The Adenovirus E1A Protein Targets the SAGA but Not the ADA Transcriptional Regulatory Complex through Multiple Independent Domains*

Expression of the adenovirus E1A protein in the simple eukaryote Saccharomyces cerevisiae inhibits growth. We tested four regions of E1A that alter growth and transcription in mammalian cells for their effects in yeast when expressed as fusions to the Gal4p DNA binding domain. Expression of the N-terminal/conserved region (CR) 1 or CR3, but not of the CR2 or the C-terminal portion of E1A, inhibited yeast growth. Growth inhibition was relieved by deletion of the genes encoding the yGcn5p, Ngg1p, or Spt7p components of the SAGA transcriptional regulatory complex, but not the Ahc1p component of the related ADA complex, indicating that the N-terminal/CR1 and CR3 regions of E1A target the SAGA complex independently. Expression of the pCAF acetyltransferase, a mammalian homologue of yGcn5p, also suppressed growth inhibition by either portion of E1A. Furthermore, the N-terminal 29 residues and the CR3 portion of E1A interacted independently with yGcn5p and pCAF in vitro . Thus, two separate regions of E1A target

The products of the human adenovirus type 5 E1A (early region 1A) gene function as potent regulators of growth and gene expression (1,2). There are two major E1A proteins of 289 and 243 amino acids that differ only by the presence of an internal sequence of 46 amino acids in the larger protein.
Comparison of the E1A sequence of multiple adenovirus serotypes has identified four distinct regions of sequence conservation (3), designated conserved regions (CR) 1 1, 2, 3, and 4 (see Fig. 1A). CR3 coincides almost exactly with the region unique to the 289-amino acid protein.
Genetic studies in the budding yeast Saccharomyces cerevisiae have demonstrated that one or more targets of E1A are conserved in this simple eukaryote. When introduced into many haploid strains of S. cerevisiae, E1A inhibits growth, leading to an accumulation of cells in the G 1 phase of the cell cycle (19 -21). Mutations within the N-terminal/CR1 and CR3 portions of E1A impair growth inhibition (19). The N-terminal/ CR1 and CR3 domains are both important for the transcriptional activities of E1A in mammalian cells (1), suggesting that E1A may interact with conserved regulatory pathways in both types of organisms to alter transcription. Several additional lines of evidence support this hypothesis. First, yeast with impaired cyclic AMP signaling are insensitive to growth inhibition by E1A (19). In mammalian cells, E1A is known to act in synergy with cyclic AMP to activate gene expression (22,23), suggesting a remarkable conservation of function between higher and lower eukaryotes. Second, SNF/SWI-dependent transcriptional activation is required for growth inhibition by the N-terminal/CR1 portion of E1A, suggesting that transcriptional activation by this fragment of E1A mediates toxicity (24).
In this study, we show that residues 1-29, 30 -69, or the CR3 portion of E1A were each sufficient to inhibit yeast growth when expressed as fusions to the Gal4p DNA binding domain (DBD). Growth inhibition by any portion of E1A required an intact yeast SAGA (Spt-Ada-Gcn5-acetyltransferase) complex * This work was supported in part by grants from the Canadian Institutes of Health Research (MOP#14631 and MOP#49448, the latter to P. G. W.), The London Health Sciences Centre, and The University of Western Ontario Academic Development Fund (to J. S. M.). 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.
¶ but was not related directly to transcriptional activation by E1A. We further show a physical interaction of yGcn5p and the homologous mammalian protein pCAF with residues 1-29 and the CR3 portion of E1A.

