Genetic and epigenetic mechanisms underlying arsenic-associated diabetes mellitus: a perspective of the current evidence

Abstract

Chronic exposure to arsenic has been associated with the development of diabetes mellitus (DM), a disease characterized by hyperglycemia resulting from dysregulation of glucose homeostasis. This review summarizes four major mechanisms by which arsenic induces diabetes, namely inhibition of insulin-dependent glucose uptake, pancreatic β-cell damage, pancreatic β-cell dysfunction and stimulation of liver gluconeogenesis that are supported by both in vivo and in vitro studies. Additionally, the role of polymorphic variants associated with arsenic toxicity and disease susceptibility, as well as epigenetic modifications associated with arsenic exposure, are considered in the context of arsenic-associated DM. Taken together, in vitro, in vivo and human genetic/epigenetic studies support that arsenic has the potential to induce DM phenotypes and impair key pathways involved in the regulation of glucose homeostasis.

Type 2 diabetes mellitus (DM) is a metabolic disorder characterized by disruption of the insulin-signaling pathway resulting in insulin resistance and pancreatic β-cell dysfunction and in limited uptake of glucose by cells. More than 640 million people are affected by Type 2 DM globally, with up to 40% unexplained by rises in obesity and other associated lifestyle factors [1]. Numerous studies have investigated the role of environmental contaminants in the development of DM [2,3]. Of the chemicals investigated, the 2011 National Toxicology Program Workshop on DM and the Environment found sufficient evidence for the relationship between exposure to moderate and high (≥150 µg As/l) levels of inorganic arsenic in drinking water and DM [4]. At the time, the Workshop deemed the evidence for association between moderate-to-low level (<150 µg As/l) exposure to be insufficient [4]. More recently, two meta-analyses that include 19 new studies have been conducted substantiating the association between chronic arsenic exposure at levels less than 150 µg As/l and DM [5,6]. These meta-analyses found pooled relative risks of developing DM in the presence of arsenic equal to 1.7 and 1.23, respectively [5,6].

Globally, more than 200 million people are exposed to inorganic arsenic at levels that exceed 10 µg As/l [7]. Arsenic is associated with numerous health endpoints including, including cancer of the skin and bladder, cardiovascular disease, stroke, neurological effects, peripherial vascular disease, and DM [8]. Several plausible mechanisms underlying DM have been proposed based upon in vivo and in vitro laboratory studies [8,9]. The data support the impact of arsenic on processes that govern glucose homeostasis, supporting that arsenic impairs insulin-dependent glucose uptake by inhibiting insulin-activated signal transduction pathway, upregulates gluconeogenesis and inhibits glucose-stimulated insulin secretion by pancreas [10]. These mechanisms are further supported by epidemiological studies in arsenic-exposed human populations where arsenic is associated with biomarkers of pathological processes typically associated with DM. Specifically these biomarkers represent renal damage, oxidative stress, low-grade inflammation and endothelial damage [11]. There are also provocative data suggesting a role for epigenetic alterations underlying arsenic-associated DM [12–17]. To date, a limited number of studies have examined the role of the epigenome in the context of arsenic-associated DM. As arsenic is known to modify various components of the epigenetic machinery [18], it is likely that study of the epigenome could provide key insight into mechanisms by which arsenic induces DM in humans.

In the present review, we summarize the molecular mechanisms proposed to underlie arsenic-associated DM. Additionally, existing studies focusing on arsenic-associated DM including those that assess genetic variants linked to DM as well as arsenic-associated modification of the epigenome are detailed. These genetic variants and epigenetic alterations could play a critical role in driving molecular changes that induce DM. Finally, gaps in the literature are noted and potential directions for future research are discussed.

