The Histone-Modifying Complex PWR/HOS15/HD2C Epigenetically Regulates Cold Tolerance.

Cold stress is a major environmental stress that severely affects plant growth and crop productivity. Arabidopsis (Arabidopsis thaliana) HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENE15 (HOS15) is a substrate receptor of the CULLIN4-based CLR4 ubiquitin E3 ligase complex, which epigenetically regulates cold tolerance by degrading HISTONE DEACETYLASE2C (HD2C) to switch from repressive to permissive chromatin structure in response to cold stress. In this study, we characterized a HOS15-binding protein, POWERDRESS (PWR), and analyzed its function in the cold stress response. PWR loss-of-function plants (pwr) showed lower expression of cold-regulated (COR) genes and sensitivity to freezing. PWR interacts with HD2C through HOS15, and cold-induced HD2C degradation by HOS15 is diminished in the pwr mutant. The association of HOS15 and HD2C to promoters of cold-responsive COR genes was dependent on PWR. Consistent with these observations, the high acetylation levels of histone H3 by cold-induced and HOS15-mediated HD2C degradation were significantly reduced in pwr under cold stress. PWR also interacts with C-repeat element-binding factor transcription factors to modulate their cold-induced binding to the promoter of COR genes. Collectively, our data signify that the PWR-HOS15-HD2C histone-modifying complex regulates the expression of COR genes and the freezing tolerance of plants.


Introduction
Cold stress is one of the major environmental factors that seriously limits the growth and productivity of plants.To overcome this constraint, plants have developed effective ways to increase resistance to cold stress and freezing.Cold acclimation is a process that increases freezing tolerance upon exposure to low but non-freezing temperatures.This process involves the activation or expression of cold-regulated (COR) genes, and consequent physiological and biochemical changes in response to cold stress (Guo et al., 2018;Ding et al., 2019).Over the past two decades, various effectors and regulators of stress signaling pathways have been identified (Liu et al., 2018b;Guo et al., 2018;Zhang et al., 2019;Tang et al., 2020).One of the best-characterized mechanisms is the C-repeat element-Binding Factors (CBFs, also known as Dehydration Responsive Element-Binding proteins [DREB], CBF1/DREB1B, CBF2/DREB1C, and CBF3/DREB1A) transcription factordependent cold signaling pathway (Chinnusamy et al., 2007;Guo et al., 2018;Liu et al., 2018a).These CBF transcription factors are APETALA2-like DNA binding domain proteins that bind to the conserved C-repeat element/dehydration responsive element (CRT/DRE) on the promoter regions of COR genes, such as COR15A, COR47, and COR78, and induce the expression of these genes, leading to freezing tolerance (Chinnusamy et al., 2007;Guo et al., 2018;Liu et al., 2018a).
Control of gene expression in the cold signaling pathway is also epigenetically regulated through the structural changes of chromatin (Park et al., 2018a;Ding et al., 2019;Chang et al., 2020).Heterochromatin and euchromatin are reversibly interchanged to repress or activate gene expression, respectively (Chen and Tian, 2007).These structural changes in chromatin are regulated by posttranslational modifications of histone, such as histone acetylation, methylation, ubiquitination, sumoylation, and phosphorylation.These modifications are accomplished by covalent modification of the N-terminal tails of core histones (Nathan et al., 2006;Sridhar et al., 2007;Luo and He, 2020).The structural change of chromatin mediated by histone acetylation is reversible and greatly influences the regulation of gene expression (Chen and Tian, 2007;Clapier and Cairns, 2009).
Histone acetylation catalyzed by histone acetyltransferases (HATs) reduces the charge interaction of histone and DNA, leading to exposure of the DNA and facilitated binding of transcription factors (Verbsky and Richards, 2001;Clapier and Cairns, 2009).Histone acetylation enhances gene transcription, whereas histone deacetylation catalyzed by histone deacetylases (HDACs) leads to chromatin compaction and subsequent gene silencing (Chen and Tian, 2007).
Arabidopsis (Arabidopsis thaliana) HOS15 is a WD40-repeat protein that shares high sequence similarity with human TRANSDUCIN-BETA-LIKE 1 protein (TBL1), an intrinsic component of the repressor silencing mediator for retinoic acid receptor and thyroid hormone receptor/nuclear receptor co-repressor (SMRT/NCoR) protein complex that is involved in histone acetylation (Zhu et al., 2008).In Arabidopsis, POWERDRESS (PWR) is predicted to be a plant NcoR1 homolog (Wang and Brendel, 2004).Using immunoprecipitation (IP) and tandem mass spectrometry (MS) analyses, we found that PWR interacts with HOS15 (Park et al., 2018b).Several reports have described that HOS15 functions in a complex with PWR and histone deacetylases (Park et al., 2018a;Park et al., 2018b;Suzuki et al., 2018;Mayer et al., 2019).In addition, HOS15 has been characterized to regulate the transcription of several genes, including GIGANTEA, a floral promoter in photoperiod-dependent flowering (Park et al., 2019), DA1-RELATED PROTEIN3 (DAR3), a negative regulator of cell proliferation (Suzuki et al., 2018), and WRKY53, a transcription factor involved in leaf senescence (Chen et al., 2016).Moreover, the transcription levels of some target genes of HOS15 were upregulated in pwr mutants (Chen et al., 2016;Suzuki et al., 2018;Mayer et al., 2019;Park et al., 2019).Besides this role as a co-repressor with PWR and histone deacetylases, HOS15 also functions as a substrate receptor for the CULLIN4 (CUL4)-based ubiquitin E3 ligase complex, CRL4.In response to cold stress, HOS15 mediates the degradation of histone deacetylase 2C (HD2C) and switches the chromatin structure from repressive to permissive form, thereby acting as a positive regulator of cold stress (Park et al., 2018a).This facilitates the recruitment of CBFs for the expression of COR genes and development of cold tolerance (Park et al., 2018a).However, the presumable role of PWR as a component of CRL4 ubiquitin E3 ligase in the cold stress signaling has not yet been tested.
In this study, we report that PWR forms a complex with HOS15 and HD2C and that the PWR-HOS15-HD2C complex epigenetically controls freezing tolerance in plants.Indeed, PWR regulates COR gene expression in a similar way as HOS15 does and the loss of function pwr mutant phenocopies the hos15 mutant, with no additive effects in the double mutant.Moreover, PWR is required for the binding of HOS15 and HD2C to the COR gene promoters.The cold-induced acetylation of histone 3 (H3), facilitated by HOS15-mediated HD2C degradation, is greatly reduced in the pwr mutants.Furthermore, PWR directly interacts with CBFs, and coldinduced binding of CBFs to COR chromatin is dependent on PWR.Collectively, our data provide mechanistic insight into how a histone-modifying PWR-HOS15-HD2C complex and CBFs co-regulate cold-responsive COR gene expression to promote freezing tolerance in plants.

