Sex reversal following deletion of a single distal enhancer of Sox9

See allHide authors and affiliations

Science  29 Jun 2018:
Vol. 360, Issue 6396, pp. 1469-1473
DOI: 10.1126/science.aas9408

Sox9 regulation during sex determination

Sex determination is regulated by the Sox9 gene. During testis differentiation, this gene is directly targeted by the product of the Y chromosome–encoded gene Sry. The regulatory region of Sox9 is complex, which is typical of genes with multiple roles in development. Gonen et al. find that a single far-upstream 557–base pair element is critical for up-regulating Sox9. Without it, XY mice develop as females instead of males. The 557–base pair enhancer is conserved, likely to be relevant to human disorders of sex differentiation, and probably essential because it acts early in a time-critical process, and any failure allows ovary-specific factors to dominate.

Science, this issue p. 1469


Cell fate decisions require appropriate regulation of key genes. Sox9, a direct target of SRY, is pivotal in mammalian sex determination. In vivo high-throughput chromatin accessibility techniques, transgenic assays, and genome editing revealed several novel gonadal regulatory elements in the 2-megabase gene desert upstream of Sox9. Although others are redundant, enhancer 13 (Enh13), a 557–base pair element located 565 kilobases 5′ from the transcriptional start site, is essential to initiate mouse testis development; its deletion results in XY females with Sox9 transcript levels equivalent to those in XX gonads. Our data are consistent with the time-sensitive activity of SRY and indicate a strict order of enhancer usage. Enh13 is conserved and embedded within a 32.5-kilobase region whose deletion in humans is associated with XY sex reversal, suggesting that it is also critical in humans.

The regulation of genes with important roles in embryonic development can be complex, involving multiple, often redundant enhancers, silencers, and insulators (1, 2). The genes may have a poised epigenetic state prior to their expression, and their activation or repression may involve positive or negative feed-forward loops. This complexity is likely to be amplified when the gene has functions in more than one tissue, given that the regulatory elements required for each are often interspersed and necessitate dynamic alterations in chromatin conformation (1, 2). The developing gonads constitute an interesting system in which to explore questions of gene regulation during development (3). Most of the cell lineages are bipotential, with the ability to give rise to cell types typical of either ovaries or testes, and many genes that become associated with male or female fate begin by being expressed at equivalent, although usually low, levels in supporting cell precursors of both XX and XY gonads (46).

In mammals, the Sry gene encodes a protein that is transiently expressed and initiates testis and subsequent male development by triggering cells of the supporting cell lineage to differentiate into Sertoli cells rather than granulosa cells typical of ovaries (7). Sox9, the main target of SRY, is critical for the differentiation of Sertoli cells and then functions along with other transcription factors, notably Sox8 and then Dmrt1, for Sertoli cell maintenance (46). Both gain- and loss-of-function studies in mice and humans demonstrate that Sox9 plays a key role in testis determination (813). Notably, humans heterozygous for null mutations develop campomelic dysplasia (CD) [Online Mendelian Inheritance in Man (OMIM) entry 114290] (11), a severe syndrome where 70% of XY patients show female development (12, 13).

