# UNDERSTANDING THE MOLECULAR MECHANISMS OF PLANT RESPONSES TO ABIOTIC STRESS

EDITED BY : Sang Yeol Lee, Dae-Jin Yun, Jose M. Pardo, Motoaki Seki, Yan Guo and Abel Rosado PUBLISHED IN : Frontiers in Plant Science

#### Frontiers eBook Copyright Statement

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

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

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

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

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

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

ISSN 1664-8714 ISBN 978-2-88963-491-0 DOI 10.3389/978-2-88963-491-0

#### About Frontiers

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

#### Frontiers Journal Series

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

#### Dedication to Quality

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

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

#### What are Frontiers Research Topics?

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

# UNDERSTANDING THE MOLECULAR MECHANISMS OF PLANT RESPONSES TO ABIOTIC STRESS

Topic Editors:

Sang Yeol Lee, Gyeongsang National University, South Korea Dae-Jin Yun, Konkuk University, South Korea Jose M. Pardo, Institute of Plant Biochemistry and Photosynthesis (IBVF), Spain Motoaki Seki, RIKEN, Japan Yan Guo, China Agricultural University, China Abel Rosado, University of British Columbia, Canada

Plant responses to environmental stress are governed by complex molecular and biochemical signal transduction processes, which act in coordination to determine tolerance or sensitivity at the whole plant level. Upon exposure to abiotic stress, plants express a sophisticated coordinated response to reprogram interconnected defense networks and metabolic pathways, by alterations in the transcription, translation, and post-translational modification of defense-related genes and proteins. Traditionally, physiological and phenotypic responses were the major ones to be collected in plant stress biology. However, modern studies include the identification of key genes that influence stress tolerance and plant growth under the imposing stress and the verification of gene functions using knock out mutants or overexpression lines. In addition, genomics has become a necessary tool for the understanding of plant stress responses at the whole genome levels. The identification of stress-tolerant plant resources and the investigation of the functional role of the genetic variants is also a valuable tool in this research field. Recently, the advent of CRISPR/Cas genome editing technology, enables these variations to be introduced in crops for improved stress tolerance traits.

Through the understanding of the molecular mechanisms involved in plant signaling in response to abiotic stress and crop performance characters under stress conditions, we hope to open new ways for the breeding of superior crops.

Citation: Lee, S. Y., Yun, D.-J., Pardo, J. M., Seki, M., Guo, Y., Rosado, A., eds. (2020). Understanding the Molecular Mechanisms of Plant Responses to Abiotic Stress. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-491-0

# Table of Contents

*06 Pectin Methylesterases: Cell Wall Remodeling Proteins Are Required for Plant Response to Heat Stress*

Hui-Chen Wu, Victor P. Bulgakov and Tsung-Luo Jinn


Li Hua Cui, Hye Jo Min, Mi Young Byun, Hyeong Geun Oh and Woo Taek Kim

*49 The Rice* SPOTTED LEAF4 (SPL4*) Encodes a Plant Spastin That Inhibits ROS Accumulation in Leaf Development and Functions in Leaf Senescence*

Giha Song, Choon-Tak Kwon, Suk-Hwan Kim, Yejin Shim, Chaemyeong Lim, Hee-Jong Koh, Gynheung An, Kiyoon Kang and Nam-Chon Paek

*63 Emerging Roles of LSM Complexes in Posttranscriptional Regulation of Plant Response to Abiotic Stress*

Rafael Catalá, Cristian Carrasco-López, Carlos Perea-Resa, Tamara Hernández-Verdeja and Julio Salinas

*77 Modulation of Ethylene and Ascorbic Acid on Reactive Oxygen Species Scavenging in Plant Salt Response*

Juan Wang and Rongfeng Huang

*83 Salinity and ABA Seed Responses in Pepper: Expression and Interaction of ABA Core Signaling Components*

Alessandra Ruggiero, Simone Landi, Paola Punzo, Marco Possenti, Michael J. Van Oosten, Antonello Costa, Giorgio Morelli, Albino Maggio, Stefania Grillo and Giorgia Batelli


Takeshi Fukao, Blanca Estela Barrera-Figueroa, Piyada Juntawong and Julián Mario Peña-Castro

*142 Chloroplast Redox Regulatory Mechanisms in Plant Adaptation to Light and Darkness*

Francisco Javier Cejudo, Valle Ojeda, Víctor Delgado-Requerey, Maricruz González and Juan Manuel Pérez-Ruiz

*153 Endoplasmic Reticulum Plays a Critical Role in Integrating Signals Generated by Both Biotic and Abiotic Stress in Plants*

Chang-Jin Park and Jeong Mee Park

*161 Early* Brassica *Crops Responses to Salinity Stress: A Comparative Analysis Between Chinese Cabbage, White Cabbage, and Kale*

Iva Pavlović, Selma Mlinarić, Danuše Tarkowská, Jana Oklestkova, Ondřej Novák, Hrvoje Lepeduš, Valerija Vujčić Bok, Sandra Radić Brkanac, Miroslav Strnad and Branka Salopek-Sondi

*177 Epitranscriptomic RNA Methylation in Plant Development and Abiotic Stress Responses*

Jianzhong Hu, Stefano Manduzio and Hunseung Kang

*188 Acetic Acid Treatment Enhances Drought Avoidance in Cassava (*Manihot esculenta Crantz*)*

Yoshinori Utsumi, Chikako Utsumi, Maho Tanaka, Chien Van Ha, Satoshi Takahashi, Akihiro Matsui, Tomoko M. Matsunaga, Sachihiro Matsunaga, Yuri Kanno, Mitsunori Seo, Yoshie Okamoto, Erika Moriya and Motoaki Seki

*200 A Wall-Associated Kinase Gene* CaWAKL20 *From Pepper Negatively Modulates Plant Thermotolerance by Reducing the Expression of ABA-Responsive Genes*

Hu Wang, Huanhuan Niu, Minmin Liang, Yufei Zhai, Wei Huang, Qin Ding, Yu Du and Minghui Lu

*213 The 26S Proteasome is Required for the Maintenance of Root Apical Meristem by Modulating Auxin and Cytokinin Responses Under High-Boron Stress*

Takuya Sakamoto, Naoyuki Sotta, Takamasa Suzuki, Toru Fujiwara and Sachihiro Matsunaga

*226 Genome-Wide Characterization and Expression Analysis of Soybean TGA Transcription Factors Identified a Novel TGA Gene Involved in Drought and Salt Tolerance*

Bo Li, Ying Liu, Xi-Yan Cui, Jin-Dong Fu, Yong-Bin Zhou, Wei-Jun Zheng, Jin-Hao Lan, Long-Guo Jin, Ming Chen, You-Zhi Ma, Zhao-Shi Xu and Dong-Hong Min


Changzheng Song, Yifan Yan, Abel Rosado, Zhenwen Zhang and Simone Diego Castellarin


Meng-Jie Zhao, Li-Juan Yin, Jian Ma, Jia-Cheng Zheng, Yan-Xia Wang, Jin-Hao Lan, Jin-Dong Fu, Ming Chen, Zhao-Shi Xu and You-Zhi Ma

*324 Transcriptome Analysis of the Hierarchical Response of Histone Deacetylase Proteins That Respond in an Antagonistic Manner to Salinity Stress*

Minoru Ueda, Akihiro Matsui, Shunsuke Watanabe, Makoto Kobayashi, Kazuki Saito, Maho Tanaka, Junko Ishida, Miyako Kusano, Mitsunori Seo and Motoaki Seki

# Pectin Methylesterases: Cell Wall Remodeling Proteins Are Required for Plant Response to Heat Stress

Hui-Chen Wu<sup>1</sup> , Victor P. Bulgakov<sup>2</sup> and Tsung-Luo Jinn<sup>3</sup> \*

<sup>1</sup> Department of Biological Sciences and Technology, National University of Tainan, Tainan, Taiwan, <sup>2</sup> Institute of Biology and Soil Science, Far Eastern Branch of the Russian Academy of Sciences, Vladivostok, Russia, <sup>3</sup> Department of Life Science, Institute of Plant Biology, National Taiwan University, Taipei, Taiwan

Heat stress (HS) is expected to be of increasing worldwide concern in the near future, especially with regard to crop yield and quality as a consequence of rising or varying temperatures as a result of global climate change. HS response (HSR) is a highly conserved mechanism among different organisms but shows remarkable complexity and unique features in plants. The transcriptional regulation of HSR is controlled by HS transcription factors (HSFs) which allow the activation of HS-responsive genes, among which HS proteins (HSPs) are best characterized. Cell wall remodeling constitutes an important component of plant responses to HS to maintain overall function and growth; however, little is known about the connection between cell wall remodeling and HSR. Pectin controls cell wall porosity and has been shown to exhibit structural variation during plant growth and in response to HS. Pectin methylesterases (PMEs) are present in multigene families and encode isoforms with different action patterns by removal of methyl esters to influencing the properties of cell wall. We aimed to elucidate how plant cell walls respond to certain environmental cues through cell wall-modifying proteins in connection with modifications in cell wall machinery. An overview of recent findings shed light on PMEs contribute to a change in cell-wall composition/structure. The finescale modulation of apoplastic calcium ions (Ca2+) content could be mediated by PMEs in response to abiotic stress for both the assembly and disassembly of the pectic network. In particular, this modulation is prevalent in guard cell walls for regulating cell wall plasticity as well as stromal aperture size, which comprise critical determinants of plant adaptation to HS. These insights provide a foundation for further research to reveal details of the cell wall machinery and stress-responsive factors to provide targets and strategies to facilitate plant adaptation.

#### Keywords: cell wall remodeling, heat stress response, guard cell wall, pectin, pectin methylesterase

**Abbreviations:** ABA, abscisic acid; AGA, apiogalacturonan; Ca2+, calcium ion; CaM, calmodulin; CBK3, CaM-binding protein kinase 3; CesA, cellulose synthase; CSLD5, cellulose synthase-like protein; CTL1, chitinase-like protein; DM, degree of methylesterification; EXP, expansin; GalA, galacturonic acid; HGA, homogalacturonan; HIT4, heat-intolerant 4; HS, heat stress; HSBP, HSF-binding protein; HSE, heat shock element; HSF, heat shock transcription factor; HSP, heat stress protein; HSR, heat stress response; MeOH, methanol; miRNA, microRNA; MP, movement protein; NF-YC10, nuclear factor Y, subunit C10; NPG1, no pollen germination 1; OG, oligogalacturonide; ORF, open reading frame; PAE, pectin acetylesterase; PG, polygalacturonase; PL, pectin lyase; PLL, pectate lyase-like protein; PME, pectin methylesterase; PME34, pectin methylesterase 34; PMEI, PME inhibitor; QRT1, QUARTET 1; RG-I, rhamnogalacturonan-I; RG-II, rhamnogalacturonan-II; ROS, reactive oxygen species; sHSP, small heat shock protein; siRNA, small interfering RNA; VGD1, VANGUARD 1; WAK, wall-associated kinase; XET/XTH, xyloglucan endotransglycosylase/hydrolase; XGA, xylogalacturonan.

#### Edited by:

Motoaki Seki, RIKEN, Japan

#### Reviewed by:

Niranjan Chakraborty, National Institute of Plant Genome Research (NIPGR), India Jenny C. Mortimer, Lawrence Berkeley National Laboratory (LBNL), United States

> \*Correspondence: Tsung-Luo Jinn jinnt@ntu.edu.tw

#### Specialty section:

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

Received: 03 July 2018 Accepted: 17 October 2018 Published: 06 November 2018

#### Citation:

Wu H-C, Bulgakov VP and Jinn T-L (2018) Pectin Methylesterases: Cell Wall Remodeling Proteins Are Required for Plant Response to Heat Stress. Front. Plant Sci. 9:1612. doi: 10.3389/fpls.2018.01612

#### INTRODUCTION

fpls-09-01612 November 5, 2018 Time: 7:43 # 2

Plants face challenges of extreme environmental conditions, which include various abiotic and biotic stresses, all of which exert adverse effects on plant growth and development. Being sessile organisms, plants cannot move to favorable environments; accordingly, they have developed a remarkable number of strategies to mitigate environmental impacts. Global climate change constitutes one of the most detrimental stresses to plants because it is driving an increase in ambient temperatures, which, according to the Intergovernmental Panel on Climate Change prediction (IPCC, 2012), are expected to be 2–5◦C higher than the current temperatures by the late twenty-first century. Extremely high temperatures can cause devastating damage to crops and reduce crop production; however, plant cells have been shown to have elaborate systems to respond to a variety of challenges, including HSR, which can improve crop yield under climate change conditions.

Acquired thermotolerance in plants consists of translating an initial moderate temperature increment into molecular defenses against subsequent extreme temperatures, such as preventing and repairing damage to heat-labile proteins and membranes (Larkindale and Vierling, 2008). HSR is characterized by the induction of a large set of HSPs, many of which comprise chaperone proteins that assist in protein folding and protect cellular homeostasis against heat and other stress stimuli (Morimoto, 2008). In addition to the induction of HS-responsive genes, the modification of biophysical properties of the cell wall may represent a key component in responding to environmental injuries. For example, HS at 37◦C can generate changes in cell wall polymers in coffee (Coffea arabica) leaves, resulting in ∼50% decrease in pectin and 40% increase in hemicellulose (Lima et al., 2013). Thus, the modulation of plant cell walls, which comprise a dynamic and interconnected network consisting of a heterogeneous matrix with diverse biochemical and mechanical properties, has emerged as an important strategy in plant stress responses.

Pectic polysaccharides are highly heterogeneous polymers involved in the control of cell wall porosity and constitute the major adhesive material between cells (Willats et al., 2001). When the cell is challenged with stress conditions, specific transcriptional responses affect the production of certain cell wall proteins, leading to crucial changes in cell wall architecture (Klis et al., 2006). Pectin modification is catalyzed by a large enzyme family of PMEs that reside in the cell wall and which modulate apoplastic Ca2<sup>+</sup> content in response to stresses for both the assembly and disassembly of the pectic network (Micheli, 2001; Wu and Jinn, 2010; Wu et al., 2010). Thus, the maintenance of cell wall integrity is tightly controlled and strictly coordinated with the stress response in plant cells. Pectins have been identified as key elements in plant responses to either heat or cold temperature stress in various species such as winter oilseed rape (Brassica napus var. oleifera), bromeliad (Nidularium minutum), Arabidopsis (Arabidopsis thaliana), rice (Oryza sativa), soybean (Glycine max), and coffee (Wu and Jinn, 2010; Carvalho et al., 2013; Huang et al., 2017). However, the cell wall factors that contribute to the development of plant thermotolerance remain largely unknown.

A previous review revealed that PME-mediated changes in the cell wall have played a role in various vegetative and reproductive developmental processes in Arabidopsis and other dicotyledons (Wolf et al., 2009). The DM of HGAs can be controlled by PMEs that have the capacity to remove methyl ester groups to contribute to the intercellular adhesion during plant development and stress responses (Wu et al., 2010; Le Gall et al., 2015). Thus, the DM constitutes a key element in the control of the wall stiffness and hydration status of the pectic matrix during abiotic stresses. PMEs exhibit a potential for the development of thermotolerance by maintaining apoplastic Ca2<sup>+</sup> homeostasis (Wu et al., 2010). Heat-activated PME activity is involved in the cell-wall localization of Ca2+, i.e., with the removal of apoplastic Ca2<sup>+</sup> that participates in HS signaling to induce HSP expression and cell-wall remodeling to retain plasma membrane integrity, thus preventing cellular content leakage and conferring thermoprotection (Wu and Jinn, 2010). Furthermore, through the identification of Arabidopsis PME34 mutant plants, it was verified that the thermotolerance impairment of pme34 was independent from the expression of HS-responsive genes; whereas PME34 functions in controlling stomatal movements and in regulating the flexibility of the guard cell wall required for heat response (Huang et al., 2017).

Little is known about the dynamics of the pectin matrix in the regulation of the impact of stress on plants. In previous research, we focused on the role of the PME, which is intrinsically involved in the modification of cell wall components in response to HS; and most recently demonstrated that the dynamic network of cell wall remodeling proteins with enzymatic activity is crucially important for cell wall tolerance to HS (Wu and Jinn, 2010; Wu et al., 2010, 2017; Huang et al., 2017). The present review, therefore, describes the most recent findings regarding cell wall remodeling and HSR as well as specific issues with the characterization of PME. In addition, this review also highlights the diversity of their roles during plant development and in response to diverse abiotic stresses, particularly to HS.

# REGULATION OF THE HEAT STRESS RESPONSE IN PLANTS

Heat stress causes a broad spectrum of cellular damage through the extensive denaturation and aggregation of proteins, and by modifying membrane permeability and fluidity, which subsequently disrupts the balance of metabolic processes. In nature, such HS conditions may be chronic or recurring, or both (Bäurle, 2016); therefore, plants have developed diverse systems to cope with recurring stress. HSR is a highly conserved stress response mechanism that reflects how plants respond and adapt to HS through improved thermotolerance. It defines all high temperature-related defense activities used in the cell to prevent damage and aggregation at the proteome level (Lindquist and Craig, 1988; Vierling, 1991).

The induction of HSPs constitutes one of the bestcharacterized responses in the adaptation to elevated temperature and plays an important role in the acquisition of thermotolerance. Recently, epigenetic mechanisms have been found to play important roles in the regulation of HSR, including DNA methylation (Boyko et al., 2010; Folsom et al., 2014; Lämke and Bäurle, 2017), histone modification (Min et al., 2014), histone variants (Kumar and Wigge, 2010), ATP-dependent chromatin remodeling (Mlynárová et al., 2007), and siRNAs and miRNAs (Ito et al., 2011; Ballén-Taborda et al., 2013). For example, miRNA 156, 160, and 172 modulation of HSP gene induction is required for Arabidopsis thermotolerance (Khraiwesh et al., 2012; Lin et al., 2018). Accumulation of the heat-induced retrotransposon ONSEN, which is recognized by HS transcription factors HsfA1 and HsfA2 through its HSE, is required for the regulation of HS memory (Ito et al., 2011; Cavrak et al., 2014; Ohama et al., 2017). Arabidopsis HIT4 is a chromocenter-localized protein that functions as a regulator of stress-triggered chromatin reorganization that is essential for plant heat tolerance (Wang et al., 2013). Therefore, it appears as if the epigenetic control of heat-responsive gene expression is frequently utilized to prevent heat-related damages (Liu et al., 2015; Lämke and Bäurle, 2017; Ohama et al., 2017). Furthermore, the emerging evidence indicates that cell wall remodeling plays a crucial role in the response to HS through the activation of cell wall-related genes and alteration of cell wall compositions (Wu and Jinn, 2010; Wu et al., 2010, 2017; Huang et al., 2017). Thus, the modification of cell wall structures to enhance their functions to perceive and respond to multiple environmental stresses is crucial for plants by imparting stress endurance. We summarize the current knowledge regarding plant HSR with different aspects to integrate cellular compartments and signaling networks as addressed in **Figure 1**.

#### Plant Heat Stress Transcription Factor-Mediated Heat Stress Transcriptional Network

The inherent ability of plants to tolerate temperatures above those that are optimal for growth is termed basal (or intrinsic) thermotolerance (Larkindale et al., 2005). Plants also have the ability to acquire tolerance to otherwise lethal HS, referred to as acquired thermotolerance (or HS priming). Specifically, mild HS primes a plant to subsequently withstand or to acclimate to high temperatures that would otherwise be lethal to an unadapted plant (Mittler et al., 2012). This priming response of plants can be maintained over several days after mild HS and a return to normal growth conditions; which is referred to as the maintenance of acquired thermotolerance or HS memory (Charng et al., 2006, 2007; Stief et al., 2014). However, the molecular mechanisms involved in plant HS-priming and HSmemory remain largely unknown, especially for HS memory (Bäurle, 2016). In brief, heat stress transcription factors (HSFs) act as central regulators of HS priming by recognizing the conserved HSE in the promoter of the genes encoding HSP (Scharf et al., 1990), which in turn guard the proteome from misfolding and aggregation under heat conditions. The HSBP is a negative regulator of HSR through the interaction with HSF and thus dissociates trimeric HSFs for the attenuation of HSR (Satyal et al., 1998; Fu et al., 2002; Hsu et al., 2010; Rana et al., 2012). The regulation of HSFs in response to HS is illustrated in **Figure 2A**.

Four homologs of HsfA1 (HsfA1a, b, d, e) play roles of master regulators for acquired thermotolerance in Arabidopsis (Liu et al., 2011). HsfA2 has been shown to act as a secondary regulator under the control of HsfA1s to trigger a transcriptional cascade for the induction of early and late HS-responsive genes (Busch et al., 2005; Charng et al., 2007). Arabidopsis HsfB1 can act as a transcriptional repressor during the attenuation of HSR, whereas tomato (Solanum lycopersicum) HsfB1 possesses both coactivator and repressor functions (Bharti et al., 2004; Ikeda et al., 2011). Two major regulators of HSR, HsfA1s and dehydration-responsive element-binding protein 2A, are controlled by other regulatory factors such as NF-YC10, Hsp90, Hsp70, and small HSP (sHSP) (Hahn et al., 2011; Sato et al., 2014), with regard to their activation or inactivation in the early HSR or an unstressed condition. In addition, phytohormone ABA is also involved in HS signaling through the regulation of HsfA6b for plant thermotolerance (Huang et al., 2016). In yeast (Saccharomyces cerevisiae), Hsf1 in collaboration with protein kinase PKC1 regulates heat-induced cell-wall genes, including CWP1, SPI1, HOR7, YGP1, and ZEO1, to mediate maintenance of cell-wall integrity under HS (Imazu and Sakurai, 2005). Thus, yeast Hsf1 plays a role not only in the induction of HSPs expression but also in the induction of a set of cell-wall genes involved in cell-wall formation and remodeling to prevent cell lysis at high temperature. These data raise the question whether plant HSFs function as yeast Hsf1 involved in cell wall remodeling. Further work is required to determine the role of HSFs encoding a variety of other proteins that specifically function in plant cell-wall organization under HS.

# Ca2+/Calmodulin-Mediated Heat Stress Signaling

Despite the ubiquitous nature of the HSR, little is known about how plants sense an elevating temperature to transmit a signal that results in HSP induction and acquired thermotolerance. One candidate molecule that serves as a second messenger during HS signaling is calcium (Ca2+), a ubiquitous signal in eukaryotic cells. Ca2<sup>+</sup> signals were shown to manifest through transient changes of spatio-temporal patterns of free cytosolic Ca2<sup>+</sup> content ([Ca2+]cyt) arising from the flux of Ca2<sup>+</sup> into the cytosol, defining the so-called Ca2<sup>+</sup> signature (Dodd et al., 2010). A stress-induced change in [Ca2+]cyt might constitute one of the primary transduction mechanisms whereby gene expression and biochemical events are altered to adapt to environmental stresses (Monroy and Dhindsa, 1995). The rise time, peak value, and duration of the decay back to resting levels of Ca2<sup>+</sup> transients generated by early events have been implicated in controlling different transduction processes, including changes in gene expression (Dolmetsch et al., 1997; Kim et al., 2009). Depending on the specific activation properties, Ca2<sup>+</sup> channels, Ca2+-ATPases, and Ca2+/H<sup>+</sup> antiporters as modulators of Ca2<sup>+</sup> shape the parameters and spatial characteristics of the Ca2<sup>+</sup> flux,

FIGURE 1 | Integration of cell wall remodeling and the heat response network. Plant perception of heat involves several pathways in different compartments. The cell wall is the first protective barrier in plants that is exposed to heat. Heat stress (HS)-triggered pectin methylesterases (PME) activity, accompanied by Ca2<sup>+</sup> mobilization from apoplastic sources, is involved in cell wall remodeling and is crucial for plant thermotolerance. During recovery time after HS, PME performs linear demethylesterification on highly esterified pectin residues and interacts with Ca2<sup>+</sup> to form a pectate gel lawn, which causes cell-wall stiffening. During non-lethal HS, acidic PME acts randomly on pectin and promotes the action of endo-polygalacturonases (PG) to contribute to cell-wall loosening and the release of Ca2<sup>+</sup> through Ca2+-permeable channels (green oval) in the plasma membrane, thus causing a transient increase in [Ca2+]cyt oscillation. This is followed by induction of a Ca2+/calmodulin (CaM)-dependent pathway to activate the master HS regulator HsfA1s, which directly triggers HS-responsive transcription factors, including HsfA2, HsfA7s, HsfBs, and dehydration-responsive element-binding protein 2A for downstream HsfA3 and HSP gene expression involved in the acquisition of thermotolerance. Histone modification and several epigenetic regulators, including small RNAs and transposons, are involved in the HSR and HS memory. MicroRNA156 targets the SQUAMOSA promoter-binding protein-like gene family, which downregulates HS-inducible genes and therefore maintains the expression of HsfA2 and HSP genes during recovery from HS for long-term adaptation to HS. The retrotransposon ONSEN, as a target of HsfA1s and HsfA2, can be modulated by siRNAs for the regulation of HS memory. Through heat-intolerant 4 (HIT4), HS can relax the silencing of transposons, whereas they can be silenced by deficient in DNA methylation 1 (DDM1) and Morpheus' molecule 1 (MOM1). However, the HS-induced cell wall-related transcript profile needs to be further explored with regard to the maintenance and modification of cell wall integrity.

resulting in distinct a Ca2<sup>+</sup> signature in response to different stress stimuli (Demidchik and Maathuis, 2007) (**Figure 2B**).

The stress-induced intracellular Ca2<sup>+</sup> levels can be transmitted and sensed by a toolkit of Ca2+-binding proteins such as CaMs and their related-proteins, such as CaM-like proteins, calcineurin B-like proteins, and Ca2+-dependent protein kinases, for downstream responses. CaMs are highly conserved, consisting of two globular domains, each with

FIGURE 2 | Basic function of plant HSF and HS-induced [Ca2+]cyt/nuc oscillation interpretation by CaM in response to heat. (A) Under unstressed conditions, Hsp70/HspP90 can directly regulate the function of HSF by blocking its transcriptional activity. Upon HS, non-native proteins induce the conversion of monomeric HSF into an active trimeric form, which is phosphorylated and translocated into the nucleus. HSF trimer, with high-affinity DNA binding capacity to the HSE (50 -nGAAnnTTCnnGAAn-3<sup>0</sup> ) of the HSP gene promoter region, activates HSP gene expression, whereas it is downregulated by the interaction of HSP and HSBP with the HSF trimer to attenuate HSR in plants. HSP production and relocation to the cytoplasm inhibits non-native protein misfolding and aggregation. (B) Cellular Ca2<sup>+</sup> transport is tightly controlled within all membrane-bound organisms during heat stress. An increase in [Ca2+]cyt is manifested by Ca2<sup>+</sup> influx to the cytosol, mediated by Ca2+-permeable ion channels, either from the apoplast across the plasma membrane, or from intracellular stores such as the endoplasmic reticulum or vacuole. In contrast, Ca2+-ATPases and the Ca2+/H<sup>+</sup> antiporter systems are responsible for Ca2<sup>+</sup> extrusion out of the cytosol. HS-elevated Ca2<sup>+</sup> occurs from apoplast entry to the cytosol or nucleus (either diffused from the cytosol or released from nuclear Ca2<sup>+</sup> reservoirs). The CaM responds to the elevation of [Ca2+]cyt signature to modulate the activity of numerous target proteins. (a) and (c) The Ca2+/CaM complex interacts with the HS transcription factors (HSFs) and modulates either HSF DNA-binding or transcriptional activities. (b,d) The Ca2+/CaM complex regulates the activation of HSF by modulating the phosphorylation status. The regulation is achieved by CaM-binding protein kinase (PK) or CaM binding protein phosphatase (PP). (c,d) CaM recognizes a high frequency and magnitude of the cytosolic Ca2<sup>+</sup> signature and is translocated into the nucleus for responding to the nuclear [Ca2+] ([Ca2+]nuc) to bind or regulate the status of HSF phosphorylation in the nucleus. ACAs, autoinhibited Ca2+-ATPases; APC, adenine nucleotide/phosphate carrier; CAXs, Ca2+/H<sup>+</sup> cation antiporters; CNGC, cyclic nucleotide-gated ion channels; ECAs, ER-type calcium ATPases; GLR3.5, glutamate receptor 3.5; HMA1, heavy metal translocating P-type ATPase; LETM1; leucine zipper-EF-hand-containing transmembrane protein 1; MCU, mitochondrial calcium uniporter; TPC1, two-pore voltage-gated channel 1.

two Ca2+-binding EF-hand motifs, and are considered to be multifunctional proteins. These proteins mostly act as general transducers of Ca2+-mediated signal cascades in eukaryotes submitted to various developmental and external stimuli. It was previously suggested that the transduction of environmental signals through CaM gene expression occurs in part by the elevation of [Ca2+]cyt levels (Braam and Davis, 1990). In orchard grass (Dactylis glomerata), DgHsp70, a homolog of cytosolic Hsp70, can bind to Arabidopsis CaM2 in the presence of Ca2+, whereas negative regulation of DgHsp70 decreases the ATPase and foldase activities via Ca2+/CaM binding (Cha et al., 2012). Furthermore, CaM is involved in HSR through the interaction with cytosolic maize (Zea mays) Hsp70 and sorghum (Sorghum bicolor) Hsp90 (Sun et al., 2000; Virdi et al., 2009). Increasing evidence indicates that CaM plays a crucial role in HS responses that lead to an elevation of [Ca2+]cyt signaling in various species (Gong et al., 1998; Liu et al., 2003, 2005; Wu et al., 2012). In wheat, CaM1-2 gene expression increases after HS at 37◦C for 10 min and reaches its peak expression after 20 min HS exposure, as determined by northern analysis (Liu et al., 2003). In moss Physcomitrella patens, a [Ca2+]cyt elevation for 20 min was induced by HS via putative plasma membrane Ca2+-permeable channels (Saidi et al., 2009).

The elevated [Ca2+]cyt and CaM can directly modulate the DNA-binding activity of HSF to HSE, suggesting that they are involved in the expression of HSP genes through the regulation of HSF (Mosser et al., 1990; Li et al., 2004). Arabidopsis signal responsive 1–6 genes (SR1 to SR6), a Ca2+/CaM-binding transcription factor, play roles in transcription activation through specific binding to a "CGCG box" (A/C/G)CGCG(G/T/C) in the promoter of genes that are involved in multiple signal transduction pathways, including HSR in plants (Yang and Poovaiah, 2002). CaM is involved in the modulation of transcription factors either through direct interaction with basic helix-loop-helix domains, or by the control of kinasemediated phosphorylation (Corneliussen et al., 1994; Corcoran and Means, 2001). In transgenic Arabidopsis, reporter GUS gene expression that is directed by the Hsp18.2 promoter was shown to be affected by CaCl<sup>2</sup> and CaM antagonists (Liu et al., 2005). Arabidopsis CBK3, by phosphorylating HsfA1a, enhances the binding activity to HSE, which promotes activation of HSF and HSP gene expression. Protein phosphatases, such as Arabidopsis PP7 are regulated by CaM that is dependent upon Ca2+-CaM binding, with the pp7 mutation resulting in a reduction in acquired thermotolerance (Liu et al., 2007). We identified the rice OsCaM1-1, whose expression resembles the biphasic [Ca2+]cyt signal, and showed that overexpression of OsCaM1-1 induced the expression of Arabidopsis Ca2+/HSrelated CBK3, PP7, HSF, and HSP genes, and enhanced intrinsic thermotolerance in transgenic Arabidopsis (Wu et al., 2012). Thus, OsCaM1-1 interprets the Ca2<sup>+</sup> signal by the cytosolic Ca2<sup>+</sup> concentration and by spatio-temporal Ca2<sup>+</sup> parameters under HS. Furthermore, OsCaM1-1 contains potential miRNA168a and miRNA408 target sites, and both miRNAs harbor HSE, which may regulate transcription of these miRNAs in response to HS (Wu and Jinn, 2012).

Extracellular CaM was found to be involved in the initiation of pollen germination and tube growth by a heterotrimeric G protein in the cellular signaling process in lily (Lilium longiflorum) pollen (Ma et al., 1999). However, the functions of apoplastic CaM are still poorly understood in plant cells. In Cedrus deodara, apoplastic CaM maintained the tip-focused Ca2<sup>+</sup> gradient and modulated the distribution of pectins during pollen tube growth (Wang et al., 2013). Apoplastic CaM contributed to Ca2<sup>+</sup> homeostasis and cell wall remodeling during pollen development. Thus, the interaction between Ca2<sup>+</sup> and apoplastic CaM may play a central role in the maintenance of Ca2<sup>+</sup> gradients for cell-wall modeling. Arabidopsis NPG1 is a pollen-specific CaM-binding protein that interacts with PLLs, suggesting NPG1 may modify the pollen cell-wall through the interaction with PLLs (Shin et al., 2014). In addition, the largest releasable pool of Ca2<sup>+</sup> is localized in the cell wall, reaching approximately 60–75% of the total tissue Ca2<sup>+</sup> content (Demarty et al., 1984). Thus, apoplastic Ca2<sup>+</sup> is essential for the control of cell integrity, cell wall cohesion, and plasma membrane permeability (Hirschi, 2004). It has been suggested that the increased [Ca2+]cyt elevation observed in transformed tobacco (Nicotiana tabacum) seedlings during HS arises from both apoplastic and cytosolic sources (Gong et al., 1998). Potato (Solanum tuberosum) plant growth under HS can persist at specific levels of Ca2<sup>+</sup> in the root, providing insight into the mechanism by which the zone of root Ca2<sup>+</sup> may modulate plant response to HS (Kleinhenz and Palta, 2002). In moss (Physcomitrella patens), a specific Ca2+-permeable channel in the plasma membrane, which regulated heat-inducible Ca2<sup>+</sup> influx, thereby leading to HSR (Saidi et al., 2009). Moreover, the recovery of HS-released Ca2<sup>+</sup> is essential for the acquisition of thermoprotection to mitigate lethal HS injury both in soybean and rice seedlings (Wu and Jinn, 2010; Wu et al., 2010).

Notably, the cleavage of apoplastic Ca2<sup>+</sup> bridges between pectic carboxyl groups that were created by PMEs is considered to play an important role in cell wall remodeling because it retains cell integrity during HS by preventing the plasma membrane from tearing away from the cell wall (Wu and Jinn, 2010; Wu et al., 2010). Thus, acquired thermotolerance is reported to critically depend on a preceding Ca2<sup>+</sup> transient through the plasma membrane so that the HSR is regulated by the transient entry of apoplastic Ca2<sup>+</sup> (Saidi et al., 2009; Wu and Jinn, 2012; Wu et al., 2012). Plant cells can monitor the functional integrity of cell walls, with the maintenance of cell wall integrity being an important process to relieve cellular stresses.

#### CELL WALL REMODELING IN HEAT RESPONSE

#### Plant Cell Wall Basics

The plant cell wall is a sophisticated structure formed by a complex mixture of cell wall polymers, such as polysacchariderich polymers, proteins, and pectin matrix that are assembled into a rigid, flexible, and dynamically organized network (Wu et al., 2017). Plant cell walls are multilayered and consist of three sections, including the middle lamella, primary cell wall,

**11**

and secondary cell wall. The middle lamella is a pectin layer to cement the bond between two adjoining cells. The heterogeneous mixture of wall composition and thickness of the cell wall may deviate absolutely, depending on the environmental conditions. The primary wall surrounds growing cells or cells capable of cell growth; whereas the secondary wall is a highly specialized and thickened structure containing lignin, which undergoes irreversible changes in many fully developed cells. Cellulose is composed of repeating glucose residues connected through β-1,4-D-glucan (β-glucan) bonds that are crossed intricately together to form microfibrils as the scaffold of the cell wall and interconnected by hemicelluloses (xyloglucans and xylans are the most abundant) and galacturonic-acid-rich pectins (**Figure 3A**).

Pectin, a highly structurally complex polysaccharide, constitutes the major component of primary cell walls for both monocots and dicots, and is important for both cellular adhesion and cell wall plasticity (Mohnen, 2008). For example, pectin makes up 35% of the primary cell wall in dicots and non-grass monocots, 2–10% of grass primary walls, and up to 5% of wood tissues (Mohnen, 2008). The middle lamella, a pectinaceous interface, depends on the formation of intermolecular links between pectin molecules and is important for the adhesion of neighboring cells (Jarvis et al., 2003). Pectins also present in the junction zone between cells within secondary walls in the xylem and fiber cells of woody tissue (Mohnen, 2008). Fiber length of angiosperms is determined by intrusive tip growth, which requires dissolution of the middle lamella, wall loosening between adjacent cells to create space for tip growing (Goulao et al., 2011), and therefore, the modification of pectin may be occurring during secondary wall growth of trees. Generally, pectinaceous polysaccharides have been defined into five classes (Ridley et al., 2001; Caffall and Mohnen, 2009; Harholt et al., 2010), including HGA, RG-I and -II (RG-II), XGA, and AGA; presumably, these structural elements are linked covalently to form the pectin complex as shown in **Figure 3B**. It is generally believed that these pectic polysaccharides are covalently linked to, or tightly associated with other types of polysaccharides, since chemical treatments or digestion by pectin-degrading enzymes are required to isolate HGA, RG-I, and RG-II from each other and from cell walls (Nakamura et al., 2002; Coenen et al., 2007). The results support that a model of pectic polymers, HGA, RG-I, and RG-II are linked together during synthesis (Caffall and Mohnen, 2009). For instance, the HGA backbone can be hydrolyzed by PG to produce monomeric, dimeric, or oligomeric fragments; however, HGA, RG-I, and RG-II polysaccharides failed to resolve independently by size exclusion chromatography prior to fragmentation by PG digestion (York et al., 1996). Furthermore, the stretches of α-(1,4)-linked GalA of soybean soluble polysaccharides were found flanked by RG-I fragments, providing evidence that HGA and RG-I are directly connected through backbone residues (Nakamura et al., 2002). Similarly, it has been suggested that HGA is linked to xyloglucan through fragments of XGA that were not readily solubilized from walls unless treated with PG (Talmadge et al., 1973). Therefore, the backbone of HGA is covalently linked to RG-I and RG-II. It is also hypothesized to be crosslinked to xyloglucan or possibly other wall polymers in muro. In particular, HGA is a major component of pectin and has a conformational flexibility that can be influenced by growth, development, and environmental cues (Willats et al., 2001). HGA consists of a linear α-1,4-linked D-GalA homopolymer, which is the most abundant pectin-rich polysaccharide, constituting 65% of the total pectin. A critical feature of HGA that influences its properties is the methylesterification at C6-carboxyl and acetylation at C2 or C3 position by specific HGA-modifying enzymes, which belong to large multigenic families in all sequenced species (Gou et al., 2012; Sénéchal et al., 2014).

Owing to the characteristics of pectic matter which form hydrophilic colloids, it has been stated that the primary cell wall is plastic and soft. This component is crucial for cell growth and expansion, and is thought to contribute to cell wall structural integrity, cell adhesion, and signal transduction (Ochoa-Villarreal et al., 2012). In addition, the depolymerization of cellulose and hemicellulose, along with pectin, is particularly abundant and dynamic during plant development and stress responses in terms of modifying cell-wall polysaccharides. Consequently, enzymatic cleavage of the cross-linking polysaccharides by a set of cell wall-related enzymes including β-glucosidase, XET/XTH, and PME etc., which are believed to play a role in modulating cell wall plasticity, apparently mediate cell-wall integrity during plant development and stress responses (**Figure 3C**). The details are described below.

# Revealing the Mechanism of Cell Wall Integrity Maintenance in Response to Abiotic Stresses

It has been proposed that plants are able to respond to a spectrum of abiotic stress conditions due to modifications in cell-wall composition and structure to perform their respective functions for the maintenance of cell-wall integrity. However, our understanding of the mechanisms of stress-induced changes in wall composition and structure is still limited. Some cell wallrelated genes have been shown to contribute directly to alter cell-wall composition to maintain cell-wall integrity under abiotic stress. Abiotic stress modified cell-wall constituents by CesA enzymes which alter cellulose biosynthesis (Wang et al., 2016), for instance, AtCesA8/IRX1, which encodes a subunit of a CesA complex to constitute part of the cell wall, plays an important role in drought and osmotic stress responses in Arabidopsis (Chen et al., 2005). Arabidopsis SOS6 encodes a CesA-like protein (CSLD5) which has an important role in response to osmotic stress by regulating stress-induced ROS accumulation in plant cell walls (Zhu et al., 2010). In barley (Hordeum vulgare), a mutation in the HvCslF6 gene that causes the loss of (1,3;1,4) β-D-glucan reducing mixed-linkage glucan in primary cell wall yields mutants increasingly susceptible to chilling (Taketa et al., 2012). In leaves of tomato, β-glucosidase, which is responsible for degrading cellulose to free glucose molecules, is involved in the heat-stress response (Edreva et al., 2000). Additionally, β-glucosidase is likely involved in developing drought-tolerant wheat seedlings (cultivar Hong Mang Mai) by differentially changing cell-wall polysaccharides to favor drought tolerance (Konno et al., 2008).

FIGURE 3 | Cell wall composition and enzymatic modification in response to heat. (A) The cell wall is a complex structure that is composed of cellulose and non-cellulosic neutral polysaccharides embedded in a pectin matrix. Pectins are located in the middle lamella and primary and secondary cell wall. Major primary cell wall are constituted of cellulose microfibrils (multiple chains of β-glucose with β-1,4 glycosidic bonds) which are cross-linked to hemicelluloses and to pectin. Xyloglucan is a major hemicellulose molecule that is composed of β-1,4-linked glucose residues with α-1,6-linked xylosyl side chains. In turn, these side chains can be decorated with either galactose, or fucose residues to create a complex pattern of branches. Xylan consists a backbone of β-1,4-linked xylose (Xyl) residues that can be substituted with glucuronic acid and/or arabinose. Additional substitutions such as acetyl and methyl groups can be also presented. And (B) pectins are highly complex class of polysaccharides that comprise galacturonic acid-rich, consisting of five major classes, namely: homogalacturonan (HGA), rhamnogalacturonan I (RG-I), rhamnogalacturonan II (RG-II), xylogalacturonan (XGA), and apiogalacturonan (AGA) form a structurally diverse glue which provides stiffness or flexibility relying on the chemical modification. (C) Based on the action of hydrolysis and substrate specificity, the degradation of cellulose is cleaved by β-glucosidase into two molecules of glucose; for breaking down hemicellulose, xyloglucan endotransglycosylase/hydrolase (XTH), and expansin proteins (not shown) associated with disassembly of cellulose and xyloglucan matrix may play a role in the cell wall remodeling in different aspects of plant development and stress responses. Xylanase is responsible for degrading xylan by cleaving β-1,4 xylose linkages in the backbone. β-xylosidases cleave xylose from the non-reducing end of the xylan chain, and glucuronidases cleave the α-1,2 linked glucuronic acid, and α-arabinosidases cleave the α-1,2 and α-1,3 linked arabinose from the backbone. Pectinolytic enzymes such as PME, PAE, PG, PL, and Arabinanase by hydrolysis of pectic substances, are important for cell wall remodeling. The HGA, a polysaccharide of α-1,4-linked galacturonic acid (GalA) residues, is the predominant form of pectin. A critical feature of HGA that influences its properties is the methyl-esterification and acetylation of specific carbons on GalA during synthesis of the backbone. HGA is de-methylesterified by the activity of PME, which results in random and contiguous patterns of free carboxylic residues. De-methyl-esterification randomly releases protons, which become a target for pectin-degrading enzymes such as PG, which act by hydrolyzing the α-1,4 link between GalA. The contiguous de-methylesterified HGA binds with Ca2<sup>+</sup> to induce gel formation, which can rigidify the cell wall.

In coffee, arabinose and galactose contents increased, whereas mannose, glucose, uronic acid, rhamnose, and fucose contents decreased after HS (Lima et al., 2013). The desiccated plant Myrothamnus flabellifolius had lower amounts of arabinoxylans than those in the hydrated plant, due to the increased association between cell-wall polymers under stress (Moore et al., 2006). Thus, the chemical profile and structural cellwall polymers can be modified under HS. XET/XTH and EXP family members are involved in cell wall loosening and, therefore, in cell expansion for growth and development, as well as in the regulation of the plant responses under abiotic stress (Rose et al., 2002; Cosgrove, 2015). The overexpression of Capsicum annuum XTH3 in tomato showed that increased salt tolerance involved cell-wall flexibility for alleviating stress effects (Choi et al., 2011). In maize, some cell wall-related genes were up-regulated under salinity stress, including ZmXET1, ZmEXPA1, ZmEXPA3, ZmEXPA5, ZmEXPB1, and ZmEXPB2, to hydrolyze and rejoin xyloglucan molecules during cellwall extension (Li et al., 2014). When Arabidopsis plants were exposed to boron toxicity, the expression of genes that encode CesA (CESA1, CESA4, CESA6, and CESA8), and CesAlike CSLB5, EXPs (EXPA5, EXP8, and EXPA14) were reduced, while PMEs (PME2 and PME41) showed a different expression pattern under boron stress and/or 24-epibrassinolide treatment ( ˙I¸skil and Surgun-Acar, 2018). Heat-tolerant, thermal Agrostis scabra, AsEXP1 was strongly induced by exposure to HS, is associated with thermotolerant grass germplasm (Xu et al., 2007). Overexpression of a Kentucky bluegrass (Poa pratensis) PpEXP1 in tobacco exhibited a lesser extent of structural damage to cells resulted in enhanced HS tolerance. Thus, the EXP family may play more extensive and divergent effects on cell-wall integrity during stress responses. On the other hand, Arabidopsis HOT2 encodes a CTL1 that is essential for tolerance to salt stress by preventing Na<sup>+</sup> overaccumulation (Kwon et al., 2006). In Chinese cabbage (Brassica rapa), several genes encoding XTH proteins, β-glucosidase, CesA, EXP, extensin, glycosyl transferase, pectin esterase, and xylosidase, are up-regulated up to two–threefold following non-lethal temperature treatment at 37◦C, which enables plants to survive a subsequent lethal temperature (Yang K. A. et al., 2006). Thus, these results provided evidence that cell wall-related proteins or enzymes are required for the cell-wall modifications involved in thermotolerance acquisition.

Recent studies have described that ROS and peroxidases are key players which initially cross-link phenolic compounds and extensins, causing cell-wall stiffening under drought stress (Tenhaken, 2014). In addition, OH. radicals, which are able to cleave sugar bonds in polysaccharides, cause loosening of the cell wall similar to the action of EXPs or xyloglucan modifying enzymes (Renew et al., 2005). In the review by Houston et al. (2016), a broader consideration was made of multiple cell wall-related genes appearing to respond to a given stimulus, and a defined set of stress-responsive transcription factors involved in transcriptional regulation. However, a specific target for cell-wall modifications due to different stress responses has to be explored in detail, especially in distinct species.

# Enzymatic Modification of Cell Wall Structure and Integrity

It has been reported that HGA-type pectins play crucial roles in mediating the modification of cell wall mechanical properties and controlling turgor-induced plant morphogenesis through the action of pectinolytic enzymes (Levesque-Tremblay et al., 2015; Ali and Traas, 2016). In plants, pectinolytic enzymes or pectinases, which act by hydrolysis of pectic substances through the reactions of depolymerization (hydrolases and lyases) and deesterification (esterases), comprise a heterogeneous group of enzymes, including PMEs, PAEs, PGs, and PLs (**Figure 3C**). The acetyl- and methyl-esterifications of pectins represent the key parameters for the regulation of cell wall mechanical properties. HGA chains can be deacetylated in muro by PAE, with the resulting acetylester change dynamically impacting plant growth and development. It has been demonstrated that the deacetylation of pectin can lower the hydrophobicity of the polysaccharide backbone to increase pectin solubility in water (Rombouts and Thibault, 1986). Thus, PAEs are a crucial structural factor can protect polysaccharides against enzymatic digestion (Liners et al., 1994; Chen and Mort, 1996; Bonnin et al., 2003). Black cottonwood (Populus trichocarpa) that overexpress PtPAE1 exhibit disturbed pollen tube elongation and severe male sterility; however, PtPAE1-mediated deacetylation has been shown to lower the digestibility of pectin (Gou et al., 2012). Following the identification of the Arabidopsis PAE family, it was found that pae8 and pae9 mutants led to ∼20% increase in acetate accumulation in cell walls leading to the reduction in inflorescence growth (de Souza et al., 2014). Arabidopsis acetylation 2 (rwa2) mutation, which displayed a 20% reduction in cell-wall acetylation, was observed to increased resistance to Botrytis cinerea (Manabe et al., 2011). When Medicago truncatula was grown in a CO<sup>2</sup> enriched atmosphere, PAE genes were induced in response to aluminum stress and were associated with aluminum resistance (Chandran et al., 2008). In addition, data retrieved from the eFP Browser<sup>1</sup> showed that Arabidopsis PAE2 and PAE4 were induced in response to osmotic and salt stress (Philippe et al., 2017).

Furthermore, the synthesis of HGA with a high methyl ester at C6 carboxyl residues occurs in the Golgi, which is then further exported into the cell wall in a highly methyl-esterified form of 70 ∼ 80% methylesterification (Willats et al., 2001). The action of PME temporally and spatially regulates the fine control of the DM, i.e., the hydrolysis of the methylester bond at the C-6 position of GalA in HGA, and is potentially involved in the regulation of cell wall architecture and determination of the methylesterification status of pectin. The increase of PME activity and DM are attributed to aluminum resistance in the root transition zone in pea (Pisum sativum) (Li X. et al., 2016). A limited number of investigations on the patterns of PME action in response to abiotic stresses suggest that this area is largely unknown. Thus, in subsequent discussion we focus on the function of PME to alter cell wall properties through the modification of different wall components, which plays

<sup>1</sup>http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi

an important role in the response to adverse environments, especially to heat exposure.

#### Functions of Pectin Methylesterase

Pectin methylesterases (EC 3.1.1.11), which belong to class 8 (CE8) of the carbohydrate esterases (CAZy website<sup>2</sup> ) (Cantarel et al., 2009), and whose activity is regulated by PMEIs, modify the DM of pectins (Pelloux et al., 2007). In the Arabidopsis genome, 66 ORFs have been annotated as putative PME genes that are distinctively expressed (Louvet et al., 2006); furthermore, 89 and 80 PME ORFs correspond to the protein-coding genes in the poplar (Populus spp.) and Asiatic cotton (Gossypium arboreum) database, respectively (Geisler-Lee et al., 2006; Li W. et al., 2016). Conversely, fewer PME genes, as represented by 43 putative ORFs, were found in rice (O. sativa subsp. Japonica cv.; Jeong et al., 2015) compared to those of dicots, which may be related to the differences in the structure of the respective cell wall, such as less methyl esterified HGA in grass species (Vogel, 2008; Burton et al., 2010).

Depending on PME structure, Arabidopsis PMEs are frequently organized with an N-terminal extension of PRE and PRO sequence. PMEs can be classified into types I and II based on their presence or absence of the PRO domain. Type I is characterized by the presence of the N-terminal PRO region, which show homology with PMEI domains, whereas type II is characterized by the absence of the PRO region (**Figure 4A**). The export of PME to the cell wall via the PRE domain, which can be mediated by a signal peptide or a transmembrane domain (TM or signal anchor), is required for protein targeting (Beigi et al., 2015). The PRO-region is required for correct targeting of the cell wall and supports an autoinhibitory activity of enzymes necessary for secretion of the mature PME to the apoplast (Giovane et al., 2004; Bosch et al., 2005). Type-II PME without the PRO-region and with five or six introns, has a similar structure to that of phytopathogenic organisms, such as fungi and bacteria (Pelloux et al., 2007). The localization of tobacco type-I PME, NtPPME1, was shown using a full-length product fused with GFP that is specifically expressed in the cell wall of pollen, whereas NtPPME1 lacking the PRO-region was maintained in the cytoplasm, suggesting that the PRO-region of NtPPME could assist the correct targeting of the mature PME (Bosch et al., 2005). The TM domain of tobacco PME Q9LEBO assists in the transport of PME to the cell surface and its export to the cell wall; however, the PRO-region of Q9LEBO does not affect targeting to the cell wall (Dorokhov et al., 2006).

#### Actions of Pectin Methylesterase

Pectin methylesterases function in de-esterification of the methylated carboxyl group (COOCH3) of pectin to form elastic pectins and accompany MeOH generation during division and maturation of the plant cell (Komarova et al., 2014). Three modes of action of mature PMEs on polysaccharides have been proposed: single-chain, multiple-chain, and multipleattack mechanisms (Aragunde et al., 2018). In the single-chain mechanism, the activity of PME converts all substrate sites on the polymeric chain. In the multiple-chain mechanism, PME catalyzes one reaction and then dissociates from the substrate, whereas PME catalyzes many cycles of reaction before the enzyme-polysaccharide complex dissociates in the multiple-attack mechanism (Beigi et al., 2015; Aragunde et al., 2018). Both single-chain and multiple-attack mechanisms have been proposed in plant and bacterial PMEs as these produce contiguous regions of GalAs (Christensen et al., 1998). Conversely, the random attack of fungal PMEs has been reported to be a multiple-chain mechanism (Duvetter et al., 2006).

During cell wall formation, HGA is de-methylesterified by the activity of PME, which results in contiguous and random patterns of free carboxylic residues. The contiguous demethylesterification of PME (by single-chain or multipleattack mechanism) leads to large amounts of demethylesterified GalA, the negatively charged chains of which can bind to Ca2<sup>+</sup> to promote the formation of "egg box" structures and play a significant role in the structural rigidity of the cell wall. Their enzymatic activity can be modulated by different optimal pH values to further shift the mode of action to random demethylesterification (Hocq et al., 2017). Random demethylesterification (as a multiple-chain mechanism) releases protons that become a target for pectin-degrading enzymes such as PG (EC 3.2.1.15), which act by hydrolyzing the α-1,4 link between GalA. PG acts co-operatively with PME to disassemble the pectin polymer networks and contribute to cell wall weakening (Micheli, 2001). This observation has been confirmed by the combination of PME and PG activity causing an increased opening of stomatal aperture in both maize and Asiatic dayflower (Commelina communis) (Jones et al., 2005). However, incubation of PG alone did not show the effect on stomata opening, indicating that the methylesterified HGA is crucial for guard cell wall movement (Jones et al., 2005). Under salt stress, the increased demethylesterified pectins mediated by PME activity tend to crosslink with the Ca2+, leading to solidification of the cell wall and decreased growth (Uddin et al., 2013). Hence, the degree of pectin methyl-esterification affects Ca2<sup>+</sup> cross-linking and pectate gel formation, which has dramatic consequences on cell wall texture and mechanical properties, thereby regulating cellular growth, cell shape, and defense reactions in plants (Pelloux et al., 2007).

In addition, the activity of PME is closely regulated by its endogenous inhibitor proteins, PMEIs, during plant development and growth (Micheli, 2001; Giovane et al., 2004). The additional PRO domain in type-I PME genes shares similarities with the PMEI domain of PMEI genes (Pelloux et al., 2007). PMEIs belong to plant invertase inhibitorrelated proteins, and as inhibitors, they play an important role in the regulation of metabolic enzymes (Koch, 1996). A transgenic Arabidopsis that constitutively expresses AtPMEI-1 or AtPMEI-2 demonstrates a significant reduction in PME activity and increased levels of pectin methylesterification (Lionetti et al., 2012). Overexpression of a novel AtPMEI has a direct, profound effect on the activity of PME. Furthermore, increased PMEI accumulation significantly improved plant resistance to the fungal pathogens Botrytis cinerea, Bipolaris sorokiniana, and Fusarium graminearum (Lionetti et al., 2007;

<sup>2</sup>http://www.cazy.org

Volpi et al., 2011). The pepper (Capsicum annuum) CaPMEI gene, when overexpressed in Arabidopsis, enhances tolerance to Pseudomonas syringae pv. tomato, mannitol, and methyl viologen (An et al., 2008). In addition, the overexpression of PMEI limits the movement of tobamovirus (tobacco mosaic virus) in tobacco and Arabidopsis, and reduces plant susceptibility to the virus (Lionetti et al., 2014). Arabidopsis PME3 and PMEI7 were shown to have overlapping expression patterns in the etiolated hypocotyls when undergoing HGA methylesterification during plant development (Sénéchal et al., 2015). Overexpression of Arabidopsis PME5 and PMEI3 resulted in softer and harder shoot apical meristem cell walls, respectively (Peaucelle et al., 2011). Thus, the regulation of PMEI genes in the function of PME has a connection with plant development, defense, and stress response including wounding, drought, and oxidative and osmotic stresses (Greiner et al., 1998; An et al., 2008).

# Physiological Roles of Pectin Methylesterase

Pectin methylesterases play an important role in both pectin remodeling and disassembly of the cell wall, and, therefore are involved in many physiological processes, including microsporogenesis, pollen germination, tube growth, pollen separation, seed germination, root development, stem elongation, polarity of leaf growth, and fruit softening during post-harvest fruit ripening (Wen et al., 1999; Pilling et al., 2000; Jiang et al., 2005; Francis et al., 2006; Tian et al., 2006). Moreover, over the past few years, several loss-of-function phenotypes of Arabidopsis PME have been described, as shown in **Figure 4B** and **Table 1**.

QUARTET1 (QRT1) assists in the liberation of pollen grains from tetrads during floral development (Francis et al., 2006). VANGUARD1 (VGD1) and PPME1 (PME9) promote pollen tube



growth (Jiang et al., 2005; Tian et al., 2006). PME-mediated demethylesterification is thought to be required to render HGA susceptible to PG-mediated degradation; for example, PME QRT1 potentially acts in tandem with PG QRT3 to degrade de-methylesterified HGA in pollen mother cell primary walls (Rhee et al., 2003; Francis et al., 2006). AtPME35 is responsible for the demethylesterification of pectins and is involved in regulating the mechanical strength of the supporting tissue in Arabidopsis inflorescence stems (Hongo et al., 2012). AtPME6 is abundant during mucilage secretion, acting on embryo morphology and mucilage extrusion, both of which are involved in embryo development (Levesque-Tremblay et al., 2015). In addition, AtPME58 is specifically expressed in mucilage secretory cells and plays a role in mucilage structure and organization (Turbant et al., 2016). PMEs also act as positive regulators in the control of cell elongation in dark-growth Arabidopsis hypocotyls (Pelletier et al., 2010). AtPME17 was highly coexpressed with and processed by a subtilisin-type serine protease AtSBT3.5 to release a mature apoplastic PME isoform that was involved in root development and in response to various stresses (Sénéchal et al., 2014). Pectin content, PME activity, and pectin demethylesterification are also involved in H2O2-induced cell expansion and in determining the root diameter of rice root tips (Xiong et al., 2015).

Additionally, the DM of HGA settled by PME constitutes an important decisive factor of the biological activity of OG-related signaling and the formation of MeOH, leading to the elicitation of plant defense responses (Osorio et al., 2008). The higher degree of pectin methylesterification is less susceptible to hydrolysis by fungal endo-PG, and, therefore, highly methylesterified pectin can trigger plant resistance to pathogenic fungi (Lionetti et al., 2012). Several studies have reported that PME interaction with a virus-encoded MP is required for tobamovirus, turnip vein clearing virus, and cauliflower mosaic virus infection, mediating cell-to-cell movement of the virus through the plasmodesmata (Chen et al., 2000). AtPME3 interacts with the cellulose binding protein of the cyst nematode Heterodera schachtii and enhances the susceptibility of the plant to nematodes (Hewezi et al., 2008). Furthermore, AtPME3 acts as a susceptibility factor and is necessary for the initial colonization by necrotrophic pathogens B. cinerea and Pectobacterium carotovorum (Raiola et al., 2011). Moreover, PME-mediated pectin methyl de-esterification may

influence the mediated release of pectin-derived compounds, which in turn elicits a defense response. Thus, the specific effect of PME in the pattern of pectin methylesterification plays a determinant role in plant immunity (Bethke et al., 2014). Overall, the study of PME genes revealed a considerable compatibility and differential control of regulatory pathways in plants.

In addition, some studies have described for the importance of pectin in secondary cell wall formation and modification. Pectinassociated β-1,4-galactans are detected in the secondary walls of tension and compression wood (Mellerowicz and Gorshkova, 2012). The occurrence of the pectin RG-II in the most primitive extant vascular plant groups (e.g., Pteridophytes, Lycophytes, and Bryophytes), is correlated with the upright growth of developed lignified secondary walls in vascular plants (Matsunaga et al., 2004). Additional evidence provided a clearer link between pectin modification and secondary wall formation. The expression of PMEs are involved in the expanding wood cells of poplar (Siedlecka et al., 2008), and in the stem, phloem, and xylem of Eucalyptus globulus (Goulao et al., 2011). Arabidopsis mutant lacking PME35 has been shown reduced the mechanical integrity in their stem interfascicular fibers (Hongo et al., 2012). Hence, pectin plays a role in the early stages of secondary wall deposition and has a fundamental role in secondary wall structure and function (Xiao and Anderson, 2013). However, the ability of cells to adapt to environmental changes through the regulation of PME-mediated modification in secondary cell wall for wall integrity maintenance remains a major challenge.

# Pectin Methyl Esterase Activity in Heat Responses

To date, numerous studies have revealed that PME participates in the regulation of plant development by affecting the mechanical properties of the plant cell walls; however, little is known regarding the role of PME in abiotic stresses. The effects of temperature stress on the cell wall may be revealed at various levels such as cell wall architecture and composition. It has been shown that pectin contents are related to temperature-dependent modifications, and that the DM of pectins is also involved in temperature responses (Solecka et al., 2008; Wu et al., 2010; Lima et al., 2013; Bilska-Kos et al., 2017; Huang et al., 2017).

Available data support the idea that cell wall-modifying enzymes are involved in temperature stress responses. For example, in winter oil-seed rape, the cold temperaturedependent pectin modification through the regulation of pectin methylesterification degree causes a retardation in leaf expansion that is correlated with the development of cold acclimation and fungus resistance (Solecka et al., 2008). In the leaves of chillingsensitive CM109 maize (Z. mays spp. indentata, dent), low temperatures of ∼14◦C/12◦C (day/night) result in a reduction of pectin contents and PME activity, especially after prolonged treatment for 28 h and 7 days (Bilska-Kos et al., 2017). High temperatures of 35–65◦C cause an activation of PME activity and the formation of MeOH in the intact tissue of green bean and tomato (Anthon and Barrett, 2006). In winter oilseed rape, HS-induces a nearly 10-fold reduction in PME35 (EV193389) gene expression (Yu et al., 2014). In tomato pollen, HsfA2 is an important coactivator of HsfA1a during HSR; in addition, in developing anthers of A2AS transgenic plants with suppressed HsfA2 level, approximately 25% of the genes have function codes assigned for cell wall-modifying enzymes (including several PME, PAE, and PL) under non-stress conditions. It has been suggested that cell wall-related genes might be directly regulated by HsfA2 (Fragkostefanakis et al., 2016). Thus, cell wall-related genes might be regulated by HS-associated gene expression in HSR. The demethylesterification rate of PME activity was increased substantially with increasing temperature, although the mechanism for temperature activation is less understood.

#### Pectin Methylesterase Effects on Cellular Calcium Levels

Polysaccharides and pectin present as a Ca2+-pectate gel are embedded in the primary-cell-wall matrix, providing an enormous Ca2<sup>+</sup> reservoir. Pectin contains largely demethylesterified HGA sequences cross-linked through Ca2<sup>+</sup> bridges to form egg-box structures, which are responsible for maintaining the integrity of the pectic network (Jarvis et al., 2003). The distribution of Ca2<sup>+</sup> at the cell wall is mainly the result of a plethora of binding sites for Ca2<sup>+</sup> in the cell wall, as well as the carefully regulated transport of Ca2<sup>+</sup> into the cytoplasm (Han et al., 2003). Elevated temperature may cause a loss of cell membrane integrity, which allows Ca2<sup>+</sup> leakage out from the cells into the cell wall to activate PME activity (Anthon and Barrett, 2006). It is possible that, at elevated temperatures, some changes may occur in the PME enzyme that converts it to a different or more active form or that its activity may be increased by the presence of Ca2<sup>+</sup> and other cations. In previous studies, we verified that fine-tuning of an apoplastic Ca2<sup>+</sup> mechanism was associated with PME activity on the pectin methylesteri?cation status by immunolocalization analyses of Ca2+-demethylated HGA during HSR and EGTA chelator treatment (Wu et al., 2010). The removal of apoplastic Ca2<sup>+</sup> might participate in HS signaling to induce HS protein expression and cell-wall remodeling to retain plasma membrane integrity, prevent leakage of cellular content and confer thermoprotection (Wu and Jinn, 2010). The blossom-end rot (BER) is a Ca2+-related physiological disorder that occurs in tomato fruit. It has been shown that a reduced level of PME expression and activity directly determine a correlation with changes in cellular Ca2<sup>+</sup> partitioning and distribution in fruits, leading to fruit susceptibility to BER development (de Freitas et al., 2012). The effect of PME expression and activity on the amount of esterified pectins and Ca2<sup>+</sup> bound to the cell wall is an important factor for plant development and stress responses. Thus, the tight control of the DM of pectin and the formation of Ca2<sup>+</sup> cross-linkage appears to play a major role in plant growth and act as a regulator in response to heat.

The action of PME and the level of Ca2<sup>+</sup> availability within the apoplasm has a direct impact on cell wall strength and expansion (Conn et al., 2011). Because the Ca2<sup>+</sup> binding to uronic acids is easy to exchange for H<sup>+</sup> (Sentenac and Grignon, 1981), this reaction may be involved in the acid-induced extension of the cell wall. Therefore, the carboxyl groups of pectin likely interact

with the charged H<sup>+</sup> atom that functions to acidify and loosen the cell wall to reduce injury. The cell corners, which contribute to cell adhesion via Ca2<sup>+</sup> cross-linking, bear greater tension and support the conductivity of mechanical stresses throughout the plant tissue (Ryden et al., 2003). Cleavage of the Ca2<sup>+</sup> bridges between pectic carboxyl groups in the cell wall is important for cell-wall remodeling during stresses. This suggests that the cell wall regulates the level of Ca2<sup>+</sup> concentration to make the cell more "relaxed," thereby increasing the capability to avoid the plasma membrane from detaching from the cell wall. The extra Ca2<sup>+</sup> is mobilized into the cytoplasm through Ca2<sup>+</sup> channels that were opened by depolarization. The extracellular influx of Ca2<sup>+</sup> is governed by changes in the ion binding properties within the cell wall rather than movements across the plasma membrane (Holdaway-Clarke et al., 1997). Moreover, pectin gel strength increases with increasing Ca2<sup>+</sup> concentration but decreases with increased temperature and acidity (Lootens et al., 2003). Thus, the cell wall needs to eliminate Ca2<sup>+</sup> and maintain low-level apoplastic Ca2<sup>+</sup> during HS, resulting in increasing Ca2<sup>+</sup> levels in the cytoplasm for regulating intracellular levels in response to HS (Wu et al., 2010).

Because MeOH is a product of PME action, it might serve as a volatile signal in the protection of photosynthetic machinery from photo-inhibition; stimulating the growth of C3 plants and the signaling of plant-herbivore interactions for plant defense mechanisms (Nonomura and Benson, 1992; Frenkel et al., 1998; Von Dahl et al., 2006). Furthermore, MeOH activates various patterns of gene expression that are involved in detoxification and signaling pathways, including the induction of HSP genes (Downie et al., 2004). The OGs, as pectin fragments related to PME activities that act as elicitors to stimulate the production of ROS, plasma membrane depolarization, and increased inositol triphosphate and [Ca2+]cyt, have been widely reported in plants (Moscatiello et al., 2006). It has been shown that the extracellular domain of WAK1, which functions as a potential sensor of cell wall signaling by directly binding to the Ca2<sup>+</sup> crosslinking pectin-derived OGs, is involved in cell growth, cell expansion, and disease resistance (Wagner and Kohorn, 2001; Decreux and Messiaen, 2005; Kohorn et al., 2006; Li et al., 2009). The heat-activated PME participates in pectin remodeling, which in turn keeps cells from separating and maintains plasmamembrane integrity, prevents cellular leakage, and coordinates with HS signaling to confer thermoprotection (Wu and Jinn, 2010). Together, these findings suggest that homeostasis of the apoplastic [Ca2+] through the regulation of PME activity during HSR might have a pronounced effect on the development of heat tolerance by preventing cellular leakage through Ca2+-pectate remodeling in the cell wall.

#### Guard Cell Wall Remodeling in Heat Responses

Guard cells comprise a highly developed system that is used to determine and characterize the mechanism of the early signal transduction pathway in plants. In particular, they are involved in gas exchange between the interior of the plant and the external environment through the regulation of successive openings and closures of the stomatal pore. Guard cells perceive a multitude of endogenous and environmental stimuli including hormonal stimuli, light, humidity, CO<sup>2</sup> concentration, drought, and temperature to trigger cellular responses resulting in stomatal opening or closure (Kim et al., 2010; Wu et al., 2017). High temperature increases the risk of heat damage and water shortage to plants. In response to elevated temperatures, transpiration occurs through the opening of stomatal apertures to facilitate cooling of the leaf surface through water evaporation (**Figure 5A**). In contrast, drought can cause stomatal closure and reduce transpiration rates; therefore, stomatal control is considered to be a short-term dynamic adaptation to avoid the reduction in leaf water potential (Osakabe et al., 2014).

Immunolocalization analyses of Arabidopsis leaf sections indicate highly methylesterified and Ca2<sup>+</sup> cross-linked deesterified HGA in mesophyll cells, whereas unesterified HGA constitutes the predominant form of pectin in guard-cell walls, leading the stomatal closure response (Amsbury et al., 2016). Arabidopsis PME6 and polygalacturonase involved in expansion

FIGURE 5 | Arabidopsis PME34 regulates the stomatal aperture under heat stress. (A) The inner wall of a guard cell is thicker and more elastic than the outer cell wall to facilitate the opening of the stomatal pore. The elastic property of the guard cell wall acts reversibly during stomatal opening and closing owing to differential thickening and the orientation of cellulose microfibrils (expressed in threads). The openings and closures of the stomata pore are strictly regulated by the integration of environmental stimuli and endogenous hormonal signals. (B) Comparison of elevated temperature stimulated-stomatal opening in wild-type (WT; Col) and pme34 mutant plants. Leaves (21-day-old) of WT and pme34 plants were treated with 37◦C-mild and 44◦C-lethal heat stress (LHS), respectively, as indicated. The pictogram shows the HS regime and the schematic diagram in the lower panel indicates the response of the stomatal aperture. Under normal condition, pme34 plants had a larger stomata aperture compared with that of WT plants. Under mild-HS treatment, the stomatal apertures in WT plants increased for aiding the dissipation of heat, whereas those of pme34 plants did not. Following the further 44◦C LHS at recovery time (RT), stomatal apertures of pme34 were opened wider than those of WT plants, indicating greater water loss than that in WT plants.

3 (PGX3) in the guard cells play an important role in response to stomatal opening/closure control (Amsbury et al., 2016; Rui et al., 2017). In a previous study, heat-exposed rice plants exhibited lower stomatal conductance until harvest, which can affect carbon balance, grain-filling processes, and yield production (Yang and Heilman, 1991). In C. communis, it has been shown that a 40◦C HS for 5 min in roots could lead to a significant decrease in stomatal conductance, indicating that the communication between root and shoot is mediated by longdistance signaling (Yang S. et al., 2006). In Mimosa pudica, heat stimulation triggers rapid hydro-passive stomatal opening and subsequent stomatal closure that is concomitant with a loss of net CO<sup>2</sup> uptake (Kaiser and Grams, 2006).

The highly specialized walls of guard cells enable them to undergo large and reversible deformation during the constriction of stomata (Wu et al., 2017). Therefore, it is possible that cell wall modification factors are involved in controlling stomata apertures. Recently, we found that Arabidopsis PME34 deficiency causes lower transpiration rates owing to an abnormal stomatal opening, leading to higher leaf temperatures and enhanced sensitivity to heat (Huang et al., 2017). The type-I PME gene PME34, which encodes a plasma membrane-localized and a cell wall deposited protein, functions during guard cell wall modification in response to heat. PME34 mutants have been shown to be hypersensitive to heat but independent of HSFmediated HSP gene transcriptional activation. The PME34 transcript was induced by ABA and highly expressed in guard cells, indicating that PME34 is associated with ABA-dependent stomatal movement in response to heat (Huang et al., 2017). High PME activity coincided with an increase in PG activity in pme34 plants, degrading pectin more easily, and further influencing the ability of guard cell walls to be modified in response to heat. This may support the idea proposed by Wu and Jinn (2010), who suggested that different PME isoforms exhibit distinct action patterns and pectic substrate specificity in response to HS. As highly methylesterified pectins are less susceptible to the action of PG, HS might render the cell wall to be more acidic so that it could stimulate the random demethylesterification activity of PME and promote the action of PG on pectin cleavage, further influencing the structural characteristics of guard cell walls for stomatal movement. Notably, the absence of PME34 activity in guard cells may be complemented by other PMEs and an integration with PG action to bring about the wide opening of stomata pores (**Figure 5B**).

Thus, PME34 may have a role in crosslinking with pectic polymers in the cell wall to regulate the flexibility of guard cell walls (Huang et al., 2017; Wu et al., 2017). Although PME functions to remove the methylester group from HGA to prevent stomatal opening, loss of PME34 resulted in wider stomata under lethal heat treatment. This is consistent with the observation that during drought stress, pme6 mutants have a significantly cooler leaf temperature than the wild-type plants, as well as a more restricted response to ABA (Amsbury et al., 2016). The pme34 mutant displayed a defect in the control of stomatal movement with a concomitant increase in leaf temperature. It also showed a higher transpiration rate through the more widely open stomata, which was probably due to the altered pectin methylesterification status of the guard cell wall properties. Thus, PME34 functions in controlling stomatal movements and in regulating the flexibility of the guard cell wall, which is required for the heat response. The impact of loss of PME34 on stomatal aperture may be due to Ca2<sup>+</sup> signaling or oligosaccharides released during cell-wall modification, or both, which requires further investigation (Wu et al., 2017).

#### CONCLUSION AND PROSPECTIVE

Although fine-tuning of the methylesterification of pectin through the regulation of PME activity during plant growth is relatively well understood, very little is known about stressinduced alterations of cell-wall polymers with respect to PME activity. Analysis of the patterns of pectin methylesterification in pme mutants is important to distinguish the distinct roles of individual PME genes. The evidence from genetic and transgenic plants indicated that the modification of cell wall remodeling has a pronounced effect on stress tolerance. The adjustment of the cell wall through the activity of PME under abiotic stresses is a critical determinant of plant adaptation. The change in cell wall metabolism and cell wall-modifying enzyme activity in controlling cell wall plasticity is an important physiological mechanism of plants in response to heat. The stress effect on the architecture of cell wall remodeling by PME activity may depend on the plant species, genotype, and growth stage, and also rely on the intensity and timing of the stress. In addition, the specificity of PMEI toward different PME isoforms can directly modulate the endogenous PME activity during plant development and various stress responses. In particular, complex interaction between PMEs and their inhibitors appears to be involved in a complex metabolic network and the regulation of gene expression pathways during plant growth and development as well as in stress adaptation. The additional complexity of the interaction of PME with other cellwall proteins to render a load-bearing, yet extensible primary cell wall during stress, remains an elusive issue. Much remains to be elucidated as to how the cell wall senses and transduces the signals leading to stress-induced transcriptional machinery changes and the underlying cell-wall polysaccharide deposition and modification. The role of cell wall-related genes, such as WAKs, which directly bind pectin polymers and partially depend upon the DM of pectin, and polysaccharides, has been explored during various stages of plant development (Kohorn et al., 2009; Tucker et al., 2018). The qualitative and quantitative assessment of cell wall composition at the single cell level is also required (Tucker et al., 2018). In particular, we need to elucidate single-cell responses to certain environmental changes. For instance, in the root cells of Arabidopsis, transcriptional changes were found to be directly related to alternations of cell-wall composition (Somssich et al., 2016), indicating that transcript abundance is followed by associated cell-wall modifying enzymes and proteins. Further, it is required to establish a direct connection between pectin

modification and secondary wall formation by identifying and determining the function of pectin-related genes. Consequently, PME-mediated deesterification could be a crucial mechanism for contributing the secondary wall growth of wood development. Likewise, the transcriptional regulation of pectin-modifying genes might be an important aspect of secondary cell wall formation attributed to both abiotic and microbial challenges. These insights provide a foundation for further research such as transcriptomics studies that may reveal details of the cell wall machinery and stress-responsive transcription factors to provide targets and strategies to facilitate plant adaptation to HS.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

H-CW and T-LJ conceived and wrote the manuscript. VB contributed to the final version of the manuscript.

# FUNDING

This work was supported by the National Taiwan University (Grant Nos. 101R892003–105R892003 and 106R891506) and by the Ministry of Science and Technology, Taiwan (Grant Nos. 105- 2311-B-002-033-MY3 and 107-2923-B-002-003-MY3) to T-LJ.


stresses without phenotypic defects in tomato plants (Solanum lycopersicum cv, Dotaerang). Plant Cell Rep. 30, 867–877.





stress through SPL transcription factors. Plant Cell 26, 1792–1807. doi: 10.1105/ tpc.114.123851


pectin methylesterase activity is required for thermotolerance in soybean seedlings. J. Exp. Bot. 61, 2843–2852. doi: 10.1093/jxb/erq121


temperature in Chinese cabbage. Plant Sci. 171, 175–182. doi: 10.1016/j.plantsci. 2006.03.013


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

# Overexpression of the Wheat (Triticum aestivum L.) TaPEPKR2 Gene Enhances Heat and Dehydration Tolerance in Both Wheat and Arabidopsis

Xinshan Zang, Xiaoli Geng, Kexiang He, Fei Wang, Xuejun Tian, Mingming Xin, Yingyin Yao, Zhaorong Hu, Zhongfu Ni, Qixin Sun and Huiru Peng\*

State Key Laboratory for Agrobiotechnology, Key Laboratory of Crop Heterosis and Utilization (MOE), Beijing Key Laboratory of Crop Genetic Improvement, China Agricultural University, Beijing, China

#### Edited by:

Dae-Jin Yun, Konkuk University, South Korea

#### Reviewed by:

Guangxiao Yang, Huazhong University of Science and Technology, China Klára Kosová, Crop Research Institute (CRI), Czechia

> \*Correspondence: Huiru Peng penghuiru@cau.edu.cn; penghuiru1452@163.com

#### Specialty section:

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

Received: 21 May 2018 Accepted: 02 November 2018 Published: 23 November 2018

#### Citation:

Zang X, Geng X, He K, Wang F, Tian X, Xin M, Yao Y, Hu Z, Ni Z, Sun Q and Peng H (2018) Overexpression of the Wheat (Triticum aestivum L.) TaPEPKR2 Gene Enhances Heat and Dehydration Tolerance in Both Wheat and Arabidopsis. Front. Plant Sci. 9:1710. doi: 10.3389/fpls.2018.01710 Wheat (Triticum aestivum L.) yield and quality are adversely affected by heat, drought, or the combination of these two stresses in many regions of the world. A phosphoenolpyruvate carboxylase kinase-related kinase gene, TaPEPKR2, was identified from our previous heat stress-responsive transcriptome analysis of heat susceptible and tolerant wheat cultivars. Based on the wheat cultivar Chinese Spring genome sequence, TaPEPKR2 was mapped to chromosome 5B. Expression analysis revealed that TaPEPKR2 was induced by heat and polyethylene glycol treatment. To analyze the function of TaPEPKR2 in wheat, we transformed it into the wheat cultivar Liaochun10, and observed that the transgenic lines exhibited enhanced heat and dehydration stress tolerance. To examine whether TaPEPKR2 exhibits the same function in dicotyledonous plants, we transformed it into Arabidopsis, and found that its overexpression functionally enhanced tolerance to heat and dehydration stresses. Our results imply that TaPEPKR2 plays an important role in both heat and dehydration stress tolerance, and could be utilized as a candidate gene in transgenic breeding.

Keywords: TaPEPKR2, heat stress, dehydration stress, PEP carboxylase kinase-related kinase, wheat, Arabidopsis

#### INTRODUCTION

Heat and drought stress and their combination during the growing season are major environmental factors affecting the production and quality of wheat worldwide. A 5.5% decrease in global wheat yields was caused by heat stress between 1980–2008 (Lobell et al., 2011), while drought stress caused considerable yield loss and high economic costs in more than 50% of wheat cultivation areas (Ashraf, 2010). Therefore, research into the genetic mechanism of heat and drought stress tolerance is getting increasingly important.

Plants have developed a range of response mechanisms to adjust to abiotic stress, especially molecular responses to maintain normal activities (Shinozaki and Dennis, 2003; Ashraf, 2010; Mittler et al., 2012; Qu et al., 2013). Genes responding to abiotic stress are essential for enhancing abiotic stress tolerance, and an understanding of these is crucial to developing abiotic stress-tolerant crops.

**27**

Protein kinases regulate key aspects of cellular function, including responses to external signals. In Arabidopsis, around 4% of predicted genes encode typical protein kinases (Hrabak et al., 2003). Phosphoenolpyruvate carboxylase kinase (PPCK)-related kinases (PEPKRs) are unique to plants and belong to the CDPK-SnRK superfamily (Hrabak et al., 2003). PPCKs are calcium-independent protein kinases that function in crassulacean acid metabolism and C4 plants (Vidal and Chollet, 1997; Nimmo et al., 2001; Agetsuma et al., 2004). Predicted PEPKR proteins in Arabidopsis contain both N- and C-terminal extensions with no similarity to the non-catalytic domains of other kinases in this superfamily (Hrabak et al., 2003). Thus far, their functions remain unknown.

In this study, we isolated the coding and promoter region of TaPEPKR2, and analyzed the expression pattern of TaPEPKR2 under heat and dehydration stress conditions. Finally, a transgenic approach was used to investigate the function of TaPEPKR2 in both wheat and Arabidopsis under heat and dehydration stress conditions.

# MATERIALS AND METHODS

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

The common wheat variety "TAM107," which has a heat-tolerant phenotype and was released by Texas A&M University in 1984, was used for gene cloning and expression analysis. Growth conditions and stress treatments of wheat were as previously described (Zang et al., 2017a,b). Briefly, the sprouted seeds were grown on moistened filter paper at 22◦C/18◦C (day/night), 12 h/12 h (light/dark), and 60% humidity in a growth chamber. For high-temperature treatments, seedlings were transferred to another growth chamber maintained at 40◦C. For dehydration stress treatments, water was replaced by PEG-6000 (20%). Leaves were collected from the seedlings at different time points, frozen immediately in liquid nitrogen and stored at −80◦C for RNA isolation.

Common wheat cultivar Liaochun10 (LC10) and Arabidopsis thaliana ecotype Col-0 was used for genetic transformation.

#### Cloning and Sequence Analysis

Total RNA was extracted from 7-day-old seedlings using TRIzol reagent (Invitrogen), and purified RNA was treated with DNase I. Subsequently, 2 µg of total RNA was reverse transcribed by M-MLV reverse transcriptase according to the instruction (Promega). Based on the probe sequence (Ta.10701.1.S1\_at), a pair of gene-specific primers TaPEPKR2-L/R was used to amplify TaPEPKR2. Primer sequences are listed in **Supplementary Table S1** (1, 2).

Database searches of nucleotide and deduced amino acid sequences of the TaPEPKR2 homologs were analyzed by NCBI/GenBank/Blast. Sequence alignment and similarity comparisons were performed by DNAMAN software. The functional domains of TaPEPKR2 were identified by SMART programs<sup>1</sup> .

# Expression Pattern Analysis of TaPEPKR2 in Wheat

Quantitative real-time PCR (RT-qPCR) was performed to determine the relative expression level of TaPEPKR2 with specific primers. The 2−11<sup>C</sup> <sup>T</sup> method (Livak and Schmittgen, 2001) was used to quantify the relative expression of TaPEPKR2, and the wheat β-actin gene was used as a reference. Each experiment was repeated three times independently. **Supplementary Table S1** (3, 4, 7, 8) lists the RT-qPCR primers.

#### Transgenic Constructs of TaPEPKR2 and Genetic Transformation in Wheat

The coding DNA sequence (CDS) of TaPEPKR2 (**Supplementary File S1**) driven by the maize ubiquitin promoter was inserted into the binary vector pBract806. The resulting expression constructs were utilized for genetic transformation. LC10 immature embryos were utilized for wheat transformation via the particle bombardment method. The presence of a TaPEPKR2 transgene was verified by PCR using primers listed in **Supplementary Table S1** (5, 6). In total, five transgenic events (T1–T5) were produced and three lines (T3–T5) with higher expression levels were selected for further analysis.

# Thermotolerance Assay in Wheat

The thermotolerance assay in wheat was performed as previously described (Zang et al., 2017a). Seeds of LC10 and TaPEPKR2 transgenic lines were grown in pots containing potting soil under the above-mentioned conditions. 5-day-old seedlings were transferred to a growth chamber at 45◦C for 18 h, typically beginning at 09:00 h, and were then shifted to 22◦C for recovery. The phenotypes were photographed 5 days after the treatment.

#### Generation of TaPEPKR2 Transgenic Arabidopsis Plants

The CDS of TaPEPKR2 was amplified and cloned into the binary vector pB2GW7 using the gateway method. Agrobacterium tumefaciens strain GV3101 containing this binary construct was used to transform Arabidopsis plants by the floral dip method. Transformants were selected on 1/2 MS medium containing Basta (125 µL/L), followed by PCR amplification of positive clones. In total, seven transgenic lines (L1–L7) were produced and three lines (L1, L3, and L5) were selected for further analysis.

# Thermotolerance Assay in Arabidopsis

The thermotolerance assay in Arabidopsis was performed as previously described (Zang et al., 2017a,b). Surface-sterilized seeds of WT and TaPEPKR2 transgenic lines were sown on MS solid medium. The plated 7-day-old seedlings were exposed

<sup>1</sup>http://smart.embl-heidelberg.de/

to 45◦C for 120 min in an illuminated growth chamber, then shifted to 22◦C to the previous day/night cycle for 5–7 days recovery. Phenotypes before and after heat treatment were photographically documented.

#### Ion Leakage Assay

fpls-09-01710 November 22, 2018 Time: 10:37 # 3

Electrolyte leakage was measured as previously described (Camejo et al., 2005; Zang et al., 2017a,b). Leaf segments of uniform maturity were cut into disks and washed three times with de-ionized water to eliminate external residues. Six disks were placed in test tube flasks with 20 mL of de-ionized water and incubated at 42◦C for 1 h. After incubation at room temperature for 24 h, the conductivity of the solutions was determined with a Horiba Twin Cond B-173 conductivity meter (HORIBA Ltd., Kyoto, Japan) and noted as T1. Next, the samples were boiled for 15 min to kill the tissues, followed by incubation at room temperature for 24 h. The conductivities of the solutions were then recorded as T2. Ion permeability was calculated as T1/T2. The experiment was repeated three times independently. Student's t-test in Microsoft excel was used to determine the presence of significant differences (∗P < 0.05, ∗∗P < 0.01).

#### Dehydration Tolerance Assay

In Arabidopsis, 10-day-old seedlings of WT and TaPEPKR2 transgenic lines germinated on MS medium solidified with 0.8% agar were planted in identical pots, which contained 30 g mixed soil (vermiculite: humus = 1:1) and were added the same volume of water. Then, they were cultured in the greenhouse under optimum growth conditions (16-h-light/8-h-dark cycle, 150 µmol m−<sup>2</sup> s −1 , 22◦C/18◦C, 60% humidity). For dehydration tolerance assay, seedlings were subjected to water deprivation for 25 days, and then re-watered for about ∼1 week. In wheat, seeds of LC10 and TaPEPKR2 transgenic lines were planted in water. 5-day-old seedlings of WT and TaPEPKR2 transgenic lines were placed in water and 20% PEG conditions for 5 days. Student's t-test in Microsoft excel was used to determine the presence of significant differences (∗P < 0.05).

# RESULTS

#### Cloning of a TaPEPKR2 Gene From TAM107

Microarray analysis with the Affymetrix GeneChip <sup>R</sup> Wheat Genome Array previously indicated that probe "Ta.10701.1.S1\_at" was induced 22.73-fold after 40◦C treatment for 1 h (Qin et al., 2008). Based on this probe sequence, we cloned the open reading frame (ORF) of TaPEPKR2 from wheat cultivar "TAM107" (previously named TaSTK, GenBank Accession No. GU213488.1). The complete ORF of TaPEPKR2 is 1347 bp and encodes a polypeptide of 448 amino acid residues. Based on the wheat cultivar Chinese Spring genome sequence, TaPEPKR2 was mapped to chromosome 5B. The protein sequence showed homology with PEPKR2 family members from other plant species, including Zea mays (ZmPEPKR2, 81.46%), Oryza sativa (OsPEPKR2, 81.98%), and A. thaliana (AtPEPKR2, 53.18%) (**Figure 1**). SMART analysis indicated that the amino acid sequence of TaPEPKR2 possessed a S\_TKc motif, with both N- and C-terminal extensions (**Figure 1** and **Supplementary Figure S1**).

### TaPEPKR2 Is Induced by Heat and Dehydration Stress Treatment

Relative expression levels of TaPEPKR2 in wheat under heat and 20% PEG stress conditions were determined by RT-qPCR using gene-specific primers (**Figure 2**). The relative expression of TaPEPKR2 increased and peaked at 1 h after heat treatment at 40◦C and then decreased, but increased mRNA abundance was maintained (**Figure 2A**). Following treatment with 20% PEG, TaPEPKR2 expression increased gradually and peaked at 12 h (**Figure 2B**). These results

indicate that TaPEPKR2 can be induced by heat and 20% PEG treatment.

#### Overexpression of TaPEPKR2 in Wheat Conferred Heat Tolerance at the Seedling Stage

To understand the function of TaPEPKR2, we transformed it into wheat cultivar LC10 under the control of the maize ubiquitin promoter by particle bombardment. A total of five transgenic lines were analyzed over T<sup>1</sup> and T<sup>2</sup> generations by PCR analysis with specific corresponding primers. Three lines (L3, L4, and L5) that exhibited TaPEPKR2 up-regulation in shoots at the seedling stage (**Supplementary Figure S2**) were selected for further analysis.

To examine the applicability of TaPEPKR2 for thermotolerance transgenic breeding, we characterized the phenotypes of TaPEPKR2 transgenic wheat at various developmental stages. Under optimum growth conditions, no visible difference was found between transgenic and WT plants (**Figure 3A**). However, after heat treatment and the recovery stage, WT plants wilted slightly more rapidly than TaPEPKR2 transgenic lines (**Figure 3A**). Electrolyte leakage is an indicator to reflect heat stress-induced membrane injury. Thus, we evaluated electrolyte leakage with detached leaves after heat treatment. The results showed that transgenic lines exhibited significantly reduced electrolyte leakage compared with LC10 (**Figure 3B**).

#### TaPEPKR2 Overexpression in Wheat Enhanced Tolerance to Dehydration Stress

Because TaPEPKR2 was induced by 20% PEG treatment, this suggested that TaPEPKR2 may be involved in dehydration

stress tolerance. As expected, TaPEPKR2 transgenic lines showed significantly higher total root lengths in the presence of 10% PEG than WT (**Figure 4**). These results indicate that the overexpression of TaPEPKR2 in wheat conferred dehydration stress tolerance.

# TaPEPKR2 Overexpression in Dicotyledonous Arabidopsis Also Enhanced Tolerance to Heat and Dehydration Stresses

To characterize the biological functions of TaPEPKR2, we overexpressed it in Arabidopsis. RT-qPCR results indicated that all seven transgenic lines showed high expression levels of TaPEPKR2, with highest expression detected in transgenic line L7 (**Figure 5A**). To further validate the function of TaPEPKR2 in plant tolerance to heat and dehydration stresses, we observed the phenotypes of TaPEPKR2 transgenic Arabidopsis at various developmental stages but found no morphological differences between transgenic lines and WT (data not shown).

Seven-day-old TaPEPKR2-OE and WT seedlings grown at 22◦C were subjected to heat treatment at 45◦C for 2 h (**Figure 5B**). Before the treatment, there was no evident difference between transgenic and WT lines (**Figure 5C**), whereas after the treatment, the survival rate of transgenic lines L3 and L5 was much higher than that of WT seedlings (**Figures 5D,E**). Electrolyte leakage of detached leaves was evaluated under heat stress conditions. The results showed that detached leaves of WT plants had released more electrolytes than transgenic leaves (**Figure 5F**). These results indicate that TaPEPKR2 overexpression in Arabidopsis also enhances thermotolerance.

To further investigate the performance of transgenic TaPEPKR2Arabidopsis plants, lines L1, L3, and L5 were selected for dehydration tolerance tests before bolting (**Figure 6**). After dehydration stress (1 week after watering), around 95% of WT seedlings died, whereas 35–70% of TaPEPKR2 transgenic seedlings survived (**Figures 6A,B**). We further validated relative water loss rates of detached leaves (**Figures 6C,D**). Compared with WT plants, lines L1, L3, and L5 displayed lower relative water loss rates, and the final relative water content of detached rosette leaves from lines L3 and L5 was significantly higher than that of controls. These results indicated that the overexpression of TaPEPKR2 resulted in dehydration tolerance in Arabidopsis.

# DISCUSSION

In the present study, we identified a wheat PEP carboxylase kinase-related kinase gene, TaPEPKR2, that we showed to be involved in both heat and dehydration stress tolerance. SMART analysis revealed that the TaPEPKR2 protein sequence possessed a S\_TKc motif (**Supplementary Figure S1**). These results suggested that TaPEPKR2 is a typical serine–threonine protein kinase gene. Sequence alignment indicated that TaPEPKR2 shares 81.98% similarity with OsPEPKR2, 81.46% similarity with ZmPEPKR2, and 53.18% similarity with AtPEPKR2, suggesting that it is highly conserved. The S\_TKc motif of PEPKRs showed the highest homology

with PPCKs, however, PEPKR proteins contain both Nand C-terminal extensions without the PPCK superfamily (Hrabak et al., 2003). These results indicated that PEPKR and PPCK functions have commonalities but also exhibit differences.

The upstream sequence of TaPEPKR2 was analyzed for the presence of cis-acting elements by using the PlantCARE database (**Supplementary Table S2**). We found the promoter region of TaPEPKR2 contained the putative CGTCA-motif/TGACG-motif, ABRE, Skn-1-motif/GCN4, TC-rich repeats, GARE-motif/TATC-box, TCA-element, MBS and WUN-motif, indicating the expression of TaPEPKR2 could be regulated by several types of stresses, including abiotic and biotic stresses. The expression pattern is a direct indication of gene involvement in developmental events. In this study, TaPEPKR2 expression was induced during heat and dehydration stress (**Figure 2**); however, there was no report of TaPEPKR2 homologs from other species previously. The expression pattern of TaPEPKR2 homologous genes was analyzed in the Arabidopsis Weigelworld Database<sup>2</sup> and Plant Expression Database<sup>3</sup> . In Arabidopsis, AtPEPKR2 was induced by heat and drought stress

<sup>2</sup>http://jsp.weigelworld.org/expviz/expviz.jsp <sup>3</sup>http://www.plexdb.org/

within a small range. In durum wheat cv. Cappelli, neither heat stress nor drought stress alone could induce TdPEPKR2 expression, however, this was achieved by a combination of the two. In durum wheat cv. Ofanto, TdPEPKR2 was induced by heat stress within a small range, but not by drought stress. However, their combination significantly induced TdPEPKR2 expression. In common, wheat drought sensitive cv. A24-39, TaPEPKR2 was weakly induced by drought stress, while in common wheat drought tolerant cv. Y12-3, TaPEPKR2 was significantly induced by drought stress (Krugman et al., 2010). In barley caryopsis, the orthologous gene of TaPEPKR2 (probe Barley1\_05454) was induced by 0.5, 3, and 6 h heat stress treatment (Mangelsen et al., 2010). The expression pattern of TaPEPKR2 and its orthologous genes induced by high temperature and dehydration hinted the function of PEPKR2 in heat and drought tolerance.

Our results revealed that overexpression of TaPEPKR2 in wheat imparted tolerance to heat and dehydration stresses compared to wild type plants. It was fist report about function of PEPKRs in plant. PEPKRs belonged to the CDPK-SnRK superfamily (Hrabak et al., 2003). Not surprisingly, many of these kinases have been implicated in response to both heat and/or dehydration stresses. PEPKRs also called PPCK related kinases and shared catalytic domain with PPCK. Previous studies have concluded that PPCK activity in rice positively regulates PEPC activity during exposure to osmotic stress (Feria et al., 2016; Liu et al., 2017). Furthermore, PPCK gene expression and activity are regulated by signaling molecules such as Ca2<sup>+</sup> and H2O<sup>2</sup> (Liu et al., 2017). Are there PEPRKs regulatory proteins which play important role in calcium and ROS homeostasis under abiotic stress? What potential specific substrates they have? Further studies are needed to determine the molecular mechanisms involved in enhancing heat and dehydration stresses tolerance in TaPEPKR2-OE plants. Relative expression levels of the genes functioning in ABA biosynthesis (ABA1, AT5G67030), signaling (ABI3, AT3G24650), heat shock protein 70 (HSP70, AT3G12580), and 17.6A (HSP17.6A, AT5G12030) were investigated in TaPEPKR2 transgenic Arabidopsis plants (**Supplementary Figure S3**). The expressions of ABA1 and HSP70 were found to be unchanged, however, ABI3 and HSP17.6A were elevated constitutively. These results present some clues of TaPEPKR2 regulating ABA signaling and heat shock proteins in heat and dehydration stresses.

#### AUTHOR CONTRIBUTIONS

QS, HP, and XZ designed the research. XZ, XG, KH, FW, and XT performed the research. XZ, XG, HP, ZN, YY, ZH, and

MX analyzed the data. XZ and HP wrote the paper. All authors read and approved the final manuscript.

#### FUNDING

This work was supported by the National Natural Science Foundation of China (31571747) and National Key Project for Research on Transgenic (2016ZX08002-002).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01710/ full#supplementary-material

# REFERENCES


FIGURE S1 | The functional domain of the TaPEPKR2 protein sequence identified by the SMART program.

FIGURE S2 | PCR analysis of TaPEPKR2 transgenic wheat plants. (A) Confirmation of TaPEPKR2 insertion in LC10 by PCR analysis of H2O, LC10, PC, L3, L4, and L5 transgenic plants. PC: Ubi::TaPEPKR2 vector was used as the positive control. (B) Relative expression of TaPEPKR2 as shown by RT-qPCR. β-actin was used as the internal control. Data are the mean ± SD of three independent biological replicates.

FIGURE S3 | Relative expression level of ABA1 (A), ABI3 (B), HSP70 (C), and HSP17.6A (D) in 10-day-old TaPEPKR2 transgenic Arabidopsis plants, as determined by RT-qPCR. Data represent the mean of three replicates ± SD.

TABLE S1 | Primer sequences used in this study.

TABLE S2 | Cis-elements present in the promoters of TaPEPKR2.

FILE S1 | Coding sequence of TaPEPKR2.


**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 Zang, Geng, He, Wang, Tian, Xin, Yao, Hu, Ni, Sun and Peng. 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.

# OsDIRP1, a Putative RING E3 Ligase, Plays an Opposite Role in Drought and Cold Stress Responses as a Negative and Positive Factor, Respectively, in Rice (Oryza sativa L.)

Li Hua Cui1,2† , Hye Jo Min1,2† , Mi Young Byun<sup>3</sup> , Hyeong Geun Oh1,2 and Woo Taek Kim1,2 \*

<sup>1</sup> Department of Systems Biology, College of Life Science and Biotechnology, Yonsei University, Seoul, South Korea, 2 Institute of Life Science and Biotechnology, Yonsei University, Seoul, South Korea, <sup>3</sup> Unit of Polar Genomics, Korea Polar Research Institute, Incheon, South Korea

#### Edited by:

Dae-Jin Yun, Konkuk University, South Korea

#### Reviewed by:

Byeong-ha Lee, Sogang University, South Korea June M. Kwak, Daegu Gyeongbuk Institute of Science and Technology (DGIST), South Korea

> \*Correspondence: Woo Taek Kim wtkim@yonsei.ac.kr

†These authors have contributed equally to this work

#### Specialty section:

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

Received: 09 October 2018 Accepted: 19 November 2018 Published: 05 December 2018

#### Citation:

Cui LH, Min HJ, Byun MY, Oh HG and Kim WT (2018) OsDIRP1, a Putative RING E3 Ligase, Plays an Opposite Role in Drought and Cold Stress Responses as a Negative and Positive Factor, Respectively, in Rice (Oryza sativa L.). Front. Plant Sci. 9:1797. doi: 10.3389/fpls.2018.01797 As higher plants are sessile organisms, they are unable to move to more favorable places; thus, they have developed the ability to survive under potentially detrimental conditions. Ubiquitination is a crucial post-translational protein modification and participates in abiotic stress responses in higher plants. In this study, we identified and characterized OsDIRP1 (Oryza sativa Drought-Induced RING Protein 1), a nuclearlocalized putative RING E3 ubiquitin (Ub) ligase in rice (Oryza sativa L.). OsDIRP1 expression was induced by drought, high salinity, and abscisic acid (ABA) treatment, but not by low temperature (4◦C) stress, suggesting that OsDIRP1 is differentially regulated by different abiotic stresses. To investigate its possible role in abiotic stress responses, OsDIRP1-overexpressing transgenic rice plants (Ubi:OsDIRP1-sGFP) were generated, and their phenotypes were analyzed. The T4 Ubi:OsDIRP1-sGFP lines showed decreased tolerance to drought and salt stress as compared to wild-type rice plants. Moreover, Ubi:OsDIRP1-sGFP progeny were less sensitive to ABA than the wild-type during both germination and post-germination growth. In contrast, Ubi:OsDIRP1-sGFP plants exhibited markedly higher tolerance to prolonged cold (4◦C) treatment. These results suggest that OsDIRP1 acts as a negative regulator during drought and salt stress, whereas it functions as a positive factor during the cold stress response in rice.

Keywords: abscisic acid, cold stress, drought stress, opposite response, rice (Oryza sativa), RING E3 Ub ligase

# INTRODUCTION

As sessile organisms, higher plants are constantly exposed to diverse environmental stresses, such as water deficit, high salinity, and extreme temperatures, throughout their life cycle that they cannot avoid. These detrimental conditions can disrupt plant growth, and on a larger scale, reduce global crop production (Ray et al., 2015). In addition to abiotic stress in plants, rapid increases in population also affect global food security (Lesk et al., 2016). Rice is a primary source of food for more than half of the world's population. Thus, improving the stress tolerance of rice is crucial to maintain high crop productivity and fulfill the growing demand for food.

Plant stress response and acclimation are regulated by a complex network of cellular factors, including stress perception and related signaling pathways, changes in hormonal and metabolic balances, and transcriptional and post-transcriptional regulation of stress-associated genes (Osakabe et al., 2014; Guerra et al., 2015). Ubiquitination is a post-translational modification of diverse cellular proteins. In higher plants, the ubiquitination pathway is critically involved in the abiotic stress response (Santner and Estelle, 2010; Lyzenga and Stone, 2012; Yu et al., 2016). Ubiquitination is a multistep process, in which ubiquitin (Ub) is conjugated to target proteins by successive reactions catalyzed by three types of enzymes, E1 Ub activating enzymes, E2 Ub conjugating enzymes, and E3 Ub ligases. In general, E3 ligases specifically recognize the target protein (Smalle and Viestra, 2004). E3 Ub ligases are divided into single- and multi-subunit types depending on their structure. Single-subunit E3 Ub ligases are further categorized into distinct families based on the presence of specific domains, such as the really interesting new gene (RING), U-box, and homology to E6AP C terminus (HECT) domains. Skp1-cullin-F-box (SCF) and anaphase-promoting complex (APS) belong to the multi-subunit E3 Ub ligases (Lee and Kim, 2011).

Abscisic acid (ABA) is a well-known plant stress hormone that plays a pivotal role in the response to environmental stimuli (Zhu, 2002; Yamaguchi-Shinozaki and Shinozaki, 2006). Cellular levels of ABA are rapidly increased under stress conditions, which induces various physiological responses, such as stomata closure, growth inhibition, and the production of proline and sugars. Different kinds of E3 Ub ligases participate in both ABA production and the ABA-mediated stress response. For example, the Arabidopsis U-box E3 Ub ligase SAUL1/PUB44 negatively regulates the ABA biosynthetic enzyme aldehyde oxidase 3 (AAO3), whereas the RING E3 Ub ligase XERICO promotes ABA production by enhancing the expression of 9-cis-epoxycarotenoid dioxygenase 3 (NCED3) (Ko et al., 2006; Raab et al., 2009). In addition, E3 Ub ligases regulate ABA perception and the signal transduction cascade by mediating the degradation of ABA receptors, SnRK protein kinases, and transcription factors (Chen et al., 2013; Lyzenga et al., 2013; Bueso et al., 2014; Irigoyen et al., 2014; Zhao et al., 2017).

RING E3 Ub ligases widely exist in eukaryotic organisms. Compared with Saccharomyces cerevisiae, which has 47 RINGtype E3 encoding genes, rice and Arabidopsis contain 378 and 499 genes, respectively (Kraft et al., 2005; Stone et al., 2005; Mazzucotelli et al., 2006; Li et al., 2008; Du et al., 2009). Recent studies have revealed the cellular roles of RING E3 Ub ligases in response to abiotic stress in the dicot model plant Arabidopsis (Cho et al., 2017; Stone, 2018). In contrast, our knowledge concerning the roles of RING E3 ligases in the monocot cereal rice in response to abiotic stresses is relatively rudimentary. SALT-AND DROUGHT-INDUCED RING FINGER1 (SDIR1) is a positive regulator of ABA-mediated drought stress response in Arabidopsis (Zhang et al., 2007; Zhang et al., 2015). OsSDIR1 is a rice ortholog of SDIR1 and positively involved in the drought stress response in rice plants (Gao et al., 2011). Oryza sativa Drought-Induced SINA protein1 (OsDIS1), a rice RING E3 Ub ligase, plays a negative role in drought stress tolerance through posttranslational regulation of Oryza sativa NIMA-related kinase 6 (OsNek6), which is a tubulin complex-related serine/threonine protein kinase, and transcriptional regulation of stress-related genes (Ning et al., 2011). Recently, rice Oryza sativa Chloroplast Targeting RING E3 ligase 1 (OsCTR1) and Oryza sativa Arsenic-Induced RING E3 ligase 1 (OsAIR1) were identified as positive factors in the drought stress response and post-germination growth under arsenate stress conditions, respectively (Lim et al., 2014; Hwang et al., 2016). Furthermore, knock-down of Oryza sativa Stress-Related RING Finger Protein 1 (OsSRFP1) resulted in increased tolerance to high salt and cold stress in rice (Fang et al., 2015).

In this study, we identified and characterized Oryza sativa Drought-Induced RING Protein 1 (OsDIRP1), which is a nuclear-localized putative RING-type E3 Ub ligase in rice. OsDIRP1 was induced by drought, high salinity, and ABA treatments, but not by cold stress (4◦C). OsDIRP1-overexpressing T4 transgenic rice plants (Ubi:OsDIRP1-sGFP) exhibited reduced tolerance to drought and salt stress as compared to wild-type rice plants. Ubi:OsDIRP1-sGFPs plants were less sensitive to ABA than wild-type plants during both the germination and postgermination stages. In contrast, the Ubi:OsDIRP1-sGFP progeny displayed markedly increased tolerance to cold stress relative to the wild-type plants. These results indicate that OsDIRP1 acts as a negative regulator of the drought and salt stress responses, while it works as a positive factor of the cold stress response in rice.

# MATERIALS AND METHODS

#### Plant Materials and Growth Conditions

Dry seeds of the rice (Oryza sativa L.) japonica variety 'Dongjin' were washed with 70% ethanol and subsequently sterilized with a 0.4% NaClO solution. Sterilized seeds were germinated and grown on half-strength Murashige and Skoog (MS) medium (Duchefa Biochemie, Haarlem, Netherlands) supplemented with 3% sucrose and 0.75% phytoagar (pH 5.6–5.8) for 10–12 days. Germinated seedlings were transplanted to soil and grown at 25–30◦C under long-day conditions (16 h light and 8 h dark) in a greenhouse. Whole seedlings of 10-day-old wild-type rice plants were subjected to drought (dried on filter paper for 0, 0.25, 0.5, 1, 2, and 4 h), salt (soaked in a 200 mM NaCl solution for 0, 1, 4, and 6 h), cold (incubated at 4◦C for 0, 12, and 24 h), and 100 µM ABA (0, 3, 6, 12, and 24 h) treatments.

#### RNA Extraction, RT-PCR, and Real-Time Quantitative RT-PCR (qRT-PCR) Analyses

Total RNA was extracted from various tissues of wild-type rice plants and stress-treated seedlings by using the Easy Spin Plants Total RNA Extraction kit (iNtRON Biotechnology, Korea) according to the manufacturer's protocol. RNA was quantified using a spectrophotometer (NanoDrop 1000; Thermo Scientific, United States). Total RNA (3 µg) was used to synthesize cDNA by using TOPscript Reverse Transcriptase (Enzynomics, Korea) and oligo (dT) primers. The PCR amplification procedure was as follows: 5 min of denaturation and enzyme activation at 95◦C, followed by 28–33 cycles of 30 s at 95◦C, 30 s at 55◦C, and 20 s

at 72◦C. PCR products were separated on a 2% agarose gel and visualized under UV light. The OsUbiquitin gene was included as a loading control. OsRab16b was used as a positive control for the drought, salt, and ABA treatments, whereas OsDREB1A was used as a positive control for cold stress. qRT-PCR was carried out using PikoReal real-time PCR (Thermo Scientific, United States) with SYBR Premix Ex Taq II (Takara, Japan) as described previously (Kim et al., 2017). The primer sequences used for RT-PCR and qRT-PCR are listed in **Supplementary Table S1**.

#### In vitro Self-Ubiquitination Assay

An in vitro self-ubiquitination assay was performed as described in a previous study (Bae et al., 2011). Bacterially expressed GST-OsDIRP1 recombinant fusion protein was incubated at 30◦C for 1 h in ubiquitination reaction buffer (10 mM ATP, 0.5 mM DTT, 2.5 mM MgCl2, 50 mM Tris-HCl, pH 7.5, and 0.5 µg Ub) in the presence or absence of E1 (Arabidopsis UBA1) and E2 (Arabidopsis UBC8). The reaction products were subjected to immunoblot analysis with anti-GST and anti-Ub antibodies as described previously (Bae and Kim, 2014).

### Subcellular Localization

The full-length coding region of OsDIRP1 was tagged with synthetic green fluorescent protein (sGFP) in-frame and inserted into the pBI221 binary vector. Isolation of rice protoplasts and transfection of the vectors into the protoplasts were performed according to the method of Yoo et al. (2007), with modifications as described by Byun et al. (2017). The 35S:sGFP, 35S:OsDIRP1 sGFP, and 35S:NLS-mRFP constructs were transfected or cotransfected into the isolated protoplasts and incubated overnight. The fluorescent signals of the expressed proteins were observed by fluorescence microscopy (BX51; Olympus, Japan) as described previously (Seo et al., 2016). NLS-mRFP was used as a nuclear marker protein.

#### Generation of OsDIRP1-Overexpressing (Ubi:OsDIRP1-sGFP) and RNAi-Mediated Knock-Down (Ubi:RNAi-OsDIRP1) Transgenic Rice Plants

The Ubi:OsDIRP1-sGFP and Ubi:RNAi-OsDIRP1 chimeric constructs were transformed into Agrobacterium tumefaciens strain LBA4404 via electroporation as described by Park et al. (2016). Transgenic rice plants were produced by using pGA2897 binary vector plasmids that contained the maize ubiquitin promoter (Ubi). All rice transformation procedures were performed as described previously by Byun et al. (2017). Generated transgenic rice T0 plants were transplanted in the soil under greenhouse condition and further propagated in the field condition. The harvested transgenic seeds of individual plant were germinated in hygromycin containing media to select the homozygous plants in T2 generation. Independent homozygous T4 OsDIRP1-overexpressing (Ubi:OsDIRP1-sGFP lines #1, #2, and #3) and homozygous T4 RNAi knock-down (Ubi:RNAi-OsDIRP1 lines #1, #2, and #3) transgenic rice plants were used for phenotypic analyses.

# Genomic DNA Extraction and DNA Gel Blot Analysis

Total genomic DNA was extracted from the leaves of 5-weekold wild-type and transgenic rice plants by using the CTAB (2% CTAB, 20 mM EDTA, 1.4 M NaCl, 2% PVP-40, and 100 mM Tris-HCl, pH 8.0) method. Total genomic DNA was digested with HindIII restriction enzyme (Thermo Scientific, United States) and separated by electrophoresis on a 0.7% agarose gel. Then, the DNA on the gel was transferred to a Hybond-N nylon membrane (GE Healthcare, United Kingdom) by the capillary transfer method, and the blot was hybridized to a <sup>32</sup>P-labeled hygromycin B phosphotransferase (Hph) probe as described by Byun et al. (2018). The autoradiography signals were visualized using a BAS2500 Bio-Imaging Analyzer (Fuji Film, Japan).

### Phenotype Evaluation of Ubi:OsDIRP1-sGFP and Ubi:RNAi-OsDIRP1 Transgenic Rice Plants Grown in a Paddy Field

The agronomic traits of the T4 Ubi:OsDIRP1-sGFP and T4 Ubi:RNAi-OsDIRP1 transgenic rice plants grown under normal field conditions were analyzed as described by Park et al. (2016). The following agronomic traits were scored: the number of panicles, panicle length, number of primary branches per panicle, 1000-grain weight, and total grain weight.

#### Stress Treatment of Wild-Type, Ubi:OsDIRP1-sGFP, and Ubi:RNAi-OsDIRP1 Rice Plants

For the drought stress treatment, 5-week-old wild-type, T4 Ubi:OsDIRP1-sGFP (independent lines #1, #2, and #3), and T4 Ubi:RNAi-OsDIRP1 (independent lines #1, #2, and #3) plants were grown in the same pot without watering for 8–9 days until the leaves were wilted. After 8–9 days of dehydration, the plants were re-watered, and their growth profiles were monitored at different time points during stress recovery. Plants that resumed growing, with green and healthy leaves, were regarded as having survived. The survival rates and total leaf chlorophyll content were determined at 15–20 days of recovery. Data were obtained from at least six independent biological experiments. To measure the rate of leaf water loss, detached leaves from 6-week-old wild-type and transgenic rice plants were placed on a filter paper at room temperature and weighed after different time intervals (15, 30, 60, 90, 120, 180, 240, and 300 min). The water loss rate was calculated as the percentage of the initial fresh weight.

For the salt stress treatment, 5-week-old wild-type and transgenic rice plants, which had been grown in the same pot under normal conditions, were transferred to a tray containing a solution of 200 mM NaCl. After 16–18 days of salt treatment, the plants were recovered by normal irrigation. Plants that resumed growing, with green and healthy leaves, were regarded as having survived. The survival rates and total leaf chlorophyll content were determined after 1 month of recovery. At least five independent biological experiments were performed.

For the cold stress treatment, 5-week-old plants grown in the same pot at 28◦C under long-day conditions (16 h light and 8 h dark) were transferred to a cold room at 4◦C for 6–8 days, after which the plants were recovered at 28◦C for 25–30 days, and their growth patterns were monitored as described previously (Byun et al., 2015). Electrolyte leakage was analyzed using 8-dayold seedlings at different time points before and after cold stress treatment (0, 5, and 10 days at 4◦C). Seedlings of wild-type and transgenic plants were soaked in a test tube containing 35 mL of distilled water on an orbital shaker (200 rpm) at 28◦C overnight. The electrolyte conductivity of each sample was determined before and after autoclaving by using a conductivity meter (Orion Star A212; Thermo Scientific, United States) according to the method of Min et al. (2016).

#### Leaf Disk Assay and Measurement of Total Leaf Chlorophyll Content

Leaf disks (0.5 cm in diameter) prepared from 5-week-old wildtype and transgenic plants were floated in various concentrations (0, 200, 400, 600, and 800 mM) of NaCl for 3 days under long-day conditions (16 h light and 8 h dark). Representative photos were taken after 3 days of incubation, and the total leaf chlorophyll content of each sample was measured.

Total leaf chlorophyll (chlorophyll a + chlorophyll b) content was measured according to Lichtenthaler (1987), with modifications as described by Min et al. (2016). The amounts of chlorophyll a + chlorophyll b were measured at 664.2 nm and 648.6 nm, respectively, with an ELISA microplate reader (VERSAmax, Molecular Devices, United States) and normalized to the dry weight of the leaves of each genotype.

#### ABA-Dependent Germination and Post-germination Tests

Wild-type and T4 Ubi: OsDIRP1-sGFP transgenic (lines #1, #2, and #3) seeds were germinated and grown on half-strength MS medium supplemented with different concentrations (0, 3, and 5 µM) of ABA (Sigma-Aldrich, Germany) at 28◦C under long-day conditions (16 h light and 8 h dark). After 7 days of germination, shoot and root lengths were measured. More than 50 seeds were used in each assay, and three independent biological assays were performed.

For the post-germination assays, wild-type and transgenic seeds were germinated on half-strength MS medium for 2 days. Then, the germinated seedlings were transferred to half-strength MS medium containing 0, 3, 5, or 10 µM ABA and incubated for another 6 days. At least 100 seeds were used in each assay, and four independent biological assays were performed.

#### RESULTS

#### Identification and Characterization of OsDIRP1 in Rice

To identify the RING E3 Ub ligases that participate in abiotic stress responses in rice, we analyzed the expression patterns of the RING-type E3 ligase gene family under drought stress based on publicly available rice microarray data<sup>1</sup> . Os06g47270 was identified to be induced by water deficit and was named Oryza sativa Drought-Induced RING Protein 1 (OsDIRP1). OsDIRP1 encodes a 375 amino acid protein with a single C3H2C3-type RING motif in the C-terminal region, a nuclear localization sequence (NLS), and a putative beta-ketoacyl synthase active site in the N-terminal region (**Figure 1A**). OsDIRP1 was 48%, 47%, and 43% identical to putative RING E3 Ub ligases in monocot millet (Setaria italica), maize (Zea mays), and sorghum (Sorghum bicolor), respectively (**Supplementary Figures S1A,B**).

OsDIRP1 transcripts were detected in all examined rice tissues, including the callus, 11-day-old seedlings, developing and mature leaves, stems, roots, flowers, and panicles (**Figure 1B**). Furthermore, OsDIRP1 was induced by drought (0.25–4 h), high salt (300 mM NaCl for 1–8 h), and ABA (100 µM for 3–24 h) treatments, but not by cold temperature (4◦C for 12 and 24 h) (**Figure 1C**). The original figures were added in **Supplementary Materials** indicated as **Supplementary Presentation S1**.

The subcellular localization of OsDIRP1 was investigated via the protoplast transient expression system. The 35S:OsDIRP1 sGFP chimeric construct was expressed in wild-type rice protoplasts with or without 35S:NLS-mRFP, and the expressed proteins were visualized by fluorescence microscopy. The results revealed that the fluorescence signal of OsDIRP1-sGFP was predominantly located in the nucleus, where it overlapped with the nuclear marker protein NLS-mRFP (**Figure 1D**), which suggests that OsDIRP1 is a nuclear protein.

#### Generation and Molecular Characterization of OsDIRP1-Overexpressing and RNAi-Mediated Knock-Down Transgenic Rice Plants

To investigate the cellular role of OsDIRP1 in abiotic stress responses, OsDIRP1-overexpressing (Ubi:OsDIRP1-sGFP) and RNAi-mediated OsDIRP1 knock-down (Ubi:RNAi-OsDIRP1) transgenic rice plants were generated. Under normal growth conditions, there was no detectable morphological difference between the wild-type and transgenic rice plants (**Figure 2A**). The results of genomic Southern blot analysis showed that there were three independent lines of each genotype (**Figure 2B**). In the OsDIRP1-overexpressing transgenic lines, the amount of OsDIRP1-sGFP transcript was markedly increased as compared to the level in wild-type plants under normal growth conditions as measured by RT-PCR (**Figure 2C**). Overexpression of OsDIRP1-sGFP protein was confirmed by immunoblot analysis with an anti-GFP antibody (**Figure 2D**). The original figures were added in **Supplementary Materials** indicated as **Supplementary Presentation S1**. The level of OsDIRP1 transcript was partially suppressed in the Ubi:RNAi-OsDIRP1 knock-down lines relative to that in the wild-type plants under both normal and drought conditions (**Figure 2E**). The agronomic traits of Ubi:OsDIRP1 sGFP and Ubi:RNAi-OsDIRP1 T4 progeny grown in a paddy field condition were analyzed with respect to the number

<sup>1</sup>http://signal.salk.edu/cgi-bin/RiceGE

of panicles, panicle length, number of primary branches per panicle, 1000-grain weight, and total grain weight. As shown in **Table 1** and **Supplementary Figure S2**, there was no significant difference among the wild-type, Ubi:OsDIRP1-sGFP, and Ubi:RNAi-OsDIRP1 plants. Thus, overexpression and knockdown of OsDIRP1 did not alter growth during the vegetative and reproductive stages. These OsDIRP1-overexpressing and RNAimediated OsDIRP1 knock-down T4 transgenic plants were used for subsequent phenotypic analysis of their response to abiotic stress.

#### OsDIRP1-Overexpressing Transgenic Rice Plants Exhibited Reduced Tolerance to Drought and Salt Stress Compared to Wild-Type Plants

Wild-type and Ubi:OsDIRP1-sGFP T4 transgenic (independent lines #1, #2, and #3) rice plants were grown at 28◦C for 5 weeks under long-day conditions (16 h light and 8 h dark). The plants were then subjected to drought stress by withholding water for 9 days. After 9 days of dehydration, these plants were rewatered, and their growth patterns were monitored for 15–20 days after initiation of re-watering. Of the wild-type rice plants, 61.0 ± 5.7% resumed growth after rehydration (**Figure 3A**). In contrast, most of the Ubi:OsDIRP1-sGFP lines exhibited pale green and yellowish leaves, and only 10.7 ± 2.9% – 19.0 ± 3.0% of the OsDIRP1-overexpressing lines survived after rehydration (**Figure 3B**).

Mature leaves were detached from plants of each genotype to measure the chlorophyll content (chlorophyll a + chlorophyll b) before and after the drought treatment. Before drought treatment, the total leaf chlorophyll amounts of the wild-type and Ubi:OsDIRP1-sGFP plants were indistinguishable. However, the OsDIRP1-overexpressing progeny contained much lower amounts of chlorophyll (1.1 ± 0.4 – 1.6 ± 0.7 mg/g DW) as compared to those of the wild-type plants (6.9 ± 0.9 mg/g DW) after recovery from drought treatment (**Figure 3C**). In addition, the detached leaves of OsDIRP1-overexpressors lost their water content faster than the wild-type leaves. After a 3-h incubation at room temperature, wild-type and Ubi:OsDIRP1-sGFP leaves lost 58.1 ± 4.1% and 59.5 ± 3.3% – 61.3 ± 1.8% of their fresh weight, respectively (**Figure 3D**). After a 6-h incubation, 51.2 ± 3.5% and 46.0 ± 3.0% – 47.7 ± 2.2% of the initial fresh weight were retained in

FIGURE 2 | Molecular characterization of OsDIRP1-overexpressing and RNAi-mediated knock-down transgenic rice plants. (A) Morphology of 2-month-old wild-type (WT), T4 Ubi:OsDIRP1-sGFP, and T4 Ubi:RNAi-OsDIRP1 rice plants grown under long-day conditions (16 h light and 8 h dark). (B) Genomic Southern blot analysis. Total leaf genomic DNA was isolated from wild-type (WT), T4 Ubi:OsDIRP1-sGFP (lines #1, #2, and #3), and T4 Ubi:RNAi-OsDIRP1 (lines #1, #2, and #3) rice plants. The DNA was digested with HindIII and hybridized to a <sup>32</sup>P-labeled hygromycin B phosphotransferase (Hph) probe under high stringency conditions. (C) RT-PCR analysis of the wild-type (WT) and T4 Ubi:OsDIRP1-sGFP (independent lines #1, #2, and #3) transgenic rice plants to examine OsDIRP1 transcript levels. OsUbiquitin was used as a loading control. (D) Immunoblot analysis of wild-type (WT) and T4 Ubi:OsDIRP1-sGFP plants. Total proteins were isolated using 2x SDS sample buffer and immunoblotted with anti-GFP antibody. Rubisco was used as an equal loading control. (E) RT-PCR analysis of the wild-type (WT) and T4 Ubi:RNAi-OsDIRP1 plants. RNA was isolated from whole seedlings of non-drought-treated (0 h) and drought-treated (4 h) wild-type (WT) and Ubi:RNAi-OsDIRP1 (lines #1, #2, and #3) plants. OsUbiquitin was used as a loading control.



Data are shown as mean ± SD from 10 plants of each genotype.

the wild-type and Ubi:OsDIRP1-sGFP leaves, respectively (**Figure 3D**). Thus, overexpression of OsDIRP1 resulted in reduced tolerance to dehydration stress, indicating that OsDIRP1 negatively influences the drought stress response in rice.

Because OsDIRP1 was induced by high salinity (**Figure 1B**), the salt tolerance of the Ubi:OsDIRP1-sGFP plants was evaluated. Wild-type and OsDIRP1-overexpressing rice plants were grown for 5 weeks under normal conditions and then irrigated with water containing 200 mM NaCl. After 16–18 days of salt treatment, these plants were transferred back to normal irrigation conditions and were allowed to grow for at least 1 month to recover, and their growth patterns were monitored. As shown in **Figures 4A,B**, the Ubi:OsDIRP1-sGFP transgenic lines displayed more evident developmental anomalies with markedly retarded growth as compared to the wild-type plants, and the survival rates of the Ubi:OsDIRP1-sGFP lines after salt treatment were 27.6 ± 5.7% – 47.9 ± 7.8%, while that of the wild-type plants was 72.6 ± 8.9%.

A leaf senescence assay was performed with leaf disks (0.5 cm in diameter) prepared from 5-week-old rice plants. The leaf disks were floated in a solution supplemented with different concentrations (0, 200, 400, 600, and 800 mM) of NaCl for 3 days under long-day conditions (16 h light and 8 h dark). The results

showed that the wild-type leaf segments retained approximately 78%, 50%, 38%, and 21% of their total leaf chlorophyll content when they were exposed to 200, 400, 600, and 800 mM NaCl, respectively (**Figures 4C,D**). On the other hand, the transgenic leaf segments were more sensitive to high salinity than the wild-type leaves, and approximately 55–74%, 19–22%, 14–16%, and 10–14% of the total leaf chlorophyll content was retained in response to 200, 400, 600, and 800 mM NaCl, respectively (**Figures 4C,D**). These results indicated that the OsDIRP1 overexpressing plants are less tolerant to high salinity than the wild-type plants.

