Arabidopsis CAPRICE-LIKE MYB 3 (CPL3) controls endoreduplication and flowering development in addition to trichome and root hair formation Free

The specification and patterning of cell types is a crucial feature of development in multicellular organisms. In Arabidopsis thaliana, the differentiation of epidermal cells has been used extensively as a relatively simple model for analyzing cell fate specification. Several types of epidermal cells are differentiated in Arabidopsis. For example, root epidermal cells differentiate into either root hair cells or hairless cells(Dolan et al., 1993). Leaf epidermal cells of Arabidopsis can differentiate into trichomes,stomate guard cells and pavement cells. Several regulatory factors are known to be involved in this epidermal cell differentiation. For example, the glabra 2 (gl2) and werewolf (wer) mutant phenotypes show conversion of hairless cells to root hair cells(Masucci et al., 1996; Lee and Schiefelbein, 1999). GL2 encodes a homeodomain leucine-zipper protein, and WERencodes an R2R3-type MYB protein that activates GL2 expression(Rerie et al., 1994; Di Cristina et al., 1996; Masucci et al., 1996; Lee and Schiefelbein, 1999). GLABRA 3 (GL3) and ENHANCER OF GLABRA 3(EGL3) encode basic helix-loop-helix (bHLH) proteins that affect hairless cell differentiation in a redundant manner(Bernhardt et al., 2003). There are two other bHLH genes, AtMYC1(Urao et al., 1996) and TRANSPARENT TESTA 8 (TT8)(Nesi et al., 2000), that are in the same subgroup as GL3 and EGL3(Heim, 2003).

CAPRICE (CPC) encodes a small protein with an R3 MYB motif, and promotes root hair cell differentiation in Arabidopsis(Wada et al., 1997). CPC protein moves from hairless cells to neighboring hair-forming cells and represses expression of GL2 (Wada et al., 2002). Arabidopsis has several additional CPC-like MYB sequences in the Arabidopsis genome, including TRIPTYCHON (TRY), ENHANCER OF TRY AND CPC 1, 2 and 3 (ETC1, ETC2 and ETC3)(Schellmann et al., 2002; Kirik et al., 2004a; Kirik et al., 2004b; Esch et al., 2004; Simon et al., 2007). The clustered trichome phenotype of the try mutant indicates that TRY protein has a regulatory role in trichome formation(Hulskamp et al., 1994; Schellmann et al., 2002). ETC1 and ETC2 have been characterized and their relationship with CPC and TRY genetically examined(Kirik et al., 2004a; Kirik et al., 2004b; Esch et al., 2004). We have recently identified a fourth CPC-like MYB, At4g01060, independently of Simon et al. (Simon et al.,2007), and have named it CPC-LIKE MYB 3(CPL3).

Here, we examine the functions of the CPL3 gene in Arabidopsis. CPL3 redundantly regulates root hair and trichome development along with other CPC homologs. Notably, among the homologs, only CPL3 has pleiotropic effects on flowering development and epidermal cell size through the regulation of endoreduplication.

The Arabidopsis thaliana Col-0 ecotype was used as wild type. The cpc-2 mutant used in this study was described previously(Kurata et al., 2005). The cpl3-1 mutant was isolated from a Wisconsin T-DNA population. We backcrossed cpl3-1 to wild type and confirmed that all of the phenotypes associated with cpl3-1 co-segregate with the T-DNA insertion. We isolated the try-29760, etc1-1 and etc2-2mutants from a SALK T-DNA population. All mutants were in the Col-0 background. Double, triple and quadruple mutants of cpc, try, etc1,etc2 and cpl3 were screened from F₂ progeny using PCR to identify homozygous cpc-2, try-29760, etc1-1, etc2-2 and cpl3-1 plants. Selected double, triple and quadruple mutants were checked and documented in the F₃ generation. The 35S::CPCand CPCp::GUS transgenic lines were described previously(Wada et al., 1997; Wada et al., 2002). Seeds were surface-sterilized and sown on 1.5% agar plates as described previously(Okada and Shimura, 1990) and grown out for observation of seedling phenotypes. Seeded plates were kept at 4°C for 2 days and then incubated at 22°C under constant white light. For each mutant line, at least ten individual 5-day-old seedlings were assayed for root epidermis changes, and at least five 2-week-old third leaves were assayed for trichomes. Plants were grown in soil at 22°C under continuous light for determining flowering time, leaf size and ploidy. For flowering-related gene expression analyses (CO, FT and SOC1), plants were grown in soil under long-day conditions (16 hours light/8 hours dark).

Primer sequences are listed in Table 1.

