The ATP-dependent chromatin-remodeling complexes do not perform covalent modifications of the DNA or histones but rather use energy derived from ATP hydrolysis to move, destabilize, eject or restructure nucleosomes. There are four different families of ATP-dependent chromatin remodeling complexes: switching defective/sucrose non-fermenting complexes (SWI/SNF); imitation switch complexes (ISWI); chromodomain, helicase, DNA binding complexes (CHD) and inositol-requiring 80 complexes (INO80; Li, 2002; Yang and Seto, 2008; Clapier and Cairns, 2009; Ho and Crabtree, 2010). Although all members of each family have distinct flanking domains, they all share an evolutionarily conserved SWI-like ATPase catalytic domain that serves as vehicle to adjust histone-DNA contacts for DNA movement and chromatin restructuring. In turn, the other domains act in the recognition of covalently modified histones, modulation of ATPase activity and/or interaction with other chromatin and transcription factors. Consequently, these unique domains and their associated proteins determine the genomic targeting specificity and biological functions of each family of chromatin remodelers. In fact, chromatin modification through ATP-dependent enzymes is associated with regulation of expression of distinct gene programs in organ development and adaptation (Ho and Crabtree, 2010).

DNA methylation is the most common epigenetic modification in the mammalian genome. This long-term stable epigenetic modulation involves the addition of a methyl group to the 5′ carbon of a cytosine by DNMT enzymes (Figure 1) and mostly occurs at the CpG (cytosine preceding guanosine) dinucleotide sequences, also known as CpG islands, in the mammalian genome (Feinberg, 2008). CpG islands, in contrast to the remainder genome, are cytosine-guanosine-rich sequences (CpG-rich), generally not methylated (Deaton and Bird, 2011), and mostly acting as sites of transcription initiation once they are associated with promoter regions of genes (~70% of gene promoters; Li et al., 1993; Saxonov et al., 2006). DNA methylation is known to be catalyzed by three different DNMTs: DNMT1, DNMT3a and DNMT3b (Broadbent et al., 2008), where DNMT1 is the core enzyme in mammals. Methylation of DNA is considered a maintenance function of DNMTs as it results in post-replicative restoration of hemi-methylated sites to full methylation (Laird, 2003). Reduction of DNMT1 activity may result in demethylation and recent studies even showed that this is an active process (Bhutani et al., 2011). However this has not been shown yet for the cardiovascular system.

DNA methylation is, generally, attributed to gene silencing by hampering the accessibility of cis-DNA binding elements present in the promoter regions of genes of the transcriptional machinery (Suzuki and Bird, 2008) and plays a crucial role in the regulation of chromatin structure including X chromosome inactivation, genomic imprinting, silencing of repetitive DNA elements and transposon transcription (Li et al., 1993; Panning and Jaenisch, 1996; Li, 2002). Moreover, DNA methylation has been linked to biological processes underlying various diseases from cancer (Feinberg and Tycko, 2004) to CVD, such as hypertension (Millis, 2011), diabetes (Ling and Groop, 2009; MacFarlane et al., 2009), atherosclerosis and inflammation (Wierda et al., 2010).

The eukaryotic DNA is tightly compact and organized in chromatin. The nucleosome is the central unit of chromatin and is composed of an octomer center of two copies of each histone protein (H2A, H2B, H3, and H4; Jenuwein and Allis, 2001) around which a DNA segment of 14–150 base pairs is looped. Each histone has an amino-terminal tail that protrudes from the surface of the nucleosome and which can be subjected to various posttranscriptional modifications such as phosphorylation, sumoylation, ubiquitination, methylation, ADP-ribosylation, proline isomerization, deimination and acetylation (Handy et al., 2011). These modifications lead to conformational changes in the chromatin resulting in altered gene expression (Margueron and Reinberg, 2010) depending on whether DNA becomes accessible (euchromatin) or inaccessible (heterochromatin) for transcription (Figure 1).

Histone acetylation occurs at the lysine residues of the histone tails resulting in de-condensation of the chromatin structure and serving as a binding site for bromodomain proteins and transcriptional activators, and eventually leading to transcriptional activation (Elliott and McKenna, 2004; Rodriguez et al., 2009). Conversely, histone deacetylation induces chromatin condensation and therefore transcriptional repression (Clayton et al., 2006; Shahbazian and Grunstein, 2007; Figure 1). Acetylation of histones is a dynamic process mediated by two counteracting enzyme families, the histone acetyltransferases (HATs) and histone deacetylases (HDACs). The harmony between the activities of these two sets of enzymes is a crucial element during regulation of gene expression and its deregulation is linked to several pathological conditions varying from cancer to CVD (Ordovas and Smith, 2010; Burgess, 2012).

