Characterization of a p53-related Activation Domain in Adr1p That Is Sufficient for ADR1-dependent Gene Expression*

The yeast transcriptional activator Adr1p controls expression of the glucose-repressible alcohol dehydrogenase gene (ADH2), genes involved in glycerol metabolism, and genes required for peroxisome biogenesis and function. Previous data suggested that promoter-specific activation domains might contribute to expression of the different types ofADR1-dependent genes. By using gene fusions encoding the Gal4p DNA binding domain and portions of Adr1p, we identified a single, strong acidic activation domain spanning amino acids 420–462 of Adr1p. Both acidic and hydrophobic amino acids within this activation domain were important for its function. The critical hydrophobic residues are in a motif previously identified in p53 and related acidic activators. A mini-Adr1 protein consisting of the DNA binding domain of Adr1p fused to this 42-residue activation domain carried out all of the known functions of wild-type ADR1. It conferred stringent glucose repression on the ADH2 locus and on UAS1-containing reporter genes. The putative inhibitory region of Adr1p encompassing the protein kinase A phosphorylation site at Ser-230 is thus not essential for glucose repression mediated byADR1. Mini-ADR1 allowed efficient derepression of gene expression. In addition it complemented anADR1-null allele for growth on glycerol and oleate media, indicating efficient activation of genes required for glycerol metabolism and peroxisome biogenesis. Thus, a single activation domain can activate all ADR1-dependent promoters.

Activation domains of transcription factors transmit signals to the transcriptional machinery of a cell to ensure proper gene expression. When tethered to a DNA binding domain, either covalently or via protein-protein interaction, they ensure that signals are transmitted to appropriate genes to activate their transcription. Although the specificity of gene expression is determined primarily by the DNA binding domain of transcription factors, activation domains can contribute to this specificity (1)(2)(3)(4).
Activation domains function by contacting other proteins that are components of the transcriptional machinery (5). The proteins contacted by activation domains include various subunits of TFIID, including TATA-binding protein itself and TATA-binding protein-associated factors, and other general transcription factors such as TFIIB, TFIIH, and members of adaptor or mediator complexes. By contacting these proteins, they recruit RNA polymerase II to the promoter and facilitate initiation and elongation of transcription (5)(6)(7)(8)(9)(10)(11).
Deletion studies of ADR1 suggested that its activation domains were promoter-specific (20,25). Deletion of the 3Јhalf of ADR1, including TADIV, prevented cells from growing on oleate, an inducer of peroxisome proliferation, whose utilization requires ADR1 (14). This ADR1 deletion mutant could still support growth on glycerol, but further deletion of TADIII prevented growth on glycerol and still allowed ADH2 derepression (14,20).
Here we describe the results of our search for transcriptional activation domains in Adr1p by fusing different parts of ADR1 to the DNA binding domain of GAL4 (GBD). In contrast to the results obtained using LexA fusions, our analyses indicate the presence of a single major TAD corresponding to TADIII. When TADIII was deleted from ADR1, the rate of ADH2 derepression decreased more than 10-fold without altering the stability of Adr1p, indicating that it plays an important role in transcription activation by native Adr1p. We have identified the region of TADIII that is important for activation as amino acids 420 -462. Within this region, the pattern of hydrophobic amino acids matches that in the transcription activation domains of p53 and related acidic activators. Mutagenesis showed that hydrophobic amino acids within the consensus sequence were particularly important for transcription activation, suggesting a conserved role for this motif in mammalian and yeast cells.
We examined the function of a gene we call mini-ADR1, a fusion of TADIII to the DNA binding domain of Adr1p. Mini-ADR1 activated ADH2 expression and growth on glycerol or oleate as well as wild-type, full-length Adr1p. Importantly, mini-ADR1 conferred stringent glucose repression on the ADH2 gene and on UAS1-containing reporters. The same amino acids that were important for transcription activation by GBD-TADIII were important for the activity of mini-ADR1.

EXPERIMENTAL PROCEDURES
Strains-Yeast strains are listed in Table I. Growth of Yeast Cultures-Yeast strains were grown in YPD or synthetic medium (SM) prepared according to standard methods (31). Glucose-repressed cultures were started in YP or supplemented SM containing 5% glucose. Derepressing cultures were inoculated into YP or supplemented SM containing 2% ethanol, 2% glycerol, 1% lactate, and 0.05% glucose.