Mechanisms of arsenic-associated DM

DM is a group of metabolic diseases characterized by hyperglycemia that results from disruption of key processes involved in the regulation of glucose homeostasis. The sustained hyperglycemia is the result of the failure to produce or secret insulin in the pancreatic β-cells and/or the failure of peripheral tissues to respond to insulin signaling. Type 1 DM is the result of an autoimmune condition where pancreatic β-cells are destroyed resulting in a loss of insulin and the subsequent inability of tissues to utilize blood glucose. In contrast, Type 2 DM is characterized by impaired insulin signaling (i.e., insulin resistance) in peripheral tissues, and subsequent β-cell dysfunction. Both Type 1 and 2 DM are characterized by hyperglycemia. Interestingly, it has been demonstrated that laboratory animals exposed to inorganic arsenic develop phenotypes consistent with DM, namely increased fasting blood glucose, impaired glucose tolerance and/or insulin resistance Supplementary Box 1 [19–25]. Additionally, prenatal exposure to inorganic arsenic predisposes offspring to the development of DM in laboratory studies [26,27]. In 11 of the 13 currently published rodent studies, the doses used to promote DM in laboratory animals were much higher than those typically associated with the environmental exposures Supplementary Box 1. It is important to note that these higher doses may be necessary as mice are less susceptible to arsenic-induced health effects than humans due to more efficient metabolism and clearance [28].

In vitro and in vivo laboratory studies support four major mechanisms by which inorganic arsenic and its methylated metabolites are able to influence DM-related processes (Box 1, Supplementary Box 1 & Supplementary Table 2). The first mechanism is the inhibition of insulin signaling, resulting in insulin resistance and in an impaired glucose uptake and utilization. The second mechanism is the induction of apoptosis of pancreatic β-cells, the primary site of insulin synthesis. The third mechanism is the inhibition of glucose-stimulated insulin secretion by pancreatic β-cells at noncytotoxic concentrations. The fourth mechanism is the induction of gluconeogenesis. While these processes are also involved in Type 2 DM [25], the ability of arsenic to dysregulate these processes supports that it is a diabetogen (Box 1).

Box 1. . Major mechanisms underlying arsenic-associated diabetes.

Mechanism 1: arsenic inhibits insulin-dependent glucose uptake

• Inhibition of GLUT4 localization to the plasma membrane.

• Inhibition of Akt signaling to mobilize GLUT4.

• Altered gene expression of the Akt pathway.

• Upregulation of antioxidant pathways leading to decreased secondary messengers.

• Altered development of the glucose transport system.

Mechanism 2: β-cell damage

• Induction of apoptosis due to production of ROS.

• Induction of chronic inflammation due to production of ROS.

Mechanism 3: β-cell dysfunction

• Failed glucose-stimulated insulin secretion due to increased ROS.

Mechanism 4: stimulation of gluconeogenesis

• Induced expression of phosphoenolpyruvate carboxykinase.

Akt: Protein kinase B; GLUT4: Glucose transporter type 4; ROS: Reactive oxygen species.

In addition to inorganic arsenic exposure, different methylated metabolites of arsenic influence processes involved in the development of DM to different degrees, suggesting that metabolism of arsenic is a determinant of DM development. Inorganic arsenic is metabolized by the enzyme arsenic (+3-oxidation state) methyltransferase (AS3MT) into different arsenicals. AS3MT first methylates and reduces inorganic arsenic to produce monomethyl arsenicals (MAsIII, MAsV). This process is repeated to produce the dimethyl arsenicals (DMAsIII, DMAsV). Importantly, it has been shown that the trivalent arsenicals are much more potent that the pentavalent counterparts as inhibitors of insulin signaling and insulin secretion by β-cells, suggesting that the trivalent species may be the main drivers of arsenic-associated DM [29]. However, it is important to note that limited evidence on pentavalent arsenical species hinders the ability to understand their roles in the induction of arsenic-associated DM. The use of biomarkers of cell injury and altered glucose signaling provides support that arsenic-induced injuries resulting in DM occur in humans. Taken together, the following studies of the effect of arsenic on glucose homeostasis pathways provide substantive evidence that arsenic is an etiologic factor in the development of DM.

Mechanisms of arsenic-associated DM: inhibition of insulin-dependent glucose uptake (insulin resistance)

