To assess GR expression in the developing female germline, we performed immunofluorescence (IF) on mouse fetal ovaries ranging from embryonic day (E) 12.5 through E18.5. Female germ cells, identified by either Oct4-GFP (Pou5f1-ΔPE-eGFP transgene) or TRA98, showed robust levels of GR starting at E13.5, which waned as development progressed (Figure 1A). GR consistently localized to the nucleus of fetal germ cells, irrespective of developmental stage, suggesting that GR may function as a transcription factor within the female germline (Figure 1B). To rigorously quantify GR expression dynamics in germ cells, we normalized GR expression levels to DAPI signal at the individual cell level, using the membrane marker wheat germ agglutinin (WGA) for segmentation (Figure S1A). These studies confirmed that GR expression in the germline decreased from E13.5 to E18.5 (Figure 1C).

At postnatal day 0 (PN0), IF revealed a small number of oocytes at the cortex of the ovary with nuclear GR, with expression declining and virtually absent by PN2 (apart from sporadic cortical oocytes with cytoplasmic GR) (Figure S1B). GR was not expressed in postnatal oocytes between PN5 and PN21, nor in adult oocytes of all follicular stages (Figure S1C). In the somatic compartment, GR was virtually absent from FOXL2⁺ granulosa cells, but was strongly expressed in the theca cell layer (marked by α-SMA ³⁵) from as early as PN7 through adulthood (Figure S1D).

It has been hypothesized that spatiotemporal differences in GR expression across different tissues can be due to the use of alternative, non-coding Nr3c1 exon 1 splice isoforms in different cell types ³⁶. To explore this possibility, we looked for evidence of exon 1 alternative splicing in our paired-end RNA-seq data of E15.5 sorted Oct4-GFP⁺ germ cells and GFP⁻ somatic cells (see Figure 2D for details). Sashimi plots visualizing the junctions between exon 2 and known exon 1 variants demonstrated that while somatic cells exclusively use exon 1B, 1D and 1F, female germ cells use a much wider range of known exon 1 variants, including 1A, 1B, 1C, 1D and 1F (Figure 1D). Surprisingly, we also found evidence of three novel splice isoforms exclusively in the germ cells, labeled here as predicted exons 1α, 1β, and 1γ (Figure 1D, blue lines). These cell type-specific patterns of exon 1 isoform usage were validated using RT-PCR on bulk and sorted populations, with primers designed specifically to amplify each of the known and predicted exon junctions (Figure 1E). This wider diversity of Nr3c1 exon 1 variants in the female germline suggests that alternative promoters may be cooperatively regulating the expression of GR, thus leading to the more dynamic temporal regulation of GR seen in the germline in comparison to the soma.

The strong, nuclear localization of GR in germ cells of the fetal ovary led us to hypothesize that GR may be functioning as a transcriptional regulator of early oocyte development. This possibility was raised in a single-cell RNA-seq study of the developing mouse ovary that noted that GR expression is highly correlated with progression of fetal oocytes through prophase I ²³. In zebrafish, genetic ablation of GR led to accelerated ovarian aging and decline in fertility later in life ^37–39. To interrogate the potential role of GR in meiosis, we utilized a mouse line with a constitutive genetic deletion of GR. Nr3c1 exon 3 floxed mice were crossed to a constitutive β-actin Cre mouse line to generate a constitutive null allele of GR (henceforth referred to as “KO”). Nr3c1 exon 3 deletion has previously been shown to result in efficient loss of functional GR protein (reviewed extensively elsewhere ¹⁶), which we confirmed by IF at E17.5 (Figure 2A), immunoblot at E13.5 (Figure 2B), and qRT-PCR at E15.5 (Figure 2C). To functionally test whether meiotic progression was delayed or disrupted following loss of GR, we scored meiotic prophase I in spreads for E15.5 WT and KO ovaries stained for SYCP3, SYCP1 and γH2AX (Figure 2D). No significant changes in substage distribution could be detected between WT and KO nuclei, suggesting that loss of GR does not disrupt meiotic progression. qRT-PCR on E15.5 WT and KO bulk ovary tissue for a panel of genes known to have important roles in meiotic progression similarly revealed no changes in the transcription of meiotic genes (Figure 2E).

To assess whether GR regulates the transcriptional landscape of alternative cellular pathways in the female germline, we employed two orthogonal transcriptomic approaches. We first performed low-input RNA-seq on FACS-sorted Oct4-GFP⁺ germ cells from individual WT and KO ovaries at E17.5. While our differential gene expression analysis pipeline revealed robust expression differences between WT female and WT male germ cells (Figure 2F, i), we saw no statistically significant differentially expressed genes between female WT and KO germ cells (Figure 2F, ii). Conditional deletion of GR specifically in the germ cells using Oct4-CreERT2 yielded the same result (Figure 2F, iii). In tandem, we performed single-cell RNA-seq on E15.5 WT and KO germ cells. All expected cell types of the fetal ovary were detected in both WT and KO embryos (Figure S2A), and total Nr3c1 transcript was depleted in all cell types of the ovary (Figure S2B). Differential gene expression analysis of WT and KO cells within the germ cell cluster revealed an extremely low number of differentially expressed genes with minor fold changes (Figure S2C), corroborating our bulk RNA-seq results that the loss of GR has little effect on the transcriptional landscape of fetal oocytes.

