The phytochrome-interacting factor PIF7 negatively regulates DREB1 expression under circadian control in Arabidopsis.

Transcription factors of the DRE-Binding1 (DREB1)/C-repeat binding factor family specifically interact with a cis-acting dehydration-responsive element/C-repeat involved in low-temperature stress-responsive gene expression in Arabidopsis (Arabidopsis thaliana). Expression of DREB1s is induced by low temperatures and is regulated by the circadian clock under unstressed conditions. Promoter sequences of DREB1s contain six conserved motifs, boxes I to VI. We analyzed the promoter region of DREB1C using transgenic plants and found that box V with the G-box sequence negatively regulates DREB1C expression under circadian control. The region around box VI contains positive regulatory elements for low-temperature-induced expression of DREB1C. Using yeast one-hybrid screens, we isolated cDNA encoding the transcriptional factor Phytochrome-Interacting Factor7 (PIF7), which specifically binds to the G-box of the DREB1C promoter. The PIF7 gene was expressed in rosette leaves, and the PIF7 protein was localized in the nuclei of the cells. Transactivation experiments using Arabidopsis protoplasts indicated that PIF7 functions as a transcriptional repressor for DREB1C expression and that its activity is regulated by PIF7-interacting factors TIMING OF CAB EXPRESSION1 and Phytochrome B, which are components of the circadian oscillator and the red light photoreceptor, respectively. Moreover, in the pif7 mutant, expression of DREB1B and DREB1C was not repressed under light conditions, indicating that PIF7 functions as a transcriptional repressor for the expression of DREB1B and DREB1C under circadian control. This negative regulation of DREB1 expression may be important for avoiding plant growth retardation by the accumulation of DREB1 proteins under unstressed conditions.

Plants respond and adapt to abiotic stresses, such as low temperature, drought, and high salinity, at the molecular and cellular levels. The expression of many genes is induced by these stresses, and their gene products function directly in stress tolerance and in the regulation of gene expression and signal transduction in stress responses. The cis-acting elements that function in stress-responsive gene expression have been analyzed to elucidate the molecular mechanisms of gene expression in response to these stresses (Thomashow, 1999;Chinnusamy et al., 2004;Yamaguchi-Shinozaki and Shinozaki, 2006). The dehydration-responsive element (DRE) containing the core sequence A/GCCGAC was identified as a cis-acting element that regulates gene expression in response to both dehydration and low temperature in Arabidopsis (Arabidopsis thaliana; Yamaguchi-Shinozaki and Shinozaki, 1994). DREs are also found in the promoter regions of many droughtand low-temperature-inducible genes (Thomashow, 1999;Yamaguchi-Shinozaki and Shinozaki, 2006). A similar motif was identified as the C-repeat and lowtemperature-responsive element in promoter regions of low-temperature-inducible genes (Thomashow, 1999;Yamaguchi-Shinozaki and Shinozaki, 2006).
Arabidopsis cDNAs encoding the ethyleneresponsive element-binding factor/APETALA2-type DRE-binding (DREB) proteins, CBF1, DREB1A, and DREB2A were isolated by yeast one-hybrid screens (Stockinger et al., 1997;Liu et al., 1998). We isolated two cDNA clones homologous to DREB1A/CBF3 (DREB1B/ CBF1 and DREB1C/CBF2) and one cDNA homologous to DREB2A (DREB2B; Liu et al., 1998;Shinwari et al., 1998). Both DREB1/CBF and DREB2 proteins bind to the DRE to activate the expression of stress-responsive genes. But DREB1/CBFs are thought to function in low-temperature-responsive gene expression, whereas DREB2s are involved in dehydration-responsive gene expression. Overexpression of DREB1/CBFs driven by the cauliflower mosaic virus (CaMV) 35S promoter increases stress tolerance to freezing, dehydration, and high salinity in transgenic Arabidopsis (Jaglo-Ottosen et al., 1998;Liu et al., 1998;Kasuga et al., 1999). More than 100 downstream target genes of DREB1A have been identified by microarrays (Seki et al., 2001;Fowler and Thomashow, 2002;Maruyama et al., 2004Maruyama et al., , 2009Vogel et al., 2005). Moreover, the levels of many metabolites (e.g. carbohydrates, amines, and organic acids) were reported to increase significantly in both lowtemperature-exposed plants and transgenic lines overexpressing the DREB1A gene (Kaplan et al., 2007;Maruyama et al., 2009). However, overexpression of the DREB1A protein also caused severe growth retardation under normal growth conditions. Use of the stress-inducible rd29A promoter instead of the constitutive CaMV 35S promoter for the overexpression of DREB1A is reported to minimize the negative effects on plant growth (Kasuga et al., 1999).
Expression of the DREB1A gene is rapidly and significantly induced by low temperature, and its induction is gated by the circadian clock (Shinwari et al., 1998;Fowler et al., 2005). Three DREB1 genes lie in tandem in the order DREB1B, DREB1A, and DREB1C, and their promoter sequences from translation initiation sites up to 2500 bp are highly conserved. Shinwari et al. (1998) reported six conserved motifs, boxes I to VI, found in the DREB1 promoters. On the other hand, Doherty et al. (2009) reported seven other conserved motifs, CM1 to CM7, between the DREB1C/CBF2 and ZAT12 promoters. These conserved regions are thought to include cis-acting elements of DREB1 expression. Zarka et al. (2003) identified two sequences that contribute to lowtemperature induction, ICEr1 and ICEr2, in the conserved regions of the CBF/DREB1 genes, which are identical to boxes IV and VI, respectively. On the other hand, some proteins, including transcription factors, have been reported to regulate low-temperature induction of DREB1s directly or indirectly. ICE1/SCRM and MYB15 are basic helix-loop-helix (bHLH)-and Mybtype transcription factors, respectively, that regulate DREB1A expression under low-temperature conditions (Chinnusamy et al., 2003;Agarwal et al., 2006). CAMTA3/AtSR1, a calmodulin-binding transcription factor, binds to the CG-1 element in ICEr2 and activates the expression of DREB1B and DREB1C (Doherty et al., 2009). ICE1 was also reported to function in stomatal differentiation, and CAMTA3 negatively regulates the accumulation of salicylic acid (Kanaoka et al., 2008;Du et al., 2009).
