The E3 Ubiquitin Ligase HOS1 Regulates Arabidopsis Flowering by Mediating CONSTANS Degradation Under Cold Stress*

Background: Intermittent cold stress delays flowering. This results from interaction between the cold and photoperiodic pathways. Results: CO protein is degraded through an HOS1-mediated ubiquitination mechanism during brief cold treatments. Conclusion: CO acts as a molecular hub that integrates photoperiod and cold signals into the flowering genetic pathways. Significance: The CO-HOS1 module is crucial for fine-tuning of photoperiodic flowering under short term temperature fluctuations. The timing of flowering is coordinated by a web of gene regulatory networks that integrates developmental and environmental cues in plants. Light and temperature are two major environmental determinants that regulate flowering time. Although prolonged treatment with low nonfreezing temperatures accelerates flowering by stable repression of FLOWERING LOCUS C (FLC), repeated brief cold treatments delay flowering. Here, we report that intermittent cold treatments trigger the degradation of CONSTANS (CO), a central activator of photoperiodic flowering; daily treatments caused suppression of the floral integrator FLOWERING LOCUS T (FT) and delayed flowering. Cold-induced CO degradation is mediated via a ubiquitin/proteasome pathway that involves the E3 ubiquitin ligase HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENE 1 (HOS1). HOS1-mediated CO degradation occurs independently of the well established cold response pathways. It is also independent of the light signaling repressor CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1) E3 ligase and light wavelengths. CO has been shown to play a key role in photoperiodic flowering. Here, we demonstrated that CO served as a molecular hub, integrating photoperiodic and cold stress signals into the flowering genetic pathways. We propose that the HOS1-CO module contributes to the fine-tuning of photoperiodic flowering under short term temperature fluctuations, which often occur during local weather disturbances.

The timing of flowering is coordinated by a web of gene regulatory networks that integrates developmental and environmental cues in plants. Light and temperature are two major environmental determinants that regulate flowering time. Although prolonged treatment with low nonfreezing temperatures accelerates flowering by stable repression of FLOWERING LOCUS C (FLC), repeated brief cold treatments delay flowering. Here, we report that intermittent cold treatments trigger the degradation of CONSTANS (CO), a central activator of photoperiodic flowering; daily treatments caused suppression of the floral integrator FLOWERING LOCUS T (FT) and delayed flowering. Cold-induced CO degradation is mediated via a ubiquitin/proteasome pathway that involves the E3 ubiquitin ligase HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENE 1 (HOS1). HOS1-mediated CO degradation occurs independently of the well established cold response pathways. It is also independent of the light signaling repressor CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1) E3 ligase and light wavelengths. CO has been shown to play a key role in photoperiodic flowering. Here, we demonstrated that CO served as a molecular hub, integrating photoperiodic and cold stress signals into the flowering genetic pathways. We propose that the HOS1-CO module contributes to the fine-tuning of photoperiodic flowering under short term temperature fluctuations, which often occur during local weather disturbances.
Light is a key environmental factor that profoundly affects seasonal flowering in plants. In Arabidopsis, flowering is pro-moted under long days (LDs) 2 but delayed under short days (SDs) (1,2). Therefore, plants monitor seasonal changes in light wavelengths and day lengths through photoreceptors and the circadian clock to determine the timing of flowering. The rhythmic light signals generated by the clock are mediated by GIGANTEA (GI) and CONSTANS (CO) to regulate floral integrators, such as FLOWERING LOCUS T (FT) and SUPPRES-SOR OF EXPRESSION OF CO 1 (SOC1), constituting the photoperiodic flowering pathway (1,2).
The coordinated actions of CO and FT play a critical role in photoperiodic flowering. CO promotes flowering by activating FT transcription in the leaf vasculature (3)(4)(5)(6). FT protein acts as a mobile florigen that moves from the leaves to the shoot apex, where it activates a floral meristem identity gene APETALA1 (AP1) (7)(8)(9)(10)(11). Accordingly, mutations of the CO and FT genes result in late flowering under LDs, whereas their overexpression leads to early flowering under both LDs and SDs (3). CO activity is regulated by light signals at both the gene transcriptional and protein levels (1,2). CO transcription is regulated by the clock; its mRNA level reaches a peak around 12 h after dawn and remains high until the following dawn (12)(13)(14)(15)(16). CO protein is stable during the light period but is rapidly degraded in darkness (17). Arabidopsis flowering is thus induced under LDs because light stabilizes the CO protein in the evening when the CO mRNA level is high. Dark-induced CO degradation is mediated by a ubiquitin/proteasome system, in which the RING finger E3 ubiquitin ligase CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1) and members of the SUP-PRESSOR OF PHYA-105 (SPA) protein family are involved (18 -21). CO levels are also regulated by light wavelengths (17). Blue and far-red lights prevent COP1-mediated CO degradation in the evening, causing accelerated flowering (17)(18)(19). Red light delays flowering by promoting CO degradation in the morning in a COP1-independent manner (17).
