The SWI/SNF ATP-Dependent Chromatin Remodeling Complex in Arabidopsis Responds to Environmental Changes in Temperature-Dependent Manner

SWI/SNF ATP-dependent chromatin remodeling complexes (CRCs) play important roles in the regulation of transcription, cell cycle, DNA replication, repair, and hormone signaling in eukaryotes. The core of SWI/SNF CRCs composed of a SWI2/SNF2 type ATPase, a SNF5 and two of SWI3 subunits is sufficient for execution of nucleosome remodeling in vitro. The Arabidopsis genome encodes four SWI2/SNF2 ATPases, four SWI3, a single SNF5 and two SWP73 subunits. Genes of the core SWI/SNF components have critical but not fully overlapping roles during plant growth, embryogenesis, and sporophyte development. Here we show that the Arabidopsis swi3c mutant exhibits a phenotypic reversion when grown at lower temperature resulting in partial restoration of its embryo, root development and fertility defects. Our data indicates that the swi3c mutation alters the expression of several genes engaged in low temperature responses. The location of SWI3C-containing SWI/SNF CRCs on the ICE1, MYB15 and CBF1 target genes depends on the temperature conditions, and the swi3c mutation thus also influences the transcription of several cold-responsive (COR) genes. These findings, together with genetic analysis of swi3c/ice1 double mutant and enhanced freezing tolerance of swi3c plants illustrate that SWI/SNF CRCs contribute to fine-tuning of plant growth responses to different temperature regimes.

SWI/SNF complexes assembled by distinct combinations of core and accessory subunits are implicated in the regulation of different cellular processes [1,[8][9][10][11]. SWI3A and SWI3B are essential genes, their inactivation leads to embryo arrest at the globular stage of embryogenesis. SWI3B also plays an important role during gametophyte development. By contrast, SWI3C and SWI3D appear to be dispensable for embryogenesis. The swi3c and swi3d mutants are viable, although they exhibit strong developmental aberrations. The swi3d mutation causes a complete sterility, whereas swi3c mutant plants are characterized by dramatic root shortening and branching, semi-dwarfism, altered leaf and flower development, and reduced fertility [8]. However, these developmental defects are less severe when swi3c plants are grown at [14][15][16] • C compared to 20-24 • C [12]. Intriguingly, inactivation of SWP73 SWI/SNF subunit in yeast is similarly reported to cause sensitivity to elevated temperature [13]. In Arabidopsis, BRM-containing SWI/SNF CRC interacts with the histone deacetylase HD2C, which is implicated in the repression of a battery of heat-activated genes [14]. These findings suggest that the SWI/SNF CRCs are involved in temperature-dependent control of transcription but so far, the underlying molecular mechanisms are largely unclear.
Here, we show that developmental defects of the Arabidopsis swi3c mutant, including embryo arrest at early stages and defective root elongation, are partially reverted when plants are grown at 14 • C. Our data indicate that SWI3C-containing SWI/SNF CRCs modulate the expression of several genes involved in low temperature signaling including ICE1, CBF1-3, MYB15, RAP2.6, ZAT10, and ZAT12. SWI3C-containing SWI/SNF CRCs were found to bind to the ICE1, MYB15 and CBF1 loci and their locations on these target genes is changed by the temperature conditions. Characterization of the swi3c/ice1 double mutant suggests a genetic interaction between SWI3C and ICE1. The swi3c mutation also influences the expression of downstream-acting cold-responsive (COR) genes and confers enhanced freezing tolerance. In conclusion, our data illustrate that the SWI3C-containing SWI/SNF CRCs are implicated in temperature-dependent regulation of plant growth and developmental responses.

