ARC5 is a dynamin-related GTPase essential for the division of chloroplasts in plants. The arc5 mutant frequently exhibits enlarged, dumbbell-shaped chloroplasts, indicating a role for ARC5 in the constriction of the chloroplast division site. In a screen for chloroplast division mutants with a phenotype similar to arc5, two mutants, cpd25 and cpd45, were obtained. CPD45 was identified as being the same gene as FHY3, a key regulator of far-red light signaling recently shown to be involved in the regulation of ARC5. CPD25 was previously named FRS4 and is homologous to FHY3. We found that CPD25 is also required for the expression of ARC5, suggesting that its function is not redundant to that of FHY3. Moreover, cpd25 does not have the far-red light-sensing defect present in fhy3 and far1. Both FRS4/CPD25 and FHY3/CPD45 could bind to the FBS-like ‘ACGCGC’ motifs in the promoter region of ARC5, and the binding efficiency of FRS4/CPD25 was much higher than that of FHY3/CPD45. Unlike FHY3/CPD45, FRS4/CPD25 has no ARC5 activation activity. Our data suggest that FRS4/CPD25 and FHY3/CPD45 function as a heterodimer that cooperatively activates ARC5, that FRS4/CPD25 plays the major role in promoter binding, and that FHY3/CPD45 is largely responsible for the gene activation. This study not only provides insight into the mechanisms underlying the regulation of chloroplast division in higher plants, but also suggests a model that shows how members of a transcription factor family can evolve to have different DNA-binding and gene activation features.
Chloroplasts, the endosymbiotic organelles derived from cyanobacteria, are the major site of photosynthesis and of numerous important biosynthetic pathways in plant cells (Dyall et al., 2004; Howe et al., 2008). Due to the endosymbiotic origin of chloroplasts, many genes involved in chloroplast division [e.g. FILAMENTOUS TEMPERATURE SENSITIVE Z (FtsZ), MINICELL D (MinD), and MINICELL E (MinE)] were identified by searching for homologs of bacterial cell division genes in databases of sequenced plant genomes (Osteryoung and Vierling, 1995; Osteryoung et al., 1998; Colletti et al., 2000; Itoh et al., 2001). However, many cyanobacterial cell division genes do not appear to have homologs in higher plants, indicating significant differences in the mechanisms of division between chloroplasts and cyanobacteria (Gao and Gao, 2011). A eukaryotic host genome-derived gene, ACCUMULATION AND REPLICATION OF CHLOROPLASTS5 (ARC5), was identified by mutant screening and positional cloning as having roles in chloroplast division (Gao et al., 2003). ARC5, a member of the eukaryotic dynamin family, is localized to the division furrow of chloroplasts on the cytosolic surface of the chloroplast's outer envelope. Many dynamin-related proteins are membrane-bound GTPases involved in the membrane remodeling that occurs during cellular processes such as endocytosis, mitochondrial fission and fusion, and phragmoplast formation (Praefcke and McMahon, 2004; Kuroiwa et al., 2008). Later, PLASTID DIVISION1 (PDV1) and PLASTID DIVISION2 (PDV2), which are homologous proteins, were identified as being involved in chloroplast division (Miyagishima et al., 2006). These proteins exist only in land plants and were suggested to be important for the localization of ARC5. FtsZ, ACCUMULATION AND REPLICATION OF CHLOROPLASTS3 (ARC3), ARC6, PDVs, and ARC5 are structural proteins that function as major components of the chloroplast division machinery. MinD, MinE, CDP1, and MULTIPLE CHLOROPLAST DIVISION SITE1 (MCD1) are required for the proper positioning of the chloroplast division machinery.
Numerous other genes have been identified as being involved in chloroplast division. The proteins encoded by these genes are probably not directly involved in the division process and their mechanisms of action are often not clear. For example, two MscS-like (MSL) proteins from Arabidopsis thaliana, MSL2 and MSL3, which are involved in osmotic shock sensitivity, were recently shown to affect chloroplast division through the Min system (Haswell and Meyerowitz, 2006). Mutations in these two genes indirectly affect the placement of the FtsZ ring in chloroplasts. Arabidopsis SUPRESSOR OF LON A (SulA) [also named GIANT CHLOROPLAST1 (GC1)] is a plastid-localized protein anchored to the stromal surface of the chloroplast inner membrane (Maple et al., 2004; Raynaud et al., 2004). Either loss of function or overexpression of AtSulA results in greatly enlarged chloroplasts. However, the exact mechanism by which this protein operates is unclear. CRUMPLED LEAF (CRL), CDK-DEPENDENT1 (CDT1) and FZO-LIKE (FZL) are further examples of Arabidopsis genes known to be involved in chloroplast division (Asano et al., 2004; Raynaud et al., 2004; Gao et al., 2006; Simkova et al., 2012).
