Vernalization-triggered expression of the antisense transcript COOLAIR is mediated by CBF genes

To synchronize flowering time with spring, many plants undergo vernalization, a floral-promotion process triggered by exposure to long-term winter cold. In Arabidopsis thaliana, this is achieved through cold-mediated epigenetic silencing of the floral repressor, FLOWERING LOCUS C (FLC). COOLAIR, a cold-induced antisense RNA transcribed from the FLC locus, has been proposed to facilitate FLC silencing. Here, we show that C-repeat (CRT)/dehydration-responsive elements (DREs) at the 3′-end of FLC and CRT/DRE-binding factors (CBFs) are required for cold-mediated expression of COOLAIR. CBFs bind to CRT/DREs at the 3′-end of FLC, both in vitro and in vivo, and CBF levels increase gradually during vernalization. Cold-induced COOLAIR expression is severely impaired in cbfs mutants in which all CBF genes are knocked-out. Conversely, CBF-overexpressing plants show increased COOLAIR levels even at warm temperatures. We show that COOLAIR is induced by CBFs during early stages of vernalization but COOLAIR levels decrease in later phases as FLC chromatin transitions to an inactive state to which CBFs can no longer bind. We also demonstrate that cbfs and FLCΔCOOLAIR mutants exhibit a normal vernalization response despite their inability to activate COOLAIR expression during cold, revealing that COOLAIR is not required for the vernalization process.


Introduction
Appropriate timing of flowering provides evolutionary advantage of plant reproductive success. As sessile organisms, plants have evolved mechanisms through which seasonal cues coordinate the transition to flowering. One of the significant environmental factors affecting flowering time of plants adapted to temperate climates is the temperature changes during the seasons, and plants have evolved complex sensory mechanisms to monitor the surrounding temperature to properly control the timing of flowering and tolerate thermal stress (Went, 1953;Penfield, 2008;Ding and Yang, 2022).
In contrast to cold acclimation, vernalization, a floral-promotion process that occurs during winter, requires an extended cold period (Napp-Zinn, 1955;Chouard, 1960;Michaels and Amasino, 2000). eLife digest Long spells of cold winter weather may feel miserable, but they are often necessary for spring to blossom. Indeed, many plants need to face a prolonged period of low temperatures to be able to flower; this process is known as vernalization.
While the molecular mechanisms which underpin vernalization are well-known, it is still unclear exactly how plants can 'sense' the difference between short and long periods of cold. Jeon, Jeong et al. set out to explore this question by focusing on COOLAIR, one of the rare genetic sequences identified as potentially being able to trigger vernalization. COOLAIR is a long noncoding RNA, a partial transcript of a gene that will not be 'read' by the cell to produce a protein but which instead regulates how and when certain genes are being switched on. COOLAIR emerges from the locus of the FLC gene, which is one of the main repressors of flowering, and it gradually accumulates in the plant when temperatures remain low for a long period. While some evidence suggests that COOLAIR may help to switch off FLC, other studies have raised some doubts about its involvement in vernalization.
In response, Jeon, Jeong et al. examined the FLC gene in a range of plants closely related to A. thaliana, and in which COOLAIR also accumulates upon cold exposure. This helped them identify a class of proteins, known as CBFs, which could bind to sequences near the FLC gene to activate the production of COOLAIR when the plants were kept in cold conditions for a while. CBFs were already known to help plants adapt to short cold snaps, but these experiments confirmed that they could act as both short-and long-term cold sensors.
This work allowed Jeon, Jeong et al. to propose a model in which CBF and therefore COOLAIR levels increase as the cold persists, until changes in the structure of the FLC gene prevent CBF from binding to it and COOLAIR production drops. Unexpectedly, examining the fate of mutants which could not produce COOLAIR revealed that these plants could still undergo vernalization, suggesting that the long noncoding RNA is in fact not necessary for this process.
These results should prompt other scientists to further investigate the role of COOLAIR in vernalization; they also give insight into how coding and noncoding sequences may have evolved together in various members of the A. thaliana family to adapt to the environment.
Several long-term cold-induced factors have been shown to play crucial roles in the epigenetic silencing of FLC. VERNALIZATION INSENSITIVE 3 (VIN3) family genes, which are upregulated by prolonged cold, encode plant homeodomain (PHD) proteins that recognize H3K9me2 enriched in FLC chromatin during vernalization (Sung and Amasino, 2004b;Kim and Sung, 2013;Kim and Sung, 2017a). These proteins mediate the recruitment of PRC2 and the subsequent deposition of H3K27me3 at the FLC nucleation region (Sung and Amasino, 2004b;De Lucia et al., 2008;Kim and Sung, 2013). In addition, vernalization-induced long noncoding RNAs (lncRNAs) are involved in such histone modifications. COLDAIR and COLDWRAP, the two lncRNAs transcribed from the first intron and promoter region of FLC, respectively, are required for H3K27me3 deposition in response to longterm cold (Heo and Sung, 2011;Kim and Sung, 2017b). COLDAIR and COLDWRAP are also thought to affect the formation of the intragenic chromatin loop at the FLC, which may be a part of the FLC silencing mechanism (Kim and Sung, 2017b). Unlike COLDAIR and COLDWRAP, which are transcribed in the sense direction of FLC, another lncRNA, COOLAIR, is an antisense transcript expressed from the 3′-end of FLC (Swiezewski et al., 2009). The gradual accumulation of COOLAIR reaches a maximum within a few weeks of cold exposure, whereas COLDAIR and COLDWRAP show peaks at a later phase of vernalization (Csorba et al., 2014;Kim and Sung, 2017b). COOLAIR was reported to remove active histone marks from FLC chromatin (Liu et al., 2007;Csorba et al., 2014;Fang et al., 2020;Xu et al., 2021a;Zhu et al., 2021). Particularly, in summer annuals, phase-separated RNA-processing complexes favor co-transcriptional proximal polyadenylation of COOLAIR (Marquardt et al., 2014;Wang et al., 2014;Fang et al., 2019;Wu et al., 2020;Xu et al., 2021b). The COOLAIR-processing machinery exhibits transient and dynamic interactions with an H3K4 demethylation complex, leading to FLC suppression at warm temperatures (Liu et al., 2007;Fang et al., 2020;Xu et al., 2021a). COOLAIR is also likely to be involved in reducing H3K36me3 at the FLC during vernalization process (Csorba et al., 2014). COOLAIR was reported to promote the sequestration of the FRI complex from the FLC promoter by condensing it into phase-separated nuclear bodies . This has been suggested to cause the inactivation of FLC, which is probably accompanied by the silencing of FLC chromatin through the removal of H3K36me3. However, a previous study has raised the issue that COOLAIR appears not to be necessary for vernalization Luo et al., 2019).
