Arabidopsis thaliana plants used in this study are in Col-0 ecotype background. The following mutants and transgenic lines used in the study have been previously described: bri1-116 (Friedrichsen et al., 2000), bin2-1 (Li and Nam, 2002), det2-9 (Lv et al., 2018), co-9 (Balasubramanian et al., 2006), , 35S:Flag-BIN2 (He et al., 2019). Seeds were grown on half-strength Murashige and Skoog medium and stratified at 4°C for 2 d, then grown in long-day (16 h light/8 h dark) conditions at 22°C. Time-course analyses were performed on 14-d-old seedlings. Nicotiana benthamiana plants were grown under LD (16 h light/8 h dark) conditions at 24°C.

Analyses of flowering time were performed as previously described (Liu et al., 2017). Flowering time was scored as the number of days from germination to the first appearance of buds at the apex (days to bolting). More than 15 plants were counted and averaged for each measurement.

For Gateway cloning, all the gene sequences were cloned into the entry vector pQBV3 (Gateway) and subsequently introduced into certain destination vectors following the Gateway technology (Invitrogen) (He et al., 2019). For ligase-independent ligation, the ligation free cloning mastermix (abm) was used following the application handbook. For generation the mutant forms of BIN2 and CO, their coding sequences were first cloned into the pQBV3 vector, and then the point mutations were introduced by the specifically designed primers as previously described (Zheng et al., 2004), following the PCR amplification program: preheating at 94°C for 3 min, 16 cycles of 94°C for 1 min, 55°C for 1 min and 68°C for 4 min, finally, added DpnI and incubated at 37°C for 4 h. All details of DNA constructs and the primers used in this study are listed in Supplemental Table 1 .

For the generation of 35S:YFP-CO transgenic line, the full-length CO coding sequence was cloned into p35S-YFP vector to generate the 35S:YFP-CO construct. The 35S:YFP-CO construct was then transformed into the Agrobacterium strain GV3101 and introduced into Col-0 by the floral-dip method (Clough and Bent, 1998). For the generation of pCO : Flag-CO and pCO : Flag-CO^T280A transgenic lines, the fragment containing the CO promoter and the CO/CO^T280A coding region were cloned into the pCAMBIA1301 vector to generate the pCO : Flag-CO and pCO : Flag-CO^T280A constructs, respectively. Then the pCO : Flag-CO and pCO : Flag-CO^T280A constructs were transformed into the Agrobacterium strain GV3101 and introduced into the co-9 mutant background by the floral-dip method.

The 35S:YFP-CO/bin2-1 line was generated through genetic crossing between the 35S:YFP-CO and bin2-1 lines. The 35S:YFP-CO/35S:Flag-BIN2 double transgenic plants were generated by genetic crossing between the 35S:YFP-CO and 35S:Flag-BIN2 transgenic lines.

Total RNAs were extracted using a plant total RNA extraction kit (Zoman). Total RNA concentration was measured by Nanodrop 2000 (Thermo Fisher Scientific NanoDrop). About 2 μg of total RNA was used for reverse-transcribing to cDNA by abm reverse transcriptase kit. The cDNA was diluted to 100 μL with water in a 1:5 ratio, 2μL diluted cDNA was used for qRT-PCR. And qRT-PCR was performed with SYBR Premix Ex Taq (TaKaRa) on Light Cycler 96p (Roche). ACTIN7 was used as an internal control. Experiments were performed independently three times with similar results. All the primers used for qRT-PCR are listed in Supplemental Table 1 .

The LCI assays were performed in N.benthamiana leaves as described previously (He et al., 2019; Yang et al., 2021). Briefly, the full-length or truncated forms of the genes were cloned into the pCAMBIA1300-nLUC and pCAMBIA1300-cLUC vectors. Different constructs were transformed into the Agrobacterium strain GV3101 and then co-infiltrated into N. benthamiana leaves. Then plants were incubated for 12 h under dark and 36 h under light, and the LUC activities were analyzed using NightSHADE LB985 (Berthold). For each protein-protein interaction assay, at least 5 independent N. benthamiana leaves were infiltrated and analyzed, and 3 independent biological replications were performed for each assay (Sun et al., 2013).