Plasmid Construction-The base plasmid for GAL4-ADR1 gene fusions was pMA424 (32) an episomal, HIS3 plasmid containing the ADH1 promoter and terminator, the GAL4 DNA binding domain (amino acid residues 1-147), and a polylinker containing unique EcoRI, BamHI, and SalI sites at the 3Ј end of the GAL4 fragment. Fragments of ADR1 were introduced into the polylinker after cutting with EcoRI, EcoRI and NruI, EcoRI and BamHI, or EcoRI and SalI. The correct reading frame at the EcoRI site in the vector and in ADR1 was obtained using the Klenow fragment of E. coli DNA polymerase or mung bean nuclease, as required. The junction sequence was verified by DNA sequence analysis using a GAL4-specific oligonucleotide.
Large fragments of ADR1 containing internal EcoRI sites were cloned using partial digestion with EcoRI or serial cloning using unique restriction sites within intermediate constructs. The source of ADR1 was pADR1-4b, which contains ADR1 flanked by SalI sites introduced at the NciI and ScaI sites immediately flanking the 5Ј and 3Ј ends of the ORF, respectively, cloned into pUC19. The ADR1 fragments in plasmid pJSL33 were derived from E. coli expression vector pCQV229 (16). Other ADR1 fragments were made by PCR and utilized primers that generated BamHI and SalI sites at the 5Ј and 3Ј ends of the fragments, respectively. The PCR products were cut with these enzymes and cloned into pMA424 cut with the same two enzymes.
A deletion of ADR1 representing amino acids 419 -467 in ADR1 was made by recombinant PCR. This deletion was introduced into a GBD-ADR1 gene fusion representing ADR1-encoded amino acids 21-753 (deleted for amino acids 419 -467) and also into non-fusion ADR1 genes by swapping DNA fragments, to create plasmids pJSL127 and pACh3. These are TRP1-CEN-based plasmids based on pBC3T1 (33) and pRS314 (34), respectively, in which ADR1 is expressed from its own promoter.
Mini-ADR1 was constructed in several steps. A 6.2-kilobase pair fragment containing the ADR1 gene from the SspI site at position Ϫ1067 (from the start of translation) to the PstI site at position 5226 was assembled between the EcoRV and PstI sites in pRS314 to create pJS20. The ADR1 gene in pJS20 was truncated at amino acid 172 by digesting the plasmid with EcoRI and BamHI and inserting a 482-base pair EcoRI-BamHI fragment from pADR1-(172)-lacZ (16) to create pJS21. The truncated ADR1 gene in pJS21 was fused to ADR1-encoded amino acids 420 -462 to create pJS22. This was done by cutting pJS21 with XbaI, blunt-ending the termini with the Klenow fragment of DNA polymerase, cutting with BamHI, and recombining the larger DNA fragment with a BamHI-EcoRV fragment cut from plasmid pJSL105 ( Fig. 1). This plasmid was then cut with BamHI and the termini blunt-ended and religated. The BamHI site in pJSL105 is in the pMA424 polylinker, and the EcoRV site is in the ADH1 terminator. There are 9 and 14 non-ADR1-encoded amino acids between amino acids 172 and 420 and at the C terminus, respectively, that are due to the polylinkers used for cloning.
Mutations were introduced into ADR1-encoded amino acids 420 -462 using mutagenic primers and recombinant PCR or the Quick Change mutagenesis kit obtained from Stratagene. The sequences of the mutagenic primers are available on request. Several mutant clones from each construction were sequenced. Two clones with the identical mutation were usually tested to reduce the possibility that a cryptic mutation introduced during the PCR amplification had an effect on the observed phenotype. The construction of pJS23 (ADR1-VP16 in pRS314) is described elsewhere. 2 Enzyme Assays-ADH enzyme activity was visualized after electrophoresis of protein extracts on non-denaturing polyacrylamide gels (30). ␤-Galactosidase activity was measured either in permeabilized cells or in cell extracts (36). Generally three transformants were assayed for each plasmid construct, and the experiment was repeated at least twice. All of the GAL4-ADR1 fusions were assayed in at least two different strains that carried different GAL reporter genes. Although the absolute values of the activity differed significantly in the different strains, the relative activity was similar in all strains. In particular, none of the GAL4-ADR1 fusions was inactive in one strain and active in another.
