Transcriptional activation mediated by binding of a plant GATA-type zinc finger protein AGP1 to the AG-motif (AGATCCAA) of the wound-inducible Myb gene NtMyb2


For correspondence (fax +81 298 38 7408;


NtMyb2 is a regulator of the tobacco retrotransposon Tto1 and the defense-related gene phenylalanine ammonia lyase (PAL), which are induced by various stress stimuli such as wounding or elicitor treatment. NtMyb2 is also induced by wounding or elicitor treatment and is regulated at the transcriptional level. In this study, mutational analysis of the promoter of NtMyb2 and gain-of-function analysis in vivo showed that the sequence AGATCCAA, named the AG-motif, is a cis-element sufficient to confer responsiveness to wounding and elicitor treatment. Furthermore, by using the south-western method, we cloned cDNAs encoding a GATA-type zinc finger protein, which can specifically bind to the AG-motif, named AG-motif binding Protein (AGP1). Domain analysis revealed that not only the GATA-type zinc finger region but also the downstream His2 motif of AGP1 is required for binding activity, showing that the AGP has a novel GATA-type zinc finger domain. AGP1 can activate expression from promoters containing the AG-motif in tobacco protoplasts, indicating that AGP1 is a positive regulator of NtMyb2. We also found that the AGP1 binding activity is highly enhanced by adenine methylation of the AG-motif by bacterial dam methylase.


Plants have many Myb-related genes, whose DNA binding domains are related to that of the mammalian proto-oncogene c-Myb (Majello et al., 1986). Plant Myb-related genes are classified into three groups depending on the number of Myb repeats. Among them, the R2R3-type Myb subgroup, which has the R2 and R3 Myb repeats, forms the largest family in plants. Plant R2R3-type Myb genes regulate various processes (reviewed by Martin and Paz-Ares, 1997; Strack et al., 2001). For example, maize P and C1, and Antirrhinum majus Myb305 regulate the pigmentation of seed aleurone and petal, respectively. Tobacco Myb1, barley GA-Myb, and rice JA-Myb are thought to be involved in responses to salicylic acid, gibberellin, and jasmonic acid, respectively (Gubler et al., 1995; Lee et al., 2001; Yang and Klessig, 1996).

The R2R3-type Myb protein NtMyb2 mediates positive regulation of the tobacco retrotransposon Tto1 and the phenylalanine ammonia lyase (PAL) gene in response to wounding and elicitor treatment (Sugimoto et al., 2000). PAL is a key enzyme of secondary metabolism involved in defense against pathogens. NtMyb2 binds to the 13-bp motif/box L of the Tto1 LTR and PAL promoters and activates transcription of the genes. NtMyb2 activity is itself regulated at the transcriptional level; NtMyb2 mRNA, NtMyb2 protein, and NtMyb2 binding activity are all induced by wounding and elicitor treatment, and overexpression of NtMyb2 in transgenic tobacco induced the constitutive expression of Tto1 and the PAL gene in the absence of wounding or elicitor stimuli. Induction of NtMyb2 expression upon wounding is not inhibited by the protein synthesis inhibitor cycloheximide. Furthermore, NtMyb2 is induced by treatment with okadaic acid, which is known to be an inhibitor of protein phosphatase and an activator of protein kinases negatively regulated by phosphatase 1 and 2A in plants (Raz and Fluhr, 1993). These observations suggested that positive regulator(s) of NtMyb2 might not require de novo protein synthesis and might be activated via phosphorylation by a protein kinase.

As the activity of NtMyb2 is regulated at the transcriptional level in response to wounding and elicitors, we analyzed the promoter of NtMyb2 to elucidate the signal transduction pathways involved. In this study, we identified a cis-element, named the AG-motif, that is sufficient to confer increased transcription in response to these stimuli. Furthermore, we cloned a cDNA encoding an AG-motif binding Protein (AGP), whose DNA binding domain resembles a GATA-type zinc finger structure and which works as a positive regulator in tobacco protoplasts. Interestingly, AGP binding activity was enhanced by adenine methylation of the target DNA, which may provide us a molecular basis for the finding that gene expression is affected by adenine methylation in plants (Higo and Higo, 1996; Rogers and Rogers, 1995; Torres et al., 1993).


The AG-motif in the NtMyb2 promoter is a functional cis-regulatory element in tobacco protoplasts

The NtMyb2 promoter fragment was cloned by inverse-PCR, and its transcription initiation point was determined by primer extension. One major and a few minor initiation sites were detected, and two TATA-box-like sequences corresponding to these sites were found (Figure 1a). Potential cis-elements in this promoter sequence were predicted using the place program (Higo et al., 1999). Sequences resembling Box A of the PAL gene promoter, which is associated with wound responses and related to the ethylene responsive element of tomato (ERF E4), were found at −599 to −594, −246 to −241, and −69 to −75 (Logemann et al., 1995; Montgomery et al., 1993;Figure 1a). Other cis-elements were found at around −110 to −101 and −111 to −103 that are related to the root-specific element (RSE) of the bean Grp1.8 gene (Keller and Heierli, 1994) and to the rbeGA3 cis-element, which mediates expression of GA3 in roots and is recognized by Repression of shoot growth (RSG; Fukuzawa et al., 2000;Figure 1b).

Figure 1.

Sequence of the NtMyb2 promoter and defense-related cis-elements.

(a) The NtMyb2 promoter region. Sequences related to cis-elements involved in defense responses are shown as Box A (Logemann et al., 1995), ERF E4 (Montgomery et al., 1993), and P binding site (Grotewold et al., 1994). The 169- and 48-bp regions shown in Figure 2 are indicated by a box and a bar, respectively. The AG-motif, also shown in Figure 2, is indicated by the shaded box. Two putative TATA boxes are indicated by open boxes. The major and minor transcription initiation sites are indicated by closed and open arrows, respectively.

(b) Alignment of the potential root-specific cis-element of NtMyb2 with the rbeGA3 element of the Arabidopsis GA3 gene (Fukazawa et al., 2000) and the RSE of the bean Grp1.8 gene (Keller and Heierli, 1994). Colons indicate identical nucleotides.

