Auxin-induced Stress Potentiates trans-activation by a Conserved Plant Basic/Leucine-zipper Factor*

The promoter element activation sequence-1 (as-1) confers tissue-specific and signal-responsive transcription in plants. Hormone and chemical stress cues are thought to activate as-1-dependent transcription through specific basic/leucine-zipper proteins, termed TGA factors, that bind this element. We report here that a highly conserved TGA factor of tobacco, TGA1a, can selectively activate transcription in response to micromolar concentrations of auxin hormones or their analogs. This induction is chemically specific, as a range of other compounds tested at similar concentrations had little or no effect. Auxin was found to augment the trans-activation potential of TGA1a through carboxyl-terminal residues. The amino-terminal domain of TGA1a, by gain-of-function assays, was found to both constitutively activate transcription and maximize the response to auxin. Further evidence indicates that the trans-activation potential of this domain in TGA1a is repressed, under basal conditions, by carboxyl-terminal residues. Because TGA1a and endogenous TGA factors are stimulated by auxin only at concentrations that inhibited cell growth, this response is likely to involve chemical stress.

confers tissue-specific and signal-responsive transcription in plants. Hormone and chemical stress cues are thought to activate as-1-dependent transcription through specific basic/leucine-zipper proteins, termed TGA factors, that bind this element. We report here that a highly conserved TGA factor of tobacco, TGA1a, can selectively activate transcription in response to micromolar concentrations of auxin hormones or their analogs. This induction is chemically specific, as a range of other compounds tested at similar concentrations had little or no effect. Auxin was found to augment the transactivation potential of TGA1a through carboxyl-terminal residues. The amino-terminal domain of TGA1a, by gain-of-function assays, was found to both constitutively activate transcription and maximize the response to auxin. Further evidence indicates that the trans-activation potential of this domain in TGA1a is repressed, under basal conditions, by carboxyl-terminal residues. Because TGA1a and endogenous TGA factors are stimulated by auxin only at concentrations that inhibited cell growth, this response is likely to involve chemical stress.
The expression of nuclear genes in response to cellular and environmental cues is largely regulated by trans-acting factors and their cognate cis-elements. In plants, auxin hormones (e.g., indole-3-acetic acid) promote growth and affect a number of cellular processes. Auxins are believed to mediate at least some of their effect by enhancing the expression of specific genes through target cis-elements (1,2). The prototype for one class of auxin-responsive element is activation sequence-1 (as-1), 1 which was first identified by Lam et al. (3) in the cauliflower mosaic virus (CaMV) 35 S promoter. Homologs of as-1 (e.g., ocs and nos), moreover, occur in the promoters of plant-transforming T-DNA genes of agrobacteria (4 -6). Apart from regulating these genes in plants, a broader role for as-1 elements has been suggested by their functional presence in plant glutathione S-transferase genes (7)(8)(9). For example, auxin-responsive expression of at least one member of the tobacco GST gene family, GNT35, is mediated by as-1 (8,9). Unlike other auxin-responsive elements, as-1 requires physiologically high (i.e. micromolar) concentrations of auxins to activate transcription, suggesting that this response may involve chemical stress. This view is further supported by the observation that biologically inactive analogs of auxins are also strong inducers of as-1-dependent activity.
Multiple TGA factors that bind selectively to as-1 and related motifs have been cloned from several plant species, and some or all of these factors are likely to mediate as-1-dependent transcription. Based on their amino acid homology, at least three subclasses of TGA factors occur in plants, with multiple members in each class. Heterodimerization between these factors may generate new combinations with potentially distinct functional properties (10). Posttranscriptional regulation may also determine the contribution of each factor to transcription (11). To date, the only TGA factor that is known to activate transcription is TGA1a of tobacco (12). Although originally suggested to be a plant homologue of CREB, TGA1a lacks several distinguishing features of this animal bZIP transcription factor (13,14).
Although little is known about how and where TGA1a functions in plants, its evolutionary conservation implies an important regulatory role in gene expression. To test this view, trans-dominant inhibitors of TGA1a and of its homologue PG13 have been developed and expressed in plants (10,15). These inhibitors are believed to act by forming inactive "heterodimers" with their endogenous counterparts, thus markedly reducing their availability and activity. Although the PG13 and TGA1a inhibitors decreased as-1 binding activity, they had little or no effect on either plant growth or development. One explanation for these findings is that the residual as-1 binding activity that survives inhibition is sufficient for normal functions, or that as-1 acts in conjunction with other cis-elements in most cellular promoters. Recent efforts to completely disrupt the functions of individual TGA factors by homologous recombination may provide alternative approaches for investigating in planta contributions by these factors (16).
