A Single Amino Acid Substitution in IIIf Subfamily of Basic Helix-Loop-Helix Transcription Factor AtMYC1 Leads to Trichome and Root Hair Patterning Defects by Abolishing Its Interaction with Partner Proteins in Arabidopsis*

Background: IIIf subfamily of bHLH members GL3/EGL3 involved in epidermal cell fate determination through interacting with MYBs/TTG1. Results: A substitution of R173H in the AtMYC1 homolog of GL3/EGL3 abolished its function in trichome and non-hair cell fate control. Conclusion: The functionally conserved Arg residue in AtMYC1/GL3/EGL3 was crucial for protein interaction and proper biological function. Significance: A novel critical and functionally conserved site in IIIf bHLH proteins was identified. Plant trichomes and root hairs are powerful models for the study of cell fate determination. In Arabidopsis thaliana, trichome and root hair initiation requires a combination of three groups of proteins, including the WD40 repeat protein TRANSPARENT TESTA GLABRA1 (TTG1), R2R3 repeat MYB protein GLABRA1 (GL1), or WEREWOLF (WER) and the IIIf subfamily of basic helix-loop-helix (bHLH) protein GLABRA3 (GL3) or ENHANCER OF GLABRA3 (EGL3). The bHLH component acts as a docking site for TTG1 and MYB proteins. Here, we isolated a mutant showing defects in trichome and root hair patterning that carried a point mutation (R173H) in AtMYC1 that encodes the fourth member of IIIf bHLH family protein. Genetic analysis revealed partial redundant yet distinct function between AtMYC1 and GL3/EGL3. GLABRA2 (GL2), an important transcription factor involved in trichome and root hair control, was down-regulated in Atmyc1 plants, suggesting the requirement of AtMYC1 for appropriate GL2 transcription. Like its homologs, AtMYC1 formed a complex with TTG1 and MYB proteins but did not dimerized. In addition, the interaction of AtMYC1 with MYB proteins and TTG1 was abrogated by the R173H substitution in Atmyc1-1. We found that this amino acid (Arg) is conserved in the AtMYC1 homologs GL3/EGL3 and that it is essential for their interaction with MYB proteins and for their proper functions. Our findings indicate that AtMYC1 is an important regulator of trichome and root hair initiation, and they reveal a novel amino acid necessary for protein-protein interactions and gene function in IIIf subfamily bHLH transcription factors.

Plant trichomes and root hairs are powerful models for the study of cell fate determination. In Arabidopsis thaliana, trichome and root hair initiation requires a combination of three groups of proteins, including the WD40 repeat protein TRANSPARENT TESTA GLABRA1 (TTG1), R2R3 repeat MYB protein GLABRA1 (GL1), or WEREWOLF (WER) and the IIIf subfamily of basic helix-loop-helix (bHLH) protein GLABRA3 (GL3) or ENHANCER OF GLABRA3 (EGL3). The bHLH component acts as a docking site for TTG1 and MYB proteins. Here, we isolated a mutant showing defects in trichome and root hair patterning that carried a point mutation (R173H) in AtMYC1 that encodes the fourth member of IIIf bHLH family protein. Genetic analysis revealed partial redundant yet distinct function between AtMYC1 and GL3/EGL3. GLABRA2 (GL2), an important transcription factor involved in trichome and root hair control, was down-regulated in Atmyc1 plants, suggesting the requirement of AtMYC1 for appropriate GL2 transcription. Like its homologs, AtMYC1 formed a complex with TTG1 and MYB proteins but did not dimerized. In addition, the interaction of AtMYC1 with MYB proteins and TTG1 was abrogated by the R173H substitution in Atmyc1-1.

We found that this amino acid (Arg) is conserved in the AtMYC1 homologs GL3/EGL3 and that it is essential for their interaction with MYB proteins and for their proper functions. Our findings indicate that AtMYC1 is an important regulator of trichome and root hair initiation, and they reveal a novel amino acid necessary for protein-protein interactions and gene function in IIIf subfamily bHLH transcription factors.
Plant trichomes and root hairs are derived from epidermal cells in above-and under-ground tissues, respectively. Whereas trichomes protect plants from UV radiation, wind, frost, and insects (1,2), root hairs enhance water and nutrient absorption, help anchor plants to the soil, and facilitate communication with biotic and abiotic factors in the soil (3). For biologists, the differentiation and morphogenesis of trichomes and root hairs from initially equivalent cells have provided an important insight into cell fate determination (2)(3)(4). In Arabidopsis roots, epidermal cells differentiate in a position-dependent manner (5)(6)(7)(8)(9). Epidermal cells in contact with two cortical cells develop into hair cells, whereas epidermal cells in contact with one cortical cell become non-hair cells (2,3).
