EDITED BY : Minghui Lu, Yule Liu, Jie Zhou, Hanjo A. Hellmann, Wei Wang and Sophia Stone PUBLISHED IN : Frontiers in Plant Science

#### Frontiers Copyright Statement

© Copyright 2007-2018 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-558-4 DOI 10.3389/978-2-88945-558-4

#### About Frontiers

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

#### Frontiers Journal Series

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

#### Dedication to Quality

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

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

#### What are Frontiers Research Topics?

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

# PROTEIN QUALITY CONTROLLING SYSTEMS IN PLANT RESPONSES TO ENVIRONMENTAL STRESSES

Topic Editors:

Minghui Lu, Northwest A&F University, China Yule Liu, Tsinghua University, China Jie Zhou, Zhejiang University, China Hanjo A. Hellmann, Washington State University, United States Wei Wang, Henan Agricultural University, China Sophia Stone, Dalhousie University, Canada

Environmental stress factors negatively affect plant growth by inducing proteins dysfunction. As coping strategies, plant have developed a comprehensive protein quality controlling system (PQCS) to keep proteins homeostasis. In this research topic of "Protein Quality Controlling Systems in Plant Responses to Environmental Stresses", some latest researches and opinions in this field, including heat shock proteins (HSPs), unfolded protein response (UPR), ubiquitin-proteasome system (UPS)and autophagy, were reported, aiming to provide novel insights for increasing cropproduction under environmental challenges.

Citation: Lu, M., Liu, Y., Zhou, J., Hellmann, H. A., Wang, W., Stone, S., eds. (2018). Protein Quality Controlling Systems in Plant Responses to Environmental Stresses. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-558-4

# Table of Contents

*05 Editorial: Protein Quality Controlling Systems in Plant Responses to Environmental Stresses*

Minghui Lu, Hanjo A. Hellmann, Yule Liu and Wei Wang

*08 Responses of Plant Proteins to Heavy Metal Stress—A Review* Md. Kamrul Hasan, Yuan Cheng, Mukesh K. Kanwar, Xian-Yao Chu, Golam J. Ahammed and Zhen-Yu Qi

#### SECTION 1

#### INVOLVEMENT OF HEAT SHOCK PROTEINS (HSPS) IN PLANT RESPONSE TO ENVIRONMENTAL STRESS


#### SECTION 2

#### INVOLVEMENT OF UNFOLDED PROTEIN RESPONSE (UPR) IN PLANT RESPONSE TO ENVIRONMENTAL STRESS


#### SECTION 3

#### INVOLVEMENT OF UBIQUITIN-PROTEASOME SYSTEM (UPS) IN PLANT RESPONSE TO ENVIRONMENTAL STRESS


Hong-Mei Yuan, Ying Sheng, Wei-Jie Chen, Yu-Qing Lu, Xiao Tang, Mo Ou-Yang and Xi Huang

#### SECTION 4

#### INVOLVEMENT OF AUTOPHAGY PATHWAY IN PLANT RESPONSE TO ENVIRONMENTAL STRESS

*94 Autophagy: An Important Biological Process That Protects Plants From Stressful Environments*

Wenyi Wang, Mengyun Xu, Guoping Wang and Gad Galili

*98 The AMP-Activated Protein Kinase KIN10 is Involved in the Regulation of Autophagy in* Arabidopsis

Liang Chen, Ze-Zhuo Su, Li Huang, Fan-Nv Xia, Hua Qi, Li-Juan Xie, Shi Xiao and Qin-Fang Chen

*109 TOR-Dependent and -Independent Pathways Regulate Autophagy in*  Arabidopsis Thaliana

Yunting Pu, Xinjuan Luo and Diane C. Bassham

*122 Endocytosis of AtRGS1 is Regulated by the Autophagy Pathway After D-Glucose Stimulation*

Quanquan Yan, Jingchun Wang, Zheng Qing Fu and Wenli Chen

*133 Autophagy is Rapidly Induced by Salt Stress and is Required for Salt Tolerance in Arabidopsis*

Liming Luo, Pingping Zhang, Ruihai Zhu, Jing Fu, Jing Su, Jing Zheng, Ziyue Wang, Dan Wang and Qingqiu Gong

#### SECTION 5

#### PROTEOMIC STUDY OF PROTEIN QUALITY CONTROLLING SYSTEMS (PQCS) INPLANT RESPONSE TO ENVIRONMENTAL STRESS

*146 Proteomic and Physiological Analyses Reveal Putrescine Responses in Roots of Cucumber Stressed by NaCl* 

Yinghui Yuan, Min Zhong, Sheng Shu, Nanshan Du, Jin Sun and Shirong Guo

*165 Comparative Proteomic Analysis Provides Insight Into the Key Proteins Involved in Cucumber (*Cucumis sativus *L.) Adventitious Root Emergence Under Waterlogging Stress*

Xuewen Xu, Jing Ji, Xiaotian Ma, Qiang Xu, Xiaohua Qi and Xuehao Chen

*179 Proteomic Analysis Reveals the Positive Roles of the Plant-Growth-Promoting Rhizobacterium NSY50 in the Response of Cucumber Roots to Fusarium*  Oxysporum *f. sp.* Cucumerinum *Inoculation*

Nanshan Du, Lu Shi, Yinghui Yuan, Bin Li, Sheng Shu, Jin Sun and Shirong Guo

*198 Proteomic Analysis Reveals the Positive Effect of Exogenous Spermidine in Tomato Seedlings' Response to High-Temperature Stress* Qinqin Sang, Xi Shan, Yahong An, Sheng Shu, Jin Sun and Shirong Guo

# Editorial: Protein Quality Controlling Systems in Plant Responses to Environmental Stresses

Minghui Lu<sup>1</sup> \*, Hanjo A. Hellmann<sup>2</sup> \*, Yule Liu<sup>3</sup> \* and Wei Wang<sup>4</sup> \*

*<sup>1</sup> College of Horticulture, Northwest A&F University, Shaanxi, China, <sup>2</sup> School of Biological Sciences, Washington State University, Pullman, WA, United States, <sup>3</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>4</sup> College of Life Sciences, State Key Lab of Wheat & Maize Crop Science, Henan Agricultural University, Zhengzhou, China*

Keywords: plant, heat shock proteins, unfolded protein response, proteasome, autophagy, abiotic stress

**Editorial on the Research Topic**

#### **Protein quality controlling systems in plant responses to environmental stresses**

#### Edited and reviewed by:

*Elison B. Blancaflor, Noble Research Institute, LLC, United States*

#### \*Correspondence:

*Minghui Lu xnjacklu@nwsuaf.edu.cn Hanjo A. Hellmann hellmann@wsu.edu Yule Liu yuleliu@mail.tsinghua.edu.cn Wei Wang wangwei@henau.edu.cn*

#### Specialty section:

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

Received: *29 May 2018* Accepted: *08 June 2018* Published: *29 June 2018*

#### Citation:

*Lu M, Hellmann HA, Liu Y and Wang W (2018) Editorial: Protein Quality Controlling Systems in Plant Responses to Environmental Stresses. Front. Plant Sci. 9:908. doi: 10.3389/fpls.2018.00908* In nature, plants are routinely exposed to adverse environmental conditions, such as elevated temperature, drought, salinity, heavy metal, etc., which are among the main causes for declining crop productivity worldwide and lead to billions of dollars of annual losses (Dhankher and Foyer, 2018). These stressors negatively affect plant growth and development by inducing misfolding, denaturation, oxidation and aggregation of proteins. Evolutionally, plants have developed a comprehensive protein quality controlling system (PQCS) to maintain protein homeostasis, mainly including heat shock proteins (HSPs), unfolded protein response (UPR), ubiquitin-proteasome system (UPS) and autophagy. This research topic aims to summarize and report novel findings on the identification, functional analysis, signal transduction, transcriptional and post-transcriptional regulation, and protein interaction of candidate components in the above systems.

HSPs are abundantly expressed under abiotic stress conditions, and function as molecular chaperones to promote proper protein folding and to prevent denatured proteins from selfaggregation (Reddy et al., 2016). Yu et al. identified 42 putative SlHSP20 genes from tomato (Solanum lycopersicum), and found that their transcript levels were profusely induced by abiotic stresses such as heat, drought, salt, but also by the fungal pathogens Botrytis cinerea, and tomato spotted wilt virus (TSWV). In addition, a total of 76 putative CaDnaJ/HSP40 genes were identified in pepper (Capsicum annuum L.), and more than 80% of them responded to heat stress treatment (Fan et al.). These studies underscore the potential involvements of HSP genes in mediating the response of plants not only to elevated temperatures but also to a broader range of environmental stress conditions.

Endoplasmic reticulum (ER) is the major organelle for folding and assembling of secretory proteins. When plants are subjected to environmental stresses, the unfolded or misfolded proteins accumulate in the ER which is referred to as ER stress (Schröder and Kaufman, 2005), and which further activates UPR to enhance the operation of the ER protein-folding machinery (Duwi Fanata et al., 2013). Bao and Howell summarized in this research topic the latest progresses in UPR. The authors discuss recent findings that this pathway is not only associated with abiotic stress response, but is also required during normal vegetative and reproductive development. In addition, it fulfills critical roles in plant immunity, affecting bacterial and viral infections. Evidence that the UPR in multicellular organisms acts in a tissue specific manner comes from Cho and Kanehara. The authors measured expression of the immunoglobulinbinding protein gene BiP3, a marker for ER-stress. This gene was strongly up-regulated after treatment with the ER-stress inducer tunicamycin (TM). Interestingly, BiP3 expression in the plant was not uniformly increased but more tissue specific. For example, the mRNA abundance of BiP3 was preferentially increased in the vascular tissues, and leaf hydathodes. In the root tip, high expression was specifically observed in the columella, and the epidermal cell layer of the elongation zone. These findings indicate that in response to TM, plants emphasize certain tissues and/or organs to maintain ER homeostasis.

When stressed protein repair or folding demands exceed the cellular capacities, protein degradation systems such as UPS and autophagy are activated to remove misfolded proteins (Liu and Howell, 2016). The UPS marks proteins for degradation by attaching polyubiquitin chains to target proteins, which in turn leads to their degradation via the 26S proteasome (Liu and Howell, 2016). But degradation of misfolded proteins is only one aspect of the UPS function. The pathway represents a central regulatory tool that affects most cellular processes in plants. For example, ICE1 (INDUCER OF CBF EXPRESSION 1) is involved in chilling and freezing tolerance by promoting expression of the CBF3 (C-REPEAT-BINDING FACTOR 3) transcription factor, and other cold-responsive genes (Chinnusamy et al., 2003). However, after ICE1 facilitated a cold-shock response, it becomes ubiquitinated by the E3 ligase HOS1 (HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENE 1) followed by proteasomal degradation (Dong et al., 2006).

In this research topic, Yuan et al. cloned HbICE1 from rubber trees (Hevea brasiliensis) and showed that its overexpression in Arabidopsis enhances cold tolerance. Fan et al. reported the identification of MaSINA1 (SEVEN IN ABSENTIA1), an E3 ligase from banana (Musa acuminata), which interacts with MaICE1, and promotes its degradation. Consequently, the authors suggested that MaSINA1 functions as a negative regulator of cold stress response in banana.

While UPS is targeting single proteins for degradation, autophagy is active on a broader scale and responsible for the degradation of single proteins, as well as protein aggregates or even whole organelles (Zientara-Rytter and Sirko, 2016). Autophagy is characterized by the de novo formation of a double membrane organelle (called autophagosome), to transport the targeted cargo components to the vacuole for degradation (Batoko et al., 2017). Autophagy-related proteins (ATGs) or their complexes recognize the target components by specific cargo receptors. In a review article, Wang et al. summarized the identification and functional characterization of three potential cargo receptors involved in plant abiotic stress, including NBR1 (NEIGHBOR OF BRCA1), TSPO (TRYPTOPHAN-RICH SENSORY PROTEIN), ATI1 (AUTOPHAGY INTERACTING PROTEIN1).

The timely degradation of misfolded proteins is important for the development of plant tolerance to abiotic stress. Luo et al. suggested that rapid protein turnover through autophagy is a prerequisite for the establishment of salt tolerance in Arabidopsis. They found that after salt treatment autophagosome formation is induced shortly, and the level of autophagy peaks within 30 min. Accordingly, within 3 h of salt treatment, accumulation of oxidized proteins is alleviated, and then contents of soluble sugar and some compatible solutes such as proline are enhanced. However, these processes are not observed or kept at lower levels in mutants such as atg2 or atg7 that are defective in autophagy. The authors propose that autophagy under salt stress is a critical requirement for bulk protein turnover.

The TOR (TARGET OF RAPAMYCIN) protein kinase is a major controller of growth-related processes in all eukaryotes. Under favorable conditions, TOR positively regulates cell and organ growth but restrains autophagy processes (John et al., 2011). However, Pu et al. reported that the modulation of autophagy by TOR was stress-type dependent. They found that the overexpression of the TOR kinase inhibited autophagy activation by nutrient starvation, salt and osmotic stress, but not by oxidative or ER stress. A similar result was observed after the treatment with the auxin NAA (1-naphthaleneacetic acid), a phytohormone that upregulates TOR activity. Since NAA treatment was unable to overcome blocked autophagy induced by a TOR inhibitor, it was suggested that auxin acts upstream of TOR in the regulation of autophagy.

Chen et al. found that KIN10 (KINASE HOMOLOG 10), a plant ortholog of the mammalian AMPK (AMP-ACTIVATED PROTEIN KINASE), acts as a positive regulator of autophagy by affecting the phosphorylation of ATG1 proteins in Arabidopsis. In KIN10 overexpression lines (KIN10-OE), the stress-induced formation of autophagosomes were accelerated. In addition, leaf senescence was delayed, while the tolerance to nutrient starvation, drought and hypoxia treatments was increased. Furthermore, carbon starvation (transfer of seedlings to continuous darkness) enhanced the level of phosphorylated ATG1a in KIN10-OE lines.

Another aspect of autophagy in this research topic was investigated by Yan et al. by studying the impact of autophagy and D-glucose on the endocytosis of RGS1 (REGULATOR OF G-PROTEIN SIGNALING 1). Under normal conditions, RGS1 interacts with and arrests the GTPase activity of the heterotrimeric G-protein subunit Gα subunit (GPA1). However, D-glucose recruits WNK8 (WITH-NO-LYSING KINASE 8) to phosphorylate AtRGS1, which in turn causes its endocytosis. The endocytosis of RGS1 physically uncouples its inhibitory activity from GPA1, and then activates the G protein-mediated sugar signaling (Urano et al., 2012). Yan et al. reported that D-glucose induced RGS1 endocytosis is needed for the formation of autophagosomes likely by activating ATG8 phosphatidylethanolamine (PE) and ATG12/ATG5 conjugation systems. The autophagy pathway on the other hand is needed for RGS1 endocytosis as RGS1 remains associated with GPA1 in atg2 and atg5 autophagy mutants, even in the presence of Dglucose. The findings show a nice interplay between endocytotic and autophagy pathways, and shed new light on sugar signaling in plant cells.

The development of plant tolerance to abiotic stress always requires the simultaneous participation of different PQCSs. Heavy metals negatively affect plant cell viability mainly by disturbing protein folding and stimulating protein aggregation. In the review article of Hasan et al. the authors summarized the recent advances on the involvement of PQCSs in plant tolerance to heavy metal stress, including ion detoxification

The proteomics study of Xu et al. provides us with new insights into the involvement of PQCS in establishing plant tolerance under adverse environmental conditions. Based on iTRAQ-quantitative proteomics approach, the authors compared the cucumber (Cucumis sativus) proteomes in adventitious roots under control and waterlogging conditions. They identified a total of 146 differentially regulated proteins (DRPs), of which 13 belonged to the categories of posttranslational modification, protein turnover and chaperones.

Polyamines such as putrescine (Put), spermidine (Spd) and spermine (Spm), are suggested to maintain the function and structure of cellular components in plant response to stress (Liu et al., 2015). After treatment with exogenous Put, Yuan et al. analyzed the DRPs of cucumber under salt stress by MALDI-TOF/TOF MS, and identified 62 DRPs, of which 15 functioned in protein metabolism, 15 in defense responses, 12 in carbohydrate metabolism, and 9 in amino acid metabolism. In a similar study in tomato with exogenous Spd, 67 DRPs were identified after high temperature treatment. The percentage of the identified proteins played roles in photosynthesis was 27%, followed by 24% of cell rescue, and defense. However, a significant amount was also related to protein synthesis, folding and degradation (22%) as well as energy and metabolism (13%) (Sang et al.).

The plant growth-promoting rhizobacterium (PGPR) can induce resistance against a broad spectrum of pathogens by simultaneously activating salicylic acid and jasmonate/ethylenedependent signaling pathways (Niu et al., 2011). With a new potential strain NSY50, Du et al. investigated the mechanisms

#### REFERENCES


of PGPR protecting cucumber from the attack of Fusarium oxysporum f. sp. Cucumerinum (FOC) by a proteomic approach. Among the 56 DRPs, 14 belonged to the protein metabolism category and two to the HSP70 family, which suggests a functional connection between the PGPR and PQCS under biotic stress.

With the unprecedented global climate changes, extreme weather conditions are more likely to occur, and which will severely impact plant growth and crop production. A better understanding of the mechanisms of how plants are able to cope with and alleviate environmental stresses is essential for crop breeders to develop efficient strategies for maintaining our current agricultural productivity and to secure a sustainable agriculture. The research topic summarized here may provide some novel insights that can help to address these eminent challenges and to further increase crop production and secure yield in the upcoming decades.

#### AUTHOR CONTRIBUTIONS

ML prepared the first draft of this editorial. HH, YL, and WW revised it. All authors listed approved it for publication.

#### ACKNOWLEDGMENTS

We thank Prof. Jie Zhou (Zhejiang University, China) and Prof. Sophia Stone (Dalhousie University, Canada) for their outstanding contribution to the edit of manuscripts submitted to this Research Topic. ML is thankful to the support from the National Natural Science Foundation of China (Grant No. 31572114).

activating salicylate- and jasmonate/ethylene-dependent signaling pathways. Mol. Plant Microbe Interact. 24, 533–542. doi: 10.1094/MPMI-09-10-0213


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

# Responses of Plant Proteins to Heavy Metal Stress—A Review

Md. Kamrul Hasan1, 2†, Yuan Cheng3†, Mukesh K. Kanwar 1†, Xian-Yao Chu<sup>4</sup> , Golam J. Ahammed<sup>1</sup> \* and Zhen-Yu Qi <sup>5</sup> \*

*<sup>1</sup> Department of Horticulture, Zhejiang University, Hangzhou, China, <sup>2</sup> Department of Agricultural Chemistry, Sylhet Agricultural University, Sylhet, Bangladesh, <sup>3</sup> State Key Laboratory Breeding Base for Zhejiang Sustainable Pest and Disease Control, Institute of Vegetables, Zhejiang Academy of Agricultural Sciences, Hangzhou, China, <sup>4</sup> Zhejiang Institute of Geological Survey, Geological Research Center for Agricultural Applications, China Geological Survey, Beijing, China, <sup>5</sup> Agricultural Experiment Station, Zhejiang University, Hangzhou, China*

#### Plants respond to environmental pollutants such as heavy metal(s) by triggering the expression of genes that encode proteins involved in stress response. Toxic metal ions profoundly affect the cellular protein homeostasis by interfering with the folding process and aggregation of nascent or non-native proteins leading to decreased cell viability. However, plants possess a range of ubiquitous cellular surveillance systems that enable them to efficiently detoxify heavy metals toward enhanced tolerance to metal stress. As proteins constitute the major workhorses of living cells, the chelation of metal ions in cytosol with phytochelatins and metallothioneins followed by compartmentalization of metals in the vacuoles as well as the repair of stress-damaged proteins or removal and degradation of proteins that fail to achieve their native conformations are critical for plant tolerance to heavy metal stress. In this review, we provide a broad overview of recent advances in cellular protein research with regards to heavy metal tolerance in plants. We also discuss how plants maintain functional and healthy proteomes for survival under such capricious surroundings.

Keywords: heavy metals, phytochelatins, metallothioneins, protein quality control system, ubiquition proteasome system, autophagy

#### INTRODUCTION

Proteins are functionally versatile macromolecules that constitute the major workhorses of living cells. They function in cellular signaling, regulation, catalysis, intra and inter cellular movement of nutrients and other molecules, membrane fusion, structural support and protection (Amm et al., 2014). The function of a protein is basically determined by its structure, which is acquired following ribosomal synthesis of its amino acid chain. In addition, the conformation of a protein largely depends on the physical and chemical conditions of the protein environment as affected by extreme temperatures, reactive molecules, heavy metal (HM) ions and other stresses that not only disrupt the folding process of a newly synthesized protein, but also induce the mis-folding of already existing proteins (Goldberg, 2003; Amm et al., 2014; Zhou et al., 2016).

Over the last several decades, the emission of pollutants into the environment has been increased tremendously due to rapid industrialization, urbanization and excessive usage of agricultural amendments. Being sessile, plants are routinely confronted by a wide array of biotic and/or abiotic stresses including HM stress (Al-Whaibi, 2011). HMs are thought to obstruct the biological

#### Edited by:

*Minghui Lu, Northwest A&F University, China*

#### Reviewed by:

*Shabir Hussain Wani, Michigan State University, United States Liang Xu, Nanjing Agricultural University, China*

#### \*Correspondence:

*Golam J. Ahammed ahammed@zju.edu.cn Zhen-Yu Qi qizhenyu@zju.edu.cn*

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

#### Specialty section:

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

Received: *30 May 2017* Accepted: *11 August 2017* Published: *05 September 2017*

#### Citation:

*Hasan MK, Cheng Y, Kanwar MK, Chu X-Y, Ahammed GJ and Qi Z-Y (2017) Responses of Plant Proteins to Heavy Metal Stress—A Review. Front. Plant Sci. 8:1492. doi: 10.3389/fpls.2017.01492*

**8**

functions of a protein by altering the native conformation through binding on it (Hossain and Komatsu, 2013). For example, in yeast, methyl-mercury (MeHg) strongly inhibits L-glutamine: D-fructose-6-phosphate aminotransferase, and overexpression of this enzyme confers tolerance to MeHg (Naganuma et al., 2000). Similarly, cadmium (Cd) can inhibit the activity of thiol transferase leading to oxidative damage, possibly by binding to cysteine residues in its active sites. In Brassica juncea, Cd-dependent changes in beta carbonic anhydrase result in the enhancement of photorespiration which may protect photosystem from oxidation (D'Alessandro et al., 2013). The modifications caused by Cd disrupt the stabilizing interactions associated with changes in the tertiary structure and cause loss of promising functions of that protein (Chrestensen et al., 2000). Fallout dysfunction of protein stimulates the danger of protein aggregation.

The biosynthesis of metal binding cysteine rich peptides that function to immobilize, sequester and detoxify the metal ions is thought to be the central for detoxification of HMs (Clemens, 2006; Viehweger, 2014). Nonetheless, under extreme conditions, metal ions profoundly affect cellular protein homeostasis by interfering with their folding process and stimulate aggregation of nascent or non-native proteins, leading to the endoplasmic reticulum (ER) stress and a decreased cell viability. To restrict the aggregation as well as to mend them is-folded proteins, cells initiate different quality control systems that fine-tune protein homeostasis. In the center of the system, a typical set of proteins, called heat shock proteins (HSPs; Amm et al., 2014), function as surveillance mechanisms, which are preferentially expressed under stress to maintain functional and healthy proteomes. In contrast, the damaged proteins that fail to achieve their native conformations are subjected to degradation through the ubiquitinproteasome process (UPS), called as ER-associated degradation (ERAD) or through autophagy to minimize the accumulation of misfolded proteins in cells (Liu and Howell, 2016). Although a significant progress has been made in our understanding of protein quality control systems, information on plant system, especially pertaining to HMs stress still remain scanty. In this review, we aim to provide a better insight into the protein quality control system in plants with regards to heavy metal tolerance. We also discuss how plants try to ensure functional and healthy proteomes under HM stress.

#### HEAVY METALS (HMs) DETOXIFICATION

Toxic metal ions at cellular level, evoke oxidative stress by generating reactive oxygen species (ROS; Li et al., 2016a). They promote DNA damage and/or impair DNA repair mechanisms, impede membrane functional integrity, nutrient homeostasis and perturb protein function and activity (Tamás et al., 2014). On the other hand, plant cells have evolved a myriad of adaptive mechanisms to manage excess metal ions and utilize detoxification mechanisms to prevent their participations in unwanted toxic reactions. In the first line of defense, plants utilize strategies that prevent or reduce uptake by restricting metal ions to the apoplast through binding them to the cell wall or to cellular exudates, or by inhibiting long distance transport (Manara, 2012; Hasan et al., 2015). In contrast, when present at elevated concentrations, cells activate a complex network of storage and detoxification strategies, such as chelation of metal ions with phytochelatins and metallothioneins in the cytosol, trafficking, and sequestration into the vacuole by vacuolar transporters (**Figure 1**; Zhao and Chengcai, 2011).

#### Phytochelatins: Structure, Regulation and Function in Heavy-Metal Stress Tolerance

In order to reduce or prevent damage caused by HMs; plants synthesize small cysteine-rich oligomers, called Phytochelatins (PCs) at the very beginning of metal stress (Ashraf et al., 2010; Pochodylo and Aristilde, 2017). Notably, PC syntheses play the most crucial role in mediating plant tolerance to HMs (Clemens, 2006; Emamverdian et al., 2015). It has been well documented that the biosynthesis of PCs can be regulated at post-translational level by metal(oid)s in many plant species. However, the over-expression of phytochelatin synthase (PCS) gene in plants does not always result in an enhanced tolerance to HM stress. For instance, over expression of AtPCS1 in Arabidopsis, paradoxically shows hypersensitivity toward Cd and Zn; although, PCs production is increased by 2.1-folds, when compared with wild type plants (Lee et al., 2003). In reality, excess PCs levels in mutant plants accelerate accumulation of HMs like Cd without improving plant tolerance (Pomponi et al., 2006; Furini, 2012). This phenomenon possibly indicates some additional roles of PCs in plant cells, such as their involvement in essential metal ion homeostasis, antioxidant mechanisms, and sulfur metabolism (Furini, 2012). Therefore, prevention of the free circulation of toxic metal inside the cytosol exhibits a potential mechanism for dealing with HM-induced toxicity (Hasan et al., 2016).

The mechanism of HMs detoxification is not only limited to the chelation, but also involves accumulation and stabilization of HM in the vacuole through formation of high molecular weight (HMW) complexes with PCs (**Figure 1**; Jabeen et al., 2009; Furini, 2012). Generally, sequestration of metal ions is a strategy adopted by organisms to ameliorate toxicity. The arrested metal ions are transported from cytosol to the vacuole for sequestration via transporters. vacuolar sequestration is the vital mechanism to HM homeostasis in plants, which is directly driven by ATP-dependent vacuolar pumps (V-ATPase and V-PPase) and a set of tonoplast transporters (Sharma et al., 2016). RNA-Seq and de novo transcriptome analysis showed that different candidate genes that encode heavy metal ATPases (HMAs), ABC transporter, zinc iron permeases (ZIPs) and natural resistance-associated macrophage proteins (NRAMPs) are involved in metal transport and cellular detoxification (Xu et al., 2015; Sharma et al., 2016). A classic example of such protein in Cd uptake in A. thaliana is the Fe (II) transporter iron-regulated transporter 1 (IRT1) belonging to the ZIP family (Connolly et al., 2002). Furthermore, NRAMPs members such as NRAMP5 is recognized as an important transporter for Mn acquisition and major pathway of Cd entry into rice roots, which

is localized at the distal side of exodermis and endodermis of the plasma membrane of cells (Clemens and Ma, 2016). Interestingly, another transporter HMA2 localized in the plasma membrane of pericycle cells is thought to transport Cd from the apoplast to the symplast to facilitate translocation via the phloem in rice, whereas HMA3 in the tonoplast sequesters Cd into vacuoles by serving as primary pump (Clemens and Ma, 2016; Sharma et al., 2016). The HM transporter 1 (HMT1) was first identified in 1995 in the yeast S. pombe, as a vacuolar PC transporter required for Cd tolerance (Mendoza-Cózatl et al., 2011). The HMT1 gene encodes ATP-binding cassette (ABC) membrane transport proteins; therefore, both HMT1 and ATP are required for the translocation of LMW PC-Cd complexes into the vacuole (**Figure 1**; Cobbett and Goldsbrough, 2002). In progression, two ABCC subfamily members of ABC transporters, ABCC1 and ABCC2 were also identified as additional vacuolar metal-PC complex transporter in S. pombe and Arabidopsis (Mendoza-Cózatl et al., 2010; Park et al., 2012). Using double mutants, Song et al. (2014) demonstrated that vacuolar sequestration by ABCC1 and ABCC2 is necessary for complete detoxification of Arsenic (As) and Cd in Arabidopsis. Interestingly, they also reported that the addition of necessary metal ions, such as zinc (Zn), copper (Cu),manganese (Mn) and iron(Fe) to the transport assay further enhances PC<sup>2</sup> transport efficiency in barley vacuoles, suggesting that PCs might contribute to both the homeostasis of essential metals and detoxification of non-essential toxic metal(loid)s in plants (Song et al., 2014). Although the mechanism how the transporters regulate the sequestration of metal-PCs conjugates to vacuoles is not clear. Very recent, Zhang et al. (2017), for the first time provided evidence that phosphorylation-mediated regulation of ABCC1 activity is required for vacuolar sequestration of As. They found that Ser<sup>846</sup> phosphorylation is required for the As resistance function of ABCC1 in Arabidopsis.

#### Metallothioneins (MTs): Structure, Regulation and Functions in HM Tolerance

Alike PCs, MTs are also naturally-occurring intracellular cysteine-rich major metal-binding proteins, which are used by cells to immobilize, sequester, and detoxify metal ions (Capdevila and Atrian, 2011). Although plant MTs have been discovered over last three decades, the precise physiological functions of MTs have not yet been fully elucidated (Liu et al., 2015a). The proposed roles of MTs include (a) participation in maintaining the homeostasis of essential transition metal ions, (b) sequestration of toxic HMs, and (c) protection against intracellular oxidative damage induced by stress (Hossain et al., 2012a).

Transition metals such as Cu, Fe, Mn and Zn are essential for all organisms because they play critical roles in a variety of physiological processes. For example, Cu is required for photosynthesis, respiration, ethylene perception, ROS metabolism and cell walls in plants (Burkhead et al., 2009; Peñarrubia et al., 2010). A number of studies suggested the involvement of plant MTs in the participation of metal ion homeostasis, especially for Cu, during both vegetative growth and senescence. For example, Benatti et al. (2014) demonstrated that the MTs deficient mutants accumulate 45% and 30% less Cu in shoot and root, compared to the WT, while there are no obvious differences in the life cycle between WT and quad-MT mutant plants under various growth conditions. Again, at early vegetative stage, there is no significant difference in Cu uptake in leaves of 4-week-old WT and MT-deficient mutants. However, the concentration of Cu remains twice in leaves of 12-week-old MT-deficient plants compared to leaves of WT. In contrast, the Cu concentration in seeds of MT-deficient plants was less than half compared to the seeds of WT (Benatti et al., 2014). All these results suggest that MTs are not required to complete life cycle, but are important for essential ions homeostasis and distribution in plants.

In general, sequestration of intracellular HMs in eukaryotes also involves binding of HMs with cytosolic cysteine-rich MTs peptides as well as compartmentalization (Sácký et al., 2014). The combination of low kinetic stability and high thermodynamic is the main features of metal-MT complexes, which bind the metals very firmly, while a part of the metal ions is easily exchanged for other proteins (Maret, 2000). Transgenic plants overexpressing MTs genes have been scored for enhanced metal tolerance and they demonstrate modified metal accumulation or distribution strategies (Gu et al., 2015; Liu et al., 2015a; Tomas et al., 2015). In plants, vacuole is considered as the final destination of detoxification of HMs. Although the chelation of metal ions by MTs is well documented, a little is known about the mechanisms of transport of metals-MT complex from the cytoplasm to the vacuole (Yang et al., 2011). Surprisingly, in ThMT3 (hispida metallothionein-like ThMT3) transgenic material, the expression of four genes (GLR1, GTT2, GSH1, and YCF1) which aid transport of HMs from the cytoplasm to the vacuole is not induced by Cd, Zn, or Cu stress (Yang et al., 2011). These results advocate that metals are not transported into vacuoles, and thatThMT3 may only regulate HMs accumulation in the cytoplasm.

The biologic functions of MTs have been a perplexing topic ever since their discovery. Many studies have suggested that in addition to the chelation or metal ion homeostasis, MTs play an important role in cellular redox homeostasis under diverse stress conditions (Kang, 2006). Abiotic stresses like HMs induce excessive accumulation of ROS in plants, and cause damages to the cellular macromolecules such as proteins, leading to metabolic and physiological disorders in cells or even cell death (Hasan et al., 2016). Interestingly, MTs have been proposed as an alternative tool by which plants protect themselves from stress-induced oxidative damage (**Figure 1**; Hassinen et al., 2011; Ansarypour and Shahpiri, 2017). Although many reports have indicated the roles of MTs in abiotic stress tolerance as ROS scavengers, the mechanisms through which MTs mediate ROS homeostasis remain unclear (Hassinen et al., 2011). It has been proposed that during ROS scavenging, metals are released from MTs and ROS moiety is bounded to the Cys residues of the same. A number of studies have also advocated that the released metals might be involved in the initiation of signaling cascade required for ROS scavenging (Hassinen et al., 2011). For example, normal cellular functioning requires Zn mobilization and its transfer from one location to another or from one Zn-binding site to another. The released Zn from MT mobilized by an oxidative reaction may either constitute a general pathway by which Zn is distributed in the cell or be restricted to conditions of oxidative stress, where Zn is essential for antioxidant defense systems (Kang, 2006), suggesting an important role of MTs in ROS homeostasis and protection of cellular macromolecules from stress-induced ROS. Additionally, the different classes of MTs have distinct tissue-specific expression patterns in plants. As example, GUS reporter constructs explored that MT1a and MT2b are expressed in the phloem, whereas MT2a and MT3 in the mesophyll cells of young leaves and in root tips (Hassinen et al., 2011). Likewise, Liu et al. (2015a) also demonstrated that OsMT2c gene encoding for type 2 MT expressed in the roots, leaf sheathes, and leaves of rice, whereas its weak expression was observed in seeds. Considering their diversified role and tissue specific expression, recently Irvine et al. (2017) showed an excellent effort to develop a low-cost MT-biosensor that can dramatically increase the signal associated with a metal of interest. Such a simple sensor technology could be potentially used in environmental monitoring specially in the areas with the metal contamination problems.

#### REPAIRING OF DAMAGED PROTEINS

Proteins are the primary targets of HMs. They either form a complex with functional side chain groups of proteins or displace essential ions from metallo proteins, leading to impairment of physiological functions (Tamás et al., 2014). In addition, HMs interfere with the native confirmations of proteins by inhibiting folding process of nascent or non-native proteins that manifest both in a quantitative deficiency of the affected proteins and in the formation of proteotoxic aggregates (Bierkens, 2000; Tamás et al., 2014). Interestingly, plants inherently respond to stress by triggering the activation of the genes involved in cell survival and/or death in contaminated environments (Hossain et al., 2013). As a part of this plant response ubiquitously involves a set of genes, commonly termed as stress genes, are induced to synthesize a group of proteins called HSPs (Gupta et al., 2010). In stress conditions, the induced synthesis of HSPs plays a significant role in maintaining the cellular homeostasis by assisting accurate folding of nascent and stress accumulated misfolded proteins, preventing protein aggregation or by promoting selective degradation of misfolded or denatured proteins (Hüttner et al., 2012; Park and Seo, 2015).

In fact, HSPs functions as molecular chaperones; proteins which are involved in "house-keeping" inside the cell (Sørensen et al., 2003). Several classes of HSP have been identified in plants (**Table 1**) and the HSP proteins having molecular weights ranging from 10 to 200 KD are characterized as chaperones which participate in the induction of the signal in stress conditions (Schöffl et al., 1999). For example, in endoplasmic reticulum (ER) all the nascent polypeptides, firstly, are stabilized by chaperones (HSP40 and HSP70-like proteins) such as ERdj3 and binding protein (BiP) before they are properly modified and folded, which prevents aggregation and helps their proper folding (**Figure 2**; Howell, 2013).

### Role of HSPs in Plant Tolerance to HM Stress

HM stress often causes disturbance to the cellular homeostasis by inactivating essential enzymes and by suppressing proteins functioning (Hossain et al., 2012b). Hence, the induction of HSPs proteins is thought-out as a critical protective, ecophysiologically adaptive and genetically conserved response of organisms to the environmental anxiety. Thus, they accomplish a key function in the hostility of stress by re-establishing normal protein conformation and cellular homeostasis (Rhee et al., 2009). Among the major categories of HSPs, HSP70 family members have extensively been studied. Functional characterization of HSP70 revealed that HSP70 is accumulated in response to environmental stressors in a wide range of plant species (Gupta et al., 2010). The specific members of this family are localized into the cytosol, mitochondria and endoplasmic reticulum(ER) and are constitutively expressed as well as regulated to maintain cellular homeostasis. An example to cite, the 70-KDa heat shock cognates(HSC70) are constitutively expressed in cells and often assist in the folding of de novo synthesized polypeptides and import or translocations of precursor proteins (Wang et al., 2004).

The recent advancements in proteomics research have enabled us to identify the functional genes or proteins involved in the responses of plants to HM stress at molecular levels (Ahsan et al., 2009). Transcript analysis in many plant species showed that HSP70 is highly expressed under a variety of metal stress (**Table 1**). Although many studies showed that the overexpression of HSP70 genes is positively correlated with the acquisition of tolerance to various stresses, including HMs, but the cellular mechanisms of HSP70 function under stress situation are not completely understood (Wang et al., 2004). HSP70 chaperones, together with their co-chaperones like DnaJ, make a set of prominent cellular machines to prevent accumulation of newly synthesized proteins as aggregates and ensure the proper folding of protein during their transfer to the destination (Al-Whaibi, 2011; Park and Seo, 2015). In transportation of precursor protein, the HSC70 is essential for cell-to-cell transport through interaction with the plasmodesmatal translocation pathway (Aoki et al., 2002). The induction of HSP70 not only limits the proteotoxic symptoms of metalions, but also helps the sequestration and detoxification of these ions by MTs (Haap et al., 2016). While the entire mechanism of HSPs-induced metal detoxification via MT has yet to be explored, only few studies pointed out that HSP60 might participate in protein folding and aggregation of many other proteins that are transported to organelles such as mitochondria and chloroplasts (Al-Whaibi, 2011). With our increasing understanding of the proteome, it is becoming clear that HSP60 is essential for cellular functions both at normal or stress environments, including metal stress (**Table 1**). Interestingly, proteomics analysis also revealed that the induction HSP60 chaperones prevents the denaturation of proteins even in the presence of metal ions in the cytoplasm (Sarry et al., 2006; Rodríguez-Celma et al., 2010). Similarly, a good number of studies showed the induction of HSP90 family proteins by different metals in many plant species (**Table 1)** which play a major role in protein folding and regulating signaltransduction networks, cell-cycle control, protein degradation and protein trafficking (Pratt and Toft, 2003; Al-Whaibi, 2011). Interestingly, they have also been found in association with several other intercellular proteins, including calmodulin, actin, tubulin and some other receptors and signaling kinases (Wang et al., 2004; Gupta et al., 2010; Park and Seo, 2015). The multiple sites of localization and high accumulation in combination with other intercellular proteins lead to the suggestion that these polypeptides perform a general mode of cellular activities (Prasad et al., 2010). This family of proteins might provide genetic buffering and contribute to the evolutionary adaptation of plant both in normal and stressful conditions (Wang et al., 2004). By contrast, there is no substantial evidence implicating HSP100/Clp proteins in HMs tolerance in plants (Agarwal et al., 2003). Recently, few studies reported that many members of this family are induced in response to metal treatments (**Table 1**),and they accomplish house keeping functions necessary for cellular homeostasis (Lee et al., 2006).

Most of the members of sHSPs are strongly inducible and some are also constitutively expressed under environmental stress conditions. One of the featured functions of this family of protein is the degradation of the proteins not suitable for folding (Gupta et al., 2010). Similar to other HSPs, sHSPs also function as molecular chaperones, however, the important characteristic that distinguishes sHsps from other chaperone classes, such as DnaK or ClpB/DnaK is that their activity is independent of ATP (Sun


TABLE 1 | Five major classes of heat shock proteins (HSPs) that are induced in response to heavy metal stress in plants.

*(Continued)*


synthesized proteins are translocated into the ER, where proteins folded into their native three dimensional structures with the help of chaperones. The correctly folded proteins are then transported to the Golgi complex, followed by delivery to the destination where they eventually function. While exposure of plants to stress causes oxidative stress by generating over ride of ROS and stimulating the misfolding of proteins. The incorrectly folded proteins are then detected by quality control system, which stimulates another pathway called unfolded protein response (UPR). The terminally misfolded proteins are then eliminated through the endoplasmic reticulum associated degradation (ERAD) pathway, where they initially ubiquitinated and then degraded in the cytoplasm by proteasome system (UPS) or subjected to autophagy. Adopted from Dobson (2003) with modifications.

et al., 2002; Mogk and Bukau, 2017). sHSPs maintain denatured proteins in a folding-competent state and allow subsequent ATP-dependent disaggregation through the HSP70/90 chaperone system, thereby facilitating their subsequent refolding (Wang et al., 2004). Recently, it has been shown that BAG3 protein acts as a modulator, brings the chaperone families together into a complex and coordinates the potentiality of Hsp22 and Hsp70 to refold the denatured proteins (Rauch et al., 2017). Additionally,

this family of HSPs is also involved in signaling similar to their counterparts, where they become phosphorylated by stresskinases, and increase the amount of reduced glutathione in the cytoplasm (Arrigo, 1998). A number of studies suggest a strong correlation between sHSP accumulation and plant tolerance, particularly to HMs stress (**Table 1**; Wang et al., 2004). In Arabidopsis, 13 different sHsps have been identified in distinct cellular compartments, which probably function as mediators of molecular adaptation to stress conditions and are unique to plants. The differential regulation of chloroplastic i.e., Cp-sHSPs or HSP26.13p in C. album plays a dual role in protecting the plant from heat and metals like Ni, Cu, and Cd stress (Haq et al., 2013).

Notably, most of HSPs are activated early in the cascade of cellular events following toxic exposure even below the lethal dose and thus, considered as strong biomarkers for environmental pollution (Bierkens et al., 1998). The excessive accumulation of ROS at cellular level is a common consequence of abiotic stress such as HMs. Interestingly, the high HSP levels protect plants against abiotic stresses not only by preventing irreversible protein aggregation, but also by promoting cellular redox homeostasis through stimulating antioxidant systems (Mu et al., 2013). Recently, Cai et al. (2017) revealed that HSPsinduced metal tolerance in plants has a strong correlation with melatonin (N-acetyl-5-methoxy tryptamine) biosynthesis, which is regulated by heat-shock factor A1a (HsfA1a). Therefore, all the above presented results suggest that the inducible HSPs are important and beneficial for fitness in normal, as well as under capricious environment. As a part of protein quality control system, HSPs play a major role in maintaining the functionality of cellular machinery under environmental stress. Therefore, research on the functional and structural aspects, cross-talk with other chaperones and relationship between different HSP expressions along with the physiological stress response should be expanded to better understand their functions.

#### Heavy Metals-Induced ER Stress

In cells, both recently synthesized and preexisting polypeptides are at constant risk of misfolding and agreegation. It has been estimated that one-third of recently synthesized proteins are misfolded under ambient conditions (Schubert et al., 2000). In addition, cells continuously face environmental challenges such as mutations, heat, active oxygen radicals and HM ions, which not only disrupt protein folding but also cause the misfolding of already folded protein, (Amm et al., 2014). The disruption of proper functioning of the ER is particularly relevant under stress conditions, whereas the demand for secreted proteins exceeds its working capacity and disrupts normal functioning of ER, termed as ER stress (Schröder and Kaufman, 2005). A number of studies have shown that HMs and metalloids inhibit refolding of chemically denatured proteins in vitro, obstruct protein folding in vivo and stimulate aggregation of nascent proteins in living cells (Sharma et al., 2011; Jacobson et al., 2012). For example, Cr has been shown to trigger oxidative protein damage and protein aggregation in yeast by enhancing mRNA mistranslation (Sumner et al., 2005; Holland et al., 2007). Likewise, Cd has also been shown to cause the widespread aggregation of nascent protein and ER stress in yeast (Gardarin et al., 2010), whilst the mechanistic details of protein misfolding in the ER and cytoplasm remain to be elucidated (Tamás et al., 2014). But somehow, this could be related to metal-induced structural alteration of ER. Recently, Karmous et al. (2015) demonstrated that Cu treatment in embryonic cells of Phaseolus vulgaris results in prevalently swollen cisternae of smooth endoplasmic reticulum and vesicles with electrondense contents. Although this phenomenon is often not well recognizable, it robustly inhibits cellular homeostasis. While the toxicity of metals, like Cd, As, Pb, Hg, and Cr is unquestionable and their interference with protein folding in living cells is well-documented, the potency of accumulation of misfolded and aggregated proteins appears to be different (Tamás et al., 2014). In yeast cells, the accumulations of aggregated proteins occur in the order As>Cd>Cr upon exposure of equi-toxic concentrations of these metals (Jacobson et al., 2012). The in vivo potency of these environmental threats to prompt protein aggregation possibly depends on the efficiency of their cellular uptake/transport and their distinct modes of biological action (Tamás et al., 2014). Under such circumstances, a coordinated adaptive program called the unfolded protein response (UPR) is commonly initiated.

The UPR is a homeostatic response to alleviate ER stress through transcriptional and translational events. The induction of UPR has three aims; initially to restore the normal function of the cell by halting production of secreted and membrane proteins, removal of misfolded proteins through ER-associated degradation (ERAD) systems, and activation of the signaling pathways that lead to increased production of molecular chaperones involved in protein folding. If these cellsparing activities are not achieved within a certain time span or the disruption is prolonged, then the UPR aims toward programmed cell death (PCD) which is called apoptosis (Deng et al., 2013; Liu and Howell, 2016). The UPR is a sensitive surveillance mechanism that monitors the ER loading capacity. If persistently misfolded proteins exceed the ER loading capacity, cellular communication between the ER and nucleus occurs for ER homeostasis, leading to the transcriptional activation that up-regulates the cellular chaperones (Brandizzi et al., 2014). Recently, it has been reported as two arm process in plants, where proteolytic processing of membrane-associated bZIP TFs and RNA splicing factor inositol-requiring enzyme-1 (IRE1) act as transducers in ER stress or UPR signaling pathway. Structural details of bZIP TFs, bZIP17, bZIP28 and IRE1/ bZIP60 and their underlying principles of mobilization in response to ER stress suggest that the UPR signaling pathway is assisted by these factors in order to protect plants from ER stress(Deng et al., 2013; Sun et al., 2013; Yang et al., 2014). This UPR transcriptional activation enhances the production of molecular chaperones such as binding protein (BiP) and glucose-regulated protein 94 (GRP94), involved in increased ER protein-folding potentiality (Yoshida et al., 1998). The molecular chaperones binding proteins(BiP) are the master regulatory elements of these quality control systems and a classical marker of UPR activation (Malhotra and Kaufman, 2007). For example, Xu et al. (2013) demonstrated that heterologous expression of AtBiP2 protein in BY-2 act as a negative regulator of Cd-induced ER stress and PCD. Recently, Guan et al. (2015) also showed that ER chaperone binding protein BiP acts as a positive regulator in Cd stress tolerance. To explore the mechanism, the author also examined the transcript level of GSH gene in LcBiP-overexpressed tobacco. Interestingly, the transcript levels of the GSH and the ER stress marker gene are not induced, meaning that BiP may enhance the folding capacity of GSH in the ER in plants. In contrast, proteins that fail to achieve their native conformation and aggregate in ER are eliminated through the ERAD pathway via a series of tightly coupled steps that include substrate recognition, targeting of terminally misfolded proteins, retro-translocations, and ubiquitin-dependent proteasomal destruction (Liu and Howell, 2016). Liu et al. (2015b) discover a plant-specific component of ERAD system in Arabidopsis. They uncovered that EBS7 (methanesulfonate-mutagenized brassinosteroid insensitive 1 suppressor 7) interacts with the ER membrane-anchored ubiquitin ligase AtHrd1a, one of the central components of the Arabidopsis ERAD machinery, whose mutation destabilizes AtHrd1a to reduce polyubiquitination. Similarly, very recently, Liu et al. (2016) demonstrated that α1, 6-linked mannose is necessary for luminal N-glycoproteins degradation via ERAD system in yeast where Htm1p-Pdi1p complex acts as a foldingsensitive mannosidase for catalyzing the first committed step. Thus, any defects in this sophisticated ERAD machinery system raise death and life issues of cells survival, especially under stressed environments. For instance, Van Hoewyk (2016) demonstrated that Arabidopsis HRD1 and SEL1L mutant plants showed decreased tolerance to selenate (Se) stress. As-Se toxicity causes both oxidative stress and protein misfolding due to the substitution of a cysteine to Se-cysteine (Van Hoewyk, 2013), whereas selenium enhances Cd tolerance in tomato plants (Li et al., 2016b). Actually, this is an alternative adaptive mechanism, involved in actively controlled and precise degradation of cellular components, and selective elimination of harmful, unwanted or damaged cells in eukaryotes (Tuzhikov et al., 2011; Brandizzi et al., 2014). Such selective suicide or PCD is paradoxically a crucial event which eventually provides survival benefits for the whole organism under extreme environmental conditions like HMs stress (Yakimova et al., 2007; Adamakis et al., 2011). However, the details of involved mechanisms are still obscure and remain to be further disclosed.

#### DEGRADATION OF METAL-INDUCED DENATURED PROTEINS

Much of plant physiological processes, growth and development are controlled by the selective degradation of unwanted misfolded or damaged proteins in order to maintain cellular homeostasis. In plants, protein degradation or proteolysis occurs either by ubiquitin proteosome system (UPS) or by autophagosomes induction (Liu and Howell, 2016). The UPS is a fundamental, highly regulated multistep enzymatic cascade that tightly controls the cellular protein homeostasis. It is a very rapid and effective method for a precise degradation of an unwanted protein at a particular time, whereas most of the times a protein is degraded only in response to a specific cellular signal or event (Pines and Lindon, 2005). The recent progress in molecular genetics revealed that like a tip of iceberg, UPS components regulate critical processes in plants (Sadanandom et al., 2012). It has been found that over 6% of Arabidopsis protein-coding genes are dedicated to the UPS (Vierstra, 2009). In eukaryotic cells proteins that destined to degrade, firstly, labeled by ubiquitin and then the ubiquitylated protein digested to small peptides by the large proteolytic complex, the 26S proteasome (Goldberg, 2003). Ubiquitylation is an energy dependent key regulatory process, which is executed by three different E1-E2-E3 enzymes conjugation cascade (Maupin-Furlow, 2013). In the first step of this cascade, ubiquitin is covalently conjugated to ubiquitin-activating enzyme (E1) in an ATP-dependent reaction, and then this activated ubiqutin is transferred from E1 to E2 (or Ub-conjugating enzyme) by transesterification. Finally, E3 (ubiquitin ligase) catalyzes the transfer of the ubiquitin from the E2 enzyme to a lysine residue on the target protein. After initial polyubiquitination, this step is repeated to form polyubiquitinated chains on the target protein and designated for degradation by the 26S proteasome (Ruschak et al., 2011; Sadanandom et al., 2012). Upon delivery, the deubiquitinating enzymes remove the polyubiquitin chain before unfolding, import and proteolysis of targeted substrates (Hartmann-Petersen et al., 2003).

On the other hand, autophagy is a biological self-destruction process, by which eukaryotic cells maintain their cellular homeostasis by turning over damaged proteins or organelles into the vacuole during developmental transitions and under stress conditions (Liu and Bassham, 2012; Wen-Xing, 2012). Upon the induction of an autophagic pathway, the cytoplasmic components that designate to degradation, are surrounded by a double membrane structure, called autophagosomes (**Figure 2**). The autophagosome then delivers its cargo materials to the vacuole, where the outer membrane of the autophagosome initially fuses with the vacuole membrane, after which the cargo materials are degraded by vacuolar hydrolases in the vacuole and recycled (Yang et al., 2016). There are four different types of autophagy detailed out in eukaryotes, including microautophagy, macroautophagy, chaperone-mediated autophagy and organellespecific autophagy (Liu and Bassham, 2012). Out of these, microautophagy and macroautophagy (hereafter referred as autophagy), have been shown to occur in plants (Han et al., 2011).

#### UPS-Dependent Proteasome Activity and Metals Stress

Extreme environmental conditions such as HMs pollution often adversely affect proteins quality by increasing free radicals that encourage denaturation and damage, thus causing the misfolding of proteins. Cells under stress, need to prevent further damage by initiating defense machinery to repair the damaged proteins or remove them when irreparable. In such circumstances, the UPS plays a crucial role in plant response and adaptation to changing environmental conditions(Stone, 2014). The UPS, functions both in cytoplasm and nucleus, responsible to modulate the levels of regulatory proteins and to remove most abnormal peptides and short-lived cellular regulators which may accumulate following exposure to abiotic stress (Lyzenga and Stone, 2012). The significant role of the UPS in the cellular response to HM stress has been recognized few years ago and is evident to increased expression of polyubiquitin genes (Genschik et al., 1992; Jungmann et al., 1993). It is believed that the expression of polyubiquitin genes under stress conditions is one of the important indications that the UPS is involved in regulation of plant HMs stress tolerance (Sun and Callis, 1997; Chai and Zhang, 1998). The genome-wide transcription analysis of rice plants showed that low concentrations of Cd treatment induces polyubiqutin expression both in root and shoot (Oono et al., 2016). In contrast, elevated metal concentrations induce the disruption of proteasomal activity, resulting in the accumulation of abnormal proteins in the cytosol, which alter cellular protein homeostais and thereby activating apoptosis (Yu et al., 2011). For instance, proteome analysis in different plant species has shown that ubiquitin activity can significantly be reduced by Cd, Co, Cu, Cr, Hg, Ni and Pb at 100µM but not by Al and Zn, whereas low concentrations can induce 26S proteosome activity (Aina et al., 2007; Pena et al., 2007, 2008). Although Co, Cu, Cr, Hg, Ni, and Pb induce accumulation of ubiquitin conjugated proteins, the abundance of 20S core protein in UPS system is not changed (Pena et al., 2008). In contrast, Karmous et al. (2014) showed that Cu treatment (200µM) has a strong effect on UPS pathway and can inhibit about 88% of the 20S proteasome activity in the cotyledons of germinating bean seeds. This implies that the effect of HMs on proteolytic system can not be generalized; however, the functional impairment-induced decrease in proteases activities appears to be a common aspect of metal toxicity in plant.

In extreme environments, over-expression of genes involved in UPS cascade, enhances tolerance to multiple stresses without any adverse effects on growth and development in plants (Guo et al., 2008). For example, Lim et al. (2014) showed the E3 or ubiquitin ligase enzyme is an important regulator for the removal of aberrant proteins under metal stress. In addition, the heterogeneous expression of rice E3 ligase enzyme synthesis RING domain OsHIR1 gene in Arabidopsis has been found to be decreased with As and Cd accumulation in both root and shoot (Dametto et al., 2015). Although the mechanism is not clear, it could probably be regulated by its substrate protein, since the OsHIR1 protein positively interacts with the OsTIP4;1 related to As and Cd uptake. Therefore, the development of strategies to reduce metal uptake and translocation as well as to improve cellular protein homeostasis with the ubiquitin/proteasome 26S system in plants seems to be a promising approach to ensure crop yield and food safety.

#### Autophagy in Metal Stress Responses

Autophagy has shown to be involved in the adaptation of plants to a wide range of drastic environmental stresses such as nutrient starvation, oxidative stress, heat stress, drought, salt, and pathogen invasion (Han et al., 2011; Wang et al., 2015; Xu et al., 2016; Yang et al., 2016). However, its pivotal roles in plants, particularly in HMs stress and adaptive responses, have perhaps not received the attention they deserve and thus remains elusive (Chiarelli and Roccheri, 2012; Pérez-Martín et al., 2015). Fascinatingly, in recent years scientists begun to explore, the involvement of autophagy in plant toward metal tolerance and the mechanism of adaptation is the same as in human and yeast. Firstly, Zhang and Chen (2010), afterwards, Zheng et al. (2012) demonstrated that autophagy is induced in metal treated plants. In progression, the expression profile analysis in tobacco seedlings after five different HMs (Cu, Ni, Zn, Cd and Mn) treatments showed that among the 30 ATGs genes, 18 ATGs genes are upregulated by more than two folds by at least one HM. They also explored that among the 18 ATGs, 11 ATGs are commonly up-regulated in seedlings by all five metals, and the expression is more sensitive to Zn treatment than others. Recently, Abd-Alla et al. (2016) for the first time demonstrated that Ag-NPs treatments result in the induction of autophagy in root nodules of Rhizobium leguminosarum as a mechanism of detoxification and surveillance. Taken together, recent studies explored the involvement of autophagy as a sophisticated regulator of surveillance under HMs stress, nevertheless, the mechanisms, especially how metals regulate autophagy, still remain to be elucidated in the future (Zhou et al., 2015).

Interestingly, the cellular sites of ROS production and signaling are thought to be primary targets of autophagy, which leads to either survival or death of cells (Scherz-Shouval and Elazar, 2007; Minibayeva et al., 2012). It has now been undoubtedly established that ROS can function as specific second messengers in signaling cascade. Environmental toxicants, like HMs are known to be strong inducers of oxidative stress due to excessive accumulation of ROS that alter the cellular homeostasis. In general excess accumulation of ROS causes many types of cellular injuries including damage to proteins, lipids and DNA, whereas some of which can result in apoptotic cell death and autophagy (Farah et al., 2016). Several lines of evidence suggest that metal-induced intracellular ROS production function in the signal transduction pathways, leading to induction of autophagy (Zhang and Chen, 2010; Pérez-Martín et al., 2015; Farah et al., 2016). Although an extensive number of studies showed that oxidative stress stimulates the autophagic induction to relieve the plants from oxidative damage, but the mechanistic information still remain limited. Very recently, Yang et al. (2016) put forward an excellent effort to open-up the mechanism for activation of autophagy. They demonstrated that unfolded proteins accumulation in the ER is a trigger for autophagy under conditions that cause ER stress. They showed that the reduction of unfolded proteins accumulation by PBA or TUDCA addition or BiP over-expression inhibits the autophagy in Arabidopsis. Whereas, the introduction of the constitutively unfolded proteins zeolin or CPY∗ activates the UPR and autophagy via IRE1b (inositol-requiring enzyme 1b) dependent manner. But how the ER stress activating these signaling cascades remains to be further revealed.

In addition to their core function, autophagic induction has also been shown to be involved in the regulation of metal uptake. For instance, Li et al. (2015) showed that induction of autophagy with mono-ubiquitination under Fe excess condition affects the functional activity and stability of exogenous Malus xiaojinensis iron-regulated transporter (MxIRT1) in yeast, thereby preventing Fe uptake via this root transporter. They also showed that in Fe led conditions, the transcript levels of ATG8 and ATG8-PE protein significantly increased, which resulted in enhanced MxIRT1 degradation, while the inhibition of autophagic initiation has opposite effects. Therefore, the development of strategies to regulate metal uptake by promoting autophagy under excess metal conditions could have potential implication in increased or even safe food production. However such kind of assumption in plant is still a matter of speculation and thus requires further extensive investigation.

#### CONCLUDING REMARKS AND FUTURE PERSPECTIVES

The present review outlines the impact of HM stress on cellular protein homeostasis and illustrates the diverse mechanistic approach that operates inside cells to regulate quality control systems toward functional and healthy proteomes. Proteins are major workhorse of cells and directly involved in plant stress response. HMs can trigger the cellular pathways that are broadly classified as death and survival signals. As surveillance mechanism, the ubiquitous plants response to HM stress is the chelation of toxic ions in the cytosol by cysteine rich peptides such as PCs and MTs, compartmentalization of metals in the vacuole by tonoplast located transporters, and the process that involves repair of stress-damaged proteins. In-depth review of recent research works revealed that MTs are not only required to complete the plants life cycle, but also play significant roles in ionic homeostasis and distribution in plants as well as cleanup of ROS and sequestration of metals as that of PCs (Benatti et al., 2014; Tomas et al., 2016). Whilst in extreme conditions metals profoundly affect cellular protein homeostasis by interfering with the folding process, they also stimulate aggregation of nascent or non-native proteins leading to ER stress and decreased cell viability. However, there is a typical set of proteins, called stress

#### REFERENCES


proteins or HSPs proteins, which are preferentially expressed under stress. HSPs restrict aggregation of nascent or nonnative proteins but trigger repair of misfolded proteins. In contrast, the damaged proteins which fail to achieve their native conformations, are removed from ER by the activation of ERAD machinery of ERQC system, leading to proteosomal (UPS) or autophagic degradation.

Recent advances in protein research, summarized herein, show that as core degradation process of misfolded or damaged polypeptides, the over-expression of E3 enzyme in UPS pathway and the autophagic induction with mono-ubiquitination prevent metal accumulation in plants (Dametto et al., 2015; Li et al., 2015). However, we are still far away from the complete understanding of the mechanism of subsequent signaling cascades that regulate metal accumulation. The development of strategies to reduce metal uptake and translocation by manipulating cellular protein quality control system in plants seems to be a promising approach that can potentially ensure increased yield as well as food safety.

#### AUTHOR CONTRIBUTIONS

MH and GA conceived the idea and designed the outlines of the article. MH, YC, MK and GA wrote the article. GA, XC, and ZQ revised the article.

#### ACKNOWLEDGMENTS

Research in the authors' laboratories is supported by grants from the National Key Research and Development Program of China (2016YFD0201001), the Geological Exploration Foundation of Zhejiang Province, China (2014002-02) and the National Natural Science Foundation of China (31401877, 31550110201, 31772294), and the China Postdoctoral Science Foundation (517000-X91608, 2015M580515).

seedlings exposed to acute cadmium toxicity. C. R. Biol. 330, 735–746. doi: 10.1016/j.crvi.2007.08.001


cadmium, hydrogen peroxide and ethanol tolerance. Braz. J. Microbio. 48, 537–543. doi: 10.1016/j.bjm.2016.10.024


leads to hypersensitivity to cadmium stress. Plant Physiol. 131, 656–663. doi: 10.1104/pp.014118


and rice as putative co-chaperones of Hsp90/Hsp70. PLoS ONE 5:e12761. doi: 10.1371/journal.pone.0012761


substrate preferences in barley and Arabidopsis. Plant Cell Env. 37, 1192–1201. doi: 10.1111/pce.12227


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

Copyright © 2017 Hasan, Cheng, Kanwar, Chu, Ahammed and Qi. 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) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Genome-Wide Identification and Expression Profiling of Tomato *Hsp20* Gene Family in Response to Biotic and Abiotic Stresses

Jiahong Yu1, 2 †, Yuan Cheng1 †, Kun Feng<sup>1</sup> , Meiying Ruan<sup>1</sup> , Qingjing Ye<sup>1</sup> , Rongqing Wang<sup>1</sup> , Zhimiao Li <sup>1</sup> , Guozhi Zhou<sup>1</sup> , Zhuping Yao<sup>1</sup> , Yuejian Yang<sup>1</sup> and Hongjian Wan<sup>1</sup> \*

*<sup>1</sup> State Key Laboratory Breeding Base for Zhejiang Sustainable Pest and Disease Control, Institute of Vegetables, Zhejiang Academy of Agricultural Sciences, Hangzhou, China, <sup>2</sup> College of Chemistry and Life Science, Zhejiang Normal University, Jinhua, China*

#### *Edited by:*

*Jie Zhou, Zhejiang University, China*

#### *Reviewed by:*

*Chu-Yu Ye, Zhejiang University, China Jianfei Kuang, South China Agricultural University, China*

*\*Correspondence:*

*Hongjian Wan wanhongjian@sina.com*

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

#### *Specialty section:*

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

*Received: 12 June 2016 Accepted: 02 August 2016 Published: 17 August 2016*

#### *Citation:*

*Yu J, Cheng Y, Feng K, Ruan M, Ye Q, Wang R, Li Z, Zhou G, Yao Z, Yang Y and Wan H (2016) Genome-Wide Identification and Expression Profiling of Tomato Hsp20 Gene Family in Response to Biotic and Abiotic Stresses. Front. Plant Sci. 7:1215. doi: 10.3389/fpls.2016.01215* The Hsp20 genes are involved in the response of plants to environment stresses including heat shock and also play a vital role in plant growth and development. They represent the most abundant small heat shock proteins (sHsps) in plants, but little is known about this family in tomato (*Solanum lycopersicum*), an important vegetable crop in the world. Here, we characterized heat shock protein 20 (*SlHsp20*) gene family in tomato through integration of gene structure, chromosome location, phylogenetic relationship, and expression profile. Using bioinformatics-based methods, we identified at least 42 putative *SlHsp20* genes in tomato. Sequence analysis revealed that most of *SlHsp20* genes possessed no intron or a relatively short intron in length. Chromosome mapping indicated that inter-arm and intra-chromosome duplication events contributed remarkably to the expansion of *SlHsp20* genes. Phylogentic tree of *Hsp20* genes from tomato and other plant species revealed that *SlHsp20* genes were grouped into 13 subfamilies, indicating that these genes may have a common ancestor that generated diverse subfamilies prior to the mono-dicot split. In addition, expression analysis using RNA-seq in various tissues and developmental stages of cultivated tomato and the wild relative *Solanum pimpinellifolium* revealed that most of these genes (83%) were expressed in at least one stage from at least one genotype. Out of 42 genes, 4 genes were expressed constitutively in almost all the tissues analyzed, implying that these genes might have specific housekeeping function in tomato cell under normal growth conditions. Two *SlHsp20* genes displayed differential expression levels between cultivated tomato and *S. pimpinellifolium* in vegetative (leaf and root) and reproductive organs (floral bud and flower), suggesting inter-species diversification for functional specialization during the process of domestication. Based on genome-wide microarray analysis, we showed that the transcript levels of *SlHsp20* genes could be induced profusely by abiotic and biotic stresses such as heat, drought, salt, *Botrytis cinerea,*

**24**

and Tomato Spotted Wilt Virus (TSWV), indicating their potential roles in mediating the response of tomato plants to environment stresses. In conclusion, these results provide valuable information for elucidating the evolutionary relationship of *Hsp20* gene family and functional characterization of the *SlHsp20* gene family in the future.

Keywords: heat shock protein 20, gene organization, phylogenetic relationship, expression profile, abiotic and biotic stresses

#### INTRODUCTION

Plants live in a complex environment, where multiple biotic and abiotic stresses may seriously restrict their growth and development (Cramer et al., 2011). In the recent years, due to unprecedented global warming caused by various factors, high temperature has appeared as one of the most severe abiotic stresses around the world. To survive and acclimatize under the adverse environment conditions, plants have established self-defense mechanisms during the course of long-term evolution. Heat shock proteins (Hsps), acknowledged as evolutionarily conserved ubiquitous proteins in all organisms, were first discovered in Drosophila melanogaster in response to the elevated temperature stress (Ritossa, 1962; Neta-Sharir et al., 2005; Cashikar et al., 2006). Previous studies have shown that high temperature as well as other environmental cues (cold, salinity, drought, heavy metals, anoxia, pathogens, etc.) could induce the occurrence of Hsps (Lindquist and Craig, 1988; Wang et al., 2003). In addition, the Hsps were also found to be associated with plant growth and development, such as embryogenesis, seed germination, and fruit maturation (Neta-Sharir et al., 2005).

According to sequence homology and molecular weight, Hsps in eukaryotes can be grouped into six families such as Hsp110, Hsp90, Hsp70, Hsp60, small heat shock protein (sHsp) of 15–42 kDa (or Hsp20) and ubiquitin (Carper et al., 1987; Sarkar et al., 2009). Out of these six groups of Hsps, sHsps are the most primary and abundant proteins under thermal stimulation in many species (Vierling et al., 1989; Vierling, 1991). Notably, among eukaryotes, the higher plants possessed more quantities of Hsp20s (Vierling, 1991). As Hsp20 is encoded by a multigene family, it is considered as the most ample and complicated member in Hsps (Vierling, 1991). The characteristic feature of Hsp20 is the presence of a carboxylterminal conserved domain of 80–100 amino acid residues, which can be defined as the α-crystallin domain (ACD). This highly conserved ACD, which is flanked by a short carboxyl-terminal extension and a variable amino-terminal domain, is believed to comprise two hydrophobic β-sheet motifs that are separated by a hydrophilic α-helical region of variable length (de Jong et al., 1998). Moreover, the Hsp20 gene family also exhibits extensive sequence variability and evolutionary divergence, which is remarkably different from other families of Hsps (Basha et al., 2012).

An earlier study showed that Hsp20 proteins in eukaryotes, which are collectively known as molecular chaperones, function as multimeric complexes ranging from 8 to 24 or more subunits (Van Montfort et al., 2002). These chaperons can selectively bind to partially folded or denatured proteins in an ATPindependent manner, which can prevent proteins from the irreversible aggregation and facilitate them folding properly (Lee and Vierling, 2000; Sun et al., 2002; Van Montfort et al., 2002). Recent studies revealed that these chaperones were important for disease resistance triggered by resistance (R) proteins and played a fundamental role in plant immunity (Botër et al., 2007; Shirasu, 2009).

To date, the Hsp20 gene families have been investigated in several plant species, including Arabidopsis, rice, soybean, pepper, and Populus trchocarpa (Scharf et al., 2001; Waters et al., 2008; Ouyang et al., 2009; Sarkar et al., 2009; Lopes-Caitar et al., 2013; Guo et al., 2015). In addition, some key features of Hsp20 and biologic function of several Hsp20 genes had been identified (Nautiyal and Shono, 2010; Goyal et al., 2012; Huther et al., 2013; Mahesh et al., 2013; Arce et al., 2015; Zhang et al., 2016). Although the availability of the tomato whole-genome sequence provides valuable resources for getting into an in-depth understanding of Hsp20s (Sato et al., 2012), little information is available on the integrated Hsp20 family at whole genomic level in tomato.

In the current paper, the members of SlHsp20 gene family in tomato were identified using a bioinformatics method and characterized by integration of sequence features, chromosome location, phylogenetic relationship, evolutionary origin, and expression patterns. These results provide valuable information that can be implicated in elucidating the evolutionary relationship of Hsp20 gene family in higher plants and functional characterization of the SlHsp20 gene family in the future.

#### MATERIALS AND METHODS

#### Retrieval and Identification of *Hsp20* Genes in Tomato

In this paper, the predicted SlHsp20 genes were identified as follows: firstly, the tomato genome sequences were downloaded from the database Sol Genomics Network (SGN, Release 2. 5, http://solgenomics.net/) and used to set up a local database by the software "DNATOOLs." Secondly, the Hidden Markov Model (HMM) profile of Hsp20 domain (PF00011) from PFam (http://pfam.sanger.ac.uk/) was employed to search against the local database using BlastP method (e < 1e-5). Furthermore, Hsp20 candidates with incomplete Hsp20 domain might be missed using HMM profile. Name search using the word "hsp20" as a keyword also applied to retrieve in SGN database. The redundant sequences were manually removed. Finally, all these predicted genes were examined for the Hsp20 domain in SMART (http://smart.embl-heidelberg.de/smart/batch.pl), Pfam and InterProScan (http://www.ebi.ac.uk/interpro/), and those without the common Hsp20 domain were excluded.

### Sequence Analysis and Structural Characterization

Information of candidate SlHsp20 genes was obtained via searching the SGN database (http://solgenomics.net/search/ locus), including chromosome locations, intron numbers, genomic sequences, coding sequences (CDS), and amino acid sequences. Intron-exon structure was determined by alignment of genome DNA and full-length cDNA sequence using Gene Structure Display Server 2.0 (http://gsds.cbi.pku.edu.cn/) (Hu et al., 2015). Molecular weight, theoretical isoelectric point (theoretical pI), and instability index (II; with the value >40 classified as unstable) of SlHsp20 proteins were analyzed by using ProtParam tool (http://web.expasy.org/protparam/).

The putative protein sequences were subjected to MEME program (http://meme-suite.org/tools/meme) to investigate conserved motifs with the following parameters: site distribution—any number of repetitions, number of motifs—10, the motif width between 6 and 200.

### Phylogenetic Analysis

To illuminate evolutionary relationship of Hsp20 gene family, the representative Hsp20 genes from Arabidopsis, rice, soybean, bluebunch wheatgrass, barley, common wheat, Eurasian aspen, together with SlHsp20 genes from tomato, were selected for constructing phylogenetic tree (Ouyang et al., 2009; Sarkar et al., 2009; Lopes-Caitar et al., 2013). Multiple sequence alignment of Hsp20 proteins was conducted using ClustalX 1.83. An unrooted Neighbor-joining phylogenetic tree was constructed using MEGA 7.0 software with default settings (Kumar et al., 2016). The bootstrap test was performed by 1000 replications.

Four Online tools were employed to predict subcellular localization, including, Predotar (https:// urgi.versailles.inra.fr/Tools/Predotar), Wolf Psort (http://www. genscript.com/psort/wolf\_psort.html), TargetP (http://www. cbs.dtu.dk/services/TargetP/), and MultiLoc (http://abi.inf.unituebingen.de/Services/MultiLoc2). The prediction of signal peptide and transmembrane domain was performed with SMART program (http://smart.embl-heidelberg.de/smart/ batch.pl).

#### Chromosome Localization and Gene Duplication

Chromosome mapping of the candidate SlHsp20 genes was viewed using the software MapDraw V2.1 (Liu and Meng, 2003). Tandem duplication and segmental duplication were also further investigated. The former was confirmed with the following criteria: (1) an array of two or more SlHsp20 genes within a range of 100 kb distance; (2) the alignment had a coverage rate more than 70% of the longer gene; (3) and the identity of the aligned region was no less than 70% (Li et al., 2010; Huang et al., 2012; Wei et al., 2016). The latter was identified based on Plant Genome Duplication Database (PGDD, http://chibba.agtec.uga. edu/duplication/index/locus).