For yeast two-hybrid (Y2H) analysis, the full-length or truncated forms of the genes were cloned into vectors pGADT7 or pGBKT7, respectively. Different constructs were transformed into AH109 yeast cells, and grown on agar plates with SD–L/W (synthetic dextrose medium lacking Leu and Trp). Further, the yeast cells were screened on SD–L/W/H medium (synthetic dextrose medium lacking Leu, Trp and His). Plates were kept at 30°C for 3 to 4 days. The empty vectors pGADT7 and pGBKT7 were used as negative controls.

For the co-immunoprecipitation (Co-IP) assays, the 35S:YFP-CO and 35S:YFP-CO/35S:Flag-BIN2 transgenic plants were used for detecting the interaction between CO and BIN2. Samples were harvested at ZT16 and then grounded to fine powder in liquid nitrogen. The total proteins were extracted by the lysis buffer (50 mM Tris-HCl at pH 7.5, 150 mM NaCl, 5 mM EDTA at pH 8.0, 0.1% Triton X-100, 0.2% NP-40) with freshly added PMSF (phenylmethylsulphonyl fluoride, 10 mM), Roche protease inhibitor cocktail (Roche, 11873580001) and MG132 (20 μM). After centrifugating at 4°C for 20 min, Extracts were mixed with Protein-G (Invitrogen, 00621751) for 1 h to reduce nonspecific immunoglobulin binding, and then incubated with anti-GFP magnetic beads (MBL, M047-8) overnight at 4°C. The precipitated samples were washed at least five times using the lysis buffer and then added 2X SDS protein loading buffer boiling for 5 min. The beads were detected by anti-GFP (1:2000; Roche, 11814460001) and anti-Flag (1:5000; abmart, M20026) antibodies.

The full-length coding sequences of CO and BIN2 were cloned into the pGEX4T-1 and pMAL2X-1 vectors to generate GST-CO and MBP-BIN2, respectively. These fusion proteins were expressed and purified from E. coli strain Transetta-DE3 (Transgen biotech, CD801). The pull-down assays were performed as described previously (Yang et al., 2021). For the assay, prewashed amylose resin in the column buffer (20 mM Tris [pH 8.0], mM NaCl, 1 mM EDTA) was used to pull down protein complexes. The mixture with proteins and amylose resin was incubated overnight at 4°C, then washed for 5 times with wash buffer, and added 2X SDS protein loading buffer boiling for 5 min. Eluted proteins were analyzed by immunoblotting using anti-MBP (1:3000; CWBIO, CW0288) and anti-GST (1:3000; CWBIO, CW0144) antibodies.

For protein level analysis, every 100 mg of Arabidopsis seedlings or N. benthamiana leaves added 200 μl protein extraction buffers. The extracted buffer used for extracting plants is described previously (He et al., 2019; Yang et al., 2021). The supernatant was separated on SDS-PAGE gels, and transferred to polyvinylidene fluoride membranes. For the protein detection, anti-Flag (1:5000; abmart, M20026) were used to detect Flag-CO proteins. Anti-Actin (1:5000; CWbiotech, CW0264) was used as a loading control.

For Phosphorylation assay, 1 μg BIN2-MBP/MBP and 1 μg CO-GST/mutated CO fusion proteins were added into kinase reaction buffer (25 mM Tris at pH 7.4, 12 mM MgCl₂, 1 mM DTT, and 1 mM ATP). Then the mixture was incubated at 37°C for 1 h and boiled with 5×SDS loading buffer. The boiled reaction products were separated by SDS-PAGE with or without 50 μM Phos-Tag (NARD, AAL-107). The phosphorylated CO was detected with anti-GST antibody.

The ChIP assays were carried out following a published protocol (Liu et al., 2017). Arabidopsis seedlings were grown on ½ MS for 10 d, 1.5g materials were collected and crosslinked in crosslinking buffer, then their chromatin was isolated. Anti-Flag ChIP grade antibody (ABclonal, AE005) and protein G plus agarose were used to immunoprecipitate the protein–DNA complex. Finally, the Flag-specific enrichment of the fragments from FT promoter was analyzed by qPCR using specific primer sets listed in Table S1 . The enrichment fold of a certain fragment was calculated by normalizing to the amount of no antibody-immunoprecipitate DNA samples.

The coding sequences of CO, CO^T280Aand Q were cloned into the pEarly-101 vector to generate YFP-CO, YFP-CO^T280A and mCherry- Q constructs, respectively. Then the constructs were transformed into the Agrobacterium strain GV3101 and infiltrated into N. benthamiana leaves. The fluorescence signal of yellow fluorescent protein (YFP) was observed at 48 h post-infiltration.