Unlike their root counterparts, trichomes are distributed on the rosette leaves, the stem, cauline leaves, and sepals without any spatial correlation to morphological landmarks (1,4). However, trichomes are regularly separated and rarely clustered in the wild type, suggesting the existence of a patterning mechanism (10,11).
Years of genetic and molecular studies have revealed that the cell fate determination of trichomes and root hair cells is controlled by similar molecular mechanisms (2,3). Three factors, the WD40 repeat protein TRANSPARENT TESTA GLABRA1 (TTG1), R2R3 repeat MYB protein GLABRA1 (GL1; for the trichomes), or WEREWOLF (WER; for the root non-hair cells) and the basic helix-loop-helix (bHLH) 2 protein GLABRA3 (GL3) or ENHANCER OF GLABRA3 (EGL3) constitute an active R2R3 MYB-bHLH-WD40 complex that initiates trichome and non-hair cell differentiation (1,2,4,(12)(13)(14)(15). GLABRA2 (GL2), a homeodomain transcription factor, is the immediate downstream target of the active complex and is involved in morphological development and maturation of the trichome and non-hair cell (16 -20). Mutations in these factors result in fewer or even no trichomes production and/or ectopic root hairs formation.
It has been shown that bHLH proteins serve as a docking site in multiprotein interactions (36). There are roughly 133 members of the bHLH family, making it one of the largest transcription factor superfamilies in Arabidopsis (37,38). A comprehensive analysis classified these genes into 12 subfamilies. Based on the functions of known members of this transcription factor family, it was speculated that different members participate in distinct plant developmental processes (37). Among them, members of the IIIf subgroup, including GL3, EGL3, and TT8, function in trichome initiation, root hair patterning, flavonoid/ anthocyanin metabolism, and/or mucilage biosynthesis (12-14, 35, 39 -41). This subgroup is homologous to the bHLH transcription factors R and B in Zea mays, which also function together with the MYB proteins C1 and Pl to control anthocyanin production (37,(42)(43). AtMYC1, the fourth member of the R/B-like IIIf subfamily of bHLH transcription factors in Arabidopsis, was first cloned by Urao et al. in 1996 (44); subsequently it was characterized as the direct target as GL3 by Morohashi and Grotewold (20), and it was shown to interact with GL1 and WER in yeast (45), and after we finished this manuscript, it was reported that AtMYC1 is an important source of variation for trichome cell fate determination in different ecotypes of Arabidopsis thaliana (46).
Here, we characterized the function of AtMYC1 in root hair development in addition to in trichome cell fate determination through the isolation of a mutant carrying a point mutation in AtMYC1. We found that the function of AtMYC1 is partially redundant yet distinct with that of GL3/EGL3 in trichome and root hair development, probably through the regulation of GL2 expression. Furthermore, we identified an amino acid residue that is functionally conserved among the R/B-like IIIf subfamily of bHLH transcription factors and which is necessary for the interaction between bHLH and MYB proteins in trichome and root hair patterning.

EXPERIMENTAL PROCEDURES
Plant Materials and Growth Conditions-The A. thaliana stocks described in this work were of the Columbia (Col) ecotype. Atmyc1-2, Atmyc1-3, and gl3-11 correspond to SALK_056899, SAIL_227_H01, and SALK_118201, respectively. Seeds were surface-sterilized in 75% ethanol, washed with sterile water, kept at 4°C in a chamber for 2 days, and then planted on Murashige and Skoog medium. After 8 days, the seedlings were transferred to soil and grown in a growth chamber under long-day conditions (16 h of light/8 h of dark) at 65% relative humidity.
Positional Cloning-The Atmyc1-1 mutant (Col background) was crossed with wild-type Landsberg erecta (Ler) plants. The F2 population was used for mapping. Simple sequence length polymorphism (SSLP), cleaved amplified polymorphic sequence (CAPS), and derived CAPS (dCAPS) markers were designed according to differences announced publicly on the TAIR website (47,48).
Microscopy-Root hair cell number and cell type pattern analysis were carried on with toluidine blue-stained 4-day-old seedling roots, viewed with differential interference contrast optics. At least 20 seedling roots were analyzed for each strain. Epidermis with visible protrusion was taken as root hair cells (12).
RNA Extraction and Real-time PCR-Total RNA for quantitative real-time PCR was prepared using TRI Reagent Solution (Ambion) according to the manufacturer's handbook. After digestion with RNase-free DNase (Promega) to eliminate contaminating DNA, 3 mg of total RNA were used for reverse transcription (Fermentas). Real-time PCR was carried out using Takara SYBR Premix Ex Taq in a 7500 real-time PCR instrument (Applied Biosystems).
