Bisphenol A Exposure Disrupts Metabolic Health Across Multiple Generations in the Mouse

Abstract

Accumulating evidence has suggested that a suboptimal early life environment produces multigenerational developmental defects. A proposed mechanism is stable inheritance of DNA methylation. Here we show that maternal bisphenol A (BPA) exposure in C57BL/6 mice produces multigenerational metabolic phenotypes in their offspring. Using various methods including dual-energy X-ray absorptiometry analyses, glucose tolerance tests, and perifusion islet studies, we showed that exposure to 10 μg/kg/d and 10 mg/kg/d BPA in pregnant F0 mice was associated with higher body fat and perturbed glucose homeostasis in F1 and F2 male offspring but not female offspring. To provide insight into the mechanism of the multigenerational metabolic abnormalities, we investigated the maternal metabolic milieu and inheritance of DNA methylation across generations. We showed that maternal glucose homeostasis during pregnancy was altered in the F0 but not F1 female mice. The results suggested that a compromised maternal metabolic milieu may play a role in the health of the F1 offspring but cannot account for all of the observed multigenerational phenotypes. We further demonstrated that the metabolic phenotypes in the F1 and F2 BPA male offspring were linked to fetal overexpression of the imprinted Igf2 gene and increased DNA methylation at the Igf2 differentially methylated region 1. Studies in H19^Δ3.8/+ mouse mutants supported the role of fetal Igf2 overexpression in altered adult glucose homeostasis. We conclude that early life BPA exposure at representative human exposure levels can perturb metabolic health across multiple generations in the mouse through stable inheritance of DNA methylation changes at the Igf2 locus.

Epidemiological and animal studies have demonstrated that the early life environment plays a critical role in adult metabolic health (1). Accumulating evidence has further suggested that the impacts of the early life environment can be transmitted across multiple generations. A proposed mechanism is epigenetic inheritance, possibly through stable changes in DNA methylation (2–5). Nevertheless, key components of the developmental origins of health and disease concepts remain to be addressed, including the identification of developmentally relevant loci that are vulnerable to environmental perturbations and evidence that the early environment can alter the DNA methylation profiles of these loci across multiple generations.

We previously reported that endocrine-disrupting chemical (EDC) exposure perturbs imprinted gene regulation (6). Imprinted genes harbor epigenetic modifications, including differential DNA methylation at imprinting control regions (ICRs), which results in parental allele–specific gene expression. Consistent with reports that exposure to the EDC bisphenol A (BPA) is associated with modification of DNA methylation in rodents (7), we showed that maternal BPA exposure at levels representative of human exposure (10 μg/kg/d or 10 mg/kg/d) in female (F0) mice before and during pregnancy altered DNA methylation of ICRs in placentas and embryos from midgestation mouse conceptuses (6). In utero BPA exposure in these offspring was associated with loss of imprinting of several imprinted genes on chromosome 7 (6). Because some of the aberrantly expressed imprinted genes are implicated in growth and/or glucose homeostasis, including insulin-like growth factor 2 (Igf2) and cyclin-dependent kinase inhibitor 1c (Cdkn1c), we postulated that BPA-induced altered imprinted gene expression disrupts the growth potential and the future metabolic health of the F1 offspring. Moreover, BPA-induced DNA methylation changes at imprinted loci can be inherited stably across generations, potentially leading to multigenerational transmission of altered phenotypes. Furthermore, the germ cells of the F1 progeny are developing at the time when in utero BPA exposure occurs, and exposure can reprogram the germ cell epigenome. Because these germ cells represent the future F2 offspring, we additionally hypothesized that the health of the subsequent generation may be affected.

In the current study, we tested the hypothesis that early developmental BPA exposure alters growth and metabolic phenotypes in the F1 and F2 offspring. Dual-energy X-ray absorptiometry (DEXA) analyses, glucose tolerance tests (GTTs), and insulin secretion studies revealed that F1 and F2 adult male offspring from BPA exposure groups were fatter, more glucose intolerant, and had altered insulin secretion relative to controls. These effects were highly sex specific because the females were unaffected. Further analysis demonstrated that F2 embryos inherited a subset of the imprinting features of the F1 embryos, including overexpression of the Igf2 gene and elevated DNA methylation at the Igf2 differentially methylated region 1 (DMR1). To test whether Igf2 contributed to the perturbed glucose homeostasis in the F1 and F2 BPA mice, we studied the metabolic phenotype of another mouse model, the H19^Δ3.8/+, which also exhibits increased Igf2 expression. Our data suggest that Igf2 overexpression during early development is responsible, at least in part, for abnormal glucose homeostasis later in life.

