Epigenetic Modifications in Cardiovascular Aging and Diseases

Aging is broadly defined as a time-dependent functional decline. Aging is an inexorable movement of life and unfortunately poses the greatest threat for several cardiovascular diseases (CVDs), including atherosclerosis, coronary artery diseases, hypertension, heart failure, atrial fibrillation, myocardial infarction, and cardiac hypertrophy.

With age, the heart and vasculature gradually show homeostatic imbalance, vascular stiffening, and fibrosis, as well as increased left ventricular (LV) wall thickness leading to accentuated tissue adaptations and decreased stress tolerance (Figure 1).^1,2 Continuous efforts have been made to delineate the crucial mechanisms behind the deterioration of cardiovascular functions during aging. From these studies, epigenetic alterations have emerged as a crucial link between the intrinsic genetic landscape and extrinsic environmental influences. Various risk factors for CVDs, such as diabetes mellitus, nutrition, smoking, stress, hypertension, and circadian rhythm, often manifest as modifications of epigenetic marks.³

• Download figure
• Download PowerPoint

Epigenetics represents a wide range of changes in gene expression independent of changes in DNA sequence. These changes can occur via DNA methylation or hydroxymethylation, histone modifications, and chromatin remodeling. Recent studies have identified histone variants, microRNAs (miRNAs), and long noncoding RNAs (lncRNAs) as additional epigenetic markers (Figure 1).^4–7 Epigenetic dysregulation is associated with CVD pathophysiologies during aging. For example, at multiple stages, atherosclerosis is closely associated with epigenetic changes in key players, such as RELA, NOS3, KLF4, and APOE, which are regulated through various layers of epigenetic modifications.^8,9

In this article, we review the mechanisms underlying how epigenetic changes contribute to cardiovascular aging and CVDs. We focus on a subset of CVDs, including atherosclerosis and cardiomyopathies, to elucidate the central mechanisms of diverse epigenetic modifications and related enzymes in the pathogenesis of these symptoms. New therapeutic approaches that target the epigenetic machineries for prevention or treatment of CVDs are also discussed.

Interplay Between Cardiovascular Aging and Diseases

Cardiac aging is a complex process characterized by decreased heart functions and ventricular and atrial remodeling. This process includes LV wall thickening because of cardiomyocyte hypertrophy, increased left atrial size, and vascular intimal thickening and stiffening because of collagen and calcium deposition (Figure 1).¹⁰ Among the well-defined hallmarks in aging, including genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, dysregulated nutrient sensing, mitochondrial dysfunction, cellular senescence, stem cell exhaustion, and altered intercellular communication,¹¹ many features are often observed in CVDs as well. For example, telomere shortening is related to CVDs. Telomere attrition reflects the cumulative burden of inflammatory, oxidative, and mechanical stress on the cardiovascular system,¹² thereby linking aging with CVDs.

Arteriosclerosis is referred to as the degenerative stiffness of the artery walls. Notably, a specific form of arteriosclerosis, atherosclerosis, is the cumulative result of endovascular inflammation, lipid oxidation, and plaque formation. Arteriosclerosis is a progressive, diffuse, age-related process.¹³ With aging and atherosclerosis regulating similar biochemical pathways in vascular alterations, vessel aging may be considered a prodromal stage of atherosclerotic diseases, or, conversely, atherosclerosis may be considered a form of accelerated arterial aging. In addition to atherosclerosis, aortic stiffening is among the earliest age-related manifestations that affect blood vessel structure and function¹³; increased aortic wall stiffness may emerge in childhood and progress with age.¹³ Additionally, the prevalence of atrial fibrillation increases with age-related progressive cardiac fibrosis as profibrotic signals are upregulated by aging.¹⁰

DNA Methylation and Demethylation

The most well-known epigenetic DNA modification is methylation at 5-cytosine, which is essential for proper gene expression, transposon silencing, alternative splicing, and genome stability.^14,15 DNA methylation patterns are routinely maintained by DNMT (DNA methyltransferase) 1, while de novo DNA methylation is typically mediated by DNMT3A and DNMT3B. DNA methylation can be reversed via the oxidation of 5-methylcytosine to 5-hydroxymethylcytosine (5-hmC) by the TET (ten-eleven translocation) proteins.¹⁶ There are a variety of strategies for profiling genome-wide DNA methylation, such as whole genome bisulfite sequencing, reduced representation bisulfite sequencing, methylated-DNA immunoprecipitation, and methyl CpG (C-phosphate-G) immunoprecipitation with different depths. Tet-assisted bisulfite sequencing technology enables single-base-resolution detection of 5-hmC¹⁷; furthermore, a new method called oxidative bisulfite sequencing can discriminate between 5-methylcytosine and 5-hmC markers in genomic DNA at single-base resolution.¹⁸ These modifications were initially considered to be stable markers for gene silencing; however, based on these technologies, DNA methylation is now known to change with chronological age.

