Control of Plant Trichome and Root-Hair Development by a Tomato (Solanum lycopersicum) R3 MYB Transcription Factor

In Arabidopsis thaliana the CPC-like MYB transcription factors [CAPRICE (CPC), TRIPTYCHON (TRY), ENHANCER OF TRY AND CPC 1, 2, 3/CPC-LIKE MYB 3 (ETC1, ETC2, ETC3/CPL3), TRICHOMELESS 1, 2/CPC-LIKE MYB 4 (TCL1, TCL2/CPL4)] and the bHLH transcription factors [GLABRA3 (GL3) and ENHANCER OF GLABRA 3 (EGL3)] are central regulators of trichome and root-hair development. We identified TRY and GL3 homologous genes from the tomato genome and named them SlTRY and SlGL3, respectively. Phylogenic analyses revealed a close relationship between the tomato and Arabidopsis genes. Real-time reverse transcription PCR analyses showed that SlTRY and SlGL3 were predominantly expressed in aerial parts of developing tomato. After transformation into Arabidopsis, CPC::SlTRY inhibited trichome formation and enhanced root-hair differentiation by strongly repressing GL2 expression. On the other hand, GL3::SlGL3 transformation did not show any obvious effect on trichome or non-hair cell differentiation. These results suggest that tomato and Arabidopsis partially use similar transcription factors for epidermal cell differentiation, and that a CPC-like R3 MYB may be a key common regulator of plant trichome and root-hair development.

Although both tomato and Arabidopsis have trichomes, tomato trichomes are distinct from Arabidopsis trichomes. Arabidopsis has non-glandular three-branched unicellular trichomes that form on stem and leaf surfaces to an extent that depends on the ecotype [23,24]. On the other hand, tomato trichomes are highly diverse in morphology and chemistry [25][26][27]. Tomato trichomes are classified into types I-VII, with types I, IV, VI and VII being glandular, and types II, III and V being non-glandular [26,28]. Glandular trichomes contain various sticky or toxic chemicals that may resist herbivores [27], whereas non-glandular trichomes may function in defense by physically limiting herbivores [29].
Trichome morphology and root-hair patterning are different in tomato and Arabidopsis. Arabidopsis root-hair cells are located over two underlying cortical cells, whereas non-hair cells are positioned over a single cortical cell [14,30]. This positiondependent pattern results in rows of root-hair cells along the longitudinal root axis and has been found in Brassicaceae and other eudicot families [31][32][33][34]. This striped root-hair pattern (Type 3) is one of three types of root-hair cell distribution patterns [31][32][33][35][36][37][38]. Tomato belongs to the Type 1 root-hair pattern group, in which all the root epidermal cells have the potential to produce root-hairs. This type of pattern appears to be the most widespread in plants [39].
In this study, we have identified Arabidopsis TRY and GL3 homologous genes from tomato. Transformants expressing the tomato TRY homologous gene (SlTRY) in Arabidopsis had no trichomes and a greater number of root-hairs, a phenotype similar to that seen in over-expressors of CPC-like MYB genes. On the other hand, transformants expressing the tomato GL3 homologous gene (SlGL3) in Arabidopsis had no obvious GL3-like effects on trichome and non-hair cell differentiation. We concluded that tomato and Arabidopsis use similar transcription factors for trichome and root-hair cell differentiation and that the SlTRY-like R3 MYB may be a key common regulator of plant trichome and root-hair development.