While homeostatic GR signaling does not appear to regulate fetal oocyte gene expression or meiotic progression, we next asked whether the GR agonist dexamethasone (dex) could elicit a transcriptional response in the female germline. Pregnant dams were administered either 10 μg dex / g or saline vehicle by intraperitoneal (IP) injection daily from E12.5 to E15.5, coinciding with the window of highest GR expression in the female germline. We validated this dosing regimen by sufficiency to induce expression of the canonical GR response gene Fkbp5 in bulk ovary, testis, and lung tissue of E15.5 embryos (Figure S3A), confirming dex was able to transit the placenta to the embryos. Accordingly, we performed bulk RNA-seq on sorted Oct4-GFP⁺ germ cells and total GFP⁻ somatic cells from E15.5 ovaries of dex vs saline-treated dams. While the somatic cells showed a robust transcriptional response following dex treatment (1477 total differential genes with adjusted p-value ≤ 0.05), by comparison, the response was severely dampened in germ cells (156 total differential genes with adjusted p-value ≤ 0.05; Figure 3A). GO term analysis of differentially expressed genes from the somatic cells showed low fold enrichment across a wide range of unrelated categories, likely due to the heterogeneity of the GFP⁻ population (Figure S3B). The set of upregulated (i.e. dex-induced) genes in the germ cells did not show any significant GO term enrichment (Figure S3B). The few downregulated germ cell genes showed an enrichment for ribosomal assembly and translation (Figure S3B), although we did not detect significant changes in protein levels of any germ cell marker assessed. In somatic gonad cells, we verified that the dex-induced transcriptional changes led to changes at the protein level using an ex vivo culture system. Fetal ovaries dissected at E14.5 and cultured in hormone-depleted medium for 48 hours with and without 1 μM dex confirmed upregulation of PLZF (Zbtb16, the most highly upregulated gene in the somatic cells by RNA-seq) specifically in the Tg:Oct4-GFP⁻ soma and not the GFP⁺ germ cells (Figure 3B, i). Utilizing the same culture system, we analyzed meiotic spreads and confirmed that dex treatment does not alter meiotic prophase, in line with the lack of transcriptional response (Figure S3C).

The observed lack of a transcriptional response in the germ cells to both deletion of GR and dex treatment in contrast to adjacent somatic cells led us to hypothesize that the female germline may be resistant to GR signaling. As the function of GR as a transcription factor requires nuclear localization, we tested whether its subcellular localization is altered in response to ligand. In ovaries cultured ex vivo without dex, GR localized to the cytoplasm of virtually all germ cells; in contrast, dex-treated cultures retained nuclear localization similar to fetal oocytes in vivo (Figure 3B, ii). As GR localization still dynamically responded to the presence or absence of ligand, despite not robustly altering transcription, an alternative modification to the GR protein specifically in the germ cells could be accounting for this attenuated activity.

GR is a highly modified protein, with a wide range of transcriptional isoforms, translational isoforms, and post-translational modifications described to date ⁴⁰. Multiple translation initiation sites in the Nr3c1 gene lead to a variety of protein isoforms with varying truncations in the regulatory N-terminal domain of the protein ⁴¹. The most truncated forms of GR (the “GR-D” isoforms) can exert an inhibitory dominant negative effect ⁴¹. We used an antibody designed specifically to recognize all known GR isoforms to immunoblot E13.5 ovary lysate but did not detect any GR-D isoforms (∼50 kDa or less), confirming that this dominant negative isoform is not the cause of decreased GR activity in the germline (Figure 3C). In mice, alternative splicing and subsequent retention of Nr3c1 intron 8 results in a GRβ isoform that functions as a dominant negative regulator of the predominant GRα isoform ⁴². However, we saw no evidence of intron 8 retention in our paired-end RNA-seq reads (Figure 3D), suggesting that the GRβ isoform is also not the cause of attenuated GR activity in fetal oocytes.

We next characterized the expression pattern of GR in the testis by IF from E12.5 through E18.5. In stark contrast to the female, germ cells of the fetal testis (marked by Tg:Oct4-GFP) showed no GR expression during early sex differentiation (Figure 4A). It was not until E17.5 that male germ cells began to express GR, and by E18.5, over 90% of all germ cells harbored GR in the nucleus (Figure 4A, 4B). We employed a similar quantitative imaging analysis as in the female to confirm that GR expression peaked in the male at E18.5 (Figure 4C).

As GR levels in prospermatogonia peaked at the end of fetal development, we next asked whether expression was maintained into postnatal development. Staining between PN0 (not shown) and PN2 revealed that strong nuclear GR expression was maintained in the TRA98⁺ germ cells shortly after birth (Figure S4A). At PN1, we also observed a wider diversity of Nr3c1 exon 1 variants in the germ cells when compared to the soma (Figure S4C), consistent with observations in fetal oocytes. Staining at PN7, PN10 and PN14 revealed that GR expression was maintained in both PLZF⁺ and c-KIT⁺ spermatogonia, although at lower levels than at PN2 (Figure S4A, S4B). As anticipated based on prior reports ^{17–19,43,44}, GR expression appeared high in the surrounding peritubular myoid cells and Leydig cells of the interstitium at all timepoints observed. By PN21, GR became highly restricted to the undifferentiated (PLZF⁺) and differentiating (c-KIT⁺) spermatogonia, yet was absent from more mature spermatocytes or spermatids (Figure S4A, S4B). Furthermore, this spermatogonia-restricted expression pattern was maintained into adulthood (Figure 4D), suggesting a potential stage specific role for GR in the male germline.