Protein Extracts and Western Blotting-Protein extracts from yeast cells were prepared by disrupting the cells with glass beads (400 m) in 50 mM Tris (pH 6.8), containing 1 mM phenylmethylsulfonyl fluoride, 2 mg/ml aprotenin, 1 mg/ml leupeptin, 2 mg/ml pepstatin A, and 1% 2-mercaptoethanol at 4°C. Debris was removed by centrifugation. A portion of the extract was brought to 10% glycerol, 1% sodium dodecyl sulfate and heated to 95°C for 5 min. Protein concentration in the remaining extract was determined using Bio-Rad Protein Assay solution (Bio-Rad). Denatured protein (50 g) was separated by SDS-polyacrylamide gel electrophoresis. The proteins were transferred to nitrocellulose. Adr1p was detected with polyclonal antibodies as previously described (37). GAL4-ADR1 fusion proteins were detected with GAL4 GBD (DNA binding domain) monoclonal antibody (Stratagene, catalog number 5399-1).
DNA Electrophoretic Mobility Shift Assays-DNA binding and the electrophoretic mobility shift assays were performed as described (38).

Identification of a Single ADR1 Activation Domain
Using GBD-ADR1 Gene Fusions-Gene fusions between the GAL4 DNA binding domain (GBD; amino acids 1-147) and different fragments of ADR1 were made in the plasmid pMA424 (39). This is a high copy plasmid containing HIS3 for selection in yeast. The gene fusion is expressed from the strong ADH1 promoter. A subset of the plasmids generated is depicted in Fig.  1, which shows the region of Adr1p present and the activities that were induced by their presence in a strain containing a GAL7-lacZ reporter gene on a high copy plasmid (p632-17b-2 (40)).
The first set of gene fusions ( Fig. 1, pJSL68 through pJSL33, and others not shown) suggested the presence of a strong activation domain between amino acids 336 and 505. Deletion of the C-terminal 582 amino acids had no significant effect on activation. More importantly, GBD-ADR1 containing the Cterminal 583 amino acids was inactive (pJSL66). Plasmids encoding the N-terminal 336 or 229 amino acids, both of which contain TADI, were inactive. We tested the effect of mutating Ser-30 to Ala, which leads to hyper-activation of ADH2 expression. We saw no effect of this mutation at a UAS GϪ containing reporter in three different GBD-ADR1 gene fusions. 3 Plasmids pJSL97 through pJSL116 ( Fig. 1) were used to more precisely localize the region responsible for transcriptional activation by GBD-ADR1 fusions. Their activities indicated that the region responsible for transcriptional activation MATa ade2 leu2 his3 tyr1 ⌬gal4 Dgal80 ura3::SV15 (URA3) (35) was contained between amino acids 420 and 462, the most acidic 40-amino acid stretch within Adr1p (41). This TAD corresponds to TADIII that was identified by LexA-ADR1 gene fusions. We deleted this region from a large GBD-ADR1 fusion (Adr1p amino acids 21-753). The activity of the mutant fusion protein decreased more than 10-fold (pJSL116). The wild-type and deletion mutant proteins were present at comparable levels. 3 No activity was detected in strains with pJSL104 (Adr1 amino acids 154 -424) or pJSL102 (Adr1p amino acids 468 -753). These plasmids encode GBD-TADII and GBD-TADIV. Because these plasmids appeared to be unstable in GGY1, we continued our analysis in strain PJ69-4A (42), which carries a GAL2-ADE2 reporter gene as well as a GAL7-lacZ reporter gene. The ␤-galactosidase activities in the transformants were similar to those shown in Fig. 1. 3 Only those transformants that were ␤-galactosidase ϩ were also Ade ϩ , as expected ( Fig.  2A). However, the Ade ϩ colonies showed strong sectoring, both of the Ade phenotype and for viability (Fig. 2B), indicating that the active fusion protein is deleterious to the cell. Western blotting of extracts of transformants carrying these plasmids showed low levels of some of the fusion proteins 3 that could be due to this toxic effect.
