Mediated Electrochemical Probing and Machine Learning for Cysteine and Reduced Monoclonal Antibody Quantification

Biopharmaceutical manufacturing requires robust analytics and process controls throughout production to insure high yield of quality products. New methodologies for rapidly accessing and integrating data-rich information from complex dynamic biological environments are of great interest. We suggest electronic detection of biological redox signatures based on mediated electrochemical probing (MEP) as an innovative, simple, and rapid modality for interrogating these complex systems. We have previously shown that by leveraging the redox properties of interchain-disulfide bonds within antibodies one can interrogate antibody structure, in particular, the occurrence of partially reduced forms that can occur during production processes. In this work, we expanded on this method to decrease sample-to-answer time and increase detection reliability by optimizing electrical measurements and implementing a machine learning pipeline that intakes the electrochemical data and quantifies free cysteine concentrations as well as reduced antibody fragment levels in growth media. In doing so, we demonstrate a simple method-development platform for electrochemical dataset generation, feature selection, and model optimization that may be transferable to other biological production processes. Further, the developed method offers opportunities for at line digital integration for monitoring complex product attributes throughout bioprocessing.