Effects of oral exposure to bisphenol A on gene expression and global genomic DNA methylation in the prostate, female mammary gland, and uterus of NCTR Sprague-Dawley rats

Abstract

Bisphenol A (BPA), an industrial chemical used in the manufacture of polycarbonate and epoxy resins, binds to the nuclear estrogen receptor with an affinity 4–5 orders of magnitude lower than that of estradiol. We reported previously that “high BPA” (100,000 and 300,000 μg/kg body weight (bw)/day), but not “low BPA” [2.5–2700 μg/kg bw/day], induced clear adverse effects in NCTR Sprague-Dawley rats gavaged daily from gestation day 6 through postnatal day 90. The “high BPA” effects partially overlapped those of ethinyl estradiol (EE₂, 0.5 and 5.0 μg/kg bw/day). To evaluate further the potential of “low BPA” to induce biological effects, here we assessed the global genomic DNA methylation and gene expression in the prostate and female mammary glands, tissues identified previously as potential targets of BPA, and uterus, a sensitive estrogen-responsive tissue. Both doses of EE₂ modulated gene expression, including of known estrogen-responsive genes, and PND 4 global gene expression data showed a partial overlap of the “high BPA” effects with those of EE₂. The “low BPA” doses modulated the expression of several genes; however, the absence of a dose response reduces the likelihood that these changes were causally linked to the treatment. These results are consistent with the toxicity outcomes.

1. Introduction

Bisphenol A (BPA), a high-production industrial chemical used in the manufacture of polycarbonate plastics and epoxy resins, can be found in many consumer products, including beverage and food containers, medical devices, dental sealants, and recycled and thermal paper. BPA is a ubiquitous environmental contaminant and the majority of the human population has some exposure to BPA (Calafat et al., 2008), primarily orally (FAO/WHO (Food and Agriculture Organization/World Health Organization), 2011; FDA, 2014). The estimated median daily intake for the overall US population is approximately 34 ng BPA/kg body weight (bw)/day (Lakind and Naiman, 2011).

BPA is known to exert estrogenic activity by interacting with the nuclear estrogen receptors alpha and beta (ERα and β) with a binding affinity that is 4–5 orders of magnitude lower than that of estradiol (Gould et al., 1998; Kuiper et al., 1998). Reported effects of BPA on well-characterized estrogenic endpoints in animal models include increased uterine weight and modulation of the expression of estrogen-sensitive genes in different tissues and cells (e.g., An et al., 2002; Bosquiazzo et al., 2010; Diel et al., 2000; Kim et al., 2012). BPA was shown also to interact with the ER-related receptor gamma (ERRγ, Takayanagi et al., 2006), with an affinity about 100-fold higher versus ERs, and G-protein-coupled ER (GPER, Thomas and Dong, 2006), with a relative binding affinity 2% that of estradiol. In vitro studies suggest further that BPA induces rapid, non-genomic responses via membrane-bound ERs, at similar levels as estradiol (Jeng and Watson, 2011; Prins et al., 2014). In addition, effects of BPA on epigenetic mechanisms have been reported, including modulation of global and gene-specific DNA methylation, histone modifications, the expression level of chromatin-modifying genes, including DNA methyltransferases, and microRNA expression (e.g., Bromer et al., 2010; Dhimolea et al., 2014; Ho et al., 2006; Kim et al., 2014; Kovanecz et al., 2014; Kundakovic et al., 2013; Patel et al., 2013; Tang et al., 2012).

The current study incorporated elements of study design previously identified as critical for data interpretation of BPA studies (Goodman et al., 2006; Richter et al., 2007; Shelby, 2008), including the use of the oral route of exposure; use of a sufficient sample size; use of a wide range of BPA doses appropriately spaced to characterize the dose–response curve; inclusion of a concurrent reference estrogen control; use of the litter as the statistical unit; minimization and monitoring of exposure of the study animals to exogenous BPA and other xenoestrogens; and concurrent assessment of the internal dosimetry of BPA. In addition, neonates were dosed directly by gavage rather than via lactation, to augment the perinatal exposure to BPA (Doerge et al., 2010b). The effects of BPA and ethinyl estradiol (EE₂) on body and organ weights, markers of puberty, histopathology, serum hormone levels, clinical chemistry, levels of ER in the brain, and life-stage internal dosimetry of BPA and EE₂ in the same cohort of animals used in the present study were reported previously (Churchwell et al., 2014; Delclos et al., 2014; Rebuli et al., 2014). Seven “low BPA” doses (2.5–2700 μg/kg bw/day, spaced half-log) were evaluated to cover a range where effects of BPA have been reported in various research studies (Shelby, 2008), but not in the guideline-compliant studies (Tyl et al., 2002, 2008). No adverse effects were detected in the “low BPA” dose range (Delclos et al., 2014), although statistically significant changes were observed in the expression level of ERα- and ERβ-coding genes in the hypothalamus (Rebuli et al., 2014).

To assess further the potential of the “low BPA” doses to induce changes at the molecular level and characterize the dose response of such changes, here we evaluated the effects of this BPA dose range on epigenetic, estrogenic, and other molecular pathways in the prostate and female mammary gland. Prostate and mammary gland were selected based on the conclusions of the NTP Center for the Evaluation of Risks to Human Reproduction and the joint Food and Agriculture Organization/World Health Organization expert panels, which identified these tissues as potential targets of BPA (FAO/WHO, 2010; Shelby, 2008). Tissues were evaluated at postnatal day (PND) 4, the time when the circulating levels of aglycone (bioactive) BPA was highest (Churchwell et al., 2014), and at PND 90, the time of longest exposure in the study. Evaluation of the PND 90 uterus was also included because the response of this organ to estrogenic compounds is well characterized in adult rats. We also evaluated the effects of two high doses of BPA (100,000 and 300,000 μg/kg bw/day), which were found to induce adverse effects in multiple toxicity endpoints in this animal cohort (Delclos et al., 2014). In addition, two doses of the reference estrogen EE₂ (0.5 and 5 μg/kg bw/day) were included, to characterize further the sensitivity of the animal model to this estrogenic compound and compare its effects with those of BPA. Finally, comparison of a naïve (not gavaged) and vehicle control allowed the evaluation of the effects of the daily gavage. The status of global genomic DNA methylation and the gene expression level of DNA methyltransferases and of several receptors to which BPA has been shown to have binding affinity (ERα and β, ERRγ, and GPER) were evaluated. Since limited effects were detected in these targeted molecular endpoints, we further assessed the genome-wide gene expression levels in prostate and female mammary gland at PND 4, an age when the levels of bioactive aglycone BPA available to reach molecular targets were highest, due to the immature phase II metabolic capability of the pups (Churchwell et al., 2014; Doerge et al., 2010a).