To try to achieve more stable expression of the gene fusions, and to confirm the results shown in Fig. 1, we transferred a selected set of GBD-ADR1 fusions to a different multicopy plasmid containing TRP1 as a selectable marker. The set represented all portions of ADR1 except the first 21 amino acids. These plasmids appeared to be more stable as indicated by an absence of lethal sectoring. The activities of the gene fusions in the new vector assayed in PJ69-4A confirmed the initial results (Fig. 3). The two transformants containing fusions that include TADIII (pGBA6 and pGBA8) had the highest ␤-galactosidase activities. Activation of gene expression could also be scored by growth phenotypes in PJ69-4A since it has GAL2-ADE2 and GAL1-HIS3 reporter genes. In addition, the Ade phenotype could be scored by color. As expected the ␤-galactosidase ϩ transformants were white and the ␤-galactosidase Ϫ transformants were red, indicating that they were Ade ϩ and Ade Ϫ , respectively. The ␤-galactosidase ϩ , Ade ϩ transformants were also His ϩ , indicating that TADIII was able to activate three different GAL4-dependent promoters. The transformant containing pGBA7, which encodes amino acids 21-753 with an internal deletion of TADIII, was weakly Ade ϩ and His ϩ . However, ␤-galactosidase activity in these transformants was about 20-fold lower than that in the transformants containing pGBA8. Since the ␤-galactosidase activities in all the other transformants were even lower, we conclude that TADI, TADII, and TADIV are less than 5% as active as GBD fusions as TADIII.
All of the GBD-Adr1p fusion proteins could be detected with anti-GBD antisera with the exception of the fusion containing amino acids 154 -420 of Adr1p. This fusion protein was detected using antisera directed against the internal portion of Adr1p. 3 Although GBD-Adr1-(154 -420) may be less abundant than some of the others, we do not believe that it contains an important activation domain when fused to GBD, since a fusion protein containing this region but lacking TADIII was 20-fold less active than the fusion containing TADIII (compare pGBA7 with pGBA8 in Fig. 3).
Deletion of TADIII Reduces the Rate and Extent of Derepression-If TADIII is an important activation domain at UAS1, its deletion should reduce derepression significantly at promoters bearing Adr1p-binding sites. This was first tested using the GAL4-ADR1 fusions pGBA7 and pGBA8. These gene fusions  1-147). The plasmids are derived from pMA424. The gene fusions are expressed from the ADH1 promoter and are followed by the ADH1 terminator. Plasmids pJSL68 through pJSL33 were made using restriction enzyme sites; plasmids pJSL97 through pJSL116 were made using using PCR amplification as described under "Experimental Procedures." The numbers following the plasmid names correspond to the ADR1-encoded amino acids fused to GBD. The plasmid pJSL116 contains an in-frame deletion of amino acids 419 -467. The host strain was JSY55 containing the multicopy plasmid 632-17b-2 which contains two copies of GAL4 site 3 upstream of a UAS-less GAL1 promoter fused to lacZ. Similar activities were obtained in a related strain (GGY1::SV15) containing an integrated reporter containing a single UAS G consensus site upstream of a CYC1/lacZ fusion. ␤-Galactosidase activities are in Miller units, and the average deviation was about 30%.
should be active at both UAS G -and UAS1-containing promoters since they contain both GBD and ABD. Table II shows that the UAS1-containing promoters tested were strongly dependent on GBD-ADR1 and that deletion of the activation domain reduced activity 5-12-fold.
To test the effect of deleting TADIII in its normal Adr1p context, an ADR1 gene was constructed lacking this region (pJSL127, ADR1⌬TADIII). The mutant gene was present as a low copy plasmid and expressed from the native ADR1 promoter. The rate of derepression of a UAS1-driven lacZ reporter gene was reduced about 15-fold in the presence of ADR1⌬TADIII compared with wild-type ADR1. The final level of activation was 5-fold lower than the level in the presence of wild-type Adr1p (Fig. 4). Western blot analysis of the wild-type and mutant proteins indicated that they were present at comparable levels and at the same level as Adr1p expressed from its normal chromosomal locus (inset to Fig. 4).
TADIII was not required for growth on glycerol or oleate. A strain containing ADR1⌬TADIII could grow on media containing these carbon sources, 3 indicating that other portions of ADR1 could compensate for loss of TADIII.
