Histone Modifications Form Epigenetic Regulatory Networks to Regulate Abiotic Stress Response^{1  [OPEN]}

BACKGROUND

Sessile land plants have evolved the ability to respond to a variety of environmental cues. Numerous research reports have identified the molecular components (transcription factors, transporters, signal transduction, and so on) that function in regulating stress response in plants. More recently, a reversible epigenetic regulatory system has been reported to have the capability of orchestrating genomic, transcriptional, translational, and metabolic responses for different biological processes (Kouzarides, 2007; Choudhary et al., 2014; Holoch and Moazed, 2015; Kinnaird et al., 2016). In general, the mode of action of epigenetic regulation is highly conserved among eukaryotes. Therefore, epigenetic regulation represents a fundamental component of omnifarious biological processes (Allis and Jenuwein, 2016). Research on the epigenetic regulation of abiotic stress response has revealed a portion or a complete picture of plant stress response that is coordinated by epigenetic elements (Kim et al., 2015; Asensi-Fabado et al., 2017; Luo et al., 2017). Histone modifications, histone variants, chromatin remodeling, regulatory RNAs (such as noncoding RNA), and DNA methylation are all elements of epigenetic regulation (Goldberg et al., 2007). This review focuses mainly on recent progress made in understanding the role of histone modifications in abiotic stress response.

In fact, the existence of an epigenetic regulatory complex responsible for abiotic stress response has been demonstrated. The analysis of histone modifications began in the middle of the 20^th century. The hypothesis of a positive relationship between transcriptional activation and increased levels of histone acetylation and methylation was raised by Allfrey et al. (1964). Numerous studies conducted since then have supported his original premise and have identified other posttranslational modifications (PTMs), such as phosphorylation and ubiquitination in transcriptional regulation in Arabidopsis (Arabidopsis thaliana; Bergmüller et al., 2007; Zhang et al., 2007; Mahrez et al., 2016; Füssl et al., 2018). Histone acetylation and methylation are now recognized as two types of important and ubiquitous epigenetic marks in gene expression (Xu et al., 2017). Initially, research on isolating enzymes that function as the “writers” and “erasers” of these modifications was vigorously pursued. Writers are enzymes that catalyze the addition of chemical moieties onto histone tails or core domains and include acetyltransferases (HATs), methyltransferases, kinases, and ubiquitinases. Erasers are enzymes that remove these modifications and include deacetylases (HDACs), phosphatases, demethylases (HDMs), and deubiquitinases (Fig. 1; references in Xu et al., 2017).

In general, histone acetylation marks (especially H3 and H4 acetylation) increase the DNA access due to the neutralization of the basic charge in histones, which results in weakening, with a few exceptions, the interaction of histones with DNA (cis effects; Allis and Jenuwein, 2016; Onufriev and Schiessel, 2019). Modifications involving histone methylation in Arabidopsis represent both repressive (symmetric H4R3me2, H3K9me2/3, and H3K27me3) and active marks (asymmetric H4R3me2, H3K4me3, and H3K36me2/3; Fig. 1; Liu et al., 2010; Wang et al., 2016). In contrast to acetylation, histone methylation retains the electron charge of Lys and has no impact on the electrostatic properties of histone proteins. The mode of action (trans effects) of histone methylation mark is probably coordinated through hydrophobicity; however, this premise is not conclusive, and other possibilities have been suggested (Musselman et al., 2012). The presence or absence of methylation of Lys and/or Arg amino acids in histones alters their association with reader proteins, leading to modifications in chromatin structure that result in either transcriptional repression or activation (Teperino et al., 2010). Various domains that recognize both unmethylated and methylated Lys or Arg residues have been reported, including ADD (ATRX-DNMT3-DNMT3L), ankyrin, bromo-adjacent homology, chromo-barrel, chromodomain (CD), double chromodomain, malignant brain tumor, plant homeodomain (PHD), Pro-Trp-Trp-Pro (PWWP), SAWADEE, tandem Tudor domain, Tudor, WD40, and the zinc finger CW (Fig. 2; Taverna et al., 2007; Bannister and Kouzarides, 2011; Allis and Jenuwein, 2016; Andrews et al., 2016; Xu et al., 2017). Documentation of the affinity of these elements to Lys or Arg methylation in plants, however, still remains elusive. The listed reader proteins appear to be highly conserved in a broad range of eukaryotic organisms; however, EMSY-LIKE1 H3K4me2/3 reader protein containing a plant-specific single Tudor domain was identified in Arabidopsis (Zhao et al., 2018). Four domains capable of recognizing acetylated Lys have been identified in human (Homo sapiens) cells (Musselman et al., 2012). Among these domains, genes encoding proteins containing bromodomain or tandem-PHD have been identified in Arabidopsis (Xu et al., 2017). Enzymatic activity of proteins having a tandem-PHD domain, however, has never been demonstrated in any homologs encoded in the Arabidopsis genome. These Arabidopsis homologs appear to lose enzymatic activity because they lack the key residues responsible for recognizing histone acylation, including acetylation (Zhao et al., 2018). Thus, the collective data indicate that epigenetic elements have diversified during the evolution of plant-specific lineages.