Compared with the signaling pathway of cold acclimation, how a long-term cold signal is transduced to trigger the induction of VIN3 and lncRNAs is less well understood. It is marginally known that some chromatin modifiers, NAC WITH TRANSMEMBRANE MOTIF 1-LIKE 8 (NTL8), and CCA1/LHY act as positive regulators of VIN3 during the vernalization process (Kim et al., 2010;Jean Finnegan et al., 2011;Zhao et al., 2020;Kyung et al., 2022). However, little is known about the upstream regulators of COOLAIR required for cold-induction. Recent reports have shown that an NAC domaincontaining protein, NTL8, and the WRKY transcription factor, WRKY63, can bind to the promoter of COOLAIR and activate its expression (Zhao et al., 2021;Hung et al., 2022). However, whether NTL8 and WRKY63 are necessary for the full extent of COOLAIR induction during vernalization has not been thoroughly addressed. In this study, we identified that CRT/DREs at the 3′-end of the FLC are required for the long-term cold response of COOLAIR. Additionally, we show that CBFs, which accumulate during the long-term winter cold, act as upstream regulators of COOLAIR during vernalization.

Results
A CRT/DRE-binding factor, CBF3, directly binds to CRT/DREs at the 3′end of the FLC The proximal promoter region of COOLAIR is highly conserved among FLC orthologs from A. thaliana relatives (Castaings et al., 2014). Thus, we assumed that the cis-element conferring a long-term cold response would exist within that block. A comparison of the region near the transcriptional start site (TSS) of COOLAIR revealed the conservation of two CRT/DREs among six FLC orthologs from five related species of the Brassicaceae family ( Figure 1A). CRT/DRE, the core sequence of which is CCGAC, is a regulatory element that imparts cold-or drought-responsive gene expression (Baker et al., 1994). CBF1, 2, and 3 encode APETALA 2 (AP2) domain-containing proteins that can bind to CRT/DRE and are usually present in the promoters of cold-and drought-responsive genes Medina et al., 1999). Consistent with this, the Arabidopsis cistrome database and the genome-wide chromatin immunoprecipitation sequencing (ChIP-seq) results from a previous study suggested that CBFs bind to the 3′-end sequence of FLC containing CRT/DREs (Figure 1-figure supplement 1; O'Malley et al., 2016;Song et al., 2021).
We performed an electrophoretic mobility shift assay (EMSA) to confirm this binding using probes harboring CRT/DREs (named DRE1 or 2) from the COOLAIR promoter. The mobility of these two Cy5labeled probes was retarded by maltose-binding protein (MBP)-fused CBF3 ( Figure 1B, Figure 1figure supplement 2). The band shift was competed out by adding an excess amount of unlabeled DRE1 or 2 oligonucleotides. In contrast, competitors containing mutant forms of DRE1 or 2 (DRE1 m or 2 m , respectively) failed to compete ( Figure 1B). We also tested the binding of other CRT/DRE-like sequences in the FLC locus. Two (DREa and b) were present in the FLC promoter, while the other two (DREc and d), were present in the first and last exons, respectively (Figure 1-figure supplement 2). Only DREc competed with the band shift caused by the CBF3-DRE1 interaction ( Figure 1B, Figure 1-figure supplement 2). The presence or absence of bases that determine the binding affinity between CBFs and CRT/DRE could explain the differences in CBF3-binding patterns among CRT/DRE-like sequences at the FLC locus (Maruyama et al., 2004).
