A HY5-COL3-COL13 regulatory chain for controlling hypocotyl elongation in Arabidopsis

CONSTANS-LIKE (COL) family members are commonly implicated in light signal transduction during early photomorphogen-esis. However, some of their functions remain unclear. Here, we propose a role for COL13 in hypocotyl elongation in Arabidopsis thaliana. We found that COL13 RNA accumulates at high levels in hypocotyls and that a disruption in the COL13 function via a T-DNA insertion or RNAi led to the formation of longer hypocotyls of Arabidopsis seedlings under red light. On the contrary, overexpression of COL13 resulted in the formation of shorter hypocotyls. Using various genetic, genomic, and biochemical assays, we proved that another COL protein, COL3, directly binds to the promoter of COL13, and the promoter region of COL3 was targeted by the transcription factor LONG HYPOCOTYL 5 (HY5), to form an HY5-COL3-COL13 regulatory chain for regulating hypocotyl elongation under red light. Additionally, further study demonstrated that COL13 interacts with COL3, and COL13 promotes the interaction between COL3 and CONSTITUTIVE PHOTOMORPHOGENIC1 (COP1), suggesting a possible COP1-dependent COL3-COL13 feedback pathway. Our results provide new information regarding the gene network in mediating hypocotyl elongation.

BBXs family has 32 members (Kumagai et al., 2008), which are divided into five groups based on whether their respective proteins contain one or two BBX motifs and whether or not they possess a CCT domain (Khanna et al., 2009). BBX family members, some of which have been characterized and implicated in light signal transduction during early photomorphogenesis (Cheng & Wang, 2005;Graeff et al., 2016;Li et al., 2014;Park et al., 2011;Preuss et al., 2012;Wang, Guthrie, Sarmast, & Dehesh, 2014;Xu, Jiang, Li, Holm, & Deng, 2018;Xu et al., 2016;Yang et al., 2014). The first BBX protein identified in Arabidopsisthaliana was Constructs for GUS-staining assays: To construct the pCOL13 -GUS-2000 construct, a region comprising the 2000-bp promoter sequence of COL13 was cloned and inserted into the pBI121 vector between the Hin dIII and Bam HI sites. To construct the pCOL13 -GUS-2812 construct, a region comprising the 2812-bp promoter sequence of COL13 was cloned and inserted into the "1301 vector" between the Sac I and Sal I sites. To construct the pCOL3 -GUS construct, a region comprising the 967-bp promoter sequence of COL3 was cloned and inserted into the pBI101 vector between the Hind III and Xba I sites.
Constructs for GFP, CFP, and YFP assays: To construct theCOL13 -GFP construct, the full-length COL13 coding region was cloned and inserted into the pBEGFP vector between the Xba I and Kpn I restriction sites. To construct the COL3-CFP, COP1-CFP, COL13-CFP, COL3-YFP, COP1-YFP, and COL13-YFP constructs, the full-length coding regions of COL3 , COP1, and COL13 were cloned and inserted into the pBluescript II Phagemid vector (Y. Liu et al., 2016), as appropriate.
Constructs for Co-IP assays: To construct the 35S:COL3-HA construct, the full-length COL3 cDNA was cloned and inserted into the pCAMBIA1390-HA vector (Fang et al., 2019).
Constructs for dual-luciferase assays: Fragments of the COL3 or COL13 promoter were cloned into pGREEN0800-LUC to generate reporter vectors. A modified pBluescript vector (pBS) was used as an effector (Han et al., 2017).
The primers used are listed in Supplementary Table 1.

Plant transformation
Constructs in binary vectors were introduced into theAgrobacterium tumefaciens strain LBA4404 and transformed intoArabidopsis wild type (WT) or mutant plants by the floral-dip method (Clough & Bent, 1998). Approximately 30 T1 transgenic plants for each transgene were screened on MS medium supplemented with the appropriate antibiotics, and phenotypic analyses were performed on T2 or later generations.
Semi-quantitative reverse transcriptase (RT)-PCR and quantitative reverse transcriptase (qRT)-PCR Semi-quantitative RT-PCR and qRT-PCR analyses were performed as previously described (Zhang, Liu, et al., 2014). RNA was extracted from 5-d-old seedlings. Three biological and three technical repetitions were performed for each combination of cDNA samples and primer pairs. The primers used are listed in Supplementary Table 1.