Lower Temperature Alleviates Phenotypic Defects Caused by Mutations of the SWI3C Core Subunit Gene of Arabidopsis SWI/SNF CRC
Inactivation of genes encoding the SWI3-type subunits of SWI/SNF CRCs results in distinct effects on Arabidopsis development. The swi3a and swi3b mutations cause lethality at the early (globular) stage of embryo development. By contrast, the swi3c and swi3d mutants are viable but exhibit severe developmental defects including reduced fertility of swi3c and complete sterility of swi3d plants grown under optimal conditions [8]. We previously observed that reducing the ambient temperature to 14-16 • C substantially improved the fertility of the swi3c mutant, which displayed an enhanced elongation of siliques containing viable seeds [12]. This observation has prompted us to examine how lower growth temperature affects the phenotypic traits conferred by mutations of all four Arabidopsis SWI3 genes. When grown at 14 • C, the number of white translucent seeds carrying aborted embryos was reduced from 24.8% to 6.8% and 14.8% to 0.8% in siliques of swi3a/+ and swi3b/+ plants, respectively. Gametophyte lethality of swi3b/+ line decreased from 34.5% to 11.95% as a result of decreasing the growth temperature (Table 1). Further analysis of mature seeds by PCR-based genotyping of isolated embryos confirmed that the reduction of growth temperature partially suppressed the defect of embryo development at the globular stage. In the progeny of swi3a/+ plants, we identified mature swi3a embryos, although their cotyledons appeared to be underdeveloped and degenerated. Similar inspection of swi3b/+ offspring identified swi3b embryos that displayed a torpedo-stage-like developmental status. Nonetheless, germination of seeds collected from swi3a/+ or swi3b/+ plants grown at 14 • C did not result in viable homozygous swi3a or swi3b progeny. This indicated that reduction of the growth temperature permitted the swi3a and swi3b embryos to reach later stages of development but did not completely suppress the block of embryo development by the swi3a or swi3b mutations ( Figure 1A). Decreasing the growth temperature resulted in notable alleviation of the block of root elongation ( Figure 1B) and enhanced root branching [12] of the swi3c mutant. At 14 • C, the length of siliques and the yield of viable seeds increased considerably from 24 to 66% and 5 to 31% compared to wild type [12], respectively, although no gross changes in overall differentiation and development of flower organs and growth of swi3c plants were detected with the exception of slightly reduced leaf curling ( Figure 1C,D, Supporting Information Figure S1). In comparison, lower growth temperature resulted in remarkable growth retardation ( Figure 1E) and did not restore the fertility of swi3d mutant. Nonetheless, we found a nearly two-fold extension of swi3d silique length in plants growing at 14 • C (Supporting Information Figure S2A,B). Despite the extension of silique length, the swi3d mutant displayed complete sterility under both low and normal temperature conditions. In summary, except for swi3d where only siliques became longer, characteristic developmental defects caused by mutations of SWI3 core subunit genes of SWI/SNF CRCs were partially suppressed at lower temperature. As only the swi3c mutant produced viable homozygous progeny, further analysis of the mechanisms underlying temperature sensitivity of swi3 mutants addressed the question how inactivation of SWI3C gene influences the expression of genes implicated in cold temperature signaling and downstream responses.

