Regulation of DNA Binding and trans-Activation by a Xenobiotic Stress-activated Plant Transcription Factor*

As-1-typecis-elements augment transcription of both nuclear and pathogen genes in response to stress and defense cues in plants. Basic/leucine zipper proteins termed “TGA factors” that specifically bind as-1 elements are likely candidates for mediating these transcription activities. Our earlier work has shown that 2,4-dichlorophenoxyacetic acid-induced xenobiotic stress enhancestrans-activation by a chimeric fusion protein of the yeast Gal4 binding domain and TGA1a, a TGA factor of tobacco. Here we demonstrate that xenobiotic stress also enhances the ability of native TGA1a to bind as-1 and activate transcription of a known target gene. In addition, the previously identified xenobiotic stress-responsive domain of TGA1a was found to inhibit this factor'strans-activation potential by a mechanism that appears to involve stimulus-reversible interactions with a nuclear repressor protein. Results from these and other studies can now be placed in the context of a working model to explain basal and xenobiotic stress-induced activities of TGA1a through its cognatecis-acting element.

As-1-type elements contribute to the expression of both pathogen and nuclear genes in plants (1)(2)(3)(4)(5)(6)(7)(8)(9). When inserted upstream of a minimal promoter and ␤-glucuronidase reporter gene, these elements drive ␤-glucuronidase expression predominantly in primary and lateral root tips of tobacco and Arabidopsis seedlings, and in the root and shoot vascular system of older plants (2,10,11). This transcription activity is further augmented in response to plant defense hormones (e.g. salicylic and jasmonic acids), wounding, and xenobiotic stress, thus indicating that diverse stimuli affect the activity of one or more cognate transcription factors through these elements.
A number of genes have been cloned that encode for as-1binding proteins, both in the same and different plant species. These "TGA factors" share considerable homology and belong to the basic/leucine zipper (bZIP) 1 family of transcription factors. Efforts to understand the contributions of TGA factors to as-1-dependent transcription will require knowledge of their cellular distribution and molecular properties. To this end, we initially chose to study TGA1a because close homologues of this tobacco TGA factor exist in other plants, suggesting a conserved and perhaps important biological role. In prior studies, we found that a chimeric protein comprised of the yeast Gal4 binding domain and TGA1a could potentiate transcription through the GAL4 cis-element in response to xenobiotic stress. These data implicate TGA1a in the expression of plant genes involved in chemical defense (12). Consistent with this notion, TGA1a transcripts and protein are preferentially coexpressed in root meristem cells with transcripts of as-1-regulated genes (e.g. GNT35) (11) that encode for type III glutathione S-transferase (GST) isoenzymes. Although the biological function of these as-1-regulated GSTs is unknown, this class of enzymes has been specifically implicated in xenobiotic detoxification and resistance in plants (13)(14)(15).
Gain of function assays with the yeast Gal4 binding domain led to the identification of specific TGA1a domains involved in basal and xenobiotic stress-activated transcription (12). By this assay, it was shown that the amino-terminal (NT) domain (residues 1-86) of TGA1a confers constitutive trans-activation, whereas the carboxyl-terminal (CT) domain (residues 142-373) of this factor largely enhances transcription in response to xenobiotic stress. Here we show that xenobiotic stress rapidly and transiently affects the ability of native TGA1a to bind its cognate as-1 element and to activate transcription through a mechanism involving stimulus-reversible repression and this factor's regulatory CT domain. Additional evidence suggests that this regulatory mechanism is likely to occur through stimulus-reversible interactions of the CT domain with a putative corepressor protein.

Tobacco Suspension Cell Cultures
Cultures (100 ml) of tobacco BY-2 suspension cells were grown and maintained in flasks at 28°C in the dark as described previously (12). For the preparation of labeled nuclear proteins, cells were cultured for 4 days and adjusted to a packed volume of 50%, and 4 ml of the cell culture were transferred to each of 5 wells in an 8-well culture plate (Costar). After 24 h, 200 Ci of [ 35 S]methionine (6000 Ci/mmol; PerkinElmer Life Sciences) were added to each sample, which was then incubated for an additional 20 h. Where indicated, cells were exposed to xenobiotic stress induced by treatment for 0 -8 h with 100 M 2,4dichlorophenoxyacetic acid (2, 4-D) in 0.1% ethanol carrier, collected by vacuum filtration, frozen in liquid nitrogen, and stored at Ϫ80°C.