Materials and Methods

Mouse information

Six-week-old virgin C57BL/6 females were purchased from The Jackson Laboratory and assigned to the following diets: (1) modified AIN 93G diet (TD 95092 with 7% corn oil substituted for 7% soybean oil; Harlan Teklad) as “control”; (2) modified AIN 93G diet supplemented with 50 μg/kg BPA (TD 110337; Harlan Teklad) as “lower dose”; or (3) modified AIN 93G supplemented with 50 mg/kg BPA (TD 06156; Harlan Teklad) as “upper dose”. Teklad Diets (Harlan Laboratories Inc) provided all ingredients except BPA (>99% in purity; Sigma-Aldrich). Based on the average body weight (bw) of an adult mouse (25 g) and daily food consumption (5 g), we estimated exposure levels per mouse to be 10 μg/kg bw/d (lower dose) and 10 mg/kg bw/d (upper dose). The doses were selected based on our previous study demonstrating the effects of exposure at these doses on DNA methylation and RNA expression at imprinted loci (6). Unconjugated BPA in serum from pregnant mice measured in this previous study indicated levels representative of human exposure (6). The lower dose represented the safe human exposure level (≤50 μg/kg bw/d) (8). After 2 weeks of treatment, females were time-mated to C57BL/6 males, and the day a plug was detected was assigned as embryonic day (E) 0.5. Pregnant females were designated as the “F0.” At E16.5 to E17.5, a subset of the pregnant F0 mice was included in the glucose tolerance studies. The remaining F0 females were allowed to deliver and raise their offspring until weaning; the offspring were designated as the “F1.” At postnatal day (PND) 21, all F1 mice were weaned on to the TD 95092 (control) diet, so these mice were exposed to BPA during gestation and lactation periods only. Body weight was recorded weekly starting at PND 1 until PND98 and at the time DEXA was performed (between PND98 and PND117). A subset of adult F1 females was time-mated to control C57BL/6 males and included in the glucose tolerance studies at E16.5 to E17.5 or allowed to deliver offspring naturally. These offspring were designated as the “F2.” For the molecular analysis of Igf2, we mated F1 females to Cast7 males as described previously (6). For our mutant mouse studies, the H19^Δ3.8/+ deletion mice were derived as described previously (9, 10). To generate the H19^Δ3.8/+ offspring, H19^Δ3.8/+ heterozygous female mice were mated to wild-type male mice. All mouse work was conducted with the approval of the University of Pennsylvania Institutional Animal Care and Use Committee.

Igf2 mRNA expression analysis

Total RNA was extracted from E9.5 to E10.5 tissues using the TRIzol reagent according to the manufacturer's guidelines and quantified using a NanoDrop spectrophotometer. Using Invitrogen SuperScript Transcriptase III and random hexamers, we subsequently generated cDNA. Allele-specific expression analysis has been described previously (6). For total expression, real-time quantitative PCR analysis was conducted using an Applied Biosystems 7900HT Fast Real Time PCR system with the following protocol: 5 μL of Master Mix, 0.12 μL of reverse and forward primers, and 2.76 μL of water and 2 μL of cDNA (50 μg). The following primers were used: 5′CGCTTCGTTTGTCTGTTCG3′ (forward) and 5′GCA GCACTCTTCCACGATG3′ (reverse) with product size of 94 bp. Three reference genes, Arppo, Gapdh, and Nono, were used in this study.

DNA methylation analysis

DNA was extracted using the standard phenol-chloroform method and quantified using a NanoDrop spectrophotometer. One microgram of DNA was bisulfite treated using the EpiTect Bisulfite Kit (QIAGEN) following the manufacturer's protocol. Pyrosequencing was conducted to measure the methylation status of the promoter region of the Igf2 DMR1 (chr7: 149851180–149851655; NCBI37/mm9) as described previously (6). In brief, 50 ng of bisulfite-treated DNA was used in the PCR. The following primers were used: 5′-TGAGGTTAGATTAGGTTGTAAGTT-3′ (forward) and 5′-CTTCCCTACCCCTTAAACC-3′ (biotinylated reverse). For the sequencing, the following primers were used: 5′-GGATTTTGTTAGGTAGGA-3′ (CpG sites 1 and 2) and 5′-TTTTAGAGGTTTTTGGAGAA-3′ (CpG sites 3 and 4).

