Portal de Periódicos da CAPES

Continuous and widespread roles for the Swi-Snf complex in transcription

1999; Springer Nature; Volume: 18; Issue: 8 Linguagem: Inglês

10.1093/emboj/18.8.2254

1460-2075

Stephen R. Biggar, Gerald R. Crabtree,

NF-κB Signaling Pathways

Article15 April 1999free access Continuous and widespread roles for the Swi–Snf complex in transcription Stephen R. Biggar Stephen R. Biggar Departments of Developmental Biology and Pathology, Stanford University Medical School, Stanford, CA, 94305 USA Search for more papers by this author Gerald R. Crabtree Corresponding Author Gerald R. Crabtree Departments of Developmental Biology and Pathology, Stanford University Medical School, Stanford, CA, 94305 USA Search for more papers by this author Stephen R. Biggar Stephen R. Biggar Departments of Developmental Biology and Pathology, Stanford University Medical School, Stanford, CA, 94305 USA Search for more papers by this author Gerald R. Crabtree Corresponding Author Gerald R. Crabtree Departments of Developmental Biology and Pathology, Stanford University Medical School, Stanford, CA, 94305 USA Search for more papers by this author Author Information Stephen R. Biggar1 and Gerald R. Crabtree 1 1Departments of Developmental Biology and Pathology, Stanford University Medical School, Stanford, CA, 94305 USA *Corresponding author. E-mail: [email protected] The EMBO Journal (1999)18:2254-2264https://doi.org/10.1093/emboj/18.8.2254 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Chromatin presents a significant obstacle to transcription, but two means of overcoming its repressive effects, histone acetylation and the activities of the Swi–Snf complex, have been proposed. Histone acetylation and Swi–Snf activity have been shown to be crucial for transcriptional induction and to facilitate binding of transcription factors to DNA. By regulating the activity of the Swi–Snf complex in vivo, we found that active transcription requires continuous Swi–Snf function, demonstrating a role for this complex beyond the induction of transcription. Despite the presumably generalized packaging of genes into chromatin, previous studies have indicated that the transcriptional requirements for the histone acetyltransferase, Gcn5, and the Swi–Snf complex are limited to a handful of genes. However, inactivating Swi–Snf function in cells also lacking GCN5 revealed defects in transcription of several genes previously thought to be SWI–SNF- and GCN5-independent. These findings suggest that chromatin remodeling plays a widespread role in gene expression and that these two chromatin remodeling activities perform independent and overlapping functions during transcriptional activation. Introduction The precise mechanisms by which chromatin represses transcription remain unclear. Attention has focused primarily on the ability of histones to inhibit transcriptional initiation since the formation of nucleosomes on DNA templates blocks initiation by RNA polymerase II and reduces the ability of the transcriptional activators to bind DNA (Lorch et al., 1987; Matsui, 1987; Workman and Roeder, 1987; Knezetic et al., 1988; Pina et al., 1990; Taylor et al., 1991; Workman et al., 1991). Allowing RNA polymerase and activators to bind DNA prior to nucleosome assembly permits transcription in the presence of histones, suggesting that chromatin restricts access to binding sites in DNA. This model of nucleosome repression seems to hold for promoters in vivo, at which the binding of transcriptional factors to critical promoter elements appears to be blocked by nucleosomes (Almer et al., 1986; Axelrod et al., 1993; Wolffe, 1994). Activities that disturb nucleosome structure or position to permit transcription factor binding would be expected to facilitate transcriptional activation from promoters repressed by chromatin. Consistent with this hypothesis, transcriptional activation from several promoters is accompanied both by nucleosome disruption and by binding of transcription factors to previously inaccessible sites in the promoters (Almer et al., 1986; Axelrod et al., 1993; Wolffe, 1994). The identification of transcription factors whose activities appear to be dedicated to disrupting chromatin during transcriptional activation