DREB1A expression under unstressed conditions is regulated by the circadian clock (Harmer et al., 2000;Maruyama et al., 2004). In addition, the expression of many low-temperature-inducible genes, including DREB1s and their downstream genes, is also regulated by the circadian clock (Bieniawska et al., 2008). Mutants of pseudoresponse regulators (PRR9, PRR7, and PRR5) and phytochromes (PhyB and PhyD) upregulate the expression of DREB1s and the downstream genes under circadian control (Franklin and Whitelam, 2007;Nakamichi et al., 2009). PRRs and phytochromes are important components of light signaling pathways and the circadian oscillator (Yanovsky and Kay, 2001;Mizuno and Nakamichi, 2005). The light and circadian signaling pathways are also important for the regulation of DREB1 expression. Phytochrome-Interacting Factor (PIF) family proteins are bHLH transcription factors that directly interact with the photoactivated phytochromes under specific light conditions (Castillon et al., 2007). They are negative regulators of the phytochrome-mediated signaling pathway and function in photomorphogenic development, such as hypocotyl elongation and chloroplast development. Most of them accumulate in the dark and rapidly degrade in response to light by the 26S proteasomal pathway. Particularly, the accumulation of both PIF4 and PIF5 mRNA and protein is regulated under light/dark (LD) cycles (Huq and Quail, 2002;Fujimori et al., 2004). Furthermore, they interact with TIMING OF CAB EXPRESSION1 (TOC1)/PRR1, a component of the light signaling pathway or the circadian oscillator. Some of them are suggested to function in the rhythmic growth pattern of hypocotyl elongation under circadian regulation (Nozue et al., 2007;Niwa et al., 2009).
In this study, we analyzed the promoter region of DREB1C and found that the conserved box V sequence functions as a negative regulatory element for circadian expression of DREB1C. The region around box VI contains positive regulatory elements for its lowtemperature-induced expression. We isolated cDNA for one of the PIF family proteins, PIF7, which interacts with the G-box sequence in box V as detected using yeast one-hybrid screens. We characterized the PIF7 gene in plant cells and analyzed the role of PIF7 in the expression of DREB1 genes under circadian control. constructed a chimeric gene with the promoter region from 246 to the translation initiation site of DREB1C, including a 5# untranslated region, fused to the GUS reporter gene as the minimal promoter. Then, four 5# DREB1C promoter regions deleted to 2394, 2256, 2190, and 2113, which were tandemly repeated twice, were fused to the minimal promoter. The fused genes were transformed into Arabidopsis, and at least 10 independent transgenic lines for the expression of the GUS reporter gene of each construct were examined using RNA gel-blot analysis. Transgenic Arabidopsis plants grown for 2 weeks were treated at 4°C for 3 h and used for the analysis of GUS expression (Supplemental Fig. S1). Low-temperature treatment clearly increased accumulation of the GUS transcript in the transgenic plants carrying the 2394, 2256, 2190, and 2113 constructs but did not increase its accumulation in the transgenic plants carrying the minimal promoter. These results indicate that the 65-bp fragment containing boxes V and VI between positions 2113 and 247 of the DREB1C promoter is important for its lowtemperature-responsive gene expression (Supplemental Fig. S1).
To analyze the 65-bp promoter fragment of DREB1C in more detail, we constructed chimeric genes with the fragment (wild type) and its base-substituted fragments, which are tandemly repeated five times, fused to the minimal promoter and the GUS reporter gene (Fig. 1A). We analyzed four mutated 65-bp fragments (M1-M4) with one or two base substitutions. In this experiment, we analyzed more than 10 independent transgenic lines for each construct. The M1 mutation is in a G-box sequence of box V, and the M2 mutation is in the region between boxes V and VI. The M3 and M4 mutations are in the 5# and 3# regions of box IV, respectively. In the transgenic plants carrying the M1 fragment, GUS expression under low-temperature conditions was similar to that of the wild type. However, the GUS transcript had already mildly accumulated in the plants before the low-temperature treatment. The M2 fragment functioned in a manner similar to that of the wild-type fragment. In contrast, neither the M3 nor the M4 fragment allowed for GUS expression at all under low-temperature conditions (Fig. 1B).