Monoubiquitination of H2A (H2Aub) and H2B (H2Bub) is also considered to be both an active and repressive mark for transcription in eukaryotes (Fig. 1). H2AK121 monoubiquitination in Arabidopsis tends to be colocalized with H3K27me3 in an independent manner but not cooperatively with POLYCOMB REPRESSIVE COMPLEX2 (PRC2), which is needed to maintain H3K27me3 in Arabidopsis (Bratzel et al., 2010; Zhou et al., 2017b). Deubiquitination of H2B is required for DNA methylation and heterochromatic histone H3 methylation (Sridhar et al., 2007). More specifically, H2B monoubiquitination activates transcription through H3K4me3 deposition (Geng et al., 2012).

Over the past 30 years, dozens of studies identified fast (half-life of a few minutes) and slow turnover rates (half-life of ∼30 min) in all four core histones from yeast (Saccharomyces cerevisiae) to plant and animal cells (Waterborg, 2002; Zheng et al., 2013). The turnover rate of histone acetylation is much faster than that of methylation in HeLa cells (Zee et al., 2010). Although it remains unclear if it applies to plant histones, these data suggest that rapid response to an ever-changing environment (i.e. abiotic stress) via acetylation could serve as a beneficial adaptation. On the other hand, it is possible that methylation might have an advantage for a long-period response, as observed in the case of flowering and inheritance of transgenerational stress memory (priming) due to their slower half-life. In accordance with this concept, previous studies have identified various epigenetic regulators for acetylation involved in abiotic stress response (Asensi-Fabado et al., 2017; Luo et al., 2017). On the other hand, epigenetic regulations in longer term responses, such as flowering, stress memory, and stress priming via histone and/or DNA methylation with other PTMs such as Ser-5P Pol II in some cases (Ding et al., 2012), have been reported (out of scope in this review, see references Sani et al., 2013; Lämke et al., 2016; Schuettengruber et al., 2017; Annacondia et al., 2018; Friedrich et al., 2019; Liu et al., 2019).

The reversible property of these epigenetic modifications involves the cooperative action of readers, writers, and erasers (Xu et al., 2017). The identification of the epigenetic components that participate in stress response and elucidating the molecular network that they coordinate is essential to fully understand epigenetic regulation. How epigenetic regulators are recruited to specific locations in chromatin to carry out their function is one of the most interesting aspects of understanding the molecular mechanism responsible for the activation of stress response in plants. TF (transcriptional factor)-mediated, long noncoding RNA-mediated, and self-targeting models have all been proposed to act as recruiters for the localization of epigenetic regulators to specific chromatin sites (Deng et al., 2018). In abiotic stress response, recruiters associated with stress response have been categorized into a TF-mediated model. For example, rice (Oryza sativa) INDETERMINATE SPIKELET1 (IDS1) and Arabidopsis MYB96 have been identified as HDAC recruiters in salinity and drought stress response, respectively (Cheng et al., 2018; Lee and Seo, 2019), poplar (Populus trichocarpa) ABSCISIC ACID (ABA)-RESPONSIVE ELEMENT BINDING PROTEIN1 (AREB1) as a HAT recruiter in drought stress response (Li et al., 2019), and rice OsbZIP46CA1 (OsbZIP46) as both an H2B ubiquitinase and deubiquitinase recruiter in drought stress response (Ma et al., 2019).

In the past decade, increased evidence on the interaction between several epigenetic components localized in chromatin has accumulated, which has increased our understanding of the epigenetic regulation in abiotic stress. Recent progress in the identification of epigenetic modifiers, such as erasers, readers, writers, and recruiters, involved in abiotic stress response in flowering plants is covered in this review (Table 1; Fig. 3).

Components of histone modifications that alter the abiotic stress response phenotype in flowering plants

RECENT ADVANCES IN UNDERSTANDING THE INVOLVEMENT OF EPIGENETIC REGULATORS IN ABIOTIC STRESS RESPONSE

Role of HATs in Drought, Salinity, and Heat Stress Response in Arabidopsis, Chinese Cabbage (Brassica rapa), Poplar, Rice, and Tomato (Lycopersicon peruvianum)

The integrated activity of HATs and HDACs regulates acetylation levels, and recent studies have documented their role in abiotic stress response. The Arabidopsis genome encodes 12 HAT genes, representing four HAT families (GENERAL CONTROL NONDEREPRESSIBLE5 [GCN5]-like [GCN5/HISTONE ACETYLTRANSFERASE OF THE GNAT FAMILY {HAG}1, 2, 3], MYST-like [HISTONE ACETYLTRANSFERASE OF THE MYST FAMILY (HAM)1, 2], p300/CBP [CREB binding protein]-like [HISTONE ACETYLTRANSFERASE OF THE CBP FAMILY {HAC}1, 2, 4, 5, 12], and TAF_II250-like [HISTONE ACETYLTRANSFERASE OF THE TAF_II250 FAMILY {HAF}1, 2]; Pandey et al., 2002; Earley et al., 2007). As described below, their functional role has been demonstrated in recent studies on GCN5 in plants subjected to salinity and heat stress.