MATERIALS AND METHODS
Yeast Strains, Media, and Plasmid Constructions-Strains used in this study are shown in Table I. Yeast culture media were prepared using standard techniques (25). The yeast expression vector pAS1U (URA3 marker) was constructed from pAS1 (TRP1 marker) (26) by subcloning the XbaI-NaeI fragment of pAS1 into the same sites of pRS426 (27). Similarly, pAS1L (LEU2 marker) was constructed by subcloning a PvuII fragment of pAS1 into the same site of pRS425 (27). The N-terminal/CR1 domain of adenovirus type 5 E1A (amino acids 1-82) was expressed as a fusion with the Gal4p DBD (amino acids 1-147) by subcloning an EcoRI-BamHI fragment from pMA424.82T (19) into pAS1U and pAS1L. The C-terminal domain of E1A (residues 187-289) was expressed as a fusion with the Gal4p DBD by subcloning an EcoRI-BamHI fragment from pMA-Ex2 (28) into pAS1U and pAS1L. The sequences encoding residues 93-138 (CR2), 139 -204 (CR3), 1-29, 30 -69, or 70 -82 of adenovirus type 5 E1A were PCR-amplified and subcloned as EcoRI-BamHI, EcoRI-SalI, or EcoRI-XhoI fragments into pAS1U and pAS1L. Plasmid pADHVP16 (29), which expresses the herpes simplex virus VP16 transcriptional activation domain fused to the Gal4p DBD, was obtained from Dr. M. M. Smith (University of Virginia, Charlottesville, VA). Fragments of pCAF encoding residues 1-352 and 310 -832 were generated by PCR with specific oligonucleotide primers and cloned into the vectors pJG4 -5 (OriGene Technologies Inc., Rockville, MD) or pMAL-c2X (New England Biolabs, Mississauga, Ontario, Canada). A vector expressing residues 310 -832 of pCAF was also constructed using a double point mutant of pCAF that lacks histone acetyltransferase activity as a template (30). For growth inhibition experiments, in combination with pCAF, the coding region for the N-terminal/CR1 and CR3 portions of E1A were subcloned into the yeast vector pRS426-GAL, which contains the GAL1 rather than the ADH1 regulatory region controlling expression of amino acids 1-147 of Gal4p. The yeast expression vector pGuN and the variant pGuN-55, which expresses the adenovirus type 2 E1B 55-kDa protein (31), were obtained from Dr. A. Tartakoff (Case Western Reserve University, Cleveland, OH).
Growth Suppression Assay-Yeast transformations were performed using a modified lithium acetate procedure as described previously (32). Cells were plated onto appropriate synthetic complete omission plates and incubated at 30°C. For assays of growth inhibition by E1A or E1B, plates were photographed 48 or 72 h posttransformation using a Foto/Eclipse Fotodyne gel doc system (Fotodyne Inc., Hartland, WI).
␤-Galactosidase Assays-Colonies of the yeast strain Y190 transformed with plasmids expressing various portions of E1A fused to the Gal4p DBD were picked and used to inoculate 5 ml of glucose-supplemented synthetic complete medium lacking uracil and leucine. Cultures were grown overnight at 30°C with agitation. Cells were collected by centrifugation at 4000 ϫ g for 5 min and washed twice with 5 ml of double distilled H 2 O. ␤-Galactosidase assays were performed as described previously (25). Enzymatic activity was calculated as (A 420 )/ (A 600 ϫ culture volume (ml) ϫ reaction time (min)).
