DNA methylation and transcriptome aberrations mediated by ERα in mouse seminal vesicles following developmental DES exposure

Significance

Early developmental exposure to endocrine active compounds causes late-stage effects and alterations in the reproductive tract of adult mice. Unexpectedly, estrogen receptor alpha (ERα) plays a pivotal role in mediating these developmental effects. As a model outcome from these developmental effects, we present transcriptome and DNA methylation profiling of the seminal vesicles (SVs) following neonatal diethylstilbestrol (DES) exposure. ERα mediates transcriptome aberrations in SVs of adult mice that impact developmental reprogramming at adulthood. DNA methylation dynamically changes during development, and methylation is greater in ERα knockout mice compared with wild type. Expression levels of DES-altered genes are associated with their DNA methylation status. These findings provide unique evidence for understanding the developmental actions and mechanisms of endocrine-disrupting chemicals in human health.

Abstract

Early transient developmental exposure to an endocrine active compound, diethylstilbestrol (DES), a synthetic estrogen, causes late-stage effects in the reproductive tract of adult mice. Estrogen receptor alpha (ERα) plays a role in mediating these developmental effects. However, the developmental mechanism is not well known in male tissues. Here, we present genome-wide transcriptome and DNA methylation profiling of the seminal vesicles (SVs) during normal development and after DES exposure. ERα mediates aberrations of the mRNA transcriptome in SVs of adult mice following neonatal DES exposure. This developmental exposure impacts differential diseases between male (SVs) and female (uterus) tissues when mice reach adulthood due to most DES-altered genes that appear to be tissue specific during mouse development. Certain estrogen-responsive gene changes in SVs are cell-type specific. DNA methylation dynamically changes during development in the SVs of wild-type (WT) and ERα-knockout (αERKO) mice, which increases both the loss and gain of differentially methylated regions (DMRs). There are more gains of DMRs in αERKO compared with WT. Interestingly, the methylation changes between the two genotypes are in different genomic loci. Additionally, the expression levels of a subset of DES-altered genes are associated with their DNA methylation status following developmental DES exposure. Taken together, these findings provide an important basis for understanding the molecular and cellular mechanism of endocrine-disrupting chemicals (EDCs), such as DES, during development in the male mouse tissues. This unique evidence contributes to our understanding of developmental actions of EDCs in human health.

Normal development and function of reproductive tract organs are dependent on a highly sensitive and appropriate response to hormone signaling (1). Developmental exposure to estrogenic and antiandrogenic endocrine disrupting chemicals (EDCs) is associated with reproductive dysfunctions in adulthood (2). EDCs include natural hormones, synthetic estrogens, phytoestrogens, plasticizers, and pesticides whose activities are thought to be mediated through the estrogen receptors (ERs) (3). ERs, including ERα and ERβ, are members of a large superfamily of nuclear receptors and can act as ligand-inducible transcription factors (TFs) (4). The main mechanism involves ER directly bound to DNA estrogen-response elements (EREs) of target genes to regulate gene expression (5, 6). EDC’s activities have been shown to be mainly through the ERs to regulate many ER-dependent genes (7).

Developmental origins of adult reproductive disease are associated with EDC exposure in both human and animal studies (2). Diethylstilbestrol (DES) is a potent synthetic estrogen which was historically prescribed to pregnant women to prevent miscarriage (8, 9). Usage was suspended, however, when it was determined that maternal exposure caused vaginal and breast tumors in young women (10, 11). Several pathologies associated with developmental exposure to DES in humans have been replicated experimentally in mice (12, 13). A mouse model of neonatal DES exposure was widely used to study the effects of EDCs on the reproductive organs with emphasis on critical developmental periods (14).

ER-knockout (ERKO) mouse studies demonstrate that ERα plays a critical role in mediating the toxicological effects of neonatal DES exposure in female and male reproductive organs (15, 16). Some of the DES toxicity effects result in adult onset atrophy of the seminal vesicles (SVs) and aberrant gene expression (15). Recently, we reported that neonatal DES exposure induces SV toxicity in adult mice and it is primarily mediated through ERα, which alters gene expression of Svs4 (seminal vesicle secretory protein IV) and causes aberrant expression of a uterine protein Ltf (lactoferrin) (17). DNA methylation is a well-characterized epigenetic modification and is important for gene regulation, transcriptional silencing, development, and tumorigenesis (18–20). Our studies indicated that DES alters the DNA methylation status in certain CpGs of the Svs4 and Ltf gene promoters and the methylation status correlates with the levels of gene expression (21). Based on our previous findings, we hypothesized that neonatal DES exposure alters the transcriptome and DNA methylation profiles in the mouse SVs and ERα plays a role in these alterations.