Plant Materials and Growth Conditions
Tomato, Solanum lycopersicum L. cv. Micro-Tom, was used. Seeds were surface-sterilized with 10% commercial bleach including a detergent (Kitchen Haiter, Kao, Tokyo, Japan), for 20 min and then rinsed with sterilized water three times for 5 min each and sown on 1.5% agar plates containing 0.5xMS medium [40]. Seeded plates were kept at 4uC for 2 d and then incubated at 25uC under constant white light (50-100 mmol m 22 s 21 ) for 7 days to produce seedlings for DNA and RNA extraction. Some 7-day-old seedlings were transplanted into soil and grown in a photoperiod of 16 h light at 25uC for 4 additional weeks to produce mature plant tissues for RNA extraction.
Arabidopsis thaliana ecotype Columbia (Col-0) and, cognate cpc-2 [41] and gl3-7454 [42] mutant plants were used. Seeds were surface-sterilized, sown on 1.5% agar plates as described previously [43] and propagated to observe seedling phenotypes. Seeded plates were kept at 4uC for 2 d and then incubated at 22uC under constant white light (50-100 mmol m 22 s 21 ). For each transgenic line, at least ten individual 5-day-old seedlings were assayed for root-hair number, and at least five individual 2-weekold third leaves were assayed for trichome number.

Transgenic Plants
Gene constructs were introduced into Agrobacterium tumefaciens C58C1. Arabidopsis plants (wild-type Col-0, cpc-2, and gl3-7454) were transformed by the floral dipping method [45] and screened on 0.8% agar plates containing diluted (50% v/v) Murashige and Skoog medium and 50 mg/L (for Col-0, and gl3-7454 background) or 100 mg/L (for cpc-2 background) kanamycin sulfate. Homozygous transgenic lines were selected based on kanamycin resistance. We isolated at least twenty T1 lines for each construct and selected at least ten T2 and five T3 lines on the basis of their segregation ratios for kanamycin resistance.

Real-time Reverse Transcription PCR Analysis
Total RNA from tomato or Arabidopsis tissues was extracted with MagDEA RNA 100 (GC) (PSS, Chiba, Japan) using a Magtration System 12 GC (PSS, Chiba, Japan). To remove contaminating genomic DNA, RNA samples were treated with DNase I (Ambion, Austin, TX, USA) according to the Magtraction System protocol. Plant tissue (100 mg) was homogenized using a TissueLyser II (Qiagen, Valencia, CA, USA) with 100 ml of RLT buffer (Qiagen, Valencia, CA, USA). Sample supernatants were applied to the instrument, and RNA was eluted with 50 ml of sterile distilled water.
First-strand cDNA was synthesized from 1 mg total RNA in a 20 ml reaction mixture using the Prime Script RT Master Mix (Perfect Real Time) (Takara, Tokyo, Japan). Real-time PCR was performed using a Chromo4 Real-Time IQ5 PCR Detection System (Bio-Rad, Hercules, CA, USA) with SYBR Premix Ex Taq II (Takara, Tokyo, Japan). PCR amplification employed a 30 s denaturing step at 95uC, followed by 5 s at 95uC and 30 s at 60uC with 40 cycles for SlTRY, SlGL3, LeActin, GL2, GFP and ACT2. Real-time PCR was used to analyze the mRNA expression level of each transcript encoding SlTRY and SlGL3 in tomato, and GL2 and GFP in Arabidopsis transformants. The relative expression of each transcript was calculated by the DDCT method [46]. The expression levels of SlTRY and SlGL3 were estimated after being normalized to the endogenous control gene LeActin (TC116322). The expression levels of GL2 and GFP were estimated after being normalized to the endogenous control gene ACT2 (AB026654). The primers were: RTSlTRY-F and RTSlTRY-R for SlTRY; RTSlGL3-F and RTSlGL3-R for SlGL3; LeActin-F and LeActin-R for LeActin [47]; GL2-F and GL2-R for GL2 [48]; GFP-F and GFP-R for GFP; and Act2-F and Act2-R for ACT2 [49].

