The Gcn5·Ada Complex Potentiates the Histone Acetyltransferase Activity of Gcn5*

The Gcn5 histone acetyltransferase (HAT) is part of a large multimeric complex that is required for transcriptional activation in yeast. This complex can acetylate in vitroand in a Gcn5-dependent manner both nucleosomal and free core histones. For this reason it is believed that part of the function of the Gcn5·Ada complex is chromatin remodeling effected by histone acetylation. The roles of the other subunits of this complex are not yet known. We have generated mutated Gcn5 proteins with severely attenuated in vitro HAT activities. Despite their apparent loss in HAT activity, these GCN5 derivatives complemented all the defects of a gcn5 strain. We have shown that when these mutated proteins were produced in yeast cells in the absence of another component of the complex, Ada2, their activity was still compromised. By contrast, when produced in the wild type context, they were partially capable of acetylating free histones and were even more active when nucleosomal arrays were used as substrates. Kinetic enzymatic analyses showed that the rate of catalysis by Gcn5 was enhanced when the mutated proteins were produced in yeast in the presence of Ada2. Because Ada2 is required for the assembly of Gcn5, we conclude that one role for components of the Gcn5·Ada complex is the potentiation of its HAT activity.

The Gcn5 histone acetyltransferase (HAT) is part of a large multimeric complex that is required for transcriptional activation in yeast. This complex can acetylate in vitro and in a Gcn5-dependent manner both nucleosomal and free core histones. For this reason it is believed that part of the function of the Gcn5⅐Ada complex is chromatin remodeling effected by histone acetylation. The roles of the other subunits of this complex are not yet known. We have generated mutated Gcn5 proteins with severely attenuated in vitro HAT activities. Despite their apparent loss in HAT activity, these GCN5 derivatives complemented all the defects of a gcn5 strain. We have shown that when these mutated proteins were produced in yeast cells in the absence of another component of the complex, Ada2, their activity was still compromised. By contrast, when produced in the wild type context, they were partially capable of acetylating free histones and were even more active when nucleosomal arrays were used as substrates. Kinetic enzymatic analyses showed that the rate of catalysis by Gcn5 was enhanced when the mutated proteins were produced in yeast in the presence of Ada2. Because Ada2 is required for the assembly of Gcn5, we conclude that one role for components of the Gcn5⅐Ada complex is the potentiation of its HAT activity.
A multitude of cellular operations is accomplished through the coordinated function of multisubunit protein complexes. Individual subunits within such complexes satisfy at least one of three general requirements: an enzymatic activity, a targeting function that ensures action in the proper cellular location with the appropriate molecular contacts, and a regulatory function that controls the manifestation of these activities. It follows that a prerequisite for the mechanistic and biological elucidation of the function of a multisubunit complex is the understanding of the role that each subunit serves. Transcription of genes by the eucaryotic RNA polymerase II is one example of an operation that depends on the concerted action of a number of multisubunit complexes. Although elegant genetic and biochemical experiments have led to the molecular identification of these complexes and to the comprehension of the general aspects of their requirement, for only a small number of subunits the function is known.
Polymerase II transcription involves promoter recognition effected by TFIID, TFIIB, and TFIIA, recruitment of polymerase II-mediated by TFIIF, and transcription initiation cata-lyzed by TFIIE and TFIIH (reviewed in Refs. 1 and 2). An additional complex, the mediator, is required to convey activating signals from transcriptional activators to the basic transcriptional machinery and is part of the polymerase II holoenzyme (3,4). Finally, the function of a third group of complexes is required to alleviate the repressing effect that the nucleosomal organization of DNA imposes on all the above interactions. Such chromatin remodeling complexes include the Swi/Snf (5)(6)(7)(8), NURF (9), RSC (10), CHRAC (11), ACF (12) and allegedly the Gcn5⅐Ada complex (13,14).