To test our hypothesis, we profiled genome-wide transcriptome and DNA methylation of the wild-type (WT) and ERα-knockout (αERKO) mouse SVs following neonatal DES exposure. We performed similar experiments on uterine samples as a comparison for tissue specificity. We investigated the changes of DNA methylation caused by DES exposure during development in the mouse SVs. Furthermore, we explored the association of DES-altered transcriptome expression and DNA methylation status in the SVs of adult mice.

Results

αERKO Mice Are Resistant to the Developmental Effects of DES in the Mouse SV.

Our previous studies showed that neonatal DES exposure reduced SV weight of WT adult mice and αERKO mice exhibited resistance to this developmental effect (15). To explore the role of ERα in the mouse SV during development, we collected the SVs from WT and αERKO mice at 3 wk (before puberty), 5 wk (puberty), and 10 wk (adult) after neonatal DES exposure (Fig. 1A). In the WT SVs, neonatal DES exposure reduced SV weight by over 60% during development compared with the vehicle (veh) at all three time points (Fig. 1B). In contrast, this reduction in SV weight did not occur in αERKO mice after DES exposure (Fig. 1B). We also found that neonatal DES exposure resulted in significant histological alterations at week 5 in the SV of WT males but not in αERKO (Fig. 1C). The histology of WT SV DES samples showed a significant increase in thickness of the smooth muscle layer. In contrast, αERKO SV DES had similar histology with the vehicle sample as shown in Fig. 1C. To examine the effects of neonatal DES exposure on the Esr1 (encodes ERα protein) and Ar [encodes androgen receptor (AR) protein] gene expression during development, we performed qPCR analysis in both WT and αERKO male mice at ages 3, 5, and 10 wk treated with either vehicle or DES. We found an increase of Esr1 and a lower amount of Ar in the WT SV during development. However, there was no significant alteration observed in either Esr1 or Ar expression following DES exposure (Fig. 1D). As expected, Esr1 was absent in all αERKO samples. These observations indicate that αERKO mice exhibit resistance to the developmental effects of DES exposure on mouse SVs during development (before adulthood).

Fig. 1.

Experimental designs and changes in SV weight and gene expression during development after neonatal exposure. (A) Timeline for neonatal treatment (corn oil vehicle or DES, 2 μg per day) and collection of tissues. (B) SV weights of male WT and αERKO mice (intact) for the three time points after neonatal treatment on days 1–5 with vehicle (veh) or DES. The mouse number in each group was n = 10–15. WT-veh was compared at each time point by two-way ANOVA using a Dunnett’s multiple comparison test, ***P < 0.001. (C) Pathology of SVs of male WT and αERKO mice at week 5 (veh or DES-treated samples) stained with hematoxylin and eosin (magnification, 10x). (D) The expression levels of Esr1 and Ar genes in SVs of WT and αERKO mice. Total RNA samples were extracted from frozen SV tissues of three individual mice for WT-veh, αERKO-veh, and αERKO-DES, or three pools (n = 5–8) for WT-DES samples. The expression levels were quantified by qPCR and normalized against the 18S housekeeping gene. Data shown represent mean fold change (±SEM) relative to SVs from week 3 WT-veh. *P < 0.05, ****P < 0.0001 by two-way ANOVA using a Tukey’s multiple comparison test.

ERα-Mediated Aberration of the Transcriptome in the SV of Adult Mice After Neonatal DES Exposure.

To investigate the role of ERα in mediating DES-altered SV gene expression, we profiled genome-wide transcriptome of adult mouse (week 10) SVs in the WT and αERKO using RNA-sequencing (RNA-Seq) and microarray analyses. For RNA-Seq, differentially expressed (DE) genes were analyzed by examining fold changes with a ±1.5-fold cutoff (Fig. 2A). As for DES’s effect on the WT SVs, a total of 2,162 DE genes were found (Fig. 2A, comparison 1). In contrast, only 20 DE genes were found in DES-treated αERKO SVs (Fig. 2A, comparison 2). As for ERα’s effect, 116 DE genes were found between the WT vehicle and αERKO vehicle SVs (Fig. 2A, comparison 3). When comparing the WT vehicle to the αERKO DES-treated SVs, 121 DE genes were found (Fig. 2A, comparison 4). Additionally, a total of 2,442 DE genes were identified when comparing the WT DES and αERKO DES samples and 2,598 genes were identified in the comparison between WT DES and αERKO veh (SI Appendix, Fig. S1A). When we compared the expression heatmap of 2,162 (1,678 induced and 484 repressed) DE genes, DES-altered genes were only in the WT SVs but were not seen in αERKO SVs (Fig. 2B).