The potential involvement of GCN5 in salt stress response was first characterized in maize (Zea mays) roots. The up-regulation of cell-wall-related genes, such as ZmEXPANSIN B2 and ZmXYLOGLUCAN ENDOTRANSGLUCOSYLASE/HYDROLASE1, is associated with an increase in H3K9 acetylation in both the promoter and coding regions of genes. The acetylation is thought to be necessary for a response to high-salinity conditions to occur in maize roots. The up-regulation of ZmEXPANSIN B2 and ZmXYLOGLUCAN ENDOTRANSGLUCOSYLASE/HYDROLASE1 genes has been suggested to be mediated by two HAT genes (ZmHATB and ZmGCN5), since mRNA expression of these HAT genes increases under salt stress conditions (Li et al., 2014). In support of this suggestion, mRNA expression of GCN5 in Arabidopsis is activated in response to salinity stress, and gcn5 mutants exhibit increased sensitivity to salinity stress due to a defect in cell wall integrity. CTL1, which encodes a chitinase-like (CTL) protein, is a direct target of GCN5 and plays a pivotal role in cell wall biosynthesis and salt stress tolerance. GCN5 activates CTL1 expression through H3K9/K14 acetylation (Zheng et al., 2019). Notably, gcn5 mutant plants also exhibit serious defects in thermotolerance in response to heat stress. GCN5 appears to positively regulate thermotolerance through the enrichment of H3K9/K14ac in the promoter regions of HEAT SHOCK TRANSCRIPTION FACTOR A3 and ULTRAVIOLET HYPERSENSITIVE6 genes (Hu et al., 2015). The occurrence of nonenzymatic acetylation reactions is possible; however, to what extent nonenzymatic acetylation occurs in the nucleosome in plants is unclear (Choudhary et al., 2014). Recent progress identified a large number of acetylation sites from nonhistone proteins (references therein Füssl et al., 2018), and some were found to have a potential role in abiotic stress response. Importantly, these data do suggest that enzymatic acetylation, mediated through HAT activity, is indispensable for abiotic stress tolerance in Arabidopsis. Although it should be noted, especially in regards to histone H3 acetylation, that GCN5/HAG1 primarily acetylates H3 and only marginally acetylates H4 or H2A/B in vitro (Earley et al., 2007)

In addition to Arabidopsis, expression analysis of HAT genes (rice, 12 OsHAT genes; Chinese cabbage, 15 BraHAT genes) has been conducted in rice under drought stress conditions and in Chinese cabbage under salinity and drought stress conditions, in which hyperacetylation of histone H3 was detected (Fang et al., 2014; Eom and Hyun, 2018). In both crops, significant alterations of expression during exposure to each stress has been detected. In rice, drought stress increased mRNA expression of OsHAC703, OsHAG703, OsHAM701, and OsHAF701 (Fang et al., 2014). In Chinese cabbage, drought stress repressed the expression of BraHAC5, whereas it strongly activated that of BraHAC7, BraHAG2, and BraHAG5. In the case of BraHAG2, an opposite expression pattern was observed between 2 d and 4 d after the drought stress treatments. Salinity stress has been shown to activate 13 BraHATs, with the exception of BraHAG4 and BraHAG6, at several time points (5 h, 1 d, 2 d) after the stress treatments (Eom and Hyun, 2018). These data suggest that HATs play a fundamental role in abiotic stress response in not only Arabidopsis but also crops as well, although the details in crops still remain unknown at this time.

HAC1 is a member of the p300/CBP-like family and lacks a bromodomain motif in plants (Pandey et al., 2002), suggesting that HAC1 needs a recruiter for nuclear localization. In tomato, the heat stress transcription factor CLASS B HEAT SHOCK FACTOR B1 (HsfB1), interacts with HAC1 and is subsequently recruited to chromatin, raising the possibility that HsfB1 may help to maintain and/or restore the expression of certain viral or housekeeping genes during extended periods of heat stress (Bharti et al., 2004). The association of HsfB1 with HAC1 appears to function as a recruiter to control in the level of histone acetylation.

Instead of the use of a transcription factor to recruit a histone acetyltransferase (AtHAC1) to chromatin, a synthetic strategy can also be used. Specifically, this strategy involves using the catalytically inactive form of clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein9 (Cas9; dCas9) as a recruiter of AtHAC1 and is an effective way to activate a drought-stress-tolerance-related gene in Arabidopsis. The fusion of dCas9 with HAT enzyme (dCas9^HAT) enables locus-specific activation of a drought-stress-tolerance-related gene in Arabidopsis, thereby enhancing drought stress tolerance. In plants expressing dCas9^HAT, this fusion is designed to be recruited and localized at the locus encoding the AREB1, a key positive regulator of drought stress response. RD29A expression is positively regulated by AREB1, and this results in an enhancement of drought stress tolerance (Roca Paixão et al., 2019). In poplar, the formation of an AREB1-ALTERATION/DEFICIENCY IN ACTIVATION 2B-GCN5 ternary complex is required for the activation of NAC genes involved in increasing drought stress tolerance through H3K9 acetylation (Li et al., 2019). ALTERATION/DEFICIENCY IN ACTIVATION 2B is a transcriptional coactivator of GCN5-containing complexes (Kaldis et al., 2011). AREB1-targeted dCas9^HAT may have a stronger impact on histone acetylation than the internal recruiting system, although it is unclear whether the ternary complex is conserved in Arabidopsis. Furthermore, ABA-dependent multisite phosphorylation controls the transcriptional activity of AREB1 (Furihata et al., 2006). In addition, the phosphorylation status of AREB1 within the ternary complex is an interesting component underlying the molecular mechanisms of abiotic stress response.