As fetal oocytes appeared to be resistant to both loss and overactivation of GR signaling, we next asked whether the male germline was similar. We performed bulk RNA-seq on sorted Oct4-GFP⁺ germ cells and GFP⁻ somatic cells isolated from PN1 testes after administering dex at E17.5, E18.5 and PN0 (as outlined in Figure 5A and Materials & Methods section). In contrast to the female, differential expression analysis revealed a pronounced transcriptional response in male germ cells following dex treatment, within the same order of magnitude as the GFP⁻ soma (Figure 5A). GO term analysis on differentially expressed genes in the germ cells following dex treatment revealed a strong enrichment for genes related to the regulation of mRNA splicing (Figure S5A). The regulation of RNA splicing in germline cells is crucial for the proper progression of meiosis and spermatogenesis ^45–57, and is conserved across species ^58–60. Downregulated genes also showed an enrichment for mitotic cell cycle progression, which is of interest at this PN1 time point given that the mitotically-arrested male germ cells will begin to re-enter mitosis at approximately PN2 ^61–63. GO analysis of differentially expressed genes in treated somatic cells showed a wider variety of terms, which is likely due to the heterogeneity of GFP⁻ cells collected. Downregulated genes were broadly enriched for extracellular matrix organization and cellular adhesion, whereas upregulated genes showed enrichment for canonical glucocorticoid response genes.

Given the crucial role of transcript splicing in spermatogenesis, we first validated this potential link between GR and splicing. To our knowledge, no prior studies have implicated GR in the regulation of mRNA splicing in any cell type to date. To quantify any dose-dependent changes in splice factor gene expression, we performed qRT-PCR on bulk PN2 testis tissue from mice treated with three different doses of dex (0, 1, and 10 μg / g). Both Tra2b (a regulator of exon inclusion/skipping known to be expressed in the testis ⁶⁴) and Srsf7 (a member of the SR-rich family of pre-mRNA splicing factors) showed a significant and dose-dependent decrease in expression in response to dex, confirming our RNA-seq results (Figure 5B). To test whether this dex-dependent decrease in a subset of splicing factors resulted in any changes in transcript isoforms within the germ cells, we utilized rMATS ⁶⁵ to specifically quantify differential splicing events from our paired-end RNA-seq data. This analysis revealed 63 splicing events that were significantly altered in response to dex (Figure 5C), with the vast majority categorized as skipped exon events. Together, these results confirmed a dex-dependent decrease in splice factor expression that may lead to exon skipping events in the male germline.

Given this link between GR signaling and splicing and the known role of transcript splicing in regulating spermatogenesis ^45–57, we next sought to assess whether loss of GR would impact male fertility. Because full body deletion of GR results in lethality at birth due to defects in lung maturation ⁶⁶, we generated a conditional deletion of GR in the germline using a transgenic Blimp1-Cre line ⁶⁷. GR^flox/flox females were bred to GR^KO/+ males harboring the Blimp1-Cre transgene to obtain conditional knockout (cKO) pups with the genotype GR^KO/flox ; Blimp1-Cre⁺, (where GR^+/flox ; Blimp1-Cre^neg served as wildtype controls). IF staining at PN30 confirmed clear germ cell-specific loss of GR in the c-KIT⁺ prospermatogonia (Figure 6A), as well as the PLZF⁺ prospermatogonia (Figure S6A). To assess the fertility of GR cKO mice, six week old males were crossed to wildtype females, and the presence of vaginal plugs and litter size were monitored for eight weeks. All genotypes were able to mate and form vaginal plugs normally (data not shown). Control mice containing the Blimp1-Cre transgene, but functionally wild-type for GR (GR^flox/+) showed a significant decrease in the number of sired pups (Figure 6B), consistent with previously reported subfertility with this allele ⁶⁸. GR cKO males were still fertile and produced viable pups (Figure 6B). Although GR cKO males sired significantly less pups than WT controls, this was not significantly different from the effect produced by the presence of the Blimp1-Cre allele alone (Figure 6B). We therefore conclude that conditional deletion of GR in the germline does not further exacerbate the subfertility phenotype caused by the Blimp1-Cre allele, and ultimately yields viable progeny.

Our time course of GR expression in the fetal ovary revealed a peak in GR expression in E13.5 fetal oocytes (Figure 1C), a result highly consistent with a scRNA-seq study that found GR was expressed in the developing oocytes (at both the transcript and protein level) between E12.5 to E14.5 ²³. To our knowledge, there is no published evidence of GR protein in the adult rodent ovary to date. We were unable to detect any appreciable levels of GR by IF in the adult oocytes of any follicular stage (Figure S1C), which contrasts with prior GR staining in caprine ovaries that showed expression in primordial and antral oocytes ⁶⁹. It is unclear whether this discrepancy is due to genuine species-specific differences, or due to differences in sample preparation or antibodies used. In humans, while GR has been detected in fetal oocytes at gestational week 9.2 ²², the expression of GR in the adult ovary remains an open question.

