Epigenome alterations in aortic valve stenosis and its related left ventricular hypertrophy

Aortic valve stenosis is the most common cardiac valve disease, and with current trends in the population demographics, its prevalence is likely to rise, thus posing a major health and economic burden facing the worldwide societies. Over the past decade, it has become more than clear that our traditional genetic views do not sufficiently explain the well-known link between AS, proatherogenic risk factors, flow-induced mechanical forces, and disease-prone environmental influences. Recent breakthroughs in the field of epigenetics offer us a new perspective on gene regulation, which has broadened our perspective on etiology of aortic stenosis and other aortic valve diseases. Since all known epigenetic marks are potentially reversible this perspective is especially exciting given the potential for development of successful and non-invasive therapeutic intervention and reprogramming of cells at the epigenetic level even in the early stages of disease progression. This review will examine the known relationships between four major epigenetic mechanisms: DNA methylation, posttranslational histone modification, ATP-dependent chromatin remodeling, and non-coding regulatory RNAs, and initiation and progression of AS. Numerous profiling and functional studies indicate that they could contribute to endothelial dysfunctions, disease-prone activation of monocyte-macrophage and circulatory osteoprogenitor cells and activation and osteogenic transdifferentiation of aortic valve interstitial cells, thus leading to valvular inflammation, fibrosis, and calcification, and to pressure overload-induced maladaptive myocardial remodeling and left ventricular hypertrophy. This is especcialy the case for small non-coding microRNAs but was also, although in a smaller scale, convincingly demonstrated for other members of cellular epigenome landscape. Equally important, and clinically most relevant, the reported data indicate that epigenetic marks, particularly certain microRNA signatures, could represent useful non-invasive biomarkers that reflect the disease progression and patients prognosis for recovery after the valve replacement surgery.


Background
Aortic valve stenosis (AS) is the most frequent heart valve disease among adults in the Western societies with ever increasing prevalence due to the rapidly aging population [1][2][3]. According to the recent populationbased studies performed in Europe and North America the pooled prevalence of total and symptomatic severe AS cases in the general elderly population (≥ 75 years of age) is estimated to 12.4, and 3.4%, respectively [4]. Moreover, with current trends in the population demographics by the year 2050 there will be an estimated 2.1 million European and 1.4 million North American patients with symptomatic severe AS [4]. Furthermore, the even more pronounced demographic changes in Africa, Asia, and South America will further increase the absolute number of AS patients [2,4,5]. Therefore, in a very recent future, AS is likely to become a major health and economic burden facing the worldwide societies [2,4,5].
The major cause of AS is thickening, fibrosis, and calcification of a previously normal tricuspid valve (TAV) or a congenital bicuspid valve (BAV), while a historically prevailing rheumatic heart valve disease, and still the most common cause of AS in developing countries accounts for only 10% of diagnosed cases [2,6,7].
No matter the cause, development of AS starts with the risk of leaflet changes and progresses over many years from early lesions to subsequent narrowing (stenosis) of the aortic valve orifice [8]. During that time, the genetic predisposition or otherwise induced faulty valve repair system in concordance with continuous blood born mechanical forces and proatherogenic risk factors (i.e., hyperhomocysteinemia, hyperlipidemia, abnormal calcium metabolism, smoking, metabolic syndrome, diabetes, hypertension, chronic renal failure, male gender, age) leads to endothelial dysfunctions followed by disruption of the subendothelial basement membrane, extracellular accumulation of plasma-derived atherogenic lipoproteins and infiltration/activation of monocyte-macrophage cells, mast cells, and T lymphocytes [3,[9][10][11][12][13][14]. That leads to intracellular lipid deposition, generation of oxidative stress with accumulation of oxidized lipids and apolipoproteins, foamy cells formation, and upregulation of various pro-fibrotic and proinflammatory factors with concomitant inhibition of plasma derived or locally presents anti-calcific proteins. Acting together, these factors promote extensive extracellular matrix remodeling, and activation of signaling pathways that promote neovascularization, inflammation and calcification [3,[9][10][11][12][13][14]. Concomitant transformation of normally quiescent valvular interstitial cells (qVICs) to active myofibroblastic (aVICs) phenotype in the valve interstitium and their subsequent differentiation to osteoblast-like cells (obVICs) with activation of pro-osteogenic signaling pathways is thought to be one of the central mechanisms contributing to the initiation and progression of AS [15]. In addition, a subset of aortic valvular endothelial cells (VECs) undergoing endothelial-to-mesenchymal transition (EMT) and/or circulating osteoprogenitor cells (COPCs) may also contribute to valvular calcification/ossification either by redifferentiating to an osteoblast-like phenotype or by promoting VICs activation through paracrine signaling [16][17][18].
Based on this timely dependent change of tissue organization, the disease has been divided in two successive functional categories: aortic valve sclerosis (ASc) and aortic valve stenosis. ASc represents the initial, clinically mostly silent stage of disease with calcification and mild fibrous thickening of the aortic leaflets without marked functional obstruction of left ventricular outflow [8,19]. Contrary, aortic valve stenosis as the more advanced stage of the disease comprise serious impairment of leaflet motion with subsequent obstruction of blood flow resulting in maladaptive left ventricular hypertrophy (LVH), myocardial fibrosis (MF), and a propensity for systolic and diastolic dysfunction and heart failure (HF) [3,8,20].
Currently there are no effective pharmacological remedies to prevent or slow the progression of AS and aortic valve replacement (AVR) either surgical or less invasive transcatheter (TAVR) approach is still the only clinical therapy at hand for its successful treatment [21,22,23]. Thus, a better understanding of mechanisms involved in the pathogenesis and progression of AS could lead to novel diagnostic, prognostic and therapeutic targets and eventual development of noninvasive therapeutic options.
Recent observations suggest that the full pathological spectrum of AS lesions cannot be entirely accounted for by hereditary predisposition or growing list of differentially expressed genes. Moreover, traditional genetic views do not sufficiently explain the well-known link between AS, proatherogenic risk factors, and disease-prone environmental influences. Thus, it has become clear that other regulatory mechanisms are essential, and a compelling argument for an epigenetic contribution is rapidly emerging [11,14].
