Transcriptomic Changes Across the HPG Axis Following Prenatal Exposure to the EDC Mixture NeuroMix

Abstract

Endocrine-disrupting chemicals (EDCs) are exogenous chemicals that are ubiquitous in our environment and found in everyday items. We previously reported that prenatal exposure of rats to a human-relevant mixture of EDCs, NeuroMix (NMX), led to alterations in physiological and behavioral phenotypes. Here, we used hypothalamic-pituitary-gonadal (HPG) tissues from these same male and female rats and conducted 3′ Tag-based RNA sequencing (TagSeq) to investigate underlying molecular mechanisms. TagSeq revealed unique tissue- and sex-specific differentially expressed genes (DEGs). In males, among the HPG tissues, NMX had the greatest effects in the hypothalamic arcuate nucleus (ARC), with 613 DEGs. Gene ontology (GO) enrichment analysis revealed that genes upregulated in the ARC of NMX males were involved in synaptic plasticity, while genes downregulated related to responses to estradiol and glucocorticoids. In females, prenatal NMX exposure induced the largest transcriptome change in the ovaries, with 1295 DEGs. GO-enrichment analysis revealed upregulation of genes involved in cilium organization and movement, while genes downregulated in this region were related to immune-related processes. Using Qiagen Ingenuity Pathway Analysis, we identified the β-estradiol pathway to be activated in all NMX female tissues and the NMX male pituitary, and inhibited in NMX male ARC, ventromedial nucleus, and testes. To our knowledge, this is one of the first studies to conduct transcriptomic profiling across HPG tissues, with these results demonstrating that prenatal exposure to NMX affects gene expression across the HPG axis in a sex-dependent manner.

Exposure to endocrine-disrupting chemicals (EDCs), either singly or as mixtures, can affect hormone signaling and actions (1, 2). Humans are exposed to changing mixtures of EDCs throughout their lives from contact with personal care products and medical tubing to children's toys, food packaging, pesticides, industrial waste, and other sources, in a manner that depends on geography, diet, lifestyle, and socioeconomic status. EDC exposure is associated with a wide range of health issues, such as reproductive disorders (3), neurodevelopmental issues (4), immune system dysfunction (5), and certain cancers (6).

Among the endocrine systems affected by EDCs are the body's neuroendocrine systems, including the hypothalamic-pituitary-gonadal (HPG) axis, which controls reproduction in both females and males (7, 8). HPG function begins with the hypothalamic secretion of GnRH from nerve terminals into the portal capillary vasculature, which targets the anterior pituitary gland release of the gonadotropins, FSH, and LH (9). Pituitary gonadotropins in circulation travel to the ovaries or testes to promote the synthesis of sex steroid hormones, as well as ovulation in females and spermatogenesis in males (9). Feedback from gonadal steroids modulates hypothalamic GnRH and pituitary gonadotropin output to close the loop on the HPG axis.

Animal models have been used to explore the developmental origins of health and disease hypothesis and have established causal connections between exposure to EDCs early in life to later reproductive (10-13) and metabolic (14, 15) disorders. Additional molecular studies have provided insight into how prenatal EDC exposure affects the expression of neuroendocrine genes and proteins across the HPG axis in adulthood (16, 17). Previous work has revealed that several EDCs (eg, polychlorinated biphenyls, phthalates, bisphenols) alter reproductive physiology and behavior and change gene expression in the control of reproduction (17-20).

Despite these findings, there are several important gaps in knowledge about EDC effects that we sought to address. First, although the hypothalamus, pituitary, and gonad have been studied separately (19-22), little research has included all 3 HPG levels in a single study, and even fewer have included head-to-head comparisons of males and females. Here, we evaluated all 3 HPG tissues from rats of both sexes. Second, with few exceptions, most molecular work to date has examined a limited number of gene targets; in the current study, we sequenced across the entire transcriptome. Third, the majority of research has used a single EDC or a mixture of EDCs within the same class [eg, a mix of polychlorinated biphenyls or a mix of phthalates (23)]; however, mixtures of EDCs may have effects that cannot be observed or predicted by investigating one chemical at a time. Moreover, doses of EDCs given individually may, in combination, have adverse effects even when given below the no observed adverse effect level (24-26). Furthermore, mixtures represent what animals and humans are constantly exposed to across their lives.

Therefore, this study aimed to identify how prenatal exposure to a low-dose mixture of environmentally relevant EDCs, named NeuroMix (NMX) for its selection based on neurobiological impacts (24, 27-30), affects gene expression across the HPG axis. To do so, we used 3′ Tag-based RNA sequencing (TagSeq), a global transcriptomic approach to identify potential gene targets, in 2 hypothalamic nuclei, the arcuate nucleus (ARC) and the ventromedial nucleus (VMN), the pituitary, and the gonads of male and female rats that were prenatally exposed to NMX or vehicle control treatments.

Methods

Breeding, EDC Treatment, and Exposure Paradigm

All animal protocols followed National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee at The University of Texas at Austin. Animals described in this study were a subset of those described in a previous publication (24); the tissues used here had been frozen and stored at −80 °C. Therefore, methods are described in brief. All adult Sprague-Dawley rats (dams and sires) were purchased from Envigo and were allowed to acclimate to the facility for a minimum of 14 days prior to any experimental manipulation. They were given standard husbandry and fed a low phytoestrogen diet (Teklad Global Diet, 2019, Indianapolis, IN) ad libitum. While we acknowledge that transportation and commercial breeding conditions can create stress, animals were sourced from the same supplier under consistent conditions to reduce variability across litters, generations, and cohorts.

Virgin females were bred, with the day of mating designated as embryonic day 0. If a sperm-positive sample was detected by vaginal smear, dams were randomly assigned to treatment groups, receiving 100 μL of NMX (n = 14 dams) or vehicle (n = 12 dams) on a quarter of a Nilla wafer fed to dams daily from embryonic day 8 to 18. On postnatal day (P) 1, the day after birth, body weight and anogenital distance were measured, and litter sex ratios were determined. Litters were culled to 5 males and 5 females. At P21, rats were weaned into same-sex cages, 2 to 3 per cage. Body weight, anogenital distance, and observations of puberty onset (date of vaginal opening or preputial separation) were recorded, and estrous cyclicity was monitored in females. Euthanasia took place at ∼P91 by rapid decapitation, with females euthanized on proestrus; brains, pituitary, ovaries, and testes were immediately removed, frozen on dry ice, and stored at −80 °C until further processing.

NMX

The NMX composition is shown in Table 1 and fully described in Gore et al (24). Briefly, NMX was formulated to account for common EDCs found in humans, which are known for their neurobiological effects, as shown by epidemiological and experimental animal studies (31-35). Dosages were chosen to be below the reported no observed adverse effect level. The vehicle solution comprised 0.06% dimethyl sulfoxide and 0.0037% ethanol in sesame oil. A uniform stock solution was used to dose all experimental subjects to prevent batch variation.

Table 1.