To identify functional cis-elements, deletional analysis of the NtMyb2 promoter sequence was performed using transient expression in tobacco BY2 protoplasts. The NtMyb2 promoter is highly active in protoplasts because of elicitor activities present in the protoplasting enzyme cellulase ONOZUKA (Sugimoto et al., 2000; Takeda et al., 1999). The intact NtMyb2 promoter, from −694 to +80, or 150-bp internal deletion mutants thereof were fused to a reporter gene encoding chloramphenicol acetyl transferase (CAT) and introduced into protoplasts by electroporation. Whereas the intact −694 to +80 promoter showed relatively strong activity (about half of the strong 35S promoter), deletion of the region from −218 to −72 reduced reporter gene activity dramatically (Figure 2a). When a 169-bp fragment (−231 to −62), including the −218 to −72 region, was inserted upstream of the 35S core promoter (−40 to +5), 30% of the activity of the full-length promoter construct was recovered (Figure 2b). Analysis of smaller internal deletions in the −231 to −62 region showed that a 48-bp region (from −133 to −86; deleted in pLIP(40)-CAT-65 and -66) is required for this activity (Figure 2b). Analysis of additional deletions and point mutations in the 48-bp region revealed that an 8-bp sequence (AGATCCAA), named the AG-motif, is required for high expression in protoplasts (Figure 2c). Gain-of-function analysis was performed using plasmids with eight copies of the AG-motif, or mutant derivatives, fused to the 35S core promoter and the reporter gene. The results showed that the repeated AG-motif is a cis-element sufficient for reporter gene expression in protoplasts (Figure 2d).

Figure 2.

Promoter activity of the intact NtMyb2 promoter and deletion mutants in tobacco protoplasts.

The intact NtMyb2 promoter, internal deletion mutants, and base-substituted fragments, fused to the CAT reporter gene, were introduced into tobacco BY2 protoplasts by electroporation.

(a) The intact NtMyb2 promoter compared with derivatives having approximately 150 bp large deletions. CAT activity in the transient expression assay is expressed relative to that of the intact promoter, which was taken as 1. Each value is the mean of three independent experiments.

(b) The activity of the putative NtMyb2 enhancer element (pLIP(40)CAT-6; −231 to −62 of the NtMyb2 promoter), fused to the 35S core promoter, compared with that of the intact NtMyb2 promoter (pLIP-CAT-0) and the core promoter (p35S(40)-CAT) on its own. The 169-bp enhancer region and small internal deletion derivatives are also compared.

(c) The sequence required for enhancer activity, as determined by using base-substitution and deletion mutants.

(d) The activities of the AG-motif (p35S(40)-AG8W-CAT) and its mutant derivative (p35S(40)-AG8M-CAT) compared with those of the entire 169-bp enhancer region (pLIP(40)-CAT-6) and the core promoter (p35S(40)-CAT) alone.

The NtMyb2 promoter and AG-motifs fused to the CaMV 35S core promoter confer responsiveness to wounding and elicitor treatment in vivo

The expression from endogenous NtMyb2 gene by wounding seems to be regulated by its promoter activity; however, we cannot exclude the possibility that the NtMyb2 mRNA level is regulated by stability rather than by rate of transcription from the NtMyb2 promoter. To address this point, we examined the activity of NtMyb2 promoter and roles of the AG-motif in gene expression in response to wounding and elicitor treatment in vivo, by using tobacco transformed with constructs carrying the native NtMyb2 promoter, or AG-motif plus the 35S core promoter, fused to the GUS reporter gene (Figure 3a). Three independent lines of NtMyb2::GUS plant and four independent lines of AG-motif::GUS plant were used for analysis. When the leaves of these transgenic lines were detached, cut into segments, and incubated for 1 h, expression of the GUS gene was induced (Figure 3b, W). GUS expression was also induced when detached leaves were subjected to elicitor treatment by dipping their petioles into a cellulase ONOZUKA solution (1 mg ml−1) (Figure 3b, E). Weak induction was detected in control leaves dipped in MES buffer, possibly because of wounding stress induced when the leaf was excised (Figure 3b, M). In contrast, induction of the GUS gene was not detected in transgenic lines carrying fusions to the mutant AG-motif (Figure 3b, AGx8M). These results, taken together, show that the AG-motif is sufficient for conferring responsiveness to wounding and elicitor treatment. The kinetics of wound-inducible expression of NtMyb2::GUS and AG-motif::GUS were compared with that of the endogenous NtMyb2. The results showed that GUS gene expression from the NtMyb2 promoter highly coincides with the endogenous NtMyb2 expression (Figure 3c, Myb2). On the other hand, the GUS gene fused to the AG-motif promoter was induced slightly faster than NtMyb2 (Figure 3c, AGx8W).

Figure 3.

Activities of the intact NtMyb2 promoter and AG-motifs in response to wounding and elicitor treatment.

(a) The structures of the reporter constructs introduced into tobacco.

(b) Total RNAs were prepared from transgenic tobacco leaves (Myb2, AGx8W, and AGx8M), which were non-treated (0) or wounded (W) and then incubated for 1 h. RNAs were also prepared from transgenic tobacco leaves, whose petioles were dipped into MES buffer (M) or MES buffer with elicitor and then incubated for 1 h (E). A RNA gel blot of these RNAs (5 µg per lane) was probed with NtMyb2-specific and GUS probes (Sugimoto et al., 2000). 28S rRNA stained with methylene blue is shown under each RNA gel blot to demonstrate equal loading of RNA.

(c) The kinetics of wound-inducible expression of NtMyb2::GUS (Myb2 with GUS probe) and AG-motif::GUS (AGx8W with GUS probe) was compared with that of the endogenous NtMyb2 (Myb2 probe). RNAs were prepared from non-treated leaves (0) and wounded leaves incubated for 15 min, 30 min, 1 h, 2 h, 4 h, 8 h, and 24 h (1/4, 1/2, 1, 2, 4, 8, and 24) and probed in the same way as described for (b).

(d) RNAs were prepared from non-treated leaves (0), and wounded leaves incubated with (W/X) or without (W) cycloheximide and probed as for (b) and (c).

NtMyb2 activity is regulated mainly at the transcriptional level. As a protein phosphatase inhibitor, which can activate a kinase activity indirectly (Usami et al., 1995), activates NtMyb2 activity and as NtMyb2 expression itself is not inhibited by the protein synthesis inhibitor cycloheximide, the transcription factor or factors involved in NtMyb2 expression may be regulated by a post-translational mechanism such as phosphorylation (Sugimoto et al., 2000). To confirm this point, we next examined the effect of cycloheximide treatment on the wound responsiveness of reporter gene expression from the NtMyb2 or AG-motif promoter. Cycloheximide treatment did not inhibit the wound inducibility of the NtMyb2 and AG-motif GUS gene fusions. This result shows that NtMyb2 expression is regulated by the post-translational mechanism involving AG-motif and its binding factor, which does not require the de novo protein synthesis (Figure 3d).

cDNAs isolated by the south-western method encode proteins that interact with the AG-motif