Work described here is part of an ongoing effort to elucidate the molecular mechanism(s) by which as-1-dependent transcription is regulated in response to environmental cues. Toward this goal, we chose to study the functional properties of TGA1a as it is highly conserved in plants, suggesting an important role in as-1-dependent transcription. In addition, TGA1a is a member of a large group of bZIP transcription factors, the activity of which in many cases is governed by cellular and environmental cues (17)(18)(19). To investigate functional properties of this tobacco factor, we employed a homologous transient transfection assay with protoplasts, an approach that has been previously used to characterize as-1-dependent transcription (20). By co-transfection and gain-offunction assays, we show here that trans-activation by TGA1a is stimulated by auxins and their analogs via chemical stress.

EXPERIMENTAL PROCEDURES
Construction of Effector and Reporter Plasmids-An oligonucleotide containing the Ϫ90 to ϩ4 bp region of the CaMV 35 S promoter was ligated into HindIII and XbaI sites of pUC19 containing bacterial chloramphenicol acetyltransferase (CAT) coding sequence (21) to create the Ϫ90-CAT reporter. The Ϫ76-CAT reporter was constructed by ligating a Ϫ76 to ϩ4 bp PCR fragment of the 35 S promoter through flanking HindIII and XbaI restriction sites into pUC19-CAT vector (21). Other reporter plasmids used were CHS-CAT (pCHC1-CAT; Ref. 24) and GAL-CAT (22). Effector cDNAs were derived from PCR-amplified products of a TGA1a cDNA clone (13). In each case, the 5Ј amino terminus of the cDNA was engineered by PCR to contain the Kozak consensus start site for translation (23) and FLAG epitope (Eastman Kodak Co.), and the 3Ј-terminus contained a UAA stop codon. DNA fragments containing amino (amino acids 1-64) and carboxy (amino acids 143-373)-terminal sequences of TGA1a, as well as a mutant version of TGA1a (TGA1a⌬bZIP) that lacks the bZIP domain (amino acids 86 -142), were generated by PCR amplification with primers that contained unique EcoRI and ClaI restriction enzyme sites. PCR products were purified, digested with these restriction enzymes, and subcloned into the Bluescript SKII(ϩ) vector (Promega). GAL4 chimeric factors were created with an EcoRI -containing PCR product that encodes the first 147 amino acids of the yeast GAL4 factor. This fragment was inserted in-frame into the EcoRI site adjacent to the FLAG epitope. Effector constructs for transient transfection studies were sequenced and analyzed for expression of full-length protein with a coupled transcription/ translation kit (Promega). pMON effector DNAs were made by excising sequences for the FLAG-epitope tagged factors with BamHI and ClaI, filling in the ends with Klenow DNA polymerase, and ligating the blunt-ended fragments into the vector pMON 999 at Klenow-modified BglII and EcoRI sites.
Plant Cell Cultures-Tobacco variety Bright Yellow 2 (BY-2) suspension cells were maintained at 26°C in 0.5ϫ Murashige and Skoog's medium supplemented with 3% sucrose, B 5 vitamins, and 0.9 M of 2,4-dichlorophenoxyacetic acid (2,4-D) as described (24). Cells were subcultured weekly by transferring 10 ml of a mid-log phase culture to 100 ml of fresh medium. The packed cell volume of the suspension culture at transfer was ϳ30%.
Preparation of Tobacco Protoplasts-For transfection experiments, BY-2 cells were cultured for 4 days, harvested by centrifugation at 500 ϫ g for 5 min, washed twice with auxin-free medium, and incubated in the same medium for 2 days. To prepare protoplasts, ϳ 20 ml (packed volume) of BY-2 cells were washed twice with 0.4 M mannitol and incubated in 50 ml of 1% (w/v) Cellulase Onozuka RS and 0.1% Pectolyase Y-23 (Karlan) in 0.4 M mannitol, pH 5.5, for 60 -90 min at 28°C. Every 30 min of incubation, the cells were gently aspirated and expelled with a 25-ml wide-mouth pipette to promote cell wall digestion. Protoplasts were collected by centrifugation as described above and washed several times with 5 volumes of 0.4 M mannitol. The concentration of protoplasts was determined with a hemocytometer, and their viability was ascertained by vital fluorescent staining with fluorescein diacetate. Just prior to electroporation, protoplasts were equilibrated in electroporation buffer (5 mM morpholinoethanesulfonic acid, pH 5.8, 70 mM KCl, and 0.3 M mannitol) and placed on ice.