Subunits of the yeast Gcn5⅐Ada complex were first identified by genetic screens aiming in the isolation of transcriptional co-activators (15,16). Genetic and biochemical studies have further indicated that Gcn5, Ada2, and Ada3 physically interact and that it is this complex that mediates transcriptional activation by a number of acidic activators (17)(18)(19)(20). It is believed that this complex is brought to the promoters through physical interactions between the Ada2 component and the activation domains (21,22). The finding that the Gcn5 component is a histone acetyltransferase (13) promoted the hypothesis that at least part of the function of the complex was the acetylation of the N termini of nucleosomal histones. This modification has been proposed to effect a conformational change in chromatin structure, thereby facilitating the process of transcription (23,24). This notion was in agreement with numerous previous observations that correlated active transcription with acetylated chromatin (reviewed in Ref. 25). Indeed, it was recently shown that the in vivo HAT 1 activity of Gcn5 is essential for the function of the complex in transcriptional activation (26,27) and that this activity correlates with the acetylation state of the chromatin within the promoter region of genes (26).
Recent biochemical analyses offered support for a second function of this complex. Gcn5, Ada2, and Ada3 have been found to co-purify in at least three biochemically distinct multisubunit complexes, 0.2, 0.8, and 2.0 MDa in size (14,28,29), all possessing Gcn5-dependent HAT activity. In addition, the largest complex was found to contain a number of proteins, among them members of the Spt family of proteins (14). It was known from genetic and biochemical studies that this Spt group of proteins have broader roles in transcriptional activation compared with the Gcn5⅐Ada proteins, and some of them interact directly with TATA-binding protein (30). All this evidence suggested that this large complex functionally links acidic activation domains (through Ada2) with the TATA-binding protein (through the Spts), whereas the HAT activity of the complex modifies nucleosomes, further stabilizing the binding of activators and TATA-binding protein on chromosomal DNA templates.
Apart from the Gcn5 HAT activity and the molecular interactions that have been defined for some components of the complex (17,19,20,(31)(32)(33), no role has yet been assigned for any other subunits of the Gcn5⅐Ada complex. Gcn5 can acetylate nucleosomal histones H3, H2B, and H4 in vitro only when located within its native complexes (14,34). By contrast, both a bacterially produced Gcn5 and a Gcn5⅐Ada2/Ada3 reconstituted trimmer were unable to modify nucleosomal substrates but were efficient in acetylating free core histones (14). These findings have suggested that other subunits within the complex regulate the Gcn5 HAT activity. In this study we demonstrate that indeed the assembly of Gcn5 within its native complex enhances the catalytic properties of its HAT activity.

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-The gcn5 and gcn5ada2 strains were obtained by appropriate gene disruptions of a GAL2 S288C parental strain with the genotype MATa trp1-1 leu2-3 ura3-52, as described in Ref. 18. Complete medium (yeast/peptone/glucose) and minimal medium supplemented with the required amino acids were used for yeast growth and transformations (35). Yeast cells were transformed using the lithium chloride method (36).
Plasmid Constructions and Protein Expression-The GCN5 coding region, generated by polymerase chain reaction, was cloned in the pALTER-Ex1 vector (Promega), and site-directed mutagenesis was performed according to the manufacturer's protocol. The mutagenic oligonucleotides were: for the His-145 3 Ala mutation (HA) 5Ј-CAGCC-ATGGAAAGTGCACTTCGATCATAGAC-3Ј and for the Cys-252 3 Ala mutation (CA) 5Ј-GGTAACATAGAAGCTTGCATCAGCGTACC-3Ј, whereas the double mutant was generated using the CA mutagenic oligonucleotide on the HA-GCN5 template. Verification of the mutagenesis was accomplished by sequence analysis.
For bacterial protein expression, wild type GCN5 and each one of the mutant genes were subcloned into the pQE-30 vector (Qiagen) and produced in Escherichia coli cells (XL1-Blue) as fusion proteins with 6-His residues at the NЈ terminus. Crude extracts from isopropyl-1-thiob-D-galactopyranoside-induced cultures were prepared by sonication in lysis buffer (50 mM Tris-HCl, pH 7.5, 20% glycerol, 1 mM EDTA, 10 mM ␤-mercaptoethanol, 0.05% Triton X-100), and recombinant proteins were purified by nickel nitrilotriacetic acid-agarose chromatography according to the technical manual (Qiagen). Constructions of the LexA-GCN5 fusion derivatives were done as in Ref. 18. Native (untagged) Gcn5 protein derivatives were produced through the pYX142 yeast expression vector (Novagen). The LexA-ADA2 containing plasmid is described in Ref. 18.
