Predictive mechanistic model for separation of monoclonal antibody, fab fragment, and aggregate species on multimodal chromatography.

Abstract

The specific selectivities offered by multimodal ligands drive the increased application of multimodal chromatography in the purification of complex new "multispecific" antibodies, which requires improved understanding of the protein-multimodal ligand interaction mechanism. In the present study, a mechanistic model is developed to predict monoclonal antibody (mAb1)-Fab fragment (Fab) and heterogeneous aggregates separation on Capto™ MMC ImpRes multimodal resin based on the general rate model coupled with the proposed preferential interaction (PI) analysis-based Langmuir non-linear binding model. The model input value of binding parameters is obtained from Perkin et al. developed PI model, fit to the characteristic 'U'-shaped curve for isocratic retention factors of mAb1, Fab, and aggregates as a function of NaCl salt concentrations. The model successfully simulates mAb1 and Fab elution peaks, whereas in the absence of deconvoluted peaks of heterogeneous aggregates, aggregates are modeled as a single species, giving satisfactory prediction of elution peak position, describing the average of the multiple (majority as double peaks) aggregate elution peaks. The physical significance of model estimated binding parameters is obtained from model estimated total number of released counter salt ions and water molecules for each species during binding, found to be consistent with their isocratic retention data. The underlying mechanism of double peak elution of aggregates during linear gradient elution was investigated based on mechanistic model estimated equilibrium constant. The proposed predictive mechanistic model was successfully validated by predicting mAb1, Fab, and aggregates elution peaks for the multimodal column operated in hydrophobic interaction mode and can be successfully implemented for process development.