Epigenetics refers to mitotically and meiotically stable (heritable), and functionally relevant DNA and chromatin modifications that are not caused by alterations (mutations) in the primary DNA sequence itself [24]. They can be either inherited or accumulated throughout a lifetime. Furthermore, given that each cell in our organism contains the identical genomic DNA the epigenetic mechanisms are fundamental to proper lineage commitment, cell fate determination, organogenesis, and ultimately, whole body homeostasis. More specifically, by altering the chromatin architecture and the accessibility of DNA coding and regulatory regions they orchestrate the spatiotemporal gene expression in a cell-type and even allele-specific (maternally or paternally imprinted loci) manner [25].
Four major epigenetic mechanisms ( Fig. 1) have been characterized: DNA methylation and hydroxymethylation, covalent histone modifications and incorporation of histone variants, ATP-dependent chromatin remodeling, and chromatin and gene regulation by non-coding RNAs (ncRNAs) including small micro RNAs (microRNAs/ miRs) and long non-coding RNAs (lncRNAs) [26][27][28][29][30][31]. These mutually interdependent epigenetic alterations, collectively named the epigenome, have profound effects on cellular repertoire of active genes [32,33]. Furthermore, the well-known characteristics of epigenetic regulatory mechanisms: stability, adaptability, and reversibility, are all equally important in maintaining and changing cellular phenotype and function in both health and disease.
First, through epigenetic modifications genes are switched on and off in a more durable fashion than by any other mechanisms of gene regulation. Secondly, epigenetic alterations undergo dynamic changes during development and in response to the various nutritional, behavioral and environmental stimuli [34]. Notably, early changes of epigenetic regulatory mechanisms caused by fetal environment may influence the adult phenotype, including an individual's susceptibility to cardiovascular diseases (CVD) later in life, and the late onset of CVDs may well be linked to agerelated alterations of epigenetic marks [35][36][37][38][39]. Finally, and clinically most relevant, all known epigenetic marks are reversible, thus opening the possibility for prophylactic or therapeutic intervention and reprogramming of cells even in the early stages of disease progression.
Herein, we provide a comprehensive overview of currently known epigenetic mechanisms involved in the control of gene expression in the native and infiltratory aortic valve cells, and discusses their roles in the pathogenesis and the progression of AS. Myocardial epigenome alterations due to pressure overload (PO) LVH induced by AS will also be covered.

DNA methylation
The data examining the role of DNA methylation changes in etiology of AS are only beginning to emerge.
Additional proof for the involvement DNA methylation changes in etiology of AS was reported by Radhakrishna et al. [42]. Their comparative DNA methylation analysis of neonatal dried blood spots obtained from newborns with AS disease and gestational-age matched controls revealed 59 significantly altered (hypomethylated or dimethylated) CpG methylation sites in the coding and/or promoter regions of 52 genes [42]. More specifically, they observed a significant methylation changes in APOA5 (apolipoprotein A5), PCSK9 (proprotein convertase and subtilisin/knexin-type 9), DUSP27 (dual-specificity phosphatase 27), RUNX1 (runt-related transcription factor 1), and TXNRD2 (thioredoxin reductase 2) gene thus concluding that their altered expression is likely responsible for congenital AS [42]. Importantly, many of these differentially methylated CpG sites demonstrated good to excellent diagnostic accuracy for the prediction of AS status, thus raising the possibility to be used as molecular screening markers for non-invasive risk estimation and disease detection [42].
Importantly, Gilsbach et al. reported that altered methylation pattern of CpG sites may also contribute to regulation of LVH as a response to chronic PO induced by AS [43]. They identified 1280 differentially methylated CpG sites in myocardial biopsies from AS patients with cardiac hypertrophy (CH) and 1365 CpG sites in patients with HF, with 523 of them significantly altered in both patient groups [43]. In addition, 496 of these differentially methylated CpG sites were concordantly altered both in hypertrophic and in HF tissue samples [43].
The first piece of evidence for the role of gene specific alteration of DNA methylation marks in AS was reported by Nagy and Back [44]. They showed that treatment of human AVICs with the DNA methyltransferase inhibitor 5-Aza-2′deoxycytidine increases ALOX5/5LO (5-lipoxygenase) transcriptional levels and the production of the proinflammatory mediator LTB4 (leukotriene B4) [44]. These in vitro findings were confirmed in surgically explanted calcified aortic valves exhibiting reduced promoter methylation of ALOX5 accompanied with significantly higher transcriptional levels compared with non-calcified valve tissue [44]. The same group has also previously reported that the local upregulation of 5-lipoxygenase pathway (ALOX5, ALOXAP/FLAP/5-LO activating protein, LTA4H/Leukotriene A4 hydrolase and LTC4S/Leukotriene C4 synthase) in human aortic valves leads to leukotriene-induced effects on aortic VICs (enhanced leukocyte recruitment, inflammation, increased reactive oxygen species /ROS/ production, LTB4-induced matrix metalloproteinase /MMP/ secretion, matrix remodeling, and calcification) and correlates significantly with the expression of osteogenic marker genes (BMP2/6 and runt-related transcription factor/RUNX2) and severity of stenosis [45].
Gene-specific DNA-methylation changes in human AVICs were also reported in promoter region of H19 (imprinted maternally expressed non-protein coding transcript) by Hadji et al. [46]. They showed that promoter region of this lncRNA is heavily methylated in healthy aortic valves resulting with no expression of its transcripts. Contrary, promoter hypomethylation observed in stenotic valves leads to increased H19 expression which, in turn, decreases NOTCH1 transcription (prevents the recruitment of P53 to NOTCH1 promoter), and consequently increases the level of NOTCH1 downstream targets RUNX2, BMP2, and BGLAP/OCN (bone gamma-carboxyglutamate (Gla) protein; osteocalcin) thus promoting osteogenic reprograming of AVICs [46].
Finally, reduced expression of EGFR (epidermal growth factor receptor) associated with hypermethylation and dysregulation of its 5-hydroxymethylation pattern has also been linked with abnormal valve differentiation leading to the calcific AS and LVH in mice [47]. Since EGFR protein normally suppresses BMP pathway, attenuation of EGFR signaling may predispose differentiation of VICs to a calcifying cell phenotype [47].
Data accumulating from other tissues also suggest that DNA methylation may be involved in transcriptional regulation of genes implicated in AS, especially the ones involved in osteoblastic transdifferentiation of VICs.
