Environmental epigenetics in metal exposure

Abstract

Although it is widely accepted that chronic exposure to arsenite, nickel, chromium and cadmium increases cancer incidence in individuals, the molecular mechanisms underlying their ability to transform cells remain largely unknown. Carcinogenic metals are typically weak mutagens, suggesting that genetic-based mechanisms may not be primarily responsible for metal-induced carcinogenesis. Growing evidence shows that environmental metal exposure involves changes in epigenetic marks, which may lead to a possible link between heritable changes in gene expression and disease susceptibility and development. Here, we review recent advances in the understanding of metal exposure affecting epigenetic marks and discuss establishment of heritable gene expression in metal-induced carcinogenesis.

Introduction

Chronic environmental exposure to some metal compounds, including arsenic, nickel, chromium and cadmium, has been known to induce cancers and other diseases in exposed individuals.1–4 This ability has been long confirmed in cell culture and animal models.5,6 However, most of these metals, with the exception of chromium, are mildly mutagenic or non-mutagenic in conventional bacterial and mammalian mutagenesis assays.7,8 Their ubiquitous and indestructible nature makes these metals the most resilient form of environmental pollutants. While it is thought that these metals disturb a vast array of cellular processes, such as redox state and various intracellular stress-signaling pathways, their ability to induce diseases remains poorly understood.9–11 Sources of potential environmental exposure to these metals include occupational exposure and environmental contamination from massive industrial production with poor practices in the disposal (i.e., the largest toxic sites known as Superfund sites in the US).12

Recent evidence from animal studies shows the important role of epigenetics in response to environmental factors including nutritional supplements and xenobiotic chemicals.13–15 Furthermore, emerging epidemiological studies show that the carcinogenic potential of some toxic metals may involve epigenetic changes, including silencing of DNA repair and tumor-suppressor genes.16–18 It is now becoming apparent that a full understanding of the effects of environmental metals on carcinogenesis will require the consideration of epigenetic-based mechanisms in addition to genetic-based mechanisms.19,20

Various epigenetic marks regulate the dynamic nature of chromatin structure and are essential components for regulating heritable gene expression.21,22 In particular, DNA methylation and histone post-translational modifications have received much attention as mediators of epigenetic processes. Other components of chromatin are also potential mediators of epigenetic inheritance, including histone variants, non-coding RNAs, non-histone chromatin binding proteins, and the higher-order structure of chromatin itself.23,24 In this Review, we focus in recent evidence from experimental and epidemiological studies implicating that some toxic metal exposures can alter epigenetic marks and gene expression patterns.

Arsenic

Arsenic (As) is a widely spread environmental contaminant that can be found in the soil and water and as airborne particles. Epidemiological studies have linked arsenic exposure to the development of lung, bladder, kidney and liver cancer.25,26 Exposure to arsenic occurs in the form of either arsenite (As^III) or arsenate (As^V).27 The increased cancer risk to arsenic is attributed to arsenite rather than arsenate, possibly due to the cell's ability to take up arsenite at a faster rate than arsenate.27,28 Although arsenic has been proposed to disturb a variety of cellular processes including cellular redox status and signal transduction, recent evidence suggests that arsenic may promote cellular transformation through chromatin-based mechanisms.29

DNA methylation.

Epidemiological studies show that arsenic exposure leads to gene-specific DNA hypermethylation in human subjects. In West Bengal, analysis of whole blood samples from individuals chronically exposed to arsenite in the drinking water shows that DNA hypermethylation in the promoter regions of the Cyclin-Dependent Kinase Inhibitor 2A (CDKN2A/p16^INK4a) and p53 genes correlates with their gene silencing30 (Table 1). Similarly, a study on a population of New Hampshire individuals with bladder cancer identifies a correlation between arsenic exposure and increased DNA methylation in the promoter regions of Ras Association Domain Family Protein 1A (RASSF1A) and Serine Protease 3 (PRSS3),31 although the expression levels of these genes are not known. A recent analysis of bladder cancer samples of Taiwanese individuals living in areas with known arsenic contamination indicates a correlation between arsenic exposure and Death-Associated Protein Kinase (DAPK) promoter DNA hypermethylation with loss of DAPK expression.32 Also, DNA methylation has been implicated in silencing of the p21/CDKN1A and Metallothionein-1 (MT-1) genes in cells chronically exposed to arsenite, as treatment of these cells with the DNA methyltransferase (DNMT) inhibitor 5-azacytidine (5-aza) partially restores expression of these genes.33 Currently, the mechanisms whereby arsenic treatment induces gene-specific DNA hypermethylation are not well understood.

