Histone Acetylation

Acetylation of chromatin by histone acetyltransferases (HATs), using acetyl coenzyme A as the acetyl donor, occurs at the e-amino groups of particular lysine side chains in the N-terminal basic tails of the core histones, with varying consequences. Histone deacetylases (HDACs) remove the acetylation. Histone acetylation has roles in chromatin assembly linked to DNA replication, in gene transcription, where the histones are hyperacetylated, and, apparently paradoxically, in heterochromatin (1). In the centric heterochromatin of the polytene chromosomes of the fruit fly Drosophila (2) and at the silent mating type loci in the yeast Saccharomyces cerevisiae (3), the histones are hypoacetylated, as might be expected, but an acetyl group on Lys12 in the exposed N-terminal tail of H4  is essential for heterochromatin formation and may act as a recognition site for the assembly of other heterochromatin proteins. The importance of the pattern of site-specific acetylation, rather than the overall number of acetyl groups, is becoming increasingly clear, but is still poorly understood. Elevated levels of Lys16 acetylation on H4 in the male X-chromosome in Drosophila appears to be important, in some way, in dosage compensation, which results in a doubling of transcription from X-linked genes in the male (4).

 Histone H4 deposited with H3 at the replication fork by the core histone molecular chaperone CAF-1 (chromatin assembly factor 1) during chromatin assembly is acetylated at Lys5 and Lys12 primarily, by a so-called B-type (deposition-linked) acetyltransferase. Deacetylation occurs soon after deposition. Transcription-related hyperacetylation occurs in chromatin at lysines 5, 8, 12, and 26 in H4, and lysines 9 and 14 in H3. Its role appears to be to “open” chromatin, primarily by destabilizing higher-order structure, thus permitting access by the transcription machinery. Suggestions of the interactions that might be disrupted have come recently from the high-resolution structure of the nucleosome core particle. Because the N-terminal tails are not bound within the nucleosome core, acetylation is unlikely to act at this level. The facilitation of transcription factor binding due to acetylation observed in in vitro assays with mononucleosomes (eg, Ref. 5) may be due to disruption of nonspecific interactions of the histone tails with the DNA gyres, in the absence of any possibilities for higher-order structure formation. Direct evidence that acetylation per se does result in increased transcription has come from studies of the hsp26 gene reconstituted using a Drosophila embryo extract into chromatin with either unacetylated or acetylated histones (6). In nuclei, large domains of chromatin-encompassing active genes appear to be acetylated; this state precedes transcription, and the chromatin is said to be in a transcriptionally competent state. In at least one well-documented case (7), the chicken b-globin gene cluster, the genes are embedded in a domain of greater than 30 kbp, where the histones are hyperacetylated and which shows increased sensitivity to the nuclease DNase I, suggesting a less compact structure.

 1. Histone Acetylases

The finding that the Tetrahymena HAT, HAT-A, is homologous to a yeast transcriptional co-activator, GCN5 (8), solved the mystery of how acetylation might be targeted to particular regions of chromatin. It has since been shown that the acetyltransferase activity of GCN5 (which functions as the GCN5-Ada2-Ada3 complex) is essential for its activator activity (9, 10) and that acetylation occurs in the promoter region. GCN5 also functions with Ada proteins in the yeast SAGA complex, which also contains Spt proteins (transcriptional regulators) (SAGA = Spt3–Ada–GCN5–Acetyltransferase) and is involved in the regulation of many yeast genes. It is now clear (1) that a growing number of proteins known to be transcriptional co-activators and co-activator-associated proteins have HAT activity [eg, p300/CBP, P/CAF (a human homologue of GCN5), the TFIID component TAFII250, which is likely to be associated with all promoters during transcriptional initiation, and many others]. (CBP = cAMP response element binding protein; CAF = CBP–associated factor.) Most, including GCN5, function within large complexes, where one subunit recognizes a DNA-bound transcription factor and another has acetyltransferase activity. p300/CBP interacts with transcription factors in many different signaling pathways (eg, nuclear hormone receptors, c-Fos, c-Jun/v-Jun), and acetylation is therefore the end point of many signal transduction pathways. Acetylation of different histones, and different sites within a particular histone, may turn out to require different HATs; GCN5 in vivo acetylates preferentially Lys8 and Lys16 of H4 and Lys14 of H3 (11). Histones may not be the only targets of all of the enzymes designated as histone acetyltransferases, for example, HMG 1 and HMG I(Y) are known to be acetylated in vivo. The transcription factor/tumor suppressor p53 is acetylated both in vitro and in vivo by p300, its coactivator. Two general transcription initiation factors, TFIIEb and TFIIF, can also be acetylated by P/CAF and p300 (12). Because the HAT activities that have recently been studied are tethered to sequence-specific DNA binding proteins at promoters (and enhancers), their effect will necessarily be local. It is not clear how this relates to whole domain acetylation over tens of kilobases. One possibility is that the loosening/disruption of the chromatin structure close to the promoter might be transmitted down the 30-nm filament  through disruption of cooperative interactions, and that the temporarily loosened tails become targets for a pool of untargeted acetylases. There are other possibilities, and this area needs further study. Various models for how histone acetylases and deacetylases selectively affect gene expression have recently been discussed (13).

