4 DISCUSSION
We previously showed that CRCT is a positive regulator of starch synthesis in rice, but the underlying molecular function of CRCT remained elusive. In this study, it was found that CRCT forms a complex with 14-3-3 proteins (Figures 7, 8 and 9; Table 1). Although 14-3-3 proteins are well characterized and play a role in the regulation of metabolism and signal transduction for various developmental processes and biological responses, the physiological significance of this CRCT/14-3-3 complex is unknown.
BiFC analysis suggests that the CRCT/14-3-3 complex should mainly occur in the nucleus (Figure 9). 14-3-3 proteins in plants are observed in various subcellular compartments including the nucleus (Bihn et al., 1997). It is well known that florigen such as rice Heading date 3a forms a complex with transcription factor FD1 and 14-3-3 protein to control flowering related genes in the nucleus (Taoka et al., 2011). In this case, FD1 binds to DNA in the form of a complex, and 14-3-3 proteins plays the role of adaptor. Among CCT proteins, CONSTANS was shown to interact with 14-3-3 isoforms by yeast two-hybrid and immunoprecipitation assays (Mayfield, Folta, Paul, & Ferl, 2007). Actually, CONSTANS contains a 14-3-3 binding motif upstream of its CCT domain, just like as CRCT (Table S3). Knockout mutants of 14-3-3 isoforms showed a late flowering phenotype similar to the knockout mutant of CONSTANS (Mayfield, Folta, Paul, & Ferl, 2007), suggesting that the interaction with 14-3-3 proteins are necessary for the proper function of CONSTANS. Thus, it is possible that interactions with 14-3-3 proteins are prerequisite for CRCT to control the expression of starch synthesis-related genes in the nucleus. However, the molecular weight of the complex containing CRCT in vivo was estimated to be 270 kDa by gel filtration chromatography (Figure 6). The monomers of GF14A and GF14B are estimated to be 29.0 kDa and 29.9 kDa, respectively, based on their amino acid sequences. In general, 14-3-3 proteins forms dimersin vivo . According to BiFC analysis, CRCT is possibly present as monomer (Figure 9). As the CRCT monomer is 34.2 kDa, the complex of a CRCT monomer and 14-3-3 dimer should be 92.2-94.0 kDa. This molecular weight is lower than the estimation by gel filtration chromatography. The estimation of molecular mass by gel filtration is affected by the shape of protein complex. Thus, there is a possibility that actual molecular mass of CRCT complex can be smaller than our estimation. As another possibility, there are other interacting factors necessary for CRCT to function that are not yet identified.
The up-regulation of CRCT by sugar was reduced by nitrogen treatment (Morita et al., 2015). Considering this C/N response, some functional relation to CRCT can be expected with ATL31, a ubiquitin ligase. InArabidopsis , the stability of the 14-3-3 protein regulated by ALT31 determined the resistance to C/N stress (Sato et al., 2011). According to their proposed mechanism, the interaction of 14-3-3 protein may inactivate CRCT. The overexpression of CRCT may lead to increases in the level of unbound free CRCT, which enable the up-regulation of the starch synthesis-related genes. In accordance with this hypothesis, it was reported that the starch content was significantly decreased in the leaves of Arabidopsis by the overexpression of 14-3-3 proteins (Diaz et al., 2011). Furthermore, antisense knockdown of 14-3-3 proteins greatly increased the starch accumulation (Sehnke et al. 2001). However, the expression level of 14-3-3 proteins can also affect the activity of key enzymes involved in carbon and nitrogen metabolism (Sehnke et al. 2001, Diaz et al., 2011). Therefore, the contribution of CRCT to starch content in 14-3-3 overexpression line and knockdown line requires more clarification.
The 14-3-3 family comprises eight members in rice (Chen, Li, Sun, & He, 2006; Yao, Du, Jiang, & Liu, 2007). These can be divided into ε-like groups and non-ε groups. Non-ε groups including GF14A and GF14B are highly homologous to each other but show different patterns in organ specific expression and stress response. The expression of GF14Bwas reported to be high in roots and low in reproductive organs, whereas the expression of GF14A was low in all organs (Yao, Du, Jiang, & Liu, 2007). According to a rice gene expression database, GF14Ais actually expressed in most organs at a relatively constant level (Figure S10). Most non-ε 14-3-3 proteins other than GF14C were induced by pathogen infection, whereas GF14C was markedly induced by abiotic stress such as salinity or drought (Chen, Li, Sun, & He, 2006). These observations suggest that 14-3-3 proteins share functions depending on organs and conditions. Among rice non-ε group 14-3-3 proteins, GF14A and GF14B are divided into different clades (Chen, Li, Sun, & He, 2006), both of which can interact with CRCT (Figure 8). In addition, MS analysis of protein that coimmunoprecipitated with CRCT detected other 14-3-3 proteins such as GF14C, GF14E and GF14F (Table 1). From these observations, it is likely that CRCT can interact with 14-3-3 proteins other than GF14A and GF14B. Therefore, it is difficult to further discuss the significance of the CRCT/14-3-3 complex from the information of 14-3-3 proteins.
In this study, ChIP analyses suggested that all of starch synthesis related genes analyzed can be target genes of CRCT (Figures 5). However, the cell specific expression of these genes analyzed by the promoterGUS were clearly different from CRCT (Figure 2; Figures S3, S4 and S5). The cell specific expression of CRCT was rather similar to that of OsGPT2 and OsBEI (Figure 2; Figures S3, S4 and S5). It was also puzzling that the expression of CRCTtranscript in the leaf sheath was lower than that in the leaf blade, whereas the protein level of CRCT was higher in the leaf sheath than in the leaf blade (Figure 3). Some mobile transcription factors have been reported in plants. For example, a GRAS transcription factor SHORT-ROOT and homeobox KNOTTED1 proteins were shown to move from one cell to another (Lucas et al., 1995; Wu, & Gallagher, 2014). We hypothesize that CRCT protein may move somehow between cells, and the place where CRCT is expressed may be different from the place where it functions. However, it is also possible that CRCT mRNA moves between cells, or the stability of CRCT protein in the leaf blade is lower than that in the leaf sheath. Further study is needed to find out the mechanism.
The starch content in the leaf sheath was evidently correlated with the expression level of CRCT in the CRCT overexpression and knockdown lines (Morita et al. 2015; Morita et al. 2016). Thus, we consider that CRCT must be involved in starch accumulation, and may be related to the natural variation of starch content in the leaf sheath of rice (Goda et al., 2016). However, in our previous study, we constitutively overexpressed CRCT using Actin promoter (Morita et al., 2015). Thus, increased starch content in our overexpression lines may be an effect of ectopic expression. In this study, the difference in the starch content was not correlated with the expression level of CRCT among six rice cultivars (Figure 1). Nevertheless, we found a significant correlation between the expression of CRCT and starch synthesis-related enzymes (Figure 1). Therefore, we presume that this contradiction could be due to the difference in expression level of CRCT between cultivars being smaller than the difference in the expression level in transgenic lines. Although there is an inconsistency in the place of expression, CRCT must be an important factor controlling starch synthesis. Moreover, this study provide important progress towards the understanding of the mechanism of how CRCT controls starch synthesis in rice.