PWR and HOS15 function in the same pathway in response to cold
PWR is a homolog of the repressor complex protein NCoR1 of mammals that in Arabidopsis interacts with HOS15 (Park et al., 2018b;Suzuki et al., 2018;Mayer et al. 2019).Loss-of-function mutants hos15-1 (C24) and hos15-2 (Col-0) have a freezing-sensitive phenotype (Zhu et al., 2008;Park et al., 2018a).Based on these facts, we hypothesized that PWR, being a HOS15-binding protein, would also be involved in cold stress response.To investigate this hypothesis, pwr-2 (SALK_071811C) and pwr-3 (SALK_006823) homozygote mutants were isolated (Supplemental Fig. S1).Mutant pwr-2 bears a T-DNA inserted at the second exon of PWR, and pwr-3 has a T-DNA insert at the first intron of PWR (Supplemental Fig. S1A).Both pwr mutants exhibited morphological phenotypes similar to those of the hos15-2 mutant.Similarities included short and blunt-ended siliques, small plant size, shorter hypocotyl lengths, and early flowering phenotypes (Supplemental Fig. S1D, E, F and G) (Yumul et al., 2013;Kim et al., 2016).These common morphological phenotypes support the possibility that PWR and HOS15 proteins function together in the same developmental processes.To observe the phenotypes under freezing stress, 2-week-old seedlings grown on MS agar were exposed to freezing temperature before and after cold acclimation (4°C, 7 days).Under both conditions, pwr mutants were sensitive to freezing compared to wild type (WT) (Fig. 1A; Supplemental Fig. S2A).The survival rates were calculated by counting numbers of seedlings from Fig. 1A and Supplemental Fig. S2A.Survival rates of pwr mutants were significantly lower than those of WT (Fig. 1B, Supplemental Fig. S2B).
Additionally, the freezing tolerance of plants grown in soil was reduced in the pwr mutants compared to that of WT (Fig. 1C, Supplemental Fig. S2C).Consistent with these freezing-sensitive phenotypes, pwr-2, pwr-3, and hos15-2 mutants displayed substantially higher electrolyte leakage than WT at freezing temperatures regardless of cold acclimation conditions (Fig. 1D, Supplemental Fig. S2D).To test the cold stress responses in pwr mutants, we investigated the transcript level of COR genes in pwr-2 and pwr-3 lines.Cold-induced expression of COR genes was significantly decreased in the pwr-2 and pwr-3 mutants compared to that in WT (Fig. 2A).We also examined whether the expression levels of the CBF transcription factors that regulate the expression of cold-responsive genes through direct binding to the promoter regions of COR genes (Chinnusamy et al., 2007) were altered in the pwr mutants.Transcript levels and protein abundance of CBF transcription factors were not changed significantly in the pwr mutants compared to that in WT (Fig. 2A and B).
These results suggested that PWR, similarly to HOS15, contributes to cold stress signaling by regulating the expression of the COR genes, but not of CBFs.Furthermore, we found that the pwr-3 hos15-2 double mutant exhibited the same phenotype as each single mutant (Supplemental Fig. S3), suggesting that HOS15 and PWR regulate cold stress signaling in the same pathway.