Other key modulator of posttranslational regulation is histone methylation which can occur on all basic amino acid residues of the histone tail; arginines, lysine and histidines (Cheung and Lau, 2005). In addition, different amino acids can be methylated to a different extent and while lysine can be subjected to mono-, di- and trimethylation, arginine residues can only become mono- or dimethylated (Cheung and Lau, 2005). Methylation of histones is a dynamic process mediated by histone methyltransferases (HMTs) and histone demethylases (HDMs; Teperino et al., 2010) and, unlike acetylation, histone methylation can induce either activation or repression of gene expression depending on the target sites and degree of methylation (Lachner and Jenuwein, 2002; Figure 1). In contrast to histone acetylation, histone methylation governed mainly by HMTs SUV39H1 and G9a (Martin and Zhang, 2005; Shi and Whetstine, 2007), has long been considered to be a permanent epigenetic mark (Jenuwein and Allis, 2001). However, the discovery of new players such as HDMs has shifted the paradigm and, in fact, several studies showed that histone methylation is tightly regulated in inflammatory and metabolic disorders (Saccani and Natoli, 2002; Villeneuve et al., 2008; Brasacchio et al., 2009).

It is now proven and accepted that the majority of the genomic DNA is transcribed as non-coding RNAs and that such RNA species play pivotal regulatory roles during development (Sayed and Abdellatif, 2011), in response to environmental adversity (Ferguson, 2011), and at the onset and progression of disease (Sayed and Abdellatif, 2011). In this context, many studies were directed at revealing the role of non-coding RNAs in physiological and pathological processes.

There are two main classes of non-coding RNAs: infrastructural (small nuclear and nucleolar RNAs, ribosomal RNAs) and regulatory RNAs (miRNAs, long non-coding RNAs, small interfering RNAs and Piwi-interacting RNAs). To date, only miRNAs have been associated with epigenetic regulatory mechanisms in HF. Epigenetic regulation through long non-coding RNAs have been extensively studied in cancer but have also been associated with cardiovascular disease, mainly in maintenance of vascular homeostasis (Robb et al., 2004; Li et al., 2010).

MicroRNAs were first described in the nematode Caenorhabditis elegans, in the early 1990s (Lee et al., 1993). From then on, a multitude of miRNAs have been identified and investigated, and presently there are ~1600 human miRNA sequences annotated at miRBase19 (Kozomara and Griffiths-Jones, 2011).

MiRNAs are transcribed as primary transcripts (pri-miRNA) from intergenic, intronic or exonic regions in the genome, by RNA polymerase II. These pri-miRNAs fold into an hairpin shape with a five prime (5′) capped (mGpppG) and a polyadenylated tail which is subsequently cleaved by an enzyme complex composed of the RNase III endonuclease Drosha and the dsRNA binding protein Pasha (also known as DiGeorge critical region 8; DGCR8; Lee et al., 2003, 2004). The resulting shorter (70–100 nucleotide in length) hairpin-shaped precursor miRNA (pre-miRNA) is transported from the nucleus into the cytoplasm by Ran-GTP and exportin-5 (Kim, 2004). In the cytoplasm, pre-miRNAs are further processed by a RNase III enzyme, Dicer, into a short (20–25 nucleotides in length) transient double stranded RNA molecule. At this stage, the formed mature RNA molecule is included in a protein complex – the so-called RNA-inducing silencing complex (RISC), while the passenger strand is degraded (Winter et al., 2009). The RISC-miRNA complex specifically targets mRNA sequences leading to negative regulation of protein synthesis or mRNA degradation (Winter et al., 2009). One miRNA can regulate a vast number of mRNAs simultaneously (Lewis et al., 2005) by predominantly acting through destabilization of target mRNAs and subsequently leading to reduced protein output (Guo et al., 2010). Therefore, decreased protein production can result from a combination of mRNA destabilization and translational inhibition. MiRNAs have been shown to be involved in different pathological processes such as cancer and CVD (Lujambio and Lowe, 2012; Quiat and Olson, 2013). While in cancer epigenetic mechanisms have been widely associated with silencing of miRNA-encoding genes and thus recognized to greatly influence the expression of genetic information, only recently the importance of such mechanisms have started to be addressed in CVD, and more specifically in HF.