2. Materials and methods

2.1. Animal handling procedures

Procedures involving care and handling of rats were reviewed and approved by the NCTR Institutional Animal Care and Use Committee and are thoroughly described in Delclos et al. (2014). Briefly, Sprague-Dawley rats from the NCTR breeding colony were maintained in polysulfone cages with glass water bottles and hardwood chip bedding, and provided water and a soy- and alfalfa-free feed (Purina 5K96) ad libitum. F₀ rats were placed in study housing conditions at wean (PND 21) and mated when they were 10–15 weeks-of-age. When a female was determined to be pregnant, the dam was housed individually. Starting at gestation day 6, the pregnant rats were weighed and orally gavaged daily until the beginning of parturition; naïve control dams were weighed daily, but were not gavaged. On the day after birth (day of birth = PND 0), the F₁ litters were culled randomly up to 5 pups/sex/litter. Each pup was weighed and orally gavaged daily with the same dose that its dam had received during gestation; naïve control pups were weighed daily, but were not gavaged. The gavage needle did not enter the esophagus for pups younger than PND 5. There were thirteen treatment groups: naïve (untreated control); vehicle control [0.3% carboxymethylcellulose (CMC) in water]; seven “low BPA” (2.5, 8, 25, 80, 260, 840, and 2700 μg/kg bw/day); two “high BPA” (100,000 and 300,000 μg/kg bw/day); and two reference estrogen controls (EE₂, 0.5 and 5.0 μg/kg bw/day). For the molecular endpoints reported here, ten animals per dose group (1/sex/litter, randomly assigned) were euthanized at PND 4 or 90 ± 5; these animals are littermates of animals used in the study reported by Delclos et al. (2014). One pup/sex/litter was sacrificed at PND 4, if there were 5 live same-sex littermates on the day of sacrifice, and at PND 90 ± 5, if there were 2 live same-sex littermates on the day of sacrifice. The use of a single pup/sex/litter per time point obviated the need to control for litter effects in the statistical analyses. Daily vaginal smears were collected starting on PND 85 in an attempt to necropsy the cycling females at estrus, to minimize biological variability due to the estrous cycle. A terminal vaginal smear confirmed all cycling females used in this study were sacrificed in estrus, except one in the 8 μg BPA/kg bw/day dose, which was euthanized in diestrus, and two in the 300,000 μg BPA/kg bw/day dose and each of the two EE₂ doses, which were euthanized in an intermediate stage between diestrus and estrus. PND 4 and PND 90 animals were exsanguinated by cardiac puncture and retro-orbital bleeding, respectively, after CO₂ asphyxiation. The PND 4 and 90 prostate (ventral, dorsal, and lateral lobes combined), PND 4 and 90 fifth inguinal mammary gland, and PND 90 uteri (both horns combined) were used. Tissues were immediately flash-frozen in an isopentane/dry ice slurry upon dissection and were used without pooling. Frozen PND 90 tissues were macerated to a fine powder using a mortar and pestle, while being kept frozen with liquid nitrogen. Samples were stored at −80 °C until further processing; extreme care was taken not to thaw the samples prior to addition of a lysis buffer.

2.2. Genomic DNA isolation and global genomic DNA methylation status analysis by cytosine extension assay

Genomic DNA was isolated from the PND 90 prostate, female mammary gland, and uterus using a DNeasy Blood & Tissue Mini kit, following the manufacturer’s instructions (Qiagen, Valencia, CA). Frozen tissue powder (20–30 mg) was homogenized in lysis buffer using a pellet pestle grinder (Kimble Chase Kontes, Vineland, NJ). Approximately 2 μg of genomic DNA digested with HpaII restriction endonuclease (New England Biolabs, Beverly, MA) were evaluated using the [³H]dCTP extension assay described in Pogribny et al. (1999).

2.3 Quantification of cytosine and 5-methylcytosine by liquid chromatography tandem mass spectrometry (LC-MS/MS)

The quantification of cytosine and 5-methylcytosine was determined using 2 μg of genomic DNA dissolved in 20 μL of Tris-EDTA buffer, pH 8.0, by LC-MS/MS as described in Pogribny et al. (2013). The percentage of methylation was calculated as the quotient between the number of moles of 5-methylcytosine and the sum of the number of moles of cytosine and 5-methylcytosine in each sample.

2.4. Total RNA isolation and quantitative real-time reverse-transcription polymerase chain reaction (qRT-PCR)

Total RNA was isolated from the prostate and uterus using a RNeasy Mini kit and from the female mammary gland using a RNeasy Lipid Tissue Mini kit, following the manufacturer’s instructions (both Qiagen). The entire frozen PND 4 tissues or 10–20 mg of frozen PND 90 tissue powder were individually homogenized using a 5 mm stainless steel bead and a TissueLyser II (Qiagen) for 2 × 2 min at 20 Hz. The uterus homogenate was further digested with proteinase K prior to RNA isolation. RNA was treated with DNase I in-column. The RNA purity and concentration of all samples were assessed using a Nanodrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA) and the RNA integrity was assessed in randomly selected samples using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA). All RNA samples had a 260/280 nm absorbance ratio greater than 2.0 and the RNA integrity number (RIN) was 9.63 ± 0.46 [mean ± standard deviation (SD); n = 288], confirming that the RNA samples were of high quality (Schroeder et al., 2006). One microgram of total RNA was reverse-transcribed using random hexamer primers and an Advantage RT-for-PCR kit (Clontech, Palo Alto, CA), following the manufacturer’s protocol. The cDNA was diluted 1:40 with nuclease-free water and used as a template in the qRT-PCR assays, as described in Camacho et al. (2011). Supplementary Table S1 lists the TaqMan assays (Applied Biosystems, Foster City, CA) used and the range of threshold cycle (Ct) values. Each cDNA sample was run in duplicate. No-template controls, in which the cDNA sample was replaced by an equal volume of nuclease-free water, were included in duplicate in each plate and found to be negative in all cases. Gapdh (NM_017008) was used as the endogenous control gene for all tissues, except PND 4 mammary gland because its expression was not stable with these samples (data not shown); instead, Mrlp19 (NM_001029898) was used. At least two independent qRT-PCRs were performed to quantify the expression level of the endogenous control genes and the average expression was used to normalize the expression of each gene of interest. Data were analyzed using the ΔΔCt method (Livak and Schmittgen, 2001).

2.5. Gene expression analysis using microarray technology

An Agilent rat whole genome 4x44k (v3) microarray platform (Agilent Technologies) was used to determine the gene expression profiles in the PND 4 prostate and female mammary glands. Seven RNA samples per dose group were used, which were randomly selected from the 10/sex/dose group that were available; the RIN range was 9.7–10 and 9.2–10 for the PND 4 prostate and female mammary gland RNA samples, respectively. The sample size was chosen based on a priori power calculations that showed that a sample size of 7/dose group would provide greater than 80% statistical power to detect a 1.5-fold change in gene expression level versus vehicle control under the current experimental error. The sample labeling and microarray process were performed according to Agilent “One-Color Microarray-Based Gene Expression Analysis” protocol (Version 5.5). Briefly, 500 ng of total RNA were reverse-transcribed and amplified to Cy3-labeled cRNA. A total of 1.6 μg of labeled cRNA from each sample was used to hybridize to Agilent microarray for 17 hrs at 65 °C. The hybridized slides were washed in Washing buffer I at room temperature and Washing buffer II at 37 °C before being scanned in an Agilent G2505C DNA Microarray scanner (Agilent Technologies) at 5 μm resolution. The resulting images were analyzed by using the Agilent Feature Extraction Software (version 10.7) to determine the Cy3 fluorescence intensity of all gene spots (features) on each array. The raw data were uploaded into the ArrayTrack database (Fang et al., 2009) and normalized using quantile scaling normalization before further statistical analysis.

2.6. Statistical analyses

The qRT-PCR and global genomic DNA methylation data are presented as mean ± standard error of the mean (SEM). Samples from male and female rats and from different tissues and PNDs were analyzed separately. Outliers in these assays were identified as observations that fell outside of a pre-defined boundary (25% quantile minus 2 times the range between 25–75% quantiles and 75% quantile plus 2 times that range) for log2 transformed data. Five and fifteen samples (out of 390) were identified as outliers in qRT-PCR and global genomic DNA methylation assays, respectively; since their exclusion did not change the statistical outcome, the data presented include all samples analyzed (n = 10/dose/sex/PND). One-way ANOVA followed by a Dunnett’s post-hoc test was used to detect statistically significant differences between each dose group, including the naïve, and the vehicle control. A two-sided p-value < 0.05 was considered statistically significant.

