Arabidopsis TRANSPARENT TESTA GLABRA2 Is Directly Regulated by R2R3 MYB Transcription Factors and Is Involved in Regulation of GLABRA2 Transcription in Epidermal Differentiation

Determination of the TTG2 Coding Region

TTG2 is predicted to encode a 428–amino acid protein, although a cDNA clone covering the whole predicted coding region has not yet been reported (Johnson et al., 2002). To confirm this prediction, we performed rapid amplification of cDNA ends (RACE)–PCR. The longest available TTG2 cDNA was 1816 nucleotides long, with an open reading frame (ORF) of 428 amino acids, as predicted previously (Figure 1A ) (Johnson et al., 2002). A comparison of the cDNA clone with the genomic sequence indicated that a 253-bp sequence upstream of the first ATG is divided into 245 bp and 8 bp by a 184-bp intron (Figure 1A). Sequencing of several RACE-PCR products revealed that alternative splicing occurs at the third intron (Figure 1A). In one of eight 5′ RACE-PCR products and one of four 3′ RACE-PCR products, the acceptor site of the third intron was 6 bp downstream of the previously reported site, resulting in a TTG2 protein that is shortened by two amino acids. It is unknown whether both versions of TTG2 protein are active or whether they both exist in the cell, and if they are, which version is more active.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Determination of TTG2 Gene Structure and Localization of the TTG2 Protein.

(A) Schematic of the TTG2 gene. Boxes represent exons; closed boxes represent coding regions, and the gray bar in the fourth exon is absent from cDNA with alternative splicing. The ATG at codon 81 is indicated by lowercase letters. T-DNA is inserted into the second exon in the ttg2-3 mutant. LB, left border.

(B) to (E) Scanning electron micrographs of trichomes of the wild type (Landsberg erecta [Ler]) (B), ttg2-1 (C), ttg2-1 transformed with the FC construct (D), and ttg2-1 transformed with the FxC construct (E).

(F) to (I) Seed coat colors of the wild type (Ler) (F), ttg2-1 (G), ttg2-1 transformed with the FC construct (H), and ttg2-1 transformed with the FxC construct (I).

(J) and (K) Confocal images of trichome and root expressing TTG2:2xGFP. Cell wall stained with propidium iodide is colored in magenta in (K).

Bars in (B) to (E) and (J) and (K) = 100 μm, and bar in (F) (for [F] to [I]) = 1 mm.

To determine whether the first ATG is the actual translation start site, constructs were made using either an intact genomic TTG2 clone containing the sequence from 1977 bp upstream of the first ATG to 426 bp downstream of the stop codon (FC) or construct FC with a CTC in place of the first ATG (FxC). In the FxC construct, the next ATG was at codon 81 (Figure 1A). These constructs were introduced into the ttg2-1 mutant, which is likely to be a null mutant because of the transposition of Tag1 into the third exon (Johnson et al., 2002). In the wild type, most leaf trichomes are three-branched and the seed color is brown (Figures 1B and 1F). ttg2-1 mutants have mostly unbranched trichomes and pale brown seeds (Figures 1C and 1G). Construct FC restored wild-type trichome development and seed coat color (Figures 1D and 1H). The ttg2-1 line transformed with construct FxC had trichomes that were two-branched, and its seed coat color was intermediate between the wild type and ttg2-1 (Figures 1E and 1I). This result suggests that translation from the first ATG is needed for full function of the TTG2 protein. The TTG2 protein is a WRKY transcription factor that has a putative nuclear localization signal. We made the ProTTG2:TTG2:2xGFP construct to show that the TTG2:2xGFP fusion protein accumulates in trichomes and root N cell nuclei (Figures 1J and 1K). Also, this construct complemented the ttg2 mutant, suggesting that the TTG2:2xGFP fusion protein is functionally equivalent to TTG2.

Double Mutant Phenotype

Mutations in TTG2 cause defects in trichome development but not in roots. Several transcriptional regulators are known to be involved in the early stages of trichome development in Arabidopsis. To make it clear where in the regulatory network TTG2 falls, gl3-1 ttg2-1, gl3-2 ttg2-1, and try-82 ttg2-1 double mutants were generated. The ttg1-1 ttg2-1, gl1-1 ttg2-1, and gl2-1 ttg2-1 double mutants have already been analyzed (Johnson et al., 2002). Phenotypes of ttg1-1 ttg2-1 and gl1-1 ttg2-1 are the same as in the single mutants ttg1-1 and gl1-1, respectively. Trichome development in gl2-1 ttg2-1 is more defective than in either single mutant. These results indicate that TTG2 functions downstream of TTG1 and GL1 in the regulatory cascade and shares functions with GL2 in trichome development.

Both gl3 mutant alleles (gl3-1 and gl3-2) produce reduced trichome numbers, and most trichomes have two branches (Figures 2A and 2E ). In the gl3 ttg2 double mutant, a few small pointed outgrowths were observed, which may be rudimentary trichomes (Figures 2B and 2E). This result suggests that TTG2 and GL3 cooperatively regulate trichome initiation and outgrowth.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Trichome Phenotypes of Double Mutants.

(A) to (D) Scanning electron micrographs of trichomes of gl3-2 (A), gl3-2 ttg2-1 (B), try-82 (C), and try-82 ttg2-1 (D). The arrow indicates a trichome-like structure in (B). Bars = 100 μm.