Postural tachycardia syndrome (POTS) has multiple symptoms, one of such being tachycardia. Dysfunction of the norepinephrine transporter (NET) gene has previously been implicated in POTS, with a reported coding mutation in the NET gene (SLC6A2; Bayles et al., 2012). Head-up tilt experiments in POTS patients and showed that the expression of norepinephrine transported is lower in POTS patients compared to healthy subjects. In the absence of altered SLC6A2 gene sequence or promoter methylation, the observed reduced expression of norepinephrine was directly correlated with chromatin modifications. Changes in expression were attributable to increased binding of the repressive methyl CpG-binding protein 2 (MeCP2) regulatory complex, in association with an altered histone modification composition at the promoter region of the SLC6A2 gene (Bayles et al., 2012).

The KCNQ1 gene is located on chromosome 11 in a region that contains a cluster of 6 genes that are expressed from either only the maternal or the paternal allele. In mice, the KCNQ1 overlapping transcript (KCNQ1ot1) is transcribed from a promoter located in intron 10 of the KCNQ1 gene. This promoter region is a CpG island and undergoes methylation on the maternal chromosome, preventing transcription, and therefore allowing expression of the gene cluster. However, this promoter region is not methylated on the paternal chromosome allowing expression of the regulatory transcript and suppressing the expression of the gene cluster (Mancini-DiNardo et al., 2003). The maternal allele is transcribed in early embryogenesis with the paternal allele being progressively methylated and therefore only activated during late embryogenesis.

Variable imprinting of the KCNQ1 gene provides a possible explanation for the existence of long QT syndrome (LQTS) in the absence of a coding sequence mutation in KCNQ1. Paternal imprinting is probably relieved in cardiac tissue, meaning that during differentiation methylation of the paternal chromosome must occur to block production of the suppressive KCNQot1 transcript. Mutations that disrupt the CpG island could prevent methylation and silence the paternal allele in the heart (Mancini-DiNardo et al., 2003; Bokil et al., 2010). A more recent study by Fatima et al. (2012) associates epigenetic modifications with regulation of the ATP sensitive potassium channel (KATP). In cardiac myocytes, different isoform combinations of the SURx (SUR1, SUR2) and Kir6.2 (KCNJ11) will be responsible for distinct physiological and pharmacological properties, depending on the isoforms expressed. Promotor CpG methylation appears to be one of the regulators of SURx isoform expression and therefore, regulated or aberrant CpG methylation might play a role in controlling channel structure and function under different conditions (Fatima et al., 2012).

Several studies have been conducted to look at the association between miRNAs and arrhythmias. MiRNA expression profiles were shown to differ in right atrial disease, with 47 miRNAs being differentially expressed between disease and control states, whereas similar changes in expression could not be found in left atrial disease (Cooley et al., 2012; Kim, 2013). In a different study, miRNAs that were differentially expressed between AF and sinus rhythm in patients with mitral stenosis were showed by microarrays (Xiao et al., 2011; Kim, 2013). These data indicate that miRNAs play a role in regulating cardiac conduction and in the induction of arrhythmias.

Multiple studies have shown that miR-208a plays an important role in action potential conduction. Overexpression of miR-208a leads to arrhythmia, cardiac fibrosis and hypertrophy, and is a strong predictor of cardiac death (Oliveira-Carvalho et al., 2013). Genetic deletion of miR-208a, on the other hand, also leads to an increased risk of AF and other arrhythmias, due to aberrant conduction mainly caused by dysregulation of Cx40 (Callis et al., 2009; Oliveira-Carvalho et al., 2013). Similarly, also miR-328 is upregulated not only in animal models of AF but also in human tissue samples from AF patients. Overexpression of miR-328 in mice increased vulnerability to AF as confirmed by diminished L-type Ca²⁺ current and shortened atrial APD. AF vulnerability could be reversed by concomitant inhibition of the miRNA by an antagomir (Lu et al., 2010; Kim, 2013).

The most well established cardiac conduction-related miRNA is by far, miR-1. This miRNA plays a role in myotonic dystrophy, a disease where degeneration of the conduction system occurs and increased CACNA1C (CAV 1.2) expression, a cardiac L-type Ca²⁺ channel gene, is observed (Kim, 2013). The involvement of miR-1 in electrocardiophysiology was further confirmed by a targeted deletion of miR-1-2 by Zhao et al. (2007), which lead to a high rate of sudden death, caused by conduction blockade due to direct targeting of Irx5, a transcription factor that regulates cardiac repolarization. In rats, induction of MI by occlusion of the left anterior descending artery results in miR-1 upregulation and arrhythmia exacerbation but treating the animals with an antisense inhibitor could abrogate these effects. Furthermore, miR-1 also directly targets KCNJ2, which encodes for the calcium channel subunit Kir 2.1, providing a possible mechanism for increase of arrhythmias in MI (Yang et al., 2007). The role of miR-1 in arrhythmogenesis was further confirmed in humans where atrial cells from AF patients display a 86% decrease in miR-1expression, a subsequent increased Kir 2.1 protein expression and an increase in I_k1 density (Girmatsion et al., 2009; Kim, 2013). MiR-1 is also involved in cardiac electrical remodeling in viral myocarditis where it is upregulated, resulting in decreased Cx43, which is required for transfer of electric activation, and indicating that miR-1 plays a role in intercellular communication.