Spatiotemporal localization of GR in the male germline here, however, is consistent with prior studies. In the somatic compartment, we observed strong and reproducible staining for GR in the peritubular myoid cells and interstitial Leydig cells of the fetal, postnatal and adult testis (Figures 4A, 4D, S4A, S4B), consistent with previous studies ^{17–19,43,44}. In the late postnatal and adult testis, we observed a highly specific pattern of GR expression in the germline, where it became restricted to undifferentiated (PLZF⁺) spermatogonia and differentiating (c-KIT⁺) spermatogonia (Figures 4D, S4A, S4B). This result is consistent with multiple rodent and human studies showing staining in the spermatogonia ^18–21.

Previous studies attribute the dynamic spatiotemporal regulation of GR across different cell types to the use of alternative promoters and exon 1 variants in the 5’ UTR of Nr3c1 ^36,70. In both the ovary and the testis, we observed that Nr3c1 transcripts produced by the germ cells use a wider combination of exon 1 variants (both previously annotated, and novel germ cell-specific variants) than their corresponding somatic cells (Figure 1D, S4C), which was surprising given the cellular heterogeneity of the GFP⁻ somatic cell fractions. This suggests that a larger diversity of upstream regulatory DNA sequences may be available for regulating GR expression more dynamically in the germline compared to the soma. We also observed that in both the testis and ovary, GR is expressed just prior to the onset of meiotic initiation in male and female germ cells. This suggests GR may have an upstream regulatory element that is either regulated by a meiotic transcription factor, or that is sensitive to changes in the chromatin landscape induced in germ cells upon meiotic initiation.

Given that GR is a potent transcription factor with a wide variety of functions across cell types ¹⁶, it was striking to find that both complete genetic deletion of GR and robust pharmacologic activation led to minimal changes in the transcriptional landscape of the female germline (Figures 2F, S2C, 3A). Furthermore, the robust response of the ovarian soma, yet not of the germ cells, to dex administration between E12.5 - E15.5 (Figure 3A) indicated that this was not a technical artifact, and instead suggested that the transcriptional response to changes in GR signaling was buffered in the cellular context of fetal oogonia. This is an interesting premise, as it may be advantageous for the germ cells to evolve protection against rampant GR-induced transcriptional changes in response to stress hormones experienced during gestation. Why the female germline has retained temporally restricted expression of GR given its apparent lack of a functional role, however, remains unclear. We have shown there is no evidence for inhibitory isoforms of GR, including both known truncated protein isoforms, as well as the GRβ transcriptional isoform (Figure 3C, 3D). We did, however, observe that GR is able to change its subcellular localization in response to ligand (Figure 3B, ii), suggesting that the ligand binding and nuclear translocation abilities of GR remain intact despite this attenuated function in female germ cells.

A possible mechanism for insulating fetal germ cells from response to GR signaling involves post-translational modifications (PTMs) to the receptor that could interfere with its ability to bind DNA, interact with other transcription factors, or recruit other cofactors required for its transcriptional activity. While a wide variety of PTMs to GR have been documented ⁴⁰, the limiting number of germ cells that can be obtained from the fetal ovary has made the detection of these PTMs via biochemical methods difficult. One particular PTM of interest is acetylation of GR by the circadian histone acetyltransferase protein, CLOCK. Acetylation of lysines within the hinge region of GR by CLOCK has been shown to result in normal ligand binding and nuclear translocation, but a complete loss of the ability of GR to both activate and repress transcription ⁷¹, very similar to what is seen in fetal oocytes. Interestingly, our single-cell RNA-seq data shows a strong enrichment for Clock expression in the germ cells in comparison to the soma (data not shown), further bolstering the idea that CLOCK could acetylate and inactivate GR specifically in the germ cells. Further biochemical analyses will be required to determine whether GR is being modified in this manner, or by any other PTMs.

In contrast to the lack of dex-induced transcriptional changes in the female germ cells, the male germline showed a considerable transcriptional response to dex treatment at PN1 (Figure 5A). While the somatic cells of the testis showed a more stereotyped glucocorticoid response with induction of canonical GR response genes such as Tsc22d3, Klf9, and Per1, germ cells did not show changes in these canonical genes. Instead, the downregulated transcripts showed a strong enrichment for genes involved in RNA processing and RNA splicing (Figure S5A). We validated that GR suppression of splice protein expression is dex dose-dependent (Figure 5B), and found that treatment with dex led to an increase in exon skipping events at specific loci across the genome (Figure 5C). To our knowledge, this is the first time GR has been implicated in regulating the expression of genes involved in RNA splicing. The ability of stress hormones to induce changes in transcript splicing through GR could pose an evolutionary advantage by increasing the overall transcript diversity of the germline in times of stress, allowing for increased adaptation of the germline to selective pressures from the environment.