Activity and Regulation by a mini-ADR1 Gene-To assay the activity of TADIII at ADR1-dependent promoters in the absence of other potential TADs, a mini-ADR1 gene was constructed. Mini-ADR1 was expressed from the ADR1 promoter on a low copy plasmid to ensure normal expression levels. The activation domain was fused in-frame to the N-terminal 172 amino acids of Adr1p, which contains the DNA binding domain FIG. 2. TADIII causes plasmid instability. The phenotype of plasmids containing GBD-ADR1 was assessed in strain PJ69-4A (Table I) which contains GAL4-responsive reporter genes driving expression of lacZ, ADE2, and HIS3. The ADR1 amino acids encoded in the gene fusion is listed next to the sector containing a transformant with the respective plasmid. Most of the plasmids tested are shown in Fig. 1. A, 3. An ADR1 activation domain is necessary and sufficient for activation from multiple GAL promoters. ADR1 is represented as in Fig. 1. GAL4-ADR1 fragments are present in a derivative of pRS304 to which a 2-m origin of replication has been added (gift of G. Zhu). The host strain was PJ69-4A. ␤-Galactosidase activities are in enzyme units (micromoles of ONPG hydrolyzed per min per mg of protein). The average deviation was about 30%. Cultures were grown to mid-log phase in trp Ϫ synthetic medium containing 5% glucose. Protein extracts were prepared as described under "Experimental Procedures." Ade and His phenotypes were scored on trp Ϫ ade Ϫ and trp Ϫ his Ϫ drop-out plates. The color of the colonies on trp Ϫ plates is recorded in parentheses in the Ade phenotype column: R, red (Ade Ϫ ); W, white (Ade ϩ ). The his Ϫ plate contained 10 mM 3-aminotriazole to increase the stringency of the test. and nuclear targeting signal. Mini-ADR1 was tested for its ability to regulate expression of a UAS1-containing reporter gene and to regulate ADH2 itself. The reporter gene was tightly glucose-repressed and was activated to the same high level in the presence of either wild-type ADR1 or mini-ADR1 when glucose was removed from the media (Table III). ADH2 itself was also tightly repressed and efficiently derepressed by mini-ADR1 as shown by analysis of ADHII activity after separation of the ADH isozymes by electrophoresis (Fig. 5). In contrast, ABD itself was completely inactive.
The ability of mini-ADR1 to complement an ADR1 deletion mutant for growth on media containing glycerol or oleic acid as the sole carbon source was also tested (Fig. 6). Mini-ADR1 could complement both of these defects as well as wild-type, full-length ADR1. As with activation of ADH2 expression, ABD alone was inactive in these assays.
Identification of Critical Amino Acid Residues in TADIII-TADIII is the major activation domain in Adr1p as defined by the GAL4-ADR1 fusions. It is a member of the class of acidic activators based on its high content of acidic residues (11 acidic versus 1 basic residue) and the absence of amino acids such as Pro or Gln. Changes in TADIII were made first in GBD-TADIII, and the activity of the mutants was assessed using a GAL4-dependent reporter gene. To determine the importance of charged residues in TADIII for activity, clusters of mutations were made in three regions (Fig. 7A). The charged residues were changed to Ala, except for Lys-446 which was inadvertently changed to Thr in three of the mutants (CR8, CR9, and CR10). Mutating all of the acidic residues in the N-terminal part of TADIII resulted in a 2-fold reduction in activity (CR1), whereas a 4-fold reduction in activity resulted when four acidic residues were changed to Ala in the middle of TADIII (CR4). The most severe reduction in activity, about 10-fold, resulted when all five charged residues were mutated at the C terminus of TA-DIII (CR10). It appears that negative charge at the C terminus TABLE II TADIII is important at UAS1-containing reporters Strain TYY303 was transformed with the reporter plasmids pADCY1 (ADH2/CYC1/lacZ), pHDY10 (CYC1(UAS1)/lacZ), YIpT2Z (ADH2(RV)/ lacZ), and pLG(S/X) (CYC1/lacZ). The GBD-based activator genes were introduced on the plasmids pJSL56 (ADR1: GBD-ADR1-(21-741)), pJSL116 (⌬TADIII: GBD-ADR1(21-753⌬TADIII), and pMA424 (Ϫ, GBD alone). Three transformants were grown in derepressing medium that was selective for both both reporter and activator plasmids as described under "Experimental Procedures." ␤-Galactosidase activity is given in Miller units. The values are the averages from three transformants, and the range of values was Ϯ30%. ADR1 and TADIII dependence is calculated as the ratio of ␤-galactosidase activities Ϯ ADR1 and ϮTADIII, respectively.  (Table III) were grown in ura Ϫ trp Ϫ synthetic medium containing 5% glucose. Cells were collected by centrifugation and resuspended in duplicate cultures of synthetic medium containing derepressing carbon sources at time 0. At the indicated times samples were withdrawn and assayed for ␤-galactosidase activity in Miller units. The values shown indicate the average enzyme activities from samples taken from the duplicate cultures. Squares, wild-type ADR1; triangles, ADR1 ⌬TADIII; circles, vector alone. The inset shows a Western blot of Adr1p after 12 h derepression. Extracts were made from strain TYY303 containing pKD8 (vector, pJSL127 (ADR1⌬TADIII, lane 3), or from strain W303-1A which contains wild-type ADR1 at its chromosomal locus (lane 4).