Because DREB1A has been shown to be regulated by the circadian clock (Harmer et al., 2000;Maruyama et al., 2004), we tried to detect the GUS transcripts under circadian control. The transgenic Arabidopsis plants carrying the wild-type, M1, and M4 constructs were grown for 2 weeks under 12-h-light/12-h-dark conditions (LD cycle); we analyzed GUS expression every 3 h in the LD cycle and after transferring to continuous light (LL cycle). The GUS reporter gene driven by the wild-type fragment showed rhythmic expression in both the LD and LL cycles, similar to the endogenous DREB1A expression (Fig. 1C). The transcript levels peaked at 1 or 4 h after the subjective dawn (Zeitgeber time [ZT] 1 or 4 h). The GUS reporter gene driven by M4 showed the same rhythmic expres-sion as that driven by the wild type, although the lowtemperature-induced expression was completely lost (Fig. 1C). On the other hand, GUS expression was also rhythmic in the plants carrying M1, but the expression level was much higher than that in the plants carrying the wild type (Fig. 1C). ) and its mutated fragments (M1-M4). The GUS reporter construct contains five copies of the 65-bp fragment and the minimal TATA box of the DREB1C promoter fused to the GUS reporter gene. Nos-T indicates the nopaline synthase terminator. The wild-type G-box sequence is boxed. B, Expression of the GUS reporter gene in response to low temperature. Two representative lines are shown. Each lane was loaded with 7.5 mg of total RNA from 2-week-old seedlings. rRNAs are shown as equal loading controls. N and C indicate the treatments for 3 h at 22°C and 4°C, respectively. EtBr, Ethidium bromide. C, Expression of the GUS reporter gene under circadian regulation. Two-week-old seedlings grown in LD cycle conditions were then transferred to LL cycle conditions. Zero marks the start of the LL cycles. The lines are the same as those shown in B, and each lane was loaded with 15 mg of total RNA.

PIF7 Is Isolated as a Protein Bound to the 65-bp Region of the DREB1C Promoter
To isolate cDNAs encoding DNA-binding proteins that interact with the 65-bp fragment (2113 to 247) of the DREB1C promoter including boxes V and VI, we performed yeast one-hybrid screens. We first constructed a parental yeast strain carrying the dual reporter genes with integrated copies of HIS3 and lacZ with four-times tandemly repeated 65-bp fragments. The yeast cells were then transformed with expression libraries containing cDNA fragments of mRNAs prepared from plants that were untreated or treated at low temperature (4°C) for 1 h. The cDNA fragments were fused to the transcriptional activation domain of yeast GAL4. We screened 1.5 3 10 6 yeast transformants of the mixed cDNA libraries. Thirty-six clones were resistant to 15 mM 3-amino-1,2,4-triazole. Eleven of them induced lacZ activity and formed blue colonies on filter papers containing X-gal (Supplemental Fig.  S2). All 11 positive cDNA clones encoded the bHLHtype transcription factor PIF7, which is reported to be a negative regulator of PhyB signaling (Leivar et al., 2008).

PIF7 Recognizes the G-Box in Box V of the 65-bp Fragment of the DREB1C Promoter
The bHLH-type transcription factors recognize DNA with the consensus sequence CANNTG (Meshi and Iwabuchi, 1995), and PIF7 has been reported to recognize the G-box sequence, CACGTG (Leivar et al., 2008). In the 65-bp fragment of the DREB1C promoter, there is one potential element for PIF7 binding around box V. To confirm that PIF7 recognizes this sequence, the binding ability of the PIF7 recombinant protein was examined by an electrophoresis mobility shift assay. A fragment (residues 1-223) of PIF7, including the DNA-binding domain, was expressed and purified as a glutathione S-transferase (GST) fusion protein in Escherichia coli. The binding ability of the GST-PIF7 fusion protein to the wild-type, M1, and M4 fragments was examined. With the wild-type and M4 fragments, shifted bands were detected and the addition of competitors led to their disappearance, indicating that the fusion protein could bind the wild-type and M4 fragments ( Fig. 2A). In contrast, the fusion protein could not bind the M1 fragment, which might be due to the mutation in the G-box sequence ( Fig. 2A).
The G-box sequence is also found in the DREB1B promoter but not in the DREB1A promoter (TACGTG; Fig. 2B). The binding ability of PIF7 to the fragments containing boxes V and VI of the other DREB1 promoters was also examined using competition experiments. The fragment (2111 to 247) of the DREB1B promoter competed with that of the DREB1C promoter for binding to the PIF7 fusion protein, but that (2114 to 248) of the DREB1A promoter did not (Fig. 2C), suggesting that PIF7 binds to the G-box in the DREB1B and DREB1C promoters but not as efficiently in the DREB1A promoter.

Expression of the PIF7 Gene and Subcellular Localization of the PIF7 Protein
To compare the expression patterns of PIF7 and DREB1, we analyzed their transcript levels under circadian control and low-temperature conditions, respectively. We grew Arabidopsis plants at 22°C under LD cycle conditions for 10 d and then transferred them to LL cycle conditions. We harvested the aerial parts of the plants every 3 h in the LD (211 to 22 h; Supplemental Fig. S3) and subsequent LL (1-46 h) cycles. We measured the transcript levels using quantitative reverse transcription (RT)-PCR and found that the DREB1A and DREB1B genes showed circadian expression both in the LD and LL cycles (Supplemental Fig. S3). Their transcript levels peaked at 7 h after the subjective dawn (ZT 7 h). In contrast, the DREB1C transcripts displayed two peaks at ZT 1 and 7 h in the LD cycle but only one peak was exhibited in the LL cycle, similar to the DREB1A and DREB1B transcripts (Supplemental Fig. S3). On the other hand, PIF7 also showed circadian expression, and its peak occurred at ZT 4 h, similar to PIF4 and PIF5/PIL6, homologous genes to PIF7 (Yamashino et al., 2003). Next, we subjected Arabidopsis plants to a low temperature (4°C) from ZT 2 h under dim light. However, the expression pattern of PIF7 was almost the same as that under room temperature (22°C; data not shown).