HAT Assays-Liquid HAT assays were performed using either bacterially expressed Gcn5 proteins or immunoprecipitated complexes from yeast extracts (see below). Typical 20-l reactions contained 10 g of core histones (free or nucleosomal), 0.25 Ci of [ 3 H]acetyl-CoA (5Ci/ mmol), and HAT buffer (50 mM Tris-HCl, pH 8, 10% glycerol, 1 mM dithiothreitol, 0.5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride). Reactions were incubated at 30°C for 10 min and stopped by the addition of SDS-sample buffer, and the mixture was loaded on a 12% SDS-PAGE followed by staining with Coomassie Blue and fluorography. Under these conditions the incorporation of radioactivity was linear for the first 10 min, and both the amount of histones as well as the concentration of [ 3 H]acetyl-CoA (2.5 M) were saturating. Kinetic experiments were performed using histone H3. Reactions were initiated by the addition of [ 3 H]acetyl-CoA and stopped by spotting on Whatman P81 cellulose membranes. Membranes were first washed with 50 mM carbonate buffer, pH 9.2, then briefly with acetone, and dried, and the radioactivity was counted in scintillation fluid. In-gel HAT activity assays were performed as described in Ref. 37.
Immunoprecipitations-Crude yeast extracts were prepared from 50 ml of exponentially grown cultures by vortexing pelleted cells with glass beads in extraction buffer (10 mM Tris-HCl, pH 8, 150 mM NaCl, 0.5% Triton X-100, 1 mg/ml leupeptin, 1 mg/ml pepstatin). Extracts were cleared by centrifugation (13000 ϫ g for 30 min at 4°C) and then incubated with protein A-Sepharose beads (Amersham Pharmacia Biotech) previously reacted with LexA-antibody. Sepharose beads with bound immunocomplexes were washed three times with extraction buffer and three times with HAT buffer before use in liquid acetylation assays.
Nucleosome Reconstitution-In vitro reconstitution of mono-or oligonucleosomes was carried out using a 200-base pair (generated by polymerase chain reaction) or a 1100-base pair BglI DNA fragment derived from pUC19. The DNA template was mixed with purified core histones (Boehringer Mannheim) at an equimolar histone/DNA ratio in 2 M NaCl. Gradient salt dialysis reconstitution was performed by decreasing the salt concentration from 2 to 0.25 M NaCl (38). Nucleosomes were fractionated by centrifugation (35,000 rpm for 19 h at 4°C on a Beckman SW41 rotor) through a 5-30% linear sucrose gradients in 10 mM Hepes, pH 8.0, 1 mM EDTA, and 0.1% Nonidet P-40. 1-ml fractions were collected, and those containing reconstituted nucleosomes were concentrated using NanoSpin-Plus centrifugal filters (Gelman Sciences) and used either as substrate in liquid acetylation assays or for micrococcal nuclease digestions. The latter were performed in a buffer containing 15 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1.4 mM CaCl 2 , 0.2 mM EGTA, 0.2 mM EDTA, and 5 mM ␤-mercaptoethanol using 0.015-0.03 units/l of micrococcal nuclease (Boehringer Mannheim) for 1 min at 25°C. The reactions were terminated by adding 2 mM EDTA and 0.5% SDS, were phenol/chloroform-extracted, and the DNA was resolved in 1.5% native agarose gel. After DNA transfer onto nylon membranes, the digestion pattern was detected by hybridization using the 1.1-kilobase BglI fragment of pUC 19.

Generation of Mutated GCN5 Proteins with Severely Diminished in Vitro HAT Activity-To investigate the role of the gcn5
HAT activity for the function of the Gcn5⅐Ada complex, we designed and generated mutations that affected this activity. We have targeted the mutations on two amino acids, His-145 and Cys-252, which were previously suggested to be important for HAT activity (13). Both of these amino acids are within the HAT domain of Gcn5 (Fig. 1A). Three mutants were generated: His-145 3 Ala (HA), Cys-252 3 Ala (CA), and the double mutant (HACA). The mutated proteins were produced in E. coli as fusions with the 6-His tag and purified using nickel nitrilotriacetic acid-agarose chromatography, and their HAT activity was assayed in vitro using saturating amounts of free histones as substrate. As shown in Fig. 1B, under these assay conditions the wild type Gcn5 protein preferentially acetylated histone H3 and to a lesser extent histone H2B (Fig. 1B, middle panel). For equal amounts of proteins, as judged by immunoblot analysis (Fig. 1B, upper panel), the HA and CA derivatives exhibited considerably diminished activity, whereas the HACA derivative had completely lost the ability to acetylate free histones. We have noticed that the in vitro activity of recombinant Gcn5 varied depending on the relative impurities of the nickel nitrilotriacetic acid-agarose chromatography procedure. To control for this we used the in-gel HAT activity assay through which the activity of electrophoretically resolved HATs can be monitored. As shown in Fig. 1C, this method confirmed the attenuated character of the Gcn5 mutant derivatives.