GTTs

At PND98 to PND117 (for the adult F1 and F2 offspring) or at day 16.5 to 17.5 of gestation (for the pregnant F0 and F1 females), mice were fasted overnight and subsequently injected with 2 g/kg bw of glucose intraperitoneally. At 0, 15, 30, 60, and 120 minutes, blood was sampled from the tail vein and analyzed by a handheld glucometer.

Insulin measurement and islet perifusion study

Fasting blood insulin levels were measured by ELISA (ALPCO). Isolation of pancreatic islets, perifusion, and insulin assays were performed as described previously (11, 12). In brief, islets were isolated by collagenase digestion and cultured with 10 mM glucose in RPMI 1640 medium (Sigma-Aldrich) for 2 days. Islets were perifused with a Krebs-Ringer bicarbonate buffer containing 0.25% BSA at a flow rate of 1 mL/min. Glucose ramps were performed at increments of 0.5 mM/min. For maximum insulin release, 30 mM KCl was used. Insulin was measured by homogeneous time-resolved fluorescence technology (Cisbio kit).

DEXA

To assess body composition, DEXA scans were performed (GE Lunar PIXImus x-ray densitometer) on a subset of male and female adult (PND98–PND117) F1 and F2 mice as described previously (13). In brief, each mouse was anesthetized throughout the duration of the scan (∼5 minutes) using isoflurane. Body fat, lean mass, bone mineral content, and bone mineral density were measured.

Statistical analysis

The significance of differences among multiple groups and between 2 groups was examined using ANOVA and the Student t test, respectively. All values are presented as means ± SEM. A P value of <.05 was considered significant. All data were analyzed using Prism data analysis software.

Results

BPA-exposed F1 male offspring have lower birth weights and develop obesity in adulthood

To assess whether BPA exposure in utero produced a multigenerational metabolic phenotype, we had to establish whether the F1 offspring in our experimental model exhibited impaired glucose homeostasis. We first determined whether BPA exposure during gestation and lactation (see Materials and Methods) affected the postnatal growth of C57BL/6 offspring. At PND1, F1 lower dose BPA male offspring had significantly reduced body weights relative to those of controls (P = .01) (Figure 1A). The body weights of F1 male offspring from this group remained lower than those of controls at PND14 and PND21 (P = .008 and P = .02, respectively) (Figure 1A). In contrast, the body weights of F1 male offspring from upper dose BPA exposure groups did not differ from those of controls at these time points (Figure 1A). Notably, F1 lower dose BPA males exhibited accelerated weight gain postweaning, and by PND28, weights were similar in all exposure groups. Between PND98 and PND117, F1 lower dose BPA adult mice had significantly increased body weights relative to those of controls (30.3 ± 1.2 and 25.9 ± 1.0 g, respectively; P = .002). The upper dose BPA males showed a trend toward increased body weights compared with those of controls (27.3 ± 0.6 g; P = .07). We did not observe any difference in food intake of F1 male mice from the different exposure groups (Supplemental Figure 1). Consistent with their body weight profiles, F1 male offspring from lower and upper dose BPA exposure groups had significantly higher body fat content between PND98 and PND117 relative to that of control males (P < .05) (Figure 1B). Interestingly, upper dose BPA mice had significantly reduced bone mineral density and bone mineral content (Supplemental Figure 2, A and B). We did not detect significant differences in body weight, body fat content, and bone mineral content and density in the F1 females at any time point analyzed (Supplemental Figures 2, A–C and 3A).

Figure 1.

Metabolic profiles of F1 control and BPA male offspring. A, F1 weekly body weight measurement in the F1 male offspring from control, lower dose, and upper dose BPA exposure groups (*, P < .05). B and C, At PND98 to PND117, F1 adult male offspring from BPA exposure groups are fatter and more glucose intolerant than controls. D and E, perifusion islet studies showing basal and glucose-stimulated insulin secretion.

