|Home | About | Journals | Submit | Contact Us | Français|
As the tide of chemicals born of the Industrial Age has arisen to engulf our environment, a drastic change has come about in the nature of the most serious public health problems.
Rachel Carson, Silent Spring, 1962
Worldwide rates of diabetes and other metabolic diseases have exploded over the last several decades. Globally, more than 170 million individuals currently suffer from diabetes, and this number is projected to reach a staggering 366 million by 2030 (1). This scourge results in significant individual morbidity and mortality while contributing to the economic fragility of healthcare systems across the globe. In the U.S. alone, annual costs associated with diabetes are estimated to be $174 billion (2). As such, every effort must be made to understand the factors underlying this emerging metabolic disaster in order to mitigate its deleterious impact on the individual and society. Recently, an expanding body of scientific evidence has begun to link exposure to synthetic chemicals with a wide variety of diseases, including reproductive tract disorders and neurobehavioral diseases. The present work discusses epidemiological links between chemical exposure and disorders of glucose homeostasis, experimental data demonstrating chemical-induced changes in insulin action, and challenges facing the field of metabolic disruption as well as approaches for addressing those challenges.
Originally articulated in the early 1990s, the environmental endocrine disruptor theory proposes that some exogenous chemicals interfere with endogenous hormonal axes (3). The recognition of this potential mechanism of action was a paradigm shift in toxicology that had previously focused on a chemical’s capacity to induce acute toxicity or to cause cancer via mutagenesis. The Environmental Protection Agency (EPA) defines an endocrine disrupting chemical (EDC) as “an exogenous agent that interferes with the production, release, transport, metabolism, binding, action, or elimination of natural hormones in the body responsible for the maintenance of homeostasis, reproduction, development, and/or behavior” (4). Putative EDCs include structurally diverse chemicals including organic pollutants, heavy metals, pharmaceuticals, and phytochemicals, with humans exposed through agricultural goods and consumer products, as well as water and air contaminated with industrial waste (Fig. 1). Early studies of EDCs focused on identifying chemicals with the capacity to modulate sex steroid and thyroid hormone signaling; however, recent work suggests that some chemicals may disturb signaling pathways critical for energy homeostasis (5). Despite the potential importance of EDCs in the pathogenesis of metabolic diseases, the contribution of synthetic chemical exposure to the diabetes epidemic remains largely unrecognized and underappreciated even though U.S. diabetes rates have increased in concordance with the national production of synthetic organic chemicals (Fig. 2). While such correlations are crude, emerging data supports a biologically plausible causative link between diabetes and chemical exposure. Here, we present data suggesting a role for some synthetic chemicals in the pathogenesis of diabetes that merits comprehensive efforts to address the contribution of environmental pollutants to this burgeoning metabolic catastrophe.
An unequivocal contributor to the diabetes epidemic is the metabolic stress induced by rising rates of obesity. Increased adiposity is closely linked to the development of insulin resistance, an important predisposing factor in the development of type 2 diabetes. Over the last several decades, obesity rates have exploded, with more than a third of the adult U.S. population now obese (6). The society-wide accumulation of body fat is undoubtedly a consequence of a widening gap between caloric intake and caloric expenditure resulting from myriad social forces; however, the magnitude and rapidity with which obesity rates have increased raise concerns about other pathogenic factors. In 2002, Baillie-Hamilton (7) proposed a link between the post–World War II increase in synthetic chemical production and the obesity epidemic. This correlation, coupled with experimental evidence demonstrating that certain environmental pollutants induce adipogenesis and weight gain in experimental models, led to the environmental obesogen hypothesis that posits a causative role for synthetic chemicals in the pathogenesis of obesity (rev. in 8).
While environmental obesogens have rightfully received much discussion, it is important to recognize that obesity per se may not lead to abnormalities in glucose homeostasis. An important distinction in obesity research is the differentiation between metabolically deleterious obesity and the “fit fat” (9). Thus, while increased fat mass may contribute to the development of diabetes, obesity is not a necessary or sufficient condition. Insulin resistance can arise independent of obesity, and the onset of frank diabetes necessitates a deficit in β-cell insulin production, as either the primary defect or the failure to compensate for diminished insulin sensitivity. Therefore, the search for pollution-induced diabetes should include a specific focus on compounds with the capacity to induce insulin resistance and/or impair β-cell function.
