Abstract
While physiologic stress has long been known to impair mammalian reproductive capacity through hormonal dysregulation, mounting evidence now suggests that stress experienced prior to or during gestation may also negatively impact the health of future offspring. Rodent models of gestational physiologic stress can induce neurologic and behavioral phenotypes that persist for up to three generations, suggesting that stress signals can induce lasting epigenetic changes in the germline. Treatment with glucocorticoid stress hormones is sufficient to recapitulate the transgenerational phenotypes seen in physiologic stress models. These hormones are known to bind and activate the glucocorticoid receptor (GR), a ligand-inducible transcription factor, thus implicating GR-mediated signaling as a potential contributor to the transgenerational inheritance of stress-induced phenotypes. Here we demonstrate dynamic spatiotemporal regulation of GR expression in the mouse germline, showing expression in the fetal oocyte as well as the perinatal and adult spermatogonia. Functionally, we find that fetal oocytes are intrinsically buffered against changes in GR signaling, as neither genetic deletion of GR nor GR agonism with dexamethasone altered the transcriptional landscape or the progression of fetal oocytes through meiosis. In contrast, our studies revealed that the male germline is susceptible to glucocorticoid-mediated signaling, specifically by regulating RNA splicing within the spermatogonia, although this does not abrogate fertility. Together, our work suggests a sexually dimorphic function for GR in the germline, and represents an important step towards understanding the mechanisms by which stress can modulate the transmission of genetic information through the germline.
Introduction
The impact of stress on mammalian reproductive capacity has long been appreciated. The hypothesis that psychological stress not only limits the ability to conceive, but may also impact the health of offspring has gained considerable attention in recent years. A growing body of work has demonstrated that rodent models of stress can induce a variety of neurologic and behavioral phenotypes that may persist across generations, even in generations unexposed to the initial stressor 1–11. The persistence of such phenotypes for up to three generations suggests that stress signals can induce lasting epigenetic changes in the germline. As the committed precursors of eggs and sperm established during early embryogenesis, germ cells are responsible for transmitting genetic as well as epigenetic information across generations. Perturbations of the messenger can potentially corrupt the transmission of this information, thus understanding the mechanism by which stress can lead to epigenetic alterations in the germline remains a crucial unanswered question.
It has been demonstrated that treatment of rodents with synthetic glucocorticoid stress hormones is sufficient to recapitulate the transgenerational phenotypic inheritance seen in physiologic stress models 12–15. These hormones are known to bind and activate nuclear hormone receptors such as glucocorticoid receptor (GR; encoded by the Nr3c1 gene), thus implicating GR-mediated signaling in transgenerational inheritance of stress. The engagement of GR with its ligand in the cytoplasm promotes translocation to the nucleus where it then binds DNA response elements and recruits transcriptional cofactors to modulate target gene expression. As such, GR functions as a ligand-inducible transcription factor, allowing for robust changes to gene expression in response to stress hormones. While GR is known to be a potent transcriptional regulator in a wide variety of cell types 16, its function in the cells of the gonad remains less clear.
Prior evidence for GR expression in the germ cells is sparse. In the adult mouse testis, one study suggested GR expression was limited to primary spermatocytes 17, while others suggested that GR is expressed in both spermatogonia as well as spermatocytes 18,19. In humans, multiple studies have shown robust GR expression in differentiating spermatogonia, as well as low expression in primary spermatocytes 20,21. In the female, a single study of the human fetal ovary found evidence for GR expression by IHC in the oocytes of a nine week old embryo 22, although expression was heterogeneous across the tissue. A more recent study of mouse fetal oocyte development using single-cell RNA-sequencing revealed increasing expression of GR in the oocytes between E12.5 to E14.5 23. Together these data suggest that GR may be expressed in the rodent / human germline, but in limited developmental windows. However, no studies to date have directly assessed the cell-intrinsic role of GR in the germ cells. While several publications describe a variety of defects to the male and female germline in response to glucocorticoid treatment 12,22,24–34, it is unclear whether the phenotypes observed are due to the direct action of glucocorticoids on germ cells, or the indirect effects of systemic glucocorticoid treatment on hormone production by gonadal somatic cells and/or the hormone producing cells of the hypothalamus and pituitary.
