[gmx-users] Binding energy of membrane protein to the membrane lipids

Justin Lemkul jalemkul at vt.edu
Sun Feb 23 18:13:58 CET 2014

On 2/23/14, 11:54 AM, sojovictor wrote:
> Thanks, Justin! That's really helpful indeed.
> You are correct, I want to know what's the energy change in replacing one
> lipid with the other, hypothesising that going to the "wrong" lipid will
> imply an energetic cost.
> Via umbrella sampling, I would thus:
> 1) Set up a full system with protein, lipids, solvent, and ions.
> 2) Pull the protein perpendicularly out of the membrane and into the solvent
> (which means I'd need a lot of solvent in the direction of the pull, such
> that there's always enough space to fit the protein and still avoid boundary
> interactions, as you explain in you tutorial, thanks!).
> 3) Do the exact same thing with the other system, independently.
> 4) See just by eye what the difference is and hope it matches my prediction.

I don't understand what "by eye" means here.  You'll get a free energy 
difference between the embedded and solvated states, i.e. the binding/insertion 
energy.  There's nothing qualitative about that.  That's deltaG for a particular 
lipid type, and the deltadeltaG is simply the difference between the two.

> Now, my question about using umbrella sampling for this purpose: is it
> reasonable to use the whole membrane (say, the group of lipids) as my
> immobile reference? In your tutorial you use a chain, in which all members
> are covalently bound to each other, so I didn't know whether you could use a
> group of independent molecules instead. I assume it will be fine, then.

Don't assume from the tutorial that you need an immobile reference.  The fact 
is, you don't.  You need to define a sensible reaction coordinate that describes 
embedded and water-solvated states.  Neither of those intrinsically requires any 
sort of position restraint.

What I had was a very specific case, and if you read my paper from which the 
tutorial was derived, you will find that the restraints there are used for a 
special purpose to mimic fibril stability.  That's not the case in most umbrella 
sampling runs.  During the generation of configurations, you may need some 
restraints to prevent perturbation of the bilayer structure, but that depends on 
how you generate those configurations.  A more pertinent example would be 



Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441


More information about the gromacs.org_gmx-users mailing list