For example, it is well known that obVICs are characterized by markers of osteoblastic differentiation, such as induction of ALP (alkaline phosphatase) [48]. Interestingly, DNA methylation plays important role in regulation of ALP expression in cells of the osteoblastic lineage, especially in its progressive transcriptional silencing during the osteoblasts to osteocyte transition [49,50]. Notably, the degree of ALP promoter methylation is inversely associated with the transcriptional levels of ALP positive cells, with osteocytes, which do not express ALP showing high CpG island methylation [49].
Recent studies have also identified SOST (sclerostin; regulates osteoblast activity and serves as a Wnt /wingless-type MMTV integration site family member/pathway antagonist and a potent inhibitor of bone formation) gene expression as a novel marker of valvular calcification [51][52][53]. Notably, AS patients have significantly higher serum levels of SOST protein when compared to healthy subjects [51,54]. Furthermore, tissues close to the calcified regions exhibit positive sclerostin staining, which is not observed in non-calcified control valves. In addition, SOST mRNA is significantly upregulated in calcified valves compared to non-calcified leaflets [51]. This increase in the SOST mRNA and protein levels is accompanied by the expression of prototypic markers (RUNX2 and BGLAP/OCN) of osteogenic transdifferentiation [51].
Importantly, SOST protein is produced by osteocytes and not osteoblasts and is regulated by DNA methylation during the final stage of osteoblast-to-osteocyte transition [55]. In addition, the SOST methylation marks are inversely correlated with its expression in osteoblastic cells but in this case with heavily methylated promoter in osteoblasts, which do not express SOST, while this same region is largely hypomethylated in human osteocytes [55][56][57].
Furthermore, the promoter regions of DLX5 and OSX genes that exhibit increase expression in human VICs were also found methylated in non-osteogenic and unmethylated in osteogenic cell lines expressing these genes [58]. In addition, Arnsdorf et al. demonstrated that mechanical stimulation, essential for maintaining bone homeostasis, may reduce DNA methylation at the SPP1/ OPN (osteopontin) promoter accompanied with corresponding increase in gene expression [59]. Interestingly, significant upregulation of SPP1 expression was found in human AS valves in relation to preoperative transvalvular pressure gradient as well as in cultured VICs in relation to the mechanical strain applied [60,61].
However, it should be emphasized that even though obVICs and cells of osteoblastic lineage share similar phenotypic markers, the applicability of the above findings to AS is currently unknown and remains largely to be established. A side by side comparison between VICs and several cell types in different stages of osteoblastic lineage commitment and differentiation performed by Monzack and Masters have shown stark differences in both the level and pattern of gene expression [62]. Whether they also include the noticeable dissimilarity in patterns of epigenetic gene regulation is currently quite unknown.
Major epigenetic markers in human aortic VICs are presented in Fig. 2.
DNA methylation changes may also regulate gene expression pattern in aortic valve endothelial cells. Thus for example, White et al. reported DNMT3B as one of the genes affected by shear stress and decreased NOTCH1 (Notch/drosophila/homolog 1/translocation-associated) signaling in primary human AVECs [63]. It is well known that NOTCH1 regulates calcification related gene networks in human vascular and valvular endothelium [63]. Notably, NOTCH1 signaling is higher on the ventricular side, and it is considered protective for calcification [64,65]. Furthermore, human NOTCH1 mutations can cause both familial and sporadic AS disease [66].
Possible NOTCH1-dependent transcriptional and epigenetic mechanisms underlying this phenomenon are quite recently revealed by Theodoris et al. [67]. They used hiPSC (human-induced pluripotent stem cell)-derived ECs generated from two patients with nonsense mutations in NOTCH1 and related unaffected individual to investigate how NOTCH1 haploinsufficiency may cause aortic valve calcification [67].
Upon shear stress NOTCH1 (+/−) ECs exhibited dysregulated epigenetic state and aberrant upregulation of proosteogenic, pro-oxidative and pro-inflammatory signaling pathways, thus recapitulating observed AVECs phenotype in AS [67]. Contrary, wild-type (WT; NOTCH1+/+) ECs cells under the same conditions suppressed these and activated homeostatic anti-atherogenic gene networks [67]. This was accompanied with corresponding alterations of DNA methylation marks, with so called 'CpG Island shores' (sequences up to 2 kb distant to CpG islands in promoter regions) showing the largest enrichment of methylation changes [67]. Their findings agree with previous reports of CpG shore regions displaying the most methylation differences in the context of specific tissue types regardless of species of origin [68].
Concludingly, all the above data, although some of them mostly inferential (osteoblastic transdifferentiation of VICs) and yet to be experimentally validated, strongly suggests that DNA methylation might be an important regulatory mechanism involved in the initiation and progression of AS.

Histone and chromatin marks
Emerging reports suggest that various HAT/HDAC (histone acetyltransferase/histone deacetylase) enzyme complexes have also an important roles in the patohistogenesis of AS.
For instance, Carter et al. have reported reduced mRNA and protein levels of class III histone deacetylase SIRT1 (silent information regulator-two (1) in AS patients when compared to controls [69]. In addition, the levels of SIRT1 expression were inversely correlated with transcriptional activity of RETN (resistin) gene in infiltrated macrophages, thus possibly contributing to the development of ASassociated inflammation [69]. Interestingly, Mohty et al. reported that higher resistin plasma levels in the elderly AS patients (≥ 70 years) are associated with lower LDL cholesterol and increased valvular calcification and inflammation compared to younger middle-aged patients [70]. This may well be associated with corresponding age-related changes in SIRT1 expression.
Roos et al. reported a significant age-related decrease of SIRT1, −2, −3,-4, −6 and SIRT7 expression in aortic valves of both SOD2 (superoxide dismutase (2) haploinsufficient (SOD2−/+) and normal (SOD2+/+) mice and increased expression of SIRT5 in aged, compared to young SOD2 +/− mice [71]. They also reported markedly reduced expression of SIRT6 in aortic valves from AS patients while knockout of SIRT6 expression in mice resulted in a dramatic progeroid phenotype, with a much greater propensity for development of cardiovascular calcification and AS [72]. The same group showed that SIRT6 inhibition promotes the formation of calcified nodules, induces amplification of pro-osteogenic signaling pathways and favors osteogenic differentiation of cultured mouse aortic VSMCs and VICs obtained from SIRT6-deficient mice, porcine, or human patients [73]. Also, SIRT6-deficient cells exhibited increased p-SMAD1/5/8 levels while the global SIRTs inhibition resulted with increased histone acetylation and protein levels of osteogenic markers SP7 and RUNX2. Furthermore, SIRTs inhibition has also enhanced the transcriptional responses of SP7 and RUNX2 gene in BMP2 treated VSMCs and VICs [73,74].
