Rethinking ion exchange chromatography modeling using the colloidal particle adsorption (CPA) model
In the early days of modeling, mechanistic models for ion exchange chromatography were mostly based on stoichiometric equations. The 2021 release of the GoSilico™ Chromatography Modeling Software has introduced a new adsorption isotherm class that is based on a colloidal description of proteins, the colloidal particle adsorption model (CPA). In comparison to the standard steric mass adsorption (SMA) model, the new colloidal models achieves increased accuracy at high column loading.
The steric mass action model
For the past three decades, mechanistic models mainly relied on simple stoichiometric equations to describe protein adsorption in ion exchange (IEX) processes. The most prominent and widely used model is the steric mass action (SMA) isotherm. The SMA model describes the interaction between proteins and IEX resins by a reversible exchange of adsorber counter-ions by proteins in the mobile phase, while explicitly considering the influence of the ionic strength on protein adsorption. The protein binding capacity of an IEX adsorber is thereby physically constrained by its ionic capacity. The nonlinear adsorption behavior at high protein loads is further described by a steric shielding of adsorber ligands (Figure 1).
Fig 1. Depiction of repulsion and shielding effects that are covered by the assumptions of the SMA model
SMA vs reality: how to explain the peak shoulders?
Despite its popularity, in recent years we have seen shortcomings of the SMA model in describing industrial process behavior. Discrepancies between the SMA model and experimental data are particularly pronounced at high protein load densities combined with linear gradient elution. As illustrated by the following simulations using increasing loads, we thereby often observe the formation of a characteristic peak shoulder or “shark fins” in the front part of the elution peak that cannot be accurately described by the SMA model.
Fig 2. Formation of peak shoulder or ”shark fins” in front the elution peak for increasing loads.
Quote: "For the last three decades, our understanding of IEX chromatography has been based predominantly on principles of stoichiometry. Given the increasing number of processes that do not follow the behavior proposed by these models, we started to consider a non-stoichiometric description of protein adsorption."
Till Briskot, Former Research Engineer at GoSilico GmbH (now part of Cytiva).
The colloidal particle adsorption model CPA
To overcome the shortcomings of the SMA model, we introduced an alternative model for describing protein adsorption in IEX columns that is not based on stoichiometric equations. Using the colloidal nature of proteins, the new model allows a more fundamental description of interactions between proteins and charged ion adsorbers. Non-linear adsorption effects are thereby ascribed to steric hindrance at the adsorber surface and electrostatic interactions between adsorbed proteins (Figure 3). In contrast to the SMA model, the maximum protein binding capacity of the adsorber is physically constrained by the adsorber surface area accessible to proteins and not by its ionic capacity.
Fig 3. Schematic representation of the mechanistic understanding behind the CPA model.
Fig 4. Visualization of antibodies bindings to ligands inside a resin bead
Using the CPA model to describe the elution of a mAb
In a recent case study, we demonstrated the ability of the CPA model to describe complex elution behavior of a monoclonal antibody (mAb) on the Capto™ S ImpAct cation exchanger. Figure 5 shows three gradient elution experiments (rows) at different load densities (columns). In addition to charge variants of the mAb (second column), the CPA model accounted for different product-related and process-related impurities including low-molecular and high-molecular weight species (third column) as well as host cell proteins and leached Protein A (fourth column). At very high protein load densities close to or beyond the protein breakthrough (second and third row), elution peaks show a distinct shoulder that cannot be explained by the SMA model but can be reproduced by the CPA model.
Fig 5. Results of gradient elution experiments at different load densities. Adapted from Analysis of complex protein elution behavior in preparative ion exchange processes using a colloidal particle adsorption model.
Quote: "The improvement in model performance compared to the SMA approach is stunning and addresses a problem we’ve been struggling with. The ability to describe the strange peak shapes at high loading, both with respect to product related and process related impurities, suggests that better processes can be developed with the colloidal model."
Gunnar Malmquist, Senior Principal Scientist at Cytiva.
The future of IEX modeling and GoSilico™ Chromatography Modeling Software
With the upgrade of GoSilico™ Chromatography Modeling Software in early 2021, the CPA model was made commercially available. Within GoSilico™ Chromatography Modeling Software, the CPA can be activated using a switch between SMA and CMA models in the model selection menu.
The results achieved show the ability of the CPA model to describe complex elution behavior. Based on mechanistic principles, the CPA model enables a causal interpretation which can help to better understand industrial IEX processes and to support the development of these processes in a model-based approach.