Transcriptional regulation by methyltransferases and their role in the heart: highlighting novel emerging functionality

Abstract

Methyltransferases are a superfamily of enzymes that transfer methyl groups to proteins, nucleic acids, and small molecules. Traditionally, these enzymes have been shown to carry out a specific modification (mono-, di-, or trimethylation) on a single, or limited number of, amino acid(s). The largest subgroup of this family, protein methyltransferases, target arginine and lysine side chains of histone molecules to regulate gene expression. Although there is a large number of functional studies that have been performed on individual methyltransferases describing their methylation targets and effects on biological processes, no analyses exist describing the spatial distribution across tissues or their differential expression in the diseased heart. For this review, we performed tissue profiling in protein databases of 199 confirmed or putative methyltransferases to demonstrate the unique tissue-specific expression of these individual proteins. In addition, we examined transcript data sets from human heart failure patients and murine models of heart disease to identify 40 methyltransferases in humans and 15 in mice, which are differentially regulated in the heart, although many have never been functionally interrogated. Lastly, we focused our analysis on the largest subgroup, that of protein methyltransferases, and present a newly emerging phenomenon in which 16 of these enzymes have been shown to play dual roles in regulating transcription by maintaining the ability to both activate and repress transcription through methyltransferase-dependent or -independent mechanisms. Overall, this review highlights a novel paradigm shift in our understanding of the function of histone methyltransferases and correlates their expression in heart disease.

INTRODUCTION

Methyltransferases (MTs) are a large group of enzymes that catalyze the transfer of a methyl group from a donor molecule, most commonly S-adenosylmethionine (AdoMet), to a specific methyl acceptor. These enzymes (>90%) can be grouped into three major classes based on their structural features, which are also reflected by unique amino acid sequences. These classes have enabled the identification of previously unrecognized enzymes: class I enzymes contain a seven-strand twisted β-sheet structure (25), class II contain a SET-catalytic domain (199), and class III is exemplified by membrane-associated enzymes with multiple membrane-spanning regions (157). In addition, these MTs can also be classified on the basis of the target substrates they methylate: specifically, natural product methyltransferases, DNA/RNA methyltransferases, and protein methyltransferases. The first comprehensive compilation of the human methyltransferome was accomplished by Petrossian and Clarke (149), who employed bioinformatic approaches to interrogate the genome, ultimately categorizing 208 proteins as characterized or putative methyltransferases, including 38 previously unannotated proteins. We have updated this list of methyltransferases in this review by removing seven proteins that are no longer included in the Uniprot database (likely due to redundancy or lack of supporting data) and two additional proteins that have duplicate entries, leaving us with 199 total human methyltransferases. This included 86 protein MTs, 58 DNA/RNA MTs, 19 natural product MTs, and 36 with unknown methylation targets (Fig. 1). We have used this categorization as a baseline for this review, with one exception: in the 9 years since their analyses were published, 40 additional protein methyltransferases have been functionally characterized and, therefore, migrate from the “putative” to “characterized” category. Although there are many functional studies, which have been carried out on individual methyltransferases describing their methylation of unique targets and the subsequent effects on biological processes, no analyses exist describing the spatial distribution across tissues of this enzyme class or their differential expression in the diseased heart. For this review, we performed tissue profiling in publicly available protein databases of 199 confirmed or putative methyltransferases to demonstrate the ubiquitous or unique tissue-specific expression of these individual proteins. In addition, we examined transcript data sets from human heart failure patients and murine models of heart disease to identify methyltransferases, which have been shown to be differentially regulated in the heart. Lastly, we focused the remainder of the review on the largest functionally characterized subset: protein methyltransferases, the majority of which have reported roles in regulating gene expression. Interestingly, our analysis showed that 76 of the 86 protein methyltransferases have confirmed roles in methylating histone residues by targeting either lysine or arginine side chains, and present a newly emerging phenomenon in which 16 of these enzymes have been shown to play dual roles in regulating transcription by maintaining the ability to both activate and repress transcription.

Fig. 1.

Human methyltransferome. A: classification of all characterized (green) and putative (blue) methyltransferases in humans based on functional domains and bioinformatics predictions from a 2011 study by Petrussian and Clarke, which we have modified to reflect newly characterized protein methyltransferases (blue-green hash) over the last 9 years. B: a total of 86 protein methyltransferases have been hypothesized to exist in humans, 36 of these had been functionally validated in 2011, and an additional 40 protein methyltransferases have been functionally validated since then, bringing the total number of those characterized to 76 (88%). MT, methyltransferase.

EXPRESSION OF METHYLTRANSFERASES IN HUMAN TISSUE AND CELLS

To determine the extent of methyltransferase expression across a wide variety of mammalian tissues, we submitted our list of 199 methyltransferases [adapted from Petrossian et al. (149), as detailed above] to the Human Proteome Map database (92), which contains mass spectrometry-based (LC-MS/MS) protein expression data from 17 adult tissues, 6 primary hematopoietic cells, and 7 fetal tissues. We generated expression profiles for both characterized and putative subsets, shown in Fig. 2. While the Human Proteome Map contains direct evidence of translation for over 17,000 genes, covering >84% of the annotated protein-coding genes, it is important to recognize that because of the nature of biological mass spectrometry, low abundant proteins may escape detection in this data set. Additionally, as provided by the Human Proteome Map database, all samples used to generate these data were obtained from histologically normal samples, so the information provided in this data set is limited only to the basal human physiological state of tissue or cells. Interestingly, our analyses show a wide variety of expression patterns among methyltransferases with some being ubiquitously expressed across tissues, while others display a more restricted expression pattern found in only one or a few tissue types. Indeed, the restricted expression of most methyltransferases was somewhat surprising to us, because of their rather generic role in methylating proteins, DNA, and RNA found in virtually all cells; however, we expect that as we continue to learn about these proteins, the specific molecular mechanisms regulating their specificity in different cells will become more apparent. Of note, the following methyltransferases are absent from Fig. 2 because they are either not present in the Human Proteome Map database but are still maintained in the Uniprot database (A: BMT2, CSKMT, EEFAKMT1, EEF1AKMT2, EEFAKMT3, EEF2KMT, ETFBKMT, TEDC2; B: BUD23, CARNMT1, METTL26, METTL27, MRM3, SPOUT1, TRMT9B), or, were not identified in any of the tissue or cells listed (A: ASMT, HEMK1, KMT2B, METTL11B, METTL22, PRDM12, PRDM16, PRDM7, PRDM9, SETD4, TRDMT1; B: BCD1N3D, ERC2-IT1, FAM86B1, FAM86B2, FBLL1, METTL15P1, METTL21P1, METTL23, METTL8, FTSJ2 [MRM2], NSUN3, NSUN5D1, NSUNP2, SETD9). Specifically, in cardiac tissue, we observe expression of 88 MTs (51 characterized, 37 putative) in fetal heart and 54 in adult heart tissue (35 characterized, 19 putative). Of these MTs expressed in fetal or adult heart tissue 42 (27 characterized, 15 putative) were expressed at both developmental stages while 55 (29 characterized, 26 putative) displayed expression exclusive to one stage. Interestingly, there were several methyltransferases that were uniquely expressed in only 1 tissue/cell, including PRDM8, which is only expressed in retina, and FAM173B, which is only expressed in platelets. In this data set, only one methyltransferase, MECOM, shows exclusive expression to the heart (at any stage), although data from the literature suggest this protein is also expressed in hematopoietic cells (60, 122). While the expression of these methyltransferases in various tissues can be helpful to guide characterization of their role in regulating cellular processes in fetal or adult tissue, it is limited only to basal states, and it does not include information about expression changes during various stages of development or disease.

