Deciphering the molecular impact of ALDH5A1 missense variants in succinic semialdehyde dehydrogenase deficiency through combined in vitro and in silico approaches

Succinic semialdehyde dehydrogenase is a key enzyme involved in the catabolism of g-aminobutyric acid. Biallelic variants in the ALDH5A1 gene cause a rare monogenic neurometabolic disorder, SSADH deficiency, characterized by a total or partial enzyme loss-of-function. Patients present symptoms from infancy on and mainly show developmental delay and intellectual disabilities, with seizures and/or autistic spectrum disorders as the main distinctive traits associated with the severe forms of the disorder. A valuable instrument to interpret variant function is provided by bioinformatic tools that are based on evolutionary and structural data. However, the results need to be complemented by other experimental approaches to dissect the molecular basis for the defect and be supportive of the clinical phenotype of patients. Here, we validate a bioinformatic prediction of loss-of-function using two recombinant enzyme variants and/or transient transfection in a ALDH5A1 knockout HEK-293T cell model. These SSADH protein variants are synthesized by two functionally hemizygous SSADH deficiency patients. The known loss-of-function p.Gly176Arg variant was predicted by bioinformatic analyses to affect oligomerization. Indeed, we demonstrate that the variant acquires a tetrameric altered structure, possibly responsible for the loss-of-function. The pathogenicity of the new p.Pro234Ser variant is mainly due to its failure to reach the mature form, possibly due to its intrinsic instability rather than a defect in the maturation apparatus, thus preventing a correct folding process. Altogether, we provide evidence that combining recombinant protein studies with structural predictions can offer valuable insights into loss-of-function, which is essential for precision therapy.