Reply To: Modeling phosphate ion binding site in protein.

[Anonymous]

So I’m assuming that you’re doing the enzyme design in a scaffold protein that already has the cysteine and aspartate amino acids in the appropriate sequence location, and that your docking trials from Autodock have given you a reasonable location for the location of phosphate.

One issue you’re going to run up against is that your reaction involves a change in geometry around the phosphate. Although the substrate, covalent intermediate and product are tetrahedral (although with a Walden inversion between the substrate/product and covalent intermediate), the transition state is trigonal bipyramidal. If you go by the transition state theory, the role of an enzyme is to maximize stabilization of the transition state, so when doing enzyme design, you’ll want to account for the transition state geometry (Published enzyme designs with Rosetta have mainly worked off of the “theozyme” concept from the Ken Houk lab – this includes not just the transition state itself, but also the geometry of the surrounding amino acids.) … Although you don’t want to only model the transition state, as the enzyme also has to accommodate the substrates, the products, and the intermediates. Usually this is handled by designing toward a multi-state hybrid molecule.

When water has been an important part of the design (e.g. the retroaldolases), the water has been explicitly incorporated into the transition state hybrid with “dummy” bonds. In your case, I’m not sure if you’ll need to model a separate water molecule, as the water in the second step is just performing the reverse of the tyrosine oxygen in the first step (that is, the tyrosine oxygen and the water oxygen will probably occupy the same position in the two halves).

My recommendation is to make a transition state/substrate/product hybrid structure which incorporates all the key features you want to accommodate/stabilize in your enzyme, and then decide what the permissible geometries of the interactions to the cys & asp are. From that you’ll create your ligand structure and constraint files (don’t forget to mark the cys-phosphate interaction as covalent) to use in the enzyme design application. You then need to prep your starting structure – align your hybrid molecule in the appropriate starting location (the key thing is to get the constrained geometry correct – don’t worry too much about clashes with sidechains, as the enzyme design application should fix those) and place the appropriate remark lines at the top of the PDB, telling which amino acids correspond to which constraints. From there, run the enzyme design application and tweak as necessary to emphasize what you think is important.

By the way, I get the impression that you may be starting from a structure which already catalyzes this or a related reaction. If that’s the case, I would recommend minimizing the amount of alterations you do to the catalytic machinery. Although you should do all of the transition state and constraint modeling to make sure all the appropriate geometry is preserved, focus your attention on the parts of the structure which are involved with the differences between your reaction and the native reaction. (Look into resfiles for limiting which amino acids can change.) One of the things we’ve found is that you tend to get better results in enzyme design if you minimize the number of changes from the native scaffold.