The GCN5 Mutant Derivatives Complement the gcn5 Phenotypes-To investigate the in vivo properties of the mutated Gcn5 proteins, the three mutant GCN5 genes were expressed in a gcn5 yeast strain either as LexA fusions (driven by the ADH1 promoter) or as native genes (driven by the TPI promoter). The LexA-Gcn5 fusion protein is functional in yeast (18). All three mutant genes, despite the apparent compromised HAT activity of the corresponding proteins in vitro, complemented the known gcn5 phenotypes to their full extent. Thus, these GCN5 derivatives complemented the slow growth rates observed in gcn5 strains grown under all nutritional conditions ( Fig. 2A), and the inability of gcn5 strains to grow under histidine starvation conditions (Fig. 2B). These observations were further quantified by testing the ability of the Gcn4 transcriptional activator to fully induce the expression of appropriate genes. For both natural (through Northern blot analysis of the HIS3 mRNA) and artificial genes (through ␤-galactosidase assays), we observed that the Gcn5 mutant derivatives fully supported the function of Gcn4 (data not shown).
The Ability of the Mutated Gcn5 Proteins to Acetylate Histones H3 and H2B Is Enhanced When Produced in Yeast in the Presence of Ada2-The above results suggested that the HAT activity of Gcn5 might not be essential for its in vivo function. To investigate this possibility, we monitored the HAT activity of the Gcn5 derivatives produced in yeast cells. This was accomplished by immunoprecipitating LexA fusion derivatives from yeast extracts using a LexA antiserum and protein-A-Sepharose as immunoadsorbent. Subsequent assays were performed in the Sepharose beads using a mixture of all four core histones as substrate as described under "Experimental Procedures." Based on the fact that LexA fusion derivatives of either Ada2 or Gcn5 are functional in yeast, we anticipated that such immunocomplexes should contain Gcn5⅐Ada complexes. To test this, we produced immunoprecipitates from ada2 or ada2gcn5 strains transformed with the LexA-ADA2 gene. Although the method produced a background acetylation on histones H2B and H4, even in the presence of just the LexA protein, a Gcn5dependent acetylation of histones H3 and H2B was clearly evident (Fig. 3A), demonstrating that the endogenous Gcn5 protein was co-immunoprecipitated with the LexA-Ada2 protein. By contrast, when acetylation of histones was effected by LexA-Gcn5 produced in the absence of Ada2, mainly histone H3 was modified.
Next, we tested the activity of immunoprecipitated com-plexes from a gcn5 or a gcn5ada2 strain expressing the various LexA-GCN5 derivatives. As shown in Fig. 3B (upper panel), the immunoprecipitates containing the wild type fusion protein produced in either the gcn5 or the gcn5ada2 strain efficiently acetylated histone H3. By contrast immunoprecipitates containing the HACA derivative were weakly active for H3 acetylation only when produced in the presence of the ADA2 gene. Similarly, the ability of the HA and CA derivatives to acetylate histone H3 was notably enhanced when those were produced in the presence of ADA2, with the CA mutant exhibiting an appreciable activity even when produced in its absence. To control for the amount of fusion protein present in each immunoprecipitate used in these assays, we performed immunoblot analysis on the upper half of the same gel used to resolve the acetylated histones. As shown in Fig. 3B (lower panel), the amount of the fusion protein present in each pair of strains demonstrated that the above acetylation activities were not due to major quantitative differences but reflected the specific activities of the Gcn5 derivatives. Due to the persisting background acetylation, differences observed in these experiments concerning the acetylation of H2B could not easily be assessed. To resolve this, immunoprecipitates were also tested for HAT activity histone solely using H2B as a substrate. The resulting acetylation pattern confirmed that the HACA fusion protein was inactive when produced in the absence of Ada2 but gained part of its HAT activity toward H2B when produced in the wild type context (Fig. 3C). Analogous results were obtained using the other mutated derivatives (not shown). We concluded that the HAT activity of the mutated Gcn5 proteins was enhanced when produced in wild type strains in an Ada2-dependent manner.