Genes Involved in Low Temperature Responses Show Altered Expression in the Swi3c Mutant
Transcription responses to cold temperature are controlled by the ICE1 (Inducer of CBF Expression 1) Myc transcription factor (TF), which stimulates the expression of three C-Repeat Binding Factors (CBFs) of APETALA2/ETHYLENE RESPONSE FACTOR TF family. The function of ICE1 master regulator of CBFs is regulated post-translationally [15][16][17]. The HOS1 ubiquitin ligase targets ICE1 for proteasomal degradation, which is counteracted by SIZ1-mediated sumoylation of ICE1 [18]. CBFs are central activators of cold-responsive (COR) genes, which are either positively or negatively regulated by the CBF-controlled TFs RAP2.1, RAP2.6 and RAP2.7 [19]. CBF expression is repressed by MYB15, which is inactivated by ICE1 under cold stress, as well as by the C2H2 zinc finger TF ZAT12 [20,21]. Another zinc finger TF ZAT10 appears to be activated by CBF3 and acts as a repressor of a set of COR genes [22]. Quantitative real-time PCR measurement indicated that the ICE1 transcript levels were comparable in wild type and swi3c mutant plants at 14 • C, although ICE1 mRNA levels were 2-fold lower at 22 • C in the swi3c mutant ( Figure 2A). Similarly, HOS1 and SIZ1 mRNA levels were similar in wild type and swi3c plants under both temperature conditions. Although MYB15 showed 4.5-fold higher expression in swi3c plants under normal growth condition, its mRNA level did not differ from wild type at 14 • C. In comparison, expression levels of RAP2.6, ZAT10 and ZAT12 were elevated (respectively 2, 4 and 4.5-fold) at 22 • C but showed a notable down-regulation (50, 3 and 6.5-fold, respectively) at 14 • C in swi3c compared to wild type plants. Transcript levels of CBF2 and CBF3 did not differ at normal temperature, but CBF1 expression was somewhat (2-fold) higher in swi3c plants as in wild type. However, at 14 • C all three CBF genes showed higher (2.5, 1.5 and 4.0-fold, respectively) induction in the swi3c mutant compared to wild type. In conclusion, this data indicated that the swi3c mutation elevates the steady-state levels of CBF mRNAs, while decreases in parallel the transcript levels of ZAT12 CBF repressor and downstream-acting ZAT10 and RAP2.6 negative regulators of subsets of COR genes at 14 • C. In addition, differential up-regulation of MYB15, RAP2.6, ZAT10 and ZAT12 expression levels at 22 • C in the swi3c mutant suggested that SWI3C-containing SWI/SNF CRCs could play a role in temperature-dependent modulation of transcription responses. To map chromatin localization of SWI3C-containing SWI/SNF CRCs on genes involved in cold signaling, the swi3c mutant was complemented by a SWI3C-YFP-HA gene construct carried by the Agrobacterium binary vector pEarley101 [23]. Constitutive expression of SWI3C-YFP-HA restored at both 22 • C and 14 • C the phenotypic defects, such as leaf curling (Supporting Information Figure S3A), block of root elongation and sterility of the swi3c mutant to wild type and resulted in the production of intact SWI3C-YFP-HA protein of 115 kDa detected by western blotting with anti-GFP antibody (Supporting Informationn Figure S3B). Chromatin immunoprecipitation followed by quantitative PCR (ChIP-qPCR) was performed with nuclear extracts prepared from formaldehyde cross-linked swi3c/ SWI3C-YFP-HA and control wild type plants using primers for amplification of 120-160 bp segments of 5 -UTRs located between positions +450 and −1200 relative to the transcription start site (TSS). In the promoter region of ICE1 gene, SWI3C-YFP-HA was localized between positions +449 and +571 at 22 • C ( Figure 2B). The peak value of SWI3C-YFP-HA increased approximately 3-fold and its localization was shifted to position −104 at 14 • C ( Figure 2C). In the 5 -UTR of MYB15, SWI3C-YFP-HA was localized around position of +458 at 22 • C ( Figure 2D) but at 14 • C the level of SWI3C-YFP-HA decreased approximately 2-fold and its localization was shifted to position −817 ( Figure 2E). A similar change in positioning of SWI3C-YFP-HA was detected in the 5 -UTR region of the CBF1 gene, in which the peak of ChIP-signal was detected around position +343 at 22 • C ( Figure 2F), while at 14 • C the localization was shifted to position −695 in parallel with approximately 2.5 fold increase of signal intensity ( Figure 2G). The ChIP-PCR data thus indicated that in the case of ICE1, lowering the temperature resulted in a shift toward proximal position of TSS, whereas in the 5 -UTR of MYB15 the localization of SWI3C-YFP-HA was changed to a more distant upstream position. On the other hand, the localization of SWI3C-YFP-HA both downstream and upstream of TSS of the CBF1 gene suggested that SWI3C-containing SWI/SNF CRCs might act as negative regulators because CBF1 transcript levels were 2 and 2.5-fold higher in the swi3c mutant compared to wild type. Finally, localization of SWI3C-YFP-HA closer to TSS did not seem to affect ICE1 transcription, which is comparable between wild type and the swi3c mutant at 14 • C. Nevertheless, more upstream localization of SWI3C correlated with a 2-fold reduction of ICE1 mRNA levels at 22 • C. In summary, these data indicated that the localization of SWI3C-SWI/SNF CRCs is dynamically changed in the promoter regions of examined genes in a temperature-dependent manner.
Analysis of predicted transcription factor binding sites in the SWI3C-binding region using the MEME software [25] identified several potential cis-regulatory elements in the SWI3C-targeted regions of MYB15, ICE1 and CBF1 genes (Supporting Information Table S2). However, subsequent GOMo analysis did not confirm potential biological roles of motives derived from the MEME analyses (Supporting Information Table S2). Additional predictions obtained by using the AGRIS server (AGRIS; http://arabidopsis.med.ohio-state.edu/) suggested potential conservation of a MYB4 binding site (ACCTAAC, [26] at mapped positions of SWI3C binding to CBF1 locus at 22 • C. In addition, a SWI3C-binding domain mapped around position −695 relative to the TSS of CBF1 at 14 • C was predicted to carry RAV1-A (CAACA), SORLIP2 (GGGCC) and LFY (LEAFY) binding (CCAGTG) motives. The RAV1-A binding site is recognized by the ABI3/VP1 transcription factors family [27], the SORLIP2 motif is involved in the control of light-regulated gene expression [28], whereas the LFY consensus is recognized by the LEAFY transcription factor, which is reported to mediate the recruitment of SWI/SNF CRCs to their target site on e.g., AP3 (APETALA 3) promoter [29]. The 5 -UTRs of ICE1 and MYB15 genes however lacked these motives. The only known predicted TF-binding site in these 5 -UTRs was a BELLRINGER motif (AAATTAAA), which was identified in the SWI3C-binding region in MYB15 (position −817) at 14 • C. This motif is used by homeobox/MYB transcription factors of regulatory networks controlling development, metabolism and responses to biotic and abiotic stresses in Arabidopsis [30] (Supporting Information Table S2).  [24]. Asterisks indicate significant difference from wild type (*): p < 0.05 according to t-test. Bars refer to SD.