Plasmid Construction and Plant Transformation-To construct pAtMYC1::AtMYC1 for our complementation analysis, a 2581-bp 5Ј-regulatory sequence together with the coding sequence of AtMYC1 was inserted into the binary vector pCAMBIA1300 and introduced into Atmyc1-1 by Agrobacterium tumefaciens-mediated floral transformation (49). To construct pAtMYC1::GUS, the above promoter was inserted upstream of GUS in the pCAMBIA 1300, and the construct was transformed into wild-type plants.
GUS Staining-GUS staining was performed as described by Cao et al. (50). Trichome staining in mature leaves was done using T1 plants; all other tissues were taken from T3 plants.
Scanning Electron Microscopy-Fresh 10-day-old seedlings were fixed in 50% ethanol, 5% acetic acid, and 3.7% formaldehyde overnight at 4°C then dehydrated in an ethanol series (once in 30, 50, 70, and 95% and twice in 100%). The samples were then freeze-dried overnight in tertiary butyl alcohol. The dried seedlings were mounted on scanning electron microscopy stubs with double-sided mounting tape, then coated with gold and examined by scanning electron microscopy (Hitachi S-2460, Tokyo, Japan).
Protoplast Transient Expression and Bimolecular Fluorescence Complement (BiFC) Assays-To generate the necessary constructs, the full-length coding sequences of the genes were inserted into pUC-SPYNE or pUC-SPYCE (51) using XbaI and BamHI. The plasmids were extracted and purified using a Plasmid Maxprep kit (Vigorous) according to the manufacturer's protocol. Protoplasts were produced from 3-4-week-old Arabidopsis leaves grown under short-day conditions (8 h of light/16 h of dark) and transformed with purified plasmid using polyethylene glycol (52). After 12-18 h of incubation at 22°C in the dark, the transformed protoplasts were checked for YFP fluorescence by confocal laser scanning microscopy (LSM 510 META; Zeiss). To confirm that AtMYC1 was expressed normally, protoplasts transformed with the corresponding plasmids were collected, and their proteins were probed with antic-Myc antibodies by Western blotting.
Yeast Two-hybrid Interaction Assay-To generate the plasmids for this assay, the coding sequences of the tested genes were cloned into pGADT7 and pGBKT7. After confirmation by sequencing, the plasmids were transformed into yeast strain AH109. Yeast transformation, growth on S.D. medium, yeast protein extraction by the urea/SDS method, and quantitative ␤-galactosidase assays were carried out according to the Clontech Yeast Protocols Handbook.

RESULTS
Isolation of Arabidopsis Mutant rtd1-1 (reduced trichomes density)-As trichomes are an easily accessible model for investigating cell fate determination (1, 3, 4), we focused on mutants with trichome developmental defects. In a screen of a T-DNA insertion library, a mutant named rtd1-1 with fewer rosette leaf trichomes than in wild type was identified ( Fig. 1A; Table 1). The defect in trichome number remained obvious when the plants bolted; few trichomes were found at the base of the main stem (Fig. 1B). Besides the trichome defect, rtd1-1 had a moderate increase of root hairs (about 16.4% more than the wild type), due to the change of about 30.3% non-hair to hair cells ( Fig. 1C; Table 2). The flanking sequence of rtd1-1 was obtained by TAIL-PCR; however, it did not link to the mutant phenotype. So it suggested that there must be another mutation.
Phenotypes of rtd1-1 Are Attributed to Point Mutation in bHLH Gene AtMYC1-To determine the genetic background of rtd1-1, the mutant (Col background) was crossed with wildtype Col plants. Among 96 B3 lines (i.e. third backcross generation), 28 lines (15 plants were tested per line) showed reduced trichome density defect. Therefore, we concluded that the defect was caused by a recessive mutation.
To identify the mutated gene responsible for these defects, the rtd1-1 was crossed with wild-type Ler plants for positional cloning. Fine mapping with 183 plants pinpointed the mutation to a 26-kb region between BAC clones F5I10 and F6N23 on chromosome 4. Sequencing revealed a point mutation (G518A) in the open reading frame of AtMYC1 (At4g00480), resulting in an R173H substitution in the protein ( Fig. 2A). We, therefore, renamed the mutant Atmyc1-1.