The Presence of Ada2 Augments the Catalytic Rate of Gcn5
HAT-To assess what enzymatic property of Gcn5 was improved when produced in the presence of ADA2, we performed kinetic enzymatic analyses using immunoprecipitates from gcn5 and gcn5ada2 strains producing the LexA-HA derivative. This derivative was chosen because it was measurably active when produced in the absence of Ada2. The kinetic experiments were performed using saturating amounts of [ 3 H]acetyl-CoA and equivalent amounts of enzyme from the two strains as indicated by immunoblot analysis (not shown). Using histone H3, we first determined that for both enzymes this substrate was saturating at the same concentration (20 M). At this substrate concentration, the rate of catalysis was estimated through a time course analysis. As shown in Fig. 4A, this rate was lower (almost half) when the HA derivative was produced in the absence of the ADA2 gene. In addition, for both enzymes we measured the rate of radioactivity incorporation effected by different substrate concentrations. A double reciprocal plot of these measurements (Fig. 4B) showed that although the apparent V max for both enzymes differed significantly, their apparent KϽitinf;mϾ values were equivalent (ϳ10 M). These results strongly suggested that when this Gcn5 derivative was produced in the presence of Ada2, the rate of HAT catalysis was enhanced, but the substrate recognition was unaffected.
The HACA Gcn5 Derivative Produced in Wild Type Strains Efficiently Acetylates Histones on Nucleosomal Arrays-The results obtained so far suggested that the observed partial HAT activity of the HACA-Gcn5 protein could be sufficient for the complete function of the Gcn5⅐Ada2 complex. Since the presumed natural substrates of Gcn5 are nucleosomal histones, we tested the HAT activity of the complexes using these substrates. Nucleosomes were assembled in vitro through gradient salt dialysis of free histones onto either a 200-base pair or a Extracts from these strains were immunoprecipitated using the LexA antibody, and HAT assays were preformed using all four free histones as substrate. Lower panel, immunoblot analysis of the top half of the same gel that indicates the amount of LexA fusion protein used in each assay. C, upper panel, fluorograph of an SDS-PAGE showing the acetylated patterns of histone H2B after HAT assays of LexA immunoprecipitates. gcn5 (WT) or gcn5ada2 (ada2) strains were transformed with LexA gene derivatives fused with either the wild type (GCN5) or the mutant HACA GCN5 derivative. Extracts from these strains were immunoprecipitated using the LexA antibody, and HAT assays were preformed using histone H2B as substrate. Lower panel, immunoblot analysis of the top half of the same gel that indicates the amount of LexA fusion protein used in each assay.

FIG. 4. Kinetic analysis of the LexA-HA HAT.
A, the rate of incorporation of [ 3 H]acetyl groups was measured using saturating amounts of histone H3 as substrate. Extracts from either a gcn5 (squares) or a gcn5ada2 (circles) strain transformed with the LexA-HA gene derivative were immunoprecipitated using the LexA antibody and immunoprecipitates were used for HAT assays. At the specified time points, aliquots were withdrawn, and the incorporated radioactivity was measured. Points on the graph represent an average of three independent experiments. B, double reciprocal plot of substrate concentration versus the rate of radioactivity incorporation in HAT assays using immunoprecipitates of LexA-HA fusion protein produced either in gcn5 (squares) or gcn5ada2 (circles) strains. Assays were performed using equal aliquots of protein A-Sepharose-bound immunoprecipitates for 15 min, and the incorporated radioactivity was measured. 1100-base pair DNA fragment. The formation of oligonucleosomes was confirmed by analyzing the partial micrococcal nuclease digestion patterns of the resulting DNA-protein complexes. As shown in Fig. 5A, such digestions produced a ladder of DNA fragments that corresponded to the sizes protected by mono-to tetranucleosomes (the latter was not adequately resolved from the input DNA). Thus, this analysis showed that onto the large DNA fragment, up to five nucleosomes were assembled. These assembled oligonucleosomes were used as substrates in acetylation assays using immunoprecipitates from cells producing either the wild type Gcn5 or the HACA LexA fusion proteins. As shown in Fig. 5B, both of these immunoprecipitates acetylated nucleosomal histone H3 and to a much lesser extent histone H2B, but only when produced in the presence of Ada2. This specificity for nucleosomal substrates was identical to the one previously observed for the biochemically purified native Gcn5⅐Ada complexes (14). More importantly, these results showed that the observed differences in the acetylation efficiencies between the two derivatives on free histones were minimized with the use of these nucleosomal