Higher-Order Structure and Solution Interactions of Antibodies

National Institute Of Biomedical Imaging And Bioengineering, Bethesda

Investigators

Linked publications & trials

Abstract

Monoclonal antibodies (mAbs) are a rapidly growing class of protein therapeutics. Beyond the obvious high-affinity interaction of antibodies with their antigen target, there are three additional types of antibody interactions that are of interest in pharmaceutical formulations. First, short- and long-range electrostatic and hydrodynamic forces modulate protein distance distributions in solution. This may result in net repulsive or attractive interactions without formation of protein complexes. Though very subtle, such interactions are key to macroscopic solution behavior and long-term protein stability. They are commonly measured through nonideality coefficients of sedimentation, diffusion, and thermodynamic virial coefficients. A second type of antibody interactions is their propensity to self-associate through formation of short-lived, reversible protein-protein complexes. These have been associated with high solution viscosity. Both nonideality and self-association properties are concentration-dependent and cannot be measured well in dilute solution. A third class of interactions form long-lived oligomers, often due to misfolded protein. These can be immunogenic and therefore must be tightly controlled. To minimize the volume required for therapeutic administration, mAbs are typically formulated at concentrations on the order of 100 mg/ml, which is far above the range of existing particle sizing methods. While sedimentation velocity analytical ultracentrifugation has become the gold standard for measuring higher-order structures in dilute solutions, new opportunities have arisen for characterizing mAb solutions much closer to formulation conditions with the new technique of nonideal sedimentation velocity (SV), recently developed in our laboratory. Using a panel of monoclonal antibodies, in collaboration with AstraZeneca, we have established concentration limits of nonideal SV for IgG mAbs to be 45 mg/mL, approximately a factor 10 higher than previously possible. We were able to show that with such highly concentrated samples it is possible to reliably measure nonideality coefficients of sedimentation. Simultaneously the new nonideal SV approach provides a simple and sensitive means to characterize self-association of antibodies. This work has highlighted a theoretical problem in the impact of weak self-association on the measurement of nonideality coefficients with different biophysical methods, which may lead to grossly inconsistent nonideality coefficients, as frequently found in the literature. In theoretical analysis we have clarified the origin of these discrepancies in the choice of thermodynamic reference frames customarily used in different methods. To overcome this problem, we have extended our global multi-method analysis (GMMA) software to allow global analysis of nonideality from different data sources at once. This bridges gaps in sensitivity and resolution between different techniques, and permits the use of a self-consistent thermodynamic reference frame that links nonideality parameters up with statistical fluid dynamics theory. In the reporting period we have further explored the parameter correlation in these GMMA models, and their sensitivity to different assumptions on the self-association mechanism. In addition, we have further refined the analysis software for dissemination of this approach. In particular, we have expanded the user interface to custom-build self-association and hetero-association models including nonideality or multi-method analysis. This will facilitate the dissemination and wider application of the GMMA approach. Beyond the characterization of antibodies, GMMA of weak interactions and nonideality provides an experimental platform for related interests in DMAS in the field of concentrated protein solutions, including eye lens crystallins. Antibody formulations can contain long-lived aggregates from misfolded protein that can form with time. These are potentially immunogenic and therefore trace quantities must be monitored and reported to FDA. In order to examine what impact the higher concentration limits of nonideal SV have on the limits of quantitation for trace protein dimers, we have previously embarked on a study of heat-denatured NISTmAb reference antibody, in collaboration with the laboratory of Dr. John Schiel (NIST). We have carried out mixing experiments of concentrated native and heat-denatured NISTmAb in parallel by nonideal SV and size exclusion chromatography. In the reporting period, we have further refined the data analysis protocols used in this study to optimize trace aggregate detection. Finally, in order to improve the application of SV to trace aggregate detection in pharmaceutical antibody preparations, we have collaborated with a working group of colleagues from the biotechnology industry to develop recommended best practices. This can serve as a guide to standardize the application of the technique, and as educational reference material for novices. Furthermore, we have enhanced our SV data analysis software for improved compatibility with regulatory requirements.