Fig. 2.

ERα-mediated aberrations of the transcriptome in the SVs of adult mice following developmental DES exposure. (A) Differentially expressed (DE) gene analysis for week 10 SVs RNA-Seq (n = 4 for each group). DE genes from the four comparisons are shown as fold change with a cutoff of >1.5-fold, fragments per kilobase million (FPKM) >1 and q < 0.05. (B) Heatmap depicting 2,162 DE genes from comparison 1 WT-DES vs. WT-veh (1,678 induced and 484 reduced genes). (C) IPA analysis of 1,678 induced or 484 reduced genes.

To understand the biological function underlying the dramatic gene changes by DES exposure during development in the WT SVs, ingenuity pathway analysis (IPA) was performed for both DES-induced and -repressed genes. As for the 1,678 induced genes, they appeared to be related to immunological, inflammatory, and hematological diseases. There were 750 genes related to the hepatic system and 585 genes related to lymphoid tissue structure developmental function (Fig. 2C, Left). The top upstream regulators were lipopolysaccharide, TNF, IFNG, IL4, and IL1B and the pathway networks were associated with endocrine system disorders, cell death, and developmental disorders (Fig. 2C, Left). As for the 484 repressed genes, they appeared to be related to developmental disorders and metabolic disease. There were 50 genes related to organ morphology and 35 genes related to organismal developmental function. Interestingly, 18 DES-repressed genes were related to reproductive system developmental function (Fig. 2C, Right). Some major upstream regulators were XBP1, AR, and androgen, through pathway networks associated with metabolic disease and cellular function and maintenance (Fig. 2C, Right). In addition, we observed a similar transcriptome with the same samples using GeneChip, a microarray analysis. Around 68% of DE genes overlapped between RNA-Seq and GeneChip data, suggesting that the two analyses were comparable. Using chromosome view analysis, we found that those DE genes were distributed throughout all chromosomes (SI Appendix, Fig. S1B). These findings indicate that ERα mediates aberrations of the transcriptome in the SV of adult mice following DES exposure and many of the DES-altered genes are related to developmental functions and diseases.

Comparison of Altered Genes in the SV and Uterus of Adult Mice After Neonatal DES Exposure.

DES-induced toxicity in the mouse uterus has been well documented and many estrogen-response genes have been identified (22, 23). To explore whether DES exposure results in similar or tissue-specific gene expression differences, we performed RNA-Seq analysis of the uterus of adult mice following the same neonatal DES exposure as used for the SV study and then compared DES-altered gene profiles between the two tissues. In a comparison of WT vehicle and WT DES-treated samples, 425 DE genes were found in the adult uterus (Fig. 3A, Left). When overlapping this gene profile with the 2,162 SV DE gene profile, we found that 155 DES-altered genes were shared in the two tissues (SI Appendix, Table S2). Interestingly, 15 well-known estrogen-response genes were found as listed in Fig. 3A. Ingenuity pathway analysis showed the differences of the top five diseases and disorders occurring between SV and uterine genes following DES exposure (Fig. 3A, Right). Of note, 111 overlapping genes (out of 155) were related to reproductive system diseases.

Fig. 3.

Comparison of altered genes in the SVs and uterus of adult mice following neonatal DES exposure. (A) Venn diagram of the overlapping between DES-altered 2,162 SV and 425 uterus DE genes (Left). IPA analysis for the 2,007 SV, 270 uterine, and 155 overlapping DE genes (Right). (B) Gene pattern analysis for 155 overlapping genes using EPIG with the expression value of FPKM. See also SI Appendix, Table S1.

To determine whether DES may have differential effects on shared genes in the adult mouse SV and uterine tissues, we performed gene pattern analysis for the 155 overlapping genes. As shown in Fig. 3B, six patterns with a varying number of genes were identified. Most genes belong to pattern 2 and pattern 3. Pattern 2 showed genes that were induced in both SV and uterus and include genes known to be altered following DES, such as Ltf, Ccna2, and Six1. This pattern also showed gene expression that could contribute to feminization in the male with genes being expressed in the SV after DES exposure at a similar level as in the control uterus. Interestingly, in pattern 3, the well-known estrogen-response genes, Dcn, Igfbp3, Inhbb, and Tgfbi showed differential changes in the SV and uterus, which were induced in the SV but repressed in the uterus after DES exposure. These findings suggest that DES exposure impacts the development of differential pathological changes between male and female tissues and this is due to a set of DES-altered genes that appears tissue-specific during mouse development.

