An Emerging Role for Epigenetic Dysregulation in Arsenic Toxicity and Carcinogenesis

Background Exposure to arsenic, an established human carcinogen, through consumption of highly contaminated drinking water is a worldwide public health concern. Several mechanisms by which arsenical compounds induce tumorigenesis have been proposed, including oxidative stress, genotoxic damage, and chromosomal abnormalities. Recent studies have suggested that epigenetic mechanisms may also mediate toxicity and carcinogenicity resulting from arsenic exposure. Objective We examined the evidence supporting the roles of the three major epigenetic mechanisms—DNA methylation, histone modification, and microRNA (miRNA) expression—in arsenic toxicity and, in particular, carcinogenicity. We also investigated future research directions necessary to clarify epigenetic and other mechanisms in humans. Data sources and synthesis We conducted a PubMed search of arsenic exposure and epigenetic modification through April 2010 and summarized the in vitro and in vivo research findings, from both our group and others, on arsenic-associated epigenetic alteration and its potential role in toxicity and carcinogenicity. Conclusions Arsenic exposure has been shown to alter methylation levels of both global DNA and gene promoters; histone acetylation, methylation, and phosphorylation; and miRNA expression, in studies analyzing mainly a limited number of epigenetic end points. Systematic epigenomic studies in human populations exposed to arsenic or in patients with arsenic-associated cancer have not yet been performed. Such studies would help to elucidate the relationship between arsenic exposure, epigenetic dysregulation, and carcinogenesis and are becoming feasible because of recent technological advancements.


Review
The International Agency for Research on Cancer (IARC) classified arsenic, a toxic metal loid, as a group 1 carcinogen > 20 years ago (IARC 1987). It is widely accepted that exposure to arsenic is associated with lung, bladder, kidney, liver, and non melanoma skin cancers (IARC 2004;Pershagen 1981;Smith et al. 1992;Smith and Steinmaus 2009). High levels of arsenic have also been associated with the development of several other diseases and deleterious health effects in humans, such as skin lesions (dyspigmentation, keratosis), peripheral vascular diseases, reproductive toxicity, and neurological effects (Abernathy et al. 1999).
Exposure to arsenic typically results from either oral arsenic consumption through contaminated drinking water, soil, and food, or arsenic inhalation in an industrial work setting. Arsenic-contaminated drinking water has been associated with increased mortality of bladder and lung cancer in Chile (Marshall et al. 2007) and with increased mortality of both non cancerous causes and cancers in Bangladesh (Sohel et al. 2009). In the human arsenic metabolic pathway, inorganic pentavalent arsenic (As V ) is converted to tri valent arsenic (As III ), with subsequent methylation to mono methylated and dimethylated arsenicals (MMA, DMA, respectively) (Drobna et al. 2009). The general scheme is as follows: Methylated arsenicals, especially MMA III , are considered more toxic than inorganic As III both in vivo (in animals) (Petrick et al. 2001) and in vitro (human cell lines) (Styblo et al. 2002). Several mechanisms by which arseni cal compounds induce tumori genesis have been proposed, including oxidative stress (Kitchin and Wallace 2008), geno toxic damage and chromosomal abnormalities (Moore et al. 1997a;Zhang et al. 2007a), and cocarcinogenesis with other environmental toxicants (Rossman et al. 2004); epi genetic mechanisms, in particular, have been reported to alter DNA methylation (Zhao et al. 1997).
It is generally believed that arsenic does not induce point mutations, based on negative findings in both bacterial and mammalian mutagenicity assays (Jacobson-Kram and Montalbano 1985;Jongen et al. 1985). Arsenic does induce deletion mutations, but arsenical compounds vary in their potency (Moore et al. 1997b). With respect to arsenic's ability to induce chromosomal alterations in humans, studies in the early 1990s showed that the cell micro nucleus assay could be used as a biological marker of the genotoxic effects of arsenic exposure (Smith et al. 1993). Later studies validated this assay and demon strated higher frequencies of micro nuclei in indi viduals who were chronically exposed to arsenicals (Moore et al. 1997a). Analysis of chromosomal alterations in DNA from bladder tumors of 123 patients who had been exposed to arsenic in drinking water showed that tumors from patients with higher estimated levels of arsenic exposure had higher levels of chromosomal instability than did tumors from patients with lower estimated levels of exposure, suggesting that bladder tumors from arsenic-exposed patients may behave more aggressively than do tumors from unexposed patients (Moore et al. 2002). Based on these overall findings, a plausible and generally accepted mechanism for arsenic carcino genicity is the induction of structural and numeri cal chromosomal abnormalities through indirect effects on DNA. However, as has been demon strated for several tumors, including uro thelial and hematological malignancies (Fournier et al. 2007;Muto et al. 2000), it is likely that inter related genetic and epi genetic mechanisms together contribute to the toxicity and carcinogenicity of arsenic (Hei and Filipic 2004;Zhao et al. 1997).

