Epigenetics in Cardiac Hypertrophy and Heart Failure.

Heart failure (HF) is a complex syndrome affecting millions of people around the world. Over the past decade, the therapeutic potential of targeting epigenetic regulators in HF has been discussed extensively. Recent advances in next-generation sequencing techniques have contributed substantial progress in our understanding of the role of DNA methylation, post-translational modifications of histones, adenosine triphosphate (ATP)-dependent chromatin conformation and remodeling, and non-coding RNAs in HF pathophysiology. In this review, we summarize epigenomic studies on human and animal models in HF.

The word “epigenetics” is composed of a Greek prefix, “epi,” meaning “upon,” “on,” or “around” and the word “genetics.” Therefore, epigenetics can be defined as the mechanism that affects heritable changes in gene expression and function without alternating the sequence of the genome. It plays fundamental roles in regulating the architecture of chromatins and gene expression at various molecular levels for maintaining cell identity and controlling cell differentiation, which are critically important for normal development and diseases. Changes of the epigenome are hallmarks of various cancers and several potentially molecules have been identified as potential new drug therapies.

The study of epigenetics in cardiovascular diseases is a relatively new field. Heart failure (HF), being the leading cause of death worldwide, occurs when the myocardium has long been recognized to undergo structural and functional remodeling. Such processes may be the cause and/or consequence of various genomic and transcriptional reprogramming of the cardiomyocytes and other nearby cells. Hence, better understanding of how epigenetic regulations are involved in HF may open a new perspective for translational research into new diagnostic tools as well as novel strategies in drug design and discovery.

Epigenetic regulation can be classified at 4 different molecular levels: 1) DNA methylation, 2) post-translational modifications of histones, 3) adenoisine triphosphate (ATP)-dependent chromatin conformation and remodeling, and 4) non-coding RNAs. In this review, we will summarize the recent advances and provide an overview of the epigenomics studies focusing on HF in humans and animal models.

DNA Methylation: A Potential Biomarker for HF

The methylation of DNA in eukaryotes predominantly occurs at the fifth carbon of the pyrimidine ring of the cytosine (5mC) and followed by a guanine dinucleotide where cytosine and guanine are separated by a phosphate (CpG). DNA methylation plays a crucial role in gene regulation, particularly on transcriptional repression that depends on where methylation is located. Increased methylation of CpG-enriched regions, known as CpG islands (CGI) at the promoter region of genes, is often associated with gene silencing, whereas methylated CpG found in the gene body is usually related to gene activation (1). In many pathological conditions, particularly in the cancer genome, alteration of DNA methylation of genes is the first epigenetic hallmark associated with the disease process (2). For example, a typical signature in many types of cancer cells is the hypermethylation at CGI of tumor suppressor genes causing transcriptional silencing of these genes. The silencing of these tumor suppressor genes affects the progression of the tumorigenesis (3).

The pioneering genome-wide studies on DNA methylation in the failing human myocardium were performed a decade ago 4, 5. Movassagh et al. first demonstrated that a large population of CGI and promoters were hypomethylated in the end-stage failing heart, and such differential DNA methylation patterns correlated with differential expression of angiogenic factors (4). Using methylated DNA immunoprecipitation followed by sequencing, they determined that the differential DNA methylation between non-failing and end-stage failing hearts did not occur evenly across the genome, but concentrated at promoter CGIs, intragenic CGIs, and gene bodies (5). In 2013, Haas et al. identified a set of candidate genes with altered DNA methylation status that may be involved in HF by using a lower-resolution method (6). Among these candidate genes, 2 genes displayed differential gene expression in dilated cardiomyopathy (DCM): adenosine receptor A2A (Adora2A) and lymphocyte antigen 75 (Ly75). Interestingly, when they validated these 2 genes by knockdown either Ly75 or Adora2A using morpholino in zebrafish, both morphants (zebrafish mutants) developed severe HF similar to that observed in humans (6).

A more extensive human study on DNA methylation in HF was published recently. Meder et al. generated a genome-wide DNA methylation profile in patients with DCM and in donor left-ventricular biopsies and whole peripheral blood using a high-resolution epigenomic-wide method with a large cohort (7). They identified 59 CpG loci with significant changes in DNA methylation in the myocardium of patients with DCM compared with clinical controls. Among these CpG loci, 29 and 30 were hypomethylated and hypermethylated in DCM, respectively. Using multi-omics approaches, they linked a subset of 517 epigenetic loci with DCM and cardiac gene expression. By further examining the methylome of peripheral blood cells, a differential display of 217 methylation sites between controls and patients with DCM was observed. Moreover, when they compared methylome between myocardium and peripheral blood cells, they identified distinct epigenetic methylation patterns that are conserved between 2 tissues. For example, the NPPA (natriuretic peptide A, also known as ANP) and NPPB (natriuretic peptide B, also known as BNP) loci were demethylated in DNA in heart tissues and peripheral blood cells from patients with DCM (7).

Using a cell-sorting technique coupled with a whole-genome bisulfite sequencing approach, Gilsbach et al. showed that only a few genomic regions within gene body of 6 genes exhibited differential DNA methylation between non-failing and failing human cardiomyocytes (8). However, those genes did not express differentially between the 2 groups. Additionally, the global DNA methylation patterns are not significantly different between non-failing and failing cardiomyocytes. Therefore, the authors concluded that methylated CpG is relatively stable in chronic HF (8). However, this observation differs from other studies in which more differential DNA methylated regions and gene loci were identified. In fact, Meder et al. found 59 differential CpG regions in DCM comparing to non-failing hearts from a total of 72 left ventricular biopsies. Moreover, a recent study by Glezeva et al. using capture-based bisulfite sequencing method also identified 151 differential methylation regions in DCM in comparison with non-failing hearts (9). Such discrepant findings in DNA methylation patterns in the failing heart could be due to the different technologies for DNA methylation that were used in studies (whole genome bisulfite sequencing [WGBS] vs. bead array vs. capture-based bisulfite sequencing). Another reason could be that the study by Gilsbach et al. (8) used sorted cardiomyocytes whereas other studies used bulk left ventricle tissues 4, 5, 7, 9, and Gilsbach’s study only used 5 pairs of samples using WGBS method and the pooled biological replicates for DNA methylation analysis.

Taken together, all the published studies to date have demonstrated that alternation of DNA methylation is highly associated with DCM. Recently, coupling RNA-seq and genome-wide DNA methylation approaches, Pepin et al. linked DNA methylation to metabolic reprogramming in men with end-stage ischemic cardiomyopathy (ICM). They also identified the differential DNA methylation between ICM and non-ischemic cardiomyopathy (NICM) (10). Specifically, the authors observed that 12.6% of CpG sites were differentially methylated between ICM and NICM, and that hypermethylation within promoter-associated CpG islands was observed in ICM samples. These hypermethylated gene loci are involved in oxidative metabolism and are downstream of an epigenetic repressor Enhancer of zeste homolog 2 (EZH2, a component of the polycomb repressive complex 2 and a histone methyltransferase) as well as Kuppel like factor 15 (KLF15), which is also an EZH2 target. These observations implied that human ischemic HF may display a distinctive DNA methylation signature associated with oxidative metabolism and anaerobic glycolysis. Furthermore, differential DNA methylation is likely mediated by EZH2-DNA methyltransferase complex to affect its downstream gene targets that are involved in metabolic reprogramming, such as KLF15.

