The purpose of this review is to introduce the exciting field of epigenetics and to describe how it could explain the mechanisms by which environmental changes induce pathological gene expression and determine cell phenotype and function in IBD. We outline how epigenetics research in the context of a variety of clinical conditions, but mainly in cancer, has begun to define the role of multiple combinations of modifications to chromatin, diverse families of enzymes, and non-coding RNAs in determining transcriptional outcomes. These findings are applicable to understanding the context-specific events that underlie the expression of genes in diseases like IBD and have the potential to reveal new targets for improved IBD therapy. The current status of epigenetics-based therapies is also summarized. (Inflamm Bowel Dis 2012)
For clinicians and scientists interested in inflammatory bowel disease (IBD), defining new factors in disease pathogenesis remains a major challenge. In this review we will demonstrate how the rapidly expanding field of epigenetics has the exciting potential to explain the causes of pathological gene expression (specifically at the level of gene transcription) and the mechanism by which environmental changes determine cellular function and phenotype. Defining the role of epigenetics in the context of the inflamed intestine would both advance our knowledge of IBD pathogenesis and enhance prospects for new effective therapeutic strategies to be developed. An explanation will be given of heritable/epigenetic changes versus changes to chromatin per se that regulate gene transcription. The review will be focused on the role of chromatin in gene expression and not cover its known roles in DNA replication, DNA repair, and cell cycle progression. The potential implications for future IBD therapy will also be discussed.
CHRONIC INFLAMMATION AND IBD PATHOGENESIS
Inflammation is a major protagonist in the initiation and perpetuation of autoimmune and inflammatory diseases such as IBD. An inflammatory response is beneficial (physiological inflammation) if it is self-limiting, does not induce significant tissue damage, and eventually clears the insult resulting in tissue repair. However, in diseases like IBD the inflammatory response becomes detrimental (pathological inflammation) when it becomes dysregulated or chronic, leading to tissue damage and loss of function. Chronic intestinal inflammation associated with IBD may be a secondary consequence of innate immune deficiency or dysfunction.1–3 Abnormal innate immune responses to the intestinal microbiota are a major theme of IBD and are thought to result from a breakdown in tolerance to the microbial flora, although the evidence for this in Crohn's disease (CD) is limited and in ulcerative colitis (UC) is still lacking.2, 3
A dysregulated and inappropriately persistent inflammatory response is central to the development of both major types of IBD: CD and UC.1, 3 Both are chronic, relapsing, and progressive inflammatory conditions. CD affects the entire gastrointestinal tract, whereas UC affects the colonic mucosa, and both forms are associated with increased risk of colon cancer. Both conditions have similar symptoms but only CD affects the small bowel, is characterized by transmural inflammation, and complicated by the occurrence of strictures and fistulae. The general consensus is that a combination of multiple factors contributes to CD and UC: genetic susceptibility, an abnormal response to the commensal gut flora, and specific changes in the environment induced by social and economic advances.2 These will be described prior to outlining how epigenetic and chromatin states can contribute to our understanding of IBD pathogenesis.
Both UC and CD are associated with the altered expression of regulatory and proinflammatory cytokines.4 A clear separation of the cytokines produced in the two types of IBD has not been reported, presumably due to the heterogeneity of IBD and the large number of cytokines that are likely to be involved. CD is associated with increased production of IL-12/IL-23 and interferon-gamma (IFN-γ) / interleukin (IL)-17. Recent identification of genetic susceptibility loci implicates the IL-23 pathway as mediating the inflammatory response in UC as well as CD.5 IL-23 and IL-6 along with other cytokines such as IL1-β stimulate T helper (Th)17 cells to produce IL-17A and IL-22 and inhibit Foxp3+ T regulatory (Treg) cell function.1 Consistent defects in Treg function have not been described in IBD to date. Excess IL-13 production has been described in UC that impairs epithelial barrier integrity and indicates that colonocytes are central to the amplification of intestinal inflammation.2
Genome-wide association studies (GWAS) along with meta-analyses by international IBD consortia have resulted in the number of genetic susceptibility loci reaching close to 100.5 The identified loci involve a wide range of biologic pathways that affect innate immunity, adaptive immunity, endoplasmic reticulum (ER) stress, autophagy, and metabolic pathways. These loci include mutations in autophagy genes and promote the elimination of bacteria, e.g., NOD2, ATG16L1, and IRGM in CD and UC-specific genes that are important in epithelial barrier integrity, e.g., CDH1, LAMB1, HNF4A. Many others have been associated with CD and UC that are important in T-cell differentiation and inflammation, e.g., IL-23R, IL-12B, the PTPN2 and CD40 ligand genes, which are modulators of the inflammatory response, and the ER stress transcription factor XBP1 that maintains IEC homeostasis.1, 5 While the large number of loci has provided insights into new pathways involved in pathological intestinal inflammation, most of these gene variants are not specific to IBD and the low penetrance of these genes limits their use in screening for susceptible individuals.
Abnormal Response to the Gut Microbiota
Studies in animal models of IBD have shown that experimental colitis does not develop in germ-free conditions but occurs once the bowel is colonized with commensal microbiota, indicating that intestinal microbes are indispensable for the development of IBD.6 Based on current evidence, it has been proposed that the immune system reacts inappropriately to the microbes of the gut and this is thought to be central for the development of IBD. However, an abnormal response to the gut microbiota may be secondary to the disruption of the epithelial barrier.1
The discordance of IBD among monozygotic twins and increased incidence of IBD in countries undergoing rapid Westernization suggests environmental factors are likely to be important in disease development.2 Links between smoking, diet, medication use, stress and high socioeconomic status have been reported.2 These observations together with the observed low penetrance of gene susceptibility loci and the phenotypic plasticity of cells involved in intestinal inflammation strongly implicates epigenetic changes as key contributors to IBD pathogenesis. These factors are described below along with a glossary to supplement definitions of relevant terms in the text (see Appendix).
