Associations between Blood and Urine Arsenic Concentrations and Global Levels of Post-Translational Histone Modifications in Bangladeshi Men and Women

Abstract

Introduction

Throughout the world, > 140 million people are exposed to arsenic-contaminated drinking water (Bagchi 2007); in Bangladesh alone, ≤ 57 million individuals are exposed (Kinniburgh et al. 2003). Chronic exposure to arsenic causes bladder, lung, and skin cancers and is also associated with numerous noncancer health outcomes [National Research Council (NRC) 2013]. Susceptibility to many of these arsenic-related health outcomes differs by sex, with some outcomes preferentially afflicting males and others females (NRC 2013). For example, males are more likely to develop arsenic-induced skin lesions (Ahsan et al. 2006b; Watanabe et al. 2001) and skin, liver, and bladder cancers (Chen and Wang 1990; Chen et al. 2003; Leonardi et al. 2012), whereas females may be more susceptible to arsenic-induced developmental outcomes (Gardner et al. 2013; Hamadani et al. 2011; Saha et al. 2012) and cardiovascular disease (CVD) (Moon et al. 2013). However, the mechanisms underlying these sex differences remain unknown.

Experimental studies and observational studies in human populations have demonstrated that arsenic alters epigenetic modifications, including global 5-methylcytosine (5-mC) (Niedzwiecki et al. 2013; Pilsner et al. 2007; Ren et al. 2011; Tellez-Plaza et al. 2014) and 5-hydroxymethylcytosine (5-hmC) (Niedzwiecki et al. 2015; Zhang J et al. 2014), and there is evidence that these effects differ by sex (Broberg et al. 2014; Niedzwiecki et al. 2015; Nohara et al. 2011; Pilsner et al. 2012). In vitro and rodent studies have also shown that arsenic alters global (percent) post-translational histone modifications (PTHMs) in tissues or in cell lines derived from tissues that are targets of arsenic toxicity, such as the lung (Zhou et al. 2008), bladder (Chu et al. 2011), and brain (Cronican et al. 2013), and the effects of arsenic on %PTHMs in the brain have been shown to be sex-dependent in mice (Tyler et al. 2015). An epidemiological study of 63 male steel workers reported that arsenic exposure via inhalation was associated with increased global levels of histone H3 lysine 4 dimethylation in white blood cells (WBCs) (Cantone et al. 2011); however, because this study only included men, potential differences by sex could not be examined.

In a previous study of 40 Bangladeshi adults, we observed sex-specific associations between arsenic, measured in urine, and several %PTHMs (H3 lysine 4 trimethylation, H3 lysine 27 trimethylation, and H3 lysine 27 acetylation) in peripheral blood mononuclear cells (PBMCs) (Chervona et al. 2012). We have also previously observed sex-specific associations between arsenic exposure and global levels of both 5-mC and 5-hmC in PBMC DNA (Niedzwiecki et al. 2015). Thus, we now present data on three PTHMs: histone H3 lysine 79 dimethylation (H3K79me2), selected because this modification has been shown to regulate the expression of Tet1 (Huang et al. 2013; Williams et al. 2011), which converts 5-mC to 5-hmC (Ito et al. 2011), and because it is dysregulated in cancers (Bernt et al. 2011; Zhang L et al. 2014), and histone H3 lysine 36 di- and trimethylation (H3K36me2 and H3K36me3, respectively), selected because these two PTHMs have been shown to be altered by arsenic in vitro (Zhou et al. 2008) and because they are also dysregulated in several types of cancer (Duns et al. 2010; Fontebasso et al. 2013; He et al. 2011; Tamagawa et al. 2013). This study utilized a subset of PBMC samples collected from participants enrolled in the Folic Acid and Creatine Trial (FACT) (Peters et al. 2015). First, we evaluated sex-specific associations between arsenic and our candidate PTHMs using baseline FACT samples. Then, because all participants in the trial were provided with arsenic-removal water filters at enrollment, we evaluated whether %PTHMs were altered after reduction of arsenic exposure; this evaluation was performed using samples collected at baseline and at week 12 from participants who did not receive a dietary supplement. The data reported herein add to a growing body of evidence indicating that arsenic induces epigenetic dysregulation on a global level and, moreover, that this dysregulation often occurs in a sex-specific manner.