To confirm that the phenotypes we observed were caused by the mutation in AtMYC1, two additional T-DNA alleles of AtMYC1, Atmyc1-2 and -3, were characterized. The T-DNA insertion in Atmyc1-2 was in exon 3; no full-length transcript was detected in homozygous plants. The insertion in Atmyc1-3 was in the 3Ј-UTR; semiquantitative and quantitative real-time PCR showed that it was a knock-down allele (Fig. 3, A and B).  Both alleles showed similar defects to Atmyc1-1, including decreased trichomes (Fig. 3, C and D; Table 1) and increased root hairs ( Fig. 3E; Table 2). Furthermore, all these defects were successfully rescued by introducing the wild-type AtMYC1 coding sequence driven by its native promoter into Atmyc1-1. Among 54 independent lines, 36 had similar numbers of trichomes as wild type. Re-1-1 and -2-14 are two representative recovered lines (Fig. 3, C-E; Tables 1 and 2). Therefore, we concluded that the defects in the mutants were caused by a loss of function of AtMYC1. AtMYC1 Is Required for Full GL2 Expression-We next focused on the epidermal cell determination defects in Atmyc1 mutants. To further explore the role of AtMYC1 in trichome and root hair determination, we examined the expression level of GL2 in Atmyc1, GL2 is a crucial transcription factor in trichome and non-hair cell development (16 -20), and it has been shown to be a direct target of the GL1-GL3-TTG1 complex (13,53). We, therefore, examined the expression level of GL2 in Atmyc1 by quantitative real-time PCR. As expected, GL2 was down-regulated in all three alleles of Atmyc1 (Fig. 4A) in accordance with the reduction in trichomes and increase in root hairs in Atmyc1 (Fig. 3C-E; Tables 1 and 2). This finding demonstrates that the lesion in AtMYC1 causes the incomplete GL2 expression and AtMYC1 is a positive regulator of GL2.
Genetic Interaction between AtMYC1 and GL3/EGL3-To assess the genetic relationship between AtMYC1 and its homolog GL3/EGL3, Atmyc1-1 was crossed with the T-DNA insertion mutant gl3-11 and egl3-7 (Col background) (54). gl3-11 TABLE 2 Effect of AtMCY1, GL3, and EGL3 mutation on cell type pattern in the root epidermis Data, including S.D., were obtained from at least 20 four-day-old seedling roots from each strain. In all strains ϳ40% of epidermal cells in the Hair cell position.

Genotype Hair cells in epidermis
Hair cell position Non-hair cell position exhibited a moderate decrease in trichome number and an increase in hair cell production, whereas the egl3-7 largely behaved like the wild type (Fig. 4, B and C; Table 2). The gl3 egl3 double mutant produces both glabrous aerial organs and extremely hairy underground organs (14, 54, Fig. 4, B-D; Table 2), which suggested the significant functional redundancy between GL3 and EGL3. For the leaf trichome development, although the Atmyc1-1 gl3-11 and Atmyc1-1 egl3-7 double mutants showed a reduction in the trichomes number on the first and second pair of leaves than Atmyc1-1 (t test: p ϭ 0.048 and 0.001, respectively; Fig. 4, B and C), the degree of reduction was much less than that of the gl3 egl3 double mutant. However, Atmyc1-1 gl3-11 show glabrous inflorescence stem (Fig. 4D) and hairy root defects with 99.6% hair cells in root epidermis because of the misspecification of nonhair to hair cells (Table 2). These results suggest that AtMYC1 and GL3 function redundantly in stem trichome and root hair cell fate control and only partially redundant in leaf trichome control.

Hair cells Non-hair cells Hair cells Non-hair cells
Furthermore, we analyzed the Atmyc1-1 egl3-7 double mutant, which had a slight decrease of leaf trichome (Fig. 4, B and C), moderate increase of root hair number (Table 2), and almost no change of stem trichomes compared with the parental line Atmyc1-1 (Fig. 4D). These results indicate that the degree of redundancy between AtMYC1 and EGL3 is even less than that of AtMYC1 and GL3.