Transient Transfection of Protoplasts-Reporter and effector plasmid DNA were prepared by standard alkaline lysis and cesium chloride gradient purification or with a plasmid isolation kit (Qiagen). Approximately 2 ϫ 10 6 viable protoplasts were transfected with an electroporator (Bio-Rad) as described (24) with 5 g of the reporter gene and 0 -80 g of effector plasmids plus pMON 999 to 85 g of total DNA. Transfected protoplasts were incubated in protoplast culture medium at 26°C for 20 h with 0.1% EtOH as a control, or with 50 M of either auxin, salicylic acid, or methyl jasmonate in 0.1% EtOH. Auxins and other chemicals were from the following sources: Life Technologies, Inc., 2,4-D and ␣-napthylene acetic acid; ICN, 2,3-dichlorophenoxyacetic acid (2,3-D); Sigma, salicylic acid (SA); and Serva, methyl jasmonate (MJ). Benzyl isothiocyanate, dimethyl fumarate, and hydroquinone were provided by Paul Talalay (Johns Hopkins University Medical School).
CAT Assay-Protoplast lysates were prepared in CAT Buffer A (25) as described previously (21). Protein in these lysates was quantified with Bradford reagent (Bio-Rad), and 1-20 g was used for determining CAT activity by a radioanalytic method (25). In this assay, [ 14 C]chlor-amphenicol was used as substrate and was separated from mono-and diacetylated reaction products by thin-layer chromatography (TLC) with 5% methanol and 95% chloroform. Labeled substrate and products were quantified following TLC with a PhosphorImager (Molecular Dynamics). The percentage of conversion of [ 14 C]chloramphenicol to its acetylated forms was used to calculate CAT activity. The mean and S.E. of results of Ն3 independent experiments were determined to evaluate the reproducibility of transient transfection results.
Western Blots-Twenty hours posttransfection, 2 ϫ 10 6 protoplasts were harvested by centrifugation at 500 ϫ g and lysed in one volume of 2ϫ Laemmli loading buffer. Protein samples were fractionated by SDSpolyacrylamide gel electrophoresis (PAGE), electrophoretically transferred to nitrocellulose, and blocked for 30 min with 3% bovine serum albumin in Tris-buffered saline plus 0.5% Tween-20 (TTBS). The blots were subsequently incubated with 1 g/ml of anti-FLAG (M2) monoclonal antibody (Kodak) in 1% bovine serum albumin in TTBS, washed three times for 10 min each with TTBS, and then incubated with a 1:2000 dilution of a goat anti-mouse antibody conjugated to alkaline phosphatase (Bio-Rad). After washing three times for 10 min each with TTBS and once for 10 min in Tris-buffered saline, the blots were developed using a colorimetric procedure for alkaline phosphatase according to the manufacturer's instructions (Bio-Rad).
In Vivo Labeling and Immunoprecipitation-Transfected protoplasts (ϳ2 ϫ 10 6 ) were incubated in 2 ml of protoplast culture medium containing 200 Ci of [ 35 S]methionine (Ͼ1000 Ci/mmol) for 20 h. Protoplasts were collected by brief centrifugation at 4000 ϫ g, washed with 1 ml of protoplast culture medium, and resuspended in 200 l of ice-cold RIPA buffer (26) containing 1 g/ml of phenylmethylsulfonyl fluoride, pepstatin A, aprotinin, and leupeptin (Sigma). Samples were resuspended for 30 s with a vortex mixer, incubated on ice for 30 min, and pelleted by centrifugation. The supernatant was then transferred to a new Eppendorf tube and incubated with 2 g of anti-FLAG (M2) monoclonal antibody on ice for 1 h, followed by incubation on ice with 10 l (packed volume) of GammaBind Sepharose (Amersham Pharmacia Biotech) at 4°C for 30 min with gentle shaking. The beads were recovered by centrifugation, washed three times with 1 ml each of RIPA buffer, and resuspended in 20 l of Laemmli loading buffer (Bio-Rad). The protein samples were fractionated by SDS-PAGE. Gels were treated with Enhance, as described by the manufacturer (NEN Life Science Products), and then analyzed by fluorography.