BPA-exposed F1 male but not female offspring have impaired glucose homeostasis

Epidemiological, clinical, and experimental animal studies have demonstrated that perinatal growth restriction followed by postnatal accelerated growth is linked to the development of adult obesity and glucose intolerance (14). To determine whether increased body fat content in BPA-exposed F1 male offspring was associated with perturbed glucose homeostasis, we performed intraperitoneal GTTs and measured fasting insulin levels. Fasting glucose concentrations in F1 upper dose and lower dose BPA male offspring were not significantly different relative to those of control mice (data not shown). In contrast, there was a trend in increased insulin levels in the F1 upper dose mice (1.29 ± 0.82 ng/mL; n = 5; P = .08) and significantly elevated levels in the lower dose BPA males (0.44 ± 0.15 ng/mL; n = 4; P = .04) relative to those of controls (0.24 ± 0.08 ng/mL; n = 6). On a GTT, glucose area under the curve (AUC) was greater in F1 upper dose BPA male offspring than in control male offspring (P = .04) (Figure 1C), consistent with the phenotype of glucose intolerance. F1 lower dose BPA male offspring showed a trend toward an increase in glucose AUC compared with that of control males (P = .06) (Figure 1C). In contrast to the males, F1 upper dose and lower dose female offspring had glucose AUCs similar to those of control offspring (Supplemental Figure 3B), suggesting that BPA exposure-induced glucose intolerance was highly sex specific.

BPA exposure alters glucose-stimulated insulin secretion

Impaired glucose tolerance is characterized by reduced insulin secretion and/or peripheral insulin resistance. To investigate whether impaired glucose tolerance in BPA-exposed F1 male offspring is associated with insulin secretory defects, we measured insulin secretion in response to a glucose gradient and KCl in isolated islets (15, 16). F1 upper dose BPA males exhibited significantly elevated basal rates of insulin release relative to those of controls (P < .01) (Figure 1D), consistent with insulin resistance. In response to an increasing glucose gradient, we did not detect a significant difference in insulin release between F1 control and upper dose BPA males (Figure 1E). Furthermore, KCl-induced insulin release was not different in the islets of upper dose BPA male mice relative to those of controls, consistent with an intact insulin secretory apparatus. These findings suggest that glucose intolerance in F1 upper dose BPA mice is not linked to impaired insulin secretion but rather is due to insulin resistance. In contrast, lower dose BPA males had similar basal rates of insulin release (Figure 1D), but significantly reduced maximal glucose-stimulated insulin release compared with that of controls (P < .01) (Figure 1E).

Effects of BPA-induced genomic imprinting defects are partially transmitted to the F2 offspring

After establishing that the F1 offspring had altered body composition, glucose tolerance, and insulin secretion, we subsequently studied the phenotypes of the next generation. An important question with in utero environmental exposures is whether offspring of the F1 exposed mice (ie, the F2 mice) also exhibit altered phenotypes in the absence of further exposure. In the current model, the F2 offspring were exposed to BPA as developing germ cells. Previously, we demonstrated that upper dose BPA exposure resulted in biallelic expression of the imprinted Igf2 gene and increased total mRNA expression in the F1 embryos (6). To determine whether genomic imprinting defects were observed in the F2 offspring, a subset of C57BL/6 F1 females were time-mated to unexposed Cast7 males, and embryos were collected at E9.5 to E10.5. We assessed allele-specific and total mRNA expression of the Igf2 gene. In contrast to the F1 embryos, we did not observe biallelic expression of the Igf2 gene in the F2 embryos (data not shown). F2 upper dose BPA embryos, however, significantly overexpressed total Igf2 mRNA relative to those of controls (Figure 2A).

Figure 2.

E10.5 F2 embryos have elevated total Igf2 mRNA expression and increased DNA methylation at CpG site 2 of the DMR1. A, Total mRNA expression of Igf2 in whole embryos from control, lower dose, and upper dose BPA exposure groups relative to that of the reference genes. All values were normalized to controls. B, DNA methylation percentages at each CpG site of the Igf2 DMR1. Increased DNA methylation was detected at CpG site 2 in F2 upper dose BPA embryos.

Our previous study revealed that Igf2 overexpression in the F1 upper dose BPA embryos was linked to increased average DNA methylation at 4 CpG sites of the DMR1 (6), specifically at CpG sites 1, 2, and 4 (Supplemental Figure 4). In contrast to the F1 embryos, we did not detect significant changes in the average DNA methylation levels at these 4 CpG sites of the DMR1 in the F2 embryos (data not shown). When individual CpG sites were further analyzed, the F2 upper dose BPA embryos showed elevated DNA methylation at CpG site 2 relative to that of controls (Figure 2B).

