ESPE Abstracts

Background: Androgens are steroid hormones necessary for human sex development. Testosterone (T) and the more potent dihydrotestosterone (DHT) are maybe the best known androgens, which exert their effect by binding and activating the androgen receptor. Steroid reductases 5α (SRD5As) catalyse the conversion of T to DHT in the classic androgen production pathway, or from 17-hydroxyprogesterone to 17OH-dihydroprogesterone, and androstenedione to androstanedione in alternate pathways leading to DHT. There are two enzymes with differential expression, of which SRD5A2 is expressed in reproductive organs and liver, and catalyses the reaction of T to DHT more efficiently than SRD5A1. Human SRD5A2 loss-of-function mutations are known, and cause severe 46XY undervirilization, while gain-of-function variants have been suggested in androgen excess syndromes such as premature adrenarche, the polycystic ovary syndrome or prostate tumors, but they have not been found so far.

Aim: Therefore, we aimed to search for gain-of-function mutations in the human SRD5A2 gene.

Methods: For that, we searched databases for candidate variants and performed bioinformatic and functional tests on selected variants. After conservation analysis of SRD5A2, a novel 3D protein model was constructed to locate the exact position of amino acids in the tertiary structure and predict their effect on protein function and substrate interaction. We then collected 116 coding SNPs in SRD5A2 from OMIM, dbSNP, Pubmed, Clinvar, HGMD and Uniprot databases. These SNPs were ranked according to their association with phenotypes, physical location in our 3D model, and molecular dynamics simulation studies. Finally, we selected 9 coding SNPs for in vitro studies. These SNPs were located within or close to highly conserved areas that form the binding cavities for substrates or cofactor NADPH. SRD5A2 variants were expressed in HEK293 cells and activity to convert T to DHT was assessed and compared to wild-type.

Result: Variants R50A and P173S decreased enzymatic activity, while variants A49T, P106L, P106A, N122A, L167S, R168C and R227Q significantly reduced activity. As predicted in our in silico analysis, all coding SNPs affected enzyme activity in vitro, however none of them showed gain-of-function.

Conclusion: In conclusion, we provide a novel protein model for studies of SRD5A2. No gain-of-function variants were identified, but we have characterized 9 human SRD5A2 variants, which might be of clinical relevance for their enzyme activity loss. It is possible that individuals carrying these SNPs show a minor phenotype that is not yet identified. Alternatively, SRD5A1 may compensate? Genotype-phenotype studies would be able to solve this question.