Subsequently, a chromatin immunoprecipitation (ChIP) assay using a CBF3-overexpressing transgenic plant, pSuper:CBF3-myc (Liu et al., 2017), was conducted to determine whether CBF3 is associated with the FLC region containing CRT/DREs in vivo. For ChIP assay to show binding affinity of CBF3 protein to the COOLAIR promoter during vernaliation, we used pSuper:CBF3-myc instead of pCBF3:CBF3-myc because endogenous CBF3 transcript level is changed during vernalization as shown afterward. Consistent with the in vitro results, CBF3-myc protein was enriched at the 3′-end of FLC and the first exon where DREc was located ( Figure 1C). Without vernalization (NV), CBF3-myc was highly enriched at both the 5′-and 3′-ends of FLC (P1 and P3), indicating that CBF3 protein can activate FLC under warm conditions, as previously reported (Seo et al., 2009). However, enrichment in the P1 region rapidly disappeared during the vernalization period. In contrast, enrichment on P3 was maintained until 10 days of vernalization (10V) and was subsequently reduced. This is coincident with the expression pattern of COOLAIR, which declines during the late phase of vernalization (Csorba et al., 2014). COOLAIR is one of the CBF regulons in Arabidopsis CBF genes are rapidly and transiently induced upon exposure to cold (Medina et al., 1999). CBF regulons, which are CBF-targeted genes, are correspondingly up-or down-regulated after cold intervals      of a few hours (Gilmour et al., 1998;Liu et al., 1998;Fowler and Thomashow, 2002). As COOLAIR contains CBF3-binding sites in its promoter (Figure 1), we analyzed whether COOLAIR shows CBF regulon-like expression. The expression of genes assigned to the CBF regulon is up-or down-regulated by CBF overexpression under warm conditions (Park et al., 2015). Similarly, we found that the total transcript level of COOLAIR increased in CBF3-overexpressing transgenic plants (pSuper:CBF3-myc) grown at room temperature (22 °C) compared to that in the wild type, Columbia-0 (Col-0) ( Figure 2A). Such upregulation was mainly due to the type I.i or type II.i COOLAIR isoforms. As the levels of both proximal (type I) and distal (type II) variants of COOLAIR were higher in the CBF3 overexpressor than in the wild type, it is likely that COOLAIR transcription, instead of the 3′-processing events, is affected  Relative transcript levels of total COOLAIR and COOLAIR variants were normalized to that of UBC and have been represented as mean ± SEM of three biological replicates. Dots and squares represent each data point. Asterisks indicate a significant difference as compared to the wild type (*, p < 0.05; **, p < 0.01; ***, p < 0.001; unpaired Student's t-test). (B) Expression dynamics of total COOLAIR after short-term (0, 0.5, 1, 1.5, 2, 3, 6, and 24 hr) cold treatment. Wild types (FRI Col) were subjected to 4 °C cold and harvested at each time point. Relative transcript levels of total COOLAIR to UBC were normalized to that of non-cold treated wild type. The values have been represented as mean ± SEM of three biological replicates. The blue shading indicates periods under cold treatment. Significant differences have been marked using different letters (a, b; p < 0.05; one-way ANOVA followed by Tukey's post-hoc test). (C) Transcript levels of total COOLAIR and COOLAIR isoforms in wild type and cbfs-1 mutant before (−Cold) and after (+Cold) a day of 4 °C cold treatment. Relative levels of total COOLAIR and COOLAIR variants were normalized to that of PP2A. Values have been represented as mean ± SEM of three biological replicates. Dots and squares indicate each data point. Significant differences have been marked using different letters (a, b; p < 0.05; two-way ANOVA followed by Tukey's post-hoc test).
The online version of this article includes the following figure supplement(s) for figure 2: by CBF3 (Figure 2A; Liu et al., 2010;Marquardt et al., 2014). It has been reported that the targets of CBF1 or 2 are not entirely identical to those of CBF3, although CBF genes show high sequence similarity (Novillo et al., 2007). Thus, we compared the expression levels of all COOLAIR isoforms in transgenic plants overexpressing CBF1, 2, or 3 (Seo et al., 2009). Overall, all CBF-overexpressing plants showed increased levels of COOLAIR ( Figure 2-figure supplement 1). However, each CBFoverexpressing plant showed a subtly different effect on COOLAIR expression. The CBF2 overexpressor showed the highest level of the proximal COOLAIR isoform, whereas the CBF3 overexpressor showed the highest level of distal COOLAIR. In contrast, the CBF1-overexpressing plants showed a slightly smaller increase in proximal COOLAIR levels and a negligible increase in distal COOLAIR, compared to those in the wild type, Wassilewskija-2 (Ws-2). Such results imply that the three CBFs regulate COOLAIR transcription subtly differently. Another characteristic of CBF regulons is that their expression is activated by a day of cold exposure (Park et al., 2015). Thus, we analyzed if COOLAIR is also induced by short-term cold exposure. We treated wild-type (FRI Col) plants with 0, 0.5, 1, 1.5, 2, 3, 6, and 24 hr of cold (4 °C), then measured the levels of total COOLAIR. The results showed that cold treatment for less than 3 hr failed to induce COOLAIR expression, although CBF transcripts reached a peak at 3 hr of cold exposure ( Figures 2B and 3A; Gilmour et al., 1998). However, after 6 hr of cold treatment, when CBF3 protein levels peaked, COOLAIR was strongly induced (Figures 2B and 3D). Thus, short-term cold-triggered COOLAIR expression dynamics were similar to those of other CBF-targeted genes; that is, the expression was highly increased after CBF transcript levels reached a peak upon cold exposure (Gilmour et al., 1998;Fowler and Thomashow, 2002). To confirm whether CBFs are responsible for the rapid cold response of COOLAIR, we examined COOLAIR induction after a day of cold treatment in the wild type and cbfs-1 mutant, in which all three CBFs were knocked out ( Figure 2C; Jia et al., 2016). All isoforms, as well as total COOLAIR, failed to be induced by short-term cold exposure in the cbfs mutant, although a drastic increase was observed in the wild type. Thus, these results strongly suggest that CBFs are required for COOLAIR induction, even under a short period of cold exposure.

CBFs accumulate during vernalization
We subsequently investigated whether CBFs are also responsible for vernalization-induced COOLAIR activation. It has been reported that COOLAIR is gradually upregulated as the cold period persists and peaks after 2-3 weeks of vernalization (Csorba et al., 2014). However, most studies on CBF expression have been performed within a few days since the function of CBFs has been analyzed only in the context of short-term cold (Gilmour et al., 1998;Liu et al., 1998;Medina et al., 1999;McKhann et al., 2008). Therefore, we investigated the expression patterns of CBFs before and after long-term cold exposure to determine the correlation between the expression of CBFs and COOLAIR during vernalization. As previously shown, the levels of all three CBFs peaked within 3 hr of cold exposure and then decreased rapidly ( Figure 3A). However, CBF levels increased again as the cold period was prolonged (> 20V), suggesting that both short-term and vernalizing cold treatments upregulated CBF expression ( Figure 3B).
Because CBFs exhibit rhythmic oscillations (Dong et al., 2011), we also investigated whether the rhythmic expression of CBFs is affected by vernalization. Wild-type (Col-0) plants were collected every 4 hr under a short-day photoperiod (SD, 8 hr light/16 hr dark) under both NV and 40V conditions. As shown in Figure 3C, although the rhythmic pattern of each CBF was variable, the overall transcript levels of all three CBFs were much higher at 40V than under NV. Consistent with the transcript level, the level of CBF3 protein in vernalized plants was higher than that in NV plants during all circadian cycles (Figure 3-figure supplement 1).