To understand the regulatory mechanisms of AG-motif-dependent gene expression, cDNAs encoding proteins that associate with the AG-motif were isolated by using a south-western method. A cDNA library was prepared from protoplasts of the tobacco cell line BY2 and screened with a probe containing the AG-motif sequence. Twenty positive clones were obtained from 4.0 × 105 independent clones. The end sequences of the cDNAs showed that the 20 clones were derived from three different sequences; these were named AGP1, AGP2, and AGP3, and were represented by 2, 16, and 2 clones, respectively (Figure 4a). A homology search using blast showed that the AGP amino acid sequences predicted from the cDNA sequences have a GATA-type zinc finger structure (Figure 4b). Computer analysis by amino acid sequence maximum matching showed that AGP1 is 36% identical to NTL1, which is reported as a possible regulator of nitrate metabolism (Daniel-Vedele and Caboche, 1993). AGP2 and AGP3 are 34 and 31% identical to NTL and 35 and 33% identical to AGP1, respectively. Other cDNA clones, encoding additional AGPs (AGP4 and AGP5), were obtained by screening cDNA libraries using AGP1 as a probe. AGP1 has a zinc finger DNA binding domain in the C-terminal region that is conserved among GATA factors (Figure 4b). Comparison of the AGP showed that not only the GATA-type zinc finger domain but also a His2 motif is highly conserved in the C-terminal region. This conserved region was named the AGP domain. In contrast, the N-terminal regions are highly variable, suggesting that these factors are structurally and functionally divergent. Existence of the conserved putative DNA binding domain suggests that the target sequence may be conserved among the AGP. As tobacco (Nicotiana tabacum) has an amphidiploid genome and AGP5 shows greater similarity to AGP1 (94% identical) than to any of the other AGPs, AGP5 may be orthologous to AGP1. When the amino acid sequence of the AGP domain is compared with the sequences of other GATA-type zinc finger proteins, the most similar protein outside of the plant kingdom is WC-2, which is involved in the blue light response of Neurospora (Figure 4b). Other putative motifs found in AGP1 are a bipartite nuclear localization signal (NLS)- and a nuclear export signal (NES)-like sequence in the N-terminal region (Fischer et al., 1995; Robbins et al., 1991).

Figure 4.

Amino acid sequences of AGP and sequence comparisons among GATA factors.

(a) Alignment of the amino acid sequences of AGP1–AGP5, and NTL1. NLS-like sequences are indicated by bars. A region conserved among the AGP is indicated by an open box.

(b) Comparisons among GATA factor zinc finger domains and the AGP1 domain. Amino acids matching those in AGP1 are indicated by dots. GATA-factors included in this comparison are NTL1 from N. tabacum (Daniel-Vedele and Caboche, 1993); WC-1, WC-2, and NIT2 from N. crassa (Ballario et al., 1996; Fu and Marzluf, 1990; Linden and Macino, 1997); and hGATA-1 from human (Trainor et al., 1990).

Optimal binding site sequence for AGP1 and the effect of methylation on binding activity

As AGP1 was cloned using a probe with an AG-motif-like sequence (AGATCT), the optimal binding site was determined by the binding site selection method. Maximum binding affinity was reached after five and additional two rounds of enrichment by binding selection from a non-specific DNA mixture, and 19 of the bound DNA sequences were determined (Figure 5a). These sequences clearly show that AGP1 can recognize AGATCTA or AGATCCA, the latter being identical to the AG-motif in the NtMyb2 promoter. We examined the binding activity of AGP translated using the reticulocyte lysate system. A 36-bp region, from −113 to −82 of the NtMyb2 promoter, was used as a probe for the electrophoretic mobility shift assay (EMSA; Figure 5b). The probe was prepared from both a DNA adenine methylase (dam+), dcm+ strain (XL1blue) and a dam, dcm strain (JM110), as the adenine residues of the AG-motif could be methylated by bacterial Dam and preliminary data suggested that methylation of the AG-motif changed its affinity for AGP. A mutant AG-motif (AGATCC to AGCGCC) was also tested to further confirm the target specificity (Figure 5b). EMSA using AG-motif as a probe and mutant motifs as unlabeled competitors showed that AGP1 can recognize the AG-motif and not recognize the mutant AG-motif (Figure 5c). As the binding site selection experiment showed that AGP1 can recognize AGATCC and also AGATCT, we compared the affinities of both sequences. The result showed that the affinities of both sequences are nearly the same, suggesting that both sequences may be the target of AGP1 in vivo. EMSA, using the probes shown in Figure 5(b), showed that AGP1–AGP3 can recognize the AG-motif with or without adenine methylation (G6mATC and GATC; Figure 5d), though a difference in the affinity for the DNA probe was observed among these proteins. Interestingly, the affinity for the DNA probe was enhanced by adenine methylation (Figure 5d). None of the proteins could recognize the mutant AG-motif (GCGC; Figure 5d), demonstrating their binding specificity.

Figure 5.

Determination of AGP1′s optimal binding sequence by binding site selection, and EMSA of the AG-motif using nuclear extracts and in vitro-translated AGP.

(a) The binding specificity of AGP1 was determined using random 28-mer oligonucleotides. The selected sites were aligned around the conserved core sequence GATC. The sequences arising from the random sequence are shown in uppercase and the flanking sequence is shown in lowercase. The homologous sequences in the selected clones are boxed.

(b) Four sequences used for EMSA. G6mATC, GCGC, and G6mATCT DNAs were prepared from XL1blue (dcm+ and dam+) and GATC was prepared from JM110 (dcm and dam).

(c) Competition assays for binding of in vitro-translated AGP1 to the probe (G6mATC). EMSA was performed after pre-incubating 25-fold (left lane under the triangle) or 125-fold (right lane under the triangle) excess amounts of unlabeled competitor DNA fragments (G6mATC, GCGC, and G6mATCT). N: no translate was added.

(d) EMSA experiments performed with in vitro-translated AGP1–AGP3. The full-length AGP1–AGP3 DNA complexes are indicated as S. To assure equal loading of the in vitro-translated AGP, these proteins were labeled with 35S-methionine and detected by autoradiography. N: no translate was added.

(e) EMSA experiments performed with nuclear extracts, prepared from non-treated leaf (0) or from leaves that had been wounded and incubated for 45, 90, and 180 min (45, 90, 180). Nuclear extracts were incubated with 32P-labeled probe and electrophoresed. Pre-immune antiserum (1 : 100 dilution; Pre-Im) or anti-AGP1 serum (1 : 100 dilution; AGP1-Im) was added to the binding reaction. Free probe (F) and bands shifted by nuclear factors (S1 and S2) are indicated. N: no extract was added.

To further characterize the AGP, their activity in nuclei was examined by EMSA using nuclear proteins prepared from leaf tissues after wounding (Figure 5e). Although AG-motif binding activity was detected in the nuclear proteins prepared from untreated as well as wounded tissues, the activity was increased by wounding (Figure 5e). The binding activity was enhanced by adenine methylation, as in the case of the in vitro translated AGP (data not shown). These results strongly suggest that the AG-motif binding activity in the nuclei was derived from AGPs. In order to confirm this possibility, anti-AGP1 antiserum was added to the EMSA binding reaction (Figure 5e). The specificity of the AGP1 immune serum is shown in Figure 7(c). The level of one of the shifted bands, indicated by S1, was diminished by addition of AGP1 immune serum (AGP1-Im), indicating that the S1 DNA–protein complex contains AGP1 protein.