Immunofluorescent Staining-After incubation for 20 h, transfected protoplasts were harvested by centrifugation at 500 ϫ g and fixed for 1 h at 20°C in 2% (w/v) paraformaldehyde in PBS containing 0.3 M mannitol. Fixed protoplasts were washed with PBS, collected on glass slides, and air dried. After blocking with 3% (w/v) nonfat powdered milk (Carnation) in PBS for 30 min, the protoplasts were incubated in 1 g/ml anti-FLAG (M2) monoclonal antibody in PBS overnight at 4°C, washed with PBS, and incubated with a 1:2000 dilution in PBS of donkey anti-mouse antibody conjugated to fluorescein (The Jackson Laboratory) for 2 h at 20°C. Protoplasts were washed with PBS, mounted in Fluoromount-G medium (Southern Biotechnology Associates), and photographed under UV light.

TGA1a
Potentiates Transcription through as-1-To investigate basal and signal-responsive properties of TGA1a, one or more effector and reporter genes ( Fig. 1) were transiently transfected into protoplasts of tobacco BY-2 suspension cells. Changes in reporter gene activity were monitored by analyzing the activity of the encoded CAT enzyme. To minimize potential effects of physiological heterogeneity on the results, only protoplasts from cells grown to mid-log phase were used in these studies. To evaluate whether as-1-dependent transcription is regulated by auxins in transfected BY-2 tobacco protoplasts, we fused a truncated version (Ϫ90-CAT) of the 35 S promoter of CaMV, which contains a single as-1 element centered Ϫ72 bp from the start site of transcription, to the CAT reporter gene. This truncated 35 S promoter has previously been shown to confer basal and auxin-induced transcription through as-1 in transgenic tobacco (9,27). To determine whether as-1 is required for transcription of the Ϫ90-CAT reporter, we studied the activity of a mutant version of this promoter (Ϫ76-CAT) that specifically lacks the upstream TGAC motif of as-1. Prior studies have shown that this mutation inhibits binding by TGA1a (11,13) and as-1-dependent transcription (3,9,27,28).
To determine the specificity of the transcriptional response to auxin, we analyzed the behavior of a chalcone synthase promoter (CHS-CAT), the activity of which in tobacco is regulated by elements other than as-1 (21,29).
As expected, expression of CHS-CAT (Fig. 2, lanes 1-3) and Ϫ76-CAT (lanes 4 -6) in transfected protoplasts was unresponsive to the synthetic auxin 2,4-D. In contrast, expression of Ϫ90-CAT, which contains as-1, was induced Ͼ7-fold by this compound (lanes 7-9). Having shown that auxin induces as-1dependent transcription in BY-2 protoplasts, presumably through endogenous TGA factors, we next examined whether TGA1a, when overexpressed, could enhance this transcriptional response. At the highest amount of effector DNA tested (80 g), the activity of Ϫ76-CAT was unresponsive to TGA1a (lanes 10 -12). In contrast, TGA1a enhanced basal and auxinresponsive expression of Ϫ90-CAT (lanes [13][14][15]. Although this factor had only a modest effect on Ϫ90-CAT basal activity (compare lanes 7 and 13), the response to auxin was augmented Ͼ600% by TGA1a (compare lanes 8 and 14), resulting in a higher overall fold induction (compare lanes 9 and 15). One possible explanation for the selective affect of TGA1a on auxin-responsive transcription might be that the synthesis or degradation of the factor is governed by auxin. To test this point, we analyzed the relative amounts of (FLAG epitopetagged) TGA1a from control and auxin-treated protoplasts. Western immunoblot analysis with an ␣-FLAG antibody of total protein extracts revealed that FLAG-TGA1a accumulated in transfected protoplasts to a similar degree in the absence and presence of auxin (data not shown). Alternatively, differences between basal and auxin-induced activities might reflect a change in the subcellular compartmentalization of TGA1a. To test this hypothesis, protoplasts were fixed and analyzed by immunofluorescent staining with ␣-FLAG antibody for the presence of FLAG-TGA1a. This factor was localized almost exclusively to nuclei, as determined by its co-localization with 4,6-diamidino-2-phenylindole (DAPI) fluorescent staining, in both control and auxin-treated protoplasts (data not shown).