Epigenetic Modifications Induced by Arsenic
Epigenetic alteration, which is not a genotoxic effect, leads to heritable phenomena that regulate gene expression without involving changes in the DNA sequence (Feinberg and Tycko 2004) and thus could be considered a form of potentially reversible DNA modification. Recent mecha nistic studies of arsenic carcinogenesis have directly or indirectly shown the potential involvement of altered epi genetic regulation in gene expression changes induced by arsenic exposure. We recently showed that urinary defensin, beta 1 (DEFB1) protein levels were significantly decreased among men highly exposed to arsenic in studies conducted Background: Exposure to arsenic, an established human carcinogen, through consumption of highly contaminated drinking water is a worldwide public health concern. Several mechanisms by which arsenical compounds induce tumorigenesis have been proposed, including oxidative stress, genotoxic damage, and chromosomal abnormalities. Recent studies have suggested that epi genetic mechanisms may also mediate toxicity and carcinogenicity resulting from arsenic exposure. oBjective: We examined the evidence supporting the roles of the three major epi genetic mechanisms-DNA methylation, histone modification, and microRNA (miRNA) expression-in arsenic toxicity and, in particular, carcinogenicity. We also investigated future research directions necessary to clarify epi genetic and other mechanisms in humans. data sources and synthesis: We conducted a PubMed search of arsenic exposure and epi genetic modification through April 2010 and summarized the in vitro and in vivo research findings, from both our group and others, on arsenic-associated epi genetic alteration and its potential role in toxicity and carcinogenicity. conclusions: Arsenic exposure has been shown to alter methylation levels of both global DNA and gene promoters; histone acetylation, methylation, and phosphorylation; and miRNA expression, in studies analyzing mainly a limited number of epi genetic end points. Systematic epi genomic studies in human populations exposed to arsenic or in patients with arsenic-associated cancer have not yet been performed. Such studies would help to elucidate the relationship between arsenic exposure, epigenetic dysregulation, and carcinogenesis and are becoming feasible because of recent technological advancements.  (Hegedus et al. 2008). DNA methylation is thought to play a role in regulating DEFB1 expression (Sun et al. 2006). Follow-up studies are under way in our labora tory to determine if reduced levels of DEFB1 in exposed populations are due to arsenic-induced targeted gene silencing. Several studies have observed extensive changes in global gene expression in individuals after arsenic exposure (Andrew et al. 2008;Bailey et al. 2009;Bourdonnay et al. 2009;Xie et al. 2007). Further, maternal exposure to arsenic has been shown to alter expression of transcripts in the mouse fetus (Liu et al. 2008) and human newborn (Fry et al. 2007). Because epi genetic processes are major regulators of gene expression, these findings suggest that dys regulation of epi genetic processes could contribute mechanistically to arsenic-induced changes in gene expression and cancer, affecting both people exposed to arsenic directly and those of future generations in a heritable manner, without directly altering the genome. Dysregulation of epi genetic processes could also contribute to vascular disease (Yan et al. 2010) and neurological disorders (Urdinguio et al. 2009).
Many groups have directly examined the association of arsenic exposure on epi genetic phenomena; because the technologies used to study the various epi genetic modifications are developing rapidly, we believe that a review of current findings from the literature is warranted. We conducted a PubMed search (National Center for Biotechnology Information, U.S. National Library of Medicine, Bethesda, MD) through April 2010 and identified studies using variable keywords, such as "arsenic AND DNA methyla tion," "arsenic AND microRNA," "arsenic AND histone modification," and "arsenic AND epigentics AND epigenomics." Our goal was to include all the studies we could find, and thus the reference lists of the identified studies were also reviewed to identify other relevant studies. Although epi genetic alterations may contribute to effects of arsenic on both cancer and non cancer outcomes, in this article we summarize the recent in vitro and in vivo research findings on the potential role of arsenicmediated epi genetic alterations in arsenicinduced toxicity and carcinogenicity. We discuss three major epi genetic mechanisms proposed to play roles in arsenic-induced carcino genesis: altered DNA methylation, histone modification, and microRNA (miRNA) expression. We also propose future directions that can further inform our understanding of the epi genetic and overall mechanisms underlying the effects of arsenic.