PWR is involved in HOS15-mediated degradation of HD2C
To understand the function of the PWR protein, we compared the sequences of Arabidopsis PWR to those of PWR homologs within plant species and of NcoR1 and 2 of mammals.The Swi3, Ada2, N-Cor, TFIIIB (SANT) domains, and the Nterminal nuclear localization signals (NLS) were highly conserved, but the intervening sequences between these functional domains were relatively variable between proteins (Supplemental Fig. S4A and S5).The SANT domains of the Arabidopsis PWR protein are highly homologous to those of human NCoR1 (Kim et al., 2016).However, NcoR has the SANT domains closer to the N-terminus, whereas PWR has them in the C-terminal half of the protein.Outside these conserved domains, alignment analysis of the full amino acid sequence of PWR showed little similarity (13.1% similarity and 7.05% identity) to the NcoR1 protein (Supplemental Fig. S5).
The HOS15 protein acts as a substrate receptor in the CUL4-based E3 ligase complex targeting HD2C in the nucleus for proteasome-dependent degradation (Park et al., 2018a).The NCoR protein, a PWR homolog in animals, is degraded by TBL1, a HOS15 homolog, upon a specific signal by the thyroid hormone (Perissi et al., 2008).Therefore, the amount of PWR protein, a HOS15 interacting protein in plants, was determined in the hos15-2 mutant under cold stress.PWR protein and transcript levels were unchanged by cold stress in the WT (Supplemental Fig. S6A).
Unexpectedly, the amount of PWR protein in the hos15-2 mutant decreased, while the transcript level increased (Supplemental Fig. S6B).Thus, we reasoned that PWR binding to HOS15 did not result in PWR degradation but could instead affect the stability of HD2C, which also interacts with HOS15.Hence, we tested whether coldinduced degradation of HD2C protein (Park et al., 2018a) was facilitated by PWR.
Even though the transcript level of HD2C was not affected in hos15-2 and pwr mutants, cold-induced degradation of the HD2C protein was impaired in pwr mutants, similarly to the hos15-2 mutant (Fig. 3A).The physical interaction of PWR with HD2C was tested using a split-luciferase (LUC) complementation assay, which is based on the reconstituted LUC activity when two proteins fused with N-and C-terminal LUC fragments (NLuc and CLuc) physically interact in vivo (Chen et al., 2008).Nicotiana benthamiana leaves that were transiently co-expressing CLuc-PWR and HD2C-NLuc displayed luminescence signals above background but not as intense as the positive control (Fig. 3B, Supplemental Fig. S7A).The interaction of PWR with HD2C was further tested in a yeast two-hybrid assay.In contrast to the split-luciferase assay (Fig. 3B), the yeast two-hybrid assay did not show an interaction of PWR with HD2C (Fig. 3C).Together, these results suggest that PWR interaction with HD2C is indirect, probably limited by the amount of adaptor protein(s) of plants with no yeast homologs.In line with this hypothesis, a co-immunoprecipitation assay in Arabidopsis showed that the interaction of PWR and HD2C in the WT was diminished in the hos15-2 mutant (Fig. 3D), indicating that HOS15 is required for the interaction of PWR and HD2C.In other words, HOS15 acts as a bridge for HD2C-PWR complex formation.To test whether the interaction between HOS15, HD2C, and PWR was altered by cold stress, the co-IP assay was repeated after treatment with the proteasome inhibitor MG132 to prevent HD2C protein degradation under cold stress, and with cycloheximide (CHX) to prevent de novo protein synthesis.The interaction between HOS15, HD2C, and PWR did not differ significantly with and without cold stress (Fig. 3E).Again, the amount of HD2C co-IPed by PWR was substantially reduced in the hos15-2 mutant background compared to the WT.In these experiments, the amount of pre-existing PWR (input) was always lower in the hos15-2 mutant compared to the WT and hd2c-1 mutant, regardless of the temperature, suggesting that HOS15 may enhance PWR stability.Consistent with the nuclear localization of HOS15 and HD2C (Park et al. 2018a), GFP-fused PWR was predominantly localized in the nucleus (Supplemental Fig. S8).From these results, PWR appears to regulate HD2C stability through HOS15, rather than by direct interaction with the HD2C protein.This is consistent with the demonstrated role of HOS15 as a substrate receptor for the CUL4-based ubiquitin E3 ligase complex under cold stress (Park et al., 2018a).
PWR is required for the association of HOS15 and HD2C to the promoter regions of COR genes HD2C and HOS15 indirectly bind to the CRT/DRE region of COR gene promoters to which CBF proteins associate (Park et al., 2018a).When HD2C is degraded by HOS15 in response to cold stress, histones are acetylated to promote COR gene expression for cold tolerance (Park et al., 2018a).To determine the effect of PWR on the association of HD2C and HOS15 to COR gene promoters, we performed chromatin immunoprecipitation (ChIP) assays with two amplicon regions of COR15A.Amplicon COR15A-II (region II) contains the CBF binding cis element, and COR15A-I (region I) was used as mock sequence (Fig. 4A).HD2C appeared to bind the amplicon region II in the WT (Col-0) at 22°C, and cold treatment reduced the amount of HD2C associated to this region (Fig. 4B, Park et al., 2018a).The association of HD2C to the CBF binding region in the COR15A promoter was significantly lower in the hos15 and pwr single and double mutants under warm temperature compared to the WT, and was reduced further upon cold stress to reach a low level similar to that of the WT (Fig. 4B).HOS15 also bound to the same region of COR15 promoter as HD2C (i.e., COR15A-II), while its association to COR15A-II increased in the cold condition (Fig. 4C).However, the cold-induced association of HOS15 protein to the COR15A-II region was decreased in pwr mutants (Fig. 4C), suggesting that PWR enhances the recruitment of HOS15 to the COR15A promoter under cold, which then results in HD2C degradation.HD2C deacetylates histone H3 (Luo et al., 2012a;Buszewicz et al., 2016).
As PWR indirectly interacts with HD2C and regulates COR gene expression (Fig. 2A, 3B-E), we investigated how PWR is functionally linked with HD2C in terms of chromatin-dependent regulation of COR gene expression.In the WT, the level of acetylated H3 (AcH3) on COR15A and COR78 promoters was increased by cold treatment, as expected from relaxed chromatin (Fig. 4D, Supplemental Fig. S9C).
However, cold treatment failed to induce the acetylation of H3 on COR genes in hos15-2 and pwr plants (Fig. 4D, Supplemental Fig. S9C).Similarly, the levels of H3K9Ac and H3K14Ac, which are regulated by PWR and HD2C (Luo et al., 2012a;Tasset et al., 2018), were significantly increased upon cold in WT, but not in hos15-2 and pwr mutants (Fig. 4E and F).Taken together, these results indicate that the PWR-HOS15 complex regulates the expression of COR genes through HD2C protein stability and histone acetylation in response to cold.