Sequences That Flank the bZIP Domain of TGA1a Confer Basal and Auxin-responsive Activation-One means by which auxin could govern the expression of the Ϫ90-CAT reporter is by altering the affinity of TGA1a for as-1. If this were so, then . Acidic and glutamine-rich sequences and the dimer stabilization domain are as indicated. a, effector DNAs. cDNAs encoding the TGA1a proteins were inserted between the Ϫ645 to ϩ4 bp region of the CaMV 35 S promoter and the nos 3Ј-polyadenylation/termination sequences in pMON 999. The 5Јend of each cDNA was fused to a 30-bp sequence containing the consensus sequence for translational initiation (23) upstream of the FLAG octapeptide epitope (Kodak). Wild-type or mutant versions of TGA1a cDNA were also fused in-frame to a sequence that encodes for the 147-amino acid DNA binding domain of the yeast GAL4 transcription factor. b, reporter DNAs. All promoters were inserted upstream of a bacterial CAT gene and agrobacterial nos 3Ј-polyadenylation/termination sequences (21)(22). The following promoters, the nucleotide end points of which are indicated in the figure, were studied: a Ϫ90 (to ϩ4 bp) CaMV 35 S promoter, which contains a single as-1 element and TATA-box (3); a Ϫ76 (to ϩ4 bp) CaMV 35 S promoter, which lacks the upstream TGAC motif of as-1; CHS, a chalcone synthase promoter (Chs15) from frenchbean (21) that contains a single G-box, H-box, and TATA box element; and GAL, a synthetic promoter with nine GAL4 elements fused to the TATA-box of the Ϫ40 to ϩ4 bp CaMV 35 S promoter (22). TGA1a might not be expected to confer auxin-responsive activation through a heterologous DNA binding domain and target cis-element. To test this hypothesis, we fused TGA1a to a 147-amino acid region of the yeast GAL4 factor, which confers binding to the GAL4 element, homodimerization, and nuclear import (30). In tobacco protoplasts, as shown here (Fig. 3) and elsewhere (22), this truncated GAL4 factor is relatively inactive on a minimal promoter that contains multiple GAL4 ciselements (GAL-CAT), and was unaffected by auxin (Fig. 3,  lanes 1-3). By contrast, GAL4-TGA1a markedly enhanced the activity of GAL-CAT in auxin-treated, but not control, protoplasts. Because the basal activity of GAL4-TGA1a was below that of GAL4 alone, we surmised that the presence of multiple DNA binding, dimerization, and nuclear import sequences in the chimeric factor interferes with its activity, expression, or stability. This notion was tested by removing the bZIP domain of TGA1a to create GAL4-TGA1a⌬bZIP. In protoplasts, GAL4-TGA1a⌬bZIP showed significantly higher basal and auxin-responsive trans-activation than those seen with GAL4-TGA1a (Fig. 3, lanes 4 -5 versus lanes 7-8). In vivo labeling experiments confirmed that this difference is due, at least in part, to the higher accumulation of GAL4-TGA1a⌬bZIP versus GAL4-TGA1a (data not shown).
Basal trans-activation by TGA1a-By GAL4 gain-of-function assays similar to those described above, we identified the contribution of specific portions of TGA1a to basal and signalresponsive trans-activation. Earlier work has shown that the first 80 amino acids of TGA1a contain multiple acidic residues that are essential for trans-activation in vitro and in vivo (12,28). To confirm these findings and to determine whether amino-terminal residues contribute to the response to auxin, we fused the GAL4 binding domain to the first 86 amino-terminal (NT) residues of TGA1a to create GAL4-TGA1a-NT. This chi-meric factor stimulated GAL-CAT reporter activity to a similar degree in the presence or absence of 2,4-D (Fig. 3), implying that amino-terminal residues are not likely to be a direct regulatory target of auxin.