Several studies have associated DNA methylation with cardiovascular aging and diseases. Cardiovascular risk factors, such as smoking, low dietary folate, and increased plasma homocysteine, induce dysregulated DNA methylation. The methylation status of hundreds of CpG sites has been used as an epigenetic clock to measure human aging rates.^19–21 DNA methylation-based estimation is well correlated with the incidence of CVDs in a 10-year follow-up study of 832 participants, all aged 70 years at the time of examination.²²

In addition, global DNA hypomethylation is observed in atherosclerotic lesions in humans and animal models,^23–25 whereas the promoter regions of atheroprotective genes, such as ESR1/2, ABCA1, and KLF4, are often hypermethylated in atherosclerosis.^3,26–28 To date, the mechanisms underlying the changes in DNA methylation in cardiovascular aging and CVDs remain obscure and controversial. For instance, increased DNMT1 in macrophage is correlated with decreased PPAR-γ (peroxisome proliferator-activated receptor gamma) and increased proinflammatory cytokines in ApoE-knockout mice fed an atherogenic diet, as well as in patients with atherosclerosis.²⁹ Consistently, the treatment of human umbilical vein endothelial cells with LDL (low-density lipoprotein) cholesterol—a major risk factor for coronary heart diseases—results in DNMT1 activation and aberrant gene expression. Conversely, DNMT1 expression is diminished in human atherosclerotic plaques.³⁰

Interestingly, the DNA methylation patterns in CVDs share some common features with developmental alterations. The methylation patterns of cardiomyocytes isolated from failing mouse hearts partially resemble those of neonatal mice.³¹ The 5-hmC landscape also shifts toward a neonatal-like distribution pattern in pathologically hypertrophic hearts. It has been suggested that TET2 may be a master epigenetic regulator of key cardiovascular genes, including MYH7, MYOCD, SRF, and KLF4.³² Additionally, the DNA hydroxymethylation level decreases on the induction of cardiac hypertrophy. Notably, the enrichment of 5-hmC at retrotransposon-LINEs is specific to hypertrophic conditions, indicating genomic instability.³³ Yet, how TETs are implicated in cardiovascular aging and many other CVDs remains elusive.^16,33

Histone Modifications and Chromatin Remodeling

Chromatin remodeling refers to the process by which the eukaryotic genomes wrap into nucleosome-based chromatin. Nucleosome remodeling changes the architecture of chromatin and affects the accessibility of DNA to elements facilitating transcription initiation and elongation, replication, recombination, and repair. One form of chromatin remodeling is the covalent modification of histones by histone acetyltransferases, deacetylases, methyltransferases, and kinases.⁷ Another form of chromatin remodeling is the noncovalent modification mediated by ATP-dependent chromatin remodeling complexes that physically exchange, move, or remove nucleosomes along DNA. Post-translational modifications (PTMs) of core histone components play vital roles in the activation and repression of gene transcription.³⁴ Histone H2A, histone H2B, histone H3, and histone H4 can be modified by methylation, acetylation, ubiquitination, phosphorylation, SUMOylation, GlcNAcylation, carbonylation, and ADP-ribosylation, collectively constituting the histone code.⁷ The protein complexes that add, remove, and recognize various PTMs are referred to as histone writers, erasers, and readers, respectively.^7,35 Growing evidence shows that the dysregulation of epigenetic regulators in histone modifications is a predisposing factor for cardiac aging and diseases.

For monitoring histone modifications and binding of chromatin remodeling factors, chromatin immunoprecipitation combined with high-throughput sequencing is used to infer the distributions of proteins (such as core histone components with specific PTMs, transcriptional factors, and epigenetic enzymes) bound to or associated with DNA.³⁶ This technique is a powerful tool to profile the genome-wide patterns of various histone modifications and to determine chromatin states, improving the current understanding of DNA-protein interactions at a genome-wide scale. In addition, the CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/clustered regularly interspaced short palindromic repeat–associated 9) system serves as a platform for designing artificial transcriptional activators or repressors targeting almost any designated sequence and as an artificial bait for chromatin immunoprecipitation combined with high-throughput sequencing.

Using new multiomics technologies, cardiomyocyte-specific global histone modification signature profiling in human failing hearts reveals that pathological gene expression is often accompanied by the alterations of active histone marks (H3K9ac, H3K27ac, H3K4me3, and H3K36me3) and repressive marks (H3K27me3).³⁷ Chromatin immunoprecipitation combined with high-throughput sequencing against 7 types of histone marks in hypertrophic mouse cardiomyocytes shows that 3 marks are associated with active regulatory regions (H3K9ac, H3K27ac, and H3K4me3), 3 are associated with repressed regulatory regions (H3K9me2, H3K9me3, and H3K27me3), and 1 is preferentially distributed within transcribed genes (H3K79me2).³⁷ Notably, 9.1% of the chromatin of these cardiomyocytes shows changes in the enrichment of at least 1 type of histone mark. These chromatin remodeling changes are responsible for the gene regulation contributing to cardiac hypertrophy. For example, the downregulation of genes clustered in oxidative stress and gene transcription regulation pathways are associated with epigenetic changes, including decreased H3K9ac, increased H3K9me3, and increased H3K27me3.³⁷

