I built a homology model for HDAC protein, and wanted to run MD simulation for it with a particular docked ligand using NAMD. I prepared the input files and parameterized the ligand and got the topology files from CHARMM GUI. However, after two attempts for running for 100ns, I found out that ligands leave the binding pocket during the run. The active site is not deep in the protein and ligands bind in an exposed mode to the surface.
- What possibly could be the reason behind that?
- Is it scientifically accepted to apply distance constraints since the purpose is it to assess the stability of the whole complex?
This observation is not at all surprising. The interplay between entropy (TS) and enthalpy (H) governs this. When it is bound to the protein it loses entropy but gains enthalpy. Whereas in the bulk it loses some enthalpy and gains a lot of entropy. Now between H and TS which one wins is decided by the interaction parameters in the force field.
I would suggest you to first get an estimation of the free energy barrier of ligand dissociation from simulations (using same parameters). The barrier height shall tell you the lifetime of bound complex.
You can try another force field.
Thanks
I do not particularly have experience with NAMD. However, I would suggest you to first check your protein-ligand complex forcefields. Did you check your ligand RMSD or ligand-protein distance (please make index files and check that). Also, visualisation might indicate that the ligand is moving away, however, in reality this might be just because of periodic boundary conditions in the movie file. All the best.
If you are starting from a docking pose you should consider the additional conformations in the top five ranks. Also, I suggest you start with shorter runs e.g., 20-25 ns to assess the relative stability of the pose before going further. Additionally you should check the complex during the minimization and equilibration stages. If the starting pose is reasonable enough, it shouldn't change that much during this process.
Just as Pallab Kumar Borah said, a quick glance at RMSD or ligand distance to protein gives you a hint on what's going on.
For further reference, you could check this paper:
10.1021/acs.jcim.7b00412
This observation is not at all surprising. The interplay between entropy (TS) and enthalpy (H) governs this. When it is bound to the protein it loses entropy but gains enthalpy. Whereas in the bulk it loses some enthalpy and gains a lot of entropy. Now between H and TS which one wins is decided by the interaction parameters in the force field.
I would suggest you to first get an estimation of the free energy barrier of ligand dissociation from simulations (using same parameters). The barrier height shall tell you the lifetime of bound complex.
You can try another force field.
Thanks
Pallab Kumar Borah , Fernando Prieto-Martínez , Sayantan Mondal Thank you all for your precious replies and for the invaluable suggestions. I appreciate your help.
i would start here: Homology modeling and molecular docking failed to reproduce the real structure of the protein and protein-ligand complex, respectively.
Other potential problems: ligand parametrization (especially the partial atomic charges), not taking into account long-range electrostatic interactions, equilibrating MD protocol (heating gradient, timestep gradient, bath-coupling gradient).
And last but not least: the longer the MD simulation the higher the likelihood of the system drifting from reality to fiction.
---
To assess the protein-ligand complex stability: calculate the protein-ligand binding free energy (e.g., using linear interaction energy or linear response approximation methods); and for this you need two sets of MD simulations (generated by varying the initial conditions, e.g., the assignmet of initial velocities): (1) protein-ligand complex in water and (2) free ligand in water. If the ligand prefers the bulk water over the protein binding site then you have a big problem . . . (probably with the structures).
Martin Klvana Thank you so much for your answer. Actually what I did so far was like this, a verified Homology modeling, followed by molecular docking several known inhibitors to assess the validity of the protein structure.
The homology modeled structure (free) was submitted to MD simulation. After the system reached the stability, the protein was used for molecular docking with several known inhibitors and reproduced protein-ligand complexes.
I use ParamChem CGenFF server to parameterize ligands.
For some reason, after two attempts, the ligands failed to remain in the active site.
- Badges
- Science topic
- Similar topics
- Filing