Dual-luciferase assay
Protoplasts were isolated, and the dual-luciferase assay was performed as previously described (Han et al., 2017). Transformed protoplasts were incubated at room temperature (˜22) for 20-22 h and luciferase activities were measured using the dual-luciferase assay system (Dual-Luciferase ® Reporter Assay, Promega, United States) according to the manufacturer's instructions. Firefly luciferase activity was normalized to Renilla luciferase activity. Three biological replicates were performed for all experiments.

Electrophoretic mobility shift assay (EMSA)
EMSA was performed as previously described with the LightShift Chemiluminescent EMSA Kit (Thermo Scientific, United States) (Han et al., 2017). The dual-luciferase assay mapped the COL3 binding site to a 1059 bp region of the COL13 promoter, located between 676 and 1675 bp upstream of the transcription start site (ATG). This promoter region was used as a 5' end biotin-labeled probe and the same fragment, but unlabeled, was used as a competitor. To investigate the core-binding motif of the 1059 bp region, a series of EMSAs involving deletions of this region were performed. We divided the 1059 bp promoter sequence into five overlapping regions: -1675 to -1391 bp (probe 1), -1421 to -1184 bp (probe 2), -1201 to -1040 bp (probe 3), -1060 to -868 bp (probe 4), and -898 to -616 bp (probe 5). The sequence of probes is listed in Supplementary Table 2.

Yeast assays
Yeast one-hybrid assays were performed as described by Lin et al. (2007). The yeast two-hybrid and three-hybrid assays were performed using a Clontech kit (PT3024-1 (PR973283)). For yeast two-hybrid assays, the bait vectors (pGBKT7 plus candidate genes) and prey vectors (pGAGT7 plus candidate genes) were transformed into the gold and Y187 yeast strains, respectively, and then each colony was picked for mating, and the mating solution was sprayed on SD/-Trp/-Leu/X-α-Gal/AbA (DDO/X/A) agar plates, and positive results were confirmed by growing them on SD/-Ade/-His/-Trp/-Leu/X-α-Gal/AbA (QDO/X/A) agar plates. For the yeast three-hybrid assay, COP1-pGADT7 and COL3-COL13-pBbidge constructs were co-transformed into the gold yeast strain. Each colony was picked and grown in SD/-Leu/-Met/-Trp and SD/-Leu/-Trp solutions, respectively. Normalized Miller units were calculated as a ratio of α-galactosidase activity in yeast. For all yeast assays, we used empty vectors as controls.
A 0.5 g sample of Arabidopsis seedlings were ground in liquid nitrogen, 1 ml (2 volumes) of lysis buffer was added (50 mM Tris pH 8.0, 150 mM NaCl, and 1 mM EDTA, containing 0.01 volume of 1× Protease Inhibitor Cocktail [Sigma]) or protease inhibitors (1 mM PMSF, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin). It was spun at 16,000 g for 20 min at 4°C, and the supernatant was transferred to a new microcentrifuge tube. Protein concentration was determined using the Bradford reagent (Bio-Rad) and bovine serum albumin was used as a standard. For immunoprecipitation (IP) reactions, 1 mg of total protein was incubated with 5 μl of polyantiserum (or pre-immune serum) in a total volume of 1 ml of lysis buffer and incubated for 1 h to overnight at 4°C, and the sample was mixed by inversion. During incubation, a 40 μl wash per reaction of Protein-A-Agarose beads with 1 ml of cold lysis buffer or phosphate-buffered saline (PBS) was used by gentle vortexing and spinning in a microcentrifuge at 12,000 gfor 30 s. The supernatant was carefully removed by aspiration. This process was repeated twice. A 50 μl sample of fresh lysis buffer or PBS was added. The beads were ready to be used. To precipitate the immune complexes, 50 μl of Protein-A-Sepharose (Amersham Pharmacia Biotech) slurry was added and incubated for 2 h to overnight at 4°C, mixing the sample head-over-tail. The beads were washed in the microcentrifuge tube by resuspension in 700 μl of lysis buffer. It was then centrifuged at 12,000 g for 15-30 s at 4°C. The effluent was discarded. Steps 7 and 8 were repeated three more times. Next, 40 μl of 2× SDS sample buffer was added to the beads. The beads were gently mixed (no vortexing) to avoid spreading the beads on the column walls. The sample was heated at 95°C for 5 min to ensure that the Protein-A-Sepharose complex was within the heating well. The sample was then centrifuged at 12,000 g for 30 s. A sample of 10 μl of the eluted immunoprecipitate was loaded on an SDS-PAGE gel (Bio-Rad). For standard western blotting analysis, 50 μg of total protein was loaded. Proteins were electroblotted onto PVDF membranes (Amersham), blocked for at least 1 h at room temperature in PBS-Tween containing 5% (wt/vol) non-fat dried milk. Primary antibodies were added to PBS-Tween containing 5% (wt/vol) non-fat dried milk and incubated for 1 h. Blots were developed using an ECL Kit from Amersham Pharmacia Biotech.