It is well known that reduced HDAC1 expression and its decreased recruitment to the promoters of osteoblast marker genes represent an important step for osteoblasts differentiation and vascular calcification [79,80]. It has been also reported that HDAC3 physically interacts with RUNX2 thus suppressing its transcriptional activity [81] Furthermore, dysregulation of HDAC3 mediated epigenetic silencing of TGFβ1 (transforming growth factor beta 1) is required during valve development to maintain VICs quiescence and may represent a predisposing factor for development of BAV and other congenital heart diseases [82]. In addition, HDAC5 is involved in RUNX2 degradation while HDAC7 represses its activity during osteoblastogenesis in deacetylase independent manner [82][83][84][85]. Furthermore, since HDAC7 suppresses the expression of MMP10, its downregulation observed in VICs may play a role in biomechanical and HOTAIR lncRNA dependent upregulation of procalcific and inflammatory genes related to BAV induced AS [78,86,87]. Nevertheless, despite all of the above data, the exact role, if any, of these HDAC enzymes in development and progression of AS has yet to be experimentally validated.
The importance of histone code alterations in the development of AS was also reinforced by previously mentioned analysis of hiPSC-derived ECs reported by Theodoris et al. [67]. They showed a clear relationship between the expression of oxidative, inflammatory and calcification related genes in WT (NOTCH1+/+) and haploinsufficient (NOTH1+/−) ECs mediated by fluid-flow conditions and NOTCH1-dependent distribution of activated (H3K4me3, H3K27ac, and H3K4me1) and repressive (H3K27me3) histone marks [67]. More specifically, the shear stress response in WT ECs resulted in activation of anti-osteogenic and anti-oxidation genes and repression of proinflammatory loci that was correlated with the occupancy of their enhancers by repressive H3K27me3 or active H3K27ac histone marks [67]. Quite contrary, upon shear stress Notch1 +/− ECs exhibited aberrant upregulation of pro-osteogenic genes associated with increased H3K4me3, H3K27ac, and H3K4me1 histone marks while H3K27ac marks were increased in active enhancers of pro-inflammatory genes under both static and shear flow conditions [67]. Moreover, anti-calcific and anti-atherogenic genes downregulated by NOTCH1 haploinsufficiency showed decreased H3K27ac active enhancer marks as well as increased repressive H3K27me3 marks [67]. Importantly, alterations of histone code were in most cases accompanied with corresponding changes of DNA methylation marks [67]. More specifically, the regions hypermethylated in Notch1 haploinsufficient ECs lost H3K4me3 or H3K27ac activating marks present in WT ECs while regions hypomethylated in N1+/− ECs gained H3K4me3 or H3K27ac marks [67].
Although it remains to be determined whether the used hiPSC-derived ECs accurately model molecular pathways induced in diseased human AVECs, these data presented by Theodoris et al. represent a significant step forward in deciphering the complexity and mutual interconnection of epigenome alterations underlying the pathogenesis of AS.
Other genes involved in regulation of epigenetic histone marks and/or chromatin remodeling that were found differentially expressed in stenotic aortic valves are presented in Table 1. Given the important role the mechanical forces, proatherogenic risk factors and intrinsic (epi)genetic variants, as well as, age and gender are playing in disease onset and progression, the known transcriptional profile of corresponding histone and chromatin remodeling factors induced by them in experimentally modified aortic valve endothelial and interstitial cells are presented as well. Specified genes are selected based on reanalysis of published data using DAVID (database for annotation visualization and integrated discovery), EpiFactors (database for epigenetic factors, corresponding genes and products) and Genecard (the human gene database) free online databases [88][89][90][91].

Histone/chromatin alteration in AS-induced LVH
Alteration of histone marks and chromatin remodeling might also contribute to the pathological response of LV myocardium to increased afterload in AS, thus leading to CH and HF [92]. Although the initial hypertrophic responses seem to be an adaptation to those stimuli, the sustained stress may lead to reactivation of fetal genes, which is possible due to the interplay of transcription factors, HAT/ADAC and ATP-dependent chromatin remodeling complexes [92].
As already known, BRG1 is highly expressed in the developing heart while BRM/SMARCA2 (Brahma; SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily a, member 2) represents the prominent catalytic unit of BAF chromatin remodeling complex in adult cardiomyocytes [96]. However, upon hypertrophic stimuli adult cardiomyocytes switch to a fetal-like state and BRG1 is reactivated and with its embryonic partners (HDACs and poly (ADP ribose) polymerases /PARPs/) forms BRG1-HDAC-PARP chromatin remodeling complex to induce a pathological gene expression [96].
The importance of BRG1 for aortic valve development was also previously reported by Akerberg et al. [97]. They showed that endocardial BRG1-deficient embryos develop thickened aortic valves that are frequently bicuspid and progress to a myxomatous, disease-like state but does not become overtly calcified [97].
Taken together, these data provide new insight into the complexity of epigenetic gene regulation in diseased aortic valve cells and affected myocardium. Without any doubt, we currently see only the very tip of histone and chromatin alterations lying under the surface of AS patohistogenesis with numerous new players expected to emerge.

LncRNAs
Long noncoding RNAs are also emerging as powerful epigenetic regulators of AS. The recent comparative transcriptome analysis of calcified aortic valves and 32 other human tissues reported by Wang et al. revealed a total of 725 aortic valve-specific lncRNAs [98]. However, the exact role of lncRNA transcripts in patohistogenesis of AS is still poorly known and limited to few examples.
