The transcriptional adaptor protein Gcn5 has been identified as a nuclear histone acetyltransferase (HAT). Although recombinant yeast Gcn5 efficiently acetylates free histones, it fails to acetylate histones contained in nucleosomes, indicating that additional components are required for acetylation of chromosomal histones. We report here that Gcn5 functions as a catalytic subunit in two high-molecular-mass native HAT complexes, with apparent molecular masses of 0.8 and 1.8 megadalton (MD), respectively, which acetylate nucleosomal histones. Both the 0.8-and 1.8-MD Gcn5-containing complexes cofractionate with Ada2 and are lost in gcn5A, ada2A, or ada3A yeast strains, illustrating that these HAT complexes are bona fide native Ada-transcriptional adaptor complexes. Importantly, the 1.8-MD adaptor/HAT complex also contains Spt gene products that are linked to TATA-binding protein (TBP) function. This complex is lost in spt20/ada5A and spt7A strains and Spt3, Spt7, Spt20/Ada5, Ada2, and Gcn5 all copurify with this nucleosomal HAT complex. Therefore, the 1.8-MD adaptor/HAT complex illustrates an interaction between Ada and Spt gene products and confirms the existence of a complex containing the TBP group of Spt proteins as demonstrated by genetic and biochemical studies. We have named this novel transcription regulatory complex SAGA (_Spt-Ada-Gcn5-Acetyltransferase). The function of Gcn5 as a histone acetyltransferase within the Ada and SAGA adaptor complexes indicates the importance of histone acetylation during steps in transcription activation mediated by interactions with transcription activators and general transcription factors (i.e., TBP).[Key Words." Acetyltransferase; nucleosome; transcription; Spt; Ada; Gcn5] Received March 28, 1997; revised version accepted May 15, 1997.Chromatin structure has an intricate role in the regulation of eukaryotic gene transcription. Nucleosomes suppress basal transcription initiation in vivo and in vitro increasing the dependence of transcription on the function of sequence-specific activator proteins (for review, see Grunstein 1990; Owen-Hughes and Workman 1994). Chromatin structures are remodeled before or during transcription activation generating DNase I hypersensitive regions (DHSs) at transcription control elements (Hager et al. 1995;Steger and Workman 1996;Svaren and Horz 1996). Multiprotein complexes have been impliSCorresponding authors.

Yeast GCN5 is one component of a putative adaptor complex that includes ADA2 and ADA3 and functionally connects DNA-bound transcriptional activators with general transcription factors. GCN5 possesses histone acetyltransferase (HAT) activity, conceptually linking transcriptional activation with enzymatic modification at chromatin. We have identified the minimal catalytic domain within GCN5 necessary to confer HAT activity and have shown that in vivo activity of GCN5 requires this domain. However, complementation of growth and transcriptional activation in gcn5- cells required not only the HAT domain of GCN5, but also interaction with ADA2. The bromodomain in GCN5 was dispensable for HAT activity and for transcriptional activation by strong activators; however, it was required for full complementation in other assays. Fusion of GCN5 to the bacterial lexA DNA binding domain activated transcription in vivo, and required both the HAT domain and the ADA2 interaction domain. These results suggest that both functions of GCN5, HAT activity and interaction with ADA2, are necessary for targeting and acetylation of nucleosomal histones.

Transcriptional adaptor proteins are required for full function of higher eukaryotic acidic activators in the yeast Saccharomyces cerevisiae, suggesting that this pathway of activation is evolutionarily conserved. Consistent with this view, we have identified possible human homologs of yeast ADA2 (yADA2) and yeast GCN5 (yGCN5), components of a putative adaptor complex. While there is overall sequence similarity between the yeast and human proteins, perhaps more significant is conservation of key sequence features with other known adaptors. We show several functional similarities between the human and yeast adaptors. First, as shown for yADA2 and yGCN5, human ADA2 (hADA2) and human GCN5 (hGCN5) interacted in vivo in a yeast twohybrid assay. Moreover, hGCN5 interacted with yADA2 in this assay, suggesting that the human proteins form similar complexes. Second, both yADA2 and hADA2 contain cryptic activation domains. Third, hGCN5 and yGCN5 had similar stabilizing effects on yADA2 in vivo. Furthermore, the region of yADA2 that interacted with yGCN5 mapped to the amino terminus of yADA2, which is highly conserved in hADA2. Most striking is the behavior of the human proteins in human cells. First, GAL4-hADA2 activated transcription in HeLa cells, and second, either hADA2 or hGCN5 augmented GAL4-VP16 activation. These data indicate that the human proteins correspond to functional homologs of the yeast adaptors, suggesting that these cofactors play a key role in transcriptional activation.

RNA polymerase II transcription requires functional interactions between activator proteins bound to upstream DNA sites and general factors bound to the core promoter. Accessory transcription factors, such as adaptors and coactivators, have important, but still unclear, roles in the activation process. We tested physical interactions of the putative adaptor ADA2 with activation domains derived from acidic activator proteins and with certain general transcription factors. ADA2 associated with the herpesvirus VP16 and yeast GCN4 activation domains but not with the activation domain of yeast HAP4, which previously was shown to be independent of ADA2 function in vivo and in vitro. Furthermore, the amino terminus of ADA2 directly interacted with the VP16 activation domain, suggesting that ADA2 provides determinants for interaction between activation domains and the adaptor complex. Both TATA-binding protein (TBP) and TFIIB have previously been shown to interact directly with the VP16 activation domain in vitro (Stringer, K. F., Ingles, C. J., and Greenblatt, J. (1990) Nature 345, 783-786; Lin, Y. S., Ha, I., Maldonado, E., Reinberg, D., and Green, M. R. (1991) Nature 353, 569-571). Interestingly, when binding was tested between VP16 and these general factors in yeast nuclear extracts, both factors interacted with VP16, but only the TBP/VP16 association was dependent on ADA2. In addition, ADA2 physically associated with TBP, but not with TFIIB. These results suggest that the role of ADA2 in transcriptional activation is to promote physical interaction between activation domains and TBP.