Mette Louise Skjødt:
Yeast surface display is an effective tool for antibody affinity maturation because yeast can be used as an all-in-one workhorse to assemble, display and screen diversified antibody libraries. By employing the natural ability of yeast Saccharomyces cerevisiae to efficiently recombine multiple DNA fragments by in vivo homologous recombination large combinatorial antibody libraries can easily be generated. We have optimized ordered assembly of three CDR fragments into a gapped vector and observed increased transformation efficiency in a yeast strain carrying a deletion of the SGS1 helicase-encoding gene. The SGS1 deletion was combined with deletions of the three protease-encoding genes PEP4, PRB1 and YPS1 to increase stability of the displayed antibody molecules. A display system based on cell surface anchoring via the cell wall protein Flo1 was developed and using a model antibody we analyzed surface expression of various antibody formats in the generated knockout strain. Functional scFv and scFab fragments were efficiently displayed on yeast whereas impaired chain assembly and heavy chain degradation was observed for display of full-length IgG molecules. To identify the optimal polypeptide linker for yeast surface display of scFv and scFab fragments, we compared a series of different Gly-Ser-based linkers in display and antigen binding proficiency. We show that these formats of the model antibody can accommodate linkers of different lengths and that introduction of alanine or glutamate residues into the linker sequences does not affect display or antigen binding. For affinity maturation, libraries of diversified antibody CDRs were generated by random mutagenesis and assembled into scFv and scFab formats by in vivo homologous recombination. From these libraries we successfully isolated variants with improved antigen binding affinities of up to 4.5-fold. Sequences analysis of the isolated variants revealed multiple overlapping mutations between the scFv and scFab libraries and prevailing amino acid substitutions at positions previously mapped as affinity maturation hotspots by in silico modeling. The isolated mutations were located to three main regions within the antigen-antibody structure: the central antigen-antibody interface, the apparent interface between the light and heavy chain variable domains and a peripheral antigen interaction region in the distal part of the heavy chain CDR-2. Based on the presented data we suggest that affinity maturation of the model antibody proceeds through multiple incremental steps of subtle improvements. We moreover conclude that the best affinity improved candidates are likely to be obtained through optimization of both the antigen-antibody interface and the antibody intraface.