Different etiologies of HF may generate different DNA methylation patterns. Using a targeted bisulfite sequence approach for patients with HF with various etiologies, Glezeva et al. identified 5 unique differentially methylated regions (DMR) in hypertrophic obstructive cardiomyopathy (HOCM), 151 DMRs in DCM and 55 DMRs in ICM and a total of the 209 genes were associated with these DMRs (9). Further validation for the genes associated with these DMRs confirmed that 6 protein-coding genes and 2 microRNA (miRNA) displayed significantly altered gene expression in at least 1 of the disease groups comparing with the control group. Among them, 2 novel HOCM-related genes were identified from this study, namely HEY2 and MSR1. Both gene loci were hypermethylated in patients with HOCM, and their gene expression level was reduced. In addition, 2 genes involved in the regulation of the extracellular matrix, CTGF and MMP2, were found to be hypomethylated with increased gene expression level in DCM. Moreover, COX17, MYMO3, and miR24-1 were found to be hypermethylated with reduced gene expression, whereas CTGF miR155 was hypomethylated with elevated gene expression in ICM (9).

Post-Translational Modifications of Histones: Dynamic Processes Accomplished by Writers and Readers Enzymatically

Histone proteins are the component of the chromatin complex that forms chromosomes. The histone octamer contains 2 copies of H2A, H2B, H3, and H4 proteins and forms the core of a nucleosome, wrapped 1.67 times by approximately 147 base pair (bp) of DNA. This conserved core complex is linked with a short linker DNA in a range between 20 and 80 bp and assembling into higher-order structures. The structures are stabilized by the linker histone H1, which usually binds to the entry/exit site of DNA on the surface of the nucleosomal core and wraps another 20 bp of DNA (11).

Unlike DNA methylation, in which the location of the methylation in the genome has a big impact on its functions, the regulatory function of the epigenetic modifications on histones are more delicate. The functions of histone modification are influenced not only by the location of these modifications (i.e., which amino acid residue and of genomic regions), but also by the type and the number of modifications on histones. The post-translational modifications (PTMs) of histones include acetylation, methylation, phosphorylation, ubiquitylation, and sumoylation at different amino acid residues of the canonical histone proteins as well as variant histones such as H3.1,H3.3, H2A.Z, and macroH2A 12, 13. These modifications affect the gene expression by changing chromatin structure to influence DNA accessibility or by recruiting various regulatory molecules such as histone modifiers, chromatin regulators, DNA repair molecules, and transcription factors. Studies from the past 2 decades have established a series of epigenetic codes for various histone modifications associated with the transcriptional status of genes. For example, acetylation of histone H3 at lysine 27 (K) residue marks active enhancers and promoter regions, whereas tri-methylation of histone H3 at K4 residues (H3K4me3) is associated with the active promoter region of genes. The comprehensive details of the histone codes associated with the regulation of gene expression have been reviewed elsewhere (12).

The PTMs of histones and their roles during cardiomyocyte differentiation and heart development for tissue regeneration have been studied extensively 14, 15, 16, 17, 18. However, our understanding of these regulations in HF remains unclear. The PTMs of histones is achieved by the actions of histone “writers” and “erasers.” They are enzymes that either add (write) PTM to histones or remove (erase) PTM from the histone proteins. Currently, the best understood the regulation of PTM on histones is on acetylation, methylation, phosphorylation (19). In this review, we will focus on 2 major PTMs to histone proteins – methylation and acetylation.

Histone acetylation and deacetylation

Acetylation commonly occurs on the lysine residues of the histone, which neutralizes the positive charge of the lysine and reduces the ionic interaction between histone proteins and DNA. This increases the DNA accessibility to transcription factors, chromatin remodelers, and modifiers and is frequently associated with transcription activation (20). Histone acetylation is controlled by histone acetyltransferase (HAT [writer]) and deacetylase (HDAC [eraser]). HATs can be divided into 2 different classes based on their localization in the cell (21). Type A HATs are in the nucleus playing an essential role in activation of gene transcription, whereas type B HATs are located in the cytoplasm mainly involved in the acetylation of newly synthesized histone. There are 4 major classes of HDAC classified by their catalytic domain 20, 22. Class I HDACs, HDAC1, 2, 3, and 8, are widely expressed, whereas Class II (HDAC4, 5, 7, and 9 belong to the IIa group, and HDAC6 and 10 belong to the IIb group) are cell-type restricted. Class III HDACs are sirtuins 1-7 and they are NAD(+) dependent 20, 22.

Evidence from animal models shows that HAT and HDAC are highly linked to cardiac hypertrophy. When CREB-binding protein/p300, a well-known HAT, was overexpressed in mouse myocardium, the mouse developed left ventricular myocyte hypertrophy, dilatation, and dysfunction (23). Cardiac-overexpression of p300 in mouse also acetylated GATA4, a zinc finger transcription factor known for mediating myocyte hypertrophy and increased DNA binding activity of GATA4 to its downstream target. This stimulated the expression of the GATA4-dependent hypertrophy-responsive genes including Nppa, prepro-endothelin-1, and β-myosin heavy chain (Myh7). In contrast, in inhibition of p300 activity with curcumin (a p300-specific HAT inhibitor) in a phenylephrine-induced hypertrophy rat model, it was observed that cardiomyocytes decreased acetylation of histone 3 and histone 4 as well as GATA4 DNA binding activity. In a hypertension salt-sensitive Dahl rat model as well as a surgically induced myocardial infarction (MI) rat model, oral curcumin administration prevented deterioration of systolic function and HF (24). Therefore, deacetylation or inhibition of HAT may play a role in HF.

Consistent with this notion, several studies have manipulated the expression of histone deacetylase in mouse hearts and in primary cardiomyocytes. Single knockout for deacetylase class II members (HDAC II) Hdac5 or 9 in mice were sensitive to cardiac stress signals, such as pressure overload and calcineurin stimulation, and developed age-dependent cardiac hypertrophy 25, 26. Moreover, expression of cardiac hypertrophic genes (such as Nppa and Myh7) was enhanced in response to the stress signals in Hdac5- or Hdac9-null mice. This cardiac phenotype was caused by the activation of a prohypertrophic transcription factor, myocyte enhancer factor-2 (MEF2C), which can be inhibited by HDAC II (27). When infecting with a mutant form of Hdac9 or Hdac5 with more potent repressive activity on gene expression in phenylephrine-induced hypertrophic cardiomyocytes, it was demonstrated using chromatin immunoprecipitation coupled with quantitative PCR (ChIP-qPCR) assay that the acetylation state of histone 3 was significantly reduced at promoter regions of Nppa and Myh7. However, the gene loci that were not regulated during hypertrophy (e.g., Gapdh) were not affected. Furthermore, reduction of the acetylation on histone 3 correlated with the gene expression of Nppa and Myh7 but not Gapdh (25). The results indicated that HDAC II has an inhibition role for cardiac hypertrophy.