Effects on gene-specific DNA methylation have also been reported in mice chronically exposed to arsenite. Similar to the human study, silencing of the CDKN2A and RASSF1A genes is correlated with the promoter DNA hypermethylation in lung tissue and tumors from mice chronically exposed to arsenate.34 In contrast to these findings, the promoters of Estrogen Receptor alpha (ER-α) and Cyclin D1 are hypomethylated in liver tissue from arsenite-exposed mice, and these genes are expressed at 3- to 5-fold higher level than control animals.35 Similar DNA hypomethylation of these genes has also been reported in mice exposed to arsenic during gestation.36 Interestingly, the ER-α and Cyclin D1 expression profile in men chronically exposed to arsenic is similar to that reported in mice, although the promoter DNA methylation status is not known in human samples.36

Both short and chronic exposures to arsenite lead to changes in global DNA methylation.37–39 Analysis of liver tissue from mice exposed to arsenite during gestation reveals decreased levels of DNA methylation at GC-rich areas of the genome and global deregulation of gene expression, including downregulation of members of the cytochrome p450 family and upregulation of members of the glutathione S-transferase family.40 However, the mechanisms by which global DNA hypomethylation increases and decreases gene expression are not known. Also, global DNA hypomethylation by chronic exposure is correlated to the cell transformation and the ability of transformed cells to induce tumors in nude mice.38 In contrast, analysis from peripheral blood of individuals chronically exposed to arsenic in the drinking water indicates mild global DNA hypermethylation.41 Therefore, DNA methylation patterns by arsenite exposure are not always consistent, and the discrepancy on global DNA methylation among studies is not well understood. Arsenic detoxification generally involves methylation by the arsenic methyl transferase AS3MT that utilizes S-adenosyl methionine (SAM) as the methyl donor.42 This is of importance given that SAM is the universal methyl donor for all methylation processes in the cell including DNA and histone methylation.43 Therefore, high concentration of arsenite exposure can lead to reduction of global DNA methylation. This notion is generally consistent with studies in cell and animal model systems, where the concentration tends to be high. However, evidence supporting this notion is generally lacking, as the level of SAM is not concomitantly determined with DNA methylation. Moreover, most epidemiological human studies show DNA hypermethylation in a gene-specific manner, where the concentration of arsenite would be low with chronic exposure. Therefore, it is possible that the effect of arsenite on DNA methylation could be dependent on dose or other factors (i.e., AS3MT activity), which may affect the intracellular SAM levels perhaps in a cell/tissue-specific manner. Furthermore, at least in some cases, the observed global DNA hypomethylation is suggested to be a result of reduced expression of the DNMT1 and DNMT3A/B although the mechanisms behind these observations remain unclear.39

Histone modifications.

Recent evidence indicates that arsenite treatment regulates gene expression through changes in histone modifications. In cells carrying the Mouse Mammary Tumor Virus (MMTV), arsenite exposure inhibits dexamethasone-induced dimethylation of arginine 17 of histone H3 (H3R17me2; a marker of transcription activation) at the nucleosome B of the MMTV promoter and suppresses its expression44 (Table 1). This is correlated with reduced SacI accessibility to the nucleosome B of the MMTV promoter as well as with inhibition of the Co-activator-Associated Arginine Methyltransferase-1 (CARM1) recruitment to this nucleosome. In addition, arsenite treatment is shown to increase the level of phosphoacetylation of histone H3 at lysine 9/serine 10 (H3K9AcS10P; a marker of transcription activation) at the promoters of the Heat Shock Protein 70 (HSP70), Mitogen-Activated Kinase Phosphatase 1 (MKP-1), c-Jun and c-Fos genes. The level of H3K9AcS10P is correlated to increased gene expression in the p38^MAPK-dependent pathway.11,45,46

Interestingly, analysis of parental and arsenite-transformed cell lines indicates upregulation of histone H3 trimethylation at lysines 4, 9 and 27 (H3K4me3, H3K9me3 and H3K27me3) and constitutive binding of the transcription factor MTF-1 at the Metallothionein-3 (MT-3) promoter.47 The simultaneous presence of epigenetic marks for transcription activation (H3K4me3) and inactivation (H3K9me3 and H3K27me3) at promoters is thought to “mark” genes that are silent but poised for rapid activation.48 MT-3 expression is silenced in both cell lines but can be induced by treatment with a histone deacetylase (HDAC) inhibitor. Interestingly, arsenite-transformed cells produce up to five-fold more MT-3 transcript as compared to the parental cells by treatment with the HDAC inhibitor, and this is correlated to downregulation of both H3K9me3 and H3K27me3, with concomitant upregulation of H4 acetylation and H3K4me3.47 These findings indicate that arsenic-induced cell transformation may involve establishing a poised promoter chromatin state of the MT-3 gene for rapid activation and expression through heritable alterations of the promoter histone modifications.