A 0.8 kb PCR-amplified linear CPL3 genomic fragment (primers TW1167/TW1168) was subcloned into pBluescript SK+ (pBS; Stratagene)using Pyrobest DNA polymerase (Takara, Tokyo, Japan) to make pBS-CPL3. Next, the Acc65I to SalI fragment of pBS-CPL3 was ligated into the Acc65I and SalI sites of the pCHF3 binary vector(Jarvis et al., 1998) to create 35S::CPL3. PCR-generated constructs were completely sequenced following isolation of the clones to check for amplification-induced errors. Finally, the amplified and ligated constructs were cloned into transformation vector pJHA212K (Yoo et al.,2005).

A 3.0 kb PCR-amplified linear CPL3 genomic fragment (primers RT71/RT72) was digested with SalI and EcoRV and ligated into the SalI and EcoRV sites of pBS-2xGFP(Kurata et al., 2005) to create pBS-CPL3:2xGFP. PCR-generated constructs were completely sequenced following isolation of the clones to check for amplification-induced errors. Finally, the SalI to SacI fragment of pBS-CPL3:2xGFP was ligated into the SalI and SacI sites of the pJHA212K binary vector(Yoo et al., 2005) to create CPL3p::CPL3:GFP.

A 2.4 kb PCR-amplified promoter region of CPL3 (primers RT50/RT51)was digested with NotI and AccI and subcloned into pBS to create pBS-CPL3p. PCR-generated constructs were completely sequenced following isolation of the clones to check for amplification-induced errors. The SalI and BamHI-digested fragment of pBS-CPL3p was ligated into the SalI and BamHI sites of binary vector pBI101 (Clontech Laboratories,CA) to create the CPL3p::GUS construct.

A 1.9 kb PCR-amplified promoter region of ETC1 (primers RT46/RT47), a 3.0 kb PCR-amplified promoter region of ETC2 (primers RT48/RT49) and a 3.0 kb PCR-amplified promoter region of TRY (primers RT88/RT89), were digested with NotI and AccI and subcloned into pBS to create pBS-ETC1, -ETC2 and -TRY. The SalI and BamHI-digested fragment of pBS-ETC1 was ligated into the SalI and BamHI sites of binary vector pBI101 (Clontech Laboratories) to create ETC1p::GUS. The SalI and XbaI-digested fragments of pBS-ETC2 and pBS-TRY were ligated into SalI and XbaI of binary vector pBI101 to create ETC2p::GUS and TRYp::GUSconstructs. At least three T3 lines were isolated on the basis of their segregation ratios for kanamycin resistance for each transgenic line.

Plant transformation was performed by a floral dip method(Clough and Bent, 1998), and transformants were selected on a 0.5×MS agar plates containing 50 mg/l kanamycin. Homozygous transgenic lines were selected by kanamycin resistance. We isolated at least twelve T1 lines for each construct and selected at least six T2 and three T3 lines on the basis of their segregation ratios for kanamycin resistance. For each transgenic line, at least ten individual 5-day-old seedlings were assayed for root hair numbers, and at least five 2-week-old third leaves were assayed for trichome numbers. Promoter::GUS transgenic lines were analyzed by PCR (primers GUS+00+/GUS+09-). At least three individual plants were assayed for GUS activity in each of the three transgenic lines. The CPL3p::GUSconstruct was introduced into the cpc-2, try-29760, etc1-1, etc2-1and cpl3-1 mutants by conventional crosses and F2 seedlings were analyzed by PCR. At least five plants from each transgenic line were assayed for GUS activity.

Promoter::GUS plants were excised from the growth medium and immersed in X-Gluc solution containing 1.0 mM X-Gluc(5-bromo-4-chloro-3-indolyl-β-glucuronide), 1.0 mM K₃Fe(CN)₆, 1.0 mM K₄Fe(CN)₆, 100 mM NaPi (pH 7.0), 100 mM EDTA and 0.1% Triton X-100. Primary roots of 5-day-old seedlings were incubated at 37°C overnight. Cotyledons of 5-day-old seedlings, 2-week-old rosette leaves and hypocotyls, and 4-week-old inflorescences and siliques were incubated at 37°C for 3.5 hours.

In situ hybridization was as described(Kurata et al., 2003). DIG-labeled antisense RNA probes for CPC, ETC1, ETC2 and CPL3 were generated by transcribing pBS-cDNA (pBS-CPC, -ETC1, -ETC2 and -CPL3) digested with HindIII, SpeI, XhoI and SpeI, respectively. T3 polymerase was used for CPC, ETC1 and CPL3 probes, T7 polymerase for the ETC2 probe.

RNA extraction and semi-quantitative RT-PCR reactions were as described(Kurata et al., 2003). The CPL3 fragment was amplified with the RT73/RT92 primer pair. EF (At1g07930) was amplified with the EF1α-F/EF1α-R primer pair as described (Kurata et al.,2005).

