Transcriptional activation by yeast PDR1p is inhibited by its association with NGG1p/ADA3p.

NGG1p/ADA3p forms a coactivator/repressor complex (ADA complex) in association with at least two other yeast proteins, ADA2p and GCN5p, that is involved in regulating transcriptional activator proteins including GAL4p and GCN4p. Using a two-hybrid analysis, we found that the carboxyl-terminal transcriptional activation domain of PDR1p, the primary regulatory protein involved in yeast pleiotropic drug resistance, interacts with the amino-terminal 373 amino acids of NGG1p (NGG1p1-373). This interaction was confirmed by coimmunoprecipitation of epitope-tagged derivatives of NGG1p and PDR1p from crude extracts. An overlapping region of the related transcriptional activator PDR3p was also found to interact with NGG1p. Amino acids 274–307 of NGG1p were required for interaction with PDR1p. This same region is required for inhibition of transcriptional activation by GAL4p. The association between NGG1p1-373 and PDR1p may be indirect, possibly mediated by the ADA complex since the two-hybrid interaction required the presence of full-length NGG1. A partial requirement for ADA2 was also found. This suggests that an additional component of the ADA complex, regulated by ADA2p, may mediate the interaction. Transcriptional activation by a GAL4p DNA binding domain fusion of PDR1p was enhanced in ngg1 and ada2 disruption strains. Similar to its action on GAL4p, the ADA complex acts to inhibit the activation domain of PDR1p.

Transcriptional activator proteins are regulated both positively and negatively by associated factors. As a group, factors that positively influence activator function are called coactivators and include components of the RNA polymerase II holoenzyme (1) and the TATA-binding protein (TBP) 1 -associated factor complex (2). Several examples of activator proteins being regulated by negative factors also exist (3). From a mechanistic standpoint dual function regulators, with the ability to activate or repress transcription depending upon different promoter contexts or environment signals, represent particularly interesting models (4). The yeast ADA complex (5) is such a dual function regulator.
We initially isolated NGG1 based on its requirement for the full inhibition of transcriptional activation by GAL4p in glucose media (6). Expression of a GAL10-lacZ reporter is increased 300-fold in glucose media in a gal80 ngg1 background. Approximately 10 -15-fold of this effect is attributable to ngg1. We believe that GAL4p is the direct target for NGG1p action since inhibition is seen for independent GAL4p binding sites, requires GAL4, but does not require the GAL4 promoter (6). In addition, the synergistic effect with gal80 is consistent with NGG1p regulating the DNA binding or activity of GAL4p. Independently, NGG1/ADA3 was isolated by the Guarente laboratory based on the ability of mutations to suppress the toxic effects of overexpression of the viral activator VP16 in yeast (7). Four additional ADA genes were isolated in these screens (ADA1 to ADA5; Refs. 8 and 9). As the toxicity of VP16 was thought to arise from squelching of essential transcription factors, the ADA genes were predicted to encode coactivators required for transcriptional activation by VP16 (8). ADA2, NGG1/ADA3, GCN5/ADA4, and ADA5 are in fact required for transactivation by a set of transcription factors that includes the chimeric molecules GAL4p-VP16 and LexA-GCN4p (7)(8)(9)(10). GCN5p had also been identified because it is required for maximal activation by GCN4p (11). Using genetic and in vitro biochemical techniques, the Guarente and Thireos laboratories have shown that the ADA proteins probably act in a complex that contains at least ADA2p, NGG1p, and GCN5p (7,9,10,12). Direct interaction, in vitro, has been observed between ADA2p and both GCN5p and the carboxyl-terminal 250 amino acids of NGG1p (12).
Based on the finding that single and double disruptions of ngg1 and ada2 have similar effects on inhibition of GAL4p, we suggested that the same or related ADA complexes are involved in transcriptional activation and repression (13).
As coactivators the ADA proteins were predicted to interact with a component of the basal transcriptional machinery (8).
