Assessing human health risk to endocrine disrupting chemicals: a focus on prenatal exposures and oxidative stress

Abstract

Understanding the health risk posed by endocrine disrupting chemicals (EDCs) is a challenge that is receiving intense attention. The following study criteria should be considered to facilitate risk assessment for exposure to EDCs: 1) characterization of target health outcomes and their mediators, 2) study of exposures in the context of critical periods of development, 3) accurate estimates of human exposures and use of human-relevant exposures in animal studies, and 4) cross-species comparisons. In this commentary, we discuss the importance and relevance of each of these criteria in studying the effects of prenatal exposure to EDCs. Our discussion focuses on oxidative stress as a mediator of EDC-related health effects due to its association with both EDC exposure and health outcomes. Our recent study (Veiga-Lopez et al. 2015)¹ addressed each of the 4 outlined criteria and demonstrated that prenatal bisphenol-A exposure is associated with oxidative stress, a risk factor for developing diabetes and cardiovascular diseases in adulthood.

Keywords:

• EDCs
• oxidative stress
• organizational effects
• prenatal programming
• risk assessment

Introduction

Endocrine disrupting chemicals (EDCs), such as bisphenol A (BPA) and phthalates (e.g. diethylhexyl phthalate, DEHP), are ubiquitously present in the environment and humans, and have been the subject of rigorous scientific investigation in recent years due to the potential for a variety of adverse health outcomes.² EDCs act as hormone agonists or antagonists, interfere with signaling mechanisms, and consequently disrupt hormonal homeostasis and developmental processes. Importantly, in utero exposure to EDCs have the potential to alter developmental trajectories of offspring, thus influencing health and disease status later in life, illustrating the concept of the developmental origins of health and disease (DOHaD).³

There is continuing debate regarding the health risks posed by exposures to EDCs.⁴ To better understand human risk from developmental exposure to EDCs, the following criteria should be considered: 1) characterization of target health outcomes and their mediators, 2) study of exposures within the context of critical periods of development, 3) accurate estimates of human exposure and use of human-relevant doses in animal studies, and 4) cross-species comparisons for establishing weight-of-evidence for adverse health outcomes (Fig. 1). Our recent study¹ investigating perinatal BPA exposure in mouse, rat, and sheep models, as well as in human pregnancy and infant samples, applied these principles in evaluating oxidative stress and free fatty acid (FFA) outcomes, thus providing a framework for conducting future studies aimed at evaluating implications of developmental EDC exposures on life course health effects. Here we review the criteria outlined above to describe its importance in evaluating human health risk to EDCs, with a focus on oxidative stress as a mediator of adverse health outcomes.

Figure 1. Criteria to Establish Human Risk from Developmental EDC Exposures. Cross species comparisons are necessary at each stage of experimental design from critical periods of exposure to dose to adverse outcome measurements to establish weight of evidence for translation to human health and determination of human risk of disease. Using animal models that have similar, measureable adverse health outcomes, such as sensitive measure of oxidative stress, in the pathway to chronic disease development are important for understanding disease development and progression. Mediation by oxidative stress is shown via arrows A and B. In order for oxidative stress to be a mediator, it must be significantly associated with both EDC dose and chronic disease outcome and must be in the pathway from exposure to disease (arrow C).

Oxidative Stress As a Mediator of EDC-Related Health Outcomes

Identifying and assessing mediators of health outcomes, which are effectors in the pathway of exposure to disease, is essential for characterizing the impact of developmental insults on health outcomes. For example, inflammation has been identified and extensively studied as a potential mediator of obesity, diabetes, and premature birth.^5,6 Recently, oxidative stress has emerged as an investigative mechanism for EDC toxicity.

Oxidative stress is classically defined as an imbalance in oxidant and antioxidant species within a system in which oxidant species are predominant.⁷ Interestingly, although oxidative stress and exposure to EDCs have been associated with many of the same health effects, they are rarely studied in parallel. For example, both oxidative stress and EDC exposures have been associated with metabolic syndrome, insulin resistance, diabetes, obesity, and cardiovascular complications.^8,9 In one study, Houstis et al.¹⁰ provided evidence of a causal relationship between increased production of reactive oxygen species (ROS), molecules that cause oxidative damage and induce oxidative stress, and insulin resistance in mice. In another study, Alonso-Magdalena et al.¹¹ demonstrated that prenatal exposure to BPA induced insulin resistance in mice. Both oxidative stress and EDCs have also been implicated in intrauterine growth restriction and preterm birth, 2 risk factors associated with adult onset metabolic diseases.^12,13 However, only a few studies have simultaneously examined oxidative stress pathways mediating the impact of prenatal EDC exposure on health outcomes using the sensitive measures for measuring markers of oxidative stress described here.