Here, we characterized the expression of GR in the germline of the developing and adult mouse gonads, and determined the role of GR in normal germline formation and function. We discovered that GR is expressed in the female germline exclusively during fetal development, but absent from the adult oocyte. We demonstrated that the female germline is, surprisingly, resistant to changes in GR signaling. Negligible changes in transcription or meiotic progression of fetal oocytes resulted from either genetic deletion or agonism of GR with dexamethasone (dex), suggesting that the female germline is intrinsically buffered from changes in GR signaling. In contrast, we found that GR is expressed in pro-spermatogonia of the perinatal testis and spermatogonia of the adult testis, and transcriptomic analysis of germ cells from dex-treated males revealed a potential role of GR in regulating RNA splicing. Together our data confirm the dynamic spatiotemporal regulation of GR expression in both the male and female germ cells, and suggest a sexually dimorphic role for this receptor in the mammalian germline.
Results
Spatiotemporal expression of GR and novel isoforms in fetal oocytes
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.
Genetic deletion of GR leads to minimal changes in the female germline
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.
Developing ovarian somatic cells, but not germ cells, show GR transcriptional activation following exogenous glucocorticoid treatment
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).
GR in fetal oocytes relocalizes in response to ligand, but resistance to glucocorticoid signaling cannot be explained by the presence of inhibitory isoforms
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.
Spatiotemporal expression of GR in the male germline
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.
Germ cells of the perinatal testis show GR transcriptional regulation following exogenous glucocorticoid treatment
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.
Regulation of mRNA splicing in the early postnatal testis
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.
Conditional deletion of GR in the male germline does not impact fertility
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. GRflox/flox females were bred to GRKO/+ males harboring the Blimp1-Cre transgene to obtain conditional knockout (cKO) pups with the genotype GRKO/flox ; Blimp1-Cre+, (where GR+/flox ; Blimp1-Creneg 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 (GRflox/+) 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.
Discussion
The growing evidence that stress-induced phenotypes can persist across multiple generations has sparked considerable interest in understanding the mechanisms by which stress can act on the germline. Our efforts to characterize the role of GR-mediated signaling in male and female germ cells represent an important step in understanding the germ cell intrinsic effects that stress hormones can play on the integrity of the germline and subsequent transmission of genetic information across generations. In this study, we confirmed that germ cells of the testis and the ovary both show dynamic temporal regulation of GR expression over fetal and adult development, with distinct sex-specific differences in expression. We demonstrated that in the context of pharmacologic activation or genetic deletion of GR, fetal oocytes show minimal transcriptional changes and no significant changes in progression through meiotic prophase I, suggesting that the fetal oocytes are somehow resistant to changes in GR signaling. In contrast, we show that prospermatogonia downregulate genes important for RNA splicing in response to GR activation, suggesting a potential role for stress signaling in regulating transcript diversity in the male germline. Together, our work demonstrates a sexually dimorphic response to GR signaling within the germline, and suggests alternative mechanisms by which the male and female germline may be differentially affected by stress.
Glucocorticoid receptor immunofluorescence in the ovary and testis: discrepancies and similarities with the literature
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.
Discovery of novel germ cell-specific exon 1 isoforms in the 5’ UTR of Nr3c1
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.
The female germline is insulated from changes in GR-mediated signaling
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.
GR in the regulation of RNA splicing in the male germline
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.
Conditional deletion of GR using Blimp1-Cre does not affect male fertility
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.
Implications for understanding sex-specific differences in the transgenerational epigenetic inheritance of stress induced phenotypes
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.
Materials and Methods
Mouse husbandry
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.
Mouse timed pregnancies
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).
Mouse fertility test
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.
Genotyping
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.
Mouse lines used, genotyping primers, and genotyping reactions
Dex in vivo dosing
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.
Ex vivo gonad culture
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.
Embryonic gonad digestion
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).
Fluorescence Activated Cell Sorting (FACS)
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.
Tissue Fixation
All mouse gonad tissue was fixed with fresh 4% paraformaldehyde (PFA) in PBS at 4°C with rocking. Fixation times were as follows:
Tissue fixation parameters
Tissue Embedding and Sectioning
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.
Section Immunofluorescence
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.
Preparation of Meiotic Spreads from Embryonic Ovaries
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.
Staining of Meiotic Spreads – SYCP1, SYCP3 and γH2AX co-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).
Confocal Microscopy
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.
Quantitative Image Analysis using Imaris
All analyses described below were performed using Imaris Microscopy Image Analysis Software v8.3.1 (Oxford Instruments).
