PROPOSAL NUMBER: 0731055 PRINCIPAL INVESTIGATOR: Erik Fernandez INSTITUTION: University of Virginia
PROPOSAL TITLE: Relating Protein Structure to Stability in the Solution and Adsorbed Phases
The goal of the project is to establish predictive relationships between the structural and molecular properties of proteins in solution and their unfolding behavior on hydrophobic interaction chromatography (HIC) surfaces. There are three elements of the research effort: experiment, computation, and modeling. The first aim is measuring batch adsorption, chromatography, and hydrogen-deuterium isotope exchange (HX) to establish the relevant equilibrium, kinetic, and structural aspects of a set of single and multidomain proteins on different adsorbent surfaces. Residue-level comparisons of solvent accessibilities in solutions and on surfaces will be made to reveal expected patterns of local stability propensities under the different environments. The second aim is to match these experimentally detected patterns with predicted local stabilities in solution computed by the COREX software based on statistical mechanical ensembles of protein conformation to confirm its capability for obtaining distributions of microstate energies. The third aim is to formulate in ensemble terms a four-state model that has been successful in correlating static and dynamic stability behavior of such adsorbing systems, and use the experimental and simulated results to develop robust relationships for protein adsorption and stability. The intended outcome is to provide techniques for predicting when and how proteins unfold upon adsorption, including elucidation of local regions of instability.
The most direct and critical application of such a predictive capability is to therapeutic protein purification. In that context, removing misfolded, unfolded, and aggregated protein from therapeutic proteins is of growing importance. These predictive tools will ultimately enable process development engineers to minimize unfolding of native protein and maximize selectivity between folded and misfolded or unfolded proteins. Controlling the behavior of proteins on surfaces is central to drug delivery, biomaterials, biosensors, protein arrays, and nanoscale devices. In all of these situations, controlling the degree and rate of protein adsorption, the changes in protein conformation, and the appearance of aggregation are critical. Undergraduates recruited through the University of Virginia Center for Diversity in Engineering will participate in the research. A multidisciplinary collaboration with Prof. Vincent Hilser (U. Texas Medical Branch) will enable both graduate and undergraduate students to apply important biophysical and statistical mechanical approaches to problems of fundamental and practical importance. Collaborative interactions with supporting biopharmaceutical companies will provide relevant multidomain proteins for the work, disseminate the results to practitioners, and expose students to protein adsorption and stability issues in a commercially relevant context. Finally, a new module for biochemical engineering courses will be developed to incorporate aspects of this research, including computational and modeling activities. This module will be disseminated to other chemical engineering faculty via an electronic journal website being set up jointly with San Jose State University.
Intellectual merit. The goal of the research has been to establish predictive relationships between structural and molecular properties of proteins and their unfolding behavior on hydrophobic interaction chromatography (HIC) surfaces. There were three elements of the research effort: experiment, computation, and modeling. The first aim was to use batch adsorption, chromatography, and hydrogen-deuterium isotope exchange (HX) to establish the relevant equilibrium, kinetic, and structural aspects of a set of single and multidomain proteins on different adsorbent surfaces. The second aim was to match these experimentally detected patterns with predicted local stabilities in solution computed by the COREX statistical mechanical model. The third aim was to apply a four-state model to describe the adsorption and conformational behavior, and then use the experimental and calculational results to develop robust relationships for protein adsorption and stability that should require few, if any, data. The intended outcome was to provide techniques for predicting when and how proteins unfold upon adsorption, including, for the first time, local regions of instability. Using hydrogen-deuterium isotope exchange detected by mass spectrometry, we have shown that folding on hydrophobic chromatography surfaces remains 2-state, and that the unfolded state can be partially unfolded. Further, this partially unfolded state seems to be similar to molten globule states of the same protein observed under solution conditions. We used COREX to predict hydrogen exchange patterns for lysozyme, a protein we studied on very hydrophobic reversed phase chromatography surfaces. The results were consistent with those from previous residue-level hydrogen exchange solvent accessibility patterns in solution. Even more interesting, the solution phase solvent accessibility patterns predicted by COREX capture elements of the patterns of partial unfolding on reversed phase chromatography surfaces. However, tests with other proteins were less successful. Consequently, we are currently pursuing alternate simulation approaches to accomplish this. The two-conformation, four-state model was successfully applied to the model unstable protein α-lactalbumin in HIC under marginally stable conditions. Below the melting temperatures of the wild-type protein and the variants, reduction of a single disulfide bond was shown to significantly increase the apparent adsorption strength due to increased instability of the protein. Further, literature data for one of the disulfide-bonded variants was used with the four-state model to accurately predict the apparent adsorption strength. The results have applications to the purification of misfolded and mis-disulfide bonded proteins. This model was also applied to investigate adsorption kinetics and reversibility. In addition, a new protein adsorption isotherm was developed to account for the effects of hydrophobic ligand type and density. Broader impacts. Three undergraduate researchers participated in the project, and two have gone on to graduate study in chemical engineering. A multidisciplinary collaboration with Prof. Vincent Hilser (U. Texas Medical Branch), enabled both graduate and undergraduate students to apply important biophysical and statistical mechanical approaches to problems of fundamental and practical importance. Collaborative interactions with supporting biopharmaceutical company BiogenIDEC provided relevant multidomain proteins for the work, disseminated the results to practitioners, and exposed students to protein adsorption and stability issues in a commercially relevant context. In particular, in collaboration with BiogenIDEC, graduate student Robert Deitcher applied various experimental techniques and models to characterize monomeric and aggregate forms of an Fc-fusion protein undergoing a HIC process.
