Epigenetic Modifications: Basic Mechanisms and Role in Cardiovascular Disease

DNA methylation

In differentiated mammalian cells, the principal epigenetic tag found in DNA is that of covalent attachment of a methyl group to the C5 position of cytosine residues in CpG dinucleotide sequences (referred to as CpG throughout this review)³. Recent findings suggest that in undifferentiated stem cells, cytosines, other than those in CpG, can be methylated, as well, ⁴ and that methylation of non-CpG cytosines is crucial for gene regulation in embryonic stem cells in particular. CpG methylation is, however, an important mechanism to ensure the repression of transcription of repeat elements and transposons, and also plays a crucial role in imprinting and X-chromosome inactivation ⁵. Transcriptional gene silencing by CpG methylation also restricts the expression of some tissue-specific genes during development and differentiation by repressing them in non-expressing cells.

During development, the pattern of CpG methylation changes in a predictable manner. In early embryogenesis, methylation is erased throughout the genome and then reestablished in all but CpG islands (regions of the genome with a concentration of CpG residues). CpG islands remain hypomethylated until later in development when some of them become methylated ^{6, 7}. Subsequent methylation of cytosines in CpG islands and at other CpG dinucleotides is associated with transcriptional repression ^{6, 8}, especially when these methylated sites involve promoter or other gene regulatory regions ³. DNA methylation may, however, be activating if it prevents binding or limits expression of transcriptional repressors. Recent studies defining the degree of methylation in mammalian promoters indicate that methylation occurs at only a small percentage of CpG dinucleotides and inhibits transcription of only a small subset of genes in differentiated cells. Many of these repressed genes are germline-specific ⁸, including pluripotency genes, suggesting that methylation is a crucial mechanism by which to suppress key genes during differentiation ⁸.

CpG methylation can supress transcription by several mechanisms. First, the presence of the methyl group at a specific CpG may directly block DNA recognition and binding by some transcription factors. For example, several studies have shown that transcriptional activation at GC-boxes is inhibited by methylation, which excludes binding of Sp1 and Sp3 transcription factors, at least in some promoter contexts ^{9, 10}. Methylation has also been shown to block the ability of the nuclear factor, Hif1, from inducing erythropoietin transcription under hypoxic conditions ¹¹. Alternatively, other factors may preferentially bind to methylated DNA, blocking transcription factor access. For example MeCP2 and other family members ¹² bind to methyl CpG and contribute to transcriptional repression by the recruitment of histone-modifying proteins, such as histone deacetylases (HDAC). Subsequently, histone deacetylation promotes chromatin condensation, further repressing transcription ^{13, 14}. This sequence of events illustrates how DNA methylation and certain histone modifications function together to contribute to the transcriptional on or off state of genes subject to epigenetic modification (Figure 1).

A family of DNA methyltransferase enzymes (DNMTs) is involved in de novo DNA methylation and its maintenance. During embryogenesis, de novo methylation is carried out by DNMT3A and DNMT3B ¹⁵. Although some studies suggest an ongoing role for DNMT3A and DNMT3B in maintaining methylation status in some cell types ^{16, 17}, the ubiquitously expressed DNMT1 is predominantly responsible for maintaining cellular levels of CpG methylation. Interestingly, transcription from alternative promoters results in expression of a truncated oocyte-specific DNMT1 isoform, DNMT1o, that is essential for early embryogenesis ¹⁸. DNMT1 functions in a complex to recognize hemi-methylated DNA and to add methyl groups to the non-methylated daughter strand formed during replication ¹⁹. The base pairing of CpG allows for the reciprocal maintenance of methylation during subsequent replication cycles. In this manner, a non-genetic trait (DNA methylation) can be passed from cell to cell and, with it, the contextual effects on gene expression. Thus, methylation can be considered a long-term, relatively stable, epigenetic trait, the effects of which can contribute to maintaining the cellular phenotype.