(E) Number of trichomes on fourth leaves of each genotype. Means ± sd of five plants of each genotype are shown. Trichomes are classified depending on the number of branch points. Plants were grown on soil for 21 d under continuous light.

Trichomes on try-82 mutants are often clustered, and many trichomes have more than three branches (Figures 2C and 2E). Most of the trichomes on the try-82 ttg2-1 double mutant were unbranched and similar to those of ttg2-1 (cf. Figures 1C and 2D). The number of trichomes was also similar to that in ttg2-1 (Figure 2E). Although some trichomes of ttg2 were also clustered, there was no obvious increase of trichome clustering in try-82 ttg2-1. These results suggest that ttg2 is epistatic to try in trichome development.

TTG2 Expression in Trichome and Root Hair Mutants

TTG2 is expressed in young leaves, petioles, trichomes, N cells of developing roots, and seed coats (Figures 3A and 3D ) (Johnson et al., 2002). Previous analysis of TTG2 expression in ttg1-1, gl1-1, gl2-1, and ttg2-1 mutant lines indicates that TTG1 is necessary for TTG2 expression in young leaves and in the endothelium of seed coats (Johnson et al., 2002). To find other genes that regulate TTG2 expression, we examined TTG2 expression in several trichome and root hair mutants. We used a ProTTG2:GUS line, containing the pXH2.2 construct, or ProTTG2(2.0):GFPGUS, which is similar to pXH2.2 but with a green fluorescent protein:β-glucuronidase (GFP:GUS) fusion protein reporter rather than GUS (Johnson et al., 2002).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

TTG2 Expression in Mutants.

TTG2 expression in mutants was examined in ProTTG2:GUS lines containing construct pXH2.2 [(A) to (C)] and ProTTG2(2.0):GFPGUS [(D) to (J)].

(A) GUS staining pattern of the pXH2.2 line in the wild type (Ler). Staining was detected in root N cells (left) and in young leaves, petioles, and trichomes (right).

(B) No GUS staining was detected in wer-1 roots.

(C) In the Pro35S:CPC line, some patchy GUS staining was detected in roots (left). GUS staining of young leaves was very weak, and a few GUS-stained trichomes were observed (inset).

(D) GUS staining pattern of the ProTTG2(2.0):GFPGUS line in the wild type (Col-0). Leaf blades stained more intensely than those of the pXH2.2 line. Other tissues stained similarly to those of the pXH2.2 line.

(E) In cpc-2, ectopic GUS staining was detected in some root H cells. Staining in aerial parts was similar to that in the wild type.

(F) In try-29760, staining was similar to that in the wild-type.

(G) In the cpc-2 try-29760 double mutant, GUS staining was detected in all epidermal root cells. GUS staining in young leaves and trichomes was similar to that in the wild type, but expression in trichomes did not last as long as in the wild type.

(H) In gl3-7454, GUS staining was similar to that in the wild type.

(I) In egl3-5712, GUS staining of roots was the same as in the wild type. The staining of leaf blades and petioles was reduced, but trichomes were fully stained.

(J) In the gl3-7454 egl3-5712 double mutant, GUS staining was detected in leaf and cotyledon petioles but not in roots.

Bars in root images = 100 μm, and bars in seedling images = 0.5 mm.

There was no GUS staining in the roots of wer-1 and the gl3-7454 egl3-5712 double mutant (Figures 3B and 3J). Conversely, ectopic GUS staining was detected in some H cells of the cpc-2 single mutant and in all H cells of the cpc-2 try-29760 double mutant (Figures 3E and 3G). These results suggest that WER, GL3, and EGL3 positively regulate TTG2 expression in N cells of roots and that CPC and TRY repress TTG2 expression in H cells. Repression of TTG2 expression by CPC was confirmed with Pro35S:CPC transgenic plants, in which GUS staining in roots, young leaves, leaf blades, and petioles was reduced (Figure 3C). GUS staining of leaf blades and petioles was also reduced in egl3-5712, and the reduction was enhanced by the addition of the gl3-7454 mutation (Figures 3I and 3J). These results also suggest that GL3 and EGL3 positively and cooperatively regulate TTG2 expression in leaf blades. Lastly, GUS staining was observed in the trichomes of all mutants if they formed trichomes.

Analysis of the TTG2 Promoter

Previous analysis of the TTG2 promoter suggested that a 5′ regulatory region extending from −703 to −300 regulates expression in trichomes, seed coats, and other tissues and that a 5′ region from −1000 to −703 regulates expression in roots (Johnson et al., 2002). TTG2 expression was examined in a series of transgenic plants with deletions or base substitutions in the 5′ regulatory region of the TTG2 gene to define elements of the upstream region that are critical for regulation.