Three distinct low temperature responses have been defined in the literature, namely vernalization (22,23), ambient temperature (24,25), and short term cold stress (26,27). In Arabidopsis, the effects of cold stress have been investigated by applying intermittent cold treatments, in which plants are exposed to 4°C for several hours during the day (28,29). Intermittent cold delays flowering and activates the floral repressor FLOWER-ING LOCUS C (FLC). Accordingly, the flowering of FLC-deficient mutants is not influenced by intermittent cold (28). Arabidopsis flowering under cold stress is also mediated by FVE (29). The fve mutants, which exhibit enhanced freezing tolerance and late flowering, are insensitive to intermittent cold, and FLC is up-regulated in these mutants. The previous studies indicate that FLC plays a critical role in the signaling cross-link between flowering timing and the cold stress response. However, it remain largely unknown how the flowering genetic pathways are interrelated with cold stress responses and whether FLC is a sole regulator in this process.
The RING finger E3 ubiquitin ligase HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENE 1 (HOS1) negatively regulates cold stress responses (30 -32). Under cold stress, HOS1 triggers the degradation of INDUCER OF CBF EXPRESSION 1 (ICE1), a cold-activated transcription factor functioning upstream of C-REPEAT (CRT)/DEHYDRATION-RESPONSIVE ELEMENT (DRE)-BINDING FACTOR (CBF) genes (32,33). Notably, HOS1-deficient mutants exhibit early flowering under both LDs and SDs, and FLC expression is suppressed in the mutants (30,31,34). Recently, the early flowering phenotype of these mutants was shown to be strongly suppressed by co mutation, and HOS1 was shown to regulate CO abundance during the light period (34), suggesting that the cold signaling attenuator HOS1 regulates photoperiodic flowering via CO.
In this work, we demonstrate that controlled degradation of CO by HOS1 plays a role in the cold regulation of Arabidopsis flowering. Under cold stress, CO was degraded through an HOS1-mediated ubiquitination mechanism, causing suppression of FT and thus delaying flowering. Our data support that CO acts as a molecular link that integrates cold signals into the photoperiodic flowering pathway, providing an adaptive strategy that prevents precocious flowering under fluctuating temperature conditions.

EXPERIMENTAL PROCEDURES
Plant Materials and Growth Conditions-Arabidopsis thaliana lines used in this study were of the Columbia (Col-0) background, unless specified otherwise. Arabidopsis plants were grown in soil or on 1 ⁄ 2ϫ Murashige & Skoog (MS)-agar plates under LDs (16-h light and 8-h dark). White light illumination (120 mol photons m Ϫ2 s Ϫ1 ) was provided by fluorescent FLR40D/A tubes (Osram, Seoul, Korea). In CO-ox and FT-ox transgenic plants, CO-and FT-MYC gene fusions, in which MYC-coding sequences were fused in-frame to the 3Ј end of the CO and FT genes, respectively, were overexpressed driven by the cauliflower mosaic virus 35S promoter. 172-ox transgenic plants overexpressing the MIR172b gene under the control of the cauliflower mosaic virus 35S promoter and the tDNA insertional loss-of-function mutant toe1-2 have been described previously (35). svp-41 and flc-3 mutants have been described previously (28). The autonomous flowering pathway mutants fca-9 and fve-3 have been described previously (35). The photoperiod pathway mutants gi-2 and co-101 have been described previously (35). cop1-4 and cop1-6 mutants have been described previously (18). Additionally, the photoreceptor mutants, such as phyA-211, phyB-9, cry1-2, and cry2-1, have been described previously (28). The early flowering hos1-3 mutant (SALK-069312) was isolated from a pool of tDNA insertion lines deposited in the Arabidopsis Biological Resource Center (Ohio State University).
To examine the effects of cold stress on gene expression, 10-day-old plants grown on MS-agar plates at 23°C under LDs were either maintained at 23°C or exposed to cold (4°C) at Zeitgeber time (ZT) 10 for 6 h, and whole plants were used for extraction of total RNA. ZT represents a standardized 24-h notation of the phase in an entrained circadian cycle, in which ZT0 designates the beginning of day.
Flowering Time Measurement and Intermittent Cold Treatment-Plants were grown in soil at either 23 or 4°C under LDs until flowering. Flowering times were determined by counting the number of rosette and cauline leaves at bolting. Fifteen to 20 plants were counted and averaged for each measurement. For intermittent cold treatments, plants were placed at 4°C for 4 h between ZT8 and ZT12 every day until flowering. White light illumination was provided during cold treatments.
Analysis of Gene Transcript Levels-Gene transcript levels were determined by quantitative real time RT-PCR (qRT-PCR). Preparation of total RNA samples, reverse transcription, and quantitative polymerase chain reaction were carried out based on recently proposed guidelines to ensure reproducible and accurate measurement of mRNA levels (37). Extraction of total RNA from appropriate plant materials and RT-PCR conditions have been described previously (38). Total RNA samples were pretreated extensively with an RNase-free DNase to remove any contaminating genomic DNA before use.
qRT-PCR was carried out in 96-well blocks with an Applied Biosystems 7500 Real-Time PCR system using the SYBR Green I master mix in a final volume of 20 l. PCR primers were designed using the Primer Express Software installed with the system and are listed in supplemental Table S1. The two-step thermal cycling profile used was 15 s at 94°C and 1 min at 68°C. An eIF4A gene (At3g13920) was included in the assays as an internal control for normalizing the variations in cDNA amounts used (39). qRT-PCR was carried out in biological triplicates using total RNA samples extracted from three independent plant materials grown or treated under identical conditions. The comparative ⌬⌬C T method was used to evaluate the relative quantities of each amplified product in the samples. The threshold cycle (C T ) was automatically determined for each reaction by the system set with default parameters. The specificity of the PCR was determined by melt curve analysis of the amplified products using the standard method installed in the system.