THE CONCEPT OF EPIGENETICS
All cells in the body have the same genome, but exhibit dramatically different phenotypes and functions. These differences arise from the epigenome and the entirety of epigenetic features possessed by an organism. The epigenome can be viewed as a system of chemical tags that attach to DNA and its associated histone proteins. These tags are retained through cell division to regulate the access and recruitment of proteins that switch genes on and off during development, cell differentiation, and disease.7 Epigenetics is now defined as “the inheritance of variation (-genetics) above and beyond (epi-) changes in the DNA sequence.”8 Epigenetic mechanisms and chromatin states are crucial, as they allow the cell to adapt to environmental cues, adjust its phenotypic development, and evolve accordingly. Bona fide epigenetic changes are those changes to chromatin that persist throughout mitosis and that are therefore inherited through cell division, although the mechanisms by which this occurs are largely unknown. Thus, epigenetics bridges the gap between genotype and environment and might explain the phenotypic discordance between monozygotic twins, age of disease onset disease severity, gender and parent-of-origin effects, and sporadic incidents of disease.9
Three distinct and interconnected mechanisms regulate the epigenome: chromatin structure modulation, DNA methylation, and noncoding RNAs, all of which are important contributors to the regulation of pathological gene expression.
CHROMATIN STRUCTURE MODULATION
Chromatin is an exquisitely organized and dynamic structure that responds to environmental cues to alter the accessibility of genes to the transcription machinery.10, 11 The basic unit of chromatin is the nucleosome, containing 147 basepairs (bp) of DNA wrapped around an octamer of core histone proteins. This consists of two of each type of core histone: Histone 2A (H2A), H2B, H3, and H4. Based on the classical model of chromatin structure, transcription is repressed when nucleosomes are condensed, i.e., packed tightly together, into what is known as heterochromatin and activated within a more relaxed or open chromatin structure when the nucleosomes are further apart, so-called euchromatin. A more complex variation of nucleosome positioning is now recognized and the functional significance of this for context specific gene regulation is under investigation.12 Specific families of regulatory proteins such as transcription factors or activators interact with their target gene sequences to activate transcription. Coregulators and other transcription proteins are recruited in a specific and temporally sequential manner10, 11 (Fig. 1).
To date, two major and distinct types of enzymatic activities have been identified as capable of altering chromatin structure: chromatin modifying enzymes that perform a large variety of posttranslational modifications (PTMs),13 both of histones and nonhistone proteins, and chromatin remodeling enzymes that alter the position of nucleosomes in an adenosine triphosphate (ATP)-dependent manner resulting in chromatin that is permissive or repressive to gene transcription. Recently, histone variant exchange has also been recognized as an important determinant of chromatin structure.14
Histones Modifications and Enzymes
Histone PTMs can modify the chromatin structure by altering the physical properties (electrostatic charge) of the chromatin fiber, thus directly affecting the accessibility of the transcription machinery, or by recruiting/stabilizing the localization of protein complexes to chromatin that recognize or perform specific posttranslational modifications.13 All these events are crucial to the regulation of gene transcription (see below).
The core histones bear an N-terminal amino acid tail of ≈20–35 residues in length and rich in basic amino acids. The N terminal and C terminal tails extend from the surface of the nucleosome and play an important role in folding of nucleosomal arrays into a higher-order chromatin structure. More than 60 residues (usually lysine, arginine, serine, and threonine) within the N-terminus of histones are capable of being posttranslationally modified. More recently, 67 additional modifications have been demonstrated, some of which are within histone globular domains. These include acetylation, methylation, ubiquitination, phosphorylation, SUMOylation (these are the most studied and are discussed below), and the little understood propionlylation, butyrylation, formylation, citrullination, proline isomerization, ADP ribosylation, tyrosine hydroxylation, glycosylation, and lysine crotonylation.13, 15, 16 The combination of site and type of modifications generate diverse local chromatin structures and functional outcomes with the potential to define disease-specific gene expression mechanisms. The issue of whether histone modification patterns constitute a “histone code” has become a controversial one.11, 12 The general consensus is that combinations of histone PTMs provide a platform for the recruitment of transcription factors/coregulators that determine the transcriptional state of chromatin and elicit downstream biological outcomes.11, 17, 18 Most modifications have been found to be dynamic and several families of histone-modifying enzymes that introduce, maintain, and reverse the modifications have been identified. Space constraints preclude the inclusion of anything other than a brief summary of the key features of each of the major chromatin modifications that have been most studied and the enzyme families that regulate them.
Histone Acetylation and Deacetylation
The acetylation and deacetylation of histones occurs due to the activities of histone acetylases (HATs) and histone deacetylases (HDACs).19–21 Four main families of histone acetyltransferases (HATs) have been identified (GNAT, MYST, CBP/p300, and the Nuclear/Steroid receptor family) based on primary structure homology.21, 22 HDACs occur in four classes (class I, II, III, or Sirtuins and IV) designated by sequence homology to the yeast HDACs.23 Class I are integral components of corepressor complexes, e.g., NuRD and CoREST. Class II bind transcription factors and control transcription by recruiting corepressors and coactivators. Class III comprises the NAD-dependent Sir2 related deacetylases also called Sirtuins and Class IV is HDAC11.
HATs and HDACs employ two major mechanisms. Both families of enzymes affect the level of transcription by directly recruiting transcriptional activators or repressors to the upstream regulatory sequences of genes and by exerting broad global functions in the genome. They function in complexes whose constituents determine lysine and substrate specificity and function. Histone acetylation is a major determinant of chromatin conformation and plays a critical role in the transition between maintenance of open/relaxed state or a compact/folded state. Acetylation of most individual lysines in H3 and H4 tails correlates positively with gene transcription. Although published evidence indicates a connection between acetylation and gene transcription, the mechanism by which this occurs is unclear.11, 12 The transfer of an acetyl group from the cofactor acetyl coenzyme A to histone lysine residues catalyzed by acetylases neutralizes positive charges on the histones, leading to opening of chromatin. This event promotes both access of the transcription machinery and the passage of RNA polymerase II. Acetylation may also provide binding sites for proteins involved in gene activation such as those containing bromodomains like BRG1 (catalytic subunit of SWI/SNF) that binds acetylated H4K8.
Histone methyltransferases fall into two categories: those that methylate lysine residues (protein lysine-methyltransferase, PKMTs) and those that add methyl groups to arginine (protein arginine-methyltransferases, PRMTs).21, 24 Both classes of methylase catalyze methyl transfer from the cofactor S adenosyl L methionine to a nitrogen atom of their target amino acid. Of the PKMTs, some selectively target a particular histone lysine residue, e.g., SET7/9 targets lysine 4 in histone H3; however, more than one enzyme can methylate the same lysine residue, e.g., methylases MLL1 to 5, hSET1A, HSET1B, and ASH1 can all methylate H3K4, a histone mark associated with activation of transcription. Histone methylation has also been linked to repression of transcription depending on the location of the methylated lysine. The most important targets for these enzymes with respect to gene regulation are likely to be histones, as methylation of these correlates with chromatin remodeling and levels of gene transcription.