Differential Alteration of Estrogen-Responsive Genes During Development in the Mouse SV Following Neonatal DES Exposure.

Our previous studies showed that ERα played a role in mediating the aberrant Ltf (an estrogen-response gene) expression in the adult SV following DES exposure (17, 21). To investigate the specificity of the effects of DES exposure on estrogen-responsive genes during development in the SVs, we selected eight well-known estrogen-responsive genes found in the mouse uterus (24) from the 155 overlapping genes and performed qPCR analysis at the three developmental stages, including week 3 (before puberty), week 5 (puberty), and week 10 (adult). Ltf and C3 are normally expressed in female tissues and therefore both basal levels were lower or nondetectable in WT vehicle and αERKO vehicle SVs (Fig. 4A). However, DES strongly induced Ltf and C3 expression in WT SVs compared with vehicle at all time points, and increasing levels were observed as the mouse aged. In contrast, there was no induction observed in the αERKO SVs following DES exposure (Fig. 4A). DES exposure weakly induced the expression levels of Six1 and Padi4 from the early developmental stages (weeks 3 and 5) and levels were significantly increased when mice reached adulthood (week 10) (Fig. 4A). Igfbp3 and Krt15 had a dynamic change during development (Fig. 4A). Interestingly, Wnt4 and Krt5 had expression pattern changes associated with the developmental stage when mice reached adulthood (Fig. 4A). These data suggest that further hormonal regulation at the different developmental stages could affect expression levels of estrogen-responsive genes in the SV following DES exposure.

Fig. 4.

Differential alterations of estrogen-response genes during development in the mouse SV following neonatal DES exposure. (A) The qPCR data for developmental changes. Total RNA samples were extracted from SV frozen tissues of three individual mice for WT-veh, αERKO-veh, and αERKO-DES, or three pools (n = 5–8) for WT-DES samples. The expression levels of Ltf, C3, Six1, Padi4, Igfbp, Krt15, Wnt4, and Krt5 were quantified by qPCR and normalized to the 18S housekeeping gene. Data shown represent mean fold change (±SEM) relative to SVs from week 3 WT-veh, ***P < 0.001, ****P < 0.0001 by two-way ANOVA using a Dunnett’s multiple comparison test. (B) The qPCR data for analyzing cell-type specificity in gene changes. Total RNA was isolated from the epithelial and stromal cells of whole SV tissues in WT adult mice using LCM as described in Materials and Methods. The quantification of RNA was shown in SI Appendix, Fig. S2. The expression levels were quantified as above. Data shown represent mean fold change (±SEM) relative to SVs from the epithelial cells (Epi) or stromal cells (Strom) WT-veh, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by two-way ANOVA using a Dunnett’s multiple comparison test.

To examine the cell type specificity in changes of estrogen-responsive genes, we used laser capture microdissection (LCM) to isolate the epithelial and stromal cells from whole SV adult tissues (Fig. 4B, Left). The qPCR results showed that DES exposure significantly induced the expression levels of Ltf, Six1, and Krt5 genes in epithelial cells, but there was no change in stromal cells. C3 gene expression was induced in both cell types in the DES-treated samples. However, an induction of Igf-1 expression was only observed in SV stroma (Fig. 4B, Right). These results indicate that certain gene changes in the mouse SVs following DES exposure appear to be cell-type specific.

DNA Methylation Reprogrammed in the Mouse SV After DES Exposure.

