We identified 19 patients (8.5%) with NR5A1 mutations among a total of 223 46, XY DSD patients, including 10 previously reported patients (I10-I19) [27] and 9 newly described patients (I1-I9). The ages of the 19 patients ranged from 1 to 28 years, with a median age of 15 years. The patients were distributed into different developmental stages, namely, childhood, adolescence and young adulthood, and included 8 prepubertal patients and 11 pubertal and postpubertal patients. Among the 19 patients, 11 patients were of the female social gender with chromosome karyotypes 46, XY, and 8 patients were of the male social gender (Table S3).

Ten variants (6 missense: p.T29K, p.G35V, p.G212S, p.E367G, p.C370Y and p.S430I, 1 nonsense: p.C283*, 2 frameshift: p.R89fsX10 and p.E148fsX295 and 1 deletion: p.N44del) were identified in previously reported cases [27]. Eight variants, including 3 missense (p.Q42E, p.R84C and p.R313C), 1 nonsense (p.Y5*), 1 frameshift (p.Y404fs) and 3 indels (p.L359_L363del, exon 6 duplication and 0.36 Mb CNV chr9:127,213,317-127570245_del containing NR5A1, NR6A1, MIR181A2HG, MIR181A2, MIR181B2 and OLFML2A) were identified in 9 new patients from 9 unrelated families (Fig. 1). p.R84C was found in both I3 and I4. According to the pedigree analysis (Fig. S1), the mutations identified in I1, I6, I8 and I9 were proven to be de novo. The p.Y5*, p.Q42E, p.L359_L363del, chr9:127,213,317-127570245_del (0.36 Mb) CNV and exon 6 duplication of NR5A1 were novel mutations identified in this study for the first time. The p.R84C [33, 34], p.R313C [34,35,36,37,38] and p.Y404fs [39, 40] mutations have been reported in previous studies. Functional studies have also been performed on p.R84C and p.R313C, so they served as positive controls in the following functional studies. The mutation sites were distributed mainly in the DBD and LBD (Fig. S2). The pathogenicity of these variants, assessed according to the ACMG guidelines, is summarized in Table 1. Variants in other phenotypically associated genes carried by patients are listed in Table S4.

NR5A1 heterozygous mutations detected in this study. c.15C > A; p.(Tyr5*) was identified in patient I1. c.124C > G; p.(Gln42Glu) was identified in patient I2. c.250C > T; p.(Arg84Cys) was identified in both patient I3 and patient I4. c.937C > T; p.(Arg313Cys) was identified in patient I5. c.1075_1089del; p.(Leu359_Leu363del) and c.1211_1212insTA; p.(Tyr404fs) were identified in patients I6 and I7, respectively. Patient I8 had an exon 6 duplication in NR5A1. Patient I9 had chromosome 9: 127,213,317-127570245_del (0.36 Mb del), containing NR5A1, NR6A1, MIR181A2HG, MIR181A2, MIR181B2 and OLFML2A

All 19 patients were normotensive, and had no clinical manifestations of adrenal insufficiency, such as skin pigmentation, nausea, vomiting or fatigue. These patients presented with variable severity of micropenis (18/19, 95%) and different types of hypospadias (18/19, 95%). The anatomical locations of the testes could be intrascrotal (10/36, 28%), inguinal (22/36, 61%) or abdominal (3/36, 8%). The EMS of patients reared as females was significantly lower than that of patients reared as males (P < 0.001). The clinical features of the 19 patients are presented in Tables S3 and S4. Two patients (I1 and I13) changed social gender from female to male and underwent cryptorchidopexy.

Sex hormones were analyzed in 15 patients; I2, I3 and I6 underwent surgery for cryptorchidectomy before the hormone assay and data of I12 were not available. The testicular function of 8 prepubertal patients was assessed via hCG stimulation tests. After hCG stimulation, the post-hCG T levels of four patients (I8, I14, I15 and I17) significantly increased by 6–61 times compared with the basal T, reaching the level of > 110 ng/dL. Conversely, the remaining four patients (I4, I5, I11 and I16) demonstrated deficiencies in testosterone biosynthesis. Among the 7 pubertal and postpubertal patients, 5 patients (I1, I7, I9, I13 and I19) presented normal levels of serum testosterone, and two patients (I10 and I18) presented lower levels. The level of testosterone was variable, with a median of 2.54 (0.15, 4.07) ng/mL. Notably, patient I10 presented significantly reduced testosterone levels, indicating severe gonadal dysgenesis. The levels of gonadotropic hormones were elevated in all 7 pubertal and postpubertal patients, indicating impaired negative feedback inhibition to the pituitary. The increase in the serum level of FSH was more prominent than that of LH. Moreover, 77% of patients (10/13) exhibited lower AMH levels than the reference range for the respective age groups.