All human PKMTs, (apart from Dot1L/KMT4) contain a 130 amino acid catalytic domain called the SET domain.25 The SET domain PKMTs have been divided into related families based on sequence alignment. Seven families have been identified, the major ones include: Suppressor of variegation 3-9 (SUV39) also known as the SET1/MLL family; the SET2/NSD family; the retinoblastoma protein-interacting zinc finger protein (RIZ) or PRDM family; the SET and MYND domain containing family (SMYD); the enhancer of zeste family (EZ); the SUV4-20 family; a family that comprises the enzymes SETD7 and SETD8 (PRSET7) and DOT1L. The number of SET domain containing enzymes in humans has recently been extended to 51 (51 SET domain proteins plus DOT1L). Histone methylation can be reversed by two families of lysine demethylases: the LSD1 and jumonji domain containing enzymes: JmjC.26
There are eleven human PRMTs, eight of which have been reported to have methylase activity.27 CARM1 (PRMT4), and PRMT1 have a known role in nuclear factor kappa B (NF-κB)-dependent gene transcription. Sequence conservation among the PRMTs is low. There may be as many as 50 PRMT enzymes in humans but little is known about them. Two types of putative arginine demethylases have been reported: JMJD6 and the peptidylarginine deiminases (PADIs), but whether these are true demethylases of histone arginine residues has been questioned.
In the last few years rapid progress has been made in understanding the role of kinase-dependent phosphorylation of histones and nonhistone proteins in the regulation of gene transcription.28 Histone kinases integrate upstream signals and transmit them to downstream effectors. As with the acetylases and methylases, more than one kinase can modify the same histone residue. For example, four kinases are known to modify threonine 3 in histone H3 and seven enzymes can phosphorylate H3S10, the best-characterized histone phosphorylation event involved in the transcriptional activation of inflammatory genes (see Epigenetic Control of Inflammation). It is known that crosstalk occurs between phosphorylation and other modifications such as acetylation and methylation when they are in close proximity, resulting in parallel or synergistic effects on gene transcription.28 H3S10 phosphorylation inhibits the repressive H3K9 methylation promoting H3K4 methylation and H3K14 acetylation. This process involves several kinases that phosphorylate H3S10, e.g., Aurora B, VRK1, MSK1/2, PIM1, RSK2, PKb/AKt, and IKKα. Interestingly, coexistence of the repressive mark H3K27me3 and S28ph displaces the polycomb group of proteins bound on H3K27me3 to activate gene transcription,29 offering a functional explanation for the increasingly frequently observed simultaneous presence of activation and repression marks on gene promoters.
The addition of ubiquitin (Ub) comprises the attachment of a single molecule to one of multiple lysines, i.e., mono-ubiquitination or multiple Ub molecules in the form of chains, i.e., poly-ubiquitination.30, 31 Ubiquitination of H2A is linked with gene activation or repression depending on the chromatin context and H2B ubiquitination plays a role in gene silencing.31–33 Ubiquitination of chromatin substrates alters their ability to regulate gene transcription, by influencing their localization or by regulating other chromatin modifications such as phosphorylation and methylation. Two examples are H2A ubiquitination at lysine 119 that inhibits basal expression of some LPS-inducible genes34 and H2B ubiquitination that is a prerequisite for H3K4 and H3K79 trimethylation. Deubiquitinating (DUB) enzymes remove Ub molecules from ubiquitinated histones.31 Five DUB families have been identified that have an increasing number of substrates including H2A and H2B that they share with ubiquitinating enzymes, However, the activities of these two enzyme families result in distinct functional outcomes suggesting they differ in their mechanisms of action and regulation.
SUMO (small ubiquitin-related modifier) is a 100 amino acid polypeptide that in a manner similar to ubiquitination, is covalently attached to target proteins.35 Four SUMO proteins have been described in mammals. SUMOylation has been associated with transcriptional activation but it may also have a role in gene repression. SUMOylation of H4, H2B, and the H2A variant H2A.Z has been demonstrated in yeast. Even though histone SUMOylation is not heritable and therefore strictly not an epigenetic modification, it is nevertheless an important contributor to chromatin structure and function. Links between acetylation and sumoylation and between SUMO and HDACs have been shown in which HDAC1 SUMOylation modulates its biological activity.36 Also, SUMOylation of transcription repressors and corepressors is important in their recruitment and function.
Chromatin remodeling complexes are specialized multiprotein machines that use ATP hydrolysis to evict, slide, or rotate nucleosomes across gene regulatory regions.37, 38 These events enable the ordered access of transcription factors to specific genes. They also exchange canonical histones with variant ones that alter the biochemical and biophysical properties of nucleosomes and determine nucleosome composition. Five families of ATP-dependent remodeling complexes exist, classified on the nature of their ATPase (SWI/SNF, ISWI, NuRD/Mi-2/CHD, INO80, and SWR1). These consist of a catalytic ATPase domain belonging to the SF2 helicase superfamily via which specific interactions occur with nucleosomal substrates. These enzymes couple ATP hydrolysis to protein conformational changes in order to control nucleosome positioning and regulation of gene expression.
Variant versions of core histones and the linker histone H1 exist in low levels and are thought to carry out specialized functions in certain areas of the genome and in pathological conditions that differ from canonical histones.14, 39 They exhibit distinct protein structure, sequence, posttranslational modifications, and biological functions. In eukaryotes, histone H3 has three variants: a testis-specific variant (H3t), H3.3, and CENPA (localized to centromeres) and histone H2A has four variants: H2A.X, H2ABbd, H2A.Z, and MacroH2A. Specialized histone H3 and H2A variants can be modified and alter key biochemical and biophysical properties inherent to nucleosomes. Variants of histones H4 and H2B have been reported in some eukaryotes.14, 40 Histone variant H3.3 has been associated with transcriptionally active chromatin, while H2A.X and H2A.Z are preferentially enriched on inactive promoters, although H2A.Z has also been associated with activation of gene transcription. Their deposition can impact chromatin structure by altering nucleosome stability, high-order chromatin structure, or carrying specific PTMs.14 MacroH2A is the largest H2A variant that is posttranslationally modified and generally enriched in transcriptionally silent regions.