Histochemical GUS staining, GFP, and fluorescence resonance energy transfer (FRET) experiments
Histochemical GUS staining, GFP microscopy, and FRET were performed as previously described with some modifications (Datta et al., 2006;Hou, Wu, & Gan, 2013;Zhang, Zhang, et al., 2014). For the GUS-staining assay, the young seedlings (4 d after germination) were fixed and incubated in GUS-staining solution for 24 h at 37°C. The stained samples were then cleaned with 75% ethanol and observed under a dissecting microscope.
For the GFP assay, the fusion COL13 -GFP constructs were transformed into protoplasts for transient expression as previously described ). Ten stable transgenic plants with COL13 -GFP were obtained using the floral-dip method. Photographs of GFP were taken using a confocal microscope Olympus.
For the FRET assay, images were acquired using an Olympus confocal microscope, and protoplasts were visualized 16 h after transformation. The CFP was excited by a laser diode 405 laser and YFP, by an argonion laser. The target regions were bleached with 100 iterations using an argon-ion laser at 100% power.

Statistical analysis
Experimental data were analyzed using an ANOVA, and the statistical significance of any differences between treatments was tested using Duncan's test or t-tests. All analyses were conducted using SPSS for Windows.

COL13 RNA accumulates at high levels in hypocotyls
By searching the gene expression information in the Arabidopsis Information Resource (TAIR) database (Klepikova, Kasianov, Gerasimov, Logacheva, & Penin, 2016), we found that COL13 (AT2G47890 ) is highly expressed in the hypocotyl. qRT-PCR analysis confirmed thatCOL13 was expressed in most plant organs, with higher expression in the hypocotyl and stem (Fig. 1a). To determine the spatial patterns of COL13 expression in more detail, transgenic lines expressing GUS driven by the 2812 bp COL13 promoter fragment were generated. As shown in Fig. 1b, GUS expression was predominantly in the hypocotyl.

COL13 regulates hypocotyl elongation under red-light conditions
To characterize the role of COL13 in plants, we obtained the corresponding Arabidopsis T-DNA insertion mutant (GK-657F04-023194, termed col13 in the following; Fig. S1a) from GABI-Kat, Max Planck Institute for Plant Breeding Research (Rosso et al., 2003). The mutation was verified by PCR (Fig. S1b), which amplified theSulphonamide(sul ) gene by using the primers listed in Supplementary Table 1. To confirm that the phenotype of the col13 mutant was indeed caused by disruption of the COL13 gene, we generated COL13 overexpression (OX) (Fig. 2a) and COL13 RNAi transgenic lines (Fig. 2b) for comparison.
To examine whether COL13 was involved in light responses, the WT, COL13 RNAi, andcol13 seedlings were germinated and grown under different light wavelengths (white, red, blue), as well as under dark conditions. As shown in Fig. S1c, under white or red light, the COL13 RNAi andcol13 seedlings had longer hypocotyls than that of the WT, whereas in blue light or dark conditions, the hypocotyl length of all seedlings was not significantly different. Therefore, our research focused on red light. For further study, COL13 OX, COL13 RNAi, col13, and WT seedlings were germinated and grown under red light. We found that the COL13 OX seedlings had shorter hypocotyls than the WT seedlings under red light (Fig. 2c,d). In contrast, the COL13 RNAi and col13 seedlings had longer hypocotyls than the WT seedlings under the same conditions (Fig. 2c, e). These findings suggested that COL13 acts as a positive regulator of red-light-mediated inhibition of hypocotyl elongation.