To further analyze the functional similarity and divergence of AtMYC1 and its homologs, we designed a promoter exchange assay. A 2.5-and 3-kb intergenic region upstream of ATG of GL3 and EGL3, respectively, were fused before the coding region of AtMYC1, and a 2.58-kb intergenic region upstream of start codon of AtMYC1 was fused ahead the open reading region of GL3 or EGL3. The constructs pAtMYC1::GL3 and pAtMYC1::EGL3 were introduced to Atmyc1-2, whereas pGL3::AtMYC1 and pEGL3::AtMYC1 were transformed into gl3-11 egl3-7 double mutant. We observed that both GL3 and EGL3 driven by AtMYC1 promoter successfully rescued the reduced trichomes defect of Atmyc1-2 (Fig. 4E, lower panel); however, AtMYC1 driven by pGL3 and pEGL3 were not able to complement the gl3 egl3 double mutant phenotype at all (Fig.  4E, lower panel). As controls, pAtMYC1::AtMYC1, pGL3::GL3, and pEGL3::EGL3 were able to recover the corresponding mutant trichome defects (Fig. 4E, top panel), and the pGL3/EGL3::AtMYC1 constructs were functional because they can rescue the Atmyc1-2 phenotype (Fig. 4E, lower panel). Similar results were obtained for the root hair cell fate determination (Table 3). These results indicate that GL3/EGL3 can exert the activity of AtMYC1 in leaf trichome specification, whereas AtMYC1 cannot substitute either GL3 or EGL3 in leaf trichome and root hair development in planta. These results together demonstrate that the function of AtMYC1 and GL3/EGL3 is both similar/redundant and divergent.
Expression Pattern of AtMYC1-To better define the expression pattern of AtMYC1 during root hair and trichome development, we created pAtMYC1::GUS construct and transformed it into the wild-type. We analyzed the transgenic plants for GUS expression pattern. Consistent with the defects in root hair spacing seen in Atmyc1, AtMYC1 was expressed in alternate cell files in the roots (Fig. 5A). Enlarged views of the meristematic and root hair zones revealed that the deeply stained cells were hair cell files. Because the stained cells files in the meristematic zone were located in a cleft between two underlying cortical cells and had a higher division rate than the neighboring unstained ones, they were in the same cell files with GUS-staining cells, producing root hairs in the mature region (Fig. 5, B and C). This expression pattern of AtMYC1 is similar to that of its homologs GL3 and EGL3 (55). In the rosette leaf, pAtMYC1::GUS expression was detected in the basal cells of trichomes (Fig. 5, D and E), in contrast to its homologs, which had higher expression in trichomes (13). Finally, we detected

AtMYC1 Mediates Trichome and Root Hair Patterning
strong GUS activity at the base of the inflorescence stem, consistent with the reduced number of trichomes in this region (Fig. 5F). In conclusion, the pattern of AtMYC1 expression shown using GUS is largely consistent with its biological functions.
AtMYC1 Functions through Its Interaction with MYB Proteins and TTG1-It was previously reported that the bHLH proteins GL3 and EGL3 interact with GL1/WER and TTG1 to promote trichome and non-hair cell identity (13-15, 56). And it was reported that AtMYC1 also interacted with these proteins

Effect of promoter exchange on cell type pattern in the root epidermis
Data, including S.D. were obtained from at least 20 four-day-old seedlings from each strain. In all strains, ϳ40% of epidermal cells in the Hair cell position.

Genotype Hair cells in epidermis
Hair cell position Non-hair cell position   APRIL 20, 2012 • VOLUME 287 • NUMBER 17

JOURNAL OF BIOLOGICAL CHEMISTRY 14115
in vitro (45). We confirmed the interaction in a yeast two-hybrid assay (Fig. 6, A and B). We next conducted an in vivo BiFC assay to confirm these results in transiently transfected Arabidopsis mesophyll cell protoplasts. For AtMYC1-YFP N /GL1-YFP C , AtMYC1-YFP N /WER-YFP C , and AtMYC1-YFP N / TRY-YFP C , about 5, 16, and 5% of the protoplasts produced strong YFP fluorescence (Fig. 6C); in comparison, no interaction signal was observed in the negative controls (AtMYC1-1-YFP N /Target protein-YFP C ). Therefore, we concluded that AtMYC1 interacts with the MYB proteins and TTG1. Because GL3 and EGL3 have the ability to form homo-and heterodimers in yeast (14, 15), we analyzed the dimerization of AtMYC1. GL3 dimerization was detected; however, AtMYC1 formed neither homo-nor heterodimers (Fig. 6D). Thus, similar to its homologs, AtMYC1 may be involved in an interaction with MYB proteins and TTG1; however, it does not dimerize. The inability of AtMYC1 to form dimer may be the reason for its failure to rescue gl3 egl3 double mutant.