Histone Methylation and Demethylation

H3K27me3—one of the most well-studied repressive marks—is governed by the histone methyltransferase of the PRC (polycomb repressive complex) 2, which contains the catalytic subunit EZH2 (enhancer of zeste homolog 2) and the H3K27-specific JmjC-domain demethylases UTX (lysine-specific demethylase 6A) and JMJD3 (lysine-specific demethylase 6B; Figure 2). EZH2 prevents cardiac pathology by repressing the expression of Six1—a homeodomain transcription factor functioning in cardiac progenitor cells that is silenced on cardiac differentiation (Figure 2).³⁸ In contrast, the cardiomyocyte-specific overexpression of Rae28—a component of PRC1—causes cardiomyocyte apoptosis, abnormal myofibrils, and dilated cardiomyopathy.³⁹ By comparison, an active histone mark, H3K79me3, is catalyzed by the histone-lysine N-methyltransferase DOT1L (histone H3-K79 methyltransferase) in mammals. DOT1L is downregulated in dilated cardiomyopathy; its cardiac-specific deletion results in the global loss of H3K79me2/3, including the loss of H3K79me2/3 enrichment on the DMD (dystrophin) gene, which encodes a membrane-associated protein whose mutation causes not only dilated cardiomyopathy but also muscular dystrophy. Consistently, DMD protein level is decreased 75% in DOT1L-deficient hearts that recapitulate the phenotypes in patients with dilated cardiomyopathy.⁴⁰ Similarly, in patients with dilated cardiomyopathy, H3K9me2/3 decreases, whereas H3K4me2 increases with increased MLL3 (myeloid/lymphoid or mixed-lineage leukemia protein 3) in the LV. Implantation of an LV assist device reverses cardiomyocyte remodeling, along with the upregulation of the H3K9me3 methyltransferase SUV39H1 (suppressor of variegation 3-9 homolog 1) and the downregulation of the H3K9me3 JMJD demethylases (Figure 2).^41,42

• Download figure
• Download PowerPoint

In the atherosclerotic plaques of patients with carotid artery stenosis, changes in histone modifications facilitate the development and progression of atherosclerosis. There are significantly reduced H3K9 and H3K27 methylation levels, particularly in smooth muscle cells (SMCs) and lymphocytes. With the progression of atherosclerosis, H3K4 methylation increases in SMCs; H3K9ac and H3K27ac also increase in atherosclerotic SMCs and macrophages, correlating with the increase in histone acetyltransferase activity of GCN5L (GCN5-like protein 1) and MYST1 (histone acetyltransferase KAT8). In addition, in endothelial cells, H3K9ac increases in atherosclerotic plaques.⁴³ Mutations in genes involved in epigenetic regulation have been associated with congenital heart defects.^44,45 In particular, mutations in human LMNA cause laminopathies. The best-characterized laminopathy is Hutchinson-Gilford progeria syndrome (HGPS)⁴⁵—a rare premature aging disease in which most patients die from atherosclerotic CVDs at ≈13 years of age. There are striking commonalities in the epigenetic defects between HGPS and physiological aging. In human stem cell-based HGPS models, the mark for facultative heterochromatin H3K27me3 and its methyltransferase EZH2 is diminished; the mark for constitutive heterochromatin H3K9me3 also decreases with its binding partner HP1α (chromobox protein homolog 5; also known as CBX5).^46–50 Progressive disorganization of heterochromatin also occurs in a Werner Syndrome (WS)–specific stem cell model.⁵⁰ Future systemic studies will depict the epigenetic changes in isogenic HGPS and WS stem cell models to elucidate the mechanisms underlying premature and normal aging and to identify effective treatments for atherosclerosis.⁵⁰

The aforementioned studies suggest that in cardiac aging, the loss of repressive marks and the gain of active marks may result in the upregulation of aging-promoting genes, aberrant activation of repetitive sequences, and an increase in transcriptional noise. These findings are consistent with the observation that at the single-cell level, cardiomyocytes isolated from 27-month-old mice exhibit considerable variations in the transcription levels of a panel of heart-specific housekeeping genes compared with those of young mice.⁵¹ The reversible nature of epigenetic modifications makes it possible to regulate and reverse some phenotypes by physical activity and dietary restriction.^52–56 In a North Texas Health study, participants with high vegetable and fruit intake have higher LINE-1 methylation levels than those on a diet of mostly meats and carbohydrates.⁵⁷ Flavonoids exhibit beneficial effects against heart diseases because of their effects on DNMT and a decrease in gene-specific methylation, demonstrating the potential of nonpharmacological adjuncts as therapies for aging individuals and patients with CVDs.⁵⁸

Although important roles for histone PTMs and chromatin remodeling in cardiovascular aging and CVDs have been revealed, how the histone code is regulated remains unknown. Continuous efforts are devoted to understanding how histone writers, erasers, and readers function in aging. For example, in normal adult hearts, the H3K9 dimethyltransferases G9a/GLP (histone-lysine N-methyltransferase EHMT1/2) interact with EZH2 and the transcription factor MEF2C (myocyte-specific enhancer factor 2C) to maintain the heterochromatin needed for repressing fetal genes,⁵⁹ thus protecting against pathological cardiac hypertrophy. However, pathological hypertrophy drives the expression of miR-217 and blocks G9a/GLP, resulting in a reduction of H3K9me2, which unlocks the terminally differentiated state of cardiomyocytes and derepresses the fetal gene program, leading to pathological remodeling of the cardiomyocyte transcriptome.⁶⁰ In another case, the forced expression of a H3K9me3 demethylase, JMJD2A (lysine-specific demethylase 4A), exacerbates cardiac hypertrophy under stress, whereas the loss of JMJD2A attenuates the hypertrophic response.⁶¹