Light Microscopy
To observe trichomes, images were recorded with a VC4500 3D digital fine microscope (Omron, Kyoto, Japan) or a digital microscope (VH-8000; Keyence, Osaka, Japan). At least five 2week-old third true leaves were analyzed for trichome number for each transgenic line. Root phenotypes were observed using an Olympus Previs AX70 microscope and an Olympus SZH binocular microscope. For each transgenic line, at least ten individual 5-day-old seedlings were analyzed for root-hair number.
To provide a framework for examining R3 MYB transcription factor evolution, we estimated the phylogeny of CPC-like R3 MYB transcription factor proteins from Arabidopsis (CPC, TRY, ETC1, ETC2, ETC3/CPL3, TCL1 and TCL2/CPL4), rice (Oryza sativa) (Os1g43180 and Os1g43230) and tomato (SlTRY) based on their deduced amino acid sequences ( Figure 1B). SlTRY was more closely related to TRY than CPC, which belongs to a cluster that includes TCL1, TCL2/ CPL4, ETC2, and TRY ( Figure 1B). ETC1, ETC3/CPL3 and CPC belong to another cluster branching from the TRY subgroup. Consistent with previous reports, phylogenic analyses using the entire amino acid sequence showed that the CPC-like MYB family can be divided into two groups: TRY, ETC2, TCL1 and TCL2/CPL4 in one group and CPC, ETC1 and ETC3 in the other [6][7][8][9]. As previously described, the two rice orthologs, Os01g43180 and Os01g43230, form a distinct clade from the Arabidopsis CPC-like R3 MYB family ( Figure 1B) [7].
The SlGL3 encoded protein is closely related to the Arabidopsis bHLH transcription factor proteins encoded by GL3 and EGL3 (Figure 2A). Alignment of the amino acid sequences showed that the full length SlGL3 protein shares 45% amino acid identity with GL3, and 46% with EGL3. The bHLH motif of SlGL3 shares 46% amino acid identity with that of GL3, and 50% with that of EGL3. Based on the amino acid sequences of the bHLH regions, the Arabidopsis bHLH transcription factors are classified into 12 groups (I-XII) [50]. Group III contains 6 subgroups (IIIa-f), and GL3 and EGL3 belong to the IIIf subgroup [50].

Expression Patterns of the SlTRY and SlGL3 Genes in Tomato
Expression of SlTRY and SlGL3 was examined in tomato tissues using real-time reverse transcription PCR. SlTRY was strongly expressed in stem and cotyledons of 7-day-old seedlings ( Figure 3A). The relative expression level of SlTRY in cotyledon tissues was approximately 5 times greater than that in seedling root tissues ( Figure 3A). The strongest expression of SlGL3 was observed in 5-week-old plant leaves ( Figure 3B). The relative expression level of SlGL3 in true leaf tissues was approximately 6 times greater than that in seedling root tissues ( Figure 3B). Both SlTRY and SlGL3 were more strongly expressed in aerial tissues (including stem, cotyledon, leaf, bud and flower) than in roots. These results suggest that both SlTRY and SlGL3 act in both shoot and root tissues and might have relatively strong functions in the aerial parts of plants.     [1,3,4,7]. Consistent with these previous observations, all homozygous CPC::SlTRY transgenic lines (#1-#5) have the no-trichome phenotype, although wild-type Col-0 produces approximately 50 trichomes on the adaxial surface of the third true leaf (Figure 4A, E, G). To see a SlTRY function more clearly, we introduced CPC::SlTRY into the cpc-2 mutant. As previously reported, the cpc-2 mutant has a greater number of trichomes than wild-type [7] (Figure 4B, F). All homozygous CPC::SlTRY in cpc-2 transgenic lines (#1-#6) show the no-trichome phenotype ( Figure 4A, G) as observed in CPC::SlTRY in a wild-type background ( Figure 4A). These results indicate that the tomato SlTRY protein has a function similar to the Arabidopsis CPC-like MYB proteins in regulating trichome development.
On the other hand, all homozygous CPC::SlTRY transgenic lines (#1-#6) produced greater numbers of root-hairs compared with wild-type Col-0, which produces approximately 40 root-hairs per mm ( Figure 4C, I, K). This result is similar to the root-hair numbers of CPC-like MYB over-expressors [1,3,4,7]. Homozygous CPC::SLTRY in cpc-2 transgenic lines (#1-#6) also showed a greater number of root-hairs compared with wild-type and cpc-2, a mutant that produces a decreased number of root-hairs. These results are similar to the previously reported results using CPC::CPC in cpc-2 or GL2::CPC in cpc-1 transgenic plants [53,54] ( Figure 4D, J, L). These results indicate that the tomato protein SlTRY has a function similar to Arabidopsis CPC-like MYB proteins in regulating root-hair development. Together, our results show that SlTRY functions similar to CPC-like MYBs both in trichome and root-hair formation in Arabidopsis.