We next examined the phenotype of Ubi:RNAi-OsDIRP1 knock-down transgenic plants in response to dehydration. Unlike the overexpressing lines, the T4 RNAi (independent lines #1, #2, and #3) plants were very similar to the wild-type plants in terms of drought stress tolerance (**Supplementary Figure S3**). We repeated this experiment, but failed to detect a difference in tolerance to drought stress between the wildtype and Ubi:RNAi-OsDIRP1 plants. These results led us to hypothesize that incomplete suppression of OsDIRP1 may not have detectable effects on stress tolerance. Alternatively, other RING E3 ligase homologs may have complemented the phenotype of the Ubi:RNAi-OsDIRP1 knock-down plants. Overall, our phenotypic analysis revealed that overexpression of OsDIRP1 reduced tolerance to both drought (**Figure 3**) and high salinity (**Figure 4**) in rice plants. With this in mind, we concluded that rice OsDIRP1 is a negative factor in the response to drought and salt stress.

#### Decreased ABA Sensitivity of the OsDIRP1-Overexpressing Transgenic Rice Plants

To examine the role of OsDIRP1 in the response to ABA, ABA-dependent germination and post-germination assays were performed. Wild-type and T4 Ubi:OsDIRP1-sGFP (lines #1, #2, and #3) seeds were germinated on half-strength MS medium supplemented with 0, 3, and 5 µM ABA. After 7 days of

incubation, the shoot and root lengths were measured. Without ABA, the germination rates of the wild-type and OsDIRP1 overexpressors were very similar (**Figures 5A,B**). In contrast, these genotypes were easily distinguishable in the presence of ABA. With 3 µM ABA, the shoot and root lengths of the wildtype seedlings were 1.6 ± 0.1 cm and 2.6 ± 0.2 cm, respectively, while those of Ubi:OsDIRP1-sGFP were 2.1 ± 0.1 cm – 3.0 ± 0.2 cm and 3.2 ± 0.1 cm – 3.8 ± 0.1 cm, respectively. These differences became even more evident at the higher concentration of ABA. In the presence of 5 µM ABA, the wildtype shoot and root lengths were 0.9 ± 0.1 cm and 1.6 ± 0.1 cm, respectively (**Figures 5A,B**). However, the shoot and root lengths of the Ubi:OsDIRP1-sGFP seedlings were 1.3 ± 0.1 cm – 1.8 ± 0.1 cm and 2.3 ± 0.1 cm – 2.9 ± 0.2 cm, respectively. These results indicated that the OsDIRP1-overexpressors were hyposensitive to ABA relative to the wild-type seedlings during the germination stage.

For the post-germination assay, wild-type and T4 Ubi:OsDIRP1-sGFP seeds were germinated on half-strength MS medium for 2 days and then transferred to medium containing 0, 3, 5, and 10 µM ABA. Then, the growth of these seedlings was monitored for 6 days after transfer. Consistent with the results of the germination assay, the OsDIRP1-overexpressing seedlings appeared to be ABA-insensitive during post-germination growth. For example, in the presence of 10 µM ABA, the shoot lengths of the wild-type and Ubi:OsDIRP1-sGFP plants were 2.3 ± 0.2 cm and 3.3 ± 0.2 cm – 3.6 ± 0.2 cm, respectively (**Figures 5C,D**). Further, the root lengths of these plants were 2.5 ± 0.3 cm and 3.2 ± 0.3 cm – 3.8 ± 0.3 cm, respectively, with 10 µM ABA. Thus, the OsDIRP1-overexpressors were insensitive to ABA during both the germination and post-germination stages, which is in agreement with the view that OsDIRP1 plays a negative role in the response to ABA. Taken together, the results of the phenotypic analyses in **Figures 3**–**5** suggest that OsDIRP1 acts as a negative factor in the ABA-mediated drought and high salt responses in rice.

#### OsDIRP1-Overexpressors Exhibited Enhanced Tolerance to Cold Stress Compared to Wild-Type Plants

Because the Ubi:OsDIRP1-sGFP lines were hypersensitive to drought and high salinity stress, we considered the possibility that Ubi:OsDIRP1-sGFP lines may also exhibit altered tolerance to low

temperature stress. To examine this possibility, wild-type and T4 Ubi:OsDIRP1-sGFP (lines #1, #2, and #3) plants were grown at 28◦C for 5 weeks under long-day conditions (16 h light and 8 h dark) and subsequently subjected to cold stress by transferring them to a cold room at 4◦C under continuous light. After 8 days of low temperature treatment, the plants were transferred back to the growth room at 28◦C and their growth patterns were monitored. Under our experimental conditions, most of the wild-type plants exhibited anomalous growth patterns, with pale green and yellowish leaves, after recovery from cold stress. They were unable to grow, and most eventually died, with a survival rate of 13.4 ± 3.6% (**Figures 6A,B**). In contrast, the OsDIRP1-overexpressors were clearly healthier, with higher survival rates (49.3 ± 8.7% for line #1, 66.5 ± 9.3% for line #2, and 69.4 ± 14.6% for line #3) than the wild-type plants.

An electrolyte leakage analysis was conducted using 8-dayold seedlings before and after cold stress treatment. Wild-type and OsDIRP1-overexpressing seedlings were soaked in 35 mL of distilled water at 4◦C, and the rates of electrolyte leakage were measured at different time points (0, 5, and 10 days). As shown in **Figure 6C**, the Ubi:OsDIRP1-sGFP seedlings exhibited lower ion leakage (10.0 ± 1.0% – 10.4 ± 1.0% at 5 days and 12.7 ± 1.1% – 14.3 ± 0.9% at 10 days) than the wild-type seedlings (12.0 ± 1.1% at 5 days and 22.9 ± 3.2% at 10 days) in response to prolonged cold stress. Consistently, the OsDIRP1-overexpressing progeny contained higher amounts of chlorophyll compared to that of the wild-type rice plants exposed to cold stress. After 1 month of recovery from cold treatment (4◦C), the leaf chlorophyll content of the wild-type plants was 1.5 ± 0.4 mg/g DW, whereas that of the Ubi:OsDIRP1-sGFP plants was 7.3 ± 1.2 – 8.0 ± 0.8 mg/g DW (**Figure 6D**).

Finally, we examined the phenotypes of T4 Ubi:RNAi-OsDIRP1 knock-down transgenic plants under low temperature stress. Similar to their drought phenotype, there was no evident phenotypic difference between the wild-type and OsDIRP1 RNAi-knock-down plants with regard to cold stress tolerance (**Supplementary Figure S4**).

#### Cold-Responsive Genes Are Upregulated in OsDIRP1-Overexpressing Transgenic Rice Plants Compared to the Levels in Wild-Type Plants

Because Ubi:OsDIRP1-sGFP plants showed markedly enhanced tolerance to cold stress, we compared the expression patterns

(D) Total leaf chlorophyll content of wild-type and T4 Ubi:OsDIRP1-sGFP (lines #1, #2, and #3) transgenic rice plants before and after cold treatment. The amounts of leaf chlorophyll (chlorophyll a + chlorophyll b) of mock-treated (before cold) and cold-treated plants were determined 1 month after recovery from cold stress. Data are means ± SE (n ≥ 3 biological independent experiments; 30 plants were used in each assay, ∗∗P < 0.01, Student's t-test).

of cold-responsive genes in the wild-type and Ubi:OsDIRP1 sGFP plants by real-time qRT-PCR. The results showed that OsDREB1A and OsDREB1B, both of which are ABA-independent cold-responsive transcription factors (Dubouzet et al., 2003; Yamaguchi-Shinozaki and Shinozaki, 2006), were upregulated in OsDIRP1-overexpressing plants relative to the levels in wildtype plants after 24 h of cold treatment. In addition, the expression levels of ABA-independent cold stress-induced genes, such as GAD (Os03g13300; glutamate decarboxylase) and MRP4 (Os01g50100; multidrug resistance protein 4) (Su et al., 2010), were also higher in Ubi:OsDIRP1-sGFP transgenic plants than in the wild-type plants under both normal and cold stress conditions. In contrast, the transcript level of OsDREB1D was slightly lower in the OsDIRP1-overexpressors than in the wildtype plants under low temperature condition. OsDREB1D is a CRT/DRE element-binding transcription factor that is involved in the ABA-dependent abiotic stress response (Zhang et al., 2009). Unlike other DREB1 genes, DREB1D is induced by ABA, drought, and osmotic stress, but not by cold in Arabidopsis (Haake et al., 2002). These results suggested that elevation of ABA-independent cold-responsive gene expression in the OsDIRP1-overexpressors leads to increased tolerance to low temperature stress. Overall, these results indicated that, contrary to the drought and salt stress responses, OsDIRP1-overexpressing plants were more tolerant to prolonged cold stress than wild-type plants, suggesting that OsDIRP1 plays a positive role in the cold stress response in rice.

# DISCUSSION

Considerable evidence has indicated that RING-type E3 Ub ligases modulate plant responses to a broad spectrum of abiotic stress (Lyzenga and Stone, 2012; Sadanandom et al., 2012). The majority of environmental stress-responsive RING E3 Ub ligases has been characterized in the dicot model plant Arabidopsis. In this study, a putative RING E3 ligase OsDIRP1 was identified and characterized in the monocot model cereal rice (**Figure 1A**). The low basal level of OsDIRP1 expression was upregulated by drought, salt, and ABA treatments (**Figures 1B,C**). However, OsDIRP1 expression was unchanged

in response to low temperature, suggesting that OsDIRP1 is differentially regulated by different abiotic stresses. Nuclearlocalized RING E3s have been shown to up- or down-regulate the expression of stress-related genes (Dong et al., 2006; Qin et al., 2008; Catalá et al., 2011; Lim et al., 2013). Consistent with these results, OsDIRP1 was predominantly localized to the nucleus (**Figure 1D**), suggesting its role in regulating gene expression in response to abiotic stress.

Constitutive upregulation of OsDIRP1 in transgenic rice plants (Ubi:OsDIRP1-sGFP) led to a severe decrease in tolerance to drought and high salinity, as evidenced by anomalous growth performance, reduced chlorophyll content, and more rapid leaf water loss rates, which resulted in markedly lower survival as compared to wild-type plants (**Figures 2**, **3**). Furthermore, the Ubi:OsDIRP1-sGFP plants exhibited reduced sensitivities to exogenously supplied ABA during both the germination and post-germination stages (**Figure 4**). These observations suggest that OsDIRP1 participates in the negative feedback loop in the rice ABA-mediated drought and salt stress responses. In contrast, Ubi:OsDIRP1-sGFP showed increased tolerance to prolonged low temperature (4◦C) treatment, as evidenced by stable growth, higher leaf chlorophyll content, and less electrolyte leakage relative to that in wild-type plants (**Figure 5**).

Given that plants are frequently exposed to a multitude of environmental stresses in different combinations at the same time, cross-talk between the cellular responses to individual stresses should occur in plant cells (Yamaguchi-Shinozaki and Shinozaki, 2006). The opposite roles of OsDIRP1 in the drought/salt and cold responses suggest that OsDIRP1 is involved in negative cross-talk between these stress tolerance mechanisms. Similar observation was reported for the Arabidopsis RING E3 ligase AtATL78. RNAi-mediated suppression of AtATL78 conferred decreased drought tolerance and enhanced cold tolerance in Arabidopsis (Kim and Kim, 2013). In addition, transgenic rice plants that ectopically overexpressed CaPUB1, a hot pepper U-box E3 Ub ligase, were hypersensitive to drought but more tolerant to cold stress than wild-type rice plants (Min et al., 2016). Thus, it appears that the ubiquitination process conducted by a single E3 Ub ligase could confer opposite results in response to different stresses. This could be due to the fact that a single E3 ligase ubiquitinates different target proteins, depending on the cellular and physiological situations (Stone, 2018). However, the detailed mechanism underlying the opposite roles of OsDIRP1 remains to be determined.

The OsDIRP1-overexpressors displayed hypersensitivity to drought and salt stresses and hyposensitivity toward ABA (**Figures 3**–**5**), but they were tolerant to prolonged cold stress (**Figure 6**). Thus, we considered the possibility that the OsDIRP1-mediated cold response is ABA-independent. This hypothesis is supported, at least in part, by the finding that the ABA-independent cold-responsive genes (OsDREB1A and OsDREB1B), but not the ABA-dependent stress-related gene (OsDREB1D), were upregulated in OsDIRP1-overexpressing plants (**Figure 7**). Because OsDIRP1 is mainly localized to the nucleus, it would be a possible scenario that OsDIRP1 ubiquitinates a positive nuclear factor(s) in an ABA-dependent

drought/salt response and a negative nuclear factor(s) in an ABA-independent cold response, which, in turn, suppresses drought/salt-responsive genes and activates cold-associated genes, respectively. To test this hypothesis, it is essential to identify the target proteins that are ubiquitinated by OsDIRP1. Thus, we are currently attempting to isolate the target proteins of OsDIRP1.

In addition to the OsDIRP1-overexpressing plants, RNAimediated knock-down transgenic rice plants (Ubi:RNAi-OsDIRP1) were also used for the phenotypic analysis. However, under our experimental conditions, these knock-down progeny did not show detectable phenotypic changes in response to water deficit and low temperature when compared to the wild-type plants (**Supplementary Figures S3**, **S4**). These results suggest that partial suppression of OsDIRP1 failed to exert detectable effects on stress tolerance. In addition, it is likely that the phenotypes of Ubi:RNAi-OsDIRP1 plants were rescued by other E3 Ub ligase(s). A database search revealed that Os06g47280 is the closest homolog of OsDIRP1. However, OsDIRP1 and Os06g47280 share relatively low amino acid identity (51%; **Supplementary Figure S1**). In addition, Os06g47280 lacks the conserved Cys and Arg residues in the C-terminus of the RING domain. In Arabidopsis, the RING E3 ligase AtAIRP2 plays a combinatory role with two other non-homologous RING E3 ligases, AtAIRP1 and AtSDIR1, and complements the drought- and salt-sensitive phenotypes of atairp1 and atsdir1 mutants, respectively (Qin et al., 2008; Cho et al., 2011; Oh et al., 2017). A similar situation may occur in these rice stress responses; thus, it is possible that an as-yet unknown E3 ligase complements the phenotypes of Ubi:RNAi-OsDIRP1 knock-down plants.

Although OsDIRP1 possesses the C-terminal RING motif (**Figure 1A**), we were unable to detect in vitro E3 Ub ligase activity using a purified GST-OsDIRP1 protein (**Supplementary Figure S5**). This result raises two possibilities. First is that OsDIRP1 requires a post-translational modification, such as phosphorylation, for activation of its enzyme activity. In fact, OsDIRP1 is predicted to possess at least six putative phosphorylation sites (**Supplementary Figure S6**). Phosphorylation was shown to be required to activate Arabidopsis ICE1 (Inducer of CBF Expression), SINA2 (SEVER IN ABSENTIA), and RFP34/CHYR1 (RING ZINC-FINGER PROTEIN34/CHY ZINC-FINGER AND RING PROTEIN 1) E3 ligases as well as the rice OsPUB15 E3 Ub ligase (Ding Y. et al., 2015; Ding S. et al., 2015; Wang et al., 2015; Chen et al., 2018). Second, OsDIRP1 may not be a bona fide E3 ligase, but instead

#### REFERENCES


may play an unidentified role in response to environmental stimuli. It is worth noting that OsDIRP1 contains a putative betaketoacyl synthase active site in the N-terminal region (**Figure 1**). This second possibility, however, seems to be unlikely, as the RING motif in the OsDIRP1 is typical of many E3 ligases and highly homologous to that of other monocot RING E3 ligases (**Supplementary Figure S1**).

Rice plants frequently confront chilling injury at seedling or early vegetative stage when they are grown in early spring in temperate region, especially in the case of rainy day. In such growth conditions, it is a possible that rice plants need to turn on the mode of defense against cold stress and to repress the drought and salt responses. In conclusion, the data presented in this report indicate that OsDIRP1 is a negative regulator of drought and salt stress responses and a positive factor in the cold stress response in rice. These results further suggest that abiotic stress tolerance responses are subject to control by reciprocal and/or antagonistic cross-talk in the stress signaling pathways in higher plants.

#### AUTHOR CONTRIBUTIONS

LHC, HJM, MYB, and HGO performed the experiments. LHC, HJM, and WTK analyzed the data. LHC, HJM, and WTK designed the project and drafted the manuscript. WTK supervised the project and complemented the writing.

### FUNDING

This work was supported by grants from the Next-Generation BioGreen 21 Program for Agriculture and Technology Development [Project Numbers PJ01113801 (NCGC) and PJ01334901 (SSAC)] funded by the Rural Development Administration and the Basic Science Research Program (Project Number 2018R1A6A1A03025607) through the National Research Foundation (NRF) funded by the Ministry of Education, South Korea to WTK.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01797/ full#supplementary-material

ligase RSL1 targets PYL4 and PYR1 ABA receptors in plasma membrane to modulate abscisic acid signaling. Plant J. 80, 1057–1071. doi: 10.1111/tpj.12708




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

# The Rice SPOTTED LEAF4 (SPL4) Encodes a Plant Spastin That Inhibits ROS Accumulation in Leaf Development and Functions in Leaf Senescence

#### Edited by:

Giha Song<sup>1</sup>†

Hee-Jong Koh<sup>1</sup>

Dae-Jin Yun, Konkuk University, South Korea

#### Reviewed by:

Woo Taek Kim, Yonsei University, South Korea Zhiguo Zhang, Biotechnology Research Institute (CAAS), China

#### \*Correspondence:

Kiyoon Kang kykang7408@snu.ac.kr Nam-Chon Paek ncpaek@snu.ac.kr †These authors have contributed

equally to this work

#### ‡Present address:

Choon-Tak Kwon, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, United States

#### Specialty section:

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

Received: 30 October 2018 Accepted: 11 December 2018 Published: 07 January 2019

#### Citation:

Song G, Kwon C-T, Kim S-H, Shim Y, Lim C, Koh H-J, An G, Kang K and Paek N-C (2019) The Rice SPOTTED LEAF4 (SPL4) Encodes a Plant Spastin That Inhibits ROS Accumulation in Leaf Development and Functions in Leaf Senescence. Front. Plant Sci. 9:1925. doi: 10.3389/fpls.2018.01925 <sup>1</sup> Department of Plant Science, Plant Genomics and Breeding Institute, Research Institute of Agriculture and Life Sciences, Seoul National University, Seoul, South Korea, <sup>2</sup> Department of Plant Molecular Systems Biotechnology, Crop Biotech Institute, Kyung Hee University, Seoul, South Korea

, Kiyoon Kang<sup>1</sup>

, Yejin Shim<sup>1</sup>

\* and Nam-Chon Paek<sup>1</sup>

, Chaemyeong Lim<sup>1</sup>

\*

,

, Choon-Tak Kwon<sup>1</sup>†‡, Suk-Hwan Kim<sup>1</sup>

, Gynheung An<sup>2</sup>

Lesion mimic mutants (LMMs) are usually controlled by single recessive mutations that cause the formation of necrotic lesions without pathogen invasion. These genetic defects are useful to reveal the regulatory mechanisms of defense-related programmed cell death in plants. Molecular evidence has been suggested that some of LMMs are closely associated with the regulation of leaf senescence in rice (Oryza sativa). Here, we characterized the mutation underlying spotted leaf4 (spl4), which results in lesion formation and also affects leaf senescence in rice. Map-based cloning revealed that the γ ray-induced spl4-1 mutant has a single base substitution in the splicing site of the SPL4 locus, resulting in a 13-bp deletion within the encoded microtubule-interactingand-transport (MIT) spastin protein containing an AAA-type ATPase domain. The T-DNA insertion spl4-2 mutant exhibited spontaneous lesions similar to those of the spl4- 1 mutant, confirming that SPL4 is responsible for the LMM phenotype. In addition, both spl4 mutants exhibited delayed leaf yellowing during dark-induced or natural senescence. Western blot analysis of spl4 mutant leaves suggested possible roles for SPL4 in the degradation of photosynthetic proteins. Punctate signals of SPL4-fused fluorescent proteins were detected in the cytoplasm, similar to the cellular localization of animal spastin. Based on these findings, we propose that SPL4 is a plant spastin that is involved in multiple aspects of leaf development, including senescence.

Keywords: spastin, microtubule severing protein, lesion mimic mutant, rice (Oryza sativa), senescence, reactive oxygen species

# INTRODUCTION

Among the defense mechanisms activated in response to pathogen attacks in plants, the hypersensitive response (HR), which induces rapid death of infected cells, prevents the spread of pathogens to adjacent cells (Morel and Dangl, 1997; Takahashi et al., 2009). Lesion mimic mutants (LMMs) exhibit spontaneous cell death in the absence of pathogen attacks and have been

**49**

isolated from plant species including barley (Hordeum vulgare) (Wolter et al., 1993), maize (Zea mays) (Hoisington et al., 1982), tomato (Solanum lycopersicum) (Badel et al., 2006), Arabidopsis (Arabidopsis thaliana) (Dietrich et al., 1994), and rice (Oryza sativa) (Kim et al., 2009). The autonomous lesions in LMMs tend to be accompanied by excessive levels of reactive oxygen species (ROS), which lead to accelerated cell death (Van Breusegem and Dat, 2006).

Based on the lesion phenotype of LMMs, some of the underlying genes have been cloned and functionally characterized. A spotted leaf gene Spl7 encodes a heat shock protein and its mutation is responsible to lesion formation in the rice leaves (Yamanouchi et al., 2002). Mutation of SPL5 encoding a putative splicing factor 3b subunit 3 (SF3b3) continuously developed small reddish-brown necrotic lesions on the rice leaves (Chen et al., 2012). The probenazole-induced protein (PBZ1) of which expression is ectopically induced in spl1 mutant is localized in theseed aleurone layer and associated with programmed cell death (PCD) (Kim et al., 2008). Impairment of COPROPORPHYRINOGEN III OXIDASE (CPOX) in the rice lesion initiation 1 (rlin1) mutant causes the formation of necrotic lesions in rice leaves and stems due to excessive ROS accumulation (Sun et al., 2011). SPL11 encodes a U-box/armadillo repeat protein conferring E3 ubiquitin ligase activity and the spl11 mutant displays a spontaneous cell death phenotype and enhanced resistance to fungal and bacterial diseases in rice (Zeng et al., 2004).

Recent studies have reported that a few of LMMs are associated with the regulatory pathways of leaf senescence. SPOTTED LEAF3 (SPL3) encodes MITOGEN-ACTIVATED PROTEIN KINASE KINASE KINASE1 (MAPKKK1), and the spl3 mutant causes to cell death due to lack of ROS scavenging activity (Wang et al., 2015). This mutant also exhibits delayed abscisic acid-mediated leaf senescence. On the contrary, the mutation of SPL28, encoding CLATHRIN-ASSOCIATED ADAPTOR PROTEIN COMPLEX 1 MEDIUM SUBUNIT µ1 (AP1M1), promotes leaf yellowing during senescence (Qiao et al., 2010). Most recently, Wang et al. (2017) have found that SPL33 encodes a eukaryotic translation elongation factor 1 alpha (eEF1A)-like protein. The spl33 mutant exhibits both phenotype of PCD-mediated cell death and early senescence. Mutation of spotted leaf sheath (sles) encoding a putative expressed protein containing kinase domain exhibits precocious senescence (Lee et al., 2018).

Notably, among more than 40 LMMs which have been reported in rice (Wu et al., 2008), mutation of the LESION MIMIC RESEMBLING (LMR) locus encoding a microtubule interacting and transport (MIT) protein causes the LMM phenotype along with excess ROS accumulation (Fekih et al., 2015). Microtubules (MTs) are dynamic cytoskeletal polymers that play essential roles in cell division (de Keijzer et al., 2014), morphogenesis (Mathur and Hülskamp, 2002), and cell migration (Villari et al., 2015). In the plant cell, MTs coordinate the deposition of cellulose microfibrils in the cell wall, which affects growth and development in plants. They also play key roles in the responses to hormones, pathogens, and environmental stresses (Buschmann and Lloyd, 2008).

MT arrays are reorganized according to the need of cells in response to internal cues and external stimuli (Shibaoka, 1994). This plasticity includes MT growth, stabilization, destabilization, and interaction with other cellular organelles (Wade, 2009), and is regulated by interactions with MT-associated proteins (MAPs), and MT-severing proteins (Gouveia and Akhmanova, 2010; Díaz-Valencia et al., 2011). Three classes of MT-severing proteins have been identified; katanin (McNally and Vale, 1993), spastin (Hazan et al., 1999), and fidgetin (Cox et al., 2000). They all have a highly conserved C-terminal AAA domain that contributes to the formation of hexameric rings or dodecameric stacked rings to interact with MTs (Neuwald et al., 1999; Vale, 2000; Lupas and Martin, 2002).

In humans, hereditary spastic paraplegia (HSP) is a devastating neurodegenerative disorder characterized by a progressive spasticity and lower limb weakness (Crosby and Proukakis, 2002). However, approximately 40% of HSP diseases are caused by the malfunction of the spastin protein, SPASTIC PARAPLEGIA 4 (SPG4) (Hazan et al., 1999; Fonknechten et al., 2000). Unlike the many studies of animal spastin implicated in this severe hereditary disease, little is known about the functions of plant spastins, even though putative spastin proteins are conserved in plants (Fekih et al., 2015).

Here, we characterized the rice spl4-1 LMM, which shows autonomous lesions accompanied by ROS accumulation in the leaf blades. By map-based cloning, we found that the SPL4 locus encodes a MT-interacting-and-transport spastin protein containing an AAA-type ATPase domain. Mutation of SPL4 resulted in delayed senescence: photosynthetic proteins remained abundant in the detached leaves under dark-induced senescence conditions. Confocal microscopic observation implied that SPL4 is localized to the cytoplasm. Collectively, our results point to a novel function for plant spastin in the inhibition of lesion formation in the leaves during vegetative growth and promotion of leaf yellowing during senescence.

# MATERIALS AND METHODS

#### Plant Materials, Growth Conditions, and Dark Treatment

The spl4-1 mutant was previously isolated from a mutant pool produced by γ-ray irradiation to the Japanese japonica rice (Oryza sativa) cultivar "Norin 8" (Iwata et al., 1978). The T-DNA insertional knockout mutant of SPL4 (LOC\_Os06g03940; PFG\_3A-16679, designated as spl4-2) was derived from the Korean japonica rice cultivar "Dongjin"<sup>1</sup> (Jeon et al., 2000; Jeong et al., 2006). Rice plants were cultivated in the paddy field under natural long day (NLD) conditions ( ≥ 14 h light/day, 37◦N latitude, Suwon, Korea). For the dark treatment, the detached leaves of rice plants grown in the paddy field under NLD conditions were incubated on 3 mM MES (pH 5.8) buffer with the abaxial side up at 28◦C in complete darkness. The leaf disks were sampled at the specified DDI for each experiment.

<sup>1</sup>http://signal.salk.edu/cgi-bin/RiceGE

scale bars = 6 cm. (B,D,F) Representative leaves of N8 and spl4-1 plants. The Roman numerals I, II, III, and IV represent the first, second, third, and fourth leaves from the top of the plants, respectively. Black triangles and boxes in (B) indicate the location and enlargement of autonomous lesions. Black scale bars = 2 cm. (G) The plant height was measured at 162 DAS. (H) Total chlorophyll concentration was determined from the first leaves (I) of N8 and spl4-1 plants shown in (F). Mean and standard deviations were obtained from more than three biological replicates. Asterisks indicate a statistically significant difference between N8 and spl4-1 plants according to Student's t-test (∗∗∗P < 0.001). These experiments were repeated twice with similar results.

#### Detection of Reactive Oxygen Species (ROS)

Hydrogen peroxide (H2O2) in the rice leaves was detected using 3,3'-diaminobenzidine (DAB) as previously described (Kim et al., 2018). The rice leaves grown in the paddy field under NLD conditions were sampled at 60, 70, 80, and 125 days after sowing (DAS) and incubated in 0.1% (w/v) DAB (Sigma) solution for 12 h at 28◦C with gentle shaking (40 rpm). Chlorophyll was then completely removed by incubation in 90% ethanol at 80◦C. H2O<sup>2</sup> was visualized as reddish-brown stains. Detection of singlet oxygen (1O2) was conducted as previously described with some modifications (Kwon et al., 2017). The rice leaves grown in the paddy field for 133 DAS were vacuum infiltrated with 10 mM sodium phosphate buffer (pH 7.5) containing 50 µM Singlet Oxygen Sensor Green reagent (SOSG, Invitrogen). After incubation for 30 min in the dark, the fluorescence emission of SOSG was detected by a laser scanning confocal microscope (Carl Zeiss LSM510). The excitation and emission wavelengths were 480 and 520 nm, respectively. The red autofluorescence emission from chlorophyll was detected following excitation at 543 nm.

#### Map-Based Cloning

The SPL4 locus was previously mapped to the short arm of chromosome 6 (Iwata et al., 1978). In this study, a mapping population of 798 F<sup>2</sup> individuals was generated by crossing the japonica-type spl4-1 mutant and the tongil-type Milyang23 (an indica/japonica hybrid cultivar, M23). To determine the chromosomal localization of the SPL4 locus, we initially performed a small-scale mapping using 100 spl4 homozygous F<sup>2</sup> plants, 11 simple sequence repeat (SSR) markers, and 5 sequence-tagged site (STS) markers distributed on chromosome 6 (**Supplementary Table S1**). The SSR marker information is available in GRAMENE<sup>2</sup> . For fine mapping, 8 additional STS markers were designed by comparing the genomic DNA sequences of the spl4-1 mutant with those of the M23 cultivar (**Supplementary Table S1**). Using these 8 STS markers and the 198 F<sup>3</sup> individuals with the spontaneous lesion phenotype, the SPL4 locus was fine-mapped to the 77-kb region between the STS8, and STS9 markers on chromosome 6.

#### Chlorophyll Quantification

To measure the total chlorophyll levels, pigments were extracted from equal fresh weights of leaves with 80% ice-cold acetone. The concentration of total chlorophyll was determined using a UV/VIS spectrophotometer (BioTek) and calculated as previously described (Lichtenthaler, 1987).

<sup>2</sup>http://www.gramene.org

grown in the field for 160 d after sowing. The picture (A) and the length (B) of panicles and internodes. Mean and standard deviations of all agronomic traits were obtained from twenty plants. Asterisks indicate statistically significant differences between N8 and spl4-1 plants according to Student's t-test ( ∗∗P < 0.01, ∗∗∗P < 0.001). The plant photos shown are representative of three independent observations.

# Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE) and Immunoblot Analysis

Total proteins were extracted from the detached leaves of 2 month-old plants that were incubated in complete darkness. Leaf tissue (10 mg) was homogenized with 100 µl of SDS sample buffer [50 mM Tris, pH 6.8, 2 mM EDTA, 10% (w/v) glycerol, 2% SDS, and 6% 2-mercaptoethanol]. Then, 4 µl of each protein extract was subjected to 12% SDS (w/v) PAGE, and the resolved proteins were electroblotted onto an Immobilon-P Transfer Membrane (Millipore). Antibodies against photosynthetic proteins (Lhca2, Lhca3, Lhcb2, Lhcb6, PsaA, and PsbD) and the large subunit of rubisco (RbcL) (Agrisera) were used for immunoblot analysis. Horseradish peroxidase activity of secondary antibodies (Sigma) was detected using the ECL system (WESTSAVE, AbFRONTIER) according to the manufacturer's instructions.

# Reverse Transcription and Quantitative PCR (RT-qPCR) Analysis

Total RNA was extracted from leaves using a Total RNA Extraction Kit(MGmed) according to the manufacturer's protocols. First-strand cDNA was synthesized from 2 µg of total RNA in a 100 µl volume using oligo(dT)<sup>15</sup> primers and M-MLV reverse transcriptase (Promega). The transcript levels of SPL4 were detected by qPCR using SPL4-specific primers (**Supplementary Table S1**). Rice UBIQUITIN 5 (OsUBQ5) (AK061988) was used as an internal control for normalization (**Supplementary Table S1**; Jain et al., 2006). The 20– µl total reaction volume included 2 µl of cDNA mixture, 2 µl of 0.5 µM primer, and 10 µl of 2X GoTaq qPCR Master Mix (Promega). PCR was performed with a LightCycler 480 (Roche) using the following conditions: 95◦C for 2 min followed by 45 cycles of 95◦C for 10 s and 60◦C for 1 min.

# Plasmid Construction

The full-length cDNA of SPL4 was amplified by PCR using genespecific primers (**Supplementary Table S1**), and inserted into the pCR8/GW/TOPO vector (Invitrogen). Then, the SPL4 cDNA was transferred into the pMDC43 and pEarleyGate 104 (pEG104) gateway binary vectors using Gateway LR Clonase II Enzyme Mix (Invitrogen), resulting in 35S::GFP-SPL4, and 35S::YFP-SPL4 constructs, respectively.

# Subcellular Localization of SPL4

Rice protoplast isolation was carried out as previously described with some modifications (Yoo et al., 2007). The leaf sheaths of 10-day-old etiolated seedlings of the japonica-type rice cultivar "Dongjin" were chopped and transferred into a digestion solution [0.5 M mannitol, 10 mM MES (pH 5.7), 1.5% (w/v) Cellulase ONOZUKA R-10 (Yakult, Japan), 0.75% (w/v) Macerozyme R-10 (Yakult, Japan), 0.1% (w/v) BSA, 10 mM CaCl2, and 5 mM 2-mercaptoethanol]. After vacuum infiltration for 10 min, the tissues were digested for 4.5 h with gentle shaking (40 rpm) at 28◦C. Following the enzymatic digestion, the protoplasts were released with W5 solution [154 mM NaCl, 125 mM CaCl2, 5 mM KCl, and 2 mM MES (pH 5.8)]. Then, the protoplasts were

adjusted to 10<sup>7</sup> to 10<sup>8</sup> cells per 1 ml of MMG solution [0.5 M mannitol, 15 mM MgCl2, and 4 mM MES (pH 5.7)] using a hemocytometer. The 50 µl of protoplasts were incubated with 110 µl of PEG solution [0.2 M mannitol, 100 mM CaCl2, and 40% (w/v) PEG 4000 (Fluka)] containing 15 µg of plasmids (35S::GFP-SPL4 or 35S::GFP) for 15 min in the dark at 28◦C. Then, the protoplasts were washed twice with W5 solution and resuspended in 1.5 ml of an incubation solution [0.5 M mannitol, 20 mM KCl, and 4 mM MES (pH 5.7)]. After incubation for 12 h in the dark, the protoplasts were examined by a confocal laser scanning microscope (Carl Zeiss LSM710).

To examine the subcellular localization of SPL4 in onion (Allium cepa) epidermal cells, the plasmids (35S::YFP-SPL4 or 35S::YFP) were transiently expressed in the onion epidermal cell layers using a DNA Particle Delivery System (Biolistic PDS-1000/He, Bio-Rad). After incubation on a phytoagar plate containing Murashige and Skoog medium for 18 h in the dark at 28◦C, the onion cells were observed with a confocal laser scanning microscope (Carl Zeiss LSM710).

#### RESULTS

#### Characterization of the spl4-1 Mutant in Rice

The spl4-1 mutant was first isolated from the γ ray-treated lines of Norin 8 (Oryza sativa ssp. japonica, hereafter referred to as N8). When N8 and spl4-1 plants were grown in the paddy field (Suwon, South Korea, 37◦N latitude), autonomous lesions began to appear in the second leaves of the spl4-1 mutant at the maximum tillering stage (87 days after sowing, DAS) (**Figures 1A,B**), and then expanded throughout the entire leaf at the heading stage (118 DAS), except for the flag leaves (**Figures 1C,D**). In addition, the height of the spl4-1 mutant was shorter than that of the N8 plant at the end of the grain filling stage (162 DAS) (**Figures 1E,G**). To investigate this height difference in more detail, we further measured the length of each internode of N8 and spl4-1 plants. All the internodes of the spl4-1 mutant were shorter than those of the N8 plant. The first internode of the spl4- 1 mutant was significantly shorter compared with the N8 plant (**Figure 2**). A number of spl4-1 leaves rolled toward the abaxial side (**Figures 3A,B**), and the width of spl4- 1 leaves was narrower than that of N8 leaves at 56 DAS (**Figure 3C**). However, the narrow phenotype of spl4-1 leaves disappeared and eventually became similar to the width of N8 leaves at 74 DAS (**Figures 3D–F**). Finally, we found that the greenness of spl4-1 leaves persisted much longer than that of N8 leaves around the grain harvest stage (162 DAS) (**Figure 1E**). Consistent with this phenotype, the total chlorophyll contents of spl4-1 leaves were higher than those of N8 leaves (**Figure 1H**). These results indicated that SPL4 functions in multiple aspects of plant development including leaf senescence.

FIGURE 4 | ROS accumulation in the spl4-1 mutant. (A) Visualization of singlet oxygen (1O2) detected by the SOSG fluorescent probe. The leaves of wild-type Norin 8 (N8) and the spl4-1 mutant were sampled at 133 d after sowing (DAS). The fluorescence of SOSG is green and chlorophyll (Chl) auto-fluorescence is red. Scale bars =5 µm. (B) DAB staining for hydrogen peroxide (H2O2) (dark brown). The leaves of N8 and spl4-1 plants were obtained at 60, 70, 80, and 125 DAS. NT, not treated. These experiments were repeated twice with similar results.

#### Reactive Oxygen Species (ROS) Accumulate in the spl4-1 Leaves

Accumulation of ROS has been observed in the lesions of several LMMs (Chen et al., 2012; Wang et al., 2015). To investigate whether the lesions of spl4-1 leaves is due to excessive ROS accumulation, singlet oxygen (1O2), and hydrogen peroxide (H2O2) were examined using singlet oxygen sensor green reagent (SOSG), and 3,3'-diaminobenzidine (DAB), respectively. The observation of green fluorescence indicated that <sup>1</sup>O<sup>2</sup> was highly accumulated in the flag leaves of 3-month-old spl4-1 mutants, but not in those of N8 plants (**Figure 4A**). Little H2O<sup>2</sup> was found in the leaf blades of the spl4-1 mutant before the onset of lesion formation at 60 DAS. However, the red-brown precipitates accumulated when the lesions spread throughout the leaf blades of the spl4-1 mutant at 70, 80, and 125 DAS (**Figure 4B**). This result suggests that, similar to other LMMs, the formation of autonomous lesions in spl4-1 leaves is closely associated with accumulation of excess ROS.

#### Map-Based Cloning of the SPL4 Locus

The spl4-1 mutation is a single recessive allele whose locus was previously mapped to an interval of 9.6 cM on chromosome 6 (Iwata et al., 1978). Mapping of the 100 F<sup>2</sup> individuals exhibiting spontaneous lesions derived from a cross between the spl4-1 mutant and Milyang23 (a Tongil-type indica/japonica hybrid cultivar) initially delineated the SPL4 locus to a 1.7-Mb region on chromosome 6 between the STS1 and RM587 markers (**Figure 5A**). Using 198 F<sup>3</sup> individuals with the spontaneous lesion phenotype, the SPL4 locus was fine-mapped to a 77-kb region flanked by the STS8 and STS9 markers (**Figure 5B**). Thirteen putative open reading frames were predicted within the candidate region according to the Rice Annotation Project Database<sup>3</sup> (**Figure 5C**). To find single nucleotide polymorphisms (SNPs), we sequenced the genomic DNA extracted from spl4-1 leaves using a whole-genome resequencing approach. Comparison of the sequence of the candidate region between Nipponbare (Oryza sativa ssp. japonica) and the spl4-1 mutant revealed that a single nucleotide substitution (G to C) at the end of the 1st intron resulted in aberrant splicing of the intron of Os06g03940 (**Figure 5D**).

When comparing the SPL4 cDNA sequences between N8 and spl4-1 plants, we found that spl4-1 mRNA had a 13-bp deletion due to a change in the alternative splicing acceptor (**Figure 5E**). SPL4 comprised 13 exons that encoded a 487-amino acid (aa) protein including a MIT\_spastin domain, and subsequently an AAA-type ATPase domain (**Figure 5F**). To confirm that the spl4- 1 mutant phenotype is caused by loss of function of SPL4, we obtained a T-DNA insertion mutant (hereafter referred to as spl4- 2) that contains the T-DNA insertion in exon 13 of Os06g03940 (**Figure 6A**). Reverse transcription quantitative PCR (RT-qPCR) analysis showed that the spl4-2 mutant lacked SPL4 transcript unlike its parental rice cultivar "Dongjin" (Korean japonica rice cultivar, hereafter referred to as DJ), while the spl4 transcript in the spl4-1 mutant with the 13-bp deletion was transcribed as much as the SPL4 transcripts in the N8 plants (**Figure 6B**). Consistent with the phenotypic observation of the spl4-1 mutant, the leaf blades of the spl4-2 mutant contained spontaneous lesions (**Figure 6C**). Taken together, these results indicated that a null mutation of the SPL4 locus results in the LMM phenotype.

#### Loss of Function of SPL4 Results in Delayed Leaf Yellowing Under Dark-Induced Senescence Conditions

The observation of more green pigments in the leaves of the spl4-1 mutant than in the leaves of N8 plants at 162 DAS

<sup>3</sup>http://rice.plantbiology.msu.edu

#### FIGURE 5 | Continued

fpls-09-01925 December 24, 2018 Time: 10:25 # 8

the STS8 and STS9 markers. Numbers below the line indicate the number of F<sup>3</sup> recombinants at the marker regions. (C) Candidate genes in the 77-kb region. (D) The G to C substitution in SPL4 in the spl4-1 mutant. Black and white bars represent the exon and untranslated region, respectively. The black line represents the intron. The black arrow indicates the G to C substitution position in the spl4-1 mutant. (E) SPL4 nucleotide sequence. The black shading represents the SPL4 mRNA. Red arrows and the red character indicate the splicing sites and the G to C substitution, respectively. (F) Domain structure of SPL4 containing the MIT\_spastin and AAA-type ATPase domains. Numbers indicate the amino acid position of SPL4.

(**Figure 1H**) suggested that SPL4 plays an important role in the regulation of leaf senescence. SPL4 mRNA levels were dramatically upregulated in the fully senescing flag leaves of DJ plants at 162 DAS (**Supplementary Figure S1**). To analyze the leaf senescence phenotype more effectively, we used detached leaves and artificially induced senescence by a dark treatment at 28◦C (Kim et al., 2006; Kong et al., 2006). When the green leaves detached from 4-week-old DJ plants turned yellow, at 4 days after dark incubation (DDI; **Figure 7A**), the expression levels of SPL4 also increased rapidly (**Figure 7B**).

To explore the potential role of SPL4 in leaf senescence, the leaves of 4-week-old spl4-1 and spl4-2 mutants and the parental cultivars N8 and DJ, respectively, were subjected to dark incubation. The green color of the leaves of both spl4 mutants was retained much longer than that of the parental cultivars at 4 DDI (**Figure 7C**). Consistent with this observation, the total chlorophyll contents of both spl4 mutants were significantly higher than those of the parental cultivars at 4 DDI (**Figure 7D**). To compare the levels of photosynthetic proteins between the plants at 0 and 4 DDI, we performed western blotting with antibodies against core proteins (PsaA and PsbD) of photosystem I (PSI) and PSII, and antenna proteins of the light-harvesting complexes (Lhca2, Lhca3, Lhcb2, and Lhcb6). The levels of the large subunit of Rubisco (RbcL), which is involved in carbon fixation, were also examined. While there was no remarkable difference in the levels of photosynthetic proteins between the parental cultivars and the spl4 mutants at 0 DDI, the levels of all proteins were much higher in the spl4 mutants than those in the parental cultivars at 4 DDI (**Figure 7E**). These results indicated that the mutation of SPL4 results in delayed leaf yellowing under both natural and dark-induced senescence conditions.

#### Cytoplasmic Localization of SPL4

To examine the subcellular localization of SPL4, we generated a transgenic construct in which the coding region of SPL4 was fused in-frame with the 3<sup>0</sup> -terminus of the green fluorescent protein (GFP) reporter gene and performed a transient expression assay using rice protoplasts. The GFP signal from the 35S::GFP-SPL4 construct was observed in the cytoplasm, whereas GFP alone showed ubiquitous distribution throughout the cell (**Figure 8A**). To definitively identify the localization of SPL4, we generated a construct expressing a yellow fluorescent protein (YFP)-SPL4 recombinant protein and introduced this construct into onion (Allium cepa) epidermal cells using a bombardmentmediated transformation method. The signal of YFP-SPL4 was punctate in the cytoplasm on the periphery of the nucleus, whereas YFP alone was observed throughout the cell (**Figure 8B**).

To observe the speckles in more detail, we zoomed in on the DAPI-stained nucleus where the punctate structures overlapped. The enlargement indicated that the localization of YFP-SPL4 is similar to that of animal spastin spots in the cytoplasm, as previously reported (**Figure 8C**; Claudiani et al., 2005).

FIGURE 7 | The spl4 mutants exhibit delayed leaf yellowing during dark-induced senescence. (A,B) Detached leaves of 2-month-old wild-type Dongjin (DJ) grown in the field were incubated on 3 mM MES (pH 5.8) buffer at 28◦C with the abaxial side up in complete darkness. The leaf yellowing phenotype (A) and expression profile of SPL4 (B) were determined at 0, 2, and 4 days after dark incubation (DDI). The transcript levels of SPL4 were determined by RT-qPCR analysis and normalized to that of OsUBQ5. Different letters indicate significant differences according to one-way ANOVA and Duncan's least significant range test (P < 0.05). (C–E) Detached leaves of Norin 8 (N8), DJ, and the spl4 mutants (spl4-1 and spl4-2) grown in the field for 2 months were subjected to dark conditions as shown in (A). The leaf yellowing phenotype (C) and total chlorophyll contents (D) were observed at 0 and 4 DDI. (E) An immunoblot assay was performed using antibodies against photosynthetic proteins (Lhca2, Lhca3, Lhcb2, Lhcb6, PsaA, and PsbD) and the large subunit of rubisco (RbcL). Mean and standard deviations were obtained from more than three biological replicates. Different letters indicate significant differences according to one-way ANOVA and Duncan's least significant range test (P < 0.05). Asterisks indicate statistically significant differences between the wild-type plants and spl4 mutants according to Student's t-test (∗∗P < 0.01, ∗∗∗P < 0.001). These experiments were repeated twice with similar results. Chl, chlorophyll; FW, fresh weight.

Proteins that translocate to the cytoplasm from the nucleus have a conserved nuclear export signal (NES) sequence, 8-x(2,3)- 8-x(2,3)-8-x-8, where 8 is L, I, F, V, or M, and x is any amino acid (Bogerd et al., 1996; la Cour et al., 2003). Human spastin has three putative NES sequences that contribute to its subcellular localization in the cytoplasm (Claudiani et al., 2005). Based on the comparison of conserved plant spastin sequences in several plant species (Fekih et al., 2015), we found that a putative NES consensus, 8-x2-8-x3-8-x2-8, was highly conserved in the N-terminal region of SPL4 between amino acids 4 and 14 (**Figure 8D**; Kosugi et al., 2008). Thus, it is possible that SPL4 encodes a spastin protein localized in the cytoplasm.

# Defective Grain Yield of the spl4-1 Mutant

To examine whether the spl4-1 mutation affects grain production, we evaluated the yield components, including 500-grain weight, grain yield per plant, panicle length, panicles per plant, spikelets per main panicle, and the seed setting rate, in N8 and spl4-1 plants grown in the field. Interestingly, among the yield components, the spl4-1 mutants showed a 12% increase in the 500-grain weight compared with N8 plants (**Figure 9A**). However, the spl4- 1 mutants had a relatively low seed setting rate and number of panicles per plant compared with N8 plants, resulting in lower grain yield per plant (**Figures 9B–E**). Moreover, the shorter length of the main panicle in the spl4-1 mutant caused a decrease of the number of spikelets per main panicle compared to N8 plants (**Figures 9F,G**). Thus, the increase of 500-grain weight did not lead to an improvement of grain yield in spl4-1 mutants.

# DISCUSSION

The mutation of the SPL4 locus has been previously reported as lesion mimic resembling (lmr) and lesion resembling disease 6-6 (lrd6-6) in rice (Fekih et al., 2015; Zhu et al., 2016). These mutants, lacking the AAA-type ATPase activity, exhibited an enhanced resistant phenotype against the rice blast fungus (Magnaporthe oryzae) and bacterial blight (Xanthomonas oryzae pv. oryzae) by inducing pathogenesis-related genes and promoting the biosynthesis of antimicrobial compounds. The studies of most LMMs have been focused on defense mechanisms against pathogen attacks due to the phenotypic similarity between lesion mimicry and the HR (Qiao et al., 2010; Feng et al., 2013; Li et al., 2014). In this study, we characterized the spl4-1 mutant and report a novel function of SPL4 in the regulation of leaf senescence. Consistent with most LMMs isolated from various plant species (Hoisington et al., 1982; Wolter et al., 1993; Dietrich et al., 1994; Badel et al., 2006; Kim et al., 2009), the spl4-1 mutants exhibit autonomous lesions along with excessive ROS accumulation (**Figures 1**, **4**).

Based on map-based cloning and a domain search in the NCBI database<sup>4</sup> , the conserved amino acid sequences of SPL4 were

<sup>4</sup>https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi

predicted to include the MIT\_spastin, and AAA-type ATPase domains (**Figure 5F**). Notably, the rice genome does not contain any other gene that is homologous to the MIT\_spastin domain of SPL4. Spastin is well-studied in animal cells because of its important role in HSP disease (Crosby and Proukakis, 2002). Human spastin, encoded by SPG4, localizes in cytoplasmic spots on the periphery of the nucleus (Claudiani et al., 2005). The localization of SPL4 in plant cells was similar to that of SPG4 in animals. Moreover, when YFP-SPL4 was transiently expressed in onion epidermal cells, fluorescent signals were detected in the cytoplasm around the nucleus (**Figures 8B,C**). These observations indicate that SPL4 encodes a putative plant spastin, possibly functioning in inhibition of lesion formation during leaf development.

The NES is the essential conserved amino acid residues that allow the protein to export from the nucleus to the cytoplasm. There are three putative NES sequences in human spastin (Claudiani et al., 2005). Chromosome region maintenance 1 (CRM1), a member of the importin β superfamily, recruits the CRM1-NES cargo-RanGTP complex, and subsequently recognizes the NES consensus sequence of human spastin to facilitate the transportation (Ossareh-Nazari et al., 2001; Fornerod and Ohno, 2002; Claudiani et al., 2005; Hutten and Kehlenbach, 2007). In Arabidopsis, EXPORTIN 1 (AtXPO1/AtCRM1) forms the export complex with a plant cofactor, RAN1. This complex interacts with NES residues that are derived from animal proteins and transports the target proteins from the nucleus to the cytoplasm (Haasen et al., 1999). NES consensus sequences are typically composed of hydrophobic conserved residues separated by a variable number of amino acids, given by 8-x(2,3)-8-x(2,3)-8-x-8. However, Kosugi et al. (2008) defined a new NES consensus sequence through screening of a Mtys1 knockout yeast strain that can be rescued by tyrosyltRNA synthetase 1 (TYS1), which can anchor to random NES sequences (Kosugi et al., 2008). Among the six NES consensus patterns identified from Kosugu's selection system, two NES sequences are inconsistent with the traditional rule; one is 8-x-8-x2-8-x-8 and the other is 8-x2-8-x3-8-x2-8 (Kosugi et al., 2008). These diverse NESs may reflect the wide spectrum of binding specificity of CRM1. Interestingly, both monocots and dicots have highly conserved spastin proteins (Fekih et al., 2015). Through our attempt to find the NES sequences in plant spastin proteins, we defined the plant-specific NES sequence as 8-x2-8-x3-8-x2-8 (**Figure 8D**), which differs from that of human spastin. Thus, our results suggest a putative plant-specific NES sequence that may contribute to the cytoplasmic localization of SPL4.

Leaf yellowing is a visual marker for estimating the degree of leaf senescence (Lim et al., 2007). The spl4-1 mutant exhibited delayed leaf yellowing compared to its parental japonica rice cultivar (Norin 8) under both natural and darkinduced senescence conditions (**Figures 1F,H**, **7C,D**). Based

on these observations, we hypothesized that alteration of MT arrays, which is regulated by MT-associated proteins (MAPs) and MT-severing proteins including spastin (Hazan et al., 1999), katanin (McNally and Vale, 1993) and fidgetin (Cox et al., 2000), is a key factor for the regulation of leaf senescence. Keech et al., 2010 showed that the MT network in epidermal and mesophyll cells of Arabidopsis leaves are degraded during natural and dark-induced leaf senescence. This MT destabilization is closely connected with the expression of genes that encode the MAP proteins. While the genes encoding the MT-stabilizing proteins (MAP65 family and MAP70-1) were repressed during natural and dark-induced leaf senescence, transcripts of MAP18, encoding a MT-destabilizing protein, are strongly upregulated in senescing Arabidopsis leaves (Keech et al., 2010). Thus, MT plasticity is closely linked to the stability of epidermal and mesophyll cells that are involved in photosynthesis.

Although expression of SPL4 significantly increased in wild-type leaves in the dark-induced senescence treatment (**Supplementary Figure S1** and **Figures 7A,B**), the underlying regulatory mechanisms responsible for leaf senescence remain unknown. The expression of katanin p60 subunit genes, encoding proteins of another MT-severing family, is significantly upregulated under dark-induced senescence (Keech et al., 2010). Arabidopsis plants overexpressing the katanin p60 subunit (AtKSS) form numerous bundles of MTs, resulting from the severing of MTs by AtKSS, and then the MTs are ultimately depolymerized (Stoppin-Mellet et al., 2006). The bot1-1 mutant, with a mutation in BOT1 encoding katanin, survives much longer than wild-type plants (Bichet et al., 2001). Here, our analysis showed that photosynthetic proteins remained much more abundant in the spl4 mutants than in the parental cultivars at 4 DDI (**Figure 7E**). By analogy to the effects of the MT-severing protein katanin, we hypothesize that senescence-induced spastin

severs the MTs of rice epidermal and mesophyll cells, followed by promoting the degradation of photosynthetic proteins.

Although ROS accumulation generally accelerates leaf senescence (Jajic et al., 2015; Lin et al., 2016), our results showed that excessive accumulation of ROS in spl4-1 mutant does not link to the promotion of leaf yellowing during dark incubation (**Figures 4**, **7**). Recently, Wang et al. (2015) reports that the LMM phenotype by mutation of SPOTTED LEAF3 (SPL3) is due to excessive ROS. However, ABA hyposensitivity of spl3 mutant leads to delaying leaf yellowing during both natural and darkinduced senescence. Since leaf senescence is a complex process involving numerous regulators (Guo et al., 2004; Buchanan-Wollaston et al., 2005), plants may have alternative senescence pathways that are mediated by spastin independent of ROS signaling.

Finally, mutation of SPL4 affects plant morphology, including short internodes (**Figure 2**). In maize, internodal cells originate from a portion of undifferentiated cells which have randomly arranged cortical MTs (Nemoto et al., 2004). These random cortical MTs are transversely reoriented in incipient internode, followed by rearranging to longitudinal direction during internode elongation. The severing activity of spastin may contribute to reorganization of cortical MTs (Shibaoka, 1994). In this scenario, it is highly possible that impairment of spastin function in spl4-1 mutant inhibit the polarization of cortical MTs in internode, resulting in semi-dwarfism.

# REFERENCES


# AUTHOR CONTRIBUTIONS

GS and C-TK performed all the experiments. H-JK, KK, and N-CP designed the research. S-HK assisted in the phenotypic characterization. YS and CL performed the leaf senescence analyses. GA developed the spl4-2 mutant and provided advice about the manuscript. GS, C-TK, KK, and N-CP wrote and edited the manuscript. All authors have read and approved the content of the final manuscript.

#### FUNDING

This research was supported by the Co-operative Research Program for Agricultural Science and Technology Development (PJ013130 to N-CP) and the Rural Development Administration, South Korea, Basic Science Research Program through the National Research Foundation (NRF) of Korea funded by the Ministry of Education (NRF-2017R1A2B3003310 to N-CP and NRF-2016R1D1A1B03933357 to KK).

#### SUPPLEMENTARY MATERIALS

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01925/ full#supplementary-material


signals in Arabidopsis thaliana. Plant J. 20, 695–705. doi: 10.1046/j.1365-313X. 1999.00644.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 © 2019 Song, Kwon, Kim, Shim, Lim, Koh, An, Kang and Paek. 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.

# Emerging Roles of LSM Complexes in Posttranscriptional Regulation of Plant Response to Abiotic Stress

Rafael Catalá\*, Cristian Carrasco-López, Carlos Perea-Resa† , Tamara Hernández-Verdeja† and Julio Salinas\*

Departamento de Biotecnología Microbiana y de Plantas, Centro de Investigaciones Biológicas-CSIC, Madrid, Spain

#### Edited by:

Yan Guo, China Agricultural University, China

#### Reviewed by:

Nobuhiro Suzuki, Sophia University, Japan Ligeng Ma, Capital Normal University, China

> \*Correspondence: Rafael Catalá catala@cib.csic.es Julio Salinas salinas@cib.csic.es

#### †Present address:

Carlos Perea-Resa, Department of Molecular Biology, Massachusetts General Hospital, Boston, MA, United States Tamara Hernández-Verdeja, Umeå Plant Science Centre, Department of Plant Physiology, Umeå University, Umeå, Sweden

#### Specialty section:

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

Received: 08 November 2018 Accepted: 31 January 2019 Published: 19 February 2019

#### Citation:

Catalá R, Carrasco-López C, Perea-Resa C, Hernández-Verdeja T and Salinas J (2019) Emerging Roles of LSM Complexes in Posttranscriptional Regulation of Plant Response to Abiotic Stress. Front. Plant Sci. 10:167. doi: 10.3389/fpls.2019.00167 It has long been assumed that the wide reprogramming of gene expression that modulates plant response to unfavorable environmental conditions is mainly controlled at the transcriptional level. A growing body of evidence, however, indicates that posttranscriptional regulatory mechanisms also play a relevant role in this control. Thus, the LSMs, a family of proteins involved in mRNA metabolism highly conserved in eukaryotes, have emerged as prominent regulators of plant tolerance to abiotic stress. Arabidopsis contains two main LSM ring-shaped heteroheptameric complexes, LSM1-7 and LSM2-8, with different subcellular localization and function. The LSM1- 7 ring is part of the cytoplasmic decapping complex that regulates mRNA stability. On the other hand, the LSM2-8 complex accumulates in the nucleus to ensure appropriate levels of U6 snRNA and, therefore, correct pre-mRNA splicing. Recent studies reported unexpected results that led to a fundamental change in the assumed consideration that LSM complexes are mere components of the mRNA decapping and splicing cellular machineries. Indeed, these data have demonstrated that LSM1-7 and LSM2-8 rings operate in Arabidopsis by selecting specific RNA targets, depending on the environmental conditions. This specificity allows them to actively imposing particular gene expression patterns that fine-tune plant responses to abiotic stresses. In this review, we will summarize current and past knowledge on the role of LSM rings in modulating plant physiology, with special focus on their function in abiotic stress responses.

Keywords: LSM complexes, abiotic stress responses, Arabidopsis, posttranscriptional regulation, mRNA decapping, pre-mRNA splicing

#### INTRODUCTION

Due to their sessile nature, plants have evolved sophisticated adaptive mechanisms to correctly decipher external signals and deploy the corresponding adequate responses. This is of capital importance when facing adverse environmental conditions, since activating the right response can make the difference between life and death. In fact, abiotic stresses (i.e., extreme temperatures, drought or high salt concentration in the soil) are among the factors that most limit plant growth and development (Boyer, 1982; Bray et al., 2000). Plant responses to abiotic stresses, therefore, must be very precisely regulated. Results obtained in the last years have showed that these responses are mainly controlled by a wide reprogramming of gene expression (Seki et al., 2001;

**63**

Kreps et al., 2002; Shinozaki et al., 2003). Several layers of regulation seem to be involved in shaping this reprogramming (Barrero-Gil and Salinas, 2013; Guerra et al., 2015). Among them, transcriptional control has attracted most of the attention so far and numerous transcription factors and cis-acting elements functioning in plant adaptation to abiotic stresses have been described (Khan et al., 2018). Nevertheless, different reports have pinpointed that posttranscriptional regulation also plays an essential role in modulating plant response to these challenging situations (Mazzucotelli et al., 2008; Floris et al., 2009; Nakaminami et al., 2012). In particular, the control of mRNA stability and precursor-mRNA (pre-mRNA) splicing, two crucial pathways of RNA metabolism, appear to fine tune plant adaptation to adverse environments.

The Sm-like proteins (LSMs) are implicated in numerous aspects of RNA metabolism in eukaryotes. The LSMs are evolutionary conserved RNA-binding proteins, typically arranged in two heteroheptameric ring-shaped complexes known as LSM1-7 and LSM2-8 (**Figure 1**) (Wilusz and Wilusz, 2013). The LSM1-7 ring is localized in the cytoplasm and is a structural component of the decapping machinery involved in exonucleolytic mRNA decay, while the LSM2-8 complex is localized in the nucleus and is a core component of the spliceosome (Wilusz and Wilusz, 2013). Using Arabidopsis thaliana (Arabidopsis) as a model plant, different investigations unveiled that LSM proteins are highly conserved in plants and that, as in other eukaryotes, are organized in cytoplasmic (LSM1-7) and nuclear (LSM2-8) complexes (Wang and Brendel, 2004; Cao et al., 2011; Perea-Resa et al., 2012; Golisz et al., 2013). Moreover, recent studies evidenced that both LSM rings actively participate in regulating plant responses to abiotic stress conditions (Perea-Resa et al., 2016; Carrasco-López et al., 2017), which constitutes an unanticipated novel function for the eukaryotic LSMs. In this review, we will summarize the current state of the art knowledge on the activity of LSM proteins, paying special attention to their role in modulating abiotic stress responses. First, in order to situate the LSM complexes into their own context, we will provide a general view about the RNA metabolic pathways in which they participate (i.e., the exonucleolytic mRNA decay and the pre-mRNA splicing), discussing the implication of their corresponding intermediates in plant response to abiotic stresses. Then, the function of LSM complexes in controlling plant adaptation to these adverse situations will be reviewed. Finally, we will propose and comment on future research directions to better understand the role of LSM complexes as master integrators of plant adaptation to their ever-changing environment.

### POSTTRANSCRIPTIONAL REGULATION OF PLANT RESPONSE TO ABIOTIC STRESS

After transcription, mRNAs are subjected to different maturation and surveillance processes, which are indispensable to yield the functional transcripts. Differential control of the mechanisms implicated in these processes strongly influences not only the accumulation but also the structure of the final transcripts, significantly increasing the complexity of the information encoded by eukaryotic genomes (Schaefke et al., 2018). Plants may benefit from this layer of regulation since it provides a precise and reliable method to control gene expression, which, in turn, would ensure a timely response to environmental challenging situations. The LSM1-7 and LSM2-8 complexes are core components of two of the most influential posttranscriptional regulatory mechanisms namely, the mRNA decay and the pre-mRNA splicing processes, respectively. Before outlining the activity of the LSM complexes in mRNA decay and splicing, we will briefly describe the components of these two mechanisms, with special emphasis on their implication in plant response to abiotic stress.

### The Role of mRNA Decay Pathways in Plant Response to Abiotic Stress

The rate of mRNA decay ranges from minutes to several hours, depending on the transcripts (Chen and Coller, 2016). Control of mRNA decay provides a rapid instrument to regulate gene expression by modulating the stability of mRNAs. Two major pathways, the endonucleolytic and the exonucleolytic ones, govern transcript degradation (Garneau et al., 2007). The endonucleolytic cleavage pathway includes quality control mechanisms, such as the nonsense-mediated decay (NMD), that functions to prevent translation of error-containing mRNAs, or the posttranscriptional gene silencing (PTGS), mainly involved in the control of gene expression (Almeida and Allshire, 2005; Nasif et al., 2018). On the other hand, the exonucleolytic pathway is characteristic of transcripts that have performed their function and are no longer needed (Kilchert et al., 2016). This route starts with the shortening of the poly(A) tail positioned at the 3 0 end of the mRNAs, a process named deadenylation (Abbasi et al., 2013). Subsequently, transcripts can be degraded in a 3<sup>0</sup> -50 direction by the exosome or the exoribonuclease SUPPRESSOR OF VCS (SOV)/DIS3L2 (Abbasi et al., 2013; Soma et al., 2017). Alternatively, mRNAs can lose their 5<sup>0</sup> N7-methylguanosine (m7GDP) cap (5<sup>0</sup> CAP), by the action of the decapping complex, and then be degraded in 5<sup>0</sup> -3<sup>0</sup> direction through 5<sup>0</sup> -30 exoribonucleases (XRNs) (Jonas and Izaurralde, 2013).

#### mRNA Deadenylation

In eukaryotes, three major pathways control the deadenylation of poly(A) RNAs. They are defined by the participation of the PAN2-PAN3 (PAN2/3) complex, the Carbon Catabolite Repressor 4 (CCR4)/CCR4-Associated Factor 1 (CAF1)/NOT (CCR4/CAF1/NOT) complex, or the poly(A) ribonuclease PARN (Abbasi et al., 2013) (**Figure 2**). While PAN2/3 complex activity has not yet been reported in plants, recent data have revealed the implication of the two other pathways in different plant physiological processes. Nevertheless, data demonstrating their role in plant adaptation to abiotic stress remain scarce. Arabidopsis has a functional poly(A) ribonuclease, known as AtPARN, that mediates the deadenylation of transcripts induced during embryo development (Chiba et al., 2004; Reverdatto et al., 2004). The induction of AtPARN transcription by osmotic and salt stresses (Nishimura et al., 2005) suggests a possible function

of this ribonuclease in plant tolerance to these challenging situations, although no data has been reported in this regard. In yeast, the CCR4/CAF1/NOT complex consists of a catalytic center composed by three DEDD-type nucleases (CAF1, CAF40, CAF130) and one EEP-type endonuclease (CCR4), and five non-catalytic NOT proteins (NOT1 to NOT5) (Goldstrohm and Wickens, 2008; Abbasi et al., 2013). Plant homologs to NOT proteins have not yet been reported. Suzuki et al. (2015), however, demonstrated that Arabidopsis has two functional CCR4 proteins, AtCCR4a and AtCCR4b, which have been described to control the deadenylation of starch-biosynthesisrelated genes and, coherently, the starch levels. Whether AtCCR4a and AtCCR4b are involved in plant response to abiotic stress remains to be determined. On the other hand, Arabidopsis contains 11 genes encoding proteins with high identity to CAF1 (Walley et al., 2010), and two of them, AtCAF1a and AtCAF1b, display deadenylation activity (Liang et al., 2009). Interestingly, both proteins are implicated in the photoxidative stress response, and, furthermore, AtCAF1a in plant tolerance to salt stress controlling mRNA decay (Walley et al., 2010).

#### The 3<sup>0</sup> -5<sup>0</sup> Degradation Pathway

After poly(A) shortening by deadenylases, transcripts can be degraded in 3<sup>0</sup> -5<sup>0</sup> direction through the exosome (Chlebowski et al., 2013) or the SOV pathways (Zhang et al., 2010; Sorenson et al., 2018) (**Figure 2**). The eukaryotic exosome is a highly conserved macromolecular complex whose core is composed by nine proteins that are accompanied by different accessory proteins for target recognition, such as RNA-binding proteins and RNA helicases (Chlebowski et al., 2013; Thoms et al., 2015). The plant exosome has been involved in the control of different physiological processes, including cuticular wax biosynthesis (Hooker et al., 2007), seed germination and early seedling growth (Yang et al., 2013), and female gametogenesis and embryo development (Chekanova et al., 2007). Whether it operates in plant response to abiotic stress conditions remains to be assessed. SOV is a 3<sup>0</sup> -50 exoribonuclease that was identified as a suppressor of null mutations in VARICOSE (VCS), a key component of the decapping complex (see below) (Zhang et al., 2010). Recent discoveries indicated that SOV and VCS would control the decay of numerous transcripts through an elaborated feedback regulatory system (Sorenson et al., 2018). Null mutations in SOV do not produce any significant morphological phenotype, but their tolerance to abiotic stress has not been studied and, thus, a possible implication cannot be ruled out.

#### The 5<sup>0</sup> -3<sup>0</sup> Degradation Pathway

Following deadenylation, transcripts can also be degraded in 5<sup>0</sup> -3<sup>0</sup> direction by the action of XRNs (**Figure 2**). This pathway controls the decay of around 68% of all transcripts in Arabidopsis and, thus, is considered the most influential mRNA decay system in plants (Sorenson et al., 2018). Prior to the action of the XRNs, the 5<sup>0</sup> CAP that protects the 5<sup>0</sup> end of mRNAs from degradation needs to be removed by the

decapping complex. During this process, the decapping complex generates mRNAs with a monophosphate nucleotide in their 5 0 end that are the preferential substrates of XRNs (Nagarajan et al., 2013; Kurihara, 2017). The decapping reaction and the subsequent mRNA degradation occurs in discreet cytoplasmic foci known as processing bodies (P-bodies), where the target mRNAs and the degradation factors assemble (Sheth and Parker, 2003; Maldonado-Bonilla, 2014). The components of eukaryotic decapping machinery can be divided into two groups depending on their function. The first group contains the holoenzyme formed by DCP2 and DCP1. DCP2 has specific pyrophosphatase activity for removing the 5<sup>0</sup> CAP, and DCP1 functions as an activator of DCP2 (Coller and Parker, 2004). The second group is composed by different regulatory proteins that are required for efficient decapping, including the Protein Associated with Topoisomerase II (PAT1), the DEAD box helicase Dhh1p, and the RNA-binding complex LSM1-7 (Coller and Parker, 2004). Xu et al. (2006) showed that Arabidopsis has DCP2 and DCP1 proteins with similar roles as those of other eukaryotes, and that VCS, a previously described Arabidopsis protein (Deyholos et al., 2003), is the functional homolog of the human DCP2 activator Hedls/Ge-1. DCP5, in turn, was demonstrated to be the Arabidopsis homolog of the Dhh1p protein (Xu and Chua, 2009). The Arabidopsis protein DGP SPIRRING (SPI), furthermore, has been described to interact with DCP1 and, consequently, to be part of the decapping complex (Steffens et al., 2015) (**Figure 2**).

Different studies have involved the decapping complex in plant response to abiotic stress. For example, DCP2 and DCP1 accumulate to P-bodies in response to heat stress and low temperature, respectively (Perea-Resa et al., 2016). Nonetheless, these studies did not provide experimental evidence on the implication of these proteins in plant tolerance to abiotic stress. Interestingly, when plants are exposed to osmotic stress, the phosphorylation of DCP1 by the MAP protein kinase 6 (MPK6) promotes its interaction with DCP5 to control the decay rate of stress-related transcripts and, consequently, Arabidopsis tolerance to this adverse situation (Xu and Chua, 2012). SPI positively controls plant response to salt stress mediating the proper localization of salt stress-related transcripts at the P-bodies (Steffens et al., 2015). Whether this function is mediated through its interaction with DCP1 remains unknown.

VCS has been recently reported as a SnRK2s substrate when plants are exposed to osmotic stress (Soma et al., 2017). The phosphorylation of VCS by SRK2G/SnRK2.1 is required, under osmotic stress, for the adequate mRNA decay of transcripts encoding regulators of plant tolerance to drought (Soma et al., 2017). Finally, it has been well documented that Arabidopsis has a functional LSM1-7 complex that plays important roles in the control of plant adaptation to several abiotic stresses by governing the decay rate of transcripts corresponding to key regulators of plant responses to those conditions (Perea-Resa et al., 2012; Golisz et al., 2013; Perea-Resa et al., 2016; Wawer et al., 2018).

After the 5<sup>0</sup> mRNA decapping, XRN proteins degrade transcripts (**Figure 2**). The XRN family is highly conserved in eukaryotes and is mainly characterized by the existence of one or more nuclear enzymes (XRN2/RAT1 and XRN3), and one cytoplasmic enzyme (XRN1/PACMAN or XRN4) (Nagarajan et al., 2013). In Arabidopsis, there is not a homologous protein to XRN1 but there are three proteins with high identity to XRN2 (AtXRN2, AtXRN3, and AtXRN4), at both structural and functional levels (Kastenmayer et al., 2000). AtXRN2 and AtXRN3 are localized in the nucleus while AtXRN4 is localized in the cytoplasm (Kastenmayer et al., 2000). Only AtXRN4 has been related with plant tolerance to abiotic stresses, in particular to heat stress (Merret et al., 2013; Nguyen et al., 2015). Recent data have also implicated AtXRN4 in Arabidopsis sensitivity to ABA, suggesting that it could play a more general function in abiotic stress responses (Wawer et al., 2018).

# The Role of Pre-mRNA Splicing in Plant Response to Abiotic Stress

Introns were identified in the late 1970's as non-coding DNA sequences interrupting the coding sequence of adenovirus genes (Berget et al., 1977; Chow et al., 1977). Nowadays, it is known that introns are not exclusive of adenovirus but are widely present in all eukaryotic genomes (Matera and Wang, 2014).

#### The Core of the Spliceosome

Lerner et al. (1980) proposed that introns were spliced from the pre-mRNAs by a highly dynamic association of five small nuclear ribonucleoproteins (snRNPs), namely U1, U2, U4, U5, and U6, in a higher order complex known as spliceosome. Further investigations confirmed this assumption and revealed that these five snRNPs are associated with more than 300 proteins that coordinately participate in the control of spliceosome activity (Wahl et al., 2009; Matera and Wang, 2014). snRNPs are evolutionary conserved and their core is composed of a small uridine-rich nuclear RNA (snRNA), which defines the complex (i.e., U1, U2, U4, U5, and U6 snRNAs), and an accompanying heteroheptameric protein complex (Wilusz and Wilusz, 2013). Depending on the complex, the snRNPs can be divided in two groups. The first one, which has the Sm complex as protein moiety, includes the U1, U2, U4, and U5 snRNPs. The second group is formed by the U6 snRNP, which is accompanied by the LSM2-8 complex (Wilusz and Wilusz, 2013). The main function of the Sm complex is to guarantee the correct levels of U1, U2, U4, and U5 snRNAs (Will and Lührmann, 2011). The LSM2-8 complex participates in the biogenesis of the U6 snRNA, ensuring

its stability and adequate levels (Wilusz and Wilusz, 2013). Recent reports revealed that plants genomes encode proteins for all components of the Sm and LSM2-8 complexes (Wang and Brendel, 2004; Cao et al., 2011; Perea-Resa et al., 2012; Golisz et al., 2013). The characterization of the Arabidopsis LSM2-8 ring not only confirmed that plants have a functional LSM2-8 ring but also that it displays unexpected functions in controlling the spliceosome activity (Perea-Resa et al., 2012; Carrasco-López et al., 2017) (see below).

#### The Splicing Reaction

In the model of the splicing reaction accepted so far, intron scission from a pre-mRNA implies several heavily regulated steps (**Figure 3**) (review in Matera and Wang, 2014). The first one is the choice of the splice site (SS). This step is mediated by different cisacting elements present in the sequence of the pre-mRNA, which are recognized by trans-acting factors (i.e., Serine/argininerich protein (SR) or heterogeneous nuclear ribonucleoproteins (hnRNPs)). These factors seem to be the ultimate responsible for attracting the first components of the spliceosome, the U1 and U2 snRNPs. The U1 snRNP binds to the 5<sup>0</sup> SS, and the U2 snRNP, with associated factors, to the 3<sup>0</sup> SS and the branch point to define the so called intron defining complex. After the creation of this complex, the U4/U6.U5 tri-snRNP is recruited to give rise to the precatalytic complex (complex B). Then, the complex B is activated (complex B<sup>∗</sup> ) by the action of different RNA helicases, resulting in the liberation of the U1 and U4 snRNPs and, more relevant, in the first catalytic step that renders a free 5<sup>0</sup> exon and an intron-3<sup>0</sup> exon lariat intermediate. The second catalytic step is mediated by the action of several RNA helicases and ends up with the generation of the post-catalytic complex containing the intron lariat and the joined exons. After this second reaction, all snRNPs are released so they can follow a subsequent round of splicing.

#### Alternative Splicing in Plant Response to Abiotic Stress

The alternative selection of SS, a process known as alternative splicing (AS), is emerging as one of the most versatile regulatory systems in eukaryotes. By means of this process, different functional mRNAs can be generated from a particular premRNA through different arrangements of introns and exons, significantly increasing the protein diversity (Black, 2003; Nilsen and Graveley, 2010). Moreover, the AS can also participate in controlling the levels of functional mRNAs because of the generation of error-containing transcripts that are targets of mRNA surveillance pathways like the NMD (Staiger and Brown, 2013). In plants, AS has a pivotal role in the regulation of gene expression affecting about 60% of all intron-containing genes (Zhang et al., 2017). In response to abiotic stresses, AS seems to have a particular important function since several key regulatory genes of plant tolerance to such stresses have been shown to be prone to AS events (Deng et al., 2011; Seo et al., 2012, 2013; Leviatan et al., 2013; Liu et al., 2013; Ding et al., 2014a; Calixto et al., 2018). Ding et al. (2014a) reported that around 49% of all intron-containing genes in Arabidopsis are subjected to AS in response to salt stress. Similarly, when plants are exposed to

abiotic stress.

fpls-10-00167 February 19, 2019 Time: 13:3 # 6

low temperature, more than 2000 genes display changes in their splicing patterns, including some important regulators of cold tolerance (i.e., PHYB or PIF7) (Calixto et al., 2018). Interestingly, among the 2000 genes showing differential AS when plants are exposed to low temperature, some are well known splicing factors, such as RCF1, GEMIN2, or STA1 (Calixto et al., 2018), indicating that cold also modulates the spliceosome activity. Accordingly, several splicing factors have been reported to play key functions in plant response to abiotic stress (extensively revised in Laloum et al., 2017; Calixto et al., 2018) (**Figure 3**). Thus, Arabidopsis RS40 and RS41, two SR proteins likely involved in SS recognition, are required for the correct splicing of several pre-mRNAs under high salt conditions and, moreover, for plant tolerance to this adverse situation (Chen et al., 2013). Similarly, AtU1A, the Arabidopsis homolog of human U1A, a component of the U1 snRNP, promotes salt tolerance and determines the 5<sup>0</sup> SS selection, shaping the splicing patterns in response to high salt (Gu et al., 2018). PRP31, the Arabidopsis homolog of a subunit of the U4/U6.U5 tri-snRNP in human and yeast, has been demonstrated to ensure the correct splicing of a number of pre-mRNAs under high salt, mannitol and low temperature and plant tolerance to these abiotic stresses (Du et al., 2015). Other components of this complex, such as STA1 and ZOP1, have also been implicated in controlling pre-mRNA splicing and tolerance to abiotic adverse conditions, including low temperature, high salt and osmotic stress (Lee et al., 2006; Du et al., 2015). Furthermore, SKIP, a component of the Arabidopsis complex B<sup>∗</sup> (Li et al., 2016), has been shown to regulate plant tolerance to high salt by determining the correct splicing pattern under this challenging environment (Feng et al., 2015).

# THE EUKARYOTIC LSM PROTEINS

The eukaryotic LSM proteins belong to the large family of "Smlike" proteins (Tharun, 2008). They were identified as antigens recognized by antibodies of systemic lupus erythematosus, and were named after the patient that provided the serum [i.e., Stephanie Smith (Sm)] (Tan and Kunkel, 1966). Later on, Lerner et al. (1981) reported for the first time that LSM proteins are complexed with different snRNAs to form snRNPs. They are evolutionary conserved from Archaebacteria and prokaryotes to eukaryotes, indicating that they could be already present in the last universal common ancestor (LUCA) (Anantharaman et al., 2002). Typically, the LSM proteins are small peptides (∼10–25 kDa) with common structural features. Their sequences contain a highly conserved bipartite domain, the Sm-domain, spanning over 100 conserved residues interrupted by a non-conserved region of up to 30 amino acids (Hermann et al., 1995; Séraphin, 1995; Achsel et al., 1999;

Catalá et al. LSM Complexes in Abiotic Stress Responses

Kambach et al., 1999; Salgado-Garrido et al., 1999). The tertiary structure displays a characteristic fold containing an N-terminal helix adjacent to a strongly bent five-stranded antiparallel β-sheet, the so-called "Sm-fold" (Kambach et al., 1999). The LSM proteins tend to arrange in ring shaped hetero-heptameric complexes with RNA-binding capability (Raker et al., 1996; Mayes et al., 1999; Salgado-Garrido et al., 1999). These rings ensure the correct levels of U6 snRNAs and facilitate the assembly of snRNP complexes (Tharun, 2008; Wilusz and Wilusz, 2013).

The LSM protein family can be divided into the Sm and the LSM subfamilies. The Sm subfamily is mainly composed by seven proteins (SmB/B<sup>0</sup> , SmD1, SmD2, SmD3, SmE, SmF, and SmG) assembled in heteroheptameric complexes around the snRNAs of the major (U1, U2, U4, and U5 snRNAs) and minor (U11, U12, and U4atac snRNAs) spliceosomes (Hermann et al., 1995; Will et al., 1999). On the other hand, the LSM subfamily is composed of eight proteins (LSM1-LSM8) that are also organized in heteroheptameric complexes (Zhou et al., 2014). Six of them are homologous to the Sms (LSM2, LSM3, LSM4, LSM5, LSM6, and LSM7), LSM8 is weakly related to SmB/B<sup>0</sup> , and LSM1 does not display significant similarity to any Sm protein (Salgado-Garrido et al., 1999). LSM1 and LSM8 proteins are mutually exclusive and determine the subcellular localization of the complexes and, more relevant, their function (Salgado-Garrido et al., 1999; Tharun et al., 2000) (**Figure 1**). LSM1 promotes the constitution of an LSM1-7 complex in the cytoplasm that plays a crucial role in the decapping complex (Salgado-Garrido et al., 1999). In contrast, LSM8 directs the formation of the LSM2-8 complex that localizes in the nucleus and forms, together with the U6 snRNA, the U6 snRNP (Séraphin, 1995; Salgado-Garrido et al., 1999). Both LSM complexes share preference for the 3<sup>0</sup> ends of RNAs, the LSM1-7 ring for 3<sup>0</sup> oligoadenylated tracts, and the LSM2-8 for 3 <sup>0</sup> oligouridinylated (oligo-U) tracts (Chowdhury et al., 2007; Zhou et al., 2014). In addition to the canonical LSMs, there are at least three classes of larger proteins having Sm-domains and other functional domains with RNA-binding capability. These LSM proteins are not well characterized, although it seems that some of them are related to mRNA translational control (Decker and Parker, 2006).

# The LSM1-7 Complex

When the eukaryotic LSM proteins were identified, it soon became clear that one of them, the LSM1, did not bind to the U6 snRNA (Salgado-Garrido et al., 1999). Furthermore, the deletion of LSM1 neither affected the levels of U6 snRNA nor the splicing efficiency (Mayes et al., 1999; Salgado-Garrido et al., 1999), indicating that LSM1 has a different function than the LSM2-8 complex. Indeed, it is a component of the 5<sup>0</sup> -30 mRNA degradation pathway mediating the decapping of several mRNAs (Boeck et al., 1998) (**Figure 1**). On the other hand, further analyses revealed that, in yeast, the proteins LSM2 to LSM7 participate in the control of mRNA decapping, while LSM8 does not (Tharun et al., 2000). Consistent with these results, LSM1 to LSM7 proteins co-localize and co-immunoprecipitate with DCP1, PAT1 and different mRNAs (Bouveret et al., 2000; Tharun et al., 2000). Similarly, the human LSM1-7 complex associates with PAT1 and XRN1 to modulate mRNA degradation

(Ingelfinger et al., 2002). It was proposed that LSM1-7 proteins are arranged in a complex different than that of the LSM2-8 nuclear one, which would be a component of the decapping complex (Tharun et al., 2000). This hypothesis was validated by crystallization experiments, which demonstrated that the yeast LSM1-7 complex resembles a thick donut, generated by the sequential interaction of the seven proteins (LSM1-LSM2-LSM3- LSM6-LSM5-LSM7-LSM4) through their Sm domains (Sharif and Conti, 2013) (**Figure 1**). The ring organization of the complex yields a lumen where the spacing between subunits matches the space between nucleotides in the RNA, allowing its interaction with single-stranded RNAs (Khusial et al., 2005). The LSM1-7 complex holds high affinity for mRNAs containing oligo-A tracts of 6 or more nucleotides (oligoadenilated) at its 3<sup>0</sup> terminal end (Chowdhury et al., 2007). In the model proposed for its function in regulating the decapping reaction, the complex would bind to the 3<sup>0</sup> end of the oligoadenylated mRNAs in P-bodies (Chowdhury et al., 2007; Tharun, 2008) (**Figure 4**).

#### The LSM1-7 Complex in Plants

A bioinformatic approach allowed the first report about plant LSM proteins. Séraphin (1995) found that the genomes of alfalfa, Arabidopsis, Brassica campestris and rice contained genes encoding proteins with high sequence identity to the Sm proteins of yeast and animals. The Arabidopsis genome, in particular, has been described to contain around 42 genes encoding proteins with the characteristic Sm domain (Wang and Brendel, 2004; Cao et al., 2011; Perea-Resa et al., 2012; Golisz et al., 2013). Eleven Arabidopsis genes encode the eight canonical LSM proteins,

LSM1-LSM8, from yeast and animals (Perea-Resa et al., 2012; Golisz et al., 2013). Genes LSM1, LSM3, and LSM6 are duplicated and encode pairs of redundant proteins (LSM1A, B; LSM3A, B and LSM6A, B). As in yeast and animals, Arabidopsis LSM proteins are arranged in two main complexes, the cytoplasmic LSM1-7 and the nuclear LSM2-8 ones (Perea-Resa et al., 2012; Golisz et al., 2013). Fortunately, Arabidopsis null mutants lsm1alsm1b and lsm8 resulted to be viable, providing unique genetic tools to approach the functional characterization of the two LSM complexes (Perea-Resa et al., 2012; Golisz et al., 2013).

The functional characterization of lsm1alsm1b double mutants revealed that, as in other eukaryotes, Arabidopsis LSM1 proteins are essential for the constitution of the LSM1-7 cytoplasmic ring (Perea-Resa et al., 2012). The Arabidopsis LSM1-7 complex co-immunoprecipitate with PAT1 and, moreover, co-localize with DCP2 and VCS in P-bodies, strongly suggesting that, together with DCP1, DCP5 and SPI, it participates in the Arabidopsis decapping machinery (Perea-Resa et al., 2012; Golisz et al., 2013) (**Figure 4**). Indeed, it has been shown to be essential for the correct decapping of a plethora of selected mRNAs, ensuring their precise turnover (Perea-Resa et al., 2012; Golisz et al., 2013). Interestingly, among the LSM1-7 targets, several transcripts corresponding to genes having key roles in Arabidopsis development were found (Perea-Resa et al., 2012; Golisz et al., 2013), which would account for the developmental alterations displayed by the lsm1alsm1b plants (Perea-Resa et al., 2012; Golisz et al., 2013). All these data highlight the relevance of the LSM1-7 complex in shaping plant physiology by modulating mRNA turnover.

#### The LSM1-7 Complex in Plant Response to Abiotic Stress

Genetic and molecular evidence unveiled that the LSM1-7 complex has a pivotal role in regulating plant response to abiotic stress. The expression levels of LSM1 significantly increase when plants are exposed to low temperature and, accordingly, the levels of the corresponding protein augment as well (Perea-Resa et al., 2016). In contrast, the levels of LSM1 transcripts do not change in response to other abiotic stresses, such as high temperature, high salt or drought (Okamoto et al., 2016; Perea-Resa et al., 2016). The functional characterization of Arabidopsis LSM5 and LSM4 proteins showed that they are implicated in plant response to abiotic stress (Xiong et al., 2001; Zhang et al., 2011). A point-mutation allele of LSM5, sad1, displayed reduced tolerance to salt and drought stresses compared to wild-type plants (Xiong et al., 2001). Similarly, a null mutant for LSM4 showed decreased salt stress tolerance (Zhang et al., 2011). Although the implication of these two proteins in the decapping reaction under abiotic stress was not evaluated, it was later demonstrated that LSM5/SAD1 promotes mRNA degradation under control conditions by removing the 5 <sup>0</sup>CAP in Arabidopsis (Golisz et al., 2013). Nonetheless, taken into account that LSM4 and LSM5/SAD1 are shared components of the cytoplasmic and nuclear LSM complexes, it is difficult to discriminate if their function in abiotic stress response is mediated through the LSM1-7 complex, the LSM2-8 complex or both. The characterization of the lsm1alsm1b double mutant, however, provided definitive evidence that the LSM1-7 complex is necessary for the correct adaptation of plants to situations of abiotic stress. It negatively regulates the ability of Arabidopsis to cold acclimate and tolerate drought, but functions as a positive regulator of Arabidopsis tolerance to salt stress (Perea-Resa et al., 2016). Genome-wide expression analysis of lsm1alsm1b plants unveiled that the LSM cytoplasmic complex differentially regulates Arabidopsis response to abiotic stresses by differentially controlling the levels of stress-inducible transcripts depending on the stress (Perea-Resa et al., 2016). Additional characterization of lsm1alsm1b plants unveiled an unexpected functional plasticity of the LSM1-7 complex to modulate the interaction of plants with their environment. In fact, depending on the abiotic stress conditions, the complex interacts with selected stress-inducible transcripts, such as LEA7, ZAT12, ABR1, ANAC019, AHK5, or ANAC092, targeting them for decapping and subsequent degradation, ensuring the appropriate patterns of downstream stress-responsive gene expression that are required for plant adaptation (Perea-Resa et al., 2016) (**Figure 5**). This stressdependent differential control of mRNA turnover represents a new layer of regulation in plant adaptation to unfavorable environmental conditions. Remarkably, it was demonstrated that the LSM cytoplasmic complex is required for the constitution of P-bodies in plants under abiotic stress conditions (Perea-Resa et al., 2016). Furthermore, the exposure of Arabidopsis plants to these conditions promotes the accumulation of the LSM1- 7 complex in P-bodies (Perea-Resa et al., 2016). Given that other P-body constituents such as DCP1 and VCS have also been shown to accumulate there in response to abiotic stresses (Perea-Resa et al., 2016), it seems reasonable to hypothesize that all components of the decapping machinery concentrate in these cytoplasmic foci when plants are confronted to adverse environmental conditions to govern mRNA decay.

#### The Role of LSM1-7 Complex in the Control of Stress-Induced ABA Biosynthesis

The increase of abscisic acid (ABA) levels is one of the primary signals triggering adaptive responses when plants are exposed to abiotic stresses such as low temperature, drought or high salt (Yang et al., 2017). ABA biosynthesis is tightly regulated at different levels, the posttranscriptional one being among the most relevant (Yang et al., 2017). Xiong et al. (2001) evidenced the pivotal role of LSM proteins in regulating ABA biosynthesis and signaling. The characterization of mutant plants with altered function of LSM5/SAD1 revealed that this LSM subunit controls the levels of transcripts corresponding to key intermediates in ABA biosynthesis (i.e., AAO3 or ABA3) and signaling (i.e., PP2C) (Xiong et al., 2001). Nonetheless, as already mentioned, the participation of LSM5/SAD1 in LSM1-7 and LSM2-8 complexes prevent to determine the actual mechanisms through which such control is carried out. Recent analyses of the lsm1alsm1b mutant have shed some light to this conundrum. Perea-Resa et al. (2016) reported that lsm1alsm1b displays increased levels of ABA in response to low temperature and high salt, but not in response to water stress, indicating that the LSM1-7 complex differentially regulates stress-induced ABA biosynthesis. They demonstrated that, depending on the stress situation, the LSM ring exerts

lines indicate positive and negative regulation, respectively.

this function by differentially controlling the decapping of NCED3 and NCED5 mRNAs, two transcripts encoding key ABA biosynthetic enzymes, and, therefore, their decay rate. The LSM1- 7 complex attenuates ABA biosynthesis under cold conditions by interacting with NCED3 and NCED5 mRNAs, and under salt stress with NCED5 mRNA, promoting their degradation. None of them, however, is target of the complex in response to water stress (Perea-Resa et al., 2016). On the other hand, it has been proposed that ABA perception and signaling through the canonical PYL/PYR/RCAR-PP2C-SnRK2 pathway is governed by the decapping complex (Wawer et al., 2018). Thus, the decay rate of the mRNAs encoding the ABA receptor PYR1 and the ABA-unresponsive SnRK2 protein kinases would be determined by the LSM1-7 complex, DCP5, and XRN4 (Wawer et al., 2018). Still, whether this activity of the decapping machinery is also involved in plant adaptation to abiotic stress conditions remains to be investigated. All these results provide genetic and molecular evidence that the LSM cytoplasmic complex contributes to establish the appropriate levels of ABA in Arabidopsis plants exposed to different abiotic stresses.

#### The LSM2-8 Complex

As already mentioned, the eukaryotic LSM2-8 complex was initially identified in the nucleus, associated to the U6 snRNA (Salgado-Garrido et al., 1999) (**Figure 1**). The complex, composed by proteins LSM2 to LSM8 sequentially ordered (LSM2-LSM3-LSM6-LSM5-LSM7-LSM4-LSM8) in a ring shape (Zhou et al., 2014), displays preference for oligo-U tracts (Achsel et al., 1999; Mayes et al., 1999), a typical feature of RNAs transcribed by the RNA polymerase III (Pol III). It has been described that the LSM nuclear complex participates in different crucial processes of pre-mRNA splicing, such as the biogenesis of the U6 snRNA, the constitution of U4/U6 di-snRNP and U4/U6.U5 tri-snRNP complexes, or the regeneration of the spliceosome (reviewed in Didychuk et al., 2018) (**Figure 6**). The U6 snRNA is part of the catalytic core of the spliceosome and, thus, its levels should be subjected to a tight control. Just after its transcription by the RNA Pol III, the chaperone-like La/Lhp1 protein interacts with the U6 snRNA Poly(U) tract to promote its retention in the nucleus. This interaction is weakened by the binding of the RNA-binding protein Prp24, allowing the access of the 3<sup>0</sup> to 5<sup>0</sup> RNA exonuclease MPN1/USB1/UBL1 that removes the last uridine moiety and leaves a phosphate in the 3<sup>0</sup> end of the transcript. Then, the presence of this phosphate favors the interaction of the LSM2-8 complex with the 3<sup>0</sup> end of the U6 snRNA. The binding of this complex inhibits U6 snRNA degradation, retains the transcript in the nucleus, and is required for the proper formation of the U6 snRNP and the subsequent composition of di- and tri-snRNPs (**Figure 6**).

#### The LSM2-8 Complex in Plants

Perea-Resa et al. (2012) reported for the first time the existence of a full functional LSM2-8 complex in plants. Arabidopsis has a LSM2-8 ring with identical structure to the one reported in yeast and metazoans and, as expected, the assembly of this complex in the nucleus is directed by the LSM8 protein (Perea-Resa et al., 2012). Experimental evidence demonstrate that the LSM nuclear complex also ensures the correct levels of U6 snRNA by promoting its stability (Perea-Resa et al., 2012; Golisz et al., 2013). Moreover, LSM8 co-immunoprecipitates with the Arabidopsis homologues of La/Lhp1 and Prp24, confirming that the LSM nuclear complex is part of a canonical U6 snRNP in plants (Golisz et al., 2013). The characterization of Arabidopsis lsm4 and lsm5/sad1 mutants suggested that LSM4 and LSM5 proteins are likely implicated in the control of pre-mRNA splicing (Xiong et al., 2001; Zhang et al., 2011; Cui et al., 2014). Nevertheless, as already mentioned, the fact that these proteins belong to both LSM1-7 and LSM2-8 complexes, hinders the possibility of attributing that function to one or another complex. Highcoverage RNA-seq analysis using null lsm8 mutants conclusively demonstrated that the LSM nuclear complex participates in the control of both constitutive and AS of a number of pre-mRNAs (Carrasco-López et al., 2017). This unexpected ability of the LSM2-8 complex to control the splicing of just a discrete number of pre-mRNAs indicates that the core components of eukaryotic spliceosome contribute, together with the associated proteins, to determining the spliceosome activity specificity. Moreover, this role is essential to establish the adequate gene-expression landscape in Arabidopsis (Perea-Resa et al., 2012; Golisz et al., 2013). Interestingly, several pre-mRNA targets of the LSM2-8 complex correspond to development-related genes (Perea-Resa et al., 2012; Golisz et al., 2013), suggesting that it is involved in plant development. Indeed, it was demonstrated that the LSM2- 8 complex regulates different developmental processes (Perea-Resa et al., 2012). A recent study described that LSM4 and LSM5 control the splicing pattern of pre-mRNAs corresponding to important components of the clock and set the adequate length of the circadian period (Perez-Santángelo et al., 2014),

the binding of the La protein to its oligo-U tract. The binding to the body of the snRNA of Prp24 and the excision of the last U by MPN1/USB1/UBL1 (MPN1) promote the substitution of MPN1 by the LSM2-8 complex. The U6 snRNA, Prp24 and the LSM2-8 complex constitute the U6 snRNP that, subsequently, favors the formation of U4/U6 di- and U4/U6.U5 tri-snRNPs. In addition, after the splicing reaction a fully functional LSM2-8 complex is essential for the regeneration of the spliceosome.

which indicates that the LSM2-8 complex might also regulate the circadian rhythm in plants. Again, the participation of LSM4 and LSM5 in both LSM complexes hampers to clearly discerning the involvement of each complex in this regulation.

# The LSM2-8 Complex in Plant Response to Abiotic

negative regulation, respectively.

Stress Arabidopsis LSM8 transcripts and the corresponding protein accumulate in response to low temperature (Carrasco-López et al., 2017), which suggest that the LSM nuclear complex might also have a role in plant response to abiotic stress. On the other hand, we have already mentioned that LSM4 and LSM5 positively regulate Arabidopsis tolerance to salt stress (Xiong et al., 2001; Zhang et al., 2011; Cui et al., 2014), but if this function is carried out throughout the LSM2-8 complex remains uncertain. Concluding experimental evidence on the implication of the Arabidopsis LSM2-8 complex in abiotic stress response has been recently attained by functionally characterizing lsm8 null mutant plants. These analyses revealed that the LSM nuclear complex differentially regulates Arabidopsis tolerance to abiotic stress. It functions as a negative regulator of the cold acclimation process, while positively controlling tolerance to salt stress (Carrasco-López et al., 2017). Deep RNA-seq experiments using lsm8 mutant plants subjected to cold and salt stresses unveiled that the LSM2-8 complex operates by ensuring the efficiency and accuracy of the splicing of selected pre-mRNAs, depending on the adverse environmental conditions. Thus, under low temperature conditions, the complex ensures the correct splicing of a selected subset of pre-mRNAs enriched in cold-related genes, such as MYB96, PRR5 or RVE1. In contrast, in response to high salt it guarantees the adequate splicing of a different group of premRNAs, which is enriched in salt stress-related genes, such as

SAT32, NHX1 or SIS (**Figure 7**). In both cases, moreover, the pre-mRNAs are distinct to those whose splicing is controlled by the complex under standard conditions (Carrasco-López et al., 2017). It is worth mentioning that miss-splicing of most LSM2-8 targeted pre-mRNAs leads to the generation of NMD signatures, indicating that the complex also warrants correct levels of the corresponding functional transcripts (Carrasco-López et al., 2017). Hence, the unanticipated specificity of the nuclear complex for particular pre-mRNA targets upon plant exposure to different abiotic stress conditions seems to ensure the adequate transcriptional patterns for each stress situation and, consequently, a correct adaptive response.

In lsm8 mutants exposed to low temperature or high salt, intron retention is, by far, the most abundant detected category of altered splicing events, evidencing that the main function of the Arabidopsis LSM2-8 complex in splicing, when plants are exposed to abiotic stress, is ensuring the full processing of introns (Carrasco-López et al., 2017). An important question that arises from these results is which are the molecular determinants underlying the specificity of the complex to target the introns that are going to be spliced under distinct adverse environments. Most genes containing targeted introns are not differentially transcribed under cold or high salt conditions, which excludes the possibility that the introns selected by the LSM nuclear complex belong to genes only or highly transcribed under a particular external stimulus. Regarding the possibility that the selected introns may include particular sequence motifs, no enrichment of sequence motifs in specific introns or particular frequencies of nucleotide sequences around their 5<sup>0</sup> and 3<sup>0</sup> SSs, or in their branch sites, have been found either. Interestingly, however, there are significant differences in GC content and/or length between some subsets of introns specifically spliced by the complex in response to distinct stress conditions (Carrasco-López et al., 2017). It has been proposed that secondary structures at the 5<sup>0</sup> splice site of introns disfavor its recognition by the spliceosome and, thus, would affect their correct splicing (Ding et al., 2014b). It can be predicted that different GC content and length would end up in distinct secondary structures in these regions of the transcripts. Hence, it is tempting to speculate that particular secondary structural features of the introns could determine their specific selection by the LSM2-8 complex and, thus, by the spliceosome, in response to a given abiotic stress. The characterization of the lsm8 mutant plants, therefore, has revealed that the core components of the spliceosome, such as the LSM2-8 complex, may regulate the activity specificity of this macromolecular machinery in an environmental conditiondependent manner, which represents a novel functional capacity for those components. Remarkably, furthermore, this function constitutes a new layer of posttranscriptional regulation in response to external stimuli in eukaryotes that seems to be essential for plant adaptation to adverse surroundings.

#### FUTURE PERSPECTIVES

Research advances in recent years have significantly expanded our understanding about the function of the LSM complexes in posttranscriptional regulation of plant response to abiotic stress. It is obvious, however, that we are still far from envisaging the molecular mechanisms regulating their function and the molecular determinants of their specificity. Data suggest that the regulatory mechanisms of plant LSM complexes take place at different levels. The methylation of LSM4 promotes its function in pre-mRNA splicing and plant tolerance to salt stress (Zhang et al., 2011). On the other hand, the activity of some components of the decapping machinery in response to osmotic stress has been shown to be governed by their phosphorylation status (Xu and Chua, 2012; Soma et al., 2017). Interestingly, LSM1 and LSM8 display amino acid sequence motives characteristic of MPK targets, and coimmunoprecipitate with some components of the MPKs family (Perea-Resa et al., 2012; Golisz et al., 2013), suggesting that the function of LSM complexes could be shaped by their differential phosphorylation pattern under a particular stress condition. Determining whether the LSM complexes may be differentially methylated, phosphorylated, or undergo some other type of posttranslational modification in response to different abiotic stresses to better adapt to challenging environments represents a research line worth to be developed in the future.

One of the most remarkable features of eukaryotic LSM complexes is that they have six common proteins. Hence, LSM1 and LSM8 have to compete for LSM2-LSM7 in order to constitute their corresponding complexes. Results obtained in yeast indicate that this competition would allow a co-regulatory mechanism of nuclear and cytoplasmic RNA processing under stressful situations (Spiller et al., 2007; Luhtala and Parker, 2009). Exploring whether this mechanism exists in plants, and if it has a role in modulating mRNA decapping or pre-mRNA splicing depending on the abiotic stress conditions is another exciting line of investigation that requires to be developed.

Another intriguing issue that needs to be approached in the future is the identity of the molecular determinants controlling the specificity of the LSM complexes to select their RNA targets in response to different adverse environmental conditions. It is well known that the levels of epigenetic marks in the chromatin influence pre-mRNA splicing in yeast and metazoans (Luco et al., 2011). In Arabidopsis, a link between the levels of the epigenetic mark histone H3 lysine 36 trimethylation (H3K36me3) and pre-mRNA splicing has been established (Pajoro et al., 2017). Indeed, this study revealed that both, the enzymes involved in the deposition of the H3K36me3 mark and the readers of the mark, contribute to determine the patterns of AS. Considering these data and the fact that different abiotic stress conditions induce specific patterns of epigenetic marks, it is reasonable to speculate that the LSM2-8 complex could select their pre-mRNA targets in each condition depending on the particular epigenetic marks present in the chromatin of the corresponding genes. Remarkably, it has been reported that uncapped-degrading mRNAs and alternatively spliced pre-mRNAs show a nocoincident preference for some particular chemical modifications in Arabidopsis (Vandivier et al., 2015). Whether RNA chemical modifications may also significantly contribute to determine the specificity of the LSM complexes in mRNA decapping and/or pre-mRNA splicing is another interesting possibility that deserves to be explored.

#### AUTHOR CONTRIBUTIONS

fpls-10-00167 February 19, 2019 Time: 13:3 # 12

All authors designed the review. CC-L, RC, and JS designed the figures. RC and JS wrote the manuscript. All authors read and approved the final version of the manuscript.

#### REFERENCES


#### FUNDING

This work was supported by grant BIO2016-79187-R from AEI/FEDER, UE to JS.

#### ACKNOWLEDGMENTS

We acknowledge support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI).




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

Copyright © 2019 Catalá, Carrasco-López, Perea-Resa, Hernández-Verdeja and Salinas. 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.

# Modulation of Ethylene and Ascorbic Acid on Reactive Oxygen Species Scavenging in Plant Salt Response

Juan Wang1,2 and Rongfeng Huang1,2 \*

<sup>1</sup> Biotechnology Research Institute, Chinese Academy of Agricultural Sciences, Beijing, China, <sup>2</sup> National Key Facility for Crop Gene Resources and Genetic Improvement, Beijing, China

Salt stress causes retarded plant growth and reduced crop yield. A complicated regulation network to response to salt stress has been evolved in plants under high salinity conditions. Ethylene is one of the most important phytohormones, playing a major role in salt stress response. An increasing number of studies have demonstrated that ethylene modulates salt tolerance through reactive oxygen species (ROS) homeostasis. Ascorbic acid (AsA) is a non-enzymatic antioxidant, contributing to ROS-scavenging and salt tolerance. Here, we mainly focus on the advances in understanding the modulation of ethylene and AsA on ROS-scavenging under salinity stress. We also review the regulators involved in the ethylene signaling pathway and AsA biosynthesis that respond to salt stress. Moreover, the AsA pool is affected by many environmental conditions, and the potential role of ethylene in AsA production is also extensively discussed. Novel insights into the roles and mechanisms of ethylene in AsA-mediated ROS homeostasis will provide critical information for improving crop salt tolerance.

#### Edited by:

Yan Guo, China Agricultural University, China

#### Reviewed by:

Tse-Min Lee, National Sun Yat-sen University, Taiwan Yongqing Yang, China Agricultural University, China

> \*Correspondence: Rongfeng Huang rfhuang@caas.cn

#### Specialty section:

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

Received: 01 September 2018 Accepted: 27 February 2019 Published: 18 March 2019

#### Citation:

Wang J and Huang R (2019) Modulation of Ethylene and Ascorbic Acid on Reactive Oxygen Species Scavenging in Plant Salt Response. Front. Plant Sci. 10:319. doi: 10.3389/fpls.2019.00319 Keywords: ethylene, AsA, salt stress, ROS scavenging, homeostasis

# INTRODUCTION

According to the report from the Food and Agricultural Organization of the United Nations, there will be challenges related to the productivity of crops to supply more food for an additional 2.3 billion people by 2050. Crop yield is greatly affected by various abiotic stresses like drought, high salinity, cold, and heat (Zhu, 2016). For instance, high salinity stress disturbs plant physiological processes through osmotic stress and ionic toxicity, causing reductions in both crop growth and yield (Yang and Guo, 2018a,b). For better utilization of salt-affected lands, it is of great help to develop crops with improved salt tolerance through molecular-assistant breeding to reveal the underlying mechanisms involved in plant response to salt stress.

Salt stress responses generally correlate with the regulations of phytohormones, including abscisic acid (ABA), jasmonic acid, gibberellin, and ethylene (Julkowska and Testerink, 2015). Although there exists opposite modulation between monocot and dicot plants (Peng et al., 2014; Yang et al., 2015), increasing investigations have revealed that ethylene-conferred salt tolerance is mediated by deterring reactive oxygen species (ROS) homeostasis (Jiang et al., 2013; Li et al., 2014; Peng et al., 2014; Yang and Guo, 2018b). Under salt stress, ROS, including hydrogen peroxide, superoxide anions and hydroxyl radicals, accumulate and damage cellular structure (Ahanger et al., 2017). ROS plays a dual role in response to stresses as toxic by products and major signal

(Qi et al., 2018), and the excess ROS could be scavenged through enzymatic and non-enzymatic antioxidant defense systems (You and Chan, 2015). Accumulating investigations have revealed that ascorbic acid (AsA) is an essential compound of non-enzymatic antioxidant in plants, functioning in plant growth, hormone signaling, and stress response (Bulley and Laing, 2016; Mellidou and Kanellis, 2017; Vidal-Meireles et al., 2017). AsA plays especially critical roles in the fine control of ROS homeostasis to improve salt tolerance (Smirnoff, 2000; Shalata and Neumann, 2001; Zhang et al., 2012; Wang J. et al., 2013), implying that AsA has an essential modulation in salt response. Considering there are some reviews about ethylene-modulated salt response (Zhang M. et al., 2016), this mini review will focus on the advances in understanding the modulation of ethylene on AsA biosynthesis and ROS-scavenging under salinity stress.

# The Regulation of Ethylene on ROS Homeostasis Is Tightly Associated With the Plant Response to Salt Stress

Gaseous phytohormone ethylene plays an important role in mediating numerous specific growth and development processes (Wang et al., 2018), especially in response to various stress conditions (Wang F. et al., 2013; Dubois et al., 2018). The biosynthesis and signaling pathway of ethylene have been well established (Guo and Ecker, 2004). After recognition of ethylene by endoplasmic reticulum membrane-associated receptors, the interaction of ethylene receptors with CONSTITUTIVE TRIPLE RESPONSE1 (CTR1) will be released, and the phosphorylation of CTR1 on ETHYLENE INSENSITIVE 2 (EIN2) will be liberated. Then, the C-terminal of EIN2 is generated by an unknown mechanism and is transported to cytoplasmic processing-body (P-body) to repress translation of EIN3 BINDING F-BOX1/2 (EBF1/2), which mediates the proteasomal degradation of EIN3 and EIN3-LIKE 1 (EIL1), resulting in the stability of EIN3/EIL1 proteins and promotion of ethylene response (Li et al., 2015). APETALA2/ETHYLENE RESPONSE FACTORS (AP2/ERFs) are one of the most important transcription factor families, regulating multiple developmental and stress response processes (Phukan et al., 2017), most of which are downstream targets of ethylene signaling (Liu et al., 2016).

Ethylene has long been known for modulating salt stress response (Cao et al., 2007). For instance, blocked ethylene signaling confers reduced salt tolerance to Arabidopsis (Achard et al., 2006; Peng et al., 2014). ein3 eil1 double mutants and other ethylene signaling-related mutants showed enhanced sensitivity to salt stress. In contrast to this modulation, ethylene displays a negative role in rice (Yang et al., 2015). OsEIL1 and OsEIL2 RNAi transgenic plants displayed increased salt tolerance. The regulation of ethylene biosynthesis also plays different roles in salt tolerance between Arabidopsis and rice (Jiang et al., 2013; Li et al., 2014). Ethylene Overproducer 1 (ETO1) plays a positive role in salt response through promoting ROS generation, followed with Na+/K<sup>+</sup> homeostasis modulation in Arabidopsis. However, SALT INTOLERANCE 1 (SIT1) negatively regulates salt response due to activation on MITOGEN-ACTIVATED PROTEIN KINASE 3/6 (MPK3/6) in rice, which promotes ethylene and ROS overproduction (**Table 1**). Thus, the different mechanisms of ethylene-directed salt response between monocot and dicot plants remain in need of research. Subsequent advances indicate that ROS homeostasis is essential for ethylene regulation of plant growth and stress response (Steffens, 2014; Zhong et al., 2014; Yang et al., 2017). ROS is a double-edged sword during salt stress response. On the one hand, ROS act as important signal molecules to activate downstream metabolic pathways. Previous studies demonstrate that ROS burst via RESPIRATORY BURST OXIDASE HOMOLOG D (RbohD) and RbohF is essential for the Na+/K<sup>+</sup> homeostasis in Arabidopsis (Ma et al., 2012), and ethylene-induced ROS production through transcriptional regulation on AtRbohF confers enhanced salt tolerance to the ethylene overproduced mutant eto1 (Jiang et al., 2013). On the other hand, ethylene signaling component EIN3/EIL1 activates ROS-scavenging gene expression to deter excess ROS accumulation and to increase salt tolerance (Peng et al., 2014). Similarly, the effects of ethylene signaling downstream factors on ROS are inconsistent during different stages of various stresses. For example, ERF74 promotes ROS burst in the early stages of various stresses through the regulation of gene expression of RbohD, followed with induction of ROS-scavenging-related genes (Yao et al., 2017). However, ethylene inducible factor TERF1 improves stress tolerance through reduced ROS content (Zhang H. et al., 2016). Therefore, fine-tuning of ethylene biosynthesis and signaling on ROS homeostasis are critical for salt tolerance.

Additionally, our previous studies verified several downstream regulators of ethylene signaling in salt response and ROS homeostasis. For example, ETHYLENE AND SALT INDUCIBLE ERF GENE 1 (ESE1), a direct target gene of EIN3, positively regulates salt tolerance and coordinates with EIN3 to activate downstream salt-related gene expression in Arabidopsis (Zhang et al., 2011); and JERF3, an ethylene-induced gene, enhances salt tolerance via direct modulation on the gene expressions of SUPEROXIDE DISMUTASE (SOD) and CARBONIC ANHYDRASE (CA) in tomato to eliminate ROS, which also confers drought and osmotic stress tolerance to transgenic rice with heterologous expression of JERF3 (Wu et al., 2008; Zhang et al., 2010; **Table 1**). Thus, identification of more ethylene signaling downstream regulators participating in ROS homeostasis under salt stress is necessary for elucidating the regulation of ethylene on ROS and salt response.

#### The Scavenging Role of AsA on ROS Homeostasis Contributes to Salt Tolerance

AsA, also known as vitamin C, is a low molecular weight antioxidant, functioning as a component of non-enzymatic scavenging of ROS in plant growth and stress tolerance (Smirnoff, 2000; Conklin, 2004; Akram et al., 2017). It has been reported that AsA improves salt tolerance in various species, including rice, potato, tomato, and citrus (Shalata and Neumann, 2001; Hemavathi et al., 2010; Kostopoulou et al., 2015; Qin et al., 2016). The L-galactose pathway is the main pathway of AsA biosynthesis in plants, and most of the genes in this pathway have been identified (Bulley and Laing, 2016). Investigations


TABLE 1 | Genes involved in ethylene- and AsA-mediated ROS homeostasis in response to salt stress.

also have elucidated the regulation via the L-galactose pathway of AsA biosynthesis, including the modulations on the AsA biosynthesis enzyme activities and stabilities at transcriptional and translational levels (Laing et al., 2015). One of these regulators is calmodulins-like 10, which interacts with AsA biosynthesis enzyme phosphomannomutase (PMM) to modulate enzyme activities and AsA pool (Cho et al., 2016), which suggested the role of calcium (Ca2+) in AsA biosynthesis. It has been known that Ca2<sup>+</sup> signaling is triggered by ROS accumulation (Rentel and Knight, 2004) and Ca2<sup>+</sup> wave is induced under salt stress (Choi et al., 2014; Liu et al., 2018). A chloroplast protein, QUASIMODO1 (QUA1), functions upstream of a thylakoid-localized Ca2<sup>+</sup> sensor, CAS, to mediate Ca2<sup>+</sup> signaling under salt stress (Zheng et al., 2017). Additionally, AsA could trigger increase of cytosolic Ca2<sup>+</sup> in Arabidopsis as a signaling molecule (Makavitskaya et al., 2018), suggesting the association between Ca2<sup>+</sup> sensor and AsA-mediated ROS scavenging during salt responses, and feedback regulation of Ca2<sup>+</sup> signaling and ROS homeostasis.

The regulation factors of AsA biosynthesis also play a role in salt response. Our previous investigations have found that ethylene-induced factor AtERF98 enhances salt tolerance due to transcriptional activation on gene expressions of AsA biosynthesis enzymes, especially direct binding to the promoter of a key enzyme of AsA biosynthesis encoding gene VTC1 (Zhang et al., 2012). Meanwhile, we also identified the posttranscriptional modulation of COP9 SIGNALOSOME SUBUNIT 5B (CSN5B) on VTC1 in Arabidopsis (Wang J. et al., 2013; Li et al., 2016), elucidating a mechanism of light/dark effects on AsA contents. Loss-of-function mutant csn5b, with more AsA content and less ROS pool, displays increased salt tolerance, suggesting the positive regulation of AsA on salt response. Recent studies showed that salt induced zinc-finger protein SIZF3, which interferes with the interaction between CSN5B and VTC1, simultaneously promotes AsA accumulation and enhances salt tolerance (Li et al., 2018; **Table 1**). Thus, increasing AsA content is a potential approach for improving plant salt tolerance.

#### The Integration of Ethylene in AsA Production Finetunes ROS Homeostasis Under Salt Stress

As discussed above, both ethylene and AsA could enhance salt tolerance via regulation of ROS homeostasis. Previous reports have indicated that ethylene in many cases maintains a low level of ROS contents under salt stress through the enzymatic pathway (Wu et al., 2008; Peng et al., 2014; Zhang W. et al., 2016). Moreover, the non-enzymatic pathway of scavenging ROS also participates in ethylene-mediated salt response, such as AtERF98, suggesting that the modulation of ethylene on ROS elimination is alternatively dependent on non-enzymatic antioxidant (Zhang et al., 2012).

There are many environmental factors affecting AsA biosynthesis, such as light (Fukunaga et al., 2010), circadian rhythm (Dowdle et al., 2007), and high temperature (Richardson, 2004). CSN5B, identified in our previous studies (Wang J. et al., 2013), is a subunit of photomorphogenic COP9 signalosome (Gusmaroli et al., 2004), which acts together with COP1, COP10, and DET1 to repress photomorphogenesis (Yanagawa et al., 2004). This research suggest that CSN5B-regulated AsA biosynthesis is a part of photomorphogenesis. Intriguingly, ethylene has functions in COP1 nucleocytoplasmic partitioning (Yu et al., 2013, 2016), indicating a possible link between ethylene and light-regulated AsA biosynthesis. It was reported that ABA-INSENSITIVE 4 (ABI4) mediates AsA-regulated plant growth (Kerchev et al., 2011) and ethylene production via transcriptional

protein levels, activating gene expression of EIN3 direct binding targets (ESE1, SIED1, and POD) and ethylene response factors (JERF3 and ERF98) to regulate salt tolerance via ROS scavenging. ERF98 positively regulates salt tolerance via transcriptional activation of AsA biosynthesis gene VTC1. Moreover, CSN5B, a subunit of photomorphogenic COP9 signalosome, contributes to AsA biosynthesis and salt responses due to modulation on VTC1 degradation. SIZF3 also confers salt tolerance through mediating the interaction between CSN5B and VTC1. This research indicates that ROS accumulation under salt stress could be eliminated through enzymatic and non-enzymatic pathways, in both of which ethylene signaling is involved. However, the understanding of ethylene roles in AsA biosynthesis is yet limited. ABI4 negatively regulates ethylene synthesis and AsA production, which supply a possible mechanism coordinating ABA and ethylene to regulate AsA biosynthesis under salt stress. Ca2<sup>+</sup> signaling could be induced by both ROS signaling and participates in AsA biosynthesis modulation through PMM. Arrows and lines with bars indicate activation and inhibition, respectively. Dotted lines indicate indirect regulations.

repression of ACS in Arabidopsis (Dong et al., 2016). In this regard, ethylene seems to have crosstalk with ABA to modulate AsA production (**Figure 1**). However, the mechanisms for these modulations are yet to be elucidated.

# CONCLUSION AND PERSPECTIVES

Emerging evidence provides the understanding of the roles of ethylene and AsA in salt tolerance through fine-tuning ROS homeostasis. Ethylene biosynthesis could be induced under salt stress, followed with ROS accumulation through transcriptional activation of Rbohs gene expression, in which ROS functions as a signal to regulate Na+/K<sup>+</sup> homeostasis. Excessive ROS is toxic to plants, and ethylene also performs a scavenging role on ROS homeostasis under salt stress through signaling pathways, including the stability of EBF1/EBF2 and transcriptional regulation of EIN3/EIL1 on downstream direct or indirect regulators such as ESE1, SIED1, POD, JERF3, and ERF98. Ca2<sup>+</sup> signaling, as a second messenger, could be induced by both ROS and AsA-mediated ROS balance, and participates in AsA biosynthesis modulation (**Figure 1**). The antagonistic effect of ethylene on ROS synthesis and scavenging under salinity stress is due to different functions of ROS at different developmental stages and in different tissues (Jiang et al., 2013; Peng et al., 2014). Additionally, non-enzymatic antioxidant AsA and the modulators involved in AsA biosynthesis confer to salt tolerance through reduced ROS accumulation (**Figure 1**; Wang J. et al., 2013; Qin et al., 2016; Li et al., 2018). However, the individual or crosstalk of ethylene and AsA regulation mechanisms on salt responses remain in need of further research. For example, although the components of the ethylene signaling pathway are conserved in Arabidopsis and rice (Yin et al., 2017), the underlying mechanisms of ethylene signaling in response to salt stress are different. Similarly, the effect of ethylene on plant growth is opposite in light and dark, such as hypocotyl elongation (Yu and Huang, 2017). Moreover, emerging research demonstrates that light plays a pivotal role in AsA synthesis

(Fukunaga et al., 2010; Wang J. et al., 2013). These findings suggest a complex network regulated by ethylene signaling under different growth conditions. Further engagement is needed to determine whether ethylene and light coordinate AsA production to maintain ROS homeostasis during salt response. Furthermore, it is widely recognized that ABA and ethylene are simultaneously involved in stress responses (Kumar et al., 2016). ABA signaling component ABI4 mediates AsAregulated plant growth (Kerchev et al., 2011) and inhibits ethylene biosynthesis (Dong et al., 2016; **Figure 1**). Nevertheless, the crosstalk between ethylene and ABA in the control AsA pool is unclear. Proper redox homeostasis is necessary for plant growth under salt stress; thus, making clear the detailed mechanisms of ethylene and AsA in maintaining ROS homeostasis will provide new insights for salt-tolerant genetic improvement.

#### REFERENCES


### AUTHOR CONTRIBUTIONS

RH proposed the concept. JW organized and drafted the manuscript. RH contributed to the editing of the manuscript. Both authors read and approved the manuscript.

# FUNDING

This work was supported by the National Key Research and Development Program of China (2016YFD0100604), the National Natural Science Foundation of China (31470366), Agricultural Science and Technology Innovation Program of the Chinese Academy of Agricultural Sciences, and the Fundamental Research Funds for Central Non-profit Scientific Institution (1610392018003).


ripening-associated ERF genes and their link to key regulators of fruit ripening in tomato. Plant Physiol. 170, 1732–1744. doi: 10.1104/pp.15.01859


negatively affect salt tolerance in rice. Plant Physiol. 169, 148–165. doi: 10.1104/ pp.15.00353


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

Copyright © 2019 Wang 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) 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.

# Salinity and ABA Seed Responses in Pepper: Expression and Interaction of ABA Core Signaling Components

Alessandra Ruggiero1,2, Simone Landi<sup>1</sup> , Paola Punzo<sup>1</sup> , Marco Possenti<sup>3</sup> , Michael J. Van Oosten<sup>2</sup> , Antonello Costa<sup>1</sup> , Giorgio Morelli<sup>3</sup> , Albino Maggio<sup>2</sup> , Stefania Grillo<sup>1</sup> and Giorgia Batelli<sup>1</sup> \*

#### Edited by:

Abel Rosado, University of British Columbia, Canada

#### Reviewed by:

Jorge Lozano-Juste, Instituto de Biología Molecular y Celular de Plantas (IBMCP), Spain Simone Diego Castellarin, University of British Columbia, Canada

#### \*Correspondence:

Giorgia Batelli giorgia.batelli@ibbr.cnr.it; giorgiabatelli@gmail.com

#### Specialty section:

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

Received: 31 October 2018 Accepted: 25 February 2019 Published: 19 March 2019

#### Citation:

Ruggiero A, Landi S, Punzo P, Possenti M, Van Oosten MJ, Costa A, Morelli G, Maggio A, Grillo S and Batelli G (2019) Salinity and ABA Seed Responses in Pepper: Expression and Interaction of ABA Core Signaling Components. Front. Plant Sci. 10:304. doi: 10.3389/fpls.2019.00304 <sup>1</sup> National Research Council of Italy, Institute of Biosciences and Bioresources (CNR-IBBR), Reaserch Division Portici, Portici, Italy, <sup>2</sup> Department of Agriculture, University of Naples "Federico II", Portici, Italy, <sup>3</sup> Council for Agricultural Research and Economics, Research Centre for Genomics and Bioinformatics (CREA-GB), Rome, Italy

Abscisic acid (ABA) plays an important role in various aspects of plant growth and development, including adaptation to stresses, fruit development and ripening. In seeds, ABA participates through its core signaling components in dormancy instauration, longevity determination, and inhibition of germination in unfavorable environmental conditions such as high soil salinity. Here, we show that seed germination in pepper was delayed but only marginally reduced by ABA or NaCl with respect to control treatments. Through a similarity search, pepper orthologs of ABA core signaling components PYL (PYRABACTIN RESISTANCE1-LIKE), PP2C (PROTEIN PHOSPHATASE2C), and SnRK2 (SUCROSE NONFERMENTING1 (SNF1)-RELATED PROTEIN KINASE2) genes were identified. Gene expression analyses of selected members showed a low abundance of PYL and SnRK2 transcripts in dry seeds compared to other tissues, and an upregulation at high concentrations of ABA and/or NaCl for both positive and negative regulators of ABA signaling. As expected, in hydroponically-grown seedlings exposed to NaCl, only PP2C encoding genes were up-regulated. Yeast two hybrid assays performed among putative pepper core components and with Arabidopsis thaliana orthologs confirmed the ability of the identified proteins to function in ABA signaling cascade, with the exception of a CaABI isoform cloned from seeds. BiFC assay in planta confirmed some of the interactions obtained in yeast. Altogether, our results indicate that a low expression of perception and signaling components in pepper seeds might contribute to explain the observed high percentages of seed germination in the presence of ABA. These results might have direct implications on the improvement of seed longevity and vigor, a bottleneck in pepper breeding.