Recently an interaction between components of the ADA complex and TBP has been demonstrated by affinity chromatography and immunoprecipitation (14). 2 In this ability to associate with TBP, the ADA complex resembles other coactivators including the TBP-associated factor complex (15,16), SUG1p (17,18), and SPT3p (19), as well as repressor proteins such as MOT1p (20,21), which are biochemically and/or genetically linked to TBP. In its role as a coactivator/repressor, the ADA complex may provide a regulatory link between the activator protein and TBP. In addition to the genetic evidence that NGG1p interacts with GAL4p (6), biochemical evidence for the bridging function of the ADA complex exists as components of the ADA complex associate with the activation domains of VP16, GCN4p, and GAL4p (14,22,23). The ability of recombinant ADA2p to interact with VP16 in vitro suggests that ADA2p may have a principal role in these interactions (14).
We have used a two-hybrid screen (24) to identify molecules * This work was supported in part by funds from the Medical Research Council (MRC) of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In general, yeast strains were grown at 30°C in liquid suspension or on 2% Bactoagar plates in YPD broth (1% yeast extract, 2% peptone, 2% glucose) or in minimal medium (0.67% yeast nitrogen base without amino acids, 2% glucose and supplemented with additional amino acids when required). Plasmid DNA was transformed into yeast cells treated with lithium acetate (30).
DNA Constructs-Molecules were constructed using standard cloning techniques and verified by restriction and sequence analysis. NGG1 derivatives were cloned in pAS1 (Ref. 26, provided by S. Elledge) to allow expression of GAL4p DNA binding domain fusion proteins. GAL4 DBD -NGG1  was generated by cloning a NdeI-NsiI(blunt) fragment of NGG1, encoding the initiator ATG to codon 373 into the NdeI-SmaI site of pAS1. GAL4 DBD -NGG1 ⌬274 -307 was constructed from GAL4 DBD -NGG1 1-373 by digestion of the NGG1 coding sequence with BglII and religation. A PCR fragment, encoding amino acids 276 -376 of NGG1p, was ligated into the NdeI-BamHI sites of pAS1 to generate GAL4 DBD -NGG1 276 -376 .
GAL4 DBD -PDR1 813-1063 was constructed by ligating a BglII fragment from GAL4 AD -PDR1 813-1063 , encoding the carboxyl-terminal 251 amino acids of PDR1 (see below), into the BamHI site of pAS1. ⌴yc-PDR1 was constructed as follows. A NotI-NsiI fragment including PDR1 sequence from the initiator ATG to codon 406 was generated by PCR. A NsiI-SstI, fragment encoding the remaining 657 amino acids of PDR1p, was transferred from a genomic clone of PDR1 (isolated from a yeast genomic library, kindly supplied by J. Archambault and J. Friesen). These two fragments were cloned into pDMYC (13) to allow expression of Myc-tagged PDR1p from the DED1 promoter. The ded1-Myc-PDR1 fusion was transferred to the 2 plasmid YEplac181 (29) as a SalI-SstI fragment.
A DNA fragment encoding the carboxyl-terminal 212 amino acids of PDR3 (31) was generated by PCR from genomic DNA and ligated into the BamHI-SalI sites of pACTII (26) to give GAL4 AD -PDR3  .
Two-hybrid Analysis-Yeast strain Y190 containing the plasmid GAL4 DBD -NGG1 1-373 was transformed with a yeast cDNA library fused to the GAL4p activation domain in pACT (Ref. 26; received from S. Elledge). Approximately 1.25 ϫ 10 5 transformants were plated on selective media containing 50 mM 3-aminotriazole. After 3-5 days at 30°C, His ϩ colonies were replicated onto Trp Ϫ Leu Ϫ media containing 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside. cDNAs encoding potential NGG1p interacting proteins were identified as His ϩ colonies that were blue. Positive colonies were purified and retested for activated GAL-lacZ expression. Yeast strains containing only the pACT-cDNA plasmids were obtained by growing strains to saturation then plating on media lacking leucine, followed by screening for tryptophan auxotrophy. The growth rates of diploids generated by mating these strains with Y187 derivatives expressing either GAL4 DBD -SNF4p, GAL4 DBD -p53, or GAL4 DBD -lamin, were assayed to determine the specificity of the interaction. DNA sequence was determined using the T7 polymerase kit from Pharmacia Biotech Inc.