Arg-173 in AtMYC1 Is Critical for Its Interaction with MYB Proteins and TTG1-Because the only mutation in Atmyc1-1 was R173H, we sought to determine the reason for its failure in function. We proposed several hypotheses; the mutation might alter the stability of AtMYC1 mRNA, the subcellular localization of AtMYC1-1 might be altered, or its ability to interact with FIGURE 6. Physical interaction between AtMYC1 and MYB proteins, TTG1. A, physical interaction between AtMYC1 and MYB proteins is shown by a yeast two-hybrid assay. Positive control: pGADT7-T/pGBKT7-53; negative control: pGADT7-T/pGBKT7-Lam. The interaction between AtMYC1/AtMYC1-1 and MYB proteins was assessed by growth on SD/ϪAde/ϪHis/ϪLeu/ϪTrp medium. Wild-type and mutated AtMYC1 were expressed in yeast and examined by Western blotting. B, physical interaction between AtMYC1/ AtMYC1-1 and TTG1 in yeast is shown. C, a BiFC assay to test the interaction between AtMYC1/ AtMYC1-1 and MYB proteins is shown. AtMYC1 and AtMYC1-1 expression in Arabidopsis protoplasts was examined by Western blotting. D, shown is yeast two-hybrid analysis of AtMYC1 dimerization. B and D, protein-protein interactions were assessed by growth on SD/ϪLeu/ϪTrp/ϪHis medium containing 5 mM 3-AT.
its binding partners might be affected. To test these hypotheses, we first assessed the level of expression of AtMYC1 in Atmyc1-1 by semiquantitative and quantitative PCR. We found that the expression level of AtMYC1 was unaffected (Fig. 3, A and B). The subcellular localization of the wild-type and mutated forms of AtMYC1 was also analyzed. Both the wild-type and mutant proteins could localize to the nucleus in tobacco leaf cells after Agrobacterium infiltration (Fig. 5G). Thus, the subcellular localization of AtMYC1-1 was not altered by the mutation either.
Next, protein-protein interactions were examined. As described above, AtMYC1 interacted with GL1, WER, TRY, and TTG1 (Fig. 6. A-C). Thus, we examined the interaction between AtMYC1-1 and MYB proteins, TTG1. Wild-type AtMYC1 was able to interact with GL1, WER, TRY, and TTG1 as shown by its growth on SD/ϪAde/ϪHis/ϪLeu/ϪTrp medium (Fig. 6, A and B). However, this was not true for AtMYC1-1 carrying the R173H substitution. This mutation resulted in an inability to grow on SD/ϪAde/ϪHis/ϪLeu/ ϪTrp medium (Fig. 6B), indicating abolishment of the interaction between AtMYC1 and MYB proteins, TTG1. This result was confirmed in vivo by a BiFC assay in transiently transfected Arabidopsis mesophyll cell protoplasts. Protoplasts cotransfected with AtMYC1-YFP N /GL1-YFP C , AtMYC1-YFP N /WER-YFP C , and AtMYC1-YFP N /TRY-YFP C exhibited strong YFP fluorescence (Fig. 6C). In terms of AtMYC1-1-YFP N /GL1-YFP C , AtMYC1-1-YFP N /WER-YFP C , and AtMYC1-1-YFP N / TRY-YFP C , no YFP signal was produced using the first two pairs, whereas about 2% of the AtMYC1-1-YFP N /TRY -YFP C cells showed YFP fluorescence (reduced by 60% relative to its wild-type counterpart) (Fig. 6C). These results indicated that the inability of AtMYC1-1 to interact with MYB proteins and TTG1 may account for its loss of function and the trichome and root hair patterning defects we observed. These results further suggest that the Arg-173 residue in AtMYC1 is required for its interaction with its binding partners and for its biological function.
Low Complexity Regions (LCRs) and Interaction of AtMYC1 with Its Partners-To explore the possibility that the R173H mutation affected the domain architecture of AtMYC1, we analyzed the mutated protein using the SMART and NCBI protein domain prediction websites. There are three LCRs in wild-type AtMYC1 besides the bHLH domain, and the R173H mutation abolished the second one (Fig. 7A). Because proteins with LCRs tend to have more partners than those without, indicating a role for LCRs in protein-protein interactions (57), we examined the influence of the LCRs in AtMYC1 on its ability to interact with MYB proteins. Several point mutations were introduced into the LCRs to enable us to test their role in protein-protein interactions by yeast two-hybrid analysis. Among the mutations introduced were: R178H, which was located in the same LCR as R173; S174A, another residue in the same LCR; N106A and I234A in the first and third LCRs, respectively (Fig. 7A). AD-GL1 and the different BD-AtMYC1s were cotransformed into yeast and analyzed for ␤-galactosidase activity. Only the R178H mutation resulted in partial reduction of enzyme activity. The other three mutations did not affect the amount of ␤-galactosidase activity despite the fact that they abolished the LCRs (Fig. 7A).
