PROGRAM | Chemical Engineering

By: Chase Herman Chair: Abraham Lenhoff

ABSTRACT

A central paradigm in chemical engineering is to connect molecular structure with continuum properties and use that information for process design. This has been well achieved in the processing of relatively small-molecule products but much less successful in biopharmaceutical manufacturing due to the complexity of macromolecules and their interactions. However, there is increasing incentive to integrate fundamental insights and mechanistic models into process development and control to mitigate problems that arise during empirical process development, such as the appreciable co-purification of impurities that is observed with some monoclonal antibody (mAb) candidates. The goal of this work is to develop such insights into the chromatographic behavior of protein impurities in mAb processing, including host-cell proteins (HCPs) and aggregates. The first part of this work uses primarily computational techniques to investigate the transport and thermodynamic contributions to protein behavior in ion-exchange (IEX) chromatography and the second part uses various experimental techniques to characterize HCP-rich aggregates and to study their process behavior.

Flow-through IEX is often used to remove trace HCPs before mAb formulation but it is experimentally challenging to study HCP clearance on-column given the numerous species that may be present at dilute concentrations. Column modeling was used in this work to illustrate that the breakthrough of HCPs can be much more diffuse than is typically observed under overloaded conditions (e.g., for mAb products). Transport contributions to this behavior can be appreciable and may be described generally using the Graetz number.

Literature data were also consolidated to search for trends in IEX adsorption equilibria and a correlation was discovered between stoichiometric displacement model (SDM) parameters, which may be useful for reducing the experimental burden of estimating characteristic charge parameters for modeling retention. Continuum electrostatics modeling at the protein scale was then used in part as an attempt to explain the empirically-observed SDM parameter correlation. Pre-existing mean-field models of IEX adsorption equilibria were found to be incapable of reproducing the correlation but an anisotropic model developed in this work was able to capture the essential features. However, all of the continuum models were incapable of accurately estimating protein IEX adsorption equilibria ab initio.

Any modeling effort must necessarily balance the competing demands of fidelity and computational tractability but identifying what chemistry and physics is required to have an accurate representation may be nontrivial. To investigate this, the ability of different molecular dynamics models to reproduce an experimental trend in IEX resin retentivities was assessed. The inclusion of electronic polarizability was found to be essential for reproducing the retentivity trend; this granularity may be required for modeling IEX interactions because ion pairing emerges from competition between two large potentials, namely electrostatic attraction and solvent opposition, and it is the fine balance between them that determines association strengths in solution.

In the second part of this work, HCPs were investigated experimentally as a multicomponent system. Recent indications that aggregates may contain appreciable HCP content prompted a cross-digest proteomic analysis of aggregates from harvest cell culture fluid (HCCF) and protein A eluate that had been viral-inactivated and neutralized (PAVIN). This analysis showed that HCPs are much more conserved across HCCF and PAVIN aggregates than observed previously; these aggregate structures appear to mediate the majority of HCP persistence through the protein A capture step and difficult-to-remove HCPs tend to be generally more concentrated than their counterparts in HCCF. A primary analysis was then performed of aggregate process behavior in steps that are typically implemented for HCP clearance, including the protein A wash step and flow-through IEX. Investigations at the column and resin length scales revealed that HCP-rich aggregates adsorb specifically in protein A chromatography and that absorptive competition from mAb molecules appears to be implicated in the efficacy of protein A washes. It was also discovered that relatively large aggregates that persist into the PAVIN and harbor most of the HCP mass are differentially retained on IEX resins and this appears to depend primarily on the resin surface chemistry and pore structure. The persistence of these aggregates correlates generally with HCP clearance and may serve as a convenient albeit imperfect proxy for informing process development decisions.

Back >