To clarify the tissue-specific expression of PIF7, the GUS reporter gene was expressed under the control of the PIF7 promoter in transgenic plants. GUS staining was detected in cotyledons and rosette leaves, but it was not clearly expressed in the roots or hypocotyls (Fig. 3A). To analyze the subcellular localization of PIF7, the PIF7-sGFP fusion gene was expressed under the control of the PIF7 promoter in the pif7-2 mutant background (Leivar et al., 2008), and two generated transgenic lines were used for further analysis (lines R6 and R7; Fig. 3B). In the transgenic Arabidopsis plants, which were grown for 2 weeks under 16-hlight/8-h-dark conditions, GFP fluorescence was clearly observed in the nuclei on the upper side of the leaves but not on the lower side (Fig. 3C). Whereas the fluorescence was diffused in the nuclei under the dark, speckles were formed in the nuclei after transferring to light conditions (Fig. 3D). When plants were kept under continuous light for 24 h, the speckles became unclear but the fluorescence could still be observed. Under the low-temperature condition, there was no significant change in the fluorescence (data not shown).
We also constructed an effector plasmid containing PIF7 fused to the VP16 transcriptional activation domain (AD) under the control of the CaMV 35S promoter as the prey (effector 2; Supplemental Fig. S4A). The GAL4-GUS plasmid, which contains nine copies of a GAL4-binding site and the minimal promoter of Figure 3. Tissue distribution and localization of PIF7. A, Patterns of GUS expression driven by the PIF7 promoter in 2-week-old seedlings: whole plant (left), rosette leaves (middle), and roots (left) are shown. Bars = 1 mm. B, The RT-PCR analysis of PIF7 expression in Col-0, pif7-2, vector control (VC), and two representative lines of complementation plants of pif7-2 (R6 and R7). Act2 was detected as a control in each experiment. C, Fluorescence of PIF7-GFP driven by the PIF7 promoter on the top (top row) and bottom (bottom row) sides of leaves. Images shown are differential interference contrast (DIC), fluorescence by 4#,6-diamino-phenylindole (DAPI) staining, GFP fluorescence, and the merged image. Bars = 20 mm. D, Response of PIF7-GFP to light in 2-week-old seedlings: seedlings in the dark (left) and exposed to white light for 1 h (middle) and 24 h (right). Bars = 5 mm.
CaMV 35S fused to the GUS reporter gene, and the 35S-LUC plasmid, which has the luciferase (LUC) reporter gene under the control of the CaMV 35S promoter, were used as a reporter and an internal control, respectively (Fujita et al., 2005). These constructs were transiently expressed in Arabidopsis mesophyll protoplasts (Supplemental Fig. S4A), and the transfected protoplasts were incubated overnight under white light. Coexpression of PIF7 or PhyB activated GUS activities by more than a factor of 20 (Supplemental Fig. S4B). Coexpression of TOC1 also activated GUS activity, but the level of activation was not as high as that of PIF7 or PhyB. We suspected that the interaction between PIF7 and TOC1 was indirect and mediated by other factors such as PhyB. To investigate the involvement of PhyB in the interaction between PIF7 and TOC1, we performed the experiment using PIF7 mutated in its APB motif, which is the PhyB-interacting domain, as a prey (Khanna et al., 2004;Castillon et al., 2007). This PIF7 mutant had a clearly reduced interaction with PhyB, but its association with TOC1 was not significantly affected (Supplemental Fig. S4C).
To confirm the interaction between PIF7 and some potential binding partners in the plant cells, we conducted a bimolecular fluorescence complementation (BiFC) assay. We constructed PIF7 and the candidates (PIF7, PIF4, PhyB, TOC1, and ICE1) fused to the N-terminal region of yellow fluorescent protein (YFP; YFP N 1-155; SPYNE) or to the C-terminal region of YFP (YFP C 156-239; SPYCE) under the control of the CaMV 35S promoter. These constructs were transfected into Arabidopsis mesophyll protoplasts and transiently expressed. We cotransfected cyan fluorescent protein (CFP) fused to H2B and driven by the CaMV 35S promoter as the positive control of transfection and nuclear localization (Dixit et al., 2006). For PIF7-PIF7 (SPYNE-SPYCE), PIF7-PIF4, PIF7-PhyB, and TOC1-PIF7, YFP fluorescence was merged with CFP in the nuclei; however, fluorescence was not detected with PIF7-ICE1 (Fig. 4).

Transcriptional Activity of PIF7
In our analysis of the DREB1C promoter, mutation of the PIF7-binding site caused activation of GUS expression (Fig. 1, B and C). Therefore, we suspected that PIF7 functions as a transcription repressor for the DREB1C promoter. To examine the transcriptional activity of PIF7, we performed a transactivation assay using mesophyll protoplasts. We constructed fulllength and deletion fragments of PIF7 fused to GAL4-DBD under the control of the CaMV 35S promoter as effectors (Fig. 5A). These constructs were transiently expressed with the reporter and internal control in Arabidopsis mesophyll protoplasts and were incubated overnight under white light. Transient expression of PIF7 (FL; 1-366) reduced GUS activity as compared with that of the DBD control, suggesting that PIF7 can function as a transcriptional repressor. Coexpression of N-terminal deletion mutants of PIF7 (89-366, 89-327, and 328-366) also reduced the activities. In contrast, coexpression of C-terminal deletion mutants of PIF7 (1-327, 1-223, and 1-88) increased GUS activity comparable to DBD (Fig. 5B). These results suggest that PIF7 has a repression domain in the C-terminal region and an activation domain in the N-terminal region.