Based on our previous findings that DNA methylation status in certain CpGs of the Svs4 and Ltf gene promoters associated with the levels of gene expression in the adult mouse SVs (21), we hypothesized that altered DNA methylation may play a role during development in the mouse SVs following DES exposure. To test this hypothesis, we performed methyl-CpG binding domain capture sequencing (MBD-Seq) analysis in the SVs of WT and αERKO mice at the two later developmental stages, puberty (week 5) and adult (week 10). Targeted bisulfite-Seq (TBS-Seq) analysis was used to validate the MBD-Seq data. A set of well-known estrogen-responsive genes (24) was selected for testing and the results of seven genes are shown in SI Appendix, Fig. S3. Differential DNA methylation regions (DMRs) were identified for each pair of samples and summarized in Fig. 5A. We found that DES increased both the loss and gain of DNA methylation between week 5 and week 10 in WT and αERKO SVs. Of note, there were many more gains of DMRs in αERKO DES vs. vehicle (16,632) compared with WT DES vs. vehicle (5,236) at week 5, suggesting the absence of ERα may have altered the methylation status in general (Fig. 5A, comparisons 1 and 2). To further investigate the role of ERα, we compared the loss and gain of DMRs between αERKO and WT vehicle SVs (Fig. 5A, comparison 3). Lack of ERα alone was enough to change the methylation status at many locations (both loss and gain of methylation), suggesting ERα may be important for developmentally setting the DNA methylation status in the SV. In addition, we found more gains of DMRs at week 5 (20,331) with both DES treatment and the absence of ERα compared with WT SVs, suggesting ERα may protect certain regions or these are sensitive to ERα presence (Fig. 5A, comparison 4).

Fig. 5.

DNA methylation dynamics during development in the mouse SVs following neonatal DES exposure. (A) Differentially methylated regions (DMRs) analysis for week 5 and week 10 SV MBD-Seq using MEDIPS. The numbers of loss or gain of methylation DMRs are shown with the four comparisons. (B) Distribution of DMRs in the mouse genome (mm10). The seven genomic regions are compared between weeks 5 and 10 in WT or αERKO samples.

Next, we investigated distribution of DMRs in the genome and found that the seven types of genomic regions were comparable between weeks 5 and 10 in the WT and αERKO SVs (Fig. 5B). In the WT loss of DMRs, there was a developmental switch from a reduction in the introns to an induction in the intergenic regions. However, it was opposite in the WT gains of DMRs (Fig. 5B, Top). In contrast, there was no switch in the mouse genome with either loss or gain of DMRs in the αERKO SVs (Fig. 5B, Bottom). The data suggest that a difference in DNA methylation pattern may occur during development in the mouse SVs that is governed by the presence of ERα.

Comparison of DNA Methylation Changes During Development in the WT and αERKO SVs After Neonatal DES Exposure.

To determine the DNA methylation changes in the two developmental stages, we overlapped the loss or gain of DMRs between week 5 and week 10 (Fig. 6A). We only found a small percentage of overlapping DMRs in all four comparisons, suggesting that there was a developmental change in DNA methylation in the mouse SV following DES exposure. In WT, only a small percentage of the overlapping DMRs in both losses (408) and gains (269) were obtained between weeks 5 and 10. In contrast, the overlapping gains of methylation DMRs in αERKO (5,264) was far above the number of losses (205), suggesting that ERα is playing a more significant role in addition to methylation than demethylation during development. To investigate whether ERα has a role in DNA methylation reprogramming by neonatal DES exposure in the adult SVs, we compared both loss and gain of DMRs for the WT and αERKO SVs at week 10 and found that only a small number of DMRs overlapped between the two genotypes (Fig. 6B), suggesting that DNA methylation dynamically changes in different loci of the mouse genome between the WT and αERKO SVs.

Fig. 6.

Comparison of DMRs and the enriched motifs in the WT and αERKO SVs following developmental DES exposure. (A) Venn diagrams depict the overlap between week 5 and week 10 DMRs with the loss and gain of methylation in four comparisons. (B) Venn diagrams depict the overlap between week 10 WT and αERKO SVs with the loss or gain of methylation DMRs. (C) Venn diagrams depict the overlap of motifs between WT and αERKO DMRs with the loss or gain, or WT loss and αERKO gain of DNA methylation.

To explore the molecular mechanisms that differentiate DNA methylation patterns in WT and αERKO SVs, we performed TF motif analysis for the loss and gain of DMRs. The results indicated 64 enriched motifs in WT losses, 40 in αERKO losses, 52 in WT gains, and 64 in αERKO gains (Fig. 6C). When overlapping the motifs of WT and αERKO, we found that there were more gain motifs shared than losses in the two genotypes. Interestingly, there were over 70% overlapping motifs in WT losses and αERKO gains (Fig. 6C). These results suggest that lack of ERα may cause more methylation, resulting in gene silencing in the adult SVs after DES exposure.

The Correlation of DES-Altered Gene Expression and DNA Methylation Status in the Adult Mouse SV.

