Improved virus isoelectric point estimation by exclusion of known and predicted genome-binding regions

Accurate prediction of the isoelectric point (pI) of viruses is beneficial for modeling virus behavior in environmental transport and physical/chemical treatment applications. However, the empirically measured pIs of many viruses have thus far defied simple explanation, let alone prediction, based on the ionizable amino acid composition of the virus capsid. Here, we suggest an approach for predicting virus pI by excluding capsid regions that stabilize the virus polynucleotide via electrostatic interactions. This method was applied first to viruses with known polynucleotide-binding regions (PBRs) and/or 3D structures. Then, PBRs were predicted in a group of 32 unique viral capsid proteome sequences via conserved structures and sequence motifs. Removing predicted PBRs resulted in a significantly better fit to empirical pI values. After modification, mean differences between theoretical and empirical pI values were reduced from 2.1 ± 2.4 to 0.1 ± 1.7 pH units. Importance This model is the first to fit predicted pIs to empirical values for a diverse set of viruses. The results suggest that many previously-reported discrepancies between theoretical and empirical virus pIs can be explained by coulombic neutralization of PBRs of the inner capsid. Given the diversity of virus capsid structures, this nonarbitrary, heuristic approach to predicting virus pI offers an effective alternative to a simplistic, one-size-fits-all charge model of the virion. The accurate, structure-based prediction of PBRs of the virus capsid employed here may also be of general interest to structural virologists.