Cellular segmentation with WGA
In order to segment individual cells within a tissue section, we utilized fluorophore-conjugated forms of the lectin wheat germ agglutinin (WGA) (Biotium, 29022-1, 29062-1, 29025-1) to outline all cellular membranes. Slides were stained overnight with WGA (1:50) using the standard section immunofluorescence protocol outlined above. Confocal .lif files were imported into Imaris, and individual cells were computationally segmented using the “Cell” module with the following settings: { Detection Type: “Detect Cell Only” // Region: “Whole Image” // Cell Type: “Cell Membrane” // Cell Smallest Diameter: “7 μm” // Cell Membrane Detail: “0.75 μm” // Cell Filter Type: “Smooth” // Quality Threshold: “Manual” // Number of voxels: >20.0 }. The resulting segmented regions were then filtered to remove “empty” (non-cell-containing) regions by applying the following filters: { Size Filter: “25 – 400 μm2” // Cell Intensity Mean (DAPI channel): >20.0 }. Lastly, regions containing doublet cells (based on DAPI), were manually removed. To further subset out germ cells, an additional filter was applied based on the germ cell marker used: { Cell Intensity Mean (Germ cell marker channel): manually determined }.
Calculating normalized GR protein levels
After performing cellular segmentation with WGA as described above, the mean fluorescence intensity values of all channels were exported for each individual cell. Mean fluorescence intensity values for the GR channel were normalized to those of the DAPI channel on a per cell basis, and the resulting normalized values were plotted.
Protein Extraction and Quantification
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.
Western Blotting
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.
RNA isolation
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.
SYBR Green qRT-PCR Primer Design
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 Tm 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 Tm 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.
Quantitative RT-PCR
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 ΔΔCT method. Briefly, ΔCT values were calculated by normalizing CT values for individual target probes to the average CT value of a standard housekeeping gene (typically 18S ribosomal RNA); ΔΔCT values were calculated by normalizing sample ΔCT values to the reference sample of choice (here usually “WT” or dex untreated); Fold change values were calculated by taking 2-ΔΔCT.
Qualitative RT-PCR
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.
Single-cell RNA sequencing
Sample preparation
To generate embryos homozygous for the GR knockout allele (GRKO/KO), heterozygous females (GRKO/+) were crossed to heterozygous males also carrying an Oct4-GFP transgene (GRKO/+ ; Tg:Oct4GFP/GFP) to facilitate FACS sorting of germ cells. Pregnant dams were dissected the morning of E15.5, and tail clips were taken from each embryo to determine GR genotype. Fetal ovaries were dissected on ice, the mesonephroi removed, and the ovaries were digested for FACS sorting as described above. Fetal ovaries were pooled based on genotype (n=4 ovary pairs for WT GR+/+, and n=2 ovary pairs for GRKO/KO). In order to enrich for germ cells (relative to the predominant somatic cells), live germ cells were FACS sorted based on Tg:Oct4GFP expression, and somatic cells spiked back into the germ cell population at a ratio of 60 : 40, germ : soma. The final cell suspension was resuspended in 0.04% BSA at a concentration of 1000 cells / μL, and was used for 10X single-cell capture as described below.
10X Capture, Library Preparation, and Sequencing
The following sample preparation and sequencing was all performed by the UCSF CoLabs Initiative. All samples were processed with the standard Chromium 10X Single-Cell 3’ Reagent Kit v3 workflow. In summary, final cell suspensions were loaded onto the 10X Chromium microfluidics chip along with reaction master mix, partitioning oil, and Single-Cell 3’ v3 Gel Beads in order to generate Gel Beads-in-emulsion (GEMs) containing individual cells. RT reactions were performed within individual GEMs to generate cell-barcoded full-length cDNA, followed by GEM breakdown, cDNA pooling, cDNA PCR amplification, cleanup, and QC. Amplified cDNA was enzymatically fragmented, end repaired, A-tailed, size selected, and ligated with adapters. Following a sample indexing PCR reaction and final size selection, sample libraries were pooled, QC’ed, and sequenced on an Illumina NovoSeq 6000 platform. Resulting paired-end sequencing reads were processed using the 10X Cell Ranger software to generate feature-barcode matrices.
Data Analysis
All single-cell RNA-seq data processing was performed in house using Seurat91 v3.2.3. Count matrices for both WT and KO conditions were read into Seurat to create Seurat Objects, and were then filtered as follows: WT: { nFeature_RNA: 1500 – 6500 // nCount_RNA: 5500 – 40000 // % mitochondria: < 10% }; KO: { nFeature_RNA: 1200 – 5500 // nCount_RNA: 3000 – 30000 // % mitochondria: < 10% }. Both WT and KO Seurat objects were combined using the “merge” function, and the resulting objects were processed for dimensional reduction using the standard Seurat workflow using the following functions: NormalizeData, FindVariableFeatures, ScaleData, RunPCA, RunUMAP, FindNeighbors, and FindClusters. During ScaleData, the following variables were regressed out: nFeature_RNA, nCount_RNA, percent.mito, and percent.ribo. The cell type identities of each cluster were determined by comparing top cluster markers to a previously annotated list E16.5 ovarian cell type specific markers92. Pairwise differential gene expression analyses were performed between WT and KO cells independently for each of the main cell type clusters (germ, theca, granulosa, and epithelia) using the FindMarkers function using the default Wilcoxon Rank Sum test. Genes with an adjusted p-value less than 0.05 and a log fold change greater than 0.25 were considered significant.
