loading page

In Silico Process Development via Computational Modeling: Insights into Molecular Biophysics to Advance and Improve Biologics Purification
  • +2
  • Francis Insaidoo,
  • John Welsh,
  • Karol Lacki,
  • Thomas Linden,
  • David Roush
Francis Insaidoo
Merck & Co Inc
Author Profile
John Welsh
Merck & Co Inc
Author Profile
Karol Lacki
Avitide, Inc
Author Profile
Thomas Linden
Takeda Bio Development Center Ltd
Author Profile
David Roush
Merck & Co Inc
Author Profile


The goal of this research is to leverage computational molecular biophysics to guide process development, reduce experimental burden and focus purification activities on feasible targets. Here, we distill a complex separation problem (e.g. chromatographic retention of monoclonal antibodies) into a tangible model (ligand/protein complex), which is computationally feasible while preserving enough detail (atomistic level for interaction site) to support industrially relevant separation challenges. Computational docking, coupled with molecular dynamics simulation, produces results that are directionally consistent with chromatography for proteins (mAb). This approach is generalizable and can be applied to a range of ligands (AEX, CEX, and Mixed Mode). A detailed model of the chromatography base matrix (agarose) was constructed to obtain a biophysical understanding of potential protein/base matrix interactions. The base matrix was then modified in silico with ligands over a range of ligand densities representative of commercial chromatography resins to generate an agarose/ligand complex. A generic approach was developed to model the impact of avidity and ligand density on mAb/ligand interaction. The results revealed that increasing ligand density mask contributions of base matrix binding. Increasing the number of ligands that can interact with mAb results in more favorable free energy of binding or ΔG (more negative) with a limited incremental increase in ΔG by increasing N (number of ligands per agarose cluster) above three. Additionally, for protein/ligand interactions at each binding site, not all ligands contribute equally to the binding affinities or interaction energies and a redistribution of binding interactions/energies occur as N increases. These observations yield insights into the impact of avidity on retention (macroscopic affinity measurement via k’). The generic approach described in this manuscript can be leveraged to inform resin selection and design as well as targeted ligand selection/purification development in a rational manner.

Peer review status:IN REVISION

22 May 2020Submitted to Biotechnology and Bioengineering
23 May 2020Assigned to Editor
23 May 2020Submission Checks Completed
26 May 2020Reviewer(s) Assigned
20 Jun 2020Review(s) Completed, Editorial Evaluation Pending
20 Jun 2020Editorial Decision: Revise Major