Interestingly, when the activity of HDAC class I (HDAC I) was abolished, it had effects opposite to that of HDAC II. The inactivation for HDAC I member, Hdac2, in mice genetically or chemically, resulted in less sensitivity to hypertrophic stimuli, whereas the mice with Hdac2 overexpression developed cardiac hypertrophy (28). The expression of cardiac hypertrophy genes, namely, Nppa, Mhy7, and Acta1, was not significantly different between control littermate and Hdac2 mice following transverse aortic constriction (TAC) or isoproterenol infusion (28). However, the state of histone acetylation was not investigated in this study. Cardiac-specific inactivation of Hdac3, another member of HDAC I, in mice caused cardiac hypertrophy as well as affected the genes associated with the fatty acid metabolism. Although there are no global changes of histone acetylation in Hdac3 conditional knockout (cKO) hearts, histone acetylation was found to be increased at the promoter region of myocardial energetic genes and glucose use such as Ucp2 and 3 (uncoupling protein 2 and 3), fatty acyl-CoA synthetase, fatty acid transport protein, and Pyruvate dehydrogenase kinase 4 (29). These data suggested that HDAC3 may have a complementary role in regulating cardiac growth and myocardial energy metabolism (29).

Histone methylation and demethylation

Histone methylations can be associated with gene activation or repression, depending on which residue on histone the modification occurred. For example, the H3K4me3 is usually associated with active promoter activity whereas the tri-methylation of histone 3 lysine 9 (H3K9me3) is usually linked to transcriptional repression 12, 14. Methylation of histones also plays a critical role in the progress of HF as evidenced by the different effects in modulating of histone demethylase (HDM,[eraser]), co-factor or histone methyltransferase (HMT,[writer]) in cardiomyopathy mouse models. Zhang et al. demonstrated that JMJD2A, a HDM, promotes cardiac hypertrophy in TAC mice (30). Inactivation of Jmjd2a in the mouse myocardium resulted in an attenuated hypertrophic phenotype after the TAC procedure, whereas the overexpression of Jmjd2a increased the hypertrophic response to TAC-induced hypertrophy. This increasing development of hypertrophy is due to the activation of four-and-a-half LIM domain1 (FHL1), which is a key molecule in developing hypertrophy in the TAC mouse model (31). It was found that JMJD2A binds to the promoter region of FHL1 and reduced the level of H3K9me3 in the heart after the TAC procedure. Interestingly, JMJD2A was significantly up-regulated in patients with HOCM (30).

Another example of histone methylation can be found in disruption of dystrophin-glycoprotein complex in DCM. When a disruptor of telomeric silencing (DOT1l), the H3K79 methyltransferase, was specifically deleted in the mouse heart, mutant mice developed various abnormalities leading to cardiac remodeling and increased cardiomyocyte cell death (32). Further analysis of these mice found that the cardiac abnormalities were due to dysregulation of dystrophin (Dmd) in hearts. Consistent with its methyltransferase function, DOT1L was found to be directly bound to the Dmd gene locus in the postnatal mouse heart. This binding correlated positively with the enrichment of H3K79me2/3 in the Dmd gene body, whereas such enrichment was abolished in the Dot1l cKO mouse hearts (32).

Moreover, a recent study showed that G9a/EHMT2, another HMT, was found to be required for cardiomyocyte homeostasis to silence the fetal gene program in the adult heart (33). Conditional aberration of G9a in 2-month-old mouse myocardia caused various cardiac dysfunctions within the first 4 weeks of knockout induction. The results indicate that G9a is required for the proper cardiac function in normal mouse hearts. Further investigation of G9a-KO mice in comparison with control showed that G9a suppresses the key cardiac regulatory genes via H3K9me2. G9a also interacts with EZH2 (an HMT) and forms a complex with MEF2C to regulate MEF2C-dependent gene expression. Interestingly, G9a was found to be up-regulated during the initial stages of cardiac hypertrophy. Furthermore, chemical inactivation of G9a in TAC mice improved cardiac function and prevented the development of hypertrophy suppressing the expression of anti-hypertrophy genes in cardiomyocytes (33). Taken together, G9a functions as an anti-hypertrophic regulator in healthy hearts, while serving as a hypertrophy activator in stressed hearts to promote cardiac hypertrophy.

Finally, the misregulation of the cofactor of histone methylation also showed effects on the regulation of HF. In vivo cardiac-specific deletion of the PAX transcription activation domain interacting protein (PTIP), a cofactor of H3K4 methylation, in adult mice caused the reduction of H3K4me3 (34). Using RNA microarray analysis, it was shown that a total of 221 genes were significantly altered in the Ptip cKO mouse hearts. The altered genes include Kv channel-interacting protein 2 (Kcnip2), a regulator for cardiac repolarization current that was previously shown to be down-regulated in a failing heart. Further analysis using ChIP-qPCR to detect H3K4me3 enrichment at 5’ regulatory region of Kcnip2 in control and Ptip cKO mouse hearts confirmed that the Ptip deletion directly affects the reduction of H3K4me3 at Kcnip2 gene region (34). These findings support the notion that the methylation of histones plays a crucial role in modulating cardiac hypertrophy and HF.

One interesting study done by Hohl et al. showed that, although HDAC4 belongs to histone deacetylase, HDAC4 may play a central role for rapid modifications of H3K9 methylation in response to cardiac load. Using human left ventricular tissues in ChIP-qPCR assays, Hohl et al. found that the promoter regions of Nppa and Nppb, 2 hallmark genes for maladaptive remodeling of the left ventricle, displayed a higher level of activation marks for H3K9ac and H3K27ac in failing hearts, whereas the repressive marks, H3K9me3 and H3K9me2, were substantially reduced (35). Using a similar method, the authors also observed that heterochromatin protein 1 (HP1), a transcriptional repressor and H3K9me3 binding protein, was dissociated from the promoter regions, which may activate both Nppa and Nppb gene expression. The H3K9 demethylation and up-regulation of NPPA and NPPB in human failing myocardium correlated with the nuclear export of phosphorylated HDAC4. Inactivation of HDAC4 in the mouse hearts resulted in a reversal of these events in cardiomyocytes (35). How does a histone deacetylase “promote” demethylation of histone? Apparently 2 HMTs, JMJD1A and 2A, were up-regulated in failing human hearts. The up-regulation of JMJD1A was positively correlated with the increased NPPA and NPPB expression. In addition, both demethylases bind to the promoter region of NPPA. HDAC4 is associated with the transcriptional repression complex with the H3K9me3 binding protein, HP1, and the histone methyltransferase, SUV39H. This complex is dissociated in response to the stress condition by the calmodulin-dependent protein kinase II δB–induced phosphorylation and nuclear export of HDCA4. Therefore, it is conceivable that the dissociation of the HDAC4-HP1-SUV39H complex opens up space at the promoter region of Nppa and Nppb, allowing the entry of the HMTs, JMJD1 and 2, to achieve the demethylation machinery in failing myocardium (35).