Targeting DNA methylation: Role of histone modification, DNA-binding proteins and RNA

Owing to its heritability, DNA methylation is a powerful means by which to suppress the expression of unwanted or excess genes. Several basic questions remain unanswered, such as the mechanisms that promote targeting of specific CpG sites for methylation or prevent their modification. Many of the insights into the mechanistic aspects of targeting CpG methylation come from studies of imprinting and X-inactivation, where CpG methylation represses gene expression in chromosomal regions. X-inactivation occurs in somatic cells of females to limit the expression of most X-chromosome genes to those from one chromosome. Given the random nature of X-chromosome inactivation, female carriers may display a wide variation in phenotypic expression of X-linked disorders ²⁰. Similarly, imprinting regulates autosomal gene expression to genes from only one parental allele in males and females. The importance of imprinting and gene dosage regulation in normal development can be appreciated by the consequences of imprinting disturbances that cause a number of human syndromic disorders, such as Prader-Willi, Angleman, Silver-Russell, and Beckwith–Wiedermann (reviewed in ^{21, 22}).

In imprinting, clusters of genes in a chromosomal region are coordinately inhibited by methylation of an imprinting center; these centers are also referred to as differentially methylated regions (DMRs), and DMRs often overlap CpG islands. Recent studies suggest that transcription of DMRs in the oocyte may target them for subsequent CpG modification by maintaining an open chromatin structure accessible to de novo methylation ²³. In both imprinting and X-inactivation, the expression of long non-coding RNAs ²⁴, such as the Xist transcript in X-chromosome inactivation ²⁵, may also play a regulatory role. Chromatin conformation is apparently essential for the imprinting of some maternally determined imprinted genes. In particular, histone H3-lysine 4 (H3K4) demethylase, LSD1, has been found to be essential for these processes, and deficiency of this enzyme results in embryonic stem cell lethality during early differentiation ^{26, 27}. To date, the mechanisms involved in specifying methylation of DMRs during imprinting include recognition of specific DNA sequence motifs in imprinting centers, the expression of factors that are oocyte-specific to mark maternal chromosomes, chromatin remodeling, and/or combinations of these events. Similar mechanisms may also target methylation of specific gene promoters during differentiation.

CpG methylation is erased in a predictable manner during gametogeneis and following zygote fertilization. Recent findings suggest that these processes require the action of cytidine deaminases, such as AID, as well as DNA repair mechanisms ^{28, 29}. Thus, enzymatic deamination of 5-methylcytosine leads to formation of thymine and T:G base-pair mismatches: base excision repair mechanisms subsequently delete thymine and restore C:G base pairing during epigenetic reprogramming. Spontaneous deamination of 5-methylcytosine also requires base excision repair mechanisms to repair base pairing mismatch, a process that is highly inefficient in most differentiated cells. Thus, spontaneous deamination of 5-methylcytosine has resulted in an overall depletion of CpG dinucleotide sequences in mammalian genomes (over evolutionary time)³⁰. In differentiated, adult cells, CpG methylation is considered long-lasting and refractory to elimination, except by alterations in the expression or activity of DNTM1 (Figure 2) (or following spontaneous deamination and mismatch repair). Recently, however, a novel mechanism involving specific DNA demethylation in response to hormone stimulation has been discovered ³¹. In this system, the cytochrome p45027B1 (CYP27B1) gene is repressed by vitamin D-interacting repressor (VDIR)-mediated recruitment of DNA methylases and MBD4, a methyl DNA binding protein. Transcriptional suppression can be relieved by parathyroid hormone (PTH)-induced demethylation of promoter CpGs. Apparently, in the presence of PTH-induced phosphorylation, MBD4 can mediate demethylation of promoter CpGs through a base-excision repair mechanism that removes methylcytosine, apparently without deamination. This hormone-induced mechanism contrasts with excision-repair mechanisms described above in that it is targeted to a specific promoter by hormone action and it does not require deamination ³² (Figure 2). Other recent studies suggest a role for DNMT3 in active demethylation at estrogen receptor-α-responsive promoters that also exhibit hormone-induced alterations in CpG methylation status ^{33, 34}. Although further studies are needed to confirm these mechanisms, the concept of hormone-induced methylation switching adds a new twist to epigenetic regulation. It remains to be seen whether other genes can be similarly regulated by methylation switching utilizing these or other, as yet, unknown mechanisms.