Total RNA was extracted using the RNeasy Plant Mini Kit (Qiagen). On-column DNase I digestion was performed during RNA purification following the protocol described in the RNeasy Mini Kit handbook. First-strand cDNA was synthesized from 1 μg total RNA in a 20 μl reaction mixture using the Prime Script RT Regent Kit (Takara). Real-time PCR was performed in a Chromo4 Real-Time PCR Detection System (Bio-Rad, Hercules, CA) using SYBR Premix Ex Taq (Takara). PCR amplification employed a 30 second denaturing step at 95°C, followed by 5 seconds at 95°C and 30 seconds at 60°C with 45 cycles for CPL3 and 40 cycles for CYCA2;1, CYCA2;2, CYCA2;3,CYCA2;4, CYCA1;1, SIM, CO, FT, SOC1 and ACT2. Relative mRNA levels were calculated by iQ5 software (Bio-Rad), and normalized to the concentration of ACT2 mRNA. The primers were: CYCA2;1-F and CYCA2;1-R for CYCA2;1; CYCA2;2-F and CYCA2;2-R for CYCA2;2; CYCA2;3-F and CYCA2;3-R for CYCA2;3; CYCA2;4-F and CYCA2;4-R for CYCA2;4; CYCA1;1-F and CYCA1;1-R for CYCA1;1; SIM-F and SIM-R for SIM; CO-F and CO-R for CO; FT-Fand FT-R for FT; SOC1-F and SOC1-R for SOC1; and ACT2-F and ACT2-R for ACT2(Czechowski et al., 2004; Huang et al., 2005; Yoshizumi et al., 2006).

ANOVA (analysis of variance) was performed to determine the significance of differences between cpl3, 35S::CPL3 or CPL3p::CPL3, and wild type.

Nuclei were extracted and stained with CyStain UV Precise P (Partec,Münster, Germany) following the manufacturer's protocol. Flow cytometric analysis was performed by a Ploidy Analyzer PA flow cytometer (Partec),according to the manufacturer's instructions.

Root phenotypes were observed using an Olympus Provis AX70 microscope and an Olympus SZH binocular microscope. For each mutant or transgenic line, at least ten individual 5-day-old seedlings were analyzed for root hair number and root GUS activity. For the observation of trichomes, images were recorded with a VC4500 3D digital fine microscope (Omron, Kyoto, Japan) or digital microscope (VH-8000; Keyence, Osaka, Japan). At least five 2-week-old third true leaves were analyzed for trichome number and GUS activity for each mutant or transgenic line. For measurement of epidermal cell numbers, five 2-week-old third leaves were cleared with chloral hydrate:glycerol:water (8:2:1, w:v:v)(Yadegari et al., 1994), and visualized using a Zeiss Axio-plan2 microscope (Carl Zeiss, Germany).

CPL3p::CPL3:GFP transgenic lines were stained with 5 μg/ml propidium iodide (PI) for 30 seconds and mounted in water. Confocal images were obtained with a 40× water-immersion objective on a Zeiss LSM-Pascal or a Zeiss LSM-510 Meta confocal laser scanning microscope using 488 nm laser lines for GFP excitation. Image processing was with Photoshop version 7.0(Adobe Systems, CA).

To observe the phenotype of trichomes, rosette leaves or inflorescences were attached to the stage and cooled in liquid nitrogen. Observations were made in low vacuum with a scanning electron microscope (model JSM5610-LV;JEOL, Akishima, Japan).

Vectors and yeast strains were obtained from Clontech (Mountain View, CA;MATCHMAKER Two-Hybrid System). CPC, TRY, ETC1, ETC2 and CPL3 full-length proteins were fused to the GAL4 DNA-binding domain in pGBT9. GL3, EGL3, AtMYC1 and TT8 full-length proteins were fused with the GAL4 activation domain in pGAD424. Yeast strain Y187 was transformed with the appropriate plasmids using carrier DNA and the lithium acetate method(Kallal and Kurjan, 1997). Following the Yeast Protocols Handbook (Clontech), a β-galactosidase assay was performed on each transformant using O-nitrophenylβ-d-galactopyranoside (Sigma) as substrate.

A search of the Arabidopsis genome sequence revealed four MYB gene sequences with high homology to CPC: TRY, ETC1, ETC2 and CPL3. The CPL3-encoded protein is closely related to the proteins encoded by CPC-like MYB family members CPC, TRY,ETC1 and ETC2, and two rice (Oryza sativa) homologs,Os01g43180 and Os01g43230 (Fig. 1A). A phylogenetic tree based on the alignment of amino acid sequences with the R2R3-type MYB family members WER, GL1 and MYB23, indicated that the two rice homologs are more closely related to the CPC-like MYB family than to the R2R3-type MYB family, and that CPL3 is more closely related to ETC1 than to the other CPC-homologous proteins(Fig. 1B).