Arsenic and its metabolites can alter glucose homeostasis through insulin resistance and impaired glucose uptake [28–30]. Arsenite (AsIII), and its trivalent metabolites, methylarsonous acid (MAsIII) and dimethylarsinous acid (DMAsIII), inhibit glucose transporter type 4 (GLUT4) recruitment to the plasma membrane despite insulin stimulation in vitro (Figure 1) [29]. Furthermore, in vitro studies have demonstrated that AsIII and MAsIII inhibit the insulin-dependent activation of protein Akt, a key signaling step required for GLUT4 translocation to the plasma membrane, by preventing its phosphorylation by PDK-1 (Figure 1) [23,31]. DMAsIII works downstream of Akt to prevent the recruitment of GLUT4, but the exact mechanism remains unclear (Figure 1). The process of Akt inhibition is compounded by the fact that arsenic can influence the expression of the PI3K transcription factor that is upstream of Akt, and that regulates other important steps in insulin signaling in vivo and in vitro [31,32]. Additionally, the expression of many of the genes in this pathway, including GLUT4, and Akt are decreased in response to arsenic exposure in vivo and in vitro [30,32–33]. In mice, arsenic exposure is associated with generation of reactive oxygen species (ROS) and upregulation of cellular antioxidant pathways including Nrf2 [34,35]. The constitutive upregulation of Nrf2 and the consequent increase in antioxidant enzyme expression could inhibit insulin-stimulated glucose uptake as the insulin pathway may also require ROS as signaling molecules [33].

Figure 1. . Arsenic (iAsIII) and its metabolites inhibit insulin-dependent glucose uptake by disrupting the insulin-activated signaling cascade.

Specifically, iAsIII and MAsIII have been shown to inhibit the phosphorylation of Akt by PDK, resulting in inhibition of GLUT4 translocation to plasma membrane; DMAsIII inhibits signaling downstream of PDK/Akt, but the exact target has not been identified.

Figure adapted from [23].

In addition to directly inhibiting the process of insulin-dependent glucose uptake, arsenic can interfere with adipogenesis, the differentiation of pre-adipocytes to adipocytes, as well as differentiation of myoblast into myotubes in skeletal muscle at doses between 0.5 and 5 µM [30,36–37] Supplementary Table 2. Adipocytes in white adipose tissue and myotubes in skeletal muscle tissue play critical roles in glucose utilization and in the maintenance of blood glucose levels. Arsenic inhibits adipocyte differentiation by altering the expression of key genes and transcription factors involved in adipogenesis, specifically PPAR-γ and CEBP-α [31,36,38–39]. This indirectly affects numerous other proteins involved in adipocyte differentiation including p21 and A-FABP [40]. Alterations of adipogenesis likely lead to decreased lipid-storage capacity and insulin resistance, as it has been shown that defects in adipogenesis can result in insulin impairment and Type 2 diabetes [36,41].

Arsenic may also influence the development of the insulin-responsive glucose transport system [30] by decreasing the phosphorylation of mechanistic target of mTOR and p70, key regulators in this process [42]. Notably, in humans, urinary total arsenic (U-tAs; a biomarker measure of arsenic exposure) is inversely associated with insulin sensitivity levels [43]. However, it has also been shown that arsenic exposure is inversely related to fasting plasma insulin and to the measure of insulin resistance (HOMA-IR) [44]. Similarly, mice that were exposed to arsenic and fed a high-fat diet displayed impaired glucose tolerance and a dose-dependent decrease in HOMA-IR, likely due to lower fasting insulin and higher fasting blood glucose than in mice fed a low-fat diet [25]. Insulin response to glucose challenge was also impaired in these mice. Thus, both impaired insulin sensitivity and impaired insulin secretion in response to high blood glucose may underlie the diabetogenic effects of arsenic exposure.

Mechanisms of arsenic-associated DM: β-cell damage

There is a growing body of evidence that arsenic can damage pancreatic β-cells. It has been shown that arsenic metabolites accumulate in the pancreata of mice exposed to arsenic [19,24,28]. Arsenic accumulation induces apoptosis of pancreatic β-cells in vitro, likely through the production of ROS, and induction of PARP and Caspase-3, as well as through other mechanisms [37,45–47]. In addition to causing cell death, this accumulation of ROS has been shown to result in chronic inflammation. Such chronic inflammation can result in pancreatitis in vivo, which can lead to the development of DM [47]. Moreover, excessive levels of ROS result in upregulation of antioxidant pathways. It has been proposed that a persistent activation of Nrf2, a key transcription factor in the antioxidant system, reduces glucose-triggered ROS-mediated signaling and thus inhibits glucose-stimulated insulin secretion in vitro [48]. Exposure to AsIII and MAsIII have also been shown to lead to inhibition of TXNRD1 in vivo, a key member of the cellular anti-oxidant defense, thus resulting in apoptosis [19,49].