Preparation of Recombinant Proteins-The HOS1 cDNA was subcloned into the pMAL-c2X Escherichia coli expression vector (New England Biolabs, Ipswich, MA) containing a maltose binding protein (MBP)-coding sequence. Recombinant MBP and MBP-HOS1 proteins were synthesized in E. coli Rosetta2 (DE3) pLysS strain (Novagen, Madison, WI) and partially purified as described previously (40). Protein synthesis was induced by adding 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside at 20°C overnight. Cells were harvested and resuspended in MBP buffer (20 mM Tris, pH 7.4, 200 mM sodium chloride, 1 mM EDTA, 10 mM 2-mercaptoethanol, 1 mM PMSF, and protease inhibitor mixture (Sigma)). Cell lysates were prepared by three cycles of freezing and thawing followed by centrifugation. The supernatants containing soluble MBP or MBP-HOS1 fusion proteins were stored at Ϫ80°C until use. Fusion proteins were purified according to the manufacturer's procedure. The purity and quantity of recombinant proteins were determined before use by analyzing by 10% SDS-PAGE and staining with Coomassie Brilliant Blue R-250.
CO cDNA was subcloned into the pET-41a(ϩ) E. coli expression vector (Novagen) containing a glutathione S-transferase (GST)-coding sequence. Fusion constructs were transformed into E. coli Rosetta2 (DE3) pLysS strain (Novagen). Induction of protein synthesis and processing of protein samples were carried out as described above with the MBP-HOS1 fusion protein.
In Vitro Pulldown Assays-To examine the interaction of CO with HOS1, CO and HOS1 cDNAs were amplified by RT-PCR and subcloned into the pGBKT7 vector, which contains the T7 RNA polymerase promoter upstream of multiple cloning sequence. 35 S-Labeled polypeptides were produced by in vitro translation using the TNT-coupled reticulocyte lysate system (Promega, Madison, WI). Five l of the 35 S-labeled CO and HOS1 protein solutions was incubated with 1 g of purified recombinant MBP-HOS1 and GST-CO fusion proteins, respectively, bound to MBP beads in 1 ml of binding buffer (50 mM Tris, pH 8.0, 100 mM sodium chloride, 10% glycerol, 1% Triton X-100, 1 mM PMSF, and protease inhibitor mixture) containing 5% skim milk for 12 h at 4°C. The beads were washed 10 times with TN buffer (25 mM Tris, pH 8.0, 100 mM sodium chloride). Bound proteins were eluted with 1ϫ SDS-PAGE loading buffer by heating for 10 min at 70°C and subjected to SDS-PAGE and autoradiography.
Yeast Two-hybrid Assays-Yeast two-hybrid assays were carried out using the BD Matchmaker system (Clontech). The pGADT7 vector was used for GAL4 activation domain, and the pGBKT7 vector was used for GAL4 DNA-binding domain. Yeast strain AH109 (leuϪ, trpϪ, adeϪ, and his), which has chromosomally integrated reporter genes lacZ and HIS under the control of the GAL1 promoter, was used for transformation. CO and HOS1 cDNA sequences were subcloned into pGBKT7 and pGADT7 vectors. Transformation of vector constructs into AH109 cells was performed according to the manufacturer's instructions. Colonies obtained were streaked on selective medium without His, Ade, Leu, and Trp.
Histochemical Assays-For histochemical analysis of GUS activities, plant materials were incubated in 90% acetone for 20 min on ice, washed twice with rinsing solution (50 mM sodium phosphate, pH 7.2, 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide), and subsequently incubated at 37°C for 18 -24 h in rinsing solution containing 2 mM 5-bromo-4chloro-3-indolyl-␤-D-glucuronide (Duchefa, Harlem, The Netherlands). They were then incubated in a series of ethanol solutions ranging from 15 to 80% to remove chlorophyll from plant tissues. The plant samples were then mounted on microscope slide glasses and visualized using a DIMIS-M digital camera (JMTECH, Seoul, Korea).
Effects of Cold Stress and Light on CO Protein Abundance-CO-ox transgenic plants were grown on MS-agar plates for 10 days and subjected to a single cold exposure for appropriate time periods. Whole plants were used for extraction of total proteins. CO proteins were detected immunologically using an anti-MYC antibody (Millipore, Billerica, MA). The intensities of bands on the blots were quantified by densitometry of images using LabWorks image acquisition and analysis software (Media Cybernetics, San Diego).