Another prominent miRNA in the regulation of cardiac conduction is miR-133. Matkovich et al. (2010) showed that an increase in miR-133a leads to prolonged QT intervals. This miRNA is highly and preferentially expressed in cardiac and skeletal muscle and is known to regulate cardiac ion channels such as Kv4- encoded I_to,f (Kcnip2; Matkovich et al., 2010; Kim, 2013). Furthermore, the catalytic and regulatory subunits of protein phosphatase 2A (PP2A) are decreased in cardiomyocytes in chronic HF and were shown to be targets of both miR-1 and miR-133. Because pharmacologic inhibition of PP2A leads to altered diastolic Ca²⁺ waves this indicates a role for these two miRNAs in calcium handling (Belevych et al., 2011; Kim, 2013).

Interestingly, a relation between nicotine abuse and cardiac arrhythmias has been suggested by several studies. Nicotine treatment of canine atrial fibroblasts, resulted in upregulation of transforming growth factor beta-1 (TGF-β1) and TGF- beta receptor type II levels (TGF-βRII), with concomitant decreased levels of miR-133 and miR-590, both directly targeting TGF-β1 and TGF-βRII. This effect was abolished by synthetic downregulation of both miRNAs (Shan et al., 2009; Kim, 2013).

Apart from miR-1 and miR-133, there are several other miRNAs that have been associated with regulation of cardiac conduction to some extent. This is the case for miR-212 that seems to regulate inward rectifier K⁺ current density by targeting Kir 2.1 (Kim, 2013), and for miR-21 which is increased in the left atria of patients with AF and which abrogation leads to repression of atrial fibrosis and AF (Adam et al., 2012; Cardin et al., 2012; Kim, 2013). Furthermore, conditional overexpression of miR-17-92 in cardiac and smooth muscle tissue results in both dilated, HCM as well as in arrhythmias. An increase in atrial and ventricular ectopy, as well as increased susceptibility to arrhythmia was observed in homozygous and heterozygous animals. After programmed electrical stimulation all transgenic animals developed sustained and lethal ventricular tachycardia (VT) or ventricular fibrillation (VF) and these effects were mainly caused by dysregulation of two downstream targets of miR-17-92, the lipid phosphatase and tensin homolog PTEN and Cx43 (Danielson et al., 2013). Likewise, also miR-155 and miR-181 have been associated with cardiac conduction defects. Circulating levels of miR-155 are upregulated in patients with specific angiotensin receptor type 1 (AT1R) polymorphisms that have been shown to be associated with an increased risk of ventricular tachyarrhythmias and sudden death (Blanco et al., 2012). In turn, miR-181a seems to play a role in VT after MI (Li et al., 2009). Altogether, the data available regarding the relation between miRNAs and arrhythmias establish miRNAs as crucial players in regulating cardiac electrophysiology and electric potential conduction through an array of different mechanisms.

