Many human diseases are the result of inappropriate changes in gene expression resulting in deleterious phenotypes of specific cells. For example, loss of expression of tumour suppressors and/or ectopic expression of oncogenes underlie many cancers, a switch from an adult to a fetal gene-expression profile in cardiac myocytes results in cardiac hypertrophy and changes in the expression of many ion channel genes leads to a phenotypic switch from contractile to proliferative smooth muscle cells in vascular diseases such as neointimal hyperplasia and atherosclerosis. Understanding the molecular mechanisms responsible for these changes in gene expression is a major goal, in order to identify novel therapeutic targets.
Eukaryotic DNA is wrapped around histone proteins forming a complex known as chromatin. Although initially thought to solely serve a structural role, facilitating the packing of DNA into a single cell nucleus, more recent studies have uncovered a multitude of post-translational histone modifications, including acetylation , methylation , phosphorylation , ubiquitination  and SUMOylation  that play a role in regulating transcription. Individual modifications have been proposed to contribute to a histone code , and, with at least 39 modifications currently identified within the core histone octamer, there is huge potential for the encoding of information. However, although some specific modifications have been associated with particular gene locations and/or transcriptional states, deciphering such a code has remained an elusive goal. Many different chromatin-modifying enzymes have been identified, and it is becoming increasingly apparent that many of these enzymes show functional interactions and many coexist as part of multiprotein complexes within the cells or can be recruited to an individual gene by a single transcription factor. Analysis of the mechanism by which transcription factors that recruit multiple chromatin-modifying enzymes regulate transcription would provide insight into the functional interactions between specific histone modifications and chromatin-modifying enzymes.
One transcription factor that recruits multiple chromatin modifying enzymes is REST [RE1 (repressor element 1)-silencing transcription factor], also known as NSRF (neuronal restrictive silencer factor) [6–8]. REST recognizes a specific DNA sequence, RE1 [also known as NRSE (neuron restrictive silencer element)], via eight zinc fingers and recruits two separate co-repressor complexes via interactions with mSin3 and CoREST by N- and C-terminal repression domains respectively [9–13]. The chromatin-modifying enzymes recruited as part of these complexes or directly by REST include class I HDACs (histone deacetylases), HDAC1 and HDAC2 [9–12], class II HDACs, HDAC4 and HDAC5 , the ATP-dependent chromatin-remodelling enzyme BRG1 , the lysine-specific demethylase LSD1  and the histone methyltransferase G9a . Work from several independent groups has uncovered functional interactions between these enzymes and has highlighted the effects of some post-translational modifications in regulating the activity of individual enzymes.
Chromatin-remodelling enzymes use energy derived from ATP hydrolysis to alter DNA–nucleosome interactions, and their association with transcriptional activation was thought to result from an opening of the chromatin structure, facilitating access of transcription factors and RNA polymerase II. In addition to its involvement in transcriptional activation, one such chromatin remodelling enzyme, BRG1, is also recruited by REST and enhances transcriptional repression . BRG1 is also found associated with other co-repressor complexes, including NuRD (nucleosome remodelling and deacetylase), SMRT (silencing mediator of retinoid and thyroid receptors) and NCoR (nuclear receptor co-repressor) [18–20], although its mechanism of action within these complexes is not known. BRG1 contributes to transcriptional repression by REST by stabilizing the binding of REST to RE1 sequences . REST recruitment to RE1 sites is reduced subsequent to expression of a mutant BRG1 which is incapable of hydrolysing ATP, suggesting that the BRG1 chromatin-remodelling activity isz required to enhance REST interactions with DNA, possibly by creating a more accessible binding site . BRG1 also contains a bromodomain which can recognize acetylated lysine residues. Specifically, BRG1 has been reported to recognize acetylated histone H4 Lys8 (H4K8) within chromatin and acetylation of H4K8 is required for BRG1 recruitment to the IFN-β (interferon β) gene . In vitro, BRG1 can also recognize acetylated H3K9/K14  and acetylated H4K12 . Increased acetylation of histone H4 results in enhanced REST recruitment, which is dependent on both the chromatin-remodelling activity and the presence of the BRG1 bromodomain , thus REST can be recruited more efficiently to RE1 sites present within acetylated chromatin. It has been a long-held belief that increased acetylation of histone tails is associated solely with transcriptional activation; however, these data suggest that increased H4K8 acetylation could also be associated with transcriptional repression. Perhaps paradoxically, binding of REST may remove the chromatin mark that was responsible for its initial recruitment, as REST is associated with histone deacetylase activity and recruitment of REST has been shown by several groups to reduce acetylation of local chromatin [9–12,25]. However, these studies have examined global histone H4 acetylation, and it is possible that the H4K8 acetylation mark is protected from deacetylation because of its interaction with BRG1. As antibodies that recognize specific acetylated lysine residues are available , such a hypothesis could be tested directly.
