Computational design and experimental characterization of metallopeptides as proteases for bioengineering applications

Enzymes are highly versatile catalysts present in biological systems and with high technological potential. Zinc metalloproteases are a major class of enzymes currently being employed in e.g. food, detergent, biopharmaceutical industries. In order to increase their robustness and range of biological and technological applications, metalloenzymes can be redesigned by exploring the chemical versatility of different metals along with protein sequence modifications. In this work, the computational design of new zinc metalloproteases was approached to test if proteolytic activity can be recapitulated in small scaffolds tailored for bioengineering applications. Structural and dynamical aspects of metalloproteases were first addressed to identify catalytically-relevant interactions. Models of the active site were developed to screen a set of 43 small scaffolds with the Rosetta software for their ability to recapitulate the native enzyme functionality. Two candidate scaffolds were selected for enzyme design and experimental characterization, namely the Sp1 zinc finger 2 and the villin headpiece subdomain. While metal coordination was achieved (binding constants KZnP,app in the 105 M-1 range), the scaffolds presented low stabilities (thermal unfolding bellow 50 °C) most likely due to perturbations introduced by the 4 to 10 sequence modifications. The metallopeptides presented catalytic activity towards ester substrates within the range of values found for other small scaffolds in the literature (second-order rate constants k2 in the 10-1 M-1s-1 range). The design approach developed in this work was successful in achieving catalytically-active metallopeptides, although target metalloprotease activity could not be achieved. Molecular dynamics simulations in microsecond regimes were subsequently used to detect design flaws related with high scaffold flexibility. This work contributes to the improvement of the computational enzyme design approaches by pointing out the need for a dynamical treatment of the designs in longer time-scales, and through the development of fast methods to rank and evaluate a large number of potential biocatalysts.