Fig. 2.

Expression of methyltransferases in human tissue and cells. The relative expression profiles of characterized (A) and putative (B) methyltransferases in human tissue and cells. Protein expression profiles were generated from the Human Proteome Map (92), which maintains mass spectrometry-based identification and normalized spectral counting-based quantification of proteins (listed under their corresponding gene identifier). Red indicates high expression, pink indicates low expression, white indicates no detectable expression. Green denotes fetal tissue, yellow denotes adult tissue, and blue denotes hematopoietic cells.

METHYLTRANSFERASES DIFFERENTIALLY EXPRESSED IN THE DISEASED HEART

To identify methyltransferases that are potentially involved in the pathophysiology of cardiac disease, regardless of whether their methylation target or specific role has been characterized, we analyzed publicly available data sets from five studies (three using human tissue and two using mouse tissue), which quantified changes in transcript abundance from cardiac tissue via RNA-sequencing to identify transcripts that are differentially expressed in human heart failure patients or mouse models of cardiac disease (Tables 1 and 2). This is not a comprehensive analysis of all data publicly available in humans and mice from individual studies but represents the most recent analysis of human and animal models in which large-scale global analysis was performed via RNA-sequencing, where the entire data set was available for us to examine. Information regarding the number of samples analyzed, sex, and region of the heart used for analysis from these studies can be found in Table 3. Each study applied different filters for determining significance within their data set, which are briefly described below. Ultimately, we included transcripts in this review that were found to be significant on the basis of the criteria of the individual publication (all of which included a P value ≤ 0.05). It is important to note that these data sets are a comparison of diseased to normal tissue where transcripts found to be differentially regulated during disease were included (those which were detected but showed no change in abundance were excluded in the original publications). Currently, there are no large-scale analyses performed at the protein level to quantify changes in methyltransferases in these same disease settings; therefore, protein data sets are not included in this analyses, but analysis of these data sets would be helpful to carry out in future investigations.

Table 1.

Transcripts of characterized or putative methyltransferases differentially expressed in humans with ischemic cardiomyopathy, nonischemic cardiomyopathy, or dilated cardiomyopathy

	Yang et al. (197)		Sweet et al. (172)		Liu et al. (118)
Gene ID (Uniprot)	ICM	NICM	DCM	ICM	DCM	ICM
SMYD1	−1.7	−1.8	−1.6	−1.41
SMYD2	1.3
SMYD5	1.2
PRMT2	1.6	1.4
EZH1	1.3
SETD7		1.2			2.8	2.2
EHMT2	1.6	1.3
PRMT7			1.3	1.3
DOT1L					−3.0	−3.8
SETD1B					2.7	−2.3
HNMT	1.9	1.8
RNMT	1.3	1.4	1.5	1.6
COMT	1.5	1.4
COQ3		−1.3	−1.2	−1.3
PNMT		−2.3			−4.8
LCMT1	1.2
ICMT			−1.3	−1.3
INMT			2.0	2.2	2.3
SETD3			−1.4	−1.5
AS3MT						−13.3
GNMT					−17.0	−26.8
NNMT					−3.8
ASH1L					3.0
METTL21A			−1.6	−1.7
PCMTD1			−1.4	−1.7	2.7
FAM173B		1.3
METTL5			−1.3	−1.4
NSUN6P2			1.6	1.8
FTSJ3	1.2		−1.2	−1.2
TRMT61B		−1.7	−1.5	−1.5
ASMTL			1.4	1.3
TRMT1L			−1.6	−2.3
RRNAD1			1.3	1.6
FAM173A			1.5	2.1
EMG1			1.2	1.4
COMTD1	−1.7	−1.7
METTL9	1.2
ELP3	1.2
METTL23	1.4
METTL7B	−2.6	−2.5

Values indicate fold change for transcripts with a P ≤ 0.05. Negative value corresponds to downregulated transcripts and positive value to upregulated transcripts. Values without boldface indicate characterized methyltransferase transcripts, whereas boldface values indicate putative methyltransferase transcripts. DCM, dilated cardiomyopathy; ICM, ischemic cardiomyopathy; NICM, nonischemic cardiomyopathy.

Table 2.

Transcripts of characterized or putative methyltransferases differentially expressed in mouse models of cardiac disease

	Lee et al. (105)	Wang et al. (189)
Gene ID (Uniprot)	TAC	ISO
DOT1L		−1.2
SETMAR		−1.2
SETD6	−1.5
PRMT2	1.7
HNMT	1.7
COQ3	−1.5
INMT	−4.4
AS3MT	−1.7
NNMT	2.7	1.3
METTL1		−1.2
COMTD1	1.5
NSUN5		−1.2
RSAD1		−1.2
NRM	1.6
PCMTD2	−1.5

Values indicate fold change for transcripts with a P ≤ 0.05. Negative value corresponds to downregulated transcripts and positive value to upregulated transcripts. Values without boldface denote transcripts of characterized methyltransferases, where boldface values denote transcripts of putative methyltransferases. ISO, isoproterenol infusion via mini-osmotic pump; TAC, transverse aortic constriction.

Table 3.

List of five transcriptome studies analyzed in this work with additional information on the number of samples analyzed, species, sex, and indication of what region of the heart was used for analysis

Publication	n Value	Sex	Species	Region of Heart Analyzed	Notes
Yang et al. (197)	Ischemic: n = 8 Nonischemic: n = 8 Nonfailing: n = 8	Ischemic: 8 M Nonischemic: 3 F, 5 M Nonfailing: 1 F, 7 M	Human	left ventricle
Sweet et al. (172)	Dilated: n = 37 Ischemic: n = 13 Nonfailing: n = 14	Dilated: 7 F, 30 M Ischemic: 3 F, 10 M Nonfailing: 3 F, 9 M	Human	left ventricle	Ischemic cohort had significantly greater proportion of patients taking statins, having coronary artery disease, hyperlipidemia, and diabetes mellitus.
Liu et al. (118)	Dilated: n = 2 Ischemic: n = 1 Nonfailing: n = 3	n/a	Human	left ventricle
Lee et al. (105)	1 wk TAC: n = 6 1 wk Sham: n = 6 8 wk TAC: n = 4 8 wk Sham: n = 4	M	Mouse	left ventricle
Wang et al. (189)	104 ISO-treated strains, ~4 mice from each strain	F	Mouse	left ventricle	Fully controlled the age, environment, severity, and timing of cardiac injury

F, female; ISO, isoproterenol infusion via mini-osmotic pump; M, male; n/a, not applicable; TAC, transverse aortic constriction.

