Therapeutic proteins such as antibodies constitute the most rapidly growing class of pharmaceuticals for use in diverse clinical settings including cancer, chronic inflammatory diseases, kidney transplantation, cardiovascular medicine, and infectious diseases. Unfortunately, they tend to aggregate when stored under the concentrated conditions required in their usage. Aggregation leads to a decrease in antibody activity and could elicit an immunological response. Using full antibody atomistic molecular dynamics simulations, we identify the antibody regions prone to aggregation by using a technology that we developed called spatial aggregation propensity (SAP). SAP identifies the location and size of these aggregation prone regions, and allows us to perform target mutations of those regions to engineer antibodies for stability. We apply this method to therapeutic antibodies and demonstrate the significantly enhanced stability of our mutants compared with the wild type. The technology described here could be used to incorporate developability in a rational way during the screening of antibodies in the discovery phase for several diseases.aggregation ͉ antibody ͉ genetic engineering ͉ molecular simulation T herapeutic antibodies are glyco-proteins tailor-made to seek out and attach themselves to specific antigens, such as biomarkers on the surface of cancer cells. Because of their potential in the cure of various diseases, antibodies currently constitute the most rapidly growing class of human therapeutics (1). Since 2001, their market has been growing at an average yearly growth rate of 35%, the highest rate among all categories of biotech drugs (2). One of the major problems encountered in antibody-based therapies is that these antibodies tend to aggregate under the high concentration formulations required for disease treatment (3). Aggregation leads to a decrease in antibody activity and could elicit an immunological response (4, 5). It also has implications for regulatory approval and for delivery methods. Stabilization of therapeutic antibodies, however, is generally performed during the development phase using trial and error methods. These are both costly and time consuming. Thus, there is a need for a screening tool that will assess aggregation behavior. Such a tool could be used, for example, to evaluate the developability of a bio-pharmaceutical coming during the discovery phase. Protein aggregation in vivo is also shown to be directly responsible for many diseases such as type II diabetes and Alzheimer's (6, 7). Thus, an understanding of the fundamental mechanisms and regions involved in protein aggregation is of importance both for stabilizing protein therapeutics and for devising strategies to prevent in vivo aggregation.Although there are many possible mechanisms for aggregation, hydrophobic interactions were shown to be the predominant interactions in extensive studies of protein folding and protein-protein binding (8-13). Our results below show that these interactions also play a key role in antibody...

Therapeutic proteins such as antibodies are playing an increasingly prominent role in the treatment of numerous diseases including cancer and rheumatoid arthritis. However, these proteins tend to degrade due to aggregation during manufacture and storage. Aggregation decreases protein activity and raises concerns about an immunological response. We have recently developed a method based on full antibody atomistic simulations to predict antibody aggregation prone regions [Proc. Natl. Acac. Sci. 2009, 106, 11937]. This method is based on "spatial-aggregation-propensity (SAP)", a measure of the dynamic exposure of hydrophobic patches. In the present paper, we expand on this method to analyze the aggregation prone regions over a wide parameter range. We also explore the effect of different hydrophilic mutations on these predicted aggregation prone regions to engineer antibodies with enhanced stability. The mutation to lysine is more effective than serine but less effective than glutamic acid in enhancing antibody stability. Furthermore, we show that multiple simultaneous mutations on different SAP peaks can have a cumulative effect on enhancing protein stability. We also investigate the accuracy of various cheaper alternatives for SAP evaluation because the full antibody atomistic simulations are highly computationally expensive. These cheaper alternatives include antibody fragment (Fab, Fc) simulations, implicit solvent models, or direct computations from a static structure (i.e., a structure from X-ray or homology modeling). The SAP evaluation from the static structure is 200,000 times faster but less accurate compared to the SAP from explicit atom simulations. Nevertheless, the SAP from a static structure still predicts most of the major aggregation prone regions, making it a potential approach for use in high-throughput applications. Thus, the SAP technology described here could be employed either in high-throughput developability screening of therapeutic protein candidates or to improve their stability at later stages of manufacturing.

No abstract

No abstract