Data linking diabetes to environmental pollutants have come from a number of epidemiological studies performed in a variety of experimental contexts (Table 1). Environmental disasters such as the chemical plant explosion in Seveso, Italy, have suggested a link between dioxin exposure and diabetes (10), while rice oil contamination in Yucheng, China, has implicated polychlorinated biphenyl ethers (PCBs) and furans (11). Exposure of military personnel to dioxins during the Vietnam War has been associated with a higher prevalence of diabetes and a reduced latency to disease development (12). Several studies of occupational exposure have suggested links between diabetes and organochlorine pesticides (13) or dioxins (14). Recreational contact via consumption of sport fish from the Great Lakes in the U.S. tied diabetes incidence with p,p’-diphenyldichloroethene (DDE), the principle metabolite of the insecticide p,p’-dichlorodiphenyltrichloroethane (DDT) (15). A variety of international studies demonstrated diabetogenic links to organochlorine pollutants (16) and heavy metals (17), with some studies suggesting a specific defect in insulin secretion but not in overall glucose tolerance (18). In addition to diabetes, epidemiological studies have associated various pollutants with other measures of disturbed glucose homeostasis, including prediabetes (impaired fasting glucose and/or impaired glucose tolerance) (16), the metabolic syndrome (19), and insulin resistance (20).
Many of the above studies focused on specific populations (i.e., those exposed occupationally, recreationally, or through specific industrial accidents) that may not reflect the risk posed to the general population; however, a series of recent investigations have examined the connection between various pollutants and measures of glucose homeostasis using data from the National Health and Nutrition Examination Survey (NHANES), which provides a representative sampling of the U.S. population (21). NHANES-based studies have shown associations between phthalates and various persistent organic pollutants (POPs) with insulin resistance, the metabolic syndrome, and diabetes (20,22). In addition, urinary levels of the ubiquitous plasticizer, bisphenol-A (BPA), have been detected in >95% of the NHANES cohort (23) and positively correlate with diabetes prevalence and other metabolic derangements (24). Thus, there is intriguing evidence suggesting possible connections between pollutants and the development of diabetes.
There are, however, caveats that must be considered in interpreting these studies. One significant challenge is the common use of cross-sectional design to correlate disease prevalence with current EDC levels. Such analyses are particularly problematic for chemicals that metabolize more rapidly and exhibit fewer propensities to bioaccumulate (e.g., BPA and phthalates) because their current levels may differ from concentrations during disease development. Additionally, issues related to coexposures to confounding compounds, selection of control populations, and variability in statistical analyses complicate data interpretation and extrapolation to the general population. Furthermore, there is heterogeneity in the definition of diabetes and insulin resistance used in these studies. Collectively, these challenges underscore the need for expanded longitudinal studies that can follow chemical exposures throughout disease development in order to better relate specific chemicals to the pathogenesis of diabetes.
The shortcomings of epidemiological investigations can be overcome by studying suspected diabetogenic chemicals using animal models. A number of chemicals have been shown to elicit biological effects that alter glucose homeostasis (Table 2). For instance, acute exposure of male mice to BPA was found to reduce the rise in plasma glucose during an intraperitoneal glucose tolerance test; however, sustained exposure (more similar to human exposure) resulted in hyperinsulinemia, a worsening of glucose tolerance, and a concomitant reduction in insulin sensitivity (25). Interestingly, the impairment in insulin action occurred despite a demonstrated increase in β-cell insulin content after both in vivo and in vitro BPA exposure (26). One explanation for these findings is that BPA operates through multiple mechanisms that independently increase insulin synthesis/secretion while simultaneously inducing peripheral insulin resistance. Alternatively, higher insulin levels induced by BPA may result in a compensatory insulin resistance to limit hypoglycemia. Regardless of the process, the overall effects of chronic BPA exposure on glucose homeostasis suggest that it may be a diabetogenic factor (27).