The metabolic phenotypes of the F1 offspring were transmitted to the next generation

Because the F2 offspring shared the molecular phenotypes of the previous generation, we asked whether the F2 offspring would exhibit a similar metabolic phenotype. To determine whether the F2 offspring also exhibited an altered metabolic profile, a subset of F1 females was mated to unexposed C57BL/6 males, and adult F2 offspring were examined for body weight, body fat, glucose tolerance, and insulin secretion. The body weights of lower and upper dose male offspring were statistically not different from those of controls at PND1 (1.43 ± 0.05, 1.59 ± 0.04, and 1.49 ± 0.05 g, respectively; P > .05) or at adulthood (28.0 ± 0.8, 28.8 ± 0.9, and 27.8 ± 1.1 g, respectively; P > .05). Similar to the F1 offspring (Figure 1B), however, F2 male offspring from the upper dose BPA group were significantly fatter than controls at PND98 to PND 117 (P < .05) (Figure 3A). Despite a higher body fat content in upper dose BPA F2 mice, fasting glucose (data not shown) and insulin concentrations (0.33 ± 0.06 and 0.35 ± 0.03 mg/mL in control and upper dose BPA F2 males, respectively; P > .05) did not differ between groups. Glucose AUC, however, were significantly greater in the F2 upper dose BPA males than in control males (P < .01) (Figure 3B). In contrast, the F2 lower dose BPA males showed glucose AUCs similar to those of controls. These results suggest that, similar to the F1 offspring, F2 upper dose BPA males also exhibit glucose intolerance.

Figure 3.

Metabolic profiles of F2 adult male offspring from control and BPA exposure groups. A and B, At PND98 to PND117, F2 upper dose BPA males are fatter and more glucose intolerant than controls. C and D, Perifusion islet studies showing normal basal insulin secretion but reduced glucose-stimulated insulin secretion in F2 lower dose males.

To determine whether glucose intolerance in the F2 BPA offspring was due to impaired insulin secretion as we observed in the F1 offspring, we conducted perifusion studies to measure basal and glucose-stimulated insulin release. We did not detect significant differences in basal insulin secretion among groups (Figure 3C). The F2 upper dose males showed glucose-stimulated and KCl-induced insulin secretion similar to those of controls (Figure 3D). These results were consistent with our observation that the F1 upper dose males showed normal insulin secretion (Figure 1E). In contrast, islets from lower dose males had significantly reduced glucose-stimulated insulin secretion compared with those of controls (P < .01) (Figure 3D). Furthermore, in the presence of KCl, similar to islets from the F1 lower dose BPA males (Figure 3D), islets from the F2 lower dose males showed a trend toward reduced insulin release relative to that of controls. These data show that the F2 male offspring inherit the metabolic phenotypes of the previous generation.

Maternal effects contributed to the perturbed metabolic health in F1 but not F2 offspring

A previous study revealed that BPA exposure induces gestational glucose intolerance in mice (17). In addition, elevated fetal Igf2 expression increases maternal glucose concentrations during pregnancy, leading to gestational diabetes (18). These studies suggest that BPA exposure and embryonic Igf2 overexpression in our mouse model may cause gestational glucose intolerance during pregnancy. Because gestational glucose intolerance is a risk factor for obesity and altered glucose homeostasis in the offspring, it is possible that our F1 and F2 male offspring become fatter than controls and develop glucose intolerance due to compromised maternal metabolic health. To test this possibility, we performed GTTs in F0 and F1 pregnant mice. Pregnant F0 mice from both the lower and upper dose exposure groups were glucose intolerant when tested between gestation days 16.5 and 17.5 (P < .05 and P < .01, respectively) (Figure 4A). In contrast to BPA-exposed F0 pregnant females, F1 females from upper dose BPA exposed and control groups showed no significant differences in glucose tolerance during pregnancy (Figure 4B). We did not observe a significant difference in maternal food intake in F0 and F1 pregnant mice among the different exposure groups (Supplemental Figure 5, A and B). These results showed that although maternal gestational glucose intolerance could explain the abnormal glucose homeostasis in the BPA-exposed F1 male offspring, it did not contribute to the metabolic phenotypes of the F2 male offspring.

Figure 4.