The Swi3c Mutation Changes Nucleosome Positioning on the ICE1, MYB15 and CBF1 Loci
To correlate the SWI3C chromatin localization data with alterations of nucleosome positioning in the swi3c mutant, micrococcal nuclease protection qPCR (MNase-qPCR) assays were performed using a set of primers positioned between +600 and −1100 relatively to TSSs of the ICE1, MYB15 and CBF1 genes. In wild type plants grown at 22 • C five nucleosomes protected from MNase digestion were mapped to positions −200 (−1 nucleosome), −360 (−2 nucleosome), +225 bp (+1), + 360 bp (+2) and +530 (+3) relative to the TSS of ICE1. In the swi3c mutant the TSS proximal nucleosome exhibited a slight shift to position −140 and fuzziness, as well as increased occupancy, correlating with lower than wild type transcript levels at 22 • C. In addition, the swi3c mutation resulted in enhanced positioning of nucleosome +2 and fuzziness of nucleosome +3, both located close to mapped position of SWI3C at normal temperature ( Figure 3A). By contrast, wild type and swi3c plants grown under decreased temperature did not show a change in nucleosome occupancy correlating with no difference in ICE1 transcription between wild type and swi3c plants under this condition ( Figure 3B).
In case of the MYB15 locus four nucleosomes protected from MNase digestion were mapped in wild type plants grown at 22 • C. Two of these were located in the distal promoter region centered to positions −816 (−1 nucleosome) and −1130 (−2 nucleosome), whereas the other two nucleosomes were mapped into the gene body around positions + 394 bp (+1) and +574 (+2). In correlation with a 4.5-fold increase of MYB15 transcript levels at 22 • C, in the swi3c mutant we found a complete loss of nucleosomes −1 and −2, as well as decreased occupancy of nucleosome +1 located in the vicinity of mapped SWI3C-binding region at position +458 ( Figure 3C). At 14 • C, the positioning of nucleosomes was slightly shifted to positions −726 bp (−1) 1000 (−2), +244 (+1) and 544 (+2). Compared to wild type, in the swi3c mutant the positioning of nucleosomes was fuzzier, and nucleosomes +1 and −2 were slightly shifted ( Figure 3D). However, no striking difference was found in positioning of nucleosomes correlating with no difference in MYB15 mRNA levels between wild type and swi3c plants at 14 • C.
In both wild type and swi3c plants grown under normal temperature four nucleosomes were protected from MNase in the CBF1 locus. Two nucleosomes were located in the distal promoter region centered to positions −815 (−1) and −990 (−2), while others were situated at positions the +120 (+1) and + 325 (+2) in the gene body. In correlation with a 2-fold increase of CBF1 mRNA levels compared to wild type at 22 • C, the −1 nucleosome exhibited a slight fuzziness, and nucleosomes +1 and +2 in the vicinity of identified SWI3C binding site at position +343 ( Figure 3E) showed decreased occupancy in the swi3c mutant ( Figure 3F). At 14 • C, the positions of nucleosomes were shifted to positions −680 (−1), −855 bp (−2) −1000 (−3) and +270. In contrast to normal temperature, at 14 • C the positioning of nucleosome +1 did not change. However, a complete loss of nucleosome −2 was detected close to the location of mapped SWI3C target region at position −695. This observation, together with a 2.5-fold increase of CBF1 mRNA levels in swi3c plants indicated that lack of SWI3C caused two different, temperature-dependent changes to generate more open chromatin structure in the promoter region of CBF1gene ( Figure 3F).
As this analysis did not show a clear correlation between nucleosome positioning, SWI3C occupancy and ICE1, MYB15 and CBF1 gene expression, we reanalyzed the data and compared nucleosome positioning and transcript level detected in the WT 22 • C vs WT 14 • C and swi3c 22 • C vs swi3c 14 • C datasets (Supporting Information Figure S4A,B). This indicated that both WT and swi3c mutant plants responded to decreased temperature in a similar way concerning ICE1 and CBF1 genes (Supporting Information Figure S4B), although some nucleosome positioning changed in temperature-dependent manner (Supporting Information Figure S4A). The nucleosome positions differed between WT and swi3c mutant at 22 • C and CBF1 expression in the mutant plants was much higher than in WT in both temperature conditions (Figure 2A). Interestingly, the highest changes in transcription for swi3c mutant vs WT at 22 • C were observed for MYB15 gene (Figure 2A). In the swi3c mutant, the MYB15 gene responded to decreased growth temperature in different way than in WT, where MYB15 expression was elevated at 14 • C, whereas swi3c plants exhibited similar levels of MYB15 mRNA regardless of temperature conditions (Supporting Information Figure S4B). Moreover, we found that the nucleosomes −1 and −2 at 22 • C on MYB15 in the in swi3c mutant were missing from promoter region close to TSS, which could explain the almost 5-fold highest expression at 22 • C compared to WT. At lower temperature, the positioning of nucleosomes in this region is observed and the activation of transcription and proper response (elevated expression) occurred in WT plants. By contrast, the MYB15 expression level in the swi3c mutant exposed to lower temperature resembled WT under the same conditions. Thus, although in the MYB15 promoter region two nucleosomes were located in cold, the swi3c mutation did not seem to alter MYB15 transcription in response to low temperature.
In conclusion, the MNase nucleosome mapping data only partially supported the results of transcript measurements and SWI3C ChIP-qRT-PCR assays revealing that the lack of SWI3C subunit of SWI/SNF CRCs results in distinct temperature-dependent changes in the chromatin structure of genes involved in cold signaling. This suggests that likely other remodelers (i.e., other classes of SWI/SNF complexes) are also involved in transcriptional control of genes in the cold response pathway in response to decreased temperature. However, replacement of the missing SWI3C-SWI/SNF complex by alternative CRCs remains to be confirmed by further studies.

