Tuning Phage Display and Yeast Surface Display for Panning for Novel Ligands on Tumor Cells

Tuning Phage Display and Yeast Surface Display for Panning for Novel Ligands on Tumor Cells

Lawrence A. Stern, Ian A. Schrack, Aakash Deshpande, Sadie M. Johnson, and Benjamin J. Hackel

Yeast surface display and phage display have proven to be effective tools in the evolution and screening of ligands with novel binding activity.  Generally, selections for binding activity are carried out using immobilized recombinant domains of target molecules.  While this method often provides ligands with apparent high affinity and specificity for the target of interest, translatability to true cellular target in vitro is a common concern.  Direct selections on mammalian cell surfaces can be carried out either exclusively or in combination with soluble target-based selections to further direct toward evolution of ligands with affinity for true cellular target.  Further, this method can be applied to simultaneous discovery of molecular targets and new binding ligands using poorly characterized cell types.  This study aims to optimize direct cellular selections using multiple formats of phage display and yeast surface display by using an epidermal growth factor receptor (EGFR) model system that encompasses multiple tiers of affinity and cellular target expression, and to directly compare these methods for applicability to binder selection from mock libraries.  Cell-based selections with standard yeast surface display can be successful with up to mid-nanomolar affinity, yielding 15-30% of inputted binders with enrichment ratios reaching 96 ± 84.  However, this level of enrichment requires at least one million targets per cell, with yields and enrichment ratios decreasing dramatically for a 10-fold decrease in targets per cell.  One hypothesis for this decrease in enrichment is a lack of accessibility of relevant epitopes due to steric hindrance by glycosylation and other membrane-bound proteins on cell surfaces. The impacts of displayed protein linker length, receptor epitope, and receptor identity will be discussed. Mathematical modeling of phage display selections highlights the importance of (a) binding strength, which is being evaluated with various intrinsic affinities and four modes of valency – monovalent pIII, bivalent pIII using a previously validated antibody hinge - leucine zipper fusion approach (Lee, et al., J. Immun. Meth. 2004), avid pIII from phage vector, and multivalent pVIII via pVIII engineering – and (b) fraction of phage displaying ligand, which is being evaluated using the aforementioned display approaches. Western blotting techniques are used to quantify displaying fraction. Mock library selections are used to quantify enrichment. Optimal conditions, and their consistency with mathematical modeling, will be discussed as well as ongoing efforts in target-ligand co-discovery.