For each of the TGA1a mutants, the carboxyl-terminal primer used to make the final construct contained a TAA stop codon in-frame after the last TGA1a residue. All PCR reactions were run as described, and the resultant products were treated with proteinase K, digested with EcoRI and ClaI, agarose gel-purified, and ligated into cognate restriction sites in KS-FLAG vector (12). The TGA1a constructs were sequenced and tested for their ability to express the expected FLAG epitope-tagged protein using a coupled in vitro transcription/translation system (Promega). The FLAG-tagged cDNA was excised with BamHI and ClaI, blunt-ended with Klenow enzyme and deoxynucleotide triphosphates, and subcloned into the plant expression vector pMON999 as described previously (12).

Protoplast Transient Transfection Assay
Protoplasts were prepared from BY-2 suspension culture cells and transfected with the indicated plasmid reporter and effector DNAs as described previously (12). Transfected protoplasts were collected by brief centrifugation and then treated with lysis buffer according to the manufacturer's instructions (Promega). Lysates were assayed for reporter gene activity by luciferase or chloramphenicol acetyl transferase (CAT) enzyme assays (12,16). To correct for differences in transfection efficiency, reporter gene activity of -90-CAT was normalized to that of CHS-LUC, a luciferase reporter gene that is transcribed from a bean chalcone synthase promoter (17). Data shown are the mean and standard error from three or more independent experiments.

Nuclear Run-on Assay
Six days after transfer into fresh medium, BY-2 cells were incubated with 100 M 2,4-D in 0.1% ethanol carrier for 0 -8 h, collected by vacuum filtration, frozen in liquid nitrogen, and stored at Ϫ80°C. Nuclei were isolated essentially as described by Dröge-Laser et al. (18); nascent transcripts from these nuclei were radiolabeled as described by Lawton and Lamb (19) and isolated according to the work of McKnight and Palmiter (20). Labeled transcripts (ϳ4 ϫ 10 9 cpm) were hybridized as described by Lawton and Lamb (19) against Nytran-immobilized full-length cDNAs (200 ng) of the tobacco GNT35 and TGA1a genes. Radioactivity associated with hybridized transcripts was quantified with a PhosphorImager (Molecular Dynamics).

Preparation of Nuclear Extracts
BY-2 cells were ground with a mortar and pestle to a fine powder under liquid nitrogen. Nuclear proteins were extracted from this material as described previously (7), followed by dialysis against two changes of 500 ml of dialysis buffer (20 mM HEPES, pH 7.9, 25 mM NaCl, 1 mM EDTA, 20% (v/v) glycerol, and 1 mM dithiothreitol). Dialyzed extracts were clarified at 5,000 ϫ g for 5 min, divided into aliquots, frozen in liquid nitrogen, and stored at Ϫ80°C.

Preparation of Recombinant TGA1a
Recombinant TGA1a was synthesized using a coupled TnT/T7 transcription/translation system (Promega) with pBluescript KS vector containing full-length cDNA of TGA1a as template (12).

DNA-affinity Chromatography
Preparation of Labeled Nuclear Extracts-BY-2 cells cultured for 6 days were incubated with 200 Ci/ml [ 35 S]methionine (6000 Ci/mmol; PerkinElmer Life Sciences) for 24 h and then treated for 1 h with either 100 M 2,4-D in 0.1% ethanol carrier (xenobiotic stress) or 0.1% ethanol carrier alone (mock). Labeled nuclear protein extracts were prepared as described above.
Preparation of DNA-affinity Resins-Oligonucleotide concatamers of wild-type or mutant as-1 sequences were coupled to cyanogen bromideactivated Sepharose according to the manufacturer's instructions (Amersham Pharmacia Biotech). The coupling efficiency of these oligonucleotides to Sepharose was ϳ4.5 g DNA/ml resin. For each binding reaction, 125 ng of immobilized DNA were used. Immobilized DNA was equilibrated in binding buffer (20 mM HEPES, pH 7.9, 100 mM NaCl, 0.2 mM EDTA, 25 g/ml poly(dI⅐dC), 0.5 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, and 20% (v/v) glycerol) for 15 min on ice.