#### Tissue-Specific Expression Analysis

In this study, RNA-seq data from Tomato Functional Genomics Database (TFGD, http://ted.bti.cornell.edu/cgi-bin/TFGD/ digital/home.cgi) were used to investigate expression patterns of putative SlHsp20 genes in different tissues of cultivated tomato (Solanum lycopersicum) and the wild relative (Solanum pimpinellifolium). Different tissues in cultivated tomato included leaves, roots, flower buds, fully opened flowers, 1 cm, 2 cm, 3 cm, mature green, breaker, and breaker+10 fruits. In the wild species (S. pimpinellifolium), nine tissues and organs were selected for analysis, including leaves, whole root, hypocotyl, cotyledons, flower buds 10 days before anthesis or younger, flowers at anthesis, 10 days post anthesis (DPA) fruit, 20 DPA fruit and breaker stage ripening fruit. Digital gene expression analysis of the putative SlHsp20 gene family was performed using software MultiExperiment Viewer (MeV) (Howe et al., 2010).

#### Expression Profile of *SlHsp20* Genes under Different Stress Conditions

To get insight into the expression profiles of the SlHsp20 gene family under different environmental stresses, microarray analysis was performed. Whole genome microarray data for diverse environment stresses such as heat, drought, salt, Botrytis cinerea, and Tomato Spotted Wilt Virus (TSWV) was downloaded from database TFGD (http://ted.bti.cornell.edu/). The array platforms for microarray data included TOM2 oligo array and Affymetrix genome array. For TOM2 oligo array, the probe sets of the SlHsp20 genes were identified through BlastN analysis in the database "TOM2 oligo sequences." For Affymetrix genome array, Probe Match tool in NetAffx Analysis Center (http://www.affymetrix.com) was used to obtain probe sequences. Average value was considered for SlHsp20 genes that had more than one probe set. The expression values of SlHsp20 genes that were up- or down-regulated more than two-fold with P < 0.05 were considered as differently expressed.

## RESULTS

### Identification of Hsp20 Family Members in Tomato

Name search and HMM analysis showed a total of 42 candidate SlHsp20 genes, four of which were identified to contain incomplete Hsp20 domains. For convenience, the SlHsp20 genes were named according to their molecular weight in our study. Details on gene name, locus name, chromosome location, open reading frame (ORF) length, intron number, protein length, molecular weight, isoelectric point (pI), and instability index were listed in **Table 1**.

The four SlHsp20 genes with incomplete Hsp20 domain encoded truncated proteins (67–129 aa) and could be nonfunctional or pseudogenes. Therefore, these SlHsp20 genes were excluded in phylogenetic tree construction. Molecular weight of the remaining predicted SlHsp20 genes ranged from 15.2 to 49.3 kDa, except for SlHsp11.9 and SlHsp14.5

#### TABLE 1 | Features of *SlHsp20* genes in tomato.


that were less than 15 kDa. Molecular weights of these SlHsp20 proteins had a large variation. Isoelectric points ranged from 4.53 (SlHsp9.0) to 9.8 (SlHsp23.8A), and protein length ranged from 67 (SlHsp7.8) to 441 aa (SlHsp49.3). The instability index indicates that 18 of the predicted SlHsp20 proteins were deemed to be stable proteins, while others were unstable.

#### Gene Structure of the *SlHsp20* Genes

Structure and phases of introns/exons were determined by alignment of genomic DNA and full-length cDNA sequences of SlHsp20 genes (**Supplementary Figure S1**). This information was available on Sol Genomics Network (**Supplementary Table S1**). It was found that among the total 42 SlHsp20 genes, 13(30.95%) were noted to be intronless, while 22 genes (52.38%) had one intron, and only 7(16.67%) had two or more introns. Based on the number of introns, we divided the genes into three patterns (pattern one has no intron, pattern two has one intron, and pattern three has more than one) (Ouyang et al., 2009). It was evident that most SlHsp20 genes belonged to pattern 1 and 2. Additionally, a relatively short intron length was found in SlHsp20 proteins in tomato, which was similar to the results reported on sHsp20 proteins in rice (Sarkar et al., 2009).

Using the MEME tool, 10 putative conserved motifs (motif 1 to motif 10) in SlHsp20 gene family were identified (**Supplementary Table S2**). The lengths of these conserved motifs varied from 15 to 57 aa (P < 0.0001). Details of all the putative motifs are outlined in **Table 2**. Among the 10 motifs, motif 2 appeared in all putative SlHsp20 genes, except for SlHsp15.5. Based on the analysis of Pfam, the full sequences of motif 1, 2, 3, 4, 5, and 8 were corresponding to the region of conserved ACD. We also found that motif 1 was in the C-terminal regions, while motif 6 appeared in the N-terminal regions generally. In addition, motif 9 was also distributed in the C-terminal regions of several genes. Notably, four SlHsp20 genes (SlHsp11.9, SlHsp17.3A, SlHsp14.5, and SlHsp15.8) present similar patterns of motif distribution. Also, the same scenario was found in other four SlHsp20 genes (SlHsp17.6C, SlHsp17.7A, SlHsp17.6B, and SlHsp17.6A) on chromosome 6. These results suggest that these genes may have relatively high conservation. Nevertheless, the functions of these highly conserved amino acid motifs still remain elusive.

It was reported that highly conserved ACD domain might be associated with the formation of multimeric complexes that are crucial for the chaperone activity of Hsp20 genes (Waters et al., 1996; Sun et al., 2002). In the current paper, in line with a previous report (Waters et al., 1996), the highly conserved ACD in SlHsp20 proteins was divided into two parts (consensus I and consensus II) based on multiple sequence alignment (**Figure 1**). These two conserved regions were separated by a hydrophilic domain with variable length and characterized by residue Pro-X (14)-Gly-Val-Leu and Pro-X (14)-X-Val/Leu/Ile-Val/Leu/Ile, respectively (Caspers et al., 1995). Intriguingly, consensus I matched well with the motif 1 located on carboxyl terminal presumed by MEME.

TABLE 2 | List of the putative motifs of SlHsp20 proteins.


### Phylogenetic Analysis of *SlHsp20* Gene Family

An unrooted N-J phylogenetic tree was constructed from a complete alignment of amino acid sequences of Hsp20 proteins in tomato and other seven plant species (**Figure 2**). All these Hsp20 proteins were grouped into 17 distinct subfamilies. However, the SlHsp20 proteins were distributed into 13 out of 17 subfamilies, including previously identified subfamilies like CI, CII, CIII, P, Px, ER, and MI (Waters et al., 1996), recently defined CV, CVI, CVII, CIX, and CXI subfamilies (Siddique et al., 2008; Sarkar et al., 2009) and a new subfamily, CXII, identified by in silico prediction of subcellular localization (**Supplementary Table S3**). Besides, we also found two orphan SlHsp20 genes in tomato that lack homologs genes in all seven organisms. The SlHsp20 genes from 13 subfamilies were distributed to a variety of cellular organelles: 29 were nucleo-cytoplasmic (C) SlHsp20 genes (9 subfamilies), 3 were endoplasmic reticulum (ER) genes, 2 were plastidial (P) genes, and 1 each for mitochondrial (M) and peroxisomal (Px) genes.

Notably, all three SlHsp20 genes belonging to ER subfamily have a signal peptide in the N-terminal region, which was consistent with the result showed by Lopes-Caitar et al. (2013) (**Table 1**). The signal peptides were reported to play a positive role in facilitating the process of protein synthesis via guiding the proteins into rough endoplasmic reticulum (Bauvois, 2012). Moreover, eight SlHsp20 genes, one from CVI subfamily and seven from CXII, had a transmembrane domain in the Cterminal region (**Table 1**).

We also found a close relationship between intron pattern and phylogenetic classification (**Supplementary Figure S1**). The result showed that the CI, CII, and ER subfamilies lacked introns. All the CVII members had more than one intron, indicating a particular phylogenetic status (Ouyang et al., 2009). The CIII, CV, CVI, CXI, MI, and Px subfamilies, together with most members of the CXII subfamily, had one intron, which may indicate a close phylogenetic relationship. In addition, gene structure may provide clues for evolutionary relationship of SlHsp20 family.

#### Chromosomal Localization and Gene Duplication

Out of the 42 predicted SlHsp20 genes, 41 are randomly distributed across the 12 tomato chromosomes, except for the SlHsp11.9 (**Figure 3**). Majority of the SlHsp20 genes were located on the distal ends of the chromosomes and mainly on the lower arms. A maximum number of eight predicted SlHsp20 genes scattered in three clusters, were present on chromosome 1.

We further performed chromosome mapping to determine the gene duplication of SlHsp20 genes on the 12 tomato chromosomes. As shown in **Figure 3**, two groups of SlHsp20 genes (SlHsp49.3/SlHsp39.4 and SlHsp17.7A/SlHsp17.6A/ SlHsp17.6B/SlHsp17.6C) can be identified as tandem duplication genes (**Supplementary Table S4**). One group (SlHsp49.3 and SlHsp39.4) was from the CXII subfamily and located on chromosome 1. The other (SlHsp17.7A, SlHsp17.6A, SlHsp17.6B, and SlHsp17.6C) was from one branch of CI subfamily and juxtaposed compactly on chromosome 6, implying that the

amino acid residues within these regions are indicated by asterisks.

high density of SlHsp20 genes on this chromosome was mainly due to the tandem duplication events. The chromosome location of tandemed genes showed that these four pairs of SlHsp20 genes (SlHsp49.3/SlHsp39.4, SlHsp17.7A/SlHsp17.6A, SlHsp17.6A/SlHsp17.6B, SlHsp17.6B/SlHsp17.6C) was intervened by no more than one gene and was close in distance on the chromosomes (approximately 4.6, 3.8, 8.2, and 3.9 kb) (**Supplementary Table S5**).

On the other hand, three segmental duplication groups were found to scatter in seven chromosomes (**Figure 3**; **Supplementary Table S6**). A duplicated segment of the SlHsp37.0 region on the distal part of chromosome 4 was present on the same location of chromosome 12, where SlHsp27.2 was located. SlHsp25.7A on chromosome 1 also showed synteny to SlHsp27.5 localized on a duplicated segment of chromosome 11. In addition, three SlHsp20 genes (SlHsp17.7A, SlHsp24.5, and SlHsp27.1) regions showed segmental duplication and were present on chromosome 6, 9, and 10, respectively.

#### Genome-Wide Expression Analysis of *SlHsp20* Genes

In silico expression patterns of the putative SlHsp20 genes at different tissues and development stages of tomato cultivar Heinz and the wild species S. pimpinellifolium were analyzed (**Figure 4**). It showed that most of the genes (83.3%) were expressed in at least one stage (tissue) from at least one genotype. Fourteen genes (SlHsp17.7A, SlHsp17.6B, SlHsp17.6C, SlHsp24.5, SlHsp18.2, SlHsp16.1A, SlHsp17.7B, SlHsp17.6A, SlHsp25.7B, SlHsp15.2, SlHsp21.6A, SlHsp17.6D, SlHsp16.1B, SlHsp23.8B) were expressed constitutively in all the stages analyzed, whereas the transcripts of 12 genes (SlHsp23.8A, SlHsp17.3A, SlHsp17.9, SlHsp9.1, SlHsp39.4, SlHsp37.0, SlHsp21.5B, SlHsp11.9, SlHsp14.5, SlHsp15.8, SlHsp7.8, and SlHsp15.5) were at almost undetectable levels. Among these genes, SlHsp17.6B had the highest expression level in the 30 DPA fruit.

When the expression patterns of SlHsp20 genes in vegetative organs (leaf and root) and reproductive organs (flower bud and flower) were compared between the 2 tomato genotype, 17 genes were either highly-induced (3) or barely expressed (14) (**Figure 4**). Conversely, two genes (SlHsp25.7B and SlHsp17.6A) displayed significantly differential expression in the various genotypes. Further, seven genes exhibited varied expression in vegetative and reproductive organs of tomato cultivar Heinz, while only two of them expressed differentially in S. pimpinellifolium. Notably, expression of seven genes was restricted to the leaf (SlHsp26.5) and root (SlHsp23.7, SlHsp49.3,

SlHsp25.7, SlHsp26.8, SlHsp27.5, and SlHsp15.5) in Heniz, and only one gene was noted to root (SlHsp7.8) in S. pimpinellifolium. It indicates that these genes are regulated in a tissue-specific manner.

Expression levels of SlHsp20 genes at breaker stage fruits were higher than that in other development stages in both genotypes. In tomato cultivar Heniz, expression of several SlHsp20 genes (SlHsp25.7B, SlHsp26.2, SlHsp21.5A, SlHsp21.5C, and SlHsp15.6)

was not almost detected in young tomato fruits (1 cm-, 2 cm-, 3 cm-, and MG fruit), but a distinct expression was observed in the breaker fruits (**Figure 4A**). In S. pimpinellifolium, two SlHsp20 genes, SlHsp17.9, and SlHsp15.6, were only expressed in 20 DPA and 33 DPA fruits (**Figure 4B**). Furthermore, expression of three SlHsp20 genes (SlHsp23.8A, SlHsp23.7, and SlHsp39.4) derived from CXII subfamily showed tissue specificity in hypocotyl. Besides, by analyzing the expression profiles of tandem and segmental duplication of SlHsp20 genes in two genotypes, we found that two groups of tandemly SlHsp20 genes (SlHsp49.3/SlHsp39.4; SlHsp17.7A/SlHsp17.6A/ SlHsp17.6B/SlHsp17.6C) displayed a more similar expression pattern, while difference in expression was observed for SlHsp20 genes (SlHsp27.1/SlHsp17.7A/SlHsp24.5; SlHsp37.0/SlHsp27.2) in segmental duplication regions.

#### Expression Profiles of *SlHsp20* Genes Induced by Different Biotic and Abiotic Stresses

To further explore the expression profiles of SlHsp20 genes under various abiotic and biotic stresses, microarray analysis were performed (**Figure 5**). In the present paper, five tomato microarray data sets, belonging to two array platforms (TOM2 oligo array and Affymetrix genome array), were obtained from Tomato Functional Genomics Database (TFGD, http://ted.bti.cornell.edu/). A total of 24 probes (57%) corresponding to SlHsp20 genes were identified, while five probes (LE17D07, LE26P10, LE26F10, LE13N06, Les.4004.1.S1\_a\_at) showed cross-reactivity with 12 SlHsp20 genes (**Supplementary Table S7**).

Microarray-based expression analysis of tomato under various abiotic stresses revealed that expression of most of the SlHsp20 genes were highly variable (**Figures 5A–C**). Expression of 13 of all tested SlHsp20 genes, especially SlHsp25.7B, SlHsp15.2, SlHsp21.5C, and SlHsp16.1B, was drastically enhanced in resistant and susceptible tomato plants under high temperature condition, except for SlHsp15.7 that was down-regulated in susceptible plants in response to heat stress (**Figure 5A**). Half of the analyzed SlHsp20 genes in susceptible plants showed a higher expression level than those in tolerant plants. Under salt treatment condition, 12 SlHsp20 genes displayed highly elevated expression, whereas SlHsp15.7 was shown to be significantly down-regulated

(**Figure 5B**). PI365967, a more salt-tolerant tomato genotype, showed relatively more responsive genes compared to tomato cultivar Moneymaker. Under drought stress condition, transcript levels of 10 and 4 SlHsp20 genes were up-regulated and downregulated in drought-tolerant tomato lines (IL2-5 and IL9-1) and drought-sensitive cultivar M82, respectively (**Figure 5C**). Among these genes, SlHsp26.5 showed a drastic enhancement of the transcript level (more than 16 fold) in three tested tomato genotypes. The expression of SlHsp26.2 was dramatically upregulated in M82 (a drought-sensitive cultivar), which was much higher (four times) than that in IL2-5 and IL9-1.

Invoked by wound and the invasion of B. cinerea, transcript levels of three genes (SlHsp15.7, SlHsp23.8B, and SlHsp16.1B) were remained unaltered, up- and down-regulated levels, respectively, in all tested samples (**Figure 5D**). In mature green fruit, most of the SlHsp20 genes displayed a stronger expression in wounded fruits than that in fruits of wound-inoculated with B. cinerea. In red ripe fruit, however, the expression patterns of SlHsp20 genes showed a reverse pattern between the wounded and wound-inoculated with B. cinerea fruits. Interestingly, seven genes (SlHsp17.6D, SlHsp17.7A, SlHsp17.6A, SlHsp17.6B, SlHsp17.6C, SlHsp9.1, and SlHsp26.2) displayed differential expression between mature green- and red ripe fruits. In addition, under TSWV infection condition, three SlHsp20 genes (SlHsp11.9, SlHsp17.6D, and SlHsp27.1) were upregulated in tomato roots, three genes (SlHsp15.2, SlHsp17.7B, and SlHsp15.7) were down-regulated, and expression of the remaining genes remained unchanged (**Figure 5E**). In shoots of tomato, almost half SlHsp20 genes were enhanced, six genes were reduced and expression levels of the remaining six genes were unaltered.

#### DISCUSSION

#### Identification and Phylogentic Relationship of *SlHsp20* Gene Family

Using in silico methods to search for Hsp20 genes in the S. lycopersicum genome, at least 42 putative SlHsp20 genes were

increased (red) transcript levels relative to the respective control. (A) Expression profiles of *SlHsp20* genes in tolerant and susceptible tomatoes under heat stress condition. (B) Expression profiles of *SlHsp20* genes under salt stress in a wild tomato genotype "PI365967" (salt-tolerant) and cultivated tomato var. moneymaker (salt-sensitive). (C) Expression profiles of *SlHsp20* genes under drought stress condition in two drought-tolerant lines (IL2-5 and IL9-1) and a drought-sensitive cultivar (M82). (D) Expression profiles of *SlHsp20* genes infected by wound and wound-inoculated with *Botrytis cinerea* in mature green (Mg) and red fruits (Rr). (E) Expression profiles of *SlHsp20* genes in tomato leaves and roots infected by tomato spotted wilt virus (TSWV).

identified (**Table 1**). Among them, four predicted SlHsp20 genes were excluded in the phylogenetic tree construction due to incomplete Hsp20 domain. Previously, Hsp20 gene family in Arabidopsis was categorized into 12 subfamilies (CI, CII, CIII, CIV, CV, CVI, CVII, MI, MII, P, ER, and Px) (Scharf et al., 2001; Siddique et al., 2008). Likewise, four novel nucleocytoplasmic subfamilies (CVIII, CIX, CX, and CXI) were also reported on OsHsp20 genes in rice (Sarkar et al., 2009). In the present paper, to reveal phylogenetic relationship of Hsp20 genes, Hsp20 genes from seven plant species, together with Hsp20 genes from tomato, were used to construct phylogenetic tree (**Figure 2**). The results showed that Hsp20 genes from the tested plant species could be grouped into 17 subfamilies, according to the method described previously (Scharf et al., 2001; Siddique et al., 2008; Sarkar et al., 2009). The Hsp20 genes from tomato were grouped into 13 out of 17 subfamilies. A novel subfamily, CXII, was identified using in silico localization prediction, which was composed of 9 Hsp20 genes from tomato (**Supplementary Table S3**). No SlHsp20 genes were grouped into the remaining four subfamilies (CIV, CVIII, CX, and MII subfamilies). Previous study showed that CIV and MII subfamilies may play an important role in response to various stress conditions and were developmentally regulated (Siddique et al., 2008). Members of CVIII subfamily could be heat-induced and CX subfamily might be involved in specific housekeeping functions under normal growth conditions (Sarkar et al., 2009). Similarly, several subfamilies of the Hsp20 genes in pepper also lacks, including CIV, CV, CVIII, CIX, CX, and CXI subfamilies. (Guo et al., 2015). Furthermore, the CIV and CVII subfamilies in rice Hsp20 family were absent (Sarkar et al., 2009). Thus, we were tempting to speculate that gene gain and gene loss events were occurred widely in plant species. The lack of subfamilies exists not only in tomato, but also in other species. It also revealed a wide range of genetic diversity in dicotyledons and monocotyledons.

Furthermore, we also found that most of Hsp20 genes (69.0%) were clustered into nine nucleocytoplasmic subfamilies, which also had been described in Arabidopsis and rice (Scharf et al., 2001; Sarkar et al., 2009). Among these subfamilies, CXII was the largest subfamily with nine members. Therefore, we speculated that cytoplasm, as a site mainly for proteins to be synthesized, could be the primary place for Hsp20 interacting with denatured proteins to avoid them from inappropriate aggregation and degradation (Lopes-Caitar et al., 2013). In addition, most subfamilies included Hsp20 genes from multiple plant species and formed a mixture of groups, suggested that diversification of SlHsp20 subfamilies predated the divergence of these species from the ancestral species.

#### Organization of *SlHsp20* Genes

It was reported that gene organization played a vital role in the evolution of multigene family (Xu et al., 2012). Our results showed that SlHsp20 genes of CI, CII, and ER subfamilies had no intron (pattern one) (**Supplementary Figure S1**). Members of CIII, CV, CIX, MI, and PX subfamilies as well as most members of CXII subfamily had one intron (pattern two). Furthermore, these SlHsp20 genes within the same subfamily shared a similar motif arrangement (**Supplementary Table S2**). This correlation between motif arrangement, intron numbers and phylogeny can be served as an additional support to their classification.

The results also showed that most of SlHsp20 genes had no intron or one intron and the length of introns of SlHsp20 genes with one intron is relatively short. This is concurrent with the result reported previously by the researcher that plants were prone to retain more genes with no intron or a short intron (Mattick and Gagen, 2001). In addition, the instability index of most proteins was equal or greater than 40, signifying that SlHsp20 proteins can be identified as unstable proteins (**Table 1**). The instability is also considered as a common trait of stressactivated proteins and thus sheds a brilliant light on the rapid induction of the Hsp20 genes (Rao et al., 2008).

#### Evolution of the *SlHsp20* Gene Family

Previous research reported that Hsp20 proteins were ubiquitous from single-celled creatures including bacteria to higher organisms like human (Kim et al., 1998; Waters et al., 2008; Li et al., 2009). It implied that Hsp20 proteins may have evolved in early stages of the life's history which predated to the divergence of the three domains of life (Eukarya, Bacteria, and Archaea). However, the formation of individual Hsp20 subfamilies was various. There were no Hsp20 subfamilies but only cytosolic sHsps in the green algae and then CI, CII, and P subfamilies appeared in mosses (Waters and Vierling, 1999a,b). The process generated gene families attribute to gene duplication, gene loss, and recombination (including gene conversion) (Nei and Rooney, 2005; Flagel and Wendel, 2009). The sHsps might had undergone duplication events early in the history of land plant (Waters, 2013).

Gene duplication was reported as one of the primary forces that drive the evolution processes of genetic systems and genomes (Moore and Purugganan, 2003). In this paper, chromosome mapping showed that 41 SlHsp20 genes were located unevenly on 12 tomato chromosomes (**Figure 3**). The localization of most of the SlHsp20 genes was on the terminal regions of the chromosomes, which might contribute to the occurrence of duplication events in tomato Hsp20 gene family.

In our study, a total of 12 SlHsp20 genes were demonstrated to be involved in gene duplication, including tandem duplication and segmental duplication (**Figure 3**). Two tandem duplication events (SlHsp49.3/SlHsp39.4 and SlHsp17.7A/SlHsp17.6A/ SlHsp17.6B/SlHsp17.6C) and three segmental duplication events (SlHsp25.7A/SlHsp27.5, SlHsp17.7A/SlHsp24.5/SlHsp27.1, and SlHsp37.0/SlHsp27.2) were observed in CI, CIX, and CXII subfamilies. Intriguingly, SlHsp17.7A on chromosome 6 was shown to participate in both tandem and segmental events. It shared a duplicated region with SlHsp27.1 at the similar position on chromosome 10 and SlHsp24.5 on the upper arm of chromosome 9, while SlHsp27.1 and SlHsp24.5 had no tandem duplicated genes in their surrounding region. Comparative analysis revealed that sequence similarity between SlHsp17.7A and SlHsp27.1 was lower than that between SlHsp17.7A and SlHsp17.6A, suggesting that segmental duplication event predated the tandem duplication event in the SlHsp17.7A cluster, and the tandemly organized genes close to SlHsp27.1 might be lost after the segmental duplication event (**Supplementary Tables S5**, **S6**). Similar scenario was observed for SlHsp17.7A and SlHsp24.5. Together, these results indicated that both tandem duplication and segmental duplication made significant contributions to the expansion of the SlHsp20 gene family in tomato.

### Expression Patterns of *SlHsp20* Genes in Different Tissues

Based on RNA-Seq atlas, a spatio-temporal regulation of SlHsp20 gene family was observed in various tissues and development stages. Under normal growth conditions, a high or preferential expression of 11 SlHsp20 genes was found, which showed tissue- and development-specific expression in leaf, root, hypocotyl, and breaker fruit (**Figure 4**). All these tissue- and development-preferential expressed genes may play a critical role in growth and development of tomato and their functions still deserve further investigation. In addition, the expression behavior of some SlHsp20 genes differed in various tissues and development stages, indicating that the SlHsp20 proteins may play diverse functional roles. We also found that four genes (SlHsp17.7A, SlHsp17.6B, SlHsp17.6C, and SlHsp24.5) were highly expressed in all the investigated tissues, implying that they might be implicated in specific housekeeping activity of tomato cell under normal growth conditions. In vegetative and reproductive organs, two SlHsp20 genes (SlHsp25.7B and SlHsp17.6A) displayed differential expression levels between cultivated tomato and the wild relative S. pimpinellifolium, which indicated that interspecies divergence of gene expression was occurred and it might lead to functional specialization.

#### Expression Patterns of *SlHsp20* Genes under Abiotic and Biotic Stresses

Under various stress conditions, it's evident that the SlHsp20 genes were induced to a larger extent when tomato suffered from abiotic stresses, including heat, salt, and drought treatments (**Figure 5**). Furthermore, we found that the heat stress inducibility of SlHsp20 genes in susceptible plants was stronger than that in tolerant plants, which also had been demonstrated in CaHsp20 genes (Guo et al., 2015). This finding indicated that a more efficient mechanism might have been built in the tolerant plants so that fewer Hsp20 genes were sufficient to reduce the damage from heat shock.

Previous research had reported that the duplicated genes were easier for increasing the diversity of gene expression than single-copy genes (Gu et al., 2004). Here, we found that expression patterns of two groups of the tandemly duplicated genes were highly similar, which reflected that these SlHsp20 genes might share similar induction mechanisms and network (Ouyang et al., 2009). Actually, compared with tandem duplications, SlHsp20 genes in segmental duplicated group showed a more differential expression behavior. For example, the different expression between SlHsp37.0-CXII and SlHsp27.2-CXII suggested that segmentally duplicated genes may also exhibit divergent expression patterns (Waters, 2013). Earlier study revealed that distantly related duplicate genes may exist functional redundancy and have more chances to acclimatize than the single-copy genes (Gu et al., 2003). Thus, the duplicated SlHsp20 genes might go through crucial diversification after duplication events, which eventually caused neo-functionalization (Ouyang et al., 2009). It was also reported as a means for the retention of those duplicated genes in a genome (Force et al., 1999).

### CONCLUSIONS

In the current study, 42 putative SlHsp20 genes were identified in tomato. Subsequently, characterization of SlHsp20 genes was performed through integration of comprehensive sequence, genome organization and expression profile analysis among different tissues and under different stresses (heat, drought, salt, TSWV, and B. cinerea) by using RNA-seq and microarray atlas. This study provided a comprehensive understanding of the SlHsp20 gene family in tomato and made a basis for working out the functional roles of the Hsp20 genes in the Solanaceae family in the future.

### AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: HW and YY. Performed the experiments: JY, YC, MR, QY, RW, ZL, GZ, and ZY. Analyzed the data: JY, YC, and KF. Wrote the paper: JY and YC. All authors have read and approved the manuscript.

#### ACKNOWLEDGMENTS

Research is supported by the National Natural Science Foundation of China (31301774, 31272156, and 31071800), Zhejiang Provincial Natural Science Foundation of China (Q15C150010) and Young Talent Cultivation Project of Zhejiang Academy of Agricultural Sciences (2015R23R08E07, 2015R23R08E09), State Key Laboratory Breeding Base for Zhejiang Sustainable Pest and Disease Control (No. 2010DS700124-KF1518), Public Agricultural Technology Research in Zhejiang (2016C32101, 2015C32049), and Technological System of Ordinary Vegetable Industry (CARS-25-G-16).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01215

Supplementary Figure S1 | Intron/exon configurations of *SlHsp20* genes. The numbers (0, 1, 2) represented introns phases, which indicated three different splicing patterns.

Supplementary Table S1 | Genomic sequences, coding sequences and amino acid sequences of *SlHsp20* genes.

Supplementary Table S2 | Organization of putative motifs in SlHsp20 proteins. Ten putative motifs are represented by numbered color boxes. Names of all the proteins were indicated on the left side.

Supplementary Table S3 | Predicted subcellular location of *SlHsp20* genes.

Supplementary Table S4 | Homology matrixes of predicted tandem *SlHsp20* gene sequences.

Supplementary Table S5 | Tandemly duplicated *SlHsp20* genes.

#### REFERENCES


Supplementary Table S6 | *SlHsp20* genes localized on duplicated segments of the tomato genome.

Supplementary Table S7 | Information of probe sets used for microarray expression analysis.

high and low levels of mitochondrial sHSP23. 6 protein. Plant Growth Regul. 70, 175–185. doi: 10.1007/s10725-013-9790-y


**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 CY and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2016 Yu, Cheng, Feng, Ruan, Ye, Wang, Li, Zhou, Yao, Yang and Wan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

## The DnaJ Gene Family in Pepper (Capsicum annuum L.): Comprehensive Identification, Characterization and Expression Profiles

FangFei Fan<sup>1</sup>† , Xian Yang<sup>1</sup>† , Yuan Cheng<sup>2</sup> , Yunyan Kang<sup>1</sup> \* and Xirong Chai<sup>1</sup>

<sup>1</sup> College of Horticulture, South China Agricultural University, Guangzhou, China, <sup>2</sup> State key Laboratory Breeding Base for Zhejiang Sustainable Pest and Disease Control, Institute of Vegetables, Zhejiang Academy of Agricultural Sciences, Hangzhou, China

Edited by: Jie Zhou, Zhejiang University, China

#### Reviewed by:

Sergey Morozov, Moscow State University, Russia Hao Li, Northwest A&F University, China

\*Correspondence: Yunyan Kang kangyunyan@scau.edu.cn †These authors have contributed equally to this work.

#### Specialty section:

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

Received: 21 January 2017 Accepted: 13 April 2017 Published: 01 May 2017

#### Citation:

Fan F, Yang X, Cheng Y, Kang Y and Chai X (2017) The DnaJ Gene Family in Pepper (Capsicum annuum L.): Comprehensive Identification, Characterization and Expression Profiles. Front. Plant Sci. 8:689. doi: 10.3389/fpls.2017.00689 The DnaJ proteins which function as molecular chaperone played critical roles in plant growth and development and response to heat stress (HS) and also called heat shock protein 40 based on molecular weight. However, little was reported on this gene family in pepper. Recently, the release of the whole pepper genome provided an opportunity for identifying putative DnaJ homologous. In this study, a total of 76 putative pepper DnaJ genes (CaDnaJ01 to CaDnaJ76) were identified using bioinformatics methods and classified into five groups by the presence of the complete three domains (J-domain, zinc finger domain, and C-terminal domain). Chromosome mapping suggested that segmental duplication and tandem duplication were occurred in evolution. The multiple stress-related cis-elements were found in the promoter region of these CaDnaJ genes, which indicated that the CaDnaJs might be involved in the process of responding to complex stress conditions. In addition, expression profiles based on RNA-seq showed that the 47 CaDnaJs were expressed in at least one tissue tested. The result implied that they could be involved in the process of pepper growth and development. qRT-PCR analysis found that 80.60% (54/67) CaDnaJs were induced by HS, indicated that they could participated in pepper response to high temperature treatments. In conclusion, all these results would provide a comprehensive basis for further analyzing the function of CaDnaJ members and be also significant for elucidating the evolutionary relationship in pepper.

Keywords: DnaJ, heat shock protein 40, chromosomal localization, stress-related cis-elements, expression patterns

#### INTRODUCTION

With the increase of global warming, high temperature has become one of the most vital abiotic stresses on crop plants (Glazebrook, 1999). Pepper (Capsicum annuum L.) which originated in the tropical regions of Latin America had been widely cultivated around the world as an important vegetable crop nowadays and is sensitive to high temperature during plant growth and development, especially in reproductive stage (Guo et al., 2014). The optimum temperature

of growing pepper is 20–30◦C. Over 32◦C can bring about serious effects on pollination and fertilization, and results in blossom and fruit dropping which can cause a significant reduction of pepper fruit yield and quality.

In the long-term evolution, plants have evolved a complicated response mechanism to respond to heat stress (HS). Previous researches had reported that heat shock response (HSR) was induced in many plant species under HS condition (Vierling, 1991). Among them, a great deal of ubiquitous and evolutionaryconserved proteins was identified as heat shock proteins (Hsps), one of the main products of the HSR (Vierling, 1991). The Hsp was first discovered in Drosophila melanogaster in response to HS (Ritossa, 1962). In the following years, more Hsps were identified in other plants (Agarwal et al., 2002; Sarkar et al., 2009; Lopes-Caitar et al., 2013; Mulaudzi-Masuku et al., 2015). According to approximate molecular weight and sequence homology, the Hsps were classified into five families, including the Hsp100, Hsp90, Hsp70, Hsp60 and small Hsps (Wang et al., 2004; Kotak et al., 2007; Gupta et al., 2010).

The Hsp40, one of the important plant Hsps, was first identified in Escherichia coli, generally existed in organisms as 41 kDa Hsps (Georgopoulos et al., 1980; Bukau and Horwich, 1998; Craig et al., 2006). The Hsp40s, also known as DnaJ proteins or J-proteins, generally consisted of the J-domain, a proximal G/F-domain, a distal zinc finger (CxxCxGxG) domain, and followed by less conserved C-terminal sequences (Caplan et al., 1993; Silver and Way, 1993). The characteristic feature of the J-proteins was the presence of evolutionarily conserved J-domain which located nearby the N-terminus and composed of approximately 70 amino acids residues (Cyr et al., 1992). The invariant tripeptide (HPD) was the hallmark of J-domain. It stimulated the ATPase activity of Hsp70 and was crucial for keeping J-protein's function (Kampinga and Craig, 2010). Previously, Cheetham and Caplan (1998) attempted to separate these proteins into three groups. Group I J-proteins were characterized by the J-domain, G/F-domain, and zinc finger domain. Group II would have the J-domain plus either a G/F-domain or zinc finger domain. Group III J-proteins only comprised the J-domain (Ohtsuka and Hata, 2000).

In recent years, it has been found that plant DnaJ proteins played important roles in response to both biotic and abiotic stresses, such as pest, pathogenic bacterium, drought, salt, and heat. In 2007, a J-domain virulence effecter of Pseudomonas syringae remolded host chloroplasts when responded to pathogen (Jelenska et al., 2007). The researchers reported that overexpression of tomato (Solanum lycopersicum) chloroplasttargeted J-protein, LeCDJ1, facilitated heat tolerance in transgenic tomatoes (Kong et al., 2014a) and further found that it also played important role in maintaining photosystem II under chilling stress (Kong et al., 2014b). Subsequently, the study has also demonstrated that this gene could enhance tolerance to drought stress and resistance to P. solanacearum in transgenic tobacco (Wang et al., 2014). In addition, Xia et al. (2014) reported that a putative J-proteins ortholog from Nicotiana tabacum could be involved in drought stress response and its over-expression enhanced drought tolerance possibly through regulating expression of stress-responsive genes.

Up to now, many J-proteins in organisms were identified, such as Arabidopsis thaliana (89) (Miernyk, 2001), yeast (22) (Walsh et al., 2004) and human (41) (Craig et al., 2006). Despite ongoing efforts to characterize the members of J-protein from other organisms (Sarkar et al., 2009; Kong et al., 2014a,b; Xia et al., 2014), none of its from pepper has been identified at the genomic level. Fortunately, the pepper whole genomic sequences were completely available (Kim et al., 2014; Qin et al., 2014), which provided an opportunity for identifying candidate J-protein genes at the genomic level. In the present work, the J-protein gene family members were identified in pepper through bioinformatics method and analyzed by integration of gene structure, conserved motifs, chromosomal localization, ciselement and expression patterns.

#### MATERIALS AND METHODS

#### Genome-Wide Identification of CaDnaJ Genes in Pepper

The genomic sequences of pepper downloaded from the Pepper Genome Database (PGD<sup>1</sup> ) (Kim et al., 2014) were used to build the local database on the software BioEdit 7.0. The Hidden Markov Model (HMM) profile of J-domain (PF00226) downloaded from the Pfam protein family database<sup>2</sup> was used as query sequence to search against putative pepper J-protein genes with e-value <10−<sup>5</sup> . Subsequently, each of all putative pepper J-protein genes was used to identify the presence of J-domain on Pfam<sup>3</sup> . The protein sequences of identified pepper J-protein gene family members were analyzed with EXPASY PROTOPARAM<sup>4</sup> to obtain molecular weight and theoretical isoelectric point (pI).

#### Multiple Alignment and Chromosomal Location

In this paper, the J-proteins in pepper were classified based on structural features. In each class, the full amino acid sequences of pepper J-proteins were aligned using the software Clustal X 2.01 (Larkin et al., 2007). Each of the J-protein genes was mapped on chromosomes using MapDraw2.1 (Liu and Meng, 2003) based on information in PGD. Two duplication events, tandem duplication and segmental duplication, were also further elaborated. For tandem duplication, three criteria were adopted. Firstly, two or more pepper DnaJ genes were arrayed within a range of 100 kb distance. Secondly, the multiple alignments of these DnaJ genes had a high coverage rate of the longer gene (more than 70%). Thirdly, the identity of the aligned region in these DnaJ genes was also more than 70% (Li et al., 2009; Huang et al., 2012; Wei et al., 2016). The segmental duplication was investigated according to Plant Genome Duplication Database (PGDD<sup>5</sup> ).

<sup>1</sup>http://peppergenome.snu.ac.kr/

<sup>2</sup>http://pfam.xfam.org/family/PF00226.29

<sup>3</sup>http://pfam.xfam.org/

<sup>4</sup>http://www.expasy.org/tools/protparam.html

<sup>5</sup>http://chibba.agtec.uga.edu/duplication/

## Promoters Analysis of Pepper CaDnaJ Genes

The upstream regions (1.5 kb) of the DnaJ gene sequences were downloaded from PGD, and were used to search for regulating factor such as gibberellins (GA), abscisic acid (ABA), salicylic acid (SA), ethylene, drought, salt and heat.

#### Tissue-Specific Expression of CaDnaJ Genes Based on RNA-Seq

In this paper, RNA-seq data reported by previous researchers (Kim et al., 2014) were used to investigate expression patterns of putative CaDnaJ genes in pepper. Different tissues were selected: root, stem, leaf and pericarp at 6 days post-anthesis (DPA), 16 DPA, 25 DPA, respectively, and mature green (MG), breaker (B), 5 days post-breaker (B5), B10. RPKM (Reads Per Kilo bases per Million mapped Reads) values of CaDnaJ genes were log2-transformed (Wei et al., 2012). Heat maps of CaDnaJ genes in different tissues were performed using software MultiExperiment Viewer (MeV) (Howe et al., 2010).

#### Plant Materials and Heat Stress Treatment

A hot pepper hybrid (zhejiao 3#), which developed by Zhejiang Academy of Agricultural Sciences, was selected in the experiments. Seeds were sterilized for 5 min using 10% hypochlorous acid solution and washed three times using distilled water. These seeds were further placed in water-saturated filter paper to germinate, then cultivated in Hoagland solution in a growth chamber which was maintained at a 16 h light at 26◦C and 8 h dark at 19◦C. At the stage of 6–8 true leaves, plants were treated with 42◦C for 4 h and plants grown at 25◦C were used as the control group. Young leaves were collected and immediately frozen with liquid nitrogen for total RNA extraction. Each treatment was conducted with three biological replicates, and samples from five plants were collected for each replicate.

### RNA Extraction and qRT-PCR Analysis

Total RNA was extracted using Total RNA kit (Tiangen Biotech, Beijing, China) and reverse-transcribed using FastQuant RT Kit (Tiangen Biotech, Beijing, China), the operational procedure followed the manufacturer's procedure.

For quantitative RT-PCR analysis, we amplified PCR products in triplicate using 2 × Taq Master Mix (Vazyme, Nanjing, China) in 20 µL qRT-PCR reactions. PCR was performed using the ABI step-one plus 96-well real-time PCR Detection System (Bio-Rad) and cycling conditions consisted of denaturation at 94◦C for 5 min, followed by 33 cycles of denaturation at 94◦C for 30 s, annealing at 55◦C for 30 s and extension at 72◦C for 30 s. The UBI gene was used as an internal control (Wan et al., 2011). Gene-specific primers were designed and used for amplification as described in **Supplementary Table S1**. Analysis of relative gene expression data was performed using the 2−11ct method (Livak and Schmittgen, 2001).

## RESULTS

### Genome-Wide Identification of CaDnaJ Genes in Pepper

A total of 85 putative sequences of pepper DnaJ genes were gotten from PGD by HMM search. Among them, nine sequences without a complete J-domain (CA03g32700, CA06g22620, CA00g87730, CA12g09540, CA02g21540, CA05g10320, CA11g03100, CA06g00080, and CA05g07590) were removed. The remaining 76 genes were assigned as pepper DnaJ genes. As a matter of convenience, the 76 DnaJ proteins were named as CaDnaJ01 to CaDnaJ76 according to their location on chromosome (**Table 1**).

The length of CaDnaJ proteins ranged from 130 (CaDnaJ23) to 1272 (CaDnaJ45) amino acids, and the predicted molecular weights were between 15.597 kDa (CaDnaJ59) and 157.85 kDa (CaDnaJ25). The CaDnaJs shared a conserved J-domain comprised about 70 amino acids, in which CaDnaJ57 owned the shortest J-domain with 39 amino acids, while J-domain of CaDnaJ25 was the longest (84 amino acids). The predicted pI-values of CaDnaJ proteins ranged from 4.56 (CaDnaJ74) to 9.87 (CaDnaJ24), indicating acidic and alkaline proteins. Besides, it was also found that 25 (32.89%) of the total 76 CaDnaJ genes had no introns, 11 genes (14.47%) had a single intron, while only CaDnaJ03 had 11 introns (**Table 1**).

#### Classification and Sequence Alignment of CaDnaJ Genes

The **Ca**DnaJ protein usually contains conserved J-domain, zinc finger domain, and uncharacterized C-terminal domain (Cheetham and Caplan, 1998). According to the presence of the complete three domains, the CaDnaJ genes were classified into five groups (A, B, C, D, and E), including 9, 8, 53, 1 and 5 members, respectively. Group A CaDnaJ proteins are characterized by the J-domain, zinc finger domain and a less conserved C-terminal. The difference between Group A and B was lack of the zinc finger domain. Group C CaDnaJ proteins only comprised the J-domain, otherwise, Group D would both have the J-domain and zinc finger domain. Group E contains a CaDnaJ protein lacked of HPD motif, which have been described as J-like proteins (Walsh et al., 2004).

Based on the classification above, sequence alignment of CaDnaJ genes was performed separately (**Supplementary Table S2**). It was found that nine members in group A possessed a conserved HPD motif in J-domain and two zinc finger domains (CxxCxGxG). Eight members in group B were lack of the zinc finger domain but owned conserved HPD motif in J-domain. The largest group (group C) which included 53 members only consisted of complete J-domain. Group D which contained merely one member (CaDnaJ36) had both the conserved HPD motif and zinc finger domain. The last group (group E), which comprised five members, held the least conservation among five groups. All the members in group E possessed J-domains lacked of HPD motif.

#### TABLE 1 | The list of CaDnaJ members identified in pepper.


(Continued)

#### TABLE 1 | Continued

fpls-08-00689 April 27, 2017 Time: 14:49 # 5


<sup>a</sup>Group A are characterized by the J-domain, zinc finger domain and a less conserved C-terminal; Group B are characterized by the J-domain and a less conserved C-terminal; Group C only comprised the J-domain; Group D would both have the J-domain and zinc finger domain; Group E was described as J-like proteins contained a J-protein lacked of HPD motif. MW, Molecular weight; pI, isoelectric points.

#### Chromosomal Location and Gene Duplication

All these CaDnaJ genes (9 members in group A, 8 members in group B, 53 members in group C, 1 members in group D and 5 members in group E) in pepper were uneven distributed on 12 chromosomes (**Figure 1**). Among them, eleven and nine genes were located on chromosome 8 and 5, respectively. Seven genes on each of chromosome 1, 3, 7, six genes on chromosome 4, five genes on each of chromosome 11 and 12, four genes on each of chromosome 2, 9, three genes on each of chromosome 6 and 10.

We further analyzed the gene duplication of CaDnaJ genes in pepper. As shown in **Figure 1**, one tandem duplication event (CaDnaJ41/CaDnaJ42) was identified on chromosome 7. The chromosome location of this pair of CaDnaJ genes was close in distance and was inserted by less than one gene. In addition, two segmental duplication events were detected. CaDnaJ04 on chromosome 1 presented synteny to CaDnaJ44 localized on a duplicated segment of chromosome 8. Similar scenario was observed for CaDnaJ19 on chromosome 4 and CaDnaJ67 on chromosome 12. These results suggested that these duplication events made contributions to expansion of pepper CaDnaJ gene family.

#### Analysis of Stress-Related cis-Elements in Pepper CaDnaJ Promoters

The upstream regions (1.5 kb) of the CaDnaJ sequences were searched for regulating factor on different stress conditions. The stress-related cis-elements were not been found in the promoter region of two CaDnaJ genes (CaDnaJ26 and 27), and the rest of 74 CaDnaJ genes possessed multiple cis-elements. Among them, the CaDnaJ 28 and 50 possessed the maximum types of stress-related cis-elements (10). On the contrary, only two types of stress-related cis-elements were held on CaDnaJ10. Besides, CaDnaJ 34 has the most stress-related cis-elements (23), including twelve methyl jasmonic acid (MeJA) related, six drought and indoleacetic acid (IAA) related, four ABA related and one GA related (**Figure 2** and **Supplementary Table S3**).

To further explore the possible regulation mechanism of CaDnaJ genes to HS, the heat stress responsiveness elements (HSEs) were also searched in the promoter region of all these CaDnaJ genes. The result showed that 66.21% (49 out of 74) CaDnaJ genes have HSEs in the promoter regions. The maximum numbers of HSEs (6) were identified in CaDnaJ25 and CaDnaJ28. Only one HSEs was found in fifteen CaDnaJ genes (CaDnaJ01, 03, 04, 08, 22, 33, 37, 39, 41, 42, 47, 52, 55, 68, and 72). In addition, other stress-related cis-elements were also detected. There were 107 TC-rich repeats in 55 genes, 99 MBS in 49 genes, 33 LTR in 26 genes, 79 TCA-element in 47 genes, 36 TGA-element in 27 genes, 87 GARE-motif in 49 genes, 138 CGTCA-motif in 41 genes, 37 ERE in 27 genes and 55 ABRE in 28 genes.

#### Expression Patterns of CaDnaJ Genes in Different Tissues

Based on RNA-seq data of different pepper tissue (root, stem, leaf, and pericarp) published previously (Kim et al., 2014), expression

profiles of CaDnaJ genes were revealed (**Figure 3**). A total of seven different stages of pericarp [6 DPA, 16 DPA, 25 DPA, MG, breaker (B), 5 days post-breaker (B5), B10] were selected for expression analysis in the present study. As shown in **Figure 3**, 29 out of 76 CaDnaJ genes were barely expressed in the tested tissues, including CaDnaJ01, 03, 05, 09, 12,14, 15, 16, 19, 22, 26, 29, 30, 38, 39, 48, 49, 51, 57, 58, 59, 63, 64, 65, 67, 69, 71, 72 and 75. The left 47 CaDnaJ genes could be detected at least in one tissue. Ten genes (CaDnaJ06, 08, 21, 27, 28, 47, 50, 55, 56 and 76) were expressed in all tested tissues. Of them, four genes (CaDnaJ06, 21, 28 and 76) were expressed at relatively high levels. Besides, tissue-specific expression was also found in some CaDnaJ genes. The CaDnaJ66 was specifically expressed in leaf, CaDnaJ07 and 13 were specifically expressed in root. During the stage of pericarp development, the expression of CaDnaJ02 was significantly up-regulated, while CaDnaJ74 and CaDnaJ45was obviously down-regulated.

#### Expression Patterns of CaDnaJ Genes in Response to Heat Treatments

To gain more insight into the role of CaDnaJ genes under HS condition, expression profiles of CaDnaJ genes in pepper response to high temperature based on qRT-PCR technique were performed. In this study, a total of 67 CaDnaJ genes were used to design successfully specific primers for expression analysis. As shown in **Figure 4**, expression of these tested CaDnaJ genes was significantly changed under high temperature stress treatments. Expression levels of five CaDnaJ genes (CaDnaJ4, 50, 59, 63, and 72) were down-regulated, and 49 CaDnaJ genes were upregulated. For the remaining 13 CaDnaJ genes (CaDnaJ1, 17, 27, 31, 43, 48, 49, 53, 54, 58, 60, 73 and 75), no difference was observed. Notably, among the CaDnaJ genes up-regulated, the expression levels of 71.4% (35/49) were increased to three folds. In total, expression of most of CaDnaJ genes was significantly altered under HS, indicating that the CaDnaJ genes were involved in plants response to high temperature stress.

#### DISCUSSION

As an important vegetable crop all over the world, pepper is deeply loved by a large population since its major ingredient in cuisines, essential vitamins and other healthy nutrients (Kim et al., 2014). High temperature has become one of the important environmental stresses and affected seriously the growth and development in pepper. The DnaJs, one of the significant Hsps, was produced in the process of plant responding to HS. Up to now, functional identification of DnaJ has been reported in

mature green; B, breaker; B5, 5 days post-breaker; B10, 10 days post-breaker; 6 DPA, 6 days post-anthesis; 16 DPA, 16 days post-anthesis; 25 DPA, 25 days post-anthesis. Fragments per kilobase of exon model per million mapped (FPKM) values were log2-transformed and heat maps with hierarchical clustering were exhibited using the software Mev 4.9.0.

many plant species (Miernyk, 2001; Qiu et al., 2006; Bekh-Ochir et al., 2013; Fristedt et al., 2014; Kong et al., 2014a,b). At the whole genome level, it has been found that Arabidopsis has 89 members and encoded multiple gene family (Miernyk, 2001). Recently, the whole pepper genome has released and provided an opportunity for identifying putative DnaJ homologous. In the current paper, a systematic analysis of CaDnaJ gene family was performed using bioinformatics methods, which focused on gene structure, chromosomal localization, stress-related cis-elements, and expression profiles in different tissues. The results would be significant for further analyzing the function of CaDnaJ members and illuminating the evolutionary relationship in pepper.

Since the first DnaJ proteins were isolated from E. Coli as 41 kDa Hsps (Georgopoulos et al., 1980), many DnaJ proteins have been subsequently reported in other life species (Bukau and Horwich, 1998; Miernyk, 2001; Walsh et al., 2004; Craig et al.,

2006). Generally, DnaJ proteins contained one to four domains (J-domain, G/F-domain, zinc finger domains and less conserved C-terminal) (Silver and Way, 1993). Initially, the DnaJ proteins were classified into three groups (A, B, C) based on domain composition (Cheetham and Caplan, 1998; Miernyk, 2001). In this study, more complex structure of DnaJ genes was observed in pepper. For example, the HPD tripeptide which was crucial for J-domain function (Kampinga and Craig, 2010) was not found in CaDnaJ23, 28, 37, 53 and 73. Therefore, given the high diversity of CaDnaJ genes, a more systematic classification was proposed in our study. A total of 76 CaDnaJ genes were classified into five groups (A, B, C, D and E). Group A was characterized by the J-domain, zinc finger domains and a less conserved C-terminal. Group B were lack of the zinc finger domains but contained the J-domain and a less conserved C-terminal. Group C only comprised the J-domain. Group D would both have the J-domain and zinc finger domains but lack of C-terminal, and group E which have been described as J-like proteins with the J-domain lack of HPD motif (**Table 1** and **Supplementary Table S2**).

It was reported that gene duplication was become one of the primary evolution forces during the processes of genetic systems and genomes (Moore and Purugganan, 2003). In our study, chromosome location showed that all these CaDnaJ genes were mapped unevenly on 12 pepper chromosomes (**Figure 1**). Six of them were involved in gene duplication, including tandem duplication and segmental duplication (**Figure 1**). One tandem duplication event (CaDnaJ41/CaDnaJ42) and two segmental duplication events (CaDnaJ04/CaDnaJ44, and CaDnaJ19/CaDnaJ67) were observed. These results indicated that both tandem duplication and segmental duplication played role in expansion of the CaDnaJ gene family in pepper.

It has been demonstrated that cis-elements participated in responding multiple abiotic and biotic stresses. For instance, several cis-elements, such as ABRE, DRE, CRT, SARE and SURE, had been identified for responding to ABA, dehydration, cold, SA, and sulfur, respectively (Sakuma et al., 2002; Maruyama-Nakashita et al., 2005; Shi et al., 2010; Osakabe et al., 2014; Feng et al., 2016). In this paper, we identified cis-elements in the promoter regions of CaDnaJ genes using the PlantCARE server (Lescot et al., 2002). Two major groups of cis-elements were observed, including stress-responsive and hormone-responsive. The former contained HSE, LTR, and TC-rich cis-elements, which was responsive to heat, low-temperature, and defense, respectively. The latter was composed of TCA-element, TGAelement, GARE-motif, CGTCA-motif, ABRE and ERE, which was responsive to SA, IAA, GA, MeJA, ABA and ethylene, respectively (**Figure 2** and **Supplementary Table S3**). These results implied that CaDnaJ genes could be involved in the process of plant respond to multiple stresses. Especially, one of the most important cis-element, HSEs, which kept AAAAAATTTC as the core sequence, accounted for 13.64% of all the cis-elements. As we all known that the expression of Hsp was controlled by heat shock transcription factors that bind to HSEs in the promoter region of the Hsp genes (Hancock et al., 2009). A total of 49 (66.21%) CaDnaJ genes have HSEs in the promoter region. Thus, the results will contribute to further understand the vital function role of CaDnaJ genes under HS condition in the further.

To obtain more insights into the expression profiles of CaDnaJ genes in different tissues, RNA-seq data were acquired from leaf, root, stem and different stage of pericarp (Kim et al., 2014). Based on RNA-seq, we found that all these CaDnaJ genes exhibited three different expression patterns: (1) barely expression or too low expression level to detect; (2) constitutive expression; (3) tissue-specific expression patterns. The expression of these CaDnaJ genes differed in tissues tested, indicating that the CaDnaJ proteins may play different functional roles. Expression of some CaDnaJ genes showed tissue- and development-specific in root, stem, leaf and pericarp, suggested that they may participated in growth and development of pepper. In addition, we also found that 10 genes (CaDnaJ06, 08, 21, 27, 28, 47, 50, 55, 56 and 76) were highly expressed in all the tested tissues, implying that they might be involved in specific housekeeping action under normal growth conditions in pepper (**Figure 3**).

It has been known that plant growth and development were frequently affected by biotic and abiotic stresses under natural conditions (Xia et al., 2014). Previous researchers had reported that the DnaJ protein was involved in plant response to heat and drought stresses (Xia et al., 2014; Wang et al., 2015). To further comprehend the putative roles of CaDnaJ genes in pepper response to HS, expressions patterns of CaDnaJ genes under HS treatment conditions were analyzed by qRT-PCR. Expression analysis revealed that most CaDnaJ genes were changed in response to HS (**Figure 4**). Among them, almost half of these CaDnaJ genes were up-regulated for three folds. We also found that there are multiple types and numbers of cis-elements in these CaDnaJs promoter, which is reported to involved in abiotic stresses (Feng et al., 2016), including TC-rich, HSEs, MBS motif. All these results suggested that these CaDnaJ genes could involved in plant response to HS.

#### REFERENCES


## AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: FF and YK. Performed the experiments: XY, XC, and YC. Analyzed the data: FF and YK. Wrote the paper: FF. All authors have read and approved the manuscript.

#### ACKNOWLEDGMENTS

The research is supported by National Natural Science Foundation of China (No. 31301767), Foundation for Distinguished Young Talents in Higher Education of Guangdong, China (No. YQ2015028), and Modern agricultural industry technology system construction of Guangdong, China (No. 2016LM1109).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.00689/ full#supplementary-material

#### TABLE S1 | Specific primer for qRT-PCR of each CaDnaJ gene.

TABLE S2 | Multiple sequence alignment of the pepper CaDnaJs. The sequences of 76 CaDnaJs were aligned according to the classification. Names of all the 76 members are showed on the left side of the figure. Conserved amino acid residues are highlighted in black. The sequences which boxed in red rectangle represented the conserved HPD motif, and sequences in yellow rectangle were zinc finger domains.

#### TABLE S3 | Stress-related cis-elements in pepper CaDnaJ promoters.


domestication and specialization. Proc. Natl. Acad. Sci. U.S.A. 111, 5135–5140. doi: 10.1073/pnas.1400975111


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

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

fpls-08-00689 April 27, 2017 Time: 14:49 # 11

# Endoplasmic Reticulum Stress Response in Arabidopsis Roots

Yueh Cho1,2,3 and Kazue Kanehara1,2,4,5 \*

1 Institute of Plant and Microbial Biology, Academia Sinica, Taipei, Taiwan, <sup>2</sup> Molecular and Biological Agricultural Sciences Program, Taiwan International Graduate Program, Academia Sinica and National Chung-Hsing University, Taipei, Taiwan, <sup>3</sup> Graduate Institute of Biotechnology, National Chung-Hsing University, Taichung, Taiwan, <sup>4</sup> Biotechnology Center, National Chung-Hsing University, Taichung, Taiwan, <sup>5</sup> Muroran Institute of Technology, Muroran, Japan

Roots are the frontier of plant body to perceive underground environmental change. Endoplasmic reticulum (ER) stress response represents circumvention of cellular stress caused by various environmental changes; however, a limited number of studies are available on the ER stress responses in roots. Here, we report the tunicamycin (TM) -induced ER stress response in Arabidopsis roots by monitoring expression patterns of immunoglobulin-binding protein 3 (BiP3), a representative marker for the response. Roots promptly responded to the TM-induced ER stress through the induction of similar sets of ER stress-responsive genes. However, not all cells responded uniformly to the TM-induced ER stress in roots, as BiP3 was highly expressed in root tips, an outer layer in elongation zone, and an inner layer in mature zone of roots. We suggest that ER stress response in roots has tissue specificity.

Keywords: Arabidopsis thaliana, ER stress response, root, stress response, UPR

#### Edited by:

Hanjo A. Hellmann, Washington State University, USA

#### Reviewed by:

Alessandro Vitale, National Research Council, Italy Zhaojun Ding, Shandong University, China

\*Correspondence: Kazue Kanehara kanehara@gate.sinica.edu.tw

#### Specialty section:

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

Received: 11 August 2016 Accepted: 24 January 2017 Published: 01 March 2017

#### Citation:

Cho Y and Kanehara K (2017) Endoplasmic Reticulum Stress Response in Arabidopsis Roots. Front. Plant Sci. 8:144. doi: 10.3389/fpls.2017.00144 INTRODUCTION

Roots are the frontier of plant body to perceive underground environmental change. In response to environmental stimuli, a crucial set of molecular processes is induced that maintains cellular homeostasis and thus circumvents fatal defects caused by the stresses. Among various organelles involved in the cellular homeostasis, the ER plays a decisive role in protein folding and secretion.

The ER is the gateway for the eukaryotic protein secretory pathway. Secretory proteins are translocated into the ER and enter protein-folding cycles to fold and assemble themselves (Anelli and Sitia, 2008). Only properly folded proteins are allowed to leave the ER by a surveillance system, collectively termed the ER quality control. The ER quality control is wellconserved molecular mechanisms among eukaryotic cells including animals, yeasts and plants (Iwata and Koizumi, 2005, 2012; Howell, 2013). When aberrant proteins are accumulated in the ER, the ER quality control recognizes these aberrant proteins and responds to maintain the ER homeostasis using multiple strategies such as UPR and ER-associated degradation (ERAD) (Kanehara et al., 2007; Walter and Ron, 2011; Ruggiano et al., 2014). The ER membrane-localized ribonuclease inositol-requiring enzyme 1 (IRE1) is one of major signal transducers in the UPR. The IRE1 senses protein-folding status in the ER and transmits signals into the nuclei by catalyzing unconventional cytoplasmic splicing of bZIP60 mRNA in plants (XBP1 in mammals and HAC1 in yeasts) followed by activation of UPR target genes including a molecular chaperone BiP (Nagashima et al., 2011; Walter and Ron, 2011). BiP is one of the most abundant chaperones in the ER lumen and is thought to bind nascent

**Abbreviations:** ER stress, endoplasmic reticulum stress; TM, tunicamycin; BiP, immunoglobulin-binding protein; UPR, unfolded protein response.

peptides to prevent protein aggregation. BiPs belong to the heat shock protein 70 family that binds ATP and operates in conjunction with J-domain-containing proteins (J proteins) (Fewell et al., 2001). Arabidopsis genome encodes three BiP genes, BiP1 (At5g28540), BiP2 (At5g42020), and BiP3 (At1g09080) (Noh et al., 2003). BiP1 and BiP2 encode ubiquitously expressed proteins whose amino acid sequences are 99% identical to each other. BiP3 encodes a less conserved protein, whose expression is limited under ER stress conditions in young seedlings (Noh et al., 2003). BiPs are also master regulators of the ER stress response in Arabidopsis (Srivastava et al., 2013). Recent studies showed that BiP3 functions in pollen development and female gametogenesis (Maruyama et al., 2014, 2015).

Extensive studies across different model organisms have revealed details of molecular mechanisms underlying the ER stress response. In plant research, however, ER stress response is an emerging subject despite its high relevance to general plant stress response studies (Liu and Howell, 2010). In addition to the conserved molecular mechanisms, recent studies have suggested that plants may have plant-specific ER stress responses, involving heterotrimeric G proteins and phosphoinositide signaling (Wang et al., 2007; Chen and Brandizzi, 2012; Kanehara et al., 2015b). Our understanding about the molecular mechanisms of the ER stress response in plants has been based mostly on the studies using whole Arabidopsis seedlings. Because root biomass is marginal to the total seedling biomass, effect of root-specific changes may be diluted to be non-measureable in the whole seedling sample even if the root and shoot differentially responds to the ER stress. In fact, previous studies explored ER stress responses in roots of various plant species. For example, the tissue-specific transcriptional regulation of soybean BiPs, gsBiP6 and gsBiP9 in transgenic Nicotiana tabacum plants was reported using gsBiPs promoter-GUS chimeric reporter genes (Buzeli et al., 2002). In roots of Pisum sativum, the expression levels of BiP-D, bZIP28, and bZIP60 were elevated during tungsten treatment, which is known to affect plant growth (Adamakis et al., 2011). In addition, rice OsBiPs and its co-chaperones, OsERdjs, were transcriptionally upregulated under the ER stress conditions in roots of rice seedlings (Ohta et al., 2013). However, little is known about a detailed molecular mechanism of ER stress response in roots despite that root is an important organ to perceive environmental stresses. Although tissue specificity of the UPR in plants has been investigated in the gametophyte development (Maruyama et al., 2010, 2014; Deng et al., 2013), it remains elusive whether an individual cells of multicellular organisms responds uniformly or differentially to the ER stress caused by external environmental stresses. The Arabidopsis root is an excellent model to investigate tissue type- and cell type-specific response in vivo because it is a transparent organ and each tissue/cell type has been characterized well.

In an effort to explore the ER stress response in the plant root system and address tissue-specific response in an intact multicellular organism, current study investigated ER stress response in Arabidopsis roots. To monitor the ER stress responses, a well-described UPR gene BiP3 has been employed because of the extremely low expression under non-stress condition but acute induction upon ER stress (Noh et al., 2003; Srivastava et al., 2013; Cho et al., 2015). Based on the timecourse observation of stable transgenic plant expressing the ProBiP3:BiP3-GUS-HDEL or ProBiP3:mRFP, we found that BiP3 was differentially expressed in root under the ER stress condition: the root tip including columella, outer layers in elongation zone and inner layers in mature zone were highly responsive to the ER stress. Our results suggest that the ER stress response has tissue-type and cell-type specificity and not all cells may respond uniformly to the ER stress in the Arabidopsis roots.

#### MATERIALS AND METHODS

#### Plant Materials and Growth Condition

Arabidopsis plants (Columbia-0 ecotype) were grown under continuous light at 22◦C. Murashige and Skoog (MS) media was used at half-strength concentration for plant culture (Murashige and Skoog, 1962). Seeds of bip3-1 (SALK\_024133) were obtained from Nottingham Arabidopsis Stock Centre (NASC). Homozygous T-DNA mutant plants were isolated by PCR-based genotyping with the specific primers (KK200/KK201, LB1.3/KK201). Position of T-DNA insertion was determined by sequencing to be located within the protein coding sequence of BiP3 (**Figure 3A**). For TM treatment, seedlings were immersed in liquid MS media containing 5 µg/ml TM for indicated time. For detection of aggregated proteins, seedlings were immersed in liquid MS media containing 10 mM MG-132 for 16 h. DMSO was used as negative controls for both TM and MG-132 treatments.

#### Sequence Alignment of BiPs

The amino acid sequences of three BiP isoforms were adopted from TAIR database (protein accession numbers: BiP1: 1009129411, BiP2: 1009134007, and BiP3: 5019479994). A multiple alignment of the protein sequences for BiPs was assembled using ClustalW<sup>1</sup> and BoxShade<sup>2</sup> .

### Plasmid Vector Construction and Plant Transformation

A 4 kbp fragment of the genomic sequence for BiP3 was amplified by PCR with oligonucleotide primers KK131 and KK132, and cloned into the pENTR/D-TOPO plasmid vector (Invitrogen, Carlsbad, CA, USA) to obtain pCC38. To create the GUS reporter construct (ProBiP3:BiP3-GUS-HDEL), SmaI site was inserted at the position immediately before the ER retention sequence HDEL of BiP3 by PCR-based site directed mutagenesis with primer KK152 (Sawano and Miyawaki, 2000). Then, a GUS cassette was inserted into the SmaI site to produce pCC71, which was recombined to a pBGW destination vector by use of LR Clonase (Karimi et al., 2005). The resulting pCC67 was transformed into wild-type (WT) plants via Agrobacterium GV3101-mediated gene transformation. Twenty-four transformed plants were selected by spraying 0.1% Basta solution to the seedlings on soil. The T1 seeds were screened by Basta, and the resistant plants

<sup>1</sup>http://clustalw.ddbj.nig.ac.jp/

<sup>2</sup>http://www.ch.embnet.org/software/BOX\_form.html

harboring ProBiP3:BiP3-GUS-HDEL were selected by PCRbased genotyping with primers (KK98/KK200). ProBiP3:BiP3- GUS-HDEL line No. 17 was selected as a representative line for observation. For the fluorescent reporter construct (ProBiP3:mRFP), the 0.9 kbp promoter region of BiP3 was amplified with primers KK131 and KK172, and cloned into pENTR/D-TOPO plasmid vector (Invitrogen, Carlsbad, CA, USA) to obtain pCC76. This was recombined into a destination vector pGWB653 (Nakamura et al., 2010) by use of LR Clonase and the resulting plasmid pCC79 was transformed into WT plants via Agrobacterium GV3101-mediated gene transformation. Then, 16 plants were selected by spraying 0.1% Basta solution to the seedling on soil. The T2 seeds were screened by Basta, and the resistant plants harboring ProBiP3:mRFP were selected by PCR-based genotyping with primers (KK202/KK133). ProBiP3: mRFP line No. 11 was selected as a representative line and used for observation by confocal laser-scanning microscopy. The sequence of primers used were listed in Supplementary Table S1.

#### Quantitative RT-PCR (qRT-PCR)

Quantitative RT-PCR analysis was performed as previously described using total RNA was isolated from 7-day-old seedlings (Lin et al., 2015). The means and standard deviations of 11CT were calculated from three independent biological replicates for whole seedlings. Six independent biological replicates were used for roots. The primers used for qRT-PCR are listed in Supplementary Table S1.

#### Histochemical GUS Staining

Gus staining was performed as previously described by Kanehara et al. (2015a). Briefly, seedling samples were immersed in GUS staining solution (10 mM EDTA, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 0.1% [w/v] Triton

X-100, and 0.5 mg/ml X-Gluc [5-bromo-4-chloro-3-indolyl-β-D-glucuronide] in 100 mM phosphate buffer), and incubated at 37◦C. Then, the reaction was stopped by replacing the solution with 70% ethanol. For colored tissues, pigments were removed

by immersing the tissue in 6:1 (v:v) ethanol : acetic acid. The images were obtained using a stereomicroscope (Zeiss Stemi 2000) equipped with a Nikon D7000 camera and an upright microscope (Zeiss Axio Imager A2) equipped with a Canon EOS 500D camera.

#### Preparation of Anti-BiP3 Antibody and Immunoblotting

To avoid a cross-reaction with BiP1 and BiP2, which show high amino acid similarity to BiP3, a polypeptide consisting of the C-terminal 19 amino acid residues of BiP3, VYEKTEGENEDDDGDDHDE, was synthesized and used to raise an anti-BiP3 polyclonal antibody in rabbits (LTK BioLaboratories, Taoyuan, Taiwan). For immunoblotting, total cell lysate from seedlings or roots was extracted in a lysis buffer [50 mM Tris-HCl (pH6.8), 2% SDS, 10 mM β-mercaptoethanol, 1% v/v protease inhibitor cocktail (Sigma)]. Protein samples were separated by 10% acrylamide SDS-PAGE and transferred to a polyvinylidene difluoride membrane for immunoblotting with

treated with 5 µg/ml TM for the indicated time. The merged images of mRFP fluorescence and DIC are shown. The fluorescent images were provided at the left side of each merged image. Scale bars, 10 µm.

rabbit polyclonal anti-BiP3 antibodies; 1:2,000, followed by goat anti-rabbit IgG peroxidase conjugates (Santa Cruz); 1:10,000. BiP3 proteins were visualized by use of chemiluminescence detection reagent (SuperSignal West Pico, Pierce) and Image Quant LAS4000 (GE Healthcare). SDS-PAGE gel was also stained with 0.1% Coomassie Brilliant Blue for 1 h at room temperature.

#### Confocal Laser-Scanning Microscopy

Fluorescence of mRFP in seedlings of ProBiP3:mRFP was observed under a microscope (LSM 510 Meta; Carl Zeiss) equipped with Objectives C-Apochromat 40×/1.2-numerical aperture (NA) and C-Apochromat 63×/1.2-NA. Images were captured using an LSM 510 v3.2 confocal microscope (Carl Zeiss) with filter (543-nm laser, band-pass 560–615 nm). Cell boundaries were visualized by differential interference contrast (DIC) images.

### Detection of Aggregated Protein

Aggregated proteins were stained by Proteostat <sup>R</sup> Aggresome Detection Kit (Enzo: ENZ-51035) according to manufacturer's instruction with a slight modification. Briefly, seedlings after the chemical treatments were fixed in 4% paraformaldehyde in the assay buffer (Proteostat <sup>R</sup> Aggresome Detection Kit, Enzo) for 30 min at room temperature, which was then washed by phosphate-buffered saline (PBS) three times. Seedlings were transferred into a permeabilizing solution (0.5% Triton X-100, 3 mM EDTA, pH 8) for 30 min. After washing with PBS buffer, seedlings were incubated with Proteostat <sup>R</sup> dye (Proteostat <sup>R</sup> Aggresome Detection Kit, Enzo) at 1:5000 dilution for 1 h in the dark.

#### Immunolocalization Analysis of BiP3 and Aggregated Proteins

After chemical treatments, seedlings were fixed in 4% paraformaldehyde (Merck) in PBS for 90 min with vacuum dry. To break down cell walls, the seedlings were incubated with 1.5% Driselase (Sigma) in PBS at 37◦C for 50 min. For plasma membrane penetration, the samples were incubated with a PBS solution containing 3% IGEPAL CA-630 (Sigma) and 10% DMSO for 30 min at room temperature. To decrease non-specific binding, samples were blocked in PBS containing 3% BSA for 3 h at room temperature. After overnight incubation with anti-BiP3 antibodies (1:100) at 4◦C, samples were washed with PBS six times, and incubated with secondary antibody, goat anti-rabbit Alexa Fluor 488 (1:1,000, Life Technologies) for 3 h at room temperature. To detect aggregated proteins, samples were incubated with Proteostat <sup>R</sup> dye (Proteostat <sup>R</sup> Aggresome Detection Kit, Enzo; 1:5000) for 1 h at room temperature. After three times washing by PBS, samples were mounted in a drop of 9:1 (v/v) glycerol: PBS, and observed under a microscope (LSM 510 Meta; Carl Zeiss) equipped with objectives C-Apochromat 40×/1.2-NA and C-Apochromat 63×/1.2-NA. Images were captured using an LSM 510 v3.2 confocal microscope (Carl Zeiss) with filter (488-nm laser, band-pass 505–530 nm) for BiP3, and with filter (543-nm laser, band-pass 560–615 nm) for aggregated proteins. Cell boundaries were visualized by DIC images.

## RESULTS

#### Observation of the ER Stress Response in Roots via the BiP3-GUS Reporter System

To observe the ER stress response in roots of intact plants, we employed Arabidopsis BiP3 gene as a reporter. BiP3 is a widely

used marker gene for the ER stress response, whose expression is extremely low at either RNA or protein levels under nonstress conditions but is highly up-regulated upon the ER stress in Arabidopsis young seedlings (Noh et al., 2003; Cho et al., 2015). We established a transgenic Arabidopsis plant that stably expresses BiP3-GUS fusion protein containing the ER retention signal (HDEL) for ER localization, which is driven by the own promoter (ProBiP3:BiP3-GUS-HDEL in WT background). We treated 7-day-old seedlings of the ProBiP3:BiP3-GUS-HDEL plants with TM for 0 to 8.5 h and observed the expression of GUS reporter by histochemical staining (**Figure 1A**). TM inhibits protein N-glycosylation and thus induces ER stress (Helenius and Aebi, 2004). As can be seen, GUS staining emerged after 2-h treatment mainly at vascular bundles in roots and root tip but no staining in leaves (**Figures 1A,C**). At 3 h, an obvious staining appeared first in hydathodes of leaves (**Figures 1A,B**), which was extended to the entire cotyledons at 4 h (**Figures 1A,B**). In roots, the GUS expression was enhanced from the vasculature to the outer layer by extending the duration of TM treatment (**Figure 1C**). Despite the exogenous chemical treatment by TM, the first GUS expression was detected in the root vasculatures, an innermost tissue. This observation suggests that roots respond to TM treatment more rapidly than leaves, and that the vasculatures and root tips are the initial sites of TM-induced ER stress response in Arabidopsis roots.

