I want to do an MD simulation to see the transition of a ligand molecule between two hosts. My first host is a cyclodextrin (Drug carrier), the second host is Albumin. I would like to see if the drug will leave the cyclodextrin carrier easily and go to bind the Albumin. I want to compare the PMF between two different drugs. I know that I may simulate each system alone and calculate PMF with respect to the dissociation from cyclodextrin and the association to Albumin, but, do you think it is reasonable to do the two simulation simultaneously and trace the PMF to see which is easier to happen. I have looked at the literature and i did not find something similar, perhaps I am missing something here. Do you have any suggestions for such simulations. I am comfortable to work with AMBER. Do you think that steered MD will be a better option than Adaptive biasing force for such simulation. Any recommendation will be appreciated.
Thanks in advance.
Marawan
My expertise is actually in molecular modelling based on experimental NMR data.
That said, I feel that your problem should definitely be treated as two separate bat similar problem. You need to know the dissociation energy of your drug molecule with both albumin and cyclodextrin.
I would start by finding the best binding modes for each of those complexes and try to force their dissociation. Probably steered MD would be the most straight forward for this purpose. I think you should go for dissociation for both complexes as it will be easier to set up and you can start from the correct binding mode. You will get more headaches if you try to model the inverse association reaction.
Of course the real complication comes when realising that this happens in solution and not in vacuum and there is a competition between your drug molecule and water to bind the pockets. I am no expert in how you would treat energies associated with solvation, so you might want to ask the community.
Anyway I would be sceptic about the numbers you will get for such complex problems and would suggest experimental evidence to back it up. Definitely try using a variety of force fields and parameters to get a feel about how much these energies vary.
One easy way would be to calculate the PMF using Umbrella Sampling along a direction equivalent to removing the ligand molecule from the pocket. You could do this for the two molecules seperately and compare the PMFs.
You could also use a SMD method such as the FR method to calculate the PMF. Using the FR method would also give you a diffusion coefficient for entering the pocket.
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/umbrella/index.html
FR method
Article Calculating potentials of mean force and diffusion coefficie...
Another alternative is to compute the PMF using the (MW)^2-XDOS [Multiple-Walker Multiple-Window eXpanded ensemble Density Of States], a flat-histogram Monte Carlo approach [L. Janosi, M. Doxastakis, J. Chem. Phys. 131, 054105 (2009)] based on the density of states algorithm proposed by Wang and Landau [Phys. Rev. Lett. 86, 2050 (2001)]. It is a fully parallelized method that accelerates flat-histogram PMF calculations by various approaches (e.g., preferential sampling Monte Carlo) and can easily be applied to calculate PMFs in difficult biological systems (like association in membranes):
(1) association of GpA [L. Janosi, A. Prakash, M. Doxastakis, Biophys. J. 99 (1), 284 (2010)];
(2) association of ErbBs [A. Prakash, L. Janosi, M. Doxastakis, Biophys. J. 99 (11), 3657 (2010), A. Prakash, L. Janosi, M. Doxastakis, Biophys. J. 101(8), 1949 (2011)].
NAMD + adaptive biasing force will work for this. It's well developed, and will give you better converged answers than steered molecular dynamics.
http://www.ks.uiuc.edu/Research/namd/2.6b1/ug/node34.html
http://www.ks.uiuc.edu/~char/tutorials/ABF/
- Badges
- Science topic