Sox9 functions in many embryonic and adult cell types (14), and genetic and molecular evidence suggests that its regulatory region is spread over a gene desert of at least 2 Mb 5′ to the coding sequence (15). The only enhancer known to be relevant for expression in Sertoli cells was TES, a 3.2-kb element mapping 13 kb 5′ from the transcriptional start site, and its 1.4-kb core, TESCO (16). Targeted deletion of TES or TESCO reduced Sox9 expression levels in the early and postnatal mouse testis to about 45% of normal but did not result in XY female development (17). We therefore used several unbiased approaches to systematically screen for additional gonad enhancers upstream of mouse Sox9. We used deoxyribonuclease I hypersensitive site sequencing (DNaseI-seq) data obtained with embryonic day 13.5 (E13.5) and E15.5 sorted Sertoli cells (18). From 33 putative enhancers, we chose only those positive at both stages (14 enhancers) for in vivo validation by transgenic assays (Fig. 1A and fig. S1). In parallel, we carried out ATAC-seq (assay for transposase-accessible chromatin using sequencing) on XY and XX gonads, which permitted the use of fewer sorted cells at E10.5, an early bipotential stage, and E13.5, when gonadal sex is already determined (figs. S1 and S2 and methods). Most putative enhancers discovered by DNaseI-seq were evident in the E13.5 XY ATAC-seq data; however, we used this assay to include two more putative enhancers in the in vivo screen: enhancer 1 (Enh1) and Enh14 (Fig. 1A and fig. S1). Chromatin immunoprecipitation sequencing (ChIP-seq) was also performed for H3K27ac, a histone modification that marks active enhancers (fig. S1).

Fig. 1 Enh13 is a testis-positive enhancer of Sox9 located within the XY SR region.

(A) A schematic representation of the gene desert upstream of the mouse Sox9 gene and the locations of the putative enhancers identified by ATAC-seq and DNaseI-seq that were screened in vivo with transgenic reporter mice. Enhancers that did not drive gonad expression of LacZ are shown in gray. Enhancers that drove testis-specific and ovary-specific LacZ expression are shown in blue and pink, respectively. The mouse regions that show conserved synteny with the human XY SR and REV SEX are depicted in green and purple boxes, respectively. (B) Enh13 (gray box) is located at the 5′ side of the 25.7-kb mouse equivalent XY SR locus (heavy black line). Data from DNaseI-seq (black) on E15.5 and E13.5 XY sorted Sertoli cells and ATAC-seq on E13.5 sorted Sertoli cells (blue) and granulosa cells (purple), as well as E10.5 sorted somatic cells, at the Enh13 genomic region are presented. Peaks correspond to nucleosome-depleted regions and are marked by a black horizontal line if they are significantly enriched compared to flanking regions, as determined by model-based analysis of ChIP-seq, and present in at least two biological replicates. The gray box overlaying the peak indicates the cloned fragment. Green areas represent sequence conservation among mice, humans (Hu), rhesus monkeys (Rh), cows (Co), and chickens (Ch) (sequence conservation tracks were obtained from the University of California–Santa Cruz). (C) β-Gal staining (blue) of E13.5 testes and ovaries from two representative independent stable Enh13 transgenic (Tg) lines. Scale bars, 100 μm.

All 16 putative enhancers were cloned upstream of an Hsp68 minimal promoter and the reporter gene LacZ and used to generate transgenic mice (2, 19) (table S1). For initial screens, we performed transient analyses at E13.5. Twelve enhancers failed to produce any gonadal β-galactosidase (β-Gal) activity, although many showed staining in other tissues in which Sox9 is normally expressed, such as chondrocytes, brain, and spinal cord (fig. S3). The remaining four showed gonad expression, and these constructs were reinjected to generate stable lines in order to better study their activity in both males and females during development. Enh8 [672 base pairs (bp) long, 838 kb 5′] conferred robust β-Gal activity in the ovary, whereas it was barely present in the testis at E13.5 (Fig. 1A and fig. S4B). This may be due to Enh8 being taken out of its original genomic context; notably, ATAC-seq revealed a much stronger peak in granulosa cells than in Sertoli cells (fig. S4A).

In contrast, Enh14 (1287 bp long, 437 kb 5′) showed robust testis-specific β-Gal activity (Fig. 1A and fig. S5B). DNaseI-seq, ATAC-seq, and H3K27ac ChIP-seq data all suggest that this enhancer is active and open only in Sertoli cells (figs. S1 and S5A). To test this candidate, we used genome editing to delete Enh14. However, Enh14 deletion did not alter expression of Sox9; its target gene Amh; or Foxl2, a marker of granulosa cells, in E13.5 XY gonads (fig. S5D), indicating that Enh14 has a redundant role, at least in the embryo. Enh32 (970 bp long, 10 kb 5′) is also testis specific but very weak and restricted to a domain close to the mesonephros (Fig. 1A and fig. S6, B and C). ATAC-seq, DNaseI-seq, and H3K27ac ChIP-seq data suggest that Enh32 is a Sertoli cell enhancer, although weak peaks were seen in the granulosa cell samples (figs. S1 and S6A).