has confirmed the notion that nucleosome disruption facilitates transcription. Two major types of these chromatin remodeling activities have been identified: multisubunit ATPases such as the Swi–Snf complex and histone acetyltransferases such as Gcn5. The yeast SWI and SNF genes originally were identified in yeast screens for genes required for expression of HO (Stern et al., 1984) and SUC2 (Neigeborn and Carlson, 1984). A subset of SWI–SNF genes encode subunits of a complex believed to remodel chromatin during transcriptional activation. swi–snf mutations impair transcription and alter chromatin structures at promoters in vivo and, consistent with a role for SWI–SNF in chromatin remodeling, mutations in histone genes suppress these defects (Sternberg et al., 1987; Kruger and Herskowitz, 1991; Hirschhorn et al., 1992; Kruger et al., 1995). In addition, the purified Swi–Snf complex alters nucleosome structure in vitro (Cote et al., 1994, 1998). The mechanisms by which chromatin remodeling by Swi–Snf stimulates transcription remain unresolved. Several in vitro studies suggest that the Swi–Snf complex functions to augment DNA binding by transcription factors (Cote et al., 1994; Owen-Hughes et al., 1996), and the enhanced affinity of transcription factor binding persists even after Swi–Snf has been inactivated (Cote et al., 1998; Schnitzler et al., 1998). These results suggest a sequence of events in which Swi–Snf acts to remodel chromatin, and transcriptional activators subsequently bind their enhancer sites. In support of this model, augmenting transcription factor binding by increasing the number of activator-binding sites or the affinities of these sites relieves the SWI–SNF dependence of promoters in vivo (Laurent and Carlson, 1992; Burns and Peterson, 1997). Contrary to this model, however, the Swi–Snf complex appears to act after activator binding at some Swi–Snf-dependent promoters since transcription factor binding at these promoters is impaired minimally in swi–snf mutants (Burns and Peterson, 1997; Ryan et al., 1998). Thus, the physiological function of chromatin remodeling has not been determined conclusively. Acetylation of the N-termini of histones represents another means of altering chromatin structure during transcriptional activation. Several studies have established a correlation between histone acetylation and transcription (Turner, 1991), and multiple transcription factors have been found to possess histone acetyltransferase activity (Kuo et al., 1996; Mizzen et al., 1996; Candau et al., 1997; Pennisi, 1997). The cloning of a histone acetyltransferase from Tetrahymena revealed that this protein had significant homology to the transcription factor Gcn5 (Brownell et al., 1996). Gcn5 functions together with Ada2 and Ada3 to mediate activation by certain transcriptional activators (Berger et al., 1992; Marcus et al., 1994). Further experimentation confirmed that Gcn5 possesses histone acetyltransferase activity and that this activity is essential for the function of GCN5 as a transcriptional co-activator (Kuo et al., 1996; Candau et al., 1997). Furthermore, multisubunit complexes containing Gcn5 (referred to here as Gcn5–Ada complexes) are capable of acetylating nucleosomal histones (Grant et al., 1997). The Swi–Snf and Gcn5–Ada complexes both appear to facilitate transcription by remodeling nucleosomes. Similar to Swi–Snf activity, histone acetylation facilitates binding of transcription factors to sites within nucleosomes (Lee et al., 1993; Juan et al., 1994; Vettese-Dadey et al., 1996), and Swi–Snf and Gcn5–Ada share at least some common target genes (Pollard and Peterson, 1997). However, the functional relationship between Swi–Snf and Gcn5–Ada remains unknown. Understanding whether the two complexes act in the same pathway to activate transcription or function independently is critical to elucidating the mechanisms of chromatin remodeling. We used genetic approaches to explore the interactions between Swi–Snf and Gcn5–Ada, and also to dissect the physiological functions of the Swi–Snf complex in vivo. Interestingly, Swi–Snf function was required continuously for transcription of several genes. Furthermore, inactivating Swi–Snf in gcn5–ada mutants revealed that the two chromatin remodeling