The construct ProTTG2(0.7):GFPGUS, which has the same 5′ regulatory and translated regions as pEH0.7, was used to confirm previous results (Johnson et al., 2002). Strong GUS staining was detected in trichomes, petioles, and young leaves of ProTTG2(0.7):GFPGUS plants. Patchy staining was also detected in leaf blades (Figure 4A ). These staining patterns were quite similar to those seen in ProTTG2(2.0):GFPGUS. Although it was reported that the pEH0.7 construct generated only very weak staining in root N cells even after 24 h of staining, 1 h was sufficient to give strong staining in the ProTTG2(0.7):GFPGUS construct (Figures 4A and 5 ) (Johnson et al., 2002). GUS staining was also detected in the trichomes, petioles, leaf primordia, and root N cells of ProTTG2(0.6):GFPGUS (Figure 5). GUS staining in ProTTG2(0.5):GFPGUS was not detected in trichomes themselves, but at their basal regions, and was restricted to the root cap (Figures 4C and 5). In ProTTG2(0.7Δ4):GFPGUS, GUS staining in leaves was similar to that of ProTTG2(0.5):GFPGUS, and it was weak and patchy in N cells and the root cap (Figures 4E and 5). These results strongly suggest that at least one cis-acting element responsible for TTG2 expression in trichomes and roots resides in a region between −609 and −495. We searched for cis regulatory elements in the PLACE (for Plant cis-Acting Regulatory DNA Elements) database and found four putative MYB binding sites within this region (Higo et al., 1999). Because two putative MYB binding sites (−568 to −563 and −567 to −560) partially overlap, the −568 to −560 sequence was designated M1 (AAACCAAAC) (Figure 5). Other putative binding sites located from −531 to −525 and −514 to −508 were designated M2 and M3, respectively (Figure 5). The sequences of M2 and M3 are identical (AACTAAC).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

GUS Expression with Modified TTG2 Promoters.

Seed samples were sectioned and observed under dark-field conditions, in which GUS staining appears pink ([B], [D], [F], [H], [J], [L], and [P]).

(A), (C), (E), (G), (I), (K), (M), (N), and (O) GUS staining of root tips (left) and aerial parts of seedlings (right) of the ProTTG2(0.7):GFPGUS, ProTTG2(0.5):GFPGUS, ProTTG2(0.7Δ4):GFPGUS, ProTTG2(0.7m2m3):GFPGUS, ProTTG2(0.7Δ5):GFPGUS, Pro8xMST:GFPGUS, Pro8xMSTm2:GFPGUS, Pro8xMSTm3:GFPGUS, and Pro8xMSTm2m3:GFPGUS lines, respectively.

(B), (D), (F), (H), (J), (L), and (P) Longitudinal sections of GUS-stained young seeds of the ProTTG2(0.7):GFPGUS, ProTTG2(0.5):GFPGUS, ProTTG2(0.7Δ4):GFPGUS, ProTTG2(0.7m2m3):GFPGUS, ProTTG2(0.7Δ5):GFPGUS, Pro8xMST:GFPGUS, and Pro8xMSTm2m3:GFPGUS lines, respectively.

Bars in root and seed images = 100 μm, and bars in seedling images = 0.5 mm.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Summary of TTG2 Promoter Analysis.

A series of deletions and base substitutions in the 5′ regulatory region of the TTG2 gene and their functions are summarized. At least 10 independent transgenic plants for each construct were examined. White boxes and arrowheads indicate the 5′ regulatory region of TTG2 for each construct. Black stripes indicate the positions of putative MYB binding sites at which base substitutions were introduced. Gray boxes indicate the cauliflower mosaic virus 35S minimal promoter. ^a Relative levels of GUS expression are denoted as follows: ++, strong; +, moderate; + −, patchy (root) or faint (seed); −, not detected; n.a., not analyzed. ^b Whether or not trichome branching is rescued was examined: +, rescued; −, not rescued; n.a., not analyzed. ^c Among 49 T1 transformants, one plant had wild-type trichomes and others had ttg2-like trichomes.

To determine which of the sites are important for TTG2 expression, reporter gene expression under the control of the 5′ regulatory region from −703 to −1 with base substitutions was measured. M1 was mutated from AAACCAAAC to AAtttAAAC, and M2 and M3 were changed from AACTAAC to gcagAAC. GUS staining in ProTTG2(0.7m1):GFPGUS, ProTTG2(0.7m2):GFPGUS, ProTTG2(0.7m3):GFPGUS, ProTTG2(0.7m1m2):GFPGUS, and ProTTG2(0.7m1m3):GFPGUS was similar to that in ProTTG2(0.7):GFPGUS (Figure 5). In ProTTG2(0.7m2m3):GFPGUS and ProTTG2(0.7m1m2m3):GFPGUS, root staining was similar to that in ProTTG2(0.7Δ4):GFPGUS (Figures 4G and 5). These results suggest that either M2 or M3 would be necessary to express TTG2 in the N cells of roots. No difference was observed between GUS staining in ProTTG2(0.7m2m3):GFPGUS and ProTTG2(0.7m1m2m3):GFPGUS, indicating that M1 is not necessary as a cis regulatory element. To confirm this conclusion, we made construct ProTTG2(0.7Δ5):GFPGUS, whose promoter lacks the intervening region from M2 to M3. The GUS staining pattern was similar to that of ProTTG2(0.7Δ4):GFPGUS, suggesting that either M2 or M3 is necessary for TTG2 expression in the N cells of roots, but M1 is not (Figure 4I).