Arsenic Exposure and DNA Methylation
DNA methylation is tightly regulated in mammalian development and is essential for maintaining the normal functioning of the adult organism (Schaefer et al. 2007). Altered DNA methylation has been associated with several human diseases (Robertson 2005). Global genomic DNA hypo methylation is a hallmark of many types of cancers (Esteller et al. 2001), resulting in ille gitimate recombination events and causing transcriptional deregulation of affected genes (Robertson 2005). In mammalian systems, DNA methyla tion occurs predominantly in cytosine-rich gene regions, known as CpG islands, and serves to regulate gene expression and maintain genome stability (Yoder et al.   1997). DNA methyl transferases (DNMTs) are responsible for transferring a methyl group from the S-adenosyl methio nine (SAM) cofactor to the cytosine nucleo tide, producing 5´-methylcytosine and S-adenosyl homocysteine ( Figure 1) (Razin and Riggs 1980). Three different families of DNMT genes have been identified so far: DNMT1, DNMT2, and DNMT3 (Robertson and Wolffe 2000).
Mechanisms of arsenic-induced changes in DNA methylation. An association between arsenic-induced carcinogenesis and DNA methylation was proposed because arsenic methylation and DNA methylation both use the same methyl donor, SAM ( Figure 1). SAM is a coenzyme involved in > 40 metabolic reactions that require methyl group transfers (Chiang et al. 1996;Loenen 2006;Reichard et al. 2007). Because SAM is the unique methyl group donor in each conversion step of biomethylation of arsenic, long-term exposure to arsenic may lead to SAM insufficiency and global DNA hypo methylation (Coppin et al. 2008;Goering et al. 1999;Zhao et al. 1997). Further, because SAM synthesis requires methio nine, an essential amino acid in humans, dietary methyl insufficiency could exacerbate effects of arsenic on DNA methylation ( Figure 1) (McCabe and Caudill 2005). Indeed, human exposure to arsenic often occurs in relatively resource-poor populations in developing countries that also may have low dietary intakes of methionine (Anetor et al. 2007). In addition to its effect on SAM availability, arsenic can directly interact with DNMTs and inhibit their activities. Several studies have shown that arsenic exposure leads to a dose-dependent reduction of mRNA levels and activity of DNMTs both in vitro and in vivo, including DNMT1, DNMT3A, and DNMT3B (Ahlborn et al. 2008;Cui et al. 2006b;Fu et al. 2007;Reichard et al. 2007).
Arsenic and global DNA hypomethylation. Global DNA hypomethylation is expected to result from arsenic exposure through both SAM insufficiency and reduction of DNMT gene expression (Reichard et al. 2007). Arsenic exposure has been reported to induce DNA hypomethylation in vitro and in animal studies (Table 1). For example, rats (Uthus and Davis 2005) and mice Okoji et al. 2002;Xie et al. 2004) exposed to As III for several weeks displayed global hepatic DNA hypo methylation. Similarly, exposure of fish to As III for 1, 4, or 7 days resulted in sustained DNA hypomethylation compared with nonexposed fish (Bagnyukova et al. 2007). Studies in cell lines in vitro yielded similar results, with a reduction in global genomic DNA methylation resulting from As III exposure (Table 1) (Benbrahim-Tallaa et al. 2005;Coppin et al. 2008;Reichard et al. 2007;Sciandrello et al. 2004;Zhao et al. 1997). In contrast to the animal and in vitro findings, there are limited human population studies available. A cross-sectional study of 64 people reported by Majumdar et al. (2010) indicated that exposure to arsenic-contaminated water (250-500 µg/L) was associated with global DNA hypermethylation. However, the partici pants in the highest estimated exposure group (> 500 µg/L) had methylation levels that were comparable with those in the two lowest groups. The one possible reason for this inconsistency may be that the actual intake of arsenic into the body is different in the participants whose exposures were estimated based on the concentrations in their drinking water. In another well-designed nested case-control study, Pilsner et al. (2007) assessed the relation ship between arsenic and DNA methylation in 294 participants and observed a positive association between urinary arsenic and DNA hyper methylation. Plasma folate level apparently has a significant effect on the level of DNA methylation because a dose-response relation was evident only among participants with adequate folate levels (≥ 9 nmol/L) when estimates were stratified according to plasma folate level after controling for other factors. In a separate but closely related nested casecontrol study, Pilsner et al. (2009) found that individuals with hypo methylation of peripheral blood leukocyte (PBL) DNA were 1.8 (95% confidence interval, 1.2-2.8) times more likely to have skin lesions 2 years later after adjusting for age, urinary arsenic, and other factors. Pilsner et al. (2009) speculated that Adequate folate may be permissive for an adaptive increase in genomic methylation of PBL DNA associated with [arsenic] exposure, and that individuals who are similarly exposed but in whom the increase in genomic DNA methylation does not occur (or cannot be sustained) are at elevated risk for skin lesions.