HDACs

Another focus of the pioneering studies on histone modifications in cardiovascular aging includes HDACs (histone deacetylases). There are 4 distinct subtypes of HDACs: class I HDACs (HDACs 1, 2, 3, and 8), class II HDACs (IIa: HDACs 4, 5, 7, and 9; IIb: HDACs 6 and 10), class III nicotinamide adenine dinucleotide-dependent SIRT (sirtuin) enzymes (Sirt1–7), and the sole class IV enzyme HDAC 11.^35,62 The activities of these HDACs are directly modulated by metabolic intermediates, such as nicotinamide adenine dinucleotide, acetyl-CoA, and β-hydroxybutyrate, which in turn transduce the impact to epigenetic changes.^35,63 How these HDACs and SIRTs trigger the transcriptional shift in heart or vascular failure is partially elucidated.⁶⁴

Various studies in mouse models with HDAC deletion or overexpression have elucidated the function of HDACs in the regulation of cardiac hypertrophy. Simultaneous cardiomyocyte-specific deletion of HDAC1 and HDAC2 results in dilated cardiomyopathy and neonatal lethality.⁶⁵ Mice with a cardiac-specific loss of HDAC3 develop massive cardiac hypertrophy and survive only 3 to 4 months.⁶⁶ Mice lacking either HDAC5 or HDAC9 are sensitive to hypertrophic stimuli, and simultaneous knockout of both genes leads to spontaneous cardiac enlargement. Loss of HDAC7 results in the thinning of myocardial walls and dilation of atria.³⁵ Thus, HDACs are critical for heart homeostasis. Notably, most of the aforementioned phenotypes attribute to abnormalities in the growth and maturation of the cardiomyocytes. To date, little is known on how specific HDACs regulate epigenetic modifications and affect the gene expression involved in aging and aging-related diseases. Sporadic studies demonstrate that HDAC9 expression is increased 3-fold in carotid plaques compared with that in normal arteries; a similar upregulation occurs in aortic and femoral plaques.⁶⁷ In addition, various polymorphisms of HDAC9 are associated with a risk of large-vessel stroke via promoting atherosclerosis.^68–70

The beneficial effects of SIRTs as deacetylases against a variety of cardiovascular pathologies, including endothelial dysfunction, atherothrombosis, myocardial infarction, and LV hypertrophy, have been well documented and extensively discussed elsewhere^64,71–78 and are, therefore, not discussed in detail in this review. Interestingly, 8 types of nonacetyl short-chain Lys acylations on histones utilizing acyl-CoA metabolites as substrates have been identified: succinylation, crotonylation, malonylation, butyrylation, propionylation, 2-hydroxyisobutyrylation, glutarylation, and β-hydroxybutyrylation.⁷⁹ In addition to acting as deacetylases, SIRTs can mediate short-chain Lys deacylations, such as depropionylation and debutyrylation. In particular, SIRT5 has little deacetylase activity but is a robust desuccinylase. SIRT5 knockout mice develop hypertrophic cardiomyopathy with protein lysine succinylation predominantly accumulating in the heart. Further investigations will help understand how different SIRTs orchestrate the proportional mixture of distinct histone acylations in response to different metabolic stimuli.

Chromatin Remodeling

There are at least 4 distinct families of chromatin remodelers: SWI/SNF (switch/sucrose nonfermentable), ISWI (imitation SWI of drosophila melanogaster), NuRD (nucleosome remodeling deacetylase), and INO80 (chromatin-remodeling ATPase INO80). These complexes can dissect the original histone-DNA contacts, thus leading to nucleosome repositioning. Emerging evidence has implicated chromatin remodelers in cardiovascular aging and CVDs. For example, a Ser818Cys mutation in INO80D—a subunit of INO80—accelerates arterial aging in humans.⁸⁰ In addition, several NuRD complex components are downregulated during HGPS premature and normal aging.⁴⁵

The most well-studied chromatin remodeling complex SWI/SNF, also called the BAF complex, consists of 12 protein subunits with either BRG1 (ATP-dependent helicase SMARCA4) or BRM (ATP-dependent helicase SMARCA2) as the essential ATPase subunit. Accumulating data have shown that the SWI/SNF complex is highly associated with cardiovascular function and pathology, positioning this complex as a potential therapeutic target for pathological cardiac hypertrophy and heart failure. A 70% decrease in BRG1 expression is reported in the myocardium of patients with congenital heart diseases. In mouse and human hypertrophic hearts, BRG1 is activated and assembles with G9a/GLP and DNMT3 to repress the expression of MYH6 (myosin-6) by H3K9/CpG methylation in its promoter region (Figure 3).⁸¹ BRG1 also interacts with its embryonic partners, PARP (poly ADP ribose polymerase) and HDACs, to regulate gene expression during cardiac hypertrophy in mice (Figure 3).⁸² In addition to BRG1, in hypertrophic hearts resulting from salt-induced hypertension, several other SWI/SNF subunits, including Baf180 and Baf60c, that are essential for heart development are upregulated.⁸³ Moreover, alterations in the association of the SWI/SNF complex with genomic DNA have been observed in injured adult mouse hearts.⁸⁴ Therefore, on BRG1 activation, all cardiac stimuli ultimately converge on the BAF-mediated regulation of key cardiac function genes, such as MHC, MYH6, and ABCA1, and the cardiac-specific lncRNA MHRT (myosin heavy chain-associated RNA transcript) to induce CVDs.^85,86 Conversely, the inhibition of BRG1 or other SWI/SNF components, such as Baf60c and Baf180, blunts cardiac regeneration.⁸⁷ The controversy between these findings suggests that the precise mechanisms of how the SWI/SNF complex contributes to the aging process require further investigations.