The first human study that we analyzed by Yang et al. (197) quantified transcripts in cardiac tissue from patients with either ischemic or nonischemic cardiomyopathy undergoing LVAD implantation and compared these to healthy donor tissue. In this study, 21 methyltransferases (14 characterized and 7 putative) were differentially expressed and considered significant on the basis of a fold change greater than 1.2 and P value < 0.5. The second study, by Sweet et al. (172) analyzed 64 human left ventricle tissue samples obtained at the time of heart transplantation that were categorized as dilated cardiomyopathy (37), ischemic (13), and nonfailing (14) hearts (172). Among the 2,993 differentially expressed transcripts, 18 were methyltransferases (9 characterized, 9 putative), which displayed significant changes in expression. The third human study from Liu et al. (118) reported gene expression signatures in heart tissue obtained from one patient with ischemia, two patients with diagnosed dilated cardiomyopathy, and three healthy controls. In their analysis of around 2,000 differentially regulated transcripts for both disease conditions, 10 were methyltransferases (all characterized).

In mouse models of cardiac disease, the first report that we analyzed by Lee et al. (105) examined transcripts from cardiac tissue in mice after transverse aortic constriction (TAC)-induced heart failure. Their data set identified a total of 1,478 differentially expressed transcripts in the diseased heart, 10 of which were methyltransferases (seven characterized and three putative). The second study by Wang et al. (190) used mini-osmotic pumps to infuse isoproterenol (ISO) in 104 inbred mouse strains over 21 days, to induce cardiac dysfunction and then quantified changes in transcripts. Their analysis identified 1,502 differentially expressed transcripts across all strains examined and included 6 methyltransferases (4 characterized, 2 putative).

Together, these five studies identified 40 methyltransferases (26 characterized, 14 putative) that are differentially expressed in human heart failure patients and 15 mouse-models of cardiac disease (10 characterized, 5 putative). Of the transcripts found to be differentially expressed, only eight of them were identified in both human patients and mouse models, five that display the same directional change in expression (DOT1L, PRMT2, HNMT, COQ3, and AS3MT) and three that display opposite changes in expression in different studies (NNMT, COMTD1, INMT). Of the methyltransferases that were identified collectively in these five studies, 12 of them have previously been examined in the heart and were shown to play a specific role in cardiac physiology and pathology [SMYD1 (181), SMYD2 (181), SMYD3 (181), EZH1 (5), SETD7 (49), EHMT2 (146) DOT1L (34, 138, 139), AS3MT (30, 48, 70), NNMT (116), COMT (74), PNMT (179), and LCMT1 (120)]. However, a majority of them (35) have never been examined within the heart (PRMT2, PRMT7, SETD1B, HNMT, RNMT, COQ3, ICMT, INMT, SETD3, GNMT, ASH1L, METTL21A, PCMTD1, FAM173B, METTL5, NSUN6P2, FTSJ3, TRMT61B, ASMTL, TRMT1L, RRNAD1, FAM173A, EMG1, COMTD1, METTL9, ELP3, METTL23, METTL7B, SETMAR, SETD6, METTL1, NSUN5, RSAD1, NRM, and PCMTD2). Therefore, future studies will be needed to determine their function in the cardiovascular system and to fully understand the variability in abundance of these enzymes across different conditions and species examined in Tables 1 and 2.

HISTONE METHYLTRANSFERASES IN TRANSCRIPTIONAL REGULATION

We next became interested in examining the largest characterized subset of methyltransferases, that of protein methyltransferases, 88% (76 of 86) of which have established roles in regulating gene expression through their ability to methylate lysine or arginine side chains on histone proteins. This subset is, therefore, the focus of the remainder of this review.

Historically, methylation of lysine residues was first reported in 1959 by Ambler and Rees (8), who described differences in the amino acid composition of flagellar proteins in Proteus vulgaris and Salmonella typhimurium, and showed the latter contained a new amino acid, ε-N-methyl lysine. Several years later, this modification (ε-N-methyl lysine) was identified on histone proteins (135) and was characterized as a posttranslational modification (occurring subsequent to peptide bond formation) (93). Subsequently, two studies initiated an interest in histone methylation and gene regulation, as it is studied today. First, Hwang and Bonner (82) reported that adding histones to DNA-dependent RNA polymerase attenuates RNA synthesis, and second, Allfrey et al. (7) discovered that histone residues can be acetylated and methylated, influencing RNA synthesis. The discovery of the first methyltransferases (the enzymes responsible for adding this modification) came soon after, for both arginine (142) and lysine (143) methyltransferase classes. However, with these discoveries in hand, it took more than two decades to decode the applicability of these methylation reactions and to provide evidence that the methylation of histones, indeed, plays a significant role in the regulation of gene expression (144). In addition, the discovery of the human SUV39H1 histone lysine methyltransferase spurred a new era of methylation research and influenced the growth of epigenetics by providing the very first link between histone methylation and the formation of higher-order chromatin structure, such as heterochromatin formation (155). It is now appreciated that both arginine and lysine residues on histones are substrates for methylation and that their side chain length (the longest side chains of all naturally occurring amino acids) and positive charge make them an attractive target for protein-protein interactions (15). Methylation of these residues enables them to contribute to changes in higher-order chromatin structure by modifying the charge state of histones and/or their interaction with DNA by influencing interactions with other hydrogen bond donors, which, in turn, regulates accessibility for transcription factors and other transcriptional machinery (72, 125). Moreover, histone methylation can influence the recruitment of proteins that specifically recognize, or “read”, these posttranslational modifications and their associated binding partners or inhibit binding of proteins to chromatin. In addition to being methylated or unmethylated, an additional level of complexity arises from the ability of these amino acids to be monomethylated, demethylated, or trimethylated on lysine or monomethylated or dimethylated (symmetrically or asymmetrically) on arginine; however, how this specificity is acquired to enable an enzyme to selectively participate in monomethylation, demethylation, or trimethylation is unknown, but it is an important area of research that will be necessary for us to fully understand the enzymatic function of these proteins. For many years, histone methylation, as opposed to acetylation or phosphorylation, was thought to be irreversible and biochemically stable. However, it was not until the discovery of the first histone demethylase in 2004 that this notion was challenged, and ever since then, we have continued to learn more about how this dynamic process of histone methylation influences a wide array of biological processes (75, 159).