In Vivo Ubiquitination Assays-For in vivo detection of ubiquitinated proteins, 10-day-old CO-ox transgenic plants were pretreated with 50 M MG132 (Sigma), a potent proteasome inhibitor (32,41), for 24 h and exposed to cold for 24 h. Nuclear extracts were obtained from the CO-ox seedling samples as described previously (42). Nuclear extract and anti-MYC antibodies coupled to agarose beads (Millipore) were mixed in extraction buffer (20 mM Tris, pH 7.4, 100 mM sodium chloride, 0.5% Nonidet P-40, 0.5 mM EDTA, 0.5 mM PMSF, and protease inhibitor mixture) containing 50 M MG132 and were incubated for 2 h at 4°C. The beads were recovered by centrifugation (5000 ϫ g, 4°C, 1 min) and washed for 1 min five times with fresh extraction buffer. The bound proteins were eluted with 1ϫ SDS-PAGE loading buffer by boiling for 5 min and subjected to SDS-PAGE. Immunological analysis was performed using anti-ubiquitin (Santa Cruz Biotechnology, Santa Cruz, CA) and anti-MYC antibodies.
Bimolecular Fluorescence Complementation (BiFC) Assays-BiFC assays were carried out as described previously (43). CO cDNA was fused in-frame to the 3Ј end of a gene sequence encoding the C-terminal half of enhanced YFP in the pSATN-cEYFP-C1 vector (E3082). HOS1 cDNA was fused in-frame to the 3Ј end of a gene sequence encoding the N-terminal half of enhanced YFP in the pSATN-nEYFP-C1 vector (E3081). The YFP N -HOS1 and YFP C -CO vectors were cotransformed into Arabidopsis mesophyll protoplasts by polyethylene glycol-calcium transfection (44). Sixteen hours after transfection, reconstitution of YFP fluorescence was monitored by fluorescence microscopy using a Zeiss LSM510 confocal microscope (Carl Zeiss, Jena, Germany) with the following YFP filter setup: excitation 515 nm, 458/514 dichroic, and emission 560 -615 BP filter.

Cold Stress Disrupts Rhythmic Expression of FT-Intermit-
tent cold is frequently employed to define the effects of cold stress on the timing of flowering in Arabidopsis, because it profoundly delays flowering without plant growth alterations (28,29). FLC plays an important role in cold regulation of flowering DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 43279 (28,29). Notably, FLC expression is induced slowly correlating to days of intermittent cold (28). We therefore suspected that a floral regulatory mechanism other than FLC would rapidly respond to cold stress imposed by intermittent cold treatments in flowering time control.

HOS1 Control of CO Under Cold Stress
To obtain clues as to how cold stress influences flowering time at the molecular level, we systematically examined the effects of intermittent cold on the expression of flowering genes by exposing Col-0 plants to 4°C for 6 h between ZT10 and ZT16 under LDs. This type of cold stress markedly delayed flowering but without any discernible morphological defects (supplemental Fig. S1), as has been reported previously (28,29), supporting that intermittent cold treatment is relevant for the examination of cold effects on flowering time in our assay conditions.
Under cold conditions, the expression of CO and SOC1 genes, which promote flowering, was induced, although FT expression was significantly suppressed (Fig. 1A). In addition, the diurnal oscillation pattern of FT mRNA abundance was disrupted under identical conditions (Fig. 1B). These observations suggested that FT suppression mediates the delayed flowering observed under cold stress.
We next examined the effects of cold stress on the tissuespecific expression patterns of FT using transgenic plants expressing the FT promoter GUS gene fusion (5). Histochemical assays revealed that cold stress significantly reduces GUS activity in the vascular tissues of the basal leaf area (Fig. 1C), where photoperiodic induction of FT occurs (3)(4)(5)(6). Notably, cold suppression of FT was diminished in photoperiod flowering pathway mutants, such as gi-2 and co-101, but was unaffected in other flowering mutants (Fig. 1, D and E). These observations indicate that FT-mediated cold signals are related to photoperiodic flowering. The floral repressor TARGET OF EAT1 (TOE1) was induced under cold conditions (Fig. 1A). However, FT was still suppressed by cold stress in the loss-offunction toe1-2 mutant, indicating that FT is independent of TOE1 in the cold regulation of flowering.
Cold Stress Reduces CO Abundance-Our data indicate that cold stress suppresses FT expression during the later daylight period under LDs, when CO protein accumulates to a high level (12)(13)(14)(15)(16)(17). We therefore hypothesized that cold suppression of FT would result from the reduced abundance of CO protein.
We produced transgenic plants overexpressing the CO-MYC fusion gene (CO-ox), in which the MYC-coding sequence was fused in-frame to the 3Ј end of the CO gene driven by the cauliflower mosaic virus 35S promoter. Ten-day-old CO-ox transgenic plants grown at 23°C were transferred to 4°C, and vice versa, and levels of CO protein were measured throughout the study period. Immunological quantification of CO protein revealed that although the CO level was rapidly reduced after transfer to 4°C ( Fig. 2A), it is significantly elevated when coldtreated plants were transferred to 23°C (Fig. 2B). In contrast, levels of CO mRNA were not discernibly affected by cold temperatures (supplemental Fig. S2, A and B). To further examine the effects of cold stress on CO abundance, the green fluorescence protein (GFP)-coding sequence was fused in-frame to the 3Ј end of the CO gene, and the CO-GFP fusion construct was transformed into Col-0 plants. Green fluorescence was significantly diminished in the root cells at 4°C (supplemental Fig.  S3), indicating that cold stress reduces CO abundance.