Epigenetic regulation can take the form of modification of the genomic DNA itself.41, 42 DNA can be modified by covalent addition of methyl groups on cytosine residues of CpG dinucleotides (i.e., cytosine followed by a guanine), catalyzed by enzymes known as DNA methyltransferases (DNMTs). In mammals, five members of the DNMT family have been reported: DNMT1, DNMT2, DNMT3a, DNMT3b, and DNMT3L, but only DNMT1, DNMT3a, and DNMT3b possess methyltransferase activity.43 The catalytic members of the DNMT family are customarily classified into de novo DNMTs (DNMT3a and DNMT3b) and maintenance DNMTs (DNMT1). The activity of the de novo DNMTs is influenced by sequences next to the target CG sites and occurs during embryonic development. DNMT1 is essential for maintenance of cytosine methylation, preferentially methylates hemi-methylated DNA during DNA replication and cell division, and contributes to the retention of genome stability. DNMT1 also has reported de novo activity in human cancer cells. DNMTs can read histone modifications, which lead to their recruitment to nucleosomes. There is a strong inverse correlation between DNA methylation and, for example, H3K4 methylation.44 Accumulating evidence suggests that DNA methylation is regulated and targeted to specific regions of the genome by histone modifications, chromatin remodeling, and non-coding RNAs.7 It has been shown recently that DNMTs form RNA-protein complexes with RNAs, like Piwi-pRNA complexes and miRNAs, and in this way DNMTs are directed to specific gene loci to establish methylation patterns.41 Interestingly, the activity and function of DNMTs are modulated by acetylation, methylation, phosphorylation, ubiquitination, and SUMOylation.
Notably, CpG dinucleotides are underrepresented in the human genome, but can be found in clusters—so-called “CpG islands”—which in turn are mainly found in promoter regions. The term “CpG island shores” has recently been coined and refers to regions of lower CpG density that lie in close proximity (≈2 kb) upstream of CpG islands. The DNA methylation of these regions correlates with transcriptional repression, interfering with the access of transcription factors to promoter regions. Surprisingly, DNA methylation within the gene body is a feature of actively transcribed genes.45, 46 Aberrant DNA methylation is a classic hallmark of cancer and a number of other diseases in which loss of DNA methylation as well as hypermethylation of specific promoters occur.41, 42 Also, dysregulated expression of DNMTs has been reported in human cancers, e.g., colorectal, liver, prostate, and breast cancers.41
In addition to 5-methylcytosines, a new DNA modification has been observed: the 5-hydroxymethyl-2′-deoxycytidine (5hmC)47 formed by oxidation of cytosines through the activity of the enzyme ten-eleven translocation 1 (TET1). It has been suggested that oxidation is part of a demethylation pathway of DNA. It has a distinct distribution from methylated cytosine and has been found in promoters, within gene bodies, and has been associated with gene expression and polycomb-mediated gene silencing. The highest levels of 5hmC have been found in the brain and the evidence to date suggests it may have an important role in epigenetic reprogramming of neuronal chromatin and gene expression.48
It is now known that 98%–99% of the human genome is transcribed as noncoding RNA (recently reviewed by Esteller).49 Noncoding RNAs (ncRNAs) comprise endogenous small RNAs usually of about 22 nucleotides in length and are present at lower levels than mRNA, suggesting they have a regulatory function. ncRNAs target the 3′untranslated region (UTR) of mRNAs, with which they share partial sequence complementarity, leading to posttranscriptional gene silencing through translational repression. When a microRNA (miRNA) has complete sequence complementarity with a target mRNA, it directs cleavage of the transcript. So far, ≈700 different miRNAs have been identified in the human genome and may regulate as much as 30% of all human genes.50 The numbers and types of noncoding RNA are continuously increasing and knowledge of their contribution to regulating gene expression in disease is rapidly expanding and has been reviewed elsewhere.49 These currently include short noncoding RNAs (such as Piwi interacting RNAs), mid-size RNAs (such as promoter associated RNAs), and long noncoding RNAs.
Substantial evidence demonstrates that there is crosstalk between the aforementioned epigenetic processes: histone modification, DNA methylation, and miRNAs. For example, the histone methylase enzyme PRMT5 can recruit specific DNA methyltransferases for gene silencing.51 It is clear that defining these cooperative interactions and mechanisms of recruitment of epigenetic factors will provide a major contribution to understanding the molecular basis of pathological gene expression.
EPIGENETIC CONTROL OF INFLAMMATION
Inflammation itself creates a microenvironment able to trigger epigenetic reprogramming that can lead to a distorted immune response. Epigenetic modifications lie at the heart of gene-specific control of the inflammatory response.34, 52, 53 Much of the capacity for selective regulation of inflammatory genes appears to be established at the chromatin level during development.54 Chromatin structure and histone modifications have been shown to be essential selective regulators of a large range of inflammatory genes,34, 52, 53 demonstrating the potential importance of these chromatin-based events in amplifying and perpetuating the inflammatory response (see below).
Regulation of Inflammation-induced Transcriptional Activation
Transcriptional activation involves a gene-specific multistep recruitment of distinct families of proteins to form an active transcription complex or enhanceosome.53, 55 In general, the first step requires a defined set of transcriptional activators to interact with specific DNA sequences present in the promoter and enhancers of genes. This leads to the recruitment of coactivator complexes containing chromatin-modifying enzymes and chromatin remodeling complexes, which in turn facilitate the recruitment and assembly of the preinitiation complex. This comprises RNA polymerase II and general transcription factors to render genes transcriptionally competent (Fig. 1). Then, in order to initiate RNA synthesis and productive elongation of RNA, the sequential phosphorylation of a number of serines at the carboxy-terminal domain of RNA polymerase II is required.55, 56 This creates binding sites for other transcription proteins that are important for mRNA processing and elongation.
A well-studied example of this model of transcriptional activation is given by human interferon-β (IFN-β) gene expression.57, 58 Upon viral infection, a multiprotein complex assembles on the regulatory region of the IFN-β promoter through the binding of inducible transcription factors including NF-κB, interferon regulatory factors, and ATF2/c-jun. This leads to the sequential recruitment of the p300 and GCN5 HATs and the SWI/SNF nucleosome remodeling complex, resulting in nucleosome displacement and recruitment of general transcription factors and RNA polymerase II to activate gene transcription.