Adrenal hormone levels were evaluated in 15 patients. The 8 am ACTH levels were elevated in 5/15 patients (I1, I5, I9, I14 and I17). Based on both the 8 am cortisol levels and cortisol levels following ACTH stimulation, 8 of the 15 patients showed no significant adrenal cortical dysfunction, while the remaining 7 did not meet the threshold, precluding a direct exclusion of adrenal cortical insufficiency. Among the 9 patients from whom samples were collected after adrenarche, 3 patients (I3, I7, and I9) had low DHEA-S levels. Notably, DHEA-S levels in patients I3 and I9 remained significantly below the age-related lower limits even after stimulation. The stimulated serum levels of P, 17-OHP and AD were found to be elevated to various degrees. The laboratory features of the 19 patients are presented in Tables S5, S6 and S7.

Considering the accuracy of the steroid assay, we assessed 14 steroids by LC‒MS/MS, including 11-oxygenated steroids, which could also reflect part of adrenal function. Our findings revealed that the median values of 11OHT, 11KA4 and 11OHA4 in 11 patients (postadrenarche age) were lower than the reference values reported by Davio et al.[41, 42], which were based on a cohort of 69 men aged between 18 and 39 years (Fig. S3 and Table S8). Notably, individual I9, who exhibited extremely low DHEA-S levels, also presented reduced levels of 11-oxygenated steroids, indicating inadequate production of adrenal-derived C19 steroids.

All patients had ultrasound or radiological abnormalities of the internal genitalia at diagnosis. Three of the six patients (I3, I9 and I10) had delayed bone age. Notably, 6/8 patients (I2, I4, I5, I9, I17 and I19) had abnormal development of the spleen. I2, I5, I9 and I19 all had a spleen characterized by anomalous morphology and size with strip-shaped calcifications. I4 and I17 had normal spleen sizes and morphologies with punctate calcifications. Specifically, abnormalities in spleen size and morphology were observed in 4 out of 6 cases, whereas calcifications were present in all 6 cases. I9 exhibited an atypical (32 HU) and centimetric nodular lesion on the right adrenal gland, with a reduced volume of the remaining adrenal gland. The radiological features and typical radiographs at diagnosis are shown in Table S9 and Fig. S4.

Three promoter regions of the target genes (CYP11A1, CYP19A1 and AMH) of NR5A1 were included in the luciferase reporter assay. Compared with WT NR5A1, p.R84C and p.R313C have been reported to have decreased DNA-binding affinity and transcriptional activity [33, 35, 36], and they served as positive controls here. Compared with those in cells transfected with WT NR5A1, the variants constructed from 3 missense (p.Q42E, p.R84C and p.R313C), 1 frameshift (p.Y404fs), 1 nonsense (p.Y5*) and 1 deletion (p.L359_L363del) exhibited varying degrees of impaired transcription of the target gene (Fig. 2A–C).

NR5A1 gene transcriptional activity assays and protein expression analysis. (A-C) NR5A1 gene transcriptional activity assays of WT NR5A1 and its variants. A luciferase reporter assay was used to assess the transcriptional activities of WT NR5A1 and its variants. Three promoter regions of the target genes (CYP11A1, CYP19A1 and AMH) of NR5A1 were included. (D) Western blot analysis of the protein expression levels of WT NR5A1 and its variants. Anti-GAPDH was used as a control. (E) Quantitative graph of the western blot analysis (D). One-way ANOVA was used. Data were represented as mean ± SD. ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001

Compared with that in cells transfected with WT NR5A1, the expression of normal NR5A1 was lower in cells transfected with p.Y5*, p.R313C and p.Y404fs variants. The protein expression of p.Y5* was comparable to that of the 293 T cells transfected with or without empty vectors. The expression of p.L359_L363del was comparable to that of the wild type. We also observed the expression of the truncated protein p.Y404fs, which has a lower protein molecular weight than the wild-type protein. Interestingly, the two variants p.Q42E and p.R84C resulted in increased protein expression levels (Fig. 2D, E).