At first glance, the reduction of CO abundance under cold conditions appeared to be contradictory to the cold induction of CO transcription (Fig. 1A). Measurements of CO mRNA levels throughout the ZT period revealed that cold induction of CO occurred only during ZT10 -16 (supplemental Fig. S4B), when both FT transcription and CO accumulation are thought to be high (12)(13)(14)(15)(16)(17). In addition, the effects of cold stress on CO mRNA accumulation were significantly reduced in the loss-offunction co-101 mutant (supplemental Fig. S4C), in which functional CO protein is not produced (5). It is likely that under cold conditions, CO transcription is elevated under cold conditions to compensate for reduced CO levels, possibly via feedback regulation.
We next examined the biochemical nature of the reduction of CO abundance under cold conditions by employing the proteasome inhibitor MG132 and protease inhibitors (PIs). CO-ox transgenic plants were exposed to 4°C with or without MG132 and PI treatments, and CO levels were immunologically determined. Although PIs had no discernible effect on CO abundance, the effects of cold stress on CO abundance were largely compromised by MG132 (Fig. 2C), suggesting that CO protein was degraded via the ubiquitin/proteasome pathway under cold stress. Ubiquitination assays in vivo showed that a series of high molecular weight polypeptide bands corresponding to polyubiquitinated CO-MYC fusion proteins was detected in CO-ox transgenic plants grown at 23°C (Fig. 2D). Notably, the levels of polyubiquitinated CO-MYC were significantly elevated in coldtreated transgenic plants. In addition, the kinetic patterns of FT expression were correlated with those of CO abundance in the presence of MG132 under cold conditions (Fig. 2E), indicating that the ubiquitin/proteasome-dependent degradation of CO regulates the FT suppression under cold stress.
To examine whether the FT suppression is a major cause of delayed flowering under cold conditions, intermittent cold treatments were carried out for 4 h between ZT8 and ZT12 under LDs, when CO levels begin to increase (12)(13)(14)(15)(16)(17). By a single cold shock, FT mRNA levels decreased to basal levels (supplemental Fig. S5). In contrast, FLC expression was not discernibly altered, suggesting that immediate suppression of FT expression by cold stress is independent of FLC. Notably, SOC1 mRNA levels were elevated under identical conditions, unlike FT suppression. These data suggested that SOC1 is not directly related to cold regulation of CO abundance.
To further examine whether the cold-induced FT suppression caused by CO degradation is associated with flowering time under cold stress, transgenic plants overexpressing CO or FT (CO-ox and FT-ox, respectively) were treated with intermittent cold. Consistent with the effects of cold stress on CO abundance and thus FT expression, flowering time measurements showed that intermittent cold delayed the flowering of CO-ox plants but did not influence that of FT-ox plants (Fig. 2F), supporting that the effects of cold treatment on CO abundance mediate the cold-induced regulation of flowering time.
Cold-induced CO Degradation Occurs Independently of COP1-The E3 ubiquitin ligase COP1 mediates the dark-induced degradation of CO (17)(18)(19). Therefore, we next investigated whether COP1 is also involved in regulating the coldinduced degradation of CO.
Ten-day-old CO-ox transgenic plants grown at 23°C were either exposed to 4°C in the light or maintained at 23°C in complete darkness, and CO levels were examined immunologically. The CO level decreased rapidly in plants exposed to 4°C in the light (Fig. 3A). However, the rate of decrease was smaller than that in plants maintained at 23°C in complete darkness, suggesting that the effects of darkness exceeded those of cold temperatures on CO abundance.
We also examined the diurnal fluctuation of CO protein levels in CO-ox transgenic plants grown under LDs at either 23 or 4°C. At 23°C, the CO levels increased gradually after dawn and reached a peak between ZT12 and ZT16 (Fig. 3B), as has been reported previously (17)(18)(19). At 4°C, CO levels increased to a degree after dawn but then decreased back to basal levels. Consistent with the reduction in CO levels at 4°C, FT mRNA levels were also reduced under identical conditions (Fig. 3C). In addition, the effects of cold stress on CO abundance were maintained under various light wavelengths and in different photoreceptor mutants (supplemental Fig. S6, A and B, respectively), indicating that the cold-induced CO degradation does not depend on light conditions. We next examined the effects of cold stress on CO abundance in the COP1-deficient cop1-4 mutant. The CO-MYC fusion gene was transformed into the mutant, and the resulting transgenic plants were exposed to 4°C. Immunological assays showed that the COP1 mutation has no effect on CO degradation under cold conditions (Fig. 3D). In addition, FT expression was suppressed in two independent COP1-deficient mutants, cop1-4 and cop1-6, to a similar degree as observed in coldtreated Col-0 plants (Fig. 3E), demonstrating that the cold regulation of CO accumulation and FT transcription occurs independently of COP1.