Other pollutants also disrupt glucose homeostasis in experimental models. Exposure of rats to the flame retardant polybrominated diphenyl ether significantly increased lipolysis while reducing insulin-stimulated glucose uptake (28). Diethylhexyl phthalate, a common plasticizer, reduced insulin levels and raised serum glucose levels in exposed rats (29), while mice treated with tributyl tin (TBT), a fungicide and antifouling agent, demonstrated hepatic steatosis and hyperinsulinemia (30). Recently, rats fed fish oil naturally contaminated with a variety of POPs demonstrated impaired glucose homeostasis, with several chemicals in the contaminated fish oil found to suppress insulin-stimulated glucose uptake in 3T3-L1 adipocytes (31).
These results are similar to findings that 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) treatment of primary murine adipose tissue impaired insulin-stimulated glucose uptake, likely by reducing glucose transporter 4 transcript levels (32). In a separate model, mice exposed to TCDD had reduced glucokinase gene expression (33), predicting a rise in blood glucose levels analogous to that seen in maturity-onset diabetes of the young type 2. Others have suggested that the diabetogenic effects of TCDD are mediated through an antagonism of peroxisome proliferator–activated receptor-γ (PPARγ) action (34) or through upregulation of the inflammatory adipokine tumor necrosis factor-α (TNF-α) in adipocytes (35). While these data are consistent with epidemiological observations linking TCDD exposure to diabetes, other studies have shown that TCDD has hypoglycemic effects. In a rat model of diabetes incorporating high-fat diet coupled with streptozotocin treatment, TCDD treatment reduced plasma glucose levels (36). However, this study may reflect an alternative metabolic disruption of quasi-starvation mediated through TCDD suppression of gluconeogenesis via inhibition of PEPCK (37). Furthermore, the hypoglycemic effects of TCDD occurred at concentrations within an order of magnitude of the known lethal dose for rat. The apparent incongruence between hypoglycemic and hyperglycemic observations likely reflects dose-dependent effects. Such findings underscore the need for mechanistic studies over wide concentration ranges that reflect both variability in human exposure and the potential for different mechanisms to predominate at different concentrations.
Historically, EDC research has focused on the ability of exogenous chemicals to modulate the activity of classic nuclear hormone receptors, including those for estrogens, androgens, and thyroid hormone. Several of these pathways appear to be critically important for energy regulation in general and glucose homeostasis in particular. For example, knockout models of aromatase and the estrogen receptor-α demonstrate the capacity of estrogens to augment glucose tolerance and insulin sensitivity (38,39). However, the effects of estrogen on insulin action may be context-dependent, as conditions associated with estrogen levels that are both high (e.g., pregnancy) and low (e.g., menopause) correlate with insulin resistance. BPA is known to have estrogenic properties, and as mentioned, prolonged treatment of male mice with this EDC induces changes consistent with a diabetic phenotype (25). Furthermore, the augmentation in β-cell insulin content after BPA exposure appears to be a direct result of its estrogenic properties, as the effect was not observed in estrogen receptor-α–knockout animals (26). Because estrogens can have divergent effects on insulin action, estrogenic EDCs may modulate insulin action differently depending on the background hormonal milieu. Thus, the experimental effects may differ between males and females as well as among females at various stages of their reproductive lives (i.e., prepubertal, postmenopausal, or reproductive age).
Androgens also appear to modulate insulin sensitivity. For example, emerging data suggests that low androgen levels in men correlate with insulin resistance. In the TIMES2 trial, testosterone treatment of hypogonadal men with diabetes or the metabolic syndrome improved insulin sensitivity as assessed by homeostasis model assessment of insulin resistance (HOMA-IR) (40). In contrast, exposure to androgens can also adversely affect glucose tolerance. Rhesus monkeys prenatally exposed to androgens show evidence of insulin resistance, with the females having features consistent with the polycystic ovarian syndrome (PCOS) phenotype (41). In humans, insulin resistance is an important clinical feature of PCOS. Interestingly, recent data suggests that women with PCOS have higher levels of BPA than control subjects, and among these PCOS patients, BPA levels correlated with measures of insulin resistance (42). Various synthetic chemicals have the capacity to function as both androgen agonists and antagonists (43), suggesting their capacity to disrupt glucose homeostasis. Importantly, these data also emphasize the potential importance of the timing, context, and relative balance of EDCs on the overall impact of chemical exposure on diabetes risk.