To verify whether the increased transcription of CBF3 led to protein accumulation, we measured the level of CBF3 at each time point of cold treatment using the pCBF3:CBF3-myc plant (Jia et al., 2016). Following the rapid and transient induction of CBF3 under short-term cold conditions ( Figure 3A; Medina et al., 1999), CBF3-myc protein levels peaked within 6 hr of cold exposure and then decreased ( Figure 3D). Thus, the peak of the CBF3 transcript and that of the CBF3 protein showed a gap of few hours. Notably, the peak of the CBF3 protein correlated with the time when COOLAIR was rapidly induced (Figures 2B and 3D). During vernalization, the level of CBF3-myc protein gradually declined until 10V, but then increased again ( Figure 3E). Thus, the level at 40V is similar to that at 1V. When the plants were transferred to room temperature for 4 d after 40V (40VT4), Wild-type plants were subjected to 0, 1, 3, 6, 12, and 24 hr of 4 °C cold treatment. Relative levels of CBF1, 2, and 3 were normalized to that of UBC. Values have been represented as mean ± SEM of three biological replicates. The blue shadings indicate cold periods. Significant differences have been marked using different letters (a-c; p < 0.05; one-way ANOVA followed by Tukey's post-hoc test). (B) Transcript levels of CBF1, 2, and 3 after 20V and 40V. Wild-type plants were treated with 4 °C vernalization under an SD cycle and collected at ZT4. Relative levels of CBF1, 2, and 3 were normalized to that of UBC. Values have been represented as mean ± SEM of three biological replicates. The blue shadings denote periods under cold. Significant differences have been marked using different letters (a-c; p < 0.05; one-way ANOVA followed by Tukey's post-hoc test). (C) Daily rhythms of CBF1, 2, and 3 transcript levels in NV or 40V plants. Col-0 plants grown under an SD cycle were collected every 4 hr between ZT0 and ZT24. Relative transcript levels of CBF1, 2, and 3 were normalized to that of PP2A. Values have been represented as mean ± SEM of three biological replicates. The white and black bars represent light and dark periods, respectively. Asterisks indicate a significant difference between NV and 40V (***, p < 0.001; two-way ANOVA). (D) Dynamics of CBF3 protein level under short-term cold exposure. The pCBF3:CBF3-myc transgenic plants were subjected to 0, 1, 3, 6, 12, and 24 hr of 4 °C cold treatment. CBF3 proteins were detected using anti-myc antibodies. Rubisco was considered the loading control. Numbers below each band indicate relative signal intensity compared to 6 hr. The mean values of two biological replicates are presented. (E) Increase of CBF3 protein level during the vernalization process. The pCBF3:CBF3-myc transgenic plants, subjected to 4 °C vernalization, were collected at ZT4 of the indicated time point. CBF3 proteins were detected using anti-myc antibodies. Rubisco was considered the loading control. Numbers below each band indicate relative signal intensity compared to 1V. The mean values of three biological replicates are presented.  (10V and 20V). For the first frost treatment, wild type and cbfs-1 mutant were treated with an additional 8 hr of −1 °C (+F) under dark, and then the whole seedlings were collected at ZT22 for analysis. All the plants were grown under an SD cycle. The gray shadings denote dark periods. Total COOLAIR levels have been represented as mean ± SEM of three biological replicates in the lower panel. Dots and squares indicate each data point. Relative levels of total COOLAIR were normalized to that of PP2A. Asterisks indicate a significant difference (*, p < 0.05; ***, p < 0.001; two-way ANOVA followed by Tukey's post-hoc test). ns, not significant (p ≥ 0.05).
The online version of this article includes the following figure supplement(s) for figure 4:      the protein level rapidly decreased. These results suggest that CBFs are responsible for the progressive upregulation of COOLAIR, at least during the early phase of vernalization (Csorba et al., 2014).

CBFs are involved in vernalization-induced COOLAIR expression
To clarify whether CBFs are responsible for long-term cold-mediated COOLAIR induction, we quantified COOLAIR levels in wild type and cbfs mutant during vernalization. As shown in Figure 2B and C, the COOLAIR level was considerably increased by a day of low-temperature treatment in the wild type but quickly declined afterward ( Figure 4A). The COOLAIR level increased again after 4V and reached a secondary peak at approximately 10V to 20V. It then decreased during the remaining vernalization period, as similar to previous results (Csorba et al., 2014). The expression dynamics of COOLAIR were similar to those of two well-known CBF targets, COR15A and RESPONSIVE TO DESICCATION 29A (RD29A), although their levels remained high until the end of the vernalization period (Figure 4figure supplement 1). In contrast to the wild-type plants, the cbfs mutant showed severely reduced COOLAIR expression, as well as COR15A and RD29A levels, during all the period of vernalization ( Figure 4A, Figure 4-figure supplement 1). Analysis of RNA sequencing (RNA-seq) data from previous studies also revealed that the levels of total COOLAIR, and proximal and distal COOLAIR variants were reduced in the cbfs mutant in Col-0 or SW ecotype background when exposed to either short-term (3 and 24 hr) or long-term (14V) cold, which supports our results (Figure 4-figure supplement 2; Park et al., 2018a;Song et al., 2021). All these data support that CBFs are necessary to fully induce COOLAIR in the early phase of vernalization.
Consistent with the reduced COOLAIR levels in cbfs mutants, the CBF3 overexpressor, pSuper:CBF3-myc, showed a much higher COOLAIR level than the wild type (Col-0), throughout the vernalization period ( Figure 4B). It is noteworthy that COOLAIR expression in pSuper:CBF3-myc was further upregulated by cold, especially during the early vernalization phase (10V), and then suppressed during the late phase. This result implies that both the transcriptional and post-transcriptional regulations of CBFs are involved in the long-term cold response of COOLAIR.