SIGL3 does not Function in Trichome and Root-hair Development in Arabidopsis
To see if SlGL3 is functionally similar to Arabidopsis GL3 or EGL3, we introduced SlGL3 into Arabidopsis wild-type Col-0 under the control of the GL3 promoter (GL3::SlGL3). The GL3 and EGL3 genes are thought to be involved in trichome and roothair formation. GL3 or EGL3 over-expressors produce greater numbers of trichomes and reduced numbers of root-hairs [12,13]. Most of homozygous GL3::SlGL3 transgenic lines produced the similar number of trichomes as that of wild-type ( Figure 5A, F). Unlike the previous observation from complementation analysis of the gl3-1 mutant by GL3::GL3 [12], homozygous GL3::SlGL3 in gl3-7454 transgenic plants did not have an increased number of trichomes compared with the gl3-7454 mutant (Figure 5B, E, G). On the contrary, one of the GL3::SlGL3 in gl3-7454 transgenic lines (#5) had significantly fewer trichomes in comparison with the gl3-7454 mutant ( Figure 5B). These results suggest that tomato SlGL3 may not have a function same to Arabidopsis GL3.
In addition to the effects on trichome number, GL3 is known to affect the trichome branching phenotype [12]. As observed in the gl3-1 mutant (Ler background) [12], trichomes of the gl3-7454 mutant (Col-0 background) had fewer branches than wild-type Col-0 ( Table 2). Introduction of the GL3::SlGL3 gene did not rescue the decreased branch number phenotype of the gl3-7454 mutant trichomes (Table 2). This result suggests that the SlGL3 gene does not have a function similar to GL3 in the induction of trichome branching.

Expression of the GL2 Gene in SlTRY expressing Plants
To determine whether CPC::SlTRY functions ( Figure 4) were due to epistatic effects of SlTRY on GL2 activity, we carried out real-time reverse transcription PCR analyses using GL2 primers ( Figure 6). The GL2 gene is thought to act downstream of the MYB-bHLH transcriptional complex to promote trichome formation and inhibit root-hair formation [1,[14][15][16][17]. Consistent with the CPC::SlTRY transgene phenotype ( Figure 4A), GL2 expression was strongly repressed in all CPC::SlTRY transgenic lines ( Figure 6A). In the CPC::SlTRY in cpc-2 transgenic lines, GL2 expression was also strongly repressed compared with wild-type and cpc-2 as was the case in the wild-type background ( Figure 6B). To compare gene expression levels of the introduced gene among transgenic lines, we checked GFP expression since GFP was fused to the C-terminal region of SlTRY ( Figure S1A). Although the transgene expression levels varied depending on the lines ( Figure  S1A), expression in all lines was strong enough to repress GL2 expression.