Given the central role of thyroid hormone in energy metabolism, disruption of normal thyroid hormone action may facilitate the development of a diabetic phenotype. Many chemicals can disrupt the thyroid hormone axis (44), and levels of several thyroid disruptors have been correlated with diabetes in epidemiological studies, including PCBs (45). Likewise, glucocorticoids are known modulators of energy metabolism, and recent data suggest that some EDCs may have the capacity to stimulate signaling through the glucocorticoid receptor (46) or by altering glucocorticoid synthesis or activation (47,48). EDCs with glucocorticoid-like activity would be predicted to diminish insulin sensitivity and foster a diabetic phenotype.
Other ligand-activated nuclear hormone receptors are important for energy regulation and have been implicated as EDC targets. Of particular interest are EDCs activating the PPARs. For example, TBT promotes adipogenesis by stimulating PPARγ and its obligate heterodimeric partner retinoid X receptor (RXR) in mouse models (49) and human mesenchymal stem cell cultures (50). Conversely, TCDD inhibits adipogenesis through a suppression of PPARγ (51). The proadipogenic effects of TBT and other EDCs serve as the basis for the environmental obesogen hypothesis. Nevertheless, while PPARγ promotes fat accumulation, its activation also increases insulin sensitivity; this is the rationale for using thiazolidinediones to treat diabetes. Despite this, TBT may impair insulin sensitivity (30); however, this may reflect its promiscuous activation of heterodimeric partners of RXR other than PPARγ. EDC-mediated effects on nuclear hormone signaling are emerging as important mechanisms of metabolic disruption; however, work remains to clarify whether these compounds alter signaling directly at the ligand binding site or whether indirect mechanisms such as coactivator/corepressor recruitment, ligand activation, allosteric effects, targeted receptor degradation, or others are the principle modes of action.
In addition to the traditional hormone receptors, the superfamily of ligand-activated nuclear hormone receptors includes several members that function primarily in the sensing and detoxification of foreign compounds, i.e., xenobiotics. These include the aryl hydrocarbon receptor (AhR), the pregnane X receptor, and the constitutive androstane receptor. In addition to their role in the induction of drug metabolizing enzymes, these receptors have modulating effects on lipid and glucose metabolism through their interaction with a wide array of other nuclear receptors (e.g., thyroid hormone receptor, glucocorticoid receptor, PPARα, PPARγ) and transcription factors (e.g., CREB, FOXO1, PGC1α) involved in energy regulation; moreover, the xenobiotic receptors appear to influence inflammatory responses (52). Interestingly, AhR was originally identified as the receptor for dioxin, one of the chemicals most frequently associated with diabetes in epidemiological studies. It is intriguing to speculate that xenobiotic receptors evolved in part to adjust metabolic pathways to environmental stressors, and that the proliferation of anthropogenic chemicals in the environment has overwhelmed these adaptive processes, thereby contributing to the onset of metabolic diseases.
In line with the “developmental origins of adult disease” hypothesis, a recent emphasis in EDC research has been focused on the effects of in utero and early postnatal chemical exposure on the genesis of adult diseases through modulation of the epigenome. Exposure to a variety of pollutants appears to modify the epigenome (53), and concerning evidence demonstrates that chemical-induced epigenetic changes can be heritable. In a rat model, exposure of pregnant dams to the fungicide vinclozolin led to transgenerational epigenetic modifications into at least the F4 generation (54). Intriguingly, there is now data demonstrating the epigenetic regulation of various genes influencing metabolic diseases, including diabetes (55). While links between EDC exposure and epigenetic alterations of genes controlling energy metabolism have yet to be described, current evidence supports the contention that exposure to EDCs may influence the metabolic state of an individual, with the potential for these effects to be transmitted to subsequent generations.