The microarray data were transformed to transcript level and the technical replicates were averaged before quantile normalization. The correlation between technical replicates was 0.985, indicating a high consistency and good experimental quality. A one-way ANOVA was used to identify the overall statistically significant differentially expressed genes. A p-value < 0.005 was used to filter the genes with no overall changes. False discovery rate (FDR) adjustment (Benjamini and Hochberg, 1995; Storey, 2003) was applied to identify the genes with a FDR under 10%. The genes identified to be significantly modulated by one-way ANOVA were further analyzed by a Dunnett’s post-hoc test to identify the statistically significant differences between each dose group and the vehicle control; a p-value < 0.05 was considered statistically significant. Principal component analysis (PCA) and hierarchical clustering were performed. Categorization of the gene ontology (GO) biological process of genes modulated significantly versus vehicle control was performed using the GOFFA (Gene Ontology for Functional Analysis) tool of ArrayTrack (Sun et al., 2006) and further confirmed using the PANTHER (protein annotation through evolutionary relationship) classification system (Mi et al., 2013). In addition, the Gene Ontology enRIchment anaLysis and visuaLizAtion tool (Eden et al., 2009) was used for enrichment analysis. The p-value and the corresponding FDR q-value of each mapped GO term were calculated; enriched GO terms with a p-value < 0.001 and more than three representative genes in the dataset were considered to be significant. One sample from the 300,000 μg BPA/kg bw/day dose group (mammary gland # 8110; arrow in Fig. 2) was consistently identified as an outlier, but its exclusion did not change the results; hence, the data presented include all samples analyzed (n = 7/dose/tissue).

Fig. 2.

Principal component analysis of the genes significantly (one-way ANOVA p-value < 0.005) modulated by BPA or EE₂versus vehicle control, in the prostate and female mammary gland of Sprague-Dawley rats treated by oral gavage daily with vehicle (0.3% CMC in water), BPA (2.5, 8, 25, 80, 260, 840, 2700, 100,000, or 300,000 μg/kg bw/day), or EE₂ (0.5 or 5.0 μg/kg bw/day) from gestation day 6 until PND 4. n = 7/dose group/tissue. The arrow indicates a statistical outlier.

3. Results

3.1. Global genomic DNA methylation status at PND 90 and expression of DNA methyltransferase coding genes at PND 4 and 90

The effect of BPA and EE₂ on the global genomic DNA methylation status in the PND 90 prostate, female mammary gland, and uterus was determined by a cytosine extension assay (Table 1). The one-way ANOVA suggested that there was a global genomic DNA methylation effect in the PND 90 uterus (p-value = 0.021); however, pairwise comparisons using the Dunnett’s test did not indicate a significant effect for any individual dose group versus the tissue-matched vehicle control. No other statistically significant differences were found. Similarly, direct quantification of 5-methylcytosine by LC-MS/MS DNA did not indicate a difference between dose groups (Table 1). We assessed further the effect of BPA and EE₂ on the expression of the DNA methyltransferase-coding genes (Dnmt1, Dnmt3a, and Dnmt3b) in PND 4 prostate and female mammary gland and in PND 90 prostate, female mammary gland, and uterus (Supplementary Tables S2–S4). The only significant effect observed was on the expression of Dnmt1 in the PND 90 female mammary gland (one-way ANOVA p-value = 0.028); however, pairwise comparisons to the vehicle control using the Dunnett’s test failed to identify a significant effect on any individual dose group.

Table 1.

Status of global genomic DNA methylation in the prostate, female mammary gland, and uterus of PND 90 Sprague-Dawley rats untreated (naïve control) or treated by oral gavage daily with vehicle (0.3% CMC in water), BPA (2.5, 8, 25, 80, 260, 840, 2700, 100,000, or 300,000 μg/kg bw/day), or EE₂ (0.5 or 5.0 μg/kg bw/day) from gestation day 6 until PND 90 ± 5. The decays per minute (cytosine extension method) or the % 5-methylcytosine (LC-MS/MS method) are shown. Data shown as mean ± SEM, n = 10/dose group/tissue.

Assay and tissue	Dose (μg/kg bw/day)
	Naïve	Vehicle	BPA 2.5	BPA 8	BPA 25	BPA 80	BPA 260	BPA 840	BPA 2700	BPA 100K	BPA 300K	EE2 0.5	EE2 5.0
Cytosine extension method (decays per minute)
Prostate	31,965 ± 1020	33,158 ± 1090	36,505 ± 2903	32,163 ± 1095	30,885 ± 1429	31,667 ± 1872	33,030 ± 996	34,656 ± 897	38,494 ± 2266	35,226 ± 1385	37,137 ± 1848	35,414 ± 1445	33,638 ± 1715
Female mammary gland	5604 ± 122	5441 ± 92	5106 ± 157	5343 ± 135	5397 ± 228	5461 ± 220	5522 ± 113	5256 ± 72	5398 ± 155	5498 ± 147	5367 ± 102	5399 ± 124	5428 ± 161
Uterus	42,260 ± 1528	41,791 ± 2196	44,626 ± 2772	39,373 ± 1762	43,961 ± 1849	47,972 ± 3740	42,294 ± 1859	47,482 ± 1677	44,290 ± 1813	43,384 ± 2009	44,575 ± 2297	39,596 ± 1827	36,620 ± 1879
LC-MS/MS method (% 5-methylcytosine)
Uterus	2.50 ± 0.04	2.51 ± 0.03	2.50 ± 0.04	2.53 ± 0.03	2.48 ± 0.02	2.54 ± 0.02	2.53 ± 0.03	2.54 ± 0.03	2.49 ± 0.04	2.52 ± 0.04	2.50 ± 0.03	2.49 ± 0.03	2.50 ± 0.03

3.2. Expression of estrogen receptor and estrogen receptor-related receptor genes at PND 4 and 90

The effect of BPA and EE₂ on the expression level of the genes encoding the nuclear ERs (Esr1 and Esr2), G-protein-coupled ER (Gper), and members of the ERR family (Esrra, Esrrb, and Esrrg) was assessed in the prostate and female mammary glands at PND 4 and PND 90, and in the uterus at PND 90. Supplementary Tables S5 and S6 show the prostate and female mammary gland data, respectively, and Fig. 1 shows the uterus data. There were no differences between the naïve and vehicle controls in the expression level of any of the genes analyzed, at either PND or in any tissue. One-way ANOVA identified a significant effect in the expression of Esrrg in PND 4 female mammary gland (p-value = 0.015) and in the expression of Esrra in PND 4 prostate (p-value = 0.045), but pairwise comparisons to the vehicle control using the Dunnett’s test did not identify any significant differences in individual dose groups. In the PND 90 prostate, the expression of Esr1 was down-regulated 1.4-fold (p-value = 0.048) by 840 μg BPA/kg bw/day versus the vehicle control, but not by any other dose. In the PND 90 female mammary gland, Esrrb was up-regulated 3-fold by the low dose of EE₂ (p-value = 0.013) and Esr2 was up-regulated about 2.5-fold by both doses of EE₂versus the vehicle control (p-value = 0.001, Supplementary Table S6). Esrrg was up-regulated about 2.5-fold by both doses of EE₂ in the PND 90 uterus (p-value = 0.003 and 0.006 for 0.5 and 5 μg EE₂/kg bw/day, respectively, Fig. 1).

Fig. 1.