␤-Galactosidase Assays-Yeast strains were grown in liquid culture in minimal media to an A 600 of 1.0 -1.5. Cells were pelleted, washed and concentrated 2-50-fold in LacZ buffer (33). ␤-Galactosidase activity was determined after disruption of the cells with glass beads as described by Himmelfarb et al. (34). Activity was standardized to protein concentrations determined by the Bio-Rad protein assay. For yeast strains expressing potent activator proteins, ␤-galactosidase activity was assayed according to Ausubel et al. (35) and standardized to cell density.
Coimmunoprecipitation of PDR1p and NGG1p-Yeast whole cell extract was prepared as described by Schultz et al. (36) with minor modifications. 2 For immunoprecipitation of Myc-tagged proteins, 50 mg of whole cell extract was rotated for 1 h at 4°C with 200 l of a slurry of Sepharose CL-4B (Pharmacia) equilibrated in IP buffer (Ref. 12; 50 mM HEPES, pH 7.3, 100 mM sodium glutamate, 6 mM magnesium acetate, 1 mM EGTA, 0.1% Nonidet P-40, 0.5 mM dithiothreitol, 500 mg/ml bovine serum albumin, 10% glycerol). Unbound protein was mixed with 7 l of an anti-Myc ascites fluid derived from the Myc1-9E10 cell line (37) and rotated at 0°C for 1 h. 100 l of Protein A-Sepharose beads (Pharmacia), preincubated with 20 g of rabbit anti-mouse IgG and IgM (Jackson ImmunoResearch Labs, Inc.) and equilibrated in IP buffer, were added to the mixtures and rotated 3 h at 4°C. Protein-bound beads were pelleted by centrifugation and washed four times in IP buffer without bovine serum albumin. Immunoprecipitated complexes were separated from the beads by boiling for 5 min in SDS gel loading buffer. Protein was separated by SDS-PAGE and analyzed by Western blotting. For immunoprecipitation of hemagglutinin (HA)-tagged proteins, a similar protocol was followed substituting anti-HA monoclonal antibody, clone 16B12, covalently bound to N-hydroxysuccinimide-activated Sepharose (Berkeley Antibody Co.), for the Myc ascites fluid and Protein A-Sepharose.
Western Blot Analysis-Western blotting with anti-Myc primary an- tibody from ascites fluid derived from the Myc1-9E10 cell line (37) using polyvinylidene difluoride membrane was performed as described previously (13). Probing with anti-HA monoclonal primary antibody, clone 12CA5 (Ref. 38; Boehringer Mannheim) was done at a dilution of 1:5000.

Isolation of NGG1p
Interacting Proteins-We have used the yeast two-hybrid system (24) to identify proteins that interact with the amino-terminal 373 amino acids of NGG1p (NGG1p 1-373). This segment of NGG1p contains a region essential for inhibition of GAL4p (amino acids 274 -307; Ref. 13). By focusing our analysis on the amino-terminal half of the protein, problems associated with the independent activation by the full-length protein (12) could be reduced. The chimeric gene, GAL4 DBD -NGG1  , which expresses a GAL4p DNA binding domain fusion protein (Fig. 1A), was transformed into yeast strain Y190 together with a yeast cDNA library expressed as GAL4p activation domain (GAL4 AD ) fusions in the LEU2 plasmid pACT (26). Leu ϩ Trp ϩ transformants were first selected for induction of the GAL-HIS3 reporter by growth in minimal media containing 50 mM 3-aminotriazole. After selection, transformants were screened for expression of GAL-lacZ by replicating on media containing 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside. From 1.25 ϫ 10 5 transformants (approximately 15% of the cDNA library), six GAL4 AD -cDNA plasmids reproducibly resulted in 3-aminotriazole resistance and showed elevated expression of ␤-galactosidase. To test if increased expression of GAL-HIS3 by GAL4 AD -cDNA fusions was dependent on and specific to GAL4 DBD -NGG1 1-373 , strains containing only the GAL4 AD -cDNA plasmids were mated to yeast strain Y187 expressing GAL4p DNA binding domain fusions of either NGG1p 1-373 , SNF1p, p53, or lamin (the latter three represent nonspecific bait proteins). For the six plasmids isolated, increased expression of GAL-HIS3, indicative of the two-hybrid interaction, was specific to coexpression with GAL4 DBD -NGG1 1-373 (data not shown). Restriction patterns of the isolated GAL4 AD -cDNA plasmids revealed four distinct cDNAs. Expression of GAL-lacZ for Y190 containing GAL4 DBD -NGG1 1-373 after retransformation with each of the four GAL4 AD -cDNAs and GAL4 AD -SNF4 (nonspecific target) is shown in Fig. 1B.