Brief exposures to arsenite are shown to increase the global levels of histone H3 phosphorylation at serine 10 (H3S10P), acetylation at lysine 9 (H3K9Ac) and H3K9AcS10P in several systems, whereas chronic exposure increases the global levels of H3K4me3 and H3K9me2.11,46,49–52 The short treatment in some of these studies suggests that the differences observed are likely to be a consequence of activated signal transduction pathways leading to the expression of the immediate stress response genes, including HSP70 and c-jun/c-fos.

Nickel

Nickel (Ni) is a toxic and carcinogenic metal of environmental concern due to its wide use in industry including the production of jewelry, coins, plating, welding, batteries and medical devices. In addition, nickel is currently used as a catalyst in carbon nanoparticle synthesis. Inhalation of nickel particles represents the major route of human exposure.53 Nickel has been known to cause multiple respiratory system conditions from mild irritation and inflammation to pulmonary edema and cancers.54 Due to the poor mutagenic ability of nickel compounds in bacterial and mammalian cell culture systems, it is thought that this metal exerts its toxic effects through non-genotoxic mechanisms.55

DNA methylation.

The absence of chromosomal aberrations in nickel-transformed cells suggests that DNA methylation may be a mechanism of transformation by nickel treatment.5,56,57 Furthermore, chromosome transfer experiments demonstrate that the unlimited proliferative character of a nickel-transformed cell line could be reversed by induction of senescence. The senescence gene(s) is located in the donor X-chromosome, and its expression is regulated by DNA methylation, as treatment of cells with 5-aza restores the senescence capacity of donor cells. These data indicate that transformation by nickel may involve DNA methylation and silencing of a senescence gene(s) in the X chromosome.58

Recent evidence shows that nickel transformation results in silencing of the DNA repair gene O6-methylguanine DNA methyltransferase (MGMT) expression in lung cancer cells.17 MGMT silencing correlates to DNA hypermethylation of its promoter (along with histone H3 deacetylation and enrichment of H3K9me2). Subsequent binding of the methyl-CpG-binding protein-2 (MeCP2) protein at the promoter indicates chromatin condensation of the MGMT promoter. Importantly, the epigenetic character of these alterations is strengthened by the fact that MGMT expression and partial reversal of repressive chromatin modifications could be achieved by treatment of cells with 5-aza and siRNA-mediated knockdown of DNMT1.17 In addition, DNA hypermethylation is shown to occur at the bacterial xanthine guanine phosphoribosyltransferase (gpt) transgene promoter.59 The gpt transgene is integrated near a heterochromatic area in cells but is normally expressed with gene activation-associated acetylated histones H3 and H4 in its promoter. However, transformation of these cells with nickel results in gpt silencing associated with increased DNA methylation and chromatin condensation.59 In addition, nickel treatment leads to subsequent binding of MeCP2 at the gpt promoter.60,61 Importantly, these phenomena could be partially reverted by pretreatment of cells with 5-aza, further implicating that nickel treatment may exert gene silencing through altering DNA methylation.59,60

A small number of studies have focused on the direct effects of nickel on mononucleosome and nucleosome array structures. X-ray crystal structures of reconstituted mononucleosomes in the presence of millimolar concentrations of nickel reveal up to 43 putative metal-binding sites in the nucleosome with preference towards GG motifs.62 Nickel binding induces alterations in the DNA-histone interaction through stretching of the DNA near the dyad axis, resulting in protrusion of two DNA bases off the nucleosome. These alterations are thought to condense higher-order chromatin structures. Indeed, previous studies using micrococcal nuclease-digested chromatin incubated with various divalent cations indicate that nickel promotes nucleosome array condensation at lower concentrations than magnesium.63 Additional evidence from atomic force microscopy and circular dichroism measurements suggests that nickel is effective at promoting condensation of reconstituted nucleosome arrays.64

Histone modifications.

