(69h) Hybrid Soluble/Cellular Target Selection Schemes Improve Discovery of Translatable Ligands

Precision medicine necessitates molecular targeted agents to match the growing diversity of clinical biomarkers. The current standard, high-throughput screening against soluble domain fragments of target molecules, can result in failure when selected ligands are tested for binding to genuine, complete, membrane-bound target. Multiple factors may contribute to this lack of translatability including: 1) improper folding of the soluble domains due to lack of transmembrane domain, 2) differential post-translational modification by the production host, 3) binding to a non-natural epitope resulting from chemical or biological tags, or 4) lack of accessibility of the bound epitope in the cellular context. This work compares five selection strategies to incorporate translatability as a selective pressure in ligand discovery campaigns against CD276 and Thy1, two biomarkers that are expressed in tumor vasculature.

Combinatorial libraries of fibronectin domains and affibodies were displayed on the yeast surface using an extended linker. Binders to CD276 and Thy1 were selected by (a-c) magnetic bead sorting and (a) FACS with soluble extracellular domains, (b) FACS with detergent solubilized cell lysates, or (c) panning on mammalian cell surfaces; or (d,e) direct panning on mammalian cell surfaces with or without depletion of non-specific protein binders. Enriched subpopulations were characterized by clonally assessing target-specific binding in a cell-based phage ELISA or yeast panning assay.

Enrichment of binders to recombinant target followed by direct cell panning on mammalian cell lines yielded the best outcomes as three of four campaigns (all but affibodies against Thy1) exhibited strong selectivity for target-positive cells (ranging from 3:1 to 86:1 selectivity). 51% to 85% of lead molecules were successful hits in a clonal cell assay. Enrichment to recombinant target followed by cell lysate FACS yielded modest selectivity (and a 23% lead hit rate) for CD276-targeted affibodies and Thy1-targeted fibronectins but not the other two scaffold/target combinations. Enrichment only with recombinant target has yielded a 35% lead hit rate in the affibody/CD276 effort. None of the campaigns directly using cell panning yielded ligands with a strong preference for target-expressing cells. Sequence diversity, affinity, stability, and structural characterization of lead molecules will be presented.

In conclusion, selections utilizing soluble target methods followed by direct cell panning yield optimal selectivity for the membrane-bound targets of interest. Several small protein ligands for CD276 and Thy1 have been identified.