As indicated above, most of the chloroplast division genes identified to date encode a chloroplast protein rather than a transcription factor or regulator of gene expression. We have identified two chloroplast division mutants, chloroplast division25 (cpd25) and chloroplast division45 (cpd45), which have enlarged and dumbbell-shaped chloroplasts, like those found in the arc5 mutant. Initially, the proteins encoded by CPD25 and CPD45 were presumed to interact with ARC5, which is thought to constrict the chloroplast outer envelope during chloroplast division. Map-based cloning of CPD45 indicated that it is the same gene as FAR-RED ELONGATED HYPOCOTYL3 (FHY3) and map-based cloning of CPD25 indicated that it is FAR1-RELATED SEQUENCE 4 (FRS4). FHY3, which is known to be a key regulator of the far-red light signal transduction pathway (Wang and Deng, 2002; Lin et al., 2007), was recently reported to regulate the expression of ARC5 in a reverse genetics study (Ouyang et al., 2011). FAR1-RELATED SEQUENCE 4 (FRS4), which encodes a protein of unknown function, is a homolog of FHY3. Our results suggest that FRS4/CPD25 is not involved in far-red light signaling and that FRS4/CPD25 and FHY3/CPD45 are not redundant to each other. We propose that FRS4/CPD25 and FHY3/CPD45 work together as a heterodimer that binds to a pair of FBL motifs in the promoter region of ARC5 and activates its expression. Our data also suggest that FRS4/CPD25 plays the major role in DNA binding, whereas FHY3/CPD45 plays the major role in gene activation.
Mature leaves of cpd25 and cpd45 have dumbbell-shaped and enlarged chloroplasts, similar to those of arc5
In an effort to identify novel chloroplast division genes, we conducted a screen for mutants with chloroplast division defects. cpd25 and cpd45 are two chloroplast division mutants characterized by elongated or dumbbell-shaped chloroplasts that are enlarged and have a constriction defect (Figure 1b,e). The chloroplast division phenotype of the cpd25 and cpd45 mutant is similar to, but weaker than, that of the previously reported arc5 mutant (Pyke and Leech, 1994; Gao et al., 2003). Because the arc5-1 mutant was derived from the Landsberg erecta (Ler) ecotype and cpd25 and cpd45 were in the Columbia (Col) background, we used a newly identified allele of ARC5 in the Col ecotype, arc5-3, in our mutant screen (Figure S1), for a better comparison of the phenotypes. In the fully expanded leaves of 5- to 6-week-old plants, the wild type has an average of 96 chloroplasts per cell (Figure 1a,g), the arc5 mutant has an average of five chloroplasts per cell (Figure 1d,g), the cpd25 mutant has an average of 15 chloroplasts per cell (Figure 1b,g), and the cpd45 mutant has an average of 27 chloroplasts per cell (Figure 1e,g).
Identification of CPD25 and CPD45
CPD25 was initially mapped to a region of approximately 190 kb with an F2 mapping population of approximately 5800 plants (Figure 2a, Table S1). This region was sequenced and a nonsense mutation was found in At1g76320, which encodes FRS4, a predicted transcription factor of unknown function (Figure 2b,c). A wild-type copy of At1g76320 was cloned and transformed into the cpd25 mutant. Among the 16 T1 plants, 11 were fully rescued (Figure 1c), three were partially rescued, and two were not rescued. Thus, CPD25 corresponds to At1g76320.
To identify CPD45, a similar strategy was used. A typical ethyl methanesulfonate (EMS)-induced point mutation, i.e. a C to T transition, was identified in the second exon of At3g22170 (Figure S2b), which was previously annotated as FHY3 (Wang and Deng, 2002) and recently reported as being involved in the regulation of ARC5 (Ouyang et al., 2011).
The cpd25 and cpd45 mutants identified initially were subjected to functional analysis, and three new mutant alleles of cpd25, i.e. cpd25-2, cpd25-3, and cpd25-4, and a new mutant allele of cpd45, i.e. cpd45-2, were identified (Figures 2b,c and S2b,c). These mutants all have chloroplast division phenotypes similar to those of cpd25 and cpd45 (Figure S3).