Fold-change versus vehicle control of the expression level of ER-, GPER-, and ERR-coding or estrogen-responsive genes in the uterus of Sprague-Dawley rats untreated (naïve control, NC) or treated by oral gavage daily with vehicle (0.3% CMC in water), BPA (2.5, 8, 25, 80, 260, 840, 2700, 100,000, or 300,000 μg/kg bw/day), or EE₂ (0.5 or 5.0 μg/kg bw/day) from gestation day 6 until PND 90 ± 5. Data shown as mean ± SEM, n = 10/dose group, except Esr2 assay, where n = 6–9 due to undetermined Ct values. * indicates one-way ANOVA p-value < 0.05 and Dunnett’s test p-value < 0.05 versus vehicle control.

3.3. Expression of estrogen-responsive genes in PND 90 uterus

The expression of four estrogen-responsive genes, coding complement component 3 (C3), progesterone receptor (Pgr), calbindin D9K (S100g), and vascular endothelial growth factor A (Vegfa), was assessed in the adult uterus (Fig. 1). The expression of C3 was not affected by any dose of BPA or EE₂. Both EE₂ doses up-regulated Pgr and S100g, while Vegfa was significantly up-regulated by the high EE₂. Pgr was up-regulated 4.5-fold (p-value = 0.010) and 3.9-fold (p-value = 0.025) by 0.5 and 5 μg EE₂/kg bw/day, respectively, while S100g was up-regulated 2.5-fold (p-value = 0.001) and 2.8-fold (p-value = 0.001) by 0.5 and 5 μg EE₂/kg bw/day, respectively. Vegfa was up-regulated 1.5-fold (p-value = 0.016) by 5 μg EE₂/kg bw/day versus vehicle control. No significant differences were detected between the naïve or BPA dose groups and the vehicle control.

3.4. Genome-wide gene expression in PND 4 prostate and female mammary glands

Since limited effects were detected at the targeted assays described above, to characterize further the effect of BPA (and EE₂) on the gene expression of prostate and female mammary glands, genome-wide gene expression was analyzed by DNA microarray technology. Tissues from PND 4 animals were used, since the internal concentration of aglycone (bioactive) BPA was highest due to the immature phase II metabolic capability of the pups (Churchwell et al., 2014; Doerge et al., 2010a) and due to the potential for developmental disruption at this more sensitive age.

Seven samples per dose group were used, based on a priori statistical power analysis. To reduce the number of samples, but still allow the characterization of the dose–response curve in the “low BPA” dose range, four “low BPA” doses were used (2.5, 25, 260, and 2700 μg BPA/kg bw/day), along with the vehicle control, the highest dose of BPA (300,000 μg BPA/kg bw/day), and the two doses of EE₂ (0.5 and 5.0 μg EE₂/kg bw/day). To verify the reproducibility of the microarray data and characterize further the dose–response curve of the gene expression changes observed in the “low BPA” dose range, qRT-PCRs were performed; ten samples per dose group from all thirteen dose groups available in the study were used in these assays. As shown in Figs. 3 and 4 (columns A and C), there was good correlation between the microarray and qRT-PCR data.

Fig. 3.

Comparison of the fold-change versus vehicle control of the expression level of selected genes assessed by genome-wide microarray and qRT-PCR in the prostate Sprague-Dawley rats untreated (naïve control, NC) or treated by oral gavage daily with vehicle (0.3% CMC in water), BPA (2.5, 8, 25, 80, 260, 840, 2700, 100,000, or 300,000 μg/kg bw/day), or EE₂ (0.5 or 5.0 μg/kg bw/day) from gestation day 6 until PND 4. Columns A and C compare microarray (gray bars) and qRT-PCR (black bars) data in the same subset of samples (n = 7/dose group). Columns B and D show qRT-PCR data using all samples (n = 10/dose group, except Gzm where n = 5–7, Lefty2 and Slpil2 where n = 6–7, and S100g where n = 4–7 due to undetermined Ct values). Data shown as mean ± SEM.

Fig. 4.

Comparison of the fold-change versus vehicle control of the expression level of selected genes assessed by genome-wide microarray and qRT-PCR in the female mammary gland of Sprague-Dawley rats untreated (naïve control, NC) or treated by oral gavage daily with vehicle (0.3% CMC in water), BPA (2.5, 8, 25, 80, 260, 840, 2700, 100,000, or 300,000 μg/kg bw/day), or EE₂ (0.5 or 5.0 μg/kg bw/day) from gestation day 6 until PND 4. Columns A and C compare microarray (gray bars) and qRT-PCR (black bars) data in the same subset of samples (n = 7/dose group). Columns B and D show qRT-PCR data using all samples (n = 10/dose group, except Ccl5 where n = 6–7/dose group due to a computer failure). Data shown as mean ± SEM.

One-way ANOVA (p-value < 0.005) identified 218 genes differentially modulated in the prostate, 8 of which had a FDR below 10%, versus tissue-matched vehicle control (Supplementary Table S7). In the female mammary gland, the expression of 257 genes was differentially modulated, 18 of which had a FDR below 10%, versus tissue-matched vehicle control (Supplementary Table S8). Regardless of the statistical method used, the high EE₂ dose modulated the largest number of genes both in the prostate and female mammary gland, followed by the low EE₂ dose. Consistently, PCA using the significantly modulated genes showed a clear separation of the high EE₂ from the vehicle control, as did the low EE₂ samples (Fig. 2 and Supplementary Fig. S1). In addition, the “high BPA” samples tended to segregate from the vehicle control in the female mammary gland. No obvious separation was observed among the “low BPA” dose groups and the vehicle control.

Tables 2 and 3 list the genes significantly modulated at least 1.5-fold by BPA or EE₂versus vehicle control in PND 4 prostate and female mammary gland, respectively, and their gene ontology process; the complete list of genes is shown in Supplementary Tables S9 (prostate) and S10 (female mammary gland). The majority of these genes were involved in cell proliferation, cell death, cell adhesion, cell signaling, and metabolic processes in both tissues. In addition, genes involved in inflammation and immune response and in oxidation-reduction processes were modulated in the mammary gland. GO term enrichment analysis in the female mammary gland confirmed the over-representation of genes associated with immune system process and cytosolic calcium ion homeostasis in the low EE₂ dose group and showed an over-representation of genes involved in cellular response to estrogen stimulus in the high EE₂ dose group (Supplementary Table S11). In the prostate, the high and low EE₂ doses had an over-representation of genes involved in axis specification and in the regulation of the non-canonical Wnt signaling pathway, respectively. The only BPA doses that yielded significant GO term enrichments were the 2.5 μg BPA/kg bw/day and 25 μg BPA/kg bw/day dose groups. In the female mammary gland, these related to cellular response to stimuli and the metabolic process, respectively. In the prostate, 25 μg BPA/kg bw/day modulated more than expected genes involved in the oxidation-reduction process, while the 2.5 μg BPA/kg bw/day dose group modulated genes involved in the regulation of kinase activity and assembly of cell projection (Supplementary Table S11).

Table 2.

List of genes significantly modulated ≥ 1.5-fold by BPA or EE₂versus vehicle control in the prostate of PND 4 Sprague-Dawley rats treated by oral gavage daily with vehicle (0.3% CMC in water), BPA (2.5, 25, 260, 2700, or 300,000 μg/kg bw/day), or EE₂ (0.5 or 5.0 μg/kg bw/day) from gestation day 6 until PND 4. N = 7/dose group. Genes are grouped according to their gene ontology biological process. The fold-change versus vehicle control is shown; the p-value of the pairwise comparison is shown between parentheses if gene modulation was significant versus vehicle control (one-way ANOVA p-value < 0.005 and Dunnett’s test p-value < 0.05). (0.00) indicates a p-value < 0.01. Genes with a false discovery rate below 10% and ≥ 1.5-fold change versus vehicle control are marked with an asterisk.

