Purpose: Protein aggregation can be a major pathway of degradation for a therapeutic monoclonal antibody (mAb), reducing potency and drug product quality. The self-association of two mAb monomers to form a dimer is the first step towards nucleation. Since this process is usually reversible, a better understanding of the structure of these early precursors of aggregates would be valuable in designing stable mAb formulations. Here, we present an in silico protocol for predicting the formation of self-associated mAb dimers and using such three dimensional structural information of the aggregation prone regions (APRs) to tailor design excipients with greater specificity to modulate aggregation.
Methods: Hydrophobicity is a major factor in protein aggregation as APRs are usually characterised by an enrichment in hydrophobic residues. We have employed Spatial Aggregation Propensity (SAP) (Chennamsetty et al., 2009) to calculate the hydrophobicity of these exposed surface patches of “mAb 1” from its crystal structure (PDB ID: 5JZ7) (Lowe et al., 2016). The region that has the highest hydrophobicity was then selected to conduct a virtual screening using AutoDock Vina (Morris et al., 2009) against a library of commercially available compounds from ZINC database. The binding configurations and binding energy prediction of mAb 1 along with the effect of the novel specific excipient were further examined by Molecular Dynamics (MD) simulations based on the structural information of the Fv fragment of mAb 1 using MARTINI force field (de Vries et al., 2007) through the GROMACS MD engine (Pronk et al., 2013). Replicates were generated with DAFT (Böckmann et al., 2015). Liquid formulations of mAb1 with or without the novel specific excipient were prepared. The effect of the novel excipient is evaluated experimentally using DynaPro to determine diffusion interaction parameter (KD).
Results: CDR on mAb 1 has the highest hydrophobicity and this region was selected for conducting a virtual screening. A compound (CPX) was found to have the highest score of -8.1 from AutoDock Vina towards the region compared with trehalose which achieved a score of -4.5. CPX was selected for further studies and parameterised using MARTINI force field. The addition of either CPX or trehalose changes the binding conformation and reduce the binding energies in the MD simulations. A reduction of self-association with KD analysis between mAb 1 molecules with the proposed excipient CPX was observed experimentally compared with mAb1 alone.
Conclusion: We have demonstrated in silico the discovery of a specific novel excipient against APR interactions in a mAb. Given the high similarity in the structure of antibodies, development of APR-customised virtual screening methods coupled with MD stimulations can potentially be beneficial in the design of novel formulations of biopharmaceuticals.
Steve Brocchini– Professor of Pharmaceutics, Department of Pharmaceutics, UCL School of Pharmacy and National Institute for Health Research Biomedical Research Centre, Moorfields Eye Hospital, LONDON, England
Christopher van der Walle– MedImmune Inc.
Ajoy Velayudhan– Professor, University College London