Glutathione S-transferase (GST) Pull Downs-GST-E1A fusion proteins were expressed in Escherichia coli BL21 bacteria and purified as per the protocol provided by the affinity resin manufacturer (Amersham Biosciences). In vitro-translated [ 35 S]methionine-labeled full-length yGcn5p (obtained from Dr. S. Berger, The Wistar Institute, Philadelphia, PA) or the carboxyl portion of pCAF spanning amino acids 310 -832 were prepared using the TNT T7 coupled transcription/translation system (Promega) according to the supplied protocol. To immobilize GST-E1A fusion proteins, 10 g of the appropriate GST-E1A fusion protein was incubated with 25 l of glutathione-Sepharose beads in PC buffer (50 mM Tris/HCl, pH 7.4, 300 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 5 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 0.1% Triton X-100) in a total volume of 200 l for 20 min at 4°C. For GST pull-down assays, immobilized GST-E1A fusion proteins were incubated with 100,000 cpm of in vitro-translated [ 35 S]methionine-labeled yGcn5p or pCAF in pull-down buffer (50 mM HEPES/KOH, pH 7.5, 150 mM KCl, 1 mM EDTA, 1 mM Na 3 VO 4 , 10 mM NaF, 10% glycerol, 0.1% Nonidet P-40, 2 g/l bovine serum albumin) containing Complete protease inhibitor mixture (Roche Diagnostics) in a total volume of 200 l for 1 h at 4°C with constant rotation. After extensive washes with pull-down buffer, interacting proteins were eluted by boiling for 2 min in 10 l of SDS sample buffer, resolved on 10% SDS-polyacrylamide gels, and visualized by autoradiography.
Histone Acetyltransferase (HAT) Assays-GST, GST-E1A fusions, maltose-binding protein (MBP), and MBP-pCAF fusion proteins were expressed in and purified from the BL21 strain of E. coli as per protocols provided by the affinity resin manufacturers (Amersham Biosciences and New England Biolabs) and dialyzed against MBP column buffer. 5 g of recombinant MBP-pCAF was mixed with 100 g of the appropriate GST-E1A fusion proteins and incubated for 15 min on ice. Reaction mixtures were then mixed with 10 l of HAT buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 0.1% (v/v) Nonidet P-40), 1.0 l of acetyl-[ 3 H] CoA (ICN Biomedical Research Products, Costa Mesa, CA), and 15 g of histone II-A (Sigma) or 15 g of bovine serum albumin as a negative control. After 45 min at 30°C, supernatant for each reaction was spotted onto circles of p81 filter paper (Whatman Nuclepore Canada, Toronto, Ontario, Canada). The air-dried filters were soaked in sodium carbonate/bicarbonate solution (50 mM NaHCO 3 , 50 mM Na 2 CO 3 ) for 30 min at 37°C, washed once in 30 ml of a 1:1:1 mixture of methanol:chloroform:acetone for 10 min, and washed twice more for 5 min. The filters were air-dried, and the amount of bound [ 3 H]acetyl was measured by liquid scintillation counting.

The N-terminal/CR1 or CR3 Portions of E1A Can Inhibit
Growth Independently-Expression of the adenovirus type 5 E1A protein suppresses growth when expressed in haploid strains of S. cerevisiae (19 -21). Mutations within the N-terminal/CR1 and CR3 portions of E1A impair growth inhibition, although the N-terminal/CR1 portion of E1A is sufficient on its own to block growth when fused to the Gal4p DBD (19 -21, 33).
To determine whether other portions of E1A could inhibit growth independently of the N-terminal/CR1 region, we expressed the CR2, CR3, and C-terminal regions of E1A as fusions with the Gal4p DBD and tested them for their effect on yeast growth (Fig. 1). Like the N-terminal/CR1 portion of E1A, expression of the Gal4p DBD-CR3 chimera inhibited growth. In contrast, the Gal4p DBD or the Gal4p DBD fused to the CR2 or C-terminal regions of E1A did not affect growth (Fig. 1B). These results demonstrate that CR3, like the N-terminal/CR1 Mat a his3⌬1 leu2⌬0 met15⌬0 ura3⌬0 Invitrogen Corporation 1799 Mat a his3⌬1 leu2⌬0 met15⌬0 ura3⌬0 ahc1ϻkanMX Invitrogen Corporation E1A Interaction with the SAGA Complex portion of E1A, is sufficient to deregulate yeast growth independently.
