COMPUTATIONAL MODELLING OF BIOMOLECULAR SYSTEMS: APPLICATIONS TO THE STUDY OF MOLECULAR RECOGNITION PROCESSES

In this thesis a broad range of computational modelling techniques has been used to study biomolecular systems. In particular two main topics have been addressed: (i) the conformational analysis of unnatural glycopeptides, in which Monte Carlo methods have been used to sample different molecule conformations in order to analyze and characterize their structural, dynamical and functional properties and (ii) the analysis of protein-protein interactions (PPIs) in classical cadherins and the design of small peptidomimetic inhibitors. Here, molecular dynamics techniques (both biased and unbiased) together with computer-aided drug design were used to study the structural properties and the mechanism of the cadherin homophilic binding and to design the first class of small molecule inhibitors of their interaction. The conformational studies performed on a series of α-N-linked glycopeptides revealed their preference for an extended conformation, a feature which was later confirmed by NMR data, thus demonstrating the predictive ability of the calculations. PPIs studies on classical cadherins led to the development of the first rationally designed inhibitors of the homophilic cadherin-cadherin interaction and also allowed to gain more insight into the cadherin binding mechanism.