How NR5A1 gene variants drive 46,XY sex development disorders

Variants in a single “master switch” gene called NR5A1 have long been known to disrupt male sexual development, but why one child is severely affected while another with a similar mutation escapes with milder features has remained one of the most stubborn puzzles in pediatric endocrinology. A new study published in the journal Biology of Sex Differences now offers a detailed molecular explanation, tracing how four different patient-derived variants of the gene cripple the protein it produces, derail its journey into the cell nucleus, and silence a network of genes essential for making male hormones. In the process, the researchers have uncovered two genes — AMHR2 and STARD8 — that appear to be newly recognized targets of NR5A1, a discovery that could reshape how clinicians interpret genetic test results in children with differences of sex development.

NR5A1, also known as steroidogenic factor-1 or SF-1, sits on chromosome 9q33 and encodes a 461-amino-acid nuclear receptor that functions as a transcription factor — a protein that binds specific DNA sequences and switches other genes on. The protein is built from several functionally distinct parts: a DNA-binding domain with two zinc-finger motifs, an auxiliary hinge region, a ligand-binding domain, and two activation-function domains that recruit the cellular machinery needed to drive transcription. During embryonic development, NR5A1 acts upstream of nearly every major event in male sex determination. It is expressed in both Leydig cells, which produce testosterone, and Sertoli cells, which nurture developing germ cells and secrete anti-Müllerian hormone, the signal that causes the Müllerian ducts — the precursors of the uterus and upper vagina — to regress in male embryos. Its expression even precedes that of SRY, the gene that initiates testis formation. Variants in NR5A1 are inherited in an autosomal dominant manner and account for roughly 10 to 15 percent of 46,XY disorders of sex development, a spectrum of conditions in which individuals with one X and one Y chromosome develop genitalia that do not align with typical male anatomy.

The research team, led by investigators at Shanghai Children’s Hospital affiliated with Shanghai Jiao Tong University, identified four naturally occurring NR5A1 variants in children evaluated between 2018 and 2023. Two of the variants — p.Cys65Ser and p.His310Arg — had never been reported before, while the other two, p.Cys30Ser and p.Gln329*, were previously known but had never been functionally characterized in depth. All four patients presented with severe proximal hypospadias, in which the urethral opening sits abnormally far down the penis, and a bifid scrotum, reflecting incomplete fusion of the genital folds. Three of the four also had undescended testes confined to the groin. Endocrine testing painted a consistent picture of primary gonadal insufficiency: follicle-stimulating hormone levels were elevated at baseline in all four children and climbed sharply after stimulation, while anti-Müllerian hormone — a marker of Sertoli cell mass — was measurable in every patient but varied widely, from 11.82 to 46.45 pmol/L. The child carrying the novel p.Cys65Ser variant had the lowest AMH and inhibin B values and the most severe external masculinization score, suggesting the greatest compromise of Sertoli cell function. Notably, ultrasound found no Müllerian structures in any patient, indicating that even the lowest AMH level had been sufficient during fetal life to trigger complete duct regression. Two children initially assigned female at birth were later reassigned and raised as boys after hormonal testing confirmed functional testicular tissue and an adequate androgen response to human chorionic gonadotropin stimulation.

Before moving into the laboratory, the team examined the evolutionary and structural context of each mutation. All four altered amino acids sit at positions that are highly conserved across species, a hallmark of functional importance. Computational modeling using SWISS-MODEL and visualization in PyMOL showed that the p.Gln329* nonsense variant — which introduces a premature stop codon — truncates the protein by 132 amino acids at its C-terminus, deleting two beta-sheets, six alpha-helices, and the entire activation-function-2 domain needed to recruit co-activator proteins. The two cysteine substitutions, p.Cys30Ser and p.Cys65Ser, strike the first and second zinc-finger regions of the DNA-binding domain respectively; while chemically subtle, each substitution introduces new polar interactions, including one with the zinc ion that stabilizes the DNA-binding fold, that may subtly distort the geometry required for high-affinity sequence recognition. The p.His310Arg substitution in the ligand-binding domain swaps a histidine for a bulkier arginine while preserving most local hydrogen bonds, but adds an unexpected new contact with a neighboring valine residue.

