In the perspective of using enzymes as biocatalysts to produce non-natural compounds, the starting substrates are usually non-natural as well. Since the enzyme-substrate interactions are not naturally optimized in such cases, it is expected to be much room for the improvement of the catalytic efficiency. Identifying the regions of the enzyme structure most sensible to the binding and reactivity of a given non-natural substrate is crucial for the redesign of the enzyme.
We will present here a computational algorithm (BindScan) that exhaustively casts all the positions on a given protein structure by individually mutating each position and measuring the effect on the binding affinity and reactivity to a given compound. The positions of those mutants showing an improvement of any of these metrics with respect to the wild-type enzyme are considered as “hot-spots” sensible to the binding or reactivity of the new compound. This information can then be used to experimentally design single point mutations or to guide directed evolution experiments for the improvement of substrate specificity in the working enzyme.
Different benchmarks of the algorithm on CAZymes will be shown based on mutational data already published for different glycosidases and transglycosidases. The algorithm can further be applied to other families of enzymes and receptors in which there is a need to tune molecular recognition and/or reactivity.