Lysine Methylation

Methylation of lysine residues is regulated by lysine-specific methyltransferases (KMTs) primarily in an AdoMet-dependent manner. Except DOT1L, all lysine-specific histone methyltransferases share a SET [Su(var), Enhancer of Zeste, Trithorax] domain that is responsible for catalysis of AdoMet onto the target lysine, resulting in monomethylation demethylation, or trimethylation (abbreviated as me1, me2, and me3, respectively) (144). The established paradigm has been that histone lysine methylation is associated with either activation or repression of gene expression, depending on the location and degree of methylation. For example, all three methylation states of histone H3K4 mark active genes, while trimethylation of histone H4K20, in general, designates silenced genes. However, for some residues, the resulting biological consequence is not always straightforward; for instance, monomethylation of some residues leads to transcriptional activation (e.g., histone H3K9me1), while trimethylation at the same site leads to gene silencing (e.g., histone H3K9me3) (13). What dictates whether a methyltransferase will carry out monomethylation, dimethylation, or trimethylation of a specific residue is unknown, but it has been hypothesized to be influenced by protein binding partners or other posttranslational modifications. For example, DOT1L is known to regulate monomethylation, demethylation, and trimethylation of H3K79, which have opposing transcriptional functions (activation vs. repression), and while it is not known what influences DOT1L to carry out monomethylation versus trimethylation, it is known that this occurs in a nonprocessive manner (requiring the enzyme to dissociate from the substrate between each addition of a methyl group) (193). Even though the addition of a methyl group does not change the net charge of the protein, as it does with lysine acetylation, it does influence hydrophobicity, which, in turn, affects protein-protein interactions, including that of transcriptional repressors and activators. For instance, trimethylation of histone H3K4 recruits the chromatin remodeling proteins CHD1 and BPTF (63, 110), which opens the chromatin structure and prevents the binding of the repressive NuRD and INHAT complexes (140, 163). The most well-known sites of histone methylation occur on histone H3 (on lysine residues 4, 9, 14, 27, 36, and 79) and H4 (on lysine residues 20 and 59). However, there are also many other lysine residues on histone H1, H2A, and H2B that have been identified (79).

Apart from histones, lysine methylation also occurs on a wide variety of proteins involved in regulating transcription and chromatin structure, as well as other biological processes. Specifically, the methylation of nonhistone proteins is known to regulate protein-protein interactions, subcellular localization (e.g., nuclear localization of Hsp70), the promoter binding affinity of transcription factors at target genes (e.g., NF-κB activation by lysine methylation), various signaling pathways (NF-κB, MAPK, Wnt, ER, and p53) and DNA damage and stress response, among others, to influence the development and differentiation in various cell types (131). We will not discuss in detail the methylation of nonhistone proteins in this review, except when it is applicable to the regulation of gene expression by the methyltransferases described below.

Arginine Methylation

Arginine methylation is becoming increasingly more important in our understanding of the dynamic regulation of gene expression. Since the discovery of the first arginine methyltransferase decades ago, this family has expanded to include 11 members with known roles in various biological processes, including signaling, RNA splicing, DNA damage repair, and transcriptional regulation (90). Indeed, it has been estimated that roughly 0.5% of all arginine residues in mammalian cells are methylated, although a large majority of these are thought to occur on nonhistone proteins, in addition to the significant number of sites on histones, where arginine residues are known to be methylated (69).

Arginine methylation is carried out by protein arginine methyltransferases (PRMTs) that consist of two conserved domains (AdoMet binding- and COOH-terminal domains), which modify arginine residues in an AdoMet-depending manner. This results in three types of modifications (distinct from those generated by lysine methyltransferases). Specifically, monomethylation occurs on the terminal N atom that can only be carried out by PRMT7. The remaining family members can catalyze subsequent methylation reactions on either the same N atom to generate symmetric (s) dimethylation (via PRMT1, PRMT2, PRMT3, PRMT4 [also known as CARM1], PRMT6 and PRMT8) or on the adjacent N atom to generate asymmetric (a) dimethylation (via PRMT5, PRMT7, and PRMT9) (125). Ultimately, these methylation marks either activate or repress transcription directly or indirectly by recruiting other chromatin binding proteins or transcriptional regulators. In general, methylation of histone H3 and H4 on H4R3me1, H4R3me2a, H3R2me2s, H3R17me2a, or H3R26me2a results in transcriptional activation, while methylation on H3R2me2a, H3R8me2a, H3R8me2s, H3R38me2a, or H4R3me2s leads to transcriptional repression. For instance, PRMT1 has been shown to methylate histone H4R3me2a, which results in recruitment of the CBP/p300 complex, which, in turn, enhances acetylation of histone H3K9 and H3K14 and promotes binding of transcriptional machinery (14). On the other hand, histone H4R3me2s methylation by PRMT5 recruits DNMT3a to chromatin, which results in DNA methylation and gene silencing (211). Additional examples of arginine methylation regulating gene expression include the recruitment of PRMTs to the promoters of various transcription factors, including p53 (9), YY1 (156), and NF-κB (47), and their methylation of transcriptional coactivators (including CBP/p300 (43), RUNX1 (212), CERC (56), HIV Tat (27)), transcription elongation factors [SPT5 (100)], and chromatin structural proteins [HMGA (166)]. Similar to lysine methyltransferases, arginine methyltransferases have also been shown to methylate a wide variety of proteins, which are not involved in transcriptional regulation. Most prominently, they are known to methylate RNA binding proteins to regulate processing, folding, stabilization, localization, and translation of RNAs (40) and to regulate DNA repair (54), signal transduction (16, 102), or nuclear/cytoplasmic shuttling (168), although this review will focus only on their ability to regulate transcription.

The role of protein arginine methyltransferases in stem cell biology (23, 24, 208), development (76), and adult homeostasis (15, 98) is unquestionable; however, their role in human disease has only more recently emerged. Implications of arginine methylation to disease is most apparent in cancer, where it has been shown that PRMTs are usually upregulated, making them an attractive target for the development of small-molecule inhibitors (80). In addition, evidence now exists, suggesting a role in other human conditions, including neurodegenerative diseases, metabolic diseases, and aging (23, 73).

EMERGING MULTIFUNCTIONALITY OF METHYLTRANSFERASES IN TRANSCRIPTIONAL REGULATION

The classical dogma regarding histone methylation and gene expression suggests that each methyltransferase modifies a single and specific (nonrandom) amino acid residue and, thereby, influences gene expression in one direction, by either activating or repressing transcription (Fig. 3A). In this way, these enzymes were thought to maintain functional specificity for regulating transcription within the genome. However, as our understanding of methyltransferases has expanded over the years, we have come to recognize that the function of so-called “histone” methyltransferases is much broader than this. Indeed, evidence now exists showing that the same methyltransferase can modify different amino acids on histones to both activate and repress transcription (based on the residue modified), even in the same cell type (Fig. 3B). In addition, these enzymes have also been shown to regulate transcription by methylating additional, nonhistone proteins, including transcription factors, transcriptional machinery, and chromatin remodeling proteins (which regulate chromatin structure and accessibility, Fig. 3C). Lastly, in addition to these methyltransferase-dependent mechanisms (Me), many of these enzymes have also been shown to regulate transcription in a methyltransferase-independent (Me-I) manner, in which they interact with other proteins, but do not methylate them (Fig. 3D) or through a separate and currently unknown mechanism (Uk). Therefore, to systematically review the ability of methyltransferases to function through these mechanisms, we compiled data for all 76 known histone methyltransferases and specifically documented the manner by which they regulate gene expression. Our analysis identified 16 methyltransferases that maintain the ability to both activate and repress transcription by participating in at least two of the four modes of regulation mentioned above and shown in Fig. 3. These 16 enzymes (SMYD1, SMYD2, SMYD3, EHMT1/EHMT2 (G9a/GLP), EZH2, KMT2D, DOT1L, CARM1, PRMT5, PRMT6, ASH1L, SETDB2, SETD3, SETD6, and SET8) and the specific mechanisms by which they activate and repress transcription are listed in Table 4. It is important to note that this table is not comprehensive in listing all known methylation targets for these enzymes, but instead focuses on those by which they have been shown to regulate gene expression. In addition, of the 16 methyltransferases, 10 have been shown to play significant roles in cardiac physiology and pathology, as described in Table 5. An additional 6 enzymes (PRMT6, ASH1L, SETDB2, SETD3, SETD6 and SET8) have never been examined in the heart; thus, detailed information about their transcriptional regulation is not included below; however, information about their histone methylation targets and unique nonhistone targets and mode of regulation is included in Table 4. Interestingly, two of these methyltransferases (ASH1L and SETD3) with unknown roles in the cardiovascular system are expressed in both the developing and adult heart (Fig. 2); therefore, further studies would be helpful to elucidate their roles in the heart physiology and pathology.