Glucose AUC measurement in pregnant F0 and F1 mice and in the H19^Δ3.8/+ male mice. A and B, Pregnant F0 females (A), but not F1 females (B), have significantly increased glucose concentrations when tested between E16.5 and E17.5. C, H19^Δ3.8/+ male mice have significantly higher glucose concentrations than wild-type mice.

Overexpression of Igf2 was associated with altered glucose homeostasis

In addition to the well-characterized role of Igf2 in promoting fetal growth (19), increasing evidence has suggested that it influences adult energy metabolism (20–23). To determine whether fetal Igf2 expression contributed to the metabolic phenotypes in our F1 and F2 BPA offspring, we asked whether the metabolic profiles of other mouse models that demonstrate elevated Igf2 expression would be altered in a manner similar to that in BPA-exposed mice. A well-studied mouse model of Igf2 overexpression is the H19^Δ3.8 model, which contains an engineered 3.8-kb deletion of the H19/Igf2 locus including the ICR (9). The ICR is unmethylated on the maternal allele and serves as an insulator that blocks access of Igf2 to enhancers that are shared between H19 and Igf2. Mice that inherit this mutation maternally (ie, H19^Δ3.8/+) exhibit 2.0- to 2.5-fold higher levels of Igf2 at E9.5 (10). If Igf2 overexpression contributes to the metabolic profile of our BPA-exposed mice, we would expect that the H19^Δ3.8/+ mice exhibit glucose intolerance. Consistent with this prediction, GTTs performed between PND98 and PND117 revealed that H19^Δ3.8/+male mice were more glucose intolerant than wild-type mice (P = .05) (Figure 4C). DEXA scanning revealed that these mice showed a trend in increased body weight and body fat content relative to those of wild-type mice (data not shown). Interestingly, H19^Δ3.8/+ females did not develop glucose intolerance (data not shown). These results demonstrate that Igf2 overexpression contributes, but is not exclusively linked, to abnormal glucose homeostasis in the mouse.

Discussion

Increasing evidence has suggested that the early life environment can have a significant impact on the future health of an individual and his or her offspring. In this study, we found that early life exposure to the EDC BPA in the mouse was associated with sex-specific abnormal metabolic health in multiple generations (Supplemental Table 1). Both F1 and F2 lower and upper dose male offspring became fatter than controls, and in the upper dose group, the males developed glucose intolerance during adulthood. Our perifusion islet studies revealed that in F1 and F2 lower dose BPA male mice, glucose-stimulated insulin secretion was impaired. Interestingly, we did not observe perturbed insulin release with upper dose BPA exposure. Both F1 and F2 upper dose males exhibited normal glucose-stimulated insulin secretion, suggesting that the glucose intolerance these mice exhibit may be due to insulin resistance rather than to defects in insulin secretion per se. An increase in basal insulin release in islets of F1 upper dose male mice was further suggestive of an insulin-resistant phenotype. Insulin biosynthesis and release are partly regulated by estrogen receptors (ERs). Previous studies have demonstrated that subcutaneous injection of 100 μg/kg/d BPA alters pancreatic β-cell function and increases insulin release in OF1 male mice and that the effects are dependent on an ERα signaling pathway (24). It is possible that BPA has a nonmonotonic dose response on β-cell function, and exposure to 10 μg/kg/d, but not 10 mg/kg/d, leads to distinct dose-response effects on insulin release in our study. Furthermore, we cannot rule out the possibility that BPA exposure also alters insulin signaling pathways at key target tissues such as liver, skeletal muscle, or adipose tissue (25) in a dose-dependent manner, leading to insulin resistance.

Previous studies have reported similar effects of maternal BPA exposure on glucose homeostasis of the F1 offspring in mouse strains other than the C57BL/6 strain used here (17, 26, 27); however, to our knowledge, this is the first study to demonstrate that the metabolic phenotype can be transmitted to the F2 offspring in the mouse. Li et al (28) recently reported glucose intolerance and insulin resistance in the F2 offspring in the rat after maternal F0 exposure. The species, route of administration, and doses used in our study were different from those of Li et al. Importantly, in addition to these differences, Li et al had generated their F2 mice through paternal transmission (ie, by mating F1 male offspring), which contrasts to the maternal transmission (ie, by mating F1 female offspring) used in our study. Use of maternal transmission, as opposed to paternal transmission males with demonstrated metabolic problems, limits the contribution of the paternal metabolic milieu to the health of the F2 offspring.

