Elucidating aggregation mechanisms in antibody fragment-based therapeutics to improve their manufacturability

Recent therapies for age-related diseases such as rheumatoid arthritis, macular degeneration, Crohn's disease, and some cancers are engineered forms of biological protein molecules called antibodies that form part of our own natural immune system. Such therapeutic proteins are being derived increasingly from simpler fragments of these antibodies with the hope that this will improve their behaviour in the body, reduce the frequency of injections required, allow them to target new regions of the body, and also allow them to be combined with other biological molecules without becoming too large or unstable. However the manufacturing of therapeutic proteins is extremely challenging due to their delicate and complex nature. Manufacturing processes aim to separate the protein molecules from the rest of the cellular components in which they were synthesised, to obtain extremely pure therapeutic material that is suitable for use in humans as a therapy. However, the processes available for large-scale manufacture place a great deal of stress on the protein due to changes in temperature or acidity, the addition of salts, the use of mechanical agitation, rapid changes in the rate of flow through machinery, and the interaction of proteins with air bubbles. This frequently causes the protein to deform slightly and to subsequently stick together to form tiny particles called aggregates. While these are often not visible to the naked eye, their presence in therapies can be hazardous to patients as they may cause severe inflammation and potentially more deadly immune responses. Therefore, one of the key challenges that the bioprocess development and therapeutic protein manufacturing industries would like to address is to be able to either predict the conditions that cause a protein to aggregate, or to increase their robustness so that they aggregate less frequently during their manufacture. We aim to carry out and demonstrate a suite of rapid experimental measurement techniques that allow a new therapeutic protein to be evaluated quickly for the conditions in which they have a greater tendency to form aggregates. The conditions to be tested will be same as those used throughout bioprocess manufacturing, and will therefore allow bioprocess engineers to rapidly identify the conditions in which their manufacturing processes will be best operated, or whether the protein is unlikely to be manufacturable. Having quickly determined the conditions at which the protein begins to form small and soluble aggregates, we will also carry out a detailed molecular analysis of the structure of proteins at these conditions and also those either side in which the protein remains in solution as a single molecule, and where it forms larger aggregates. This will allow us to see what changes in the protein structure occur before, during, and after the aggregation is initiated and therefore deduce which events are on the critical path to aggregate formation. Having achieved this we will then be able to target changes to the protein called mutations that will interfere with and suppress the aggregation process. Finally, by comparing a related set of therapeutic antibody fragment proteins, we will gain insight into those factors that are specific to each protein type, and those that occur more generally and hence become useful targets for the future engineering of therapeutic protein designs. It will also allow others to improve their mathematical modelling methods that aim to predict whether proteins will aggregate under certain conditions.

Biopharmaceuticals are derived increasingly from Fab and ScFv antibody fragments. A major challenge for their manufacture is their tendency to aggregate during bioprocessing due to increasing protein concentrations, low pH, high salt, shear or surface effects in centrifugation, filtration, chromatography or viral inactivation steps. Aggregates are hazardous to patients and yet their necessary removal leads to longer and more costly bioprocess development times. Characterisation of the molecular events that occur as antibody-fragments aggregate under typical bioprocess conditions will provide a fundamental basis for designing more robust antibody fragment scaffolds and for improving current predictive tools. Several protein engineering efforts have selected antibody fragments resistant to pH or heat induced aggregation using libraries that vary only in the CDR loops, though these regions are also required for the selection of antigen binding affinity. Sequence algorithms have improved the prediction of sites with high aggregation propensity for small globular proteins by combining secondary-structure propensity (an indirect measure of local protein unfolding and beta-sheet forming potential) with amino-acid hydrophobicity and charge indices. However, these do not yet include the role of specific protein-protein interactions or the relative conformational positions of multiple domains in larger proteins which can both influence the initial formation of small soluble oligomers prior to aggregation. We aim to characterise the protein conformations, global and local unfolding, and specific interactions that occur within protein monomers and the small soluble aggregates formed as bioprocess conditions are changed. We will use a combination of microscale experiments, protein engineering, AUC, NMR and SAXS combined with constrained protein modelling. Identifying key protein sites and structural effects will fundamentally improve the design of robust therapeutic scaffolds.

'WHO WILL BENEFIT FROM THE RESEARCH?' UK-based companies within the BRIC community will benefit from research which allows them to more effectively predict lead antibody-fragment candidates that are likely to aggregate during bioprocess development and manufacturing scale up, using high-throughput tools for a 'screen early - fail early' approach. A propensity to aggregate directly increases manufacturing process development costs due to time spent on finding bioprocesses that minimise it, and also due to absorbing the cost of failing to find a suitable bioprocess with earlier lead candidate molecules. Characterisation of aggregation under bioprocessing conditions for a variety of antibody-fragment based molecules will provide a broader and better mechanistic understanding of how aggregation is initiated, and ultimately how protein engineering can be used to intervene and produce scaffolds that are generally more robust to aggregation. This will in turn decrease the time, cost and risk of product development. Potential patients will benefit because the research will significantly aid reduction in development times of antibody-fragment based medicines, which is particularly crucial for those addressing previously unmet clinical needs. Benefits to the NHS relate to the possibility of constraining costs. Proteins are innately complex and labile so that bioprocess development times and hence costs tend to be high. The capacity to treat conditions such as rheumatoid arthritis much more effectively in ageing populations is vital but it still poses a problem with respect to stretched NHS budgets. 'HOW WILL THEY BENEFIT FROM THE RESEARCH?' The research will fundamentally characterise the early structural events that lead to antibody-fragment aggregation at a wide range of bioprocess conditions. This will reveal commonalities between several antibody-fragment types, as well as their idiosyncrasies. This improved understanding will better direct protein engineering and predictive tools in future for industry to minimise the aggregation propensity of new molecules with a significantly higher consistency. The research will also demonstrate simple microscale screening tools to rapidly identify the range of solution conditions in which bioprocesses can be operated safely without causing protein aggregation. These will be examined as indicators for molecular robustness at a broader range of in-process stresses including shear, surface interactions, air-liquid interfaces, freezing and freeze-drying. Together, the rapid evaluation tools and greater understanding of the structural mechanisms that initiate antibody-fragment aggregation will enable the UK biotechnology industry to design more efficient bioprocesses and robust protein scaffolds that minimise aggregation. This will continue to improve as academic beneficiaries in the UK will also be able to refine their predictive models for aggregation propensity by including the effects of solution conformation and specific protein-protein interactions elucidated here. Finally we will also compare different approaches to guide protein engineering for minimising aggregation, including existing predictive tools, previously identified mutations from other antibody models and guidance from the NMR spectra determined by us. The UK economy will benefit because academic research will complement the country's strength in bioscience discovery. Collaboration between bioprocess engineers and protein biophysicists on industrially relevant therapeutic proteins will ensure effective knowledge and skills transfer between the science and engineering base and UK industry. This will expand their position in the global healthcare market and attract further R&D investment from global business which recognises the UK as a good place to conduct these activities. Such retention of expertise, know-how and intellectual property will aid the capacity to remain internationally competitive.