Inflammatory gene transcription requires precise combinations of specific regulatory mechanisms based on cell type, stimulus, and individual genetic variation and is also associated with specific histone modifications.59 The induction of some primary response genes (such as IL6, IL8, MCP-1, and IL-12p40) by Toll-like receptor (TLR) signaling is marked by phosphorylation of histone 3 at serine 10 (H3S10), methylation at H3K4 and acetylation at H3K9/K14.60–62
Recent studies exploiting the lipopolysaccharide (LPS)-inducible transcriptional program as a model for understanding the regulation of inflammatory genes have shed light on the role of chromatin structure as a crucial regulator of the kinetics of NF-κB target gene activation.63, 64 Two classes of TLR-induced genes have been identified that are activated with differential kinetics. The first class are primary/immediate early genes that are rapidly induced and protein synthesis-independent, e.g., IL-1β and TNF-α in resting macrophages are packaged in a poised structure that facilitates rapid gene activation under physiological stimulatory conditions.60, 63, 64 These genes have a chromatin structure that resembles that of constitutively active genes and inducible transcription factors that enhance transcription initiation, elongation, and pre-mRNA processing. Evidence for the abnormal formation of a poised structure at cytokine promoters under disease conditions has been reported.65 The epigenetic variability underlying the variation in cytokine production in the human population may explain the differences in the inflammatory response between individuals with inflammatory/autoimmune diseases, including IBD.
Transcriptional activation of the second class of genes, the so-called secondary (delayed and protein synthesis dependent) genes, requires that a chromatin barrier is overcome, e.g., in the case of NF-κB target genes after LPS stimulation of macrophages.60, 63, 66 In other cases, the removal of corepressor complexes, such as NCoR and SMRT, is required, which contain HDACs and other subunits that serve to maintain a repressive chromatin structure.67 Hence, the assembly of promoters into nucleosomes provides a barrier to transcriptional activation and the requirement for nucleosome remodeling is gene-dependent. Another subset of LPS-inducible genes has been identified that undergoes demethylation of H3K9 and H3K27 as a prerequisite for induction.61, 68
Interestingly, some cytokines have been reported to induce changes in chromatin states. IL-6 can increase DNMT1 expression and total DNMT1 activity, suggesting that mediators of chronic inflammation can reprogram genes by altering chromatin components.69
Epigenetic Regulation of Tolerance
Endotoxin tolerance is a state of hyporesponsiveness to LPS (and other inflammatory stimuli) that is induced during conditions of excessive inflammation. Its purpose is to limit inflammation-induced associated pathology.70 Some cell types in specific compartments in the body are naturally tolerant: for instance, intestinal macrophages, although constantly exposed to LPS, make almost no proinflammatory cytokines due to downregulation by stromal cell-derived factors.71 Hence, LPS-tolerant cells are refractory to activation by TNF-α and IL-1β, and the available evidence suggests that this is, at least in part, mediated by epigenetic mechanisms. Studies of endotoxin-tolerant human monocytic cell lines have shown that Rel B directly interacts with the methylase G9a, impairs NF-κB p65 binding, and causes silencing of the IL-1β gene.72, 73 Silencing of TNF-α occurs by sustained binding of heterochromatin protein 1 alpha (HP1α) and methylation of H3K9, reduced phosphorylation of H3S10, and disrupted recruitment of p65 NF-κB to the promoter.74 Dynamic repositioning of nucleosomes from permissive to repressive locations has also been reported to be a contributory factor in silencing of TNF-α during sepsis.75 Similarly, association of HMGB1 and HP1 with the promoters of TNF-α and IL-1β genes promotes Rel B- mediated silencing of these cytokines in tolerant cells.76
Chromatin modifications can allow the selective control of distinct sets of genes after repeated exposure of murine macrophages to inflammatory stimuli.77 For example, LPS can induce the expression of both antimicrobial and proinflammatory genes, but restimulation with LPS induces only antimicrobial genes, suggesting that this is due to persistent gene-specific histone modifications that prime cells to respond to subsequent LPS stimulation. In contrast, the antiinflammatory genes do not retain their histone marks after initial LPS stimulation and so these genes are not expressed upon restimulation with LPS. Hence, defining how epigenetic/chromatin states regulate antimicrobial and inflammatory genes will provide important new insights into IBD pathogenesis as well as potential new targets for IBD therapy.
Aberrant epigenetic/chromatin changes have also been reported to mediate the breakdown of tolerance to specific self-antigens that can lead to a variety of autoimmune diseases such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA) (see below).
Epigenetic Regulation of T-cell Lineage Commitment
Recent advances have highlighted the importance of epigenetic mechanisms in determining the gene expression pattern and specification of T-cell lineage commitment.78 The proper differentiation and development of naïve T lymphocytes into effector T cells (Th1, Th2, Th17, or Treg) regulates immune responses by producing lineage-specific effector cytokines. The dysregulation of these T-cell subtypes can result in a variety of pathological conditions including IBD.1 The specification of T-cell subsets is controlled by epigenetic factors, chromatin modifications, and networks of lineage-specifying transcription factors. An example of this is the Th1/Th2 differentiation process that depends on chromatin modifiers such as histone methylation, histone acetylation, and DNA methylation at conserved noncoding regions in close proximity to specific cytokine loci. Th2 differentiation is dependent on epigenetic silencing of the Th1 Ifng cytokine locus, and differentiation of Th1 cells is determined by silencing of the Th2 cytokine locus that comprises the Il4, Il13, Rad5, and Il5 genes.79
Epigenetic regulation has also proven essential for Th17 cells and Tregs, even though extensive investigations are still needed to elucidate the role of epigenetic reprogramming and histone-modifying factors in the generation of these two T-cell subtypes. IL-17A and IL-17F are typical cytokines representative of differentiated Th17 cells in vivo and in vitro. Histone H3 acetylation and H3K4me3 at the Il17/Il17f promoter and intronic regions have been described during Th17 differentiation.80 The Forkhead box transcription factor Foxp3 is the major transcription factor that determines the fate and identity of Treg and its enhanced expression is required for regulatory effector function level.81, 82 Recent work indicates that both induction and stabilization of Foxp3 expression is under the control of DNA methylation at the promoter level.82
EPIGENETICS AND DISEASE
By definition, epigenetic signals should be sufficiently permanent to be heritable; however, the majority of changes to chromatin are potentially reversible. This is illustrated by the counterbalancing actions of the various enzymes that are responsible for maintaining the epigenome and specific chromatin states.13 Such plasticity renders a cell able to quickly reprogram its phenotype and function in order to adapt to altered physiological conditions, such as chronic inflammation.83 Autoimmune and inflammatory disease increase in frequency with increasing age, with epigenetic dysregulation due to loss of the normal balance between gene-promoting and gene-silencing factors proposed as a potential explanation.84 Indeed, epigenetics and chromatin states are likely to play a leading role in transcriptional reprogramming in the etiology of cancer and chronic inflammatory diseases like IBD.