AGP1 and AGP3 can activate reporter constructs having the AG-motif in BY2 protoplasts

The activity of AGP1–AGP3 was examined in BY2 protoplasts. The reporter construct consisted of a CAT coding region fused to the 169-bp region, or eight copies of the AG-motif or the mutant motif, along with the 35S minimal promoter (Figure 6a; pLIP(40)-CAT-6, p35S(40)-AG8W-CAT, and p35S(40)-AG8M-CAT). For the effector constructs, AGP1AGP3 cDNAs, and luciferase (LUC) as a negative control, were fused to the 35S promoter (Figure 6a; p35S-AGP1, p35S-AGP2, p35S-AGP3, and p35S-LUC). CAT activity expressed from pLIP(40)-CAT-6 or p35S(40)-AG8W-CAT in BY2 protoplasts was stimulated five- or sixfold when it was co-transfected with the AGP1 expression plasmid (CAT-6 + AGP1, Figure 6b; AG8W + AGP1, Figure 6b,c). Less stimulation was seen for the reporter construct having the mutant AG-motif (AG8M + AGP1; Figure 6c). The effect of AGP2 and AGP3 overexpression was examined in the same way. Co-transfection of the AGP2 effector construct could not induce reporter gene expression, whereas the AGP3 construct could induce the reporter gene to the same levels as the AGP1 construct (Figure 6d). These results, taken together, show that AGP1 and AGP3 can activate NtMyb2 in tobacco protoplasts by binding to the AG-motif.

Figure 6.

AGP1-mediated trans-activation depends on binding to the AG-motif.

(a) Structures of the reporter and effector plasmids. The 169-bp region (−231 to −62; pLIP(40)-CAT-6), eight copies of the AG-motif (p35S(40)-AG8W-CAT), or the mutant derivative (p35S(40)-AG8M-CAT) was fused to the 35S core promoter (−40 to +5) and the CAT gene and used as reporter constructs. The effector plasmids consisted of the whole 35S promoter fused to AGP1–AGP3 cDNA (p35S-AGP1–p35S-AGP3). A plasmid consisting of the 35S promoter fused to the LUC cDNA was used as a negative control.

(b) The effector plasmid (p35S-AGP1, AGP1) or the control plasmid (p35S-LUC, LUC) was co-transfected with the reporter construct pLIP(40)-CAT-6 (CAT-6) into tobacco protoplasts.

(c) The effector plasmid (p35S-AGP1, AGP1) or the control plasmid (p35S-LUC, LUC) was co-transfected with the reporter constructs p35S(40)-AG8W-CAT (AG8W) and p35S(40)-AG8M-CAT (AG8M) into tobacco protoplasts.

(d) One of the effector plasmids, p35S-AGP1 (AGP1), p35S-AGP2 (AGP2), or p35S-AGP3 (AGP3), or the control plasmid (LUC) was co-transfected with the reporter construct p35S(40)-AG8W-CAT (AG8W) into tobacco protoplasts. For (b,c), the average relative CAT activity obtained from three independent experiments is shown with SDs.

AGP RNAs are constitutively expressed in tobacco leaf

As the AG-motif confers wound responsiveness (Figure 3), the expression of AGP1AGP3 in response to wounding was examined. RNA gel blots prepared from leaf segments that had been incubated on 0.05% MES buffer (pH 5.7) for 0.25, 0.5, 1, 2, 4, 8, and 24 h were probed with AGP1–AGP3 cDNA fragments. All three RNAs were expressed in non-treated leaves, but the expression patterns in response to wounding were different (Figure 7a). AGP1 RNA was induced slightly by wounding, but later its expression level was reduced. AGP2 RNA was constitutively expressed even after wounding. AGP3 expression was diminished when NtMyb2 expression level was high.

Figure 7.

RNA gel blot analysis of AGP transcripts and protein gel blot analysis of AGP1.

(a) Each lane was loaded with 5 µg of total RNA from leaf segments harvested at 0, 0.25, 0.5, 1, 2, 4, and 12 h after cutting and probed with NtMyb2-, AGP1-, AGP2-, and AGP3-specific DNA fragments. The 28S rRNA band stained with methylene blue is shown under each RNA gel blot to demonstrate equal loading of RNA.

(b) Each lane was loaded with 5 µg of nuclear protein extract, prepared from non-treated leaf (0) or wounded leaf incubated for 45, 90, or 180 min after wounding (45, 90, and 180). The protein blot was probed with anti-AGP1 antibody and detected using the ECL system (Amersham-Pharmacia).

(c) To examine the specificity of the anti-AGP1 antibody, 1 µl samples of in vitro-translated AGP1–AGP3 were blotted on the membrane and probed with anti-AGP1 antibody. To assure equal loading of in vitro-translated AGP, these proteins were labeled with 35S-methionine and detected by autoradiography.

We further examined the levels of AGP1 protein in the nuclear extract used in Figure 5(e) by protein gel blot using anti-AGP1 antibody. Consistent with its RNA level, the AGP1 protein was present even under normal conditions and the level was slightly reduced by wounding (Figure 7b). The specificity of the anti-AGP1 antibody was examined using in vitro translation products of AGP1–AGP3 cDNA. The result indicates that this antibody can distinguish AGP1 from AGP2 and AGP3 (Figure 7c). These results suggest that the AGP1 protein present in the nucleus in non-treated leaf tissues is inactive and is activated by wound stimuli. As for tissue specificity, AGP1 RNA was detected at the same levels in non-treated leaf, suspension-cultured cells, flower, and root (data not shown).

Mapping of the DNA binding domain in vitro and the activation domain in BY2 protoplasts

The putative DNA binding domain of AGP1 (AGP domain) is similar to GATA-type zinc finger structures, whereas the core of its target sequence is GATC, not the GATA to which GATA-type factors typically bind (see Discussion). Outside of the GATA-type zinc finger, the AGP domain shows less homology with known transcription factors, suggesting that the AGP domain is a novel subtype of the GATA-type DNA binding domain. To determine the DNA binding and activation domains of AGP1, mutational analysis was performed.