Keywords: Capsicum annuum L., abscisic acid, PYR/PYL/RCARs, PP2Cs, SnRK2s, seed germination, seed viability

# INTRODUCTION

fpls-10-00304 March 15, 2019 Time: 17:20 # 2

Pepper (Capsicum spp.) is an economically important genus of the Solanaceae family, with a global production of 38.4 million tons including green fruits and dried pods, harvested from 3.7 million hectares in 2016<sup>1</sup> . In sweet pepper (Capsicum annuum L.), seed vivipary and short longevity are two major challenges of commercial production (Marrush et al., 1998; Lanteri et al., 2000; De, 2004; Sano et al., 2016).

Abscisic acid (ABA) plays a major role in the adaptation to stresses during the vegetative phase, as well as in the establishment of seed dormancy, seed longevity, and inhibition of germination under unfavorable conditions (Zhu, 2002; Finkelstein, 2013; Ruggiero et al., 2017). Seed germination starts with the uptake of water by imbibition of the dry seed, followed by embryo expansion. The uptake of water is triphasic with a rapid initial uptake (phase I, i.e., imbibition) followed by a plateau phase (phase II). A further increase in water uptake (phase III) happens only when germination occurs, as the embryo axis elongates and breaks through the covering layers, typically the endosperm and the testa (Manz et al., 2005). While in Arabidopsis thaliana testa and endosperm rupture occur simultaneously, in many Solanaceae seeds these two events are temporally distinct (Petruzzelli et al., 2003). Mechanical resistance from testa and endosperm dormancy appears to be the cause of nondeep physiological dormancy in seed model systems such as A. thaliana and Solanaceae species (Leubner-Metzger, 2003). Enzymes that facilitate testa rupture can be released by the endosperm and/or the radicle. β-1,3-glucanases (βGlu) facilitate endosperm rupture by breaking intercellular adhesion and causing cell separation. In pepper, the accumulation of βGlu occurs prior to radicle emergence. Unfavorable osmotic potentials, darkness and ABA inhibit endosperm rupture and βGlu accumulation in the micropylar cap of the seeds, the site of radicle emergence (Petruzzelli et al., 2003).

In non-dormant seeds, exogenous ABA also inhibits the transition from water uptake phase II to III and late embryo cell expansion, but does not affect phase I and II and testa rupture (Müller et al., 2006).

In Arabidopsis, ABA inhibition of germination is mediated by members of the core ABA signaling pathway as well as by ABFtype transcription factors.

Components of this signaling pathway have been isolated in the past decades, and include the PYRABACTIN RESISTANCE (PYR)/PYL (PYR1-LIKE)/REGULATORY COMPONENT OF ABA RESPONSE (RCAR), (hereafter referred to as PYLs), PROTEIN PHOSPHATASE2C (PP2C), and SNF1- RELATED PROTEIN KINASE2 (SnRK2) (Ma et al., 2009; Park et al., 2009). In the absence of ABA, clade A PP2Cs, a family of major negative regulators of ABA responses, bind to and inhibit SnRK2 kinases. The association keeps the kinases inactive by blocking their catalytic cleft and by dephosphorylating the activation loop (Soon et al., 2012). In response to environmental or developmental cues, ABA is perceived by PYL receptors and induces closure of two highly conserved β-loops that function as a gate and latch, and lock the pocket, creating a binding surface for the PP2Cs (Cutler et al., 2010; Moreno-Alvero et al., 2017). A conserved tryptophan residue in the PP2C inserts directly between the gate and latch, and functions to further lock the receptor in a closed conformation (Melcher et al., 2010; Santiago et al., 2012; Moreno-Alvero et al., 2017). SnRK2 kinases are thus released from PP2C inhibition, and can then phosphorylate many downstream effectors (Fujii et al., 2009), including the basic leucine-zipper transcription factors such as ABI5 (ABSCISIC ACID INSENSITIVE 5). A double snrk2.2/2.3 mutant displayed highly ABA-insensitive germination and increased seed dormancy (Fujii et al., 2007). Seed germination was resistant to 50-100 µM ABA in a sextuple pyr1/pyl1/2/4/5/8 mutant (Gonzalez-Guzman et al., 2012). Imbibed abi5 mutant seeds were able to perform the first stages of seed germination, but were arrested before the radicle penetration of the inner testa and endosperm (Lopez-Molina et al., 2002).

Genomic tools are now available to analyze the molecular details of these phenomena in pepper (Kim et al., 2014; Qin et al., 2014; Kim et al., 2017; Hulse-Kemp et al., 2018). The elucidation of the core ABA signaling pathway in model systems and the identification of key secondary regulatory proteins allows this knowledge to be applied to crop species (Klingler et al., 2010; Hou et al., 2016). Previous work has already characterized core components in Solanum lycopersicum (Gonzalez-Guzman et al., 2014; Chen et al., 2016) and a pair of pepper PP2C/PYL proteins has been previously functionally analyzed (Lim and Lee, 2016). Different studies have shown the applications of ABA signaling core components to enhance plant stress tolerance, through either genetic engineering or chemical approaches in crops (Okamoto et al., 2013; Rodriguez and Lozano-Juste, 2015; Zhang et al., 2016; Vaidya et al., 2017). Pepper presents a potential practical application of these methods for genetic improvement that can directly benefit breeders, producers, and consumers.

Here, we report on ABA and NaCl sensitivity in pepper seeds and plants as well as the identification of components of the C. annuum PYL, PP2C, and SnRK2 gene families. We have characterized pepper's sensitivity to ABA and NaCl at the germination stage by evaluating gene expression and the potential for in vivo interaction among the ABA signaling components. We performed yeast two hybrid assays among putative pepper core components and with Arabidopsis orthologs to confirm the ability of these proteins to function in the ABA signaling cascade. We also verified interactions in protoplasts using a bi-molecular florescence assay.

#### MATERIALS AND METHODS

#### Plant Material

Capsicum annuum L. Quadrato D'Asti giallo (2480) seeds were provided by "S.A.I.S. Spa" (Italy), while seeds of the genotypes Corno di Toro rosso (QSB294TS), (refer to as

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

Corno rosso), Corno di Toro giallo (QSB296TS) (refer to as Corno giallo), Friariello (Nocera selection, PGM283), Nocera rosso (Japanese selection, QPA1909TS), Nocera giallo (Japanese selection, QPA1310TS), and Marconi rosso (R15020) were kindly gifted by "La Semiorto Sementi s.r.l." (Italy).

For germination analyses, seeds were sown on solid MS medium (1X MS Salts including vitamins, 15 g/L Plant agar, pH 5.7) in the presence of ABA (1, 5, and 10 µM) or NaCl (25, 50, and 100 mM). Germination was scored daily in terms of radical emergence and fully expanded cotyledons.

For seedling sensitivity to ABA and NaCl, seedlings were transferred from germination media after 9 days and were grown on vertical plates on solid MS medium (1X MS Salts including vitamins, 5 g/L Sucrose, 15 g/L Plant agar, pH 5.7) in the presence of 100 mM NaCl or 20 µM ABA. Root length was scored every 2 days, shoot weight and photographs were taken after 5 days of growth.

In hydroponic culture, pepper seedlings were grown in solution containing: 1.5 mM Mg(NO3)2·6H2O, 3.4 mM Ca(NO3)2·4H2O, 1 mM KNO3, 1.8 mM K2SO4, 1.5 mM KH2PO4, and 14 mg/L Hidromix (Valagro, Italy), pH 6.2. The solution was changed weekly. After 17 days of culture, NaCl 200 mM was added to the solution to impose salt stress. After 3 h of salt treatment, shoots and roots of each condition were collected separately.

#### RNA Isolation, cDNA Synthesis, and qRT-PCR

Total RNA was extracted from seeds, shoots and roots (100 mg) using RNeasy <sup>R</sup> Plant Mini kit (Qiagen, Germany) following the manufacturer's instructions. RNA quantity was measured spectrophotometrically by NanoDrop ND-1000 Spectrophotometer (NanoDropTechnologies, United States), and integrity was verified on a denaturing agarose gel. One microgram of total RNA was DNase-treated and reverse transcribed using QuantiTect <sup>R</sup> Reverse Transcription kit (Qiagen, Germany) according to manufacturer's instructions. For gene expression analyses, the complementary DNA was diluted 1:20 and 2 µL of diluted cDNA were used for each qRT-PCR reaction, performed with 6.25 µL of 1X Platinum <sup>R</sup> SYBR <sup>R</sup> Green qPCR SuperMix (Thermo Fisher Scientific, United States) and 1.75 µL of primer mix (4.28 µM) in a 12.5 µL PCR reaction. Primers used are listed in **Supplementary Table S1**. Reactions were performed with ABI 7900 HT (Applied Biosystems, United States). Cycling conditions were: 10 min at 95◦C, followed by 40 cycles of 95◦C for 15 s and 60◦C for 1 min. Three or four biological replicates per treatment, each with three technical replicates were tested. PCR product melting curves were analyzed to confirm the presence of a single peak, indicative of one PCR product per primer couple assayed. For relative quantification of gene expression, Capsicum annuum eukaryotic initiation factor 5A2 (EIF5A2) (Acc.no. AY484392) was used as endogenous reference since its expression was found stable in all the analyzed tissues and treatments as also reported by Wan et al. (2011). Quantification of gene expression was carried out using the 2 <sup>−</sup>11Ct method (Livak and Schmittgen, 2001) and reported as relative expression levels, compared to control conditions as internal calibrator.

For absolute qRT-PCR total RNA was extracted from dry seeds, germinating seeds (i.e., seeds incubated for 4 days on control medium), shoots and roots from seedlings grown in hydroponic system in control condition. RNA isolation, cDNA Synthesis and qRT-PCR were performed as already mentioned above. All primer pairs (reported in **Supplementary Table S1**) were tested by PCR. A single product of the correct size for each gene was confirmed by agarose gel electrophoresis. The amplified fragment of each gene was subcloned into the pGEM <sup>R</sup> -T Easy vector (Promega, United States) and used to generate standard curves by serial dilutions. Results were analyzed using the ABI PRISM 7900HT Sequence Detection System, Version 2.3. Analysis of variance (ANOVA) on absolute qRT-PCR data was carried out using the SPSS software package (SPSS 19 for Windows, SPSS Inc., an IBM Company, United States). When ANOVA indicated that a single factor or their interaction was significant, mean separation was performed using the Duncan's multiple range test at p < 0.05 on each of the significant variables measured.

#### Yeast Two-Hybrid Assay

For yeast two-hybrid experiments, the prey plasmid pGADT7 and the bait plasmid pGBKT7 (Clontech, United States) were used. The full-length coding sequence of CaABI and CaHAI were PCR amplified using RNA prepared from seeds and cloned in frame into pGADT7 between EcoRI and XhoI, SmaI and XhoI restriction sites, respectively. The full-length coding sequence of CaPYL2 and CaPYL4 were PCR amplified using RNA prepared from leaves and cloned in frame with the GAL4 binding domain (BD) of pGBKT7 digested with SmaI and PstI. The full-length coding sequence of CaPYL8, CaSnRK2.3, and CaSnRK2.6 were PCR amplified using RNA prepared from seeds and cloned in frame into pGBKT7 between EcoRI and BamHI, EcoRI and SalI, EcoRI, and BamHI restriction sites, respectively. Plasmids were sequenced to rule out PCR-induced mutations. Primers used for PCR amplification of the mentioned genes are listed in **Supplementary Table S2**. The bait and prey plasmids containing Arabidopsis thaliana genes were previously described (Park et al., 2009; Hou et al., 2016).

The bait and prey plasmids were transformed into yeast strain AH109 (Clontech, United States) using the Lithium acetate/Polyethylene glycol method (Bai and Elledge, 1997). The self-activation test was performed prior to the testing of combinations of interest as reported in Docimo et al. (2016). After verifying that the bait and prey plasmids were not showing self-activation, co-transformations to verify interactions were performed. Transformed colonies containing bait and prey plasmids were selected on synthetic drop-out medium lacking tryptophan and leucine (-W/-L). Co-transformants were grown overnight in liquid culture lacking tryptophan and leucine (-W/- L). For the interaction between bait and prey, an equal amount of cells was spotted on medium lacking tryptophan, leucine, histidine and adenine (-W/-L/-H/-A). Positive and negative controls were also performed as indicated in the figure legend.

# Bimolecular Fluorescence Complementation Assay

fpls-10-00304 March 15, 2019 Time: 17:20 # 4

The CDS of CaPYL2 and CaPYL4 were fused downstream of N-terminal region of YFP, while CaHAI was fused downstream of the C-terminal region of YFP, using pUGW0 vectors (Nakagawa et al., 2007). Leaf protoplasts were prepared and transformed according to Pedrazzini et al. (1997), using 3 weeks old N. tabacum plants. DNA (40 µg of each construct) was introduced into 1 × 10<sup>6</sup> protoplasts by PEG-mediated transfection. After 16 h incubation in the dark at 25◦C, each interaction was split in two and one was treated with 50 µM of ABA. After the incubation time, YFP fluorescence in protoplast cells was detected by confocal microscopy.

# Confocal Imaging

Confocal microscopy analyses were performed on an Inverted Z.1 microscope (Zeiss, Germany) equipped with a Zeiss LSM 700 spectral confocal laser-scanning unit (Zeiss, Germany). Samples were excited with a 488 nm, 10 mW solid laser with emission split at 505 nm for YFP.

# Bioinformatics

Sequences of pepper (Capsicum annuum) ABA receptors (PYL/PYR/RCAR), PP2Cs and SnRK2s were found using a multiple database search to identify potential members of these families. Tomato (Solanum lycopersicum) and Arabidopsis thaliana sequences were previously identified by other authors (Schweighofer et al., 2004; Gonzalez-Guzman et al., 2014) and obtained at https://solgenomics.net and https://www.arabidopsis. org, respectively. Pepper sequences were obtained using a BLAST P approach at http://pepperhub.hzau.edu.cn/pegnm/ database, using each A. thaliana and tomato sequence as queries. The obtained hits were filtered using a blast score cut-off ≥ 150. Alignments and phylogenetic analyses were performed using the software MEGA version 6 (Tamura et al., 2013). Sequence alignment was achieved using the MUSCLE algorithm, using a maximum number of interactions equal to 32. Phylogenetic trees were constructed using the maximum likelihood method with the substitution JTT model gamma distributed. The test of phylogeny was performed using the bootstrap method with a number of replications equal to 100.

For cloned PP2Cs, alignments of multiple amino acid sequences were carried out using Clustal W<sup>2</sup> . The alignment results were marked using BOXSHADE 3.21 software<sup>3</sup> .

# RESULTS

#### Pepper Seed Germination in Presence of ABA and NaCl

To verify the sensitivity of Capsicum annuum to abiotic stress and ABA treatments at the seed stage, we scored the germination percentage of commercial seeds of Quadrato D'Asti giallo (QA),

<sup>2</sup>https://www.genome.jp/tools-bin/clustalw

<sup>3</sup>http://www.ch.embnet.org/software/BOX\_form.html

a high yielding variety among the most commonly cultivated for marketing in Italy and abroad. The germination rates in terms of radicle emergence and cotyledon expansion of seeds in presence of ABA or NaCl were analyzed in detail (**Figure 1**). After 15 days, seeds treated with 1 or 5 µM ABA reached virtually 100% germination, while those treated with 10 µM ABA showed 62% seed germination (**Figure 1A**). The seeds incubated with 10 µM ABA had complete inhibition of cotyledon expansion, while the presence of 5 µM ABA resulted in a 6 day delay compared to controls or 1 µM ABA (**Figure 1B**). The radicle emergence was not affected by salt treatments except for 100 mM NaCl, which caused a 1 day delay of germination compared to no or lower NaCl concentrations (**Figure 1C**). Sodium chloride also delayed cotyledon expansion. At 100 mM NaCl, only 63% of the seedlings had fully expanded cotyledons after 15 days of incubation, while at 0, 25, and 50 mM 90% of the cotyledons were fully expanded (**Figure 1D**). The delayed cotyledon expansion caused by the highest concentrations of ABA and NaCl was also visible early as 8 days (**Figure 1E**). Therefore, the presence of NaCl or ABA had little effect on QA seed germination at low and medium concentrations with a clear effect only at high concentrations. A similar behavior was observed in other tested varieties, which showed significant germination reduction at 10 µM ABA and 100 mM NaCl (**Supplementary Figure S1**).

We also tested response of seedlings to ABA and NaCl to verify sensitivity at different developmental stages. We therefore scored root growth of QA seedlings germinated on control media and subsequently transferred to plates containing NaCl or ABA for 5 days (**Figure 2**). As shown in **Figures 2A,B**, root growth was inhibited at 20 µM ABA. In particular, 2 days after transfer, the seedlings grown in presence of 20 µM ABA showed a 16% increase of the initial root length compared to a 33% increase in 100 mM NaCl and a 40% increase in plants grown in the control medium. After 5 days of incubation, there was a 30% difference in root growth between seedlings grown on ABA and no or NaCl treatments. Measurements of shoot weight at the end of the experiment allowed for a discrimination of the treatments, showing a reduction in plants treated with NaCl 100 mM or ABA 20 µM compared to controls (**Figure 2C**).

#### Capsicum annuum PYL, PP2C, and SnRK2 Identification

Using annotated tomato and Arabidopsis protein sequences, we identified putative orthologs in pepper genome (C. annuum L\_Zunla-1, Qin et al., 2014) of ABA signal transduction core components: PYR/PYL/RCAR (PYLs), clade A PP2Cs and SnRK2s (**Figure 3**). A high bootstrap was observed for the three analyzed families, with C. annuum homologs showing a closer similarity to the tomato counterparts. Ten PYL genes (**Figure 3A**) were identified and clustered in the classical three subfamilies (Ma et al., 2009), with the exception of Capana02g001761, which clustered with Solyc02g076770. A multiple aminoacid sequence alignment of Ca- and AtPYLs showed that functional residues in the conserved loops (CL1-CL4) were well conserved (**Supplementary Figure S2**).

Seven genes encoding putative clade A PP2Cs showed a partition in 2 subfamilies (**Figure 3B**); the first included Capana06g000398, Capana05g002193 and Capana03g002491, clustering with Arabidopsis AIP1, HAI3, HAI1, AHG3, and AHG1; Capana03g000145, previously characterized as CaADIP1 grouped with Arabidopsis ABI1, ABI2, HAB2, and HAB1, as previously shown (Lim and Lee, 2016) and with Capana07g000875, Capana12g000483, and Capana08g000504. We focus our studies on three CaPP2Cs representatives: Capana03g002491, Capana06g000398, and Capana07g000875 (circled in red, **Figure 3B**). A multiple alignment of Capana03g002491 deduced amino acid sequences with AtAHG1, Solyc03g006960, AtABI1 and AtHAB1 showed that most of the functional residues were well conserved, including those necessary for the Mn/Mg++ interaction and the interaction with PYLs (**Supplementary Figure S3**). Similarly to AtAHG1, Capana03g002491 lacks the tryptophan that in other clade A PP2Cs participates to the binding with ABA (W385 in AtHAB1, W300 in AtABI1), which in Arabidopsis appears to be a unique feature of AHG1. Thus, we named Capana03g002491 as CaAHG1 (**Figure 3B**).

To study protein interactions, Capana06g000398 and Capana07g000875 were isolated and cloned using RNA extracted from seeds. The proteins encoded by the cloned coding sequences showed only a few conservative amino acid replacements with similar properties (**Supplementary Figure S4**). A notable exception is the substitution of a phenylalanine involved in the Van der Waal contacts with PYLs (F306 in ABI1, Yin et al., 2009), which in sequences from Solanaceous plants is replaced by a serine (S), indicating possible modifications in the binding with regulatory PYLs. Furthermore, the cloned Capana07g000875, which corresponds to XM\_016724784.1, one of the four predicted splicing variants of this gene, has a shorter protein sequence compared to other PP2Cs aligned, therefore it lacks two conserved D residues required for the Mn/Mg++ ions interaction (**Supplementary Figure S4**). On the basis of protein homology and conserved functional residues we named Capana06g000398 as a putative CaHAI and Capana07g000875 as

CaABI (circled in red, **Figure 3B**). Nine genes encoding SnRK2s were clustered in the three classical sub-families (**Figure 3C**, Umezawa et al., 2010), and were all already annotated in the genome. Similarly to tomato, two genes encoding subfamily III SnRK2s appear to be present in the genome, compared to three genes in Arabidopsis. Alignments of the domain II in the C-terminal region of the SnRK2 proteins separate pepper SnRK2s in SnRK2a and SnRK2b subfamilies (**Supplementary Figure S5**).

# PYL, PP2C, and SnRK2 Gene Expression in Different Tissues

For gene expression analyses, we selected three PYLs, one for each subfamily (CaPYL2, CaPYL4, CaPYL8), three PP2Cs from different subclades (CaABI, CaAHG1, CaHAI), the two SnRK2s from subclass III (CaSnRK2.3, CaSnRK2.6) and one from subclass I (CaSnRK2.4) (**Figure 3**, red circles).

To verify organ specific expression, we performed absolute quantification through qRT-PCR in dry seeds, germinating seeds, shoots and roots (**Figure 4**). CaPYL2 showed low expression in seeds, both dry and germinating, and roots, with highest levels of expression detected in shoots (about 10,000 copies/µL cDNA). CaPYL4 had low expression in dry seeds (about 400 copies/µL cDNA), higher in germinating seeds and shoots, and the highest expression among the tested CaPYLs in roots (about 48,000 copies/µL cDNA) (**Figure 4A**). Regarding expression of PP2Cs in dry seeds, CaABI and CaHAI showed the lowest (about 470 copies/µL cDNA) and highest (about 10,000 copies/µL cDNA) amount, respectively (**Figure 4B**), while similar expression values in the other organs were detected. All three selected SnRK2 genes showed low expression in dry seeds. CaSnRK2.4 had low expression also in germinating seeds (around 400 copies/µL cDNA) and roots (around 800 copies/µL cDNA) while CaSnRK2.6 showed the highest steady state level (21,000 copies/µL cDNA) in shoots (**Figure 4C**).

# Regulation by NaCl and ABA of PYL, PP2C, and SnRK2 Expression

We studied the expression of selected genes in QA seeds incubated for 4 days with ABA (1, 5, 10 µM) or NaCl (25, 50, 100 mM). ABA or salt treatments resulted in up-regulation of positive ABA signaling components such as PYLs and SnRK2s (**Table 1**). In particular, CaPYL4 expression was significantly upregulated by the highest concentration of ABA, whereas CaPYL8 was upregulated by all NaCl treatments. CaPYL2 expression level was below detection. CaSnRK2.6 was up-regulated in all tested treatments except for the lowest ABA concentration, while

CaSnRK2.3 was induced by ABA 10 µM, NaCl 50 and 100 mM. CaSnRK2.4 was up-regulated by ABA 10 µM and all NaCl treatments. CaHAI was the only PP2C gene up-regulated by ABA and all NaCl treatments while CaAHG1 was up-regulated only by ABA 1 µM. We also tested shoot and root expression of QA seedlings grown in hydroponic solution and subjected to 3 h salt stress treatments (**Table 2**). In this case, the three tested PP2C genes were significantly up-regulated in both tissues. By contrast, positive regulators were less responsive; CaSnRK2.6 was the only SnRK2 up-regulated in shoot. Notably, CaPYL4 was strongly down-regulated after NaCl treatment in both shoots and roots.

#### Interaction Between PYLs, PP2Cs, and SnRK2s

The key function of ABA receptors upon ABA binding is their ability to interact with and inhibit PP2Cs, releasing SnRK2s from inhibition (Fujii et al., 2009; Park et al., 2009). We therefore tested whether selected CaPYLs, CaPP2Cs and CaSnRK2s were capable of interaction with each other or with Arabidopsis members of the ABA signaling cascade in the yeast two hybrid assay. CaPYLs and CaSnRK2s were fused to the GAL4 Binding Domain (BD) and the PP2Cs CaABI and CaHAI to the GAL4 activation domain (AD) and used to test multiple combinations (**Figure 5**). CaPYL4 interacted with AtABI1 and AtPP2CA irrespective of the presence of ABA, while CaPYL2 interaction with AtABI1, AtABI2 and AtPP2CA was only observed in presence of the hormone (**Figure 5A**). The ability of CaHAI to interact with AtPYLs in presence or not of ABA was also tested (**Figure 5B**). Yeast co-transformed with CaHAI and, respectively, AtPYL6, AtPYL7, AtPYL8, AtPYL9, AtPYL11, AtPYL12 and AtPYL13 was able to grow on selective media (–W/–L/–H/–A). Interactions between AtPYL6, AtPYL11 and CaHAI were weak without ABA, but increased markedly when ABA was

present. Furthermore on selective media containing ABA, the yeast co-transformed with AtPYL3/CaHAI, AtPYL4/CaHAI and AtPYL10/CaHAI combinations was able to grow in addition TABLE 2 | Relative quantification of gene expression measured by qRT-PCR of selected genes in roots and shoots of Quadrato D'Asti giallo seedlings grown in hydroponic system after 3 h of 200 mM NaCl treatment.


All data are expressed relative to control treatment as mean ± SD (n = 3). Asterisks represent significance levels using Student t-test; <sup>∗</sup>P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.005. B.D., below detection.

to other combinations mentioned above. The combination AtPYL5/CaHAI showed a weak growth on the selective media with ABA (**Figure 5B**). No interaction was observed for any transformation with the cloned CaABI and each of the AtPYLs (**Supplementary Figure S6**) or CaPYL2/4/8 (**Figure 5C**). In the test between CaPYLs and CaHAI, we observed interaction in presence of ABA with CaPYL2/4/8, while without ABA only the combination CaPYL8/CaHAI showed growth (**Figure 5C**). The interactions between PP2Cs and SnRK2s of C. annuum and Arabidopsis were also analyzed in yeast (**Figures 5D–F**). When we tested combinations between AtSnRK2s and CaPP2Cs, only the yeast co-transformed with the member of subfamily III AtSnRK2.6 and CaHAI was able to grow on the selective media (**Figure 5D**). Similarly, we observed interaction between CaSnRK2.3/2.6 and AtABI1 and AtPP2CA (**Figure 5E**). A positive interaction between CaPP2Cs and CaSnRK2s was observed when yeast was cotransformed with CaHAI and CaSnRK2.3 or CaSnRK2.6, while the cloned CaABI was incapable of binding to SnRK2s in yeast (**Figure 5F**). CaABI did not interact with any of the Arabidopsis or pepper ABA receptors and SnRK2s in yeast (**Figures 5C,D,F** and **Supplementary Figure S6**), although the

TABLE 1 | Relative quantification of gene expression measured by qRT-PCR of selected genes in seeds of Quadrato D'Asti giallo incubated for 4 days on control medium and media with different concentrations of ABA or NaCl.


All data are expressed relative to control treatment as mean ± SD (n = 4). Asterisks represent significance levels using Student t-test; <sup>∗</sup>P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.005.

FIGURE 5 | Interaction test of PP2Cs with PYLs and SnRK2s in the yeast two-hybrid assay. (A) AtPP2Cs fused to the GAL4 Activation Domain (AD) were co-transformed in yeast with CaPYLs cloned in frame with the GAL4 binding domain (BD) of pGBKT7, with combinations shown in figure. (B) CaHAI fused to the AD in pGADT7 was expressed in yeast with AtPYLs in pBDGAL4, the different combinations are shown. (C) CaPP2Cs were fused to the AD in pGADT7 and co-transformed with CaPYLs cloned in pGBKT7. (D) CaPP2Cs in pGADT7 were co-transformed in combination with AtSnRK2s cloned in pGBKT7. (E) AtPP2Cs in pACT2 were co-transformed in yeast with CaSnRK2s in pGBKT7, with combinations shown in figure. (F) Co-transformants of yeast containing CaPP2Cs fused to the AD in pGADT7 in combination with CaSnRK2s in pGBKT7. Co-transformations of the pepper PYL/SnRK2/PP2C orthologs with appropriate complementary empty vectors are shown as negative controls. Yeast cells grown on synthetic media (–W/–L) and on synthetic, selective media without (–W/–L/–H/–A) or, where indicated, with 50 µM ABA (–W/–L/–H/–A+ABA) are shown. Pictures were taken after 3 days of incubation at 30◦C.

protein was produced in yeast, as verified through westernblot by the presence of a signal of the expected molecular weight (**Supplementary Figure S7**). We confirmed in a plant system the interaction between CaHAI and CaPYL2/4 in presence of ABA. We used a split-reporter system in tobacco protoplasts by transient co-expression of CaHAI and CaPYL2/4 fused to complementary fragments of the YFP. A reconstituted fluorescence signal was observed for both interactions only in presence of ABA (50 µM, **Figure 6**), thus confirming the results obtained in yeast. The interaction CaHAI-CaPYL2 was observed within 5 min of incubation with ABA, while a 2 h incubation was allowed to detect the CaHAI-CaPYL4. All negative controls with or without ABA are reported in **Supplementary Figure S8**.

We also investigated the subcellular localization of the proteins CaABI, CaHAI, CaPYL2, and CaPYL4 (**Supplementary Figure S9**). CaABI, CaPYL2, and CaPYL4 showed a nuclear and cytoplasmic localization, while for YFP-CaHAI the fluorescence signal was present only in the nucleus (**Supplementary Figure S9**).

#### DISCUSSION

Seed germination and seedling growth are adversely affected by ABA and NaCl. Capsicum annuum is classified as a moderately salt-sensitive species (De Pascale et al., 2003; Aktas et al., 2006; Penella et al., 2016). Here, we have shown that pepper cv. Quadrato D'Asti giallo (QA) had reduced seed germination and a complete inhibition of cotyledon expansion at 10 µM ABA. Similarly, the presence of 100 mM NaCl caused a 1 day delay in seed germination and reduced cotyledon expansion. Therefore, QA seeds displayed an inhibited germination only at the highest NaCl or ABA concentrations applied (**Figure 1**). At seedling stage, we observed a reduction of root length and shoot weight in seedlings grown in the presence of 100 mM NaCl (**Figure 2**), consistent with previous reports of a reduced vegetative growth in presence of NaCl (Yildirim and Güvenç, 2006).

NaCl-induced inhibition of germination in pepper is mainly due to the osmotic component of salt stress (Chartzoulakis and Klapaki, 2000; Demir and Mavi, 2008), which is counteracted in plants by employing ABA-dependent mechanisms (Cutler et al., 2010; Nakashima and Yamaguchi-Shinozaki, 2013).

Differences in sensitivity to ABA/NaCl between developmental stages/tissues, which may also contribute to explain the known phenomenon of pepper vivipary and low seed longevity, might be at least partially due to the expression/activity of ABA signaling components. Therefore, through a sequence similarity search, we identified putative pepper orthologs of PYLs, clade A PP2Cs and SnRK2s. Reflecting a high degree of conservation of these families among land plants, cluster analysis showed the classical organization in 3 subfamilies for pepper PYLs and SnRK2s. However, similarly to the tomato PYLs, no close relative for the Arabidopsis subgroup AtPYL11/12/13 was found (Gonzalez-Guzman et al., 2014). Capana02g001761, annotated as CaPYL12 like, did not cluster in any of the classical three subfamilies (**Figure 3A**), and was in a clade with the putative tomato ortholog Solyc02g076770, also ungrouped in the study of Gonzalez-Guzman et al. (2014). Capana02g001761, as its tomato ortholog, lacked key aminoacids, including residues of conserved loop 2 (CL2), CL3, and CL4 (**Supplementary Figure S2**). In contrast, presence of these conserved residues, as well as the ABA-dependent interaction with PP2C observed in yeast and protoplasts suggests that CaPYL2 and CaPYL4 are functional ABA receptors.

Additionally, the functional residues or domains were well conserved in the deduced sequences of three selected CaPP2Cs (**Supplementary Figures S3**, **S4**). Similarly to Arabidopsis AHG1 and Solyc03g006960, CaAHG1 lacks the tryptophan (W385 in HAB1, W300 in ABI1) that in other clade A PP2Cs participates to the binding with ABA (Weiner et al., 2010).

The domain II in the C-terminal region of the SnRK2 proteins is critical for PP2C interaction and ABA responsiveness (Kobayashi et al., 2005; Yoshida et al., 2006). Seven pepper SnRK2s (Capana06g001594, Capana01g000732, Capana08g002441, Capana02g003135, Capana05g000287, Capana12g002655, Capana04g000996) belong to subgroup 3 and 2, which in Arabidopsis can be strongly or weakly activated by ABA; while Capana08g001898 and Capana00g002391 cluster with subgroup 1, containing SnRK2s that may not be activated by ABA treatment (**Figure 3C** and **Supplementary Figure S5**; Kulik et al., 2011).

Quantification of gene expression of selected genes in different organs showed differences among family members (**Figure 4**).

CaPYL2 was highly expressed in shoots, while CaPYL4 and CaPYL8 showed highest expression in roots. These data are consistent with previous studies in tomato (Gonzalez-Guzman et al., 2014) and soybean (Bai et al., 2013). Interestingly, differences in transcript levels of the tested PP2Cs were observed only in seeds. CaHAI had highest steady state expression levels in dry seeds. By contrast, CaABI had the lowest expression in dry or germinating seeds and similar expression in all vegetative tissues. Finally, SnRK2 kinases, including the two subclass III members, also had lowest expression in dry seeds. In other species, SnRK2s were shown to be highly expressed in seeds (Fujii et al., 2007; Nakashima et al., 2009; Lou et al., 2017). Considering the established role of these kinases in plant responses to abiotic stress (Zhu, 2016), the low SnRK2 expression in seed suggests an ineffective mechanism of ABA signaling that may affect sensitivity to ABA at the germination stage. However, it is possible that other SnRK2s genes, not tested here, might be highly expressed in seeds. In addition, activity of ABA signaling components is regulated post-translationally through ABA binding, protein interaction and/or phosphorylation/dephosphorylation. Thus, the gene expression in specific organs or in response to treatments only provides an initial indication of the effectiveness of this mechanism, which will have to be verified by measurements of protein activity.

In seeds, 10 µM ABA treatment induced up-regulation of PP2C CaHAI, indicating a possible negative feedback

ABA. Scale bars = 20 µm in all panels.

at high ABA concentrations, as well as of all the tested SnRK2s. ABA-induced up-regulation of SnRK2s has been reported in the case of vegetative tissues, and is thought to derive from a self-amplification mechanism of ABA signal transduction components (Wang et al., 2018). Interestingly, 10 µM ABA treatment also induced CaPYL4 expression, while in vegetative tissues PYLs are commonly down-regulated by ABA (Wang et al., 2018).

In seedlings, CaPYL4 was significantly down-regulated after NaCl treatment in shoots and roots, while CaPYL8 did not change significantly (**Table 2**). Similarly, in tomato seedlings exposure to dehydration caused repression of SlPYL4 expression, while SlPYL8 showed no change in gene expression (Sun et al., 2011). By contrast, all genes within the SlPP2C gene family were significantly up-regulated by dehydration (Sun et al., 2011). In rice, expression of 10 OsPP2C genes in plants treated with 150 mM NaCl was up-regulated (Xue et al., 2008). Consistently, we observed NaCl-induced up-regulation for CaABI, CaHAI, and CaAHG1 (**Table 2**).

Protein-protein interaction assays through Y2H between pepper core signaling components or with Arabidopsis orthologs showed ABA-dependent and -independent interactions (**Figure 5**), as observed in examples from other crops (Wang et al., 2018). CaHAI showed an interaction pattern similar to the Arabidopsis orthologs (Bhaskara et al., 2012). In BiFC, interaction between CaHAI and CaPYL2 or CaPYL4 were both ABA-dependent and localized to the nucleus and cytoplasm, indicating that both transcriptional and immediate responses to ABA can be regulated by these complexes (**Figure 6**).

Although all the data here gathered points toward CaPYL2/4, CaHAI, and SnRK2.3/2.6 as ABA receptors, PP2C phosphatases and SnRK kinases, respectively, in vitro activity assays are needed to truly probe this point.

Notably, a CaABI transcript, annotated as splicing isoform 4 (XM\_016724784.1), amplified from dried/germinating seeds, encoded a truncated protein lacking two conserved D residues required for the Mn/Mg++ ions interaction. When used for interaction assays was not capable to complex with PYLs or

SnRK2s, suggesting that this variant may not be functional. Alternative splicing of pre-mRNAs constitutes an additional layer of regulation of gene expression and a way to expand the proteome diversity or regulate protein functionality in specific developmental/environmental situations, or cell types (Laloum et al., 2018). In plants, intron retention is indicated as the most common alternative splicing type (Ner-Gaon et al., 2004). Stress and ABA-response regulatory genes are known to be most prone to alternative splicing phenomena, which may produce active vs. inactive protein variants (Laloum et al., 2018). Specifically, splicing variants have been described in the case of the Arabidopsis HAB1 PP2C (Ling et al., 2017). The ratio between the full-length, functional HAB1 protein and the truncated splice form resulting from intron retention are regulated by the ABA levels and are involved in the fine-tuning of the ABA signaling cascade (Wang et al., 2015). The splicing variant of CaABI expressed in seeds may result from a similar mechanism of balance of phosphatase activity, and may reflect a developmental rather than environmental regulation between protein isoforms.

#### CONCLUSION

In conclusion, pepper seeds retained high germination percentages in presence of NaCl and ABA. This may be due to low expression in seeds of ABA signaling components such as CaABI, CaPYL2, CaPYL4, CaSnRK2.3, CaSnRK2.6. The expression of tested PYLs and SnRK2s was comparatively higher in vegetative tissues, and this correlated with restored sensitivity to exogenous ABA and salinity. Expression of a splice variant of CaABI encoding a truncated protein unable to interact with other tested signaling components may also impair ABA signaling cascade in seeds. The alleles controlling the expression of these signaling components in maturing and dry seed are potential traits that can be manipulated for the purposes of improving pepper low seed storability through molecular breeding and gene editing.

#### REFERENCES


# DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the **Supplementary Files**.

#### AUTHOR CONTRIBUTIONS

AR, SG, and GB designed the work. AR, SL, PP, and MP performed the experiments and prepared the figures. AR, SL, MVO, and AC analyzed the data. GM and AM participated in the experimental design and data interpretation. AR, MVO, and GB wrote the manuscript, with inputs from all authors. All the authors have read and approved the manuscript.

### FUNDING

This work was supported by the Italian Ministry of University and Research, under Grant No. PON02\_00395\_3215002 (GenHORT).

#### ACKNOWLEDGMENTS

We wish to thank Dr. V. Cirillo (University of Naples, Federico II) for help with the statistical analyses and Mr. G. Guarino (CNR-IBBR) for help with figure preparation.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00304/ full#supplementary-material



levels of abscisic acid (ABA). J. Am. Soc. Hortic. Sci. 123, 925–930. doi: 10.21273/ JASHS.123.5.925



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

Copyright © 2019 Ruggiero, Landi, Punzo, Possenti, Van Oosten, Costa, Morelli, Maggio, Grillo and Batelli. 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.

#### Regulation of K+ Nutrition in Plants

*Paula Ragel1,2 , Natalia Raddatz1 , Eduardo O. Leidi3 , Francisco J. Quintero1 and José M. Pardo1 \**

*1Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas y Universidad de Sevilla, Seville, Spain, 2Centre for Organismal Studies, Universität Heidelberg, Heidelberg, Germany, 3Instituto de Recursos Naturales y Agrobiologia de Sevilla, Consejo Superior de Investigaciones Cientificas, Seville, Spain*

Modern agriculture relies on mineral fertilization. Unlike other major macronutrients, potassium (K+ ) is not incorporated into organic matter but remains as soluble ion in the cell sap contributing up to 10% of the dry organic matter. Consequently, K+ constitutes a chief osmoticum to drive cellular expansion and organ movements, such as stomata aperture. Moreover, K+ transport is critical for the control of cytoplasmic and luminal pH in endosomes, regulation of membrane potential, and enzyme activity. Not surprisingly, plants have evolved a large ensemble of K+ transporters with defined functions in nutrient uptake by roots, storage in vacuoles, and ion translocation between tissues and organs. This review describes critical transport proteins governing K+ nutrition, their regulation, and coordinated activity, and summarizes our current understanding of signaling pathways activated by K+ starvation.

*Edited by:* 

*Lam-Son Tran, RIKEN, Japan*

#### *Reviewed by:*

*Manuel Nieves-Cordones, Center for Edaphology and Applied Biology of Segura, Spanish National Research Council (CSIC), Spain Fernando Aleman, The Scripps Research Institute, United States*

#### *\*Correspondence:*

*José M. Pardo jose.pardo@csic.es*

#### *Specialty section:*

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

*Received: 07 December 2018 Accepted: 20 February 2019 Published: 20 March 2019*

#### *Citation:*

*Ragel P, Raddatz N, Leidi EO, Quintero FJ and Pardo JM (2019) Regulation of K+ Nutrition in Plants. Front. Plant Sci. 10:281. doi: 10.3389/fpls.2019.00281*

Keywords: plant nutrition, potassium, nitrate, regulation, long-distance transport

# INTRODUCTION

Potassium (K+ ) is of paramount importance in plant cell physiology. K+ is an essential macronutrient that fulfills critical functions related to enzyme activation, osmotic adjustment, turgor generation, cell expansion, regulation of membrane electric potential, and pH homeostasis (Hawkesford et al., 2012). While the K+ concentration in the soil solution may vary widely from 0.01 to 20 mM, plant cells maintain a relatively constant concentration of 80–100 mM in the cytoplasm (Rodriguez-Navarro, 2000). Moreover, plants accumulate large amounts of K+ in their vacuoles, surpassing purely nutritional requirements. Hence, K+ is the most abundant cation in plant cells, comprising up to 10% of plant dry weight and often exceeding the ca. 2% that supports near-maximal growth rates (White and Karley, 2010). There is a steep curvilinear relationship between the tissue concentration of K+ and plant growth, from which a critical concentration of K+ supporting 90% of maximum yield can be determined. Above this concentration, growth has no correlation with the increased K+ content, but at lower K+ concentrations, growth declines rapidly. Consequently, K+ fertilization is common practice in modern agriculture and about 40–60% of crop yields are attributable to commercial fertilizer use (Stewart et al., 2005). However, agricultural fertilization is far from being fine-tuned with nutritional requirements.

K+ is taken up from the soil solution by root epidermal and cortical cells. Once K+ is inside the root symplast, it may be stored in vacuoles, where it fulfills osmotic functions, or is transported to the shoot *via* xylem (Pardo and Rubio, 2011). In turn, shoot cells may also supply stored K+ for redistribution *via* phloem. In this transit from the soil to the different

**97**

plant organs, K+ crosses various cell membranes through K+ specific transport systems (**Figure 1**). Coordinated operation of the different transport systems within the plant to secure K+ uptake from the soil and delivery to the different plant organs requires complex K+ sensing and signaling mechanisms. Because of the extraordinary diversity of K+ transporters in plant cells and the physiological and developmental processes in which they are involved, this review is focused on the molecular mechanisms mediating K+ uptake and release at the plasma membrane level, with an emphasis on K+ absorption from the soil and distribution throughout the plant due to the relevance of these processes in plant nutrition. Storage of K+ into vacuoles is treated only briefly and readers are referred to other comprehensive reviews describing transport systems operating at the tonoplast (Martinoia et al., 2012; Ahmad and Maathuis, 2014; Eisenach and De Angeli, 2017; Martinoia, 2018). Last, because of the extensive interactions of nitrogen and K+ in plant mineral nutrition, we summarize the coordinated regulation of NO3 − and K+ uptake and long-distance transport in *Arabidopsis*.

Uptake and distribution of K+ in plant cells is carried out by a variety of transporter proteins categorized into several families with varied structures and transport mechanisms that comprise the channel families *Shaker*-like voltage-dependent, the tandem-pore (TPK), and the two-pore channels (TPC) (Hedrich, 2012), the carrier-like families KT/HAK/KUP (Nieves-Cordones et al., 2014a; Li et al., 2018), HKT uniporters and symporters (Hamamoto et al., 2015), and cation-proton antiporters (CPA). The CPA family is the largest one and includes the NHX, CHX, and KEA antiporters (Sze and Chanroj, 2018). In this review, we describe the structure and diversity of the main K+ transporter families whose members contribute substantially to K+ nutrition. Other proteins with uncertain roles or descriptions of transport activities without candidate proteins have been omitted.

#### TRANSPORT PROTEIN FAMILIES INVOLVED IN K+ NUTRITION

#### K+ -Selective Channels

The first K+ transporter with a role in nutrient uptake was the *Shaker*-like, voltage-gated, and K+ -selective channel AKT1 (Hirsch et al., 1998). Although voltage-gated (VG) channels of plants are phylogenetically related to animal *Shaker* channels, they are distinct and include additional functional domains (Jegla et al., 2018). The basic architecture of VG channels consists of four α-subunits surrounding a central aqueous pore for K+ permeation. Each subunit contains six transmembrane segments, named S1–S6, which can be divided into two different modules: the first four α-helices form a voltage-sensor domain that contains multiple positively charged residues that moves within the membrane in response to voltage. This movement is directly coupled to the opening or closing of the channel. The segments S5, S6, and the pore loop, form the pore domain, named P, where each of the four subunits contributes equally to the permeation pathway. Moreover, plant α-subunits have a long C-terminal region constituting more than half of the protein (**Figure 2**). This cytosolic tail includes several functional domains: (1) a linker region (C-linker) proximal to the pore

FIGURE 1 | Transporters involved in K+ uptake by roots and inter-organ partition. HAK5, AKT1, and non-selective cyclic nucleotide-gated cation channels (CNGC) all contribute to K+ nutrition, albeit at different ranges of substrate concentrations, from low- to high-availability, respectively. K+ efflux through the outward-rectifying GORK channel facilitates the fine-tuning of plasma membrane electrical potential, and allows repolarization under circumstances that promote depolarization, such as salinity stress. In the root stele, the outward-rectifying SKOR channel releases K+ into the xylem vessels for nutrient delivery to the shoots. The nitrate transporter NRT1.5 facilitates K+ uploading into the xylem either by electrical coupling with other K+ -selective transporters or directly acting as K+ /H+ antiporter. In aerial tissues, an array of K+ -influx channels and KT/HAK/KUP carriers allow the uptake of the incoming K+ into green cells. K+ is stored inside vacuoles by NHX exchangers and released back to the cytosol by TPK and TPC1 channels, and possibly also by KT/HAK/KUP carriers at the tonoplast (the vacuole in root cells is omitted for simplicity). The plasma membrane outward K+ channel AKT2 releases K+ into the phloem for returning K+ to the root and to facilitate the uploading of photosynthates into the phloem sap.

that transduces conformational changes that gate the channel and that may also determine the target membrane (Nieves-Cordones et al., 2014b; Jegla et al., 2018); (2) a conserved and essential cyclic nucleotide-binding homology domain (CNBHD) whose function is not the binding of cNMP but to mediate the interactions between subunits within the channel tetramer; (3) an ankyrin domain (found in only six out of the nine *Arabidopsis* VG channels), which may mediate the binding of interacting proteins (Michaely and Bennett, 1992); and (4) a distal KT/KHA domain rich in hydrophobic and acidic residues, that is unique to plant K+ channels, and is involved in channel tetramerization and clustering at the membrane (Daram et al., 1997; Ehrhardt et al., 1997; Zimmermann et al., 2001; Dreyer et al., 2004).

Plant voltage-gated K+ channels are divided into three subfamilies regarding their response to the membrane potential (Dreyer and Uozumi, 2011): (1) Inward-rectifying (Kin) channels that in *Arabidopsis* include AKT1, AKT6, KAT1, and KAT2; they open at hyperpolarized membrane potentials allowing the uptake of K+ . (2) Outward-rectifying (Kout) channels that mediate K+ release because they open at depolarized membrane potentials; this group is composed of SKOR and GORK channels. (3) Weakly rectifying (Kweak) channels that can mediate both K+ uptake and release, and whose *Arabidopsis* representative is AKT2. In addition, the *Arabidopsis* KC1 (KAT3) is an electrically silent *Shaker*-like protein that interacts with and regulates functionality of the Kin channels AKT1, KAT1, KAT2, and AKT2, but not the Kout channels (Jeanguenin et al., 2011). This interaction negatively shifts the activation threshold of Kin channels and decreases the macroscopic inward conductance compared to that of homomeric channels (Dreyer et al., 1997; Duby et al., 2008; Geiger et al., 2009). Heteromerization of different subunits of Kin channels is of great importance to increase the functional diversity and regulation of different cell types (Dreyer et al., 1997; Véry and Sentenac, 2003; Xicluna et al., 2007; Jeanguenin et al., 2008; Lebaudy et al., 2008; Lebaudy et al., 2010). Although this behavior has also been suggested for outward-rectifier (Kout) channels, heteromerization has been reported only among subunits of Kin or Kout, preventing formation of heteromeric structures between the two subunit types (Dreyer et al., 2004).

#### K+ -Uptake Carriers

Proteins of the KT/HAK/KUP family are present in plants, fungi, bacteria, and even viruses (Greiner et al., 2011; Santa-Maria et al., 2018), and they are often associated with K+ transport across membranes and K+ supply. In bacterial genomes, K+ carriers of this family are encoded by single-copy genes named *kup*. In *Escherichia coli*, kup is a constitutive low-affinity uptake system that operates as K+ -H+ symporter (Zakharyan and Trchounian, 2001). In fungi, the homologous proteins are encoded by *HAK1*-like genes present as one- or two-copy in most species. In contrast to bacterial *kup*, fungal *HAK* genes are strongly induced by K+ starvation and the encoded proteins mediate high-affinity K+ transport (Benito et al., 2011). In plants, these transporters are known as KT, HAK, or KUP (KT/HAK/KUP family) and they are represented by multiple genes in their genomes. Members of this family have been widely associated with high-affinity K+ uptake from the soil, while others may function in both low-affinity and/or highaffinity transport (Luan et al., 2009; Very et al., 2014) and other roles related, for example, to K+ translocation, control of water movement at the plant level, salt tolerance, osmotic/ drought responses, transport of other alkali cations, and developmental processes in plants, such as root hair growth and auxin distribution (Li et al., 2018; Santa-Maria et al., 2018). These diverse functions of KT/HAK/KUP transporters may all result from their critical roles in cellular K+ homeostasis. *KT/HAK/KUP* genes are not present in animal cells, what could indicate that they are crucial for K+ transport in organisms facing external solutions with fluctuating and very low K+ concentrations, often in the μM range (Ashley et al., 2006). Based on the present knowledge, *KT/HAK/KUP* genes are present in all plant genomes, which contrasts with that in the other kingdoms, where they are only present in certain species (Grabov, 2007; Greiner et al., 2011). This difference may reflect the importance of these transporters for the plant's way of life.

KT/HAK/KUP transporters of land plants are classified according to their sequence homology into six clusters or clades (I–VI), with clade VI including only members of bryophytes (Santa-Maria et al., 2018). Phylogenetic analysis shows that all KT/HAK/KUPs from algae diverge from land plant clades, suggesting that the diversification into these groups took place after the colonization of land by green organisms (Santa-Maria et al., 2018). The KT/HAK/KUP group from angiosperms displays a high and rather variable number of members in the different plant species genomes that have been sequenced so far. For instance, there are 13 genes in *Arabidopsis*, 16 in peach, 17 in grapevine, 20 in *Medicago*, 21 in *Cassava*, 27 in rice, maize, and *Brachypodium,* and 57 in *Panicum virgatum* (Song et al., 2015; Nieves-Cordones et al., 2016a,c; Ou et al., 2018). Members of KT/HAK/KUP family in angiosperm are classified among clades I–V (Nieves-Cordones et al., 2016c).

The KT/HAK/KUP transporters involved in K+ uptake from the soil are clustered into a distinct subgroup of clade I, termed Ia (Nieves-Cordones et al., 2016c) and that we call herein HAK1-like transporters by analogy with the fungal counterparts. This subgroup includes barley HvHAK1 (Santa-María et al., 1997; Gierth and Mäser, 2007), *Arabidopsis* AtHAK5 (Rubio et al., 2000; Gierth et al., 2005), rice OsHAK1 and OsHAK5 (Bañuelos et al., 2002), pepper CaHAK1 (Martinez-Cordero et al., 2004), tomato LeHAK5 (Nieves-Cordones et al., 2007), and *Thellungiella* ThHAK5 (Alemán et al., 2009b). High-affinity K+ transport has been demonstrated for all the HAK1-like transporters in heterologous expression systems (Nieves-Cordones et al., 2014a). On the other hand, AtKUP7, belonging to clade V, could be involved in K+ uptake from low to moderate external K+ concentrations (Han et al., 2016), and thus the participation in K+ uptake from soil of proteins from different clades should not be discarded. Interestingly, other members of clade I could be related with K+ uptake by cells from specialized tissues. DmHAK5 from Venus flytraps is implicated in the uptake of K+ released from the digested prey in the bi-lobed capture organ (Scherzer et al., 2015), whereas the quinoa CqHAK5-like drives K+ influx into cells of the leaf salt bladders to contribute to the osmotic balance of the cytosol against the osmotic pressure of the salt-containing vacuoles (Bohm et al., 2018).

KT/HAK/KUP transporters are phylogenetically related to the superfamily of acid-polyamine-organocation (APC) transporters that comprises secondary active transport proteins responsible for uniport, symport, and antiport of a wide range of substrates (Vastermark et al., 2014). Taking as template crystal structures of prokaryotic APC transporters, computational 3D modeling of AtKUP7 (Ahn et al., 2004; Al-Younis et al., 2015; Santa-Maria et al., 2018), AtKUP4/TRH1 (Daras et al., 2015), OsHAK1 (Rai et al., 2017), AtKUP1/TRH1 (Santa-Maria et al., 2018), AtHAK5 (Santa-Maria et al., 2018), and HvHAK1 (Santa-Maria et al., 2018) has been reported. The structural models (**Figure 2**) show the presence of common attributes among all of them: (1) a hydrophobic core containing 10–14 transmembrane (TM) segments; and (2) three cytosolic domains—the N- and C-termini and a region containing approximately 70 residues situated between second and third TMs (loop II–III). Although, the structure of the pore region has not been described yet, several works have analyzed the effect of mutations on the function of these transporters (Santa-Maria et al., 2018). So far, results indicated that several parts of the protein may contribute to setting the Vmax of the transporter and that the region including from N-terminus to loop II–III may contribute in determining its Km. Furthermore, sequence alignments show that, although there is not extensive sequence conservation, 40 amino acid residues are conserved in exactly the same position in all the eukaryotic HAK transporters and in the Kup bacterial transporters (Rodriguez-Navarro, 2000). Six of these conserved residues are included in a highly conserved motif in the first transmembrane domain whose consensus sequence is **G**VVY**GD**LGTS**PLY** (the amino acids conserved in all proteins are in bold) (Rodriguez-Navarro, 2000). A helical-wheel representation of this transmembrane fragment locates three glycine residues on the same side of the helix, which in the case of a tetrameric structure may operate as a substrate selectivity filter analogous to the GXGYGD motif highly conserved in K+ channels. Regarding this, it has been suggested that AtKUP4/TRH1 may form homodimers (Daras et al., 2015), likely involving the interaction between C-terminus domains and less likely between loops II–III.

Contrary to VG channels that are all targeted to the plasma membrane, KT/HAK/KUP transporters have been reported in different subcellular compartments (**Table 1**). The majority of the characterized transporters of the KT/HAK/KUP family are located in the plasma membrane, although not all of them are involved in K+ nutrition. For instance, AtKUP4/TRH1 seems to participate in auxin transport related with root gravitropism and root hair development (Rigas et al., 2013), whereas AtKUP6 acts in lateral root initiation and development in the auxin and ABA crosstalk signaling pathways (Osakabe et al., 2013).


TABLE 1 | Sub-cellular location of selected *Arabidopsis*, barley, rice, and *Physcomitrella patens* KT/HAK/KUP transporters.

#### HKT Proteins

The ample repertoire of transporters encoded in the genome of plants includes proteins that are collectively known as High affinity K+ Transporters (HKTs; **Figure 2**) despite the fact that these proteins facilitate Na+ -selective uniport or Na+ -K+ symport with a channel-like activity (Benito et al., 2014). Phylogenetic and functional analyses distinguished two HKT subfamilies (Platten et al., 2006). Members of subfamily I (HKT1) are ubiquitous in plants, Na+ -selective, and mostly involved in Na+ recirculation through vascular tissues, as best exemplified by AtHKT1;1 (Sunarpi et al., 2005). Members of subfamily II (HKT2) have been found only in monocotyledonous species. Although they are all K+ -permeable, mechanistically HKT2s can operate as either Na+ -K+ symporters or K+ -selective uniporters [reviewed by Benito et al. (2014)]. HKT2-like proteins of cereals have been involved in K+ nutrition.

#### K+ UPTAKE BY ROOTS

The uptake of K+ by roots (often measured with rubidium as tracer) exhibits a biphasic kinetics in response to increasing external concentrations corresponding to high- and low-affinity transport systems, which work at low (<1 mM) and high (>1 mM) external K+ concentrations respectively (Epstein et al., 1963; Gierth and Mäser, 2007). At high concentration in the soil solution, K+ crosses the membrane mostly through channels. The channels simply give a path for the ions allowing them to move down the electrochemical gradient. At low K+ concentration, active transporter systems are needed in order to pull K+ inside the cell against its electrochemical gradient. However, studies in several plant species have shown that channels may be involved in K+ uptake in the high-affinity range of K+

concentrations (Rubio et al., 2010) as long as the membrane is sufficiently hyperpolarized, i.e. highly electronegative inside (Hirsch et al., 1998; Spalding et al., 1999; Gierth and Mäser, 2007; Rubio et al., 2010).

The sensitivity to NH4 + is an important characteristic of highaffinity K+ uptake mediated by carriers that has been used as a tool for the identification of additional high-affinity transport systems (Santa-María et al., 1997; Nieves-Cordones et al., 2007). NH4 + -sensitive and -insensitive components of high-affinity K+ uptake have been identified in *Arabidopsis* (Spalding et al., 1999), barley (Santa-Maria et al., 2000), pepper (Martinez-Cordero et al., 2005), and rice (Bañuelos et al., 2002; Chen et al., 2015). Results indicate that the NH4 + -sensitive component of K+ uptake is likely mediated by KT/HAK/KUP transporters (HAK1-like transporters), whereas inward-rectifier K+ channels (AKT1-like channels) constitute the NH4 + -insensitive pathway (Santa-Maria et al., 2000; Nieves-Cordones et al., 2014a). Together, AKT1-like channels and HAK1-like transporters are now thought to constitute the main systems for K+ uptake in plants under low-K+ concentrations (**Table 2**). However, the NH4 + -sensitive and -insensitive pathways appear to contribute differently to highaffinity K+ uptake depending on the plant species and the ionic external concentration of transported substrates, mainly K+ , NH4 + , and Na+ (Aleman et al., 2011; Nieves-Cordones et al., 2016c).

Among the inward-rectifying K+ channels of *Arabidopsis*, only AKT1 and AtKC1 are abundantly expressed in root tissues (Reintanz et al., 2002). AtKC1 expressed alone remains in the endoplasmic reticulum, but it can be recruited to the plasma membrane to regulate AKT1 activity (Duby et al., 2008; Geiger et al., 2009; Honsbein et al., 2009; Wang et al., 2010). In addition, AKT1 is positively regulated by the protein kinase complex comprising the kinase CIPK23 and one of the two alternative calciumdependent regulatory subunits CBL1 and CBL9 (Li et al., 2006;

TABLE 2 | Comparison of AKT1 channels and HAK1/HAK5 transporters from *Arabidopsis* and rice working at different ranges of external K+ concentrations. The uptake systems working in addition to AKT1 and HAK1/HAK5 likely include CHX exchangers (Zhao et al., 2008) and cyclic nucleotide-gated channels (CNGC) that may contribute to K+ absorption when the external K+ concentration is sufficiently high (Caballero et al., 2012).


Xu et al., 2006). AKT1 possesses an intrinsic K+ sensor reducing channel conductance at submillimolar external K+ concentrations. Despite this K+ sensor, upon activation by the CIPK/CBL complex at low external K+ , the homomeric AKT1 channels open at voltages positive of *E*K, a condition potentially resulting in cellular K+ leakage (Geiger et al., 2009). Incorporation of the *At*KC1 subunit into the channel complex, however, shifts the voltage dependence of AKT1 toward more negative potentials (ca. −70 mV) to prevent K+ loss (Geiger et al., 2009; Wang et al., 2010; Wang et al., 2016). In other words, AKT1/KC1 heteromerization renders the channel more efficient at blocking K+ permeation in the outward direction. The physical interaction of the CIPK23/CBL1 complex is specific for AKT1 channels and does not involve the AtKC1 subunit. The gain-of-function mutation *AtKC1-D* (G322D substitution in transmembrane S6) was recovered in the *cipk23* mutant background. *AtKC1-D* enhanced the inhibition of AKT1 channel activity and restricted K+ leakage through AKT1 under low-K+ conditions, thereby increasing the tolerance to nutrient stress (Wang et al., 2016). Although the double mutant *akt1 KC1-D* was sensitive to low-K+ , indicating that KC1-D action is through AKT1, an additional indirect effect of mutation KC1-D through HAK5 cannot be ruled out. By inhibiting AKT1 and shifting its voltage dependence toward a more negative direction, the plasma membrane could become hyperpolarized in the *KC1-D* mutant, thereby enhancing the expression and activity of HAK5 and improving net K+ uptake.

Several mechanisms for AKT1 deactivation have been proposed. The PP2C-type protein phosphatase AIP1 interacts with and inactivates the AKT1 channel, counteracting the activation by CIPK23 in oocytes (Lee et al., 2007). In principle, these findings are evidence of a phosphorylation/dephosphorylation switch that regulates AKT1 channel activity, but it should be noted that no phosphorylation of AKT1 by CIPK23 and dephosphorylation by AIP1 has been demonstrated conclusively (Hashimoto et al., 2012). Instead, four components, CIPKs, CBLs, PP2Cs, and AKT1, appear to interact mutually and form a molecular complex whose specific composition could ultimately regulate channel activity (Lan et al., 2011). In this model, PP2C phosphatases interact with the kinase domain of CIPKs to counteract kinasemediated activation of AKT1. Upon calcium signaling, CBLs interact with PPC2C to inhibit their phosphatase activity while simultaneously activating the partnering CIPKs. On the other hand, CBL10, a regulatory subunit of CIPK24/SOS2 but not of CIPK23, also interacts directly with AKT1 and negatively modulates AKT1 activity by competing with CIPK23 to bind AKT1 (Ren et al., 2013). Since CBL10 function is related to salinity stress rather than to mineral nutrition (Kim et al., 2007; Quan et al., 2007; Lin et al., 2009), this cross-regulation may constitute a mechanism to prevent salinity-induced K+ loss though AKT1. In line with this, the nitric oxide (NO) that accumulates under salinity stress also inhibits the K+ uptake mediated by AKT1. The link is indirect since NO triggered the accumulation of pyridoxal 5′-phosphate (PLP), an active form of vitamin B6, that in turn repressed the activity of AKT1 in *Xenopus* oocytes and *Arabidopsis* root protoplasts (Xia et al., 2014).

In *Arabidopsis*, the voltage-gated channel GORK (guard cell outward-rectifying K+ ) is the major outward-rectifying K+ channel in guard cells where it contributes to K+ efflux for decreasing turgor and stomatal closure (Ache et al., 2000; Hosy et al., 2003). In addition, GORK is expressed in root outer cell layers (epidermal, root hairs, and cortex) of *Arabidopsis* and thus GORK is considered a major pathway for stressinduced K+ leakage from root cells, e.g. by exposure of roots to high salt (Ivashikina et al., 2001; Demidchik et al., 2010; Demidchik et al., 2014). Production of hydroxyl radicals (HO˙) in salinized roots stimulates a dramatic K+ efflux mediated by GORK from root cells (Demidchik et al., 2010). The oxidative and salt stresses cause programmed cell death (PCD) and collapse membrane potential in root cells of *Arabidopsis thaliana* in a K+ -dependent manner. Accordingly, the *Arabidopsis gork1-1* mutant showed no K+ outwardly directed currents in response to HO˙. Besides, after exposure to high NaCl levels, the mutant *gork1-1* displayed lower activity of proteases and endonucleases for PCD, which in the wild type was dramatically enhanced by K+ loss in root cells (Demidchik et al., 2010).

Both the expression level and channel activity of GORK are significantly upregulated by increasing levels of the abscisic acid (ABA) and jasmonate. Stimuli that elevated endogenous ABA concentrations, e.g. drought, osmotic stress, or cold, led to the up-regulation of *GORK* transcripts (Becker et al., 2003; Suhita, 2004) while treatment with salicylic acid inhibited the presence of active GORK channels and improved salinity tolerance through prevention of K+ efflux. Recent studies demonstrated that calcium-dependent protein kinase 21 (CPK21) phosphorylated GORK and suggested that 14-3-3 proteins control GORK activity through binding with CPK21. This kinase phosphorylates three amino acid residues in the C-terminus of GORK, T344, S518, and S649. Binding of 14-3-3 to CPK21 strongly stimulated its kinase activity and increased Ragel et al. Regulation of K+ Nutrition in Plants

GORK phosphorylation (van Kleeff et al., 2018). On the other hand, the phosphatase AtPP2CA interacts physically with GORK inhibiting its current (Lefoulon et al., 2016). Thus, AtPP2CA could have an antagonist role to CPK21 on the regulation of GORK (van Kleeff et al., 2018). These results imply that the salinity-induced membrane depolarization together with the Ca2+- and CPK21-dependent phosphorylation act together to activate GORK and to repolarize the plasma membrane by means of releasing part of the cytosolic K+ . Moreover, the peak of the salt-induced K+ -efflux in the *aha2* mutant, devoid of a major isoform of the plasma membrane H+ -ATPase, was stronger and more sustained than in the wild-type, suggesting that H+ -pumps take over membrane repolarization after the initial K+ -loss to re-enact K+ uptake (van Kleeff et al., 2018). Recently, Saponaro et al. (2017) showed that 14-3-3 proteins are also capable of modulating KAT1, although in this case 14-3-3 bound directly to the KAT1 C-terminus affecting both the voltage dependency of the channel and the number of channel molecules in the membrane (Sottocornola et al., 2008).

K+ -H+ symport has long been considered the likely catalytic mechanism of plant KT/HAK/KUP transporters based on the demonstration that K+ -H+ symport operates in K+ -starved *Neurospora crassa* and on thermodynamical considerations regarding the steep K+ gradient that KT/HAK/KUP proteins are able to achieve across cell membranes that exceeds what could be reached by coupling the K+ uptake to the membrane potential solely (Rodriguez-Navarro, 2000). Until recently, efforts to express plant KT/HAK/KUP proteins in *Xenopus* oocytes to measure K+ currents had failed, but work with the DmHAK5 transporter from Venus flytraps showed that co-expression of the corresponding cRNA with that of CBL9/CIPK23 (but not DmHAK5 alone) generated inward K+ and Rb+ currents in *Xenopus* oocytes that were stimulated by low external pH (Scherzer et al., 2015). Moreover, salt bladders of the halophyte *Chenopodium quinoa* that accumulate salts to very high concentrations express a HAK-like activity driving high-affinity and selective K+ uptake that was dependent on acidic external pH and by the CIPK23/CBL1 kinase module of *Arabidopsis* (Bohm et al., 2018). Electrophysiological recordings in rice roots showed that the activity of OsHAK1 was strongly electrogenic and depolarizing. Plots of the OsHAK1-dependent K+ -induced membrane depolarization had a slope of 29 mV per decade of external K+ concentration, suggesting the co-transport of two monovalent cations (a 59 mV slope is to be expected from an uniprot transport moving only single K+ ions) (Nieves-Cordones et al., 2017). Together, these data strongly suggest that plant KT/ HAK/KUP proteins operate as K+ -H+ symporters. Residues involved in K+ binding and/or transport have not been identified; however, mutant proteins with residue substitutions of members of the KT/HAK/KUP family have been described as showing modified affinity for K+ , Na+ , and/or Cs+ , or increased Vmax (Aleman et al., 2014).

HAK1-like transporters are subject to complex transcriptional and post-translational regulations, although studies have been carried out almost exclusively in *Arabidopsis* AtHAK5 (Jung et al., 2009; Rubio et al., 2014; Ragel et al., 2015). Under any stress conditions that directly affect K+ acquisition, such as K+ deprivation or salinity, high-affinity K+ uptake systems should be transcriptionally or post-translationally activated in order to maintain the K+ supply and K+ /Na+ homeostasis. Accordingly, all characterized HAK1-like transporters exhibit low expression levels in roots under control conditions, are highly up-regulated upon K+ deprivation and rapidly down-regulated when K+ is resupplied (reviewed by (Li et al., 2018)). Furthermore, it has been commonly observed that other ions, particularly NH4 + , NO3 − , Na+ , and Pi, also regulate the expression of *HAK1*-like genes and not always in the same way (Nieves-Cordones et al., 2019). For example, NH4 + reduces the transcriptional induction by K+ starvation of the pepper *CaHAK1* (Martinez-Cordero et al., 2005) and *Arabidopsis AtHAK5* (Qi et al., 2008), but enhances the expression of *LeHAK5* in tomato (Nieves-Cordones et al., 2007). The presence of NaCl prevents the induction of *LeHAK5* by K+ starvation (Nieves-Cordones et al., 2007), but provokes a strong and transient up-regulation of *HvHAK1* (Fulgenzi et al., 2008). Thus, the *Arabidopsis* model cannot be completely extended to other plant species, crops among them. In contrast to HAK1-like transporters, KT/HAK/KUP proteins belonging to clusters II–V show diverse expression patterns and most of them do not exhibit transcriptional regulation in response to K+ deficiency (Ahn et al., 2004; Li et al., 2018). For example, *AtKUP7* (cluster V, plasma membrane) transcript is not induced by low-K+ (Han et al., 2016) and *AtKUP12* (cluster III, chloroplast) is down-regulated after K+ resupply (Armengaud et al., 2004).

Regarding the transcriptional regulation of genes encoding HAK1-like transporters, it has been shown that the effect of the nutrient deficiency and salt stresses on transcriptional expression of *AtHAK5* and *LeHAK5* is associated with changes in the root cell membrane potentials (Nieves-Cordones et al., 2008; Rubio et al., 2014); the hyperpolarization of the plasma membrane of root cells induces transcription of both genes. Supporting this, *ThHAK5* of *Thellungiella halophila* (salt cress, a.k.a. *Eutrema salsuginea*) is expressed to higher levels than *AtHAK5* under salt stress, while roots of *T. halophila* maintained a more negative membrane potential than *Arabidopsis* roots (Volkov and Amtmann, 2006; Alemán et al., 2009b; Rubio et al., 2014). Besides membrane hyperpolarization, the expression of *AtHAK5* is also induced, under K+ -limiting conditions, as result of signaling cascades that involve ROS production, phytohormones, and transcription factors. Low-K+ stress, alike other nutrient-deprived conditions, promotes an increase of ethylene that positively regulates ROS production in roots (Shin and Schachtman, 2004; Jung et al., 2009). Roots deprived of K+ induce the expression of genes involved in ethylene biosynthesis and signaling, and in ROS metabolism, promoting two-fold higher levels of ethylene and the increase in hydrogen peroxide (H2O2) concentrations. Both ethylene and ROS give rise to enhanced transcription of *HAK5* in *Arabidopsis* and tomato (Rodenas et al., 2018). In *Arabidopsis*, H2O2 produced by the NADPH oxidase RHD2/RbohC regulates the expression of *AtHAK5* in response to K+ deficiency (Shin and Schachtman, 2004) and it has been proposed that peroxidase RCI3 (Rare Cold Inducible gene 3) contributes to ROS production during *Arabidopsis* root response to K+ deficiency (Kim et al., 2010). In the case of ethylene-induced *AtHAK5* transcription, the intermediaries in ethylene signaling CTR1 (Constitutive Triple Response (1) and EIN2 (Ethylene Insensitive (2) are partially involved. Results also suggest the existence of other signaling pathways or an EIN2-independent ethylene route that may play an important role in low-K+ signaling (Jung et al., 2009). Genetic hierarchy indicates that ethylene signaling acts upstream of ROS when plants are deprived of K+ (Jung et al., 2009). Nevertheless, it has also been speculated that a positive feedback may stimulate ethylene-induced ROS production (Wang et al., 2002). Other hormones have been shown to be involved in K+ deprivation signaling and response, for instance jasmonic acid (Armengaud et al., 2004), auxin (Jung et al., 2009; Hong et al., 2013), ABA (Kim et al., 2010), cytokinins (Nam et al., 2012), and gibberellins through DELLA proteins (Oliferuk et al., 2017). Cytokinins are known to regulate macronutrient homeostasis by controlling the expression of nitrate, phosphate, and sulfate transporters. Cytokinin content decreases under K+ -starved conditions, and cytokinin-deficient mutants, under same conditions, display enhanced accumulation of both ROS and *AtHAK5* transcripts (Nam et al., 2012). By contrast, cytokinin-receptor mutants lost the responsiveness to low-K+ , including ROS accumulation and root hair growth. Interestingly, the cytokinin/ethylene ratio is positively correlated with tomato shoot biomass, suggesting that the balance between both hormones is important in determining the plant vigor at low-K+ supply, but with an inverse role in tomato compared to *Arabidopsis*, where cytokinin/ethylene ratio was negatively correlated with tolerance to K+ deprivation (Jung et al., 2009; Nam et al., 2012).

In addition to low nutrient conditions, salt stress (and presumably other abiotic stresses) results in modifications of *AtHAK5* expression or the low-K+ response. Mild salt stress does not induce *AtHAK5* expression but its expression levels gradually increased following an increase in NaCl concentrations (Ahn et al., 2004; Hong et al., 2013). This suggests that plants may recognize high Na+ levels as K+ deprivation. However, the induction of *AtHAK5* expression by low K+ was suppressed by salt stress in *Arabidopsis* (Nieves-Cordones et al., 2010), but not in *T. halophila* (Alemán et al., 2009a). As discussed above, under salt stress conditions, *T. halophila* registers a more negative root membrane potential than *A. thaliana* (Volkov and Amtmann, 2006), which may explain the expression of *ThHAK5* under these conditions (Alemán et al., 2009a).

In recent years, several transcription factors (TFs), and their target sequences, have been identified in the *AtHAK5* promoter. Among them, ARF2 (Auxin Response Factor 2) is the only one described so far to work as negative regulator of *AtHAK5* transcription (Zhao et al., 2016). Interestingly, ARF2 has been found to be involved in many phytohormone-signaling pathways, but it seems not to participate in auxin signaling. Under K+ sufficient conditions, channel-mediated K+ uptake would be energetically more favorable than symport through AtHAK5, and hence *AtHAK5* should be shut down (Zhao et al., 2016). In those conditions, ARF2 binds to the auxin-responsive elements (AuxREs) within the *AtHAK5* promoter and represses transcription. When plants are subjected to low-K+ stress, ARF2 is rapidly phosphorylated by an unknown kinase and loses DNA binding activity. ARF2 is removed from the *AtHAK5*

promoter, which relieves the repression on *AtHAK5* transcription. In turn, other TFs bind to the *AtHAK5* promoter and activate its transcription. These TFs up-regulating *AtHAK5* expression under K+ starvation include RAP2.11, which binds to the ethylene-responsive element (ERE) and the GCC-box of the *AtHAK5* promoter, and whose expression is stimulated by ethylene and ROS, alike *AtHAK5* (Kim et al., 2012). TFs DDF2, JLO, bHLH121, and TFII\_A also interact with the upstream region of *AtHAK5*, but the specific binding motif for each of them has not been identified yet (Hong et al., 2013). All of these transcription factors are sufficient to activate *AtHAK5* expression in heterologous systems, but none of them is absolutely required. When K+ is resupplied, ARF2 becomes dephosphorylated again and represses *AtHAK5* expression (Zhao et al., 2016). Thus, it is apparent that regulation of the activity of TFs acting on *AtHAK5* transcription (positively or negatively) is necessary to determine cooperatively the accumulation of the corresponding transcripts (Santa-Maria et al., 2018).

Although a general nutrient deprivation stimulus is sufficient for the transcriptional activation of *AtHAK5* and *LeHAK5* genes, a reduction of internal K+ is required for the induction of a functional HAK5-mediated high-affinity K+ uptake in *Arabidopsis* and tomato roots (Rubio et al., 2014), suggesting the existence of post-transcriptional regulation *in planta*. Recently, it was shown that activation of high-affinity K+ uptake mediated by AtHAK5 (Ragel et al., 2015), DmHAK5 from Venus flytraps (Scherzer et al., 2015), and CqHAK from quinoa (Bohm et al., 2018) is mediated by the CBL-interacting protein kinase (CIPK)/ calcineurin B-like protein (CBL) complex comprising CIPK23 and CBL1/9 proteins of *Arabidopsis*. Notably, this CIPK23/CBL1,9 module also activates AKT1 channel, that together with AtHAK5 constitutes the main K+ uptake pathway in *Arabidopsis* roots (Xu et al., 2006; Lee et al., 2007). Both the protein kinase AtCIPK23 and the Ca2+ sensor AtCBL1 are necessary and sufficient for activation of the high-affinity K+ transporter AtHAK5 in yeast (Ragel et al., 2015). Besides AtCBL1, other CBLs (AtCBL8/9/10) are able to bind AtCIPK23 and activate AtHAK5 to complement K-uptake defective yeast growth. The reduction in the K+ concentration produces a specific Ca2+ signature in the cytosol (**Figure 3**) (Behera et al., 2017) that would be recorded by AtCBL1, promoting CIPK23/CBL1 complex formation, and the activation of AtHAK5 by phosphorylation at the cytosolic N-terminus (Ragel et al., 2015), in a similar way that was described for AKT1 (Xu et al., 2006; Lee et al., 2007). The enhancement of growth at low-K+ of yeast cells co-expressing AtHAK5, AtCIPK23, and AtCBL1 seems to result from modification of the kinetic properties of the transporter (Km decrease and Vmax increase), likely through the phosphorylation-induced conformational changes of AtHAK5 (Ragel et al., 2015). However, since physical interaction between CIPK23/CBL1 and AtHAK5 is also required for full activation of AtHAK5 in yeast, the trafficking of the transporter to plasma membrane has been proposed as a second mechanism of AtHAK5 regulation by CIPK23/CBL1 complex. Supporting this idea, the AtHAK5 protein was mainly detected in the endoplasmic reticulum of K+ -sufficient plants, while K+ starvation produced an enrichment of AtHAK5 protein in the plasma membrane (Qi et al., 2008).

In heterologous systems, the *Arabidopsis* CIPK23/CBL1,9 complex enabled the activation of various members from clade I of KT/ HAK/KUP transporters, such as pepper CaHAK1 (Ragel et al., 2015) and Venus flytrap DmHAK5 (Scherzer et al., 2015), but not of tomato SlHAK5 or the *Eutrema salsuginea* EsHAK5 (Ragel et al., 2015). These results suggested that the activation mechanism by CIPK23/CBLs complexes is evolutionarily conserved, but not the phosphorylation site and/or the target sequence recognition, which may vary among distant plant species. Accordingly, a quimeric tomato HAK5 protein that contained the 15 first amino acids of CaHAK1 could be activated by CIPK23/CBL1 in yeast (Ragel et al., 2015).

Since the two main contributors to K+ uptake in *Arabidopsis*, AKT1 and HAK5, are regulated by the CIPK23/CBL1,9 complex, the coordinated regulation of these transport systems deserves attention. Under K+ -sufficient conditions, K+ uptake by HAK5 would be energetically more expensive than permeation through the AKT1 channel. The H+ -pumping activity of plasma membrane ATPases, which is used for many secondary transport processes, creates *per se* an electrical charge (negative inside) that suffices to draw significant amounts of K+ into the cytosol. At a regular steady membrane potential of −120 to −180 mV, root epidermal cells could sustain a 100–1,000-fold inward-directed gradient of K+ . However, coupling K+ uptake to H+ influx not only returns H+ to the cytoplasm but also is more depolarizing than simple K+ permeation, which in turn imposes a greater demand on the H+ -pumps and ATP consumption. We speculate that under such conditions of K+ sufficiency, the plasma membrane is not hyperpolarized (or not enough), signaling phytohormones are not produced, and therefore transcription of *HAK5* is not activated. AKT1 would be operational through the physical interaction with CIPK23 (and possibly other CIPKs) (Lee et al., 2007). As the K+ concentration outside decreases, the function of AKT1 becomes increasingly hampered and full activation by the Ca2+ dependent CIPK23/CBL1,9 complex is required to sustain K+ uptake, while the KC1 safeguard prevents K+ leakage through AKT1 (Wang et al., 2016). CIPK23 is known to display different states of activation, depending on factors that affect the activation of CIPKs by upstream kinases (Barajas-Lopez et al., 2018) and CBL binding (Chaves-Sanjuan et al., 2014). Thus, a mild K+ deprivation may produce a partially activated CIPK23 that would be competent for activating AKT1 but not HAK5, whereas severe K+ deprivation leads to *HAK5* transcription and to full activation of CIPK23, which would then be competent for activating HAK5 (Ragel et al., 2015). The *CIPK23* is itself induced transcriptionally by low-K+ stress (Xu et al., 2006), which could also enhance the response to nutritional stress.

The participation of Ca2+ sensors in high-affinity K+ uptake could mechanistically connect K+ starvation with other abiotic stresses, for instance: salinity, water availability, oxygen deficiency (hypoxia) or absence (anoxia), mechanical stress, cold stress, heavy metal stress, and other nutrient deprivations, all sharing cytosolic free Ca2+ as a second messenger (Wilkins et al., 2016). Current thinking is that the specificity of Ca2+ signaling is determined by the amplitude and duration (and possible oscillation) of the cytosolic Ca2+ increase, often referred to as the "calcium signature" that is elicited by the stimulus. K+ deficiency evokes two successive Ca2+ signals in roots exhibiting different spatial and temporal specificity (Behera et al., 2017). The first one is characterized by a transient and fast Ca2+ increase within 1 min in the postmeristematic elongation zone (most prominently in the vascular tissue and endodermis), followed by a Ca2+ return nearly to basal concentrations in <7 min. The second wave (secondary Ca2+

response) occurs after several hours (18 h) as sustained Ca2+ elevation in defined tissues of the elongation and root hair differentiation zones. It has been proposed that this secondary Ca2+ elevation would contribute to long-term adaptation responses (Behera et al., 2017). These differential Ca2+ signatures could also underpin the step-wise activation of CIPK23/CBL1,9 discussed above. Ca2+-sensing proteins with diverse Ca2+ affinities, subcellular localizations, and downstream target specificities may perceive Ca2+ signatures differentially and transduce them into adequate downstream signaling responses (Hashimoto and Kudla, 2011).

Besides the CIPK23/CBL1,9 module, evidence for a second Ca2+-dependent pathway acting on AtHAK5 during low-K+ signaling has been recently provided (Brauer et al., 2016). The INTEGRIN LINKED KINASE1 (ILK1) interacts with AtHAK5 and promotes its accumulation, likely by phosphorylation of AtHAK5, albeit this could not be proved *in vitro*. ILK1 activity contributes to growth during extreme K+ limitation where AtHAK5 is the only transport system contributing to K+ uptake. ILK1 interacts with the calmodulin-like protein 9 (CML9), which invokes Ca2+ signaling, and both proteins were needed to promote AtHAK5 accumulation in the membrane fraction of *Nicotiana benthamiana* leaves, leading to the suggestion that ILK1 and CML9 promote HAK5 maturation and transport from the endoplasmic reticulum to the plasma membrane. It remains to be investigated whether ILK1 functions in parallel or in coordination with the CIPK23/CBL pathway to modify AtHAK5 activity.

In addition to the interaction with kinases, the modulation of KT/HAK/KUP transporters through physical interactions among different subunits (homomers or heteromers formation) has been suggested. Physical self-interaction among subunits of AtKUP4 has been described (Daras et al., 2015). In the event that KT/ HAK/KUP proteins formed a tetrameric structure similar to that of *Shaker* channels, the conserved sequence GVVYGDLGTSPLY in the first TM fragment of each subunit would line the pore, where the three conserved glycine residues of each subunit may operate as a selectivity filter (Rodriguez-Navarro, 2000).

In summary, the integration of post-translational regulation, including a specific Ca2+ signature, protein interactions, phosphorylation events, and reallocation of transporters to the plasma membrane, as well as the transcriptional regulation governed by plasma membrane hyperpolarization, ROS production, and hormonal response pathways, all of them integrate in the regulation of HAK1-like K+ uptake systems in plant cells under different K+ starvation conditions. **Figure 3** summarizes the transcriptional and post-translational regulation of AtHAK5.

The *Arabidopsis* mutant lacking HAK5 and AKT1 still takes up K+ and shows residual growth at external K+ concentration above 1 mM, indicating the existence of additional compensatory transport system(s) (Pyo et al., 2010; Rubio et al., 2010). Because this low-affinity uptake is largely sensitive to Ca2+ and other divalent metals, permeates Cs+ , and is inhibited by cyclic nucleotides, non-selective cyclic nucleotide-gated cation channels (CNGC) are likely candidates for this alternative system (Caballero et al., 2012). Indeed, genetic data have implicated AtCNGC3 and AtCNGC10 in K+ uptake because knock-out and knockdown lines have reduced K+ contents (Kaplan et al., 2007). Moreover, AtCHX13, a plasma membrane cation/proton antiporter up-regulated by K+ starvation, has been involved in root K+ uptake (Zhao et al., 2008). AtKUP7 is preferentially expressed in *Arabidopsis* roots, and may also be instrumental in K+ uptake and in K+ loading into xylem sap, affecting K+ translocation from roots to shoots (Han et al., 2016). In rice, OsHAK1 and OsAKT1 fulfill the same functions than the *Arabidopsis* counterparts HAK5 and AKT1 (Li et al., 2014; Chen et al., 2015; Nieves-Cordones et al., 2017). In addition, OsHAK5 has been related to high-affinity K+ uptake and in the release of K+ into the xylem (Nieves-Cordones et al., 2016b). **Table 2** summarizes the range of K+ concentrations in soil at which each AKT1-like channel and HAK1-like transporter contributes to high-affinity K+ uptake in *Arabidopsis* and rice (Aleman et al., 2011; Nieves-Cordones et al., 2016a).

Of note is that HKT2-like transporters have been implicated in K+ nutrition in cereals. The rice protein OsHKT2;1 provides a major pathway for root high-affinity Na+ uptake that supports plant growth under limiting K+ supply (Horie et al., 2007; Haro et al., 2010). Under K+ starvation, Na+ can partially compensate for K+ as osmoticum and change balance (Rodriguez-Navarro and Rubio, 2006; Alvarez-Aragon and Rodriguez-Navarro, 2017), and indeed high-affinity Na+ uptake has been observed in roots of several species (Haro et al., 2010). Rates of Rb+ influx between WT and *hkt2;1* roots did not reveal significant differences, but *hkt2;1* had lower Na+ contents (Horie et al., 2007). Thus, the possible involvement of OsHKT2;1 in root K+ uptake could not be verified. Although the activity of HKTs may influence the K+ status of plants, particularly under saline stress, HKTs appear to be determinants for salt tolerance with no significant role in K+ nutrition (Hamamoto et al., 2015).

#### COMPARTMENTATION AND STORAGE INTO VACUOLES

Once K+ is inside the root symplast, it may be stored in vacuoles locally, or transported to the shoot *via* xylem and accumulated in aerial tissues. Plants accumulate large amounts of K+ in their vacuoles, surpassing purely nutritional requirements. In K+ -sufficient plants, K+ content can reach up to 10% of plant dry weight, thereby exceeding the ca. 2% that supports near-maximal growth rates (White and Karley, 2010). The vacuolar K+ pool plays a chief biophysical function, i.e. the lowering of osmotic potential to draw water, generate turgor, and drive cell expansion. Because vacuoles occupy most of the intracellular volume of plant cells and are the main cellular reservoir for K+ , changes in tissue K+ concentration are largely a reflection of the dynamics of the vacuolar pool. Cytosolic K+ concentration will decline below the optimal set point of 80–100 mM only when the vacuolar K+ reserve has been depleted below the thermodynamic equilibrium with the cytosolic pool (Walker et al., 1996). Conversely, surplus K+ is placed into the vacuole to maintain cytosolic K+ within narrow limits independently of K+ abundance in the growth medium.

In contrast to the plasma membrane, accumulation of K+ in the vacuole depends on the coordinated activity of tonoplast H+ pumps and secondary K+ transporters that link K+ fluxes to the dissipation of the pH gradient or the electrical membrane potential created by the asymmetric distribution of charges. The vacuolar H+ -ATPase (V-ATPase) and the pyrophosphatase of the tonoplast (PPase) pump H+ toward the vacuolar lumen and generate pH gradients of 1–2 pH units (acidic inside) and an electrical charge (membrane potential) of 20–40 mV that is positive in the vacuolar lumen relative to the cytosol. This means that positively charged K+ ions are excluded from K+ -replete vacuoles unless transport is coupled to an energydependent uptake mechanism, whereas efflux is driven by vacuolar channels permeating K+ downhill its electrochemical gradient. A K+ /H+ antiporter energized by the pH gradient across the tonoplast was long suggested to catalyze vacuolar K+ accumulation (Walker et al., 1996; Carden et al., 2003), but the molecular identity of the underlying transporter(s) has remained elusive until recently. Vacuolar NHX-type exchangers have been shown to serve this critical function in plant cells (Venema et al., 2002; Venema et al., 2003; Leidi et al., 2010; Bassil et al., 2011; Barragan et al., 2012).

NHX exchangers were originally described as Na+ /H+ antiporters able to confer salt tolerance by driving the sequestration of excess Na+ into vacuoles (Blumwald, 2000). However, the underlying mechanism remained uncertain because the salt tolerance of transgenics overexpressing NHX proteins from various sources did not always correlate with enhanced Na+ accumulation (Jiang et al., 2010). Moreover, biochemical studies established that NHX proteins catalyze Na+ /H+ and K+ /H+ exchange with similar affinities (Venema et al., 2002). The recent meta-analysis of a large number of publications reporting tolerance phenotypes imparted by exchangers of the Cation/Proton Antiporter Family 1 (CPA1, which includes NHX proteins) concluded that the effect on K+ status was generally more pronounced than on Na+ content (Ma et al., 2017). An informative work showed that overexpression of the AtNHX1 in tomato induced K+ -deficiency symptoms despite transgenic plants having greater K+ contents than controls (Leidi et al., 2010). The intense sequestration of K+ in NHX1-overexpressing plants reduced cytosolic K+ activity, primed the induction of the high-affinity K+ uptake system, and elicited an array of metabolic and hormonal disorders related to K+ deprivation (Leidi et al., 2010; De Luca et al., 2018). Notwithstanding these unintended effects resulting from NHX overexpression, NHX proteins do increase salt tolerance, presumably because retention of cellular K+ is a requisite for adaptation to a saline environment (Jiang et al., 2010). Salinity stress elicits depolarization of the root plasma membrane and ROS production, both of which open outward-rectifying K+ channels that discharge K+ to re-build the membrane potential (Shabala and Pottosin, 2014). The salinity-induced K+ loss implies the need to replenish the cytosolic K+ pool by withdrawing K+ stored in vacuoles (Cuin et al., 2003; Leidi et al., 2010).

Deletion of *NHX1* and *NHX2* genes encoding the two major vacuolar NHX isoforms resulted in the inability to compartmentalize K+ and, surprisingly, in sensitivity to K+ supply at concentrations that did not compromise the growth of control plants (Bassil et al., 2011; Barragan et al., 2012). Moreover, *nhx1 nhx2* mutant lines showed dysfunctional stomatal activity, with impaired opening and closure (Barragan et al., 2012; Andrés et al., 2014). The rapid uptake and release of K+ and anionic organic acids by guard cells, mostly in the vacuolar compartment, drives the movements of stomata. Changes in the volume and shape of guard cells run in parallel with intense remodeling of vacuoles (Gao et al., 2005; Tanaka et al., 2007). Disruption of K+ accumulation in the guard cells of *nhx1 nhx2* mutant plants correlated with more acidic vacuoles and the disappearance of the highly dynamic remodeling of vacuolar structure associated with stomatal movements (Andrés et al., 2014).

Electrophysiological recordings of channel activities in the tonoplast have identified fast vacuolar (FV), slow vacuolar (SV), and K+ -selective vacuolar (VK) cation channels that mediate the release of vacuolar K+ (Hedrich, 2012). The VK currents have been assigned to two-pore K+ (TPK) channels (Gobert et al., 2007). TPK1, 2, 3, and 5 of *Arabidopsis* are located in the tonoplast, while TPK4 is in plasma membrane. TPK1 currents are independent of the membrane voltage but sensitive to cytosolic Ca2+ and regulated by calcium-dependent protein kinases (CDPKs) and 14-3-3 protein binding (Latz et al., 2013). Also in *Arabidopsis*, the TPC1 channel accounts for the SV current (Peiter et al., 2005). TPC1 is voltage-dependent and non-selective, allowing K+ and Na+ to permeate toward the cytosol. Whether TPC1 also permeates Ca2+ or Ca2+ is only an effector of TPC1 gating is a matter of controversy (Hedrich et al., 2018). TPC channels are activated by a decrease in transmembrane potential and increased cytosolic Ca2+, and inhibited by low luminal pH and Ca2+. Structurally, TPC1 resembles two subunits of voltage-dependent *Shaker*-like channel fused in tandem, and two cytosolic EF hands in between (Guo et al., 2016; Kintzer and Stroud, 2016). The ubiquitous nature of TPC channels and the magnitude of the SV/TPC currents are such that TPC channels are capable of contributing substantially to cellular K+ homeostasis. Accordingly, the transcriptome of the *tpc1* loss-of-function mutant of *Arabidopsis* is reminiscent of profiles that were obtained under K+ limitation (Bonaventure et al., 2007b). However, plants lacking TPC1 function are not impaired in growth and development. This may indicate that the TPC1 channel is closed most of the time and opens upon specific inputs or under stress. Current thinking is that TPC1 is part of a Ca2+/ROS relay that propagates stress signals (Choi et al., 2014; Evans et al., 2016). The gain-of-function mutant *fou2* results in a hyperactive TPC1 channel with an altered voltage-dependent gating behavior that increases the probability of the channel to be open under physiological vacuolar potentials. As a consequence of this "leaky" channel, the *fou2* mutant plant behaves as being wounded and shows elevated levels of the stress hormone jasmonate (Bonaventure et al., 2007a).

Several KT/HAK/KUP transporters have been localized to the tonoplast (**Table 1**). They are thought to force the energetically uphill release of K+ into the cytoplasm under chronic K+ starvation, in which the cytosolic concentration of K+ could be low enough to impair the discharge of luminal K+ by tonoplast channels (Ahmad and Maathuis, 2014).

#### LONG-DISTANCE TRANSPORT AND INTER-ORGAN K+ PARTITIONING

Potassium absorbed by peripheral root cells and not compartmentalized in vacuoles must be transported to the upper parts of the plant through the xylem (Gaymard et al., 1998; Park et al., 2008; Ahmad and Maathuis, 2014). This step is critical in the long-distance distribution of K+ from roots to the upper parts of the plant, and is driven by negative pressure (pulling) created by evaporation of water from leaves. The osmotic water uptake that is caused by nutrient absorption in the root also provides a positive force, known as root pressure, from roots to xylem vessels. Under regular K+ supply, symplastic K+ diffusion to the xylem through the stele may contribute sufficiently to K+ transport from root to shoot (Yang et al., 2014). Moreover, K+ is highly mobile within plants, exhibiting cycling between roots and shoots *via* xylem and phloem (Ahmad and Maathuis, 2014). In this section, we will review K+ channels, KT/HAK/KUP carriers, and HKT transporters that are involved in long-distance transport of K+ in plants.

Potassium channels SKOR and AKT2 play an important role in K+ translocation *via* xylem and phloem. SKOR (Stelar K+ Outward Rectifier) is expressed in root stele cells (pericycle and xylem parenchyma cells) of *Arabidopsis*, where it mediates K+ secretion by xylem parenchyma cells of roots and toward the xylem vessels (Gaymard et al., 1998). SKOR, being an outward-rectifying channel, opens upon membrane depolarization to allow cytosolic K+ efflux. In addition, the gating of SKOR is sensitive to extracellular K+ concentration, with a maximum activity around 10 mM K+ . In the presence of ample external K+ , the channel opens at less negative membrane voltages, thereby minimizing the risk to serve as an undesirable K+ influx pathway. This behavior of SKOR is achieved by a complex interplay between the pore region and the "S6 gating domain" localized in the last transmembrane segment, which contains three amino acid residues, D312-M313-I314, that acquire great relevance in coupling K+ sensitivity and gating of the channel (Johansson et al., 2006). At high external K+ concentration, the pore region is more rigid and strongly interacts with the S6 gating domain stabilizing the channel in a closed state. On the other hand, with a low external K+ concentration, the pore region is less occupied, more flexible, and does not interact with the surrounding transmembrane domains anymore.

The expression of *SKOR* is inhibited by abscisic acid (ABA) synthesized during water stress. This suggests that diminished K+ transport to the xylem in response to ABA allows osmotic adjustment of roots to the drying soil (Gaymard et al., 1998). Intra- and extra-cellular acidification also induced a decrease of SKOR currents at the macroscopic and single-channel levels without affecting macroscopic gating parameters and the single channel conductance. This decrease of SKOR currents could be due to a reduction in the number of channels available for activation, which could help preventing K+ loss from roots toward the shoot tissue (Lacombe et al., 2000a). Hydrogen peroxide exhibits the opposite effect on SKOR. Treatment with H2O2 increases SKOR outward currents and decreases its half activation time. Analyses in heterologous expression systems showed that SKOR sensibility to ROS is a voltage-dependent process produced by oxidation of Cys168 located on the S3 α-helix within channel (Garcia-Mata et al., 2010). Thus, upon acute depolarization of plasma membrane induced by salinity, SKOR in xylem parenchyma cells can be rapidly activated to mediate K+ loading into the xylem. After the plasma membrane potential is restored by increased H+ -ATPase activity, SKORdependent K+ release from root stelar cells to the xylem by membrane depolarization is suppressed. Then, accumulated ROS under salinity could, in turn, activate SKOR channels to allow xylem K+ loading. This may require a highly coordinated mechanism to ensure efficient xylem K+ loading in saltstressed plants.

Large quantities of K+ recirculate from roots to shoots *via* the xylem and subsequently return to the roots *via* the phloem (Thompson and Zwieniecki, 2005; De Schepper et al., 2013). The magnitude of the K+ flux recirculated from the shoots to the roots would constitute a signal by which the growing shoots could communicate to roots their K+ requirement and regulate K+ secretion into the xylem sap (and eventually root K+ uptake). *AKT2* is mainly expressed in the phloem both in leaves and roots (Deeken et al., 2000; Lacombe et al., 2000b), where the AKT2 channel protein plays a dual role by loading K+ in source tissues and unloading K+ in sink organs (Gajdanowicz et al., 2011). AKT2 is the only weak inward-rectifier characterized in *Arabidopsis* (Dreyer et al., 2001; Cherel et al., 2014). The protein phosphatase PP2CA interacts with AKT2 to induce both the inhibition of the channel current and the enhancement of its inward rectification (Cherel et al., 2002). When expressed in mammalian COS cells and *Xenopus* oocytes, AKT2 channel exhibited two gating modes that were dependent on phosphorylation by endogenous cAMP-dependent protein kinase A (PKA) (Michard et al., 2005a,b). In mode 1, the non-phosphorylated channel behaves as a weak inward-rectifier. In mode 2, phosphorylated AKT2 is permanently open and able to conduct K+ in the inward and the outward directions. Two serine residues, S210 and S329, located in the pore inner mouth that likely undergoes conformational changes on voltagedependent movements, were identified as targets for phosphorylation (Michard et al., 2005a). Nonetheless, it was proposed that post-translational modifications in these positions alone are not enough to completely convert AKT2 from an inward-rectifying to a non-rectifying channel. A lysine within the voltage sensor enables AKT2 to sense its phosphorylation status and to change between the two modes. Replacement of the lysine by serine or arginine displays an AKT2 inwardrectifier (Michard et al., 2005a; Sandmann et al., 2011). Thus, AKT2 can modulate the membrane voltage by switching between its modes of an inward or a non-rectifying channel, respectively, and phosphorylation acts as a tool for fine-tuning (Deeken et al., 2000; Michard et al., 2005b). Depending on the cellular context, the phosphorylation status of the AKT2 channels may change, enabling them to drive either inward or outward K+ fluxes (Michard et al., 2005a). Of note is that ABA reduces *SKOR* expression in the xylem while increasing that of *AKT2*

in the phloem (Pilot et al., 2003). This dual effect reduces K+ transport to the shoots and increases delivery of K+ to the roots *via* the phloem, thus helping in maintaining a low osmotic potential in water-deprived and salt-stressed roots.

The protein kinase(s) targeting AKT2 remains to be identified. The presence of PKA-like activity in plant cells is poorly documented and awaits confirmation. Held et al., 2011, demonstrated the association of AKT2 with CIPK6 and CBL4 and the effect of this assembly on macroscopic AKT2 currents. However, in contrast to the AKT1-CIPK-CBL complexes, no phosphorylation event could be detected *in vitro*. Instead, it was proposed the Ca2+-dependent targeting of AKT2 to the plasma membrane depended solely on the physical interaction of AKT2 with CIPK6/CBL4. The plasma membrane localized receptor-like pseudo-kinase MRH1/MDIS2 also interacts with AKT2 (Sklodowski et al., 2017). MRH1 appears to be essential for AKT2 function since the phenotype of *mrh1-1* and *akt2* knockouts were similar in energy-limiting conditions. However, electrophysiological analyses showed that MRH1 did not affect AKT2 transport. Moreover, the putative kinase domain of MRH1 lacks essential sites that are indispensable for a functional kinase suggesting that MRH1 is a pseudo-kinase and that MRH1 and AKT2 are parts of a bigger protein complex in which MRH1 might help to recruit other unknown partner(s), that might post-translationally regulate AKT2 (Sklodowski et al., 2017).

Because phloem comprises living cells, the K+ content in phloem is inherently high (50–150 mM) and the pH is near neutrality (Ahmad and Maathuis, 2014). Consequently, secondary transport in and out the sieve tubes cannot be extensively linked to the H+ -motive force. K+ and photoassimilates are loaded together in source tissues and downloaded in sinks. At source tissues, H+ -coupled sucrose transporters load the sugar into the phloem. The influx of H+ leads to membrane depolarization of the phloem cells, thereby reducing the driving force for further sucrose loading. Depolarization is prevented by the release of K+ by AKT2 (Deeken et al., 2002). Thus, K+ in the phloem stimulates sugar loading into the phloem sap. Moreover, the post-translational regulation of AKT2 channel activity described above might play a role in the fine-tuning of photoassimilate distribution within the plant by way of controlling the membrane potential through the modulation of K+ fluxes into the phloem (Michard et al., 2005a; Gajdanowicz et al., 2011). Accordingly, the expression level of *AKT2* increases in the presence of light and CO2 assimilates (Deeken et al., 2000; Lacombe et al., 2000b). In summary, through reversible posttranslational modifications, AKT2 taps a "potassium battery" providing additional energy for transmembrane transport processes besides energization by the plasma membrane H+ -ATPase.

Members of the KT/HAK/KUP family, e.g. AtKUP7 and OsHAK5, have been proposed to facilitate long-distance K+ transport from root to shoot, presumably by mediating K+ uptake into the xylem parenchyma cells (Yang et al., 2014; Han et al., 2016). This function of KT/HAK/KUP transporters would be relevant under K+ deprivation, when apoplastic K+ levels could be below the operational range of channels. In rice, OsHAK1 (as well as OsAKT1) seems also to participate in the root-to-shoot transfer of K+ and grain yield (Chen et al., 2015); however, it is likely that OsHAK5 dominates K+ translocation from roots to shoots at low-K+ supply. A role in K+ transfer from root to shoot has been also proposed for OsHAK21 under salt stress condition, but not under K+ starvation situations (Shen et al., 2015). The *Vitis* VvKUP1, VvKUP2, and the voltagegated channels VvK1.1 and VvK1.2 have been described to function in K+ accumulation during grape berry development (Davies et al., 2006; Cuellar et al., 2010; Cuellar et al., 2013).

As mentioned earlier, HKT channel-like proteins are primarily involved in Na+ fluxes both in roots (monocots) and vascular bundles (monocots and dicots) (Hamamoto et al., 2015). However, they often have a significant impact in maintaining high K+ /Na+ ratio in aerial parts during salinity stress and genetic diversity in HKT proteins meditating long-distance transport of Na+ and K+ have a great impact on the salt tolerance of cereals (Ren et al., 2005; Munns et al., 2012; Zhang et al., 2018). Notwithstanding the above generalization, the class-II HKT transporter of maize ZmHKT2 used K+ as the preferred substrate, was mainly expressed in the root stele, and regulated root-to-shoot K+ delivery. Domain-swapping between natural variants of ZmHKT2 imparting contrasting salt tolerance indicated that the amino acid variant A130G accounted for differential rates of K+ transport to shoots (Cao et al., 2018). On the other hand, mutants in *AtHKT1;1* of *Arabidopsis* and *OsHKT1;5* of rice accumulated significantly less K+ in shoots and xylem sap under salinity stress despite the fact that these transporters are Na+ -selective (Uozumi et al., 2000; Sunarpi et al., 2005; Kobayashi et al., 2017). The uptake of Na+ into xylem parenchyma cells by AtHKT1;1 and OsHKT1;5 possibly causes depolarization of the plasma membrane that triggers K+ secretion into the xylem vessel *via* outward-rectifying K+ efflux channels (Hauser and Horie, 2010). Support for this proposal still requires the analysis of genetic interactions between *hkt1*-like and *skor*-like mutants.

K+ is the preferred counter ion for root-to-shoot translocation of NO3 − in the xylem of crops and *Arabidopsis* (Engels and Marschner, 1993; Zhang et al., 2010; Rodenas et al., 2017). NRT1.5, a member of the Nitrate Transporter 1/Peptide Transporter Family (NPF7.3), is important for the NO3 − -dependent K+ translocation in *Arabidopsis* (Lin et al., 2008; Drechsler et al., 2015; Meng et al., 2016). Lack of NRT1.5 resulted in K+ deficiency in shoots under low NO3 − availability, whereas the root elemental composition was unchanged (Lin et al., 2008; Drechsler et al., 2015). Mutant analyses revealed that both NRT1.5 and SKOR contributed additively to K+ translocation; SKOR activity was dominant under high NO3 − and low K+ supply, and NRT1.5 was required under low NO3 − (Drechsler et al., 2015; Li et al., 2017). Accordingly, the *Arabidopsis* mutant *lks2*, unable to grow in low-K+ , is a loss-of-function mutant in *NRT1.5* (Li et al., 2017). Together, these data indicate that NRT1.5 facilitates K+ release out of root parenchyma cells and K+ loading into xylem vessels. NRT1.5 is a plasma membrane protein that in *Xenopus* oocytes behaved as a low-affinity, pH-dependent bidirectional nitrate transporter (Lin et al., 2008). Surprisingly, NRT1.5 has also been shown to release K+ from *Xenopus* oocytes and yeast in a pH-dependent manner, and has been proposed to function as a K+ /H+ antiporter (Li et al., 2017. If confirmed by additional research, the data of Li et al. (2017) imply that the linkage between NO3 − and K+ transport is more intimate than the mere balancing of charges as previously thought.

#### CO-REGULATION OF K+ AND NITROGEN UPTAKE

Plants take up numerous mineral nutrients from the soil; some of them are essential (as K+ or NO3 − ), while others can be toxic at high concentrations (as Na+ or NH4 + ). Adaptive responses to varying mineral nutrient conditions in the soil, particularly low-nutrient environments, involve multiple signaling pathways whose integration allows plants to grow and adjust their development to each specific nutritional situation (Kellermeier et al., 2014). Thus, changes in the concentration of one nutrient trigger a signaling cascade that modify not only the amount, localization, and/or activity of this nutrient-specific transporter/channel, but also transporters/ channels related with other nutrients. N-K interactions are important for root architecture (Kellermeier et al., 2014). In the previous section, we have discussed the linkage of NO3 − and K+ in long-distance transport. Here, we present the recent knowledge gained about the coordinated regulation of K+ and NO3 − uptake and nutrition.

K+ starvation is required for triggering high-affinity HAK5 mediated K+ uptake in roots of *Arabidopsis* and tomato. However, limitation of K+ , N, or P, all induced hyperpolarization of the plasma membrane of root cells and enhanced *HAK5* transcription (Rubio et al., 2014), a response that could be due to maintenance of electrical balance since single N and P starvation, probably resulting in lower NO3 − and PO4 3− contents, led to a concomitant reduction of the K+ content (Kellermeier et al., 2014). Alternatively, the transport of a nutrient could become inhibited if another nutrient is limiting growth (Nieves-Cordones et al., 2019). In line with this, NO3 − , PO4 3−, and SO4 2− deficiencies reduced root K+ uptake (Rodenas et al., 2017). Furthermore, comparison of the transcriptional responses to single or multiple nutrient deprivations showed that N starvation had a dominant effect over P and K starvation. In other words, the transcriptional landscape of combined K+ and N limitation was mainly driven by the N-starvation response.

For most plants, nitrate (NO3 − ) and ammonium (NH4 + ) are the two major nitrogen sources (Crawford, 1995; Gazzarrini et al., 1999). In general, in aerobic soil conditions, nitrate is the primary nitrogen source, while under anoxic conditions ammonium is (Ho and Tsay, 2010). To be assimilated, NO3 − has to be taken up from the soil and converted into ammonium by nitrate and nitrite reductases, and then incorporated into amino acids *via* the glutamine-synthetase and glutamatesynthase (GS-GOGAT) pathway. Therefore, ammonium is the preferred nitrogen source in plants, but ammonium uptake *via* the roots is tightly controlled because elevated ammonium concentrations in the cytosol are toxic (Straub et al., 2017). The mechanisms underlying ammonium toxicity are not fully understood, but acidification of the external environment, disruption of the acid/base balance, and the energy lost exporting excess ammonium may be key factors. NO3 − and NH4 + uptake systems involve different families of proteins and a complex regulation not fully understood yet. Here, we will focus on *Arabidopsis* dual-affinity NO3 − transporter AtNRT1.1 (Nitrate Transporter 1, also called AtCHL1 and AtNPF6.3) and high-affinity NH4 + transporters AtAMT1, because they share part of their post-translational regulation with AtAKT1 and AtHAK5.

The CIPK23/CBL1,9 protein kinase complex is key factor in the coordination of plant nutrition, regulating iron, NO3 − , and K+ uptakes (Ho et al., 2009; Ragel et al., 2015; Straub et al., 2017; Dubeaux et al., 2018). The transport and regulatory protein AtNRT1.1 is involved in both high-affinity and low-affinity nitrate uptake. Unphosphorylated AtNRT1.1 is a low-affinity nitrate transporter working as a dimer, and its phosphorylation by CIPK23/CBL1,9 leads to dimer dissociation. Phosphorylated AtNRT1.1 monomers show a higher nitrate affinity than the dimers (Ho et al., 2009; Sun et al., 2014). On the other hand, AtAMT1 ammonium transporters work as trimers and the phosphorylation by CIPK23/CBL1 (and not CBL9) of a single monomer exhibits an allosteric effect, leading to the cooperative closure of all three pores in the trimer (Straub et al., 2017). Together, these data indicate that CIPK23 and CBL1 are major regulators of NO3 − , K+ , and NH4 + homeostasis in *Arabidopsis*. As **Figure 4** shows, under K+ - and NO3 − -sufficient and NH4 + moderate (non-toxic) conditions, CIPK23 is in the cytoplasm and inactive because of the interaction with protein phosphatases of the PP2C family (Lee et al., 2007; Cherel et al., 2014). In this situation, AtNRT1.1 will work as dimeric low-affinity NO3 − transporter and AtAMT1 trimers will be active, unphosphorylated AtAKT1 will be less active and *AtHAK5* will be transcriptionally repressed. Thus, N demand is covered by NO3 − and NH4 + uptake, and K+ is supplied by AKT1. Under K+ and/or NO3 − low concentrations or toxic NH4 + conditions, AtCIPK23 is recruited by AtCBL1 or AtCBL9 to the plasma membrane, which allows the phosphorylation of CIPK23 target transporters. Phosphorylated AtAMT1 trimers are inactive and N demand is covered by high-affinity NO3 − transport through phosphorylated AtNRT1.1 monomers. Phosphorylated AtAKT1 will enhance K+ influx, serving as a balancing counter ion for NO3 − . If the K+ concentration is low enough, *AtHAK5* will be transcribed and the AtHAK5 protein activated by CIPK23/CBL1,9 complex.

Formation of second-order lateral roots in *Arabidopsis* was significantly stimulated by K+ starvation, but low-N inhibited this effect of low-K+ . Mutation of *AKT1* or *CIPK23* also cancelled the formation of lateral roots under low-K+ (Kellermeier et al., 2014). These nitrate-specific effects occurred over a concentration range that triggers phosphorylation of NRT1.1 by CIPK23. These results suggest that N and K+ availability determines root architecture and that CIPK23 serves as the regulatory node acting through both AKT1 and NRT1.1.

It is noteworthy that AtNRT1.1 has been described as a transceptor (i.e., a transporter that is also serving as sensor/ receptor for its substrate) (Ho et al., 2009; Wang et al., 2009), but its uptake activity is not required for the sensing function.

As a NO3 − sensor, when plants are exposed to high concentrations of NO3 − , dephosphorylated AtNRT1.1 leads to the primary NO3 − -response, consisting in the rapid expression of nitrate assimilatory enzymes and nitrate transporters to prepare the plant to assimilate NO3 − . In response to low concentrations of NO3 − , CIPK23 phosphorylates AtNRT1.1 that in turn will downregulate the primary response. Therefore, the phosphorylation status of AtNRT1.1 not only switches the transport modes, but also induces different levels of primary nitrate response, and CIPK23 works as a negative regulator of the primary nitrate response. How the phosphorylation status of AtNRT1.1 modulates the transcriptional levels remains unknown (Ho and Tsay, 2010). Transcriptional responses to low-N showed a strong additive effect by low-K+ (Kellermeier et al., 2014). Moreover, NO3 − deprivation increased *AtHAK5* and *LeHAK5* expression to a similar level than K+ deprivation (Rubio et al., 2014). The transcription factors and signaling pathways involved in this coordinated response to nutrient deprivation have not been characterized.

Finally, as mentioned before, the transporters of the NPF group AtNRT1.5 and AtNRT1.8 may transport K+ besides or instead NO3 − (Li et al., 2017). AtNRT1.5 plays a crucial role in K+ translocation from root to shoot and is involved in the coordination of K+ /NO3 − distribution in plants (Drechsler et al., 2015). Furthermore, the putative orthologs of AtNRT1.5 from rice (OsNPF7.9) and maize (ZmNPF7.10) can also function as K+ efflux transporters (Li et al., 2017), suggesting that the K+ transport function of NRT1.5-like proteins is conserved in vascular plants. These results lay an additional layer to the coordinated transport of N and K+ besides the co-regulation by CIPK23 of NO3 − - and K+ -specific transporters.

# CONCLUSIONS AND PERSPECTIVES

Although plant responses to K+ starvation are well documented at the physiological and transcriptional levels, the sensing and regulatory mechanisms underlying these changes need to be clarified. Despite all the progress made regarding the regulation of individual K+ transporters and the signaling pathways involved, how depletion of cellular K+ is sensed remains poorly understood. Hyperpolarization of the plasma membrane under limiting K+ seems to be a key factor in the transcriptional activation of genes encoding high-affinity K+ uptake, but other early signaling events activating the known cascades of phytohormones, ROS production, and posttranslational modifications of K+ transport proteins remain unclear. It has been proposed that AKT1 may act as a K+ sensor in the root architecture response to nutrient supply, possibly by linking auxin transport to plasma membrane potential (Kellermeier et al., 2014). Moreover, it is also worth noting that transient and localized cytosolic K+ spikes have been proposed to work, together with Ca2+ and ROS waves, as messengers that signal and shape plant adaptive responses to stress (Shabala, 2017). Nonetheless, fundamental questions regarding the sensing of and responses to K+ deprivation are of central importance for plant nutrition and deserve additional research. The availability of novel genetically encoded K+ sensors (Bischof et al., 2017) that could be targeted to various cellular compartments of predefined cells and tissues will be powerful tools to monitor the dynamics of cellular K+ with unprecedented spatiotemporal resolution.

#### AUTHOR CONTRIBUTIONS

All authors have contributed to literature search, discussion, and writing of the manuscript. PR and JP assembled all sections. PR and FQ prepared the Figures. All authors checked and approved the manuscript.

#### REFERENCES


#### FUNDING

This work was supported by grant BIO2015-70946-R to FQ, and by grants BFU2015-64671-R and BIO2016-81957-REDT from AEI-MINECO (co-financed by the European Regional Development Fund), and the SSAC grant PJ01318205 from the Rural Development Administration, Republic of Korea, to JP.

#### ACKNOWLEDGMENTS

We acknowledge support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI).


into the phloem of Arabidopsis. *Planta* 216, 334–344. doi: 10.1007/ s00425-002-0895-1


pathway abundance and novel protein functions. *Curr. Biol.* 14, 354–362. doi: 10.1016/j.cub.2004.02.039


translocation in tomato and Arabidopsis plants. *Physiol. Plant.* 160, 410–424. doi: 10.1111/ppl.12558


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

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

# Submergence and Waterlogging Stress in Plants: A Review Highlighting Research Opportunities and Understudied Aspects

#### Takeshi Fukao<sup>1</sup> , Blanca Estela Barrera-Figueroa<sup>2</sup> , Piyada Juntawong<sup>3</sup> and Julián Mario Peña-Castro<sup>2</sup> \*

<sup>1</sup> School of Plant and Environmental Sciences, Virginia Tech, Blacksburg, VA, United States, <sup>2</sup> Laboratorio de Biotecnología Vegetal, Instituto de Biotecnología, Universidad del Papaloapan, Tuxtepec, Mexico, <sup>3</sup> Center for Advanced Studies in Tropical Natural Resources, National Research University – Department of Genetics, Faculty of Science, Kasetsart University, Bangkok, Thailand

#### Edited by:

Dae-Jin Yun, Konkuk University, South Korea

#### Reviewed by:

Angelika Mustroph, University of Bayreuth, Germany Markus Teige, University of Vienna, Austria

#### \*Correspondence:

Julián Mario Peña-Castro julianpc@unpa.edu.mx; julianp@prodigy.net.mx

#### Specialty section:

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

Received: 17 November 2018 Accepted: 05 March 2019 Published: 22 March 2019

#### Citation:

Fukao T, Barrera-Figueroa BE, Juntawong P and Peña-Castro JM (2019) Submergence and Waterlogging Stress in Plants: A Review Highlighting Research Opportunities and Understudied Aspects. Front. Plant Sci. 10:340. doi: 10.3389/fpls.2019.00340 Soil flooding creates composite and complex stress in plants known as either submergence or waterlogging stress depending on the depth of the water table. In nature, these stresses are important factors dictating the species composition of the ecosystem. On agricultural land, they cause economic damage associated with longterm social consequences. The understanding of the plant molecular responses to these two stresses has benefited from research studying individual components of the stress, in particular low-oxygen stress. To a lesser extent, other associated stresses and plant responses have been incorporated into the molecular framework, such as ion and ROS signaling, pathogen susceptibility, and organ-specific expression and development. In this review, we aim to highlight known or suspected components of submergence/waterlogging stress that have not yet been thoroughly studied at the molecular level in this context, such as miRNA and retrotransposon expression, the influence of light/dark cycles, protein isoforms, root architecture, sugar sensing and signaling, post-stress molecular events, heavy-metal and salinity stress, and mRNA dynamics (splicing, sequestering, and ribosome loading). Finally, we explore biotechnological strategies that have applied this molecular knowledge to develop cultivars resistant to flooding or to offer alternative uses of flooding-prone soils, like bioethanol and biomass production.

Keywords: hypoxia, anoxia, biotechnology, cell signaling, stress perception, submergence, waterlogging

# INTRODUCTION

The interaction of plants with the environment is a continuous process in which water plays a central role. At a global scale, indicators such as potential evapotranspiration and the number of wet days per year (amount and temporal occurrence of rainfall) directly determine the distribution of plants and species richness in a geographical context (Kreft and Jetz, 2007). In addition to

**118**

rainfall, depth of the groundwater table, which is the interface between oxygenated soil and water-saturated aquifers, has a significant influence on maintaining aquatic ecosystems in dry periods, impeding drainage and defining wetland habitats for plants adapted to waterlogged soils (Fan et al., 2017). Together, rainfall and groundwater table depth influence soil hydrology, controlling physiological characteristics in plants to adapt to changes in water availability, such as rooting depth (Fan et al., 2017).

Agriculture around the world is constantly challenged by the rising incidence of adverse weather events as a consequence of global warming. Extreme events that alter water availability, like droughts and floods, constitute big threats to food security (Food and Agriculture Organization of the United Nations [FAO], 2017). Flooding of agricultural fields is generally caused by intensive and/or extensive rainfall over a period of time, but it may also result from overflowing of a body of water over land. Floods have a negative impact on economic and social aspects. Floods cause loss of livestock and seed stocks, destruction of infrastructure, machinery and tools, food shortage, diseases, and loss of agricultural productivity. This makes affected farmers and the general population vulnerable, causing poverty and migrations (Saldaña-Zorrilla, 2008).

On a global scale, floods were the cause of almost two-thirds of all damage and loss to crops in the period between 2006 and 2016, with a value of billions of dollars (Food and Agriculture Organization of the United Nations [FAO], 2017). For example, in Pakistan, extreme monsoon rains between 2010 and 2014 caused floods on large extensions of land that were responsible for the significant loss of at least 11 billion tons of rice, sugarcane, maize, and cotton, contributing to a total economic loss of over 16 billion dollars (Rehman et al., 2015). In addition, depending on the extent of the damage, resuming farming after a flooding event may require following an efficient recovery plan in order to remediate productive lands, which includes removing sediments, repairing physical and nutrient soil properties, reverting loss of beneficial soil microbial activity, and if the plant damage was lethal, replanting (Singh et al., 2013; Rey et al., 2019).

Based on the height of the water column produced, flooding can be classified as waterlogging, when it is superficial and covers only the root, or as submergence, when water completely covers the aerial plant tissues (Sasidharan et al., 2017). Both types of flooding disrupt the movement of oxygen from the air to plant tissues (Lee et al., 2011), producing a natural condition known as hypoxia (<21% O2) (Sasidharan et al., 2017).

At the agronomic level, there are different strategies to cope with submergence or waterlogging, such as developing standard models for the prediction and assessment of crop loss to floods for risk management, economic insurance, and decision making. Systems have been designed to predict flood-affected and flooddestroyed crop areas based on correlations to standard climate indices (Zhang Q. et al., 2016) or to estimate flooded crop acreage, crop damage and flood frequency and use algorithms to support post-flood crop loss estimation (Di et al., 2017). These systems constitute useful tools for planning, managing and applying solutions for the mitigation of damage to crops caused by floods.

In addition, the use of biological knowledge to breed superior cultivars can act as "genetic insurance" to guarantee a superior yield in the face of a flooding event, or even in multi-stress conditions (Mickelbart et al., 2015). In this sense, an excessive water supply induces hypoxia in plants (Lee et al., 2011), increases the vulnerability to pathogen attack (Hsu et al., 2013), and limits the flow of light to the plant (Jackson and Ram, 2003). During recovery after a flooding event, plants experience oxidative stress (Yeung et al., 2018) and must remobilize nutrients to achieve a normal homeostatic state (Tsai et al., 2016). Plants respond to flooding and the associated stress by changes in gene expression that are finely regulated at a multilevel scale from epigenetics (Tsuji et al., 2006) to transcriptional (Mustroph et al., 2010; Lee et al., 2011) and translational regulation (Juntawong et al., 2014a).

In recent years, flooding stress and its derivatives like submergence, waterlogging, hypoxia, and anoxia, were investigated extensively in plants, mainly in Arabidopsis and rice, to identify molecular elements that may play a role in tolerance to flooding. In this review, we intend to highlight aspects of this compound stress that present a niche of opportunity for expanding our knowledge and identifying genomic determinants in order to develop climate-resilient varieties that can meet the food demands of a rising world population.

#### ADVANCES ON STRESS MOLECULAR SENSING AND SIGNALING

Submergence and waterlogging trigger dramatic changes in gene expression, which coordinate morphological and metabolic adaptations to the stress (Bailey-Serres and Voesenek, 2008; Mustroph et al., 2010). Group VII ethylene response factors (ERF-VIIs) is a class of ERF transcription factors (TFs) that regulate the expression of a wide range of genes involved in adaptive responses to flooding and low oxygen (Voesenek and Bailey-Serres, 2015). In Arabidopsis, five ERF-VII genes, HRE1, HRE2, RAP2.2, RAP2.3, and RAP2.12, were recognized as key regulators for flooding and low-oxygen tolerance. These ERF-VIIs regulate a similar set of hypoxia-responsive genes, but constitutively expressed RAP2.2, RAP2.3, and RAP2.12 are more powerful activators than stress-inducible HRE1 and HRE2 according to transactivation studies (Bui et al., 2015; Gasch et al., 2016). In rice, an ERF-VII TF gene, SUB1A, is known as the master regulator of submergence tolerance, allowing plants to endure complete submergence for 14–16 days (Fukao et al., 2006; Xu et al., 2006). Other rice ERF-VII genes, SNORKEL1 and SNORKEL2, were found specifically in deepwater rice (Hattori et al., 2009). These tandemly repeated ERF-VIIs are responsible for enhanced internode elongation under flooded conditions, enabling plants to outgrow floodwaters. The importance of ERF-VIIs in flooding responses and tolerance is also indicated in Rumex and Rorippa, dicot species from flood-prone environments (van Veen et al., 2014).