Assay Conditions-Wild-type or mutant immobilized DNAs were then added to nuclear extracts (1.0 ϫ 10 6 cpm incorporated [ 35 S]methionine) from BY-2 cells treated for 1 h with either 100 M 2,4-D in 0.1% ethanol carrier (xenobiotic stress) or 0.1% ethanol carrier alone (mock). After incubation on ice for 1 h, immobilized DNA-protein complexes were washed repeatedly with binding buffer. Bound proteins were then eluted with RIPA (21) and analyzed for the presence of TGA1a by immunoprecipitation.

Immunoprecipitation
Protein extracts were clarified by brief centrifugation at 14,000 ϫ g and incubated on ice for 1 h with 5 l of either preimmune sera or immune sera prepared against NT residues 57-72 of TGA1a (11). Immune complexes were recovered with 10 l of Gammabind Plus Sepharose (Amersham Pharmacia Biotech) in RIPA buffer solution and 2% bovine serum albumin (Sigma) during gentle mixing for 30 min on ice. Bound protein was resuspended in 25 l of Laemmli loading buffer, denatured by boiling, fractionated by SDS-PAGE on a 10% running gel, and detected by fluorography with En 3 hance according to the manufacturer's instructions (PerkinElmer Life Sciences).

GST Binding Assay
GST Fusion Proteins-PCR amplification of TGA1a and its derivatives was done using the following sets of primers and KS-TGA1a as template (12): (a) for TGA1a-CT, 5Ј-ACTTGGGAATTCTCTTCAACG-TACACCCAATTT-3Ј and 5Ј-TTTATCGATGTCAGGTAGGCTCACGT AGACG-3Ј; and (b) for TGA1a-NT, 5Ј-TTTGAATTCTCTCGTCGTG-CATCTGTTAATTCTTCAACGTAC-3Ј and 5Ј-TTTATCGATGTCAT-TCATATCTGTTAGAAGT-3Ј. The PCR products were gel-purified, digested with EcoRI and ClaI, and subcloned into pBluescript KS ϩ . After amplification in the XL-1 blue strain of Escherichia coli, plasmid DNA was isolated and digested with EcoRI and XhoI. Inserts were sequenced to confirm their identity and subcloned into pGEX-4T1 (Amersham Pharmacia Biotech). GST fusion proteins were expressed and isolated from an E. coli BL-21 strain according to the manufacturer's instructions (Amersham Pharmacia Biotech).
GST Binding Conditions-Reactions (40 l) contained Binding buffer (20 mM HEPES, pH 7.9, 25 mM NaCl, 1 mM EDTA, 20% (v/v) glycerol, 0.1 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol), 2 ϫ 10 8 cpm of labeled nuclear protein extract, and 10 g of GST fusion protein on glutathione-Sepharose resin (Pharmacia). After gentle mixing for 1 h, the Sepharose resins were recovered by brief centrifugation at 2000 g and repeatedly washed with Binding Buffer. Bound proteins were eluted with Laemmli loading buffer and fractionated by SDS-PAGE. Staining with Coomassie blue was used to detect GST proteins. Labeled proteins in gels were detected by fluorography using En 3 hance according to the manufacturer's instructions (NEN), with an intensifying screen at-80°C and 3-7 days exposure to x-ray film (Kodak).

