Calcium-dependent protein kinase CPK32 mediates calcium signaling in regulating Arabidopsis flowering time

Abstract Appropriate flowering time is critical for the reproductive success of plant species. Emerging evidence indicates that calcium may play an important role in the regulation of flowering time. However, the underlying molecular mechanisms remain unclear. In this study, we demonstrate that calcium-dependent protein kinase 32 (CPK32) regulates flowering time by affecting the alternative polyadenylation of FLOWERING CONTROL LOCUS A (FCA) and altering the transcription of FLOWERING LOCUS C (FLC), a central repressor of flowering time. The knockdown of CPK32 results in an obvious late flowering phenotype and dramatically enhanced FLC transcription. CPK32 interacts with FCA, and phosphorylates the serine592 of FCA in a Ca2+-dependent manner. Moreover, the ratio of abundance of the FCA transcripts (FCA-D and FCA-P) changes significantly in the cpk32 mutant, which subsequently affects FLC expression and consequently regulates floral transition. The present evidence demonstrates that CPK32 modulates flowering time by regulating FCA alternative polyadenylation and consequent FLC expression.


INTRODUCTION
Flowering time is critical for flowering plant reproduction, species maintenance, adaptation and domestication, and is also a key element in crop breeding and variety selection because of its importance for crop growth vigor and yield [1,2]. Molecular analysis and genetic studies of flowering time regulation have made great strides in deciphering the regulatory mechanism of floral transition.
FLOWERING LOCUS C (FLC) is the core regulatory gene mediating flowering time in the autonomous and vernalization pathways [3,4]. Alternative polyadenylation (APA) generates mRNAs with distinct 3 ends and has emerged as a pervasive regulatory mechanism in gene expression [5,6]. Several floral regulators modulate FLC expression through APA, such as the conserved RNA-binding proteins FLOWERING CONTROL LOCUS A (FCA) and FLOWERING LOCUS PA (FPA) [7][8][9][10][11]. Interaction between FCA and FLOWERING LOCUS Y (FY) is required for the efficient selection of the 3' end of FCA transcripts and FLC repression [9,12]. Integrative genome-wide analysis revealed that the heterogeneous nuclear ribonucleoprotein (hnRNP) A1-like protein 1 (HLP1) regulates flowering by modulating the APA of FCA [13]. Despite the wealth of knowledge about the FCA-FLC module in the autonomous pathway, little is known about the molecular mechanism that regulates the APA of FCA during flowering.
Plants have evolved sophisticated signaling cascades that perceive, integrate and respond to endogenous and external stimuli, thereby ensuring their survival and the proper timing of flowering for species maintenance. Calcium (Ca 2+ ) is a ubiquitous second messenger in living organisms [14]. Previous reports have shown that the environmental factors that influence flowering time, such as light quality [15], day length, light intensity [16] and low temperatures [17,18], also affect the cytosolic Ca 2+ concentration in Arabidopsis roots, leaves or seedlings.
Ca 2+ signals are perceived and decoded by a series of Ca 2+ -binding proteins or Ca 2+ transducers. Among calcium-binding proteins, Ca 2+ -dependent protein kinases (CPKs/CDPKs) are plant-specific protein kinases that participate in a variety of important physiological activities, including ion channel regulation [19,20], stomatal movements [21], cell expansion [22] and responses to pathogens [23]. In the present study, we present evidence for the involvement of CPK32 in flowering time regulation. The CPK32 knockdown mutant cpk32 exhibited an obvious late-flowering-time phenotype, accompanied by significant enhancement of FLC expression. It was further revealed that CPK32 interacts with FCA, and phosphorylates the serine 592 of FCA in a Ca 2+ -dependent manner. As a result, the alternative splicing of FCA pre-mRNA is regulated, and in turn FLC transcription is regulated. The presented evidence demonstrates that CPK32 regulates FCA pre-mRNA levels to fine-tune flowering time through the autonomous pathway.

