(21b) Engineering Pan-Reactive VEGF Antagonists for Neovascular Eye Diseases

Introduction: Neovascularization in the eye drives the pathogenesis of diseases including age-related macular degeneration and diabetic retinopathy, two of the leading causes of blindness in adults. The Vascular endothelial growth factor (VEGF) ligands are major stimulants of neovascularization. Suppression of VEGF ligands binding with their receptors (VEGFRs) have shown clinical benefit in treatment of neovascular eye diseases. Current FDA-approved therapies all primarily target VEGF-A, the most prominent regulator of blood vessel formation, in order to prevent it from binding to VEGFR-2 and subsequently activating signaling. However, VEGF-C is now increasingly implied in pathological angiogenesis, and has been shown to be upregulated following anti-VEGF-A treatment as well as being as potent as VEGF-A in driving retinal neovascularization under hypoxic conditions. No current therapies are bind both VEGF-A and VEGF-C to provide a more complete blockade of neovascularization. We designed an error-prone mutagenic library of the domains 2&3 of VEGFR-2 (VEGFR-2 D23), which binds both VEGF-A and -C, using the yeast surface display platform in order to isolate and characterize high-affinity clones that could bind both VEGF-A and VEGF-C that could act as superior therapeutics for the treatment of neovascular eye diseases.

Materials and Methods: VEGF-A and VEGF-C dimers were produced in a mammalian cell expression system and biotinylated for use as targets in library selections and titrations. We cloned an error-prone mutagenic library of VEGFR-2 into the pCT yeast expression plasmid and transformed the resulting DNA into electrocompetent EBY-100 yeast. We performed five rounds of magnetic- and fluorescence-activated cell sorting (MACS and FACS) against VEGF-A to select for functionally- expressing, high-affinity variants. Variants that bound VEGF-A were then tested for VEGF-C binding. Sequences were extracted from clones that bound both ligands, and the evolved VEGFR-2 D23 mutants were titrated against VEGF-A and VEGF-C in order to determine their affinities for each ligand.

Results and Discussion: VEGFR-2 D23 was found to display on the surface of yeast cells when plasmid DNA encoding the protein sequence with a C-terminal cmyc epitope tag was transformed into competent yeast, but binding was unable to be detected for either VEGF-A or VEGF-C when the biotinylated ligands were incubated with fluorescent streptavidin. An error-prone library with a mutation rate of 1.9 amino acids/variant was cloned and transformed into yeast. We demonstrated the enrichment of the mutagenic library against VEGF-A over 5 rounds of selections (Figure 1), and were able to identify several variants with a range of affinities that were able to bind both VEGF-A and VEGF-C with high affinity.

Conclusions: We have identified and characterized high-affinity VEGFR-2 D23 variants that engage VEGF-A and VEGF-C. These clones are being further characterized and will be the basis for a site-directed library based on the crystal structures of VEGFR-2 interacting with VEGF-A and VEGF-C aimed at further increasing the affinity of the antagonist for both ligands.