Leveraging clinical epigenetics in heart failure with preserved ejection fraction: a call for individualized therapies

Abstract

Described as the ‘single largest unmet need in cardiovascular medicine’, heart failure with preserved ejection fraction (HFpEF) remains an untreatable disease currently representing 65% of new heart failure diagnoses. HFpEF is more frequent among women and associates with a poor prognosis and unsustainable healthcare costs. Moreover, the variability in HFpEF phenotypes amplifies complexity and difficulties in the approach. In this perspective, unveiling novel molecular targets is imperative. Epigenetic modifications—defined as changes of DNA, histones, and non-coding RNAs (ncRNAs)—represent a molecular framework through which the environment modulates gene expression. Epigenetic signals acquired over the lifetime lead to chromatin remodelling and affect transcriptional programmes underlying oxidative stress, inflammation, dysmetabolism, and maladaptive left ventricular remodelling, all conditions predisposing to HFpEF. The strong involvement of epigenetic signalling in this setting makes the epigenetic information relevant for diagnostic and therapeutic purposes in patients with HFpEF. The recent advances in high-throughput sequencing, computational epigenetics, and machine learning have enabled the identification of reliable epigenetic biomarkers in cardiovascular patients. Contrary to genetic tools, epigenetic biomarkers mirror the contribution of environmental cues and lifestyle changes and their reversible nature offers a promising opportunity to monitor disease states. The growing understanding of chromatin and ncRNAs biology has led to the development of several Food and Drug Administration approved ‘epidrugs’ (chromatin modifiers, mimics, anti-miRs) able to prevent transcriptional alterations underpinning left ventricular remodelling and HFpEF. In the present review, we discuss the importance of clinical epigenetics as a new tool to be employed for a personalized management of HFpEF.

Graphical abstract

Role of environmental factors and epigenetic processing in the pathogenesis of HFpEF.

Heart failure with preserved ejection fraction

Heart failure (HF) is the most common cause of hospitalization in the Western societies. While the incidence of HF with reduced ejection fraction (HFrEF) has declined over the last decade, the prevalence of HF with preserved ejection fraction (HFpEF) is on the rise, and such increase seems to reflect the current pandemic of cardiometabolic disorders such as obesity, metabolic syndrome, type 2 diabetes, and arterial hypertension.^{1  ,  2}

One of the main features of HFpEF is the marked increase in left ventricular (LV) end-diastolic pressure and pulmonary venous pressure after small changes in LV end-diastolic volumes, which frequently leads to exertional dyspnoea.³ Diagnosis of HFpEF remains challenging and requires signs or symptoms of congestion, preserved or mildly abnormal LV systolic function (ejection fraction >50%, LV end-diastolic volume index <97 mL/m²), and diastolic dysfunction.⁴ Heart failure with preserved ejection fraction associates with cardiac and non-cardiac comorbidities, which synergistically result in abnormalities of ventricular and vascular structure and function.⁵ Comorbidities drive chronic systemic inflammation and increased oxidative stress, thus triggering endothelial and cardiomyocyte dysfunction as well as extracellular matrix (ECM) remodelling⁶ (Figure 1). Systemic metabolic inflammation, also accompanied by an increased activity of inducible nitric oxide synthase and augmented nitrosative stress, plays an important role in the pathophysiology of obesity-associated HFpEF.^{6  ,  7} Since increased inflammation is linked with the development of HF,⁸ it is not surprising that interventions that reduce inflammation are being explored as potential treatment in HF. With an annual mortality rate of 22%, the prognosis of HFpEF is as grim as the prognosis of HFrEF and its mortality rate is higher than most forms of cancer.⁹ Nevertheless, evidence-based therapeutic strategies are still lacking and HFpEF treatment remains largely empirical.² Indeed, drugs proven to be beneficial in HFrEF have failed so far in patients with HFpEF and this could be related to a poor understanding of the underlying molecular and cellular mechanisms that distinguish HFpEF from HFrEF.^{1  ,  2} This suggests that a ‘one size fits all’ strategy may be ill-suited to HFpEF and rather supports the use of personalized interventions specifically tackling comorbidity-related HFpEF phenotypes and gender-based differences.

Figure 1.

Main risk factors and molecular alterations underlying HFpEF. The clustering of risk factors precipitates metabolic abnormalities, inflammation, and oxidative stress leading to structural and functional alterations underlying HFpEF. Accumulation of free radicals and inflammatory molecules in different cardiac cells affect cardiac remodelling and function at several levels, including microvascular function, cardiomyocyte stiffness, and fibrosis. eNOS, endothelial nitric oxide synthase; ICAM-1, intercellular adhesion molecule 1; NO, nitric oxide; ROS, reactive oxygen species; TNFα, tumour necrosis factor alpha; VCAM-1, vascular cell adhesion molecule 1.

Structural and functional changes in heart failure with preserved ejection fraction

Regardless of ejection fraction, cardiac remodelling in human and experimental models of HF is characterized by impaired cardiomyocyte contraction and relaxation, as well as myocardial hypertrophy and accumulation of collagen in the ECM^{10  ,  11} (Figure 1). Structural and functional adaptation of the left ventricle suffering either chronic pressure overload or acute myocardial infarction has a major role in HF pathophysiology. Extracellular matrix changes are mainly determined by collagen and consist of an increased collagen volume fraction, a relative abundance of the stiff collagen type 1 and more collagen crosslinking.^{12  ,  13} Any of these properties can potentially be altered in heart disease and modify diastolic stiffness.^{11  ,  14}

Diastolic function is often conceptualized as the entirety of an active process that comprises mid-ventricular ejection, pressure decline to early filling,¹⁵ which relates to myofilament dissociation and calcium re-uptake, and to the passive properties of the myocardial wall, including cardiomyocyte stiffness, ECM changes, chamber geometry, and the pericardium.¹⁶ Cardiomyocyte diameters have also been observed to be larger in HFpEF than in other phenotypes.¹⁰ On top of the observed structural changes, cardiomyocyte function also significantly contributes to HFpEF pathophysiology,^{7  ,  17–19} which is mainly attributed to the titin filament network of the cardiomyocytes that modulates myocardial diastolic stiffness. As the most crucial factor contributing to LV diastolic stiffness, titin is responsible for the passive elasticity of cardiac muscle through isoform switching or post-translational modifications (PTMs) such as phosphorylation and oxidation. Pathological alterations in HFpEF do not occur only in the myocardium but also in the surrounding adipose tissue. Accumulation of epicardial adipose tissue (EAT) in cardiometabolic patients is indeed associated with more profound haemodynamic derangements at rest and exercise, and poorer exercise capacity.²⁰ Epicardial adipose tissue can affect cardiac function mainly via two mechanisms: (i) a mechanical obstacle to LV filling (pericardial restraint); and (ii) secretion of pro-oxidant (H₂O₂) and pro-inflammatory mediators [interleukin (IL)-6, IL-1β, tumour necrosis factor (TNF)-α] acting on the surrounding myocardium and leading to ECM remodelling, hypertrophy, and defective autophagic flux.²¹