Previously, Carrion et al. have reported decreased transcript levels of HOTAIR (HOX transcript antisense RNA) in human BAVs as compared to the normal TAVs and in VICs exposed to cyclic stretch [78]. They also detected significant upregulation of several genes associated with calcification of VICs subjected to cyclic stretch (ALP and BMP2) and HOTAIR siRNA treatment (ALP, BMP 1/4/6, bone morphogenic protein receptor type 2/BMPR2, endothelin 1/EDN1, periostin/POSTN, SOST, matrix γcarboxyglutamate (Gla) protein/MGP, and MMP2/10/12) [78]. Furthermore, the mechanoresponsive HOTAIR expression and epigenetic regulation (recruitment of PRC2 complexes which mediate trimethylation of lysine 27 on histone H3/H3K27me3 to ALP and BMP2 promoters) of calcification related genes were significantly repressed by WNT β-catenin signaling pathway [78]. Interestingly HOTAIR siRNA treatment of cultured VICs resulted with decreased expression of DNMT3B and altered expression of number of genes involved in histone and chromatin modification ( Table 1) and genes encoding various miR (miR-1228, miR-1978 and miR-586) and lncRNA transcripts ( Table 2). Implication of lncRNAs in AS development is also demonstrated by the role of intergenic lncRNA MALAT1 (metastasis-associated lung adenocarcinoma transcript 1) which is associated with osteogenic transdifferentiation of human VICs. Its upregulated expression in AS compared to control valves was first reported by Bosse et al. [99]. Contrary to them, Zhu et al. detected significantly lower levels of MALAT1, both in calcific valves and osteoblastic differentiating VICs accompanied by lower levels of TWIST1 (twist-related protein 1) protein [100]. Furthermore, they showed that MALAT1 directly interacts with TWIST1 and could enhance its stability while the TWIST1 overexpression prevents VICs calcification induced by inhibition of MALAT1 [100]. Possible mechanism was previously indicated by study of Zhang et al. demonstrating that TWIST1 negatively regulates osteoblastic transdifferentiation of human VICs through direct inhibition of RUNX2 [101].

LncRNAs associated with AS-induced LVH
Recent studies had shown that lncRNAs may also play an important role in regulation of AS-induced cardiac hypertrophy. Thus, Viereck et al. reported a significant upregulation of CHAST (cardiac hypertrophy-associated lncRNA transcript) lncRNA in hypertrophic heart tissue from AS patients and in human embryonic stem cellderived cardiomyocytes upon hypertrophic stimuli [102]. Furthermore, Ounzain et al. reported the upregulation of CARMEN (cardiac mesoderm enhancer-associated noncoding RNA) in patients with idiopathic dilated cardiomyopathy (DCM) and AS [103]. The same group quite recently reported WISPER (WISP2 super-enhancer associated RNA) lncRNA as a novel cardiac fibroblast enriched super-enhancer related lncRNA, whose expression was significantly correlated with cardiac fibrosis both in a murine model of myocardial infarction and in AS patients [104]. Furthermore, previous study by Ounzain et al. reported differential expression of Novlnc6, Novlnc23, and Novlnc44 lncRNA in LV biopsy samples from dilated cardiomyopathy (DCM) and AS patients when compared to control LV tissue [105]. In contrast to DCM group, patients with AS were not associated with downregulation of enhancer derived lncRNA Novlnc6, or its predicted target gene NKX2-5 (NK2 Transcription Factor Related, Locus 5). However, they detected significant downregulation of Novlnc44 while the downregulation of Novlnc23 was not statistically significant [105]. Contrary to the above, Peters et al. indicate no role of MALAT1 in PO-induced CH, myocardial remodeling and HF in mouse model(TAC-transverse aortic constriction, and angiotensin II-induced CH) of AS despite its reported role in regulation (binding and inhibition) of antihypertrophic miR-133 [106].
Other lncRNAs transcripts differentially regulated in AS compared to control aortic valve tissue are presented in Table 2. They are selected based on bioinformatic reanalysis of published data using DAVID database for annotation visualization and integrated discovery; NCBI Gene website; Genecard, the human gene database; lncrnadb, the reference database for functional long noncoding RNAs, and HGNC web base for gene nomenclature [88,89,[107][108][109][110]. However, functional roles for most of these lncRNA transcripts in the vasculature and other human tissues are largely unknown.
For example, Balderman et al. have reported that BMP2 decreases microRNA-30b and microRNA-30c thus promoting calcification of VSMCs [140,141]. In addition, Ding et al. reported that miR-30e represses the osteogenic program (reduction of the osteogenic panel: dermatopontin/ DPT, decorin/DCN, RUNX2, BMP4, SPP1/OPN, IGF2/insulin-like growth factor 2, ALP, and BGLAP/OCN) in bone marrow-derived mesenchymal stem cells and SMCs by targeting IGF2 and drives their differentiation into adipogenic or SMC lineage [125]. They showed that injections of antimiR-30e increases IGF2 expression in the mouse aorta and significantly enhances OPN protein expression and calcium deposition in aortic valves [125]. They also showed that NFYC (nuclear transcription factor gamma subunit c) gene and its hosted miR-30e transcripts are down-regulated with age and atherosclerosis and inversely proportional to the expression of the osteogenic genes RUNX2, OPN, and IGF2 [125]. MiR-30b/c/d transcripts were also predicted to target RUNX2, while miR-29a and miR-29c were known to target collagen production and miR-2861 has been shown to affect RUNX2 activity (inhibitor of RUNX2 protein) in mouse osteoblasts by targeting HDAC5 [116,142]. In addition, Hu et al. and Xia et al. have detected the regulatory feed-back loop between RUNX2, miR-3960 and miR-2861 acting during the osteoblastic differentiation and in osteogenic transdifferentiation of VSMCs, correspondingly [143,144]. Their results suggested that RUNX2 could directly induce the expression of miR-3960/miR-2861 cluster (located at the same loci) by binding to the putative binding site of its promoter [143,144]. Furthermore, Goettsch et al. reported the involvement of miR-125b in osteogenic transdifferentiation of VSMCs in in vitro and in vivo experimental settings by targeting the osteoblast transcription factor SP7/OSX while When et al. showed that miR-125b regulates transdifferentiation and calcification of VSMCs in a high-phosphate environment by targeting ETS1 (ETS proto-oncogene 1), a known transcription factor involved in osteoblastogenesis and extracellular matrix (ECM) mineralization [145,146]. However, the applicability of these findings in the settings of AS largely remains to be established.