CPK32 is a positive regulator of plant flowering
The CPK family in the Arabidopsis genome includes 34 members, and they cluster into four subgroups based on protein sequence homology [24]. To investigate the potential function of CPKs in the floral transition in Arabidopsis, the flowering time phenotype of 32 CPK mutant lines (Supplementary Fig. S1 and Supplementary Table 1) were screened in long-day (LD) photoperiod conditions. At the time of bolting, the majority of cpk mutants showed a similar flowering time phenotype compared with the wild-type (WT) plants, except cpk23, cpk32 and cpk33 mutants, which displayed a late flowering phenotype, and the cpk28 mutant (Salk 112540, an overexpression line), which displayed an early flowering phenotype (Fig. 1A). In Arabidopsis, flowering time is closely associated with leaf number, which is widely used to quantify the time of the floral transition [12,25]. At the time of flowering, these three late flowering mutants developed more leaves than the WT plants (Fig. 1B). CPK32 belongs to subgroup III, CPK23 and CPK33 belong to subgroup II, and CPK28 belongs to subgroup IV of the CPK super-family [24]. Their divergent homology suggests that they may regulate the flowering of Arabidopsis via different mechanisms. RT-qPCR experiments revealed that transcription of FLC was dramatically enhanced in the cpk32 mutant, while FLC transcription was not changed in cpk23, cpk28 and cpk33 mutants (Fig. 1C). Considering FLC is a central repressor of flowering [3,4], further investiga-tion was focused on characterizing CPK32 function in flowering time regulation.
The cpk32 mutant line used in our experiments is a knock-down allele ( Supplementary Fig. S2). To validate that the late-flowering-time phenotype was caused by decreased CPK32 expression, three independent complementation lines (COM9, COM15 and COM17) were generated (Supplementary Fig. S2C and D). All complementation lines displayed a flowering time similar to the WT in LD photoperiod conditions ( Fig. 2A and B). These results confirmed that the decreased expression of CPK32 in the cpk32 mutant is responsible for its late-flowering-time phenotype and suggests that CPK32 may positively regulate the floral transition in Arabidopsis.

Expression profiling of CPK32
To gain insight into the function of CPK32 during floral initiation, CPK32pro : GUS transgenic lines were used and the GUS staining showed that CPK32 was expressed in embryos and emerged radicles, roots, floral organs, anthers and stigma (Supplementary Fig. S3). The results also showed that CPK32 is expressed in the shoot apex ( Supplementary  Fig. S3D), indicating that CPK32 may play a role in floral initiation.

CPK32 regulates flowering time through the autonomous pathway
Four major pathways regulating flowering time have been defined by classic genetics approaches in Arabidopsis [26,27]. To identify the pathway that CPK32 is involved in, the flowering-time phenotypes of the cpk32 mutant were tested under different photoperiods, gibberellic acid (GA) treatment and vernalization treatments. Compared with LD conditions, short-day (SD) conditions extended the vegetative growth period of all lines indicated with their increased leaf number, and the cpk32 mutant still flowered significantly later than the WT plants and the complementation lines (Fig. 2C). These results indicate that cpk32 mutants respond to changes in photoperiod in a similar way to WT plants, except for their delayed flowering. The experiments conducted with GA and vernalization treatments in LD photoperiod conditions showed that all plants flowered earlier than their corresponding controls, indicating that the cpk32 mutant is similar to WT in its response to GA and vernalization ( Fig. 2D and E). Taken together, these data demonstrate that CPK32 does not promote floral transition through the photoperiod, GA and vernalization pathways but most likely through the autonomous pathway.  To confirm the involvement of CPK32 in the autonomous pathway, the FLC transcription in various lines was analyzed. As shown in Figs 1C and 2F, FLC transcriptional expression in the cpk32 mutants was much higher than in WT plants, while the complementation lines showed similar transcription to the WT (Fig. 2F). Consistently with this, FLOWERING LOCUS T (FT) and SUPPRESSOR OF OVEREXPRESSION OF CO 1 (SOC1), which are downstream targets of FLC in flowering-time control [28,29], were down-regulated in the cpk32 mutant compared to the WT plants and the complementation lines ( Fig. 2G and H).
To further confirm CPK32 involvement in floral transition through its effect on FLC expression, the cpk32 flc double mutant was generated (Supplementary Fig. S2E) and it showed the similar early flowering phenotype as the flc single mutant in LD photoperiod conditions ( Fig. 3A and B). These results support our hypothesis that CPK32 regulates flowering by repressing FLC expression.  [27,30]. Among these genes, seven, including FLK, FLD, FVE, FPA, FY, REF6 and LUMINIDEPENDENS, showed no obvious changes in their expression in the cpk32 mutant compared to the WT ( Supplementary Fig. S4A). FLC transcription in the double mutants cpk32 prmt5, cpk32 prmt10, cpk32 ref6, cpk32 fve and cpk32 fld was higher than that in corresponding single mutants, respectively, except the cpk32 fca double mutant ( Fig. 3C and Supplementary Fig. S4B and C). FLC transcription in the cpk32 fca double mutant was nearly the same as that in the fca single mutant (Fig.  3C), suggesting that CPK32 may regulate FLC expression through FCA.