Further studies are required to determine if exposure to As III has differential effects on the status of DNA methylation across tissues, cells, and species.
Arsenic and gene promoter methylation. Although the effects of arsenic exposure on global genomic DNA methylation remain unclear, DNA hypo methylation or hypermethylation of promoters of some genes has been reported in human skin cancer (Chanda et al. 2006) and bladder cancer Marsit et al. 2006c) associated with arsenic exposure. It has also been observed  Fu and Shen 2005;Jensen et al. 2008;Mass and Wang 1997), animal cell lines , Takahashi et al. 2002, animals (Cui et al. 2006a;Okoji et al. 2002;Waalkes et al. 2004), and humans (Chanda et al. 2006;Chen et al. 2007;Marsit et al. 2006b;Zhang et al. 2007b) exposed to arsenic (Table 2). Although this gene-specific effect observed in these studies could be due to study bias, because researchers examined only a small group of genes, the similar methylation pattern repeatedly reported in the same genes after arsenic exposure might also suggest that arsenic could selectively target specific genes. However, little is known about how DNA methylation is targeted to specific regions (Jones and Baylin 2002). Hypo-and hypermethylation of genes could mediate carcinogenesis through up-regulation of oncogene expression or down-regulation of tumor suppressor genes, respectively. Both observations have been reported. Hypomethylation of the promoter region of oncogenic Hras1 and an elevated Hras1 mRNA level was demonstrated in mice treated with sodium arsenite (Okoji et al. 2002). Similar results on mRNA expression and promoter hypo methylation of Hras1 and c-myc were also observed in vitro Takahashi et al. 2002). The evidence has linked over expression of Esr1 (estrogen receptor 1) gene with estrogeninduced hepato cellular carcinoma in mice (Couse et al. 1997). Arsenic exposure leads to over expression of the Esr1 gene resulting from hypo methylation of its promoter region, indicating an association between over expression of Esr1 and arsenic hepatocarcinogenesis Waalkes et al. 2004).
Dose-dependent hypermethylation at the promoter region of several tumor suppressor genes [e.g., p15, p16, p53, and death-associated protein kinase (DAPK)] was induced by arsenic exposure in vitro and in vivo (Boonchai et al. 2000;Chanda et al. 2006;Fu and Shen 2005;Mass and Wang 1997;Zhang et al. 2007b). In a population-based study of human bladder cancer in 351 patients, RASSF1A and PRSS3 promoter hyper methylation was positively associated with toenail arsenic concentrations, and promoter hypermethylation in both genes also was associated with invasive (vs. non invasive low grade) cancer (Marsit et al. 2006b). This outcome was recapitulated in arsenic-induced lung cancer in A/J mice, in which the arsenic exposure reduced the expression of RASSF1A resulting from hyper methylation of its promoter region and was associated with arsenic-induced lung carcino genesis (Cui et al. 2006a). DAPK is a positive mediator of γ-interferon-induced programmed cell death and a tumor suppressor candidate. In a study of 38 patients with urothelial carcinoma, Chen et al. (2007) reported hypermethylation of DAPK in 13 of 17 tumors in patients living in arsenic-contaminated areas compared with 8 of 21 tumors from patients living in areas not contami nated with arsenic. This hyper methylation of DAPK was also observed in an in vitro study when immortalized human uro epithelial cells were exposed to arsenic ). The increase of DNA hyper methylation of promoter in p16 was observed in arseniasis patients compared with people with no history of arsenic exposure (Zhang et al 2007b). In another study Chanda et al. (2006) examined the methyla tion status of promoters in p53 and p16 in DNA extracted from peripheral lymphocytes and observed an increase of methylation in both p53 and p16 associated with an estimated arsenic exposure in a dosedependent manner. However, this same study also showed that the subjects from the highest arsenic exposure group exhibited hypomethylation of both p53 and p16. Chronic exposure to arsenic in vitro has been shown to induce malignant trans forma tion in several human cell types (Benbrahim-Tallaa et al. 2005;Zhao et al. 1997) in which the alteration of DNA methylation level has been shown to be involved (Jensen et al. 2008(Jensen et al. , 2009aZhao et al. 1997).
