Dear Bruno,

if you really want to elucidate the effect of mutations on enzyme activities or substrate binding behavior, you should always perform an in-depth analysis and comparison of the wild type and mutant structures.

First of all you need a good structure of your wild type or of any mutant as starting point from the pdb, optimally as a complex structure with bound substrate or product. If a crystallographic structure is not available use the structural homology search tools from the pdb to find another enzyme with the closest structural and sequence similarities to your target protein.

Now use homology modeling (e.g.SWISSmodel via expasy) to derive structures for your mutants. Alternatively you can also directly change the respective residues using programs such as coot, ucsf chimera or pymol. If you have some experience in structural biology, you can also have a look into the electron density files that can be fetched from the pdb, as they often give valuable hints how a mutated residue might be oriented.

Finally you have to perform a structural alignment of all your structures to see the possible effects of the mutations. There are several questions you should answer for each mutation to predict its influence on enzyme activity.

- Does the mutation change the active site regarding shape, hydrophobicity, polarity, charge size or flexibility ?

- Does the mutation change catalytically important residues?

- Does the mutation change the enzymes flexibility?

- Does the mutation change the hydrogen bonding and water network around the active site?

Keep in mind that even residues with considerable distance from the active site can change an enzymes properties, e.g. via hydrogen bonding network or flexibility (I.e. loop regions connecting two helices can be locked nicely via proline or ionic interactions with other residues).

In the case of beta-lactamases I solved the crystal structures of VIM-31, one of the many different VIM variants and only two mutations had important influence on the active site by disrupting hydrogen bonds or even by introducing inhibitory metal binding site effects on the active site.

The quintessence of this long post is, don't trust the result of such mutation analysis servers too much. They can provide valuable hints to find interesting mutation sites but they cannot replace your deep and fundamental investigation of the structural effects a mutation can result in. Use your scientific competence and structural knowledge and you will get valuable and robust information a server can never provide.

Thanks Konrad!