PWR promotes binding of CBF proteins to COR promoter regions
We have shown that HOS15 likely bridges a complex of PWR and HD2C that associates with CRT/DRE regions of COR genes, where the binding of CBF proteins enhances COR gene expression (Fig. 3 and 4, Supplemental Fig. S9B) (Park et al., 2018a;Mayer et al., 2019;Park et al., 2019).To test the interaction of PWR and CBF proteins, yeast two-hybrid, Co-IP, and split-LUC complementation assays were performed (Fig. 5A-C, Supplemental Fig. S7B).The results demonstrated that all three CBF isoforms directly interacted with PWR but not with HD2C.Next, we examined whether HOS15 and PWR affected the binding of CBF proteins to COR gene promoters in response to cold stress (Fig. 5D, Supplemental Fig. S10).Cold treatment (4°C, 24 h) greatly enhanced the binding of CBF proteins to the CRT/DRE regions of COR15A and COR78 in WT (Col-0).However, cold-induced CBF binding was drastically reduced in hos15-2 and pwr mutants (Fig. 5D, Supplemental Fig. S10), indicating that HOS15 and PWR facilitate the binding of CBF to the chromatin of COR genes in response to cold stress.In addition, CBF binding to the CRT/DRE region of COR15 gene promoter was not significantly different between the pwr-3 hos15-2 double mutant and each single mutant plant (Supplemental Fig. S11).
These results suggest that the PWR-HOS15 complex is required for the binding of CBF transcription factors to CRT/DRE regions of COR gene promoters during cold stress.