Genetic Interactions Between SWI3C and ICE1
ICE1 is a key positive regulator of the cold signaling pathway. To test how inactivation of ICE1 affect the cold temperature-dependent suppression of swi3c developmental and fertility defects, the swi3c-1 mutation [8] was introduced by crossing into the ice1 (SALK_068119) T-DNA insertion mutant. In the SALK_068119 mutant a T-DNA insertion in exon 4 of ICE1 resulted in a C-terminal truncation of the coding domain, which showed still active transcription 5 upstream of the insertion site (Supporting Information Figure S5A,B). Nonetheless, the mutation appeared to cause at least a partial loss of ICE1 function because in the F2 offspring of swi3c/+ice1/+ hybrid we failed to identify homozygous swi3c ice1 double mutants. In the siliques of swi3c/+ice1 F2 progeny we did not observe embryo lethality, which suggested that swi3 ice1 homozygous seeds were either not viable or incapable of germinating in soil under normal conditions. When germinating 216 F2 seed of a swi3c/+ice1 plant on 0.5 MS medium only 8 (3.7%) swi3 ice1 seedlings were identified by PCR-based genotyping indicating a significant deviation from the expected 3:1 ratio. Upon transfer into soil, the double mutant plants displayed an enhanced severity of swi3c developmental defects, including enhanced dwarfism and reduction of leaf blade size (Supporting Information Figure S6A) and complete sterility. Comparative microscopic analysis of developmental status of embryos using Nomarski optics indicated retarded development of embryos in both swi3c and ice1 mutants compared to wild type plants. However, both swi3c and ice1 single mutants could produce viable progeny, whereas in the double swi3c ice1 mutant pollen tubes could not germinate into the embryo sac and fertilization did not occur resulting in a failure of embryo formation (Supporting Information Figure S6B).

Inactivation of SWI3C Alters the Expression of Cold-Responsive (COR) Genes
The expression of cold-responsive maker genes ADH1 (Alcohol Dehydrogenase 1), COR15A  Figure S6C). By contrast in swi3c plants growing at 14 • C we detected 2.0-fold and 1.5-fold lower expression of ADH1 and COR15A genes compared to WT.. We also found a discrete up-regulation of COR47 and 2.0-fold elevated expression of RD22 and RD29A genes in the swi3c mutant compared to WT plants grown at 14 • C. In summary, these results indicated an involvement of SWI3C in transcriptional regulation of COR genes under both normal and low temperature conditions.

Swi3c Mutant Exhibits Enhanced Freezing Tolerance
Overexpression of CBFs in cooperation with other cold-regulated first-wave transcription factors was demonstrated to increase freezing tolerance [33]. Therefore, we examined the freezing tolerance trait of swi3c mutant by performing a freezing-survival assays. Wild type and swi3c plants were moved to 4 • C, and then within 30 min the temperature decreased to 0 • C. Subsequently, the temperature decreased by 1 • C every hour until it reached −8 • C. Next, the plants were moved to a growth chamber with long day and 22 • C conditions. After 7 days recovery period, the number of surviving plants was determined. We found that the survival rate for wild type plants was approximately 20% while for swi3c it reached up to −90% ( Figure 4A,B) indicating enhanced freezing tolerance. We also found that the constitutive expression of SWI3C-YFP-HA restored, upon freezing, the phenotypic defects exhibited by the swi3c plants (Supporting Information Figure S7). To further confirm enhanced freezing tolerance of swi3c mutant, we performed electrolyte leakage assays [34] (and determined the EL50 value (the temperature at which 50% of cellular electrolytes was released due to freezing tolerance). For wild type plants, the EL50 value was~−6 • C, whereas the swi3c mutant exhibited only~30% of cellular electrolyte release at the same temperature ( Figure 4C). Collectively, these results demonstrated that the swi3c mutant exhibited increased freezing tolerance.