Fig. 3.

Schematic diagram depicting mechanisms by which methyltransferases regulate gene expression. A: classically, histone methyltransferases (MT; yellow) were thought to only modify a single and specific amino acid (pink circle) and, thereby, maintain specificity for regulating gene expression in one direction within the genome. B: however, more recently, studies have shown that the same methyltransferase can modify multiple amino acids (purple circle) on histones to both activate and repress transcription, even within the same cell. In addition, the same methyltransferase can also regulate gene expression by methylating other chromatin binding proteins (C) or through methyltransferase-independent mechanisms (D) (typically as a component of a chromatin binding complex). MT, methyltransferase.

Table 4.

Methyltransferases capable of both activating and repressing transcription through methyltransferase-dependent (methylating histones or other proteins), methyltransferase-independent (binding but not methylating proteins), and/or uncharacterized mechanisms

Gene ID (Uniprot)	Histone Residue Methylated	Activation/Repression	Mechanism of Regulation: Methylation (Me), Methylation-Independent (Me-I), Unknown (Uk)	Activation/Repression
SMYD1	H3K4me3 (19, 175, 192)	Activation (19, 175)	Nppa (Uk) (65), Tgfβ3 (Uk) (65)	Repression (65)
			skNAC (Me) (19, 147, 213), TRB3 (Me) (154)	Activation (19, 154)
SMYD2	H3K4me3 (3)	Activation (3)	RB (Me) (160), p53 (Me) (81)	Repression (81, 160)
	H3K36me2 (31)	Repression (31)	PARP1 (Me) (150), HSP90 (Me) (52)	Activation (52, 150)
SMYD3	H3K4me2/3 (75)	Activation (75, 183)	MAP3K2 (Me) (124)	Activation (124)
	H4K20me3 (64, 186)	Repression (64, 186)
EHMT1/EHMT2 (G9a/GLP)	H3K9me1/2 (113, 174)	Both (108, 113, 174)	G9a-p53 (Me-I) (152), G9a-CARM1-p300 (Me-I) (22)	Activation (22, 152)
	H3K27me1/2 (173)	Both (173)	Jarid1a (Me) (36)	Repression (36)
EZH2	H3K27me1 (61)	Activation (61)	GATA4 (Me) (78)	Repression (78)
	H3K27me2/3 (61, 201, 210)	Repression (61, 201, 210)	Jarid2 (Me) (161), RelA/RelB (Me-I) (106), SWI/SNF (Me-I) (91)	Activation (91, 106, 161)
KMT2D	H3K4me1 (41, 66, 88, 104)	Both (41, 66, 88, 104)	CBP/p300 (Uk) (101), ERα (Me-I)	Activation (101)
	H3K4me2/3 (66, 104)	Activation (66)
DOT1L	H3K79me1/2 (193)	Activation (193)	ENL/MLLT1, AF4/MLLT2, AF9/MLLT3 and AF10/MLLT10 (Me-I) (21, 130, 133), RNAPII (Me-I) (94), c-Myc (Me-I) (44)	Activation (21, 44, 94, 148)
	H3K79me3 (193)	Repression (193)	Sir3 (Me-I) (62)	Repression (62)
CARM1	H3R17me2a (69, 164, 194) H3R26me2a (69, 164) H3R42me2a (33, 69)	Activation (33, 69, 164, 194)	Sox2 (Me) (209), CBP/p300 (Me) (136) SRC-3 (Me) (59), GRIP1/p300 (Me) (107)	Activation (136, 209) Repression (59, 107)
PRMT5	H3R2me2s (127)	Activation (10, 15, 98, 127)	hnRNP A1 (Me) (68)	Activation (68)
	H2AR3me2s (10, 15, 98, 176) H3R8me2s (145, 190, 204) H4R3me2s (190, 204)	Repression (10, 15, 98, 176) Both (145, 190, 204)
PRMT6	H2AR11me2a (188) H2AR29me2a (188)	Repression (188)	HMGA1a (Me) (119, 128), DNA Pol b (Me) (54)	Activation (54, 119, 128)
	H3R2me2a (26, 126, 141)	Both (26, 126, 141)
	H3R42me2a (33)	Activation (33)
ASH1L	H3K4me3 (200) H3K36me2 (202)	Activation (200) Repression (202)	MMP (Me) (17)	Repression (17)
SETDB2	H3K9me3 (96, 99, 162, 180)	Repression (96, 99, 162, 180)	Insig2 (Me-I) (158)	Activation (158)
SETD3	H3K4me1/2 (55) H3K36m1/2 (55)	Activation (55)	Myod1 (Me) (55), p53 (Me) (1) FoxM1 (Me) (46)	Activation (1, 55) Repression (46)
SETD6	H2AZK7me1 (20)	Repression (20)	RelA (Me) (134), PLK-1 (Me) (58), DJ1 (Me-I) (37), PAK4 (Me) (185)	Repression (37, 58, 134) Activation (185)
SET8	H4K20me1 (87, 198)	Both (87, 198)	TWIST (Me) (196) p53 (Me) (169)	Both (196) Repression (169)

Currently, 16 of the 86 total protein methyltransferases have been shown to be capable of this dual functionality. Mechanism indicated in parenthesis: Me, methylation-dependent; Me-I, methylation-independent; Uk, unknown.

Table 5.