Molecular fingerprints of heart failure with preserved ejection fraction

Emerging evidence indicates a key role of oxidative stress in HFpEF-associated hypertrophy, fibrosis, and myocardial stiffness^{11  ,  22–26} (Figure 1). Oxidative stress can damage cells, proteins, and DNA, thus contributing to premature senescence, deregulation of cytosolic heat shock response, mitochondrial and endoplasmic reticulum unfolded protein response, the ubiquitin-proteasome system, and autophagy.¹⁹ In healthy and diseased hearts, the protein quality control (PQC) system and its underlying pathways (chaperones and co-chaperones) play an essential role as key facilitators of protein folding and function. Protein quality control targets unfolded/misfolded or aggregate substrates for degradation, e.g. via the ubiquitin-proteasome system or autophagy. The constant mechanical and metabolic stress in cardiomyocytes places a high demand on PQC eventually leading to alterations of molecular chaperones and co-chaperones, accumulation of misfolded protein, cardiac dysfunction, and HF.^{22  ,  25  ,  27} Of note, overexpression of various chaperones, such as heat shock proteins (HSPs), results in significantly reduced aberrant protein aggregation in desminopathy and αB-crystallinopathy^27–29 and can ameliorate pathological protein aggregation and improve muscle function.^{25  ,  29} Previous work demonstrated that high cardiomyocyte stiffness in HFpEF human myocardium is corrected by α-B crystallin, probably through relief of titin aggregation.³⁰ α-B crystallin and HSP27 can translocate to the titin spring segment of stressed muscle tissue, including myopathic muscle, to protect unfolded titin domains from aggregation^{25  ,  31} and prevents a pathological increase in titin stiffness.^{25  ,  30} Therefore, titin oxidation can alter diastolic stiffness resulted from titin damage by oxidative or physical stress.^{22  ,  27} Other molecular chaperones are induced by various forms of stress such as heat and nutrient deprivation, inflammation, and other disease processes to protect for example against stimulated ischaemia and metabolic stress,³² attenuate the conjugation of HSPs, decrease macrophage infiltration and the inflammatory cytokine expression,³³ resulting in an attenuation of pressure overload-induced cardiac hypertrophy and fibrosis.¹⁹ Heat shock proteins counter this risk by covering the displayed hydrophobic domains of unfolding proteins until cellular conditions return to normal. Therefore, pathological myocardial stress, including oxidative stress and inflammation, increases the amount of misfolded protein that needs to be refolded to a native state or removed to avoid protein aggregation. To counteract this stress and the accumulation of misfolded protein, one response is the rapid induction of HSP expression. Hence, the current lack of understanding HFpEF stems also from a poor characterization of the relation with oxidative stress in HFpEF, and any beneficial effects in terms of therapy will require targeting this upstream modulator of this pathway. Therefore, deepening our basic understanding of the pathophysiology of disease associated with oxidative stress, inflammation, PQC, and chaperones will be invaluable to provide firm foundations for clinical innovation.

Epigenetics: linking environmental factors to gene expression

Although inborn genetic variation is a central aspect, the non-genetic regulation of cardiovascular disease (CVD) explains a large part of the observed variability in disease phenotypes and might be relevant to understand heterogeneous and multifactorial diseases such as HFpEF. Epigenetic signals or epi-mutations are plastic modifications occurring at the level of DNA and histones, which do not affect the genetic code but rather influence gene activity by modulating the accessibility of transcription factors to gene promoters.³⁴ Given their ability in orchestrating chromatin conformation, also non-coding RNAs (ncRNAs) are currently recognized as part of the epigenetic modifications. The epigenetic information acquired during life reflects individual experiences, lifestyle changes (physical inactivity) and exposure to environmental cues (early life stress, maternal nutrition, pollution, smoking, noise), and explains why individuals with an identical genetic background (i.e. monozygotic twins) display different characteristics and different cardiovascular risk profiles when grown-up under different conditions^{35  ,  36} (Figure 2). Acquisition of the epigenetic make-up starts very early, already during the intrauterine life. Of note, epigenetic changes are heritable (both from paternal and material lines) and can be transmitted across multiple generations. This implies that inherited epigenetic signals in the offspring may lead to premature transcriptional alterations and early CVD phenotypes (endothelial dysfunction, diastolic dysfunction, LV hypertrophy, skeletal muscle abnormalities), which can start to manifest already during adolescence.³⁷ Epigenetic remodelling may contribute to the current pandemic of cardiometabolic disturbances, inflammatory changes, and comorbidities, all causally implicated in HFpEF development (Graphical abstract).³⁸ Recent evidence also suggests that epigenetic processing, namely by sirtuins and histone deacetylases (HDACs), is strongly implicated in skeletal muscle metabolism and function.³⁹ These alterations may contribute to explain the impairment of exercise capacity in HFpEF patients.

Figure 2.

Environmental factors and epigenetic processing in HFpEF. Over the course of life, several factors (environmental, foetal, placental maternal) induce epigenetic signals (DNA methylation, histone modifications, non-coding RNAs), which alter chromatin accessibility and transcription of genes implicated in left ventricular hypertrophy, fibrosis, inflammation, endothelial dysfunction, and metabolic dysfunction. These modifications may significantly contribute to the development of comorbidities (obesity, insulin resistance) and HFpEF. HFpEF, heart failure with preserved ejection fraction.

Classification of epigenetic signatures

Epigenetic changes include three main classes of modifications: (i) DNA methylation; (ii) histone PTMs; and (iii) ncRNAs.

DNA methylation

DNA methylation—a process driven by a family of DNA methyltransferases (DNMT1, DNMT2, DNMT3A, DNMT3B, and DNMT3L)—consists in the addition of a methyl group to the 5C of cytosine.³⁴ In mammals, DNA methylation mainly occurs at the level of CpG islands in regulatory regions and induces (in 98% of cases) transcriptional repression by impeding the binding of transcription factors or by acting as a binding site for transcriptional repressors (i.e. HDACs).

Histone modifications

Several histone PTMs, including acetylation, methylation, ubiquitination, and SUMOylation, contribute to regulate chromatin structure and gene transcription.⁴⁰ Histone acetylation is regulated by histone acetyltransferases (HATs)—which utilize acetyl-coenzyme A as a substrate to transfer an acetyl group to the histone residue—and HDACs, involved in erasing acetyl groups from histone tails. While histone acetylation is generally associated with enhancer remodelling and active transcription, the effects of histone methylation may vary based on which residue is being modified. For example, methylation of lysine 4 and 36 of histone 3 (H3K4me and H3K36me3) enhance gene transcription while methylation of lysine 9 and 27 of histone 3 (H3K9me H3K27me) promote transcriptional silencing.⁴¹ As for histone acetylation, families of chromatin-modifying enzymes namely histone methyltransferases and demethylases regulate methylation. These histone writing-and-erasing enzymes interact selectively with DNA methylated regions and thus enable gene repression or transcription.⁴⁰

Non-coding RNAs

Non-coding RNAs can be categorized in two classes based on their size: (i) small ncRNA (<200 nucleotides) that include microRNAs (miRNAs) and endogenous short interfering RNAs, and (ii) long ncRNA (lncRNAs), which can reach a length of 2 kb.⁴² MicroRNA have shown to be key fine tuners of several biological processes occurring in the heart and the vasculature.⁴³ It was recently reported that ncRNAs also include thousands of circular RNAs, which have shown to play an essential role in the regulation of gene expression, especially acting as miRNA sponges.⁴⁴ The advent of high-throughput sequencing technologies coupled with mass spectrometry and bioinformatics techniques has led to the discovery of ≈4000 lncRNAs in the human genome.⁴⁵ Long ncRNAs are classified into multiple groups based on their genomic location, such as intergenic/intervening RNAs and intronic or exonic lncRNAs.⁴⁶ Emerging evidence shows that lncRNAs recruit epigenetic factors and trigger chromatin remodelling, thereby leading to the activation or repression of genes in the nucleus. Although ncRNAs take part to the epigenetic process, they mainly act as indirect modulators of DNA/histone complexes by orchestrating the expression and activity of chromatin-modifying enzymes.