Subcellular localization analysis was performed to assess the nuclear localization abilities of the four variants. As shown in Fig. S5, in p.EGFP-N2-vector-transfected cells, EGFP green fluorescence was distributed in the cytosol and nucleus, whereas EGFP-tagged WT NR5A1 was localized in the nucleus with a uniform distribution. Certain variants, such as the p.T29K and p.N44del variants, are known to have abnormal subnuclear localization, which showed nuclear localization with fine subnuclear puncta in our previous study [27]. The WT and p.R84C NR5A1 proteins were previously shown to be expressed at comparable levels and localized exclusively to the nucleus by Reuter et al. [33]. Compared with the abnormal variants, the variants p.Q42E, p.R313C and p.L359_L363del presented a nuclear localization pattern similar to that of WT NR5A1 and the positive control p.R84C. They did not alter trafficking or stability.

Multiple sequence alignments of the NR5A1 protein from various species revealed that R84 and R313 presented greater evolutionary conservation than Q42 (Fig. S6).

Three-dimensional front and dorsal views of the wild-type NR5A1 protein are shown in Fig. S7A and B. p.Y5* led to a premature stop codon, yielding a mere 4-amino-acid truncation (Fig. S7C, D). Compared with Q42, the E42 residue retained interactions (hydrogen bonds) with K38 (1.9 Å) and R87 (2.4 Å) but lost the interaction with R84 (1.9 Å) (Fig. S7E). The change from R84 (basic amino acid) to C84 (sulfur-containing amino acid) resulted in the loss of interaction with G90 (1.8 Å), G91 (1.9 Å), and R92 (2.4 Å), with only Q42 remaining as a bonding partner at a distance of 2.0 Å (Fig. S7F). R313 interacted with three amino acid residues, E237 (1.7 Å and 1.8 Å), D309 (2.1 Å) and H317 (2.0 Å), whereas C313 interacted with D309 (2.0 Å), H310 (2.9 Å), Q316 (2.8 Å) and H317 (2.0 Å) (Fig. S7G). In general, the p.Q42E, p.R84C, and p.R313C mutations resulted in a single amino acid substitution, thereby altering the local conformation of the wild-type protein. Conversely, the deletion of 5 amino acids (p.L359_L363del) led to a truncated α-helical structure with a disordered coil region. (Fig. S7H). A truncated protein of 429 amino acid residues was observed in p.Y404fs (Fig. S7I).

Eight patients (aged 5 to 22 years) underwent testicular biopsy (I1 and I16) or orchiectomy (I2, I5, I6, I7, I9, I11 and I16). They all presented with cryptorchidism, in which the testes were significantly reduced in size. In eight patients, histological examination via HE staining revealed the predominant presence of Sertoli cells, with a limited number of spermatogonia observed within the seminiferous tubules. A few spermatids were observed in two patients (I2, 18 years old; I7, 22 years old). Additionally, the absence of spermatozoa and Leydig cell hyperplasia were also noted. Both I2 and I7 had Johnsen scores of 6, whereas the remaining patients had scores of 3. Anti-SOX9 and anti-CYP17A1 antibodies were used to label Sertoli cells and Leydig cells, respectively. Notably, all eight subjects presented comparable expression levels of SOX9 in Sertoli cells. However, variations in CYP17A1-positive Leydig cell staining were observed among them. Specifically, the expression of CYP17A1 was significantly lower in two prepubertal patients (I11, 2.9% and I16, 10.7%) than in the other six individuals (I1, I2, I5, I6, I7 and I9, ranging from 19.0% to 46.3%). The HE and IHC results are listed in Table 2. Representative histologic images of typical patients and normal males are shown in Fig. 3 and Fig. S8.

Histopathological testicular tissue analysis of patients with NR5A1 mutations. Representative histologic images of testicular tissue from four typical patients (I5, I6, I11 and I16) are shown, while the histologic images of normal male testicular tissue are presented in Fig. S8. Pathological sections were stained with HE (left), and Leydig cells and Sertoli cells were positively stained by IHC with antibodies against CYP17A1 (middle) or SOX9 (right). Testicular seminiferous tubules were lined with Sertoli cells, and a few spermatogonia were observed. CYP17A1 ( +)-Leydig cell staining differed among patients (I5-29.2%, I6-45.7%, I11-2.9%, I16-10.7%), whereas there was little difference in SOX9 ( +)-Sertoli cell staining. Scale bar:100 μm