By: Olivier Beyens, Sam Corthaut, Sarah Peeters, Pieter Van Der Veken, Ingrid De Meester, Hans De Winter

Drug discovery involves searching for molecules binding to a protein target, where the modulation of this protein target leads to a biological effect. Drug discovery research can be performed via two general paradigms: ligand-based or structure-based. In ligand-based drug design, new molecules are designed based on the known binding affinity of already tested molecules. In structure-based drug discovery, new molecules are designed based on the structure of the target protein, thus allowing for increased rationalization of the design of new molecules.
We focused on a structure-based approach called “cosolvent molecular dynamics”. In cosolvent molecular dynamics simulations, the structure of the target protein is placed in a virtual box with a variety of common fragments of drug molecules. After adding these fragments, the remainder of the box is filled with water. The molecular dynamics simulation subsequently computes the movement of every atom in this system. By tracking where the fragments move in the target protein, a map can be built of the favorite locations of these fragments within the target protein. If the fragments are chosen in such a way that they represent common interactions, the resulting maps can be used as a guide to design new drug molecules.
We had two objectives when applying these cosolvent molecular dynamics simulations: to further development of the technology and to apply the technique to two protein targets, namely DPP8 and DPP9. In terms of technology development, we built a new tool to help in the setup of cosolvent MD simulations with hydrophobic fragments and we improved an existing protocol to post-process the results. Using these new tools, we computed qualitative DPP8 and DPP9 fragment affinity maps (example: Figure 1). We have made these fragment affinity maps freely available to other researchers, further enabling the rational design of new DPP8 and DPP9 inhibitors.

These molecular dynamics simulations can become computationally demanding, especially for long run times. Due to the high computational requirements, access to the resources provided on the VSC and LUMI high-performance computing systems was crucial for benchmarking our technological developments and for computing qualitative fragment affinity maps for DPP8 and DPP9. As both the new setup tool for hydrophobic fragments and the fragment affinity maps were made freely available, the provided computational resources have enabled further research in both the field of cosolvent molecular dynamics and in the field of DPP8 and DPP9 drug design.
Read the full article in ACS Publications here