To examine whether M2 or M3 alone would be sufficient for TTG2 expression, we made four reporter constructs. An 8-mer of the 53-bp fragment from −546 to −494, which includes both M2 and M3 (MST), was fused to the cauliflower mosaic virus 35S minimal promoter to drive the reporter gene (Figure 5). Mutated MSTs (MSTm2, MSTm3, and MSTm2m3) were prepared to make similar constructs (Figure 5). GUS staining in Columbia (Col-0) transformants was detected in Pro8xMST:GFPGUS, Pro8xMSTm2:GFPGUS, and Pro8xMSTm3:GFPGUS transgenic lines but not in Pro8xMSTm2m3:GFPGUS (Figures 4K and 4M to 4O). Staining in Pro8xMST:GFPGUS was observed in trichomes and leaf blades (Figure 4K). Some lines showed GUS staining in N cells, but others did not (Figure 4K). GUS staining in the trichomes and leaf blades of Pro8xMSTm2:GFPGUS was similar to that of Pro8xMST:GFPGUS (Figure 4M). However, GUS staining in the roots of Pro8xMSTm2:GFPGUS lines was stronger than that in Pro8xMST:GFPGUS (Figures 4K and 4M). In Pro8xMSTm3:GFPGUS, staining was also observed in trichomes and leaf blades, although leaf blade staining appeared to be stronger than that of Pro8xMST:GFPGUS or Pro8xMSTm2:GFPGUS (Figures 4K, 4M, and 4N). No staining was observed in roots of Pro8xMSTm3:GFPGUS (Figure 4N).

Strong GUS staining was also observed in young embryos and all of the seed coat cell layers of ProTTG2(0.7):GFPGUS, Pro8xMST:GFPGUS, Pro8xMSTm2:GFPGUS, and Pro8xMSTm3:GFPGUS (Figures 4B, 4L, and 5). In ProTTG2(0.7m2m3):GFPGUS, ProTTG2(0.7m1m2m3):GFPGUS, and ProTTG2(0.7Δ5):GFPGUS lines, GUS intensity was weaker, but the pattern was similar to that of ProTTG2(0.7):GFPGUS (Figures 4H, 4J, and 5). Very weak GUS expression was detected in ProTTG2(0.7Δ4):GFPGUS (Figures 4F and 5), and no staining was observed in the embryos or seed coats of ProTTG2(0.5):GFPGUS or Pro8xMSTm2m3:GFPGUS (Figures 4D, 4P, and 5). These results suggest not only that M2 and M3 are required but that other elements are also important for TTG2 expression in young ovules and developing seed coats.

The trichome defect of the ttg2 mutant was rescued when TTG2 was driven by the Pro8xMST, Pro8xMSTm2, or Pro8xMSTm3 promoter (Figure 5). On the other hand, TTG2 driven by Pro8xMSTm2m3, ProTTG2(0.7Δ4), or ProTTG2(0.7Δ5) did not rescue the defects (Figure 5). Interestingly, when TTG2 was driven by ProTTG2(2.0Δ4), in which the region from −609 to −495 was removed from the full-length promoter, trichome defects were also rescued (Figure 5). Because the sequence from −949 to −943 is identical to M2, this region may be a MYB binding site.

Analysis of transgenic lines using the MST fragment suggests that MYB transcription factors bind to MST. WER is one of the binding candidates, because there was no TTG2 expression in the roots of wer-1 mutants. GL1 is also a binding candidate, because it is an early regulator of trichome development and because it is functionally equivalent to WER (Lee and Schiefelbein, 2001). TT2 may also be a candidate, because both TT2 and TTG2 are involved in proanthocyanidin accumulation in seed coats (Nesi et al., 2001; Johnson et al., 2002).

WER and TT2 bound strongly to MST and MSTm3 and significantly to MSTm2 in the yeast one-hybrid system assay (Figure 6 ). GL1 also bound significantly to MST (β-galactosidase units of pGAD-GL1 and pGAD424 transformants were 7.0 ± 0.6 and 1.2 ± 0.5 [mean ± sd], respectively) and apparently bound to MSTm2 and MSTm3 (Figure 6). Because the background activity of β-galactosidase is high in yeast containing the pLacZi-8xMSTm2m3 construct, it is difficult to determine whether or not WER, GL1, and TT2 bound to MSTm2m3. These results suggest that TTG2 expression is directly regulated by WER in roots, by GL1 in trichomes, and by TT2 in developing seed coats. Although WER and GL1 are interchangeable in plants, their binding strengths to MST in yeast were very different (Lee and Schiefelbein, 2001). DNA binding activity and/or the stability of R2R3 MYB protein may differ, because R2R3 MYB proteins are solely expressed as chimeric proteins with the GAL4 activator domain in yeast one-hybrid assays, whereas R2R3 MYB proteins are thought to form a complex with bHLH proteins in planta (Payne et al., 2000; Bernhardt et al., 2003).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Yeast One-Hybrid Assays with R2R3 MYB Transcription Factors and the TTG2 Promoter.

Yeast one-hybrid assays were used to examine the binding activities of WER, GL1, and TT2 to the MST, MSTm2, MSTm3, and MSTm2m3 fragments. pGAD424 is a negative control for effector plasmids, and pLacZi is a negative control for reporter plasmids. Data are presented as means ± sd from three independent experiments.

Expression of the Chimeric TTG2 Repressor Resulted in ttg1 Mutant Phenocopy

To determine whether TTG2 regulates other transcription factors that regulate epidermal cell fate, ProGL1:GUS, ProWER:GFP, ProCPC:GUS, and ProGL2:GUS reporters were introduced into ttg2-1 mutants. There were no apparent differences in GUS or GFP expression patterns between the wild type and ttg2-1 mutants (Figure 7 ). It is thus unlikely that these genes function downstream of TTG2 in the regulatory cascade.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Gene Expression in the ttg2 Mutant.