3’ Tag-Seq
Sample preparation
To generate embryos homozygous for the GR knockout allele (GRKO/KO), heterozygous females (GRKO/+) were crossed to heterozygous males also carrying an Oct4-GFP transgene (GRKO/+ ; Tg:Oct4GFP/GFP) to facilitate FACS sorting of germ cells. Pregnant dams were dissected the morning of E17.5, and tail clips were taken from each embryo to determine GR genotype. Fetal ovaries were dissected on ice, the mesonephroi removed, and the ovaries were digested for FACS sorting as described above. Tg:Oct4-GFP+ germ cells from individual embryo ovary pairs were FACS sorted directly in QIAgen RLT+ buffer and stored at -80 until ready for RNA extractions as outlined above. To generate embryos with a conditional deletion of GR (GRcKO/cKO), females homozygous for an exon 3 floxed GR allele (GRflox/flox) were crossed to similar males that were also heterozygous for the germ cell specific Oct4CreERT2 allele (GRflox/flox ; Oct4CreERT2/+ ; Tg:Oct4GFP/GFP). Pregnant dams were injected at E10.5 with 125 μg / g tamoxifen to induce recombination and deletion of GR exon 3 specifically in the germ cells. Pregnant dams were dissected the morning of E17.5 and processed for FACS sorting as described for GRKO/KO females above.
Library Preparation and Sequencing
All library preparations and sequencing were performed by the University of California, Davis DNA Technologies & Expression Analysis Core. Gene expression profiling was carried out using a 3’ Tag-RNA-Seq protocol. Barcoded sequencing libraries were prepared using the QuantSeq 3’ mRNA-Seq Library Prep FWD kit (Lexogen) for multiplexed sequencing according to the manufacturer recommendations. The library fragment size distribution was determined using microcapillary gel electrophoresis on a Bioanalyzer 2100 (Agilent), and libraries were quantified using a Qubit fluorometer (LifeTechnologies). Final libraries were pooled in equimolar ratios and sequenced on an Illumina HiSeq 4000.
Data Analysis
Data analysis of 3’ Tag-Seq data was performed in house. Illumina universal adapters (AGATCGGAAGAG) were trimmed from fastq files using cutadapt93. Paired-end reads were aligned to the mm10 genome using STAR94 v2.6.0, and counts files were generated using featureCounts95 v1.6.3. Differential expression analysis was performed using edgeR96 v3.28.1 and limma97 v3.42.2. The resulting p-values were adjusted using the Benjamini and Hochberg’s approach for controlling the false discovery rate. Genes with an adjusted p-value ≤ 0.05 were assigned as differentially expressed.
Bulk RNA-seq on in vivo dex-dosed gonads
Sample preparation
For analysis of female germ cells by RNA-seq, pregnant dams were injected by IP at approximately 09:00 with either saline or 10 mg dex / kg weight daily at E12.5, E13.5, E14.5 and E15.5. Embryos were dissected at E15.5 at approximately 15:00, and fetal ovaries placed in ice-cold 0.4% BSA in PBS. Mesonephroi were carefully microdissected away from the ovary, and ovaries were then digested for FACS sorting as described above. For analysis of male germ cells by RNA-seq, pregnant dams were injected via IP at approximately 09:00 with either saline of 10 mg dex / kg weight daily at E17.5 and E18.5. Following delivery the evening of E18.5, PN0 pups were injected the following morning subcutaneously with either saline or 10 mg dex / kg bodyweight. Pups were then sacrificed at approximately 11:00 on PN1, and testes placed in ice-cold 0.4% BSA in PBS. The tunica vaginalis was carefully microdissected away from each testis, and testes were then digested for FACS-sorting as described above.
Library Preparation and Sequencing
All library preparations and sequencing were performed by the company Novogene. mRNA was purified from total RNA using poly-dT magnetic beads. Following mRNA fragmentation, first strand cDNA synthesis was carried out using random hexamer primers, followed by second strand cDNA synthesis. Fragment ends were repaired, A-tailed, and ligated with sequencing adapters, followed by size selection, PCR amplification, and purification. Final libraries were quantified using Qubit and run on an Agilent Bioanalyzer system to ensure proper size distribution. Quantified libraries were pooled and sequenced on an Illumina platform. Clustering of the index-coded samples was performed according to the manufacturer’s instructions, and library preparations were subsequently sequenced to generate paired-end reads.