### Expression Patterns of UPR Genes in Roots

To investigate whether the same set of UPR genes is induced by TM treatment in roots, we treated 7-day-old WT seedlings with TM and compared the gene expression of BiP3, BiP1/2, CALRETICULIN (CRT1: At1g56340), and CALNEXIN (CNX1: At5g61790) at 2 and 5 h after the TM treatment by qRT-PCR (**Figure 2**). The expression of all of these genes was induced both in the whole seedlings (**Figures 2A–D**) and roots (**Figures 2E–H**) upon TM treatment. These data suggest that a similar set of UPR genes was induced by TM treatment in roots as compared to the whole seedlings. Notably, however, some minor difference was observed including higher fold increase in BiP3, BiP1/2 and CNX1 at 2 h (**Figures 2A,B,D,E,F,H**). These results are consistent with the GUS reporter assay (**Figure 1**), in which roots robustly responded upon TM treatment.

### Accumulation of BiP3 Protein by TM Treatment

To detect endogenous BiP3 protein in WT plants, we produced polyclonal anti-BiP3 antibodies against a synthetic peptide corresponding to 19 amino acid residues at C-terminus of BiP3 protein because of low sequence similarity of this region with those in BiP1 and BiP2 (**Supplementary Figure S1**). To examine the specificity of the antibodies, we extracted total protein from 7-day-old seedlings of the WT and the bip3-1 mutant (**Figure 3A**) treated with TM for 5, 14, and 24 h, then performed immunoblot analysis. The bip3-1 mutant was previously reported as a null mutant (Maruyama et al., 2010). As shown in **Figure 3B**, a specific band at approximately 75 kDa was detected at 5 h after TM treatment in the WT, whose intensity was further increased at 14 h and 24 h. Because these bands were not detected in TM-treated bip3-1 mutant, the result indicated that anti-BiP3 antibodies we raised recognized BiP3 specifically. We detected BiP3 in roots as well as in the whole seedlings treated with TM for 10 h (**Figure 3C**). Thus, both roots and whole seedlings accumulate endogenous BiP3 in response to TM treatment.

### Spatiotemporal ER Stress Response in Arabidopsis Roots

Next, to observe spatiotemporal ER stress responses in intact Arabidopsis roots, we established a transgenic Arabidopsis plant that stably expresses a fluorescent mRFP protein under endogenous BiP3 promoter (ProBiP3:mRFP) in the WT background. The mRFP protein was expected to localize at cytoplasm. No mRFP signals were observed in roots without the TM treatment (at 0 h in **Figure 4**). Time-course observation of the ProBiP3:mRFP plants up to 5 h after TM treatment showed that fluorescent signal first appeared at 3.5 h (**Figure 4** and **Supplementary Figure S2**). At 5-h treatment, obvious signals were detected into the three regions, a root tip in the meristematic zone, an outer layer in the elongation zone and an inner layer in the mature zone (**Figure 4**). This pattern became more obvious at 8-h treatment (**Supplementary Figure S3**). At 24 h after TM treatment, fluorescent mRFP signals were strongly detected at inner layers including stele in the mature zone, columella and lateral root cap of the root tip (**Supplementary Figure S3**) in the meristematic zone. Since we cannot exclude a possibility that time-dependent changes and cell type-specific features of BiP3 detection might in part reflect the kinetics of TM uptake by intact plantlets, we compared mRFP signals of the ProBiP3:mRFP plants treated with dithiothreitol (DTT),

another known ER stress inducer. As shown in **Supplementary Figure S4**, the patterns with DTT treatment were similar to those with TM at 5- and 8-h treatments although the intensity of signal was slightly lower than that with TM. These localizations are in agreement with the result of GUS staining (**Figure 1**) at the early time point after TM treatment. Nevertheless, the other cell did not show the mRFP signal at 24 h, suggesting that specific cells may respond to the ER stress in Arabidopsis roots.

### Distribution of Misfolded Protein via Detection of Aggresome in Arabidopsis Roots

It has been known that the accumulation of misfolded protein in the ER triggers the ER stress response. To investigate the distribution of misfolded protein in Arabidopsis roots under ER stress conditions, we first tested detection of an aggresome by use of a commercially available aggresome detection kit (Proteostat <sup>R</sup> Aggresome Detection Kit, Enzo), which was previously reported to detect aggresome formed by adding proteasome inhibitor MG-132 in mammalian cells as well as plant cultured cells (Kothawala et al., 2012; Nakajima and Suzuki, 2013). When we treated 7-day-old WT seedlings with MG-132 for 16 h, signals for aggregation were detected in root epidermis cells (**Figures 5D–F** and **Supplementary Figures S5A– C**), but not in the cells with mock treatment (**Figures 5A– C**). This result indicates that the assay works for the intact Arabidopsis roots. Next, we treated 7-day-old WT seedlings with TM to examine whether this assay detects misfolded proteins under the ER stress condition. As can be seen in (**Figures 5J–L** and **Supplementary Figures S5D–F**), the signals were detected in root epidermal cells treated with TM, but not without the treatment (**Figures 5G–I**). Hence, the assay can be used for the detection of misfolded proteins in Arabidopsis roots.

### Localization of BiP3 and Misfolded Proteins under the ER Stress Condition

To observe a possible co-localization of misfolded protein and BiP3 under the ER stress in Arabidopsis roots, we performed immunofluorescence analysis using anti-BiP3 antibodies and fluorescent-labeled secondary antibodies with the abovementioned assay to detect misfolded proteins. When the seedlings were treated with TM for 24 h, BiP3 and misfolded proteins co-localized well at columella in root tips (**Figures 6C,G** and **Supplementary Figure S6**), indicating the co-localization of ER stress response represented by BiP3 and misfolded protein accumulation.

## DISCUSSION

Current study explored ER stress response in roots of Arabidopsis. BiP3 was more rapidly and intensively induced in root tissues compared to leaves (**Figure 1**). Indeed, the induction of BiP3 was more significant in roots for not only BiP3 but also BiP1/2 after 2 h of TM treatment (**Figure 2**). This rapid upregulation was also found for CNX1 (**Figure 2**). Between 2 h and 5 h after TM treatment, the relative fold changes of UPR genes expression were less significant in roots than whole seedlings. Moreover, high abundance in BiP3 proteins was detected in roots after 10 h of ER stress treatment (**Figure 3C**). Of note, no obvious morphological changes were found in roots during the shortterm TM treatment (**Figure 4** and **Supplementary Figure S2**), while the long-term TM treatment is known to cause the root growth defect (Cho et al., 2015; Kanehara et al., 2015b).

Interestingly, BiP3 was not ubiquitously expressed among different tissues in young seedlings under the ER stress condition. Our GUS reporter assay revealed that BiP3 is primarily expressed in vascular tissues and apical meristem in roots as well as hydathodes in leaves (**Figure 1**). This observation in roots was further elaborated by BiP3 promoter reporter assay with mRFP, which not only supports the result of GUS staining but

also detailed the location of BiP3 expression at root tips in the meristematic zone, the outer layer in the elongation zone and the inner layer in the mature zone (**Figure 4**). In addition, the co-localization of misfolded proteins and BiP3 was observed in the root tip (**Figure 6**). Although hydathodes, root caps, and root apical meristems are known for their secretory activity, root hair cells that are another highly secretory cells did not show a high BiP3 expression to the TM-induced ER stress in our observation (Battey et al., 1996; Tsugeki and Fedoroff, 1999; Komarnytsky et al., 2000; Pilot et al., 2004; Preuss et al., 2004; Cole et al., 2014). This suggests that cell-type specific responses do not simply reflect different secretory activity of individual cell types.

Taken together these observations, we propose a schematic model in which a specific response to ER stress is suggested based on the reporter assay of BiP3 following TM treatment at tissue (**Figure 7A**) or cellular (**Figure 7B**) levels. Molecular mechanisms on the ER stress response have been extensively studied in unicellular models; however, how individual tissues or cells are orchestrated in intact multicellular organisms is an enigma to date. This specific response to TM may correspond to the differential strategy or priority of ER stress response among different tissues, whose mechanistic details await future investigation based on our current study. In conclusion, we suggest that ER stress response in roots has tissue specificity.

### AUTHOR CONTRIBUTIONS

KK conceived the research and designed the experiments; YC performed the experiments and analyzed the data; YC and KK wrote the manuscripts; both authors commented on the manuscript and approved the contents.

#### FUNDING

This study was financially supported by the grants from Institute of Plant and Microbial Biology, Academia Sinica to KK and the JSPS KAKENHI (grant no. 16K07388) to KK.

#### REFERENCES


#### ACKNOWLEDGMENTS

We thank Chia-En Chen (Institute of Plant and Microbial Biology, Academia Sinica, Taiwan) for technical assistances and Dr. Tsuyoshi Nakagawa (Shimane University) for providing Gateway binary vector pGWB653 that contains the bar gene, which was identified by Meiji Seika Kaisha, Ltd.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.00144/ full#supplementary-material

FIGURE S1 | Multiple alignment of amino acid sequences of the BiPs. The carboxyl-terminal sequences used to raise BiP3-specific antibodies are underlined.

FIGURE S2 | Spatiotemporal ER stress response in Arabidopsis roots. Observation of the ProBiP3:mRFP signal in roots of 7-day-old seedlings treated with 5 µg/ml TM for the indicated time. The merged images of mRFP fluorescence and DIC are shown. Scale bars, 10 µm.

FIGURE S3 | Spatiotemporal ER stress response in Arabidopsis roots. Observation of the ProBiP3:mRFP signal in roots of 7-day-old seedlings treated with 5 µg/ml TM for the time indicated. Merged images of mRFP fluorescence and DIC are shown. The fluorescent images were provided at the left side of each merged image. Scale bars, 10 µm.

FIGURE S4 | Spatiotemporal ER stress response in Arabidopsis roots treated with DTT. Observation of the ProBiP3:mRFP signal in roots of 7-day-old seedlings treated with 5 µg/ml TM or 2 mM dithiothreitol (DTT) for the time indicated. Merged images of mRFP fluorescence and DIC are shown. The fluorescent images were provided at the left side of each merged image. Scale bars, 10 µm.

FIGURE S5 | Detection of aggregated proteins and misfolded proteins in roots. (A–C) Detection of aggregated proteins using MG-132 in roots. Seven-day-old WT seedlings were treated with MG-132 for 16 h. (D–F) Detection of misfolded proteins after TM treatment in roots. Seven-day-old WT seedlings were treated with TM for 8 h. Staining of aggregated or misfolded proteins (A,D), DIC images (B,E), and merged images (C,F). Scale bars, 10 µm.

FIGURE S6 | Immunofluorescence analysis of BiP3 and colocalization with misfolded proteins in roots. Four-day-old WT seedlings were treated with TM for 24 h. BiP3 localization detected by anti-BiP3 antibodies (A), misfolded proteins detected by Aggresome dye ProteoStat <sup>R</sup> (B), merged image (C), and DIC image (D). Scale bars, 10 µm.



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

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

# The Unfolded Protein Response Supports Plant Development and Defense as well as Responses to Abiotic Stress

Yan Bao and Stephen H. Howell\*

Plant Sciences Institute and the Department of Genetics, Development and Cell Biology, Iowa State University, Ames, IA, USA

The unfolded protein response (UPR) is a stress response conserved in eukaryotic organisms and activated by the accumulation of misfolded proteins in the endoplasmic reticulum (ER). Adverse environmental conditions disrupt protein folding in the ER and trigger the UPR. Recently, it was found that the UPR can be elicited in the course of plant development and defense. During vegetative plant development, the UPR is involved in normal root growth and development, the effect of which can be largely attributed to the influence of the UPR on plant hormone biology. The UPR also functions in plant reproductive development by protecting male gametophyte development from heat stress. In terms of defense, the UPR has been implicated in virus and microbial defense. Viral defense represents a double edge sword in that various virus infections activate the UPR, however, in a number of cases, the UPR actually supports viral infections. The UPR also plays a role in plant immunity to bacterial infections, again through the action of plant hormones in regulating basal immunity responses.

Keywords: protein folding, ER (endoplasmic reticulum) stress, IRE1, bZIP28, regulated-IRE1 dependent RNA decay (RIDD), auxin, brassinosteroid, plant virus

#### INTRODUCTION

The unfolded protein response (UPR) is widely regarded as a stress response which is activated by stress conditions in the endoplasmic reticulum (ER) (Hartl and Hayer-Hartl, 2009; Gardner et al., 2013). ER stress is brought about by a variety of different conditions that can lead to the accumulation of misfolded or unfolded proteins in the ER (Duwi Fanata et al., 2013). These conditions include abiotic stresses, such as high temperature, salt stress or biotic agents, such as viral or bacterial pathogens. The UPR is also activated under protein synthesis overload conditions when the need for protein folding simply cannot meet demands (Liu and Howell, 2010b, 2016; Kørner et al., 2015).

Stress conditions in the ER are communicated to the nucleus via the UPR signaling pathway. There are two arms to this pathway in plants (Howell, 2013a) (**Figure 1**). One arm is mediated by RNA splicing factor, IRE1, an ER transmembrane protein with its N-terminus facing the ER lumen and its C-terminus, bearing both its protein kinase and ribonuclease domains, facing the cytosol. The lumenal domain of IRE1 senses the protein status in the ER. The primary target of IRE1 in plants is bZIP60 mRNA which is spliced in response to stress (Deng et al., 2011). In Arabidopsis, IRE1's cleavage of bZIP60 mRNA results in the excision of a 23 base-pair intron

Edited by:

Minghui Lu, Northwest A&F University, China

#### Reviewed by:

Richard Strasser, University of Natural Resources and Life Sciences, Vienna, Austria Nozomu Koizumi, Osaka Prefecture University, Japan

#### \*Correspondence:

Stephen H. Howell shh@iastate.edu

#### Specialty section:

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

Received: 08 December 2016 Accepted: 27 February 2017 Published: 15 March 2017

#### Citation:

Bao Y and Howell SH (2017) The Unfolded Protein Response Supports Plant Development and Defense as well as Responses to Abiotic Stress. Front. Plant Sci. 8:344. doi: 10.3389/fpls.2017.00344

**59**

(Deng et al., 2011; Nagashima et al., 2011). The unspliced form of bZIP60 mRNA encodes a membrane-anchored transcription factor, however, splicing causes a frame shift eliminating the transmembrane domain, yielding a form of bZIP60 (bZIP60s) targeted to the nucleus (Deng et al., 2011). Under normal growth conditions, the unspliced form of bZIP60 (bZIP60u) is transcribed and translated (Iwata et al., 2008; Nagashima et al., 2011), but it is not yet clear what its function might be.

Beside its splicing function, IRE1 also attacks other mRNAs in response to stress in a process called regulated-IRE1 dependent RNA decay (RIDD) (Hollien and Weissman, 2006; Hollien et al., 2009). Studies by Mishiba et al. (2013) showed that RIDD in Arabidopsis largely targets mRNAs encoding secretory pathway proteins. Thus, IRE1 is a major factor shaping the stress transcriptome, upregulating genes by promoting the production of a potent transcription factor (bZIP60s) and by degrading other transcripts through RIDD (Mishiba et al., 2013).

stresses interfere with protein folding in the ER, leading to an accumulation of unfolded or misfolded proteins in the ER, activating the UPR. The splicing of bZIP60(u) mRNA introduces a frameshift, such that the resultant spliced form bZIP60(s) mRNA is translated into a transcription factor targeted to the nucleus. ER stress also provokes the mobilization of the membrane-anchored transcription factors from ER to Golgi, where they are processed to bZIP17(p) and bZIP28(p) by Golgi resident S1P and S2P proteases, releasing their cytosolic transcription factor domains. The factors from both branches are targeted to the nucleus, and either homodimerize or heterodimerize to bind to the promoters and regulate the expression of stress response genes. Based on Howell (2013b).

The other arm of the ER stress signaling pathway is mediated by ER membrane-associated transcription factors, bZIP17 and bZIP28 (**Figure 1**). Under unstressed conditions, these factors are retained in the ER by their association with binding protein (BiP). In response to stress, when misfolded proteins accumulate in the ER, BiP is competed away and disassociates from bZIP28 (Srivastava et al., 2013). Once liberated, the factors relocate from the ER to the Golgi, where they are cleaved by two Golgiresident proteases, Site-1 and Site-2 protease (S1P and S2P) (Liu et al., 2007a; Che et al., 2010). Cleavage by S2P in the Golgi membrane releases the cytosolic-facing components of bZIP17/bZIP28 from the Golgi allowing for their transport into the nucleus to upregulate the expression of stress response genes (Liu et al., 2007a,b; Liu and Howell, 2010a).

#### THE UPR IN VEGETATIVE DEVELOPMENT

The UPR has been extensively studied in the context of ER stress, although recently more attention has been paid to the role of the UPR in plant development and defense. The UPR has been found to play roles in both vegetative and reproductive development. Vegetative development studies have focused on root development, and under normal growth conditions, root growth is inhibited in ire1a ire1b double mutants that knock out both IRE1 isoforms in Arabidopsis (Chen and Brandizzi, 2012). IRE1a and IRE1b have overlapping functions, however, the extent of overlap has not been fully resolved (Nagashima et al., 2011; Chen and Brandizzi, 2013; Howell, 2013b). In some studies IRE1b appears to be more active in response to ER stress agents (Deng et al., 2011),while IRE1a is reported to play a more prominent role in certain biotic stress responses (Moreno et al., 2012).

IRE1 is a multifunctional protein with ribonuclease and protein kinase domains, and IRE1b has been dissected with sitespecific mutations in an effort to learn which of its functional domains is required for normal root growth (**Figure 2A**). One of the site-specific mutations knocks out the ribonuclease activity of IRE1b (N820A) while two other affect activities associated with the protein kinase domain (Deng et al., 2013). Of the latter two, one blocks the catalytic activity of the protein kinase (D628A), while the other double mutant prevents nucleotide binding (D608N, K610N), which is required for activating the ribonuclease activity of IRE1b. In complementation experiments with these site-specific IRE1b mutations, it was found that neither the mutation in the ribonuclease domain (N820A) or the nucleotide binding domain (D608N, K610N) could restore normal root growth in an ire1a ire1b double mutant (Deng et al., 2013). However, the mutation in catalytic site of the protein kinase domain (D628A) complemented root growth in the ire1a ire1b mutant, meaning that the ribonuclease function of IRE1 is necessary for normal root growth, but that the catalytic activity of IRE1's protein kinase is dispensable.

The principal target of IRE1's RNA splicing activity is bZIP60 mRNA, and so it is curious that the short root phenotype is not observed in bzip60 single or bzip28 bzip60 and bzip17 bzip60 double mutants. This finding argues that the impact of

IRE1 on root development is dependent on the ribonuclease activity of IRE1, but independent of its principal RNA splicing target bZIP60 mRNA (Deng et al., 2013). The only ribonuclease activity of IRE1 that is known to be independent of bZIP60 is its RIDD activity, the promiscuous ribonuclease activity of IRE1 in attacking other mRNAs encoding secretory proteins (Mishiba et al., 2013). Interestingly, the short root phenotype of ire1a ire1b is also observed in bzip28 ire1b seedlings, implying that the two branches of the UPR signaling pathway coordinately influence root development (Deng et al., 2013).

### THE UPR IN PLANT HORMONE BIOLOGY

The UPR is reported to play an unsuspected role in hormone biology, which may explain the influence of the UPR on vegetative growth and development. Chen et al. (2014) recently examined the possible relationship between the UPR and auxin regulation in Arabidopsis. They found, quite unexpectedly, that ER stress down-regulates the expression of genes encoding ER- and PM-localized auxin efflux transporters (PIN-formed or PIN proteins) and the auxin receptors TIR1/AFBs. The extent of down-regulation was modest, and the mechanism by which this happens is unclear. It does not require IRE1, which eliminates the possibility that RIDD might be involved in the down-regulation. One consequence of the down-regulation of auxin receptors TIR1/AFBs was the possible stabilization of AUX/IAA proteins. Chen et al. (2014) analyzed the levels DII-VENUS, a fluorescently tagged AUX/IAA surrogate that contains the degron responsible for auxin-induced TIR1/AFBmediated protein degradation. They found that the levels of DII-VENUS increased in response to ER stress suggesting that under stress, auxin responsive genes may remain repressed (or otherwise controlled) by AUX/IAA in the presence of auxin.

On the other hand, Chen et al. (2014) also reported that activation of UPR requires certain auxin regulators and ERlocalized PINs, such as PIN5 and PIN6. They found that in response to ER stress, loss-of-function pin5, pin6 and even abf1 mutants showed reduced expression of some common UPR biomarker genes, such as BIP1 and 2 and PROTEIN DISULFIDE ISOMERASE6 (PDI6). Again, the effect was modest and there was no clear explanation for this phenomenon, although the authors speculated that the mutants might affect organellar distribution of free auxin, which might influence ER stress responses in some undefined way.

The UPR has also been implicated in brassinosteroid (BR) mediated responses. The relationship appears to involve the membrane-associated transcription factors, bZIP17 and bZIP28, and not the RNA splicing arm of the UPR signaling pathway. The relationship was uncovered in mutants of S2P, a gene encoding a Golgi-resident protease that processes bZIP17 and bZIP28 when mobilized by ER stress (Che et al., 2010). Mutants in S2P have a short root phenotype, which can be overcome by expressing a preprocessed form of bZIP17 (bZIP171C) or bZIP28 (bZIP281C). The relationship of this phenomenon to BR signaling derives from the fact that high levels of a BR agonist,

brassinolide (BL), inhibit root growth in WT seedlings, but not in s2p mutants and that the root growth inhibition by BL in WT can be overcome by expressing bZIP171C or bZIP281C.

Che et al. (2010) also found that bZIP171C or bZIP281C assists in activating BR responses in bri1-5 mutants. The BR receptor in bri1-5 mutants is functional, but defective in trafficking to the cell surface. Thus, the expression of bZIP171C or bZIP281C likely aids the BR receptor in bri1-5 in trafficking, but not in other aspects of BR signaling. That notion was reinforced by the finding that bZIP171C or bZIP281C expression did not rescue bri1-6 and det2 mutants, defective respectively in BR perception and synthesis. Nonetheless, to show that bZIP171C or bZIP281C expression helped to make the BR receptor in bri1-5 operational, the authors demonstrated that the expression of either of the two bZIP1Cs partially restored BES1 dephosphorylation (a measure of BR signaling by the BRI1 receptor) and the upregulation of BR-induced genes in response to BL.

#### THE UPR IN REPRODUCTIVE DEVELOPMENT

Recently, it was shown that the UPR also plays a role in protecting plant reproductive development from elevated temperature. Plants are vulnerable to heat stress during the reproductive phase in their life cycle, and Deng et al. (2016) showed the RNA splicing arm of the UPR guides Arabidopsis reproductive development in such a way so as to protect it from elevated temperature. The authors found that the double ire1a ire1b mutant and a mutant in the immediate downstream target, bZIP60, were sterile at elevated temperature. Through, reciprocal crosses it was revealed that the temperature sensitive sterility was a male trait and impacted pollen production. The defect was also found to be sporophytic in nature, and at elevated temperature it affected the structure of the tapetum, which is ER-rich nurse tissue for the developing male gametophyte. The tapetum contributes materials for the formation of the pollen wall and coat, and it was observed that the defect dramatically affected the proper deposition of the pollen coat.

In another study, Deng et al. (2013) dissected IRE1b as described in the previous section with the intent of finding out what IRE1 activities protect male gametophyte development. Using site-specific IRE1b mutants in complementation experiments, the authors demonstrated that, both kinase and RNase functions of IRE1 are required to promote and protect male reproductive development. Deng et al. (2016) further found that bZIP60, the downstream target of IRE1, is also required for temperature protection of male gametophyte production. IRE1 splices bZIP60 mRNA to make an active transcription factor, and there were major expression changes in genes that likely contribute to pollen wall/coat construction, such as small cysteine rich pollen coat proteins. Quite surprisingly, the expression of SEC31A, a protein involved in COPII vesicle formation, restored fertility of ire1a ire1b at elevated temperature. SEC31A is the gene most highly dependent on IRE1a IRE1b function during ER stress in vegetative tissue. The basis for SEC31A's fertility restoration activity in ire1a ire1b mutants at elevated temperature is not known, but it is speculated that its contribution to ER to Golgi trafficking may compensate for other defects in the double mutant (Deng et al., 2016).

The role of the UPR in normal plant development was unsuspected because the UPR is thought to be quiescent under normal conditions and is only activated by stress. Does the unspliced form of bZIP60 produced under normal conditions, but upregulated in response to stress, have some function that we are not aware of? On the other hand, the UPR may have a low level of activity under "normal" conditions, sufficient to play a supporting role in plant development. What activates IRE1 under these conditions is not known, although it is speculated that the heavy demand for protein synthesis and/or secretion during development may activate the UPR.

### THE UPR AND MICROBIAL IMMUNITY

The UPR is also reported to play roles in bacterial immunity. Tateda et al. (2008) found that when Nicotiana benthamiana was inoculated with a non-host pathogen, Pseudomonas cichorii, and a host-specific pathogen, Pseudomonas syringae, the nonhost pathogen led to the upregulation in expression of bZIP60, while the host-specific pathogen did not. The authors silenced N. benthamiana bZIP60 using virus induced gene silencing (VIGS) and found that the plants became more susceptible to the non-host pathogen. Therefore, the authors conclude that the UPR, and more specifically the expression of bZIP60, is an important component of immunity to host pathogens.

Likewise, Moreno et al. (2012) reported that the UPR confers bacterial immunity to Arabidopsis. In their analysis, they observed that ire1a ire1b double mutants and a mutant in their downstream target, bzip60, were more susceptible to P. syringae avrRpt2. In addition, the mutants were less able to establish systemic acquired resistance (SAR) to the bacteria when treated with salicylic acid (SA), the only plant hormone which is known to induce UPR in Arabidopsis (Nagashima et al., 2014). A signature of SAR in Arabidopsis is the secretion of Pathogenesis Related Protein 1 (PR1), and low levels of secreted PR1 were found in SA-treated ire1a ire1b double mutants and also in the ire1a single mutant. From this, Moreno et al. (2012) argued that IRE1-bZIP60 branch of UPR is involved in SA-mediated plant immune responses and that mutants compromised in the UPR are more susceptible to bacterial pathogens.

## THE UPR AND VIRUS INFECTIONS

The UPR also plays a significant role in plant virus infection and immunity (Ye et al., 2011). However, the role of UPR is a double edge sword in that on the one hand the UPR appears to bolster plant immunity, but on the other hand the UPR assists in virus infection. Tobacco mosaic virus (TMV) and the tobacco N-gene is a classic case of virus resistance in plants, which appears to involve the UPR in strengthening plant immunity. The N-gene was identified and cloned by Dinesh-Kumar et al. (1995) as

a resistance R gene. During N-mediated defense, a number of genes characteristic of the UPR are upregulated including protein disulfide isomerases, ERp57 and P5, calreticulin 3, glucoseregulated protein 78 (GRP78) and BiP5 (Caplan et al., 2009). To determine whether the upregulation of these genes in tobacco was of consequence to TMV infection, VIGS was used to suppress their expression. Silencing of these genes did, indeed, result in a loss of virus containment in inoculated leaves but did not fully prevent the programmed cell death caused by the virus (Caplan et al., 2009). Thus, this observation implicates the UPR in bolstering N-mediated TMV immunity.

However, there are far more examples for how the UPR supports viral infection. ER membrane expansion is an integral part of the UPR, and Brome mosaic virus (BMV), Tobacco etch virus (TEV), Cowpea mosaic virus (CPMV), Red clover necrotic mosaic virus (RCNMV), Grapevine fan leaf virus (GFLV), and Potato virus X (PVX) are all known to induce proliferation and invaginations of the ER (Ritzenthaler et al., 1995; Schaad et al., 1997; Carette et al., 2002; Lee and Ahlquist, 2003; Lee et al., 2003; Turner et al., 2004). The ER membranes serve as a scaffold for plant virus replication and movement complexes and/or they support virion maturation (Verchot, 2016).

Infection of N. benthamiana plants with potato virus X (PVX) induces a number of genes associated with the UPR including BIP, PDI, calreticulin (CRT) and calmodulin (CAM) (Ye et al., 2011). The viral component responsible for the activation has been traced to the triple gene block protein 3, TGBp3, a viral membrane movement protein. TGBp3 delivered on its own by a tobacco mosaic virus vector will also upregulate UPRrelated factors. Not only does PVX upregulate the UPR, but the UPR helps to support PVX infection. This was revealed by silencing N. benthamiana bZIP60 and finding that the silencing inhibits virus replication in protoplasts and delays virus systemic accumulation in plants.

The UPR also supports turnip mosaic virus (TuMV) infections demonstrated by the fact that double ire1a ire1b mutant suppresses TuMV symptoms in Arabidopsis. Since the major splicing target for IRE1a and IRE1b is bZIP60 mRNA, Zhang et al. (2015) reported that a knockout in bZIP60 also suppressed viral symptoms and that the suppression of symptoms in the bZIP60 knockout could be overcome by the transgenic expression of an activated form of bZIP60. They further showed that bZIP60 is spliced in response to virus infection and that a viral membrane protein, 6K2, on its own could elicit bZIP60 mRNA splicing in a N. benthamiana transient expression system (Zhang et al., 2015).

How could it be that TGBp3 in PVX and 6K2 in TuMV elicit the UPR? There is precedence for the expression of certain

#### REFERENCES


proteins causing ER stress. In Arabidopsis chronically misfolded forms of carboxypeptidase Y, CPY<sup>∗</sup> (Finger et al., 1993) and zeolin, a fusion between two storage proteins, zein and phaseolin (Mainieri et al., 2004) produce ER-stress induced autophagy (autophagy that can be reversed by chemical chaperones) (Yang et al., 2016). Thus, TGBp3 in PVX and 6K2 in TuMV may be interpreted by the ERQC system as chronically misfolded proteins or they may interfere with the folding of other proteins.

The virulence determinant in soybean mosaic virus (SMV), a potyvirus, also elicits UPR in its host, but appears to do so by a different mode (Luan et al., 2016). The virulence factor in the potyvirus is the P3 protein, which is involved in a variety of functions including viral replication, movement and pathogenesis. Luan et al. (2016) showed that SMV P3 interacts with soybean translation elongation factor 1A (eEF1A). Using VIGs, the authors knocked down the expression of eEF1A, which reduced the ability of the plants to induce ER stress and rendered the plants more resistant to SMV.

#### CONCLUSION

The UPR, which has been long associated with stress, also functions during normal development, defense and viral infection (**Figure 2B**). The conditions that elicit the response and the consequence of its induction are current subjects of investigation.

#### AUTHOR CONTRIBUTIONS

All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.

### FUNDING

YB was supported by a grant from the National Science Foundation IOS 1353867. SH was funded by the Plant Sciences Institute at Iowa State University.

### ACKNOWLEDGMENT

We acknowledge Diane C. Bassham for reviewing the article and providing suggestions.


in Arabidopsis. Plant J. 69, 266–277. doi: 10.1111/j.1365-313X.2011. 04788.x


and nuclear relocation of a membrane-associated transcription factor, bZIP28. Plant Cell 19, 4111–4119. doi: 10.1105/tpc.106.050021


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

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

# The Banana Fruit SINA Ubiquitin Ligase MaSINA1 Regulates the Stability of MaICE1 to be Negatively Involved in Cold Stress Response

Zhong-Qi Fan, Jian-Ye Chen, Jian-Fei Kuang, Wang-Jin Lu and Wei Shan\*

State Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources/Guangdong Provincial Key Laboratory of Post-harvest Science of Fruits and Vegetables, College of Horticulture, South China Agricultural University, Guangzhou, China

#### Edited by:

Hanjo A. Hellmann, Washington State University, United States

#### Reviewed by:

Xiaosa Xu, Cold Spring Harbor Laboratory, United States Caiji Gao, South China Normal University, China

> \*Correspondence: Wei Shan shanwei@scau.edu.cn

#### Specialty section:

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

Received: 19 January 2017 Accepted: 26 May 2017 Published: 12 June 2017

#### Citation:

Fan Z -Q, Chen J -Y, Kuang J -F, Lu W -J and Shan W (2017) The Banana Fruit SINA Ubiquitin Ligase MaSINA1 Regulates the Stability of MaICE1 to be Negatively Involved in Cold Stress Response. Front. Plant Sci. 8:995. doi: 10.3389/fpls.2017.00995 The regulation of ICE1 protein stability is important to ensure effective cold stress response, and is extensively studied in Arabidopsis. Currently, how ICE1 stability in fruits under cold stress is controlled remains largely unknown. Here, we reported the possible involvement of a SEVEN IN ABSENTIA (SINA) ubiquitin ligase MaSINA1 from banana fruit in affecting MaICE1 stability. MaSINA1 was identified based on a yeast two-hybrid screening using MaICE1 as bait. Further yeast two-hybrid, pull-down, bimolecular fluorescence complementation (BiFC) and co-immunoprecipitation (CoIP) assays confirmed that MaSINA1 interacted with MaICE1. The expression of MaSINA1 was repressed by cold stress. Subcellular localization analysis in tobacco leaves showed that MaSINA1 was localized predominantly in the nucleus. In vitro ubiquitination assay showed that MaSINA1 possessed E3 ubiquitin ligase activity. More importantly, in vitro and semi-in vivo experiments indicated that MaSINA1 can ubiquitinate MaICE1 for the 26S proteasome-dependent degradation, and therefore suppressed the transcriptional activation of MaICE1 to MaNAC1, an important regulator of cold stress response of banana fruit. Collectively, our data reveal a mechanism in banana fruit for control of the stability of ICE1 and for the negative regulation of cold stress response by a SINA E3 ligase via the ubiquitin proteasome system.

#### Keywords: banana fruit, cold stress, E3 ligase, ICE1, ubiquitination

### INTRODUCTION

Banana (Musa acuminata) is one of the most popular fruit crops worldwide (Paul et al., 2016; Shan et al., 2016). As a typical climacteric fruit, bananas have a very limited shelf-life due to rapid softening (Han et al., 2016). Practically, low temperature storage is an effective technology to maintain post-harvest banana fruit qualities and to extend its shelf-life. However, being tropical fruits, banana fruit are highly sensitive to cold stress, and storage at low temperatures (<13◦C) results generally in peel browning and failure of ripening, causing severe post-harvest losses (Chen et al., 2008), which restricts the application of low temperature storage during transportation of banana fruit.

Cold stress is one of the major abiotic stresses limiting plant growth and development, productivity, product quality, post-harvest life and geographic distribution (Knight and Knight, 2012; Shan et al., 2014; Janmohammadi et al., 2015). To overcome this constraint, plants have developed sophisticated responses at the physiological and biochemical levels (Hu et al., 2013; Huang et al., 2015; Jia et al., 2016). Over the last decades, enormous progress has been achieved in the identification of important components involving in the cold signaling network, among which the INDUCER OF CBF EXPRESSION (ICE)–CREPEAT BINDING FACTOR/DRE BINDING FACTOR–COLD REGULATED (ICE-CBF-COR) transcriptional cascade is pretty well understood (Medina et al., 2011; Zhou et al., 2011; Wisniewski et al., 2014; Shi et al., 2015). In this pathway, CBFs are induced rapidly by cold stress, and in turn activate downstream COR genes to increase plant cold tolerance. ICEs encode MYC-type bHLH transcription factors (TFs) that can activate CBFs gene expression via binding to their promoters (Chinnusamy et al., 2003; Shi et al., 2015).

It is well-known that ICE-CBF-COR pathway is positively or negatively controlled by many important regulators at transcriptional, post-transcriptional, and post-translational levels. Among these regulators, CAMTA3 (calmodulin-binding transcription activator 3) (Doherty et al., 2009), SIZ1 (for SAP and Miz1) (Miura et al., 2007) and OST1 (OPEN STOMATA 1) (Ding et al., 2015) are positive regulators, while MYB15 (Agarwal et al., 2006), HOS1 (HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENES1) (Lee et al., 2001; Dong et al., 2006; Jung et al., 2014) EIN3 (ethylene insensitive 3) (Shi et al., 2012) and JA ZIM-domain 1/4 (JAZ1/4) (Hu et al., 2013) function as negative regulators of ICE-CBF-COR pathway. For example, HOS1 ubiquitinates and degrades ICE1 protein via the 26S proteasome pathway, indicating that HOS1 attenuates cold responses by triggering ICE1 degradation through the ubiquitin-proteasome system (UPS) (Lee et al., 2001; Dong et al., 2006; Jung et al., 2014). On the contrary, a small ubiquitinrelated modifier (SUMO) E3 ligase, SIZ1 sumoylates ICE1, antagonizing the polyubiquitination of ICE1 to facilitate its stability, thus causes enhanced cold tolerance (Miura et al., 2007). More recently, the protein kinase OST1 is also shown to phosphorylate ICE1 to enhance its stability and transcriptional activity, resulting in increased cold tolerance (Ding et al., 2015). These findings suggest that the regulation of ICE1 protein stability is important to ensure effective cold stress response. Although the UPS-mediated protein degradation is an important post-translational regulatory mechanism for controlling the abundance of key regulators, and has emerged as an integral player in plant response and adaptation to environmental stresses, its involvement in regulating ICE1 stability in relation to cold stress response of economical fruits, such as bananas, needs to be investigated.

Giving the increasing demand of cold storage and the cold sensitivity of banana fruit, we are aiming at the molecular mechanism(s) of the cold response in banana fruit, which will contributes to genetic improving cold tolerance, fruit quality and storage potential. Our previous studies have shown that two banana fruit MYC2 proteins act together with ICE1, which is related to the methyl jasmonate (MeJA)-induced chilling tolerance (Zhao et al., 2013). In addition, a cold-responsive NAC (NAM, ATAF1/2, and CUC2) TF MaNAC1, is a novel direct target of MaICE1 and may be associated with cold stress through interacting with MaCBF1 (Shan et al., 2014). Nevertheless, the factors controlling ICE1 protein stability associated with cold stress response of banana fruit are far from being clearly elucidated. In this study, we report that a SEVEN IN ABSENTIA (SINA) E3 ligase MaSINA1 interacts with and ubiquitinates MaICE1, leading to the degradation of MaICE1 and the attenuation of its transcriptional activity. Our study thus reveals that MaSINA1 may negatively regulate cold stress response of banana fruit via controlling MaICE1 stability.

#### MATERIALS AND METHODS

#### Plant Materials and Treatments

Pre-climacteric banana (M. acuminata, AAA group, cv. Cavendish) fruit at 75–80% maturation (about 12 weeks after anthesis) were harvested from a local commercial plantation near Guangzhou, China. Each banana hand was cut into individual fingers. Banana fruit of uniform weight, shape and maturity, and free of visual defects, were used for this study. For cold stress, fruit were stored immediately at 7◦C for 5 days, while for control, fruit were directly stored at 22◦C. Samples were taken at 0, 6, and 12 h and 1, 3, and 5 days of storage. Banana peel was collected, frozen in liquid nitrogen and stored at −80◦C for further use.

Tobacco (Nicotiana benthamiana) plants were planted in a growth chamber of 22◦C under long day conditions (16-h light/8-h dark), and 4- to 6-week-old plants were selected for analysis.

#### Yeast Two-Hybrid (Y2H) Screening/Assay

Yeast two-hybrid screening/assay was performed using the MatchmakerTM Gold yeast two-hybrid system (Clontech, Cat. No. 630489) following the User Manual. Briefly, to screen the interacting proteins, the coding sequence of MaICE1 was cloned into pGBKT7 vector to fuse with the DNA-binding domain (DBD) as bait, and transformed into yeast strain Gold Y2H by the lithium acetate method. The cDNA library (2.0 × 10<sup>9</sup> cfu/ml) was generated by TAKARA BIOTECHNOLOGY (DALIAN) CO., LTD using poly (A)<sup>+</sup> mRNAs extracted from banana fruit that were stored under cold stress, fusing to pGADT7 with activation domain (AD) and was transformed into Gold Y2H carrying the MaICE1 bait. The transformed cells (approximately 6.0 × 10<sup>6</sup> cfu) were placed on DDO medium (minimal media double dropouts, SD medium with -Leu/- Trp), and positive clones among the transformants were identified by scoring growth on QDO medium (minimal media quadruple dropouts, SD medium with -Leu/-Trp/- Ade/-His). Plasmids of positive clones was extracted from the yeast cells using a TIANprep yeast plasmid DNA kit (Tiangen) and then transformed into Escherichia coli for sequencing.

To confirm the MaSINA1-MaICE1 interaction, the coding sequences of MaSINA1 and MaICE1 were inserted into pGBKT7 or pGADT7 vector as bait and prey, respectively, and were co-transformed into Gold Y2H. Yeast cells were grown on DDO medium for 3 days, then transformed colonies were plated onto QDO medium, as well as QDO media containing 4 mg mL-1 Xα-Gal (α-Gal) for blue color development, to verify the possible interaction between MaSINA1 and MaICE1 according to their growth status and the activity of α-galactosidase. Primers used for Y2H assay are listed in Supplementary Table 1.

#### Bimolecular Fluorescence Complementation (BiFC) Analysis

To create constructs for BiFC assay, the coding sequence of MaICE1 or MaSINA1 fusing with YNE or YCE, was cloned into the pEAQ-HT vector (Sainsbury et al., 2009). The resulting constructs were then introduced into the Agrobacterium tumefaciens strain GV3101, and co-infiltrated into the abaxial side of 4- to 6-week-old tobacco (N. benthamiana) leaves using a 1-mL needleless syringe as described previously (Fan et al., 2016). Infected tissues were analyzed at 48 h after infiltration. YFP fluorescence was captured using the Confocal Spectral Microscope Imaging System (Leica TCS SP5), with an argon blue laser at 488 nm, a beam splitter for excitation at 500 nm, and a spectral detector set between 515 and 540 nm. Primers used for generating the constructs are listed in Supplementary Table 1.

#### In Vitro GST Pull-Down Assay

The full-length cDNA of MaSINA1 was cloned into the pMAL-c2X expression vector (New England Biolabs) (primers are listed in Supplementary Table 1). The maltose binding protein (MBP)-tagged MaSINA1 fusion protein was expressed in BM Rosetta (DE3) and purified by affinity chromatography using amylose resin (New England Biolabs, Cat. No. E8021S) according to the manufacturer's instructions. GST-MaICE1 protein was obtained as described previously (Shan et al., 2014). In vitro GST pull-down assay was performed as described previously with minor modifications (Liu et al., 2013). Briefly, GST or GST-MaICE1 recombinant protein was incubated with 30 µL of Glutathione resin in 1× PBS buffer for 2 h at 4◦C, the binding reaction was washed three times with 1× PBS buffer and then the MBP-MaSINA1 recombinant protein was added and incubated for an additional 2 h at 4 ◦C. The beads were washed five times with wash buffer (1× PBS, 0.1% Triton X-100), following the elution by boiling with SDS loading buffer, separated by 10% SDS-PAGE, and subjected to western blotting analysis using the anti-MBP antibody (Abcam, Cat. No. ab9084) and the anti-GST antibody (Abcam, Cat. No. ab9085) respectively, with secondary goat anti-rabbit IgG peroxidase antibody (Thermo Scientific, Cat. No. 32460). Detection was carried out using the chemiluminescent substrate SuperSignal West Pico (Thermo Scientific, Cat. No. 34080) for horse-radish peroxidase and imaged on a ChemiDocTM MP Imaging System (Bio-Rad Laboratories).

### Semi-In Vivo Coimmunoprecipitation (CoIP) and Ubiquitination Assays

To create MaSINA1-His and MaICE1-GFP constructs, full-length MaSINA1 or MaICE1 was inserted into pEAQ-HT-His and pEAQ-HT-GFP vectors (primers are listed in Supplementary Table 1), respectively (Sainsbury et al., 2009; Peyret and Lomonossoff, 2013), and were introduced into A. tumefaciens strain GV3101, following co-infiltrated into the abaxial side of 4- to 6-wk-old tobacco leaves using a 1-mL needleless syringe. After 36 h of infiltration, 10 µM MG132 (Merck, Cat. No. 474790) was injected into tobacco leaves to prevent protein degradation. After 36 h, tobacco leaves were harvested and the protein was extracted as described by Han et al. (2016) and Ye et al. (2016), as well as the following CoIP assay. A 10 µl volume of anti-GFP antibody (Abcam, Cat. No. ab290) was added to 1 mL of cell lysates. Then, binding was gently shaked at 4◦C for 4 h, and 50 µl of protein A agarose beads (Roche, Cat. No. 11134515001) was added. After 3 h of incubation at 4◦C, the precipitated samples were washed, separated by SDS-PAGE and then performed western blotting analysis as described above using 4000-fold diluted anti-His antibody (Abcam, Cat. No. ab9108) and anti-GFP antibody (Abcam, Cat. No. ab290) respectively, for CoIP assay, and anti-ubiquitin antibody (Sigma–Aldrich, Cat. No. U119) for examining the ubiquitination of MaICE1.

### Gene Isolation, Sequence and Expression Analysis

Frozen banana peel was ground in liquid nitrogen using a mortar and pestle. Total RNA was extracted using the hot borate method of Wan and Wilkins (1994), and the extract was treated with DNAse I digestion using an RNAse-free kit (Promega, Cat. No. M6101). The DNA-free total RNA was used as template for reverse-transcription PCR. The first-strand cDNA of the product was applied to PCR amplification. Quantitative realtime PCR (qRT-PCR) was carried out on a Bio-Rad CFX96 Real-Time PCR System using the GoTaq <sup>R</sup> qPCR Master Mix Kit (Promega, Cat. No. A600A) following the manufacturer's instructions. MaACT1 was used as the reference gene to normalize the gene expression levels (Chen et al., 2011). Primers for gene isolation and qRT-PCR are listed in Supplementary Table 1.

Alignments were carried out on ClustalX (version 1.83) and GeneDoc software, and a phylogenetic tree was constructed using the Neighbor-Joining method in the MEGA5 program.

### Subcellular Localization Assay

The complete Open Reading Frame (ORF) of MaSINA1 was amplified and inserted into the pEAQ-GFP vector (primers are listed in Supplementary Table 1). The MaSINA1-GFP plasmid was electroporated into the A. tumefaciens strain GV3101, and injected into the abaxial side of 4- to 6-week-old tobacco leaves as described above. pEAQ-GFP was employed as the positive control. After 48 h of infiltration, GFP signal was visualized with a fluorescence microscope (Zeiss Axioskop 2 Plus) with a beam splitter for excitation at 500 nm.

#### Promoter Isolation and Activity Analysis

Genomic DNA of banana leaves was extracted using the DNeasy Plant Mini Kit (Qiagen, Cat. No. 69104). The MaSINA1 promoter region was amplified by PCR using the specific primers listed in Supplementary Table 1. Conserved cis-element motifs in the promoter were predicted using Plant-CARE<sup>1</sup> database. The PCR product was inserted into the pGreenII 0800-LUC double reporter vector (Hellens et al., 2005) to fuse it with the Firefly luciferase (LUC) reporter gene (MaSINA1 pro-LUC). A Renilla luciferase (REN) droved by the 35S promoter at the same vector was used as an internal control. The construct CaMV35S-REN/MaSINA1 pro-LUC was infiltrated into tobacco leaf protoplasts by polyethylene glycol (PEG) methods as described previously (Shan et al., 2014; Fan et al., 2016).

The promoter activity was assayed according to Fan et al. (2016). The transformed protoplasts were incubated at room temperature (22◦C) or cold (7◦C). After 36 h, LUC and REN activities were measured on the Luminoskan Ascent Microplate Luminometer (Thermo) using the dual luciferase assay kits (Promega, Cat. No. E1910), with a 5-s delay and 15-s integrated measurements. The promoter activity is indicated by LUC/REN ratio. At least six assay repeats were included for each.

### In Vitro Ubiquitination Assay

The ubiquitination assay was generally conducted as described by Cheng et al. (2012). For E3 ubiquitin ligase activity assay of MaSINA1, 500 ng MBP-MaSINA1 recombinant protein was incubated for 2 h in the presence or absence of 50 ng of human E1 (Boston Biochem, Cat. No. E305), 250 ng of human E2 (Boston Biochem, Cat. No. E2-622), 2 mg of ubiquitin (Boston Biochem, Cat. No. u–100sc). The reaction products were subjected to western blotting using anti-MBP (Abcam, Cat. No. ab9084) and anti-Ub antibodies (Sigma–Aldrich, Cat. No. U119). For the in vitro MaSINA1-mediated MaICE1 ubiquitination assay, MBP-MaSINA1 and GST-MaICE1 proteins were coincubated as described above. The reaction products were analyzed using the anti-GST, -MBP and -Ub antibody, respectively.

### Semi-In Vivo Cell-Free Proteasomal Degradation Assay

To construct pEAQ-MaICE1-GFP-LUC, firstly the coding region of firefly LUC was amplified from pGreenII 0800-LUC as template, and inserted into pEAQ-GFP, generating pEAQ-LUC-GFP. Then the full length of MaICE1 was cloned into pEAQ-LUC-GFP to fuse in frame with LUC and GFP to give rise to pEAQ-MaICE1-LUC-GFP. The full-length cDNA of MaSINA1 without the stop codon was cloned into pEAQ vector, generating pEAQ-MaSINA1. The specific primers used for construction these vectors are shown in Supplementary Table 1. The constructs were electroporated into the A. tumefaciens strain GV3101, growing in YEP medium (1% peptone, 1% yeast extract, and 0.5% NaCl) supplemented with 100 mg/L kanamycin and 10 mg/L rifampin overnight at 28◦C. Cells were centrifuged, resuspended to OD<sup>600</sup> = 0.6 in infiltration buffer [10 mM MgCl2, 10 mM MES (pH 5.6), 100 µM acetosyringone]. After 4 h of incubation at room temperature, equal volumes (1:1) of different combinations of the strain harboring each construct were mixed and co-infiltrated into tobacco leaves as described above.

The cell free proteasomal degradation assay was performed as described previously (García-Cano et al., 2014). For MG132 treatment, 10 µM MG132 or distilled water, was injected into tobacco leaves, respectively. After 4 h of injection, leaves were harvested and ground into fine powder in liquid nitrogen. Total protein extracts from 12.5 mg of fresh leaf weight, prepared by bead-beating the tissue in 25 µL of degradation/DNase digestion buffer [10 mM Tris-HCl (pH 7.6), 0.5 mM CaCl2, 10 mM MgCl2, 5 mM DTT, 5 mM ATP, and 1× plant protease inhibitor mixture (Sigma–Aldrich)], were incubated at room temperature for 30 min in a final reaction volume of 120 µl. The protein concentration was determined using the Bradford reagent (Bio-Rad Laboratories). Reactions were terminated by boiling in SDS sample buffer, and subjected to SDS-PAGE followed by western blotting using anti-GFP antibody (Abcam, Cat. No. ab290), as described above. Putative Rubisco large chains (∼55 kDa) was adopted as the loading control (Huang et al., 2013). LUC activity was also measured using the dual luciferase assay kit, as described above.

#### Semi-In Vivo Analysis of MaICE1 Transactivational Activity

This transient expression assay was performed in tobacco leaves as described above. The MaNAC1 promoter was cloned into the pGreenII 0800-LUC double reporter vector, while MaICE1 or MaSINA1, was cloned into the pEAQ vector as effectors (Primers are listed in Supplementary Table 1). The LUC and REN activity were recorded as described above. The trans-activation ability of MaICE1 to the MaNAC1 promoter is indicated by the LUC/REN ratio. The experiments were repeated at least six times, yielding similar results.

### RESULTS

#### MaSINA1 Physically Interacts with MaICE1

In our previous study, ICE1–CBF pathway is acknowledged to be involved in cold tolerance of banana fruit (Peng et al., 2013; Zhao et al., 2013; Shan et al., 2014). In Arabidopsis, it has been well documented that ICE1 protein stability is vital for cold tolerance (Miura et al., 2011; Ding et al., 2015). To gain a deeper insight into regulators that will affect ICE1 stability of banana fruit, we screened a banana fruit cold stress-related cDNA library using MaICE1 as the bait. Among the eight candidate interacting proteins of MaICE1, one of the promising interactors (GSMUA\_Achr9G05880\_001; XP\_009416451) was particularly interesting because it encodes a SEVEN IN ABSENTIA (SINA) ubiquitin ligase (termed MaSINA1), and it was studied further. The deduced protein of this gene contains conserved RING finger and zinc finger motifs, and has high sequence identity (85%) with Arabidopsis SINAT5 (Xie et al., 2002) (**Figure 1**), which

<sup>1</sup>http://bioinformatics.psb.ugent.be/webtools/plantcare/html/

was identified as a homolog of Drosophila SINA (Carthew and Rubin, 1990). The interaction between MaSINA1 and MaICE1 was further examined by yeast two-hybrid assay. As shown in **Figure 2A**, positive α-galactosidase activity confirmed the interaction between MaSINA1 and MaICE1. Then in vitro GST pull-down assay was performed to further confirm the interaction between MaSINA1 and MaICE1. The results showed that MBP-MaSINA1 protein was pulled down by GST-MaICE1 (**Figure 2B**), also supporting that MaSINA1 interacts with MaICE1.

Subsequently, we used BiFC to determine where MaSINA1- MaICE1 complex formation occurs within the plant cell. For this assay, MaSINA1 and MaICE1was fused with the C-terminal half and N-terminal half of the YFP, respectively. Both constructs were then transiently expressed in tobacco (N. benthamiana) leaves. Reconstituted YFP fluorescence was captured in the nucleus of the tobacco leave cells when MaSINA1-YC was coexpressed with MaICE1-YN, but not with the control combinations (**Figure 2C**), revealing that MaSINA1 interacts with MaICE1 in the nucleus.

To further validate this interaction in the plant cell, CoIP assay was performed in tobacco leaves expressing MaICE1-GFP and MaSINA1-His. The protein extraction was immunoprecipitated with the anti-GFP antibody, the immunoprecipitated protein complex was analyzed by western blotting using anti-GFP and anti-His antibodies. As shown in **Figure 2D**, the MaSINA1-His protein was only detected in the immunoprecipitated complex from the leaf tissue expressing MaSINA1-His and MaICE1-GFP. No MaSINA1-His protein was appeared in the immunoprecipitated complex from the leaf tissue expressing MaSINA1-His with the vector control (empty-GFP) (**Figure 2D**). These results together clearly demonstrate that MaSINA1 directly interacts with MaICE1.

### Molecular Characterization of MaSINA1

Unlike MaICE1 expressed constitutively in the banana fruit under cold stress (Zhao et al., 2013), qRT-PCR analysis showed that MaSINA1 was down-regulated by cold stress. Compared with the expression of MaSINA1 in control fruit, the expression of MaSINA1 in the fruit directly stored at 7◦C (cold stress) decreased at 6 h, and was ∼14% of control on day 5 (**Figure 3A**).

To better understand the MaSINA1 expression in response to cold stress, MaSINA1 promoter with 1059 bp length was isolated from the genome of M. acuminata. Based on the Plant-CARE database, one low-temperature responsive element, CCGAC, termed LTRECOREATCOR15 was found in the MaSINA1 promoter at −573 bp to −577 bp from the initiation codon (Supplementary Figure 1), indicating that MaSINA1 promoter might be cold-responsive. Furthermore, a transient assay using the dual luciferase reporter system in tobacco leaf protoplasts showed that MaSINA1 promoter activity was inhibited by cold stress (**Figure 3B**). These results indicate that MaSINA1 is repressed by cold.

To investigate the subcellular location of MaSINA1, we fused the GFP with MaSINA1 protein, and transient expressed it in tobacco leaves. The GFP fluorescence of MaSINA1 fusion protein was localized predominantly in the nucleus, and some in the cytoplasm and plasma membrane (**Figure 3C**), which is similar with the localization of SINAT5 in Arabidopsis (Xie et al., 2002) and OsDIS1 in rice (Ning et al., 2011).

Previous studies showed that many RING motif-containing proteins possess E3 ubiquitin ligase activity (Stone et al., 2005). MaSINA1 contains a C3HC4-type RING finger motif at the N terminus with conserved Cys and His residues (**Figure 1**). To test whether MaSINA1 is a functional E3 ligase enzyme, we produced MaSINA1 in E. coli as a fusion with the MBP tag and purified the tagged protein (Supplementary Figure 2). Human E1, E2 and ubiquitin were used for the in vitro E3 ubiquitin ligase activity assay. Ubiquitination activity was detected using anti-MBP antibody and anti-ubiquitin antibody. As shown in **Figure 3D**, self-ubiquitination of MaSINA1 (polyubiquitinated smear ladders) was observed when Ub, E1, and E2 were present (**Figure 3D**, lane 5), but not in any negative controls that missed any one of the necessary components for the reaction (**Figure 3D**, lanes 1–4). Thus, MaSINA1 functions as an E3 ubiquitin ligase.

### MaSINA1 Targets MaICE1 for Ubiquitination Degradation

As MaSINA1 interacts with MaICE1 (**Figure 1**), and MaSINA1 functions as an E3 ubiquitin ligase, we sought to investigate whether MaICE1 protein could be ubiquitinated by MaSINA1, an in vitro ubiquitination assay was also carried out using MaICE1 protein as a substrate. Recombinant GST-MaICE1 and MBP-MaSINA1 proteins were co-incubated in the presence or absence of Ub, E1, and E2 at 30◦C for 2 h. The reaction mixture was analyzed by immunoblotting using anti-GST antibody. Co-incubation of GST-MaICE1 and MBP-MaSINA1 in the presence of Ub, E1, and E2 gave rise to a high-molecular-mass band, while lacking of Ub, E1, or E2 in the reaction mixture abolished the ubiquitinated band (**Figure 4A**), suggesting that MaSINA1 is a ubiquitin ligase capable of ubiquitinating MaICE1. The ubiquitination of MaICE1 by MaSINA1 was further confirmed in planta. MaICE1-GFP and MaSINA1-His constructs were co-expressed in N. benthamiana leaves. After immunoprecipitation with the anti-GFP antibody matrix, the immunoprecipitated complex was verified by Western blotting using the anti-ubiquitin antibody. As shown in **Figure 4B**, a smear banding representing the polyubiquitinated

absence of MG132. The resulting protein extracts were analyzed using an anti-GFP antibody. Putative Rubisco large chains (∼55 kDa) was adopted as the loading control for total protein. Bottom panel, LUC activity in each sample used in Western blot analysis. Each value represents the mean ± SE of six biological replicates. Different letters above bars indicate a statistical difference at the 5% level compared with the empty.

MaICE1 protein was detected by the anti-ubiquitin antibody in the anti-GFP-immunoprecipitated complex from the co-expression of MaICE1-GFP and MaSINA1-His, but not in the immunoprecipitated complex from the co-expression of empty-GFP and MaSINA1-His control (**Figure 4B**). Together, these data reveal that MaSINA1 is able to ubiquitinate MaICE1.

To test whether MaSINA1 can promote MaICE1 degradation through ubiquitination, the stability of MaICE1 protein in plant cells was next examined using a cell free proteasome degradation assay. The MaICE1 fused to a GFP and a LUC tag was inserted into pEAQ vector (**Figure 4C**). Total proteins were extracted from tobacco leaves transiently co-expressing GFPand LUC-tagged MaICE1 and/or MaSINA1, and the stability of MaICE1-GFP was tested by western blotting with anti-GFP antibody. As shown in **Figure 4C**, MaICE1 amounts declined substantially in the presence of MaSINA1, whereas without MaSINA1, MaICE1 remained relatively stable. This MaSINA1 mediated destabilization of MaICE1 most likely occurred by proteasome degradation pathway, because it was inhibited by MG132, a known selective inhibitor of proteasomal activity (Yang et al., 2004). Quantification of LUC activity also showed that expression of MaSINA1 protein resulted in the degradation of MaICE1 protein, while in the presence of MG132, the degradation of MaICE1 was eliminated (**Figure 4C**). Together, these findings strongly demonstrate that MaSINA1 and MaICE1 form a protein module by which MaSINA1 ubiquitinates and controls the steady protein level of MaICE1.

### MaSINA1 Attenuates the Trans-activation of MaICE1 to MaNAC1

Our previous study showed that MaICE1 bound specifically to the MYC recognition sequence in the MaNAC1 promoter and positively regulated MaNAC1 expression, which is an important transcriptional regulator of cold stress response of banana fruit (Shan et al., 2014). The result that MaSINA1 targets MaICE1 for ubiquitination degradation led us to investigate whether or not MaSINA1 interferes with transactivation of MaNAC1 by MaICE1. To this end, we performed transient expression assays using the dual-luciferase reporter system. In this experiment, the MaNAC1 promoter-driven

LUC (MaNAC1 pro-LUC) and CaMV35S promoter-driven REN (CaMV35S-REN; as an internal control) were constructed in the same vector, together with an effector plasmid expressing MaICE1 or MaSINA1 (**Figure 5A**), and expressed in the tobacco leaves. The in vivo transcriptional activity of MaICE1 was reflected by the LUC/REN ratio. Compared with the control that was co-transfected with the empty construct, co-expression of MaICE1 with MaNAC1 Pro-LUC significantly increased the LUC/REN ratio, while this increase was abolished when MaSINA1 was co-expressed (**Figure 5B**), indicating that MaSINA1 attenuates the trans-activation of MaICE1 to MaNAC1. Importantly, in the presence of MG132, the attenuation of MaSINA1 on MaICE1 transcriptional activation of MaNAC1 was diminished (**Figure 5B**). These data indicate that MaSINA1 might act as a negative regulator of cold stress response of banana fruit via controlling the stability and trans-activation of MaICE1.

#### DISCUSSION

Cold storage is effectively applied to maintain the post-harvest qualities and extend the shelf life of many horticultural fruits and vegetables. However, tropical and subtropical fruits, like bananas, are easily susceptible to chilling injury, resulting in quality deterioration and substantial losses, which significantly shortens storage life (Chen et al., 2008; Zhao et al., 2013; Shan et al., 2014). Thus, revealing the molecular mechanism(s) of the cold response in banana fruit is important for the genetic improvement of cold tolerance and improving fruit quality and storage potential. Currently, the most well-known cold signaling pathway is ICE-CBF-COR cascade (Ding et al., 2015; Shi et al., 2015). Accordingly, we also previously found that ICE-CBF pathway is positively involved in cold stress response of banana fruit (Zhao et al., 2013; Shan et al., 2014). However, regulators that act upstream in this cascade have not been identified. In this study, we sought to identify a SINA ubiquitin ligase MaSINA1 from banana fruit, and show that MaSINA1 controls MaICE1 protein stability via the UPS.

The UPS serves as a versatile post-translational modification, and has been implicated in almost all aspects of growth and development, as well as in responses to biotic and abiotic stress in plants (Lyzenga and Stone, 2012; Stone, 2014; Sharma et al., 2016). Specificity of the UPS is controlled mainly by the substrate-recruiting E3 ubiquitin ligases, and consequently, a large number of E3 ligases, have been isolated and shown to be involved in plant stress responses by modulating the abundance of key downstream stress-responsive TFs (Lyzenga and Stone, 2012; Stone, 2014; Sharma et al., 2016). However, only a few ubiquitin ligases such as HOS1, CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1), carboxyl terminus of Hsc70-interacting protein (CHIP) and Arabidopsis F-box protein 7 (FBP7), have been implicated in cold stress response (Barrero-Gil and Salinas, 2013). Arabidopsis HOS1 was the first RINGtype E3 ubiquitin ligase to act as a negative regulator of cold responses (Jung et al., 2014). Under normal growth conditions, HOS1 did not affect the nuclear localization and the abundance of GFP-ICE1, while HOS1 can interact with and ubiquitinate ICE1 for degradation under cold stress (Dong et al., 2006). Consistent with a role in mediating ICE1 degradation, overexpression of HOS1 results in reduced expression of coldresponsive genes and increased sensitivity to cold stress (Lee et al., 2001), suggesting HOS1 functions to attenuate stress signaling. Strangely, no interaction was found between banana fruit MaHOS1 and MaICE1 (data not shown). Indeed in the present work, another RING-type E3 termed MaSINA1 was identified from banana fruit. MaSINA1 showed high sequence similarity to Arabidopsis SINAT5 (**Figure 1**), and was repressed by cold

combinations by Student's t-test.

stress (**Figure 3**). We also found that MaSINA1 interacted with and ubiquitinated MaICE1 and promoted MaICE1 degradation via the UPS (**Figures 2**, **4**). In addition, we demonstrated that MaSINA1 attenuated the trans-activation of MaICE1 to MaNAC1 (**Figure 5**). Based on these results, it could be speculated that MaSINA1 negatively regulates cold stress response of banana fruit by promoting the ubiquitination-mediated degradation of MaICE1 protein. To the best of our knowledge, it is the first report that except HOS1, another novel E3 ubiquitin ligase controls ICE1 protein stability in fruits. Previously, we also showed that unlike Arabidopsis ICE1, cold-induced phosphorylation of MaICE1 might not be necessary for its transcriptional activity, but could enhance its trans-activation ability (Shan et al., 2014). The present data also reveal that the regulators of controlling ICE1 stability via UPS during cold stress response might be different in banana fruit and Arabidopsis. However, whether MaICE1 stability controlled by MaSINA1 is cold-dependent or -independent needs to be elucidated. In addition, it should be pointed out that ubiquitination is a reversible post-translational modification and de-ubiquitination enzymes (ubiquitin proteases) are also involved in modulating protein function (Barrero-Gil and Salinas, 2013). Therefore, the participation of ubiquitin proteases in plant responses to cold stress will be a subject of future research.

Besides ubiquitination, protein phosphorylation and sumoylation also play an important role in the regulation of ICE1 activity under cold stress (Chinnusamy et al., 2007; Barrero-Gil and Salinas, 2013). In contrast with HOS1, SIZ1 mediates sumoylation of ICE1, which reduces the polyubiquitination of ICE1 to enhance its stability (Miura et al., 2007). Although a very earlier research hypothesized that cold stress induces phosphorylation of ICE1 (Chinnusamy et al., 2007), until recently, the hypothesis is confirmed by Ding et al. (2015). They found that under cold stress, OST1 (OPEN STOMATA 1), a wellknown Ser/Thr protein kinase in ABA signaling, phosphorylates ICE1 to antagonize the degradation of ICE1 mediated by HOS1. More interestingly, the OST1 protein also competes with HOS1 to bind to ICE1, thus releasing ICE1 from the HOS1-ICE1 complex. Thus, the dual role of OST1 contributes to the enhancement of ICE1 stability (Ding et al., 2015). Our previous study also indicated that the trans-activation ability of MaICE1 is enhanced by MaICE1 phosphorylation. Given that ICE1 protein stability

#### REFERENCES


is regulated by various post-translational modifications, it is of interest to find out how these modifications are coordinately balanced to maintain ICE1 homeostasis in response to cold stress.

### CONCLUSION

In summary, a banana fruit SINA ubiquitin ligase MaSINA1 was identified. MaSINA1 can ubiquitinate MaICE1 for the 26S proteasome-dependent degradation, and therefore suppresses the transcriptional activation of MaICE1. Our findings reveal a novel mechanism in banana fruit for control of the stability of ICE1 and for the negative regulation of cold stress response by a SINA E3 ligase.

### AUTHOR CONTRIBUTIONS

J-YC, W-JL, and WS designed the research. Z-QF, and WS performed the experiments. Z-QF, J-YC, J-FK, W-JL, and WS wrote the manuscript.

### FUNDING

This work was supported in part by the National Key Research and Development Program (grant no. 2016YFD0400103) and the China Agriculture Research System (grant no. CARS-32-09).

### ACKNOWLEDGMENT

We thank Professor George P. Lomonossoff (Department of Biological Chemistry, John Innes Centre, Norwich Research Park) for the generous gift of pEAQ series vectors.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.00995/ full#supplementary-material

tolerance in banana fruit. Physiol. Plant. 132, 318–328. doi: 10.1111/j.1399- 3054.2007.01013.x


factor MaWRKY26 involved in the regulation of JA biosynthetic genes. Sci. Rep. 6:23632. doi: 10.1038/srep23632


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

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

## Overexpression of Hevea brasiliensis HbICE1 Enhances Cold Tolerance in Arabidopsis

Hong-Mei Yuan<sup>1</sup> \*, Ying Sheng<sup>2</sup> , Wei-Jie Chen<sup>1</sup> , Yu-Qing Lu<sup>1</sup> , Xiao Tang<sup>1</sup> , Mo Ou-Yang<sup>1</sup> and Xi Huang<sup>1</sup> \*

*<sup>1</sup> Hainan Key Laboratory for Sustainable Utilization of Tropical Bioresources, Institute of Tropical Agriculture and Forestry, Hainan University, Haikou, China, <sup>2</sup> State Key Laboratory of Hybrid Rice, College of Life Sciences, Wuhan University, Wuhan, China*

Rubber trees (*Hevea brasiliensis*) were successfully introduced to south China in the 1950s on a large-scale; however, due to the climate, are prone to cold injury during the winter season. Increased cold tolerance is therefore an important goal, yet the mechanism underlying rubber tree responses to cold stress remains unclear. This study carried out functional characterization of *HbICE1* (Inducer of *CBF* Expression 1) from *H. brasiliensis*. A nucleic protein with typical features of ICEs, HbICE1 was able to bind to MYC recognition sites and had strong transactivation activity. *HbICE1* was constitutively expressed in all tested tissues, with highest levels in the bark, and was up-regulated when subjected to various stresses including cold, dehydration, salinity and wounding. When overexpressed in *Arabidopsis*, *35S::HbICE1* plants showed enhanced cold resistance with increased proline content, reduced malondialdehyde (MDA) metabolism and electrolyte leakage, and decreased reactive oxygen species (ROS) accumulation. Expression of the cold responsive genes (*COR15A*, *COR47*, *RD29A,* and *KIN1*) was also significantly promoted in *35S::HbICE1* compared to wild-type plants under cold stress. Differentially expressed genes (DEGs) analysis showed that cold treatment changed genes expression profiles involved in many biological processes and phytohormones perception and transduction. Ethylene, JA, ABA, as well as ICE-CBF signaling pathways might work synergistically to cope with cold tolerance in rubber tree. Taken together, these findings suggest that *HbICE1* is a member of the *ICE* gene family and a positive regulator of cold tolerance in *H. brasiliensis*.

Keywords: hevea brasiliensis, ICE1, cold stress, CBF pathway, bHLH, reactive oxygen species

#### INTRODUCTION

Cold stress is one of the most devastating environmental factors adversely affecting plant growth and development, significantly constraining geographic distribution and agricultural productivity. Cold stress interferes with various physiological and biochemical processes via direct inhibition of metabolic reactions and indirect induction of osmotic, oxidative and other stresses. Hevea brasiliensis, with its high rate of production and superior rubber quality, is the sole commercial source of natural rubber. A perennial tropical tree species originating from the Amazonian forests of Brazil, rubber trees were traditionally planted within a restricted region between 15◦ north and 15◦ south latitudes. Though rubber trees were successfully introduced to south China in the 1950s

#### Edited by:

*Jie Zhou, Zhejiang University, China*

#### Reviewed by:

*Abidur Rahman, Iwate University, Japan Ren Maozhi, Chongqing University, China*

#### \*Correspondence:

*Hong-Mei Yuan yuanhongmei@hainu.edu.cn Xi Huang xihuang@hainu.edu.cn*

#### Specialty section:

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

Received: *05 April 2017* Accepted: *07 August 2017* Published: *22 August 2017*

#### Citation:

*Yuan H-M, Sheng Y, Chen W-J, Lu Y-Q, Tang X, Ou-Yang M and Huang X (2017) Overexpression of Hevea brasiliensis HbICE1 Enhances Cold Tolerance in Arabidopsis. Front. Plant Sci. 8:1462. doi: 10.3389/fpls.2017.01462*

**77**

on a large-scale, they are prone to cold injury during the winter season. Cold stress not only affects rubber production, but also threatens the survival of rubber trees in China. Increased cold tolerance is therefore a major aim of rubber tree breeding programs.

Plants have evolved various physiological, biochemical and molecular strategies aimed at adaptation to adverse situations (Nakashima et al., 2009; Thomashow, 2010; Theocharis et al., 2012). In the past decade, significant progress has been made in deciphering the key components of the cold signaling pathway in the model plants Arabidopsis and other species (Wisniewski et al., 2014; Shi Y. et al., 2015). In Arabidopsis, three C-repeat-binding factors (CBFs), CBF1, CBF2, and CBF3, have been functionally characterized (Jaglo-Ottosen et al., 1998; Medina et al., 1999). Under low temperatures, cold stress rapidly and transiently induces CBF expression, stimulating expression of cold-responsive (COR) genes by binding to Crepeat binding factor (CRT)/dehydration-responsive element (DRE) cis-elements in the promoters of COR genes, thereby increasing accumulation of proline and total sugar and protecting membranes and proteins from damage (Stockinger et al., 1997; Liu et al., 1998; Thomashow, 1999; Chinnusamy et al., 2007; Maruyama et al., 2012; Shi Y. et al., 2015). ICE1 (inducer of CBF expression 1), a constitutively expressed MYC-like bHLH transcriptional activator, also functions during cold acclimation by inducing CBF expression via binding of MYC recognition elements in CBF promoters (Chinnusamy et al., 2003, 2010; Fursova et al., 2009). The mutant ice1 was found to exhibit reduced plant tolerance to chilling and freezing stresses due to repression of AtCBF3 expression and subsequent decreases in expression of various downstream COR genes (Chinnusamy et al., 2003; Fursova et al., 2009). In contrast, plants overexpressing either AtICE1 or AtICE2 display improved freezing tolerance via enhanced AtCBF3 and AtCBF1 expression(Chinnusamy et al., 2003; Fursova et al., 2009). Thus, the ICE-CBF-COR transcriptional regulatory cascade is a wellestablished plant response to cold stress (Chinnusamy et al., 2007; Shi Y. et al., 2015). Cold acclimation, which is thought to enhance freezing tolerance after exposure to low temperatures (Thomashow, 1999), is one of the major mechanisms of plant adaptation to cold stress. But whether cold acclimation can affect the cold resistance of rubber tree has not been reported so far. Besides transcriptional regulation of ICE gene, several important components have been found to regulate cold acclimation by modulating ICE-CBF pathway at posttranslational levels. The small ubiquitin-related modifier (SUMO) E3 ligase, SIZ1 (SAP and Miz 1) and the RING finger E3 ligase HOS1 (high expression of osmotically responsive genes1) have been shown to modify ICE1 posttranslationally and function in the ICE-CBF/DREB1 signaling pathway (Dong et al., 2006; Miura et al., 2007). Recent reports further suggest that ICE1 is also regulated by various other factors such as jasmonate signaling proteins JAZ 1/4 (JASMONATE ZIM-DOMAIN 1/4), which inhibit ICE1 transcriptional activity (Hu et al., 2013). Moreover, Ding et al. (2015) discovered that the protein kinase OST1 (open stomata 1), a key component in ABA signaling, interacts with and phosphorylates ICE1 protein under cold stress, stabilizing and activating ICE1 and thereby enhancing plant tolerance to freezing temperatures. ICE1 is also degraded by the E3 ligase HOS1 (high expression of osmotically responsive gene 1)-mediated 26Sproteasome pathway (Dong et al., 2006). More recently, Huang X. S. et al. (2015) reported that ICE1 from Poncirus trifoliate functions in cold tolerance by altering polyamine accumulation via interaction with arginine decarboxylase. ICE1 is therefore not only a central component in cold signaling, but also serves as a convergence point, integrating signals to regulate cold tolerance in plants.