The p300/CBP-like and MYST-like family contain PHD and CD reader domains, respectively, which have the potential to recognize methylated lysines (PHD[H3K4me2/3 and H3K9me3]; CD[H3K9me2/3 and H3K27me2/3]). As mentioned below, at least H3K4 methylations are involved in abiotic stress response (Musselman et al., 2014; Andrews et al., 2016). A proposed model suggests that multiple enzyme complexes containing several types of histone- and non-histone-modifying activities work in concert with other chromatin remodeling machines to regulate gene transcription (Strahl and Allis, 2000). It is possible that the p300/CBP-like family might be involved in the coordination of stepwise regulation mediated through a multiple enzyme complex, including acetylation and methylation activity, to activate stress response. Phylogenetic trees show that the p300/CBP-like family forms the largest gene family in HAT, which might indicate functionally redundant roles between them (Pandey et al., 2002; Eom and Hyun, 2018). Further analyses are warranted to uncover the role of the p300/CBP-like family in abiotic stress response.

Roles of HDACs in Cold, Drought, Salinity, and Heat Stress Response in Arabidopsis, Brachypodium (Brachypodium distachyon), and Rice

HDACs are categorized into zinc-dependent and NAD (NAD(+)) types based on their catalytic domain. The REDUCED POTASSIUM DEPENDENCY3 (RPD3)-like and the SILENT INFORMATION REGULATOR2-like gene families are zinc dependent and NAD(+) dependent HDACs, respectively. The RPD3-like family is further divided into four classes (class I [HISTONE DEACETYLASE {HDA}6, 7, 9, 19], class II [HDA5, 14, 15, 18], an unclassified class [HDA8, 10, 17], and class IV [HDA2]), based on their homology to yeast HDACs. Plants have also evolved a plant-specific HDAC (HD-tuin) family (TYPE-2 HDAC(HD2)A-D; Bolden et al., 2006; Hollender and Liu, 2008; Seto and Yoshida, 2014; Ueda et al., 2017). Recent studies have indicated that several class I and II RPD3-like family genes and HD-tuin family proteins are involved in cold, drought, heat, and salinity stress response.

HDA9 and HDA19 negatively regulate salt stress tolerance (Mehdi et al., 2016; Zheng et al., 2016; Ueda et al., 2017), while HDA6, HD2C, and HD2D positively regulate salinity tolerance (Chen and Wu, 2010; Chen et al., 2010; Luo et al., 2012; Han et al., 2016). Regarding hda9, enrichment of histone H3K9 acetylation at promoters of 14 genes was observed among a collection of randomly selected genes that respond to water deprivation stress in wild-type plants and was detected in response to both salt and drought stress. Their up-regulation in hda9 appears to contribute to enhanced salinity stress tolerance (Zheng et al., 2016). HDA19 controls ABA signaling by binding to the chromatin of ABA receptor genes (PYRABACTIN RESISTANCE1 [PYR1]-LIKE [PYL]4, PYL5, and PYL6) through H3K9ac, thus affecting the expression level of ABA receptor genes when its activity is altered (Mehdi et al., 2016). In hda19-3 (Col-0 background), plants significantly accumulate ABA in young seedlings (Ueda et al., 2019). These data suggest that HDA19 seems to be linked to the control of ABA signaling in several steps.

Genetic analysis has revealed that HDA6 and HD2C cooperatively regulate salt stress response through the expression of ABA-responsive genes, such as ABSCISIC ACID INSENSITVE1 and ABSCISIC ACID INSENSITVE2. Their physical interaction has been confirmed, and mRNA expression of these ABA-responsive genes is activated through increased levels of H3K9K14 acetylation in single and double hd2c/hda6 mutants (Luo et al., 2012). In rice, the overexpression of HDA705, a homolog of Arabidopsis HDA6 or HDA7 (Fu et al., 2007), decreased ABA levels and salt stress tolerance during seed germination (Zhao et al., 2016). Plants expressing AtHD2D, whose substrate is considered to be histone H3K27ac (Lee and Cho, 2016), accumulate malondialdehyde more slowly and exhibit a more gradual increase in electrolyte leakage compared to wild-type plants, both of which are indications of increased tolerance to abiotic stresses, including drought, salt, and cold stress (Han et al., 2016). Notably, hda19 mutant plants also exhibit increased tolerance to multiple abiotic stresses relative to wild-type plants (Ueda et al., 2018a). Details of the epigenetic molecular mechanism responsible for HD2D and HDA19 increases in tolerance to multiple stresses are still unknown. Thus, further research will be needed to uncover the details related to the specific mechanism.