Mechanisms of arsenic-associated DM: β-cell dysfunction

Arsenic and its methylated metabolites can inhibit the gene transcription of insulin and/or secretion in the pancreas. Some studies suggested that AsIII may decrease the levels of insulin mRNA produced by pancreatic β-cells. Decreased insulin mRNA production would produce the same effect as destruction of these cells. Still, there have been conflicting data on the subject, likely due to dose differences employed across studies [50,51]. The destruction of cells by ROS-mediated mechanisms, as well as other cellular targets of AsIII, directly impacts the ability of the pancreases to produce insulin. Recent in vitro research has demonstrated that pancreatic islets exposed to low arsenic concentrations continue to produce insulin, but fail to secrete it. The result is a blunted glucose-stimulated insulin secretion [50,51]. Consistent with these findings are results of a recent population study that showed arsenic exposure to be associated with a decrease in the insulin secretion index, HOMA2%B [52]. An additional in vitro study using a β-cell line suggested that this is likely due to the ability of AsIII to inhibit calcium oscillation, which is necessary for glucose-stimulated insulin secretion [53]. Interestingly, both MAsIII and DMAsIII are more potent inhibitors of glucose-stimulated insulin secretion by pancreatic islets than AsIII, though the mechanisms for this are unclear [51]. It has also been shown in vitro that the cellular adaptation to chronic oxidative stress caused by AsIII exposure lowers ROS production and impairs glucose-stimulated insulin secretion, which may require low-level ROS as one of the regulatory mechanisms [54].

Mechanisms of arsenic-associated DM: stimulation of gluconeogenesis

In addition to inhibiting insulin secretion, gene expression changes induced by arsenic in vivo can also induce hepatic gluconeogenesis. Hepatic gluconeogenesis is the pathway for synthesis of glucose from noncarbohydrate sources during fasting. Arsenic induces increased expression of phosphoenolpyruvate carboxykinase, a rate-limiting enzyme in gluconeogenesis resulting in fasting hyperglycemia [55,56]. In summary, these data detail the manner by which arsenic is able to regulate glucose hemostasis resulting in hyperglycemia and DM.

Genetic & epigenetic underpinnings for arsenic-associated DM

Multiple studies have found that differences in the metabolism of inorganic arsenic and the distribution and excretion of arsenic metabolites are associated with likelihood of diabetes development [12,57–60]. As detailed previously, inorganic arsenic is methylated primarily by AS3MT to form the monomethylated and dimethylated arsenic metabolites, MAs and DMAs. Studies of arsenic-exposed populations have found that higher inorganic arsenic and MAs in urothelial cells and higher DMAs in urine are more strongly associated with DM than drinking water measures of arsenic or U-tAs [44,59,61–62]. These differences in arsenic metabolite profiles represent a potential indicator of the risk of developing DM and could be mediated by differences in genotypes, specifically single nucleotide polymorphisms (SNPs), of critical genes such as AS3MT [63], CAPN-10 – a calcium-dependent protease that plays a key role in exocytosis of insulin-containing vesicles in β-cells [64], GSTO1 – an enzyme that contributes to the reduction of arsenic from the pentavalent to trivalent form [57] and NOTCH2 – a member of a signaling cascade involved in cell differentiation [57–58,64–65]. Interestingly, the majority of SNPs associated with differences in arsenic metabolism occur in non-coding regions of the genes, suggesting a potential role for regulatory factors or splice variants.

In addition to the role of polymorphic variants of key genes, the processes of glucose homeostasis, as well as arsenic metabolism, may also be altered by epigenetic regulation. In support of this, exposure to arsenic induces alterations to the epigenome, including changes in CpG methylation and miRNA expression [18,66–67]. The epigenome constitutes potentially heritable changes that influence gene expression but are not contained within the DNA sequence [18]. These changes result in altered gene expression and subsequent changes in protein levels. In contrast to SNPs or other DNA sequence variations, epigenetic alterations can influence gene and protein expression, but do not alter amino acid sequence. Of the epigenetic modifications studied in relationship to arsenic, DNA methylation has been the best characterized in human populations [18].

