding genes whose products trigger the innate plant defense mechanism (Kang et al., 2001; Farman et al., 2002). Thus, understanding the molecular biology of the repetitive elements in the rice blast fungus not only offers an insight into their effects on genome evolution process, but also sheds light on the mechanisms associated with pathogenic variation found at the strain level (Dean et al., 2005). Consistent with this assumption, over 10% of the genomes RML-29 and RP-2421 were covered with repetitive sequences in our study. Similar content of presence of repetitive elements in the genomes of M. oryzae were reported in the previous studies as well (Xue et al., 2012; Gowda et al., 2015). However, slightly lower percentage of the repetitive elements was documented in the genomes of 70-15 and 98-06 isolates (Dean et al., 2005; Dong et al., 2015). In present study, the maximum genome sequences masked with repeats was found in BTGP1b (18.50% of genome) and higher percentage of repeats were identified in all the long reads assembled genomes, except BTJP4\_1 (BR32, BTGP1b, BTGP6F, BTMP13\_1, CD156, FJ81278, FR13, US71) than the short reads assembled genomes (ASM80585v1, FJ81278, P131, Y34, RP-2421, RML-29). One thing is very clearly we observed by our repeat analysis of 15 different Magnaporthe isolates and other published reports (Dean et al., 2005; Xue et al., 2012; Gowda et al., 2015; Dong et al., 2015) that the varying degree of repeat occurrences in the genome of Magnaporthe has role in the pathogen survival and in the disease causing ability of the pathogen. Varying copy numbers of different repeats occur in RML-29 and RP-2421 isolates. Surprisingly, the short interspersed nuclear elements (SINEs) show high difference in its copy number distributions between RP-2421 and RML-29. The distribution of this repeat is approximately 25 times more frequent in RP-2421 than that of RML-29. The high copy number of SINEs in M. oryzae genome was also reported by Kachroo et al. (1995). They suggested two possible reasons for the high SINE copy numbers in M. oryzae genome. One of the reasons is a fusion or insertion of the repeat elements in the genomic region of M. oryzae followed by elimination of a large part of the inserted sequence from the genome and second reason is recombination event occurring between the repeat sequence and the genomic sequence of M. oryzae. Premature termination of reverse transcriptase could be a main cause for the generation of SINE in M. oryzae genome as well as in other eukaryotic genomes (Kachroo et al., 1995). Further, in Magnaporthe, SINEs were found to evolve before the divergence of the two host-specific forms, since these are found in both M. oryzae and M. grisea (Kachroo et al., 1995). The amplification and extensive occurrence of the SINE all over the genome may have foremost impact on structural and functional genomics in M. oryzae and could possibly be one of the major mechanisms by which the pathogenic variability and adaptability are generated among the isolates of M. oryzae (Kachroo et al., 1995). Similar hypothesis was also given by Xue et al. (2012) that transpositions of repeat elements (TEs) may play an important role in several adaptation processes of the rice blast fungus during host-pathogen interaction, such as transcriptional regulations, duplication of genomic fragments, modulating the genomic content via the addition or deletion of isolate specific sequences.

# CONCLUSION

fmicb-10-00966 May 6, 2019 Time: 12:57 # 15

Molecular understanding of repeat elements in the rice blast fungus not only offers an insight into their effects on genome evolution process, but also sheds light on the mechanisms of pathogenic adaptation attained by different strains. In the pan-genome analysis, we found several isolate-specific genes (ranging from 19 to 554) identified from each of the 12 different Magnaporthe isolates that might be gained or lost during the course of evolution and might have roles in their pathogenicity events and their host-pathogen interactions. RP-2421 possessed the highest isolate-specific gene density (1041.66 genes/Mb region) among the isolates. Higher number of pathogenicity related genes annotated in RP-2421 (154) than RML-29 (145) might justify the correlation between pathogenicity related genes and virulence spectrum of a M. oryzae isolate. Also a very high copy number (2029) of short interspersed repeat element (SINE) were found in RP-2421 in comparison with RML-29 (79), and they might have direct correlation with the virulence variability within the isolates. More than six hundreds candidate effectors were identified from both RML-29 (669) and RP-2421 (675) through secretome analysis and most of them (approximately 48%) could not be annotated. For better understanding about the strain specific behavior, in term of pathogenicity, these unknown candidate effectors should be functionally validated. In gene evolution rate analysis, 10013 single copy orthologous genes of five strains, RML-29, RP-2421, 70-15, P131, and Y34 were estimated by pairwise comparison of the strains and we found that all the strain-pair combinations, except two are under positive selection pressure. The resources generated and the knowledge gained through the present comparative study of virulent and avirulent filed isolates of M. oryzae can give us

# REFERENCES


understandings about the genome variation process in the fungus under different environmental conditions, and will help in the effective management of rice blast disease.

# AUTHOR CONTRIBUTIONS

TS conceived and managed the project. PS and TS wrote and revised the manuscript. AM and PJ helped in the data analysis. RR provided fungal cultures. VS gave input in the manuscript preparation. All authors read and approved the final manuscript.

# FUNDING

This research was supported by funding from National Agricultural Innovative Project (NAIP), Indian Council of Agricultural Research, New Delhi, India. The Funders had no role in study design, the collection, analysis and interpretation of data, or writing of the manuscript.

# ACKNOWLEDGMENTS

PS thanks Dr. Archana Kumari for her help throughout the run of the Illumina sequencing. He is also grateful to Mr. Jatesh Sachadeva (Premas Biotech Pvt. Ltd.) for his technical assistance during the sequencing experiments. TS is thankful to the Department of Science and Technology, Govt. of India for JC Bose National Fellowship.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2019.00966/full#supplementary-material

and potential virulence effectors. BMC Genomics 14:887. doi: 10.1186/1471- 2164-14-887



may contribute to parasite fitness and defeat of plant resistance. Plant Cell 18, 2402–2414. doi: 10.1105/tpc.106.043307


Leptosphaeria maculans is affected by genomic environment and exposure to resistance genes in host plants. PLoS Pathog. 6:e1001180. doi: 10.1371/journal. ppat.1001180


**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 © 2019 Singh, Mahato, Jain, Rathour, Sharma and Sharma. 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.