CPK32 regulates FLC expression through FCA
Previous reports have shown that FCA functions as a posttranscriptional regulator, and that the fca mutant has a late-flowering phenotype [7,25]. The act in the same pathway, and the upregulated expression of FLC in the cpk32 mutant is most likely mediated by FCA.

CPK32 interacts with FCA
As previous reports showed that FCA interacts with FY to downregulate FLC expression [12] and CPK32 interacts with Abscisic acid responsive element-Binding Factors (ABFs) [31], FCA-FY and CPK32-ABF1 were used as the positive controls in Yeast Two-Hybrid (Y2H) experiments to test the proposed interaction between CPK32 and FCA. Indeed, CPK32 strongly interacted with FCA (Fig. 4A), while FY, FLK and FVE had no interaction with CPK32 (Supplementary Fig. S5A). All the other seven subgroup III CPK members had no interaction with FCA ( Supplementary Fig. S5B). The protein pull-down assays further confirmed the interaction between CPK32 and FCA. The Nterminal of FCA containing two RNA recognition  motif (RRM) domains and the C-terminal of FCA containing the tryptophan-tryptophan (WW) domain were purified (Fig. 4B). CPK32 interacted with the C-terminal region of FCA (FCA-C-His), but not with its N-terminal region (Fig. 4C). It should be noted that CPK32 did not interact with CKS-FCA-C-His, suggesting that CKS-tag, a 29 kD fragment, may weaken the interaction between CPK32 and the FCA-C terminal. In addition, bimolecular fluorescence complementation (BiFC) assays showed specific CPK32-FCA interaction signals in the nucleus (Fig. 4D). These results demonstrate that CPK32 directly interacts with FCA both in vitro and in vivo.

CPK32 phosphorylates FCA in a Ca 2+ -dependent manner
It was hypothesized that CPK32, as a protein kinase, may regulate FCA function by phosphorylation. Truncated FCA segments (Fig. 5A) were the substrates. Autoradiography showed that FCA-B, FCA-E, FCA-F, FCA-G and FCA-H fragments were phosphorylated by CPK32, and the strongest signal detected was in the FCA-G fragment (Fig. 5B), which mainly contains the WW domain required for FCA autoregulation [9]. The potential phosphorylation sites, including serine, threonine and tyrosine in the WW domain, were substituted with alanine to mimic an un-phosphorylation status for the phosphorylation assay. Autoradiography results showed that the phosphorylation signal of FCA S592A -E significantly decreased (Fig. 5C), suggesting that the serine 592 is the target site of phosphorylation by CPK32.
Furthermore, FCA phosphorylation by CPK32 is Ca 2+ -dependent. Chelation of Ca 2+ by addition of EGTA in the reaction medium resulted in weaker signals for all phosphorylated bands (Fig. 5D). The C terminus of CPK protein contains a Ca 2+ -binding domain constituted by EF-hand motifs [24]. To determine whether CPK32-mediated phosphorylation of FCA is Ca 2+ binding dependent, a series of Natl Sci Rev, 2022, Vol. 9, nwab180 SD/-T-L  single, double and multiple EF-hand motif mutations of CPK32 were generated as described previously [32], rendering these motifs unable to bind Ca 2+ . Using an FCA-E fragment as the substrate, phosphorylation signals were abolished for all CPK32 proteins with mutated EF-hands (Fig. 5E).

SD/-T-L-H-A C P K 3 2 -A D / A B F 1 -B D C P K 3 2 -A D / F C A -B D F Y -B D / F C A -A D C P K 3 2 -A D / F Y -B D A B F 1 -B D / A D F C A -B D / A D F Y -B D / A D F C A -A D / B D C P K 3 2 -
The results further demonstrated that phosphorylation of FCA by CPK32 is Ca 2+ dependent.
To further confirm correlation of CPK32mediated Ca 2+ signaling and flowering-time regulation, the transgenic lines (cpk32 CPK32-EF-M) with four EF-hands mutations [32] were generated (Supplementary Fig. S6) and used to test the flowering-time phenotype. As shown in Fig. 5F and G, the cpk32 CPK32-EF-M plants displayed the same delayed flowering-time phenotype as the cpk32 mutant. Moreover, FLC expression in these EF-hand-mutated transgenic lines was similarly enhanced as in the cpk32 mutant (Fig. 5H).