substrates. Reconstituted mononucleosomes were not suitable substrates for both these immunoprecipitates (not shown). We concluded that the severely attenuated in vitro HAT activity of the double-mutated Gcn5 derivative was further enhanced when assayed on nucleosomal substrates. DISCUSSION We have initiated an investigation on the in vivo role and the regulation of the Gcn5 HAT activity. Toward this goal we have generated loss-of-function Gcn5 derivatives by site-directed mutagenesis. Three mutants were obtained, in two of which the HAT activity of the recombinant protein was attenuated, whereas in the third one, this activity was completely compromised. Despite these in vitro defects, all three proteins provided wild type Gcn5 function in yeast cells. This was confirmed by examining growth rates and transcriptional competence, which are both affected in gcn5 strains. All three mutated Gcn5 derivatives were able to support normal growth rates and proper Gcn4-dependent transcriptional activation through natural as well as artificial promoters. In addition, the function of both Gal4 and VP16 transcriptional activation domains (which is compromised in gcn5 strains (18)) was normal in strains harboring the GCN5 mutant genes (data not shown). These results were in contradiction with the notion that the Gcn5 HAT activity is important for the function of the Gcn5⅐Ada complex. This issue was resolved by measuring the HAT activities of the mutated molecules produced in yeast cells in the presence or absence of Ada2 and by using nucleosomal substrates. We propose that all mutants were functional in vivo due to the potentiation of their enzymatic activity when assembled into native complexes. This in turn reveals a regulatory role for components of the Gcn5⅐Ada complex on its HAT activity.
In this study Gcn5⅐Ada complexes were isolated through immunoprecipitations from yeast extracts containing LexAtagged functional Ada2 or Gcn5 proteins. This immunoprecipitation approach produced native Gcn5⅐Ada complexes with HAT activities comparable with complexes characterized by others biochemically (14,28). This was supported by the following observations. (i) When immunoprecipitates were challenged with nucleosomal substrates, the resulting profile of acetylated histones was identical to that of two Gcn5-containing native complexes that were biochemically isolated by Grant et al. (14). These complexes also predominantly acetylated histone H3 and to a lesser extent, H2B, in a Gcn5-dependent manner. By contrast it is unlikely that the immunoprecipitates contained the 200-kDa Gcn5⅐Ada complex reported by Pollard and Peterson (28), since this complex predominantly acetylated nucleosomal or free histone H4. (ii) When immunoprecipitates were produced through the LexA-Ada2 protein, they exhibited a Gcn5-dependent HAT activity with specificity for free histones H3 and H2B, which was quantitatively and qualitatively different from the activity of immunoprecipitates produced through LexA-Gcn5 in the absence of Ada2 (see Fig. 3A). This was evident despite the persisting background HAT activity that was notably absent when nucleosomal substrates were employed. This latter observation suggested that this contaminating acetylation activity might involve either cytoplasmic HATs or acetyltransferases with other substrate specificities.
Using this immunoprecipitation assay we observed differences in the HAT activity of the various Gcn5 derivatives, which was dependent on whether they were produced in the presence or absence of Ada2. When produced in the absence of this protein, the wild type Gcn5 preferentially acetylated histone H3. This specificity was also observed for both CA and HA derivatives, with the latter exhibiting a severe reduction in its HAT activity. By contrast, the HACA derivative showed no activity when produced in gcn5ada2 strains. These profiles of acetylated free histones were drastically altered when the GCN5 derivatives were expressed in the presence of the ADA2 gene. The ability of HA and to a lesser extent, of CA proteins, to acetylate histone H3 was clearly augmented, whereas acetyltransferase activity was even observed for the HACA protein directed toward histones H3 and H2B. A slight enhancement of the HAT activity was also observed when the wild type Gcn5 protein was produced in the presence of Ada2. These results demonstrated a regulatory role for the Ada2 on the Gcn5 HAT activity, which was clearly evident when the attenuated Gcn5 (WT) or gcn5ada2 (ada2) strains were transformed with LexA gene derivatives fused with either the wild type (GCN5) or the mutant HACA GCN5 derivative. Extracts from these strains were immunoprecipitated using the LexA antibody, and HAT assays were preformed using nucleosomal arrays as substrate. Upper panel, immunoblot analysis of the top half of the same gel, which indicates the amount of LexA fusion protein used in each assay. derivatives were used.