To address how PIF7 regulates the DREB1C promoter, we conducted a transactivation assay with PIF7 and its interacting factors PhyB and TOC1 as effectors under the control of the CaMV 35S promoter. We used the GUS reporter gene under the control of the 1-kb region upstream from the translation initiation site of DREB1C as a reporter (Fig. 5C). Coexpression of PIF7 reduced GUS activity as compared with the control without the effector. In particular, coexpression of both PIF7 and TOC1 reduced GUS activity significantly more than PIF7 expression alone (Fig. 5D).

DREB1 Expression in the pif7 Mutant
To characterize the role of PIF7 in DREB1 expression, we compared the expression levels of DREB1s between wild-type plants (ecotype Columbia [Col-0]) and the pif7-2 mutant. We grew both plants under LD cycle conditions for 10 d and then transferred them to LL cycle conditions. We harvested the aerial parts of the plants every 4 h in the LD cycle (216 to 24 h; Fig.  6) and subsequent LL cycles (0-36 h) and measured the transcript levels by semiquantitative PCR. In the LD cycle, we could not observe significant expression differences between the wild type and the pif7-2 mutant for the three DREB1 genes (Fig. 6). However, between 12 and 24 h in the LL cycle, expression levels of DREB1s, especially DREB1B and DREB1C, in the mutant were clearly higher than those in the wild-type plants. Expression patterns of DREB1s in the mutant after 24 h were similar to those in the wild-type plants.
We also measured transcripts of other circadiancontrolled genes. In the mutant AtGolS3, which is downstream of DREB1 (Maruyama et al., 2004), a similar expression pattern of DREB1 was displayed in both the wild type and the mutant. In contrast, expression levels of HFR1, which is downstream of PIF4 and PIF5 (Lorrain et al., 2008), were lower in the mutant than in the wild type, except at 16 and 20 h in the LL cycle. Expression patterns of CCA1, which is a component of the central oscillator (Green and Tobin, 2002), did not show significant differences between the wild type and the mutant.
To confirm that the elevated DREB1B and DREB1C gene expression in the mutant was caused by a disruption of the PIF7 locus, we used plants expressing the PIF7-sGFP fusion gene under the control of the PIF7 promoter in the pif7-2 mutant as a means to complement the gene expression (Fig. 3B). In two independent complementation lines (R6 and R7), none of the DREB1 genes exhibited different expression patterns as compared with the wild-type plants (Supplemental Fig.  S5B). Recovery of PIF7 gene function clearly complemented the expression of DREB1 genes. We also performed expression analysis of DREB1s under lowtemperature conditions, because induction of DREB1 expression by low-temperature stress has been reported to be gated by the circadian clock . However, under low-temperature conditions for 2 h, which began at subjective morning (ZT 2 h) or evening (ZT 14 h), no significant difference was observed in the expression of DREB1 (Supplemental Fig. S5C).

DISCUSSION
Previously, we reported that there are six conserved sequences from boxes I to VI in the three DREB1 promoters and hypothesized that these conserved Figure 5. Transactivation activity of PIF7 in mesophyll protoplasts. A, Schematic diagram of the effector and reporter constructs used in the transactivation analysis with the GAL4-DBD. The effector constructs contain the CaMV 35S promoter fused to the GAL4-DBD and PIF7 fragments with or without deletion. The GUS reporter construct contains nine copies of the GAL4-binding site fused to the CaMV 35S minimal TATA box. To normalize for transfection efficiency, the LUC reporter gene driven by the CaMV 35S promoter was cotransfected as a control in each experiment. B, Transactivation or repression domain analysis of PIF7 using deletion constructs. Relative activity indicates the multiples of expression compared with the value obtained with the vector control. Error bars indicate the SD of three technical replicates. C, Schematic diagram of the effector and reporter constructs used in the transactivation analysis using the DREB1C promoter. The effector construct contains the CaMV 35S promoter fused to the PIF7, PhyB, and TOC1 fragments. The reporter construct contains the 1-kb region upstream of the translation initiation site of DREB1C fused to the GUS reporter gene. D, Transactivation analysis of PIF7 with interacting factors of the DREB1C promoter. Relative activity indicates the multiples of expression as compared with the value obtained with the vector control. Error bars indicate the SD of three technical replicates. sequences include cis-acting elements involved in DREB1 expression (Shinwari et al., 1998). In this study, we analyzed the promoter region of DREB1C using transgenic plants and found that a 65-bp region containing boxes V and VI is sufficient for its induction by low temperature (Supplemental Fig. S1A). Moreover, we demonstrated that this 65-bp fragment responded not only to low temperature but also to the circadian clock (Fig. 1, B and D). We analyzed the promoter activity of the mutated fragments and found that the region around the M1 mutation functions as a negative regulatory element for circadian expression and that the regions around the M3 and M4 mutations function as positive regulatory elements for low-temperatureinducible gene expression of the DREB1C gene (Fig. 1,  B and D). In the plants carrying M1, GUS expression levels were much higher than those in the plants carrying the wild type, indicating that the G-box sequence in the fragment negatively regulates DREB1C expression under circadian control. However, we still detected rhythmic expression and significant accumulation of GUS expression at the start of the dark period. This anomalous expression pattern of GUS in the plants carrying M1 suggests the presence of negative regulatory elements that influence circadian expression other than the G-box in the 65-bp fragment.
