OPTIMIZATION AND APPLICATION OF COMPUTATIONAL METHODS FOR THE DESIGN OF PROTEIN-PROTEIN INTERACTIONS MODULATORS

In the wide field of PPIs, this PhD project has been focused on the optimization and application of computational methods for the design of PPIs modulators, with a particular interest toward peptide modulators targeting PPIs involving helical motifs. In this contest, the first part of the project has been aimed to define the rationales behind the helical secondary structure stabilization and the helical screw sense selectivity exerted by chiral Cα-tetrasubstituted amino acids (cCTAAs) through REMD simulations and QTAIM analyses, and the mechanisms responsible of the helical screw sense inversion through PNEB simulations. In detail, it has been found that the helical motif is stabilized by two complementary mechanisms: the first depends on the steric hindrance exerted by the cCTAA in an area parallel to the peptide helix axis and downstream of the cCTAA itself, whereas the second consists in the strengthening of the helical H-bond network thanks to peculiar C-H···O=C interactions. Analogously, P-helical screw sense selectivity is ascribable to the cCTAA steric hindrance parallel to the peptide helix axis, without particular preferences for the region downstream and upstream of the cCTAA, together with quite strong noncovalent interactions, consisting of classical N – H···O=C H-bonds and weak C – H···O=C interactions. Furthermore, PNEB simulations performed on achiral peptides of different lengths suggest that the helical screw sense inversion requires the formation of γ-turns, although a preferential screw sense inversion direction was not found. Therefore, the knowledge gained from these studies could be helpful in designing stable helical peptides, having a preferential screw sense and that can be in principle activated in situ by inducing a conformational switch from P to M helix or vice versa. Conversely, the second part of the project has been focused on the optimization of an MMGBSA based method, called Nwat-MMGBSA, aimed to improve the correlation between predicted binding energies of PPI complexes and experimental data. This approach, consisting in the inclusion, as part of the receptor, of hydration shells around the ligand during the MMGBSA calculations, was initially tested on classical receptor-ligand complexes and, then, automatized, optimized and tested on PPI complexes. This approach turned out to be good for the evaluation of PPI modulators activities, from different points of view. First of all, when water played a significant role in mediating protein-ligand interactions, the application of Nwat-MMGBSA improved the correlation between predicted and experimental data. On the other hand, if the solvent does not explicitly participate to the interaction, it did not give detrimental results compared to those obtained with the standard approach. In addition, the protocol proved to be robust and reproducible, giving equivalent results by using different setups. Furthermore, although an optimal number of water molecules to include in the hydration shell could not be found, in the case of PPI interactions inhibited by small molecules the inclusion of 50 – 60 water molecules appears to be a good choice. A non-negligible advantage of this approach is represented by the possibility to automatize it, making it applicable for drug design/discovery purposes. Therefore, although further evaluations are needed, most of all on larger datasets, the knowledge coming from the combination of both parts of the project can be exploited for the design of stable non-natural peptides targeting PPIs.