The N-terminal/CR1 or CR3 Domains of E1A Can Activate Transcription in Yeast-Studies in the simple eukaryote S. cerevisiae have shown that fusion of a strong transcriptional activation domain to the Gal4p DBD can inhibit yeast growth (29). We reasoned that growth inhibition by the Nterminal/CR1 and CR3 domains E1A could be produced by a similar effect, as both of these regions of E1A function as transcriptional activation domains when fused to a DNA binding domain in mammalian cells (34,35). We tested the ability of the N-terminal/CR1, CR2, CR3, and C-terminal regions to function as activation domains in yeast when fused to the Gal4p DBD (Fig. 2). We transformed yeast strain Y190, which contains an integrated Gal4p-dependent ␤-galactosidase reporter gene, with expression vectors for the Gal4p DBD or the four Gal4p DBD-E1A fusions. The N-terminal/CR1 fragment of E1A strongly activated transcription in this system, and a lesser degree of transcriptional activation was observed for CR3. In contrast, neither the CR2 nor C-terminal regions activated ␤-galactosidase expression. Thus, the regions of E1A that inhibit growth in yeast (Fig. 1B) also function to activate transcription, raising the possibility that these two activities are related.
Growth Inhibition Mediated by E1A Requires the SAGA Transcriptional Activation Complex-Growth inhibition mediated by fusion of the herpes simplex virus VP16 transcriptional activation domain to the Gal4p DBD has been proposed to result from sequestration of limiting components of the transcription apparatus at genomic sites (29). This is supported by the observation that a Gal4p DBD-VP16 chimera does not inhibit growth in yeast strains disrupted for components of the SAGA yeast transcriptional activation complex (29,36,37). We tested the ability of the Gal4p DBD N-terminal/CR1 and CR3 chimeras to inhibit growth in yeast strains with disruptions of NGG1, GCN5, and SPT7 (ngg1⌬, gcn5⌬, and spt7⌬, respectively), which encode components of the SAGA complex (Fig. 3). As reported previously (36 -38), Gal4p DBD-VP16 inhibited growth in wild-type yeast strains but not in strains disrupted for NGG1 or GCN5. Interestingly, although Gal4p DBD fused to the N-terminal/CR1 or CR3 domains was toxic in matched wild-type strains, neither chimera inhibited growth in ngg1⌬, gcn5⌬, or spt7⌬ strains. However, growth inhibition by the N-terminal/CR1 or CR3 regions was not affected by deletion of AHC1 (Fig. 4), which encodes a component specific to the yeast ADA transcriptional activation complex (39). These results demonstrate that growth inhibition mediated by either the N-terminal/CR1 or CR3 domains of E1A depends on the presence of an intact SAGA complex but not the related ADA complex. Mutation of SAGA components relieved growth suppression by E1A and VP16 specifically, as disruption of NGG1 had no effect on growth inhibition by the adenovirus E1B 55-kDa protein (data not shown).
The N-terminal/CR1 or CR3 Domains of E1A Interact Independently with yGCN5p and pCAF-Our observation that growth inhibition by either the N-terminal/CR1 or CR3 domains of E1A requires an intact SAGA complex suggests that both these portions of E1A interact with components of the complex. The N-terminal/CR1 portion of E1A is known to interact with pCAF, a mammalian homologue of yGcn5p (4), and E1A has been shown to coprecipitate an HAT activity when expressed in yeast (40), suggesting that this region of E1A may also target yGcn5p. We expressed the N-terminal/CR1, CR2, CR3, and C-terminal portions of E1A as fusions to GST and tested their ability to interact with in vitro-transcribed and -translated yGCN5p (Fig. 5A). An interaction was observed between yGcn5p and the N-terminal/CR1 and CR3 but not the CR2 or C-terminal portions of E1A. Otherwise identical experiments performed using in vitro-transcribed and -translated pCAF yielded similar results (Fig. 5B). These experiments show that the N-terminal/CR1 portion of E1A can interact not only with pCAF but with the homologous yGcn5p, as well. Importantly, this observation also suggests that CR3 can target these two related proteins independently of the N-terminal/ CR1 portion of E1A.