• Download figure
• Download PowerPoint

In addition, BRG1 is upregulated in patients with aortic dissection and dilation. In vitro BRG1 overexpression upregulates genes clustered in RRAD (Ras-related associated with diabetes) and inhibits migration and proliferation of human SMCs.⁸⁸ Another component of the SWI/SNF complex, Baf60a, is a diet-sensitive regulator of cholesterol homeostasis; liver-specific Baf60a inactivation protects mice from diet-induced atherosclerosis (Table 1).⁸⁹

In summary, nucleosome dynamics, occupancy, and positioning within chromatin is influenced by chromatin remodelers. Enzymatic complexes, such as the BAF complex, reposition chromatin to facilitate access to condensed genomic DNA by gene-regulatory elements. The loosening state of chromatin regions may correlate with dysregulated gene transcription in cardiac aging and diseases. By using novel genome-wide chromatin conformation capture (Hi-C) and transposase-accessible chromatin sequencing technology, open chromatin landscapes during aging can be profiled, thereby enhancing current understanding of how dysregulation of a chromatin remodeling complex contributes to cardiovascular aging and diseases.

miRNAs, lncRNAs, and Circular RNAs

A large part of the human genome is transcribed into noncoding RNAs (ncRNAs). These ncRNAs play important roles in age-associated CVDs and have been viewed as potential biomarkers and therapeutic targets. There are several subfamilies of ncRNAs, such as linear small-size miRNAs (<200 nucleotides) and lncRNAs (>200 nucleotides), as well as circular RNAs (circRNAs) consisting of a covalently closed continuous loop. Deep RNA sequencing followed by analyses with different algorithms can be used to detect genome-wide mRNAs, miRNAs, lncRNAs, and circRNAs.⁹⁰ Here, we mainly focus on lncRNAs and circRNAs in cardiac aging and diseases.

LncRNAs

lncRNAs regulate multiple biological pathways in cardiac development, aging, and diseases. Different from miRNAs, lncRNAs are poorly conserved across species. There are ≈100 000 of lncRNAs in the human genome, and many of these molecules are cardiac enriched.⁸⁶ LncRNAs are classified into 6 categories based on their genomic locations, including the dominant intergenic lncRNAs between 2 protein-coding genes, intronic lncRNAs in the introns of protein-coding genes, sense and antisense lncRNAs in the sense or antisense strands of protein-coding genes, bidirectional promoter lncRNAs in promoter regions, and enhancer lncRNAs in the enhancer regions of the genome.

In recent years, many studies have profiled the changes in lncRNAs in CVDs and aging by deep sequencing.⁹¹ lncRNA profiling of mouse hearts with myocardial infarction has identified hundreds of novel cardiac-specific lncRNAs proximal to cardiac-specific enhancers, contributing to maladaptive cardiac pathological remodeling.⁹² Some of these lncRNAs are conserved in humans and dynamically modulated in CVDs, including dilated cardiomyopathy and aortic stenosis. In nonischemic myocardial biopsies from patients with dilated ischemic cardiomyopathy and heart failure, 9 lncRNAs (CDKN2B-AS1/ANRIL, RMRP, RNY5, SOX2-OT, SRA1 EGOT, H19, HOTAIR, and LOC285194/TUSC7) are significantly modulated. RMRP, H19, and HOTAIR are also upregulated in hypertrophic hearts in a mouse model.⁹³ H19 binds to miR-103/107 and targets the FADD (Fas-associated protein with death domain) pathway in a myocardial ischemia/reperfusion injury model,⁹⁴ whereas HOTAIR is a competing endogenous RNA that regulates PTEN (phosphatidylinositol 3,4,5-trisphosphate 3-phosphatase and dual-specificity protein phosphatase) by inhibiting miRNA-19 and miRNA 29b.^93,95 Intriguingly, CDKN2B-AS1/ANRIL, HOTAIR, and LOC285194/TUSC7 are similarly upregulated in the peripheral blood mononuclear cells of these patients, suggesting their potential as novel blood-based predictive biomarkers for heart failure.⁹⁶ Other candidate lncRNA markers include MIAT (myocardial infarction-associated transcript) and SENCR (smooth muscle and endothelial cell-enriched migration/differentiation-associated lncRNA), which is related to diastolic dysfunction and myocardial infarction.^97,98 Single-cardiomyocyte nuclear transcriptomes of normal and failing hearts further reveal nodal lncRNAs as key regulators of cardiomyocyte cell cycle.⁹⁹ Importantly, Wisper (Wisp2 superenhancer-associated RNA)—a cardiac fibroblast-enriched lncRNA—correlates with the severity of fibrosis in diseased human hearts. Wisper controls the expression of profibrotic lysyl hydroxylase 2, which in turn regulates extracellular matrix deposition and the proliferation and survival of heart fibroblasts. Depletion of Wisper attenuates cardiac fibrosis in mice,¹⁰⁰ highlighting its therapeutic potential for cardiac fibrosis.¹⁰⁰