Studies on cold stress in rubber trees were mainly focused on changes in physiological parameters and alleviating injury after cold stress, the major components of cold signaling of rubber tree were nearly unidentified except HbCBF1and the molecular mechanisms underlying how rubber tree responses to cold stress were still very poorly understood (Cheng et al., 2015). In this study, the H. brasiliensis ICE1 (HbICE1) gene was cloned and functionally characterized for its role in cold tolerance, revealing its location in the nucleus and ability to bind to MYC recognition sites. Overexpression of 35S::HbICE1 in Arabidopsis enhanced cold resistance probably due to increased proline level, reduced MDA content and electrolyte leakage, and decreased reactive oxygen species (ROS) accumulation under cold stress. Furthermore, the cold responsive genes (COR15A, COR47, RD29A, and KIN1) were significantly activated in overexpressing plants compared to the wild-type (WT) under cold conditions. Taken together, these data suggest that HbICE1 is a functional member of the ICE gene family, playing a positive role in cold resistance in H. brasiliensis.

### MATERIALS AND METHODS

#### Plant Material and Treatments

Reyan 7-33-97 rubber trees (H. brasiliensis) cultivated at the experimental plantation of Hainan University, Hainan Province, China, were used in this study. The plants were pruned annually. To examine tissue-specific expression of HbICE1, samples from six tissues (the latex, leaves, stem, bark, stamen and pistil) were collected for RNA extraction from 17-year-old mature trees tapped for the previous two years. To determine expression of HbICE1 in response to NaCl and dehydration treatment, four batches of 12 seedlings per treatment were selected. Three batches were treated with 200 mM NaCl or 10% polyethylene glycol (PEG), and one with ddH2O then the leaves collected for RNA isolation. For cold treatment, seedlings were transferred to a culture room at 4◦C under a 12 h light/12 h dark cycle with 80% humidity for 1, 3, 6, 12, or 24 h then the leaves collected for RNA isolation. For wounding treatment, leaves of 5-year-old mature virgin (untapped) trees were wounded with a hemostat then the leaves collected for RNA extraction.

To generate 35S::GFP-HbICE1 transgenic Arabidopsis plants, the coding region of HbICE1 was amplified using primers 5′ gaattcATGCTTGATACCGACTGGTATGATA-3′ and 5′ -gaattcT

**Abbreviations:** GFP, green fluorescent protein; ICE1, Inducer of CBF Expression 1; MDA, malondialdehyde; qRT-PCR, quantitative real-time PCR; ROS, reactive oxygen species.

CACATCATTCCATGAAAGCCAGCT-3′ . The coding sequence fragment was then subcloned into the ECOR1 site of the pEGAD vector and the recombinant plasmid introduced into Arabidopsis ecotype Col-0 via Agrobacterium tumefaciens (GV3101)-mediated transformation using the floral dip method (Clough and Bent, 1998). Arabidopsis plants were sown in vermiculite in pots at 23◦C under 16 h light/8 h dark conditions with 75% humidity and a light intensity of 150 mmol m−<sup>2</sup> s −1 .

To determine cold tolerance of the transgenic plants, a cold treatment assay was performed as described previously with slight modifications (Ding et al., 2015). Fourteen-dayold seedlings cultivated in 1/2 MS medium at 23◦C were transferred to 4◦C for 2 days then exposed to −8 ◦C for 4 h and subsequently returned to normal conditions. Survival rates were measured after 7 days. To determine cold tolerance of the rubber tree seedlings, seedlings with or without cold acclimation (1 day at 4◦C) were exposed to −16◦C for 0.5 and 1 h and subsequently returned to normal conditions. The freezing treatment experiment was performed in triplicate. For Arabidopsis, T3 or T4 homozygous transgenic plants were used in this study.

### Isolation and Bioinformatics Analysis of HbICE1

The CDS of HbICE1 was predicted using Bioedit software (http:// www.mbio.ncsu.edu/BioEdit/bioedit.html) and confirmed using BLASTP on the NCBI BLAST server (http://blast.ncbi.nlm. nih.gov/Blast.cgi). The detailed Insilico cloning procedure was carried out as previously described (Hong et al., 2015). Molecular weights (MW) and isoelectric points (pI) were predicted using ExPASy (http://www.expasy.org/tools), nuclear localization signals were predicted using the online server (http://www.predictprotein.org/), and protein secondary domains were predicted with a Motif scan (http://myhits.isbsib.ch/cgi-bin/motif\_scan). Sequence alignment was performed using DNAMAN software and a phylogenetic tree constructed using MEGA 5.1 software.

#### Subcellular Localization of HbICE1

The full-length CDS of HbICE1 was fused to the C terminus of the green fluorescent protein (GFP) of the pEGAD vector, driven by the 35S promoter. GFP in the roots of homozygote transgenic Arabidopsis 35S::GFP-HbICE1 plants was examined using confocal microscopy (excitation and emission: 488 and 515 nm, respectively; 22◦C).

### Transcriptional Activation Assay and MYCR-Binding Assay of HbICE1

For the transcriptional activation assay, the open reading frame (ORF) of HbICE1 was amplified by PCR using primers P1 (cgGAATCCatgcttgataccgactggta) and P2 (aaCTGCAGcatcattccatgaaagccag). The coding sequence fragment was then subcloned into the EcoRI and PstI restriction sites of the pGBKT7 vector, giving pGBKT7-HbICE1. The recombinant vector pGBKT7-HbICE1, as well as pGBKT7- 53+pGADT7-T (positive control) and pGBKT7 (negative control), were transformed into yeast strain AH109 according to the manufacturer's instructions (Clontech, PT4084-1). The transformed yeast was placed on plates containing SD/- Trp-Leu/X-α-Gal/Aureobasidin A, SD/-Trp-Leu, SD/-Trp, SD/-Trp-His or SD/-Trp-His-Ade medium then incubated at 28◦C for 3–4 days for analysis of transformant growth.

To investigate whether HbICE1 was able to bind to the MYC recognition sequence, a yeast one–hybrid experiment (Y1H) was performed according to the manufacturer's instructions (Clontech). The ORF of HbICE1 was fused to the pGADT7 vector digested with EcoRI to create pGADT7-HbICE1. A 66 bp DNA fragment containing triple tandem repeats of the sequence containing the MYC recognition sequence (CACATG) was inserted into the pHIS2 vector, generating the recombinant vector pHIS2-MYCR. pHIS2-MYCR was then transformed into yeast strain Y187 on plates containing SD/-Trp, SD/-Trp/-His or SD/-Trp/-His/10mM 3-AT medium to verify autoactivation of pHIS2-MYCR. Both pGADT7-HbICE1 and pHIS2-MYCR were subsequently co-transformed into Y187 to verify the DNA sequence MYCR and HbICE1 protein interactions. The transformed yeast strains were placed on plates containing SD/- Trp/-Leu, SD/-Trp/-Leu/-His, or SD/-Trp/-Leu/-His/10mM 3- AT medium then incubated at 28◦C for 3–4 days for analysis of transformant growth.

### RNA Extraction and Quantitative Real-Time PCR

Hevea RNA was extracted from leaves of H. brasiliensis as described previously (Xia et al., 2011). Arabidopsis RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA) according to the instruction manual. All RNA samples were treated with RNase-free DNase I (Promega, Madison, WI, USA) to digest genomic DNA. The concentration and quality of DNaseI-treated total RNA was determined by spectrophotometry and agarose gel electrophoresis. As a template for first-strand cDNA synthesis, 2 µg DNase I-treated RNA was used according to the manufacturer's instructions (RevertAid First Stand cDNA Synthesis kit; Fermentas, Vilnius, Lithuania). Quantitative realtime PCR (qRT-PCR) was carried out using ABI-7500 Real-Time PCR apparatus with SYBR Green I dye (Takara, Tokyo, Japan) as follows: 95◦C for 3 min followed by 40 cycles of 95◦C for 20 s, 58◦C for 15 s and 72◦C for 20 s. The efficiency of each primer pair was evaluated prior to PCR using the primers listed in **Table S1**. Relative levels were calculated as 2−11CT. Each biological sample was performed with three technical repetitions, and data analysis carried out using three independent biological replicates. The whole process from plant material treatment, RNA extraction, cDNA synthesis to qRT-PCR was repeated three times per treatment as three independent biological replicates. Values were statistically analyzed by ANOVA or the Student's t-test.

### Analysis of Malondialdehyde (MDA) and Proline Contents, and Electrolyte Leakage

Three-week-old seedlings of transgenic 35S::HbICE1 and WT Arabidopsis were subjected to cold stress treatment then leaves harvested to determine MDA and proline contents, and the degree of electrolyte leakage. Proline accumulation was determined as described previously (Shi H. T. et al., 2012). Briefly, 0.25 g leaf samples were harvested and extracted in 3% sulfosalicylic acid then centrifuged at 12,000 × g for 10 min. The supernatant (2 mL) was incubated with 2 mL ninhydrin reagent [2.5% (w/v) ninhydrin, 60% (v/v) glacial acetic acid, and 40% 6M phosphoric acid] and 2 mL glacial acetic acid at 100◦C for 40 min, and the reaction terminated in an ice bath. MDA levels were measured using the thio-barbituric acid (TBA) method as described previously (Cai et al., 2015). Electrolyte leakage was measured as described previously (Lin et al., 2012). Eight leaves from transgenic and WT plants treated with and without cold treatment were placed in a bottle with 40 mL ddH2O2, shaken at 120 rpm for 3 h then conductivity (C1) measured using an ion leakage meter. Conductivity (C2) was measured after boiling the leaves for 30 min and shaking for 1 h. Electrolyte leakage was calculated as (C1/C2) × 100%.

#### Screening of Differentially Expressed Genes (Degs) Based on RNA-Seq

Total RNA was isolated from the rubber tree seedlings treated at 4◦C for 3 and12 h, including three biological repeats for each condition. RNA-Seq was performed by the Beijing Genomics Institute (Shenzhen, China). Oligo (dT) magnetic beads were used to select mRNA with polyA tail, followed by DNase I reaction to remove DNA probe. The purified mRNA was used reverse transcription to double-strand cDNA (dscDNA) by N6

random primer. End of dscDNA was repaired with phosphate at 5 ′ end and stickiness "A" at 3′ end, then ligated and with adaptor. Two specific primers of adaptor were used to amplify the ligation product. The PCR product was denatured by heat and the single strand DNA was cyclized by splint oligo and DNA ligase. The prepared library was sequenced by (Illumina HiSeqTM 2000).

Clean reads were mapped to the hevea contigs assembly using SOAPaligner/soap2 mismatches; no more than 2 bases were allowed in the alignment. The number of clean reads for each gene was calculated and then normalized to Reads Per Kb per Million reads (RPKM), which associates the read number with the gene expression levels. Further, deep analysis were performed based on DEGs, including Gene Ontology (GO) enrichment analysis, KEGG pathway enrichment analysis, cluster analysis, protein-protein interaction network analysis and finding transcription factor.

### RESULTS

#### Cloning and Bioinformatics Analysis of HbICE1

Plant ICE1-like genes play a critical role in cold tolerance in a number of different plants; however, the ICE1 gene has yet to be identified in H. brasiliensis. To clone the HbICE1 gene

from H. brasiliensis, the A. thaliana ICE1 protein sequence was used as a query sequence in a blast search against the Hevea EST and nucleotide database. The matching sequence with the lowest E-value was subsequently selected as a backbone for further insilico assembly of full-length cDNA. The predicted full-length HbICE1 cDNA contained an ORF of 1410 bps, as validated by PCR amplification and sequencing. The predicted ORF encoded a protein of 469 amino acid residues with an estimated MW of 51 kDa and a pI of 5.30. Multiple alignments of HbICE1 and ICE1 proteins from other plants indicated that the C terminus of all ICE1 proteins tested was highly conserved, whereas the N terminus varied (**Figure 1**). Furthermore, HbICE1 contained a MYC-like basic helix-loop-helix (bHLH) domain, a serine-rich region (S-rich), a zipper region (ZIP) and an ACT-UUT-ACR like domain, all of which are typical features of ICE proteins (**Figure 1**). A phylogenetic tree was reconstructed using the deduced amino acid sequence of HbICE1 and other plant ICEs, revealing that HbICE1 is most closely related to JcICE1 of Jatrophacurcas (**Figure 2**).

#### HbICE1 Is Nuclear-Localized and Has Transactivation Activity in Yeast

As a possible transcriptional factor, HbICE1 should be located in the nucleus for transcription regulation. To verify this, we created a construct expressing the GFP-HbICE1 fusion protein by fusing GFP in-frame to the 5′ end of the HbICE1 ORF under control of the cauliflower mosaic virus 35S promoter then transforming the construct into WT Arabidopsis using the floral dip method. Following selection by Basta resistance, independent transgenic lines were selected for verification by genomic PCR and qRT-PCR analysis. QRT-PCR data revealed high expression of HbICE1 in six lines (L1, L2, L3, L6, L9, and L10), three of which (L1, L3, and L6) showing highest

FIGURE 2 | Phylogenetic analysis of HbICE1 and ICEs from other plant species. The neighbor-joining method was used to construct the tree. GenBank accessions of the predicted ICE protein sequences used are as follows: JcICE1 (NP\_001306848.1), VaICE1 (AGP04218.1), AtICE1 (NP\_189309.2), RcICE1 (XP\_015570780.1), GrICE1 (XP\_012489464.1), MdICE1 (XP\_008379053.1), PeICE1 (XP\_011040262.1), PtICE1 (XP\_002318166.1), NtICE1 (XP\_009625133.1) and BjICE1 (AEB97375.2).

transcript levels were selected for further analysis (**Figure 3A**). Roots of homozygote 35S::GFP-HbICE1 plants were examined using confocal microscopy revealing exclusive expression of the GFP-HbICE1 protein in the nucleus (**Figure 3B**). This finding confirmed that HbICE1 is a nuclear protein, consistent with a previous report suggesting that Arabidopsis ICE1 is a transcription factor (Chinnusamy et al., 2003).

Next, we assayed transactivation activity of HbICE1, another important feature of transcription factors. As shown in **Figure 4A**, yeast transformed with both pGBKT7-53 and pGADT7-T (CK+) grew well on SD/-Trp-Leu and SD/-Trp-Leu/X-α-Gal/Aureobasidin A medium. Moreover, while yeast cells transformed with pGBKT7-HbICE1 grew well on SD/-Trp medium, they grew normally on SD/-Trp-His-Ade medium, exhibiting fairly strong β-galactosidase activity. In contrast, yeast cells transformed with the negative control (pGBKT7) did not grow on SD/-Trp-His medium (**Figure 4A**). These results further confirm that HbICE1 has transcriptional activation activity.

Since Arabidopsis ICE1 can bind to the MYC-recognition element (Chinnusamy et al., 2003), we also explored whether HbICE1 could bind to a sequence containing the MYC recognition element using a yeast one hybrid (Y1H) assay. As shown in **Figure 4B**, yeast cells co-transformed with pHIS2-MYCR and pGADT7-HbICE1 grew as well as those containing the positive control (p53HIS2+pGAD-Rec2-53) on SD/-Trp-Leu-His medium with and without 10 mM 3-AT. In contrast, cells co-transformed with the negative control (pHIS2MYCR+pGADT7) did not grow under this condition, further suggesting that HbICE1 binds to MYC recognition sites, activating the transcription of report genes in yeast.

### Analysis of HbICE1 Expression Patterns

To investigate tissue-specific expression of HbICE1, HbICE1 expression in the latex, leaves, stem, bark, stamen and pistil was examined by qRT-PCR. HbICE1 was universally expressed in all tissues tested with the highest expression in bark (**Figure 5A**). We also examined expression profiles of HbICE1 under various abiotic stresses including cold, dehydration, wounding, and salinity. HbICE1 was significantly induced at 3 h after cold treatment, but expression decreased to a low level at 24 h (**Figure 5B**). Under dehydration, HbICE1 mRNA was markedly induced by nearly 4.5-fold at 6 h then quickly returned to the normal level at 12 h post treatment (**Figure 5C**). After wounding treatment, the HbICE1 transcripts showed strong induction at 1 and 3 h (**Figure 5D**), while under salt stress, expression was induced 3 h after treatment (**Figure 5E**). These results indicate that HbICE1 expression is regulated during plant responses to multiple abiotic stresses.

### Over-Expression of HbICE1 in Arabidopsis Increases Cold Tolerance

To further determine the role of HbICE1 in cold tolerance, HbICE1-overexpressing lines and WT plants were subjected to freezing treatment. As shown in **Figure 6A**, HbICE1 overexpressing lines displayed less freezing damage and increased survival ratecompared with the WT after 1-week recovery. The survival rate of WT plants was only 2.8%, significantly lower than

that of the transgenic lines (77.1% for L1, 55.6% for L3, and 22.2% for L6; **Figure 6B**), suggesting that overexpression of HbICE1 conferred cold tolerance in transgenic Arabidopsis.

#### Over-Expression of HbICE1 Affects Proline Content, Electrolyte Leakage, MDA Metabolism and H2O<sup>2</sup> Accumulation under Cold Stress

Many physiological parameters such as proline content, electrolyte leakage, MDA metabolism and ROS accumulation are known indicators of tolerance to a wide variety of abiotic stresses (Feng et al., 2012; Xu et al., 2014; Cai et al., 2015; Huang X. S. et al., 2015). Thus, these physiological parameters were subsequently examined in HbICE1-overexpressing lines under cold stress. Proline accumulation is considered an adaptive response to various kinds of environmental stress, conferring stress tolerance by promoting osmotic adjustment, protecting membranes and proteins, and inhibiting ROS production. Under normal growth conditions, the proline content of the HbICE1 overexpressing lines was similar to that of the WT. However, after 24 and 48 h at 0◦C, levels were significantly higher in the transgenic lines compared to the WT. At 48 h after treatment, proline levels were 2-fold higher in the transgenic compared to the WT plants (**Figure 7A**).

Electrolyte leakage is another reliable indicator of cell membrane damage during the plant stress response (Feng et al., 2012; Huang X. et al., 2015; Cheng et al., 2016). Under normal growth conditions, the transgenic lines and WT plants showed similar levels of electrolyte leakage ranging from 13 to 15%. However, during cold treatment, electrolyte leakage was considerably less in the transgenic lines compared to the WT, suggesting a reduction in cell membrane damage in the HbICE1 overexpressing lines (**Figure 7B**), and therefore, improved cold tolerance. MDA is also an important index of plant oxidative stress and cell injury in response to stress conditions. While no significant differences in MDA content were observed between transgenic lines and WT plants under normal conditions, lower

taken before and after 7-d recovery. In (B), data represent the means of three replicates ± SD. Asterisks indicate significant differences compared with the WT under the same treatment condition (\*\*\**P* < 0.005, student's *t*-test ).

values were observed in the transgenic compared to WT plants after cold treatment (**Figure 7C**).

ROS-induced oxidative damage is associated with much of the stress-induced damage that occurs at the cellular level (Chinnusamy et al., 2007). Stress-induced ROS production results in degradation of polyunsaturated lipids, resulting in formation of MDA. In other words, the reduction in MDA content in the transgenic plants might reflect a decrease in ROS accumulation under cold stress. We therefore assayed ROS accumulation in the 35S::HbICE1 plants using DAB staining. As expected, stronger staining was observed in the WT plants compared with the transgenic plants under cold conditions (**Figure 7D**), suggesting higher levels of damage in the WT. Taken together, these findings suggest that the increased cold tolerance of 35S::HbICE1 plants is the result of increased proline accumulation, reduced electrolyte leakage and MDA metabolism, and a decrease in H2O<sup>2</sup> accumulation under cold conditions.

#### HbICE1 Positively Regulates Cold-Responsive Gene Expression under Cold Conditions

To further elucidate the molecular mechanism of the 35S::HbICE1 plant response to cold stress, we examined transcript levels of cold-responsive genes using qRT-PCR. These cold response genes (COR15A, COR47, KIN1, and RD29A), which contain DRE or related motifs, are downstream target genes of CBF (Stockinger et al., 1997). Under normal conditions, expression levels of COR15A, COR47, KIN1, and RD29A were

relatively low in both the transgenic and WT plants. In contrast, after cold exposure for 12 h, expression levels of all tested genes were significantly up-regulated in all three 35S::HbICE1 lines compared with the WT (**Figures 8A–D**), suggesting that overexpression of HbICE1 positively regulates cold-responsive gene expression, thereby contributing to improved cold tolerance.

### Analysis of Differentially Expressed Genes (DEGs) Based on RNA-Seq of Rubber Tree in Response to Cold Stress

In order to gain a global view of transcript expression in rubber tree in response to cold stress, RNA-Seq was used to analyze the differentially expressed genes, generating 24,108,424 raw sequencing reads and 24,050,817 clean reads after filtering low quality (**Table S2**). 96.50% reads of the control samples matched to a unique or multiple genomic locus, whereas 96.60% of the 3 h cold treatment sample and 96.16% of 12 h cold treatment matched, respectively (**Table S3**). A total of 8077genes showed differential expression (low false discovery rate [FDR] < 0.001 and P-value < 0.05), including 4096 genes at 3 h (1389 up-regulated, 2707 downregulated) and 6060 genes at 12 h (3188 up-regulated, 2872 down-regulated) after cold treatment (**Figure 9**, **Tables S4**, **S5**). More DEGs appeared at 12 h than at 3 h after treatment. All the DEGs could be categorized into three main clusters, e.g., biological process, cellular component and molecular function according to GO classification. In the cluster of "biological process," 17 GO terms of were significantly enriched, of which the GO terms "metabolic process," "single-organism process," "cellular process," "location," "response to stimulus" were most evidently enriched (**Figure S1**), suggesting that the these biological processes were responsive to cold stress in rubber tree.

### Pathways Enrichment Analysis of DEGs

To further reveal the biological functions of cold responsive genes in rubber tree, pathway enrichment analysis of DEGs based on KEGG database were performed, generating a scatter plot of the top 20 of KEGG enrichment (**Figure 10**). Most DGEs were enriched in "metabolic pathway" and "biosynthesis of secondary metabolites." Interestingly, "plant-pathogen interaction," and "plant hormone signal transduction" pathways were also significantly enriched. Noteworthy, the DGEs enriched in "plant hormone signal transduction" pathway showed most differentially expression in both samples, suggesting that phytohormones involved in the perception and transduction of cold induced signaling in rubber tree. In total, 50 DEGs associated with biosynthesis and/or signal transduction of the phytohormones, including 13 genes related with auxin, 13 genes related with ethylene, 9 genes related with gibberellin (GA), 9 genes related with jasmonic acid (JA), 4 genes related with abscisic acid (ABA) and 2 genes related with cytokinin (CK) (**Table 1**). Interestingly, ABA 8′ -hydroxylase 2 (scaffold0153\_318575), ABSCISIC ACID-INSENSITIVE 5-like protein 2 isoform X1 (scaffold0430\_516715), and two ABA receptors (scaffold2344\_2623, scaffold0748\_467762), were down-regulated by cold treatment at both 3 h and 12 h. Furthermore, 12 ethylene-responsive transcription factors (ERFs) showed differential expression, of which nine ERFs (scaffold0359\_512732, scaffold0636\_609282, scaffold0668\_410096, scaffold0668\_421116, scaffold0770\_505 198, scaffold0770\_519202, scaffold0782\_27868,

scaffold1267\_104008, scaffold2594\_1826) were up-regulated and three ERFs (scaffold0024\_3294824, scaffold0447\_369296, scaffold1195\_120325) were down-regulated at 12 h after cold treatment. Additionally, four genes related with JA biosynthesis, two encoding allene oxide cyclase (AOC, scaffold1038\_209563, scaffold1142\_120639), one encoding 12-oxophytodienoate reductase (OPR, scaffold0150\_641) and one encoding latex allene oxide synthase (AOS, scaffold1632\_19091), were up regulated at 12 h after cold treatment. Furthermore, three JAZ genes and one MYC2 were also significantly up-regulated, strongly suggesting that JA signaling pathway involved in cold response of rubber tree.

### Identification of Transcription Factors (TFs) in Response to Cold Stress

Numerous families of transcription factors (TFs ) are known to play crucial roles in signal transduction and regulation when plants are subjected with various abiotic stresses, including the AP2/EREBP, MYB, MYC, WRKY, NAC, and bHLH families (Agarwal and Jha, 2010; Liu et al., 2014). In the current study, 105 putative TFs were found to be exhibited differential expression in response to the cold treatment, of which those encoding AP2/EREBP domaincontaining proteins constituted the largest group (22.8%), followed by MYB proteins (16.2%), WRKY proteins (8.6%), NAC proteins (5.7%), C2H2 proteins (4.7%), bHLHs (3.8%) (**Table 2**). The largest group of the cold-mediated TFs belonged to the AP2/EREBP family and was composed of 24 members. Of these, two genes (scaffold0997\_153922, scaffold1276\_47774) were annotated as CBF/ DREB genes, which have been shown to play important roles in cold acclimation leading to freezing tolerance (Nakashima et al., 2009). In addition, four members (scaffold0103\_29681, scaffold0749\_398268, scaffold0914\_54940, scaffold1218\_8594) of bHLH family were found to be exhibited differential expression in response to the cold treatment. Of those, the expression of an ICE1-like TF (scaffold1218\_8594) was observed to up-regulated by cold treatment. Our results therefore indicated that the ICE-CBF pathway is conserved in rubber tree responses to the cold stress. Among the differentially expressed C2H2 TFs, all of the four members (scaffold0248\_1439020, scaffold0540\_449380, scaffold0760\_285386, scaffold0827\_2391, scaffold2824\_5887) were up-regulated by cold treatment. We also noticed that the cold stress mediated the expression of TFs in the WRKY (scaffold0378\_852023, scaffold0447\_61208, scaffold0447\_61719, scaffold0653\_241031, scaffold0800\_278767, scaffold0821\_417450, scaffold0844\_76279) family. The roles of members of these families in cold tolerance have been well established in numerous plants (Chinnusamy et al., 2010).

pairwise and Y axis means number of screened DEGs. Blue bar denotes down-regulated genes and orange bar for the up-regulated.

### DISCUSSION

Cold is the major environmental abiotic stress adversely affecting the growth and geographical distribution of plants. Hevea brasiliensis is native to tropical rainforests in the Amazonian basin, but was expanded to sub-optimal environments worldwide in the late 1970s, prominently northeast India, southern China, highland and coastal areas of Vietnam, and southern Brazil. Cold stress has therefore become a limiting factor not only of rubber production, but also survival of rubber trees. Chilling temperatures (0–15◦C) interfere with a number of metabolic and physiological processes such as chloroplast and mitochondria integrity, plastid membrane composition and photosynthetic electron transport, resulting in leaf wilt and lesions, bark breakdown and latex leakage, root system withering and frostbite, and ultimately cell death. Increased cold tolerance is therefore a major aim of rubber tree breeding programs.

As an upstream regulator, ICE1 plays a key role in cold signaling pathways in a wide range of plants species such as Arabidopsis (Shi Y. et al., 2012; Lang and Zhu, 2015), wheat, rice, Phalaenopsis aphrodite, Trifoliate orange, banana and grapevine (Badawi et al., 2008; Nakamura et al., 2011; Peng et al., 2013, 2014; Xu et al., 2014; Huang X. S. et al., 2015). It is therefore reasonable to expect that ICE1 homologs also play a key role in cold tolerance in H. brasiliensis. Characterization and functional analysis of the ICE1 gene in rubber trees is therefore a significant step in determining its cold signaling pathway.

Multiple sequence alignments indicate that HbICE1 shares typical ICE1 protein domains with other plant species such as the MYC-like basic helix-loop-helix (bHLH) domain, serine-rich

FIGURE 10 | Statistics of pathway enrichment of DEGs in pairwise of Control-VS-Cold3 h (A) and Control-VS-Cold12 h (B). Rich Factor is the ratio of differentially expressed gene numbers annoted in this pathway term to all gene numbers annoted in this pathway term. Greater rich Fator means greater intensiveness. *Q*-value is corrected *p*-value ranging from 0 to 1, and less *Q*-value means greater intensiveness. We just display the top 20 of enriched pathway terms.

#### TABLE 1 | Hormone-related genes that were differentially expressed during the cold treatment.


*(Continued)*



region (S-rich), zipper region (ZIP) and ACT-UUT-ACR-like domain; however, it possesses a varied N terminus (**Figure 1**). A potential sumoylation site, which is reportedly crucial for AtICE1 activation and stability (Miura et al., 2007), was also found in HbICE1 (**Figure 1**), suggesting that HbICE1 activity is mediated by the SUMO E3 ligase. Furthermore, HbICE1 was confirmed as being nuclear localized, and able to bind to the MYC-recognition element. These observations imply that HbICE1 is a novel putative ICE1 homolog.

Expression profiles revealed that, like AtICE1, HbICE1 is expressed constitutively in all tissues (Chinnusamy et al., 2003). Furthermore, HbICE1 was induced by multiple abiotic stresses including cold, dehydration, wound and salinity (**Figure 5**). HbICE1 was only slightly induced by cold, consistent with a previous report suggesting that cold stress induces little transcriptional alteration, instead resulting in posttranslational modification of ICE1 to active the CBF pathway (Chinnusamy et al., 2003; Ding et al., 2015). Upregulation of HbICE1 following dehydration was consistent with previous reports in P. trifoliate and Pyrus ussuriensis (Huang X. et al., 2015; Huang X. S. et al., 2015), but differs from AtICE1, which is not triggered by dehydration (Chinnusamy et al., 2003). This disparity between HbICE1 and AtICE1 might be attributed to the inherent differences between plant species.

Cold acclimation is one of the major mechanisms for plant to adapt to cold stress (Thomashow, 1999). Indeed, cold acclimation was found to enhance freezing tolerance rubber tree after exposure to low temperature (**Figure S2**).The electrolyte leakage was less in the cold-acclimated (CA) rubber trees to the nonacclimated (NA) rubber tree, and CA rubber trees experienced less cold stress-induced H2O<sup>2</sup> accumulation compared to the NA plants, suggesting an enhanced membrane integrity and the lower cold stress-induced H2O<sup>2</sup> accumulation in the CA rubber tree seedlings to NA plants. Overexpression of HbICE1 in Arabidopsis enhanced cold tolerance only after cold acclimation (**Figure 6**), indicating that other co-factors associated with cold acclimation are essential for HbICE1 mediated cold tolerance. Similar results were reported for wheat ICE genes (Badawi et al., 2008), where overexpression of TaICE87 or TaICE41 in Arabidopsis enhanced freezing tolerance only after cold acclimation (Badawi et al., 2008). These results suggest that other co-factors induced by cold acclimation are essential for HbICE1-mediated cold tolerance. Previous reports also confirmed this hypothesis. Chinnusamy et al. (2003) showed that cold-induced modification of the AtICE1 protein or a transcriptional cofactor is necessary for AtICE1-induced activation of CBF expression. Furthermore, Miura et al. (2007) showed that SIZ1 (SAP and Miz1), mediates sumoylation of ICE1, which reduces the polyubiquitination of ICE1 to enhance its stability. A potential sumoylation site was also found in HbICE1 protein (**Figure 1**), suggesting that HbICE1 activity could be regulated by sumoylation via the SUMO E3 ligase.

Physiological parameters such as electrolyte leakage and contents of MDA, chlorophyll and proline are closely related to cold tolerance under the regulation of COR, which is triggered by the ICE transcription factor (Badawi et al., 2008; Peng et al., 2014; Xu et al., 2014; Liu et al., 2017). In many plant species, overexpression of ICE1 is sufficient to alter physiological parameters and enhance cold tolerance (Feng et al., 2013; Huang X. et al., 2015). In our study, the HbICE1ox transgenic lines showed improved survival rates and freezing tolerance as a result reduced electrolyte leakage and MDA metabolism, and increased proline accumulation (**Figure 7**), suggesting that HbICE1 plays a positive regulatory role in the response to cold stress. These findings suggest that the ICE-CBF-COR transcriptional cascade, which influences the freezing tolerance capacity of plants, exists not only in Arabidopsis but other species such as H. brasiliensis.

H2O<sup>2</sup> is often associated with biotic and abiotic stresses (Yuan and Huang, 2016; Cao et al., 2017; Lu et al., 2017; Yuan et al., 2017). Our study further suggests that the 35S::HbICE1 plants experienced less cold stress-induced H2O<sup>2</sup> accumulation compared to the WT, despite similar contents under normal conditions. Consistence with the decrease in MDA content and electrolyte leakage under cold stress, these data imply that enhanced membrane integrity and decreased levels of lipid peroxidation caused the lower cold stressinduced H2O<sup>2</sup> accumulation in the transgenic compared to WT plants.

Transcription factors families such as AP2/EREBP, MYB, WRKY, C2H2, NAC, and bHLH are well known to be involved in stress tolerance in plants (Chinnusamy et al., 2010), but how they work synergistically to cope with cold tolerance requires

#### TABLE 2 | Differential expression transcription factor (TF) in response to the cold treatment.


*(Continued)*

#### TABLE 2 | Continued


*(Continued)*

#### TABLE 2 | Continued


determination. In this work, two genes (scaffold0997\_153922, scaffold1276\_47774) annotated as CBF/DREB and one ICE1 member of bHLH family (scaffold1218\_8594) were up-regulated by cold treatment (**Table 2**), indicating that ICE-CBF pathway is conserved in rubber tree responses to the cold stress. The phytohormones auxin, ABA, JA, and ethylene are known to play important roles in the regulation of plant growth and abiotic stress responses (Shi Y. et al., 2015). In current study, the DGEs data suggested that many phytohormones related genes were responsive to cold treatment in rubber tree. ABA integrates various stress signals and modulates stress responses, but whether it is involved in cold responses is still debated (Tuteja, 2007). It has been suggested that plant's response to the cold stress may be ABA-independent, but increasing evidences of contrary were reported (Lang et al., 1994; Wang et al., 2015). We noticed that ABA 8′ -hydroxylase, a key enzyme in ABA catabolism, was down-regulated by cold treatment. Down-regulation of the ABA 8 ′ -hydroxylase gene implies that catabolism of ABA might be attenuated in rubber tree under cold stress, which is consistent with the previous report that ABA levels increase slightly in response to low temperature (Lang et al., 1994). In addition, we detected the decreased expression of ABA receptor PYR1 and PYL2 (**Table 1**), how ABA mediated signaling is involved in the cold responses of rubber tree remains further investigation. Ethylene has been well documented in the cold stress response (Shi Y. et al., 2012, 2015). In our study, the expression of ERFs was significantly regulated by cold treatment. JA was recently reported to significantly enhance plant freezing tolerance with or without cold acclimation, (Hu et al., 2013). We observed the increased expression of several key genes for JA biosynthesis, including OPR (scaffold0150\_641), AOS (scaffold1038\_209563) and AOC (scaffold1632\_19091), and several essential genes (e.g., JAZs and MYC) in JA signaling pathways at 12 h after cold treatment, strongly suggesting that cold stress triggers JA biosynthesis and responses in rubber tree, which is consistent with the study of A. thaliana (Hu et al., 2013). JA was therefore proposed to be a positive regulator of cold responses in rubber tree.

In conclusion, a novel ICE-like transcription factor, designated HbICE1, was isolated from H. brasiliensis and functionally characterized. This nuclei protein, which has typical features of ICE proteins, was found to have transactivation activity via binding to MYC recognition sites. 35S::HbICE1 plants showed enhanced cold resistance via increased proline content, a reduction in MDA metabolism and electrolyte leakage, and a decrease in ROS accumulation, promoting expression of the cold-responsive genes. These findings suggest that HbICE1 is a member of the ICE gene family and positive regulator of cold tolerance. Differentially expressed genes (DEGs) analysis showed that cold treatment changed genes expression profiles involved in many biological processes and phytohormones perception and transduction. Ethylene, JA, ABA, as well as ICE-CBF signaling pathways might work synergistically to cope with cold tolerance in rubber tree. These findings will help elucidate the cold signaling network in H. brasiliensis, ultimately aiding breeding programs aimed at improving cold stress tolerance.

### AUTHOR CONTRIBUTIONS

HY conceived and directed this study, designed and performed the experiments, analyzed the data, wrote and revised the

#### REFERENCES


manuscript; YS, YL, MO, XT, and WC performed the experiments, analyzed the data; XH revised the manuscript. All authors have read and approved the final version submitted.

#### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China [grant number 31560197, 31260170], the Hainan Education Department Foundation [grant number Hnky2015-4], Hainan University Youth Foundation (hdkyxj201703) and the Startup Funding of Hainan University [grant number kyqd1437].

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017. 01462/full#supplementary-material

Figure S1 | GO functional classification on DEGs in pairwise of Control-vs.-Cold 3 h (A) and Control-vs.-Cold12 h (B). X axis means number of DEGs (the number is presented by its square root value). Y axis represents GO terms. All GO terms are grouped in to three ontologies: blue is for biological process, brown is for cellular component and orange is for molecular function.

Figure S2 | Electrolyte leakage (A) and DAB staining (B) of the rubber tree seedlings with indicated freezing temperatures. Seedlings were treated at −16◦C for 0.5 h or 1 h for nonacclimated (NA) and cold-acclimated (CA) plants (CA; 1 day at 4◦C). Error bars show SD from three replicates. (∗∗*P* < 0.01, ∗∗∗*P* < 0.005, student's *t*-test).

Table S1 | List of the primers used for qRT-PCR.

Table S2 | Summary of sequencing data for each sample.

Table S3 | Alignment statistics of reads align to reference genome.

Table S4 | Differentially Expressed Genes (using Noiseq) in pairwise of Control-VS-Cold 3 h.

Table S5 | Differentially Expressed Genes (using Noiseq) in pairwise of Control-VS-Cold 12 h.


MabHLHs that interact with MaICE1 in banana fruit. Planta 238, 937–953. doi: 10.1007/s00425-013-1944-7


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

# Autophagy: An Important Biological Process That Protects Plants from Stressful Environments

Wenyi Wang1, 2, Mengyun Xu<sup>2</sup> , Guoping Wang<sup>2</sup> \* and Gad Galili <sup>1</sup> \*

*<sup>1</sup> Department of Plant Science, Weizmann Institute of Science, Rehovot, Israel, <sup>2</sup> College of Horticulture, South China Agricultural University, Guangzhou, China*

Keywords: autophagy, protein degradation, abiotic stress, nbr1, TPSO, ATI1

Plants are sessile organisms that cannot escape from stressful environments, such as drought, high salinity, high temperature, and shortage of essential minerals in the soil. Hence, plants have evolved processes that protect them from these harmful conditions. One of these major processes is autophagy (which means, "self-eating"), a mechanism that destroys specific compounds that participate in efficient growth and requires extensive energy input and on the other hand stimulates biological processes that protects from the stress. Autophagy can be either a bulk process, turning over bulk amounts of various components in response to major stresses, such as serious accumulation of damaging compounds in the soil, or a selective process turning over specific components in response to specific and/or relatively minor environmental cues, such as minor shortage of rain and/or non-significant shortage of minerals in the soil (Han et al., 2011; Avin-Wittenberg et al., 2012; Liu and Bassham, 2012; Michaeli et al., 2016).

#### Edited by:

*Minghui Lu, Northwest A&F University, China*

#### Reviewed by:

*Shi Xiao, Sun Yat-sen University, China Xiaohong Zhuang, The Chinese University of Hong Kong, Hong Kong*

\*Correspondence:

*Guoping Wang gpwang@scau.edu.cn Gad Galili gad.galili@weizmann.ac.il*

#### Specialty section:

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

Received: *01 November 2016* Accepted: *19 December 2016* Published: *11 January 2017*

#### Citation:

*Wang W, Xu M, Wang G and Galili G (2017) Autophagy: An Important Biological Process That Protects Plants from Stressful Environments. Front. Plant Sci. 7:2030. doi: 10.3389/fpls.2016.02030*

#### THE AUTOPHAGY-RELATED GENES (ATGs) INVOLVED IN ABIOTIC STRESS IN PLANTS

The identification of autophagy-related genes (ATGs) was an important milestone in the understanding of the mechanism of autophagy. Thus far, over 30 ATGs have been identified in Arabidopsis, rice, tobacco, and pepper based on comprehensive, genome-wide analysis (Xia et al., 2011; Zhou et al., 2015; Zhai et al., 2016). ATGs can be divided into three categories with respect to their function: (i) ATG1-ATG13 comprising the kinase complex, an upstream regulator that initiates autophagosome formation (Suttangkakul et al., 2011); (ii) The ATG9 and ATG6/vps30 complexes involved in vacuolar protein sorting in which ATG9 interacts with ATG2 and ATG18, boosting phagophore expansion. This operates via diverse shuttling of endomembranes, such as those of the endoplasmic reticulum (ER), Golgi, and mitochondria (Tooze and Yoshimori, 2010; Yang and Klionsky, 2010; Kang et al., 2011). The ATG6/vps30 complex recruited by ATG14, which localizes to the pre-autophagosomal structure (PAS) as well as the vacuolar membrane, to generate the autophgosomes (Tooze and Yoshimori, 2010); (iii) ubiquitin like conjugation systems (ATG5-ATG12 complex and ATG8-PE complex), which is essential for autophagosome formation. The ATG8–PE complex has been proven to recruit the cytoplasmic cargo to ensure autophagosome maturation and closure, and it is subsequently transported to the vacuole for degradation (Bassham, 2007; Michaeli et al., 2016). Most of complexes participating in each selective autophagy process have been identified in yeast and animal systems. However, it is still in infancy in plants (Michaeli et al., 2016).

Among the ATG genes, ATG8 is the central protein involved in autophagy and also a marker for the autophgosome (Shpilka et al., 2011). It has been shown to participate in various processes, such as diverse intracellular trafficking, post-mitotic Golgi reassembly, cargo receptor recognition, conjugation to phosphatidylethanolamine, etc. (Kwon and Park, 2008; Tooze and Yoshimori, 2010; Yang and Klionsky, 2010). Moreover, ATG8 plays an important role in the sensitivity of plants

**94**

to abiotic stresses. There are nine isoforms of ATG8 in Arabidopsis (ATG8a to ATG8i). The over-expression of AtATG8f led transgenic Arabidopsis plants to be more sensitive to mild salt and/or osmotic stress. This kind of sensitivity was accompanied by the modification of root architecture (Sláviková et al., 2005). Furthermore, the overexpression of SiATG8a improved the tolerance to nitrogen starvation and drought stress in transgenic Arabidopsis plants (Li et al., 2015).

The other ATGs also play critical roles in the stress response, especially the response to carbon and nitrogen starvation (Han et al., 2011).Two classical autophagy-related mutants, atg5 and atg7, showed hypersensitivity to carbon and nitrogen starvation (Phillips et al., 2008; Yoshimoto et al., 2009). Autophagydefective RNAi-AtATG18a plants displayed enhanced sensitive to salt, drought and methy viologen treatments compared with wild-type plants (Xiong et al., 2007; Liu et al., 2009). Moreover, atg13 double-knockout (atg13a atg13b) and atg11 knockout plants also showed a classical atg mutant phenotype, which exhibited increased sensitivity to carbon and nitrogen starvation (Suttangkakul et al., 2011). Rice Osatg10b mutants were sensitive to treatments with high salt (250mM) and methyl viologen (MV) (Shin et al., 2009). Additionally, a series of autophagydeficient mutants, such as atg2-1, atg5-1, atg7-3, atg10-1, were hypersensitive to submergence stress (Chen et al., 2015).

#### NBR1-MEDIATED SELECTIVE AUTOPHAGY MAKES THE HIGHLY UBIQUITINATED SOLUBLE PROTEINS PRONE TO AGGREGATION DURING ABIOTIC STRESSES

Out of a number of autophagy cargo receptors, the cargo receptor NBR1 (NEIGHBOR OF BRCA1 GENE 1) is one of the critical components of the autophagy process. NBR1 was identified in yeast, mammals and plants (Johansen and Lamark, 2011). NBR1 was the first selective cargo receptor, which was found to be responsible for sequestration of ubiquitinated proteins to the vacuole for their degradation inside this organelle. However, the involvement of the cargo receptors in autophagy is poorly understood. Recently, functional homologs of NBR1 proteins were identified in Arabidopsis (AtNBR1) and tobacco (JOKA2) plants (Svenning et al., 2011; Zientara-Rytter et al., 2011).

The Arabidopsis NBR1 is a homolog of the mammalian autophagic adaptor NBR1, with an ubiquitin-association domain binds selectively to six of the nine Arabidopsis ATG8 protein isoforms. The nbr1 mutant exhibits sensitivity to a spectrum of abiotic stresses, similar to the autophagy-deficient atg5 and atg7 mutants, but had no obvious effect onthe response to carbon starvation and resistance to a necrotrophic fungal pathogen. This indicates that AtNBR1 participates in the response to abiotic stress (Zhou et al., 2013). Under high heat conditions, an insoluble highly ubiquitinated detergent-resistant protein was shown to be prone to aggregation, thereby enabling recognition by NBR1 and its subsequent transport to the vacuole for degradation. Moreover, Rubisco activase and a number of catalases are linked to the response of plants to a wide variety of abiotic stresses and these enzymes accumulated in the nbr1 mutant (Zhou et al., 2014). These findings suggest that NBR1 mediated selective autophagy pathway plays a critical role during abiotic stresses (**Figure 1**).

#### ABIOTIC STRESS-INDUCED TSPO-RELATED PROTEIN REDUCES AQUAPORIN PIP2;7 THROUGH AUTOPHAGIC DEGRADATION

To decipher the involvement of autophagy in abiotic stress in plants, the identification of the cargo receptor in the process of autophagy is crucial. This is because of the following reasons (i) under normal conditions, autophagy maintains a basal level for homeostasis; (ii) under abiotic stress, cargo receptors are active to remove the damaged or unwanted materials or to recycle the materials for providing anabolic substrates and metabolites to the cells. In plants, one of the critical processes of autophagy is the transport of the unnecessary components to the vacuole for degradation, which is a selective process requiring the cargos.

To minimize the effects of abiotic stresses, plants have developed a sophisticated protein quality control system that maintains the protein homeostasis. When subjected to abiotic stresses, such as heat and drought, the earliest response is inhibition of protein synthesis and increase in protein folding and processing. Recently a new Arabidopsis cargo receptor was identified and was named TRYPTOPHAN-RICH SENSORY PROTEIN (TSPO). It is a multi-stress regulator that is transiently induced by abiotic stress and is originally described as a heme-binding protein, which interacts with ATG8 to enable the degradation of porphyrins via an autophagy-dependent degradation mechanism (Vanhee et al., 2011). Moreover, TSPO interacts intracellularly with the plasma membrane aquaporin PIP2;7 (PLASMA MEMBRANE INTRINSIC PROTEIN 2;7) and downregulates it in the cell. The coexpression of TSPO and PIP2;7 led to decreased levels of PIP2;7 in the plasma membrane and abolished the membrane water permeability mediated by the overexpression of PIP2;7 in transgenic seedlings. Furthermore, ABA treatment activates TSPO and triggers the degradation of PIP2;7 through the autophagic pathway. These findings suggest that TSPO acts as a selective plant-specific autophagy cargo receptor during abiotic stress (Hachez et al., 2014). Remarkably, the autophagy-mediated reduction in the quantity of PIP2;7 modulates the osmotic water permeability of membranes, which is important during heat and drought stress (**Figure 1**).

### ATI1, A STRESS-ASSOCIATED PROTEIN ASSEMBLES INTO DIFFERENT TYPES OF NOVEL BODIES ASSOCIATED WITH EITHER ER OR PLASTIDS

With the aim of elucidating the biological processes in plants that are associated with selective autophagy in plants, our lab identified a number of plant-specific ATG8f binding proteins based on a yeast two-hybrid analysis. One of these

process of TSPO-mediated selective autophagy pathway, TSPO is transcriptionally upregulated during heat and drought stress, then interacts with PIP2;7 and ATG8 to enable the degradation of porphyrins *via* an autophagy-dependent degradation mechanism (A). The second pathway is ATI1 interact with chloroplast protein in the ATI1-body, then this compound transport their cargo proteins to autophagosomes, which eventually transport to the vacuole for degradation (B). In the NBR1-mediated selective autophagy pathway, an insoluble highly ubiquitinated detergent-resistant protein and ATG8 were shown to be prone to aggregation to form autophgosome, thereby enabling recognition by NBR1 and its subsequent transport to the vacuole for degradation (C).

proteins, named "Autophagy Interacting Protein 1" (ATI1), was further subjected to detailed studies. When grown under regular, non-stress conditions, ATI1 was partially associated with the endoplasmic reticulum(ER) membrane. Furthermore, upon exposure of the plants to carbon or nitrogen starvation, ATI1 was assembled into two different types of novel bodies that were associated with either the ER or the plastids (Honig et al., 2012).

When the plants were exposed to carbon or nitrogen starvation, ATI1 was incorporated into novel bodies that were either moved along the ER network, or localized inside the plastids. These novel bodies were then transported into the central vacuole in which their contents were apparently being turned over inside the plastids (**Figure 1**). Interestingly, the seedlings of the over-expressing ATI1 plants germinated faster and showed increased tolerance to carbon starvation and salt stress, whereas the plants with suppressed expression of ATI1 showed reduced tolerance to carbon starvation and salt stress, indicating that the biological processes using ATI1 confer faster growth and increased stress tolerance to the germinating seedlings (Michaeli et al., 2014).

The above results imply the ATI1 is a multifunctional protein, which is associated with ER-to-vacuole and plastid-to-vacuole trafficking by ATG8-mediated selective autophagy. Moreover, ATI1 is also involved in an autophagic system that promotes the seedling organization under abiotic stress conditions.

### CONCLUDING REMARKS

Despite the identification of three potential cargo receptors in plants, a detailed understanding of these cargos is imperative. The questions that need to be addressed are whether there are links between the different cargo-mediated autophagy pathways, the mechanisms for the recognition and delivery of the misfolded and damaged proteins and organelles, and how the cargos are trafficked. We believe that with an in-depth research on the involvement of autophagy in abiotic stress, particularly in crop plants, it is possible to open new avenues for the enhancement of stress tolerance by using genetic and/or genetic engineering approaches, ultimately leading to enhanced production.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

WW and GG conceived and designed the project, and wrote the manuscript. MX and GW helped with figure revision. The manuscript was approved by all other authors.

favourable growth conditions and under starvation stresses in Arabidopsis plants. J. Exp. Bot. 56, 2839–2849. doi: 10.1093/jxb/eri276


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

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

# The AMP-Activated Protein Kinase KIN10 Is Involved in the Regulation of Autophagy in Arabidopsis

Liang Chen1,2, Ze-Zhuo Su<sup>1</sup> , Li Huang<sup>1</sup> , Fan-Nv Xia<sup>1</sup> , Hua Qi<sup>1</sup> , Li-Juan Xie<sup>1</sup> , Shi Xiao<sup>1</sup> \* and Qin-Fang Chen<sup>1</sup> \*

<sup>1</sup> State Key Laboratory of Biocontrol and Guangdong Provincial Key Laboratory of Plant Resources, School of Life Sciences, Sun Yat-sen University, Guangzhou, China, <sup>2</sup> College of Life Sciences, South China Agricultural University, Guangzhou, China

#### Edited by:

Minghui Lu, Northwest A&F University, China

#### Reviewed by:

Taijoon Chung, Pusan National University, South Korea Martin Dickman, Texas A&M University, United States

#### \*Correspondence:

Shi Xiao xiaoshi3@mail.sysu.edu.cn Qin-Fang Chen chenqf3@mail.sysu.edu.cn

#### Specialty section:

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

Received: 30 December 2016 Accepted: 26 June 2017 Published: 10 July 2017

#### Citation:

Chen L, Su Z-Z, Huang L, Xia F-N, Qi H, Xie L-J, Xiao S and Chen Q-F (2017) The AMP-Activated Protein Kinase KIN10 Is Involved in the Regulation of Autophagy in Arabidopsis. Front. Plant Sci. 8:1201. doi: 10.3389/fpls.2017.01201 Autophagy is a highly conserved system in eukaryotes for the bulk degradation and recycling of intracellular components. Autophagy is involved in many physiological processes including development, senescence, and responses to abiotic and biotic stress. The adenosine 5'-monophosphate (AMP)-activated protein kinase AMPK positively regulates autophagy in mammals; however, the potential function of AMPK in plant autophagy remains largely unknown. Here, we identified KIN10, a plant ortholog of the mammalian AMPK, as a positive regulator of plant autophagy and showed that it acts by affecting the phosphorylation of ATG1 (AUTOPHAGY-RELATED GENE 1) proteins in Arabidopsis. Transgenic Arabidopsis lines overexpressing KIN10 (KIN10- OE) showed delays in leaf senescence, and increased tolerance to nutrient starvation, these phenotypes required a functional autophagy pathway. Consistent with KIN10 having a potential role in autophagy, the nutrient starvation-induced formation of autophagosomes and cleavage of GFP-ATG8e were accelerated in the KIN10-OE lines compared to the wild type. Moreover, the KIN10-OE lines were less sensitive to drought and hypoxia treatments, compared with wild type. Carbon starvation enhanced the level of phosphorylated YFP-ATG1a in the KIN10-OE lines compared to that of wild type. Together, these findings suggest that KIN10 is involved in positive regulation of autophagy, possibly by affecting the phosphorylation of ATG1s in Arabidopsis.

#### Keywords: ATG1, AMPK, autophagy, KIN10, phosphorylation

#### INTRODUCTION

Autophagy is the process of degradation and recycling of cytoplasmic organelles, proteins, and macromolecules, and is highly conserved among eukaryotes. Autophagy is activated by a variety of stress factors, such as nutrient deprivation, hypoxia, reactive oxygen species, and infection by pathogen (Kroemer et al., 2010; Han et al., 2011). Autophagy plays an essential role in the maintenance of cellular homeostasis under changing nutrient conditions. Among the three types of autophagy, macroautophagy (hereafter referred to as autophagy) is the predominant form (Klionsky, 2007). During autophagy, double-membrane vesicles, called autophagosomes, are formed from the expanding membranes of preautophagosomal structures; these autophagosomes sequester the enclosed components and deliver them to the lysosome/vacuole for degradation.

**98**

The serine/threonine protein kinase ULK1(Unc-51-like kinases 1, mammalian homologs of ATG1) activates autophagy in response to developmental cues or stress signals by initiating autophagosome formation (Mizushima, 2010; Wirth et al., 2013; Wong et al., 2013). In mammalian cells, ULK1 activity is directly controlled by the target of rapamycin (TOR) and the AMPdependent protein kinase (AMPK) (Kim et al., 2011; Shang and Wang, 2011; Alers et al., 2012). Under nutrient-rich conditions, the activated TOR kinase disrupts the ULK1-ATG13 complex by phosphorylating the ATG13 subunit, and thereby inhibits autophagy. However, under nutrient starvation conditions, AMPK directly phosphorylates ULK1 at the Ser 317 and Ser 777 residues, subsequently activating autophagy (Kim et al., 2011). Alternatively, AMPK may activate autophagy by inhibiting TORC1 (TOR complex 1) activity (Gwinn et al., 2008; Lee et al., 2010). ULK1 may also be involved in the termination of autophagy. Specifically, ULK1 represses AMPK activity through a negative regulatory feedback loop (Löffler et al., 2011). Similarly, another study suggests that the KLHL20-mediated ubiquitination and degradation of ULK1 contributes to the termination of autophagy (Liu et al., 2016).

In plants, the Snf1-related kinase 1 (SnRK1), a homolog of the yeast Snf1 and mammalian AMPK, is a highly conserved energy sensor and is activated under energy deprivation (Polge and Thomas, 2007; Baena-González and Sheen, 2008; Emanuelle et al., 2015). SnRK1 is composed of one catalytic α subunit (KIN10 and 11 in Arabidopsis) and two regulatory subunits, β and γ (Polge and Thomas, 2007; Emanuelle et al., 2015). Overexpression of KIN10 delays flowering and leaf senescence in Arabidopsis, suggesting that KIN10 play a positive role in the regulation of growth and development as well as energy signaling (Baena-González et al., 2007). Although Snf1/AMPK likely promotes autophagy by directly activating ATG1/ULK1 in yeast and animals, the role of SnRK1 in plant autophagy is not well characterized.

Our study demonstrated that Arabidopsis KIN10 is a positive regulator of autophagy. Under nutrient starvation, transgenic plants overexpressing KIN10 (KIN10-OE) showed enhanced autophagosome formation and increased tolerance to nutrient deprivations. Furthermore, the level of starvation-induced phosphorylation of ATG1 increased in the KIN10-OE lines, suggesting that KIN10 is likely involved in positive regulation of autophagy, possibly by affecting the phosphorylation of ATG1 proteins.

### MATERIALS AND METHODS

#### Plant Materials, Growth Conditions, and Treatments

The KIN10 overexpression lines (OE-1 and OE-2) and KIN10 RNAi lines (RNAi-1 and RNAi-7) used in this study were in the Arabidopsis Landsberg erecta (Ler) background (Baena-González et al., 2007). The autophagy-related mutants atg5-1 and atg7- 3 (Thompson et al., 2005; Lai et al., 2011; Chen et al., 2015; Col-0 ecotype) were backcrossed twice to the Ler wild-type plants to obtain atg5-L and atg7-L plants. The atg5-L and atg7-L mutants were further crossed to the OE-1 line to generate OE-1 atg5-L and OE-1 atg7-L lines. Transgenic lines expressing GFP-ATG8e and YFP-ATG1a, driven by the CaMV 35S promoter, have been previously described (Xiao et al., 2010; Suttangkakul et al., 2011). All Arabidopsis seeds were surface-sterilized with 20% bleach containing 0.1% Tween-20 for 20 min, and then washed 5 times with sterile water. Seeds were sown on Murashige and Skoog (MS) medium, followed by cold treatment in the dark for 3 days. Seven days after germination, seedlings were transplanted into soil and grown in a plant growth room with a16-h-light/8-h-dark cycle at 22◦C.

For the carbon starvation treatment, 7-day-old MS-grown seedlings or 4-week-old soil-grown plants were transferred to continuous darkness for the indicated duration followed by recovery under normal growth conditions for 7 days. Samples were collected or photographed at the indicated time points. To calculate the survival rate after darkness, at least 18 plants per genotype were dark-treated followed by a 7-day recovery. The number of surviving plants, where survival if defined as the capability to produce new leaves, was recorded. For the nitrogen starvation treatment, 7-day-old seedlings grown on MS medium were transferred to solid MS or nitrogen-deficient solid MS medium and grown for 5 days. Chlorophyll contents were measured and calculated after the recovery.

#### RNA Extraction and Quantitative Reverse-Transcription PCR Analysis

Total RNA extraction and quantitative reverse-transcription PCR (qRT-PCR) analysis were performed as previously described (Chen et al., 2015). Briefly, the isolated RNA was reverse transcribed using the PrimeScript RT Reagent Kit with gDNA Eraser (Takara, RR047A) following the manufacturer's instructions. The qPCR was carried out using SYBR Green master mix (Takara, RR420A) on a StepOne Plus real-time PCR system (Applied Biosystems). The conditions for the qPCR were: initial denaturation at 95◦C for 5 min followed by 40 cycles of PCR (denaturing at 95◦C for 10 s, annealing at 55◦C for 15 s, and extension at 72◦C for 30 s). Three experimental replicates were used for each reaction. ACTIN2 was used as the reference gene. Gene-specific primers used for the qPCR analysis are listed in Supplementary Table S1.

#### Laser Scanning Confocal Microscopy

The stable transgenic lines expressing a GFP-ATG8e fusion protein were used to monitor autophagosome formation (Xiao et al., 2010). Seven-day-old GFP-ATG8e seedlings grown in MS solid medium were transferred to MS medium or MS medium lacking sugars (MS-C) under darkness for the indicated times. After treatment, the primary root cells were observed using an LSM 780 NLO laser scanning confocal microscope (Carl Zeiss).

#### Western Blot Analysis

Total protein extraction was performed as previously described (Chen et al., 2015). Briefly, 4-week-old plant leaves or 7-day-old whole seedling were ground in liquid nitrogen and homogenized in ice-cold extraction buffer (50 mM sodium phosphate, pH

7.0, 200 mM NaCl, 10 mM MgCl2, 0.2% β-mercaptoethanol and 10% glycerol) supplemented with protease inhibitor cocktail (Roche, 04693132001). Total homogenates were placed on ice for 30 min, and then centrifuged for 30 min at 11,000 g. The supernatant was transferred to a new microfuge tube for SDS-PAGE electrophoresis.

For immunoblot analysis, total proteins were subjected to SDS-PAGE and electrophoretically transferred to a Hybond-C membrane (Amersham, 10600016). Anti-GFP antibodies were used to detect GFP as previously described (Chen et al., 2015). YFP was detected with rabbit anti-GFP polyclonal antibodies (Abcam, ab290).

#### Phosphatase Treatment

Phosphatase treatment was performed according to Suttangkakul et al. (2011) with minor modification. Seven-day-old YFP-ATG1a and YFP-ATG1a/KIN10-OE seedling were homogenized in ice-cold protein extraction buffer supplemented with 1 mM phenylmethylsulfonyl fluoride and protease inhibitor cocktail (Roche). Samples were placed on ice for 30 min, and then centrifuged for 30 min at 11,000 g. The supernatant was incubated with λ protein phosphatase (New England Biolabs) with or without addition of phosphatase inhibitor PhosSTOP (Roche) for 30 min at 30◦C. 2 × SDS-PAGE sample buffer was added to the total sample and heated to 95◦C for 5 min.

#### Statistical Analysis

Data are reported as means ± SD of three independent experiments unless otherwise indicated. The significance of the differences between groups was determined by a two-tailed Student's t-test. P-values < 0.05 or < 0.01 were considered significant.

#### Accession Numbers

Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank databases under the following accession numbers: KIN10 (At3g01090), ATG1a (At3g61960), ATG1b (At3g53930), ATG1c (At2g37840), ATG2 (At3g19190), ATG5 (At5g17290), ATG6 (At3g61710), ATG7 (At5g45900), ATG8a (At4g21980), ATG8e (At2g45170), ATG9 (At2g31260), ATG10 (At3g07525), ATG13a (At3g49590), ATG13b (At3g18770), ATG18a (At3g62770), and PI3K (At1g60490).

#### RESULTS

#### Transgenic Plants Overexpressing KIN10 Showed Delayed Leaf Senescence and Enhanced Tolerance to Nutrient Starvations

The KIN10 overexpression lines (OE-1 and OE-2) showed delayed leaf senescence (Baena-González et al., 2007). To examine the potential role of Arabidopsis KIN10 in autophagy, we further examined the response of the KIN10-OE lines and KIN10 RNA interference lines (RNAi-1 and RNAi-7) to naturally induced senescence and nutrient deficiency. The RNA and protein level of KIN10 in the KIN10-OE and KIN10-RNAi lines were first confirmed by qRT-PCR and western blot analyses (**Supplementary Figure S1**). Under normal growth conditions, both KIN10-OE lines displayed slower growth and delayed natural senescence compared to the wild type, while the KIN10-RNAi lines showed similar phenotypes to the wild-type plants (**Figure 1A**). The level of chlorophyll in the leaves of 6-week-old KIN10-OE lines was much higher than that of the wild type (**Figure 1B**).

The KIN10-OE lines showed enhanced tolerance to carbon starvation induced by constant darkness for 7 days followed by a 7-day recovery, while the KIN10-RNAi lines appeared similar to the wild-type plants after this treatment (**Figures 1C,D**). For the nitrogen deficiency treatment, 7-day-old MS-grown seedlings were transferred to MS or MS-N solid medium for 5 days. The cotyledons of the wild-type plants and KIN10-RNAi lines were significantly yellowed as indicated by the reduced chlorophyll contents (**Figures 1E,F**). In contrast, the KIN10-OE lines were more resistant to nitrogen deficiency and had significantly higher chlorophyll levels compared to that of the wild type (**Figures 1E,F**). These findings suggest that overexpression of KIN10 can delay natural senescence and improves tolerance to carbon and nitrogen starvation.

#### The Enhanced Tolerance of the KIN10-OE Lines to Nutrient Starvation Is Dependent on a Functional Autophagy Pathway

Given that the KIN10-overexpression lines showed delayed leaf senescence and enhanced tolerance to nutrient starvation (**Figures 1**, **2**), and that the autophagy-deficient mutants had the opposite phenotype (Baena-González et al., 2007; Li and Vierstra, 2012), we used these plants to further assess the genetic connection between KIN10 and the autophagy pathway. The atg5-L and atg7-L mutants (atg5 and atg7 mutants in the Ler background) were crossed to KIN10-OE-1 (OE-1) to generate OE-1 atg5-L and OE-1 atg7-L lines. We then tested the tolerance of the 4-week-old wild-type, OE-1, OE-1 atg5-L, OE-1 atg7-L, atg5-L, and atg7-L plants to carbon starvation. Compared to the wild-type plants, the OE-1 plants showed enhanced tolerance and the atg5-L and atg7-L plants showed decreased tolerance to carbon starvation. Interestingly, the OE-1 atg5-L and OE-1 atg7-L lines displayed similar sensitivities to carbon starvation to that of the atg5-L and atg7-L mutants (**Figure 2A**). In addition, the enhanced resistance of the OE-1 line to nitrogen deficiency was attenuated by the loss-of-function of ATG5 and ATG7 (**Figure 2B**). The enhanced tolerance to starvation in the OE-1 line was further supported by the higher survival rates (**Figure 2C**) and higher relative chlorophyll contents (**Figure 2D**) in this line. Together, these results indicate that the enhanced tolerance to nutrient starvation in the KIN10-OE lines is dependent on a functional autophagy pathway. The evidence that the autophagy-associated phenotypes in the KIN10-OEs were primarily recovered by the autophagy deficient mutants, suggesting that autophagy acts downstream of KIN10 to affect

FIGURE 1 | Overexpression of KIN10 delays senescence and enhances tolerance to nutrient starvation in Arabidopsis. (A) Naturally induced senescence of the wild type (WT) and the KIN10-OE (OE-1 and OE-2) and KIN10-RNAi lines (RNAi-1 and RNAi-7). Photos were taken at 4 and 6 weeks after germination. (B) Relative chlorophyll contents in the leaves of 4- and 6-week-old WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants. The data are means ± SD (n = 3) calculated from three biological replicates. For each experiment, a population of 14 plants was recorded per genotype. ∗∗P < 0.01 by Student's t-test. (C) Phenotypes of the WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants after carbon starvation. Four-week-old WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants grown under normal growth conditions were transferred to constant darkness for 7 days and photos were taken after a 7-day recovery. (D) Survival rates of the WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants described in (C). The data are means ± SD (n = 3) calculated from three biological replicates. For each biological replicate, a population of 18 plants was recorded per genotype. ∗∗P < 0.01 by Student's t-test. (E) Phenotypes of the WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants after nitrogen starvation. One-week-old WT, OE-1, OE-2, RNAi-1, RNAi-7 seedlings grown on MS solid medium were transferred to MS or MS-N solid medium and photos were taken after 5 days of treatment. (F) Relative chlorophyll contents of the WT, OE-1, OE-2, RNAi-1, and RNAi-7 seedlings described in (E). The data are means ± SD (n = 3) calculated from three biological replicates. For each biological replicate, a population of 20 plants was recorded per genotype. ∗∗P < 0.01 by Student's t-test.

plant growth and stress responses. Given that KIN10 is a wellknown master regulator in energy signaling in Arabidopsis, we therefore proposed that it governs a cellular switch between plant growth and stress responses by modulating various downstream signaling pathways, including autophagy.

### The KIN10-OE Lines Are Tolerant to Drought and Submergence

The autophagy-defective mutants are hypersensitive to abiotic stresses such as drought and submergence (Liu et al., 2009; Chen et al., 2015). To further assess the role of KIN10 in

FIGURE 2 | The enhanced tolerance of the KIN10-OE lines to nutrient starvation is dependent on a functional autophagy pathway. (A) Phenotypes of the wild type (WT), OE-1, OE-1/atg5-L, atg5-L, OE-1/atg7-L, and atg7-L plants after carbon starvation. Four-week-old WT, OE1, OE-1/atg5-L, atg5-L, OE-1/atg7-L, and atg7-L plants grown under normal conditions were transferred to constant darkness for 7 days and photographs were taken after 7 days of dark treatment. (B) Phenotypes of the WT, OE-1, OE-1/atg5-L, atg5-L, OE-1/atg7-L, and atg7-L plants after nitrogen starvation. One-week-old WT, OE-1, OE-1/atg5-L, atg5-L, OE-1/atg7-L, and atg7-L seedlings grown on solid MS medium were transferred to MS or MS-N solid medium and photos were taken after 5 days of treatment. (C) Survival rate of the WT, OE-1, OE-1/atg5-L, atg5-L, OE-1/atg7-L, and atg7-L plants described in (A). The data are means ± SD (n = 3) calculated from three biological replicates. For each biological replicate, a population of 18 plants was recorded per genotype. ∗∗P < 0.01 by Student's t-test. (D) Relative chlorophyll levels in the leaves of the WT, OE-1, OE-1/atg5-L, atg5-L, OE-1/atg7-L, and atg7-L seedlings described in (B). The data are means ± SD (n = 3) calculated from three biological replicates. For each biological replicate, five technical replicates (each replicate was pooled with 20 seedlings) were measured per genotype. ∗∗P < 0.01 by Student's t-test. "a" indicates values that are significantly higher than that of the WT; "b" indicates values that are significantly lower than that of the WT.

autophagy-related stress responses, the wild-type (Ler), the KIN10-OE lines (OE-1 and OE-2), and the KIN10-RNAi lines (RNAi-1 and RNAi-7) were subjected to drought and submergence treatments. As shown in **Figure 3**, no significant differences were observed between the wild type and the KIN10-OE or KIN10-RNAi lines under normal growth conditions. However, after a 14-day drought treatment, the leaves of the wild type and the KIN10-RNAi lines turned yellow and wilted, while the leaves of the KIN10-OE lines remained green and turgid (**Figure 3A**). After a 4-day recovery by rehydration, the survival rates of the KIN10-OE lines were significantly higher than those of the wild type and the KIN10-RNAi lines (**Figure 3B**). In addition, the KIN10-OE lines were much more tolerant than the wild type and the KIN10-RNAi plants to a 6-day submergence in water (under light conditions) followed by a 6-day recovery (**Figure 3C**), which was supported by the improved survival rates of the KIN10-OE lines compared to the wild-type plants after submergence (**Figure 3D**).

re-watering for 4 days. The data are means ± SD (n = 3) calculated from three biological replicates. For each biological replicate, a population of 18 plants was recorded per genotype. ∗∗P < 0.01 by Student's t-test. (C) Phenotypes of the WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants after the submergence treatment. Four-week-old WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants were submerged for 6 days and photos were taken after a 6-day recovery. (D) Survival rates of the WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants described in (C) after the 6-day recovery. The data are means ± SD (n = 3) calculated from three biological replicates. For each biological replicate, a population of 18 plants was recorded per genotype. ∗∗P < 0.01 by Student's t-test.

### Overexpression of KIN10 Activates the Formation of Autophagosomes

To examine the potential involvement of KIN10 in regulating autophagosome formation, we first tested the abundance of ATG transcripts (ATG2, ATG5, ATG7, ATG8a, ATG10, and ATG18a) in the wild type, KIN10-OE lines (OE-1 and OE-2), and the KIN10-RNAi lines (RNAi-1 and RNAi-7). qRT-PCR analyses showed no significant changes in the expression levels of ATG7, ATG10, and ATG18a among the wild type, KIN10-OE lines, or the KIN10-RNAi lines, while the expression of ATG2, ATG5, and ATG8a was slightly upregulated in the KIN10-OE lines in comparison to the wild type (**Supplementary Figure S2**).

To further investigate the role of KIN10 in the induction of autophagy, we examined autophagosome formation in the wild type, KIN10-OE and KIN10-RNAi lines using green fluorescent protein (GFP)-tagged ATG8e (Xiao et al., 2010). Seven-day-old GFP-ATG8e (wild-type background), GFP-ATG8e/KIN10-OE and GFP-ATG8e/KIN10-RNAi seedlings were transferred to solid MS medium (MS) or MS-C under darkness for 6 h, and the GFP fluorescence of root cells was subsequently observed using confocal microscopy. As shown in **Figure 4**, under both MS and MS-C conditions, the numbers of GFP-ATG8e labeled punctate structures significantly increased in the KIN10-OE lines in comparison to the wild type and the KIN10-RNAi lines (**Figures 4A,B**). Upon nutrient starvation, the GFP-ATG8e fusion protein is degraded to release a free, relatively stable GFP, and the accumulation of GFP signals reflects the level of induction of autophagy (Li et al., 2014; Klionsky et al., 2016). As shown in **Figure 4C**, the degradation of the GFP-ATG8e fusion protein in the GFP-ATG8e/KIN10-OE line was faster than in the GFP-ATG8e or GFP-ATG8e/KIN10-RNAi line. Consistent with this, the ratio of free GFP to GFP-ATG8e in the GFP-ATG8e/KIN10-OE line was higher than that of in the GFP-ATG8e or GFP-ATG8e/KIN10-RNAi line (**Figure 4D**), suggesting that overexpression of KIN10 enhances autophagic activity.