activities act in independent and redundant pathways to activate transcription. Results Conditional alleles of SWI2 regulate Swi–Snf function in vivo To explore the physiological functions of chromatin remodeling complexes in more detail, we generated temperature-sensitive alleles of SWI2. SWI2 encodes a component of the yeast Swi–Snf complex that is required for Swi–Snf function (Laurent et al., 1993). Conditional Swi2 function would therefore allow the temporal coordination of Swi–Snf inactivation with different stages of transcriptional activation. This means of regulating Swi2 activity would also permit the growth of strains requiring SWI2 for viability, and thereby afford the opportunity to make precise assessments of the roles of the Swi–Snf complex in transcription in these strains. Swi2 contains a consensus ATPase domain, and mutations that reduce Swi2 ATPase activity debilitate transcription of SWI–SNF-dependent genes in vivo and abolish chromatin remodeling by a purified mutant complex in vitro (Laurent et al., 1993; Cote et al., 1994). Thus, Swi2 ATPase activity is crucial for Swi–Snf function. Temperature-sensitive alleles of SWI2 were generated using PCR to mutagenize the consensus ATPase domain of Swi2 within the context of the otherwise wild-type gene. Mutants were isolated on the basis of temperature-sensitive growth on glycerol media. Further analysis of one temperature-sensitive allele, swi2-6.3, suggested that temperature sensitivity was limited to the ATPase domain. At the restrictive temperature, the growth phenotype of swi2-6.3 mimics that of swi2K798A, a mutant allele of SWI2 specifically lacking ATPase activity, instead of behaving like a SWI2 deletion (Figure 1A). Consistent with these results, cells carrying either swi2-6.3 or wild-type SWI2 contained similar Swi2 protein levels at the permissive and restrictive temperatures (Figure 1B). Figure 1.(A) swi2K798A and swi2-6.3 both supported wild-type growth rates. CY120 carrying the indicated alleles of SWI2 was plated on minimal glucose media at 30 or 37°C for 3 days. (B) Swi2 protein levels were not altered by temperature in SWI2 and swi2-6.3 strains. CY120 carrying SWI2 or swi2-6.3 was grown at 26.5°C and then either shifted to 37°C or maintained at 26.5°C for 2 h. Whole-cell extracts were then prepared and Swi2p levels were determined by Western blotting. (C) Temperature-sensitive transcriptional activity of swi2-6.3. CY120 carrying SWI2 (wt), swi2-6.3 (ts) or swi2K798A (K798A) was grown at 24 or 37°C and then assayed for HO–LacZ or SUC2 expression at the same temperatures by measuring β-galactosidase or invertase activity, respectively. Fold activation relative to swi2K798A cells is plotted, and standard error is shown. Download figure Download PowerPoint The distinct growth phenotypes associated with the swi2Δ allele and the swi2-6.3 and swi2K798A alleles demonstrated that the latter mutations did not eliminate SWI2 function completely. However, SWI2 deletions and mutants lacking ATPase activity display similar transcriptional defects, suggesting that Swi2 ATPase activity is crucial for transcriptional activation by Swi–Snf (Laurent et al., 1993). To confirm that swi2-6.3 conferred temperature-sensitive transcriptional defects, we measured expression from the HO and SUC2 promoters, for which transcription depends on the Swi–Snf complex (Neigeborn and Carlson, 1984; Stern et al., 1984). Secreted invertase activity in yeast corresponds to the rate of SUC2 transcription (Carlson and Botstein, 1982) and, therefore, SUC2 expression was assessed by measuring invertase activity in cells carrying different alleles of SWI2. HO promoter activity was analyzed by measuring β-galactosidase activity in cells carrying different SWI2 alleles and an integrated HO–LacZ fusion that joins the HO promoter with the β-galactosidase open reading frame. At the permissive temperature, swi2-6.3 drove HO and SUC2 expression at 41% and 56%, respectively, of wild-type levels. At the restrictive temperature, however, swi2-6.3 cells expressed HO and SUC2 at levels similar to those seen in cells carrying swi2K798A (Figure 1C). Expression from a third SWI–SNF-dependent promoter, a variant of the GAL1 promoter (see below for details on this promoter), was