The four cDNAs were sequenced and compared with sequences in the GenBank data base. Interestingly, one of the cDNAs encoded the carboxyl-terminal 251 amino acids of the transcriptional activator protein PDR1p (39). PDR1p contains a DNA binding domain related to GAL4p and activates expression of genes required for pleiotropic drug resistance (39 -41). Based on the role of NGG1p and the ADA complex in regulating transcriptional activators (6 -9, 12, 13), we chose to further characterize the interaction between NGG1p and PDR1p and its functional significance as described below.
In addition to PDR1p, we identified three cDNAs encoding proteins that interact with the amino-terminal 373 amino acids of NGG1p. The NIF3 (NGG1p interacting factor) cDNA contained sequence not found in the data base. The protein fragment responsible for the interaction with NGG1p 1-373 was probably a 54-amino acid GAL4p fusion since a frameshift mutation at the GAL4 DNA activation domain junction resulted in a loss of interaction. A second cDNA was in frame with the coding sequence of SYGP-ORF43 (accession no. L11119), an open reading frame coding for 357 amino acids with no described function. The third cDNA coded for the carboxyl-terminal 180 amino acids of a yeast ubiquitin-specific protease, UBP3p (42). These gene products are being further analyzed to determine the functional significance of their interaction with NGG1p.
Mapping the Region of NGG1p Required for Interaction with PDR1p-To localize the region within the amino-terminal 373 amino acids of NGG1p that is able to recruit PDR1p, we analyzed internal deletions of GAL4 DBD -NGG1 1-373 in a two-hybrid analysis with GAL4 AD -PDR1 813-1063 (Fig. 2). Because GAL4 DBD -NGG1 fusions are transcriptional activators (12,13), the GAL4 DBD -NGG1 deletion derivatives were also coexpressed with GAL4 AD -SNF4, which does not interact with NGG1p 1-373 (Fig. 1), to control for changes in GAL10-lacZ induction related to its function as an independent activator. tion of GAL4p (6), is essential for the interaction with PDR1p.
Coimmunoprecipitation of NGG1p and PDR1p-Coimmunoprecipitation was used to verify the two-hybrid interaction between NGG1p and PDR1p. Whole cell extracts were prepared from yeast strains expressing NGG1p tagged with a HA hemagglutinin epitope (43,44) with or without coexpression of PDR1p tagged with a Myc epitope (37). Protein complexes were immunoprecipitated with anti-Myc antibody. After extensive washing, proteins were eluted from the matrix, separated by SDS-PAGE, and subsequently Western-blotted with anti-HA antibody to detect NGG1p (Fig. 3A). HA-NGG1p migrates with an apparent size of 116 kDa; this band is only detected in extracts prepared from strains expressing HA-NGG1p (compare lanes 1 and 2 with 3). A 116-kDa anti-HA-reactive band, corresponding to HA-NGG1p, was found to coimmunoprecipitate with Myc-PDR1p (compare lanes 4 and 5). Its identity as HA-NGG1p was confirmed by the absence of this band in an immunoprecipitate from a strain expressing Myc-PDR1p, but lacking HA-NGG1p (data not shown).