A recent study showed that the first seasonal frost (< 0 °C) during the winter season strongly induces COOLAIR expression (Zhao et al., 2021). Therefore, we tested whether CBFs are also required for the strong induction of COOLAIR triggered by freezing temperatures. Wild-type and cbfs plants were subjected to NV, 10V, and 20V, with or without an additional 8 hr of −1 °C freezing treatment. Irrespective of the pre-treatment with non-freezing cold, the freezing temperature increased the levels of both COOLAIR and CBF in the wild type ( Figure 4C, Figure 4-figure supplement 3). However, the cbfs mutants showed a much smaller increase in the COOLAIR level than the wild type when exposed to 8 hr of sub-zero cold after 10V and 20V ( Figure 4C). In particular, COOLAIR levels were not elevated by freezing treatment in NV cbfs. Thus, CBFs seem responsible for both the gradual increase of COOLAIR during vernalization and the strong COOLAIR induction triggered by freezing temperatures.
CRT/DREs at the 3′-end of FLC are necessary for CBFs-mediated induction of COOLAIR during vernalization Since CBF3 could bind to the CRT/DREs in the first exon (DREc) and the 3′-end of FLC (DRE1 and 2) ( Figure 1B and C), we investigated whether CRT/DRE is responsible for the CBF-mediated long-term cold response of COOLAIR. To address this, we performed an A. thaliana protoplast transfection assay using the 35S:CBF3-HA effector construct and the pCOOLAIR DRE :LUC reporter construct, in which the 1 kb wild-type COOLAIR promoter was fused to the luciferase reporter gene. The luciferase activity assay showed that CBF3-HA protein activated the transcription of the COOLAIR promoter ( Figure 5A). In contrast, CBF3-HA failed to increase luciferase activity when the mutant version of the COOLAIR promoter with mutations in the DRE1 and 2 sequences (pCOOLAIR DRE m :LUC) was used as a reporter. Additionally, co-transfection of the 35S:CBF3-HA effector and pFLC:LUC reporter construct, which harbors the 1 kb sequence with the promoter and first exon of FLC, did not show increased luciferase activity. These results strongly indicate that CBF3 can activate COOLAIR transcription through DRE1 and 2 located in the COOLAIR promoter, but not via the first exon of FLC.
To further determine whether the 3′-end sequence of FLC is required for cold-induction of COOLAIR expression, we measured COOLAIR expression before and after vernalization in the FLC ΔCOOLAIR mutant (FLC ΔCOOLAIR-1 ), that lacks a 324 bp portion of the COOLAIR promoter region (Luo  Reverse transcription PCR (RT-PCR) analysis showed that neither proximal nor distal COOLAIR variants were detected in this mutant, and the levels did not increase after 14 day cold exposure ( Figure 5B). Using the CRISPR-Cas9 system, we generated additional FLC ΔCOOLAIR mutant lines, FLC ΔCOOLAIR-3, and FLC ΔCOOLAIR-4 ( Figure 5B, Figure 5-figure supplement 1). FLC ΔCOOLAIR-3 has a 685 bp deletion in the COOLAIR promoter region where both CRT/DREs are located. As expected, this mutant failed to express COOLAIR even after 14V, which is similar to the FLC ΔCOOLAIR-1 mutant. In contrast, the FLC ΔCOOLAIR-4 mutant, which bears a 301 bp deletion in the distal COOLAIR promoter region (upstream of CRT/DREs) and a 13 bp sequence replacement downstream of CRT/DREs, showed a normal vernalization-induced increase in all COOLAIR variants, similar to the wild type. This strongly suggests that CRT/DREs located at the COOLAIR promoter are critical for CBF-mediated COOLAIR induction during vernalization. Notably, the low-abundance convergent antisense transcripts (CAS; Zhao et al., 2021) were similarly reduced by 14V in all genotypes, that is, the wild type and FLC ΔCOOLAIR mutants ( Figure 5B, Figure 5-figure supplement 2), suggesting their regulation is independent of COOLAIR.
Although the common COOLAIR TSS was eliminated in FLC ΔCOOLAIR-1 (Figure 5B), the proximal COOLAIR isoform (type I.i) was still slightly detected by quantitative PCR (qPCR), probably due to alternative TSSs ( Figure 5C). However, the expression of the proximal COOLAIR variant in the FLC ΔCOOLAIR-1 mutant was not induced throughout the vernalization period, while the wild type showed type I.i COOLAIR peaks at 1V and 20V ( Figure 5C). Furthermore, CBF3 overexpression did not significantly increase type I.i COOLAIR levels in the FLC ΔCOOLAIR-1 mutant, while it caused a strong increase in proximal COOLAIR in the wild-type background (Col-0) ( Figure 5-figure supplement 3). Taken together, our results strongly suggest that the COOLAIR promoter region containing the two CRT/ DREs is necessary for the long-term cold response of COOLAIR.

CBFs-mediated COOLAIR induction during vernalization is not required for FLC silencing
Finally, we investigated the role of COOLAIR induced by CBFs in the vernalization-triggered FLC silencing process. COOLAIR was reported to facilitate the removal of H3K36me3 from FLC chromatin during vernalization, whereas the other lncRNAs, COLDAIR and COLDWRAP, are required for H3K27me3 deposition (Csorba et al., 2014;Berry and Dean, 2015;Kim and Sung, 2017b;Zhu et al., 2021). Therefore, we compared the enrichment of H3K4me3, H3K36me3, and H3K27me3 on FLC chromatin in non-vernalized or vernalized wild-type and cbfs plants. As reported (Bastow et al., 2004;Sung and Amasino, 2004b;Yang et al., 2014), the enrichments of H3K4me3 and H3K36me3 were reduced, but that of H3K27me3 was increased in FLC chromatin at 40V in the wild type ( Figure 6A-C). However, although COOLAIR levels were significantly reduced in the cbfs mutant ( Figure 4A), vernalization-mediated reductions in H3K4me3 and H3K36me3 levels and an increase in H3K27me3 levels in the cbfs mutant were comparable to those in the wild type ( Figure 6A-C). This result suggests that CBF-mediated COOLAIR induction is not required for vernalization-induced epigenetic changes on FLC chromatin.