To assess the function of the CPL3 gene, we identified a T-DNA mutant allele, cpl3-1, in the Wisconsin T-DNA collection. cpl3 plants have a T-DNA insertion in the first exon(Fig. 1C). Using semi-quantitative RT-PCR, no CPL3 mRNA could be detected in the cpl3 mutant, indicating that transcription of the CPL3 gene is disrupted by the T-DNA insertion (Fig. 1D). We also screened several T-DNA-tagged pools for knockouts of CPC-like MYB genes, and identified the mutant lines etc1-1(Kirik et al., 2004a) and etc2-2. Additionally, ecotype Col-0 alleles of cpc and try mutants were found and named cpc-2(Kurata et al., 2005) and try-029760. Thus, all of the mutants used in this study are in a Col-0 background.

The root hair phenotype of the cpc-1 (WS background)(Wada et al., 1997) and cpc-2 (Col-0 background) mutant lines is characterized by the formation of approximately one-fourth as many root hairs as the wild type (see Fig. S1A in the supplementary material), respectively. In the cpl3-1line, the relative number of root hairs was about 80% that of wild type(Fig. 2B, Table 2). There is a distinct possibility that functional redundancy provided by the presence of similar genes obscures the phenotype of each individual knockout mutant. Therefore, we made cpl3 double mutants with each of the other CPC-like MYB mutants (Table 2). It had already been reported that the cpc-1 try-82 and cpc-1 etc1-1double mutants have very few root hairs(Schellmann et al., 2002; Kirik et al., 2004b). We confirmed these observations using the cpc-2, try-29760 and etc1-1 alleles (see Fig. S1A in the supplementary material). Root hair production in the cpl3 cpc double mutant was about 50% of that in the cpc single mutant (see Fig. S1A in the supplementary material). However, root hair production in cpl3 try, cpl3 etc1 and cpl3 etc2 double mutants was not significantly different to that in wild type (Table 2).

Trichome formation in the double mutants is more complicated than root hair formation because of the three developmental aspects of trichome formation:number, clustering and branching. There were more trichomes on cpc-2than on wild type (see Fig. S1B in the supplementary material), as had been reported using the WS allele, cpc-1(Schellmann et al., 2002). As was the case with cpc mutant alleles, cpl3 plants also had more trichomes: roughly 80% more than wild type(Table 3, Fig. 2N). The cpl3 cpcand cpl3 etc2 double mutants had more trichomes than the parental single-mutant lines (Table 3, Fig. 2O,Q). The combination of try-EM1 (Folkers et al.,1997) with cpc-1 resulted in an increase in trichome clustering (Schellmann et al.,2002). The cpl3 try double mutant had a slightly increased percentage of clustered trichomes compared with the trysingle mutant (Table 3). By contrast, there were no clusters on the cpl3 single mutant, cpl3 cpc, cpl3 etc1, or cpl3 etc2 double mutants(Fig. 2N,O,Q, Table 3).

The etc1-1 try-82 cpc-1 triple mutant produced a large number of trichomes (Kirik et al.,2004b). In our experiment, the cpl3 try etc1 triple mutant had more clusters, though the trichome number was similar to that of cpl3 (Fig. 2R, Table 3). The cpl3 cpc try triple mutant had a more extreme phenotype, with heavy marginal clusters (Fig. 2S). Each cluster in cpl3 cpc try was larger than any on cpc try, but it was difficult to distinguish individual trichomes so they could not be accurately counted (see Fig. S2 in the supplementary material). The leaves of the cpl3 cpc try etc1 quadruple mutant were entirely covered by trichomes (Fig. 2T), which is similar to that seen in the transformant line 35S:GL1 35S:R(Larkin et al., 1994). Unlike the etc1-1 try-82 cpc-1 triple mutant(Kirik et al., 2004b), the cpl3 cpc try etc1 quadruple mutant developed a large trichome cluster covering even the midvein. Almost all of the adaxial epidermal cells appeared to have differentiated into trichomes (Fig. 3A). Differential interference contrast (DIC) images of the quadruple mutant showed that the epidermal layer was completely made up of trichome cells, to the exclusion of pavement cells, socket cells and guard cells (Fig. 3C). Hypocotyls and inflorescences were also covered with trichomes(Fig. 3B and see Fig. S3A in the supplementary material), and almost every hypocotyl epidermal cell had differentiated into trichomes (see Fig. S3B in the supplementary material). SEM images of the adaxial surface of a true leaf showed that there was a large variation in the size and branch number of individual trichomes on the quadruple mutant (Fig. 3D).

try mutant trichomes have increased DNA content and more branches than wild type (Hulskamp et al.,1994). By contrast, cpl3 mutant trichomes had consistently fewer branches than wild type, with 55% of cpl3trichomes having two branches (Table 4, Fig. 3E,F). These trichome phenotypes were also found on plants grown in soil (see Fig. S4 in the supplementary material). Double mutants with cpl3 and the other CPC homologs did not significantly further reduce the branching observed in cpl3 (Table 4).