The first structure of a HAT catalytic subunit (Hat1 from the yeast Saccharomyces cerevisiae, which was the first HAT to be isolated) has recently been determined, complexed with acetyl CoA (14); in vivo Hat1 exists as a heterodimer with the regulatory subunit HAT2. Hat1's sequence preference for acetylation is Lys12>Lys5, and it acetylates free rather than nucleosomal histones, so it seems likely to be a deposition-related HAT. Nonetheless, Hat1 in both yeast and human cells appears to be primarily nuclear, rather than cytoplasmic as is generally assumed (see Refs. 14 and 15). Hat1 has sequence motifs in common with the GCN5-related superfamily of N-acetyltransferases (the GNAT superfamily) and should therefore be an excellent model for the other HATs. Indeed the Hat1 structure explains mutagenesis data on other HATs, such as Gcn5 and CBP; some of the mutations map to the acetyl CoA binding site. The structure of a HAT with a histone or histone tail bound is now needed to reveal the basis of the site specificity of acetylation, which is in the meantime suggested by modeling (14); the structure of a Hat1/Hat2 complex would also be of considerable interest. Hat2 is similar to the p48 subunit of CAF-1 (see Chromatin) and to similar subunits in histone deacetylase HD1 in humans and a related deacetylase in Drosophila. These related proteins may be histone-binding subunits.

 2. Histone Deacetylases

Histone deacetylases (HDACs) are enzymes that remove acetyl groups, which generally results in repression of transcription. They also appear to be targeted to particular sites by gene regulatory proteins, in this case repressors, in organisms from yeast to mammals. (Intriguingly, the yeast deacetylase RPD3, which is also a repressor of silencing at telomeres and mating type loci, specifically deacetylates Lys5 and Lys12 in H4 in vitro, consistent with the requirement for Lys12 in an acetylated form for heterochromatin.) Like the HATs, HDACs function in the context of large complexes in which interaction of the HDAC and repressor appears to be mediated by corepressors. In yeast the deacetylase RPD3 is complexed with the corepressor sin3. The mammalian homologue, mSin3, has four paired amphipathic helix (PAH) domains that can interact with several repressors and corepressors. One is the Mad/Max repressor (Mad antagonizes the transcriptional activation and transformation functions of the Myc oncoprotein ); another is N-CoR, a repressor for the thyroid hormone receptor, which exists in a complex with mSin3A and the mammalian homologue of RPD3. Progress in this rapidly evolving field has been succinctly summarized (1, 16). The finding that the 5-methylcytosine binding protein, MeCP2, recruits Sin3 (17, 18) and a deacetylase suggests a way of ensuring that chromatin in methylated (inactive) regions of the genome remains unacetylated and in a stable higher-order structure.

References

1. M. Grunstein (1997) Nature 389, 349–352. 

2. B. M. Turner, A. J. Birley, and J. Lavender (1992) Cell 69, 375–384. 

3. M. Braunstein et al. (1996) Mol. Cell. Biol. 16, 4349–4356. 

4. J. R. Bone et al. (1994) Genes Dev. 8, 96–104. 

5. M. Vettese-Dadey et al. (1996). EMBO J. 15, 2508–2518. 

6. K. P. Nightingale, R. E. Wellinger, J. M. Sogo, and P. B. Becker (1998) EMBO J. 17, 2865–2876.

7. T. R. Hebbes, A. W. Thorne, and C. Crane-Robinson (1988) EMBO J. 7, 1395–1402. 

8. J. E. Brownell et al. (1996) Cell 84, 843–851. 

9. M.-H. Kuo et al. (1998) Genes Dev. 12, 627–639. 

10. L. Wang, L. Liu, and S. Berger (1998) Genes Dev. 12, 640–653. 

11. M.-H. Kuo et al. (1996) Nature 383, 269–272. 

12. A. Imhof et al. (1997) Curr. Biol. 7, 689–692. 

13. K. Struhl (1998) Genes Dev. 12, 599–606. 

14. R. N. Dutnall, S. T. Tafrov, R. Sternglanz, and V. Ramakrishnan (1998) Cell 94, 427–438. 

15. A. Verreault, P. D. Kaufman, R. Kobayashi, and B. Stillman (1997) Curr. Biol. 8, 96–108. 

16. M. J. Pazin and J. T. Kadonaga (1997) Cell 89, 325–328. 

17. X. Nan et al. (1998) Nature 393, 386–389. 

18. P. L. Jones et al. (1998) Nat. Genet. 19, 187–191.