|Home | About | Journals | Submit | Contact Us | Français|
The yeast GCN5 gene encodes the catalytic subunit of a nuclear histone acetyltransferase and is part of a high-molecular-weight complex involved in transcriptional regulation. In this paper we show that full activation of the HO promoter in vivo requires the Gcn5 protein and that defects in this protein can be suppressed by deletion of the RPD3 gene, which encodes a histone deacetylase. These results suggest an interplay between acetylation and deacetylation of histones in the regulation of the HO gene. We also show that mutations in either the H4 or the H3 histone gene, as well as mutations in the SIN1 gene, which encodes an HMG1-like protein, strongly suppress the defects produced by the gcn5 mutant. These results suggest a hierarchy of action in the process of chromatin remodeling.
Nuclear processes, including transcription, require that enzymes gain access to the eukaryotic DNA template despite the fact that it is complexed with histone and nonhistone proteins to form chromatin. Genetic studies with Saccharomyces cerevisiae have identified two groups of genes that appear to link transcriptional regulation to chromatin structure (40). The first group encodes components of the SWI/SNF complex, which has been proposed to antagonize the repressive effects of chromatin on transcription (24). SWI/SNF genes were identified in genetic screens for mutants defective in the expression of various genes, including the HO and SUC2 genes (2, 18, 22, 32). The second group of genes includes various SPT and SIN genes, which were defined as suppressors of various types of transcriptional defects (40). The sin2-1 mutation was found to lie in the HHT1 gene, which encodes histone H3. Five additional different point mutations, two in histone H3 and three in histone H4, also displayed a Sin−/Spt− phenotype (12, 25). These mutations affect residues that are believed either to contact DNA or to be involved in histone-histone contacts within the histone octamer (39). The SIN1 gene was found to encode a protein with similarities to mammalian HMG1, a structural component of chromatin (11). Although the precise role of yeast SIN1 is not known, the similarity of sin1 and sin2-1 mutant phenotypes has led to the inference that these two genes have related physiological functions.
Recently, a group of genes involved in acetylation and deacetylation of histones has been recognized. Histone acetylation has long been correlated with the modulation of gene activity (37). Acetylation of lysines in histone amino-terminal tail domains reduces the positive charge, thereby weakening histone-DNA interactions, destabilizing higher-order structure, and rendering nucleosomal DNA more accessible to transcriptional factors (4, 14). Yeast Gcn5 was originally identified as a regulatory factor required for function of the yeast activator Gcn4 (5), and recently it has been shown that Gcn5 is a histone acetyltransferase (3, 13, 28) that is part of at least two high-molecular-weight complexes called ADA (8) and SAGA (7). The recruitment of these complexes to DNA is thought to direct the local destabilization of nucleosomes, producing more efficient transcriptional activation on a promoter. Aside from transcriptional regulators that function as histone acetyltransferases, there are also regulators that deacetylate the histones (29, 36). These deacetylases comprise part of a transcriptional repression pathway conserved from yeast to vertebrates and provide a molecular mechanism whereby transcription can be continually controlled (19, 38).
In this paper we show that the expression of the HO gene is affected by defects in histone acetylation and deacetylation. Previous work has shown that the SWI/SNF complex and structural components of chromatin also affect HO expression (11, 12, 16, 22). We present an analysis of single and double mutations in the genes encoding several of these components, and the results suggest a hierarchy in the chromatin remodeling process.
All strains of S. cerevisiae used in this study are described in Table Table1.1. Complete medium (yeast extract-peptone-dextrose [YEPD]) and minimal medium supplemented with the required amino acids were used for yeast growth and transformations (26). Histidine limitation was accomplished by supplementing minimal media with 10 mM 3-amino-1,2,4-triazole (3-AT) (5).
Single mutants were obtained either by gene disruptions performed by using the one-step replacement method (27) or by gene conversions carried out by a two-step gene replacement procedure (31). Double and triple mutants were obtained by crossing single mutants of opposite mating types and selecting segregants carrying the desired mutations.
A strain carrying a swi5::LEU2 null allele was generated as described in reference 34. The gcn5::hisG strain was generated as described in reference 15. The HO-lacZ fusion allele is described in reference 30. The histone mutations were introduced in the chromosome by a two-step replacement procedure (31) using integrating plasmids marked with the URA3 gene (obtained from R. K. Tabtiang and I. Herskowitz); as these mutations are partially dominant, it is possible to observe their effects, even in the presence of another histone gene copy (12). The rpd3Δ::LEU2 strain was generated by transforming yeast with pDM176 digested with BamHI. This plasmid carries the RPD3 locus with a replacement of the entire RPD3 open reading frame (ORF) with the LEU2 gene (15a). Correct integration was tested by PCR analysis using oligonucleotides flanking the RPD3 locus. The sin1Δ::TRP1 deletion strains were generated by transforming yeast with pUC-SIN1::TRP linearized with EcoRI-SphI. This plasmid carries a replacement of the SIN1 ORF with the TRP1 gene (11). Correct integration was tested by PCR analysis using oligonucleotides flanking the SIN1 locus.