The functional consequences were unambiguous. When the researchers expressed wild-type or mutant NR5A1 in HEK293T cells — a line chosen in part because it does not produce NR5A1 endogenously, allowing clean measurement of transfected constructs — western blotting revealed that all four variants yielded substantially less protein than the wild-type receptor. Confocal immunofluorescence microscopy then showed that whereas the normal protein accumulates exclusively in the nucleus, where it must reside to bind DNA, all four mutant proteins showed markedly reduced nuclear signal and a corresponding shift into the cytoplasm. A transcription factor trapped outside the nucleus is, in effect, a transcription factor lost: the dual defect in protein abundance and localization provides a direct mechanistic account of why each variant behaves as a loss-of-function allele despite sometimes modest changes at the amino acid level.

To map the downstream wreckage, the team turned to the p.Gln329* truncating variant, which had shown the strongest effects in earlier assays and carries clear clinical relevance. RNA sequencing of cells expressing the mutant versus the wild-type protein identified 359 differentially expressed genes — 85 upregulated and, tellingly, 274 downregulated. Among the most strongly suppressed were canonical pillars of steroidogenesis: CYP11A1, which catalyzes the first committed step converting cholesterol into pregnenolone inside mitochondria; STAR, which shuttles cholesterol to that enzymatic machinery; and CYP17A1 and CYP21A2, which carry steroid synthesis through subsequent steps. Gene Ontology and KEGG pathway analyses confirmed that the downregulated genes clustered overwhelmingly in steroid biosynthesis, steroid and cholesterol metabolism, cortisol and aldosterone synthesis pathways, and cholesterol transport — exactly the molecular programs a developing testis depends on to masculinize the fetus.

The decisive technical advance came from combining the transcriptomic data with CUT&Tag — Cleavage Under Targets and Tagmentation — a technique that uses an antibody-guided transposase to map where a protein of interest contacts the genome. CUT&Tag on the wild-type receptor revealed binding peaks enriched in promoter regions and a recurring “CGCC” motif that likely represents part of the NR5A1 consensus binding element. When the same experiment was run with the p.Gln329* mutant, locus-specific enrichment collapsed at the promoters of two genes that had also emerged as strongly downregulated in the RNA-seq data: AMHR2, which encodes the receptor that mediates anti-Müllerian hormone signaling, and STARD8, a less-studied member of the steroidogenic acute regulatory-related lipid transfer domain family. Quantitative PCR validated the reduced messenger RNA levels of both genes, and the convergence of expression loss with loss of genomic binding made AMHR2 and STARD8 the study’s prime candidates for novel NR5A1 targets.

Dual-luciferase reporter assays then tested whether the variants actually impair the ability of NR5A1 to activate these promoters. Wild-type NR5A1 robustly increased the activity of the CYP11A1, AMHR2, and STARD8 promoter constructs compared with an empty vector. All four patient variants significantly diminished activation of the CYP11A1 and AMHR2 promoters, confirming these as targets vulnerable across the variant spectrum. STARD8 told a subtler story: the three missense variants left its promoter activity essentially untouched, while only the truncating p.Gln329* variant caused a marked decrease. This variant-specific pattern is intriguing because it suggests that STARD8 regulation is exquisitely sensitive to the integrity of the receptor’s C-terminal activation-function-2 region, and it may help explain why truncating variants can produce clinical phenotypes that missense variants in the same domain do not.

The clinical implications run in several directions. AMHR2 dysfunction is already known to cause persistent Müllerian duct syndrome, and the finding that NR5A1 variants suppress AMHR2 expression may clarify why a minority — roughly 17 percent — of NR5A1-related patients retain Müllerian structures despite measurable AMH: the hormone may be present, but its receptor may be transcriptionally starved. STARD8 has a more mysterious history; a previous case report described two sisters with 46,XY gonadal dysgenesis carrying variants in the gene, and work in fruit flies supports a conserved role for the protein’s family in testis development, but direct functional evidence in humans had been lacking. The new data position STARD8 as a context-dependent NR5A1 target that may participate in Leydig cell differentiation and lipid handling during steroidogenesis. The authors caution that their experiments were performed in a non-gonadal cell line, that CUT&Tag demonstrates physical association rather than definitive functional binding, and that the transcriptomic and CUT&Tag analyses covered only the truncating variant — validation in physiologically relevant gonadal cell models and larger patient cohorts will be needed. Even so, by connecting specific molecular lesions to reduced protein output, failed nuclear import, and the selective silencing of hormone-synthesis and sex-development genes, the study converts a list of genetic variants into a mechanistic narrative — and takes a substantial step toward predicting, from a DNA sequence alone, how deeply a given NR5A1 mutation will disturb the intricate choreography of human sex development.