Composition of NeuroMix

Chemical (Abbreviation)	Dose
Aroclor 1221 (A1221)	100 µg/kg
Bisphenol A (BPA)	2.5 µg/kg
Bisphenol S (BPS)	2.5 µg/kg
Di(2-EthylHexyl)-phthalate (DEHP)	20 µg/kg
Di-n-butyl phthalate (DBP)	20 µg/kg
Perfluorooctane sulfonate (PFOS)	20 µg/kg
Polybrominated diphenyl ether 47 (PBDE-47)	100 µg/kg
PCB-153	100 µg/kg
Vinclozolin	100 µg/kg

Estradiol and Testosterone Radioimmunoassay

Banked serum samples from the same animals used for TagSeq were used to quantify circulating estradiol (E2; females) and testosterone (males) concentrations. Serum E2 concentrations were detected in duplicate in females (n = 11 vehicle, n = 12 NMX) using an RIA kit, following the manufacturer's protocol (Cat. No. DSL-4800, Beckman Coulter, Brea, CA). Estradiol was measured in a single RIA, which had a range of 2.2 to 750 pg/mL. The E2 RIA has an assay sensitivity of 2.2 pg/mL, and the assay had a mean intra-assay variability of 3.2%. Serum testosterone was similarly run in duplicate in a single assay using an RIA kit (Cat. No. 07189102, MP Biomedicals, Santa Ana, CA). The assay range was 0.1 to 10 ng/mL, assay sensitivity was 0.03 ng/mL, and intra-assay variability was 8.0%. Two samples were removed from testosterone RIA analysis, 1 due to high intra-assay variability (>25%) and the other because it was identified as an outlier by Grubb's test. This led to an n of 10 in the vehicle group and an n of 12 in the NMX group. Linear mixed-effect models (LMM) were used to determine the effect of treatment on serum hormone levels. Separate models were run for E2 and testosterone; each included cohort as a random factor. P-values lower than .05 were considered significant.

Tissue Preparation, RNA Isolation, and 3′ TagSeq

Frozen brains were sectioned into 500 μm coronal sections on a Thermo Fisher Scientific Cryostar NX50 cryostat. The ARC and VMN of the hypothalamus were identified using a rat brain atlas (36), bilaterally punched using a 0.75 mm diameter Palkovits punch (Stoelting, Wood Dale, IL), and stored at −80 °C until processed further. All tissues were lysed using a RNeasy Lysis buffer (Qiagen, Germantown, MD) with β-mercaptoethanol (Thermo Fisher Scientific, Waltham, MA). For the ARC, VMN, and pituitary, tissue was submerged in 600 μL of buffer, and a 1 cc syringe with a 22-gauge needle was used to lyse the tissue by homogenizing up and down about 10 times. For ovaries and testes, 1 stainless steel bead (5 mm diameter, Qiagen, Hilden, Germany) was placed in a 2 mL microcentrifuge tube with 450 μL of buffer before tubes were placed in the TissueLyser (Qiagen, Germantown, MD) for 4 minutes at 20 to 30 Hz, rotating the rack of tubes after the initial 2 minutes. Sixty  μL of the homogenized tissue was placed into a fresh 1.5 mL tube with 290 μL additional buffer. For all HPG tissues, once lysed, total RNA was extracted using an RNAeasy Mini Kit (74104; Qiagen, Germantown, MD) according to the manufacturer's protocol with added DNAse I to remove genomic DNA. RNA quality was determined using a RNA6000 Pico Kit Assay on a BioAnalyzer. Processed samples had an average RIN of 8.5, well above the recommended 7 cutoff (Agilent Technologies, Santa Clara, CA). RNA concentrations were determined with a Quantifluor RNA System Kit (E3310; Promega, Madison, WI). ARC and VMN samples were diluted to 15 ng/μL, pituitary samples were diluted to 50 ng/μL, and ovaries and testis were diluted to 25 ng/μL and stored at −80 °C before being submitted to The University of Texas Genomic Sequencing and Analysis Facility for library preparation and sequencing. In total, 155 extracted RNA samples were submitted and processed in 3 separate runs: 1 for ARC and VMN samples, a second for pituitary samples, and a third for ovary and testis samples for tag-based RNA sequencing (37). Table 2 shows the number of samples in each treatment group for each tissue type. Libraries were constructed, according to (38). Reads were sequenced on the NovaSeq 6000 SR100 with minimum reads of 4 million and target reads per sample of 5 million.

Table 2.

Number of samples for each tissue type and treatment group

	Hypothalamic ARC		Hypothalamic VMN		Pituitary		Ovary		Testis
	NMX	Veh	NMX	Veh	NMX	Veh	NMX	Veh	NMX	Veh
Males	7	6	10	10	12	12	—	—	11	12
Females	5	6	9	9	12	11	12	11	—	—

Hypothalamic-pituitary-gonadal tissues were selected from the same subset of animals.

Abbreviations: ARC, arcuate nucleus; Veh, vehicle; VMN, ventromedial nucleus.

Bioinformatics

Preprocessing

Raw reads were processed to obtain gene count data by following the TagSeq data processing pipeline based on Lohman et al (38) and Meyer et al (39). Briefly, a customized Perl script utilizing FASTX-Toolkits and CUTADAPT 2.8 (40) was used to remove reads with a homo-polymer run of “A” ≥8 bases and retain reads with a minimum 20 bases and to remove PCR duplicates. The quality of raw sequence data was checked with FastQC (41). All samples passed quality control and were used in the analysis. Processed reads were then mapped to the Rattus norvegicus reference genome (mRatBN7.2, National Center for Biotechnology Information) using Star (42). The average mapping rates for ARC, VMN, and anterior pituitary were 87.5%, 88.6%, and 89.4%, respectively. Ovaries and testes had an average mapping rate of 70.0%. All genomic data associated with this study have been deposited in the National Center for Biotechnology Information database under accession number PRJNA1303210 (43).

Analysis of differentially expressed genes

All differentially expressed gene (DEG) analyses were conducted using the Bioconductor package “limma” (44). Gene counts obtained from STAR were organized by tissue type, and separate differential gene expression analyses were performed. First, we filtered out genes with <10 counts across all samples in each tissue type. Next, filtered reads were normalized by the trimmed mean of the M-values normalization method (43) to adjust for differences in library size among samples. Males and females were normalized together to reduce the likelihood of unequal variances among groups for the ARC, VMN, and pituitary. After normalization, counts for males and females were analyzed separately in limma using weighted least squares linear models for each gene, and then contrasts were made of these fitted linear models to calculate log2 fold change and significance for genes between NMX and vehicle groups for each sex. Raw P-values for each DEG analysis were adjusted via empirical false discovery rate (eFDR; 45). We permutated sample labels 5000 times and obtained a null distribution of P-values to estimate the eFDR. We defined DEGs as those with an absolute log₂ fold change of at least 0.2 (corresponding to a 15% change in expression) and an eFDR below 5%. Enriched gene ontology (GO) analysis was conducted to explore functional differences among the different comparisons of DEGs. We used the “clusterProfiler” R package (46) to test the statistical overlap between annotated biological and functional significance in each DEG set.

Rank–rank hypergeometric overlap

To compare the sexes, we used the R package “RedRibbon” to perform a rank–rank hypergeometric overlap analysis to detect the overlap between genes differentially expressed in the same or opposite directions (47). A pairwise comparison of DEG lists in males and females was conducted for the ARC, VMN, and pituitary. Each gene list was ranked according to its fold change. The colors of each square in the rank–rank hypergeometric overlap map indicate –log10(P-value) of the overlap. The bottom left and top right quadrants in each map represent the overlap of shared downregulated and upregulated genes, respectively. In contrast, the top left and bottom right quadrants depict discordant overlaps.

Hypothesis testing

To test our hypotheses that NMX influenced estrogen and testosterone pathways, we selected a set of a priori genes known to be related to E2 and testosterone in each tissue type [Supplemental Table S1 (48)]. Using linear mixed models, we investigated the relationship of normalized gene expression of these genes with serum E2 for females and serum testosterone for males, using serum hormones as the fixed factor and cohort as a random factor.