EDC effects on other molecular mechanisms implicated in the development of diabetes, e.g., inflammation and oxidative stress, have only recently been considered. For example, PCB-77 has been shown to promote expression of the proinflammatory adipokines interleukin-6 (IL-6) and TNF-α, leading to impaired insulin signaling in endothelial cells (56). BPA treatment of human adipose tissue explants also augments secretion of IL-6 and TNF-α while simultaneously inhibiting the release of the insulin-sensitizing adipokine adiponectin (57). Interestingly, the detoxification of exogenous chemicals by the cytochrome P450 enzymes generates oxidative stress, which may promote hepatic insulin resistance due to the liver’s dual role in energy and drug metabolism (58). Other diabetogenic mechanisms such as induction of endoplasmic reticulum stress, implicated in arsenic-induced β-cell apoptosis (59), are intriguing but remain poorly studied. With the plethora of structurally diverse compounds present in the environment, these and additional mechanisms may be relevant in the disruption of energy regulation. Characterizing the relevant mechanisms is critical for identifying potential pharmaceutical targets to treat environmentally induced diabetes.
As alluded to above, there are a number of challenges limiting our understanding of the impact of synthetic chemicals on metabolic diseases that relate to the chemicals themselves, the exposed individuals, and the experimental approach used to study EDC effects on glucose homeostasis (Table 3). The tens of thousands of unique chemicals released into the environment create an enormous analytical challenge in quantifying human exposure while the physical properties of some compounds contribute to their bioaccumulation and persistence in human tissues long after the exposure has terminated. This contributes to the near ubiquity of certain EDCs in the U.S. population (e.g., hexachlorobenzene and DDE) (23) and raises important questions about the threshold of exposure necessary to elicit a disease phenotype. The experimental challenge is further complicated by the lack of clear structure-function relationships that preclude in silico prediction of adverse health effects, thereby necessitating the use of bioassays to characterize the physiological effects of chemical exposure.
Inter-individual variation in gene-environment interactions may also modify the deleterious effects of synthetic chemicals. For example, the adverse consequences of occupational exposure to pesticides among Costa Rican banana farmers was found to be influenced by whether they inherited “favorable” or “unfavorable” metabolizing genes (60). Other predisposing factors, such as obesity or a family history of diabetes, may also accentuate the diabetogenic effects of some chemicals (61), while high-fat diets may augment exposure to lipophilic EDCs. As discussed, EDCs that modulate sex steroid action may have divergent metabolic effects depending on the background hormonal milieu leading to sexually dimorphic effects that are also influenced by changes over the life span. Finally, evidence of in utero programming and transgenerational effects suggest that an individual’s disease-related exposure may have taken place before birth or even in a prior generation.
Lastly, experimental design may influence the observed biological effect or fail to accurately recapitulate real-world scenarios. Particularly vexing are mixtures of compounds that may exert additive, antagonistic, or even synergistic biological effects. Consequently, the ultimate metabolic phenotype may differ considerably from studies of single chemicals in isolation. Furthermore, nonmonotonic dose-response relationships are seen with some chemicals (e.g., BPA  and polybrominated diphenyl ethers ), thus mandating studies across wide concentration ranges that also account for effects at extremely low doses. Finally, animal models and humans may have divergent responses to EDCs (63), and the phytochemical content of animal feeds modulates EDCs effects (64). Thus, careful selection of experimental models and appropriate controls are critical for understanding the metabolic effects of synthetic chemicals.