A majority of lncRNAs are present in the nucleus. Emerging evidence shows that lncRNAs recruit epigenetic factors and trigger chromatin remodeling, thereby leading to the activation or repression of genes in the nucleus.¹⁰¹ Expression of the Myheart or MHRT—a highly sequentially and structurally conserved lncRNA transcribed divergently to MYH6 and overlapping with MYH7 in the antisense direction—gradually increases from the fetus to adulthood and serves as a safeguard for cardiac health in mice and humans.¹⁰² MHRT binds to the helicase domain of BRG1, antagonizing BRG1 occupancy on its chromatin targets and suppressing the pathological Myh6-to-Myh7 switch, thus protecting mouse hearts from cardiomyopathies (Figure 3). Under pathological stress, activated BRG1 forms a chromatin repressor complex with HDAC2/9 and PARP to repress MHRT transcription, and thus the heart loses the protection from MHRT (Figure 3).¹⁰² In addition, MHRT ameliorates cardiac hypertrophy via myocardin—a strong smooth and cardiac muscle-specific transactivator that binds to promoters harboring CArG box motif and activates a panel of downstream muscle-specific genes, including TAGLN (transgelin), ACTA2 (aortic smooth muscle), and MYH11 (myosin-11). MHRT recruits HDAC5 to myocardin, resulting in the deacetylation or direct inhibition of MYOCD gene transcription (Figure 3). Conversely, as part of a positive feedback loop, myocardin can promote the expression of MHRT by binding to its promoter.¹⁰³ In addition to MHRT, BRG1 interacts with MANTIS—a scaffolding lncRNA—to stabilize the SWI/SNF chromatin remodeling complex that mediates the transcription of important angiogenesis genes, such as SOX18 (transcription factor SOX-18), SMAD6 (mothers against decapentaplegic homolog 6), and COUP-TFII (COUP transcription factor 2), by ensuring efficient RNA polymerase II binding.¹⁰⁴ Furthermore, a cardiac-enriched lncRNA, cardiac hypertrophy-associated epigenetic regulator, is implicated in maladaptive cardiac remodeling via direct interaction with the catalytic subunit of PRC2 and inhibition of histone H3 lysine 27 methylation, thus regulating genes involved in cardiac hypertrophy.¹⁰⁵

In summary, these paradigms for lncRNA-chromatin interactions show that lncRNAs play roles in shaping genome organization and controlling gene expression levels in the cardiovascular system.^101,106 New technologies, such as CRISPR/Cas9-based gain/loss of function, RNA immunoprecipitation, and sequencing, will provide a comprehensive understanding of the lncRNA biology and identify therapeutic targets for the prevention and treatment of CVDs within reach.¹⁰⁷ Given the different expression of lncRNAs among various species, it will be necessary to focus on the pathways regulated by lncRNAs or to directly study the lncRNAs in primates.

circRNAs

In recent years, circRNAs have added another layer of complexity to ncRNA biology. circRNAs are pervasively expressed in mammalian tissues and often act as transcriptional/translational regulators or miRNA sponges.^90,108–111 Approximately 15 318 and 3017 cardiac circRNAs have been identified in the human and mouse, respectively.¹¹² In human and mouse failing hearts, the abundance of circRNAs increases along with the re-expression of fetal genes, such as NPPA and MYH7.¹¹³ By microarray expression profiling, 63 differentially expressed circRNAs in diseased mouse hearts have been identified.¹¹⁴ Several circRNAs (such as circRNAs from CHD7 (chromodomain-helicase-DNA-binding protein 7) and DNAJC6 (putative tyrosine-protein phosphatase auxilin) are dynamically expressed in response to stress in human stem cell-derived cardiomyocytes.¹¹⁵ Although multiple studies have profiled the expression patterns of circRNAs in the cardiovascular system, their functions have yet to be defined.¹¹²

The circAmotl1 is identified as a circRNA interacting with PDK1 (3-phosphoinositide-dependent protein kinase 1) and AKT1 (proline-rich AKT1 substrate 1), leading to phosphorylation and nuclear translocation of AKT1 and thus enhancing cardiomyocyte survival.¹¹⁶ This circRNA dramatically decreases in aged human hearts. Inversely, a circRNA generated from FOXO3 (Forkhead box protein O3), which forms a complex with p21 (cyclin-dependent kinase inhibitor 1) and CDK2 (cell division protein kinase 2) and abolishes the function of CDK2 to promote the cell cycle,¹¹⁷ is highly expressed in the heart tissues of aged mice and human. The ectopic expression of circFoxo3 induces cellular senescence in cultured mouse embryonic fibroblasts¹¹⁸ and aggravates cardiomyopathy, whereas silencing circFoxo3 alleviates these symptoms.