Ingenuity pathway analysis

All genes normalized in the limma analysis were further analyzed using the Qiagen Ingenuity Pathway Analysis (IPA) software (Qiagen Inc., https://digitalinsights.qiagen.com/IPA). The data were uploaded into the IPA platform, and genes were filtered to only include log2 fold change higher than 0.2 with an eFDR above .05. Then a core analysis was conducted for each tissue in both sexes to identify key biological processes affected by NMX exposure.

Weighted gene coexpression analysis

Weighted gene coexpression network analysis (WGCNA) was performed in the ovary and testis using the “WGCNA” R package (49) because these tissues had an adequate sample size (>20). The sample sizes for VMN, ARC, and pituitary were insufficient to do this analysis. Ovaries and testes were run in 2 separate WGCNA analyses created for each tissue type. Both WGCNA models were constructed in signed hybrid mode with a minimum module size of 100 genes. To construct a signed hybrid correlation network in the ovaries, a power value of 4 was set [Supplemental Fig. S1A (48)]. WGCNA identified 25 modules with the 23 combined ovary samples [Supplemental Fig. S1B and 1C (48)]. For the 24 testis samples, a WGCNA model was constructed with a power value of 4, leading to the identification of 11 modules [Supplemental Fig. S2 (48)]. After identifying modules of coexpressed genes within each network, we calculated the differentially expressed module eigengene (deMEGs). The deMEGs is the first principal component of the gene expression levels of all genes in the module, summarizing the overall gene expression profile. Further, we calculated module membership (MM) for each gene in modules of interest. MM is measured as the Pearson correlation between the gene expression level and the module eigengene, and the absolute value of module membership close to 1 indicates that the gene is highly connected to other genes in the module. Genes with MM above 0.8 were considered hub genes in their modules. Linear mixed effects model was performed with treatment and serum hormone level, either E2 for ovary or testosterone for testis, as fixed factors and deMEGs as the outcome for all the modules. All models included cohort as a random factor to determine the association between each deMEGs and treatment. Modules were considered significant if they showed a P-value lower than .05.

Results

Steroid Hormones

Estradiol

Serum E2 concentrations were measured in NMX and vehicle-exposed adult female rats at ∼P91 (euthanized on proestrus). Using a linear mixed effect model with cohort as a random factor, we found no significant differences in serum E2 between treatments (β = 7.8 ± 6.3, P = .21, Fig. 1).

Figure 1.

Box plots showing median and interquartile range of serum estradiol in females (left) and testosterone in males (right). Points represent individual rats. *P < .05.

Testosterone

For males, we investigated the effects of prenatal NMX exposure on serum testosterone in adulthood (∼P91). A linear mixed effect model with cohort as a random factor found serum testosterone to be significantly higher in the NMX-treated males (β = .74 ± .33, P = .04, Fig. 1).

Number of DEGs in Hypothalamus and Anterior Pituitary

The transcriptomes of the hypothalamic ARC, hypothalamic VMN, and pituitary were examined first to compare the effects of NMX on the transcriptomes and to leverage the ability to make head-to-head comparisons between female and male rats. The total number of DEGs for each tissue type can be found in Table 3.

Table 3.

Total number of differentially expressed genes between NMX and vehicle animals

		ARC		VMN		Pituitary
	Log2 fold change	↓↓	↑↑	↓↓	↑↑	↓↓	↑↑
Males	≥0.2 (total)	296	317	85	118	110	84
	0.2-0.5	172	241	75	102	106	77
	0.5-1	88	69	6	6	4	6
	>1	36	7	4	10	0	1
Females	≥0.2 (total)	113	79	131	138	101	110
	0.2-0.5	109	75	102	121	96	104
	0.5-1	3	4	12	15	4	6
	>1	1	0	17	2	1	0

Data were split across 4 different log2 fold change thresholds for both males and females. The top row (≥0.2) includes all genes meeting this threshold. The next 3 rows show these genes distributed from lowest to highest for log2-fold changes. Up arrows indicate upregulation by NMX compared to vehicle; down arrows indicate downregulation.

Abbreviations: ARC, arcuate nucleus of the hypothalamus; NMX, NeuroMix; VMN, ventromedial nucleus of the hypothalamus.

ARC transcriptomics

NMX affected gene expression in the ARC, with more effects in males than females (Fig. 2A). At our cutoff of log2 fold changes >0.2, males had 613 DEGs (6.9% of the total genes in our analysis) whereas females had 192 DEGs (2.2% of the total genes in our analysis; Fig. 2B). This reflects 3.2-fold more genes being differentially expressed in the ARC of NMX males than females. A full list of all genes including DEGs found in the ARC of NMX males and females compared to same-sex controls can be found in Supplemental Table S2 (48).

Figure 2.

Volcano plots represent log2 fold change by −log10 of eFDR .05 for male (A) ARC, (C) VMN, (E) pituitary, and (G) testes, and female (B) ARC, (D) VMN, (F) pituitary, and (H) ovaries. Each volcano plot is annotated with genes with the largest fold changes. Orange circles on the left indicate downregulation, blue circles on the right indicate upregulation.

Abbreviations: ARC, arcuate nucleus; eFDR, empirical false discovery rate; PIT, pituitary gland; VMN, ventromedial nucleus.

To investigate the biological underpinning of these DEGs, we performed GO-enrichment analysis on those genes found to be up- or downregulated by NMX in males and females compared to same-sex vehicle controls. In NMX males, GO-enrichment analysis revealed genes upregulated in the ARC were involved mainly in regulating synaptic plasticity, learning and memory, and dendrite development [Fig. 3A; Supplemental Table S3 (48)]. Genes related to these processes showing differential expression in NMX males included brain-derived neurotrophic factor (Bdnf), SH3 and multiple ankyrin repeat domains 1 (Shank1), glutamate receptor metabotropic 1 (Grm1), along with multiple GABA receptors: gamma-aminobutyric acid receptor subunit alpha-1 (Gabra1), subunit α-2 (Gabra2), and subunit α-4 (Gabra4; Fig. 4). The same GO-enrichment analysis indicated genes downregulated in the NMX males are those involved in responses to E2 and glucocorticoids. This includes GHRH (Ghrh), leptin receptor (Lepr), galanin (Gal), and IGF binding protein 6 (Igfbp6; Fig. 4). These genes along with others related to steroid hormone signaling did not correlate with serum testosterone levels [Supplemental Fig. S3A (48)]. To further assess the reductions in body weight associated with NMX exposure, including attenuated weight gain and lower adult body weight reported in Gore et al (24), we evaluated whether expression levels of the metabolic signaling genes presented in Fig. 4 were associated with body weight at P1 and P84 [adulthood; Supplemental Fig. S4 (48)]. Of the genes examined, only Ghrh exhibited a significant positive correlation with body weight at both developmental timepoints, such that lower expression levels were associated with lower body weight regardless of treatment condition. No significant associations were observed for the other metabolic genes.

Figure 3.

Significantly enriched GO terms related to biological processes are displayed for male ARC (A), VMN (C), and PIT (E), as well as female ARC (B), VMN (D), and PIT (F). The Y-axis represents GO term descriptions, while the X-axis shows the gene ratio, which indicates the proportion of input genes associated with each GO term. Dot size corresponds to −log10(eFDR), and color represents gene expression direction: open yellow circles for GO terms from downregulated genes and filled purple circles for those from upregulated genes. All effects are shown for NMX compared to controls.