(A) ProGL1:GUS expression in the wild type. GUS staining was observed in young leaves and trichomes.

(B) ProGL1:GUS expression in the ttg2-1 mutant. The GUS staining pattern was similar to that of the wild type (A).

(C) ProWER:GFP expression in the wild type. GFP fluorescence was observed in N cells. Cell wall stained with propidium iodide is colored in magenta.

(D) ProWER:GFP expression in the ttg2-1 mutant. The GFP fluorescence pattern was similar to that of the wild type (C).

(E) ProCPC:GUS expression in the wild type. GUS staining was observed in N cells (left) and trichomes (right).

(F) ProCPC:GUS expression in the ttg2-1 mutant. The GUS staining pattern was similar to that of the wild type (E).

(G) ProGL2:GUS expression in the wild type. GUS staining was observed in N cells (left) and trichomes (right).

(H) ProGL2:GUS expression in the ttg2-1 mutant. The GUS staining pattern was similar to that of the wild type (G).

Bars in root images = 100 μm, and bars in seedling images = 0.5 mm.

Although TTG2 is expressed in N cells, no phenotypic defects were observed in ttg2 roots. This suggests that one or more other gene(s) compensates for the loss of TTG2 function if TTG2 plays a role in root development. To examine whether TTG2 plays such a role in roots, we made transgenic plants expressing a chimeric protein in which TTG2 was fused with the modified EAR motif repression domain (SRDX), which should act as a dominant repressor (Hiratsu et al., 2003). No trichomes were formed on leaves, anthocyanins did not accumulate in hypocotyls, and ectopic root hairs were formed in the roots of 83 of 106 ProTTG2:TTG2:SRDX T1 transgenic lines (Figures 8I to 8K ). Several glabrous lines were grown to examine their seed coats, which were pale yellow (Figure 8L). These phenotypes resembled those of the ttg1 mutants (Figures 8M to 8P).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Phenotypes of ProTTG2:TTG2:SRDX Transgenic Plants.

Comparison of phenotypes of the ProTTG2:TTG2:SRDX transgenic plant ([I] to [L]) with that of the wild type (Col-0) ([A] to [D]), ttg2-3 ([E] to [H]), and ttg1-11 ([M] to [P]).

(A) Hairless cell files separate root hair cell files in wild-type roots.

(B) Wild-type leaves with three-branched trichomes.

(C) Anthocyanin is accumulated at the junction of the hypocotyl and cotyledons in wild-type seedlings.

(D) Seed coat color of the wild type is dark brown.

(E) Distribution of root hairs in ttg2-3 is similar to that in the wild type (A).

(F) The number of trichomes per leaf is reduced and most trichomes are unbranched in ttg2-3.

(G) Anthocyanin accumulation in ttg2-3 is similar to that of the wild type (C).

(H) Seed coat of ttg2-3 is pale brown.

(I) Root hairs are formed from all cell files in ProTTG2:TTG2:SRDX transgenic plants.

(J) Glabrous leaves of ProTTG2:TTG2:SRDX transgenic plants.

(K) Anthocyanin is not accumulated in ProTTG2:TTG2:SRDX transgenic plants.

(L) Seed coat color of ProTTG2:TTG2:SRDX transgenic plants is much paler than that of ttg2-3.

(M) Root hair cell files are adjacent in ttg1-11.

(N) Glabrous leaves of ttg1-11 are similar to those of ProTTG2:TTG2:SRDX transgenic plants.

(O) Anthocyanin is not accumulated in ttg1-11 or in ProTTG2:TTG2:SRDX transgenic plants.

(P) Seed coat color of ttg1-11 is similar to that of ProTTG2:TTG2:SRDX transgenic plants.

Bars = 0.5 mm.

Real-time RT-PCR was used to compare the expression of seven genes known to influence root epidermal cell differentiation in the ProTTG2:TTG2:SRDX line with the wild type (Col-0) and ttg2-3 (Figure 9A ). The expression of six of the seven genes was similar between the wild type and ttg2-3, with TTG2 as the exception. GL2 expression was markedly reduced in ProTTG2:TTG2:SRDX lines compared with the wild type. CPC and endogenous TTG2 expression were significantly reduced, whereas GL3 expression was significantly higher, in at least one ProTTG2:TTG2:SRDX line. WER expression was somewhat lower, and EGL3 expression was slightly higher, than in the wild type. TTG1 expression was the same as in the wild type. Expression of four reporter constructs in ProTTG2:TTG2:SRDX lines was also examined. ProGL2:GUS was not detected in ProTTG2:TTG2:SRDX (Figure 9B), and expression of ProTTG2:GFP:GUS was reduced and nonspecific to N cells (Figure 9C). In the case of ProCPC:GUS, GUS staining was detected in the stele but not in the epidermis (Figure 9D). ProWER:GFP gave strong GFP fluorescence in N cells but weak fluorescence in H cells (Figure 9E). These results suggest that TTG2 regulates the expression of CPC, GL2, and itself and that the repression of GL2 by the TTG2:SRDX chimeric protein results in ectopic root hair formation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Gene Expression in Roots of ProTTG2:TTG2:SRDX Transgenic Plants.

Expression of genes involved in regulating the differentiation of root epidermis was analyzed by real-time RT-PCR and reporter constructs.