Moreover, differential expression of miRNAs was also detected in circulatory osteoprogenitor cells that play a significant role in pathogenesis of AS disease. Thus, a recent study reported by Takahashi et al. detected a higher levels of pro-osteogenic miR-30c in the AS group compared to controls and lower levels of miR-31, miR-106a, miR-148a, miR-204, miR-211, and miR-424, previously reported as negative regulators of pro-osteogenic pathways in mesenchymal stem cells [147][148][149][150]. Also, the degree of aortic valve calcification in their samples was weakly positively correlated with the number of COPCs and miR30c levels. Furthermore, the number of COPCs and the level of BGLAP/OCN protein in these cells was positively correlated with miR-30c and negatively correlated with the levels of remaining miRNAs [147]. Importantly, after surgical procedure both the number of  COPCs and the levels of miR30c were decreased while the levels of the other miRNAs remained the same [147]. Moreover, the observed changes in miRNAs levels were greater after AVR than TAVR procedure. That may be explained by the procedural differences between these two surgical procedures, with TAVR leaving the residual diseased tissue around the prosthetic valve, which may continue to activate osteogenic processes via dysregulation of ossification related miRNAs [147]. Finally, Li et al. recently reported the role of macrophage miRNAs in regulation of AS [134]. They found that M1 macrophage-derived microvesicles deliver miR-214 promote pro-osteogenic differentiation of VICs and subsequent aortic valve calcification through downregulation of TWIST1, a direct target of miR-214 [134]. The upregulation of miR-214 of aortic valve samples was accompanied with both valve calcification and M1 macrophage polarization. In addition, the functional involvement of miR-214/TWIST1 in the osteogenic differentiation of VICs was further confirmed by intravenous treatment of hypercholesterolemic APOE−/− mice with a miR-214 inhibitor, which significantly suppressed valve calcification and resulted in the upregulation of TWIST1 [134].

Circulatory microRNAs
The utility of circulating miRNAs, as potential diagnostic, and prognostic biomarkers of AS, has also attracted considerable attention over the recent years. They have been correlated both with LV structural and functional impairment and with the outcome of AS patients after surgery.
For instance, the role of circulatory miR-21 as an indicator of LV fibrosis in AS patients in response to PO was supported by findings of Villar et al. and Fabiani et al. while Coffey et al. reported the higher levels of circulatory miR-21-5p in AS patients without coronary artery disease (CAD), but no apparent difference was found between groups with CAD [151][152][153]. Moreover, they showed that plasma levels of miR-21 and miR-21-5p correlated directly with the mean transvalvular gradients thus underscoring the value of circulating miR-21 as a biomarker for MF [151][152][153].
Coffey et al. detected higher circulatory levels of miR-451a and miR-22-3p, and lover levels of miR-24-3p and miR-382-5p in AS patients compared to controls that also remained significantly after adjusting for age [153]. However, after qRT-PCR validation only miR-22-3p and miR-382-5p had levels expected from microarray analysis. In addition, miR-22-3p and miR-24-3p were increased, whereas miR-382-3p was reduced in AS participants with CAD [153]. Furthermore, similar to miR-21-5p, the circulatory levels of miR-382-5p in AS patients also showed a significant correlation with maximum transvalvular velocity and mean gradient, but not with LV mass index [153].
Significantly lower circulating levels in AS patients compared to healthy controls were also reported for miR-1, miR-133, and miR-378 by Chen et al. [154]. In addition, their results indicated that the lower levels of miR-378 may serve as predictor for LVH independent of the pressure gradient [154]. Similarily, Garcia et al. reported that the higher preoperative plasma levels of miR-133a can predict the reversibility of LV hypertrophy after AVR, while Røsjø et al. detected positive association between higher circulating levels of miR-210 and increased mortality during follow-up of AS patients independently of established risk indices, including NT-proBNP (N-terminal proBrain Natriuremic Peptide) levels [155,156].
Quite recently Martínez-Micaelo et al. reported that plasma circulating miRNA expression profile in AS patient may also reflect the morphology of the aortic valve (bicuspid/tricuspid) [124]. Their miRNA-wide microarray and qRT-PCR comparison between the groups (healthy TAV subjects without aortic dilation, BAV patients without aortic dilation and BAV patients with aortic dilation) revealed that the expression levels of circulating miR-122, miR-130a and miR-486 are significantly influenced by the morphology of the aortic valve (BAV vs. TAV), while the expression pattern of miR-718 was inversely correlated with the aortic diameter and thus may possibly be used an independent predictor of aortic dilation [124]. In addition, the targeted gene prediction and putative function analysis of selected miRNAs revealed that the bicuspid valve morphology-associated miRNAs (miR-122, miR-130a and miR-486) most probably affects the TGFβ1 signaling pathway (a total of 32 targeted genes) while the dilatation related miR-718 may be associated with focal adhesion and blood vessel remodeling processes [124].
The possibility that circulating miRNAs are differentially expressed in the blood of hypertrophic cardiomyopathy patients and those with LVH induced by AS has been further investigated by Roncarati et al. [158]. Among the miRNAs significantly increased in the plasma of their patients (miR-27a, miR-199a-5p, miR-26a, miR-145, miR-133a, miR-143, miR-199a-3p, miR-126-3p, miR-29a, miR-155, miR-30a, and miR-21) they found that correlation with LVH holds true for only miR-199a-5p, miR-27a, and miR-29a, whereas only miR-29a was significantly associated with both hypertrophy and fibrosis, thus identifying it as a potential biomarker for myocardial remodeling assessment in CH [158]. The similar trend was detected for only five miRNAs (miR-21, miR-26a, miR-27a, miR-30a, and miR-133a) in AS patients, whereas mir-29a levels showed no increased in AS relative to control, thus suggesting specific miR signatures for these two pathological conditions [158].
Interestingly, Varrone et al. reported that plasma levels of mRNA protein targets in AS patients may also be used for indirect measurement of myocardial miRNAs expression. They detected inverse relationship between myocardial expressions of miR-1 and circulating levels of (FABP3, heart-type fatty acid-binding protein-3) in AS patients with LVH [159]. Specifically, myocardial miR-1 expression was decreased while FABP3 levels were increased compared to controls [159]. Furthermore, the decrease of myocardial wall stress following TAVR procedure have led to downregulation of FABP3 levels to values comparable to ones for control subjects [159]. Moreover, the increased level of circulating IGF1 (insulin-like growth factor 1) in AS patients were also significantly blunted by the TAVR procedure, thus suggesting that this reciprocal relationship between miR-1 and FABP3 protein may be tightly controlled by the IGF1/ miR-1/FABP3 signaling axis [159]. It seems that the hypertrophic response in AS patients is followed by the increase in IGF1 plasma levels and downregulation of miR-1 [159]. Eventually, the reduction of cellular miR-1 levels relieves its negative control over the expression of FABP3, thus leading to a prompt protein release into the circulation where it might be used for indirect measurement of myocardial miR-1 activity [159].