Genome-wide studies on PTMs of histones

Several genome-wide studies on PTMs of histone in myocardium have been conducted in animal models and humans. The first genome-wide study on methylation of histone in both human and rodent myocardium was published in 2009. Using a chromatin immunoprecipitates using genomic tiling arrays (ChIP-on-ChIP) technique, Kaneda et al. showed that the distribution patterns of H3K4me3 and H3K9me3 were significantly different between healthy and failing hearts (36). The differential distribution patterns of the 2 histone marks were found mainly at the gene loci involved in calcium signaling and cardiac contractility during the development of HF. In 2011, Movassagh et al. demonstrated that H3K36me3, a histone marker for active transcription, exhibited a differential methylation pattern and enrichment in cardiomyopathic and normal human hearts. The differential displays of H3K36me3 were mainly found in non-coding RNA loci, suggesting the involvement of the non-coding RNA in cardiomyopathy (4).

The most comprehensive genome-wide studies on PTMs of histones in cardiac hypertrophy and HF in an animal model was published in 2013 by Papait et al. Using ChIP followed by sequencing analysis for mouse cardiomyocytes isolated from sham and TAC mice for 7 histone epigenetic marks, Papait et al. identified that the level of histone acetylations (e.g., H3K9ac and H3K27ac) was decreased around the gene loci whose expression was down-regulated in TAC mice (37). In addition, there was a correlation between the gene expression level for those differentially expressed in TAC mice and the enrichment of histone methylation repression marks (e.g., H3K27me3, H3K9me2, and H3K9me3). The more active the transcription, the less enrichment for histone repression marks they observed (37). The authors also identified 9,207 cardiac-hypertrophy-specific-active enhancers that have DNA-binding motifs for MEF2C and its closely related family member, MEF2A. These results were validated using ChIP-qPCR analysis. The data suggested that MEF2A and 2C stimulate the cardiac hypertrophy gene expression via enhancers (37).

Although several genome-wide studies in the myocardium in both animals and humans have been published, their measurements were performed in bulk tissue specimens rather than in microdissected cells. Therefore, the data interpretation was based on several mixed-cell populations. Because epigenetic regulation is critical for cell-fate specification and differentiation, the effect of such regulation is highly cell-type–specific. Thus, it is necessary to determine the epigenetic signatures in a specific cell type. Recently, Gilsbach et al. used a nuclear staining with fluorescence-assisted sorting method to specifically isolate human cardiomyocytes from normal fetal, infant, and adult hearts as well as failing ones for a genome-wide epigenetic study (8). They showed that the whole-transcriptome regulation for normal cardiomyocytes during normal development is established by dynamic methylated CpG and common histone signatures at distal cis-acting elements of genes and gene bodies. One interesting finding from this study is that the expression of failing myocardium-related genes is modulated by active enhancer histone marks but not by DNA methylation. This observation is different from what was previously reported, and needs to be validated in a larger cohort study. Furthermore, because a heart is made up of other cell types besides cardiomyocytes (such as cardiac fibroblasts), it is necessary to generate epigenetic maps for other heart cell types.

In summary, our understanding of histone modifications on adult and failing hearts may have improved compared with a decade ago. However, we still lack systematic studies for normal and failing hearts in humans. Moreover, the currently published histone modification human studies have focused mainly on DCM and predominantly in males. Different types of cardiomyopathy may have their own distinct regulatory mechanisms. Sexual dimorphism can also link to different genetic and epigenetic regulations (38).

ATP-Dependent Chromatin Conformation, Remodeling, and HF

The mammalian chromatin architecture and organization are highly regulated. The dynamic modification of the epigenome, including post-translational modification on histone tails and DNA methylation, causes chromatin remodeling and re-organization. This is an important process for controlling cell-fate determination, differentiation, and organ development (14). The chromatin regulators, such as histone modifiers, co-factors, and transcriptional regulators, can recruit or be recruited by other chromatin regulators and transcription factors to regulate gene expression. By recruiting all necessary regulators, the distal cis-acting elements, such as enhancers, can form chromatin loops to interact with the proximal promoter regions of genes, and thus initiate and enhance the transcriptional activity of genes with the basal transcriptional complex (e.g., RNA polymerase II).

The histone writers and erasers achieve the PTMs on histones. These PTMs serve as “marks” to be recognized and interpreted by “readers” that carry on the action for regulating gene expression. The role of readers in HF is just emerging. Studies on modulating the function of bromodomain (BET) proteins, an acetylated-lysine binding protein and an epigenetic reader, have suggested that chromatin structure and organization are important in regulating the kinetics of gene expression in failing hearts 39, 40. The BET family comprises BRD2, BRD3, BRD4, and BRDT. Among them, BRD4 protein expression was increased during cardiac hypertrophy. Inhibition of BET activity with the small molecule JQ1, either in mice or primary cell culture, suppressed phenylephrine-mediated cellular hypertrophy and pathological gene induction. Spiltoir et al. showed that the hypertrophic stimuli stimulated recruitment of BRD4 to the transcriptional start site of the gene Nppa. Binding of BRD4 to the ANF transcription start site was associated with increased phosphorylation of local RNA polymerase II. Recently, Duan et al. showed that administration of the JQ1 has therapeutic effects in animal HF models (41). Using an unbiased ingenuity pathway analysis, the authors showed that BET inhibition preferentially suppressed innate inflammatory and profibrotic transcriptional networks, namely, Nuclear Factor κB and Transforming Growth Factor–β signalings. Taken together, these studies suggested that BET proteins can be good pharmacological targets for treating HF.

In addition to the histone writers and readers and co-activators mentioned above, ATP-dependent chromatin remodeling complexes (ADCRs) are another group of regulators that may play a critical role in chromatin remodeling. ADCRs contain an adenosine triphosphate (ATPase) subunit that belongs to the SNF2 superfamily of proteins. The complex uses ATP hydrolysis as an energy source to alter or disrupt the histone-DNA interaction. In mammals, 4 ADCR families (the SWI/SNF, ISWS, CHD, and INO80) have been identified (42). Among these families, the roles of the SWI/SNF group in cardiac development and hypertrophy have been studied extensively during the past decade 43, 44, 45, 46, 47, 48. These studies have found that BAF complex, the mammalian analog of the SWI/SNF complex, is usually abundant in the embryonic heart but down-regulated in adult myocardium. The complex modulates gene expression of an important and motor molecule of the heart, myosin heavy chain (MHC). Two isoforms of MHC, α- and β-MHC, are specifically expressed in the mammalian hearts, and they are located on the same chromosome (49). Alpha-MHC has higher ATPase activity than β-MHC. Alpha-MHC is expressed in adult cardiomyocytes primarily, whereas β-MHC is expressed in embryonic cardiomyocytes.

