381122 Combined Effect of Saline and Organic Modifiers on HIC and Rpc Separation of Insulin Variants
With the introduction of the first biopharmaceuticals in the early 1980s, the pharmaceutical industry was transformed. Since then, the market for medicines based on recombinant proteins has grown rapidly, and today approximately a fourth of all new pharmaceuticals belong to this group.  However, the complexity of the fermentation broth from the production of recombinant proteins calls for sophisticated purification methods. One of the most widely used and most successful purification methods in this area is preparative chromatography.  Since proteins are prone to denaturize at increasing levels of organic modifier, the relatively mild conditions in hydrophobic interaction chromatography (HIC) are favorable, and this chromatographic mode is very commonly applied for protein purification. Reversed-phase chromatography (RPC) is the chromatographic mode-of-choice for analytical purposes, but it can also be a powerful tool for separation of proteins which can tolerate the harsher conditions, e.g. insulin and peptides. [3, 4]
Both HIC and RPC are based on hydrophobic interactions between the ligands and the proteins, but the two modes differ in the degree of hydrophobicity, and in the type of mobile phase modifier used. In HIC, the hydrophobicity of the resin is low, and thus, addition of a salt is sufficient to modify the retention. In RPC, on the other hand, the resin is highly hydrophobic and elution is achieved by adding an organic modifier, such as ethanol, to the mobile phase. Nevertheless, HIC and RPC are based on the same fundamental physical phenomena, and in principle, both saline and organic modifiers can be used to modify both types of chromatography. These similarities suggest that a common mechanistic model for HIC and RPC can be developed.
As chromatographic experiments are expensive as well as time-consuming, model-aided process development and optimization is becoming increasingly attractive to the pharmaceutical industry. An undisputable prerequisite for model-aided process development and optimization is the existence of a model structure which describes the process well. In contrast to ion-exchange chromatography (IEC), for which the steric mass action (SMA)  and the Langmuir mobile phase modulator (MPM)  models are well-established, the debate on how to describe the hydrophobic phenomena involved in HIC and RPC is still on-going. The effects have, for example, been assigned to changes in the mobile phase surface tension , in the salt-solute interactions , and in the solute activity coefficient .
Assuming that the latter is correct, a change of resin would only affect the adsorption equilibrium constant, and for the same combination of solute and solvent system, the effect of the modifier concentrations would be the same. One interesting route towards a more general and predictive model is to attempt to develop a combined model for HIC and RPC, in which the modifier effect is attributed to changes in the activity coefficient of the solute. This is the motivation for this work; we want to investigate, understand and describe the common factors and dividers for the two modes of chromatography. The aim of our present study is to investigate whether or not similarities in modifier effects can be observed, and thus if the suggested route is feasible.
The proteins chosen for this study were three insulin variants, with very similar physicochemical properties. A mock feed was prepared by adding the insulins to a buffer without any salt and organic modifiers. All experiments were linear range isocratic pulse experiments, performed at constant temperature and pH. The HIC resins used were two methacrylate-based ones, one with butyl ligands and one with phenyl ligands. A C4 and a C18 resin, on silica backbone, were used for the RPC experiments. The mobile phase modifiers chosen were ethanol and KCl.
For the HIC resins, three different experiment series with varying KCl concentrations were run; at 0, 5, and 10 wt% ethanol. For some of the series, the solubility of KCl in the ethanol-water system was limiting, and in these cases, high retention volumes could not be reached. For the RPC resins, five different experiment series were run within the range 0.1-0.7 mol KCl/kg solvent, and with varying ethanol content. Each series ranges from almost non-retained conditions, up to retention volumes above 30 CV. The total column porosity and the system dead volume were determined from salt pulse experiments.
From the retention volumes, the natural logarithm of the thermodynamic retention factor (A) was calculated for each solute in each experiment. For the HIC data, lnA was plotted against κ2, the squared inverse of the Debye length, which is directly proportional to the ionic strength. All series display a positive linear dependence on κ2, but with different slopes for different ethanol content. This is consistent with Kirkwood’s theory for the effect of salt on proteins . Many salts have a salting-out effect at high concentrations and a salting-in effect at low concentrations, resulting in a minimum in retention. However, for the combination of solutes and solvent system studied, no salting-in effect was observed on the HIC resins.
For the RPC data, lnA was plotted against the volume fraction of ethanol in the mobile phase. The result was parallel, slightly convex curves with a negative dependence on the ethanol content. In accordance with the assumption that the modifier concentrations only affect the solute activity coefficient, quadratic polynomials with different intercepts but the same slope and curvature could be fitted to all RPC series, independent of resin. Additionally, the intercepts for each resin decrease linearly with increasing salt concentration. Thus, the solutes display a salting-in behavior on the RPC resins, within the interval 0.1-0.7 mol KCl/kg solvent.
The next step is to calibrate a linear isotherm, in which the modifier effects are attributed to changes in the solute activity coefficient, to the data collected. As the results agree well with Kirkwood’s salting-in and salting-out theories, the ability of these theories to describe the changes in the solute activity coefficient will be investigated. When the linear isotherm has been established, we will study the non-linear range, with the aim to develop a full model which can be used as a tool in process development and optimization.
The Swedish Foundation for Strategic Research (SSF) and the Process Industry Centre at Lund University (PIC-LU) are gratefully acknowledged for financial support. Nanna Mikkelsen at Novo Nordisk in Bagsværd, Denmark has provided invaluable help with the experimental equipment.
1. Walsh, G., Second-generation biopharmaceuticals. Eur. J. Pharm. Biopharm., 2004. 68: p. 185-196.
2. Jungbauer, A., Chromatographic media for bioseparation. J. Chromatogr. A, 2005. 1065: p. 3-12.
3. Freitag, R., Chromatographic Techniques in the Downstream Processing of Proteins in Biotechnology, in Animal Cell Biotechnology - Methods in Molecular Biology, R. Pörtner, Editor 2014, Humana Press: New York, NY, USA. p. 419-458.
4. Graumann, K. and A.-A. Ebenbichler, Development and Scale up of Preparative HIC for the Purification of a Recombinant Therapeutic Protein. Chem. Eng. Technol., 2005. 28(11): p. 1398-1407.
5. Brooks, C.A. and S.M. Cramer, Steric Mass-Action Ion Exchange: Displacement Profiles and Induced Salt Gradients. AIChE J., 1992. 38(12): p. 1969-1978.
6. Melander, W.R., S. El Rassi, and C. Horváth, Interplay of hydrophobic and electrostatic interactions in biopolymer chromatography: Effects of salts on the retention of proteins. J. Chromatogr., 1989. 469: p. 3-27.
7. Melander, W. and C. Horváth, Salt Effects on Hydrophobic Interactions in Precipitation and Chromatography of Proteins: An Interpretation of the Lyotropic Series. Arch. Biochem. Biophys., 1977. 183: p. 200-215.
8. Arakawa, T. and S.N. Timasheff, Preferential Interaction of Proteins with Salts in Concentrated Solutions. Biochemistry, 1982. 21: p. 6545-6552.
9. Mollerup, J.M., A Review of the Thermodynamics of Protein Association to Ligands, Protein Adsorption, and Adsorption Isotherms. Chem. Eng. Technol., 2008. 31(6): p. 864-874.
10. Kirkwood, J., Protein, Amino Acids and Peptides as Ions and Dipolar Ions, in American Chemical Society Monograph Series, E.J. Cohn and J.T. Edsall, Editors. 1943, Reinhold Publishing Corporation: New York, USA.