Pericardial fluid miRNAs
Several vascular-enriched miRNAs have also been detected in pericardial fluid (PF) of AS patients undergoing surgery. For instance, Miyamoto et al. reported significantly higher levels of miR-423-5p in PF compared to its plasma values [160]. Contrary, the levels of muscleenriched (miR-133a) and vascular-enriched (miR-126 and miR-92a) miRNAs were found unaltered [160].
In another experiment, Kuosmanen et al. detected more than 70 miRNAs in PF collected from AS and other HF patients (coronary artery disease, mitral valve insufficiency, aortic valve insufficiency, and other cardiovascular disease) during open-heart surgery, with let-7b-5p, miR-16-5p, miR-21-5p, miR-125b-5p, and miR-451a being the most abundant [161]. However, despite the differences in disease etiologies (ischemic vs. nonischemic) or the stage of the HF the overall miRNA profiles were quite similar between the groups [161].

Myocardial miRNAs
Expressional profiling of myocardial biopsies from AS patients further confirms the regulatory role of miRNAs in development of LVH and fibrosis induced by AS [163].
The role of myocardial miR-21 in these processes was supported by several investigators. Villar et al. reported that AS patients featured higher myocardial expression levels of both primary (pri-miR-21) and mature miR-21 transcripts. They were restricted to the interstitial cells (cardiac fibroblasts) within the ECM, with no or very weak miR-21 signals detected in the cardiomyocytes of AS patients or control samples, correspondingly [151]. Moreover, both the myocardial and circulating levels of miR-21 were positively correlated with the myocardial expression levels of genes encoding collagen I, collagen III, fibronectin, TGFβ1 and effectors of TGFß signaling (SMAD2 and MAP3K7/TAK1 mitogen-activated protein kinase kinase kinase 7) together with negative correlation with the miR-21 targets (PTEN, phosphatase and tensin homolog; TIMP3, tissue Inhibitor of metalloproteinases; RECK, reversion inducing cysteine rich protein with kazal motifs, and PDCD4 programmed cell death 4/neoplastic transformation inhibitor), thus suggesting a link between the severity of the maladaptive remodeling and miR-21 upregulation in the myocardium [151]. Another study reported by Lorenzen et al. detected the link between increased OPN expression and activation of the transcription factor AP-1, with subsequent miR-21 induction and regulation of downstream antifibrotic targets (PTEN and SMAD7) in angiotensin 2 (ANGII/AGT)-induced cardiac cells and in LV biopsies from AS patients with myocardial fibrosis [164]. In addition, Garcia et al. reported a new TGFβ-dependent regulatory mechanism involved in the miR-21 transcription and posttranscriptional maturation, through the interaction of p-SMAD2/3 effectors with the ribonuclease DICER1 processor machinery [165]. They showed that this TGFβ-dependent facilitator mechanism could contribute to the pathogenesis of POinduced myocardial remodeling both in the experimental mouse model with transverse aortic constriction (TAC), and in patients with AS [165].
The association of miRNAs with PO-induced cardiac hypertrophy and heart failure in mouse model (TAC) of AS has also been reported by Eskildsen et al. [166]. Together with several well-described cardiac disease related miRNAs (miR-21, miR-29, miR-133a, and miR-208b), they identified altered expression of three novel miRNAs; miR-24, miR-301a, and miR-335 in the left ventricle of TAC affected mice compared to controls [166]. Interestingly, the increased expression of miR-24, miR-301a, and miR-335 was not found in an animal model of myocardial infarction thus suggesting that their regulation is specific for AS and is not part of a general cardiac disease response [166]. Furthermore, the importance of miR-133a in regulation of AS induced LVH was also reported by Duisters et al. [167]. They found reduced miR-133 and miR-30c expression levels in several forms of PO-induced LVH including the hearts of AS patients [167]. In addition, the expression of both miRNAs was inversely related to the protein level of CTGF (connective tissue growth factor), a key molecule in the process of fibrosis and a powerful inducer of extracellular matrix (ECM) synthesis [167]. Moreover, several lines of evidence provided by this study confirmed a negative regulatory role of miR-133 and miR-30c on the levels of CTGF gene expression, not only by repressing CTGF translation but also by degrading its mRNA [167]. Interestingly, 34 of the 42 mammalian collagen genes are also predicted targets of miR-133, thus suggesting a major role for this miRNA in preventing cardiomyocyte collagen synthesis, and the quality of their surrounding ECM [167].
Recently, another study reported by Jiang et al. detected the negative correlation between miR-133 and lncRNA-ROR (regulator of reprogramming, also named lncRNA-ST8SIA3) in mouse (TAC) model of pressure overload induced cardiac hypertrophy [168]. Thus, it seems that lncRNA-ROR serves as the miR-133 sponge while on the other hand overexpression of miR-133 successfully attenuates the expression of lncRNA-ROR reversing its prohypertrophic influence leading to markedly decreased expression of fetal NPPA/ANP (natriuretic peptide A) and NPPB/BNP (natriuretic peptide B) genes [168]. In addition, Renaud et al. reported that members of class I and IIb HDACs may also play a role in the regulation of POinduced miR-133a downregulation in mouse (TAC) model of cardiac hypertrophy [169]. More specifically, the treatment with the class I and IIb HDAC inhibitor Vorinostat (also known as SAHA-suberoylanilide hydroxamic acid) significantly attenuated TAC-Induced downregulation of miR-133a and diminished the upregulation of CTGF protein abundance and collagen deposition, thus suggesting that the effect of HDAC inhibition on miR-133a expression is reflected on its downstream fibrotic targets [169].
These two reports clearly indicate the importance of understanding the complex crosstalk between miRs and other epigenetic regulators such as lncRNAs and HDACs, thus providing the ground for innovative therapies to reset the epigenome alterations in AS and other heart diseases.
MiR-133a also emerged as a key element of the postoperative reverse remodeling process of LVH in the report of Villar et al. [170]. They identified the microRNA-133a levels in intraoperative biopsies as a significant positive predictor of left ventricular mass normalization in AS patients while β-myosin heavy chain expression and BMI constituted negative predictors [170].