By yeast one-hybrid screening, PIF7 was isolated as cDNA for a protein that binds to the 65-bp fragment of the DREB1C promoter (Supplemental Fig. S2). PIF7 was previously reported to function as a negative regulator of PhyB signaling in a deetiolated seedling under prolonged red light, similar to PIF3 and PIF4 (Leivar et al., 2008). PIF3 is known to function in anthocyanin accumulation and chloroplast development, and PIF4 plays a role in shade avoidance (Monte et al., 2004;Shin et al., 2007;Lorrain et al., 2008;Stephenson et al., 2009). However, the physiological function of PIF7 has not yet been revealed in detail. PIF7 contains the bHLH-type DNA-binding domain and can bind specifically to the G-box sequence around the M1 mutation in the 65-bp fragment of the DREB1C promoters ( Fig. 2A). Therefore, we hypothesized that PIF7 might regulate the expression of DREB1s, particularly the negative regulation of DREB1B and DREB1C, under circadian control, as the G-box sequence is not completely conserved in the DREB1A promoter (Fig. 2B). Both GUS activity and GFP fluorescence driven by the PIF7 promoter were mainly detected in cotyledons and rosette leaves (Fig.  3, A and C). GUS activity driven by the DREB1 promoters was weakly detected in cotyledons but not at all in rosette leaves under unstressed conditions (Novillo et al., 2007). Expression of PIF7 was regulated by the circadian clock but not by low temperature (Supplemental Fig. S3). These findings are consistent with the hypothesis that PIF7 negatively regulates DREB1 expression under circadian control. This negative regulation of DREB1 expression may be important for avoiding plant growth retardation by the accumulation of DREB1 proteins under unstressed Figure 6. The expression of DREB1 and circadian-controlled genes in the pif7 mutant under circadian control. We analyzed the expression profile in DREB1 and some circadian-controlled genes with semiquantitative RT-PCR under circadian control. Ten-day-old seedlings of wildtype (WT; Col-0) and pif7 mutant plants were grown in LD cycle conditions and then transferred to LL cycle conditions. Zero marks the start of the LL cycles. Black circles and white squares indicate Col-0 and the pif7 mutant, respectively. The relative amount indicates the multiples of expression as compared with the value obtained with the expression peaks. conditions, as overexpression of the DREB1A protein is reported to cause severe growth retardation under normal growth conditions (Kasuga et al., 1999).
Furthermore, we showed that the PIF7-sGFP protein driven by the PIF7 promoter was localized in the nuclei of the leaves and formed speckles in the nucleus in response to short-term light, similar to other PIFs ( Fig. 3D; Leivar et al., 2008). Fluorescence speckles of PIF-GFP were reported to be the sites for the degradation of PIF proteins (Chen, 2008). PIF7-sGFP in the speckles might also be degraded under long-term light. However, diffuse fluorescence of PIF7-sGFP in the nucleus was still detected after 24 h of incubation under white light (Fig. 3D). These results coincide with the report that the PIF7 protein is stable under long-term light, whereas most other PIF proteins are rapidly degraded under the long-term light condition (Castillon et al., 2007;Leivar et al., 2008), suggesting that the PIF7 function is dominant under light conditions. Under the subjective night condition in the LL cycle (12-24 h in LL), expression of DREB1B and DREB1C in the pif7-2 mutant was significantly higher than in the wild type ( Fig. 6; Supplemental Fig. S5A). However, expression levels of these genes did not exceed the expression peak (at 8 h in LL). These results indicate that expression of DREB1B and DREB1C is not repressed in the mutant under the light condition. A PIF7 mutation affected their expression only in LL cycles but not in LD cycles. In the LD cycles, especially the dark conditions, the other PIFs, such as PIF4, seem to function redundantly to repress the expression of these DREB1 genes. Indeed, we showed that PIF7 formed not only homodimers but also heterodimers with PIF4, using two-hybrid and BiFC systems in Arabidopsis protoplasts ( Fig. 4; Supplemental Fig.  S4B). Nakamichi et al. (2009) reported that expression of the DREB1 genes in LD cycles is up-regulated in the triple mutant of PRR9, PRR7, and PRR5 (d975). The expression patterns of DREB1s in the d975 mutant were similar to that of the GUS reporter gene driven by the M1 fragment of the DREB1C promoter ( Fig. 1C; Nakamichi et al., 2009). Therefore, we supposed that these PRRs are upstream factors of PIF7 that regulate the circadian expression of DREB1C. We analyzed the interactions between PIF7 and each PRR protein with the two-hybrid assay using Arabidopsis protoplasts, but direct association was not detected in the plant cells (Supplemental Fig. S4B). On the other hand, TOC1 expression has been reported to be completely repressed in the d975 mutant (Nakamichi et al., 2009). Our two-hybrid and BiFC assays revealed that PIF7 interacts with TOC1 directly in plant cells ( Fig. 4;  Supplemental Fig. S4B). Moreover, coexpression of TOC1 with PIF7 showed stronger repression of the reporter gene driven by the DREB1C promoter (Fig.  5D), suggesting that up-regulation of DREB1s in d975 is partly caused by PIF7 interacting with TOC1.