Epigenetic control of pathways involved in diastolic dysfunction and heart failure with preserved ejection fraction

Cardiac fibrosis

DNA methylation

Several studies demonstrated a causal role of epigenetics in cardiac fibrosis.⁴⁷ In LV myocardial specimens from HF patients, Movassagh et al.⁴⁸ found reduced DNA methylation at the promoter of up-regulated genes with critical functions in cardiac fibrosis (i.e. AMOTL2, ARHGAP24, and PECAM1). Aberrant methylation of RASAL1 (a Ras-GTPase-activating protein) promoter—an epigenetic signature responsible for endothelial to mesenchymal transition (EndMT) —was observed in experimental mouse models of cardiac fibrosis as well as in HF patients.⁴⁹ Interestingly, DNMT3a and TET3 were shown to modulate RASAL1 promoter methylation leading to changes in fibrotic response.^{50  ,  51} Increased DNMT3a activation in rat cardiac fibroblasts was associated with up-regulation of ERK1/2 and down-regulation of RASSF1a, a gene involved in fibroblast activation.⁵² Interestingly, both gene silencing and pharmacological inhibition of DNMT3a were able to restore the expression of RASSF1a (Ras association domain family 1 isoform A) thus preventing fibroblast proliferation.⁵² Modulation of DNA methylation by 5-azacytidine (5-Aza) also exerts anti-fibrotic and anti-hypertrophic actions by reducing myocardial collagen levels and myocyte size in spontaneously hypertensive rats.⁵³ Furthermore, in atrial appendages from patients with ischaemic cardiomyopathy, up-regulation of DNMT1 and DNMT3B led to increased DNA methylation, and up-regulation of collagen 1 and alpha-smooth muscle actin (α-SMA) genes.⁵⁴

Histone modifications

Histone PTMs also play a key role in cardiac fibrosis. Metabolic perturbations in HFpEF patients reduce the availability of the sirtuin cofactor nicotinamide adenine dinucleotide (NAD+), leading to impaired activity of Sirt1, Sirt3, and Sirt6 and subsequent deregulation of transcriptional programmes involved in fibroblast activation.⁵⁵ Of interest, pharmacological reactivation of Sirt1 (either by resveratrol or SRT2104) prevents isoproterenol-induced cardiac fibrosis by regulating the transforming growth factor (TGF)-β/Smad2/3 pathway.⁵⁶ In mouse cardiac endothelial cells, TGFβ stimulation causes chromatin changes (H3K4me3) leading to increased transcription of plasminogen activator inhibitor-1, a key mediator of collagen synthesis and ECM remodelling in the heart.⁵⁶ Transforming growth factor-β treatment also alters the expression of the HAT p300 thus promoting EndMT and fibrosis.⁴⁹ Moreover, p300 mediates IL-1β-dependent activation of HDACs in cardiomyocytes, thus leading to cardiac fibrosis and maladaptive remodelling.⁵⁷ Furthermore, cardiomyocyte-specific deletion of the histone methyltransferase DOT1L—involved in K79 methylation at histone 3—associates with increased cardiac fibrosis.⁵⁸

Non-coding RNAs

Several miRNAs have also shown to orchestrate fibrosis and LV remodelling in HF.⁵⁹ Two different studies performed in large cohorts of HF patients showed that a panel of miRNAs (miR-193a, miR-30, miR-106a, miR-191, miR-486, miR-181a, miR-660, and miR-199b) implicated in ECM remodelling, and fatty acid biosynthesis showed high discriminative power for distinguishing HFpEF from HFrEF.^60–62 Up-regulation of miR-208a causes cardiac fibrosis and hypertrophy by targeting the TGFβ co-receptor endoglin and beta-myosin heavy chain (βMHC),⁶³ whereas its inhibition prevents perivascular fibrosis and cardiac dysfunction in Dahl salt-sensitive rats.⁶⁴ Along the same line, miR-503 increases collagen production via the apelin-13-TGFβ pathway.⁶⁵ On the other hand, down-regulation of miR-24 promotes cardiac fibrosis, while its overexpression attenuates cardiac ECM remodelling.⁶⁶ Other miRNAs involved TGFβ-dependent cardiac fibrosis include miR-21, miR-29b, miR-27b, and miR-155.^{49  ,  67} In particular, miR-21 promotes fibroblast activation via the ERK/MAPK pathway,^{68  ,  69} while its inhibition protects against TGFβ-mediated fibrosis.⁷⁰ Deregulation of miR-29b and miR-133 also associates with cardiac fibrosis in Ang II-treated rats.⁷¹ In addition, the imprinting gene delta like 1 homolog (DLK1) and miR-370 within the 14q32 cluster were reported to regulate myocardial fibrosis via targeting TGFβ/Smad3 signalling.⁷² Emerging evidence also indicates a prominent role for lncRNAs in cardiac fibrosis.⁷³ Long ncRNAs Meg3 and Wisper are potent modulators of cardiac fibrosis in HF^{74  ,  75} whereas MALAT1 and GATA6-AS facilitate TGFβ-mediated EndMT^{76  ,  77} (Figure 3).

Figure 3.

Main epigenetic alterations potentially involved in HFpEF. Alterations of DNA methylation, histone modifications, and non-coding RNA landscape promote transcriptional changes leading to key HFpEF features namely cardiac fibrosis, hypertrophy, and microvascular dysfunction. The main epigenetic signals implicated in this progression are shown.

Hypertrophic remodelling and myocardial stiffness

DNA methylation

DNA methylation significantly affects cardiac hypertrophy.⁵⁵ Indeed, pharmacological inhibition of DNA- methylating enzymes blunts hypertrophic growth in experimental models of pressure overload.⁷⁸ Moreover, changes in DNA methylation are causally involved in the development of diabetes-induced HFpEF.⁷⁹ In line with these experimental findings, failing human hearts display aberrant CpG methylation of genes involved in cardiac contractility, hypertrophy, and diastolic dysfunction (i.e. Ly75 and ERBB3).⁸⁰ Oxidative stress was recently shown to impair mitochondrial DNMT1 activity thus promoting mitochondrial DNA methylation⁸¹ and transcriptional changes ultimately affecting mitochondrial biogenesis and hypertrophic response.^{82  ,  83}

Histone modifications

Among different chromatin modifiers, sirtuins are strongly deregulated in cardiac hypertrophy and failure.⁸⁴ In rat cardiomyocytes stimulated with endothelin-1, overexpression of Sirt1 prevents cardiac hypertrophy via degradation of H2A.Z.⁸⁵ Consistently, Sirt1 activation attenuates cardiomyocyte hypertrophy in Ang II-treated mice and spontaneously hypertensive rats.^{86  ,  87} Along the same line, Sirt3 KO mice show an increased hypertrophic and fibrotic response,⁸⁸ whereas overexpression of Sirt3 protects against cardiac hypertrophy by promoting FoxO3-dependent transcription of antioxidant genes.⁸⁹ Moreover, Sirt6 KO mice develop cardiac hypertrophy and HF while Sirt6 overexpression prevents these alterations by modulating IGF/Akt signalling.⁹⁰ Of clinical relevance, Sirt6 is also down-regulated in patients with HF.⁹⁰ Ample evidence supports the role of HDACs in hypertrophic growth. Specifically, class II HDACs exert anti-hypertrophic activity whereas class I HDACs mediate the hypertrophic response.⁹¹ In a rat model of hypertrophy, treatment with magnesium valproate—a class I HDAC inhibitor—prevented cardiac hypertrophy by reducing oxidative stress.⁹² In line with these findings, HDAC2 KO mice are protected against isoproterenol-induced cardiac hypertrophy via transcriptional regulation of Inpp5f, a negative modulator of Akt/Gsk3β pathway.⁹³ In contrast, HDAC2 transgenic mice are hypersensitive to hypertrophic stimuli.⁹³ Unlike HDAC2, depletion of HDAC5 or HDAC9 in mice leads to hypertrophic remodelling and HF via inhibition of the transcription factor MEF2C.^{94  ,  95} Similarly, genetic deletion of HDAC4—another class II HDAC—leads to cardiac hypertrophy.⁹⁶ Hyperactivation of histone acetyl-transferases [i.e. CREB binding protein (CBP) and p300] is mostly associated with hypertrophy and LV remodelling in mice.⁹¹ Indeed, inhibition of either CBP or p300 prevented phenylephrine-induced hypertrophy, while their overexpression in cardiomyocytes promoted hypertrophic growth.⁹⁷ In contrast, the expression of the acetyltransferase MOF was reduced in hypertrophic murine hearts and in failing human hearts, and its overexpression protected against transverse aortic constriction (TAC)-induced hypertrophy via activation of the antioxidant enzyme catalase and MnSOD.⁹⁸ While the role of histone acetylation in HF has been largely investigated, the role of histone methylation remains poorly understood. Genome-wide studies performed in rats and humans showed that H3K4me3 and H3K9me3 are strongly increased in HF.^{99  ,  100} Another work demonstrated that JMJD2A—responsible for the demethylation of H3K9me3 and H3K36me3—is up-regulated in cardiac hypertrophy while its genetic deletion is protective.¹⁰¹ Of interest, the methyltransferase Smyd2 was recently found to methylate the chaperone HSP90, thus promoting the interaction of a Smyd2–methyl-HSP90 complex with the N2A-domain of titin.¹⁰² Loss of function of Smyd2 was associated with increased myocardial stiffness and cardiac dysfunction in experimental models.¹⁰²