Gene symbol	GenBank accession number	BPA 2.5	BPA 25	BPA 260	BPA 2700	BPA 300,000	EE2 0.5	EE2 5.0
Cell adhesion
Hapln2	NM_022285	1.06	1.04	0.98	1.02	1.02	1.16	1.60 (0.00)
Rxfp1	NM_201417	1.35	1.06	0.91	1.13	1.11	1.56	2.58 (0.00)
Cell proliferation and cell death
Axin2	NM_024355	1.17	0.92	0.80	0.89	1.06	0.91	0.61 (0.01)
Barhl1	NM_057109	1.19	0.83	1.50 (0.04)	1.01	1.09	0.97	0.86
Cp110	NM_001108501	0.99	1.04	1.15	0.84	0.90	0.67 (0.03)	1.21
Tnfsf15	NM_145765	1.64 (0.03)	1.25	1.27	0.93	1.54	1.6 (0.05)	0.98
Cell signaling and cytoskeleton
Angpt2	NM_134454	0.90	0.95	0.92	0.98	0.90	1.13	1.55 (0.01)
Kb23*	NM_001008813	1.09	0.98	1.08	1.02	1.09	1.10	1.63 (0.00)
Lefty1	NM_001109080	1.02	0.98	1.02	1.08	0.99	1.00	2.05 (0.00)
Lefty2	NM_001007556	1.00	1.07	1.18	1.03	0.97	1.23	2.52 (0.00)
Sfrp5	NM_001107591	0.84	1.05	1.11	0.93	0.71	0.63 (0.04)	1.14
V1rf4	NM_001009522	0.99	1.00	1.01	1.01	1.01	1.62 (0.00)	1.02
Ion transport
RGD1563815*	NM_001127556	0.96	0.92	1.00	1.15	1.07	2.09 (0.00)	3.33 (0.00)
Scn11a	NM_019265	0.45	0.95	1.04	0.69	0.43	0.31 (0.01)	1.11
Metabolism
Bco2	NM_001127712	2.70 (0.00)	1.79	2.55 (0.01)	2.55 (0.01)	2.57 (0.00)	2.71 (0.00)	2.46 (0.01)
Cyp4x1	NM_145675	0.93	1.62 (0.01)	1.28	1.05	1.07	0.93	1.44
Dnase2b	NM_021664	0.57	0.83	0.93	0.87	0.41	0.34 (0.01)	0.21 (0.00)
Elovl4	NM_001191796	0.87	1.08	1.18	0.83	0.8	0.43 (0.02)	1.41
Gzmb	NM_138517	1.02	1.03	0.85	1.46	0.44 (0.02)	0.49	0.73
Gzmc	NM_134332	0.87	0.75	0.66	0.68	0.38 (0.01)	0.36 (0.01)	0.42 (0.02)
Lbx2	NM_001109244	1.15	1.05	1.02	0.93	1.03	1.98 (0.00)	1.21
Mgat4c	NM_001135814	1.19	1.05	1.11	1.27	1.11	1.19	1.72 (0.00)
Pnliprp1	NM_032081	0.72	0.97	0.67	0.63	1.18	0.74	3.01 (0.03)
S100g*	NM_012521	0.85	0.82	2.83	1.06	7.11	8.73	49.09 (0.00)
Slpil2	NM_001008872	0.71	0.90	0.94	0.89	0.62	0.59	0.38 (0.00)
Other (including uncharacterized)
Ano4	NM_001106778	0.92	1.22	1.09	1.04	1.04	1.25	1.80 (0.00)
Ctxn3	NM_001134696	0.94	1.01	0.94	1.00	0.94	1.04	2.28 (0.00)
Hist1h2aa	NM_021839	1.04	0.76	0.50 (0.03)	0.72	0.80	1.27	0.58
LOC362795	BC098733	1.11	1.15	1.77 (0.04)	1.09	1.67	2.10 (0.00)	1.45
Nlgn3	NM_134336	0.46 (0.03)	0.95	1.35	0.88	0.72	0.37 (0.00)	1.19
Olr1491	NM_001000717	0.92	0.95	0.98	1.25	0.98	0.94	2.68 (0.00)
Plcxd2	NM_001134481	0.93	1.16	1.03	0.86	0.67	0.45 (0.01)	1.00
Tmem154	NM_001108553	1.17	0.67 (0.04)	1.32	0.83	0.91	0.92	1.09

Table 3.

List of genes significantly modulated ≥ 1.5-fold by BPA or EE₂versus vehicle control in the female mammary gland of PND 4 Sprague-Dawley rats treated by oral gavage daily with vehicle (0.3% CMC in water), BPA (2.5, 25, 260, 2700, or 300,000 μg/kg bw/day), or EE₂ (0.5 or 5.0 μg/kg bw/day) from gestation day 6 until PND 4. N = 7/dose group. Genes are grouped according to their gene ontology biological process. The fold-change versus vehicle control is shown; the p-value of the pairwise comparison is shown between parentheses if gene modulation was significant versus vehicle control (one-way ANOVA p-value < 0.005 and Dunnett’s test p-value < 0.05). (0.00) indicates a p-value < 0.01. Genes with a false discovery rate below 10% and ≥ 1.5-fold change versus vehicle control are marked with an asterisk.