To confirm the interaction, the reciprocal experiment was performed. HA-NGG1p was immunoprecipitated from whole cell extracts with anti-HA antibody coupled to Sepharose beads, and the presence of Myc-PDR1p was assayed by Western blotting with anti-Myc antibody (Fig. 3B). A Myc-reactive band of 120 kDa that corresponds to Myc-PDR1p (compare lanes 1 and 2), coimmunoprecipitates with HA-NGG1p (lane 4). The coimmunoprecipitation of PDR1p and NGG1p from whole cell extracts verifies the interaction that was detected in the two-hybrid analysis. This interaction may be indirect since PDR1p was not found in immunoprecipitates with HA-tagged NGG1p 1-373 when the proteins were transcribed and translated in vitro (data not shown).
NGG1p Also Interacts with the Carboxyl-terminal Activation Domain of PDR3p-Mutations in a second genetic locus, PDR3, result in similar multidrug resistance phenotypes as for PDR1 (31,45,46). PDR3p and PDR1p share 36% amino acid identity (31) and are thought to recognize the same promoter elements (45,47). Two regions of PDR3p, amino acids 1-109 and 795-976, activate transcription when fused to the LexA DNA binding domain (31). The latter activation domain contains two stretches of 21 and 24 amino acids that are 62% and 71% identical, respectively, with the carboxyl-terminal domain of PDR1p (Fig. 4). Based on the sequence similarity and the apparent functional overlap of PDR1p and PDR3p, we tested for interaction between NGG1p 1-373 and the carboxyl-terminal region of PDR3p by two-hybrid analysis. Y190 containing either GAL4 DBD -NGG1p 1-373 or the nonspecific bait protein, GAL4 DBD -SNF1p, were transformed with a plasmid expressing the GAL4p activation domain fusion, GAL4 AD -PDR3p 765-976 . This 212-amino acid region of PDR3p corresponds to the carboxyl-terminal 258 amino acids of PDR1p. Resulting transformants were assayed for two-hybrid interaction in direct comparison to strains expressing GAL4 AD -PDR1p 813-1063 (Table  II). Coexpression of GAL4 AD -PDR3p 765-976 with GAL4 DBD -NGG1p 1-373 resulted in increased expression of GAL-lacZ approximately 5-fold above that with GAL4 DBD -SNF1p; results equivalent to those for GAL4 AD -PDR1p 813-1063 . This indicates that NGG1p interacts with both PDR1p and PDR3p and suggests that their carboxyl-terminal activation domains share a related protein motif responsible for the interaction.
Interaction of PDR1p with NGG1p 1-373 Is Dependent on NGG1 and ADA2-To evaluate whether the ADA complex is involved in mediating the interaction between NGG1p 1-373 and PDR1p 813-1063 , we performed the two-hybrid assay in yeast strains lacking components of the complex, JY335 (ngg1-1) and CY933 (ada2). Introduction of ngg1-1 and disruption of ada2 necessitated that we substitute a GAL10-lacZ reporter (6)  with GAL4 AD -PDR1 813-1063 , resulted in an 8-fold increase in expression of the GAL10-lacZ reporter as compared to coexpression with GAL4 AD -SNF4 (negative control; Fig. 5). A role for the ADA complex in the interaction was suggested by the finding that in the ngg1-1 strain, coexpression of GAL4 DBD -NGG1  and GAL4 AD -PDR1 813-1063 , did not show the same increase in expression of GAL10-lacZ. NGG1 is thus required to observe the two-hybrid interaction between PDR1p 813-1063 and NGG1p 1-373 . In contrast, GAL10-lacZ expression is only marginally decreased in the ngg1-1 strain expressing GAL4 DBD -SNF1p and GAL4 AD -SNF4p, two proteins know to interact in a two-hybrid analysis (24). This latter finding is particularly important since reduced GAL10-lacZ expression could occur from a decrease in the physical interaction between NGG1p 1-373 and GAL4 AD -PDR1 or from reduced activity of the GAL4p activation domain in the ngg1-1 background. As an additional test to verify that ngg1-1 does not affect the GAL4p activation domain, we determined the expression of GAL10-lacZ in a yeast strain expressing only a fusion of the GAL4p DNA binding domain with the carboxyl-terminal activation domain (GAL4p ⌬148 -762 ). Agreeing with results previously reported by Horiuchi et al. (12), we found that the activity of the GAL4p activation domain was only slightly reduced in JY335. We conclude that it is the interaction between NGG1p 1-373 and the carboxyl-terminal 251 amino acids of PDR1p that requires NGG1.