Genetic interaction and physiological characterization of hypocotyl elongation
Given that the phyB , hy5, col3, and cop1 mutations can affect hypocotyl elongation under red-light conditions (Datta et al., 2006;Lee et al., 2007;Peter H Quail, 2002;, we investigated the expression of COL13 in the absence ofPHYB , COL3 , HY5, and COP1 . Semi-quantitative RT-PCR and qRT-PCR analysis revealed that the expression of COL13 in phyB , col3, or hy5 knockout plants was significantly reduced compared with that of the WT, whereas the expression of COL13 in the cop1 mutants was increased (Fig. 3a, b). As the expression of COL13 decreased the most in the col3 mutant, we generated transgenic lines expressing GUS under the control of the COL13 promoter with the col3 mutant background. Interestingly, although the COL13 promoter was active in the hypocotyls and cotyledons in the WT seedlings, GUS expression was not detected in the hypocotyl in the col3 mutant background (Fig.  3c).
To understand the functional relationship and genetic interaction between COL13 and COL3 and their role in the regulation of hypocotyl growth, we generated a col13 col3 double mutant and examined hypocotyl length under red-light conditions. Given thatcol13 was in Col-0 and col3 was in the WS background, crossing lines from different backgrounds would likely affect hypocotyl length. To reduce the effect of the background, we used the F1 hybrid of Col-0x WS as the WT. We found that, although hypocotyl length in the double-mutant col13 col3 was longer than in the WT seedlings, it was not significantly different from hypocotyl length in the single mutants, col13 or col3. (Fig. 3d). To confirm this result, we created the RNAi lines of COL13 in the col3 mutant background (Fig. 3e), and we obtained the same result as in Fig.  3d. Additionally, we also generated a COL13 -OX line in thecol3 mutant background and showed that the hypocotyl length in this strain was similar to that of the WS and significantly shorter than that of the col3 mutant (Fig. 3e). In other words,COL13 overexpression rescued the phenotype exhibited by thecol3 mutant. Taken together, our results suggest thatCOL13 might be downstream of COL3 in the red-light-mediated signaling pathway.

HY5-COL3-COL13 regulatory chain
Based on the genetic data, the col3 hy5 double mutant behaved like the hy5 mutation (Datta et al., 2006), and COL13 might be downstream of COL3 in regulating hypocotyl elongation. In addition, it has been well established that HY5 binds to "CACG" and "GACGTG" in the promoters of light-responsive genes (Lee et al., 2007;Nawkar et al., 2017). COL3 is a light-responsive gene, and we examined the promoter sequence of the COL3 gene and found that it contains two "CACG" and one "GACGTG". Thus, we hypothesized that there would be an HY5-COL3-COL13 regulatory chain for controlling hypocotyl growth. To test this hypothesis, the HY5 and COL3 coding sequences, as well as a deletion series of the COL13 promoter, were cloned into the dual-luciferase system (Fig. 4a). As shown in Figure 4b, these dual-luciferase experiments and yeast-one hybrid assays confirmed the ability of HY5 to bind to the COL3 promoter and COL3 to bind to the COL13 promoter. Additionally, these experiments also mapped the COL3 target regions (1059 bp) to between -1675 bp and -616 bp of the COL13 promoter (Fig. 4b). To investigate the core-binding motif of the 1059 bp region, a series of EMSAs involving deletions of this region were performed. We divided the 1059 bp promoter sequence into five overlapping regions (Fig. 4c): -1675 to -1391 bp (probe 1), -1421 to -1184 bp (probe 2), -1201 to -1040 bp (probe 3), -1060 to -868 bp (probe 4), and -898 to -616 bp (probe 5), and showed that probe 2 (-1421 to -1184 bp) was essential for binding of COL3 to the COL13 promoter (Fig. 4d). The in vivo interaction of COL3 with probe 2 was further confirmed by EMSA competition experiments that were conducted by adding excess amounts of the competitor (5-, 10-, and 25-fold higher amounts) (Fig. 4e). Next, we analyzed the sequence of probe 2, and found that in the 238 bp fragment, there were three light responsive elements (ATCT-motif, G-Box and TCT-motif) and one core promoter element for transcription start (TATA-box) (Fig. 4f). Then we wondered if these promoter elements were required for binding with COL3. To address this question, yeast one-hybrid assays were performed and the results showed that a 49-bp region of COL13 promoter containing G-Box and TCT-motif were the binding requirement (Fig, 4g).

COL13 is located in the nucleus
Transformation of Arabidopsis protoplasts with a construct expressing COL13-CFP indicated that COL13 is located in the nucleus (Fig. 5a), and a similar result was obtained when the root apical cells of stable COL13-GFP transgenic plants were examined (Fig. 5b).