Non-coding RNAs

Down-regulation of miR-1 leads to cardiac hypertrophy and HFpEF development by regulating the calcium signalling components calmodulin, Mef2a, and GATA4.^{103  ,  104} In cultured cardiomyocytes, several miRNAs (i.e. miR-23a, miR-23b, miR-24, miR-195, and miR-214) were found to be implicated in hypertrophic response.^{103  ,  105} Among these, miR-195 is reported to act as a key driver of myocyte growth and HFpEF in mice.¹⁰³ Up-regulation of miR-22 also participates in cardiac hypertrophy, and its inhibition prevents both Ang II and isoproterenol-induced cardiac hypertrophy in vitro and in vivo by modulating Sirt1 and HDAC4.^{106  ,  107} Other miRNAs, namely miR-27 and miR-21*, are also causally involved in hypertrophic cell growth.^{108  ,  109} Among lncRNAs, Mhrt was shown to protect the heart from hypertrophy by interacting with the transcriptional activator BRG1.¹¹⁰ The lncRNA Chast is also implicated in autophagy inhibition and hypertrophic remodelling by regulating Pleckstrin homology domain-containing protein family M member 1 (PLEKHM1).¹¹¹ Along the same line, the lncRNA CHAER interacts with the methyltransferase PRC2, thus reducing the repressive H3K27me3 marks at the promoters of genes involved in cardiac hypertrophy.¹¹² The lncRNA H19 is highly conserved and down-regulated in failing hearts from mice, pigs, and humans. H19 gene therapy prevents and reverses experimental pressure overload-induced HF and acts as an anti-hypertrophic lncRNA and represents a promising therapeutic target to combat pathological cardiac remodelling.¹¹³ A specific miRNA, miR-132, has been shown to play a crucial role in cardiac hypertrophy and remodelling¹¹⁴ and was proven efficient to improve both systolic and diastolic function in large animal models of myocardial infarction.¹¹⁵

Microvascular dysfunction

Microvascular dysfunction, mainly resulting from inflammation and oxidative stress, is a main culprit in HFpEF and is strongly regulated by epigenetic modifications.¹¹⁶ Several HFpEF-related comorbidities, namely obesity, type 2 diabetes and hypertension, are strongly implicated in modifications of the epigenetic landscape in the vasculature and may participate in HFpEF.¹¹⁶

DNA methylation

In diabetic patients, reduced DNA methylation of p66^Shc promoter results in gene up-regulation, vascular oxidative stress, and endothelial dysfunction.¹¹⁷ Reduced DNA methylation and increased H3K4me3 were found on the promoter of TNFα, a potent inflammatory cytokine involved in obesity and hypertension-related microvascular dysfunction.^{118  ,  119} Along the same line, age-dependent DNA hypomethylation regulates the expression of IL-1β, a relevant cytokine, which was recently linked to increased vascular risk and HF.^{120  ,  121} DNA hypomethylation of endothelin-1 promoter was also linked to microvascular dysfunction.¹²²

Histone modifications

Alterations of histone acetylation participate in microvascular complications by regulating inflammatory transcriptional programmes.¹²³ The HAT p300/CBP-associated factor (PCAF) was shown to epigenetically regulate the pro-inflammatory transcription factor NF-kB as well as its downstream genes CCL2, IL-6, and TNFα in the vasculature.¹²⁴ Increased acetylation at the promoter of inflammatory genes has also been reported in diabetic patients.¹²⁵ In line with these finding, HDAC6-induced histone deacetylation protects against oxidative stress-induced endothelial dysfunction.¹²⁶ Sirtuins, a class of HDACs, play a key role in epigenetic remodelling and microvascular disease.¹²⁷ In particular, the NAD⁺-dependent deacetylase SIRT1 induce chromatin changes (namely H3 deacetylation) leading to the modulation of genes implicated in nitric oxide signalling, inflammation, oxidative stress, autophagy, and vascular aging.¹²⁸ In the endothelium, Sirt1 down-regulation is associated with decreased levels of FOXO1-dependent antioxidant genes, thus promoting oxidative stress-induced endothelial dysfunction.¹²⁷ Moreover, SIRT3 overexpression improves ROS-driven endothelial dysfunction by enhancing the antioxidant defence.¹²⁹ Like Sirt1 and Sirt3, Sirt6 deficiency associates with enhanced inflammation, oxidative stress, and defective angiogenesis.¹²³ In the diabetic vasculature, aberrant histone methylation (H3K4me1, H3K9me2, and H3K9me3) at Nox4 and eNOS promoters accounts for persistent up-regulation of these genes with consequent increase of oxidative stress and endothelial dysfunction.¹³⁰

Non-coding RNAs

Several miRNAs, such as miR-92, miR-126, miR-195, miR 26a, and miR-155, are also heavily implicated in microvascular dysfunction by targeting relevant pathways underlying endothelial and smooth muscle cell damage (i.e. eNOS, RhoA, Smad1, Sirt1, Bcl-2).¹³¹ Deregulation of miR-29 and miR-34 also triggers endothelial dysfunction by derailing nitric oxide and Sirt1 signalling pathways, respectively.^{132  ,  133} Another approach is to use potentially anti-fibrotic drugs that modulate miRNAs, as recently reported.¹³⁴ In this study, miRNA regulatory natural compounds derived from a library screen promoted anti-fibrotic effects in the heart, thus improving LV relaxation in several animal models of diastolic dysfunction.¹³⁴ Finally, recent evidence suggest an involvement of lncRNA MALAT1 and MEG3 in endothelial inflammation, permeability, and dysfunction.^{135  ,  136} The most important epigenetic alterations contributing to HFpEF are reported in Figure 3.

Potential application of epigenetics for heart failure with preserved ejection fraction phenomapping

Rationale for using epigenetic information

Epigenetic signals acquired during life can be employed as a valuable tool to predict structural (i.e. fibrosis, hypertrophy) and functional (diastolic dysfunction) alterations of the heart eventually leading to HFpEF. Indeed, as compared with our genetic make-up, modifications of DNA and histones have the advantage of reflecting environmental exposures. The relevance of using epigenetic information in HFpEF is supported by the pivotal role of environmental factors and comorbidities (i.e. obesity, type 2 diabetes) in its pathogenesis. Of note, once acquired most of epigenetic modifications are relatively long-lasting and able to induce a durable ‘epigenetic memory’ affecting the transcriptional profile along the arc of aging.¹³⁷ There are several reasons why epigenetic biomarkers might represent an attractive option in this setting: (i) the epigenetic background embraces the contribution of lifestyle and environmental factors; this represents a clear advantage over genetic biomarkers, which are based only on DNA sequence alterations; (ii) DNA-based biomarkers are rather stable in both fluids ad tissue specimens; and (iii) epigenetic changes can be detected in all genomic contexts, not only in coding regions, and at early stages of diseases.¹³⁸ The dynamism and the strong cell specificity imply a rigorous approach in the detection of epigenetic signals for the prediction of HFpEF.

The advent of computational epigenetics

The recent development of computational epigenetics (a combination of computer science, statistics, physics, and computational biology) combined with deep machine learning may certainly help to organize and structure wide epigenomic data to build predictive models.^{139  ,  140} This approach has already been used in the field of cancer, stem cell, neurodegenerative, and autoimmune diseases and holds great promises for risk prediction.¹³⁸ A recent data-driven machine learning approach employing 18 highly discriminating CpGs was able to identify specific methylation signatures controlling pivotal genes implicated in autoimmune dysregulation and food allergies.¹⁴¹ In few years, computational epigenetics may help to: (i) analyse and interpret large epigenomic datasets to quantify HFpEF risk and start personalized approaches, (ii) draw meaningful inferences of epigenetic modifications in HFpEF, (iii) test existing epidrugs and develop new epigenome-editing and high-throughput experimental methodologies, and (iv) take advantage of computational methods, especially machine learning approaches, to leverage epigenomic data for HFpEF phenomapping.