Interplay between deacetylation, demethylation and methylation
REST can interact with co-repressor complexes that contain several chromatin-modifying enzymes that include histone deacetylases, a histone H3K4 demethylase and a histone H3K9 methylase. Individually, each one of these enzymes could contribute to transcriptional repression, although it is becoming apparent that the enzymes themselves show a great deal of interdependency and co-operation. Such dependency is at least partly due to the actions of one enzyme resulting in a more favourable substrate for a second enzyme. For example, demethylation of H3K4 is stimulated by the presence of HDACs within the CoREST complex because acetylated H3K9/K14 inhibits the ability of the demethylase LSD1 to demethylate H3K4 . Once demethylated, H3K4 is recognized by another component of the CoREST co-repressor complex, BHC80 (B-Raf–HDAC complex 80 kDa), via its PHD domain, and recruitment of BHC80 is important for repression of SCN1A (type 1 voltage-gated Na+ channel, α subunit), SCN3A (type 3 voltage-gated Na+ channel, α subunit) and SYN1 (synapsin I) in HeLa cells . Deacetylation of H3K9/K14 can also inhibit the recruitment of the MLL (mixed lineage leukaemia) complex which contains a H3K4 methylase activity and activates transcription [29,30]. Thus HDAC activity results in both enhanced removal of H3K4 methylation and a reduction in the activities that methylate H3K4. Some of the most widely used tools to study an involvement of histone acetylation in transcription are HDAC inhibitors such as TSA (trichostatin A). Because the HDAC activity has the potential to promote LSD1 function, one predicted effect of TSA would also be inhibition of H3K4 demethylation, resulting in a potential increase in H3K4 methylation, and functional effects of HDAC inhibitors may be due to their secondary effects on H3K4 methylation rather than on their direct effect on histone acetylation. REST also recruits the histone methylase, G9a, which methylates H3K9 to form mono- or di-methyl H3K9. G9a activity is also dependent on HDAC activity, as it cannot catalyse the addition of methyl groups on to a residue that is already acetylated. Once methylated, H3K9 is no longer a substrate for acetyltransferases (reducing the potential for MLL recruitment) and is also recognized by the heterochromatin proteins HP1α and HP1γ, which mediate chromatin condensation, adding a further layer of transcriptional repression [31,32].
The role of chromatin changes in cardiac hypertrophy
Cardiac hypertrophy results in an increased heart size because of the increase in size of individual cardiac myocytes. Although it is a physiological response by the body to generate increased heart muscle mass in response to an increased workload for the heart, excessive hypertrophy leads to heart failure. One feature of cardiac hypertrophy is a change in the gene-expression programme of individual myocytes from an adult profile to one characteristic of a fetal heart. In vivo models of cardiac hypertrophy show that REST expression is reduced in this phenotypic adaptation and that inhibition of REST is sufficient to drive cardiac hypertrophy in transgenic mice . In adult myocytes, REST represses many genes expressed during fetal development of the heart, and loss of REST allows their re-expression . Although loss of REST function is sufficient to drive hypertrophy, sustained expression of REST prevents gene changes in response to hypertrophic stimuli , suggesting that loss of REST is necessary for the hypertrophic response. Two genes whose expression in cardiac hypertrophy has been well studied are Nppb and Nppa which encode the brain and atrial natriuretic peptides, BNP and ANP, respectively. Both of these genes contain RE1 sites and are regulated by REST in cardiac myocytes [25,33,34]. Inhibition of REST function in cardiac myocytes leads to an increase in the expression of both Nppb and Nppa [25,33] and changes in the levels of histone acetylation and histone methylation at the promoters of these genes . Increased histone H4 acetylation follows loss of REST at the Nppb and Nppa RE1 sites and this is associated with increased expression of Nppb and Nppa as treatment of cardiac myocytes with the HDAC inhibitor TSA results in increased levels of Nppb and Nppa mRNA . However, in the presence of TSA, loss of REST results in a further increase of Nppb and Nppa mRNA, suggesting that REST does not repress these genes via HDAC activity alone. Loss of REST function is also associated with increased level of H3K4 dimethylation at the Nppb and Nppa promoter . The increased methylation of H3K4 is required for increased gene expression as incubation of cardiac myocytes with MTA (5′-deoxy-5′-methylthioadenosine), a