Modeling effects of human single nucleotide polymorphisms on protein-protein interactions

Abstract

A large set of three-dimensional structures of 264 protein-protein complexes with known nonsynonymous single nucleotide polymorphisms (nsSNPs) at the interface was built using homology-based methods. The nsSNPs were mapped on the proteins' structures and their effect on the binding energy was investigated with CHARMM force field and continuum electrostatic calculations. Two sets of nsSNPs were studied: disease annotated Online Mendelian Inheritance in Man (OMIM) and nonannotated (non-OMIM). It was demonstrated that OMIM nsSNPs tend to destabilize the electrostatic component of the binding energy, in contrast with the effect of non-OMIM nsSNPs. In addition, it was shown that the change of the binding energy upon amino acid substitutions is not related to the conservation of the net charge, hydrophobicity, or hydrogen bond network at the interface. The results indicate that, generally, the effect of nsSNPs on protein-protein interactions cannot be predicted from amino acids' physico-chemical properties alone, since in many cases a substitution of a particular residue with another amino acid having completely different polarity or hydrophobicity had little effect on the binding energy. Analysis of sequence conservation showed that nsSNP at highly conserved positions resulted in a large variance of the binding energy changes. In contrast, amino acid substitutions corresponding to nsSNPs at nonconserved positions, on average, were not found to have a large effect on binding affinity. pKa calculations were performed and showed that amino acid substitutions could change the wild-type proton uptake/release and thus resulting in different pH-dependence of the binding energy.

Figure 1

Distribution of ΔΔΔG_tot(nsSNP) and ΔΔΔG_el(nsSNP) in kcal/mol for OMIM and non-OMIM cases. Solid bars, OMIM; open bars, non-OMIM.

Figure 2

Illustration of nsSNPs at interface of protein-protein complexes: (a) TTR (transthyretin, gene ID: 4507725), red, A chain; blue, E chain; green, Ser in A85; yellow, F in A85; magenta, N in E63. (b) DYNLRB1 (Roadblock-1, gene ID: 7661822), red, A chain of target; light red, A chain of SNP variants; blue, B chain of target; sky blue, B chain of SNP variant; green, K in A75; yellow, E in A75; magenta, D in B61 of target; pink, D in B61 of SNP variant. (c) HBB (β-globin, gene ID: 4504349), red, B chain; blue, C chain; green, V in B34; yellow, L in B34. (d) GSTM2 (glutathione S-transferase M2, gene ID: 4504175), red, A chain; blue, B chain; green, M in A130; yellow, K in A130; magenta, M in B50.

Figure 3

MSA. Blank frame is nsSNP position. (a) HBB (β-globin, gene ID: 4504349); (b) GSTM2 (glutathione S-transferase M2, gene ID: 4504175).

Figure 4

Change of the binding energy in kcal/mol as a function of the amino acid conservation (SI%). The broken lines are guides for the eye and follow the maximal amplitude of binding energy change. (a) ΔΔΔG_tot(nsSNP); (b) ΔΔΔG_el(nsSNP).

Figure 5

Change of the binding energy in kcal/mol as a function of calculated proton uptake/release (absolute value of ΔΔq). (a) ΔΔΔG_tot(nsSNP); (b) ΔΔΔG_el(nsSNP).