FRS4/CPD25 is phylogenetically related to FHY3/CPD45, but cpd25/frs4 does not have the far-red light sensing defect observed in fhy3/cpd45
Both FRS4/CPD25 and FHY3/CPD45 belong to the FRS family. A phylogenetic analysis using the protein sequence of the DNA-binding domain of members of this gene family indicated that FHY3/CPD45 was most closely related to FAR1, and that FRS4/CPD25 was also closely related to these two proteins (Figure 3a). Both FHY3 and FAR1 had been shown to be involved in far-red light signaling (Hudson et al., 1999; Wang and Deng, 2002). However, in contrast to the fhy3/cpd45 mutant, the frs4/cpd25 mutant, which is a null allele, is not defective in far-red light sensing (Figure 3b), suggesting that, unlike FHY3 and FAR1, FRS4/CPD25 has no role in far-red light signaling.
FRS4/CPD25 is required for the activation of ARC5
The chloroplast division phenotype of frs4/cpd25 is similar to that of arc5, pdv1, and pdv2, which were previously reported to have a chloroplast division defect characterized by dumbbell-shaped chloroplasts (Pyke and Leech, 1994; Gao et al., 2003; Miyagishima et al., 2006). Therefore, RT-PCR analysis was performed to investigate the expression level of ARC5, PDV1, and PDV2 in 20-day-old plate-grown plants. Various numbers of PCR cycles were tested to avoid saturation of amplification. As shown in Figure 4(a), the expression level of ARC5 in cpd25 is lower than that of the wild type. However, there is no apparent difference in PDV1 and PDV2 expression between the frs4/cpd25 mutant and the wild type (Figure 4a). We also analyzed the expression level of these three genes in 40-day-old soil-grown wild-type and cpd25 plants, and in the frs4/cpd25 mutant complemented with CPD25. Similarly, the level of ARC5 was lower in the mutant plants, whereas that of PDV1 and PDV2 was unaffected (Figure 4b). In complemented T1 transgenic plants, the expression level of ARC5 was similar to that in wild-type plants (Figure 4b).
We quantified the relative expression of ARC5, PDV1, and PDV2 using quantitative RT-PCR analysis. As shown in Figure 4(c,d), the expression of ARC5 in 20-day-old plate-grown frs4/cpd25 plants was 10.8% of that of wild-type plants and the expression of ARC5 in 40-day-old soil-grown frs4/cpd25 plants was 19.7% of that of wild-type plants. The expression level of PDV1 and PDV2 in frs4/cpd25 and fhy3/cpd45 was close to that of wild-type and complemented T1 transgenic plants (Figure S4). Therefore, the reduced expression of ARC5 is likely to be the main reason for the chloroplast division defect in frs4/cpd25. The enlarged and dumbbell-shaped chloroplasts in frs4/cpd25, which are similar to those observed in arc5, also support this notion.
Constitutive overexpression of ARC5 rescues the chloroplast division defects in frs4/cpd25
ARC5 was localized to the cytosolic side of the chloroplast division furrow and probably does not have a chloroplast transit peptide (Gao et al., 2003). A GFP-ARC5 fusion driven by the constitutively active 35S promoter, whose expression is unlikely to be regulated by FRS4/CPD25, was transformed into arc5 mutant plants. Constitutive overexpression of GFP-ARC5 in the arc5 mutant rescued the chloroplast division defect (Figure 5a–e), suggesting that the fusion protein was functional even in the presence of an N-terminal GFP tag. Constitutive overexpression of GFP-ARC5 in the independently transformed frs4/cpd25 mutant also rescued the chloroplast division phenotype (Figure 5g–j). Localization of GFP-ARC5 in fully rescued frs4/cpd25 mutant plants (Figure 5h–j) was similar to that in arc5 mutant plants complemented with 35Spro-GFP-ARC5 (Figure 5c–e) or the previously reported ARC5pro-GFP-ARC5 (Gao et al., 2003). Thus, rescue of the chloroplast division defects in frs4/cpd25 was simply due to the recovered expression of ARC5. Therefore, we conclude that the chloroplast division defect in frs4/cpd25 is due to the lack of ARC5 activation.
Identification of the 5′ -untranslated region and promoter region of ARC5
It has been suggested that FHY3 directly binds to the promoter of ARC5, and yeast one-hybrid analysis demonstrated that a motif (CACGCGC) −65 bp upstream of the ATG start codon is critical for this binding (Ouyang et al., 2011). However, our results indicate that the 5′ -untranslated region (UTR) of ARC5 probably has a length of between 118 and 124 bp (Figure S5).