Epigenetic biomarkers

Over the last few years, several studies contributed to unmask the epigenetic landscape in HF patients (Table 1). A recent pilot study employing powerful whole-genome methyl-binding domain-capture sequencing showed that peripheral blood DNA methylation was able to discriminate between coronary artery disease patients with and without HF.¹⁴⁸ Another study unveiled several epigenetic loci associated with HF phenotypes. Of note, several of these loci could be replicated in independent cohorts, thus underlining the role of epigenetics in regulating key transcriptional routes in the heart. Moreover, these methylation changes were conserved across different tissues (myocardial samples, peripheral blood monocytes).¹⁴⁹ Along the same line, targeted DNA methylation sequencing revealed specific epigenetic patterns able to discriminate among different HF subtypes.¹⁴² In this study, hypomethylation of CTGF and MMP-2 was among the most important epigenetic modifications. Given the potential involvement of these genes in HFpEF, as outlined by strong associations with cardiac fibrosis,¹⁵⁰ future studies should investigate whether epigenetic remodelling at the promoter of CTGF and MMP-2 (DNA methylation, histone modifications) confers additional information in predicting HFpEF development over time. In contrast, other studies did not confirm significant variations of CpG methylation in failing vs. non-failing human hearts.¹⁵¹ Such discrepant findings in DNA methylation patterns in the failing heart could be due to the different technologies used to detect DNA methylation (i.e. whole-genome bisulphite sequencing vs. bead array vs. capture-based bisulphite sequencing).¹⁵²

Table 1.

Potential epigenetic biomarkers in heart failure with preserved ejection fraction

Epigenetic signature	Target	Detectability in fluids/tissues	Open issues	Relevance to HFpEF
CpG methylation of gene loci	LY75, PTGES, CTNNAL1, TNFSF14, MRPL16, KIF17, CTGF, and MMP-2	Whole blood	Not all studies have confirmed a significant variation of CpG methylation in failing vs. non-failing human hearts. Such discrepant findings could be due to the different technologies for DNA methylation used in studies (i.e. WGBS vs. bead array vs. capture-based bisulphite sequencing).	Reduced CpG methylation of CTGF and MMP-2 are potentially important biomarkers in HFpEF given their strong involvement in cardiac fibrosis.142
CpG methylation of gene loci	COL9A1 IGF2BP1, GATA4, and TET1	Visceral adipose tissue	The applicability of this approach is limited giving the difficulties in obtaining biopsies of adipose tissues.	These changes are casually involved in metabolic dysfunction and systemic inflammation, key hallmarks of HFpEF.143 Investigation of CpG signatures in EAT could be also relevant.
Somatic mutations of chromatin modifiers (CHIP)	DNMT3A, TET2, and ASXL1	Whole blood	Whether CHIP is a causal biomarker in HFpEF remains elusive. Further studies are needed to confirm its predictive value in this setting.	Gene mutations associated with CHIP were independently associated with incident HFpEF in a cohort of 5214 postmenopausal women.144
Histone PTM	H3K4me3, H3K9me3, and H3K36me3	Myocardial biopsies, PBMCs, circulating cell-free nucleosomes (?)	Histone PTM can be only assessed in tissues or in isolated cells; this affects the applicability of this method. Circulating cell-free nucleosomes can be used to overcome this limitation.	These chromatin marks can be employed as causal biomarkers of LV hypertrophy and remodelling in HFpEF and could be used to direct and personalized intervention.
miRNA	miR-183–3p	Plasma/serum	Lack of specificity for HFpEF (also down-regulated in HFrEF), further validation is required.	miR-183–3p is down-regulated in patients with HFpEF.145
miRNA	miR-190a	Plasma/serum	Non-causal biomarker which requires further validation in multiple cohorts.	Useful to detect HFpEF, and to discriminate HFpEF from HFrEF.145
miRNA	miR-193b–5p, miR-494-3p, miR-454, and miR-500	Plasma/serum	Further validation required.	Down-regulated in HFpEF patients. miR-494-3p is a causal biomarker involved in lipotoxic damage and myocardial steatosis.146
miRNA	miR-30c, -miR-146a, miR-221, miR-328, and miR-375	Plasma/serum	Most of these miRs are not affected in HFpEF patients (only miR-375 is significantly down-regulated).	Causal biomarkers useful to differentiate HFpEF from HFrEF. Incremental prognostic value beyond NT-proBNP.60
miRNA	miR-545-5p, miR-671-5p, miR-1233, and miR-1246	Plasma/serum	Further validation required.	Up-regulated in HFpEF patients.145  ,  147

CHIP, Clonal hematopoiesis of indeterminate potential; EAT, epicardial adipose tissue; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; PTM, post-translational modifications.

Somatic mutations of chromatin modifiers: the emerging role of CHIP in heart failure with preserved ejection fraction

Somatic mutations of chromatin modifiers regulating DNA methylation, namely DNMT3A and TET2, were recently linked to the emergence of leucocyte clones (clonal haematopoiesis of indeterminate potential, CHIP) that not only confer increased risk of haematologic malignancies, but also CVD such as myocardial infarction and stroke.¹⁵³ The likelihood of CHIP increases with age, and CHIP mutations are present in around 10% of people aged between 71 and 80 years. A recent cohort study showed that mutations of DNMT3A and TET2 were highly prevalent in patients with HF (18.5%) and independently associated with death and HF re-hospitalization.¹⁵⁴ Given the strong link between aging and HFpEF, these mutations may play an important causal role in HFpEF. A very recent analysis of 5214 postmenopausal women in the Women’s Health Initiative showed that gene mutations associated with CHIP (DNMT3A, TET2, and ASXL1) were independently associated with incident HFpEF.¹⁴⁴ Consistent with these observations in humans, experimental models of CHIP in mice displayed high levels of myocardial inflammation and dysfunction.¹⁵⁵ Taken together, these findings suggest that age-driven loss of function of DNA methyl-writing enzymes skews immune cells towards a pro-inflammatory phenotype and might significantly participate in triggering immune dysregulation, relevant comorbidities (obesity, type 2 diabetes, chronic obstructive pulmonary disease), and eventually HFpEF.¹⁵⁶

Sex-specific epigenetic signals in heart failure with preserved ejection fraction

HFpEF has emerged as a sex-biased disease with higher prevalence (>60%) in female patients, who are generally postmenopausal and have one or more risk factors such as diabetes, hypertension, and inflammation.¹⁵⁷ The mechanisms underlying this HFpEF sex difference are currently unknown. However, it is conceivable to hypothesize that sex-specific genes, hormones, chromosomes, epigenetics, and their intertwined interactions may orchestrate the sex-biased modulation of cardiac structure and function between men and women. Women with HFpEF develop less ventricular dilation but have a small but significant increase in ventricular stiffness, likely due to differential susceptibility to accumulation of fibrotic tissue or dissimilar relaxation parameters and calcium handling, leading to a greater level of impaired diastolic filling.¹⁵⁸ For instance, female patients with HFpEF display higher levels of the propeptide for type I collagen and have increased partition coefficients, extracellular volumes, and pre-contrast T1 (factors associated with myocardial fibrosis) than men.¹⁵⁸ Oestrogen deficiency may certainly contribute to explain these differences. It is our hypothesis that an interplay between sex hormones and sex-specific epigenetic mechanisms underlie the gender dimorphism in HFpEF prevalence. Oestrogen and androgen receptors bind to hormone response elements and recruit the HATs CBP and E1A binding protein p300 (EP300) to the DNA,¹⁵⁹ strongly suggesting that sex hormone may directly regulate DNA and histone-modifying enzymes to impact the epigenetic processes in a sex-specific manner in the development of the spectrum of HFpEF and its course. However, there is urgent need to further this area of research.

Epigenetic indicators of metabolic inflammation: early warning signals?