Gene symbol	GenBank accession number	BPA 2.5	BPA 25	BPA 260	BPA 2700	BPA 300,000	EE2 0.5	EE2 5.0
Cell adhesion and cell growth
Cbln1*	NM_001109127	1.43	1.24	1.18	1.17	0.91	2.32 (0.01)	3.67 (0.00)
Lgals4	NM_012975	0.88	1.58	1.56	1.68	2.44 (0.01)	2.09 (0.03)	2.06 (0.03)
Lppr4	NM_001001508	0.82	1.91 (0.04)	0.88	0.97	0.89	0.69	0.63
Matn4	NM_001106539	0.96	0.61 (0.03)	0.79	0.71	0.93	0.76	0.51 (0.00)
Nov	NM_030868	0.86	0.89	0.91	0.79	0.84	0.89	0.60 (0.00)
Ntng1	NM_001106465	1.64	0.95	1.28	1.62	1.83 (0.04)	2.01 (0.01)	1.78 (0.05)
Omd*	NM_031817	0.90	0.75	0.83	0.93	1.10	1.12	1.87 (0.00)
Snip	NM_019378	0.86	1.08	1.03	1.13	1.35	1.32	2.17 (0.00)
Spon1*	NM_172067	0.96	0.92	0.96	0.95	1.11	1.35 (0.03)	1.94 (0.00)
Cell proliferation and cell death
Apln	NM_031612	0.91	0.86	1.03	0.84	0.76	0.62	0.47 (0.00)
Esm1	NM_022604	1.09	0.74	1.08	0.96	0.74	0.86	0.59 (0.01)
Fas	NM_139194	2.35 (0.00)	1.13	1.05	1.44	1.31	1.80 (0.04)	2.08 (0.01)
Ppm1f	NM_175755	1.14	1.16	1.06	1.05	0.90	0.92	0.67 (0.00)
Cell signaling and cytoskeleton
Kalrn	NM_032062	1.30	1.42	1.23	1.08	1.20	1.88 (0.00)	1.02
Kctd16	NM_001172155	0.74	0.84	1.12	0.85	1.52	1.20	1.96 (0.03)
Necab2	NM_133415	0.90	0.85	0.95	1.27	1.14	1.23	1.74 (0.00)
Nfe2	NM_001012224	1.08	0.99	0.97	0.93	0.92	0.97	0.65 (0.00)
Olr499	NM_001000931	0.91	0.91	0.90	0.91	1.56 (0.01)	1.02	1.01
Rab39	NM_001108148	0.78	1.05	2.36 (0.00)	0.90	0.89	0.93	0.83
Ramp3*	NM_020100	1.01	0.84	0.95	0.91	1.12	1.17	1.50 (0.00)
Rasa4	XM_002724808	1.44 (0.01)	1.21	1.13	1.38 (0.04)	1.57 (0.00)	1.47 (0.01)	1.31
Spta1	NM_001011908	1.33	1.16	1.04	1.13	0.78	0.87	0.57 (0.02)
Trhr2	NM_181364	0.80	1.13	1.02	1.00	1.12	1.95	2.06 (0.04)
Vom2r5	NM_001099462	0.73 (0.04)	0.66 (0.00)	0.66 (0.00)	0.65 (0.00)	0.65 (0.00)	0.69 (0.01)	0.65 (0.00)
Inflammation and immune response
Ccl11*	NM_019205	0.77	0.61 (0.00)	0.82	0.84	0.95	0.8	0.52 (0.00)
Ccl5	NM_031116	1.83 (0.04)	0.82	1.19	0.92	1.42	1.67	1.08
Cd163l1	NM_001106312	1.55	0.93	1.14	1.27	2.00 (0.00)	1.33	0.93
Clec2d2	NM_001085402	1.33 (0.04)	1.44 (0.01)	1.26	1.41 (0.01)	1.34 (0.04)	1.61 (0.00)	1.41 (0.01)
Il4ra*	NM_133380	1.14	0.98	1.02	1.03	1.12	1.48 (0.01)	1.71 (0.00)
Mca32	NM_021585	0.86	0.63 (0.00)	0.79	0.77	0.8	0.89	0.60 (0.00)
Orm1	NM_053288	1.00	2.83 (0.02)	1.10	2.07	0.78	1.19	0.84
Prtg	NM_001037651	1.13	0.95	1.10	1.20	1.21	1.41	1.77 (0.00)
Serpinb9	NM_001007732	1.27	1.04	1.30	1.06	1.30	1.54 (0.00)	1.40 (0.02)
Ion transport
Atp2b2	NM_012508	0.52	0.43	1.25	0.40	0.72	0.26 (0.01)	0.36
Naglt1	NM_176080	1.56 (0.00)	0.99	0.98	0.98	0.99	1.05	1.00
Scn1a	NM_030875	1.07	1.32	1.65	2.91 (0.00)	2.05 (0.04)	1.25	1.41
Slc26a1	NM_022287	1.04	1.08	1.06	0.98	0.79	0.99	0.65 (0.01)
Slc4a1	NM_012651	1.10	1.20	1.04	1.08	0.92	0.85	0.62 (0.01)
Metabolism
Ak7	NM_001108055	1.26	2.13 (0.00)	0.94	1.54	0.98	1.37	0.82
B3gnt3	NM_001106068	0.65	2.81 (0.01)	1.07	1.21	0.91	1.10	1.54
Ces1a	NM_001190375	1.01	0.82	1.08	1.05	1.46	1.83 (0.01)	1.49
Clock*	NM_021856	1.11	1.25	0.96	1.15	1.06	1.22	0.57 (0.00)
Cpeb1	NM_001106276	1.01	1.31	0.92	1.90 (0.01)	0.97	1.17	0.89
Ctrc	NM_001077649	1.25	1.05	1.11	0.98	1.36 (0.04)	1.55 (0.00)	1.26
Elovl2	NM_001109118	1.54	1.13	1.21	1.05	1.24	1.68 (0.01)	1.79 (0.00)
Gck	NM_012565	1.69 (0.00)	1.31	1.05	1.34	0.90	1.46	1.06
Ghr	NM_017094	0.99	1.84 (0.02)	1.01	1.39	0.85	1.04	0.89
Inmt	NM_001109022	0.95	0.73	0.94	0.63	2.31 (0.04)	1.28	1.46
Kel	NM_001191611	1.03	1.15	1.10	1.06	0.82	0.89	0.54 (0.00)
Nt5e*	NM_021576	0.92	0.99	1.01	0.94	1.08	1.51 (0.00)	1.73 (0.00)
RGD1565002	XM_001080634	1.11	1.12	1.01	1.13	1.04	1.29	1.53 (0.00)
S100g	NM_012521	0.67	0.72	1.33	1.41	0.73	1.63	4.30 (0.01)
Oxidation-reduction process
Akr1c14	NM_138547	1.51	0.76	0.79	0.81	1.04	1.69	2.61 (0.02)
Akr1cl1*	NM_001109900	0.90	1.38	0.76	1.07	1.14	2.49 (0.00)	4.79 (0.00)
Cyp2r1	NM_001108499	0.86	1.69 (0.04)	0.96	1.04	0.77	0.97	0.73
Dio3	NM_017210	1.05	1.10	0.92	0.85	1.71	1.91	3.64 (0.00)
LOC685171	NM_001109459	0.56 (0.01)	0.73	0.54 (0.00)	0.53 (0.00)	0.52 (0.00)	0.68	0.83
Other (including uncharacterized)
LOC302576	NM_001013959	3.84 (0.01)	1.19	0.90	0.83	1.10	0.99	2.71
LOC304027	XR_005615	0.92	1.14	0.99	1.24	0.85	0.87	0.60 (0.02)
LOC500265	XR_007267	3.74 (0.00)	1.24	1.18	1.23	1.45	1.25	1.42
LOC679566	NM_001109372	0.99	0.91	1.03	0.89	1.46	1.13	1.58 (0.02)
LOC680214	XM_001056138	1.04	1.65 (0.00)	1.03	1.04	1.05	1.06	1.00
LOC680967	XM_002730137	0.26 (0.00)	0.81	0.67	0.75	0.73	0.61	1.18
LOC685106*	XM_001062312	0.39	0.91	1.48	1.52	0.07 (0.02)	0.03 (0.00)	2.06
LOC690402	NM_001109588	0.32 (0.01)	0.34 (0.01)	0.29 (0.00)	0.38 (0.03)	0.26 (0.00)	0.25 (0.00)	0.25 (0.00)
Mansc1	NM_001109603	0.84	1.00	0.88	0.97	1.04	1.81 (0.01)	0.85
RGD1311378	NM_001106547	0.92	1.33	0.96	0.97	0.61 (0.04)	0.98	0.61 (0.04)
RGD1562378	XM_575704	1.00	1.61 (0.04)	1.07	1.07	0.77	0.95	0.82
RGD1562720	NM_001109240	0.24 (0.00)	0.30 (0.01)	0.61	0.26 (0.00)	0.26 (0.00)	0.49	0.34 (0.03)
RGD1564859	XM_347061	0.97	1.30	1.02	0.88	0.81	0.98	0.56 (0.00)
Scube1*	NM_001134884	0.92	0.90	0.93	1.00	1.07	1.43 (0.01)	2.03 (0.00)
Thsd7b	NM_001191669	0.84	1.03	0.81	1.00	0.54	0.91	0.35 (0.00)
Tmcc2	XM_223107	0.64	1.06	1.04	0.89	0.70	0.71	0.39 (0.00)
Tmem144	NM_001108551	1.03	0.92	0.91	0.91	0.97	1.54	2.88 (0.00)