Abbreviations: ARC, arcuate nucleus; eFDR, empirical false discovery rate; GO, gene ontology; NMX, NeuroMix; PIT, pituitary; VMN, ventromedial nucleus.

Figure 4.

Heatmap of gene expression in male and female rats prenatally exposed to NMX or vehicle. Eleven genes related to cilia assembly and organization, 15 genes involved in endocrine systems, 7 genes affiliated with stress, 10 genes related to immune processes, 13 genes associated with metabolic signaling, and 8 genes involved in synaptic signaling have expression mapped in the 8 tissue types: M-ARC, F-ARC, M-VMN, F-VMN, M-PIT, F-PIT, ovaries, and testes. The comparison for all tissue types was animals receiving NMX compared to same-sex controls receiving vehicle. A scale is shown, with red representing relatively higher gene expression and blue representing relatively lower gene expression. The significantly different genes (P < .05) for each comparison in each tissue type are represented with an asterisk (*); genes trending in significance (P .05-.09) are represented with a plus sign (+).

Abbreviations: F-ARC, female arcuate nucleus; F-PIT, female pituitary; F-VMN, female ventromedial nucleus; M-ARC, male arcuate nucleus; M-PIT, male pituitary; M-VMN, male ventromedial nucleus; NMX, NeuroMix.

For females, GO enrichment identified genes upregulated in animals receiving NMX related to glucocorticoid response and canonical Wnt signaling [Fig. 3B; Supplemental Table S4 (48)]. Genes downregulated in NMX females compared to vehicle were associated with synaptic plasticity, learning and memory, and dendrite morphogenesis [Fig. 3B; Supplemental Table S4 (48)]. We also investigated whether genes known to play a role in estrogen signaling correlated with serum E2 in these animals but found no significant differences [Supplemental Fig. S5A (48)]. These genes were also not significantly different between treatment groups in the female ARC (Fig. 4).

VMN transcriptomics

Males and females had similar numbers of DEGs when compared to same-sex controls (Fig. 2C and 2D). In males, 203 genes (1.9% of the total genes in our analysis) were differentially expressed; in females this was 269 genes (2.5% of the total genes in our analysis; Table 3). A complete list of genes including DEGs identified in the VMN of NMX males and females, compared to same-sex controls, is available in Supplemental Table S5 (48).

GO-enrichment analysis performed on the DEGs in the VMN revealed genes upregulated in NMX males compared to vehicle were involved in synaptic plasticity, memory and learning, and neuron differentiation [Fig. 3C; Supplemental Table S3 (48)]. Significantly different genes included in these processes include glutamate receptors glutamate receptor, metabotropic 1 (Grm4), and glutamate receptor-interacting protein 1 (Grip1), as well as Bdnf and synaptopodin (Synpo). Notably, oxytocin (Oxt), a gene that plays a role in modulating synaptic plasticity as well as social behaviors, is downregulated in the VMN of males receiving NMX compared to control males. These synaptic plasticity-related genes identified in the male VMN overlap with those previously observed in the male ARC following NMX exposure. In contrast, expression of these same genes in females showed a nonsignificant but opposite pattern in both the ARC and VMN (Fig. 4). Four genes—Bdnf, cholecystokinin (Cck), glutamate ionotropic receptor NMDA type 2A (Grin2a), and cholinergic receptor nicotinic α-7 (Chrna7)—were found to overlap across synaptic plasticity-related GO terms in the male ARC, male VMN, and female ARC. These genes were upregulated in male tissues but downregulated in female tissue, although statistical significance was only reached in male tissues. Genes downregulated in NMX males' VMN were involved in responses to E2 and adaptive immune response, including Wnt family member 6 (Wnt6) and member 4 (Wnt4; Fig. 4). The genes showing the largest upregulation in the VMN of NMX males were Cck and transthyretin (Ttr), both involved in thyroid and metabolic signaling (Fig. 4). Genes related to steroid hormone signaling in the male VMN gene expression results were not significantly different between treatment groups (Fig. 4) and did not correlate with serum testosterone [Supplemental Fig. S3B (48)].

For females exposed to NMX, the GO-enrichment analysis showed genes upregulated in VMN compared to controls were related to reproductive system development, hormone metabolic processes, and responses to E2 and glucocorticoids [Fig. 3D; Supplemental Table S4 (48)]. Significant DEGs involved in these processes include lysine methyltransferase 5A (Kmt5a), TIMP metallopeptidase inhibitor 1 (Timp1), IGF-binding protein 3 (Igfbp3), family with sequence similarity 107 member A (Fam107a), and FK506 binding protein 5 (Fkbp5). Genes downregulated in NMX females compared to vehicle were involved in cilium assembly, organization, and movement (Fig. 4) and included: cilia and flagella associated protein 161 (Cfap161), dynein axonemal heavy chain 10 (Dnah10), (Dnaaf11), and sperm flagellar 1 (Spef1) (Fig. 4). Pro-melanin concentrating hormone (Pmch) is another notably highly downregulated gene in NMX females as it is important in neuroendocrine function and metabolic signaling (Fig. 4). We did not find other key endocrine signaling genes to be differentially expressed between treatment groups in the female VMN (Fig. 4). There were also no significant relationships found between serum E2 and genes known to play a role in estrogen signaling [Supplemental Fig. S5B (48)].

Pituitary gland transcriptomics

In the pituitary gland, we found less robust differential expression between the NMX rats compared to same-sex control counterparts. NMX males had a total of 194 (1.9% of the total genes in our analysis; Fig. 2E), and NMX females had 211 (2.0% of the total genes in our analysis; Fig. 2F) genes differentially expressed with log2 fold changes >0.2 compared to same-sex controls. The full list of genes including DEGs in the pituitary of NMX males and females, compared to same-sex controls, is available in Supplemental Table S6 (48). Using GO-enrichment analysis, no significant GO terms were found to represent the genes downregulated in the pituitary of NMX males. Genes upregulated in the pituitary of NMX males were related to neurogenesis, gliogenesis, responses to growth factor β, and E2 [Fig. 3E; Supplemental Table S3 (48)]. GO-enrichment analysis in females showed that genes upregulated in the pituitary of NMX females were associated with response to glucocorticoids, neurogenesis, and gliogenesis [Fig. 3F; Supplemental Table S4 (48)]. Since GO terms related to neurogenesis and gliogenesis appeared in both male and female pituitary datasets, we examined the genes associated with these terms across sexes. We identified 3 genes in common: triggering receptor expressed on myeloid cells 2 (Trem2), EPH receptor A4 (Epha4), and smoothened (Smo), which were upregulated in both males and females. However, none of these genes remained statistically significant after correction for multiple hypothesis testing. Genes downregulated in the pituitary of NMX females were associated with antigen processing and presentation [Fig. 3F; Supplemental Table S4 (48)]. There were also no significant changes in the expression of steroid hormone signaling genes in the male pituitary. In contrast, in the female pituitary, androgen receptor (Ar) was upregulated and IGF-1 (Igf1) was downregulated in response to NMX exposure (Fig. 4). However, none of these hormone-related gene expression changes correlated with circulating hormone levels in either sex [Supplemental Figs. S3C and S5C (48)].