In contrast to salt stress response, hda6 and hd2c exhibit an opposite response to cold stress (acclimation). An analysis of hda6 and hd2c mutants indicated that the mutants exhibited decreased and increased freezing tolerance relative to cold-treated, wild-type plants, respectively (To et al., 2011; Park et al., 2018). Why hda6 and hd2c would exhibit different phenotypes in response to cold versus salinity stress is still unclear. A switch in the interactor of HD2C, however, may provide a possible explanation for their contrasting activity, since HD2C interacts with HDA9 in addition to HDA6 (Park et al., 2019). A growth-stage- or stress-type-specific interactome analysis may provide additional insights into the mechanism underlying the role of these HDACs to different abiotic stresses.

The HDACs mentioned above play a pivotal role in drought and heat stress response. The interaction between Arabidopsis HD2C deacetylase and a BRAHMA (BRM)-containing SWITCH/SUC NONFERMENTING chromatin remodeling complex has been confirmed. HD2C is considered to be a subunit of the chromatin remodeling complex. Notably, hd2c and brm plants exhibit a better recovery after stress treatments as indicated by measurements of rosette diameter of Arabidopsis plants subjected to heat stress. Analysis of mutants suggests that HD2C and BRM act in a common genetic pathway to negatively regulate heat-stress-responsive genes (Buszewicz et al., 2016). hda6 plants were hypersensitive to heat exposure (Popova et al., 2013). On the other hand, defects in HDA6 enhance tolerance to drought stress. HDA6 specifically binds to genes whose expression is activated by the jasmonic acid signaling network induced as a response to water deficit (Kim et al., 2017). In contrast, plants deficient in HDA15, a class II HDAC, exhibit reduced ABA sensitivity and enhanced sensitivity to drought stress. HDA15 forms a complex with MYB96, and this complex binds to the promoters of a subset of RHO GTPASE OF PLANTS (ROP) genes, namely ROP6, ROP10, and ROP11, and represses their expression through deacetylation of H3K9K14ac and H4K5K8K12K16ac from cognate regions, particularly in the presence of ABA (Lee and Seo, 2019).

In addition to Arabidopsis, the role of HDACs in drought stress response has been reported in the monocot plant Brachypodium. Overexpression of BdHD1, a HDA19 homolog, causes Brachypodium plants to exhibit a hypersensitive-to-ABA phenotype and better survival under drought conditions. In contrast, BdHD1 RNA-interference plants are insensitive to ABA and exhibit low survival under drought stress conditions (Song et al., 2019).

Role of Histone Methyltransferases in Dehydration, Drought, and Salinity Stress Response in Arabidopsis

There is a greater number of regulators for histone methylation than acetylation, which may consequently serve as a potential system for fine-tuning stress response. Arg and Lys residues are methylated by different proteins, namely Arg methyltransferases (PRMTs) and histone Lys methyltransferases (HKMTs), respectively. While previous genetic studies in eukaryotes indicated that histone acetylation sites function redundantly and have cumulative effects on transcriptional expression in eukaryotes (Yun et al., 2011), current research indicates that each methylation site clearly has a unique role in transcriptional regulation. Two Arg methylation sites (H3R17 and H4R3) and five Lys methylation sites (H3K4, H3K9, H3K27, H3K36, and H4K20) have thus far been identified in plants (Liu et al., 2010; Pontvianne et al., 2010). The Arabidopsis genome encodes 41 genes for SET (SET: Su(var)3-9, E(z), and Trithorax) domain proteins (or 49 genes for putative SET domain-containing proteins). SET domain proteins are putative candidates for five classes of HKMTs and nine PRMT genes (Liu et al., 2010; Pontvianne et al., 2010). Target sites for each HKMT and PRMT have been described: H3K4 (SET DOMAIN GROUP [SDG]4/8, ARABIDOPSIS TRITHORAX [ATX]1/2/3/4/5) methylation; H3K9 (SU(VAR)3–9 HOMOLOGS [SUVH]1/2/3/4/5/6/7/8 and SU(VAR)3–9 RELATED [SUVR]1/2/4/5) methylation; H3R17 (AtPRMT4a/4b) methylation; H3K27 (ATXR5/6, SWINGER, MEDEA, and CURLY LEAF) methylation; H3K36 (SDG4/8/25/26) methylation; H4R3 (AtPRMT1a/1b/5/10) methylation; and H4K20 (SUVH2) methylation. Among them, the biochemical activity of AtPRMT1a/1b/4a/4b/5/10, ATX1/2/5/6, SUVH1/4/5/6, SUVR4, and SDG8/25 HKMTs has been demonstrated (references in Liu et al., 2010; Pontvianne et al., 2010). ATX4/5 and AtPRMT5 methylases are involved in drought stress tolerance (Fu et al., 2013; Liu et al., 2018). AtPRMT5 appears to regulate salt stress response via methylation of nonhistone proteins, in addition to histone methylations (Zhang et al., 2011).

ATX1 mediates H3K4me3 (Pien et al., 2008), and the atx1 loss-of-function mutant exhibits sensitivity to dehydration stress. This sensitivity is attributed to its effect on the expression of several stress-responsive genes, including 9-cis-EPOXYCAROTENOID DIOXYGENASE3; suggesting that ATX1 regulates plant response to dehydration and osmotic stress (Ding et al., 2011). ATX4 and ATX5 play an essential role in drought stress response through their function as an active mark (H3K4me3). Single and double mutants of ATX4 and ATX5 exhibit increased tolerance to drought stress relative to wild-type plants and significantly lower levels of H3K4me3. Deficient expression of these genes decreases the expression of ABA-HYPERSENSITIVE GERMINATION3 (AHG3), which encodes a phosphatase 2C protein, an essential negative regulator of ABA signaling. ATX4/5 redundantly regulate AHG3 expression through their direct binding to the AHG3 locus and the enrichment of H3K4me3 (Liu et al., 2018).

