Human sex reversal is caused by duplication or deletion of core enhancers upstream of SOX9

Abstract

Disorders of sex development (DSDs) are conditions affecting development of the gonads or genitalia. Variants in two key genes, SRY and its target SOX9, are an established cause of 46,XY DSD, but the genetic basis of many DSDs remains unknown. SRY-mediated SOX9 upregulation in the early gonad is crucial for testis development, yet the regulatory elements underlying this have not been identified in humans. Here, we identified four DSD patients with overlapping duplications or deletions upstream of SOX9. Bioinformatic analysis identified three putative enhancers for SOX9 that responded to different combinations of testis-specific regulators. All three enhancers showed synergistic activity and together drive SOX9 in the testis. This is the first study to identify SOX9 enhancers that, when duplicated or deleted, result in 46,XX or 46,XY sex reversal, respectively. These enhancers provide a hitherto missing link by which SRY activates SOX9 in humans, and establish SOX9 enhancer mutations as a significant cause of DSD.

Fig. 1

Duplication or deletion of the human SOX9 testicular enhancer eSR-A is associated with DSD. a The 600 kb genomic region upstream of human SOX9 showing the XYSR, RevSex and TESCO candidate regulatory regions. b The XYSR region was defined by deletions in two previously published 46,XY DSD patients (blue). Two novel duplications in 46,XX DSD patients (grey), allowed us to redefine the minimal overlap to 5.2 kb (green). c Sub-cloning strategy, dark blue lines indicate sub-clones analysed for enhancer activity in the pGL4.10 Beta-globin (βg) plasmid using luciferase assays. Predicted transcription factor binding motifs for SF1 (SF1-a and SF1-b) and SOX9 are shown with vertical light blue and magenta lines. Bioinformatic tracks from the UCSC genome browser are shown including the ENCODE track of enhancers present in human mammary epithelial cells (HMEC) (yellow denotes a weak enhancer) and DNaseI hypersensitivity data from human foetal testis and ovary (ROADMap). This shows a testis-specific peak over the a4 fragment. The 100-vertebrate conservation track shows a spike of conservation beneath the a4 fragment DNaseI peak. d Enhancer activities of sub-cloned fragments a1-a5 as measured by luciferase assays transfected with SF1 and SOX9 (n = 4). e Mutation of the SF1-a (eSR-A ΔSF1-a), and SOX9 (eSR-A ΔSOX9) binding motifs either separately or together results in a loss of enhancer activity compared to non-mutated eSR-A as assessed by luciferase assays with co-transfection of SF1 and SOX9 (n = 4). f. Co-transfection of FOXL2 with SF1 + SOX9 shows a repression of eSR-A activity compared to SF1 + SOX9 (n = 3). All luciferase assays carried out in COS7 cells. Error bars are s.e.m. P-values (two-tailed t tests): *P ≤ 0.05. **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Source data are provided as a Source Data file.

Fig. 2

Identification and analysis of the SOX9 enhancer eSR-B. a Schematic of a number of previously published 46,XX DSD duplication/triplication (grey)^,,, and 46,XY DSD deletion (blue)^, patients upstream of SOX9. The minimal overlap (RevSex) of the six patients (24 kb) in magenta. b Unbiased luciferase tiling strategy of the minimal RevSex region (16 constructs—blue) to test for enhancer activity using the PGL4-βg vector. c Two fragments, b8 and b9, showed significant enhancer activity and their overlapping fragment, eSR-B was sub-cloned (blue lines). A conserved SOX9 binding site was identified (magenta line and box) and UCSC genome browser ENCODE data (HMEC) highlighted a strong active enhancer site (orange), conservation tract indicated a conserved sequence (blue peaks). No DNaseI peaks were detected in human foetal testis or ovaries (green). d Enhancer activities of small fragments (b1-b16) of the minimal overlap in COS7 cells transfected with SF1 and SOX9 in luciferase assays. Only the fragments b8 and b9 showed significant enhancer activity compared to the empty vector PGL4-βg (n = 3). e Enhancer activities of b8 and b9 overlapping region eSR-B as assessed by luciferase assays in COS7 cells co-transfected with SF1 (white bars) and SOX9 alone (grey bars) and in combination (blue bars) (n = 4). Significance is compared to the PGL4-βg empty vector control. f Enhancer activity of eSR-B fragment with mutation of the SOX9 (eSR-B ΔSOX9) binding site motif in response to SOX9 transfection (n = 7) significance compared to the eSR-B non-mutated fragment. g Luciferase reporter assays show FOXL2 transfected with SOX9 represses eSR-B enhancer activity compared to SOX9 alone (n = 4). All Luciferase assays are carried out in COS7 cells. Error bars are s.e.m. P-values derived from two-tailed t tests: *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Source data are provided as a Source Data file.

