Biochemical Characterization of Disease-Associated Variants of Human Ornithine Transcarbamylase

Human ornithine transcarbamylase deficiency (OTCD) is the most common ureagenesis disorder in the world. OTCD is an X-linked genetic deficiency in which patients experience hyperammonemia to varying degrees depending on the severity of the genetic mutation. More than two-thirds of the known mutations are caused by single nucleotide substitutions. In this paper, partial order optimum likelihood (POOL), a machine learning method, is used to analyze single nucleotide substitutions in OTC with varying disease phenotypes and predicted catalytic efficiencies. Specifically, we used a computed metric, μ₄, a measure of the degree of coupling between an ionizable residue and its neighbors, calculated for the catalytic residues, to identify which protein variants were most likely to have impacted catalytic activities. From this analysis, 17 disease-associated variants were selected plus one additional variant, representing a range of μ₄ values and POOL ranks. Then μ₄ predictions were compared with established bioinformatics tools, SIFT, PolyPhen-2, Provean, FATHMM, MutPred2, and MutationTaster2. The bioinformatics tools predicted that most of these mutations are deleterious. The variants were biochemically characterized using kinetics assays, size exclusion chromatography, and differential scanning fluorimetry. POOL combined with μ₄ analysis was able to predict correctly which variants were catalytically hindered in vitro for 17 out of 18 variants. Then by expressing a subset of these proteins in cell culture, mechanisms for disease were proposed. Analysis using μ₄ is a complementary method to the sequence-based bioinformatics tools for predicting the effects of mutation on catalytic function.