also temperature sensitive in swi2-6.3 strains (see Figure 5B). Thus swi2-6.3 provided a means of using temperature to control the transcriptional activity of Swi2, and thereby the Swi–Snf complex, in vivo. In these experiments, HO expression demonstrated significant temperature sensitivity in wild-type SWI2 cells. The minimal effects of temperature on SUC2 expression in a wild-type SWI2 background suggested that the effects of temperature on HO expression reflected alterations in activities other than the Swi–Snf complex at the HO promoter. Continuous role for Swi2 in transcription Many, if not all, SWI–SNF-dependent promoters identified to date (including HO, SUC2 and INO1) share the property of rapid transcriptional induction in response to specific signals. This observation suggests that the Swi–Snf complex may be dedicated to initiating gene expression. Consistent with this possibility, transient Swi–Snf activity imparts long-term changes to chromatin structure in vitro (Owen-Hughes et al., 1996; Logie and Peterson, 1997; Cote et al., 1998; Schnitzler et al., 1998). Swi–Snf establishes structural alterations in nucleosomes that persist for up to 4 h following Swi–Snf inactivation. To determine whether the Swi–Snf complex plays a role in maintaining transcription beyond initiation in vivo, we examined the effects of inactivating Swi2 after transcription had been activated. We initially focused on SUC2 transcription. In wild-type cells, glucose represses SUC2 expression (Carlson and Botstein, 1982). Growth in media lacking glucose induces SUC2 expression, but the extremely long half-life (>180 min) of the SUC2 mRNA under these conditions precludes straightforward measurements of the rate of SUC2 transcription. However, the SUC2 message has a very short half-life in glucose media (≤10 min; Cereghino and Scheffler, 1996) and, therefore, measuring message levels in cells grown in glucose would provide accurate assessments of on-going rates of transcription. Glucose represses SUC2 transcription through the two repressors, Mig1 and Mig2 (Lutfiyya and Johnston, 1996). Therefore, to permit SUC2 expression in glucose media, our experiments were conducted using yeast carrying MIG1 and MIG2 deletions. In cells lacking MIG1 and MIG2, expression of SUC2 in glucose media depended on Swi2 activity (Figure 2A). To determine the role of Swi2 in on-going transcription, SUC2 expression was measured following a shift to the restrictive temperature in mig1Δ mig2Δ cells carrying either wild-type SWI2 or swi2-6.3. In SWI2 cells, a shift to 37°C had little effect on SUC2 expression (Figure 2A). However, when swi2-6.3 cells were shifted to 37°C, SUC2 message was reduced rapidly to the level seen in swi2K798A cells. These data demonstrate that an activated promoter still requires Swi–Snf activity for its expression. SUC2 mRNA levels dropped 8-fold after 1 h at 37°C. Given the half-life of the SUC2 message (∼10 min) and the fact that a period of 15–20 min was required for the temperature of the media to reach 37°C, these results establish SUC2 as a direct transcriptional target of Swi–Snf activity. Figure 2.(A) Swi2 activity was required continuously to maintain ongoing transcription of SUC2. Top: SUC2 expression was compared in mig1Δ mig2Δ strains (YSB16) carrying SWI2 or swi2K798A and grown in glucose media at 30°C. Total RNA was isolated and SUC2 mRNA levels were determined by RNase protection. Bottom: mig1Δ mig2Δ strains carrying SWI2 or swi2-6.3 were grown in glucose media at 26.5°C and then shifted to 37°C. Total RNA was isolated at the indicated times following the shift to 37°C and SUC2 mRNA levels were determined by RNase protection. Levels of the cytoplasmic invertase message (inv), whose expression does not depend on SWI–SNF (Neigeborn and Carlson, 1984), were measured as a control. (B) Swi2 activity was required continuously to maintain ongoing transcription of a GALp*–STE3 fusion. Left: CY120 carrying an integrated GALp*–STE3 reporter gene (YSB21) and either SWI2 or swi2K798A was grown in glucose, and then shifted into galactose (gal) for 8 h or maintained in glucose (glu). Total RNA was isolated and STE3 mRNA levels were determined by Northern blotting. The bands corresponding to the GALp*–STE3 and the