Methods

Results

Discussion

In an adult population in Bangladesh, we examined associations between arsenic exposure and three %PTHMs (%H3K36me2, %H3K36me3, and %H3K79me2), which were selected because they are dysregulated in several types of cancer (Bernt et al. 2011; Duns et al. 2010; Fontebasso et al. 2013; Tamagawa et al. 2013; Zhang L et al. 2014) and are altered by arsenic in vitro (Zhou et al. 2008) or because they regulate 5-hmC (Huang et al. 2013; Williams et al. 2011). Percent 5-hmC has been shown to be altered by arsenic in male rats (Zhang J et al. 2014) and in humans in a sex-dependent manner (Niedzwiecki et al. 2015). We observed that arsenic was associated with higher levels of %H3K36me2, but only in men. Interestingly, the use of arsenic-removal water filters, which was associated with significant reductions in both bAs (Peters et al. 2015) and uAsCr, was also associated with significant reductions in %H3K36me2 in the whole sample. However, in sex-stratified analyses, although we observed that %H3K36me2 declined in both men and women, the decline only achieved statistical significance among women. Given that ethical considerations did not allow for a comparison group that did not receive arsenic-removal water filters, we cannot rule out the possibility that the decline in %H3K36me2 may have been caused by extrinsic factors. Additional studies will be needed to confirm the changes we observed in %H3K36me2 in association with arsenic removal. In cross-sectional analyses, arsenic exposure was not associated with %H3K36me3. Additionally, %H3K36me3 did not change over time despite reductions in arsenic exposure. Thus, arsenic did not appear to alter %H3K36me3 in histones derived from PBMCs. Although in cross-sectional analyses arsenic exposure was not associated with %H3K79me2, which has been shown to regulate the expression of Tet1 (Huang et al. 2013; Williams et al. 2011), we did observe a suggestive (p = 0.07) decline in %H3K79me2 over time after individuals received arsenic-removal water filters; this observation needs to be confirmed in a larger sample.

Although neither %H3K36me3 nor %H3K79me2 changed significantly over the 12-week period, we cannot make definitive conclusions about the stability of these marks because all participants received arsenic-removal water filters at baseline and were thus subject to an intervention; furthermore, these marks did vary over time in some participants. Little is known about the stability of %PTHMs in human PBMCs. One study by Zhang et al. examined the stability of PTHMs during adipogenesis, and the authors observed that although gene-specific levels of PTHMs were highly dynamic, global levels of PTHMs were remarkably stable (Zhang et al. 2012). However, because Zhang et al. used murine adipocyte cell lines, it is unclear if these findings are relevant to %PTHM stability in human PBMCs. One previous epidemiological study measured %PTHMs in PBMCs collected at three 1-week intervals from 15 nickel refinery workers and from 15 individuals who had not been exposed occupationally to nickel, and the authors observed that the interindividual variances in %PTHMs were much higher than the intraindividual variances, suggesting that %PTHMs are relatively stable over time in human PBMCs (Arita et al. 2012). However, because this evaluation was made over a short duration and in a small number of participants, this area requires additional investigation.