Concordant and discordant gene expression between NMX males and females

Because this study was fully powered for analyses in males and females, we used this opportunity to compare relationships of effects of NMX between the sexes in common tissues (ARC, VMN, pituitary). Using the rank-rank hypergeometric overlap approach, we evaluated the similarities and differences between the gene expression patterns induced by NMX in male and female samples when compared to same-sex vehicle controls. Notably, there was a strong significant discordance in gene expression in the ARC between NMX males and females (Fig. 5A). In the VMN, there was still some discordance in gene expression, particularly in the genes that were upregulated in NMX females but downregulated in NMX males (Fig. 5B). However, in the pituitary we saw a predominantly shared pattern of gene expression across the sexes (Fig. 5C).

Figure 5.

RedRibbon hypergeometric map of genes in male and female NMX samples. The rank-rank hypergeometric overlap algorithm was conducted for each pairwise analysis (males vs females) in the (A) ARC, (B) VMN, and (C) pituitary. The genes were ranked according to their fold change, and the color represents the −log(adjusted P-values). Numbers indicate the enrichment correlation of the gene list with P-adjusted values in parentheses. Overlaps in shared upregulated genes are found in the top-right quadrant, and shared downregulated genes are found in the bottom-left quadrant. Discordant gene expression, upregulated by NMX females but downregulated by NMX males are found in the top left quadrant, whereas genes upregulated in NMX males but downregulated in NMX females are found in the bottom right quadrant.

Abbreviations: ARC, arcuate nucleus; NMX, NeuroMix; VMN, ventromedial nucleus.

To further explore concordant and discordant gene expression patterns, we identified genes that were differentially expressed in both males and females within each tissue, rather than relying solely on overall expression trends as in the rank-rank hypergeometric overlap analysis. Overall, there were few overlapping DEGs between sexes, but the pituitary showed the highest number, with 8 shared genes, 7 of which were concordantly regulated [Supplemental Fig. S6 (48)]. Specifically, Kmt5a, Pip4k2a, and Usf3 were upregulated in both NMX males and females, while Pnpia7, Sbf1, Zfp384, Snfa1, and Cytip were consistently downregulated.

Gonads

Testes transcriptomics

The testes were less robustly affected by NMX than the ovaries, with 268 differentially expressed testicular genes identified (2.79% of the total genes in our analysis; Fig. 2G). Of the 268 DEGs, 116 of them were upregulated in NMX males, while the other 152 genes were downregulated in NMX males compared to control males. A list of all genes including those that were differentially expressed in the testes of NMX compared to controls can be found in Supplemental Table S7 (48). Few endocrine-related genes were differentially expressed in the testes, with the exception of Igf1 and IGF receptor 1 (Igf1r), both of which were downregulated (Fig. 4). Notably, expression of these genes also did not correlate with serum testosterone levels in males [Supplemental Fig. S3D (48 )]. However, GO-enrichment analysis revealed that genes upregulated in the testes of NMX males compared to controls were associated with regulation of RNA splicing and mRNA processing. Genes downregulated in NMX males were related to intracellular glucose homeostasis, response to sugars, carbohydrates, and E2 [Fig. 6A; Supplemental Table S3 (48)]. Interestingly, the GO term response to E2 was enriched in 3 of the 4 male tissues examined: the ARC, VMN, and testes. Among the genes associated with this GO term, only 2—hemoglobin subunit α 2 (Hba-a2) and collagen type I α 1 chain (Col1a1)—were shared across all 3 tissues. Both genes were downregulated in these regions; however, after correction for multiple hypothesis testing, only Col1a1 remained significantly downregulated in NMX males in the ARC and testes.

Figure 6.

Significantly enriched GO terms related to biological processes are displayed for testes (A) and ovaries (B). The Y-axis represents GO term descriptions, while the X-axis shows the gene ratio, which indicates the proportion of input genes associated with each GO term. Dot size corresponds to −log10(eFDR), and color represents gene expression direction: open yellow circles for GO terms from downregulated genes and filled purple circles for those from upregulated genes. All effects are shown for NMX compared to controls.

Abbreviations: eFDR, empirical false discovery rate; GO, gene ontology; NMX, NeuroMix.

WGCNA of testes

In the testes, a sufficient sample size (n = 23) allowed for WGCNA analysis, which detects correlations among genes with similar expression patterns and organizes them into modules designated by arbitrary colors. In the testes, WGCNA identified 11 distinct modules [Supplemental Fig. S2 (48)]. Two of the 11 modules—turquoise (β = .19 ± .08, P  = .039) and red (β = .20 ± .08, P  = .023)—showed significantly different deMEGs between control and NMX testis samples (Fig. 7). The turquoise module had significantly lower deMEGs in the NMX samples, with the 1301 genes in this module related to energy metabolism as determined by GO-enrichment analysis [Supplemental Table S9 (48)]. Eighty-five of the 1301 genes were considered hub genes in the turquoise module, but the only hub gene that was significantly lower in the NMX testis compared to control was heat shock protein β-9 [Hspb9; MM = 0.90, log2 fold change = −0.20; Supplemental Table S10 (48)]. The red module had significantly higher deMEGs in the NMX samples. Thirty-six of the 713 genes (5%) in the red module were considered hub genes, with 20 hub genes being upregulated in the NMX testes compared to controls [Supplemental Table S9 (48)]. GO-enrichment analysis revealed genes in this module were associated with RNA processing [Supplemental Table S10 (48)].

Figure 7.

Summary of WGCNA in testis. Box plots represent the median and IQR of module eigengene expression by treatment, and points represent individual subjects. For each module, the biological processes GO terms are listed. *P-values between .01 and .04, †P-values between .05 and .08.

Abbreviations: GO, gene ontology; IQR, interquartile range; WGCNA, weighted gene coexpression network analysis.

Two other modules—brown (β = .17 ± .09, P  = .071) and blue (β = .17 ± .09, P  = .066)—were trending to have an effect of treatment (Fig. 7). The brown module contained 840 genes and had significantly higher deMEGs in NMX samples. Of the 60 hub genes in the brown module, only 10 (6%) were significantly upregulated in NMX testes compared to controls [Supplemental Table S9 (48)]. GO-enrichment analysis revealed genes in the brown modules to be related to epigenetic regulation and DNA processing [Supplemental Table S10 (48)]. The blue module had significantly lower deMEGs in NMX testis samples compared to controls. One hundred twenty-five of the 1259 (10%) were considered hub genes [Supplemental Table S9 (48)]. The only hub gene that was significantly downregulated in NMX testis was exoribonuclease family member 3 (Eri3; MM = 0.82, log2 fold change = −0.24).

Ovary transcriptomics

Of the HPG tissues in females, the ovaries showed the most robust differential expression between NMX and vehicle control females. Notably, when accounting for all significant genes with a log2 fold change above 0.2, there were a total of 1295 DEGs (12.6% of the total genes in our analysis; Fig. 2H). A complete list of genes including DEGs found in the ovaries of NMX compared to control females can be found in Supplemental Table S8 (48). As in the brain, serum E2 levels did not correlate with endocrine gene expression in the ovaries [Supplemental Fig. S5D (48)]. Only one endocrine-related gene, liver receptor homolog 1 (Nr5a2), was differentially expressed (upregulated) in females exposed to NMX (Fig. 4).