CPK32 regulates the alternative polyadenylation of FCA by affecting FCA-FY interaction
Previous studies have shown that FCA regulates its own expression by promoting premature cleavage and polyadenylation of its own pre-mRNA [7]. The FCA pre-mRNA can be alternatively spliced to form four different transcripts and only the FCA-γ can complement the late-flowering phenotype of the fca mutant [8]. CPK32 phosphorylated the WW domain, which is required for FCA autoregulation in a calcium-dependent manner (Fig. 5B, D and E). To test whether CPK32 plays a role in processing FCA mRNA, the distal (FCA-D, mainly FCA-γ ) and proximal (FCA-P, mainly FCA-β) polyadenylation of FCA pre-mRNA were analyzed. Compared to the WT, the cpk32 mutant had decreased levels of FCA-D transcripts and increased levels of FCA-P transcripts,  while COM9 had approximately the same levels as WT (Fig. 6A and B). Obviously, disruption of CPK32 expression in the cpk32 mutant resulted in changes in the ratio of the FCA-D to FCA-P transcripts. Previous reports have shown that the functional FCA-γ (FCA-D) transcripts result in accelerated flowering [7,8], which together with the present results suggests that the late flowering-time phenotype of the cpk32 mutant may result from the decreased functional distal polyadenylation FCA transcript. The transcripts of FCA-P and FCA-D in EFhand-mutated complementation lines did not re-sume at WT levels ( Fig. 6A and B), demonstrating that CPK32 activity plays an important role in regulating FCA pre-mRNA processing. The WW domain is required for FCA autoregulation of RNA 3 polyadenylation and mediates the interaction with FY, and this interaction significantly affects flowering time [12]. As shown in the BiFC assays, the FCA S592D , a persistent phosphorylated FCA form, showed an interaction with FY, while the FCA S592A , an unphosphorylated form of FCA, did not show the interaction between FCA and FY (Fig. 6C). These results suggest that CPK32 may regulate flowering time by maintaining the FCA-FY interaction via phosphorylation modification.

DISCUSSION
In this study, we showed that CPK32 modulates flowering time by regulating FCA pre-mRNA processing in Arabidopsis. CPK32 interacts with FCA, phosphorylates the WW domain of FCA and regulates the APA of FCA transcripts (Figs 4-6). FCA plays an important role in alternative polyadenylation of antisense RNAs and 3 end formation of FLC and its transcript [11,33]. As a result, FLC expression is suppressed and the inhibition on flowering transition is released (Fig. 6D).
Among CPK family members, the present data showed that CPK32 is specifically involved in flowering-time control by repressing FLC expression (Fig. 1C). We also observed a late flowering-time phenotype for the mutant cpk33 ( Fig. 1A and B), which is consistent with previously reported results [34,35]. In addition to CPK32, CPK23, CPK28 and CPK33 are also involved in the regulation of flowering time ( Fig. 1A and B), although these CPKs do not affect FLC expression (Fig. 1C). Further analysis of their functions would increase the understanding of the Ca 2+ signaling mechanism's role in regulation of flowering time.
CPKs have been identified as regulators involved in various physiological responses in plants [36]. Previous reports have shown that CPK32 can interact with ABFs and regulate ABA-responsive gene expression [32]. A genome-wide survey of CPK function in pollen tubes demonstrated that CPK32 overexpression caused the formation of a bulge at the tip of the pollen tube [37]. CPK32 interacted with cyclic nucleotide-gated channel 18 (CNGC18) and promoted Ca 2+ entry during the elevation phase of the Ca 2+ oscillations that occurred during polar growth of pollen tubes [38]. In addition, CPK32 was one of the main regulators of nitrate-CPK-NIN-LIKE PROTEIN (NLP) signaling networks [39]. CPK32 also regulated the activity of ammonium transporter 1;1 (AMT1;1) [40]. It seems that CPK32 may be a multi-functional regulator and play an important role in integrating multiple signaling processes and optimizing plant growth and development.
In conclusion, this study has revealed that the Ca 2+ -dependent protein kinase CPK32 functions as a positive regulator of flowering time acting through the autonomous pathway, and this finding provides new insight into the complex regulatory network of flowering time and calcium signaling in plants.

MATERIALS AND METHODS
Detailed descriptions of materials and methods are available as supporting information at NSR online.

SUPPLEMENTARY DATA
Supplementary data are available at NSR online.