Summary. Arsenic does not fall into the classic model of carcinogenesis because it is not efficient at inducing point mutations or initiating and promoting the development of tumors in experimental animals. One likely mechanism by which arsenicals operate is through the disruption of normal epi genetic control at specific loci, which may result in aberrant gene expression and cancer (Andrew et al. 2008;Xie et al. 2007). Although there is increasing evidence that arsenic exposure alters methylation levels in both global DNA and promoters of some genes, the current available studies are essentially descriptive and difficult to interpret because of the complexity of the study populations and limited information provided in the reports. Studies are needed that systematically investigate DNA methylation on a genomewide level in arsenic-exposed cell lines and in target tissues, such as exfoliated bladder cells, from well-characterized arsenic-exposed human populations, or in tumor tissue from arsenic-associated cancers. Such studies would help to clarify potential effects of arsenic exposure on DNA methylation and carcinogenesis. Abbreviations: ERα, estrogen receptor α; Hyper, hypermethylated; Hypo, hypomethylated; NA, not available. a Study subjects were grouped based on historical arsenic concentration in drinking water, and the range of arsenic concentration in drinking water was < 50 µg/L to > 300 µg/L. b The estimated toenail arsenic concentration of study subjects was < 0.01 µg/L to > 50 µg/L.

Arsenic Exposure and Histone Modification
Chromatin is structured within the cell nucleus in units called nucleosomes, in which DNA is packaged within the cell. The nucleo some core particle consists of stretches of DNA (~ 146 bp) wrapped in left-handed superhelical turns around a histone octamer consisting of two copies each of the core histones H2A, H2B, H3, and H4 (Luger et al. 1997). Although H1 does not make up the nucleosome "bead," H1 plays a role in keeping in place the DNA that has wrapped around the nucleosome (Figure 2). From a structural and functional perspective, histones have different characteristics depending on the number of amino acids and the number and type of covalent modi fi ca tions in these residues. These covalent modifications, found in the tails of the histone chains, influence many fundamental biological processes including acetylation, methylation, phosphorylation, citrullination, ubiquitination, sumoylation, ADP ribosylation, deimination, and proline isomerization (Kouzarides 2007) (Figure 2). To date, published studies on histone modifications and arsenic toxicity have focused on acetylation, methylation, and phosphorylation. Histone acetylation. Histone acetylation is a dynamic and reversible event (Glozak and Seto 2007), in which the acetylation status of lysine residues in the histone tail is regu lated by two antagonistic enzyme classes, histone acetyl transferases (HATs) (Sterner and Berger 2000) and histone deacetylases (HDACs) (Cress and Seto 2000). Using acetyl coenzyme A as an acetyl group donor, HATs enzymatically transfer a single acetyl group to the ε-amino group of specific lysine side chains within the histone's basic N-terminal tail region, whereas HDACs remove the acetyl group from the lysine residues.
Evidence for an association between altered histone acetylation and arsenic-induced toxicity continues to be strengthened. In the early 1980s, arsenic exposure was shown to significantly reduce histone acetylation in Drosophila (Arrigo 1983). More recently, changes in histone H3 acetylation have been observed in association with As III -and MMA III -induced malignant transformation of human urothelial cells in vitro; these modifications apparently are arsenic specific because the co-occurring changes in both As III -and MMA III -induced malignant transformation are significantly more frequent than those occurring by random chance (Jensen et al. 2008). Further, Jensen et al. (2008) reported DNA hyper methylation in a number of the hypo acetylated promoters identified in the study, suggesting that arsenic coordinately targets genes through dys regulation of different epi genetic mechanisms contributing to malignant transformation. Recently, we showed that the global level of H4K16 acetylation in human bladder epithelial cells was reduced in a dose-and time-dependent manner by both As III and MMA III treatment (Jo et al. 2009). Moreover, knockdown of MYST1, the gene responsible for H4K16 acetylation, resulted in increased cytotoxicity from arsenical exposure in human bladder epithelial cells, suggesting that H4K16 acetylation may be important for resistance to arsenic-induced toxicity.
Interestingly, As III exposure has also been shown to induce elevated histone acetylation, which was reportedly responsible for the up-regulation of genes involved in apoptosis or the response to cell stress after exposure to arsenic (Li et al. 2002(Li et al. , 2003. This result probably is mediated by HDACs. As III has been shown to inhibit HDAC genes that correlate with increased global histone acetylation (Ramirez et al. 2008). The level of inhibition is comparable with that of the well-known HDAC inhibi tor tricho statin A (Drummond et al. 2005). Together, these studies clearly provide evidence that histone acetylation is dys regulated by arsenic exposure, but further work is needed to understand the under lying mechanisms and to clarify the net effect of altered histone acetylation on arsenic-induced toxicity and carcinogenesis.