Data Analysis
Initial data QC, genome alignment, and differential expression analysis was all carried out by Novogene.
Data QC
Raw fastq reads were processed using custom perl scripts to remove reads containing adapter sequences, reads containing poly-N, and low quality reads (based on Q20, Q30 and %GC scores).
Alignment
The mm10 reference genome was indexed using Hisat298 v2.0.5. Cleaned, paired-end reads were aligned to the mm10 genome using the splice-aware aligner Hisat2 v2.0.5.
Expression Quantification
Counts of reads mapping to each gene were determined using featureCounts95 v1.5.0-p3. FPKM values were calculated for each gene based on the length of the gene, sequencing depth, and read counts mapping to the gene.
Differential Expression Analysis
Differential expression analysis of pairwise conditions was performed using DESeq299 (v1.20.0). The resulting p-values were adjusted using the Benjamini and Hochberg’s approach for controlling the false discovery rate. Genes with an adjusted p-value ≤ 0.05 were assigned as differentially expressed.
Gene Ontology Enrichment Analysis
GO Term enrichment analysis was performed using the GO Consortium’s online tool at: http://geneontology.org/. A list of significant genes (adjusted p-value ≤ 0.05) was uploaded and used to look for enriched biological processes, utilizing a Fisher’s Exact test followed by Bonferroni correction for multiple testing. Lists for upregulated and downregulated genes were run independently. The results were sorted based on highest fold enrichment scores, and the top ten processes for each category plotted.
Differential splicing analysis using rMATS
Differential transcript splicing analysis was performed using rMATS65 version 4.1.2, with genome-aligned BAM files generated from paired end RNA-seq reads used as input. Comparisons were made between saline treated and dex treated germ cells from PN1 testes (Figure 2.12A), using the rMATS settings: -t paired ; --readLength 150. Results were visualized as sashimi plots, generated with the program rmats2sashimiplot.
Antibodies Used
RT-PCR Primers Used
qRT-PCR Primers Used
Acknowledgements
The authors would like to thank thesis committee members Barbara Panning, Kaveh Ashrafi and Brian Feldmen for critical feedback over the course of this project. Thank you to Ryan Samuel for consulting on scRNA-seq analysis, and thank you to Bikem Soygur for exceptional scientific guidance and technical assistance on this project. We also thank the UC Davis DNA Technologies & Expression analysis core for 3’ Tag-Seq library prep and sequencing, the UCSF CoLabs Initiative for performing Illumina 10X capture, library prep, and sequencing, the UCSF Parnassus Flow Cytometry Core for assistance with cell sorting, as well as A. Rajkovic for generously providing Nobox antibody.
Funding
S.A.C was supported by the Ruth L. Kirschstein National Research Service Award Individual Predoctoral F31 fellowship NIH 5F31HD101234, N.R. is supported by the NIH Endocrinology T32-DK007418 Fellowship, M.H.F. is supported by NIH 1F31HD110208-01 and the Hillblom/BARI Graduate Student Fellowship Award, and D.J.L. is supported by P30-ES030284, R01ES028212, and R01GM122902, the W.M. Keck Foundation, and the UCSF Program for Breakthrough Biomedical Research.
Competing interests
The authors have no competing interests to declare.
Data availability
All data required to evaluate the conclusions put forth have been included in the paper and/or supplementary materials. Genomics data has been deposited in the Gene Expression Omnibus (GEO) under the accession code GSE234681.