(A) Real-time RT-PCR analysis of WER, CPC, GL3, EGL3, TTG1, endogenous TTG2 [TTG2 (endo)], and GL2 expression in root tissue of the wild type (Col-0), ttg2-3, and two independent ProTTG2:TTG2:SRDX lines (SRDX#8 and SRDX#12). Expression levels are given in relative units, with the wild-type level serving as 1 unit. The mean ± sd of four separate reactions (two replicates of each of two separate biological replicates) is indicated for each sample.

(B) Comparison of ProGL2:GUS expression in roots of the wild type (left) with that of ProTTG2:TTG2:SRDX transgenic plants (right).

(C) Comparison of ProTTG2:GFPGUS expression in roots of the wild type (left) with that of ProTTG2:TTG2:SRDX transgenic plants (right).

(D) Comparison of ProCPC:GUS expression in roots of the wild type (left) with that of ProTTG2:TTG2:SRDX transgenic plants (right).

(E) Comparison of ProWER:GFP expression in roots of the wild type (left) with that of ProTTG2:TTG2:SRDX transgenic plants (right). Cell wall stained with propidium iodide is colored in magenta.

Bars in (B) to (E) = 100 μm.

The N-Terminal Region of TTG2 Is Important for Its Function

Although the TTG2 gene had been shown to encode a WRKY transcription factor in a previous study, no full-length TTG2 cDNA has been isolated (Johnson et al., 2002). In this study, RACE-PCR revealed that TTG2 has an additional noncoding exon and that alternative splicing occurs at the third intron. An ORF of the longest TTG2 cDNA corresponded to the predicted coding region (Johnson et al., 2002). The construct FxC, in which the first ATG of the ORF was changed to CTC, did not rescue the phenotypic defects of mutant line ttg2-1 completely, providing evidence that, except for the possibility of a rare start codon, the first ATG of the ORF is the translation start site (Figures 1E and 1I).

The construct FxC only partially rescued trichome defects in ttg2-1 mutants, as shown by the growth of two-branched trichomes (Figure 1E). Bifurcated trichomes were also observed in transformants of the ttg2-1 line, which contained constructs in which a genomic TTG2 fragment from the third to the fifth exons was driven by the cauliflower mosaic virus 35S promoter in the sense direction (data not shown). These results raise the possibilities that translation also occurs from the codon 81 ATG of the coding region and that a truncated protein that lacks the N-terminal region of TTG2 was produced from the FxC construct (Figure 1A). As the N terminus is rich in Ser and Thr, it is thought to be a transcription activation domain (Johnson et al., 2002). However, because the truncated TTG2 protein has other putative activation domains and motifs, such as WRKY domains and a nuclear localization signal, transcription activity may still be reduced (Johnson et al., 2002).

TTG2 Expression Is Regulated by MYB, bHLH, and TTG1

TTG2 plays a pleiotropic role in plant development and is expressed in several disparate organs. According to a previous study, TTG2 expression in the ttg1-1 mutant is patchy and less regular in roots than in the wild type, suggesting that TTG1 positively regulates TTG2 expression (Johnson et al., 2002). Our analyses demonstrate that WER, GL3, and EGL3 positively regulate TTG2 expression in N cells of wild-type roots (Figures 3B and 3J). The two identical MYB binding sites (M2 and M3) in the MST region of the TTG2 promoter are clearly important for TTG2 expression, and WER, a R2R3 MYB protein, binds to them (Figures 4 to 6⇑⇑). Because the WER protein interacts with bHLH proteins, GL3, and EGL3, and GL3 and EGL3 interact with TTG1 in yeast, transcriptional complexes containing WER, GL3/EGL3, and TTG1 could directly promote TTG2 expression in roots (Payne et al., 2000; Bernhardt et al., 2003; Zhang et al., 2003). In contrast with WER, GL3, and EGL3, CPC and TRY negatively regulate TTG2 expression (Figures 3E and 3G). Both CPC and TRY encode R3 MYB proteins missing a domain that activates transcription (Wada et al., 1997; Schellmann et al., 2002). It has been proposed that R3 MYB proteins compete with R2R3 MYB transcription factors for binding to bHLH transcription factors to inhibit the formation of the regulatory complex (Esch et al., 2003, 2004). Accumulation of CPC and TRY proteins should inhibit the formation of WER-GL3/EGL3-TTG1 complexes and repress TTG2 expression in H cells of wild-type roots and in all epidermal root cells of Pro35S:CPC plants.

TTG2 expression in trichomes, leaf blades, and seed coats is also likely to be directly promoted by R2R3 MYB-bHLH-TTG1 complexes. In trichomes, GL1-GL3/EGL3-TTG1 complexes should promote TTG2 expression, as shown by the binding of GL1 to the MST region in yeast assays (Payne et al., 2000). In seed coats, TT2-TT8-TTG1 complexes should promote TTG2 expression, because TT2 protein also binds the MST region in yeast assays (Baudry et al., 2004). TTG2 expression in leaf blades was repressed in the egl3, gl3 egl3, and ttg1 mutants (Johnson et al., 2002). In these mutants, anthocyanin accumulation is reduced in hypocotyls, cotyledons, and young leaves (Zhang et al., 2003). GL3 and EGL3 interact with the MYB domains of PAP1 and PAP2, which regulate anthocyanin production (Zhang et al., 2003). Because TTG2 is upregulated in a PAP1-overexpressing line, TTG2 transcription should be positively regulated by a PAP1/PAP2-GL3/EGL3-TTG1 complex in leaf blades (Tohge et al., 2005).