ATX1 and ATX5 have both PHD and PWWP reader domains, whereas the PWWP domain is not contained within ATX4 (Pontvianne et al., 2010). As mentioned above, these methylases have been proven to regulate abiotic stress response through the active histone mark H3K4me3. The PHD reader domain recognizes H3K4me2/3 and H3K9me3, whereas PWWP recognizes H3K36me3, H4K20me1/3, and H3K79me3 (Musselman et al., 2014; Andrews et al., 2016). Arabidopsis lacks the H3K79me3 mark and appears to have lost a homolog H3K79 methyltransferase Dot1, which is found in mammals (references in Roudier et al., 2011). In addition to H3K4me3, the multiple site recognition of PHD and PWWP reader domains suggest that an additional mark might be involved in abiotic stress response. In particular, the role of H4K20me1/3 is poorly understood in plants and has not yet been investigated in these atx mutants. It is reasonable to consider that additional analyses might uncover a novel methylation site that is involved in abiotic stress response.

AtPRMT5/SHK1 BINDING PROTEIN1 (SKB1)/CALCIUM UNDERACCUMULATION1 (CAU1) and AtPRMT10 mediate symmetric and asymmetric dimethylation of H4R3, respectively. Notably, symmetric and asymmetric H4R3me2 are considered as repressive and active marks, respectively (references in Liu et al., 2010; Bobadilla and Berr, 2016). Additionally, cau1 and skb1-2 (atprmt5-2) and atprmt5 alleles provide increased tolerance to drought stress. The repressive mark (symmetric H4R3me2) decreases under NaCl- or ABA-treated conditions (Zhang et al., 2011). Drought stress also decreases the accumulation of AtPRMT5/SKB1/CAU1 proteins (Fu et al., 2018), suggesting that the release of the repressive mark plays a pivotal role in the activation of stress-tolerance-related genes. Actually, decreased levels of H4R3me2 are required for the activation of NAC055 whose overexpression contributes to increased drought stress tolerance (Fu et al., 2013, 2018). In contrast, skb1-1 (atprmt5-1) plants exhibit hypersensitivity to salt stress (Zhang et al., 2011). H4R3 is one of the catalytic substrates of PRMT5; however, PRMT5 also mediates methylation of Arg residues of nonhistone proteins such as ARGONAUTE2 and small nuclear ribonucleoprotein Sm-LIKE4 (LSM4; Zhang et al., 2011; Hu et al., 2019). The hypersensitive phenotype of skb1-1 plants is partly explained by methylation of AtPRMT5 and LSM4. HDA6 mediates the acetylation of GSK3-like kinase BR-INSENSITIVE2, which acts as a key negative regulator in the BR signaling pathway (Hao et al., 2016). Therefore, the functional role of nonhistone protein modifications is receiving greater attention. Clearly, the various modifications to nonhistone proteins and their functional impact need to be investigated to better understand the mode of action of erasers and writers in abiotic stress response.

Role of HDMs in Dehydration and Salinity Stress Response in Arabidopsis

HDMs are divided into two classes, Lys-specific demethylases (LSD) and hydroxylation by JumonjiC (JmjC) domain-containing proteins (JMJ). These proteins facilitate the removal of methyl groups from methylated Lys residues in an independent catalytic reaction. The Arabidopsis genome contains 4 LSD and 21 JMJ genes (Liu et al., 2010). Some JMJ proteins appear to have lost demethylase activity. For example, JMJ24, which has E3 ubiquitin ligase activity, ubiquitinated a DNA methyltransferase, CHROMOMETHYLASE3, in vitro and destabilized it in vivo. Therefore, its demethylase activity remains unclear (Deng et al., 2016). The target sites of LSD and JmjC are follows: H3K4 (FLOWERING LOCUS D, LSD 1-LIKE1/2, and JMJ14/15/17/18) demethylation; H3K9 (IBM1/JMJ25) demethylation; H3K27 (JMJ11/EARLY FLOWERING6, JMJ12/RELATIVE OF EARLY FLOWERING6, JMJ/30/32) demethylation; and H3K36 (JMJ30) demethylation, although their substrates are still disputable (Liu et al., 2010; Xiao et al., 2016; Huang et al., 2019). To date, JMJ15 and JMJ17 demethylases have been shown to function in salinity and dehydration stress response, respectively (Shen et al., 2014; Huang et al., 2019).

JMJ15 can only demethylates H3K4me3 (Liu et al., 2010; Xiao et al., 2016). Gain-of-function mutants (jmj15-1 and jmj15-2) exhibited increased tolerance to salinity stress, while a loss-of-function mutant (jmj15-3) exhibited increased sensitivity to salinity stress. Overexpression of JMJ15 down-regulates many genes that are preferentially marked by H3K4me3 and H3K4me2; however, the direct targets responsible for increased tolerance to salinity stress that are JMJ15 dependent have not yet been identified (Shen et al., 2014).