Although significant gene expression changes with at least a 1.5-fold change were observed in all “low BPA” doses versus the tissue-matched control and significant GO term enrichments were observed in some “low BPA” dose groups, it should be noted that seven of the 8 (88%) prostate genes and 21 of the 27 (78%) female mammary gland genes were modulated by a single “low BPA” dose level (Tables 2 and 3). Analysis of all genes significantly modulated in the “low BPA” dose range versus vehicle control, regardless of the fold-change of the modulation, yielded a similar result, with twenty-six of the 31 (84%) genes and 42 of the 51 (82%) genes found to be modulated by a single “low BPA” dose level in the PND 4 prostate and female mammary gland, respectively (Supplementary Tables S9 and S10). The remaining genes (Bbs7, Bco2, Mdh1b, Mtfr1, and Ube2 in the prostate, and Clec2d2, LOC100302465, LOC685171, LOC690402, Ndufa10l1, Rasa4, RGD1562720, Slitrk6, and Vom2r5 in the female mammary gland) were modulated by multiple “low BPA” dose levels, but the gene expression fold-change was not correlated with the BPA dose nor was it further amplified by the high BPA and/or EE₂ doses (Tables 2 and 3 and Supplementary Tables S7 and S8). The majority of the genes modulated by multiple “low BPA” dose levels had a very low expression level; in fact, only Ube2a, Clec2d2, and Rasa4 were expressed at sufficient level to be detected by qRT-PCR. qRT-PCR failed to confirm the modulation of these three genes by the “low BPA” dose levels (Figs. 3 and 4). The difference in outcome between microarray and qRT-PCR assays may be due to the existence of multiple Ube2a transcripts (a second microarray probe did not indicate modulation of Ube2a gene, Supplementary Table S7) or the sub-optimal design of the Clec2d2 Taqman assay (which targets not only Clec2d2, but also Clec2dl1 and LOC689757, Supplementary Table S1). qRT-PCR using all study samples confirmed the statistically significant up-regulation of Rasa4 by 300,000 μg BPA/kg bw/day (1.51-fold, p-value = 0.04) and 0.5 μg EE₂/kg bw/day (1.58- fold, p-value = 0.01), but not by 2.5 or 2700 μg BPA/kg bw/day (Fig. 4).

4. Discussion

The current study assessed the effects of a wide range of oral BPA doses (2.5–300,000 μg/kg bw/day) at the molecular level in rat prostate and female mammary gland. These two tissues were selected because important data gaps regarding BPA toxicity were identified previously, including poor characterization of the molecular pathways modulated by BPA at doses below those needed to activate the ERs (FAO/WHO, 2011; FDA, 2014; Shelby, 2008). The adult rat uterus, a well-characterized estrogen-sensitive tissue, was also included. The tissues analyzed were harvested from litter-mates of animals used in the study reported by Delclos et al. (2014), which summarized the effects observed at the body and organ weight, clinical chemistry, estrous cyclicity, sperm parameters, and histopathology level, and Churchwell et al. (2014), which summarized the findings at the internal dosimetry level. In addition to the nine BPA dose groups analyzed, two negative controls (naïve and vehicle), and two reference estrogen controls (EE₂, 0.5 and 5.0 μg/kg bw/day) were also included. All together, these aspects of study design resulted in a very robust and well-controlled study.

To circumvent the poor lactational transfer of BPA due to its efficient metabolism and excretion by the dam (Doerge et al., 2010b), which would result in an attenuated exposure during the critical neonatal period, BPA was administered directly to the pups by gavage from the day after birth. The effect of this dosing procedure and of the vehicle solution was assessed by comparing a naïve (not gavaged) control with the vehicle control. No statistically significant differences were detected between the two negative controls in any molecular assay, regardless of tissue or PND. Together with the lack of differences between naïve and vehicle controls reported in this animal cohort by Delclos et al. (2014), our data support further the lack of effect of the daily gavage on the endpoints analyzed in this study.

We first sought to determine the effects of BPA (and EE₂) on targeted molecular endpoints, selected based on previous reports. These included global genomic DNA methylation and expression levels of DNA methyltransferases and BPA-binding receptors coding genes. We analyzed further the effects of BPA on the genome-wide gene expression level of PND 4 prostate and mammary gland by microarray technology. As reported in Churchwell et al. (2014), despite the meticulous care taken to minimize exposure of the study animals to exogenous sources of BPA during housing and dosing, both naïve and vehicle controls had detectable levels of circulating BPA glucuronide, a metabolite of BPA produced in vivo. The mean serum levels of BPA glucuronide in the naïve and vehicle controls and in animals dosed with 2.5 μg BPA/kg bw/day were undistinguishable, thus precluding the assessment of treatment-related effects of the lowest BPA dose in the study versus the negative controls; however, the margin of exposure between animals treated with >8 μg BPA/kg bw/day and vehicle controls was sufficient to interpret any effect in the context of the BPA treatment (Delclos et al., 2014; Rebuli et al., 2014).

We assessed the effect of BPA and EE₂ on the PND 90 global genomic DNA methylation using both an enzymatic and a non-enzymatic method. The non-enzymatic LC-MS/MS method measures the absolute DNA methylation status and is considered as a “gold” standard in global DNA methylation analyses. The enzymatic cytosine extension method detects the extent of the methylation status of the cytosine residues within the recognition sequence of the HpaII restriction endonuclease, which are randomly distributed across the genome (Rocha et al., 2010). Regardless of the method, we did not detect a statistically significant PND 90 global genomic DNA methylation change in either BPA- or EE₂-treated animals versus the vehicle control. To the best of our knowledge, this is the first study examining the effect of BPA on the global genomic DNA methylation in the prostate; however, we cannot exclude the possibility of local region-specific DNA methylation changes induced by BPA- or EE₂-treatment. Ho et al. (2006) and Tang et al. (2012) reported that the subcutaneous injection of Zivic-Miller Sprague-Dawley rats with 10 μg BPA/kg bw/day in the perinatal period induced changes in the methylation status of specific gene promoters in the dorsal prostate. These authors further reported the modulation of the gene expression level of Dnmt3a/b in the dorsal prostate by BPA, an effect not observed in the current study. The effects of BPA in the ventral and lateral prostate lobes were not reported by Ho et al. (2006) or Tang et al. (2012). The use of the combined ventral and dorsolateral lobes in the current study versus the use of the dorsal lobe by Tang et al. (2012) may explain the differences observed, since the prostatic lobes exhibit different sensitivity to different estrogenic compounds (Ofner et al., 1992; Tam et al., 2008) and the use of combined lobes would dilute a lobe-specific effect. In addition, the use of a younger age in our study, when the rat prostate is still undergoing arborization and the different lobes are becoming anatomically distinct (Hayashi et al., 1991) could contribute to the difference in findings. Recently, the effect of BPA on the epigenome of the Harlan Wistar-Furth rat female whole mammary gland was reported by Dhimolea et al. (2014). These authors demonstrated transient region-specific DNA methylation changes induced by BPA treatment; however, similar to our results, they did not identify any significant hypo- or hypermethylation trends in genomic DNA methylation.

Several treatment-related effects were observed in the EE₂ dose groups at the gene expression level. The expression of previously proposed molecular biomarkers of estrogenic exposure, including Vegfa, Pgr, and S100g (An et al., 2002; Diel et al., 2002; Jung et al., 2012; Schmidt et al., 2006) were up-regulated by one or both EE₂ doses in the PND 90 uterus. In addition, S100g was up-regulated by EE₂ at PND 4 in both the prostate and female mammary glands. One or both doses of EE₂ further modulated the expression of ER or ERR-coding genes in the PND 90 uterus and female mammary gland. PCA of the statistically significant microarray data showed a segregation of the high and low EE₂ from the vehicle control at PND 4 in both prostate and female mammary glands, and GO term enrichment analysis showed an over-representation of genes involved in response to estrogen stimulus in the genes modulated by the high EE₂ in the female mammary gland. Of the genes modulated by one or both doses of EE₂ in the PND 4 tissues analyzed, several are known estrogen-responsive genes. Genes up-regulated by EE₂ in the current study included Angpt2 in the prostate and Akr1c13, Il4ra, Ramp3, Serpinb9, Spon1, and Trhr2 in the female mammary gland. These genes were previously shown to be up-regulated by estrogen (Aenlle et al., 2009; Hong et al., 2004; Jiang et al., 2008; Kimura et al., 1994; Rincón-Rodríguez et al., 2013; Ye et al., 2002). Conversely, Apln and Clock were found to be down-regulated by EE₂ in the female mammary gland in the current study, in agreement with previous reports (Damdimopoulou et al., 2011; Heneweer et al., 2007). In all cases, the direction of the gene expression change was consistent with those reported previously. Taken together, our data confirmed the sensitivity of the NCTR Sprague-Dawley rat to the reference estrogen EE₂, consistent with our previous studies (Delclos et al., 2014; NTP (National Toxicology Program), 2010a, 2010b; Rebuli et al., 2014).