Our study has several potential limitations. First, given the cross-sectional nature of some of the analyses, we need to consider the possibility of reverse causality. Because several previous experimental studies have shown that arsenic influences %PTHMs, it is unlikely that reverse causality would explain our findings. However, it is possible that %PTHMs influence the expression of genes involved in arsenic metabolism, such as arsenic (+3 oxidation state) methyltransferase, which could thereby influence the excretion of arsenic, thus altering bAs and uAs_Cr concentrations. Although residual confounding is another important consideration, the associations between arsenic measures and %PTHMs were quite robust even after adjusting for additional covariates. Another important consideration for our study is the fact that %PTHMs were measured in human PBMCs, which comprise a mixed population of cell types. However, global DNA methylation levels, which are closely related to %PTHMs, have been shown to be very similar between blood cell types (reviewed in Smith and Meissner 2013). Additionally, a cross-sectional study of 63 male steel workers that examined associations between inhalation exposure to occupational toxicants, including arsenic, and %PTHMs in total WBCs evaluated the influence of cell-type distribution on these associations, and the authors observed that although adjusting for the proportion of granulocytes influenced their results, adjusting for other cell types did not have a major impact on the findings (Cantone et al. 2011). Because we measured %PTHMs in PBMCs, which do not include granulocytes, potential shifts in the proportion of granulocytes could not explain the associations we observed between arsenic and %PTHMs. However, we cannot rule out the possibility that alterations in the proportion of monocytes, natural killer cells, T cells, B cells, or their subpopulations may have affected our findings, and this is an area that merits additional investigation. Finally, an important limitation of our study is that the sample sizes for prospective analyses were small. Therefore, we may have had insufficient statistical power to formally examine sex differences in the influence of arsenic removal on %PTHMs.

Despite some of the limitations of this study, our findings support those of a previous experimental study performed in A549 cells that examined the effects of arsenite (2.5 and 5 μM) on %H3K36me2 and %H3K36me3 (Zhou et al. 2008); in that study, the authors observed that arsenic decreased %H3K36me2 and increased %H3K36me3. Although we observed a positive association between uAs_Cr and %H3K36me2 in men and did not observe an association between uAs_Cr and %H3K36me3, we studied a population that had been exposed to arsenic-contaminated drinking water for years to decades, whereas Zhou et al. measured %PTHMs in a cell line that had been exposed to arsenic for a 24-hr period. Furthermore, our study participants were healthy individuals, whereas A549 cells are alveolar basal epithelial cells derived from a male human lung tumor; arsenic may have distinct effects in different tissues and may also have different effects in normal versus cancerous cells. Additionally, in vitro studies are limited in that they cannot account for the numerous systemic differences associated with sex in vivo. Nevertheless, it is quite interesting that arsenic appears to target H3K36me2 in such diverse models.

Although the consequences of arsenic-induced increases in %H3K36me2 are unknown at the present time, H3K36me2 has been implicated in oncogenic programming (Kuo et al. 2011). Furthermore, some of the enzymes responsible for regulating this mark, such as methyltransferase NSD2, are overexpressed in multiple cancers, including those caused by arsenic, such as bladder, lung, and skin cancers (Hudlebusch et al. 2011). A global increase in H3K36me2 leads to widespread increases in this mark across the genome, thereby altering its typical distribution (Popovic et al. 2014), which may have profound effects on both gene expression and genomic stability. For example, a global increase in H3K36me2 is associated with increased levels of H3K36me2 within gene bodies, which in turn is associated with increased expression of genes involved in oncogenic programming (Kuo et al. 2011).

Similar to our findings, several studies have observed sex-specific effects of arsenic on other epigenetic marks, such as DNA methylation (Broberg et al. 2014; Nohara et al. 2011; Pilsner et al. 2012), including our previous finding that arsenic exposure is associated with increased %5-mC and %5-hmC in men but not in women (Niedzwiecki et al. 2015). Sex-specific effects of other environmental contaminants, such as cadmium and lead, on DNA methylation have also been observed (Faulk et al. 2013; Kippler et al. 2013). Because PTHMs can direct DNA methylation patterns (Cedar and Bergman 2009), PTHMs may mediate the effects of environmental contaminants, such as arsenic, on DNA methylation marks. In addition to the sex-specific findings for uAs_Cr and %H3K36me2 reported herein, our group previously observed sex-specific correlations between uAs_Cr and several other %PTHMs (Chervona et al. 2012). Additionally, Tyler et al. recently demonstrated that arsenic alters %PTHMs in a sex-dependent manner in mouse brain (Tyler et al. 2015). Although the mechanisms are not fully understood, arsenic also altered the expression of corresponding histone-modifying enzymes, including Mll and Kdm5b, in a sex-dependent manner (Tyler et al. 2015). Many histone-modifying enzymes have also been shown to interact with the androgen receptor (Heemers and Tindall 2007), and some histone demethylases are dosage-sensitive regulators that are coded for by genes that reside on the Y chromosome and are highly conserved across mammalian species and broadly expressed across tissues and cell types (Bellott et al. 2014). Thus, both hormonal influences and genetic differences may contribute to the sex-specific effects of arsenic on %PTHMs.