HOS1 Is Responsible for Cold-induced Ubiquitination of CO-
Next, we investigated how cold-induced degradation of CO is linked with known cold signaling pathways. We found that CO abundance was reduced in the ICE1-deficient ice1-1 mutant and CBF3-overexpressing plants to a similar degree as that observed in Col-0 plants under cold conditions (supplemental Fig. S7A). Diurnal expression patterns of CBF3 and COLD-REGULATED 15A (COR15A) genes were also unchanged in co-101 mutant and CO-ox transgenic plants (supplemental Fig.  S7B). Furthermore, mutations of the genes that mediate temperature regulation of flowering, such as FCA, FVE, and SHORT VEGETATIVE PHASE (SVP) (45)(46)(47), had no effects on the cold regulation of CO abundance (supplemental Fig. S8). These observations suggested that cold induction of CO degradation is not mediated by cold stress or ambient temperature response genes.
The E3 ubiquitin ligase HOS1 is a cold signaling attenuator that mediates the degradation of ICE1 (32). Therefore, we suspected that HOS1 would mediate CO degradation under cold stress.
We first carried out yeast two-hybrid assays to examine whether CO interacts with HOS1 using a series of deletion forms of CO and HOS1 (Fig. 4, A and B, respectively). Notably, we found that CO directly interacts with HOS1 in yeast cells (Fig. 4, C and D). These HOS1-CO interactions were mediated by the C-terminal region of CO (residues 106 -373) containing the CCT (CONSTANS, CONSTANS-like, and TOC1) domain, which is required for COP1-CO interactions (18,19), and the C-terminal region of HOS1 (residues 457-927), which mediates HOS1-ICE1 interactions (24). These HOS1-CO interactions were also verified by in vitro pulldown assays (Fig. 4, E and F) and in vivo BiFC assays (Fig. 4G).
We next crossed CO-ox transgenic plants with a HOS1-deficient hos1-3 mutant in the Col-0 background. The resulting plants were exposed to 4°C, and CO abundance was examined. At 23°C, CO levels were higher (ϳ2-fold) in the hos1-3 background compared with that in the Col-0 background (Fig. 4H). Notably, the effects of cold stress on CO degradation were sig-nificantly reduced in the hos1-3 background. To further examine the role of HOS1 in the cold-induced regulation of CO abundance, the HOS1 gene was expressed under the control of a ␤-estradiol-inducible promoter in CO-ox transgenic plants, and CO levels were analyzed in the presence or absence of the inducer. CO levels rapidly decreased after application of the inducer under cold conditions (Fig. 4, I and J). These observations demonstrated that cold-induced CO degradation largely, if not completely, depends on HOS1.
We also carried out in vivo ubiquitination assays using CO-ox transgenic plants in Col-0 and hos1-3 backgrounds exposed to cold stress. CO ubiquitination was diminished in the hos1-3 background (Fig. 4K), indicating that HOS1 is responsible for the cold-induced ubiquitination of CO.
Based on our observations, we predicted that hos1-3 flowering would not be affected by intermittent cold treatments. Indeed, flowering time measurements revealed that the mutant flowering was not discernibly delayed in response to intermittent cold ( Fig. 4L and supplemental Fig. S9), further supporting the role of HOS1 in the flowering response to cold stress. In addition, the suppressive effects of cold stress on FT expression were markedly reduced in the hos1-3 mutant (supplemental Fig. S10). The incomplete recovery of FT expression in the mutant suggested that FT is not regulated entirely by the HOS1-CO module under cold stress.
Low ambient temperatures (12-16°C) profoundly delay flowering (45)(46)(47)(48). Therefore, we investigated whether the HOS1-CO module mediates the effects of ambient temperature on flowering initiation. Notably, the hos1-3 flowering was not significantly delayed at 16°C, unlike Col-0 flowering (supplemental Fig. S11). However, it is unlikely that the HOS1-CO module is involved in the thermal control of flowering because CO is not required for delayed flowering at low ambient temperatures (see "Discussion") (45,46,48).

HOS1-CO Module Integrates Cold Stress and Light Signals into Arabidopsis
Flowering-Recently, it has been suggested that HOS1 induces CO degradation in response to PHYTO-CHROME B (PHYB)-mediated light signals (19,26). Therefore, FIGURE 3. Cold-induced regulation of CO abundance occurs independently of COP1. A, B, and D, detection and quantification of CO proteins were carried out as described in Fig. 2. C and E, FT mRNA levels were examined by qRT-PCR. Bars indicate the mean Ϯ S.E. Ten-day-old plants grown on MS-agar plants at 23°C were subjected to cold treatments. A, effects of dark and cold on CO abundance. CO-ox transgenic plants were exposed to 4°C in the light (L) or transferred to complete darkness (D). Whole plants were harvested at the indicated time points for preparation of protein extracts. h, hours. B and C, kinetic measurements of CO abundance (B) and FT mRNA levels (C) after cold treatment under LDs. D, cold-induced regulation of CO abundance in the cop1-4 mutant. The CO-MYC fusion gene was transformed into cop1-4 mutant, and the resulting transgenic plants were exposed to 4°C for 12 h. Two independent lines were examined. CO-ox transgenic plant was used as control. E, FT expression in cop1 mutants after cold treatment. The cop1 mutants were exposed to 4°C at ZT10, and whole plants were harvested at the indicated time points for extraction of total RNA. Rub, ribulose-bisphosphate carboxylase/oxygenase (Rubisco).
we investigated whether the HOS1-mediated CO degradation observed under cold stress is associated with PHYB signals.
