Srd5a1 is Differentially Regulated and Methylated During Prepubertal Development in the Ovary and Hypothalamus

Abstract

5α-reductase-1 catalyzes production of various steroids, including neurosteroids. We reported previously that expression of its encoding gene, Srd5a1, drops in murine ovaries and hypothalamic preoptic area (POA) after early-life immune stress, seemingly contributing to delayed puberty and ovarian follicle depletion, and in the ovaries the first intron was more methylated at two CpGs. Here, we hypothesized that this CpG-containing locus comprises a methylation-sensitive transcriptional enhancer for Srd5a1. We found that ovarian Srd5a1 mRNA increased 8-fold and methylation of the same two CpGs decreased up to 75% between postnatal days 10 and 30. Estradiol (E₂) levels rise during this prepubertal stage, and exposure of ovarian cells to E₂ increased Srd5a1 expression. Chromatin immunoprecipitation in an ovarian cell line confirmed ESR1 binding to this differentially methylated genomic region and enrichment of the enhancer modification, H3K4me1. Targeting dCas9-DNMT3 to this locus increased CpG2 methylation 2.5-fold and abolished the Srd5a1 response to E₂. In the POA, Srd5a1 mRNA levels decreased 70% between postnatal days 7 and 10 and then remained constant without correlation to CpG methylation levels. Srd5a1 mRNA levels did not respond to E₂ in hypothalamic GT1-7 cells, even after dCas9-TET1 reduced CpG1 methylation by 50%. The neonatal drop in POA Srd5a1 expression occurs at a time of increasing glucocorticoids, and treatment of GT1-7 cells with dexamethasone reduced Srd5a1 mRNA levels; chromatin immunoprecipitation confirmed glucocorticoid receptor binding at the enhancer. Our findings on the tissue-specific regulation of Srd5a1 and its methylation-sensitive control by E₂ in the ovaries illuminate epigenetic mechanisms underlying reproductive phenotypic variation that impact life-long health.

Keywords: 5α reductase-1; Srd5a1; epigenetic; hypothalamus; methylation; ovary.

Figure 1.

Srd5a1 is differentially regulated in ovaries and hypothalamus across the lifespan. (A, B) Srd5a1 mRNA levels in (A) ovaries and (B) the hypothalamic preoptic area (POA) of mice at various ages (from 1 to 4 litters at each time point; for >50 days old, mice in each group were not identical ages and the average age is shown). The mRNA levels were normalized to those of Rplp0 and are shown relative to levels at the first time point; mean ± standard error of the mean (SEM; some of the ovarian data are from [9]). (C-F) The % DNA methylation (% of cytosines methylated out of the total number sequenced at the same site) measured by bisulfite conversion followed by MiSeq deep sequencing at the (C, D) first and (E, F) second CpG in (C, E) ovaries and (D, F) POA. In all graphs, n-values at each point are shown; P > .05 (ANOVA, Tukey-Kramer t-test) for groups sharing the same letter. Shaded boxes mark periods of significant change.

Figure 2.

E₂ increases Srd5a1 expression in ovarian granulosa cells, and estrogen receptor (ESR1) binds the locus of the differentially methylated CpGs at a transcriptional enhancer. (A) Srd5a1 mRNA levels in mouse ovarian primary cell culture (mice were 30 days old) after exposure to E₂ (100 nM) for 24 hours; mean ± SEM, n = 3, 4; **P < .01. (B) Chromatin immunoprecipitation (ChIP) for ESR1 in murine ovarian granulosa KK-1 cells with or without 2 hours exposure to E₂ (10 nM) followed by qPCR for loci across the 5′ end of the gene and first intron, with Pgr as positive control. IP/input levels are mean ± SEM (n = 7, except at 223-371, 676-828, and Pgr where n = 4). Student t-test compared treated and nontreated groups, *P < .05. (C) Schematic (adapted from UCSC genome browser; http://genome.ucsc.edu) showing the locus of the 5′ end of the Srd5a1 gene (in continuous dark bar: UTR [thin bar], first exon [thick bar], and part of first intron [thin line]) in the mouse genome, with the location of CpG1 and CpG2, and ½ ERE motif sites marked (arrows). The CpG island and several regions identified by ENCODE as proximal enhancer-like sequences (pELS) are shown, as well as the regions we found enriched for ESR1 or H3K4me1 in KK-1 cells (diamonds, centered on the amplicon center and in accordance with the resolution determined by sonication). (D) ChIP assay for H3K4me1, performed and presented as in Fig. 2B (n = 4); ^###P < .001 (ANOVA, Tukey-Kramer t-test) compared with all other means.