endogenous STE3 messages are indicated. The STE3-specific probe also recognized a slow migrating band (marked by an asterisk) whose presence correlated with GALp* inactivity. Right: YSB21 carrying either SWI2 or swi2-6.3 was grown in glucose at 26.5°C and then moved into galactose media at 26.5°C for 8 h. The cells were then shifted to 37°C. Total RNA was isolated at the indicated times following the shift to 37°C and STE3 mRNA levels were measured by Northern blotting. The bands corresponding to the GALp*–STE3 and endogenous STE3 messages are indicated, as is the band corresponding to GALp* inactivity (*). The graph shows densitometric quantitation of the data from the right side of the figure. mRNA levels from cells grown under non-inducing conditions in the absence of galactose are also shown. Download figure Download PowerPoint To extend this analysis, we examined the effects of inactivating Swi2 on transcription from a variant of the GAL1 promoter. Unlike the SUC2 promoter, for which no classical transcriptional activators have been identified, expression from the GAL1 promoter depends on the prototypical DNA-binding transcriptional activator, Gal4. Interestingly, although expression from the full-length GAL1 promoter, which has four Gal4-binding sites, proceeds independently of Swi–Snf function, transcription from a truncated version of the promoter, which contains only two Gal4-binding sites, displays pronounced transcriptional defects in swi–snf mutants (Burns and Peterson, 1997; Gaudreau et al., 1997). The truncated GAL1 promoter is referred to here as GALp*. To assess the effects of inactivating Swi2 on on-going transcription from GALp*, the ORF and 3′-untranslated region (3′-UTR) of STE3 were fused downstream of this promoter to generate a GALp*–STE3 fusion. Since the half-life of the STE3 mRNA is short (≤4 min; Herrick et al., 1990), measuring STE3 message levels permitted accurate assessments of on-going rates of transcription from GALp*. The GALp*–STE3 transcript exceeded the length of the endogenous STE3 message by ∼150 bases but failed to produce full-length Ste3 protein. Consistent with the physiological regulation of the GAL1 promoter, expression of the GALp*–STE3 gene remained repressed in glucose media and was strongly induced in galactose (Figure 2B). Furthermore, GALp*–STE3 expression required Swi2 function (Figure 2B). In wild-type SWI2 cells grown in galactose, a shift to 37°C reduced GALp*–STE3 mRNA levels (Figure 2B). However, a reduction in the levels of the endogenous STE3 message was also seen following a shift to 37°C. The similarity between the changes in the expression of endogenous STE3 at 37°C in SWI2 and swi2-6.3 backgrounds suggests that the changes in STE3 and GALp*–STE3 mRNA levels in wild-type cells represented temperature effects that were independent of changes in Swi–Snf activity. In swi2-6.3 cells grown in galactose, a shift to 37°C rapidly extinguished GALp*–STE3 expression (Figure 2B). These results confirmed that the maintenance of on-going transcription from active promoters depends on continuous Swi2 activity. Furthermore, similar to SUC2, the rapid reduction in transcription from GALp* suggests that this promoter is a direct target of Swi–Snf function. Swi2 and the histone acetyltransferase, Gcn5, function in independent pathways to activate transcription We were interested in exploring the functional relationship between the Swi–Snf complex and the Gcn5–Ada complex during transcription. Similar to the Swi–Snf complex, the Gcn5–Ada complex has been proposed to facilitate transcription by remodeling chromatin (Pollard and Peterson, 1997). The ability of Gcn5 to acetylate nucleosomes may be responsible for, or facilitate alterations in chromatin structure necessary for gene expression (Kuo et al., 1996; Candau et al., 1997; Grant et al., 1997). Histones acetylated by Gcn5, for example, may become substrates for the Swi–Snf complex or vice versa, or these two complexes may act independently to activate transcription. Mutations in SWI–SNF genes are synthetically lethal with mutations in GCN5–ADA genes (Pollard and Peterson, 1997; Roberts and Winston, 1997), suggesting that these two complexes may perform independent functions. However, given the profound