Histone regulation: readily reversible epigenetic changes

DNA methylation tags promote the persistence of certain histone states, such as deacetylation, thus providing a mechanism for perpetuating post-translational histone modifications. Histones can be post-translationally modified to restructure chromatin in many ways, including phosphorylation, ubiquitination, acetylation, and methylation^{35, 36}. In fact, the ‘Histone Code Hypothesis’ suggests that different combinations of histone modifications may regulate chromatin structure and transcriptional status ^{37, 38}. Details, such as the location of nucleosomes relative to the transcriptional start site of a gene, together with specific combinations of sites, types, and extents of histone modifications, add to the complexity of the histone code (reviewed in ^{1, 39}). Nonetheless, efforts have been made to characterize patterns of histone modifications that contribute to cell-type specific regulation of genes in differentiated cells, such as those associated with smooth muscle-specific genes regulated by serum response factor binding to the CArG box ⁴⁰.

Of the many described histone modifications, histone acetylation, at the ε-amino group of lysine residues in H3 and H4 tails, is most consistently associated with promoting transcription. However, this description oversimplifies a complex process, as acetylated, open-chromatin structure may also allow access of transcriptional repressors. For example, some bromodomain-containing factors, such as BRG1 and Brd4, target to acetylated histones where they can mediate the formation of repressor (or activator) complexes ^{41, 42}. Acetylation is targeted to regions of chromatin by the recognition and binding of DNA sequence-specific transcription factors that recruit one of a growing family of histone acetyl transferase (HAT) cofactors such as CREB binding protein (CBP), and p300, MYST, and GNAT (see Table 2 for a list of histone modifying enzymes) ⁴³. Disruption of the normal acetylation activity of CBP/p300 family members is associated with Rubenstein-Taybi syndrome, an autosomal dominant syndrome ^{44, 45}, highlighting the importance of these cofactors in regulating the proper expression of gene combinations important in development and differentiation.

Deacetylation of histones correlates with CpG methylation and the inactive state of chromatin. There are 4 classes of histone deacetylase enzymes (HDACs), with members capable of deacetylation of histones and/or other protein targets ⁴⁶. These regulatory proteins are themselves subject to regulation by acetylation, phosphorylation, and sumoylation ⁴⁷, which can affect their function, subcellular distribution, and protein-protein associations ⁴⁸. Several studies suggest that interactions with sequence-specific DNA binding proteins and co-repressor complexes can target certain HDAC proteins to histones in a gene-specific manner ^{49, 50}.

Histone lysine methylation patterns and their effects on transcription are more complex than acetylation, in that some methylation sites are associated with transcriptionally permissive chromatin (euchromatin) and some are repressive, fostering heterochromatin formation. In addition, ε-amino groups of lysine residues can be mono-, di-, or tri-methylated. Overall, the H3K27me3 and H3K9me states are associated with silencing, whereas the H3K4me3 and H3K36me3 states are transcriptionally permissive modifications (see Table 3 for a list of histone methylation sites). Importantly, methylation marks recruit effector proteins that play essential roles in maintaining the transcriptional state of the chromatin, for example H3K9me recruits HP1, contributing to heterochromatin formation.

Most histone lysine methyltransferases have a SET homology domain, a vast family of proteins that can be grouped into 7 subfamilies based on their structural similarities ³⁹. SET1 family members specifically foster active chromatin by methylating H3K4. Other histone lysine methyltransferase families can methylate several histone targets. In addition, some of these methyl transferases have additional domains that specify binding to methylated DNA or to other proteins, such as CBP ³⁹.

Until recently, histone methylation was considered a long-term epigenetic marker as the only mechanism for its removal was histone turnover; however, recent studies confirm the existence of multiple histone demethylases capable of demethylating histone lysine methyl groups. These enzymes include lysine-specific demethylase 1-(LSD1), which removes mono- or di-methyl groups from H3K4. The tri-methylated modification is targeted for removal by the Jumonji C-(JmjC) domain-containing demethylases ⁵¹ (see Table 2). Similar to histone methylases, LSD1 and JmjC family proteins may demethylate histones in a gene-specific manner, directed, in part, by interactions between demethylases and DNA sequence-specific nuclear factor complexes. In addition, recent studies show that specific histone demethylases may regulate androgen-mediated transcriptional responses and osteoblast differentiation ^52–54.