The stability of most ERF-VII proteins is regulated by the N-end rule of proteolysis (NERP; Gibbs et al., 2011; Licausi et al., 2011a; van Dongen and Licausi, 2015). This sequential pathway

has been recognized as a direct oxygen-sensing mechanism in plants because one of the enzymes in this process, plant cysteine oxidase (PCO), requires molecular oxygen as a cosubstrate (White et al., 2017). The distinct expression patterns, ERF-VII selectivity, and catalytic capability of five Arabidopsis PCO isoforms suggest that they may have different biological roles in stress response and developmental regulation (White et al., 2018). Under ambient oxygen conditions, ERF-VII proteins are constitutively degraded via the N-end rule pathway. Under low oxygen, however, the oxygen-dependent PCO reaction is inhibited, resulting in the escape of ERF-VII proteins from targeted proteolysis. All Arabidopsis ERF-VIIs are substrates of the N-end rule pathway due to their conserved N-terminal MC motif (MCGGAI) that is recognized by the NERP pathway enzymes (Bailey-Serres et al., 2012; White et al., 2018). However, some rice ERF-VIIs such as SUB1A and SUB1C are not degraded via this pathway in vitro (Gibbs et al., 2011). In the case of SUB1A, the N-degron is protected from recognition through proteinprotein interactions, and act during submergence through other downstream ERF-VIIs that are transcriptionally activated and susceptible to degradation by the NERP pathway (Lin et al., 2019). The stability of SUB1A protein in the presence of oxygen reflects enhanced recovery after de-submergence and increased tolerance to post-submergence injury in rice genotypes carrying SUB1A (Fukao et al., 2011; Alpuerto et al., 2016). SUB1C does not have an MC motif. Both SNORKEL ERFs have a minimum MC motif (Gibbs et al., 2011) but they have not been investigated as NERP substrates.

A recent study revealed that accumulation of the AtRAP2.12 protein (an Arabidopsis ERF-VII) is enhanced by a Raf-like mitogen-activated protein kinase kinase kinase (MAPKKK), HCR1, under combined oxygen-deprived and K+-sufficient conditions (Shahzad et al., 2016). Without exogenous application of K+, the effect of HCR1 on RAP2.12-mediated hypoxia responses, such as increased expression of core hypoxia-response genes (CHG) and reduced root hydraulic conductivity, was not observed even under oxygen deprivation. Under flooded conditions, the concentration of K<sup>+</sup> in the soil is substantially lowered, and the uptake of K<sup>+</sup> by plants is also restricted due to reduced hydraulic conductivity. Thus, this K+-dependent process under hypoxia suggests that low oxygen and flooding can increase internal K<sup>+</sup> concentration in specific tissues and celltypes due to altered ion channel activities. Low oxygen regulates the function of various K<sup>+</sup> channels in mammals (Wang F. et al., 2017). Further studies are required to elucidate the significance of K <sup>+</sup> regulation in ERF-VII-mediated low-oxygen responses.

Another mechanism that controls the participation of ERF-VIIs in transcriptional activation is sequestration of the TFs to the plasma membrane (Licausi et al., 2011a). Under normoxia, RAP2.12 protein physically interacts with plasma membrane-localized acyl-CoA-binding proteins, ACBP1 and ACBP2, allowing RAP2.12 to escape from the N-end rule pathway (**Figure 1A**). In contrast, low oxygen promotes the removal of RAP2.12 from the plasma membrane, leading to relocalization of the TF to the nucleus (Kosmacz et al., 2015).

The transactivation capability of RAP2.12 is also regulated by a hypoxia-inducible trihelix TF, HRA1 (Giuntoli et al., 2014). HRA1 physically binds to RAP2.12, reducing the expression of CHGs. Interestingly, RAP2.12 stimulates mRNA accumulation of HRA1, indicating that HRA1 and RAP2.12 are regulated by a negative feedback loop. It was proposed that this regulatory mechanism can contribute to the avoidance of RAP2.12 overaccumulation, preventing rapid depletion of carbohydrate reserves under oxygen deprivation. Indeed, artificial stabilization of RAP2.12 by N-terminal modification (N-end rule insensitive mutation) reduced tolerance to low oxygen in Arabidopsis (Paul et al., 2016), emphasizing the significance of fine-tuning ERF-VIIs in hypoxia adaptation.

Many other protein–protein interactions of ERF-VIIs are described in the literature, opening a research area that may render much information (**Figure 1B**). SUB1A was reported to have dozens of protein partners in rice named Sub1A Binding Proteins (SAB; Seo et al., 2011). Only two of those interactions were characterized, with SAB23 and SAB18, and it was concluded that they improve the crosstalk between submergence stress and pathogen defense and the modulation of elongation, respectively (Seo et al., 2011). SUB1A also interacts in a positive feedback loop with mitogen-activated protein kinase 3 (MAPK), a central regulator of primary signaling cascades (mainly ROS), and allows submergence to suppress gibberellin action to inhibit elongation under submergence (Singh and Sinha, 2016). Fukao and Bailey-Serres (2008) demonstrated that SUB1A enhances inhibition of gibberellic acid (GA) signaling by the up-regulation of rice DELLA proteins, known negative regulators of GA. Recently, it was demonstrated that the Arabidopsis ERF-VII RAP2.3 physically interacts with DELLA proteins through a domain located on the N end of RAP2.3, regulating apical hook development (Marín-de la Rosa et al., 2014). When domains are analyzed in the structure of Arabidopsis ERF-VIIs, a division in low and high domain complexity can be observed (**Figure 1C**) (Nakano et al., 2006; Bui et al., 2015; Papdi et al., 2015). If we add that some of these ERFs are under transcriptional (Syed et al., 2015; Rivera-Contreras et al., 2016) and translational regulation by light (Loreti et al., 2018), we have a scenario of high versatility of potential protein–protein interactions providing the plant cell with ERF-VII regulation plasticity, but which has yet to be fully understood.

### OXIDATIVE STRESS AND ENERGY SIGNALING AND MANAGEMENT

#### Stress/Post-stress Management of Oxidative Stress Symptoms

During flood events, plants are partially or completely submerged in water, but as floodwaters subside, plants suddenly encounter reoxygenation. Reoxygenation has been recognized as an abiotic stress that can injure plants post-submergence. For example, when rice plants were de-submerged from 7 days of inundation, an indicator of oxidative damage, malondialdehyde (MDA), along with superoxide anion and H2O2, was accumulated in the leaves (Fukao et al., 2011). Re-exposure to atmospheric oxygen after 7–10 days of submergence also induced leaf

GAI, gibberellic acid insensitive; GDH, glutamate dehydrogenase; HB, hemoglobin; HCR1, hydraulic conductivity of the root; HRE1, hypoxia response attenuator; Int., interactor protein; MPK3, MAP kinase 3; NiR, nitrite reductase; NO, nitric oxide; NTR, N-terminal route; PCO, plant cell oxidase; SAB, Sub1A binding.

dehydration in rice (Setter et al., 2010; Fukao et al., 2011). These observations indicate that a plant's survival of a flood event requires tolerance to multiple stresses such as submergence, reoxygenation, and dehydration. Indeed, the major submergence tolerance regulator in rice, SUB1A, confers tolerance to oxidative stress and dehydration through activation of ROS detoxification

RAP2.3, a low-complexity ERF-VII TF (others being HRE1 and HRE2), and details of demonstrated domain functions. 2-OG, 2-oxoglutarate; ACBP, acyl-CoA binding protein; ALAT, alanine aminotransferase; ARC, amidoxime reducing component; CHGs, hypoxia core genes; ERF, ethylene response factor; FAE, fatty acid elongase;

and abscisic acid (ABA) responsiveness (Fukao et al., 2011). Similarly, in Arabidopsis many ERF-VIIs are involved in the adaptation to submergence, oxidative stress, and dehydration (Tamang and Fukao, 2015). A recent study demonstrated that jasmonate is a pivotal hormone to activate ROS detoxification systems under reoxygenation in Arabidopsis (Yuan et al., 2017). Exogenous application of jasmonate increased tolerance to reoxygenation, whereas jasmonate-deficient mutants exhibited intolerance to this stress.

#### Nitric Oxide as a Central Homeostatic Regulator of Hypoxic Stresses

The accumulated knowledge on the biological roles of nitric oxide (NO) in plants fits well with a modern model where ROS are increasingly recognized as an integral part of the cellular signaling process (Foyer et al., 2017). This view is reinforced when taking into account the environment in which actual flooding/submergence stresses occur. In the field, plants still must balance the presence and absence of varying degrees of illumination. That rarely will create a complete shutdown of aerobic and anabolic pathways in favor of fully anaerobic and catabolic pathways, but rather a complex situation where balance is intended. In addition, the stress will not arrive as a sudden insult; it will build up gradually around the plant.

Recent research in the unicellular alga Chlamydomonas reinhardtii highlighted the enzyme nitrate reductase (NR) as a versatile platform to achieve this balance. Using genetic and pharmacological tools, it was demonstrated that NR provides the structure for a short, cytosolic electron transport chain that is able to distribute electrons, not only through its canonical pathway from NAD(P)H to nitrite, but also to NO through the interacting protein amidoxime reducing component (ARC), and to NO and O<sup>2</sup> through hemoglobin (HB) to form a cycle back to nitrate (Chamizo-Ampudia et al., 2016) (**Figure 1A**). These authors proposed a hypothetical structural model that would spatially accommodate all these alternative electron transfers in a dimeric NR association (Chamizo-Ampudia et al., 2017).

Interestingly, blocking NO production through the genetic knockout of NR genes is sufficient to stabilize ERF-VII proteins even under normoxic conditions, indicating that NR is the main source of NO in plant cells (Gibbs et al., 2014). Therefore, being a cytoplasmic complex interacting with soluble and diffusible electron carriers (Chamizo-Ampudia et al., 2016) with gene homologs in both dicot and monocot model plants (Tejada-Jiménez et al., 2013), NR/ARC/HB would be able to connect most of the known hypoxia/anoxia responses such as glycolysis, N assimilation, NO signaling, and ERF-VII activity (**Figure 1A**). The finding that Arabidopsis ARC is able to catalyze the in vitro production of NO (Yang et al., 2015) opens the possibility of further exploration of the biochemistry and applications of the NR/ARC/HB electron chain transport.

Acting as a meeting hub of survival strategies, NR/ARC/HB has several levels of plasticity. Transcriptionally, both in Arabidopsis and Brachypodium, NR transcripts are downregulated in shoots by submergence, HB transcripts are strongly up-regulated and are an integral part of the CHGs, and ARC transcripts (At1g30910, At5g44720, Bradi1g32350, and Bradi4g37990) are mildly down-regulated (Lee et al., 2011; Rivera-Contreras et al., 2016). However, in tomato and tobacco, despite the down-regulation of NR mRNA under root hypoxia, the activity of NR is increased to a level where external nitrate protects against stress damage even in aerial parts (Allègre et al., 2004). On the other hand, excessively low NR activity in tobacco roots leads to high lactic acid and ethanolic fermentation, but without the protective effect of nitrite against H<sup>+</sup> cytoplasmic acidification (Stoimenova et al., 2003; Libourel et al., 2006). At the protein level, these observations indicate a stress scenario where the core NR structure production is limited, its activity maintained, and its two alternative electron acceptors, ARC and HB (paradoxically working in sequence), remain to compete with a strong balance toward NO removal and the expected consequence of homeostatic regulation of anaerobic metabolism (**Figure 1A**). In Jatropha curcas this logic is not followed since NR transcripts are up-regulated and this was considered a probable cause of waterlogging sensitivity (Juntawong et al., 2014b).

The abundance of NR protein under normoxia is positively regulated by nitrate and negatively by ammonia (Kim et al., 2018). These observations concur with measurements in hypoxic roots where nitrate is actively reduced to nitrite and NO, but this activity is inhibited by external ammonia, and as a consequence fermentative lactate production increases (Oliveira et al., 2013). Studies have shown that, during hypoxia, both nitrate and ammonia tend to increase; ammonia improves plant fitness, indicating the need to balance the canonical NR activity to support nitrogen fixation (Allègre et al., 2004).

The stability of NR protein in normoxia is also under different post-translational controls with positive effects like sumoylation, or negative ones like phosphorylation and proteasome degradation (Park et al., 2011), and intriguingly, by partitioning its subcellular location to the nucleus (Kim et al., 2018). Interestingly, it was found that NR is destabilized by CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1), a ubiquitin ligase (Park et al., 2011). In Rumex palustris, a plant that can tolerate submergence, higher COP1 and HB expression were measured concomitant with lower ammonia and higher nitrate and protein concentrations (even in normoxia) when compared to Rumex acetosa, which is sensitive to submergence, thus indicating that posttranslational suppression of NR may be a relevant tolerance mechanism (van Veen et al., 2013). Therefore, NR/ARC/HB provides an interesting genetic platform to develop biotechnological strategies for improving plant performance under stress (Llamas et al., 2017).

The biological relevance of the NO cycle (Igamberdiev and Hill, 2004) to fix NO by NR and channel it to nitrogen assimilation pathways was tested in vivo. Under nitrogen-limiting conditions it was shown that externally applied NO can overcome N deficiency and was improved by HB overexpression (Kuruthukulangarakoola et al., 2017). Therefore, under hypoxia, NO recycling can simultaneously provide NAD<sup>+</sup> for glycolysis and to allow alanine synthesis to compete with the fermentative pathways and conserve nitrogen (**Figure 1A**). An important nitrogen recycling pathway seems to be that of glutamate dehydrogenase (GDH), because under

anoxic stress the Arabidopsis gdh double mutant accumulates more 2-oxoglutarate, has lower gamma-aminobutyric acid (GABA) reserves, and is deficient in alanine consumption during reoxygenation (Tsai et al., 2016). These metabolic defects have dramatic negative consequences for plant survival after submergence.

Although GDH is known to operate in vivo in the deamination of glutamate (Labboun et al., 2009), a role for amination activity under oxygen stress is suggested not only by the accumulation of 2-oxoglutarate in the gdh mutant but also because <sup>15</sup>NH<sup>4</sup> is barely incorporated into glutamine in soybean roots (implying that ATP limits glutamine synthase, the regular ammonium assimilation pathway) but still is efficiently transferred to GABA and alanine (António et al., 2016). Since plants have covered their needs to balance carbon and nitrogen with at least four major enzymes (alanine aminotransferase, aspartate amino transferase, GABA transaminase and GDH), untangling this metabolic network requires experimental models that take into account the nitrogen source, the use of external sucrose in the analysis buffers, methodological invasiveness, and microenvironment control. Recently, an elegant multidisciplinary approach was used to demonstrate the role of GDH in recycling ammonia to sustain the proliferation of cancer cells (Spinelli et al., 2017).

#### Energy and Calcium Signaling

In Arabidopsis, the translocation of ERF-VII protein RAP2.12 to the nucleus is dependent on the oxygen concentration (Kosmacz et al., 2015) and the dioxygenase enzymatic activity of PCO isoforms reaches its peak at 20% oxygen, further declining almost linearly as O<sup>2</sup> decreases; the precise relationship between NO and O<sup>2</sup> in vivo is still unknown, however (White et al., 2018). This recent knowledge indicates that plant cells have evolved efficient mechanisms to sense oxygen levels and to transduce those concentrations into a molecular response suited to face different flooding and hypoxic scenarios.

Recent reviews have proposed that this oxygen-sensing capacity must be accompanied by other sensing and transduction mechanisms, especially aimed at detecting a cellular energy crisis. By comparison to animal cells, several candidate mechanisms were proposed to react at a faster rate than the oxygen sensing capability of ERF-VII, and most of them rely on membrane proteins acting as ionic transporters (Wang F. et al., 2017). Igamberdiev and Hill (2018) have presented convincing arguments for the role of calcium signaling as a strong, fast and flexible primary signal of energy stress, in particular the increase in the level of cytoplasmic calcium [Ca2+]cyt from internal and apoplastic sources, but more importantly, directly from altered ATP/ADP ratios.

Temporary spikes of [Ca2+]cyt and subsequent cellular efforts to return to a homeostatic level were measured in anoxic and recovery conditions more than 20 years ago (Sedbrook et al., 1996). Later, it was demonstrated that an H2O2-mediated cellular rheostat composed of Rop G-proteins and its interactor RopGAP involves [Ca2+]cyt-activated NADPH oxidases and that a misbalance of any component leads to defects in alcohol dehydrogenase (ADH) expression with large impacts on plant survival after hypoxic stress (Baxter-Burrell et al., 2002). Calcium can also regulate ATP-Mg/Pi mitochondrial transporters, providing a direct pathway to flexibly regulate the ATP/ADP pool (Stael et al., 2011).

Successful return to a homeostatic state of [Ca2+]cyt is also important for plant fitness, since knockout mutants for Ca2<sup>+</sup> transporters of ATPases and proton antiporter families (ACA/CAX) present aberrant concentrations of Na+, K+, and Ca2+. Remarkably, this inability to correct [Ca2+]cyt leads to lower biomass after waterlogging stress (Wang et al., 2016). The importance of crosstalk between [Ca2+]cyt-sensitive systems and H2O<sup>2</sup> modulation was demonstrated by the characterization of the CHG coding for HRU1, a protein belonging to the archaic universal stress protein (USP) family, which is capable of physically interacting with the RopGAP rheostat and whose misregulation leads to severely sensitive phenotypes under submergence and anoxic stresses (Gonzali et al., 2015). Remarkably, the expression of Arabidopsis HRU1 is dependent on ERF-VII, allowing it to get feedback from oxygen sensing. In tomato, another member of the USP family is able to interact with a Ca2+-dependent protein kinase to physically transduce [Ca2+]cyt to ROS signaling (Gutiérrez-Beltrán et al., 2017).

Finally, downstream transducers of [Ca2+]cyt, like calmodulin-like protein, connect this elemental signal to more complicated processes such as the formation of multimolecular ribonucleoprotein complexes that, when disrupted, have large effects on plant survival of hypoxia (Lokdarshi et al., 2016) (see section "RNA Dynamics Under Stress"), most likely by the incorporation of efforts by plant cells to save energy through the conservation of mRNAs that are useful during the expected reoxygenation (Sorenson and Bailey-Serres, 2014). These complexes have multiple connections to other stresses and cellular homeostatic processes (Chantarachot and Bailey-Serres, 2018).

With all this research it is becoming more clear that this multilevel signaling and transduction network is in agreement with the evidence that anaerobic responses are not unique to hypoxic stress and provide a unifying genetic framework to generically respond to stress conditions that decrease energy availability (Greenway and Armstrong, 2018) and, at the same time, provide mechanisms to sense O<sup>2</sup> concentrations (Sasidharan et al., 2018), thus creating an adjustable response to the evolutionary range of stress intensities found in nature by combining in time energy stress, waterlogging and submergence peculiarities and light/dark cycles.

Recently, by means of high-definition live imaging, genetically encoded fluorescent sensors and genetics, observations on the scale of seconds could be made of the fast dynamics of [Ca2+]cyt bursts, signal distribution, and homeostatically set points in intact plants in the context of wounding/herbivore stress and how it is mediated, almost exclusively, by glutamate and glutamate transporters (Toyota et al., 2018). This amino acid is a pivotal metabolite of hypoxic stresses (Barding et al., 2013). This imaging research strategy should shed light on the dynamic connections of the many signaling hubs reviewed

here, from both the biological and biotechnological aspects of flooding stress.

#### MOLECULAR MECHANISMS OF MULTICOMPONENT STRESS

#### Root Architecture and Responses

Root architecture and plasticity play an important role in the adaptation to submergence and waterlogging stress. The formation of aerenchyma and adventitious roots is a morphological characteristic of waterlogging-tolerant species. Aerenchyma is known to enhance internal oxygen diffusion from the aerial parts to the waterlogged roots that allows the roots to maintain aerobic respiration (Armstrong, 1980).

Two types of aerenchyma can be found in roots: primary aerenchyma, consisting of lysigenous aerenchyma, and schizogenous aerenchyma in the roots of rice, maize and wheat, and secondary aerenchyma in soybean roots (reviewed in Yamauchi et al., 2018). The formation of lysigenous aerenchyma results from the selective death and subsequent lysis of root cortical cells, while schizogenous aerenchyma formation is caused by cell separation, without the occurrence of cell death. Secondary aerenchyma develops from phellogen, forming spongy tissue filled with air spaces outside of the stem, hypocotyl, and roots.

Gene expression analysis was applied to study the molecular mechanisms of lysigenous aerenchyma formation in roots using a microarray combined with laser microdissection demonstrating that the expression of genes involved in calcium signaling, cell-wall modification, ethylene and reactive oxygen species (ROS) changes during lysigenous aerenchyma formation under conditions of low oxygen (Rajhi et al., 2011; Yamauchi et al., 2011). However, the precise molecular mechanism controlling both primary and secondary aerenchyma formation has not yet been characterized.

In addition, the spatial arrangement of roots and their components is crucial for the adaptation to submergence and waterlogging stress. In Arabidopsis, low-oxygen-induced primary root bending is controlled by hypoxia-induced auxin flux at the root tips but negatively regulated by the hypoxia-induced group-VII ERFs (Eysholdt-Derzsó and Sauter, 2017). Moreover, the hypoxia-induced group-VII ERFs also promote adventitious root (AR) elongation, while ethylene inhibits adventitious root formation (Eysholdt-Derzsó and Sauter, 2018). In Solanum dulcamara, a dicot species that constitutively develops dormant primordia that can be reactivated upon flooding, transcriptome analysis and hormonal treatment were used to investigate the signaling pathway controlling AR primordium reactivation (Dawood et al., 2016). Flooding increased ethylene accumulation and subsequently a drop in ABA level in AR primordia tissue. ABA treatment inhibited activation of AR primordia by flooding and blocking of ABA was able to reactivate AR primordia in the absence of flooding (Dawood et al., 2016). Both flooding and ethylene induced polar auxin transport, but auxin treatment alone was not sufficient for AR emergence indicating the interplay among ABA, auxin, and ethylene contributing to the regulation of AR development under flooding. In waterlogging-tolerant legumes of the genus Trifolium, its ability to tolerate waterlogging coordinates with higher root porosity and the ability to form lateral roots (Gibberd et al., 2001). However, little is known about the molecular mechanism in waterlogging-tolerant crops controlling lateral root formation induced by waterlogging.

Plants are able to respond to different levels of flooding and systemic signals can travel between underground and aerial organs, causing differential transcriptomic responses (Hsu et al., 2011); ethylene is one of these signals affecting many processes, including the characteristic hyponastic response in the leaves of many plants (Polko et al., 2015). However, many other signals can also come into play, for example, carbohydrates, ABA, ions, and amino acids, even ratios of different closely related metabolites, notably T6P and sucrose whose enzymatic control (through trehalose phosphate phosphatase) is expressed in young heterotrophic tissues, especially in the roots (Kretzschmar et al., 2015).

Therefore, understanding root architecture, development and plasticity under submergence and waterlogging stress could further clarify plants' responses to the stress. Supporting this idea, Mustroph (2018) recently reviewed the vast literature of reported quantitative trait loci (QTL) associated with flooding stress and found that, remarkably, most QTLs associated with waterlogging resistance in economically relevant tolerant cultivars of grasses were related to aerenchyma and adventitious root formation, and require further molecular understanding.

# Carbohydrate Depletion Sensing

Under flooded conditions, plants experience energy and carbohydrate deprivation due to reduced photosynthesis and aerobic respiration. Plants encounter a similar energy status when exposed to prolonged darkness. Therefore, it is not surprising that plants tolerant to submergence can also endure prolonged darkness. Constitutive and stress-induced expression of SUB1A increases tolerance to submergence in rice through the restriction of carbohydrate breakdown and elongation growth (Fukao et al., 2006; Fukao and Bailey-Serres, 2008). The same genotypes exhibit stronger viability under prolonged darkness by reduced degradation of carbohydrate reserves and chlorophyll (Fukao et al., 2012). Under both submergence and constant darkness, SUB1A restricts the biosynthesis of ethylene, a hormone promoting leaf senescence. Similarly, mutation of prt6 and its allelic variant, ged-1, increased both submergence and dark tolerance in Arabidopsis through the reduced breakdown of starch reserves (Riber et al., 2015). These results indicate that common signaling and metabolic pathways are involved in the regulation of submergence and dark tolerance in plants. Comparative microarray analysis revealed that 29.2% of genes up-regulated upon submergence in shoots are also induced by prolonged darkness in Arabidopsis (Lee et al., 2011), suggesting the existence of common regulatory pathways for submergence and dark tolerance.

Sucrose non-fermenting-1-related kinase 1 (SnRK1) is the central regulator of energy homeostasis in plants where its

kinase activity is induced in response to energy starvation (Emanuelle et al., 2016). Besides the energy status, SnRK1 activity is also regulated by the redox status (Wurzinger et al., 2017). In rice, SnRK1A plays a critical role in seed germination under anaerobic (flooded) conditions (Lee et al., 2009). Submergence-induced sugar starvation triggers mRNA accumulation of calcineurin B-like protein-interacting protein kinase 15 (CIPK15), enhancing the accumulation of SnRK1A proteins. CIPK15 also physically interacts with SnRK1A, resulting in MYBS1-mediated alpha-amylase induction (Lee et al., 2009). This SnRK1A-mediated process is also pivotal for rice seed germination under non-stress conditions (Lu et al., 2007) because localized sugar starvation occurs in actively growing tissues even under aerobic conditions. Altering SnRK1 signaling in Arabidopsis had strong repercussions on plant tolerance, disrupting AMP/ATP ratios, T6P and amino acid homeostasis (Cho et al., 2016). Not surprisingly, starch is a crucial nutrient reservoir for tolerance to submergence; both inabilities to use and synthesize starch are detrimental for plant survival (Loreti et al., 2018). However, signaling of sugar availability works through gene redundancy of SnRK and the expression of basic leucine zipper TFs (bZIP TFs) in a separate pathway from the transcriptional regulation of CHGs (Loreti et al., 2018).

The role of SnRK1 in the regulation of carbohydrate and nitrogen metabolism under starvation conditions has been characterized. For example, SnRK1 directly phosphorylates HMG-CoA reductase, NR, sucrose phosphate synthase, 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase, and trehalose-6-phosphate synthase to regulate their enzymatic activities (Mackintosh et al., 1992; Douglas et al., 1997; Sugden et al., 1999; Kulma et al., 2004; Harthill et al., 2006). SnRK1 also activates S1-bZIPs such as bZIP1, bZIP11, and bZIP53, which regulate starch and branched-chain amino acid catabolism, trehalose-6-phosphate biosynthesis, and nitrogen signaling in Arabidopsis under starved conditions (Baena-González et al., 2007; Dietrich et al., 2011; Ma et al., 2011; Para et al., 2014). More recently, SnRK1 was reported to phosphorylate C-type bZIPs and promote the formation of C/S<sup>1</sup> bZIP heterodimers (Mair et al., 2015; Pedrotti et al., 2018). The C/S<sup>1</sup> bZIP–SnRK1 complex binds to target promoters and regulates chromatin structure via histone acetyltransferases (Pedrotti et al., 2018). The downstream genes directly or indirectly regulated by the SnRK1–bZIP network are associated with crucial metabolic pathways for plant survival under energy starvation.

A similar mechanism mediated by SnRK1 may regulate the expression of hypoxia-responsive genes. Indeed, SnRK1 directly binds to the promoter regions of hypoxia-inducible genes in response to submergence, which correlates with the increased expression of these genes in Arabidopsis (Cho et al., 2012). Because SnRK1 contains no DNA-binding domains, the authors suggested that protein complexes consisting of SnRK1 and TFs (e.g., C/S<sup>1</sup> bZIPs) bind to the target promoters and modify their chromatin structure. Identification of TFs interacting with SnRK1 and target gene chromatins will advance the molecular understanding of SnRK1-mediated transcriptional regulation under submergence and waterlogging.

# Bacterial Damage and Preemptive Transcriptomic Preparation

Consistently, transcriptomic studies under submergence stress have found up-regulated gene clusters associated with plant responses to fungi and bacteria. Expressed genes code for pattern recognition receptor proteins (PRRs), wall-associated kinases (WAKs), leucine-rich repeats (LRRs) and lectin-DUF26 proteins (Lee et al., 2011; Hsu et al., 2013; Rivera-Contreras et al., 2016). The number of genes inside these clusters, ranging up to hundreds, suggests that complex plant–microbe relationships are expected by the plant cell to be counteracted/established concomitant to flooding. These can be detrimental, for example, an increased number of microbial cells with potential pathogenic capacities. On the other hand, the induction and stabilization of beneficial partnerships may also be possible.

Both of these possibilities were explored experimentally. Hsu et al. (2013) demonstrated in Arabidopsis that strong upregulation of innate immunity genes can be detected as early as 1 h after submergence. Interestingly, a 2-h submergence stress treatment conferred the plant tolerance to post-submergence inoculation with Pseudomonas syringae. When T-DNA knockout mutants of TF WRKY22 were simultaneously confronted with submergence and the bacterial pathogen, mutant plants suffered more visual damage and harbored a bacterial population of twice the size of that on wild-type plants. WRKY TFs are transcriptional modulators of plant–microbe interactions and have as genetic targets different LRRs, PRRs, and WAK proteins. Parallel to this pathway, Vicente et al. (2018) used mutations in PRT6, a pivotal N-terminal rule enzyme, and also observed increased resistance to P. syringae (in Arabidopsis) and to P. japonica and Blumeria graminis (in barley), most likely by stabilization of ERF-VII TFs (Vicente et al., 2018). Therefore, during submergence/waterlogging, WRKYs and ERF-VII TFs create a positive pathogen defense genetic loop.

The hundreds of pathogen-related genes expressed may be a reflection of an evolutionary history of exposition to probable pathogens, since some of these genes (e.g., lectins) were demonstrated to specifically recognize chemical patterns of certain species (Miyakawa et al., 2014). On the other hand, positive plant–bacteria relationships have also been documented during a flooding event. In rice, the diversity of nitrogen-fixing bacteria was studied before and after intentional flooding as part of normal agricultural practices (Ferrando and Fernández Scavino, 2015). The results indicate that flooding helped to establish a more diverse facultative/anaerobic diazotrophic community, especially in the rhizosphere. Breidenbach and Conrad (2014) also reported that planted rice soils have a stable population throughout plant development which changes in response to draining and crop rotation to maize. Although not directly related to flooding, a report in Arabidopsis (Lundberg et al., 2012) demonstrated that bacterial populations in the rhizosphere are distinct from those detected in soil and, more importantly, are characteristic of different genotypes,

predicting that the signaling for bacterial recruitment in roots is genetically coded.

The works mentioned above successfully used available masssequencing technologies to overcome the methodological barriers imposed by traditional microbiology to study the complex rhizosphere environment. Although, the study in rice may represent the case of a domesticated plant adapted for flooding events where positive interactions develop, the experimental working frame employed by the authors may shed light on other crops and wild plants that may not be readily prepared to establish positive interactions in a flooding event and where defense triggering mechanisms may be more important.

Another unexplored pathway of plant–microbe interactions in submergence is the input of jasmonate-zim-domain protein 3 (JAZ3) to submergence survival given that it is a CHG in Arabidopsis (Gasch et al., 2016), a family member of firsttransducers of [Ca2+]cyt (Toyota et al., 2018) with important roles as both pathogen defense mediator and as a natural target for chemically induced susceptibility by fungal pathogens mimicking jasmonic acid signaling (Chini et al., 2018).

#### Heavy Metals and Salinity

Transcriptomic studies of submergence stress have frequently reported the altered expression of transcripts coding for proteins of heavy-metal homeostasis (Lee et al., 2011; Rivera-Contreras et al., 2016). Also, a common pattern of ethylene and ROS bursts and responses was observed in heavy-metal stress and flooding stress signaling, pointing to probable co-evolution (Steffens, 2014). Measurements of ion concentrations during a flooding event indicate that Na, Al, B, Mn, and Fe tend to accumulate (Setter et al., 2009; Zeng et al., 2013) and, interestingly, that plants are able to restrain traffic of toxic ions to their biomass (Kashem and Singh, 2001). The ability to restrain this traffic is also differential and characteristic for different genotypes, as demonstrated in wheat (Setter et al., 2009).

The relevance of these observations to submergence/flooding tolerance is highlighted by the finding that transcripts coding for proteins of iron homeostasis are more strongly down-regulated in Arabidopsis roots of tolerant ecotypes (van Veen et al., 2016). Using RNA-Seq, Du et al. (2017) analyzed the response to waterlogging in roots of tropical maize lines and found a mRNA coding for a heavy-metal transporter as differentially expressed in tolerant lines. The genomic location of this gene coincided with a previous QTL associated with more root biomass under waterlogging in maize seedlings (Osman et al., 2013). This particular gene has not been molecularly characterized; therefore, its promising role under stress is poorly understood.

#### RNA DYNAMICS UNDER STRESS

#### Cytoplasmic RNA Dynamics

Regulation of gene expression is an important factor controlling most plant biological processes, including cellular differentiation, organ development, and environmental adaptation to the constantly changing environment. In plants, the fundamental role of transcriptional regulation is universally recognized, and the general mechanism participating in transcriptional regulation is well understood. In recent years, post-transcriptional control has been embraced as an essential process controlling plant gene expression (Chantarachot and Bailey-Serres, 2018).

Submergence and waterlogging stress typically limit oxygen availability and energy production in plant cells. Several pieces of evidence demonstrate that post-transcriptional gene regulation participates in the metabolic adjustments that aid a plant's surviving conditions of low oxygen. Cellular mRNAs typically associate with RNA-binding protein (RBP) forming mRNA– ribonucleoprotein (mRNP) complexes. These mRNP complexes determine the activity of individual RNAs through the regulation of mRNA processing, localization, translation, sequestration, and degradation (Bailey-Serres et al., 2009).

Plant cytoplasmic ribosome complexes are among the best characterized mRNP complexes. The ribosome complexes contribute to the initiation, elongation, and termination of translation. During initiation, the large subunit (60S) of the ribosome joins the small subunit (40S) at the AUG initiation codon, forming an 80S ribosome. The sequential recruitment of ribosomes onto a single mRNA generates polyribosome (polysome) complexes (Juntawong et al., 2014a). Using Arabidopsis expressing a FLAG-epitope tagged ribosomal protein L18, translating ribosome affinity purification (TRAP) technology can be used to isolate the mRNA populations associated with at least one 80S ribosome that could be further analyzed by high-throughput genomic approaches (Reynoso et al., 2015). TRAP allows quantitative examination of the plant translatome for changes in mRNA translation. A study using TRAP and microarrays demonstrated that a condition of low oxygen reduces the cellular ATP content and results in selective translation and global translational repression, which is rapidly reversible following reoxygenation in Arabidopsis (Branco-Price et al., 2008). Interestingly, a large proportion of mRNAs that display no changes in abundance and some low-oxygen-induced mRNAs are not translated until reoxygenation, suggesting these mRNAs could be sequestrated in non-translated mRNP complexes. The contribution of selective mRNA translation to low oxygen was further assessed in specific cell populations of Arabidopsis seedlings using the cell-type-specific TRAP technology combined with microarrays (Mustroph et al., 2009). The cell-type-specific translatome exposed a common general response in most cells but with complex and yet unexplored local molecular responses and adaptation to low oxygen.

Because a single mRNA can associate with more than one ribosome, ribosome profiling (Ribo-Seq) technology that includes the nuclease digestion of mRNAs associated with ribosomes, deep-sequencing of ribosome-protected footprints (RPFs) and mapping of RPFs to the transcripts, allows the quantification of translation rates of individual mRNAs (Juntawong et al., 2015). Evaluation of mRNA translation using Ribo-Seq in Arabidopsis seedlings subjected to low oxygen revealed several aspects of translational regulation (Juntawong et al., 2014a). Firstly, low oxygen globally reduces initiation but increases termination. Secondly, the translational efficiency of individual mRNAs is selectively controlled under conditions of low oxygen. Thirdly, translation of alternatively spliced mRNAs

and non-coding (nc) RNAs could contribute to the proteomic diversity in response to low-oxygen conditions. Lastly, features such as an upstream Open Reading Frame (uORF) on individual mRNAs affect mRNA translation under conditions of low oxygen, including bZIP TFs, are pivotal to sugar signal transduction under different scenarios and are regarded as paradigms of uORF translation regulation (Dröge-Laser et al., 2018).

mRNA sequestration is considered a mechanism that contributes to the control of mRNA translation under low oxygen. In mammalian cells, energy and nutrient stresses result in global repression of translation followed by the recruitment of non-translationally active mRNAs into the stress granule (SG) mRNP complexes (Chantarachot and Bailey-Serres, 2018). SGs are typically recognized as mRNA storage sites that are involved in the dynamic exchange of mRNAs with translating ribosomes and processing bodies (PB) where the mRNAs are degraded (reviewed in Bailey-Serres et al., 2009). In plants, oligouridylate binding protein 1C (UBP1C), a triple RNA Recognition Motif (RRM) ortholog of the animal SG nucleated protein T-cell intracellular antigen 1 (TIA-1), is dynamically and reversibly associated with translationally inactive mRNAs forming UBP1C-SGs under conditions of low oxygen (Sorenson and Bailey-Serres, 2014). In addition, UBP1C knockdown lines display a lowoxygen-sensitive phenotype, indicating the significance of SG formation and mRNA sequestration as a mechanism controlling the response to low oxygen in plants.

Evidence of the functional roles of RNA dynamics in fine-tuning low-oxygen gene expression and the molecular mechanisms underlying these modes of regulation has been emerging gradually. Recent advances in high-throughput approaches that allow the identification and quantification of ribosome footprints, RNA structure, and protein–RNA interaction will facilitate the understanding of the mechanism controlling RNA dynamics toward low-oxygen response.

#### Expression and Modulation of miRNAs

MicroRNAs (miRNAs) are endogenous, small non-coding RNAs that act as post-transcriptional regulators of gene expression and have been extensively described as being involved in biological processes in plants such as development and the response to environmental stimuli (Chen, 2009). In plants, miRNAs recognize target mRNAs by almost perfect base pairing and direct them for RNA cleavage or inhibition of translation (Rogers and Chen, 2013). When a specific miRNA is accumulated, the expression of the target mRNA is expected to be decreased, and vice versa. MiRNA expression is highly sensitive to environmental clues that prepare them to function as dynamic regulators of response to stress in a fine-tuned manner. Moreover, miRNAs act as regulatory nodes in complex networks to interconnect the response to biotic and abiotic stress with plant development (Rubio-Somoza and Weigel, 2011). With the advent of NGS technologies, several miRNAs were described, uncovering a diversity of biological functions in response to stress in plants.

miRNAs have an active role in response to flooding in plants as revealed by genome-wide studies in response to hypoxia (Moldovan et al., 2010; Licausi et al., 2011b), waterlogging (Liu et al., 2012; Zhai et al., 2013), and submergence (Zhang et al., 2008; Jeong et al., 2013; Jin et al., 2017; Li et al., 2017; Franke et al., 2018). Despite the fact that reports of miRNA's responsiveness to flooding stress are still scarce in comparison to reports on other abiotic stress, existing studies have revealed that miRNAs regulate four main lines of response to flooding stress including morphological adaptation, management of energy supply, control of flowering, and oxidative stress response (**Figure 2**).

#### Morphological Adaptation

miRNAs promote and control the balance between the response to flooding stress and development through hormonal signaling pathways. The analysis of cis-regulatory elements in the promoters of flooding-responsive miRNAs has made evident that ethylene, GA, ABA, and auxin signal transduction pathways are main components interconnecting the complex network of response to flooding (Zhang et al., 2008; Liu et al., 2012). Downstream, several flooding-responsive miRNAs regulate target mRNAs, encoding TFs involved in hormonal signaling pathways that orchestrate adaptation to flooding through morphological changes in the different plant tissues.

For example, in the roots of maize, miR159 was up-regulated by waterlogging and was responsible for silencing two mRNAs encoding GAMYBs, MYB33, and MYB101 homologs (Liu et al., 2012). MYB33 and MYB101 together with MYB65 are controlled by miR159 in Arabidopsis to inhibit primary root growth (Xue et al., 2017). Hence, up-regulation of miR159 in maize roots under waterlogging may be important to clear GAMYB mRNAs and arrest primary root growth. Also, down-regulation of GAMYBs could limit responses related to ABA signaling pathways, in which MYB factors have an active role. On the contrary, miR159 has been found to be down-regulated by flooding in other tissues, such as submerged Lotus seedlings, where degradome and transcriptome analysis confirmed several GAMYB targets of miR159 (Jin et al., 2017) that might be involved in the regulation of petiole elongation mediated by GA.

Upon submergence or waterlogging, tolerant plants such as Alternanthera philoxeroides respond rapidly to oxygen limitation by initiating adventitious rooting on nodes of the submerged stem (Ayi et al., 2016). Auxin signaling is central to morphological changes in roots through the activity of miRNAs and auxin response factor TFs (ARF; Meng et al., 2010). It has been demonstrated that miR166 is involved in modulating root growth by targeting HD-ZIP III transcripts as part of a complex phytohormone response network including auxins (Singh et al., 2017). The over-expression of miR166 resulted in down-regulation of HD-ZIP III and increased root growth under normal conditions in Arabidopsis (Singh et al., 2014). MiR166 was up-regulated in response to short-term waterlogging in maize roots, where it was demonstrated experimentally to cause down-regulation of its target, the maize HD-ZIP III family member rolled leaf 1 (Zhang et al., 2008). It is possible that during waterlogging, miR166 participates in the modulation of hormone signaling pathways to promote lateral and adventitious rooting through modulation of HD-ZIP III. On the other hand, down-regulation of miR166 was observed in Arabidopsis roots under hypoxia stress (Moldovan et al., 2010). Since hypoxia

synthase; GH, glycosyl hydrolase; LAC, laccase; LRR, leucine-rich repeat; ME, malic enzyme; PLCL, plantacyanin-like protein; POD, peroxidase; SOD, superoxide

causes the cessation of root growth, which is followed by the rapid development of adventitious roots when normoxia is reestablished, it has been hypothesized that miR166 could play a key role in the pathways that integrate critical levels of hypoxia signals during flooding, like calcium spikes and an increase of ROS, to redirect root growth toward branching or adventitious rooting when normoxia returns.

dismutase; UKN, unknown function protein; β-amy, beta-amylase.

In addition, in maize roots, miR167 was found to be upregulated by short-term submergence (Zhang et al., 2008) and short-term waterlogging (Liu et al., 2012). In contrast to short-term stress, miR167 was down-regulated by long-term waterlogging in maize roots, revealing differential regulation during the progression of stress (Zhai et al., 2013). Two miR167 targets were experimentally confirmed in maize roots encoding ARF16 and ARF18 transcription factors (Liu et al., 2012). Based on the previous knowledge that miR167 controls the expression of ARFs to regulate adventitious and lateral root development (Gutierrez et al., 2009), regulation of ARFs by miR167 in maize roots could have a balancing effect between suppressing primary root growth and promoting adventitious rooting in response to flooding in a time-scale manner. MiR167 was also up-regulated in the internodes of Alternanthera plants (Li et al., 2017) and Populus seedlings (Ren et al., 2012) and down-regulated in Lotus seedlings under submersion (Jin et al., 2017), probably to control some of the diverse roles of ARF transcription factors in different tissues (i.e., shoot/petiole elongation) in response to flooding.

In roots of Arabidopsis exposed to hypoxia, which is a component of flooding stress, miR390, another miRNA involved in the regulation of auxin response pathways, was up-regulated (Moldovan et al., 2010). It was demonstrated experimentally that miR390 acts by its target mRNA, TAS3, from which small transacting RNAs are produced (tasiRNA3) that negatively control the expression of ARF TFs. In a study of Arabidopsis mutants with altered levels of TAS3a, it was found that miR390 and tasiRNA3 were accumulated at the sites for lateral root initiation to block the expression of ARFs and then increase the length of lateral roots. At the same time, ARFs control the auxin-inducible expression of miR390, thus constituting a loop for feedback

regulation that maintains ARF expression in a fine-temporal and spatial regulation of the optimal control of root development under oxygen-limited conditions (Marin et al., 2010).

Other miRNAs associated with auxin regulation pathways are miR393 and miR164. MiR393 was down-regulated in response to submergence in Lotus seedlings. As validated by degradome and small RNA sequencing analysis, their targets are mRNAs encoding TIR/F-BOX, which end up being upregulated by down-regulation of miR393 to promote the auxin response in submerged seedlings (Jin et al., 2017). Conversely, miR393 was up-regulated in waterlogged maize roots (Liu et al., 2012) and submerged Brachypodium aerial tissue (Franke et al., 2018). On the other hand, miR164 was up-regulated in maize roots in response to flooding (Liu et al., 2012). It was experimentally demonstrated that in Arabidopsis miR164 represses the expression of NAC/NAM domain proteins, and thus provides fine control of lateral root growth promoted by auxins (Guo et al., 2005). This suggests that up-regulation of miR164 in maize roots could have a similar effect on the control of root growth in response to waterlogging.

MiR156 was up-regulated by hypoxia in Arabidopsis roots and by submergence in Lotus seedlings and Brachypodium aerial tissues (Moldovan et al., 2010; Jin et al., 2017; Franke et al., 2018). MiR156 regulates several mRNAs encoding members of the SQUAMOSA PROMOTER BINDING PROTEIN-LIKEs (SPLs) gene family, which participate in a diversity of roles like control of root growth, transition from the juvenile to the adult phase, and branching and maturation of the shoot (Schwarz et al., 2008). For example, SPL10 is one of the main repressors of root development in Arabidopsis (Yu et al., 2015). Under normal conditions, overexpression of miR156 causes the down-regulation of SPL10, thus promoting lateral rooting in Arabidopsis (Gao et al., 2018). This suggests that the miR156:SPL10 module might function in regulating the adaptive response of roots in flooding conditions.

During flooding, ethylene functions as a primary signal to induce morphological and metabolic adjustments in plants. APETALA2/Ethylene Responsive Element (AP2/ERF) is a large family of TFs regulated by ethylene that play roles in the control of floral organ identity, in the development of shoot meristem, and in primary and secondary metabolism, among other plant growth and developmental processes (Licausi et al., 2013), and are targeted for posttranscriptional regulation by miR172. Longterm waterlogging in maize roots down-regulated the expression of miR172 and accumulated AP2/ERF mRNAs according to degradome analysis (Zhai et al., 2013), suggesting an increase of the ethylene signaling pathways to promote the development of crown roots as an adaptive morphological response.

#### Management of Energy Supply

Flooding creates an environment where an excess of water limits aerobic respiration and photosynthesis, causing a decrease in the energy supply. An alternative for maintenance of the energy supply in such conditions is limiting the futile synthesis of starch and inducing starch breakdown. Submergence in maize roots induces miR399, down-regulating its predicted target, a granule-bound starch synthase (Zhang et al., 2008), suggesting a mechanism to limit the synthesis of starch and avoid wasting energy. In Populus, miR399 is also induced in plantlets by submergence and, different to maize, its target mRNA is predicted to encode a Major Facilitator Superfamily protein (MFS) that functions as a transporter of diverse substrates (Ren et al., 2012). In Lotus seedlings, submergence down-regulates several members of the miR399 family, resulting in the accumulation of their predicted target gene encoding a beta-amylase (Jin et al., 2017). Beta-amylase is an enzyme required for starch breakdown active during stress (Wang X. et al., 2017) that could participate in the remobilization of carbohydrates in Lotus to provide energy during flooding.

In maize roots, short-term hypoxia treatment down-regulated the expression of miR159, miR395, and miR474, suggesting the accumulation of their predicted targets encoding enzymes such as vacuolar ATPase, glycoside hydrolase, ATP sulfurylase and malic enzyme (Zhang et al., 2008). These targets are involved in fundamental processes under energy stress, such as ATP production sensing, pH homeostasis, enhancement of sulfur assimilation, carbohydrate breakdown, and pyruvate regulation. However, when hypoxia treatment is continued for more than 24 h and oxygen becomes limiting, miR395 and miR474 are up-regulated to decrease the expression of those enzymes.

#### Control of Flowering

Flowering and the development of floral organs are energyconsuming processes for plants under flooding stress. Thus, delaying flowering and flower development could be seen as an alternative strategy to save energy under submergence stress (Peña-Castro et al., 2011). Despite studies on miRNAs responsive to flooding being mostly performed on roots or seedlings, several miRNAs were described that are involved in controlling flowering during flooding stress. One of them, miR319, is induced in Lotus seedlings to down-regulate the expression of its target TCP4, which is critical for petal growth and development (Jin et al., 2017). In this same line of control, miR156 is up-regulated by submergence in Brachypodium and Lotus seedlings, and by hypoxia in Arabidopsis roots. MiR156 controls the expression of SPLs, which are a rich group of TFs involved in a diverse group of functions in plant tissues. While in roots miR156 could be responsible for the regulation of SPLs involved in root development, in the rest of the plant miR156 may control different SPLs to promote the phase change from vegetative to reproductive growth and other developmental processes like floral organ size and fruit development (Chen et al., 2010).

miR172 also takes part in the response regulating flowering under flooding stress. MiR172 is up-regulated by short-term hypoxia in Arabidopsis roots (Moldovan et al., 2010) and by short-term waterlogging in maize roots and Lotus seedlings (Liu et al., 2012; Jin et al., 2017), with regulatory activity over experimentally validated Apetala2 (AP2) AP2/ERF and AGAMOUS TFs, respectively. As SPLs, AP2/ERFs have diverse functions in plant growth depending on the tissue (i.e., in the root), and the severity of the stress. Some targets of miR172 are repressors of flowering, indicating that miR172-mediated down-regulation may promote flowering in plants during the first hours of waterlogging or hypoxia (Zhu and Helliwell, 2011). However, during long-term submergence miR172 is downregulated, implicating that the target mRNA will be up-regulated, with the consequent repression of flowering. miR5200 is another miRNA involved in flowering that accumulates in Brachypodium in response to submergence; it targets FLOWERING LOCUS (FT), a TF with a central role in determining the flowering time, and causes delayed flowering as a strategy for the control of flowering in response to stress (Jeong et al., 2013). In this way, miR5200 contributes to the saving of energy necessary under stress.

#### Oxidative Stress Response

fpls-10-00340 March 21, 2019 Time: 16:28 # 13

During flooding stress, respiration is impaired, leading to the production of ROS. In order to maintain ROS homeostasis, plants under flooding respond by the activation of oxidation/reduction enzymes, such as cupredoxins. Consistent with this, the most represented biological process predicted from miRNAs responsive to submergence in Lotus seedlings was oxidation– reduction (Jin et al., 2017). Plantacyanin, laccase and superoxide dismutase contain cupredoxin domains and are conserved targets of miR408, miR528, and miR397. In Lotus seedlings, miR408 and miR397 were down-regulated by submergence (Jin et al., 2017). In maize roots under long-term submergence, miR528 was also down-regulated (Zhang et al., 2008). Interestingly, up-regulation was observed for miR408 and miR528 in maize roots undergoing short-term waterlogging (Liu et al., 2012), and for miR397 under short-term hypoxia in Arabidopsis roots (Licausi et al., 2011b), reflecting a balance between ROS signaling and toxicity.

In addition to the processes outlined above, other functions could be regulated by miRNAs that await being addressed, such as the potential interplay between responses to flooding and plant–pathogen interaction. In Lotus, two targets encoding putative LRR receptor-like protein kinases, which are recognized for playing a role in response to pathogens, were predicted as targets of miR159 and miR9774. These miRNAs were downregulated in response to submergence (Jin et al., 2017). On the other hand, miR171 showed up-regulation upon submergence in maize roots and Brachypodium (Zhang et al., 2008; Franke et al., 2018), and down-regulation in Lotus seedlings (Jin et al., 2017). MiR171 was predicted to target a WRKY TF in maize, which belongs to a large family of TFs that participate in disease resistance networks in plants (Pandey and Somssich, 2009). In Brachypodium, miR171 was predicted to target a GRAS TF. Among other functions, GRAS TFs are important in the defense against stress (Mayrose et al., 2006). In Lotus, where miR171 was down-regulated, however, its predicted targets are proteins with unclear or uncharacterized function (Jin et al., 2017).

Fine-temporal regulation of miRNAs is important for the response to flooding stress. Many miRNAs show fluctuating expression profiles during the time-course of hypoxia or submergence treatments (Licausi et al., 2011b; Li et al., 2017). In maize roots, long- and short-term waterlogging accumulated a series of miRNAs with opposite expression profiles depending on the duration of the stress. Most miRNAs up-regulated in the short-term treatment (Liu et al., 2012), were down-regulated under long-term waterlogging (Zhai et al., 2013). Based on the reports of responsive miRNAs to flooding, there is a remarkable trend for down-regulation of miRNAs in mid- and long-term treatments (Zhang et al., 2008; Jin et al., 2017).

When flood-tolerant and flood-sensitive inbred lines of maize were compared, long-term waterlogging down-regulated miRNAs in both lines. However, during the first few hours of waterlogging, in contrast with the sensitive line, the tolerant line up-regulated most of the responsive miRNAs (Liu et al., 2012). This suggests that upon early induction of miRNAs, the repression of functions associated with their target mRNAs, regarded as response signals to hypoxia (e.g., aerenchyma formation, lateral root development and cell detoxification), is a determinant for waterlogging tolerance through the efficient management of energy in oxygen-limited environments. In contrast, in the waterlogging-sensitive line, the down-regulation of miRNAs activated the hypoxia response signals already in the first hours of treatment, thus leaving the plant without the energy and oxygen required to withstand long-term stress.

Understanding the roles of miRNAs in response to flooding can be challenging due to the diversity of mature miRNAs that arise from a single family of miRNAs. Moreover, the mRNAs that are targeted by members of a miRNA family can regulate diverse functions in specific plant tissues. For example, several miRNAs with roles in flowering have been found in the roots of plants during flooding. The functions of these miRNAs and their targets in roots during flooding need to be addressed in order to increase our understanding of the functional diversity of miRNAs. Adding to this complexity, miRNAs function as common elements, connecting the response to different kinds of stress. For example, several miRNAs identified as responsive to flooding stress have also been described in other abiotic stress conditions. In general, all miRNAs related to morphological adaptation and oxidative homeostasis are common to other forms of abiotic stress. Characterizing in detail the expression patterns and function of such miRNAs in the acquisition of tolerance traits in plants represents an opportunity to improve tolerance to flooding in combination with other environmental stress.

The function of miRNAs in flooding stress is estimated by the identity and function of their target mRNAs and consequent downstream effects. In many of the studies mentioned above, the interactions of the miRNA:target modules were confirmed and validated by experimental approaches, while in others, targets were only predicted and need experimental validation. The application of combinatorial approaches, including transcriptome data, degradome, ARGONAUTE immunoprecipitation, and translatome analyses (Jeong et al., 2013; Zhai et al., 2013; Juntawong et al., 2014a; Franke et al., 2018), will be helpful for identifying and consolidating miRNA:target modules and downstream regulation.

In some cases, novel non-conserved miRNAs were identified that are responsive to flooding. These miRNAs are interesting because functional characterization has uncovered new functions potentially explaining group or species-specific traits in plants that have not been described in classic model systems (Ren et al., 2012; Jin et al., 2017; Li et al., 2017). In addition, the identification of miRNAs involved in flooding tolerance and differential strategies for tolerance could be facilitated by analysis of the natural variation using different ecotypes with contrasting

tolerance to flooding (Liu et al., 2012) and in flooding-tolerant models like Sesbania (Ren et al., 2017), Oryza (Nishiuchi et al., 2012), Rumex, and Rorippa (Voesenek et al., 2014).

Quantitative trait loci associated with morphological traits of interest can also be useful for identifying miRNAs involved in tolerance. For example, in maize, miR166, miR167, and miR319 are co-located with QTLs for tolerance to flooding (Osman et al., 2013). The association of miRNA genes and other genes with QTLs is a powerful strategy for obtaining insight into the molecular basis of quantitative traits and when considering miRNAs as new candidate genes for marker-assisted selection.

Many of the miRNAs responsive to flooding were described and their functions characterized in natural processes like ROS protection and development. The value of some miRNAs for biotechnology was explored, such as of miR156, miR164, miR166, miR172, miR319, and miR159, which have been directed to the improvement of biomass for biofuel production (Chuck et al., 2011; Trumbo et al., 2015). All these miRNAs are regulated by submergence, waterlogging, or hypoxia. However, those miRNAs have not been functionally characterized in detail in the context of tolerance to flooding. This knowledge will open a new area of opportunity for the use of next-generation genomic technologies for the improvement of tolerance to flooding in crop plants mediated by miRNAs.

#### Transposon Activation in Submergence/Hypoxia Stress

Much of the current knowledge about plant responses to flooding was obtained through transcriptional studies. It is increasingly recognized that other processes supporting the response can also be important and relevant for agricultural biotechnology (Sanchez, 2013). One of the unexplored pathways in flooding stress is the commonly found, but frequently overlooked, expression of retrotransposons (Mustroph et al., 2010; Rivera-Contreras et al., 2016). Stress has long been known to change the expression patterns of retrotransposons, and the significance of this is hypothesized to have many biological possibilities for acclimatization (Wessler, 1996). With the advent of NGS and accompanying statistical analyses (Xu et al., 2017), a picture has been drawn where transposons may act as transcriptomic enhancers, anchors for novel chromatin methylation patterns, and temporary inducers of expression plasticity in vegetative tissues (Springer et al., 2015; Negi et al., 2016). This knowledge is starting to be compiled as a viable biotechnological strategy to achieve crop improvement (Springer and Schmitz, 2017).

Few reports have explored the role of retrotransposons in flooding stress. Twenty years ago, it was reported that a transposon insertion in the promoter of a tomato 1 aminocyclopropane-1-carboxylate synthase gene (ACC) changed its expression pattern when compared to other members of the ACC family (Shiu et al., 1998). Transposon activity is also characteristic of rice and was detected in several family members (Mustroph et al., 2010).

Transposable elements have been studied in more depth in other kinds of stress. Under drought stress, transposons can also act as miRNA-coding RNAs by acting as promoters of anti-sense transcripts, as observed in maize (Xu et al., 2017). However, given the potential mutagenic action of expressing transposable elements, they are under tight regulation (Li et al., 2010), and natural mechanisms exist that limit their heritability after a stress event, for example, high temperature (Gaubert et al., 2017). The elimination of these controls in plants through different genetic and physiological strategies (including classics like tissue culture), can be a potential biotechnology strategy to induce superior cultivars (Paszkowski, 2015).

However, this elimination not always occurs through traditional expression enhancement, but yet to be characterized mechanisms may be involved that include epigenetic control (Virdi et al., 2015). Recently, an example of a non-canonical role for transposon expression was described in rice roots, showing that their mRNAs can be "miRNA sponges" that act as spurious targets of miRNAs and consequently protect the original target mRNA, and when those mRNAs code for TFs, the effect is amplified (Cho and Paszkowski, 2017). These studies indicate that research of flooding stress response can benefit from testing epigenetic mutation, retrotransposon activity and miRNA bioinformatics.

#### USE OF GENETIC DIVERSITY TO UNCOVER INTEGRATED MECHANISMS

Excessive water damages most terrestrial plants. However, genetic diversity in the tolerance to submergence and waterlogging was recognized in many plant species. For example, most rice cultivars die within 7 days under complete submergence (Bailey-Serres et al., 2010). However, a limited number of rice accessions including FR13A can endure complete inundation for up to 14 days (Fukao and Xiong, 2013). QTL analyses identified SUBMERGENCE-1 (SUB1), which positively affects submergence tolerance, on chromosome 9S of FR13A (Xu and Mackill, 1996; Nandi et al., 1997; Sripongpangkul et al., 2000; Toojinda et al., 2003). Introgression of this locus into submergence-intolerant japonica and indica accessions through marker-assisted backcrossing significantly increased submergence tolerance (Xu et al., 2006; Septiningsih et al., 2009). Map-based cloning and allelic survey revealed that the SUB1 locus encodes two to three ERF-VII genes, SUB1A, B and C, of which SUB1A is the determinant of submergence tolerance (Xu et al., 2006). The major function of SUB1A in submergence tolerance is to restrict carbohydrate consumption and amino acid metabolism by the suppression of ethylene production and GA responsiveness and to activate brassinosteroid synthesis, resulting in the restriction of shoot elongation and the avoidance of carbohydrate starvation underwater (Fukao et al., 2006; Fukao and Bailey-Serres, 2008; Barding et al., 2012, 2013; Schmitz et al., 2013; Tamang and Fukao, 2015). Transcriptome analyses indicated that SUB1A up-regulates genes involved in ROS detoxification and anaerobic respiration, but down-regulates genes associated with macromolecule biosynthesis (Jung et al., 2010; Mustroph et al., 2010; Locke et al., 2018).

Some rice accessions promote internode elongation in response to flooding, enabling the plants to outgrow gradually

rising floodwaters (Fukao and Xiong, 2013). QTL analysis revealed a major locus regulating this response, in which the tandemly repeated ERF-VII genes SNORKEL1 and 2 are located (Hattori et al., 2009). Consistent with SUB1A (Fukao et al., 2006), SNORKEL1 and 2 are ethylene-inducible (Hattori et al., 2009). It is likely that these deepwater-rice-specific ERF-VIIs are involved in gibberellin biosynthesis (Ayano et al., 2014). Recent genome-scale gene expression analysis using deepwater and non-deepwater accessions revealed that genes associated with gibberellin biosynthesis and responsiveness, cell wall loosening and extension, and disease resistance are up-regulated in deepwater rice, whereas genes involved in lignin synthesis and oxidative respiration are down-regulated (Minami et al., 2018).

The ability of rice to elongate its seedling shoot under complete submergence is critical for direct seeding. A Myanmar landrace, Khao Hlan On, exhibits strong tolerance of anaerobic germination (Ismail et al., 2009), and QTL analysis identified a major locus responsible, qAG-9-2 (Angaji et al., 2010). This locus contains a gene encoding trehalose-6-phosphate phosphatase, TPP7, and a near-isogenic line containing this locus showed increased coleoptile growth underwater (Kretzschmar et al., 2015). Trehalose-6-phosphate functions as an inhibitor of SnRK1 activity in growing tissues (Zhang et al., 2009; Paul et al., 2010; Delatte et al., 2011). The proposed function of TPP7 is to promote the conversion of trehalose-6-phosphate to trehalose, activating the SnRK1-dependent signaling cascade to increase sink strength in growing embryos and coleoptiles (Kretzschmar et al., 2015). In contrast to this TPP7-mediated mechanism, SUB1A is suggested to restrict the SnRK1-dependent pathway to suppress carbohydrate breakdown in elongating leaves (Locke et al., 2018).

These advances in rice research have not been matched in other crops. Still, studies on QTL identification shed valuable light into what mechanisms, of those known, serve not only as "hardware" of the flooding stress response, but also as natural hubs to increase tolerance. For example, Subtol6, a QTL associated with constitutive higher expression of the HEMOGLOBIN gene in tolerant maize cultivars (Campbell et al., 2015) supports the role of NO signaling, a result that was replicated via RNA-Seq of sensitive and tolerant ecotypes of Brachypodium (Rivera-Contreras et al., 2016).

Thorough genetic characterization of the dozens of QTLs associated with populations that have contrasting anatomical adaptations useful for flood stress tolerance, especially in roots, is expected to help to identify novel components (Mustroph, 2018). It is not uncommon to find genes with unknown function in the associated locus (even in model Arabidopsis), or multiple and closely located genes promising at the physiological level (e.g., ROS and P signaling; Akman et al., 2017), and awaiting transgenic and introgression analysis (Osman et al., 2013; Zhang et al., 2017).

#### BIOTECHNOLOGICAL SUCCESS AND EXPLORATIONS

#### SUB Locus Introgressions

Identification of the SUB1 locus and its flanking markers allowed rice breeders to develop rice cultivars that are submergence-tolerant through marker-assisted backcrossing. The submergence-tolerance gene, SUB1A, is expressed only during submergence, and its mRNA abundance quickly drops when floodwaters subside (Xu et al., 2006; Fukao et al., 2011). Due to this regulation in time, rice varieties in which the SUB1 region was introgressed have the same yield, key agronomic traits, and grain quality as the background varieties under regular growth conditions (Singh et al., 2013). On the other hand, SUB1 rice varieties show greater grain yield after 1–2 weeks of submergence. SUB1-introgressed lines were generated by many rice-breeding programs in various countries. Currently, more than four million rice farmers grow SUB1 rice in Asia (Ismail et al., 2013).

#### N-Terminal Rule Mutants

Despite the fact that submergence stress is a multicomponent stress, many mutants that were characterized and provided information on the biological mechanisms expressed during submergence/hypoxia have tolerant phenotypes. However, taking aside the well-known case of SUB1A, few have been biotechnologically explored in crops. Notably, barley PRT6 mutants developed by TILLING or RNAi, showed increased tolerance to waterlogging, and maintain pre-stress seed weight (Mendiondo et al., 2016). Further challenging of these mutants proved they are also tolerant to salinity and drought (Vicente et al., 2017), and pathogens (Vicente et al., 2018), although susceptible to Fusarium.

### Fruit Conservation and Hypoxia

For a long time, it has been known that fruits respond differentially to ethylene, and this has allowed classifying fruits in two groups depending on their response. Those responsive to ethylene bursts by displaying high respiration rates are named climacteric if the response is strong enough to induce fast ripening, or non-climacteric if the response is milder or slow (Paul and Pandey, 2014). In this way, hypoxic atmospheres have long been used to improve fruit quality, by empirically adjusting the optimal parameters of oxygen and temperature for each genotype, to achieve the desired inhibitory feedback of ethylene production (Lum et al., 2017). Recently, plant signaling and expression studies have reported that CHGs underlie this technology; a network of interacting TFs, including ERF-VIIs, are active and responsible for fermentative metabolism and tannin depletion in persimmon fruit (Diospyros kaki), with the practical consequence of decreased astringency (Zhu et al., 2018).

Given that this technology for ripening delay depends on the strength of ethylene feedback inhibition, not only inhibitors of ethylene perception are effective (Lum et al., 2017) but also NO fumigation (Mahajan et al., 2014), a result that is logic now that the role of NO in hypoxia modulation is clearer (see section "Nitric Oxide as a Central Homeostatic Regulator of Hypoxic Stresses") (Gibbs et al., 2014). However, an excess of hypoxia stress leads to the expression of CHGs and flavor alteration by the accumulation of fermentative products (Cukrov et al., 2016). As mentioned before, the threshold between conservation and fermentation is highly variable in fruits and vegetables (Beaudry, 1999); therefore, a solution to further monitor this balance is

the use of metabolomic fingerprinting by combining state-of-theart analytical methods, informatics, and agronomical knowledge, as recently shown for coffee (Gamboa-Becerra et al., 2017). The use of tissue or developmental stage-specific promoters to drive the timely expression of RNAi constructs against transcripts of ethylene receptors or fermentative enzymes may also render interesting biotechnological prototypes for fruit conservation technology.

#### Biomass for Biofuel Production in Flood-Prone Lands

An interesting alternative proposed for employing flood-prone lands in the tropics is the cultivation of sugarcane, which was reported as a flooding-tolerant plant that may even benefit from short flooding periods in the form of increased sugar yields (Glaz and Gilbert, 2006; Ray et al., 2010). The genetic potential of sugarcane was demonstrated in a multi-harvest fieldtest where family clones of sugarcane, valued for their sugar or biomass yields, were able to retain productivity of both in the face of intermittent flooding (Viator et al., 2012). This ability led to sugarcane being proposed as a component of wastewater treatments in constructed wetlands, where it thrived and removed phosphorus (Mateus et al., 2014).

The use of sugarcane on these types of low-quality soils is being promoted in connection with bioenergy production from the obtained biomass (Viator et al., 2012) since it would not compete in the "food vs. fuel" paradox (Peña-Castro et al., 2017). Research is also being performed to expand the climate range where sugarcane can be cultivated through hybridization with Miscanthus (Glowacka et al., 2016), a flood- and chilling-tolerant grass (Mann et al., 2013).

A phenotype that limits the use of other tall grasses as bioenergy crops is lodging (Rueda et al., 2016), i.e., reduction of the stalk's angle of growth to the soil. Lodging is also a negative consequence of flooding in rice (Jackson and Ram, 2003) and lodging resistance is an agronomic trait that rice breeders intensely focus on (Hirano et al., 2017). A gene coding for gibberellin 2-oxidase, a GA deactivating enzyme, was found to improve lodging resistance (Liu et al., 2018) and members of this gene family are down-regulated in SUB1 rice varieties (Jung et al., 2010). Another example is the allele SEMIDWARF1 that encodes a gibberellin 20-oxidase, an enzyme of GA biosynthesis, that when active promotes node elongation in deepwater rice in response to flooding, and when silent, is a historic allele that allowed the development of the short high-yield rice of the Green Revolution (Kuroha et al., 2018). Both contrasting examples represent an interesting connection between the fields of bioenergy, submergence tolerance and food security.

In Jatropha curcas, a promising biodiesel crop that can produce high oil content in seeds and can be grown on marginal land without competing with other food crops, waterlogging results in significant reduction of growth and biomass yield, implying it is highly sensitive to waterlogging (Gimeno et al., 2012; Verma et al., 2014). Due to its narrow genetic background, the genetic diversity of waterlogging-tolerant Jatropha has not yet been reported. However, based on transcriptomic analysis of waterlogged Jatropha roots, several candidate genes, including NR, NiR, and ERF-VIIs, could be targeted for genetic engineering of waterlogging tolerance (Juntawong et al., 2014b).

#### The Potential of Genome Editing for Developing Biotechnological Applications for Submergence/Waterlogging Stress

The advent of editing technologies has opened a new era of discovery in many fields of biotechnology including agricultural research (Doudna and Charpentier, 2014). However, there is only one reported example of gene editing in the field of submergence/waterlogging stress. Yamauchi et al. (2017) edited rice RESPIRATORY BURST OXIDASE HOMOLOG truncated genes (RBOHH) to demonstrate the crucial role that peroxide produced by this enzyme has on waterlogging signaling and aerenchyma formation. Unfortunately, homozygous plants proved to be sterile and this prevented further phenotyping.

A recent report provides an interesting framework for genome editing for plant biotechnological purposes. Miao et al. (2018) employed Cas9 to create mutants for all PYRABACTIN RESISTANCE 1 (PYL) ABA receptors in rice and explored all combinations in search of genotypes of biotechnological interest. Of all possible combinations, pyl1/4/6 maximally increased productivity and growth by being released from ABA-related natural growth restrictions but without losing ABA-controlled positive phenotypes, like unwanted seed sprouting, presented by many other mutant combinations. The accumulated knowledge of submergence/flooding stress seems to us to be mature enough to implement this type of powerful gene and genome editing strategies to many of its controlling branches, notably the complex network of TF families.

These promising innovation technologies can produce varieties in which single bases are edited (Chen et al., 2018) up to gene fusions and substitutions (Miki et al., 2018) and even transgene-free genomes (Zhang Y. et al., 2016). The precision and type of induced mutations are less disruptive than many of the mutations empirically conserved during plant domestication history and that prevail in the most valuable traits of major crops of all continents (Meyer and Purugganan, 2013). Regulators must take this into account to avoid disappointing decisions on restraints to the release of gene-edited varieties (Callaway, 2018) and maintain incentives for innovations in research fields where their full potential has not been thoroughly tested, as is the case with submergence/flooding stress.

#### Research Structures and Dissemination of Technology

The discovery of the SUB1 locus and its wide implementation in Asia is a success story in the field of plant biotechnology (Bailey-Serres et al., 2010; Dar et al., 2013; Mickelbart et al., 2015; Dar et al., 2018) and plant molecular biology (Haak et al., 2017). In addition, it should also be a paradigm for research structures that allow the discovery, characterization, and implementation of agricultural phytotechnologies. The research programs of the International Rice Research Institute (IRRI) are solidly based on

the diversity of wild and domesticated rice germplasm collected and maintained—for almost a century in an objective-guided multi-year research plan, and by a committed international scientist nucleus with expertise from agriculture economics to molecular biology (International Rice Research Institute [IRRI], 2017).

Internationally, other CGIAR centers, notably the International Maize and Wheat Improvement Center (CIMMYT), have started to develop similar integral strategies for agricultural research through the agricultural innovation system called Seeds of Discovery–MasAgro (Pixley et al., 2018). This is a strategy aimed at fully employing the maize and wheat germplasm collection using high-throughput genotyping (mostly by NGS) and integrated bioinformatics tools. This framework has produced large-scale genetic data related to flowering time in >4,000 maize landraces (Romero et al., 2017), genetic diversity and its relationship with drought and heat stress of >8,000 wheat accessions in Mexico (Vikram et al., 2016), and the screening of 400 inbred lines of maize and expression analysis of waterlogging-tolerant and -intolerant lines (Du et al., 2017).

The implementation of this specialized genetic characterization as a technology is challenging, especially in countries with a vast range of microclimates such as, e.g., Mexico. Therefore, MasAgro is an innovation platform that connects CIMMYT high-end research and large genetic resources with field-testing, but not in fixed institutionally owned locations, but in a series of hundreds of micro-climate locations where the partner is usually a local company challenging global companies with improved seeds developed for specific locations, or communal associations (Camacho-Villa et al., 2016; Hellin and Camacho, 2017). The result is that genetic information is screened simultaneously in many locations and the most valuable hybrids and landraces are kept and commercialized. These agricultural innovation systems are expected to fill the gap between research and social impact (Westermann et al., 2018). From a scientific standpoint, the flow of genetic material from the lab to the field and back will help develop plant genetic resources valuable for discovering integrated molecular mechanisms that are relevant across environments and flood-related stresses, or to clarify the relative importance, in the context of large genetic-diversity screening, of all those mechanisms herein reviewed and described.

#### REFERENCES


#### CONCLUSION

Scientific understanding of plants' responses to submergence and waterlogging has dynamically evolved from the pioneering works dealing with fermentative metabolic changes to current state-ofthe-art research creating a picture of interconnected perception, transduction and signaling events aimed to support the plant cell in the transit from stress up to the — always expected recovery phase. These events were demonstrated to involve all possible molecular mechanisms, from mineral signaling to large ribonucleoprotein complexes, in continuous crosstalk, each step adding fitness to the response.

Still, more connections await characterization. In this review we have noted modulation of miRNAs, RNA dynamics and primary energy signaling as some that may benefit from new cutting-edge technologies, developed both inside the field or in parallel areas of plant biology.

The accumulated knowledge prompts us to affirm that the field is sufficiently mature to start moving much of this knowledge to explicit biotechnological tests directed at mitigating the deleterious effects of flooding in the farmer's economy. The success story of SUB1-introgressed lines provides much experience to adapt efforts to other regions of the world commonly affected by this stress.

# AUTHOR CONTRIBUTIONS

All authors contributed to the research, writing, and review processes for this article.

#### FUNDING

The review was supported by research grants from the Consejo Nacional de Ciencia y Tecnología - México (Ciencia Básica 287137) to JMP-C, Virginia Corn Board, Virginia Agricultural Council, the Virginia Agricultural Experiment Station, and the Hatch Program of the National Institute of Food and Agriculture, United States Department of Agriculture to TF, Thailand Research Fund (MRG5980033), and Kasetsart University Research and Development Institute (KURDI) to PJ.




or saline water in Jatropha curcas. Environ. Exp. Bot. 78, 47–55. doi: 10.1016/j. envexpbot.2011.12.014


flood-prone rainfed lowland areas in Asia. Field Crops Res. 152, 83–93. doi: 10.1016/j.fcr.2013.01.007


Arabidopsis thaliana ecotype Columbia. New Phytol. 190, 457–471. doi: 10. 1111/j.1469-8137.2010.03590.x






regulation of metabolic pathways by trehalose-6-PHOSPHATE1. Plant Physiol. 149, 1860–1871. doi: 10.1104/pp.108.133934


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

Copyright © 2019 Fukao, Barrera-Figueroa, Juntawong and Peña-Castro. 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.

# Chloroplast Redox Regulatory Mechanisms in Plant Adaptation to Light and Darkness

Francisco Javier Cejudo\*, Valle Ojeda, Víctor Delgado-Requerey, Maricruz González and Juan Manuel Pérez-Ruiz

Instituto de Bioquímica Vegetal y Fotosíntesis, Consejo Superior de Investigaciones Científicas, Universidad de Sevilla, Seville, Spain

Light is probably the most important environmental stimulus for plant development. As sessile organisms, plants have developed regulatory mechanisms that allow the rapid adaptation of their metabolism to changes in light availability. Redox regulation based on disulfide-dithiol exchange constitutes a rapid and reversible post-translational modification, which affects protein conformation and activity. This regulatory mechanism was initially discovered in chloroplasts when it was identified that enzymes of the Calvin-Benson cycle (CBC) are reduced and active during the day and become rapidly inactivated by oxidation in the dark. At present, the large number of redox-sensitive proteins identified in chloroplasts extend redox regulation far beyond the CBC. The classic pathway of redox regulation in chloroplasts establishes that ferredoxin (Fdx) reduced by the photosynthetic electron transport chain fuels reducing equivalents to the large set of thioredoxins (Trxs) of this organelle via the activity of a Fdx-dependent Trx reductase (FTR), hence linking redox regulation to light. In addition, chloroplasts harbor an NADPH-dependent Trx reductase with a joint Trx domain, termed NTRC. The presence in chloroplasts of this NADPH-dependent redox system raises the question of the functional relationship between NTRC and the Fdx-FTR-Trx pathways. Here, we update the current knowledge of these two redox systems focusing on recent evidence showing their functional interrelationship through the action of the thioldependent peroxidase, 2-Cys peroxiredoxin (2-Cys Prx). The relevant role of 2-Cys Prxs in chloroplast redox homeostasis suggests that hydrogen peroxide may exert a key function to control the redox state of stromal enzymes. Indeed, recent reports have shown the participation of 2-Cys Prxs in enzyme oxidation in the dark, thus providing an explanation for the long-lasting question of photosynthesis deactivation during the light-dark transition.

Keywords: chloroplast, hydrogen peroxide, light, darkness, peroxiredoxin, photosynthesis, redox regulation, thioredoxin

# INTRODUCTION

Photosynthesis is the process that allows the use of light energy for biomass production using water as source of reducing power; hence, chloroplasts have the function of providing the metabolic intermediates that support plant growth. These intermediates include molecules with signaling function such as hormones and hormone precursors, so that chloroplast performance also plays an

#### Edited by:

Sang Yeol Lee, Gyeongsang National University, South Korea

#### Reviewed by:

David Kramer, Michigan State University, United States Pascal Rey, Commissariat à l'Energie Atomique et aux Energies Alternatives (CEA), France

> \*Correspondence: Francisco Javier Cejudo fjcejudo@us.es

#### Specialty section:

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

Received: 05 December 2018 Accepted: 12 March 2019 Published: 04 April 2019

#### Citation:

Cejudo FJ, Ojeda V, Delgado-Requerey V, González M and Pérez-Ruiz JM (2019) Chloroplast Redox Regulatory Mechanisms in Plant Adaptation to Light and Darkness. Front. Plant Sci. 10:380. doi: 10.3389/fpls.2019.00380

**142**

important role in the harmonization of the growth of the different organs during all stages of plant development. Therefore, light is a key environmental factor for plant growth and development. Some of the changes in light availability, i.e., intensity and quality, vary in a predictable manner and plants, like other organisms, are able to anticipate these changes by the circadian clock. However, in nature, light availability changes continuously and, consequently, chloroplast performance needs to be rapidly adapted to these unpredictable conditions.

Central for the ability of chloroplast metabolism to rapidly respond to changes in light intensity is thiol-based redox regulation, which relies on the extraordinary properties of the thiol group of cysteines (Cremers and Jakob, 2013). Considering random distribution of amino acids in proteins, cysteine is underrepresented in all organisms though its presence appears to correlate with complexity ranging between 2.26% in mammals to 0.5% in some archaebacteria (Miseta and Csutora, 2000). The thiol group of cysteines is very sensitive to oxidant conditions, being able to react with hydrogen peroxide so that it may be oxidized as sulfenic (−SOH), sulfinic (−SO2H), and even sulfonic acid (−SO3H) (Cremers and Jakob, 2013). In its reduced form, cysteine may react with another cysteine forming a disulfide bridge. Cysteine residues involved in dithiol-disulfide redox exchange usually show a high degree of conservation in redox-sensitive proteins, the conformation and activity of which is deeply affected by the redox state of these pairs of cysteines.

Dithiol-disulfide exchange constitutes a universal regulatory mechanism present in all types of organisms from bacteria and fungi to plants and animals. The reduction of disulfide bridges in redox-regulated proteins relies on the protein disulfide reductase activity of thioredoxins (Trxs) and glutaredoxins (Grxs) (Meyer et al., 2012). However, most studies in different types of organisms have focused on the regulatory properties of Trxs, small polypeptides of 12–14 kDa with a well-conserved active site (WCGPC), which were initially described as cofactors of ribonucleotide reductase from Escherichia coli (Laurent et al., 1964). In heterotrophic organisms, the reducing power required for Trx activity is provided by NADPH via the action of an NADPH-dependent Trx reductase (NTR) (Jacquot et al., 2009; **Figure 1**). Interestingly, despite the large number of proteins that undergo redox regulation in heterotrophic organisms, the gene families encoding NTRs and Trxs is rather low, having typically two, at most three, members (Meyer et al., 2012). Redox regulation is also very important in plant chloroplasts; however, in these organelles, this regulatory mechanism shows remarkable differences as compared to heterotrophic organisms. In this review, we discuss the complex redox regulatory network of plant chloroplasts, focusing on the relevance of redox regulation for the rapid adaptation of chloroplast metabolism to light availability.

#### THIOL-DEPENDENT REDOX REGULATION IN PLANT CHLOROPLASTS

Redox regulation of enzyme activity was first discovered by the seminal work of Prof. Buchanan at Berkeley ( Buchanan, 2016). The initial key finding was that photo-reduced Fdx is required for the activation of the Calvin-Benson cycle (CBC) enzyme fructose bisphosphatase (FBPase) (Buchanan et al., 1967). Further analyses uncovered the participation of two components, Trx and an Fdx-dependent Trx reductase (FTR), which fuel reducing equivalents from photo-reduced Fdx for the reduction of a regulatory disulfide of FBPase, which rendered a more active form of the enzyme (Buchanan et al., 1971; Wolosiuk and Buchanan, 1977). Different approaches based on biochemical analyses soon identified two isoforms of Trxs in chloroplasts, f and m, named on their ability to preferentially activate FBPase and NADP-malate dehydrogenase, respectively (Buchanan, 2016). More recently, the enormous availability of genome sequence data from a large number of plant species has shown that chloroplasts harbor a complex set of up to twenty Trxs and Trx-like proteins (Buchanan et al., 2012; Meyer et al., 2012; Balsera et al., 2014; Geigenberger et al., 2017). In addition, mass spectrometry analyses in conjunction with techniques for trapping proteins interacting with Trxs have allowed the identification of a large number of putative Trx targets from chloroplasts (Montrichard et al., 2009). Although the redox regulation of many of these proteins awaits in vivo experimental confirmation, the number of putative Trx-interacting enzymes so far identified suggests that redox regulation may virtually affect any process occurring in the chloroplast, being thus a regulatory mechanism essential for the rapid adaptation of chloroplast performance to ever changing light conditions.

The classical view of redox regulation in chloroplasts, based on the FTR-Trx system, relies on Fdx reduced by the photosynthetic electron transport chain, which thus directly links redox regulation in the organelle to light (Schürmann and Buchanan, 2008). This is in clear contrast with heterotrophic organisms, where redox regulation relies on NADPH and the NTR-Trx redox system (Buchanan, 2017). Therefore, a major difference in redox regulation between chloroplasts and heterotrophic organisms is the source of reducing power used for this regulatory mechanism (**Figure 1**). This notion was modified by the discovery of NTRC, a novel NTR with a joint Trx domain at the C-terminus (Serrato et al., 2002, 2004). NTRC is exclusively found in organisms that perform oxygenic photosynthesis, namely plants, in which it is encoded by one or two genes, algae and some, but not all, cyanobacteria (Pascual et al., 2010; Nájera et al., 2017). In plants, NTRC shows localization in all types of plastids, either from photosynthetic or non-photosynthetic tissues (Kirchsteiger et al., 2012); however, it is a relatively abundant protein in chloroplasts where it shows stromal localization (Serrato et al., 2004; Moon et al., 2006; Pérez-Ruiz et al., 2009). Initial biochemical analyses of NTRC, based on the study of truncated polypeptides containing the NTR or the Trx domain of the enzyme, confirmed the NTR and Trx activities, respectively, of these domains (Serrato et al., 2004). Thus, based on these results, we suggested that NTRC is a bi-functional enzyme, which might function either as NTR or as Trx in the chloroplast. Soon afterward, it was reported that NTRC is able to efficiently reduce 2-Cys peroxiredoxins (2-Cys Prxs) (Moon et al., 2006; Pérez-Ruiz et al., 2006; Alkhalfioui et al., 2007). Indeed, the incubation of NTRC and 2-Cys Prx allows the rapid reduction of hydrogen peroxide

metabolism to ever changing light availability. Chloroplasts harbor a complex set of Trxs and Trx-like proteins, which rely on photo-reduced Fdx via the participation of FTR, thus linking redox regulation of enzyme activity to light. In addition, chloroplasts contain an additional redox system, NTRC, which relies on NADPH (see below, Figure 2).

using NADPH as source of reducing power, hence showing that the enzyme is able to conjugate both NTR and Trx activities for the efficient reduction of these thiol-peroxidases. Further biochemical analyses using mutated versions of the enzyme suggested that the catalytic active form of NTRC is a dimer arranged in a head-to-tail conformation, which interacts with 2- Cys Prxs through the Trx domain (Pérez-Ruiz and Cejudo, 2009; Bernal-Bayard et al., 2012). Overall, these evidences uncover the use of NADPH to support the antioxidant function of 2-Cys Prxs in chloroplasts (Pérez-Ruiz et al., 2006). It is known that 2- Cys Prxs, which are among the most abundant proteins of the chloroplast stroma (Peltier et al., 2006), show a very efficient hydrogen peroxide scavenging activity (Dietz, 2011; Perkins et al., 2015; Liebthal et al., 2018). In addition to NTRC, chloroplast Trxs and Trx-like proteins also show capacity of 2-Cys Prx reduction in vitro (Broin et al., 2002; Collin et al., 2003, 2004; Dangoor et al., 2012; Eliyahu et al., 2015; Hochmal et al., 2016; Vaseghi et al., 2018; Yoshida et al., 2018). In vitro assays comparing several chloroplast Trxs indicated that type-x Trx is the most efficient 2-Cys Prx reductant (Collin et al., 2003). The comparison of the in vivo redox state of 2-Cys Prxs in NTRC and Trx x Arabidopsis knock out mutants showed a severe impairment of the redox state of the 2-Cys Prxs in the ntrc mutant, whereas in the trxx mutant it was indistinguishable of the wild type (Pulido et al., 2010). Based on these results, it was proposed that NTRC is the most relevant reductant of 2-Cys Prxs in vivo. Because NTRC interacts with 2-Cys Prxs by the Trx domain, it could be considered that NTRC is a Trx that incorporates its own reductase, which would explain the high catalytic efficiency of this enzyme (Cejudo et al., 2012). However, the presence of an NTR domain in NTRC raises the question of whether this enzyme also enables the transfer of reducing

equivalents from NADPH to plastidial Trxs. The overexpression of the NTR domain of NTRC in the Arabidopsis ntrc mutant partially recovered the wild type phenotype, suggesting that NTRC displays NTR activity and might interact with chloroplast Trxs, in particular with Trx f (Toivola et al., 2013). This notion, however, is questioned by the fact that NTRC is unable to reduce neither of the chloroplast Trxs in vitro (Bohrer et al., 2012). Therefore, whether NTRC interacts with its targets exclusively through the Trx domain, as it does with 2-Cys Prxs, or through the NTR domain, as it was proposed for Trx f, remains an open issue.

# THE CHLOROPLAST REDOX SYSTEMS NTRC and Fdx-FTR-Trx ACT CONCERTEDLY

The identification of NTRC as an NADPH-dependent redox system established the presence of two redox pathways in chloroplasts, hence raising the question of the functional relationship between them. This issue is being addressed through reverse genetic approaches in Arabidopsis thaliana. The deficiency of different chloroplast Trxs results in a surprising low effect on plant growth. This is the case of knock out mutants for Trx x (Pulido et al., 2010) or Trxs y (Laugier et al., 2013), which show growth phenotypes very similar to the wild type. More intriguingly, despite the major role proposed for Trxs f in light-dependent redox regulation of CBC enzymes (Michelet et al., 2013), the double knockout mutant of Arabidopsis lacking Trxsf 1 and f 2 show almost wild type growth phenotype (Yoshida et al., 2015; Naranjo et al., 2016a). An in-depth analysis of the in vivo redox state of FBPase in this mutant revealed that the light-dependent redox regulation of the enzyme was only partially affected (Naranjo et al., 2016a), indicating that other chloroplast Trxs contribute to the redox regulation of FBPase. In this regard, it was shown that m-type Trxs have a relevant function in the light-dependent redox regulation of CBC enzymes (Okegawa and Motohashi, 2015). There are four isoforms of type m Trxs in Arabidopsis (Okegawa and Motohashi, 2015). Single mutants deficient in Trxs m1 and m4 show growth phenotypes similar to the wild type (Courteille et al., 2013; Laugier et al., 2013), though the deficiency of Trx m4 causes up-regulation of the NADH dehydrogenase-like complexdependent plastoquinone reduction pathway of photosynthetic electron transport (Courteille et al., 2013). Notably, mutant plants devoid of Trx m3, the less abundant m-type Trx (Okegawa and Motohashi, 2015) showed unaffected chloroplast performance but impaired symplastic trafficking (Benítez-Alfonso et al., 2009). To our knowledge, no mutants completely devoid of the four m-type Trxs have been reported, however, approaches based on gene silencing generated Arabidopsis plants with very decreased contents of Trxs m1, m2 and m4, which allowed to propose a function of these m-type Trxs in photosystem II (PSII) biogenesis (Wang et al., 2013). Plants impaired in the expression of the variable subunit of FTR, which provides electrons to plastidial Trxs, display marked phenotype traits such as sensitivity to oxidative stress, impaired light-dependent reduction of Trxregulated enzymes and increased contents of 2-Cys Prxs (Keryer et al., 2004). The latter trait giving further support to the tight relationship between FTR and NTRC redox systems. Finally, it should be mentioned that the deficiency of Trx z affects chloroplast transcription, thus compromising chloroplast biogenesis (Arsova et al., 2010). However, it is not clear whether the role of Trx z in the expression of plastid-encoded genes is redox-dependent (Wimmelbacher and Bornke, 2014).

In clear contrast, the Arabidopsis NTRC knockout mutant, ntrc, shows a characteristic phenotype consisting in retarded growth and pale green leaves, with about 70–75 % of the chlorophyll contents of the wild type plants (Serrato et al., 2004; Lepistö et al., 2009). Interestingly, the growth phenotype of the ntrc mutant is highly dependent on light availability; i.e., growth retard is aggravated when plants are grown under shortday conditions, as compared with long-day conditions (Pérez-Ruiz et al., 2006; Lepistö et al., 2009), and fluctuating light intensities (Thormählen et al., 2017). In addition, the Arabidopsis ntrc mutant shows high sensitivity to different abiotic stresses including high salt (Serrato et al., 2004), prolonged darkness (Pérez-Ruiz et al., 2006), and high temperature (Chae et al., 2013), as well as to biotic stress (Ishiga et al., 2012, 2016). As 2-Cys Prxs are very efficient hydrogen peroxide scavengers and NTRC is the most efficient reductant of these enzymes, the increased sensitivity of the ntrc mutant to biotic and abiotic stresses might be in agreement with the antioxidant function proposed for the enzyme. Therefore, the antioxidant function proposed for NTRC implies that NADPH serves as source of reducing power in chloroplasts, at least to support the hydrogen peroxide scavenging activity of 2-Cys Prxs (Spínola et al., 2008). This notion not only suggested a relevant role for NADPH in chloroplast redox homeostasis, it changes the paradigm of redox regulation linked to light, as NADPH is produced in chloroplasts from photo-reduced Fdx but also from sugars by the oxidative pentose phosphate pathway during darkness.

In summary, the approaches based on the analysis of Arabidopsis mutants show that while the deficiency of NTRC causes a severe effect on plant growth and development, the deficiency of different types of chloroplast Trxs, with the exception of Trx z, has low or no effect on plant growth, indicating the functional redundancy of these enzymes. Moreover, the fact that the absence of NTRC has such severe phenotype suggests the participation of this enzyme in different aspects of chloroplast redox homeostasis, besides its proposed antioxidant function as electron donor to 2-Cys Prxs. Key chloroplast metabolic pathways, such as chlorophyll and starch biosynthesis, and light energy utilization, which were known to be regulated by the canonical FTR/Trx pathway, are also affected by NTRC. A first indication of the effect of NTRC on redox-regulated processes was the finding that the pathway of chlorophyll biosynthesis is impaired in the ntrc mutant (Stenbaek et al., 2008). Initially, it was proposed that the positive effect of NTRC is exerted by the reduction of 2-Cys Prxs, which protects the aerobic cyclase activity of oxidant conditions. Further analyses revealed a direct role of NTRC on the redox regulation of the chlorophyll biosynthesis enzyme

MgP methyltransferase (CHLM) and direct interaction of NTRC with CHLM and glutamyl-transfer RNA reductase1 (GluTR1) (Richter et al., 2013). Moreover, the content of both CHLM and GluTR enzymes were decreased in the ntrc mutant, suggesting that NTRC also affects the stability of these enzymes (Richter et al., 2013). In line with this observation, the I subunit (CHLI) of Mg chelatase, a redox-regulated enzyme that catalyzes the first dedicated step of chlorophyll biosynthesis (Ikegami et al., 2007), was shown to be regulated by NTRC (Pérez-Ruiz et al., 2014). Taken together, these results strongly suggest the participation of NTRC in the redox regulation of the chlorophyll biosynthesis pathway. In addition, NTRC participates in the redox regulation of starch biosynthesis (Michalska et al., 2009). A key regulatory step of this pathway is catalyzed by ADP glucose pyrophosphorylase (AGPase), a heterotetrameric enzyme formed by two small and two large subunits. Under illumination, AGPase is activated by the reduction of a disulfide bridge that links the two small subunits, a process that was described to be regulated by Trxs (Geigenberger, 2011). Indeed, Arabidopsis mutant plants deficient in Trx f 1 show decreased light-dependent activation of AGPase (Thormählen et al., 2013). Interestingly, the Arabidopsis ntrc mutant shows decreased starch contents in leaves, and

impaired reduction of AGPase in response to light (Michalska et al., 2009). Moreover, NTRC interacts and triggers the reduction of the enzyme in vitro, thus indicating the participation of NTRC in the light-dependent reductive activation of AGPase (Michalska et al., 2009), a notion further confirmed by the analysis of the regulation of starch metabolism in the ntrc mutant grown under short-day conditions (Lepistö et al., 2013). Altogether, these studies show that redox regulation of chlorophyll and starch biosynthesis, two key metabolic pathways previously shown to be regulated by Trxs, are also regulated by NTRC, indicating overlapping functions of the FTR-Trxs and NTRC redox systems in plant chloroplasts. Finally, the analysis of photochemical parameters revealed that mutant plants lacking NTRC show high non-photochemical quenching (NPQ) (Thormählen et al., 2015; Carrillo et al., 2016; Naranjo et al., 2016b). Although the ntrc mutant shows altered xanthophyll cycle (Naranjo et al., 2016b) and impaired reduction of the γ subunit ATP synthase under low irradiance, which causes higher lumen acidification and lower electron transport rate (Carrillo et al., 2016), the ultimate reason to explain how NTRC affects the efficiency of light energy utilization remains unknown. Nevertheless, these results reveal that NTRC does not only participate in the light-dependent redox

equivalents from NADPH to support the hydrogen peroxide scavenging activity of these thiol-dependent peroxidases. Though at lower efficiency, chloroplast Trxs are also able to reduce 2-Cys Prxs. Different studies have shown the participation of NTRC in the upstream photochemical reactions of photosynthesis and the redox regulation of the γ subunit of ATP synthase (γ), and of downstream targets including enzymes of the Calvin-Benson cycle, starch and chlorophyll biosynthesis. Moreover, NTRC might display NTR activity reducing chloroplast Trxs such as Trx f.

regulation of metabolic pathways including the biosynthesis of starch and chlorophyll, it does also participate in the regulation of photochemical reactions such as NPQ. Therefore, the action of NTRC in chloroplast performance is exerted by the antioxidant function of the enzyme due to its capacity to reduce 2-Cys Prxs, but also on upstream photochemical reactions and on downstream redox-regulated targets (**Figure 2**).

A useful approach for understanding the functional relationship between the FTR-Trxs and NTRC redox systems was the analysis of Arabidopsis mutants simultaneously deficient in both pathways. While mutant plants knock out for Trx f 1 (Thormählen et al., 2013), or Trxsf 1 and f 2 (Yoshida et al., 2015; Naranjo et al., 2016a) showed slight growth phenotype effects, the combined deficiencies of Trxs f and NTRC, the ntrc-trxf1f2 mutant, produced a very severe growth inhibition phenotype (Thormählen et al., 2015; Ojeda et al., 2017). The dramatic growth inhibition phenotype of these mutants is in agreement with the severe decrease of light energy utilization efficiency (Ojeda et al., 2017). Similarly, the lack of Trx x has almost no effect on plant growth (Pulido et al., 2010), whereas the simultaneous deficiency of Trx x and NTRC causes as well a dramatic effect of plant growth, much more severe than that of the trxx or ntrc single mutants (Ojeda et al., 2017). Interestingly, a remarkable feature of the ntrc-trxf1f2 and the ntrc-trxx mutants is the high mortality at the seedling stage (Ojeda et al., 2017), suggesting that chloroplast redox regulatory mechanisms are essential at the cotyledon-to-true leaf transition, which is a critical stage of plant development. The fact that the lack of chloroplast Trxs causes such a dramatic effect when combined with the lack of NTRC was further supported by the finding that mutant plants simultaneously devoid of NTRC and the catalytic subunit of FTR are inviable (Yoshida and Hisabori, 2016).

It has been reported that NTRC interacts with well-established Trx targets such as phosphoribulokinase (PRK) and FBPase (Nikkanen et al., 2016). Moreover, the analysis of the redox state of FBPase, a well-known redox-regulated enzyme of the CBC, showed decreased level of light-dependent reduction in the ntrc (Ojeda et al., 2017), and the trxf1f2 (Naranjo et al., 2016a) mutants. Intriguingly, the level of light-dependent reduction of FBPase in the trxx and the trxf1f2 mutants was very similar (Ojeda et al., 2017), despite the fact that Trx x is considered not involved in the redox regulation of enzymes of the CBC (Collin et al., 2003). The light-dependent redox reduction of FBPase is even more affected in the ntrc mutant, and essentially undetectable in mutant plants simultaneously lacking Trxs f or x and NTRC (Ojeda et al., 2017). Thus, these results support the notion that both Trxs (f and x) and NTRC concertedly participate in the redox regulation of FBPase. A possibility to explain the concerted action of the NTRC and the FTR-Trxs redox systems is that both have overlapping regulatory activity on the different redox regulated targets. Remarkably, in vitro assays with the purified proteins confirmed the high efficiency of Trxs f 1 and f 2, as compared with Trx x, in FBPase reduction, but revealed that NTRC is unable to reduce the enzyme, despite the fact that the light-dependent reduction of FBPase is severely affected in the ntrc mutant ( Ojeda et al., 2017). Therefore, these studies show that the activity of NTRC is needed for the function of different chloroplast Trxs; though for the redox regulation of FBPase the effect of NTRC seems to be exerted indirectly.

### THE CENTRAL ROLE of 2-Cys Prxs: INTEGRATING ANTIOXIDANT AND REDOX REGULATORY MECHANISMS

The puzzling question arising is how NTRC exerts such pleiotropic effects on different chloroplast redox regulated processes while, at least for the light-dependent redox regulation of FBPase, the effect is exerted without the direct interaction of both enzymes. The clarification of this conundrum came from the analysis of Arabidopsis lines combining NTRC and 2-Cys Prxs mutations. Although these plants are devoid of the redox regulatory function of NTRC, the phenotype of these mutants resembles that of the wild type (Pérez-Ruiz et al., 2017). This counterintuitive result indicated that the deficiency of 2-Cys Prxs exerts a suppressor effect of the ntrc mutant phenotype. Moreover, the overexpression of 2-Cys Prxs in the ntrc mutant background provoked the aggravation of the growth inhibition phenotype of these transgenic plants while the overexpression of the 2-Cys Prxs in the wild type background exerted no or low effect (Pérez-Ruiz et al., 2017). These results indicated that the increase of the content of 2-Cys Prxs becomes toxic for plant growth only when NTRC is not present. Although the absence of NTRC is expected to severely impair the antioxidant capacity of 2-Cys Prxs, it is important to take into account that, as stated above, different chloroplast Trxs are able to transfer reducing equivalents to 2-Cys Prxs, albeit much less efficiently than NTRC (Broin et al., 2002; Collin et al., 2003, 2004; Dangoor et al., 2012; Eliyahu et al., 2015; Hochmal et al., 2016; Vaseghi et al., 2018; Yoshida et al., 2018). Indeed, the light-dependent reduction of 2-Cys Prxs observed in the ntrc mutant (Pérez-Ruiz et al., 2017) suggests that these enzymes deplete reducing power from the pool of Trxs. Based on these results, it was hypothesized that in wild type plants the redox state of 2-Cys Prxs is maintained by NTRC, which is the most efficient reductant of the enzyme. Thus, the drainage of reducing equivalents from the other chloroplast Trxs would be very low and, consequently, the redox state of the pool of Trxs allows the light-dependent redox regulation of downstream targets (**Figure 3A**). In contrast, the ntrc mutant lacks the major source of reducing equivalents for 2-Cys Prxs, hence causing the accumulation of the oxidized form of these enzymes. Despite the lower efficiency of electron transfer of Trxs, the increased level of oxidized 2-Cys Prxs in plants lacking NTRC would provoke higher drainage of reducing equivalents from the pool of Trxs, compromising the light-dependent reduction of downstream targets (**Figure 3B**). Finally, in plants combining NTRC and 2-Cys Prx mutations, the decreased levels of 2-Cys Prxs would alleviate the drainage of reducing equivalents from the pool of Trxs, hence keeping sufficiently reduced the stromal Trxs for the light-dependent reduction of downstream targets, which would be thus the molecular basis of the suppression

accumulate in oxidized form, however, as the amount of 2-Cys Prxs is low, the drainage of reducing equivalents from the pool of Trxs is also low, and hence the

of the ntrc phenotype (**Figure 3C**). The analysis of the redox state of Trxs f, FBPase and PRK in the suppressed line fully confirmed this hypothesis hence uncovering the central function of 2-Cys Prxs in chloroplast redox regulation (Pérez-Ruiz et al., 2017). Moreover, the suppressed line also recovered wild type levels of chlorophyll biosynthesis enzymes (Richter et al., 2018), indicating that the suppressor effect caused by decreased levels of 2-Cys Prxs is exerted beyond the enzymes of the CBC. Based on these results, we proposed a model according to which the redox balance of 2-Cys Prxs, which is maintained by NTRC and NADPH, modulates the light-dependent reduction of redox regulated enzymes based on Fdx reduced by the photosynthetic electron transport chain, which is fueled to stromal Trxs via the action of FTR (Pérez-Ruiz et al., 2017). This model integrates the redox exchange of Trx and redoxregulated targets with hydrogen peroxide via the action of 2- Cys Prxs.

redox state of the pool of Trxs is appropriate for the reduction of downstream targets.

As mentioned above, the simultaneous lack of NTRC and Trx x or Trxs f cause dramatic loss of photosynthetic efficiency, light-dependent redox regulation of CBC enzymes and, consequently, severe growth inhibition phenotypes (Ojeda et al., 2017). Thus, to test the robustness of the function of the couple NTRC-2-Cys Prxs in chloroplast redox regulation, mutants plants deficient in NTRC, and Trxs x or f were combined with decreased contents of 2-Cys Prxs (Pérez-Ruiz et al., 2017; Ojeda et al., 2018b). Decreased contents of 2- Cys Prxs recovered growth phenotypes in plants lacking NTRC and Trx x (Ojeda et al., 2018b) or NTRC and Trxs f (Pérez-Ruiz et al., 2017), thus extending the suppressor effect of the deficiency of 2-Cys Prxs to the dramatic growth inhibition phenotypes caused by the simultaneous lack of NTRC and Trxs. Altogether, these results confirm the essential function of the NTRC-2-Cys Prxs redox couple in chloroplast performance and plant growth.

### 2-Cys Prxs PARTICIPATE IN CHLOROPLAST ENZYME OXIDATION IN THE DARK

The redox regulation of chloroplast enzymes in response to light was discovered in the sixties of the past century; since then, most studies have focused on the light-dependent enzyme activation by reduction, achieving a comprehensive knowledge of this regulatory mechanism (Buchanan, 2016). At the same time, it became clear that redox regulated enzymes of the CBC such as FBPase (Leegood and Walker, 1980) and glyceraldehyde phosphate dehydrogenase (GAPDH) (Sparla et al., 2002) were rapidly oxidized in the dark; however, the mechanism of enzyme oxidation has remained unknown. Our model of chloroplast redox regulation proposes a central role of 2-Cys Prxs in maintaining the redox state of the pool of Trxs that participate in the light-dependent redox regulation of chloroplast metabolic pathways including enzymes of the CBC (Pérez-Ruiz et al., 2017; Ojeda et al., 2018b) and the chlorophyll biosynthesis pathway (Richter et al., 2018). An important consequence of this model is that hydrogen peroxide acts as sink of reducing equivalents and, thus, could be an effective way of relieving reducing equivalents of the pool of Trxs in the dark, hence providing a possible explanation for the long-standing question of how chloroplast enzymes become oxidized in the night. Our group addressed this possibility by analyzing the redox state of well-studied redoxregulated enzymes, FBPase, GAPDH and the γ subunit of

ATPase in the Arabidopsis 2cpab mutant, which is knock out for 2-Cys Prxs A and B (Ojeda et al., 2018a). The enzymes under analysis were detected almost fully reduced in both wild type and 2cpab mutant plants adapted to light. While in the wild type these enzymes were rapidly oxidized in the dark, oxidation was significantly delayed in the 2cpab mutant, indicating the participation of 2-Cys Prxs in the shortterm oxidation of chloroplast enzymes in the dark (Ojeda et al., 2018a). Independently, Vaseghi et al. (2018) reported the participation of 2-Cys Prxs in the oxidation of reductively activated chloroplast enzymes FBPase, PRK, and NADPHdependent malate dehydrogenase (MDH) and showed that oxidation was compromised in Arabidopsis mutant plants devoid of 2-Cys Prxs. The analysis of the oxidative activity of the Trxlike2 (Trx-L2) protein in vitro, in conjunction with the use of a severe 2-Cys Prxs knock down mutant of Arabidopsis led Hisabori's group to propose the participation of an oxidant pathway formed by Trx-L2/2-Cys Prxs in enzyme oxidation in the dark (Yoshida et al., 2018). The identification of the function of 2-Cys Prxs in the oxidation of thiol groups is in line with the previous proposal of the oxidizing activity of this enzyme in response to moderate light intensity (Dangoor et al., 2012) and for oxidation of the small subunit of AGPase (Eliyahu et al., 2015).

Therefore, after decades of research focused on the mechanisms leading to enzyme activation in the light, three

FIGURE 4 | The NTRC-2-Cys Prx couple controls the reduction/oxidation balance of redox-regulated targets in the day and in the night. During the day, photo-reduced Fdx fuels reducing equivalents via FTR and Trxs for the reduction (activation) of redox-regulated enzymes. The redox state of the pool of Trxs is maintained by the couple NTRC-2-Cys Prxs, which relies on NADPH as source of reducing power. In the dark, the input of reducing equivalents via reduced Fdx ceases and Trxs mediate the oxidation of reduced targets transferring electrons through the activity of 2-Cys Prxs to hydrogen peroxide, which acts as final sink of electrons. The oxidant efficiency of different Trxs is presented as proposed by Yoshida et al. (2018) and Vaseghi et al. (2018).

independent studies (Ojeda et al., 2018a; Vaseghi et al., 2018; Yoshida et al., 2018) identified the participation of 2-Cys Prxs in enzyme oxidation in the dark. The characterization of the 2-Cys Prx interactome revealed the interaction of 2-Cys Prxs with a large number of chloroplast proteins including FBPase (Cerveau et al., 2016). Therefore, a possibility to be taken into account is that FBPase oxidation occurs by the direct transfer of electrons from the enzyme to 2-Cys Prxs. However, studies performed in vitro with recombinant enzymes (Ojeda et al., 2018a; Yoshida et al., 2018) or with extracts from chloroplast stroma (Vaseghi et al., 2018) showed that reduced FBPase was not directly oxidized in the presence of 2-Cys Prx or Trxs. In contrast, the addition of 2-Cys Prxs and Trxs resulted in FBPase oxidation (Ojeda et al., 2018a), indicating the participation of these Trxs in the process of oxidation. This notion was further supported by Vaseghi et al. (2018), who determined the MDH oxidation efficiency by different Trxs, including canonical m- fand x-types and Trx-like proteins such as chloroplast droughtinduced stress protein of 32 kDa (CDSP32). Finally, it was shown that Trx-L2, which has the less negative reducing potential and is unable to reduce FBPase, is the most efficient chloroplast Trx for transfer of reducing equivalents of reduced enzymes to 2-Cys Prxs and H2O<sup>2</sup> (Yoshida et al., 2018). Altogether, these studies proposed a pathway of oxidation involving Trxs, which would act as intermediates between the reduced targets and 2-Cys Prxs (**Figure 4**). This finding is an important advance in the current understanding of the control of chloroplast photosynthetic metabolism by light and darkness. However, it should be noted that, albeit delayed, enzyme oxidation still takes place in mutant plants lacking 2-Cys Prxs, indicating the participation of additional mechanism(s) in the process of enzyme oxidation in the dark. In search of these additional mechanisms, our group tested the participation of Prx Q and Prx IIE, monomeric Prxs present in Arabidopsis chloroplasts (Dietz, 2011). However, Arabidopsis triple mutants combining the lack of 2-Cys Prxs with severely decreased levels of either Prx Q or Prx IIE showed similar growth phenotype and rates of enzyme oxidation in the dark than the 2cpab mutant, suggesting that these Prxs have no relevant function in enzyme oxidation (Ojeda et al., 2018a).

# CONCLUDING REMARKS AND FUTURE PROSPECT

While redox regulation in heterotrophic organisms involves a low number of Trxs and NTRs, and depends on NADPH, redox regulation in chloroplasts involves two redox systems: NTRC, which relies on NADPH, and the FTR-Trxs system, in which a large set of Trxs relies on photoreduced Fdx. Therefore, redox regulation in chloroplasts is much more complex than in heterotrophic organisms, probably reflecting the sessile life style of plants and their necessity to rapidly respond to natural fluctuations in light. A recent report has identified the central role of 2-Cys Prxs integrating disulfide-dithiol exchange of redox-regulated enzymes and hydrogen peroxide detoxification

(Pérez-Ruiz et al., 2017). Moreover, the participation of 2-Cys Prxs in the short-term oxidation of chloroplast enzymes in the dark has been recently reported (Ojeda et al., 2018a; Vaseghi et al., 2018; Yoshida et al., 2018), thus starting to solve the long-lasting question of the mechanism that allows the rapid oxidation of chloroplast enzymes in the night (Jacquot, 2018). Nevertheless, the participation of additional mechanism(s), yet to be identified, as well as the role of chloroplast Trxs in the oxidation process need to be clarified. Finally, the function of NTRC could be exclusively the maintenance of the redox balance of 2-Cys Prxs for proper chloroplast function, however whether this enzyme hase additional regulatory functions deserves further attention in next few years.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

FJC designed and wrote the initial version of the manuscript. VO, VD-R, MG, and JMP-R contributed to the final version.

# FUNDING

Work in our laboratory was supported by European Regional Development Fund-co-financed grant (BIO2017-85195-C2- 1-P) from the Spanish Ministry of Economy, Industry, and Competiveness (MINECO) and BIO-182 from Junta de Andalucía, Spain.

system in plastid redox regulation and signalling. FEBS Lett. 586, 2974–2980. doi: 10.1016/j.febslet.2012.07.003



their loss affects growth under short-day conditions in Arabidopsis thaliana. J. Exp. Bot. 67, 1951–1964. doi: 10.1093/jxb/erw017



in vivo function of reductase and thioredoxin domains. Front. Plant Sci. 4:389. doi: 10.3389/fpls.2013.00389


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

Copyright © 2019 Cejudo, Ojeda, Delgado-Requerey, González and Pérez-Ruiz. 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.

# Endoplasmic Reticulum Plays a Critical Role in Integrating Signals Generated by Both Biotic and Abiotic Stress in Plants

*Chang-Jin Park1,2 \* and Jeong Mee Park3,4 \**

*1 Department of Bioresources Engineering, Sejong University, Seoul, South Korea, 2 Plant Engineering Research Institute, Sejong University, Seoul, South Korea, 3 Plant Systems Engineering Research Center, Korea Research Institute of Bioscience and Biotechnology (KRIBB), Daejeon, South Korea, 4 Department of Biosystems and Bioengineering, University of Science and Technology (UST), Daejeon, South Korea*

#### *Edited by:*

*Sang Yeol Lee, Gyeongsang National University, South Korea*

#### *Reviewed by:*

*Kyun Oh Lee, Gyeongsang National University, South Korea Pradeep Kachroo, University of Kentucky, United States*

#### *\*Correspondence:*

*Chang-Jin Park cjpark@sejong.ac.kr Jeong Mee Park jmpark@kribb.re.kr*

#### *Specialty section:*

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

*Received: 21 December 2018 Accepted: 15 March 2019 Published: 04 April 2019*

#### *Citation:*

*Park C-J and Park JM (2019) Endoplasmic Reticulum Plays a Critical Role in Integrating Signals Generated by Both Biotic and Abiotic Stress in Plants. Front. Plant Sci. 10:399. doi: 10.3389/fpls.2019.00399*

Most studies of environmental adaptations in plants have focused on either biotic or abiotic stress factors in an attempt to understand the defense mechanisms of plants against individual stresses. However, in the natural ecosystem, plants are simultaneously exposed to multiple stresses. Stress-tolerant crops developed in translational studies based on a single stress often fail to exhibit the expected traits in the field. To adapt to abiotic stress, recent studies have identified the need for interactions of plants with various microorganisms. These findings highlight the need to understand the multifaceted interactions of plants with biotic and abiotic stress factors. The endoplasmic reticulum (ER) is an organelle that links various stress responses. To gain insight into the molecular integration of biotic and abiotic stress responses in the ER, we focused on the interactions of plants with RNA viruses. This interaction points toward the relevance of ER in viral pathogenicity as well as plant responses. In this mini review, we explore the molecular crosstalk between biotic and abiotic stress signaling through the ER by elaborating ER-mediated signaling in response to RNA viruses and abiotic stresses. Additionally, we summarize the results of a recent study on phytohormones that induce ER-mediated stress response. These studies will facilitate the development of multi-stress-tolerant transgenic crops in the future.

Keywords: ABA, ER stress, heat stress, osmotic stress, RNA virus, SA

# INTRODUCTION

Being sessile organisms, plants rely on their interactions with various organisms to adapt to environmental changes. Most terrestrial plants establish symbiotic relationships with microorganisms such as fungi and bacteria (Parniske, 2008; Finkel et al., 2017). Interactions of plants with these microorganisms play an important role not only in protecting plants against various pathogens but also in defense against abiotic stresses (Lau and Lennon, 2012). Plant molecular biologists have been studying plant responses to individual environmental stresses for a long time, and attempts have been made to apply the results of these studies to agriculture. However, in the field, plants are exposed to a variety of stresses simultaneously, and the responses of plants to these stresses are often different from those predicted in the laboratory (Pandey et al., 2015). Rapid ongoing global climate change is increasing the need to study plant responses to simultaneous stresses.

The endoplasmic reticulum (ER) is one of the largest, most functionally complex, and architecturally variable organelles discovered in eukaryotic cells (Schuldiner and Schwappach, 2013). It is a highly dynamic and complex cytoplasmic membrane system composed of two structurally distinct subdomains: the nuclear envelope enclosing the nucleus and an interconnected network called the peripheral ER, comprising a series of flattened sacs and tubules (Voeltz et al., 2002; Westrate et al., 2015). The ER is a central organelle that regulates stress responses in both plant and animal cells (Ellgaard and Helenius, 2003; Schroder and Kaufman, 2005). Stresses affecting protein folding lead to ER stress, which is communicated to the nucleus *via* the unfolded protein response (UPR), a cellular homeostatic response to ER stress (Fontes et al., 1991; Ellgaard and Helenius, 2003). Although the molecular mechanism of ER stress in plants is not as well understood as in animals, the expansion and diversity of ER stress-related genes revealed by genome sequencing of various plant species suggests that plants use more ER stress responses to adapt to the environment than animals (Liu and Howell, 2010b; Howell, 2017).

In plants, two main types of ER stress sensors, which regulate different UPR signaling pathways, have been identified: ER membrane-associated basic leucine zipper (bZIP) transcription factors, bZIP28 (Srivastava et al., 2014) and bZIP17 (Liu et al., 2008), and an ER resident transmembrane protein, inositol-requiring enzyme 1 (IRE1) (Koizumi et al., 2001) (**Figure 1**). Under unstressed normal conditions, bZIP17/28 are retained in the ER by their association with the binding protein (BiP), which is a master regulator of UPR. Under stress conditions, when unfolded proteins accumulate, BiP is sequestered away and released from bZIP17/28 (Gao et al., 2008; Srivastava et al., 2014). bZIP transcription factors are then transported from the ER to the Golgi apparatus, where they are proteolytically cleaved. The cytosol-facing regions of bZIP17/28 are then transported from the Golgi apparatus to the nucleus, to upregulate the expression of stress response genes, and to restore ER homeostasis (Liu et al., 2007a; Liu and Howell, 2010a). The IRE1 harbors both kinase and ribonuclease domains (Koizumi et al., 2001). In Arabidopsis (*Arabidopsis thaliana*), activated IRE1 splices bZIP60 mRNA, resulting in a frame shift and yielding a bZIP60 variant targeted to the nucleus (Deng et al., 2011; Hayashi et al., 2012, 2016; Liu and Howell, 2016). In the nucleus, bZIP60 plays a role in expression of ER stress response-related genes.

In this mini review, we summarize ER stress responses in plants to RNA viruses, abiotic stresses, and stress-related hormones to understand plant cell perception of simultaneous biotic and abiotic stresses and the responses to these stresses through the ER.

# EXPLOITATION OF ER MEMBRANES BY RNA VIRUSES

The majority of plant viruses that contains a positive-sense RNA genome and amplifies RNA in specific membrane-associated regions of host organelles are called replication complexes (RCs). Viral RCs play an important role in the enrichment of host cellular components for viral replication and prevent the activation of specific host defense mechanisms triggered by double-stranded RNA (dsRNA) intermediates of viral replication (Heinlein, 2015; Romero-Brey and Bartenschlager, 2016). Over the past decade, visualization of viral RC formation using advanced cell fluorescence imaging and microscopy techniques has enhanced our understanding of the process of virus replication in plant cells and the host factors involved (Jin et al., 2018). Although cell membrane origin and replication components and processes vary among plant RNA viruses, the RCs associated with plant RNA viruses can be divided into two types: double membrane vesicles and spherules/ invaginations, as in animal viruses. This suggests that the process of viral infection is evolutionarily conserved among plant and animal viruses (Jin et al., 2018).

The ER is connected to the membrane of most other cellular organelles and the plasma membrane through membrane contact sites (MCSs) (Pérez-Sancho et al., 2016). Recent research suggests that inter-organelle communications *via* ER-driven MCSs helps communicate stress signals faster and more accurately than via vesicles or molecular diffusion (Pérez-Sancho et al., 2016). In plants, various membrane-bound organelles have been targeted as viral RC sites by one or more viral species, although the significance of interactions between these viruses and host cell organelles remains unclear (Grangeon et al., 2012). Tomato bushy stunt virus (TBSV) replicates on peroxisomes but is capable of forming viral RCs on the ER in the absence of peroxisomes, suggesting that a particular organelle is not a limiting factor for plant RNA virus infection (Jonczyk et al., 2007; Chuang et al., 2014). However, a particular type of lipid has been shown to affect viral RC formation and function. For example, the phosphatidylcholine (PC) content is high at sites of viral RCs of Brome mosaic virus (BMV), and blockade of PC synthesis inhibits BMV replication both in the native host barley and in the alternative host yeast, indicating that PC is important for BMV replication (Zhang et al., 2016). In addition, experiments using plants, yeast, and artificial phospholipid vesicles revealed that phosphatidylethanolamine is required at the site of TBSV replication (Xu and Nagy, 2015). Further research is needed to understand the effect of virus-specific lipid requirement on the host immune response, lipid metabolism in ER, and inter-organelle communication.

# ER STRESS SIGNALING INDUCED BY VIRUS INFECTION

Overexpression of viral proteins and changes in the lipid membrane structure during virus infection induce ER stress

variant (bZIP60-S) targeted to the nucleus. When BiP is sequestered away, bZIP17/28 transcription factors are released from the ER to the Golgi apparatus, where they are cleaved. The cytosol-facing regions of bZIP17/28 are transported from the Golgi apparatus to the nucleus. After translocation, bZIP transcription factors upregulate the expression of stress response genes. Expression of bZIP60 is increased by viral infection. SA induces IRE1-mediated splicing of bZIP60 mRNA and proteolytic processing of bZIP28. SA regulates the expression of stress response genes via an NPR1-depenedent pathway. ABA promotes proteolytic processing of bZIP17. Dashed arrows indicate yet-uncharacterized molecular pathways.

and UPR signaling by perturbing protein homeostasis in both plant and animal hosts (Rutkowski and Kaufman, 2004; Zhang and Wang, 2012). The importance of ER stress signaling for infection of animal viruses has been highlighted in recent years (Li et al., 2015). In particular, much work has focused on mammalian viral proteins that interact directly with components of the UPR signaling pathway to inhibit or induce this response, thereby improving viral replication or transmission in mammalian cells (Blázquez et al., 2014). More recently, increasing evidence indicates that perturbation of the lipid bilayer composition also activates UPR signaling through IRE1 and PERK independently of its effect on protein folding homeostasis in the ER lumen (Volmer et al., 2013; Halbleib et al., 2017). IRE1 and PERK mutants lacking the luminal domain, which is required to sense unfolded protein stress, are activated by deprivation of inositol in yeast and by treatment with saturated fatty acids in mammalian cells (Volmer et al., 2013; Halbleib et al., 2017). Although the molecular mechanism by which UPR sensors recognize lipid bilayer stress has not been elucidated, the hepatitis C and West Nile viruses, which belong to the family Flaviviridae, induce lipid-dependent UPR signaling through nonstructural viral proteins to facilitate viral replication and to avoid host immune surveillance (Leier et al., 2018).

The molecular components involved in ER stress and UPR signaling were identified later in plants than in animals. Consequently, associations between plant RNA virus infections and ER stress responses were mostly investigated by measuring the expression levels of ER stress-related genes such as *BiPs* and protein disulfide isomerase and calreticulin (Zhang and Wang, 2012). Overexpression of several plant viral proteins such as Turnip mosaic virus (TuMV) 6K2, Potato virus X (PVX) triple gene block protein 3 (TGBp3), Rice black-streak dwarf virus P10, and Garlic virus X p11 induces the expression of ER stress response-related genes in plants (Heinlein, 2015; Lu et al., 2016; Howell, 2017). These proteins also bind to the ER membrane and enhance expression of the *bZIP60* and *BiP* genes in tobacco (*Nicotiana benthamiana*) and Arabidopsis (Ye et al., 2011, 2012, 2013). The Arabidopsis *bZIP60* knockout mutants show a low viral titer when infected with PVX and TuMV (Zhang et al., 2015). These viruses associate with different host subcellular organelles; TuMV forms viral RCs with the peripheral membrane of peroxisomes or chloroplasts, while PVX associates with the ER. These results suggest that activation of ER stress by viral proteins through bZIP60 benefits pathogenesis of plant viruses regardless of their RC-associated subcellular organelle. In addition, eukaryotic translation elongation factor 1 alpha (eEF1A) has been known to interact with viral RNA-dependent RNA polymerase of potyviruses, tobamoviruses, and tombusviruses, which is essential for viral replication (Nishikiori et al., 2006; Yamaji et al., 2006; Thivierge et al., 2008; Li et al., 2010). A recent report demonstrated that the UPR is activated in soybean during Soybean mosaic virus (SMV) infection and that this promotes accumulation of the virus (Luan et al., 2016). The authors showed that eEF1A, which interacts with the P3 protein of SMV, plays an important role in UPR activation and viral replication, although the molecular mechanism by which eEF1A activates the UPR is unclear (Luan et al., 2016). The P3 protein is known to reside in ER as a virulence factor of SMV, but its exact biochemical function is unknown (Chowda-Reddy et al., 2011). IRE1 activates splicing of bZIP60 in plants; therefore, it would be interesting to determine whether plant viral proteins targeting ER, similar to animal viral proteins, disturb ER lipid homeostasis and utilize the ER stress response through IRE1–bZIP60 signaling to support viral RCs.

Although the mutual regulatory mechanisms of UPR and programmed cell death (PCD) are well known in animals, it is not yet clear how plant UPR is linked to pathogen-induced cell death (Shore et al., 2011). In mammalian cells, a protein kinase R (PKR)-like ER kinase (PERK) induces PCD by phosphorylating the α-subunit of eukaryotic initiation factor 2 (eIF-2α), which prevents general protein synthesis and ultimately induces apoptosis (Liu et al., 2015b). However, few studies have investigated the induction of cell death by ER stress in plants. The existence of PERK, an ER stress sensor, in plants remains controversial, and the relevance of PERK-mediated phosphorylation of eIF-2α to the general plant defense response, including that against virus infection, has not yet been clarified (Zhang and Wang, 2012). A genome-wide search for eIF-2α kinases suggests that Arabidopsis and rice (*Oryza sativa*) plants lack a mammalian PERK homolog and possess only a homolog of the yeast general control non-derepressible-2 (GCN2) (Zhang et al., 2008). Indeed, phosphorylation of eIF-2α by AtGCN2 in Arabidopsis was observed under stress conditions and unlike in animals, it was not observed during viral infection (Zhang et al., 2008). Silencing of the gene encoding p58IPK, a putative plant ortholog of the mammalian PKR inhibitor, causes cell death following viral infection in tobacco (Bilgin et al., 2003). In animals, p58IPK acts as an inhibitor of the antiviral protein PKR, which induces apoptosis by phosphorylating eIF-2α upon recognition of viral dsRNA (Tan et al., 1998). Bilgin et al. (2003) demonstrated a direct interaction between tobacco p58IPK and helicases of TMV and Tobacco etch virus and an increase in eIF-2α phosphorylation in p58IPK-silenced plants. These results suggest the presence of a PKR- or PERK-like kinase-mediated pathway for eIF-2α phosphorylation, although so far no PKRor PERK-like kinases have been found in plants using sequence similarity searches with mammalian counterparts.

