Conclusion
In conclusion, our data shows that computational modeling can be used to
gain meaningful insight into structural properties of an antibody that
is critical for process development. This high through-put structural
information can be used as an additional parameter to select antibodies
that will lead to successful development downstream. While the overall
structural features are important, a better mechanistic understanding of
protein-chromatographic resin is gained by coupling in silicodocking studies to molecular dynamics and experimental data. We
demonstrated using this approach that different regions of an antibody
contribute at different levels to the overall column retention and
selectivity. This is because different high affinity binding sites were
observed base on the ligand density of the agarose-ligand complex.
Further, we addressed the impact of ligand density on overall
protein-resin binding and how the attached ligands to the resin
compensate to maximize its interaction with the protein. Higher overall
binding affinities are achieved at higher ligand densities by a
cumulative effect of lower individual binding affinities (avidity)
(Figure 4). This computational approach to evaluating protein ligand
interaction has broader implications for biologics development (from
lead candidate selection through purification to formulation) and
commercial chromatographic resin design and head groups selection and
optimization.