Methyltransferases with expanded functionality and their roles in the fetal and/or adult heart, as well as reported cardiac phenotype in zebrafish, mouse, or human

Methyltransferase	Role in Fetal Heart	Role in Adult Heart	Cardiac Phenotype of KD/KO
SMYD1	Morphogenesis (71, 154); Myofibrillogenesis (154, 175); Cell proliferation (154)	Inhibit hypertrophic growth (65)	Zebrafish: Impaired myofibril formation (175), no heartbeat (86). Mouse: Disrupted right ventricle formation and cardiomyocyte maturation (71, 154) Human: Hypertrophic cardiomyopathy (57)
SMYD2	Dispensable (51); Myofibrillogenesis (187)	No phenotype (51)	Zebrafish: Impaired myofibril formation (187) Mouse: Normal heart development (51)
SMYD3	Morphogenesis, maturation/proliferation of differentiated myocytes (67)	Unknown	Zebrafish: Abnormal looping of heart tube, pericardial edema (67)
EHMT1/EHMT2 (G9a/GLP)	Morphogenesis (83)	Inhibit hypertrophic growth, regulates cardiac homeostasis (146)	Mouse: Neonatal lethality and severe cardiac defects (atrioventricular septal defects) (83, 146) Human: Kleefstra syndrome (97)
EZH2	Cell proliferation (38); Apoptosis (77)	Maintain cardiac homeostasis (50); Dispensable for cardiac regeneration (4)	Mouse: Failure of myocardial compaction, hypertrabeculation, and ventricular and atrial septal defects (38)
KMT2D	Cell differentiation, embryonic development, calcium handling/ion homeostasis (11)	Unknown	Zebrafish: Abnormal development of the atria and/or ventricle, prominent bulging of the myocardial wall (184) Mouse: Embryonic lethal, disorganized interventricular septum (11) Human: Kabuki syndrome (165, 191), congenital heart defects (203)
DOT1L	Cell differentiation (34)	Regulate growth/shape, maintains normal cardiac function (138)	Mouse: Dilated cardiomyopathy, conduction abnormalities, increased mortality (138, 139)
CARM1	Unknown	Regulates inflammation (117); Cardiac autophagy (109)	Unknown
PRMT5	Dispensable for cell growth (207)	Regulate hypertrophic growth via GATA4 (39); Muscle stem cell proliferation during regeneration/repair (207)	Unknown
PRMT6	Unknown	Unknown	Unknown
ASH1L	Unknown	Unknown	Unknown
SETDB2	Unknown	Unknown	Unknown
SETD3	Unknown	Unknown	Unknown
SETD6	Unknown	Unknown	Unknown
SET8	Unknown	Unknown	Unknown

KD/KD, knockdown or knockout.

The SMYD family of methyltransferases consists of five members that share a similar catalytic SET domain, which is split by a MYND domain thought to be involved in protein binding. SMYD1, a myocyte-specific lysine methyltransferase, was originally shown to methylate histone H3 at K4 (a known mark of gene activation), which was also thought to be the hallmark function of this family. However, since then, it has been demonstrated that SMYD enzymes can methylate several other unique histone residues and nonhistone proteins, with the ability to activate or repress transcription at different genomic loci. Specifically, SMYD1 can induce transcriptional activation of PGC1α through histone H3K4me3 (192), SMYD2 influences transcription of the Tacc2 gene by binding to its promoter (3), and SMYD3’s ability to dimethylate and trimethylate histone H3K4 results in recruitment of histone acetyltransferases and Sp1 to the promoter region of the androgen receptor (114). In contrast, these enzymes have been shown to repress transcription through histone and/or nonhistone methylation or methylation-independent mechanisms: SMYD1 has been shown to be a histone deacetylase-dependent transcriptional repressor (71) and can repress transcription of nppa and tgfβ3 in cardiomyocytes (65). SMYD2 can dimethylate histone H3K36 in vitro and repress transcription through interaction with the Sin3A histone deacetylase complex (31) and SMYD3 plays an important role in negative transcriptional regulation by trimethylating histone H4K20 (64). Regarding their methylation of nonhistone proteins, SMYD1 can regulate transcription by methylating skNAC (19) or TRB3 (154). SMYD2 methylates retinoblastoma in response to DNA damage (160) and p53 (81), both resulting in transcriptional repression, or SMYD2 methylates PARP1 in cancer cells (150) and Hsp90 in muscle cells to activate transcription (52, 187). SMYD3 plays a critical role in various forms of cancer by methylating MAP3K2, which results in activation of the Ras/Raf/MEK/ERK signaling pathway and tumorigenesis (124).

In the heart, the SMYD family has been shown to play key roles in the regulation of growth and differentiation (181). SMYD1, SMYD2, and SMYD3 have been shown to be involved in morphogenesis (67, 71, 154), SMYD1 and SMYD2 in myofibrillogenesis (86, 111, 187), and SMYD3 in the maturation of myocytes (67). Specifically, cardiac-specific deletion of SMYD1 in the mouse heart removes repression of developmental genes that leads to transcriptional remodeling, hypertrophy growth, and fulminate heart failure (65). The repressive transcriptional role of SMYD1 was examined by ChIP and luciferase reporter assays, where SMYD1 was shown to bind directly to the promoters of tgfb3 and nppa genes and inhibit their expression, independent of its methyltransferase activity toward histone H3K4me3, emphasizing an unknown mechanism for this transcriptional regulation (65). SMYD2 was shown to be a key regulator of myofibrillogenesis and sarcomere organization in zebrafish although knockout in the mouse heart gave no overt phenotype (51, 52). SMYD3 is necessary for the development of cardiac and skeletal muscle in zebrafish (67); however, the specific mechanism for its functionality in the heart was never reported.

G9a (also known as EHMT2 or KMT1C) is responsible for the majority of monomethylation and dimethylation of histone H3 at K9 in mammalian cells, and it has also been shown to methylate histone H3 on K27 (173). This methyltransferase uses its catalytic activity to efficiently repress developmentally regulated genes during embryonic stem cell differentiation (174), in addition to regulating T-cell differentiation (108) and muscle differentiation (113).

G9a and its coregulatory G9a-like protein/GLP (also known as EHMT1 or KMT1D) regulate transcription as either a homomeric or heteromeric complex; however, the heteromeric complex seems to be the functional form in vivo due to its increased stability (113). In addition, while G9a and GLP can both independently methylate histones and possess the same substrate specificity in vitro, when individually knocked down they cannot compensate for the levels of histone H3K9me2, further suggesting that they primarily function as a complex (113, 170). The G9a/GLP heterocomplex can also regulate transcription by partnering with cofactors, including the histone demethylase Jarid1a to repress gene expression during erythroid cell differentiation in a methyltransferase-dependent manner (36). Additionally, recent studies have demonstrated that the G9a/GLP heterocomplex can also regulate transcription through methyltransferase-independent mechanisms by either recruiting CARM1 and p300 to glucocorticoid receptor genes to activate transcription (22) or by interacting with p53 and influencing transcriptional activation in the absence of p53 methylation; however, the specific mechanism has not yet been established (152).

In the heart, in the conditional, cardiac-specific G9a knockout mice, G9a was shown to be an important regulator of cardiomyocyte homeostasis by mediating the repression of genes regulating cardiomyocyte function, including antihypertrophic genes through its methyltransferase activity on histone H3K9 and interaction with EZH2 and also by forming a complex with the transcription factor MEF2C (146). Additionally, Thienpont et al. (177) also demonstrated that loss of EHMT1 and EHMT2 enzymatic activity is sufficient to induce pathological cardiac hypertrophy in the adult cardiomyocyte and that activation of this pathway can protect against pathological hypertrophy in vivo and in vitro.