To further test the effect of these mutations, the mutant form of AtMYC1s were driven by its native promoter and transformed into Atmyc1-2. AtMYC1-1 could not rescue Atmyc1-2 phenotype among 81 independent T1 lines (Fig. 7B), indicating full abolishment of the AtMYC1 function by the R173H mutation. However, the other four mutant forms of AtMYC1s could successfully recover mutant phenotype (Fig. 7B), demonstrating that these mutations did not obviously affect protein activity in vivo. Thus, the Arg-173 residue in AtMYC1, other than the LCRs, is crucial for the interaction of AtMYC1 with MYB proteins.
Conserved Arg Residue Is Also Essential for Interaction of AtMYC1 Homologs with Their Partners-Because our alignment showed that residue Arg-173 was conserved among AtMYC1 and its closest homologs (GL3, EGL3, and TT8) (Fig.  2B), we predicted that the corresponding Arg in AtMYC1 homologs may contribute to their protein-protein interactions. To test this, we examined the effect of the conserved Arg mutation on EGL3 and GL3 protein-protein interactions. As before, the Arg residue was changed to a His and inserted into yeast two-hybrid vectors. Like full-length GL1, GL3 has strong selfactivation when fused to the DNA-binding domain of GAL4. Because the N terminus of GL1 (GL1NT), which contains the MYB repeat domain, has been shown to successfully interact with GL3/EGL3 without autoactivation (15), we constructed BD-GL1NT and AD-GL3/EGL3 and tested their interactions by yeast two-hybrid analysis and the measurement of ␤-galactosidase activity. The Arg-to-His mutation in EGL3 affected its interaction with GL1; reduced ␤-galactosidase activity relative to that in the negative control was observed (Fig. 8A). In the case of GL3, ␤-galactosidase activity was decreased by about 70% (Fig. 8A). In our BiFC assays, about 10 and 0.01% of the co-transformed protoplasts exhibited strong YFP fluorescence in their nuclei for wild-type EGL3-YFP N /GL1-YFP C and EGL3R158H-YFP N /GL1-YFP C , respectively, indicating that the Arg residue substitution abolished the EGL3/GL1 interaction in vivo (Fig. 8B). Similar results were obtained for GL3 and GL3R162H; the percentage of cells exhibiting fluorescence was 19 and 5%, respectively, when co-transfected with GL1 (Fig.  8B).
We further tested the importance of this conserved Arg in planta by introducing the point mutation form of protein into gl3 egl3 double mutant. Consistently, among 18 independent T1 lines, no trichome was produced in pEGL3::EGL3R158H transgenic plants (Fig. 8C). As a control, the pEGL3::EGL3 recovered trichome production in 12 independent lines among 17 (Fig. 8C). Similarly, pGL3::GL3R162H failed to complement the glabrous defect of gl3 egl3 double mutant among 22 independent lines, whereas the pGL3::GL3 induce trichome production in 13 independent lines among 20 (Fig. 8C). TT8 was not examined here, as it mainly functions in seed pigment synthesis and does not interact with GL1 (58). Taken together, we identified a conserved Arg residue IIIf bHLH family member that is essential for the proper function and interaction of R/Blike IIIf bHLH proteins with their binding partners.

DISCUSSION
AtMYC1 Controls Trichome and Root Hair Cell Fate Determination in Arabidopsis-We isolated an AtMYC1 mutant with defects in trichome and root hair pattern formation ( Fig. 1; Tables 1 and 2). The characterization of two additional T-DNA insertion alleles and the transgenic complementation assay confirmed that AtMYC1 is involved in trichome and root hair patterning in Arabidopsis (Fig. 3). Our results show that AtMYC1 functions in trichome and root hair control together with MYB proteins and TTG1 (Fig.  6). GL2, the immediate downstream target of MYB-GL3-TTG1, was down-regulated in all alleles of Atmyc1 (Fig. 4A), consistent with the observed phenotypes of Atmyc1 (Fig. 3, C-E; Tables 1 and 2). We speculate that GL2 is also a downstream target of AtMYC1.