Non-coding RNA, transcriptional gene silencing, and epigenetic modifications

As discussed above, long non-coding RNAs (lncRNAs) play an essential role in imprinting and X-chromosome inactivation. The gene silencing effects of these lncRNAs and other lncRNAs, such as HOTAIR in the human homeobox loci, are due, in part, to their recruitment of remodeling complexes such as the polycomb complex that foster histone methylation (notably the inhibitory H3K27me3) ^55–57. In addition, lncRNAs can also suppress transcription by other mechanisms, such as the recruitment of RNA-binding proteins that interfere with histone deacetylation or exclude TFIID promoter association ^{58, 59}.

There is also a growing literature on the role of small noncoding RNAs (sncRNAs) and their effects on transcriptional gene silencing. This group of RNAs includes the dicer-dependent microRNAs (miRNAs) and small inhibitory RNAs (siRNAs) formed by RNA-interference pathways, as well as the piRNAs, that are formed in a dicer-independent manner and specifically associate with the PIWI subfamily of argonaute proteins. Although each of these classes of sncRNAs have been shown to mediate epigenetic DNA and histone modifications, the piRNAs appear to have a distinct function of repressing transposon expression in germline cells by fostering de novo DNA methylation ⁶⁰. A role for siRNA in RNA-mediating DNA methylation and transcriptional gene silencing was first discovered in plants ⁶¹ and has been found to exist in many species, including mammals ⁶². Recent findings in mammalian cells suggest that synthetic siRNAs and endogenous miRNAs that target gene promoters may direct transcriptional gene silencing by recruiting specific argonaute proteins and forming epigenetic remodeling complexes that suppress gene expression by fostering histone deacetylation, histone methylation (H3K9 and H3K27), and DNA methylation ^63–65. In fact, although the mechanism has not been completely elucidated, dicer-deficient ES cells exhibit defects in differentiation that correlate with a loss of de novo DNA methylation and loss of miRNAs⁶⁶, suggesting a role for endogenous miRNAs in regulating necessary epigenetic changes during differentiation. Other studies indicate that some siRNAs targeted to TATA-box sequences may block transcriptional inititation without causing epigenetic modifications to histones and DNA ⁶⁷. Overall, these studies illustrate the important role of ncRNAs in modulating gene transcriptional silencing.

DNA methylation and single nucleotide polymorphisms

Theoretically, SNPs that create CpG sites may be targets for epigenetic modifications, just as loss of these sites will prevent DNA methylation. The consequences of a polymorphism resulting in a CpG in the promoter region of the NDUFB6 gene illustrates this intersection between genetic and epigenetic regulation. NDUFB6 is a respiratory chain protein with diminished expression in type 2 diabetes. In muscle biopsies from elderly patients, NDUFB6 expression inversely correlates with the degree of DNA methylation, suggesting that the presence of a CpG site confers more risk for decreased expression (and potentially disease risk) than not having the site ⁷¹. Taken together, these findings support the concept that epigenetic modifications can influence risk in complex diseases.

Epigenetics, nutrition, and environment

Barker and colleagues hypothesized that environmental factors in crucial periods of early life (during fetal development, for instance) can influence risks for cardiovascular and metabolic diseases later in life. This concept is supported by a number of studies that have associated low birth weight in human populations with increased risk of cardiovascular disease (see for instance reviews ^{72, 73}). For example, individuals prenatally exposed to famine during the Dutch Hunger Winter (1944–45) experienced higher prevalence of obesity and coronary heart disease as adults, when compared to adults born before or conceived after that period ⁷⁴. In Barker’s seminal work, it was found that low birth-weight babies who survived infancy had an increased risk of coronary heart disease later in life, and that increasing birth weight was associated with a graded decrease in risk ⁷⁵. In addition, in utero exposure to hypercholesterolemia has been associated with higher incidence and accelerated progression of lesions in humans, rabbits, and mice ^{76, 77}. Exposure to different behavioral patterns during early postnatal life has also been shown to influence epigenetic modifications in experimental animal models ⁷⁸. Thus, it has been suggested that these long-lasting changes arise, at least in part, from epigenetically mediated alterations in gene expression that occur very early in life ⁷⁹. Applying these concepts to human populations, it has recently been proposed that social and environmental stresses during development may influence epigenetic processes that contribute to the adult race-based US health disparities in diseases like hypertension, diabetes, stroke, and coronary heart disease ⁸⁰.