Expression of pCAF Suppresses Growth Inhibition by E1A in Yeast-Based on the results presented above, growth inhibition by either the N-terminal/CR1 or CR3 portions of E1A likely results from an interaction with yGcn5p. We next tested whether expression of pCAF, a mammalian homologue of yGcn5p, would restore growth to yeast expressing these regions of E1A. We cotransformed yeast with vectors expressing either the N-terminal/CR1 or CR3 portions of E1A and vectors expressing either the amino portion of pCAF (pCAF 1-352 ) or the carboxyl portion shown previously to interact with E1A (pCAF 310 -832 ) (4). Yeast expressing E1A and pCAF 1-352 continued to grow slowly, whereas expression of pCAF 310 -832 restored rapid growth to yeast expressing either fragment of E1A  Fig. 1A were prepared as fusions to GST and used in GST pull-down assays with 35 S-labeled full-length yGcn5p or the Cterminal portion of pCAF as described under "Materials and Methods." Proteins were recovered with glutathione-Sepharose and analyzed by SDS-PAGE and radiography. 1/10 of the input of 35 S-labeled proteins used in the binding reactions is shown for comparison. (Fig. 6). The observation that pCAF expression suppresses growth inhibition mediated by either the N-terminal/CR1 or CR3 portions of E1A supports our observation that pCAF interacts independently with each of these regions (see Fig. 3 and Fig. 5). Interestingly, expression of the carboxyl portion of pCAF containing two point mutations that abolish HAT activity was still able to restore rapid growth to yeast expressing E1A. As suppression of E1A induced growth inhibition was independent of the enzymatic activity of pCAF, it likely does not result from a complementation of yGcn5p activity by mammalian pCAF. Instead, the expression of pCAF may compete with endogenous yGcn5p for interaction with E1A, effectively preventing binding to yGcn5p.
E1A Reduces Acetyltransferase Activity of pCAF in Vitro-Full-length E1A has been shown to inhibit the acetyltransferase activity of pCAF in vitro (41). We next examined the consequences of the interaction with either the N-terminal/CR1 or CR3 portions of E1A on the acetyltransferase activity of pCAF. Recombinant pCAF fused to the MBP was purified from E. coli and used in an in vitro HAT assay alone or in combination with portions of E1A fused to GST (Fig. 7). As observed previously for full-length E1A, either the N-terminal/CR1 or CR3 portions of E1A, which we show interact with pCAF (Fig.  5), reduced the ability of pCAF to acetylate free histones (Fig.  7). This reduction in acetylation was greater than that observed with the CR2 or C-terminal portions of E1A, which do not interact with pCAF.
yGcn5p and pCAF Interact with Residues 1-29 of E1A, and This Region Is Sufficient for Growth Inhibition in Yeast-We next determined the subregion of the N-terminal/CR1 fragment of E1A required for interaction with pCAF. Portions of E1A corresponding to amino acid residues 1-29, 30 -69, and 70 -82 of E1A were expressed as fusions to glutathione S-transferase and tested for interaction with in vitro-transcribed and-translated yGcn5p or pCAF. Residues 1-29 were sufficient for inter-action with either yGcn5p or pCAF (Fig. 8, B and C). These same portions of E1A were expressed in yeast as fusions to the Gal4p DBD and examined for their ability to inhibit growth (Fig. 9B) and activate transcription (Fig. 9C). The fragment corresponding to residues 70 -82 had no effect on growth. Residues 1-29 reduced yeast growth, confirming further that targeting of yGcn5p by E1A is sufficient to deregulate growth in yeast. Interestingly, the fragment of E1A spanning residues 30 -69 was also able to inhibit growth independently. Growth inhibition by either residues 1-29 or 30 -69 of E1A was lost