GO-enrichment analysis revealed that those genes upregulated in NMX females were involved in cilium assembly, organization, and movement, represented by significant DEG including dynein axonemal heavy chain 12 (Dnah12), dynein axonemal heavy chain 11 (Dnah11), dynein axonemal intermediate chain7 (Dnai7), cilia and flagella associated protein 65 (Cfap65), cadherin EGF LAG seven-pass G-type receptor 2 (Celsr2), glutamate-rich protein 3 (Erich3), and others (Fig. 4). These GO terms also appear to reflect genes that are downregulated in the female VMN. When examining genes associated with cilium-related GO terms across both tissue types, we found 29 overlapping genes. Among these, 4 genes—Cfap161, Spef1, coiled-coil domain containing protein 39 (Ccdc39), and TRAF3 interacting protein 1 (Traf3ip1)—were significantly differentially expressed in both the female VMN and the ovary. Notably, these genes were downregulated in the VMN and upregulated in the ovary. To assess whether mate preference behavior was associated with this gene expression, we examined the relationship between hormone stimulus preference, measured by time spent nose-touching during the mate preference test, and the expression of cilia-related genes. This behavioral metric was chosen based on previous findings indicating it was significantly altered in females prenatally exposed to NMX. Significant group-by-expression interactions were observed for several cilia-related genes in the VMN but not in the ovary [Supplemental Fig. S7 (48)]. Specifically, Ccdc40 (β = 1.49 ± .54, P = .016), Cfap65 (β = 2.66 ± .74, P = .003), and Dnah11 (β = 1.78 ± .56, P = .006) exhibited opposite patterns across treatment groups: lower expression was associated with reduced hormone preference in NMX-exposed females but with increased hormone preference in controls. Genes downregulated in NMX ovaries were involved in adaptive immune response and leukocyte-mediated immunity according to a GO-enrichment analysis, with genes including RT1 class II, locus DMb (RT1-DMb), locus DMa (RT-DMa), locus Db1 (RT-Db1), macrophage expressed 1 (Mpeg1), C-C motif chemokine ligand 7 (Ccl7), coronin 1A (Coro1a), and others (Fig. 4).

WGCNA of ovaries

Again, in the ovary, there was an adequate sample size (n = 23) to conduct WGCNA analysis. WGCNA identified 25 distinct modules [Supplemental Fig. S1 (48)]. Five of these modules—turquoise (β = .219 ± .075, P = .008), blue (β = .201 ± .080, P < .020), yellow (β = .219 ± .080, P = .013), salmon (β = .217 ±.080, P = .013), and midnight blue (β = .213 ± .082, P = .018)—showed an effect of treatment (Fig. 8). The turquoise module was the only one to have significantly greater differentially expressed module eigengenes (deMEGs) in the NMX females. Two hundred ninety-two of the 1601 genes in this module were identified as hub genes, with 221 (76%) being upregulated in the NMX females compared to vehicle [Supplemental Table S11 (48)]. GO-enrichment analysis revealed these genes to be involved in cilia organization and assembly [Supplemental Table S12 (48)]. The rest of the modules showed downregulation in NMX compared to vehicle females. For the blue module, 157 of the 952 genes in this module were considered hub genes, and 40 (25%) of the hub genes were significantly downregulated in the NMX group [Supplemental Table S11 (48)]. GO-enrichment analysis reported these genes to be involved in RNA-related processes [Supplemental Table S12 (48)]. The yellow module, with 142 of its 804 genes considered hub genes, also had significantly lower deMEGs in the NMX females [Supplemental Table S11 (48)]. Of these 142 hub genes, 108 (76%) showed lower differential expression in the NMX females. GO-enrichment analysis revealed these genes were involved in muscle processes [Supplemental Table S12 (48)]. The salmon module contained 237 genes, with 37 considered hub genes, again with significantly lower deMEGs in the NMX females [Supplemental Table S11 (48)]. Only 11 of the 37 hub genes were differentially expressed and downregulated in the NMX females. GO-enrichment analysis revealed these genes to be involved in ribosome biogenesis regulation of chromosome organization [Supplemental Table S12 (48)]. Midnight blue was the last module that showed an effect of treatment. 21 of the 228 genes in the midnight blue module were considered hub genes with 12 of these hub genes downregulated in the NMX females [Supplemental Table S11 (48)]. GO-enrichment analysis showed genes in this module were related to adaptive immune response and the cytokine-mediated signaling pathway [Supplemental Table S12 (48)].

Figure 8.

Summary of WGCNA in ovaries. Box plots represent median and IQR of module eigengene expression by treatment, scatter plots indicate module eigengene relationship with serum estradiol, and points represent individual subjects. For each module, the biological processes GO terms are listed. *P-values between .01 and .05, **P-value < .01.

Abbreviations: GO, gene ontology; IQR, interquartile range; WGCNA, weighted gene coexpression network analysis.

Two additional modules, black and brown, showed an effect of treatment and had a relationship with serum E2 (Fig. 8). The black module had significantly higher deMEGs (β= .209 ± .060, P = .003) in the NMX females compared to controls as well as a positive associated with serum E2 (Linear Mixed Effects Model [LMM]: β= .007 ± .002, P = .003, correlation: R² = 0.57, P < .001). The black module contained 350 genes; 35 of the 42 (83%) hub genes found in the black module were upregulated in the NMX females. GO-enrichment revealed genes in the black module were associated with regulation of the leukocyte apoptotic process. The brown module had significantly lower deMEGs (β = .164 ± .071, P = .033) in the NMX females and a negative association with serum E2 (LMM: β = .006 ± .002, P = .020, correlation: R² = 0.51, P = .01). One hundred eighty-eight of the 806 genes in the brown module were considered hub genes, with 14 of them also downregulated in the NMX females. GO-enrichment analysis showed genes in the brown module were involved in metabolites and energy production and cellular and aerobic respiration.

Two modules, dark green and cyan, had a significant relationship only with serum E2 with no effect of treatment (Fig. 8). The dark green module contained 160 genes with only 5 hub genes displaying a positive relationship with serum E2 (LMM: β = .006 ± .003, P = .029, correlation: R² = 0.52, P = .01), and genes in this module were revealed to be involved in cellular response to peptide hormone stimulus as responses to insulin and E2 by GO-enrichment analysis. The cyan module, which contains 233 genes, showed a negative relationship with serum E2 (LMM: β = .006 ± .003, P = .045, correlation: R² = 0.43, P = .03). No hub genes or GO terms were identified in the cyan module.

IPA

IPA is another tool for analyzing -omics data to identify DEGs and pathways. A full list of upstream regulators and pathways for each tissue type for both sexes can be found in Supplemental Tables S13 to S20 (48). These results were found from DEGs in NMX animals compared to same-sex controls in each tissue.

From IPA, of all the pathways identified, the one that was affected in most tissues of both sexes was the β-E2 pathway. In males, this pathway was significantly inhibited by NMX in male rats in the ARC (z = −2.704, P < .001), VMN (z = −1.653, P = .015), and testis (z  = −0.699, P = .001), and it was activated in the NMX male pituitary (z = 2.274, P = .024). In females, the β-E2 pathway was activated in the ARC (z = 1.026, P = .015), VMN (z = 3.522, P < .001), pituitary (z = 1.468, P = .012), and ovary (z = 0.183, P  < .001) tissues. This pathway is important for a variety of physiological functions including reproductive health. In the male ARC, which was the tissue most affected by NMX in this sex, the testosterone pathway was also inhibited (z = −2.271, P = .006). Interestingly, in the testes, genes upstream of the dihydrotestosterone (z  = −1.938, P = .053) pathway were inhibited in NMX males compared to controls. Notably, many endocrine genes, although not significant after multiple comparison correction, are downregulated in the male ARC (Fig. 3). Endocrine genes did not show consistent patterns across other tissue types in either sex (Fig. 3).

