[gmx-users] questions about long-range electrostatic interaction in membrane systems

Jianguo Li ljggmx at yahoo.com.sg
Mon Dec 6 09:16:21 CET 2010


Dear All,

I am working on the interaction of a positively charged peptide (18AA)  and the 
bacterial membrane consisting of a mixture of zitterionic and  negatively 
charged lipids. Our experiments show that the peptide can  disrutp the membrane.
One problem puzzed me is that which method to use in treating the  long-range 
electrostatic interactions for my system. The following is my  understanding, 
please correct me if I am wrong.

There are two options for treating long-range electrostatic interactions for my 
system:
(a) use a normal 3-D PME (by setting ewald_geometry=3d, which is the  defaut 
option); and (b) use a slab PME in 2-D by setting  ewald_geometry=3dc (thanks 
for David Bostick's suggestions)

If I use normal PME in 3-D (ewald_geometry=3d), and put the peptide on  top of 
the membrane, I cannot increase the peptide concentration. For  example, if I 
put 3 peptide on top of the membrane, 1 peptide will move  to the lower leaflet 
of top image box. I believe this is because of the  repulsion between the 
peptide and also the attraction between the  peptide and the lower leaflet of 
the membrane in the top image box. Also as mentioned in the previous posts, 3D 
PME has artifact (i.e., due to the dipole moment of the simulation box which is 
significant as pointed out by David Bostick through emails since I have a 
positively charged peptide and a negatively charged membrane).

When using 2D PME with slab geometry (ewald_geometry=3dc), the membrane is 
treated as an infinite charged slab, the  positively charged peptide is not only 
interacts with the membrane in  the central box, but also interacts with the 
membrane in the image  boxes. Although the interaction between the peptide and 
the membrane in  the image boxes far from the central box is very weak, but the 
overal  sum of all these interactions could be very strong since there are  
infinite number of image boxes. Therefore it will over-estimate the intearction 
betwen the peptide and the membrane. 


In fact, I got very different results when setting ewald_geometry being  3d or 
3dc. Using ewald_geometry=3d, nothing happens except a minor  deformation of the 
membrane in the vicinity of the peptide. However,  when using 
ewald_geometry=3dc, the peptide readily induce a water pore  and penetrate into 
the membrane. I am not sure if the results using 2d  PME is correct or not, as 
it has some artifact for my system.

I am wondering how big the artifact is due to the infinity of the 2D PME  for 
slab geometry. Any suggestions for modeling the finite size of the  cell 
membrane are appreciated!

Another question is about the parameter epsilong_surface in Gromacs which is the 
dipole correction of Ewald summation. Is it suitable to use this parameter in my 
simulations? If possible, which value I need to use for this parameter?

Sorry for so many questions, thanks for your time!  

best regards,
Jianguo
Postdoc Research Fellow
Bioinformatics Institute and Singapore Eye Research Institute, Singapore

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20101206/fe963658/attachment.html>


More information about the gromacs.org_gmx-users mailing list