Virus-like particles (VLPs) are biotherapeutics that can trigger immune reactions against their parental virus or against foreign epitopes displayed on the chimeric VLP surface. Chimeric VLPs are obtained by insertion of antigenic epitopes into viral capsid proteins. Development of chimeric VLPs requires screening of a large number of epitope insertions to meet immunogenicity and manufacturability requirements. Manufacturability is usually assessed by high-throughput screening (HTPS) experiments to determine the effects of environmental factors such as pH on critical manufacturability parameters such as solubility. These effects are highly influenced by the protein’s physicochemical properties. Knowledge of these properties helps in HTPS design and in predictive assessment of manufacturability. This knowledge can be obtained computationally using 3-D structures to extract protein surface descriptors. The current work presents a high-throughput pipeline for the preparation and refinement of 3-D structures that supports structure screenings with a fully automated and data-dependent workflow. As case study, in silico 3-D structure preparation, refinement and surface charge description of 3 chimeric HBcAg dimers was performed. Quality, speed, stability of the molecular dynamics (MD) simulation and prediction of experimental VLP zeta potential data was evaluated with 2 different force fields (YASARA2 and AMBER03) and 2 different pKa estimation methods (YASARA and H++). To prevent subjective simulation control by visual inspection of the deviation of atom coordinates over time, an automated and data-dependent workflow was developed. Step-wise MD simulation reduced computational effort. MD simulations were terminated automatically by identifying a 2 ns Window of Stability (WoS). The WoS was also used to calculate surface charge based on 1000 structural snapshots. The median surface charge and median absolute deviation in the WoS were found to be a more reliable descriptor than surface charge based on one snapshot alone. Evaluation of the performance indicates the applicability of the proposed computational pipeline for early chimeric VLP candidate assessment and design optimization of HTPS manufacturability experiments. Minimal user interaction, ease of use and data-dependent MD simulation control increase reproducibility and robustness. In turn this may enhance model development for the prediction of physicochemical surface properties of VLPs and other proteins.

High-throughput computational pipeline for 3-D structure preparation and in silico protein surface property screening: A case study on HBcAg VLP surface charge

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.