(7c) Genome-Wide Knockout Screens Reveal New Engineering Targets That Enhance Protein Secretion in a Non-Model Yeast

The yeast Komagataella phaffii (Pichia pastoris) is commonly used for the production of recombinant proteins due to its rapid growth to high cell densities and its ability to secrete proteins into the culture medium. Large-scale clinical manufacturers routinely achieve industrially relevant titers for protein therapeutics like insulin and subunit vaccines including for the COVID-19 pandemic. There is growing interest in K. phaffii as an alternative host for the manufacturing of more complex, high-value protein therapeutics like monoclonal antibodies (mAbs), including several recent product approvals. Early engineering efforts yielded humanized yeast strains capable of secreting mAbs with specific quality attributes and post-translational modifications. Secreted titers must be increased, however, to realize cost-saving or scalability advantages over current hosts like CHO cells.

While integration of exogenous genes can confer new cellular functions and post-translational modifications, engineering of the endogenous host-cell genome may expand overall secretory capacity through deletion of unneeded proteins and pathways or expansion of secretory organelles. In a non-model host like K. phaffii, only 50-70% of the genome is functionally annotated, limiting the extent of rational genomic engineering. Tools for unbiased functional genomics are needed to discover new engineering targets that may improve protein secretion.

Here, we present the application of two CRISPR-Cas9 knockout libraries for the discovery of new engineering targets in K. phaffii. First, we targeted all endogenous secreted host cell proteins and identified non-essential genes in the yeast secretory pathway. Deletion of ~10 non-essential HCPs yielded strains with increased secreted titer and purity of several recombinant proteins. Second, we created a pooled genome-wide knockout library. We employed a novel flow-cytometric assay to identify library members with enhanced secretion of a mAb. Subsequent validation of discovered knockouts yielded new targets for strain engineering. These unbiased approaches to serve as powerful demonstrations for the genomic engineering of poorly characterized manufacturing hosts for complex phenotypes.