### Overexpression of KIN10 Enhances the Phosphorylation of ATG1 Proteins

Given that AMPK phosphorylates ULK1 to activate autophagy in mammalian cells (Egan et al., 2011; Kim et al., 2011), we hypothesized that KIN10 may be involved in autophagy by

directly or indirectly phosphorylating ATG1. To confirm this, ATG1 phosphorylation was first tested using λ protein phosphatase and phosphatase inhibitor PhosSTOP in a yellow fluorescent protein (YFP)-tagged ATG1a transgenic plant (YFP-ATG1a) (Suttangkakul et al., 2011). As shown in **Figure 5A**, two species of YFP-ATG1a were detected using anti-GFP antibodies by western blot. Consistent with a previous study (Suttangkakul et al., 2011), the λ phosphatase treatment of total protein extracted from the YFP-ATG1a transgenic plant reduced the levels of the higher molecular weight species, while PhosSTOP blocked this shift (**Figure 5A**), To demonstrate the role of KIN10 in the regulation of ATG1, we crossed the YFP-ATG1a line to the OE-1 line to generate the YFP-ATG1a/KIN10-OE lines. Immunoblot analysis showed that the level of YFP-ATG1a fusion protein was significantly higher in the YFP-ATG1a KIN10-OE line than in the YFP-ATG1a line (**Figure 5B**). Upon carbon starvation, the phosphorylation status of ATG1 was enhanced in the YFP-ATG1a KIN10-OE line in comparison to the YFP-ATG1a line (**Figures 5B,C**).

To determine whether the accumulation of YFP-ATG1a was caused by the higher transcription of YFP-ATG1a in the YFP-ATG1a KIN10-OE line, we analyzed the total transcript level of ATG1a and YFP-ATG1a during carbon starvation by qRT-PCR. As shown in **Supplementary Figure S3**, The total transcript level of ATG1a was enhanced in the YFP-ATG1a KIN10-OE line but not in the YFP-ATG1a line, while the YFP-ATG1a transcript level did not change much in either line. Interestingly, the total expression of both ATG1a and YFP-ATG1a was slightly higher in the YFP-ATG1a KIN10-OE line than in the YFP-ATG1a line.

### DISCUSSION

As a mammalian ortholog of yeast ATG1, ULK1 is phosphorylated by AMPK to activate autophagy or phosphorylated by TOR to repress autophagy (Kim et al., 2011; Shang and Wang, 2011). It has been proposed that AMPK can activate autophagy by directly phosphorylating ULK1 or by suppressing the activity of mTORC1, which thereby inhibits ULK1 activity by phosphorylation (Kim et al., 2011). In this study, we present several lines of evidence to support the hypothesis that KIN10 is involved in the regulation of autophagy in Arabidopsis. First, the overexpression of KIN10 (KIN10-OE)

resulted in delayed leaf senescence and enhanced tolerance to nutrient deficiencies and abiotic stresses in Arabidopsis (**Figures 1**, **3**). Second, the enhanced tolerance to nutrient starvation in the KIN10-OE lines is dependent on a functional autophagy pathway (**Figure 2**). Third, the expression of ATGs and autophagosome formation and degradation were enhanced in the KIN10-OE lines (**Figure 4**). Last, the phosphorylation of ATG1a was enhanced in the KIN10-OE lines in comparison to that of the wild type (**Figure 5**). Taken together, these results demonstrate that KIN10 is a positive regulator of autophagy activation, possibly by enhancing the phosphorylation of ATG1, a mechanism that seems to be conserved in plants and animals.

KIN10 is an energy sensor in plants that has diverse functions in the regulation of plant metabolism, development, and stress responses (Polge and Thomas, 2007; Baena-González and Sheen, 2008; Jossier et al., 2009). KIN10 may be essential for maintaining the cell's energy balance during nutrient starvations (Baena-González et al., 2007). For example, the overexpression of KIN10 delays natural and nitrogen starvation-induced senescence, and the plants where KIN10 and KIN11 have been targeted by virus-induced gene silencing (KIN11 is a functionally redundant homolog of KIN10 in Arabidopsis) have an early senescence phenotype (Baena-González et al., 2007). Here, we investigated the autophagy-associated senescence phenotypes of the KIN10-OE and KIN10-RNAi lines and suggested that the phenotypes observed in the KIN10-OE lines were genetically linked to the autophagy pathway. Though we observed significant phenotypic differences in the KIN10-OE lines compared to the wild type in response to the treatments, we did not observe significant differences in the autophagy-associated phenotypes, gene expression and autophagosome formation in the KIN10-RNAi lines after the treatments. It is not feasible to use the kin10 kin11 virus-induced gene silencing lines in autophagy-related phenotypic analyses due to the severe growth inhibition of these silenced lines (Baena-González et al., 2007). It is still unknown whether KIN11 plays a redundant role in the regulation of autophagy, and generation of transgenic lines with double knockdown of KIN10 and KIN11 expression will be useful for future investigation of the functions of KIN10 and KIN11 in autophagy induction.

Autophagy plays an important role in the plant's response to various stress conditions, such as oxidative stress (Xiong et al., 2007a,b), nutrient deficiency (Doelling et al., 2002), hypoxia (Chen et al., 2015), and pathogen infection (Liu et al., 2005; Wang et al., 2011). The autophagy-defective (atg) mutants, such as atg2-1, atg5-1, atg7-1, and atg10-1, are frequently used to study the role of autophagy in stress responses in plants. In contrast to the situation in animal systems, little is known about the effects of constitutive activation of autophagy in plants. TOR has been suggested to be a negative regulator of plant autophagy (Liu and Bassham, 2010). To circumvent the embryo lethal phenotype of TOR loss-of-function mutant,

RNA interference (RNAi) was used to reduce TOR transcript levels in the RNAi-AtTOR plants, which show constitutive autophagy (Liu and Bassham, 2010). Given the fundamental roles of TOR in plant growth and metabolism (Xiong and Sheen, 2014), it is difficult to distinguish whether the phenotype of the RNAi-AtTOR line is caused by increased autophagy or is due to the reduced expression of TOR. In comparison, we suggested that overexpression of KIN10 enhances tolerance to nutrient deficiencies in Arabidopsis (**Figure 1**), and this enhanced tolerance was blocked by a deficiency in autophagy (**Figure 2**), which demonstrates that KIN10 improves tolerance to nutrient starvations by directly activating the autophagy pathway. Moreover, we showed that the KIN10-OE lines were more tolerant to drought and submergence (**Figure 3**), indicating that KIN10 is a potential candidate for genetic improvement of plant responses to nutrient deficiencies and water-related stresses. However, we cannot exclude the possibility that KIN10 may indirectly regulate autophagy by inhibiting TOR, since the mammalian AMPK has been reported to regulate autophagy by negative modulation of mTORC1 (Kimura et al., 2003). Further investigations of the coordinated functions of KIN10 and TOR will deepen our understanding of the upstream energy signals that regulate autophagy initiation in Arabidopsis.

The roles of AMPK in the regulation of autophagy have been extensively studied in animals (Gwinn et al., 2008; Lee et al., 2010; Kim et al., 2011; Alers et al., 2012; Mack et al., 2012), but the relationship between the plant AMPK and autophagy is still unknown. In our study, KIN10 overexpression activate autophagy pathway (**Figure 4**). Y2H assays suggested that KIN10 interacts with ATG1a and ATG13a in vitro (**Supplementary Figure S4**). However, we were unable to obtain further evidence of this interaction with BiFC and CoIP assays in planta (**Supplementary Figure S5**). We conclude that KIN10 positively regulates autophagy pathway through a possible unknown mechanism bypass ATG1/ATG13 protein complex. Alternatively, the activation of autophagy pathway by KIN10 overexpression may also be caused by inhibiting TOR activity, which is well-known to function as a negative regulator in autophagosome formation (Liu and Bassham, 2010). As suggested by Suttangkakul et al. (2011), the extent of ATG1a phosphorylation was highly regulated by the nutritional state through the action of upstream kinases and/or ATG1 autophosphorylation. Our results showed that, in response to starvation, overexpression of KIN10 enhanced the phosphorylation of ATG1a (**Figure 5**) which supports the idea that autophagy may function downstream of the KIN10 kinase by directly or indirectly targeting ATG1 proteins for phosphorylation. In conclusion, our findings demonstrated that KIN10 is a positive regulator of autophagy in Arabidopsis.

### AUTHOR CONTRIBUTIONS

SX and Q-FC designed the study. LC, Z-ZS, LH, F-NX, HQ, and L-JX carried out the experiments. SX, Q-FC, and LC analyzed the data. SX and LC wrote the manuscript.

#### FUNDING

This work was supported by the National Natural Science Foundation of China (Projects 31370298 and 31461143001), Program for New Century Excellent Talents in University (Project NCET-13-0614), the Foundation of Guangzhou Science and Technology Project (201504010021), and Sun Yat-sen University (Start-up fund to SX).

### ACKNOWLEDGMENTS

We thank the ABRC for providing the T-DNA seed pools, J. Sheen (Harvard Medical School) for the KIN10-OEs and KIN10-RNAi lines, F.Q. Li (South China Agricultural University) for the YFP-ATG1a line, and M. L. Chye (University of Hong Kong) for the GFP-ATG8e line.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.01201/ full#supplementary-material

FIGURE S1 | Molecular identification of the KIN10-OE and KIN10-RNAi transgenic plants. (A) qRT-PCR analysis of KIN10 transcript levels in 4-week-old wild type (WT), OE-1, OE-2, RNAi-1, and RNAi-7 plants. Transcript levels relative to the wild type were normalized to the levels of ACTIN2. The data are means ± SD (n = 3) calculated from three biological replicates. ∗∗P < 0.01 by Student's t-test. (B) Immunoblot analysis of KIN10 in 4-week-old WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants. Anti-KIN10 antibodies were used for immunoblotting. Coomassie blue-stained total proteins (Rubisco) are shown below the blot to indicate the amount of protein loaded per lane.

FIGURE S2 | Overexpression of KIN10 activates autophagy-related gene expression. Expression patterns of ATGs in the WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants. Total RNA was isolated from 4-week-old WT, OE-1, OE-2, RNAi-1, and RNAi-7 plants under normal growth conditions. Transcript levels relative to the WT were normalized to that of ACTIN2. The data are means ± SD (n = 3) calculated from three biological replicates. ∗∗P < 0.01 by Student's t-test.

FIGURE S3 | ATG1a transcript levels in the YFP-ATG1a and YFP-ATG1a/KIN10-OE plants in response to carbon starvation. Total RNA was isolated from 7-day-old YFP-ATG1a and YFP-ATG1a/KIN10-OE transgenic plants grown on MS medium followed by carbon starvation for 0, 3, 6, 12, and 24 h. Transcript levels relative to YFP-ATG1a at 0 h were normalized to that of ACTIN2. The data are means ± SD (n = 3) calculated from three biological replicates. ∗∗P < 0.01 by Student's t-test. Light gray bars indicate gene expression in the YFP-ATG1a, dark gray bars indicate gene expression in the YFP-ATG1a/KIN10-OE.

FIGURE S4 | Yeast two-hybrid assays showing the physical interactions of KIN10 with autophagy-related proteins (ATGs). (A) Y2H assay of the interaction between KIN10 and ATG proteins (ATG1a, ATG1b, ATG1c, ATG6, ATG8e, ATG9, PI3K, ATG13a, and ATG13b). ATG1a, ATG1b, ATG1c, ATG6, ATG8e, ATG9, PI3K, ATG13a, and ATG13b bait constructs were fused to the DNA-binding domain (BD), and full-length KIN10 was fused to the activation domain (AD). Vectors containing the AD and BD were co-expressed in yeast strain YH109. Protein interactions were determined by a growth assay in a medium lacking Trp, Leu, His, and Ade, with 30 mM 3-amino-1,2,4-triazole which was added to repress self-activation. The vector containing the AD or BD alone served as the negative control. (B) Y2H assay of the interaction between the functional domains of KIN10 and ATG1a and ATG13a. ATG1a and ATG13a bait constructs were fused to the BD, and prey constructs were fused to the AD. The vector containing the BD alone served as the negative control. Full-length KIN10 contained a protein kinase

domain (CD), a ubiquitin-associated domain (UBA) and a kinase associated domain 1 (KA1). Protein interaction was determined by a growth assay in a medium lacking Trp, Leu, His, and Ade (SD-Trp-Leu-His-Ade) supplemented with 30 mM 3-amino-1,2,4-triazole.

FIGURE S5 | In vivo assays showing no interaction of KIN10 with ATG1a and ATG13a. (A) CoIP assay of the association between KIN10 and ATG1a/ATG13a. FLAG-tagged ATG1a/ATG13a and HA-tagged KIN10 (KIN10-HA) was transiently

#### REFERENCES


expressed in protoplasts from wild-type Arabidopsis and immunoprecipitated by FLAG affinity agarose beads. (B) BiFC assay of KIN10 interaction with ATG1a and ATG13a in Arabidopsis protoplast cells. (C) BiFC assay of the interaction between functional domains of KIN10 and ATG1a/ATG13a in Arabidopsis protoplast cells. The split nYFP and cYFP fused to KIN10 and ATG1a/ATG13 were coexpressed in leaf protoplasts. nYFP/cYFP and ATG6-nYFP/TRAF1a-cYFP vectors were similarly co-expressed as negative and positive controls. Confocal images obtained from YFP, auto-fluorescent chlorophyll, and bright-field are shown. Bars = 20 µm.


promotes starvation-induced and age-dependent leaf senescence. Plant Cell 22, 1463–1482. doi: 10.1105/tpc.110.075333


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

Copyright © 2017 Chen, Su, Huang, Xia, Qi, Xie, Xiao and Chen. 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) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# TOR-Dependent and -Independent Pathways Regulate Autophagy in Arabidopsis thaliana

Yunting Pu1,2, Xinjuan Luo1,3 and Diane C. Bassham1,2,4 \*

<sup>1</sup> Department of Genetics, Development and Cell Biology, Iowa State University, Ames, IA, United States, <sup>2</sup> Interdepartmental Genetics Program, Iowa State University, Ames, IA, United States, <sup>3</sup> College of Life Sciences, Northwest A&F University, Yangling, China, <sup>4</sup> Plant Sciences Institute, Iowa State University, Ames, IA, United States

Autophagy is a critical process for recycling of cytoplasmic materials during environmental stress, senescence and cellular remodeling. It is upregulated under a wide range of abiotic stress conditions and is important for stress tolerance. Autophagy is repressed by the protein kinase target of rapamycin (TOR), which is activated in response to nutrients and in turn upregulates cell growth and translation and inhibits autophagy. Down-regulation of TOR in Arabidopsis thaliana leads to constitutive autophagy and to decreased growth, but the relationship to stress conditions is unclear. Here, we assess the extent to which TOR controls autophagy activation by abiotic stress. Overexpression of TOR inhibited autophagy activation by nutrient starvation, salt and osmotic stress, indicating that activation of autophagy under these conditions requires down-regulation of TOR activity. In contrast, TOR overexpression had no effect on autophagy induced by oxidative stress or ER stress, suggesting that activation of autophagy by these conditions is independent of TOR function. The plant hormone auxin has been shown previously to up-regulate TOR activity. To confirm the existence of two pathways for activation of autophagy, dependent on the stress conditions, auxin was added exogenously to activate TOR, and the effect on autophagy under different conditions was assessed. Consistent with the effect of TOR overexpression, the addition of the auxin NAA inhibited autophagy during nutrient deficiency, salt and osmotic stress, but not during oxidative or ER stress. NAA treatment was unable to block autophagy induced by a TOR inhibitor or by a mutation in the TOR complex component RAPTOR1B, indicating that auxin is upstream of TOR in the regulation of autophagy. We conclude that repression of auxin-regulated TOR activity is required for autophagy activation in response to a subset of abiotic stress conditions.

#### Keywords: autophagy, TOR signaling, stress responses, auxin, Arabidopsis

#### INTRODUCTION

Plants have evolved many response mechanisms to adapt to various growth conditions, including abiotic stresses. One such mechanism is autophagy, a major pathway for degradation and recycling of cytoplasmic materials in all eukaryotes (Liu and Bassham, 2012; Yang and Bassham, 2015). Autophagy is active at a low basal level even under normal conditions, and numerous human diseases are linked to autophagy defects, including cancer and various neurodegenerative diseases

#### Edited by:

Yule Liu, Tsinghua University, China

#### Reviewed by:

Shi Xiao, Sun Yat-sen University, China Liwen Jiang, The Chinese University of Hong Kong, Hong Kong

> \*Correspondence: Diane C. Bassham bassham@iastate.edu

#### Specialty section:

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

Received: 08 April 2017 Accepted: 26 June 2017 Published: 11 July 2017

#### Citation:

Pu Y, Luo X and Bassham DC (2017) TOR-Dependent and -Independent Pathways Regulate Autophagy in Arabidopsis thaliana. Front. Plant Sci. 8:1204. doi: 10.3389/fpls.2017.01204

such as Parkinson's, Huntington's, and Alzheimer's diseases (Cai et al., 2016; Davidson and Vander Heiden, 2017). In plants, autophagy functions in the response to both abiotic and biotic stress, and is induced during senescence and nutrient deficiency (Doelling et al., 2002; Hanaoka et al., 2002), salt and drought stresses (Liu et al., 2009), oxidative stress (Xiong et al., 2007b), endoplasmic reticulum (ER) stress (Liu et al., 2012), and pathogen infection (Liu et al., 2005).

When autophagy is activated, a double-membrane cup-shaped structure named a phagophore is formed. The phagophore expands to form a double-membrane vesicle called an autophagosome, while engulfing cellular components to be degraded. Autophagosomes are delivered to and fuse with lysosomes in mammalian cells or the vacuole in plant or yeast cells, where the cargo is degraded into small molecules by vacuolar hydrolases and recycled (Yang and Bassham, 2015). Studies in yeast have identified more than 30 autophagy-related (ATG) genes, many of which have also been found in plants (Tsukada and Ohsumi, 1993; Yang and Bassham, 2015). A key protein involved in autophagosome formation is ATG8, which can be used as a marker for autophagosomes when fused with a fluorescent protein (Yoshimoto et al., 2004; Contento et al., 2005). ATG8 is attached to the autophagosome membrane through a covalent bond to phosphatidylethanolamine (PE) via two ubiquitin-like conjugation systems that include the E1-like activating enzyme ATG7 (Ichimura et al., 2000). Knockout of ATG7 therefore prevents autophagosome formation, leading to plants being hypersensitive to both abiotic and biotic stress conditions (Doelling et al., 2002; Lenz et al., 2011; Zhou et al., 2013).

The target of rapamycin (TOR) complex is a key regulator of autophagy, and is composed of TOR itself and two binding partners, regulatory-associated protein of TOR (RAPTOR), and Lethal with Sec Thirteen 8 (LST8) (Yang et al., 2013; Dobrenel et al., 2016). TOR is a Ser/Thr protein kinase in the phosphatidylinositol-3-kinase (PI3K) – related kinase (PIKK) family (Noda and Ohsumi, 1998; Menand et al., 2002), whereas RAPTOR recruits substrates to the complex for phosphorylation by TOR (Hara et al., 2002), and LST8 stabilizes the complex (Yang et al., 2013). The TOR signaling pathway both positively regulates cell growth and metabolism and negatively regulates autophagy in yeast, mammals, and plants (Dobrenel et al., 2016). In Arabidopsis thaliana, a null mutation in TOR is embryo lethal (Menand et al., 2002), whereas decreased TOR expression due to RNA interference leads to autophagy induction (Liu and Bassham, 2010), and arrested plant growth and development (Deprost et al., 2007). Active-site TOR inhibitors (asTORis) that disrupt TOR activity by competition for ATP-binding also result in plant growth defects (Montane and Menand, 2013). Consistent with this, overexpression of TOR enhances growth and osmotic stress resistance (Deprost et al., 2007; Ren et al., 2011).

Two RAPTOR genes exist in Arabidopsis, RAPTOR1A and RAPTOR1B (Anderson et al., 2005; Deprost et al., 2005). A raptor1b null mutant has growth defects, including delayed leaf initiation and growth, late bolting and flowering, and short roots, while raptor1a knock out mutants have no major developmental phenotypes, possibly due to the higher expression of RAPTOR1B in most plant tissues (Anderson et al., 2005; Deprost et al., 2005). A raptor1a raptor1b double knockout mutant has minimal meristem growth, indicating that RAPTOR1A and RAPTOR1B might have some distinct functions, but is not embryo-lethal, and TOR must therefore retain some of its function in the absence of RAPTOR (Anderson et al., 2005). Two LST8 genes have also been identified in Arabidopsis, LST8-1 and LST8-2, although only LST8-1 appears to be expressed (Moreau et al., 2012). The null mutant lst8-1 has strong growth defects and impaired adaptation to long day conditions (Moreau et al., 2012). Mutation of lst8-1 or raptor1b, or disruption of TOR activity with asTORis, causes hypersensitivity to abscisic acid (ABA) and decreased ABA synthesis (Kravchenko et al., 2015), indicating that the TOR complex may also play a role in hormone signaling.

Target of rapamycin signals through phosphorylation of downstream substrates (Raught et al., 2001; Ren et al., 2011). Several TOR substrates have been identified in Arabidopsis, including the p70 ribosomal protein S6 kinase (S6K) (Mahfouz et al., 2006; Xiong and Sheen, 2012), the E2Fa transcription factor, which activates cell cycle genes (Xiong et al., 2013; Li et al., 2017), and TAP46, a regulatory subunit of protein phosphatase type 2A (PP2A), which was suggested to regulate plant growth and autophagy (Yorimitsu et al., 2009; Ahn et al., 2011). Arabidopsis has two S6K paralogs with 87% sequence identity, S6K1 and S6K2, both of which are phosphorylated by TOR. The activity of plant S6Ks increases in response to auxin and cytokinins (Turck et al., 2004).

Upstream regulation of TOR signaling in plants is still poorly understood. Auxin can enhance TOR activity to promote the translation reinitiation of mRNAs via S6K1, and deficiency in TOR signaling impaired auxin-mediated root gravitropism (Schepetilnikov et al., 2013). Auxin regulation is mediated by the small GTPase ROP2, which directly binds to and activates TOR (Li et al., 2017; Schepetilnikov et al., 2017). These studies indicate that auxin might regulate plant growth, development and stress responses through the TOR signaling pathway. In this study, we first confirm that the TOR complex is a negative regulator of autophagy in Arabidopsis, and demonstrate a role for RAPTOR1B in this regulation. We show that TOR regulates autophagy induced by nutrient starvation, salt or osmotic stress, but not oxidative or ER stress, indicating that TOR-dependent and -independent pathways for regulation of autophagy exist in plants. In addition, exogenous auxin has similar effects on stress-induced autophagy as TOR overexpression, suggesting a mechanism by which auxin interfaces with stress responses in plants through regulation of TOR activity.

### MATERIALS AND METHODS

#### Plant Materials and Growth Conditions

Arabidopsis thaliana seeds of WT (Col-0) or other indicated genotypes were sterilized with 33% (v/v) bleach and 0.1% (v/v) Triton X-100 (Sigma) for 20 min, followed by five washes of 5 min each with sterile water. Sterilized seeds were stored at 4◦C in darkness for at least 2 days to allow stratification before plating on solid <sup>1</sup>/<sup>2</sup> MS medium (2.22 g/L Murashige-Skoog vitamin and

salt mixture [Caisson Laboratory, MSP09], 1% [w/v] sucrose, 0.6% [w/v] Phytoblend agar [Caisson Laboratory], 2.4 mM 2-morphinolino-ethanesulfonic acid [MES, Sigma], pH 5.7). Seedlings were grown under long-day conditions (16 h light) at 22◦C for 7 days. Plants for transient expression in leaf protoplasts were grown in soil in a humidity-controlled growth chamber with 50% humidity at 20–23◦C under long-day conditions for 4–6 weeks. T-DNA insertion mutants used in this study are: raptor1a (SALK\_043920c), raptor1b (SALK\_078159) (Anderson et al., 2005), S7817 (SALK\_147817), G166 (GABI\_166C06), G548 (GABI\_548G07) (Deprost et al., 2007), and atg7 (GABI\_655B06) (Chung et al., 2010). Transgenic plants used in this study are: GFP-ATG8e (Xiong et al., 2007b), TOR-OE1 and TOR-OE2 (Ren et al., 2011).

#### Stress and Drug Treatments

For sucrose and nitrogen starvation, 7-day-old seedlings grown on solid <sup>1</sup>/<sup>2</sup> MS medium were transferred to solid <sup>1</sup>/<sup>2</sup> MS medium lacking sucrose or nitrogen for an additional 3 days (Doelling et al., 2002). Sucrose starvation plates were kept in the dark after transfer. For salt and mannitol treatment, 7-day-old seedlings grown on solid <sup>1</sup>/<sup>2</sup> MS medium were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium with 0.16 M NaCl or 0.35 M mannitol for 6–8 h. For oxidative and ER stress, 7-day-old seedlings grown on solid <sup>1</sup>/<sup>2</sup> MS medium were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium with 5 mM H2O<sup>2</sup> (Sigma) for 2–3 h, or with 2 mM dithiothreitol (DTT, Fisher) or 5 µg/mL tunicamycin (Sigma) for 6–8 h. For AZD8055 treatment, 7-day-old seedlings grown on solid <sup>1</sup>/<sup>2</sup> MS medium were transferred to solid <sup>1</sup>/<sup>2</sup> MS medium with 2 µM AZD8055 (LC Laboratories) for 1 day, or liquid <sup>1</sup>/<sup>2</sup> MS medium with 1 µM AZD8055 for 2–3 h.

For auxin treatment, 7-day-old seedlings were transferred to solid <sup>1</sup>/<sup>2</sup> MS medium supplemented with 20 nM 1-naphthaleneacetic acid (NAA, Sigma-Aldrich, N0640) with or without starvation for an additional 3 days, or in liquid <sup>1</sup>/<sup>2</sup> MS medium with 20 nM NAA for 6–8 h with or without stress treatments as described above. For BTH treatment, 7-day-old seedlings were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium supplemented with 100 µM acibenzolar-S-methyl (BTH, Sigma-Aldrich, 32820) with or without 20 nM NAA for 8 h.

For concanamycin A treatment, 7-day-old GFP-ATG8e seedlings were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium with DMSO or 1 µM concanamycin A (Sigma) with or without other stress or drug treatments for 6–8 h.

### Autophagy Detection by Fluorescence Microscopy

Arabidopsis seedling roots were stained with monodansylcadaverine (MDC) as described previously (Contento et al., 2005). MDC-stained seedlings were observed with a Zeiss Axio Imager.A2 upright microscope (Zeiss) equipped with Zeiss Axiocam BW/color digital cameras using a DAPI-specific filter at the Iowa State University Microscopy and Nanoimaging Facility. GFP-ATG8e transgenic seedlings were observed and photographed with the same microscope system with a GFP-specific filter. Cells within the root elongation zone were photographed and the number of autophagosomes in each image was counted and averaged from at least 10 images per sample. Confocal microscopy images of autophagosomes in root cells and leaf protoplasts were taken using a Leica SP5 × MP confocal/multiphoton microscope system (Leica) with a 63x/1.4 oil immersion objective at the Iowa State University Roy J. Carver High Resolution Microscopy Facility (Pu and Bassham, 2016).

#### Transient Expression in Protoplasts

GFP-ATG8e was transiently expressed in Arabidopsis leaf protoplasts as previously described (Sheen, 2002; Liu et al., 2012). 25–30 µg of GFP-ATG8e plasmid DNA was introduced into protoplasts using 40% (w/v) polyethylene glycol (PEG, Sigma-Aldrich). Protoplasts were washed and incubated in W5 solution (154 mM NaCl, 125 mM CaCl2, 5 mM KCl, 2 mM MES, pH 5.7). For starvation treatment, protoplasts were incubated in W5 solution without sucrose or with 0.5% (w/v) sucrose as control at room temperature in darkness for 2 days in total. For other stress treatments, protoplasts were incubated in W5 solution with treatments as described in the Stress and Auxin Treatment section. Protoplasts were observed by fluorescence microscopy (Nikon Eclipse E200) using a FITC filter, and protoplasts with more than three visible autophagosomes were counted as active for autophagy (Yang et al., 2016). A total of 100 protoplasts were observed per genotype for each condition, and the percentage of protoplasts with induced autophagy was calculated and averaged from three independent experimental replicates.

### Generation of RAPTOR1B Construct

The RAPTOR1B cDNA sequence was divided into two fragments, and each fragment was amplified from Col-0 cDNA using CloneAmp HiFi PCR Premix (Takara). The 5<sup>0</sup> fragment of RAPTOR1B was amplified with forward primer 5<sup>0</sup> - CACCGAGCTCGAATTCATGGCATTAGGAGACTTAATGGT GTCTC-3<sup>0</sup> (inserted SacI restriction site underlined), and reverse primer 5<sup>0</sup> -GTCAAACCCAATATCAAGCAAGGTACCCA-3<sup>0</sup> , digested with SacI and KpnI (within the RAPTOR1B cDNA sequence), and ligated into the pPZP212 binary vector (Hajdukiewicz et al., 1994; Li et al., 2009), which has a 35S promoter sequence at the 5<sup>0</sup> end and a MYC tag sequence at the 3<sup>0</sup> end of the insert. The 3 0 fragment was amplified with forward primer 5<sup>0</sup> -TG GGTACCTTGCTTGATATTGGGTTTGAC-3<sup>0</sup> and reverse primer 5<sup>0</sup> -CACCGTCGACTCTTGCTTGCGAGTTGTCGTGGG TG-3<sup>0</sup> (inserted SalI restriction site underlined), digested with KpnI and SalI, and ligated into the pPZP212 vector containing the 5<sup>0</sup> fragment to complete the full sequence. The entire construct was confirmed by sequencing.

#### Accession Numbers

Sequence data from this article can be found in the Arabidopsis Genome Initiative under the following accession numbers: TOR, AT1G50030; RAPTOR1A, AT5G01770, RAPTOR1B, AT3G08850; ATG8e, AT2G45170; ATG7, AT5G45900.

### RESULTS

fpls-08-01204 July 8, 2017 Time: 15:17 # 4

## Inhibition of TOR Signaling Leads to Constitutive Autophagy

We have shown previously that decreased TOR expression via RNA interference induces autophagy in Arabidopsis, suggesting that TOR is a negative regulator of autophagy in plants (Liu and Bassham, 2010). To confirm that autophagy is induced by inhibition of TOR kinase activity (Montane and Menand, 2013), we examined autophagy activity after application of the asTORis AZD8055 (Dong et al., 2015). WT and atg7 seedlings, a previously characterized knockout mutant that is unable to form autophagosomes (Doelling et al., 2002), were grown under standard conditions for 7 days, followed by 1 µM AZD8055 treatment in liquid <sup>1</sup>/<sup>2</sup> MS medium for 2–3 h. Roots of seedlings were stained with monodansylcadaverine (MDC), an acidotropic dye that can stain autophagosomes (Biederbick et al., 1995; Contento et al., 2005), and examined by fluorescence microscopy. Autophagosomes appear as rapidly moving fluorescent puncta, and the number of visible puncta in each image were counted for quantification. As expected, compared to the basal level of autophagy in the control, inhibition of TOR activity by AZD8055 led to a significant increase in the number of autophagosomes (**Figure 1A**), while no autophagosomes were detected in the atg7 mutant. This confirmed that TOR negatively regulates autophagy in Arabidopsis, and that the kinase activity of TOR is critical for this regulation.

Previous studies have shown that down-regulation of TOR or its binding partners RAPTOR and LST8 leads to defects in plant growth and development (Anderson et al., 2005; Deprost et al., 2007; Moreau et al., 2012; Montane and Menand, 2013), suggesting that RAPTOR and LST8 are critical for TOR-regulated plant growth. To test whether inhibition of TOR complex activity by disruption of RAPTOR also induces autophagy, WT, raptor1a and raptor1b knockout mutant seedlings (Anderson et al., 2005) were grown on <sup>1</sup>/<sup>2</sup> MS medium with sucrose for a week, and autophagy in root cells was examined by MDC staining followed by fluorescence microscopy (**Figures 1B,C**). Compared to the basal level of autophagy in WT seedlings, the number of autophagosomes in the raptor1a mutant appeared slightly higher, but this difference was not statistically significant, possibly due to the variability between seedlings. The raptor1b mutant had a significantly higher number of autophagosomes, suggesting that the raptor1b mutant has constitutive autophagy, and that RAPTOR1A and RAPTOR1B may not function equally in autophagy regulation.

To confirm that the raptor1b mutant has increased basal autophagy under standard conditions, the autophagosome marker GFP-ATG8e was expressed transiently in WT, raptor1a and raptor1b leaf protoplasts (**Figure 1D**). The GFP-ATG8e fusion protein has been used extensively as a specific marker of autophagosomes and autophagic bodies (Yoshimoto et al., 2004; Contento et al., 2005; Pu and Bassham, 2016), and active autophagy is defined as more than three visible autophagosomes in a protoplast (Yang et al., 2016). WT protoplasts maintain a basal level of autophagy with a low percentage with active autophagy. Consistent with the MDC staining results, the percentage of raptor1b protoplasts with active autophagy was significantly higher than that of WT protoplasts (**Figure 1E**). However, raptor1a also had a significantly higher percentage of active autophagy in leaf protoplasts, although significantly lower than raptor1b. RAPTOR1A may therefore be more important for autophagy regulation in leaves than in roots.

To confirm that the constitutive autophagy in the raptor1b mutant is specifically due to the mutation in RAPTOR1B, the RAPTOR1B cDNA was transiently expressed under a 35S promoter together with GFP-ATG8e in raptor1b knock out mutant leaf protoplasts. Autophagy was assessed as described above using fluorescence microscopy (**Figure 1F**). Note that the number of autophagosomes in the raptor1b mutant is variable between experiments, depending most likely on the growth conditions and age of the plants, room temperature, etc. The percentage of protoplasts with active autophagy in the raptor1b mutant expressing the RAPTOR1B cDNA was substantially lower than for the mutant protoplasts alone, and was not significantly different from WT. This indicates that the increased basal autophagy observed in the raptor1b mutant was suppressed by expression of the RAPTOR1B cDNA, confirming that the constitutive autophagy phenotype is indeed due to the disruption of RAPTOR1B.

#### Overexpression of TOR Blocks Autophagy upon Starvation, Salt and Drought Stress

Previous studies and our data have shown that the TOR complex negatively regulates autophagy in Arabidopsis (Liu and Bassham, 2010) (**Figure 1**), but the conditions under which TOR is important in plants are unknown. In other organisms TOR is well-described as regulating autophagy in response to nutrients (Dobrenel et al., 2016), with a decrease in TOR activity during nutrient deficiency leading to activation of autophagy. We therefore hypothesized that overexpression of TOR might prevent activation of autophagy by nutrient deficiency, and that autophagy induction by other stresses might be TOR-independent. To test this hypothesis, we obtained several previously characterized Arabidopsis lines with T-DNA insertions in the TOR upstream region (S7817, G166, and G548) and two transgenic lines with TOR expressed from a 35S promotor (TOR-OE1 and TOR-OE2). All lines have overexpression of TOR and enhanced growth (Deprost et al., 2007; Ren et al., 2011), with the exception of S7817, which has decreased TOR expression in leaves and overexpression of TOR in roots (Deprost et al., 2007). Seeds of WT and the five TOR overexpression lines were germinated and grown on solid <sup>1</sup>/<sup>2</sup> MS medium plus sucrose for 1 week, followed by transfer to solid <sup>1</sup>/<sup>2</sup> MS medium plus or minus nitrogen in the light, or minus sucrose in the dark for an additional 3 days. Autophagosomes were detected by MDC staining followed by fluorescence microscopy (**Figures 2A,B**). Representative images of one of the TOR overexpression lines are shown in **Figure 2A**. Quantification of autophagosomes indicated that both WT and the TOR overexpression lines had a low basal level of autophagy under

FIGURE 1 | Inhibition of TOR or RAPTOR leads to constitutive autophagy. (A) The TOR inhibitor AZD8055 induces autophagy. 7-day-old WT (Col-0) and atg7 mutant seedlings were treated with DMSO or AZD8055 (AZD) for 2–3 h, stained with MDC and then observed and imaged by fluorescence microscopy. The number of puncta in each image was counted and averaged from at least 10 images per genotype for each condition. (B,C) Autophagy is induced in raptor1b mutant root cells under standard growth conditions. (B) 7-day-old WT, raptor1a and raptor1b knockout mutant seedlings were stained with MDC and observed by fluorescence microscopy. The number of puncta in each image was quantified as in (A). (C) Representative confocal images of MDC-stained WT, raptor1a and raptor1b mutant seedlings. MDC-stained autophagosomes appear as white puncta within cells as indicated by white arrows. Scale bar = 20 µm. (D,E) Leaf protoplasts of raptor1a and raptor1b mutants have constitutive autophagy. (D) Transient expression of a GFP-ATG8e fusion protein in leaf protoplasts of WT and RAPTOR mutants, observed by confocal microscopy. GFP-tagged autophagosomes appear as green puncta within leaf protoplasts in the left column as indicated by white arrows. The middle and right columns show DIC and merged images respectively. Scale bar = 10 µm. (E) Quantification of D. Protoplasts were observed using epifluorescence microscopy. The percentage of protoplasts with more than three visible GFP-tagged autophagosomes was calculated, with 100 protoplasts observed per genotype for each condition. (F) Expression of the RAPTOR1B cDNA complements the raptor1b constitutive autophagy phenotype. A GFP-ATG8e fusion protein was transiently expressed in raptor1b mutant leaf protoplasts with or without full-length RAPTOR1B, expressed from a 35S constitutive promoter, or in WT protoplasts as a control. Protoplasts were observed using epifluorescence microscopy. The percentage of protoplasts with more than three visible GFP-tagged autophagosomes was quantified as in (E). For all graphs, error bars indicate means ± standard error (SE) from three independent replicates. Asterisks or different letters indicate statistically significant differences (P < 0.05) using Student's t-test compared with WT under control conditions.

FIGURE 2 | Overexpression of TOR blocks autophagy induced by nutrient starvation, salt and osmotic stresses. (A) Representative confocal images of MDC-stained WT and TOR-OE2 seedlings after the indicated stress treatment. For nutrient starvation, 7-day-old seedlings of WT and TOR-OE2 transgenic lines were transferred to solid <sup>1</sup>/<sup>2</sup> MS medium for an additional 3 days with or without nitrogen in the light, or without sucrose in the dark. For salt and osmotic stress, 7-day-old WT and TOR-OE2 seedlings were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium plus or minus 0.16 M NaCl or 0.35 M mannitol for 6–8 h. Scale bar = 20 µm. (B,C) Quantification of autophagosome number in WT and TOR overexpression lines after sucrose or nitrogen starvation (B), salt, or osmotic stress (C), treated as in (A). MDC-stained autophagosomes were observed by fluorescence microscopy and photographed. The average number of autophagosomes was calculated from 10 images per genotype for each condition. (D,E) TOR overexpression lines fail to activate autophagy under sucrose starvation (D), salt or osmotic stress (E) in leaf protoplasts. A GFP-ATG8e fusion protein was transiently expressed in leaf protoplasts of WT and TOR overexpression lines. Protoplasts were incubated in the dark plus or minus 0.5% (w/v) sucrose for 2 days (D), or plus or minus 0.16 M NaCl or 0.35 M mannitol for 1 day (E). Protoplasts were observed using fluorescence microscopy. The percentage of protoplasts with more than three visible GFP-tagged autophagosomes was calculated from 100 protoplasts observed per genotype for each condition. For all graphs, error bars indicate means ± SE from three independent replicates. Asterisks indicate statistically significant differences (P < 0.05) using Student's t-test compared with WT under control conditions.

control conditions. The average number of autophagosomes in WT seedlings after sucrose or nitrogen starvation was significantly higher than in control conditions, whereas the TOR overexpression lines had no significant activation of autophagy. This indicates that overexpression of TOR can repress autophagy induced by nutrient starvation, suggesting that repression of TOR activity is required for activation of autophagy in response to nutrient depletion.

While previous studies have shown that TOR is involved in nutrient sensing (Dobrenel et al., 2016), the extent to which TOR regulates stress responses other than nutrient deficiency is not known, although a link to osmotic stress resistance in Arabidopsis has been suggested (Mahfouz et al., 2006; Deprost et al., 2007). Autophagy is activated in Arabidopsis by salt and osmotic stresses (Liu et al., 2009). Therefore, we also tested whether overexpression of TOR affects autophagy induced by salt or osmotic stress (**Figures 2A,C**). WT and the TOR overexpression lines were germinated and grown on solid <sup>1</sup>/<sup>2</sup> MS medium for 1 week, and then transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium containing 0.16 M NaCl or 0.35 M mannitol for 6–8 h. Autophagy in seedling roots was assayed by MDC staining followed by fluorescence microscopy (**Figure 2C**). Autophagy in

salt or mannitol treated WT seedlings was significantly higher than the basal level of autophagy seen under control conditions. As for nutrient deficiency, autophagy was not induced in TOR overexpression lines under salt or osmotic stress, indicating that TOR can also repress autophagy induced by these stresses.

To confirm these results, we measured autophagy by transient expression of GFP-ATG8e in leaf protoplasts from WT and TOR overexpressing plants under sucrose starvation, salt and osmotic stresses (**Figures 2D,E**). As the protoplast incubation buffer contains nitrogen, it was not possible to test nitrogen deficiency using our standard protocol. After transformation with GFP-ATG8e constructs, protoplasts were incubated with or without sucrose for 2 days (**Figure 2D**), or plus or minus 0.16 M NaCl or 0.35 M mannitol for 1 day (**Figure 2E**), after which autophagy was observed using fluorescence microscopy. The percentage of protoplasts with active autophagy was calculated, with 100 protoplasts observed per genotype for each condition. WT and TOR overexpression lines had a low level of autophagy under control conditions, except for the S7817 line which had constitutive activation of autophagy. In this line, TOR expression is decreased in leaves, potentially explaining this observation (Deprost et al., 2007). While WT protoplasts had a significantly higher level of autophagy under sucrose starvation, salt and osmotic stresses, autophagy in the TOR overexpression lines, with the exception of S7817, remained at a low basal level indistinguishable from that in control conditions. We conclude that TOR is a regulator of autophagy in response to salt and osmotic stress in addition to nutrient deficiency.

#### Overexpression of TOR Has No Effect on Oxidative Stress- or ER Stress-Induced Autophagy

Autophagy is also induced by oxidative stress and ER stress in plants (Xiong et al., 2007b; Liu et al., 2012; Yang et al., 2016). Oxidative stress is triggered when cells accumulate excessive reactive oxygen species (ROS), and oxidized proteins and lipids are degraded through autophagy (Xiong et al., 2007a,b). ER stress is generated when unfolded or misfolded proteins exceed the capacity of protein folding or degradation systems, causing accumulation of proteins in the ER (Howell, 2013). It can be triggered by heat stress, or experimentally by chemicals such as dithiothreitol (DTT) or tunicamycin (Howell, 2013). To determine whether TOR regulates autophagy upon oxidative or ER stress, 7-day-old WT and TOR overexpression lines were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium plus or minus 5 mM H2O<sup>2</sup> for 2–3 h to cause oxidative stress, or plus 2 mM DTT or 5 µg/mL tunicamycin for 6–8 h to cause ER stress. Autophagy in seedling roots was detected by MDC staining followed by fluorescence microscopy (**Figures 3A–C**). Representative images of one of the TOR overexpression lines are shown in **Figure 3A**. WT and TOR overexpression lines had a low level of autophagy under control conditions, and WT seedlings had significantly higher autophagy induction after oxidative or ER stress treatment. Unlike nutrient, salt or osmotic stresses, TOR overexpression had no effect on autophagy induction, as overexpression lines remained able to activate autophagy under these stresses, suggesting that autophagy is activated via a pathway that does not require inhibition of TOR activity.

To confirm that autophagy remains induced in TOR overexpression lines under oxidative and ER stress, GFP-ATG8e was transiently expressed in leaf protoplasts of WT and TOR overexpression lines. Protoplasts were incubated with or without 5 mM H2O2, 2 mM DTT, or 5 µg/mL tunicamycin for 1 day, and observed using fluorescence microscopy (**Figures 3D,E**). WT protoplasts had a significantly higher level of autophagy after oxidative or ER stress treatment. In accordance with the MDC staining results, TOR overexpression lines also had a significantly higher percentage of protoplasts with active autophagy after oxidative or ER stress treatment, with no significant difference compared to WT under the same stress conditions. This demonstrates that overexpression of TOR is unable to repress autophagy induced by oxidative or ER stress, suggesting that oxidative- or ER stress-induced autophagy might be regulated through a TOR-independent pathway.

The effects of oxidative stress on autophagy can be difficult to interpret, as autophagy is also triggered by signaling ROS produced by NADPH oxidase (Liu et al., 2009). Salicylic acid (SA) has also been shown to enhance ROS signaling and induce autophagy in plants (Yoshimoto et al., 2009); we therefore tested whether SA-induced autophagy is dependent on TOR. Seven-day-old WT and TOR overexpression lines were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium with 100 µM benzo-(1,2,3)-thiadiazole-7-carbothioic acid S-methyl ester (BTH), an SA agonist, or 80% ethanol as control for 8 h. Autophagy in seedling roots was detected by MDC staining followed by fluorescence microscopy (Supplementary Figure 1A). WT and TOR overexpression lines had a low level of autophagy under control conditions, while both WT and TOR overexpression lines had significantly increased autophagy induction upon BTH treatment, suggesting that SA-induced autophagy, as for H2O2-induced autophagy, is not TOR dependent.

### Auxin Represses Stress-Induced Autophagy through TOR

Target of rapamycin activity in Arabidopsis can be enhanced by exogenous addition of the auxin 1-naphthaleneacetic acid (NAA) (Schepetilnikov et al., 2013), indicating that auxin might regulate plant growth and development via the TOR signaling pathway. We hypothesized that auxin might repress stress-induced autophagy in plants through the TOR pathway. To confirm the existence of TOR-dependent and -independent pathways for activation of autophagy, NAA was added exogenously to GFP-ATG8e transgenic plants to activate TOR, and the effect on autophagy under different conditions was assessed (**Figures 4A–D** and Supplementary Figure 1B). For nutrient deficiency, 7-day-old GFP-ATG8e transgenic seedlings were transferred to solid <sup>1</sup>/<sup>2</sup> MS medium with or without sucrose or nitrogen and plus 20 nM NAA or DMSO for an additional 3 days. For salt, osmotic and ER stress, and BTH treatment, 7-day-old seedlings were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium plus 0.16 M NaCl, 0.35 M mannitol, 2 mM DTT or 5 µg/mL

autophagosome number in WT and TOR overexpression lines under oxidative stress (B) or ER stress (C) treated as in A. MDC-stained autophagosomes were observed by fluorescence microscopy and photographed. The average number of autophagosomes was calculated from 10 images per genotype for each condition. (D,E) Autophagy was induced in leaf protoplasts of TOR overexpression lines under oxidative stress (D) or ER stress (E). A GFP-ATG8e fusion protein was transiently expressed in leaf protoplasts of WT and TOR overexpression lines. Protoplasts were incubated in the dark plus or minus 5 mM H2O<sup>2</sup> for 2–3 h (D), or 2 mM DTT or 5 µg/mL tunicamycin (TM) for 6–8 h (E). Protoplasts were observed using fluorescence microscopy. The percentage of protoplasts with more than three visible GFP-tagged autophagosomes was calculated from 100 protoplasts observed per genotype for each condition. For all graphs, error bars indicate means ± SE from three independent replicates. Asterisks indicate statistically significant differences (P < 0.05) using Student's t-test compared with WT under control conditions.

tunicamycin, or 100 µM BTH and plus 20 nM NAA or DMSO for 6–8 h. For oxidative stress, 7-day-old seedlings were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium plus 20 nM NAA or DMSO for 6–8 h, with 5 mM H2O<sup>2</sup> added only during the last 2–3 h to avoid cell death. To more clearly observe GFP-ATG8e-labeled autophagic bodies in the vacuoles by confocal microscopy, 1 µM concanamycin A was added to block degradation of autophagic bodies prior to imaging of the vacuoles (Dröse et al., 1993; Liu and Bassham, 2010) (**Figure 4A**). In control conditions, root cells had few autophagic bodies within the vacuole, whereas all stresses tested led to accumulation of large numbers of autophagic bodies. In the presence of auxin, autophagic body accumulation was inhibited in nutrient deficiency, salt and osmotic stress, but accumulation was still observed in oxidative and ER stress and upon BTH treatment. These results also indicate that NAA reduces the number of autophagosomes observed by blocking autophagosome formation, rather than by accelerating autophagosome degradation. Autophagy was quantified by counting the number of autophagosomes under each condition, averaged from 10 images per genotype for each condition (**Figures 4B–D** and Supplementary Figure 1B). Compared to the

FIGURE 4 | Auxin represses autophagy induced by nutrient starvation, salt and osmotic stress through the TOR signaling pathway. (A–E) NAA represses autophagy induced by nutrient starvation, salt and osmotic stresses. (A) Representative confocal images of GFP-ATG8e transgenic seedlings after NAA and stress treatments. Concanamycin A was included under all conditions to allow accumulation of autophagic bodies inside the vacuole, facilitating visualization. For nutrient starvation, 7-day-old GFP-ATG8e seedlings were transferred to solid <sup>1</sup>/<sup>2</sup> MS medium plus DMSO or 20 nM NAA for an additional 3 days with or without nitrogen in the light, or without sucrose in the dark. Treated seedlings were then transferred to liquid medium under the same conditions plus 1 µM concanamycin A for an additional 6–8 h. For all other stresses, 7-day-old GFP-ATG8e seedlings were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium with 1 µM concanamycin A and DMSO or 20 nM NAA for 6–8 h, together with 0.16 M NaCl, 0.35 M mannitol, 2 mM DTT, or 5 µg/mL tunicamycin (TM) for 6–8 h, or 5 mM H2O<sup>2</sup> or 1 µM AZD8055 during the last 2–3 h of DMSO or NAA treatment. Scale bar = 20 µm. (B–E) Quantification of autophagic body number in GFP-ATG8e transgenic seedlings under sucrose or nitrogen starvation (B), salt, osmotic stress or oxidative stress (C), ER stress (D), or AZD8055 treatment (E), treated as in (A). GFP-tagged autophagosomes in each condition were observed by fluorescence microscopy and photographed. The number of autophagosomes was counted and averaged from 10 images per genotype for each condition. (F,G) Auxin cannot repress the constitutive autophagy seen in a raptor1b mutant. (F) Representative confocal images of MDC-stained WT and raptor1b mutant seedling roots. 7-day-old WT and raptor1b seedlings were treated in liquid <sup>1</sup>/<sup>2</sup> MS medium with DMSO or 20 nM NAA for 6–8 h. Scale bar = 20 µm. (G) Quantification of (F). The average number of autophagosomes was calculated from 10 images per genotype for each condition. For all graphs, error bars indicate means ± SE from three independent replicates. Asterisks indicate statistically significant differences (P < 0.05) using Student's t-test compared with WT under control conditions.

basal level of autophagy under control conditions, autophagy was significantly higher after stress treatments. In the presence of NAA, autophagy was still significantly induced by oxidative and ER stress conditions and in the presence of BTH, but no significant difference compared to control conditions was observed under nutrient starvation, salt and osmotic stresses. This suggests that NAA represses autophagy induced by sucrose and nitrogen starvation, salt and osmotic stresses, but not oxidative stress or ER stress, consistent with the results from overexpression of TOR.

To further confirm that addition of auxin represses stress-induced autophagy through activation of TOR, we examined whether auxin can inhibit the constitutive autophagy seen upon disruption of the TOR signaling pathway by chemical inhibition or genetic mutation (**Figures 4E–G**). To inhibit TOR kinase activity, 7-day-old GFP-ATG8e seedlings were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium with or without 20 nM NAA for 6–8 h, with DMSO or 1 µM AZD8055 added during the last 2–3 h of treatment. GFP-labeled autophagic bodies in roots after concanamycin A treatment were examined using confocal microscopy (**Figure 4A**). AZD8055 as expected led to a high accumulation of autophagic bodies in the vacuole, and NAA had no effect on this accumulation, suggesting that NAA acts upstream of TOR in the autophagy pathway. The extent of

autophagy was quantified by counting root autophagosomes, and AZD8055 caused accumulation of autophagosomes both in the presence and absence of NAA, with no significant difference in autophagy induction (**Figure 4E**).

As an alternative approach, the effect of NAA upon inhibition of TOR complex function via knockout of RAPTOR1B was tested. Seven-day-old WT and raptor1b seedlings were transferred to liquid <sup>1</sup>/<sup>2</sup> MS medium with or without 20 nM NAA for 6–8 h, followed by MDC staining and autophagy detection by fluorescence microscopy (**Figures 4F,G**). NAA had no significant effect on the constitutive autophagy seen in the raptor1b mutant. Taken together, these results suggest that auxin acts upstream of TOR in the regulation of autophagy.

#### DISCUSSION

The TOR signaling pathway is a critical pathway for balancing cell growth and survival (Dobrenel et al., 2016). TOR was suggested to function as a complex with RAPTOR and LST8 based on studies in yeast and mammals (González and Hall, 2017), and previous studies in plants (Anderson et al., 2005; Deprost et al., 2005; Moreau et al., 2012). A knock out mutation in TOR is embryo-lethal in Arabidopsis (Menand et al., 2002), and down-regulation of TOR via RNA-interference arrests plant growth and induces autophagy. This suggests that TOR is a positive regulator of growth and development, and a negative regulator of autophagy in plants (Deprost et al., 2007; Liu and Bassham, 2010). To confirm that the TOR complex negatively regulates autophagy in Arabidopsis, we used a TOR inhibitor, AZD8055 (Chresta et al., 2010; Montane and Menand, 2013; Dong et al., 2015), which led to a significant induction of autophagy (**Figure 1A**). We also disrupted the TOR signaling pathway via a knockout mutant in RAPTOR, a binding partner of TOR. RAPTOR1B is the most highly expressed isoform of RAPTOR in Arabidopsis (Deprost et al., 2005), and raptor1b has a much more severe growth defect than raptor1a (Anderson et al., 2005). The raptor1b knockout line exhibited constitutive autophagy in both roots and leaf protoplasts, whereas a raptor1a mutation had only minor effects on autophagy, suggesting that RAPTOR1B is the primary RAPTOR isoform for repression of autophagy under our conditions.

Autophagy is induced by numerous stresses, including nutrient deficiency, salt, drought, oxidative and ER stresses (Doelling et al., 2002; Hanaoka et al., 2002; Xiong et al., 2007b; Liu et al., 2009, 2012). TOR has been well-characterized as regulating autophagy in response to nutrients in yeast and mammals, and down-regulation of TOR leads to growth defects and autophagy induction (Dobrenel et al., 2016). Therefore, we hypothesized that nutrient deficiency induces autophagy through the TOR signaling pathway in plants. As expected, overexpression of TOR repressed autophagy upon sucrose or nitrogen starvation, suggesting that TOR regulates nutrient deficiency-induced autophagy. Many upstream regulators of TOR have been identified in yeast and mammals, although many of them are not conserved in plants. One upstream kinase that is conserved throughout eukaryotes, named AMPK in mammals and Snf1 in yeast, senses energy status and activates autophagy in response to low energy (Hulsmans et al., 2016). A homolog of AMPK and Snf1 in plants, SnRK1, has been identified, and is also activated under stress conditions (Dobrenel et al., 2016; Nukarinen et al., 2016). SnRK1 phosphorylates RAPTOR1B, potentially decreasing TOR activity, although whether this affects autophagy is as yet unknown.

Salt and drought are two major environmental stresses encountered by plants; both lead to osmotic stress, while salt stress also leads to ionic stress. Surprisingly, overexpression of TOR and activation of TOR by auxin represses autophagy in both conditions, indicating that activation of autophagy upon salt and drought stress is also dependent on TOR. A substrate of TOR, S6K, shows reduced expression and activity under salt and osmotic stress (Mizoguchi et al., 1995; Mahfouz et al., 2006), suggesting that salt and osmotic stress reduce TOR activity. However, it is unclear how salt and osmotic stresses signal to the TOR signaling pathway. Salt, osmotic stress and nutrient deficiency all increase cellular ROS levels, which might function as signaling molecules or lead to oxidative stress (Zhu, 2016). A major source of signaling ROS is generated by plasma membrane NADPH oxidases (Miller et al., 2009), and we have shown previously that NADPH oxidase inhibitors block autophagy during nutrient deficiency and salt stress, but not osmotic stress (Liu et al., 2009). Osmotic stress activation of autophagy is therefore independent of NADPH oxidase. NADPH oxidase inhibitors also fail to inhibit the constitutive autophagy caused by down-regulation of TOR by RNA interference (Liu and Bassham, 2010). TOR may therefore act downstream of NADPH oxidase in regulating autophagy, or possibly in a parallel pathway that is independent of NADPH oxidase. SA has been shown to increase ROS signaling, and autophagy is induced by the SA analog BTH (Yoshimoto et al., 2009). However, overexpression of TOR or increasing TOR activity with auxin failed to inhibit BTH-induced autophagy, suggesting that SA-induced autophagy is TOR-independent. Excessive ROS also cause oxidative stress, and increasing TOR activity by overexpression or auxin addition failed to repress autophagy induced by H2O2, suggesting that oxidative stress activates autophagy through a TOR-independent pathway. It is still unclear whether signaling ROS regulate autophagy through TOR, and further work is needed to identify the stress sensors that trigger activation of autophagy.

Salt, drought, and heat stresses can also cause accumulation of excessive unfolded or misfolded proteins within the ER, known as ER stress. ER stress has been shown to induce autophagy in Arabidopsis (Liu et al., 2012; Yang et al., 2016). However, our data indicate that TOR overexpression has no effect during ER stress, suggesting that ER stress-induced autophagy is independent of TOR. Upon ER stress, the plant ER stress sensor inositol-requiring enzyme-1 (IRE1) splices the mRNA encoding the transcription factor membrane-associated basic leucine zipper 60 (bZIP60) to activate the unfolded protein response (UPR). The UPR aids proper folding or degradation of unfolded and misfolded proteins via upregulation of UPR-related genes (Howell, 2013). In Arabidopsis, induction of autophagy by ER stress is triggered by unfolded and misfolded proteins (Yang et al., 2016) and is dependent on one of the IRE1 isoforms, IRE1b,

but not on IRE1a or bZIP60 (Liu et al., 2012). ER stress therefore appears to activate autophagy through IRE1b, in a pathway that is independent of TOR. However, how IRE1b regulates autophagy upon ER stress, and whether other UPR response genes are involved, requires further investigation.

Auxin has long been studied for its critical role in plant growth regulation (Enders and Strader, 2015). Auxin increases TOR activity, and auxin-mediated root gravitropism is impaired when TOR signaling is disrupted (Schepetilnikov et al., 2013). Auxin is unable to restore hypocotyl growth in estradiol-inducible tor mutants (Zhang et al., 2016), and many auxin response genes have reduced expression upon inhibition of TOR (Dong et al., 2015), suggesting that TOR is involved in auxin-regulated plant growth. Recent studies identified a small GTPase, ROP2, that mediates the activation of TOR by auxin (Li et al., 2017; Schepetilnikov et al., 2017). We hypothesized that enhancing TOR activity with auxin might repress stress-induced autophagy via the TOR signaling pathway. Indeed, as in the TOR overexpression lines, autophagy induced by nutrient starvation, salt or osmotic stresses was repressed by addition of NAA, whereas oxidative and ER stress-induced autophagy was not affected. NAA was unable to repress the autophagy induced by inhibition of TOR activity with the inhibitor AZD8055 or in a raptor1b knockout line. Exogenous application of the synthetic auxin 2, 4-D failed to restore growth of raptor1b, although raptor1b mutants can sense exogenous auxin normally (Anderson et al., 2005), supporting the conclusion that TOR signaling acts downstream of auxin.

In summary, we have demonstrated that autophagy can be regulated through TOR-dependent or -independent pathways, depending on the type of stress, and that auxin regulates plant stress responses through the TOR signaling pathway. Future

#### REFERENCES


work is required to identify the upstream stress sensors that repress TOR activity to allow activation of autophagy and the components of the TOR-independent autophagy activation pathway.

#### AUTHOR CONTRIBUTIONS

YP and DB designed the experiments. YP and XL conducted the experiments and analyzed data. YP and DB wrote the manuscript.

### FUNDING

This work was supported by grant no. 1R01GM120316-01A1 from the United States National Institutes of Health to DB and by the Iowa State University Plant Sciences Institute.

#### ACKNOWLEDGMENTS

We thank Drs Maureen Hanson for providing raptor1a and raptor1b mutant seeds, Raju Datla for TOR-OE1 and TOR-OE2 transgenic seeds, Richard Vierstra for atg7-2 mutant seeds, Yanhai Yin for the pPZP212 vector and Margaret Carter for assistance with confocal microscopy.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.01204/ 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 © 2017 Pu, Luo and Bassham. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Endocytosis of AtRGS1 Is Regulated by the Autophagy Pathway after D-Glucose Stimulation

Quanquan Yan<sup>1</sup> , Jingchun Wang<sup>1</sup> , Zheng Qing Fu<sup>2</sup> and Wenli Chen<sup>1</sup> \*

<sup>1</sup> Ministry of Education Key Laboratory of Laser Life Science and Institute of Laser Life Science, College of Biophotonics, South China Normal University, Guangzhou, China, <sup>2</sup> Department of Biological Sciences, University of South Carolina, Columbia, SC, United States

Sugar, as a signal molecule, has significant functions in signal transduction in which the seven-transmembrane regulator of G-protein signaling (RGS1) protein participates. D-Glucose causes endocytosis of the AtRGS1, leading to the physical uncoupling of AtRGS1 from AtGPA1 and thus a release of the GAP activity and concomitant sustained activation of G-protein signaling. Autophagy involves in massive degradation and recycling of cytoplasmic components to survive environmental stresses. The function of autophagy in AtRGS1 endocytosis during D-glucose stimulation has not been elucidated. In this study, we investigate the relationship between autophagy and AtRGS1 in response to D-glucose. Our findings demonstrated that AtRGS1 mediated the activation of autophagy by affecting the activities of the five functional groups of protein complexes and promoted the formation of autophagosomes under D-glucose application. When the autophagy pathway was interrupted, AtRGS1 recovery increased and endocytosis of ATRGS1 was inhibited, indicating that autophagy pathway plays an important role in regulating the endocytosis and recovery of AtRGS1 after D-glucose stimulation.

Keywords: Arabidopsis thaliana, D-glucose, AtRGS1, endocytosis, autophagy

## INTRODUCTION

In addition to integrating multi-faceted internal and external cues to gain nutrient homeostasis to build and fuel cells, sugars have significant hormone-like functions as primary messengers during signal transduction (Rolland et al., 2001, 2002; Ramon et al., 2008; Kang et al., 2010). In plants, a G-protein-coupled pathway is involved in sugar signaling. The Arabidopsis thaliana genome encodes an Arabidopsis thaliana regulator of G-protein signaling 1 (AtRGS1) protein that has an N-terminal seven transmembrane (7TM) domain and a catalytic RGS box at its C-terminal domain (Chen et al., 2003; Johnston et al., 2007; Urano et al., 2012; Bradford et al., 2013; Urano and Jones, 2014). In Arabidopsis thaliana, the heterotrimeric G-protein complex consists of one canonical Gα subunit (GPA1), one Gβ subunit (AGB1), and three Gγ subunits (Huang et al., 2015). AtRGS1 interacts with AtGPA1 and stimulates the rate-limiting GTPase activity of AtGPA1 in a D-glucose-regulated manner (Jeffrey et al., 2008). Previous studies by Urano et al. (2012) indicate that D-glucose recruits a with-no-lysing kinase (AtWNK) to phosphorylate AtRGS1 at least two C-terminal serines and that the phosphorylation of AtRGS1 is sufficient to cause endocytosis of

#### Edited by:

Yule Liu, Tsinghua University, China

#### Reviewed by:

Jirong Huang, Shanghai Institutes for Biological Sciences (CAS), China Gian Pietro Di Sansebastiano, University of Salento, Italy

> \*Correspondence: Wenli Chen chenwl@scnu.edu.cn

#### Specialty section:

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

Received: 11 April 2017 Accepted: 29 June 2017 Published: 12 July 2017

#### Citation:

Yan Q, Wang J, Fu ZQ and Chen W (2017) Endocytosis of AtRGS1 Is Regulated by the Autophagy Pathway after D-Glucose Stimulation. Front. Plant Sci. 8:1229. doi: 10.3389/fpls.2017.01229

AtRGS1 (Fu et al., 2014). Huang lab's data reveals that deletion of the C-terminus (AtRGS1−413) has no effect on AtRGS1 localization at the plasma membrane (Hu et al., 2013). In our study, we would like to investigate whether autophagy pathway is involved in AtRGS1-mediated sugar signaling pathway.

Autophagy ('self-eating') is a highly conserved mechanism among eukaryotes for degrading and recycling intracellular materials during survival to several environmental stresses (Fujioka et al., 2008; Mizushima et al., 2011; Wang P. et al., 2016). Despite the clear role of autophagy in the innate immune system of the plant during infection by necrotrophic pathogens, recent studies have demonstrated that autophagy plays an important role in degrading and recycling intracellular molecules to regulate catabolic processes (Bassham, 2007; Li and Vierstra, 2012; Gao et al., 2016). When autophagy is induced, double-membrane vesicles (autophagosome) generate and engulf the cytoplasmic components, then transport them to the lysosome (animals) or vacuole (yeast and plants) for degradation. The degradation products of these cellular materials are released into the cytoplasm for recycling (Yang and Bassham, 2015; Wang X.D. et al., 2016). Autophagy likely has essential functions as protective mechanisms that assist plants to survive unfavorable growth conditions. Several distinct autophagic types have been defined in many species and classified as microautophagy, macroautophagy, chaperonemediated autophagy, and organelle-specific autophagy (He and Klionsky, 2009; Liu and Bassham, 2012; Chanoca et al., 2015). Macroautophagy (hereafter, referred to as autophagy), as the most extensively studied autophagic type, is generated at the certain stage of development or upon encountering environmental stresses.

Several studies point to the relationship between RGS and autophagy pathway in animals. The Gα-interacting protein (GAIP), a member of a novel RGS family, is known to interact with the heterotrimeric G protein Gαi3, to regulate inhibitory regulative G-protein (Gi) signaling pathways (Ogier-Denis et al., 1997). In human intestinal cells, GAIP/RGS19 acts as a GTPase-activating protein and inactivates the Gi protein, stimulating the lysosomal-autophagic pathway (Garcia-Marcos et al., 2011). In HT-29 cells, amino acids can mediate autophagy by inhibiting extracellular signal-regulated kinase1/2 (ERK1/2)-dependent GAIP phosphorylation (Ogier-Denis et al., 2000). Both GAIP and the Gαi3 protein are part of a signaling pathway that controls lysosomal autophagic catabolism. Phosphorylation of GAIP depends on the activation of the Erk1/2 MAP kinases. Activated ERK also abolishes the autophagy-inhibitory effects of the heterotrimeric Gi protein isoform Gi3 (Pattingre et al., 2003; Law et al., 2016).

Whether the AtRGS1-mediated sugar signaling pathway is related to the autophagy pathway during D-glucose stimulation remains elusive. Here, we show that autophagy is essential for regulation of the AtRGS1-mediated sugar signaling pathway in response to D-glucose in Arabidopsis thaliana. Furthermore, autophagy could promote the endocytosis of AtRGS1 and is involved in the recovery of AtRGS1 under D-glucose treatment. These results provide deep insights to the mechanism of AtRGS1 endocytosis upon D-glucose stimulation.

### MATERIALS AND METHODS

#### Plant Materials

The plants used in all experiments were Arabidopsis thaliana ecotype Columbia 0 (Col-0). The transgenic seeds (GFP tagged ATG8a, GFP-ATG8a adriven by 35S promoter) were provided by Dr. Li Faqiang (Biochemistry, City University of New York, New York, NY, United States). The T-DNA knockout lines atg2 (SALK\_076727), atg5 (SALK\_020601) were obtained from the Arabidopsis Biological Resource Center (ABRC). The 35S::AtRGS1-YFP (35S::AtRGS1 was subcloned into pEarleyGate101) Agrobacterium and Atrgs1–2 (SALK\_074376) mutant seeds were kindly provided by Dr. Alan Jones (University of North Carolina, Chapel Hill, NC, United States). The overexpression transgenic seeds of AtRGS1-YFP, atg5/AtRGS1-YFP and atg2/AtRGS1-YFP were created by Dr. Wenli Chen in the laboratory of Alan Jones. Homozygousities of all DNA-insertion were confirmed by PCR analysis of genomic DNA with the primer sets listed in Supplementary Table S1 and **Figures S1B,C,E**. The transgenic lines were screened on MS plates with Basta (10 µg·mL−<sup>1</sup> ) (**Supplementary Figure S1A**). The specific primers listed in Supplementary Table S1 were used to examine the gene expression levels of AtRGS1 described in **Supplementary Figure S1D**.

### Growth Conditions and Treatment

Approximately, 100 seeds of Arabidopsis thaliana were sterilized by sequential treatments with 75% (v·v −1 ) ethanol (1 min) and 1% (v·v −1 ) NaClO (10 min), followed by washing with sterile distilled water six times, vernalizing in dark conditions for 3 days at 4◦C for better germination, and sown in liquid Murashige and Skoog (MS, without sucrose) medium with 1% D-glucose (pH 5.8). Seedlings were incubated in a plant growth chamber under dim continuous light (25 µmol m−<sup>2</sup> s −1 ) at 23◦C, shaking at 100 r·min−<sup>1</sup> for another 5 days (Normal conditions). To sugar starve seedlings, the seedlings were then transferred to 500 mL flasks with 100 mL liquid MS medium without D-glucose or any other sugar and allowed to grow on a shaker (100 r·min−<sup>1</sup> ) in the dark for 2 h. Following sugar starvation, seedlings were removed from the sugar-free MS medium and incubated with liquid MS media containing 1% or 6% D-glucose for the indicated time periods (0, 0.5, and 2 h) on a shaker (100 r·min−<sup>1</sup> ) (Jeffrey et al., 2008; Urano et al., 2012).

#### Gene Expression Analysis

Seedlings (0.1 g) were collected and frozen in liquid nitrogen and stored at −80◦C by use of the TRIzol reagent according to the manufacturer's instructions. The total isolated RNA was treated with primeScript RT Master Mix (Takara) according to the manufacturer's instructions to synthesize the cDNA (Mackey et al., 2003; Caplan et al., 2008). Then, cDNA synthesis was

performed by adding 5× PrimeScript RT Master Mix (Takara, 1× final concentration), total RNA (0–500 ng) and RNAsefree dH2O to a final volume of 10 µl, incubating the samples at 37◦C for 15 min, followed by incubating at 85◦C for 5 s to terminate the reactions. We amplified different autophagyrelated genes to quantify transcript levels by using gene-specific primers (Supplementary Table S1) in seedlings exposed to different treatments. Real-time PCR (qRT-PCR) was performed in Applied Biosystems 7500 with the following thermocycler program: 1 min of pre-incubation at 95◦C followed by 35 cycles of 15 s at 94◦C, 30 s at 55◦C, and 35 s at 72◦C. SYBR Green dye fluorescence was used at the end of the annealing phase. To confirm the presence of single products, a melting curve from 65 to 95◦C was used. The level of relative expression was analyzed by using the 2−11Ct analysis method (Sun et al., 2012).

#### Confocal Microscopy

Root cells of seedlings located approximately in the elongation region were imaged using a Zeiss LSM 710 META system (LCSM; Carl-Zeiss, Jena, Germany). Confocal microscopy with excitation at 488 nm (a multi-Ar ion laser) and emission at 505–550 nm was used to detect the GFP-ATG8a fusion protein. YFP signals were excited by a 514-nm argon laser and its emission was detected at 520–550 nm by a photomultiplier detector, and for obtaining image quantification, ImageJ plugins was used. The digital Images were captured with a 40× oil immersion objective and analyzed with Aim Image Browser Image Processing software (Carl Zeiss) (Ishida et al., 2008; Zhang and Chen, 2011; Sheng et al., 2012).

#### Protein Isolation and Immunoblotting

Plant samples (0.4 g) were ground in liquid nitrogen, re-suspended in ice-cold protein extraction buffer (10 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.2% Triton X-100, 0.5% Nonidet p-40, 1 mM PMSF, 3 mM DTT, 0.5 mM CaCl2, 1% ASB-14 [For AtRGS1-YFP protein] and inhibitors), and then incubated on ice for 20 min. Samples were centrifuged at 12000 × g at 4◦C for 20 min. The supernatants were transferred to new 1.5 ml eppendorf tubes. Protein concentrations were measured by the Bradford method after the total protein was diluted 100 times. Sample buffer (5×, 25 µl) with 20 mM DTT was added to 100 µl of extracted proteins. The extracted proteins were vortexed lightly and incubated in 90◦C boiling water for 2–5 min. Total proteins were then separated by SDS-polyacrylamide gel electrophoresis (PAGE) and transferred onto polyvinylidene difluoride membranes as described in earlier studies (Wang X.D. et al., 2016). The resulting membranes were blocked in TBST buffer with 5% (w/v) skim milk and incubated with the primary antibodies, followed by the corresponding secondary antibodies (Lianke <sup>R</sup> , Hangzhou, China). TBST with 5% (w/v) skim milk was used to dilute antibodies for immunoblotting. Antibodies for GFP (AG279) were purchased from the Beyotime Institute of Biotechnology (Shanghai, China); Antibodies for YFP (632381) was obtained from Clontech; Antibodies against ATG7 (ab99001) was purchased from Abcam (Shanghai, China). The levels of plant-actin were analyzed as controls for different treatments by western analysis using plant-actin antibodies (E12-053, Enogene <sup>R</sup> Biotech, New York, NY, United States) (Yue et al., 2012). All digital images were analyzed in Image J when necessary.

### Statistical Analysis

All results were repeated at least three times. Statistical analysis was performed with an ANOVA in SPSS software. Statistical significance was accepted at the level of <sup>∗</sup>P < 0.05, ∗∗P < 0.01.