A comparable analysis was performed in the ada2 strain, CY933. Disruption of ada2 has no effect on the GAL4p activation domain (GAL4p ⌬148 -762 ; Fig. 5); however, the two-hybrid interaction between NGG1p 1-373 and GAL4 AD -PDR1 decreases approximately 3-fold in CY933. This result indicates that ADA2 is required for the full interaction between NGG1p 1-373 and PDR1p. Interestingly, disruption of ada2 does not have as pronounced an effect as disruption of ngg1. This suggests that ADA2p is not directly involved in contacting PDR1p, but rather may be modulating the activity or expression of a component of the complex that is directly involved.
NGG1p and ADA2p Inhibit Transcriptional Activation by PDR1p-NGG1p and ADA2p inhibit transcriptional activation by GAL4p (6,13); in contrast, these proteins are required for full activity of LexA-GCN4p and LexA-VP16 (7)(8)(9)(10). If the interaction between NGG1p and PDR1p is functionally significant, disruption of NGG1 should alter transcriptional activation by PDR1p. We analyzed expression of the GAL10-lacZ reporter (6) in wild type and ngg1-1 strains containing GAL4 DBD -PDR1p 813-1063 , a fusion of the GAL4p DNA binding domain with amino acids 813-1063 of PDR1p (Fig. 6A). The presence of a transcriptional activation domain at the carboxyl terminus of PDR1p was confirmed by the 16-fold increase in expression of GAL10-lacZ by GAL4 DBD -PDR1p 813-1063 as compared with GAL4 DBD -SNF1p in the wild type strain CY922. In the ngg1-1 strain (JY335) expression of GAL10-lacZ increased approximately 4.5-fold in the presence of GAL4 DBD -PDR1p 813-1063 as compared to CY922. This effect was not due to the influence of the GAL4p DNA binding domain because a similar increase in expression was not found in JY335 for GAL4 DBD -SNF1p. The increase in activity of GAL4 DBD -PDR1p 813-1063 in the ngg1-1 background is comparable to that for full-length GAL4p integrated into CY922 and JY335. These results demonstrate that, similar to GAL4p, the activity of the transcriptional activation domain of PDR1p is inhibited by NGG1p.
To determine if other components of the ADA complex are involved in inhibiting PDR1p-mediated activation, a similar analysis was performed with the ada2 disruption strain, CY933. For GAL4 DBD -PDR1p 813-1063 and GAL4 DBD -SNF1p, a GAL4p-regulated lacZ reporter with five tandemly repeated GAL4p DNA binding sites, his3-G4-lacZ (6) was used to increase the relative expression of ␤-galactosidase. Similar to ngg1-1 strains, disruption of ada2 resulted in a 4.5-fold increase in GAL4 DBD -PDR1p 813-1063 -mediated expression of his3-G4, while expression mediated by GAL4 DBD -SNF1p was relatively unaffected (compare CY933 with CY922; Fig. 6B). Therefore, both ADA2p and NGG1p, are involved in negatively regulating transcriptional activation by PDR1p. Based on the known interaction between NGG1p and ADA2p (12), 2 and the involvement of ADA2p and NGG1p in the interaction between PDR1p 813-1063 and NGG1p 1-373 , this suggests that the ADA complex inhibits PDR1p. DISCUSSION Using the two-hybrid system, we have found that the carboxyl-terminal 251 amino acids of the transcriptional activator protein, PDR1p, associate with amino acids 1-373 of NGG1p. This was confirmed by the coimmunoprecipitation of epitopetagged derivatives of NGG1p and PDR1p from yeast extracts. Based on the 5-fold increase in activation by GAL4 DBD -PDR1p 813-1064 in both ngg1 and ada2 strains, we also conclude that, in common with GAL4p (6, 13), NGG1p and ADA2p inhibit transcriptional activation by the activation domain of PDR1p.   ) and transformed into yeast strain Y190 containing GAL4 DBD -NGG1p 1-373 or GAL4 DBD -SNF1p (control for NGG1p-independent expression of the reporter). Transformants were grown in minimal media containing 2% glucose and assayed for expression of the GAL-lacZ reporter in direct comparison to equivalent strains containing GAL4 AD -PDR1p 813-1063 . Each value represents the average ␤-galactosidase activity of at least 3 transformants, standardized to total protein. The far right column indicates the ratio of ␤-galactosidase activity for strains containing GAL4 DBD -NGG1 1-373 versus GAL4 DBD -SNF1 as a measure of the two-hybrid interaction. (ngg1-1), and CY933 (ada2). Cells were grown in minimal media containing 2% glucose and assayed for expression of a GAL10-lacZ reporter introduced into each strain on a ADE2 centromeric plasmid. Except for strains containing GAL4p ⌬148 -762 , ␤-galactosidase assays were performed after glass bead disruption of cells as described by Himmelfarb et al. (34) and standardized to protein concentration. For strains containing GAL4p ⌬148 -762 , ␤-galactosidase activity was measured according to Ausubel et al. (35) and standardized to cell density. Each value represents the average from a minimum of three independent transformants. nd, not determined.