Epigenetics and Cancer
In the past 20 years the epigenetics of human cancer has become increasingly visible with the growing identification of cancer-specific epigenetic mechanisms including global and gene-specific hypomethylation, hypermethylation, loss of imprinting, and changes in chromatin modifications.85 The expression of many oncogenes in tumors is activated through hypomethylation. By contrast, tumor suppressor gene silencing has been linked to promoter hypermethylation. Recent findings suggest that most of the aberrant DNA methylation occurs in CpG island shores, as observed in colon cancer, with most changes associated with sites that vary in tissue differentiation.86 This is consistent with the “epigenetic progenitor model” of cancer in which epigenetic alterations that affect tissue-specific differentiation are a major factor in cancer pathogenesis.
With respect to genome-wide changes in histone-modifications, e.g., loss of acetylation at lysine 16 and trimethylation at lysine 20 of histone H4, is a common hallmark of human cancer,87, 88 and global histone modification patterns predict the risk of cancer recurrence. Several human PKMTs and PRMTs are associated with human cancers, e.g., overexpression of the polycomb group protein EZH2, a H3 lysine-27 histone methyltransferase, is found in many cancers such as bladder, breast, and prostate and is associated with widespread transcriptional repression.89, 90 In contrast, loss of H3K27 methylation in other cancers suggests that the level of histone modifications depends on the type and origin of the tumor cells.91 In addition, mutations in a number of histone methyltransferases and acetyltransferases genes have been demonstrated in a variety of cancers including a number of gastrointestinal tumors and B-cell lymphomas.92, 93
Both chromosomal rearrangements and mutations, in cancer disrupt chromatin. For instance, chromosomal translocation of the mixed lineage leukemia (MLL) oncoproteins leads to aberrant patterns of H3K79 and H3K4 methylation, resulting in altered gene expression of MLL targets in nearly 80% of childhood leukemias.94 BAF47 (SNF5), a subunit of the SWI/SNF chromatin remodeling complex and tumor suppressor, undergoes heterozygous deletion in chronic myeloid leukemia.95 In addition, other members of SWI/SNF, e.g., the BRG1 and BRM ATPases, are deleted in a variety of cancers including lung and breast cancer.96, 97
miRNAs can also be abnormally expressed or mutated in cancers where they may function as oncogenes or tumor suppressor genes.98 This may be due to the location of miRNAs in regions of chromosomal instability or in close proximity to chromosomal breakpoints. The miRNA downregulation in human tumors is often caused by hypermethylation at the miRNA promoters or by alteration of miRNA biogenesis/processing pathway components.98, 99
Epigenetic Studies of Human Chronic Inflammatory Diseases
Less information is currently available on the role of dysregulated epigenetic mechanisms and chromatin states in human inflammatory states but this area is now emerging.83 Persistent chronic inflammation involves the downregulation of antiinflammatory genes and the upregulation of proinflammatory genes that, as already discussed, are likely to be regulated by epigenetic/chromatin changes. Examples of specific diseases in which such abnormal events occur include human studies of chronic gastritis that show chronic inflammation is associated with increased DNA methylation.100 Also, a strong association has been demonstrated between inflammatory biomarkers, such as C-reactive protein and IL-6, and aberrant DNA methylation in peripheral blood leukocytes of chronic kidney disease patients.101 Experiments on T cells from patients with SLE and RA have concluded that DNA hypomethylation is associated with alterations in gene expression and autoreactivity.102 In fact, several studies have now mapped changes in DNA methylation and histone modifications in SLE and shown global deacetylation of H3 and H4 and loss of DNA methylation.103 The activities of HATs and HDACs are abnormal in asthma and chronic obstructive lung disease (COPD). In COPD, patients have a progressive reduction in total HDAC activity that reflects the severity of the disease104 and bronchial biopsies from asthmatic patients show increased HAT activity and reduced HDAC activity,105 indicating that a disruption in the balance of these enzymatic activities may be important in disease development. A number of other examples have been reported of abnormal DNA methylation, histone modifications, and lysine/arginine methyltransferase activities in inflammatory and neurodegenerative diseases (reviewed7).
Epigenetics and IBD
Surprisingly little information is available on the role of epigenetic changes or chromatin states in IBD. For a decade the literature was restricted to reports of changes in DNA methylation in a small number of genes in colon cancer and UC. The first such report106 showed that methylation of a CpG island in the estrogen receptor gene increased with age in nonneoplastic colorectal epithelium, and was highly methylated in colorectal epithelium from UC. Since then, methylation of IBD susceptibility genes, such as gene-specific methylation of mutator L homolog 1 (MLH1) and hyperplastic polyposis 1 (HPP1), have been linked to colon cancer and IBD,107 and more recently methylation of protease-activated receptor 2 (PAR2) and multidrug-resistance gene 1 (MDR1) has been linked to UC.108, 109 A recent DNA methylation profiling study of matched diseased and nondiseased intestinal tissues from 26 IBD patients identified seven CpG sites that are differentially methylated in inflamed intestinal tissues.110 The first genome-wide DNA methylation profiling of patients with ileal CD was completed in the last year and demonstrated multiple sites of differential DNA methylation in genes involved in the immune response and in genes near to CD susceptibility loci.111 A handful of reports have shown changes in the expression of miRNAs associated with CD and UC.112
Epigenetics and Fibrosis
Fibrosis is a pathological consequence of many chronic inflammatory diseases that manifests as a progressive deposition of excessive extracellular matrix mainly by activated fibroblasts, leading to irreversible tissue damage and organ failure. This is clearly relevant to IBD and particularly to CD, where transmural fibrosis often leads to stricture formation and potentially bowel obstruction, with significant morbidity and the need for surgical intervention. Evidence is now emerging that the fibrogenic process is also orchestrated at a chromatin/epigenetic level. Many recent studies have suggested that HDAC inhibitors trichostatin A, phenylbutyrate, suberoylanilide hydroxamic acid, and valproic acid reduce collagen type I gene transcription, deposition, and fibrosis as well as expression of a smooth muscle actin in different fibroblast types113–118 and various models of fibrogenesis,118, 119 although at least some of the HDAC effects may be mediated by nonhistone proteins. The involvement of HDACs has been reported in epithelial-to-mesenchymal transition, but the role and identity of specific histone deacetylases in mesenchymal transformation and fibroblast differentiation remains unclear. Hypermethylation of specific genes has been shown to be a mechanism of perpetuating fibroblast activation and kidney fibrogenesis.120 Gene-specific hypermethylation and histone hypoacetylation are responsible for fibroblasts activation and diminished cyclooxygenase-2 expression, respectively, in idiopathic pulmonary fibrosis.121, 122 In the liver, the activated fibroblasts phenotype and in vivo fibrosis appear to be controlled by the concerted action of MeCP2 (methyl-CpG binding protein 2), the histone methyltransferase EZH2, and miR132.123 Another recent study indicates that specific histone modifications but not DNA methylation are altered globally by transforming growth factor-β1 (TGF-β1) during epithelial-to-mesenchymal transition of AML12 mouse hepatocytes in vitro.124 Our own studies of intestinal endothelial-to-mesenchymal transition have produced evidence for histone modification changes in association with the induction of type I collagen gene transcription, suggesting for the first time that epigenetic/chromatin changes may be important regulators of intestinal fibrosis (Sadler T, Scarpa M, Rieder F, West G, Stylianou E. Distinctive and persistent epigenetic modifications upon induction of the Col1A2 gene in intestinal endothelial-to-mesenchymal transition, American Journal of Pathology).