Thus, we next examined the diurnal patterns of CO accumulation in plants overexpressing the CO-MYC fusion gene in Col-0, phyB-9, and hos1-3 backgrounds grown at either 23 or 4°C. CO abundance was significantly reduced in both Col-0 plants and phyB-9 mutant plants under cold conditions in comparison with those of corresponding plants grown at 23°C (Fig.  5, A and B). However, CO levels were higher in the phyB-9 mutant than in Col-0 plants grown at both temperatures (Fig.  5B). Together with the occurrence of elevated CO levels in the phyB-9 mutant that exhibits early flowering (17), these observations suggested that PHYB regulates CO abundance in the light period, possibly by modulating HOS1 activity (34).
Notably, overall CO levels were higher in the hos1-3 background than those in the Col-0 background but were similar to those in the phyB-9 background when grown at 23°C (Fig. 5, C  and D), supporting that HOS1 regulates CO abundance in the light period. However, under cold stress, although the hos1 mutation largely eliminated the negative effects of cold on CO accumulation, the phyB mutation only slightly influenced these cold effects (Fig. 5, B-D). These observations indicated that PHYB and HOS1 are both involved in regulating CO abundance at 23°C but act independently of cold-induced CO degradation.
To further examine how the cold regulation of CO abundance is related to PHYB signals, CO levels were analyzed in plants grown under red light conditions (Fig. 5E). At 23°C, CO levels were significantly elevated in phyB-9 and hos1-3 backgrounds. Under cold conditions, the effects of cold on CO degradation were maintained to a degree in the phyB-9 background but were significantly decreased in the hos1-3 background (Fig.  5E), suggesting that cold-induced regulation of CO abundance occurs independently of PHYB.
Altogether, our data indicated that HOS1 triggers CO degradation in response to cold stress, linking the cold stress response with Arabidopsis flowering (Fig. 6). In this working scenario, two ubiquitin/proteasome pathways, the light signaling attenuator COP1 and the cold stress signaling attenuator  Fig. 2. J, HOS1 mRNA levels were examined by qRT-PCR (t test, *, p Ͻ 0.01), as described in Fig. 1. K, in vivo detection of ubiquitinated CO in hos1-3 mutant. The assays were carried out as described in Fig. 2D. L, effects of intermittent cold on flowering time in hos1-3 mutant. Plants were treated with intermittent cold for 4 h between ZT8 and ZT12 under LDs until flowering. Rosette leaves of 15 plants were counted for each plant genotype and analyzed for statistical significance using the Student's t test (*, p Ͻ 0.01). Bars indicate mean Ϯ S.E. Rub, ribulose-bisphosphate carboxylase/oxygenase (Rubisco). DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 HOS1, regulate the flowering promoter CO. When plants are exposed to cold stress, CO degradation is induced even in the light, resulting in delayed flowering.

DISCUSSION
Flowering Initiation Under Cold Stress-The effects of short term cold stress on flowering times have often been investi-gated using intermittent cold treatments, which have been shown to delay floral transition, regardless of the time-of-day of cold exposure (this study and Refs. 28,29). The short term cold stress response would be necessary to cope with abrupt fluctuations in ambient temperature by preventing precocious flowering until stable temperature conditions are guaranteed. In this regard, it is likely that this cold stress signaling serves as a means of daily cold sensing and thus apparently differs from the vernalization, which mediates a long term cold responses that promote flowering.
The effects of intermittent cold stress on flowering timing are regulated mainly by FLC, and the flowering of FLC-deficient mutants is largely, although not completely, unaffected by intermittent cold (28). FVE and SVP also play a role in the cold regulation of flowering. Moreover, plant responses to intermittent cold stress are also associated with the well known CBF cold signaling pathway that induces freezing tolerance (26,49,50). The CBF genes are induced by cold stress (26,27), and this induction leads to delayed flowering via FLC (28,49), supporting that delayed flowering is a part of cold stress response in plants.
It has been reported that expression of FLC is down-regulated in the hos1 mutants (30,31,34), suggesting that the HOS1-mediated cold regulation of flowering time is associated with FLC. Meanwhile, it has been proposed that FLC is not the sole regulator of the cold stress response in flowering time control (28). FLC expression is slowly induced with correlation to the duration of cold exposure with only a 2-fold increase in the FLC transcript level after a 2-week intermittent cold treatment (45,47). In addition, flc mutants still show slightly delayed flowering under identical temperature conditions, suggesting that an as-of-yet characterized signaling pathway mediates the rapid response of flowering to low temperatures. . Cold-induced regulation of CO abundance is linked with phyB signaling via HOS1. CO-ox transgenic plants were crossed with Col-0 plants, phyB-9 mutant, and hos1-3 mutant, and the resulting plants were subjected to cold treatment. Detection and quantification of CO protein were carried out as described in Fig. 2. A-C, kinetic measurements of CO abundance after cold treatments under LDs. D, comparison of relative CO levels. Quantification of CO protein levels was carried out using the immunological blots in A-C. E, effects of cold stress on CO abundance under red light conditions. Ten-day-old plants grown on MS-agar plates at 23°C were either maintained at 23°C or exposed to 4°C for 12 h under red light conditions. Rub, ribulose-bisphosphate carboxylase/oxygenase (Rubisco). Under LDs, phyB signals promote CO degradation in the morning via HOS1. In the evening, blue and far-red light signals suppress COP1 activity, leading to elevation of CO abundance. In darkness, CO protein is rapidly degraded through the COP1-mediated ubiquitination pathway. When plants are exposed to cold stress, HOS1 promotes CO degradation. In this regard, CO acts as a molecular link that integrates light and cold stress signals into photoperiodic flowering.