Histone methylation. Like acetylation, histone methylation is also a reversible process. However, unlike acetylation, which occurs only on lysine residues at the histone tail, histone methylation occurs on both lysine and arginine residues (Martin and Zhang 2005;Wysocka et al. 2006). In mammals, histone methylation is usually found on histone H3 and H4, although it also occurs on H2A or H2B. Arginine methylation is catalyzed by the enzyme arginine N-methyltransferase (Wysocka et al. 2006), whereas lysine methylation is catalyzed by two different classes of proteins, the SET-domain-containing protein family and the non-SET-domain proteins DOT1/DOT1L (Martin and Zhang 2005). Histone methylation can occur in the mono methyl, symmetrical dimethyl, and asymmetrical di methyl states and in the tri methyl group states, in contrast to the single acetyl group added during acetylation (Klose and Zhang 2007). Histone methylation was considered a static modification until recent years, when enzymes were found to be capable of antagonizing histone arginine methylation or directly removing a methyl group from a lysine residue of histone (Klose and Zhang 2007). These enzymes include peptidyl arginine deiminase enzymes and amine oxidase-and JmjC domain-containing histone demethylase enzymes.
Accumulating evidence implicates the aberrant loss or gain of histone methylation in tumori genesis (Schneider et al. 2002). Arrigo (1983) first reported that exposure to arsenic in Drosophila cells led to a complete abolishment of methylation of histones H3 and H4, and the effect on H3 was later confirmed by other investigators Tanguay 1986, 1988). The response to arsenic exposure in the mammalian cell is more complex, and As III treatment can lead to differential effects on the methylation of H3 lysine residues, including increased H3 lysine 9 dimethyla tion (H3K9me2) and H3 lysine 4 tri methyla tion (H3K4me3) and decreased H3 lysine 27 trimethylation (H3K27me3) (Zhou et al. 2008). Zhou et al. (2009) showed that 1 µM arsenite  volume 119 | number 1 | January 2011 • Environmental Health Perspectives significantly increased H3K4me3 after 24-hr or 7-day exposures in human lung carcinoma A549 cells. Importantly, H3K4me3 remained elevated, apparently inherited through cell division, 7 days after the removal of arsenite. Elevated H3K9me2, mediated by increased levels of histone methyl transferase G9a protein (Zhou et al. 2008), correlates with transcriptional repression (Peterson and Laniel 2004) and has been shown to be involved in the silencing of tumor suppressers in the cancer cell lines (Esteve et al. 2007;McGarvey et al. 2006). However, data on the patterns of histone methylation induced by arsenic exposure are limited, and further studies are required to decipher the relationship between altered histone methylation and gene expression, as well as its effect on arsenic-induced carcinogenesis. Histone phosphorylation. All four core histone proteins, H2A, H2B, H3, and H4, and the linker histone H1 can be posttranslationally modified by phosphorylation. Cyclin-dependent kinases are believed to be responsible for H1 phosphorylation (Swank et al. 1997). Several kinases are able to phosphorylate H2A and H2B, such as ataxia telangiectasia mutated for H2AX (Burma et al. 2001). Phosphorylation of H3 has been specifically implicated in cell cycle progression and regulation of gene expression (Houben et al. 2007). Similarly, phosphorylation of histone H4 (serine 1) increases during the cell cycle and is believed to be regulated by casein kinase 2 ( Barber et al. 2004).

Scheme of nucleosome
Histone phosphorylation may also contribute to arsenic-induced carcinogenesis. Although all four core histones (H2A, H2B, H3, and H4) are targets of protein kinases (Peterson and Laniel 2004), the best-studied histone phosphorylation event is that of H2AX, a form of H2A that represents up to 25% of the total H2A pool in mammals. Zykova et al. (2006) demonstrated that arsenic tri oxide induces apoptosis by up-regulation of phosphorylated H2AX and may be one of the mechanisms by which arsenic tri oxide acts as an anti neoplastic agent (Figure 2). Little is known about histone phosphorylation and arsenic carcino genesis. Studies have suggested that H3 phosphorylation induced by arsenic exposure might be responsible for the up-regulation of the oncogenes c-fos and c-jun (Li et al. 2003) and induction of a proto apoptotic factor, caspase 10 (Li et al. 2002). Nickel, another important metal with epi genetic effects, has been shown to induce phosphorylation of histone 3, specifically H3S10 (serine 10) via the activation of the JNK/SAPK (c-jun N-terminal kinase/stressactivated protein kinase) pathway . Because arsenite exposure is known to activate JNK and p38/Mpk2 kinase by inhibition of the corresponding protein phosphatases (Cavigelli et al. 1996), phosphorylation of histone H3 via the JNK/SAPK pathway might be a common mechanism of metal-induced histone modification.