To identify the promoter region and 5′-UTR of ARC5, a set of primers upstream of the start codon was designed based on the 5′ expressed sequence tag (EST) sequences of ARC5 genes from a variety of plant species to amplify the cDNA of ARC5 in Arabidopsis (Figure S5, Table S3). The two most critical primers are f2, which starts at −124 bp, and f3, which starts at −118 bp (with −1 being defined as the nucleotide immediately upstream of the start codon of ARC5; Figure S5a). These primers were used to amplify ARC5 cDNA with primer r2. For a better comparison of their amplification, the PCR was not saturated. The cDNA was strongly amplified with f3 and r2, but not with f2 and r2 (Figure S5c). The cDNAs could not be amplified with primers further upstream than −124 bp. Primer f1, which starts at −176 bp and contains an FBS motif (−169 to −163), did not amplify the cDNA (Figure S5c). Primers f1, f2, and f3 all amplified a genomic DNA fragment when used with primer r1, with almost equal amplification ability (Figure S5b). Therefore, the 5′-UTR of ARC5 probably has a length of between 118 and 124 bp and the promoter occurs upstream of this region.
We analyzed the genomic sequence of Arabidopsis and found that the intergenic region between ARC5 and its upstream gene is at most 323 bp. Then, we analyzed the promoter region of ARC5 and identified one FBS motif and two ‘ACGCGC’ FBS-like (FBL) motifs (Figures S5a and 6a) (Lin et al., 2007). The two FBL motifs (from −232 to −227 and −182 to −177 bp) are in the forward direction, whereas the single FBS motif (−169 to −163 bp) is in the reverse (Figure 6a).
FHY3/CPD45 can bind to the promoter of ARC5 to activate its expression
To understand the mechanism by which ARC5 is regulated, we first conducted an electrophoretic mobility shift assay (EMSA) to establish whether FHY3/CPD45 could bind directly to the promoter region of ARC5. Since full-length FHY3/CPD45 expressed in Escherichia coli localized to inclusion bodies, we expressed only the first 200 amino acid residues fused to an N-terminal yellow fluorescent protein (YFP) tag (YFP-CPD45N1). This fusion protein was soluble. The first 200 amino acids of FHY3/CPD45 contain the DNA-binding domain, and it has previously been shown that an N-terminal glutathione S-transferase (GST) fusion protein that includes this sequence can bind to the promoter region of FHY1 and FHL (Lin et al., 2007). A region of the ARC5 promoter that includes the FBS and FBL motifs was PCR amplified for analysis (Figure 6a). To establish whether these motifs in the promoter region of ARC5 are responsible for binding to FHY3/CPD45, the ‘cgcg’ sequences in these motifs were mutated to ‘tttt’ (Figure 6a).
The results of the EMSA are presented in Figure 6(b). As expected, YFP-CPD45N1 caused a strong up-shift of the biotin-labeled wild-type ARC5 promoter (lane 2). When a 50-fold excess of unlabeled competitor DNA was added to the reaction, only a fraction of the labeled ARC5 promoter DNA was up-shifted as before (lane 3), indicating strong competition for binding to CPD45N1. To rule out the possibility of non-specific binding, a 50-fold excess of an unlabeled mutated ARC5 promoter region containing m1, m2, and m3 mutations was added for competition analysis (lane 4). Almost no competition was observed. These results indicate that YFP-CPD45N1 specifically binds to the promoter region of ARC5. Furthermore, since no up-shift was observed when DNA of the mutated ARC5 promoter was incubated with YFP-CPD45N1 (lane 9), the FBS and FBL motifs in the promoter region are essential for DNA binding.
A mutant version of YFP-CPD45N1, which contains the S90F mutation present in cpd45, was incubated with both the wild type (lanes 5, 6, and 7 of Figure 6b) and the mutant ARC5 promoter (lane 10 of Figure 6b). No up-shift was observed, suggesting that this mutation abolishes the DNA-binding activity of YFP-CPD45N1 in vitro. Most homologs of FHY3/CPD45 in other species have an A residue at the position corresponding to S90 in FHY3/CPD45, which is structurally more similar to the S residue than the F residue (Figure S6). The S90 residue exists in a conserved FAR1 DNA-binding domain (Figure S2c) and many of the residues flanking S90 are highly conserved (Figure S6). Since the S to F mutation could cause a significant change in protein structure, it may severely affect the structure of FHY3/CPD45 and it is likely that the mutated FHY3/CPD45 also loses its DNA-binding activity in vivo.
A yeast one-hybrid assay was performed to analyze the interaction between FHY3/CPD45 and the promoter region of ARC5. Unfortunately, the ARC5 promoter is active in yeast and drives the expression of the HIS3 reporter gene (Figure S7). Therefore, the yeast one-hybrid assay is not suitable for evaluating this interaction.
A yeast two-hybrid assay was also performed to analyze the interaction between FHY3/CPD45 and FRS4/CPD25. Surprisingly, the yeast strain AH109 transformed only with pGBKT7-CPD45 could grow on medium that lacked histidine (Figure S8). Thus, when FHY3/CPD45 was fused to the Gal4 DNA-binding domain, it could activate the expression of the reporter gene, HIS3. This suggests that FHY3/CPD45 has gene activation activity in yeast.