As proper RNA splicing is crucial for meiotic progression in the spermatogenic lineage ^45–57, perturbations to RNA splicing as a result of elevated GR signaling could have important consequences for fertility. To our surprise, genetic deletion of GR specifically in germ cells of the testis did not appear to alter fertility in an appreciable manner (Figure 6B). While the subfertility caused by the presence of the Blimp1-Cre transgene alone could be masking a more subtle subfertility phenotype caused by loss of GR, it is also possible that homeostatic levels of GR signaling do not regulate splicing. While pharmacologic agonism of GR signaling in the adult testis could lead to more pronounced disruptions to splicing and subsequent fertility than deletion of GR, we were unfortunately unable to test this hypothesis. Both chronic stress and dex administration have been shown to inhibit testosterone production by Leydig cells ^72–81, and given that testosterone is essential for regulating spermatogenesis ⁸², it was not possible with current tools to tease apart the cell intrinsic effects of dex administration and GR activation specifically in spermatogonia. Future studies using transgenic overexpression of GR in the spermatogonia could address this effect of elevated GR signaling on fertility.

Together, our work demonstrates that the male and female germline have differential susceptibilities to GR signaling, which may have important implications for understanding sex-specific differences in the transgenerational inheritance of stress induced phenotypes. In this study, we observed that the male germline shows a robust transcriptional response to GR signaling in comparison to the attenuated response of the female, which may suggest that the male germline is inherently more vulnerable to GR-mediated stress signals. In fact, multiple studies have shown a more pronounced transmission of stress-induced and glucocorticoid-induced phenotypes through the paternal lineage ^{1,4,6,8,9,12–15,83,84}, leading to the hypothesis that epigenetic changes to the sperm are a likely causative factor. While it has been shown that some histone modifications such as H3K4me3 can be transmitted transgenerationally through sperm ^85,86, there is also a growing body of evidence demonstrating that regulation of non-coding RNA within the sperm is potentially a major mechanism through which the stress-induced phenotypes are transmitted ^3,29,87–90. Our finding that dex-mediated agonism of GR activity in male germ cells leads to an alteration in RNA splicing could represent a mechanism by which stress hormones alter the transcriptional milieu within the spermatogonia, impacting the overall RNA landscape within the germline. Future studies looking into the effect of GR signaling specifically on small non-coding RNAs will be imperative as we attempt to further elucidate the germ cell intrinsic impact of stress hormones on the germline and subsequent transgenerational epigenetic inheritance.

All animal work was performed under strict adherence to the guidelines and protocols set forth by the University of California San Francisco’s Institutional Animal Care and Use Committee (IACUC), and all experiments were performed in an animal facility approved by the Association for the Assessment and Accreditation of Laboratory Animal Care International (AAALAC). All mice were maintained in a temperature-controlled animal facility with 12 hour light dark cycles, and were given access to food and water ad libitum.

All matings were set in the evenings (15:00 or after), and the presence of a vaginal plug the morning after mating (08:00 – 11:00) was denoted embryonic day 0.5 (E0.5). Pregnant females were dissected at various timepoints, the uterine horns removed into ice-cold 0.4% BSA in PBS, and embryos dissected and staged based on canonical morphologic features. For all postnatal timepoints, postnatal day 0 (PN0) was assigned as the morning a litter was first seen (where litters were dropped the night of E18.5).

Adult male mice (age 6 weeks) were paired with WT, mixed background females (age 6 weeks), and females were monitored daily for the presence of a vaginal plug to indicate mating. For each pairing, number of litters produced and total number of pups per litter were recorded over an 8 week period.

Total genomic DNA was extracted from ear punches or tail tips by boiling in alkaline lysis buffer (25 mM NaOH ; 0.2 mM EDTA) for 45 minutes at 95°C, cooling to 4°C, and then neutralizing with an equal volume of 40 mM Tris-HCl. All genotyping reactions were performed with KAPA HotStart Mouse Genotyping Kit (Roche, 07961316001) with 1 μL of gDNA and a final primer concentration of 5 μM each. Cycler conditions are listed below each table. All PCR products were separated by electrophoresis on a 2% agarose gel in TBS stained with ethidium bromide (VWR, E3050), and genotypes determined based on sizes of DNA products.

For all in vivo dex dosing experiments, a water-soluble form of dex (Sigma, D2915) was utilized to allow for delivery in a saline vehicle (0.9% NaCl w/v in ddH₂O). The developmental time windows of dex administration were determined based on GR’s germline expression profile, and differed for the male and female germline. Dex dose and time widows varied between experiments, and are outlined for each experiment in the Results section. For all injections, the volume given to both saline and dex treated animals was fixed, where the concentration of dex in the injected solution was scaled to ensure precise dose delivery (normalized to body weight) in the fixed volume. For gestational timepoints (E12.5 - E18.5), pregnant dams were injected via IP injection at approximately 09:00 with 100 μL saline/dex. For postnatal timepoints (PN0 - PN21), pups were weighed and injected subcutaneously (in the back flank) with 20 μL of saline or dex. See experimental schematics for exact doses and timing.

Fetal ovaries were dissected at E14.5 and pooled into ice cold 0.4% BSA in PBS as described above. Ovaries were cultured in Millicell 24-well hanging inserts, 1.0 μm PET (Millipore, MCRP24H48) at 37°C, 5% CO₂. Ovaries were cultured in DMEM / F12 base medium (Gibco, 11330-032) supplemented with 10% charcoal-stripped FBS (Sigma, F6765), 1 mM sodium pyruvate (Gibco, 11360070), 0.5X MEM Nonessential Amino Acids (Corning, 25-025-CI), and 100 U/mL pen/strep (Gibco, 15140-122). Bottom chambers were filled with 1.3 mL of media, top inserts with 200 μL of media, and the media was changed daily. For experimental conditions, dex (Sigma, D1756) dissolved in DMSO was added to cultures at a final concentration of 1.0 μΜ.