A series of N-terminal and C-terminal deletion mutants were made by PCR, and the mutant proteins produced in vitro (D1–D9 and R1–R4; Figure 8a). Activities of the in vitro translated mutant proteins were compared by EMSA using a 36-bp probe carrying the AG-motif (N-terminal deletions, Figure 8b; C-terminal deletions, Figure 8c). Up through the D7 mutant, binding activity was gradually reduced as the size of the N-terminal deletion was increased. However, for the D8 mutant, which had the largest deletion but lacked a small basic region having a putative bipartite NLS-like sequence, binding activity was suddenly restored, and its activity is even stronger than that of the wild type (D8; Figure 8b). Analysis of the C-terminal deletions showed that a 30-amino acid deletion had no effect, whereas a 40-amino acid deletion eliminated DNA binding activity (R4; Figure 8c), suggesting that, at a minimum, the region up to Asn313 is required for DNA binding activity. The region from Ser223 to Asn313, corresponding to the AGP domain, is sufficient for the DNA binding (Figure 4a). In general, the GATA-type zinc finger has two Cys2 zinc binding sites, whereas the AGP1 DNA binding domain retains two Cys2 motifs and one His2 motif, all of which are potential zinc binding sites. Inhibition experiment using the chelator of zinc ion o-phenanthroline shows that zinc ion(s) is required for DNA binding activity (Figure 8d). To determine which site is involved in DNA binding, three mutations (Cys246Ala, Cys269Ala, and His294Ala) were introduced into the AGP1 protein. None of the mutant proteins were able to bind to the 36-bp probe, suggesting that not only the two Cys2 motifs but also the downstream His2 motif is required for DNA binding (Figure 8e). Based on these data, we propose the AGP domain as a novel DNA binding domain whose putative structure is Cys2/Cys2/His2-(Zn2+)2 (Figure 8f). A yeast GAL4 DNA binding domain also forms a binuclear metal cluster involving six Cys residues in a cloverleaf-like array (Pan and Coleman, 1990). In contrast to loop structures of GAL4 DNA binding domain which form alpha-helix, putative loops of AGP domain are predicted to form beta-sheet or coil by the secondary structure prediction algorithm (Chou and Fasman, 1978). This prediction suggests that the number of chelating zinc ion of AGP domain is the same as that of GAL4 DNA binding domain; however, the structure of AGP domain may be different from that of GAL4.

Figure 8.

Mapping of DNA binding domain and activation domain of AGP1.

(a) The second amino acid of each truncated AGP1 protein of the N-terminal deletion series is shown (D1–D9). The end points of the C-terminal deletion derivatives are also shown (R1–R4). The positions of acidic amino acids (E, D) are indicated by the upper bars and basic amino acids (K, R) by the lower bars. An NES, Ser-rich sequence (S cluster), NLS, and a region conserved among the AGP (AGP domain) are indicated.

(b) The binding activity of in vitro translation products from pT7-AGP1, and from pT7-AGP1-D1 to pT7-AGP1D9 were examined by EMSA using the 36 W probe whose sequence is shown in Figure 5(b). The signals corresponding to the protein-DNA complex (S) or free probe (F) are indicated. To assure equal loading of in vitro-translated AGP, these proteins were labeled with 35S-methionine and detected by autoradiography.

(c) The binding activity of in vitro-translated full-length AGP1 or its C-terminal truncated derivatives (R1–R4) were examined by EMSA using the 36 W probe. The signals corresponding to the protein-DNA complex (S) or free probe (F) are indicated.

(d) The requirement of zinc ion(s) for the binding activity was examined by EMSA. Five nanograms of AGP1 protein was pre-incubated with 0, 1 or 5 mm of o-phenanthroline on ice for 30 min and used for the binding reaction. Non: none of AGP1 protein was added.

(e) The binding activity of AGP1 (AGP1) and AGP1 mutants (C246A, C269A, and H294A), in which cysteine or histidine residues in the putative zinc binding site were substituted with alanine were examined by EMSA using the 36 W probe.

(f) Putative structure of AGP domain of Cys2/Cys2/His2-(Zn2+)2 is shown.

(g) The effector plasmid (p35S-AGP1 and p35S-AGP1D1 to -AGP1D9, AGP1 and D1–D9) or the control plasmid (p35S-LUC, LUC) was co-transfected with the reporter construct p35S(40)-AG8W-CAT (AG8W) into tobacco protoplasts.

To determine what other region is required for trans-activation besides the DNA binding domain, the trans-activation activity of AGP1 mutants was examined by co-transfection of AGP1 expression constructs and the reporter construct (Figure 8g). The deletions in mutants D3–D9 abolished the trans-activation activity (Figure 8g), even though the DNA binding activity was retained in vitro (Figure 8b). These results suggest that a 93-amino acid region corresponding to an acidic domain (Acidic-I; Figure 8a) is required for trans-activation. However, we cannot exclude the possibility that the loss of trans-activation activity was a result of the decrease of translation efficiency of AGP1 deletion mutants and/or loss of function of nuclear targeting.


AGP1 and its target sequence AG-motif are involved in NtMyb2 regulation

Previously, we have shown that NtMyb2 is involved in the wound and elicitor response of the retrotransposon Tto1 and the defense-related gene PAL through binding to the 13-bp/box L (Sugimoto et al., 2000). Here, we report the identification of a cis-regulatory element, named the AG-motif that is sufficient for the expression of NtMyb2. We also report the molecular cloning of five GATA-type zinc finger proteins that specifically bind to the AG-motif, named AGP1–AGP5. Mutational analysis of the NtMyb2 promoter in tobacco protoplasts revealed that the AG-motif (AGATCCAA) is a cis-element sufficient for gene expression in protoplasts. Overexpression of AGP1 and AGP3 induced activation of a reporter gene fused to a promoter containing the AG-motif in protoplasts, suggesting that AGP1 and AGP3 are positive regulators of the AG-motif promoter (Figure 6c,d).

Maximum accumulation of the reporter gene transcript expressed from the AG-motif promoter occurred slightly faster in response to wounding than that of transcript expressed from the native NtMyb2 promoter (Figure 3c), suggesting that a cis-element other than the AG-motif is also involved in the transcriptional regulation of NtMyb2. As shown in Figure 1, there are two other potential regulatory elements in the NtMyb2 promoter: an rbeGA3 element, which is thought to be involved in the regulation of the GA3 gene (Helliwell et al., 1998), and an RSE, which is involved in the expression of the bean grp1.8 gene in root (Keller and Baumgartner, 1991). These elements partially overlap the AG-motif. Even if these elements are not directly related to the wounding response, it is possible that the expression of NtMyb2 is affected by an interaction between AGP and rbeGA3 and/or RSE binding protein. In particular, the constitutive expression of NtMyb2 in roots raises the possibility that RSE works as a regulatory cis-element of NtMyb2 (Sugimoto et al., 2000).

We searched for genes having the AG-motif or AG-motif-related sequences as putative targets of AGP. The octamer motif, which is an essential cis-element for S-phase-specific gene expression of the histone H3 gene, has the AG-motif-like sequence GATCC (Taoka et al., 1999). However, a reporter gene fused to a promoter having a wheat type-I octamer motif without the AG-motif-like sequence retains S-phase-specific expression, suggesting that the contribution of this AG-motif-like sequence may be limited.