Discussion
Plants, as sessile organisms, are constantly exposed to various abiotic and biotic stresses. To be able to survive in a changing environment, plants evolved numerous defense mechanisms allowing them to survive under stress conditions. The phenotypic plasticity plays a crucial role in adaptation to changing environmental conditions, thanks to which plants developed adaptive defense reactions to response to unfavorable conditions. Epigenetic changes occurring at the chromatin level play a pivotal role in the genotype-environment interaction [35].
Exposure to low temperature stimulates signaling pathways that control the expression of genes conferring plant tolerance to cold stress [36]. ICE1 (Inducer of CBF Expression 1), also called SCREAM1 [37], is a key positive regulator of the cold-responsive pathway, which is activated by sumoylation. ICE1 is critical for induction of the CBF (C-Repeat Binding Factor) regulon and inhibition of the CBF repressor MYB15 [20]. ICE1 is negatively regulated through ubiquitination by the HOS1 (High Expression of Osmotically Responsive Genes 1) RING finger-type ubiquitin E3 ligase, which stimulates its proteasomal degradation [18]. CBFs cross-regulate their own transcription [22] and bind to cis-elements in the promoters of COR genes and activate their transcription [38]. The ICE1-MYB15-CBF signaling module plays a major role in adaptation and acclimation of plants to cold stress [39].
Adaptation to abiotic stresses is accompanied by either activation or inactivation of a large number of genes, implying changes in their chromatin structures. Alterations in the chromatin structure are induced by DNA methylation, modification of core histone tails, exchange of histone variants, and altered positioning of nucleosomes [40,41]. Histone deacetylase HD2C was reported recently to play an important role in plant adaptation to temperature changes by modulating the transcription of cold-responsive genes COR15A and COR47 [42]. HD2C is known to co-operate with SWI/SNF chromatin remodeling complexes in response to high temperature in Arabidopsis [14]; however, it is yet unknown whether HD2C interaction with SWI/SNF CRCs plays also a role in plant responses to cold temperature, which is suggested by the observation that reduced fertility of plants carrying mutations of the SWI3C core subunit of SWI/SNF CRCs is largely alleviated by decreasing the growth temperature to 14 • C [12].
In this report, we investigated cold temperature sensitivity of developmental defects caused by mutations of SWI3 core subunit genes of Arabidopsis SWI/SNF CRCs. We showed that decreasing the growth temperature to 14 • C permitted the swi3a and swi3b mutant embryos to develop close to final maturation, although their development was arrested at the globular stage at 22 • C. Similarly, the SWI3B-dependent female gametophytic defect is partially restored by lowering the growth temperature to 14 • C. In case of the swi3c mutant, in addition to improved fertility, lowering the temperature partially suppressed the block of root elongation and enhanced root branching, and leaf curling. As exception, developmental defects of the swi3d mutant were not restored by decreased growth temperature. The plants growing at decreased temperature exhibited even more retarded growth but displayed a slight extension of siliques compared to swi3d plants grown at 22 • C. In contrast to the swi3c mutant, swi3d plants did not improve their fertility at 14 • C and remained fully sterile. The differential behavior of swi3c and swi3d mutants further supports our previous findings that SWI3C and SWI3D are present in different classes of SWI/SNF CRCs [8] and suggests that SWI3D-containing SWI/SNF CRCs class acts distinctly from temperature signaling pathways.
Inactivation of SWI3C was found to alter transcriptional regulation of several genes acting in the cold signaling pathway, including CBF1-3, ICE1, MYB15, RAP2.6, ZAT10 and ZAT12. Interestingly, the transcription of these genes was differentially deregulated in the swi3c mutant at 22 and 14 • C. This suggested that the function of SWI3C-containing SWI/SNF CRCs is differently regulated depending on the ambient temperature. The swi3c mutation was found to enhance transcription of the CBF1-3 genes at 14 • C, which normally correlates with enhanced cold tolerance [43]. Subsequent chromatin immunoprecipitation study indicated direct targeting of SWI3C to the ICE1, MYB15 and CBF1 loci, and in all three cases SWI3C was localized at 22 • C 3 downstream of TSS in these genes. Although unexpected, this observation is in line with recent mapping studies of human [44] and Arabidopsis [45] SWI/SNF CRCs, which were located in various gene regions. Similarly, SWI/SNF CRCs in Drosophila were observed to bind gene body regions affecting transcription elongation [46] and RNA polymerase II pausing [47]. The location of SWI3C on the ICE1, CBF1 and MYB15 loci has remarkably changed in a temperature-dependent manner and was shifted 5 -upstream of TSSs in the promoter regions of ICE1, MYB15 and CBF1 genes.
In the case of the CBF1 gene, SWI3C-binding under control conditions occurred in close proximity to a MYB4-binding motif, which is targeted by various MYB transcription factors including MYB15 [20], another regulatory target of SWI3C-SWI/SNF CRCs. By contrast, under decreased temperature SWI3C shifted from this site in the gene body to the putative RAV1-A, SORLIP2 and LFY (LEAFY) consensus binding sites in the distal CBF1 promoter region. While the LEAFY transcription factor is a known binding partner of SWI/SNF CRCs, there is so far no report on the regulatory interdependence of RAV1 and SWI/SNF. The SORLIP2 motif was identified in the promoter regions of phytochrome A-regulated genes [28] and the BRM ATPase subunit of SWI/SNF CRCs is reported to interact directly with PIF1 (PHYTOCHROME-INTERACTING-PARTNER 1) [48]. Nonetheless, so far, no regulatory link was identified between SORLIP2 and SWI/SNF. A different localization of SWI3C on the CBF1 gene indicated two different temperature-dependent modes of transcriptional regulation by SWI/SNF-mediated chromatin remodeling at 22 and 14 • C. By contrast, different localization of SWI3C-SWI/SNF CRCs on the ICE1 locus, which is not coupled to altered transcription at 14 • C, might reflect altered chromatin loop formation, transcription factor binding, histone modifications, etc. [47].
On the CBF1, ICE1, and MYB15 loci the chromatin nucleosome structure varies in and between wild type and swi3c mutant plants depending on the growth temperature. Lower transcription of ICE1 in the swi3c mutant at 22 • C correlates with altered (enhanced) positioning and shift of nucleosomes. An opposite effect is found on the MYB15 locus, where we observed an absence (promoter region) and decreased positioning (gene body) of particular nucleosomes. Decreased positioning of nucleosomes in the gene body and slight fuzziness of nucleosomes in the promoter region correlate with enhanced transcription of CBF1 and MYB15 genes under normal growth condition in the swi3c mutant. These findings are in line with the results showing that the presence of SWI/SNF CRCs in the gene body plays an important role in the control of nucleosome density during transcriptional elongation in Drosophila [46]. The contrasting situation was observed under a lower growth temperature where the chromatin structure on ICE1 did not differ from the wild type, whereas only some fuzziness of nucleosomes was detected on MYB15, while the expression of these genes was not altered in the swi3c mutant compared to wild type. These results suggest that inactivation of SWI3C has no effect on the MYB15 gene expression at decreased temperature. However, the comparative analysis of MYB15 expression levels in WT and swi3c plants growing at 22 • C vs. 14 • C indicated that indeed WT and swi3c exhibit substantially different activation modes of the MYB15 gene. In WT plants growing at decreased temperature, MYB15 activation occurs but the swi3c plants are unable to elevate MYB15 expression under these conditions. By contrast, the swi3c mutant was still able to elevate CBF1 expression under decreased temperature condition compared to swi3c plants grown at 22 • C, although under both temperature conditions the expression of MYB15 was higher in swi3c than in WT plants. Interestingly, on the CBF1 gene, a loss of nucleosome-2 was observed at 14 • C, corresponding to the position of potential cis-regulatory sequences and enhanced transcription of CBF1 in the swi3c mutant. Collectively, these findings indicate that SWI/SNF CRCs are likely implicated in chromatin remodeling of ICE1, MYB15 and CBF1 transcription units under changing ambient temperature.
The analysis of swi3c/ice1 double mutant shows that simultaneous inactivation of ICE1 and SWI3C has a dramatic effect on Arabidopsis development, i.e., the plants were severely dwarfed and completely sterile, while both swi3c and ice1 single mutants produced some seeds [8,49]. Consistently, further investigation of expression of cold-responsive genes indicated their temperature-dependent deregulation in the swi3c mutant plants. These results suggest that SWI3C is involved in fine-tuning of cold responses in Arabidopsis ( Figure 5A-C), in addition to its previously documented roles in modulation of leaf and flower development and hormone signaling pathways [1,11,12]. This conclusion is supported by the fact that swi3c plants exhibit an enhanced freezing tolerance, which was demonstrated by their increased survival rate and decreased electrolyte leakage at −8 • C. It was recently shown that overexpression of CBFs increases freezing tolerance in Arabidopsis [33]. Thus, enhanced freezing tolerance of the swi3c mutant is consistent with observed overexpression of CBF1-3 genes under decreased temperature.
Seeds were sown on soil or plated on half-strength MS medium (Sigma-Aldrich) containing 0.5% sucrose and 0.8% agar (pH 5.8). Plants were grown under long day (LD) conditions (16h light/8h dark) at 22 • C or 14 • C, 70% humidity and 140 mM m -2 s -1 light intensity. For qRT-PCR, ChIP-qPCR and MNase-qPCR analyses plant material was collected six hours after the onset of the light period from the aerial part of plants in principal stage growth with 12 rosette leaves > 1mm length according to Boyes et al. (2001) [24]. For each experiment, three biological and three technical replicates were used. For the assessment of phenotypic alterations three biological replicates were used.