Fetal gonads were dissected into ice cold 0.4% BSA in PBS, washed once, and maintained on ice until ready for digestion. 0.4% BSA solution was removed and replaced with 100 μL of 0.25% Trypsin-EDTA (Fisher Sci, 25200056) per ovary pair, or 150 μL per testis pair. Samples were incubated in a 37°C water bath for 30 minutes, with gentle pipetting every 10-15 minutes to facilitate the dissociation. After 30 minutes, DNase I (1 mg / mL) was added at a 1:10 dilution, and samples were incubated another 10 minutes at 37°C. Samples were pipetted to ensure complete digestion, and then an equal volume of ice-cold FBS (Gibco, 10437028) was added to inactivate trypsin. To prepare for FACS sorting, Sytox Blue viability dye (Invitrogen, S34857) was added to samples at 1:1000 dilution, and then samples were filtered through a 35 μm filter into FACS tubes (Falcon, 352235).

To isolate germ cells expressing the Pou5f1-ΔPE-eGFP transgene from the GFP⁻ somatic compartment, we utilized either a BD FACSAria II or III system. Briefly, cells were gated to remove doublets/multiplets via FSC height vs area comparison, and then dead cells were gated out based on uptake of the Sytox Blue viability dye. GFP⁺ germ cells and GFP⁻ somatic cells were sorted directly into 300 μL of QIAgen RLT Plus Buffer (with βME). Upon completion of the sort, samples were immediately vortexed for 30 sec and flash frozen.

All mouse gonad tissue was fixed with fresh 4% paraformaldehyde (PFA) in PBS at 4°C with rocking. Fixation times were as follows:

Following PFA fixation and PBS washes, embryonic gonads were incubated overnight in 30% sucrose (in PBS) at 4°C. The following day, the tissue was embedded in OCT (Tissue-Tek, 4583) and blocks stored at −80°C until ready for sectioning. For all postnatal and adult tissues, gonads were first incubated in 10% sucrose (in PBS) for about 1-2 hours at 4°C until the tissue sank to the bottom of the tube. Gonads were then transferred to incubate overnight in 30% sucrose (in PBS) at 4°C. The following day, the tissue was transferred to 50/50 OCT / 30% sucrose and allowed to equilibrate for 6 hours at 4°C prior to embedding in 100% OCT. Blocks were similarly stored at −80°C until ready for sectioning. All blocks were sectioned on a Leica 3050S Cryostat at a thickness of 5 – 10 μm depending on the tissue.

For all cryosectioned slides, slides were thawed to room temperature and then washed three times (5 min each) with 1X PBS to remove residual OCT. Slides were blocked for one hour at room temperature in 10% heat inactivated donkey serum + 0.1% Triton X-100 in PBS. All primary antibody incubations were performed at 4°C overnight in a humidified chamber, with primary antibodies diluted accordingly in blocking buffer. The next day, slides were washed three times with 1X PBS (5 min each), followed by a one hour incubation at RT with secondary antibodies diluted in blocking buffer. Samples were washed three times with 1X PBS (5 min each), mounted in VECTASHIELD Antifade Mounting Medium (Vector Laboratories, H-1000), and sealed with a coverslip.

After dissection, ovary pairs were transferred to 1 mL centrifuge tubes containing 100 μL of dissociation buffer (0.025% Trypsin, 2.5 mg/mL collagenase, and 0.1mg/mL DNase). Samples were incubated in a 37°C water bath for 30 minutes, and were pipetted vigorously every 10 minutes using a P-200 pipette to facilitate digestion. After 30 minutes, trypsin activity was stopped by adding an equal volume of FBS and mixing well. Next, and equal volume of freshly prepared hypotonic buffer (30 mM Tris pH 8.2, 50 mM sucrose, 17 mM sodium citrate, 5 mM EDTA, 0.5 mM DTT, 0.5 mM PMSF, at a final pH of 8.2) was added per sample, and samples were incubated for 30 minutes at room temperature. Samples were centrifuged for 10 min at 1000 rpm at room temperature, the supernatant removed, and then resuspended in 100 mM sucrose (+ 7.5 mM Boric acid, final pH 8.2). Samples were typically resuspended in 60 μL per ovary, which usually made three slides per ovary. Positively charged slides were pre-cleaned with 70% ethanol, and then dried gently with a Kimwipe. Using a hydrophobic pen, a small (∼2cm x 2cm) square was drawn on each slide, and the squares were each coated with approximately 20 μL of fixative solution (1% PFA, 0.15% Triton X-100, 3 mM DTT, 7.5 mM Boric acid, final pH 9.2). Using a P-20 pipette, 20 μL of the cell suspension was dropped onto the square from approximately one foot above. Slides were incubated for 1 hour in a covered humidified chamber at room temperature to allow fixation of nuclei to slide, and then incubated uncovered for approximately three hours to allow slides to fully dry. Once dry, slides were washed twice with 0.4% Photoflo in H₂O (Kodak, 146 4510) by fully submerging slides for two minutes each. Finally, slides were allowed to fully dry, then stored at −80°C until ready to stain.

