Rational design of glycoengineered interferon-β analogs with improved aggregation state: experimental validation

Recombinant human interferon-β (rhIFN-β) used clinically has lower efficacy than expected due to protein instabilities such as aggregation. Increasing molecular stability, glycoengineering has been used to improve clinical efficacy for a number of therapeutics; however, often labor-intensive trail-and-error approaches are used to identify additional glycosylation sites. In this study two rhIFN-β analogs with one additional glycosylation site, L6T and S75N, identified by a rational in silico approach, were characterized. These rhIFN-β analogs were synthesized in parallel with a Chinese hamster ovary (CHO) codon-optimized natural human IFN-β (Opt-IFN-β) and expressed in CHO cells using the same expression system. The molecular weights for both analogs were observed to be higher than Opt-IFN-β, consistent with hyper-glycosylation. The in vitro biological assay showed the hyper-glycosylated analogs and the Opt-IFN-β had similar activity. The aggregation studies demonstrated that both analogs had lower tendencies to aggregate compared to the Opt-IFN-β. These experimental studies validate the in silico strategy to predict suitable glycosylation sites that would be glycosylated, while maintaining biological function. Moreover, this work describes hyper-glycosylated rhIFN-β analogs with improved solubility (i.e. lower aggregation). These findings, together with the rational in silico design, will allow us to increase protein glycosylation with the goal to enhance therapeutic efficacy.