#### ER STRESS SIGNALING IN RESPONSE TO ABIOTIC STRESS

#### Heat Stress

Over the past several decades, extensive studies have elucidated several molecular mechanisms of plant responses to high temperature, mainly focusing on flower development, circadian clock modulation, and immune response (Liu et al., 2015a). Although great progress has been achieved in the identification of molecular mechanisms of plant thermotolerance, how plants sense and transduce the heat signal remains unknown. Recently, it was reported that phytochromes, which sense the ratio of red to far-red light, function as thermosensors. Indeed, phytochrome-null plants display a constitutive warm temperature response (Jung et al., 2016). In addition, elongated hypocotyl 5 (HY5), a bZIP transcription factor and a downstream component of phytochrome-mediated light signaling, negatively regulates UPR by competing with bZIP28, which upregulates the expression of stress response genes (Nawkar et al., 2017). These reports shed light on the molecular mechanism of crosstalk between UPR and thermal sensing, mediated by HY5, which positively mediates light signaling but negatively regulates UPR gene expression.

The most detrimental effect of heat shock is the accumulation of unfolded proteins in both the cytosol and ER. Therefore, unfolded protein sensors in the ER and cytosol are proposed to play an essential role in thermotolerance. One of the UPR components, bZIP28, contributes to the upregulation of heat responsive genes, leading to heat tolerance of Arabidopsis (Gao et al., 2008). Knockout mutants of bZIP28 are sensitive to high temperature, suggesting an essential role of UPR in the general heat stress response and thermotolerance (Gao et al., 2008). Chromatin immunoprecipitation coupled with high-throughput sequencing revealed 133 putative direct targets of bZIP28 in Arabidopsis seedlings subjected to heat stress (Zhang S. S. et al., 2017). Another UPR component, IRE1, is also reported to be involved in heat stress response in Arabidopsis. Heat activated IRE1 splices bZIP60 mRNA, which is required for the upregulation of BiP3 in response to ER stress (Deng et al., 2011). Furthermore, IRE1 regulates the stress transcriptome by degrading various mRNAs (Mishiba et al., 2013; Maurel et al., 2014). Therefore, phytochromes are major regulators of heat stress response and thermotolerance in plants through the adjustment of downstream molecular components and different types of ER stress sensors.

#### Osmotic Stress

Osmotic stress caused by drought and high salinity has a major impact on plant growth and crop production, which highlights the importance of osmotic stress tolerance in plants. Drought and high salinity elicit many common and interactive downstream effects such as high levels of abscisic acid (ABA) (Nambara and Marion-Poll, 2005) and stressresponsive gene expression (Zhu, 2002). In Arabidopsis, the salt stress signaling pathway is reported to resemble an ER stress response (Liu et al., 2007b). Salt treatment activates the Golgi apparatus-resident site-1 protease (S1P) and cleaves ER membrane-associated transcription factor bZIP17. The released cytosol-facing region of bZIP17 is translocated to the nucleus, where it activates the expression of salt stress response genes. In this pathway, the activated bZIP17 transcription factor upregulates the expression of downstream genes, which desensitize ABA signaling (Zhou et al., 2015). Another major UPR component, BiP, plays an important role in osmotic stress tolerance in soybean (*Glycine max*) and tobacco plants *via* an unknown mechanism (Alvim et al., 2001; Valente et al., 2009). Expression profile analyses of soybean plants treated with ER stress inducers (tunicamycin/ azidothymidine) or an osmotic stress inducer (polyethylene glycol) suggest a link between ER stress and osmotic stress pathway (Irsigler et al., 2007). In wheat (*Triticum aestivum*), the expression of BiP is upregulated during osmotic stressrelated cell death caused by tauroursodeoxycholic acid, an apoptosis inhibitor (Zhang L. et al., 2017).

In addition to the ER-resident proteins, NAC (NAM/ATAF1/2/ CUC2) domain-containing transcription factors are also receiving attention in osmotic stress response. Rice (151), soybean (152), and Arabidopsis (117) harbor a large family of NAC domaincontaining proteins that are involved in multiple stress responses (Le et al., 2011; Nakashima et al., 2012; Shao et al., 2015). In soybean, GmNAC81 (also known as GmNAC6) has been identified as a component of ER stress- and osmotic stressinduced cell death response (Costa et al., 2008; Faria et al., 2011); this cell death response is synergistically activated by GmNAC30 and GmNAC81 (Mendes et al., 2013).

# ER STRESS INDUCED BY STRESS-RELATED HORMONES

#### Salicylic Acid (SA)

SA is a key signaling component of both local defense response at infection sites and systemic resistance (Vernooij et al., 1994; Klessig et al., 2000). Functional crosstalk between SA and ER stress was first observed in Arabidopsis (Wang et al., 2005). Treatment of Arabidopsis with SA alters the level of many ER proteins required for protein folding and secretion, including BiP2 (Wang et al., 2005) and BiP3 (Pajerowska-Mukhtar et al., 2012). SA-induced expression of some of the ER stress-related genes is regulated by the heat shock factor-like transcription factor, *TL1*-binding transcription factor 1 (TBF1), which is genetically dependent on the non-expressor of pathogenesisrelated genes 1 (NPR1), a master regulator of SA signaling (Wang et al., 2005; Pajerowska-Mukhtar et al., 2012). Because TBF1 regulates only selected ER genes, it was initially suggested that plants have evolved a specific mechanism for regulating SA-induced expression of ER stress-related genes (Pajerowska-Mukhtar et al., 2012). It was later shown that SA induces these genes *via* two main UPR signaling pathways: proteolytic processing of bZIP28 and IRE1-mediated splicing of bZIP60 mRNA (Nagashima et al., 2014). However, because it is unlikely that SA directly inhibits protein folding in the ER, the mechanism of activation of the UPR pathway by SA remains unclear. In Arabidopsis, constitutive expresser of pathogenesis-related genes 5 (CPR5), a plant-specific growth and stress regulator, acts as a negative modulator of SA in the early signal transduction steps, downstream of pathogen recognition and upstream of SA (Bowling et al., 1997). Recently, CPR5 was reported to act as a negative regulator of bZIP28 and bZIP60 through protein– protein interactions (Meng et al., 2017). Consequently, CPR5 suppresses the function of ER stress-induced bZIP28 and IRE1 bZIP60 pathways in the homeostatic control of SA-mediated plant growth (Meng et al., 2017).

#### ABA

ABA regulates many aspects of plant growth and development and plays a central role in the response to heat and osmotic stress (Shinozaki and Yamaguchi-Shinozaki, 2000; Raghavendra et al., 2010; Hauser et al., 2011; Huang et al., 2016). For example, in Arabidopsis, high temperature treatment upregulates ABA biosynthesis-related genes and downregulates ABA degradation-related genes (Toh et al., 2008). Similarly, in rice, heat treatment increases the level of ABA (Wu et al., 2016). Despite the increasing number of reports on ABA and heat stress, only a few studies have been conducted to investigate the effect of ABA on ER stress. Arabidopsis plants overexpressing the maize ortholog of Arabidopsis bZIP17 (ZmbZIP17) upregulates the expression of ER stress response genes. ER stress inducers such as dithiothreitol and tunicamycin induce ZmbZIP17 and its translocation to the nucleus (Yang et al., 2013). In addition, ZmbZIP17 interacts with ABA-responsive cis-elements in the promoter of ABA-responsive genes in yeast. Considering these reports, it is very likely that bZIP17 is involved in ABA-mediated ER stress response.

The mechanism of activation of bZIP17, which triggers ABA signaling in response to ER stress, has been elucidated in a study on seed germination in Arabidopsis; the authors showed that the Golgi apparatus-resident site-2 protease (S2P) cleaves and activates bZIP17, thus regulating downstream target genes that encode negative regulators of ABA signaling (Zhou et al., 2015). The level of ABA is also elevated by salt stress, which induces a signaling cascade involving the processing of bZIP17 by S1P, translocation of bZIP17 to the nucleus, and upregulation of salt stress genes (Nambara and Marion-Poll, 2005; Liu et al., 2007b). However, whether bZIP17-mediated ABA response is directly linked to the ER stress response remains unclear. It is possible that salt or ABA treatments promote protein misfolding, and S1P/S2P-mediated bZIP17 processing plays an essential role in desensitizing the plant to ER stress.

#### CONCLUSION AND PERSPECTIVES

Changes in the agricultural environment, such as changes in temperature and water availability caused by climate change, can cause enormous reductions in crop yields *via* increased biological stress (Pandey et al., 2017). Global warming, for example, increases temperature stress in plants while at the same time increasing insect populations, which results in the spread of insect-borne viruses and their expansion to new host areas (Bebber, 2015). To help crops withstand newly emerging stresses, it is important to understand the mechanisms that plants have evolved to counteract various stress factors. Here, we have summarized recent advances in our understanding of ER responses to RNA viruses, abiotic stresses, and hormone responses (**Figure 1**). Of the three known ER stress sensors, IRE1 and bZIP28 are both involved in ER stress responses to viral infection and abiotic stress, whereas bZIP17 appears to be only an abiotic stress-specific sensor. Prasch and Sonnewald (2013) have shown that Arabidopsis plants exposed simultaneously to multiple stresses, such as heat, drought, and TuMV infection, have different responses to those of plants exposed to only one stress. For instance, defense-related genes induced by viral infection were not observed in Arabidopsis

#### REFERENCES


plants exposed to three stresses, but the ER stress response, which was not observed during TuMV infection, was induced (Prasch and Sonnewald, 2013). These results suggest that the mechanisms by which plants adapt to both external and internal stresses must be elucidated to understand the molecular adaptation of plants to multiple external stresses.

#### AUTHOR CONTRIBUTIONS

JMP designed the outline of the article and C-JP and JMP wrote the article.

#### FUNDING

This work was supported by the KRIBB Initiative Program and the Basic Research Program of the National Research Foundation of Korea (NRF-2017R1D1A1B03032215 to C-JP and NRF-2017R1A2B4012820 to JMP) funded by the Ministry of Science and ICT.

#### ACKNOWLEDGMENTS

We acknowledge Bioedit Ltd for critically reviewing the paper and for language assistance.

factor involved in the unfolded protein response in Arabidopsis. *Proc. Natl. Acad. Sci. U. S. A.* 108, 7247–7252. doi: 10.1073/pnas.1102117108


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

*Copyright © 2019 Park and Park. 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.*

# Early Brassica Crops Responses to Salinity Stress: A Comparative Analysis Between Chinese Cabbage, White Cabbage, and Kale

Iva Pavlovic´ 1,2† , Selma Mlinaric´ 3† , Danuše Tarkowská<sup>2</sup> , Jana Oklestkova<sup>2</sup> , Ondrej Novák ˇ 2 , Hrvoje Lepeduš4,5, Valerija Vujci ˇ c Bok ´ 6 , Sandra Radic Brkanac ´ 6 , Miroslav Strnad<sup>2</sup> and Branka Salopek-Sondi<sup>1</sup> \*

#### Edited by:

Sang Yeol Lee, Gyeongsang National University, South Korea

#### Reviewed by:

Honghong Wu, University of California, Riverside, United States María Serrano, Universidad Miguel Hernández de Elche, Spain

#### \*Correspondence:

Branka Salopek-Sondi salopek@irb.hr

†These authors have contributed equally to this work as first authors

#### Specialty section:

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

Received: 23 December 2018 Accepted: 25 March 2019 Published: 11 April 2019

#### Citation:

Pavlovic I, Mlinari ´ c S, ´ Tarkowská D, Oklestkova J, Novák O, Lepeduš H, Vujci ˇ c Bok V, ´ Radic Brkanac S, Strnad M and ´ Salopek-Sondi B (2019) Early Brassica Crops Responses to Salinity Stress: A Comparative Analysis Between Chinese Cabbage, White Cabbage, and Kale. Front. Plant Sci. 10:450. doi: 10.3389/fpls.2019.00450 <sup>1</sup> Department of Molecular Biology, Ruder Boškovi ¯ c Institute, Zagreb, Croatia, ´ <sup>2</sup> Laboratory of Growth Regulators, Institute of Experimental Botany, The Czech Academy of Sciences, Palacký University, Olomouc, Czechia, <sup>3</sup> Department of Biology, Josip Juraj Strossmayer University of Osijek, Osijek, Croatia, <sup>4</sup> Faculty of Humanities and Social Sciences, Josip Juraj Strossmayer University of Osijek, Osijek, Croatia, <sup>5</sup> Faculty of Dental Medicine and Health, Josip Juraj Strossmayer University of Osijek, Osijek, Croatia, <sup>6</sup> Division of Botany, Department of Biology, Faculty of Science, University of Zagreb, Zagreb, Croatia

Soil salinity is severely affecting crop productivity in many countries, particularly in the Mediterranean area. To evaluate early plant responses to increased salinity and characterize tolerance markers, three important Brassica crops – Chinese cabbage (Brassica rapa ssp. pekinensis), white cabbage (B. oleracea var. capitata) and kale (B. oleracea var. acephala) were subjected to short-term (24 h) salt stress by exposing them to NaCl at concentrations of 50, 100, or 200 mM. Physiological (root growth, photosynthetic performance parameters, and Na+/K<sup>+</sup> ratio) and biochemical parameters (proline content and lipid peroxidation as indicated by malondialdehyde, MDA, levels) in the plants' roots and leaves were then measured. Photosynthetic parameters such as the total performance index PItotal (describing the overall efficiency of PSI, PSII and the intersystem electron transport chain) appeared to be the most salinity-sensitive parameter and informative stress marker. This parameter was decreased more strongly in Chinese cabbage than in white cabbage and kale. It indicated that salinity reduced the capacity of the photosynthetic system for efficient energy conversion, particularly in Chinese cabbage. In parallel with the photosynthetic impairments, the Na+/K<sup>+</sup> ratio was highest in Chinese cabbage leaves and lowest in kale leaves while kale root is able to keep high Na+/K<sup>+</sup> ratio without a significant increase in MDA. Thus Na+/K<sup>+</sup> ratio, high in root and low in leaves accompanying with low MDA level is an informative marker of salinity tolerance. The crops' tolerance was positively correlated with levels of the stress hormone abscisic acid (ABA) and negatively correlated with levels of jasmonic acid (JA), and jasmonoyl-L-isoleucine (JA-Ile). Furthermore, salinity induced contrasting changes in levels of the growthpromoting hormones brassinosteroids (BRs). The crop's tolerance was positively correlated with levels of BR precursor typhasterol while negatively with the active BR

**161**

brassinolide. Principal Component Analysis revealed correlations in observed changes in phytohormones, biochemical, and physiological parameters. Overall, the results show that kale is the most tolerant of the three species and Chinese cabbage the most sensitive to salt stress, and provide holistic indications of the spectrum of tolerance mechanisms involved.

Keywords: Chinese cabbage, kale, salinity stress, photosynthetic performance, stress hormones, brassinosteroids, tolerance, white cabbage

#### INTRODUCTION

Global warming and associated climate changes are imposing severe abiotic stresses that are seriously impairing crop yields and quality in many affected areas. One of these stresses (exacerbated by various human activities) is soil salinity. Over 7% of the world's total land and approximately 20% of irrigated land is affected by high salinity. The problem is particularly severe in the Mediterranean, semi-arid and arid areas (Zhang et al., 2014; Munns and Gilliham, 2015).

Plants' growth rates and productivity depend on photosynthetic efficiency. Thus, it is highly important for them to adjust their photosynthetic apparatus in accordance with environmental stresses. Photosystem II (PSII) is the most sensitive part of the apparatus to salt stress (Kalaji et al., 2011; Jajoo, 2014; Oukarroum et al., 2015). However, salinity stress has complex effects on photosynthetic activity, depending on the species or cultivar, duration of the stress and salt concentration. Low salt concentrations usually induce adaptations of photosynthetic activity, overall connectivity of the photosystem units and functional antenna size that maintain or even increase photosynthetic efficiency (Mehta et al., 2010; Da¸browski et al., 2016). In contrast, higher salinity usually causes photoinhibition of both photosystems (PSII and PSI), inhibits overall electron transport chain activity and increases non-photochemical quenching (Jajoo, 2014).

Salt concentrations exceeding taxa-specific thresholds induce three kinds of interacting stresses that are collectively called salinity stress. These are ionic stress caused by toxic concentrations of ions (mainly Na+), osmotic stress caused by associated reductions in water uptake, and oxidative stress mainly driven by increases in levels of reactive oxygen species (ROS) (Liang et al., 2018). Energy-efficient osmotic adjustment of cell turgor by the accumulation of Na<sup>+</sup> and Cl<sup>−</sup> in leaves and roots is a characteristic response of salt-tolerant species and halophytes. In addition, maintaining low Na<sup>+</sup> (and Cl−) concentrations in the cytoplasm of cells, optimizing the concentration of essential K <sup>+</sup> and efficient sequestration of salt ions in vacuoles are key elements of "tissue tolerance" (Munns et al., 2016). In fact, Na<sup>+</sup> competing with the binding site of K<sup>+</sup> on proteins e.g., channels and enzymes, is the main reason for Na<sup>+</sup> toxicity (Almeida et al., 2017). High Na<sup>+</sup> concentrations in cellular cytoplasm can inactivate enzymes and metabolic processes. Furthermore, the accumulation of Na<sup>+</sup> in leaves can inhibit photosynthesis if ions are not adequately compartmentalized at the cellular or subcellular level (Julkowska and Testerink, 2015). Moreover, more salt-sensitive species respond by de novo synthesis of osmolytes and allocation of growth assimilates to osmotic adjustment, with consequent increases in energy costs and reductions in growth rates (Munns et al., 2016). In addition, the oxidative stress associated with high salinity necessitates induction of antioxidant mechanisms to detoxify ROS, as inefficient ROS removal can result in damage to essential macromolecules (proteins, lipids, and nucleotides) and cell death.

One of the key mediators and modulators of plants' responses to all environmental factors are phytohormones, the main plant signaling molecules (Verma et al., 2016; Raja et al., 2017). Complex networks of interacting phytohormones [abscisic acid (ABA), salicylic acid, jasmonates (JAs), brassinosteroids (BRs), cytokinins, ethylene, auxins, and gibberellins] play crucial roles in plants' physiological responses and adaptation to salinity stress (Fahad et al., 2015). One of these phytohormones, ABA, is a key mediator of osmotic stress responses (Fahad et al., 2015). ABA regulates transpiration rates and maintains cellular turgor by controlling stomatal opening and closure (in concert with other phytohormones), induces osmoprotectant accumulation, activates ROS detoxification mechanisms, and modifies ion transport (Finkelstein, 2013; Fahad et al., 2015). JAs also modulate plants' growth, development and abiotic stress responses (Ahmad et al., 2016). They play key roles, inter alia, in the development of embryos, seedlings and floral organs, seed germination, growth inhibition, and senescence. JAs also participate interactively with other phytohormones in crops' performance affecting adaptations to changes in environmental conditions (Per et al., 2018). Inter alia, crosstalk between ABA and JAs regulates stomatal closure and transpiration under drought conditions, and in saline conditions, JAs are involved in the control of uptake of sodium ions (Riemann et al., 2015). BRs also have been recently reported roles in plant tolerance to diverse stress factors (e.g., salt, drought, and temperature stresses) (Jiroutova et al., 2018). BRs participate in the control of cell cycling and growth, modification of cell wall architecture, and adjustment of membrane systems. Moreover, they contribute to the maintenance of cells' redox systems and the regulation of stomatal aperture in drought and salinity responses (Sharma et al., 2017).

To meet problems caused by salinity, knowledge of major crops' tolerance levels and mechanisms is clearly important. Currently, most Brassica crop species are classified as moderately salt tolerant. However, the amphidiploid species B. juncea, B. napus, and B. carinata reportedly have a somewhat higher tolerance than the diploids B. oleracea, B. nigra, and B. rapa (Purty et al., 2008). The reasons for this are uncertain, and there is little knowledge of tolerance mechanisms in the family. To assist

efforts to elucidate mechanisms responsible for salt tolerance in Brassicaceae, three Brassica crops with global economic importance were selected for this study: white cabbage (Brassica oleracea var. capitata), kale (B. oleracea var. acephala), and Chinese cabbage (B. rapa L. pekinesis). Given the complexity of tolerance mechanisms, a holistic approach was chosen in efforts to elucidate salinity responses of the three Brassica crops.

We hypothesized that early, initial salinity responses may be essential for long-term salinity tolerance and provide insights into the involvement of different defense mechanisms. Moreover, we hypothesized that fine tuning of hormonal status plays major roles in early salt stress responses and further adaptation of the selected brassicas. We first determined their sensitivity/tolerance to applied salinity stresses by root-growth bioassays and biomass production. Then photosynthetic parameters of hydroponically grown plants were measured following short-term salinity stress. In parallel, biochemical parameters involved in responses to oxidative stress (the level of lipid peroxidation), osmotic stress (proline accumulation), and ionic stress (Na+/K<sup>+</sup> ratio) were measured in leaves and roots of treated plants in comparison to corresponding controls. We also measured changes in levels of ABA, JAs, and BRs to explore their roles in the regulation of salinity responses in the selected crops. Finally, Principal Component Analysis (PCA) of acquired data enabled us to draw correlations regarding traits of the crops and identify key contributors to differences in their salinity responses.

#### MATERIALS AND METHODS

#### Plant Material and Experimental Conditions

Seeds of Chinese cabbage (B. rapa L. ssp. pekinensis (Lour.) Hanelt cv. Cantonner Witkrop), white cabbage (B. oleracea var. capitata cv. Varaždinski) and kale (B. oleracea var. acephala cv. IJK9) were purchased from ISP International Seed Processing GmbH, Quedlinburg, Germany, the Agricultural Advisory Service of Varaždin Region, Croatia, and Institute for Adriatic Crops and Karst Reclamation, Split, Croatia, respectively.

The level of salinity tolerance/sensitivity of the selected Brassica crops was determined using the root-growth bioassay described by Pavlovic et al. (2018a) ´ . The effect of NaCl on the seedlings biomass production was evaluated after 7 days of treatments in comparison to the corresponding controls.

For hydroponic growth, seeds were germinated on 1% agar plates and then several-day-old seedlings were placed in a home-made hydroponic growth system supplying commercially available nutrient solutions (Flora Series and GHE Hydroponics) according to the manufacturers' instructions. Plants were grown in 5.5 L dark pots in a growing chamber at 21◦C, with 16/8 h light (115 µmol m−<sup>2</sup> s −1 )/dark cycles.

At the four fully developed leaf stage (after 3–4 weeks, depending on the cultivar) sets of the plants were subjected to salinity stress, by incrementally increasing concentrations of NaCl for 25 mM and 50 mM in the nutrient solution (in order to avoid shock). Salinization was performed at 2 h intervals to final concentrations as follows: 50 mM NaCl (25 mM/two steps), 100 mM Na (25 mM/four steps), and 200 mM NaCl (25 mM/four steps and 50 mM/two steps), respectively. The nutrient solution of controls of each species remained unchanged. After 24 h of exposure to the final salt concentrations, the plants were harvested together with controls. The experiment was performed with four biological replicates, each consisting of eight plants from a pot, unless otherwise stated.

Plant material for biochemical analysis was stored at −80◦C until use while material for hormonal profiling and determination of ion contents was freeze-dried until analysis. Photosynthesis measurements (fast chlorophyll a fluorescence kinetics) were performed on leaves in vivo, as described below.

#### Fast Chlorophyll a Fluorescence Kinetics

Fast chlorophyll a fluorescence kinetics were measured on 18 randomly selected, dark-adapted leaves of each Brassica cultivar using a Handy-PEA fluorimeter (Hansatech, United Kingdom). After 30 min of dark adaptation, the leaves were exposed to a pulse of saturating red light (3200 µmol m−<sup>2</sup> s −1 , peak at 650 nm). OJIP transients were measured by recording data from 50 µs (F0) to 1 s (Fm). Data extrapolated from the acquired OJIP curves were subjected to JIP-tests to calculate parameters according to Strasser et al. (2000). Calculations are shown in **Supplementary Table S1**.

The OJIP transients were double-normalized between O (50 µs) and P steps and presented as relative variable fluorescence, WOP = (Ft−F0)/(FP−F0). Fluorescence data were plotted on a logarithmic time scale, and the O, J, I, and P steps were marked in plots. Normalization between O and K (300 µs) steps revealed L-band (150 µs) which was presented as variable fluorescence WOK = (Ft−F0)/(FK−F0) and then plotted with difference kinetics 1WOK = WOK−(WOK)ref. Normalization between O and J (2 ms) steps revealed K-band, presented as variable fluorescence WOJ = (Ft−F0)/(FJ−F0) and plotted with difference kinetics 1WOJ = WOJ−(WOJ)ref. Values measured in control plants were used as referent value (WOK)ref (Strasser et al., 2004; Yusuf et al., 2010).

#### Biochemical Stress Parameters

Levels of MDA and proline were determined spectrophotometrically using previously reported methods (Radic et al., ´ 2009). Plant material for determination of MDA levels (250 mg of fresh weight of leaves and roots) was extracted in 2 ml of potassium phosphate buffer (50 mM, pH 7.0, 0.1 mM EDTA) with the addition of polyvinylpolypyrrolidone (PVPP). MDA levels in the extracts were estimated by MDA reaction with thiobarbituric acid, subtracting the absorbance at 600 nm from the absorbance at 532 nm, and using an extinction coefficient of 155 mM−<sup>1</sup> cm−<sup>1</sup> .

Proline was extracted from 100 mg samples (FW) in 1.5 ml of 3% sulfosalicylic acid and determined spectrophotometrically at 520 nm using ninhydrin. Absorbance values were adjusted using a calibration curve constructed with L-proline as a standard and the results were expressed in nmol proline per g FW.

A Specord 40 spectrophotometer (Analytik Jena, Jena, Germany) was used for all absorbance measurements, with four replicates per assay.

# Contents of Sodium and Potassium Ions

Contents of sodium and potassium ions in roots and leaves of the Brassica crops were measured by high-resolution inductively coupled plasma mass spectrometry, using a Thermo Fisher Scientific HRICP-MS Element 2 instrument (Thermo Fisher Scientific, Bremen, Germany) equipped with an ESI-a SC-2 DX FAST autosampler (Elemental Scientific, United States) using indium as an internal standard. Typical instrumental conditions and measurement parameters used throughout the work have been previously reported (Fiket et al., 2016). Before analysis, powdered lyophilized tissue samples were subjected to microwave-assisted acidic digestion in HNO3/HF (60:1, v/v) using a Multiwave 3000 (Anton Paar, Graz, Austria) at 1400 W.

# Hormonal Profiling

#### Stress-Related Hormones

The stress-related phytohormones, JAs (jasmonic acid, JA, and jasmonoyl-L-isoleucine, JA-Ile) and ABA were determined as previously described (Floková et al., 2014) with minor modifications. Briefly, lyophilized samples (5 mg dry weight, DW) were homogenized with a MM 301 vibration mill (Retsch GmbH, Haan, Germany), extracted in 1 mL 50 mM sodium phosphate buffer (pH 7.0) containing 1% sodium diethyldithiocarbamate and stable isotope-labeled internal standards (10 pmol [2H6]JA, 10 pmol [2H6]ABA and 0.1 pmol [ <sup>2</sup>H2]JA-Ile; OlchemIm, Olomouc, Czechia) then alkalized by adding 1 mL of 5% NH4OH/H2O (v/v). The resulting solution was purified by passage through a mixed-mode anion exchange column (Oasis <sup>R</sup> MAX column, 1 cc/30 mg, Waters, Milford, MA, United States) conditioned with 100% MeOH and equilibrated with H2O and 5% NH4OH (1 ml of each solution). After sample loading, the column was washed with 2 mL 5% NH4OH followed by 2 mL 100% MeOH, then the acidic phytohormones were eluted using 2 mL 2% HCOOH in 100% MeOH (v/v). The samples were evaporated to dryness under a stream of nitrogen and stored in a freezer at −20◦C until analysis.

#### Brassinosteroids

Samples' BR contents were analyzed as previously described (Oklestkova et al., 2017) with a few modifications. Briefly, lyophilized samples (40 mg DW) were homogenized to a fine consistency using 3 mm zirconium oxide beads and an MM 301 vibration mill at a frequency of 30 Hz for 3 min (Retsch, Haan, Germany). The samples were then extracted overnight with stirring at 4◦C using a benchtop laboratory rotator (Stuart SB3; Bibby Scientific, Cole-Parmer, Staffordshire, United Kingdom) after adding 1 mL of icecold 60% acetonitrile and 30 pmol of [2H3]brassinolide (BL), [ <sup>2</sup>H3]castasterone (CS), [2H3]typhasterol (TY), [2H3]24-epiBL, [ <sup>2</sup>H3]24-epiCS, [2H3]28-norBL, and [2H3]28-norCS (OlchemIm, Olomouc, Czech Republic) as internal standards. The samples were then centrifuged, purified using DPA-6S SPE columns (Supelco, Bellefonte, PA, USA) and evaporated to dryness in vacuo. They were then dissolved in 75 µl 100% MeOH with sonication, made up to 1 ml with PBS buffer (pH 7.2) then loaded on an immunoaffinity column (IAC) coated with anti-BR monoclonal antibodies. After washing the IAC with 9 mL H2O, BRs were eluted using 3 mL ice-cold MeOH (−20◦C), evaporated in vacuo using a CentriVap <sup>R</sup> acid-resistant benchtop concentrator (Labconco Corp., MO, USA) and stored at −20◦C until analysis.

#### UHPLC-MS/MS Analysis

For ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) analysis, all samples were dissolved in 40 µL of the mobile phase and analyzed using an Acquity UPLC System (Waters, Milford, MA, USA) coupled to a XevoTM TQ-S MS triple quadrupole mass spectrometer (Waters, MS Technologies, Manchester, United Kingdom). Stable isotope-labeled internal standards were used as references and concentrations of analytes were quantified by the isotope dilution method. Previously published instrument settings were used for profiling stress-related phytohormones (Floková et al., 2014) and BRs (Oklestkova et al., 2017). Four independent replicate analyses were performed for each hormone group if not stated otherwise.

#### Statistical Analysis

Between-treatment differences in measured variables of each of the Brassica crops were evaluated using factorial analysis of variance (ANOVA) followed by the post hoc Tukey Honest Significant Difference (HSD) test. Differences were considered significant if p < 0.05. To minimize bias in comparisons of the varieties, which differed in salinity tolerance, all data were normalized to control values. Data presented in the text, figures, and tables are means ± standard deviation of four replicates (n = 4) for biochemical markers and phytohormones, and means ± standard deviation of 18 replicates (n = 18) for fluorescence measurements.

# Principal Component Analysis

Correlations among the measured physiological, biochemical, and hormonal parameters, treatments and responses of the Brassica crops were explored by PCA. PCA was performed using the correlation matrix of the average values of traits after standardization (autoscaling). Linear correlations among variables were determined by Pearson coefficients (p < 0.05). XLSTAT software (ver. 2017.01.40777) implemented in Microsoft Office Excel 2010 was used for all statistical procedures.

# RESULTS

# Salinity Tolerance Evaluation by Root-Growth Bioassay

The inhibition of root-growth during exposure to NaCl at 50–200 mM for 24 h was used as a quick, convenient bioassay to evaluate the salinity tolerance of the selected brassicas (**Figure 1**). As shown in **Figure 1**, the treatments caused dose-dependent root-growth inhibition in all three varieties, but most strongly in Chinese cabbage. Root-growth inhibition rates were similar in white cabbage and kale. Seedlings biomass production upon prolonged salinity stress (up to 7 days) showed statistically significant inhibition in Chinese cabbage and white cabbage

at higher salt concentrations (100 and 200 mM NaCl), while there is no significant inhibition in biomass production in kale (**Supplementary Figure S1** in the **Supplementary File**).

# Salinity Effects on Photosynthetic Performance

To acquire information about the selected brassicas' photosynthetic performance, we measured fluorescence transients of hydroponically grown plants exposed to 50, 100, or 200 mM NaCl (**Figures 2D–F**). Normalized OJIP transients, indicating differences between control and stressed plants in PSII and PSI functionality, demonstrated that the fluorescence intensity at defined transient steps successively increased with increases in NaCl concentration in all three brassicas. Spider plots presenting normalized values of NaCl-treated plants relative to the controls are shown in **Figures 2A–C**. The plots show clear between-taxa differences in total photosynthetic performance index, PItotal, a sensitive parameter reflecting the functional activity of PSII, PSI, and intersystem electron transport chain. The PItotal of Chinese cabbage decreased 26.6, 47, and 66.6% after exposure to 50, 100, and 200 NaCl, respectively, while decreases were only detected after exposure to 100 and 200 mM NaCl in white cabbage (23.8 and 46.2%, respectively) and kale (26.7 and 41.9%, respectively). The density of reaction centers (RCs) per unit chlorophyll a, RC/ABS, in Chinese cabbage decreased 8% at the highest NaCl concentration. In contrast, 100 and 200 mM NaCl induced significant increases in RC/ABS in white cabbage (8.4 and 11.5%, respectively) and kale (8.1 and 8.9%, respectively). Flux ratio trapping per dissipation, TR0/DI0, and electron transport further than QA, ET0/(TR0−ET0), significantly declined in Chinese cabbage exposed to 100 and 200 mM NaCl (by 8.8 and 14.9%, respectively). In white cabbage and kale, these parameters decreased significantly only at the highest NaCl concentration, relative to controls (by 12.2 and 7.9%, respectively).

The probability of end electron acceptor reduction, RE0/ (ET0−RE0), decreased in Chinese cabbage by 20.6, 33.3, and 48.6% after exposure to 50, 100 and 200 NaCl, respectively. However, this parameter decreased only after exposure to 100 and 200 mM NaCl, relative to controls, in white cabbage (by 25.5 and 39.5%, respectively) and kale (by 30.5 and 38.9%, respectively).

Fluorescence intensity at 50 s (F0) was not affected in any cultivar by any of the salt treatments. The maximum fluorescence intensity (Fm) was significantly decreased by exposure to 100 and 200 mM NaCl in Chinese cabbage (by 5.9 and 12.4%, respectively), while in white cabbage and kale it only significantly decreased at 200 mM (by 6.4 and 6.5%, respectively) relative to controls. Similarly, the maximum quantum yield of primary photochemistry, TR0/ABS, significantly decreased in Chinese cabbage at 100 and 200 mM NaCl relative to controls (by 1.7 and 3.1%, respectively), while in white cabbage and kale it decreased only at the highest NaCl concentration (by 1.5 and 1.2%, respectively).

Variable fluorescence at 150 µs (VL) showed no significant changes in any cultivar after exposure to salt at any concentrations relative to controls. Variable fluorescence at 300 µs (VK) significantly increased (by 5.3%) at the highest NaCl concentration in Chinese cabbage, while 100 and 200 mM NaCl caused significant reductions (7.2 and 9.2%, respectively) of this parameter in kale. There were significant increases in variable fluorescence at the J step (VJ) in Chinese cabbage at 100 and 200 mM NaCl (5.9 and 10.7%, respectively), but only at 200 mM in white cabbage and kale (9.1 and 5.8%, respectively). Similarly, variable fluorescence at 30 s (VI) increased after exposure to 50, 100, and 200 mM NaCl in Chinese cabbage (by 5.0, 8.4, and 13.0%, respectively), but only after exposure to 100 and 200 mM NaCl in white cabbage (by 7.0 and 12.2%, respectively) and kale (by 8.2 and 11.1%, respectively).

There were clear differences in L-bands (**Figures 3A–C**) and K-bands (**Figures 3D–F**), obtained from the normalized O-K and O-J curves between Chinese cabbage and the two other Brassica crops. Bands for Chinese cabbage were positive and their amplitude was highest after exposure to 200 mM NaCl. In contrast, both bands for white cabbage and kale were negative after exposure to NaCl at all concentrations, but again their amplitude was highest at 200 mM NaCl.

# Effects of Salinity on Biochemical and Physiological Stress Parameters

To evaluate effects of the treatments to major known biochemical salinity stress markers, the Na+/K<sup>+</sup> ratio, and levels of proline and MDA (**Figure 4**) were measured in roots and leaves of the selected Brassica crops.

Maintenance of the potassium to sodium ionic ratio and accumulation of proline are salinity defense strategies. The increase of sodium ions and decrease of potassium ions (**Supplementary Table S2**) resulted in higher Na+/K<sup>+</sup> ratio upon salinity (**Figures 4A,B**). The Na+/K<sup>+</sup> ratio was affected more strongly in roots than in leaves, and exposure to the highest salt concentration (200 mM) resulted in 254-, 291-, and 589-fold changes in Chinese cabbage, white cabbage, and kale

roots, respectively, relative to controls (**Figure 4A**). The highest difference in Na+/K<sup>+</sup> ratio in leaves between control and salt treatments was 43-fold, in Chinese cabbage after exposure to 200 mM NaCl (**Figure 4B**), which induced ca. 10-fold increases in the Na+/K<sup>+</sup> ratios of the other two brassicas' leaves. The osmotic effect of NaCl was evaluated by monitoring changes in levels of proline (**Figures 4C,D**), which successively increased after exposure to 100 and 200 mM NaCl in both roots and leaves of the three crops. We detected 9.6-, 5.7-, and 8.2-fold increases in proline levels in roots of Chinese cabbage, white cabbage, and for each species.

fpls-10-00450 April 10, 2019 Time: 17:49 # 7

as difference kinetics 1WOJ = WOJ–(WOJ)ref in the 0.05–2 ms time range. (WOK)ref and (WOJ)ref reference values were obtained from measurements of control plants

kale, respectively, after exposure to 200 mM NaCl, relative to controls (**Figure 4C**), and much higher increases in their leaves (15.6-, 39.7-, and 13.9-fold, respectively) (**Figure 4D**). To evaluate the oxidative effects of the treatments, MDA contents were measured (**Figures 4E,F**). The treatments did not significantly affect the MDA content in kale roots but significantly affected it in Chinese cabbage roots at all salt concentrations and white cabbage roots at higher salt concentrations (100 and 200 mM NaCl) (**Figure 4E**). Moreover, the two highest salt concentrations (100 and 200 mM) increased MDA levels in leaves of all three brassicas relative to corresponding controls (**Figure 4F**).

#### Effects of Salinity on Stress Hormones

Changes in stress-related hormones (ABA, JA, and JA-Ile) induced by the treatments are shown in **Figure 5**. Exposure to 100 and 200 mM NaCl caused significant increases in ABA levels in roots and leaves of all three brassicas (**Figures 5A,B**). The strongest increase in roots (2.85-fold relative to controls) was induced by exposure to 100 mM NaCl in roots of kale. However, the highest salinity (200 mM NaCl) decreased ABA levels in roots of Chinese cabbage (to 62% of control levels), although it increased them (approximately twofolds relative to controls) in white cabbage and kale roots. In leaves, the highest increase in ABA levels (25.5-fold) was also induced by 100 mM NaCl in Chinese cabbage (**Figure 5B**). The 200 mM NaCl treatment resulted in a smaller increase in ABA levels in Chinese cabbage, but similar and further increases in levels relative to those induced by 100 mM NaCl in leaves of white cabbage and kale, respectively. Levels of JAs (JA and JA-Ile) in roots and leaves (**Figures 5C–F**) showed similar responses to the salt stress treatments in all three crops. In mild stress conditions (50 mM NaCl) JA and JA-Ile (**Figures 5D,F**) remained at control levels in leaves of Chinese and white cabbage while levels of both hormones were significantly lower in kale. In root tissue, reductions in JA and JA-Ile contents (**Figures 5C,E**) were more pronounced in Chinese cabbage than in the other crops at 50 mM NaCl. Higher salinity resulted in significant decreases in levels of both hormones in roots and leaves of all three brassicas.

# Effects of Salinity on Brassinosteroids

Changes in levels of BRs in the brassicas induced by the treatments are presented in **Figure 6**. In both roots and leaves, only three BRs (TY, CS, and BL) were detected. Reductions in levels of the precursor TY in roots and leaves (**Figures 6A,B**) were accompanied by increases in BL (**Figures 6E,F**) in both organs of Chinese cabbage exposed to NaCl at all concentrations, relative to controls. In contrast, TY, CS, and BL levels were moderately reduced in leaves and not affected in roots of white cabbage. The three stress treatments induced significant increases in levels of TY, CS, and BL in roots of kale (except in CS levels at 100 mM NaCl), but either no significant changes or reductions in kale leaves.

# Principal Component Analysis (PCA)

To investigate correlations among the brassicas in terms of the measured stress-related parameters, the data were subjected to PCA based on a matrix of Pearson correlation coefficients (p < 0.05). Average values of measured traits were standardized

prior to analysis by autoscaling. The resulting correlation matrix, eigenvalues, factor loadings and factor scores are presented in **Supplementary Tables S6–S20**. Positions of the brassicas and relations among the measured parameters under the salinity treatments, in roots and leaves are shown in PCA biplots in **Figures 7A,B**, respectively. The first two Principal Components, F1 and F2, explained 61.24 and 65.70% of the cumulative variability of measured traits in the roots and leaves, respectively. Detailed PCA of correlations of photosynthetic parameters with the cultivars and treatments is presented in **Supplementary Figure S2** in the **Supplementary File** and **Supplementary Tables S16–S20**. The first two Principal Components explained 79.85% of the cumulative variability of photosynthetic traits in leaves.

The separation of the cultivars, particularly Chinese cabbage, in the root biplot (**Figure 7A**) clearly shows that their salinity responses differ. Kale and white cabbage treatments were grouped close to their controls in the left quadrants, while Chinese cabbage treatments were positioned far away from the corresponding control. The PCA plot also shows relations of the measured parameters in the brassicas' responses (factor loadings are presented in **Supplementary Tables S8**, **S13** for roots and leaves, respectively). The phytohormones BL and ABA, and stress marker MDA are positioned close to 50 and 100 mM NaCl-treated Chinese cabbage roots, while proline is grouped together with 100 and 200 mM NaCl-treated white cabbage roots and Chinese cabbage at 200 mM NaCl, indicating that they are the main response parameters to these treatments. Parameters positioned in the same quadrant as kale roots and white cabbage under control and 50 mM NaCl treatments are TY and Na+/K<sup>+</sup> ratio.

In the leaf biplot (**Figure 7B**), cultivars exposed to the control and mild stress (50 mM NaCl) treatments are grouped together on the left side of the plot, and the severe salinity treatments shifted them to the right. Like the pattern in the root biplot, Chinese cabbage leaves were positioned far from kale leaves, and white cabbage leaves exposed to the salinity treatments were in intermediate positions, indicating differences in their salinity

responses and hence sensitivity to salt stress. The parameters positioned closest to Chinese cabbage exposed to severe salinity are the phytohormones ABA and BL, together with Na+/K<sup>+</sup> ratio, while proline, and MDA are positioned closer to the other two Brassica varieties.

and mild salinity-treated B. oleracea varieties (white cabbage and kale), while the V<sup>J</sup> , VL, and V<sup>I</sup> parameters were positioned closer to Chinese cabbage.

There were also significant correlations between phytohormones, biochemical, and selected photosynthetic parameters (PItotal and TR0/ABS) that reflect the overall efficiency of photosynthesis in leaves (**Supplementary Table S11**). These photosynthetic parameters were negatively correlated with the Na+/K<sup>+</sup> ratio, proline and MDA, as well as ABA.

Based on photosynthetic parameters, B. oleracea cultivars (controls and salinity-treated) grouped close together and far from Chinese cabbage (**Supplementary Figure S2** in the **Supplementary File**). PItotal was positively correlated with its constituents: TR0/DI0, ET0/(TR0−ET0), RE0/(ET0−RE0), and TR0/ABS, but negatively correlated with RC/ABS. It was also positively correlated with F<sup>m</sup> but negatively correlated with V<sup>J</sup> and V<sup>I</sup> . PItotal and its constituents were grouped close to control

#### DISCUSSION

Soil salinization is a major agricultural problem, which impairs crops' growth, yields, and quality. Soils with electrical conductivity (EC) of >4 dS/m are characterized as saline. However, under some environmental conditions, soil EC values can exceed 20 dS/m (corresponding to 200 mM NaCl) (Shannon and Grieve, 1999; Zhang et al., 2014). Thus, there are urgent needs to characterize crops' salinity responses and the mechanisms involved in salinity tolerance. To aid such efforts we have examined, in detailed responses of selected brassicas to short-term salinity: Chinese cabbage (B. rapa), white cabbage (B. oleracea var. capitata) and kale (B. oleracea var. acephala). All three brassicas were grown in a hydroponic

system until they reached the four fully developed leaf-stage, at which they are frequently transferred into fields in conventional cultivation systems and thus exposed to different environmental conditions. The selected salt concentrations (50, 100, and 200 mM NaCl) correspond to concentrations in naturally occurring saline soils. Diverse physiological, biochemical, and hormonal parameters were measured to gain insights into the brassicas' responses to short-term salinity and correlations between their responses and tolerance.

# Salinity Tolerance and Photosynthetic Performance of the Selected Brassicas

An earlier study on Brassica crops identified root-growth inhibition as a significant biomarker of salt sensitivity (Zhang et al., 2014). Thus, we used a simple root-growth bioassay to characterize the selected cultivars' salt sensitivity and found that Chinese cabbage was the most sensitive, while kale and white cabbage were more tolerant under the applied experimental conditions. These results are consistent with previous findings that cabbages (including B. oleracea var. capitata) have a salinity threshold of 180 mM NaCl, kales (such as B. oleracea var. acephala) can grow after exposure to 230–550 mM NaCl following re-watering (Shannon and Grieve, 1999), while many B. rapa sub-species have been described as salt-sensitive (Jan et al., 2016). As a component of salinity stress is osmotic stress, the results are also consistent with our recently published finding of a corresponding pattern of tolerance to drought conditions (Pavlovic et al., 2018b ´ ).

Photosynthesis, and hence crops' growth and productivity can be inhibited by salinity, to degrees that depend on the crops' tolerance. Therefore, we tested the influence of salinity on the

three brassicas' photosynthetic performance by measuring their fast chlorophyll a fluorescence kinetics and determining several photosynthetic parameters. It is well known that OJIP kinetics, which reflects the redox state of Q<sup>A</sup> and QB, are extremely sensitive to salt stress (Mehta et al., 2010; Da¸browski et al., 2016; Duarte et al., 2017). The O-J phase is a light-dependent phase of OJIP kinetics that reflects reduction of the acceptor side of PSII (Schansker et al., 2014) and provides information on antenna size and connectivity between PSII RCs. Therefore, the significantly higher increase in variable fluorescence at the J step (VJ) observed in Chinese cabbage in response to salinity stress than in the other two varieties can be attributed to a stronger decrease in re-oxidation of QA. Between the O and J steps, two additional steps can be distinguished, designated the L- and K-bands (Yusuf et al., 2010). Positive L-bands obtained for salt-stressed Chinese cabbage, relative to controls (**Figure 3A**), indicate that its connectivity and system stability were impaired more than in the other two brassicas, for which negative L-bands were obtained. In addition, positive K-bands were obtained for salt-stressed Chinese cabbage (**Figure 3D**), presumably due to impaired electron flow between OEC and the acceptor side of the RC, while negative K-bands were obtained for salt-stressed white cabbage and kale (Yusuf et al., 2010; Chen et al., 2016). Negative K-bands have been recognized as signs of tolerance to stress conditions (Krüger et al., 2014; Zurek et al., 2014 ˙ ; Begovic´ et al., 2016), indicating that PSII antennae are functionally intact (Venkatesh et al., 2012). Further, the J-I phase reflects a partial reduction of the intersystem electron carriers, while the I-P phase reflects reduction of the acceptor side of PSI (Yusuf et al.,

2010; Schansker et al., 2014). The increase in V<sup>I</sup> induced by all salinity treatments in Chinese cabbage (**Figure 2A**) is attributable to faster accumulation of reduced Q<sup>A</sup> and QB-non-reducing PSII centers that cannot transfer electrons further along the electron transport chain (Tomek et al., 2001), suggesting that even the mildest salt stress partially inactivated its PSII centers. In contrast, indications of restricted electron transport between Q<sup>A</sup> and Q<sup>B</sup> (increases in VI) in white cabbage and kale were only detected following exposure to higher NaCl concentrations (**Figures 2B,C**). Furthermore, salinity exposure decreased the photosynthetic system's ability to convert excitation energy to electron transport beyond QA. This is consistent with a recent demonstration that increases in V<sup>I</sup> are due to failure of PSI to oxidize reduced plastoquinone (Goltsev et al., 2016).

Exposure to salt stress did not induce changes in the F<sup>0</sup> parameter in any Brassica, while the F<sup>m</sup> parameter decreased more strongly in Chinese cabbage than in the others (**Figure 2**) suggesting inhibition of electron flow through PSII. Reductions in the maximum yield of PSII primary photochemistry, TR0/ABS (**Figures 3A–C**) was obtained in all three Brassica cultivars, indicating impairment of PSII photochemical efficiency. This is supported by a recent report of similar changes in F0, F<sup>m</sup> and TR0/ABS parameters in two perennial ryegrasses exposed to salt stress (Da¸browski et al., 2016).

The most sensitive parameter of the JIP-test is the total photosynthetic performance index, PItotal (**Figures 2A–C**), which indicates the overall functional activity of both photosystems, PSII and PSI, as well as the intersystem electron transport chain (Yusuf et al., 2010; Krüger et al., 2014; Da¸browski

et al., 2016; Goltsev et al., 2016; Kalaji et al., 2017). This multiparametric expression reflects the overall efficiency of light energy absorption (RC/ABS), quantum yield of excitation energy trapping (TR0/DI0), probability of a trapped exciton moving an electron further along the electron transport chain than Q<sup>A</sup> [ET0/(TR0−ET0)] and the probability of PSI reducing its end acceptors [RE0/(ET0−RE0)] (Yusuf et al., 2010; Krüger et al., 2014). PItotal declined in all three Brassica crops exposed to salinity (**Figures 2A–C**), but most strongly in Chinese cabbage. The most sensitive PItotal component to the salinity treatments was the RE0/( ET0−RE0) parameter, reflecting the efficiency of processes involving PSI and its ability to reduce its end acceptors. In addition, reductions in TR0/DI<sup>0</sup> and ET0/(TR0−ET0) were observed in all three brassicas at the highest NaCl concentration, and in Chinese cabbage reductions also occurred at 100 mM NaCl. This suggests that the higher NaCl concentrations caused structural damage to thylakoids, thereby reducing trapping of excitation energy and its conversion to electron transport. Similarly, Da¸browski et al. (2016) reported that diminution of PSII activity and impairment of PSI function significantly decreased the total performance index in ryegrass, and Oukarroum et al. (2015) detected salt-induced inhibition of both PSII and PSI electron transport activities in duckweed. In addition, Mehta et al. (2010) suggested that salt stress decreased efficient electron transport further than the primary acceptor, and it can significantly decrease values of all parameters related to electron transport in tomato leaves (Zushi and Matsuzoe, 2017).

# Salt Stress-Related Biochemical Parameters in the Selected Brassicas

An important component of plants' salt tolerance is the maintenance of an appropriate cytosolic Na+/K<sup>+</sup> ratio, and the tolerance of Brassica genotypes reportedly correlates inversely with Na+/K<sup>+</sup> ratios in their roots, steam, and leaves when exposed to salinity. Salt tolerance is also reportedly correlated with transcription levels of genes involved in ionic homeostasis, including components of the Salt Overly Sensitive (SOS) response and vacuolar NHX1 Na+/H<sup>+</sup> antiporter responsible for pumping Na<sup>+</sup> into vacuoles from the cytoplasm (Chakraborty et al., 2012). Furthermore, efficient compartmentalization of Na<sup>+</sup> into vacuoles and regulation of Na<sup>+</sup> transport from roots to shoots is crucial for plant survival since Na<sup>+</sup> is more toxic to leaves (Almeida et al., 2017; Assaha et al., 2017). These reports are consistent with our findings that white cabbage and kale had lower Na+/K<sup>+</sup> ratios in their leaves than the salt-sensitive Chinese cabbage, mainly due to increases in Na<sup>+</sup> contents. Observations of lower Na+/K<sup>+</sup> ratios in leaves and higher ratios in roots of white cabbage and kale suggest that control of Na<sup>+</sup> transport from roots to leaves is more efficient in these cultivars.

In addition to ionic homeostasis, osmoregulation of cell turgor is another important physiological process under salt stress. We observed a positive correlation between the Na+/K<sup>+</sup> ratio and proline content. Osmoregulatory accumulation of proline under saline conditions is reportedly a common response of glycophytes, including many brassicas (Kumar et al., 2009; Mittal et al., 2012; Jan et al., 2016). Higher concentrations of proline accumulated in leaves than in roots, while the Na<sup>+</sup> levels were much higher in roots of the selected brassicas. Similar salinity responses have been recorded in tomato plants, where proline accumulation positively correlated with expression of the P5CS (proline biosynthesis) gene and accumulation of Na<sup>+</sup> in leaves, but not in roots (Almeida et al., 2014). The cited authors concluded that there is no link between proline accumulation and a plant tolerance index, suggesting that proline "is not, per se, the driving force of tolerance." Similarly, we found that kale leaves accumulated less proline than the other two, more sensitive brassicas. Some studies suggest that proline contributes to redox homeostasis, and increases in its biosynthesis help maintain functionality of the photosynthetic apparatus and induce increases in activities of antioxidant enzymes (Puniran-Hartley et al., 2014; Surender Reddy et al., 2015). However, Pearson's coefficients we obtained show a negative correlation between the PItotal photosynthetic index and proline, suggesting that proline does not play a major protective role in photosynthesis in the selected Brassica cultivars under our experimental conditions. Furthermore, increases in MDA levels (a marker of lipid peroxidation) suggest that salt stress increases ROS production in leaves of all three brassicas more than in roots. The pro-oxidant character of high endogenous proline levels (Soshinkova et al., 2013) in leaves or insufficient antioxidant machinery could potentially explain our findings.

### Roles of Phytohormones in Salinity Responses of the Selected Brassicas

Early stress responses are mediated by complex crosstalk of phytohormones (including synergistic and antagonistic interactions), among which ABA is reportedly the main salinity messenger (Fahad et al., 2015). Links between ABA and salinity tolerance are widespread in the plant kingdom but highly dependent on genotype, developmental stage, and stress conditions. Increases in ABA in responses to salinity have been detected in sensitive genotypes of tomato (Duan et al., 2012), basil (Mancarella et al., 2016), and barley (Witzel et al., 2014). However, significant increases in ABA have also been recorded in early responses of a tolerant barley variety (Kamboj et al., 2015) and sustained responses of a tolerant tomato cultivar (Gharbi et al., 2017). According to Ellouzi et al. (2014), the halophytes Cakile maritima and Thellungiella salsuginea can accumulate ABA more efficiently than the glycophyte A. thaliana under extreme salinity (400 mM NaCl). These findings are consistent with our results. The moderately tolerant white cabbage and tolerant kale maintained or dose-dependently increased levels of ABA in their leaves in response to salt. In contrast, ABA levels were lower than control levels in roots of the saltsensitive Chinese cabbage at 200 mM NaCl, and lower in its leaves at 200 mM than at 100 mM NaCl. Moreover, observed decreases in the photosynthetic parameter PItotal, particularly in Chinese cabbage leaves, are consistent with control of transpiration in salinity responses through ABA-induced stomatal closure and associated reductions in photosynthetic rates (Ashraf and Harris, 2013).

Furthermore, the observed accumulation of proline following exposure to salinity can be partially explained by increased levels of ABA since proline biosynthesis is regulated by an ABAdependent and an ABA-independent pathway (Strizhov et al., 1997; Verslues and Bray, 2006). Thus, the observed discrepancy between proline accumulation and changes in ABA levels implies that the selected brassicas respond to salinity through different osmoregulatory mechanisms, which are organ- and cultivarspecific. Moreover, ABA is involved in the SOS pathway and prevention of Na<sup>+</sup> transport from roots to shoots through xylem (Zhu et al., 2017). Reduced ABA levels in Chinese cabbage roots, accompanied by a sharp increase in the Na+/K<sup>+</sup> ratio in leaves at the highest salinity, show that Chinese cabbage has less ability to control Na<sup>+</sup> leakage to shoots than the other two more tolerant cultivars.

In addition to ABA, JAs contribute to abiotic stress responses, although their roles in responses to biotic stress have received the most attention. There are also conflicting indications of their roles in salinity responses. Increases in JA levels following exposure to salinity, with associated increases in tolerance, have been recorded in several species, including tomato, iris, and rice (Wang et al., 2001; Pedranzani et al., 2003; Tani et al., 2008). However, impaired biosynthesis of the JA precursor OPDA reportedly enhances the stress tolerance of Arabidopsis aoc mutants by increasing ROS scavenging activity and delaying senescence. These mutants also reportedly accumulate less Na<sup>+</sup> in shoots than wild type controls when exposed to salinity, while their roots' ion contents are not affected (Hazman et al., 2015). Similarly, we found that in leaves of more tolerant cultivars exposed to severe salinity, contents of JA-Ile (the most bioactive jasmonate) were lower than in corresponding controls, while JA-Ile levels in leaves of salt-sensitive Chinese cabbage were not influenced by salinity. Furthermore, Kurotani et al. (2015) found that the bioactive JAs negatively affect rice viability under salt stress in experiments with a transgenic mutant. Enhanced expression of CYP94C2b, which contributes to the conversion of bioactive JA-Ile to the inactive forms 12OH-JA-Ile and 12COOH-JA-Ile, did not affect natural senescence but delayed salt-induced senescence in an overexpressing rice mutant. The repression of JA-signaling is maintained at low levels of JA-Ile, while increases in its levels trigger JA-signaling (Wasternack and Strnad, 2016). Therefore, the considerably lower amounts of JA and JA-Ile we observed in leaves of white cabbage and kale may potentially contribute to a postponement of JA-mediated senescence in saline conditions.

The role of BRs in salt stress has been explored through monitoring effects of exogenous application of active BRs, which reportedly increases plants' tolerance, mainly through induction of H2O<sup>2</sup> signaling and antioxidant machinery (Sharma et al., 2017). BRs can also mitigate some negative effects of salt stress on photosynthesis (Siddiqui et al., 2018). However, little is known about the roles of endogenous BRs in salinity stress responses. The Arabidopsis BR signaling impaired mutants BRdeficient det2-1 and BR-insensitive bin2-1 reportedly have higher sensitivity to salt stress during seed germination and seedling growth than wild type counterparts and have reduced levels of proline and transcripts of salt- and ABA-induced genes (Zeng et al., 2010). However, improved drought tolerance and delays in senescence during drought have been observed in transgenic creeping bentgrass (Agrostis stolonifera L.) with reduced levels of BRs (Han et al., 2017). Correlations between reduced levels of BL and improvements in drought tolerance have also been recorded in Arabidopsis plants (Northey et al., 2016). Moreover, we recently found that more BL accumulated in drought-sensitive Chinese cabbage than in the more tolerant kale and white cabbage in drought conditions (Pavlovic et al., 2018b ´ ). Furthermore, a positive correlation between ABA and BL was noted. It is known that stomata closure can be mediated by BR and stress hormone (ABA, ethylene) crosstalk (Ha et al., 2016; Jiroutova et al., 2018), which may lead to reduced photosynthetic rates under salinity stress (Ashraf and Harris, 2013). Since photosynthesis was most strongly affected in Chinese cabbage, changes in BR levels could be linked to the osmotic component of salt stress. Accumulation of BRs was also observed in roots; kale accumulated three times more CS at the highest salinity while BL levels in Chinese cabbage were increased up to 60-fold, depending on the treatment. The high accumulation of BL in this salt-sensitive cultivar correlates with its stronger inhibition of root growth observed in the root-growth bioassay. In addition, experiments with Arabidopsis bes1-D mutants, which have enhanced BR signaling, and treatment of wild type plants with high levels of BL, have revealed that BRs promote overall root growth inhibition (González-García et al., 2011). Maintenance of optimal levels of active BRs in roots may contribute to tolerance mechanisms since it has been proposed that BR signaling is temporarily inhibited in early phases of root growth response to salinity (Geng et al., 2013).

# CONCLUSION

Based on our data, Chinese cabbage is the most sensitive, white cabbage moderately while kale the most tolerant to salinity stress among selected brassicas. Reductions in photosynthetic efficiency (PItotal) were observed in all three brassicas exposed to higher salt concentrations, indicating that salt stress reduced the capacity of the photosynthetic system for efficient energy conversion, particularly in Chinese cabbage. This is in agreement with biomass reduction upon prolonged salinity stress (7 days) which was the most prominent in Chinese cabbage, then in white cabbage and finally kale. ABA levels were enhanced in all three brassicas under salt stress which is consistent with control of transpiration in salinity responses through ABA-induced stomatal closure and associated reductions in photosynthetic rates. It was shown that more tolerant varieties were able to sustain or even increase their ABA levels, while those of the sensitive Chinese cabbage declined under the most severe salt conditions. Better tolerance of kale was accompanying with better ability to control Na<sup>+</sup> leakage from roots to shoots; Na+/K<sup>+</sup> ratio was high in root and low in leaves with low MDA level at the same time in comparison to more sensitive brassicas. Furthermore, the considerably lower amounts of JA and JA-Ile measured in leaves of more tolerant white cabbage and kale may potentially contribute to the postponement of

JA-mediated senescence in saline conditions. Finally, enhanced salinity tolerance in selected brassicas is accompanying with a higher level of TY (precursor of active BRs) and lower level of active BL. The high accumulation of BL in salt-sensitive cultivar correlates with its stronger root growth inhibition and reduced photosynthetic rates under salinity stress.

To our knowledge, this is the first report on comparative research on some Brassica crops to gain insights into correlations between initial salinity tolerance and diverse physiological, biochemical and hormonal parameters in Brassicaceae. However, long-term exposure of Brassica crops to salinity stress need to be investigated and compared to these initial salt responses. Further molecular-level research is needed to establish more precise conclusions and global understanding of salinity tolerance in brassicas.

#### AUTHOR CONTRIBUTIONS

IP and BS-S designed the research. IP performed the salinity stress experiments, analyzed the levels of stress hormones, conducted the statistical and principle component analyses, and designed the figures. BS-S performed the root-growth bioassays and biomass production experiments. JO and DT performed the brassinosteroid instrumental and data analyses. ON and MS supervised all other hormonal measurements and data analysis. VVB and SRB performed the stress diagnostic experiments and data analysis. HL performed the photosynthesis measurements.

# REFERENCES


SM was responsible for analysis of photosynthesis data, statistics, and figure design. IP and SM drafted the manuscript. All authors discussed the results and implications, edited the manuscript, and approved the final manuscript.

# FUNDING

This work was supported by the Croatian Science Foundation (Project No. IP-2014-09-4359), the Ministry of Education, Youth and Sports of the Czech Republic (European Regional Development Fund-Project "Plants as a tool for sustainable global development" No. CZ.02.1.01/0.0/0.0/16\_019/0000827), and the Czech Science Foundation (Project No. 17-06613S).

#### ACKNOWLEDGMENTS

We thank Branimir Urlic, Ph.D., and MS Mara Bogovi ´ c for ´ providing kale and white cabbage seeds, respectively, used in this research and MS Ivan Petˇrik for assistance in performing phytohormone analysis.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00450/ full#supplementary-material


stress response in Arabidopsis. Plant Cell 25, 2132–2154. doi: 10.1105/tpc.113. 112896


basilicum. Environ. Exp. Bot. 130, 162–173. doi: 10.1016/j.envexpbot.2016. 06.004



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

Copyright © 2019 Pavlovi´c, Mlinari´c, Tarkowská, Oklestkova, Novák, Lepeduš, Vujˇci´c Bok, Radi´c Brkanac, Strnad and Salopek-Sondi. 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.

# Epitranscriptomic RNA Methylation in Plant Development and Abiotic Stress Responses

#### Jianzhong Hu† , Stefano Manduzio† and Hunseung Kang\*

Department of Applied Biology, College of Agriculture and Life Sciences, Chonnam National University, Gwangju, South Korea

Recent advances in methylated RNA immunoprecipitation followed by sequencing and mass spectrometry have revealed widespread chemical modifications on mRNAs. Methylation of RNA bases such as N 6 -methyladenosine (m6A) and 5-methylcytidine (m5C) is the most prevalent mRNA modifications found in eukaryotes. In recent years, cellular factors introducing, interpreting, and deleting specific methylation marks on mRNAs, designated as "writers (methyltransferase)," "readers (RNA-binding protein)," and "erasers (demethylase)," respectively, have been identified in plants and animals. An emerging body of evidence shows that methylation on mRNAs affects diverse aspects of RNA metabolism, including stability, splicing, nucleus-to-cytoplasm export, alternative polyadenylation, and translation. Although our understanding for roles of writers, readers, and erasers in plants is far behind that for their animal counterparts, accumulating reports clearly demonstrate that these factors are essential for plant growth and abiotic stress responses. This review emphasizes the crucial roles of epitranscriptomic modifications of RNAs in new layer of gene expression regulation during the growth and response of plants to abiotic stresses.

#### Edited by:

Sang Yeol Lee, Gyeongsang National University, South Korea

#### Reviewed by:

Andreas Bachmair, University of Vienna, Austria Byungho-Ho Kang, The Chinese University of Hong Kong, China

> \*Correspondence: Hunseung Kang hskang@jnu.ac.kr

†These authors have contributed equally to this work

#### Specialty section:

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

Received: 14 October 2018 Accepted: 01 April 2019 Published: 17 April 2019

#### Citation:

Hu J, Manduzio S and Kang H (2019) Epitranscriptomic RNA Methylation in Plant Development and Abiotic Stress Responses. Front. Plant Sci. 10:500. doi: 10.3389/fpls.2019.00500 Keywords: abiotic stress, epitranscriptome, RNA metabolism, RNA methylation, RNA modification

# INTRODUCTION

Epigenetic regulation of gene expression via DNA methylation and histone modifications is an important strategy for living organisms to achieve fine-tuned regulation of developmental processes or responses to environmental cues. Similar to DNA methylation in epigenetic regulation, posttranscriptional RNA modifications are emerging as important "epitranscriptomic" regulatory networks in recent years (Saletore et al., 2012; Meyer and Jaffrey, 2014). Over 150 different chemical modifications on mRNAs, tRNAs, and rRNAs are currently known for all kingdoms of life (Cantara et al., 2010; Boccaletto et al., 2018). Among diverse modifications found on mRNAs, N6-methyladenosine (m6A) is the most prevalent modification in both plants and animals (Liu and Pan, 2016; Covelo-Molares et al., 2018). Recent advances in methylation RNA sequencing (Met RNA-seq) and deep RNA sequencing have revealed transcriptome-wide m6A methylation patterns in plants as well as in animals (Luo et al., 2014; Wang et al., 2015a; Cui et al., 2017). These modifications within mRNAs can affect multiple steps of transcript's fate, including splicing (Haussmann et al., 2016; Xiao et al., 2016), nucleus-to-cytoplasm export (Zheng et al., 2013), RNA turnover (Du et al., 2016; Mauer et al., 2017; Wei et al., 2018), and translation (Meyer et al., 2015; Wang et al., 2015b; Choi et al., 2016).

The level and status of RNA methylation in cells depend on two crucial proteins: RNA methyltransferase (MT) designated as "writer" and RNA demethylase (DMT) designated as "eraser" (**Figure 1**). In addition to these two essential proteins required for the addition and removal of methyl groups on RNAs, a third protein designated as "reader" is involved in the recognition and processing of methylated RNAs (reviewed in Meyer and Jaffrey, 2014). In animals, genes encoding m6A writer (Ping et al., 2014; Schwartz et al., 2014), reader (Luo and Tong, 2014; Xu et al., 2014), and eraser (Jia et al., 2011; Zheng et al., 2013) proteins have been identified and characterized (**Table 1**). Notably, mutants lacking specific m6A writer, reader, or eraser have displayed abnormal development and altered response to hypoxia and high temperatures, suggesting crucial roles of RNA methylation in animal development and adaptation to changing environmental cues (reviewed in Meyer and Jaffrey, 2014; Yue et al., 2015; Hsu et al., 2017).

Although these recent studies clearly point to the importance of RNA methylation and essential roles of writers, readers, and erasers in the development of animals, functions of these proteins in plants are just beginning to be uncovered. Arabidopsis contains functional orthologs of m6A writer complex components, erasers,

FIGURE 1 | Roles and structural characteristics of m6A RNA methylation-related proteins. (A) Cellular factors introducing, deleting, and interpreting m6A marks are methyltransferase ("writers"), demethylase ("eraser"), and RNA-binding protein ("reader"), respectively. (B) The writer complex consists of five components: MTA/B (methyltransferase A/B), FIP37 (FKBP12 interacting protein 37), VIR (Virilizer), HAKAI (for "destruction" in Japanese, a c-Cb1-like protein), erasers belong to AlkB-homology (ALKBH) family proteins, and readers are YT512-B homology domain (YTHD) proteins. Numbers at the C terminus indicate the number of amino acid residues in each Arabidopsis protein. MT\_A70, methytransferase\_A70; FE2OG\_OXY, Fe2<sup>+</sup> 2-oxoglutarate dioxygenase domain; WTAP, WT1-associated protein.

and reader proteins, some of which have been found to play essential roles in normal plant development (Bodi et al., 2012; Shen et al., 2016; Ruži ˚ cka et al., 2017 ˇ ; Arribas-Hernández et al., 2018; Scutenaire et al., 2018; Wei et al., 2018). All these aforementioned studies have emphasized the essential roles of RNA methylation in plant development. However, the identity and functions of most writers, readers, and erasers in plants are currently unclear. In this review, we systematically identified potential m6A writers, readers, and erasers in Arabidopsis and rice (Oryza sativa) by comparing sequence homology to animal counterparts. We also reviewed multiple functions and potential significance of m6A RNA methylation in the development and response of plants to diverse abiotic stresses.

#### DIVERSE MODIFICATIONS FOR EUKARYOTIC RNAS

Over 150 different internal modifications on RNAs have been identified (Cantara et al., 2010; Boccaletto et al., 2018), with different degree, topology, and kinds of modifications between mRNAs, tRNAs, and rRNAs. For instance, approximately 17% of total nucleotides in tRNAs are modified, whereas only 2% of nucleotides in rRNAs are modified (Jackman and Alfonzo, 2013). Among diverse modifications identified for tRNAs and rRNAs, 2 0 -O-ribose methylation and pseudouridilation of rRNAs and 5 methylcytosine (m5C) and 1-methylguanidine (m1G) of tRNAs are the most abundant (Chou et al., 2017). Despite emerging roles of mRNA modifications in its processing and function, mRNA is less densely modified compared to tRNAs and rRNAs (Gilbert et al., 2016). Only a handful of different methylations have been identified so far in mRNAs, with N6-methyladenosine (m6A) being the most abundant (Liu and Pan, 2016; Covelo-Molares et al., 2018). These methylations of bases can influence the structure of RNAs by increasing its hydrophobicity and disrupting the canonical Watson-Crick base pairing (Oerum et al., 2017; Väre et al., 2017).

Importantly, all organisms have evolved to cluster methylation marks in functionally critical positions rather than randomly distributing them along RNA molecules. Most of these modified bases in rRNAs are located at the interface between ribosomal large and small subunits corresponding to P-site and A-site (Sharma and Lafontaine, 2015; Sloan et al., 2017). Wobble positions 34 and 37 of the anticodon loop in tRNAs are the most frequently and diversely modified (Väre et al., 2017). These conserved modification patterns reflect the essential role of RNA methylation in ribosome structure and biogenesis, codon recognition and decoding, and translation initiation or elongation (Jackman and Alfonzo, 2013; Chou et al., 2017; Sloan et al., 2017; Väre et al., 2017). Similar to rRNAs and tRNAs, mRNAs are also methylated in specific regions. For instance, m6A maps preferentially to the transcription start site, the stop codon, and the 3<sup>0</sup> UTR (Dominissini et al., 2012; Luo et al., 2014; Meyer and Jaffrey, 2014), while m5C is predominantly found in 3<sup>0</sup> UTR and coding regions (Squires et al., 2012; David et al., 2017). Several studies have shown that m1A methylation is frequently found in the start codon and the first splicing

site which influences translation (Dominissini et al., 2016; Safra et al., 2017). Clearly, the degree, topology, and non-random distribution of RNA modifications are crucial for its specific cellular functions.

### WRITERS, ERASERS, AND READERS INVOLVED IN m6A RNA METHYLATION AND RECOGNITION

#### Writers

Genes encoding m6A writer complexes have been identified and characterized firstly in animals. Several proteins including methyltransferase-like 3 (METTL3) and METTL14, Wilms' tumor 1-associating protein (WTAP), and Vir like m6A methyltransferase-associated (VIRMA; KIAA1429) are known to form multicomponent m6A writer complexes in animals (Shah and Clancy, 1992; Ping et al., 2014; Schwartz et al., 2014; **Table 1**).

Methyltransferase-like 3 is the principal enzyme exerting methyltransferase activity, while METTL14 has a supporting role forming a METTL3-METTL14 heterodimer (Sledz and Jinek, 2016; Wang et al., 2016). After the identification of METTL3 in mammals as a homolog of yeast methyltransferase IME4 (Shah and Clancy, 1992), its orthologs were identified in different species including Arabidopsis and Drosophila. At present, Arabidopsis orthologs of animal m6A writer components have been identified, including MTA (ortholog of METTL3) and MTB (ortholog of METTL14).

TABLE 1 | List of writers, readers, and erasers involved in RNA methylation in Arabidopsis thaliana and rice (Oryza sativa).


Wilms' tumor 1-associating protein functions as a stabilizer for the heterodimer localized in nuclear speckle (Ping et al., 2014; Lence et al., 2016). VIRMA plays a role in guiding the methyltransferase complex to the selective target region of mRNAs (Niessen et al., 2001; Yue et al., 2018). Arabidopsis VIR and FIP37 were identified as a ortholog of VIRMA and WTAP, respectively (Zhong et al., 2008; Bodi et al., 2012; Shen et al., 2016; Ruži ˚ cka et al., 2017 ˇ ).

Recently, zinc finger CCCH domain-containing protein 13 (ZC3H13), the latest component of methyltransferase complex, was found to be essential for localization of methyltransferase complex in mammals and Drosophila (Guo et al., 2018; Knuckles et al., 2018). However, the existence and molecular function of ZC3H13 in plants remain unknown. Interestingly, Arabidopsis contains E3 ubiquitin ligase HAKAI as an additional m6A writer component (Ruži ˚ cka et al., ˇ 2017; **Table 1**). Although knockdown of its expression can decrease m6A level (Ruži ˚ cka et al., 2017 ˇ ), the primary role of HAKAI in methyltransferase complexes has yet to be investigated.

#### Erasers

Removal of methylation marks on RNAs is carried out by α-ketoglutarate-dependent dioxygenase (AlkB) homolog (ALKBH) proteins that can erase alkyl and methyl groups from DNAs, RNAs, and proteins (Fedeles et al., 2015; Alemu et al., 2016). Mammals have nine ALKBH family members: ALKBH1 to ALKBH8 and fat mass- and obesityassociated protein (FTO) (Ougland et al., 2015; **Table 1**). Although ALKBH2 and ALKBH3 have been identified as main DNA repair enzymes, ALKBH3 also shows activity on m1A and m3C of RNAs (Ueda et al., 2017). Interestingly, ALKBH1 acts on a wide range of substrates in DNAs, RNAs, and histones (Westbye et al., 2008; Ougland et al., 2012; Wu et al., 2016). In addition to its role in cytoplasm, human ALKBH1 targets several m1A methylated tRNAs in mitochondria, influencing the organellar translation and function (Kawarada et al., 2017; Müller et al., 2018). ALKBH8, another tRNA DMT, interestingly contains both methyltransferase and demethylase domains, unlike other family members (Pastore et al., 2012).

Only two m6A erasers, ALKBH5 and FTO, have been identified in animals so far. Both enzymes were originally shown to be involved in demethylation of m6A (Jia et al., 2011; Zheng et al., 2013). However, recent studies have suggested that FTO has a much higher activity toward N<sup>6</sup> , 2<sup>0</sup> -Odimethyladenosine (m6Am) compared to that for m6A (Meyer and Jaffrey, 2017; Mauer et al., 2017; Mauer and Jaffrey, 2018). ALKBH5 and FTO have been found to be involved in alternative splicing, 3<sup>0</sup> -UTR processing, mRNA stability, translation, and amino-acids deprivation response pathway (Zheng et al., 2013; Zhao et al., 2014; Bartosovic et al., 2017; Tang et al., 2018). Arabidopsis contains several putative m6A eraser ALKBH family proteins (**Table 1**), among which only two eraser proteins ALKBH9B and ALKBH10B have been functionally characterized in viral infection and floral transition (Duan et al., 2017; Martínez-Pérez et al., 2017). In summary, although increasing number of erasers targeting specific methylation marks have been identified, the activity and substrate RNAs of most ALKBH family members in plants and animals are yet to be determined.

#### Readers

Interpretation of methylation marks is tightly related to posttranscriptional regulation of mRNA metabolism which requires reader proteins to recognize methylated transcripts and ultimately determine their fates. In recent years, several RNA-binding proteins (RBPs) that can recognize m6A marks on mRNAs have been identified in animals by RNA-protein immunoprecipitation using synthetic m6A-containing RNAs (Dominissini et al., 2012; Xu et al., 2014; Arguello et al., 2017; Edupuganti et al., 2017; Wu et al., 2017). YT521-B homology (YTH) domain family (YTHDF) protein was first identified as an m6A-binding protein (Xu et al., 2014). Recently, human and mouse YTHDF proteins including YTHDF1, YTHDF2, YTHDF3, YTHDC1, and YTHDC2 were found to possess a specific binding pocket for m6A nucleotides and exhibit significantly high affinity to methylated RNAs, suggesting their role as m6A readers (Dominissini et al., 2012; Hsu et al., 2017; Xiang et al., 2017; Liao et al., 2018). YHHDF2 can bind to m6A-modified RNAs and play a distinct role in mRNA degradation by recruiting the CCR4- NOT deadenylase complex (Wang et al., 2014; Zhou et al., 2015; Du et al., 2016). YTHDF1 was found to recognize the 50UTR of m6A-modified mRNAs in the cytosol, which promotes translation of target transcripts in a cap-independent manner (Wang et al., 2015b; Shi et al., 2017). YTHDC1 is involved in exon-selective gene splicing in the nucleus (Xiang et al., 2017). Interestingly, YTHDC2 also contains RNA helicase domain (Jain et al., 2018). Arabidopsis and rice genomes encode 13 and 12 YTHD proteins, respectively (Li et al., 2014a; **Table 1**). Contrary to extensive study on YTHD proteins in animals, only three evolutionarily conserved c-terminal region (ECT) family proteins have recently been functionally characterized in Arabidopsis as YTHD homologs (Arribas-Hernández et al., 2018; Scutenaire et al., 2018; Wei et al., 2018; **Table 1**).

Besides YTHD proteins, two other proteins containing different RNA-binding domains that can recognize m6A marks in animal cells have been reported. One is a heterogeneous nuclear ribonucleoprotein A2/B1 (HNRNPA2B1) which regulate RNA splicing in the nucleus through a well-characterized RNArecognition motif (Alarcon et al., 2015). Notably, instead of directly binding to m6A site as YTHD proteins, HNRNPA2B1 might bind to altered structures right after the m6A site (Alarcon et al., 2015). Insulin-like growth factor 2 mRNAbinding protein (IGF2BP) contains tandem K-homology (KH) domains to recognize m6A sites and enhance target mRNA stability, storage, and translation in an m6A-dependent manner (Nicastro et al., 2015; Huang et al., 2018). Eukaryotic initiation factor 3 (eIF3) can also promote translation of mRNAs depending on m6A modification (Meyer et al., 2015). Clearly, more reader proteins recognizing other RNA modifications as well as m6A marks should be uncovered to fully understand

cellular roles of epitranscriptomic RNA modifications in both plants and animals.

#### RNA METHYLATION IN ANIMAL DEVELOPMENT AND DISEASES

m6A methylation has been demonstrated to affect all fates of mRNA metabolism, including pre-mRNA processing and intron splicing in the nucleus, nucleus-to cytoplasm RNA export, translation, and RNA decay in the cytoplasm (**Figure 2**). Analysis of different mettl mutants demonstrated the essential role of m6A methylation in cell development, proliferation, differentiation, and motility by regulating mRNA stability and splicing pattern of diverse transcripts (Wang et al., 2014; Chen et al., 2015; Geula et al., 2015; Park and Hong, 2017; Widagdo and Anggono, 2018). Loss of FTO can inhibit differentiation of primary myoblasts and skeletal muscle in mice, suggesting that m6A demethylase FTO plays a crucial role in somatic and neural stem cell differentiation (Wang et al., 2017). A larger number of gene encoding clock genes and clock output genes are enriched in m6A methylation (Fustin et al., 2013; Hastings, 2013) and changes in m6A levels can affect circadian rhythms, cellular growth, and survival (Fustin et al., 2018).

Notably, recent studies have demonstrated that alteration in m6A levels is closely associated with various diseases, especially cancer (reviewed in Dai et al., 2018; Pan et al., 2018). For example, FTO affects m6A level and translation of Angptl4 mRNA, which regulates fatty acid mobilization in adipocytes and body weight (Wang et al., 2015a). Low m6A level in total RNA is related to type 2 diabetes mellitus (Shen et al., 2014). Considering that aberrant cell growth and differentiation cause cancer, it is worth noting that cancer cells may improve their survival rate and progression by modulating aberrant methylation of target RNAs. Several studies have shown that expression of FTO or ALKBH5 can decrease m6A level, resulting in enhanced cancer cell growth (Zhang et al., 2016; Li et al., 2017; Zhang et al., 2017). METTL3 acts as an oncogene in cancer cells, enhancing the translation of cancer-inducing genes by interacting with translation initiation factor (Lin et al., 2016).

In addition to m6A methylation, m5C is also involved in cell development and diseases. This modification is deeply associated

FIGURE 2 | Diverse cellular processes affected by m6A RNA methylation. Splicing of mRNAs in the nucleus and diverse RNA metabolism in cytoplasm, including cap-dependent and cap-independent translation, RNA decay in cytosol and P-body, and RNA storage, is affected by m6A RNA methylation. Specific "reader" proteins recognizing m6A marks on mRNAs play essential roles in these cellular processes. Writers (MTA, MTB, FIP37, VIR, and HAKAI), erasers (ALKBH9B/10B), and reader (YTH09) identified in Arabidopsis are shown.

with testis differentiation and tumor cell proliferation. A previous study has shown that NOP2/sun RNA methyltransferase family member 2 (NSUN2), an m5C writer, is highly expressed in tumor cells and its depletion decreases levels of Ddx4, Miwi, and Tudor domain-containing proteins, suggesting an essential role of m5C RNA methylation in male germ cell differentiation (Frye and Watt, 2006). Moreover, loss of NSUN2 causes an accumulation of progenitors, decreases in upper-layer neurons, and increases in tRNA fragment accumulation in the brain, resulting in damage to neural stem cell differentiation and motility (Flores et al., 2017). Although these studies clearly demonstrate the importance of m6A and m5C in cell proliferation and diseases, biological functions of other RNA methylations in animal development and pathogenesis are yet to be elucidated.

### RNA METHYLATION IN PLANT DEVELOPMENT AND ABIOTIC STRESS RESPONSES

Although our understanding of writers, readers, and erasers in plants is far behind their animal counterparts, accumulating reports clearly demonstrate that these factors are essential for plant growth and abiotic stress responses. Herein, we will summarize and discuss characterized and potential writers, readers, and erasers (**Table 1**) in plants.

### m6A Writers

Genome-wide m6A methylation patterns have been mapped in barley, Arabidopsis, and rice (Li et al., 2014b; Luo et al., 2014). However, key enzymes responsible for this methylation have only been studied in Arabidopsis. Analysis of mta (Arabidopsis ortholog of human METTL3) knockdown mutants has revealed that MTA is required for m6A mRNA methylation and essential for normal growth and development, such as shoot and root growth as well as leaf and floral development (Zhong et al., 2008; Bodi et al., 2012). Moreover, MTA was found to interact with MTB, an Arabidopsis ortholog of human METTLl4. Knockdown of MTB showed a similar but less severe phenotype compared to mta mutants, indicating that both writers are essential for plant development (Ruži ˚ cka et al., 2017 ˇ ). The Arabidopsis m6A writer complex also includes an ortholog of human WTAP named FIP37. Depletion of FIP37 results in embryo lethality while its partial loss causes huge overproliferation of shoot meristems by increasing the stability of shootmeristemless (STM) and WUSCHEL (WUS) (Shen et al., 2016). Vir and Hakai are other m6A writer components in Arabidopsis. They affects root and shoot growth as well as cotyledon development, similar to other m6A writer mutant phenotypes (Ruži ˚ cka et al., 2017 ˇ ). However, the molecular mechanism underlying Vir and Hakai functions is yet to be elucidated.

Despite increasing understanding of the roles of m6A writers in plant growth and development, reports demonstrating their involvement and functions in plant response to abiotic stresses are lacking. Our analysis of publically available microarray data using GENEVESTIGATOR revealed that expressions levels of writers in Arabidopsis and rice are differently modulated by diverse abiotic stresses (**Figure 3**). In Arabidopsis, levels of most m6A writer components were not significantly modulated by abiotic stresses. Levels of MTA and FIP37 were only marginally increased by cold and heat stress, respectively. In rice, the level of OsFIP was increased by cold stress whereas levels of OsMTA, OsMTB, and OsVIR were decrease by cold, drought, or salt stress. The constant expression of m6A writer components under normal and stress conditions suggests the fundamental role of m6A methylation in plant development and stress responses.

# m5C Writers

Although m5C methylation in DNA has been studied for many years, its cellular and molecular functions in RNAs is just beginning to be uncovered. Due to advancement in RNA sequencing, m5C RNA methylation could be mapped to mRNAs in both animals and plants (Schaefer et al., 2009; Hussain et al., 2013; Song et al., 2018). Overall, m5C RNA methylation is a less abundant modification of mRNA than m6A methylation. In Arabidopsis, two enzymes, TRM4A and TRM4B, are responsible for m5C RNA methylation. Both enzymes are orthologs of human m5C methyltransferase NSuns2. However, TRM4A contributes to tRNA m5C methylation while TRM4B targets mRNA for m5C methylation. Loss of TRM4A does not exhibit any visible phenotype while loss of Trm4B reduces root length, suggesting the role of mRNA m5C methylation in root development regulation (David et al., 2017). In accordance to this, loss of m5C RNA methylation affects the stability of short hypocotyl 2 (SHY2) and indoleacetic acid-induced protein 16 (IAA16), two critical genes related to root development (Cui et al., 2017). Our analysis showed that expression levels of Arabidopsis TRM4B are marginally increased by cold stress, although they decrease under heat stress. In contrast, expression levels of rice TRM4A and TRM4B are not altered in response to abiotic stresses (**Figure 3**). Although these expression patterns suggest potential roles of m5C writers in abiotic stress response, the relevance of m5C methylation to abiotic stress responses awaits further investigation.

# m6A Erasers

Among protein factors involved in RNA methylation in plants, erasers are so far the least studied, although new knowledge is gained rapidly. Thirteen Arabidopsis ALKBH family members have been identified by bioinformatic analysis (Mielecki et al., 2012). However, only a few of them have been studied so far (**Table 1**). Among them, ALKBH9A, 9B, 9C, 10A, and 10B show the highest amino acid sequence similarity with human ALKBH5. Other family members are numbered based on their sequence similarity to human orthologs (**Table 1**). Like animal counterparts, most erasers are localized in the nucleus and cytoplasm whereas ALKBH1D is also present in chloroplasts. Interestingly, some of them show relocalization to the nucleus in response to methylating agents (Mielecki et al., 2012). ALKBH10B was identified as the principal mRNA m6A eraser influencing floral transition by controlling transcript levels of SPL3, SPL9, and FLOWERING LOCUS T (Duan et al., 2017). Another demethylase, ALKBH9B, was shown to revert m6A from single-stranded RNA in vitro (Martínez-Pérez et al., 2017).

Although alkbh9b knockout mutants do not show differences in plant RNA m6A methylation level (Duan et al., 2017), its depletion results in hypermethylation of alfalfa mosaic virus (AMV) RNA, mediating systemic infection by interacting with viral cap proteins (Martínez-Pérez et al., 2017).

Expression of ALKBH9A is highly induced in roots under salt stress but not in response to ABA (Ma et al., 2006). Its level is much lower than ALKBH9 and ALKBH10 under normal conditions (Duan et al., 2017). ALKBH10A is down-regulated by heat stress (Merret et al., 2015) whereas ALKBH10B is upregulated in response to karrikins (Nelson et al., 2010). Although these previous studies suggest a specific role of ALKBHs in stress responses as well as plant development, nothing is known about their actual roles. Our analysis showed that expression levels of ALKBH members were marginally up- or down-regulated in Arabidopsis by different abiotic stresses (**Figure 3**). Notably, levels of ALKBH1 in rice were highly increased upon drought, cold, or ABA treatment whereas expression levels of ALKBH6, ALKBH8B, and ALKBH10A were decreased by drought, ABA, or cold (**Figure 3**). These data suggest that ALKBHs could play important roles in abiotic stress responses, although this awaits further investigation.

#### m6A Readers

Although several RBPs interpreting m6A marks have been identified in animals, roles of only three YTHD m6A reader proteins have very recently been determined in Arabidopsis (Arribas-Hernández et al., 2018; Scutenaire et al., 2018; Wei et al., 2018). YTHD09 (ECT2) is involved in trichome development. Moreover, cytoplasmic-localized YTHD09 relocates to stress granules upon heat exposure, suggesting its role in mRNA fate control under stress conditions (Scutenaire et al., 2018). By using single and double mutants, it has been demonstrated that YTHD09 (ECT2), YTHD13 (ECT3), and ECT4 regulate the timing and execution of plant organogenesis (Arribas-Hernández et al., 2018). Moreover, a molecular study revealed that ECT2 targets a large number of m6A-containing transcripts, including TTG1, ITB1, and DIS2, which are involved in trichome development (Wei et al., 2018). Further sequencing analysis suggested that ECT2 increases the stability of these

FIGURE 3 | Heatmap showing stress-responsive expression patterns of writers, readers, and erasers in Arabidopsis and rice. Red or green colors represent upregulated and downregulated expression level, respectively. Microarray data were obtained from GENEVESTIGATOR, and expression levels of each gene under stress conditions were calculated relative to control levels.

transcripts and influences trichome development (Wei et al., 2018). Although these studies clearly point to important roles of YTHD readers in plant development, more in-depth and focused efforts are needed to identify and characterize potential reader proteins (**Table 1**) that can recognize not only m6A modification, but also other methylation marks in plants.

No reports demonstrating the involvement or functions of any RNA methylation reader proteins in plant response to abiotic stresses have been published so far. However, a previous study and our current analysis showed that the expression of YTHDs in Arabidopsis and rice is highly regulated by different abiotic stresses (Li et al., 2014a; **Figure 3**). In Arabidopsis, levels of YTHD05, YTHD06, and YTHD07 are increased by heat, cold, hypoxia, or submergence stress. In contrast, the expression level of YTHD10 decreases under cold, drought, salt, or osmotic stress whereas YTHD08 level is reduced by heat stress. In rice, YTHDs responded differently to various abiotic stresses (Li et al., 2014a; **Figure 3**). Expression levels of YTHD05, YTHD06, YTHD07, and YTHD09 are downregulated by cold stress whereas levels of YTHD03 and YTHD08 increase under submergence and heat stress, respectively. Notably, none of these rice YTHDs showed altered expression under salt stress whereas YTHD01, YTHD02, YTHD03, YTHD04, or YTHD08 does not respond to cold stress. The fact that m6A reader proteins respond more to abiotic stresses than writers and erasers suggests that decoding of methylation marks is much more important than introducing or removing these marks during stress adaptation process in plants. It would be interesting to characterize roles of reader proteins in RNA metabolism and its consequence in stress responses.

#### CONCLUDING REMARKS AND PERSPECTIVES

Chemical modifications of RNAs are invaluable ways to expand decoding capacity of RNA transcripts beyond genetic

#### REFERENCES


information inherent to genome sequences. They are crucial for posttranscriptional gene regulatory events such as mRNA splicing, stability, and translation. The ability to regulate the fate of RNA molecules through nucleotide modifications is vital to plant survival and fitness under adverse as well as favorable environmental conditions. Despite the increasing discovery of cellular components essential for chemical modification and decoding of modified RNA molecules, our knowledge regarding physiological roles of proteins involved in these processes is far from sufficient. Several key questions remain to be further investigated. Are there any other internal RNA modifications not identified so far? How these components are regulated depending on developmental stages and/or in response to changing environmental cues? What guides the specificity of interactions between these components with target transcripts? Are these components conserved between dicots and monocots, especially in crop species? Addressing these questions will greatly expand our knowledge on the process of chemical modifications of RNAs and its effects on plant survival and fitness under stressful conditions. Such studies could provide potential new targets for engineering crop plants with higher adaptability to adverse environmental conditions.

#### AUTHOR CONTRIBUTIONS

HK designed the concept. JH and SM compiled and analyzed data. JH, SM, and HK contributed to the writing of this review.

#### FUNDING

This work was supported by grants from the Next-Generation BioGreen21 Program (PJ01314701 and PJ01312201), funded by Rural Development Administration, Republic of South Korea.


translational regulation. Mol. Cell 68, 978–992. doi: 10.1016/j.molcel.2017. 11.002


fpls-10-00500 April 15, 2019 Time: 17:39 # 9


methylation at internal and 50 sites. Cell Rep. 8, 284–296. doi: 10.1016/j.celrep. 2014.05.048


fpls-10-00500 April 15, 2019 Time: 17:39 # 10

3-methylcytosine in DNA and RNA. J. Biol. Chem. 283, 25046–25056. doi: 10.1074/jbc.M803776200


ALKBH5-mediated m6A-demethylation of NANOG mRNA. Proc. Natl. Acad. Sci. U.S.A. 113, E2047–E2056. doi: 10.1073/pnas.1602883113


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

Copyright © 2019 Hu, Manduzio and Kang. 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.

# Acetic Acid Treatment Enhances Drought Avoidance in Cassava (Manihot esculenta Crantz)

Yoshinori Utsumi<sup>1</sup> , Chikako Utsumi1,2, Maho Tanaka1,3, Chien Van Ha1,2 , Satoshi Takahashi1,3, Akihiro Matsui1,3, Tomoko M. Matsunaga<sup>4</sup> , Sachihiro Matsunaga2,5 , Yuri Kanno<sup>6</sup> , Mitsunori Seo<sup>6</sup> , Yoshie Okamoto<sup>1</sup> , Erika Moriya<sup>1</sup> and Motoaki Seki1,2,3,7 \*

<sup>1</sup> RIKEN Center for Sustainable Resource Science, Yokohama, Japan, <sup>2</sup> Core Research for Evolutional Science and Technology, Japan Science and Technology, Kawaguchi, Japan, <sup>3</sup> RIKEN Cluster for Pioneering Research, Wako, Japan, <sup>4</sup> Research Institute for Science and Technology, Tokyo University of Science, Noda, Japan, <sup>5</sup> Department of Applied Biological Science, Graduate School of Science and Technology, Tokyo University of Science, Noda, Japan, <sup>6</sup> Dormancy and Adaptation Research Unit, RIKEN Center for Sustainable Resource Science, Yokohama, Japan, <sup>7</sup> Kihara Institute for Biological Research, Yokohama City University, Yokohama, Japan

#### Edited by:

Vasileios Fotopoulos, Cyprus University of Technology, Cyprus

#### Reviewed by:

Yasunari Fujita, Japan International Research Center for Agricultural Sciences, Japan Jitendra Kumar, University of Allahabad, India Fidele Tugizimana, University of Johannesburg, South Africa

> \*Correspondence: Motoaki Seki motoaki.seki@riken.jp

#### Specialty section:

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

Received: 30 November 2018 Accepted: 04 April 2019 Published: 24 April 2019

#### Citation:

Utsumi Y, Utsumi C, Tanaka M, Ha CV, Takahashi S, Matsui A, Matsunaga TM, Matsunaga S, Kanno Y, Seo M, Okamoto Y, Moriya E and Seki M (2019) Acetic Acid Treatment Enhances Drought Avoidance in Cassava (Manihot esculenta Crantz). Front. Plant Sci. 10:521. doi: 10.3389/fpls.2019.00521 The external application of acetic acid has recently been reported to enhance survival of drought in plants such as Arabidopsis, rapeseed, maize, rice, and wheat, but the effects of acetic acid application on increased drought tolerance in woody plants such as a tropical crop "cassava" remain elusive. A molecular understanding of acetic acid-induced drought avoidance in cassava will contribute to the development of technology that can be used to enhance drought tolerance, without resorting to transgenic technology or advancements in cassava cultivation. In the present study, morphological, physiological, and molecular responses to drought were analyzed in cassava after treatment with acetic acid. Results indicated that the acetic acid-treated cassava plants had a higher level of drought avoidance than water-treated, control plants. Specifically, higher leaf relative water content, and chlorophyll and carotenoid levels were observed as soils dried out during the drought treatment. Leaf temperatures in acetic acid-treated cassava plants were higher relative to leaves on plants pretreated with water and an increase of ABA content was observed in leaves of acetic acidtreated plants, suggesting that stomatal conductance and the transpiration rate in leaves of acetic acid-treated plants decreased to maintain relative water contents and to avoid drought. Transcriptome analysis revealed that acetic acid treatment increased the expression of ABA signaling-related genes, such as OPEN STOMATA 1 (OST1) and protein phosphatase 2C; as well as the drought response and tolerance-related genes, such as the outer membrane tryptophan-rich sensory protein (TSPO), and the heat shock proteins. Collectively, the external application of acetic acid enhances drought avoidance in cassava through the upregulation of ABA signaling pathway genes and several stress responses- and tolerance-related genes. These data support the idea that adjustments of the acetic acid application to plants is useful to enhance drought tolerance, to minimize the growth inhibition in the agricultural field.

Keywords: ABA, acetic acid, cassava, drought avoidance, drought response

# INTRODUCTION

fpls-10-00521 April 22, 2019 Time: 17:40 # 2

Abiotic stresses are one of the crucial constraints to food security and crop production. Among them, heat and drought are the two most important stresses with a negative impact on growth and productivity of crops (Fahad et al., 2017). Due to increasing demands for greater food production, it is important to develop desirable and strategic crops, such as cassava, to increase food productivity (Kamanga et al., 2018; Varshney et al., 2018). Drought tolerance is an important research subject pertaining to cassava, as climate change has raised concerns about global drought problems and also places significant demands on breeding programs (Aina et al., 2007; De Oliveira et al., 2017).

Cassava is a tropical crop that serves as an important source of food and industrial materials. It is a vital economic resource for poor farmers in marginal areas where its production is constrained by drought. Although cassava is one of the most drought-tolerant crops, the underlying mechanism for its ability to survive under drought has been investigated recently. For example, physiological association among cassava varieties to drought has been evaluated using plants on field condition and tissue cultures in several countries (Chemonges et al., 2013; Oliveira et al., 2015; Jolayemi and Opabode, 2018). The droughtresponsive genes and proteins have been identified in order to understand the physiological mechanisms of drought survival (Utsumi et al., 2012; Turyagyenda et al., 2013; Zhao et al., 2015; Fu et al., 2016). Cassava plants can activate a combination of drought avoidance and tolerance mechanisms that help to maintain optimum growth, development, and metabolism (El-Sharkawy, 2007). Increasing the growth efficiency and survivability of cassava plants subjected to drought, could help the plant survive in marginal growing areas that are subject to stress conditions (Okogbenin et al., 2013).

A variety of different approaches are currently being used to minimize the negative effects of abiotic stresses on plants. Application of various stress priming agents has been presented as a potential strategy to increase biotic and abiotic stress tolerance in plants (Savvides et al., 2016). Recent studies have confirmed that the application of chemical stress priming agents represents a promising approach to manage plant stress adaptation under field conditions by activating their innate stress-adaptive mechanisms (Kinoshita and Seki, 2014).

The application of exogenous methyl jasmonate has been reported to enhance abiotic stress tolerance in soybean (Yoon et al., 2009). The application of exogenous strigolactone enhanced drought tolerance in Arabidopsis (Ha et al., 2014). The application of salicylic acid was shown to be beneficial for plants either in optimal or stress environments, through the regulation of various plant metabolic processes and through the modulation of the production of varied osmolytes, secondary metabolites and the plant-nutrient status, to protect plants under abiotic stress conditions (Khan et al., 2015). Chemicals such as ethanol, mandipropamid, melatonin, polyamines, and sodium nitroprusside enhanced abiotic stress tolerance in plants (Park et al., 2015; Savvides et al., 2016; Nguyen et al., 2017). The epigenetic inhibitors such as Ky-2, Ky-9, and Ky-72 increased salinity stress avoidance in Arabidopsis (Sako et al., 2016; Nguyen et al., 2018a,b). The in vitro cassava plants enhanced salinity stress tolerance by suberoylanilide hydroxamic acid (SAHA) treatment (Patanun et al., 2017). Recently, the application of acetic acid was also reported to enhance drought tolerance in a variety of plant species, including Arabidopsis, rice, maize, rapeseed, and wheat by activating the JA-signaling pathway (Kim et al., 2017). However, the effects of acetic acid application on increased drought tolerance in woody plants such as a tropical crop "cassava" remain elusive. In the present study, physiological- and molecular-responsive mechanisms were analyzed to elucidate the effect of acetic acid on cassava plants subjected to drought. Results of the study indicated that the acetic acid treatment increased drought avoidance in cassava. This was accomplished by the ability of acetic acid to regulate the rate of transpiration and maintain the water content in cassava leaves.

#### MATERIALS AND METHODS

#### Cassava Cultivar and Plant Preparation

The African cassava cultivar "60444 (Manihot esculenta Crantz)" was obtained from the International Institute of Tropical Agriculture (IITA, Nigeria) in vitro cassava germplasm collection. The in vitro cassava plantlets were acclimated to ambient atmospheric conditions and were subsequently maintained under a greenhouse condition (50% humidity, 28◦C, and 16 h supplemental lighting). Cassava plants grow well under the greenhouse condition. Stem cuttings (approximately 3 cm) were obtained from individual plants and propagated. After reaching an approximate stem length of 15 cm from the soil surface, the plants were transferred to a plastic pot (7.9 diameter × 6 height cm) filled with vermiculite. After transfer, the plants were grown under a greenhouse condition for 2 weeks and then used in the drought experiment. The treatment with 10–20 mM acetic acid solution was effective in rice, maize, wheat, rapeseed and Arabidopsis (Kim et al., 2017). In this study, the treatment with 10 mM acetic acid solution was performed in cassava plants because wilting leaves were observed by an application of 20, 30-, and 50-mM acetic acid solution. Also, to examine the effect of lower concentration of acetic acid on cassava plants, we also evaluated 1 mM acetic acid-treated plants with regarding to the drying test and measurement of net photosynthesis rate. The plants were watered with acetic acid or plain water (control) for 7 days, and then exposed to a drought for 14 days under a greenhouse condition to remove all water from soil pot. The phenotype of wilting leaves can be observed during 14 days of drying.

#### Quantification of Leaf Wilting

Cassava plants without the application of acetic acid were placed on a rotating table, and images of the plant were taken from 360 degrees to quantify the extent of leaf wilting due to the imposed drought. Each individual cassava plant had 6–10 leaves with leaves numbered in order from the top, e.g., 1, 2,. . . , 10 (**Supplementary Figure S1A**). An image of each leaf in which the petiole was parallel to the camera was selected and

analyzed. For each leaf, the midrib line was drawn between the base and the midpoint of the midrib (**Supplementary Figures S1B,C**). The angle made by the midrib line of the central leaflet with the vertical axis was measured using ImageJ software and used as an indicator of the level of drought or wilting (**Supplementary Figures S1D,E**).

#### Determination of Fresh and Dry Weight, and Relative Water Content in Waterand Acetic Acid-Treated Plants

Leaves of stressed and non-stressed plants were collected separately and used to measure fresh weight and dry weight, and relative water content (RWC) using a previously described protocol (Nishiyama et al., 2011).

### Determination of Chlorophyll and Carotenoid Content

The fifth leaf was selected for the measurement of relative water and chlorophyll contents to determine the effect of acetic acid treatment on drought. Because the RWC in cassava leaves was decreasing from the bottom to the top of stem (**Figure 1A**). The chlorophyll and carotenoid in approximately 0.5 g fresh weight of the fifth leaf of cassava plants were extracted by shaking (200 rpm) them overnight in the dark in 30 mL of an 80% acetone solution. Subsequently, 1.0 mL of the extracted solution was used to measure absorbance of chlorophyll at 645 nm and 663 nm for carotenoid at 470 nm in a spectrophotometer. Chlorophyll and carotenoid levels were calculated as described by Mostofa et al. (2015).

#### Thermal Imaging

Thermal images were captured using an R500EX-S infrared camera equipped with a standard lens (Nippon Avionics Co., Ltd.). To measure the leaf temperature in narrow range, the camera was mounted vertically at approximately 60 cm above the leaf canopy for observations. To observe the leaf temperature in broad range, the camera adjusted with angle of approximately 50 degree was set an approximately 100 cm above leaf canopy. Thermal images were stored by interval of a minute and subsequently analyzed for temperature determination on a custom python script.

#### Total RNA

For total RNA preparation, cassava plants were grown and treated with 10 mM acetic acid or water (control) for 7 days. The first to third leaves from the top of stems were collected and frozen in liquid nitrogen for total RNA purification. Frozen leaves from water- and acetic acid-treated plants were pulverized in liquid nitrogen using a Multi Beads Shocker system (Yasuikiki). Total RNA was extracted from 100 mg of leaves using the method described by Utsumi et al. (2017). Total RNA samples were subsequently purified using a Plant RNA Reagent (ThermoFisher) according to the manufacturer's instructions, and samples of total RNA were treated with DNase I (Takara), and an RNase inhibitor, for 30 min at 37◦C to eliminate genomic DNA contamination in the samples. Samples of total RNA were purified using a RNeasy Plant Mini Kit (QIAGEN) according to the manufacturer's instructions and RNA quality was evaluated by electrophoresis using a Bioanalyzer (Agilent). After extraction, the total RNA samples were stored at −80◦C until further use.

#### Oligo-Microarray Analysis of Gene Expression and Statistical Analyses

Total RNA was used to evaluate gene expression using a cassava DNA oligo-microarray that included more than 30,000 probes as described by Utsumi et al. (2016). Leaves were collected from four separate water- and four 10 mM acetic acid-treated cassava plants and treated as one biological replicate. The presented gene expression data were collected

from three independent biological replicates. A total of six expression microarray data sets were analyzed using GeneSpring GX (Agilent Technologies, United States). A 75 percentile normalization of the expression level was performed on all six samples. The normalized signal intensity was transformed into a log<sup>2</sup> ratio for display and analysis. The normalized signal intensity of transcripts from each sample group (experiment) was used in the statistical analysis. An analysis of variance test was conducted to determine the effect of treatment (water treatment vs. 10 mM acetic acid treatment). Changes in gene expression were statistically analyzed using an unpaired t-test for the two treatment groups. A false discovery rate (q-value) was calculated by a Westfall–Young multiple testing correction based on an unpaired t-test. The information from the oligo-DNA microarray was deposited in the Gene Expression Omnibus (GEO) at NCBI. The accession numbers are: Platform, GPL22197; Series, GSE122140 Manihot esculenta microarray; Samples, GSM3455975-GSM3455980.

#### Statistical Analyses

Physiological and biochemical data were analyzed using a oneway ANOVA and differences among means were analyzed by Scheffé's method (p < 0.05) (StatPlus 5 pro AnalystSoft Inc., United States) or a Duncan's multiple range test (P < 0.05) (using IBM SPSS software package 21.0).

#### Phytohormone Measurements

Endogenous ABA, JA, and JA-Ile were extracted with 80% (v/v) acetonitrile containing 1% (v/v) acetic acid from the first to third leaves from the top of stems treated with 10 mM acetic acid or water for 7 days after freeze-drying.

Hormone contents were determined using a UPLC-MS/MS system consisting of a quadrupole/time-of-fight tandem mass spectrometer (Triple TOF 5600, SCIEX, Concord, ON, Canada), and a Nexera UPLC system (Shimadzu Corp., Kyoto, Japan) by Kanno et al. (2016).

# Phylogenetic Analysis of Amino Acid Sequences

Homology between Arabidopsis and cassava on amino acid sequences was determined by BLASTX<sup>1</sup> where the query nucleotide sequences were aligned with homologous sequences retrieved from Phytozome v12.1.6<sup>2</sup> . Phylogenetic trees were generated by the neighbor-joining method using CLUSTAL\_W 2.0.12 software, and branch significance was analyzed by bootstrap with 1,000 replicates. Phylogenic trees were visualized using GENTYX-Tree 2.2.5 (Genetyx) software.

# RESULTS

# Relationship Between Angle of Leaf Wilting and Water Loss in Leaves

As the water from the soil in the potted cassava plants decreased, cassava leaves wilted and dropped from the stem. The RWC of leaves gradually declined from the top leaf downward after drying (**Figure 1A**). While the majority of the leaves face upward, however, the angle between the midrib and the vertical axis did not correctly indicate the leaflet angle in leaves which faced the front of the camera. Therefore, images in which the width of the leaflet exceeded 10% of the length of the midrib were excluded

<sup>1</sup>http://blast.st-va.ncbi.nlm.nih.gov/Blast.cgi

from the analyzed dataset. For example, the width of the leaflet in **Supplementary Figure S1D** is 5.2% of the midrib length, but 19.9% in **Supplementary Figure S1E**. In this case, the latter leaf was not used in the analyzed dataset. The uppermost leaves were too small to measure the angle accurately, and the lower leaves which wilted due to the drought often rolled up and the angle was also unable to be measured. Thus, only leaves 2 through 5 were used in the analysis (**Supplementary Figure S1A**). The angles of 30 leaves before and after imposed drought were analyzed. The angle between the midrib and the vertical axis decreased significantly in response to the drought. The average degree angle changed from 85.6 to 26.9 (**Figure 1B**).

# Effect of the Acetic Acid Treatment on Leaf Wilting and Leaf Water Status

To determine the effect of acetic acid treatment on drought in cassava, the fifth leaf was selected for the measurement of relative water and chlorophyll contents because the RWC in cassava leaves decreased from the bottom to the top of the stem (**Figure 1A**). Results indicated that the acetic acidtreated plants exhibited a less wilted leaf phenotype in response to the drought, relative to the water-treated, control plants (**Figure 2A**). To further investigate the response of acetic acidtreated cassava plants to drought, RWC was measured. As illustrated in **Figures 2B,C**, RWC was unchanged under nonstress conditions in the acetic acid-treated group relative to the water control group. In response to the drought imposed by the drying soil treatment, however, the RWC of leaves in the acetic acid-treated cassava plants was significantly higher than in the water-treated plants at 6 days, but not 4 days, after the onset of the drought (**Figure 2C**). These results suggest that the enhancement of drought avoidance induced by the

<sup>2</sup>https://phytozome.jgi.doe.gov/pz/portal.html

FIGURE 4 | Thermal images of whole plants and the net photosynthetic rate of cassava plants treated with either 10 mM acetic acid or water (control). (A) Thermal images of cassava plants treated with either 10 mM acetic acid or water (control) after 3 days under sunlight. (B) Leaf temperature differences in cassava plants treated with either 10 mM acetic acid or water (control) and placed in the sunlight. The mean leaf temperature was calculated based on temperature of five places on each leaf indicated by asterisks in (A). The color gradation on the bar indicates the photosynthetically active radiation (PAR). (C) Thermal images of cassava plants treated with either 10 mM acetic acid or water (control) after 5 days under the cloudy condition. (D) Differences in leaf temperature of cassava plants treated with either 10 mM acetic acid or water (control) and placed under the cloudy condition. The mean leaf temperature was calculated based on the temperature of five places on each leaf indicated by asterisks in (C). The color gradation on the bar indicates the PAR. (E) Net photosynthetic rate, stomatal conductance and transpiration rate of first to third leaves from the top of the stem of cassava plants treated with 10 mM acetic acid or water (control) at 7 days after the commencement of the drought. Mean ± SD of nine leaves from three plants.

acetic acid treatment is associated with maintaining the water status in plants.

# Decreased Degradation of Chlorophyll and Carotenoid in Acetic Acid-Treated Plants Subjected to Drought

The imposed drought resulted in a considerable decline in the level of total chlorophyll (Chl) and carotenoids. Total Chl and carotenoid levels, however, were higher in acetic acidtreated plants than in water-treated plants subjected to drought (**Figure 2D**). These results suggest that the negative effect of drought on photosynthetic pigments and carotenoids were substantially minimized by the acetic acid treatment. The level of total Chl and carotenoids was unchanged under non-stressed conditions and after 4 days of drought. After 6 days of drought, however, total Chl and carotenoid levels decreased in the watertreated control plants but not in the acetic acid-treated plants (**Figure 2D**). These results are consistent with the RWC and leaf wilting phenotype (**Figures 2B,C**) data obtained in response to the drying soil treatment.

### The Acetic Acid Treatment Increases Leaf Temperature and Decreases Net Photosynthesis and Stomatal Conductance

The acetic acid treatment delayed leaf wilting in cassava plants subjected to drought. To determine the physiological effect of the acetic acid treatment on leaves, leaf temperature and gas exchange in acetic acid-treated and non-treated plants was measured. Leaf temperature in detached acetic acid-treated leaves rapidly increased in comparison to the temperature in detached watertreated leaves (**Figure 3**). Leaf temperatures within intact, acetic acid-treated plants were measured under sunny (**Figures 4A,B**) or cloudy conditions (**Figures 4C,D**). The acetic acid-treated plants exhibited a higher leaf temperature than leaves in watertreated plants (**Figures 4A,C**). Leaf temperatures in acetic acidtreated plants under sunny conditions were also continuously higher than in leaves of water-treated plants (**Figures 4B,D**). Leaf temperature under the high photosynthetically active radiation (PAR) of a sunny day increased more significantly than under the low PAR of a cloudy day (**Figures 4B,D**). The net photosynthesis rate, stomatal conductance and transpiration rate in acetic acid-treated plants decreased considerably after 7 days of the acetic acid treatment in comparison to control (**Figure 4E**). We also examined the net photosynthetic rate, conductance and transpiration rate with 1 mM or 10 mM acetic acid treated-plants. The net photosynthetic rate, conductance and transpiration rate in 10 mM acetic acid-treated plants were decreased to 40%, 15%, and 38%, respectively, in comparison to that in control under 2,000 µmol mol−<sup>2</sup> s −1 . In 1 mM acetic acid treated-plants, the values were kept at 64%, 58%, and 50%, respectively (**Figure 4E**). These results suggest that acetic acid might induce stomatal closure and the degree of physiological effect to acetic acid might depend on the concentration of acetic acid used.

### Increase of Abscisic Acid (ABA) Level and Expression of ABA Signaling Pathway Genes in Leaves of Acetic Acid-Treated Cassava Plants at 7 Days After the Imposition of a Drought

Acetic acid-treatment delayed the decrease in RWC in leaves of cassava plants exposed to a drought. In Arabidopsis, research on the effect of acetic acid has suggested that acetic acid operates through jasmonate (JA) signaling to respond to drought and maintains the RWC in plants. Therefore, to investigate the effect

#### TABLE 1 | Acetic acid-responsive genes in cassava.

fpls-10-00521 April 22, 2019 Time: 17:40 # 8


(Continued)

#### TABLE 1 | Continued

fpls-10-00521 April 22, 2019 Time: 17:40 # 9


<sup>a</sup>Encoded proteins/other features indicate the putative functions of the gene products that are expected from sequence similarity. The information for the NCBI protein reference sequence with the highest sequence similarity with the probes is shown. <sup>b</sup>Log<sup>2</sup> ratio = log<sup>2</sup> (10 mM acetic acid/control). <sup>c</sup>The information for AGI locus ID with the highest sequence similarity with the probe is shown. <sup>d</sup>The information for cassava gene code with the highest sequence similarity with the probe is shown.

of the acetic acid, the measurement of ABA, JA and jasmonate-Ile (JA-Ile) and transcriptome analysis of acetic acid-treated- and control leaves were conducted. The ABA content from 10 mM acetic acid-treated leaves was significantly increased 1.9 folds in comparison to control one but JA and JA-Ile contents were not changed (**Figure 5**). In the transcriptome analysis, results indicated that there was an upregulation of genes involved in the ABA signaling pathway and heat shock proteins (HSPs) in the acetic acid-treated leaves relative to control leaves of cassava plants (**Table 1**). The microarray analysis identified 2,399 genes that showed significant differences (FDR < 0.0001) in expression as determined with Westfall–Young multiple testing correction based on an unpaired t-test. Among the differentially expressed genes, 227 and 330 genes were twofold up- or down-regulated,

Frontiers in Plant Science | www.frontiersin.org

the Arabidopsis database (TAIR10), the total number of Arabidopsis genes with GO annotations in TAIR10 and the p-value in parenthesis. Box colors in PAGE

indicate levels of statistical significance: yellow < 0.01; orange < 0.001. White boxes are shown as non-significant terms.

respectively, by the acetic acid treatment. Functional category classification using the gene ontology (GO) of the Arabidopsis Information Resource (TAIR10) for Biological Process by agriGO was performed on the 1,945 of the differentially expressed cassava genes that could be annotated based on the genome from Arabidopsis thaliana. These genes were classified into the GO terms "response to abscisic acid stimulus," "response to hydrogen peroxide," "response to heat," "response to high light intensity," "tRNA metabolic process," and "positive regulation of cellular process" for biological process (**Figure 6**). Here, we focused on "response to abscisic acid stimulus," "response to hydrogen peroxide," "response to heat," "response to high light intensity" due to show the increase of ABA level in acetic acid-treated leaves. The stress-upregulated cassava genes classified in these GO terms include the homolog of a key gene involved in response to stress and ABA; ABA INSENSITIVE 2, ABI2(AT5G57050.1), which encodes a member of a protein phosphatase 2C (PP2C) (**Table 1**) that plays a role in ABA signal transduction as negative regulators (Leung et al., 1997); several HSPs, such as HSP21 (AT4G27670.1), HSP70 (AT3G12580.1), HSP18.2 (AT5G59720.1), and HSP20 (AT1G54050.1 and AT2G29500.1), which play important roles in cells by preventing aggregation of destabilized proteins exposed to high heat and light intensity (Basha et al., 2012); OPEN STOMATA 1, OST1 (AT4G33950.2), which encodes the ABA-activated protein kinase, a member of SNF1-related protein kinases (SnRK2) that are involved in stomatal closure through ABA induction (Umezawa et al., 2010) (**Supplementary Figure S2A**); ABI five binding protein 2 (AFP2) (AT1G13740.1), whose members interact with the transcription factor, ABA-Insensitive5 (ABI5) which is mediated in ABA response (Lynch et al., 2017); and two different kinds of PP2C genes (AT5G51760.1 and AT5G59220.1) that may play a role in ABA signal transduction along with ABI2.

Although several genes related to the ABA signaling pathway were identified, the expression of key genes in the ABA biosynthetic pathway, such as the 9-cis-epoxycarotenoid dioxygenases (NCEDs), were not dramatically changed in response to the acetic acid (**Supplementary Figure S2B** and **Supplementary Table S1**) (Iuchi et al., 2001). Upregulation of JA biosynthetic genes, such as allene oxide cyclases (AOCs), and signaling pathway genes, such as COI1, JAZ, and MYC families, was also not observed in cassava plants treated with acetic acid (**Supplementary Table S2**). In the current study of cassava, the expression of genes involved in the ABA signaling pathway were upregulated by the acetic acid treatment and sunlight, suggesting that activation of the ABA signaling pathway may play a role in acetic acid-mediated drought avoidance in cassava plants through the increased ABA level.

#### DISCUSSION

Results of the present study indicate that an application of acetic acid enhances drought avoidance in cassava. Analyses of microarray experiments revealed that acetic acid upregulates the expression of ABA signaling pathway-related genes, such as ABA-insensitive 2 (ABI2; Finkelstein and Somerville, 1990) and OST1 (Mustilli et al., 2002) homologs, and stress response and tolerance-related genes, such as HSPs (**Figure 5** and **Table 1**). Acetic acid treatment increased leaf temperature and accumulation of ABA and decreased stomatal conductance and transpiration rates; suggesting that the acetic acid treatment retained the RWC in cassava plants subjected to drought through the ABA-mediated control of stomatal aperture.

Previous studies have reported that treatment of plants with acetic acid enhances drought tolerance in Arabidopsis, rice, maize, wheat, and rapeseed and that activation of JA signaling is a key event in the enhancement of drought tolerance in Arabidopsis (Kim et al., 2017). Activation of ABA biosynthesis and the ABA signaling pathway was not observed in Arabidopsis in response to an acetic acid treatment (Kim et al., 2017); although the expression of ABA signaling-related genes, such as ABI2, OST1, ABA-hypersensitive germination 1 (AHG1; Nishimura et al., 2007), ABA-insensitive five-binding protein 2 (AFP2; Lynch et al., 2017) were upregulated in cassava in response to an acetic acid treatment (**Table 1**). In the present study, increased expression of JA biosynthesis and signaling pathway genes was not observed in response to the acetic acid treatment, although the expression of JA biosynthetic enzyme AOC3 was upregulated and the level of JA increased transiently in response to an acetic acid treatment in Arabidopsis (Kim et al., 2017). Differences in the molecular response of cassava vs. Arabidopsis may be due to the following reasons: (1) innate differences in the response of woody plants, such as cassava, and herbaceous plants, such as Arabidopsis, to the exogenous application of acetic acid; (2) use of different experimental conditions in the studies conducted in Arabidopsis vs. cassava. For example, high light (in the range of 100–1,300 µmol m−<sup>2</sup> s −1 ) vs. low (∼100 µmol m−<sup>2</sup> s −1 ) light intensity was used in studies on cassava vs. Arabidopsis, respectively (Kim et al., 2017; Rasheed et al., 2018).

The application of acetic acid increased the expression of various stress response and stress tolerance-related genes, such as heat shock family proteins, HSP20 (Muthusamy et al., 2017), HSP21 (Zhong et al., 2013), HSP70 (Tang et al., 2016; Leng et al., 2017) and HSP90 (Krishna and Gloor, 2001), ER stress-related gene, HOP3 (At4g12400; Fernández-Bautista et al., 2017a,b), catalase 1 and catalase 2 (Zou et al., 2015), multiprotein-bridging factor 1c gene (MBF1c; Alavilli et al., 2017), and an outer membrane tryptophan-rich sensory protein (TSPO) (Guillaumot et al., 2009). HSP20 and HSP70 family members function as molecular chaperones and help plants adapt to environmental stress (Tang et al., 2016; Leng et al., 2017; Muthusamy et al., 2017). HSP21 is involved in the protection against oxidative stress (Zhong et al., 2013). HOP3, a member of the HOPs [heat shock protein 70 (HSP70) and heat shock protein 90 (HSP90) organizing proteins] family in Arabidopsis plays an essential role in the response of the endoplasmic reticulum (ER) in plants to abiotic stress (Fernández-Bautista et al., 2017a). HOP3 has been suggested, through its role in the alleviation of ER stress, to play an important function in plant response to various stresses (Fernández-Bautista et al., 2017a). It is plausible to suggest that acetic acid treatment enhances drought avoidance in cassava through the regulation of the ER stress response. Overexpression of P. alpinum multiprotein-bridging factor 1c gene (PaMBF1c)

enhanced salt tolerance in Arabidopsis (Alavilli et al., 2017). Catalase plays an important role in ABA- and H2O2-mediated signal transduction and in maintaining H2O<sup>2</sup> homeostasis in response to drought (Zou et al., 2015). TSPO expression is upregulated by drought in both shoots and roots, and the TSPO promoter has been shown to be useful for the genetic engineering of drought tolerant plants to drought (Rasheed et al., 2018).

#### CONCLUSION

Our study demonstrates that the external application of acetic acid under high PAR leads to stomatal closure which is induced by the activation of the ABA-dependent stress response pathway in cassava, by increasing the ABA level. Our study also indicates the possibility of avoiding drought without the inhibition of plant growth, through the external application of acetic acid with a lower concentration. Although the detailed mechanisms underlying acetic acid-mediated enhancement of drought avoidance in cassava remain to be elucidated, the findings of the current study suggest that treatment of plants with simple, easily available and low cost compounds, such as acetic acid, may have beneficial effects on the growth of plants subjected to drought.

#### AUTHOR CONTRIBUTIONS

YU and MoS conceptualized the study. YU, CU, MT, CH, and ST measured the physiological parameters. TM and SM measured the leaf angle. YK and MiS performed the phytohormone

#### REFERENCES


measurements. YU, MT, and AM performed the microarray analysis. CU, YO, and EM performed the propagation of plants. YU, CH, and MoS wrote the original draft.

# FUNDING

This project was financially supported by the Japan Science and Technology Agency (JST), Core Research for Evolutionary Science and Technology (CREST; JPMJCR13B4 to MoS), and grants from RIKEN (to MoS).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00521/ full#supplementary-material

FIGURE S1 | Leaf wilting phenotype of representative water-treated (control) cassava plant after being subjected to a drought (A), representative cassava plant not subjected to a drought (soil drying) (B,D,E), representative cassava plant subjected to a drought stress (C). Images in which the width of the leaflet (blue line) exceeded 10% of the midrib length (red line) were excluded from the analyzed dataset.

FIGURE S2 | Phylogenetic tree of SnRK2 and NCED amino acid sequences from cassava (Manihot esculenta, Me) and Arabidopsis (Arabidopsis thaliana, At). (A) SnRK2 and (B) NCED amino acid sequences. The distance of the branches denotes the bootstrap majority consensus values on 1,000 replicates.

TABLE S1 | Expression of NCED genes that play a key role in ABA biosynthesis.