Furthermore, Beaumont et al. have reported that myocardial down-regulation of miR-122 might also be involved in the development of myocardial fibrosis in AS patients, most probably through the up-regulation of TGFβ1 [171]. They also reported differential expression of 118 miRNAs (99 down-regulated and 18 up-regulated) in AS patients with severe myocardial fibrosis (SF) compared with the non-SF AS group [171]. The role of miRs in regulation of TGFβ pathway in the cardiomyocyte hypertrophy and interstitial fibrosis in the settings of AS was also reported by Tijsen et al. [172]. They showed that the members of the miR-15 family (miR-15a/b, miR-16, miR-195, miR-497, and miR-322), that are abundantly expressed in the heart and up-regulated in the diseased myocardium, directly or indirectly inhibit TGFβ-pathway by targeting TGFBR1 and several other genes (p38, SMAD2, SMAD3, SMAD7, and Endoglin/ENG) within this pathway [172].
Specific contribution of miRNAs to LV cardiomyopathy induced by AS was further confirmed by Ikeda et al. [173,174]. They reported 87 differentially expressed miRNAs (among 428 examined) in LV samples of AS patients compared to samples from diseased hearts with DCM and ischemic cardiomyopathy/ICM and the non-failing control group [173,174]. Among the miRNAs with known cardiac-enriched expression (miR-1, miR-133, and miR-208), miR-1 was downregulated in DCM and AS and tended to be downregulated in ICM patients while the expression of miR-133 and miR-208 were not significantly changed [173,174]. In addition, the miR-214 exhibited the up-regulation in all three disease groups while the miR-19a and miR-19b were the most down-regulated miRs in DCM and AS, but not in ICM patients [173,174]. MiR-24 has also been identified as significantly upregulated in AS patient group [173,174]. Importantly, differential expression of 13 of these miRNAs was specific to AS, while eight miR-NAs exhibited differential expression in cardiomyopathy groups (ICM and DCM) that did not overlap with the expression of AS specific miRs thus suggesting that altered expression of these miRNAs reflects distinct disease mechanisms or disease stage in AS compared with cardiomyopathy samples [173,174].
Another miRNA-wide microarray study reported by Gallego et al. detected differential expression of 70 miR-NAs (64 downregulated and 6 upregulated) between control subjects and two clusters of AS patient identified according to cardiomyocyte apoptotic index [175]. Among them miR-10b, miR-125b-2* and miR-338-3p were inversely correlated with cardiomyocyte apoptotic index [175]. They concluded that myocardial downregulation of miR-10b may be involved in increased cardiomyocyte apoptosis in AS patients, probably through the upregulation of apoptotic peptidase activating factor 1 (APAF1) gene expression, thus contributing to cardiomyocyte damage and to the development of heart failure [175].
Finally, Beaumont et al. reported reduced expression of miR-133 and miR-19b in the myocardial and serum samples of AS patients when compared to controls [176]. Moreover, both myocardial and serum miR-19b levels were found inversely correlated with expression levels of lysyl oxidase (LOX), collagen cross-linking and left ventricular stiffness in AS patients, particularly in patients with heart failure [176]. In addition, the expression levels of miR-29b, miR-1, miR-208a and miR-499-5p was under the limit while no differences were found for miR-21 levels between serum samples from AS patients and control subjects [176].
Taken together the above data provide clear evidence that aberrantly expressed miRNAs are implicated in a diverse spectrum of pathophysiological pathways leading to AS development and progression, and consequently to pressure overload-induced myocardial remodeling and hypertrophy. This is obviously just a beginning of story with the list of novel miRNA players and the previously unknown roles for the already existent ones growing at a steady pace.

Conclusion
Herein we presented a growing body of experimental evidences to support the key role of epigenetic alterations in the etiology and progression of AS. Evidently they participate in crucial disease-prone phenotype changes of aortic valve cells, and regulate key processes underlying ASinduced valvular tissue remodeling and maladaptive myocardial hypertrophy, i.e., fibrosis, calcification, LV remodeling, and inflammation. As can be seen, alteration of DNA methylation marks may contribute to production of proinflammatory mediators (ALOX5 induced LTB4 production), and have role in osteogenic transformation of VICs (H19 induced NOTCH1 downregulation and subsequent upregulation of RUNX2, BMP2, and OCN) while some of them demonstrate promising biomarker potentials for the prediction of AS status (analysis of dried blood spots of neonatal AS patients). Proinflammatory and osteogenic role was also demonstrated for histon code marks (mediated by alterations of SIRTs gene expression), and their involvement in AS-induced pathological remodeling of left ventricular myocardium was also clearly established (e.g., reexpression of the fetal gene programs induced by altered expression of JARID2 and DPF3 genes). Apparently most studies were focused on the role of small non-coding miRNAs demonstrating their essential role in some of the key processes responsible for disease progression such as the phenotypic alterations of valvular endothelial and interstitial cells under pathologically altered blood born mechanical forces, induction of proinflammatory pathways, and osteogenic transdifferentiation of VICs (Table 4). Moreover, they seem to be crucial for the disease-prone role of monocytemacrophage and COPC cells and have established role in regulation of myocardial fibrosis and LVH. Most importantly, specific miRNA signatures (e.g., plasma levels of miR-21, miR-210, and miR-378), in combination with clinical and functional imaging parameters, could represent useful non-invasive biomarkers of disease progression or recovery after aortic valve replacement. Less is known about the ATP-dependent chromatin remodeling processes and long noncoding RNAs and their extent and potential role in subclinical and clinical manifestation of AS have yet to be examined and experimentally validated in both small and large scale human and animal studies. Also, we currently see only the very tip of histone and DNA methylation marks lying under the surface of AS patohistogenesis. Since all epigenetic mechanisam in a given cell and tissue structure are mutually interdependent as clearly represented by Theodoris et al. A better understanding of epigenome landscape in native and infiltrating aortic valve cells and affected myocardium will certainly shed new insights into all aspects of AS pathology and add important incremental diagnostic and prognostic informations useful for risk stratification and patient management. Even more, since all known epigenetic marks are potentially reversible this perspective is especially exciting given the potential for development of successful and non-invasive therapeutic intervention and reprogramming of cells at the epigenetic level even in the early stages of disease progression.