Evidence indicates that the Sprouty Homolog 2 (Spry2), gpt and MGMT genes are linked to nickel-induced gene silencing with changes in histone modifications.17,59,65 Nickel treatment inhibits expression of the Spry2 gene, a negative regulator of receptor tyrosine kinases.66 The Spry2 silencing is correlated with increased level of H3K9me2, a marker of transcription inactivation, at its promoter.65 Interestingly, increased level of H3K9me2 at the Spry2 promoter is thought to arise because of nickel inhibition of the H3K9-specific, iron-dependent histone demethylase Jumonji domain-containing protein 1A (JMJD1A). In vitro, nickel directly inactivates JMJD1A at an equal molar ratio, suggesting a mechanism involving iron replacement in the catalytic center.67

Similar to results from the DNA methylation study of the gpt transgene, transformation of these cells with nickel results in gpt silencing associated with histone deacetylation, increased level of H3K9me2 and chromatin condensation.59 These events can be partially reversed by co-treatment of cells with the HDAC inhibitor Trichostatin A (TSA), implicating that nickel-induced gpt silencing involves modulation of the local gpt chromatin structure. This notion is reinforced by the fact that gpt is insensitive to nickel-induced silencing if integrated away from heterochromatic areas.68 However, alterations of epigenetic marks at the gpt promoter chromatin triggered by nickel exposure may only be temporal, as removal of nickel from media eventually leads to expression of gpt.69 Also, similar to results from the DNA methylation study, the MGMT gene silencing correlates to histone H3 deacetylation and increased H3K9me2, which is partially reversed by treatment of cells with TSA.17

Nickel exposure has also been shown to affect the global levels of histone modifications, implicating global deregulation of gene expression. Treatment of nickel-transformed cells with TSA has been shown to reverse the phenotype to that of untransformed cells along with the gene expression profile, suggesting that histone acetylation might be inhibited during nickel-induced transformation.70 However, it is not clear whether this reversible transforming phenotype is associated with the altered gene expression profile. Further support for nickel inhibition of histone acetylation comes from experiments where continuous nickel exposure leads to histone H4 hypoacetylation in yeast and human cells.71 Nickel binding to the H4 N-terminal peptide at His 18 is thought to interfere with the interaction of histone acetyltransferase complexes (HAT) with chromatin.72

Other histone modifications that are also globally affected by nickel treatment include H3K9me2, H3K4me3 and ubiquitination of histones H2A and H2B (H2A/H2Bub).49,60,61,73,74 In some cases, the increase in H2A/H2B ubiquitination has been correlated to upregulation of the Ubiquitin-Conjugating Enzyme H6 (UbcH6) E2 ligase at the protein level or inhibition of an unidentified histone deubiquitinating activity.73,74 However, the mechanisms whereby nickel affects the levels of these global histone modifications and their significance on epigenetic heritance remain unexplored.

Extended exposure to nickel is shown to result in cleavage of C- and N-terminal peptides of H2B and a C-terminal octapeptide of H2A, possibly due to nickel-induced histone oxidation.75–77 Interestingly, nickel has been shown to form a stable complex with the C-terminal ELAKHA hexapeptide of H2B.78 Currently, the mechanism of peptide cleavage and its biological significance remain elusive.

Chromium

Chromium (Cr) exposure to humans is widespread, given that multiple industries including stainless steel welding, chrome plating and ferrochrome manufacturing utilize this metal. Chromium is also found in the environment in the form of airborne particles arising from automobile catalytic converters.12 Chromium exposure has been linked to respiratory conditions including nasal ulceration and lung cancers.2,79 The major source of exposure for humans is chromium-contaminated water, typically located in the vicinity of industrial areas. It is speculated that tens of millions of people are exposed to chromium contamination worldwide. Although chromium can be found in various oxidative states including +6, +3 and 0, Cr(VI) is the prevalent form of chromium that has been epidemiologically linked to development of respiratory cancers.12 The major mechanisms of chromium-induced cytotoxicity are thought to be genotoxic, as it is well documented that chromium can lead to oxidative stress, DNA strand breaks, DNA-protein crosslinks and formation of stable chromium-DNA adducts.80,81

DNA methylation.