## RESULTS

#### D-Glucose Activates the Autophagosomes

AtRGS1 serves as a receptor for glucose which induces its endocytosis (Grigston et al., 2008; Urano et al., 2012). Autophagy is a regulated vacuolar degradation pathway (Bassham, 2007). ATG8a protein is used as an alternative marker of autophagy. The green punctate structures, which could be labeled by GFP-ATG8a, are usually considered to be autophagosomes and their intermediates. The ATG8a protein always adheres to the autophagic vacuoles in the process of autophagic transport (Contento et al., 2005; Yoshimoto et al., 2009). To assess how the autophagy-related signaling pathway was activated in signal transduction pathways mediated by D-glucose, we examined root cells of GFP-ATG8a transgenic seedlings to see if autophagosomes were induced by D-glucose in Arabidopsis thaliana. Therefore, before and after D-glucose treatment, the green punctate structures were examined using laser confocal scanning microscopy. Concanamycin A (CA) is an inhibitor of the vacuolar type H+-ATPase (V-ATPase) that has been used to inhibit degradation by reducing the activity of vacuoles to stabilize autophagic cargoes (Kim et al., 2013; Shin et al., 2014). Before observation, the starved seedlings were treated with CA.

As observed in **Figure 1**, in the root cells of GFP-ATG8a plants, a small number of autophagosomes were detected in starved seedlings for 2 h (**Figure 1B**) and 2.5 h (**Figure 1C**). Addition of D-glucose led to dramatic changes. There was a significant increase of autophagosomes. These results indicate that D-glucose could cause AtRGS1 endocytosis by acting as a signaling factor to activate the formation of autophagosomes.

#### Induction of Autophagic Flux by D-Glucose

Autophagic flux was measured by western blotting using GFP antibodies in GFP-ATG8a transgenic Arabidopsis (Gao et al., 2016). During transport to the vacuolar lumen, GFP-ATG8, as a marker of autophagy, is often degraded to release free GFP, which was used to measure autophagic flux (Li et al., 2014). As shown in **Figure 2**, compared to the seedlings D-glucose treated for 0 h, seedlings treated for 0.5 and 2 h showed a higher level of GFP, in a time dependent manner (**Figure 2**). Our data indicate autophagic flux was induced by D-glucose in the plants of GFP-ATG8a, demonstrating that GFP-ATG8 was transported to the vacuolar lumen.

starved seedlings treated with 1% <sup>D</sup>-glucose for 0 h at <sup>∗</sup>P < 0.05 or ∗∗P < 0.01. Error bar represent SD obtained from three independent replicates.

Autophagy-Related Genes Expressions Rise in Wild-Type after D-Glucose Treatment

Recently, several Arabidopsis thaliana autophagy-related (ATG) genes have been identified and their functions of them have been well-assessed (Liu and Bassham, 2012). Five functional groups, ATG1 composite enzyme complex, Beclin1-phosphatidylinositol 3-kinase (PI3K), ATG9 complex and two ubiquitination-like conjugation systems including ATG12-ATG5 and ATG8-PE, participate in the process of autophagosome formation (Hofius et al., 2011; Shpilka et al., 2015). To investigate the function of autophagy in the AtRGS1-mediated sugar signaling pathway in Arabidopsis thaliana, we studied the D-glucose-induced ATG gene expression in both Atrgs1–2 mutants and wild-type (WT) plants. qRT-PCR analyses showed that after D-glucose treatment, the transcript levels of all seven ATG genes were induced to a much higher level in WT plants than in Atrgs1–2 null plants, indicating that AtRGS1 is required for the expression of these seven ATG genes upon D-glucose stimulation (**Figure 3**). In addition, the induction of ATG was slower than Atrgs1–2 null plants than in WT plants. The expression level of ATG1, ATG4b, and ATG12a rose at 2 h in Atrgs1–2 (**Figures 3A,D,G**). However, in WT, the expression level of the ATG genes reduced when treated with D-glucose for 2 h, but still higher than the expression when treated with D-glucose for 0 h except ATG4b (**Figure 3D**). Therefore, we believe that AtRGS1 plays an essential role in D-glucose-induced expression of autophagy genes.

In the complicated process of the autophagosome formation, five functional groups of ATG proteins are needed to form autophagosomes at the certain stage (Thompson et al., 2005;

Seay et al., 2006). The ATG1 which encodes a protein kinase participates in the autophagy activation and autophagosome initiation (Kamada et al., 2010; Suttangkakul et al., 2011; Noda and Fujioka, 2015), and ATG6/ PI3K is essential for the nucleation of autophagic vacuoles (Patel and Dinesh-Kumar, 2008). We found a significant increase in the transcript levels of ATG1 and ATG6 in WT after 0.5 h D-glucose induction (**Figures 3A,F**), suggesting that autophagy was induced and autophagic vacuoles were nucleated. The ATG9 complex was sufficient for the membrane formation of autophagosomes. ATG2 transcript level dramatically increased in response to treatment with D-glucose in WT (**Figure 3B**). The ATG12- ATG5 and ATG8-phosphatidylethanolamine (PE) conjugation systems are two ubiquitination-related conjugation system of autophagy, which are essential for the formation and closure of autophagic vesicles and the subsequent delivery to the vacuole (Thompson et al., 2005; Li et al., 2012; Ohsumi, 2014). As shown in **Figures 3C–E,G**, the increase of ATG4a, ATG4b, ATG5, and ATG12a expression after 0.5 h of stimulation suggested that the two ubiquitin-like protein-conjugating pathways were activated. Autophagy-related gene expression increased in WT, indicating that D-glucose had an important role in the formation of the five functional groups, and promoted the formation of autophagosomes.

### The Expression of ATG7 Protein Rises in Wild-Type upon D-Glucose Stimulation

The E1-like ATG7-activating enzyme is required to form ATG8 phosphatidylethanolamine (PE) and ATG12-ATG5 conjugation systems, subsequently E2-conjugating enzymes, ATG3 and ATG10 combine ATG8 and ATG12 with PE and ATG5 (Wang P. et al., 2016). To elucidate the response of autophagy to D-glucose, we examined the autophagosome generation under D-glucose; ATG7 (76 kDa) protein levels were analyzed by western blot using an anti-ATG7 antibody. In WT, the expression of ATG7 protein significantly increased in response to Dglucose (**Figures 4A,B**). On the contrary, in Atrgs1–2 mutants, the expression of ATG7 only slight increased after D-glucose application (**Figures 4A,B**). The data obtained in the experiment

suggested that AtRGS1 promoted the expression of ATG7 after D-glucose treatment.

#### ATG2 and ATG5 Inhibit the Recovery of AtRGS1 in Response to D-Glucose

Autophagy is a precise and complex machinery used to degrade and recycle intracellular molecules in plants during response and survival to environmental stresses (Kulich et al., 2013; Bassham and Crespo, 2014). The disruption of the ATG2 and ATG5 gene blocks the formation of autophagosomes (Ishida et al., 2008; Phillips et al., 2008; Lee et al., 2014). AtRGS1, as a D-glucose receptor, mediates sugar signaling pathway by endosomes. To investigate the cellular function of autophagy during the AtRGS1 response to D-glucose, the levels of AtRGS1-YFP protein were assessed by the immunoblot analysis using an anti-YFP antibody in AtRGS1-YFP, atg2, and atg5 autophagy-deficient mutants.

Our data showed that AtRGS1-YFP protein expression levels were reduced after starvation, but, increased after D-glucose application in the plants of AtRGS1-YFP and the two autophagydeficient mutants (**Figures 5A–D**). However, after 2 h starvation, AtRGS1-YFP protein in AtRGS1-YFP plants reduced more than that in the two mutants (**Figure 5D**). That means ATG2 and ATG5 promote the decrease of AtRGS1 protein under starvation treatment. After glucose application, AtRGS1-YFP protein recovery in AtRGS1-YFP plants was slower than that in atg2 and atg5 mutants (**Figure 5D**). These results demonstrated that ATG2 and ATG5 inhibit the AtRGS1 protein recovery after D-glucose treatment.

#### ATG2 and ATG5 Promote the Endocytosis of AtRGS1 in Response to D-Glucose

It has been proven that D-glucose induces AtRGS1 endocytosis (Grigston et al., 2008). To determine whether autophagy functions to mediate endocytosis of AtRGS1 after D-glucose application, we analyzed the subcellular distribution of YFP fluorescence by using laser confocal scanning microscopy in AtRGS1-YFP, atg2 and atg5 plants. In AtRGS1-YFP plants, endocytosis occurred after D-glucose application (**Figure 6A**). But, in atg2 and atg5 mutants, AtRGS1-YFP was mainly localized to the plasma membrane, and the internalization rate increased only slightly in response to D-glucose (**Figures 5B–D**). The results were consistent with the movement of AtRGS1- YFP in AtRGS1-YFP, atg2 and atg5 plants (**Supplementary Figure S2**). These results revealed that under the stimulation of D-glucose, when the autophagy pathway was deficient in atg2 and atg5 plants, the endocytosis of AtRGS1 was inhibited. AtRGS1 is mainly localized on the plasma membrane. Our data demonstrated that means ATG2 and ATG5 promote AtRGS1 endocytosis.

## DISCUSSION

AtRGS1, which is proposed to be an extracellular receptor for D-glucose, accelerates the hydrolysis of AtGPA1 to negatively regulate G-protein signaling (Urano et al., 2012). Our study indicated that D-glucose could activate autophagy and that autophagy pathway did not function well in Atrgs1–2 mutants. The rapid recovery of AtRGS1 in atg2 and atg5 mutants suggested that the autophagy pathway might be involved in inhibition of the recovery of AtRGS1 in response to D-glucose. AtRGS1 endocytosis was inhibited and it mainly located on the plasma membrane in plants of atg2 and atg5, indicating that ATG2 and ATG5 promote the endocytosis of AtRGS1. In this study we have demonstrated that autophagy is required for regulation of the AtRGS1-mediated sugar signaling pathway in response to D-glucose.

### The Autophagy Pathway Is Induced by D-Glucose

In plants, glucose as a metabolite or signaling molecule affects gene and protein expressions, and growth and developmental programs (Sheen, 2014; Huang et al., 2015). Autophagy is an intracellular degradation system conserved in eukaryotic cells that consists of the formation of double-membrane structures called autophagosomes that engulf and sequester the cytoplasmic components and then fuse them with the endosome/vacuole and finally break them down in the vacuole (Bassham, 2007; Wang P. et al., 2016).

Punctate-like structures labeled by GFP-ATG8a, which represent autophagosomes and their intermediates, appear in large numbers after treatment with D-glucose (**Figure 1**). Autophagic flux is often utilized to measure the degradation activity of autophagic substrates within the vesicular system (Mizushima et al., 2010; Loos et al., 2014). The increase of the accumulation of free GFP dependent on time implied that autophagic flux was induced by D-glucose in GFP-ATG8a plants (**Figure 2**), suggesting that autophagosomes were transported to the vacuole for breakdown. The assembly of five functional groups of ATG proteins is required for the formation of autophagy (Obara and Ohsumi, 2011). The increase of the expression autophagy-related gene expression in WT indicated that D-glucose promoted formation of autophagosomes (**Figure 3**). E1-like ATG7-activating enzyme is essential for the formation of ATG8-phosphatidylethanolamine (PE) and ATG12-ATG5 conjugation systems (Wang P. et al., 2016). After induction with D-glucose, the increased level of ATG7 protein in WT promotes the formation of two ubiquitination conjugation systems (**Figure 4**). Previous research has shown that the autophagic flux was inhibited in atg7 mutants in Arabidopsis thaliana (Shin et al., 2014; Wang X.D. et al., 2016). In WT, the increased level of ATG7 protein induced by Dglucose will likely promote the autophagic flux (**Figure 4**). Taken together, these results indicate that D-glucose induces autophagy, which plays an important role in intracellular homeostasis.

#### AtRGS1 Is Required for Autophagy Pathway Induced by D-Glucose

In Arabidopsis thaliana, three signaling pathways that are sensitive to glucose have been analyzed. (1) The hexokinase 1 (AtHXK1) pathway: AtHXK1 is recognized as glucose sensor involving in the metabolic and physiological processes by mediating gene transcription in response to glucose (Sheen, 2014). (2) The glycolysis-dependent SNF1-RELATED KINASE1/TARGET OF RAPAMYCIN (SnRK1/TOR) pathway: the glycolysis-dependent SnRK1/TOR pathway acts as cellular energy sensors regulated growth and development (Lastdrager et al., 2014). (3) The AtRGS1-dependent G-protein-coupled signaling pathway: G-protein-coupled pathways involving AtRGS1 are involved in sugar signaling (Grigston et al., 2008). Genetic evidence suggests that D-glucose causes AtRGS1 endocytosis or a sugar metabolite regulates AtRGS1 activity toward AtGPA1 (Urano et al., 2012).

Our results demonstrate that D-glucose induces autophagy pathway in WT Arabidopsis plants (**Figures 1**–**4**). However, in Atrgs1–2 mutants after D-glucose application, the analysis of qRT-PCR data indicate that the autophagy pathway did not function well (**Figure 3**), and the expression of ATG7 increase only slightly (**Figure 4**). The E1-like ATG7 activating enzyme is required to form the two ubiquitin-like protein-conjugating pathways (Wang P. et al., 2016), so we hypothesize that D-glucose induces the formation and activation of ATG8-phosphatidylethanolamine (PE) and ATG12-ATG5 conjugation systems, which are closely related to AtRGS1. A lot of evidence (**Figures 3**, **4**) has been provided in support of the relationship between autophagy and AtRGS1 protein. Autophagy has been shown to be involved in AtRGS1-mediated sugar signaling pathway.

### ATG2 and ATG5 Regulate the Recovery and Endocytosis of AtRGS1 after D-Glucose Stimulation

Autophagy appears to be the main contributor to the maintenance of the equilibrium with different environmental stresses and the degradation and reuse of nutrients (Li et al., 2012; Liu and Bassham, 2012; Ohsumi, 2014; Wang P. et al., 2016). Endocytosis is a major route of entry for plasma membrane proteins or extracellular materials into the cell for recycling back to the plasma membrane, or degradation in the vacuole. Endocytosis plays an important role in the process of cellular responses to environmental stimuli and signaling transduction (Fan et al., 2015; Michaeli et al., 2016).

In mammalian cells, autophagosomes probably fuse with endosomes of GPCRs induced by agonists for degradation, or endosomes of GPCRs involve recycling via the endosome. In plants, the AtRGS1 interacted with some relative proteins stimulates the endocytosis of AtRGS1 and promotes downstream signaling pathway (Urano et al., 2012). AtWNK kinases recruited by D-glucose phosphorylate AtRGS1 to mediate endocytosis (Urano et al., 2012). The interactions between AtRGS1 and small GTPases molecules of RAB (Ras-like small GTP binding) and ARF (ADPribosylation factor) are hypothesized to regulate AtRGS1 to be endocytosed into the cytoplasm, then recycled back to the plasma membrane via recycling endosomes (Jaiswal et al., 2015).

In Arabidopsis thaliana, D-glucose induced the increase of the levels of AtRGS1-YFP protein expression in the plants of AtRGS1-YFP and atg2 and atg5 mutants (**Figures 5A–D**). After D-glucose application, AtRGS1-YFP proteins recovered more rapidly in atg2 and atg5 mutants than that in AtRGS1-YFP plants, indicating that ATG2 and ATG5 inhibited the AtRGS1 protein recovery after D-glucose treatment. Therefore, we conclude that autophagy pathway regulates AtRGS1 recovery (**Figure 5**).

However, in autophagy-deficient atg2 and atg5 mutants, the endocytosis of AtRGS1 was inhibited, indicating that ATG2 and ATG5 promote the endocytosis of AtRGS1 in response to D-glucose (**Figure 6**). Endocytosis of AtRGS1 physically uncouples the GTPase-accelerating activity of AtRGS1 from the G protein, which permits sustained activation (Urano et al., 2012). Perhaps inhibition of endocytosis of AtRGS1 leads to a physical coupling of AtRGS1 with AtGPA1 and concomitant inhibition of G-protein signaling.

In plants, endocytic and autophagic pathways interplay. Autophagosomes fuse with the endosome/vacuole, and degrade in the vacuole (Zhuang et al., 2015). So we hypothesize endocytosis of AtRGS1 and autophagic pathways interplay in Arabidopsis thaliana, and autophagic pathways have an effect on

the recycling back to the plasma membrane, or degradation in the vacuole of AtRGS1.

Further study aims to determine whether AtRGS1-YFP colocalized with autophagic bodies after D-glucose stimulation. This study is essential for extending our knowledge of the relationship between AtRGS1 and autophagy.

#### AUTHOR CONTRIBUTIONS

QY and WC conceived and designed the experiments. QY and JW performed research. QY, ZF, and WC analyzed the data. QY and ZF wrote the manuscript.

#### FUNDING

This research was supported by the National Natural Science Foundation of China (Grant Numbers 31570256 and 31170250), and the Natural Science Foundation of Guangdong Province, China (Grant Number, 2014A030313420).

### ACKNOWLEDGMENTS

We thank Professor Alan M. Jones for providing seeds from Atrgs1–2 mutant and the 35S::AtRGS1-YFP (35S::AtRGS1 was subcloned into pEarleyGate101) Agrobacterium. We also

### REFERENCES


Chanoca, A., Kovinich, N., Burkel, B., Stecha, S., Bohorquez-Restrepo, A., Ueda, T., et al. (2015). Anthocyanin vacuolar inclusions form by a microautophagy mechanism. Plant Cell 27, 2545–2559. doi: 10.1105/tpc.15.00589


thank Dr. Li Faqiang for providing the transgenic Arabidopsis expressing GFP-ATG8a.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.01229/ full#supplementary-material

FIGURE S1 | (A) The transgenic lines (AtRGS1-YFP, atg2/AtRGS1-YFP and atg5/AtRGS1-YFP) and WT were screened on MS plates with Basta (10 µg/mL). DNA was respectively isolated from eight seedlings (1–8) of atg2/AtRGS1-YFP (B), atg5/AtRGS1-YFP (C), and Atrgs1–2 (E). WT acts as a control. Homozygous for DNA-insertion were confirmed by PCR of genomic DNA with the primers. The products were electrophoresed on a 1% gel for 30 min, then detected by using UV illumination. (D) qRT-PCR was performed to analyze transcription level of AtRGS1 in AtRGS1-YFP, atg2/AtRGS1-YFP and atg5/AtRGS1-YFP lines grown on MS plates for 7 days. Data represent mean and SD of at least three independent experiments. The asterisk indicates a significant difference from WT (∗∗P < 0.01).

FIGURE S2 | Movement of AtRGS1-YFP in the plant of AtRGS1-YFP (A), atg2 (B), atg5 (C). Root cells of seedlings located approximately in the elongation region were imaged using a Zeiss LSM 510 confocal laser scanning microscope (LCSM, LSM 510/ConfoCor 2, Carl-Zeiss, Jena, Germany). The normal seedlings were observation in response to D-glucose. Scale bars represent 10 µm. (D) Quantification of AtRGS1-YFP internalization rate. Internalization rate was calculated by dividing the internalization of the normal seedlings treated with D-glucose for 5, 10, or 15 min by the internalization of the normal seedlings treated with D-glucose for 0 min. Experiments were performed three independent replicates with similar results. Asterisks, significant differences from the starved seedlings treated with <sup>D</sup>-glucose for 0 min, <sup>∗</sup>P < 0.05 or ∗∗P < 0.01.

cell death in Arabidopsis. Acta Physiol. Plant 38, 150. doi: 10.1007/s11738-016- 2160-4


Zhuang, X. H., Cui, Y., Gao, C. J., and Jiang, L. W. (2015). Endocytic and autophagic pathways crosstalk in plants. Curr. Opin. Plant Biol. 28, 39–47. doi: 10.1016/j.pbi.2015.08.010

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

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

# Autophagy Is Rapidly Induced by Salt Stress and Is Required for Salt Tolerance in Arabidopsis

Liming Luo† , Pingping Zhang† , Ruihai Zhu, Jing Fu, Jing Su, Jing Zheng, Ziyue Wang, Dan Wang\* and Qingqiu Gong\*

Tianjin Key Laboratory of Protein Sciences, Department of Plant Biology and Ecology, College of Life Sciences, Nankai University, Tianjin, China

Salinity stress challenges agriculture and food security globally. Upon salt stress, plant growth slows down, nutrients are recycled, osmolytes are produced, and reallocation of Na<sup>+</sup> takes place. Since autophagy is a high-throughput degradation pathway that contributes to nutrient remobilization in plants, we explored the involvement of autophagic flux in salt stress response of Arabidopsis with various approaches. Confocal microscopy of GFP-ATG8a in transgenic Arabidopsis showed that autophagosome formation is induced shortly after salt treatment. Immunoblotting of ATG8s and the autophagy receptor NBR1 confirmed that the level of autophagy peaks within 30 min of salt stress, and then settles to a new homeostasis in Arabidopsis. Such an induction is absent in mutants defective in autophagy. Within 3 h of salt treatment, accumulation of oxidized proteins is alleviated in the wild-type; however, such a reduction is not seen in atg2 or atg7. Consistently, the Arabidopsis atg mutants are hypersensitive to both salt and osmotic stresses, and plants overexpressing ATG8 perform better than the wildtype in germination assays. Quantification of compatible osmolytes further confirmed that the autophagic flux contributes to salt stress adaptation. Imaging of intracellular Na<sup>+</sup> revealed that autophagy is required for Na<sup>+</sup> sequestration in the central vacuole of root cortex cells following salt treatment. These data suggest that rapid protein turnover through autophagy is a prerequisite for salt stress tolerance in Arabidopsis.

#### Keywords: autophagy, Arabidopsis, salt stress, autophagic flux, ATG8, NBR1, vacuoles

### INTRODUCTION

Soil salinity is a major abiotic factor that limits crop yield (Flowers, 2004; Munns and Tester, 2008; Hanin et al., 2016). Globally, about 900 million hectares of land were estimated to be saline, and more than 30% of the irrigated crops were salt-affected (FAO<sup>1</sup> ) (Flowers, 2004; Schroeder et al., 2013). High concentrations of NaCl inhibits plant water uptake, and Na<sup>+</sup> and Cl<sup>−</sup> accumulated in the cytosol lead to ion toxicity. As results, photosynthetic rates are reduced, leading to energy depletion and accumulation of excessive reactive oxygen species (ROS) (Zhu, 2002; Deinlein et al., 2014; Golldack et al., 2014; Julkowska and Testerink, 2015). In order to alleviate the osmotic and ionic stresses, plants close their stomata to minimize water loss (Munemasa et al., 2015), reduce their growth rates (Julkowska and Testerink, 2015), and limit intracellular Na<sup>+</sup> concentration by

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

#### Edited by:

Minghui Lu, Northwest A&F University, China

#### Reviewed by:

Caiji Gao, South China Normal University, China Zhaojun Ding, Shandong University, China

#### \*Correspondence:

Dan Wang wangdan629@nankai.edu.cn Qingqiu Gong gongq@nankai.edu.cn †These authors have contributed equally to this work.

#### Specialty section:

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

Received: 31 December 2016 Accepted: 04 August 2017 Published: 22 August 2017

#### Citation:

Luo L, Zhang P, Zhu R, Fu J, Su J, Zheng J, Wang Z, Wang D and Gong Q (2017) Autophagy Is Rapidly Induced by Salt Stress and Is Required for Salt Tolerance in Arabidopsis. Front. Plant Sci. 8:1459. doi: 10.3389/fpls.2017.01459

compartmentalization in the vacuole (Munns and Tester, 2008). Gradually, relocation of Na<sup>+</sup> away from young tissues takes place to exclude Na<sup>+</sup> from growing organs (Munns and Tester, 2008; Deinlein et al., 2014), and plants may enter a growth recovery phase to resume growth at a reduced steady rate (Munns, 2002; Julkowska and Testerink, 2015).

With engineering of salt-tolerant crops in mind, numerous studies have been carried out to elucidate salt stress response, with emphases on transcription regulation and ABA signaling (Urano et al., 2010; Osakabe et al., 2014). Enormous progress have also been made in understanding the ion transport mechanisms, such as the discovery of Salt Overly Sensitive (SOS) signaling cascade and the in-depth studies on the HKT and NHX transporters (Qiu, 2012; Ji et al., 2013; Volkov, 2015). Additionally, the molecular basis for tissue tolerance has been revealed, which includes biosynthesis of osmolytes, such as glycinebetaine, sugar alcohols, polyamines, and proline (Tarczynski et al., 1993; Szabados and Savoure, 2010; Chen and Murata, 2011).

Compared with the topics above, how salt-challenged plants manage to maintain their energy level and re-allocate the limited resources is less well understood. One pathway that may contribute to salt-elicited nutrient recycling is macroautophagy (hereafter autophagy) (Han et al., 2011; Li and Vierstra, 2012; Liu and Bassham, 2012).

Autophagy is a bulk degradation pathway that helps maintain cellular homeostasis (He and Klionsky, 2009; Mizushima et al., 2011; Liu and Bassham, 2012; Ohsumi, 2014). In this pathway, obsolete proteins and damaged organelles are enveloped by an expanding double-membraned vesicle, the isolation membrane/phagophore, which matures to a sealed autophagosome before fusing with the lytic vacuole (Li and Vierstra, 2012; Lamb et al., 2013; Shibutani and Yoshimori, 2014; Michaeli et al., 2016; Zhuang et al., 2016). The inner membrane of the autophagosome along with the cargo, termed the autophagic body, is then degraded by vacuolar hydrolases, and amino acids and other macro molecules are released back into the cytosol through transporters (Kuma and Mizushima, 2010; Shibutani and Yoshimori, 2014).

The hallmark of autophagy is the formation of the autophagosome (Xie and Klionsky, 2007; Shibutani and Yoshimori, 2014; Zhuang et al., 2016). As key players in this process, autophagy-specific ubiquitin-like proteins (UBLs) ATG8/LC3/GABARAP act as protein scaffolds to mediate phagophore expansion (Nakatogawa et al., 2007; Xie et al., 2008; Weidberg et al., 2010). Conjugation of ATG8 to the phagophore requires the activity of several other ATG proteins (Shibutani and Yoshimori, 2014). Firstly, newly synthesized ATG8 is truncated by the cysteine protease ATG4 to expose the C-terminal Glycine residue (Kirisako et al., 2000; Woo et al., 2014). The Glycine is then conjugated to the amino group of Phosphatidylethanolamine (PE) in a ubiquitin-like conjugation reaction catalyzed by ATG7 as the E1, ATG3 as the E2 (Ichimura et al., 2000), and the ATG12-ATG5- ATG16 complex as the E3 enzyme (Hanada et al., 2007). PE-conjugated ATG8 stably associates with both phagophore and completed autophagosomes, hence is commonly used as a marker for microscopic study of autophagy (Kabeya

et al., 2000; Yoshimoto et al., 2004; Contento et al., 2005). In addition, PE-conjugated ATG8 moves faster than the unconjugated ATG8 in SDS-PAGE gels, hence the amount of lipidated ATG8 is indicative of autophagic activity (Kabeya et al., 2000). Furthermore, by comparing the ATG8-PE levels in the presence and absence of vacuolar protease inhibitors, such as E-64d, or tonoplast H+-ATPase inhibitors, such as concanamycin A (Con A), autophagic flux can be quantified (Mizushima and Yoshimori, 2007). The selective autophagic flux can be detected by measuring Neighbor of BRCA1 (NBR1) degradation (Mizushima and Yoshimori, 2007; Svenning et al., 2011). During selective autophagy, NBR1 binds both ATG8 and mono- and (especially) polyubiquitin, linking the ubiquitinated cargoes to the autophagy machinery. Then it is transported together with the cargoes inside the autophagosome to the lytic vacuole for degradation (Mizushima and Yoshimori, 2007; Svenning et al., 2011). Hence NBR1 serves as both a receptor and a selective substrate of autophagy, and the degradation of NBR1 is indicative of selective autophagic flux (Mizushima and Yoshimori, 2007). Therefore, immunoblotting with antibodies against ATG8, NBR1, or epitope tags fused to ATG proteins provides more quantitative information for detecting autophagy (Bao et al., 2016).

To see if the level of autophagic flux correlates with salt tolerance, Arabidopsis mutants defective in autophagy and mutants with reduced level of autophagy, as well as transgenic plants that have increased level of autophagy were selected for further analyses (**Supplementary Figure S1**). The five ATGs belong to the core autophagy machinery (Xie and Klionsky, 2007). ATG5, ATG7, and ATG10 are required for ATG8-lipid adduct and autophagosome formation, and ATG5 have been shown to localize at the outer surface of the cortical endoplasmic reticulum (ER) to recruit ATG8 for phagophore assembly and expansion (Le Bars et al., 2014). Both atg5 and atg7 are autophagy-deficient mutants (Thompson et al., 2005; Inoue et al., 2006; Shin et al., 2014). ATG2 supposedly forms a complex with ATG18 to mediate the shuttle of ATG9 vesicles, which are a source of autophagosomal membranes (Yamamoto et al., 2012). Autophagic flux is known to be reduced rather than absent in atg9 (Inoue et al., 2006; Shin et al., 2014; Zhuang et al., 2017). Transgenic Arabidopsis over-expressing GmATG8c (ATG8-OX) was used to represent plants with high levels of autophagy (Xia et al., 2012).

As a house-keeping pathway, autophagy is generally maintained as a basal level and can be quickly induced by nutrient deprivation and various stresses (He and Klionsky, 2009). Several lines of evidence suggest that autophagy is positively involved in plant salt stress adaptation. NaCl treatments have been shown to induce transcriptions of several Autophagy-related (ATG) genes, especially ATG8s and ATG18s in Arabidopsis, salt cress, rice, wheat, tobacco, pepper, and foxtail millet (Gong et al., 2005; Liu et al., 2009; Xia et al., 2011; Pei et al., 2014; Zhou et al., 2015; Li et al., 2016; Zhai et al., 2016). In Arabidopsis, RNAi-AtATG18a plants were hypersensitive to salt and osmotic stress in

germination and seedling growth (Liu et al., 2009). The autophagy-deficient mutants, atg5 and atg7, as well as the ATG8-interacting autophagy cargo adaptor mutant nbr1, exhibited sensitivity toward drought and salt treatments (Zhou et al., 2013). Mutation in ATG8-Interacting Protein 1(ATI1) led to salt sensitivity during germination (Michaeli et al., 2014). Intriguingly, it was reported that over-expression of GFP–AtAtg8f-HA led to reduced tolerance toward mild osmotic and salt stresses, but not to stronger stresses (Slavikova et al., 2008).

Many questions remain on how autophagy contributes to salt stress tolerance in plants. When does autophagic flux peak following salt stress? Does autophagy participate in the clearance of salt-induced oxidized proteins? Does it contribute to osmolyte production? Does it have a role in Na<sup>+</sup> uptake or sequestration in the vacuole?

Here we show that the autophagic flux is rapidly induced by salt treatment. Then autophagy deficient (atg5 and atg7, atg10), defective (atg9 and possibly also atg2), and enhanced (Pro35S:GmATG8c, ATG8-OX) lines were selected to see if the levels of autophagy correlate with salt stress tolerance. Physiological analyses revealed that the atg mutants accumulated less soluble sugars and proline, whereas ATG8-OX plants accumulated more osmolytes. Oxidized protein is alleviated after treated with NaCl in the wild-type within 3 h. We also show that Na<sup>+</sup> sequestration in the lytic vacuole of salt-stressed root cortex cells is correlated with the level of autophagy. Our observations suggest that salt stress rapidly triggers autophagy to facilitate bulk protein turnover, thus providing macromolecules and energy required for plant survival.

### MATERIALS AND METHODS

#### Accession Numbers

ATG2, At3g19190; ATG5, At5g17290; ATG7, At5g45900; ATG8a, AT4G21980; ATG9, At2g31260; ATG10, AT3G07525; NBR1, AT4G24690.

#### Plant Materials and Growth Conditions

Arabidopsis (ecotype Columbia-0) was grown as described (Xiong et al., 2016). Generally, seeds were surface-sterilized with 75% ethanol for 5 min, 100% ethanol for 1 min, rinsed with ddH2O for five times, then stratified at 4◦C for 2 days before plated on 1/2 Murashige and Skoog (1/2 MS) medium (Sigma– Aldrich, United States) containing 0.8% (w/v) agar (Sigma– Aldrich, United States), 1% (w/v) sucrose (Sigma–Aldrich, United States), pH5.7. Liquid 1/2 MS medium was prepared in the same way, only without agar. The plants were grown at 16 h (22◦C)/8 h (18◦C) with a photosynthetic photon flux density at 90 µE m−<sup>2</sup> sec−<sup>1</sup> . The T-DNA insertion mutants atg2-1 (Salk\_076727), atg5-1 (SAIL\_129B07, CS806267), atg7 (SAIL\_11H07, CS862226), atg9 (SAIL\_527\_A02, CS874564), and atg10-1 (Salk\_084434) were obtained from ABRC (Sessions et al., 2002; Alonso et al., 2003), and transgenic lines carrying Pro35S:GmATG8c were as described (Xia et al., 2012). All mutants and transgenic plants were verified by genomic PCR. Primers used are listed in **Supplementary Table S1**. All lines have been freshly propagated to ensure wild-type level germination rates on control medium. Phenotypes were documented as described (Xiong et al., 2016). All images were analyzed with Image J<sup>2</sup> , and statistical analyses (F-test, Student's t-test) were done with Microsoft Excel 2010.

#### Generation of an Autophagic Marker, ProATG8a:GFP-ATG8a

To construct ProATG8a:GFP-ATG8a, an 1199 bp fragment upstream of the start codon (ATG) of ATG8a (−1199 To 1) was PCR-amplified and inserted between Pst I and Nco I of pCAMBIA1302. Then a genomic fragment of ATG8a (942 bp) was inserted after GFP by homologous recombination with a ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China). The construct was verified by sequencing before introduced into Agrobacterium tumefaciens (GV3101) for floral dipping (Clough and Bent, 1998). Primary transformants were selected by antibiotic resistance and verified by PCR. The T3 homozygous marker line L5-1 was introduced into atg10 by crossing, and atg10/GFP-ATG8a plants were identified by genotyping of F2 individuals.

#### Immunoblotting

To quantify autophagic flux, 10-day-old vertically grown seedlings (approximately 200 mg) were transferred to liquid 1/2 MS medium with 150 mM NaCl for 0, 0.5, 1, 3, and 6 h before harvested. To block autophagosome degradation in the lytic vacuole, a parallel set of seedlings were transferred to medium containing NaCl plus 0.5 µM Concanamycin A (ConA, Sigma–Aldrich, United States) and harvested at the same time points. Protein extraction, quantification, and immunoblotting were done as described (Xia et al., 2012). All SDS-PAGE gels were prepared with 6 M urea. Primary antibodies used were anti-GmATG8c (1:3000) (Xia et al., 2012), anti-NBR1 (Agrisera, 1:3000), and anti-tubulin (Utibody, 1:5000). Each experiment was repeated for at least three times, and one representative result was shown. Quantification of immunoblots was done with Image J and statistical analyses (F-test, Student's t-test) were done with Microsoft Excel 2013.

#### Oxidized Protein Analysis

Oxidized protein analysis was performed following a previous report (Xiong et al., 2007). Ten-day-old seedlings grown on 1/2 MS vertical plates were transferred to liquid 1/2 MS medium with 150 mM NaCl for 0, 0.5, 1, 3, and 6 h before harvested for protein extraction. The extraction buffer contains 0.1 M Tris-HCl, pH 7.5, 0.3 M Sucrose, 1 mM EDTA, 0.1 mM phenylmethylsulphonylfluoride (PMSF), and 1% [v/v] β-mercaptoethanol. Extracts were centrifuged at 1,000 g for 10 min and supernatants were collected. Oxidized proteins were detected using an OxyBlot protein oxidation detection kit (Abcam, United Kingdom) according to the manufacturer's instructions. Dinitrophenylhydrazine (DNP) signals (entire lane)

<sup>2</sup>http://rsb.info.nih.gov/ij/

were quantified by densitometry in Image J and normalized to the wild-type 0 h control value, which was set as 1. Each experiment was repeated for at least three times, and one representative result was shown. Coomassie blue staining of total proteins was used as the loading control.

### Quantification of Proline, Total Soluble Sugar, and Reducing Sugar content

Ten-day-old seedlings grown on 1/2 MS vertical plates were transferred to liquid 1/2 MS containing 150 mM NaCl for 0, 8, and 24 h. Harvested samples were weighed, and free proline was quantified with ninhydrin assay at A520 nm with a kit (Comin Biotech, Suzhou, China) following manufacturers' instructions. Total soluble sugar and reducing sugar contents were determined by the anthrone reagent method and dinitrosalicylic acid (DNS) method with a kit (Yuanye Biotech, Shanghai, China) following manufacturers' instructions. The sugar contents were determined against a standard curve prepared with glucose (Sigma–Aldrich, United States). Three biological replicates were done in each case with consistent results, and representative results are shown.

#### Amino Acid Analysis

Ten-day-old seedlings grown on 1/2 MS vertical plates were transferred to liquid 1/2 MS containing 150 mM NaCl for 0, 8, and 24 h. A parallel set of seedlings were treated with NaCl plus 10 mM Chloroquine. Harvested samples were weighed and quantified for soluble amino acid on a membraPure Amino Acid Analyzer A300 (Germany) following manufacturer's instructions.

#### Stress Treatments

For germination assay, stratified seeds were sown on 1/2 MS medium with or without NaCl or mannitol. Four biological replicates were done, which are consist of four seed populations obtained from self-fertilization of independent parents. More than 100 seeds were used in each seed population. Radical protrusion was scored as germination at a 4-h interval during first 3 days, then daily until day 7.

#### Laser Scanning Confocal Microscopy (LSCM)

To observe NaCl-induced autophagosome formation, 4- or 5-days-old vertically grown GFP-AtATG8a seedlings were immersed in 100 mM NaCl, or 0.5 µM ConA, or both NaCl and ConA, for 30 min and 1 h, respectively, and scanned with a Leica SP5 (Leica, Germany) with a same set of scanning parameters. For each condition, at least 30 seedlings from three to five biological replicates were imaged, and >30 root cortex cells in the root hair zone were quantified for the numbers of autophagosomes/autophagic bodies.

For imaging of Na+, 5-days-old Arabidopsis seedlings were transferred to liquid 1/2 MS (control) or 1/2 MS containing 100 mM NaCl (salt stress) for 6 h, and then CoroNa Green AM (Thermo Fisher Scientific) was added into the medium to a final concentration of 5 µM. Two hours later, the seedlings were stained with 5 µM FM4-64 (Thermo Fisher Scientific) for 5 min, washed thoroughly, and scanned with a SP5 (Leica, Germany) confocal microscope following previous reports (Meier et al., 2006; Oh et al., 2010). At least 20 seedlings from three biological replicates were imaged for each line in each condition, and >30 root cortex cells in the meristematic zone were quantified for the CoroNa Green AM fluorescent intensities in Image J.

### Gene Expression Analysis

Ten-day-old seedlings were transferred to liquid 1/2 MS medium containing 150 mM NaCl for 0 (control), 30 min, and 3 h. RNA extraction, reverse transcription, RT-PCR and q-RT-PCR were performed as described (Xiong et al., 2016). EF1a (At5g60390) was used as an internal control. Three biological replicates consisting of three technical repeats each were done. Transcript levels of the nine salt-inducible genes following NaCl treatment in ATG8-OX, atg5, and atg9 were compared with both WT control and with the same germplasm control using Student's t-test. Primers used are listed in **Supplementary Table S1**.

## RESULTS

#### NaCl Treatment Rapidly Induces Autophagy in Arabidopsis Seedling Root Cortex Cells

To see if the autophagic flux is affected by salt stress, we firstly observed autophagosome formation and autophagic body turnover by looking at the autophagosome marker GFP-ATG8a in transgenic Arabidopsis seedlings carrying ProATG8a:GFP-ATG8a (**Supplementary Figure S2**). The marker line has a similar growth rate to the WT both under normal growth conditions and under nitrogen or carbon limitation, or salt and osmotic stresses (**Supplementary Figure S2**). The vacuolar proton pump inhibitor Concanamycin A (ConA) was used to preserve the autophagic bodies in the central vacuole. Under controlled conditions, GFP-ATG8a signals were mainly observed in the cytoplasm, with sporadic, relatively large (1–2 µm in diameter) puncta representing autophagosomes also detected (**Figure 1A**, DMSO). In the presence of ConA, additional smaller puncta representing autophagic bodies were observed in the vacuole (**Figure 1A**, +ConA 0.5 h). When treated with 100 mM NaCl for 30 min, large puncta accumulated (**Figure 1A**, +NaCl 0.5 h). In the presence of ConA, the numbers of autophagosomes and autophagic bodies increased even more, indicative of induced autophagic flux (**Figure 1A**, +NaCl +ConA 0.5 h). Interestingly, the number of ATG8a puncta was slightly reduced at 1 h of salt treatment (**Figure 1A**, lower panel), suggesting that the autophagic flux peaked shortly after salt stress. Quantification of the number of autophagic bodies confirmed that the autophagic flux, represented by the difference between the numbers of autophagic bodies treated with NaCl only and that of NaCl plus ConA treatment, peaked at 0.5 h (**Figure 1C**). Imaging of an atg10 mutant carrying GFP-ATG8a under the same conditions showed that few autophagosome/autophagic bodies could be detected in the mutant with or without NaCl or NaCl plus ConA (**Figure 1B**). The imaging results indicated that

autophagic flux can be rapidly induced by NaCl treatment in Arabidopsis.

#### Autophagic Flux Is Elevated Shortly after NaCl Treatment in the Wild-Type

To confirm that NaCl treatment can rapidly induce autophagy, immunoblotting of ATG8s was performed. We first examined

whether the anti-GmATG8c antisera could detect both the lipidated and the non-lipidated ATG8s following a previous report (Suttangkakul et al., 2011). Total membrane fraction from the wild-type and atg7 seedlings were collected, and the solubilized samples either treated or not with phospholipase D (PLD) were analyzed with immunoblotting. Unfortunately, the antibodies preferentially recognize un-lipidated ATG8s (**Supplementary Figure S3**). Hence the induction of autophagy was analyzed by comparing the difference in ATG8 protein levels between Concanamycin A (ConA)-treated and untreated samples at the same time points. Consistent with the imaging results (**Figure 1**), the levels of ATG8s were significantly induced at 0.5 and 1 h of NaCl plus ConA treatment (**Figure 2**).

Since the autophagic flux can also be detected by measuring NBR1 degradation (Mizushima and Yoshimori, 2007), changes in NBR1 levels following salt stress were monitored. The clear reduction in NBR1 protein level at 0.5 h of salt stress, in combination with the relative constant levels of NBR1 at the same time points in the presence of ConA (**Figure 2**), indicated that NBR1-dependent, selective autophagy peaked at 0.5 h following NaCl treatment.

Salt-induced ATG8 flux, represented by an elevation in ATG8 levels following NaCl treatment and an even higher induction in NaCl plus ConA treatment, was not observed in atg5 or atg7 mutants (**Supplementary Figure S4**). ATG8 levels were induced in atg2 and atg9 only at 0.5 h (**Supplementary Figure S4**). NBR1 degradation was observed at 0.5 and 1 h in atg9, and at 6 h in atg2, however, not in atg5 or atg7 (**Supplementary Figure S4**).

#### Oxidized Protein Levels Are Transiently Reduced after NaCl Treatment in the Wild-Type

Oxidized proteins induced by ROS are known substrates for autophagy (Xiong et al., 2007). To see whether autophagy may contribute to the clearance of oxidized proteins generated upon salt stress, levels of oxidized proteins were analyzed over a time-course of 6 h. In the WT, oxidized proteins accumulated significantly within 0.5 h of salt treatment, got back to the control level at 3 h, before returning to the 0.5–1 h level by 6 h. atg9 performed better than the WT in the clearance of oxidized proteins, with a reduction clearly observed at 1 h. atg2 and atg7 both had higher-than-WT levels of oxidized proteins before stress, which then got further induced by salt stress. Only at 6 h did the levels of oxidized proteins drop slightly in atg2 and atg7. Overall, the results indicate that autophagy could have a positive role in transiently reducing the level of oxidized proteins generated by salt stress (**Figure 3**).

### The Autophagy Mutants Are Hypersensitive to Salt Stress during Germination

To see if autophagy is positively involved in salt stress adaptation, germination percentage of the autophagy mutants and ATG8-OX plants were documented in a time course on control and NaCl-containing plates. Under controlled condition, ATG8-OX

<sup>∗</sup>p < 0.05. Bar = standard error.

germinated significantly faster than the wild-type (p < 0.01, paired student's t-test), whereas the autophagy mutants were statistically similar to the wild-type (**Figure 4A**). In the presence of NaCl, germination was slowed down in all lines, especially at higher concentrations (**Figures 4B–D**). The two autophagydeficient lines, atg5 and atg7, germinated significantly slower than the wild-type on all salt stress conditions (**Figures 4B–D**). atg9 and atg2 germinated faster than the wild-type on 50 mM NaCl, but not on higher concentrations (**Figures 4B–D**).

Similar germination curves were observed in the atg mutants on mannitol-containing plates (**Supplementary Figure S5**). These physiological data suggested that the level of autophagy

is positively correlated with the germination rate at salt- and osmotic- stress conditions.

Consistent with the germination phenotypes, root-bending assay performed on 150 mM NaCl showed that the WT and the ATG8-OX lines responded normally to gravity stimulation on vertical plates supplemented with 150 mM NaCl. In contrast, atg5 and atg7 had less bending in their primary roots on NaCl plates (**Supplementary Figure S6**). Altogether, the phenotypes confirmed that autophagy is required for the adaptation of seedlings toward salt stress.

#### Accumulation of Osmolytes Is Correlated with the Level of Autophagy upon Salt Stress

Autophagy may cater to the timely production of saltinduced osmolytes, such as proline and soluble sugars. The

levels of proline, soluble sugar, and reducing sugar were measured in control and salt-stressed (0, 8, and 24 h) seedlings (**Figures 5A–C**). Before salt treatment, ATG8-OX had significantly higher levels of proline and reducing sugar than the wild-type, whereas several atg mutants had significantly lower levels of the osmolytes compared to the wild-type. Following salt treatment, all lines accumulated more osmolytes. ATG8- OX generally had much higher levels of proline and sugars, especially at 24 h (**Figures 5A–C**). In contrast, the atg mutants accumulated significant less osmolytes compared with the wildtype (**Figures 5A–C**), suggesting that the production of proline and soluble sugars indeed relies on autophagy.

#### Expression of Salt-Inducible Genes Was Similar in Autophagy-Reduced and Autophagy-Enhanced Lines

To see if the salt-induced transcription is affected by the level of autophagy, we compared the expression patterns of saltinducible genes in atg5 and atg9 mutants, ATG8-OX, and the wild-type with quantitative RT-PCR (**Figure 6**). The selected genes are known to be induced by salt stress either dependent or independent of ABA (Ma et al., 2006). Since we observed an induction of autophagic flux at 0.5 h of NaCl treatment, two relatively early time points (0.5 and 3 h) were selected for the transcript analysis. Expression of most markers was induced at 0.5 h in the wild-type, atg9, and ATG8-OX, and was further induced at 3 h (**Figure 6**). The expression patterns observed in the wild-type, atg9, and ATG8-OX indicated that autophagy may not affect salt-inducible gene expression directly. The extra high induction in gene expression observed in atg5 (**Figure 6**), however, is unexpected, and might have resulted from a yet unknown regulatory mechanism.

#### Sequestration of Sodium Ions in the Vacuole of Root Cells Is Correlated with Autophagy Levels

To see whether autophagy might play a role in Na<sup>+</sup> compartmentation upon salt stress, sodium ions were visualized with CoroNa Green, a fluorescent dye specific for Na+-imaging, with a same set of scanning parameters on a confocal microscope. The lipophilic dye FM4-64 was used to stain the PM, outlining the cells. Without NaCl, the fluorescence was barely visible (**Supplementary Figures 7**, **8**) as reported (Oh et al., 2010). After 8 h of 100 mM NaCl treatment, striking differences in fluorescent intensity was observed among the lines (**Figure 7** and **Supplementary Figure S7**). As described (Oh et al., 2010), Na<sup>+</sup> accumulated in the vacuoles of root cortex cells in all lines, however, at very different quantities. Compared with the wild-type, very weak signals were detected in atg2, atg5, and atg7, whereas much stronger signals were observed in ATG8-OX (**Figure 7** and **Supplementary Figure S8**). atg9 had lower-than-WT yet clearly visible signals (**Figure 7** and **Supplementary Figure S9**). The observation suggested that sodium compartmentation in the root cortex cells positively correlates with the level of autophagy.

FIGURE 6 | Expression of salt-inducible genes in WT, ATG8-OX, atg5, and atg9 following salt treatment. Ten-day-old seedlings were incubated in liquid 1/2 MS containing 150 mM NaCl for 0 (control), 30 min, and 3 h. Transcript levels of selected salt-inducible genes were profiled and normalized to the wild-type control. <sup>∗</sup>p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, respectively. Red star = compared with WT at time point 0; Black star = compared with the same germplasm at time point 0. Three biological replicates, each consisting of three technical repeats were done, and one representative replicate is shown. Bar = standard error.

### DISCUSSION

#### ProATG8a:GFP-ATG8a as a New Marker Line for Autophagy

In the past decade, several methods have been established for observing autophagy in plants (Klionsky et al., 2016; Pu and Bassham, 2016), including staining acidic vesicles with the fluorescent dye monodansylcadaverine (MDC), electron microscopy of cells, etc. The most widely used method is to observe the presence and distribution of Pro35S:GFP-ATG8s and ProUBQ10:GFP-ATG8s in plant cells (Yoshimoto et al., 2004; Contento et al., 2005; Suttangkakul et al., 2011; Shin et al., 2014). Studies had revealed, however, that ectopic expression of ATG8s could have measurable impact on plant development. The Pro35S:GFP-ATG8 plants generally grow faster to larger sizes (Slavikova et al., 2008; Xia et al., 2012). In this study, we constructed GFP-ATG8a driven by its own promoter, and generated T3 homozygous lines. Line 5-1 had phenotypes indistinguishable from the WT under normal growth conditions, and shared the same starvation- and stress- induced phenotypes with the WT. A bonus is that the T-DNA insertion site in this line had been revealed from TAIL-PCR, so that the marker line can be easily genotyped after crossing with other lines. We also noticed that in this line, the strong nuclear ATG8 signals commonly observed in both Pro35S:GFP-ATG8 and ProUBQ10:GFP-ATG8 under controlled conditions (Zhou et al., 2013; Pei et al., 2014) is absent (**Figure 1** and **Supplementary Figure S2**), indicating that the ATG8a level in this line is closer to the endogenous level.

#### Autophagy Is Rapidly Induced by Salt Stress

An interesting finding in this study is that the autophagic flux peaks as early as 0.5 h upon salt stress in Arabidopsis seedlings (**Figures 1**, **2**). Similarly, it has been reported that autophagic flux was induced at 30 min following hypertonic

stress (350 and 500 mOsmol/kg NaCl, approximately 150 to 250 mM NaCl) on in LLC-PK<sup>1</sup> renal proximal tubule-like cells (Nunes et al., 2013). Such speedy induction of autophagy is more likely based on post-translational modification rather than de novo synthesis of core autophagy proteins. What the possible modifications are remain to be explored. Apart from the possible regulation on ATG8 deacetylation (Huang et al., 2015), a known determinant on ATG8 activity, the redox-controlled ATG4 activity (Perez-Perez et al., 2014, 2016), might also regulate salt-induced autophagy. The ATG4 protease not only cleaves nascent ATG8 at its conserved C-terminal Glycine residue to promote conjugation of ATG8 to PE during autophagosome formation, but cleaves the amide bond between ATG8 and PE on completed autophagosome to recycle ATG8 (Yu et al., 2012). The two Arabidopsis ATG4s have been demonstrated to have different substrate preference in vivo and in vitro (Woo et al., 2014), and the activity of Chlamydomonas ATG4 has recently been shown to be inhibited by oxidation at a Cysteine residue (C400) conserved between plants and yeasts (Perez-Perez et al., 2016). It can be postulated that the oxidative stress generated during salt stress may modulate ATG8-PE formation by regulating ATG4 activity.

roots (three biological replicates), were quantified. Bar = 25 µm in (A).

#### Autophagy Is Likely Required for the Efficient Establishment of Salt Tolerance

∗∗∗p < 0.001. Red star = lower than WT; black star = higher than WT in (B).

The germination assay clearly indicates that autophagy plays a positive role in ensuring timely germination upon salt stress treatment. There are at least two explanations for such observation. Firstly, autophagy is known to contribute to nutrient remobilization from source to sink (Guiboileau et al., 2012; Xia et al., 2012), and autophagy mutants had insufficient protein degradation in their rosette leaves (Guiboileau et al., 2013). Therefore, during seed maturation, the acquisition of seed storage proteins, free amino acids, fatty acids, and other macromolecules is likely defective in the autophagy mutants (Galili et al., 2014). When challenged by salt stress, the germinating atg mutants could have insufficient osmolyte production, thus exhibiting sensitivity toward osmotic challenge. In contrast, the ATG8- OX seeds could have benefited from higher levels of storage proteins and other macromolecules during germination. Such differences become clearer as the stress conditions become more severe (**Figure 4** and **Supplementary Figure S5**), especially on the two autophagy deficient mutant, atg5 and atg7. Indeed, a global analysis on etiolated (carbon starvation) autophagy mutant seedlings showed that, amino acids, organic acids, and protein

levels were significantly decreased in atg5 (Avin-Wittenberg et al., 2015). Secondly, autophagy participates in the clearance of damaged organelles such as peroxisomes (Kim et al., 2013; Shibata et al., 2013; Yoshimoto et al., 2014), and salt stress is known to generate such damages. Salt stress also leads to accumulation of protein aggregates and oxidized proteins (Mano et al., 2014), both of which have been reported to be substrates of autophagy (Toyooka et al., 2006; Xiong et al., 2007). On the other hand, autophagy is certainly not the only degradation pathway responsible for the elimination of damaged organlles, oxidized proteins, and protein aggregates. Our observation that the level of oxidized proteins are only transiently reduced in the wildtype indicated that autophagy may not be the major pathway involved in the long-term clearance of oxidized proteins. It has been reported that proteins oxidized in salt-stressed Arabidopsis are versatile not only in their functions but also in their subcellular localizations (Mano et al., 2014). A number of them are predicted to localize to the apoplast, the plasma membrane, and the nucleus, thus are likely to be degraded by other pathways. A recent report showed that in severely salt-stressed tomato roots (250 mM NaCl, 6 h), two of the three catalytic subunits, β2 and β5, of the proteasome are transiently modified to generate the stress proteasome (Kovacs et al., 2017). At 24 h, a normal proteasome profile reappeared. Such a transient apperance of the stress-induced active proteasome coincites with oxidized protein degradation (Kovacs et al., 2017) and is likely a parallel pathway to autophagy in short-term salt stress response.

Why does the autophagic flux, reflected by the turnover of both ATG8 and NBR1 in the lytic vacuole, take place at such an early time point? Considering that salt-induced transcription generally starts at a similar time point, it can be hypothesized that the efficient translation of stress-responsive proteins may rely on a reliable pool of amino acids. Indeed, the quantification of free amino acids following salt stress supported this hypothesis (**Supplementary Figure S9**). At the 8-h time point, the ATG8- OX seedlings had significantly lower levels of free amino acids, indicating that it was able to remobilize more amino acids for protein synthesis at an earlier time point. As a bulk degradation pathway, autophagy is likely a good candidate for maintaining such a pool. One may wonder which organelles or proteins are preferentially "eaten" during salt stress, since various substrates have been discovered in autophagy induced by different abiotic stresses. For instance, upon heat stress, autophagy is required for the clearance of the ubiquitinated protein aggregates (Zhou et al., 2013). Hypoxia-induced autophagy, on the other hand, may use chloroplast proteins as substrates, as the autophagy mutants die quickly following submergence and their detached leaves turn yellow faster upon ethanol treatment (Chen et al., 2015).

Interestingly, autophagy also regulates the compartmentation of sodium ions in root cells, although the underlying mechanism remains uncovered. A model summarizing our findings is presented (**Figure 8**). The interesting fact that autophagy gets induced very rapidly following salt treatment and is required for successful salt adaptation may be explored in molecular breeding of salt tolerant crops in future.

#### AUTHOR CONTRIBUTIONS

QG, LL, and PZ conceived and designed the study. LL, PZ, DW, and QG analyzed the data and drafted the manuscript. LL, PZ, RZ, JF, JZ, JS, and ZW carried out the experiments. All authors have read and approved the final manuscript.

#### FUNDING

This work was supported by the National Natural Science Foundation of China (31401179, 31671419), and in part by the Fundamental Research Funds for the Central Universities and the Funds for National basic science personnel training (J1103503).

#### ACKNOWLEDGMENTS

We apologize to colleagues whose works are not cited due to space limitations. We thank the Arabidopsis Biological Resource

#### REFERENCES


Center (ABRC) for mutant lines; Dr. Fuchen Shi, Dr. Lei Chen, Ms. Ruming Liu, Ms. Lihong Zu, Mr. Jinxin Zheng, Mr. Siyuan Yan, Mr. Sheng Liu, and Mr. Wei Xu for technical assistance, and Dr. Zhiping Xie for discussions. We also thank the reviewers for their helpful suggestions.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.01459/ full#supplementary-material

FIGURE S1 | Illustration of materials used in this study.

FIGURE S2 | ProATG8a:GFP-ATG8a as a marker line for autophagy.

FIGURE S3 | The Anti-GmATG8c antisera preferentially detect the non-lipidated ATG8s.

FIGURE S4 | Autophagy is not induced by salt treatment in autophagy mutants.

FIGURE S5 | Germination of atg mutants, WT, and ATG8-OX on mannitol.

FIGURE S6 | Root bending assay of atg mutants, WT, and ATG8-OX on 150 mM NaCl.

FIGURE S7 | CoroNa Green staining of root tips of ATG8-OX lines.

FIGURE S8 | CoroNa Green staining of root tips without NaCl.

FIGURE S9 | Changes in free amino acid levels in lines with altered autophagy levels upon salt stress.

TABLE S1 | Primers used in this study.


Guiboileau, A., Yoshimoto, K., Soulay, F., Bataille, M. P., Avice, J. C., and Masclaux-Daubresse, C. (2012). Autophagy machinery controls nitrogen remobilization at the whole-plant level under both limiting and ample nitrate conditions in Arabidopsis. New Phytol. 194, 732–740. doi: 10.1111/j.1469-8137.2012.04084.x


Na<sup>+</sup> imaging. J. Neurosci. Methods 155, 251–259. doi: 10.1016/j.jneumeth.2006. 01.009


fpls-08-01459 August 18, 2017 Time: 13:9 # 12

seedlings to hormonal controls and abiotic stresses. J. Exp. Bot. 59, 4029–4043. doi: 10.1093/jxb/ern244


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

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

fpls-08-01459 August 18, 2017 Time: 13:9 # 13

# Proteomic and Physiological Analyses Reveal Putrescine Responses in Roots of Cucumber Stressed by NaCl

Yinghui Yuan<sup>1</sup> , Min Zhong<sup>1</sup> , Sheng Shu<sup>1</sup> , Nanshan Du<sup>1</sup> , Jin Sun1, 2 and Shirong Guo1, 2 \*

*<sup>1</sup> Key Laboratory of Southern Vegetable Crop Genetic Improvement, Ministry of Agriculture, College of Horticulture, Nanjing Agricultural University, Nanjing, China, <sup>2</sup> Suqian Academy of Protected Horticulture, Nanjing Agricultural University, Suqian, China*

Soil salinity is a major environmental constraint that threatens agricultural productivity. Different strategies have been developed to improve crop salt tolerance, among which the effects of polyamines have been well-reported. To gain a better understanding of the cucumber (*Cucumis sativus* L.) responses to NaCl and unravel the underlying mechanism of exogenous putrescine (Put) alleviating salt-induced damage, comparative proteomic analysis was conducted on cucumber roots treated with NaCl, and/or Put for 7 days. The results showed that exogenous Put restored the root growth inhibited by NaCl. Sixty-two differentially expressed proteins implicated in various biological processes were successfully identified by MALDI-TOF/TOF MS. The four largest categories included proteins involved in defense response (24.2%), protein metabolism (24.2%), carbohydrate metabolism (19.4%), and amino acid metabolism (14.5%). Exogenous Put up-regulated most identified proteins involved in carbohydrate metabolism, implying an enhancement in energy generation. Proteins involved in defense response and protein metabolism were differently regulated by Put, which indicated the roles of Put in stress resistance and proteome rearrangement. Put also increased the abundance of proteins involved in amino acid metabolism. Meanwhile, physiological analysis showed that Put could further up-regulated the levels of free amino acids in salt stressed-roots. In addition, Put also improved endogenous polyamines contents by regulating the transcription levels of key enzymes in polyamine metabolism. Taken together, these results suggest that Put may alleviate NaCl-induced growth inhibition through degradation of misfolded/damaged proteins, activation of stress defense, and the promotion of carbohydrate metabolism to generate more energy.

Keywords: Cucumis sativus L., metabolic processes, proteome, putrescine, root, salt stress

### INTRODUCTION

Soil salinity is one of the most serious environmental constraints to farming production, owing to both natural reasons and inappropriate agricultural operations (Munns and Tester, 2008). Although variable strategies existed for dealing with salt stress, most crops could not grow well on saline soil. Salt tolerance is a complex trait, involved in osmotic adjustment, toxic ions exclusion, and compartmentalization, as well as morphological changes (Munns, 2005). Most studies focused

#### Edited by:

*Jie Zhou, Zhejiang University, China*

#### Reviewed by:

*Qaisar Mahmood, COMSATS Institute of Information Technology, Pakistan Qinghua Shi, Shandong Agricultural University, China*

> \*Correspondence: *Shirong Guo srguo@njau.edu.cn*

#### Specialty section:

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

> Received: *19 May 2016* Accepted: *01 July 2016* Published: *15 July 2016*

#### Citation:

*Yuan Y, Zhong M, Shu S, Du N, Sun J and Guo S (2016) Proteomic and Physiological Analyses Reveal Putrescine Responses in Roots of Cucumber Stressed by NaCl. Front. Plant Sci. 7:1035. doi: 10.3389/fpls.2016.01035* on salt-induced responses in the shoot tissues, since minimizing toxic ions accumulation in leaf is essential for plant growth and productivity (Julkowska et al., 2014). However, root is the first organ exposed to salt stress, and sometimes even shows greater growth reduction than shoot (Bernstein et al., 2004; Contreras-Cornejo et al., 2014). Root growth inhibition is the most obvious change caused by salt stress. Arabidopsis root exposed to NaCl exhibited a reduced meristem size and smaller mature cells, thus resulting in primary root growth inhibition (West et al., 2004). Excessive salt ions accumulating in the soil exert osmotic pressure to roots, which leads to the inhibition of water absorption capacity, over-accumulation of reactive oxygen species (ROS; Jiang et al., 2016), and interruption of membranes (Gupta and Huang, 2014). In addition, large amount of Na<sup>+</sup> entering into root cells is always accompanied by sustained leakage of cytosolic K <sup>+</sup>, resulting in severe ion imbalance and metabolic disorders (Witzel et al., 2009).

Polyamines (PAs) are biologically ubiquitous aliphatic amines in all living cells. In plants, PAs have been implicated in various processes about plant growth and development, and play essential roles in abiotic, and biotic stress responses (Alcázar et al., 2010; Gill and Tuteja, 2010). The tetraamine spermine (Spm), triamine spermidine (Spd), and their precursors, diamine putrescine (Put), are the most common PAs studied in plants. Apart from the free forms, PAs in plants also occurred as conjugates with small molecules such as hydroxycinnamic acids and phenolic, and bound to macromolecules like proteins, and nucleic acids (Quinet et al., 2010; Tiburcio et al., 2014). In most plant species, Put is synthesized either from L-Arginine or L-Ornithine by arginine decarboxylase (ADC), or ornithine decarboxylase (ODC; Dalton et al., 2016), and then converted to Spd and Spm by the action of spermidine synthase (SPDS) and spermine synthase (SPMS). S-adenosylmethionine decarboxylase (SAMDC) provides aminopropyl groups for the synthesis of Spd and Spm (Bagni and Tassoni, 2001; Fuell et al., 2010). Polyamines catabolic process is catalyzed by diamine oxidases (DAO) for Put and polyamine oxidases (PAO) for SPD/Spm. Polyamine oxidases also catalyze the back-conversion of Spm to Spd (Moschou et al., 2008). The mutual conversion among endogenous Put, Spd, and Spm makes it difficult to ascertain their individual functions (Mattoo et al., 2010).

Numerous studies using mutants, ansgenic gene lines, or exogenous application demonstrated the positive role of PAs in salt tolerance (Urano et al., 2004; Wi et al., 2006; Yamaguchi et al., 2006; Kamiab et al., 2014). PAs inhibit the opening and induce closure of stomata by regulating voltage-dependent inward K <sup>+</sup> channel (Liu et al., 2000) or H2O<sup>2</sup> signal (Konstantinos et al., 2010), thus restricting transpiration and reducing water loss. PAs also regulated ROS homeostasis during salt stress by activating antioxidant defense system and directly scavenging O − 2 and ·OH (Li et al., 2014; Saha et al., 2015). Interactions of PAs with ion channels especially contributed to the ion homeostasis under salinity conditions (Pottosin et al., 2014). Apart from these physiological changes, PAs are also reported to interact with proteins, the final executants of life activities, to control Plant adaptations. Previous studies reported that PAs could bind to charged spots at protein interfaces and thus regulate protein function by modulating electrostatic protein–protein interactions (Berwanger et al., 2010). PAs covalent binding with proteins by transglutaminase (TGase) helps stabilizing cell structure and function (Campos et al., 2013). Comparative proteome could provide comprehensive analysis of plant responses to environmental stresses regulated by PAs (Witzel et al., 2009). Proteomic analysis conducted on cucumber leaves revealed that increased salt tolerance by Spd was attributed to the increased levels of proteins involved in protein biosynthesis, antioxidant defense reaction, and energy metabolism (Li et al., 2013). Shi et al. (2013) found that pretreatment of Put, Spd, and Spm enhanced salt and drought tolerance of bermudagrass by regulating carbon fixation, electron transport, energy, and defense pathways, and also activiating accumulation of osmolytes. All PAs effectively depressed protein tyrosine nitration and carbonylation by eliminating reactive nitrogen and oxygen species in leaves of citrus exposed to salinity stress (Tanou et al., 2014).

Cucumber (Cucumis sativus L.) as world popular vegetable is sensitive to soil salinity. Studies about exogenous substances such as calcium (He et al., 2012), salicylic acid (Hao et al., 2012), 24-epibrassinolide (An et al., 2016), and Spd (Li et al., 2013) induced proteome alterations in cucumber have been wellreported. Our previous study also reported the photosynthetic associated proteins regulated by Put (Shu et al., 2015). Little information about Put regulating proteomic changes in roots of cucumber exposed to salt stress is available. In this present work, we conducted 2-dimensional gel electrophoresis to analysis the root proteins responding to Put under NaCl stress conditions. A comprehensive analysis among Put regulated proteins and related metabolic processes was then carried out to explore the mechanism of Put in alleviating the damage of salt stress.

## MATERIALS AND METHODS

### Plant Material and Treatment

Cucumber (Cucumis sativus L. cv. Jinyou No. 4) seeds were sown in quartz sand, cultured in a greenhouse, and transplanted to plastic containers at the leaf two stage as previously described (Yuan et al., 2014). After 2 days pre-culture, the seedlings were treated as follows: (a) control, seedlings grown in full-strength Hoagland nutrient solution; (b) Put, seedlings grown in fullstrength Hoagland nutrient solution containing 0.8 mM Put; (c) NaCl, seedlings grown in full-strength Hoagland nutrient solution containing 75 mM NaCl; (d) NaCl + Put, seedlings grown in full-strength Hoagland nutrient solution containing 75 mM NaCl and 0.8 mM Put. The experiment was arranged in a randomized complete block design with three replicates per treatment, that is, each treatment included three containers of total 36 plants. The concentrations of NaCl and Put were selected on the basis of previous experiment (data not shown). All the nutrient solutions were renewed every 2 days.

### Root Growth Determination and Viability Staining

The length of primary root was measured every day to calculate the average increment during the whole treatment period. After 7 days of treatment, cucumber roots were cut off and photographed. Root cell viability was then assessed by fluorescein diacetate (FDA)–propidium iodide (PI) double staining method as described by Bose et al. (2014). Root segments, including the root tip (1 cm), were cut off and stained with 5µg mL−<sup>1</sup> FDA for 3 min followed by 3µg mL−<sup>1</sup> PI for 10 min. Then the stained roots were observed using a Leica DM2500 microscope (Leica Microsystems, Wetzlar, Germany). Excitation wavelengths of 488 and 594 nm were used for FDA and PI imaging, respectively. Images were acquired with a digital camera (Leica DFC495, Leica Microsystems) equipped with a LAS V3.8 (Leica Microsystems) software.

#### Measurement of Free Amino Acids Levels

After 7 days of treatment, root samples were harvested, oven dried, and acid hydrolyzed to analyze free amino acids by an automatic amino acid analyzer (Hitachi L-8900, Tokyo, Japan) according to Norden et al. (2004). Dried samples of root were suspended in 6 M HCl and then hydrolyzed at 110◦C for 24 h. The hydrolyzed samples were centrifuged at 3000 × g for 5 min, and the supernatant was collected and evaporated to dryness. The dried samples were then redissolved in 0.1 M sodium citrate buffer (pH 2.2). The resulting amino acids solution was analyzed with 17 kinds of L-amino acid (sigma) as the standard. Results were determined as mg per 100 mg of dry samples (% DW).

#### Analysis of Endogenous Polyamines

Endogenous polyamines levels were analyzed by highperformance liquid chromatography (HPLC) according to Shu et al. (2015) with small modifications. After 7 days of treatment, root samples were harvested and homogenized in 5% (W/V) cold HClO<sup>4</sup> and incubated on ice for 1 h. After centrifugation for 20 min at 12,000 × g, the supernatant was used to determine free, and conjugated polyamines and the pellet was used to determine bound polyamines. For conjugated and bound polyamines, samples were hydrolyzed at 110◦C for 18 h in 6 M HCl and then evaporated at 70◦C. The residue was re-suspended in 5% HClO<sup>4</sup> and used to measure polyamines contents with the non-hydrolyzed supernatant. The resulting solutions were mixed with 2 M NaOH and benzoyl chloride, and vortexed vigorously.

FIGURE 1 | Effects of exogenous putrescine (Put) on root morphology (A), average growth rate (B), and root viability (C) of cucumber seedlings in hydroponics with or without 75 mM NaCl for 7 d. Each histogram represents a mean ± *SE* of six independent experiments (*n* = 6). Different letters indicate significant differences between treatments (*P* < 0.05) according to Duncan's multiple range tests. Control, seedlings cultured in normal nutrient solution; Put, seedlings cultured in nutrient solution with 0.8 mM Put; NaCl, seedlings cultured in nutrient solution supplemented with 75 mM NaCl; NaCl + Put: Both NaCl and Put added to the nutrient solution.

After incubation for 30 min at 37◦C, saturated NaCl solution was added to terminate the reaction. The benzoyl polyamines were extracted with cold diethyl ether. Then the diethyl ether phase was evaporated to dryness and re-dissolved in 64% (v/v) cold methanol. Finally, polyamines were assayed by HPLC 1200 series system (Agilent Technologies, Santa Clara, CA) with a C18 reverse phase column (4.6 by 250 mm, 5µm Kromasil) and a two solvent system including a methanol gradient (36–64%, v/v) at a flow rate of 0.8 mL min−<sup>1</sup> . Standard Put, Spd, and Spm (Sigma) were treated in similar way.

#### Protein Extraction and 2-Dimensional Gel Electrophoresis (2D) Analysis

After 7 days of treatment, total root proteins were extracted using a trichloroacetic acid (TCA)-acetone precipitation method modified from Hurkman and Tanaka (1986). Fresh root samples (2 g) were ground to fine powder with liquid nitrogen and resuspended in 6 mL of ice-cold extraction buffer which contains 20 mM Tris-HCl (pH 7.5), 1 mM ethylenebis(oxyethylenenitrilo)tetraacetic acid (EGTA), 1 mM dithiothreitol (DTT), and 1 mM phenylmethyl sulfonyl fluoride (PMSF). After standing on ice for 20 min, the homogenate was centrifuged at 15,000 × g at 4◦C for 20 min. The resulting supernatant was transferred into a new centrifuge tube and precipitated with five volumes of ice-cold acetone containing 10% TCA and 0.07% β-mercaptoethanol at −20◦C for at least 4 h. The resulting protein-containing suspension was centrifuged at 20,000 × g for 25 min. The protein pellet was washed three times with acetone containing 0.07% β-mercaptoethanol at −20◦C for 2 h. The protein sample was air-dried and

FIGURE 2 | Representative 2-DE gel images from root samples treated with NaCl and/or putrescine (Put). Total proteins were extracted and separated by IEF/SDS-PAGE then stained with Coomassie Brilliant Blue (R-250). An equal amount (800µg) of total proteins was loaded onto 18 cm gel strip (pH 4–7, linear). The pI and molecular mass standards are indicated on the top and left side of each gel image. Sixty two differentially expressed protein spots are marked with arrows and numbers, and annotated according to the numbering in Table 1. Control, seedlings cultured in normal nutrient solution; Put, seedlings cultured in nutrient solution with 0.8 mM Put; NaCl, seedlings cultured in nutrient solution supplemented with 75 mM NaCl; NaCl+Put: Both NaCl and Put added to the nutrient solution.


(Put).




 

*bNumber of identified peptides.*

*cPercentage of sequence coverage by matched peptides. dThe*

 *values higher than 1.5 or lower than 0.67 indicate significant changes.*

rehydrated in a rehydration buffer consisted of 7 M urea, 2 M thiourea, 4% 3-[(3-cholanidopropyl) dimethylammonio]-1 propanesulfonic acid (w/v), 40 mM DTT, 0.5% (v/v) immobilized pH gradient (IPG) buffer 4–7, and 0.01% (w/v) bromophenol blue. Protein concentrations were quantified using the Bradford method (Bradford, 1976). Bovine serum albumin (BSA) was used as the standard.

Isoelectric focusing (IEF) was performed with pH 4–7, 18 cm IPG linear gradient strips (GE Healthcare, USA). IPG strips were loaded with 350µL of protein sample containing 800µg protein in a rehydration tray for 12–16 h at 25◦C. After rehydration, IEF was accomplished at 20◦C on an Ettan IPGphor 3 (GE Healthcare, USA) with the following conditions: 100 V for 1 h, followed by 200 V for 1 h, 200 V for 1 h, 500 V for 1 h, 1000 V for 1 h, 4000 V for 1 h, a gradient of 10,000 V for 1 h, and then 10,000 V rapid focus, reaching a total of 75,000 V h. The electric current during IEF was no more than 50µA per strip.