Results of epigenetic studies of arsenic exposure suggest that the epigenome could play a key role in the development of arsenic-associated DM. Four studies of arsenic-exposed non-diabetic individuals have shown that arsenic disrupts gene expression, miRNA expression and DNA methylation profiles in a manner consistent with DM [12–16]. Andrew et al. assessed gene expression in the blood of individuals exposed to arsenic in New Hampshire, USA, where the highest level of exposure was 74.4 µg/l of inorganic arsenic in drinking water [14]. Importantly, many of the genes identified as dysregulated in association with arsenic were involved in Type 1 DM signaling [14]. These genes included members of the HLA cluster that account for 40–50% of familial aggregation of Type 1 DM, as well as other genes associated with hereditary Type 1 DM. These genes are responsible for regulation of the immune response, a primary mechanism of β-cell destruction in Type 1 DM. In addition to changes in mRNA expression, it has been shown that arsenic is associated with altered miRNA expression. Rager et al. analyzed miRNA profiles from the Biomakers of Exposure to ARsenic pregnancy cohort in Gòmez Palacio, Mexico, where arsenic ranged from 0 to 240 µg/l. It was found that miRNAs with altered expression in response to prenatal arsenic exposure were enriched for roles in DM [16]. Specifically, arsenic-associated changes in miR-107 and miR-20b were observed, both of which have previously been associated with DM. Furthermore, ten downstream targets of these miRNAs were identified as differentially expressed in response to arsenic exposure. In addition to these data in a different study, miR-25, a regular of glycemic control, was altered in rats exposed to arsenic at 0.1–100 mg/l [17]. These data highlight that arsenic influences the altered expression of both mRNA and miRNAs associated with DM signaling.

In addition to expression of miRNAs and mRNAs, arsenic has been shown to perturb DNA (CpG) methylation. Smeester et al. assessed blood-based DNA methylation patterns associated with arsenicosis in adults from Zimapán, Mexico. The exposure of these individuals was characterized by urinary As levels (27.5–90.8 µg/l) and arsenicosis by the presence of skin lesions [13]. Using a genome-wide assay with a methyl-binding protein bound to methylated DNA followed by array hybridization, a total of 183 genes displayed methylation levels that were significantly associated with U-tAs levels. Among these genes are those that are involved in DM including CENPE, which was previously identified through a genome-wide association study as being associated with risk of DM development [68]. Subsequent analysis of these data performed in Bailey et al. assessed DNA methylation changes as they related to arsenic metabolism (i.e., proportions of the methylated metabolites in urine) [12]. Specifically, enrichment for DM-associated genes was observed in methylated probes associated with the dimethylated fraction of arsenic (%DMAs) in urine. Notably, these genes were associated with pancreatic β-cell function and apoptosis, as well as with insulin signaling and insulin resistance. Genes regulating insulin production (PDX1, INS), and those involved in glucose-stimulated insulin secretion (VAMP2) were among those that were differentially methylated in relationship to arsenic exposure. A limitation of the studies of the Zimapán population was the lack of gene expression measures, thus it is unclear what the specific impact of these CpG alterations were on transcript level. Lastly, DNA methylation changes that resulted in functional gene expression changes in cord blood were assessed in the Biomakers of Exposure to ARsenic cohort [15]. Among the set of genes that was identified with altered methylation in relation to arsenic exposure was the imprinted gene, KCNQ1. KCNQ1 is a key protein in the glucose-stimulated cascade that leads to insulin secretion by β-cells [69]. Interestingly, SNPs in this gene have been associated with increased odds of development of Type 2 DM, gestational diabetes and glucose tolerance [70–72]. Furthermore, previous studies have linked hypomethylation of this gene to increased odds of DM development [70,72]. Taken together, these studies suggest that changes to the epigenome, specifically CpG methylation and miRNA expression, are tied to altered gene expression of DM-associated genes. The genes identified in these studies are involved in multiple DM processes, including destruction of β-cells, glucose-stimulated insulin secretion and insulin secretion by β-cells. In future research, the precise role of the epigenome in the development of arsenic-associated DM should be further explored.

In addition to determining associations of arsenic exposure and DM, population-based studies have demonstrated sex-dependent differences in the development of arsenic-associated disease [73]. In particular, some studies have found that women have higher standardized mortality ratios and hazard ratios for DM-related death and disease development in association with arsenic [74–76]. These findings have been substantiated by in vivo animal studies where female rodents are more susceptible than male rodents to the diabetogenic effects of arsenic exposure [20,56]. While these data suggest that sex influences the development of arsenic-associated DM, further research is needed to fully elucidate the underpinnings of this relationship. Interestingly, this could further support the role of epigenetic alterations as a potential driver of arsenic-associated DM as multiple studies have shown sex-dependent DNA methylation patterns [77–79].