FIG. 6. Mini-ADR1 complements an ADR1-null allele for growth on glycerol or oleate. Cultures of strain TYY303 (adr1⌬1) containing pJS22 (mini-ADR1), pJS20 (ADR1), pJS21 (ABD), or pRS314 (vector) were grown in derepressed conditions and then streaked on trp Ϫ drop-out plates containing glucose, glycerol, or oleate as the sole carbon source. of TADIII is most important for activation, but the other clusters of acidic residues also contribute modestly to activity.
The most important hydrophobic amino acids appeared to be in the N terminus of TADIII (Fig. 7B). Changing pairs of hydrophobic residues in this region reduced activity 20-fold (HA1 and HA2). The effect of mutating hydrophobic residues in the rest of TADIII was not as deleterious, although all of the mutations reduced activity significantly.
To test the effect of activation domain mutations in their normal context, and at different UAS1-containing promoters, hydrophobic residues and the cluster of charged amino acids at the C terminus of TADIII were substituted by other amino acids, usually Ala, in mini-ADR1 and full-length ADR1. The effect of the mutations on a UAS1-containing reporter gene (Table IV) showed that hydrophobic residues Phe-425 and Leu-428 and Leu-429 are particularly important. The cluster of charged amino acids between Lys-446 and Asp-454 was essential for activity of mini-ADR1, as it was for GBD-TADIII.
The mini-ADR1 mutants were also assessed for their ability to grow on glycerol or oleate as a carbon source (summarized in Table IV). The mutations had an effect on growth that was similar to their effect on activation of the reporter gene. Mini-ADR1 containing substitutions in Phe-425, and Leu-428, Leu-429, or the cluster of charged resides between Lys-446 and Asp-454 was unable to complement an ADR1-null allele for growth on glycerol-or oleate-containing media. Thus, the same residues are probably important at all UAS1-containing promoters. A qualitatively similar effect of these mutations was seen at the ADH2 promoter as assessed by separation of the ADH isozymes by electrophoresis. 3 The level of mini-ADR1 mutant proteins in yeast was assessed by Western blotting. All of the proteins could be detected and were present at comparable levels, except the two with the mutations CR9 and CR10, which were present at lower levels than the others. 3 To demonstrate that the mutant proteins were properly folded, DNA electrophoretic mobility shift assays were performed. As shown in Fig. 8, all of the defective mutant proteins were able to bind DNA. The mutant proteins containing multiple changes of charged residues (CR9 and CR10) were present at reduced levels as judged by this assay. However, the active mutant (CR9) was reduced to the same extent as the inactive mutant (CR10), suggesting that the loss of activity was not due to lower abundance.
The activity of full-length Adr1p containing many of these same mutations was determined by activation of a reporter gene (Table IV). Its activity was significantly reduced by several of these mutations, but the impact of the mutations was not as strong as in the context of mini-ADR1. DISCUSSION GAL4(GBD)-ADR1 fusions identified a single, strong activation domain comprised of only 42 amino acids (residues 420 -462) within Adr1p. A previous analysis using LexA-ADR1 gene fusions identified four activation domains including one between residues 359 and 571. The latter region was designated TABLE IV Mutations in TADIII alter ADR1-dependent phenotypes Mutations were introduced into the plasmids pJS22 (mini-ADR1) and pJS20 (ADR1) and assayed in strain TYY303 (adr1⌬) transformed with the reporter plasmid pHDY10 (CYC1(UAS1)/lacZ). Mutations HA2, HA3, HA5, CR9, and CR10 are described in Fig. 7. The triple HA mutant is F425A,L428A,L429A. Transformants were grown selectively (trp Ϫ ura Ϫ derepressing medium) as described under "Experimental Procedures." ␤-Galactosidase activities (Miller units) of three transformants are reported as the percentage of the value for WT. The growth phenotype was tested for TYY303 transformed with pJS22 (mini-ADR1) and mutant derivatives of miniADR1 on SM (trp Ϫ ) plates containing either glycerol (at 37°C, 2 days) or oleate (at 25°C, 3 days). Growth was scored as follows: ϩϩϩ, growth equivalent to WT; ϩϩ and ϩ, intermediate growth; ϩ/Ϫ, hardly detectable growth; Ϫ, no growth. N C, no construct.