Despite the overlaps in health outcomes, oxidative stress has only recently been studied as a potential mediator of EDC-related outcomes. Recent studies have reported associations between a variety of different EDCs and oxidative stress.^14-16 For example, phthalate-induced reproductive toxicity has been linked to oxidative stress in both rats and humans,^13,17 while perinatal exposure to thimerosal, tributyltin and benzene have been implicated in production of oxidative stress in rats and mice.^14-16 Given its role in disease development together with increasing evidence supporting a link to EDC exposures, oxidative stress is worthy of further study, particularly in the context of prenatal exposures.

Measures of oxidative stress

A wide variety of measures are used to evaluate oxidative stress, each with distinct applications and interpretations. Many measures of oxidative stress can be obtained using minimally invasive methods (e.g., blood draw and urine collection), making them relatively easy to implement in both animal and human studies. Others involve direct analysis of a target organ or tissue. Careful consideration of the EDC and health outcome(s) of interest is essential in both choosing the measurement to implement and in interpreting results. While elevated ROS levels represent a hallmark of oxidative stress, ROS are short-lived and difficult to accurately measure and interpret.¹⁸ Therefore, more stable measures or markers of oxidative stress, such as covalently modified amino acids and proteins, are needed for accurate characterization of oxidative stress. The stability of these markers is dependent on a variety of factors, but under steady state conditions, they can generally be interpreted as representative of the oxidative stress environment of the cells or tissues at the time of collection.

Historically, immunoassay-based me-thods for assessing markers of oxidative stress such as malondialdehyde (MDA) and thiobarbituric acid reactive substances (TBARS) have been widely used to measure oxidative stress. However, state of the art analytical techniques have been developed in recent years for quantification of specific biomarkers of oxidative stress and the link between EDC exposures and chronic disease outcomes. Below, we discuss a wide range of these highly sensitive analytical methods available for measuring stable markers of oxidative stress and their relevance in studying developmental exposures to EDCs. Many of these methods are yet to be applied for the study of developmental programming following prenatal EDC exposure (see Table 1).

Table 1. Sensitive Markers of Oxidative Stress and Their Use in Studies on Representative EDCs^Δ

Measure of Oxidative Stress	Target Biomolecule	Sensitive Analytical Methods^*	Sample Matrix	EDCs Studied - Direct Exposure	EDCs Studied - Fetal Exposure
o,o-dityrosine (DiY), 3-chlorotyrosine (ClY), and 3-nitrotyrosine (NY)	Amino acids	LC-MS/MS; GC-MS/MS	Plasma, serum, urine, tissues	NS	BPA^Citation1
F2-isoprostanes	Lipids	LC-MS; GC-MS	Plasma, urine	PCBs,^Citation28 OP pesticides,^Citation29 PAHs^Citation53	NS
Free Fatty Acids (FFA)	Lipids	GC; GC-MS	Plasma, tissues	OP pesticides^Citation38	BPA,^Citation1 phthalates^Citation37
8-hydroxydeoxyguanosine (8-oxodG)	DNA	HPLC-ECD	Urine, tissues	phthalates,^Citation17 BPA,^Citation54 PAHs^Citation55	BPA^Citation56,57
8-hydroxyguanosine (8-oxoGuo)	RNA	HPLC-ECD	Urine, tissues	NS	NS
Cellular redox potentials	Cell	HPLC	Plasma, serum, blood, tissue	NS	NS

* Only the most sensitive analytical methods reported

NS: Not yet studied in published literature

^Δ Representative EDCs considered for this table included bisphenol a (BPA), phthalates, polychlorinated biphenyls (PCBs), lead, organophosphate (OP) and organochlorine (OC) pesticides, organotins, flame retardants, and polycyclic aromatic hydrocarbons (PAHs)

Oxidation products of amino acids

Recent studies have utilized products of oxidized aromatic amino acids in proteins as a measure of oxidative stress. These products are commonly measured in plasma, serum, urine, and tissues, and can be representative of either systemic or target organ oxidative stress.¹⁸ However, to accurately analyze products of amino acid oxidation, the target tissue should be perfused with an antioxidant buffer prior to tissue harvest and sample collection, and samples stored in this buffer at −80°C until analysis.¹⁸ Oxidized amino acid products can then be measured using liquid chromatography (LC) or gas chromatography (GC) coupled with tandem mass spectrometry (MS/MS), and with high performance liquid chromatography (HPLC). HPLC may result in coelutions of similar compounds, and is therefore not as specific as MS.^1,18