References
- 1.Epigenetic transmission of the impact of early stress across generationsBiol Psychiat 68
- 2.Transgenerational impairment of hippocampal Akt-mTOR signaling and behavioral deficits in the offspring of mice that experience postpartum depression-like illnessProg Neuro-psychopharmacology Biological Psychiatry 73:11–18
- 3.Implication of sperm RNAs in transgenerational inheritance of the effects of early trauma in miceNat Neurosci 17:667–669
- 4.Influence of early stress on social abilities and serotonergic functions across generations in micePlos One 6
- 5.Inheritable Effect of Unpredictable Maternal Separation on Behavioral Responses in MiceFront Behav Neurosci 5
- 6.Chronic social instability induces anxiety and defective social interactions across generationsBiol Psychiat 73
- 7.Transgenerational effects of social stress on social behavior, corticosterone, oxytocin, and prolactin in ratsHorm Behav 65:386–393
- 8.Transgenerational inheritance of chronic adolescent stress: Effects of stress response and the amygdala transcriptomeGenes Brain Behav 18
- 9.Pathological brain plasticity and cognition in the offspring of males subjected to postnatal traumatic stressMol Psychiatr 20:621–631
- 10.Brain metabolic alterations in mice subjected to postnatal traumatic stress and in their offspringJ Cereb Blood Flow Metabolism 37:2423–2432
- 11.Transgenerational disruption of functional 5-HT1AR-induced connectivity in the adult mouse brain by traumatic stress in early lifeMol Psychiatr 22
- 12Elevated paternal glucocorticoid exposure alters the small noncoding RNA profile in sperm and modifies anxiety and depressive phenotypes in the offspringTransl Psychiat 6
- 13.Prenatal Glucocorticoid Exposure Modifies Endocrine Function and Behaviour for 3 Generations Following Maternal and Paternal TransmissionSci Rep-uk 7
- 14.Antenatal Glucocorticoid Exposure Results in Sex-Specific and Transgenerational Changes in Prefrontal Cortex Gene Transcription that Relate to Behavioural OutcomesScientific Reports 9
- 15.Prenatal Glucocorticoid Exposure Results in Changes in Gene Transcription and DNA Methylation in the Female Juvenile Guinea Pig Hippocampus Across Three GenerationsSci Rep-uk 9
- 16.Glucocorticoid Signaling in Health and Disease: Insights from Tissue-Specific GR Knockout MiceEndocrinology 159
- 17.Localization of the glucocorticoid receptor in testis and accessory sexual organs of male ratMol Cell Endocrinol 95:115–120
- 18.Glucocorticoid receptor distribution in rat testis during postnatal development and effects of dexamethasone on immature peritubular cells in vitroAndrologia 32:23–30
- 19.Corticosterone administration to rat pups, but not maternal separation, affects sexual maturation and glucocorticoid receptor immunoreactivity in the testisPharmacol Biochem Be 73:95–103
- 20.Possible involvement of the glucocorticoid receptor (NR3C1) and selected NR3C1 gene variants in regulation of human testicular functionAndrology-us 5:1105–1114
- 21.The Glucocorticoid Receptor NR3C1 in Testicular Peritubular Cells is Developmentally Regulated and Linked to the Smooth Muscle-Like Cellular PhenotypeJ Clin Medicine 9
- 22.Dexamethasone Induces Germ Cell Apoptosis in the Human Fetal OvaryJ Clin Endocrinol Metabolism 97:E1890–E1897
- 23.Dissecting the initiation of female meiosis in the mouse at single-cell resolutionCell Mol Life Sci 78:695–713
- 24.Toxic effects of dexamethasone on mouse testicular germ cellsAndrologia 42:247–253
- 25.Dysfunction in Sertoli cells participates in glucocorticoid-induced impairment of spermatogenesisMol Reprod Dev 88:405–415
- 26.Effect of Prednisolone upon Normal and Pathologic Human SpermatogenesisFertil Steril 17:500–513
- 27.Apoptosis of testicular germ cells induced by exogenous glucocorticoid in ratsHum Reprod 15:1917–1920
- 28.Dexamethasone effects on Bax expression in the mouse testicular germ cellsFolia Histochem Cyto 47:237–41
- 29.Single paternal dexamethasone challenge programs offspring metabolism and reveals multiple candidates in RNA-mediated inheritanceIscience 24
- 30.The effect of glucocorticoids on ERK-1/2 phosphorylation during maturation of lamb oocytes and their subsequent fertilization and cleavage ability in vitroReprod Toxicol 29:198–205
- 31.Dexamethasone or Triamcinolone increases Follicular Development in Immature Female RatsJpn J Pharmacol 84:281–286
- 32.Corticosterone Injection Impairs Follicular Development, Ovulation and Steroidogenesis Capacity in Mice OvaryAnimals Open Access J Mdpi 9
- 33.Glucocorticoids impair oocyte competence and trigger apoptosis of ovarian cells via activating the TNF-α systemReproduction 160:129–140