To further understand the relationship between gene expression and DNA methylation, we mapped the week 10 MBD-Seq peaks to the location of DES-altered genes in the mouse genome within ±100 kb of the transcriptional start site (TSS). When the 1,678 DES-induced genes were checked against the 11,271 loss-of-methylation DMRs, 855 DES-induced genes were associated with 1,582 DMR losses (Fig. 7A, Left). When the 484 DES-repressed genes were checked against 19,381 gain of DMRs, 365 DES-repressed genes were associated with 1,170 DMR gains (Fig. 7A, Right). These findings demonstrate that DNA methylation status is associated with the expression levels of a subset of altered genes in the SVs of adult mice following neonatal DES exposure.

Fig. 7.

Correlation of gene expression and DNA methylation status in the adult mouse SVs following developmental DES exposure. (A) Data integration of DES-altered DE genes and DMRs which are mapped to the closest gene TSS ± 100 kb. (B) The top enriched motifs in 1,582 loss or 1,170 gain of methylation DMRs. See also SI Appendix, Tables S2–S4. (C) UCSC Genome Browser screen shots represent tracks of RNA-Seq, MBD-Seq, and TBS-Seq analysis for DES induced with the loss DMR gene (Mmp8) and DES repressed with the gain DMR gene (Shisa4).

Currently, our group has reported that functionality of ERα in the binding sites of target genes is restricted to EREs, which vary from the consensus palindromic element by one or two nucleotides (24). To identify ERα ERE-mediated genes in the adult mouse SV after developmental DES exposure, we searched for consensus ERE sites (ERE, GGTCAnnnTGACC or 1-nt mutation) in DES-altered genes that contained DMRs within a distance of ±100 kb of the TSS. After integrating the three datasets, including DE genes, DMRs, and EREs (SI Appendix, Fig. S4, Left), we found 29 DES-induced genes with a loss in methylation and 25 DES-repressed genes with a gain in methylation that contained EREs (SI Appendix, Fig. S4, Right). To determine whether ERα had direct interactions with the ERE binding site after DES exposure, as an example, we selected Epha2 (a DES-induced and loss-of-methylation gene) which contained an ERE site in the promoter region for further analysis (SI Appendix, Fig. S5). MBD-Seq data showed that DES caused the loss of DNA methylation at week 10 in the two regions of Epha2 and this status was correlated with the gene expression by RNA-Seq and qPCR analyses. Using ChIP-qPCR analysis, ERα and Pol II enrichments were significantly increased in an Epha2-ERE region in DES samples, which was expected. These results indicate that DNA methylation status, as exemplified with Epha2, correlates with gene expression and ERα can directly bind to an ERE of Epha2 in the SVs of adult mice, which is made accessible after neonatal DES exposure.

To further explore the molecular mechanisms that differentiate possible DES-altered gene expression via DNA methylation, we performed the TF motif analysis for a subset of gene-mapped loss and gain of DMRs. The results induced 25 enriched motifs from the known motif analysis, 30 motifs from the de novo motif analysis in the loss of DMRs (SI Appendix, Tables S3 and S4), four motifs from the known motif analysis, and nine motifs from the de novo motif analysis in the 1,170 gain-of-methylation DMRs (SI Appendix, Table S5). HRE, E2A, FOXP1, MafA, CRX, Foxo1, STAT6, and Smad3 motifs were enriched in the losses (Fig. 7B, Left) and HIF-1a, MYB, GATA3, Nrf2, Nkx2, Mycn, IRF4, and TFAP2A motifs were enriched in the gains (Fig. 7B, Right). Interestingly, AR half-site motifs were identified in both the gain and loss lists. This suggests that differential androgen signaling could occur from the DES treatment in WT SVs and play a significant role in altered gene expression related to the toxicity. In addition, associations between expression and DNA methylation status for the two genes, Mmp8 (DES-induced and loss methylation) and Shisa4 (DES-repressed and gain methylation) are shown in Fig. 7C. Using TBS-Seq, we confirmed the loss-of-methylation status of Mmp8 in the two regions from the two MBD peaks (7 CpGs in chr9: 7,560,386–7,560,591; 4 CpGs in chr9: 7,567,566–7,567,676) (Fig. 7C, Left) and the gain-of-methylation status of Shisa4 in a region from a MBD peak (16 CpGs in chr1: 135,372,809–135,373,184) (Fig. 7C, Right). Genes that have an association between expression and DNA methylation status are listed in Fig. 7C, Bottom. These results suggest that DNA methylation correlates with and regulates both induced and reduced gene expression with differentially enriched motifs in the adult SV following DES exposure.