To analyze whether FHY3/CPD45 could activate the expression of ARC5 in vivo, a construct containing an ARC5 promoter (which is 657 bp long and includes part of the upstream gene) fused to a GUS reporter gene (uidA) was made and bombarded into the leaves of fhy3/cpd45 plants. GUS was hardly expressed (Figure 6c). However, when this construct was co-transformed with a construct containing FHY3/CPD45 driven by a 35S promoter, GUS was strongly expressed (Figure 6c). GUS was not expressed when all FBS and FBL motifs in the ARC5 promoter were mutated, even when the promoter constructs were co-transformed with 35Spro-CPD45 (Figure 6c). These data further suggest that FHY3/CPD45 can activate the expression of ARC5 in vivo and that the FBS and FBL motifs in the promoter region are essential for the binding and gene activation.
Characterization of the FHY3/CPD45-binding site in the ARC5 promoter
To further understand how FHY3/CPD45 activates the expression of ARC5, the FBS and FBL motifs in the ARC5 promoter were mutated individually or in pairs and analyzed by EMSA (Figure 7a). For the promoter regions with a single mutation, all the probes were shifted in a manner similar to that of the wild type, except that they migrated somewhat faster through the gel. The mobility of these three probes varied slightly. Probes bearing the m3 mutation ran fastest and those with the m1 mutation ran slowest. Three mutated probes each containing a pair of mutations (i.e. only one FBS or FBL motif was not mutated) were also shifted and ran even faster. A part of the probe with the m1m3 mutation pair or with the m2m3 mutation pair was not shifted. The three probes bearing pairs of mutations also ran at different speeds. The probe with the m2m3 mutation ran faster than the other two probes and contained more unshifted DNA. The probe with the m1m2 mutations ran slowest and had no unshifted DNA. The increased migration of these probes could be due to reduced binding of FHY3/CPD45 and earlier disassociation. These data indicate that FHY3/CPD45 can bind to all three motifs, but that the strength of binding varies, with binding to the third motif (−169 to −163) being strongest and to the first (−232 to −227) weakest.
To further understand the role of the FBS motif and the two FBL motifs in FHY3/CPD45-mediated gene activation, the single and double mutant forms of the ARC5 promoter described above were fused to GUS and the effect of these mutations on the ability of the promoter to drive gene expression was tested (Figure 7b). Whereas the m1 and m2 mutations had almost no effect on the activity of the promoter, the m3 mutation resulted in reduced promoter activity. Furthermore, the m1m2 and m1m3 mutation pairs also reduced the activity of the promoter, and the m2m3 mutation abolished promoter activity. These data are in agreement with the EMSA results and suggest that the second and third FHY3/CPD45-binding sites are essential for the activity of the ARC5 promoter.
FRS4/CPD25 binds the ARC5 promoter more strongly than does FHY3/CPD45, but does not exhibit gene activation activity
To understand the role of FRS4/CPD25 in the activation of ARC5, the N-terminal DNA-binding domain of FRS4/CPD25 (the first 126 amino acid residues, corresponding to the DNA-binding domain of FHY3/CPD45) fused to an N-terminal YFP tag (YFP-CPD25N1) was expressed in E. coli and purified for EMSA (Figure 8). The DNA probes used were the same as those used for FHY3/CPD45. We found that CPD25N1 could bind to the wild-type ARC5 promoter. The m1, m2, or m1m2 mutation reduced the DNA-binding activity of CPD25N1. In contrast to CPD45N1, CPD25N1 did not bind to the probes with m3, m1m3, and m2m3 mutations, suggesting that the third motif is essential for the DNA-binding activity of CPD25N1.
The above results show that both CPD25N1 and CPD45N1 can bind to the promoter region of ARC5 (Figures 6-8). However,their binding strengths were quite different. For the EMSA results shown in Figure 8(a), the concentration of the proteins used was only one-fifth and the concentration of the probes used only one-tenth of that used in Figures 6(b) and 7(a). At this concentration, the majority of wild-type probe was shifted by CPD25N1, whereas no shift was observed for CPD45N1. Thus, the binding strength of CPD25N1 to the ARC5 promoter is much stronger than that of CPD45N1.
A GUS reporter assay similar to the one presented in Figure 8(b) was carried out. 35S promoter-driven FRS4/CPD25 and GUS genes driven by various versions of the ARC5 promoter were co-transformed into the leaves of the fhy3/cpd45 mutant. Although FRS4/CPD25 could strongly bind to the wild-type ARC5 promoter, as shown above, almost no GUS expression was observed in the leaves co-transformed with 35Spro-CPD25 and ARC5pro.WT-GUS. Therefore, FRS4/CPD25 does not have the gene activation activity exhibited by FHY3/CPD45.