The fact that GAL4-TGA1a-NT showed considerable basal activity when compared with GAL4-TGA1a⌬bZIP suggested that the NT domain in the latter factor is latent or repressed. In vivo labeling of these two factors revealed that their relative accumulation was similar, either in the presence or absence of auxin (Fig. 4, lanes 3 and 4 versus lanes 7 and 8), at an effector concentration (40 g) that maximized GAL-CAT reporter activity (Fig. 5). Differences between the basal activities of GAL4-TGA1a-NT and GAL4-TGA1a⌬bZIP thus result from intrinsic transcriptional properties of these factors.
To further delineate sequences required for basal trans-activation, the carboxyl-terminal (CT) region (amino acids 142-373) of TGA1a was fused to GAL4 to create GAL4-TGA1a-CT. Basal activity of GAL4-TGA1a-CT was comparable to that of GAL4-TGA1a⌬bZIP (Fig. 3). However, correcting for the fact that GAL4-TGA1a-CT was expressed nearly 4 times more than GAL4-TGA1a⌬bZIP (Fig. 4), the basal activity of the CT domain is likely to be similar to background levels seen with GAL4 alone.
Results with the CHS-CAT reporter (Table I) showed that the GAL4 chimeric factors do not generally effect RNA polymerase II-dependent transcription or have cytotoxic effects. These data imply that GAL4 chimeric factors affected gene expression through the GAL4 element and not by an effect on CAT reporter enzyme activity or cell viability.
The Carboxyl-terminal Domain of TGA1a Mediates the Response to Auxin-trans-activation by GAL4-TGA1a-CT under basal conditions was stimulated Ͼ6-fold by auxin (Fig. 3). At saturating amounts of effector DNAs, GAL4-TGA1a-CT had less than half the activity of GAL4-TGA1a⌬bZIP (Fig. 5). This result was not simply due to relative differences in the accumulation of these factors, as the amount of GAL4-TGA1a-CT was ϳ4 times higher than that observed with GAL4-TGA1a⌬bZIP (Fig. 4) at the highest amount of effector DNA tested. These data imply that NT and CT domains in TGA1a may both be required for achieving maximal trans-activation by this factor in response to auxin.
Chemical Cues That Induce Transcription through as-1 and TGA1a-as-1 type elements confer gene expression in response to the phytohormones auxin, SA, and MJ, whereas other plant hormones, including the cytokinin benzyl-aminopurine and abscisic acid, are inactive (5,9,20,27,31). In contrast to these results, we found that trans-activation by GAL4-TGA1a⌬bZIP or endogenous TGA factors was induced by either natural (i.e. ␣-napthylene acetic acid and indole-3-acetic acid) or synthetic (i.e. 2,4-D) auxins to similar degrees, whereas SA and MJ were found to be relatively inactive (data not shown). We have no good explanation for these conflicting results or for the nature of the apparent defect in SA and MJ signaling in BY-2 protoplasts. These findings, however, are consistent with the notion that auxin may stimulate transcription through a signal pathway that is distinct from those used by SA and MJ.
It has been previously shown that as-1 and related elements confer transcription in response to biologically active and inactive analogs of auxin, suggesting that this response may involve chemical stress cues (8,20). To further investigate this possibility, we tested the ability of the biologically active auxin 2,4-D and its inactive analog 2,3-D to stimulate transcription from the Ϫ90-CAT reporter. We found that 50 M 2,3-D was ϳ60% as effective as 2,4-D at inducing as-1-dependent transcription or GAL4-TGA1a⌬bZIP trans-activation, thus suggesting that auxins stimulate these transcriptional responses through chemical stress.
Depending on their concentration, auxins can act as mitogens and promote cell growth or show herbicidal and inhibitory activity. To further investigate the nature of the auxin signal involved here, we tested whether 2,4-D stimulates transcription at concentrations that inhibit or promote cell growth. In these experiments, protoplasts transfected with Ϫ90-CAT, or with GAL-CAT and GAL4-TGA1a⌬bZIP, were incubated with 2,4-D over a range of concentrations. To determine the concentrations of 2,4-D that promote or inhibit growth, auxin-starved BY-2 cells were incubated with 2,4-D over a range of concentrations for 1 week, after which time the fresh weight of the cells was determined. The optimal growth of the culture under these conditions required 0.1 M 2,4-D, a concentration that has little effect on the activities of either the Ϫ90-CAT reporter or GAL4-TGA1a⌬bZIP (Fig. 6). Significantly, 2,4-D augmented transcription only at concentrations that inhibited cell growth, thus implying that these transcriptional responses to auxin are mediated by chemical stress. DISCUSSION In plants, as-1 and related elements have been shown to confer transcription in response to such diverse cues as auxins, salicylic acid, methyl jasmonate, heavy metals, wounding, and xenobiotic stress. Little is currently known about the individual contributions of as-1 binding TGA factors to these transcriptional responses. Of these factors, only TGA1a has been shown to activate plant transcription in vivo (28). Results here extend these findings by demonstrating that transiently expressed TGA1a can activate as-1-dependent transcription in response to auxin.
