Unraveling the binding mechanism of macrocycle peptides to PD-L1 through computational approaches

3 Abstract PD-1, and its ligand PD-L1, represent a well-known immune checkpoint involved in the silencing of T-cells in the tumor environment. For this reason, they are the target of several mAb that are clinically used for cancer treatment with extraordinary results in some cases. Small molecule inhibitors of PD-L1 are under investigation as well, but they have been demonstrated to cause the dimerization of PD-L1. In the present work, we focused on peptide macrocycles that combine the specificity of mAb with smaller dimensions, better bioavailability, and lower production costs. In the attempt to understand the leading mechanism driving the binding of the known macrocycles to PD-L1, we focused on co-crystallized macrocycles (PDB IDs: 6PV9 and 5O4Y). These two ligands differ for just one residue (serine and sarcosine) but this difference accounts for an activity gap of two orders of magnitude (pIC50 8.79 and 6.24, respectively). As the analysis of crystallographic binding geometry does not provide explanations, we carried out a 500 ns molecular dynamics simulation on both complexes and the PD-L1 apo-form, aimed to get more insight into the binding process. The MD simulation revealed a different behavior of the two peptides: the most active resulted stable while the less active detaches from the target macromolecule maintaining a hydrophobic interaction with PD-L1 Tyr123. Interestingly, the same site was also detected by the analysis carried out with TRAPP (TRAnsient Pockets in Proteins), indicating it as a relevant hot spot to be exploited in the PD-L1 ligand design.