CPL3 was expressed under the control of the 35S promoter to produce an overexpressing line for comparison with the 35S::CPC, 35S::TRY,35S::ETC1 and 35S::ETC2 lines(Wada et al., 1997; Schellmann et al., 2002; Kirik et al., 2004a; Kirik et al., 2004b). As with the overexpression lines of its homologs, 35S::CPL3 had more than the normal number of root hairs, and no trichomes(Fig. 2I,J,U,V, Tables 2, 3, and see Fig. S5 in the supplementary material). These results indicate that each of the CPC-like MYB homologs has a similar function for root hair and trichome formation when overexpressed under the control of the 35S promoter. When the CPL3 gene's own promoter was used to increase CPL3expression (CPL3p::CPL3), a relatively small amount of ectopic root hair formation was observed (Fig. 2K,L, Table 2, and see Fig. S5A in the supplementary material), and there were no trichomes produced (Fig. 2W,X, Table 3, and see Fig. S5B in the supplementary material).

Previous analyses in yeast have shown that CPC protein binds to the bHLH domain of the maize R protein (Wada et al., 2002), and that CPC, TRY, ETC1 and ETC2 bind to the bHLH-domain-containing GL3 protein (Kirik et al., 2004a; Kirik et al.,2004b; Esch et al.,2003). To test whether CPL3 can also interact with GL3, we performed a yeast two-hybrid analysis. All of the CPC homologous proteins(CPC-BD, TRY-BD, ETC1-BD, ETC2-BD and CPL3-BD) bound strongly to GL3-AD, and there also was significant binding to EGL3-AD and AtMYC1-AD (see Fig. S6 in the supplementary material).

CPL3 expression was examined in plant tissues using real-time PCR. CPL3 was more strongly expressed in shoots (including a few small true leaves) than in roots of seedlings(Fig. 4A). The relatively strong expression of CPL3 in mature plants was specifically observed in siliques and buds (Fig. 4A). Shellman et al. reported that in trichomes, TRY expression was strongest, followed by CPC(Schellmann et al., 2002). In our hands, in situ hybridization of shoot meristem tissue indicated that the expression of CPC was stronger than that of ETC1 in trichomes (Fig. 4B,C). We could not detect ETC2 or CPL3 expression in young trichomes(Fig. 4D,E).

To analyze expression at the cellular level, we made CPL3promoter-GUS fusions. CPL3p::GUS was expressed in young leaves and mostly restricted to stomate guard cells in leaves, cotyledons and hypocotyls(Fig. 4F-H). We could not detect CPL3p::GUS in trichomes. CPL3 was detectable in roots by real-time PCR (Fig. 4A), but no CPL3p::GUS was detected in roots(Fig. 4I). Consistent with real-time PCR, strong CPL3p::GUS expression was observed in inflorescences and developing seeds in siliques(Fig. 4J-M).

Protein localization was determined using protein-2XGFP fusion constructs driven by the CPL3 promoter (Fig. 4N). In CPL3p::CPL3:GFP transgenic plants, a strong GFP signal was observed in the guard cells(Fig. 4N), which was also consistent with the expression pattern of the CPL3p::GUS construct(Fig. 4F). We could not check for GFP localization in trichomes of these transgenic plants because increased CPL3 gene dosage prevents the formation of trichomes(Fig. 2W,X, Table 3, and see Fig. S5B in the supplementary material). There was also no GFP signal in root epidermis of CPL3p::CPL3:GFP plants, which is consistent with the CPL3p::GUS data.

The guard cell-specific protein localization of CPL3(Fig. 4N) indicates some specific involvement with stomate initiation or development, but there was no difference in the distribution or cluster formation of guard cells among cpl3, 35S::CPL3, CPL3p::CPL3 or wild-type plants(Table 5, and see Tables S1 and S2 in the supplementary material). Also, the double mutant cpl3 etc2did not show any aberrant stomata phenotype (see Table S3 in the supplementary material). Similar to root hairs, stomates of hypocotyls in wild-type Arabidopsis develop from epidermal cells that overlie two cortical cell files (Berger et al.,1998; Hung et al.,1998), a location known as the `S' (stomate) position. Therefore,we also checked the position of stomates on hypocotyls. About 84% of stomates are located in the S position and 16% in the `N' (non-stomate) position in Col-0 plants (Table 6). There was no significant difference in the positions of hypocotyl stomata among cpl3, CPL3p::CPL3 and wild-type plants(Table 6, and see Table S4 in the supplementary material). 35S:CPL3 transgenic lines,however, showed a significant difference in hypocotyl stomate distribution,with 54% of its stomata at the S position and 46% at the N position(Table 6, and see Table S5 in the supplementary material). These observations suggest that CPL3 is involved in the distribution of hypocotyl guard cells.

