Mutational analysis of the complex of human RNase inhibitor and human eosinophil-derived neurotoxin (RNase 2)

RNase inhibitor (RI) binds diverse proteins in the pancreatic RNase superfamily with extremely high avidity. Previous studies showed that tight binding of RNase A and angiogenin (Ang) is achieved primarily through interactions of hot spot residues in the 434-460 C-terminal segment of RI with the enzymatic active site; Asp435 of RI forms key hydrogen bonds with the catalytic lysine in both complexes, whereas the other contacts are largely distinctive. Here we have investigated the structural basis for recognition of a third ligand, eosinophil-derived neurotoxin (EDN), by single-site and multisite mutagenesis. Surprisingly, Ala replacement of Asp435 decreases affinity for EDN only by 14-fold, as compared to the several hundred-fold decreases with RNase A and Ang, and individual mutations of three other hot spot residues-Tyr434, Tyr437, and Ser460-have essentially no effect. Ala substitutions of nine additional residues, selected by examining a computational model of the RI.EDN complex, also have no marked impact. Overall, the losses in affinity for the single-residue variants examined account for only approximately 25% of the free energy of binding for the complex. However, multisite mutagenesis of RI reveals strong superadditivity of mutational effects, indicating that part of this shortfall reflects negative cooperativity. Replacement of Tyr434 together with Asp435 or Tyr437 increases K(i) by 540- and 290-fold, respectively. Thus, the C-terminal region of RI again plays an important role in ligand recognition, although probably smaller than for binding RNase A and Ang. Simultaneous substitutions of three neighboring tryptophans (261, 263, and 318) on RI attenuate affinity even more dramatically (by 4900-fold), indicating that the interactions of this RI region also contribute a considerable amount of the binding energy for the EDN complex. These findings highlight the potential importance of cooperativity in protein-protein interactions and the consequent limitations of single-site mutagenesis for assessing interface energetics.