More recent studies in animal models have begun to characterize epigenetic modifications that are influenced by the intrauterine environment. For example, feeding a low-protein diet to pregnant rats causes low birth weight, hypertension, and endothelial dysfunction in the offspring. Studies have shown a role for the renin-angiotensin system in this phenotype as treatment of pregnant mothers with angiotensin converting enzyme inhibitors or angiotensin receptor (AT1R) antagonists ⁸¹ alleviates hypertension in the offspring. Consistent with these earlier findings, offspring of pregnant mothers fed low protein diets were found to have hypomethylated AT1bR gene promoters along with increased adrenal expression of AT1bR ⁸², suggesting a role for specific hypomethylation in regulating elevated blood pressure in this model. Other studies have reported that a low protein diet during pregnancy in rat results in overexpression of the hepatic glucocorticoid (GR) and peroxisomal proliferator-activated (PPARα) receptors in offspring ^{83, 84}. Interestingly, it had been previously reported that DNMT1 contains a GR response element ⁸⁵. Of importance, these studies established an underlying epigenetic mechanism that involves a decrease in the methylation patterns of GR and PPARα promoters, a decrease in DNMT1 expression, and an increase in the levels of transcription-permissive histone modifications. GR and PPARα receptors are key transcription factors whose altered expression may modulate the activity of a large number of metabolic pathways, and have been implicated in the pathology of numerous diseases, including obesity, diabetes, and atherosclerosis ^{86, 87}.

As discussed further in the following section, the methyl-group responsible for DNA and histone methylation originates from S-adenosyl methionine (AdoMet), via Met biosynthesis through folate-dependent or -independent pathways of homocysteine (Hcy) remethylation (Figure 3). Of importance, supplementing a protein-restricted maternal diet in rats with methyl groups by addition of folate or glycine has been shown to decrease hypertension ⁸⁸, improve endothelium-dependent vasodilation, and increase endothelial NO synthase mRNA levels ⁸⁹, and to restore both the expression and promoter methylation status of the hepatic GR and PPARα receptors in offspring ⁸³. These data support the hypothesis that folate and other methyl group donors can influence fetal development and the risk of cardiovascular disease in the next generation. Interestingly, supplying folate to the offspring, rather than the pregnant mothers, increased the methylation status of some, but not all, of the genes modified by maternal protein restriction ⁹⁰, suggesting that some epigenetic modifications may not be reversible by nutritional interventions in the offspring.

Other modifiers of methylation may also influence epigenetic tags. For instance, dietary status of choline (a betaine precursor that is involved in folate-independent pathways of methionine synthesis) was shown to affect DNA methylation ⁹¹. Moreover, in a study using apo-E deficient mice to assess the efficacy of dietary intervention in retarding atherogenesis, it was shown that betaine supplementation attenuated atherosclerotic lesion formation and growth ⁹². Treatments that alter epigenetic modifications may also have utility in human atherosclerosis where DNA methylation changes in target genes, such as the estrogen receptor α and β genes ^{93, 94}, may contribute to the underlying pathogenic mechanisms (discussed further in ⁷⁷).

Regulation of vascular nitric oxide (NO) production by endothelial nitric oxide synthase (eNOS) may influence atherogenesis as well as thrombosis. Interestingly, the restricted expression of eNOS to vascular endothelium is determined, in part, by DNA methylation and histone modifications ⁹⁵, as treatment of non-expressing cell types with a DNMT inhibitor induced demethylation of the eNOS promoter and increased the levels of eNOS mRNA. Furthermore, local enrichment of acetylated H3 and H4 histones across the native eNOS locus was found in eNOS-expressing cells compared to non-endothelial cells, whose treatment with a HDAC inhibitor upregulated eNOS mRNA expression. Recent data suggests that NO itself may be an epigenetic factor, based on its regulatory function upon chromatin and gene expression, via modification (S-nitrosation and tyrosine-nitration) of nuclear factors, HDACs, and histones ⁹⁶.