Notably, jmj17 loss-of-function mutants display dehydration stress tolerance and ABA hypersensitivity in regards to stomatal closure. JMJ17 specifically demethylates H3K4me1/2/3 and directly binds the promoter and gene body of OPEN STOMATA1 (OST1). This suggests that OPEN STOMATA1 mRNA abundance is regulated by H3K4me3 demethylation, and thus the latter modulates dehydration stress response (Huang et al., 2019).

Role of Histone Phosphorylation and Ubiquitination in Arabidopsis and Rice

The induction of H3 phosphorylation in response to abiotic stress has been observed in cultured cells, however, the specific molecular mechanisms of the response are not clearly understood. An Arabidopsis mutant defective in two paralogues of MUT9-LIKE KINASE1/2, which encode closely related Ser/Thr protein kinases, exhibits pleiotropic phenotypes, including dwarfism and hypersensitivity to osmotic (Polyethylene glycol) and salt stress. The double mutant has reduced global levels of H3T3ph, while polyethylene glycol (drought-like) treatments increase the repressive mark in wild-type plants (Wang et al., 2015).

Monoubiquitination of histones H2A and H2B has been generally detected in eukaryotes, and its role in abiotic stress response is gradually being discovered. Earlier studies revealed that H2B monoubiquitination (H2Bub) regulates stress response in Arabidopsis and rice. The Arabidopsis genome contains genes for two RING E3 ligases (HISTONE MONOUBIQUITINATION [HUB]1/2) and three E2 conjugases (UBIQUITIN CARRIER PROTEIN [UBC]1/2/3) for histone H2B monoubiquitination (Cao et al., 2008). Further research indicated that hub1 and hub2 mutants exhibit a loss of H2Bub and a sensitivity to salinity stress phenotype. H2Bub1 regulates salt-stress-induced depolymerization of microtubules. Furthermore, the PTP-MPK3/6 signaling module is responsible for integrating the signaling pathways that regulate microtubule stability, which is critical for plant salt stress tolerance (Zhou et al., 2017a). LONG-CHAIN ACYL-COA SYNTHETASE2, ABERRANT INDUCTION OF TYPE THREE1, and HOTHEAD, which are involved in cutin biosynthesis, and ECERIFERUM1, which is involved in wax biosynthesis, are downregulated in hub1 and hub2 mutants. Consequently, this down-regulation results in an alteration of wax composition and reduces cutin 16:0 dicarboxylic acid (Ménard et al., 2014). Thus, H2Bub may be required for drought stress tolerance, as the accumulation of waxes has been associated with increased protection against water loss (Patwari et al., 2019). Consistent with this premise, the ectopic expression of AtHUB2 increases histone H2B monoubiquitination and enhances drought tolerance in transgenic cotton (Gossypium hirsutum; Chen et al., 2019). Furthermore, overexpression of OsHUB2 in rice revealed that H2Bub positively modulates ABA sensitivity and drought resistance. OsHUB2 interacts with OsbZIP46, and this complex regulates ABA signaling. In addition, OsbZIP46 also interacts with MEDIATOR of OsbZIP46 DEACTIVATION and DEGRADATION to form an indirect complex with a putative deubiquitinase, rice OTUBAIN-LIKE DEUBIQUITINASE. OsbZIP46 plays a pivotal role as a recruiter in regulating drought stress response via ABA signaling (Ma et al., 2019).

Considering the research above, it appears that histone phosphorylation and ubiquitination are as essential for stress response in plants as histone acetylation and methylation.

Modification of Histones by Histone Regulators Affect Abiotic Stress Response: An Added Level of Complexity

Each histone modification profile is very dynamic and complex. Adding to this complexity, the cross talk between different epigenetic regulators has been reported (Liu et al., 2014). This level of complexity makes it much more difficult to understand epigenetic systems involved in the coordination of abiotic stress response. For example, HDA19 forms a complex with MULTICOPY SUPRESSOR OF IRA1 (MSI1) to fine-tune salinity stress response in Arabidopsis (Mehdi et al., 2016). MSI1 also connects LIKE HETEROCHROMATIN PROTEIN1 (LHP1) to PRC2. The LHP1-MSI1 interaction functions as a positive feedback loop to recruit PRC2 to chromatin to maintain H3K27me3 during replication (Derkacheva et al., 2013). In humans, ENHANCER of ZESTE HOMOLOG2 (EZH2) is a subunit of PRC2 and the only enzymatic component of the PRC2 complex that catalyzes H3K27me3 (Simon and Lange, 2008). CURLY LEAF and SWINGER in Arabidopsis are considered to be homologs of EZH2 (Hennig and Derkacheva, 2009). Details on the components of Polycomb group proteins have been presented by Förderer et al. (2016). HDAC activity is required for the EZH2-dependent entry of the repressing mark into chromatin. HDA19 deficiency may lead to changes in H3K27me3 deposition, in addition to its effect on the level of histone acetylation. Further research for the role of genes for subunits forming a HDA19-containing complex may provide new evidence on the coordination of abiotic stress response by HDA19, although the role of some subunits such as Arabidopsis ortholog of human SIN3 ASSOCIATED POLYPEPTIDE18 (AtSAP18) and HISTONE DEACETYLASE COMPLEX1 (HDC1) have been discovered in previous studies (Song and Galbraith, 2006; Perrella et al., 2013).