Maladaptive epigenetic processing as the result of environmental cues and lifestyle may contribute to trigger the combination of a pro-inflammatory and dysmetabolic phenotype, a condition known as metabolic inflammation (meta-inflammation).¹⁶⁰ Of interest, recent studies pinpoint meta-inflammation as a major culprit in HFpEF pathogenesis.¹⁶⁰ Very likely, the epigenetic routes involved in cardiometabolic alterations are the same that, in the long term, will foster the development of HFpEF. Chromatin remodelling in adipose tissue might also participate in cardiometabolic phenotypes, systemic inflammation, and HFpEF. Growing evidence indicates that transcriptional changes in visceral and epicardial tissue (EAT) may alter the adipocyte secretome eventually fostering myocardial apoptosis, stiffness, and diastolic dysfunction. Genome-wide epigenetic analysis in visceral adipose tissue from obese patients revealed important variations in the methylation of CpG sites at the promoter of several genes, namely COL9A1, IGF2BP1, GATA4, and TET1.¹⁴³ Other studies support the role of epigenetics in the switch from brown to white adipose tissue, a detrimental event leading to adipogenesis and secretion of pro-inflammatory chemokines (i.e. IL-6) acting on the surrounding myocardium.²¹

Histone maps

Only few studies have explored the potential utility of histone PTM as biomarkers of cardiac remodelling and HF. Using high-throughput pyrosequencing performed with ChIP products, Kaneda et al.⁹⁹ showed that the distribution patterns of H3K4me3 and H3K9me3 were significantly different between healthy and failing hearts and these marks were found in the gene loci involved in calcium signalling and cardiac contractility. Papait et al.¹⁰⁰ also reported a set of promoters with an epigenetic pattern that distinguishes specific functional classes of genes regulated in hypertrophy and identified 9207 candidate active enhancers whose activity was modulated. Specific histone modifications, namely reduced H3K9me3 or H3K27me3, were strongly associated with up-regulation of genes involved in cardiac hypertrophy, namely myocyte enhancer factor (MEF)2C.¹⁶¹ Another study performed in human myocardial samples from HF patients showed that H3K27ac and H3K36me3 were among the predictive marks for relevant transcriptional changes occurring in the failing myocardium (i.e. CTGF).¹⁵¹ Future studies should appraise whether specific histone pattern may provide early information on transcriptional programmes preceding LV hypertrophy and diastolic dysfunction.

MicroRNAs as potential fingerprints of heart failure with preserved ejection fraction

Circulating miRNAs could help to stratify HFpEF risk, thus opening avenues for individualized treatment. Watson et al.⁶⁰ showed that various miRNA combinations (miR-30c, −146a, −221, −328, and −375) are useful biomarkers for HF and are also effective in the differentiation of HFpEF from HFrEF. The biological role of these miRs, which regulate ECM remodelling, myocardial inflammation, and fibrosis, fits well with a potential involvement in HFpEF. In 2018, Chen et al.¹⁴⁷ reported the identification of two highly up-regulated miRNAs, miR-3135b and miR-3908, in HFpEF and highlighted their potential as markers differentiating HFrEF from HFpEF. Computational analyses of miRNA signatures in this setting showed that ECM receptor and fatty acid biosynthesis were the pathways mainly contributing to the pathophysiological differences between HFrEF and HFpEF.⁶² Consistent with these observations, miR-494-3p was recently shown to regulate fatty acid synthase and cardiac steatosis in obese patients with severe diastolic dysfunction and preserved ejection fraction.¹⁴⁶ From a clinical standpoint, it would be also relevant to search for miRNAs able to identify preclinical diastolic dysfunction rather than manifest HFpEF or its differentiation with HFrEF. This approach would allow the identification of high-risk patients and would enable preventive therapies, and disease monitoring over time.¹⁴⁵

Current challenges for individual epigenetic profiling

Cell specificity of epigenetic information poses the challenge of which tissue/fluid is appropriate to unveil specific biomarkers in the setting of HFpEF. The study of epigenetic landscape can be applied to a wide range of samples obtained by non-invasive techniques (saliva, stool, urine, or blood) to invasive protocols (tissue biopsy). The use of liquid biopsies is particularly recommended for the identification of biomarkers on a large scale. Several studies have already shown that CpG methylation in whole blood or peripheral blood mononuclear cells or monocytes can reliably mirror epigenetic alterations occurring in cardiac cells. Tissue samples can be used in selected cases to monitor myocardial epigenetic alterations over the course of treatment. Another issue is that not all epigenetic changes can be easily detected in clinical practice. For example, histone PTMs can only be assessed in tissues or in isolated cells; this obviously reduces the applicability of this method. To overcome this limitation, a recent study reported that circulating cell-free nucleosomes (cf-nucleosomes) can be employed to detect histone marks. These circulating nucleosomes—which are generally released by apoptotic or necrotic cells into the bloodstream—can be detected in patients with cancer or CVD. Of interest, the pattern of histone modifications in circulating cf-nucleosomes can be associated with the cell-type of origin, a feature that increases its potential as a biomarker.¹⁶² This is a clear advantage in clinical cardiology as it allows to detect cardiomyocyte-specific histone modifications from blood. Last but not least, only few epigenetic marks have been adequately validated for routine clinical use. Important current limitations of epigenetic biomarkers are that assays are diverse, comparative data are few, and inconsistencies among published papers are frequent.¹⁶³ Moreover, reproducibility in their assessment remains an important issue as different technologies are being used to detect DNA methylation (i.e. whole-genome bisulphite sequencing vs. bead array vs. capture-based bisulphite sequencing).¹⁵² There remain indeed important steps to be taken before leveraging epigenetic information in the clinical context of HF. Many epigenetic biomarkers need a systematic validation and subsequent configuration into clinical assays and regulatory approvals. A stronger collaboration between industry and academia may pave the way for a faster validation of potential new epigenetic biomarkers in real-world applications.

Epidrugs for personalized therapies in heart failure with preserved ejection fraction

The growing understanding of chromatin structure has led to the design of specific drugs and discovery of dietary compounds with ability to erase or write epigenetic signatures eventually resetting the cell transcriptome in disease states.¹⁶⁴ Notably, several of these compounds have been already approved by the Food and Drug Administration (FDA) for the treatment of cancer, neurological and autoimmune diseases. Here we describe a series of potential pharmacological intervention to re-programme adverse epigenetic modifications underlying pathological cardiac remodelling and HFpEF (Table 2).

Table 2.

Epidrugs with potential application in heart failure with preserved ejection fraction