EZH2 (Enhancer of Zeste 2) is one of the most studied lysine methyltransferases. It belongs to the Polycomb group family, functions as an active catalytic subunit of the Polycomb Repressive Complex 2 (PRC2), and is responsible for the repression of many genes involved in development and cell differentiation (28, 29). EZH2 contains a SET domain that targets histone H3 K27. Initially, it was demonstrated that EZH2 plays a critical role in silencing tumor suppressor genes due to its trimethylation of histone H3K27 (123, 201). However, recently Ferrari et al. (61) reported that EZH2 and PRC2 complex can control all three stages of histone H3K27 methylation, and while dimethylation and trimethylation is associated with gene repression, H3K27 monomethylation was found to be enriched at the transcriptional start site of active genes. In addition to EZH2’s catalytic activity, there have been reports suggesting new noncanonical roles of this enzyme. Specifically, EZH2 has been shown to methylate GATA4, which lead to its transcriptional deactivation (78) or Jarid2, a known (1) cofactor of PRC2 that regulates chromatin structure by regulating H3K27me3 during cell differentiation (161). Moreover, several studies have reported that EZH2 can act independently of the PRC2 complex in a methyltransferase-independent manner. For example, EZH2 has been shown to activate NF-κB target gene expression in ER-negative basal-like breast cancer cells by interacting with RelA and RelB (106). Lastly, Kim et al. (91) demonstrated that when EZH2 interacts with the SWI-SNF complex, it can activate target genes via a PRC2-independent mechanism and can activate androgen receptor gene transcription by binding directly to its promoter (89).

Ablation of EZH2 in the fetal heart leads to decreased cardiomyocyte proliferation and increased apoptosis, and enzymatic inactivation of EZH2 in the fetal heart results in lethal congenital malformations or perinatal death (38, 77). However, in the neonatal heart, it has been shown to be dispensable for heart regeneration (4).

KMT2D (also known as MLL4 and MLL2 in humans) belongs to a family of histone lysine methyltransferases. It is ubiquitously expressed in adult tissues and plays an important role in embryonic development, cellular differentiation, metabolism, and tumor suppression. Additionally, it was implicated in human congenital heart disease and Kabuki syndrome (66). Molecularly it was first demonstrated that KMT2D plays an essential role in transcriptional activation due to histone H3K4 methylation; however, more recently, it has also been shown to negatively regulate gene expression. Specifically, KMT2D’s catalytic SET domain has been shown to methylate histone H3K4 generally at enhancer regions and maintain global levels of this methylation mark (104). More specifically, monomethylation of histone H3K4 by human KMT2D (MLL4) activates transcription of peroxisome proliferator-activated receptor γ in overnutrition-induced hepatic steatosis (88). KMT2D can also associate with WRAD, NCOA6, PTIP, PA1, and UTX demethylase in a multiprotein complex, where it is necessary to activate complex formation and stabilization of UTX during its histone H3K27 demethylation activity (45, 84). Additionally, KMT2D can interact with ERα in the presence of estrogen and regulate ERα transcriptional activity (129) or with CBP/p300 at enhancers during cell differentiation (101). Lastly, MLL4 in complex with MLL3 is known to monomethylate histone H3K4 at specific promoter regions, leading to conditional repression of muscle and inflammatory response genes in myoblasts (41).

It is well established that KMT2D is essential for cardiac cell differentiation and embryonic development (11). KMT2D is also implicated in cardiac disease, where it was first found to be differentially expressed in patients with congenital heart disease and later confirmed functionally using KM2TD knockout mouse that develop severe cardiac defects (11, 203). In addition, aberrant trimethylation by KMT2D can lead to Kabuki syndrome (137).

DOT1L (disruptor of telomeric silencing 1-like) is a unique lysine methyltransferase because instead of having a canonical SET catalytic domain, it contains a SET binding motif similar to arginine and DNA methyltransferases (193). Molecularly, DOT1L targets histone H3K79, which is located on the nucleosome surface, as opposed to the NH₂-terminal tail where most posttranslational modifications are deposited. It catalyzes the monomethylation, demethylation, and trimethylation of K79, but must dissociate and reassociate again with the residue after each methylation event to add additional methyl groups. Monomethylation and dimethylation of histone H3K79 is associated with open chromatin and transcriptional activation, while H3K79me3 is a mark of transcriptional repression (193). DOT1L plays a critical role in transcriptional elongation, cell cycle progression, development, somatic reprogramming, and DNA damage repair. DOT1L was also found to play a role in tumorigenesis, as it interacts with the mixed lineage leukemia (MLL) fusion proteins ENL/MLLT1, AF4/MLLT2, AF9/MLLT3, and AF10/MLLT10 and forms a complex that is necessary for histone H3K79 methylation, euchromatin formation, and expression of downstream oncogenes and leukemogenesis (21, 130, 133). Additionally, Cho et al. (44) reported that DOT1L binds to the oncogene c-Myc before epigenetic activation of the epithelial-mesenchymal transition (EMT) regulators in breast cancer progression. DOT1L can interact directly with the CTD domain of RNA Polymerase II and recruit this enzyme to the promoter regions of genes to be transcribed (94). DOT1L was also implicated in telomeric silencing, where methylation of histone H3K79 prevents binding of Sir3 (62).

In the heart, DOTL1 has been shown to play a critical role during cardiomyocyte differentiation (34, 151). Additionally, cardiac-specific knockout of DOT1L in mice leads to downregulation of histone H3K79, the development of dilated cardiomyopathy, conduction abnormalities, and increased mortality (138).

CARM1 (coactivator-associated arginine methyltransferase 1, also known as PRMT4 or Protein Arginine Methyltransferase 4) was the first PRMT that was functionally linked to regulation of gene expression (195) and has been shown to methylate several arginine residues on histone H3, which lead to transcriptional activation, including R2, R17, R26, and R42 (although H3R2Me2a has only been shown to occur in vitro) (69, 90, 121, 164). Specifically, PRMT4 was found to asymmetrically dimethylate R17 on histone H3, which leads to recruitment of the transcription elongation-associated PAF1 complex that, in turn, activates a subset of genes, including targets of estrogen signaling (194). CARM1 can also methylate arginine residues on histone H3 in the core of the structure, including asymmetrical dimethylation of R42, which signals transcriptional activation by disrupting histone:DNA interactions (33). In addition, CARM1 can methylate several nonhistone substrates, including transcription factors (40, 209), coregulators (59, 136), and RNA polymerase II (171), which result in either positive or negative transcriptional regulation. Specifically, CARM1 methylation of Sox2 leads to transactivation in ESC cells (209), and methylation of CBP/p300 is required for the activity and stability of the transcriptional coactivator p/CIP/SRC-3 in hormone signaling (136). Conversely, methylating the GRIP1 binding domain of p300 by CARM1 blocks its interaction with GRIP1 and inhibits estrogen receptor-mediated transcription (107). PRMT4 can also repress transcription by methylating the steroid receptor coactivator (SRC-3), which leads to decreased ERα-mediated transcription and termination of hormone signaling by disassembling the SRC-3/CARM1 coactivator complex (59).