As in plants, auxin induces a change in the activity of TGA factors of BY-2 protoplasts to enhance transcription from a truncated Ϫ90 (to ϩ4 bp) 35 S promoter that contains a single as-1 element centered at Ϫ72 bp (3). Moreover, as-1 is the primary determinant of this truncated promoters activity in tobacco root and suspension culture cells (9,32). In tobacco leaves (27) and roots (data not shown), as-1 confers transcription from the Ϫ90 promoter in response to exogenous auxin. In BY-2 protoplasts, basal and auxin-responsive transcription from Ϫ90-CAT requires as-1, as evidenced by the fact that a mutant version of this promoter (Ϫ76-CAT) that lacks the upstream TGAC motif of this element was inactive (Fig. 2).
Having shown that the Ϫ90-CAT reporter gene was selectively activated by auxin, we next examined whether TGA1a might potentiate the activity of this promoter. Although transiently expressed TGA1a enhanced basal and auxin-responsive transcription through as-1, it had a more pronounced effect on the latter activity. The ability of TGA1a to differentially affect Ϫ90-CAT activity was explored by a number of ways. First, reporters that lack as-1 (e.g., Ϫ76-CAT and CHS-CAT) were unresponsive to transiently expressed TGA1a, thus demonstrating that the affect of TGA1a on Ϫ90-CAT activity is through as-1. Results here with CHS-CAT also show that a G-box element that shares homology in its core motif to as-1, but binds poorly to TGA1a in vitro (13,33), is unresponsive to this factor.
We also explored the possibility that TGA1a might be sequestered in the cytoplasm and subsequently mobilized to the nucleus in response to auxin. This type of regulation is observed with the stimulus-coupled nuclear import of NF-B and STAT transcription factors in animal cells (17). By immunofluorescent staining, we observed that (epitope-tagged) TGA1a is predominantly localized to the nucleus in control and auxintreated protoplasts. Because no change in the cellular amount of this factor was found in response to auxin, we conclude that TGA1a is constitutively present in the nucleus and that its transcriptional activity is latent in the absence of a stimulus.
If binding to as-1 is regulated by auxin, we reasoned that a chimeric fusion protein of TGA1a and the heterologous DNA binding domain of the yeast GAL4 transcription factor might promote constitutive, rather than signal-responsive, transcription. Contrary to this expectation, we found that GAL4-TGA1a conferred both activities. This prompted us to investigate whether as-1-specific binding, and perhaps other bZIP-encoded functions, are required for trans-activation by this factor. We found that the activity of a mutant form of GAL4-TGA1a that lacks the bZIP domain (i.e. GAL4-TGA1a⌬bZIP) was enhanced by auxin, thus demonstrating that bZIP-encoded functions (e.g. as-1-specific binding, dimerization, and nuclear import) in TGA1a are not the primary regulatory target of this stimulus. Removal of the bZIP domain from GAL4-TGA1a was also found to increase its overall basal and auxin-responsive activities. Although the reason for this effect is unknown, we speculate that the presence of multiple and distinct DNA binding domains in GAL4-TGA1a may result in a chimeric factor that is functionally unstable or that binds to plant as-1 elements, thus decreasing its effective concentration on the GAL-CAT reporter. Data from in vivo labeling experiments favor the former possibility, because the accumulation of GAL4-TGA1a is at least one-fourth that of GAL4-TGA1a⌬bZIP at effector DNA concentrations that maximize transcription.