Mutations
␤-Galactosidase activity Growth phenotype  7. Alanine mutagenesis of TADIII. Changed residues are shaded. Strain PJ69-4A was transformed with plasmids carrying GBD-ADR1 gene fusions with the mutations shown. The base plasmid for mutagenesis was pGBA6 (Fig. 3). ␤-Galactosidase activity was measured in Miller units and is shown as a percentage of the activity measured for the wild-type sequence. The average deviation was about 30%. A, mutation of charged residues. Numbers 1-3 above the amino acid sequences indicate the regions where clusters of charged amino acids, Glu or Asp except in the case of Lys-446, were mutated. Dots above the sequences indicate the charged residues. B, mutation of bulky hydrophobic residues. Dots above the sequences indicate the phenylalanine and leucine residues that were changed. TADIII (20). Based on our unpublished results, 3 a later report (25) tested residues 420 -462 and showed that they were sufficient to activate transcription when fused to LexA. TADIII can function at three natural GAL4-dependent promoters, at artificial LexA-and GAL4-dependent promoters, and at ADR1dependent promoters. Surprisingly, TADs I, II, and IV did not significantly activate transcription from GAL4-dependent promoters.
TADIII is sufficient when fused to the nuclear targeting and DNA binding domains of ADR1 to carry out all of the known functions of ADR1. The observation that mini-ADR1 could fully complement an ADR1-null allele was unexpected since the encoded protein is only 24 kDa, much smaller than full-length Adr1p (150 kDa). Recently we have found that when TADIII is replaced by the activation domain from the herpesvirus VP16 gene the resultant ADR1-VP16 fusion behaved in an identical manner to mini-ADR1. 2 These results argue strongly that regulation mediated by ADR1 is due to the DNA binding domain of the protein or its binding site, UAS1 (40). Previous studies of Adr1p have failed to find a regulatory domain that mediates glucose repression. In particular, attempts to demonstrate a regulatory role for phosphorylation at the cAPK site at amino acid 230 have not met with success (20,27,29,30,37). Our studies with mini-ADR1 and with ADR1-VP16 2 demonstrate that neither the Ser-230 region nor the other portion of Adr1p that functioned as an inhibitory domain in Lex-ADR1 gene fusion (20) are essential for glucose repression mediated by Adr1p.
The observation that mini-ADR1 or an ADR1 (ABD)-VP16 gene fusion can complement an ADR1-null allele also demonstrates that all ADR1-dependent promoters can be activated by a single, small activation domain. Thus, it seems unlikely that wild-type Adr1p employs promoter-specific activation domains. Also surprising was the high level of activity shown by TA-DIII even when expressed from the weak ADR1 promoter on a low copy plasmid. TADIII was as strong as the activation domain of VP16. This provided us with a sensitive assay to measure the effect of mutations within the activation domain on the activation of UAS1-containing reporters and on ADH2 itself.
Mutational analysis of the activation domain allowed us to identify a potential target of TADIII. Several activation domains from multicellular eukaryotes possess a conserved motif, FXX⌽⌽, where X is any amino acid and ⌽ is a hydrophobic amino acid (43). This motif is also present in TADIII, as shown in Fig. 9. Related sequences, FXLLX and GXXLLX, are also present. Most importantly, the conserved hydrophobic amino acids have been shown by mutational analysis to be important for the function of the activation domain. The conservation of this motif and its importance in transcriptional activation domains from mammals and yeast suggest that its function may be conserved as well.
Co-crystal structures of the activation domains of VP16 and p53 complexed with hTaf31 and Mdm2, respectively, have demonstrated a role for the motif FXX⌽⌽ in interaction of an induced ␣ helix of the activation domain with its partner (43,44). These results suggest that this motif in ADR1 might interact with a protein related to either Mdm2 or Taf31. The yeast homolog of TAF31 is TAF17, an essential yeast gene (45). We are currently investigating the possibility that TADIII interacts TAF17.