Consistent with the epigenetic changes, the FLC level in the cbfs mutant was gradually suppressed during the vernalization process, and the suppression was maintained after 40VT4 similar to that observed in the wild type ( Figure 6D). Such result is consistent with the RNA-seq data previously reported using cbfs mutants in an SW ecotype (Park et al., 2018a). RNA-seq data analysis showed the transcriptional start site of COOLAIR. The thick black arrow below denotes exons of the type I.i COOLAIR variant. The position of the amplicon used for the qPCR analysis is marked with blue arrows. The result of qPCR analysis is presented in the lower panel. Relative transcript levels were normalized to that of PP2A. Values have been represented as mean ± SEM of three biological replicates. The blue shading indicates periods under cold treatment. Asterisks indicate a significant difference, as compared to NV (*, p < 0.05; **, p < 0.01; ***, p < 0.001; two-way ANOVA followed by Tukey's post-hoc test).
The online version of this article includes the following source data and figure supplement(s) for figure 5: Source data 1. Uncropped labeled gel images and the original image files for the RT-PCR results.
Relative Expression (/PP2A) 0 ns ** ns * ns ns ns ns ns Figure 6. Vernalization-triggered epigenetic silencing of FLC is not affected by CBFs-mediated COOLAIR induction. (A-C) Enrichments of H3K36me3 (A), H3K4me3 (B), and H3K27me3 (C) on the FLC locus in NV or 40 V wild-type and cbfs-1 plants. The whole seedlings were collected at ZT4 under an SD cycle. Modified histones were immunoprecipitated with anti-H3K36me3, anti-H3K4me3, or anti-H3K27me3 antibodies. H3K36me3, H3K4me3, and H3K27me3 enrichments of the IP/5% input were normalized to those of ACT7 or FUS3. Relative enrichments have been represented as mean ± SEM of three biological replicates. One-way ANOVA and Tukey's post-hoc test were performed on the ChIP results obtained using two primer pairs corresponding to the FLC nucleation region . Asterisks indicate a significant difference (*, p < 0.05). ns, not significant (P ≥ 0.05). (D) Transcript levels of FLC mRNA in wild-type, cbfs-1 and FLC ΔCOOLAIR-1 plants during and after vernalization. Blue shading denotes cold periods. Relative levels were normalized to that of PP2A. Values have been represented as mean ± SEM of three biological replicates. ns, not significant between wild type and mutants, as assessed using two-way ANOVA followed by Tukey's post-hoc test (p ≥ 0.05). (E) Transcript levels of FLC mRNA in wild type, FLC ΔCOOLAIR-1 , FLC ΔCOOLAIR-3 , and FLC ΔCOOLAIR-4 during and after vernalization. Blue shading denotes cold periods. Relative levels were normalized to that of PP2A, and then normalized to NV of each genotype. Values have been represented as mean ± SEM of three biological replicates. ns, not significant between wild type and mutants, as assessed using two-way ANOVA followed by Tukey's post-hoc test (p ≥ 0.05). (F) Effect of the first frost-mimicking treatment (8 hr freezing) on the FLC mRNA level in wild type and cbfs-1 after vernalization. NV, 10V, 20V, NV +F, 10V+F, and 20V+F plants were subjected to the treatments described in Figure 4C. Relative levels were normalized to that of PP2A. Values have been represented as mean ± SEM of three biological replicates. Dots and squares represent each data point. Asterisks indicate a significant difference (*, p < 0.05; **, p < 0.01; two-way ANOVA followed by Tukey's post-hoc test). ns, not significant (p ≥ 0.05).
that the cbfs mutant in the SW background exhibited reduced FLC levels after 14V, similar to the wild type (Figure 4-figure supplement 2).
Because our results indicate that COOLAIR induction is not required for FLC silencing and the function of COOLAIR in vernalization-induced FLC silencing is still debated Csorba et al., 2014;Zhao et al., 2021;Zhu et al., 2021), we analyzed whether the loss-of-function mutations FLC ΔCOOLAIR-1 and FLC ΔCOOLAIR-3 result in any defect in the vernalization response. In contrast to the transgenic lines used in previous reports, FLC  , which have small deletions in the COOLAIR promoter region, are unlikely to cause any unexpected effects on FLC chromatin. FLC  , which has a deletion outside of the CRT/DRE region, exhibited a gradual decrease and eventual silencing of FLC by long-term cold, similar to the wild type ( Figure 6E). Interestingly, both FLC ΔCOOLAIR-1 and FLC ΔCOOLAIR-3 mutants exhibited similar decreases and silencing of FLC by vernalization ( Figures 5B, 6D and E). These results show that FLC is normally silenced by vernalization regardless of COOLAIR induction by cold.
Consistent with this, the cbfs mutant, compared to the wild type, did not show any difference in the reduction of FLC levels upon exposure to a freezing cold for 8 hr after NV, 10V, and 20V ( Figure 6F). However, the same freezing cold treatment caused the failure of strong induction of COOLAIR in the cbfs ( Figure 4C). These results also support that CBF-mediated COOLAIR induction is not required for FLC silencing by vernalization. As expected from the FLC levels, vernalization-mediated promotion of flowering was comparable in the wild type and cbfs ( Figure 7A and B), indicating that cbfs exhibits a normal vernalization response. Consistently, the flowering time of FLC ΔCOOLAIR-1 and FLC ΔCOOLAIR-3 was accelerated by vernalization, similar to that of the wild type ( Figure 7A, C and D). These results show that FLC silencing occurs regardless of whether COOLAIR is induced during a cold exposure that results in vernalization. In summary, our data strongly support that CBFs, upregulated by long-term cold exposure, induce COOLAIR expression ( Figure 8); however, cold-mediated COOLAIR induction is not required for FLC silencing during vernalization.

