Protein stability and aggregation are major, and still largely unsolved, issues affecting the development and production of biopharmaceuticals. Besides their impact on development costs, the safety of biopharmaceuticals is also significantly affected by these issues. Protein aggregation is a major element behind the immunogenicity, and perhaps also the toxicity, of many bioactive polypeptides. Traditionally, protein misfolding and aggregation problems have been approached by implementing costly DSP and formulation strategies in an attempt to minimise the impact of protein mis-assembly in the development and viability of polypeptide drugs.

A predictive in silico platform (AggreSolve™) has been developed to overcome protein stability and aggregation issues and provide solutions to improve the quality of biotherapeutics. This platform can be used to predict intrinsic aggregation propensity of virtually any polypeptide, giving an idea of the tendency of a given sequence to aggregate during expression or production. The method makes possible to extract critical information by calculating solvent accessibility, structural preferences, aggregation propensity, and potential inter-molecular interactions. This information can be combined to describe potential stability and aggregation issues that might appear during development and propose potential modifications that could contribute to stabilise a polypeptide by removing aggregation ‘hot-spots'.

We show that in silico predictive methodologies can be used to screen out polypeptides with aggregation and stability issues early on in the development process. Similar approaches can be applied to generating improved biopharmaceuticals through protein engineering. We have successfully applied such tools to identify polypeptides with aggregation and stability problems and to generate improved versions of therapeutic antibodies.

As one of the potential applications of the technology we show how AggreSolve™ can be applied to re-engineer full antibodies that show stability and aggregation problems. In a particular case, we show how potential aggregation ‘hot-spots' were identified and more than 1,000 possible substitutions were screened in silico before producing a reduced number of variants to be analysed in vitro. Where modifications aggregation hot-spots affected CDRs, an alternative method involving protein engineering ‘by proxy' was employed. The short-listed variants were expressed and fully characterised and exhibited a significant reduction in aggregation levels, whilst activity remained largely unaffected. The results show how protein engineering can be used successfully applied to reduce aggregation problems in complex proteins such as full antibodies.

Extended Abstract Status: Not Uploaded