I am planning to calculate binding affinities for a set of ligands to my protein of interest, both the wild type as well as mutated form. The crystal structure for the particular conformation of the protein with required mutations are not available. I am confused as to which one should be my starting structure:
1. the ligand is docked using Autodock to my protein and the structure having the most stable energy is chosen (it may not correspond to the same binding pose as in the crystal structure of the same ligand to the protein of interest in some other mutation)
2. Superimpose the required protein to the available crystal structure and select the inhibitor binding pose (using say, PyMol).
On proper equilibration and MD production run for ~50ns, I will calculate the binding energies for the ligands to the protein. Can I expect similar results from both cases? If not, which one should be the correct starting structure?
Thanks in advance for your help.
There's not good answer to your question. There are instances where a single-point mutation changes the conformation of the entire protein, as well as changes the ligand binding mode and there are instances where multiple mutations do not change much in the conformation of the protein.
If the mutation(s) are far from the ligand binding site: do the MD of the wild-type and mutant protein and check if there are any large differences, particularly in the binding site. If there aren't, you can probably justify superimposing the ligand from the crystal structure of the wild-type.
If there are changes or if the mutation(s) are within the binding site: you should probably take the docked ligand poses - but make sure that the docking is done correctly.
As a sanity check: make sure, that Autodock can recreate the pose from the wild-type crystal structure.
i dont think so. results would certainly varry in both the cases.
I suggest you to first check for the variation in wild type and mutant. if deviations are large than going with superimposing method seems to be quite logical.
There's not good answer to your question. There are instances where a single-point mutation changes the conformation of the entire protein, as well as changes the ligand binding mode and there are instances where multiple mutations do not change much in the conformation of the protein.
If the mutation(s) are far from the ligand binding site: do the MD of the wild-type and mutant protein and check if there are any large differences, particularly in the binding site. If there aren't, you can probably justify superimposing the ligand from the crystal structure of the wild-type.
If there are changes or if the mutation(s) are within the binding site: you should probably take the docked ligand poses - but make sure that the docking is done correctly.
As a sanity check: make sure, that Autodock can recreate the pose from the wild-type crystal structure.
You could try both. Anyhow, when you run MD simulations you can run several trajectories of each pose that you want to try. They might converge to the same average structure and then that will be a good confirmation of the stability of your pose. If you want to calculate binding affinities that will be the important structure for you.
You might get inspired from this paper:
http://pubs.acs.org/doi/abs/10.1021/ci500151f
Just to add to the previous comments: free energy calculations are very problematic and, even methods such as FEP depend on the magnitude of conformational differences between the states, so there is no reliable method for all cases. And if you have only two different systems to study (wild type and mutated), you won't be able to trace a reliable correlation. However while the prediction of affinity is problematic, the observation of associated conformational events is not (considering a sufficiently large sampling and proper force field). So a safer way to understand your system pass through conformational modulation instead of affinity. Also, take special care on the atomic charges you use on both redocking the complexed inhibitor and docking your compound, as well as on the choose of the degrees of freedom to be sampled. Both can interfere in your result.
Start with a crystal structure of a protein with close structure to your protein. Change the structure to your desired structure with a convenient software. Then you can optimize it using a QM method.
- Badges
- Science topic