The fact that ADR1 encoded-TADs interact with multiple components presumably allows a higher level of transcription activation than would be possible if it interacted with a single factor. Therefore, it is surprising that mini-ADR1, containing only a single activation domain, was as active as the entire protein. However, TADIII cannot be the only TAD active at ADR1-dependent promoters since its deletion from the entire protein reduced (about 10-fold) but did not abolish expression from these promoters. We speculate that in the context of the entire protein some of the ADR1-encoded TADs may be unavailable initially for interaction with components of the transcription apparatus and that during assembly of an active initiation complex (49), or during DNA binding (50), different TADs may become accessible to contribute to activation. By exposing TADs as artificial proteins, such as in mini-ADR1, they may exhibit higher activities than they possess in their natural environment. This may explain why mutations that severely impair the activity of mini-ADR1 are less deleterious in the context of the native protein.
Mutational analysis of TADIII identified important acidic and hydrophobic residues (Fig. 9). Changing hydrophobic amino acids to Ala had a more detrimental effect on activation than individual acidic-to-Ala substitutions, as has been observed for other acidic activators (46,(51)(52)(53)(54)(55)(56). Mutation of a cluster of charged (four acidic and one basic) residues at the C terminus of TADIII inactivated it. Mutating other clusters of acidic residues also significantly reduced transcription activation. Based on biochemical studies, it has been proposed that negatively charged regions of an activation domain might provide electrostatic interactions between an activation domain and its target, allowing weak initial binding that is supplanted by an induced fit interaction between hydrophobic residues on the two partners (43). There appears to be a correlation between negative charge of TADIII and transcriptional activity, but further mutagenesis and analysis is needed to test this rigorously. No promoter-specific defects were noted for any of the mutants tested. They had similar defects when tested in the context of a GBD-TADIII fusion or in mini-ADR1 and at different GAL4-and ADR1-dependent promoters. Thus, it is likely that they are defective in steps common to all of the FIG. 9. A conserved activation domain motif is present in TA-DIII. The gene name, position of the motif, and its sequence are shown. At the bottom of the figure the amino acid sequence of TADIII and its position in the ADR1 ORF are shown with the conserved motifs overlined. Circles below the sequence indicate hydrophobic residues shown by mutagenesis to be important for function. Brackets indicate clusters of charged, mostly acidic, residues whose mutation significantly reduced transcriptional activation. Residues indicated by closed symbols were more important for function than residues indicated by open symbols.
Why was only one ADR1-encoded TAD evident when assayed in the GAL4 gene fusions whereas four were found using LexA-ADR1 fusions? Several explanations are possible. (a) TADI, TADII, and TADIV might be promoter-specific, activating Lex operator-containing reporters but not UAS G -containing reporters(4). (b) LexA-Adr1p fusions might be more abundant than GBD-Adr1p fusions. As shown here, ABD itself is inactive when expressed at physiological levels. However, when expressed at very high levels, its activity can be detected in derepressed conditions (16) or in strains with mutations in ADR7, ADR8, or ADR9 (57). Thus, TADI seems to require very high levels in order to display its activity. (c) TADI, TADII, and TADIV might be cryptic activation domains whose activity is strongly dependent on the particular fragment of ADR1 that is fused to the DNA-binding fragment. This is true for GAL4-ADI (39) and for ADR1 TADIV (25). (d) The GAL4-ADR1 fusions encompassing TADI, TADII, and TADIV might have been inactive because they were misfolded. This seems unlikely because deletion of TADIII reduced 10 -20-fold the activity of a GBD-ADR1 gene fusion that would still encode TADs I, II, and IV and yet the mutant protein was as stable as the wild-type fusion protein.
The interpretation we favor to explain these different results invokes the ability of activation domains to function with different efficiencies at different core promoters (4). We suggest that the LexA promoter used in the other studies recruits a different constellation of RNA polymerase II-associated factors than the GAL4-and ADR1-specific promoters we used or that the relative importance of the various factors differs at different core promoters (4). Since Adr1p deleted for TADIII can still activate transcription, albeit weakly, contacts mediated by other TADs are sufficient for activated transcription.
Why does ADR1 encode 1323 amino acids if its essential functions can be carried out by mini-ADR1, encoding only 210 amino acids? A similar conundrum is presented by Gal4p, another large transcription factor (881 amino acids) whose essential features are present in a mini-GAL4 comprised of the DNA binding and activation/Gal80p-association regions (39,58). Dissection of GCN4 also revealed redundant activation domains, only one of which was necessary and sufficient for nearly wild-type function (52). In a natural setting the other parts of these transcription factors may be important. Evolution may have favored the formation of large transcription factors so that they could perform different or overlapping functions at different promoters. Our laboratory-based experiments may only survey a small fraction of these functions. Global surveys of Adr1p function should allow us to test this hypothesis using the different ADR1 alleles described here.