- 34.Glucocorticoids impair oocyte developmental potential by triggering apoptosis of ovarian cells via activating the Fas systemSci Rep-uk 6
- 35.Combined iDISCO and CUBIC tissue clearing and lightsheet microscopy for in toto analysis of the adult mouse ovary†Biol Reprod 102:1080–1089
- 36.Tissue specific glucocorticoid receptor expression, a role for alternative first exon usage?Biochem Pharmacol 72:1529–1537
- 37.Knockout of the Glucocorticoid Receptor Impairs Reproduction in Female ZebrafishInt J Mol Sci 21
- 38.Loss of the glucocorticoid receptor causes accelerated ovarian ageing in zebrafishProc Royal Soc B 287
- 39.nr3c1 null mutant zebrafish are viable and reveal DNA-binding-independent activities of the glucocorticoid receptorSci Rep-uk 7
- 40.Glucocorticoid receptor control of transcription: precision and plasticity via allosteryNat Rev Mol Cell Bio 18
- 41.Translational Regulatory Mechanisms Generate N-Terminal Glucocorticoid Receptor Isoforms with Unique Transcriptional Target GenesMol Cell 18:331–342
- 42.Discovery of glucocorticoid receptor-beta in mice with a role in metabolismMol Endocrinol 24
- 43.Covalent affinity labeling, radioautography, and immunocytochemistry localize the glucocorticoid receptor in rat testicular leydig cellsAm J Anat 186:369–377
- 44.In vivo actions of the Sertoli cell glucocorticoid receptorEndocrinology 155
- 45.Male sterility and enhanced radiation sensitivity in TLS−/− miceEmbo J 19:453–462
- 46.Ewing sarcoma gene EWS is essential for meiosis and B lymphocyte developmentJ Clin Invest 117:1314–1323
- 47.RBM5 Is a Male Germ Cell Splicing Factor and Is Required for Spermatid Differentiation and Male FertilityPlos Genet 9
- 48.The splicing landscape is globally reprogrammed during male meiosisNucleic Acids Res 41:10170–10184
- 49.RNA Binding Protein Ptbp2 Is Essential for Male Germ Cell DevelopmentMol Cell Biol 35:4030–4042
- 50.An Orchestrated Intron Retention Program in Meiosis Controls Timely Usage of Transcripts during Germ Cell DifferentiationDev Cell 41:82–93
- 51.BCAS2 is involved in alternative mRNA splicing in spermatogonia and the transition to meiosisNat Commun 8
- 52.Mettl3-mediated m6A regulates spermatogonial differentiation and meiosis initiationCell Res 27:1100–1114
- 53.Impaired Spermatogenesis, Muscle, and Erythrocyte Function in U12 Intron Splicing-Defective Zrsr1 Mutant MiceCell Reports 23:143–155
- 54.An ancient germ cell-specific RNA-binding protein protects the germline from cryptic splice site poisoningElife 8
- 55.DDX5 plays essential transcriptional and post-transcriptional roles in the maintenance and function of spermatogoniaNat Commun 10
- 56.Yuan, S. et al. hnRNPH1 recruits PTBP2 and SRSF3 to cooperatively modulate alternative pre-mRNA splicing in germ cells and is essential for spermatogenesis and oogenesis. (2021) doi:10.21203/rs.3.rs-1060705/v1.hnRNPH1 recruits PTBP2 and SRSF3 to cooperatively modulate alternative pre-mRNA splicing in germ cells and is essential for spermatogenesis and oogenesis https://doi.org/10.21203/rs.3.rs-1060705/v1
- 57.SYMPK Is Required for Meiosis and Involved in Alternative Splicing in Male Germ CellsFrontiers Cell Dev Biology 9
- 58.Autoregulation of the splicing of transcripts from the transformer-2 gene of DrosophilaGene Dev 5:786–796
- 59.Major spliceosome defects cause male infertility and are associated with nonobstructive azoospermia in humansProc National Acad Sci 113:4134–4139
- 60.Precursor RNA processing 3 is required for male fertility, and germline stem cell self-renewal and differentiation via regulating spliceosome function in Drosophila testesSci Rep-uk 9
- 61.Proliferative activity of gonocytes, Sertoli cells and interstitial cells during testicular development in miceReproduction 93:233–243
- 62.Spermatogonial morphology and kinetics during testis development in mice: a high-resolution light microscopy approachReproduction 142:145–155
- 63.Reproliferation and relocation of mouse male germ cells (gonocytes) during prespermatogenesisAnat Rec 258:210–220
- 64.Identification of Evolutionarily Conserved Exons as Regulated Targets for the Splicing Activator Tra2β in DevelopmentPlos Genet 7
- 65.MATS: a Bayesian framework for flexible detection of differential alternative splicing from RNA-Seq dataNucleic Acids Res 40:e61–e61
- 66.Targeted disruption of the glucocorticoid receptor gene blocks adrenergic chromaffin cell development and severely retards lung maturationGene Dev 9
- 67.Blimp1 is a critical determinant of the germ cell lineage in miceNature 436
- 68.PRDM1/BLIMP1 is widely distributed to the nascent fetal–placental interface in the mouse gastrulaDev. Dyn 246:50–71