For many arsenic-induced health outcomes, susceptibility differs by sex (NRC 2013). For example, men are more susceptible to developing arsenic-induced skin lesions (Ahsan et al. 2006b; Watanabe et al. 2001) and cancers of the skin, liver, and bladder (Chen and Wang 1990; Chen et al. 2003; Leonardi et al. 2012). In contrast, several studies have reported that early-life exposure to arsenic is associated with impaired intellectual function and other developmental outcomes among female, but not male, children (Gardner et al. 2013; Hamadani et al. 2011; Saha et al. 2012). Additionally, in the United States, the arsenic-associated risk for CVD was found to be higher among women (Moon et al. 2013). Animal studies have also revealed sex-specific effects of arsenic for many outcomes. For example, female mice are more susceptible to arsenic-induced changes in locomotor activity (Bardullas et al. 2009) and are more likely to develop lung tumors as a result of prenatal exposure to arsenic, whereas males are more likely to develop liver and adrenal tumors in response to such exposures (Waalkes et al. 2003).

Although some of the sex-specific effects of arsenic observed in human populations may be explained by sex differences in coexposures (e.g., cigarette smoking, UV exposure, and nutritional deficiencies), the dramatic sex differences observed in well-controlled animal studies of arsenic toxicity suggest that coexposures are not solely responsible for these differences. One consideration is that women have a higher capacity to fully methylate inorganic arsenic to dimethyl arsenic species, which facilitates arsenic elimination in urine (Hopenhayn-Rich et al. 1996; Hsueh et al. 2003; Lindberg et al. 2007), which should generally reduce arsenic toxicity for women. However, some arsenic-related health outcomes preferentially afflict women; thus, there are likely other contributing factors. Epigenetic dysregulation has been implicated in the development of arsenic-induced health outcomes, including skin lesions and cancers of the skin and bladder (Chanda et al. 2006; Pilsner et al. 2009; Smeester et al. 2011; Wilhelm et al. 2010). Thus, epigenetic dysregulation may be one important mechanism contributing to the sex differences observed for multiple arsenic-related health outcomes. Previous studies examining the sex-specific effects of arsenic on epigenetics have focused on DNA methylation. However, the results of this study and those of our previous study (Chervona et al. 2012) suggest that arsenic also induces sex-specific alterations in %PTHMs.

Conclusions

Our findings have two major implications that warrant further investigation. First, arsenic exposure was associated with %H3K36me2 in a sex-dependent manner in our study population of adults in Bangladesh. Although it is tempting to speculate that these findings may explain some of the observed sex differences in susceptibility to arsenic-induced diseases, the impact of %H3K36me2 and other %PTHMs on health outcomes will require further study. Second, the arsenic-associated increase in %H3K36me2 observed in men decreased, albeit non-significantly, after the use of arsenic-removal water filters. However, because we did not have a comparison group that did not receive water filters, and because %H3K36me2 decreased significantly in women, future studies will be needed to evaluate the effects of arsenic removal on %H3K36me2 and to investigate whether downstream effects of alterations in %PTHMs, such as changes in DNA methylation patterns, persist over time.

Supplemental Material

Footnotes

References