(49f) Engineered Hepatocyte Growth Factor Fragments Function as Met Receptor Antagonists by Inhibiting Ligand-Induced Dimerization

Jones, D. S. - Presenter, Stanford University
Cochran, J. R. - Presenter, Stanford University

The Met tyrosine kinase receptor is a critical cancer target that is overexpressed on many solid tumors, and molecules that inhibit Met activation hold significant promise for cancer therapy. The natural Met receptor ligand, the hepatocyte growth factor (HGF), is composed of multiple domains that interact with and activate the receptor through heparin-mediated homodimerization. The minimal subunit of HGF that can activate Met is composed of its N-terminal domain and first Kringle domain (termed NK1). To create a Met receptor antagonist for cancer therapy, we adopted a combined approach of combinatorial and rational engineering to transform the NK1 fragment of HGF from a weak agonist into a potent Met antagonist. First, using directed evolution through yeast surface display, we evolved mutants of the NK1 fragments with marked improvements in stability and Met binding affinity. Engineered NK1 mutants exhibited significant improvements in binding affinity (Kd= 1.7 ? 3.3 nM) and high thermal stability (Tm~60-70 °C) compared to wild-type NK1 (50 ? 150 nM; Tm=51 °C). Next, we rationally introduced point mutations into NK1 to transform the resulting mutants into antagonists by disrupting the ligand dimerization interface that is highlighted in the NK1 crystal structure. Intriguingly, the NK1 mutants isolated directly from the directed evolution studies functioned as Met antagonists in cell-based assays without the need for further rationally-designed modifications. To determine the molecular mechanism for this observation, we explored the oligomeric state of the NK1 mutants, since wild-type NK1 induces dimerization and activation of the Met receptor through a heparin-induced NK1 homodimer. Analytical size exclusion chromatography experiments revealed that wild-type NK1 homodimerized in the presence of heparin, as previously reported in the literature. In contrast, both the parental mutants and those harboring the rationally-designed point mutations remained monomeric in the presence of heparin, even though they still retained the ability to bind to heparin. This lack of heparin-induced ligand dimerization explains why our engineered NK1 mutants do not induce Met receptor dimerization and activation, and thus function as competitive HGF antagonists. Although rational engineering was not required for this particular system, combinatorial and rational engineering approaches for transforming a natural receptor agonist into a potent receptor antagonist is a useful complement to the traditional antibody engineering approach for development of therapeutic agents.

This project was supported by NIH NCI R21 CA131706.