TABLE S2 | Expression data of JA signaling pathway genes, such as COI1, MYCs, JAZs, and JA biosynthesis pathway genes, such as AOCs.


stress in Solanum lycopersicum. Adv. Crop Sci. Technol. 6:3. doi: 10.4172/2329- 8863.1000362


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

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

# A Wall-Associated Kinase Gene CaWAKL20 From Pepper Negatively Modulates Plant Thermotolerance by Reducing the Expression of ABA-Responsive Genes

Hu Wang, Huanhuan Niu, Minmin Liang, Yufei Zhai† , Wei Huang, Qin Ding, Yu Du and Minghui Lu\*

College of Horticulture, Northwest A&F University, Yangling, China

#### Edited by:

Sang Yeol Lee, Gyeongsang National University, South Korea

#### Reviewed by:

Yong Hwa Cheong, Sunchon National University, South Korea Haitao Shi, Hainan University, China

#### \*Correspondence: Minghui Lu

xnjacklu@nwsuaf.edu.cn

†Present address: Yufei Zhai, College of Horticulture, Nanjing Agricultural University, Nanjing, China

#### Specialty section:

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

Received: 23 December 2018 Accepted: 23 April 2019 Published: 14 May 2019

#### Citation:

Wang H, Niu H, Liang M, Zhai Y, Huang W, Ding Q, Du Y and Lu M (2019) A Wall-Associated Kinase Gene CaWAKL20 From Pepper Negatively Modulates Plant Thermotolerance by Reducing the Expression of ABA-Responsive Genes. Front. Plant Sci. 10:591. doi: 10.3389/fpls.2019.00591 Heat stress has become a major threat to crop production due to global warming; however, the mechanisms underlying plant high-temperature sensing are not well known. In plants, the membrane-anchored receptor-like kinases (RLKs) relay environmental signals into the cytoplasm. In a previous study, we isolated a wall-associated RLK-like (WAKL) gene CaWAKL20 from pepper (Capsicum annuum L.). Here, the amino acid sequence of CaWAKL20 was characterized and found to consist of conserved domains of WAK/WAKL family, including an extracellular region containing a GUB-WAK binding domain and a degenerated EGF2-like domain; a transmembrane region; and an intercellular region with an STKc catalytic domain. Moreover, CaWAKL20 transcription was inhibited by heat stress, whereas it was induced by both ABA and H2O<sup>2</sup> treatments. Silencing of CaWAKL20 enhanced pepper thermotolerance, while overexpression decreased Arabidopsis thermotolerance. Additionally, Arabidopsis lines overexpressing CaWAKL20 showed less sensitivity to ABA during seed germination and root growth. Finally, the survival rate of Arabidopsis seedlings under heat stress treatment was enhanced by ABA pre-treatment, while it was compromised by the overexpression of CaWAKL20. Furthermore, the heat-induced expression of several ABA-responsive genes and some key regulator genes for thermotolerance was decreased in Arabidopsis CaWAKL20-overexpression lines. These results suggest that CaWAKL20 negatively modulates plant thermotolerance by reducing the expression of ABA-responsive genes, laying a foundation for further investigation into the functional mechanisms of WAKs/WAKLs in plants undergoing environmental stresses.

Keywords: pepper, heat stress, wall-associated receptor-like protein kinase, abscisic acid, H2O<sup>2</sup>

# INTRODUCTION

Plants frequently face adverse environmental conditions, including extreme temperature, drought, salinity, and flood that can negatively affect plant growth, development, production, and even survival. As global warming worsens, heat waves are occurring with increased frequency and longer duration, especially in lower latitude regions (IPCC, 2014), and the resultant heat stress is becoming an increasingly significant problem for crop production and world food security (Ohama et al., 2017).

Heat stress can cause protein denaturation and aggregation, reduce cellular functions, and even result in cell death (Driedonks et al., 2015). To defend against these negative effects, as sessile organisms, plants have developed a conserved heat stress response (HSR) system to induce the expression of stress-related genes (Mittler et al., 2012). However, relatively little is known about how plants sense high temperature, which then hinders the establishment of systems that protect against heat stress in crop breeding and/or cultivation.

Under heat stress, the HSR can be triggered by increased fluidity of the plasma membrane (Sangwan et al., 2002) or accumulation of unfolded proteins (Rütgers et al., 2017). The plasma membrane is suggested to be a primary heat sensor in plants (Hofmann, 2009), and changes in cytomembrane fluidity can be sensed by integral membrane proteins, such as ion channels and/or transporters, as well as membrane-anchored receptor-like kinases (RLKs) (Zhu, 2016). While the involvement of membrane calcium channels in plant HSR has been confirmed (Saidi et al., 2009; Finka et al., 2012), the contribution of RLKs to plant thermotolerance is poorly understood.

Plant RLKs are classed as transmembrane proteins containing an N-terminal extracellular domain and a C-terminal intracellular kinase domain. Receptor-like kinases are thought to participate in a diverse range of life processes including growth, development, hormone signalings, plant-microbe interactions, and responses to environmental stimuli (Shiu and Bleecker, 2001; Ye et al., 2017).

As an RLK subfamily, wall-associated receptor kinase (WAK) and WAK-like (WAKL) are characterized by the presence of an extracellular epidermal growth factor (EGF)-like domain. These WAKs/WAKLs physically link the cell wall to the plasma membrane, directly transmitting extracellular signals into the cytoplasm to regulate cell growth and stress responses (Anderson et al., 2001; Kohorn, 2016). Zhang et al. (2017) even suggested that "ZmWAK functions as a hub in the trade-off between maize growth and defense."

Since firstly reported in Arabidopsis (He et al., 1998), the contributions of WAKs/WAKLs to plant immunity have been widely studied, including AtWAKL22 in Arabidopsis resistance to Fusarium oxysporum (Diener and Ausubel, 2005), SlWAK1 in tomato resistance to Pst (Rosli et al., 2013), OsWAKs in rice resistance to fungal blast and bacterial blight disease (Li et al., 2009; Delteil et al., 2016; Harkenrider et al., 2016), ZmWAKs in maize resistance to northern corn leaf blight and head-smut disease (Hurni et al., 2015; Zhang et al., 2017; Yang et al., 2018), and TaWAKL4 in wheat resistance to Zymoseptoria tritici (Saintenac et al., 2018).

An increasing number of studies have reported on the involvement of WAKs/WAKLs in plant tolerance to abiotic stresses. For example, Sivaguru et al. (2003) reported that overexpression of AtWAK1 enhanced Arabidopsis tolerance to heavy metal stress, whereas Hou et al. (2005) found that T-DNA insertion in the AtWAKL4 promoter increased Arabidopsis hypersensitivity to salt stress. Furthermore, Xia et al. (2018) suggested that, when subjected to excess Cu, OsWAK11 can modify cell wall properties to retain Cu at the cell wall. However, it remains to be elucidated whether WAK/WAKL genes are involved in plant thermotolerance.

In our previous work, we isolated a heat-responsive gene Capana12g000852 from the pepper plant. Due to its high amino acid sequence homology with Arabidopsis AtWAKL20, Capana12g000852 was renamed CaWAKL20. In this study, we show that silencing the CaWAKL20 gene enhanced pepper tolerance to heat stress, whereas overexpressing CaWAKL20 weakened Arabidopsis thermotolerance. In addition, Arabidopsis lines overexpressing CaWAKL20 exhibited reduced sensitivity to abscisic acid (ABA) during seed germination and seedling growth. We also show that, under heat stress treatment, the induced expression of some ABA-responsive genes was reduced in CaWAKL20-overexpressing Arabidopsis lines. Our study may contribute to the understanding of the molecular mechanisms of plant tolerance against heat stress.

### MATERIALS AND METHODS

#### Plant Materials and Growth Conditions

The pepper R9 thermotolerant line (introduced from the World-Asia Vegetable Research and Development Center, PP0042-51) and Arabidopsis ecotype Col-0 were used in this study. Plant materials were grown in a chamber under 200 µmol·m−<sup>2</sup> ·s −1 illumination intensity, a 16 h day/8 h night regime, and 70% relative humidity. The temperature was set to 26◦C/20◦C (day/night) for pepper and 22◦C/18◦C for Arabidopsis.

#### Analysis of the Deduced CaWAKL20 Amino Acid Sequence

The CaWAKL20 (Capana12g000852) amino acid sequence was downloaded from the PGD database<sup>1</sup> , and those of Arabidopsis WAKs/WAKLs were obtained from TAIR (The Arabidopsis Information Resource)<sup>2</sup> (Verica and He, 2002). The conserved domains in the CaWAKL20 amino acid sequence were identified using the online SMART tool (Simple Modular Architecture Research Tool)<sup>3</sup> . Full-length CaWAKL20 and AtWAKs/WAKLs were aligned using the online Clustal Omega program<sup>4</sup> , and the phylogenetic tree was generated by MEGA 6 software using the neighbor-joining method, based on the p-distance substitution model, and with 1000 bootstrap replicates (Tamura et al., 2013).

Arabidopsis AtWAKL20 was used to predict the CaWAKL20 protein-protein interaction network in the STRING interaction database (Search Tool for the Retrieval of Interacting Genes/Proteins)<sup>5</sup> with a 0.400 confidence score. Finally, the output was imported into Cytoscape\_v3.4.0 (National Institute of General Medical Sciences, MD, United States) to generate the network map.

<sup>1</sup> ftp://ftp.solgenomics.net/genomes/Capsicum\_annuum/C.annuum\_zunla/ annotation/

<sup>2</sup>https://www.arabidopsis.org/

<sup>3</sup>http://smart.embl-heidelberg.de/

<sup>4</sup>http://www.ebi.ac.uk/Tools/msa/clustalo/

<sup>5</sup>http://string-db.org

#### Subcellular Localization of CaWAKL20

The CaWAKL20 coding sequence (CDS) without a stop codon was amplified from the pepper line R9 using the GFP-CaWAKL20-F and GFP-CaWAKL20-R primer pair (**Supplementary Table S1**); the PCR product was then cloned into a GFP-tagged pBI221 transient expression vector. The empty vector was used as the control. Particle bombardment was used to transfect plasmid into onion epidermal cells, which were then incubated for 24 h at 28◦C in the dark. The GFP signal was visualized using A1R confocal laser scanning microscopy (Nikon, Tokyo, Japan). For plasmolysis analysis, the transformed onion epidermal cells were treated in 0.8 M mannitol solution for 15 min before observation of GFP expression (Genovesi et al., 2008).

### Virus-Induced Gene Silencing (VIGS) of CaWAKL20

To generate gene-silenced plants using VIGS, a conserved 346 bp fragment in the CaWAKL20 CDS was amplified from the pepper line R9 with the TRV2-CaWAKL20-F and TRV2-CaWAKL20-R primer pair (**Supplementary Table S1**), and the PCR product was inserted into a pMD19-T vector (Takara, Dalian, China). After being digested with EcoR I and BamH I, the resultant CaWAKL20 gene fragment was cloned into the pTRV2 virus expression vector to generate the TRV2:CaWAKL20 gene-silencing vector. The empty vector was used as the control and was referred to as TRV2:00, and the TRV2:CaPDS (phytoene desaturase gene) vector was used as the marker for successful gene silencing. The Agrobacterium tumefaciens strain GV3101 containing TRV2:CaWAKL20, TRV2:00, or TRV2:CaPDS was injected into the cotyledons of the pepper line for gene silencing. When the photo-bleaching phenotype was clearly observable in newly grown leaves of plants transformed with TRV2:CaPDS, the transcription of TRV1 and TRV2 in plants expressing TRV2:00 and TRV2:CaWAKL20 was assessed by semi-quantitative PCR using the TRV1-TL-F/TRV1-TL-R and TRV2-Coat P-F/TRV2-Coat P-R primer pairs, respectively (**Supplementary Table S1**, Tsaballa et al., 2011). The efficiency of silencing of CaWAKL20 expression in plants expressing TRV2:CaWAKL20 was assessed by qRT-PCR using the qCaWAKL20-F and qCaWAKL20-R primer pair (**Supplementary Table S1**).

#### Generation of Arabidopsis Lines Overexpressing CaWAKL20

The complete CaWAKL20 CDS was amplified from the R9 line using the CaWAKL20-F and CaWAKL20-R primer pair (**Supplementary Table S1**), and the amplification product was cloned into the pVBG2307 plant binary expression vector under the control of the CaMV35S promoter. Using A. tumefaciens strain GV3101, the 35S::CaWAKL20 vector was transformed into Arabidopsis ecotype Col-0 by the floral dip method (Clough and Bent, 1998). Transgenic plants were screened by adding kanamycin into the MS culture medium, and the T3 generation was then used for subsequent experiments.

# Experimental Treatments and Collection of Samples

For tissue-specific expression analysis of CaWAKL20, samples of seeds, roots, stems, young leaves, flower buds, and young fruits were collected from pepper plants grown under normal conditions.

For analysis of CaWAKL20 expression under stress treatments, pepper seedlings at the six-leaf stage were either incubated at 45◦C for heat stress treatment, sprayed evenly with a 0.1 mM ABA solution for ABA treatment, or immersed with roots in a 100 mM H2O<sup>2</sup> solution for H2O<sup>2</sup> treatment. The young leaves were collected at 0, 1, 3, 6, 12, and 24 h post-treatments, and all the samples were immediately frozen in liquid nitrogen and stored at −80◦C for RNA extraction.

To assess the efficiency of silencing of CaWAKL20 expression, pepper seedlings containing TRV2:CaWAKL20 and TRV2:00 were incubated at 45◦C for 1 h. To measure the thermotolerance of CaWAKL20-silenced plants, pepper seedlings containing TRV2:CaWAKL20 and TRV2:00 were incubated at 45◦C for 10 h and then allowed to recover for 24 h. The pepper leaves were sampled at the end of both treatments and either stored at −80 ◦C for gene expression analysis or used immediately to determine malondialdehyde (MDA) content and the maximal photochemical efficiency of PSII (Fv/Fm).

For thermotolerance evaluation of plants overexpressing CaWAKL20, transgenic Arabidopsis lines overexpressing CaWAKL20 (CaWAKL20-OE) or containing the empty vector (EV) were used. Ten days old Arabidopsis seedlings on MS plates were immersed in a water bath at 45◦C for 50 min and then allowed to recover for 2 days at 22◦C. Three weeks old Arabidopsis seedlings in pots were incubated at 45◦C for 12 h and allowed to recover for 7 days at 22◦C. The survival rates of Arabidopsis seedlings were calculated at the end of both treatments, and leaves from seedlings in pot were sampled at 3 h post-heat stress treatment for gene expression analysis. Three weeks old Arabidopsis seedlings in pots were treated at 45◦C for 12 h and then collected for the assessment of H2O<sup>2</sup> accumulation.

To assess the effects of CaWAKL20 overexpression on Arabidopsis sensitivity to ABA, seeds of CaWAKL20-OE and EV lines were germinated on MS plates containing 0, 0.75, and 1 µM ABA. The germination rate, green cotyledon rate, and root length were determined at 4, 9, and 11 days after treatment, respectively. All experiments were performed with three biological replicates.

To evaluate the effects of exogenous ABA on the thermotolerance of plants overexpressing CaWAKL20, 10 days old seedlings of Arabidopsis lines of EV, OE2, and OE14 were transferred to MS medium with or without 5 µM ABA (Wang et al., 2017). After pre-treatment with ABA for 2 days, the plates were immersed in a water bath at 45◦C for 50 min and then allowed to recover for 2 or 3 days at 22◦C. The rate of green seedlings was calculated at different time points after treatments.

# Determination of MDA Content, PSII Fv/Fm and H2O<sup>2</sup> Accumulation

The MDA content of the pepper leaves was measured using the thiobarbituric acid assay (Dhindsa et al., 1981). The PSII Fv/Fm of

pepper leaves was determined using the FluorCam7 fluorescence imaging system (EcoTech, China). The assessment of H2O<sup>2</sup> accumulation in Arabidopsis seedlings was performed using the diaminobenzidine (DAB) staining method (Dang et al., 2013). All experiments were performed with three biological replicates.

### Total RNA Extraction, cDNA Synthesis, and qRT-PCR Analysis

Total RNA was extracted from the leaves of pepper and Arabidopsis plants using the Trizol <sup>R</sup> kit (Invitrogen, Carlsbad, CA, United States), and the residual genomic DNA was digested with RNase-free DNase I (Promega, Madison, WI, United States). First-strand cDNA synthesis was performed using the PrimeScriptTM Kit (TaKaRa, Tokyo, Japan) according to the manufacturer's instructions. Primer pairs were designed using Primer-BLAST in NCBI<sup>6</sup> (**Supplementary Table S1**), and qRT-PCR was performed using SYBR <sup>R</sup> Premix Ex TaqTM II (TaKaRa). Relative gene expression levels were analyzed according to the 2−11CT method (Livak and Schmittgen, 2001), in which CaUBI3 and AtActin2 were used as internal controls for pepper and Arabidopsis, respectively. Significance tests for differences in gene expression between control and stress treatments were performed using the Student's t-test method at the 0.05 and 0.01 significance levels.

#### RESULTS

#### Analysis of the Deduced CaWAKL20 Amino Acid Sequence

Domains that are conserved in the WAK/WAKL protein family were identified in the amino acid sequence of Capana12g000852 (**Supplementary Figures S1A,B**) using the SMART online tool. The conserved domains included an extracellular region containing a signal peptide and a GUB-WAK binding domain (galacturonan-binding, pfam13947) at the N-terminus, a transmembrane region, and an intercellular region with a catalytic STKc domain (Serine/Threonine kinase, cd14066) at the C-terminus. The EGF domain, a marker for the WAK subfamily that distinguishes it from others in RLKs (Anderson et al., 2001), was absent in Capana12g000852 (**Supplementary Figure S1A**); however, a degenerated EGF2-like domain (Prosite: PS01187) (Verica et al., 2003) was present (**Supplementary Figure S1B**), which is also observed for AtWAKL20 (Verica and He, 2002). In addition, Capana12g000852 displayed a closer phylogenetic relationship with AtWAKL20 than with the other 25 WAK/WAKL members found in Arabidopsis (Verica and He, 2002; **Supplementary Figure S1C**), and was therefore renamed CaWAKL20.

To further understand its functional patterns, the CaWAKL20 protein-protein interaction network was predicted using the STRING online tool based on Arabidopsis interologs. Ten CaWAKL20 interaction partners were identified, including a WAK family member (WAKL7), a zinc ion binding protein, and eight protein phosphatase 2C (PP2C) family members, i.e., ABI2 (ABA Insensitive 2), AHG1 (ABA-Hypersensitive Germination 1), APD9 (Arabidopsis PP2C clade D9), EGR2 (E Growth-Regulating 2), HAI1 (Highly ABA-Induced PP2C gene 1), and PP2C74, 76, and 80 (**Supplementary Figure S1D**).

#### Subcellular Localization of CaWAKL20

To confirm the WAK/WAKL classification of CaWAKL20, the GFP-tagged CaWAKL20 CDS was transiently expressed in onion epidermal cells under the control of the CaMV35S promoter. In the cells transformed with the empty control vector, the GFP signal was distributed throughout the entire cell. In contrast, green fluorescence was detected along the edge of the cells transformed with the CaWAKL20-GFP fusion vector (**Figure 1A**).

To further test the membrane- and cell wall-binding properties of CaWAKL20, the onion epidermal cells transformed with CaWAKL20 were cultured in mannitol for plasmolysis. In plasmolyzed cells, the green fluorescence of the CaWAKL20-GFP protein was observed both in the plasma membrane and the cell wall (**Figure 1B**). This suggests that CaWAKL20 was localized at the plasma membrane and linked to the cell wall.

### Expression of CaWAKL20 in Different Pepper Plant Tissues and Under Heat Stress, ABA, and H2O<sup>2</sup> Treatments

To explore the expression patterns of CaWAKL20 in different pepper plant tissues, the roots, stems, leaves, flowers, and fruits of plants grown under normal conditions were sampled from the R9 thermotolerant line, and qRT-PCR was performed using a CaWAKL20-specific primer pair. The results indicated that CaWAKL20 was variably expressed in different tissues, and the expression levels were clearly lower in the stems and flowers than in the other tissues (**Figure 2A**).

The CaWAKL20 protein was predicted to interact with several proteins related to ABA (**Supplementary Figure S1D**), a major phytohormone essential for plant responses to a wide range of stresses (Vishwakarma et al., 2017). In addition, reactive oxygen species (ROS) also play a key role in plants' acclimation to abiotic stresses, including heat (Suzuki and Katano, 2018). Therefore, to elucidate the response of CaWAKL20 to heat stress, ABA and ROS treatments, CaWAKL20 expression was analyzed following incubation at 45◦C and exposure to 0.1 mM ABA and 100 mM H2O2, respectively. The results showed that CaWAKL20 transcription responded to all treatments. For heat stress, CaWAKL20 expression decreased continuously during the whole treatment (**Figure 2B**), while with both ABA and H2O<sup>2</sup> treatments, CaWAKL20 expression increased at 1 h, and then declined rapidly (**Figures 2C,D**).

#### Thermotolerance Was Enhanced in CaWAKL20-Silenced Pepper Seedlings

To understand the role of CaWAKL20 in pepper thermotolerance, CaWAKL20 was silenced in the thermotolerant R9 pepper line. The fragment of CaWAKL20 used for VIGS was predicted to have no off-target potential (by the VIGS tool

<sup>6</sup>https://www.ncbi.nlm.nih.gov/tools/primer-blast/

in SGN<sup>7</sup> ) and was therefore presumed to be specific for the CaWAKL20 gene. Approximately 1 month after injection for VIGS, the bleaching was clearly observed in the leaves of positive

<sup>7</sup>http://vigs.solgenomics.net/

control pepper seedlings containing TRV2:CaPDS; however, no visible phenotype was observed in leaves with either TRV2:00 or TRV2:CaWAKL20 (**Supplementary Figure S2A**). Both the RNA1 segment in TRV1 (GenBank: AF406990) and the coat protein in TRV2 (GenBank: AF406991) were almost equally

expressed in all virus-infected pepper plants (**Supplementary Figure S2B**). Compared to plants with TRV2:00, the expression of the CaWAKL20 gene was silenced to approximately 70% in the plants with TRV2:CaWAKL20, which became more evident after the heat stress of 45◦C for 1 h (**Supplementary Figure S2C**).

After exposure to heat stress at 45◦C for 10 h and recovery for 24 h under normal conditions, the leaves of the pepper plants containing TRV2:00 were severely parched, while those with TRV2:CaWAKL20 resumed growing (**Figure 3A**). Under heat stress treatment, the MDA content increased from 5.88 to 11.44 µmol<sup>∗</sup> g −1 fresh weight (FW) in leaves of plants with TRV2:CaWAKL20, whereas for plants with TRV2:00 the MDA content in the leaves increased from 5.98 to 9.15 µmol<sup>∗</sup> g <sup>−</sup>1FW. After heat treatment, the MDA content was significantly higher in the leaves of plants with TRV2:00 than those with TRV2:CaWAKL20 (p < 0.05) (**Figure 3B**). Furthermore, with prolonged heat stress, the PSII Fv/Fm declined continuously in the leaves of plants transformed with both TRV2:CaWAKL20 and TRV2:00; at the end of treatment, the Fv/Fm value was clearly higher in the former (0.60) than that in the latter (0.51) (p < 0.01) (**Figures 3C,D**).

#### Thermotolerance Was Reduced in CaWAKL20-Overexpressing Arabidopsis Lines

To further understand the role of CaWAKL20 in plant thermotolerance, Arabidopsis lines overexpressing CaWAKL20 (OE) were generated, with OE2 and OE14 being selected and used for further studies (**Supplementary Figures S2D,E**).

For 10 days old Arabidopsis seedlings on MS plates, after induction of heat stress at 45◦C for 50 min followed by recovery for 2 days, the survival rate was higher than 80% in the EV line compared to only approximately 12% for seedlings in either CaWAKL20-OE line (**Figures 4A,B**). For 3 weeks old Arabidopsis in pots, after induction of heat stress at 45◦C for 12 h and recovery for 7 days, the survival rate was higher than 80% in the EV line compared to only 31% in the OE2 line and 34% in the OE14 line (**Figures 4C,D**). Meanwhile, the seedlings from the CaWAKL20-OE lines that survived had fewer green leaves than those from the EV line. After heat stress treatment at 45◦C for 12 h, the level of DAB staining increased in all Arabidopsis seedlings. Compared with that of the EV line, the staining intensity in the seedlings of both OE2 and OE14 were obviously raised (**Figure 4E**), suggesting the higher levels of H2O<sup>2</sup> accumulation in CaWAKL20-OE lines.

#### ABA Sensitivity Decreased in Arabidopsis CaWAKL20-OE Lines

No difference was observed in the seed germination rate between CaWAKL20-OE and EV Arabidopsis lines under normal growth conditions. Under ABA treatments, the seed germination rate was reduced, but more so in the EV line than in the CaWAKL20-OE lines (**Figure 5A**). When the treatments were extended to 9 days, the growth of CaWAKL20-OE seedlings was

significant difference at the 0.01 level.

clearly better than those of the EV line (**Figure 5B**). Under 0.75 and 1 µM ABA treatments, the percentage of green cotyledons decreased to 47 and 35%, respectively, in the EV line, but to 77 and 65%, respectively, in the CaWAKL20-OE lines (**Figure 5C**). Furthermore, root growth was suppressed by ABA treatments in both the CaWAKL20-OE and EV lines; however, root length

was clearly greater in the former than in the latter (**Figure 5D**). For example, under 1 µM ABA treatment, the root length of the CaWAKL20-OE seedlings was approximately twofold greater than that of EV seedlings (**Figure 5E**).

# ABA-Enhanced Thermotolerance Was Compromised by Overexpression of CaWAKL20 in Arabidopsis

To clarify the relationship between ABA treatment and CaWAKL20 expression in the development of plant thermotolerance, the seedlings of Arabidopsis CaWAKL20-OE were pretreated with ABA for 2 days and then exposed to heat stress. The results showed that the heat stress decreased the survival rate of Arabidopsis seedlings of all three lines in both with and without ABA pre-treatment, and the survival rates in OE2 and OE14 were significantly lower than those in EV line in both the tested time points (**Figure 6**). Compared to those without ABA pre-treatment (ABA-), however, the survival rate of Arabidopsis seedlings under heat stress was enhanced by ABA pre-treatment (ABA+) in all three Arabidopsis lines of EV, OE2, and OE14 (**Figure 6A**). Furthermore, under the treatment of ABA-, in OE2 and OE14, the survival rate was lower than in EV line by about 35 and 31% after 2 days of heat stress, and about 34 and 35% after 3 days of heat stress, respectively. Under the ABA+ treatment, the survival rate was lower in OE2 and OE14 than in EV line by about 15 and 13% after 2 days of heat stress, and about 44 and 42% after 3 days of heat stress, respectively (**Figure 6B**).

### Heat Stress-Induced Expression of ABA-Responsive Genes Was Reduced in Arabidopsis CaWAKL20-OE Lines

To confirm the involvement of ABA in CaWAKL20-mediated thermotolerance, we further examined the expression of several heat-related and simultaneously ABA-responsive genes in Arabidopsis EV line and CaWAKL20-OE lines under heat stress; the genes included AREB (Fujita et al., 2005), ABF (Choi et al., 2013), HSFA6b (Huang et al., 2016), DREB, and HSFA3 (Schramm et al., 2008). All the genes were induced by heat stress in both the EV line and CaWAKL20-OE lines; however, except for AtHSFA6b, the expression levels were clearly lower in the CaWAKL20-OE lines than in the EV line (**Figure 7A**).

In addition, the expression of several key regulator genes for plant thermotolerance, HSFA1a, HSFA2, and HSFA7a (Yoshida et al., 2011), was also assessed under heat stress. The results

showed that in line with those of ABA-responsive genes, the transcription of all tested thermotolerance-regulator genes was enhanced by heat stress, whereas this enhancement was inhibited by the overexpression of CaWAKL20 in both OE2 and OE4 (**Figure 7B**).

#### DISCUSSION

In a previous study, we isolated a heat-responsive gene Capana12g000852 from the pepper plant (data not shown). Further conserved domain analysis identified all the necessary WAK/WAKL domains in the deduced Capana12g000852 amino acid sequence (**Supplementary Figures S1A,B**). In addition, Capana12g000852 showed a similar domain distribution and closer phylogenetic relationship to Arabidopsis AtWAKL20 (**Supplementary Figures S1A,C**). From these data, we suggest that Capana12g000852 is a pepper homolog of AtWAKL20 and rename it CaWAKL20. The WAK/WAKL properties of CaWAKL20 are also supported by its subcellular localization in onion epidermal cells (**Figure 1**).

Although WAKs/WAKLs participate in plant responses to various biotic and abiotic stresses as a plasma membrane localized receptor-like kinase (Anderson et al., 2001), their roles in thermotolerance are unclear. Here, we showed that CaWAKL20 expression was downregulated in a manner that was dependent on the duration of the heat stress treatment (**Figure 2B**). When CaWAKL20 was silenced, pepper plant thermotolerance was enhanced as evidenced by the smaller increase in MDA content and the lower decline in the Fv/Fm value, compared to the plants transformed with the empty TRV2:00 vector (**Figure 3**). In contrast, the Arabidopsis CaWAKL20-OE lines displayed reduced thermotolerance in terms of seedling survival rate and ROS accumulation (**Figure 4**). These data suggest that CaWAKL20 negatively modulates plant thermotolerance.

The induced expression of WAK/WAKL genes is widely believed to be a requirement for plant survival during pathogen infection or heavy metal stress (He et al., 1998; Sivaguru et al., 2003); however, some contradictory results have also been reported. Harkenrider et al. (2016) found that overexpression of OsWAK25 increased rice susceptibility to necrotrophic fungal pathogens, although it enhanced rice resistance to hemibiotrophic pathogens. OsWAK14, OsWAK91, and OsWAK92 positively regulate rice resistance to blast fungus, while OsWAK112d functions as a negative regulator (Delteil et al., 2016). When the AtWAKL4 promoter was impaired, Arabidopsis tolerance to K+, Na+, Cu2+, and Zn2<sup>+</sup> was reduced, but its tolerance to Ni2<sup>+</sup> was enhanced (Hou et al., 2005). Therefore, WAKs/WAKLs may have differential roles in plant tolerance against biotic and/or abiotic stresses, and/or the roles may be stress type-dependent. As far as CaWAKL20, its functional model in the pepper plant response to heat stress remains to be further elucidated.

When contending with adverse environments, plants activate their protective mechanisms that, conversely, often suppress plant growth to focus energy on defending against the stress. Therefore, the ability to switch from growth to defense is crucial for plants' survival under stressed conditions (Wang and Wang, 2014; Albrecht and Argueso, 2017). Several hubs for tuning plant stress signaling and development have been reported in plants, such as CDPKs (Calcium-Dependent Protein Kinases) (Schulz et al., 2013) and WAKs/WAKLs (Kohorn, 2016). Zhang et al. (2017) found that maize ZmWAK promotes cell growth in the absence of pathogens but switches to a protective role when the maize plant is attacked by Sporisorium reilianum. In our study, after heat stress treatment, CaWAKL20 expression in the pepper thermotolerant line declined continuously (**Figure 2B**); similarly, Giarola et al. (2016) also observed that CpWAK1 transcription in Craterostigma plantagineum, a plant species with tolerance to extreme desiccation, was downregulated during dehydration treatment. These results suggest that maintaining a lower level of CaWAKL20 expression to slow cell growth is beneficial for pepper tolerance to injury from heat stress. This is also supported by our transgenic data (**Figures 3**, **4**) as well as the predicted interaction between CaWAKL20 and the growth regulator EGR2 (**Supplementary Figure S1D**), although the functional mechanisms involved require further detailed study.

0 h after treatment was taken as 1.0. Error bars represent standard deviations for three replicates, and each replicate consisted of 12 seedlings. ∗∗ indicates significant difference at the 0.01 level.

Abscisic acid plays important roles in plant responses to a range of environmental stresses, including heat stress. ABA is thought to perform a number of cellular functions, such as controlling the production of protective enzymes and regulating the transfer of water, to keep plant cells alive from heat stress (reviewed by Vishwakarma et al., 2017). Heat acclimation triggers an increase in endogenous ABA content (Liu et al., 2006); conversely, exogenous ABA enhances thermotolerance by upregulating the expression of heat-shock proteins (HSPs), including transcription factors (Wang et al., 2017). In this study, CaWAKL20 was predicted to interact with three ABA signaling components – ABI2, HAI1, and AHG1 (**Supplementary Figure S1D**) – and CaWAKL20 expression was rapidly induced by ABA treatment (**Figure 2C**). Meanwhile, Arabidopsis CaWAKL20-OE lines showed a lower sensitivity to ABA than the EV line for seed germination, seedling survival, and root growth (**Figure 5**), and the enhanced thermotolerance of Arabidopsis seedlings by ABA pre-treatment was compromised in CaWAKL20-OE lines (**Figure 6**). These results suggest that CaWAKL20 functions in an ABA-related pathway. In addition, the heat-induced expression of several ABA-responsive genes, AREB, ABF, DREB, and HSFA3, and some key regulator genes for plant thermotolerance, HSFA1a, HSFA2, and HSFA7a, was reduced in Arabidopsis CaWAKL20-OE lines compared to the EV line (**Figure 7**). Interestingly, similar phenomena were observed in the plastid casein kinase 2 (CK2) knockout mutant; CK2 is a major serine/threonine-specific kinase in the chloroplast stroma, and the authors argued that CK2 positively regulated retrograde signaling from the plastid to the nucleus during plant responses to ABA and heat stress (Wang et al., 2014). These results indicate that CaWAKL20 negatively modulates pepper plant thermotolerance by repressing the expression of ABA-responsive genes.

As universal signals, ROS might integrate with hormone signaling, including ABA signaling, to tailor the cellular homeostasis under stress conditions (Suzuki and Katano, 2018). In our study, the expression was also induced by exogenous H2O<sup>2</sup> with a pattern similar to that of ABA (**Figure 2D**). Suzuki et al. (2013) reported that ROS was responsible for the spread of heat signal from the initial site to the entire plant, and ABA specifically regulated plant acclimation to heat stress. Thus, it can be hypothesized that CaWAKL20 functions to link the ROS signal and ABA pathway. In this model, both ABA and H2O<sup>2</sup> induce

the expression of CaWAKL20, and CaWAKL20 expression further negatively regulates ABA signaling by decreasing the ABA sensitivity of plant growth and inhibiting the induced-expression of ABA-responsive genes and key regulator genes for thermotolerance under heat stress. However, which node in the ABA signal pathway, such as perception, biosynthesis, degradation, or signaling, is regulated by CaWAKL20 needs more investigation to verify.

#### CONCLUSION

In this study, we report that pepper CaWAKL20 possesses the conserved domains of the WAK/WAKL family and is localized to the plasma membrane and linked to the cell wall. The expression of CaWAKL20 was downregulated by heat stress but upregulated by ABA treatment. Silencing CaWAKL20 expression enhanced pepper thermotolerance, but CaWAKL20 overexpression decreased Arabidopsis tolerance to heat stress. In addition, overexpressing CaWAKL20 reduced Arabidopsis sensitivity to ABA and decreased the heat-induced expression of ABA-responsive genes. Therefore, we suggest that CaWAKL20 negatively modulates plant thermotolerance by inhibiting ABA-responsive gene expression. Our results lay a foundation for further understanding of the functional

#### REFERENCES


mechanisms of WAKs/WAKLs during plant adaptation to environmental stress.

#### AUTHOR CONTRIBUTIONS

HW and MLu designed the research. HW, HN, MLi, YZ, and WH performed the experiments. HW and HN analyzed the data and drafted the manuscript. QD, YD, and MLu revised the manuscript and contributed reagents, materials, and analysis tools.

#### FUNDING

This work was supported by the National Natural Science Foundation of China (31572114, 31872091) and Agricultural Key Science and Technology Program of Shaanxi Province, China (2016NY-063, 2018NY-029).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00591/ full#supplementary-material

system regulate acclimation to heat. Front. Plant Sci. 6:999. doi: 10.3389/fpls. 2015.00999


leaf blight encodes a wall-associated receptor-like kinase. Proc. Natl. Acad. Sci. U.S.A. 112, 8780–8785. doi: 10.1073/pnas.1502522112

IPCC (2014). Climate Change 2014: Synthesis Report. Geneva: IPCC, 55–74.


and abscisic acid regulates rapid systemic acclimation in plants. Plant Cell 25, 3553–3569. doi: 10.1105/tpc.113.114595


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

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

# The 26S Proteasome Is Required for the Maintenance of Root Apical Meristem by Modulating Auxin and Cytokinin Responses Under High-Boron Stress

Takuya Sakamoto<sup>1</sup> , Naoyuki Sotta<sup>2</sup> , Takamasa Suzuki<sup>3</sup> , Toru Fujiwara<sup>2</sup> and Sachihiro Matsunaga<sup>1</sup> \*

 Department of Applied Biological Science, Faculty of Science and Technology, Tokyo University of Science, Noda, Japan, Department of Applied Biological Chemistry, Graduate School of Agricultural and Life Sciences, The University of Tokyo, Bunkyo, Japan, ¯ College of Bioscience and Biotechnology, Chubu University, Kasugai, Japan

#### Edited by:

Jose M. Pardo, Instituto de Bioquímica Vegetal y Fotosíntesis (IBVF), Spain

#### Reviewed by:

Taras P. Pasternak, University of Freiburg, Germany Fangsen Xu, Huazhong Agricultural University, China

> \*Correspondence: Sachihiro Matsunaga sachi@rs.tus.ac.jp

#### Specialty section:

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

Received: 31 December 2018 Accepted: 23 April 2019 Published: 14 May 2019

#### Citation:

Sakamoto T, Sotta N, Suzuki T, Fujiwara T and Matsunaga S (2019) The 26S Proteasome Is Required for the Maintenance of Root Apical Meristem by Modulating Auxin and Cytokinin Responses Under High-Boron Stress. Front. Plant Sci. 10:590. doi: 10.3389/fpls.2019.00590 Boron (B), an essential micronutrient, causes adverse effects on the growth and development of plants when highly accumulated. By the analysis of Arabidopsis mutants hypersensitive to high-boron (high-B) stress, we have shown that 26S proteasome (26SP) is required to maintain the morphology of the root apical meristem (RAM) under high-B stress. To further understand the molecular function of 26SP in tolerance to high-B stress in the RAM, in this study we investigated the pathways regulated by 26SP using a 26SP subunit mutant, rpt5a, which is hypersensitive to high-B stress. Expression of RPT5a was induced by high-B stress in the entire RAM accompanied by its strong expression in the stele, including the stem cells. Analysis of stele organization in the rpt5a mutant revealed that 26SP is especially important for maintenance of the stele under high-B stress condition (3 mM B treatment). Expression analyses of an auxin-response reporter revealed that auxin responses were enhanced in the stele and the stem cell niche by high-B stress, especially in the rpt5a mutant. In contrast, the expression of TCS::GFP representing cytokinin signaling in the stem cell niche was unchanged in the wild type and extremely weak in the rpt5a mutant, irrespective of B condition. The drastically aberrant auxin and cytokinin responses in the rpt5a mutant under high-B stress were supported by transcriptome analysis using root tips. These results suggest that the collapse of hormonal crosstalk in the stele including the stem cells occurred in the rpt5a mutant, especially under high-B stress. Treatment with the auxin signaling inhibitor α-(phenyl ethyl-2-one)-indole-3-acetic acid (PEO-IAA) reduced sensitivity to high-B stress in the wild type and restored the RAM morphology in the rpt5a mutant under the high-B stress condition. In addition, cytokinin treatment conferred the rpt5a mutant with tolerance to high-B stress in RAM morphology. It is concluded that 26SP containing RPT5a is required for maintenance of auxin/cytokinin balance in the stele, which is crucial for preventing defects in RAM morphology under high-B stress.

Keywords: auxin, boron, cytokinin, root apical meristem, 26S proteasome

# INTRODUCTION

fpls-10-00590 May 10, 2019 Time: 14:48 # 2

The micronutrient boron (B) performs indispensable roles for plant growth and development. However, B is toxic when it is accumulated to a high concentration in plant tissues, and consequently causes growth retardation. Given that the optimal range in B concentration in the soil is considered to be narrow, a limitation on crop productivity is often observed, especially in arid and semi-arid areas, including South Australia, Turkey, Mediterranean countries, California, and Chile in addition to regions with high-B soils of anthropogenic origin caused by agricultural irrigation and over-fertilization (Nable et al., 1997; Kot, 2007). Therefore, it is important to understand the mechanism of plant tolerance of high-B stress at the molecular level, which will potentially aid in breeding crops tolerant of B toxicity.

The root tip is the primary belowground target of B toxicity. In wheat roots, application of a high B concentration to mature root tissues other than the root tip has only slight inhibitory effect on root growth (Reid et al., 2004). Two fundamental developmental processes maintain root growth: mitotic cell division in the root apical meristem (RAM), and cell elongation after mitosis (Scheres et al., 2002). Inhibition of cell division in the RAM is a critical cause of B toxicity in roots. In roots of Arabidopsis thaliana, high-B treatment decreases the RAM size through inhibition of cell division activity (Sakamoto et al., 2011, 2018; Aquea et al., 2012). The reduction in cell division activity is also observed in root tips of Vicia faba treated with a high B concentration (Liu et al., 2000), The inhibition of cell division is attributed to DNA damage caused by high-B stress in the RAM (Sakamoto et al., 2011). We previously reported that a chromosomal protein complex (condensin II) and a regulatory particle AAA-ATPase 5a (RPT5a) of 26S proteasome (26SP) are essential for amelioration of high-B-dependent DNA damage and the maintenance of RAM size in A. thaliana (Sakamoto et al., 2011, 2018).

The 26SP protein complex is composed of the 20S core particle (CP) and 19S regulatory particle (RP). The complex regulates numerous biological processes, such as cell-cycle progression, plant hormonal responses, and signaling in response to abiotic and biotic stimuli through the selective degradation of polyubiquitin-tagged proteins (Sadanandom et al., 2012). The RPT5a gene belongs to the RP, which functions in the recognition and unfolding of target proteins, and translocation of target proteins to the CP, which shows protease activities (Sadanandom et al., 2012). Loss of function of RPT5a gives rise to high-B hypersensitivity that accompanies severe defects in RAM morphology, including disorder of cell alignment around the stem cell niche (Sakamoto et al., 2018). The defects in RAM morphology in the rpt5a mutant are partially attributed to failure in degradation of a chromatin remodeling factor, BRAHMA (BRM) (Sakamoto et al., 2018). In the wild type, however, enhanced accumulation of BRM increases sensitivity to high-B stress in root growth without exhibiting the defects in RAM morphology (Sakamoto et al., 2018). This finding implies that RPT5a is also involved in pathways other than the BRM-dependent pathway to maintain RAM morphology under high-B stress.

Maintenance of the stem cell niche is the basis for formation of proper RAM morphology, which is controlled by several hormonal pathways (Lee et al., 2013). Auxin and cytokinin are the principal regulators of RAM maintenance (Moubayidin et al., 2009). Auxin and cytokinin have an essentially antagonistic functional relationship; cytokinin stimulates cell differentiation by suppression of auxin signaling and transport, whereas auxin promotes cell division by inactivation of cytokinin signaling (Lee et al., 2013). Recent studies have revealed crosstalk between auxin and cytokinin signaling in the stem cell niche. The collective activities and topology of the PIN-FORMED (PIN) proteins and the AUXIN RESISTANT 1 (AUX1)/LIKE AUX1 (LAX) family transporter proteins contribute to form the auxin gradient along the RAM and the auxin maximum at the quiescent center (QC) (Blilou et al., 2005; Grieneisen et al., 2007; Ugartechea-Chirino et al., 2010), which produces stem cell initials. This auxin distribution determines the spatial patterning of PLETHORA (PLTs) expression in RAM, which is crucial for the specification of the stem cell niche and to control the proliferation of stem cell daughter cells (Scheres and Krizek, 2018). Auxin-mediated expression of PLT1 and PLT2 is regulated by AUXIN RESPONSE FACTOR 5 (ARF5) and ARF7 during the development of the embryonic root (Aida et al., 2004). The ARF5 target TARGET OF MONOPTEROS 5 (TMO5) and its homolog T5L1 promote the periclinal division of procambium cells that function as stem cells of vascular tissues in association with LONESOME HIGHWAY (LHW) in the cytokinin-dependent pathway (De Rybel et al., 2013; Ohashi-Ito et al., 2014).

The ubiquitin/26SP pathway plays critical roles in the regulation of various hormonal signaling pathways (Kelley and Estelle, 2012). The 26SP protein is involved in ARF transcription through degradation of the repressor for ARF genes, AUX/IAA family proteins ubiquitin-tagged by a E3 ligase complex SCFTIR/AFB in response to auxin perception by TRANSPORT INHIBITOR RESPONSE (TIR1) (Kong et al., 2016). The degradation of AUX/IAA proteins is also mediated by PROTEASOME REGULATOR 1 (PTRE1) dependent on promotion of 26SP activity (Yang et al., 2016). It is suggested that PLT1 and PLT2 expression in primary roots is also regulated by BRM action (Yang et al., 2015), whose degradation by 26SP requires RPT5a function (Sakamoto et al., 2018). In addition, the RP subunits RP non-ATPase 10 (RPN10) and RPN12a are involved in the degradation of ARABIDOPSIS RESPONSE REGULATOR 1 (ARR1) and ARR5 (Smalle et al., 2002, 2003; Kurepa et al., 2014). ARR1 is a transcriptional activator that promotes the expression of cytokinin-responsive genes, whereas ARR5 acts in the repression of cytokinin responses (Kurepa et al., 2014), suggesting that 26SP functions in balancing activation and inhibition and/or spatiotemporal regulation of cytokinin responses in roots.

In this study, we characterized the defects in RAM morphology exhibited by the rpt5a mutant under high-B stress and showed that 26SP containing RPT5a is involved in adjustment of the activity of TIR1/AFB-dependent auxin signaling to the appropriate level required for the maintenance of the stem cell niche, especially under high-B conditions.

In addition, 26SP containing RPT5a is also involved in the regulation of cytokinin responses, which may repress auxin responses and be crucial for stem cell proliferation and consequently the size of the stele in RAM of A. thaliana.

#### MATERIALS AND METHODS

#### Plant Materials and Growth Condition

The rpt5a-4 and rpt5a-6 mutants of A. thaliana (background Columbia; Col-0) were established previously (Sakamoto et al., 2018). For reporter analysis, rpt5a-4 was crossed with reporter lines harboring pWOX5::erGFP (Blilou et al., 2005), DR5::GFP (Ottenschläger et al., 2003), pPIN1::PIN1-GFP (Benková et al., 2003), and TCS::GFP (Zürcher et al., 2013). Double-homozygous lines were established from their F<sup>2</sup> progeny and used for the analysis. In all experiments, seeds were sown on media containing MGRL solution, 1% (w/v) sucrose, and 1.5% (w/v) gellan gum. Boric acid was used to adjust the B concentration in the medium. After 3 days incubation at 4◦C, the plates were placed vertically in a growth chamber (16-h light/8-h dark cycle, 22◦C) until analysis.

#### β-Glucuronidase (GUS) Reporter Line and GUS Staining

The promoter fragment containing 2,735 bp upstream of the start codon of RPT5a was amplified from Col-0 genomic DNA using the primers 5<sup>0</sup> -caccCTCTAGAGGTTCCCAATTAG-3<sup>0</sup> and 5<sup>0</sup> -TCTTCGAAGCTTGACGTATCG-3<sup>0</sup> , and cloned into the pENTRTM/D-TOPO <sup>R</sup> vector following the manufacturer's protocol (Invitrogen, Carlsbad, CA, United States). The cloned promoter fragment was subsequently subcloned into pMDC162, a GatewayTM destination vector containing a β-glucuronidase (GUS) gene, by LR recombination with LR clonase II (Invitrogen) following the manufacturer's protocol. The constructs were inserted into Agrobacterium tumefaciens (strain GV3101::pMP90) and used to transform Col-0 plants. Transgenic plants were selected on half-strength Murashige and Skoog medium supplemented with 1% sucrose, 20 µg/mL hygromycin B, and 250 µg/mL claforan. Transgenic T<sup>3</sup> plants harboring homozygous T-DNA were used for subsequent analyses.

To detect GUS activity, seedlings were stained with a solution containing 100 mM Na2HPO<sup>4</sup> (pH 7.0), 0.1% Triton X-100, 2 mM K3Fe[CN]6, 2 mM K4Fe[CN]6, and 0.5 mg/mL 5-bromo-4 chloro-3-indolyl-β-D-glucuronic acid for 1 h at 37◦C. The GUSstained seedlings were incubated in clearing solution (80% chloral hydrate and 10% glycerol) overnight at 4◦C. Images of GUSstained roots were captured using a stereomicroscope (SZH10; Olympus, Tokyo, Japan) equipped with a digital CCD camera.

#### Root Elongation Assay

Five-day-old seedlings pre-incubated vertically on normal MGRL medium were transferred to fresh medium containing indicated concentrations of B, indole-3-acetic acid (IAA), α-(phenyl ethyl-2-one)-indole-3-acetic acid (PEO-IAA) (Hayashi et al., 2012), and trans-zeatin (tZ). Positions of primary root tips were marked on the plates with a pen. After incubation for an additional 4 days, the length of the newly elongated primary roots from the marked positions was determined using ImageJ ver.1.51h software<sup>1</sup> .

#### Confocal Fluorescence Microscopy

To observe the RAM structure and expression patterns of reporter genes, roots of 9-day-old seedlings subjected to the 4-day B treatments were stained with propidium iodide (PI) (10 mg/mL; Molecular Probes, Eugene, OR, United States) for 5 min.

For analysis of the number of cell files in the stele, roots of 9-day-old seedlings subjected to the 4-day B treatments were fixed with 4% formaldehyde in PEM buffer [50 mM PIPES (pH 6.8), 2 mM EGTA (pH 7.0), 2 mM MgSO4] for at least 24 h at 4◦C. The fixed plants were treated with PEMT buffer [1% (v/v) Triton-X 100 in PEM buffer] for 20 min. The nuclei were stained with 1:5000 SYBRTM Green I in PEMT buffer for 10 min. After washing twice with PEM buffer, cell walls were stained with Calcofluor White (Sigma-Aldrich, Tokyo, Japan) for 30 min. After washing twice with PEM buffer, plants were mounted in 20% 2,2<sup>0</sup> -thiodiethanol in PEM for observation.

The samples were examined using a confocal fluorescence microscope (FV-1200; Olympus). Signals for PI were excited using the 559 nm wavelength and detected using a 575–675 nm band bass filter. Signals for GFP and SYBR Green I were excited using the 473 nm wavelength and detected using a 490–540 nm band bass filter. Signals for Calcofluor White were excited using the 405 nm wavelength and detected using a 430–455 nm band bass filter.

#### RNA Sequencing and Data Analysis

The primary root tips (∼1 cm from the tip) were collected from Col-0 and rpt5a-4 plants treated with 0.03 or 3 mM B. Total RNA was extracted from collected root tips using RNeasy <sup>R</sup> Plant Mini Kit (Qiagen, Valencia, CA, United States) and treated with RNase-free DNase (Qiagen). A total of 500 ng RNA was subjected to mRNA isolation and subsequent library preparation with the NEBNext <sup>R</sup> Poly(A) mRNA Magnetic Isolation Module (NEB, Ipswich, MA, United States) and NEBNext <sup>R</sup> Ultra RNA Library Prep Kit for Illumina (NEB), respectively. The prepared library was sequenced using the NextSeqTM 500 system (Illumina, San Diego, CA, United States) in the single-end mode with a read length of 75 bp. Four independent biological replicates were analyzed for each genotype. The sequencing data are deposited with DDBJ (accession number DRA008073).

The produced bcl files were converted to fastq files by bcl2fastq (Illumina). Quality-filtered reads were mapped onto the Arabidopsis reference genome (TAIR10) by bowtie 1.2.1 (Langmead et al., 2009) with the following parameters: '-all best -strata.' Up- and down-regulated genes were extracted by R ver.3.2.0 software using the R package edgeR (Robinson et al., 2010), treating biological quadruplicates as paired samples. Genes showing a p-value less than 0.05 and more than two-fold change in comparison with the data for Col-0 under the normal B

<sup>1</sup>http://rsb.info.nih.gov/ij/

condition (0.03 mM B) were identified. Gene ontology (GO) analysis was performed using GO Enrichment Analysis<sup>2</sup> .

#### RESULTS

#### RPT5a Is Involved in Maintenance of the Stem Cell Niche Especially Under High-B Stress

Previously, we revealed that RPT5a proteins are expressed in the entire RAM and are essential for maintenance of RAM morphology, particularly under the high-B condition (Sakamoto et al., 2018). To further understand the function of RPT5a in RAM maintenance under high-B stress, we assessed the promoter activity of RPT5a using plants with pRPT5a::GUS (**Figure 1**). Under the normal B condition (0.03 mM B), GUS expression was detectable only around the stele, which consists of the primary vascular system, except for the stem cell niche in the RAM. In plants treated with 3 mM B, GUS expression was also detected in the cells outside of the stele in the RAM and staining in the stele extended toward the stem cell niche. Treatment with 6 mM B further enhanced GUS expression in the same regions as observed in the 3 mM B condition. These results indicated that RPT5a in the RAM is responsive to high-B stress.

Given that strong promoter activity of RPT5a was observed in the stele even under the normal B condition, we investigated the involvement of RPT5a in stele organization by measuring the width of the stele in the RAM. The width of the stele was comparable between the wild type and rpt5a mutants under the normal B condition (**Figure 2A**). However, high-B stress caused reduction in the stele width only in rpt5a mutants

<sup>2</sup>http://geneontology.org/page/go-enrichment-analysis

(**Figure 2A**). Consistently, significant reduction in the number of cell files was observed only in rpt5a-4 treated with high-B stress (**Figure 2B**). These results suggested that RPT5a is involved in the maintenance of stele organization under the high-B condition but not under the normal B condition.

The number of cell files in the stele is determined by the regulation of the periclinal cell divisions of procambium cells produced around the stem cell niche (Levesque et al., 2006). It was expected that defects in the stele organization in rpt5a mutants under high-B stress was attributable to the disorganization of the stem cell niche where the RPT5a promoter was activated under high-B stress (**Figure 1**). Indeed, root tips of rp5a-4 showed severe defects in alignment of cell files around the stem cell niche under the high-B condition (**Figures 2B,C**) (Sakamoto et al., 2018). In addition, we observed the highly disordered expression of pWOX5::erGFP, which is a marker for QC cells (Blilou et al., 2005), in rpt5a mutants treated with high-B stress, which represented the disorganization of the stem cell niche (**Figure 2C**). Although slightly broader expression of pWOX5::erGFP was observed (**Figure 2C**), the stele width was not affected in rpt5a mutants under the normal B condition (**Figure 2A**). Taken together, these results suggested that the function of RPT5a in the maintenance of the stem cell niche is especially crucial for organization of the RAM under high-B stress.

### RPT5a Is Crucial for Maintaining Auxin Responses and Transport in the RAM Under High-B Stress

The feedback circuit between WUSCHEL RELATED HOMEOBOX (WOX5) and an auxin response repressor, IAA17, is essential for patterning of the auxin gradient in the root tip, which is crucial for organization of the stem cell niche (Tian et al., 2014). Therefore, the dysregulation of WOX5 expression led us to speculate on the collapse of proper auxin response and distribution in the rpt5a mutant. To evaluate this hypothesis, we used the DR5::GFP reporter for auxin accumulation (Ottenschläger et al., 2003), and the pPIN1::PIN1-GFP auxin transport reporter, which acts in the stele (Benková et al., 2003). High-B treatment slightly increased DR5::GFP expression in the stele of the RAM in the wild type (**Figure 3A**). In contrast, the rpt5a mutant displayed highly enhanced expression of DR5::GFP in the stele of the RAM and differentiated regions under the high-B condition (**Figure 3A**). The rpt5a-4 mutant showed reduced intensity of PIN1 expression in the stele of the RAM, under both the normal and the high-B conditions (**Figure 3B** and **Supplementary Figure S1**). Thus, we concluded that the auxin response within the RAM was altered in the rpt5a mutant, especially under the high-B condition.

The periclinal division of procambium cells, which determines the width of the stele, is regulated cytokinin (Ohashi-Ito et al., 2014). Reductions in the width and number of cell files in the stele in the rpt5a-4 mutant under high-B stress (**Figures 2A,B**) implied the alteration of cytokinin responses around the stem cell niche in this mutant. Analysis of the cytokinin signaling reporter TCS::GFP (Zürcher et al., 2013) revealed that the loss

of function of RPT5a resulted in the extremely low level of TCS::GFP expression in the stem cell niche above the QC region, irrespective of B condition (**Figure 3C**). These results suggest that

50 µm. Magenta, PI-stained cell walls; green, GFP fluorescence.

the cytokinin response is highly repressed in the rpt5a mutant, but it is unlikely to be directly associated with the disorganization of the RAM in the rpt5a mutant under high-B stress.

#### RPT5a Is Crucial for Maintaining Auxin and Cytokinin-Responsive Gene Expression Level in the Root Tip Under High-B Stress

To further evaluate the involvement of RPT5a in the regulation of auxin and cytokinin responses under the high-B condition, we performed RNA sequencing (RNA-seq) analysis using ∼1 cm-long root tips subjected to high-B treatment for 4 days. A total of 24,758 genes were analyzed. We identified 1,410 (wild type) and 3,423 (rpt5a-4) up-regulated (≥2-fold, p < 0.05) genes and 345 (wild type) and 3,085 (rpt5a-4) down-regulated genes (≤0.5-fold, p < 0.05) in response to high-B treatment (**Figure 4A**). Among the genes differentially expressed in the rpt5a-4 mutant, 2,268 up-regulated genes and 2,875 downregulated genes were not detected in the wild type subjected to high-B stress (**Figure 4A**), which indicated that the loss of function of RPT5a was predominantly responsible for the substantial changes in gene expression profiles in root tips under high-B stress. The gene ontology analysis showed that genes associated with primary and secondary metabolic processes were significantly enriched in the high-B-dependent up-regulated gene sets in both the wild type and the rpt5a-4 mutant (**Supplementary Figure S2A** and **Supplementary File S1**). In the case of the gene set down-regulated by high-B stress, genes associated with metal ion homeostasis were highly enriched in the wild type, whereas genes associated with cell wall organization were highly enriched in the rpt5a-4 mutant (**Supplementary Figure S2B** and **Supplementary File S1**).

Next, we focused on auxin- and cytokinin-responsive genes. Among genes up- and down-regulated by high-B stress, the number of genes annotated with "response to auxin" (GO: 0009733) was greater in the rpt5a-4 mutant (94 up-regulated, 68 down-regulated) compared with those in the wild type (31 up-regulated, 16 down-regulated) (**Figure 4B**). We observed that ARF5, a transcriptional activator for auxin-responsive genes that is expressed in the stele (Rademacher et al., 2011), was up-regulated even in the rpt5a-4 mutant under the normal B condition and was further induced by high-B stress in both the wild type and the rpt5a-4 mutant (**Supplementary Figure S2C** and **Supplementary File S2**). In addition, a gene set specifically up-regulated in rpt5a-4 including ARF11 which is also expressed in RAM including the stele (Rademacher et al., 2011) (**Supplementary Figure S2C** and **Supplementary File S2**). In contrast, a gene set specifically down-regulated in the rpt5a-4 mutant involved auxin transport genes, such as PIN2 and PIN4 (Blilou et al., 2005; Grieneisen et al., 2007), and negative regulators of auxin-responsive gene expression, such as IAA6 and IAA7 (Tiwari et al., 2001) (**Supplementary Figure S2D** and **Supplementary File S2**). Regarding genes annotated with "response to cytokinin" (GO: 0009735), the number of high-B-dependent up- and down-regulated genes in

FIGURE 4 | Transcriptome analysis of the rpt5a mutant under high-B stress. (A) Number of up-regulated (left) and down-regulated (right) genes in each condition. Genes that showed more than twofold change (p < 0.05) in expression relative to its expression level in Col-0 under the normal B condition were analyzed. (B,C) Number of high-B stress dependent up-regulated (left) and down-regulated (right) genes annotated with "response to auxin" (GO: 0009733) (B) and "response to cytokinin" (GO: 0009735) (C) under each condition. (D,E) Differences in expression levels of all genes annotated with "response to auxin" (D) and "response to cytokinin" (E) among all conditions. Heat maps show relative expression levels for the genes by z-scores of read counts per million mapped reads.

the rpt5a-4 mutant (26 and 38 genes, respectively) was also greater than that in the wild type (10 and 8 genes, respectively). A gene set specifically down-regulated in the rpt5a-4 mutant included the cytokinin receptor AHK5 (Kurepa et al., 2014), and positive and negative regulators for cytokinin responses, namely ARR12 and ARR8, respectively (To and Kieber, 2008) (**Supplementary Figure S2E** and **Supplementary File S2**). Although the difference in expression was not less than 0.5 fold, the expression of WOL/AHK4/CRE1, a cytokinin receptor involved in stele development (Mähönen et al., 2000), was also significantly reduced in the rpt5a-4 mutant, especially under the high-B condition (**Supplementary Figure S2E**). When we compared the differences in expression of whole genes annotated as auxin- and cytokinin-responsive genes, both expression profiles were drastically different, especially for rpt5a-4 under high-B stress (**Figures 4D,E**). In conclusion, these data further support the hypothesis that RPT5a acts in modulating auxin response and distribution and cytokinin response, especially under the high-B condition. It should be noted that we could not completely exclude the possibility that the differences in gene expression were merely attributable to the differences in composition of cells in 1-cm-long root tips.

### Repression of TIR1/AFB-Dependent Auxin Signaling Pathway by RPT5a Is Crucial for Maintenance of RAM Morphology Under High-B Stress

We investigated the association of the enhanced auxin responses with the hypersensitivity of the rpt5a mutant to high-B stress. First, we analyzed the effect of an exogenous synthetic auxin IAA on the sensitivity of root growth to high-B stress. Under the normal B condition, both rpt5a-4 and rpt5a-6 mutants displayed highly reduced root elongation, even under 1 nM IAA treatment in which the wild type grew normally (**Figure 5A**), indicating that the rpt5a mutant is more sensitive to auxin with respect to root elongation. Simultaneous treatment of IAA and 1.5 mM B reduced the apparent sensitivity of root growth to high-B stress expressed as the root elongation ratio relative to that under the normal B condition in rpt5a mutants (∼60–65% without IAA and 75–85% with 10 nM IAA treatment), but not in the wild type (∼82–87% in all conditions) (**Figure 5A**). These results suggested that B has actions similar to those of IAA in roots of the rpt5a mutant. Next, we investigated the involvement of the TIR1/AFB-dependent auxin signaling pathway in the sensitivity of root growth to high-B stress using PEO-IAA, an inhibitor of the TIR1/AFB-dependent auxin signaling pathway. The negative effect of PEO-IAA on root growth under the normal B condition was comparable between the wild type and the rpt5a mutants. However, the effect of PEO-IAA under the high-B condition differed between the wild type and the rpt5a mutants. In rpt5a-4 and rpt5a-6, 1.25 µM PEO-IAA treatment alleviated the inhibitory effect of high-B stress on root growth (**Figure 5B**). In addition, the negative effect of higher concentrations of PEO-IAA on root growth was less than that in the wild type under the high-B condition (**Figure 5B**). The present results suggested that the enhancement in auxin responses through the TIR1/AFBdependent auxin signaling pathway is a critical cause of the severe inhibition of root growth in the rpt5a mutant under high-B stress.

We observed the effect of enhanced auxin responses on RAM morphology in the rpt5a mutant under the high-B condition. Concurrently, the expression pattern of pWOX5::erGFP as a marker for maintenance of the stem cell niche was observed. Although a high concentration of IAA (20 nM) caused severe inhibition of root growth in rpt5a mutants, irrespective of B condition (**Figure 5A**), the RAM morphology and WOX5 expression in rpt5a-4 were similar to those observed without IAA treatment (**Figure 5C**). In contrast, 1.25 µM PEO-IAA treatment prevented the alteration of RAM morphology and the extent of cells expressing pWOX5::erGFP under high-B stress in the rpt5a-4 mutant (**Figure 5C**), which is likely associated with the improved root growth (**Figure 5B**). Similar effects were observed in rpt5a-4 treated with a higher concentration of PEO-IAA (5 µM), even though root growth was not recovered (**Figure 5B**). We should note that improvement of RAM maintenance by PEO-IAA treatment was not accompanied by increase in the width of the stele (**Figure 5C**), which suggested that maintenance of the stele width requires certain cellular processes subsequently for maintenance of the stem cell niche.

Analysis of DR5::GFP expression confirmed that IAA treatment did not further enhance the local auxin maximum in the RAM of rpt5a-4 upon high-B stress (**Supplementary Figure S3**), consistent with the results observed for RAM morphology (**Figure 5C**). Even 1 µM PEO-IAA treatment significantly inhibited DR5 activity (Hayashi et al., 2012). However, DR5::GFP expression remained high in the RAM of rpt5a-4 under the high-B plus PEO-IAA conditions (**Supplementary Figure S3**). Considering that DR5-inducible expression depends on TIR1/AFB (De Smet et al., 2007) and PEO-IAA inhibits TIR1/AFB activity (Hayashi et al., 2012), the increased DR5::GFP expression in rpt5a-4 under the high-B plus PEO-IAA conditions might be mediated by a different pathway other than TIR1/AFB auxin signaling. It is also likely that the inhibition of certain components of auxin responses is sufficient to prevent the defects in RAM maintenance.

Taken together, these results established that RPT5a is involved in the repression of the TIR1/AFB-dependent auxin signaling pathway, which is crucial for the maintenance of RAM morphology under high-B stress.

# RPT5a Is Involved in Regulation of Auxin/Cytokinin Balance in RAM Maintenance, Especially Under High-B Stress

Auxin and cytokinin function antagonistically in controlling cell division and differentiation for root growth (Lee et al., 2013). Given that TCS::GFP expression in the RAM was highly reduced in the rpt5a mutant, irrespective of B condition (**Figure 2A**), we speculated that defective cytokinin signaling causes an inclination of the auxin/cytokinin balance toward auxin, especially under the high-B condition. To evaluate this hypothesis, we analyzed the effect of treatment with the synthetic cytokinin tZ on root growth and morphology. Interestingly, we observed that 50 nM tZ improved root growth in the rpt5a-4 and rpt5a-6 mutants under the normal B condition, whereas root growth of the wild type was inhibited (**Figure 6A**). Similarly, 50 nM tZ treatment improved root growth in rpt5a mutants but not in the wild type (**Figure 6A**). These results indicated that reduced root growth in rpt5a mutants was attributable to reduced cytokinin signaling in the RAM. We also observed that rpt5a mutants displayed a reduced ratio in root elongation under the high-B plus tZ conditions (∼70– 75% without tZ and ∼60–70% with tZ). Therefore, with regard to the sensitivity of root growth of rpt5a mutants to high-B stress, tZ alone was ineffective in negating the hypersensitivity to high-B stress of root growth. However, confocal microscopic analyses of RAM morphology indicated that cytokinin treatment conferred the rpt5a-4 mutant with tolerance to high-B stress with regard to maintenance of the stem cell niche represented by PI staining and pWOX5::erGFP expression (**Figure 6B**) and the width of the stele (**Figure 6C**). Taken together, it is plausible that the reduced cytokinin response is a cause of enhanced auxin signaling leading to the defects in RAM maintenance in the rpt5a mutant under high-B stress. Similar to the effect of PEO-IAA treatment, the DR5::GFP expression level was increased in the RAM of rpt5a-4 under the high-B plus tZ conditions (**Supplementary Figure S4**), which similarly implied the possibility that enhancement of certain components of auxin responses may be sufficient to cause disorganization of the RAM in the rpt5a mutant under high-B stress.

means ± SE (n = 19–22, <sup>∗</sup>p < 0.05, ∗∗p < 0.01, Student's t-test). (B) Effects of tZ treatment on RAM morphology and expression patterns of pWOX5::erGFP in the RAM under the normal (0.03 mM B) and high-B (1.5 mM B) conditions. Scale bars, 50 µm. Magenta, PI-stained cell walls; green, GFP fluorescence. (C) Effects of tZ treatment on the width of the stele at the border of the RAM in Col-0 and rpt5a-4 under the normal and high-B conditions. Values are means ± SE (n = 10, p < 0.05, one-way ANOVA and Tukey's HSD).

# DISCUSSION

fpls-10-00590 May 10, 2019 Time: 14:48 # 11

In this study, the repression of certain components of auxin responses by RPT5a was indicated to be crucial for the maintenance of RAM morphology under high-B stress. In A. thaliana, in addition to the majority of other RP subunits, RPT5 is encoded by two gene copies, RPT5a and RPT5b (Book et al., 2010). Given that the rpt5b mutant is not hypersensitive to high-B stress (Sakamoto et al., 2018), it is considered that 26SP containing RPT5b does not function in the repression of auxin responses in roots under high-B stress.

Considering the well-known function of 26SP in degradation of AUX/IAA family proteins that inhibit auxin responses (Kong et al., 2016), it is expected that defects in 26SP activity lead to increased stabilization of such proteins and consequently further repress the auxin responses. Indeed, the protease activities of 26SP are reduced in rpt5a mutants, irrespective of B condition (Sakamoto et al., 2018). However, DR5::GFP expression and transcriptome data indicated that auxin responses were enhanced compared with those of the wild type, especially under the high-B condition (**Figures 3**, **4** and **Supplementary Figure S2**). This finding could be explained by the existence of certain factors, such as PTRE1, that enhance 26SP activities regarding the degradation of AUX/IAA family proteins independently of the TIR1/AFB pathway (Yang et al., 2016). Interestingly, consistent with this possibility, the expression of PTRE1 was prone to be up-regulated in the rpt5a-4 mutant (**Supplementary Figure S5**). Combined with the observed improvement of RAM morphology in the rpt5a mutant under high-B stress by PEO-IAA treatment, which stabilizes AUX/IAA proteins (**Figure 5**), it is considered that 26SP functions to fine-tune the homoeostasis of AUX/IAA proteins to maintain the auxin responses required for RAM maintenance at the appropriate level under the high-B condition.

In the rpt5a mutant, ARF5 was highly up-regulated under high-B stress (**Supplementary Figure S2**), which is upstream of TMO5 and T5L1 (De Rybel et al., 2013). TMO5–LHW and T5L1– LHW regulate both the activation and inhibition of periclinal division of procambium cells in the stem cell niche in a cytokinindependent manner (Ohashi-Ito et al., 2014). However, rpt5a mutants showed reduced width and cell files of the stele, which is determined by the number of cell files in the stem cell niche (**Figure 2**). The phenotype of the rpt5a mutant may be due to the reduced cytokinin responses, which was suggested by low TCS::GFP expression and insensitivity to external cytokinin in terms of root growth (**Figures 3**, **5**). We observed that TMO5 was up-regulated and its downstream gene AHP6, a repressor of the cytokinin responses required for periclinal cell division, was also up-regulated, whereas the expression level of activators of cytokinin responses, LOG3 and LOG4, was unchanged (**Supplementary Figure S5**). The present data also suggested that the activity of cytokinin-dependent periclinal cell divisions was reduced in the rpt5a mutant, especially under high-B stress. Promotion of cytokinin responses by treatment with tZ rescued the width of the stele in the rpt5a mutant under the high-B condition (**Figure 5**). This result suggested that 26SP containing RPT5a is involved in switching of the cytokinin response toward activation, although the molecular mechanism remains unknown. Based on this hypothesis, the reason that the width of the stele was not rescued by PEO-IAA treatment in the rpt5a mutant under high-B stress (**Figure 5**) may be naturally low cytokinin responses for stem cell division.

The formation of an auxin gradient in the RAM mediated by a series of auxin transporters is indispensable for maintenance of the stem cell niche (Scheres and Krizek, 2018). A chromatin remodeling factor, BRM, directly activates auxin efflux transporters PIN1, PIN2, PIN3, PIN4, and PIN7, which is required for stem cell maintenance (Yang et al., 2015). Our recent finding of increased accumulation of BRM in the rpt5a mutant, especially under high-B stress (Sakamoto et al., 2018), implies that auxin response is enhanced. However, this finding is partially contradictory to the present results. We observed that PIN1 expression, and PIN2 and PIN4 transcription are highly repressed in the rpt5a mutant under high-B stress (**Figure 2** and **Supplementary Figure S2**). One possible explanation for this expression pattern of PIN paralogs is the reduction of production and/or maintenance of cells that have the ability to express PIN genes, which is attributable to the defects in stele development in rpt5a mutant. Alternatively, it is possible that in the context of high-B stress, BRM might not act predominantly in gene regulation. Instead, it might act in loosening of chromatin structure, which becomes a critical cause of DNA damage under high-B stress, as we previously observed (Sakamoto et al., 2018).

The highly enhanced auxin response in the rpt5a mutant evokes the question; why is the auxin response up-regulated? Auxin is crucial for condensation of chromatin in proliferative cells, which prevents the incidence of DNA damage and affects gene regulation associated with chromatin structure (Hasegawa et al., 2018). The rpt5a mutant exposed to high-B stress accumulates a high level of DNA damage in the RAM (Sakamoto et al., 2018). Therefore, the enhanced auxin response in the rpt5a mutant may be a defensive response that modulates chromatin integrity and gene regulation in response to the incidence of DNA damage. This may also be applicable to the wild type, as the slight stimulation of auxin signaling represented by DR5::GFP and ARF5 expression was also observed in the wild type under high-B stress (**Figure 2** and **Supplementary Figure S2**).

# CONCLUSION

In conclusion, the present results have established novel roles for 26SP containing RPT5a in regulation of both auxin and cytokinin responses to maintain the stem cell niche and consequent RAM morphology. In this context, 26SP containing RPT5a may regulate multiple proteins to modulate the auxin/cytokinin balance associated with the formation of proper stem cell alignment and the activity of stem cell division that enables the proper development of vascular tissues in the root under the high-B condition.

#### AUTHOR CONTRIBUTIONS

TSa and SM conceived the project and wrote the manuscript. TSa, NS, TF, and SM designed the experiments. TSa, NS, and TSu

performed the experiments. TSa analyzed the data. All authors read and approved the final version of the manuscript.

#### FUNDING

This research was supported by a Japan Science and Technology Agency (JST) CREST grant (JPMJCR13B4) and MXT/JSPS KAKENHI (15H05955 and 15H05962) to SM.

#### ACKNOWLEDGMENTS

We thank Sayoko Mibu, Yuka Asako, and Ritsuko Kamei (Tokyo University of Science) for technical assistance. We thank Robert McKenzie, Ph.D., from Liwen Bianji, Edanz Group, China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00590/ full#supplementary-material

#### REFERENCES


FIGURE S1 | Quantitative analysis of PIN1-GFP expression. Mean intensity of PIN1-GFP signal in the RAM was measured using ImageJ software (n = 3, p < 0.05, one-way ANOVA and Tukey's HSD).

FIGURE S2 | Analysis of RNA-seq data. (A,B) Top 20 enriched Gene Ontology terms among up-regulated (A) and down-regulated (B) genes in Col-0 (left panel) and the rpt5a-4 mutant (right panel) under the 3 mM B condition (p < 0.05, Fisher's exact test). (C,D) Expression levels of selected auxin-responsive up-regulated genes (C) and down-regulated genes (D) in the RNA-seq data. (E) Expression levels of selected cytokinin-responsive down-regulated genes in the RNA-seq data. (C–E) Values are means ± SE (n = 4, <sup>∗</sup>p < 0.05, ∗∗p < 0.01, Student's t-test).

FIGURE S3 | Effects of IAA or PEO-IAA treatment on expression patterns of DR5::GFP in the root apical meristem under the normal (0.03 mM B) and high-B (1.5 mM B) conditions. Five-day-old seedlings were treated with IAA or PEO-IAA for 4 days. Scale bars, 100 µm. Magenta, PI-stained cell walls; green, GFP fluorescence.

FIGURE S4 | Effects of tZ treatment on expression patterns of DR5::GFP in the root apical meristem under the normal (0.03 mM B) and high-B (1.5 mM B) conditions. Five-day-old seedlings were treated with 50 or 100 nM tZ for 4 days. Scale bars, 100 µm. Magenta, PI-stained cell walls; green, GFP fluorescence.

FIGURE S5 | Expression levels of selected genes in RNA-seq data. Values are means ± SE (n = 4, <sup>∗</sup>p < 0.05, ∗∗p < 0.01; Student's t-test).

FILE S1 | List of significantly enriched gene ontology terms.

FILE S2 | List of genes annotated with "response to auxin" and "response to cytokinin" among differentially expressed genes.


response factor gene family. Plant J. 68, 597–606. doi: 10.1111/j.1365-313X. 2011.04710.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 © 2019 Sakamoto, Sotta, Suzuki, Fujiwara and Matsunaga. 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.

# Genome-Wide Characterization and Expression Analysis of Soybean TGA Transcription Factors Identified a Novel TGA Gene Involved in Drought and Salt Tolerance

Bo Li1,2† , Ying Liu<sup>2</sup>† , Xi-Yan Cui<sup>3</sup>† , Jin-Dong Fu<sup>2</sup>† , Yong-Bin Zhou1,2, Wei-Jun Zheng<sup>1</sup> , Jin-Hao Lan<sup>4</sup> , Long-Guo Jin<sup>2</sup> , Ming Chen<sup>2</sup> , You-Zhi Ma<sup>2</sup> , Zhao-Shi Xu<sup>2</sup> \* and Dong-Hong Min<sup>1</sup> \*

#### Edited by:

Yan Guo, China Agricultural University, China

#### Reviewed by:

Yong Hwa Cheong, Sunchon National University, South Korea Soumitra Paul, University of Calcutta, India

#### \*Correspondence:

Zhao-Shi Xu xuzhaoshi@caas.cn Dong-Hong Min mdh2493@126.com

†These authors have contributed equally to this work

#### Specialty section:

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

Received: 06 November 2018 Accepted: 10 April 2019 Published: 16 May 2019

#### Citation:

Li B, Liu Y, Cui X-Y, Fu J-D, Zhou Y-B, Zheng W-J, Lan J-H, Jin L-G, Chen M, Ma Y-Z, Xu Z-S and Min D-H (2019) Genome-Wide Characterization and Expression Analysis of Soybean TGA Transcription Factors Identified a Novel TGA Gene Involved in Drought and Salt Tolerance. Front. Plant Sci. 10:549. doi: 10.3389/fpls.2019.00549 <sup>1</sup> College of Agronomy, Northwest A&F University/State Key Laboratory of Crop Stress Biology for Arid Areas, Yangling, China, <sup>2</sup> Institute of Crop Science, Chinese Academy of Agricultural Sciences (CAAS)/National Key Facility for Crop Gene Resources and Genetic Improvement, Key Laboratory of Biology and Genetic Improvement of Triticeae Crops, Ministry of Agriculture, Beijing, China, <sup>3</sup> College of Life Sciences, Jilin Agricultural University, Changchun, China, <sup>4</sup> College of Agronomy, Qingdao Agricultural University, Qingdao, China

The TGA transcription factors, a subfamily of bZIP group D, play crucial roles in various biological processes, including the regulation of growth and development as well as responses to pathogens and abiotic stress. In this study, 27 TGA genes were identified in the soybean genome. The expression patterns of GmTGA genes showed that several GmTGA genes are differentially expressed under drought and salt stress conditions. Among them, GmTGA17 was strongly induced by both stress, which were verificated by the promoter-GUS fusion assay. GmTGA17 encodes a nuclear-localized protein with transcriptional activation activity. Heterologous and homologous overexpression of GmTGA17 enhanced tolerance to drought and salt stress in both transgeinc Arabidopsis plants and soybean hairy roots. However, RNAi hairy roots silenced for GmTGA17 exhibited an increased sensitivity to drought and salt stress. In response to drought or salt stress, transgenic Arabidopsis plants had an increased chlorophyll and proline contents, a higher ABA content, a decreased MDA content, a reduced water loss rate, and an altered expression of ABA- responsive marker genes compared with WT plants. In addition, transgenic Arabidopsis plants were more sensitive to ABA in stomatal closure. Similarly, measurement of physiological parameters showed an increase in chlorophyll and proline contents, with a decrease in MDA content in soybean seedlings with overexpression hairy roots after drought and salt stress treatments. The opposite results for each measurement were observed in RNAi lines. This study provides new insights for functional analysis of soybean TGA transcription factors in abiotic stress.

#### Keywords: soybean, TGA transcription factor, molecular characterization, abiotic stress response, drought and salt tolerance

**Abbreviations:** ABA, abscisic acid; ABRE, ABA-responsive element; ARE, anaerobic responsive element; ERE, elicitorresponse element; FW, fresh weight; GFP, green fluorescent protein; MS, Murashige and Skoog; ORF, open reading frame; P-box, pathogen-inducible box; qRT-PCR, quantitative real-time-PCR; W-box, wound-inducible box; WT, wild type.

# INTRODUCTION

fpls-10-00549 May 15, 2019 Time: 16:34 # 2

Abiotic stress, such as drought and high salinity, greatly affect plant growth and development. In order to adapt to abiotic stress, plants have evolved complex signal transduction pathways and diverse response mechanisms to protect themselves against cellular damage caused by abiotic stress (Munns and Tester, 2008; Chen and Murata, 2011; Krasensky and Jonak, 2012; Mickelbart et al., 2015; Zhu, 2016; Qi et al., 2018; Yang and Guo, 2018). Under stress conditions, changes in gene expression are the earliest responses in plants, and a number of stressresponsive genes have been noted to have important functions in drought and salt resistance. Among these genes, transcription factors are very important as the proteins they typically encode control the transcription of downstream genes (Singh et al., 2002; Li et al., 2008). Among them, the basic leucine zipper (bZIP) gene family is one of the largest transcription factor families in plants. bZIP genes were classified into 10 groups (A, B, C, D, E, F, G, H, I, and S), along with another two extra groups, J and K, based on the similarity in the basic region and additional conserved motifs (Jakoby et al., 2002; Nijhawan et al., 2008). The TGA (TGACG motif-binding factor) transcription factors belong to group D, which can recognizes as-1- type cis-elements located in the promoter region of the target genes (Katagiri et al., 1989; Zhang et al., 1999; Johnson et al., 2003). Tobacco TGA1a was the first TGA transcription factor cloned from plants and is characterized by the conserved basic region/leucine zipper domain (Katagiri et al., 1989). Subsequently, more TGA transcription factors were identified in various plants (Jakoby et al., 2002; Espín et al., 2012). For TGA proteins, the primary structure of the bZIP domain is conserved, containing an invariant motif N-x7- R/K-x9-L-x6-L-x6-L in the N-terminus, and a bZIP-D box, the motif Yx2RL[RQ]ALSS[LS]W, represents the signature domain of group D in the C-terminus (Jakoby et al., 2002).

A growing body of evidence has shown that members of TGA transcription factor are known to play crucial roles in many biological processes: defense against pathogens and plant development (Johnson et al., 2001; Thurow et al., 2005; Choi et al., 2010; Zander et al., 2012; Gatz, 2013; Wang et al., 2016). Despite this body of work, little has been reported about the functions of TGA transcription factors in plant responses to abiotic stress. The Brassica juncea BjCdR15, an orthologous gene of Arabidopsis TGA3, had an induced expression in response to cadmium and stress from other heavy metals (Fusco et al., 2005). Transgenic Arabidopsis and tobacco overexpressing BjCdR15 exhibited an enhanced tolerance to cadmium through regulation of cadmium uptake and long-distance transport (Farinati et al., 2010). The crab apple MhTGA2, which showed an induced expression in response to low temperature, NaCl, and PEG, enhanced tolerance to salt and osmotic stress in transgenic apple and tobacco (Zhang et al., 2012; Du et al., 2014). Overexpression of AtTGA4 increased tolerance to drought stress by improving nitrate transport and absorption in Arabidopsis (Zhong et al., 2015). These findings suggested that TGA transcription factors may play functions in plant adaptation to various abiotic stimuli, including drought and salt.

Soybean (Glycine max L.), an economically important oil and protein crop, is considered a moderately drought- and salttolerant plant, though its growth and productivity is adversely affected by soil drought and salinity. Given the potential importance of TGA genes in plant responses to abiotic stress, we conducted a genome-wide analysis of the soybean TGA family and investigated the potential functions of TGA genes in plant responses to drought and salt stress.

#### MATERIALS AND METHODS

### In silico Identification of Soybean TGA Transcription Factors

The sequences of known TGA proteins from Arabidopsis were retrieved from the TAIR database. These sequences were used as queries to search against the soybean genome database using the BLASTP program with a threshold E-value cutoff of 1.0 × e −5 (Espín et al., 2012). Then, each candidate soybean TGA sequence used as a query against the Pfam database to confirm its membership in the bZIP family (Finn et al., 2016). Repeat sequences were removed manually. The ExPASy server was used to predict several physio-chemical parameters of TGA proteins such as molecular weights (Mw) and theoretical isoelectric points (pI) (Gasteiger et al., 2003).

#### Sequences Analysis

The exon/intron gene boundaries were analyzed using the Gene Structure Display Server 2.0 (GSDS) tool (Hu et al., 2015). The 2.0 kb sequences upstream of the start codon of soybean TGA genes were extracted from the phytozome database as the regulatory promoter region. Putative cis-acting elements were analyzed using the PlantCARE database (Lescot et al., 2002). Multiple sequence alignments for the predicted protein sequences were performed using ClustalX software (Thompson et al., 1997). MEGA6.0 software was used to construct a phylogenetic tree based on the bootstrap neighbor-joining (NJ) approach followed by 1000 bootstrap replicates (Tamura et al., 2013).

#### Expression Analysis of Soybean TGA Genes in Different Tissues

The analysis of gene expression patterns in different tissues, including roots, root hairs, stem, leaves, nodules, seeds, and flowers, was carried out using transcriptome data obtained from the Phytozome database, and the heatmap was produced with HemI software (Deng et al., 2014).

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

The soybean variety Williams 82 was used for experiments in this study. Seedlings were grown in pots containing mixed soil (humus:vermiculite = 1:1) in a greenhouse with a 14-hlight/10-h-dark photoperiod, 28/20◦C day/night temperatures, and 60% relative humidity. Sixteen-day-old seedlings were subjected to drought and salinity treatments. For drought stress, seedlings were removed from the soil and left to dry

on filter paper. For the high-salinity treatment, the roots of seedlings were immersed in solution containing 200 mM NaCl. For treatments with exogenous ABA, leaves were sprayed with 200 µM ABA solution. Seedlings were sampled for RNA extraction at 0, 1, 3, 6, 12, and 24 h after each respective treatment.

#### RNA Isolation and Quantitative Real-Time PCR (qRT-PCR)

Total RNA was isolated from soybean seedlings and hairy roots, or from Arabidopsis seedlings using Trizol reagent (TaKaRa, Japan). The first strand cDNA was synthesized using the PrimeScript 1st Strand cDNA Synthesis Kit (TaKaRa, Japan) based on the manufacturer's instructions. qRT-PCR was performed on ABI prism 7500 Real-Time PCR system (Applied Biosystem, United States) using SYBR Green Real Master Mix (Tiangen, China) with the following PCR cycles: 95◦C for 15 min, followed by 40 cycles of amplification (95◦C for 10 s, 58◦C for 20 s, and 72◦C for 32 s). The data of qRT-PCR were determined using the 2−11C<sup>t</sup> method according to the cycle threshold (Ct) values (Livak and Schmittgen, 2001). The soybean CYP2 (GmCYP2) (Glyma.12g024700) and Arabidopsis actin2 (AtACT2) (At3g18780) genes were used for normalization. Student's t-test was used to determine significant differences. A level of 0.05 was used for statistical significance. Primer sequences for qRT-PCR analysis were designed using the software tool Primer Premier 5.0. The primer sequences are listed in **Supplementary Table S1**. All reactions were conducted with four biological replicates for each sample.

# Subcellular Localization and Transcriptional Activation Analysis

The ORF of GmTGA17, lacking the stop codon, was amplified using custom primers and fused to the N-terminal region of GFP. The ORF of NtTGA2.2 (AAF06696), encoding a nuclear-localized protein, was cloned into the N-terminal region of RFP as the positive control (Thurow et al., 2005). The genes were driven by CaMV35S promoter. The reconstruction plasmid of NtTGA2.2-RFP and GmTGA17-GFP were cotransferred into Arabidopsis protoplasts, and NtTGA2.2-RFP and the 35S::GFP vector were co-transformed as the control. The fluorescence signal was detected by confocal microscopy (Leica Microsystem, Heidelberg, Germany) after incubating in darkness at 22◦C for 16 h. The primers used are shown in **Supplementary Table S1**.

To perform the transcriptional activity assay in yeast cells, the ORF of GmTGA17 was amplified and cloned into the pGBKT7 vector. pGBKT7-AtDREB2A was the positive control based on previously reported data (Sakuma et al., 2006), while the pGBKT7 empty vector was the negative control. The plasmids were transformed into yeast strain AH109 according to the method described previously (Gietz and Schiestl, 2007). The YeastmakerTM Yeast Transformation System 2 (Clontech, United States) was used for yeast transformation. Transcriptional activity was analyzed using methods established in prior work (Yang et al., 2010).

The transcriptional activity was examined in Arabidopsis protoplast system. The reporter was a plasmid containing the firefly (LUC) gene fused with 5 × GAL4 binding sites under the control of CaMV35S promoter. The other plasmid with the renilla luciferase (REN) gene was used as the internal control (Liu et al., 2018). GAL4-BD and BD-AtDREB2A were used as negative and positive effector. The each effector plasmid was cotransformed with the two reporters. Firefly and Renilla luciferase was quantified at 18 h post-transformation by using the Dual-Luciferase Reporter Assay System following the manufacturer's instructions (Promega, United States).

#### Generation of Transgenic Arabidopsis Plants

In order to produce transgenic Arabidopsis lines, the ORF of GmTGA17, the stop codon, was amplified and cloned into the pCAMBIA1302 vector driven by the CaMV35S promoter. The expression vector pCAMBIA1302-GmTGA17 was transformed into Agrobacterium tumefaciens strain GV3101 and transferred into Arabidopsis Col-0 plants using the floral dip method (Clough and Bent, 1998). The harvested seeds were screened by hygromycin (40 mg/L) resistance. T<sup>3</sup> transformed plants were confirmed by qRT-PCR analysis and used for further study based on the expression level of GmTGA17.

### Agrobacterium rhizogenes-Mediated Transformation of Soybean Hairy Roots

To generate GmTGA17 promoter-GUS construct, a 1.9-kb fragment upstream from the initiation codon was amplified and then used to replace the CaMV35S promoter in pCAMBIA3301. The ORF of GmTGA17 was amplified and cloned into pCAMBIA3301 under the control of the CaMV35S promoter to generate the pCAMBIA3301-GmTGA17 overexpression vector. For construction of the RNAi suppression vector, a 542 bp ligated fragment containing the sequence from CDS positions 549 to 750, the first intron sequence and the reverse complement sequence of the sequence from CDS positions 549 to 750 (**Supplementary Table S1**) was synthesized (Biomed, Beijing, China) and cloned into pCAMBIA3301 to generate pCAMBIA3301 -GmTGA17- RNAi vector. All recombinant vectors were transformed into soybean hairy roots by high-efficiency Agrobacterium rhizogenes-mediated transformation as described previously (Kereszt et al., 2007).

#### Abiotic Stress Tolerance Assessments of Transgenic Arabidopsis and Soybean Hairy Roots

Three transgenic Arabidopsis lines with a higher expression of GmTGA17 were used to evaluate the drought and salt tolerance. For the root growth assay, 5-day-old seedlings were transferred to 1/2 MS medium ( containing 6%, 9% or 12% PEG6000 and 50, 75 or 100 mM NaCl, respectively) for vertical growth under a photoperiod of 16-h-light/8-hdark at 22◦C, 40 µmol m−<sup>2</sup> s −1 light. Seedling roots were scanned with Expression 11000XL scanner and analyzed for total root length with WinRHIZO software after 8 days of treatments. For drought stress in soil, water was withheld from 2-week-old soil-grown seedlings until differences in phenotype were observed, then plants were watered again and recovered for 1 week to count the survival rates. For high-salinity treatment, 2-week-old soil-grown seedlings were soaked in 250 mM NaCl solution for 1 week, and the control group continued to grow under normal conditions. Proline and MDA contents were measured under both stress conditions, and chlorophyll content was measured under high-salinity treatment as described previously (Arnon, 1949; Bates et al., 1973; Cakmak and Horst, 1991).

For water loss measurement, the leaves from 3-week-old plants grown in soil under normal conditions were excised and weighted immediately (initial weight, W0). Subsequently, the detached leaves were maintained in a growth chamber (40% relative humidity) and measured at designated time intervals. The fresh weights measured at each time point were used as Wn. Three replicates were done for each line. The water loss rate was calculated as (W0-Wn)<sup>∗</sup> 100/W0.

The transgenic hairy roots were confirmed by PCR and qRT-PCR analysis. For promoter-GUS analysis, the transgenic hairy roots were immersed in 10% PEG6000, 100 mM NaCl, or water. The GUS activity and the transcript level of GUS in soybean transgenic hairy roots were detected at 0, 3, 6, 9, 12, and 24 h posttreatments. Histochemical staining of GUS was performed as previously described (Jefferson, 1987). In addition, the expression levels of GUS were analyzed by qRT-PCR.

For abiotic stress tolerance assays, the transgenic hairy roots with higher (OE) or lower (RNAi) expression level of GmTGA17 were selected for further study. 10-day-old soilgrown plants with transgenic hairy roots were soaked in 20% (m/v) PEG6000 or 150 mM NaCl solutions for 10 days. The control group continued to grow under normal conditions. After treatments, the transgenic roots were scanned and the total root length and the total root surface were analyzed. Meanwhile, chlorophyll, proline, and MDA contents in leaves were detected as described above.

#### Stomatal Aperture Measurement

Stomatal aperture assay was performed as described previously (Kim and Kim, 2013). The guard cells were photographed using confocal microscopy (Leica Microsystem, Heidelberg, Germany), and the stomatal aperture (ration of width to length) was analyzed using Photoshop CS5 software (Adobe System).

#### Determination of ABA Content

ABA content was measured as described previously (Liu et al., 2014). The Phytodetek-ABA ELISA Kit (Agdia, United States) was used for the determination of ABA according to the manufacture's protocol.


fpls-10-00549 May 15, 2019 Time: 16:34 # 4

#### Statistical Analysis

fpls-10-00549 May 15, 2019 Time: 16:34 # 5

All experiments above were repeated at least three replicates independently. The data were subjected to Student's t-test analysis using functions in Excel 2007. The values are shown as mean ± standard deviation (SD). P-value cut-off of 0.05.

#### Primers

The primers and sequences used in this study are listed in **Supplementary Table S1**.

### RESULTS

#### Identification of TGA Transcription Factors in Soybean

In this study, full-length proteins and conserved domains of 10 Arabidopsis TGAs were used as BLAST query sequences against the soybean genome database. A total of 27 non-redundant, putative TGA genes were identified in the soybean genome, which named GmTGA01-GmTGA27 according to the Gene ID number (**Table 1**). These loci were distributed across 16 chromosomes in the soybean genome, with the exceptions of chromosomes 7, 9, 16, and 17. The size range of the predicted products of these putative GmTGAs spans 290 (GmTGA05) to 517 (GmTGA17) amino acids residues, with molecular weights (Mw) ranging from 32.37 kDa (GmTGA05) to 57.92 kDa (GmTGA11), and protein isoelectric points (pIs) ranging from 6.24 (GmTGA14) to 9.4 (GmTGA13) (**Table 1**).

To evaluate the evolutionary history of GmTGAs and their relationships with TGAs in other plants, the TGA protein sequences were aligned and compared with TGA sequences from Arabidopsis and rice, resulting in a multiple sequence alignment and phylogeny of 53 TGA proteins. The phylogenetic tree showed that 53 TGAs were divided into three major clades: Clade I, II and III (**Figure 1**), which agreed with previous reports on subclassification within TGA family proteins (Murmu et al., 2010; Espín et al., 2012). Clades I and II both contain 8 GmTGAs, while clade III has 11 members of this protein family.

#### Gene Structure and Cis-Acting Elements

To obtain some insight into the gene structure of 27 GmTGA genes, their intron-exon organization was examined. GmTGA genes were interrupted by 6–11 introns (**Supplementary Figure S1**). In order to better annotation and prediction of potential functions of GmTGA genes, we searched for ciselements in their putative promoter regions. The results revealed that all GmTGA genes contained one or more of the ABRE, ARE, MYB, MYC, MBS, or G-box elements in their putative promoter regions. Only the putative promoter region of GmTGA17 carried the dehydration-responsive element (DRE). Low-temperature responsive elements (LTRE) were only observed in the putative promoter regions of GmTGA05, GmTGA25, and GmTGA27 (**Supplementary Table S2**).

### Domain Structure of Soybean TGA Proteins

To analyze the structural characteristics of conserved domains in GmTGA proteins, a multiple sequence alignment was performed using the 27 full-length TGA amino acid sequences from soybean. Multiple alignments showed that the GmTGAs contain a typical bZIP domain, two polyglutamine domains and a unique bZIP-D box (**Supplementary Figure S2**). The bZIP domain consists of a DNA binding domain and a leucine zipper domain. DNA binding domains are located in the N-terminus. Highly conserved amino acid sequences in the DNA binding domains have Glutamine, Alanine and Serine residues, which are regularly spaced by three different amino acids. The leucine zipper domain is formed by a heptad repeat of leucine residues. The GmTGAs also contain the polyglutamine domains I (QI) and II (QII). All members of GmTGAs possess the bZIP-D box, located in the C-terminus (**Supplementary Figure S2B**). These observations are in accordance with previous reports about the structural characteristics of the TGA conserved domains (Jakoby et al., 2002; Farinati et al., 2010; Espín et al., 2012).

#### Expression Profiles of Soybean TGA Genes Are Different for Each Tissue

To analyze the expression patterns of GmTGA genes, the transcriptome data of seven soybean tissues and organs was extracted from the soybean genome database and made a heatmap of GmTGA genes expression profile (**Figure 2**). A total of 26 GmTGA genes were expressed in all evaluated tissues, whereas GmTGA01 were expressed in root, root hairs, stem, nodules, seed and flower tissue, but not in leaves. Additionally, the expression patterns were different between GmTGA genes in same tissue. For example, GmTGA04, GmTGA06, GmTGA10, GmTGA15, GmTGA17, GmTGA20, GmTGA22, and GmTGA23 were expressed at their highest levels in roots, the expressions of GmTGA07, GmTGA09, GmTGA18, and GmTGA21 were highest in stems, GmTGA01, GmTGA02, GmTGA05, GmTGA11, GmTGA14, GmTGA16, and GmTGA25 were expressed most strongly in nodules, GmTGA03, GmTGA12, and GmTGA26

The color scale is shown in the bar at bottom of figure.

( ∗∗P < 0.01) compared with the corresponding controls.

transcription was most enriched in seeds, and GmTGA08, GmTGA13, GmTGA19, GmTGA22, and GmTGA24 were found most highly expressed in flowers. The expression of GmTGA01 reached its highest level in nodules, but was not expressed in leaves. These transcriptional patterns suggested that the expression of these genes might be governed by diverse and potentially tissue-dependent regulatory mechanisms.

#### GmTGA Genes Are Involved in Response to Drought and Salt

To investigate the potential roles of GmTGAs in response to multiple abiotic stresses, the expression profiles of GmTGA genes in soybean plants treated with drought, salt or water were examined by qRT-PCR. The data revealed that the transcript levels of GmTGA genes showed no obvious difference

under non-stress conditions (data not show). Under drought treatment, twelve GmTGA genes had a significantly different transcriptional response to drought (**Figure 3**). Among them, GmTGA15, GmTGA17, GmTGA24, GmTGA26, and GmTGA27 had significant increases in transcript level after drought treatment, especially GmTGA15 (10-fold) and GmTGA17 (22 fold), which reached a peak at 24 and 12 h post-treatment, respectively. GmTGA05, GmTGA07, GmTGA10, GmTGA14, GmTGA16, GmTGA20, and GmTGA22 had significant decreases in transcript level after drought treatment. The other GmTGA genes showed no response to drought (**Figure 3**). Following salt treatment,; seven GmTGA genes had a significantly different transcriptional response to salt (**Figure 4**). Among them, GmTGA10, GmTGA13, GmTGA14, GmTGA17, and GmTGA26 had significant increases in transcript level after salt treatment, especially GmTGA13 (10-fold), GmTGA17 (26 fold), and GmTGA26 (8-fold), which had the highest transcript levels at 12 and 3 h post-treatment, respectively. GmTGA05 and GmTGA07 had significant decreases in transcript level after salt treatment. The other GmTGA genes had no significant changes under salt treatment (**Figure 4**). In light of its dramatic up-regulation in stress conditions, GmTGA17 was selected for further study of the roles of GmTGAs in abiotic stress tolerance.

# GmTGA17 Is a Nuclear Protein With Transcriptional Activation Activity

To follow the subcellular localization of GmTGA17 proteins, the ORF of GmTGA17 was fused to GFP, as well as NtTGA2.2 fused to RFP as a positive control, The fusion genes were subsequently co-transformed into Arabidopsis protoplasts. The visible fluorescence showed that GmTGA17 was located in the nucleus (**Figure 5A**). This finding was in accordance with its putative function as a transcription factor.

In order to test for transcriptional activity of GmTGA17, the fusion plasmids pGBKT7-GmTGA17, the positive control pGBKT7-AtDREB2A, and the negative control pGBKT7 were separately transformed into yeast strain AH109. The transformed yeast strain grew well in non-selective medium SD/–Trp and, using X-α-Gal, it was observed that both the positive control and the cells harboring pGBKT7-GmTGA17 displayed β-galactosidase activity, whereas the negative control exhibited no β-galactosidase activity (**Figure 5B**), suggesting that GmTGA17 possesses transcriptional activation activity in yeast cells.

To further confirm whether GmTGA17 has transcriptional activation activity, we used a dual reporter system for a transient expression assays in Arabidopsis protoplasts (Liu et al., 2018). In this system, the reporter plasmid was constructed by fusing the firefly luciferase (LUC) gene to a 5 × GAL4 binding site. The renilla luciferase (REN) gene under the control of the CaMV35S promoter was used as the internal control (**Figure 5C**). The ORF of GmTGA17 was fused to the GAL4 binding domain (GAL4- BD) as effector plasmid (**Figure 5C**). After co-transformed three plasmids into Arabidopsis protoplasts, the relative LUC activity was determined. The results showed that the relative LUC activity was significantly upregulated when GmTGA17 or the positive control AtDREB2A expressed, compared with the empty vector control (**Figure 5C**). The above findings suggested that GmTGA17 potentially acts as transcriptional activation in plant cell nucleus.

# Promoter Activity of GmTGA17 During Drought and Salt Stress

Analysis by qRT-PCR showed that the expression of GmTGA17 is responsive to drought and salt treatments. We used an established promoter-GUS construct in conjunction with a 1.9 kb GmTGA17 promoter upstream of the transcription start site, to further study the activity of the GmTGA17 promoter under drought and salt treatments. The construct was then transformed into soybean hairy roots. Staining of transgenic hairy roots showed that GUS activity was significantly increased in hairy roots after 6 h of drought or salt treatments compared with the hairy roots treated with water (**Figure 6A**). qRT-PCR date, agreed with the observations of GUS histochemical staining, showed that GUS was also up-regulated (**Figure 6B**). The above results suggested that the promoter activity of GmTGA17 was enhanced by drought and salt treatments.

#### Overexpression of GmTGA17 Improved Arabidopsis Tolerance to Drought

Since GmTGA17 was identified as a predicted stress-tolerance regulator, the drought and salt tolerance phenotypes of transgenic Arabidopsis lines grown in medium and soil were examined. The semi-qRT-PCR result was shown in **Figure 7A**, and T<sup>3</sup> transgenic Arabidopsis lines (OE-1, OE-4, and OE-6) with the high expression of GmTGA17 were selected for further experiments (**Supplementary Figure S3**). PEG6000 was used to simulate drought stress when grown in medium. In the absence of PEG, the total root length and fresh weight showed no difference between transgenic and WT plants. In contrast, when exposed

FIGURE 6 | Expression of the GUS reporter gene under the control of GmTGA17 promoter in transgenic soybean hairy roots. (A) Histochemical assay and (B) relative expression of GUS. The GmTGA17pro-GUS transgenic soybean hairy roots were treated with distilled water, 10% (m/v) PEG6000, or 100 mM NaCl for 6 h before being subjected to histochemical and expression analysis. The expression level of CYP2 was used as quantitative control. Values are means and SD obtained from four biological replicates. The asterisks indicate a statistical significance (∗∗P < 0.01) compared with the corresponding controls.

to 1/2 MS medium with 6% PEG6000 for 8 days, the growth of transgenic and WT plants was strongly inhibited, though the degree of inhibition in the transgenic plants was much lower than that of WT plants. The total root length and fresh weight of transgenic plants were significantly longer and greater than those of WT lines (**Figures 7B,C**). The total root length and fresh weight were not significantly different between transgenic lines and WT plants when grown on 1/2 MS medium supplemented with 9% and 12% PEG (**Figures 7B,C**).

To compare with the above results, we also tested the tolerance of transgenic Arabidopsis lines growing in the soil under drought stress conditions. After a 2-week water-deficit regimen, the leaves of WT plants became shriveled or died, while transgenic lines were only slightly shriveled and showed lower mortality than WT plants (**Figure 7D**). Compared to a 54.32% survival rate of WT plants, the survival rate of transgenic plants was 77.78–91.35% after 1 week of re-watering following treatment (**Figure 7E**). After a 10-day water-withholding treatment, two stress-related parameters, proline and MDA contents, were compared between the transgenic and WT seedlings. The results showed that proline and MDA contents in transgenic seedlings (17.90–18.37 µg/g FW and 11.52–12.07 µM/g FW) were significantly higher and lower than in WT plants (11.37 µg/g FW and 16.49 µM/g FW) under drought stress conditions, respectively (**Figures 7F,G**). Collectively, these results indicated that GmTGA17 may have a potential function in enhancement of the transgenic plants tolerance to drought stress.

#### Overexpression of GmTGA17 Improved Arabidopsis Tolerance to Salinity

To determine if GmTGA17 is in involved in responses to salt stress, we tested the salt tolerance of the transgenic Arabidopsis lines. When grown on 1/2 MS medium, the no difference

in phenotype was observed between the transgenic and WT seedlings (**Figure 8A**). When grown on 1/2 MS medium containing 75 mM NaCl for 1 week, the total root length of transgenic plants was significantly longer than those of WT plants (**Figure 8B**). Similarly, the fresh weight of transgenic plants was significantly greater than those of WT plants (**Figure 8C**). In response to 50 or 100 mM NaCl treatments, the total root length and fresh weight were different between the transgenic and WT seedlings, though not significant at the 0.05 level (**Figures 8B,C**).

The capacity of GmTGA17 transgenic lines to respond to high salinity stress was also assessed. The leaves of WT plants were severely wilted, whereas, the leaves of transgenic plants were slightly damaged but still remained green after treated for 1 week (**Figure 9A**). The transgenic plants displayed a significantly higher survival rate compared with WT plants under salinity conditions (**Figure 9B**). We also analyzed the chlorophyll content of all experimental plants and observed a significantly higher chlorophyll content in the transgenic plants compared with WT plants (**Figure 9C**). Together, these results demonstrated that overexpression of GmTGA17 enhanced tolerance to salt stress in transgenic Arabidopsis plants.

#### GmTGA17 Promoted ABA-Induced Stomatal Closure

A reduced water loss rate is a major factor contributing to the maintenance of moisture under drought stress. In the present study, data showed that the detached leaves from transgenic Arabidopsis lines lost water more slowly than those of WT plants (**Figure 10A**). Since the stomatal aperture has a close link with water loss, we examined the stomata of transgenic Arabidopsis and WT plants with or without ABA treatment. Under normal growth conditions, there was no obvious difference in stomatal aperture between transgenic Arabidopsis and WT plants. In the presence of 10 or 15 µM ABA, the stomatal aperture of transgenic Arabidopsis plants was significantly smaller than that of WT plants (**Figures 10C,D**). Therefore, transgenic Arabidopsis lines showed more rapid ABA-induced stomatal closure than did WT plants.

In addition, endogenous ABA content were measured in transgenic Arabidopsis lines and WT plants before and after stress treatments. As shown in **Figure 10B**, there was no significant difference in ABA content in leaves between transgenic Arabidopsis and WT plants under normal growth conditions. However, in response to drought or salt stress, the leaves of transgenic Arabidopsis lines accumulated significantly higher content of ABA than WT plants, suggesting that droughtand salt-tolerance phenotypes of transgenic Arabidopsis lines are at least partially derived from higher endogenous ABA content.

#### GmTGA17 Upregulated ABA-Responsive Genes in Transgenic Arabidopsis

To further elucidate the possible mechanism of GmTGA17 during the abiotic stress response, the transcript levels of eight known ABA-responsive genes, i.e., RD29A, RD29B, RD22, KIN1, COR15A, NCED3, COR47, and RAB18, were analyzed in transgenic Arabidopsis lines and WT plants under drought or salt

stress treatment. The qRT-PCR data showed that the expressions of RD29A, RD29B, and RAB18 were significantly up-regulated in transgenic Arabidopsis lines compared with WT plants under both non-stress conditions and stress conditions (**Figure 11**). Under non-stress conditions, no obvious differences in the expressions of RD22, KIN1, COR15A, NCED3, and COR47 were observed between transgenic Arabidopsis lines and WT plants. However, the transcript levels of RD22, KIN1, COR15A, NCED3, and COR47 were significant higher in transgenic Arabidopsis lines than those in WT plants under both drought and salt treatments (**Figure 11**).

### GmTGA17 Improves Drought and Salt Stress Tolerance in Transgenic Soybean Hairy Roots

The participation of GmTGA17 in drought and salt tolerance was further investigated by performing similar abiotic stress assays in Agrobacterium rhizogenes-mediated soybean hairy roots (Collier et al., 2005; Kereszt et al., 2007; Sun et al., 2015; Wang et al., 2015). qRT-PCR analysis showed that the expression level of GmTGA17 was significantly higher in hairy roots overexpressing GmTGA17, and significantly lower in RNAi hairy roots suppressing GmTGA17 compared with the empty vector control (**Figure 12C**).

Subsequently, we evaluated the drought and salt tolerance in plants with transgenic hairy root grown in soil watered with 20% (m/v) PEG or 150 mM NaCl solutions for 10 days. To avoid damaging the transgenic hairy roots, PEG6000 was also used for simulating drought stress instead of withholding water in this assay. All plants grew well and displayed a similar phenotype under non-stress conditions (**Figure 12A**). Under high-salinity, the leaves of all experimental plants gradually yellowed and shriveled, while under PEG treatments, the leaves of all experimental plants wilted and bleached from bottom to top. Moreover, the plants carrying the GmTGA17 RNAi hairy roots were the most sensitive to high salinity and PEG stress, while less susceptible were the empty vector control hairy roots, and least susceptible were the GmTGA17 overexpression hairy roots (**Figures 12A,B**). Under high-salinity stress, we separately observed a significantly higher and lower chlorophyll content in leaves of plants with overexpression hairy roots and RNAi hairy roots compared with that of the empty vector control (**Figure 12F**). Under both stress conditions, proline content

was higher in soybean plants with overexpression hairy roots but lower in RNAi lines relative to the empty vector control. Similarly, MDA content in plants with overexpression hairy roots were lower but higher in RNAi lines relative to the empty vector control (**Figures 12G,H**). Additionally, compared with the empty vector control, the total root length and root surface of the overexpression hairy roots were longer and greater, but the total root length of the RNAi hairy roots were shorter with decreased surface area under high-salinity and PEG stress conditions (**Figures 12D,E**). These results indicated that GmTGA17 positively regulated tolerance to drought and salt stress in transgenic soybean hairy roots.

#### DISCUSSION

TGA transcription factors have been reported to function in various biological processes in plants. However, there is no comprehensive analysis of TGA transcription factors in soybean, and especially concerning their participation in abiotic stress tolerance. In the present study, a total of 27 TGA members were identified in the soybean genome. The GmTGAs could be clustered together with TGAs from Arabidopsis and rice in the same clade (**Figure 1**), suggesting that the evolution of TGA genes are conserved between monocots and dicots. However, the number of TGAs in soybean is expanded compared to that in Arabidopsis, rice, maize, and papaya (Jakoby et al., 2002; Nijhawan et al., 2008; Espín et al., 2012; Wei et al., 2012). Gene structure analysis showed that putative GmTGA genes within the same clade share a similar intron-exon organization (**Supplementary Figure S1**), and similar patterns were observed in Arabidopsis and maize (Murmu et al., 2010; Wei et al., 2012). This indicates that the diverse status of intron and exon splicing might thus be meaningful for the evolution of TGA genes. All predicted GmTGA proteins possessed QI and QII domains, which is reported to function as a transcriptional activation domain (Schindler et al., 1992; Schwechheimer et al., 1998; Chuang et al., 1999). Additionally, we found that GmTGA17 showed transcriptional activation activity (**Figures 5A,C**). Taken together, we speculate that GmTGAs may act as transcription

activator. TGA proteins have a conserved bZIP-D box in addition to the common domains of bZIP proteins (Jakoby et al., 2002; Farinati et al., 2010). Although the bZIP-D box is a characteristic domain for TGA family, there are currently no articles of which we are aware reporting the association of this feature with any particular biological function.

It is well known that cis-elements in the promoter region of a given gene are closely related to the biological functions of gene product (Cheng et al., 2013). In the present study, the cis-acting elements analysis revealed that the putative promoter regions of all GmTGA genes possess ERE, W-box, P-box or WUN-motif pathogen-related cis-elements (Raventós et al., 1995; Choi et al., 2004), suggesting involvement of GmTGA genes in defense against pathogens. This observation was in accordance with the fact that TGA genes involved in response to pathogen attack (Jakoby et al., 2002; Wei et al., 2012). Moreover, a number of stress-related cis-elements have been identified in the putative promoter regions of GmTGA genes (**Supplementary Table S2**), with evidence demonstrating participation of these cis-elements in responses to drought, salt, ABA, and low temperature (Baker et al., 1994; Yamaguchi-Shinozaki and Shinozaki, 1994; Busk and Pagès, 1998; Gómez-Cadenas et al., 2001; Maestrini et al., 2009). The expression profiles of these genes in different plant tissues showed that 26 deduced GmTGA genes were expressed in roots, root hairs, stems, leaves, nodules, seeds, and flowers, with the exception of GmTGA01, which had no expression in leaves (**Figure 2**). It is thus evident that GmTGA genes likely play a broad role in soybean development.

We hypothesized that the existence of abundant stressrelated cis-elements suggests that GmTGA genes may participate in plant responses to a range of abiotic stresses. Our qRT-PCR data confirmed this hypothesis, with results showing that GmTGA genes are involved in drought and salt stress responses, though the response mechanisms are diverse. For example, GmTGA15, GmTGA17 and GmTGA25 were up-regulated, while GmTGA05 and GmTGA07 were down-regulated by drought and salt treatments. GmTGA10 and GmTGA14 were up-regulated by drought treatment, but down-regulated by salt treatment. Similar stress response mechanisms were observed in bZIP family genes of rice and Brachypodium distachyon (Nijhawan et al., 2008; Liu and Chu, 2015).

Using transgenic Arabidopsis and soybean hairy root assays, we further characterized the roles of GmTGA17 in abiotic stress tolerance. Our data showed that GmTGA17

could confer tolerance to drought and salinity in transgenic plants (**Figures 7**–**9**, **12**). Plant tolerance to abiotic stresses are modulated by complicated plant hormone signal transduction pathways and metabolism, especially the main stress hormone ABA (Huang et al., 2008). Drought and salt stress can induce the accumulation of ABA, and this in turn regulates the expression

of stress-responsive genes to improve the plant adaption to stress (Xiong et al., 2002). In this study, endogenous ABA content was significantly higher in transgenic Arabidopsis seedlings than WT plants under drought and salt stress treatments (**Figure 10B**). Coincidentally, the transcript level of a key ABA-biosynthesis gene, AtNCED3, was significantly up-regulated in drought- and

salt-treated transgenic Arabidopsis lines compared with WT plants (**Figure 11**). Based on our stomatal aperture assay data, we speculated that transgenic Arabidopsis plants exhibited reduced leaves water loss due to induced ABA accumulation. Additionally, qRT-PCR data revealed that GmTGA17 also had an induced expression response to ABA treatment (**Supplementary Figure S4**), and upregulated several ABA-responsive genes in transgenic Arabidopsis lines under drought and salt stress treatments (**Figure 11**). The above results suggest that GmTGA17 may participate in plant response to drought and salinity in an ABA-dependent pathway.

Previously, AtTGA1-AtTGA7 were grouped within three clades associated with plant defense, based on sequence homology (Xiang et al., 1997; Kesarwani et al., 2007). However, the other three AtTGA members were not included in this classification. Subsequently, TGAs from Arabidopsis and papaya were divided into three main clades (I, II and III) according to the similarity of their protein sequences (Murmu et al., 2010; Espín et al., 2012). Clade I contains AtTGA1, AtTGA3, AtTGA4, AtTGA7, CpTGA1, and CpTGA3. AtTGA2, AtTGA5, AtTGA6, AtPAN, CpTGA2, and CpTGA4 belong to clade II. AtTGA9, AtTGA10 clustered together with CpTGA5 and CpTGA6 in the clade III.

Our phylogenetic analysis revealed that the 27 predicted GmTGAs could also be divided into clades I, II, and III, mentioned above (**Figure 1**). In clade I, AtTGA3 and AtTGA4 participate in plant pathogen response and root development, as well as playing a vital role in the regulation of response various abiotic stresses (Kesarwani et al., 2007; Alvarez et al., 2014; Du et al., 2014; Zhong et al., 2015; Canales et al., 2017). CpTGA3 was shown to be responsive to Salicylic Acid, suggesting its potential involvement in the defense response (Espín et al., 2012). qRT-PCR data generated in this study showed that GmTGA07, GmTGA10, and GmTGA15 are involved in responses to drought or salt stress.

In clade II, AtPAN is a key regulator in the control of floral patterning (Running and Meyerowitz, 1996; Chuang et al., 1999). AtTGA2, AtTGA5 and AtTGA6 play important roles in disease resistance and development (Zhang et al., 2003; Mueller et al., 2008; Stotz et al., 2013). Our results revealed that the expression of GmTGA05, GmTGA13, GmTGA22, GmTGA24, GmTGA26, and GmTGA27 were up- or down-regulated by drought and salt stress. Interestingly, GmTGA13, GmTGA22, and GmTGA24 were expressed most strongly in flowers (**Figure 2**), this indicates that they might be also involved in floral development processes in soybean.

Of the TGAs assigned to clade III, AtTGA9 and AtTGA10 were shown to participate in anther development as well as ROS-mediated responses to pathogens (Murmu et al., 2010; Noshi et al., 2016). CpTGA5 was strongly expressed, but only in petals, suggesting an associated role in floral development (Espín et al., 2012). OsbZIP41 was found to be up-regulated in seedlings under blue light treatment while OsbZIP79 and OsbZIP83 were down-regulated under abiotic stress conditions (Nijhawan et al., 2008). Miyamoto et al. (2015) found that OsbZIP79 suppressed production of the diterpenoid phytoalexin, an antimicrobial metabolite. Among the GmTGA genes, the expression of GmTGA14, GmTGA16, GmTGA17, and GmTGA20 were activated or repressed under drought and salt stress. Moreover, GmTGA17 is a positive regulator of plant tolerance to drought and salt stress. The above findings suggested the functional diversification of a certain TGA gene or the TGA members in the same clade. However, further studies are needed to determine the specific functions of the GmTGA family genes by additional experiments.

# CONCLUSION

Twenty-seven soybean TGA genes were identified in the soybean genome. Expression analysis showed that soybean TGA family genes may participate in responses to drought and salt stress. GmTGA17 conferred tolerance to drought and salt stress in both transgenic Arabidopsis plants and soybean hairy roots. However, RNAi hairy roots silenced for GmTGA17 exhibited an increased sensitivity to drought and salt stress. Taken together, soybean TGA genes may act as important components of abiotic stress tolerance in plants.

# AUTHOR CONTRIBUTIONS

Z-SX coordinated the project, conceived and designed the experiments, and edited the manuscript. BL performed the experiments and wrote the first draft. YL and J-DF conducted the bioinformatic work and performed the experiments. W-JZ, J-HL, L-GJ, Y-BZ, MC and D-HM contributed with valuable discussions. X-YC revised and edited the manuscript. Y-ZM coordinated the team. All authors read and approved the final manuscript.

# FUNDING

This research was financially supported the National Transgenic Key Project of the Chinese Ministry of Agriculture (2018ZX0800909B and 2016ZX08002-002), the National Natural Science Foundation of China (31871624 and 31871611), and the Program of Introducing Talents of Innovative Discipline to Universities (Project 111) from the State Administration of Foreign Experts Affairs (#B18042) "Crop Breeding for Disease Resistance and Genetic Improvement."

#### ACKNOWLEDGMENTS

We are grateful to Dr. Lijuan Qiu of the Institute of Crop Science, Chinese Academy of Agricultural Sciences for kindly providing soybean seeds.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00549/ full#supplementary-material

FIGURE S1 | Phylogenetic analysis and intron/exon structures of TGA genes in soybean. The phylogenetic tree (left panel) was created using MEGA6.0 software. Exon/intron structures of TGA genes are shown in the right panel. Gene models were drawn to scale as indicated on the bottom.

FIGURE S2 | Structure and alignment of GmTGA proteins. (A) Primary structure of TGA (Jakoby et al., 2002). (B) Alignment of GmTGA amino acid sequences. Black arrows indicate Glutamine, Alanine, and Serine residues. Asterisks indicate the leucine zipper domain.

FIGURE S3 | PCR detection and the expression levels of GmTGA17 in transgenic Arabidopsis lines overexpressing GmTGA17. (A) Genomic DNA PCR detection of GmTGA17 in transgenic Arabidopsis lines (OE-1, OE-4, and OE-6). "+" represents

#### REFERENCES


positive control; "–" represents negative control (WT plant). The size of target band is 671 bp. (B) Relative expression levels of GmTGA17 in transgenic Arabidopsis lines. Actin2 was used as a control.

FIGURE S4 | Expression profile of GmTGA17 under ABA treatment. Expression pattern of GmTGA17 in soybean seedlings under ABA treatment. The expression levels were normalized to that of CYP2. All values are means and SD obtained from four biological replicates. The asterisks indicate a statistical significance ( ∗∗P < 0.01) compared with the corresponding controls.

TABLE S1 | Primers and sequences used in this study.

TABLE S2 | Cis-elements in promoters of GmTGA genes.

tolerance to salt stress in transgenic apple (Malus domestica). J. Agric. Sci. 152, 634–641. doi: 10.1017/S0021859613000130


fpls-10-00549 May 15, 2019 Time: 16:34 # 18

of soybean to study root biology. Nat. Protoc. 2, 948–952. doi: 10.1038/nprot. 2007.141


to mediate elicitor response of a maize PRms gene. Plant J. 7, 147–155. doi: 10.1046/j.1365-313X.1995.07010147.x


fpls-10-00549 May 15, 2019 Time: 16:34 # 19


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

Copyright © 2019 Li, Liu, Cui, Fu, Zhou, Zheng, Lan, Jin, Chen, Ma, Xu and Min. 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.

# Luciferase-Based Screen for Post-translational Control Factors in the Regulation of the Pseudo-Response Regulator PRR7

Yeon Jeong Kim and David E. Somers\*

Department of Molecular Genetics, The Ohio State University, Columbus, OH, United States

Control of protein turnover is a key post-translational control point in the oscillatory network of the circadian clock. Some elements, such as TOC1 and PRR5 are engaged by a well-described F-box protein, ZEITLUPE, dedicated to their proteolytic turnover to shape their expression profile to a specific time of night. For most other clock components the regulation of their protein abundance is unknown, though turnover is often rapid and often lags the cycling of the respective mRNA. Here we report the design and results of an unbiased genetic screen in Arabidopsis to uncover proteolytic regulatory factors of PSEUDO-RESPONSE REGULATOR 7 (PRR7), a transcriptional repressor that peaks in the late afternoon. We performed EMS mutagenesis on a transgenic line expressing a PRR7::PRR7-luciferase (PRR7-LUC) translational fusion that accurately recapitulates the diurnal and circadian oscillations of the endogenous PRR7 protein. Using continuous luciferase imaging under constant light, we recovered mutants that alter the PRR7-LUC waveform and some that change period. We have identified novel alleles of ELF3 and ELF4, core components of the ELF3-ELF4-LUX Evening Complex (EC), that dampen the oscillation of PRR7-LUC. We report the characterization of two new hypomorphic alleles of ELF3 that help to understand the relationship between molecular potency and phenotype.

#### Edited by:

Dae-Jin Yun, Konkuk University, South Korea

#### Reviewed by:

Sumire Fujiwara, National Institute of Advanced Industrial Science and Technology (AIST), Japan Matt Jones, University of Essex, United Kingdom

> \*Correspondence: David E. Somers somers.24@osu.edu

#### Specialty section:

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

Received: 10 January 2019 Accepted: 02 May 2019 Published: 22 May 2019

#### Citation:

Kim YJ and Somers DE (2019) Luciferase-Based Screen for Post-translational Control Factors in the Regulation of the Pseudo-Response Regulator PRR7. Front. Plant Sci. 10:667. doi: 10.3389/fpls.2019.00667 Keywords: circadian clock, pseudo-response regulator, PRR7, ELF3, ELF4, EMS mutagenesis, post-translational regulation

#### INTRODUCTION

The circadian clock system helps to coordinate daily oscillations in gene expression, metabolism and physiology to help optimize growth and reproduction under daily light/dark cycles. It is primarily comprised of interlocked autoregulatory feedback loops of gene transcription and translation, but relies strongly on numerous post-transcriptional and post-translational processes (Seo and Mas, 2014; Mateos et al., 2018). In Arabidopsis, one of the core loops involves the evening-expressed gene TIMING OF CAB EXPRESION 1 (TOC1) and the morning expressed genes CIRCADIAN CLOCK ASSOCIATED 1/LATE ELONGATED HYPOCOTYL (CCA1/LHY) which act together in a mutually repressive negative feedback.

Among the additional transcriptional repressors/co-repressors and activators/co-activators that comprise a fully functional clock is a five-member family of pseudo-response regulators (PRRs).

PRR9, PRR7, PRR5, PRR3, and TOC1 are expressed in sequential and overlapping order over the course of diel and circadian cycles (Matsushika et al., 2000; Fujiwara et al., 2008). Numerous studies have highlighted the dual role that most of these PRRs play as transcriptional repressors (Farre and Liu, 2013). At one level they act to repress transcription of certain core clock genes, helping to maintain the correct period and robustness of the central oscillator. In particular, the waveform of CCA1/LHY expression is established by the sequential and ordered expression, from morning to evening, of PRR9, PRR7, and PRR5, which results in the direct repression of these morning genes at all times except for early morning and late night (Nakamichi et al., 2010). At the same time, the precise phase-specific expression of each of the PRRs contributes to an orchestration of concomitant specific phasing of output gene expression (Nakamichi et al., 2012; Farre and Liu, 2013; Liu et al., 2013, 2016).