Sequence data in Arabidopsis from this study can be found in the Arabidopsis Genome Initiative database under the following accession numbers: BRI1 (AT4G39400), DET2 (AT2G38050), BIN2 (AT4G18710), CO (AT5G15840), FT (AT1G65480) and ACTIN7 (AT5G09810), SK12(AT3G05840), SnRK2.6(AT4G33950), MPK6(AT2G43790).

CONSTANS (CO) is a central regulator of floral initiation in response to photoperiod. In order to explore the regulatory effects of phosphorylation and specific protein kinases for CO, we tested a series of kinases for CO interaction through the luciferase (LUC) complementation imaging (LCI) assay. In the LCI assay, CO was fused with the amino-terminal part of LUC (nLUC) and kinases (BIN2, SK12, SnRK2.6 and MPK6) were ligated to the carboxyl-terminal part of LUC (cLUC) to generate nLUC-CO and cLUC-BIN2 (SK12, SnRK2.6, MPK6), respectively. The results showed that strong LUC activity could be observed when nLUC-CO/cLUC-BIN2 and nLUC-CO/cLUC-SK12 were co-expressed in Nicotiana benthamiana leaves, but no LUC activity was detected in nLUC-CO/cLUC-SnRK2.6, nLUC-CO/cLUC-MPK6 ( Figure 1A ). Since BIN2 and SK12 (a homolog of BIN2) interact with CO, we further tested the interaction between BIN2 and CO through other approaches. In the yeast two-hybrid (Y2H) assay, BIN2 was fused with pGBKT7 and CO was ligated to pGADT7 to generate BD-BIN2 and AD-CO, respectively. The AH109 strain yeast cells transformed with BD-BIN2 and AD-CO could grow well in the selection medium, but the negative controls did not ( Figure 1B ), indicating that BIN2 physically interacts with CO in yeast cells. Consistently, the MBP-BIN2 fusion proteins could retain GST-CO proteins, whereas MBP alone could not, indicating that BIN2 directly interacts with CO in the in vitro pull-down assay ( Figure 1C ). Moreover, we generated the 35S:Flag-BIN2/35S:YFP-CO double transgenic plant through genetic crossing between the 35S:Flag-BIN2 and 35S:YFP-CO lines for the Co-IP assay. The results showed that the Flag-BIN2 proteins could be precipitated by YFP-CO proteins in the double transgenic 35S:Flag-BIN2/35S:YFP-CO seedlings in the Co-IP assay, further suggesting that BIN2 interacts with CO in vivo ( Figure 1D ).

To map the interaction region of CO with BIN2, the full-length of CO protein was truncated into N-terminal part harboring the B-Box domain (CO-NT) and C-terminal part containing the CCT domain (CO-CT) ( Figure 1E ). Those two parts of CO were fused with nLUC to generate nLUC-CO-NT and nLUC-CO-CT for LCI assays, respectively. As shown in Figure 1F , obvious LUC activities were observed in the nLUC-CO-NT/cLUC-BIN2 co-expression samples, but only background level of LUC activities was observed in the nLUC-CO-CT/cLUC-BIN2 samples. These results suggest that the N-terminal part of CO containing the B-Box domain is required for the interaction with BIN2 ( Figure 1F ). Moreover, we also explored whether the kinase activity of BIN2 is required for its interaction with CO. BIN2^K69R mutant form, in which the Lys69 was replaced by Arg and the GSK kinase activity was abolished, was fused with cLUC to generate cLUC- BIN2^K69R for LCI assays. As shown in Figure 1G , less LUC activities were detected in the cLUC-BIN2^K69R/nLUC-CO co-expression samples than those in the cLUC-BIN2/nLUC-CO co-expression samples ( Figure 1G ), suggesting that the kinase activity of BIN2 is essential for its interaction with CO.

Since BIN2 physically interacts with CO, we were curious whether these two genes have similar expression patterns during photoperiod. To address this question, we analyzed the expression patterns of BIN2 and CO in Col-0 during a 24-h photoperiod. Interestingly, we found that the transcripts of BIN2 exhibited similar diurnal expression patterns with CO ( Supplementary Figure 1 ), supporting the notion that BIN2 is supposed to post-transcriptionally modify CO protein in regulating flowering time.