Discussion
The HOS15-PWR complex modulates various signaling pathways by regulating histone acetylation on target genes Chromatin structure is determined by a number of histone modifications that strongly affect gene expression under developmental programs and in response to environmental stresses (Hollender and Liu, 2008;Clapier and Cairns, 2009).The mechanisms and constituent proteins for chromatin modification are extensively conserved in eukaryotic organisms, including plants, animals, and yeast, which indicate ancient mechanisms that evolved from a common ancestor (Chen and Tian, 2007).Among epigenetic regulations, histone acetylation appears to directly control gene expression as histone deacetylases (HDACs) and histone acetyltransferases (HATs) interact with co-repressors or co-activators to form complexes that regulate the transcription of genes through changes in chromatin properties (Pandey et al., 2002;Hollender and Liu, 2008).
In mammals, this co-repressor complex interacts with HISTONE DEACETYLASE 3 (HDAC3) and negatively regulates gene expression by decreasing the level of histone acetylation of the target gene promoter (Rosenfeld et al., 2006).In Arabidopsis, based on phylogenetic analysis, RPD3-like type Class I of HDACs comprising HDA6, HDA7, HDA9, and HDA19, are most closely related to HDAC3 of mammals (Pandey et al., 2002;Hollender and Liu, 2008;Alinsug et al., 2009).
HOS15 was demonstrated to interact with HDA6, HDA9, and HDA19 (Park et al., 2018a;Mayer et al., 2019;Park et al., 2019).Moreover, HOS15-deficient mutants share a common phenotype with the hda9 mutant, including the small plant size, blunt siliques, early flowering, and short hypocotyls (Park et al., 2018a;Suzuki et al., 2018;Park et al., 2019).The hda9 mutant is also less sensitive to dehydration and salt stress, suggesting that HDA9 negatively affects the salt and drought stress response (Zheng et al., 2016).PWR is a SANT domain-containing protein that interacts with HDA9 and functions as a repressor in flowering, leaf senescence, and thermomorphogenesis (Yumul et al., 2013;Chen et al., 2016;Kim et al., 2016;Suzuki et al., 2018;Tasset et al., 2018;Mayer et al, 2019).The pwr mutants showed morphological phenotypes similar to those of hos15-2 and hda9 mutants (Supplemental Fig. S1) (Yumul et al., 2013;Kim et al., 2016;Park et al., 2019).PWR is required for H3K9 acetylation on thermomorphogenic genes, and the hypocotyl elongation induced by high ambient temperature (Tasset et al., 2018).In addition, both PWR and HDA9 interact with HOS15 (Park et al., 2018b;Mayer et al., 2019;Park et al., 2019).Moreover, the target spectra of the three proteins seem to overlap, and they are involved in ion homeostasis and the response to stresses that mainly include cold, dehydration, abscisic acid (ABA)-mediated response, and oxidative stress (Mayer et al., 2019;Park et al., 2019).Hence, there is ample evidence supporting that PWR interacts with HOS15 and HDA9 to regulate various stress responses.Here, we have shown that a similar complex with a different geometry of partners, namely PWR, HOS15, and HD2C, plays a role in defining chromatin structure at the COR gene promoters, thereby contributing towards cold-induced chromatin remodeling.
The epistatic effects of hos15-2 and hda9 mutations, together with microarray analyses, suggested that HOS15 had additional functions independently of HDA9 (Mayer et al., 2019).HOS15 also interacts with HDA6 and HDA19, which are RPD3like type Class I histone deacetylases (class I HDACs) (Park et al., 2018a).Both HDA6 and HDA19 are involved in ABA-and salt-mediated transcriptional responses, and negatively regulate the expression of stress responsive genes (Chen and Wu, 2010;Zhu et al., 2019).Our findings demonstrate the involvement of the PWR-HOS15 complex in the cold stress signaling pathway by regulating HD2C protein abundance, which appears to be mediated by the substrate receptor activity of HOS15 in the CUL4 E3 ligase complex (Park et al, 2018a).The HD2-type histone deacetylase HD2C functions as a transcriptional repressor by promoting chromatin compaction (Wu et al., 2003).The hd2c loss-of-function mutant exhibits globally increased levels of H3K9K14 acetylation and H3K4 dimethylation, and a decreased level of H3K9 dimethylation (Luo et al., 2012a).HD2C is involved in ABA and abiotic stress responses (Sridha and Wu, 2006;Buszewicz et al., 2016), and the transcript levels of HD2C are reduced upon ABA and salt treatments (Luo et al., 2012a).HD2C-deficient mutants show low germination and survival rates, while HD2C overexpressors are insensitive to ABA, salt, and drought treatments (Sridha and Wu, 2006;Colville et al., 2011;Luo et al., 2012a).HD2C interacts with DNA METHYLTRANSFERASE-2 (DNMT2) (Song et al., 2010) as well as with HDA6 and HDA19 to regulate stress-inducible genes (Luo et al., 2012a, b).Genes encoding ABA INSENSITIVE 3 (ABI3) and ABA receptors PYL4, PYL5, and PYL6, which are positive regulators of ABA signaling, seem to be direct targets of HDA19 (Ryu et al., 2014;Mehdi et al., 2016).Moreover, the expression of DELTA1-PYRROLINE-5-CARBOXYLATE SYNTHASE (P5CS), encoding a rate-limiting enzyme in proline biosynthesis, is enhanced in hda19 (Ueda et al., 2017).In summary, HDACs interact with several co-factors and play a role in various environmental stress responses.
Taken together, these data signify that the PWR-HOS15 module is involved in a variety of stress responses through the interaction with several HDACs.

