Histone acetyltransferases.

Transcriptional regulation in eukaryotes occurs within a chromatin setting and is strongly influenced by nucleosomal barriers imposed by histone proteins. Among the well-known covalent modifications of histones, the reversible acetylation of internal lysine residues in histone amino-terminal domains has long been positively linked to transcriptional activation. Recent biochemical and genetic studies have identified several large, multisubunit enzyme complexes responsible for bringing about the targeted acetylation of histones and other factors. This review discusses our current understanding of histone acetyltransferases (HATs) or acetyltransferases (ATs): their discovery, substrate specificity, catalytic mechanism, regulation, and functional links to transcription, as well as to other chromatin-modifying activities. Recent studies underscore unexpected connections to both cellular regulatory processes underlying normal development and differentiation, as well as abnormal processes that lead to oncogenesis. Although the functions of HATs and the mechanisms by which they are regulated are only beginning to be understood, these fundamental processes are likely to have far-reaching implications for human biology and disease.

Histone acetyltransferases: function, structure, and catalysis.

Histone acetyltransferases (HATs) directly link chromatin modification to gene activation. Recent structure/function studies provide insights into HAT catalysis and histone binding, and genetic studies suggest cross-talk between acetylation and other histone modifications. Developmental aberrations in mice and certain human cancers are associated with HAT mutations, further highlighting the importance of these enzymes to normal cell growth and differentiation.

Recruitment of the yeast Tup1p-Ssn6p repressor is associated with localized decreases in histone acetylation.

Posttranslational acetylation of histones is an important element of transcriptional regulation. The yeast Tup1p repressor is one of only a few non-enzyme proteins known to interact directly with the amino-terminal tail domains of histones H3 and H4 that are subject to acetylation. We demonstrated previously that Tup1p interacts poorly with more highly acetylated isoforms of these histones in vitro. Here we show that two separate classes of promoters repressed by Tup1p are associated with underacetylated histones in vivo. This decreased histone acetylation is dependent upon Tup1p and its partner Ssn6p and is localized to sequences near the point of Tup1p-Ssn6p recruitment. Increased acetylation of histones H3 and H4 is observed upon activation of these genes, but this increase is not dependent on transcription per se. Direct recruitment of Tup1p-Ssn6p complexes via fusion of Tup1p to the lexA DNA binding domain is sufficient to confer repression and induce decreased acetylation of H3 and H4 at a target promoter. Taken together, our results suggest that stable decreases in histone acetylation levels are directed and/or maintained by the Tup1p-Ssn6p repressor complex.

Identification of protein interactions by far Western analysis.

Far western blotting is a method of identifying protein-protein interactions. One protein of interest is immobilized on a solid support membrane, then probed with a non-antibody protein. Far western blots can be used to identify specific interacting proteins in a complex mixture of proteins. They are particularly useful for examining interactions between proteins that are difficult to analyze by other methods due to solubility problems or because they are difficult to express in cells. This method is performed totally in vitro, and the proteins of interest can be prepared in a variety of ways.

Identification of protein interactions by far Western analysis.

Far Western blotting is a method for detecting protein-protein interactions. This technique utilizes a nonantibody protein probe to detect interacting proteins immobilized on a membrane support. Proteins to be assayed can be prepared by multiple techniques and detected by several labeling schemes.

Ssn6-Tup1 interacts with class I histone deacetylases required for repression.

Ssn6-Tup1 regulates multiple genes in yeast, providing a paradigm for corepressor functions. Tup1 interacts directly with histones H3 and H4, and mutation of these histones synergistically compromises Ssn6-Tup1-mediated repression. In vitro, Tup1 interacts preferentially with underacetylated isoforms of H3 and H4, suggesting that histone acetylation may modulate Tup1 functions in vivo. Here we report that histone hyperacetylation caused by combined mutations in genes encoding the histone deacetylases (HDACs) Rpd3, Hos1, and Hos2 abolishes Ssn6-Tup1 repression. Unlike HDAC mutations that do not affect repression, this combination of mutations causes concomitant hyperacetylation of both H3 and H4. Strikingly, two of these class I HDACs interact physically with Ssn6-Tup1. These findings suggest that Ssn6-Tup1 actively recruits deacetylase activities to deacetylate adjacent nucleosomes and promote Tup1-histone interactions.