Exposure of potato (Solanum tuberosum) to cold stress induces enhanced chromatin accessibility via bivalent histone modifications (H3K4me3 and H3K27me3) of active genes (Zeng et al., 2019). In fact, SHORT LIFE, a plant-specific reader protein for methylation, has been recently identified that recognizes both active (H3K4me3) and repressive (H3K27me3) marks. The proposed role of this protein is to function as a switch that changes the status of chromatin between an active and repressive mode (Qian et al., 2018). While the specific role of the bivalent mark is unclear, bi- or multivalent marks do appear to coordinate abiotic stress response in plants, as multiple and/or various modifications act in firmly, tolerant, and binary switching mode against reader proteins, which may have an effect on chromatin structure leading to modulation of gene expression (Zhao et al., 2018). Uncovering the cross talk between histone modifications, including corresponding functional eraser and writer enzymes for them and reader proteins on stress response, will be needed to understand the epigenetic machinery for abiotic stress response.

Diversification of histones becomes more complicated in epigenetic regulation (Henikoff and Smith, 2015). For example, histone H2A.Z is required for a strong repressive effect on transcription under drought stress conditions, which may contribute to counteracting unwanted transcription in noninductive conditions in Arabidopsis (Sura et al., 2017). In Brachypodium, H2A.Z containing nucleosomes is essential for maintaining grain yield under thermal stress conditions (Boden et al., 2013). These data suggest that H2A.Z variants play a pivotal role in abiotic stress response in plants; although structural divergences or differences in the PTMs of histone H2A variants are still debatable.

Induced Increases in Abiotic Stress Tolerance by Pharmacological Inhibition of Epigenetic Modifiers

Epigenetic modifiers, including erasers, writers, readers, and recruiters for histone modifications have gained increased interest as potential therapeutic targets in human cancer (Simó-Riudalbas and Esteller, 2015; Prachayasittikul et al., 2017). In addition, they have also been associated with life span extension in yeast and mammals (Mahajan et al., 2011). Notably, small molecules, including cellular metabolites, that act as activators or inhibitors of the SILENT INFORMATION REGULATOR2 and RPD3-like HDAC family of enzymes have been identified (Mahajan et al., 2011; Choudhary et al., 2014). In plants, HDAC inhibitors contribute to increasing tolerance to salinity stress in Arabidopsis and cassava (Sako et al., 2016; Patanun et al., 2017; Ueda et al., 2017, 2018b). Furthermore, metabolites or effectors derived from competitors or pathogens have been reported to have an inhibitory effect on the enzymatic activity of histone modifiers. In brief, they appear to serve as a mediator in the interaction that occurs between organisms in an ecosystem (Ramirez-Prado et al., 2018). For example, the maize genome encodes Hm1 for NADPH-dependent reductase, which inactivates Helminthosporium carbonum (HC) toxin having HDAC activity. The HC toxin causes pathogen virulence on maize by Cochliobolus carbonum race 1. Maize usually exhibits a resistance phenotype to the HC toxin biosynthesized from Cochliobolus carbonum race 1. However, some strains lost the reductase enzymatic activity during breeding, resulting in susceptibility to the HC toxin. (Marla et al., 2018). Further analysis of the published research on this topic and additional research will provide new insights into survival strategies activated by exposure to abiotic stress conditions that involve biological interactions between metabolites and epigenetic regulators in nature.

CONCLUSION

A series of epigenetic elements that modify or regulate abiotic stress response, in particular histone modifications involving acetylation, methylation, ubiquitination, and phosphorylation, have been identified. Transcriptional factors are considered to be a fundamental component in regards to recruiters for erasers or writers. Therefore, the TF-mediated model is the most widely accepted model that has been presented thus far. It is likely that evidence that can be applied to and support the long noncoding RNA model or self-targeting model may be discovered in the near future. Tissue-specific expression is a requisite for stress response. Therefore, discovering how epigenetic modifiers are properly recruited to target loci is essential to increase our understanding of stress responses. High-resolution methods of analysis, such as a single cell analysis, will be needed to reveal the functional role of each histone modification, as most of the currently available data have been obtained from the investigation of a combination of cell and/or tissue types, which may contain a mixture of PTMs from a modification site in different cells.

Pleiotropic effects are observed when the enzymatic activity of epigenetic elements is inhibited because epigenetic factors often form a complex with various proteins and regulate different biological processes via different epigenetic marks. The use of chemical compounds, such as HDAC inhibitors, represents a promising approach to suppress the pleiotropic effects caused by the inhibition of epigenetic element activity and selectively increase tolerance to abiotic stresses. As discussed in this review, recent research on epigenetic factors has provided new potential targets for enhancing tolerance to abiotic stress. Deciphering epigenetic codes and studying molecular mechanisms will identify specific pathways responsible for regulating stress response and provide the ability to manipulate stress response in a more sophisticated and targeted manner, although many challenges and questions still need to be addressed (see Outstanding Questions).

ACKNOWLEDGMENTS

The authors would like to show their appreciation to all lab members for their support.

LITERATURE CITED

Author notes