Plant PWR functions differently from animal NCoR1
SMRT/NcoR1 co-repressor complexes have been well studied in association with nuclear hormone receptors in animals (Oberoi et al., 2011).SMRT/NcoR1 repressor complexes are recruited to ligand-unbound retinoic acid receptor and thyroid hormone receptor, which bind to response elements of target genes.Upon ligand binding, SMRT/NcoR1 and the E1A C-terminal binding protein (CtBP) are degraded by the TBL1XR1-and TBL1X-dependent proteasome, respectively (Perissi et al., 2008).In these processes, TBL1X and TBL1XR1 function as E3 ubiquitin ligases for the recruitment of the ubiquitin-proteasome system to degrade corepressors, leading to activation or transcription of specific nuclear receptorcontrolled genes.By contrast, the plant homolog of TBL1, HOS15, was found to act as a substrate receptor for the CUL4-based ubiquitin E3 ligase to degrade the plantspecific histone deacetylase HD2C rather than degrading PWR, the plant homolog of NcoR (Fig. 3A, Supplemental Fig. S6B) (Park et al., 2018a).Our finding that PWR interacts indirectly with HD2C through HOS15 is consistent with the cold-induced degradation of HD2C being reduced in pwr mutants, similarly to hos15-2 (Fig. 3).Surprisingly, HOS15 seems to enhance the stability of PWR.As exemplified by the co-IP experiments shown in Figure 3D-E and Supplemental Figure S6B, the amount of pre-existing PWR before treatments with the inhibitors of protein synthesis and degradation, CHX and MG132, were always lower in the hos15-2 mutant compared to the WT and hd2c-1 mutant, regardless of the temperature.Presumably, PWR gains stability through the interaction with HOS15.The low overall sequence similarity between of PWR and NcoR proteins may explain their analogous functions through distinct molecular mechanisms.Only two SANT domain regions (SANT1 and SANT2) of PWR share conserved primary sequence with NcoR (52% and 63% similarity, respectively) (Kim et al., 2016), whereas the full amino acid sequence of PWR shares only 13.1% similarity and 7.05% identity with the full peptide sequence of NcoR protein.The SANT1 and SANT2 domains of NcoR are necessary for HDAC activation and the binding to unmodified histone tails, respectively (Yu et al., 2003).
However, the SANT2 domain of PWR does not bind to unmodified histone, but only to the modified histone (Kim et al., 2016).Mammalian NCoR1 has three repressor domains next to the SANT domain for the repression activity, but those domains are not found in PWR.Thus, the plant proteins PWR and HOS15 seem to have similar functions to the mammalian homologs but different operating mechanisms.

HOS15-PWR complex is a co-activator with the CBF transcription factors
CBF transcription factors play a predominant role in promoting COR gene expression and the adaptation to severe cold stress (Chinnusamy et al., 2007;Liu et al., 2018a).Neither PWR nor HOS15 affected the transcriptional induction of CBF genes upon cold treatments, but these proteins physically interacted with all three CBF proteins (Fig. 2, 5A, B and 5C) (Park et al., 2018a).Our results indicate that the PWR-HOS15 complex is strongly associated with the CTR/DRE elements on the COR gene promoters, which are the binding elements for the CBF transcription factors (Fig. 4C, Supplemental Fig. S9B) (Park et al., 2018a).In addition, CBF association with the COR genes was greatly reduced in the hos15-2, pwr-2, and pwr-PWR-HOS15 complex interacts with the CBF transcription factors and that the complex is required for the association of CBF with the promoter region of the COR genes to induce expression, at least in part by promoting HD2C degradation and chromatin aperture upon cold stress.Thus, the PWR-HOS15 complex can be considered a CBF-dependent co-activator.
In ada2b and gcn5 mutants, which are deficient in histone acetyl-transferases, COR gene induction by cold stress is delayed and their expression levels are reduced (Vlachonasios et al., 2003).However, the ada2b mutant (Ws-2 ecotype) showed increased freezing tolerance, indicating that ADA2b may function in the repression of freezing tolerance through histone acetylation (Vlachonasios et al., 2003).Interestingly, despite the observation that the hos15-1 mutant (C24 ecotype) had a freezing-sensitive phenotype, the expression of the COR genes was increased.
Thus, the cold-signaling mechanism in ada2b mutants in the Col-0 background would be interesting to study, to clarify the differences among these ecotypes.In addition, GENERAL CONTROL NONDEREPRESSIBLE 5 (GCN5), which displays histone acetyl-transferase activity, is recruited by CBF1 through the transcriptional co-activator ADA2 to increase the expression of COR genes (Mao et al., 2006).Here, we show that COR gene expression was reduced in hos15-2, pwr-2, and pwr-3 mutants during cold acclimation.In these mutants, H3 acetylation levels at the COR gene promoter regions were reduced (Fig. 2A, Fig. 4D, Supplemental Fig. S9C).In addition, HD2C protein stability was negatively regulated by the PWR-HOS15 complex.We suggest that the PWR-HOS15 complex is important for the formation of an epigenetic complex to regulate histone acetylation and deacetylation.
Subsequently, this altered chromatin structure affects the expression of COR genes during cold acclimation.
Epigenetic regulation plays an important role in many aspects of biological processes.In this study, we have discovered a complex composed of PWR, HOS15, HD2C, and CBF transcription factors that regulate COR genes expression and cold tolerance.Our earlier report (Park et al. 2018a) and the present findings have been incorporated into the model shown in Figure 6.In normal conditions, PWR-HOS15 forms a complex with HD2C by which the COR chromatin is hypoacetylated and the COR gene expression is repressed.In response to cold stress, PWR-HOS15 recruits CUL4 to form a CRL4 HOS15 complex for HD2C degradation, resulting in the enhanced acetylation of H3 on the COR chromatin.Consequently, the COR chromatin is  (Park et al., 2018a).