A recent pioneering study has linked circRNAs to the expression of aging-associated genes. circANRIL is the first circRNA found in the p16INK4A locus.^119–121 The expression of circANRIL is correlated with p16INK4A/p14ARF transcription, as well as the risk of atherosclerotic vascular diseases. Initially identified as a marker of aberrant RNA splicing with no specific functions, circANRIL binds to a core subunit of the 60S-preribosomal assembly factor and impairs ribosome biogenesis in vascular SMCs. Consequently, circANRIL induces nucleolar stress and p53 activation, thus slowing down the overproliferation of SMCs in atherosclerosis. In addition, the expression of circANRIL increases in patients with coronary artery diseases, indicative of an atheroprotective role.¹¹⁹ In addition to circANRIL, circRNA hsa_circ_0010729 targets the HIF-1α (hypoxia-inducible factor 1-alpha) pathway by sponging miR-186,¹²² and circRNA WDR77 (methylosome protein 50) targets FGF-2 (fibroblast growth factor 2) by antagonizing miR-124; both of these molecules regulate vascular endothelial cell proliferation (Table 2).¹²³

circRNA is an extracellular, stable, and detectable molecule in the bloodstream, suggesting its potential as a candidate noninvasive biomarker of CVDs.¹²⁴ In fact, some lncRNAs and circRNAs have not only been identified as vital biomarkers but also as promising therapeutic targets in heart diseases.^124,125 Based on the global transcriptomic analyses of plasma RNAs from patients with myocardial infarction, circulating long noncoding RNA LIPCAR has been identified as a novel biomarker for cardiac remodeling and for predicting survival in patients with heart failure.¹²⁶ In peripheral blood mononuclear cells, Hsa_circ_0005836, hsa_circRNA_025016, and hsa_circ_0124644 were identified as diagnostic biomarkers for various CVDs.^127,128 With increasingly more functions of circRNAs discovered, the use of biochemical enrichment strategies and deep sequencing coupled with novel bioinformatics approaches will further provide a comprehensive understanding of circRNAs in mammalian cardiac aging and diseases.

mRNA Modifications—a New Era of the Epigenome

In recent years, the discovery of mRNA modifications has opened a new realm of posttranslational regulations, also called the RNA code or epitranscriptome. N⁶-methyladenosine (m⁶A), referring to the methylation of the adenosine base at the nitrogen-6 position, is the most prevalent internal mRNA modification linked to increased RNA stability and translational efficiency and accuracy.^{52,53,129–135} The relationship between m⁶A and CVDs is just emerging. It is particularly intriguing that polymorphisms of the FTO (fat mass and obesity associated) gene—an m⁶A demethylase—are risk factors not only for obesity and metabolic abnormalities but also for CVDs.^136,137 FTO protein level decreases with increased global m⁶A in ischemic mouse and human hearts; by comparison, AAV-mediated FTO overexpression significantly improves cardiac function in mice.¹³⁸ Recent evidence shows that the NSun2 (tRNA [cytosine(34)-C(5))-methyltransferase]) methylates ICAM-1 (intercellular adhesion molecule 1) mRNA in the 5' untranslated region and 3' untranslated region, as well as the coding sequence, promoting ICAM-1 protein expression—a feature of the inflammatory response in endothelial cells.¹³⁹

Another conserved and pervasive form of RNA modification, adenosine-to-inosine RNA editing,^140,141 also contributes to atherosclerosis. In endothelial cells, adenosine deaminase ADAR1 (double-stranded RNA-specific adenosine deaminase) acts on CTSS (cathepsin S) mRNA,¹⁴² promoting the recruitment of HuR (human antigen R) and thereby stabilizing CTSS mRNA. ADAR1 knockdown downregulates CTSS transcripts, whereas ADAR1 overexpression increases the adenosine-to-inosine RNA modification of CTSS mRNA and the expression of CTSS protein. Increased adenosine-to-inosine RNA editing of CTSS is observed in patients with atherosclerotic diseases.¹⁴² Our knowledge about the biological implications of epitranscriptomic regulations and especially their spatiotemporal changes under various physiological and pathological conditions is still limited.

To profile the changes in epitranscriptomic marks during cardiovascular aging, DNA–RNA immunoprecipitation followed by sequencing has identified thousands of ncRNAs in the human genome.¹⁴³ Chromatin isolation by RNA purification may also be used to intersect interactomes of RNA and chromatin at genome scale. In fact, it has already been applied to genomic mapping of long ncRNA occupancy in the genome.¹⁴⁴ In addition, a variety of novel sequencing technologies, such as m⁶A-seq, MeRIP-seq, PA-m⁶A-seq, miCLIP, and m⁶A-LAIC-seq can be used to further depict the m⁶A modification landscape of cardiovascular tissues in a transcriptome-wide manner and reveal the fundamental roles of RNA modifications in cardiac aging and diseases.¹³³

Omics Technologies and Systemic Genomics Approaches

With the aforementioned similarities in chromatin epigenetics between cardiovascular aging and CVDs, it would be interesting to use novel 3-dimensional genome techniques to investigate whether aged cardiomyocytes or vascular cells are associated with a plastic chromatin architecture. Advances in multiomics technologies, encompassing genomics, methylomics, epigenomics, transcriptomics, proteomics, and metabolomics will facilitate the revolutionization of the current understanding of epigenetics by probing chromatin remodeling at unprecedented spatial and temporal resolutions.^145–147