The AGP domain as a novel type of zinc finger structure

The GATA-type transcription factors are widespread among eukaryotes and are involved in diverse regulatory processes. For example, GATA-1 and GATA-3 regulate gene expression in mammalian erythroid cells (Trainor et al., 1990; Tsai et al., 1989; Zon et al., 1991). The yeast dal80 gene encodes a negative regulator of multiple nitrogen catabolic genes, and NIT2 and GLN3 work as positive regulators of nitrogen metabolism in Neurospora crassa and yeast (Fu and Marzluf, 1990; Minehart and Magasanik, 1991). WC-1 and WC-2 regulate the blue light responses of N. crassa (Talora et al., 1999). In plants, NTL1 was cloned based on homology with NIT2 and was discussed as a regulator of nitrogen metabolism in plants (Daniel-Vedele and Caboche, 1993). In Arabidopsis, 28 GATA factors have been deduced from the whole genome sequence (Riechmann et al., 2000). The binding site of the GATA-type factors is usually GATA. For example, human GATA-1, with two zinc fingers, recognizes (T/A)GATA(A/G; Omichinski et al., 1993). NTL1 has been predicted to possess a single GATA-type zinc finger structure, but its target sequence and activity have not been reported.

The DNA binding domain of the AGPs is larger than the conserved GATA-type zinc finger structure; in the case of AGP1, DNA binding activity requires not only the Cys2/Cys2 but also the His2, with 25-amino acid spacing between them, and this structure is highly conserved among the AGP. We predicted that the structure of the AGP1 DNA binding domain is Cys2/Cys2/His2-(Zn2+)2-type zinc finger like that of the GAL4 DNA binding protein (Pan and Coleman, 1990) but with the large loops that are predicted to form beta-sheet or coil, different from alpha-helix of GAL4 loops (Figure 8f). AGP1 requires the His2 motif for DNA binding; however, this motif is not found in non-plant GATA factors. The co-translation of AGP1 and truncated AGP1, AGP2, or AGP3 followed by DNA binding analysis showed that AGP do not form homo- or heterodimers (data not shown). Therefore, we propose that AGP domain is a novel zinc finger structure not classified by Krishna et al. (2003), and it works as a monomer, e.g. each of the two loops recognizes a different region of the AG-motif.

Signal transduction pathway of wounding

The expression level of AGP1 mRNA is up-regulated by wounding; however, the mRNA is constitutively detected in leaf tissue and tissue cultures (Figure 6a). Consistent with this observation, AGP1 protein was constitutively detected in nuclear extracts (Figure 7b). On the other hand, the activation of NtMyb2 and the AG-motif::GUS gene is not inhibited by treatment with cycloheximide, suggesting that AGP1 is activated by a modification that does not require de novo protein synthesis. In the absence of wounding or elicitor treatment, NtMyb2 mRNA is induced by treatment with Okadaic acid (Sugimoto et al., 2000), which is known as a phosphatase inhibitor and as an activator of kinases, which are negatively regulated by phosphatase (Usami et al., 1995). These results raise the possibility that AGP1 is activated by phosphorylation in response to wounding and elicitor treatment.

There are several plant kinases that have been linked to defense responses. For example, the tobacco MAP kinases SIPK and WIPK are transiently activated by wounding and treatment with elicitors (Seo et al., 1995; Zhang and Klessig, 1998). Another class of plant kinases, the calcium-dependent protein kinases (CDPK), are activated by interaction of the resistance gene Cf-9 and its avirulence gene product AVR9, by hypo-osmotic stresses, and by non-specific elicitor (Romeis et al., 2000). MAPKs and CDPKs are thought to be involved in the defense responses; however, downstream factor(s), substrate of these kinases, is still not known. AGP might be a substrate of MAPK or CDPK in response to wounding and treatment with elicitors.

The effect of adenine methylation on binding of AGP

Cytosine methylation participates in the regulation of gene expression by chromatin modification and transcriptional silencing (reviewed by Ng and Bird, 1999). In contrast with cytosine methylation, adenine methylation seems to be limited to enteric bacteria and is involved in the regulation of DNA replication and gene expression (Crooke, 1995; Huisman and de Graaf, 1995). Although the biological significance of adenine methylation is not clear in plants, gene expression (Graham and Larkin, 1995; Higo and Higo, 1996; Rogers and Rogers, 1995; Torres et al., 1993) and flower and leaf development (van Blokland et al., 1998) were affected by adenine methylation. The finding that adenine methylation enhances both the ability of AGP1 to bind to the AG-motif (Figure 5) and the activity of the promoter in protoplasts may give a molecular basis for these observations. As the reporter gene fused to the AG-motif and the 35S core promoter was induced by wounding and elicitor treatment in transgenic tobacco (AGx8W; Figure 3), adenine methylation is not essential for the AG-motif-mediated response in vivo.

Experimental procedures

Plant materials and treatment of plant tissues

Tobacco plants (N. tabacum cv. SR1) were grown at 26°C in a greenhouse. Protoplasts were prepared from BY2 suspension cultures as described by Hirochika et al. (1996). Leaves were wounded as follows: a whole leaf was wounded by a cheese grater and then floated on the buffer (0.05% Mes/KOH, pH 5.7). Elicitor treatment was performed as follows: a petiole attached to an intact leaf was dipped into an elicitor solution of cellulase Onozuka R10 (1 mg ml−1; Yakuruto Biochemicals Co., Ltd, Tokyo, Japan).

Cloning of NtMyb2 promoter by I-PCR and primer extension

The NtMyb2 promoter fragment was cloned by I-PCR, using specific primers designed according to the 5′ and 3′ non-coding regions (110R: 5′-ATGACTCTTTAATCACTTGG-3′ and 1371F: 5′-CAACATCCTCCTAACTTAGC-3′) of the NtMyb2 cDNA sequence. PCR fragments were cloned into the BamHI site of pBlueScriptSK(+) (Stratagene, La Jolla, CA, USA), resulting in pSK-LIP. The primer extension experiment was performed using a primer extension kit (Promega, Madison, WI, USA) and 51R primer (5′-AGTGTACTGTTTCTTTTGGATGTGT-3′).