Trans-acting factors stimulate gene expression by overcoming rate-limiting steps in the recruitment of RNA polymerase II to the promoter. Within these factors, domains that are rich in acidic, glutamine, proline, or isoleucine residues commonly mediate trans-activation or repression. The NT region of TGA1a contains an acidic-rich region (amino acids 18 -66) that promotes trans-activation in vitro (13) and in vivo (28). Experiments here confirm and extended these results by showing that the first 84 residues of the amino terminus of TGA1a, when fused to the DNA binding domain of the heterologous GAL4 protein, confer basal trans-activation. Moreover, the activity of the NT domain is constitutive and unaffected by auxin, in keeping with our findings that the response to auxin is largely mediated by the carboxyl-terminal domain of TGA1a. We were surprised to find that at saturating amounts of each factor, the basal activity of GAL4-TGA1a-NT is markedly higher than that of GAL4-TGA1a⌬bZIP (Fig. 3). This difference suggests that CT residues 143-373, which flank the bZIP domain, might inhibit the basal activity of the NT region. We speculate that auxin somehow relieves this repression, thus maximizing trans-activation. This notion is also consistent with results here showing that maximal activation by GAL4-TGA1a⌬bZIP and GAL4-TGA1a-NT are quite similar in the presence of auxin. In regard to the potential regulatory role of the CT domain on NT activity, we note that trans-activation by an animal bZIP factor termed ATF-2 is regulated through intramolecular interactions (34).
Evidence here indicates that the carboxyl-terminal region of TGA1a may have multiple roles in transcription. First, acidic and glutamine residues make up over one-third of a 50-amino acid region within the CT domain (Fig. 1), thus implying the presence of a second activation domain in TGA1a. Consistent with this view, we found that the CT domain of TGA1a alone can confer basal and auxin-responsive transcription through the GAL4 binding domain, albeit at lower levels than those seen with TGA1a⌬bZIP. Further mutagenesis and gain-offunction assays may help to identify specific residues in the CT domain that contribute to these individual activities.
How might auxin affect the trans-activation potential of the CT domain? Prior studies have identified a "dimer stabilization domain" (amino acids 178 -373) within the CT of TGA1a. Although this domain enhances the ability of TGA1a to dimerize and bind as-1, it is not essential for either of these activities (35). We suggest that this domain may reversibly bind to an additional factor that mediates transcription, a situation that is analogous to that observed with other bZIP factors (36,37). Experiments here indicate that under basal conditions this mediator may bind the CT domain of TGA1a to repress transactivation through the NT domain. This inhibition is apparently relieved by auxin. We are further exploring this notion that a regulatory protein differentially associates with TGA1a in response to auxin.
Diverse plant hormones such as auxins, SA, and MJ, in addition to a number of their biologically inactive analogs, stimulate transcription through as-1 type elements (2,20). We found that 2,3-D, a biologically inactive analog of the auxin 2,4-D, stimulated as-1-dependent gene expression and the trans-activation potential of TGA1a. Lack of correspondence between the biological activity of auxins or their analogs and their ability to induce transcription has been interpreted as evidence of a response to general chemical stress (20). To further explore the nature of the auxin pathway involved, we examined whether mitogenic or toxic properties of 2,4-D mediate its affect on as-1-dependent transcription. As transcription was only induced by concentrations of 2,4-D that inhibited cell-growth, we suggest that this response involves an inhibitory chemical stress pathway. Although benzyl isothiocyanate, dimethyl fumarate, and hydroquinone induce the transcription of animal genes through chemical stress (38), we found that these compounds failed to stimulate as-1 -dependent transcription in BY-2 protoplasts over a range of concentrations (data not shown). Based on these observations, we suggest that transcriptional responses studied here evidence a high degree of selectivity for the type of chemical inducer.
A potential biological role for as-1-type elements has been suggested by their presence in the promoters of genes that encode for GST isoenzymes of plants (8, 9, 20, 39 -41). GSTs conjugate, transport, and detoxify a wide range of chemical substrates in animals and plants (42,43), and through their GSH-dependent peroxidase activity, they may protect plant cells from reactive hydroperoxides produced as a consequence of oxidative stress and defense responses (44). Intriguingly, a number of plant GSTs also bind to auxin (45)(46)(47). Animal and plant GST genes are regulated by xenobiotic and hormone cues through nearly identical cis-elements (41,48), thus raising the possibility that transcription factors that control the expression of GST genes may be evolutionarily conserved.