Drug	Epigenetic mechanism	Molecular action	Potential application in HFpEF
Folatesa	Methyl donor	Restore CpG methylation of genes regulating endothelial function, nitric oxide bioavailability (eNOS), oxidative stress (i.e. p66Shc), adipogenesis and liver steatosis (i.e. glucocorticoid receptor and PPARα).	Improvement of microvascular function and perivascular inflammation; prevention of myocardial steatosis and lipotoxic damage; possible EAT browning with anti-inflammatory effects and restoration of autophagic flux in the heart165  ,  166
Azaticidine (5-AZA)a	DNMT inhibitor	Inhibits DNMT with subsequent hypomethylation of genes implicated in vascular homeostasis (i.e. eNOS).	Possible improvement of microvascular function. The drug induces a global hypomethylation. Therefore, a cell-specific approach (ECs) is warranted.167
Sulphoraphane	HDAC inhibitor	Inhibits class IIa HDAC and HDAC2 enzyme activities in different cell types.	Effect on microvascular endothelial function, vascular inflammation and ECM remodelling. Improves glucose homeostasis (potential effects on cardiac metabolism).168  ,  169
EGCG	DNMT inhibitor; HAT inhibitor; miRNA regulator	Regulates chromatin accessibility at the promoter of inflammatory genes; modulates AMPK/mTOR signalling.	Prevents endothelial inflammation (i.e. ICAM-1) and dysfunction; restores autophagy with anti-hypertrophic and anti-fibrotic effects in the heart.170–172
Danshen	HMT inhibitor	Inhibits JMJD2A, a methyltransferase promoting cardiac hypertrophy via reduced H3K9 trimethylation and FHL1 up-regulation.	Positive effects of LV remodelling (cardiac hypertrophy and geometry, diastolic function).173
Vorinostat (SAHA)a	HDAC inhibitor	Regulates chromatin structure and transcription of genes mediating inflammation (IL-1β, IL-6, and TNFα), fibrosis, autophagy, and mitochondrial biogenesis (PGC1α). Improves functional and haemodynamic parameters in experimental HFpEF.	Prevents hypertrophic remodelling (autophagy restoration), atrial enlargement, diastolic dysfunction while improving metabolic efficiency and cardiac energetics (mitochondrial function); effects on endothelial inflammation and microvascular function.174–176
Sodium butyratea	HDAC inhibitor	Suppresses inflammatory transcriptional programmes (NF-kB signalling) and prevents metabolic alterations by enhancing oxidative phosphorylation and beta-oxidation in mitochondria.	Preservation of metabolic efficiency and cardiac energetics; anti-inflammatory action on the vasculature and the myocardium with possible beneficial effects on the microcirculation and LV hypertrophy.177
Trichostatin A	HDAC inhibitor	Suppresses TNFα transcription in the heart.	Effects on LV hypertrophy and remodelling.177
Valproic acida	HDAC inhibitor	Regulates the Foxm1 pathway and prevents MI-induced LV remodelling.177	Possible effects on myocardial fibrosis.
Apicidin	HDAC inhibitor	Regulation of Akt signalling with reduction of TAC-induced LV hypertrophy and failure.	Anti-hypertrophic and anti-fibrotic action with improvement of LV relaxation.177
Givinostat (ITF2357)a	HDAC inhibitor	Ameliorates both hypertension and aging-induced HFpEF by improving cardiac myofibril relaxation. The observed effects are independent of chromatin remodelling.	Promising candidate in HFpEF, direct action on cardiac myofibril relaxation in HFpEF models.178
Resveratrol	HDAC inhibitor	Attenuates adipogenesis, inflammation, oxidative stress, endothelial dysfunction, and LV relaxation.	Beneficial effects on diastolic function, hypertrophy, and microvascular dysfunction.179–181
Curcumina	HAT inhibitor	Proteasome-dependent degradation of the HAT p300 and the closely related CBP protein with suppression of pro-inflammatory and pro-oxidant genes. Improves endothelial function.	Potential benefits on LV hypertrophy, diastolic function, and the microcirculation.177
JQ1	BET inhibitor	Epigenetic action on NF-κB and TGFβ signalling networks with prevention of pressure overload-induced hypertrophy and fibrosis.	Effects on myocardial inflammation, fibrosis, LV remodelling, and diastolic function.182
Apabetalone (RVX-208)a	BET inhibitor	Modulates gene transcription by preventing the interaction of acetylated histones with DNA. Modulates lipid metabolism, oxidative stress, and vascular inflammation (IL-1β, IL-6, and TNFα).	Putative beneficial effects on microvascular dysfunction and metabolic inflammation.183  ,  184
MRG-201a	miR-29 mimic	miR-29 is a powerful regulator of ECM remodelling and cardiac fibrosis. MRG-201 exerts potent anti-fibrotic activity and is being tested in a phase I clinical trial.	Reduction of cardiac fibrosis with improvement of LV relaxation and exercise tolerance.185
RG-012	miR-21 antagomir	Prevents cardiac inflammation (IL-1β, IL-6, and TNFα), and maladaptive LV remodelling after MI.	Effects on LV hypertrophy, fibrosis, and microvascular function.185
CDR132L	Synthetic lead-optimized oligonucleotide inhibitor of miR-132	Improves diastolic dysfunction while reducing left atrial size and remodelling in preclinical HF models. Successfully tested in a phase 1b trial for its safety and PK profile in HF patients.186	Good candidate in HFpEF for its beneficial effects on LV relaxation, cardiac haemodynamics, and exercise tolerance.187

DNMT, DNA methyltransferase; EAT, epicardial adipose tissue; ECM, extracellular matrix; HAT, histone acetyltransferase; HDAC, histone deacetylase; HF, heart failure; HFpEF, heart failure with preserved ejection fraction; HMT, histone methyltransferase; LV, left ventricle.

^aFood and Drug Administration approved drugs.

DNA methylation editing-drugs

Folates act as modulators of epigenetic modifications as the methyl group responsible for DNA and histone methylation originates from S-adenosyl methionine. Restoration of CpG methylation by folates impact on several transcriptional programmes implicated in HFpEF-related comorbidities such as hypertension, obesity, and diabetes.^{188  ,  189} Treatment with folates has shown to restore promoter methylation of genes regulating endothelial dysfunction, nitric oxide bioavailability as well as key pathways involved in adipogenesis and liver steatosis such as hepatic glucocorticoid receptor and PPARα.¹⁶⁵ Of note, folates epigenetically regulate the transcription of the mitochondrial adaptor p66^Shc, a key driver of myocardial oxidative stress and dysfunction.¹⁶⁶ Other agents able to modulate DNA methylation—namely azacytidine—have shown to prevent endothelial apoptosis and inflammation mainly via promoter hypomethylation and restoration of eNOS transcription.¹⁶⁷ The latter mechanism could be relevant in patients with HFpEF as targeted overexpression of eNOS within the vascular endothelium attenuates diastolic dysfunction in mice.¹⁹⁰

Natural compounds as epigenetic modulators

Dietary compounds such as sulphoraphane—an organosulphur compound found in broccoli sprouts—inhibit class IIa HDAC and HDAC2 enzyme activities in different cell types and have shown to prevent vascular remodelling and fibrosis via inhibition of the Nrf2 pathway.¹⁶⁸ Sulphoraphane also blunts TNFα-mediated endothelial inflammation by suppressing endothelin-1, VCAM-1, ICAM-1, and E-selectin.¹⁶⁹ These findings gain importance in patients with HFpEF as recent work has shown a pivotal involvement of endothelial inflammation in its pathogenesis. Indeed, E-selectin and ICAM-1 are significantly up-regulated in the myocardium of HFpEF patients and ZSF1-HFpEF rats and participate in myocardial stiffness and dysfunction.^{6  ,  7} Epigallocatechin-3-gallate (EGCG)—a catechin contained in green tea—acts as DNMT inhibitor, HAT inhibitor as well as a miRNA regulator and has shown to prevent lipopolysaccharide-induced ICAM-1 and VCAM-1 up-regulation in human endothelial cells.¹⁷⁰ Moreover, EGCG promotes autophagy via modulation of the mTOR-AMPK pathway.¹⁷¹ Suppression of mTOR by EGCG could represent an important mechanism to counteract myocardial fibrosis and hypertrophy in HFpEF patients.¹⁷² Danshen—the dry root and rhizome of the herbaceous plant Salvia miltiorrhiza—inhibits the methyltransferase JMJD2A and may exert protective effects in HFpEF.¹⁷³ Indeed, JMJD2A promotes TAC-induced cardiac hypertrophy via reduced H3K9 trimethylation and FHL1 up-regulation, and its expression is significantly enhanced in patients with pathological cardiac hypertrophy.¹⁰¹

Histone deacetylase inhibitors

The FDA approved compound vorinostat (SAHA) acts as an HDAC inhibitor and has shown to prevent pathological cardiac remodelling in mice via modulation of autophagy as well as via anti-oxidant and anti-inflammatory properties. In an experimental model of hypertensive cardiomyopathy, SAHA blunted circulating levels of several pro-inflammatory cytokines, including IL-1β, IL-6, and TNFα, thus reducing cardiac hypertrophy and interstitial fibrosis.¹⁷⁴ Moreover, in obese mice SAHA regulates the expression of PGC-1α, thus preserving mitochondrial biogenesis and oxygen consumption in the ischaemic myocardium.¹⁷⁵ Together with vorinostat, the whole class of HDAC inhibitors seems to hold promises for the prevention of cardiac remodelling and HF.¹⁹¹ In a feline model of HFpEF, SAHA reduced LV hypertrophy, left atrial size while improving LV relaxation via epigenetic regulation of mitochondrial function.¹⁷⁶ Different HDAC inhibitors, including sodium butyrate, valproic acid, trichostatin A, and apicidin consistently prevented cardiac dysfunction and HF in experimental models.^{177  ,  192} The HDAC inhibitor givinostat (ITF2357) also ameliorated hypertension and aging-induced HFpEF by improving cardiac myofibril relaxation.¹⁷⁸ Overall, treatment with HDAC inhibitors could be potentially relevant in HFpEF, where inflammation, fibrosis, and myocardial stiffness play a prominent role.