Constructs for promoter analysis

The internal deletion mutants of the NtMyb2 promoter were made as follows: upstream regions of the internal deletions were made by PCR with or without point mutations to make an AseI and pSK-LIP as a template. PCR products were digested with AseI, filled in with Klenow enzyme, and digested with XbaI. The downstream regions were made by PCR, which has a mutation to generate a BamHI site in the position corresponding to +80 of the NtMyb2 promoter, and a specific primer with a mutation to generate a DraI, followed by digestion with BamHI and DraI. An appropriate set of fragments was inserted between the XbaI and BamHI sites of pLTR-CAT4 (Hirochika et al., 1996), resulting in pLIP-CAT-2–pLIP-CAT-6. The full-length NtMyb2 promoter fragment (−694 to +80) and truncated fragments (−617 to +80 and −71 to +80) were amplified by PCR, digested with XbaI and BamHI, and inserted into pLTR-CAT4 XbaI and BamHI sites, resulting in pLIP-CAT-0, pLIP-CAT-1, and pLIP-CAT-7, respectively. The deletion start and end points of the resulting plasmids pLIP-CAT-1–pLIP-CAT-7 were −694 to −620, −617 to −500, −505 to −408, −414 to −315, −320 to −222, −218 to −72, and −694 to −72, respectively, from the transcription start site. The 169-bp region from −231 to −62 was amplified by PCR and cloned into the EcoRV site of pBlueScriptSK(+), resulting in pSK-LIP6. The HindIII–PstI fragment from pSK-LIP6 was inserted between the HindIII and PstI sites of p35S(40)-CAT (Takeda et al., 1999), resulting in pLIP(40)-CAT-6. The internal small (<24 bp) deletion mutants of the 169-bp region were made as follows: upstream fragments were made by PCR using universal M13 primer F (M13-F: 5′-CGTTGTAAAACGACGGCCAG-3′) and a specific primer whose 5′-terminal 20 nt matched the upstream side of the deletion start point and 3′-terminal 10 nt matched the downstream side of the deletion end point. The downstream fragments were also made by PCR using universal M13-RV primer and a specific primer whose 5′-terminal 20 nt matched the downstream side of the deletion end point and whose 3′-terminal 10 nt matched the upstream side of deletion start point. Next, appropriate PCR products were mixed (1/10 volume each) and amplified with the two universal primers (M13-F and M13-RV) to make fused fragments, which were then digested with HindIII and PstI and ligated with a HindIII and PstI digest of p35S(40)-CAT, resulting in pLIP(40)-CAT-62, -63, -64, -65, -66, -653, -654, -655, -656, -661, -662, and -663. The PCR products amplified using primers 61F: 5′-CGATAAGCTTGTGTAATACACACTTGTCGT-3′ and M13-RV, and M13-F and 67R: 5′-CGGGCTGCAGTGTTTGCGATTTCCACGAAG-3′ were digested with HindIII and PstI and inserted into p35S(40)-CAT, resulting in pLIP(40)-CAT-61 and -67, respectively. The deletion start and end points of the resulting plasmids pLIP-CAT-61–pLIP-CAT-67 were −221 to −206, −205 to −182, −181 to −158, −157 to −134, −133 to −110, −109 to −86, and −85 to −62, respectively, from the transcription start site. The point mutations were introduced by PCR as previously described by Sugimoto et al. (2000), resulting in pLIP(40)-CAT-651 and -652. The oligonucleotides 5′-agctaa(AAGATCCAA)8aactgca-3′ and 5′-gtt(TTGGATCCTT)8tta-3′, and 5′-agctaa(AAGCGCCAA)8aactgca-3′ and 5′-gtt(TTGGCGCCTT)8tta-3′ were annealed and ligated between the HindIII and PstI sites of p35S(40)-CAT, resulting in p35S(40)-AG8W-CAT and p35S(40)-AG8M-CAT. The sequences of all of the PCR-amplified fragments or inserted oligonucleotides were re-confirmed by sequencing.

Transient-transfection assays in tobacco protoplasts

Protoplasts prepared from the BY2 cell line (106 cells in 800 µl) were mixed with 5 µg of each of the plasmids and 20 µg of sonicated calf thymus DNA and electroporated using an electroporation apparatus (Gene Pulser; Bio-Rad, Hercules, CA, USA). After the electroporation, the protoplasts were cultured for 18 h at 26°C. Lysis of the protoplasts and the CAT assay were carried out as described by Hirochika et al. (1987).

Plant transformation

The XbaI–BamHI promoter fragment from pLIP-CAT-0, which has the full-length NtMyb2 promoter, was inserted between the XbaI and BamHI sites of the binary vector pBI-101.3 to make pBI-NtMyb2-GUS. The HindIII–BamHI promoter fragments from p35S(40)-AG8W-CAT and p35S(40)-AG8M-CAT were inserted between the HindIII and BamHI sites of pBI-101.3, resulting in pBI-AG8W-GUS and pBI-AG8M-GUS, respectively. Agrobacterium tumefaciens EHA101 (Hood et al., 1986) was transformed with pBI-NtMyb2-GUS, p35S(40)-AG8W-CAT and p35S(40)-AG8M-CAT using electroporation. SR1 tobacco was transformed by the leaf-disk method (Horsch et al., 1984).

Construction of the cDNA library and south-western screening

Total RNA from tobacco BY2 protoplasts was extracted using ISOGEN (Nippongene, Tokyo, Japan). Poly(A)+ RNA was purified using OligoTex-dT30 (Roche, Basel, Switzerland). The cDNA was synthesized using a cDNA synthesis kit and cloned into the SalI–NotI sites of λ-gt22 (Gibco-BRL, Gaithersburg, MD, USA). The resulting cDNA library contained 1 × 106 independent clones.

The probe for the south-western screening was a 4-mer, a concatamer four repeats long, of the oligonucleotides 5′-CTAGATGGTAGGTGAGAT-3′ and 5′-CTAGATCTCACCTACCAT-3′, which were annealed and cloned into the XbaI site of pBlueScriptSK(+). The HindIII–EcoRI fragment of this construct was used as a probe. An AG-motif-related sequence was formed at the junctions of the annealed oligonucleotides, and the XbaI sites at the junctions were protected by methylation with bacterial dam methylase.

Using Escherichia coli strain Y1090 as a host, 2 × 104 Pfu of the cDNA library were spread on a 13 cm × 9 cm LB agarose Petri dish and screened by south-western method described by Vinson et al. (1988).

Subcloning of the cDNAs and sequence analysis

Phage DNAs prepared from positive plaques were PCR-amplified and sequenced at both ends using a Taq Dideoxy Terminator Cycle Sequencing Kit (ABI, Foster City, CA, USA). By clustering, three independent clones were selected and subcloned into the SalI–NotI sites of pBlueScriptSK(+), resulting in pSK-sAGP1, pSK-sAGP2, and pSK-sAGP3. Using these inserts as probes, longer cDNAs were cloned and the SalI–NotI fragments were subcloned into the SalI–NotI sites of pBlueScriptSK(+), resulting in pSK-AGP1, pSK-AGP2, and pSK-AGP3. AGP4 and AGP5 were cloned by a lower stringency screening using the AGP1 BbsI–BbsI fragment as a probe and were subcloned by the same process, resulting in pSK-AGP4 and pSK-AGP5. Analyses of primary sequences and multiple sequence alignments were performed with the software genetyx-win Ver. 4.00 (software development).