Different types of histone modifications have been shown to affect gene regulation and expression in a coordinated manner. For example, WNT5A gene expression is up-regulated in As III -and MMA III -induced malignant transformation in uro epithelial cells in association with the enrich ment of permissive histone modifications and reduction of repressive modi fi cations in the WNT5A promoter region (Jensen et al. 2009b). Two modifications of histone H3, dimethylation of H3K4 and acetyla tion of H3K9 and H3K14, are associated with transcriptional competency, whereas the other two modifications of histone H3, tri methylation of H3K27 and dimethyla tion of H3K9, are correlated with transcriptional repression (Peterson and Laniel 2004).
Summary. Although we are still in the early stages of elucidating the association between histone modifications induced by arsenic and their effects on arsenic carcinogenicity, newly available techniques such as mass spectrometry (MS)-based histone modification analysis and genome wide sequencing offer the potential to systematically charac terize the altered histone modifications induced by arsenicals and the subsequent changes in gene expression.

Arsenic Exposure and miRNA Expression
In the past few years, several laboratories have discovered a small class of non-protein-coding RNAs, called microRNAs (miRNAs), that participate in diverse biological regulatory events and are transcribed mainly from non-proteincoding regions of the genome (Bartel 2004;He and Hannon 2004). More than 700 human miRNAs have been identified to date, as documented in the miRBase database (Release 14; miRBase 2009), and it is predicted that many more exist. Each miRNA is thought to target several hundred genes, and as many as 30% of mammalian genes are regulated by miRNA (Lewis et al. 2005). miRNAs deactivate gene expression by binding to the 3´-untranslated region of mRNA with incomplete base pairing (Wightman et al. 1993). The exact mechanisms by which expression is repressed are still under investigation but may include the inhibition of protein synthesis, the degradation of target mRNAs, and the trans loca tion of target mRNAs into cyto plasmic processing bodies (Jackson and Standart 2007). Because of the suppressive effect of miRNA on gene expression, a reduction or elimination of miRNAs that target oncogenes could result in the inappropriate expression of those onco proteins; for example, Johnson et al. (2005) have shown that RAS oncogene is regulated by the let-7 miRNA family. Conversely, the amplification or over expression of miRNAs that have a role in regulating the expression of tumor suppressor genes could reduce the expression of such genes. A prime example of this is the observation of the miR-34 family on the p53 tumor suppressor pathway (He et al. 2007).
Altered miRNA expression and arsenic exposure. Despite the significant progress made toward understanding the bio genesis and mechanisms of action of miRNA, much less is known about the effect of environ mental exposures, especially carcinogens such as arsenic, on miRNA expression. Several studies have shown that exposure to exogenous chemi cals can alter miRNA expression (Kasashima et al. 2004;Pogribny et al. 2007;Shah et al. 2007).
In vitro exposure of cells to iron sulfate or aluminum sulfate, which generate reactive oxygen species (ROS), led to the up-regulation of a specific set of miRNAs, including miR-9, miR-125b, and miR-128 (Lukiw and Pogue 2007). ROS generation resulting from arsenic exposure is thought to play a large role in arsenicinduced carcinogenesis and toxicity (Flora et al. 2007;Hei and Filipic 2004) and could potentially alter these miRNAs in a similar manner. Marsit et al. (2006a) examined the roles that arsenic and folate deficiency play in miRNA expression; these authors found that human lymphoblast TK6 cells that had been treated with sodium arsenite and cells that had been grown in folate-deficient media over a 6-day period showed similarly altered expression of five miRNAs compared with untreated controls, suggesting a common mechanism of dys regulation. One such potential mechanism is aberrant DNA methylation occurring as a result of SAM depletion (Caudill et al. 2001;Loenen 2006), which arises under conditions of arsenic exposure and folate deficiency. However, Caudill et al. (2001) found no significant decrease in global methylation in the treated compared with the control groups, suggesting more subtle or targeted effects. The induced changes in miRNA expression were not stable and returned to baseline levels upon removal of the stress conditions, suggesting that chronic exposure may be necessary to permanently alter expression of miRNAs (Marsit et al. 2006a). Arsenic trioxide, a treatment option for acute pro myelo cytic leukemia (APL) (Zhou et al. 2005), induces the relocalization and degradation of the nuclear body protein promyelocytic leukemia (PML) protein, as well as the degradation of PMLretinoic acid receptor-α (PML-RARα) in APL cells (Shao et al. 1998). APL patients treated with all-trans retinoic acid release a group of miRNAs transcriptionally repressed by the APL-associated PML-RAR oncogene (Saumet et al. 2009), suggesting that arseni cals may produce similar effects on miRNA expression in APL patients.