EPIGENETICS-BASED THERAPEUTIC INHIBITORS AND TARGETS
Currently, a number of inhibitors have been designed for HDACs, DNMTs, HATs, and HMTs and this is a dynamic and expanding area of research with enormous therapeutic potential.21, 125, 126 It is clear that knowledge of epigenetic changes and chromatin states will continue to provide the basis for the development of novel drugs that in some cases have already shown clinical efficacy (see below). It is also evident that the well-known global inhibitors of HDAC and DNMT also modulate nonhistone targets and therefore a direct link between the clinical efficacy of such inhibitors and drug-induced modulation of specific epigenetic changes needs to be established. Targeting individual histone acetylase/methylase/deacetylase enzymes has a higher likelihood of obtaining drugs with epigenetic specificity (see below).
Histone Deacetylase Inhibitors (HDACi)
The growing interest in HDACi as antiinflammatory agents is due to their efficacy in several animal models of inflammatory, autoimmune, and neurodegenerative diseases such as colitis, arthritis, septic shock, and Huntington's disease, among others.127–131 A wide range of natural and synthetic HDAC inhibitors (HDACi), have been identified and classified into four categories based on their chemical composition as follows: 1) hydroaxmic acid derivatives, e.g., TSA, SAHA, ITF2357; 2) short chain fatty acids, e.g., butyrate and valproate; 3) benzamides, e.g., MS-275; and 4) cyclic peptides, e.g., α Apicidin and depsipeptides.132 Diverse examples of HDACi are currently in clinical trials or therapeutic use.21, 125, 126, 133 Valproic acid is in clinical trials for use in epilepsy and bipolar disorder and is used in HIV-infected patients. ITF2357 is in Phase II clinical trials for active systemic onset juvenile idiopathic arthritis. SAHA and Romidepsin (FK228) are approved for the treatment of cutaneous T-cell lymphoma and are in Phase I/II studies of patients with leukemia. Interestingly, butyrate is a naturally occurring HDACi that is also a major energy source for colonic epithelial cells and a breakdown metabolite of dietary fiber produced by the action of gut microbiota.134 It has been shown to reduce intestinal inflammation and stimulate mucosal repair in trinitrobenzene sulfonic acid (TNBS) colitis but its clinical efficacy has not been demonstrated.135
The effects of HDAC inhibition can be both pro- and antiinflammatory. The multiplicity and diversity of HDACi effects have resulted in controversial results including compromised immunity and exacerbation of atherosclerosis and inflammatory lung diseases. To reduce side effects and to develop agents that target individual deacetylases further research is needed into the modes of action of these enzymes and their inhibitors.
DNA Methyltransferase Inhibitors (DNMTi)
The two most potent DNMTi are nucleoside analogs: 5-azacytidine (5-Aza-CR) and 5-aza-2′-deoxycytidine (5-aza-CdR or decitabine).42 These are also approved by the Food and Drug Administration (FDA) for the treatment of myeloid malignancies, e.g., myelodisplastic syndrome and acute myeloid leukemia. The metabolism and mechanism of action are being investigated. They decrease DNA methylation by becoming incorporated into the DNA of actively replicating tumor cells136 and 5-Aza-CR is incorporated into RNA, where it blocks the translation of oncogenic proteins. Several other cytidine analogs have been developed that have improved stability and efficacy but these are inefficiently metabolically activated. Strategies to develop improved DNMTi are ongoing, including the development of prodrugs of nucleoside analogs such as 5-Aza-Cr where synthetic addition of chemical moieties has improved stability.42 In addition, a number of nonnucleoside analogs are being tested that are less cytotoxic and do not become incorporated into DNA. Short-chain oligos and miRNAs have been developed to block DNMT activity but have yet to be validated. Future drug development must include avoidance of global induction of demethylation by DNMTi, as this can cause malignancy in cancer patients, and therefore the development of therapies targeted at individual epigenetic factors is of major importance.
HDACi DNMTi Combinations
In addition to individual HDACi and DNMTi, a number of combinations are in Phase I/II and preclinical trials for a variety of hematopoietic and solid malignancies including leukemia, breast, ovarian, colon, and prostate cancer.126 A combination of decitabine and valproic acid has significant activity in patients with relapsed/refractory acute myeloid leukemia and myeloid dysplastic syndrome. The combined therapy approach is based on published studies that have described the crosstalk between HDAC and DNMT silencing mechanisms.137, 138 It is clear that further understanding of the interaction between these families of enzymes will continue to contribute to the development of better agents.