Strains were grown to mid-log phase in YEPD medium. Total-yeast RNA was isolated and fractionated on formaldehyde gels, transferred to nylon membranes (Genescreen; DuPont), and hybridized with random-primed 32P-labeled fragments. The DNA probes used were obtained as PCR fragments by amplification of the desired ORF with specific primers (MapPairs; Research Genetics Inc.), with the exception of the HO probe, which was obtained as a 2.6-kb HindIII fragment from the plasmid pGAL-HO (9).
HO gene expression is dependent on SWI5. This gene encodes a zinc finger DNA-binding protein which binds specifically, along with the PHO2 protein, to the upstream region of the HO promoter (1, 34). Genetic studies have described a series of extragenic suppressor mutations that permit expression of HO in the absence of the SWI5 gene product (17, 33). Two of the genes identified in this screen, RPD3 and SIN3, encode, respectively, a histone deacetylase and a protein tightly associated with it (10, 29, 35). The fact that mutations in the gene pair SIN3/RPD3 are able to suppress the absence of the Swi5 protein suggests that one of the roles of the Swi5-Pho2 heterodimer is the recruitment, either directly or indirectly, of a histone acetyltransferase activity. A likely candidate is the GCN5 gene, which encodes a protein with histone acetyltransferase activity (13). To test this idea, we examined the levels of HO mRNA produced in wild-type and isogenic gcn5 mutant strains (obtained by disruption of the GCN5 gene; see Materials and Methods). We found that a gcn5 mutant strain produced significantly less HO mRNA (Fig. (Fig.1A).1A). By contrast, the absence of the Gcn5 protein did not impair the normal levels of PHO2 and SWI5 mRNA.
In principle, SWI5 and GCN5 gene products could act in the same pathway or through different pathways to activate HO expression. If two genes act in the same pathway, then the phenotype of the double mutant should be the same as that of one of the single mutants. On the other hand, if two genes act through different pathways, then the phenotype of the double mutant should be more severe than that of either single mutant. To distinguish between these two possibilities, we measured the β-galactosidase activity produced by a chromosomal HO-lacZ gene fusion in a swi5 gcn5 double mutant and compared it to those in the single mutants (Fig. (Fig.1B).1B). HO-lacZ expression in the gcn5 and swi5 mutants was reduced 50- and 200-fold, respectively. In the double mutant, HO-lacZ expression was reduced 200-fold. The β-galactosidase values of the swi5 mutant are so low (0.5 Miller units) that we cannot make a conclusive argument about the relationship of SWI5 and GCN5. However, since both defects are suppressed by the same mutations (i.e., by rpd3, sin1, and sin2 mutations; see below) and since the levels of mRNA for SWI5 and PHO2 genes are not affected by gcn5 mutations (Fig. (Fig.1A),1A), these facts support the idea that SWI5 and GCN5 act in the same pathway to stimulate HO expression.
The results described above are compatible with the idea that histone acetylation is required for maximal HO transcriptional activation. According to this hypothesis, a mutation in a gene encoding a deacetylase should be able to suppress a gcn5 mutation. A likely candidate is the RPD3 gene, since mutations in this gene suppress the Swi5 requirement in the HO gene (35). We therefore measured HO-lacZ activity in single and double mutants carrying null alleles of the GCN5 and RPD3 genes. The results (Fig. (Fig.2)2) show that a deletion of the deacetylase gene RPD3 alleviates the requirement for the histone acetyltransferase gene GCN5 in HO gene expression. The level of suppression of a gcn5 mutation by the deletion of RPD3 is similar to that observed in the case of swi5 mutations (35) and is also similar to the suppression observed in a triple swi5 gcn5 rpd3 mutant, again supporting the view that SWI5 and GCN5 function in the same pathway.
One of the defects originally observed in gcn5 mutant strains was their inability to grow in media imposing amino acid limitation (20). Thus, a strain carrying a deletion of the GCN5 gene is defective in growth in media containing 3-AT, a condition that mimics histidine starvation (5). To address whether a deletion in the RPD3 gene suppresses other defects in gcn5 strains, we also tested the ability of the RPD3 deletion to allow growth of a gcn5 strain in the presence of 3-AT. As shown in Fig. Fig.2,2, the gcn5 strain exhibited a growth defect under such conditions compared with an isogenic wild-type strain. Deletion of the RPD3 gene indeed alleviates this defect, allowing growth of the gcn5 strain under these conditions.