Loss of Gcn5l2 leads to increased apoptosis and mesodermal defects during mouse development.

Histone acetyltransferases regulate transcription, but little is known about the role of these enzymes in developmental processes. Gcn5 (encoded by Gcn5l2) and Pcaf, mouse histone acetyltransferases, share similar sequences and enzymatic activities. Both interact with p300 and CBP (encoded by Ep300 and Crebbp, respectively), two other histone acetyltransferases that integrate multiple signalling pathways. Pcaf is thought to participate in many of the cellular processes regulated by p300/CBP (refs 2-8), but the functions of Gcn5 are unknown in mammalian cells. Here we show that the gene Pcaf is dispensable in mice. In contrast, Gcn5l2-null embryos die during embryogenesis. These embryos develop normally to 7.5 days post coitum (d.p.c.), but their growth is severely retarded by 8.5 d.p.c. and they fail to form dorsal mesoderm lineages, including chordamesoderm and paraxial mesoderm. Differentiation of extra-embryonic and cardiac mesoderm seems to be unaffected. Loss of the dorsal mesoderm lineages is due to a high incidence of apoptosis in the Gcn5l2 mutants that begins before the onset of morphological abnormality. Embryos null for both Gcn5l2 and Pcaf show even more severe defects, indicating that these histone acetyltransferases have overlapping functions during embryogenesis. Our studies are the first to demonstrate that specific acetyltransferases are required for cell survival and mesoderm formation during mammalian development.

Conservation of histone binding and transcriptional repressor functions in a Schizosaccharomyces pombe Tup1p homolog.

The Ssn6p-Tup1p corepressor complex is important to the regulation of several diverse genes in Saccharomyces cerevisiae and serves as a model for corepressor functions. To investigate the evolutionary conservation of these functions, sequences homologous to the S. cerevisiae TUP1 gene were cloned from Kluyveromyces lactis (TUP1) and Schizosaccharomyces pombe (tup11(+)). Interestingly, while the K. lactis TUP1 gene complemented an S. cerevisiae tup1 null mutation, the S. pombe tup11(+) gene did not, even when expressed under the control of the S. cerevisiae TUP1 promoter. However, an S. pombe Tup11p-LexA fusion protein repressed transcription of a corresponding reporter gene, indicating that this Tup1p homolog has intrinsic repressor activity. Moreover, a chimeric protein containing the amino-terminal Ssn6p-binding domain of S. cerevisiae Tup1p and 544 amino acids from the C-terminal region of S. pombe Tup11p complemented the S. cerevisiae tup1 mutation. The failure of native S. pombe Tup11p to complement loss of Tup1p functions in S. cerevisiae corresponds to an inability to bind to S. cerevisiae Ssn6p in vitro. Disruption of tup11(+) in combination with a disruption of tup12(+), another TUP1 homolog gene in S. pombe, causes a defect in glucose repression of fbp1(+), suggesting that S. pombe Tup1p homologs function as repressors in S. pombe. Furthermore, Tup11p binds specifically to histones H3 and H4 in vitro, indicating that both the repression and histone binding functions of Tup1p-related proteins are conserved across species.

Interactions of transcriptional regulators with histones.

Tremendous advances in the study of chromatin have revealed new classes of transcriptional regulators distinct from classical DNA-binding proteins. Many previously described transcription factors, coactivators, and adaptors are regulators of chromatin structure, interacting directly with the core histone proteins or with nucleosomes. This review describes a method used by our laboratory to examine the interactions of regulatory proteins with the core histone proteins. Far-Western analysis uses a protein probe to detect interactions with histones immobilized on membranes. Variations of this technique can detect the acetylation state of the interacting histones and whether the interaction occurs through the globular domain or the amino-terminal "tail" domain. In addition, we discuss complementary techniques for confirming histone-regulatory protein interactions.