methylase inhibitor, prevents the increase in Nppb and Nppa mRNA levels associated with inhibition of REST . REST recruits HDAC activity as part of both the mSin3 complex via the N-terminal repression domain and as part of the CoREST complex via the C-terminal domain. Recruitment of either of these complexes is sufficient to maintain repression of the Nppa mRNA via low levels of acetylation and H3K4 methylation at the Nppa promoter; however, both complexes are required to efficiently repress the Nppb mRNA . An important functional interaction between acetylation and methylation is highlighted by the observation that loss of the N-terminal repression domain, which interacts with mSin3, results in an increase in both histone acetylation and H3K4 methylation of the Nppb promoter, even though recruitment of the demethylase, LSD1, by the C-terminal repression domain is maintained. Loss of the C-terminal repression domain of REST and thus loss of LSD1 results in greater increases in histone acetylation and H3K4 methylation and a greater level of Nppb mRNA compared with the N-terminal repression domain . The ability of the N-terminal repression domain to provide some transcriptional repression of Nppb and to maintain a partly reduced H3K4 methylation in the absence of LSD1 probably reflects the ability of HDACs to remove a binding site for activating complexes such as MLL that contain H3K4 methylase activity [29,30]. The ability of REST to repress Nppb and Nppa expression will be antagonistic to the activity of transcriptional activators which are most likely to recruit histone acetyltransferases to these two genes. Possibly, REST can effectively repress the Nppa gene using only a single repression domain, while both repression domains are required for Nppb repression because the number and/or effectiveness of transcriptional activators recruited to the Nppb gene is greater than for Nppa in cardiac myocytes.
In addition to cardiac hypertrophy, alterations in REST function have been implicated in several other disease states. In colonic epithelia, REST is a tumour suppressor, and loss of the REST gene or mutations of REST are associated with colon cancer . In VSMCs (vascular smooth muscle cells), reduced REST expression is associated with the switching of the cells from a contractile quiescent phenotype to a non-contractile proliferative phenotype . Such phenotypic switching of VSMCs is important for angiogenesis and vascular repair, but also underlies many vascular proliferative diseases such as neointimal hyperplasia and atherosclerosis. Loss of REST function has also been linked with lung cancer, although, for this disease, the cause of reduced REST function appears to be a change in the splicing of the REST gene itself . Several disease states result from enhanced REST function. Normal, but not mutant, huntingtin protein can directly interact with REST and sequester REST protein in the cytoplasm of striatal and cortical neurons . The increased levels of nuclear REST in cortical neurons of patients with Huntington's disease results in the reduced expression of the survival growth factor BDNF (brain-derived neurotrophic factor), which has been proposed to contribute to the neuronal death observed in Huntington's disease . Medulloblastomas, which are tumours of the brain, show high levels of REST , and, in response to epilepsy or ischaemia, REST mRNA levels are elevated in the brain [40,41] and results in neuronal death due to reduced expression of the AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid) receptor subunit gene Gria2 . In some disease states, the target genes of REST have been identified, and how the changes in their expression levels contribute to disease is currently being investigated [33,36,41–43].
Although deciphering the complexities of the histone code still remains elusive, it is clear that neither specific chromatin modifications nor individual chromatin-modifying enzymes function in isolation to regulate gene expression. Rather, a co-ordinated activity of modifying enzymes produces a cascade of changes, resulting in the switching from one chromatin state to another and a change in the level of gene transcription. An important future goal will be to develop small-molecule inhibitors of chromatin-modifying enzymes. Combinations of such inhibitors will provide novel therapeutic strategies to treat diseases in which dysregulation of gene transcription is important.
Transcription: A Biochemical Society Focused Meeting held at the University of Manchester, U.K., 26–28 March 2008 as part of the Gene Expression and Analysis Linked Focused Meetings. Organized and Edited by Stefan Roberts (Manchester, U.K.) and Robert White (Beatson Institute, Glasgow, U.K.).