COL13 interacts with COL3 , but not COP1
According to previous reports, both COL3 and COL13 are CONSTANS (CO)-like proteins, which are related to CO (Robson et al., 2001), and as shown for COL13 above, COL3 also positively regulates red-lightmediated inhibition of hypocotyl elongation inArabidopsis (Datta et al., 2006). We also demonstrated that COL13 shares the same subcellular localization as COL3 (Fig. 5a, b). Given that COL3 can interact with BBX32 and that COL13 also belongs to the BBX zinc finger TF family, we hypothesized that COL3 might interact with COL13. This idea was supported by a two-hybrid assay revealing that COL3 interacts with COL13 protein in yeast (Fig. 6a). Next, we examined the interaction in transgenic plants expressing both COL3 and COL13 and showed that COL13 was co-immunoprecipitated with COL3 from seedling tissues (Fig.  6b). The phenotypes of 35S:COL3-HA and 35S:COL13-GFP transgenic plants were the same as 35S:COL3 and 35S:COL13 transgenic plants, respectively, which produced shorter hypocotyl than WT grown in the presence of red light (Fig. S2). The interaction between COL13 and COL3 was also demonstrated in plant cells in a FRET assay (Fig. 6c-f). As shown in Fig. 6c, both cyan fluorescent protein (CFP)-fused COL3 and yellow fluorescent protein (YFP)-fused COL13 were observed in the nucleus after excitation with a 405 nm or 514 nm laser, respectively. After bleaching an area of interest with the 514 nm laser, YFP-COL13 fluorescence was reduced dramatically, whereas there was a clear increase in CFP-COL3 emission in the same area (Fig. 6d), indicating that FRET had occurred. The relative intensities of emissions from CFP-COL3 and YFP-COL13 in the area of interest, before and after bleaching, are shown in Fig. 6e, f.

COL13 promotes interaction between COL3 and COP1
Interestingly, although COL13 and COL3 have similar structures, containing two N-terminal tandemly repeated B-box domains and a CCT domain in the C-terminal, only COL3 can interact with COP1, and COL13 does not bind to COP1 (Fig. 6a). These results were also demonstrated by the FRET assay (Fig.  S3). To investigate whether COL13 influences the interaction between COP1 and COL3, we performed a yeast three-hybrid assay. In this yeast system, the COL3-COL13-pBridge construct allowed expression of COL3-BD /bait and COL13 in yeast, and COL13 was expressed only in the absence of methionine (Met). As shown in Fig. 7a, the growth of yeast carrying indicated constructs on selective medium (+Met or -Met) along with an α-galactosidase assay that showed that COP1 and COL3 had a stronger binding activity with the expression of COL13. Based on a previous report, COP1 interacted with COL3 and inhibited the production of COL3 (Datta et al., 2006). By combining our results above, we propose a possible COP1-dependent COL3-COL13 feedback pathway (Fig. 7b), which is involved in the regulation of hypocotyl elongation.