In lung cancer specimens of individuals chronically exposed to chromium, silencing of the mismatch repair gene Human MutL Homolog 1 (hMLH1) and the tumor suppressor p16^INK4a is correlated with DNA hypermethylation of their promoters.18,82 Furthermore, DNA hypermethylation in the promoter regions of the Adenomatous Polyposis Coli (APC) tumor suppressor and MGMT is detected in these samples, although a correlation with their gene expression levels is not established.16 Chromium has also been found to induce DNA hypermethylation of the gpt promoter and subsequent silencing of its expression in cells.83

Histone modifications.

Similar to results of the DNA methylation study from human samples, chromium exposure results in increased level of H3K9me2 at the promoter of hMLH1 gene and reduces its expression in lung cancer cells.84 Interestingly, the increased level of H3K9me2 is correlated to upregulation of the H3K9 methylase G9a at both mRNA and protein levels in these cells.84 In addition, short exposure to chromium leads to a reduction in the levels of H3K4me3 and H3/H4 Ac and blocks aryl hydrocarbon receptor (AhR) and RNAP II recruitment at the Cytochrome p450 Subfamily 1 Polypeptide 1 (CYP1a1) promoter, with the corresponding inhibition of CYP1a1 expression in response to benzo(α)pyrene.85 These events are shown to correlate to the chromium-dependent crosslinking of DNMT1 and HDAC1 at the CYP1a1 promoter, suggesting that reduction of H3K4me3 and H3/H4 Ac levels at the CYP1a1 promoter is likely a result from the physical obstruction of repressor complexes. This in turn may prevent access of histone-modifying complexes and the transcriptional machinery, thus resulting in reduced CYP1a1 expression. In addition, short chromium exposure increases the global levels of H3K4me3 and H3K9me2.49 Both modifications appear to be differentially localized in the chromatin regions and have been speculated to alter gene expression patterns.

Cadmium

Cadmium (Cd) is a toxic transition metal that is both an environmental and occupational toxicant. Sources of cadmium exposure include battery manufacturing, some electroplating processes and consumption of tobacco products. Cadmium has been designated a human carcinogen by both the International Agency for Cancer Research and the National Toxicology program.86 Multiple epidemiological studies have linked cadmium exposure with pulmonary cancers, while fewer studies have linked it to bladder and pancreatic cancers.87 In addition, cadmium is a multiple tissue carcinogen in animal models.88 Its long half-life and poor excretion make cadmium a cumulative toxin. As such, previous exposures to cadmium may lead to detrimental effects during the lifetime of exposed individuals. Although the mechanisms of cadmium-induced transformation are not understood, its effects are thought to arise through the disruption of zinc-dependent cellular processes.89 This is due to the structural and physical similarities between zinc and cadmium. In addition, although cadmium can produce oxidative stress, it is poorly mutagenic and has a weak DNA binding affinity. This has forwarded the hypothesis that cadmium may promote carcinogenesis through epigenetic mechanisms.86

DNA methylation.

Gene-specific DNA hypermethylation and gene silencing have been observed in cells chronically exposed to cadmium, involving the p16^INK4a, RASSF1A and MT-1 genes.90,91 Importantly, expression of these genes can be restored by treatment of cells with 5-aza, further strengthening the correlation between DNA methylation and gene silencing. Increased expression of DNMT3B is shown to correlate with DNA hypermethylation of the p16^INK4a and RASFF1A promoters.90 However, details on the mechanisms of MT-1 silencing are lacking, as this event is only correlated to increased total cellular DNMT activity.91 In addition, recent evidence indicates that global DNA methylation is affected by cadmium exposure. Short exposures to cadmium lead to reduction of global de novo DNA methylation in chick embryos.92 The reduced DNA methylation levels are thought to be a result of the inhibitory effects of cadmium on expression of DNMT3A/B, the enzymes responsible for de novo DNA methylation during chick development. Another report suggests that DNA hypomethylation resulting from short exposures to cadmium may potentially be correlated with increased cell cycle rates in cells, although the evidence of a direct connection between these two events is lacking.93 Short exposures to cadmium is to a dose-dependent reduction on DNMT activity, resulting in DNA hypomethylation. In contrast, chronic exposures result in a dose-dependent increase in cellular DNMT activity and concomitant DNA hypermethylation, even after cadmium is removed from the medium.91 However, the discrepancy on global DNA methylation patterns between short and long exposures to cadmium in this study is not well understood. In addition, global DNA hypermethylation has been correlated with upregulation of DNMT1 and DNMT3A/B expression in cells transformed by and chronically exposed to cadmium.90,94 The overall pattern suggests that short exposures are associated with DNA hypomethylation, whereas chronic exposure is associated with DNA hypermethylation.