Modified tyrosine residues are one of the most commonly measured oxidation products due to their stability and usefulness in identifying specific oxidation pathways.¹⁸ These molecules represent both oxidized phenylalanine and tyrosine residues.¹⁸ Examples of oxidized tyrosine amino acid products include o,o-dityrosine (DiY), 3-chlorotyrosine (ClY), and 3-nitrotyrosine (NY). DiY is formed upon oxidation and radical formation of a tyrosine residue and can be formed through both nitrosative and oxidative pathways.¹⁸ Recently, DiY has been found to be elevated in metabolic pathologies, including atherosclerosis and hyperlipidemia.¹⁹ Since EDCs, such as BPA, have also been linked to similar metabolic conditions,⁸ measuring DiY following prenatal exposure to EDCs would be useful to identify oxidative stress pathways.

ClY, a tyrosine modification catalyzed by myeloperoxidase, is associated with macrophage activation and acute inflammation.^18,20 Elevated levels of ClY are associated with similar health effects as those associated with elevated DiY, such as atherosclerosis and cardiovascular disease.^20,21 Thus, ClY may also be of interest to investigators studying EDCs.

NY is formed by nitration of tyrosine and is mediated by reactive nitrogen species and is characteristic of nitrosative pathways.¹⁸ Increased levels of NY have been linked with a variety of negative health outcomes associated with preterm birth, such as perinatal asphyxia.²² Results from several studies have suggested that estrogenic compounds may impact NY formation, indicating a role for NY in response to EDCs. For example, a study performed in estrogen receptor-α (ERα) knock out mice demonstrated that impaired ERα signaling causes an increase in NY production and metabolic dysregulation.²³ This link between estrogen signaling and NY demonstrates the usefulness of measuring NY levels in exposure studies involving EDCs that bind estrogen receptors.

Oxidation products of lipids

Products of lipid peroxidation represent another set of common measures of oxidative stress. These compounds are typically less stable than amino acid products due to rapid reactions following initial oxidation. Nonetheless, lipid peroxi-dation products may have specific biological functions and are detectable in a variety of biological matrices,^18,24, but these sensitive analytical techniques have not yet been used in prenatal EDC exposure studies. Lipid oxidation products are commonly detected using a combination of LC or GC and MS.²⁴ F₂-isoprostanes are common markers of lipid peroxidation that are released into urine and plasma. They can serve as systemic markers of oxidative stress that can be measured with minimal invasiveness.²⁵ However, to prevent ex-vivo oxidation, BHT should be added to collected samples prior to freezing and storing at −80°C.²⁶ Elevated F₂-isoprostane levels, as measured by ELISA-based techniques, have been associated with metabolic dysregulation.²⁷ Elevated F₂-isoprostane levels have also been associated with direct exposure to EDCs such as polychlorinated biphenyls (PCBs) and organophosphate (OP) pesticides (Table 1).^28,29 Investigations utilizing sensitive analytical techniques to measure F₂-isoprostanes in prenatal EDC exposure studies are needed to further elucidate how EDC exposures during critical periods in development may impact lipid peroxidation and health outcomes.

Oxidation products of nucleic acids

Multiple methods are used to measure DNA and RNA oxidation. To quantify DNA and RNA oxidation, HPLC combined with electrochemical detection (HPLC-ECD) is a common sensitive method used to measure oxidized guanine residues.³⁰ Urine, plasma and tissue samples should be stored at −20°C or −80°C immediately following collection, and prior to HPLC-ECD analysis. Specialized technique with cold high salt guanidine thiocyanate, catalase and 2,2,6,6-tetramethylpiperidine-N-oxyl should be used for extraction DNA from the sample to prevent spurious oxidation.^31,32

Guanine residues are the most commonly measured oxidized nucleosides, because guanine has the lowest redox potential and is the most susceptible to a majority of oxidative processes, although thymine is the primary target of hydroxyl radicals.³⁰ Hence, 8-hydroxydeoxyguanosine (8-oxodG) is most commonly used to assess DNA oxidation and 8-hydroxyguanosine (8-oxoGuo) for RNA oxidation.³⁰ 8-oxodG and 8-oxoGuo in urine can be interpreted as measures of global oxidative stress, while in tissues, they represent local oxidative stress.³⁰ However, 8-oxodG and 8-oxoGuo levels in plasma are dictated more by kidney function than oxidative stress, and therefore cannot be used to make comparisons across individuals.³⁰