Overexpression of FRS4/CPD25 and FHY3/CPD45 have different effects on ARC5 expression and chloroplast division
It was shown that overexpression of either FHY3 or FAR1 could suppress the far-red light sensing mutant phenotype of both genes, suggesting that their functions are somehow redundant (Wang and Deng, 2002). However, transformation of the fhy3/cpd45 mutant with a 35Spro-CPD25 transgene resulted in a more severe mutant phenotype, with an average of five chloroplasts per cell (Figure 9a). Quantitative RT-PCR data indicated that the level of ARC5 in the transgenic plants was even lower than that in fhy3/cpd45 (Figure 9b). Of the 21 T1 transgenic plants analyzed, 20 have a more severe chloroplast division defect than the fhy3/cpd45 mutant. This is probably because FRS4/CPD25 has DNA-binding activity but no gene activation activity. Overexpression of FRS4/CPD25 may prevent the binding of FHY3/CPD45, which has gene activation activity, to the promoter region of ARC5. The chloroplast division phenotype was rescued when the frs4/cpd25 mutant was transformed with a 35Spro-CPD45 transgene (Figure 9a). Of nine T1 transgenic plants, four showed a wild-type phenotype. Quantitative RT-PCR analysis indicated that the level of ARC5 in the transgenic plants was recovered to a level even slightly higher than that in the wild type (Figure 9b). This is probably because FHY3/CPD45 has gene activation activity but low DNA-binding activity. Overexpression of FHY3/CPD45 increases the chance of this protein binding to the ARC5 promoter, and thereby increases the expression of ARC5.
FRS4/CPD25 and FHY3/CPD45 may function as a heterodimer to regulate ARC5
Our results indicate that both FRS4/CPD25 and FHY3/CPD45 are required for the activation of ARC5. To further explore the mechanism by which these proteins activate ARC5, we conducted a bimolecular fluorescence complementation (BiFC) analysis to analyze their interactions. Both the N-terminal and C-terminal halves of YFP were fused to FHY3/CPD45 and the fusions were transiently expressed in Nicotiana benthamiana (tobacco) leaf cells by Agrobacterium tumefaciens-mediated transformation. No YFP signal was observed when cells were co-transformed with the controls, NYFP-CPD45 and CYFP, NYFP and CYFP-CPD45, NYFP-CPD25 and CYFP, and NYFP and CYFP-CPD25 (Figure S9). As shown in Figure 10(a), YFP signals were observed in the cells co-transformed with NYFP-CPD45 and CYFP-CPD45, suggesting that FHY3/CPD45 can self-interact. Co-localization of YFP with 4,6-diamidino-2-phenylindole (DAPI) indicated that the YFP signal was localized to the nucleus. However, FRS4/CPD25 could not self-interact in our BiFC analysis (Figure 10a). In tobacco leaf cells co-transformed with NYFP-CPD45 and CYFP-CPD25 or NYFP-CPD25 and CYFP-CPD45, YFP signals were also observed and co-localized with the DAPI staining (Figure 10a), suggesting that FHY3/CPD45 and FRS4/CPD25 form a heterodimer in the nucleus.
In this work, we have identified a new chloroplast division gene in Arabidopsis, FRS4/CPD25, by mutant screening and map-based cloning. Most of the previously identified chloroplast division genes, such as FtsZ, ARC1, ARC3, ARC6, MinD, MinE, MCD1, CDP1, PDV, ARC5, and FZL, encode proteins localized to chloroplasts (Osteryoung and Vierling, 1995; Osteryoung et al., 1998; Colletti et al., 2000; Itoh et al., 2001; Gao et al., 2003, 2006; Vitha et al., 2003; Miyagishima et al., 2006; Maple et al., 2007; Nakanishi et al., 2009; Zhang et al., 2009; Kadirjan-Kalbach et al., 2012). In contrast, FRS4/CPD25 and FHY3/CPD45 encode transcription factors that regulate chloroplast division. Since many genes are involved in chloroplast division and their expression needs to be regulated, it is likely that more genes like FRS4/CPD25 and FHY3/CPD45 will emerge.
Although fhy3 has been extensively studied, we and others only recently observed the defective chloroplast division phenotype (Ouyang et al., 2011). This is probably because most of the previous studies used young seedlings grown in the dark and under far-red light, conditions which do not allow chloroplast development (Barnes et al., 1996; Wang and Deng, 2002; Lin et al., 2007, 2008).