In addition to sin3 and rpd3 mutations, defects in other genes are well-known suppressors of transcriptional deficiencies in HO. One of these genes is SIN1. This gene encodes a protein with similarities to the mammalian HMG1 protein, and it is believed to be a component of chromatin (11). We have monitored both HO-lacZ expression and the ability to grow in the presence of 3-AT of a double mutant defective in both GCN5 and SIN1. The results shown in Fig. Fig.33 indicate that the absence of Sin1 protein relieves the requirement of Gcn5 both for HO expression and for growth on 3-AT.
An explanation for the results obtained with the sin1 mutant is that the suppression we observed is caused by a defect in chromatin structure, such that this defective chromatin bypasses the requirement for histone acetylation. If this is the case, then other mutations which produce defective chromatin might also be expected to suppress the gcn5 defects. Certain amino acid changes (sin mutations) in either histone H3 or histone H4 alleviate the same set of transcriptional defects as does the sin1 mutation (12, 23). These sin mutations lie in the histone fold domain of histones H3 and H4, and they are in close proximity to one another on the surface of the histone octamer. It has been proposed that residues altered by these mutations may define a functional domain (the SIN domain) that behaves formally as a negative regulator of transcription (12).
To address if defective histones also suppress gcn5 mutations, the following histone mutant alleles were tested for their ability to suppress a deletion of the GCN5 gene: sin2-1 (R116H in HHT1), hhf2-7 (R45C in HHF4), hhf2-8 (V43I in HHF4), and hhf2-13 (R45H in HHF4). In spite of the fact that the targets for GCN5 protein are the histone tails, mutations in the histone fold are able to efficiently suppress the defects caused by the absence of the GCN5 gene product (Fig. (Fig.4A).4A).
We also determined the effects of combining a deletion of the RPD3 gene with the histone sin mutations. Levels of HO-lacZ activity were determined in single and double mutants, and we found in the double mutants a strong synergistic effect; that is, the activity displayed by the double mutant is higher than the sum of the activities displayed by the single mutants (Fig. (Fig.4B).4B). The same synergistic effect is also seen in combinations of rpd3 and sin1 mutations (data not shown).
Regulation of the yeast HO gene is complex, and many genes that regulate HO have been identified (16). These include genes encoding the SWI/SNF complex (22, 32); SIN1, which encodes an HMG1-like protein (11); SIN2, which encodes histone H3; HHF4, which encodes histone H4 (12); and SIN3, which, along with RPD3, is involved in the deacetylation of histones (35). In this paper, we show a requirement for the GCN5 gene, which encodes a histone acetyltransferase (3), for optimal transcription of the HO gene.
The identification of histone acetyltransferases and histone deacetylases as transcriptional regulators provides molecular mechanisms whereby transcription might be turned up or down (38), but so far no such interplay between acetylase and deacetylase activities at a single gene has been reported. The suppression of the gcn5 defects by deletion of one of the genes encoding a deacetylase activity provides clear support for such interplay at the HO promoter. The suppression we observed is only partial, suggesting a functional redundancy in the deacetylase activity. Another protein with deacetylase activity is encoded by the gene HDA1, and three additional ORFs with high levels of homology with RPD3 and HDA1 have also been described (29). However, we observed that deletion of HDA1 or of one of these additional ORFs (HOS1) does not suppress the GCN5 requirement in HO expression (data not shown). Another explanation for the fact that suppression is only partial is that the rpd3 deletion may destabilize additional proteins with which it is complexed (7), and these additional proteins may contribute to the activation of HO.
The pattern of genetic interactions described in this work suggests a hierarchy of gene function that pertains to chromatin components, histone acetylation, and the SWI/SNF complex. Loss of Swi5 (the major activator protein for the HO gene ) can be partially suppressed by sin1, sin2 (histone H3), sin3, and rpd3 mutations (33, 35). Loss of GCN5 (a histone acetyltransferase, also required for HO transcription) can be suppressed by these same mutations (Fig. (Fig.2,2, ,3,3, and and4A).4A). However, while defects in the SWI/SNF complex can be suppressed by sin1 (which is thought to be a target of the SWI/SNF complex  and sin2 mutations (11, 12), they cannot be suppressed efficiently by sin3 or rpd3 mutations (33, 35). These results indicate that histone acetylation at the HO promoter functions upstream of the SWI/SNF complex. Consistent with this view is the strong synergy seen between rpd3 mutations (which affect the acetylation of histone tails) and sin1 and sin2 mutations (which circumvent the need for the SWI/SNF complex) (Fig. (Fig.4B).4B). One hypothesis consistent with this genetic hierarchy is that, at the HO promoter, histone acetylation precedes and enables the action of the SWI/SNF complex. A similar view has recently been developed independently by Pollard and Peterson (24a).
We thank D. Moazed, R. K. Tabtiang, and R. Candau for providing indispensable strains and plasmids throughout the course of this work. D. Moazed is also acknowledged for critical reading of the manuscript.
This work was supported by an NIH grant to A.D.J. and by an EMBO long-term postdoctoral fellowship to J.P.-M.