in a strain disrupted for GCN5 but not in a strain disrupted for AHC1 (Fig 9B). This result indicates that residues 1-29 and 30 -69 of E1A both target the SAGA but not the ADA complex. As the fragment spanning residues 30 -69 does not bind yGcn5p in vitro (Fig. 8B), its seems likely that it targets a separate component of the SAGA complex. Surprisingly, unlike the entire N-terminal/CR1 portion of E1A, none of the small fragments, which collectively span that same region, were able to activate transcription (Fig. 9C). These results indicate that interaction with yGcn5p is not sufficient for transcriptional activation and that the ability of E1A to inhibit yeast growth is unrelated to its ability to activate transcription. DISCUSSION We report here that either the N-terminal/CR1 or CR3 domains of E1A are sufficient to inhibit yeast growth when expressed as fusions with the Gal4p DBD. In contrast, neither the CR2 nor C-terminal domains of E1A affected yeast growth when fused to the Gal4p DBD (Fig. 1B). These results agree with previous reports demonstrating that the N-terminal/CR1 domain can inhibit yeast growth (19) and with the observation that deletions within either the N-terminal/CR1 or CR3 domains impair growth inhibition (19,21). Our results suggest that the CR3 domain, like the N-terminal/CR1 domain, is targeting a cellular function(s) conserved in yeast and raises the possibility that genetic analysis of CR3 function in yeast will provide useful insight into the mechanisms by which this region of E1A functions in higher eukaryotic cells.
In mammalian cells, both the N-terminal/CR1 and CR3 domains are equally potent activators of gene expression when fused to the Gal4p DBD (35,42). When tested in yeast, these two regions also functioned to activate transcription when fused to the Gal4p DBD (Fig. 2). This raised the possibility that their ability to inhibit growth results from activation of transcription at inappropriate genomic sites. However, this is not entirely consistent with our observation that although the Nterminal/CR1 and CR3 domains were similarly proficient at inhibiting growth (Fig 1B), transcriptional activation by the CR3 domain was much weaker than that observed for the N-terminal/CR1 domain (Fig. 2). In addition, although residues 1-29 or 30 -69 of E1A are sufficient for growth inhibition (Fig.  9B), neither is capable of activating transcription (Fig. 9C). Thus, growth inhibition is clearly not related to the ability of E1A to activate transcription and likely results from trapping of limiting cellular factors.
Previous studies in yeast using the strong herpes simplex virus transcriptional activator VP16 showed that a Gal4p DBD-VP16 chimera does not inhibit growth in yeast strains disrupted for components of the SAGA yeast transcriptional activation complex (36 -38). As observed for VP16, neither the N-terminal/CR1 or CR3 portions of E1A inhibited growth in yeast strains in which the genes encoding the Gcn5p, Ngg1p, or Spt7p components of the SAGA transcriptional activation complex were disrupted. However, both portions of E1A inhibited growth in matched wild-type strains (Fig. 3) or in yeast in which the gene encoding the Ahc1p component of the ADA transcriptional regulatory complex was disrupted (Fig. 4). These results demonstrate that growth inhibition mediated by the N-terminal/CR1 and CR3 portions of E1A requires a functional SAGA complex and that growth inhibition is independent of the related ADA complex, which shares the yGcn5p, Ada2p, and Ngg1p proteins in common with SAGA (39). The requirement for the SAGA complex appears specific for growth inhibition by E1A, as growth inhibition by the unrelated adenovirus E1B 55-kDa protein was not affected by disruption of the SAGA complex (data not shown).