The NGG1p-PDR1p Association May Be Mediated by a Component of the ADA Complex-Our experiments suggest that the interaction between NGG1p and PDR1p may be indirect. First, NGG1p and PDR1p did not coimmunoprecipitate when translated in vitro. Second, both ADA2 and NGG1 were required for the interaction of NGG1p 1-373 with PDR1p. The requirement for NGG1 is particularly interesting. Full-length NGG1p may be required to form a dimer with NGG1p 1-373 , which then directly, or indirectly, associates with PDR1p 813-1063 . We think that this is unlikely based on our inability to demonstrate an interaction between NGG1p and NGG1p 1-373 by two-hybrid analysis (data not shown). Alternatively, full-length NGG1p may be required for the stability or expression of a component of the ADA complex mediating the interaction between NGG1p 1-373 and PDR1p 813-1063 . Consistent with this, we have found that the stability of ADA2p and NGG1p are dependent on the presence of each other. 2 ADA2p, itself, is a possible candidate for mediating the interaction between NGG1p and PDR1p, since it has been found to interact with the VP16 activation domain in vitro (14). While we cannot exclude this possibility, the fact that disruption of ADA2 does not completely eliminate the two-hybrid interaction, and that NGG1p 1-373 lacks the carboxyl-terminal region of NGG1p that interacts with ADA2p (12), suggests that a third component is involved. The presence of this component must be less dependent on ADA2p than NGG1p.
To date, GCN5p is the only other defined protein in the complex; however, we have recently found NGG1p and ADA2p as part of a complex that fractionates with an apparent size of greater than 1 ϫ 10 6 kDa by gel filtration. 2 Coimmunoprecipitation studies with epitope-tagged NGG1p also reveal at least five interacting proteins in addition to ADA2p. 2 Similar studies using a derivative of NGG1p lacking amino acids 274 -307, the region required for the PDR1p interaction, revealed a similar but slightly smaller complex containing NGG1p. 2 A protein that associates with NGG1p through interactions involving amino acids 274 -307 may directly contact the activation domain of PDR1p. Clearly, this region of NGG1p is involved with the function of NGG1p, since its deletion results in loss of inhibition of GAL4p by NGG1p and imparts the slow growth phenotype typical of disruption of ngg1 (13).
Interaction of NGG1p with PDR1p and PDR3p-The primary role of PDR1p in the multidrug resistance response is evident from the high frequency of PDR1 mutations that result in resistance to nearly 30 chemicals that affect cytoplasmic or mitochondrial functions (reviewed in Ref. 48). PDR1p regulates the expression of nine or more yeast genes, including membrane transporter proteins such as PDR5p (40,41) and SNQ2p (25,49). Mutations conferring similar multidrug resistant phenotypes map to PDR3 (45,46,50,51). These data (along with the finding that PDR1 and PDR3 cross-complement null alleles, have related DNA binding motifs that bind the same consensus (45,47), and overall share 36% amino acid identity (31,39,45,52)) indicate that these two regulators functionally overlap (25).
