Engineered chimeric virus-like particles (VLPs) have shown great potential as cancer immunotherapeutic. Chimeric VLPs incorporate antigenic epitopes that can alter the assembly competence, solubility and stability of the virus proteins. To reduce those effects, the epitope inserts are fused to linkers. However, integration of a linker often does not lead to the desired effect of creating an assembly competent, soluble and stable VLP, suggesting there is a dependence of these properties on the VLP platform and the structure of the inserted epitope and linker. In this paper, we propose an in silico approach to predict VLP solubility in clarified harvest based on a chimeric Hepatitis B virus core antigen (HBcAg) VLP platform and reveal critical insert properties. Recent computational advances have rendered molecular dynamics (MD) simulations applicable to more complex protein structures. From X-Ray diffraction structures of the native HBcAg dimer we generated by homology modeling and MD simulations 3D models of 27 virus protein dimers associated with chimeric HBcAg VLPs, varying in the sequence of the insert. These 27 VLPs were characterized by their solubility in clarified E. coli harvest. We created in silico models to predict VLP solubility applying QSPR modeling. Based on the protein 3D structure and amino acid sequence, molecular descriptors related to geometry, charge and hydrophobicity were calculated. Besides well-established descriptors, we implemented descriptors that were specifically designed for this study, assuming that a rational design of the QSPR model potentially decreases the probability of including unnecessary or erratic descriptors. With principal component analysis (PCA) data dimension was reduced and solubility class membership was predicted by classification of the constructs into clusters. The model assigned more than 91% of the constructs correctly in cross-validation. Analysis of the model uncovered critical insert properties that may aid the design of novel inserts or be the basis for the modification of insoluble constructs. In conclusion, we developed a rational QSPR approach to predict VLP solubility. We established a tool to reduce screening efforts in early VLP drug development, cutting down the number of required experiments including DNA synthesis, plasmid transformation and expression of potential candidates by the ability of identifying insoluble VLP species thus potentially decreasing time-to-market and R&D costs.

Predicting the solubility of VLPs: Qualitative structure property relationship (QSPR) modeling applied to HBcAg VLPs

255th National Meeting (2018)

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