The enlarged, dumbbell-shaped chloroplasts of frs4/cpd25 are similar to those of arc5, pdv1, and pdv2. We compared the expression level of ARC5, PDV1, and PDV2 between the wild type and cpd25 mutant. Among these three genes, only the level of ARC5 was greatly reduced in frs4/cpd25 mutants. Moreover, constitutive overexpression of ARC5 rescued the chloroplast division defects in frs4/cpd25. Therefore, the reduced expression of ARC5 is the main reason for the chloroplast division defect in this mutant.
According to our observations, the 5′-UTR of ARC5 starts at least 118 bp upstream of the start codon (Figures S5a and 6a). The FHY3-binding sequence analyzed by yeast one-hybrid analysis in the earlier study (Ouyang et al., 2011) is only 65 bp upstream of the ATG start codon and seems to be a part of the 5′-UTR. We used the promoter region of ARC5 (from −64 to −285 bp) to perform a yeast one-hybrid analysis and found that this promoter has strong activity in yeast (Figure S7). Thus, the ARC5 promoter is not suitable for yeast one-hybrid assays. Instead, we performed an EMSA, which showed that FHY3/CPD45 binds to the promoter region of ARC5 (Figures 6 and 7).
Upstream of the 5′-UTR at −163 and −177 bp, one FBS motif (GCGCGTG) and one FBS-like motif (ACGCGC) are present on the reverse strand (Figure 6a). However, all previous studies of FHY3 binding focused on a single FBS motif (Lin et al., 2007, 2008; Li et al., 2010; Ouyang et al., 2011). In our study, BiFC analysis indicated that FHY3/CPD45 can self-dimerize or heterodimerize with FRS4/CPD25 in plant cells (Figure 10a). Our results suggest that a pair of inverted FBS or FBS-like motifs may enable the dimerized FHY3/CPD45 and FRS4/CPD25 to better activate downstream genes.
The functions of FRS4/CPD25 and FHY3/CPD45 are not redundant, because the activation of ARC5 requires the action of both of these proteins. This is supported by the fact that both the cpd25 and cpd45 mutants show a dramatic reduction in the level of ARC5 and an obvious chloroplast division phenotype (Figures 1 and 4). Although FRS4/CPD25 and FHY3/CPD45 are homologous (Figure 3a), their exact functions in the activation of ARC5 differ. FHY3/CPD45 has ARC5 activation activity, but low DNA-binding activity (Figures 6, 8 and 9); FRS4/CPD25 has no gene activation activity, but binds strongly to the ARC5 promoter (Figures 8 and 9).
Based on our results, we propose a working model in which FRS4/CPD25 and FHY3/CPD45 activate the expression of ARC5. In this model, FRS4/CPD25 and FHY3/CPD45 form a heterodimer that binds to the promoter region of ARC5 and activates its expression (Figure 10b). In the heterodimer, FRS4/CPD25, via its DNA-binding domain (Figure 10b, labeled with a dark green color), plays the major role in binding to the ARC5 promoter and FHY3/CPD45, via its gene activation domain (Figure 10b, labeled with a red color), plays the major role in the activation of ARC5. In the frs4/cpd25 mutant, FHY3/CPD45 homodimers cannot efficiently bind to the promoter of ARC5 and activate its expression. In the fhy3/cpd45 mutant, FRS4/CPD25 can strongly bind to the promoter region of ARC5; however, since FRS4/CPD25 has no gene activation activity, ARC5 is not activated.
The enhanced chloroplast division phenotype and the reduced expression of ARC5 resulting from the overexpression of FRS4/CPD25 in the fhy3/cpd45 mutant (Figure 9) supports our view that FRS4/CPD25 has no gene activation activity for ARC5, and that its primary function is DNA binding. The finding that overexpression of FHY3/CPD45 can rescue the frs4/cpd25 mutant phenotype and increase the expression of ARC5 (Figure 9) supports our view that FHY3/CPD45 has low DNA-binding activity, and is mainly involved in gene activation. These proteins belong to a family of transposase-derived transcription factors that consists of over 13 members (Figure 3a) (Lin and Wang, 2004). One family member, FRS9 (At4g38170), was not included in our phylogenetic analysis (Figure 3a) because it does not have a DNA-binding domain. However, we cannot exclude the possibility that it is also involved in gene regulation, especially through dimerization with other proteins in the same family. Why not have only one monomeric transcription factor to regulate gene expression? The divergence of this family and the dimerization of its members not only increase the number of genes targeted and pathways regulated by these proteins, but also fine-tune the expression of the target genes. Furthermore, the increased complexity of gene regulation would improve the capacity of plants to deal with all kinds of environmental conditions and their fitness in the nature.