Mammalian GCN5 and P/CAF acetyltransferases have homologous amino-terminal domains important for recognition of nucleosomal substrates.

The yeast transcriptional adapter Gcn5p serves as a histone acetyltransferase, directly linking chromatin modification to transcriptional regulation. Two human homologs of Gcn5p have been reported previously, hsGCN5 and hsP/CAF (p300/CREB binding protein [CBP]-associated factor). While hsGCN5 was predicted to be close to the size of the yeast acetyltransferase, hsP/CAF contained an additional 356 amino-terminal residues of unknown function. Surprisingly, we have found that in mouse, both the GCN5 and the P/CAF genes encode proteins containing this extended amino-terminal domain. Moreover, while a shorter version of GCN5 might be generated upon alternative or incomplete splicing of a longer transcript, mRNAs encoding the longer protein are much more prevalent in both mouse and human cells, and larger proteins are detected by GCN5-specific antisera in both mouse and human cell extracts. Mouse GCN5 (mmGCN5) and mmP/CAF genes are ubiquitously expressed, but maximum expression levels are found in different, complementary sets of tissues. Both mmP/CAF and mmGCN5 interact with CBP/p300. Interestingly, mmGCN5 maps to chromosome 11 and cosegregates with BRCA1, and mmP/CAF maps to a central region of chromosome 17. As expected, recombinant mmGCN5 and mmP/CAF both exhibit histone acetyltransferase activity in vitro with similar substrate specificities. However, in contrast to yeast Gcn5p and the previously reported shorter form of hsGCN5, mmGCN5 readily acetylates nucleosomal substrates as well as free core histones. Thus, the unique amino-terminal domains of mammalian P/CAF and GCN5 may provide additional functions important to recognition of chromatin substrates and the regulation of gene expression.

Essential and redundant functions of histone acetylation revealed by mutation of target lysines and loss of the Gcn5p acetyltransferase.

The Gcn5p histone acetyltransferase exhibits a limited substrate specificity in vitro. However, neither the specificity of this enzyme in vivo nor the importance of particular acetylated residues to transcription or cell growth are well defined. To probe these questions, we mutated specific lysines in the N-termini of histones H3 and H4 and examined the effects of these mutations in yeast strains with and without functional GCN5. We found that in vivo, GCN5 is required either directly or indirectly for the acetylation of several sites in H3 and H4 in addition to those recognized by the recombinant enzyme in vitro. Moreover, in the absence of GCN5, cells accumulate in G2/M indicating that Gcn5p functions are important for normal cell-cycle progression. Mutation of K14 in H3, which serves as the major target of recombinant Gcn5p acetylation in vitro, confers a strong, synthetic growth defect in gcn5 cells. Synergistic growth defects were also observed in gcn5 cells carrying mutations in lysine pairs (K8/K16 or K5/K12) in histone H4. Strikingly, simultaneous mutation of K14 in H3 and K8 and K16 in H4 to arginine, or deletion of either the H3 or the H4 N-terminal tail, results in the death of gcn5 cells. Mutation of these same three sites to glutamine is not lethal. Indeed, this combination of mutations largely bypasses the need for GCN5 for transcriptional activation by Gal4-VP16, supporting an important role for histone acetylation in Gcn5p-mediated regulation of transcription. Our data indicate that acetylation of particular lysines in histones H3 and H4 serves both unique and overlapping functions important for normal cell growth, and that a critical overall level of histone acetylation is essential for cell viability.

Amino termini of histones H3 and H4 are required for a1-alpha2 repression in yeast.