Figure 3.

In ovarian KK-1 granulosa cells, methylation of the intronic enhancer CpG2 prevents E₂ stimulation of Srd5a1 expression. (A) DNA methylation (bisulfite conversion and sequencing) at the Srd5a1 promoter (−123 to +62 bp) and intronic enhancer (+870 to +1081, including CpG1 and CpG2) in ovarian KK-1 cells: each column represents a single CpG site, and each row a repeat; black circles represent CpGs that are methylated and white circles those that are not. (B) Targeted DNA methylation was performed by stable expression of a FLAG-tagged dCas9-DNMT3A catalytic domain, recruited to the enhancer of Srd5a1 by two site-specific gRNAs (thick green lines). (C) ChIP assay for FLAG peptide in KK-1 cells stably expressing dCas9-DNMT3A-FLAG, after transfection with the gRNAs or empty vector, followed by qPCR for the Srd5a1 intronic enhancer, an upstream region and Gapdh as controls; IP/input levels presented as in Fig. 2B (n = 3). Student t-test compared levels in cells with and without transfection of the gRNAs; * < 0.05. (D) Levels of DNA methylation (measured by bisulfite conversion and MiSeq) in these cells at CpG1 and CpG2 of the Srd5a1 intronic enhancer, shown relative to those in control cells (no gRNAs); mean ± SEM (n = 3). (E) Srd5a1 and Greb1 mRNA levels in these cells with or without E₂ (10 nM, 24 hours). The mRNA levels were analyzed and presented as before (n = 3); **P < .01, ***P < .001 compared with untreated controls.

Figure 4.

In GnRH neuronal GT1-7 cells, Srd5a1 mRNA levels are not affected by E₂ even after reduction in intronic enhancer CpG1 methylation. (A) DNA methylation at the intronic enhancer and ∼300 bp adjacent upstream region in the POA of 7- and 60-day-old female mice was performed and is presented as % methylation, mean ± SEM (n = 4); small or capital letters designate statistical tests for each age group separately (Kruskal Wallis, Dunn test); CpG (0) had no detectable methylation. (B) DNA methylation was assessed at the Srd5a1 promoter (−124 to +62 bp) and intronic enhancer (CpG1 and CpG2) in GT1-7 cells, as in Fig. 3A and is presented similarly. (C, D) Srd5a1 mRNA levels in (C) KK-1 and GT1-7 cells, or (D) GT1-7 cells after 10 to 100 nM E₂ exposure (Greb1 serves as positive control), measured and presented as before (n = 4); *P < .05, ***P < .001. (E) Targeted demethylation was performed by overexpression of dCas9-TET1 catalytic domain, recruited to the Srd5a1 enhancer by the same site-specific gRNAs as in Fig. 3B (thick green lines). (F) DNA methylation (bisulfite conversion and MiSeq) at CpG1 and CpG2 of the Srd5a1 intronic enhancer in the GT1-7 cells expressing the dCas9-TET1. Levels are presented relative to those in control cells (no gRNAs); mean ± SEM (n = 3); *P < .05. (G) Srd5a1 mRNA levels in similarly transfected cells, with or without exposure to E₂ (10 nM), presented as before (n = 3); P > .05 in t-test for all comparisons.

Figure 5.

The glucocorticoid, dexamethasone, represses Srd5a1 expression in GT1-7 neuronal cells, and the glucocorticoid receptor is found at the intronic enhancer. (A, B) Srd5a1 mRNA levels in GT1-7 cells after (A) 1-100 nM dexamethasone (Dex; n = 6-7) for 24 hours, with Gnrh and Fkbp5 as controls, or (B) 10 nM Dex for 24 to 72 hours (n = 3-4); data analyzed and presented as before. (C) ChIP assay for GR in GT1-7 cells after Dex exposure (10 nM, 24 hours), and qPCR for identification of binding at the Srd5a1 promoter, enhancer, and additional putative sites (from chip-atlas.org), with Fkbp5 as positive control, performed and presented as in Fig. 2B (n = 3). Student t-test compared treated and nontreated groups; ***P < .001, otherwise P > .05.