After running the first dimension, IEF strips were equilibrated in 2D equilibrium buffer [50 mM Tris–HCl, pH 8.8, 6 M urea, 30% glycerol (v/v), 2% sodium dodecyl sulfate (SDS)] containing 1% DTT for 15 min and then in the same equilibrium buffer containing 2.5% iodoacetamide for 15 min. Then the equilibrated strips were placed directly onto 12.5% polyacrylamide-SDS slab gels and sealed with 0.5% agarose solution containing bromophenol blue dye. The second dimensional SDS-polyacrylamide gel electrophoresis (SDS-PAGE) was conducted using the EttanDaltSix electrophoresis system (GE Healthcare, USA). Electrophoresis was carried out at 15 W per gel until the bromophenol blue dye front reached about 1 cm from the bottom of the gel. The resulting gels were stained with Coomassie Brilliant Blue (CBB) R-250 to visualize protein spots.

#### Image and Data Analysis

After de-staining, images of CBB-stained 2-D gels were obtained using an Image scanner III (GE Healthcare, USA) and analyzed with Imagemaster 2D Platinum version 5.0 (GE Healthcare, USA). The abundance of each protein spot was estimated by the percentage volume (vol%), which was normalized as the ratio of the volume of a single spot to the whole set of spots present in the gel. Only spots with significant (Duncan's multiple range test at the P < 0.05 level) and reproducible changes (at least 1.5 folds change in abundance) in three replicates were used for mass spectrometry.

#### Protein Identification

Differentially expressed protein spots were excised from gels and in-gel protein digestion was carried out as described by He et al. (2012). The resulting peptides were used for MALDI-TOF/TOF analysis with an ABI 5800 Proteomics Analyzer MALDI-TOF/TOF analyzer (Applied Biosystems, Foster City, CA, USA). Spectral data were used to search NCBI (http://www.ncbi.nlm. nih.gov/) and cucumber genomics database (http://cucumber. genomics.org.cn) with the software MASCOT version 2.2 (Matrix Science, London, UK). MS + MS/MS spectra search criteria in the databases were: trypsin as enzyme, one missed cleavage site, carbamidomethyl set as fixed modification, methionine oxidation allowed as dynamical modification, peptide mass tolerance within 100 ppm, the fragment tolerance set to ± 0.4 Da, and minimum ion score confidence interval for MS + MS/MS data was set to 95%.

### Protein Functional Classification

The differentially expressed proteins were classified according to the biological processes in which they are involved based on UniProtKB (http://www.uniprot.org/) and KEGG (http://www. genome.jp/kegg/) databases. Hierarchical clustering of protein expression patterns was performed using Cluster software version 3.0. Input data was calculated by dividing spot abundance at NaCl and/or Put treatment by abundance of the same protein spot at control and log<sup>2</sup> transformed. The resulting heat map was visualized by Java Treeview.

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

Root tissues were harvested after 1, 3, 5, and 7 days of treatment and extracted for total RNA using Trizol reagent (Takara, Otsu, Japan). A first-strand cDNA fragment was then synthesized using a SuperScript First-strand Synthesis System for qRT-PCR. qRT-PCR was performed according to the instructions provided by the SYBR <sup>R</sup> Premix Ex TaqTM II (Tli RNaseH Plus) kit (Takara). A 20µL reaction mixture containing 2µL of cDNA (diluted 1:5), 10µL of SYBR <sup>R</sup> Premix ExTaqTM II (2×), 0.8µL of each specific primer, 0.4µL of ROX reference dye, and 6µL of ddH2O was run in triplicate on a StepOnePlusTM Real-Time PCR System (Applied Biosystems). Thermal cycling was initialized at 95◦C for 5 min, followed by 40 cycles of 95◦C for 15 s, 60◦C for 1 min, and a final extension of 15 s at 95◦C. Relative expression levels were calculated using the 2−11C<sup>T</sup> method, where the CTvalues obtained from the amplification plots for tested genes were normalized against that of the housekeeping gene (actin) and compared with the control. Base sequences of these tested genes were identified by searching the NCBI and cucumber genomics database. The gene specific primers were designed using Beacon Designer 7 (Premier Biosoft International, CA, USA) and all primer sequences were listed in Supplementary Table 1.

#### Statistical Analysis

All data were statistically analyzed using SAS 13.0 software (SAS Institute, Inc., Cary, NC, USA) by Duncan's multiple range test at P < 0.05 level of significance.

### RESULTS

#### Root Growth

The effect of exogenous Put on root growth of cucumber seedlings exposed to NaCl stress was investigated (**Figure 1**). Seventy-five millimolars of NaCl treatment for 7 days seriously decreased cucumber root growth, and cucumber root gradually turned flaccid, filemot, and transparent (**Figure 1A**). Application of exogenous Put effectively alleviated the growth inhibition induced by NaCl stress, showing a 1.8-folds up-regulation in the average growth rate, but exerted no significant effect on roots of control plants (**Figure 1B**). Viability staining experiments

also proved that Put reduced the root cells death under NaCl stress (**Figure 1C**), thus maintained the absorption of water and nutrients.

#### Comparative Proteomic Analysis

A comparative analysis of proteome was conducted using root samples after 7 days of treatment to investigate the profiles of NaCl and/or Put-responsive, differentially expressed proteins. Approximately 400 reproducible protein spots were detected on the 2-DE gels (**Figure 2**; Supplementary Figure 1), of which 62 differentially expressed protein spots (changes ≥1.5 folds) were successfully identified by MALDI-TOF/TOF MS. These differentially expressed proteins were listed in **Table 1**.

All proteins respond to Put and/or NaCl were categorized into different classes according to their biological process (**Figure 3A**), among which one protein (spot 37) had no functional annotations in the database. The categories with a high level of expression variation are those involved in protein metabolism (24.2%), defense response (24.2%), carbohydrate metabolism (19.4%), and animo acid metabolism (14.5%). The number and overlapping of differentially expressed proteins were summarized in **Figures 3B,C**. Totally, 53 protein spots were significantly regulated by NaCl when compared with control (**Figure 3B**). Of those, 33 protein spots were upregulated in abundance, and 20 protein spots were downregulated. Most identified proteins involved in carbohydrate metabolism, protein metabolism, and defense response were upregulated and most identified proteins involved in amino acid metabolism, fatty acid metabolism, and secondary metabolism were down-regulated (**Figure 3C**). However, 54 protein spots were significantly regulated by Put under control or NaCl stress conditions, of which 27 protein spots were regulated by Put under both control and NaCl stress conditions (**Figure 3B**). Put upregulated the expression of most identified proteins involved in different biological processes, while also decreased the abundance of several proteins involved in defense response and protein metabolism (**Figure 3C**).

In order to identify proteins with similar expression patterns, hierarchical clustering was performed (**Figure 4**). Cluster A was composed of 13 proteins that were up-regulated under both NaCl and NaCl+Put treatment as compared with those of the control. Most of these proteins were involved in defense response. Cluster B involved 15 proteins that were down-regulated by Put both under the control and NaCl stress conditions, and most of these proteins were corresponding to defense response and protein metabolism. Cluster C included 19 proteins that were down-regulated by NaCl, but recovered by the application of Put. Most identified proteins participating in amino acid metabolism were included in this cluster. Proteins involved in carbohydrate metabolism, especially tricarboxylic acid cycle, were also present in this cluster. Cluster D contained 15 proteins with increased abundance by Put under both the control and NaCl stress conditions. Most identified proteins functional in glycolysis (part of carbohydrate metabolism) were included in this cluster.

#### Free Amino Acids Contents

Many proteins related to protein and amino acid metabolism were identified in this study. Considering that amino acid is the raw material for protein synthesis, we measured the free amino acid levels under different treatment (**Table 2**). Glutamic acid (Glu) was the most abundant amino acid in cucumber root, occupied over 2% of dry weight, while cysteine, the least amino acid identified, occupied <0.1% of root dry weight. Except for proline, all the other amino acids contents were increased by NaCl stress, leading to 15.2% increase in total amino acids level as compared with that of the control. Exogenous Put increased all amino acids levels in control plants, and total amino acids level was 1.23-fold higher than the control. Put applied to NaCl stressed plants further increased amino acids levels except for cysteine. Exogenous Put caused a higher increase in amino acids levels than NaCl, which reflected a promotion in amino acid metabolism and protein hydrolysis.

#### Endogenous Polyamines Metabolism

In response to NaCl stress, free, conjugated, and bound Put and Spm contents in cucumber roots were increased in comparison with those of the control (**Table 3**). Free Spd level was also increased by NaCl, whereas conjugated and bound forms of Spd were decreased significantly. Treatment with Put under NaCl stress condition caused increase in almost all kinds of PAs when compared with NaCl-only treatment, except for the bound Spm, which was decreased by Put. Exogenous Put also increased free PAs levels in control plants. Generally, NaCl stress caused increase in total Put and Spm levels, but decreased Spd level. Put treatment increased total Put and Spd contents both under control and stressed conditions, but decrease total Spm level stressed by NaCl.

Expressions of seven genes encoding enzymes involved in PAs metabolism were then analyzed after 1, 3, 5, and 7 days of treatment. NaCl stress up-regulated all tested genes involved in PAs biosynthesis (**Figures 5A–E**). PAO expression was also increased by NaCl (**Figure 5G**), while DAO was first increased and then decreased after 5 days of NaCl treatment (**Figure 5F**). Put applied to NaCl-stressed plants down-regulated the expression of ADC, ODC, SAMDC, SPDS, and SPMS induced by NaCl, but increased DAO expression. After 1 day of treatment, Put up-regulated the expression of ADC, SAMDC, SPDS, and DAO, but down-regulated SPMS and PAO expression in the control plants. In fact, Put decreased the expression of SPMS and increased DAO expression in the control plants during the whole treatment period.

### DISCUSSION

Multiple publications showed that polyamines can increase plants tolerance to various environmental stresses (Urano et al., 2004; Yamaguchi et al., 2006). In the present study, Put was found to restore cucumber root growth and root cell vitality decreased by NaCl stress (**Figure 1**). 2-DE analysis was conducted to obtain the profiling of the total proteins in cucumber roots responding to Put under NaCl stress conditions (**Figure 2**, **Table 1**). After quantitative and MALDI-TOF/TOF MS analysis, 62 differentially regulated protein spots were successfully identified. Proteins involved in protein metabolism, and defense response took up the highest percentage of differentially expressed proteins. Twelve carbohydrate metabolism related proteins and nine amino acid metabolism related proteins made the third and fourth group of identified proteins. In addition, several proteins participated in other metabolic processes were also identified. The regulation of Put and NaCl to these metabolic processes is discussed below.

#### Putrescine Regulates Protein Metabolism

Previous studies in bacteria provided direct evidence for the function of polyamines in protein synthesis (Algranati et al., 1977). Eukaryotic translation initiation factor 3 subunit J (eIF3J, spot 35) and elongation factor 2 (EF2, spot 23) are involved

normal nutrient solution; Put, seedlings cultured in nutrient solution with 0.8 mM Put; NaCl, seedlings cultured in nutrient solution supplemented with 75 mM NaCl; NaCl+Put: Both NaCl and Put added to the nutrient solution). Each row represents individual protein spot with spot number and protein name labeled on the right side. Color scales of log2 values are shown, in which red and green show the higher and lower expression levels, respectively. And gray indicated protein missing under specific treatment.

in the initiation and elongation stage of mRNA translation and protein synthesis (Kanhema et al., 2006). Put maintained the high level of eIF3J abundance induced by NaCl, and further increased EF2 expression in NaCl stressed conditions (**Table 1**, **Figure 2**), which proved the role of Put in protein synthesis. However, Put decreased the abundance of 60S acidic ribosomal protein P2-4 (RPP2D, spot 46), which is part of large subunit of ribosomes. This may be related to the function of Put in balancing ribosomal particles and protein synthesis, which was reported in E. coli by Algranati et al. (1977). Stimulating the synthesis of specific proteins by Put might play an important role in coping with salt stress.

#### TABLE 2 | Effect of exogenous putrescine (Put) on free amino acid contents (%DW) in roots of cucumber seedlings exposed to 75 mM NaCl for 7 d.


*Each value is the mean* ± *SE of three independent experiments (n* = *3). Different letters indicate significant differences between treatments (P* < *0.05) according to Duncan's multiple range tests. Control, seedlings cultured in normal nutrient solution; Put, seedlings cultured in nutrient solution with 0.8 mM Put; NaCl, seedlings cultured in nutrient solution supplemented with 75 mM NaCl; NaCl*+*Put: Both NaCl and Put added to the nutrient solution.*

Salt stress leads to protein unfolding or misfolding, which affects protein conformation and function, and causes metabolic disruption (Kumari et al., 2015). In this study, five spots were identified as protein chaperones (**Table 1**). Three out of the five spots represented heat shock 70 kDa protein (HSP70, spots 3, 9, and 11). HSP70 can be induced by various environmental stresses (Tomanek and Sanford, 2003), and this was further demonstrated in our study by 3.22- and 1.69 folds increases in abundance (spots 9 and 11) under NaCl stress as compared with that of the control. However, Put significantly decreased HSP70 (spots 9 and 11) expression both under control and NaCl stress conditions (**Table 1**). This result was consistent with Li et al. (2013), who found that Spd could down-regulated the expression of HSP70. Polyamines were reported to influence the DNA binding capacity of heat shock transcriptional factor HSF in rat cells and thus affecting the accumulation of HSP70 mRNA (Desiderio et al., 1999). Königshofer and Lechner (2002) reported that under heat stress conditions, polyamine metabolic status in tobacco, or alfalfa cells influence the HSP synthesis by affecting cell membranes integrity and properties. Mitochondrial chaperonin CPN60-2 (spot 4) belongs to HSP60 family, implicated in mitochondrial protein import, and macromolecular assembly (Tsugeki et al., 1992). In cucumber leaf, CPN60-2 was not affect by NaCl stress, but down-regulated by exogenous Spd (Li et al., 2013). In the present study, CPN60-2 in cucumber root was not changed under NaCl stress condition either, but significantly up-regulated by Put. Put also increased the abundance of protein disulfide-isomerase (PDI, spot 12), which catalyzes the formation, breakage or rearrangement of disulfide bonds during proteins folding (Gruber et al., 2006). Hence, Put may regulate chaperones to facilitate protein folding and prevent unfolding/misfolding-induced protein aggregation, thus reestablishing normal protein conformation and maintaining cellular metabolism when exposed to salt stress.

Votyakova et al. (1999) reported that proteins covalently attached with PAs became more resistant to proteolysis. Wajnberg and Fagan (1989) demonstrated that polyamines inhibited the ATP-dependent degradation of ubiquitinated proteins. Ubiquitin-conjugating enzyme E2 2 (UBC2, spot 42) participates in protein modification by catalyzing covalent attachment of ubiquitin to proteins. This process is involved in ubiquitin-mediated protein degradation by the 26S proteasome (Cui et al., 2012). In accordance, Put decreased the abundance of UBC2 under control and NaCl stress conditions (**Table 1**). This result suggested that Put not only targeting energy consumption (Wajnberg and Fagan, 1989) but also the ubiquitinated processes to inhibit protein degradation. However, Put further upregulated the NaCl-induced increase in 26S protease regulatory subunit 6A homolog (RPT5A, spot 13) and proteasome subunit alpha type-5 (PSMA5, spot 4), which were also involved in the degradation of ubiquitinated proteins. Combination with the response in chaperones, Put may accelerate the degradation of misfolded/damaged proteins caused by NaCl to alleviate salt stress induced-damage. In addition, Put regulated protein metabolism may reprogram proteome of cucumber roots in response to salt stress, which is supported by Tanou et al. (2014), thus maintaining cell homeostasis.

FIGURE 6 | Schematic presentation of main metabolic pathways regulated by putrescine (Put) in cucumber root exposed to NaCl. Changes in protein abundance (marked in red ellipse), amino acids contents (yellow), PAs levels (green), and gene expression (blue) were integrated. Arrows on the life side of the ellipses indicate changes induced by NaCl as compared with the control, and arrows on the right side indicate changes induced by Put in NaCl stressed conditions. Red or green arrows represent up-regulation or down-regulation, respectively, and the black short lines indicate no change. APX, ascorbate peroxidase; CPN60-2, chaperonin CPN60-2; DHAR2, glutathione S-transferase DHAR2; EF2, elongation factor 2; eIF3J, eukaryotic translation initiation factor 3 subunit J; FAH, fumarylacetoacetase; FBA, fructose-bisphosphate aldolase; FRK, fructokinase; HSP70, heat shock 70 kDa protein; MDH, malate dehydrogenase; PDI, protein disulfide-isomerase; PGAM, 2,3-bisphosphoglycerate-independent phosphoglycerate mutase; POD, peroxidase; PSMA5, proteasome subunit alpha type-5; RPP2D, 60S acidic ribosomal protein P2-4; RPT5A, 26S protease regulatory subunit 6A homolog; SAMs, S-adenosylmethionine synthase; SOD, superoxide dismutase; TPA, thiol protease aleurain-like; UBC2, ubiquitin-conjugating enzyme E2 2.


TABLE 3 | Effect of exogenous putrescine (Put) on contents of endogenous free, conjugated, and bound PAs in roots of cucumber seedlings exposed to 75 mM NaCl for 7d.

*Each value represents a mean* ± *SE of three independent experiments (n* = *3). Different letters indicate significant differences between treatments (P* < *0.05) according to Duncan's multiple range tests. Control, seedlings cultured in normal nutrient solution; Put, seedlings cultured in nutrient solution with 0.8 mM Put; NaCl, seedlings cultured in nutrient solution supplemented with 75 mM NaCl; NaCl*+*Put: Both NaCl and Put added to the nutrient solution.*

### Putrescine Activates Stress Defense Responses to Alleviate Salt Induced Damage

Plants activate a broad array of defense responses to protect themselves from stress induced injuries. Chitinases are usually considered as pathogenesis-related proteins, and widely distributed in plants. Despite of their role against pathogen stress, chitinases are also implicated in various abiotic stresses (Grover, 2012). In the present work, Put increased chitinase (spot 38) abundance under control conditions (**Table 1**), indicated a promotion in plant defense. Glycine-rich RNA-binding protein (GR-RBP) has a role in RNA transcription or processing during stress by acting as RNA chaperone. GR-RBP7 was also reported to be involved in plant innate immunity (Lee et al., 2012). The increase in GR-RBP3 (spot 43) abundance induced by Put further proved the involvement of PAs in plant immunity. Apart from the plant immunity, redox regulation also represents essential defense response. In the current study, a total of 12 spots (spots 14, 28, 29, 30, 34, 45, 47, 48, 51, 52, 55, and 57) were identified as antioxidant related proteins and mostly showed up-regulated expression when exposed to NaCl stress (**Table 1**). The ability of Put in increasing the activities of antioxidant enzymes, such as superoxide dismutase (SOD), peroxidase (POD), and ascorbate peroxidase (APX), to alleviate oxidative damage caused by environmental stresses has been widely studied (Verma and Mishra, 2005). In this study, Put application to NaCl-stressed seedlings further improved the abundance of SOD (spots 45 and 47), POD (spot 14), APX (spots 28, 29 and 30), and glutathione S-transferases (GST, spot 51), which implied that Put regulated both abundance and activities of antioxidant enzymes to mitigate oxidative damage induced by salt stress. Aldo-keto reductase 4 (AKR, spot 57) belongs to AKR superfamily, which play central roles in response to oxidative stress by detoxifying toxic aldehydes, and ketones to the appropriate alcohol (Hyndman et al., 2003). AKR was also up-regulated by Put (**Table 1**). These results indicated that Put could enhance plant defense responses to reduce stress injuries and gain better growth in NaCl stress conditions.

### Putrescine Promotes Energy Metabolism

Plants struggled in stress conditions need a substantial amount of energy to maintain their growth and development. Mitochondrial respiratory chain, the main provider of energy needed for cellular metabolism, has been reported to be activated by salt stress (Fry et al., 1986). Consistently, the abundance of NAD(P)H dehydrogenase (spot 53) was induced by NaCl in the present study, indicating an activation of respiratory electron transport systems. Carbohydrates metabolism, especially glycolysis (EMP), and tricarboxylic acid (TCA) cycle is also a primary source of energy for plant metabolism (Kumari et al., 2015). In the present study, seven differentially expressed spots were identified as enzymes participated in glycolysis (**Table 1**). Fructokinase-4 (FRK, spot 62), fructose-bisphosphate aldolase (FBA, spot 59), 2,3-bisphosphoglycerate-independent

phosphoglycerate mutase (PGAM, spots 6 and 7), and enolase (spots 1, 2, and 5) were all up-regulated by NaCl, suggesting that roots attempted to generate more energy to cope with salt stress. However, NaCl stress decreased the abundance of malate dehydrogenase (MDH, spots 49, 50, and 56) involved in TCA cycle.

Several studies reported the involvement of PAs in adjusting EMP-TCA metabolism under stressed environment. Jia et al. (2010) found that exogenous Spd enhanced hypoxia stress tolerance by up-regulating activities of enzymes functioning in TCA cycle. Put sprayed to leaves partly reserved the NaCl-reduced activities of phosphofructokinase (PFK), pyruvate kinase (PK), and phosphoenolpyruvate pyruvate kinase (PEPC) involved in glycolysis metabolism, as well as activities of succinate dehydrogenase (SDH), isocitrate dehydrogenase (IDH), and MDH in TCA cycle (Zhong et al., 2015). Here we found that despite of decreased FBA abundance, exogenous Put increased the abundance of FRK, PGAM, and FBA both under control and NaCl stress conditions. What's more, Put partly reserved the MDH expression decreased by NaCl. All of these changes suggested that exogenous Put could promote EMP-TCA pathway activity. In addition to directly regulating these enzymes, Put may also affect TCA cycle through its metabolite, γ-aminobutyric acid (GABA; Gill and Tuteja, 2010). Adjusting EMP-TCA pathway to product more energy may be an important mechanism for Put to alleviate salt stress induced damage.

#### Polyamines and Amino Acid Metabolism Play Key Roles in Salt Tolerance

Amino acid metabolic pathways as well as the protective metabolites produced constitute integral parts of the plant immune system (Zeier, 2013). In this study, nine spots were identified as proteins participated in amino acid metabolism (**Table 1**), while seven of these also related to PAs biosynthesis. Polyamines and amino acid metabolism play key roles in connecting C and N metabolism and several secondary metabolism pathways (Majumdar et al., 2016). We summarized the crossed metabolism connecting PAs, amino acids, and EMP-TCA cycle in **Figure 6**.

NaCl stress increased the soluble amino acid levels, which were further improved by Put (**Table 2**). Free amino acids can serve as osmolytes to regulate osmotic stress caused by salt, and their increases in response to NaCl and/or Put is largely attributed to protein hydrolysis (Yuan et al., 2012). This is further proved by the simultaneous up-regulation of glycine content (**Table 2**) and glycine cleavage system H protein 2 (spot 44) abundance (**Table 1**). Glu is the most abundant amino acid in cucumber root (**Table 2**), and is considered as the center of N metabolism in plants. Glu can be converted into ornithine (Orn), a non-protein amino acid present in low concentration in plants, which is an important precursor for Put (Majumdar et al., 2016). The increase in arginine (Arg) level, another precursor for Put, provided sufficient substrates for the synthesis of Put. In addition, Arg can be converted to Orn and urea via arginase (spot 24), which showed increased abundance under NaCl stress condition but did not respond to Put (**Table 1**). This result indicated that Put production through ODC might be enhanced by NaCl stress. The upregulated expression of ODC also proved this assumption. Glu and Orn contribute to proline (Pro) synthesis by different routes. Salt stress induced Pro accumulation is an important mechanism for osmotic regulation. However, NaCl stress exerted no significant influence on Pro level in this study (**Table 2**). The most probable reason for this result was the substantial synthesis of Put consumed Arg and Orn thus limiting Glu and Orn flow into Pro. External Put inhibited the synthesis of more Put by regulating ADC and ODC expression, and consequently more N was available for Pro production. In this situation, the direct absorption of Put from nutrient solution may contribute large part of the increase in endogenous Put level (Shu et al., 2015).

Methionine (Met) is also related to PAs synthesis through conversion into S-adenosylmethionine (SAM). Sadenosylmethionine synthase (SAMs, spots 15, 16, 17, 19, 21, and 60) catalyzes the biosynthesis of SAM from Met and ATP, and serves as a common precursor for PAs and ethylene (Gong et al., 2014). Transgenic experiments proved that SAMs contributed to stress tolerance by increasing PAs accumulation (Qi et al., 2010; Gong et al., 2014). Put up-regulated SAMs abundance both under control and NaCl stress conditions (**Table 1**), and Spd showed the similar effect (Li et al., 2013). The increase in SAMs was partly responsible for the elevation in endogenous PAs levels (**Table 3**). Although SAMs was significantly reduced by NaCl, PAs levels were still increased due to the transcriptional activation of PAs biosynthesis (**Figure 5**).

The protective roles of PAs are partly attributed to their association with low molecular weight compounds and macromolecules, therefore the conjugated and bound PAs are essential for plant stress tolerance. Quinet et al. (2010) observed that the conjugated PAs pool was positively linked with rice salt tolerance. The conjugated PAs pool was reduced by NaCl in the present study due to a large decrease in conjugated Spd level (**Table 3**). Put application increased the conjugated and bound PAs levels, thus stabilizing the intercellular small molecules and macromolecules to resist salt stress. Exogenous Put induced the accumulation of endogenous Put, also elevated the expression of DAO (**Figure 5F**), which proved the transcriptional activation reported by Quinet et al. (2010). Interestingly, the oxidation of Put is accompanied by the production of H2O2, and GABA, and the latter is key component connecting PAs and amino acid metabolism (Majumdar et al., 2016). Another protein, fumarylacetoacetase (FAH, spot 61), is involved in the multiple steps synthesis of acetoacetate and fumarate from L-phenylalanine or Ltyrosine (Herrera et al., 2010). Fumarate happened to be an intermediate of TCA cycle (**Figure 6**). Put reversed NaCl reduced FAH abundance is an indirect evidence for its regulation to TCA cycle.

Apart from the proteins mentioned above, Put was also found to regulate proteins involved in other metabolic pathways like lipid metabolism (spots 25 and 26) and nitrile formation (spots 20 and 27), indicating Put regulated various metabolic pathways to counteract salt stress-induced changes. Moreover, the identification of proteins related to cell division (spot 18) and cell cytoskeleton (spots 22, 31, and 58) proved the ability of Put to improve plant growth. We proposed a model of the NaCl stress tolerance conferred by external Put in cucumber seedlings (**Figure 7**). The application of Put to NaCl-stressed seedlings results in a series of responses from physiological level to metabolic level. Put enhanced stress defense capacity, regulated carbon and nitrogen metabolism to supply substrates and energy for various metabolic processes, and affected cell growth. All these changes contribute to the Put induced root growth promotion and salt tolerance.

### CONCLUSION

In conclusion, this work demonstrated that Put improving cucumber root growth and salt tolerance could be associated with the following points: (1) stimulating protein synthesis and degrading misfolded/damaged proteins induced by NaCl; (2) activation of stress defense response to alleviate salt induced injuries; and (3) providing more energy for various metabolic processes by up-regulating proteins involved in carbohydrate metabolism. This study provides comprehensive insights into the cell metabolism regulated by NaCl and/or Put, and would be able to better enrich our understanding of the mechanism by which Put improves the salt tolerance of cucumber seedlings.

### AUTHOR CONTRIBUTIONS

SG designed the study and guided the research. YY wrote the main manuscript text and performed the experiments. MZ and ND prepared all the figures and performed some of the experiments. SS and JS modified this manuscript. All authors reviewed and approved the manuscript.

## FUNDING

This work was supported financially by the National Natural Science Foundation of China (31401919, 31471869, and 31272209), the China Earmarked Fund for Modern Agroindustry Technology Research System (CARS-25-C-03), the Priority Academic Program Development (PAPD) of Jiangsu Higher Education Institutions, and the Research Fund for the Doctoral Program of Higher Education (20130097120015).

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01035


**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 © 2016 Yuan, Zhong, Shu, Du, Sun and Guo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparative Proteomic Analysis Provides Insight into the Key Proteins Involved in Cucumber (Cucumis sativus L.) Adventitious Root Emergence under Waterlogging Stress

#### Xuewen Xu, Jing Ji, Xiaotian Ma, Qiang Xu, Xiaohua Qi and Xuehao Chen\*

*Department of Horticulture, School of Horticulture and Plant Protection, Yangzhou University, Yangzhou, China*

#### Edited by:

*Wei Wang, Henan Agricultural University, China*

#### Reviewed by:

*Abu Hena Mostafa Kamal, University of Texas at Arlington, USA GuoJun Yu, National Cancer Institute, USA*

> \*Correspondence: *Xuehao Chen xhchen@yzu.edu.cn*

#### Specialty section:

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

Received: *12 August 2016* Accepted: *26 September 2016* Published: *13 October 2016*

#### Citation:

*Xu X, Ji J, Ma X, Xu Q, Qi X and Chen X (2016) Comparative Proteomic Analysis Provides Insight into the Key Proteins Involved in Cucumber (Cucumis sativus L.) Adventitious Root Emergence under Waterlogging Stress. Front. Plant Sci. 7:1515. doi: 10.3389/fpls.2016.01515* Waterlogging is a common abiotic stress in both natural and agricultural systems, and it primarily affects plant growth by the slow oxygen diffusion in water. To sustain root function in the hypoxic environment, a key adaptation for waterlogging tolerant plants is the formation of adventitious roots (ARs). We found that cucumber waterlogging tolerant line Zaoer-N seedlings adapt to waterlogging stress by developing a larger number of ARs in hypocotyls, while almost no AR is generated in sensitive line Pepino. To understand the molecular mechanisms underlying AR emergence, the iTRAQ-based quantitative proteomics approach was employed to map the proteomes of hypocotyls cells of the Zaoer-N and Pepino under control and waterlogging conditions. A total of 5508 proteins were identified and 146 were differentially regulated proteins (DRPs), of which 47 and 56 DRPs were specific to tolerant and sensitive line, respectively. In the waterlogged Zaoer-N hypocotyls, DRPs related to alcohol dehydrogenases (ADH), 1-aminocyclopropane-1-carboxylicacid oxidases, peroxidases, 60S ribosomal proteins, GSDL esterases/lipases, histone deacetylases, and histone H5 and were strongly overrepresented to manage the energy crisis, promote ethylene release, minimize oxidative damage, mobilize storage lipids, and stimulate cell division, differentiation and growth. The evaluations of ethylene production, ADH activity, pyruvate decarboxylase (PDC) activity and ethanol production were in good agreement with the proteomic results. qRT-PCR analysis of the corresponding 146 genes further confirmed the accuracy of the observed protein abundance. These findings shed light on the mechanisms underlying waterlogging triggered cucumber ARs emergence, and provided valuable information for the breeding of cucumber with enhanced tolerance to waterlogging.

Keywords: cucumber, waterlogging, hypocotyls, adventitious root, iTRAQ

## INTRODUCTION

Waterlogging is described as the saturation of the soil with water around the plant roots, and constitutes one of the most severe abiotic stresses for plant growth and development (Sairam et al., 2009). Nearly 16% of the fertile areas of our globe annually undergo waterlogging due to excessive rainfall, lack of soil drainage, and irregular topography, which resulting in severe economic loss (Ahsan et al., 2007; Xu et al., 2014). In particular, the availability of oxygen for respiration is blocked in waterlogged organs because of gas diffusion in water is about 10<sup>3</sup> times slower than that in the air (van Veen et al., 2014) To sustain energy supply, it is essential for the waterlogged organs to switch over to anaerobic mode for energy production (Jackson and Colmer, 2005; Xu et al., 2014). However, it is obviously inefficient to provide enough energy via glycolysis and fermentation when waterlogging is prolonged (Bailey-Serres and Voesenek, 2008; van Veen et al., 2014). To survive from longterm waterlogging stress, aerobic respiration for plants must be maintained via oxygen transport to enhance internal oxygen diffusion (Shimamura et al., 2014). Some species adapted to waterlogging stress by faster stem/hypocotyls elongation that enable the shoot to regain contact with the open atmosphere, such as Rumex and rice (Evans, 2004; Jiang et al., 2010).

Another key adaptation to waterlogging is the formation of ARs, which minimize the distance for oxygen diffusion and improve gas diffusivity (Sauter, 2013). ARs usually originated from the waterlogged part of hypocotyls or basal stem region, and such adaptation can replace the deteriorating primary roots (Bailey-Serres et al., 2012; Sauter, 2013) Therefore, the adaptive responses of ARs formation to waterlogging might be more important than those of primary roots for survival in waterlogged soil (Li et al., 2009; Yamauchi et al., 2014). Genetic, molecular, and physiological approaches have confirmed that auxin, ethylene and carbohydrate status can affect ARs formation (Changdee et al., 2009). A transgenic Arabidopsis line overexpressing the auxin response factor (ARF) 17 developed fewer ARs than wild-type plants, confirming the potential role of ARF genes in the regulation of ARs development by auxin (Sorin et al., 2005). Taramino et al. (2007) reported that the defective initiation of maize ARs in the rootless concerning crown and seminal roots mutant was caused by a mutation in the gene encoding the Lateral Oateral Boundaries (LOB) domain protein. Expression of the LOB gene was rapidly induced by the application of auxin, and the LOB protein is thought to be the direct targets of ARF transcription factor in the auxin signaling pathway. Rigal et al. (2012) have showed that overexpression of AINTEGUMENTA LIKE 1 (PtAIL1), a transcription factor belongs to APETALA2/ETHYLENE RESPONSE FACTOR family, increased number of ARs in Populus, while RNA interference lines with a reduced level of PtAIL1 transcripts had fewer ARs. However, the underlying mechanisms by which ARs formation is affected have not fully understood, and the reported findings have been contradictory. For examples, in flood-induced ARs of rice stem nodes, it is ethylene but not auxin that signals activation of the cell cycle (Lorbiecke and Sauter, 1999), which is followed by generation of ROS as measured with electron paramagnetic resonance spectroscopy (Voesenek and Bailey-Serres, 2015). Vidoz et al. (2010) found that there may be interaction between ethylene and auxin with respect to AR production in flooded tomato hypocotyls using ethylene- and auxin-insensitive mutants. These contradictory findings may be due to variation in the different plant species, growth conditions, and methods of quantifying ARs.

Cucumber is an economically important vegetable crop and is widely grown in the world with total harvest of more than two million hectares in 2016, ranking 4th in quantity of world vegetable production (FAO STAT 2016, http://faostat3.fao.org). Cucumber is also known for its sensitivity to waterlogging and easily affected by heavy rain or soil waterlogging due to its shallow root system and strict oxygen requirement (Qi et al., 2012; Xu et al., 2014, 2016). The relative tolerance of different cucumber accessions to waterlogging stress has been evaluated in our previous study (Qi et al., 2011). In greenhouse waterlogging experiment, we found that the waterlogging tolerant line Zaoer-N adapt to waterlogging treatment by developing numerous ARs in hypocotyls, while almost no ARs were generated in the sensitive line Pepino. As we known, proteomic applications provide a powerful tool for understanding plant proteome changes in response to abiotic and biotic stress (Chen and Harmon, 2006; Yang et al., 2013; Cui et al., 2015). In recent years, the iTRAQ-based quantitative proteomic approach has been widely used for the comparative analyses of proteome changes because it allows for the simultaneous identification and quantification of peptides by measuring the peak intensities of reporter ions with MS/MS (Yu et al., 2015). In the present study, an iTRAQ-based quantitative proteomics approach was employed to elucidate the detail effects of waterlogging stress on the protein expression levels in hypocotyls of the two contrasting cucumber lines differing in waterlogging tolerance ability and the capacity of ARs formation. The results will provide new insights into the physiological and molecular mechanisms associated with waterlogging stress in hypocotyls cells of cucumber seedling. To the best of knowledge, this is the first frame work that reveals the molecular mechanisms underlying AR emergence upon waterlogging on cucumber lines. Our findings provide a list of potential candidates for further elucidating waterlogging tolerance in plants.

## MATERIALS AND METHODS

## Plant Growth and Treatments

Waterlogging tolerant line Zaoer-N and sensitive line Pepino were grown in 8-cm-wide pots containing peat, vermiculite, and perlite (3:1:1, v/v/v) in a greenhouse at 28/20◦C (14/10 h) day/night temperature and a relative humidity ranging from 70 to 85%. Cucumber seedlings at the three-leaf stage (21 days after germination) were placed in plastic cups filled with water (pH 7.03, 25◦C, electrical conductivity 0.34 dS m−<sup>1</sup> , dissolved oxygen level 7.17 mg/L) to the base of the first true leaves (∼4 cm above the soil surface, soil redox potentials Eh 272 ± 4.5 mV). Water was maintained at this level for the duration of the treatment. For control treatments, plants were watered daily and allowed to drain freely. Hypocotyls (below the water surface) were collected from the Zaoer-N and Pepino seedlings after 2 days of waterlogging and control for physiological assay. Thirty seedling hypocotyls from the three replicate were pooled, immediately frozen in liquid nitrogen, and then stored at −80◦C for iTRAQ analysis and RNA extraction.

#### Protein Extraction and iTRAQ Analysis

Cucumber hypocotyls were grinded to a fine powder in liquid nitrogen and then transferred to 700µl extraction solution (8 M urea, 4% CHAPS, 2 M thiourea, 40 mM Tris/HCl, pH 8.5, 2 mM EDTA, 1 mM PMSF, 10 mM dithiothreitol). The mix was sonicated for 15 min, and then at 4◦C centrifuged 25,000 g for 20 min. The supernatants were removed to a new tube and mixed well with 5 volumes of pre-chilled acetone incubated at −20◦C for 2 h. After centrifugation, the resulting pellet was washed three times with cold acetone. The air-dried pellet was dissolved in the 250µl of 0.5 M TEAB (Applied Biosystems, Milan, Italy), and sonicated for 15 min. The centrifugation step was repeated, and the supernatant was transferred to a new tube and protein concentration was quantified by the 2-D Quant kit (GE Healthcare) according to the manufacturer's protocol.

Hundred microliter of proteins for each sample were incubated with trypsin (Promega, Madison, WI, USA) at a ratio of protein: Trypsin = 20:1 overnight at 37◦C. After digestion, the peptides were dried by vacuum centrifugation, reconstituted in 0.5 M TEAB. Labeling was performed according to the manufacturer's manual for the iTRAQ (Applied Biosystems, Foster City, CA USA) (113 tags for Zaoer-N control, 114 tags for Zaoer-N waterlogging treatment, 117 tags for Pepino control, 118 tags for Pepino waterlogging treatment). One unit of iTRAQ reagent (defined as the amount of reagent required to label 100µg of protein) was thawed and reconstituted in 70µL isopropanol. The labeled peptide mixtures were incubated for 2 h at room temperature and then dried by vacuum centrifugation. Peptides were purified using an SCX chromatography (LC-20AB HPLC pump system, Shimadzu Corporation, Kyoto, Japan). The fractions were then loaded into a Shimadzu LC-20AD nano HPLC by the auto sampler onto a 2 cm C18 trap column. Peptides were eluted at a flow rate of 300µL/min with a linear gradient of 5–30% buffer B (95% ACN, 0.1% formic acid) over 60 min followed by ramping up to 100% buffer B over 5 min and holding for 20 min. The collected peptides were subjected to 5600 TripleTOFAnalyzer (AB Sciex).

The charge states of peptides were set to +2, +3, and +4. The mass spectrometer was maintained in positive ion mode, and MS spectra were acquired over a range of 300–1800 m/z. The resolving powers of the MS/MS and MS scan at 200 m/z for the Q-Exactive were set as 17,500 and 70,000, respectively. For the acquired MS spectra, the top ten most intense signals were collected for MS/MS analysis. Ions were fragmented through higher energy collisional dissociation with normalized collision energies of 30 eV, and the isolation window was 2 m/z. The maximum ion injection times were set at 10 ms for the survey scan and 60 ms for the MS/MS scans. The automatic gain control target values for both scan modes were set to 3.0 × 10<sup>6</sup> . The dynamic exclusion duration was 25 s. The chromatograms were recorded at 214 nm. The underfill ratio was defined as 0.1% on the Q-Exactive.

#### Peptide and Protein Identification and Proteomic Data Analysis

The raw mass files were analyzed by the Proteome Discoverer 1.4 software (Thermo Fisher Scientific). Search for the fragmentation spectra was performed by the MASCOT 2.3.02 software search engine embedded in Proteome Discoverer against the cucumber 9930 reference genome database version 2 (http://www.icugi.org/cgi-bin/ICuGI/index.cgi). The following search parameters were used: The initial precursor mass tolerance was set to ±20 ppm, and the fragment ion mass tolerance was set to ±20 mmu, trypsin as the cleavage enzyme. The protein identification was calculated by at least two unique peptides, minimum peptides for protein quantification was set to one, the normalization method and protein change ratio type (upregulated or downregulated) were both set as median. The results were filtered based on a false discovery rate (FDR) of no more than 1% to guarantee the result's confidence and a Mascot probability of 95%.

#### Total RNA Extraction and qRT-PCR

Total RNA from each sample was isolated by RNAiso Plus (Takara, China), then dissolved in water-DEPC and kept at a final concentration of 1000µg/mL using Biophotometer Plus (Expander, Germany). Total RNA was reverse-transcribed by a Takara PrimeScript <sup>R</sup> RT reagent kit with gDNA eraser according to the manufacturer's manual (Takara, China). Primer sequences for qRT-PCR were designed using Primer Premier 5.0 (Premier Biosoft International, Palo Alto, CA, USA). qRT-PCR was performed using a Takara SYBR <sup>R</sup> PrimeScriptTM RT-PCR kit (Takara, China). SYBR Green PCR cycling was performed on an iQTM 5 Multicolor real-time PCR detection system (Bio-RAD, USA) using 20µL samples with the following temperature program: 95◦C for 10 s, followed by 40 cycles of 95◦C for 15 s, and then annealing at 52◦C for 30 s. The relative quantization of gene expressions were calculated and normalized to Actin. Three replicates were used for qRT-PCR.

### Physiological Determination

The chlorophyll concentration was measured by a SPAD (soil and plant analysis development) meter (SPAD-502Plus, Konica Minolta, Tokyo, Japan). 7 days after removing waterlogging treatment, cucumber plants were visually compared with wellwatered control and scored for tolerance rating (TOR) using a scale of 0–5 (Navazio and Staub, 1994), where 0 = dead plants, 1 = 75 ∼100% of wilt, 2 = 50∼74% wilt, 3 = leaves undulating and recurved, 4 = recurved leave margins and 5 = green plant with no sign of stress. The lower scale stood for waterlogging susceptibility while the higher scale stood for tolerance. Ethylene production after 48 h of waterlogging treatment was measured by a gas chromatograph device equipped with a TRB-5 capillary column (0.32 mm id, 30 m length, 0.25 lm df) and a flame ionization detector. Temperature, injection, detector and heat clamber 150, 150, and 100◦C were set. Nitrogen was employed

as the carrier gas. PDC (EC 4.1.1.1) and ADH (EC 1.1.1.1) activities were detected by spectrophotometer (Spectrum SP-752, Shanghai, China) metrically at 340 nm, following the methods described by Xu et al. (2014). Crude enzymes extracts of hypocotyls were prepared by macerating 1.0 g of the tissue in 5.0 cm<sup>3</sup> of 0.1 M cold Tris-HCl buffer (pH 8.0) as described by Mustroph and Albrecht (2003). ADH and PDC activity was determined spectrophotometrically by measuring the rate of reduction of NAD<sup>+</sup> (Waters et al., 1991). The reaction mixture contained 30 mM ethanol, 65 mM Tris-HCl buffer (pH 9.5), 1 mM NAD+, and 0.1 cm<sup>3</sup> of crude enzyme extracts. Increase in absorbance due to production of NADH was recorded at room temperature. 1 U of PDC and ADH was defined as the amount of enzyme required to decompose 1 mol of substrate per g protein per minute. For ethanol production experiment, fresh hypocotyls were harvest and frozen rapidly in liquid nitrogen and ground to a powder. 500 mg root material was homogenized with 50 Mm cold 0.1 M HCl in a Teflon-sealed test tube. After 20 min incubation at 70◦C, a 1 mL sample of headspace was injected into a gas chromatograph (Thermo Trace GC Ultra) equipped with a supel cowax 10 column (30 m × 0.32 mm × 0.25µm) and a flame ionization detector following the instructions of Kato-Noguchi and Watada (1997). Ethanol evaporation under these conditions was negligible.

### RESULTS

#### Phenotype Differences of Zaoer-N and Pepino in Responding to Waterlogging Treatment

To evaluate the phenotype differences of waterlogging stress on the tolerant Zaoer-N and the sensitive Pepino, seven representative phenotypes after 7 days of waterlogging were recorded. As is shown in **Table 1**, the tolerance rating, SPAD value, leaf flesh weight, ARs number, and flesh weight for Zaoer-N were considerably greater than those for Pepino, indicating that Zaoer-N was significantly more tolerant than Pepino (p < 0.05). However, the differences between Zaoer-N and Pepino for primary root flesh weight and hypocotyls length were not significant which suggesting the primary root injury and hypocotyls elongation may not contribute to waterlogging tolerance between the two lines.

Transferring Zaoer-N seedlings to waterlogging stress resulted in ARs primordia emergence and visible on the hypocotyls surface 2 days after waterlogging, in comparing with control plants and waterlogged Pepino (**Figure 1**). Seven days after treatment, the average AR numbers protruded at the basal part of the hypocotyls were 30.4 (**Figure S1**). To elucidate the response mechanisms related to morphological differences between the two genotypes in responding to waterlogging, an iTRAQ proteomic approach was thus applied to the hypocotyls of the two-day-waterlogged Zaoer-N and Pepino seedlings.

#### Inventory of Hypocotyls Proteins Identified by iTRAQ

Using the Mascot software, a total of 52,053 spectra matched to known spectra, 46,773 spectra matched to unique peptides, and 23,533 peptides, 22,066 unique peptides, and 5508 proteins were identified (**Table 2**). All the mass spectrometry data have been deposited into the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD004023. Among these proteins, 1048 were between 20 to 60 kDa, 129

FIGURE 1 | Adventitious root formation on hypocotyls of cucumber seedlings. (A) Zaoer-N control; (B) Zaoer-N waterlogging; (C) Pepino control; (D) Pepino waterlogging. The yellow arrows in the picture indicated the location of adventitious roots. The water level was kept at about 2 cm above the soil for 2 days and then removed for photography.

TABLE 1 | Response of waterlogging tolerant line Zaoer-N and sensitive line Pepino after 7 days of waterlogging treatment.


*TOR, tolerance rating* = *0 (dead plants)—5 (green plants with no sign of symptom); SPAD, soil plant analysis development; ARN, adventitious root numbers; HLH, hypocotyls length; LFW, leaf flesh weight; HFW, hypocotyls flesh weight; PRFW, primary root flesh weight. Means within trait followed by the same letter(s) are not significant difference at p-value* < *0.05.* between 0 to 20 kDa, 379 between 60 to 100 kDa, and 175 over 100 kDa (**Figure 2A**). 1792 proteins had one identified peptide, 1027 had two, 417 had more than 11, and the left had 3–10 (**Figure 2B**). The peptide sequence coverage was primarily less than 20% (**Figure 2C**). Ratio distributions for the identified proteins in Zaoer-N and D8 were shown in **Figures S2A,B**, respectively. Because iTRAQ quantification underestimated the amount of real fold change between two samples, protein with a fold-change >2 or <0.5 in abundance were regarded as DRPs.

TABLE 2 | Spectra, peptides, and proteins that were identified from iTRAQ proteomics by searching against the cucumber reference genome 9930 version 2 database.


Based on this criteria, 146 DRPs in the Zaoer-N and/or Pepino were selected for further analysis. Of these, 90 DRPs (62 increased and 28 decreased in abundance) changed significantly in the tolerant Zaoer-N and 99 DRPs (62 increased and 37 decreased in abundance) in the sensitive Pepino (**Table S1**). Only 43 DRPs (34 increased and 9 decreased in abundance) were shared between the two lines, while 47 DRPs were only identified in Zaoer-N, and 56 DRPs were only identified in Pepino.

### Classification of Waterlogging Responsive DRPs

Of the 146 DRPs, 131 DRPs can be assigned to 21 categories using the COG database. The largest group was general function prediction only (18.3%) followed by posttranslational modification, protein turnover, chaperones (9.9%), carbohydrate transport and metabolism (9.2%), amino acid transport and metabolism (9.2%) and Translation, ribosomal structure and biogenesis (6.1%) (**Figure 3**). To further characterize these identified waterlogging responsive proteins, the DRPs identified in Zaoer-N and Pepino were respectively passed through GO enrichment analysis using the agriGO (http://bioinfo.cau.edu.cn/agriGO). A GO term was considered

to be significantly enriched if the false discovery rate was below 0.05. Enrichment analysis of GO functions revealed that 12 GO terms were found shared between Zaoer-N and Pepino, such as oxidoreductase activity (GO:0016491), response to stress (GO:0006950) and heme binding (GO:0020037). Four GO terms were unique to Zaoer-N, including response to oxidative stress (GO:0006979), response to chemical stimulus (GO:0042221), peroxidase activity (GO:0004601), and antioxidant activity (GO:0016209). On the contrary, GO terms related to carbohydrate metabolic process (GO:0005975), hydrolase activity (GO:0016798), and cofactor binding (GO:0048037) were specially enriched in Pepino (**Figure 4**).

To characterize the functional consequences of the DRPs associate with waterlogging stress and subsequent AR formation, KEEG pathway mapping based on KEGG orthology terms for assignment was also carried out. Only significantly enriched pathway categories that had a p-value lower than 0.05 were selected. The results indicated that waterlogging could affect phenylpropanoid biosynthesis and alpha-Linolenic acid metabolism in both of the two lines. KEEG terms "glycolysis/gluconeogenesis" and "biosynthesis of secondary metabolites" were the highly enriched in the DRPs that were specifically up-regulated in the Zaoer-N, but "arginine biosynthesis," "alanine, aspartate, and glutamate metabolism,"and "biosynthesis of amino acids" were dramatically enriched in DRPs in Pepino (**Table 3**). All of the DRPs in KEGG categories were shown in **Tables S2**, **S3**.

#### Protein–Protein Interaction Analysis

Plant proteins play interrelated roles together in the context of networks. To investigate how the cucumber hypocotyls transmit waterlogging stress signaling through protein– protein interactions, the identified DRPs in Zaoer-N and Pepino were analyzed respectively using the String database (http://string-db.org). The abbreviations of the protein names in the networks were shown in **Table S4**. Eight separate interaction networks were predicted in Zaoer-N (**Figure 5A**). The ribosomal proteins, including 40S ribosomal protein (AT5G39850), 60S ribosomal proteins (RPL16A, AT5G02870, AT1G57660, AT2G19730, AT4G29410), ribosomal protein L5 (AT3G25520),

TABLE 3 | KEGG pathway enrichment analysis of differentially regulated proteins (DRPs).


*KOBAS2.0 web-based program was used for KEEG mapping (http://kobas.cbi.pku.edu.cn/).*

were found to be actively interacted with embryo defective 3031 (EMB3031). Another notable interacted protein group is composed of pyruvate kinase (AT5G08570), PDC (AT5G01320), ADH1, cinnamyl alcohol dehydrogenase 1 (ATCAD4), Prxs (PRXR1, AT2G41480, PRX52, PER64, RHS19, and AT2G37130), and phenylalanine ammonia-lyase 2 (PAL2). Argine/serine-rich zine knuckle-containing protein 33 (RS2Z33) was a key protein in a another network and interacted with pre-mRNA-splicing factor SR34, serine/arginine-rich protein RSZ22, zine finger protein RSZ21, SC35-like splicing factors (SCL30A, SCL30, and SCL28). Furthermore, there were five pairs of interactive protein-species in this network. The main two pairs were hemoglobin 3 (GLB3) and hemoglobin 1 (HB1), deoxyhypusine synthase (DHS) and eukaryotic elongation factor (ELF5A-1), lipoxygenase 2

(LOX2) and alpha-dioxygenase (DOX1), haloacid dehalogenaselike hydrolase domain-containing protein (AT5G59480) and haloacid dehalogenase-like hydrolase domain-containing protein (AT5G44730), uncharacterized protein (AT5G39890) and aluminum induced protein (AT4G27450), respectively. Seven separate interaction networks were predicted in Pepino, and some of them were overlapped with the prediction in Zaoer-N (**Figure 5B**). However, the network made up of heat shock proteins (HSP90.1, HSP23.6, ATHSP22.0, AT1G53540, and HSP21), phosphatases (SGT1A and SGT1B) and SQUINT were only found in Pepino. Although these predicted interaction networks need to be verified, they have provided a narrow pool of protein–protein interactions in cucumber hypocotyls in responding to waterlogging for our further investigations.

## Correlation of iTRAQ Data with mRNA Expression Data

The correlation of iTRAQ data and mRNA expressions were carried out to further investigate protein expression profiles under waterlogging stress. The mRNA expressions levels were obtained by qRT-PCR analysis of the corresponding 146 genes with the same experimental design. In the correlation analysis, we calculated the Pearson's correlation coefficients (PCC-value). The PCC-values of the DRPs and mRNA pairs were 0.26 in Zaoer-N (**Figure 6A**), while PCC-values were 0.18 for groups in Pepino, respectively (**Figure 6B**). As expected, significant correlations existed in both of the two groups at the 0.05 level (2-tailed). The positive correlation of iTRAQ data and mRNA expression levels suggested that the regulation of these proteins likely resulted from the transcriptional inductions of the corresponding genes in the waterlogged hypocotyls cells. However, the results also showed that some proteins and corresponding gene relative expression levels were not well correlated. The discrepancy of these protein and mRNA expression levels also has been found in other iTRAQ studies, and may be ascribed to translational and posttranslational regulatory processes or feedback loops between the processes of mRNA translation and protein degradation(Vogel and Marcotte, 2012).

#### Physiological Responses of Zaoer-N and Pepino to Waterlogging

To investigate whether these identified DRPs can result in physiological changes, we also evaluated the ethylene release, ADH and PDC activities and production of ethanol (**Figure 7**). Two days of waterlogging treatment led to a significant increase in ADH and PDC activities (p < 0.05) and both of them were higher in Zaoer-N than that in Pepino (**Figures 7A,B**). As we known, the production of ethanol closely correlated with ADH activity. Compared with non-waterlogged cucumber hypocotyls, the ethanol concentrations after 2 days of waterlogging increased significantly (p < 0.05) in both lines but to a greater degree in Zaoer-N than in Pepino (**Figure 7C)**. A 7.3-fold and 4.1 fold increase in ethylene release after 2 days of waterlogging was observed in hypocotyls of Zaoer-N and Pepino, respectively (**Figure 7D**).

## DISCUSSION

ARs formation as a result of flooding or waterlogging stress were reported for tomato (Vidoz et al., 2010), bittersweet (Dawood et al., 2014, 2016), wheat (Barrett-Lennard et al., 1988), barley (Pang et al., 2007), rice (Bleecker et al., 1986), and maize (Mano et al., 2005). In deep-water rice cultivar Pin Gaew56, AR primoridia emerged from the stem node within 8 ∼ 10 h of partial submergence (Lorbiecke and Sauter, 1999). The present study showed that ARs primordia emergence and visible on the cucumber hypocotyls surface 2 days after waterlogging, the number of ARs significantly increased in Zaoer-N compared with Pepino 7 days after waterlogging, and the SPAD value in the leaves was significantly higher in Zaoer-N than in Pepino (**Table 1**). These observations indicated that Zaoer-N adapted

better to waterlogging stress than Pepino. ARs facilitate oxygen diffusion in Zaoer-N to accelerate aerobic metabolism, and maintaining a sufficient energy supply to meet the demands of the long-term survival (Bailey-Serres and Voesenek, 2008). The process of activation of AR emergence under waterlogging, however, has been studied to a much lesser extent. To further our understanding about the molecular and physiological reprogramming that enable AR primordia to resume growth upon an abiotic trigger, we thus analyzed the proteomic response to waterlogging in cucumber at the initial stage of ARs growth.

to normalize the expression data. Each value denotes the mean relative level of

## Management of the Energy Crisis

Energy crisis is a major factor affecting waterlogged organs survival because only 2 ∼ 4 mol ATP per mol hexose were produced by glycolysis and fermentation when compared with 30 ∼ 36 mol ATP produced by aerobic respiration (Bailey-Serres and Voesenek, 2008). Previous studies have confirmed that both sensitive and tolerant plants are able to maintain their ATP production by inducing glycolytic and fermentative enzymes at the early adaptive response to waterlogging treatment (Yu et al., 2015). As expected, we also found that several enzymes, such as UDP-glucose 6-dehydrogenase (Csa1M012150.1) in hemicellulose synthesis, fructose-bisphosphate aldolase (Csa3M750920.1), and pyruvate kinase (Csa5M580610.1 and Csa6M449830.1) involved in glycolysis, and ADH (Csa7M320050.1) and PDC (Csa6M518930.1) in ethanol fermentation were significantly accumulated in both of Zaoer-N and Pepino, suggesting the common regulation of energy generation in waterlogged cucumber hypocotyls. However, our data showed that ADH (Csa7M322060.1) accumulated only in Zaoer-N (**Table 4**, **Table S1**), in accordance with the observation of ADH activity and ethanol concentrations (**Figures 7A,C**). As we known, ADH plays an essential role in the recycling of NADH to NAD+, which is pivotal for the continuation of glycolysis pathway (Ismond et al., 2013), and is the only energy source under oxygen deprivation condition. Interestingly, aldehyde dehydrogenase (ALDH, Csa1M372010.1), which converts the acetaldehyde into acetate, was accumulated only in Pepino. As postulated by Nakazono et al. (2000), the conversion of acetaldehyde to acetate by ALDH also consumes NAD+, which could potentially further block glycolysis and ATP supply. Together, these results suggested that the two lines exhibit a similar enzymes cascade as other plants species in responding to waterlogging stress, but the more efficient of regenerate ATP and NAD<sup>+</sup> and continuation of the glycolysis pathway in tolerant Zaoer-N than sensitive Pepino to cope with energy crises imposed by waterlogging. Nevertheless, high rates of glycolysis and alcohol fermentation will unavoidably lead to the fast depletion of sugar stores and carbon crisis, and the products of alcohol fermentation (acetaldehyde and ethanol) may have damaging consequences on cell integrity (Limami et al., 2008). Thus, escape from the damaging consequences appears an important issue in survival strategy, especially for those tolerant lines if waterlogging stress prolonged.

### Ethylene Production and ROS Production in Responses to Waterlogging

Among several internal changes in the waterlogged plants, it is the pervasive and rapid accumulation of ethylene that makes it an early and reliable waterlogging-triggered signal (Sasidharan and Voesenek, 2015). In our previous study, we found that the ARN in Zaoer-N hypocotyls 3 days after waterlogging was significantly inhibited when pretreated with 100 mg/ L 1 methylcyclopropene as an inhibitor of ethylene action (Xu et al., 2016), suggesting the importance of ethylene in waterloggingtriggered cucumber AR production. Ethylene production was observed in both of Zaoer-N and Pepino 2 days after treatment (**Figure 7D**). As is known to all, the gaseous phytohormone ethylene is biosynthesized from methionine, and it is produced by the activation of 1-aminocyclopropane-1-carboxylicacid (ACC)

expression of three biological replicates.

ethanol production (C), and ethylene production (D) were measured. Data represents the average values ± SE of three independent experiments. Letters above the bars indicate a statistically significant difference (*p* < 0.05) according to Duncan's multiple range test.

TABLE 4 | Information of differentially regulated proteins related to alcohol dehydrogenase, pyruvate decarboxylase, and 1-aminocyclopropane-1-carboxylate oxidase.


synthase and ACC oxidase (ACO) (Yang and Oetiker, 1994). We identified three ACOs (Csa2M000520.1, Csa4M056660.1, and Csa6M511860.1), and two of these proteins were accumulated in both lines, whereas Csa2M000520.1 accumulated only in Zaoer-N, indicating that a higher quantity of ethylene production in waterlogged Zaoer-N than those in Pepino may caused by accumulation of this protein (**Table 4**, **Table S1**). It should be pointed out that S-adenosyl-L-methionine synthetase (SAM, Csa3M016410.1) was decreased following exposure to 2 days of treatment. SAM is the key enzyme in the synthesis of S-adenosyl-L-methionine, which is an important metabolic component in the ethylene biosynthesis pathway (Yang and Hofman, 1984). The decrease of SAM in the proteome from waterlogging-treated tomato root has also been reported (Ahsan et al., 2007). The decrease of SAM observed in the current study indicated that ethylene production in cucumber hypocotyls would be reduced after 2 days of treatment. In deep-water rice, the enhanced formation of epidermal cell death at sites where ARs primordia emerge is dependent on the presence of ethylene (Mergemann and Sauter, 2000; Steffens et al., 2011). The growing ARs exert a mechanical force on the epidermal cells overlying them in a process that also requires ethylene-mediated ROS generation (Steffens et al., 2006). Despite of ROS at lower concentrations may act as signaling molecules involved in acclimation to environmental stress, a high level of these compounds are also harm to the plant cell. Heat shock proteins (HSPs) are the most recurrent of ROS related proteins that not only respond to heat stress, but also in respond to other stress, such as waterlogging, and has been proven to be associated with the presence of hydrogen peroxide (Pucciariello et al., 2012; Banti et al., 2013). Five HSPs were identified in present study, and all of them (Csa3M113300.1, Csa3M020080.1, Csa3M183950.1, Csa5M138480.1, and Csa5M591720.1) were accumulated only in Pepinos. Increase of HSPs has been proposed to be specific H2O<sup>2</sup> sensors in plants, suggesting the higher H2O<sup>2</sup> production in Pepino.

### Proteins Involved in Cell Division and Cell Growth

The process of AR primordia emergence is accompanied with extensive cell division and cell growth. Regulation of cell division and cell growth-related proteins were also apparent. The DRPs list contained 14 cell wall remolding proteins, including expansins, pectinesterase, proteinase inhibitor, xyloglucan endotransglucosylase/hydrolases, polygalacturonase, and some structural constituent of cell wall. Most of these proteins had higher fold changes in Zaoer-N than Pepino, suggesting the AR primordia emergence in Zaoer-N could not be separated with cell wall extensibility. Interestingly, nine peroxidase (Prxs), which involved in many development processes such as cell growth and differentiation (Dunand et al., 2007), were also differentially regulated. Five of them (Csa6M495000.1, Csa7M049140.1, Csa7M061710.1, Csa2M346020.1, and Csa3M736970.1) were accumulated only in the waterlogged hypocotyls of Zaoer-N. Passardi et al. (2006) found that Arabidopsis seedlings lacking Prx33 transcripts have shorter roots than the wild-type controls and roots are still shorter in the double mutant, while seedlings overexpressing AtPrx34 exhibit significantly longer roots. Studies have also provided evidences that Prxs activities play an protective role in the ROS scavenging process upon waterlogging stress (Qi et al., 2012). Thus, accumulation of Prxs in Zaoer-N will help consumption of ROS and promote cell elongation around the AR primordia. Additionly, our iTRAQ data also showed that two histone deacetylases (Csa3M748810.1 and Csa6M150540.1) were decreased in Zaoer-N specially, which might results in decondensation of the chromatin structure around the promoter regions of embryonic genes and help stimulating cell differentiation (Yoo et al., 2006). The Arabidopsis histone deacetylases HDA6 and HDA19 redundantly regulate embryonic genes and negatively affect callus formation from the hypocotyls (Tanaka et al., 2008). Thus, decrease of these histone deacetylases may also help promoting the emergence of AR primordia in Zaoer-N hypocotyls. Although we did not detect cyclin or cyclin dependent kinase in the hypocotyls of Zaoer-N at the 2 days time point, there were already a specific induction of many proteins controlling cell size and cell proliferation in Zaoer-N hypocotyls, such as 60S ribosomal proteins, which are absolutely required for protein synthesis (Bailey-Serres and Freeling, 1990), and histone H5 (Csa6M407650.1) in cell division (Bailey-Serres and Voesenek, 2008). In contrast to the situation in the primordia, a large group of cell wall modifying- and cell cycle-related proteins were decreased in the hypocotyls of Pepino. These findings suggest that cell division and growth are suppressed in Pepino hypocotyls upon waterlogging, such that these hypocotyls might enters a state of endurance.

### Proteins Related to Jasmonic Acid (JA) Content

The threonine dehydratase (TD, Csa6M448730.1) was the highest decreased (∼0.048 fold than control) protein that specifically regulated in waterlogged Zaoer-N. TD catalyzes the formation of α-keto butyrate from Thr, the first step in the biosynthesis of Ile (Kang et al., 2006). For more than a decade, TD has been recognized as a reliable marker for JA elicitation in potato and tomato (Dammann et al., 1997). This information prompted us to investigate the JA concentration during AR emergence on cucumber hypocotyls. As expected, JA concentrations in 2 day waterlogged Zaoer-N hypocotyls were significantly decreased (∼0.33 fold) than control. On the contrary, JA concentrations in Pepino hypocotyls were significantly increased (∼1.99 fold) than control after 2 days of waterlogging (**Table S5**). Sanders et al. (2000) and von Malek et al. (2002) found that JA deficient mutants opr3/dde1 and dde2-2 produced more ARs than the wild type, Gutierrez et al. (2012) showed that Arabidopsis gh3 triple mutant accumulated twice as much JA but inhibited the AR initiation compared with the wild type. It is tempting to speculate that JA is an inhibitor of cucumber adventitious rooting. The accumulation of JA in waterlogged Pepino might be caused by the higher fold changes of lipoxygenase (Csa4M286960.1, ∼6.29 fold than control, **Table S1**), which is a key enzyme in JA biosynthesis. However, due to the oxygen requirement of lipoxygenase (Arbona and Gómez-Cadenas, 2008), the increased JA biosynthesis via lipoxygenase pathway will unavoidably lead to higher risk of being injured by oxygen deprivation under waterlogging condition.

### Proteins Related to Phenylpropanoid Metabolism and Lipid Catabolism

A feature of plant responses to waterlogging is the activation of phenylpropanoid metabolism, which is also supported by our KEEG pathway mapping (**Table 3**). We found that DRPs enriched in KEEG term "phenylpropanoid metabolism" (**Table S3**) were all Prxs (as described in Proteins Involved in Cell Division and Cell Growth Section) except for a phenylalanine ammonia-lyase (PAL, 4.3.1.5). PAL catalyzes the first committed step of the core pathway of general phenylpropanoid metabolism (Gómez-Vásquez et al., 2004), and is one of the most extensively studied enzymes with respect to plant responses to abiotic stress. Higher levels of PAL activity or mRNA accumulation in response to waterlogging have been found in tomato roots, Vigna sinensis and Zea mays (Alla et al., 2002). However, in current investigation, we found that PAL (Csa6M147460.1) decreased in cucumber hypocotyls after 2 days of waterlogging treatment, and to a greater degree in Zaoer-N (0.071) than in Pepino (0.456). Politycka (1998) found that the upregulation of PAL activity resulted in cucumber root growth decreasing. The highly inhibition of PAL enzyme could result from accumulation of its intermediate product in phenylpropanoid metabolism i.e., the trans-cinnamic acid, and we thus speculated that the decrease of PAL in waterlogged Zaoer-N could benefit to waterlogging triggered AR growth.

In plants, lipid catabolism is known to be an active process that can mobilize storage lipids and can detoxify deleterious membrane lipids caused by abiotic stresses (Kim et al., 2016). Except for the lipoxygenase discussed above, our iTRAQ study also found that three GSDL esterases/lipases, Csa1M058140.1, Csa3M669600.1, and Csa3M434970.1, were differentially regulated by waterlogging stress. GDSL esterases/lipases are a newly discovered subclass of lipolytic enzymes that can break down lipids and hydrolyze triglycerides into glycerol and free fatty acids (Chepyshko et al., 2012). Thus, far, little data have apparently been published on the detailed role of GSDL esterase/lipase activity in waterlogging tolerance or AR formation. Interestingly, Csa3M669600.1was the highest accumulated (∼7.029 fold than control) protein in waterlogged Zaoer-N, and Csa1M058140.1 was induced only in Zaoer-N. These indicated enhanced lipid catabolism, possibly via elevated glycerol to provide additional carbon and energy sources for AR primordia growth or free fatty acids for membrane constituents of those newly developed cells (Li et al., 2016).

Together, this study reveals the proteomic difference of the tolerant line Zaoer-N and the sensitive line Pepino upon exposure to waterlogging stress. Our iTRAQ-based proteomic data showed that the more efficient of regenerate ATP and NAD<sup>+</sup> and continuation of the glycolysis pathway in tolerant line Zaoer-N appears to be an important way to maintain the level of available carbohydrate and energy to prevent intracellular energy shortage resulting from the waterlogging-caused oxygen depravation. Moreover, the phytohormone ethylene and ROS production might act as the response signaling in activating the apical meristem in AR primordial, whereas the JA might play an opposite role. Compare with Pepino, proteins cell division, cell growth, phenylpropanoid metabolism, and lipid catabolism are also important to facilitate protrusion of the AR primordia to the outside in Zaoer-N hypocotyls. These results indicated that proteomics with bioinformatics analysis is a good starting point for understanding the overall molecular response of cucumber hypocotyls to waterlogging. A deep and broad understanding of the DRPs identified is ongoing to make clear their specific role in ARs emergence upon waterlogging.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

XC conceived the experiment. XX and JJ performed the research. XX, JJ, and XM collected data. XX, QX, and XQ analyzed the data and wrote the manuscript.

#### ACKNOWLEDGMENTS

This research was funded by the National Natural Science Foundation of China (31372087). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors gratefully acknowledge financial support from China scholarship council and foundation of excellent doctoral dissertation of Yangzhou University.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01515

Table S1 | Detailed information of the differentially regulated proteins identified by iTRAQ.

Table S2 | KEEG pathway enrichment analysis of differentially regulated proteins in Zaoer-N.

Table S3 | KEEG pathway enrichment analysis of differentially regulated proteins in Pepino.

Table S4 | The abbreviations of the protein names showed in the networks of Figure 5.

Table S5 | Endogenous jasmonic acid concentrations in hypocotyls of Zaoer-N and Pepino after 2 days of waterlogging.

Figure S1 | Morphology of 7-day waterlogged Zaoer-N and Pepino. The water level was kept at about 2 cm above the soil for 7 days and then removed for photography.

Figure S2 | Ratio distributions for the identified proteins. The changes of protein contents in Zaoer-N (A) and Pepino (B) 48 h after waterlogging treatment were analyzed. The horizontal axis displays the base Log2-transformed ratios. The red point indicates that the ratio was greater than 1.5, and the green point indicates that the ratio was less than 0.67. The differentially regulated proteins with a fold-change >2 (red points on the right of red lines) or <0.5 (green points on the left of green lines) in abundance were further analyzed in the present study.


root of cucumber (Cucumis sativus L.). Sci. Hortic. 179, 388–395. doi: 10.1016/j.scienta.2014.10.001


**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 © 2016 Xu, Ji, Ma, Xu, Qi and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proteomic Analysis Reveals the Positive Roles of the Plant-Growth-Promoting Rhizobacterium NSY50 in the Response of Cucumber Roots to Fusarium *oxysporum* f. sp. *cucumerinum* Inoculation

#### *Edited by:*

*Jie Zhou, Zhejiang University, China*

#### *Reviewed by:*

*Yuan Huang, Huazhong Agricultural University, China Abu Hena Mostafa Kamal, University of Texas at Arlington, USA*

#### *\*Correspondence:*

*Shirong Guo srguo@njau.edu.cn † These authors have contributed equally to this work.*

#### *Specialty section:*

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

*Received: 26 September 2016 Accepted: 25 November 2016 Published: 14 December 2016*

#### *Citation:*

*Du N, Shi L, Yuan Y, Li B, Shu S, Sun J and Guo S (2016) Proteomic Analysis Reveals the Positive Roles of the Plant-Growth-Promoting Rhizobacterium NSY50 in the Response of Cucumber Roots to Fusarium oxysporum f. sp. cucumerinum Inoculation. Front. Plant Sci. 7:1859. doi: 10.3389/fpls.2016.01859* Nanshan Du1 † , Lu Shi 1 †, Yinghui Yuan<sup>1</sup> , Bin Li <sup>2</sup> , Sheng Shu1, 3, Jin Sun1, 3 and Shirong Guo1, 3 \*

*<sup>1</sup> Key Laboratory of Southern Vegetable Crop Genetic Improvement in Ministry of Agriculture, College of Horticulture, Nanjing Agricultural University, Nanjing, China, <sup>2</sup> Department of Horticulture, Shanxi Agricultural University, Taigu, China, <sup>3</sup> Suqian Academy of Protected Horticulture, Nanjing Agricultural University, Suqian, China*

Plant-growth-promoting rhizobacteria (PGPR) can both improve plant growth and enhance plant resistance against a variety of environmental stresses. To investigate the mechanisms that PGPR use to protect plants under pathogenic attack, transmission electron microscopy analysis and a proteomic approach were designed to test the effects of the new potential PGPR strain *Paenibacillus polymyxa* NSY50 on cucumber seedling roots after they were inoculated with the destructive phytopathogen *Fusarium oxysporum* f. sp. *cucumerinum* (FOC). NSY50 could apparently mitigate the injury caused by the FOC infection and maintain the stability of cell structures. The two-dimensional electrophoresis (2-DE) approach in conjunction with MALDI-TOF/TOF analysis revealed a total of 56 proteins that were differentially expressed in response to NSY50 and/or FOC. The application of NSY50 up-regulated most of the identified proteins that were involved in carbohydrate metabolism and amino acid metabolism under normal conditions, which implied that both energy generation and the production of amino acids were enhanced, thereby ensuring an adequate supply of amino acids for the synthesis of new proteins in cucumber seedlings to promote plant growth. Inoculation with FOC inhibited most of the proteins related to carbohydrate and energy metabolism and to protein metabolism. The combined inoculation treatment (NSY50+FOC) accumulated abundant proteins involved in defense mechanisms against oxidation and detoxification as well as carbohydrate metabolism, which might play important roles in preventing pathogens from attacking. Meanwhile, western blotting was used to analyze the accumulation of enolase (ENO) and S-adenosylmethionine synthase (SAMs). NSY50 further increased the expression of ENO and SAMs under FOC stress. In addition, NSY50 adjusted the transcription levels of genes related to those proteins. Taken together, these results suggest that *P. polymyxa* NSY50 may promote plant growth and alleviate FOC-induced damage by improving the metabolism and activation of defense-related proteins in cucumber roots.