How does the BAF complex modulate the expression of these 2 isoforms? Using various myocardial-specific deletion of Brg1 mouse models at different developmental stages, it was shown that BRG1, an important ATPase subunit of the BAF complex, interacts with HDACs and poly ADP-ribose polymerase (PARPs), e.g., PARP1, to suppress the adult isoform of α-MHC, encoded by Myh6 gene, and activates the fetal isoform (β-MHC), encoded by Myh7 gene, in the embryonic heart. Because the BAF complex was down-regulated in adult cardiomyocytes, the suppression of α-MHC is removed. Therefore, the expression of the adult isoform of MHC (α-MHC) is up-regulated (46). Interestingly, the down-regulation of BRG1 gene expression at birth was also observed in normal human hearts. However, unlike in adult mouse hearts where the BRG1 was undetectable in cardiomyocytes, it remains detectable in human adult cardiomyocytes, suggesting the other mechanisms may be present in regulating the switch of fetal and adult MHC in humans (50). Nevertheless, in hypertrophic and failing hearts, the subunits of BAF complexes and its binding partners, HDACs and PARPs, were increased and the expression of fetal MHC was up-regulated, whereas the adult MHC was decreased (46).

Another mechanism for the inhibition of adult Myh6 gene is that BRG1 recruited G9a, a HMT, and DNA methyltransferase (DMNT3) to the promoter region of Myh6 in the adult hypertrophic and failing hearts. The recruitment of G9a/DMNT3 complex resulted in the deposition of the H3K9me3, a repressive chromatin mark, and CpG methylation at promoter region of Myh6, which may subsequently activate the inhibition machinery for Myh6 gene (51). When Brg1 was specifically deleted in mouse hearts, the cardiac hypertrophy was diminished in the TAC mice. Therefore, it is conceivable that the proper expression and maintenance of the state of adult MHC expression can be a promising approach for treating HF. Moreover, BRG1 can directly increase the expression of osteopontin, a pro-fibrotic factor induced cardiac fibrosis (52). Thus, it seems that BRG1 acts as a stress-activated chromatin remodeler that controls fetal reprogramming of cardiac genes, e.g., the switch of adult form, Myh6, to fetal gene, Myh7, and activates hypertrophy and HF (46).

Chromatin remodeling complexes can swap histone variants, such as H3.3 and H2A.Z, with the canonical histone proteins in the nucleosome. It is clear now the incorporation of histone variants is critical for regulating the chromatin structure and gene expression. For example, a SWI/SNF chromatin-remodeling complex, SWR1, can switch H2A with H2A.Z at specific genomic loci to maintain genome integrity and to facilitate initiation of transcription, whereas INO80 chromatin-remodeling enzyme can evict H2A.Z from nucleosome to stabilize the chromatin (53).

The role of histone variants in development and diseases has been described elsewhere 54, 55, 56. Nevertheless, the role of histone variants in HF remains largely unknown. It was reported that the histone variant H2A.Z is up-regulated in TAC mice hearts and knockdown H2A.Z with shRNA in cultured rat cardiomyocytes under mechanical hypertrophic growth conditions attenuated cardiac hypertrophy and downregulated growth-related genes (namely, cyclin-dependent kinase 7 and ribosomal S6) (57). When HIRA, a histone chaperone protein responsible for incorporation of histone variant H3.3, was specifically deleted in the mouse heart, the heart developed fibrosis, cardiac hypertrophy, as well as abnormal cardiac function. In addition, the sarcolemmal integrity was compromised in the Hira-deleted mouse (58). The histone variant H3.3 was shown to be associated with gene bodies of active genes in non-pluripotent cells and enriched at both active and silence genes in embryonic stem cell and precursor cells (59). Misregulation of incorporation of H3.3 would alter the gene expression. Indeed the cardiac transcriptome of Hira-cko hearts was altered in comparison with the control hearts.

Chromatin loops participated in the same regulatory territory, forming a higher order of chromatin structure known as topologically associating domains (TAD). Recently, the importance of TAD in HF has been reported (60). Rosa-Garrido et al. inspected the chromatin configuration differences among cardiomyocytes isolated from 3 different groups of mice and their correlation to the gene expression using high-resolution chromatin mapping and RNAseq approaches. These 3 groups are as follows: (1) normal adult, (2) induced-cardiac hypertrophy by TAC, and (3) cardiac-specific deletion of CCCTC-binding factor (CTCF) (Ctcf-KO) mice. CTCF was previously shown to regulate chromatin three-dimensional (3D) structure 61, 62. Interestingly, mice with the cardiac-specific deletion of Ctcf developed HF. Moreover, the expression of CTCF was reduced in human patients with a failing heart and was increased in patients whose hearts were assisted by a left ventricle assist device. Rosa-Garrido et al. also described the alterations of chromatin architecture in healthy and diseased cardiac myocytes in mice. Significant alternations of chromatin loops and the boundary strength of TADs were found in cardiomyocytes from both Ctcf-KO and TAC mice. The enhancers and promoters interaction for major cardiac-related loci was also altered in the cardiomyocytes from failing hearts. In addition, there was a positive correlation between the alteration of gene expression levels and changes in chromatin compartmentalization in Ctcf-KO and TAC mice. It remains unclear if the similar TADs and chromatin looping structures and alterations would exist in human patients with HF.

Using a chromosome conformation capture technique approach, Hi-C, to create the 3D map of the chromosome interaction for induced pluripotent stem cell-derived cardiomyocytes, Montefiori et al. recently linked the known cardiovascular disease–associated single nucleotide polymorphism to the target genes (63). The authors identified that the majority of non-coding regions could influence genes that are far away from them and some regions may modulate the gene expression for more than 2 genes. These findings are consistent with the notions that the long distance enhancers interact with the promoter to establish a long-range control of gene expression, and it is important to consider this aspect when interpreting functional targets of disease loci. A summary of epigenetic modifications and chromatin remodeling involved in cardiomyopathy is listed in Table 1.

Table 1

Summary of Chromatin Regulator in Cardiomyopathy

Regulator	Action	Effect	Reference
CREB-binding protein (P300)	Acetylation of histone tails.	(1) Increasing acetylation of MEF2 and GATA4. (2) Cardiac hypertrophy regulation.	23, 24
HDAC2	Deacetylation of histone tails.	(1) Overexpression of HDAC2 caused stimulation of the Akt/GlaxoSmithKline3b pathway. (2) Cardiac hypertrophy regulation.	28
HDAC3	Deacetylation of histone tails.	Regulation of cardiac growth and myocardial energy metabolism.	29
HDAC4	(1) Deacetylaion of histone tails. (2) Control H3K9 demethylation and HP1 dissociation to the NPPA promoter in response to elevated preload.	Mediation of H3K9 methylation and HP1 dissociation to NPPA promoter.	35
HDAC5 and 9	Deacetlylaion of histone tails.	Inhibit the transcriptional activity of MEF2c and act as negative regulator of cardiac hypertrophy.	25, 26
DOT1L	Methylation of H3K79me.	Reduction of DOT1L activity causes DCM.	117
JMJD2A	Demethylation of H3K9me3, H3K4me3, and H3K27me3.	Activate cardiac hypertrophy and alter cardiac gene expression.	30
PTIP	(1) Co-factor of H3K4 methylation (2) Regulates the expression of Kcnip2.	Misregulation of PTIP cause cardiac hypertrophy and failure.	34
G9a/EHMT2	Methylation of H3K9me2 and H3K27me3 (lesser extent).	(1) Maintain cardiomocyte homeostasis and interact with MEF2C to silence the fetal gene program in the adult heart. (2) Promote cardiac hypertrophy in stressed hearts.	33
SWARCA4 (BRG1)	(1) Regulates PARP-1/HDAC. (2) Recruits G9a and DMNT3.	(1) Activation of Brg1 diminished cardiac hypertrophy in TAC mice. (2) Activates fetal MHC isoform expression and repress adult MHC isoform in the hypertrophic and failing adult heart.	46, 51
CTCF	Regulating chromatin 3D structure.	Inactivation of CTCF caused HF.	60