CARM1 has only been examined in two studies where CARM1’s stabilization, regulated by nuclear AMPK, can contribute to autophagy in the mature heart (109) and where it was shown to regulate inflammation in acute coronary syndrome (117); however, its regulatory role was dependent on miR-15a targeting.

PRMT5 (protein arginine methyltransferase 5) is known as a master epigenetic regulator, essential for cellular growth and development, and functions in a wide variety of cellular processes, including signal transduction, transcription, translation, and DNA recombination. Among the long list of PRMT5 substrates (167), we will focus only on those which provide evidence for its regulatory role in gene expression. Specifically, PRMT5 has been shown to regulate transcription by either modifying histone residues or by directly modulating the activity of several transcription factors to both activate and repress transcription depending on the arginine residue that is methylated and/or the promoter targeted. For instance, PRMT5-mediated symmetric dimethylation of histone H3R8 and histone H4R3 leads to silencing of several tumor suppressor genes [including Sp7, Rb, and Ptpro (6, 145, 190)] and immune response genes (182); however, methylation of these same residues can also activate transcription of Fgfr-3 and Eif4 (204) and adipogenic differentiation genes (103). PRMT5 can also methylate histone H3R2, which recruits Wdr5 and the MLL complex and stimulates histone H3K4 methylation and euchromatin maintenance (127). In association with Mep50, PRMT5 also methylates cytosolic histone H2AR3 to repress differentiation-specific genes in pluripotent ES cells (176). Lastly, PRMT5 can methylate the heterogeneous nuclear ribonucleoprotein hnRNP A1, which promotes its interaction with IRES RNA and results in IRES-dependent translation of various genes (Ccnd1, Cdkn1B, Myc, HIF1a, Mtif, and Mep50) (68).

PRMT5 is required for muscle stem cell differentiation and is important in adult muscle regeneration and repair, but shown to be dispensable for expansion of muscle progenitor cells during development (207, 208). In the heart, it can also regulate cardiac sodium channels through arginine methylation (18) and hypertrophic signaling by methylating GATA4 on three arginine residues, which inhibits the ability of GATA4 to promote transcriptional activation (39).

CONCLUDING REMARKS

Cardiovascular epigenetics defines heritable mechanisms that establish stable patterns of cardiac gene expression without modifying the base sequence of DNA. It is known that at the basis of many heart pathologies lie the vast spectrum of dynamic changes in cardiac gene expression that represent the molecular underpinnings driving morphological and physiological transformation of the myocardium, which also include both maladaptive and compensatory mechanisms aimed at mitigating disease-induced remodeling (32, 153). This includes any global changes in gene expression required for chromatin to undergo structural changes to silence or activate specific loci within the nucleus (178). Over the last decade there has been impressive progress made in our understanding of how environmental conditions affect cardiac gene expression through epigenetic mechanisms. We now know much more about the roles of chromatin modifications and chromatin regulators in the context of cardiac diseases (115). Specifically, several studies in murine models of cardiac hypertrophy have demonstrated that modulating key epigenetic factors, such as HDACs (85, 132, 205), BET domain-containing proteins (53), or specific transcription factors (112) can influence changes in gene expression to prevent pathological remodeling; however, we still lack the deep understanding of the mechanisms responsible for regulating chromatin structure during the pathology of heart disease. Among the many epigenetic mechanisms, which regulate heart physiology and pathology, several excellent reviews have recently highlighted DNA methylation, noncoding RNAs, structural remodeling of chromatin (2, 35, 95, 115), and histone posttranslational modifications (12) and can be found using these references, among others. However, the role of the enzymes that add or remove these posttranslational modifications have been less well studied, and a more global view (not focused on the heart) has been reviewed previously (42).

In recent years, the contribution of epigenetics to cardiac physiology has emerged as a key regulatory process in both cardiac development and disease and has recognized the involvement of methyltransferases as major players in transcriptional regulation. Specifically, protein methyltransferases have been linked to the development of congenital heart disease, cardiac hypertrophy and heart failure (206). For instance, aberrant trimethylation by KMT2D can lead to Kabuki syndrome (137) or inactivation of DOT1L in the mouse heart causes downregulation of H3K79 methylation that results in the development of dilated cardiomyopathy (138). However, we have only begun to scratch the surface in understanding the specific roles of lysine and arginine methyltransferases in the heart. This review highlights the many methyltransferases expressed in the heart, how their expression is differentially regulated during development or disease, and emphasizes that only a limited number of these enzymes have been functionally examined in any detail in this organ (more than one-third of these enzymes remain uncharacterized). Therefore, future studies will be needed to interrogate these enzymes to determine the implications to cardiac pathophysiology. Additional research that is focused on expanding our understanding of how epigenetic regulation is driven by these uncharacterized methyltransferases may unlock their therapeutic potential with new perspectives for translational research and new diagnostic tools, as well as novel strategies in drug design and discovery.

Indeed, we have much to learn about these enzymes, including the specific molecular mechanisms governing substrate specificity and the downstream effects of their catalytic activity. As highlighted in this review, there is now strong evidence to show that these enzymes have a much broader role in regulating transcription than has previously been appreciated. While their substrate specificity was traditionally thought to be engendered by the ability of one enzyme to methylate one amino acid residue in one manner (monomethylation, dimethylation, or trimethylation) to activate or repress transcription, the compilation of methyltransferase data in this review shows that these enzymes maintain the ability to both activate and repress transcription in a context-specific manner. Interestingly, some methyltransferases are very specific and exhibit this dual functionality with only one methylation target. For example, DOT1L is the only enzyme known to methylate K79 on histone H3, which, depending on the degree of methylation (monomethylation, demethylation, or trimethylation), can have a positive or negative effect on transcription. However, other enzymes, like G9a/GLP are more promiscuous and in addition to methylating multiple histone residues, they can also methylate other substrates within the nucleus or cytosol to positively or negatively regulate transcription. Given the growing amount of evidence, which continues to reveal new insights into the function of these enzymes, one can appreciate how little we know about their regulatory roles and how much is still to be discovered. And finally, we cannot forget about the substantial list of putative enzymes that have been classified on the basis of their unique methyltransferase domains but await functional characterization.

GRANTS

This work was supported by the National Institutes of Health Grants R01-HL-130424 and F32-HL-144034, American Heart Association Grant 20PRE35120356, and the Nora Eccles Harrison Treadwell Foundation Grant 10038331.

DISCLOSURES

No conflict of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.W.S., K.D., A.B., R.B., and S.F. prepared figures; M.W.S., K.D., and A.B. drafted manuscript; M.W.S., K.D., P.A., E.H., and S.F. edited and revised manuscript; M.W.S., K.D., A.B., P.A., R.B., E.H., and S.F. approved final version of manuscript.