Abstract
Understanding the mechanism of protein adsorption and transport inside of chromatographic adsorbent is critical for the downstream process design of the pharmaceutical industry, especially for the multicomponent system, such as protein monomer and its isoform dimer. This work forces on understanding a new form of hydroxyapatite (HAP) adsorbent, ceramic hydroxyapatite (CHT). The work determined the internal structure of two types of CHT, Type I and Type II, and the adsorption behavior of a monoclonal antibody (mAb) in monomeric and dimeric forms. Internal porosities and apparent pore radii based on inverse size exclusion chromatography are 0.73 and 30 nm for Type I and 0.70 and 49 nm for Type II. Adsorption isotherms show higher maximum capacities on Type I compared to Type II, in approximate agreement with the ratio of surface areas. The isotherms are dependent on the Na+ concentration consistent with an electrostatically driven mechanism. Mixture adsorption shows selectivity toward the dimer. Effective pore diffusivities for strong binding conditions, obtained by confocal microscopy, are much smaller than the non-binding values for Type I but essentially the same for Type II, indicating that diffusional hindrance by the bound protein is greater in the smaller pores of Type I.
The separation dynamics of monoclonal antibody monomer/dimer mixtures are first examined by frontal analysis. The binding capacity and selectivity are dependent on the CHT type and salt concentration. While the rate of protein adsorption on CHT Type I is slow and controlled largely by pore diffusion resulting in relatively poor separation, adsorption on CHT Type II is much faster and better separation is obtained than with Type I. However, comparison with predictions based on pore diffusion alone, reveals the presence of additional resistances associated with adsorption and displacement kinetics. A spreading kinetics model assuming multiple binding configurations coupled with pore diffusion was developed to describe these effects and found to be in quantitative agreement with the frontal analysis results and able to predict the separation achieved for conditions outside the range of the experiments. To help validate the assumed mechanism, isocratic elution experiments were also conducted at low protein loads. The chromatograms could be described by the solution of the spreading model coupled with pore diffusion in the linear region of the isotherm with parameters determined from the analytical moments confirming a trend of increasing tendency to spread and slower kinetics as the salt concentration is decreased and binding strength is increased.
Multicomponent separation is then examined by the gradient elution chromatography on two types of CHT adsorbents. Experimental results show that the pH drop introduced by the increased sodium concentration is significantly more extreme during a sodium chloride gradient compared to a sodium phosphate gradient. Therefore, a sodium phosphate gradient facilitates a better separation for this model mAb. The empirical interpolation (EI) method is employed in concert with the pore diffusion model coupled with lumped adsorption kinetics to predict multicomponent overloaded elution behavior. Predictions show good agreement with the experimental results for protein loads up to 85% of the column binding capacity.
Finally, a comparative study of the optimized separation process in the case of frontal analysis and gradient elution chromatography is presented using these mechanistic models developed in this work. The optimization behavior in terms of productivity and yield is examined under the constraints of monomer purity. The column simulations reveal that gradient elution exhibits higher productivity and yield compared to frontal analysis. However, with process optimization, frontal analysis can still achieve a separation process with 80% yield alongside a monomer purity cut-off at 95% which is applicable in an industrial setting.