Construction of Genetically Complemented Plants by Expression of YFP-HA Tagged SWI3C Protein
The SWI3C cDNA was cloned into pDONR207, sequenced, and moved into pEarley101 [23] using the Gateway procedure. Transgenic lines were generated by floral dip transformation of swi3c-1/+ plants using the Agrobacterium GV3101 (pMP90) strain [51,52]. Primary transformants were selected based on their BASTA (100µM) resistance. Homozygous swi3c plants complemented with the SWI3C-YFP-HA construct were identified by PCR-based genotyping and subjected to subsequent analysis of their phenotypic traits. The expression of SWI3C-YFP-HA protein in swi3c/ SWI3C-YFP-HA plants was confirmed by Western blotting using a rat anti-GFP antibody (Chromotek).

RNA Extraction and qRT-PCR Analysis
RNA was extracted using RNeasy plant kit (Qiagen). Traces of genomic DNA were removed using a TURBO DNA-free kit (Ambion). 2.5 µg of total RNA was reverse-transcribed using a first-strand cDNA synthesis kit (Roche). qRT-PCR assays were performed with SYBR Green Master mix (BioRad) and specific primers for PCR amplification. qRT-PCR data were recorded and analyzed using iQ-PCR (BioRad) or LightCycler480 (Roche) equipment and software according to manufacturers' recommendations. PP2A and UBQ5 (AT1G13320 and AT3G62250, respectively) mRNAs were used as reference. The relative transcript level of each gene was determined by the 2 -∆∆Ct method [53]. For each primer pair, the primer efficiency was measured and melting curve analyzed. For each experiment, three biological and three technical replicates were used. Primers used for qRT-PCR study are listed in Supporting Information Table S1.