As-1-dependent Activation by TGA1a Is Repressed through
Its CT Domain-Our previous work using chimeric Gal4-binding proteins led to the suggestion that the CT domain of TGA1a might function as a stimulus-reversible repressor of transcription. To further test this notion, we examined here whether removal of the regulatory CT domain from TGA1a enhanced its as-1-dependent transcriptional activity. Effector constructs of wild-type and mutant forms of TGA1a were "epitope-tagged" with the octapeptide FLAG epitope at their amino termini ( Fig.  1) to facilitate immunological monitoring of steady-state amounts of these proteins. Gel-shift binding assays revealed that a mutant TGA1a factor, termed TGA1a⌬CT, which lacks the entire CT domain (residues 142-373), was unable to bind as-1 (Fig. 2). This loss of function was due to the fact that CT residues appear to be essential for promoting formation of the DNA-binding dimer of TGA1a, as suggested by Katagiri et al. (22). Because the leucine zipper of TGA1a alone was unable to efficiently confer dimer formation, we pursued an alternative strategy to distinguish between dimer stabilization and repressor functions of the CT domain. This strategy involved exchanging the three heptad repeats that comprise the leucine zipper region of TGA1a⌬CT for three heptad repeats from the leucine zipper of the mammalian bZIP factor CREB. The resultant chimeric protein, TGA1a-CREB⌬CT, bound as-1 as well as wild-type TGA1a (Fig. 2, lane 3). As a control, we also tested whether the same leucine zipper exchange made in full-length TGA1a affected as-1-dependent binding by this factor. Results of gel-shift binding assays indicate that this protein, TGA1a-CREB, bound as-1 to a degree that was equal to or greater than that of TGA1a (Fig. 2, lane 4). These data imply that the inability of the leucine zipper of TGA1a to efficiently promote as-1 binding is due to its weaker dimerization potential compared with that of a similar length region of the CREB zipper.
Having ascertained that the modified TGA1a factors were capable of binding as-1, we next examined their effect on as-1dependent transcription. We found that TGA1a-CREB⌬CT, which lacks the regulatory CT domain, significantly enhanced as-1-dependent transcription in the absence of xenobiotic stress, i.e. under basal conditions (Fig. 3). In contrast, neither TGA1a nor TGA1a-CREB had significant effects on basal transcription, even when increasing amounts of their respective effector genes were tested (data not shown). TGA1a⌬CT, which lacks the ability to bind as-1, as expected, had little or no effect on transcription through this element.
The results above, however, might have arisen from differences in the steady-state amounts of these TGA1a proteins in transfected protoplasts. To test this possibility, we incubated transfected protoplasts with [ 35 S]methionine and prepared nu-  142-373). The position of the three heptad repeats that comprise the leucine zipper is indicated by the solid letter L. Effector constructs: cDNAs encoding for wild-type and modified TGA1a proteins fused downstream of the octapeptide FLAG epitope (Kodak) were subcloned into the pBluescript KS ϩ vector for sequencing and for in vitro protein synthesis. In both TGA1a-CREB⌬CT and TGA1a-CREB, the three heptad repeats that comprise the leucine zipper of TGA1a were exchanged for three heptad repeats from the leucine zipper of CREB (indicated by the open letter L). For plant protoplast expression studies, the cDNAs were excised from the pBluescript vector and inserted as shown between the CaMV 35S promoter and nos 3Ј polyadenylation/termination sequences in the plant expression vector pMON999. Reporter constructs: a single copy of as-1 (indicated by two tandem TGAC motifs) was placed just upstream of the TATA box and the start site of the CaMV 35S minimal promoter. This promoter fragment was then ligated upstream of the CAT reporter gene. A DNA fragment of sequences Ϫ326 to ϩ180 bp of the bean chalcone synthase promoter was ligated upstream of the LUC reporter gene (CHS-LUC) and served as an internal control for transfection efficiency.
clear extracts for immunoprecipitation analysis. Using this approach, we found that steady-state amounts of these TGA1a factors were generally similar among transfected protoplasts (Fig. 4). Thus, differences in the activities of these factors were not likely due to their relative abundance.
We were also interested in determining whether accessory proteins with potential regulatory activity interact with TGA1a because we previously observed that a 120-kDa protein was bound to transiently expressed TGA1a. 2 Immunoprecipitation assays here confirmed this observation (Fig. 4, lane 2). We also noted that TGA1a-CREB⌬CT, which lacks the regulatory CT domain of TGA1a, was not associated with the 120-kDa protein. This loss of activity was not due to the presence of the CREB leucine zipper in TGA1a-CREB⌬CT because the TGA1a-CREB control factor, which contains the same zipper substitution, bound the 120-kDa protein as well as TGA1a. These observations prompted us to test whether the CT domain alone can recruit this protein.
