Glycoengineering of CHO cells for production of recombinant therapeutics with enhanced efficacy

Conference Dates

May 6-11, 2018


Glycosylation can significantly affect the efficacy of recombinant therapeutics. Glycoprotein drugs, such as EPO, require a high degree of sialylation on their N-glycans in order to have a longer circulatory half-life. Mannose-terminated N-glycans can target the protein to dendritic cells and macrophages cells via cell surface mannose-binding receptors. Removal of core fucose from human IgG1 antibodies has been shown to significantly enhance its affinity to FcRIIIa and thereby dramatically improves its antibody-dependent cellular cytotoxicity (ADCC). Cancer cells generally express glycoproteins with shortened O-glycans. Therefore, recombinant anti-cancer vaccines carrying these short tumor-associated O-glycans are more ideal for triggering specific anti-tumor immune responses. With cytotoxic lectins and genome editing tools, namely ZFNs, TALENs and CRISPR-Cas9, we have so far created 29 CHO glycosylation mutant cell lines (CHO-gmt1 to CHO-gmt29; gmt = glycosylation mutant). In some of these mutant lines, only one gene has been inactivated (CHO-gmt1, 2, 3 4 cells), whereas in others more than 10 glycosylation genes have been knocked out in order to obtain a particular glycan structure. With these mutants, we have been able to produce EPO with highly sialylated N-glycans for extending in vivo half-life. We have developed stable cell lines to produce recombinant human β-glucocerebrosidase with mannose-terminated N-glycans (mainly Man5). This product is similar to Cerezyme, but the in vitro glycan modification is not needed. We have also developed stable cell lines to produce fucose-free rituximab and GA101 with high titers. In a cell-based ADCC assay, these fucose-free rituximab and GA101 outperformed their commercial counterparts, Rituxan and Gazyva, respectively. Furthermore, rituximab produced by more than a dozen of these CHO-gmts were purified and the glycans attached to the antibody were analyzed by MALDI-TOF MS and LC-MS. The results showed that many of these mutants were able to produce the antibody with one major N-glycan which represents 90-97% of the total N-glycans attached to the antibody. These antibodies are invaluable tools for studying the impact of glycans on antibody functions including PK/PD characteristics. Rituximab produced by these CHO-gmt cells have been compared in a cell-based ADCC assay. Their binding affinities for FcRII, FcRIIIa and the neonatal Fc receptor (FcRn) were also determined. As expected, fucose-free glycans bind FcRIIIa much tighter than others, and are most effective in activating ADCC. While most glycans did not affect the binding to FcRII and FcRn, two glycans significantly enhanced the binding for FcRII and FcRn. The in vivo cell killing ability and circulating half-life for the rituximab antibodies that carry different N-glycans are being investigated in mouse models. Taken together, we have generated a large panel of CHO glycosylation mutants. Each of these mutants produces a unique set of N- and O-glycans. They can be used to study the impacts of glycans on therapeutic efficacies of biologics. Our study on the rituximab carrying different homogeneous N-glycans represents a first systematic investigation on the impact of glycans on antibody functions.

This document is currently not available here.