In this study, we demonstrate that the E3 ubiquitin ligase HOS1 modulates CO abundance to control flowering time under cold stress conditions. Accordingly, flowering of the hos1-3 mutant was insensitive to intermittent cold treatments. Thus, we propose that the HOS1-CO module mediates the cold-induced regulation of flowering in response to daily temperature fluctuations by regulating FT transcription. Notably, suppression of FT expression by cold shock does not require FLC ( Fig. 1E and supplemental Fig. S5), showing that HOS1 mediates the regulation of flowering initiation by intermittent cold stress in an FLC-independent manner. It is therefore likely that although the HOS1-FLC pathway is involved mainly in the regulation of flowering under long term intermittent cold stress, the HOS1-CO-FT module could be functionally specialized in detecting short term cold stress signals, which may contribute to the cold regulation of photoperiodic flowering (see below). Further investigation using various mutants in flowering time and cold response assays would be required to provide insights into the genetic networks through which the cold-induced regulation of flowering is linked with the cold stress response in Arabidopsis.
Here, we report that an HOS1-mediated ubiquitination mechanism links the cold stress response with photoperiodic flowering. We demonstrated that, under cold stress, cold-activated HOS1 induces CO degradation, causing FT suppression and delayed flowering. Our findings indicated that cold signals delay flowering by triggering CO degradation and thus disrupting the diurnal oscillation of FT expression even under inductive photoperiodic conditions. Based on previous studies and our own data, we propose that the regulation of CO abundance by HOS1 serves as a molecular scheme that incorporates cold stress and photoperiodic signals into FT-mediated photoperiodic flowering (Fig. 6). The incomplete recovery of CO abundance and FT expression in the hos1-3 mutant under cold conditions suggested that HOS1-independent cold signaling pathways, which may include those regulating FT transcription, are also involved in the cold regulation of photoperiodic flowering.
What is the physiological significance of the cold-induced regulation of CO abundance in flowering time control? We found that the cold-induced CO degradation is a relatively rapid process occurring within an hour after exposure to cold temperatures. In annual plants that flower in spring, the HOS1-CO module may monitor short term changes in ambient temperature and delay flowering until spring has truly arrived. It is thus likely that the HOS1-CO module fine-tunes the timing of photoperiodic flowering (Fig. 6).
It is remarkable that the photoperiodic flowering promoter CO is regulated at the protein level by HOS1, a well known E3 ubiquitin ligase that attenuates cold responses (30,31). In addition, HOS1 is also involved in PHYB signaling to regulate CO abundance during the light period (34). Furthermore, CO abundance is regulated by photoperiodic signals via the COP1-CO module (17)(18)(19). Therefore, it is evident that CO activity is regulated at the protein level by at least two distinct ubiquitination pathways that incorporate photoperiodic and temperature signals into the flowering genetic pathway. CO activity is also regulated at the gene transcriptional level by GIGANTEA-mediated circadian rhythms (14 -16). An external coincidence model for photoperiodic flowering suggests that flowering is promoted under LDs when the maximal CO level coincides with the high CO transcription in the evening (13,17,50). Our findings further support the external coincidence model, in which photoperiodic and temperature signals are incorporated via CO.
Role of HOS1 in Thermosensory Flowering-Our data showed that HOS1 also plays a role in ambient temperature regulation of flowering. Moreover, CO is not associated with thermosensory flowering (45,46,48). It is therefore likely that the HOS1mediated regulation of thermosensory flowering is controlled by signaling pathways independent of CO.
Interestingly, it has recently been shown that the basic helixloop-helix transcription factor PHYTOCHROME INTER-ACTING FACTOR4 (PIF4) promotes flowering by directly activating FT at warm ambient temperatures (48). The PIF4-FT pathway may be responsible for the early flowering of the hos1-3 mutant even under SDs, when CO abundance is very low. Notably, PIF4 abundance is regulated by ambient temperature, which is prominent particularly under red light conditions, possibly by an as-of-yet unidentified ubiquitination pathway (36,51). It is therefore possible that PIF4 abundance is regulated by HOS1-mediated temperature signals, a hypothesis that is also supported by the role of HOS1 in ambient temperature control of flowering (supplemental Fig. S11). It will be interesting to examine whether HOS1 interacts with PIF4 during the thermal control of flowering. Together with the signaling linkage between HOS1 and PHYB (19,34), these observations indicate that HOS1 is linked directly and indirectly with various light and temperature signaling pathways in both COdependent and -independent manners.