Structural and Mutational Analysis Identifying Binding "Hot Spots" in Protein-Protein Interactions. A. A. Kossiakoff, A. deVos and J. A. Wells, Department of Protein Engineering, Genentech, Inc. 460 Point San Bruno Blvd. South San Francisco, CA 94080

The binding surfaces between the endocrine family of hormones and receptors are substantial, significantly larger than those associated with antibody-antigen interfaces. With these extensive contact surfaces, how can molecules show both binding versitility and finely honed specificity required for biological regulation? Towards understanding these issues at the molecular level, we have studied complexes in the endocrine family of hormone-receptors by complementary crystallographic, mutagenesis and biochemical analyses. Ala-scanning mutagenesis analysis involving each contact residue of both human growth hormone (hGH) and its receptor (hGHR) has shown that the binding and the specificity determinants are generally different and involve a relatively small subset of the residues in the interface. Of the approximately 30 residues of hGH that directly contact the receptor, about 85% of the total binding energy is derived through the effects of just seven of them. It was also determined that hormone-receptor binding affinity is an "off-rate" driven phenomenon. In a related experiment, a variant of hGH that binds 400 times tighter to its receptor was produced by a technique called phage display. This method allowed for a high throughput assay of several million mutants from randomly mutating 15 contact sites on hGH. The resulting binding selections and structure analysis gave two important and unexpected findings: 1) several hGH side chains that had participated in H-bonding interactions in the complex were replaced by hydrophobic groups in the selection and 2) the structure of the mutant complex to the receptor showed that structural elements of both the mutant hormone and the receptor had changed conformation. This shows that phage display mutagenesis not only probes sequence diversity but, in some types of tertiary structures, conformational diversity as well. Thus, this convolution of these two parameters makes the actual diversity available much larger than the computed 3x106 that was based on amino acid composition alone.