Discussion
Winter is challenging for sessile plants: cold weather makes reproduction difficult, and frost threatens survival. Cold acclimation and vernalization, the two types of low-temperature memories, allow plants to overcome this harsh season by making them more viable during winter or flowering in the subsequent warm spring (Thomashow, 1999;Michaels and Amasino, 2000). Although it is well understood how plants recognize and remember short-term cold exposure to increase frost hardiness, there is still much to know about the long-term cold sensing mechanism that establishes the ability to flower.
FLC orthologs from a range of A. thaliana relatives express the vernalization-induced antisense lncRNA, COOLAIR (Castaings et al., 2014). As the proximal sequence blocks within the COOLAIR promoters are conserved, they are likely targets of cold sensor modules. This study showed that CBF proteins act as cold sensors binding to the CRT/DREs in the conserved promoter region, thereby activating the transcription of COOLAIR. Based on the data presented here, we propose a working model (Figure 8) in which a prolonged low-temperature environment gradually increases the levels of CBF proteins, which activate COOLAIR transcription in the early phase of vernalization by binding to CRT/DREs at the 3′-end of FLC. However, as cold period extended, CBF proteins are excluded from the COOLAIR promoter, probably because silencing of FLC chromatin occurs regardless of CBFs level. This may account for a decrease in COOLAIR levels during the later phase of vernalization. Consistent with this, a previous report showed a continuous increase in the COOLAIR level throughout the vernalization period in the vin3 mutant, which has a defect in the maintenance of FLC silencing (Swiezewski et al., 2009). Thus, VIN3-dependent Polycomb silencing is at least partially involved in the release of CBF proteins from the COOLAIR promoter. Although CBFs are upregulated by cold to trigger COOLAIR expression, CBF-mediated COOLAIR induction during vernalization is not solely due to increased CBF levels because the CBF3 overexpression lines still showed COOLAIR induction by vernalization ( Figure 4B). This may indicate that additional factors participate in the regulation of CBFs function. As aforementioned, it has previously been reported that low temperatures cause functional activation of CBFs through the degradation of HD2C which maintains an epigenetically inactive state of downstream targets at warm temperatures (Park et al., 2018b). Moreover, cold triggers the monomerization or stabilization of CBF proteins, thereby promoting their functions Lee et al., 2021). Similarly, such an additional regulatory mechanism of CBF function may be enhanced by long-term winter cold to fine-tune COOLAIR expression.
In addition, other thermosensors are likely to facilitate COOLAIR induction by vernalization. Recently, it was suggested that NTL8, an NAC domain-containing transcription factor, acts as an upstream regulator of COOLAIR (Zhao et al., 2021). NTL8 also binds to the COOLAIR promoter region, which is not far from CBF-binding sites (O'Malley et al., 2016;Zhao et al., 2021). COOLAIR is highly expressed in both dominant mutant alleles, ntl8-Ds, and NTL8-overexpressing transgenic plants at warm temperatures. Thus, it has been proposed that slow accumulation of the NTL8 protein, due to reduced dilution during slower growth at low temperatures, causes COOLAIR induction during vernalization. In addition, a WRKY transcription factor, WRKY63, has been suggested as another inducer of COOLAIR transcription during vernalization (Hung et al., 2022). WRKY63 is enriched in the COOLAIR promoter region, and COOLAIR expression is reduced in the wrky63 mutant, suggesting that WRKY63 functions as a COOLAIR activator. Unlike NTL8, both transcript and protein levels of WRKY63 were upregulated by vernalization, which may lead to an increase in COOLAIR expression. Thus, it would be worthwhile to elucidate whether NTL8, WRKY63, and CBFs synergistically activate COOLAIR transcription during vernalization. The reason why vernalization-triggered COOLAIR induction requires more than one thermosensor is unclear. Presumably, multiple thermosensory modules ensure precise and robust COOLAIR expression during winter, when the temperature fluctuates.
Transcription of COOLAIR has been proposed to govern the FLC chromatin environment in both warm and cold temperatures (Csorba et al., 2014;Fang et al., 2020;Xu et al., 2021b). Previous studies have inferred that COOLAIR is required to remove H3K36me3 from FLC chromatin, especially under low-temperature conditions. This argument is supported by the delayed H3K36me3 removal in vernalized FLC-TEX transgenic plants, where the COOLAIR promoter is substituted by the rbcs3B terminator sequence (Csorba et al., 2014;Marquardt et al., 2014;Wang et al., 2014). In addition, COOLAIR is reported to promote condensation of FRI upon cold exposure . Thus, FRI is sequestered from the FLC promoter as a nuclear condensate, which causes the transcriptional repression of FLC. Therefore, it has been suggested that vernalization-induced COOLAIR prevents the activation of FLC by the FRI complex and causes epigenetic silencing of FLC chromatin through a decrease of H3K36me3 during the vernalization process.
However, there is some controversy as to whether COOLAIR is required for vernalization and the associated FLC silencing. In contrast to the FLC-TEX lines, some T-DNA insertion lines which lack COOLAIR induction by long-term cold exposure showed a normal vernalization response . Likewise, the results from this study show that COOLAIR transcription is not required for the epigenetic silencing of FLC. The cbfs mutant, which exhibits severely reduced COOLAIR induction, shows relatively normal vernalization responses such that flowering time is fairly well accelerated and active histone marks, H3K36me3 and H3K4me3, are reduced in FLC chromatin while the repressive histone mark, H3K27me3, is increased in FLC chromatin similar to the wild type (Figures 6  and 7). Consistently, FLC ΔCOOLAIR-1 , and FLC ΔCOOLAIR-3 mutants, which have almost undetectable levels of COOLAIR, exhibit both FLC suppression and flowering acceleration by vernalization similar to that seen in the wild type (Figures 6 and 7). Moreover, the freezing treatment mimicking the first frost did not affect FLC suppression in the cbfs, although the mutant showed defects in CBF-mediated COOLAIR induction (Figures 4C and 6F). It is noteworthy that our finding that vernalization and the associated FLC silencing do not require COOLAIR expression was obtained using cbfs and FLC ΔCOOLAIR mutant lines rather than the transgenic lines used in previous studies. Thus, we can exclude the possibility that transgenes somehow provide cryptic promoters or cause unexpected epigenetic effects.