Although it has been known that BRs play an important role in promoting flowering in Arabidopsis, the underlying mechanism remains largely unclear. To explore the biological role of BR pathway in regulating flowering time, we systematically examined the flowering time phenotypes of BR biosynthetic and signaling mutants including bin2-1, det2-9 and bri1-116 under LD (16 h light/8 h dark) conditions. The results showed that bin2-1, det2-9 and bri1-116 displayed obvious late-flowering phenotypes compared with the wild type (WT) Columbia-0 (Col-0) under LD conditions ( Figures 2A, B and Supplementary Figures S2A, B ). To further dissect the molecular effect contributing to the late flowering of BR biosynthetic or signaling mutants, we analyzed daily expression patterns of FT, a key integrator of flowering time (Putterill et al., 1995; Kobayashi et al., 1999), in these mutants using quantitative RT-PCR (qRT-PCR). The qRT-PCR assays revealed that the expression levels of FT in Col-0 plants were peaked at Zeitgeber time 16 (ZT16), consistent with previous report (Turck et al., 2008), but the peak of FT expression levels were largely abolished in the bin2-1, det2-9 and bri1-116 mutants ( Figure 2C and Supplementary Figure 2C ). These results imply that the late flowering phenotypes of these mutants were likely due to the down-regulation of FT transcription.

Considering that CO directly activates FT transcription (Onouchi et al., 2000; Samach et al., 2000), we were curious about the relationship between the BIN2 and CO in regulating flowering time. To this end, we studied the genetic interaction between BIN2 and CO. We generated the CO-overexpressing transgenic line 35S:YFP-CO and introduced the 35S:YFP-CO transgene into the bin2-1 mutant background to generate the 35S:YFP-CO/bin2-1 lines. As shown in Figures 2D, E , the 35S:YFP-CO transgenic line flowered much earlier than the wild type plant Col-0. Notably, the 35S:YFP-CO/bin2-1 line displayed obvious early-flowering phenotype compared with the bin2-1 mutant, indicating that over expression of CO can rescue the late-flowering phenotype of bin2-1 ( Figures 2D, E ). Consistent with the flowering phenotypes, the transcript levels of FT at ZT16 in 35S:YFP-CO/bin2-1 were higher than those in the bin2-1 mutant ( Supplementary Figure 3 ). This result demonstrates that BIN2 acts genetically upstream of CO in regulating flowering time.

Since the protein kinase BIN2 interacts with CO, it is important to test whether CO is a substrate of BIN2. To this end, we used a phos-tag approach, where phosphorylated proteins in the gel containing phos-tag reagent are visualized as bands with slower migration compared with the corresponding dephosphorylated proteins. As shown in Figure 3A , the slower migrated band of CO could be observed when incubated with MBP-BIN2 and ATP; whereas this phenomenon could not be observed when incubated MBP-BIN2 without ATP ( Figure 3A ). These results demonstrate that BIN2 can phosphorylate CO. Consistently, the slower migrated band of CO was not observed when incubated with MBP-BIN2^K69R (the kinase dead version of BIN2) and ATP ( Figure 3A ), suggesting that the kinase activity of BIN2 is required for the phosphorylation of CO.

In order to search for the putative phosphorylation sites of CO by BIN2, we first analyzed the putative BIN2 phosphorylation motifs in CO (Ser/Thr-X-X-X-Ser/Thr; Wang et al., 2002; Ryu et al., 2007; Ryu et al., 2010). As a result, we identified 13 putative phosphorylation motifs in CO protein ( Supplementary Figure 4 ). Next, we mutated Ser/Thr to Ala in the 1^th-3^th phosphorylation motifs of CO (CO^mN), 4^th-9^th phosphorylation motifs of CO (CO^mM) and 10^th-13^th phosphorylation motifs of CO (CO^mC) to mimic non-phosphorylated forms of CO, respectively. As shown in Figure 3B , CO^mN and CO^mM was still phosphorylated by BIN2; in contrast, CO^mC was not phosphorylated by BIN2, illustrating that the phosphorylation sites are located in the 10^th-13^th phosphorylation motifs of CO ( Figure 3B ). Further, we mutated Ser/Thr to Ala in the 10^th phosphorylation motifs of CO (CO^mC-1), 11^th phosphorylation motifs of CO (CO^mC-2), 12^th phosphorylation motifs of CO (CO^mC-3) and 13^th phosphorylation motifs of CO (CO^mC-4), respectively. The results showed that the CO^mC-1, CO^mC-2 and CO^mC-4 mutant forms were still phosphorylated by BIN2; whereas CO^mC-3 could not be phosphorylated, indicating that the phosphorylation sites are located at the 12^th phosphorylation motif of CO ( Figure 3C ). Finally, we mutated the Thr280 and Thr284 residues at the 12^th phosphorylation motif to Ala to generate the CO^T280A and CO^T284A mutant forms, respectively. As shown in Figure 3D , CO^T284A, but not CO^T280A, could be phosphorylated by BIN2, suggesting that the Thr280 residue is the specific BIN2 phosphorylation site of CO protein ( Figure 3D ). In summary, our efforts reveal that BIN2 can phosphorylate the Thr280 residue of CO protein.