In addition to the effects on root hair and trichome formation, cpl3 mutant plants were affected in flowering time. As shown in Table 7, cpl3 mutant plants flowered earlier than wild type (28.9±0.5 versus 37.6±0.5 days) and with fewer leaves (8.2±0.3 versus 17.4±0.6). By contrast, cpc, try, etc1 and etc2 mutant plants were not significantly different from wild type(Table 7). 35S::CPL3transgenic plants flowered slightly later than wild type (41.1±1.1 versus 37.6±0.5 days) and with more leaves (28.5±1.7 versus 17.4±0.6) (Table 7, and see Table S6 in the supplementary material). To clarify the effect of the CPL3 gene on flowering, the expression of some flowering-related genes was examined in plant lines with altered CPL3 expression by real-time PCR. FLOWERING LOCUS T (FT), SUPPRESSOR OF OVEREXPRESSION OF CO 1 (SOC1) and CONSTANS(CO) play a central role in controlling floral transition (flowering)(Bagnall, 1993; Lee et al., 2000; Koornneef et al., 1991). Expression of FT and SOC1 were higher in the cpl3mutant than in wild type under long-day (LD) conditions during leaf development (Fig. 5A,B). In 21-day-old leaves, expression of FT was about 9.5-fold higher, and that of SOC1 3.2-fold higher in cpl3 than in wild type(Fig. 5A,B). Expression of CO in cpl3 was not significantly different from that in wild type (Fig. 5C). These results suggest the involvement of CPL3 in flowering regulation by repressing FT and SOC1 expression.

In addition to the altered flowering phenotype, cpl3 mutant plants were much larger and CPL3-overexpressing plants were much smaller than wild type (Fig. 6A). We confirmed this dwarf phenotype of CPL3-overexpressing plants with three 35S::CPL3 and five CPL3p::CPL3 independent transgenic lines. The fresh weight of cpl3 mutants was about 50% greater than that of wild type, and CPL3p::CPL3 plants were about 30% the fresh weight of wild type (Fig. 6B). Leaf epidermal cells in the cpl3 mutant were remarkably larger, and cells from CPL3p::CPL3 were smaller than those of wild type(Fig. 6C-E). This translates into a significant difference in overall plant size. For example, the cpl3 mutant third leaf was about 13% larger, and the CPL3p::CPL3-overexpressing line was about half the size, of wild type. However, there was no significant difference in leaf cell numbers(Table 8). These observations demonstrate that enhanced growth of cpl3 was caused by hypertrophy rather than hyperplasia of leaf cells. In addition, hypocotyl cells and hypocotyls of cpl3 were elongated, but those of 35S::CPL3were rounded, resulting in slightly shortened hypocotyls(Fig. 6F-K).

Because endoreduplication is generally thought to provide a mechanism for increasing cell size, and the try mutation increases trichome endoreduplication (Szymanski and Marks,1998), we analyzed the ploidy of leaf cells from wild-type, cpl3, 35S::CPL3 and CPL3p::CPL3 plants. In the cpl3mutant, the number of 32C and 16C cells was increased in 16-day-old first leaves (Fig. 7A; the proportions of 32C and 16C were 0% and 26.9% for Col-0, and 2.6% and 29.2% for cpl3, respectively). This result suggests that the altered phenotypes of the cpl3 mutant, including hypertrophic cell growth, are associated with an increase in endoreduplication. By contrast, 35S:CPL3 and CPL3p::CPL3 had reduced DNA levels in 16-day-old first leaves (Fig. 7A; the proportions of 4C and 2C for Col-0 were 19.6% and 16.9%,for 35S::CPL3 24.3% and 27.4%, and for CPL3p::CPL3 33.1% and 26.4%, respectively). In 3-week-old third leaves, the ploidy differences were observed more clearly (Fig. 7B;the proportions of 16C for Col-0, cpl3, 35S::CPL3 and CPL3p::CPL3 were 2.1%, 12.0%, 0.7% and 1.0%, respectively). This study demonstrates that CPL3 is likely to have some role in ploidy-dependent epidermal cell growth in Arabidopsis.

Endoreduplication is a type of cell cycle that skips the cell division steps of mitosis and thus affects cell cycle-related gene expression. We examined the expression of the cell cycle-related genes CYCA2;1, CYCA2;2,CYCA2;3, CYCA2;4 and CYCA1;1, which have been reported to be involved in endoreduplication in Arabidopsis(Yoshizumi et al., 2006; Imai et al., 2006), and of SIM, which is a cell cycle regulator controlling the onset of endoreduplication (Churchman et al.,2006). Expression of these genes in Col-0, cpl3,35S::CPL3 and CPL3p::CPL3 was analyzed by real-time PCR(Fig. 8). Although expression of CYCA2;1 and CYCA2;2 in the cpl3 mutant seemed to be somewhat reduced, and in overexpressers (35S::CPL3 and CPL3p::CPL3) seemed to be increased, compared with wild type(Fig. 8A,B), no significant changes were observed for any of the CYCA genes(Fig. 8A-E). Compared with wild type, expression of SIM was higher in the cpl3 mutant and lower in 35S::CPL3 and CPL3p::CPL3 during leaf development(Fig. 8F). Significant reduction in expression was observed in 30-day-old 35S::CPL3 and 12-day-old CPL3p::CPL3 plants compared with wild type(Fig. 8F). These results suggest that increased ploidy in the cpl3 mutant and decreased ploidy in CPL3 overexpressers are involved in the function of SIM.