CAMTA3 has been reported to be a transcriptional activator to regulate the expression of DREB1B and DREB1C under low-temperature conditions, and this protein has been shown to bind the CG-1 element in their promoters (Doherty et al., 2009). The CG-1 element and G-box are proximal, especially in the DREB1C promoter (Fig. 1A). We could not detect a direct interaction between PIF7 and CAMTA3 in the plant cells (Supplemental Fig. S4B) or between PIF7 and other regulators in the low-temperatureresponsive expression of DREB1, such as ICE1 and MYB15. These results suggest that PIF7 functions independently from the transcription factors that regulate DREB1 expression in response to low temperature. These ideas are consistent with the result that PIF7 disruption did not affect low-temperatureinduced expression of DREB1 (Supplemental Fig.  S5C). However, both circadian-and low-temperatureregulated gene expression were reported to be involved in calcium signaling (Catala et al., 2003;Dodd et al., 2006;Doherty et al., 2009). A common upstream factor might coordinately regulate DREB1 expression under circadian control and the low-temperature response. Leivar et al. (2008) reported that PIF7 functions as an activator under the dark conditions in etiolated seedlings. However, in our transient experiments using mesophyll protoplasts, expression of PIF7 fused to GAL4-DBD reduced the relative reporter activity under light conditions, and deletion of the N-terminal region of PIF7 further decreased the relative reporter activity (Fig. 5B). In contrast, deletion of its C-terminal region raised the activity higher than the control. These results suggest that PIF7 contains separate transcriptional activation and repression domains and can Figure 7. A model of the regulation of DREB1 expression under circadian control. PIF7 binds to the G-box sequence in the promoter regions of DREB1B and DREB1C and represses their expression under circadian control. PIF7 forms both homodimers and heterodimers with other PIFs, such as PIF4, in Arabidopsis cells. These other PIFs may function redundantly to repress the expression of these DREB1 genes. PIF7 activity is regulated by one of its binding partners, TOC1, downstream of PRR9, PRR7, and PRR5 (Nakamichi et al., 2009). PIF7 also interacts with and is regulated by PhyB. Another transcriptional regulatory pathway is expected to exist downstream of PRR9, PRR7, and PRR5 to repress the expression of DREB1A. Thus, the expression of DREB1s is repressed by the central oscillator under the unstressed conditions. TFs, Transcription factors. function as a transcriptional activator or repressor. When the reporter gene was fused to the DREB1C promoter, PIF7 reduced the relative reporter activity under light conditions (Fig. 5D), indicating that PIF7 functions as a transcriptional repressor to DREB1 expression. In the pif7-2 mutant, expression of DREB1B, DREB1C, and their downstream genes was upregulated but that of HFR1 was down-regulated ( Fig.  6B; Supplemental Fig. S5A). On the other hand, Franklin and Whitelam (2007) reported that the circadian expression of DREB1 genes was higher under a low-red to far-red light compared with a high-red to far-red light. We showed that coexpression of both PIF7 and PhyB reduced the reporter activity of the DREB1C promoter under light conditions to a greater extent than did PIF7 expression alone (Fig. 5D), suggesting that the interaction between PIF7 and the Pfr form of PhyB enhanced the activity of PIF7 to repress DREB1 expression (Leivar et al., 2008). Transcriptional effects of PIF7 might depend on the downstream genes and/or light conditions.
In conclusion, we showed that the G-box sequence in the DREB1C promoter negatively regulates DREB1C expression under circadian control. We isolated the cDNA for PIF7, which binds to the G-box sequence and functions under circadian control as a transcriptional repressor of DREB1B and DREB1C. PIF7 forms not only homodimers but also heterodimers with other PIFs, such as PIF4, in Arabidopsis cells; these other PIFs might function redundantly to repress the expression of these DREB1 genes. Moreover, PIF7 activity is regulated by the interacting factor TOC1 downstream of PRR9, PRR7, and PRR5 (Nakamichi et al., 2009). PIF7 also interacts with the Pfr form of PhyB, which regulates the activity of PIF7 under the light conditions (Leivar et al., 2008). It is expected that other transcriptional regulatory pathways exist downstream of PRR9, PRR7, and PRR5 to repress DREB1A expression. The expression of DREB1s is strongly repressed by the central oscillator, which prevents the accumulation of DREB1 to avoid growth retardation under unstressed conditions (Fig. 7). A T-DNA insertion line of PIF7, pif7-2 (SAIL_622_G02), was obtained from Syngenta (Sessions et al., 2002). Isolation of the homozygous pif7-2 line was confirmed by PCR with Ampdirect Plus (Shimadzu; http://www.shimadzu. co.jp/) and ExTaq polymerase (TaKaRa; http://www.takara-bio.co.jp/) using the T-DNA left-border primer (5#-GCCTTTTCAGAAATGGATAAATAGCC-TTGCTTCC-3#) and the PIF7-specific primers (5#-TTGAAACAGCTACAA-GCACAAGTACAG-3# and 5#-GATTCGAAGAACTTGAAGGCATG-3#).

Yeast One-Hybrid Screening of Arabidopsis cDNA Libraries
Yeast one-hybrid screens were performed as described previously (Liu et al., 1998). The titer of the cDNA library was 4.5 3 10 6 plaque-forming units. We used the yeast strain YM4271 and the Fast Yeast Transformation Kit (TaKaRa) for yeast transformation. For isolation of the full-length cDNA, 5#-RACE was performed using the 5#-RACE system (Invitrogen; http:// www.invitrogen.co.jp/) according to the manufacturer's instructions.

Fusion Protein Preparation and Purification and Gel Mobility Shift Assay
The coding region of PIF7 was inserted into the pGEX-4T-1 vector (GE Healthcare Life Sciences; http://www.gelifesciences.co.jp/) at the BamHI site. The plasmid was expressed in Escherichia coli Rosetta (DE3) cells and purified using Glutathione Sepharose (GE Healthcare Life Sciences) according to the manufacturer's instructions with minor modifications. The fusion protein was eluted from the beads by incubation with an elution buffer (50 mM Tris-HCl [pH 8.0] and 20 mM glutathione) overnight at 4°C.