ATP-dependent chromatin remodeling complexes play crucial roles in vertebrates, mainly in organ development and adaptation. Most of them have been associated with heart development and only a few were implicated in heart disease. The BAF (brahma-associated factor) complex is the vertebrate homolog of the yeast SWI/SNF family of chromatin remodelers. In mammals, this complex contains 12 protein components from which an ATPase subunit encoded by either Brm (brahma) or Brg1 (brahma-related gene 1). These two subunits, although highly homologous, exhibit distinctive functions in vivo. While several studies have demonstrated that individual subunits of the BAF complex are essential during heart development (Lickert et al., 2004; Takeuchi et al., 2007; Takeuchi and Bruneau, 2009) and may be implicated in human congenital diseases (Kitagawa et al., 2003; Bajpai et al., 2010), BRG1 was recently involved in cardiac disease (Hang et al., 2010). In embryos, Brg1 promotes myocyte proliferation and it preserves fetal cardiac differentiation by interacting with HDACs and poly (ADP ribose) polymerase (PARP) to repress α-MHC to β-MHC shift. Brg1 (also known as Smarca4) is not expressed in the adult heart but it is reactivated by stress conditions such as pressure overload. Once reactivated, Brg1 forms a complex with its embryonic partners (HDAC and PARP), to induce the pathologic α-MHC to β-MHC shift. Adult myocardial gene deletion of Brg1 inhibited cardiac hypertrophic growth and reversed the MHC switch. Accordingly, Brg1 is activated in patients with HCM, correlating with disease severity and MHC changes (Hang et al., 2010). PPAR is a nuclear enzyme known to respond to DNA damage and facilitate repair. Besides DNA repair, PPAR-1 also modulates chromatin to control the transcriptional machinery in response to diverse stimuli. Such stimuli induce PPAR activation and PAR-dependent striping of histones from chromatin, thereby favoring the opening of chromatin to allow transcriptional regulation (Tulin and Spradling, 2003; Kim et al., 2004). PARP is activated in cardiac hypertrophy and its activity is increased in murine and human failing hearts (Pillai et al., 2005b; Xiao et al., 2005). Deletion of PARP-1 in mice or pharmacological inhibition of PARP activity decreases cardiac hypertrophy induced by angiotensin II (Pillai et al., 2006) or pressure overload (Pillai et al., 2005a; Xiao et al., 2005), delays the progression from hypertensive cardiomyopathy to HF (Bartha et al., 2009), decreases cell death and HF after MI (Palfi et al., 2006) and diminishes myocardial ischemia/reperfusion injury (Szabo et al., 2002).

Although very preliminary, there seems to be a link, at the chromatin level, between fetal hearts and adult diseased hearts, and in the future, targeting the regulation of chromatin remodeling processes may become a promising approach to prevent or maybe even reverse pathological cardiac hypertrophic growth and HF.

Unlike in other diseases such as cancer, the role of DNA methylation in HF remains elusive. Movassagh et al. (2010) compared genome-wide methylation profiles of left ventricle tissue from HF patients and healthy controls by methylated DNA immunoprecipitation-chip (MeDIP-chip), in which immunoprecipitation of methylated DNA is followed by microarray hybridization and further validated by bisulfite sequencing. As a result, three differentially methylated angiogenesis-related loci have been identified and correlated to differential expression levels of the corresponding gene (Movassagh et al., 2010). Hyper-methylation of the 5′ regulatory region of platelet endothelial cell adhesion molecule 1 (PECAM-1) and hypo-methylation of the angiomotin-like protein 2 (AMOTL2) in failing hearts correlated with reduced expression of those genes, while hyper-methylation within the Rho GTPase activating protein 24 gene (ARHGAP24) is correlated with increased expression of ARHGAP24 in failing hearts (Movassagh et al., 2010). Moreover, a follow up study (Movassagh et al., 2011a) generated a genome-wide DNA methylation map of human hearts and revealed a significant decrease in global promoter methylation of genes with increased expression in failing hearts (Movassagh et al., 2011a). The genome-wide methylation profile of patients with idiopathic dilated cardiomyopathy was recently generated (Haas et al., 2013). In an attempt to validate the methylation profiling of the 20 most differentially methylated genes, MassARRAY and Bisulfite sequencing were used in a large independent cohort (30 patients; Baccarelli and Bollati, 2009). Interestingly, 12 (out of 20) genes showed similar methylation patterns between the two independent studies. Such approach allowed the identification of two novel genes with differential methylation profiles between patient and control subjects, lymphocyte antigen 75 (ly75) and adenosine A2a receptor (adora2a). Curiously, downregulation of those genes in zebrafish by using specific morpholino technology resulted in reduced ventricular contractility and HF (Haas et al., 2013). More recently, DNA methylation was found to be responsible for the hypermutability of distinct cardiac genes. This is the case for the cardiac isoform of the myosin binding protein C gene (Mybpc3) that has a significantly higher level of exonic methylation of CpG sites than the skeletal isoform (Mybpc2; Meurs and Kuan, 2011). This suggests that there are unique aspects of the Mybpc3 gene or its epigenetic environment that are prone to generate genetic mutations.

Very recently, a report in the Journal of the American Heart Association (Bellavia et al., 2013) provided evidence for the effects of ambient particulate-matter (PM) on blood pressure (BP). In humans, exposure to fine and coarse concentrated ambient particles (CAPs) induce blood hypomethylation of Alu, a transposable repeated element, and Toll-like receptor 4 (TLR4). Hypomethylation of both factors was found to be associated with increased systolic BP after exposure. This is of great interest since many epidemiological studies (O𠆝Toole et al., 2008; Brook et al., 2010) have reported a correlation between PM exposure, cardiovascular disease and death, and may, therefore, represent a novel mechanism that mediates environmental effects on BP and indirectly cardiovascular disease and HF.