3D = 3 dimensional; CTCF = CCCTC-binding factor; DOT1l = disruptor of telomeric silencing1 like histone lysine methyltransferase; H3K4me3 = tri-methylation of lysine 4 on histone 3; H3K9me3 = tri-methylation of lysine 9 on histone 3; HDAC = histone deacetylase; HF = heart failure; HP1 = heterochromatin protein 1; Kcnip2 = Kv channel-interacting protein 2; MHC = myosin heavy chain; NPPA = natriuretic peptide A; PARP = poly ADP-ribose polymerase; PTIP = PAX transcription activation domain interacting protein; TAC = transverse aortic constriction.

Long Non-Coding RNAs: The New Players in Epigenetic Regulation for HF

Only <2% of transcribed RNAs translate into proteins 64, 65. Recent advances in transcriptomics and bioinformatics techniques revealed that untranslated RNAs (non-coding RNA [ncRNA]) are key regulators for transcription and translation in developmental as well as disease processes. In addition to the traditional, well-known non-coding RNAs (such as transfer RNA and ribosomal RNA), the rest of the non-coding RNAs can be classified into 2 categories based on their size. One category of ncRNA is the small nc-RNA, which is <200 nucleotides long. Small nc-RNA includes miRNA, piwi-interacting RNA, short interfering RNA, small nucleolar RNAs, small nuclear RNAs, extracellular RNAs, and small Cajal body-specific RNAs. The other category of ncRNA is the long ncRNAs (lncRNAs), which are longer than 200 nucleotides. lncRNAs have been linked to playing critical roles in chromatin remodeling. The epigenetic functions of small nc-RNA (especially miRNA) are extensive and have been reviewed elsewhere 66, 67, 68, 69, 70. Here we will focus on the epigenetic roles of lncRNA in HF.

Like other non-coding RNA, lncRNAs do not encode any protein product but elicit their functions via diverse mechanisms (71). For example, lncRNA can interact with chromatin regulators and transcriptional modifiers to enhance or repress expression of downstream genes. IncRNAs can also regulate gene expression by interfering with miRNA pathways. They also play essential roles on post-translational regulations that involve RNA splicing, messenger RNA (mRNA) stability, and translation. Furthermore, several lncRNAs have been identified to play vital roles during heart development (72).

Studies suggested that lncRNA plays a vital role in myocardium remodeling and pathogenesis in failing hearts. The mitochondria-derived lncRNA long intergenic non-coding RNA predicting cardiac remodeling (LIPCAR) was associated with left ventricular remodeling. This assumption is based on the observation that the LIPCAR is elevated in plasma of patients with HF after MI compared with MI patients without left ventricular remodeling (73). By lncRNA array screening, Piccoli et al. identified that cardiac fibroblast–specific lncRNA, maternally expressed gene 3, was the most abundant CF-lncRNA in the mice that underwent the TAC procedure during the first 4 weeks (74). The Wang and Xiao’s group together identified that 15 and 135 lncRNAs were dysregulated in cardiac hypertrophy and failing mouse hearts, respectively (75). The study used whole-transcriptome analysis for left ventricle samples collected from mice after 1 week of TAC (hypertrophic stage), and 8 weeks of TAC procedure (end-stage HF). Among these dysregulated lncRNAs, 2 of them are particularly interesting. The first 1 is H19 and is a highly abundant and conserved imprinted lncRNA. H19 was significantly up-regulated in HF samples in comparison with control samples. Consistent with this finding, H19 was later shown to be a negative regulator of cardiomyocyte hypertrophy (76). Another interesting lncRNA from their findings is a cardiac hypertrophy–associated epigenetic regulator (Chaer), which is a heart-enriched lncRNA (77). Chaer was shown to be involved in cardiac hypertrophy development by directly binding to the catalytic subunit of polycomb repressor complex 2 (PRC2). PRC2 is responsible for methylation of H3K27me3 for gene silencing. The interaction between Chaer and PRC2 decreased the level of H3K27me3 in the promoter region of cardiac hypertrophy–related genes, such as NPPA. Therefore, it may cause induction of these cardiac hypertrophy–related genes in hypertrophy hearts (77).

Long non-coding RNAs also regulate the chromatin-remodeling molecule to maintain cardiac functions. A cardiac-specific long-non RNA cluster, Myheart (Mhrt), was found to abound in the heart and have a protective role in pathological hypertrophy in heart (52). Mhrt is a cluster of lncRNA and was located at myh6-myh7 gene loci. Expression of Mhrt was down-regulated in TAC mice hearts, in part due to the suppression by BRG1-HDAC-PARP chromatin remodeling complexes. Genetic restoration of Mhrt779, the most abundant Mhrt species, prevented the pathological hypertrophy phenotype after TAC in the heart. In fact, Mhrt can directly bind to the helicase domain of BRG1 that prevents the binding of BRG1 to its genomic DNA targets and thus prevents activation of downstream gene targets of BRG1, such as Myh6 (52).

Using a global lncRNA expression profiling method, another cardio-specific enriched lncRNA, cardiac hypertrophy-associated transcript (Chast), was identified (78). Chast is specifically up-regulated in cardiomyocytes in TAC mice as well as in human patients who have aortic stenosis with hypertrophy heart phenotypes. Moreover, Chast seems to inhibit the expression of Pleckstrin homology domain-containing protein family member 1 (Plekhm1). PLEKHM1 is an autophagy regulator. Because inhibition of autophagy in cardiomyocytes was shown to cause cardiac remodeling, it was proposed that Chast enhances cardiac hypertrophy and myocardium remodeling by inhibiting the function of PLEKHM1 to prevent autophagy of cardiomyocytes.

Using a deep sequencing of the non-coding RNA transcriptome method, Yang et al. found that the expression of lncRNAs, but not miRNA, is differentially expressed between nonischemic and ischemic human failing left ventricles (79). Interestingly, 10% of the differentially expressed lncRNAs can be improved or normalized for patients whose hearts were assisted with a left ventricular assist device. A recent study evaluated the potential of circulating lncRNAs as biomarkers for HF (80). Xuan et al. assessed the circulating levels of 13 known to be relevant to cardiovascular disease lncRNAs. Plasma samples from 72 patients with HF and 60 from non-HF patients were tested using quantitative reverse transcription-PCR. They identified 2 lncRNAs, NRON and MHRT, that may serve as novel predictive biomarkers for a failing heart.

