Are 3D Structures Of Proteins Similar In An Inhibitor Bound State And Unbound State

Question

[Will]

I'm trying to compare the 3D distance between two AAs with their evolutionary conservation. However, I'm coming across a problem when looking for 3D crystal structures of HIV-1 proteins ... mainly that the majority of crystal structures of proteins are from experiments where the protein of interest is bound with active site inhibitors. Here are some examples for reference:

http://www.pdb.org/pdb/explore/explore.do?structureId=1S6Q

http://www.pdb.org/pdb/explore/explore.do?structureId=1S1X

http://www.pdb.org/pdb/explore/explore.do?structureId=1S1W

I'm not incredibly familiar with 3D structures but I do know that these are small-molecule inhibitors. Can I assume the majority of the structure is unchanged? Should I take many different structures of the protein (with different SMIs) and average them somehow?

I know that all of these methods will add plenty of noise to the distance measurements but I'm just looking for a "back of the envelope" calculation for an initial hypothesis test. Do you think this will add too much noise to the see any signal? If so, can you propose another idea?

[Egon Willighagen]

I have asked on the Blue Obelisk eXchange how Jmol can visualize the differences.

Answer 1

[Khader Shameer]

To answer your question: 3D structures of proteins are not identical in an inhibitor bound state and unbound state - they will have conformational differences which can be measured using standard Root Mean Square Deviation (RMSD) calculation tools (see: Protein Structural Alignment at Wikipedia).

In your specific case, These structures are co-crystallized with the small molecules and they will definitely have conformational changes due to the binding of the small-molecules. You may check in PDB and try to find a unbound form of your protein of interest and perform an RMSD calculation using SuperPose or similar RMSD calculation tools to measure the distance.

It is not clear to me what you are trying to achieve by averaging the co-crystallized structures.

[Kraut]

Will, you could try removing the ligand form the bound structure and 'relaxing' it with MD or a kinetic framework like Rosetta++.

[Will]

The problem is that there are many HIV proteins where there are NO UNBOUND examples ... I'd be limited to 1 of 15. I'm wondering if the conformational changes are drastic or minimal. Are the conformational changes local to the binding site or global?

The idea of averaging/aligning is that if there are many different examples of BOUND structures, and the conformational changes are uncorrelated, can I get an approximation of the UNBOUND state.

[Khader Shameer]

I won't be able to tell you the effect of the conformational changes with out further analysis/data. To get a rough idea to understand the effect you may do superposition/RMSD of your protein of interest in its bound and unbound form. To understand the exact effect of binding (global and local) you may need further analysis of the structure (for example you could try a Root Mean Squar Fluctuation (RMSF) analysis using GROMACS to see the exact of binding.

[Will]

any suggestions for the proteins where there are no unbound examples? For just a "back of the envelope" type calculation?

Answer 2

[rdk]

You can use http://apoholo.projekty.ms.mff.cuni.cz/ to visualize the differences between Apo and Holo protein forms.

[Mensur Dlakic]

Thank you for this useful resource. If you are the software author, as seems to be the case here, it is customary to say so as a way of disclosing your bias.