Histone modifications.

Evidence suggests that cadmium exposure may lead to heritable changes in chromatin structure for rapid transcription activation. Transformation of cells by cadmium results in the establishment and maintenance of a bivalent chromatin domain at the MT-3 promoter, involving changes in histone modifications nearly identical to those observed for arsenic-transformed cells.47 This inheritable alteration of chromatin structure leads to inducibility of MT-3 expression when cells are treated with an HDAC inhibitor. Therefore, these data indicate that transformation by cadmium may establish and stably maintain chromatin structure of the MT-3 promoter for rapid gene activation.47

Mercury

Although strictly not considered a metal, mercury is an environmental toxicant that is correlated with brain toxicity. Indeed, it has been long known that exposure to mercury results in defects in brain function.95 Due to its ability to cross the placenta, mercury is thought to be most toxic during the development of the fetus.96 In cell culture models, methylmercury (MeHg) affects expression of several neuronal growth factors, therefore affecting neural stem cell differentiation and neurite outgrowth.97–99 Various types of seafood contain high levels of mercury and become a source of exposure.

Histone modifications.

Prenatal methylmercury exposure increases the level of H3K27me3 and decreases the level of H3Ac at the Brain-Derived Neurotrophic Factor (BDNF) promoter, which results in reduced expression of BDNF in the dentate gyrus.100 This is correlated to a depression-like behavioral phenotype. Interestingly, treatment with the anti-depressant fluorexine restores BDNF expression in the dentate gyrus, and this correlates to upregulation of H3Ac, without affecting H3K27me3, at the BDNF promoter.100 Restoration of BDNF expression also correlates to reversal of the depression-like behavior. These results imply that a methylmercury-induced epigenetic silencing of the BDNF in the hippocampus may contribute to a behavioral phenotype.

Conclusion and Future Perspectives

Accumulating evidence clearly shows that toxic metal exposures lead to the induction and alteration of epigenetic marks in experimental and epidemiological studies. These studies highlight the possibility that environmental metals may exert their effects on gene expression through establishing and maintaining epigenetic chromatin states. Currently, it is not clear whether these altered epigenetic marks are linked to heritable gene expression patterns or the causative pathway to the disease (cancer) susceptibility and development.

To establish that epigenetic mechanisms underlie the molecular mechanisms of actions for metal exposures, there are a number of factors to be investigated in the future. Changes in epigenetic marks at the global level require further evaluation for their involvement in epigenetic processes. It is important to eliminate the possibility that alterations in global DNA methylation, or any other global epigenetic marks, are not a reflection of dynamic global cellular responses. This is especially critical for metal exposure, where metals can affect multiple cellular processes, including global chromatin structure. For instance, nickel is proposed to bind to and condense chromatin, and chromium can crosslink chromatin-associated proteins to DNA.62–64,85,101,102 Also, cadmium is shown to potentially promote global DNA hypermethylation by either stimulating DNMT activity or inducing its gene expression, whereas arsenite can deplete SAM, leading to DNA hypomethylation.37–39,42,90 Furthermore, studies of arsenite show that global DNA hypomethylation does not always correlate with local gene-specific DNA hypermethylation or gene expression patterns, suggesting that a gene-by-gene-based analysis is required for establishing epigenetic-based mechanisms.29,30 Given that multiple cellular processes are affected by exposure to these metals, identification of the initial epigenetic signaling pathways or factors that establish and maintain metal-induced heritable chromatin states could provide further evidence for epigenetic mechanisms.

Most of the global and local epigenetic marks mediated by metal exposure in experimental studies are generally measurable only in the continuous presence of metals. Removal of nickel treatment leads to re-expression of a nickel-silenced gene, therefore suggesting that altered epigenetic marks could be a result of dynamic cellular responses.69 Also, alterations in epigenetic marks induced by brief or short-term exposure most likely represent ongoing signal transduction and may not represent heritable epigenetic events, especially without evidence of heritable epigenetic marks correlating with gene expression patterns. In the case of cadmium, short- and long-term exposures result in the opposite effect on DNA methylation through different mechanisms, suggesting that duration, in addition to dose, of exposure affects the outcome of epigenetic marks.69,92–94 The only exception to this is the study of the MT-3 gene.47 This study suggests that regulation of epigenetic events by transient exposures to arsenite and cadmium may not necessarily involve heritable gene expression patterns but the establishment and maintenance of heritable epigenetic chromatin states, which poise the gene for rapid activation in response to a stimulus.