Homocysteine: a link between DNA methylation and vascular disease

Homocysteine (Hcy) is biochemically linked to the principal epigenetic tag found in DNA. Although increased circulating levels of Hcy are a risk factor for vascular disease, recent clinical trials that used folate and/or other vitamin B therapies to lower Hcy failed to reduce cardiovascular event rates, therefore casting doubt on homocysteine’s direct causative role in vascular disease⁹⁷. Most studies, however, have not explored the consequences of elevated Hcy on methylation processes, even though Hcy plays a crucial role in methyl-donor biosynthesis ⁹⁸. As depicted in Figure 3, the methyl group responsible for the establishment and maintenance of DNA methylation patterns originates from S-adenosyl methionine (AdoMet), an intermediate in Hcy metabolism. In addition to DNA methylation, AdoMet serves as the methyl donor for more than one hundred different cellular methyltransferase reactions, including histone methylation. Following the transfer of the methyl group, AdoMet is converted into S-adenosyl homocysteine (AdoHcy), which inhibits the majority of AdoMet-dependent methyltransferases. AdoHcy is further converted into Hcy and adenosine by AdoHcy hydrolase, which is widely distributed in mammalian tissues. This reaction is reversible and strongly favors AdoHcy synthesis rather than hydrolysis; however, both Hcy and adenosine are rapidly removed under physiological conditions, favoring the hydrolysis reaction. If Hcy accumulates, AdoHcy will accumulate as well, potentially inhibiting transmethylation reactions. Thus, increased Hcy may be regarded as a global DNA hypomethylation effector via AdoHcy accumulation.

There are many in vivo examples that suggest Hcy levels may modulate global DNA methylation. For example, in healthy humans, increased levels of plasma Hcy were associated with both increased AdoHcy concentrations and DNA hypomethylation in lymphocytes ⁹⁹. This inverse relationship between Hcy plasma concentrations and DNA methylation patterns was further confirmed in other reports ^100–102 (with one exception ¹⁰³) and extended to several animal models ¹⁰². Several studies support the concept that DNA hypomethylation may be responsible, in part, for vascular complications associated with increased circulating levels of Hcy ¹⁰⁴. For example, vascular disease patients manifested increased levels of both plasma Hcy and intracellular AdoHcy, together with decreased DNA methylation, supporting a role for hyperhomocysteinemia (HHcy) in modulating epigenetic mechanisms ¹⁰⁵. This association has also been confirmed in several animal studies in which increased circulating levels of Hcy and AdoHcy were associated with endothelial dysfunction and aberrant DNA methylation patterns ^106–108. [One human study, however, documented that coronary heart disease patients had increased global DNA methylation in the presence elevated serum Hcy levels ¹⁰⁹.] In patients with renal functional impairment (and altered homocysteine clearance), which augments the risk for vascular disease, Ingrosso and colleagues reported increased levels of Hcy together with reduced global lymphocyte DNA methylation pattern ¹¹⁰. Of importance, Ingrosso and colleagues reported that global and specific DNA hypomethylation affected the expression of two genes (SYLB, a pseudoautosomal gene, and H19, an imprinted gene), and that both global DNA methylation patterns and allelic gene expression were normalized after lowering plasma Hcy levels with folate administration. Although two later reports failed to confirm these observations regarding global hypomethylation in patients with renal failure ^{111, 112}, Ingrosso and colleagues work represents the first human study that causally linked Hcy with altered gene expression via DNA hypomethylation.

Aberrant global DNA methylation is only an index of the potential for epigenetic dysregulation. In addition to AdoHcy, a growing list of factors has been identified that can modify DNA methylation patterns. These include the rate of cell growth and DNA replication, chromatin accessibility, local availability of AdoMet, nutritional factors including folate supplementation, duration and degree of the hyperhomocysteinemic state, inflammation, dyslipidemias, oxidative stress, and aging ¹¹³. Thus, the relation between increased Hcy and DNA global hypomethylation may be masked in the clinical setting owing to the presence of these confounders, thereby possibly explaining some contradictory and counterintuitive findings reported to date. Another important aspect to consider is that DNA methylation is unequally distributed throughout chromosomes of differentiated cells ⁴. Thus, hyper- and hypomethylated regions can coexist in the genome, and global DNA methylation status need not correspond to the methylation status of specific genomic regions. For example, it has been recently shown that human cardiomyopathies of different etiologies display a unifying pattern of altered DNA methylation of three angiogenesis-related loci in which the differential (increased or decreased) methylation was correlated with the expression of the corresponding gene ¹¹⁴. It is likely that research on single gene methylation and expression may lead to a better understanding of the vascular effects of elevated Hcy ^{113, 115}.