There are at least 2 possible mechanisms that mediate the observed multigenerational phenotypic inheritance in the current study: an altered maternal metabolic milieu that affected the health of the F1 and F2 offspring and/or epigenetic inheritance. To gain insights into the contribution of maternal effects on the metabolic phenotype in the F1 and F2 offspring, we investigated gestational glucose homeostasis in the F0 and F1 females during pregnancy. Consistent with a previous report (17), we observed that BPA exposure induced gestational glucose intolerance in pregnant F0 mice (Figure 4A). Importantly, we report for the first time that F1 female mice from the upper dose BPA exposure group did not develop glucose intolerance during pregnancy (Figure 4B). This latter observation strongly suggests that in our environmental exposure mouse model, the metabolic phenotype of the F2 offspring was not caused by altered maternal metabolic milieu but indicates an epigenetic process.

Consequently, we addressed whether the observed multigenerational metabolic phenotype in this BPA exposure mouse model was associated with inheritance of altered DNA methylation. In the F1 embryos, BPA exposure significantly increased DNA methylation at the Igf2 DMR1 and elevated total Igf2 mRNA expression (6). Here, we demonstrated that the F2 embryos shared these molecular phenotypes (Figure 2, A and B, and Supplemental Table 1). We further asked whether Igf2 misregulation during embryonic development was linked to an abnormal growth profile and metabolic health later in life. The Igf2 gene is highly expressed during fetal development and the IGF2 peptide hormone is primarily involved in prenatal growth regulation. Overproduction of IGF2 during development leads to somatic overgrowth (29); mice lacking the Igf2 gene due to an engineered mutation are 60% smaller than wild-type mice (19). Additional studies have shown that IGF2 or Igf2 expression influences adult energy metabolism (20–23). In this current study, we compared the metabolic phenotype of adult F2 offspring from BPA exposure groups with that of H19^Δ3.8/+ mice, which also overexpressed Igf2 during embryonic development. We demonstrated, for the first time, that Igf2 overexpression in the H19^Δ3.8 mouse model was associated with glucose intolerance in the adult male offspring (Figure 4C), consistent with a role for Igf2 in adult metabolism. Igf2 misregulation, however, is not exclusively responsible for the development of the phenotype. H19^Δ3.8/+ females also overexpress Igf2 but do not develop glucose intolerance during adulthood. Our observations suggest that Igf2 misregulation is not the sole contributor to the observed phenotypes, but most likely represents a downstream effect contributing to altered metabolism and that there are unknown, critical upstream events that control the final health outcomes.

Sex-specific differences in developmental outcomes induced by BPA exposure have been reported in other studies (30). BPA is an estrogen mimic that has the ability to bind to ERs, although relative to endogenous estrogens, it has weaker binding affinity and potency. Human and animal studies have suggested that estrogen may provide protective effects to females, which may limit downstream metabolic effects (31). Our current study indicates that there are distinct downstream developmental susceptibilities between the sexes toward metabolic abnormalities in response to early life environmental exposure. In addition, IGF signaling is controlled by GH produced by the pituitary. Because GH pulsatility is highly sex specific, factors that influence IGF signaling may produce different outcomes between males and females, providing another sex-specific mechanism to respond to early life environmental perturbations.

In conclusion, our study demonstrates that early life exposure to environmental EDCs may disrupt the metabolic health of the developing fetus as well as its offspring. The exact mechanisms of this phenotype transmission are unclear, but our study suggests a role for Igf2, possibly through epigenetic misregulation. Limitations of the current study include the unknown role of the ERs in mediating BPA-induced multigenerational metabolic effects and the uncertainty as to whether BPA-induced epigenetic perturbation is linked to its estrogenicity. Although literature reporting unfavorable health effects of BPA is growing, BPA is considered a weak estrogen with poor affinity to the ERs. Furthermore, a recent study has questioned whether BPA levels reported in many studies are sufficient to activate the ERs (32). We demonstrated that the serum BPA level in the pregnant mice in our mouse model was within representative levels of human exposure (6); however, whether this exposure level would produce an adverse health effects in humans is an ongoing research topic. In the future, a more global approach for analyzing the effects of exposure on genome-wide DNA methylation patterns and for studying effects in the F3 offspring would provide more comprehensive mechanistic insights into the multigenerational inheritance of health outcomes in this model system.