The N-terminal/CR1 portion of E1A has been shown to interact directly with a C-terminal portion of pCAF (43). This suggested that E1A might target yGcn5p, the yeast homologue of pCAF. We confirmed that the N-terminal/CR1 portion of E1A bound pCAF using an in vitro GST pull-down approach (Fig. 5) and showed that it also binds to the related yGcn5p protein. Importantly, we observed that CR3 also bound both Gcn5p and pCAF in vitro (Fig. 5), a novel finding. We also found that expression of the E1A binding portion of pCAF restored growth to yeast expressing either the N-terminal/CR1 or CR3 portions of E1A. Importantly, the ability of this portion of pCAF to restore growth to yeast expressing E1A was independent of its intrinsic HAT activity (Fig. 6). This observation suggests that the ability of pCAF to suppress growth inhibition results from its ability to interact directly with either the N-terminal/CR1 or CR3 portions of E1A. Thus, expression of pCAF may compete with yGcn5p for interaction with E1A, effectively sequestering it from endogenous yGcn5p. Taken together, these results clearly show that two independent portions of E1A bind to yGcn5p and pCAF. As a consequence of this interaction, either the N-terminal/CR1 or CR3 portions of E1A inhibited the acetyltransferase activity of pCAF (Fig. 7). Although we did not detect any interaction of the CR2 portion of E1A with pCAF, this region also reduced the HAT activity of pCAF in vitro. This reduction was reproducible but not as pronounced as that observed with the N-terminal/CR1 or CR3 portions of E1A. A similar CR2-dependent reduction in the in vitro HAT activity of p300 has been reported (44), despite the fact that this region is not involved in p300 interaction, suggesting that it may have some general effect on this assay.
Using small portions of E1A, we determined that the Nterminal 29 residues of E1A were sufficient for interaction with Gcn5p and pCAF (Fig. 8B). A previous report indicated that a deletion spanning residues 55-60 reduces, but does not ablate, the association of E1A with pCAF (43). Reduced binding may indicate that the deleted region stabilizes the interaction with pCAF, which is supported by the observation that deletion of residues 38 -67 reduces the association of E1A with HAT activity in yeast (40). Importantly, residues 1-29 were sufficient for growth inhibition (Fig. 9B), showing that a direct interaction between this portion of E1A and yGcn5p is sufficient to inhibit growth. However, residues 1-29 were not able to acti-vate transcription, suggesting that whereas binding to yGcn5p may be necessary for transcriptional activation, it is not sufficient on its own. Thus, the recruitment of the yGcn5p or pCAF acetyltransferases must function in cooperation with additional E1A interacting factors to activate transcription.
Intriguingly, residues 30 -69 of E1A were also sufficient for growth inhibition, although they were not able to bind yGcn5p or pCAF (Fig. 8B) or activate transcription (Fig. 9C). Given that disruption of multiple components of the SAGA complex relieves growth inhibition by the entire N-terminal/CR1 por- Transformed cells were allowed to grow ϳ48 h at 30°C and photographed. C, yeast strain Y190 was transformed with vectors expressing the Gal4p DBD fusions described in A. The ability of each portion of E1A to stimulate transcription from a Gal4p-dependent ␤-galactosidase reporter gene was assayed in triplicate as described under "Materials and Methods" and is reported as percent activity of the N-terminal/CR1 fragment. Error bars indicate the standard errors. tion of E1A, it seems likely that residues 30 -69 of E1A are targeting a second distinct component of the SAGA complex. This is further supported by our observation that disruption of GCN5 blocks growth inhibition by either residues 1-29 or 30 -69 of E1A (Fig. 9B). Interestingly, an N-terminal portion of E1A spanning residues 12-54 interacts with TRRAP (45), the mammalian homologue of the Tra1p component of the SAGA complex, and the 243-amino acid E1A protein coprecipitates with fragments of Tra1p expressed as GST fusion from yeast extracts (33). These results suggest that residues 30 -69 of E1A may inhibit growth by targeting Tra1p.
In conclusion, our results demonstrate that two separate domains of E1A that function as transcriptional activators in mammalian cells retain this function in S. cerevisiae, underscoring the conservation of basic mechanisms of transcriptional regulation between yeast and higher eukaryotes. We have also shown that three independent domains of E1A have evolved to target the SAGA complex in vivo, highlighting the important role of this complex in gene regulation. This provides a new genetic system to further elucidate mechanisms by which E1A and the SAGA complex regulate transcription and growth.