Probe labeling and gel mobility shift assays were performed as described previously (Urao et al., 1993) with minor modifications. The oligonucleotides of the wild-type fragments of DREB1A, -1B, and -1C and the mutation fragments, fused to the HindIII site, were amplified by PCR and digested by HindIII. A mixture of 20,000 dpm of 32 P-labeled probe and 2 mg of the fusion proteins was incubated in 13 binding buffer [25 mM HEPES-KOH (pH 7.9), 40 mM KCl, 1 mM dithiothreitol, 1 mM ETDA, 10% glycerol, 1% bovine serum albumin, and 1 mg of poly(dI-dC)] for 30 min at room temperature. The reaction mixtures were resolved by electrophoresis through a 6% polyacrylamide gel in 0.53 Tris-borate-EDTA buffer at 50 V for 2.5 h and visualized by autoradiography. For competition experiments, the competitors were incubated with the fusion protein for 5 min at room temperature prior to the addition of labeled probes.

Construction of Transgenic Plants
For the deletion analysis, the minimal promoter region of DREB1C (198 bp upstream of the translation initiation site) was amplified from Arabidopsis genomic DNA as a template by PCR with PrimeStar HS polymerase (TaKaRa). The fragment was then digested with HindIII and XhoI and inserted into the pGK-GUS vector (Qin et al., 2008). The partial fragments of the DREB1C promoter were also amplified by PCR and inserted into the HindIII site.
The PIF7 promoter region (1,019 bp upstream of the translation initiation site) was amplified from genomic DNA by PCR. The fragment was then inserted into the XbaI and SmaI sites of the pGK-GUS vector. For the GFP fusion constructs expressed under the control of the promoter region, the PIF7 promoter and coding regions were amplified by PCR from genomic DNA and cDNA, respectively, with primer sets containing the overlapping sequence between the promoter and coding regions. The fragments were annealed and polymerized as described previously (Fujita et al., 2005). The fused fragment was inserted into the HindIII and SmaI sites of the pGK-35SsGFP vector, which had its CaMV 35S promoter removed.
All constructs were then introduced into Agrobacterium tumefaciens strain GV3101 with the pSoup helper plasmid (Hellens et al., 2000). Arabidopsis plants were transformed with these constructs as described previously (Liu et al., 1998).

RNA Gel-Blot and Quantitative RT-PCR Analyses
Total RNA from Arabidopsis plants was isolated with the RNAiso reagent (TaKaRa) according to the manufacturer's instructions. The RNA gel-blot analysis was conducted as described (Satoh et al., 2004). Probes of DREB1C and GUS for the RNA gel-blot analysis were prepared as described previously (Shinwari et al., 1998). For cDNA synthesis, 500 ng of total RNA was used with SuperScriptIII reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Real-time RT-PCR analysis was performed as described previously (Sakuma et al., 2006). The PIF7 and DREB1 primers are described in Supplemental Table S1, and the HFR1 and CCA1 primers were as described by Lorrain et al. (2008) and Yamashino et al. (2008), respectively. At4g32680 was quantified as an internal control ). In the quantitative real-time RT-PCR, all reactions were performed in triplicate.

Construction of Transient Expression Assays
For the protoplast two-hybrid system, the transcriptional activation domain was amplified by PCR from pVP16 and inserted into a BamHI site of the pBI35SV vector (pBI35SV-VP16AD; Hollenberg et al., 1995). Effector constructs for the transactivation assays and the protoplast two-hybrid system were constructed with PCR-amplified coding sequences of PIF7, which were inserted into the SpeI site of p35S-564, which contains the GAL4-DBD, or pBI35SV-VP16AD. The pGUS-558 and pBI35SV-LUC vectors were used as the reporter and internal control constructs, respectively (Fujita et al., 2005). For BiFC analysis, full-length fragments of PIF7, PIF4, PhyB, TOC1, and ICE1 were amplified by PCR and cloned into the XbaI site of the pUC-SPYNE and pUC-SPYCE vectors (Walter et al., 2004). As an experimental control, a full-length fragment of H2B was amplified by PCR and cloned into the XbaI and SmaI sites of the pGK-CFP vector ).

Transient Expression Assay Using Arabidopsis Protoplasts
The transient expression assays using protoplasts derived from Arabidopsis mesophyll were performed as described previously (Yoo et al., 2007) with minor modifications. Leaves from plants grown for 6 weeks were cut into 0.5-to 1-mm pieces using a fresh razor blade. Twenty-five to 30 leaves were digested in 25 mL of enzyme solution, which was vacuumed for 15 min, and incubated in the dark for 3.5 h at 22°C to 25°C. The protoplasts were washed and resuspended to approximately 1.2 3 10 5 cells mL 21 . For transfection, a plasmid mixture (3-5 mg each of effector and reporter and 3 mg of internal control) was added to a protoplast and polyethylene glycol solution [40% (w/v) polyethylene glycol 4000 (Fluka; http://www.sigma-aldrich.co.jp/ fluka/first.htm), 0.2 M mannitol, and 100 mM Ca(NO 3 ) 2 ]. The transfected protoplasts were incubated under light conditions (25 mmol m 22 s 21 ) for more than 16 h. LUC activity was assayed using the Bright-Glo Luciferase Assay System (Promega; http://www.promega.co.jp/). Both the luminescence and fluorescence intensities were measured with a multilabel counter (Wallac AVROsx 1420; Perkin-Elmer; http://www.perkinelmer.com/).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Deletion analysis of the DREB1C promoter fragment.
Supplemental Figure S2. Isolation of PIF7, the protein bound to the 65-bp fragment of the DREB1C promoter.
Supplemental Figure S3. Gene expression patterns of PIF7 and DREB1s.
Supplemental Figure S4. Examination of possible interactions with PIF7.
Supplemental Figure S5. Expression of DREB1s under circadian control and low-temperature conditions in the pif7 mutant.
Supplemental Table S1. Oligonucleotides used for the experiments.