Genetic analysis shows that AtMYC1 has both function redundancy and divergence with GL3/EGL3 (Fig. 4). They are involved in trichome and root hair cell fate control redundantly with different degree. The redundancy between AtMYC1 and GL3 is obvious for root hair cell and stem trichome development while less obvious for the leaf trichome (Fig. 4, B-D; Table  2). The redundancy between AtMYC1 and EGL3 is weakest, whereas the strongest is between GL3 and EGL3 (Fig. 4, B-D; Table 2). Their functional divergences are as follows. First, the phenotypes of Atmyc1 and gl3 were different in branch development. gl3 produced fewer branches (15), whereas Atmyc1 showed no change in branch number (data not shown). Second, GL3 and EGL3 formed homo-and heterodimers with each other (14,15); however, AtMYC1 was unable to dimerize under our conditions (Fig. 6D). Third, GL3 and EGL3 are capable to complement leaf trichome and root hair phenotype of Atmyc1, whereas GL3 and EGL3 are irreplaceable by AtMYC1 for leaf trichome and root hair development (Fig. 4E). Fourth, GL3 overexpression partially rescued the glabrous phenotype of ttg1, whereas AtMYC1 overexpression did not (15). In addition, Morohashi and Grotewold (20) found that GL3/GL1 bind to AtMYC1 chromatin, suggesting that AtMYC1 functions downstream of GL3/GL1 and indicating the existence of feed-forward regulation in the function of AtMYC1 and GL3. In conclusion, the functions of AtMYC1 and GL3/EGL3 are not simple redundancy. They have similar yet distinct functions during Arabidopsis epidermis cell fate determination, which is summarized in a simple model in Fig. 9.
Characterization of Atmyc1-1 Reveals Critical Amino Acid Residue for Interaction of AtMYC1 with MYB Proteins and TTG1-Formation of the MYB-bHLH-TTG1 trimeric complex, in which MYB proteins and TTG1 bind a bHLH transcription factor simultaneously, is important for proper development in Arabidopsis (e.g. trichome and root hair patterning, flavonoid/anthocyanin metabolism, and mucilage biosynthesis) (4, 12-14, 35, 39, 40). The N-terminal-most 100 amino acid residues and ϳ200 -400 amino acid residues in the bHLH protein GL3 are required for its interaction with MYB proteins and TTG1, respectively (14). Our in vitro and in vivo results (Fig. 6) and previous yeast two-hybrid or pulldown data (28,45) indicate that AtMYC1, like other bHLH factors, interacts with MYB proteins and TTG1.
We isolated an AtMYC1 mutant with a defect in trichome and root hair patterning ( Fig. 1; Tables 1 and 2). Positional cloning suggested that an R173H mutation in AtMYC1 accounted for the defects in the mutant ( Fig. 2A). By virtue of a point mutation at Arg-173 in AtMYC1, we verified that this Arg residue is essential for the interaction of AtMYC1 with MYB proteins and TTG1 in vitro and in vivo (Fig. 6). The failure to interact with its partner proteins in AtMYC1-1 resulted in altered GL2 expression, leading to defects in trichome and root hair patterning (Figs. 3 and 4A; Tables 1 and 2). Thus, this Arg residue is essential for AtMYC1 and MYBs-TTG1 interactions, downstream gene activation, and Arabidopsis epidermal cell fate determination.
Although Arg to His mutation abolished the formation of a LCR, our results showed that the other four mutations that also led to LCRs abolishment did not affect the gene function in planta (Fig. 7B). These results highlight the importance of this conserved Arg in protein interaction. It was reported that in a high throughput statistical analysis of amino acid frequencies involved in protein interaction, the arginine is the most frequent residue because of its wider radii of action and more accessible long side chains carrying the charge (59). So the Arg to His mutation may change the favorable amino acid feature for protein interaction or alter the protein structure, resulting in failure of interaction with partners.
Arg Residue Is Functionally Conserved among IIIf Subfamily bHLH Transcription Factors-An alignment of AtMYC1 with homologs of the bHLH IIIf subfamily from Arabidopsis showed that the critical Arg-173 residue was conserved among its closest homologs (EGL3, GL3, and TT8; Fig. 2B), indicating the functional conservation of this amino acid residue. By yeast two-hybrid and BiFC assays and transgenic complementation analysis we showed that the interaction between EGL3/GL3 and MYB protein GL1 and their biological function were affected by the corresponding Arg-to-His mutation (Fig. 8, A-C). Thus, our results demonstrate that the corresponding Arg residue in GL3 (Arg-162) and EGL3 (Arg-158) is important for protein interaction. This residue is not in any of the reported FIGURE 9. The comparison of AtMYC1 and its homologs in trichome and root hair cell fate determination. AtMYC1 and its homologs are indicated by ellipsoids of different gray colors. AtMYC1 cannot form dimers, whereas GL3 and EGL3 both can dimerize, which is shown by a single ellipsoid (monomer) and two overlapped ellipsoids (dimer), respectively. The functional redundancy and divergence for AtMYC1, GL3, and EGL3 in control of trichome of stem and leaf and root hair specification are indicated on the right.
MYB-or TTG1-interacting regions; thus, we revealed a novel critical and functionally conserved site in IIIf bHLH proteins for the protein-protein interaction and proper function.