Summary. Overall, these studies show that environmental carcinogen exposures can lead to altered miRNA expression profiles, which may be associated with the process of carcino genesis. Further studies are necessary to clarify whether chronic exposure to arsenic is capable of altering miRNA expression and what biological effects are related to the altered miRNA expression.

Epigenomic Approach Proposed for Future Studies
Emerging evidence suggests that arsenic acts through several epi genetic mechanisms. The characterization of genome wide patterns of DNA methylation, post translational histone modification, and miRNA expression after arsenic exposure in vitro and in vivo represents a new frontier toward our understanding of the mechanisms of arsenic toxicity and carcino genesis. Emerging epi genomic technologies such as chromatin immunoprecipitation (ChIP)-on-chip and ChIP sequencing (ChIP-seq), global methylation, and miRNA micro arrays, as well as whole genome DNA sequencing platforms, will facilitate these efforts (Schones and Zhao 2008). ChIP-on-chip and ChIP-seq, used primarily to determine how proteins interact with DNA, have the potential to clarify how epi genetic changes, particularly histone modifications, induced by arsenic exposure regulate gene expression (Park 2009). MS offers an un biased approach to mapping the combinations of histone modifications and requires highly sensitive and precise mass measurements; for example, the difference in mass between tri methylation and acetylation is only 36 mDa. Using liquid chromatography-MS, we identified acetylation of H4K16 as a histone modification that is significantly reduced after arsenic treatment, especially with longterm exposure (Jo et al. 2009).
With the rapid development of array and sequencing-based DNA-methylation profiling technologies, global DNA methylation profiling has clearly come of age. Because epi genetic modifications alter gene expression but not gene sequence, transcriptomics may eventually allow the charac teriza tion of the expression profiles of epi genetically labile genes. Identification of the genes dys regulated through epi genetic mechanisms by arsenic exposure will further elucidate the associated biological processes and disease states. Proteomics using both conventional "bottom-up" and newer cutting-edge "top-down" MS approaches to detect labile post translational modifications that are often lost in conventional MS/MS experiments will allow further clarification of the resulting pheno type. The difference between these two approaches is that the materials introduced into the mass spectrometer are either peptides generated by enzymatic cleavage of one or many proteins in the "bottom-up" approach, or intact protein ions or large protein fragments in the "top-down" approach. Integration of epi genetic, transcriptomic, and proteomic data sets generated by these techniques will facilitate a more thorough under standing of the interplay of these processes under normal conditions and during arsenic exposure. Indeed, the importance of a comprehensive understanding of the epi genome has been recognized by the scientific community and is reflected in the National Institutes of Health (NIH) Roadmap Initiative (NIH 2007) with the goal of developing comprehensive reference epi genome maps and new technologies for comprehensive epigenomic analyses.

Conclusion and Future Directions
Although experiments in suitable model systems could complement the human studies, as discussed above, there may be differences between epi genetic effects in animals and humans and between various tissues and cell types. Thus, studies in human populations exposed to high levels of arsenic will be necessary to understand how individual differences in arsenic methylation and genetic background, as well as environmental factors such as diet and age, influence the epigenetic response to chronic arsenic exposure. Studies will also be required across various tissue and cell types to identify and validate the levels and patterns of epi genetic markers in these cells. Accessible tissues such as blood may not represent a good surrogate of target tissues such as bladder, kidney, and lung. High-resolution methylation data have shown that tissues have distinct epi genetic profiles (Christensen et al. 2009;Illingworth et al. 2008), and aging and environmental exposures may alter methylation in a tissue-specific manner (Christensen et al. 2009). Thus, epigenetic profiles from disease-relevant tissues such as exfoliated bladder cells from exposed and unexposed disease-free individuals could allow early effects to be identified. Such cells could also be analyzed from individuals with arsenic-and non-arsenic-associated cancers to identify arsenic-associated tumorigenic profiles. Rosser et al. (2009) showed that it may be possible to detect bladder cancer using gene expression signatures in exfoliated bladder urothelia. Similarly, the effects of inhaled arsenic on epi genetic profiles in bronchial airway epithelial cells could be examined in exposed and unexposed disease-free individuals and those with lung cancer, as was recently done using miRNA profiling for cigarette smoke exposure (Schembri et al. 2009).
In conclusion, a comprehensive epigenomic approach may elucidate the mechanisms of arsenic-induced carcinogenesis. Such an approach would also facilitate the discovery of biomarkers of arsenic exposure and early effects, associated diseases and disease progression, and factors that confer susceptibility.