Currently, it is not clear whether specific gene silencing and associated with epigenetic marks induced by metals are transitory or heritable in both experimental and epidemiological studies. Although studies are limited, it would be of great importance to determine whether silencing of tumor suppressor and DNA repair genes could be identified as a common set of epigenetically labile genes that could play roles in metal-induced carcinogenesis. Furthermore, this identification would provide epigenetic biomarkers for detecting early-stages of cancer.

Collectively, an increasing body of evidence clearly supports the concept that epigenetics has potential for better understanding of the molecular mechanisms whereby environmental metal exposure leads to heritable epigenetic marks, correlating with gene expression patterns. Future studies of epigenetic marks need to address the heritability of these marks and correlate them to heritable gene expression patterns, possibly linking them to metal-induced carcinogenesis.

Figures and Tables

Table 1 Metal-induced epigenetic marks

Metal	Chromatin modification	Gene	Correlation with gene expression	Reference
Arsenic	DNA methylation (+)	p16^INK4a/CDKN2A, p53, RASSF1A, DAPK, p21/CDKN1A, PRSS3, MT-1	silencing	30–34, 37, 41
	DNA methylation (−)	global, ER-α, Cyclin D1	activation	35–40
	H3K9Ac	global	n.d.	52
	H3K9AcS10P (+)	global, HSP70, MKP-1, c-jun, c-fos	activation	11, 45, 46
	H3S10P (+)	global	n.d.	50, 51
	H3R17me2a (−)	MMTV	silencing	44
	H3K4me3 (+)	MT-3	poised(*)	47, 52
	H3K9me3 (+)	MT-3	poised(*)	47, 52
	H3K27me3 (+)	MT-3	poised(*)	47
Nickel	DNA methylation (+)	gpt, MGMT	silencing	17, 59–61
	H3/H4Ac (−)	global, gpt	silencing	59, 71
	H3K9me2 (+)	global, gtp, Spry2, MGMT	silencing	17, 49, 59, 60, 61, 65
	H3K4me3 (+)	global	n.d.	49, 61
	H2A/H2BUb (+)	global	n.d.	72, 73
Chromium	DNA methylation (+)	gpt, hMLH1, p16^INK4a, APC, MGMT	silencing	18, 81
	H3K9me2 (+)	global, hMLH1	silencing	49, 83
	H3K4me3 (+)	global	n.d.	49
	H3K4me3 (−)	CYP1a1	silencing	84
	H3/H4Ac (−)	CYP1a1	silencing	84
Cadmium	DNA methylation (+)	global, p16^INK4a, RASSF1A, MT-1	silencing	89, 90, 93
	DNA methylation (−)	global	n.d.	91, 92
	H3K4me3 (+)	MT-3	poised(*)	47
	H3K9me3 (+)	MT-3	poised(*)	47
	H3K27me3 (+)	MT-3	poised(*)	47
Mercury	H3K27me3 (+)	BDNF	silencing	100
	H3Ac (−)	BDNF	silencing	100

Abbreviations: (+), increase; (−), decrease; n.d., not determined; (*), evidence suggests that gene is poised for activation; ER-α, Estrogen receptor α; CDKN2A/p16^INK4a, Cyclin-dependent kinase inhibitor 2A; RASSF1A, Ras association domain family protein 1; DAPK, death-associated protein kinase; CDNK1A /p21, cyclin-dependent kinase inhibitor 1A; MT-1, Metallothionein-1; PRSS3, serine protease 3; HSP70, heat shock protein 70; MKP-1, Mitogen-activated protein kinase phosphatase 1; MMTV, mouse mammary tumor virus; MT-3, Metallothionein-3; gpt, bacterial xanthine guanine phosphoribosyl transferase; MGMT, O6-Methylguanine DNA methyltransferase; Spry2, sprouty homolog 2; hMLH1, human mutL homolog 1; APC, adenomatous polyposis coli; CYP1a1, Cytochrome p450 subfamily 1 Polypeptide 1; BDNF, brain-derived neurotrophic factor.

Acknowledgements

This work was supported by funding from the National Institutes of Health (grants ES015210 and AI64706) to H.C.H.