The summarized studies support the role for genetic variation and epigenetic modification as susceptibility factors for the development of arsenic-associated DM. Moving forward, future studies on the role of the epigenome should include assessments of functional changes in gene and protein expression to elucidate causal mechanisms important to the development of arsenic-associated DM. As some epigenetic marks are stable over time [80], they could provide evidence for the temporality of arsenic exposure and the development of DM in later life.

Conclusion & future perspective

There is strong evidence that arsenic is a diabetogen. This evidence comes from epidemiologic studies with a mechanistic basis supported by both in vitro and in vivo studies. The data demonstrate that arsenic triggers the development of DM phenotypes in laboratory animals, and impairs key pathways involved in glucose homeostasis. Some observations from in vivo studies, such as the sex-dependent nature of the relationship between arsenic and DM, and in vitro studies, such as dysfunction of β-cells, are substantiated by human population-based research. Importantly, evidence suggests a potential role for the epigenome (e.g., DNA methylation, miRNAs) as a means by which arsenic could influence the development of DM. Research that focuses on molecular measurements, such as DNA methylation, gene expression and protein expression, could provide valuable insight into how arsenic exposure affects glucose homeostasis and which of the proposed mechanisms is most likely driving the association observed in human populations. Moving forward, we foresee environmental DM research will incorporate biomarker-based DM indicators, such as measures of insulin resistance and β-cell function, to be used in conjunction with arsenic exposure biomarkers in human studies. Furthermore, researchers should focus on three primary areas. First, there is a need to develop in vivo and in vitro models to better assess low-dose effects on glucose homeostasis. Second, it is important to characterize key developmental windows of susceptibility related to DM phenotypes and assess susceptibility factors such as genetic variation and sex. Lastly, the use of standardized methods for the assessment of arsenic exposure and metabolism and for DM diagnosis should be implemented across studies to ensure comparability. These directions of study will further elucidate the mechanistic basis underlying arsenic-associated DM.

Executive summary.

Background

• Increased incidence of diabetes has been associated with arsenic exposure in numerous epidemiological studies.

• Arsenic has been shown to induce DM phenotypes in vivo and disrupt key processes responsible for glucose homeostasis in vitro.

Mechanisms of arsenic associated-diabetes mellitus

• There are four major mechanisms by which arsenic is hypothesized to induce DM.

• First, arsenic has been shown to inhibit insulin-dependent glucose uptake particularly through inhibition the glucose transporter, GLUT4.

• Second, arsenic has been shown to induce pancreatic β-cell damage as it accumulates in the pancreata of exposed laboratory animals and stimulates the production of ROS.

• Third, arsenic results in pancreatic β-cell dysfunction, specifically chronic arsenic exposure leads to inhibition of glucose-stimulated insulin secretion.

• Lastly, arsenic is thought to stimulate liver gluconeogenesis, resulting in increased glucose entering the blood stream.

Genetic & epigenetic underpinnings for arsenic-associated diabetes mellitus

• Variants in the genes AS3MT, CAPN-10, GSTO1 and NOTCH2 have be linked to increased likelihood of developing DM in the presence of inorganic arsenic.

• Epigenetic mechanisms, specifically DNA methylation and miRNA expression, likely play a key role in arsenic-associated DM.

• In human populations exposed to arsenic, genes associated with Type 1 DM display altered expression at the mRNA level.

• Arsenic has been shown to alter miRNAs associated with DM and glycemic regulation in humans and in vivo.

• Three studies have identified arsenic-associated changes in DNA methylation of DM-related genes. Two of these studies were in adult populations and one was in infants exposed to arsenic prenatally.

Conclusions & future outlook

• In vivo and in vitro studies suggest that arsenic is a diabetogen, and data from epidemiological studies of human populations support this.

• The epigenome remains understudied in the context of arsenic-associated DM and could provide valuable insight into mechanism underlying the ability of arsenic to affect glucose homeostasis in humans.

• In vivo and in vitro models that can better assess low-dose effects of glucose homeostasis should be developed.