Keywords: cucumber, *Fusarium oxysporum* f. sp. *cucumerinum*, plant-growth-promoting rhizobacteria, proteomics, root

#### INTRODUCTION

Cucumber (Cucumis sativus L.) is an important and popular vegetable cash crop that is consumed worldwide. However, the production of this plant is severely threatened by cucumber Fusarium wilt, which is caused by the soilborne fungal pathogen Fusarium oxysporum f. sp. cucumerinum (FOC), a destructive vascular disease that can cause plant death and serious economic loss (Ahn et al., 1998; Huang et al., 2012). Because there are no available effective chemical products or resistant varieties and because grafting is time consuming (Chung et al., 2008; Cao et al., 2012), other effective, economical, and environmentally friendly cultivation methods have been developed and tested.

To date, alternative methods, such as biological control agents, have been shown to be effective and have been increasingly applied in the field (Yang et al., 2010; Li et al., 2012; Lin et al., 2014; Xu et al., 2014). Among these methods, some beneficial bacteria inhabiting the plant rhizosphere, called plant growthpromoting rhizobacteria (PGPR), are directly or indirectly involved in promoting plant growth and the biological control of plant diseases (Kloepper and Metting, 1992). A variety of mechanisms have been confirmed to improve plant health and increase crop productivity for those strains, such as synthesizing different antimicrobial compounds (Tamehiro et al., 2002; Ongena et al., 2007; Chen et al., 2014), various hormones (Bottini et al., 2004; Martínez-Viveros et al., 2010; Kochar et al., 2011), increasing the availability of plant nutrients in the rhizosphere (Idriss et al., 2002; Palacios et al., 2014), and inducing systemic resistance in host plants (Niu et al., 2011; Jiang et al., 2015).

Numerous technical methods, such as DNA microarray technology and proteomic analysis, have been applied to improve our understanding of plant-PGPR interactions. A study in the Arabidopsis root transcriptome showed that extensive changes were induced in hormone- and defense-related genes after exposure to Azospirillum brasilense Sp245 (Spaepen et al., 2014). In another study, the ISR- and iron acquisition-related transcription factors in Arabidopsis roots were activated by Pseudomonas fluorescens WCS417 and its volatiles (Zamioudis et al., 2015). A dynamic protein network was designed to explain the mechanism of P. putida UW4 to release hypoxic stress and promote cucumber growth (Li et al., 2013b). Proteins involved in growth promotion and defense reactions were significantly induced in Arabidopsis following inoculation with Paenibacillus polymyxa E681 (Kwon et al., 2016). In addition, many other researchers have used proteomics to investigate plant-pathogen interactions. Wang et al. (2011) found that a switch from glycolysis to the pentose phosphate pathway may be an active response of the cotton plant against V. dahliae infection to enhance wilt tolerance or resistance. Total root protein was isolated from infected cucumber roots of the susceptible bulk and resistant bulk of cucumber generation F2, and the result showed that jasmonic acid and redox signaling components occurred in response to F. oxysporum infection in resistant plants (Zhang et al., 2016). Eighty-six differentially expressed proteins were identified from the leaves of resistant tomato cultivar "Zheza-301" and susceptible cultivar "Jinpeng-1" after TYLCV infection, which demonstrated that an interaction network between tomato leaves and TYLCV infection was established (Huang et al., 2016). To date, although several papers have addressed the use of proteomics to investigate the interactions of plants and pathogens (Konishi et al., 2001; Wang et al., 2011; Carrillo et al., 2014) or plants and PGPR (Kandasamy et al., 2009; Wang et al., 2013), little research is available on the effects of PGPB on cucumber plants under pathogenic attack.

In this present study, a promising bio-control agent Paenibacillus polymyxa-NSY50, which was originally isolated from the high suppression capacity of compost (Du et al., 2015; Shi et al., 2016), was used to better understand direct or indirect interactions among P. polymyxa NSY50, the pathogen FOC and cucumber roots. A total of 56 proteins were identified that were up- or down-regulated by the inoculation of P. polymyxa NSY50 and/or FOC, and the proteins could be classified into multiple biological functions. This work provides key insights into the promotion mechanisms of P. polymyxa NSY50 action to determine the optimal use of this beneficial PGPR in sustainable agricultural practices.

#### MATERIAL AND METHODS

#### Plant Material, Microbial Culture Conditions, and Treatments

Cucumber (Cucumis sativus L. cv. Jinchun No. 2) seeds were sown in quartz sand and cultivated in a greenhouse. Paenibacillus polymyxa-NSY50 was grown on LB medium at 28◦C for 3 days, and the cucumber Fusarium wilt pathogen F. oxysporum f. sp. cucumerinum (FOC) was incubated in PDA liquid culture for 7 days.

At the two true leaves stage, the cucumber seedlings were transplanted to tanks containing half-strength Hoagland nutrient solution (He et al., 2015). After 3 days of pre-culture, the seedlings were treated as follows: (1) control, control plants were grown in Hoagland's solution; (2) NSY50, seedlings grown in Hoagland nutrient solution containing 500-mL 1.0 × 10<sup>8</sup> NSY50 cell suspension; (3) FOC: 500 mL of a cell suspension of FOC at 1 × 10<sup>7</sup> conidia/mL was poured into the tanks after 6 days of culture; and (4) NSY50+FOC: plants were inoculated with 500 mL of NSY50 (1 × 10<sup>8</sup> CFU/mL) after 3 days of culture, and 3 days later, the plants were challenge-inoculated with 500 mL of a cell suspension of FOC (1 × 10<sup>7</sup> conidia/mL). The experiment was arranged in a randomized complete block design with three replicates for each treatment, which resulted in a total of 36 seedlings per treatment. Seedlings were reared in a greenhouse at 25–30◦C in the daytime and at 15–18◦C at night under natural lighting at 60–75% relative humidity.

#### Plant Growth Analysis

After 9 days of FOC treatment, shoot, and root were separated from one another and measured after being washed with sterile distilled water. The plant height, biomass, and volume of root in each treatment were measured as previously described (Shi et al., 2016). Each data point was the average of measurements collected in triplicate.

#### TEM Analysis

For transmission electron microscopy, roots of seedlings from different treatments were excised and immediately cut into small segments (1∼2 mm, including the tip); The samples were then fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 24 h (primary fixation) and immersed in 2% osmic acid in the same buffer for 2 h (second fixation). After dehydration in acetone and embedding in Durcupan ACM (Fluka), the resulting roots were cut to obtain ultra-thin sections, stained with uranium acetate and lead citrate in series and examined using a HITACHI transmission electron microscope (Carl Zeiss, Göttingen, Germany) at an accelerating voltage of 80 kV.

#### Protein Extraction

The cucumber roots were prepared for protein fractionation 3 days post-inoculation of FOC. Total protein extraction was performed using a trichloroacetic acid, i.e., the acetone precipitation method modified described by Hurkman and Tanaka (1986). Fresh root samples (2 g) were milled with liquid nitrogen and suspended in extraction buffer as described by An et al. (2016). The homogenate was centrifuged at 15,000 g at 4 ◦C for 20 min. An aliquot (1 mL) of the resulting supernatant was transferred to a new tube and precipitated with ice-cold acetone (10% TCA and 0.07% β-mercaptoethanol) overnight at −20◦C, and then the resulting protein sample was centrifuged at 20,000 g for 25 min. The pellet was washed three times with cold acetone (0.07% β-mercaptoethanol) and allowed to stand at −20◦C for 2 h. Finally, the protein pellet was air-dried and used for 2-DE.

#### 2-DE and Image Analysis

Isoelectric focusing (IEF) was performed using an 18-cm IPG linear gradient strip, pH 4–7 (GE Healthcare, San Francisco, USA). The dried protein pellet was rehydrated in a rehydration buffer that contained 7 M urea, 2 M thiourea, 4% 3-[(3 cholanidopropyl) dimethylammonio]-1-propanesulfonic acid (w/v), 40 mM DTT, 0.5% (v/v) IPG buffer 4–7 and 0.01% (w/v) bromophenol blue. The protein levels were quantified using the Bradford method (Bradford, 1976). The 350µL protein samples were loaded on IPG strips containing 800µg of protein and rehydrated for 12–16 h at 25◦C. After rehydration, the IPG strips were run on an Ettan IPGphor 3 (GE Healthcare, San Francisco, USA) using the following protocol: the voltage for IEF was set at 100 V for 1 h, followed by 200 V for 1 h, 200 V for 1 h, 500 V for 1 h, 1000 V for 1 h, 4000 V for 1 h, a gradient of 10,000 V for 1 h, and then a 10,000 V rapid focus, which reached a total of 75,000 V h with a maximum electric current of 50 µA per strip. After accomplishing the first dimension, IEF strips were equilibrated for 15 min in a 2D equilibrium buffer that consisted of 6 M urea, 30% glycerol (v/v), 50 mM Tris-HCl (pH 8.8), and 2% sodium dodecyl sulfate (SDS) containing 1% DTT (w/v); the strips were then incubated in 2.5% (w/v) iodoacetamide instead of DTT for 15 min. Then, the strips were placed directly onto 12.5% polyacrylamide-SDS slab gels and sealed using 1% molten agarose solution. The second dimensional separation was conducted using the EttanDalySix electrophoresis system (GE Healthcare, USA). At 15 mA per gel, the electrophoresis was not stopped until the bromophenol blue dye reached approximately 1 cm from the bottom of the gel. Protein spots were visualized using Coomassie Brilliant Blue (CBB) R-250.

The CBB-stained 2-DE gels were scanned using an Image scanner III (GE Healthcare, USA). The digitized images of three independent experiments were analyzed with Imagemaster 2D Platinum version 5.0 (GE Healthcare, USA). The abundance of each protein spot was estimated by the percentage volume (vol.%), and the spot volumes were normalized as the ratio of the total volume for all of the spots that were present. Spots with a vol.% that represented at least a 1.5-fold change and was significant at the P < 0.05 level (Duncan's multiple range tests) were used for identification.

### Protein Identification and Functional Classification

The differentially expressed protein spots were extracted from CBB-stained preparative polyacrylamide gels and then identified using an ABI 5800 Proteomics Analyzer MALDI-TOF/TOF (Applied Biosystems, Foster City, CA, USA). The data lists were used as a query to search in the NCBI (http://www.ncbi.nlm. nih.gov/) and cucumber genomics databases (http://cucumber. genomics.org.cn) using the software MASCOT version 2.2 (Matrix Science, London, UK). The following search parameter criteria were used: trypsin cleavage, one missed cleavage site allowed; carbamidomethyl set as a fixed modification; oxidation of methionines allowed as a variable modification; a peptide mass tolerance within 100 ppm; fragment tolerance set to ± 0.4 Da; and a minimum ion score confidence interval for MS/MS data set to 95%.

Identified proteins were categorized according to the biological processes with which they were involved based on Gene Ontology (GO) (http://www.geneontology.org/) and the Uniprot Protein Knowledgebase (http://www.uniprot.org/). Hierarchical clustering of the protein expression patterns was performed on the log-transformed (using log base 2) spot abundance ratios using the software Cluster version 3.0. A heat map was created using Java Treeview.

#### Western Blot Analysis

Root tissues were milled in a mortar with ice-cold extraction buffer containing 30 mM Tris—HCl (pH 8.7), 1 mM MgCl2, 0.7 M saccharose, 1 mM EDTA, 1 mM DTT, 1 mM PMSF, and 1 mM ascorbic acid. The extracted protein was quantified using the Bradford method (1976), denatured at 95◦C for 3–5 min and then stored at −20◦C until analysis.

For western blot analysis, SDS-polyacrylamide gel electrophoresis (SDS-PAGE) was conducted using the methods of Laemmli (1970). Briefly, 11 µg protein samples were transferred to a 0.45 µm PVDF membrane at 10 V for 1.5 h and were washed with TBST three times; then, the PVDF membrane was blocked with 5% nonfat dry milk for 2 h, washed with TBST three times, and incubated with monoclonal antibodies against S-adenosylmethionine synthase and enolase (produced in rabbit; Univ-bio, Shanghai, China) for 2 h. The membrane was washed with TBST and incubated at room temperature for 1 h with a Goat Anti-Rabbit IgG HRP-conjugate. The membrane was then washed with TBST three times and developed using diaminobenzidene (DAB) and H2O<sup>2</sup> (He et al., 2012).

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

Root tissues were harvested at 1, 3, 9 days after inoculation with FOC. The total RNA was extracted according to the TRI reagent protocol (Takara Bio Inc.) and then converted into cDNA following the manufacturer's instructions. Primers were designed based on the sequences obtained from NCBI using Beacon Designer 7.90 (Supplementary Table 1). qRT-PCR was performed using the SYBR PrimeScriptTM RT-PCR Kit (Takara Bio Inc.) in accordance with the manufacturer's instructions and run on a StepOneTM real-time PCR system (Applied Biosystems, Singapore). The relative gene expression was calculated using the 2−11Ct method (Livak and Schmittgen, 2001), and the methods were based on a previously described protocol (Shi et al., 2016). All reactions were conducted with three biological replicates.

#### Statistical Analysis

All data were statistically analyzed using SPSS 20.0 for Windows, and significance was assigned at the P < 0.05 level using Duncan's multiple comparisons test.

#### RESULTS

#### Plant Growth

The biomass of the cucumber seedlings was measured 9 days post-inoculation (dpi) with the pathogen. The plant height, shoot freshness and dry mass of NSY50-treated seedlings were significantly greater than those of plants grown under control conditions (**Table 1**). However, pathogen inoculation (treatment FOC) seriously decreased the growth of the cucumber seedlings, the plant height, shoot fresh weight, dry weight and root volume and the root fresh weight and dry weight, which were significantly suppressed compared to the control. Moreover, under pathogen stress, plant biomass was significantly higher in P. polymyxa NSY50 pre-treated cucumber plants, which showed a 1.2–1.5-fold up-regulation in plant growth compared to untreated plants. Thus, the results clearly indicated that the pathogen inoculation inhibited growth in cucumber seedlings and that P. polymyxa NSY50 can release at least a portion of the stress and promote plant growth.

#### Ultrastructure of the Roots

Under control conditions, the vacuoles had small sizes, and many nuclei and mitochondria could be seen clearly under the microscope (**Figure 1**). The cellular ultrastructure of cucumber roots seemed to show no difference between the control plants and plants treated with NSY50 alone (NSY50). However, the inoculation of FOC (FOC) caused remarkable structural changes. Specifically, the karyotheca appeared to have contracted, and the number of the mitochondria decreased compared to the control. Furthermore, the number of central vacuoles and autophagosomes appeared to indicate that the pathogen attack advanced the cell aging and death process. Pre-treatment with NSY50 (NSY50+FOC) apparently enhanced the stability of the cell structure.

#### Identification and Functional Classification of Proteins

Approximately 350 reproducible protein spots were detected on the 2-DE gels, and 56 differentially expressed protein spots (changes ≥ 1.5-fold) were successfully identified by MALDI-TOF/TOF MS (**Figure 2**; Supplementary Figure 1). These differentially expressed proteins are listed in **Table 2**.

The identified proteins were grouped into seven different categories on the basis of biological processes to which they were related according to the Gene Ontology and Uniprot Protein Knowledgebase (**Figure 3A**). The categories with a high level of expression variation included the defense response (28.6%), protein metabolism (25.0%), carbohydrate and energy metabolism (19.6%) and amino acid metabolism (17.9%). Among the 56 differentially expressed spots, 44 protein spots were significantly regulated by the pathogen FOC compared to the control (**Figure 3B**), and there were 16 up-regulated spots and 28 down-regulated spots (**Figure 3C**). The defense-response-related proteins had the highest upregulation rate, and the decreased proteins were involved in protein metabolism (10 spots), carbohydrate, and energy metabolism (7 spots), and amino acid metabolism (7 spots). Additionally, 32 protein spots were significantly induced by NSY50 inoculation compared to the control (**Figure 3B**). Of those spots, 25 were up-regulated, and seven were downregulated. The defense-response-related proteins (10 spots) and amino acid metabolism (5 spots) were the most highly enriched categories (**Figure 3C**). However, compared to FOC, 33 proteins had changes, including 24 proteins that increased and 9 proteins that decreased after pre-treatment with NSY50 (NSY50+FOC). The category of carbohydrate and energy metabolism had the highest up-regulation rate, whereas the decreased proteins were distributed among the defense response,


*Each value is the mean* ± *SE of three replicates. Different letters indicate significant differences at P* < *0.05 according to Duncan's multiple range tests. Control, seedlings cultured in normal nutrient solution; NSY50, control*+ *P. polymyxa NSY50 (2.5* × *106 CFU/mL); FOC, control*+ *FOC (2.5* × *105 conidia/mL); NSY50*+*FOC, pre-treated with NSY50 (2.5* × *106 CFU/mL) for 3 days, then challenged with FOC (2.5* × *105 conidia/mL).*

protein metabolism, secondary metabolism and cell-related protein categories (**Figure 3C**).

#### Clustering Analysis of Differentially Expressed Proteins

To acquire a comprehensive overview of the differentially expressed proteins that were affected by NSY50 under control conditions and FOC stress, hierarchical clustering was performed, and proteins with similar expression patterns were grouped together (**Figure 4**). Cluster A consisted of 4 proteins (spots 9, 6, 13, and 10) that were up-regulated by NSY50, down-regulated after inoculation with FOC compared to the control, but recovered by the application of NSY50 (NSY50+FOC). Cluster B was composed of 23 proteins that decreased considerably under FOC stress with/without NSY50 inoculation. Most of the identified proteins in this cluster were involved in protein metabolism, carbohydrate and energy metabolism. Cluster C involved 7 proteins that were up-regulated by NSY50 under both control conditions and FOC stress conditions, and these proteins corresponded to defense responses, carbohydrate and energy metabolism, amino acid metabolism, and protein metabolism. Cluster D included 10 proteins that were downregulated by NSY50 but up-regulated by the inoculation of FOC. Most of these proteins were involved in defense responses. Cluster E contained 4 proteins that were downregulated by NSY50 but up-regulated by the inoculation of FOC, regardless of whether they were pre-inoculated with NSY50. Cluster F included 8 proteins that increased in abundance when inoculated with NSY50 or FOC alone

total proteins were separated by IEF/SDS-PAGE and then stained with Coomassie Brilliant Blue (R-250) and loaded on each 18 cm gel strip (pH 4–7, linear). The numbers for the 56 differentially expressed proteins are marked and annotated according to the numbering in Table 2. Control, seedlings cultured in normal nutrient solution; NSY50, control+ *P. polymyxa* NSY50 (2.5 × 10<sup>6</sup> CFU/mL); FOC, control+ FOC (2.5 × 10<sup>5</sup> conidia/mL); NSY50+FOC, pre-treated with NSY50 (2.5 × 10<sup>6</sup> CFU/mL) for 3 days then challenged with FOC (2.5 × 10<sup>5</sup> conidia/mL).

but were down-regulated in combined inoculation treatment (NSY50+FOC).

### Validation of Differentially Expressed Proteins

As shown in **Table 2**, five differentially expressed protein spots belonging to ENO and eight SAMs protein spots were successfully identified by MALDI-TOF/TOF MS, and they were all significantly regulated by inoculation with NSY50 and/or FOC. This outcome indicated that ENO and SAMs were considered important plant proteins for coping with pathogen invasion. Therefore, ENO and SAMs were analyzed by western blotting to verify the proteomic data. The result showed a similar tendency, in that the expression of ENO and SAMs were slightly more highly regulated by the inoculation of NSY50 but significantly down-regulated by FOC infection compared to the control. However, the expression of both of these proteins was recovered by pre-treatment with NSY50 (NSY50+FOC) compared to FOC, regardless of whether measurements were taken 1, 3, or 9 days post-inoculation with FOC (**Figures 5A,C**; Supplementary Figures 2–5). Additionally, the level of SAMs was seriously decreased by FOC inoculation on the first day after inoculation (**Figure 5C**).

#### Expression Analysis of Several Identified Protein-Related Genes

The ENO and SAMs gene expression levels were then analyzed 1, 3, and 9 days post-inoculation. FOC up-regulated both proteins

#### TABLE 2 | Root proteins responsive to NSY50 and/or FOC identified by MALDI-TOF/TOF MS.


*(Continued)*

#### TABLE 2 | Continued


*(Continued)*

#### TABLE 2 | Continued


*<sup>a</sup>Spot number corresponding with 2-DE gel as shown in Figure 2.*

*<sup>b</sup>Number of identified peptides.*

*<sup>c</sup>Percentage of sequence coverage by matched peptides.*

*<sup>d</sup>The values higher than 1.5 or lower than 0.67 indicate significant changes.*

at all time points compared to the control (**Figures 5B,D**). The gene expression of ENO after NSY50 treatment was significantly lower compared to the FOC treatment. However, SAMs showed a contrary tendency in the early stages of inoculation (**Figure 5D**), and the transcript levels in NSY50+FOC were higher and increased more rapidly compared to those of plants that were inoculated with only FOC (FO C) from 1 to 3 days postinoculation.

Seven other genes related to the identified proteins, including GST, SAMDC, ACS, ACO1, ACO2, HSP70, and OPR1, were selected and subjected to expression pattern analysis. NSY50 up-regulated almost all of these genes (except ACS and ACO1) compared to the control (**Figure 6**). FOC stress up-regulated the expression of GST, SAMDC and ACO1 but decreased the ACS and ACO2 expression. The combined inoculation (NSY50+FOC) treatments up-regulated expression of almost all of these genes (except SAMDC and ACS) compared with the FOC treatment. We summarize the metabolism pathways that connect the expression of these genes, the proteins related to an antioxidant response, amino acid metabolism, and the EMP-TCA cycle in **Figure 7**.

#### DISCUSSION

Biological control is one of the most beneficial methods because it is both efficient and environmentally friendly for plant protection (Bargabus et al., 2003; Tjamos et al., 2005). Currently, the commercialization of bio-control agents is still limited (Szewczyk et al., 2006) due to complicated mechanisms of bio-control that have yet to be clarified (Niu et al., 2011). Paenibacillus polymyxa has been considered a bio-control agent for a wide range of plant pathogens (Dijksterhuis et al., 1999; Ling et al., 2011; Hong et al., 2016). It is an important Bacillus species because of its ability to produce different bioactive compounds with a broad spectrum of activities or induce systemic resistance against a variety of environmental stresses (Kim et al., 2015; Zhou et al., 2016). With the goal of commercializing P. polymyxa NSY50 in the future, we analyzed the proteins in cucumber roots that responded to NSY50 under FOC stress conditions to investigate the promotion and protection mechanisms of NSY50. The results of our proteome analysis are discussed below.

#### Proteins Related to Defense

Reactive oxygen species (ROS) have been demonstrated to be involved in symbiotic interactions between plants and microorganisms (Scheler et al., 2013). During pathogen invasion, plants have always suffered from ROS and reactive aldehyde bursts, which often lead to cellular damage (Niu et al., 2011; Gupta et al., 2014; Sengupta et al., 2015; Camejo et al., 2016). Fortunately, plants have developed an antioxidant enzymatic system to scavenge for and collect these toxic compounds. In the present study, a total of 16 spots were identified as defense-related proteins, and most of them were regulated by the inoculation of NSY50 and/or FOC (**Table 2**, **Figure 3C**).

Two superoxide dismutases (SOD, spots 25, and 27) were identified and were significantly up-regulated by pre-treatment with NSY50 (NSY50+FOC) compared to FOC treatment. Three peroxidases (POD, spots 19, 24, and 54) with different subunits were identified and showed different accumulation patterns. Regardless of whether they were inoculated with NSY50 or FOC, the expression levels were both up-regulated compared to the control. Additionally, spots 19 and 24 showed expression levels in NSY50+FOC that were higher than the proteins inoculated with FOC alone. Ascorbate peroxidase (APX, spot 21) was also

up-regulated by NSY50 (NSY50) compared to the control. It is known that the activation of antioxidant enzymes is a major plant cell method for protection (Youssef et al., 2016). Several studies confirmed the ability of bio-control agents to mitigate the effects of oxidative bursts by increasing the activity of antioxidant enzymes (Israr et al., 2016; Youssef et al., 2016). Thus, the results of the present study indicated that application of NSY50 played an active role in determining the abundance of antioxidant enzymes.

GSTs belong to a family of multifunctional enzymes that play critical roles in protecting tissues from oxidative damage by quenching reactive molecules (An et al., 2016). Previous studies have reported that GSTs can enhance resistance to the virulent bacterial pathogen Pseudomonas syringae pv. tomato DC3000 in Arabidopsis (Jones et al., 2006) and were increased in response to various hormones, such as salicylic acid, ethylene, methyl jasmonate, and auxin, and to biotic and abiotic stress (Lieberherr et al., 2003; Faltin et al., 2010). Additionally, Kwon et al. (2016) found that four glutathione S-transferases were significantly up-regulated by plant-growthpromoting rhizobacterium P. polymyxa E681 in Arabidopsis and suggested that the activation of defense-related proteins may the

respectively.

reason for resistance to the fungal pathogen. It was reported that the overexpression of GSTs induced by another PGPR, P. fluorescens, had an essential role in the ISR by priming rice plants and protecting cells from oxidative damage (Kandasamy et al., 2009). In the present study, two glutathione S-transferases (GST, spots 32, and 37) were identified. One of them (spot 32) was significantly up-regulated in NSY50, and the other (spot 37) showed expression levels in NSY50+FOC that were much higher than those in FOC. The qRT-PCR analysis also showed that the application of NSY50 significantly up-regulated this gene expression, regardless of whether the plants were inoculated with FOC (**Figure 6**). This result indicated that NSY50 could increase GST content to remove toxins in cucumber seedlings.

Plant thioredoxins appear to play a fundamental role in plant tolerance of oxidative stress by supplying power to reductases detoxifying lipid hydroperoxides or repairing oxidized proteins (Vieira Dos Santos and Rey, 2006). Thioredoxin H-type 1 isoform X2 (TRX, spot 52) belongs to the thioredoxins family and was up-regulated by the inoculation of NSY50 and/or FOC. This result was consistent with the results of Laloi et al. (2004), who reported that the Arabidopsis cytosolic thioredoxin h5 gene was induced by oxidative stress and its response to a

pathogen elicitor. In addition, Sun et al. (2010) found that the overexpression of NtTRXh3 (An h-type thioredoxin in tobacco) conferred resistance to Tobacco mosaic virus and Cucumber mosaic virus, both of which had reduced multiplication and pathogenicity in NtTRXh3- overexpressing plants compared to controls. These reductions indicated that TRX participated in defense mechanisms linked to the oxidative bursts resulting from pathogen attack.

In this study, a probable aldo-keto reductase 4 (AKRs, spots 50) was successfully identified, which showed that inoculation with NSY50 significantly up-regulated the expression of aldoketo reductase (AKRs), with or without FOC. The AKRs superfamily comprises a large number of primarily monomeric protein members that reduce a broad spectrum of substrates, which range from simple sugars to potentially toxic aldehydes (Sengupta et al., 2015). There are a number of interesting reports on the active involvement of plant AKRs in detoxification of stress-induced reactive carbonyls as plant defense systems against biotic stress factors, such as pathogenic attack (Tremblay et al., 2010; Santos et al., 2012; Xu et al., 2013). This may be the first report on a type of protein induced by the PGPR that actively contributes to defense against biotic stressors.

Other defense response proteins, such as heat shock protein 70 (HSP70, spots 4 and 12) and MLP-like protein 328 (MLP328, spot 26), were also identified in this study. HSP70 can be induced by various environmental stresses, such as cold, salt, high temperature, and hypoxic stress, and this protein often acts as a molecular chaperone (Xu et al., 2011; Li et al., 2013b; Yuan et al., 2016) to prevent the irreversible aggregation of other proteins and regulate the folding and accumulation of proteins (Feder and Hofmann, 1999). In this study, it was observed that NSY50 enhances the expression of HSP70 (spot 4) by 3.31-fold compared to the control. Whereas, FOC significantly decreased HSP70 expression both under control conditions and pre-treatment with NSY50 (**Table 2**), the combined inoculation (NSY50+FOC) showed much higher up-regulation than FOC, although there was no significant difference. The transcription of HSP70 showed a significant increase in NSY50+FOC compared to inoculation with FOC alone (**Figure 6**). This result indicates that the PGPG are capable of enhancing the expression of this important protein. This result was consistent with the research of Li et al. (2013b), who reported that PGPR P. putida UW4 significantly enhanced the level of plant heat shock proteins under hypoxic stress to improve stress resistance. The role of

the MLP-like protein was unclear, and it was hypothesized to be a physiological defense protein that responded to stress (Fukao et al., 2009; An et al., 2016). However, in this study, MLP328 was significantly up-regulated, with a 3.13–3.68-fold increase following the inoculation of NSY50 and/or FOC compared to the control. The mechanism of this type of physiological defense protein needs further study.

#### Protein Metabolism

Of the differentially accumulated proteins, 14 were identified as responding to FOC and/or NSY50 treatments; these responses were attributed to protein metabolism, and the proteins could be divided into five functional groups (**Table 2**). The first group includes those involved in protein folding and assembly, including four protein disulfide-isomerase (PDI, spots 14, 42, 44, and 45) and a 20 kDa chaperonin (CPN20, spot 23). Both the NSY50 and FOC treatments enhanced the expression of PDI (spot 14) by 1.90–2.42-fold compared to the control. Several studies have shown that the overexpression of PDI in plants that are challenged with pathogens has been thought to be related to the correct folding of defense proteins (Gruber et al., 2007; Palomares-Rius et al., 2011; Cipriano et al., 2015). However, the remaining three PDI proteins (spots 42, 44, and 45) were significantly decreased by pathogen attack (inoculated with FOC). This effect may be due to the stresses that were inhibited expression of this type of protein, which was consistent with previous reports (Yuan et al., 2016). However, another report showed that the application of PGPR recovered the expression level of this protein under stress conditions (Li et al., 2013b). The CPN20 (spot 23), which was among the most-represented proteins that emphasize the impact of stress on post-translation modification/ modification machinery (Zhang et al., 2015) showed similar results when inoculated with FOC.

The second group included three spots of eukaryotic translation initiation factor 5A (eIF-5A, spots 33, 38, and 39), one eukaryotic translation initiation factor 5A-2 (eIF-5A2, spot 56) and one elongation factor 2 (EF2, spot 47), which are involved in the initiation and elongation stage of mRNA translation and protein synthesis (Thompson et al., 2004; Yuan et al., 2016). Overall, the FOC inoculation significantly decreased the eIF-5A expression levels in cucumber seedling roots, whereas pre-treatment with NSY50 (NSY50+FOC) enhanced the expression of eIF-5A. However, the eIF-5A2 protein had a higher accumulation after FOC inoculation conditions both with and without NSY50 compared to non-stress conditions. The protein eIF-5A2 was reported to likely be involved in the regulation of apoptotic cell death (Feng et al., 2007). EF2 was significantly upregulated, with a 2.70-fold increase after inoculation with NSY50 compared to the control, whereas inoculation with FOC showed only a slight increase. These results indicated that the PGPR could balance ribosomal particles and protein synthesis, which was consistent with previous reports (Li et al., 2013b).

The third group included proteasome subunit alpha type-2-A (PRs, spot 22) and thiol protease aleurain-like (TPA, spot 41), which are involved in protein degradation. The proteasome was assigned a precise role in the degradation of the oxidized proteins generated by many stresses, which avoided the inhibition of the plant's metabolic pathway (Polge et al., 2009). In this study, infection with FOC down-regulated the expression level of TPA, and the plants metabolic pathway may have been damaged by this destructive fungal pathogen, which was consistent with the TEM result. The PRs significantly

decreased under FOC stress but were remarkably up-regulated by inoculation with NSY50, which indicated that the degradation of proteins with oxidative damage was enhanced (An et al., 2016). The fourth group was protein transport and included one protein called nuclear transport factor 2-like (NTF2, spot 28). The NTF2 facilitated protein transport into the nucleus (Zhao et al., 2006). Both inoculation with NSY50 and FOC were significantly up-regulated the expression of this protein. The ubiquitin-conjugating enzyme E2 variant 1C (UEV1C, spot 30) belonged to the fifth group, protein modification, and it participates in protein modification by catalyzing the covalent attachment of ubiquitin to proteins (Yuan et al., 2016). In this study, FOC significantly decreased the expression of UEV1C, whereas pre-treatment with NSY50 (NSY50+FOC) enhanced its expression. These results suggest that the application of NSY50 alleviated the damage to protein metabolism that was induced by F. oxysporum.

#### Carbohydrate and Energy Metabolism

The proteins related to carbohydrate and energy metabolism, particularly the proteins involved in the glycolysis (EMP) and tricarboxylic acid (TCA) cycle were significantly changed and are shown in **Table 2**. A total of 11 identified proteins in response to two strains were found and were divided into four functional groups. The first group focused on glycolysis and included five spots with enolase isoform (ENO, spots 1, 2, 17, 18, and 46), protein DJ-1 homolog D (DJ1D, spot 13), probable fructokinase-4 (FRK, spot 20) and fructose-bisphosphate aldolase (FBA, spot 55). The others were involved with the tricarboxylic acid cycle, including malate dehydrogenase (MDH, spot 36), sucrose metabolism protein acid beta-fructofuranosidase-like (BFRUCT, spot 43), and energy metabolism protein ATP synthase (ATPase, spot 3). Almost all of them were down-regulated by FOC and recovered by treatment with NSY50 (NSY50+FOC), except FBA and BFRUCT.

Enolases, known as 2-phospho-D-glycerate hydrolases, are one of the most important enzymes in glycolysis. They catalyze the dehydration of 2-phosphoglycerate into phosphoenolpyruvate. Five spots were identified as enolases in this study, and western blotting was used to verify proteomic data (**Table 2**, **Figure 5A**; Supplementary Figures 2, 3) and showed a similar tendency, regardless of whether measurements were taken 1, 3, or 9 days post-inoculation with the pathogen FOC. The gene expression showed a reverse trend, and inoculation with FOC significantly up-regulated its gene expression (**Figure 5B**). Interestingly, protein abundance significantly recovered along with inoculation time (**Figure 5A**). Several studies reported that both the ENO protein and the gene expression increased under salt stress and attempted to generate more energy to cope with the stress (Ndimba et al., 2005; Zhong et al., 2015; An et al., 2016; Yuan et al., 2016).

The DJ-1D protein (DJ1D, spot 13), which was reported to have several functions, including functioning as an antioxidant (Taira et al., 2004), is a redox-dependent chaperone (Krebiehl et al., 2010) and can convert glyoxals to carboxylic acids (Kwon et al., 2013). Glyoxals are reactive 2-oxoaldehydes that are found in cells under various stress conditions (Thornalley, 1996). In this study, DJ1D was significantly up-regulated, with 4.25- and 2.28-fold increases after inoculation with NSY50 and/or FOC, respectively.

Li et al. (2013b) reported that the application of PGPR P. putida UW4 adjusted the EMP-TCA metabolism under hypoxic stress. In contrast, Kwon et al. (2016) found that some of the proteins related to the glycolysis pathway, such as FBA and pyruvate dehydrogenase complex component E2, were significantly up-regulated by the PGPR strain P. polymyxa E681 in Arabidopsis roots. Carbohydrate metabolism provides the plant with required carbon, which is critical for the production of new tissues, and also can have profound effects on plant growth through modulation of cell division and expansion (Eveland and Jackson, 2012; Kwon et al., 2016). In this study, we found that the application of P. polymyxa NSY50 increased the abundance of FRK, ENO, BFRUCT, and ATPase compared to control conditions. Additionally, NSY50 partially recovered the ENO, FRK, DJ1D, MDH, and ATPase expression, which had been decreased by FOC. These results suggested that the application of NSY50 could promote EMP-TCA pathway activity. Presumably, these results imply there was an enhancement in energy generation and carbohydrate production to promote plant growth and resist stress damage.

#### Amino Acid Metabolism

Ten spots were identified as proteins involved in amino acid metabolism. Eight S-adenosylmethionine synthases (SAMs, spots 6, 7, 8, 9, 10, 15, 16, and 48) comprising different subunits were identified and markedly decreased (except spot 16) in response to FOC infection. SAMs catalyze the biosynthesis of S-adenosylmethionine (SAM) from methionine and ATP, and they are active in the biosynthesis of lignin, glycine betaine and polyamines (Fontecave et al., 2004). These proteins can be induced by various environmental stresses, such as cold, salt, drought stress and even pathogen infection (Li et al., 2013a; Huang et al., 2016), and most of them were down-regulated by stress (Wang et al., 2016; Yuan et al., 2016). This outcome was consistent with the results showing that the inoculation of FOC significantly decreased SAMs expression (**Table 2**, **Figure 5C**; Supplementary Figures 4, 5). However, pre-treatment with NSY50 up-regulated SAMs abundance both under control and pathogen infection conditions. The accumulation of SAMs in plants has been shown to be related to the enhance tolerance to various kinds of environmental stresses. Overexpression of suadea salsa SAMs gene promoted salt tolerance in transgenic tobacco (Qi et al., 2010). An increase in production of SAMs was observed in cold stressed rice (Cui et al., 2005), mechanically wounded Phaseolus lunatus (Arimura et al., 2002), salt-stressed barley (Witzel et al., 2009), and in response to cotton worm feeding in soybean (Fan et al., 2012). Hence, the remarkable expression of SAMs by NSY50 suggested that plant might promote SAM-dependant metabolism through synthesize more SAM to further enhance the synthesis of new proteins and stress tolerance.

Glutamine synthetase (GS) is a key enzyme in plant N assimilation that catalyzes the combination of ammonia and glutamate into glutamine (Nam et al., 2011; Rogic et al., ´ 2015). Two types of GS (spots 11 and 35) were identified and showed different accumulation patterns. However, they were both upregulated by NSY50, regardless of whether the plants were inoculated with FOC. This result suggested that the application of NSY50 could lead to a high accumulation of amino acid metabolism, and this was probably one way in which NSY50 promoted plant growth and intensified resistance to pathogen infection for cucumber seedlings.

### Fatty Acid Metabolism Proteins

Jasmonic acid is a signaling compound that influences multiple cellular functions and plays crucial roles in the signaling network regulating the development of induced resistance, including systemic acquired resistance (SAR) and induced systemic resistance (Glazebrook, 2001). Niu et al. (2011) found that PGPR Bacillus cereus AR156 induced ISR to Pseudomonas syringae pv. tomato DC3000 in Arabidopsis via the simultaneous activation of salicylic acid-, jasmonic acid-, and ethylene-dependent signaling pathways. In the present study, 12-oxophytodienoate reductase 1 (spot 49), which is involved in jasmonate biosynthesis, was identified to be induced by both NSY50 and FOC inoculation. Additionally, NSY50 treatment showed a 2.02-fold increase in abundance compared to FOC. A good correlation was observed between this protein and transcript abundance, as etermined by qPCR analysis (**Figure 6**). This outcome indicated that P. polymyxa NSY50 may stimulate the jasmonate signal transduction network.

Enoyl-[acyl-carrier-protein] reductase [NADH] (ACP, spot 51), which is a subunit of the fatty acid synthase complex that catalyzes the de novo synthesis of fatty acids (Mou et al., 2000), was significantly up-regulated by inoculation with FOC. It was recently reported that ACP was sensitive to the diphenyl ether cyperin produced by certain pathogenic fungi in the tissues of infected plants, which likely contributed to the virulence of these disease agents (Dayan et al., 2008). The role of the ACP during pathogen infection remains unclear, and further study is required.

#### Secondary Metabolism and Cell-Related Proteins

The NSY50 was also found to regulate proteins involved in other metabolic pathways, such as secondary metabolism (spot 34) and cell-related protein (spots 5 and 29), which indicated that NSY50 regulated various metabolic pathways to counteract FOC infection. The mechanism of action will need to be studied in future investigations.

#### CONCLUSION

To the best of our knowledge, this is the first proteomicbased research to focus on the interactions between cucumber plants, the pathogen F. oxysporum, and PGPR strain P. polymyxa NSY50. A total of 56 protein spots were identified with a change more than 1.5-fold or less than the 0.67-fold in the protein abundance ratio in cucumber seedling roots after inoculation with P. polymyxa NSY50 and/or FOC. The majority of these proteins were related to defense responses, protein metabolism, carbohydrate and energy metabolism and amino acid metabolism. These proteins might work cooperatively to enhance resistance to FOC attack and keep plant growth on a mostly even keel. Our results showed that the improved plant

### REFERENCES


growth and defenses by P. polymyxa NSY50 might be associated with the following processes: (i) activation of stress defenserelated proteins by stress injuries that are alleviated by NSY50; (ii) NSY50 stimulating protein synthesis and degrading damaged proteins induced by FOC; (iii) up-regulation of proteins in the EMP-TCA route to provide more carbohydrates and energy to promote plant growth and resist debilitation; or (iv) adjusting the plant amino acid metabolism to increase biomass of the plant and regulate other metabolic pathways, such as jasmonic acid, to enhance resistance to FOC. This research enriched our understanding of the mechanisms of how PGPR mediates the stress response in plants and promotes protection from pathogen infection.

#### AUTHOR CONTRIBUTIONS

We thank the numerous individuals who participated in this research. SG designed the study and guided the research. ND and LS wrote the main manuscript text and performed the experiments. YY prepared all the figures and performed some of the experiments. BL contributed new reagents and analytical tools. SS and JS edited this manuscript. All authors reviewed and approved the manuscript.

#### FUNDING

This work was financially supported by the National Natural Science Foundation of China (No. 31672199, No. 31471869, and No. 31401919), the China Agriculture Research System (CARS-25-C-03), and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2016. 01859/full#supplementary-material


suppression of soilborne pathogens of cucumber and pepper. Appl. Microbiol. Biotechnol. 80, 115–123. doi: 10.1007/s00253-008-1520-4


oxysporum f. sp. Cucumerium Owen. Physiol. Mol. Plant Pathol. 96, 77–84. doi: 10.1016/j.pmpp.2016.09.002


**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 © 2016 Du, Shi, Yuan, Li, Shu, Sun and Guo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proteomic Analysis Reveals the Positive Effect of Exogenous Spermidine in Tomato Seedlings' Response to High-Temperature Stress

Qinqin Sang<sup>1</sup> , Xi Shan<sup>1</sup> , Yahong An<sup>1</sup> , Sheng Shu<sup>1</sup> , Jin Sun1, 2 and Shirong Guo1, 2 \*

*<sup>1</sup> Key Laboratory of Southern Vegetable Crop Genetic Improvement in Ministry of Agriculture, College of Horticulture, Nanjing Agricultural University, Nanjing, China, <sup>2</sup> Nanjing Agricultural University (Suqian), Academy of Protected Horticulture, Suqian, China*

Edited by:

*Jie Zhou, Zhejiang University, China*

#### Reviewed by:

*Caiji Gao, South China Normal University, China Hao Wang, South China Agricultural University, China*

> \*Correspondence: *Shirong Guo srguo@njau.edu.cn*

#### Specialty section:

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

Received: *18 October 2016* Accepted: *20 January 2017* Published: *06 February 2017*

#### Citation:

*Sang Q, Shan X, An Y, Shu S, Sun J and Guo S (2017) Proteomic Analysis Reveals the Positive Effect of Exogenous Spermidine in Tomato Seedlings' Response to High-Temperature Stress. Front. Plant Sci. 8:120. doi: 10.3389/fpls.2017.00120* Polyamines are phytohormones that regulate plant growth and development as well as the response to environmental stresses. To evaluate their functions in high-temperature stress responses, the effects of exogenous spermidine (Spd) were determined in tomato leaves using two-dimensional electrophoresis and MALDI-TOF/TOF MS. A total of 67 differentially expressed proteins were identified in response to high-temperature stress and/or exogenous Spd, which were grouped into different categories according to biological processes. The four largest categories included proteins involved in photosynthesis (27%), cell rescue, and defense (24%), protein synthesis, folding and degradation (22%), and energy and metabolism (13%). Exogenous Spd up-regulated most identified proteins involved in photosynthesis, implying an enhancement in photosynthetic capacity. Meanwhile, physiological analysis showed that Spd could improve net photosynthetic rate and the biomass accumulation. Moreover, an increased high-temperature stress tolerance by exogenous Spd would contribute to the higher expressions of proteins involved in cell rescue and defense, and Spd regulated the antioxidant enzymes activities and related genes expression in tomato seedlings exposed to high temperature. Taken together, these findings provide a better understanding of the Spd-induced high-temperature resistance by proteomic approaches, providing valuable insight into improving the high-temperature stress tolerance in the global warming epoch.

Keywords: antioxidant, high-temperature stress, proteomics, spermidine, tomato

## INTRODUCTION

High-temperature stress represents one of the most frequent abiotic stresses worldwide, inducing several physiological and biochemical processes in cells, and limiting the growth and productivity of plants (Bita and Gerats, 2013). Plants respond to high temperature by reprogramming their proteome, metabolome and transcriptome to establish a new steady-state balance of metabolic processes (Kosová et al., 2011; Lin H. H. et al., 2015; Sruthi et al., 2016).

Polyamines (PAs) are ubiquitous low-molecular-weight aliphatic amines, and include putrescine (Put), spermidine (Spd), and spermine (Spm). PAs are known to participate in the regulation of physiological and developmental processes (Liu et al., 2007; Gupta et al., 2013), and they are also involved in the defense reaction of plants against various environmental stresses (Todorova et al., 2007; Berberich et al., 2015; Pál et al., 2015). The integration of environmental stimuli, signal transduction and the stress response is mediated, at least partially, by intensive cross-talk among plant hormones (Wahid et al., 2007). Recent studies indicated that polyamines act as cellular signals in the intricate cross talk with different metabolic routes and complex hormonal pathways (Pál et al., 2015). The exogenous Spd enhancement of high-temperature stress tolerance via the involvement of antioxidant ability and photosynthetic efficiency had been described (Tian et al., 2012; Mostofa et al., 2014), but little information about Spd regulating proteomic changes under the high-temperature stress is available.

As mRNA abundance is not enough to provide information about the proteins, proteomic analysis has become a powerful tool to elucidate the mechanisms of plant stress tolerance (Skalák et al., 2016). Previous studies reported that PAs could bind to charged spots at protein interfaces and modulate electrostatic protein–protein interactions to regulate the protein functions (Berwanger et al., 2010). Exogenous polyamines had been found to activate multiple pathways that conferred increased salt and drought tolerances in bermudagrass by reprogramming the proteome (Shi et al., 2013). Li et al. (2013) and Yuan et al. (2016) showed that application of Spd/Put changed the expression of proteins and contributed to counteract the damage induced by salt stress in cucumber seedlings. Igarashi and Kashiwagi (2015) reported that polyamines could stimulate the synthesis of proteins at the translation level due to the formation of a polyamine-RNA complex.

The tomato (Lycopersicon esculentum) is one of the most important vegetables from both the nutritional and economic points of view. The effects of exogenous Spd in enhancing the stress tolerance had been described in cucumber (Tian et al., 2012) and in rice (Mostofa et al., 2014). However, little information is available to explain the specific mechanisms by which PAs regulate the high-temperature stress responses through a proteomic approach. In this study, we investigated the differentially expressed proteins in tomato leaves through 2-dimensional gel electrophoresis to better understand the underlying mechanisms of Spd application in high-temperature stress resistance.

### MATERIALS AND METHODS

#### Plant Materials and Treatments

Tomato (Lycopersicon esculentum Mill. cv. Puhong 968) seeds were obtained from the Shanghai Academy of Agricultural Sciences, China. Seeds were germinated and grown in plastic nutrition pots filled with growth media (Zhenjiang Peilei Co., Ltd., China). The germinated seedlings were grown under controlled condition (light intensity, 600 µmol m−<sup>2</sup> ·s −1 ; day/night temperature, 25/18◦C; light/dark photoperiod, 14 h/10 h; relative humidity, 55–65%) in growth chambers (Ningbo Jiangnan Instrument Factory, Ningbo, China).

After the third true leaf developed, the seedlings were subjected to high-temperature (day/night temperature, 38/28◦C; light/dark photoperiod, 14/10 h; relative humidity, 55–65%). The experimental plots included four different treatments: (1) Cont; (2) Spd (1 mM); (3) HT; (4) HT+ Spd (1 mM). The concentration of Spd was selected on the basis of previous experiment (data not shown). One millimole Spd was sprayed to leaves at 17:00 every day, and the control plants were sprayed with distilled water. After 7 days of treatment, the third fully expanded tomato leaves of each treatment were stored at −80◦C for physiological and proteomic analysis. The experiment was arranged in a randomized complete block design and biological replicates were independently carried out three times.

#### Measurement of Dry Weight, Chlorophyll Content, and Net Photosynthetic Rate (Pn)

The tomato seedlings were washed with sterile distilled water. After wiped with gauze, samples were dried in an oven at 105◦C for 15 min followed by 75◦C for 72 h, until reaching a constant weight, and then weighed for dry weight. Chlorophyll was extracted with a mixture of acetone, ethanol and water (4.5: 4.5: 1 by volume) and its content was estimated using the method of Arnon (1949). Pn was measured using a portable photosynthesis system (LI-6400, LI-COR Inc, USA).

#### Protein Extraction

Protein extraction was performed according to a modified version of the trichloroacetic acid (TCA) acetone precipitation method described by Hurkman and Tanaka (1986). Frozen leaf tissues were ground in liquid nitrogen and suspended in icecold extraction buffer (8 M urea, 1% (w/v) dithiothreitol (DTT), 4% (w/v) CHAPS and 40 mM Tris). Then the homogenates were centrifuged at 15,000 × g for 20 min at 4◦C, and the supernatants were precipitated overnight with ice-cold acetone containing 10% (w/v) TCA and 0.07% (v/v) β-mercaptoethanol. The resulting protein-containing suspensions were centrifuged at 20,000 × g for 30 min at 4◦C, and then the protein pellets were washed three times with cold acetone containing 0.07% (v/v) β-mercaptoethanol. Finally, the protein pellets were air-dried at room temperature and dissolved in rehydration buffer (8 M urea, 1 M thiourea, 2% w/v CHAPS). The protein concentrations were determined by the methods of Bradford (1976) using bovine serum albumin as the standard, and then the protein was stored at −80◦C until being subjected to two-dimensional gel electrophoresis (2-DE).

### 2-DE

For first dimensional isoelectric focusing (IEF), IPG strips (GE Healthcare, San Francisco, CA, USA, 17 cm, pH 4–7 linear gradient) were used according to the methods of Li et al. (2013). The dry IPG strips were rehydrated at room temperature for 12– 16 h in 350µL rehydration solution [8 M (w/v) urea, 1 M (w/v) thiourea, 2% (w/v) CHAPS, 65 mM DTT, 0.8% (v/v) IPG buffer 4–7, and 1% (w/v) bromophenol blue)]. Following rehydration, the IPG strips were run on an Ettan IPGphor 3 (GE Healthcare, USA) with a gradient of 100 V (1 h), 200 V (1 h), 200 V (1 h), 500 V (1 h), 1000 V (1 h), 4000 V (1 h), and 10,000 V (1 h), finally reaching a value of 75,000 V h. The working temperature was maintained at 20◦C with a maximum current of 50 mA per strip. After the first dimension, the IEF strips were equilibrated for 15 min in equilibration solution I [1% (w/v) DTT, 6 M urea, 30% (v/v) glycerol, 2% (w/v) SDS, and 50 mM Tris–HCl (pH 8.8)], and then in equilibration solution II [2.5% (w/v) iodoacetamide, 6 M urea, 30% (v/v) glycerol, 2% (w/v) SDS, and 50 mM Tris–HCl (pH 8.8)] for 15 min.

The second dimensional SDS-polyacrylamide gel electrophoresis was performed on running gels (Hoefer SE600 Ruby Standard Vertical System, GE Healthcare; 12.5% polyacrylamide) as described by Laemmli (1970). The strips were embedded on the top of the SDS gel and then sealed with 1% molten agarose solution. Electrophoresis was carried out at 15 mA per gel until the bromophenol blue dye reached the bottom of the gel. After the 2-DE, the gels were stained overnight with Coomassie Brilliant Blue (CBB) R-250 solution (0.1% (w/v) of CBB R-250 in 1:4:5 (v/v) methanol: acetic acid: deionized water) and destained with a 1:1:8 (v/v) methanol: acetic acid: deionized water solution with several changes, until a colorless background was achieved.

#### Image and Data Analysis

The 2-D gels were scanned with an Image Scanner III (GE Healthcare, San Francisco, USA). Spot detection, quantification and matching were performed with ImagemasterTM 2D Platinum software (version 6.0, GE Healthcare, San Francisco, USA). The intensity of each spot on the 2-D gels was quantified based on the volumes percentage (vol. %). Only spots with significant changes (at least 1.5-fold quantitative changes, P < 0.05) were considered to be differentially expressed.

### Protein Identification

The protein spots were excised from the polyacrylamide gels, and identified using MALDI-TOF/TOF MS by an Ultraflex II mass spectrometer (Applied Biosystems, Foster City, CA, USA). The resulting peptide mass lists were searched in NCBI (http:// www.ncbi.nlm.nih.gov) using the software MASCOT version 2.1 (Matrix Science, London, UK). The parameter criteria were as follows: trypsin cleavage, one missed cleavage allowed; carbamidomethyl (C) set as a fixed modification; oxidation of methionines allowed as a variable modification; peptide mass tolerance within 100 ppm; fragment tolerance set to ± 0.4 Da; and minimum ion score confidence interval for MS/MS data set to 95%.

The classification of the identified proteins was performed by searching in the UniProt Knowledgebase (UniProtKB, http:// www.uniprot.org).

### Hierarchical Cluster Analysis and Interaction Network

The hierarchical clustering of the protein expression patterns was performed on the log<sup>2</sup> transformed vol. % of each protein spot using Cluster software (version 3.0). The complete linkage algorithm was enabled, and the results were plotted using Treeview software (version 1.60).

Mapping of the interaction network was performed using the STRING database (http://string.embl.de) based on conformed and predicted interactions.

#### Enzyme Activity Analysis

Ascorbate peroxidase (APX, EC 1.11.1.11) activity was determined according to Nakano and Asada (1981) by measuring the rate of ascorbate oxidation at 290 nm (ε = 2.8 mM−<sup>1</sup> cm−<sup>1</sup> ). Dehydroascorbate reductase (DHAR, EC 1.8.5.1) activity was calculated from the change in absorbance at 265 nm and the extinction coefficient of 14 mM−<sup>1</sup> cm−<sup>1</sup> , as described by Nakano and Asada (1981). Superoxide dismutase (SOD, EC 1.15.1.1) activity was calculated by inhibiting the photochemical reduction of NBT at 560 nm. One unit of SOD activity was defined as the amount of enzyme that caused 50% inhibition of NBT reduction rate (Becana et al., 1986).

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

The total RNA was extracted from the tomato leaf tissues as described in the TRI reagent protocol (Takara Bio Inc.). The total RNA and cDNA syntheses were performed according to the manufacturer's instructions. The primers were designed according to the NCBI (Supplementary Table 1). qRT-PCR was performed with the SYBR PrimeScriptTM RT-PCR Kit (Takara Bio Inc.) according to the manufacturer's instructions. All experiments were repeated three times and the relative gene expression was calculated by the 2−11Ct method.

### Statistical Analysis

All biochemical analyses were conducted at least three times. Data were statistically analyzed with statistical software SPSS 17.0 (SPSS Inc., Chicago, IL, USA) using Duncan's multiple range test at the P < 0.05 level of significance.

## RESULTS

#### Morphological and Physiological Responses

After 7 days' treatment with exogenous Spd, no significant differences were observed in the tomato leaves under nonstressful conditions. Phenotypic observations showed that the untreated high-temperature stressed seedlings exhibited chlorosis and yellowing, whereas the Spd-treated seedlings had a better visual appearance (**Figure 1A**). Under the hightemperature stress, the dry weight, chlorophyll content and net photosynthetic rate (Pn) decreased by 33.0, 16.4, and 58.9%, respectively. However, exogenous Spd application resulted in improvements in these parameters (**Figure 1**).

### Proteomic Analysis

To reveal the protective effect of exogenous Spd on the tomato under high-temperature stress, a total of 67 differentially expressed spots were identified using 2-DE and MALDI-TOF-MS (**Figure 2**, **Table 1**, Supplementary Figure 1). To better understand which physiological process was regulated by Spd under the high-temperature stress, the identified proteins

Cont, control plants under 25/18◦C (day/night); Spd, plants under 25/18◦C with 1 mM Spd foliar spraying; HT, plants under 38/28◦C; HT+Spd, plants under 38/28◦C with 1 mM Spd foliar spraying. Each histogram represents a mean ± SE of three independent experiments (*n* = 3). Different letters indicate significant differences between treatments (*P* < 0.05) according to Duncan's multiple range tests.

were grouped into 7 categories based on their biological functions according to Gene Ontology (**Figure 3**). Among the 67 proteins, the majority were sorted into photosynthesis (27%), followed by cell rescue and defense (24%), protein synthesis, folding, and degradation (22%), energy and metabolism (13%), amino acid metabolism (5%), signal transduction (5%), and unknown (4%).

Compared with the control, there were 33 up-regulated spots and 32 down-regulated spots in response to the hightemperature stress (**Figure 4A**). For the high-temperature stress induced proteins, the most highly enriched category was cell rescue and defense. However, exogenous Spd up-regulated 35 spots and down-regulated 26 spots compared with the untreated seedlings subjected to high-temperature stress, and of these proteins, the most prevalent category was photosynthesis (**Figure 4B**).

To obtain a comprehensive overview of the differentially expressed proteins, hierarchical cluster analysis was conducted to categorize the proteins that showed differential expression profiles affected by Spd under the normal and high-temperature stress conditions (**Figure 5**).

#### Antioxidant Enzymes and Related Genes Expression Analysis

The proteomic results revealed that the abundances of some antioxidant enzymes (spots 25, 29, 35, 36) were changed (**Figure 6A**), so we further analyzed the associated antioxidant enzyme activities (APX, DHAR, SOD) and related gene expressions (APX 2, APX 6, DHAR 1, DHAR 2, Fe SOD, Cu/Zn SOD). The activities of the enzymes showed significant decreases under high-temperature stress. However, exogenous Spd remarkably increased their activity compared with the high-temperature stress alone (**Figure 6B**). A similar trend was observed for the expression levels of most of the antioxidant enzyme related genes (**Figure 6C**).

FIGURE 2 | Coomassie Brilliant blue (R-250)-stained 2-DE gels. Spot numbers indicate the 67 identified differentially expressed proteins. The range of the molecular mass of protein markers was from 10 to 170 kDa.

#### Interaction Network Analysis

The proteins act together in the context of networks in cells, rather than performing their functions in an isolated manner (Bian et al., 2015). The STRING database provides a critical assessment and integration of protein–protein interactions, including direct (physical) as well as indirect (functional) associations. A network was used to show the interactions of the identified proteins and revealed the potential information at the protein level (**Figure 7**). Most energy metabolism related proteins (86.7%) and cell rescue and defense (68.8%) were involved in the protein–protein interaction network. Among the interaction proteins, the energy metabolism related proteins represented the highest proportion (35.1%). More importantly, GAPDH (spot 22) and phosphoglycerate kinase (spot 18) were the important junctions of interacting proteins in the network, suggesting that energy was of the utmost importance for the response to high temperature stress with exogenous Spd treatment.

#### DISCUSSION

Polyamines are known to effectively alleviate the plant growth inhibition by abiotic stress. In this study, exogenous Spd was shown to promote the growth and improve the photosynthetic capacity of the tomato under high-temperature stress (**Figure 1**), which is consistent with a previous finding in rice (Mostofa et al., 2014). 2-DE analysis was conducted, and 67 differentially regulated proteins were identified in response to high temperature and/or exogenous Spd (**Figure 2**, **Table 1**). The regulation of the metabolic processes by Spd and high temperature is discussed below.

#### Photosynthesis-Related Proteins

Photosynthesis is highly sensitive to high-temperature stress and is often inhibited before other cell functions are impaired (Mathur et al., 2014). Importantly, Rubisco and Rubisco activase (RCA) are the primary limiting factors of net photosynthesis under stress (Ahsan et al., 2007; Hu et al., 2015). In this study, we found that proteins related to Rubisco (spots 6, 28, 49) and RCA (spots 16, 64) markedly decreased in response to hightemperature stress, similar to other proteomic studies (Han et al., 2009; Lin K. H. et al., 2015). High temperature can reduce the activation state of Rubisco (Law and Crafts-Brandner, 1999), which is often attributed to the thermolability and loss of activity of RCA under high-temperature stress (Salvucci and Crafts-Brandner, 2004; Sharkey, 2005). However, exogenous Spd had positive effects on Rubisco and RCA in tomato leaves, suggesting

#### TABLE 1 | Leaf proteins responsive to high temperature stress and/or Spd identified by MALDI-TOF/TOF MS.


*(Continued)*

#### TABLE 1 | Continued


*(Continued)*

#### TABLE 1 | Continued


*<sup>a</sup>Spot numbers corresponding to spots in* Figure 1*.*

*<sup>b</sup>TpI and EpI are the theoretical isoelectric point and experimental isoelectric point, respectively.*

*<sup>c</sup>TMr and EMr are the theoretical molecular mass and experimental molecular mass, respectively.*

*<sup>d</sup>The total number of identified peptides.*

*<sup>e</sup>Percentage of sequence coverage by matched peptides.*

that the Calvin cycle and photosynthetic carbon assimilation were maintained at high levels, contributing to the biomass accumulation under high-temperature stress.

Ferredoxin-NADP reductase (spot 23) is the last enzyme in the transfer of electrons during photosynthesis from PS I to NADPH, producing NADPH for CO<sup>2</sup> assimilation (Fukuyama, 2004; Tian et al., 2015). Oxygen-evolving enhancer proteins (spots 32, 39, 57) are also involved in the light reaction of PS II, and are the most heat-susceptible part of the PS II apparatus (Vani et al., 2001). The abundances of ferredoxin-NADP reductase and the oxygen-evolving enhancer (OEE) decreased in response to hightemperature stress, but the expression significantly increased with the application of Spd compared with the stress alone, suggesting that Spd played an active role in the photosynthetic chain, resulted in a higher stability of PS II and an enhancement of oxygen evolving complex capacity, and then subsequently led to an enhancement of the photosynthetic capacity (Shi et al., 2013; Su et al., 2013).

Three spots were identified as proteins implicated in chlorophyll biosynthesis. Glutamate-1-semialdehyde 2,1 aminomutase (spots 11, 13) is an important enzyme to catalyze the formation of 5-aminolevulinic acid (ALA), a vital precursor of chlorophyll (Zhu et al., 2013). Coproporphyrinogen III oxidase 1 (spot 43) catalyzes the oxidative decarboxylation of coproporphyrinogen III to protoporphyrinogen IX in the chlorophyll biosynthesis pathways (Tian et al., 2015). Interestingly, the expression of chlorophyll biosynthesis proteins was increased under high-temperature stress, whereas the chlorophyll content was decreased (**Figure 1C**). One plausible explanation of this observation is that chlorophyll biosynthesis in plants is very complicated and co-regulated by many factors, but the temperature-related inhibition of the enzyme activity could be an important reason for the inhibition of the chlorophyll biosynthesis.

#### Cell Rescue and Defense

Plants have evolved a complex sensory mechanism to monitor and adapt to prevailing environmental conditions (Ahsan et al., 2007). Heat shock proteins (HSPs) are typically induced when cells are exposed to high-temperature stress, and are closely related to the acquired thermo-tolerance (Charng et al., 2006). In our study, three forms of HSP70 (spots 3, 4, 5) were identified and significantly up-regulated under high-temperature stress, which is a key part of the high-temperature response (Liao et al., 2014). In addition, two small heat shock proteins (sHSPs, spots 37, 67) were found to be newly induced by high-temperature stress, and were both found to be absent under normal conditions. The sHSPs were further up-regulated by exogenous Spd, suggesting that Spd played a crucial role in maintaining proper folding, facilitating the refolding and preventing the aggregation of the denatured proteins under high-temperature stress (Shi et al., 2013). In this experiment, the stimulation of the heat shock protein with the application of Spd may be relevant to the influence of polyamines on the DNA-binding capacity of heat shock transcriptional factor HSF (Desiderio et al., 1999).

Reactive oxygen species (ROS) metabolism is a universal response to environmental stresses. The stress-induced accumulation of ROS seriously damages the cellular membrane and internal function components, and plants have developed an antioxidant system to regulate the ROS level (Li et al., 2015). In the present study, five proteins were found to have antioxidant-related functions. Among them, Spd increased the abundances of stromal ascorbate peroxidase (APX, spot 25) and dehydroascorbate reductase (DHAR, spot 29) under hightemperature stress (**Figure 6A**). Further analysis revealed that

the activities of APX and DHAR were increased significantly with the application of Spd under high-temperature stress (**Figure 6B**). The enhanced activities could be largely explained by the up-regulated mRNA levels of APX2, APX6, DHAR1, and DHAR2 (**Figure 6C**). Interestingly, the expression of superoxide dismutases [Fe] (Fe SOD, spot 35) in the plastid was not in accordance with the superoxide dismutase [Cu-Zn] (Cu/Zn SOD, spot 36) in the chloroplast. Moreover, the protein expression, activities of enzymes and related mRNA levels also showed different change patterns in response to hightemperature and/or Spd treatment. The variance might be due to the post-transcriptional regulation and post-translational modification of SOD through complex mechanisms, which needs further study. Taken together, the exogenous Spd is involved in antioxidant and detoxification defense mechanisms, mitigating oxidative damage and intensifying the resistance to high-temperature stress (Mostofa et al., 2014; Sang et al., 2016).

#### Protein Synthesis, Folding and Degradation

Generally, abiotic stress causes a transient suppression of de novo protein synthesis (Capriotti et al., 2014). In this study, proteomic analysis identified two spots related to protein synthesis, including elongation factor TuB (spot 17)

and mRNA binding protein precursor (spot 66), which were markedly decreased under the high-temperature stress. However, the expression was enhanced after the application of Spd. According to previous data (Li et al., 2012), it can be hypothesized that stimulating the synthesis of specific proteins by exogenous Spd may play important roles in regulating the proteins synthesis and translational machinery, which are important components of the stress response in plants.

Two proteins (spots 38, 46) that induce proper protein folding and/or prevent the aggregation of stress-damaged proteins were preferentially upregulated under high-temperature stress. In agreement with this observation, the upregulation of peptidylpropyl cis–trans isomerase FKBP 16-3 (spot 38) had been

reported in Arabidopsis and rice in response to high-temperature stress (Palmblad et al., 2008; Gammula et al., 2011). The two proteins showed a decreasing pattern under the stress with Spd, indicating that exogenous Spd might regulate protein folding and assembly, participating in the high-temperature stress tolerance.

Proteolysis is a complex process involving many enzymes and pathways in various cellular compartments. Proteases play a central role in metabolism under abiotic stress as they are involved in protein inactivation and the degradation of damaged proteins (Capriotti et al., 2014). In our study, the cysteine proteinase 3-like (spot 52) was up-regulated under high-temperature stress, in agreement with previous studies (Koizumi et al., 1993; Callls, 1995). Interestingly, the expression of ATP-dependent zinc metalloprotease (spot 7) and proteasome subunit alpha type-2-A-like (spot 33) were decreased under high-temperature stress but increased significantly with the application of exogenous Spd. Stimulating the proteolysis of specific proteins by Spd accelerated the degradation of misfolded/damaged proteins, and made tissues more stable by covalently attaching with proteins (Li et al., 2013). Furthermore, the PAs regulated the protein metabolism and may reprogram the proteome in response to abiotic stress (Yuan et al., 2016), which may also account for the resistance to high-temperature stress of the tomato seedlings.

#### Energy and Metabolism

It is well-known that sufficient ATP is necessary in response to abiotic stress in plants (Hu et al., 2014). Three proteins associated with ATP synthase (spots 8, 9, and 53) were significantly upregulated under the high-temperature stress, suggesting a higher energy demand for the degradation and biosynthesis of proteins (Das et al., 2015). However, the ATP synthase proteins were down-regulated by the exogenous application of Spd under high-temperature stress, stabilizing the process of ATP synthesis, and energy metabolism.

ATP is mainly produced by carbohydrate metabolism, such as glycolysis, the tricarboxylic acid cycle and the pentose phosphate pathway (Hu et al., 2015). The first group included 7 proteins involved in glycolysis pathway. Among them, our results showed that fructose-bisphosphate (FBP) aldolase in the cytoplasm (spot 27), and chloroplastic (spot 42) decreased significantly under high temperature. Moreover, glyceraldehyde-3-phosphate dehydrogenase (GAPDH, spot 22), and 2,3-bisphosphoglycerateindependent phosphoglycerate mutase (spot 19) also decreased under the stress, which would inhibit the glycolysis pathway and glycolysis associated with intermediate metabolism. However, exogenous Spd up-regulated these proteins, allowing more carbohydrates to enter the glycolic pathway and maintain the normal physiological metabolism of the tomato seedlings, thereby supporting the high-temperature resistance (Shan et al., 2016). The second group included malate dehydrogenase (MDH, spots 21, 65), involved in the tricarboxylic acid cycle. In this study, MDH showed different accumulation patterns in response to high temperature, whereas exogenous Spd sprayed on the leaves maintained the MDH expression at a high level. The third group was protein participating in the pentose phosphate pathway. Under high-temperature stress, the abundance of transketolase (TK, spot 10) and ribose-5-phosphate

isomerase (spot 55) decreased. Spd application further improved the abundance of TK, whereas it affected ribose-5-phosphate isomerase unremarkably. Adjusting the EMP-TCA-PPP pathway to produce more energy may be an important mechanism for Spd to alleviate stress induced damage (Li et al., 2015).

#### Signal Transduction

Signal transduction pathways play an important role in abiotic stress at the cellular level, leading to changes in metabolic pathways and cellular processes. After the perception of the stress, a signal would be transferred from the cell surface to the nucleus, and then the responsive proteins would be translated (Guo et al., 2013). Within this functional category, we identified calreticulin (spots 1, 2) and harpin binding protein 1 (spot 56). Calreticulin, a major endoplasmic reticulum Ca <sup>2</sup>+ binding chaperone, is involved in a variety of cellular signaling pathways. Calreticulin also plays a crucial role in regulating Ca2<sup>+</sup> intracellular homeostasis (Nakamura et al., 2001). In our study, calreticulin was significantly up-regulated under the high-temperature stress, but was down-regulated by Spd. These observations suggested that Spd has a relationship with the stressinduced Ca <sup>2</sup><sup>+</sup> signal transduction, probably allowing the release of free Ca <sup>2</sup><sup>+</sup> to relieve stress. Interestingly, harpin binding protein 1 was significantly down-regulated in the tomato leaves under the high-temperature stress, which was concordant with the finding in spring soybean under cold stress (Tian et al., 2015). However, the protein level recovered the controlled level after Spd treatment, and the regulatory mechanism remains unclear.

### CONCLUSION

In conclusion, our results demonstrated that exogenous Spd improving tomato seedlings growth and high temperature

protein abundance marked in red ellipse were integrated. Arrows on the life side of the ellipses indicate changes induced by high temperature stress as compared with the control, and arrows on the right side indicate changes induced by Spd under high temperature stress conditions. Red or green arrows represent up-regulation or down-regulation, respectively, and the black short lines indicate no change. ADZM, ATP-dependent zinc metalloprotease FTSH 2; ANKRD2, ankyrin repeat domain-containing protein 2; APX, stromal ascorbate peroxidase; CA, carbonic anhydrase; CPO1, coproporphyrinogen-III oxidase 1; CYP, cysteine proteinase 3-like; DHAR, dehydroascorbate reductase; EFTuB, elongation factor TuB; FBA, fructose-bisphosphate aldolase; FNR, ferredoxin–NADP reductase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase B; GSAM, glutamate 1-semialdehyde 2,1-aminomutase; HDHDP, haloacid dehalogenase-like hydrolase domain-containing protein At3g48420; HSP70, heat shock 70 kDa protein; mBP, mRNA binding protein precursor; MDH, malate dehydrogenase; OEE, oxygen-evolving enhancer protein; PCI, peptidyl-prolyl cis-trans isomerase FKBP16-3; PGAM, 2,3-bisphosphoglycerate-independent phosphoglycerate mutase; PGK, phosphoglycerate kinase; PSα-2A, proteasome subunit alpha type-2-A-like; RPIA, ribose-5-phosphate isomerase 3; sHSPs, class I small heat shock protein; IMPCH, putative inosine monophosphate cyclohydrolase; SOD, superoxide dismutase; TK, transketolase; TPI, triosephosphate isomerase; Trx, thioredoxin-like protein CDSP32.

tolerance, could be associated with the following processes: (1) stimulating protein related to photosynthesis and energy metabolism, enhancing photosynthetic capacity, providing higher energy for various metabolic processes to cope with high-temperature stress; (2) activation of cell rescue and defense response to alleviate stress induced injuries, activating the antioxidant system; (3) stimulating protein synthesis and degrading misfolded/damaged

proteins induced by high temperature stress. Schematics (**Figure 8**) was formed to illustrate the detailed mechanism to reveal cell metabolism regulated by high temperature and/or Spd. This study provides comprehensive insights through comparative proteomics, and would be able to better enrich our understanding of the mechanism of Spd improves the tolerance of under the high-temperature stress.

#### AUTHOR CONTRIBUTIONS

SG designed the research and proposed the research proceeding. QS wrote the main manuscript text. XS and YA prepared all figures and modified this manuscript until submitted. SS and JS improved the manuscript. All authors reviewed and approved the manuscript.

#### FUNDING

This work was supported by the National Natural Science Foundation of China (No. 31471869, No. 31401919 and No.

#### REFERENCES


31272209), the Jiangsu Province Scientific and Technological Achievements into Special Fund (BA2014147), the China Earmarked Fund for Modern Agro-industry Technology Research System (CARS-25-C-03), and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017. 00120/full#supplementary-material


milky stage of rice grains during high temperature stress. J. Exp. Bot. 65, 655–671. doi: 10.1093/jxb/ert435


species, and thermos tolerance provided by isoprene. Plant Cell Environ. 28, 269–277. doi: 10.1111/j.1365-3040.2005.01324.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 © 2017 Sang, Shan, An, Shu, Sun and Guo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.