Slides were warmed to room temperature, and then immediately incubated with 0.1% Triton X-100 in PBS for 10 minutes to permeabilize. Slides were washed three times with 1X PBS for five minutes each, and then blocked for 1 hour at room temp in 5% BSA in PBS. Slides were incubated overnight at 4°C with primary antibodies diluted in 5% BSA: rabbit α-SYCP1 (Abcam, ab15090; 1:200) and mouse α-γH2AX (Millipore Sigma, 05-636-I; 1:200). The following day, slides were washed three times, ten minutes each, with 1X PBS, and then incubated for 1 hour at room temp with secondaries in 5% BSA: donkey α-mouse 555 (1:200), donkey α-rabbit 647 (1:200), and DAPI (1:1000). Next, slides were washed three times, ten minutes each, with 1X PBS, and then incubated for 1 hour at room temp with pre-conjugated mouse α-SYCP3 (Abcam, ab205846; 1:50). Slides were washed again three times, and then mounted with VECTASHIELD Antifade Mounting Medium (Vector Laboratories, H-1000).

All imaging was performed on a white-light Leica TCS SP8 inverted confocal microscope using either an HC FLUOTAR L 25×/0.95 VISIR water objective (Leica) or an HC PL APO 63x/1.40 CS2 oil objective (Leica). All images were taken at 1024 x 1024 pixel resolution, and any tile scan images were merged using Leica software.

All analyses described below were performed using Imaris Microscopy Image Analysis Software v8.3.1 (Oxford Instruments).

Proteins were extracted from either whole-gonad tissue or FACS-sorted cells using Pierce RIPA Buffer (Thermo Fisher, 89900) as per the manufacturer’s protocol. For FACS-sorted, frozen cell pellets (100-300k cells), pellets were thawed on ice and resuspended in 100 μL of RIPA buffer. Suspension was incubated on ice for 15 minutes, with vortexing every three minutes. Suspension was spun at 16,000xg for 15 min at 4°C to pellet non-soluble cellular debris, and then supernatant transferred to a fresh tube. For whole-gonad tissue, dissected gonads were washed once with 1X PBS (without BSA), and then resuspended in 100-200 μL of RIPA buffer (depending on stage / number of gonads). Tissue was homogenized using a motorized pestle, on ice, until no visible tissue pieces remained. The suspension was then incubated on ice for 15 minutes (with vortexing every three minutes), and spun at 16,000xg for 15 min at 4°C as described above. Total amount of protein was quantified using a Pierce BCA Protein Assay Kit (Thermo Fisher, 23227) with BSA protein standards, following the manufacturer protocol exactly.

Prior to loading, protein samples were diluted with sample buffer (4X Laemmli Buffer (Bio-Rad, #1610747) supplemented with 10% β-mercaptoethanol) and water to reach a final sample buffer concentration of 1X. Samples were boiled in microcentrifuge tubes in a thermocycler for 5 minutes at 95°C to denature proteins. For all blots, samples were run on Tris/Glycine Mini-PROTEAN TGX Precast SDS-PAGE gels, 4-15% (Bio-Rad #4561085), with 1X Tris/Glycine/SDS Buffer (Bio-Rad #1610732). 5 μL of either WesternSure Pre-stained Chemiluminescent Protein Ladder (LI-COR #926-98000) or Chameleon Duo Pre-stained Protein Ladder (LI-COR #928-60000) were used depending on the downstream imaging system. All gels were electrophoresed at 50V for 10 minutes, and then 100V for 50-60 minutes. All gels were transferred to Immobilon-FL PVDF Membrane (Millipore #IPFL10100) (pre-activated with 100% methanol) using the Bio-Rad Trans-Blot SD semi-dry transfer cell (Bio-Rad # 1703940) at 15V for 60 min at 4°C. Transfer buffer was prepared as a 1X final solution by mixing 7:2:1 of ddH₂O : methanol : 10X Tris/Glycine Transfer Buffer (Bio-Rad #1610734). Following transfer, membranes were blocked for 60 min with 5% non-fat milk (Bio-Rad #1706404) prepared in 1X TBST (TBS + 0.1% Tween 20 ; Bioworld #42020084-1) on a rotator at RT. Primary antibodies were diluted in 5% non-fat milk and membranes incubated at 4°C overnight with gentle rotation. The following day, membranes were washed three times with 1X TBST for 5 min at RT with gentle rotation. Secondary antibodies, HRP-conjugated goat α-rabbit IgG (Abcam #ab205718) or HRP-conjugated goat α-mouse IgG (Abcam #ab205719) were diluted 1:5000 in 5% non-fat milk and incubated with membranes for 1 hr at RT with gentle rotation, and then washed three times with 1X TBST for 5 min at RT with gentle rotation. Membranes were imaged using the Pierce ECL Western Blotting Substrate (Thermo #32109) and developed on autoradiography film.