COOLAIR is highly conserved among Arabidopsis relatives (Castaings et al., 2014); thus, our finding that COOLAIR is not required for vernalization-induced FLC silencing is surprising. There are several caveats in the studies of COOLAIR function. First, different experimental schemes may affect the vernalization response differently. A recent study of plants grown in fluctuating temperatures designed to mimic natural environments may indicate a role for COOLAIR in vernalization-mediated FLC repression (Zhao et al., 2021), and perhaps fluctuating temperatures that mimic field conditions may have different effects on the vernalization response of cbfs or FLC ΔCOOLAIR mutants as well. However, the experiments in most studies that claimed a role for COOLAIR in vernalization were performed in laboratory conditions with controlled temperatures as in this study. If indeed COOLAIR does have a role in vernalization in specific environmental conditions, the lack of a role for COOLAIR in vernalization in other conditions such as those in this study may be due to functional redundancy between COOLAIR and other FLC repressors such that other repressors, such as COLDAIR, obscure the effect of COOLAIR on FLC silencing, as has been proposed before (Castaings et al., 2014). This issue should be addressed in future studies.
Although short-term and long-term cold memories elicit different developmental responses, both are triggered by the same physical environment. Therefore, it is not surprising that a common signaling network governs both processes. CBF genes have been highlighted for their roles in cold acclimation (Gilmour et al., 1998;Liu et al., 1998;Jia et al., 2016;Zhao et al., 2016). CBFs rapidly accumulate upon short-term cold exposure, thereby activating the transcription of cold-responsive genes, such as COR15A and RD29A Liu et al., 1998). Interestingly, CBFs are also highly accumulated during vernalization and mediate long-term cold-induced COOLAIR expression. Hence, our results indicate that the cold signaling pathways that establish the short-term and long-term cold responses are not sharply distinguishable, as suggested previously (Sung and Amasino, 2004a;Seo et al., 2009).

EMSA
Maltose-binding protein (MBP) and MBP-CBF3 recombinant fusion proteins were induced by 500 μM isopropyl β-D-thiogalactoside (IPTG) in the Esecherichia coli BL21 (DE3) strain. Cell extracts were isolated with a buffer containing 20 mM Tris-HCl (pH 7.4), 1 mM EDTA, 200 mM NaCl, 10% glycerol, 1 mM DTT, and 1 mM PMSF. Proteins were purified from cell extracts using MBPTrap HP column (GE Healthcare, Chicago, IL, USA) with ÄKTA FPLC system (Amersham Biosciences, Amersham, UK). The Cy5-labeled probes and unlabeled competitors were generated by annealing 25 bp-length oligonucleotides. Electrophoretic mobility shift assay (EMSA) was performed as previously described with a few modifications (Seo et al., 2012). For each EMSA reaction, 5 μM of protein and 100 nM of Cy5labeled probe were incubated at room temperature in a binding buffer containing 10 mM Tris-HCl (pH 7.5), 1 mM EDTA, 50 mM NaCl, 5% glycerol, and 5 mM DTT. For the competition assay, the reaction mixtures were incubated in the presence of 100 fold molar excess of each competitor. The reaction mixtures were electrophoresed at 100 V after incubation. The Cy5 signals were detected using WSE-6200H LuminoGraph II (ATTO, Amherst, NY, USA).

Gene expression analysis
Total RNA was isolated from -100 mg of seedlings grown on MS agar plate using TRI Reagent (Sigma-Aldrich, St Louis, MO, USA) or TS Plant Mini Kit (Taeshin Bio Science, Gyeonggi-do, Korea). cDNA was generated using 4 μg of total RNA, 5 units of recombinant DNase I (TaKaRa, Kyoto, Japan), 200 units of RevertAid reverse transcriptase (Thermo Scientific, Waltham, MA, USA), and buffer containing 0.25 mM dNTP and 0.1 μg oligo(dT) 18 . Quantitative PCR analysis was performed using iQ SYBR Green Supermix (Bio-Rad, Hercules, CA, USA) and a CFX96 Real-Time PCR system (Bio-Rad).
To examine COOLAIR expression in the COOLAIR promoter deletion lines, Total RNAs were extracted using the RNeasy Plus Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions, followed by the digestion of residual genomic DNA by the gDNA wiper (Vazyme, Nanjing, China). 1.0 µg RNA from each sample was taken for cDNA synthesis using the HiScript III 1st Strand cDNA Synthesis Kit (Vazyme) with the transcript-specific primers (5′-T GGTT GTTA TTTG GTGG TGTG AA-3′ for COOLAIR class I and 5′-GCCC GACG AAGA AAAA GTAG -3′ for class II; Zhao et al., 2021). Semi-quantitative PCR amplifications were performed, followed by agarose gel separation of PCR products.

RNA-seq data analysis
We retrieved two sets of RNA sequencing (RNA-seq) data deposited in the National Center for Biotechnology Information (NCBI) under BioProject accession codes PRJNA416120 and PRJNA732005. PRJNA416120 contains raw reads from NV, 1V, and 14V wild-type and cbfs plants in a Sweden-ecotype (SW) background (Park et al., 2018a). PRJNA732005 contains reads from 0, 3, and 24 hr cold-treated wild-type and cbfs plants in a Col-0 background (Song et al., 2021). The reads were aligned to the Arabidopsis TAIR 10 reference genome and annotated in ARAPORT 11 using STAR version 2.7.10 a. Isoform estimation was performed using Salmon version 1.6.0.