Role of TTG2 in Epidermal Differentiation

The CRES-T system, which was applied in the TTG2:SRDX experiment, uses a chimeric transcription factor with the EAR motif repression domain that functions as a dominant repressor (Hiratsu et al., 2003). ProTTG2:TTG2:SRDX lines had an enhanced ttg2 phenotype (Figure 8), suggesting that TTG2 is a transcriptional activator and that the expression of TTG2's target genes would be repressed in ProTTG2:TTG2:SRDX lines.

In roots of the ProTTG2:TTG2:SRDX lines, GL2, CPC, and TTG2 were significantly repressed, and downregulation of GL2 resulted in ectopic root hair formation. Although they were directly downstream of TTG1/WER/GL3/EGL3 in the regulatory cascade, expression of TTG1, WER, GL3, and EGL3 were not significantly repressed (Figure 9), suggesting that TTG2:SRDX downregulates GL2, CPC, and TTG2 without TTG1, WER, GL3, or EGL3 (Koshino-Kimura et al., 2005; Ryu et al., 2005). This situation suggests that R2R3 MYB-bHLH-TTG1 complexes and TTG2 function together to regulate their common target genes (Figure 10 ). The severe trichome phenotype of gl3 ttg2 double mutants supports this possibility (Figure 2B). However, the contribution of TTG2 could be smaller than that of R2R3 MYB-bHLH-TTG1 complexes, because TTG2 expression itself depends on R2R3 MYB-bHLH-TTG1 complexes and there is no apparent difference in the expression of GL2 and CPC between the wild type and ttg2 mutants (Figures 7 and 9A). In addition, the extent of TTG2's contribution may be tissue-dependent. The effect of TTG2 in trichome development may thus be large enough that trichome defects can be observed in ttg2 mutants, whereas it may be so small in roots that root hair defects are undetectable (Johnson et al., 2002). Downregulation of GL2 by TTG2:SRDX may also cause the glabrous phenotype of TTG2:SRDX lines. However, strong ProTTG2:TTG2:SRDX lines are completely glabrous, whereas gl2 mutants form immature trichomes. This inconsistency suggests that TTG2:SRDX suppresses GL2 and other genes that have residual activity in trichome development of gl2 mutants. GL2 is one of the 16 class IV HD-ZIP (HD-ZIP IV) genes in Arabidopsis. Several members of this gene family are actually expressed in trichomes (Nakamura et al., 2006), and some of them may cooperatively regulate trichome development with GL2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

Proposed Model for TTG2 in Epidermal Cell Differentiation.

Black arrows indicate positive regulation, and the thickness of the lines indicates degree of contribution. The red arrow indicates intercellular movement of the R3 MYB protein to inhibit interaction between R2R3 MYB and bHLH in adjacent cells.

The anthocyanin-deficient and severe transparent testa phenotype of ProTTG2:TTG2:SRDX lines suggests that TTG2 is involved in the regulation of both anthocyanin and proanthocyanidin accumulation. It has been reported that ANTHOCYANINLESS2 (ANL2), a member of the HD-ZIP IV family, regulates anthocyanin accumulation in epidermal and subepidermal shoot cells (Kubo et al., 1999). However, ANL2 was not one of the upregulated genes in PAP1-overexpressing plants in which TTG2 was upregulated (Tohge et al., 2005). This suggests that ANL2 is not downstream of PAP1 in the regulation of anthocyanin biosynthesis. Another possibility is that TTG2 regulates structural genes encoding enzymes such as dihydroflavonol reductase and leucoanthocyanidin dioxygenase, which are common in both anthocyanin and proanthocyanidin biosynthesis. A comparison of structural gene expression in the seeds and vegetative tissues of the wild type and ttg2 may reveal some level of structure gene regulation by TTG2.

Plant Materials and Growth Conditions

Arabidopsis thaliana ecotypes Col-0 or Ler were used as the wild type. Except for transgenic plants generated for this study, plant materials are listed in Supplemental Table 1 online. Plants were grown in soil at 22°C under continuous light. Seeds were surface-sterilized and sown on 1.5% agar plates as described previously (Okada and Shimura, 1992) and grown out for observation of seedling phenotypes. In the experiment for anthocyanin accumulation, seeds were sown on 1.5% agar plates containing half-concentration Murashige and Skoog medium (Wako Pure Chemical) and 1.5% sucrose. Seeded plates were kept at 4°C for 4 d and then incubated at 22°C under continuous light.

RACE-PCR

Total RNA was extracted using an RNeasy plant mini kit (Qiagen). cDNA ends were amplified with the 5′ RACE system, version 2, and 3′ RACE system (Invitrogen) using total RNA from wild-type Col-0 seedlings. For 5′ RACE-PCR, primer TTG2-GSP0 was used for first-strand cDNA synthesis. Either TTG2-GSP1 or TTG2-GSP2 was used as the gene-specific primer for the first PCR amplification, and TTG2-GSP2 was used for the second amplification. The 5′ RACE-PCR product was digested with SalI and ClaI and cloned into pBluescript II SK+. For 3′ RACE-PCR, either TTG2-B or TTG2-E was used as the gene-specific primer for the first PCR amplification, and TTG2-E was used for the second amplification. The 3′ RACE-PCR product was digested with BamHI and SpeI and cloned into pBluescript II SK+. Ex Taq DNA polymerase (Takara) was used for cDNA amplification. Primer sequences are described in Supplemental Table 2 online.