In agreement with GO enrichment, IPA also found that genes downregulated in NMX ovaries were those that inhibit immune pathways for IL-12 β (Il12b; z = −2.423 P  < .001), IL-18 (Il18, z = −2.407, P  < .027), nuclear factor-kappa B (Nfkb, z = −0.819, P  < .019), and others. Notably, immune signaling pathways appear downregulated in WGCNA as well.

Discussion

Developmental exposure to the EDC mixture NMX was previously shown to affect development and behaviors related to anxiety-like behaviors, sociability, and mate preference in rats (24). The current study extended those aspects of work related to the reproductive neuroendocrine axis through transcriptomics of hypothalamic, pituitary, and gonadal tissues and measuring serum testosterone (males) or E2 (females) from the same experimental rats. Our new results demonstrated that early-life NMX is associated with hormonal and molecular changes in adulthood. Males treated with NMX had significantly higher testosterone than control males. Serum E2 in adult females was not affected by the NMX treatment (although it correlated with some gene expression changes in the ovary). Gene expression across the HPG axis was significantly altered by NMX in a tissue- and sex-dependent manner. Categories of genes commonly identified were related to cilia organization, endocrine systems, epigenetic regulation, immune processes, metabolism, and synaptic signaling. It was also informative to compare results in males and females for those tissues analyzed in both sexes. In the 2 subregions of the hypothalamus, the VMN and the ARC, NMX affected gene expression in a sex-dependent manner. In the pituitary, we found largely concordant patterns of gene expression in NMX males and females when compared to same-sex controls. For the gonads, ovary tissues had strikingly greater differential gene expression than testes. Overall these results indicate that the HPG axis of each sex is (mostly) uniquely sensitive to prenatal NMX exposure.

Sex Differences in EDC Effects in the Hypothalamus

Prior studies have reported sexually dimorphic effects of EDC exposure on gene expression in the hypothalamus. Prenatal exposure to the polychlorinated biphenyl mix Aroclor 1221 affected gene expression in the medial preoptic nucleus, particularly in female rats (50). A1221 also altered the expression of numerous estrogen-sensitive genes in the anteroventral periventricular nucleus of females but not males (51). However, the same A1221 exposure increased gene expression in the ARC in males but not females (52).

There are sex differences in effects of other EDCs on hypothalamic gene expression. Females rats exposed to diethylstilbestrol, methoxychlor, and bisphenol A (BPA) during pre- and postnatal periods had molecular changes in the ARC during adulthood (53). In male rats, prenatal exposure to di-(2-ethylhexyl) phthalate (DEHP) affected gene expression in the multiple nuclei of the hypothalamus, including the anteroventral periventricular nucleus, ARC, and medial preoptic nucleus (54).

Our current study extended previous work through bulk transcriptomics of the ARC and VMN of rats exposed to NMX or vehicle and confirmed that most gene expression changes were unique to each sex. Notably, we saw a discordant pattern of gene expression hypothalamic subregions VMN and ARC between the 2 sexes, without showing any significant difference in the number of DEGs within each sex. These findings suggest that single EDCs or a mixture of EDCs can lead to sex-specific changes in gene expression.

NMX Induced Transcriptomic Changes in Hypothalamus, Especially the Male ARC, Related to Synaptic Plasticity and Metabolism

One of the most notable effects of NMX exposure was observed for genes related to synaptic function and metabolic signaling in the ARC. The ARC is a central hub for integrating various hormonal signals to help regulate energy balance and reproduction. Changes in synaptic plasticity and in number or magnitude of synaptic connections are influenced by sex hormones and energy status (55, 56). We show that a number of GABA and glutamatergic receptors had increased gene expression in the ARC of males treated with NMX. Previous studies in rodents revealed higher GABA and glutamate levels in the whole brain as well as amygdala and hippocampus subregions after perinatal exposure to BPA (57, 58). This increase in inhibitory and excitatory receptors suggests modulations in neuron communication in the ARC of males treated with NMX, which can influence downstream signaling in the HPG axis.

Among the affected genes of interest, Bdnf (upregulated) and Oxt (downregulated) exhibited opposing patterns of expression in the ARC of NMX-treated males; notably, both are implicated in modulating synaptic strength and neurotransmission (59, 60). Other EDC work demonstrated increased Bdnf in the hippocampus after prenatal exposure to dibutyl phthalate; this increase in gene expression was associated with improved spatial learning (61). Several studies have also investigated EDC effects on the oxytocin system. Notably, EDCs upregulate the release of oxytocin (62) without altering the number of oxytocin neurons (63) in the hypothalamus. Oxytocin receptor gene expression was decreased in male mice exposed to BPA prenatally leading to a deficiency in social behaviors (64). This downregulation of Oxt in the current study may be a result of negative feedback, reducing transcription of the gene to avoid overproduction of the hormone being released to restore normal signaling in the brain.

Additionally, in the ARC of males treated with NMX, we saw many genes related to metabolic signaling downregulated. The downregulation of Ghrh, Lepr, Pomc, and other metabolic genes suggest that prenatal exposure to a mixture of EDCs may lead to a disruption in energy metabolism. In male rodents, prenatal (54) and adult (65) EDC exposure has been shown to reduce body weight and influence the downregulation of metabolic gene expression. We have previously demonstrated that prenatal NMX exposure affects male body weight during both infancy and adulthood. Notably, body weight at both timepoints was significantly correlated with Ghrh expression levels. There has been an increasing interest in how EDCs contribute to the perturbation of energy metabolism and consequently lead to health problems and metabolic disease (66-68). Our results support the hypotheses that EDC exposure can contribute to changes in metabolic health.

Compared to the male ARC, relatively few genes were differentially expressed in the VMN of NMX-treated males. Nevertheless, genes involved in synaptic plasticity and metabolic regulation were also affected in the VMN, albeit to a lesser extent than in the ARC. In female ARC, the same gene categories identified in the GO analysis were significantly altered by NMX treatment but in the opposite direction compared to males. However, individual genes did not reach statistical significance following correction for multiple comparisons, despite the overall expression trends appearing directionally opposite to those observed in the male ARC. Together, these findings suggest that the ARC is more sensitive than the VMN to NMX exposure in males and that this sensitivity is sex-specific. NMX may disrupt hypothalamic function through both region- and sex-dependent mechanisms—potentially interfering with key transcriptional programs that regulate metabolism and synaptic plasticity in males while triggering an opposing or compensatory response in females.

In the female VMN, genes involved in cilia organization and assembly emerged as a distinct category of interest, with several individual genes within this pathway also showing differential expression. In the brain, more work has been done on nonmobile primary cilia, which are found on the surface of mammalian cells and play an important role in neuronal signaling (69). The malfunction of primary cilia in the brain can contribute to a number of diseases impacting brain development, neurogenesis, and neuronal migration (69-71). Prenatal exposure to A1221 was previously shown to modulate gene expression related to cilia regulation in the hypothalamus of P1, P30, and P60 female rats (17). These findings suggest that developmental exposure to NMX might alter cilia-related pathways involved in neuronal development and signaling in females, potentially leading to long-term disruptions in hypothalamic function, an effect not observed in the male VMN.