Freezing tolerance assay and electrolyte leakage measurements
The freezing tolerance assay was performed as previously described (Yang et al., 2010;Ji et al., 2015;Park et al., 2018a) with some modifications.Briefly, 2-week-old WT plants and mutants grown on the 1/2 MS 0.9% (w/v) agar medium containing 1.5% (w/v) sucrose (pH 5.7) under long-day conditions at 23 ± 1°C were untreated or cold acclimated by incubation at 4 ± 1°C for 7 days.Plants were then placed in a RUMED 4001 freezing chamber adjusted to -1°C and programmed to reduce the temperature to 1°C for 1 h for the non-acclimated condition, and to 1°C for 2 h for cold-acclimated plants.When the desired temperatures were reached, cold-treated plants were removed from the chamber.Then plants were covered with ice and then kept at 4 ± 1°C dark room for 12 h.The plates were then incubated in the presence of light at 23°C.Survival was assessed after 4 days.For the freezing tolerance assay in soil, 7day-old WT and mutant seedlings grown on 1/2 MS medium were transferred to soil and cultivated for 2 weeks.These 3-week-old plants were either not-treated or they were acclimated to cold by incubation at 4 ± 1°C for 7 days in the RUMED 4001 freezing chamber that was programmed to decrease in temperature to 1°C for 0.5 h for the non-acclimated condition and maintained for 1h, and it dropped up to the indicated temperatures.After cold treatment in the chamber, plants were covered with ice in the dark for 24 h at 4 ± 1°C.After the ice melted, plants were placed in the presence of light at 23°C.The survival rates were determined after 4 days later at 23°C.
To measure electrolyte leakage, 3-week-old plant rosette leaflets were incubated in 200 μl deionized water in a model AP28R programmed freezing bath (Poly science) for the indicated time periods.The amount of electrolyte leakage was measured as previously described (Park et al., 2018a).

RNA extraction and quantitative RT-qPCR analysis
The total RNA from 2-week-old plants was extracted with the RNeasy plant Mini kit (Qiagen) and treated with RNase free DNase (Sigma-Aldrich).Complimentary DNA (cDNA) was synthesized by Superscript II reverse transcriptase (Invitrogen).
Reverse transcription quantitative PCR (RT-qPCR) was analyzed using SYBR Green PCR Master Mix Kit (Bio-Rad).The gene-specific primer sequences are provided in Supplemental Table S1.The relative expression levels were analyzed by the comparative cycle threshold (ΔΔCt) method.

Split-luciferase complementation imaging assay
The assay was performed with modified protocols as previously described (Chen et al., 2008).The open reading frame regions of the PWR, CBF1, CBF2, CBF3, and HD2C genes were cloned into vectors for split-luciferase complementation, which were modified and used for the Gateway system from pCambia1300-NLuc and pCambia1300-CLuc, respectively.The indicated constructs were mobilized into A. tumefaciens GV3101.The transformed GV3101 cells were grown in LB medium at 30°C overnight.Cells were then washed once with wash buffer (10 mM MgCl 2 , 10 mM MES) and resuspended in infiltration buffer (100 µM acetosyringone, 10 mM MgCl 2 , 10 mM MES).After 4-h incubation, bacterial suspensions (optical density at 600 nm = 0.5) were infiltrated into Nicotiana benthamiana leaves using a needleless syringe.After 3 days, luciferin (1 mM) was sprayed onto the leaves and light according to LUC activity was detected by an iXon CCD imaging apparatus (Andor Technology plc).

Yeast two-hybrid assay
For yeast two-hybrid assays, HOS15, HD2C, and PWR were cloned into pDEST22 (AD) and pDEST32 (BD).These plasmids were co-transformed into Saccharomyces cerevisiae strain PJ69-4A by polyethylene glycol (PEG)-mediated heat shock (42°C) transformation using the protocol of the manufacturer (Clontech).Growth of cotransformed yeast cells were observed on tryptophan, leucine, and histidine nutrientdeficient medium (-TLH).Transformation and growth of the yeast cell control was checked on -TL nutrient-deficient medium.Plates were photographed after a 5-day incubation at 30°C.The split-ubiquitin assay was performed with modified protocols as previously described (Reichel and Johnsson, 2005).In brief, PWR and CBFs were cloned into pMet-GWY-Cub-URA3p-CYC1 and pCup-NuI-GWY-CYC1 vectors modified from the original pMet and pCub vectors, respectively.Constructs were transformed into Saccharomyces cerevisiae strain JD53 by PEG-mediated heat shock (42°C).Interactions between pairs of proteins were analyzed by yeast cell growth on tryptophan, histidine, and uracil nutrient-deficient medium (-THU) and containing 5-fluoroorotic acid (5-FOA) (1 mg/ml, Zymo Research).Plates were photographed 3 to 5 days after incubation at 30°C.

