Comparison of In Vitro and In Silico Assessments of Human Galactose-1-Phosphate Uridylyltransferase Coding Variants.

Introduction Human pathogenic coding variations of the galactose-1-phosphate uridylyltransferase (GALT) gene cause classic galactosemia, a recessive disease of galactose metabolism. Unfortunately, there are many variants of uncertain significance (VUS) that need to be characterized in order to be able to predict the likelihood of classic galactosemia for all possible genotypes. There are many bioinformatic resources available that attempt to predict the pathogenicity of a human variant, but it is unclear if these methods realistically predict the consequence of these variants. To determine the clinical application of these resources, we compared the results of enzymatic assays with predictive models. Methods In all assays, we compared the activity of the three human GALT VUS (Alanine81Threonine, Histidine47Aspartate, Glutamate58Lysine) to native GALT (nGALT) and to a variant of known pathogenic clinical significance (Glutamine188Arginine). The enzymatic activities of VUS recombinant proteins were compared to the results of analytical methods. The methods included the comparison of molecular dynamic simulation root-mean-square deviation (RMSD) results and the results from predictive programs PredictSNP, evolutionary model of variant effect (EVE), ConSurf, and sorting intolerant from tolerant (SIFT). Results The enzymatic assays showed that the variants tested had diminished Vmax relative to the native protein. The VUS RMSD data for both the whole protein and individual residues in the molecular dynamics simulations were not significantly different when compared to nGALT. The other predictive programs had mixed results for each VUS and were not consistent with the enzyme activity or simulation results. Conclusions Our experiments indicated a statistically significant decrease in enzymatic activity of the VUS when compared to nGALT. These experiments also demonstrated significant differences between predictions and results. These results suggest that the in silico tools used may not be beneficial in determining the pathogenicity of GALT VUS.