Few Effects of NMX in the Pituitary Gland

Few genes of interest were differentially expressed in the pituitary of male and females treated to NMX compared to same-sex controls. Others have found that DEHP (20), BPA (72), and other bisphenols (73) lead to transcriptome-level changes in reproductive-related genes in the pituitary of males and female mice. However, these studies focused on single EDC exposures, whereas our study investigates the effects of an EDC mixture, NMX. The lack of observed gene expression changes in response to the NMX mixture may be due to potential interactions between the chemicals in the mixture, which could alter the expected biological responses compared to single-compound exposures. Interestingly, the pituitary exhibited large, concordant patterns of gene expression between NMX-treated males and females compared to their respective same-sex controls. This suggests that, in contrast to typical sex-specific transcription patterns observed under normal physiological conditions (74), both sexes experience similar gene expression changes in this region. These findings suggest that prenatal NMX exposure may be causing alterations in pituitary function that override typical sex-specific transcriptional regulation. This could result from NMX disrupting the hormonal signaling pathways that usually control sex-specific gene expression, leading to a more uniform gene expression profile across sexes in the pituitary.

In the Gonads, NMX Exposure Elicited a Much More Robust Transcriptomic Response in Ovaries Than in Testes

Extending down the HPG axis, perinatal (21) and adolescent (75) exposures to individual EDCs have been shown to alter gene expression and morphology in rodent ovaries. In agreement with this previous work, we found numerous molecular changes in the ovaries of female rats receiving NMX prenatally when compared to those receiving vehicle.

Many of the genes downregulated in the ovaries of NMX females were involved in immune regulation, a result that was consistent across all of the female analyses including differential gene expression and GO enrichment as well as in WGCNA, with 1 module involved in adapted immune response and cytokine signaling displaying a significantly lower deMEG in NMX females. Other studies have revealed effects of EDCs on immune function. BPA, phthalates, and pesticides but not polychlorinated biphenyls increased circulating immune and inflammatory markers in humans (5). Phthalate mixtures also altered inflammatory cytokines in human follicle fluid (76). DEHP (up to 8 nM) has been shown to increase expression of immune cell gene markers in human ovarian follicles in vitro (77), and intraperitoneal injections of 20 or 40 µg/kg similarly elevated immune-related gene expression in mouse ovaries (78). As a whole, this EDC research, including ours, demonstrates that exposure leads to increased inflammation and immune-regulated gene expression changes in the ovary. Further research is needed to determine whether this increase in immune-related gene expression negatively impacts reproductive health.

Interestingly, we also identified differential expression in gene modules associated with smooth muscle function. Although these changes were observed in the ovary, they may reflect broader disruptions in reproductive tract physiology, particularly within the uterus. Dysregulation of smooth muscle-related genes could interfere with key contractile processes necessary for ovulation, oocyte transport, and uterine receptivity, each essential for successful reproduction (79, 80). These results suggest that the effects of prenatal EDC exposure on fertility may not be limited to follicular development but may also involve altered smooth muscle signaling pathways that are crucial for coordinating reproductive function.

In male rats, NMX had limited overall effects on differential gene expression in the testes. However, WGCNA revealed 2 modules with significantly increased expression of deMEGs related to RNA modifications: one associated with RNA processing and the other with epigenetics and DNA methylation. In stark contrast, the WGCNA analysis in female ovaries identified a gene cluster involved in RNA processing that showed significantly reduced expression in NMX-exposed females. These opposing patterns—upregulation in males and downregulation in females—highlight a clear sex difference in the response to NMX, suggesting that EDC exposure may differentially disrupt the regulation of RNA processing and downstream gene expression pathways depending on sex.

Despite few gene expression effects in the testes of NMX males, particularly among classical steroidogenic genes, we detected significant downregulation of Igf1 and Igf1r. This pattern may reflect disrupted testicular development resulting from prenatal exposure to NMX, consistent with previous findings illustrating the critical role of these genes in normal testis function (81). These results align with broader evidence from other EDCs literature demonstrating altered gene expression regulating the production of androgens in the testis (82). Specifically, adult rats exposed to perfluorododecanoic acid (83) and fetal rats exposed to dibutyl phthalate (84) showed lower expression of many cholesterol transporter genes as well as genes involved in testosterone biosynthesis in the testes.

NMX Induced Similar Changes in Cellular Respiration But Opposing Effects on Ciliary Pathways in the Ovaries and Testes

Gene clusters related to energy metabolism and cellular respiration found by WGCNA were downregulated in both the testis of NMX males and the ovaries of NMX females compared to same-sex controls. The majority of the EDC literature focusing on energy metabolism in the gonads has been done in males, showing inhibition of glucose uptake (85) and decreases in respiratory chain complex enzymes and proteins (86, 87) in testis, with concomitant negative impacts on male infertility. There is little research in this arena on the ovary in terms of EDC effects, but the current NMX results should be considered in the context that energy homeostasis in the ovary influences female fertility (88). As a whole, our results suggest that EDC exposure negatively impacts cellular processes within the gonads.

Another interesting target revealed in the current study was cilia assembly and organization, which were upregulated across multiple analysis in the NMX females. Interestingly, in the testes WGCNA found a cilia gene cluster showing downregulation in males exposed to NMX. Mobile cilia assembly and function are essential in both male and female reproductive tracts. In males, cilia help suspend semen, and, in females, cilia help move oocytes in addition to capturing and binding semen (89, 90). Molecular changes that lead to disruptions in cilia functioning can alter fertility in both males and females (89). Notably, many of the same cilia genes were downregulated in the female VMN (see previous discussion on hypothalamus) with no changes seen in any other male tissue. Although the effects of EDCs on cilia within the HPG axis remain underexplored, our findings indicate that EDC exposure may disrupt cilia organization and assembly across the axis, potentially altering reproductive processes.

In addition to gene expression changes affecting structural processes like cilia organization, EDC exposure also alters the expression of genes involved in key hormonal signaling pathways, suggesting broader disruption of the HPG axis. Analysis of the full gene sets using IPA revealed activation of β-E2 signaling across all tissues in NMX females, while this pathway was inhibited in NMX male tissues, except for the pituitary, where it was activated. These patterns suggest sex-specific differential expression of upstream and downstream components regulating β-E2 signaling, contributing to its tissue-specific modulation. β-E2 signaling is important in reproductive systems, brain function, and metabolism and can impact gene expression and cell signaling (91). Since EDCs can affect estrogen signaling through a range of direct and indirect mechanisms, predicting their effects on this pathway is challenging (92). Research suggests that both the timing of exposure (93) and the specific EDC involved can alter hormone activity, leading to downstream disruptions of the HPG axis (94-96).

Conclusions

In conclusion, this study demonstrates that prenatal exposure to the EDC mixture NMX induces significant physiological and molecular changes across the HPG axis of rats in adulthood, with distinct sex-dependent effects. Males exhibited increased serum testosterone and the greatest number of changes in the hypothalamic ARC, with upregulation in synaptic plasticity markers and downregulation of metabolic genes. Females showed the most robust transcriptional alterations in the ovaries, particularly in immune and cilia regulation. Across the HPG axis, NMX influenced gene expression linked to synaptic signaling, metabolism, and cilia organization, with notable sex differences in hypothalamic subregions and pituitary gene expression. These findings highlight the vulnerability of the developing HPG axis to environmental perturbations and underscore the importance of considering sex-specific outcomes when evaluating the long-term impacts of EDC exposure.

Funding

Grant support: National Institutes of Health R01 ES029464, R35 ES035024.

Disclosures

A.C.G. is an expert witness on a legal case related to polychlorinated biphenyls and is on the board of directors of the Endocrine Society. The other authors have nothing to disclose.

Data Availability

Original data generated and analyzed during this study are included in this published article or in the data repositories listed under References.