Chromatin Immunoprecipitation (ChIP)
Chromatin was isolated from 2g samples of WT and swi3c/ SWI3C-YFP-HA seedlings in the principal growth stage 1.12 [24] grown at 22ºC or 14ºC, respectively, and sonicated according to [54] with the exception that instead of pepstatin A and aprotinin, 1% PIC (Sigma), 1mM PMSF and 5mM β-me were included in the extraction buffer. Cross-linked SWI3C-YFP-HA samples were subjected to immunoprecipitation using 25µL of GFP-TRAP beads (Chromotek) according to the manufacturer's protocol. The percentage of input was calculated using 2 -∆∆Ct method [53]. To facilitate the comparison of different samples, the calculated percent input of wild type was set to 1. The relative enrichment represents the fold change compared to the wild type. The exon region of TA3 was used as negative control [55]. For each experiment three biological and three technical replicates were used.

Mapping of Nucleosome Locations Using MNase Protection Assay
For the mapping of nucleosome positions, 2g samples of WT and swi3c seedlings in the principal growth stage 1.12 [56] grown at 22ºC or 14ºC, respectively, were proceeded according to Saleh et al., (2008) [54] and Zhu et al., (2013) [57] with minor modifications (i.e., the buffer contained 1% proteinase inhibitor cocktail (Sigma) instead of pepstatin A and aprotinin). 200 Kunitz units of Micrococcal Nuclease (NEB) was used for digestion of each sample. DNA was purified with phenol: chloroform extraction, ethanol precipitation and subsequently separated on 2% agarose gels. Bands corresponding to~150 bp were gel isolated, purified, and used for qPCR measurements. Relative nucleosome occupancy was represented as fraction of undigested chromatin DNA and plotted against the ICE1, MYB15 and CBF1 genes position with respect to the TSS (transcription start site) for each primer pair where the position denotes the center of each 70-110 bp amplicon. For each experiment three biological and three technical replicates were used.

Microscopic Analyses
Immature seeds were cleared in Hoyer's solution (30 mL of water, 100 g of chloral hydrate, 7.5 g of Arabic gum, and 5 mL of glycerin) on a glass slide and examined with a compound microscope equipped with Nomarski optics. For scanning electron microscopy (SEM) the plant material was dried in liquid carbon dioxide and mounted on stubs using double sided adhesive and conductive tabs. Next, plant material was coated with gold and platin before imaging with a Zeiss Supra 40VP SEM (Carl Zeiss NTS, Oberkochen, Germany). For each experiment at least three biological replicates were used.

Freezing Tolerance Assay
To determine the freezing tolerance of swi3c plants, the electrolyte leakage and survival rate measurements were used. Electrolyte leakage assay was performed according to the methods described by Guo et al., (2002) [58] and Zhao et al., (2016) [34]. Briefly, rosette leaves from 3 weeks old plants grown on 0.5 Murashige and Skoog medium (Sigma Aldrich) containing 0.5% sucrose and 0.8% agar under long day condition (LD) (16h day/8h night) were excised and placed in Eppendorf tubes containing 100 µl deionized water. To each tube a similar ice cube was added and immediately transferred to a freezing bath cooled to 0 • C. When all the tubes were placed in freezing bath the program started according to conditions 1 • C decrements every 30 min until reached −8 • C. The tubes were removed from freezing bath when samples reached desired temperature. Solutions with leaves were then transferred to a falcon tube containing 20 mL of deionized water. Then tubes were incubated overnight and electroconductivity was measured. After measurements, the tubes were incubated in −80 • C overnight and the next day electroconductivity was measured again. The electrolyte leakage was calculated as a percentage ratio of conductivity before and after −80 • C incubation. Three biological and technical replicates were used.
The survival freezing assay was performed by incubation of 3 weeks old plant grown on MS medium in freezing incubator by applying following temperature program: start temperature was set at 4 • C and then dropped to 0 • C by 30 min and then temperature decreased by 1 • C every 1 h until reached −8 • C. After treatment plants were recovered at control condition for 7 days when survival rate was counted. Three biological and technical replicates were done.

Bioinformatic Analysis of Cis-Regulatory Elements
Bioinformatic analysis of cis-regulatory elements was performed using the Multiple EM for Motif Elicitation (MEME) [25] and followed by Gene Ontology for Motifs (GOMo) [59]. MEME analysis was performed with default settings using minw = 4 and maxw = 50 parameters. GOMo analysis was done with default settings.
To get the more comprehensive picture of cis-regulatory elements the additional bioinformatic analysis of cis-regulatory elements was performed using the Arabidopsis Gene Regulatory Information Server (AGRIS; http://arabidopsis.med.ohio-state.edu/) [60][61][62].

Statistical Analyses
The method used in this paper for statistical analyses was unpaired Student's T-test.