growth defects in swi–snfΔ mutants and the mild growth defects in gcn5–adaΔ mutants (Figure 1A and data not shown), the inviability of swi–snfΔ gcn5–adaΔ double mutants may represent simply the combined growth defects of the individual mutants. Also, the inviability of cells lacking the functions of both complexes precludes critical analysis of the transcriptional defects in cells lacking both Swi–Snf and Gcn5–Ada. Thus the question of whether the Swi–Snf and Gcn5–Ada complexes serve in the same or distinct pathways during transcriptional activation remains unresolved. The conditional allele of SWI2 provided a unique opportunity to analyze quantitatively the functional relationship between the Swi–Snf and Gcn5–Ada complexes. swi2-6.3 gcn5Δ strains failed to grow at 37°C on glucose media, whereas SWI2 gcn5Δ strains grew under both permissive and restrictive conditions (Figure 3A). Neither SWI2 GCN5 nor swi2-6.3 GCN5 strains demonstrated temperature sensitivity on glucose (see Figure 1A). Since swi2-6.3 supports wild-type growth rates on glucose at 37°C in wild-type GCN5 strains, the inviability of swi2-6.3 gcn5Δ mutants does not represent merely the accumulation of growth defects, but instead suggests that the activities of Swi–Snf and Gcn5–Ada complexes combine to perform some function(s) critical for cell growth. Figure 3.(A) Temperature-sensitive growth of swi2-6.3 gcn5Δ double mutants. swi2-6.3 gcn5Δ and SWI2 gcn5Δ mutants were grown at 26.5 or 37°C on rich glucose media. (B) Swi2 and Gcn5 act independently to activate SUC2 expression. Cells with the indicated GCN5 alleles and either wild-type SWI2 (wt) or swi2-6.3 (ts) were grown in 2% glucose at 26.5°C, shifted to 37°C for 1 h, and then transferred into 0.05% glucose (lo) media at 37°C for 1.5 h. Wild-type SWI2 GCN5 cells were also maintained in 2% glucose (hi) throughout the experiment. Total RNA was isolated and SUC2 mRNA levels were determined by RNase protection. Levels of the cytoplasmic invertase message (inv) were measured as a control. Relative SUC2 mRNA levels are based on message levels from SWI2 GCN5 cells maintained in 2% glucose. Download figure Download PowerPoint To address the functional interactions between the Swi–Snf and Gcn5–Ada complexes during transcriptional activation, we compared the abilities of wild-type cells, and single and double mutants to activate SUC2 transcription (Figure 3B). Relative to wild-type cells, swi2-6.3 mutants expressed reduced levels of SUC2 at the restrictive temperature, consistent with the role of SWI2 in SUC2 expression (see Figure 1C). gcn5Δ mutants also displayed mild transcriptional defects. swi2-6.3 gcn5Δ double mutants, however, expressed SUC2 at lower levels than the gcn5Δ single mutant. Thus, inactivating Swi2 in a gcn5Δ background produced pronounced transcriptional defects, showing that Swi2 plays a role in transcription independent of GCN5. These data demonstrate that Swi–Snf and Gcn5–Ada act in distinct pathways to activate SUC2 transcription. If these two complexes functioned exclusively in the same pathway, then the transcriptional defects in the swi2-6.3 gcn5Δ double mutants should have matched the defects seen in the SWI2 gcn5Δ single mutants. The observation that swi2-6.3 GCN5 and SWI2 gcn5Δ mutants drive such different levels of SUC2 transcription further supports the conclusion that the Swi–Snf and Gcn5–Ada complexes operate in distinct pathways. Inactivating Swi2 in both GCN5 and gcn5Δ backgrounds produced apparently similar reductions in SUC2 mRNA levels. However, in both cases, message levels approximate mRNA levels under conditions that repress SUC2 expression. Thus, for SUC2 transcription, the nearly complete requirement for Swi2 activity in the presence of GCN5 limited the extent of the effects of inactivating Swi2 in the absence of GCN5. We therefore explored the roles of Swi2 and Gcn5 in the expression of other genes. Redundant roles for Swi2 and Gcn5 in GAL1 expression To explore further the functional interactions between the Swi–Snf and Gcn5–Ada complexes, we compared the abilities of wild-type cells, and single and double mutants to activate transcription of the endogenous GAL1 gene. Expression from the full-length GAL1 promoter proceeds at nearly wild-type levels