Expression of the GL2 Gene in SlGL3 expressing Plants
To determine whether CPC::SlGL3 functions ( Figure 5) were due to epistatic effects of SlGL3 on GL2 activity, we also carried out real-time reverse transcription PCR analyses using GL2 primers (Figure 7). Inconsistent with the GL3::SlGL3 transgene phenotypes ( Figure 5A), significant GL2 expression changes were observed only in GL3::SlGL3 line #1 compared with wild-type Col-0 ( Figure 7A). Apparently, SlGL3 does not have a remarkable effect on GL2 expression. In GL3::SlGL3 in gl3-7454 transgenic plants, a significant increase in GL2 expression was observed in lines #2 and #3 compared with that in the gl3-7454 mutant; however, these GL2 expression levels did not reach similar expression levels of GL2 in wild-type Col-0 ( Figure 7B). A significant decrease in GL2 expression was observed in line #5 compared with that in the gl3-7454 mutant ( Figure 7B). Thus, we checked GFP expression that should reflect the introduced SlGL3 expression levels since the SlGL3 construct was fused to GFP ( Figure S1B). The GFP expressions varied greatly among GL3::SlGL3 in gl3-7454 transgenic lines ( Figure S1B). In addition, the relative expression levels of GFP in GL3::SlGL3 in gl3-7454 lines were lower than that in the GL3::SlGL3 lines ( Figure S1B). These results suggest that SlGL3 expression was unstable in the gl3-7454 mutant background.