DNA and RNA oxidation measures are becoming increasingly popular and can be useful in measuring oxidative stress associated with EDC exposure. For example, Ferguson et al.¹³ demonstrated that increased phthalate metabolites present in urine samples from a population of pregnant women were associated with increased urinary 8-oxodG. In addition, a recent study that used more sensitive measures for 8-oxodG found that it was associated with phthalate-induced toxicity in peripubertal rats, demonstrating its usefulness in EDC exposure studies (Table 1).¹⁷

Cellular redox potentials

Another method used to evaluate oxidative stress is the direct measurement of reactive thiol species. Reactive thiols, such as glutathione (GSH) and its oxidized form glutathione disulfide (GSSG), often govern the redox state of the cell and can be directly measured in cells, tissues, blood and other biological matrices using HPLC.³³ Because GSH may rapidly auto-oxidize upon contact with air, samples must be collected and stored in reducing buffer at −80°C.³⁴ Applying the Nernst equation to the measured concentrations of each redox pair provides an estimate of the cellular redox potential. The same redox HPLC method can be used to evaluate S-glutathionylation levels, protein-bound GSH, which is another indicator of oxidative stress.³³ These measurements are extremely sensitive and may detect early oxidative stress responses prior to lipid oxidation and other measures of oxidation-related outcomes.³³ Early data from current investigations in our group indicate that these measurements may be useful in detecting changes in oxidative stress due to prenatal BPA exposure (Marchlewicz et al. unpublished data).

Free fatty acids and relationship with oxidative stress

Multiple studies have linked elevated plasma free fatty acids (FFA) levels with EDCs, metabolic syndrome, and insulin resistance, making FFAs an attractive measurement to use as an outcome measure. ^35,36 Free fatty acids (FFAs) have been demonstrated to increase the formation of oxidative species, such as hydrogen peroxide and hydroxide radicals, through activation of NAD(P)H oxidase, and thus can serve as a proxy for oxidative stress.³⁵ FFA levels are typically analyzed in plasma and tissues and can be quantified with GC.^1,35 Since FFAs are not characterized by their oxidation state, there is no need for preventative measures against auto-oxidation, although collected tissues should be immediately frozen at −80°C for storage.³⁷ Recently, we found that mice prenatally exposed to BPA had significant changes in FFAs; myristic acid was decreased and omega 6 γ-linolenic acid was increased relative to controls.¹ Other studies have found that FFA levels can be altered by exposure to OP pesticides and phthalates (Table 1).^37,38 More research is needed to elucidate the role for FFAs in EDC-mediated toxicity.

Exposures in The Context of Critical Periods and Susceptibily During Development

To assess human health risk to EDCs, it is imperative to evaluate exposure in the context of developmental stage. Critical periods of development are windows during prenatal and early postnatal life at which organ differentiation occurs. These periods are particularly susceptible to environmental insults due to the potential for inducing organizational changes that persist throughout the life course.³⁹ During fetal development, different organ systems begin to develop at different time periods. For instance, in humans, the nervous system begins developing at gestational day 18 to 19, while the reproductive system begins to develop at gestational day 27.⁴⁰ Thus, the susceptibility to EDC exposure and health outcome is dependent on the critical period for a given target organ system. Prenatal exposure to phthalates, for example, has been shown to alter Leydig cell differentiation, which occurs during sexual differentiation of the reproductive tract.⁴¹

It is important to recognize that critical periods differ among species. Some species are precocial, physiologically mature at birth, while others are altricial, physiologically immature at birth so developmental programming extends into early postnatal life.⁴² For example, humans and sheep are precocial and their ovaries develop entirely prenatally,⁴³ while rats and mice are altricial and their ovaries continue to develop after birth.⁴² These disparities in developmental timing among species should be considered in studies of EDC exposures.

