[gmx-users] Re: DPPC Bilayer Size

[chris.neale at utoronto.ca]

> I am trying to simulate a GPCR protein belonging to the rhodopsin  
> family with amino acid length of 320 residues. Also, I would be  
> using the pre-hydrated DPPC bilayer, can somebody suggest that for a  
> GPCR protein of
> roughly 320 aa, how big DPPC bilayer would be appropriate to use  
> e.g. 128 DPPC bilayer is available from Peter Tieleman’s   
> website http://moose.bio.ucalgary.ca/index.php?page=Main .
It depends what you are studying I suppose: the lipid or the protein.  
Certainly you need enough lipid so that one side of the protein  
doesn't see the other side. An argument could also be made that you  
don't want any lipids to see both sides of the protein, although I am  
not sure that this is so essential if your focus is on protein  
behaviour.

As a rough guess, assuming that your protein is roughly globular, I  
imagine that you should probably use about 300 lipids per leaflet  
prior to making the hole. There is no way that I can really know, but  
just in case you have desired this kind of answer...  For you to make  
this estimate yourself, measure the largest diameter in the xy plane  
(tilted as you expect it to be during the simulation), add 1nm for  
possible side chain extension and 1.4nm for your rvdw (or whatever you  
use) and you have a box width that would be the recommended minimum.  
Compare that to the equilibrated membrane box that you have and get a  
rough number of lipids/leaflet. However, you will need to be  
monitoring your run to see that the protein doesn't get bigger in the  
xy by more than 1nm. I personally like to make the system much too  
large at first, equilibrate using 3 or 4 cpus in parallel for about  
5ns, then analyze to see the distance fluctuations. This allows me to  
make a safe guess about the minimal box size.

With a system so large, technical details like rvdw vs. box size  
probably won't be an issue. That leaves considerations based on  
equilibration and cpu hours/ns.

1. There is some information available that a small mass ratio of  
lipid/protein can lead to reduced ability of the system to reach  
equilibrium:

Molecular Dynamics Simulations of Model Trans-Membrane Peptides in  
Lipid Bilayers: A Systematic Investigation of Hydrophobic Mismatch
Senthil K. Kandasamy and Ronald G. Larson
Biophysical journal 2006

This means that although a smaller system will complete more steps in  
a given period of cpu time, it is not necessarily true that a smaller  
system will equilibrate faster in a given period of cpu time.

2. If I remember my GPCR biology correctly, they have 7 TMhelicies?  
That's a huge number of lipids to remove. You want to either be very  
carefull and test out a couple of methods to remove the lipids (the  
make_hole version of gromacs or tieleman's groups new  
expansion/contraction method for example) or have a membrane big  
enough that if you remove too make lipids from one leaflet, it won't  
put undue strain on the other leaflet by density values (or you could  
use the p2sub1 crystal PBC in CHARMM, which I have tried  
unsuccessfully to implement in gromacs).

3. The size of the required membrane will be minimal when the  
hydrophobic mismatch is minimal, so getting the correct membrane  
environment is important.

4. Don't realy on anything that I have said. It's the best that I can  
do, but really it's just a suggestion.

Chris

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.