The Saccharomyces cerevisiae alpha2 repressor controls two classes of cell-type-specific genes in yeast through association with different partners. alpha2-Mcm1 complexes repress a cell-specific gene expression in haploid alpha cells and diploid a/alpha cells, while a1-alpha2 complexes repress haploid-specific genes in diploid cells. In both cases, repression is mediated through Ssn6-Tu1 corepressor complexes that are recruited via direct interactions with alpha2. We have previously shown that nucleosomes are positioned adjacent to the alpha2-Mcm1 operator under conditions of repression and that Tupl interacts directly with histones H3 and H4. Here, we examine the role of chromatin in a1-alpha2 repression to determine if chromatin is a general feature of repression by Ssn6-Tup1. We find that mutations in the amino terminus of histone H4 cause a 4- to 11-fold derepression of a reporter gene under a1-alpha2 control, while truncation of the H3 amino terminus has a more modest (3-fold or less) effect. Strikingly, combination of the H3 truncation with an H4 mutation causes a 40-fold decrease in repression, clearly indicating a central role for these histones in a1-alpha2-mediated repression. However, in contrast to the ordered positioning of nucleosomes adjacent to the alpha2-Mcm1 operator, nucleosomes are not positioned adjacent to the a1-alpha2 operator in diploid cells. Our data indicate that chromatin is important to Ssn6-Tup1-mediated repression but that the degrees of chromatin organization directed by these proteins differ at different promoters.

Global alterations in chromatin accessibility associated with loss of SIN4 function.

Sin4p is a component of a mediator complex associated with the C-terminal domain of RNA polymerase II and SIN4 is required for proper regulation of several genes in yeast, including the HO endonuclease gene, glucose repressible genes and MATa cell-specific genes. Previous studies indicated that SIN4 may influence transcription through changes in the organization of chromatin. We have examined a specific chromatin structure associated with MATa cell-specific repression in sin4 MATalpha cells to determine if SIN4 is required for nucleosome positioning. Although the loss of SIN4 has no effect on nucleosome location, we find that the sensitivity of bulk chromatin from sin4 cells to micrococcal nuclease digestion is strikingly increased relative to chromatin from isogenic wild-type cells. The nuclease hypersensitivity of chromatin from sin4 cells is not related to gross alterations in histone gene expression or to bulk increases in histone modification. Our experiments suggest that SIN4 directly or indirectly regulates a global aspect of chromatin accessibility, providing a molecular basis for phenotypic similarities between sin4 mutations and mutations in histones.

The subunit-exchange model of histone acetylation.

Increased histone acetylation has long been linked to gene activation, but little is known about how acetylation levels are regulated, largely because the histone acetyltransferase activities (HATs) responsible for this modification have been cloned only recently. Comparison of the biochemical nature of the Tetrahymena HAT A complex with the genetic and biochemical properties of the Saccharomyces Gcn5p-Ado complex leads us to propose that histone acetylase assemblies may be modular in nature and that this modularity may be an intimate part of the association of these enzymes with chromatin. The 'subunit-exchange' model provides a mechanism for the regulation and targeting of both histone acetylases and deacetylases and has implications for the control of cell growth, proliferation and tumorigenesis.

Transcription-linked acetylation by Gcn5p of histones H3 and H4 at specific lysines.

The yeast transcriptional adaptor, Gcn5p, is a catalytic subunit of a nuclear (type A) histone acetyltransferase linking histone acetylation to gene activation. Here we report that Gcn5p acetylates histones H3 and H4 non-randomly at specific lysines in the amino-terminal domains. Lysine 14 of H3 and lysines 8 and 16 of H4 are highly preferred acetylation sites for Gcn5p. We also demonstrate that lysine 9 is the preferred position of acetylation in newly synthesized yeast H3 in vivo. This finding, along with the fact that lysines 5 and 12 in H4 are predominant acetylation sites during chromatin assembly of many organisms, indicates that Gcn5p acetylates a distinct set of lysines that do not overlap with those sites characteristically used by type B histone acetyltransferases for histone deposition and chromatin assembly.