The remaining enhancer, Enh13 (557 bp long, 565 kb 5′), is highly conserved among mammals and is located toward the distal 5′ end of a 25.7-kb region in mice that shows conserved synteny with a 32.5-kb region upstream of human SOX9 termed XY SR, the deletion of which is associated with sex reversal (20) (Fig. 1, A and B, and fig. S1). Enh13 shows the strongest Sertoli cell–specific peak within this region in both the DNaseI-seq and ATAC-seq data. H3K27ac ChIP-seq data mark Enh13 as active in both Sertoli and granulosa cells, which may support the observation that some transgenic lines also exhibit β-Gal activity in the ovary (Fig. 1C and figs. S1 and S7A). β-Gal expression is clearly within Sertoli and granulosa cells (fig. S7C). ATAC-seq data from E10.5 genital ridges show that Enh13 is not open at this stage, irrespective of chromosomal sex (Fig. 1B and fig. S1), suggesting that it opens coincident with specification of the supporting cell lineage from SF1-positive cells of the coelomic epithelium (5).

Genome editing was used to derive mice homozygous for deletions of Enh13. Homozygous deletion always led to XY female development, whether in a TES mutant background or in a wild-type background (Fig. 2 and figs. S8 to S10). The latter result was surprising, because if TES accounts for 55% of Sox9 expression in early Sertoli cells, any additional enhancer(s) should not account for more than 45% and, when this enhancer is deleted, levels of Sox9 should remain higher than the threshold of ~25% below which sex reversal might be expected (17). Nevertheless, whereas XY Enh13+/− embryos still undergo normal testis development, XY Enh13−/− embryos produce ovaries indistinguishable from those of XX wild-type embryos, with no signs of testis cords or a coelomic vessel (Fig. 2, B and C, and fig. S10). Immunofluorescence analysis of E13.5 and 6-week-old XY Enh13+/− and Enh13−/− gonads for SOX9 and FOXL2 showed that the former are still testes whereas the latter are fully sex-reversed ovaries (Fig. 2C and fig. S10D). Similar analysis of XX gonads with Enh13 deletion did not show any obvious phenotype (fig. S11).

Fig. 2 Deletion of Enh13 leads to complete XY male-to-female sex reversal.

(A) A schematic representation of the locations of Enh13 and TES upstream of Sox9. Blue and purple arrows represent the external and internal single guide RNAs, respectively, used to delete Enh13. Black arrows represent the PCR primers used to genotype embryos and mice with Enh13 deletion. Chr11, chromosome 11. (B) Bright-field (BF) pictures and hematoxylin and eosin (H&E)–stained sections of E13.5 XY Enh13+/+, Enh13+/−, and Enh13−/− and XX Enh13+/+ gonads. (C) Immunostaining of E13.5 XY wild-type, Enh13+/−, and Enh13−/− and XX wild-type gonads. Gonads were stained for Sertoli marker SOX9 (green), granulosa marker FOXL2 (red), and 4′,6-diamidino-2-phenylindole (DAPI) (blue). Sex-reversed gonads are indistinguishable from wild-type XX gonads, whereas the heterozygous deletion does not appear to alter testis morphogenesis. Scale bars in (B) and (C), 100 μm.

The Enh13 deletion was generated in a C57BL/6J genetic background, which is sensitized toward XY female sex reversal (21). To test the strength of the deleted allele, we therefore backcrossed the deletion into a mixed C57BL/6J × CBA background. As before, XY heterozygotes presented as normal fertile males whereas homozygotes showed full male-to-female sex reversal (fig. S12).