in swi–snf mutants (Burns and Peterson, 1997; Gaudreau et al., 1997), and transcriptional activation by Gal4, which drives GAL1 expression, displays minimal defects in gcn5–ada mutants (Marcus et al., 1994). However, the observations that expression from a truncated version of the GAL1 promoter depends on Swi–Snf (Burns and Peterson, 1997; Gaudreau et al., 1997), and that the Gal4 activation domain physically interacts with Ada2 (Melcher and Johnston, 1995), suggest that Gal4 may utilize both of these complexes to activate GAL1 expression. For our experiments, cells grown in glucose media at the permissive temperature were shifted to 37°C for 1 h and then transferred into galactose media at 37°C to induce GAL1 expression. Relative to wild-type cells, cells lacking either Swi2 or Gcn5 activity had mild defects in GAL1 induction; 1.5- and 2.7-fold, respectively (Figure 4). Cells lacking both Swi2 and Gcn5 activity, however, showed a significantly greater transcriptional defect; 12.8-fold. Gal4 levels were similar in all strains under these conditions (data not shown). Thus swi2-6.3 gcn5Δ double mutants exhibited pronounced transcriptional defects relative to SWI2 gcn5Δ single mutants, confirming the conclusion that Swi–Snf and Gcn5–Ada act in distinct pathways to activate transcription. Figure 4.Loss of GCN5 revealed a requirement for Swi2 in GAL1 expression. Cells with the indicated SWI2 and GCN5 alleles were grown in glucose at 26.5°C, shifted to 37°C for 1 h, and then either maintained in glucose (glu) or transferred into galactose (gal) for 6 h at 37°C. Total RNA was isolated, and GAL1 mRNA and 25S rRNA levels were measured by Northern blotting. GAL1 mRNA levels were quantitated after correcting for relative rRNA levels. Relative GAL1 message levels were plotted on a logarithmic scale and standard error is shown. Similar results were seen following 2 and 4 h inductions in galactose. Download figure Download PowerPoint Interestingly, the effects of inactivating Swi2 depended on the GCN5 background of the yeast. In gcn5Δ mutants, inactivating Swi2 revealed significant defects in transcription of GAL1 (4.7-fold). However, inactivating Swi2 in a wild-type GCN5 background produced only minimal defects in transcription (1.5-fold). These findings indicate that Swi–Snf and Gcn5–Ada ultimately perform redundant functions during transcription, because one complex is dispensable unless cells lack the other complex. Gal4 utilizes both Swi–Snf and Gcn5–Ada to activate transcription The defects in GAL1 induction in swi2-6.3 gcn5Δ mutants strongly support the idea that Gal4 utilizes both Swi–Snf and Gcn5–Ada to activate transcription. The differential SWI–SNF dependence of the full-length and truncated GAL1 promoters can then be explained if, at full-length promoters containing four binding sites, Gal4 provides an excess of chromatin remodeling capabilities. At promoters containing only two binding sites (GALp*), the ability of Gal4 to recruit remodeling activities may become limiting and, therefore, inactivating either complex would cripple transcriptional activation from this promoter. Consistent with this proposal, inactivating either Swi2 or Gcn5 significantly reduced transcription from GALp* (Figure 5A). Glucose represses GAL1 transcription in part through an upstream repressing sequence that binds transcriptional repressors whose activities require the presence of glucose in the media (Flick and Johnston, 1990). Transcription from the GAL1 promoter further requires galactose to unveil the activation domain of Gal4. Since the induction of GAL1 expression involves inactivation of transcriptional repressors as well as transcriptional activation by Gal4, the defects in GAL1 induction seen in swi2-6.3 gcn5Δ mutants could have reflected defects in derepression or activation, or both. Several lines of evidence argue that the defects lie in activation. First, the Gal4 activation domain requires SWI2 to activate transcription from a promoter lacking the GAL1 upstream repressing sequence (Laurent and Carlson, 1992). Secondly, synthesis of Gal4, which involves the same derepression phenomenon that occurs at the GAL1 promoter, occurred normally in swi2-6.3 gcn5Δ mutants (data not shown). Thirdly, on-going transcripti