One of the profound tragedies of the diabetes epidemic is its disproportionate effect on minority groups and the economically disadvantaged. National survey data from 2007–2009 show that 11.8% of Hispanics and 12.6% of non-Hispanic blacks self-identified as having diabetes compared with only 7.1% of non-Hispanic whites (65). Furthermore, rates of diabetes are inversely related to household income and education level (66). This heightened risk may reflect disproportionate exposure to chemical pollutants among these groups (67). Some studies have specifically linked chemical exposure with diabetes among high-risk groups, including Mexican Americans (68) and Native Americans (69), while others have found stronger associations between POPs and diabetes among Hispanics (22). This strengthened association may reflect coordinate exposure to other diabetes-promoting chemicals, increased susceptibility to metabolic disruption (e.g., in 60), or an interaction between diabetogenic chemicals and other predisposing risk factors, e.g., diet or limited access to healthcare. The greater synthetic chemical exposure of poor and minority groups is likely a consequence of many factors; however, one important contributor is the historical construction of chemical production facilities and toxic waste sites in poor communities (70). The causality of these intertwined relationships is difficult to dissect; nevertheless, the synergy of coexisting poverty, poor education, and pollution likely contributes to the pathogenesis of metabolic diseases.
Mitigation of the detrimental effects of pollution on metabolic health will require the development and implementation of comprehensive strategies, including expanded research programs, improved regulation and public policy, and education efforts (Fig. 3). The EPA has recently begun analyzing synthetic chemicals for endocrine-disrupting effects through the Endocrine Disruptor Screening Program. This initiative should be expanded to include analyses of metabolic disruption in order to specifically identify diabetogenic chemicals with significant emphasis placed on characterizing the molecular mechanisms in order to better identify health-threatening chemicals, to anticipate additive/synergistic effects of mixtures, and to determine potential pharmaceutical targets. Susceptibility across the life span should be assessed with special emphasis on critical developmental windows and transgenerational effects.
Risk analysis can be enhanced by identifying groups with exposures to high-risk chemicals and achieving a better understanding of factors predisposing to disruptions in energy homeostasis, including genetic polymorphisms linked to detoxification and metabolic pathways. Exposure analysis must be expanded beyond urine and serum to include lipid-rich organs that bioaccumulate POPs (e.g., brain and adipose) as well as tissues relevant to in utero and early postnatal exposure (e.g., human breast milk, cord blood, and placental tissue). Beyond direct measurement of EDCs, development of clinical biomarkers will facilitate identification of chemical-exposed individuals who can then be monitored prospectively for the development of diabetes, enhancing efforts to establish causality.
The new toxicology paradigm of endocrine and metabolic disruption mandates a transformation in regulatory policy to limit the production and use of chemicals that threaten metabolic health. Specific attention should be paid to banning chemicals with long-lasting efforts due to either their environmental persistence or ability to induce transgenerational effects. When doubts persist about a chemical’s metabolic impact, the “precautionary principle” should be adopted and its use restricted. Whenever possible, decontamination of environments and individuals exposed to metabolic disruptors should be pursued to limit ongoing exposure. For individuals with exposures that cannot be cleared, drugs should be developed to modulate the specific pathways responsible for pollution-induced diabetes. Finally, scientists and clinicians must become advocates in educating lawmakers and the public about the threat of metabolic disruptors as well as the means to limit their impact through sound government policy and smart consumer choices.
In 1962, Rachel Carson warned of the health threat posed by environmental pollution (71). Nearly 50 years later, evidence suggests that human exposure to synthetic chemicals may be contributing to the burgeoning diabetes epidemic. While the revolution in synthetic chemistry has facilitated vast improvements in our quality of life, these benefits increasingly appear to have come with a hidden cost. This paradox of progress now mandates a reassessment of how our consumption habits negatively impact our metabolic health in order to devise effective strategies to limit the significant individual and societal toll of diabetes.
This work was supported in part by National Institutes of Health/National Institute of Environmental Health Sciences Grant K08ES019176 (to R.M.S.).
No potential conflicts of interest relevant to this article were reported.
B.A.N. and R.M.S. wrote, reviewed, and edited the manuscript.
Due to reference constraints, the authors regret that some literature could not be cited.
The authors thank Rachel P. Delaney of the U.S. Environmental Protection Agency Andrew W. Breidenbach Environmental Research Center Library in Cincinnati, Ohio, for research assistance; Yuxi Lin of the University of Chicago for editorial comments; and Paul A. Volden of the University of Chicago Committee on Molecular Metabolism and Nutrition for assistance with illustrations.