The CT Domain of TGA1a Is Necessary and Sufficient for Stimulus-responsive Recruitment of a 120-kDa Nuclear Protein-Results here indicate that the CT domain of TGA1a in-hibits trans-activation under basal conditions. This is consistent with our earlier work showing that this domain confers transcription in response to xenobiotic stress by a mechanism involving stimulus-induced de-repression (12). We thus examined here whether this regulatory domain is necessary and sufficient for recruiting the TGA1a-binding protein. In this experiment, radiolabeled nuclear proteins were incubated with GST fusion proteins of NT or CT domain polypeptides of TGA1a, washed to remove unbound material, fractionated by SDS-PAGE, and detected with fluorography (Fig. 5a). Due to its high degree of insolubility, recombinant GST-TGA1a was not tested with this assay. Comparable amounts of each GST fusion protein were programmed in the binding reactions, as evidenced by staining of the SDS-PAGE gel with Coomassie Blue (Fig. 5b). Results show that GST-CT, but not GST or GST-NT, bound a 120-kDa protein from nuclear extracts of mock-treated cells. This interaction was not detected with extracts from cells that had been treated for 0.5 h with 2,4-D, suggesting that xenobiotic stress affects recruitment by TGA1a of a 120-kDa nuclear protein.
Binding by TGA1a to a 120-kDa Protein Is Inversely Correlated with Stimulus-induced Changes in the Rate of as-1-dependent Transcription-The results above showed that recruitment of a 120-kDa nuclear protein in vitro by recombinant CT domain protein was responsive to xenobiotic stress. Here we examined whether changes in TGA1a activity, and in its association with the 120-kDa protein, are correlated. Because timecourse studies are difficult to perform with transient transfection systems, we used an alternative approach to address this question. Nuclear extracts were prepared from [ 35 S]methionine-labeled suspension cells that had been treated for 0 -8 h with 100 M 2,4-D. TGA1a in these extracts was recovered by immunoprecipitation with anti-TGA1a peptide (residues 57-72) antibodies that are monospecific for TGA1a (11). We note that although the calculated mass of TGA1a is 41 kDa, plant transiently expressed and recombinant forms of this factor appear to migrate as 46-and 47-kDa proteins, respectively, on SDS-PAGE gels (12). This small difference in mobility is thought to arise from differences in the posttranslational state of these factors. As expected, control reactions with preimmune sera did not bind either recombinant TGA1a (data not shown) or a nuclear protein with an apparent mass of 46 kDa, which is similar to that of TGA1a (Fig. 6). In contrast, immune sera bound recombinant TGA1a (data not shown; Ref. 11) and a 46-kDa nuclear protein that migrated slightly faster than recombinant TGA1a by SDS-PAGE electrophoresis. Immunoprecipitation of the 46-kDa protein was blocked (data not shown) by addition of the TGA1a peptide epitope. Immunospecificity of anti-TGA1a antibodies was also confirmed using a similar NT peptide from PG13, a tobacco TGA factor that shares a strong (Ͼ85%) overall identity with TGA1a. Although the PG13 peptide has 11 of 15 residues identical to those of the TGA1a peptide epitope, the PG13 peptide had no effect on immunoprecipitation of TGA1a. These results indicate that the 46-kDa protein in nuclear extracts is identical in both apparent mass and antigenic properties to TGA1a.
The relative amount of TGA1a that was recovered by immunoprecipitation from extracts of cells treated for 0 -8 h with 2,4-D varied little between samples, consistent with the results of earlier studies of TGA1a in planta (11). Also, a 120-kDa nuclear protein was recovered with TGA1a in nuclear extracts of mock-treated (i.e. 0 h) cells. This association was not observed in extracts obtained from cells treated for 0.5-2 h with 2,4-D. Longer treatments with 2,4-D of 4 and 8 h, respectively, led to a partial or complete restoration of the association between TGA1a and the 120-kDa protein. Other proteins with molecular masses that were slightly smaller or larger than that of the 120-kDa protein might reflect either its posttranslational modification or proteolysis or might simply reflect the presence of unrelated proteins.
We next examined whether this transient association between TGA1a and the 120-kDa protein correlates with changes in the rate of as-1-dependent transcription (Fig. 7). To this end, we used a nuclear run-on assay to monitor the rate of accumulation of de novo transcripts from GNT35, a tobacco GST gene whose expression is both enhanced by 2,4-D through as-1 and preferentially localized in planta with that of TGA1a (5,8,11). Tobacco cells treated with 2,4-D showed a strong increase in GNT35 transcription by 0.5 h, with maximal activity occurring by 2 h. Longer treatments with 2,4-D resulted in a rapid decline in the rate of GNT35 transcription. Because expression of TGA1a has been found to be unaffected by xenobiotic stress (11), de novo transcription from this gene served as the internal control. We observed that the rate of transcription of TGA1a was initially similar to that of GNT35 and unaffected by 2,4-D. Collectively, these findings indicate that changes in the rate of transcription of GNT35 were inversely correlated with the degree of association between TGA1a and a 120-kDa nuclear protein.