We could detect no difference in gonadal phenotypes between Enh13−/−:TES+/+ and Enh13−/−:TES−/− embryos or mice, suggesting that homozygosity for the Enh13 deletion alone reduced Sox9 levels well below the critical threshold required for testis development, which had been determined at E13.5 (17) (figs. S8 and S10). However, examining levels of gene expression at this stage when there is sex reversal will be uninformative because factors such as WNT4 and FOXL2 repress Sox9 once ovary development begins (5). We therefore analyzed Sox9, Sry, Sf1, and Foxl2 mRNAs at E11.5, during the brief period when gonadal sex is being determined. Real-time quantitative polymerase chain reaction (RT-qPCR) revealed that XY Enh13+/− and Enh13−/− genital ridges expressed 58 and 21% of the wild-type levels of Sox9 mRNA, whereas XY TES+/− and TES−/− genital ridges showed 55 and 50%, respectively (Fig. 3A). Control XX genital ridges contained 18% of the Sox9 mRNA levels found in XY genital ridges (Fig. 3A). Therefore, E11.5 XY Enh13−/− gonads express Sox9 at levels close to those of XX gonads at the same stage, explaining the observed complete sex reversal. Deleting one or two copies of TES had relatively little effect at E11.5, especially compared with the effect at E13.5, in contrast to the results with Enh13 deletions at E11.5 (Fig. 3A) (17). This again supports the conclusion that Enh13 plays a more substantial role than TES during early gonadal development.

Fig. 3 Enh13 regulates the expression of the Sox9 gene in vivo.

(A to D) RT-qPCR analysis of genes involved in male (Sox9, Sry, and Sf1) and female (Foxl2 and Sf1) gonadal sex determination in E11.5 XY gonads with Enh13 deletion and/or TES deletion (18 tail somites). Data are presented as mean 2−ΔΔCt values normalized to the expression of the housekeeping gene Hprt. The sample sizes (n) listed below each genotype are the numbers of individuals. Error bars show SEM of 2−ΔΔCt values. P values are represented above the relevant bars (unpaired, two-tailed t test on 2−ΔΔCt values; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). WT, wild type.

Sry expression is normally down-regulated as SOX9 levels increase, but it can persist if testis differentiation fails (22, 23). This is consistent with a direct or indirect repressive effect of SOX9 on Sry. At E11.5, Sry mRNA levels were higher than wild-type levels in both Enh13+/− and Enh13−/− XY gonads (168 and 152%, respectively) (Fig. 3B). In contrast, the TES deletion did not significantly alter Sry expression (Fig. 3B), which is expected because Sertoli cell differentiation is proceeding in TES mutants. SF1 is known to interact first with SRY and later with SOX9 to regulate Sox9 expression levels and also many of their downstream target genes (16). We found no significant changes in levels of Sf1 mRNA with any of the enhancer deletions at E11.5 (Fig. 3C). Using Foxl2 as an early marker of granulosa cell differentiation (24), we found that mRNA levels in XX wild-type gonads at E11.5 were 3.6 times as high as those in XY gonads (Fig. 3D). Compared with the latter, Enh13+/− and Enh13−/− XY gonads showed two- and threefold increases, respectively, with the homozygotes having mRNA levels very close to XX control levels. Therefore, Enh13−/− XY gonads reveal an early commitment to the ovarian pathway.

There was a 30 to 50% decrease in Sox9 mRNA levels in E11.5 XX Enh13−/− gonads compared to the wild type, as reflected by reduced immunofluorescence for SOX9 protein (fig. S11). These data indicate that Enh13 also plays a role in the very early expression of Sox9 in the XX gonad, consistent both with the small peak seen with ATAC-seq and with occasional reporter activity in the transgenic mouse assays (Fig. 1, B and C, and fig. S7).