Long-range chromatin interactions and chromatin accessibility are robustly associated with gene expression regulation. By using new genomic tools, condensed or relaxed chromatin compartments can be mapped genome wide. For example, active regulatory regions, such as promoters and enhancers, located in open regions can be systemically identified across the human genome.³⁶ Hi-C and chromatin interaction analysis with paired-end-tag sequencing are widely used to analyze chromatin interactions. Hi-C illustrates an accurate 3-dimensional architecture on a genome-wide scale, whereas chromatin interaction analysis with paired-end-tag sequencing reveals information about the 3-dimensional genome organization governed by a particular transcriptional factor or epigenetic regulator. Additionally, DNA adenine methyltransferase identification, sequencing of micrococcal nuclease sensitive sites, DNase I hypersensitive site sequencing, and transposase-accessible chromatin sequencing assays are all powerful and sensitive methods to probe genome-wide chromatin accessibility.³⁶ High-resolution mapping of the 3-dimensional epigenome by Hi-C has suggested a plastic epigenome in failing hearts. A global structural remodeling of chromatin, including a significant decrease in surrounding local chromatin interactions and remodeling of long-range interactions of cardiac enhancers, underpins heart failure.¹⁴⁸ Notably, cardiac-specific knockout of CTCF (CCCTC-binding factor)—an insulator protein in vertebrates—is sufficient to induce heart failure.¹⁴⁸

Every step in epigenetic regulation is governed by corresponding enzymes. Enzymes that alter nucleosome structure or position are at the center of gene and genome regulation. Cutting-edge proteomic technologies are used to detect global changes in the expression levels, PTMs (methylation, acetylation, succinylation, crotonylation, malonylation, and so forth), and interacting factors of these key epigenetic enzymes during cardiac aging and diseases. In particular, high-resolution, mass spectrometry-based proteomics further enables proteome-wide characterization and quantification of acetylation, ushering in the age of acetylomics.

Bulk cells have primarily been used in omics analyses during the last decades. However, different types of cells in the cardiovascular system may be highly heterogeneous during aging and disease progression. Single-cell genomics, such as massively parallel single-cell RNA-seq, facilitate detailed transcriptome analysis to identify variants of key epigenetic enzymes/pathways in specific diseased cohorts or cell types.^54,57,58,146 Altogether, new sequencing technologies have generated vast amounts of data. Systemic genomics approaches with metadimensional and multistaged algorithms will help depict a comprehensive picture of how multilayers of hierarchal chromatin organization govern gene regulation.

Conclusions and Perspectives

The current knowledge of the role of epigenetics in cardiovascular aging and diseases has substantially increased in the last decade. It has become clear that alterations in chromatin remodeling, DNA occupancy, and ncRNA expression all contribute to the progress of cardiovascular aging and diseases. However, our understanding of epigenetic regulation during aging in cardiovascular system remains fragmentary; it is debatable whether the overall changes in the epigenetic patterns return to an embryonic state or only shift to open chromatin. In illustrating the epigenetic landscape during pathological and physiological aging, the current studies have focused on the pathogenesis of CVDs but not merely the aging process itself. How aging-related epigenetic changes directly contribute to the transformation in CVDs remains to be unraveled. In addition, epigenetic factors co-occur and are closely interconnected during aging, but which are the primary driving forces and which are just consequences? Furthermore, how environmental stimuli, such as smoking and drinking, influence chromatin architecture in cardiac aging and diseases remains obscure. Environmental stimuli may influence the ROS (reactive oxygen species) level or DNA damage to affect the epigenetic stage or change the level of metabolites, such as nicotinamide adenine dinucleotide, acetyl-coenzyme A, or S-adenosylmethionine, to regulate the activity of chromatin-modifying enzymes.⁵⁵ Novel multiomics technologies combined with integrative approaches make it feasible to profile different layers of epigenetic landscapes in a spatiotemporal manner, which will be invaluable in understanding the molecular mechanisms involved in cardiac aging and diseases. Time-resolved analyses of epigenetic responses to external stimuli illustrate the environment-epigenome interactions that modulate aging. Moreover, CRISPR/Cas9 editing technology makes it feasible to target specific epigenetic enzymes and manipulate epigenetic pathways in a locus-specific manner, facilitating the elucidation of the roles of particular epigenetic modifications. Isogeneic human stem cells can be created and differentiated into heart or vessel cells carrying mutations of interest from patients with CVDs.^{46,48–50,56,149,150} New findings can be validated in vivo in human cardiac tissues or at least in nonhuman primates. For future clinical applications, profiling studies of large cohorts is needed to investigate the feasibility of using epigenetic markers for the prognosis and diagnosis of human CVDs. Hopefully, these combined approaches will provide a deeper understanding of the mechanisms underlying cardiovascular aging and diseases and eventually facilitate the development of effective interventions for human cardiovascular disorders.

Acknowledgments

Because of limitations of space, we apologized for not being able to cite all important studies in this review. This work was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA16010100), the National Key Research and Development Program of China (2015CB964800, 2017YFA0103304, 2017YFA0102802, 2014CB910503, and 2018YFA0107203), the National Natural Science Foundation of China (grant No. 81625009, 81330008, 91749202, 91749123, 81371342, 81471414, 81422017, 31671429, 81601233, 81671377, 31601109, 31601158, 81771515, 81822018, 81870228, 81801399, 81801370, 31801010, and 81701388), the National High Technology Research and Development Program of China (2015AA020307), CAS (KJZDEW-TZ-L05 and CXJJ-16M271), Program of Beijing Municipal Science and Technology Commission (Z151100003915072), the Thousand Young Talents Program of China, Advanced Innovation Center for Human Brain Protection (117212), and Beijing Municipal Commission of Health and Family Planning (PXM2018_026283_000002).

Footnotes