For bulk gonad tissue (both freshly isolated and ex vivo cultured), gonads were first homogenized in a 1.5 mL centrifuge tube in 100 μL TRIzol (Invitrogen 15596026) using a motorized pestle. After bringing the final volume to 1 mL with TRIzol, samples were vortexed on high for 30 seconds before storing at −80°C. To extract total RNA from TRIzol samples, 200 μL of chloroform was added per sample. Samples were vortexed vigorously for 30 seconds and then centrifuged at 12,000xg for 15 min at 4°C. The upper aqueous layer was carefully extracted, mixed 1:1 with an equal volume of fresh 70% ethanol, and loaded onto a QIAgen RNeasy Micro spin column for further cleanup and elution. The QIAgen RNeasy Micro Kit (cat. no. 74004) protocol was followed exactly, including the on-column DNase digestion for 15 min. For FACS-sorted cell samples that were sorted into QIAgen RLT Plus Buffer, samples were mixed with equal volume of 70% ethanol and loaded onto column without TRIzol homogenization and chloroform extraction. RNA concentrations were quantified using the Qubit RNA HS Assay Kit (ThermoFisher cat. no. Q32852) following the manufacturer’s protocol.

All primer sets used for qRT-PCR were designed from scratch. For each gene of interest, the mouse cDNA sequence of the predominant, full-length isoform was downloaded from the Ensembl database (https://www.ensembl.org). To identify potential primers with minimal off-target amplification across the mouse transcriptome, cDNA sequences were uploaded to NCBI’s Primer BLAST tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast), and primer searches were run with the following modified search parameters: PCR product size: 70-200 bp; Primer melting temperatures: Min: 60°C, Opt: 62°C, Max: 64°C, Max T_m diff: 1°C ; Database: Refseq mRNA ; Organism: Mus musculus ; Primer specificity stringency: “Ignore targets that have more than 4 or more mismatches to the primer” ; Primer GC content: 30-70% ; Max self complementarity: 3.00 ; Max pair complementarity: 3.00 ; Concentration of divalent cations: 3 ; Concentration of dNTPs: 0.2. Resulting putative primer pairs were screen manually for minimal self/pair complementarity scores, and then checked using the Beacon Designer tool (http://www.premierbiosoft.com/qOligo/Oligo.jsp?PID=1) for any potential primer-dimer events by selecting for primer pairs that had ΔG values closest to zero for Cross Dimer, Self Dimer and Hairpin scores. Lastly, the expected PCR product of any putative primer pairs was checked for potential secondary structure formation using IDT’s UNAFold program (https://www.idtdna.com/unafold). As no primer pairs were completely without potential secondary structure formation, primer pairs were deemed acceptable so long as none of the predicted secondary structures had an anticipated T_m value at or above the annealing temperature of our cycler reaction (60°C). Whenever possible, primer pairs that spanned an intron were selected to minimize any amplification from contaminating genomic DNA. All primers used were validated through a standard melt curve analysis following qRT-PCR reactions (95°C, 30 sec // 60°C, 1 min // 95°C then dec by 0.15°C every sec) to ensure the amplification of a single PCR product as anticipated, and any primer pairs that failed melt curve analysis were discarded and redesigned.

Prior to qRT-PCR, 100 – 1000 ng of RNA per sample was reverse transcribed using the qScript cDNA SuperMix (QuantaBio cat. no. 95048-100). qRT-PCR was performed using PowerUp SYBR Green Master Mix (Thermo A25776), as per manufacturer’s instructions. For each individual reaction, 1 ng total cDNA was used, with a final primer concentration of 0.5 μM each, and all reactions were performed in technical triplicate. All experiments were run on a ThermoFisher QuantStudio 5 with the following cycler conditions: (50°C, 2 min) – 1 cycle // (95°C, 2 min) – 1 cycle // (95°C, 30 sec ; 60°C, 1 min) – 40 cycles. For all experiments, qRT-PCR analysis was performed using standard ΔΔC_T method. Briefly, ΔC_T values were calculated by normalizing C_T values for individual target probes to the average C_T value of a standard housekeeping gene (typically 18S ribosomal RNA); ΔΔC_T values were calculated by normalizing sample ΔC_T values to the reference sample of choice (here usually “WT” or dex untreated); Fold change values were calculated by taking 2^-ΔΔCT.

Prior to RT-PCR, 100 – 1000 ng of RNA per sample was reverse transcribed using the qScript cDNA SuperMix (QuantaBio, 95048-100). All PCR reactions were performed using GoTaq Green Master Mix (Promega, M7123) with 0.25 ng total cDNA per reaction, with a final primer concentration of 0.5 μM each. Cycler conditions: { 95°C, 3 min – 1 cycle // (95°C, 15 sec ; (55-60)°C, 15 sec ; 72°C, 30 sec) – 35 cycles // (72°C, 10 min) – 1 cycle // 4°C hold }. For each primer pair, a temperature gradient PCR was used to determine the optimal annealing temperature. Products were electrophoresed on a 2% agarose gel at 150V for 30 minutes, and then visualized using a Bio-Rad GelDoc imager. Mouse XpressRef Universal Total RNA (QIAgen, 338114) was reverse transcribed and used at a concentration of 250 ng cDNA per reaction as a positive control for each primer set. Nuclease free water (Thermo Fisher, AM9937) was used as a negative control.