To explore the biological relevance of CO phosphorylation by BIN2, we generated the pCO : Flag-CO/co-9 and pCO : Flag-CO^T280A /co-9 transgenic plants in the co-9 mutant background. As shown in Figure 4A , the pCO : Flag-CO transgene could completely rescue the late-flowering phenotype of co-9. Notably, the pCO : Flag-CO^T280A /co-9 lines flowered obviously earlier compared to pCO : Flag-CO/co-9 and Col-0 ( Figures 4A–C ). The transcript levels of CO in these transgenic lines were comparable ( Figure 4D ). Consistent with the flowering phenotypes, the transcript levels of FT at ZT16 in the Flag-COT280A/co-9 to Flag-COT280A/co-9 plants were higher than those in the pCO : Flag-CO/co-9 and Col-0 plants ( Figure 4E ). Then we checked whether the phosphorylation of CO affect the enrichments of CO on FT. To this end, we performed chromatin immunoprecipitation-quantitative PCR (ChIP-qPCR) assays using the 10-day-old pCO : Flag-CO/co-9 and pCO : Flag-COT280A/co-9 seedlings. As expected, in the pCO : Flag-CO/co-9 transgenic plants, CO can directly bind to the promoters of FT. Meanwhile, the unphosphorylation of CO increased the enrichments of CO on FT, suggesting that the phosphorylation of CO inhibit the DNA-binding activity of CO to COREs in the FT promoter ( Figures 4F, G ). However, the CO protein levels were largely similar between the pCO : Flag-CO^T280A /co-9 and pCO : Flag-CO/co-9 transgenic plants ( Figure 4H ), illustrating that phosphorylation of CO by BIN2 does not influence CO protein stability. Next, we investigated whether the phosphorylation by BIN2 affects the interaction intensity between CO and TOE1/NF-YC4. The Y2H assays showed that the mutation of the Thr280 residue to Ala did not significantly affect the interaction intensity between CO and TOE1/NF-YC4 ( Supplementary Figure 5 ). In addition, subcellular localization assay showed that the CO-YFP and CO^T280A-YFP proteins were both localized in the nuclei, suggesting that the phosphorylation by BIN2 does not affect the subcellular localization of CO protein ( Supplementary Figure 6 ). Taken together, these observations demonstrate that the phosphorylation of CO by BIN2 inhibits the function of CO in promoting flowering through affecting the DNA-binding activity of CO.

The formation of CO dimer/oligomer is possibly involved in influencing CO activity (Graeff et al., 2016; Lv et al., 2021). Here we showed that CO could interact with itself in plant cells ( Figure 5A ), which may facilitate the formation of CO dimer/oligomer. Further, we conducted LCI assays to determine which region of CO is responsible for the interaction of CO-CO. The results showed that the N-terminal part of CO harboring the B-Box domain mediates the interaction of CO-CO ( Figure 5B ). As described above, the N-terminal part of CO also mediates the interaction of BIN2 and CO ( Figure 1F ). The findings that the B-Box domain of CO mediates the interaction of CO-CO prompted us to determine whether BIN2 competitively interferes with the CO-CO interaction. To test this idea, we co-expressed BIN2-Flag with nLUC-CO-NT and cLUC-CO-NT proteins in N. benthamiana leaves for LCI assays. The YFP protein was expressed instead of BIN2-Flag as a control. The results showed that the samples co-expressing nLUC-CO-NT/cLUC-CO-NT together with BIN2-Flag displayed weaker luminescence signals compared with those of samples co-expressing nLUC-CO-NT/cLUC-CO-NT and YFP ( Figures 5C–G ). Meanwhile, LCI assay showed that phosphorylation or unphosphorylation of CO did not significantly affect the interaction intensity between CO and CO ( Supplementary Figure 7 ). These results demonstrate that BIN2 physically inhibits the interaction of CO-CO in plant cells.