The sequence of Enh13 is highly conserved among mammals (Fig. 1B) and contains consensus binding sites for transcription factors known to regulate early gonad development and sex determination (fig. S13) (6). Mouse Enh13 contains a single consensus SRY binding site as well as a SOX9 site to which SRY can also bind (Fig. 4A and fig. S13). We performed ChIP-qPCR on E11.5 gonads dissected from Sry-Myc transgenic embryos by using a specific antibody against the MYC tag (22). SRY-MYC–positive gonads had an 11-fold enrichment versus SRY-MYC–negative gonads with primers spanning the SOX9 consensus site and a sixfold enrichment with primers spanning the SRY site, whereas primers against the strongest SRY binding site in TESCO (22) showed fivefold enrichment (Fig. 4, A and B). This reveals the strong binding of SRY to Enh13 at E11.5, with a preference for the SOX9 consensus site, possibly due to the adjacent SF1 binding site. Preferential binding of SRY to Enh13 over TESCO at E11.5 supports the hypothesis that the former is more critical because it initiates up-regulation of Sox9, whereas the latter is secondary.

Fig. 4 Enh13 is bound by SRY and SOX9 in vivo.

(A) A schematic representation of the locations of the primers used for ChIP experiments in Enh13. BS, binding site. (B) ChIP-qPCR assay of E11.5 mouse genital ridges after immunoprecipitation with anti-cMYC antibody. Data are presented as the level of enrichment of SRY-MYC–positive relative to SRY-MYC–negative genital ridges, meaning that values greater than 1 (marked by the dotted line) represent specific enrichment. The primers used span the putative SRY binding site in Enh13 and TESCO [around the SRY R6 site; see (16)] and a negative control region on chr11. Data are the means ± SD (n = 2); *P ≤ 0.05, **P ≤ 0.01, ****P < 0.0005 (Student t test). (C) ChIP-qPCR assay of E13.5 mouse testes after immunoprecipitation with anti-SOX9 antibody. The primers used span the putative SOX9 binding site in Enh13 and TESCO [around the SOX9 R1 site; see (16)] and a negative control region on chr11. Data are the means ± SD (n = 3). ****P < 0.0005; ns, not significant (Student t test). Unr ab, unrelated antibody (rabbit IgG isotype) used as a control. (D) ChIP-seq with the use of anti-SOX9 antibody and E90 fetal bovine testis. The bovine (bov) Enh13 is indicated by the gray box (at cow Bostau8 chr19: bp 60,063,628 to 60,064,165). The asterisk denotes the peaks with a false discovery rate of <0.05 in the two bovine datasets. y-axis numbers represent counts. The input tracks represent sequencing reads of chromatin input. The bovine Sox9 gene is indicated by the arrow and is 570 kb downstream (to the left) of Enh13.

ChIP assays revealed SOX9 to be bound at similar levels to both Enh13 and TESCO in cells from E13.5 testes (Fig. 4C). In addition, SOX9 ChIP-seq data obtained with bovine embryonic testis (25) revealed a strong peak localizing to the conserved syntenic region of Enh13 (537 bp long, 570 kb 5′) (Fig. 4D). This suggests that, like TES, Enh13 is used by SOX9 to autoregulate SOX9 expression and that this interaction is conserved in mammals. Unlike several other gonadal enhancers, Enh13 appears to be well conserved, has a clear role in mice to initiate up-regulation of Sox9 expression in response to SRY activity, and may contribute to maintaining Sox9 expression. This makes it very likely to play a similar role in humans, given its location within the XY SR region (20, 26). If it does play a similar role, heterozygosity for Enh13 deletions in humans should mimic heterozygosity for null mutations in SOX9, with XY female sex reversal occurring in about 70% of cases but perhaps without other CD phenotypes.