Protein Methyltransferase Inhibitors (PMTi)
PMTs have common chemical mechanisms of catalysis and their enzymatic activity has been successfully modulated. Although these enzymes are highly druggable, there is a lack of potent, selective inhibitors of PMTs.21, 24, 125 Natural protein methyltransferase inhibitors have been isolated from Streptomyces spp. and others like chaetocin for SUV39 have been discovered, but many are nonspecific and cytotoxic. Analogs of the cofactor SAM are nonspecific, as SAM is also a cofactor for DNMTs and other protein methyltransferases. Encouragingly, a few specific inhibitors have recently been discovered including DZNep, which has both anti-HIV and anti-breast cancer activity and inhibits trimethylation of H3K27 and H4K20, with little effect on normal cells. The only arginine methyltransferase inhibitor with therapeutic value to date is ellagic acid, a specific inhibitor of CARM1 (methylates H3R17) that acts through an unknown mechanism and has anti-breast and -prostate cancer activity.
The increasing evidence for important functions of lysine/arginine methylation necessitates increased understanding of these enzymes and their function.
Histone Acetyltransferase Inhibitors (HATi)
The development of specific HAT inhibitors is feasible due to a lack of homology in structure and sequence of HAT enzymes.21, 125, 139 Efforts to develop specific HAT inhibitors are ongoing, as some of the natural products including curcumin, anacardic acid, and garcinol are only moderately potent and selective. Peptide-CoA conjugates like Lys-CoA are effective HAT inhibitors but lack cell permeability. Analogs of anacardic acid have demonstrated inhibitory activity and cytotoxicity toward a variety of cancer cells but not to nonmalignant human cell lines. Curcumin, a major constituent of turmeric, is a specific inhibitor of the HAT p300/CBP and has been used to reverse cardiac hypertrophy and heart failure. CTH7A, a water-soluble derivative of curcumin, can reduce histone acetylation and tumor size in mice. Other nontoxic derivatives of the natural HATi have been synthesized and show promise.
A variety of strategies are currently being used to define genome-wide epigenetic changes in specific cancers. “Cancer signatures” or epigenotypes that characterize the transformed phenotype have been identified and such studies are also transforming our knowledge of disease-specific epigenetic modifications that underline tumorigenesis. These could provide new biomarkers for the disease and potentially have prognostic value. A recent study showed that a potent small inhibitor of Dot1L (the H3K79 methylase) was found to selectively kill cancer cells that contained the MLL mutation that causes mixed lineage leukemia, demonstrating how targeting a specific histone modification could provide a highly specific therapy.
The studies performed in cancer demonstrating that alterations to chromatin are central to the reprogramming of pathological gene expression are directly relevant to many inflammatory diseases, including IBD. Definition of the complex and diverse epigenetic profiles that underlie phenotypic plasticity would be a crucially important starting point in delineating the mechanisms of chromatin dysregulation that lead to disease. The demonstration that such changes underlie chronic inflammation would provide a quantum leap in our understanding of IBD pathogenesis. Based on such knowledge, novel and highly specific therapeutic agents that have eluded clinicians and scientists for so long could at last be developed.
Genetic material located in the cell nucleus containing DNA wrapped around histone proteins along with other proteins that help package DNA into the nucleus. Chromatin condenses to form chromosomes in a eukaryotic cell.
The basic repeating unit of chromatin that consists of 146 basepairs of DNA wound around an octamer of histone proteins. Nucleosomes are found in eukaryotic nuclei and appear as bead-like structures on DNA referred to as “beads on a string” when viewed by electron microscopy.
A structural form of chromatin that can be open (euchromatin) or closed (heterochromatin) associated with activation or repression of transcription respectively. Multiple, complex variations of chromatin structure occur that are potential determinants of different biological outcomes.
The portion of the genome that is highly condensed or closed chromatin in which the nucleosomes are tightly packed together and is enriched for specific histone marks such as histone 3 K9 trimethylation and H4K20 trimethylation.
Decondensed or open chromatin with widely spaced nucleosomes enriched in active genes and associated with specific histone marks associated with active gene transcription, e.g., histone 3 lysine 4 trimethylation and histone 3 acetylation.
Enzymatic modification of the N-terminal region of the core histones H2A, H2B, H3, and H4 by the addition of a variety of chemical groups, e.g., acetyl groups (acetylation). The extent and pattern of modification regulates the degree of chromatin condensation.
An ATP-dependent enzymatic process that alters histone-DNA interactions and regulates the position of nucleosomes within chromatin.
Histone Variant Exchange
The incorporation into chromatin and replacement of canonical histones by distinct types of histone proteins.
A sequence of 0.5–2 kb that is rich in cytosine-guanine dinucleotides (the p denotes the phosphodiester bond that connects the two bases). These sequences are located in the promoter regions upstream of a variety of genes including tissue-specific genes. In mammals, these are hypomethylated with respect to the rest of the genome.
Jumonji C Domain
A protein region or domain conserved between bacteria and humans that is present in histone demethylase enzymes that catalyse the removal of methyl groups from methylated lysine residues by a hydroxylation reaction.
ATPase Domain of Chromatin Remodeling Enzymes
The protein domain in chromatin remodeling enzymes that uses ATP to generate energy for translocation of the enzyme along DNA, leading to the movement or destabilization of nucleosomes.
A complex of transcription activators, coactivators, and RNA polymerase II that assemble on gene promoters and are required for the activation of gene transcription.
Transcription (Factor) Activator
A nuclear protein that contains domains through which it directly binds specific DNA sequences and interacts with coactivators to activate gene transcription.
Transcription (Factor) Repressor
A nuclear protein that contains domains through which it directly binds specific DNA sequences and interacts with corepressors to repress gene transcription.
Proteins that associate with DNA promoters but do not bind DNA directly and are present in large complexes. Some have enzymatic activity and can add or remove posttranslational modifications to histones. They function in large complexes with transcription activators/repressor and coactivators/repressors resulting in downstream effector mechanisms for gene regulation.
Proteins that associate with DNA promoters but do not bind DNA directly and are present in large complexes. Some have enzymatic activity and can add posttranslational modifications to histones. They couple transcription activators/repressors resulting in downstream effector mechanisms for gene activation.
Proteins that associate with DNA promoters or enhancers but do not bind DNA directly. Some have enzymatic activity and can remove posttranslational modifications on histones. They function in large complexes with other transcription proteins and couple transcription activators/repressors resulting in downstream effector mechanisms for gene repression.
A group of proteins required to silence genes encoding regulators of differentiation, development, and the immune response.
A domain named after 3 Drosophila genes: Su(var)3-9, Enhancer of zeste, and Trithorax found within a class of histone methyltransferase that catalyze.