Constructs and Plant Transformation

Plasmids for transgenic plants were constructed by standard techniques. Details are provided in the Supplemental Methods online. These constructs were subcloned into the pJHA212K binary vector (a gift from Ji Hoon Ahn) (Yoo et al., 2005) and introduced into Agrobacterium tumefaciens C58C1 or ASE. Arabidopsis plants (Col-0 wild type, ttg2-1, and ttg2-3) were transformed by the floral dipping method and screened on 0.8% agar plates containing half-concentration Murashige and Skoog medium and 50 mg/L kanamycin sulfate.

Real-Time RT-PCR

Total RNA was extracted from the roots of 5-d-old seedlings grown on 1.5% agar plates using the RNeasy plant mini kit (Qiagen). On-column DNase I digestion was done during RNA purification following the protocol in the RNeasy mini handbook. Extracted RNA was redigested with DNase I (amplification grade; Invitrogen) and then purified. First-strand cDNA was synthesized from 1 μg of total RNA in a 20-μL reaction mixture using a PrimeScript RT reagent kit (Takara). Real-time PCR was performed in a Chromo4 real-time PCR detection system (Bio-Rad) using SYBR Premix Ex Taq (Takara). PCR amplification had a 30-s denaturing step at 95°C, followed by 5 s at 95°C and 30 s at 60°C, with 40 cycles for WER, CPC, GL3, EGL3, TTG1, and EF1α and 50 cycles for endogenous TTG2 and GL2. Relative mRNA levels were calculated by iQ5 software (Bio-Rad) and normalized to the concentration of EF1α mRNA. The primers used were as follows: WER-qRTPCRf and WER-qRTPCRr for WER, CPC-qRTPCRf and CPC-R1 for CPC, GL3-qRTPCRf and GL3-qRTPCRr for GL3, EGL3-qRTPCRf and EGL3-qRTPCRr for EGL3, TTG1-qRTPCRf and TTG1-qRTPCRr for TTG1, TTG2en-qRTPCRf and TTG2-BB for endogenous TTG2, GL2-F and GL2-qRTPCRr for GL2, and RT-EF1α-F1 and RT-EF1α-R1 for EF1α (see Supplemental Table 2 online; Czechowski et al., 2004; Kwak and Schiefelbein, 2007).

GUS Staining and Observation

GUS staining of Arabidopsis transformants containing the TTG2 reporter construct series was performed by fixing samples of the aerial parts of the plants in 90% ice-cooled acetone for 15 min, washing with staining buffer [50 mM Na-phosphate, pH 7.2, 3 mM K₃Fe(CN)₆, 3 mM K₄Fe(CN)₆, and 0.5% Triton X-100], and staining overnight at 37°C in staining buffer containing 2 mM X-Gluc (5-bromo-4-chloro-3-indolyl-β-glucronidase). Chlorophyll was removed with an increasing ethanol concentration series. Seedling samples were observed as whole mounts in 70% ethanol. Ovules were observed using dark-field optics by dehydrating silique samples before embedding in Technovit 7100 (Kulzer) for 8-μm sectioning. Root samples were stained directly for 1 h at 37°C in staining buffer containing 1 mM X-Gluc and observed as a whole mount in water. For the other constructs, K₃Fe(CN)₆ and K₄Fe(CN)₆ concentrations and/or incubation times were modified depending on the intensity of staining.

Scanning Electron Microscopy

Scanning electron microscopy samples were prepared and observed as described previously (Kurata et al., 2003).

Confocal Laser Scanning Microscopy

Roots were stained with 5 μg/mL propidium iodide for 30 s and mounted in water. Confocal images were obtained on Zeiss LSM-Pascal or LSM-510 Meta confocal laser scanning microscopes using 488- and 543-nm laser lines for excitation of GFP and propidium iodide, respectively. Images were processed with Adobe Photoshop version 6.0 (Adobe Systems).

Yeast One-Hybrid Assay

The construction of reporter and effector plasmids is described in the Supplemental Methods online. Reporter plasmids pLacZi-8xMST, pLacZi-8xMSTm2, pLacZi-8xMSTm3, pLacZi-8xMSTm2m3, and the intact pLacZi as a negative control were digested with ApaI and integrated into yeast strain YM4271 at the URA3 locus. The resulting yeast strains were selected on SD medium lacking uracil. Three effector plasmids, pGAD-WER, pGAD-GL1, and pGAD-TT2, and the intact pGAD424 were transformed into yeast that had integrated each reporter plasmid. Following the Yeast Protocols Handbook (Clontech), a β-galactosidase assay was performed on each of three individual transformants using o-nitrophenyl β-d-galactopyranoside as substrate.

Accession Numbers

The GenBank/EMBL/DDBJ accession number for the TTG2 cDNA investigated in this study is AB289607. The Arabidopsis Genome Initiative locus identifiers for the genes mentioned in this study are as follows: TTG2 (At2g37260), TTG1 (At5g24520), GL2 (At1g79840), GL3 (At5g41315), EGL3 (At1g63650), GL1 (At3g27920), WER (At5g14750), TT2 (At5g35550), CPC (At2g46410), and TRY (At5g53200).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Table 1. Plant Materials Used in This Study.

• Supplemental Table 2. Primers Used in Constructs and Assays.

• Supplemental Methods. Construction of Plasmids.