It is clear from our data and those of others that the upstream regulatory region of Sox9 is very complex. Our screens with gonadal cells revealed 33 potential enhancers distributed over 1.5 Mb. Transgenic assays used to test the most promising 16 enhancers revealed 4 that produced expression in the gonads, whereas the majority did not. This “hit rate” of 25% agrees with the results of some other studies (19) and merits caution in interpreting data based solely on the accessibility of chromatin and histone marks. However, several of these putative enhancers are bona fide enhancers in other locations; for example, Enh29, mapping about 70 kb 5′, is equivalent to SOM, an enhancer active in many tissues, excluding the gonads (27). Others may have distinct roles; Enh11 contains a putative CTCF binding site. In addition, several putative enhancers, notably Enh4, -5, -8, and -9, appear to be open in both granulosa and Sertoli cells, with Enh8 even more so in the former. These may contribute to the low level of Sox9 expression seen in supporting cell precursors, but they may also represent sequences required to repress Sox9, which might not be detected by transgenic reporter assays.

The notion of redundancy or “shadow enhancers” within a regulatory region is well established (28, 29), and recent data suggest that deletion of single, even “ultraconserved” enhancers from developmentally important genes can have at most subtle if not undetectable effects (30, 31). It is therefore remarkable to see that deleting Enh13 alone phenocopies the loss of Sox9 itself within the supporting cell lineage (9, 32). Substantial evidence points to the time-dependent action of SRY on Sox9. If Sox9 fails to reach a critical threshold within a few hours, then ovary-determining and/or antitestis factors, such as Wnt signaling, accumulate to a sufficient level to repress Sox9 and make it refractory to male-promoting factors, including SRY, even though expression of the latter persists in XY gonads when Sertoli cells fail to differentiate (33). We suggest that Enh13 is an early-acting enhancer, such that without it Sox9 transcription fails to increase to a level where the other enhancers can act before the gene is silenced. It is only later that TES, and perhaps other enhancers such as Enh14 and Enh32, begins to act in a more redundant fashion, although it is conceivable that each enhancer has a major role to play during distinct phases of Sertoli cell development from the fetal to the adult testis. It will be of interest to determine how Enh13 activity cascades into the recruitment of the other enhancers.

Supplementary Materials

Materials and Methods

Figs. S1 to S13

Tables S1 to S4

References (3444)

References and Notes

Acknowledgments: We dedicate this paper to the memory of Danielle M. Maatouk, a much-missed colleague. We are grateful to the Biological Research Facility, Genetic Modification Service, Advanced Sequencing Facility, and Experimental Histopathology Facility of the Francis Crick Institute and the Flow Cytometry Core at Northwestern University for technical assistance. We thank members of our labs for advice, support, and helpful comments. Funding: This work was supported by the Francis Crick Institute, which receives its core funding from Cancer Research UK (FC001107), the U.K. Medical Research Council (FC001107), and the Wellcome Trust (FC001107), and by the U.K. Medical Research Council (U117512772). F.P. was funded by the Agence Nationale pour la Recherche (ANR blanc TestisDev). D.M.M. was funded by the Northwestern University School of Medicine. Author contributions: N.G. and R.L.-B. designed the study. C.R.F., S.A.G.-M., I.M.S., and D.M.M. performed ATAC-seq and H3K27ac ChIP-seq and cloned the enhancer-LacZ plasmids. S.C.S. helped with analyzing the reporter mice. S.W. performed cytoplasmic and pronuclear zygote injections. R.S. provided the TES mutant (TESMS)–cyan fluorescent protein (CFP) transgenic mice. F.P. performed the mouse and bovine SOX9 and SRY ChIP. N.G. performed the rest of the experiments. N.G. and R.L.-B. analyzed and interpreted the results and wrote the manuscript. All authors reviewed and added input to the manuscript. Competing interests: The authors have no competing interests. Data and materials availability: All data are available in the main text or the supplementary materials. ATAC-seq and H3K27Ac ChIP-seq data have been deposited in the Gene Expression Omnibus under accession number GSE99320.

Stay Connected to Science

Navigate This Article