Binding site selection experiment

The binding site selection experiment was performed as described by Pollock and Treisman (1990), with some modification. Random DNA was made by PCR using primer (BSS-F: 5′-ATAAGCACAAGTTTTATCCG-3′, BSS-R: 5′-GAATTCCGGATGAGCATTCA-3′) and template (BSS-N28: 5′-ATAAGCACAAGTTTTATCCG(N28)TGAATGCTCATCCGGAATTC-3′) oligonucleotides. After five rounds of enrichment were performed as described by Pollock and Treisman (1990), an immunoprecipitation step using the antibody was substituted for affinity purification using the glutathione S-transferase (GST) system. Further enrichment was performed as follows: using polynucleotide kinase, the 5′ ends of the PCR products were labeled with 32P and used for the binding reaction of EMSA. The shifted band was excised and used as a template for the next round of amplification by PCR. For the binding reaction, we used AGP1 protein (without an affinity tag) translated in vitro using reticulocyte lysate of TNT system (Promega).

Preparation of nuclear extracts and EMSA

Nuclear extracts were prepared as described by Green et al. (1987), with some modifications as described by Sugimoto et al. (2000). Nuclear extracts were prepared from leaves wounded by a cheese grater as described in the previous section and incubated for 45, 90, and 180 min.

In vitro translation was performed with the TNT Coupled Reticulocyte Lysate System (Promega) using 1 µg of template plasmid per 50 µl reaction. The translation efficiencies of each AGP and AGP mutant protein were examined by assaying 35S-methionine incorporation. A 1-µl sample of each of the in vitro translation products was mixed with 0.5 ng 32P-labeled DNA and 2 µg poly(dI•dC)•poly(dI•dC) in K/DSB (25 mm Hepes/KOH (pH 7.9), 1 mm EDTA, 50 mm KCl, 10 mm 2-mercaptoethanol, and 10% glycerol) and incubated for 30 min at 4°C. To separate the free and bound DNA, reactions were run out on a 4% polyacrylamide-bisacrylamide (40 : 1) gel in 0.5× TGE (25 mm Tris, 192 mm glycine, and 1 mm EDTA) at 4°C. The oligonucleotide pairs LIP36F (5′-CGAGATCCAACTTCGTGGAAATCGCAAACAAAGA-3′, corresponding to nucleotides −114 to −82 of the NtMyb2 promoter) and LIP36R (5′-CGTCTTTGTTTGCGATTTCCACGAAGTTGGATCT-3′), LIP36M1F (5′-CGAGCGCCAACTTCGTGGAAATCGCAAACAAAGA-3′) and LIP36M1R (5′-CGTCTTTGTTTGCGATTTCCACGAAGTTGGCGCT-3′), or LIP36M2F (5′-CGAGATCTAACTTCGTGGAAATCGCAAACAAAGA-3′) and LIP36M2R (5′-CGTCTTTGTTTGCGATTTCCACGAAGTTAGATCT-3′) were annealed and inserted into the ClaI site of pBlueScriptSK(+), resulting in pSK-LIP36W, pSK-LIP36M1, and pSK-LIP36M2, respectively. Probe and competitor DNAs were prepared as follows: a HindIII and SalI digest was labeled by filling in using Klenow enzyme, isolated by PAGE, and eluted.

Inhibition experiments using o-phenanthroline were performed as follows: AGP1 protein was prepared from E. coli extract expressing full-length AGP1 fused to calmodulin binding peptide (Affinity Protein Expression System with pCAL-n; Stratagene). Approximately 5 ng of AGP was incubated with 0, 1, and 5 mmo-phenanthroline for 30 min on ice and used for EMSA described above.

Preparation of specific antibody and protein blotting

A GST fusion of AGP1 was expressed in E. coli (XL1Blue) from pGEX-sAGP1 and purified on a glutathione-Sepharose affinity column. The AGP1 polypeptide was eluted from the affinity column by thrombin digestion and was used for the immunization of rabbits. The AGP1-specific antibody was purified using antigen-Sepharose. This purified antibody (0.5 mg ml−1) was diluted 5000-fold and used as a primary antibody for protein gel blot analysis using an ECL Detection kit (Amersham-Pharmacia, Uppsala, Sweden). For the gel-shift assay, pre-immune serum and immune serum were used.

Constructs for transcriptional activation assay and mapping of the DNA binding domain

The oligonucleotides 5′-GATCCGTCGACGAATTCCCGGGCTCGAGCGGCCGCGGTAC-3′ and 5′-CGCGGCCGCTCGAGCCCGGGAATTCGTCGACG-3′ were annealed and inserted between the BamHI and KpnI sites of p35S-CAT-N (Hirochika et al., 1996), resulting in p35S-Ter-S, which has unique XbaI, BamHI, SalI, SmaI, XhoI, NotI, KpnI, and SacI sites between the 35S promoter and the nos terminator. The SalI–NotI fragments of pSK-AGP1, pSK-AGP2, and pSK-AGP3 were inserted between the SalI and NotI sites of p35S-Ter-S, resulting in p35S-AGP1, p35S-AGP2, and p35S-AGP3, respectively. A PCR fragment, amplified with primers (164F; 5′-AAGAAAGAGTGGATCCGATGGGGTCAA-3′ and 395R; 5′-CAATATCCTCATACGGAACTG-3′) using pSK-AGP1 as the template was digested with BamHI and SacI. The AGP1 C-terminal region was excised with SacI and EcoRI from pSK-AGP1 and ligated with the BamHI–SacI fragment of AGP1 and BamHI–EcoRI digest of pRSET-B, resulting in pRSET-AGP1. A PCR fragment was amplified from the AGP1 N-terminal region, using the primers having a mutation to make NdeI site (CATATG) and the NdeI–MluI and MluI–EcoRI fragments were inserted into pRSET-B NdeI–EcoRI sites, resulting in pT7-AGP1D0–pT7-AGP1D9. PCR products amplified using the primers that have sequence substitutions to make stop codon (TTA) and T7U, 5′-TGGCCGATTCATTAATGCAG-3′, which could amplify the region from the T7 promoter to the end of the intact or C-terminal truncated AGP, were purified using a QiaQuick PCR Purification Kit (Qiagen) and used as a template for the TNT-Coupled in vitro Transcription and Translation Kit (Promega). The resulting translation products were used for the binding reaction in Figure 8. Fragments containing intact and truncated AGP1 were excised with XbaI and MluI and inserted into the XbaI and MluI sites of p35S-AGP1, resulting in p35S-AGP1D0–p35S-AGP1D9. Amino acid substitutions were introduced during PCR amplification using primers having base substitution as described by Sugimoto et al. (2000). The BbsI–BbsI fragment of pRSET-AGP1 was replaced with these PCR fragments digested with BbsI, resulting in pRSET-AGP1C246A, -AGP1C269A, and -AGP1H294A.


We thank Dr K. Nakamura for kindly providing the plasmid pBI-H1. This work was supported by a project grant from the Ministry of Agriculture, Forestry and Fisheries of Japan and an enhancement of Center-of-Excellence, special coordination funds for promoting science and technology in Japan.

DDBJ/EMBL/GenBank Accessions for the NtMyb2 promoter, AGP1, AGP2, AGP3, AGP4, and AGP5 are AB107989, AB107689, AB107690, AB107691, AB107692, and AB107693, respectively.