Discussion

In this study, we examined the developmental effects related to toxicity in the mouse SVs following synthetic estrogen DES exposure. Our data uncovered that the associated tissue effects appear linked to alterations in the mRNA transcriptome and DNA methylation profile in the mouse SVs following neonatal DES exposure. Following a series of analyses, we concluded that ERα-mediated aberrations of the transcriptome and many of the altered genes were related to and enriched for developmental functions and potential diseases in the adult SVs when male mice were exposed to DES in the early stages of development. Interestingly, we found that a set of estrogen-responsive uterine genes were also aberrantly expressed in the SV of adult male mice following DES exposure. DNA methylation was reprogrammed dynamically during development in the mouse SVs following DES exposure. The changes occurred in different loci of the mouse genome between the WT and αERKO SVs. The expression levels of about 60% of DES-altered genes were associated with the corresponding DNA methylation status.

Exposure to EDCs during development can alter susceptibility to adult diseases later in life, including many diseases related to the male and female reproductive tracts affecting fertility (25–27). The toxicity of EDCs, such as DES or genistein exposure in female mice, has been well documented as a major contributor to uterine cancer when mice reached adulthood (23, 28–30). In agreement, we showed that cancer was the number one disease association following DES exposure from the uterine gene profiling analysis in this study. We also provided the natural developmental transcriptome expression in the mouse SVs following neonatal DES exposure, which is a unique transcriptome profile in this mouse reproductive organ. The observed phenotype difference between WT-veh and WT-DES SV during development could be due to upstream regulators such as cytokines (TNF, IL4, and IL1B), kinases (EGFR, IKBKB, JAK, and MAPKs), and many transcription regulators that become activated. Many of these regulate cell death, differentiation, and proliferation, which are vital in the growth of cells and tissue. The repression of organ and tissue morphology genes such as Ggt1, Lep, and Nr4A3, and feminization protein GPRC6A are critical for cell adhesion (31). More interestingly, the 111 overlapping DES-altered genes (out of 155) between uterus and SVs are related to reproductive tract diseases. These findings provide unique mechanistic information to understand the differential toxicities in males and females following EDC exposure as shown with DES.

The Ltf gene, which is highly expressed in the female reproductive tract, was used as a marker for the early hormonal response in the developing mouse uterus (32, 33). Normally, Ltf is not expressed in adult mouse SVs (17). However, neonatal DES exposure significantly induces the basal levels of this gene’s expression and altered programming in the adult mouse SVs (15, 17). We demonstrated that neonatal DES exposure induced the Ltf gene expression continually in the SVs from week 3 to week 10. However, it did not occur in the adult mouse uterus, indicating a differential tissue selective reprogramming of this locus from the developmental DES exposure. The Wnt family proteins as secreted ligands can act through many receptors to stimulate signaling pathways. These proteins have been linked to oncogenesis and developmental processes in the female reproductive tract (34–36). Here, we report that the Wnt4 gene is highly expressed in the early SV developmental stage and has reduced expression immediately following puberty. Interestingly, this gene was only induced in the SVs of WT adult mice and not the αERKO. These findings suggest that estrogen-responsive genes regulated by ERα action, such as Ltf and Wnt4, contribute to hormonal regulation during development in a tissue-specific manner; therefore, alterations in their expression after neonatal DES exposure could result in feminization of the SV.

Epigenetic modifications, including DNA methylation and histone modification, play important roles in regulating cellular differentiation events. In mammalian cells, DNA methylation occurs at cytosine, predominantly in CpG dinucleotide contexts (37, 38). Here, we identified the DMR’s response to DES in the WT and αERKO SVs and found that there were more DNA methylation changes (gain or loss) in the αERKO than in the WT SVs after neonatal DES exposure. Neonatal DES exposure increases the level of testosterone in the adult αERKO compared with WT mice (17). The dynamic DNA methylation changes suggest that there is a possibility of other hormones, such as testosterone, involved in the DNA methylation changes. Interestingly, when we mapped the week 10 αERKO gained DMRs to the genes within 10 kb, we discovered that about 25% of DES-induced genes (423 of 1,678 genes) in the WT SVs overlapped with αERKO gained DMR genes (SI Appendix, Fig. S6). This could be an explanation as to why some of the genes were silenced in the αERKO due to the DNA methylation status. EDCs act primarily through nuclear receptors, including ER and AR (27). It is expected that different disorders are seen in males and females as a result of EDC effects that mimic estrogens and/or antagonize androgens (27). In agreement, we found that AR and androgen as the upstream regulators were inhibited in the SVs following DES exposure. A balance between ER and AR regulation and signaling during development could impact the effect of EDCs in male reproductive organs.