We argue that the observed increase in HAT activity when Gcn5 derivatives were produced in the presence of Ada2 was the result of the assembly of Gcn5 molecules into native complexes. It is known that the Ada2 protein directly interacts with Gcn5 (20) and that mutations on the Ada2 interaction domain of Gcn5 result in loss of Gcn5 function (39). As no other direct interactions have been reported for Gcn5, it is reasonable to assume that this protein assembles into the complex via the Ada2 interaction. Therefore, we postulate that immunoprecipitates obtained from strains lacking Ada2 represented uncomplexed Gcn5 molecules. This was supported by the fact that these immunoprecipitates acetylated free histones with specific activities analogous to those observed in assays utilizing the corresponding bacterially produced protein derivatives. We conclude that the assembly of Gcn5 is what potentiates its HAT activity. Further genetic and biochemical analyses will be required to determine which component(s) of the complex directly effect this potentiation.
There are two ways through which the Gcn5 HAT activity could be enhanced by components of the complex: either through an increase in its enzymatic activity or through an improvement in substrate recognition. To address this question we analyzed the enzymatic properties of uncomplexed versus complexed forms of Gcn5. The kinetic analysis performed using the HA derivative showed that assembled Gcn5 exhibited an increased rate of catalysis but had an almost identical K m with the uncomplexed form. The latter indicated that the two forms of enzyme recognize free histone substrates with equal efficiencies. Similar results were obtained using the CA derivative, and increased HAT activity was also reproducibly observed for the wild type assembled Gcn5 (not shown). Finally, although assembly into the complex clearly increased the activity of the Gcn5 derivatives, restoration to wild type levels was never observed, and the mutant character of each derivative as recombinant molecule was retained. We conclude that the observed complex-dependent potentiation of the Gcn5 HAT activity does not simply reflect a neutralization of the attenuated character of the mutated forms but rather reflects a conformational change that the assembly effects on Gcn5, which results in increased catalytic rates.
All our conclusions thus far have been based on the use of free histones as substrate. This was reasonable since unassembled Gcn5 can recognize only free histones. The use of nucleosomal arrays as substrates revealed that the HAT activity of the HACA derivative was further enhanced when compared with the corresponding activity of the wild type Gcn5. Assuming that such arrays are the physiological substrates for the Gcn5⅐Ada complex, this observation was consistent with the fact that this mutant was fully functional in vivo, and it further demonstrated the positive action of components of the complex on its HAT activity. By contrast and in agreement with previous observations concerning the Tetrahymena HAT B (37), mononucleosomes were not suitable substrates for acetylation by the Gcn5⅐Ada complex. We postulate that either the observed Gcn5 potentiation is best elaborated on polynucleosomal substrates due to better recognition or that docking of the complex on such arrays and interaction with nucleosomes further enhances HAT activity.
Modulation of enzymatic activities within native complexes have been reported in a number of instances from the well studied regulation of CDK kinases by cyclins (40) to the recently described example of enhancement of the MSH2-MSH3mediated DNA mismatch recognition by the MLH1-PMS1 complex in yeast (41). Cases relevant to transcriptional regulation include the increased Swi2/Snf2 ATPase activity when meas-ured within the Swi⅐Snf complex (6,42) and the regulation of kinase and ATPase activities of TFIIH by TFIIE (1). Finally, the activity of the major cytoplasmic HAT in yeast, Hat1, is also enhanced when associated with the Hat2 subunit, which is involved in substrate recognition (43). The observed positive effect of components of the Gcn5⅐Ada complex on its HAT activity could simply represent a structural feature of the Gcn5 holoenzyme. Alternatively it could indicate means of a biologically relevant modulation of the HAT activity in response to physiological signals. The attenuated forms of Gcn5 could serve as valuable tools to address these questions.