1. Coarse-grain modeling of lipid membranes
Mario Orsi
University of Southampton
LAMMPS workshop, Albuquerque, 9 August 2011 1

3. Membrane modeling Atomistic models: Coarse-Grain (CG) models:
Water
Lipid
bilayer
Water
Atomistic models:
Accurate but computationally demanding
Coarse-Grain (CG) models:
Orders of magnitude faster to simulate 3

4. CG models of lipids and water Klein et al
CG models of lipids and water Klein et al., J Phys Chem B (2001); Smit et al., J Phys Chem B (2003); Marrink et al., J Phys Chem B (2004, 2007); Izvekov and Voth, J Phys Chem B (2005);
Atomistic CG
Headgroup:
dipole moment ~ 20 D
Ester/glycerol:
dipole moments ~ 2 D
CG Issues:
Incomplete electrostatics
Water: generic apolar fluid
Lennard-Jones combination rules not followed: εAB ≠(εAεB)1/2
Look for atomistic model with partial charges highlighted
H2O:
dipole moment ~ 3 D

5. Our CG model [Orsi et al, J Phys Chem B, 2008]
Explicit electrostatics, relative dielectric constant r=1
Lipid: charges for headgroup and dipoles for glycerol/ester
Water: 1-site “SSD” dipolar model [Liu & Ichiye, J Phys Chem, 1996]
Tails: anisotropic “liquid-crystal” potential [Gay and Berne, J Chem Phys, 1981]
Bespoke MD code developed in-house

6. Membrane properties
J Phys: Condens Matter, 22, (2010)

7. CG/atomistic multiscale simulation
Fine chemical detail for selected molecules
“Dual-resolution” system of mixed granularities
Our CG model compatible with atomistic potentials [Michel et al, J Phys Chem B, 2008; Orsi et al, J Phys Chem B, 2009]
Electrostatics: classical Coulomb formulae 7

8. “Dual-resolution” systems
Drugs & hormones
[Soft Matter, 6, 3797 (2010)]
Antimicrobial compounds
[J R Soc Interface, 8, 826 (2011)]

9. Limitations of our methodology
Force field
Gay-Berne potential (ellipsoidal tails)
Unrealistic interdigitation when simulating solid (“gel”) bilayer phases
Software
Our bespoke code
Not parallel
Not very flexible
Using LAMMPS instead? Not straightforward:
No SSD water
Straight cutoff for dipoles is problematic

10. New “ELBA” force field ELectrostatics-BAsed coarse-graining
Simplified water: LJ + dipole
Lipid tails: LJ
Validated with in-house program
For challenging applications, the power of LAMMPS is needed!

11. LAMMPS “dipole/cut” pair style
Potential
Energy
LAMMPS “dipole/cut” pair style
Straight cutoff: discontinuities in potential and force
Poor energy conservation
Artefacts in the particles’ motion
New shifted-force style (“dipole/sf”)
Modified formulae: potential & force go to zero at the cutoff [Allen & Tildesley, 1987]:
cut sf
Force
Derivative (force)
Derivative (force)
zoomed x10
Force zoomed x10

12. LAMMPS NVE simulation of point dipoles
Shifted-force: better energy conservation + larger timesteps

13. ELBA in LAMMPS: preliminary MD results
Membrane self-assembly
Lipid/water dispersion
Lipid heads: red
Lipid tails: transparent yellow
Water: blue

14. ELBA in LAMMPS: membrane self-assembly
2 ns
Phase separation

15. ELBA in LAMMPS: membrane self-assembly
10 ns
Water channel forms

16. ELBA in LAMMPS: membrane self-assembly
30 ns
Water channel persists

17. ELBA in LAMMPS: membrane self-assembly
40 ns
Water channel thinning

18. ELBA in LAMMPS: membrane self-assembly
50 ns
Water channel disappears
Remaining defect: “thread” of lipid headgroups

19. ELBA in LAMMPS: membrane self-assembly
60 ns
Defect-free, stable bilayer

20. ELBA+LAMMPS perspectives
More lipid species
Cholesterol
Lipid mixtures
Microdomain (“raft”) formation
Dual-resolution
Atomistic proteins in CG membrane environment
[From “The inner life of the cell”] 20

21. Acknowledgements Jonathan Essex Julien Michel Funding: UK taxpayers
(University of Edinburgh)
Wendy Sanderson
Massimo Noro
Funding:
UK taxpayers
J&J
Unilever 21