Discussion
Light regulates photomorphogenesis in plants. A large number of genes that are involved in such photomorphogenesis processes have been identified as light receptors (Datta et al., 2006;Kircher et al., 2002;Peter H. Quail, 2002), signal transduction factors (Gangappa et al., 2013;Osterlund et al., 2000) or degradation proteins (Crocco, Holm, Yanovsky, & Botto, 2010;Crocco et al., 2015;Delker et al., 2014). One of the immediate questions is how these genes act in a network to mediate various light-related phenotypes. It has been shown that multiple pathways are interlinked to form a gene network of photomorphogenesis (Lau & Deng, 2012;Lee, Park, Ha, Baldwin, & Park, 2017). Among these pathways, it is worth mentioning the ones formed by a subset of family genes termed the COL genes (Cheng & Wang, 2005). These family of genes plays multiple roles in plant development (Datta et al., 2006;Graeff et al., 2016;Muntha et al., 2018;Tripathi et al., 2017;Wang et al., 2014). As an effort toward COL networking, we investigated the relationship betweenCOL3 and COL13 and provided evidence that these two COLs and HY5 were connected together to form an HY5-COL3-COL13 regulatory chain that controls hypocotyl elongation in Arabidopsis (Fig. 7b). Hypocotyl elongation is a genetically well-controlled process that responds to light. In Arabidopsis , several key genes are required for hypocotyl growth. Among these, COP1 is a negative regulator (McNellis, von Arnim, & Deng, 1994), whereas HY5 and COL3 are considered to be positive (Datta et al., 2006;Hardtke et al., 2000). A previous study showed that COL3 plays a role in flowering and hypocotyl elongation (Datta et al., 2006), and COL3 is known to interact with B-BOX32 to regulate flowering (Tripathi et al., 2017). However, there has been no research on how COL3 regulates hypocotyl elongation. In this study, we demonstrated that COL13 , whose RNA accumulated to a high level in the hypocotyl (Fig. 1), was one more positive regulator in the regulation of hypocotyl elongation under red-light conditions. For example, overexpression of COL13 or knockdown of its transcript resulted in a shorter or longer hypocotyl, respectively, (Fig. 2).
As a positive regulator under far-red, red, blue, and UV-B light conditions (Ang et al., 1998;Delker et al., 2014;Hardtke et al., 2000), HY5 mediates about one-third of genes expression throughout theArabidopsis genome (Lee et al., 2007). This study revealed that HY5 bound to the promoter of COL3 and upregulate its transcription (Fig. 4b), indicating that COL3 also acted as a downstream target of HY5. Interestingly, COL3 bind to the COL13 promoter and positively regulate its expression (Fig.4b-g), suggesting that COL3 are positive regulators of COL13. Thus, these findings suggest that HY5, COL3, and COL13 constitute a hypocotyl regulatory pathway.
BBX family members are commonly involved in photomorphogenesis and that they can interact with other BBX proteins to regulate plant growth (Tripathi et al., 2017;Wang et al., 2014). For example, COL3 belong to the BBX family and COL3 interact with BBX32 to regulate flowering (Tripathi et al., 2017). In this study, we demonstrated that COL13 could interact with COL3 (Fig. 6). Furthermore, we found that the expression of COL13 promoted the interaction between COP1 and COL3 (Fig. 7a). To our knowledge, COP1 is responsible for the degradation of several positive TFs, such as COL3, in the dark (Datta et al., 2006;Dornan et al., 2004;Duek et al., 2004;Lau & Deng, 2012;Osterlund et al., 2000;Seo et al., 2004;Seo et al., 2003). Increasing the binding activity of COP1 and COL3 would lead to the degradation of COL3. As a result, there would be less COL3 to activate the expression of COL13 (Fig. 7b). Therefore, we proposed a possible COP1-dependent COL3-COL13 feedback pathway as a hypothesis to optimize the HY5-COL3-COL13 regulatory pathway (Fig. 7b). This feedback pathway could enrich the regulation network in hypocotyl elongation.      Analysis of the binding of HY5 to the COL3 promoter and COL3 to COL13 promoter truncations. (a) Diagram of constructs used in this study. For luciferase system, the AD-HY5 or AD-COL3 fusion proteins driven by the 35S promoter produces a potential effector protein, while the AD protein alone represents a negative control for basal activity ofCOL3 promoter or each COL13 promoter truncation. TheLUC gene driven by the series of COL3 promoter orCOL13 promoter truncations tests the ability of the AD-HY5 or AD-COL3 fusion protein to bind to each promoter truncation. For yeast-one hybrid system, the GAD-HY5 or GAD-COL3 fusion proteins driven by the GAL1 (P GAL1 ) promoter serves as effectors. The GAD protein served as negative control to see if there exits the self activity of COL3 or COL13 promoters. The LacZ gene driven by COL3 or COL13 promoter truncations served as the reporter to test the binding activity of the GAD-HY5 or GAD-COL3 fusion protein to individual promoter truncations. (b) For luciferase assay, the fusion protein AD-HY5 