Several investigators have focused on target gene methylation patterns to explain some of the deleterious effects of Hcy. For example, in cultured endothelial cells, it was shown that Hcy, at physiological relevant concentrations, inhibits cell growth with a concomitant increase in intracellular AdoHcy concentration and a significant decrease in the AdoMet/AdoHcy ratio, suggesting that cellular hypomethylation could play a role in the observed phenotype ¹¹⁶. Subsequent studies suggested a role for transcriptional suppression of cyclin A in mediating Hcy-induced endothelial cell growth inhibition ^{117, 118}, and found that Hcy triggers transcriptional inhibition of cyclin A through demethylation of a specific CpG site located in the core promoter, viz., on the cell-cycle dependent element (CDE). In addition, cyclin A promoter demethylation eliminates MeCP2 binding, which, in turn, impairs HDAC association, leading to an accumulation of acetylated H3 and H4. The authors concluded that the loss of DNA methylation in the CDE repressor site and the resulting chromatin remodeling increases chromatin accessibility to repressors, resulting in inhibition of cyclin A gene transcription. Of additional interest was the observation that a physiological concentration of plasma Hcy inhibits DNMT1 activity in this cell system, providing evidence that Hcy can directly modulate specific DNA methylation reactions. In other studies, Hcy was also shown to disrupt the growth of endothelial cells by downregulating fibroblast growth factor-2 (FGF2) via an epigenetic mechanism involving transcriptional repression ¹¹⁹. Apparently, the FGF2 gene promoter encompasses a CpG island and, in contrast with the cyclin A example, the FGF2 gene was heavily methylatedat cytosine residues despite significant AdoHcy accumulation. Normal levels of FGF2 transcription were restored when the cells were simultaneouslyexposed to a DNA demethylating agent and Hcy. Taken together with other examples in the literature ^{108, 120–122}, these findings suggest that Hcy and AdoHcy accumulation can have complex effects on DNA methylation targets and their transcriptional potential.

Increasing evidence indicates that alterations in lipid metabolism may play a role in vascular pathology associated with hyperhomocysteinemia (HHcy) ^123–126, and many studies suggest that epigenetics may play a role in these processes. For example, changes in DNA methylation have been suggested as a potential mechanism for altered apoA-I and apoA-IV gene expression in mice with HHcy ^{127, 128.} In mice, both apoA-I and apoA-IV genes are contained within the apolipoprotein gene cluster on chromosome 9, and, notably, the cluster contains a CpG-rich regioncorresponding to the 3′ flanking region of the apoA-I gene andthe 5′ flanking region of the apoA-IV gene. Similarly, it has been shown that Hcy is significantly and inversely correlated with HDL-bound cholesterol and apoA-I in both human and murine models of HHcy ¹²⁹. In murine primary hepatocyte cultures, cellular hypomethylation, induced by AdoHcy accumulation, was suggested as an explanation for the Hcy-induced inhibition of apoA-I protein synthesis ¹³⁰; however, additional analysis indicated that Hcy regulation of apoA-I synthesis may also involve other, nonepigenetic mechanisms. Additional evidence for regulation of lipid biosynthesis by the epigenetic effects of Hcy was reported by Devlin and colleagues, who showed that a reduced methylation capacity (assessed by AdoMet/AdoHcy ratio) in liver from mice with mild or moderate HHcy was associated with hepatic changes in phospholipid species and impaired long-chain polyunsaturated fatty acid metabolism that could be attributed, in part, to differential CpG methylation of genes involved in these pathways ^{131, 132}. Taken together, these findings indicate that Hcy may influence gene expression by modulating epigenetic pathways. As discussed above, the list of potential modifiers of methylation is long, suggesting that there are still many questions yet unresolved about Hcy’s effects on methylation, gene expression, and cardiovascular disease risk.