Stimulus-induced Changes in the as-1 Binding Activity of TGA1a-One means by which xenobiotic stress might affect the activity of TGA1a is by inhibiting its ability to bind as-1. However, gel-shift experiments with nuclear extracts from mock-or 2,4-D-treated cells showed comparable amounts of a nuclear as-1 binding activity, termed "ASF-1" (data not shown). ASF-1 from these suspension cells is derived from the contributions of a number of different TGA factors (e.g. TGA2.1 and TGA2.2) that are more abundant than TGA1a (23). To therefore determine the specific contribution of the comparatively small amount of TGA1a to this activity, we used a sequential enrichment procedure involving DNA-affinity chromatography and anti-TGA1a immunoprecipitation (Fig. 8a). Results indicate that TGA1a from mock-treated cells failed to bind immobilized as-1 DNA (Fig. 8b, lane 4), unlike the activity of TGA1a from 2,4-D-treated cells (Fig. 8b, lane 10). Based on results with a mutant as-1 sequence, TGA1a binding was specific for as-1 (Fig. 8b, lanes 6 and 12). As expected, control reactions with preimmune sera did not recover detectable amounts of TGA1a from either of these extracts.
Interestingly, crude input fractions from mock-treated cells revealed the presence of a 120-kDa nuclear protein bound to TGA1a (Fig. 8b, lane 2), whereas TGA1a from the input nuclear fraction from 2,4-D-treated cells was not associated with this protein (Fig. 8b, lane 8). Based on the relative amounts of TGA1a in lanes 8 and 10 of Fig. 8b, we estimate that nearly all of the TGA1a present in the input fraction was recovered using as-1 DNA-affinity chromatography. The above-mentioned findings contrast sharply with the apparent inability of the TGA1a-120-kDa protein complex to bind as-1 (Fig. 8b, lane 4). DISCUSSION Plant transcription involving as-1-type cis-elements and their cognate TGA factors is activated by diverse cues including defense hormones, wounding, and xenobiotic stress. Of the known TGA factors, tobacco TGA1a is among the best characterized with regard to its contribution to transcription. TGA1a promotes as-1-dependent transcription in vitro (24) and as a transiently transfected chimeric factor in response to xenobiotic stress (12). In tobacco seedlings, TGA1a is preferentially coexpressed in root tip meristem cells along with several as-1regulated genes, including GNT35, whose activities are further enhanced by xenobiotic stress (11). Gain of function studies with a heterologous system involving the yeast Gal4-binding protein indicate that the CT domain of TGA1a confers transcription through the GAL4 cis-element in response to xenobiotic stress through a mechanism involving a change in transactivation potential (12). Evidence here significantly extends our understanding of how TGA1a is regulated by showing that the CT domain and xenobiotic stress play additional regulatory roles in TGA1a activity.
Transient transfection assays with TGA1a-CREB⌬CT have demonstrated here an inhibitory role for the CT domain in TGA1a trans-activity. The presence of the leucine zipper of CREB in TGA1a-CREB⌬CT promoted efficient dimerization and DNA binding in the absence of a "dimer stabilization" function conferred by the missing CT domain (22). This modification allowed us to identify separate contributions made by dimer stabilization and repression activities of the CT domain to as-1-dependent transcription. Enhanced trans-activity by TGA1a-CREB⌬CT was shown to be largely due to the absence of the inhibitory CT domain and not due to the presence of the leucine zipper of CREB, as evidenced by the relatively poor trans-activity of the TGA1a-CREB control. In addition, immu- noprecipitation assays indicated that the enhanced activity of TGA1a-CREB⌬CT was not due to significant comparative differences in its steady-state concentration, a finding that is further supported by our observations that further increasing expression of TGA1a or TGA1a-CREB had little additional stimulatory effect on transcription. We conclude from these data that the CT domain represses the basal activity of TGA1a.