Sirtuins

Inhibition of the HDAC SIRT1 by resveratrol has shown to attenuate adipogenesis, inflammation, oxidative stress, and to rescue obesity-related micro- and macrovascular dysfunction. Interestingly, several experimental studies in ischaemic and non-ischaemic HF have shown beneficial effects of resveratrol in improving diastolic dysfunction, LV remodelling, cardiac haemodynamics, energetics, and exercise capacity.¹⁷⁹ However, despite these preclinical studies, it is still unknown if resveratrol can improve HF in humans. In a double-blind, placebo-controlled trial involving patients with stable coronary artery disease receiving 10 mg daily of resveratrol for 3 months, resveratrol improved LV diastolic function.¹⁹³ Moreover, 20 mg of resveratrol/day administered for 60 days resulted in a significant decrease in B-type natriuretic peptide in patients with angina pectoris, suggesting improved LV function.¹⁸⁰ Although limited, these studies suggest that SIRT1 activation by resveratrol may have a direct impact on diastolic function and HFpEF in humans. Ongoing randomized trials, including the REV-HF (NCT03525379) and the RES-HF (NCT01914081) trials, will contribute to understand the efficacy of resveratrol in patients with HF.¹⁸¹ Modulation of SIRT1, SIRT3, and SIRT6 has also shown to prevent endothelial dysfunction and vascular oxidative stress. The latter aspect deserves attention as coronary microvascular rarefaction and microvascular dysfunction are emerging as major contributors in diastolic dysfunction and HFpEF.¹¹⁶

Bromodomain and extra-terminal motif inhibitors

Bromodomain and extra-terminal motif (BET) inhibitors represent an emerging class of drugs that reversibly bind the bromodomains of BET proteins BRD2, BRD3, BRD4, and BRDT, and prevent protein–protein interaction between BET proteins, acetylated histones, and transcription factors. A recent study showed that the BET bromodomain inhibitor JQ1 prevents pressure overload-induced hypertrophy and HF.¹⁸² Integrated transcriptomic analyses across animal models and human iPSC-CMs reveal that BET inhibition preferentially blocks transactivation of a common pathologic gene regulatory programme that is robustly enriched for NFκB and TGFβ signalling networks, typified by innate inflammatory and profibrotic myocardial genes.¹⁸² These findings establish that pharmacologically targeting innate inflammatory and profibrotic myocardial signalling networks at the level of chromatin is effective in animal models and human cardiomyocytes, providing the critical rationale for further development of BET inhibitors in HFpEF. The BET inhibitor apabetalone (RVX-208) has also shown to prevent endothelial inflammation (IL-1β, IL-6, and TNFα) and atherosclerosis in ApoE^−/− mice.¹⁸³ The recent phase III BETonMACE trial, designed to investigate the impact of apabetalone on cardiovascular outcomes in 2425 patients with diabetes after an acute coronary syndrome, failed to meet the primary endpoint (cardiovascular death, non-fatal myocardial infarction, or stroke).¹⁸⁴ However, the drug was associated with a reduced risk of first (29 vs. 48, P = 0.03) or first and recurrent congestive HF hospitalizations (35 vs. 70, P = 0.01). Undoubtedly, larger clinical trials are needed to explore better the safety and efficacy of apabetalone for the treatment of CVD and its potential use in the setting of HFrEF and HFpEF.

MicroRNA therapeutics

Many miRNA-based therapeutics are currently in preclinical development, and a growing number is reaching clinical trials. A phase II clinical trial (NCT03601052), which started patient recruitment in July 2018 is testing the effect of remlarsen (MRG-201, Miragen Therapeutics)—a miR-29 mimic, which has shown potent anti-fibrotic activity—in preventing or reducing keloid formation in subjects with a history of keloid scars. The results of this trial, which has been recently completed, will pave the way to investigate the effects of remlarsen on ECM remodelling and fibrosis in patients with HF. Regulus therapeutics is also developing miRNA therapeutics such as anti-miR-21, anti-miR-155, and anti-miR–33 for the treatment of fibrotic diseases, inflammation, and cardiometabolic disorders, respectively.¹⁸⁵ In a recent translational study, treatment with the anti-miR-132 was able to reverse cardiac remodelling in a pig model of HF. Of note, the compound exerted favourable pharmacokinetics, safety, tolerability, dose-dependent PK/PD relationships, and high clinical potential for the anti-miR-132 treatment scheme.¹¹⁵ Consistent with these results other studies using CDR132L, a synthetic lead-optimized oligonucleotide inhibitor of the miR-132, reported preclinical efficacy and safety in models of experimental HF. Specifically, CDR132L improved diastolic function while reducing left atrial size and remodelling, important features in HFpEF.¹⁸⁷ Of clinical relevance, a first-in-human phase 1b randomised, double-blind, placebo-controlled study recently showed that miR-132 inhibition was safe and was associated with promising beneficial effects in patients with chronic HF.¹⁸⁶

Targeting gut microbiota to modulate the host epigenome

Many gut metabolites including folates, choline, and butyrate have shown to affect the epigenome at different levels. Alterations in intestinal microbial composition have been well described in patients with HF.¹⁹⁴ Three independent HF cohorts that used sequencing techniques to characterize intestinal microbial compositions reported a consistent decrease in microbial diversity and a depletion of several butyrate producers (i.e. Faecali bacterium, Eubacterium hallii) that were inversely associated with inflammatory biomarkers. A recent study demonstrated that microbiota-derived metabolite promotes epigenetic modifications via enhanced HDAC3 activity in the gut.¹⁹⁵ These findings are of interest since increased HDAC3 expression induced concentric remodelling and thickening of ventricular myocardium in mice.¹⁹⁵ In this perspective, microbiome directed interventions could also represent a promising option to modulate the host epigenome and to suppress chronic inflammation, a key player in HFpEF.

Future perspectives and conclusions

Although epigenetic therapies hold great promises for the future management of CVD, treatment with DNA hypomethylating agents and HDAC inhibitors is generally associated with side effects due to the lack of specificity. However, clinical studies conducted so far in cancer patients did not show any significant effect of these drugs on the incidence of second malignancies or any other severe, unanticipated toxicity. Directing intervention to specific cell types is a major challenge in this setting. Emerging evidence suggests that this is possible for RNA-targeted therapeutics for gene silencing. Recent examples are represented by liver-specific modulation of PCSK9 expression, ANGPTL3, and lipoprotein(a).¹⁸⁵ Another issue related to the study of epigenetic modifications and HF phenotypes is the demonstration of a causal relation between these two factors. The causality of presumed epigenetic events and cardiovascular phenotype can be assessed by using different strategies. One possibility would be using longitudinal studies whereby the epigenome can be profiled before the onset of the disease and during follow-up.¹³⁸ Alternatively, Mendelian randomization—a process that interrogates the causal relationships between exposure, epigenetic marks, and outcome—can serve to establish meaningful hierarchies, thus helping to discriminate between epigenetic phenomena and epiphenomena.¹⁹⁶ The results of large epigenomic studies over the upcoming years will certainly help to decipher the complex link between genetics, epigenetics, and HFpEF, and define and validate the added value of epigenetic information into personalized therapies.¹⁹⁷

Supplementary material

Supplementary material is available at European Heart Journal online.

Funding

F.P. is the recipient of a Sheikh Khalifa’s Foundation Assistant Professorship in Cardiovascular Regenerative Medicine at the Faculty of Medicine, University of Zürich. The present work is supported by the Swiss National Science Foundation (310030_197557 to F.P.), the Swiss Heart Foundation, Swiss Life Foundation, the EMDO Stiftung; Kurt und Senta-Hermann Stiftung, and the Schweizerische Diabetes-Stiftung (to F.P.); the Holcim Foundation and the Swiss Heart Foundation (to S.C.). T.T. is supported by an ERC Grant Longheart and the DFG (KFO311).

Conflict of interest: T.T. is filed and licensed patents on non-coding RNAs. T.T. is founder and shareholder of Cardior Pharmaceuticals GmbH. T.T. reports fees/support by Novo Nordisk, Amicus Therapeutics, Boeringer Ingelheim, Takeda, and Sanofi-Genzyme (all outside the submitted work). C.T. has received speaker fees and/or contributions to congresses from Abbott, Abiomed, Astra Zeneca, Bayer, Berlin Chemie, Pfizer, and Servier (all outside the submitted work).