Both 300,000 μg BPA/kg bw/day and EE₂ modulated the expression level of Dclk3, Gnai2, Gzmc, LOC687090, Map4k5, and Ppic in the PND 4 prostate and Coro1c, Ctrc, Lgals4, LOC685106, Ntng1, and RGD1311378 in the PND 4 female mammary gland, and the direction of the change versus the vehicle control was the same for both compounds in all cases. In addition, PCA of the statistically significant microarray data suggested that the effect of the high BPA dose level at the genome-wide gene expression overlapped to some extent with that of the low dose EE₂, in particular in the female mammary gland. Our data thus suggest that the high BPA dose induced estrogenic effects that were more pronounced in the female than in the male tissue. These data are consistent with the overall conclusion of the 90-day subchronic toxicology study, in which 300,000 μg BPA/kg bw/day had a clear effect in multiple endpoints in female rats, some of which overlapped with the EE₂ dose groups, while the effects of this high BPA dose level in male rats were much more restricted (Delclos et al., 2014). The internal dosimetry data collected in littermates of animals used in the current study further support that treatment with the 300,000 μg BPA/kg bw/day results in levels of circulating bioactive, aglycone BPA sufficient to activate the ER and mediate estrogenic effects (Churchwell et al., 2014).

Several rodent in vivo studies reported the modulation of the gene expression level by BPA doses within or below the “low BPA” dose range as defined in the present study, including in the rat prostate and female mammary gland. These include the above mentioned Ho et al. (2006) and Tang et al. (2012) studies that report gene expression modulation in the prostate of Sprague-Dawley rats treated subcutaneously with 10 μg BPA/kg bw/day on PND 1, 3, and 5. The modulation of female mammary gland gene expression in rodent models was reported by Moral et al. (2008; Sprague-Dawley rats, 25 and 250 μg BPA/kg bw/day, daily gavage, from GD 10 to 21), Wadia et al. (2013; ERα (+/+) and ERα (−/−) C57BL/6 mice, 250 ng BPA/kg BW/day, subcutaneous osmotic pump, from GD 8 to 19), and Dhimolea et al. (2014; Wistar-Furth rats, 250 μg BPA/kg bw/day, subcutaneous osmotic pump, from GD 9 to PND 1). Similarly, in our study, several genes were found to be modulated both in PND 4 prostate and female mammary glands in the “low BPA” dose range versus the vehicle control. Of these, over 75% were affected only by a single BPA dose level, despite the close spacing between consecutive “low BPA” dose levels (~10-fold in microarray and ~3.3-fold in the qRT-PCR assays). This was observed for both the prostate and female mammary gland. Only five genes in the prostate and eight genes in the female mammary gland were modulated by multiple BPA dose levels; however, the fold-change of these genes was similar across the wide range of doses tested, despite the 120,000-fold difference between the lowest and the highest dose of BPA tested, reducing our confidence that these observations were related to the BPA treatment. The sample size used in the current study was selected to provide greater than 80% statistical power to detect a 1.5-fold change of a treated group versus the vehicle control, as determined by a priori power calculation analysis, although it is possible that a larger sample size may have resulted in the detection of additional statistically significant findings, including of dose-related changes in the “low BPA” dose range. Due to the gaps between the 2700, 100,000, and 300,000 μg BPA/kg bw/day dose groups in the present study, we cannot exclude the possibility that BPA induced further effects in this dose range that could have gone undetected; however, these doses represent approximately 5000 to 500,000 times the mean human dose of BPA (Lakind and Naiman, 2011), and thus would bear little significance in the context of human exposure. Our observation that, although genes were found to be statistically modulated in all “low BPA” dose groups versus vehicle control, none showed a clear dose–response, emphasizes the importance of incorporating multiple and appropriately spaced dose levels in the study design to better interpret the biological significance of the statistical findings.

PND 4 tissues were used to assess the effects of BPA at the genome-wide gene expression level because the perinatal period is usually recognized as a susceptible window to disruption (Macon and Fenton, 2013; Prins and Ho, 2010) and the levels of circulating bioactive aglycone BPA were highest at this age (Churchwell et al., 2014). Indeed, the PND 4 gene expression profiles were modulated in the prostate and female mammary gland by the EE₂ and high BPA dose groups, but no treatment-related changes were observed in the “low BPA” dose range. It is possible that longer exposures to “low BPA” could result in such changes. This hypothesis is being assessed currently in the NCTR Sprague-Dawley rat by other investigators within the Consortium Linking Academic and Regulatory Insights on BPA Toxicity (CLARITY-BPA). These ongoing studies will provide further insight into the molecular, functional, and morphological effects of a wide-range of oral BPA upon longer exposure periods, including at PND 21, PND 90, 6 months, and 1 year of age (Schug et al., 2013).

5. Conclusions

Our study reports that both 0.5 and 5.0 μg EE₂/kg bw/day modulated the expression level of several genes, including of known estrogenic-responsive genes, in the prostate, female mammary gland, and uterus of NCTR Sprague-Dawley rats. In addition, our PND 4 microarray data showed that 300,000 μg BPA/kg bw/day induced effects that partially overlapped with those of EE₂. Microarray data showed further the modulation of several genes in the “low BPA” dose range in both the prostate and female mammary gland, but the lack of a dose-response reduces the likelihood that these effects were causally linked to the treatment. These results are consistent with the toxicity outcomes reported previously for this animal cohort (Delclos et al., 2014). Endpoints complementary to the ones reported here are being assessed currently in NCTR Sprague-Dawley rats exposed to BPA for longer exposure periods by other investigators participating in the CLARITY-BPA study (Schug et al., 2013). Those data will help to elucidate further the effects of a wide-range of oral BPA at the epigenetic and gene expression level in this rat model and integrate the molecular findings in the context of structural and functional effects.

Abbreviations

BPA

bisphenol A

bw

body weight

C3

complement component 3-coding gene

CMC

carboxymethylcellulose

Ct

threshold cycle

Dnmt

DNA methyltransferase-coding gene

EE₂

ethinyl estradiol

ER

estrogen receptor

Esr

estrogen receptor-coding gene

FDR

false discovery rate

GO

gene ontology

GPER

G-protein-coupled ER

LC-MS/MS

liquid chromatography tandem mass spectrometry

NCTR

National Center for Toxicological Research

NTP

National Toxicology Program

PCA

principal component analysis

PND

postnatal day

Pgr

progesterone receptor-coding gene

qRT-PCR

quantitative real-time reverse-transcription polymerase chain reaction

RIN

RNA integrity number

S100g

calbindin D9K-coding gene

Vegfa

vascular endothelial growth factor A-coding gene

Appendix: Supplementary material

Supplementary data to this article can be found online at doi:10.1016/j.fct.2015.04.009.