Cellular redox environment and signaling is an important component of development and can be perturbed by EDC-induced oxidative stress.⁴⁴ EDC-induced oxidative stress during critical developmental windows would prevent the conceptus from properly responding and adapting to changes in redox state or increases in ROS.⁴⁴ For example, we found prenatal BPA exposure in humans during the first trimester, when several organ systems differentiate, induces oxidative stress in the mother and the fetus.¹

Assessing Human-Relevant EDC Exposure Levels and Tissues

Accurate characterization of human exposures to EDCs as well as utilization of human-relevant exposures with particular attention to dose levels, including low and non-monotonic dose effects,⁴⁵ in animal models is crucial for risk assessment. Of particular relevance here, some ROS can form from secondary effects rather than primary effects of EDCs, which can be dependent on dose.⁴⁶ EDCs and their metabolites are commonly measured in blood, plasma, and urine in order to estimate human exposures. Likewise, within mammalian toxicological studies, the choice of tissue of analysis for oxidative stress marker should reflect human biology, with multiple tissues assessed when possible. Our recent work evaluating mouse dams exposed during gestation and lactation to BPA in the diet revealed tissue specific alterations in blood and liver analyzed for redox potentials.⁴⁷ Ultimately, a firmer understanding of EDC toxicokinetics in both animal models and humans is essential to accurately estimate exposure levels and estimate doses for testing in animal studies. In addition, because of the ubiquitous presence of many EDCs in the environment, care needs to be taken to use contaminant-free collection, storage, and analytical materials such as that employed in a recent BPA round robin study.⁴⁸ It is also important to utilize well-validated analytical methods for measurement purposes. Appropriate blanks need to be included both during collection and measurement.⁴⁹ Recoveries need to be verified by spiking known concentrations of the EDC of interest. Once accurate estimates of human exposure have been established, animal studies should use this information in choosing appropriate doses and dosing strategies of EDC for testing, which is crucial for assessing human health risk.

Cross-Species Comparison to Determine Weight-of-Evidence on Adverse Health Outcomes

Because critical periods in development differ across species, it is imperative to interpret results within this context. Cross species comparisons enable increased evaluation of weight of evidence for risk assessment. A few investigations that have examined cross-species responses to EDCs have found distinct responses to EDCs across species. For example, Johnson et al.⁵⁰ demonstrated that in utero exposure to DEHP had different effects on Leydig cell hormone synthesis in the testes of rats, mice and humans; DEHP was found to inhibit hormone production in fetal Leydig cells in testes of rats, but not in mice. Another recent in vitro study that examined the direct effects of 6 different EDCs (mono-(2-ethylhexyl) phthalate (MEHP), cadmium, depleted uranium, diethylstilbestrol (DES), BPA, and metformin) on gametogenesis and steroidogenesis in rat, mouse, and human testes cells found that many of the compounds had species-specific effects.⁵¹ Thresholds of oxidative stress have also been shown to be different across species. For example, Hassan et al.⁵² found that rats, mice, guinea pigs, and hamsters had different sensitivities to endrin-mediated lipid peroxidation. Therefore, cross-species studies are essential for providing accurate risk assessment and help translate findings in animals to humans.

Conclusions

Our recent study, Veiga-Lopez et al.¹ follows the outlined criteria for assessing human health risk to BPA, a well-known EDC. We examined oxidative stress as a mediator of adverse health outcomes, studied BPA exposure in the context of the prenatal period when organizational effects are documented, assessed BPA levels in humans using the validated methodology, applied relevant human exposure and dose levels in animal studies, and evaluated 4 species, including humans. In following these criteria, our study demonstrated that prenatal exposure to BPA leads to oxidative stress, a risk factor for development of cardiovascular disease and diabetes in adulthood, in offspring of 3 species. Our study was not without limitations, however. The samples sizes used in this study are small, limiting the generalizability of the study. Additional large-scale studies are needed to expand these observations.

Given the growing body of evidence linking oxidative stress to EDCs and the necessity to interpret results in the context of species being studied, investigations using a cross-species approach are needed for evaluating risk from developmental exposures to EDC. The recent abundance of sensitive analytical methods that can be applied to measure oxidative stress will now allow for further elucidation of toxicity mechanisms for these compounds, thus advancing our understanding of how these chemicals contribute to human health.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by NIH grants R01 R01ES01654, R01 ES017005, R01 ES017524, P01 ES02284401, P30 ES017885, as well as US. Environmental Protection Agency (US EPA) grant RD83543601. Support for KN and EHM was provided by NIH Institutional Training Grants T32 ES007062 and T32 HD079342, respectively. The contents of this publication are solely the responsibility of the grantee and do not necessarily represent the official views of the US EPA or the NIH. Further, the US EPA does not endorse the purchase of any commercial products or services mentioned in the publication.