An original DNA methylation study indicated that prenatal DES exposure caused demethylation of three specific CpG sites from a methylated status in the Ltf gene promoter in 3-wk-old CD-1 mouse uteri (39). We found that the same three CpG sites that were unmethylated in the adult mouse SVs by neonatal DES exposure were not changed in the 3-wk-old C57BL/6 mouse SVs, which remained in the methylated status (21). In this study, we identified a CpG region in the intron of the Ltf gene that was unmethylated by DES exposure in the adult mouse SVs. These results suggest that DES exposure alters the DNA methylation pattern with a sex and/or strain specificity during development that impacts different CpG loci.

DNA methylation was traditionally known as a regulator in gene silencing, but recent studies have shown a more complex picture. More recently a study showed that DNA methylation can also have different outcomes, including activation of transcription (40). The direct relationship between DNA methylation and gene expression remains unclear (41). The association between DNA methylation and gene expression has been proposed to be either active or passive, and can depend on the context in which they occur in the genome (42). Here we found that the expression levels of genes, around 50% of DES-induced and 75% of DES-repressed genes, were correlated with their DNA methylation status. These findings suggest that DNA methylation is involved in both DES-induced and -repressed gene expression in the adult mouse SVs after neonatal DES exposure. Evidence has indicated that transcription factors can interact with methylated DNA sequences (40). In our analysis, we also found the sets of DES-induced genes with a gain of DMRs or repressed genes with the loss of DMRs (SI Appendix, Fig. S6A). Further analysis of those DMRs and genes will be needed for understanding the role of DNA methylation in gene regulation. Additionally, gene silencing may have a major impact that inhibits a required gene expression during development in the mouse SVs following neonatal DES exposure. Studies looking at natural DNA methylation variations in human populations have shown that genetic variations influence DNA methylation levels in different tissue/cell types (43, 44). Additionally, epigenetic regulation of gene expression and the association of specific histone marks with promotor sequence classes are fine tuned in a cell type-specific manner (45). To have a better understanding of the epigenetic mechanism of DES-induced toxicity in the mouse SVs, a genome-wide analysis of DNA methylation in a single cell type or analysis of histone marks should be performed in future studies.

Conclusions

Our findings demonstrate that neonatal DES exposure causes toxicity in the mouse SV from early developmental stages through adulthood in mice and many developmental genes are altered. ERα appeared to play a role in mediating the aberrant transcriptome expression induced by DES. Interestingly, we found that a set of estrogen-responsive uterine genes were also now aberrantly expressed in the SV of adult male mice following DES exposure and certain gene changes appeared to be cell-type specific. DES exposure reprogrammed DNA methylation dynamically during development in the mouse SV. These observations were seen in different loci of the mouse genome when considering both WT and αERKO SVs. Furthermore, DNA methylation status associated with the expression levels of ∼60% of DES-altered genes in the mouse SV. These findings provide unique evidence for understanding molecular and epigenetic mechanistic consequences in human health following developmental EDC exposure.

Materials and Methods

Animal and Neonatal Treatments.

All animal studies were conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals (47). The full protocol can be found in SI Appendix, Materials and Methods.

DNA Extraction, MBD-Seq, Mapping, and Analysis.

Genomic DNA samples were extracted from pooled frozen tissues of each group using a Tissue Blood Kit (Qiagen) according to the manufacturer’s protocol. Pooled genomic DNA samples were sonicated with Biorupter (Diogenode) and methylated DNAs were captured with his-tagged recombinant MBD2b along with its binding partner MBD3L1 using the MethyCollector Ultra Kit (cat. no. 55005, Active Motif). The full protocol can be found in SI Appendix, Materials and Methods.

TBS-Seq, Mapping, and Analysis.

Bisulfite conversion sequencing PCR primers were designed using the software program EpiDesigner (www.epidesigner.com/). Bisulfite conversion sequencing PCR was performed using the EZ DNA Methylation-Gold Kit (Zymo Research). The full protocol can be found in SI Appendix, Materials and Methods.

The materials and methods for H&E staining; RNA extraction and qPCR; microarray (GeneChip) and data analysis; RNA-Seq, mapping, and analysis; laser capture microdissection (LCM); ChIP-qPCR analysis; pattern analysis; motif analysis; ERE motifs contained within peaks; and Statistical analysis can be found in SI Appendix, Materials and Methods.