As mentioned earlier, lncRNAs can also elicit their function by interfering with the function of miRNAs. One example is the interaction between lncRNA (cardiac hypertrophy–related factor [CHRF]) and mir-489 (81). CHRF was shown to act as a sponge RNA to directly bind to mir-489, an anti-cardiac hypertrophy miRNA. This interaction affects the inhibition of myeloid differentiation primary response gene (myd88) by miR-489 and, therefore, causes cardiac hypertrophy.

Nutrition, Gut Microbiome, and Epigenetics: Interplay Among Host, Gut Microbiotial Metabolites, and Host Chromatin

In addition to the intrinsic factors, epigenetics is highly influenced by external factors, such as nutrients, toxins, stress, and other environmental factors directly or indirectly. Recently, attention has focused on how microbe diet interactions influence gene activity via epigenetic mechanisms that may affect an individual’s health 82, 83, 84. Many gut metabolites, such as butyrate, acetate, and folate, may influence epigenetic processes by regulating enzyme activity that is responsible for epigenetic modifications (85). Butyrate, a short-chain fatty acid, is a potent inhibitor of HDACs and thus causes histone hyperacetylation (86). The acetylation of histone increases chromatin accessibility, which allows access of transcription factors and other critical factors to the open chromatin regions allowing for gene transcription. The presence of butyrate and its related sub-product, e.g., sodium butyrate, thus can alter the gene expression via inhibition of histone deacetylase 86, 87, 88. In colon cancer research, it was demonstrated that butyrate inhibits HDAC activity and elicits its anti-inflammatory effects via suppressing the nuclear factor κB and interferon γ production and increases the expression of peroxisome proliferator-activated receptor γ (PPARγ) 89, 90. Many studies have shown that activation of the PPARγ has a beneficial role in cardiac diseases (91). Therefore, butyrate may have a direct protective role in cardiovascular disease via anti-inflammatory machinery. However, whether or not the anti-inflammatory effects of butyrate also involved epigenetic regulation in cardiovascular diseases remains elusive. In addition, Mathew et al. showed that the effect of butyrate on histone H3 modification altered the G1-specific cell cycle proteins and further resulted in the arrest of smooth muscle cell proliferation, which may provide an atheroprotective potential (92). The data suggested that butyrate may have a role in preventing cardiovascular diseases as well as in therapeutic intervention of arterial restenosis and in-stent-restenosis as a pharmacological agent, respectively (92).

Folate is a type of B vitamin that can be synthesized by gut bacteria or obtained from the diet, and is required for methionine homeostasis (85). It is a methyl donor to generate S-adenosylmethionine (SAM), which is a methyl-donating substrate for methyltransferase needed for the methylation of cytosine in DNA or lysine in histones. Folate and methionine deficiency leads to reduced SAM levels and subsequently reduces the methylation on histone 3 lysine (H3K4) in yeast and human cells as well as in an in vivo model 93, 94. In addition, it was shown that the deficiency on dietary methyl-donors had a great impact on the epigenetic profile and also resulted in the development of metabolic diseases (95). However, the link between the gut microbiome-produced folate to HF remains unclear.

Choline is another methyl-group donor for epigenetic modification and is an essential nutrient. Choline is not only used by the host but also by the bacteria residing in the intestine of the host. Gut microbes used choline to produce trimethylamine (TMA), which can be further oxidized in the host’s liver to generate trimethylamine N-oxide (TMAO), a metabolite linked to HF susceptibly, adverse prognosis, and induction of inflammation 96, 97. By manipulating choline in a gnotobiotic mouse model harboring the synthetic gut bacterial community either with the 1 can consume choline and produce TMA (CC+) or the 1 can not (CC-), Romano et al. showed that gut bacteria compete with the host for the usage of choline. This competition caused the global reduction of DNA methylation on examined organs of the host, namely, brain, heart, liver, and colon (98). Despite, the elevated TMAO, behavior changes, and exacerbation of metabolic disease, no cardiac phenotype was reported. Additionally, how TMAO influences the susceptibility of the failure and if this gut metabolite has any epigenetic role in the heart remain unclear. Although the ability of the gut microbiome to influence epigenetics has been established in cell lines and lower model organisms focusing on brain development, immune system, and cancer biology 86, 87, 88, 92, data supporting their contributory role in HF and cardiovascular diseases are still emerging.

Therapeutic Potential of Targeting Epigenetic Regulators in HF

Many drug candidates targeting epigenetic molecules have been identified for cancer therapy and other diseases. Also, several of them have been used in cardiac hypertrophy and failing heart studies in cellular and animal models (Table 2). Those epigenetic-candidate drugs may also be used to treat patients with cardiac hypertrophy and HF. One promising approach is targeting the BET complex. As mentioned earlier, BET proteins play roles in chromatin remodeling and establishing the basic transcriptional complex at the promoter region of genes to regulate gene expression. Several studies showed that administration of a BET inhibitor, JQ1, in vivo and in vitro could suppress pathological cardiac remodeling, suppress pathological hypertrophy, and block the innate inflammatory and profibrotic myocardial program 39, 41.

Table 2

Potential Epigenetics Drugs for HF

Chemical	Action	Cardiac Outcome	Current Clinical Application	Ref. #’s
5-aza	DNA MTi.	Improved cardiac function partially.	ALL; AML; sickle cells disease.	113, 114, 115
Apicidin	HDACi for class I subtypes 1, 2, and 3.	Decreased myocardial hypertrophy.	NR	109
Curcumin	HATi for p300.	Prevented deterioration of systolic function and HF.	OA; RA; diabetes.	24, 110, 111, 112
Givinostat (ITF2357)	Pan-HDACi.	Decreased inflammatory response and angiogenic effects. Reduced EMT and cardiac fibrosis.	Duchenne muscular dystrophy; juvenile idiopathic arthritis; polycythemia vera; myelofibrosis.	104, 105, 106, 107
JQ1	BET bromodomain inhibitor.	Suppressed cardiac hypertrophy and pathological cardiac remodeling. Improved cardio function in TAC mice.	Variety of cancers.	39, 40, 41
SK-7041	HDACi for class I subtypes 1 and 2.	Reduced myocardial hypertrophy.	NR.	108
Trichostatin A	HDACi for class I and II.	Suppressed cardiac hypertrophy and improve survival rate.	Variety of cancers.	99
Valpronic acid	HDACi for class I and II.	Suppressed cardiac hypertrophy and improved survival rate.	Variety of cancers.	108
Vorinostat (SAHA)	HDACi for class I and II.	Reduced infarct size and improved cardiac function	Cutaneous T-cell lymphoma.	102, 103

5-aza = 5-azacytidine; ALL = acute lymphoblastic leukemia; AML = acute myeloid leukemia; BET = bromodomain; EMT = epithelia-mesenchymal transition; HDACi = histone deacetylase inhibitor; HMTi = histone methyltransferase inhibitor; MTi = methylation inhibition; NR = no report; OA = osteoarthritis; RA = rheumatoid arthritis; SAHA = suberoylanilide hydroxamic acid. Other abbreviations as in Table 1.

