Regression model for predicting pathogenic properties of SOD1 mutants based on the analysis of conformational stability and conservation of hydrogen bonds.

Intracellular aggregation of proteins is thought to be involved in the aetiology of various neurodegenerative diseases. In particular, mutations in the SOD1 gene are linked to the familial form of amyotrophic lateral sclerosis (ALS). Recently, we developed a regression model for estimating the survival time of ALS patients carrying mutations in SOD1. This model was built based on an analysis of the stability of hydrogen bonds formed in SOD1 mutant proteins during a molecular dynamics (MD) simulation. In the present paper, the regression model was improved by taking into account a new hydrogen-bond property that reflects the conservation measure of a hydrogen bond in the space of protein conformational states. Conformational conservation of hydrogen bonds, being obtained with elastic network (EN) models, allowed us to find eight hydrogen bonds that might affect the pathogenic SOD1 mutants' properties in addition to the bonds that were found via MD in our previous work. The correlation coefficient between survival time of patients with ALS-linked mutations in SOD1 predicted within the improved model and that observed in the literature was 0.91. SOD1 amino acid residues forming these pathogenic hydrogen bonds are found in zinc-binding and electrostatic loops as well as at zinc-binding sites and are in contact with SOD1 aggregates, which implies that these regions are sensitive to perturbations from pathogenic mutations.