The ADA genes encode factors which are proposed to function as transcriptional coactivators. Here we describe the cloning, sequencing, and initial characterization of a novel ADA gene, ADA1. Similar to the previously isolated ada mutants, ada1 mutants display decreases in transcription from various reporters. Furthermore, ADA1 interacts with the other ADAs in the ADA/GCN5 complex as demonstrated by partial purification of the complex and immunoprecipitation experiments. We estimate that the complex has a molecular mass of approximately 2 MDa. Previously, it had been demonstrated that ada5 mutants displayed more severe phenotypic defects than the other ada mutants (G. A. Marcus, J. Horiuchi, N. Silverman, and L. Guarente, Mol. Cell. Biol. 16:3197-3205, 1996; S. M. Roberts and F. Winston, Mol. Cell. Biol. 16:3206-3213, 1996). ada1 mutants display defects similar to those of ada5 mutants and different from those of the other mutants with respect to promoters affected, inositol auxotrophy, and Spt- phenotypes. Thus, the ADAs can be separated into two classes, suggesting that the ADA/GCN5 complex may have two separate functions. We present a speculative model on the possible roles of the ADA/GCN5 complex.

In this report we described the cloning and characterization of ADA5, a gene identified by resistance to GAL4-VP16-mediated toxicity. ADA5 binds directly to the VP16 activation domain but not to a transcriptionally defective VP16 double point mutant. Double mutants with mutations in ada5 and other genes (ada2 or ada3) isolated by resistance to GAL4-VP16 grow like ada5 single mutants, suggesting that ADA5 is in the same pathway as the other ADA genes. Further, ADA5 cofractionates and coprecipitates with ADA3. However, an ada5 deletion mutant exhibits a broader spectrum of phenotypes than mutants with null mutations in the other ADA genes. Most interestingly, ADA5 is identical to SPT20 (S.M. Roberts and F. Winston, Mol. Cell. Biol. 16: 3206-3213, 1996), showing that it shares phenotypes with the ADA and SPT family of genes. Of the other SPT genes tested, mutants with mutations in SPT7 and, strikingly, SPT15 (encoding the TATA-binding protein) show resistance to GAL4-VP16. We present a speculative pathway of transcriptional activation involving the ADA2-ADA3-GCN5-ADA5 complex and the TATA-binding protein.

Mutations in yeast ADA2, ADA3, and GCN5 weaken the activation potential of a subset of acidic activation domains. In this report, we show that their gene products form a heterotrimeric complex in vitro, with ADA2 as the linchpin holding ADA3 and GCN5 together. Further, activation by LexA-ADA3 fusions in vivo are regulated by the levels of ADA2. Combined with a prior observation that LexA-ADA2 fusions are regulated by the levels of ADA3 (N. Silverman, J. Agapite, and L. Guarente, Proc. Natl. Acad. Sci. USA 91:11665-11668, 1994), this finding suggests that these proteins also form a complex in cells. ADA3 can be separated into two nonoverlapping domains, an amino-terminal domain and a carboxyl-terminal domain, which do not separately complement the slow-growth phenotype or transcriptional defect of a delta ada3 strain but together supply full complementation. The carboxyl-terminal domain of ADA3 alone suffices for heterotrimeric complex formation in vitro and activation of LexA-ADA2 in vivo. We present a model depicting the ADA complex as a coactivator in which the ADA3 amino-terminal domain mediates an interaction between activation domains and the ADA complex.

A selection for yeast mutants resistant to GAL4-VP16-induced toxicity previously identified two genes, ADA2 and ADA3, which may function as adaptors for some transcriptional activation domains and thereby facilitate activation. Here we identify two new genes by the same selection, one of which is identical to GCN5. We show that gcn5 mutants share properties with ada mutants, including slow growth, temperature sensitivity and reduced activation by the VP16 and GCN4 activation domains. Double mutant studies suggest that ADA2 and GCN5 function together in a complex or pathway. Moreover, we demonstrate that GCN5 binds to ADA2 both by the two-hybrid assay in vivo and by co-immunoprecipitation in vitro. This suggests that ADA2 and GCN5 are part of a heteromeric complex that mediates transcriptional activation. Finally, we demonstrate the functional importance of the bromodomain of GCN5, a sequence found in other global transcription factors such as the SWI/SNF complex and the TATA binding protein-associated factors. This domain is not required for the interaction between GCN5 and ADA2 and thus may mediate a more general activity of transcription factors.