PSEUDO-RESPONSE REGULATOR7 (PRR7) is a key component in the control of the plant circadian clock. It is one of five closely related transcriptional repressors in the Arabidopsis clock that controls not only the period of the oscillator, but also acts on core genes involved in abiotic stress (Liu et al., 2013; Kolmos et al., 2014). PRR7 occupies a unique, synergistic position in the plant circadian system: the prr7 mutant (ca. + 1 h) enhances the short period of the prr5 mutant (−1.5 h) to a much shorter period (prr5 prr7 = −5.0 h), while it also strongly enhances the long period of the prr9 mutant (+1.5 h) to be even longer (prr9 prr7 = +8 h) (Farre et al., 2005; Mizuno and Nakamichi, 2005; Nakamichi et al., 2005; Salome and McClung, 2005). These findings show that PRR7 operates centrally and together with other PRR proteins to control period, but how this occurs is unknown. PRR7 and other PRRs also act with the co-repressor TOPLESS (TPL) and histone deacetylases to form repressive complexes (Wang et al., 2013).

PSEUDO-RESPONSE REGULATOR7 also plays a central role in the abiotic stress response. PRR7 is involved in the regulation of ABA-related processes, including control of genes affecting salt and freezing tolerance. A high percentage (28%) of PRR7 targets are also ABA-regulated, with more than one third of PRR7 target genes possessing ABA-responsive elements (Liu et al., 2013). STO (AT1G06040; SALT TOLERANCE), STH (AT2G31380 salt tolerance homolog), and members of the CBF/DREB family (AT4G25470, AT4G25490, AT4G25480) are examples of genes targeted by PRR7 that are involved in salt, drought, and cold stress tolerance (Nakamichi et al., 2012; Liu et al., 2013).

Given this central role for PRR7, and since the posttranslational regulation of only two PRR family members has been well characterized (TOC1 and PRR5), we undertook a forward genetic screen to identify PRR7 protein turnover factors. A previous luciferase-based screen successfully identified ZEITLUPE (ZTL) as an F-box protein responsible for the E3 ligase-based proteolysis of TOC1 and PRR5 (Mas et al., 2003; Kiba et al., 2007; Fujiwara et al., 2008). The rapid protein turnover of the clock-related PRRs (Fujiwara et al., 2008; Nakamichi et al., 2010) suggests dedicated proteolytic factors may be associated with each to ensure their proper phasing during the circadian cycle.

Our approach employed a PRR7-luciferase translational fusion (PRR7::PRR7-LUC) and EMS mutagenesis to identify plants with aberrantly high levels of luminescence at times when PRR7 levels are normally low. We recovered multiple classes of factors that alter the luminescence profile, and characterized here are three new alleles of EARLY FLOWERING 3 (ELF3) identified from the screen.

#### MATERIALS AND METHODS

#### Plasmid Construction and Transgenic Plant

To generate the PRR7::PRR7-luciferase (PRR7::PRR7-LUC) transgenic line, PRR7 coding sequence from ATG to STOP codon was subcloned into Nco I site in pPZP-BAR DONR plasmid harboring luciferase fused to 1208 bp of the PRR7 promoter, which was kindly provided by the McClung laboratory (Dartmouth College, Hanover, NH, United States). A genomic fragment containing 2223 bp upstream of the 5<sup>0</sup> end of PRR7 was then cloned into EcoRV site upstream of the PRR7 gene to replace the 1208 bp-promoter resulting in PRR7::PRR7-LUC (Nakamichi et al., 2010). Arabidopsis thaliana plants (Col-0) were transformed with Agrobacterium tumefaciens strain GV3101 by a floral dip method (Clough and Bent, 1998). Basta-resistant primary transformants were self-pollinated, and a high amplitude cycling bioluminescence homozygous line was selected from the T<sup>3</sup> generation. After validating that the circadian oscillation in luciferase activity correlated with the abundance of PRR7-LUC protein, T<sup>4</sup> seeds were harvested and used for ethyl methanesulfonate (EMS) mutagenesis. To construct TAP-tagged ELF3WT, ELF3A37T, and ELF3P666S , the DNA fragment containing the nucleotide substitution corresponding to the mutation was subcloned to pENTR/D-TOPO (Invitrogen, K240020) and verified by sequencing. The TAP tag (2x Protein A IgG binding domain His-9x myc) was placed at the N-terminus of ELF3 by LR recombination with pN-TAPa (Rubio et al., 2005). HA-tagged-ELF4, LUX, GI, and PIF4 were obtained by cloning pENTR/D-TOPO clones into pCsVMV-HA-C-1300 vector. pENTR4-phyB was kindly supplied by the Quail laboratory (UC Berkeley, CA) and cloned into pCsVMV-GFP-N-1300 vector. GFP-TOC1 construction was described previously (Wang et al., 2013). Primers for plasmid construction are listed in **Supplementary Table S1**.

#### EMS Mutagenesis

Approximately 27,500 PRR7::PRR7-LUC seeds were EMS treated. Briefly, the seeds were soaked overnight in 0.1 % potassium chloride and transferred to 100 mM phosphate buffer containing 0.25 % EMS. After shaking incubation at room temperature for 15 h, the seeds were rinsed three times with 100 mM sodium thiosulfate and washed several times in water. The mutagenized seeds were sown on 10 soil flats and stratified at 4◦C for 4 days, and grown until seed set under 16L:8D at 22◦C. Flats were harvested as 256 pools of between 50–150 plants/pool.

### Bioluminescence Assays

fpls-10-00667 August 14, 2019 Time: 11:58 # 3

Approximately one thousand and two hundred seeds from each pool were plated on Murashige and Skoog (MS) media containing 3 % sucrose and grown in 12:12 LD white-light cycles (50 µmol m−<sup>2</sup> s −1 ) for 5 days. Seedlings were sprayed with 1 mM luciferin solution containing 0.01 % Triton X-100 and transferred to imaging chamber. Images were obtained with an Andor iKon-M 934 CCD camera (Andor Technology, Belfast, United Kingdom) for 5 min every 2 h under continuous LED red and blue light (30 µmol m−<sup>2</sup> s −1 ) at 22◦C. Luminescence signals were quantified by Image-J software.

#### Phenotypic Analyses

For flowering time measurement, dried seeds were sterilized and sown on soil flats followed by stratification at 4◦C in dark for 4 days. Plants were grown at 22◦C under a 16L:8D photoperiod (white light, 110 µmol m−<sup>2</sup> s −1 ) and watered as necessary until the plants were flowered. The number of rosette leaves were determined from the plants when the bolt reached 1 cm. For hypocotyl length analysis, surface sterilized seed were plated on MS media without sucrose and stratified at 4◦C in dark for 4 days. Germination and growth were carried out at 22◦C in continuous red LED light with different light intensities ranging from 0.52 to 20.32 µmol m−<sup>2</sup> s −1 . Hypocotyl length was measured from images of the seedlings 4 days after illumination using Image-J software.

### RNA Extraction and Quantification Real-Time PCR

Total RNA was extracted using TrizolTM reagent according to manufacturer's protocol (Thermo Fisher Scientific, Cat #155960-028) and treated with RNase-free DNase I (Ambion, Cat #AM2224) for 30 min at 37◦C. First-strand cDNA was synthesized from 3 µg of the total RNA using Oligo(dT)12−<sup>18</sup> primer and SuperScriptTM III reverse transcriptase (Thermo Fisher Scientific, Cat #18080093) followed by RNase H treatment for 20 min at 37◦C. For quantitative real-time RT-PCR, specific primers and equal volume of the template cDNA were combined with 7.5 µL iQTM SYBR-Green Super Mix (Bio-Rad, Cat #1708882), and subjected to following thermal cycling conditions: 94◦C for 2 min; followed by 44 cycles of 94◦C for 15 s and 55◦C for 34 s. The quantities of input cDNA were normalized to At5g15400, and transcript levels of target genes were analyzed by CFX ManagerTM Software (Bio-Rad). Primers used for qRT-PCR are listed in **Supplementary Table S1**.

# Protein Extraction and Immunoblotting

Protein extraction and immunoblotting were performed as described previously (Kim et al., 2003). Briefly, ground tissues were homogenized with extraction buffer (50 mM Tris–Cl, pH 7.5, 150 mM NaCl, 0.5% Nonidet P-40, 1 mM EDTA, 3 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 5 µg/ml leupeptin, 1 µg/ml aprotinin, 1 µg/ml pepstatin, 5 µg/ml antipain, 5 µg/ml chymostatin, 50 µM MG132, 50 µM MG115, and 50 µM ALLN) and centrifuged at 16,000 g at 4 ◦C for 10 min. To detect luciferase protein, supernatant proteins were concentrated by TCA precipitation and resultant pellets were resuspended in Urea/SDS loading buffer (40 mM Tris–Cl, pH 6.8, 8 M Urea, 5% SDS, 1 mM EDTA, 2% 2 mercaptoethanol). The total proteins were separated on a 8% SDS-PAGE gel (acrylamide:bisacrylamide, 37.5:1) and probed with 1:1,000 anti-luciferase (Sigma, L0159) and 1:15,000 anti-ADK antibody (gift from Dr. David Bisaro), as a loading control. For immunoprecipitated proteins, 1:2,000 anti-myc (Sigma, M4439), 1:1,000 anti-HA (Sigma, 3F10), and 1:5,000 anti-GFP (Abcam, ab6556) were used.

#### Coimmunoprecipitation

Agrobacterium tumefaciens strains GV3101 or AGL-1 harboring TAP-ELF3WT, TAP-ELF3A37T, or TAP-ELF3P666S, and HA-ELF3WT, HA-ELF3A37T, HA-ELF3P666S, HA-ELF4, HA-LUX, HA-GI, GFP-TOC1, HA-PIF4, or phyB-GFP were co-infiltrated into Nicotiana benthamiana leaves. Total proteins were extracted from the leaves collected 2–3 days after the infiltration, and the cleared supernatant was incubated with 20 ul of Human IgG-Agarose (Sigma, A6284) for 1 h at 4◦C. After washing the resin with cold extraction buffer 4–5 times, 1.5 ul of HRV3C protease (Thermo Scientific, 88947) was added for1.5 h at 4◦C to release the resin-bound immune complexes, and separated by SDS-PAGE.

# Statistical Analysis

For comparison between the plant groups, one-way Analysis of variance (ANOVA) followed by Tukey's HSD test was applied using R 3.5.0. Statistically significant differences (p < 0.05) are represented by small letters within the figures.

# RESULTS

#### Identification of Factors Altering PRR7::PRR7-LUC Expression

We identified a PRR7::PRR7-luciferase (PRR7-LUC) translational fusion line to perform an EMS-based mutant screen for PRR7 turnover factors. We chose a transgenic line in which the circadian oscillation of the PRR7-LUC protein recapitulates endogenous PRR7 phasing (**Figures 1A,B**), and which also demonstrates robust circadian oscillations in luminescence (**Figure 1C**). We reasoned that mutants exhibiting luminescence oscillations with reduced amplitude and/or higher trough levels would be candidates for a loss-of-function in factors involved in PRR7-LUC turnover. 41,433 EMS-mutagenized seedlings from 44 pools (50–150 plants/pool) were screened and 31 candidates were identified. Some mutants show essentially WT period but with significantly higher troughs (**Figure 2**), as expected if PRR7 proteolysis is reduced. Before further mapping, we first tested whether any mutant loci corresponded to known clock loci. In particular, loss-of-function mutations in evening complex (EC) genes (ELF3, ELF4, and LUX) cause circadian arrhythmia and upregulation of PRR7 transcription (Kolmos et al., 2009; Dixon et al., 2011; Kolmos et al., 2011; Mizuno et al., 2014; Choudhary et al., 2015; Huang and Nusinow, 2016).

We extracted genomic DNA from the 20 surviving lines with increased PRR7-LUC luminescence and examined the genomic sequences of EC genes. Fifteen of twenty lines, originating

from seven different pools, had single point mutations in ELF3 or ELF4 coding regions, causing amino acid substitutions or predicted premature translation termination. The remaining five lines have no mutations in the EC genes, and segregate as single gene mutations in backcrossed F2 populations. Two of these lines showed growth and development similarities to the GIGANTEA (GI) mutant, gi-2 (long period in constant light and late flowering). We sequenced GI and confirmed that the mutations are not at that locus.

Ten of the fifteen EC mutants showed single amino acid substitutions in ELF3 as either proline to serine (P666S; elf3-13), alanine to threonine (A37T; elf3-14), or premature termination (Q550X; elf3-15) (**Figures 2**, **3**). These mutations occurred at the N-terminal and C-terminal regions of ELF3, respectively, which are highly conserved among plant ELF3 orthologs (**Supplementary Figure S1**). The elf4 mutant (**Figure 2**, pool #32) is a presumed null (Trp26 STOP; TGG – >TAG).

#### Effects of Novel elf3 Mutations on Growth and Development

Each of these mutations co-segregated with high PRR7- LUC luminescence in backcrossed F2 populations. Higher levels of PRR7-LUC were also detected in the mutants at ZT1, confirming that luminescence levels arose from more PRR7-LUC protein accumulation (**Supplementary Figure S2**). To identify the effects of these novel mutations on growth and development, we backcrossed four independently isolated elf3 mutant lines (elf3-13#1 and elf3-13#2; elf3- 14#1, and elf3-14#2) to PRR7::PRR7-LUC/Col-0 and selected one individual segregant (BC1F3) from the respective F2 populations to characterize. We included Q550X premature translation termination mutants, (elf3-15#1 and elf3-15#2 BC1F3 segregants) as a controls along with the null, elf3-8 (Hicks et al., 2001).

Severe mutants of ELF3 result in arrhythmicity or near arrhythmicity in circadian oscillations of gene expression and bioluminescent reporters (Hicks et al., 1996; McWatters et al., 2000; Covington et al., 2001; Kim et al., 2005), while reduced function alleles shorten period (Kolmos et al., 2011). Both elf3-13 and elf3-14 shorten period significantly (elf3-13: 19.2–19-6 h; elf3-14: 21.6–22.0 h; and WT: 23.8 h) while elf3-15 is arrhythmic (**Figure 4A**). Using relative amplitude error (R.A.E.) as a measure of oscillation robustness (low values indicate strong rhythms, >0.6 poorly rhythmic or arrhythmic) elf3-13, with the shortest period, is more strongly compromised in function than elf3-14 (**Figure 4B**).

EARLY FLOWERING 3 (ELF3) loss-of-function mutations cause early flowering under long and short days (Zagotta et al., 1996; Hicks et al., 2001). The elf3-15 mutation results in significantly early flowering under long days (16:8) compared to wild-type (PRR7::PRR7-LUC/Col-0), but later than elf3-8 (**Figure 5**). The very slightly later flowering of elf3-15, relative to elf3-8 may be due to mild overexpression of PRR7 from the presence of the PRR7::PRR7-LUC transgene. elf3-13 also shortened flowering time, but was the least severe allele, compared to elf3-14, and the presumed null, elf3-15.

Light-dependent hypocotyl elongation is a sensitive measure of the extent to which the phototransduction pathway is intact (Gommers and Monte, 2018). Strong elf3 mutants show diminished light responsiveness, resulting in longer hypocotyls compared to wild-type (Zagotta et al., 1996; Reed et al., 2000). All three elf3 alleles cause significantly longer hypocotyls under a range of red light intensities (**Figure 6**). elf3-15 is the most severe, especially at very low light intensities, with hypocotyl lengths similar to the elf3-8 null mutant. However, at higher light intensities elf3-14 is similar to elf3-15. elf3-13 retains the most functionality at all light intensities, relative to the other two alleles, but shows strongly diminished function.

#### elf3-13 and elf3-14 Mutants Retain Some Repressive Functions

To further refine our understanding of the effects of these mutations on circadian and stem elongation processes, we examined in the elf3-13, -14, and -15 mutants the expression patterns and levels of select components of both processes known to be under ELF3 control (**Figure 7**). ELF3 is expressed at night under LD (12 h light:12 h dark) cycles and during subjective night under free-running constant light conditions (Nusinow et al., 2011; Flis et al., 2015). It represses PRR9 and PRR7 expression during those times, restricting their expression to early in the day (Dixon et al., 2011; Flis et al., 2015). The elf3 null (elf3-8) shows markedly higher expression levels of both genes during the dark, particularly for PRR7 (**Figures 7B,C**). elf3-15 is similar to elf3-8, suggesting it effectively acts as a null, consistent with its circadian arrhythmicity (**Figure 4**). elf3-13 and elf3-14 show a nearly normal expression pattern for PRR9 under LD, but


FIGURE 3 | Protein coding sequence of ELF3. Deduced amino acid sequences and nucleotide changes in known elf3 alleles are shown above and below the nucleotide sequence. Sites of novel mutants identified in this study are indicated in red. Sources for the known indicated elf3 alleles are: elf3-1, elf3-3, elf3-4, elf3-5, elf3-6, elf3-7, elf3-8, and elf3-9 (Hicks et al., 2001); elf3-12 (Kolmos et al., 2011); elf3-101, elf3-102, and elf3-103 (Yoshida et al., 2009); and elf3-201, elf3-202, elf3-203, and elf3-204 (Kinoshita et al., 2011).

**251**

a marked de-repression of PRR7 expression is seen for both mutants during the late night, though not as strongly as for elf3-15 (**Figures 7B,C**).

The sequential temporal expression of the PRR proteins contributes strongly to restricting CCA1 expression to the very late night and early morning (Nakamichi et al., 2010). ELF3 upregulates CCA1 indirectly, through repression of the repressors of CCA1, PRR9, and PRR7 (Dixon et al., 2011; Kolmos et al., 2011). As a result, elf3 null mutants (elf3-8) show lower peak CCA1 levels, and elf3-15 is very similar (**Figure 7A**; Dixon et al., 2011). elf3-13 and elf3-14 appear intermediate in effect, consistent with their effects on PRR9 and PRR7 expression (**Figures 7A,B**).

GIGANTEA (GI) is a key component in the control of flowering time and numerous other processes (Fowler et al., 1999; Mishra and Panigrahi, 2015), and a co-chaperone in the maturation of the F-bpx protein, ZEITLUPE (ZTL) (Cha et al., 2017). GI expression is strongly upregulated in elf3 null mutants (Fowler et al., 1999; Kolmos et al., 2011; Dixon et al., 2011). We confirm those findings (**Figure 7D**) and show that elf3-15 is very similar to elf3-8 in de-repressing GI at night. Similar to

our findings for PRR7, both elf3-13 and elf3-14 can repress GI expression at night, but elf3-13 is less effective (**Figure 7D**).

Phytochrome Interacting Factors (PIFs) play multiple, interacting and integrative roles in plant development (Leivar and Monte, 2014). PIF4 and PIF5 are clock-regulated at the transcript level, and light-regulated at the protein level, acting as integrators of these signals in the control of hypocotyl elongation (Nozue et al., 2007; Lorrain et al., 2008; Niwa et al., 2009). In the context of the EC complex, ELF3 represses PIF4 expression at night, which rises strikingly in elf3 null mutants, including elf3-15 (Nusinow et al., 2011; Lu et al., 2012; **Figure 7E**). PIF4 expression in elf3-13 and elf3-14 are closer to WT, but the phase of expression is slightly advanced, consistent with their short period, and expression rises markedly during the late night, especially in elf3-14 (**Figure 7E**). These findings are consistent with the phenotypes of these two hypomorphic alleles, with elf3-14 hypocotyls slightly longer than elf3-13, correlating with the greater derepression of PIF4 in elf3-14 (**Figure 7E**).

Taken together, these results indicate that the P666 and A37 residues are crucial for the normal developmental and circadian functions of ELF3 protein. Both mutations strongly diminish ELF3 activity and regulation of circadian clock output pathways and hypocotyl elongation, but significant functionality is retained, as evidenced by circadian oscillations for 4 days or more in elf3-13 and elf3-14 (**Figure 4**). elf3-13 exhibits a shorter circadian period and greater degree of derepression of PRR7 and GI than elf3-14, suggesting it is the stronger of the two mutant alleles with respect to clock function.

To better understand how elf3-13 (P666S) and elf3-14 (A37T) compromise clock and hypocotyl function, we tested their interactions with known protein partners, including ELF3, ELF4, LUX, GI, TOC1, phyB, and PIF4 (Liu et al., 2001; Yu et al., 2008; Nusinow et al., 2011; Herrero et al., 2012; Nieto et al., 2015; Huang et al., 2016). We performed co-infiltration into N. benthamiana of Agrobacterium harboring appropriate pairwise combinations of TAP-tagged ELF3 (WT, elf3-13, elf3-14) with HA-tagged or GFP-tagged ELF3, LUX, TOC1, GI, phyB, and PIF4 (**Figure 8** and **Supplementary Figure S3**). Compared with WT ELF3, self-interaction for the two alleles appeared unaffected (**Figure 8A** and **Supplementary Figure S3**), though

there is a weak statistically significant reduced interaction between WT ELF3 and elf3-14 (**Supplementary Figure S3**). The two other components of the EC, ELF4, and LUX also showed no detectable changes in their interaction with elf-13 or elf-14 protein (**Figure 8B** and **Supplementary Figure S3**).

TIMING OF CAB EXPRESION 1 has been recently identified as an ELF3 interactor, though the significance is unclear (Huang et al., 2016). The GI-ELF3 interaction has been implicated in the COP1-dependent turnover of GI protein (Yu et al., 2008). Neither elf-13 nor elf3-14 altered interactions with TOC1 or GI (**Figure 8C** and **Supplementary Figure S3**).

Phytochrome Interacting Factor4 (PIF4) transcript levels are regulated by ELF3 through its participation in the EC. ELF3 can also regulate PIF4 activity through direct binding to the PIF4 bHLH domain, suppressing PIF4 transcriptional activity (Nieto et al., 2015). PIF4 interactions with ELF3 were not detectably altered by the elf-13 and elf3-14 mutations (**Figure 8D** and **Supplementary Figure S3**). phyB interacts with ELF3 as a likely point of intersection between light signaling and the circadian system (Huang and Nusinow, 2016; Huang et al., 2016). The elf3-14 mutation reduced the interaction with phyB by half, while the elf3-13 mutation had no effect (**Figure 8D** and **Supplementary Figure S3**). The A37T mutation resides within the known ELF3-phyB interaction domain [aa 1–201; (Liu et al., 2001)], suggesting we have identified a key ELF3 residue important in the phyB-ELF3 association.

#### DISCUSSION

Here we have characterized three new alleles of ELF3 recovered from a forward genetic screen for PRR7 protein turnover factors. The approach relied on changes in the bioluminescence oscillation pattern of PRR7-LUC protein. Higher levels of PRR7- LUC during normal trough times could indicate a more stable protein, suggesting a loss-of-function in PRR7 proteolytic factors. This was observed in ztl mutants where strongly flattened rhythms of SCFZTL complex targets, TOC1 and PRR5, were seen under LD (Mas et al., 2003; Kiba et al., 2007; Fujiwara et al., 2008). Subsequent studies found that other members of the ZTL family (LKP2 and FKF1) also contribute to TOC1 and PRR5 turnover, but their contribution is less substantial than ZTL (Fujiwara et al., 2008; Baudry et al., 2010).

Reliance on changes in the PRR7-LUC waveform as the primary criterion meant that false positives that altered clock activity in ways unrelated to PRR7 proteolysis could be recovered, since the transgene was driven by the PRR7 promoter. We observed that the flattened circadian oscillations of many mutants (**Figure 2**) were reminiscent of elf3, elf4, and lux null mutants

these interactions are shown in Supplementary Figure S3.

(Hicks et al., 1996; Doyle et al., 2002; Hazen et al., 2005; Onai and Ishiura, 2005; McWatters et al., 2007). Sequencing these loci for all 20 surviving candidates revealed three novel ELF3 alleles and one novel ELF4 allele (translation termination). No mutants at the LUX locus were identified.

The elf3-13 and elf3-14 mutants are only the second and third alleles reported that retain significant but compromised function for multiple ELF3 controlled processes (Kolmos et al., 2011). In these mutants, the three primary developmental phenotypes affected by ELF3, flowering time, clock function and hypocotyl expansion are clearly intermediate between wild-type function and loss-of-function. Interestingly, all three processes are similarly compromised despite the location of the two mutations at opposite ends of the protein. Previous work suggested the N-terminal region, location of the A37T substitution of elf3-14, as important in mediating interactions with GI and phyB (Liu et al., 2001; Yu et al., 2008). Both GI and phyB play roles in flowering time, circadian period control and hypocotyl expansion (Koornneef et al., 1980; Goto et al., 1991; Somers et al., 1998; Park et al., 1999; Huq et al., 2000; Sawa et al., 2007). While the

Kim and Somers Factors Regulating PRR7-LUC Waveform

elf3-14-GI interaction is similar to WT ELF3-GI, the phyBelf3-14 interaction is strongly reduced (**Figure 8C**). The phyB-ELF3 interaction domain maps to the N-terminus of ELF3, consistent with the location of the elf3-14 A37T mutation (Liu et al., 2001). The elf3 and phyB mutants are similar in displaying long hypocotyls in red light and early flowering in long days (Liu et al., 2001). We have not measured the protein levels of phyB or ELF3 in the elf3-14 background, but assuming they are normal, their reduced interaction suggests that a phyB-ELF3 association is key to normal hypocotyl development and flowering time. Our interaction assay was performed using tissue harvested under white light, and further extraction and immunoprecipitation also in the light. Previous in vitro work indicated that the phyB-ELF3 interaction is similar for both the Pr and Pfr forms of phytochrome (Liu et al., 2001). This suggests that the primary interaction is light-independent, but light conditions could still affect the recruitment of additional factors which are dependent on the phyB form. One possibility includes members of a MUT9-LIKE KINASE clade (MLKs) that associate with ELF3 in a phyBdependent way (Huang et al., 2016).

Residue P666 resides near the C-terminus, in the PIF4 interaction domain (Nieto et al., 2015), although the P666S mutation (elf3-13) shows no detectable difference in the PIF4 elf3-13 interaction (**Figure 8**). However, this interaction is not relevant to the compromised repression of PIF4 and other genes (**Figure 7**), since all known transcriptional repressive activities of ELF3 are associated with its participation in the EC (Huang and Nusinow, 2016). ELF3 is thought to function as a scaffold, linking ELF4 with LUX, the DNA-binding partner of the tripartite complex (Huang and Nusinow, 2016; Huang et al., 2016). Our findings indicate that neither P666S nor A37T alters ELF3 binding to either partner, suggesting no effect on EC formation. However, modifications of the structure of the complex, which might alter recruitment of EC-interaction factors, or chromatin residence, may result from either or both mutations. The MLK kinases that associate with the EC (Huang et al., 2016) suggests a possible role for phosphorylation in the control of EC activity, co-factor interactions or chromatin binding. Both mutations add a potential phosphorylation site (S/T) which could result in an aberrant ELF3 phosphorylation state. ELF3 chromatin IPs in these elf mutant backgrounds would test one of these possibilities.

#### REFERENCES


Also untested is the effect of elf3-13 and elf3-14 mutations on ELF3 nucleocytoplasmic distribution. ELF3 level in the nucleus is a key determinant in the effectiveness of ELF3 function in the clock (Herrero et al., 2012; Anwer et al., 2014). Nuclear localization of ELF3 is facilitated by ELF4 (Herrero et al., 2012), The ELF3-ELF4 interaction domain has been mapped to the middle of ELF3, exclusive of the A37 and P666 residues (Herrero et al., 2012), so it is not surprising that the neither mutation affects this interaction. However, nucleocytoplasmic partitioning is a multi-factor process and there are multiple mechanisms and proteins involved. Interactions with key nuclear import or nuclear exclusion partners might be affected by these elf3 mutations (Meier and Somers, 2011; Floch et al., 2014). Examination of elf3-P666S and elf3-A37T nucleocytoplasmic distribution would test this hypothesis.

While our screen identified these new elf3 alleles, three uncharacterized lines with similar PRR7-LUC profiles (e.g., **Figure 2**, pool#33) and no mutations in EC components or GI remain. Our forward genetics approach has demonstrated utility in uncovering novel reagents useful in probing the mechanics of the clock.

#### AUTHOR CONTRIBUTIONS

YJK and DS designed the research, analyzed the data, and wrote the manuscript. YJK performed the research.

#### FUNDING

This work was supported by National Institutes of Health grant R01GM093285 and by a grant from the Next-Generation BioGreen21 Program (Systems and Synthetic Agrobiotech Center, Project PJ01327305), Rural Development Administration, Republic of Korea (DS).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00667/ full#supplementary-material


through EARLY FLOWERING 3 in Arabidopsis. Curr. Biol. 21, 120–125. doi: 10.1016/j.cub.2010.12.013



of the Arabidopsis circadian clock. Plant Cell 17, 791–803. doi: 10.1105/tpc.104. 029504


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

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

# Corrigendum: Luciferase-Based Screen for Post-Translational Control Factors in the Regulation of the Pseudo-Response Regulator PRR7

#### *Yeon Jeong Kim and David E. Somers\**

translational regulation

**A Corrigendum on**

*Department of Molecular Genetics, The Ohio State University, Columbus, OH, United States*

*Edited and reviewed by: Frontiers in Plant Science, Frontiers Media SA, Switzerland*

> *\*Correspondence: David E. Somers somers.24@osu.edu*

#### *Specialty section:*

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

*Received: 29 July 2019 Accepted: 30 July 2019 Published: 22 August 2019*

#### *Citation:*

*Kim YJ and Somers DE (2019) Corrigendum: Luciferase-Based Screen for Post-Translational Control Factors in the Regulation of the Pseudo-Response Regulator PRR7 Front. Plant Sci. 10:1057. doi: 10.3389/fpls.2019.01057*

**Luciferase-Based Screen for Post-translational Control Factors in the Regulation of the Pseudo-Response Regulator PRR7**

Keywords: circadian clock, pseudo-response regulator, PRR7, ELF3, ELF4, EMS mutagenesis, post-

*By Kim YJ and Somers DE (2019). Front. Plant Sci. 10:667. doi: 10.3389/fpls.2019.00667*

There is an error in the *Funding* statement. The correct number for the "Next-Generation BioGreen21 Program" is "PJ01327305."A correction has therefore been made to the *Funding* statement:

"This work was supported by National Institutes of Health grant R01GM093285 and by a grant from the Next-Generation BioGreen21 Program (Systems and Synthetic Agrobiotech Center, Project PJ01327305), Rural Development Administration, Republic of Korea (DS)."

The authors apologize for this error and state that this does not change the scientific conclusions of the article in any way. The original article has been updated.

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

# The Physiological Functions of Universal Stress Proteins and Their Molecular Mechanism to Protect Plants From Environmental Stresses

Yong Hun Chi<sup>1</sup>† , Sung Sun Koo<sup>1</sup>† , Hun Taek Oh<sup>1</sup> , Eun Seon Lee<sup>1</sup> , Joung Hun Park<sup>1</sup> , Kieu Anh Thi Phan<sup>1</sup> , Seong Dong Wi<sup>1</sup> , Su Bin Bae<sup>1</sup> , Seol Ki Paeng<sup>1</sup> , Ho Byoung Chae<sup>1</sup> , Chang Ho Kang<sup>1</sup> , Min Gab Kim<sup>2</sup> , Woe-Yeon Kim1,3, Dae-Jin Yun<sup>4</sup> and Sang Yeol Lee<sup>1</sup> \*

#### Edited by:

Ruth Grene, Virginia Tech, United States

#### Reviewed by:

Tong Zhang, Pacific Northwest National Laboratory (DOE), United States Klára Kosová, Crop Research Institute (CRI), Czechia

\*Correspondence:

Sang Yeol Lee sylee@gnu.ac.kr †These authors have contributed equally to this work

#### Specialty section:

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

Received: 19 December 2018 Accepted: 22 May 2019 Published: 05 June 2019

#### Citation:

Chi YH, Koo SS, Oh HT, Lee ES, Park JH, Phan KAT, Wi SD, Bae SB, Paeng SK, Chae HB, Kang CH, Kim MG, Kim W-Y, Yun D-J and Lee SY (2019) The Physiological Functions of Universal Stress Proteins and Their Molecular Mechanism to Protect Plants From Environmental Stresses. Front. Plant Sci. 10:750. doi: 10.3389/fpls.2019.00750 <sup>1</sup> Division of Applied Life Science (BK21Plus), Plant Molecular Biology and Biotechnology Research Center, Gyeongsang National University, Jinju, South Korea, <sup>2</sup> College of Pharmacy and Research Institute of Pharmaceutical Science, Gyeongsang National University, Jinju, South Korea, <sup>3</sup> Institute of Agricultural and Life Science (IALS), Gyeongsang National University, Jinju, South Korea, <sup>4</sup> Department of Biomedical Science and Engineering, Konkuk University, Seoul, South Korea

Since the original discovery of a Universal Stress Protein (USP) in Escherichia coli, a number of USPs have been identified from diverse sources including archaea, bacteria, plants, and metazoans. As their name implies, these proteins participate in a broad range of cellular responses to biotic and abiotic stresses. Their physiological functions are associated with ion scavenging, hypoxia responses, cellular mobility, and regulation of cell growth and development. Consistent with their roles in resistance to multiple stresses, USPs show a wide range of structural diversity that results from the diverse range of other functional motifs fused with the USP domain. As well as providing structural diversity, these catalytic motifs are responsible for the diverse biochemical properties of USPs and enable them to act in a number of cellular signaling transducers and metabolic regulators. Despite the importance of USP function in many organisms, the molecular mechanisms by which USPs protect cells and provide stress resistance remain largely unknown. This review addresses the diverse roles of USPs in plants and how the proteins enable plants to resist against multiple stresses in ever-changing environment. Bioinformatic tools used for the collection of a set of USPs from various plant species provide more than 2,100 USPs and their functional diversity in plant physiology. Data from previous studies are used to understand how the biochemical activity of plant USPs modulates biotic and abiotic stress signaling. As USPs interact with the redox protein, thioredoxin, in Arabidopsis and reactive oxygen species (ROS) regulates the activity of USPs, the involvement of USPs in redox-mediated defense signaling is also considered. Finally, this review discusses the biotechnological application of USPs in an agricultural context by considering the development of novel stress-resistant crops through manipulating the expression of USP genes.

Keywords: abiotic/biotic defense signaling, biotechnological application, external stress, molecular mechanism of USPs, multi-functional roles, universal stress protein

# INTRODUCTION

fpls-10-00750 June 4, 2019 Time: 10:11 # 2

Plants as sessile organisms are persistently confronted with detrimental factors that are arisen from ever-changing environment. To cope with environmental stresses that are harmful to their growth and development, plants have evolved sophisticated and delicate defense mechanisms. In fact, the external stress activates diverse defense signaling that include the production of reactive oxygen species (ROS), change in redox potential or cellular level of Ca2<sup>+</sup> ion, disruption of ion homeostasis, and adjustment of membrane fluidity (Gilroy et al., 2016; Choudhury et al., 2017). After sensing the external stress via specific receptors, plants transduce the foreign signal into intracellular downstream signaling pathways including the activation of protein kinase or phosphatase, stimulation of downstream target proteins, and biosynthesis of phytohormones for the control of plant growth/development (**Figure 1**; Sheikh et al., 2016; Akimoto-Tomiyama et al., 2018). In particular, crosstalk of these complex signaling networks precisely regulates the expression of stress responsive genes and protects plants from external stresses (Sewelam et al., 2016; Choudhury et al., 2017; Nejat and Mantri, 2017). Thus, the identification of diverse stress-resistant genes/proteins from various organisms and elucidation of their biochemical and physiological functions can provide valuable information for the preparation of valuable crops with stress tolerance and high productivity.

As a representative defense protein that protects host organisms from diverse external stresses, a novel stress-inducible protein with an estimated molecular weight of 13.5 kDa was identified from the cytosolic fraction of E. coli using matrix-assisted laser desorption/ionization (MALDI) analysis (VanBogelen et al., 1990). The protein designated 'Universal Stress Protein (USP)' is significantly overexpressed under unfavorable environmental stresses, such as nutrients starvation (deficiency of carbon, nitrogen, phosphate, sulfate, and amino acids), heat/cold shock, oxidative stress, heavy metal toxicity, uncoupler of electron transport chains, exposure to polymyxin, cycloserine, ethanol and antibiotics etc. (VanBogelen et al., 1990; Kvint et al., 2003). Following the discovery of USP from E. coli, many USP proteins containing at least one USP domain consisting of 140 to 160 conserved amino acid residues with other diverse functional motifs have been found from a wide variety of organisms including bacteria, archaea, plants, and metazoans (Foret et al., 2011; Vollmer and Bark, 2018). The USP domain (Pfam accession number PF00582) forming an α/β subdomain structure is important for numbers of cellular defense signaling (Kerk et al., 2003; Kvint et al., 2003; Tkaczuk et al., 2013) and numerous stress-resistant metabolic pathways (VanBogelen et al., 1990; Nystrom and Neidhardt, 1992; Zarembinski et al., 1998; Ndimba et al., 2005; Siegele, 2005; Persson et al., 2007; Sen et al., 2019). The functions of USPs are shown to involve in protein scaffolding, holding and preventing the denaturation of molten globular macromolecules, and cellular protein transport (Vollmer and Bark, 2018). Moreover, several USPs exhibit DNA binding, repairing, and refolding activities that can support organisms to protect their nucleic acids from external stresses (Kvint et al., 2003; Drumm et al., 2009).

Consistent with their multi-functional roles, USPs possess a variety of other functional motifs and thus show a high degree of structural diversities. USP-like protein groups also include flavoproteins, which are involved in electron transport, N-type protein phosphatase, and ATP sulfhydrylases (Aravind et al., 2002). Based on their structural homology with X-ray crystal structure of MJ0577 protein isolated from Methanocaldococcus jannaschii or with the protein structure of USPA from Haemophilus influenza, USPs are largely classified into two categories (Sousa and McKay, 2001). USPs belonging to the first group contain an ATP-binding motif at their C-terminal region [G-2X-G-9X-(S/T)] and have an α/β-core structure consisting of five β-strands and four α-helical structures (Aravind et al., 2002). By contrast, polypeptides in the second group are lacking the ATP-binding residues and unable to bind or utilize ATP (Sousa and McKay, 2001). The structural and functional diversity of USPs results in many orthologous proteins being placed in USP groups, producing a large USP superfamily (Siegele, 2005).

Whereas the physiological function and structural diversity of USPs have been extensively investigated in microorganisms, only a few studies have been made in plants, although plants also contain large numbers of USPs. Therefore, in this paper, we will examine the biochemical and molecular properties, structural characteristics, and functional diversities of plant USPs, after reviewing the bacterial USP properties. Considering the redoxmediated control of its chaperone activity (Jung et al., 2015), particular attention will also be paid to the redox-dependent functional and conformational regulation. To the best of our knowledge, this is the first review paper of plant USPs that will serve much valuable information to the plant biologists for analyzing their molecular mechanisms and development of stress tolerant crops with high productivity. Therefore, in the final section, we are focusing on their biotechnological application in agricultural research fields.

#### Functional and Structural Diversity of Bacterial USPs

Following the determination of amino acid sequence and protein structure of the first USP in E. coli, large numbers of USP homologs have been identified from bacterial sources and formed large USP families (Kvint et al., 2003). The E. coli USPs contain six different proteins including USPA, USPC, USPD, USPE, USPF and USPG. In fact, USPB was identified from stress condition and named USPB, of which gene was located immediately upstream of USPA gene (Farewell et al., 1998). However, USPB was not considered as a bona fide E. coli USP and eliminated from USP groups, because the protein was shown to be an integral membrane protein with two putative transmembrane domains and the molecular structure of the protein did not satisfy the criteria of USP structures. Finally, USPB is missed from the E. coli USP groups (Farewell et al., 1998; Tkaczuk et al., 2013; Vollmer and Bark, 2018). The six E. coli USPs are classified into four subclasses based on their structural similarity and amino acid sequence homology as follows. Whereas Class I without having the ATP binding motif includes USPA, USPC and USPD, Class II

containing ATP binding motif is composed of USPF and USPG. In contrast to Class I & II, USPE in E. coli has tandem-repeated two USP domains in a polypeptide, that are designated E1 and E2 domains corresponding to the first and second USP domains, respectively. The E1 and E2 domains of E. coli USPE are grouped into Class III and Class IV.

E. coli USP belonging to each subclass takes its own specific function in particular environmental stress, as shown in **Figure 2A** (Kvint et al., 2003; Nachin et al., 2005); USPA and USPD in Class I play their roles in the resistance against oxidative stress and iron scavenging, but USPF and USPG protein in Class II also partly participate in the protection of bacterial cells from the same oxidative stress. Thus USPD mutant exhibits a high sensitivity to streptonigrin, causing an increase in intracellular iron concentration, which suggests that USPD plays a role in cellular iron scavenging. In addition to their

FIGURE 2 | Functional roles of E. coli USPs and molecular structures of diverse bacterial USPs. (A) Functional roles of the six different E. coli USPs containing USPA, USPC, USPD, USPE, USPF, and USPG. Thick and thin arrows indicate the major and minor roles of specific USPs, respectively. T-shape arrows represent suppression of the physiological responses. USPs linked by brackets share the common physiological roles. This Figure 2A is modified from the reference of Nachin et al. (2005). (B) Molecular structures of the diverse bacterial USPs containing only a USP domain or USP domains fused with other catalytic motifs that are obtained from Pfam database (https://pfam.xfam.org/family/Usp). The numbers of USPs having the specific type of molecular structure are indicated at the right side of the figure in parenthesis. Each type of domain and motif is represented by different color-boxes.

anti-oxidative function, USPC and USPE, F, and G play in cellular adhesion, agglutination, cell motility, and swimming (Nachin et al., 2005). In contrast to the roles played by USPC and USPE, USPF and USPG belonging to Class II played different functions in cellular migration or movement. They negatively regulate bacterial mobility but positively control cell affixment and agglomeration. These results clearly indicate that the functions of various bacterial USPs are coordinated to enhance the stress tolerance of cells against harsh external circumstances.

The multiple functions of USPs in other bacteria are derived from their structural diversity. During the process of evolution, the USP domain is probably fused with other catalytic motifs to produce multi-structural USP proteins having diverse biochemical and molecular functions. Therefore, in addition to their single USP domain, USPs contain highly divergent other functional motifs, including protein kinase, Na+/H<sup>+</sup> exchanger, and amino acid permease motifs, as well as a voltage gated Cl<sup>−</sup> channel, whose function have not yet been clarified (**Figure 2B**; Kvint et al., 2003; Tkaczuk et al., 2013). The specific property of individual USPs under stress conditions might be dependent on their fused catalytic motifs. Thus, the combination of USP domain with other specific catalytic motif is likely to produce various functions that can protect host organisms from diverse external stresses. In reality, functional diversity of bacterial USPs has been demonstrated as follows; USP in Acinetobacter baumannii, a bacterium causing pneumonia and sepsis, is essential for protecting the pathogen from oxidative stress and respiratory toxins (Elhosseiny et al., 2015; O'Connor and McClean, 2017). Similarly, USP from Salmonella typhimurium plays a critical role in the bacterium's survival during oxidative stress (Liu et al., 2007), and USP in Listeria monocytogenes confers protecting activity for salt, acidity, and oxidative stress (Esvan et al., 2000; Gomes et al., 2011). Rv2623, a USP isoform in Mycobacterium tuberculosis, has an important role in mycobacterial growth and leads to chronic infection of humans (Drumm et al., 2009; Glass et al., 2017). Although physiological functions of bacterial USP are demonstrated in diverse cellular physiology and pathogenicity, the biochemical and molecular mechanism of most USPs remain largely unknown.

Similar to bacterial USPs, diverse forms of USP have been identified from different plant sources by searching the internet database, Ensembl Plants<sup>1</sup> , and found 2,141 USPs (**Table 1**). All the proteins contain at least one USP domain and other catalytic motifs, which are differentially expressed in specific tissues, organs, and developmental stages or under different stress conditions (Li et al., 2010; Wang et al., 2017). The result suggests that plant USPs exert their distinctive function in specific tissues and developmental stages under particular stress conditions. Numbers of USPs found in various plant species are summarized in **Table 1**. The largest number of USPs are found in Brassica napus which contains 142 USPs. And the genomes of Triticum aestivum, Brassica rapa, Solanum lycopersicum, Solanum tuberosum, Oryza sativa japonica, Vitis vinifera, and Zea mays have 123, 71, 42, 41, 38, 33, and 43 USP genes, respectively. Although numerous numbers of USP genes and their wide distribution in diverse plant species implicate their importance in plant growth and development, the physiological and biochemical functions of plant USPs remain largely unknown. Especially, from the fact that many USPs are found in crop plants, it may be proposed that USP genes might be multiplied during the domestication procedures. The evolutionary pathway of plant USPs, as well as their duplication and functional specificity, should be further studied to determine why plants have so many USPs.

Like the bacterial USPs, plant USPs have diverse functional motifs and a variety of structural characteristics. USPs in seven representative plant species including Oryza sativa, Medicago truncatula, Zea mays, Brachypodium distachyon, Setaria italica, Populus trichocarpa, and Arabidopsis thaliana are shown in **Figure 3A**. The most common type of USP in the plants has only a single USP domain, but the other proteins additionally contain a variety of other functional motifs. Using their amino acid sequences, phylogenetic tree of the three representative plant USPs in Oryza sativa, Zea mays and Arabidopsis thaliana is derived (**Figure 3B** and **Supplementary Figures 1, 2**). The phylogenetic tree of plant USPs strongly suggests that the functional diversity of plant USPs is much greater than those of bacterial USPs, because the latter is clustered within a very narrow range of evolutional tree. The other catalytic motifs found in plant USPs include serine/threonine kinase, tyrosine kinase, U-box, SWI2/snf2 and Mudr (SWIM)-zinc finger, HomeoDomain leucine zipper (HDzip), cation exchanger, C1 motif of Insensitive to Killer toxin3 (IKI3). The catalytic motifs of plant USPs may be derived over the course of evolution during the selective pressure against diverse stresses, which leads to the fusion of different catalytic motifs with a USP domain. The process provides plants for multiple strategies to protect them from foreign stresses (Kerk et al., 2003). The fusion of a cation exchange motif with the USP domain endows plants with protection from sodium toxicity. The cation exchange motif added to the USP domain pumps the sodium out from plant cells and prevents its accumulation. Fusion of protein kinase motif with a USP domain enables the protein to bind and

<sup>1</sup>http://plants.ensembl.org/

.

TABLE 1 | Numbers of USPs found in different plant species<sup>∗</sup>


<sup>∗</sup>The number of USPs in each plant species was obtained from the Ensembl Plants database (http://plants.ensembl.org/index.html). †Phylogenetic tree of the three representative plant USPs (Oryza sativa japonica, Zea mays, Arabidopsis thaliana) is constructed and presented in Figure 3B (Oryza sativa japonica) and in Supplementary Figures 1, 2 (Zea mays, Arabidopsis thaliana).

utilize ATP and performs the energetically unfavorable reaction. The combination of such functional motifs with a USP domain at their amino terminal or carboxy terminal region explicitly suggests that divergent USPs have evolved to play specific roles under particular stress conditions, which protects plants from kaleidoscopic circumstances. The extensive shuffling of USP domain with a wide variety of functional motifs has provided plants with diverse sophisticated tactics for their survival under variable external conditions.

#### Physiological Significance of Plant USPs in Biotic and Abiotic Defense Signaling

To protect plants from myriad of biotic and abiotic stresses caused by environmental stimuli, they have to develop highly advanced and sophisticated systems and devices (Conrath, 2011; Petrov et al., 2015; Martinez-Medina et al., 2016). Given the large numbers of plant USPs, it is reasonable to suppose that the proteins play crucial roles in diverse aspects of plant physiology and metabolism. Their functional diversity originated from the variety of other catalytic motifs fused with the USP domain is critically important for plant stress resistance. Although, only a

FIGURE 3 | Molecular structures of diverse plant USPs and the phylogenetic tree of 38 USPs in Oryza sativa japonica. (A) Molecular structures of diverse USPs in plant sources (Brachypodium distachyon, Setaria italic, Oryza sativa japonica, Medicago truncatula, Zea mays, Arabidopsis thaliana and Populus trichocarpa) containing only a USP domain or USP domains fused with other catalytic motifs that are obtained from Ensembl Plants database (http://plants.ensembl.org/index.html). The domain architectures of different USPs are obtained from the Uniprot database (http://www.uniprot.org/), and domains are predicted using the InterPro database (www.ebi.ac.uk/interpro). Numbers of USPs having the specific type of molecular structure in plants are indicated at the left side of the figure in parenthesis. Each type of domain and motif is represented by different color-boxes. (B) Phylogenetic tree of the 38 USPs in Oryza sativa extracted from the Ensembl Plants database (http://plants.ensembl.org/index.html). The tree was constructed with USP domains of 38 Oryza sativa USPs after deleting all other domain sequences with the use of Maximum Likelihood method in MEGA7 (Kumar et al., 2016). E. coil USPs, Methanocaldococcus jannaschii MJ0577 and Haemophilus influenza USPA are included as references in the phylogenetic tree. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The analysis involves 53 amino acid sequences. All positions containing gaps and missing data are eliminated. There are a total of 57 positions in the final dataset. Black, red, blue and violet color labeled USP proteins contain USP domain only, USP + Protein Kinase motifs USP + Protein Kinase\_Tyr motif and USP + Protein Kinase + U-box motif, respectively. EcUSPA, EcUSPC and EcUSPD are represented in green; EcUSPF and EcUSPG by sky blue; EcUSPE1 and EcUSPE2 by magenta; MJ0577 by gray; USPA by light green colors.

little information is available on the biochemical and functional properties of plant USPs yet, several cases are introduced on the physiological importance of individual USPs related to plant stress responses (**Table 2**).

All the forty-four USPs found in Arabidopsis genome contain an ATP-binding site and exhibit a high sequence homology to that of 1MJH protein family (Kerk et al., 2003). The proteins have diverse functions in protecting plants from different stresses as follows; HRU1, an Arabidopsis USP, regulates the intracellular level of hydrogen peroxide (H2O2) under hypoxic condition and transduces the oxygen-deficient signal into the downstream defense signaling pathway (Gonzali et al., 2015). Thus, at low oxygen concentration, it induces dissociation of the cytosolic form of dimeric HRU1 into monomeric HRU1, which translocates into the plasma membrane to interact with its partner, ROP2-RbohD, and make the HRU1-ROP2-RbohD complex (**Table 3**). The complex increase the intracellular concentration of H2O<sup>2</sup> and finely tunes H2O<sup>2</sup> level, that enables the plants to recover from anoxia. Another type of Arabidopsis USP, AtUSP, was identified from the cytosolic extracts treated with heat shock or oxidative stress. Under optimum conditions, AtUSP exists as diverse forms including monomer, dimer, trimer, and oligomeric complexes. When plants are exposed to heat shock and/or oxidative stress, the intracellular redox status is changed to make a shift of AtUSP from low molecular weight species to a high molecular weight complex (Jung et al., 2015). The protein structure of AtUSP existed as an inactive monomer or dimer under optimum conditions can be switched into a high oligomer complex in response to external stress, which provides the chance of the protein to acquire a novel function of


TABLE 2 | Physiological functions of USPs identified from different plant species.

'molecular chaperone' working at the plant cytoplasm (**Figure 4**, left panel). Then, the holdase chaperone function of AtUSP allows it to prevent denaturation of intracellular core macromolecules from heat shock or oxidative stress. This action resembles the general properties of most heat shock proteins. The heat shockmediated functional transition of AtUSP to a holdase chaperone ensures transgenic plants over-expressing AtUSP to have a strong resistance to heat or oxidative stress.

In addition to heat shock-mediated post-translational modification of AtUSP acting as a protein chaperone, mRNA level of AtUSP was significantly enhanced at low temperature, suggesting that AtUSP might have another specific role in cold stress (Melencion et al., 2017). From the subcellular localization of AtUSP in cold condition and investigation of RNAs stability with or without the presence of AtUSP, it can be concluded that the protein also gets another function of 'RNA chaperone' under cold condition functioning at the plant nucleus (**Figure 4**, right panel) (Kang et al., 2013; Melencion et al., 2017). In fact, low temperature induces over-stabilization or misfolding of RNA molecules, which are inactive to serve as a template RNAs (Lorsch, 2002; Kang et al., 2013). Then the RNA chaperone function of AtUSP enables the unwinding of over-stabilized RNA molecules and refolding them into their active structures, to provide them as RNA templates for their translation. Consequently, overexpression of AtUSP protects the transgenic plants from cold and freezing conditions. The physiological roles of AtUSP are therefore critical for protecting plants from both high and low temperatures, playing dual functions in plants to adapt for the environmental temperature fluctuation (Jung et al., 2015; Melencion et al., 2017). The result is also demonstrated in chilling tolerance of grapefruit (Citrus paradisi cv. Star ruby), that is achieved by altering the expression of its USP. The gene can be induced by exposing the fruit to a short pre-treatment with hot water and then briefly rinsing and brushing, which reduces injuries during subsequent chilling (Sapitnitskaya et al., 2006).

Apart from the temperature-associated function of AtUSP, another USP in Salicornia brachiata (SbUSP) is shown to involve in abiotic stress resistance (Udawat et al., 2016). Ectopic expression of SbUSP in tobacco plants significantly enhances salt tolerance, exhibiting an increased osmotic stress resistance through the removal of intracellular ROS. Under the salt and osmotic stress conditions, the osmosensor recognizes the cellular Na<sup>+</sup> level and activates the specific protein kinase involved in salt signaling. This protein kinase then phosphorylates serine and threonine residues of SbUSP. Next, the phosphorylated SbUSP activates the expression of downstream target genes, which causes an accumulation of osmoprotectant, alleviates intracellular ROS build-up, and protects plants from salt stress. SpUSP from tomato (Solanum pennellii) and GhUSP from cotton (Gossypium hirsutum L.) are also induced by salt stress, suggesting that the USPs play crucial roles in plant salt-stress tolerance (Loukehaich et al., 2012; Li et al., 2015). Besides the salt stress, SpUSP gene is noticeably induced by the treatment of drought, heat/cold shock, treatment of paraquat, wounding, and phytohormone (abscisic acid, gibberellic acid, and ethylene) (Loukehaich et al., 2012). In addition, transcript level of SlRd2 mRNA, another USP gene in tomato (Solanum lycopersicum), is critically enhanced by the treatment of salt and LiCl, suggesting that the physiological function of SIRd2 might be involved in salt and osmotic stress tolerance in plants (Gutierrez-Beltran et al., 2017). Thus, overexpression of SIRd2 in S. cerevisiae (BY4741) strongly exhibits osmotic tolerance as well as the expression of SlRD2 gene in uspA E. coli mutant (TN3151), displaying a highly sensitive phenotype to oxidative stress and clearly complements the mutant properties, which restores bacterial viability in the presence of 5 mM H2O2. Since the SIRd2 function is probably associated with the removal of intracellular ROS in plants, the expression of SlCipk6 and GFP-SIRd2 in tobacco plants decreases the ROS level. To carry out the assignment, SlCipk6 (Calcineurin B-like interacting protein kinase) phosphorylates the dimeric form of SIRd2, which negatively regulates the SlCipk6-mediated ROS production (**Table 3**) (Gutierrez-Beltran et al., 2017). In transgenic tomato overexpressing SpUSP, abscisic acid level is elevated under drought condition and induces stomatal closure to reduce water loss, that endows plants with drought tolerance

TABLE 3 | Interaction partners of USPs identified from plant sources and their functions.


their native forms of functional RNAs to serve them for protein translation.

and improved photosynthetic efficiency. During the procedure, SpUSP is shown to interact with annexin (AnnSp2) known as a target of calcium signaling in eukaryotes analyzed by yeast twohybrid and bimolecular fluorescence complementation (BiFC) techniques (**Table 3**) (Loukehaich et al., 2012). Annexin plays prominent roles in abiotic and biotic stress resistance in plants (Jami et al., 2008; Konopka-Postupolska et al., 2009; Szalonek et al., 2015; Konopka-Postupolska and Clark, 2017). Therefore,

similar to the plants overexpressing SpUSP, the overexpression of AnnSp2 in tomato critically enhances their drought tolerance through the stomatal closure, build-up of ABA and chlorophyll contents, reduction in water loss, elimination of ROS, decreasing the level of lipid peroxidation, increase in proline concentration and antioxidant activities (Ijaz et al., 2017).

Furthermore, transcription of GaUSP1 and GaUSP2 in Gossypium arboretum is induced by drought stress, indicating that these two GaUSPs function in the control of intracellular water content (Maqbool et al., 2008, 2009). Treatment with various stress-inducing factors, such as salt, dehydration, darkness, heavy metals, and phytohormones covering abscisic acid and gibberellic acid strongly increases the activity of cotton USP promoters. Activity of a 949 bp fragment of the cotton USP promoter is significantly increased in transgenic tobaccos during the stress treatment, as shown in USP mRNA levels (Zahur et al., 2009). Drought stress also upregulates the expression of USP genes in Amor cork tree (Phellodendron amurense) and pigeon pea (Cajanus cajan L.) (Wang et al., 2008; Sinha et al., 2015). Furthermore, in rice (Oryza sativa), OsUSP1 containing the conserved ATP-binding residues regulates phytohormone signaling to increase stress resistance and protects plants during submergence in water by regulating ethylene concentration in cells (Sauter et al., 2002).

Besides their roles in abiotic stress resistance, plant USPs also participate in defense response against pathogenic attack. Infection of the Chinese Milk Vetch (Astragalus sinicus) roots with the nodule-inducing leguminous bacterium, Mesorhizobium huakuii, results in an increased expression of USP gene AsD243, which suggests that AsD243 plays a role in nodule development (Chou et al., 2007). Following elicitation of Arabidopsis cells by treating the Phytophthora infestans zoospores or bacterial eliciting peptide, flagellin-22, two USPs including AtPHOS32 and AtPHOS34 are phosphorylated by mitogen-activated protein kinases (MAPKs). AtPHOS32, a substrate of MAP kinases 3 and 6, involves in pathogen defense signaling. Phosphorylated USPs thus appear to activate defense signaling in plants to provide protection against pathogenic attacks (Lenman et al., 2008; Merkouropoulos et al., 2008). All these results strongly suggest that plant USP genes are upregulated in response to diverse external stresses. As the wider biochemical functions and molecular properties of USPs remain largely obscure, future studies should focus on the identification of their roles in protecting plants against particular biotic/abiotic stress and also investigate their molecular mechanisms in more detail with the use of double or triple usp mutants, or mutants lacking specific catalytic motifs.

#### Redox Regulation of Plant USPs

Plants protect themselves from diverse internal and external stresses, including heat shock, freezing stress, salt, heavy metal toxicity, flooding, drought, and biotic pathogens, using complicated and dynamic strategies that are principally regulated by redox signaling (Chi et al., 2013; Geigenberger et al., 2017). Plants have developed delicate redox signaling systems that sense internal redox changes and respond by activating specific intracellular redox-mediated defense signaling pathways (Gonzalez-Bosch, 2018; Noctor et al., 2018). ROS, the byproducts of physiological O<sup>2</sup> metabolism including H2O2, superoxide anions (O<sup>2</sup> ·−), hydroxyl radical (OH·), and singlet oxygen (O2), are precisely controlled by enzymatic and nonenzymatic antioxidant defense systems (Nita and Grzybowski, 2016). They induce oxidative damage of cells and eventually result in cell death (Gill and Tuteja, 2010; You and Chan, 2015). A well-characterized typical redox system in plants includes glutaredoxin/thioredoxin proteins, which play a central role in the regulation of carbon metabolism and photosynthesis (Gutsche et al., 2015; Geigenberger et al., 2017; Nikkanen et al., 2017). In plants, redox proteins consist of multigene families with a large number of potential target molecules related to diverse aspects of cellular metabolism (Delorme-Hinoux et al., 2016; Geigenberger et al., 2017; Mata-Perez and Spoel, 2019). The proteins are core components of plant defense signaling pathways and act as a dynamic linker between stress perception and physiological responses. Redox proteins modulate target enzyme's activity by post-translational modification through the oxidation and reduction of their catalytic Cys residues in response to ROS changes. The 2-Cys peroxiredoxins (2- Cys Prxs) are ROS sensors that participate in redox signaling through their structural switching between the monomer form and oligomeric complexes (Konig et al., 2002; Jang et al., 2004; Perkins et al., 2015). Other redox proteins including thioredoxin-3, AtTDX, and AtNTRC, exhibit multiple functions in response to environmental stimuli via redox-dependent processes (Lee et al., 2009; Mayer, 2012; Chae et al., 2013; Chi et al., 2013). Besides the redox proteins, there are different transcriptional factors, such as Rap2.4a and AtbZIP16 and Non-expresser of Pathogenesis Related gene 1 (NPR1), regulating the expression of stress-responsive genes or antioxidant enzymes by redoxand structure-dependent manner (Shaikhali et al., 2008, 2012; Montrichard et al., 2009; Chi et al., 2013).

In particular, the chaperone function and structural change of AtUSP is altered by treatment with dithiothreitol (DTT) and/or H2O<sup>2</sup> (Jung et al., 2015). AtUSP with multimeric complexes in optimum conditions changed into a monomer, accompanying with a decrease in its chaperone activity by DTT treatment. In reverse, the monomer form of AtUSP shifts into high oligomeric complex by H2O2, together with an increase in its chaperone activity. These results suggest that functional and structural switching of AtUSP is regulated by redox-dependent manner like other redox proteins, transcription factors and coactivators. The functions of USPs identified from plant sources are involved in modulating ROS concentration produced by diverse environmental stress (**Table 2**). In Arabidopsis, the interaction of three USPs (HRU1, AtUSP, and At3g17020) with a redox partner, thioredoxin-h1, have been determined by affinity chromatography, yeast two-hybrid analysis, and BiFC assays (**Table 3**) (Ueoka-Nakanishi et al., 2013; Gonzali et al., 2015). The thioredoxin-h1 is known to regulate the activity of calciumdependent protein kinase 21 (AtCPK21) and reactivates the oxidized AtCPK21 under oxidative stress with a redox regulation (Ueoka-Nakanishi et al., 2013). Since the HRU1 and AtUSP are target proteins of thioredoxin-h1 as well as their amino acid sequences contain two conserved cysteine residues, the results

provide a strong possibility on the involvement of USPs in redox regulation. From the results, it can be clearly demonstrated that the structural switching of AtUSP is induced by redox change in response to external stress, accompanying with its functional alteration. The importance of the cysteine residues in this redoxdependent regulation of USPs requires further investigation.

#### Biotechnological Application of USPs for the Development of Stress-Tolerant Valuable Crops

Controlling the expression of specific plant genes results in the activation of numerous biological signaling pathways and intracellular metabolic networks that influence plant growth, development, physiology, and productivity (Finkel, 2011). It is well known that, in plants, the complex signaling networks involved in stress responses mutually regulate each other's activities through the cross-talk involved in their communication and translational and post-translational modification. Plant USPs participate in a number of cellular metabolism to regulate defense systems against diverse external stresses. Regulating the expression of USP genes may therefore provide a powerful strategy for the development of stress-tolerant varieties of crop plants. The success of such an approach requires detailed understanding of USP functions in molecular basis underlying the stress tolerance responses in plants. As plant USPs have diverse roles in defense responses in response to ever-changing environmental stresses, it may be necessary to manipulate their expression to produce highly valuable, stress-tolerant crops that have valuable application in the agricultural fields. Physiological importance of USPs in plants strongly support the idea that control of USP gene expression in important crop plants in combination with other techniques, such as molecular breeding and genetic engineering may produce novel and high productive crop varieties (Jung et al., 2015; Udawat et al., 2016; Gutierrez-Beltran et al., 2017; Melencion et al., 2017). Thus, under conditions involving unfavorable environmental stresses, such as climate change, extreme temperatures, and other severe environmental problems, projects applying an understanding of USP gene function are likely to be highly important to the preparation of future varieties of stress-tolerant crops.

#### CONCLUSION

This review highlights the important roles played by USPs in the survival of all living organisms, including bacteria, Archaea, fungi, plants, and metazoans, in the face of diverse environmental stresses. Despite the great importance of USPs, their molecular properties remain largely unknown. They are widely distributed across different cell types and species, indicating their significance

#### REFERENCES

Akimoto-Tomiyama, C., Tanabe, S., Kajiwara, H., Minami, E., and Ochiai, H. (2018). Loss of chloroplast-localized protein phosphatase 2Cs in Arabidopsis thaliana leads to enhancement of plant immunity and resistance to in plant tissues, organs, and physiology. In plants, their functions include acting as protein chaperone and RNA chaperone, nucleotide binding, and prevention of hypoxia, and thus USPs offer protection from a wide range of external stresses. As USPs have versatile structures, resulting from the fusion of the USP domain with many other catalytic motifs, it is highly likely that these proteins are involved in multiple reactions and diverse cellular processes under stressful conditions. Furthermore, the other catalytic motifs allow functional diversity by enabling structural switching from small molecular species to high molecular complexes in response to external stresses. Exposure to heat shock and oxidative shock, in particular, induces the formation of high molecular complexes that function as protein chaperones, preventing denaturation of crucial intracellular molecules due to thermal stress. As there are many USPs in plants, it will be necessary to unravel the functional specificity of individual USPs in different species. The greatest challenge facing investigators of plant USPs is the determination of their physiological and biochemical functions in relation to plant metabolism. Uncovering these functions may unlock new biotechnological applications and lead to the development of valuable, stress-resistant crops. Through an applied understanding of the function of USPs, it may be possible to develop novel varieties with high productivity under unfavorable growth conditions. This should provide a focus for future investigation.

#### AUTHOR CONTRIBUTIONS

YC, MK, W-YK, D-JY, and SL made a substantial contribution to the conception, design, and writing of this manuscript. SK, HO, EL, JP, KP, SW, SB, SP, HC, and CK collected USP gene and protein datasets from plant sources and designed the figures and tables.

#### FUNDING

This work was supported by a grant from the NG-BioGreen 21 Program (SSAC, PJ013173), RDA, Korea, and by a Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (Grant No. 2018R1D1A1B07047990).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00750/ full#supplementary-material

Xanthomonas campestris pv. campestris infection. Mol. Plant Pathol. 19, 1184– 1195. doi: 10.1111/mpp.12596

Aravind, L., Anantharaman, V., and Koonin, E. V. (2002). Monophyly of class I aminoacyl tRNA synthetase, USPA, ETFP, photolyase, and PP-ATPase nucleotide-binding domains: implications for protein


genes in barley. Theor. Appl. Genet. 121, 907–917. doi: 10.1007/s00122-010- 1359-9


Sheng Wu Gong Cheng Xue Bao 24, 198–202. doi: 10.1016/s1872-2075(08) 60010-2


function of a hypothetical protein: a test case of structural genomics. Proc. Natl. Acad. Sci. U.S.A. 95, 15189–15193. doi: 10.1073/pnas.95.26.15189

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

Copyright © 2019 Chi, Koo, Oh, Lee, Park, Phan, Wi, Bae, Paeng, Chae, Kang, Kim, Kim, Yun and Lee. 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.

# Communications Between the Endoplasmic Reticulum and Other Organelles During Abiotic Stress Response in Plants

*Linchuan Liu1,2 and Jianming Li1,2,3 \**

*1 State Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources, South China Agricultural University, Guangzhou, China, 2 Guangdong Key Laboratory for Innovative Development and Utilization of Forest Plant Germplasm, College of Forestry and Landscape Architecture, South China Agricultural University, Guangzhou, China, 3 Department of Molecular, Cellular, and Developmental Biology, University of Michigan, Ann Arbor, MI, United States*

#### *Edited by:*

*Abel Rosado, University of British Columbia,Canada*

#### *Reviewed by:*

*Giovanni Stefano, Michigan State University, United States Yohann Boutté, UMR5200 Laboratoire de biogenèse membranaire (LBM), France*

#### *\*Correspondence:*

*Jianming Li jmlumaa@scau.edu.cn; jian@umich.edu*

#### *Specialty section:*

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

*Received: 24 December 2018 Accepted: 21 May 2019 Published: 12 June 2019*

#### *Citation:*

*Liu L and Li J (2019) Communications Between the Endoplasmic Reticulum and Other Organelles During Abiotic Stress Response in Plants. Front. Plant Sci. 10:749. doi: 10.3389/fpls.2019.00749*

To adapt to constantly changing environmental conditions, plants have evolved sophisticated tolerance mechanisms to integrate various stress signals and to coordinate plant growth and development. It is well known that inter-organellar communications play important roles in maintaining cellular homeostasis in response to environmental stresses. The endoplasmic reticulum (ER), extending throughout the cytoplasm of eukaryotic cells, is a central organelle involved in lipid metabolism, Ca2+ homeostasis, and synthesis and folding of secretory and transmembrane proteins crucial to perceive and transduce environmental signals. The ER communicates with the nucleus *via* the highly conserved unfolded protein response pathway to mitigate ER stress. Importantly, recent studies have revealed that the dynamic ER network physically interacts with other intracellular organelles and endomembrane compartments, such as the Golgi complex, mitochondria, chloroplast, peroxisome, vacuole, and the plasma membrane, through multiple membrane contact sites between closely apposed organelles. In this review, we will discuss the signaling and metabolite exchanges between the ER and other organelles during abiotic stress responses in plants as well as the ER-organelle membrane contact sites and their associated tethering complexes.

Keywords: membrane contact sites, endoplasmic reticulum, unfolded protein response, lipid exchange and transport, calcium homeostasis, reactive oxygen species

#### INTRODUCTION

Plants growing under natural habitats have to deal with various environmental stresses during their growth and development. Abiotic stresses such as extreme cold and hot temperatures, drought, salinity, and nutrient deficiency can greatly affect plant growth and crop productivity. Plants have evolved various sophisticated strategies to respond to different environmental stimuli at different levels from alternations in gene expression to changes in morphology (Nakashima et al., 2009; Su et al., 2013). The sensing and transduction of the environmental signals in stressed plants were intensively studied in the past several decades, revealing potential strategies to improve plant stress tolerance and agricultural productivity. It is generally believed that plant cells sense external environmental stimuli by various sensors, which are localized on the plasma membrane (PM), in the cytosol, or inside organelles. These environmental sensors activate intracellular signaling cascades that involve Ca2+, lipids, reactive oxygen species (ROS), and phytohormones (Osakabe et al., 2013; Zhu, 2016), ultimately inducing changes in gene expression, protein production, and metabolic pathways to enhance plant stress tolerance. Therefore, coordinated signaling between various intracellular compartments with distinct biochemical processes plays an important role in maintaining cellular homeostasis for the plant stress tolerance.

The endoplasmic reticulum (ER) is a central network of interconnected tubules and flattened cisternae that extend throughout the entire cytoplasm of the eukaryotic cells (**Figure 1**). The ER network occupies a large volume of the cytoplasm, with its membrane accounting for ~50% of total cellular membranes, and functions in protein processing and folding, lipid biosynthesis, and Ca2+ storage (Stefano and Brandizzi, 2018). In eukaryote cells, about one-third of newly synthesized proteins enter the ER where they are glycosylated, folded, and/or assembled into protein complexes. The ER houses several stringent quality control mechanisms that export only correctly folded and properly assembled proteins to continue their secretory journeys (Hetz et al., 2015). However, protein folding in the ER is an errorprone process that could easily be disturbed by various abiotic and biotic stresses, leading to accumulation of mis/unfolded

plasma membrane (PM) through MCSs. The pointed extensions of a peroxisome and a chloroplast represent peroxule and stromules, respectively. Question marks indicate MCSs that have not yet characterized. MCSenriched proteins are directly involved in physical tethering; mediate organelle biogenesis; and regulate exchanges of lipids, Ca2+, ROS, and other important metabolites and signaling molecules.

proteins in the ER and causing ER stress (Angelos et al., 2017). Currently, the unfolded protein response (UPR) is widely considered as a significant intracellular signaling pathway that links the ER proteostasis with gene regulation in the nucleus to alleviate the ER stress. Given its characteristic dynamic architecture and its essential roles in producing proteins and lipids for other organelles and maintaining Ca2+ homeostasis, the ER makes numerous physical contacts with other organelles and endomembrane compartments (**Figure 1**; Stefano and Brandizzi, 2018; Wu et al., 2018). Recent studies have identified many so-called ER-membrane contact sites (MCSs) that facilitate exchanges of important metabolites and signaling molecules between the ER and various organelles (Prinz, 2014; Wang and Dehesh, 2018; Wu et al., 2018). In this review, we will discuss recent results on the inter-organellar communications between the ER and other organelles during plant abiotic stress responses as well as the ER-organelle physical contacts and their associated tethering complexes.

#### THE ENDOPLASMIC RETICULUM-NUCLEUS INTERACTION *VIA* UNFOLDED PROTEIN RESPONSE

In addition to the physical ER-nuclear envelop connection (**Figure 1**), the ER-nucleus interaction is mediated by a highly conserved signaling mechanism known as UPR, which is activated by accumulation of misfolded proteins in the ER. Because protein folding is an error-prone process that can easily be disturbed by various environmental stresses, UPR is closely connected to the plant stress tolerance (Liu and Howell, 2016). In plants, the UPR pathway is principally mediated by two major branches that are conserved in mammalian cells (Howell, 2013). One arm is mediated by two homologous ER membrane-anchored bZIPfamily transcription factors, bZIP17 and bZIP28 that are activated by regulated intradomain proteolysis (Liu et al., 2007a). bZIP17 was originally identified as a transcription factor activated by salt stress (Liu et al., 2007b), while bZIP28 was discovered to be activated by heat stress (Gao et al., 2008). Both bZIP17 and bZIP28 are type II transmembrane proteins with a single transmembrane domain (TMD) and a DNA-binding/transcriptional activation bZIP domain facing the cytosol, and a C-terminal domain inside the ER lumen (Sun et al., 2013). In response to ER stress, bZIP17 and bZIP28 dissociate from the major ER luminal chaperone, binding immunoglobulin proteins (BiPs), and traffic from the ER to the Golgi where the two bZIP proteins are proteolytically processed by the Golgi-resident site-1 and site-2 proteases (S1P and S2P), thus releasing their N-terminal cytosolic domains that move into the nucleus (Andersson et al., 2007; Liu et al., 2007b; Gao et al., 2008; Srivastava et al., 2012; Iwata et al., 2017). The nuclear-localized bZIP17/28 proteins bind to their target promoters to increase expression of genes encoding ER chaperones, folding catalysts, and components of the ER-associated degradation (ERAD) machinery, which work together to restore the ER homeostasis (Liu and Howell, 2010). Interestingly, high light intensity increases ER stress sensitivity of plants *via* a competitive inhibitory interaction of bZIP28 with LONG HYPOCOTYL5 (HY5), a bZIP protein that positively regulates light signaling but suppresses the UPR pathway (Nawkar et al., 2017). The other arm of the plant UPR pathway involves the unconventional splicing of the mRNA of another bZIP transcription factor, bZIP60, which is catalyzed by the ER membrane-anchored inositol-requiring enzyme 1 (IRE1) (Deng et al., 2011; Nagashima et al., 2011). IRE1 is the most conserved ER stress sensor among yeast, plants, and animals and is a dual-functional protein with both protein serine/threonine kinase and endoribonuclease (RNase) activities. *Arabidopsis* has two IRE1 homologs, IRE1a and IRE1b (Koizumi et al., 2001). Under the ER stress, IRE1a and IRE1b can form homodimers or heterodimers to trigger their RNase activities, which splice the *bZIP60* mRNA (Howell, 2013). The frame-shift splicing of the *bZIP60* mRNA causes production of the active form of bZIP60 (bZIP60s, s for spliced) that lacks a transmembrane domain and can thus move into the nucleus to bind promoters of its target genes (Deng et al., 2011; Nagashima et al., 2011; Iwata and Koizumi, 2012). In addition to its *bZIP60* mRNA splicing role, the *Arabidopsis* IRE1s also participate in selective degradation of certain mRNAs of secretory pathway proteins and inhibitory proteins of the ER stress-induced autophagy, a process known as regulated IRE1-dependent decay of mRNAs (RIDD) (Mishiba et al., 2013; Bao et al., 2018). In plants, the ER stress responses are closely related to abiotic stress tolerance. *Arabidopsis* mutants defective in bZIP17, bZIP28, and/or bZIP60 show increased sensitivity to various environmental stresses whereas overexpression of the active forms of the three bZIP proteins enhances the plant stress tolerance (Fujita et al., 2007; Liu et al., 2007b; Kataoka et al., 2017; Ruberti et al., 2018). A recent study also implicated bZIP17 and a component of the *Arabidopsis* ERAD machinery in salt acclimation memory that enables plants to tolerate severe salt stress (Tian et al., 2018).

# THE ENDOPLASMIC RETICULUM-GOLGI RELATIONSHIP

The ER and the Golgi apparatus are the first two membrane compartments in the protein secretory pathway. Unlike the mammalian cells in which the ER and the Golgi apparatus are separated by the ER-Golgi intermediate compartment (ERGIC, also known as the vesicular-tubular cluster or VTC), the ER and the Golgi complex are often physically attached in plant cells at ER exit sites (ERES) (**Figure 1**; Sparkes et al., 2009), although recent studies suggested the presence of an ERGIC-like compartment termed as GECCO for Golgi entry core compartment in plant cells (Ito et al., 2012, 2018). The ER-Golgi interaction involves the coat protein complex II (COPII)-mediated cargo export from the ER and the COPI-mediated retrieval of ER-resident proteins from the Golgi. Due to the existence of high stringent quality control mechanisms, only the correctly folded and properly assembled proteins can be exported from the ER into the Golgi, whereas those incompletely-/misfolded and improperly assembled proteins are retained in the ER for chaperone-assisted refolding or removal by ERAD that involves cytosolic proteasomes (Brandizzi and Barlowe, 2013; Liu and Li, 2014). In the Golgi complex that includes the *trans*-Golgi network (TGN), the ER-derived protein cargos undergo N-glycan maturation and are sorted by vesicle-dependent/independent trafficking pathways to specific destinations to carry out their cellular functions. Live cell imaging revealed that the plant Golgi apparatus is a highly dynamic organelle with dispersed stacks of cisternae that are often physically attached to the ER tubules (**Figure 1**; Sparkes et al., 2009). Additionally, the shape and architecture of the Golgi complex are flexible enough to adapt to the functional status of different plant cells (Dupree and Sherrier, 1998). These functional and physical connections between the ER and the Golgi complex not only ensure normal cellular activities but are also essential for the survival of plant cells during stress conditions.

Recent studies have shown that several *Arabidopsis* mutants deficient in the ER-Golgi/Golgi-ER vesicle trafficking exhibit the ER stress and are hypersensitive to abscisic acid (ABA) and salt stress (Zhao et al., 2013, 2018; Pastor-Cantizano et al., 2018), suggesting that the bidirectional vesicle transport between the ER and Golgi is crucial for maintaining cellular homeostasis and adaptation to environment stresses. In addition to vesicular trafficking, accumulating evidence indicates the existence of non-vesicular transport connecting the ER and Golgi. Threedimensional electron microscopy and Forster resonance energy transfer-based fluorescence lifetime imaging microscopy revealed the physical contacts between the ER subdomains and *trans*-Golgi/TGN in mammalian cells (Ladinsky et al., 1999; Venditti et al., 2019b). No ER-*trans*-Golgi/TGN (referred hereinafter as ER-TG) contact has been observed so far in plant cells, but laser trap was used to reveal the ER-*cis*-Golgi interaction in plant cells, which occurs at ERES where the mobile Golgi stacks are associated with COPII components (**Figure 1**; Dasilva et al., 2004; Hawes et al., 2008; Sparkes et al., 2009). AtCASP, a homolog of a yeast/mammalian transmembrane Golgi protein known as CCAAT-displacement protein alternatively spliced product (CASP) was recently identified as a component of a novel tethering complex that connects ERES with the *cis*-Golgi to form the so-called "mobile secretory unit" (Osterrieder et al., 2017). The *cis*-Golgi-localized AtCASP could interact with ERES-enriched proteins to mediate the ER-*cis*-Golgi tethering that likely increases the efficiency of COPII vesiclemediated cargo transport *via* the so-called "hug-and-kiss" mechanism (Kurokawa et al., 2014). Identification of potential AtCASP-binding proteins that are enriched at ERES could discover additional components of the ER-*cis*-Golgi tethering complex that might help to resolve the controversy on the mechanism of the ER-Golgi transport (Robinson et al., 2015) and explain the "sticky" nature of the plant *cis*-Golgi cisterna (Sparkes et al., 2009).

Mammalian cells lack the ER-*cis*-Golgi physical contact but contain multiple ER-TG contact sites that are implicated in the non-vesicle-mediated lipid exchange (**Figure 1**; Mesmin et al., 2019; Venditti et al., 2019b). Several lipid transfer proteins (LTPs) localized at the ER-TG interface were identified, such as CERT (ceramide-transfer protein), FAPP2 (four-phosphate adaptor protein 2), and OSBP (oxysterol-binding protein), which mediate the vesicle-independent ER-TG transport of ceramide, glucosylceramide, and cholesterol (coupled with counter-transport of phosphatidylinositol-4-phosphate), respectively (Mesmin et al., 2019; Venditti et al., 2019a). All three LTPs share similar protein domains important for the ER-TG bridging, including a TGN-binding N-terminal pleckstrin homology (PH) domain, a central FFAT (diphenylalanine in an acidic tract) motif exhibiting specific binding to the ER-localized vesicle-associated membrane protein-associated proteins (VAPs), and the C-terminal oxysteroid-binding domain. Almost nothing is known about the ER-TG contact in plant cells, but the *Arabidopsis* genome encodes multiple homologs of CERT/FAPP2/OSBP (Umate, 2011) that lack the FFAT motif and a total of 12 VAP homologs known as plant VAP homologs (PVAs) (Sutter et al., 2006). One of the *Arabidopsis* OSBP-related proteins (ORPs), ORP3a, is localized to the ER *via* its interaction with an ER-localized PVA, PVA12 through a WFDE (tryptophan-phenylalanineaspartate–glutamate) motif located on the surface of ORP3a (Saravanan et al., 2009). It remains to be investigated whether or not plant cells have the ER-TG physical contacts, and if so, whether some of the *Arabidopsis* homologs of CERT/FAPP2/ OSBP interact with ER-localized PVAs to mediate the ER-TG tethering and the ER-TG lipid/sterol exchanges.

# THE ENDOPLASMIC RETICULUM-MITOCHONDRIA CONNECTION

Mitochondrion is an intracellular double-membrane organelle found in all eukaryotic cells. It not only provides cellular energy and metabolic intermediates but also participates in many other cellular processes, such as ROS signaling, Ca2+ buffering, cell differentiation, and apoptosis (Labbé et al., 2014). Under changing environmental conditions, plants have to adjust their metabolism to balance their energy production and consumption through mitochondria. Recently, a growing body of evidence suggests that mitochondria and the ER cooperate in several biosynthetic pathways and exchange signaling molecules during stress conditions (Mueller and Reski, 2015; Wang and Dehesh, 2018). It is well known that environmental stresses, such as heat, drought, salinity, and high light intensity, increase production and accumulation of ROS in mitochondria, which not only serves an important intracellular signal (at low concentrations) to regulate various cellular pathways but also causes oxidative damage (at high concentrations) to the cellular components (Suzuki et al., 2012; Das and Roychoudhury, 2014). ROS can also be generated in the ER lumen, which has a higher redox potential (~100 mV) than that of other cellular compartments (Birk et al., 2013). The oxidative protein folding process in the ER is mediated by protein disulfide isomerases (PDIs) and a flavin adenine dinucleotide-binding protein, ER oxidoreductase 1 (Ero1), which produces H2O2 as a result of electron flow from target proteins *via* the PDI-Ero1 couple to O2 (Tu and Weissman, 2002; Santos et al., 2009; Higa and Chevet, 2012). Due to the H2O2 permeability of the ER membrane (Ramming et al., 2014), the ER-induced oxidative stress can influence the production of mitochondrial ROS likely mediated by the ER-mitochondria physical contacts (Bhandary et al., 2012; Murphy, 2013; Zeeshan et al., 2016). On the other hand,

The ER-mitochondria contact is also essential to build the membrane system of mitochondria that import most lipids from other organelles (Li-Beisson et al., 2017). The ER-mitochondria tethering allows lipid exchanges between two apposed membranes and/or permits access of the membrane-localized enzymes to lipid substrates on the tethered membrane (Michaud et al., 2017). In yeast, the ER-mitochondria encounter structure (ERMES) is the most well-defined ER-mitochondria tethering complex that facilitates the ER-mitochondria phospholipid exchanges (**Figure 1**; Michel and Kornmann, 2012; Lang et al., 2015). The yeast ER-mitochondria tethering also involves another complex known as the ER membrane complex (EMC)-translocase of outer membrane 5 kDa subunit (TOM5) complex (Lahiri et al., 2014). In mammalian cells, the ER-mitochondria interface, known as mitochondria-associated ER membrane (MAM), has more complicated protein complexes involved in physical tethering, Ca2+ regulation, lipid exchanges, mitochondrial fission, autophagy, and apoptosis (Lee and Min, 2018). In plants, despite visual evidence for the ER-mitochondria physical interaction that likely plays a role in mitochondrial fission and the ER-mitochondria coordinated biosynthesis and exchanges of phospholipids (**Figure 1**; Mueller and Reski, 2015; Michaud et al., 2017), no homologs of the yeast ERMES were found in plants that also lack homologs of a majority of known mammalian MAM proteins (Duncan et al., 2013; Michaud et al., 2017). The *Arabidopsis* genome does encode homologs of three of the six components (EMC1, 2, 3, 5, 6, and TOM5) of the EMC-TOM5 complex (Michaud et al., 2016) and homologs of mitofusin1 (MFN1), a mitochondrial fusion GTPase that interacts with its ER-localized homolog MFN2 to mediate the ER-mitochondria tethering (Detmer and Chan, 2007; de Brito and Scorrano, 2008). However, the two *Arabidopsis* MFN1/2 homologs, DRP3A/3B and FZL, are not involved in mitochondrial fusion (Arimura, 2018), and there is no report on the involvement of the three homologs of the yeast EMC-TOM5 complex in the ER-mitochondria tethering in plant cells. A recent study identified a *Physcomitrella patens* protein, MELL1 (mitochondria-ER-localized LEA-related LysM domain protein 1) that regulates the numbers of the ER-mitochondria contact sites and could thus be a component of the plant ER-mitochondria tethering complex (Mueller and Reski, 2015). It will be interesting to determine whether MELL1 is conserved in higher plants and if so, whether the MELL1 homologs are a component of the yet to be identified ERMES/MAM in higher plants and required for the phospholipid biosynthesis/exchange of the ER and mitochondria. The lipid exchanges between the ER and mitochondria also involve lipid trafficking between the inner membrane (IM) and outer membrane (OM) of the mitochondria. A recent study implicated a mitochondrial transmembrane lipoprotein (MTL) complex containing the TOM complex and IM-localized AtMIC60, an *Arabidopsis* homolog of the yeast MIC60 that is a component of the well-studied mitochondria contact site and cristae organizing system (MICOS) (Pfanner et al., 2014), in the IM-OM lipid trafficking (Michaud et al., 2016). It is thus possible that the TOM complex, through its interaction with IM-localized AtMIC60 capable of extracting membrane lipid and the ER-localized homologs of the yeast EMC-TOM5 complex, functions as a crucial component of a plant ER-mitochondria tethering complex to mediate lipid exchanges or coordinate lipid biosynthesis.

The ER-mitochondria physical contact is also essential for the Ca2+ cross talk between the two organelles, which is often influenced by ROS. In plants, a variety of environmental stimuli trigger Ca2+ transients, such as the influx of Ca2+ into the mitochondrial matrix, to regulate gene expression and metabolism (Carraretto et al., 2016). However, the ER is generally considered the main intracellular Ca2+ store. The Ca2+ channels located at the ER-mitochondria contact sites, such as the mitochondrial outer membrane-localized VDAC (voltage-dependent anion-selective channel) and the ER membrane-anchored inositol triphosphate-dependent calcium channel IP3R, are believed to mediate the transport of Ca2+ between the ER and mitochondria in response to ER stress in mammalian cells (Lee and Min, 2018). The mammalian ER-localized Ca2+ release channel ryanodine receptor is activated by Ero1-generated H2O2 (Anelli et al., 2012). It remains to be determined if the ER ROS also regulates the ER Ca2+ release in plant cells that lack the homologs of the mammalian ER Ca2+ efflux channels IP3R and ryanodine receptor (Stael et al., 2012).

Two recent studies revealed another interesting mechanism by which the ER interacts with the mitochondria in plant cells. The mitochondrial retrograde regulation (MRR), which transmits the stress-induced mitochondrial signal into the nucleus to increase production of certain mitochondrial proteins for sustaining or restoring the mitochondrial functions during stressful conditions (Dojcinovic et al., 2005), was shown to involve two ER-anchored NAC transcription factors, ANAC013 and ANAC017 (De Clercq et al., 2013; Ng et al., 2013). *ANAC013* knockdown lines and an *ANAC017* knockout mutant were hypersensitive to stress than their wild-type controls. It was hypothesized that the mitochondrial stress somehow activates yet unknown proteases that proteolytically activate the two ER-anchored ANAC proteins that can subsequently translocate into the nucleus (Wang et al., 2018c). It will be interesting to test if the proteolytic activation of the two NAC-type transcription factors occurs at ERMES/MAM in plant cells. Proteomic experiments with stressed *Arabidopsis* plants expressing non-cleavable variants of ANAC013/017 might lead to identification of potential components of the *Arabidopsis* ERMES/ MAM. It is also interesting to note that the two ANACs were recently implicated in coordinating mitochondrial and chloroplast functions *via* their physical interactions with a nuclear protein Radical-induced Cell Death1 (RCD1) that was known to be regulated by ROS (Shapiguzov et al., 2019).

#### THE ENDOPLASMIC RETICULUM-PLASMA MEMBRANE CONTACT

The plasma membrane (PM), a lipid bilayer embedded with proteins, is an essential cellular component for the plant stress tolerance. It not only serves as a physical barrier to shield cellular contents from the extracellular environment and controls the flux of solutes and macromolecules but also contains a wide range of sensors and receptors that perceive and transmit all kinds of environmental signals. As discussed above, the ER not only produces, folds, and assembles the PM-localized channels/transporters and receptors/sensors but also delivers lipids to the PM and other intracellular compartments *via* vesicle-dependent and/or independent mechanisms.

The ER-PM contact sites (EPCSs) are evolutionarily conserved microdomains that are important for the ER-PM communications, such as lipid homeostasis, and Ca2+ influx (**Figure 1**; Saheki and De Camilli, 2017). The composition of EPCSs and their molecular functions have been well established in the yeast and mammalian cells in the last decade (Stefan, 2018). The yeast EPCSs consists of six proteins: three tricalbins, Increased sodium tolerance protein 2 (Ist2), and the ER-resident protein Scs2/22 (Suppressor of choline sensitivity 2/22) (Manford et al., 2012). The mammalian EPCSs contains three tricalbin homologs known as E-Syts for extended synaptotagmin (Giordano et al., 2013) and two Scs2/22 homologs, VAP-A and VAP-B, but lacks an Ist2 homolog (Selitrennik and Lev, 2016). In plants, the EPCS complex is the best known protein tether of the plant ER MCSs and consists of VAP27, VAP-Related Suppressor of TMM (VST), an actin-binding protein NETWORKED 3C (NET3C), actin filaments, and microtubule networks (**Figure 1**; Wang et al., 2014, 2016, 2017, 2018a; Ho et al., 2016). In particular, a phospholipid-binding protein Synaptotagmin1 (SYT1), which is the plant homolog of tricalbin/E-Syts, was found in the plant EPCS complex (Perez-Sancho et al., 2015) and subsequently used as a marker for the plant EPCS for microscopic studies (McFarlane et al., 2017; Lee et al., 2019). SYT1 has been previously described as an essential component for maintaining the PM integrity, especially under conditions of high risks of membrane disruption such as osmotic shock, freezing, and salt stresses (Schapire et al., 2008). Other studies have shown that SYT1 is required for tethering the ER to the PM and plays an essential role in regulating the ER remodeling and the stability of EPCSs (Siao et al., 2016). A recent study revealed that the ER-anchored SYT1 directly binds the PM-localized phosphatidylinositol 4,5-bisphosphate [PI (4,5)P2] to establish EPCSs (Lee et al., 2019), thus revealing a physiological function of the stressed-induced PM accumulation PI(4,5)P2 (Heilmann, 2008). It is likely that the protein-lipid tether could be disrupted or strengthened by additional SYT1/PI(4,5)P2-binding proteins.

EPCSs are now widely accepted as important sites for the non-vesicular lipid transport, which appears to be the major transport route of certain lipid species (Lev, 2012). Plants exposed to abiotic stresses have to adapt their membrane lipid composition and fluidity to changing environmental conditions by adjusting the relative amounts of various lipids, such as phospholipids and galactolipids (Hou et al., 2016). It is well known that lipids synthesized in the ER need to be delivered to other membranes for assembly of biological membranes or for lipid-mediated signaling cascades. It is proposed that the lipid transfer proteins (LTPs) are localized at the EPCSs and function as dynamic tethers between the two membranes with their lipid transfer module regulating lipid exchange (Dickson et al., 2016; Quon et al., 2018). Mammalian VAPs are known to interact with proteins involved in lipid transfer (Gatta et al., 2015) while SYT1 contains a synaptotagmin-like mitochondrialipid-binding protein (SMP) domain that is implicated in lipid transfer in mammals (Schauder et al., 2014). It is likely that the plant EPCSs are also involved in the ER-PM lipid transfer and thus play important role in plant stress tolerance by modulating the composition and fluidity of the PM. The EPCS is also important for the intracellular Ca2+ homeostasis in mammalian cells. The ER-PM contacts are critically implicated in generating the cytosolic Ca2+ signals, which is likely mediated by Ca2+ release from the ER in response to the PM-perceived environmental stimuli, and in replenishing the depleted ER Ca2+ store (Chung et al., 2017). Given the importance of Ca2+ signaling in plant stress response (Ranty et al., 2016), it would be interesting to investigate the role of EPCSs in regulating the stress-triggered intracellular Ca2+ dynamics in plants.

In addition to the EPCS-mediated exchange of lipids and Ca2+, there are other mechanisms that connect the ER physiology to the PM function in plant stress response. A recent study implicated a PM-localized NAC transcription factor, ANAC062, in the ER-nucleus-mediated UPR pathway (Yang et al., 2014). It is quite possible that the ER stress could increase the EPCS formation, altering the local membrane lipid composition to enhance the proteolytic processing of the PM-anchored ANAC062 (Seo et al., 2010). The cleaved ANAC062 can then move into the nucleus to regulate UPR-related genes, thus helping to mitigate the ER stress. Other studies found that the increased cytosolic Ca2+ caused by the stress-triggered Ca2+ release from the ER could activate the PM-localized NADPH oxidase, which was known to be induced by UPR and is required to survive ER stress (Ozgur et al., 2015, 2018; Angelos and Brandizzi, 2018). It is quite tempting to speculate that the Ca2+-mediated activation of the PM-localized NADPH oxidase might require EPCSs. It is important to note that the plant NADPH oxidase is the most well-studied ROS enzymatic system and plays a key role in ROS signaling involved in plant growth, stress tolerance, and plant immunity (Marino et al., 2012).

One unique type of the plant ER-PM contact occurs at plasmodesmata (PD), which consist of the cylindrically apposed PM and the tightly compressed ER (desmotubule) with unique lipid/protein compositions (Grison et al., 2015; Leijon et al., 2018). The PD-PM and the desmotubule are connected by spokelike elements (Ding et al., 1992; Nicolas et al., 2017) whose molecular identities remain to be defined, but recent studies suggested the PD association of AtSYT1 (Levy et al., 2015) and VAP27 (Wang et al., 2016). The space between the PD-PM and the desmotubule constitutes the actual channel (the cytoplasmic sleeve) that transports a wide range of molecular cargos across cell walls of neighboring cells (Tilsner et al., 2016). Given the key role of PD in generating cytosolic and membrane continuity that are essential for growth and development, stress tolerance, and plant defense, the permeability of PD (also known as size exclusion limit), governed by the size of the cytoplasmic sleeve and distribution of spokes that creates nanochannels, is constantly regulated by various of developmental and environmental signals (Sun et al., 2019). Although PD exhibits the essential features of MCS (Scorrano et al., 2019), it remains to be investigated if the ER-PM contacts in PD play any role in inter-organelle exchanges of lipids, Ca2+, and/or other signaling molecules.

#### THE ENDOPLASMIC RETICULUM-CHLOROPLAST JUNCTION

Chloroplasts conduct photosynthesis and produce energy for plant growth, development, and defense. In addition, chloroplasts are essential for synthesizing certain amino acids, lipids, and fatty acids. Like mitochondrion, chloroplast is also a semiautonomous organelle with its own genome and a majority of chloroplast proteins are encoded by the nuclear genome and imported from the cytosol. Accordingly, the plant cells execute anterograde and retrograde communications between the chloroplast and the nucleus to respond to changing environment (Watson et al., 2018). Under stress conditions, ROS such as singlet oxygen and superoxide were generated from electron transport chain in the chloroplasts, which cause oxidative damage to the photosynthetic organelle. Consequently, the chloroplasts use ROS and several metabolites, such as 3′-phosphoadenosine 5′-phosphate (PAP) (Chan et al., 2016) and methylerythritol cyclodiphosphate (MEcPP) (Xiao et al., 2012), to relay the stress signal into the nucleus to reprogram gene expression for damage mitigation and stress acclimation (Woodson and Chory, 2012). The chloroplast-nucleus signaling might also involve chloroplastnucleus contact sites consisting of stromules, the stroma-filled tubular protrusions from the chloroplast outer membrane (**Figure 1**; Kohler and Hanson, 2000; Hanson and Hines, 2018), which facilitate translocations of chloroplast-sequestered transcription factors into the nucleus in response to various stresses (Caplan et al., 2008; Sun et al., 2011; Foyer et al., 2014). Stromules were also known to be associated with the ER, Golgi apparatus, PM, mitochondria, and peroxisomes (Kwok and Hanson, 2004; Schattat et al., 2011; Hanson and Hines, 2018); however, the physiological significance of these associations remains to be investigated in the coming years.

The ER and chloroplasts are the two major sites of lipid biosynthesis (van Meer et al., 2008; Hurlock et al., 2014) and the ER-chloroplast interaction is essential for lipid homeostasis in plant cells under normal growth condition and in response to various environmental stresses (Negi et al., 2018; Lavell and Benning, 2019). The ER-chloroplast-mediated lipid biosynthesis involving *de novo* synthesis of fatty acids (FAs) in chloroplasts, the chloroplast-ER transport of FAs, the ER-catalyzed assembly and modification of glycerolipids that move back to chloroplasts for producing galactolipids (Benning and Ohta, 2005), the major chloroplast lipids (Dormann and Benning, 2002). Studies in recent years strongly suggest that the chloroplast-ER physical contact sites, better known as plastid-associated membranes [PLAMs, (Andersson et al., 2007)], are directly involved in the lipid exchange (Tan et al., 2011; Block and Jouhet, 2015). At least two groups of proteins were detected at the ER-chloroplast membrane contact sites (Tan et al., 2011). The first group includes several members of the trigalactosyldiacylglycerol (TGD) protein family, which form a bacterial-type ABC transporter for transporting lipids from the ER to the thylakoid membrane (Xu et al., 2010; Wang et al., 2012; Fan et al., 2015). The second group includes lipid processing enzymes such as phosphatidylcholine (PC) synthase and CLIP1 lipase/acylhydrolase that directly act on lipids from the contacting ER-chloroplast membranes (Mehrshahi et al., 2013, 2014). In addition, a recent study indicated the presence of several lipid transfer proteins, including Azelaic Acid Induced 1 (AZI1), EArly *Arabidopsis* Aluminum Induced 1 (EARLI1), and Defective in Induced Resistance 1 (DIR1), at the ER-chloroplast contact site that facilitates the movement of a lipid-derived signal for systemic acquired resistance against pathogens (Cecchini et al., 2015).

Various abiotic stresses, such as high light exposure and wounding, can lead to accumulation of MEcPP in chloroplasts, which serves as a retrograde signaling metabolite that relays the chloroplast stress signal into the nucleus to alter gene expression (Xiao et al., 2012). Intriguingly, the chloroplast-synthesized MEcPP signal could activate the transcription of IRE1 and bZIP60, two key components of the ER stress-triggered UPR pathway *via* a Ca2+ dependent transcription factor calmodulin-binding transcription activator3 (Walley et al., 2015; Benn et al., 2016). In addition, a loss-of-function mutation in an *Arabidopsis* gene encoding the chloroplast stearoyl-acyl carrier protein desaturase, which introduces double bonds into FAs, constitutively activates the expression of a known ER-UPR marker gene *BIP3* (Iwata et al., 2018). A loss-of-function mutation in the *Arabidopsis SAL1* gene, which encodes a chloroplast/ mitochondria-localized bifunctional enzyme with both 3′(2′),5′-bisphosphate nucleotidase (converting PAP to AMP) and inositol polyphosphate 1-phosphatase activities, attenuated ER stress response and exhibited hyposensitivity to ER stress inducers (Xi et al., 2016). Together, these findings provide additional support for the involvement of the photosynthetic organelle in regulating the ER homeostasis.

#### THE ENDOPLASMIC RETICULUM-PEROXISOME COLLABORATION

Peroxisome is a semiautonomous single-membrane-bound organelle that participates in a wide range of biochemical processes, particularly the β-oxidation of fatty acids and metabolism of hydrogen peroxide (Smith and Aitchison, 2013). In plants, peroxisomes also perform other important functions such as the glycolate cycle and photorespiration, secondary metabolism, hormone (auxin and jasmonic acid) biosynthesis, metabolism of ROS and reactive nitrogen species (RNS) (Nyathi and Baker, 2006; Hu et al., 2012; Sandalio and Romero-Puertas, 2015). Notably, peroxisomes are highly dynamic organelles that alter their morphology, proliferation, and metabolic activities in response to environmental signals (Honsho et al., 2016; Kao et al., 2018). The membrane extensions of peroxisomes, termed as peroxules (**Figure 1**), are often observed when plants are exposed to exogenous H2O2 or high-intensity light (Sinclair et al., 2009; Barton et al., 2013; Jaipargas et al., 2016). Salt stress, heavy metals, and herbicide application were known to increase the metabolic activity and proliferation rate of peroxisomes (Palma et al., 1987; McCarthy et al., 2001; Mitsuya et al., 2010; McCarthy-Suárez et al., 2011; Fahy et al., 2017).

It has been well known that peroxisome dynamics such as elongation, fission, and degradation as well as metabolic changes require their constant collaborations and communications with other intracellular organelles (Hu et al., 2012; Del Rio and Lopez-Huertas, 2016; Kao et al., 2018). The ER-peroxisome connection has been known for many years as peroxisomes are formed by budding from specialized ER regions and/or by growth and fission of preexisting peroxisomes in yeast and mammalian cells (Hu et al., 2012; Kao et al., 2018). Although there is no clear evidence to support the ER budding model for the plant peroxisomes (Mullen and Trelease, 2006; Trelease and Lingard, 2006), the ER is at least involved in the plant peroxisome biogenesis by providing membranes, lipids, and certain peroxisome membrane proteins (PMPs) to preexisting or fission-created nascent peroxisomes (Hu et al., 2012).

The plant peroxisomes were shown to be closely associated with the ER by early microscopic observation (Huang et al., 1983) and could be physically attached to the ER as suggested by live cell imaging of dynamic behaviors of peroxisomes (and peroxules) and the ER in *Arabidopsis* (Mathur, 2009; Sinclair et al., 2009; Barton et al., 2013). However, it remains unknown whether the observed ER-peroxisome contiguity in *Arabidopsis* is mediated by the peroxisome-ER physical tether that was first described in yeast. The yeast peroxisome-ER tethering complex consists of a peroxisome biogenic protein, peroxin 3 (PEX3), localized on the ER and peroxisome, and the peroxisome inheritance factor Inp1 that serves as a bridge to link the ER and peroxisomelocalized PEX3 (Knoblach and Rachubinski, 2013). The PEX3- Inp1-PEX3 trimeric complex plays a key role in partitioning peroxisomes in dividing yeast cells and controlling the peroxisome population (Knoblach et al., 2013). The mammalian peroxisome-ER tether consists of the ER-localized VAPs and the PMPs with acyl-CoA binding domains (ACBDs) and is thought to regulate peroxisome proliferation and to facilitate the ER-peroxisome lipid exchange (Hua et al., 2017; Costello et al., 2017a,b). Despite microscopic observations of the ER-peroxule association (Sinclair et al., 2009; Barton et al., 2013), a plant peroxisome-ER tethering complex remains to be discovered. The identification of a peroxisome-ER tether is expected to shed light on the functional collaboration between the two dynamic organelles, especially the mechanisms of peroxisome biogenesis/maintenance and their dynamic responses to various environmental stresses.

It was recently suggested that peroxisomes, ER, and mitochondria could form a "redox triangle" that uses tethering complexes to assemble a hypothetical "redoxosome" that transmits intercompartmental redox signals to regulate ROS metabolism in response to cellular signals and environmental cues (Yoboue et al., 2018). A plant "redoxosome" should include protein tethering complexes of chloroplasts with the ER, mitochondria, and peroxisome. The chloroplast works together with mitochondria and peroxisomes in photorespiration involving inter-organellar metabolite exchanges while the chloroplast tubular extensions, stromules, are thought to interact with the ER, mitochondria, and peroxisomes (Mathur et al., 2012; Hanson and Hines, 2018). Fluorescent microscopic studies and proteomic experiments with a plant genetic model system such as *Arabidopsis* could make a significant contribution to our understanding of such a "redoxosome" in plants. Dynamic physical associations of multiple organelles aided by organelle extensions and tethering complexes might be a common cellular mechanism that facilitates exchanges of ROS/RNS, Ca2+, lipids, and other metabolites/ signaling molecules to mount coordinated cellular responses to changing environment.

#### THE ENDOPLASMIC RETICULUM-VACUOLE ASSOCIATION

Vacuoles are single-membrane-bound organelles that are filled with a wide range of inorganic ions and organic molecules (**Figure 1**). In plants, at least two types of vacuoles have been identified, including protein storage vacuoles (PSVs) and lytic vacuoles (LVs) (Paris et al., 1996; Zhang et al., 2014). PSVs usually serve as a warehouse for seed storage proteins that are synthesized in the ER during seed maturation, while LVs occur in the vegetative tissues and contain acidic contents and degradative enzymes with lysosome-like properties (Shimada et al., 2018). It has been shown that the vacuoles play crucial roles in storage of nutrients and metabolites, detoxification, pH homeostasis, and stress tolerance (Muntz, 2007; Viotti, 2014). Maintaining proper turgor pressure in vacuoles is required for morphological alterations of cells during plant development, and the rapid vacuolar uptake or unloading of various ions and metabolites allows plants to efficiently cope with environmental stresses. For instance, AtNHX1 is an *Arabidopsis* tonoplast-localized Na+ /H+ antiporter that moves excessive Na+ from the cytosol into the vacuole, lowering the water potential of the vacuole and driving water flow into the cells to maintain plants' growth under high salinity condition (Apse et al., 1999). It is well known that stomatal opening or closure is associated with vacuole morphology changes in guard cells, highlighting the important roles of vacuole in plant response to abiotic stresses, such as high temperature and drought (Gao et al., 2005; Tanaka et al., 2007; Bak et al., 2013).

Many vacuolar proteins and metabolites are synthesized and processed in the ER and transported to the vacuoles. One well-established pathway for vacuolar transport is the COPIImediated vesicle trafficking from the ER to the Golgi and the post-Golgi transport that involves the plant TGN and the pre-vacuolar compartment (PVC, also known as MVB for multi-vesicular body) (Xiang et al., 2013; Brillada and Rojas-Pierce, 2017). Recent studies indicated the presence of a direct Golgi-independent ER-vacuole trafficking route involving the machinery of autophagy (Viotti et al., 2013; Michaeli et al., 2014), which degrades and recycles damaged/misfolded/ aggregated proteins and defective/excessive intracellular organelles (Wang et al., 2018b). More importantly, autophagy is an integral part of the ER stress-triggered UPR. Under the ER stress, ER components bud from the ER and form autophagosome with the aid of appropriate cargo receptors, and the autophagosome subsequently fuses with the lytic vacuole to release the ER cargos for degradation *via* the classical macroautophagy pathway (Liu et al., 2012; Michaeli et al., 2014; Yang et al., 2016). A special process of autophagy, ER-phagy (Schuck et al., 2014) or reticulophagy (Liu et al., 2012), is activated to degrade damaged ER fragments when UPR fails to mitigate the ER stress. Further studies revealed that the ER stress-induced reticulophagy in *Arabidopsis* requires one of the ER-localized UPR sensor IRE1b but not bZIP60 (Liu et al., 2012).

Given the presence of a direct ER-vacuole trafficking route for transporting metabolites, proteins, and membranes in plant cells, it is quite possible that plant cells have multiple ER-vacuole contact sites that serve important cellular functions, especially when responding to environmental stresses. In yeast, the ER-vacuole contact site (**Figure 1**) [known as nuclear ER-vacuole junctions or NVJs (Pan et al., 2000)] has been well studied and is implicated in the biogenesis and transport of lipid droplets in response to metabolic stress (Hariri et al., 2018). The yeast NVJ is established by interaction between one of the two ER membrane proteins, Nvj1 and Ltc1 (lipid transfer at contact site1), and an armadillo repeat protein Vac8 that requires palmitoylation for its localization to the vacuolar membrane (Pan et al., 2000; Murley et al., 2015). The yeast NVJ tether also contains Nvj2, one of the seven SPM domaincontaining proteins that are localized at MCSs, including three at ERMES and the remaining three (tricalbins) at EPCSs (Toulmay and Prinz, 2012). Despite essential roles of the vacuoles in plant growth, stress tolerance, and plant defense (Shimada et al., 2018), little is known about the plant ER-vacuole contact sites and their associated tethering complexes. *Arabidopsis* lacks a homolog of Nvj1 or Ltc1 but contains >100 armadillo repeat proteins (Sharma et al., 2014) and five tricalbin homologs known as AtSYTA-E or AtSYT1–5 (Craxton, 2004). Live cell imaging of fluorescently tagged ER/tonoplast-localized proteins coupled with optical tweezers (Sparkes, 2018) could reveal potential ER-vacuole contact sites and their dynamic changes in response to environmental stresses. Given the widespread occurrence of SMP-containing proteins at multiple MCSs in yeast and mammalian cells (Toulmay and Prinz, 2012), identification of a plant ER-vacuole tethering complex might be facilitated by confocal microscopic examination of fluorescently tagged AtSYT1–5 followed by biochemical studies of an AtSYT localized at the ER-vacuole contact sites.

#### CONCLUSION

Accumulating evidence supports important roles of the ER-organelle interactions in plant stress tolerance, which involves exchanges of metabolites and signaling molecules at specialized MCSs with unique tethering complexes. Further studies that combine live cell imaging, proteomics, and plant genetics are needed to fully understand the composition and dynamic regulation of these MCSs in response to environmental changes and their additional physiological functions.

#### AUTHOR CONTRIBUTIONS

LL and JL discussed the writing plan, LL drafted the manuscript, and JL edited the manuscript.

#### FUNDING

This work was partially supported by grants from National Natural Science Foundation of China (NSFC31600996 to LL and NSFC31730019 to JL). The open access publication fee

#### REFERENCES


is provided by a startup fund from South China Agricultural University.

#### ACKNOWLEDGMENTS

We would like to thank Fen Su for her help in generating the figure and the handling editor and two reviewers for their constructive criticisms and helpful suggestions. We apologize to those colleagues whose works were not fully cited in this article.


phosphatase to regulate phosphoinositides. *J. Cell Biol.* 213, 33–48. doi: 10.1083/jcb.201508106


response to hydroxyl stress. *Plant J.* 59, 231–242. doi: 10.1111/j.1365- 313X.2009.03863.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 © 2019 Liu and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(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.*

# ABA Alleviates Uptake and Accumulation of Zinc in Grapevine (Vitis vinifera L.) by Inducing Expression of ZIP and Detoxification-Related Genes

#### Changzheng Song1,2, Yifan Yan<sup>2</sup> , Abel Rosado<sup>3</sup> , Zhenwen Zhang<sup>1</sup> \* and Simone Diego Castellarin<sup>2</sup> \*

#### Edited by:

Rosa M. Rivero, Spanish National Research Council (CSIC), Spain

#### Reviewed by:

Ruben Bottini, CONICET Mendoza, Argentina Sergio Ruffo Roberto, State University of Londrina, Brazil Lingfei Shangguan, Nanjing Agricultural University, China Rongrong Guo, Guangxi Academy of Agricultural Sciences, China

#### \*Correspondence:

Zhenwen Zhang zhangzhw60@nwsuaf.edu.cn Simone Diego Castellarin simone.castellarin@ubc.ca; sdcastel@mail.ubc.ca

#### Specialty section:

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

Received: 10 January 2019 Accepted: 18 June 2019 Published: 05 July 2019

#### Citation:

Song C, Yan Y, Rosado A, Zhang Z and Castellarin SD (2019) ABA Alleviates Uptake and Accumulation of Zinc in Grapevine (Vitis vinifera L.) by Inducing Expression of ZIP and Detoxification-Related Genes. Front. Plant Sci. 10:872. doi: 10.3389/fpls.2019.00872 <sup>1</sup> Shaanxi Engineering Research Center for Viti-Viniculture, College of Enology, Northwest A&F University, Yangling, China, <sup>2</sup> Wine Research Centre, The University of British Columbia, Vancouver, BC, Canada, <sup>3</sup> Department of Botany, The University of British Columbia, Vancouver, BC, Canada

Abscisic acid (ABA) is a plant hormone that can mitigate heavy metal toxicity. Exogenous ABA and ABA mimic 1 (AM1) were applied to study the influence on Zn uptake and accumulation in Vitis vinifera L. cv. Merlot seedlings exposed to excess Zn. The seedlings were treated with either normal or excess levels of Zn in combination with applications of ABA and AM1. Excess Zn exposure resulted in decreased lateral root length, decreased photosynthesis, elevated uptake, and accumulation of Zn in roots, trunks, and stems, decreased jasmonic acid content in roots and leaves, and induced the expression of Zn transportation- and detoxification-related genes. Remarkably, in the presence of toxic amounts of Zn, the exogenous application of ABA, but not of AM1, reduced the uptake and accumulation of Zn in roots and induced higher expression of both ZIP genes and detoxification-related genes in root and leaf. These results indicate that exogenous ABA enhances the tolerance of grape seedlings to excess Zn and that AM1 is not a suitable ABA mimic compound for Zn stress alleviation in grapes.

#### Keywords: abscisic acid, ABA mimic, excess Zn stress, grape, heavy metal toxicity

# INTRODUCTION

Wine grapes are produced in more than 40 countries. Climates and soils of different regions contribute to the diversity of quality and type of the grape and wines (Anderson and Aryal, 2015). In the last decades, wine grape cultivation has been rapidly developed in tropical and subtropical areas, including Mexico, Venezuela, Peru, Brazil, India, Thailand, and South China (Intarapichet et al., 2007; Jogaiah et al., 2013; Terra et al., 2013; dos Santos Lima et al., 2014), driven by local wine market demand (Anderson and Aryal, 2015; Anderson and Nelgen, 2015). However, soils in these areas are often susceptible to heavy metal toxicity because of low pH and pollution.

Zinc (Zn) is an essential micronutrient for plants. It plays a key role in photosynthetic redox reactions, and it is an essential cofactor for many enzymes involved in nitrogen metabolism and protein synthesis (Hafeez et al., 2013). Similar to other plant micronutrients, Zn is beneficial in a narrow range of concentrations, and its bioavailability in soils increases at low pH

(Alloway, 2012). In unpolluted soils, the amount of Zn is generally below 125 ppm (Baccio et al., 2003; Hussain et al., 2010). However, environmental pollution due to industrial and agricultural activities including excessive application of Zncontaining fertilizers and pesticides, manures, sewage sludge, smelters, incinerators, mines, and galvanized products has increased the concentration of Zn in many agricultural soils above the threshold of phytotoxicity (Robson, 1993a).

According to the no observed effect concentrations (NOECs), the highest Zn dose that can be added to soils without affecting plants is 32–400 mg/kg (fresh weight). Toxic effects are identified at total Zn concentrations of 100 to >1000 mg/kg (Alloway, 2012), and a tissue concentration ≥400 mg/kg (dry weight) of Zn is considered toxic for nearly all plants (Sofo et al., 2013). Generally, Zn phytotoxicity inhibits the growth of roots and stems, modifies leaf morphology, induces chlorosis, reduces photosynthesis, interferes with the uptake of other nutrients, induces stress phytohormones, and alters the expression of genes related to Zn accumulation and detoxification (Rout and Das, 2009).

Abscisic acid (ABA) is a plant hormone with important functions as a stress alleviator, particularly in responses to drought, salt, and chilling stress (Bari and Jones, 2009). ABA levels in plants are tightly controlled by environmental conditions, and high ABA concentration activates signaling cascades of other phytohormones, such as salicylic acid (SA) and jasmonic acid (JA) (Shi et al., 2015). Exogenous ABA application causes an alleviating effect on plants under heavy metal stresses (Hsu and Kao, 2003; Wang et al., 2013), and a recent study by Shi et al. (2015) suggests that exogenous ABA applications can decrease the phytotoxic effect of Zn in Populus × canescens tissues by modulating the transcriptional activity of key genes involved in Zn transport and detoxification, and by activating the antioxidative defense system.

Since the application of ABA in agricultural practice is limited by ABA's chemical instability, costly production, and rapid catabolism, a small molecule, ABA mimic (AM1), has been recently identified as an ABA surrogate based on its structural analogy to ABA (Cao et al., 2013). Similar to ABA, AM1 activates multiple members of the ABA receptor family, such as pyrabactin resistance 1 (PYR1) and PYR1-like (PYL) protein, and enhances the tolerance of plants to drought and cold stress (Cao et al., 2013; Cheng et al., 2016). However, compared with ABA, AM1 is easier to synthesize and more resistant to photolysis. Therefore, it has the potential to become an ABA replacement in agricultural practice.

In the present study, the effects of ABA and AM1 in heavy metal stress alleviation were tested by studying the uptake and translocation of Zn in "Merlot" grapevines grown under excess Zn stress. Leaf area, photosynthesis and foliar pigments, Zn localization and concentration, phytohormone level, and expression of Zn-related genes were measured to elucidate the physiological and molecular response underlying the potential mitigating effects of ABA on Zn uptake and on physiology in grapevine, and to explore strategies to mitigate Zn phytotoxicity in vineyards.

# MATERIALS AND METHODS

# Plant Cultivation and Treatments

One-year-old hardwood cuttings of "Merlot" (V. vinifera L.) with 4–6 nodes were pre-rooted in a thermostatically controlled heated container (26◦C at the base of the cuttings) in a cold room (4◦C) for 40 days. The cuttings were then transferred to pots and cultivated for 5 weeks in the Horticulture Greenhouse at The University of British Columbia (26◦C day and 20◦C night, 16 h photoperiod). Afterward, the rooted seedlings were transferred into 4 L plastic pots filled with clean sand and cultivated for 10 more weeks. The plants were irrigated with 50 mL distilled water or Hoagland solution alternately at each sunset. Thirty-two plants with similar heights were randomly divided into four groups and treated with either basal (0.765 µM Zn2+, 0.22 mg/L) or excess levels (10 mM Zn2+, 2880 mg/L) of zinc sulfate (ZnSO4.7H2O) dissolved in aqueous solution. To 2 of 3 excess Zn treatments, 10 µM ABA or 10 µM AM1 solutions were applied to the roots. This resulted in four treatments: Basal Zn, Excess Zn, Excess Zn + ABA, and Excess Zn + AM1. The treatments were applied in combination with the Hoagland solution every day for 10 days. Eight grapevines per treatment were considered. Four plants per treatment were harvested on the 4th day after the treatment began (DAT); the rest of the plants were harvested on the 10th DAT. Each plant was regarded as a biological replicate, so four biological replicates were included in each treatment.

# Leaf Area, Root Length, and Gas Exchange Measurement

For each plant, the lengths of the shoot and of the main vein of each leaf were measured using a ruler at 0, 4, 10 DAT. A regression line between the length of the main vein and the total leaf area was calculated by measuring 50 leaves of different sizes using a ruler and leaf area meter (LI-3100, LI-COR, NE, United States) (Sivilotti et al., 2016). The regression was used to estimate the leaf area of each leaf of the plants in a non-destructive manner, and the total leaf area per plant was calculated by summing the area of each leaf of the plant. The length of lateral roots was also measured at 10 DAT.

Before each sampling point, the leaf gas exchange was determined for each plant. Measurements were conducted from 9:00 am to 1:00 pm. Mature leaves with plastochron 7–9 (Lamoreaux et al., 1978) were selected for the measurement of the net photosynthetic rate (A), the stomatal conductance (gs), and the transpiration rate (E) using a Li-Cor-6400 portable photosynthesis system (Li-COR Inc., NE, United States). The light source from the lamp was set at 1500 µmol/m<sup>2</sup> /s, air flow through the sample chamber was 500 µmol/s, reference cell CO<sup>2</sup> concentration was 400 µmol/mol, and the leaf temperature was 22.0◦C.

# Sampling of Plant Material

Lateral roots, trunks, and stems between the 5th and 9th nodes from the shoot tip, and leaves between the 1st and 10th node, were harvested at 4 and 10 DAT. Subsamples of fine fresh root, trunk, stem, and leaf were harvested for histochemical

analysis. The rest of the samples were frozen in liquid nitrogen and ground into powder using RNase-free mortars and pestles, and then stored at −80◦C. This plant material was used for Zn and hormone determination and gene expression analysis. Additional leaves with plastochron 7–9 were collected and used for pigment determination.

#### Analysis of Foliar Pigment

The leaves were cleaned with distilled water after collection. The veins were removed, and an aliquot of 0.5 g of leaf sample was added into a mortar and ground with 80% acetone. The homogenate was washed out with 80% acetone, filtered into a 10 mL volumetric flask, and then filled to volume. The concentrations of chlorophylls were determined spectrophotometrically as previously described (Wellburn, 1994).

#### Determination of Zn Concentration

Powder samples, approximately 5 g each, were dried at 60◦C for 72 h to calculate the fresh to dry mass ratio. Afterward, 1 g of dried sample was weighed into a tared, oven-dried crucible. The samples were ashed in the muffle furnace for 1 h at 300◦C and 4 h at 500◦C. After the samples were cooled in a desiccator, 5 mL of 2 M HCl was carefully added. A warm sand bath was used to dissolve soluble constituents, and a tared Waterman #42 paper was used to filter the solution into a 100 mL volumetric flask. Distilled water was used to wash the filter paper and make up to 100 mL of solution. Subsequently, the concentrations of Zn and other mineral elements were determined using a Nu AttoM Inductively Coupled Plasma Mass Spectrometer (ICP-MS) (CAMECA, Gennevilliers, France) as described by Murray et al. (2000). For Zn levels in leaf, the concentrations were measured in the petiole, which has been widely used for diagnosis of Zn in previous research (Alloway, 2004; Romero et al., 2013).

#### Analysis of Zn Localization

Zn accumulation was histochemically detected in root, trunk, stem, and leaf tissues using the Zn chelating agent dithizone (diphenylthiocarbazone, 30 mg dissolved in 60 ml acetone, 20 ml distilled water and four drops of glacial acetic acid). Hand sections of fresh samples were stained for 2 min and rinsed several times with deionized H2O. Afterward, the sections were immediately analyzed with an Olympus AX-70 light microscope (Olympus Corporation, Tokyo, Japan). The red– purple Zn dithizonate complex was photographed under the light microscope at 10× or 20× magnification using an Olympus Fluoview 1000 scan head connected to a computer (Olympus Corporation, Tokyo, Japan).

#### Determination of Phytohormone Contents Chemical Standards

ABA (cat. no. 013 2701) was purchased from OlChemIm Ltd. (Olomouc, Czechia), SA (cat. no. S5922) from Sigma-Aldrich (ON, Canada), and JA (cat. no. 88300) from Cayman Chemical (MI, United States). d6-ABA (cat. no. A110002) was purchased from Toronto Research Chemicals (ON, Canada), d4-SA (cat. no. CS-O-06948) from Clearsynth (ON, Canada), and d5-JA (cat. no. D-6936) from C/D/N Isotopes Inc. (QC, Canada).

#### Phytohormone Extraction

ABA, SA, and JA were extracted according to the method published by Shi et al. (2015) with minor modifications. An aliquot of 0.4 g of fresh tissue was extracted by 5 mL of 80% (v/v with water) methanol containing 200 mg/L of butylated hydroxytoluene and 500 mg/L of citric acid monohydrate. Aliquots of 100 ng of d6-ABA, d4-SA, and d5-JA each were added to the extraction buffer as internal standards (IS). The samples were shaken for 24 h at 4◦C and subsequently centrifuged for 15 min at 10,000 × g under 4◦C. The supernatant was collected and transferred into a 20 mL flat-bottom vial using a syringe (Luer-Lok Tip Syringe, Sigma-Aldrich, ON, Canada) and filter (0.22 µm × 13 mm, PVDF Millex Filter, Sigma-Aldrich, ON, Canada). Afterward, the supernatant was concentrated by a Thermo Savant's Universal Vacuum System (UVS400) (Thermo Fisher Scientific, Waltham, MA, United States) and re-suspended in 250 µL of 80% methanol.

#### Identification and Quantification of Phytohormones

For the identification and quantification of ABA, SA, JA, and AM1, 5 µL of extract was injected into an Agilent 1100 Series high performance liquid chromatograph (HPLC) coupled to an LC/MSD Trap XCT Plus mass selective detector. Chromatography separation was carried out by an Agilent ZORBAX SB-C18 Column (1.8 µm, 4.6 mm × 50 mm). Mass spectrometry data were generated via electrospray ionization (ESI) in negative modes. The temperature of the column was maintained at 60◦C. Mobile phases consisted of Solvent A and B. Solvent A was distilled water with 0.2% formic acid; solvent B was acetonitrile with 0.2% formic acid. The LC separation used a binary solvent gradient with a flow rate of 1.00 mL/min. The gradient conditions were 1.00 min, 10.0% solvent B; 5.00 min, 90.0% solvent B; 6.00 min, 90.0% solvent B; 6.10 min; 5.0% solvent B. Extracted ion chromatography was used for ABA, SA, and JA quantification. Specifically, [MS-H+] 263 was used for ABA, [MS-H+] 269 was used for d6-ABA, [MS-H+] 137 was used for SA, [MS-H+] 141 was used for d4-SA, [MS-H+] 209 was used for JA, [MS-H+] 214 was used for d5-JA. Hormone concentrations were calculated based on internal standard-based calibration curves (**Supplementary Table S1**) that were prepared for each analyte/IS pair, as described elsewhere (Ross et al., 2004). Samples were run in random order.

