(307a) Untargeted Proteomic Study of Disulfide Bond Reduction during Monoclonal Antibody Production in Chinese Hamster Ovary (CHO) Cell (Faculty/Industry Candidate)

Authors: 
Park, S. Y. - Presenter, University of Massachusetts
Lee, K., Tufts University
Aron, K. L., Bristol Myers Squibb
Xu, X., Bristol-Myers Squibb
Egan, S., Bristol-Myers Squibb
Borys, M., Bristol-Myers Squibb
Ghose, S., Bristol-Myers Squibb
Cura, A., Bristol Myers Squibb
Depending on process conditions, monoclonal antibodies (mAbs) produced by Chinese hamster ovary (CHO) cells can undergo partial disulfide bond reduction, which may result in the formation of low molecular weight (LMW) fragments. These partially reduced fragments lack the stability and biological functions of the intact protein. Thus, formation of LMW fragments negatively affects product quality and process productivity. The reduction of disulfide bonds is thought to occur due to host cell proteins (HCPs) that accumulate in the cells. However, mechanistic understanding of this phenomenon remains limited. It is unclear which host cellular pathways contribute to disulfide reduction, and how variations in process conditions affect these pathways.

In this study, we used an untargeted proteomics approach to systematically characterize changes in HCPs under conditions that lead to formation of LMW fragments, and to identify pathways that correlate with markers of disulfide reduction. Cell samples were collected during early stationary phase (days 7, 8, and 9) from fed-batch bioreactors (n=2 per condition) operated under reducing and non-reducing conditions. Antibodies recovered from the non-reducing condition contained no LMW fragments, whereas antibodies from the reducing condition were LMW fragments. These samples were analyzed by liquid chromatography-mass-spectrometry (LC-MS) perform on a quadrupole time-of-flight (TOF) system. Principal component analysis of approximately 3,100 unique monoisotopic peptides showed clear separation between reducing and non-reducing conditions. Annotation of the peptides followed by analysis of covariance (ANCOVA) identified 85 proteins that trend significantly differently between reducing and non-reducing conditions (p<0.05). These significant proteins were clustered based on protein-protein interactions using STRING. Significant clusters were associated with carbon metabolism, protein folding, heat shock response, and protein processing in the endoplasmic reticulum (ER). These results are consistent with Fisher’s exact test performed on the differentially expressed proteins, which also identified protein folding (p<10-5) as the most significantly enriched pathway. Targeted analysis of selected proteins in the oxidative stress response and ER protein processing pathways, including the thioredoxin system, further corroborated that these pathways were elevated under the reducing condition. Taken together, our findings suggest that LMW fragments could form due to disulfide reduction associated with ER oxidative stress. It is possible that the ER stress initially results from accumulation of misfolded proteins, which could further amplify ER stress and additionally cause mitochondrial oxidative stress. Further studies are warranted to investigate, at the molecular level, these possible explanations.

In conclusion, global profiling of HCPs under reducing and non-reducing conditions, followed by multivariate analysis of annotated proteins identified key pathways that are differentially regulated and correlate with disulfide reduction. Prospectively, these proteins could be monitored during process development and scale-up to control product quality.