Historically, cpc was isolated as a mutant with few root hairs(Wada et al., 1997). Although the typical plant MYB gene encodes a R2-R3 type MYB region(Rosinski and Atchley, 1998), CPC encodes a small R3-type MYB of only 94 amino acids. Not long after isolation of CPC, TRY was isolated as a CPC-homologous gene from a trichome-clustering mutant(Schellmann et al., 2002). More recently, ETC1, ETC2 and ETC3 were isolated as enhancers of try and cpc(Kirik et al., 2004a; Kirik et al., 2004b; Esch et al., 2004; Simon et al., 2007). We also independently isolated these three genes by homology with CPC. In this paper, we describe the isolation of CPL3, and provide evidence that it has a redundant function with the other homologs of CPC. In addition to its MYB function, CPL3 has epistatic effects on a number of crucial plant development and growth mechanisms. From the observation of double, triple and quadruple mutant phenotypes, it is clear that CPCplays the dominant role in the regulation of root hair formation, and TRY plays the dominant role in trichome formation, but all five CPC-like MYB genes, including CPL3, have a redundant function in root hair and trichome formation (Figs 2, 3, Tables 2, 3, and see Fig. S1 in the supplementary material). The reduction in root hair number compared with wild type was significant in the cpl3-1 single mutant, but not in cpl3 try, cpl3 etc1 or cpl3 etc2 double mutants(Table 2). Because CPC represses its own expression (Wada et el.,2002), it is possible that CPC-homologous genes also repress CPC expression. Thus, disruption of the other CPC-homologous genes in the cpl3 mutant background may enhance CPC expression, which leads to the formation of many root hairs. The conversion of almost all adaxial surface leaf cells and hypocotyl cells into trichome cells in the cpc try etc1 cpl3 quadruple mutant indicates that CPL3 either directly or indirectly shares a similar function with its homologs, because the cpc try etc1 triple mutant had a number of epidermal cell types other than trichome cells(Fig. 3, and see Fig. S3A in the supplementary material).

When CaMV 35S is used as a promoter for the expression of CPL3,the number of root hairs is increased, and no trichomes are formed in transformant lines, as observed in 35S::CPC, 35S::TRY, 35S::ETC1 and 35S::ETC2 transformants (Fig. 2, Tables 2, 3)(Wada et al., 1997; Schellmann et al., 2002; Kirik et al., 2004a; Kirik et al., 2004b). Yeast two-hybrid analyses showed that CPC, TRY, ETC1, ETC2 and CPL3 are capable of interacting with GL3, EGL3 and AtMYC1 (see Fig. S6 in the supplementary material). These findings demonstrate that CPL3 and the other CPC-like MYB proteins have a similar binding function. Although EGL3 had higher promoter activity and RNA accumulation than GL3, the mutant phenotype of egl3 was `weaker' than that of gl3(Bernhardt et al., 2003; Bernhardt et al., 2005). The strong trichome deficiency and proliferation of root hair phenotypes of the gl3 mutant are probably due to the strong binding activity of GL3 with the CPC-like MYB proteins.

CPL3p::GUS was mainly expressed around stomata(Fig. 4F-H), but the cpl3 mutant, cpl3 etc2 double mutant and CPL3overexpressors did not have aberrant stomatal phenotypes(Table 5, and see Tables S1,S2, S3 in the supplementary material). We counted stomates on cotyledons,which do not differ much in size regardless of their genotype. Therefore,stomatal density is apparently relatively constant(Table 5). The stomatal phenotypes of true leaves were also observed several times. Thus, we concluded that CPL3 did not affect stomate formation in leaves, although the gene is expressed there. The expression patterns of CPC homologs have been roughly classified into two groups. CPCp::GUS, TRYp::GUS and ETC1p::GUS are expressed mainly in roots and trichomes, and ETC2p::GUS and CPL3p::GUS are expressed in young leaves and guard cells (see Fig. S7 in the supplementary material). Thus, GUS expression by the CPC-like MYB family is found in tissues throughout the entire plant body. CPL3p::GUS expression was reduced in the cpc and etc1 backgrounds (see Fig. S8 in the supplementary material). These results suggest that the members of this regulatory protein family play different roles. Given the complexity of regulatory cascades and the contributions of phytohormones, cell-wall-associated proteins and cytoskeleton structures, it is nonetheless likely that this gene family has fairly direct control over epidermal differentiation, development and integration for the entire plant.