#### RNA Extraction and Analysis of Transcriptional Level by Real-Time PCR

Total RNA extraction and quantitative real-time PCR (qRT-PCR) were performed as described by Savoi et al. (2016). Total RNA was extracted with the "Spectrum Plant total RNA" kit (Sigma-Aldrich, ON, Canada) from 0.1 g of sample powder. The quantity and quality of the RNA were analyzed with an ND-1000 Spectrophotometer NanoDrop (Thermo Fisher Scientific, Wilmington, DE, United States). All RNA samples were digested to remove genomic DNA and reverse-transcribed in a 20 mL

reaction mixture for cDNA synthesis using a DNase I, RNasefree, and RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, MA, United States), respectively, following the manufacturer's manuals. The expression levels of 10 Zn-related genes, whose functions will be introduced in discussion, VviZIP2, VviZIP6, VviZIP7, VviZIP13, VviHMA2, VviNAS, VviNRAMP3, VviYSL1, VviPCR2, and VvibZIP23, were determined in root and leaf tissues by using an Applied BiosystemsTM 7500 Fast Dx Real-Time PCR Instrument (Thermo Fisher Scientific, MA, United States) and a PowerUP SYBR Green Master Mix (Thermo Fisher Scientific, TX, United States). The genes were selected according to whole-genome array data organized from publicly available RNA-sequencing experiments in our previous publication (Wong et al., 2018). The primers were designed with PrimerQuest Tool (Integrated DNA Technologies, IA, United States) (**Supplementary Table S2**), and the annealing temperature was 60◦C for all primer pairs except VviUbiquitin, which annealed at 56◦C.

#### Statistical Analysis

Mean and standard deviation of values were calculated using Excel 2016 (Microsoft, Washington, DC, United States). Oneway analysis of variance (ANOVA) and two-way ANOVA were conducted using SPSS Statistics 20.0 (IBM, NY, United States). For experimental variables, the principal factors in two-way ANOVA were treatment and time of sampling. Statistical significance among averages was evaluated using the LSD's test with P < 0.05. Figures were made using the drawing software Origin 9.0 (OriginLab, MA, United States).

# RESULTS

#### Leaf Area Index and Root Length

Shoot length and leaf area were measured at 0, 4, and 10 DAT. No difference of shoot length or leaf area among treatments was observed (**Table 1**). However, the Excess Zn and Excess Zn + ABA treatments promoted the increase (from 0 DAT to 4 and 10 DAT) of leaf area. As for the roots at 10 DAT, the average length in all the excess Zn treatments decreased compared with the Basal Zn treatment, and no alleviating effect was found for the ABA or AM1 addition treatments.

#### Photosynthesis and Foliar Pigments

Leaf gas exchanges were affected by Zn treatments, but a significant Zn treatment × time of sampling interaction was also observed (**Table 2**). The CO<sup>2</sup> assimilation rate (A) in the three excess Zn treatments was lower than that in the Basal Zn treatment at both 4 and 10 DAT. As for the stomatal conductance (gs) and the transpiration rate (E), the Excess Zn and Excess Zn + ABA treatments had higher gs than Basal Zn at 10 DAT, and the Excess Zn treatment had higher E than the Basal Zn treatment at 10 DAT. The foliar pigments were not affected by treatments.

#### Zn Concentrations and Localization

Zn concentrations in tissues of "Merlot" seedlings were measured at 4 and 10 DAT (**Figure 1**). Zn treatments affected Zn concentrations in roots, trunks, and stems, but not in leaves. Major effects were observed at 10 DAT. Overall, excess Zn applications increased the Zn concentration in root, trunk, and stem tissues. In the roots, excess Zn exposure led to an increase of Zn accumulation at 4 DAT, and AM1 addition further promoted Zn uptake. At 10 DAT, the Excess Zn and Excess Zn + AM1 treatments had a sharp increase of Zn concentration; the increase was much less in the Excess Zn + ABA treatment, indicating an alleviating effect of ABA on excess Zn accumulation. In the trunk, no effects were observed in the excess Zn treatments at 4 DAT, but increases of Zn levels were observed in the excess Zn treatments at 10 DAT with no mitigating effect of ABA and with a promoting effect of AM1. In the stems, all the excess Zn treatments increased the Zn concentration, and no effect of the time of sampling or of the interaction between the treatments and the time of sampling was observed.

Zn-dithizone complexes were observed at 10 DAT in seedling tissues. As shown in **Supplementary Figure S1**, more Zn accumulated in epidermal cells in roots, pith rays, and phloem in trunks, stems, and petioles in the excess Zn treatments than in the Basal Zn treatment. According to the intensity of red– purple Zn-dithizone precipitates in the tissues, a sharp decline of Zn concentration was observed from lower (roots) to higher (petioles) tissues of the seedlings. Zn accumulation exhibited a similar pattern in the Excess Zn, Excess Zn + ABA, and Excess Zn + AM1 treatments.

#### Phytohormone and AM1 Concentrations

Excess Zn exposure without ABA addition did not lead to changes of ABA in roots and leaves at 4 DAT but slightly increased ABA in roots at 10 DAT (**Figures 2A,B**). Excess Zn + ABA remarkably increased ABA in roots at both 4 DAT and 10 DAT, and increased ABA in leaves at 10 DAT; however, interactions between the treatments and the time of sampling were observed. Excess Zn + AM1 promoted ABA in roots at 4 DAT. No effect of excess Zn was observed on SA in roots or leaves (**Figures 2C,D**). Excess Zn decreased JA in the roots compared with the Basal Zn treatment (**Figure 2E**). The addition of exogenous ABA and AM1 alleviated the effect of excess Zn. No difference of JA in roots was found among the Excess Zn + ABA, Excess Zn + AM1, and Basal Zn treatments. In the leaves at 4 DAT, the Excess Zn treatment slightly reduced JA compared with the Basal Zn treatment, while the addition of exogenous ABA severely decreased JA compared with the Basal Zn and Excess Zn treatments (**Figure 2F**). At 10 DAT, Excess Zn + ABA induced the highest JA concentration among treatments.

AM1 was detected in roots of plants treated with Excess Zn + AM1, at both 4 DAT and 10 DAT (**Supplementary Figure S2**), and the concentration at 10 DAT (283.33 µg/g FW) was higher than that at 4 DAT (63.89 µg/g FW). No AM1 was detected in leaf samples at 4 or 10 DAT.

# Expression of Genes Involved in Zn Uptake and Translocation

To study the effect of excess Zn and ABA or AM1 addition on the expression of Zn uptake- and translocation-related


TABLE 1 | Shoot length, shoot length increment from 0 days after the treatment (DAT), leaf area, leaf area increment, and lateral root length (cm) of "Merlot" (Vitis vinifera L.) seedlings exposed to basal and excess Zn in combination with ABA and AM1 (n = 4).

Two-way ANOVA was performed and the P-values of the effects of Zn treatments (Tr), time of sampling (Ti), and their interaction (TrxTi) are indicated: <sup>∗</sup>P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001; ns, P > 0.05. Different letters indicate different averages based on LSD post hoc tests. #A one-way ANOVA and a LSD post hoc test was run to test the effect of Zn treatments on lateral shoot length at 10 DAT.

TABLE 2 | Leaf photosynthesis (A), stomatal conductance (gs), transpiration rate (E) and photosynthetic pigments at 4 or 10 days after treatment (DAT) (n = 4).


Two ways ANOVA was performed and the P-values of the effects of Zn treatments (Tr), time of sampling (Ti), and their interaction (TrxTi) are indicated: <sup>∗</sup>P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001; ns, P > 0.05. Different letters indicate different averages based on LSD post hoc test.

genes, 10 genes, including the members of the ZIP family of transporters VviZIP2, VviZIP6, VviZIP7, and VviZIP13, heavy metal detoxification-related genes VviHMA2 (Heavy Metal ATPases of the P1B-type ATPases), VviNAS (NA Synthase), VviNRAMP3 (Natural Resistance-Associated Macrophage Protein), VviYSL1 (Yellow Stripe-Like), and VviPCR2 (Plant Cadmium Resistance), as well as the basic-region leucine zipper (bZIP) transcription factor VvibZIP23, were considered. Based on whole-genome array data (**Supplementary Table S3** and **Figure S3**), relative expression of a total of 68 grapevine homologs of the above genes was analyzed across tissues and experimental conditions, and the 10 selected genes were the most expressed among the homologs.

The expression of most genes was notably affected by excess Zn exposure, ABA or AM1 addition, and the sampling time in both roots and leaves (**Figure 3**). In the roots, excess Zn induced higher expression of all the ZIP genes but VviZIP13 at 4 DAT, as well as some of the heavy metal detoxification-related genes, such as VviHMA2, VviNAS, VvibZIP23, and VviPCR2. VviHMA2 and VviNAS were the most affected by the Excess Zn treatment – seven and five times higher than Basal Zn treatment, respectively. At 10 DAT, most of the above genes were similarly expressed in both the Basal Zn and Excess Zn treatments, but

VviZIP2, VviZIP13, and VviHMA2 were lower in expression in the Excess Zn than in the Basal Zn treatment. Excess Zn + ABA generally induced the expression of ZIP genes and detoxificationrelated genes than in Basal Zn and Excess Zn at 4 DAT. The detoxification genes VviHMA2, VviNAS, and VviPCR2 were the most influenced – 10 times more expressed in Excess Zn + ABA than in Basal Zn treatment. As compared with the Excess Zn treatment, the ABA addition led to a long-lasting induced expression of these genes at 10 DAT. The Zn detoxification gene VviYSL1 was the most induced by Excess Zn + ABA at 10 DAT. Expression of other genes, VviZIP6, VviZIP7, VvibZIP23, VviNRAMP3, and VviPCR2 was also upregulated in the Excess Zn + ABA treatment as compared with the Basal Zn and Excess Zn treatments. Excess Zn + AM1 affected the expression of VviZIP2, VviZIP6, VviZIP13, VvibZIP23, VviYSL1, and VviPCR2 at 4 DAT when compared with Basal Zn. The effect of AM1 addition on the expression of most genes was rarely consistent with the effect of ABA addition.

In the leaves, elevated expression of all the ZIP genes, VviNRAM3, and VviPCR2 was found in the Excess Zn treatment at 4 DAT. The Excess Zn + ABA treatment alleviated the increase of expression of VviZIP2, VviZIP7, VViZIP13, VviNRAM3, VvibZIP23, and VviPCR2 observed in Excess Zn at 4 DAT. Moreover, Excess Zn + ABA decreased VviZIP7, VviZIP13, and VviYSL1 expression as compared with the Basal Zn treatment. In the Excess Zn + AM1 treatment, the expression of VviZIP2, VviZIP6, VviNRAM3, VvibZIP23, and VviPCR2 was similar to that in the Excess Zn treatment and higher than that in the Basal Zn treatment at 4 DAT. At 10 DAT, in the Excess Zn treatment, the expression of VviZIP6, VviZIP7, VviZIP13, VviNAS2, and VviYSL1 was higher than in the Basal Zn treatment, while the expression of VviNRAMP3, VvibZIP23, and VviPCR2 was lower than Basal Zn treatment. The expression of most ZIP genes (VviZIP2, VviZIP7, VviZIP13), VviNRAMP3, VvibZIP23, and VviPCR2 in the Excess Zn + ABA treatment was higher than that in the

on LSD post hoc tests.

Basal Zn and Excess Zn treatments. Moreover, in the Excess Zn + AM1 treatment, the expression of most genes except for VviZIP6 was down-regulated compared with the Basal Zn treatment at 10 DAT.

#### DISCUSSION

#### ABA Alleviates Physiological Stress on: "Merlot" Grapevines Responding to Excess Zn

Reduced growth of roots and stunted shoot growth are general physiological responses to Zn phytotoxicity in plants (Sresty and Rao, 1999; Hirt and Shinozaki, 2004; Richard et al., 2011). In this study, we show that grapevines exposed for 10 days to toxic concentrations of Zn display reduced root length but show no growth defects or visible toxic symptoms in shoots. This result is in accordance with White et al. (1979) and suggests that shoots are more tolerant to long-term Zn exposure than roots.

Though excess Zn exposure did not affect the growth or the total chlorophyll content in grapevine leaves, it was associated with the inhibition of the net photosynthetic rate as previously reported in poplar (Shi et al., 2015) and beans (Tsonev and Cebola-Lidon, 2012). Given that the Mg2<sup>+</sup> cofactor can be replaced by Zn2<sup>+</sup> in chlorophylls (Wettstein et al., 1995), we hypothesize that the Zn-mediated inhibition of photosynthesis in grapevine is due to the impaired functioning

of the electron transport within the light harvesting complexes rather than a reduction in the total content of photosynthetic pigments in the antenna.

There is evidence that with excess Zn exposure, higher proportions of the total plant Zn accumulate in the roots, and the excess Zn mainly accumulates in root cortical cell walls or vacuoles (Robson, 1993b; Bringezu et al., 1999). A higher intensity of zinc-dithizone precipitates was consistently observed in root cortical cells, and a larger increase of Zn concentration in roots as compared with other organs analyzed under excess Zn exposure was observed in grapevine seedlings (**Supplementary Figure S1** and **Figure 1**). In the first 4 days of the excess Zn treatments, Zn levels in trunks, stems and leaves generally remained similar to Zn levels in the Basal Zn treatment, indicating that the roots prevented the upper tissues from accumulating potentially toxic Zn levels by accumulating the excess Zn (**Figure 1**). At 10 DAT, the higher Zn concentrations in roots, trunks, and stems under excess Zn treatments showed the extension of Zn toxicity from the roots to the aerial parts of grapevine seedlings. The addition of exogenous ABA mitigated the accumulation of Zn in roots. This result is consistent with those reported for poplar (Shi et al., 2015). However, in our study the difference between Excess Zn and Excess Zn + ABA was not significant for upper organs.

Plant hormones play important roles in diverse biotic and abiotic stress responses (Bari and Jones, 2009). Although the complex network of interactions of hormones and heavy metal toxicity is not fully understood, evidence has shown that ABA, SA, and JA are involved in the detoxification of heavy metals in plants (Hirt and Shinozaki, 2004; Clemens, 2006; Bari and Jones, 2009; Shi et al., 2015). Interestingly, the response of endogenous ABA concentrations to excess Zn stress varied among studies, including decrease, no change, and increase of ABA in the plant tissues (Zengin, 2006; Yang et al., 2011; Sofo et al., 2013; Shi et al., 2015). In the present study, ABA concentration was not affected by excess Zn, except in the root at 10 DAT, where it was higher in the Excess Zn than in the Basal Zn treatment (**Figures 2A,B**). On the other hand, exogenous ABA addition consistently increased ABA in roots and leaves, which coincides with previous research (Hsu and Kao, 2003; Shi et al., 2015). As in previous studies that investigated the mitigating effects of ABA on heavy metal toxicity (Hsu and Kao, 2003, 2005; Shi et al., 2015), exogenous ABA was applied simultaneously with excess Zn, and this continuous feeding of the seedlings with ABA (Excess Zn + ABA) produced a high level of ABA in the tissues considered in comparison with the Basal Zn and Excess Zn treatments. The high ABA levels in the leaves might have affected leaf physiology. However, despite a small but significant reduction in leaf photosynthesis observed in the exposed Zn + ABA treatment when compared with the Excess Zn treatment (**Table 2**), the other leaf parameters assessed were not affected by ABA addition. Transient applications of ABA have also been shown to alleviate lead stress (Zhao et al., 2009; Wang et al., 2013); hence, it would be worth assessing if transient applications would also alleviate Zn toxicity and for how long the alleviating affect would persist in grapevine.

Salicylic acid is also known as one of the key hormones that regulates plant response to biotic and abiotic stress (Maksymiec, 2007; Bari and Jones, 2009), and it plays a key role in the activation of pathogen-induced systemic acquired resistance (SAR) (Kachroo and Robin, 2013). Although there is evidence that an inverse relationship exists between SA-dependent resistance and JA-dependent resistance (Gupta et al., 2000; Turner et al., 2002), the concentration of SA was generally not affected by the treatments while that of JA was (**Figures 2C,E**). It is worth noticing that ABA treatment decreased SA in roots at 4 DAT compared with the Excess Zn treatment, even though the difference was only marginally significant (P = 0.07), indicating that in the grapevine roots exposed to excess Zn, SA signaling could be negatively regulated by ABA as reported for the plant immune response (de Torres Zabala et al., 2009; Kim et al., 2011).

Jasmonic acid plays an important role in plant defense. It is commonly biosynthesized in response to several biotic and abiotic stresses, including toxic action of heavy metals. Many studies have shown that heavy metal stress is closely connected with JA signaling (Xiang and Oliver, 1998; Cobbett, 2000; Maksymiec et al., 2005, Maksymiec, 2007; Bari and Jones, 2009), and it is likely that JA mediates heavy metal-induced gene expression. Additionally, exogenous JA could induce a crosstolerance to several heavy metal stresses (Xiang and Oliver, 1998; Maksymiec and Krupa, 2006). However, a study of the aquatic plant Wolffia arrhiza indicates that exogenous JA acts in a concentration-dependent manner, in which a high concentration of JA enhances heavy metal toxicity (Piotrowska et al., 2009). In this study, the decreased accumulation of JA by excess Zn in roots confirms the role of JA in the adaptation response to heavy metal toxicity (**Figure 2E**). Furthermore, ABA addition inhibited the decrease of JA concentration in the roots. In addition, it has been reported that JA could protect the photosynthetic apparatus against heavy metal stress (Piotrowska et al., 2009). In our study, Excess Zn + ABA promoted a strong increase of JA concentration in the leaves at 10 DAT; however, this did not mitigate the reduction of photosynthesis determined by excess Zn.

#### ABA Induced the Expression Level of Zn-Related Genes in "Merlot" Grapevines Responding to Excess Zn

Zn uptake and transport play pivotal roles in Zn homeostasis in plants, and the Zrt and Irt-like protein (ZIP) family is well characterized for its critical role in Zn uptake, transport, and homeostasis (Astudillo et al., 2013). Our analysis showed that VviZIP2, VviZIP6, VviZIP7, and VviZIP13 homologs are highly expressed in different tissues in grapevine (**Supplementary Figure S3**). The induced expression of VviZIP genes in the Excess Zn treatment supported the fast accumulation of Zn in the roots at 4 DAT. In plants, basic region/leucine zipper motif (bZIP) transcription factors regulate diverse processes including stress signaling (Jakoby et al., 2002) and constitute a group of ABA-response regulators (Hirt and Shinozaki, 2004). bZIP19 and bZIP23 were reported to regulate the uptake of Zn (Assunção et al., 2010). The highly induced expression of VvibZIP23 by ABA addition regardless of treatment time hints

at its function as an ABA-response regulator, and it corresponds with the up-regulated expression of ZIP genes in seedlings of the same treatment. The NRAMP family in plants encompasses metal transporters, and transporting ions into the vacuole is one way of reducing toxic metal levels in the cytosol (Hall, 2002). Besides, the transporter NRAMP3 is localized in the vacuolar membrane, and the elevated expression of NRAMP3 in Thlaspi caerulescens was suggested to be necessary for tolerating high Zn concentration (Oomen et al., 2009). In this study, the expression of VviNRAMP3 was consistently up-regulated in response to the Excess Zn + ABA treatment. However, no induction by excess Zn was found. The detoxification function of VviNRAMP3 might be induced by ABA alone. There are different mechanisms of tolerance to excess Zn, and there are various proteins involved in these mechanisms (Rout and Das, 2009; Sinclair and Krämer, 2012), which include the chelation of metals in the cytosol (NAS and YSL), the transport of Zn2<sup>+</sup> from the cytoplasm into vacuoles (HMA), and the pumping of Zn2<sup>+</sup> out of the cytosol for xylem loading (PCR) (Song et al., 2010; Ueno et al., 2011; Deinlein et al., 2012). The up-regulated expression of these genes by excess Zn in the roots also verified the toxicity of our excess Zn treatments. As for the ABA addition, the stronger and longer-lasting induction of the expression of detoxification genes showed the alleviating effect of exogenous ABA on Zn toxicity in grape seedlings.

Previous research has shown that the activity of antioxidant enzymes highly correlates with Zn levels in plants (Shi et al., 2015), and exogenous ABA causes increased activities of antioxidative enzymes (SOD, APX, CAT, and POD) in plants, relieving the oxidative stress caused by heavy metals (Zhao et al., 2009; Wang et al., 2013). These mechanisms and enzymes might also be involved in the response of grapevines to excess Zn.

# Root Application of AM1 Does Not Alleviate Excess Zn Stress in "Merlot" Grapevines

AM1 was not effective in mitigating Zn uptake. AM1 has been reported as a promising analog of ABA due to its ability to activate ABA receptors. Although, like ABA, AM1 could enhance the tolerance of plants to drought and cold stress, we did find contrasting effects between AM1 and ABA treatments in response to excess Zn in grapevine. Various factors could have led to this inconsistency. First, the exogenous ABA was possibly taken up by the roots and transported into upper organs. On the contrary, seedlings only accumulated AM1 in roots, and there was no evidence of transport of AM1 to other tissues. This possibly contributed to limiting its effects on Zn absorbance and transportation. Second, AM1 is an agonist only for a selective subset of ABA receptors (Cao et al., 2013). Moreover, AM1 is less potent than ABA in binding to certain ABA receptors due to their structural differences. Recently, a series of AM1 fluorine derivatives (AMFs) were verified to be more potent ABA analogs (Cao et al., 2017). Testing the effect of these AMFs on alleviating excess Zn stress in grapevines will provide useful knowledge for their potential application in agriculture.

# CONCLUSION

Excess Zn exposure led to shorter lateral roots, decreased photosynthesis, fast and high uptake and accumulation of Zn in roots, trunks and stems, decreased concentration of the heavy metal-related endogenous hormone JA, and induced expression of Zn transporter- and detoxification-related genes. Exogenous ABA addition mitigated the uptake and accumulation of Zn and led to higher induced expression of both VviZIP genes and detoxification-related genes. These results demonstrate that exogenous ABA could enhance the tolerance of grapevine to excess Zn, likely by regulating the expression of genes involved in Zn uptake and detoxification. As a heavy metal mitigating strategy, treatments with AM1 could not effectively alleviate the accumulation of Zn. Future studies should test if the observed responses are consistent across genotypes and in grafted grapevines.

# AUTHOR CONTRIBUTIONS

CS and YY performed the experiments. ZZ and SC provided all financial support and critical intellectual input in the design of the study. CS completed the manuscript in collaboration with SC and AR. All authors read and approved the final version of the manuscript.

# FUNDING

This work was supported by the China Agriculture Research System for Grape Industry (CARS-29-zp-06) and the National Sciences and Engineering Research Council (NSERC) Discovery Program (10R23082). The China Scholarship Council (CSC) supported the international cooperation of this work.

# ACKNOWLEDGMENTS

The authors thank Dr. Leslie Lavkulich for helping with the Zn analysis as well as the China Scholarship Council (CSC) and the NSERC for financially supporting the project.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00872/ full#supplementary-material

FIGURE S1 | Zn localization in root, trunk, stem, and petiole of "Merlot" (Vitis vinifera L.) seedlings exposed to Basal, Excess Zn, and Excess Zn with ABA and AM1 additions by dithizone staining at 10 days after treatment. Well stained samples with zinc-dithizone precipitates (red–purple), arrows point to Zn-dithizone precipitates.

FIGURE S2 | ABA mimic 1 (AM1) in the root and leaf tissues of "Merlot" (Vitis vinifera L.) seedlings in Excess Zn + AM1 treatment at 4 or 10 days after treatment (n = 4). Different letters indicate significantly different averages based on LSD post hoc tests.

FIGURE S3 | Relative expression of genes (Zrt and Irt-like protein, ZIP; Heavy Metal ATPases of the P1B-type ATPases, HMA; NA Synthase, NAS; Natural Resistance-Associated Macrophage Protein, NRAMP; Yellow Stripe-Like, YSL; Plant Cadmium Resistance, PCR; basic-region leucine zipper transcription factor, bZIP) involved in Zn uptake and transport in different tissues according to whole-genome array data from 14 publicly available experiments. VIT\_03s0017g02170, VviZIP2; VIT\_06s0004g05070, VviZIP6; VIT\_06s0004g06940, VviZIP7; VIT\_19s0015g00190, VviZIP13;

VIT\_11s0103g00370, VviHMA2; VIT\_14s0060g01190, VviNAS2;

#### REFERENCES


VIT\_07s0129g00620, VviNRAMP3; VIT\_13s0158g00380, VvibZIP23; VIT\_02s0025g02500, VviYSL1; VIT\_01s0011g05470, VviPCR2.

TABLE S1 | Calibration curves for quantification of phytohormones.

TABLE S2 | Primer list of tested genes involved in Zn uptake and transport in of "Merlot" (Vitis vinifera L.) seedlings.

TABLE S3 | Relative expression of 68 grapevine homologous genes across tissues and experimental conditions.



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

Copyright © 2019 Song, Yan, Rosado, Zhang and Castellarin. 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.

# OoNAC72, a NAC-Type Oxytropis ochrocephala Transcription Factor, Conferring Enhanced Drought and Salt Stress Tolerance in Arabidopsis

Huirui Guan† , Xin Liu† , Fei Niu, Qianqian Zhao, Na Fan, Duo Cao, Dian Meng, Wei He, Bin Guo, Yahui Wei\* and Yanping Fu\*

Department of Life Science, Key Laboratory of Resource Biology and Biotechnology in Western China, Northwest University, Xi'an, China

#### Edited by:

Motoaki Seki, RIKEN, Japan

#### Reviewed by:

Yong Hwa Cheong, Sunchon National University, South Korea Nobuhiro Suzuki, Sophia University, Japan

#### \*Correspondence:

Yahui Wei weiyahui@nwu.edu.cn Yanping Fu fuyanping@nwu.edu.cn †These authors have contributed equally to this work

#### Specialty section:

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

Received: 06 January 2019 Accepted: 21 June 2019 Published: 11 July 2019

#### Citation:

Guan H, Liu X, Niu F, Zhao Q, Fan N, Cao D, Meng D, He W, Guo B, Wei Y and Fu Y (2019) OoNAC72, a NAC-Type Oxytropis ochrocephala Transcription Factor, Conferring Enhanced Drought and Salt Stress Tolerance in Arabidopsis. Front. Plant Sci. 10:890. doi: 10.3389/fpls.2019.00890 The NAC proteins form one of the largest families of plant-specific transcription factors (TFs) and play essential roles in developmental processes and stress responses. In this study, we characterized a NAC domain transcription factor, OoNAC72, from a legume Oxytropis ochrocephala. OoNAC72 was proved to be localized in the nuclei in tobacco lower epidermal cells and had transcriptional activation activity in yeast, confirming its transcription activity. OoNAC72 expression could be induced by drought, salinity and exogenous abscisic acid (ABA) in O. ochrocephala seedlings. Furthermore, over-expression of OoNAC72 driven by CaMV35S promoter in Arabidopsis resulted in ABA hypersensitivity and enhanced tolerance to drought and salt stresses during seed germination and post-germinative growth periods. In addition, over-expression of OoNAC72 enhanced the expression of stress-responsive genes such as RD29A, RD29B, RD26, LEA14, ANACOR19, ZAT10, PP2CA, and NCED3. These results highlight the important regulatory role of OoNAC72 in multiple abiotic stress tolerance, and may provide an underlying reason for the spread of O. ochrocephala.

Keywords: O. ochrocephala, NAC transcription factor, drought stress, salt stress, ABA hypersensitivity, transgenic Arabidopsis

#### INTRODUCTION

Oxytropis ochrocephala Bunge, one of the toxic Oxytropis locoweeds, distributed widely among Northwest China, where the plant often suffered from stress environment such as drought, high soil salinity and low temperature. In natural grassland plant, O. ochrocephala can rapidly replace local forages grass species because of its unpalatability and strong biotic stress tolerance. Grasslands infested by O. ochrocephala lead to tremendous losses to livestock husbandry, as well as great damage to the grassland ecological equilibrium (Zhao et al., 2013; He et al., 2015). However, existing research on O. ochrocephala mainly focused on surveys, allelopathy and toxicological studies of distribution (Tulsiani et al., 1988; Zhao et al., 2013), and have not yet investigated its resistance mechanism.

In general, plants often experience some harsh environments during growth and development. As a result, plants must respond to those stresses by regulating the resistance-related genes. Transcription factors play an extremely important role in the process of stresses response

by activating or inhibiting the expression of the downstream target genes. In plants, several families of stress-responsive transcription factors have been functionally characterized under stress regulation networks, such as NAC, bZIP, WRKY, MYB/MYC, and AP2/ERF (Hénanff et al., 2013). NAC transcription factors, one of the largest plant-specific transcription factor families, play an important role in plant growth and stress response, and have become a hot spot in the research of gene regulation (Olsen et al., 2005b; Zheng et al., 2009). NAC transcriptional factors are derived from three kinds of genes containing particular domains of NAM (no apical meristem), ATAF (Arabidopsis transcription activation factor) and CUC (cup-shaped cotyledon) (Souer and Al, 1996; Aida et al., 1997).

NAC transcription factors are key regulators of plant resistance to stress by ABA-dependent or ABA-independent pathways (Puranik et al., 2012). Currently, the NAC transcription factor family have been systematically screened and analyzed in various plants, such as Arabidopsis (138), rice (158), wheat (134), canola, cotton, banana, and soybean (Hegedus et al., 2003; Ooka et al., 2003; Yujie et al., 2008; Meng et al., 2009; Tran et al., 2009; Tang et al., 2012; Jia et al., 2014; Jinpu et al., 2014; Tak et al., 2017). In Arabidopsis, Miki et al. (2010) found that RD26/ANAC072 was significantly induced by drought, high salt and ABA, and the rd26 mutant was not sensitive to exogenous ABA, revealing positive regulation by ABA signaling under drought stress. Similar study showed that ANAC096 also exhibits ABA-dependent signaling and regulates the response of transgenic Arabidopsis to osmotic stress (Xu et al., 2013). Moreover, ANAC096 was reported to have a synergistic relationship with ABRE binding factor and increased plant stress resistance (Xu et al., 2013). In rice, the expression of ONAC022 was up-regulated by various stresses (Hong et al., 2016). Transgenic rice plants overexpressing OsNAC5 and OsNAC6 enhanced dehydration, high salinity and disease tolerances (Nakashima et al., 2007; Takasaki et al., 2010; Jeong et al., 2013). The overexpression of OsNAC9 altered root architecture of rice plants, enhancing drought resistance and grain yield under field conditions (Redillas et al., 2012). In wheat, TaNAC29 and TaNAC2 can be up-regulated by different abiotic stresses, and transgenic Arabidopsis plants overexpressing these genes improved salt and drought tolerance (Mao et al., 2012, 2014; Huang et al., 2015; Huang and Wang, 2016). Additionally, in soybean, Tran et al., 2009 screened and cloned 31 soybean NAC genes, and found that nine of them were induced by drought stress. Pinheiro et al. (2009) found that the expression of GmNAC2/3/4 was significantly induced by osmotic pressure, and GmNAC3/4 was simultaneously induced by ABA, JA, and salt. Although a growing number of studies have shown that NAC transcription factors play a critical regulatory role in a variety of stress-responsive signaling pathways in higher plants, the biological function of O. ochrocephala NAC transcription factor is still unknown.

In this study, an abiotic stress-related NAC family gene OoNAC72 from O. ochrocephala were screened and characterized, and then the subcellular localization and transcriptional activation activities of OoNAC72 protein were further verified. The expression patterns of OoNAC72 in response to polyethylene glycol (PEG), salt and exogenous ABA treatments were also determined by the quantitative real-time PCR (qRT-PCR). Moreover, transgenic Arabidopsis plants over-expressing OoNAC72 (OoNAC72-OX) were measured for phenotypic and physiological characteristics under drought and salt stress conditions. Through this study, we aim to gain a more in-depth and comprehensive understanding of the OoNAC72 structure and its function. The results may provide a new insight into the mechanism for the rapid spread of O. ochrocephala.

#### MATERIALS AND METHODS

#### Plant Materials and Growth Conditions

Mature O. ochrocephala seeds were collected from Haiyuan, Ningxia Province (36◦ 290 4900N 105◦ 360 4900E 2171 mH) in July 2014. Seeds after collection were pretreated with 98% H2SO<sup>4</sup> for 6∼9 min, then they were washed in distilled water 4∼6 times and germinated on wet filter papers for 3 days in the dark on petri dishes. For hormone treatments, 3-week-old seedlings grown in a greenhouse were treated by spraying with 100 mM gibberellin (GA), 100 mM ethephon (ETH) and 100 mM abscisic acid (ABA) with equal volume of solution containing only 0.1% ethanol and distilled water as controls. For high salinity and drought treatments, roots of O. ochrocephala seedlings were soaked in 150 mM NaCl and 20% PEG-6000, respectively. O. ochrocephala seedlings treated with various chemicals and stress elicitors along with control plants were sampled at 0, 1, 3, 6, 12, 24, and 48 h post-treatment (hpt). All samples were frozen in liquid nitrogen and stored at −80◦C for RNA extraction. Three independent biological replications were performed for each experiment.

# RNA Extraction and cDNA Synthesis

Extraction of O. ochrocephala total RNA was performed with the Trizol Reagent (TIANGEN, Beijing, China) according to the manufacturer's instructions. Quality and integrity of total RNA was assessed by 1.0% agarose gel electrophoresis. RNA purity and concentration were determined on a NanoDropTM 2000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, United States). In order to perform RT-PCR and qRT-PCR, the first-strand cDNA was synthesized by reverse transcription using 3 µg total RNA in a 10 µl reaction volume according to the manufacturer's instructions using the transcription kit (Thermo Fisher Scientific, Waltham, MA, United States). The cDNA was diluted 10-fold with nuclease-free water for RT-PCR and qRT-PCR.

# Cloning of OoNAC72 and Sequence Analyses

The sequence of OoNAC72 was obtained by our research group from the O. ochrocephala's transcriptome sequencing data (He et al., 2015). Using the specific primers, we amplified the ORF of OoNAC72 (**Supplementary Table S1**). The PCR condition was as follows: 3 min at 95◦C; 34 cycles of 30 s at 95◦C, 30 s at 55◦C, and 30 s at 72◦C; and then 10 min at 72◦C. The resulting PCR products were cloned to the pGEM-T Easy Vector (TaKaRa,

Beijing, China) and sequenced by Sangon Biotech Co., Ltd., (Shanghai, China). Multiple sequence alignment of OoNAC72 with NAC TFs in other species was performed with DNAMAN 8.0. A phylogenetic tree was constructed using a neighbor-joining (NJ) method with 1000 bootstrap replicates in MEGA 5.0.

#### Quantitative RT-PCR Analyses

Expression profiles of OoNAC72 after different treatments were determined by qRT-PCR analyses with a pair of primers amplifying a 101-bp fragment (**Supplementary Table S2**). To ensure gene-specific amplification, the primers were used to amplify the OoNAC72 gene by regular PCR and sequenced. For qRT-PCR analyses, reactions were conducted following the method of Zhuang et al. (2015a). O. ochrocephala Histone H3 (KR733680.1) and Actin101 (KR822225.1) were used as the internal references (He et al., 2015). OoNAC72 expression level was calculated using the relative 2−11Ct method. Three replications were performed for each experiment.

### Sub-Cellular Localization of OoNAC72

The coding sequence without the stop codon of OoNAC72 was transferred into the pCAMBI1302-eGFP vector (Invitrogen, United States) to generate a pCAMBI1302-OoNAC72-eGFP fusion protein using a pair of primers containing Bgl II or Spe I site (**Supplementary Table S3**). The re-combinational construct pCAMBI1302-OoNAC72-eGFP and pCAMBI1302 eGFP (control vector) were infiltrated into the leaves of 6-week-old Nicotiana benthamiana, respectively (Sheludko et al., 2007). After transformation for 36–60 h, the expression location of OoNAC72:eGFP fusion protein was observed using confocal laser scanning microscopy (CLSM, Olympus FV1000, Olympus Optical Company Ltd., Japan). Laser scanning confocal microscope was used to detect the eGFP (excitation: 488 nm, emission: 510 nm) fluorescence signal. eGFP images, The 40,6-diamidino-2-phenylindole (DAPI) (excitation: 405 nm, emission: 461 nm) nuclear stain was used to determine the location of nucleus. Images were acquired with the software FV10-ASW 4.2 Viewer. Each digital image was recorded with the same camera settings and was not further processed.

# Transcriptional Activation Analysis of OoNAC72

For transactivation analysis of OoNAC72 in yeast cell, the yeast strain AH109 (Clontech) was transformed with the appropriate bait vectors. The full-length coding sequence without the stop codon of OoNAC72 was amplified using three pairs of primers with Smal I - and Pst I sites (**Supplementary Table S3**). The PCR products were digested with Smal I and Pst I and then were cloned into the GAL4 binding domain vector pGBKT7 according to the manufacture's protocol (Clontech). Empty pGBKT7 vector was used as a negative control and the pGAL4 vector was used as a positive control. Transformed yeast cells were transformed onto SD medium (SD / -Trp, SD/ -Trp-His-Ade, SD / -Trp-His-Ade / X-α-gal) to compare their survival. Plates were incubated at 28◦C for 3 days before photographing.

### Generation of Transgenic Plants

Although O. ochrocephala is widely distributed as a leguminous plant on the grassland, its current culture system with a complete growth cycle in the laboratory has not been reported, which makes the gene function research become extremely difficult with O. ochrocephala as a host plant. Therefore, OoNAC72 was transferred into the model plant Arabidopsis to reveal its potential biological functions. The full-length coding sequence without the stop codon of OoNAC72 was cloned into pCAMBI1302-OoNAC72-eGFP and transformed into Arabidopsis thaliana Columbia-0 (WT) plants according to the floral dip method using Agrobacterium tumefaciens strain GV3101 (Clough and Bent, 1998). The positive transgenic lines were screened on homomycin (50 mg/L) plates, and further identified by genomic DNA PCR, and the OoNAC72 expression level in leaves of each transgenic line was examined by qRT-PCR. The homozygous lines of T3 generation plants were used for study.

#### Analysis of Stress Tolerance

Two representative transgenic lines overexpressing OoNAC72- OX lines and wild type (WT) Arabidopsis plants were selected for abiotic stress tolerance assays. For the analysis of germination rate, surface-sterilized seeds were sown on 1/2 MS solid medium supplemented with 100 and 150 mM NaCl, 200 and 250 mM mannitol, and 1 and 3 µM ABA, respectively. The seeds were first vernalized at 4◦C for 3 days in dark, and then were incubated at 22◦C with 16 h light / 8 h dark cycle. The germination rates of seeds were calculated when the green cotyledons emerged. For seedling root length experiment, 5 day-old seedlings cultivated on 1/2 MS solid medium were transferred onto 1/2 MS solid medium supplemented with 100 and 150 mM NaCl, 200, 250 mM mannitol or 1, 3 µM ABA for vertical culture (Yuriko et al., 2013). The length of primary roots of each subset were measured after 14 days of the treatments.

For evaluation of stresses tolerance at the vegetative growth stage, 3-week-old WT and OoNAC72-OX lines grown in soil under non-stress conditions were irrigated continuously with 150 mM NaCl for 30 days. For drought tolerance assay, 40 day-old plants were continuously dehydrated until the leaves withered, and then were rehydrated. Dehydration rate of detached leaves and stomatal conductance were determined on the 10th day after the onset of drought stress. To further evaluate the response mechanism of OoNAC72-OX lines to ABA-mediated drought, the leaves of 40-day-old seedlings of the WT and OoNAC72-OX lines were sheared with 10 µM ABA for 3 h (Huang and Wang, 2016; Mao et al., 2016), and then the stomatal conductance was measured under light microscopy.

#### ABA Content Detection

Three-week-old OoNAC72-OX seedlings and WT plants grown in soil were transferred to 1/2 MS liquid medium supplemented with 20% PEG-6000 and incubated at greenhouse

FIGURE 1 | Subcellular localization of the OoNAC72 protein in leaf epidermal cells of Nicotiana benthamiana. eGFP and OoNAC72-eGFP constructs were separately expressed instantly in leaf epidermal cells of Nicotiana benthamiana and observed under a laser scanning confocal microscope after agroinfiltration for 48 h. DAPI images indicate nuclear staining. eGFP images, DAPI stained images, differential interference contrast images (DIC) and merged images were taken. Bars = 20 µm.

FIGURE 2 | Transcriptional activation assay of the OoNAC72 in yeast cell. (A) Schematic Diagram. The numbers on the left indicate the last residues of polypeptides. Vectors pGBKT7 and pGAL4 were used as negative and positive control, respectively. All plasmids were transformed into the yeast strain AH109 (B–D). Transformed yeasts were dripped on the SD/-Trp, SD/-Trp-His-Ade, and SD/-Trp-His-x-gal after being cultured for 3 days in the growth chamber.

for 2 days. Fresh Arabidopsis samples harvested at different time stages during development were immediately frozen for ABA quantification by the ABA immunoassay kit (Yang et al., 2001; Yanjuan et al., 2012).

#### Gene Expression Analysis of Endogenous Genes in Arabidopsis Leaves Under Stresses

To further investigate the molecular mechanism of stress tolerance, the expression levels of marker genes were detected in the WT and OoNAC72-OX plants. Total RNAs of the whole plants from 14-day-old WT and OoNAC72-OX plant seedlings grown on 1/2 MS solid medium supplemented with 150 mM NaCl and 200 mM mannitol were extracted with the Trizol Reagent. The qRT-PCR was performed using specific primers (**Supplementary Table S4**) for the expression levels of marker genes, and Actin8 was employed as a reference control.

#### Measurement of Physiological Changes

To determine water loss rate, leaves were harvested from 4-week-old seedlings of the WT and OoNAC72-OX plants and dehydrated on the dry filter paper (22–25◦C, humidity 45–60%) for weighing at designated time points. Images were captured at 0 and 3 h after the treatment. The water loss rate was calculated based on the initial fresh weight of the leaves. Proline and malondialdehyde (MDA) content, SOD and POD activities were measured according to Wang et al. (2010) and Moore (1998). Proline content was measured according to Bates et al. (1973). At least 20 seedlings were employed for physiological indices analysis in each sample.

#### Statistical Analysis

Data were presented as means ± SD of at least three independent replicates from one representative experiment. Analysis of

significant difference was performed by Duncan's multiple range tests in the ANOVA program of SPSS (IBM SPSS 22), taking <sup>∗</sup>P < 0.05, ∗∗P < 0.01 as critical value.

# RESULTS

#### OoNAC72 Encodes a NAC Domain Protein

A 1301 bp cDNA containing an ORF of 996 bp (65–1060) was cloned from the transcriptome of O. ochrocephala. This ORF coded a protein of 331 amino acids with a theoretical molecular weight of 37.49 kDa. The conserved domain analysis revealed that its N-terminal region had a highly conserved NAC domain (amino acid 98–220), which consisted of five subdomains A–E (**Supplementary Figure S1**). Whereas its C-terminal region had no significant similarity to any other members of the NAC family. BLASTP analysis revealed that this protein shared the highest similarity (70%) to AtNAC72 (XP\_016170502.1) in Arabidopsis. Further phylogenetic analysis confirmed that the relatedness of the predicted protein to AtNAC72 was highly homologous with those of Medicago truncatula and Cicer arietinum (**Supplementary Figure S2**). Therefore, this O. ochrocephala gene was designated as OoNAC72 (MH142381).

#### Sub-Cellular Localization of OoNAC72

The sequence analysis showed that OoNAC72 possessed a conserved nuclear localization signal (NLS, positions 74–88, 117–129 aa) (**Supplementary Figure S1**). Meanwhile, analysis by ProtComp v.9.0 indicated a high likelihood of nuclear localization for OoNAC72 protein. To confirm this prediction, an expression cassette fusing OoNAC72 with the eGFP protein was constructed. Then the fused protein was expressed transiently in Nicotiana benthamiana while the pCAMBI1302 eGFP functioned as a control. Fluorescence microscopy revealed that the OoNAC72-eGFP fusion protein was exclusively localized in the nucleus in the transformed cells, whereas the control eGFP was uniformly distributed throughout the cell (**Figure 1**). These results further confirmed that the OoNAC72 protein was a nuclear-localized protein.

### Transcriptional Activity Assay in Yeast Cells

The result of transcriptional activity analysis of OoNAC72 was illustrated in **Figure 2**. The vector pGBKT7 fused with OoNAC72 was used as the experimental set, and the empty pGBKT7 vector and pGAL4 were hired as the negative and positive controls, respectively (**Figure 2A**). All of the transformants grew well on selective SD/-Trp medium (**Figure 2B**), indicating that the three vectors were successfully transformed into the yeast cells. The GAL4-binding domain-OoNAC72 construct and pGAL4 grew well on SD/-Trp-His-Ade medium, while the transformants containing the pGBKT7 vector did not grow on the same medium (**Figure 2C**). The results of α-galactosidase activity assays showed that transformants containing GAL4 binding domain-OoNAC72 construct and pGAL4 appeared blue in color on SD/-Trp-His-Ade medium containing 5-bromo-4 chloro-3-indoxyl-α-D-galactopyranoside (X-α-Gal) (**Figure 2D**). These results indicated that OoNAC72 had transcriptional activity in yeast.

# Expression Patterns of OoNAC72 Under Various Treatments

independent replications. Asterisk indicates significant difference (∗P < 0.05; ∗∗P < 0.01) between transgenic lines and WT.

Quantitative real-time PCR was used to evaluate the expression patterns of OoNAC72 during dehydration (20% PEG-6000), salt (150 mM NaCl) and three hormone treatments (100 mM ABA, ETH, and GA) (**Figure 3**). When the seedlings were treated with dehydration, OoNAC72 mRNA abundance was slightly induced at 1 h, followed by progressive elevation until reaching the peak value at 3 h, which showed an approximately 11.9-fold increase relative to the initial level (**Figure 3A**). Treatment with 150 mM NaCl led to a quick accumulation of OoNAC72 mRNA level, which progressed until a maximum level was reached at 6 h, approximately 12.3-fold increase relative to the initial level (**Figure 3A**). Moreover, for the ABA treatment, the expression levels of OoNAC72 increased rapidly and reached the maximum level at 3 h, being 41.8-fold greater than the control (0 h), while there were no significant changes when treated with ETH and GA (**Figure 3B**). All the results suggested that OoNAC72 may respond to stress in O. ochrocephala by participating in ABA-dependent signal transduction pathways.

#### Overexpression of OoNAC72 Changed the Phenotype of Arabidopsis

To explore the function of OoNAC72 during the tolerance to abiotic stress, transgenic Arabidopsis over-expressing plants driven by the CaMV35S promoter were generated. We analyzed the growth status of two OoNAC72-OX transgenic lines (OX1, OX2) from the screening transgenic positive seedlings. Compared with wild-type plants, OoNAC72- OX transgenic lines (OX1, OX2) showed rosette leaves during the vegetative growth stage, and the number of leaves was significantly increased (**Figures 4A,B,D**); At the reproductive stage, OoNAC72-OX transgenic lines (OX1, OX2) showed significant delayed bolting and flowering (**Figures 4C,E**). OoNAC72-OX transgenic lines (OX1, OX2) showed no significant difference in leaf number, flowering plant height, dry and wet weight (**Figures 4D–G**). These

findings suggested that OoNAC72 was an important regulator of plant development.

### Overexpression of OoNAC72 Increases Tolerance to Salt and Osmotic Stresses Under Sterile Condition in Arabidopsis

To further investigate mechanisms of hypersensitivity to abiotic stress in OoNAC72-OX plants, we evaluated the stress tolerance of transgenic (OX1 and OX2 lines) and WT Arabidopsis in germination and root growth in seedlings (**Figure 5**). WT and OoNAC72-OX lines were no significant difference on 1/2 MS medium, whereas OX1 and OX2 lines better growth and longer root length under 150 mM NaCl, 200 mM NaCl, 200 mM mannitol and 250 mM mannitol stresses. These results suggested that OoNAC72-OX plants had improved tolerance to salt and drought stresses during seed germination and post-germinative growth periods.

#### Drought Tolerance and Salt Tolerance Phenotypes in Transgenic Plants Under Non-sterile Conditions and Analysis of Physiological Indices

To further elucidate the possible involvement of OoNAC72 in the response to drought and salt, we compared drought and salt tolerance of the OoNAC72-OX lines and WT plants at the vegetative growth stage (**Figure 6**). After 20 days of 150 mM NaCl stress treatment, the phenotypes of the WT and OX1 and OX2 lines began to show significant differences, in which 87.5% of the WT died, while the majority of OX1 and OX2 lines remained green,whereas OX1 and OX2 showed significantly higher survival rate 75 and 62.5% (**Figures 6A,C**). In order to

stresses. Seedlings were grown without water for 35 days and re-watered for 5 days. (C) Survival rate after recovery from drought and salt stress. Each experiment comprises 20 plants. (D–G) Proline and MDA content, SOD and POD activities in WT and OoNAC72-OX lines after drought and salt stress. Each data point is the mean of three biological replicates. Error bars indicate SD, and asterisks indicate a significant difference (∗P < 0.05; ∗∗P < 0.01) compared with WT.

detect the drought resistance of OoNAC72-OX lines, WT and OoNAC72-OX lines were simultaneously withholding water for 35 and 5 days after rehydration, the phenotypes of the WT and OoNAC72-OX lines began to show significant differences, the leaves of WT lines completely wilted and the plants survival rate was 3.33%. However, OX1 and OX2 lines survival rate were 87.5% (**Figures 6B,C**). Furthermore, there were no significant differences in proline and MDA content, POD and SOD activities between WT and OX1 and OX2 lines under normal condition (**Figures 6D–G**). Whereas, under salt and drought stresses, OX1 and OX2 lines exhibitedhigher levels of proline content, POD and SOD activities, and lower MDA level compared with those of WT (**Figures 6D–G**). Phenotypic analysis showed that the OoNAC72-OX plants had the highest drought and salt tolerance. These data demonstrated that the OoNAC72-OX plants exhibited increased tolerance to salt and drought stresses, thus we speculated OoNAC72 plays a critical role in O. ochrocephala response to salt and drought stresses.

#### Drought Resistance of WT and OoNAC72-OX Plants

To further investigate drought sensitivity of the OoNAC72-OX plants, the rate of nature water loss and stomatal apertures of leaves from the 25-day-old soil-grown WT and OoNAC72-OX plants were detected (**Figure 7**). After 3 h of air drying, the

leaves of OX1 and OX2 were slightly curled, while the WT plants were severely curled (**Figure 7A**). Additionally, OoNAC72-OX lines showed lower water loss rate compared with WT plants (**Figure 7B**), indicating that over expression of OoNAC72 had increased water retention capacity in Arabidopsis. After 10 days of drought control, stomatal apertures index of the transgenic and WT lines all appeared significant changes. OX1 and OX2 plants decreased from 0.40 and 0.49 to 0.09 and 0.08, which was significantly smaller than WT (from 0.46 to 0.27) (**Figures 7C,D**). WT and OoNAC72-OX plants were treated with 20% PEG-6000 simulated drought stress. ABA was detected at different time points. The ABA content of these OoNAC72-OX lines were significantly higher than that of the WT (**Supplementary Figure S3**). Those results indicated that OoNAC72-OX plants may reduce the loss of water by regulating the stomata closure to improve drought resistance.

# Altered Expression of Stress-Responsive Genes in Transgenic OoNAC72 Plants

To explore the underlying basis of this phenotype caused by drought or salt stress in the transgenic plants, eight stressresponsive genes: RD29A (Nakashima et al., 2006), RD29B (Nakashima et al., 2006), RD26 (Fujita et al., 2004), LEA14 (Jia et al., 2014), ANACOR19 (Tran et al., 2004), ZAT10 (Mittler et al., 2002), PP2CA (Yoshida and Hirayama, 2006) and NCED3 (Iuchi, 2001) were selected for expression pattern assays (**Figure 8**). Under normal condition, the expression levels of eight genes showed no significant difference between the OoNAC72- OX lines and WT. However, the expression levels of the eight genes in the OX1 and OX2 lines were significantly higher than those in WT plants under salt and drought conditions.

#### Increased ABA Sensitivity in OoNAC72-OX Plants

The ABA sensitivity of the OoNAC72-OX lines was assessed by analyses of seed germination and seedling growth (**Figure 9**). In the absence of ABA, no obvious difference was observed between transgenic and WT plants under normal growing conditions. However, when supplied with 1 or 3 µM ABA, seeds germination and seedling root length of the OX1 and OX2 lines were significantly inhibited compared with the WT lines (**Figures 9A– D**). Moreover, when 40-day-old plants were treated with 10 µM ABA, the stomatal apertures of both wild-type and OoNAC72- OX lines all happened to change (**Figure 9E**). The stomatal apertures index of WT plants decreased from 0.47 to 0.25, while the OX1 and OX2 lines dropped from 0.48 and 0.52 to 0.12 and 0.15 (**Figure 9F**). It indicated that the OoNAC72-OX lines were more rapid and variable in stomatal conductance. These

results indicated that overexpression of OoNAC72 gene led to increased ABA sensitivity, which resulted in retarded growth of transgenic plants.

# DISCUSSION

#### O. ochrocephala Can Be Exploited as a Pasture and Ecological Resource

O. ochrocephala is a perennial poisonous plant widely distributed in harsh environments, and livestock generally do not eat it when edible pasture is relatively abundant. In recent years, due to the emergence of extreme climatic and human overgrazing, O. ochrocephala has spread rapidly and become one of the main poisonous weeds in Western China. Studies have shown that O. ochrocephala is rich in protein and mineral elements and can be used as a good forage forb for livestock after detoxification (Shen and Mo, 2017). At the same time, as a highly resistant plant, O. ochrocephala has important ecological value in wind break and sand fixation, as well as biodiversity maintenance. In view of these characteristics, we focus on O. ochrocephala to elucidate the survival mechanism in adversity, and ultimately to provide a theoretical basis for its development and utilization.

# OoNAC72 Is a Stress-Related NAC Transcription Factor

In the present study, we characterized the OoNAC72 gene, a novel stress-related member of the NAC gene family in O. ochrocephala. OoNAC72 protein contains a typical NAC conserved domain located in the N-terminal region, which can be divided into five subdomains. As the transcriptional regulatory domain, the C-terminus is extremely variable. This result is in agreement with previous reports on other NAC TFs domains, such as PbeNAC1, ZmNAC55, CmNAC1 (Mao et al., 2016; Cao et al., 2017; Jin et al., 2017). Phylogenetic analysis confirmed the OoNAC72 had the highest homology with the two legumes, Medicago truncatula and Cicer arietinum. Subcellular localization and transactivation analysis jointly revealed that OoNAC72 was a transcription factor, which was consistent with previous research (Mao et al., 2016). Therefore, OoNAC72 act as a transcriptional activator, depending on interactions with other transcription factors or conformational changes. These results speculate that OoNAC72 is a novel stress-related member of the NAC gene family in O. ochrocephala.

# OoNAC72 Is Involved in Response to Diverse Stresses

In previous researches, many NAC genes had been identified as an important part of the progress of complex signal transduction when plants were subjected to various stresses, especially abiotic stress such as drought and salt stress (Puranik et al., 2012). In our study, we found that OoNAC72 can be induced by high salt, drought and ABA treatment, whereas not induced by GA and ETH. we speculated that OoNAC72 may respond to stress in O. ochrocephala by ABA-dependent signal transduction pathways. These abiotic stresses induced expression pattern of

OoNAC72 was supported by earlier studies, such as in pumpkin (Cao et al., 2017), Pyrus betulifolia (Jin et al., 2017), tomato (Zhu et al., 2014), chickpea (Yu et al., 2014, 2016) and Tamarix hispida (Wang et al., 2017).

# OoNAC72 Is Involved in Developmental Processes

Numerous reports have demonstrated that NAC TFs are involved in a number of biological processes, such as regulating the growth of plant cells (Hiroaki et al., 2010), seed development (Sperotto et al., 2009), embryonic development, fiber formation and development, cell differentiation and leaf senescence (Youn-Sung et al., 2007). In our study, we found that OoNAC72-OX (OX1, OX2, OX3) lines showed significant rosette leaves, delayed twitches and flowering compared with WT plants. However, the mechanism resulting from overexpression of OoNAC72 in regulating this phenotype is still unknown and waiting for further research.

#### Overexpression of OoNAC72 Improves Stress Tolerance in Arabidopsis

Overexpression or deletion mutation is the primary way to study the function of NAC transcription factors in adversity stress. In this study, several physiological changes of OoNAC72-OX lines seem to be involved in abiotic stress-resistant and molecular mechanisms. Firstly, in terms of phenotype, OoNAC72-OX lines appeared improved tolerance to salt and drought stresses during seed germination, post-germinative growth periods and vegetative growth stage. Consistent with the results of OoNAC72, CarNAC4-transgenic plants exhibited enhanced drought and salt tolerance than the WT plant, which were strongly demonstrated by both morphological and physiological changes (Yu et al.,

<sup>∗</sup>P < 0.05, ∗∗P < 0.01.

2016). Secondly, at the physiological and biochemical levels, proline and MDA content, POD and SOD activities in plants overexpressing OoNAC72 have been significantly induced under different stress conditions. Thirdly, at the molecular level, our data also demonstrated that over-expression of OoNAC72 enhanced the expression of stress-responsive genes such as RD29A, RD29B, RD26, LEA14, ANACOR19, ZAT10, PP2CA, and NCED8. Previous studies have confirmed that NAC proteins could bind to NACRS containing the core sequence "CGT[A/G]" to regulate gene expression through ABA dependent pathway for various stress response (Duval et al., 2002; Olsen et al., 2005a; Takasaki et al., 2010; Jensen et al., 2013). However, the genome of O. ochrocephala has not been sequenced and the OoNAC72 promoter sequence could not be cloned, and so it is impossible to predict these cis-acting elements associated with stress in the promoter, and resistance mechanisms need to be further investigated.

#### OoNAC72 Functions in ABA Signaling and Confers Drought Resistance to Transgenic Plants

Abscisic acid is a very important signaling molecule that respond to many adverse environmental stresses such as high salt, drought and extreme temperatures (Agarwal and Jha, 2010; Fujita et al., 2011). In plants, higher ABA sensitivity may irritate the stomata to maintain moisture and enhance drought resistance. In our study, ABA content of WT and OoNAC72-OX plants treated with 20% PEG-6000 simulating drought stress showed that ABA content of the OoNAC72-OX lines was significantly higher than that of the WT (**Supplementary Figure S3**). Furthermore, we found that the stomata in Arabidopsis overexpressing OoNAC72 had different degrees of closure after drought stress, indirectly mapping the relationship between drought stress and the ABA pathway. Similar results were found in Arabidopsis and maize (Ping-Li et al., 2007; Mao et al., 2016). Overexpression of KUP6 (K+ uptake transporter 6) in Arabidopsis increased the sensitivity of ABA to drought stress through faster stomatal closure, thereby improving the tolerance of transgenic plants to drought stress (Yuriko et al., 2013). Genes such as PP2Cs, SnRK2s, NAC and WRKY played a vital role in ABA signaling pathways induced by drought stress (Kim, 2014). Therefore, we supposed that OoNAC72 may enhance the drought resistance of transgenic Arabidopsis by participating in the ABA pathway.

# OoNAC72 Scavenges ROS Capability by Increasing ABA Content Under Stress

Under stress conditions, plants reduce the damage caused by stress-induced ROS by enhancing their antioxidant defense system (Oracz et al., 2009; Suzuki et al., 2012). MDA is one of the important parameters to measure the degree of oxidative damage in plant cells. While pro contributes to osmotic adjustment to effectively enhance the antioxidant system and reduce peroxidative damage (Zhuang et al., 2015a). In our study, transgenic plants produced less MDA and more pro under stress than wild-type plants, revealing that OoNAC72 overexpression enhanced the body's regulation of feedback regulatory substances.

Under drought stress, ABA acts as an upstream signaling of NO and they form cross-signal pathways co-regulate a balance between ROS (H2O2) and NO production (Zhuang et al., 2015b). Our resistance assays showed that the ABA contents, and SOD and POD activities were significantly induced by OoNAC72 under drought stresses condition. The increase in ABA content leads to an increase in the concentration of NO for activating NO signaling pathway. On the one hand, the low concentration of NO interacts with H2O<sup>2</sup> to stimulate the antioxidant defense system. On the other hand, the increase of SOD and POD activity further enhances plant's ability to scavenge excessive H2O2. The synergy between NO signal and ROS signal promotes the dynamic balance between NO and H2O2, and ultimately improves plant resistance tolerance (Miller et al., 2010). These results indicate that showing the positive regulation of OoNAC72 on ABA signaling pathway under drought stress.

# CONCLUSION

Plant NAC transcription factors control diverse biological processes, such as differentiation, development and abiotic stress responses. In this study, we identified a gene encoding NAC72 type transcription factor from Oxytropis ochrocephala, and characterized its role. The nuclear localization and transcriptional activity of this protein indicate the possible function of OoNAC72 as a transcription factor. In addition, expression of OoNAC72 transcript was shown to be up-regulated in response to abiotic stresses and exogenous ABA. Furthermore, analyses of transgenic Arabidopsis expressing OoNAC72 also supported the involvement of OoNAC72 in drought and salinity responses as well as in the regulation of ABA-dependent processes.

# AUTHOR CONTRIBUTIONS

HG and YF conceived and designed the study. HG, XL, and FN performed the experiments. HG, XL, and QZ analyzed the data and wrote the manuscript. FN, QZ, NF, DC, and DM reviewed the manuscript. WH, BG, YW, and YF contributed reagents, materials, and fund support.

# FUNDING

This work was supported by Special Fund for Agro-scientific Research in The Public Interest (201203062), Key Research and Development Plan Project of Shaanxi Province (2018ZDXM-SF-016) and Opening Foundation of Key Laboratory of Resource Biology and Biotechnology in Western China (Northwest University), Ministry of Education (ZSK2016003) to YW, Bureau of Science and Technology in Shaanxi Province, Xi'an City, Beilin District (GX1702) to BG, Natural Science Foundation of Shanxi Province of China (2018JQ3029) and Natural Science Foundation of China (31702159) to YF, Key Research and Development Plan Project of Shaanxi Province (2018SF-034) and Scientific Research from Shaanxi Provincial Department of Education (17JK0243) to NF.

### ACKNOWLEDGMENTS

fpls-10-00890 July 10, 2019 Time: 17:16 # 13

We thank the reviewers for their constructive suggestions. We are also particularly grateful to other members of our research group for their helpful comments to improve the article.

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00890/ full#supplementary-material

gene NCED3 in Arabidopsis thaliana. FEBS Open Bio. 3:321. doi: 10.1016/j.fob. 2013.07.006



transgenic Tamarix and Arabidopsis. Front. Plant Sci. 8:635. doi: 10.3389/fpls. 2017.00635


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

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

# The Roles of GmERF135 in Improving Salt Tolerance and Decreasing ABA Sensitivity in Soybean

Meng-Jie Zhao<sup>1</sup>† , Li-Juan Yin<sup>1</sup>†‡, Jian Ma<sup>2</sup>† , Jia-Cheng Zheng<sup>3</sup> , Yan-Xia Wang<sup>4</sup> , Jin-Hao Lan<sup>5</sup> , Jin-Dong Fu<sup>1</sup> , Ming Chen<sup>1</sup> , Zhao-Shi Xu1,2,3 \* and You-Zhi Ma<sup>1</sup> \*

#### Edited by:

1

Yan Guo, China Agricultural University, China

#### Reviewed by:

Xiuli Hu, Henan Agricultural University, China Yangrong Cao, Huazhong Agricultural University, China

#### \*Correspondence:

Zhao-Shi Xu xuzhaoshi@caas.cn You-Zhi Ma mayouzhi@caas.cn †These authors have contributed

equally to this work ‡Present address:

Li-Juan Yin, Agricultural Genomics Institute at Shenzhen, Chinese Academy of Agricultural Sciences, Shenzhen, China

#### Specialty section:

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

Received: 22 November 2018 Accepted: 04 July 2019 Published: 23 July 2019

#### Citation:

Zhao M-J, Yin L-J, Ma J, Zheng J-C, Wang Y-X, Lan J-H, Fu J-D, Chen M, Xu Z-S and Ma Y-Z (2019) The Roles of GmERF135 in Improving Salt Tolerance and Decreasing ABA Sensitivity in Soybean. Front. Plant Sci. 10:940. doi: 10.3389/fpls.2019.00940

Institute of Crop Science, Chinese Academy of Agricultural Sciences (CAAS)/National Key Facility for Crop Gene Resources and Genetic Improvement, Key Laboratory of Biology and Genetic Improvement of Triticeae Crops, Ministry of Agriculture, Beijing, China, <sup>2</sup> Department of Agronomy, Jilin Agricultural University, Changchun, China, <sup>3</sup> College of Agriculture, Anhui University of Science and Technology, Fengyang County, China, <sup>4</sup> Hebei Academy of Agriculture and Forestry Sciences, Research Center of Wheat Engineering Technology of Hebei, Shijiazhuang, China, <sup>5</sup> College of Agronomy and Plant Protection, Qingdao Agricultural University, Qingdao, China

Abscisic acid (ABA) mediates various abiotic stress responses, and ethylene responsive factors (ERFs) play vital role in resisting stresses, but the interaction of these molecular mechanisms remains elusive. In this study, we identified an ABA-induced soybean ERF gene GmERF135 that was highly up-regulated by ethylene (ET), drought, salt, and low temperature treatments. Subcellular localization assay showed that the GmERF135 protein was targeted to the nucleus. Promoter cis-acting elements analysis suggested that numerous potential stress responsive cis-elements were distributed in the promoter region of GmERF135, including ABA-, light-, ET-, gibberellin (GA)-, and methyl jasmonate (MeJA)-responsive elements. Overexpression of GmERF135 in Arabidopsis enhanced tolerance to drought and salt conditions. In addition, GmERF135 promoted the growth of transgenic hairy roots under salt and exogenous ABA conditions. These results suggest that soybean GmERF135 may participate in both ABA and ET signaling pathways to regulate the responses to multiple stresses.

Keywords: ABA, ethylene-responsive factor, hypocotyl elongation, root growth, response mechanism, salt tolerance, soybean

# INTRODUCTION

As a vital hormone in plants, abscisic acid (ABA) is essential to a myriad of aspects of plant growth and developmental processes, including plant embryo development, seed maturation, fruit maturity, and stomatal movement (Finkelstein et al., 2002). ABA signal transduction has been studied for many years (Pandey et al., 2006, 2009; Shen et al., 2006; Ma et al., 2009; Park et al., 2009; Yang and Guo, 2018), and the widely accepted molecular mechanism is that the pyrabactin resistant/PYR-like/regulatory component of ABA receptor (PYR/PYL/RCAR) acts as an ABA receptor which can bind to ABA, and then binds to and inhibits protein phosphatases type 2C (PP2Cs) (Santiago et al., 2009). In addition, the activity of SNF1-related protein kinases 2 (SnRK2s) is enhanced and can phosphorylate ABRE binding factors (AREB/ABFs) to induce physiological and biochemical changes in the process of ABA response (Hubbard et al., 2010; Yoshida et al., 2015). A recent study showed that ABFs can directly bind to the promoters of group A PP2C genes, and

rapidly induce their expression on exogenous ABA treatments (Wang et al., 2018). However, the downstream molecular mechanism is yet not clearly understood.

Many transcription factors, such as MYC/MYB, bZIP/ABRE, AP2/ERF, and NAC, are regulated by ABA. The AP2/ERF family, the largest plant transcription factor family (Okamuro et al., 1997), can be divided into three subfamilies: the AP2, ERF, and RAV subfamilies. Among them, the ERF subfamily can specifically bind to GCC-box and/or the dehydrationresponsive element/C-repeat (DRE/CRT) cis-acting elements (Allen et al., 1998).

The AP2/ERF family is involved in responses to various abiotic stresses and exogenous hormones (Xu et al., 2008, 2011; Klay et al., 2018). For example, overexpressing TaERF1 enhances tolerance of physiological and environmental stresses, such as salt, drought, low temperature, exogenous ABA, ethylene (ET), salicylic acid (SA), and disease (Xu et al., 2007). Transgenic rice plants expressing OsERF922 displayed higher susceptibility to Magnaporthe oryzae and NaCl compared to the wild type (WT), while the knockout mutant and RNAi lines enhanced resistance to these stresses (Liu et al., 2012). A recent study described how ABA modulates the expression level of ERF family, showing that the expression levels of ABA-responsive genes such as RD22, LEA3, and PODs were up-regulated after overexpressing OsERF101 in rice, which enhanced its susceptibility to ABA (Jin et al., 2018).

Transcriptomic analysis of grapevine organs treated with or without ABA showed that ERF members were involved in differently expressed genes (DEGs), and the ERF subfamily had the most significant change compared to other transcription factors (Rattanakon et al., 2016). Although the ABA signal pathway has been extensively studied, it is yet unclear how regulation of the expression of downstream genes via ERF subfamily could enhance abiotic stress tolerances in soybean. To investigate whether the ERF subfamily is modulated by the ABA signal in soybean, we studied the response to exogenous ABA and identified the functions of GmERF135 in transgenic Arabidopsis and soybean hairy roots.

### MATERIALS AND METHODS

#### Plant Materials and Treatment

Soybean cultivar "Tiefeng 8" was sown in pots containing vermiculite and grown at 25◦C for 14 days and then treated with various abiotic stresses. For the various treatments, soybean plants were exposed in the air for rapid drought, dipped into 200 mM NaCl for salt stress, placed in 42◦C/4◦C chamber for high/low temperature stress, respectively. They were also placed in an airtight container filled with ET, or dipped in 100 µM ABA, 50 µM salicylic acid (SA), or 50 µM jasmonate (JA) for exogenous hormone stresses. Leaves of these plants were collected after 0, 0.5, 1, 2, 5, 12, and 24 h treatment and then immediately stored at −80◦C for RNA extraction.

# Gene Structure and Protein Domain of ERFs in Soybean

The whole genome data of the candidate ERF genes was obtained from JGI Glyma1.0 annotation<sup>1</sup> (Goodstein et al., 2012). Gene structure was analyzed by submitting the CDS of the candidate genes and whole genomic DNA sequences to the Gene Structure Display Server (GSDS) website<sup>2</sup> . Protein Fold Recognition Server (PHYRE2)<sup>3</sup> was used for analysis of structural homology modeling of these genes, and DOG 2.0 was used to draw the protein domains.

# Quantitative Real-Time PCR (qRT-PCR)

Total RNA of the soybean plants was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA, United States). The specific primer pairs for the 16 genes were designed by Primer Premier 5.0 according to the cDNAs. qRT-PCR was conducted using the SYBR Premix Ex TaqTM kit (TAKARA, Kusatsu, Japan) and the ABI Prism 7500 real-time PCR system (Thermo Fisher Scientific, Waltham, MA, United States). The 2−11CT method was used to conduct qRT-PCR analysis (Le et al., 2011). Soybean Actin (U60506) was used as an internal control and sequence of specific primers were shown in **Supplementary Table S1**.

#### Isolation and Promoter Analysis of GmERF135

The GmERF135 gene was amplified by PCR and the primers for cloning the GmERF135 were 5<sup>0</sup> - AATCATTATGTGTGGCGGTGCC-3<sup>0</sup> and 5<sup>0</sup> -TATTCCTCGC TAATCGAAACTCCAGAG-3<sup>0</sup> . The PCR product was then cloned into the pEASY-T1 vector (TransGen, Beijing, China). 1,886 bp promotor region of GmERF135 was cloned and submitted to PLACE<sup>4</sup> and PlantCARE<sup>5</sup> databases to analyze the putative cis-acting elements in the promoter region (Lescot et al., 2002).

# Subcellular Localization Analysis

The cDNA of GmERF135 were augmented with PCR, connected to the N-terminus of humanized green fluorescent protein (hGFP) reporter gene under the control of the double Cauliflower Mosaic Virus (2 × CaMV) 35S promoter, and a recombinant plasmid was obtained (Scott et al., 1999). The recombinant plasmid was introduced into onion epidermal cells, while the onion epidermal cells with hGFP vector acted as the control. Fluorescence microscopy was used to identify hGFP expression (Xu et al., 2007; He et al., 2016).

# Generation of Transgenic Arabidopsis and Stress Treatments

AT5G47230 is the orthologue of GmERF135 in Arabidopsis, which was named ERF2 and used to investigate function in

<sup>2</sup>http://gsds.cbi.pku.edu.cn/

<sup>1</sup>https://phytozome.jgi.doe.gov/pz/portal.html

<sup>3</sup>http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index

<sup>4</sup>http://www.dna.affrc.go.jp/PLACE/

<sup>5</sup>http://bioinformatics.psb.ugent.be/webtools/plantcare/html/

responses to various stresses. The seeds of two mutants, erf2- 1 and erf2-2 (SALK\_126889, SALK\_076967) were mutated via T-DNA insertion.

The cDNA of GmERF135 was obtained by using the specific primer pairs: 5<sup>0</sup> -TGATTACGCCAAGCTTATGTGTG GCGGTGCC-3<sup>0</sup> , 5<sup>0</sup> -CCGGGGATCCTCTAGAATCGAAACTC CAGAG-3<sup>0</sup> , and then were cloned into pBI121 under the control of the CaMV 35S promoter. The recombinant plasmid 35S::GmERF135 was sequenced and transformed into wild-type (WT) and the two mutants Arabidopsis lines using the vacuum infiltration method (Bechtold and Pelletier, 1998; Liu et al., 2013). T3 seeds of transgenic lines were selected for further analysis.

For root length assay, 30 seeds of each line were sown on 1/2 strength MS growth medium with or without 6% PEG, 75 mM NaCl, 1-aminocyclopropane-1-carboxylic acid (ACC), or in dark treatment for growth, respectively. At least 30 seedlings per line were randomly selected to measure root length. The cotyledon pieces of each line were recorded every 12 h. Each treatment contained three independent replicates.

### Soybean Hairy Root Induction and Stress Treatments

The Superroot of Lotus corniculatus and Cucumopine-type Agrobacterium rhizogene strain K599 with pGFPGUSPlus were provided by Professor Tian-Fu Han (CAAS, China). Seedling growth, rooting, hairy root induction, and hairy root transformation were performed as described by Chen et al. (2004, 2014)

GFP positive (GFP+) hairy roots were induced by K599 carrying the pGFPGUSPlus-GmERF135 binary vector. These hairy roots were cultured on 1/2 MS medium supplemented with 0, 50, 85, 120, or 150 mM NaCl for salt treatment, and 50, 100, or 150 µM ABA for hormone treatment, respectively. They were then incubated at 24◦C under a 16/8 h light/dark cycle condition for 2 weeks. After 24 h incubation at 105◦C, the increase in dry weight (30 roots per unit) was measured and recorded. Each treatment contained three independent replicates.

# RESULTS

### Molecular Characterization of Soybean Targeted ERF Genes

In a previous study, 160 non-redundant soybean ERFs were identified using the Pfam database and SMART program, and these soybean ERFs were clustered into eight groups (**Supplementary Table S2**; Zhao et al., unpublished). To comprehensively understand the responses of soybean ERFs to ABA, 16 ERF genes were selected for further investigation according to the phylogenetic tree. Gene structure analysis showed that six genes had no introns, including GmERF106, GmERF132, GmERF135, GmERF41, GmERF49, and GmERF84 (**Figure 1**). The remaining 10 ERF genes contained one intron, which was distributed in each cDNA region except for GmERF103, GmERF111, and GmERF15. The protein domains of each soybean ERF gene were drawn by DOG2.0 (**Figure 2**). All the proteins contained a conservative AP2/ERF domain, which was distributed in different positions of each protein sequence. The AP2/ERF domain of the proteins in each group displayed similar locations such as Group I, III, IV, V, VII, and VIII.

An expression pattern map of soybean ERFs was drawn based on the gene-chip data downloaded from the soybean genome database<sup>6</sup> . As shown in **Supplementary Figure S1A**, the soybean ERFs were expressed at different levels in various tissues and organs. Among them, GmERF111 and GmERF135 showed high expression levels in almost all tissues and organs. GmERF75 showed high expression levels in root and nodule, and GmERF49 was highly expressed in nodule and 10 DAF pod shell. GmERF15 mainly expressed in young leaf. All the candidate genes had different expression patterns.

#### Isolation and Subcellular Localization Characterization of GmERF135

Abscisic acid plays an important role in plant growth and development which is closely related to quality and yield in plants. To investigate how ABA affected the expression pattern of the 16 soybean ERF genes, qRT-PCR was conducted. Almost all the ERFs were up-regulated by ABA, except for GmERF49 (**Figure 3**). The ERF gene GmERF135 was the highest expressed gene after exogenous ABA treatment except for GmERF75 which has been studied (Zhao et al., unpublished). The transcript level of GmERF135 was activated and showed a 17-fold increase within 2 h after treatment and then declined to normal. Considering the high expression level and ABA-responsive increase in plants, GmERF135 was selected for further study.

The subcellular localization assay was conducted to provide clues to understand intrinsic characteristics of cell activities. hGFP reporter gene fused to C-terminus of GmERF135, which could fluoresce when laser irradiated. The GmERF135::hGFP fused protein only fluoresced in nucleus, while the control fluoresced throughout the entire cell (**Figure 4**). This result showed that GmERF135 functions in nucleus.

# GmERF135 Promoter Region Comprises Diverse Stress-Responsive Elements

The promoter region is an important part of the gene which could regulate gene expression and control gene action. To investigate the potential regulation mechanism, the promoter region of GmERF135, which has 1886 bp length upstream of the start codon, was isolated. The PLACE and PlantCARE databases were used to analyze the putative regulatory elements in the promoter region. Several regulatory elements were identified to be involved in responses to abiotic stresses and plant hormones (**Table 1**).

The GmERF135 promoter region contains many ABA and stress responsive elements, including ABA-responsive elements (ABREs, 3 hits), MYBST1 core sequences (4 hits), MYB binding sites (MBS, 1 hit), and DPBF binding sites (2 hits) **(Table 1**). Except for ABREs, three hormone-responsive elements were predicted, including an ethylene- responsive element (ERE, 1 hit), MeJA-responsive element TGACGmotif (1 hit) and CGTCA-motif (1 hit), and GA-responsive

<sup>6</sup>https://soybase.org/soyseq

blue boxes represent introns, and black lines represent untranslated regions (UTRs).

element (GARE, 2 hits). In addition, TCT-motif (1 hit), G-box (2 hits), and Box 4 (1 hit) were found, which could act as light-responsive elements in the promoter region of GmERF135. These elements suggested that GmERF135 may be involved in responses to multiple abiotic stresses and exogenous hormones.

#### The Roles of GmERF135 in Responding to Multiple Stimuli

Quantitative Real-Time PCR was conducted to investigate the expression level of GmERF135 under abiotic stresses, such as drought, salt, high/low temperature, and exogenous hormones including ET, SA, and JA. GmERF135 was induced by almost all the abiotic stresses and exogenous hormones. Remarkably, the expression level of the gene was extremely up-regulated by drought, low temperature, and ET treatment. The transcription of GmERF135 peaked at 5 h under drought or low-temperature treatment which had 18-fold / 22-fold increases, respectively (**Figure 5**). The peak of GmERF135 transcription appeared at 12 h after treatment with ET, which showed a 28-fold increase. In contrast, GmERF135 rapidly reached the maximum transcript level within 0.5 h under the NaCl and JA treatments, and then declined to baseline after 1 h. Under the SA treatment, GmERF135 was



induced and reached the peak within 2 h (about 7-fold) and then declined to normal level in 24 h. These results showed that GmERF135 may be involved in responses to multiple stimuli.

#### GmERF135 Rescued Two erf2 Mutants and Affected Growth of the Root and Cotyledon

Total RNA was extracted for semi-quantitative PCR from hypocotyls, root, stem, and leaf tissues of normal growth soybean seedlings. Actin primers were used as a parallel reaction to normalize the added template amounts. GmERF135 was predominantly expressed in the leaf, less in the hypocotyl and stem, and very little in the root (**Supplementary Figure S1B**).

To assess whether GmERF135 could rescue the two mutants, the seeds of transgenic GmERF135::erf2 lines, WT, and two erf2 mutants were sown on the 1/2 MS medium for growth and the plant growth rate was recorded each 12 h. Six day later, the phenotypes were imaged (**Figures 6A,B**).

Growth rate assay showed GmERF135::erf2 lines displayed faster growth rate compared to the mutants **(Figure 6C**). WT plants required 2 d of growth to sprout the two pieces of the cotyledon while the two mutants required 4 days. After introduction of GmERF135 in erf2 lines, 3 days was required for the same process. The two mutants needed 6 days to reach 4-leaf stage, while the transgenic GmERF135::erf2 lines had similar growth rate with WT, which only needed 5 d. Root length assay showed there is an approximately 25% decrease in erf2 mutants compared to WT lines **(Figure 6D**). After overexpressing GmERF135 in the erf2 mutants, the lines had similar root lengths to WT and the phenotype of which was rescued. These results showed that overexpression of GmERF135 in the two erf2 mutants partly rescued two erf2 mutants and affected growth of root and cotyledon in Arabidopsis.

#### GmERF135 Promotes Plant Growth Under Drought and Salt Stresses in Arabidopsis

To determine whether GmERF135 confers abiotic stress tolerance to Arabidopsis plants, 3-day-old WT, erf2 mutants, and GmERF135 overexpression seedlings were transferred to 1/2 MS medium with or without 6% PEG, 75 mM NaCl, ACC, and dark environment grown for 3 days. The growth rate and the root length of 30 seedlings in each line were recorded (**Figures 7B,C**). Statistics showed that the average root length of erf2 was shorter than the WT and transgenic Arabidopsis plants under all conditions except for ACC treatment (**Figures 7A,B**). The transgenic plants displayed larger cotyledon than the WT and mutants under NaCl and PEG treatments, which suggested that GmERF135 enhanced the growth rate of Arabidopsis plants. Overexpression of GmERF135 in Arabidopsis resulted in a longer root compared to the WT and two erf2 mutants under 6% PEG, 75 mM NaCl, and dark conditions. No detectable difference was observed between the transgenic lines and WT under ACC conditions.

To further understood the response to various stresses in Arabidopsis, some marker genes, such as NCED3 (Ruggiero et al., 2004), RD29A (Yamaguchi-Shinozaki and Shinozaki, 1994), COR15A (Xu et al., 2011), DREB2A (Liu et al., 1998), ABA1, ABA2 (He et al., 2016), ACO4, and ACS2 (Kim et al., 2013) were selected for qRT-PCR (**Supplementary Figure S2**).

FIGURE 5 | Expression patterns of GmERF135 under multiple stimuli. Two-week soybean seedlings were given various abiotic stresses and exogenous hormones for 0, 0.5, 1, 2, 5, 12, and 24 h, which were used to extract RNA and obtained cDNA. The transcript levels of GmERF135 after multiple stresses were quantified by qRT-PCR, including drought, NaCl, high/low temperature, ET (ethylene), JA (jasmonate), and SA (salicylic acid). Data were shown referring to three biological replicates.

#### GmERF135 Improves Tolerance of Salt and ABA in Soybean

To investigate the roles of GmERF135 in soybean, the pGUS-GmERF135 expression vector was constructed and transformed into soybean hairy roots, which were studied on MS basal medium supplemented with different concentrations of ABA and NaCl. GmERF135 transgenic soybean hairy roots experienced greater growth under the treatments compared to the vector control, especially under the 100 µM ABA and 85 mM NaCl conditions (**Figure 8A**).

Dry weight measurement results showed that GmERF135 hairy roots increased by about 10% compared to the vector under the MS0 condition, while transgenic hairy roots displayed a 6.24–75% improvement compared to WT under NaCl and exogenous ABA at the different concentrations (**Figure 8B**). For NaCl treatment, there has a 6.24–75.03% increase in the average dry weight of hairy roots overexpressing GmERF135. For exogenous ABA, the transgenic hairy roots displayed a 36– 65% improvement compared to vector control under exogenous ABA treatment at different concentrations (**Figure 8B**). It is noteworthy that a total of 64.46% improvement in GmERF135 hairy roots was observed under the 100 µM ABA condition, and an improvement of 75.03% with 85 mM NaCl (**Figure 8B**). Results of dry weight measurements confirmed the conclusion that GmERF135 could promote plant growth under ABA and NaCl conditions.

# DISCUSSION

Abscisic acid is well-known to be involved in responses to multiple stresses, such as drought, salt, and cold, and to induce expression of stress-related genes in plants (Leung et al., 1997;

FIGURE 7 | GmERF135 transgenic Arabidopsis lines improved cotyledon growth under the NaCl and PEG conditions. (A) Phenotype of GmERF135 transgenic Arabidopsis lines under various stresses. Three-day-old GmERF135 transgenic Arabidopsis lines were transferred to 1/2 MS medium with 6% PEG, 75 mM NaCl, ACC, or grown in darkness. (B) Average root lengths of WT, erf2, and 35S::GmERF135 lines. Three-day-old seedlings were treated with various stresses for 3 days and the data were shown as the means ± SD of three biological replicates (at least 30 individuals per line). (C) Days needed for growing cotyledon. Statistic data were recorded every 12 h. Three biological replicates were performed.

Rohde et al., 2000; Rook et al., 2001; Xu et al., 2007). The interaction of the ABA and ET signal pathway is extremely complex and intricate. AREB/ABFs, members of the bZIP family of transcription factors, participate in the ABA signal pathway which can specifically bind to ABRE cis-element to modulate the expression of downstream target genes (Bonetta and McCourt, 1998; Leung and Giraudat, 1998; Li et al., 2012). Fujita et al. (2005) observed that AREB1 can act as a transcription activator in ABRE-dependent ABA signaling, which enhances drought tolerance in Arabidopsis. Overexpression of the AREB1 gene in Arachis hypogaea could enhance drought tolerance by modulating ROS scavenging and maintaining content (Li et al., 2013). Phosphorylated AREB/ABFs may bind to the ABRE ciselement of GmERF135 promotor region to activate its transcription level (Fujita et al., 2013), which could activate or repress the transcription of targeted genes. In our study, analysis of the GmERF135 promoter region showed that three ABREs were located at the promoter region, which may be binding sites of AREB/ABFs to affect the transcript of GmERF135. GmERF135 has a conversed AP2/ERF domain, which could specifically bind to GCC-box and/or dehydrationresponsive element/C-repeat (DRE/CRT) cis-acting elements to modulate the expression of pathogenesis- and abiotic stressesrelated genes (Hao et al., 1998; Chakravarthy et al., 2003; Zarei et al., 2011).

Previous studies showed there is an antagonistic interaction between ABA and JA-ET signaling pathways ABA for disease response (Beaudoin et al., 2000; Anderson et al., 2004; Martin-Rodriguez et al., 2011). A recent study showed that SlAREB1 could enhance the expression of ethylene biosynthetic genes ACS and ACO in tomato fruits, which are the two key genes of ET synthesis (Mou et al., 2018). In our study, qRT-PCR analysis showed that the transcription of both AtACO4 and AtACS2 increased (**Supplementary Figure S2**) in transgenic Arabidopsis lines, which suggested acceleration of ET production. ET acceleration could trigger a series of reactions of ET and activate the transcript of ERF responsive genes (Guo and Ecker, 2004). At the same time, the transcript levels of ABA1 and ABA2, key factors of ABA synthesis, were also upregulated after overexpressing GmERF135 in Arabidopsis lines. These results suggested that GmERF135 may participate in both ET and ABA signaling pathways, and the regulation between the two signaling pathways needs further research.

Except for ABREs, other stress-related elements were also distributed in the promotor region of GmERF135, such as the MYBST1 core sequence, MBS, and ERE (**Table 1**). The MYBST1 core sequence and MBS elements have been demonstrated to be involved in drought, low temperature, salt, and ABA stress responses (Abe et al., 2003). qRT-PCR showed that GmERF135 is induced by multiple stresses, including drought, salt, low temperature, exogenous ET, SA, and JA. These changes may be caused by some corresponding cis-elements in promoter regions which can be bound by specific transcription factors (Huang et al., 2012; Buscaill and Rivas, 2014). These results suggest that soybean ERF gene GmERF135 is a key factor which participates in multiple signaling pathways to regulate expression levels of stress-related genes.

# AUTHOR CONTRIBUTIONS

Z-SX and Y-ZM coordinated the project, conceived and designed experiments, and edited the manuscript. M-JZ and L-JY conducted the bioinformatic work, performed experiments and wrote the first draft. JM revised the manuscript and figures. J-CZ conducted the bioinformatic work. Y-XW, J-HL, and J-DF contributed with valuable discussions. MC provided analytical tools and managed reagents. All authors have read and approved the final manuscript.

# FUNDING

This research was supported by the National Natural Science Foundation of China (31871624), the National Transgenic Key Project of Ministry of Agriculture (2018ZX0800909B), and the Technological Innovation Projects of Modern Agriculture of Hebei Province (494-0402-YBN-RDC4).

# ACKNOWLEDGMENTS

We thank Professor Tian-Fu Han for providing the Superroot culture of L. corniculatus, A. rhizogenes strain K599, and the binary vector pGFPGUSPlus. We also thank Dr. Wen-Sheng Hou for plant material preparation, technical assistance, and soybean seeds.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00940/ full#supplementary-material

FIGURE S1 | Expression patterns of candidate soybean ERFs in different organs. (A) Expression patterns of candidate soybean ERFs in different organs. Normalized expression data for the soybean ERF genes were collected from the SoyBase (http://www.soybase.org/). Expression (vertical coordinates) is in transcripts per million (TPM). (B) Semi-quantitative PCR of GmERF135 in different organs.

FIGURE S2 | Express levels of various stresses-related genes after treatment. Two-week-old WT and transgenic Arabidopsis lines were used to extract total RNA. GmActin/AtActin was used for normalization. Data were shown as the means ± SDs of three experiments.

TABLE S1 | Specific primers of each gene for qRT-PCR. All the primers were designed via Primer Primer 5.0.

TABLE S2 | Classification of 160 soybean ERF genes.

#### REFERENCES

fpls-10-00940 July 19, 2019 Time: 15:31 # 10


response in reproductive tissues. Plant Mol. Biol. 98, 51–65. doi: 10.1007/ s11103-018-0762-5



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

Copyright © 2019 Zhao, Yin, Ma, Zheng, Wang, Lan, Fu, Chen, Xu and Ma. 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.

# Transcriptome Analysis of the Hierarchical Response of Histone Deacetylase Proteins That Respond in an Antagonistic Manner to Salinity Stress

*Minoru Ueda1,2,3, Akihiro Matsui1,3, Shunsuke Watanabe4, Makoto Kobayashi5, Kazuki Saito5,6, Maho Tanaka1,3, Junko Ishida1,3, Miyako Kusano5,7, Mitsunori Seo4 and Motoaki Seki1,2,3,8\**

*1 Plant Genomic Network Research Team, RIKEN Center for Sustainable Resource Science, Yokohama, Japan, 2 Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Agency (JST), Kawaguchi, Japan, 3 Plant Epigenome Regulation Laboratory, RIKEN Cluster for Pioneering Research, Wako, Japan, 4 Dormancy and Adaptation Research Unit, RIKEN Center for Sustainable Resource Science, Yokohama, Japan, 5 Metabolomics Research Group, RIKEN Center for Sustainable Resource Science, Yokohama, Japan, 6 Graduate School of Pharmaceutical Sciences, Chiba University, Chiba, Japan, 7 Graduate School of Life and Environmental Sciences, University of Tsukuba, Tsukuba, Japan, 8 Kihara Institute for Biological Research, Yokohama City University, Yokohama, Japan*

#### *Edited by:*

*Vicent Arbona, University of Jaume I, Spain*

#### *Reviewed by:*

*Keqiang Wu, National Taiwan University, Taiwan Giorgio Perrella, ENEA - Centro Ricerche Trisaia, Italy*

*\*Correspondence:*

*Motoaki Seki motoaki.seki@riken.jp*

#### *Specialty section:*

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

*Received: 18 January 2019 Accepted: 23 September 2019 Published: 18 October 2019*

#### *Citation:*

*Ueda M, Matsui A, Watanabe S, Kobayashi M, Saito K, Tanaka M, Ishida J, Kusano M, Seo M and Seki M (2019) Transcriptome Analysis of the Hierarchical Response of Histone Deacetylase Proteins That Respond in an Antagonistic Manner to Salinity Stress. Front. Plant Sci. 10:1323. doi: 10.3389/fpls.2019.01323*

Acetylation in histone and non-histone proteins is balanced by histone acetyltransferase and histone deacetylase (HDAC) enzymatic activity, an essential aspect of fine-tuning plant response to environmental stresses. HDACs in *Arabidopsis* are composed of three families (RPD3-like, SIRT, and HD-tuins). A previous study indicated that class I (HDA19) and class II (HDA5/14/15/18) RPD3-like family HDACs control positive and negative responses to salinity stress, respectively. Furthermore, quintuple *hda5/14/15/18/19* mutants (*quint*) exhibit salinity stress tolerance, suggesting that *hda19* suppresses the sensitivity to salinity stress present in quadruple *hda5/14/15/18* mutants (*quad*). In the present study, transcriptome analysis of the *quint* mutant was conducted to elucidate the hierarchical control of salinity stress response operated by RPD3-like family HDACs (HDA5/14/15/18/19). The analysis identified 4,832 salt-responsive genes in wild-type (Col-0), *hda19-3*, *quad*, and *quint* plants and revealed that 56.7% of the salt-responsive genes exhibited a similar expression pattern in both the *hda19-3* and *quint* plants. These results indicate that deficiency in HDA19 has a bigger impact on salinity stress response

than in class II HDACs. Furthermore, the expression pattern of genes encoding enzymes that metabolize phytohormones raises the possibility that a drastic change in the homeostasis of phytohormones, such as abscisic acid, brassinosteroid, and gibberellin, may contribute to increasing stress tolerance in *hda19-3* and *quint* plants. Among these phytohormones, abscisic acid accumulation actually increased in *hda19-3* and *quint* plants, and decreased in *quad*, compared with wild-type plants. Importantly, 7.8% of the salt-responsive genes in *quint* plants exhibited a similar expression pattern in *quad* plants, suggesting that some gene sets are regulated in an HDA5/14/15/18-dependent manner.

**324**

The transcriptome analysis conducted in the present study revealed the hierarchical and independent regulation of salt stress response that is mediated through HDA19 and class II HDACs.

Keywords: high salinity stress, epigenetics, histone acetylation, histone deacetylases, stress response

#### INTRODUCTION

Recent studies have demonstrated that a diverse type of histone modifications, such as acetylation, methylation, and phosphorylation, play a pivotal role in orchestrating plant response to environmental stresses (Kim et al., 2015). Acetylation levels are balanced by histone acetyltransferases (HATs) and histone deacetylases (HDACs). These enzymes have the ability to write or erase an acetylation mark, respectively. In general, the mark is positively correlated with mRNA expression. Notably, the mechanism by which histone acetylation governs plant response to environmental stresses through HDACs has been elucidated (Asensi-Fabado et al., 2016; Luo et al., 2017; Ueda et al., 2018b).

Three families of HDAC proteins (Reduced potassium dependency 3 (RPD3)-like, Silent Information Regulator 2 (SIRT), and HD-tuins) have been recognized in plants. The *Arabidopsis thaliana* genome encodes 18 genes that are members of the three HDAC families: 12 RPD3-like family proteins; 2 SIRT family proteins; and 4 HD-tuin family proteins. The RPD3-like family is further divided into three classes (I, II, and IV) based on their homology to yeast HDACs (Hollender and Liu, 2008). The RPD3-like family HDACs are pharmacological targets because their inhibition has potential application as a cancer treatment (Bolden et al., 2006; Seto and Yoshida, 2014). The potential value of their inhibition not only applies to cancer therapy but may also be applicable for increasing environmental stress tolerance in plants. *Arabidopsis* and cassava plants treated with HDAC inhibitors exhibit an increase in tolerance to salinity stress (Sako et al., 2016; Patanun et al., 2017; Nguyen et al., 2018). Pharmacological analysis indicated that inhibition of class I RPD3-like HDACs is essential for increasing salinity tolerance based on the evaluation of the survival of plants treated with class-selective inhibitors and subjected to salinity stress conditions (Ueda et al., 2017).

Consistent with the pharmacological studies, *Arabidopsis* plants deficient in class I RPD3-like HDACs (HDA6, HDA9, and HDA19) exhibit increased tolerance to environmental stresses, including different accessions of Col-0. A previous study reported that *hda19* plants (*hda19-3* and *hda19-5*) in the Col-0 background exhibit tolerance to salinity, drought, and heat stress (Ueda et al., 2017; Ueda et al., 2018a). In *hda19-3* plants, positive regulators, such as *NAC019* and *ABI5* in the ABA signaling pathway, that contribute to enhanced environmental stress tolerance are strongly upregulated (Ueda et al., 2017). HDA6 and HDA9 negatively regulate the expression of drought stress tolerance-related genes (class I RPD3-like HDAC) (Zheng et al., 2016; Kim et al., 2017). HDA9 is also involved in improving salt stress tolerance (Zheng et al., 2016). Collectively, the data indicate that inhibition of class I RPD3-like HDACs plays an essential role in improving environmental stress tolerance and that they function as a negative regulator of increased tolerance to environmental stresses, with the exception of the hypersensitive phenotype of the *hda6* mutant in response to salt stress (Chen et al., 2010). In contrast to the inhibition of class I RPD3-like HDACs, the multiple inhibition of class II RPD3-like HDACs (*quadruple hda5/14/15/18* mutant: *quad*) decreases salinity stress tolerance (Ueda et al., 2017). Thus, these previous studies imply that HDACs participate in an antagonistic manner in response to salinity stress.

Although the different classes of HDACs regulate salinity stress responses in a contrasting manner, the epistasis of *hda19-6* in *quad* has been observed. When a mutation of HDA19 is introduced in *quad*, the mutation converts the saltsensitive phenotype to a salt-tolerant phenotype (Ueda et al., 2017). The details of which quintuple *hda5/14/15/18/19* mutant (*quint*) acquired the tolerance to salinity stress, however, remain unknown. In the present study, a genomewide transcriptional profiling of *quint* was conducted to reveal the molecular mechanism underlying the epistasis of HDA19 against HDA5/14/15/18. The transcriptional profiling identified 4,823 salt-responsive genes in samples of whole seedlings that exhibited significant changes in expression in Col-0, *hda19-3*, *quad*, and *quint* plants. Further analysis indicated that 56.7% of the salt-responsive genes exhibited a similar expression pattern in both *hda19-3* and *quint* plants. Notably, genes for abscisic acid (ABA) biosynthesis were significantly upregulated in *quint* and *hda19-3* plants growing under non-stress and/or salinity stress conditions. In accordance with the transcriptional profiling, their ABA levels increased under salinity stress conditions. This was in addition to the upregulation of genes for transcriptional regulators, such as ABI5, reported in a previous study, suggesting that earlier expression of ABA signaling pathway genes contributes to increasing salinity tolerance in *hda19* and *quint*  plants. The results of the transcriptional profiling also indicate that an alteration of phytohormone homeostasis, in particular ABA, brassinosteroid (BR), and gibberellin (GA), occurs in *hda19* and *quint* plants.

The effects of salinity stress on plant growth could be divided into two types. Initially, osmotic stress has an impact on plant water uptake after onset of the salt stress, and, subsequently, specific ion toxicities cause to reduce cellular metabolism (Munns et al., 2006). Here, we mainly evaluate the effect of osmotic stress on plant growth because 5-day-old plants are subjected to higher salinity stress conditions (125 mM NaCl). The results indicate that 7.8% of the salt-responsive genes exhibit a similar expression pattern in *quint* and *quad* plants. Among them, the expression of *NAC016*, which is involved in leaf senescence (Kim et al., 2013), is strongly induced in *quad* plants and is considered to play a major role in producing the salt stress-sensitive phenotype. The upregulation of *NAC016*, however, was detected in *quint* mutant plants that exhibit similar tolerance to salinity stress to *hda19* plants, suggesting the presence of an HDA5/14/15/18 regulatory network independent from HDA19. The transcriptional profile conducted in the present study identified a set of genes that were regulated counteractively in *quad* and *quint* plants, which resulted in the enrichment of genes encoding enzymes involved in stress sensitivity (e.g., *IPT7* and *ISOPENTENYL TRANSFERASE 7*). The mode of action of HDA19 and HDA5/14/15/18 in salinity stress response and where crosstalk occurs between them during salinity stress response are discussed.

#### MATERIALS AND METHODS

#### Plant Material and Growth Conditions

*Arabidopsis thaliana* (L.) Heynh. (Columbia: Col-0), *hda19-3*, *hda5*/*14*/*15*/*18* (*quad*) mutant, and *hda5*/*14*/*15*/*18*/*19* (*quint*) mutant plants were used in this study. *quint* has the *hda19-6* allele. Details pertaining to the wild-type and mutant plants have been previously reported (Ueda et al., 2017). The T-DNA insertional mutants for the generation of the listed mutants were obtained from ABRC (Samson et al., 2002; Alonso et al., 2003). After surface sterilization with sodium hypochlorite, followed by two rinses with distilled water, seeds were floated on 1 ml liquid media (half-strength MS medium with 0.5% MES and 0.1% agar, pH 5.7) for 48 h at 4°C in 24-well tissue culture plates (TPP; Trasadingen, Switzerland). After germination, the seedlings in the 24-well tissue culture plates were placed on a cultivation rack at 22°C under a long-day photoperiod (16h:8h light/dark cycle) at 70–90 μE m− 2 s− 1. When plants were 5 days old (counted after seeds had germinated), the treated plants in each well were administered 25 μl 5 M NaCl (125 mM NaCl per well), while the controls received no salt treatment. Extraction of ABA, proline, and RNA from salt-treated and non-treated (control) plants was conducted for the following analysis at 2 h after the salt treatment was administered. In the evaluation of salinity tolerance in the *hda19-3*, *quad*, and *quint* mutants, 100 mM NaCl treatments were applied to 5-day-old plants (counted after seeds had germinated) in liquid culture and the percentage survival determined 4 days later (three biological replicates consisted of 10 plants; means ± SD). Plants with green true leaves were counted as a survival. Significant differences between the survival values of the experimental plants, relative to untreated or NaCl-treated wild-type plants, were determined using Student's *t* test (\*P < 0.05 and \*\*P < 0.01).

#### Microarray Analysis

Total RNAs were extracted from 5-day-old plants using an RNeasy Plant Mini Kit (Qiagen, Valencia, CA). All RNAs were further purified by incubation with RNase-free DNase I (Qiagen), according to the manufacturer's instructions. RNAs were reverse transcribed into cDNAs using 400 ng of total RNA. cDNA was labeled with a single color (Cy3) using a Quick Amp labeling kit (Agilent Technologies, Palo Alto, CA, USA) and hybridized to an *Arabidopsis* custom microarray (GEO array platform: GPL19830, Agilent Technologies) (Nguyen et al., 2015). Arrays were scanned with a microarray scanner (G2505B, Agilent Technologies). The resulting microarray data were deposited in and are available on the GEO website (GEO ID: GSE121225). The R program ver. 3.2.3 was used for the analysis of the microarray data. The fluorescence intensities of the microarray probes were normalized by quantile normalization using the limma package (Smyth, 2004). Genes with a significant change in expression were selected using the following criteria: an expression log\_2 ratio >0.5 and a controlled *p* value (false discovery rate, FDR; Benjamini and Hochberg, 1995) from a *t* test analysis <0.05. For construction of the heat map, a *Z* score was computed for each of the selected genes using gplots. For constructing the hierarchical cluster, pairwise distances between the expression data of significant expression changed genes were calculated using the "Euclidean" method and hierarchical clustering on this distance matrix was constructed using the "Ward" method. The resultant was divided into 10 groups by using the "cutree" function in R.

#### RT-qPCR Analysis

First-strand cDNA was synthesized from 250 ng total RNA with random primers. ReverTra Ace (TOYOBO) was used for the reverse transcription reaction according to the manufacturer's instructions. Transcript levels were assayed using THUNDERBIRD SYBR qPCR Mix (TOYOBO) and a StepOnePlus Real-Time PCR System (Applied Biosystems) according to the manufacturer's protocols. Gene-specific primers were designed using the PrimerQuest tool (http://sg.idtdna.com/primerquest/ Home/Index). Melting curve analyses were conducted to validate the specificity of the PCR amplification. At least three biological replicates were used in each reverse transcription–quantitative PCR (RT-qPCR) assay. *MON1* was used as a reference gene to normalize data because this gene has been demonstrated to be one of the best reference genes for *Arabidopsis* (Remans et al., 2008). The microarray data also support the use of this reference gene under the experimental conditions used in this study. The RT-qPCR scores and relevant primers are listed in **Supplementary Tables S1** and **S2**, respectively.

#### ABA Measurement

Endogenous ABA was extracted with 80% (*v*/*v*) acetonitrile containing 1% (*v*/*v*) acetic acid from whole wild-type and mutant seedlings after freeze drying. Hormone contents were determined using a UPLC-MS/MS system consisting of a quadrupole/timeof-flight tandem mass spectrometer (Triple TOF 5600, SCIEX, Concord, Canada) and a Nexera UPLC system (Shimadzu Corp., Kyoto, Japan) as described previously (Kanno et al., 2016).

#### Proline Measurement

Quantification of proline content was conducted as described in Kusano et al. (2007). Briefly, each frozen sample was extracted with a concentration of 25 mg fresh weight (FW) of whole seedling per milliliter of extraction medium (methanol/

chloroform/water, 3:1:1 *v*/*v*/*v*) containing 10 stable isotope reference compounds. An equivalent of the 27.8 µg FW of the derivatized sample was injected to Leco Pegasus HT gas chromatography–time-of-flight mass spectrometry (LECO, St. Joseph, MI, USA). Data processing was performed using ChromaTOF version 4.72.0.0 (LECO, St. Joseph, MI, USA). A calibration curve was generated by analyzing the derivatized proline at concentrations of 0.1, 0.5, 1.0, 3.0, 5.0, and 10 ng/µl of injection buffer, respectively. We applied specific ions at *m/z* 142 and 216 to quantify proline content. For normalization, 15 ng/µl of [13C5]-proline (Cambridge Isotope Laboratories, Andover, MA, USA) was used. We chose *m/z* 146 and 220 for it.

#### Statistical and GO Enrichment Analysis

Changes in gene expression and ABA and proline accumulations were statistically analyzed with one-way ANOVA from data obtained from three or four biological replicates. P < 0.05 was considered as significant. Gene ontology (GO) analyses of upregulated genes in *hda19* (Group 1 in **Figure 1C**) and *quad* (Group 2 in **Figure 1C**) based on results obtained from the above microarray analysis were carried out using PANTHER (Mi et al., 2013).

### Accession Numbers

Sequence data from this article can be found in the *Arabidopsis* Genome Initiative or GenBank/EMBL databases under the following accession numbers: *ABA2*; AT1G52340, *ABI5*; AT2G36270, *GA 2ox7*; AT1G50960, *GA 20ox1*; AT4G25420, *IPT7*; AT3G23630, *MON1*; AT2G28390, *NAC016*; AT1G34180, *NCED4*; AT4G19170, *UGT73C5*; AT2G36800.

# RESULTS

#### Global Transcriptome Analysis in Response to Salinity Stress

Under salinity stress conditions, the *hda19-3* and *quad* mutant plants showed a higher and a lower survival rate than wildtype (Col-0) plants, respectively. Similar to the *hda19-3* mutant plants, the *quint* plants exhibited a comparable survival rate (**Figure 1A**). To understand responses to salt stress and elucidate the details of the mechanism underlying increased tolerance to salinity stress in *quint*, transcriptional profiling using a genome-wide microarray was conducted for *quint* (*hda19-6*, an allele of *hda19*), *hda19-3*, and *quad* mutant plants, as well as wild-type plants, growing under normal and salinity stress conditions. The analysis identified 4,823 saltresponsive genes in whole seedling samples that exhibited significant expression changes in at least one of the genotypes (one-way ANOVA with Benjamini–Hochberg correction [FDR] < 0.05; **Figures 1B**, **C**). A principal component analysis (PCA) of gene expression was conducted to understand the global expression patterns of the 4,823 salt-responsive genes in the mutant and wild-type plants under salt stress and non-stressed conditions. The results indicated that the *quad*  mutant does not cluster with the wild type anymore after salt stress (open circles in **Figure 1B**). Furthermore, the clustering between *hda19-3* and *quint* is very similar between the control (filled circles in **Figure 1B**) and salt (open circles in **Figure 1B**) conditions. These results indicate that the global expression pattern in the *quint* mutant in response to salinity stress was similar to the response of the *hda19-3* mutant.

The heat map illustrates that there are significant differences in the transcriptional accumulation of the 4,823 salt-responsive genes under non-stress and salinity stress conditions in wildtype plants and each mutant genotype. A hierarchical cluster analysis revealed that differentially expressed genes in each condition can be divided into ten groups (Group 1: 729, Group 2: 281, Group 3: 903, Group 4: 183, Group 5: 607, Group 6: 498, Group 7: 304, Group 8: 239, Group 9: 193, and Group 10: 881 genes; **Figure 1C** and **Supplementary Tables S3–S12**). The lack of HDAC (*HDA19* or *HDA5/14/15/18*) expression had no significant effect on the expression of the collective 1,401 genes in Groups 3 and 6. The remainder of the 4,823 salt-responsive genes was affected under non-stress and/or salt stress conditions by the genetic defects. As indicated by the PCA, *hda19* is clearly pleiotropic over the *quad*. A total of 56.7% of the salt-responsive genes (Groups 1, 2, 5, 8, and 10) exhibited a similar expression pattern in both *hda19-3* and *quint* plants. In contrast, only 7.8% of the salt-responsive genes (Groups 4 and 9) exhibited a similar expression pattern in *quad* and *quint* plants. Notably, Group 7 enriched genes were expressed additively in *quint* plants relative to *hda19-3* and *quad* plants, suggesting that *HDA19* and *HDA5/14/15/18* redundantly act on mRNA expression. Importantly, however, HDA19 and HDA5/14/15/18 mainly regulate a set of genes in response to salinity stress in an independent manner.

#### Upregulation of ABA Signaling Pathway-Related and ABA Metabolism-Related Genes That Are Independently Regulated by HDA19 in *hda19-3* and *quint* Mutants

In general, histone deacetylases catalyze the deacetylation of histone tails, leading to DNA compaction tightening, inducing a transcriptional inactivation; hence, one would expect that mutation of HDACs has the opposite effect: transcriptional activation. GO enrichment analysis revealed that the majority of the genes in Group 1 (**Figure 1C**) containing upregulated genes under salinity stress conditions in *hda19* and *quint* compared with wild-type plants belongs to GO:0009611 (response to wounding), GO:0009414 (response to water deprivation), GO:0009651 (response to salt stress), and GO:0009737 (response to abscisic acid; **Supplementary Table S13**). Consistent with our previous study, transcription factors such as ABI5, which are considered to be a major positive regulator in the ABA signaling pathway, were strongly upregulated in both *hda19-3* and *quint* plants subjected to salinity stress conditions (Group 1 in **Figure 1C** and **Figure 2A**; **Supplementary Table S1**). In addition to the transcriptional factors involved in ABA signaling, the transcriptional profiling analysis revealed that the ABA biosynthesis genes *ABA2* and

conditions. (A) The survival rate (in percent) of each plant was evaluated 4 days after treatment with 100 mM NaCl or without NaCl (means ± SD; *n* = 3, where each biological replicate was a collection of 10 plants). *P* values were calculated using Student's *t* test (\**P* < 0.05 and \*\**P* < 0.01). (B) Principal component analysis (PCA) based on whole-genome transcriptome analysis. *Filled* and *open circles* reflect the expression profiles under normal and salinity growth conditions, respectively. (C) Hierarchical cluster analysis of salt-responsive genes. The genome-wide mRNA profiles determined by microarray analysis were obtained from 5-day-old plants treated with or without 125 mM NaCl for 2 h. Transcript data were generated from three biological replicates. The heat map represents the *Z* score, with *bars* showing values from −2 to 2. *Red* represents upregulated genes, while *blue* represents downregulated genes. Genes with a significant change in expression were selected using the following criteria: an expression log\_2 ratio >0.5 and a controlled *p* value (FDR; Benjamini and Hochberg, 1995) from a *t* test analysis <0.05.

*NCED4* were also upregulated in *quint* and/or *hda19-3* plants under non-stress and/or stress conditions (**Figures 2A**, **B** and **Supplementary Table S14**). Their upregulation was also confirmed by RT-qPCR (**Figure 2A** and **Supplementary Table S1**). In contrast, the upregulation of *ABA2* and *NCED4* was not observed in the transcriptional profiling analysis of *quad* plants (**Figure 2A** and **Supplementary Table S1**). Consistent with the transcriptome analysis, relative to wildtype plants, the ABA levels increased in *hda19-3* and *quint*

plants. Decreased accumulation of ABA in *quad* under salinity stress conditions was detected, although our transcriptome analysis could not find any signs of the decline to the best of our knowledge (**Figure 2C** and **Supplementary Table S1**). Thus, it appears that HDA19 exerts its anti-stress effect through its regulation of ABA signaling, which operates in a class II HDAC-independent manner.

ABI5 functions at the core of ABA signaling, and the induction of ABI5 expression causes growth arrest and the induction of

some of the LATE EMBRYOGENESIS ABUNDANT (LEA) proteins, such as RAB18, COR6.6, COR15A, EM1, and EM6, that prevent protein aggregation resulting from water loss at all stages of plant growth (Skubacz et al., 2016). Only the induction of *COR6.6* was almost significant (*p* < 0.058) under normal growth conditions and was clearly significant under salinity stress conditions (*p* < 0.016). Nonetheless, *EM6* was one of the direct targets (Finkelstein and Lynch, 2000), while the expression of the remaining four *LEA* genes did not increase significantly (**Supplementary Table S15**). These data suggest that an atypical activation of ABA signaling, coordinated by an absence of *HDA19* expression, may occur in *hda19* and *quint* plants.

To characterize the phenotype of mutants, we measured the accumulation of proline, known to act as a compatible osmolyte to counteract salinity stress (Szabados and Savoure, 2010). Salinity stress induced proline accumulation in all lines, and, under salinity stress conditions, there were significant declines in proline accumulation in *hda19-3* and *quint* plants compared with wild-type and *quad* plants (**Figure 2D**).

#### Altered Expression of Genes Encoding Enzymes Related to Phytohormone Biosynthesis or Catabolism

Phytohormones play an important role in fine-tuning abiotic stress response in plants (Qin et al., 2011). GA is believed to counteract the effect of ABA. Thus, fine-tuning the ABA/ GA ratio is essential to regulate both plant development and plant response to environmental stress (Vanstraelen and Benkova, 2012). Microarray data indicated that the expression of genes involved in GA biosynthesis was altered in *hda19-3* and *quint* plants. *GA2ox7*, a GA catabolic gene that degrades bioactive GA under salinity stress conditions (Magome et al., 2008; Colebrook et al., 2014), was significantly suppressed in *hda19-3* and *quint* plants under salinity stress conditions (**Figures 3A**, **C** and **Supplementary Table S1**). The expression of *GA20ox1*, which encodes a key oxidase enzyme in the biosynthesis of GA (De Vleesschauwer et al., 2012), was significantly upregulated in *hda19-3* and *quint* plants relative to wild-type plants under non-stress conditions (**Figures 3A**, **C** and **Supplementary Table S1**).

BR/ABA antagonism is known to play a role in balancing growth and stress response in plants (Clouse, 2016). BR homeostasis appeared to be altered in *hda19-3* and *quint* plants. UGT73C5, which catalyzes BL-23-*O*-glucosylation and a subsequent reduction in BR activity (Poppenberger et al., 2005), was significantly upregulated under salinity stress conditions (**Figures 3B**, **C** and **Supplementary Table S1**). These data indicate that the homeostasis of phytohormones, in particular ABA and ABA-antagonistic phytohormones such as GA and BR, may be altered in *hda19-3* and *quint* plants.

#### Presence of an HDA19-Independent Pathway and a Pathway Counteractive to HDA19 Coordinated by Class II HDACs

We have not detected any explainable genes from Group 2 (**Figure 1C**) containing upregulated genes under salinity stress conditions in *quad* compared with wild-type plants by the GO enrichment test (**Supplementary Table S16**). However, *NAC016*, whose expression is involved in senescence (Kim et al., 2013), was strongly induced in *quad* plants (Ueda et al., 2017). *Arabidopsis* plants overexpressing *NAC016* (*ANAC016-OX*) rapidly turn white when subjected to either salt or oxidative stress (Kim et al., 2013). Consistent with a previous study (Ueda et al., 2017), the *quad* plants in the current study were sensitive to salinity stress (**Figure 1A**). Unexpectedly, the significant induction of *NAC016* was retained in *quint* plants, although *quint* plants exhibited an elevated level of salinity tolerance that is similar to *hda19-3* plants (**Figure 4A**, Group 9 in **Figure 1C**, **Supplementary Table S1**). This suggests that there is an independent mechanism regulating *NAC016* expression in HDA5/14/15/18 (*quad*) plants that does not involve HDA19.

Transcriptional analysis detected a set of genes that were regulated counteractively in *hda19* alleles (*hda19-3* and *quint*) and *quad* (Groups 2 and 8 in **Figure 1C**; **Supplementary Tables S4** and **S10**) plants, which could account for the discrepancies in the upregulation of *NAC016* and the salttolerant phenotype in *quint*. Deletions of *HDA5/14/15/18* activate the expression of the cytokinin (CK) biosynthetic gene, *ISOPENTENYL TRANSFERASE7* (*IPT7*), resulting in increased levels of CK (Yanai et al., 2005) (**Supplementary Table S4**). Overproduction of endogenous cytokinin and decreased levels of CK have been reported to decrease and increase salt tolerance, respectively (Nishiyama et al., 2012; Wang et al., 2015). Downregulation of *IPT7* was observed in the current study by the introduction of the *HDA19* mutation into *quad* plants under salinity stress conditions (**Figure 4A** and **Supplementary Table S1**). This counteractive expression of *IPT7* may reflect the opposite salt sensitivity and salttolerant phenotypes observed in *quad* and *quint* plants, respectively (**Figure 4B**).

# DISCUSSION

ABA signaling is one of the major phytohormonal pathways responsible for increasing tolerance to a variety of environmental stresses, including salinity and drought (Larosa et al., 1987; Mantyla et al., 1995; Bari and Jones, 2009; Vishwakarma et al., 2017). The current study indicated that ABA signaling is strongly induced in *hda19* mutant plants. There is no clear morphological difference between *quad* and wild-type plants. On the other hand, *hda19-3* and *quint* showed lower germination rates under non-stress conditions, although statistical significance was not detected (**Figure 1A**). A wide range of developmental abnormalities such as growth inhibition was observed in *hda19-3* (Hollender and Liu, 2008). Some of them might be explainable by the enhanced acclimation of ABA. The transcriptome analyses revealed that an HDA19 defect primarily activates the ABA signaling pathway, which is independently regulated from class II HDACs. ABA signaling plays a pivotal role in the growth reduction of plants during the early phase of plant response to salinity stress (Ismail et al., 2014). In contrast, deficiencies in class II HDAC genes appear to result in a malfunction in the growth reduction because expansin 12 (a cell wall-loosening protein; **Supplementary Table S4**) (Lee et al., 2001) and TRH1 (K+ transporter for root hair elongation; **Supplementary Table S4**) (Desbrosses et al., 2003) are activated in response to salinity stress conditions in *quad* plants as in the case of *IPT7* (**Figure 4**). In *hda19-3* and *quint* plants, the upregulation of these genes could not be detected (**Supplementary Tables S4** and **S10**). The HDA19 defect, in which HDA19 is not produced,

same as shown in Figure 2A. (B) A simplified model for the counteractive and hierarchical control of the expressions of *NAC016* and *IPT7* genes mediated through HDA19 and HDA5/14/15/18 deficiencies.

may induce an earlier expression of genes involved in the ABA signaling pathway, including ABA metabolism and other transcriptional factors.

BR homeostasis appeared to be altered in *hda19* and *quint* plants. UGT73C5, which catalyzes BL-23-*O*-glucosylation leading to reduced BR activity (Poppenberger et al., 2005), was significantly upregulated under salinity stress conditions (**Figure 3C**). Under the experimental conditions used in the current study, ABA signaling may be activated due to the reduced accumulation of active BR, which results in increased salinity stress tolerance. In support of this premise, *ABI5* expression has been reported to directly suppress BR signaling through the BRASSINAZOLE RESISTANT 1 (BZR1) transcription factor (Yang et al., 2016). The upregulation of *ABI5* expression under salinity stress conditions may indicate the reduced accumulation of active BRs in *haa19* and *quint* plants. A previous study also indicated that BR signaling is involved in the suppression of GA biosynthetic genes, such as *GA20ox* and *GA3ox3*, and the induction of a GA inactivation gene, such as *GA2ox*, which leads to the suppression of GA signaling (De Vleesschauwer et al., 2012). In the current study, the transcriptome analysis indicates that GA signaling may be activated in *hda19* and *quint* plants because suppression of *GA2ox7* and the induction of *GA20ox1* were detected (**Figure 3C**). Considering the above data, a reduced accumulation of active BRs may trigger a change in the expression pattern of genes encoding phytohormonecatabolizing enzymes.

In contrast to ABA metabolism, a defect in GA metabolism generally enhances tolerance to abiotic stresses, such as salinity and drought stress (Achard et al., 2006; Colebrook et al., 2014). The results of the present study indicate that GA metabolism is activated in *hda19* and *quint* plants based on the transcriptome analysis (**Figure 3C**). These plants, however, exhibit tolerance to salinity stress. Furthermore, *hda19* plants also exhibit tolerance to drought stress (Ueda et al., 2018a). If the hormone network is altered in *hda19-3* or *quint* plants, the altered hormone network (coincidental activation of antagonistic hormone signaling of ABA and GA) may allow plants to alleviate growth arrest induced by ABA signaling. Notably, *ga2ox7* mutants that accumulate active GA exhibit less growth retardation than Col wild-type plants of *Arabidopsis* subjected to high-salinity stress (Magome et al., 2008). ABA stimulates generally an increase in proline accumulation (Szabados and Savoure, 2010). ABA increased in *hda19* and *quint*; nonetheless, a lower proline accumulation was found in them than in wild-type plants and *quad* (**Figure 2D**). The unusual response in proline accumulation might be due to the alteration in phytohormone homeostasis under salinity stress conditions. Further studies are needed to reveal how epigenetic regulation mediated by HDA19 controls stress responses and crosstalk between phytohormones.

We confirmed the altered mRNA expressions of *ABI5*, *ABA2*, *NCED4*, *GA 2ox7*, *GA 20ox1*, *UGT73C5*, *NAC016*, and *IPT7* genes under normal and/or salinity growth conditions by microarray and RT-qPCR analysis. Among them, we found tissue- or organspecific expressions, except for *ABA2* and *GA 2ox7* genes, by *Arabidopsis* eFP Browser (Winter et al., 2007) (**Supplementary Figure S1**). In the study, we have not revealed whether their alterations are due to the derepression of their tissue-specific expression or activation through the inhibition of HDAC activity. A detailed analysis of their expression is needed to understand how HDAC inhibition alters salinity stress response.

HDA19 mainly appears to mediate histone H3 acetylation (Zhou et al., 2005; Krogan et al., 2012). In class II HDACs, HDA5 participates in histone H3 acetylation (Luo et al., 2015). HDA15 participates in histone H3 and H4 acetylation (Liu et al., 2013; Gu et al., 2017). HDA14 is located in chloroplasts, and its localization to chromatin has not been documented (Alinsug et al., 2012). Synergistic upregulation of HDA19 and HDA5/14/15/18 deficiencies observed in *quint* plants may be due to the redundancy in histone or non-histone acetylation coordinated by both HDA19 and class II HDACs. As shown in **Figure 4A**, the expression of *NAC016* is perhaps independent of HDA19, as the majority of the responses described in the manuscript seem to be otherwise. In the case of HDA19 and HDA5/14/15/18-dependent regulation (HDA19- and HDA5/14/15/18-dependent expressions, categories 1, 2, 5, 8, and 10 and categories 4 and 9, respectively, in **Figure 1C**), levels in the acetylation of histones around target sites or proteins, including non-histones, may be independently coordinated by these HDACs. In actuality, multiple acetylations of non-histone proteins have been documented using HDAC inhibitors or mutants (Hao et al., 2016; Hartl et al., 2017). For example, it remains unclear whether the altered expression of genes for BR, GA, and other hormone pathways is due to the HDA19 action or a reflection of the effect on ABA signaling. Further analysis to determine whether HDACs directly regulate the gene expression of candidate signaling pathways through histone acetylation and/or are involved in the acetylation of key enzymes involved in hormonal crosstalk, as highlighted in the current study, will elucidate the detailed processes in stress responses in plants that are coordinated by HDA19 and HDA5/14/15/18.

#### AUTHOR CONTRIBUTIONS

MU and MSek designed the experiments. MU, AM, SW, MKo, MT, and JI conducted the experiments. MU, AM, SW, MKo, MKu, KS, and MSeo analyzed the data. MU, AM, SW, MKo, MKu, MSeo and MSek wrote the manuscript.

# FUNDING

This work was supported by grants to MU and MSek from the RIKEN and grants to MSek from Japan Science and Technology Agency (JST) [Core Research for Evolutionary Science and Technology (CREST, grant number JPMJCR13B4)], and KAKENHI on Innovative Areas (grant no. 16H01476, 18H04791, and 18H04705) of the Ministry of Education Culture, Sports and Technology of Japan.

#### ACKNOWLEDGMENTS

The authors would like to show their appreciation to Ms. C. Torii, Ms. K. Mizunashi, and Ms. A. Sato for their technical support. They are also grateful to Ms. S. Nakae, Ms. K. Kaneko, and Ms. F.

#### REFERENCES


Sakai in the support unit for Bio-Material Analysis and RIKEN CBS Research Resources Division for technical help with the nucleotide sequencing analyses.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.01323/ full#supplementary-material

FIGURE S1 | eFP diagrams of *ABI5*, *ABA2*, *NCED4*, *GA 2ox7*, *GA 20ox1*, *UGT73C5*, *NAC016*, and *IPT7* genes


and RAB18 proteins in *Arabidopsis thaliana. Plant Physiol.* 107, 141–148. doi: 10.1104/pp.107.1.141


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

*Copyright © 2019 Ueda, Matsui, Watanabe, Kobayashi, Saito, Tanaka, Ishida, Kusano, Seo and Seki. 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.*