Designing an enzyme able to catalyse a particular chemical reaction is a huge challenge. Different strategies adopted for this task have been developed in the last years. De novo design is one of these, where a
protein without specific catalytic properties is used as a scaffold for an active site design from scratch.
The work contained in this dissertation is focused on the theoretical investigation of the multi-step Retro-Aldol reaction mechanism of 4-hydroxy-4-(6-methoxy-2- naphthyl)-2-butanone (methodol) catalysed by three different previously published protein scaffolds; two de novo enzymes, RA95.5-8F and RA95.5-5, and one catalytic antibody, 33F12. Using different computational methods based on the use of multiscale QM/MM potentials, the first part of the study has aimed to understand in detail their reaction mechanism in terms of geometries and free energy landscape. These studies, and their comparison, allow us to identify the roles played by the amino acids around the different reaction sites, their interaction with the species involved in the reaction, and how they favour or not the stabilization of the transition states of the different chemical steps. The information derived from these studies has been exploited in the identification of specific amino acids that could be mutated in one of these protein scaffolds, the de novo enzyme RA95.5-8F, in order to improve its catalytic activity. Due to the multistep character of the reaction to be catalysed, this is a challenging goal since the activation free energy of the rate-determining step (RDS) of the reaction must be reduced but without harming the other chemical steps. As shown in the last part of this thesis, our investigation has successfully led to the proposal of a new catalytic protein for the catalysis of the Retro-Aldol reaction, with higher catalytic efficiency compared to the most efficient RA95.5-8F. This new strategy can significantly support and accelerate the experimental works on the design of new enzymes.
