The role of water in computational and experimental derivation of binding thermodynamics in SH2 domains

We have studied the role of bound interface water molecules on the prediction of the thermodynamics of SH2 domain binding to tyrosyl phosphopeptides using a method based on accessible surface area buried upon association. We studied three phosphopeptide ligands, which have been shown by Lubman and Waksman (J Mol Biol;328:655, 2003) and Davidson et al. (JACS;124:205, 2002) to have similar binding free energies but very different thermodynamic signatures. The thermodynamic model is semiempirical and applies to the crystal structure of the SH2 domain-bound forms. We explored all possible combinations of bound interfacial waters. We show that the model does not predict the binding thermodynamics of either ligand. However, we identified the empirical formula describing the heat capacity change as the source of the problem. Indeed, systematic exploration of heat capacity change values between 0 and -300 cal/mol deg results in a sharp distribution of the number of ligand/SH2/water-subset structures that provide binding thermodynamics similar to experimental values. The heat capacity change values at which the distributions peak are different for each peptide. This prompted us to experimentally determine the heat capacity change for each of the peptides and we found them to coincide with the values of the peaks. The implications of such findings are discussed.