Heterozygous mutations in SOX2 may cause idiopathic hypogonadotropic hypogonadism via dominant-negative mechanisms

Abstract

Pathogenic SRY-box transcription factor 2 (SOX2) variants typically cause severe ocular defects within a SOX2 disorder spectrum that includes hypogonadotropic hypogonadism. We examined exome-sequencing data from a large, well-phenotyped cohort of patients with idiopathic hypogonadotropic hypogonadism (IHH) for pathogenic SOX2 variants to investigate the underlying pathogenic SOX2 spectrum and its associated phenotypes. We identified 8 IHH individuals harboring heterozygous pathogenic SOX2 variants with variable ocular phenotypes. These variant proteins were tested in vitro to determine whether a causal relationship between IHH and SOX2 exists. We found that Sox2 was highly expressed in the hypothalamus of adult mice and colocalized with kisspeptin 1 (KISS1) expression in the anteroventral periventricular nucleus of adult female mice. In vitro, shRNA suppression of mouse SOX2 protein in Kiss-expressing cell lines increased the levels of human kisspeptin luciferase (hKiss-luc) transcription, while SOX2 overexpression repressed hKiss-luc transcription. Further, 4 of the identified SOX2 variants prevented this SOX2-mediated repression of hKiss-luc. Together, these data suggest that pathogenic SOX2 variants contribute to both anosmic and normosmic forms of IHH, attesting to hypothalamic defects in the SOX2 disorder spectrum. Our study describes potentially novel mechanisms contributing to SOX2-related disease and highlights the necessity of SOX2 screening in IHH genetic evaluation irrespective of associated ocular defects.

Keywords: Endocrinology; Neuroendocrine regulation; Neuroscience.

Figure 1. Family pedigrees of probands with SOX2 variants identified in the Massachusetts General Hospital IHH cohort.

Probands are identified by arrows; + indicates wild-type (WT) allele; and V indicates the variant allele. FGFR1, FGF receptor 1; KS, Kallmann syndrome; nIHH, normosmic IHH.

Figure 2. SOX2 is expressed in AVPV but not ARC kisspeptin neurons in adult mouse hypothalamus.

(A–C) Kiss-tdTomato (red), α-SOX2 (green), and DAPI (blue). White arrowheads identify cells in which Kiss-tdTomato and SOX2 are colocalized. Sections are from female AVPV, female ARC, and male ARC Kiss-tdTomato reporter mice, respectively. (D) Quantification of the percentage of SOX2 colocalized with Kiss-tdTomato. N = 3 mice. A minimum of 150 kisspeptin neurons were quantified per replicate. (E and F) Schematics of the location of kisspeptin neurons in the AVPV and ARC of mice, respectively. Kisspeptin neurons were quantified throughout both regions of the hypothalamus. Schematics depict relative position of representative images in A–C with regions highlighted in red. Original image taken at 20× magnification. Scale bar = 10 μm. Data represent mean ± SEM. Data were analyzed using 1-way ANOVA with Tukey’s multiple-comparison post hoc test. Significance indicated by ***P < 0.001.

Figure 3. SOX2 represses KISS1 transcription in immortalized kisspeptin cell lines via direct DNA binding.

(A) Relative mRNA levels of Sox2 in NIH 3T3, KTaR-1, and KTaV-3 cells using real-time quantitative reverse transcription. N = 6–7. hKiss-luc was cotransfected with Sox2 shRNA or a scrambled sequence in an shRNA vector into (B) KTaR-1 or (C) KTaV-3 cells. Luciferase expression in each condition was normalized to the scrambled vector. N = 4–6. (D) Biotin-labeled 30 bp double-stranded oligonucleotides from the indicated regions of the human KISS1 promoter were used in DNA pulldown of protein from NIH 3T3 cells transfected with a hSOX2 expression vector with an HA tag. Following precipitation of DNA/protein complexes with streptavidin magnetic beads, proteins were eluted and analyzed by Western blot with an α-HA antibody. The consensus oligonucleotide contains a 5× multimer of the full SOX2 binding sequence. The scrambled oligonucleotide contains a 5× multimer of a sequence unrelated to SOX2 binding. All conditions were run on same blot and with same exposure. The consensus and scramble lanes were moved to the left side for clarity with the schematic. Representative image from N = 3 biological replicates. Data were analyzed using Student’s t test or 1-way ANOVA with Dunnett’s multiple-comparison post hoc test. Significance indicated by *P < 0.05, ****P < 0.0001.

Figure 4. Mutations in SOX2 interfere with SOX2-mediated repression of KISS1 transcription.

(A) Schematic of the SOX2 protein with patient mutation locations indicated. Protein mutation nomenclature follows recommendations by the Human Genome Variation Society. NLS, nuclear localization signal. (B) Western blot of KTaR-1 cells transfected with SOX2-HA WT or a mutation-harboring hSOX2 plasmid. (C) Quantification of Western blot in B. Total volume for each band normalized to remove background and displayed relative the transfection control, GFP. Actin used as a loading control. N = 3. (D and E) hKiss-luc was cotransfected with SOX2, a mutation-harboring variant of SOX2, or an empty vector (EV) into KTaR-1 or KTaV-3 cells, respectively. SOX2 protein represses expression from hKiss-luc by 79% in KTaR-1 and 84% in KTaV-3 cells. Mutant SOX2 proteins that did not repress the expression from hKiss-luc are indicated in purple in A, D, and E. Values were normalized relative to EV. N = 3–9. Data represent mean ± SEM. Data were analyzed using 1-way ANOVA with Dunnett’s multiple-comparison post hoc test if ANOVA was significant. Significance indicated by *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 5. Two missense mutations in SOX2 prevent proper nuclear localization.

(A–E) KTaV-3 cells were transfected with SOX2-HA WT or an hSOX2 plasmid harboring 1 of the mutations. The 4 mutations that displayed a phenotype in Figure 4, B and C, were studied. Immunostaining was completed using α-HA to identify human SOX2, α-GFP to locate transfected cells and identify the location in cytoplasm of the cell, and DAPI to identify the location of the nucleus. Representative image from N = 3 biological replicates. Images were taken at 40× original magnification.

Figure 6. Two truncating mutations in SOX2 act in a dominant-negative fashion.

(A) Biotin-labeled SOX2 consensus oligonucleotides were used in DNA pulldown of protein from NIH 3T3 cells transfected with hSOX2 or 1 of the mutation-harboring hSOX2 plasmids. Following precipitation of DNA/protein complexes with streptavidin magnetic beads, proteins were eluted and analyzed by Western blot with an α-HA antibody. 10% input was included as a transfection control. (B) Quantification of DNA precipitation in A. Fold-change was normalized to 10% input for each protein. N = 3. (C) hKiss-luc was cotransfected with SOX2 and a variant of SOX2 harboring 1 of the mutations in apportioned amounts from 50 ng to 200 ng. Every well also included 200 ng of WT SOX2. EV was used to compensate for differences in the amount of SOX2 harboring the mutation. Results are displayed as fold-change relative to SOX2 WT–only condition. Data represent mean ± SEM. Data were analyzed using 1-way ANOVA with Dunnett’s multiple-comparison post hoc test. Significance indicated by *P < 0.05.