Discussion
In this study, we identified tomato SlTRY and SlGL3 genes that were orthologous to the Arabidopsis TRY and GL3 genes, respectively. Recently, a high-quality genome sequence of tomato was released by the Tomato Genome Consortium [55]. Since tomatoes are very distantly related to Arabidopsis evolutionarily, these sequence data may offer important information about plant evolution in the future. The functional analyses of the SlTRY and SlGL3 genes in this study provide insights into tomato trichome and root-hair evolution. Branching of the SlTRY and CPC clusters from a common trunk in the phylogenic tree suggests that the evolution of tomato and Arabidopsis CPC-like R3 MYB genes began with duplication of a single common ancestor after divergence from rice ( Figure 1B). Based on the functions of known members of the Arabidopsis bHLH transcription factor family, it was hypothesized that different members participate in distinct developmental processes [50]. Among the genes, members of the IIIf subgroup, including AtMYC1, TT8, GL3 and EGL3 function in trichome and root-hair development, flavonoid/ anthocyanin metabolism, and/or mucilage biosynthesis [13,19,20,[56][57][58][59]. Phylogenic analyses predicted that tomato SlGL3 evolved from a common ancestor to Arabidopsis GL3 and EGL3 after divergence of the IIIf subgroup from other bHLH subgroups ( Figure 2B). Thus, we expected that SlTRY and SlGL3  would have similar functions to TRY/CPC and GL3/EGL3 in trichome and root-hair differentiation. Both SlTRY and SlGL3 were shown to be expressed in all the tissues examined, especially in the aerial parts (Figure 3), suggesting that these genes function in nearly the entire tomato plant body.
In our experiment, SlTRY was demonstrated to function quite similarly to the CPC-like MYB transcription factors in Arabidopsis trichome and root-hair formation ( Figure 4). Both CPC::SlTRY and CPC::SlTRY in cpc-2 transgenic plants showed the no-trichome and increased root-hair phenotypes ( Figure 4). Previously, an R3-type MYB of TRY and an R2R3-type MYB of GL1 were reported to compete for a GL3 binding site to form different types of MYB-bHLH complexes involved in Arabidopsis trichome differentiation [60]. TRY prevents the interaction between GL1 and GL3 [19], and CPC physically interacts with GL3/EGL3 [13], suggesting a competition model for CPC and WER [13,16]. The CPC protein was proposed to disrupt the WER-GL3/EGL3 protein complex by competitive binding with WER, leading to repression of GL2 expression [18,53,54]. In this study, we showed that SlTRY also repressed GL2 expression ( Figure 6). These results suggest that the SlTRY protein may also disrupt the MYB-bHLH complex of GL1/WER-GL3/EGL3, leading to repression of GL2 expression.
In contrast to SlTRY, SlGL3 did not show clear GL3/EGL3like functions for trichome and root-hair differentiation in Arabidopsis ( Figure 5). Overexpression of GL3 and/or EGL3 induced a notable increase in trichome number and a decrease in root-hair number in Arabidopsis [12,13]. However, in our experiment, only two of five and three of five GL3::SlGL3 transgenic lines showed a significant increase in trichome number and a significant decrease in root-hair number compared with wild-type, respectively ( Figure 5A and B). It is thus possible that tomato SlGL3 has an evolutionarily conserved, highly homologous amino acid sequence and only a partially similar function to Arabidopsis GL3/EGL3. Since trichome and root-hair structure differs between Arabidopsis and tomato, the SlGL3 gene may have acquired another function from GL3/EGL3 during evolution. The functional difference between SlGL3 and GL3/EGL3 may be derived from the relatively low amino acid homology region in the bHLH motifs (Figure 2A). Only one of five GL3::SlGL3 transgenic lines showed a significant increase in the GL2 expression level compared with wild-type Col-0 ( Figure 7A). Thus, two possibilities exist. First, a low affinity of SlGL3 protein to WER/GL1 proteins may result in the formation of an incomplete MYB-bHLH protein complexthat cannot activate GL2 expression. Recently, Zhao et al. reported that a single amino acid substitution in another GL3 homologous gene AtMYC1, leads to trichome and root-hair patterning defects by abolishing its interaction with partner proteins in Arabidopsis [61]. Arginine (R173) in the AtMYC1 protein is an essential amino acid residue for interaction with MYB proteins for proper functions [61]. We confirmed that there is a conserved Arg in the SlGL3 protein as in the GL3, EGL3 and AtMYC1 proteins (Figure 2A). Thus, some amino acid substitution other than Arg may contribute to the functional difference between SlGL3 and GL3/EGL3. Second, SlGL3 might have lost either the DNA binding ability to the GL2 promoter region or the ability to activate the GL2 promoter. For example, we previously reported that WER loses its DNA binding ability by at least two amino acid substitutions [53]. Tomato SlGL3 may have lost its DNA binding ability to the GL2 promoter region during evolution.  In order to compare our results in the same Col-0 background, we used the gl3-7454 mutant for the complementary experiment.
The gl3-7454 mutant shows only a mild phenotype compared with the gl3-1 mutant (Ler:Landsberg erecta background) [12]. The gl3-7454 mutant shows no significant difference in trichome number or in root-hair number compared with wild-type Col-0 ( Figure 5B and D). Unexpectedly, one of five GL3::SlGL3 in gl3-7454 transgenic lines showed significant decreases in trichome number compared with gl3-7454 ( Figure 5B), and three of five GL3::SlGL3 in gl3-7454 transgenic lines showed significant increases in roothair number compared with gl3-7454 ( Figure 5D). Consistent with these unexpected phenotypes of GL3::SlGL3 in gl3-7454 transgenic plants, the relative expression levels of GL2 varied ( Figure 7B). Two of five GL3::SlGL3 in gl3-7454 transgenic lines showed significantly higher GL2 expression levels compared with that in gl3-7454, but the level did not reach that in the wild-type Col-0 ( Figure 7B). One of five GL3::SlGL3 in gl3-7454 transgenic lines showed significantly lower GL2 expression levels compared with that in the gl3-7454 mutant ( Figure 7B). As checked by fusion of SlGL3 to GFP, expression of introduced SlGL3 was unstable and fluctuated depending on the lines ( Figure S1). In addition, SlGL3 did not rescue the reduced number of trichome branches phenotype of gl3-7454 (Table 2). These data strongly suggest the functional divergence between tomato SlGL3 and Arabidopsis GL3/EGL3. There are 158 bHLH genes in Arabidopsis [52,62]. Tomato should have the similar or more number of the bHLH genes when the full annotations of tomato genes are determined. We concluded that there is other GL3 ortholog(s) in the unannotated tomato genomes or tomato uses other pathways to regulate the epidermal cell differentiation.Additional investigations to further determine the functions of R3-MYB and bHLH in trichome and root-hair differentiation in tomato are necessary.