Even though chloroplast division and far-red light signaling appear to be two independent pathways, it is well known that the light signal has numerous connections with chloroplast activity. For example, light is important for the greening of young seedlings; far-red light blocks the greening of Arabidopsis seedlings and plastid development via a phytochrome A-dependent mechanism (Barnes et al., 1996), and phototropins in Arabidopsis are involved in chloroplast movement (Jarillo et al., 2001; Kagawa et al., 2001; Sakai et al., 2001; Lin, 2002). FHY3/CPD45 is an activator of both FHY1, which represses hypocotyl elongation (Lin et al., 2007), and ARC5, which promotes chloroplast division. It is intriguing that these two different pathways were connected at FHY3 during evolution. ARC5 exists in a wide range of plants (Gao et al., 2003; Osteryoung and Nunnari, 2003; Gao and Gao, 2011). The existence of many chloroplasts in one cell is known to be beneficial to plants (Jeong et al., 2002; Ii and Webber, 2005; Koniger et al., 2008). It may be advantageous to have a transcription factor that regulates and links both of these pathways.
Plant materials and growth conditions
All Arabidopsis thaliana plants were of the Columbia-0 ecotype. For growth on plates, seeds were surface sterilized and sown on 1/2 MS medium. After 3 days of incubation at 4°C, the plates were transferred to a growth chamber at 22°C with cycles of 16 h of light (7 a.m. to 11 p.m.) and 8 h of darkness (11 p.m. to 7 a.m.). For far-red light treatment, plates were illuminated with light of wavelengths 720–740 nm for 4 days and then seedlings were photographed and measured for length.
Microscopy and phenotype analysis
To analyze the chloroplast division phenotype, a piece of leaf tissue from the fully expanded first true leaf was mounted in water and viewed with an Olympus CX21 microscope (http://www.olympus-global.com/). To obtain a better view, leaf tissue was fixed in 3.5% glutaraldehyde for 1 h and then placed in 0.1 m Na2EDTA (pH 9) for 2 h at 55°C to lyse the middle lamella and separate different cells.
Bright field images of chloroplasts were recorded with a charge-coupled device (CCD) camera mounted on an Olympus CX21 microscope. Fluorescence images of chlorophyll and GFP were acquired with a CCD camera coupled to a conventional Leica fluorescence microscope DM2500 (http://www.leica-microsystems.com/). Fluorescence of reconstructed YFP in the BiFC assay and DAPI staining were viewed with a conventional Olympus fluorescence microscope BX61 and captured with a CCD camera.
To quantify the chloroplast division phenotype, the mesophyll cell plan areas were measured from microscopy images using an image analysis system (Image Analysis System 10.0, Changheng, http://www.crisoptical.com/lm2_41_344.htm), and the chloroplast number in each cell was counted manually. Thirty mesophyll cells from the leaves of 40-day-old Arabidopsis plants were used for the quantification.
Mapping of CPD25 and CPD45
The cpd25 and cpd45 mutants were crossed with the Ler wild type to obtain F1 seeds. F2 seeds were harvested from self-crossed F1 plants. F2 or F3 plants were phenotyped by microscopy and genotyped using PCR markers designed to map CPD25 and CPD45. Approximately 5800 F2 and F3 plants were used for the fine mapping of CPD25. Markers developed in this study and used for the fine mapping are listed in Table S2. The final mapping interval was between CH1-28.60 and CH1-28.79, which is about 190 kb. Approximately 3000 F2 plants were used for the fine mapping of CPD45. Markers developed in this study and used for the fine mapping are listed in Table S1. The final mapping interval was between CH3-7.79 and CH3-7.88, which is about 90 kb.
The first 200 amino acids of FHY3/CPD45, which contain the DNA-binding domain, were used in a BLAST search of Arabidopsis proteins and sequences of the corresponding region in other proteins were retrieved. FRS9 does not have such a region and was not included in the analysis. Multiple alignments of amino acid sequences were performed using ClustalX (http://www.clustal.org/clustal2/). A phylogenetic tree was constructed by the neighbor-joining method using ClustalX, and 1000 bootstrap replicates were performed.
Electrophoretic mobility shift assay
Probes were PCR amplified, purified and labeled with the EMSA Probe Biotin Labeling Kit (Beyotime, http://www.beyotime.com/). Recombinant fusion proteins were expressed in E. coli, purified, and used for the EMSA, which was performed with a Chemiluminescent EMSA Kit (Beyotime, Jiangsu, China). More details are provided in Supplementary Methods.
We thank Dr Chuanyou Li for providing the pCAMBIA 1391Z plasmid, and Xin Chen for assistance. This work was supported by grants from the Natural Science Foundation of China (no. 30971439), Beijing Municipal Natural Science Foundation (no. 5102022), New Century Excellent Talents in University (no.10-0221) and the Fundamental Research Funds for the Central Universities.