can up-regulate LUC expression from the COL3 promoter, but not from COL13 promoter; and the fusion protein AD-COL3, but not AD alone, can up-regulate LUC expression from some of the COL13 promoter truncations. For Yeast-one hybrid assay, the fusion protein GAD-HY5 can strongly bind to the promoter of COL3 , but not COL13 to direct LacZ expression in yeast cells that turns 5-bromo-4-chloro-3-indolyl-b-D-galactopyranoside to blue compound; and the fusion protein AD-COL3, but not AD alone, can bind to some of theCOL13 promoter truncations. (c) The 1059-bp promoter (-1675 bp to -616 bp ) of COL13 was devided into five overlapping regions: -1675 to -1391 bp (probe 1), -1421 to -1184 bp (probe 2), -1201 to -1040 bp (probe 3), -1060 to -868 bp (probe 4), and -898 to -616 bp (probe 5). (d) Electrophoretic mobility shift assay (EMSA) analysis showing the binding of COL3 on the -1421 to -1184 bp promoter of COL13 (probe 2) in vitro. The + andrepresent the positive and negative control, respectively. (e ) Electrophoretic mobility shift assay (EMSA) analysis showing the binding of COL3 on the -1421 to -1184 bp promoter of COL13 in vitro. The black arrow indicates binding of COL3 to the biotinlabeled COL13 promoter. The+ andrepresent the presence and absence of corresponding components, respectively. Competition experiments were carried out by adding 5-, 10-and 25-fold excessive competitor. (f) In probe 2, there were three light responsive elements (ATCT-motif, G-Box and TCT-motif) and one core promoter element for transcription start (TATA-box). (g) The fusion protein GAD-COL3, but not GAD alone, can strongly bind to a promoter truncation of COL13 to direct LacZ expression in yeast cells. The 238-bp promoter region (-1421 bp to -1184 bp ) ofCOL13 was devided into four fragments: -1421 to -1356 bp, -1355 to -1307 bp, -1306 to -1242 bp, and -1241 to -1184 bp. The colour rectangles represented indicated promoter elements in (f).  Co-immunoprecipitation (Co-IP) in Arabidopsis Immunoprecipitations (IPs) were performed on proteins extracted from 10 d-old Arabidopsis seedlings grown under long-day illumination (16L: 8D) at 22 degC. Leaf tissues were harvested 1 h after the light cycle commenced. IP was performed using an anti-HA antibody and COL13 was co-immunoprecipitated with an anti-GFP antibody. A 5% input was used. Western blots were performed on 10% (wt/vol) precast gels (Bio-Rad). (c) COL3-CFP and COL13-YFP colocalize to the nucleus in protoplasts in the light and dark. Bar=5um. (d-f) FRET between CFP-COL3 and YFP-COL13 analyzed by acceptor bleaching in the nucleus. Bar=5um. The top panels in (d) show a representative pre-bleach nucleus co-expressing YFP-COL13 and CFP-COL3 excited with either a 514 or a 405 nm laser in light and dark, resulting in emission from YFP (yellow) or CFP (blue), respectively. The bottom panels in (d) show the same nucleus post-bleaching after excitation with a 514 or a 405 nm laser. The relative intensities of both YFP and CFP were measured before and after bleaching, as indicated in (e) and (f), respectively. (a) Yeast three-hybrid analysis of COP1-COL3 interaction in the presence of COL13. Normalized Miller units were calculated as a ratio of α-galactosidase activity in yeast. Additionally, normalized Miller units are reported separately for yeast grown on media with or without 1 mM methionine (Met), corresponding to induction (-Met) or repression (+Met) of Met25 promoter-driven COL13 expression, respectively. Means and standard errors of the means for three biological repetitions are shown. Lower-case letters indicate significant differences in α-galactosidase. (b) A model representing the HY5-COL3-COL13 regulatory chain and COP1-dependent COL3-COL13 feedback pathway in the regulation of hypocotyl elongation.

Supporting Information
Table S1 List of primers and their uses.    FRET between CFP-COL3 and YFP-COP1 analyzed by acceptor bleaching in the nucleus. The top panels in b show a representative pre-bleach nucleus co-expressing YFP-COP1 and CFP-COL3 excited with either a 514 or a 405 nm laser in light and dark, resulting in emission from YFP (yellow) or CFP (blue), respectively. The bottom panels in (b) show the same nucleus after bleaching following excitation with a 514 or a 405 nm laser. The relative intensities of both YFP and CFP were measured once before and twice after bleaching, as indicated in (c) and (d).
(e) COL13-CFP and COP1-YFP co-localized to the nucleus in protoplasts in light and dark. (f) FRET between CFP-COL13 and YFP-COP1 analyzed by acceptor bleaching in the nucleus. The top panels in f show a representative pre-bleach nucleus co-expressing YFP-COP1 and CFP-COL13 excited with either a 514 or a 405 nm laser in light and dark, resulting in emission from YFP (yellow) or CFP (blue), respectively. The bottom panels in f show the same nucleus after bleaching following excitation with a 514-or a 405-nm laser. The relative intensities of both YFP and CFP were measured once before and twice after bleaching, as indicated in (g) and (h).