Another potential target of epigenetic molecules is HDAC. There are many HDAC inhibitors (HDACi) and animal testing demonstrated the beneficial effects of HDACi treatments in preventing or improving cardiac functions. Treatment of either trichostatin A and scriptaid, 2 broad-spectrum HDACis, in a pressure-overload hypertrophy mouse model revealed that cardiac hypertrophy and fibrosis were suppressed, and the hypertrophy-associated switch of adult and fetal isoforms of myosin heavy chain expression was diminished (99). In addition, the same HDAC inhibitors were shown to reduce myocardial infarct size and to preserve systolic function in a Langendorff perfusion in vitro model as well as in a mouse ischemia/reperfusion model 100, 101. Using mouse and rabbit ischemic/reperfusion models, Xie et al. showed that suberoylanilide hydroxamic acid (SAHA, also known as Vorinostat), a Food and Drug Administration–approved HDACi for cancer treatment (102), also can preserve the systolic function and reduce infarct size (103). Additionally, using a simulated ischemic/reperfusion neonatal rat ventricular myocyte culture system, Xie et al. further demonstrated that the cardioprotective effects are partly due to induction of cardiomyocyte autophagy (103). Givinostat, another pan-HDACi, which is used in treating Duchenne muscular dystrophy, inflammatory diseases, myelofibrosis, and blood cancers 104, 105, 106, has been shown to attenuate inflammatory response and angiogenic effects, reduce endothelial-to-mesenchymal transition, and reduce cardiac fibrosis on an acute MI mouse model (107). Moreover, when using SK-7041, a Class I HDACi, in either angiotensin II infusion or TAC cardiac hypertrophy mouse modes, Kee et al. demonstrated that myocardial hypertrophy was reduced and the survival rate was improved in mouse and rat hypertrophy models (108). Using an in vitro screening assay, Gallo et al. identified a truncated form of apicdin, another class I HDACi, and this HDACi also demonstrated similar results as what Kee et al. found (109). These results suggested that cardioprotective effects were mainly mediated by class I HDACs.

Many small molecules for inhibition of HDAC have been developed and tested to treat cardiac diseases. However, unlike HDAC, only a few inhibitors of HAT currently show a potential protective effect against cardiac disease. One of them is curcumin, which was first used as an anti-inflammatory reagent. Currently, curcumin has been suggested for treating osteoarthritis, rheumatoid arthritis, and diabetes 110, 111, 112. It is a natural compound derived from the active ingredient of turmeric (Curcuma longa), which can act as a p300-specific histone HAT inhibitor. Oral treatment with curcumin in a rat MI model demonstrated improvement in left ventricular function and reduction of fibrosis compared with non-treatment MI rats (24).

A DNA methylation inhibitor, 5-azacytidine (5-aza), is widely used to treat leukemia and sickle cell diseases 113, 114. In a spontaneously hypertensive rat model, intraperitoneal injection of 5-aza at a dose of 10 mg/kg resulted in improvement in cardiac function as well as reduction in cardiac fibrosis (115).

Last but not least, lncRNA silencing approach using GapmeR, a type of antisense oligonucleotide, is another potential therapeutic target for HF (116). Although no clinical trials targeting lncRNAs were performed, several in vivo studies have shown that injection of lncRNA GapmeR such as GapmeR-Chast preserved or improved cardiac function in mice after pressure overload-induced cardiac hypertrophy and HF 74, 78.

Conclusions and Future Perspectives

This review summarizes recent studies that supported key contributing roles of epigenetic mechanisms in cardiac hypertrophy and failure and may have important therapeutic implications for HF (Figure 1, Central Illustartion). However, the ubiquitous characteristics of epigenetic regulators have created a significant challenge to develop targeted epigenetic therapies for cardiac diseases. First, because most epigenetic regulators modulate gene expression via conventional mechanisms, the global inhibition or activation of such regulators may have pleiotropic effects with no cardiac specificity. Second, the heart is made up of several different types of cells. Each cell type may possess its own epigenetic signature in response to disease conditions. Before generalizing such a knowledge base for cardiac therapies, larger cohort studies with cell-specific epigenetic maps for various types of cardiomyopathy are necessary. Equally important is the need to identify cardiac disease–specific genes and their corresponding epigenetic regulators and to establish networks between the epigenetic alterations and the signaling transduction cascades (specifically to cardiac hypertrophy and failing hearts). Another challenge of epigenetic therapy for HF (as is true for other diseases) is to minimize the off-target effects by developing suitable and efficient methods for delivering to the specific locations within a heart. Lastly, a potential near-term application of the current knowledge base of epigenetic regulation for cardiac diseases is the development of epigenetic biomarkers for diagnosis, prognosis, and therapy optimization, especially from easily accessible tissues (such as peripheral blood or blood cells) that may reflect cardiac-specific alterations. Generating such insights into epigenetic modifications along the natural history of human HF will catalyze the development of personalized therapies for cardiomyopathy and HF.

Figure 1

The Epigenetic Mechanisms in Heart Failure

Epigenetic regulations include DNA methylation, PTMs of histones, ATP-dependent chromatin conformation and remodeling, and non-coding RNA-mediated regulation. DNA methylated CpG-enriched region at the promoter region of genes is often associated with gene silencing, whereas methylated CpG found in the gene body is usually related to gene activation. Histone modifications associated with gene expression normally occur on histone 3. The histone writers (HAT and HMT) and erasers (HDAC and HMT) are responsible for PTMs of histone. The active histone mark, H3K27ac and repressive marks, H3K9me2 and H3k27me3, and promoter mark, H3K4me3, are shown. LncRNA can work with epigenetic regulators to effect activation or repression on chromosome remodeling and accessibility as well as the mRNA stability. In the non-failing heart, normal adult cardiac genes, adult isoform MYH6, express normally in cardiomyocyte. The cardiac hypertrophy genes, such as NPPA and CTGF, are not expressed or are at a basal level in non-failing hearts. The DNA around the promoter region of the hypertrophic-related gene loci is hypermethylated, and the chromatin state is inactive by marking H3K27me3 and/or H3K9me2. The PRC2/EZH2 and G9a-EZH2-MEF2C complexes involved in histone methylation in normal hearts are shown. It was shown that lncRNA, Mhrt, plays a critical role in maintaining MYH6 expression in the adult heart by preventing BRG1 binding to the promoter region of MYH6. However, in failing hearts, in addition to normal cardiac gene expression, the hearts return to the fetal gene program and express the fetal isoform of myosin heavy chain, MYH7. This is likely due to the re-activation of BRG1 during heart failure. The potential mechanisms are shown. Two cardiac transcription factors, GATA4, MEF2C, and CBP/p300 (HAT), and JMJD (HMT) are involved in cardiac hypertrophy and heart failure. Two examples of the actions of lncRNAs, CHRT and Chaer, involved in cardiac hypertrophy and heart failure are presented. (Please see text for details.) ATP = adenosine triphosphate; CpG = cytosine-phosphate-guanine ; HAT = histone acetyltransferase; HDAC = histone deacetylase; HMT = histone methyltransferase; mRNA = messenger RNA; MYH6 = myosin heavy chain; MYH7 = myosin heavy chain beta; PTM = post-translational modifications.

Central Illustration

The Epigenetic Mechanisms in Heart Failure