Fig. 3

Identification and analysis of the SOX9 enhancer eALDI. a Schematic of the upstream region of SOX9 highlighting human TESCO (hTESCO) and the proposed enhancer eALDI. Bioinformatic data from UCSC genome browser. ENCODE data (HMEC) indicated a strong active enhancer (orange) covering the proposed eALDI region. DNaseI peaks were observed in one foetal testis track (dark green), and strong conservation across the region was observed (blue peaks). A predicted SRY transcription factor binding motif is presented by magenta vertical line and box. By comparison ENCODE and DNaseI data for hTESCO show less enhancer potential and lower levels of conservation. b Enhancer activity of eALDI and hTESCO measured by luciferase assays in COS7 cells when co-transfected with SOX9 alone (grey bars), SOX9 + SF1 (blue bars) and SF1 + SRY (green bars). The eALDI region shows significant enhancer activity in response to both SF1 + SOX9 and SF1 + SRY transfection when compared to the empty pgl4-βg (n = 4). c Mutation of the SRY (eALDIΔSRY) transcription factor motif in the eALDI enhancer reduced activity when co-transfected with SF1 + SOX9 or SF1 + SRY compared to unmuted eALDI. Error bars represent s.e.m. P-values derived from two-tailed t tests: **P ≤ 0.01, ****P ≤ 0.0001. d Luciferase reporter assays in COS7 show FOXL2 transfected with SF1 + SOX9 represses eALDI enhancer activity compared to co-transfection with SOX9 + SF1 alone (n = 3) Error bars are s.e.m. P-values derived from two-tailed t tests: ****P ≤ 0.0001. e CRISPR/Cas9 mediated knock-out of eALDI resulted in reduction of Sox9 expression to 40% of wild type at E11.5 and f 51% of wild-type levels in mouse testis at E14.5 as assessed by qRT-PCR. At 11.5 n = 6 XY WT, XY eALDI n = 6 and XX WT n = 3. At E14.5 XY WT n = 6, XY eALDI n = 6 and XX WT n = 4. Error bars are s.e.m. P-values derived from t-tests: *P ≤ 0.05 and ***P < 0.001 compared to wild-type males. Source data are provided as a Source Data file.

Fig. 4

Testis-specific SOX9 enhancer synergy and a proposed model for SOX9 activation. Luciferase activity of three novel human SOX9 enhancers individually and in tandem for a SRY + SF1 (green) and b SOX9 + SF1 (blue) in COS7 cells. n = 5, Error bars represent s.e.m. Fold activation was determined by average luciferase activity of the tandem construct, relative to the sum of the average luciferase activity of the individual enhancers. This value is shown above the error bars. c Proposed model of SOX9 activation. In XY individuals SRY and SF1 are expressed early in the human XY foetal gonad, and initiate SOX9 expression via the eALDI enhancer. SOX9 expression is then further upregulated and maintained by SF1 and SOX9 with all three enhancers eSR-A, eSR-B and eALDI potentially via chromatin looping, allowing direct interaction between the enhancers and the SOX9 promoter. Source data are provided as a Source Data file.