Our finding that NGG1p also associates with the carboxylterminal activation domain of PDR3p suggests that PDR3p and PDR1p share a motif that is involved in recruiting NGG1p. Two candidate regions are stretches of 21 and 24 amino acids having 62% and 71% identity starting at residues 858 and 1035, respectively, of PDR1p (Fig. 4). Interestingly, several mutations of PDR1p that result in increased drug resistance cluster around amino acids 800-1000 (25). These mutations may prevent the association between NGG1p and PDR1p allowing for increased activation.
Induction of the genes involved for multidrug resistance requires an interplay between PDR1p and PDR3p. A predominant role for PDR1p in this response is suggested by the finding that deletion of PDR1 has a more pronounced effect on sensitivity to cycloheximide than deletion of PDR3 (31,45). However, PDR3p is clearly involved because the double disruption (pdr1 pdr3) has increased sensitivity as compared to either single disruption. Delahodde et al. (47) identified two sites in the PDR3 promoter, which bind either PDR1p or PDR3p. These sites are required for resistance to cycloheximide even in pdr1 strains, suggesting that autoregulation of PDR3 is involved in drug resistance. PDR1p may confer a rapid response to cellular FIG. 6. NGG1p and ADA2p inhibit transcriptional activation by PDR1p. A, transcriptional activation in ngg1-1 strains. Sequence encoding the carboxyl-terminal 251 amino acids of PDR1p (amino acids 813-1063) was fused to that for the DNA binding domain of GAL4p in pAS1 (GAL4 DBD -PDR1 813-1063 ). Transcription of a GAL10-lacZ reporter was assayed in isogenic yeast strains CY922 (wild type; stippled bars) and JY335 (ngg1-1; hatched bars) that had been transformed with plasmids containing GAL4 DBD -PDR1 813-1063 or GAL4 DBD -SNF1, or had GAL4 integrated at the ura3-52 (CY958 and CY960; see Table I). Cells were grown in minimal media containing 2% glucose and ␤-galactosidase activity quantitated for a minimum of three independent colonies from each transformation using glass bead lysis (34). The units of ␤-galactosidase labeled on the left axis correspond to values for activation by GAL4 DBD -PDR1p 813-1063 and GAL4 DBD -SNF1p, while the units on the right axis correspond to activation by GAL4p. B, transcriptional activation in ada2 strains. Transcription of his3-G4-lacZ, which contains a modified his3 promoter with five optimal GAL4p binding sites (6) was assayed in isogenic yeast strains CY922 (wild type; stippled bars) and CY933 (ada2; solid bars) that had been transformed with plasmids containing GAL4 DBD -PDR1 813-1063 and GAL4 DBD -SNF1. For the isogenic strains integrated with GAL4 (CY958 and CY959) expression of GAL10-lacZ was assayed. toxins by inducing expression of membrane transporters, such as PDR5p and enhance the response by activating expression of PDR3. Once induced PDR3p may potentiate drug resistance through further induction of membrane transporters and by activating its own autoregulatory loop (47). As inhibitors of PDR1p and probably PDR3p, NGG1p and ADA2p are involved in limiting the expression of genes involved in drug resistance, including PDR3, and may have a key role in terminating the response.
General Regulatory Role of NGG1p-Our first report on NGG1p focused on its regulation of GAL4p and in particular its action in glucose repression (6). Subsequently, we have shown that ADA2p is also required for inhibition of activation by GAL4p (13). As witnessed by their action on PDR1p, NGG1p and ADA2p can function as more general inhibitors of transcriptional activation, not limited to a role in glucose repression.
The proteins of the ADA complex are uniquely positioned to have a critical role in regulating transcription. Acting as a coactivator/repressor, this complex links components of the basal transcriptional machinery with activator proteins. The ADA proteins can provide gene-specific regulation through their association with different activators. Moreover, the complex can induce either gene-specific activation or repression as determined, presumably by the nature of the activator-complex interactions and how this influences the basal machinery.