Although the general effect of CPL3 on cell fate is similar to that of CPC, TRY, ETC1 and ETC2, CPL3 has several characteristics that make it distinct from the other CPC homologs. First, the cpl3-1 mutant itself has a reduced number of root hairs,whereas etc1-1 and etc2-2 have normal root hair numbers(Fig. 2B, and see Fig. S1A in the supplementary material). cpl3-1 also produces an increased number of trichomes, similar to cpc-2(Fig. 2N, and see Fig. S1B in the supplementary material). Secondly, most stomata are distributed in the S position in wild-type hypocotyls, but transformant line 35S::CPL3trichomes are evenly distributed in both the S and N positions(Table 6, and see Table S4, S5 in the supplementary material). Thirdly, cpl3 plants have an early flowering phenotype (Table 7,and see Table S6 in the supplementary material). Expression of the FTand SOC1 genes in the cpl3 mutant was increased compared with wild type (Fig. 5A,B). A flowering regulatory cascade model would thus include repression of FT and SOC1 by CPL3. CPL3 expression is significantly higher in 35S::CPL3 and CPL3p::CPL3transformant lines than in wild type during leaf development(Fig. 5D). If CPL3 directly represses FT and SOC1, their expression in 35S::CPL3 and CPL3p::CPL3 plants would be reduced significantly (Fig. 5A,B). One possibility is that CO or some other factor overcomes the function of CPL3 to repress FT and SOC1. Fourthly, cpl3 plants have an abnormally large growth phenotype, and CPL3 overexpressers have a dwarf phenotype (Fig. 6). The hypertrophic cell phenotype of cpl3 is associated with an increase in endoreduplication (Fig. 7). CPL3 is thus likely to play an essential role in ploidy-dependent epidermal cell growth in Arabidopsis. Endoreduplication is generally thought to provide a mechanism for increasing cell size(Sugimoto-Shirasu and Roberts,2003), although the correlation between ploidy level and cell size is not always high (Leiva-Neto et al.,2004; Gendreau et al.,1998; Schnittger et al.,2003).

Previous studies have shown that a mutation in TRY leads to increased endoreduplication in trichomes and reduced endoreduplication in the epidermis (Szymanski and Marks,1998). This is the opposite of what happens in the cpl3mutant. Because a mutation in CPL3 leads to an increase in endoreduplication in the epidermis (Fig. 7), a decrease in trichome branching might be the result of reduced endoreduplication in trichomes(Fig. 3F, Table 4). It has been postulated that TRY is expressed in trichomes, reducing endoreduplication in those cells, followed by diffusion into neighboring cells to mediate lateral inhibition (Schellmann et al., 2002). Thus, we propose a model in which CPL3 is expressed in young leaf epidermal cells and represses endoreduplication, after which it affects neighboring trichome cells by slightly promoting endoreduplication.

A2-type cyclins play an important role in regulating endoreduplication in Arabidopsis (Burssens et al.,2000; Imai et al.,2006; Yoshizumi et al.,2006). CYCA2;1 is expressed in various differentiated cells, such as guard cells (Burssens et al., 2000), and loss of CYCA2;3 increases polyploidy in mature true leaves (Imai et al.,2006). However, we could not detect any significant change in the expression of CYCA genes in cpl3 mutant and transgenic Arabidopsis plants expressing CPL3. SIM is a cell cycle regulator that controls endoreduplication onset in Arabidopsis(Churchman et al., 2006). SIM transcript levels are increased in GL3-overexpressing lines and decreased in the gl3 egl3 double mutant(Churchman et al., 2006). Because CPL3 can bind to GL3 and EGL3 (see Fig. S6 in the supplementary material), it might inhibit GL3 and/or EGL3 function to induce expression of SIM (Fig. 8F).

The CPC-like MYB gene families are thought to have evolved by gene duplication (Fig. 1C). Gene family members that have not completely diverged functionally and thus retain some functional redundancy may represent intermediate stages of regulatory specification (Thomas, 1993; Cooke et al., 1997). As such,they can provide considerable potential for adaptive or evolutionary responses to environmental changes or occupation of a new ecological niche through selection for the most advantageous cell size and flowering time.

We thank Dr Yoshizumi for technical advice, M. Sato, K. Toyooka, M. Wakazaki, Y. Miyazaki, H. Oka and T. Gohara for technical assistance, and T. Araki, S. Yamaguch, Y. Kamiya, T. Ishida and T. Kurata for useful suggestions. This work was supported in part by Grants-in-Aid from the Ministry of Education, Science, Sports and Culture of Japan (15770152).