During these studies, we observed that a 120-kDa protein bound TGA1a and that this interaction is responsive to xenobiotic stress. Evidence from several lines of investigation suggests that this TGA1a-binding protein may be a putative corepressor. First, its association with TGA1a was transiently affected by xenobiotic stress to a degree that was inversely correlated with the rate of expression of a TGA1a target gene, suggesting an inhibitory function. Second, binding by this 120-kDa protein to TGA1a specifically involved the regulatory CT domain of this factor, which mediates transcription in a heterologous system in response to xenobiotic stress. Deleting this domain from TGA1a abolished both its association with the 120-kDa protein and its ability to repress basal transcription activity, thus providing support for the notion that this TGA1abinding protein has potential inhibitory activity.
In addition to the TGA1a-binding protein identified here, several other plant proteins have been shown to bind TGA factors. A ϳ26-kDa Arabidopsis protein termed OBP1 associates with a subset of Arabidopsis TGA factors to facilitate their as-1 binding activity. OBP1 itself is a DNA-binding factor and binds to AAGG motifs in as-1-regulated promoters (25,26). A second Arabidopsis protein of ϳ66 kDa termed NPR1 (or NIM1) is a positive regulator of the as-1-regulated PR1 gene in response to pathogens and plant defense cues, such as salicylic acid. Like OBP1, the NPR1 protein binds to a subset of TGA factors (27)(28)(29). Mutations in NPR1 that disrupt this binding activity also impair salicylic acid-induced expression of PR1, thus linking NPR1 activity to as-1. Intriguingly, Zhang et al. (27) have suggested a model whereby NPR1 potentiates transcription by reversing the binding of a repressor to one or more TGA factors. The apparent mass of the 120-kDa protein and its proposed mechanism of action appear to distinguish it from either OBP1 or NPR1.
Results from this study indicate that TGA1a belongs to a subset of bZIP transcription factors whose activities are regulated through a mechanism involving stimulus-reversible repression. In the bZIP factor ATF-2, intramolecular binding under basal conditions occurs between the activation and bZIP domains and inhibits this factor's DNA binding and transactivation potential (30). In vivo, these inhibitory interactions in ATF-2 are alleviated by the coactivator CBP (31). A different type of regulatory mechanism occurs with CCAAT/enhancerbinding protein, a bZIP factor that negatively regulates its trans-activation potential via intramolecular interactions between activation and repressor domains through a cellular repressor protein (32). Transcription factors of the estrogen hormone receptor class represent yet another type of repression (33). When bound in the nucleus by the 90-kDa heat shock protein (hsp90), the estrogen receptor is unable to bind its cognate cis-element. Estrogens reverse this effect by inducing a conformational change in the receptor, thus promoting its release of hsp90 and subsequent binding to DNA.
Thus, how might xenobiotic stress affect the activity of TGA1a? One likely mechanism suggested here involves stimulus-induced changes in as-1 binding by TGA1a, perhaps through its stimulus-reversible association with a putative corepressor protein. Despite the fact that the 120-kDa protein bound to TGA1a in nuclear extracts from mock-treated cells, we were unable to detect this complex by gel-shift binding assay, suggesting that this complex may not bind DNA. This view is directly supported by evidence from an alternative TGA1a detection assay involving sequential DNA-affinity chromatography and immunoprecipitation. By this approach, xenobiotic stress was found to potentiate the as-1-binding activity of TGA1a, presumably by inducing a change in this factor's association with a 120-kDa protein. However, knowledge of the proportion of cellular TGA1a bound by this protein will be necessary before definitive conclusions regarding the regulatory contribution of this interaction to as-1 binding by TGA1a can be made.
Based on present and previous findings, we propose the following working model to explain how TGA1a affects as-1-dependent transcription. In the absence of xenobiotic stress, as-1 binding and trans-activation functions of TGA1a are inhibited. TGA1a's loss of basal transcription activity occurs through this factor's inhibitory CT domain, by a mechanism that may involve its stimulus-reversible interaction with a putative corepressor protein. In response to xenobiotic stress, TGA1a transactivation and as-1 binding activities are enhanced. These changes occur in parallel with the release of the putative corepressor protein. Longer exposure to xenobiotic stress promotes reassociation of TGA1a with this protein and a concomitant decline in as-1-dependent transcription.