- 69.Immunolocalization for glucocorticoid receptor and effect of cortisol on in vitro development of preantral folliclesVet Animal Sci 7
- 70.5′-Heterogeneity of Glucocorticoid Receptor Messenger RNA Is Tissue Specific: Differential Regulation of Variant Transcripts by Early-Life EventsMol Endocrinol 14:506–517
- 71.Circadian rhythm transcription factor CLOCK regulates the transcriptional activity of the glucocorticoid receptor by acetylating its hinge region lysine cluster: potential physiological implicationsFaseb J 23:1572–1583
- 72.Effects of restraint stress on plasma LH and testosterone concentrations, Leydig cell LH/HCG receptors, and in vitro testicular steroidogenesis in adult ratsHorm Behav 24:324–341
- 73.Role of glucocorticoids in the stress-induced suppression of testicular steroidogenesis in adult male ratsHorm Behav 26:350–363
- 74.Effects of acute and chronic immobilization stress on rat Leydig cell steroidogenesisJ Steroid Biochem Mol Biology 58:351–355
- 75.Effects of in Vivo Administration of Dexamethasone, Corticotropin and Human Chorionic Gonadotropin on Steroidogenesis and Protein and DNA Synthesis of Testicular Interstitial Cells in Prepuberal RatsEndocrinology 101:1256–1263
- 76.Chronic Corticosterone Treatment Impairs Leydig Cell 11β-Hydroxysteroid Dehydrogenase Activity and LH-Stimulated Testosteorne ProductionHorm Metab Res 32:142–146
- 77.Effects of excess corticosterone on NADPH generating enzymes and glucose oxidation in Leydig cells of adult ratsHum Exp Toxicol 25:119–125
- 78.Glucocorticoid-mediated repression of P450scc mRNA and de novo synthesis in cultured Leydig cellsEndocrinology 124:2099–104
- 79.Direct inhibitory effect of glucocorticoids upon testicular luteinizing hormone receptor and steroidogenesis in vivo and in vitroEndocrinology 108:2142–8
- 80.Corticosterone has direct inhibitory effect on the expression of peptide hormone receptors, 11 beta-HSD and glucose oxidation in cultured adult rat Leydig cellsMol Cell Endocrinol 279:52–62
- 81.Corticosterone induces steroidogenic lesion in cultured adult rat leydig cells by reducing the expression of star protein and steroidogenic enzymesJ Cell Biochem 103:1472–1487
- 82.The regulation of spermatogenesis by androgensSemin Cell Dev Biol 30:2–13
- 83.Early prenatal stress epigenetically programs dysmasculinization in second-generation offspring via the paternal lineageJ Neurosci 31
- 84.Paternal transmission of stress-induced pathologiesBiol Psychiat 70
- 85.Sperm histone H3 lysine 4 trimethylation is altered in a genetic mouse model of transgenerational epigenetic inheritanceNucleic Acids Res 48:11380–11393
- 86.Disruption of histone methylation in developing sperm impairs offspring health transgenerationallyScience 350
- 87.Alterations in sperm long RNA contribute to the epigenetic inheritance of the effects of postnatal traumaMol Psychiatr 25:2162–2174
- 88.Paternal Stress Exposure Alters Sperm MicroRNA Content and Reprograms Offspring HPA Stress Axis RegulationJ Neurosci 33:9003–9012
- 89.Transgenerational epigenetic programming via sperm microRNA recapitulates effects of paternal stressProc National Acad Sci 112:13699–13704
- 90.Reproductive tract extracellular vesicles are sufficient to transmit intergenerational stress and program neurodevelopmentNat Commun 11
- 91.Spatial reconstruction of single-cell gene expression dataNat Biotechnol 33:495–502
- 92.Two distinct pathways of pregranulosa cell differentiation support follicle formation in the mouse ovaryProc National Acad Sci 117:20015–20026
- 93.Cutadapt removes adapter sequences from high-throughput sequencing readsEmbnet J 17:10–12
- 94.STAR: ultrafast universal RNA-seq alignerBioinformatics 29:15–21
- 95.featureCounts: an efficient general purpose program for assigning sequence reads to genomic featuresBioinformatics 30:923–930
- 96.edgeR: a Bioconductor package for differential expression analysis of digital gene expression dataBioinformatics 26:139–140
- 97.. limma powers differential expression analyses for RNA-sequencing and microarray studiesNucleic Acids Res 43:e47–e47
- 98.Transcript-level expression analysis of RNA-seq experiments with HISAT, StringTie and BallgownNat Protoc 11:1650–1667
- 99.Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2Genome Biol 15
Article and author information
Author information
Version history
- Sent for peer review:
- Preprint posted:
- Reviewed Preprint version 1:
- Reviewed Preprint version 2:
- Version of Record published:
Copyright
© 2023, Cincotta et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
- views
- 858
- downloads
- 130
- citation
- 1
Views, downloads and citations are aggregated across all versions of this paper published by eLife.