Co-immunoprecipitation (Co-IP) assay
For the Co-IP assay, total proteins were extracted from 2-week-old Arabidopsis seedlings or ten-day-old Arabidopsis plants treated with or without cold (0°C) for 12 h in the presence of cycloheximide (CHX, 100 µM) and proteasome inhibitor MG132 (50 µM).For immunoprecipitation, the protein extracts were centrifuged twice at 12,000 rpm for 15 min at 4°C and incubated with anti-PWR or anti-HD2C (Agrisera) antibody fused to protein A agarose bead for 2 h at 4°C with gentle rotation.After washing with 1X PBS, proteins were mixed with 4X SDS (sodium dodecyl sulfate) buffer and heated at 90°C at 4 min.Protein were separated by SDS-PAGE (polyacrylamide gel electrophoresis), followed by immunoblotting.The immunoblots were developed using appropriate primary antibodies, including anti-PWR, anti-HD2C, anti-CBFs, and anti-HOS15 for 2 h at 23 ± 1°C or overnight at 4°C.The membranes were developed by peroxidase-conjugated secondary antibody antirabbit or anti-mouse antibody (Santa Cruz Biotechnology), and the antigenic proteins were detected with enhanced chemiluminescence using the appropriate reagent (Bio-Rad).

Subcellular localization
PWR was fused at the N-terminus of green fluorescent protein (GFP) in pMDC83.
changed from a repressive (closed) form to an active (open) form.Next, CBF transcription factors can easily access the open region of the COR promoters to induce expression of COR genes for cold tolerance.The interaction of PWR-HOS15 with CBFs further facilitates the recruitment of these transcription factors needed for enhanced COR gene expression.Basal levels of CBFs expressed under temperate conditions might help recruiting PWR-HOS15 to COR genes Both PWR-GFP and NLS-dsRed fluorescent protein (RFP) were transiently coexpressed in N. benthamiana leaves using A. tumefaciens GV3101.The fluorescence signals of GFP and RFP were captured using a confocal laser scanning microscope (Olympus FV1000).

Figure legends Fig. 1 .
Figure legendsFig.1. PWR is involved in freezing tolerance (A) Freezing phenotype of 2-week-old wild-type (Col-0), two pwr mutants (pwr-2 and pwr-3), and hos15-2 (positive control) seedlings grown on MS medium under longday condition.Two-week-old plants grown on MS media plates pretreated with cold (4°C for cold acclimation) for 7 days were exposed to -9°C freezing stress, followed by recovery at 23°C for 4 days.Photographs were taken before (Control) and after exposure to the -9°C freezing stress.(B) The survival rates of plants were calculated from (A).The data are the mean value ± SD (n=25).Asterisks indicate a statistically significant difference from Col-0 (Student's t-test, *P<0.05,**P<0.01).Similar results were obtained in three independent experiments.(C) Three-week-old plants grown on soil were cold acclimated at 4°C for 7 days, and then exposed to freezing at the indicated temperatures.Photographs were taken after the recovery at 23°C for 5 days.(D) Electrolyte leakage assay of acclimated plants was performed at the indicated temperatures.The data are the mean ± SD of biological replicates (n=3).

Fig 2 .
Fig 2. PWR positively regulates the expression of COR genes (A) COR but not CBF gene expression is reduced in pwr-2 and pwr-3 mutants.The transcript levels of cold-induced genes were quantified by RT-qPCR in Col-0 and pwr mutants.cDNA was obtained from 2-week-old plants treated at 4°C for the indicated times.ACTIN2 was used to normalize the expression of COR and CBF genes.Bars represent means ± SD from three biological replicates with three technical repeats each.Asterisks indicate a statistically significant difference from Col-0 (Student's ttest, *P<0.05,**P<0.01).(B) Cold-induced CBF protein accumulation was not significantly changed in the hos15-2, pwr-2, and pwr-2 hos15-2 mutants compared to the wild type.Total proteins were extracted from 2-week-old wild-type (Col-0), hos15-2, pwr-2, and pwr-2 hos15-2 plants treated with cold stress (4°C) for 6 h and applied to immunoblot with anti-CBF antibody cross-reacting with CBF1 to CBF3 (left).The relative band intensity of CBF proteins was calculated from blots of three biological repeats using Image Lab software (right).Values are the mean ± SD.

Fig 4 .
Fig 4. PWR is required for HOS15 function (A) Promoter scheme of the COR15A gene.Region II was predicted to contain CBFbinding cis elements (white boxes), whereas region I was not.(B) ChIP assays were performed with anti-HD2C antibody in wild-type, hos15-2, pwr-3, and pwr-3 hos15-2 plants treated with cold (0°C) for 24 h.Chromatin from 2-week-old plants treated with cold (4°C, 24 h) was immunoprecipitated with anti-HOS15 (C), anti-acetylated Histone3 (Ac-H3) (D), anti-H3K9Ac (E), and anti-H3K14Ac (F) antibodies.The amount of DNA in the immunoprecipitate complex was determined by RT-qPCR, and is presented as the fold-enrichment after normalization with a mock control using the Actin7 gene promoter.Bars represent means ± SD from three biological replicates with three technical repeats each (n=3).Asterisks indicate a statistically significant difference from Col-0 (Student's t-test, *P<0.05,**P<0.01).Similar results were obtained from three independent experiments.