Computational protocol: Membrane Binding and Self-Association of the Epsin N-Terminal Homology Domain

Similar protocols

Protocol publication

[…] Atomistic MD simulations utilized a single ENTH domain bound to the bilayer solvated with explicit TIP3P water and sufficient Na+ counterions to maintain overall charge neutrality (a). The initial coordinates for the ENTH domain were taken from the crystal structure (Protein Data Bank code 1H0A). For the lipid bilayer, an all-atom 80/20 palmitoyl-oleoyl phosphatidylcholine/palmitoyl-oleoyl phosphatidylserine mixed bilayer is used. The mixed bilayer was generated as described in a previous work. The ENTH domain/lipid bilayer system was built in the following way: The binding of PIP2 headgroup and the formation of H0 is assumed for the ENTH–bilayer bound complex. To embed H0 to the lipid bilayer interface, we removed five lipids (four PC lipids and one PS lipid). The orientation of H0 is approximately parallel to the membrane surface as observed from EPR experiments. The initial coordinates for the PIP2 headgroup bound to the ENTH domain are taken from Ford et al.Subsequent solvation, minimization, and MD equilibration protocols were carried out as follows. After the ENTH domain was placed on the lipid bilayer surface, the system is solvated with TIP3P water and neutralized by Na+ counterions. The resulting simulated system possessed ~ 97,000 atoms. Initially, all Cα and three phosphorus atoms of the PIP2 headgroup were harmonically restrained with a 5 kcal/(mol Å2) force constant and a conjugate gradient minimization of 5000 steps was applied followed by heating to 310 K followed by 100 ps constant NPT equilibration. The following three stages and a total of 70 ns constant NPT equilibration were then performed to fully relax the interaction of the ENTH domain and bilayer. In the first stage of 10 ns constant NPT equilibration, restraints were only applied on all the Cα and phosphorus atoms of the PIP2 headgroup to relax the bilayer. For the second stage of 10 ns constant NPT equilibration, the restraints on Cα were released except for the H0 part (residue 1 to residue 15). Finally, 50 ns of unrestrained simulation was performed as the final stage of equilibration. A production run of 50 ns was then performed for data analysis and the parameterization of the ENTH CG model.Atomistic MD simulations employed CHARMM22 and CHARMM27 force field parameters with the CMAP correction to describe the protein and the lipid–protein interactions, respectively. The parameterization for PIP2 lipid was as described in Lupyan et al. Simulations were performed under isothermal, isobaric conditions (constant NPT) and periodic boundary conditions. A Langevin thermostat with a damping coefficient of 0.5 ps− 1 was used to maintain the system temperature at 310 K. The system pressure was maintained at 1 atm using a Langevin piston barostat. Semi-isotropic pressure coupling was employed to retain the square shape of lipid bilayer throughout the simulation. Short-range nonbonded interactions were truncated smoothly between 10 and 12 Å, and the particle mesh Ewald algorithm was used to compute long-range electrostatic interactions at every time step. All covalent bonds involving hydrogen were constrained by the SHAKE algorithm (or SETTLE for water), permitting an integration time step of 2 fs. System minimization, equilibration, and dynamics were performed using the NAMD 2.7b1 software package. System construction and image generation were performed by using the VMD 1.8.7 software package. [...] A CG model of the ENTH domain, which retains the overall protein shape (b) and key membrane–protein interactions, was developed. The optimal locations of the CG sites were determined by the essential dynamics coarse-graining method based on the 50-ns atomistic MD trajectory of the ENTH bound to a lipid bilayer. The essential dynamics coarse‐graining method ensures that the locations of CG sites are variationally selected to optimally represent the low‐frequency dynamics as observed in atomistic MD simulations. The interactions between the CG sites within a single ENTH domain are represented by harmonic springs, parameterized by the heteroENM method. The strength of the harmonic springs obtained from the heteroENM method matches the thermal fluctuations of CG distances as mapped from atomistic MD data. Lennard-Jones (LJ) interactions were used to model all lipid–protein and protein–protein interactions as described in a previous work. Additional lipid–protein interactions were applied to membrane‐interacting sites to better reflect the difference from the non-membrane‐interacting sites on the protein. CG sites 1, 2, and 10 in b are categorized as membrane interaction sites, where CG sites 1 and 2 represent H0 and CG site 10 represents the arginine 114 loop (R114 loop). The H0–membrane interaction has a well depth of 4 kcal/mol. This value is chosen by employing an empirical transfer free energy calculation of partitioning an ENTH H0 (residues 1–15, sequence: MSTSSLRRQMKNIVH) from water to lipid bilayer interface using the software MPEx. In this calculation, N-terminal and C-terminal groups were selected as “NH3+” and “CONH2” groups, respectively. By varying total helicity of ENTH H0 from 85% to 100%, the free energy ranges from − 3.4 to − 4.3 kcal/mol; this indicates that partitioning the ENTH H0 sequence into bilayer interface as an α-helix form is a thermodynamically favorable process. It should be noted that previously reported metadynamics simulations to fold and insert the endophilin H0 helices into a membrane favors the folded state by 4.8 kcal/mol with respect to the unfolded state, and this value is comparable to what we have used here. For the interaction between the R114 loop and membrane, a well depth of 1.2 kcal/mol was applied. This interaction is obtained from atomistic MD trajectories by constructing the potential of mean force (PMF) as a function of distance between the center of mass of the residues within the CG site of the R114 loop (residues 112 to 116) and the phosphorus atom of the lipid headgroups. The PMF is computed by Boltzmann inversion of the radial distribution function, as previously described. Finally, 10% of the HAS lipid headgroups, the same fraction of PIP2 lipids used in the EPR experiments, were given an additional interaction with the PIP2 binding pocket of the ENTH domain to account for the specific interaction between the PIP2 headgroup and the ENTH domain binding pocket. The location of the PIP2 binding pocket was defined by the crystal structure in which the positively charged residues interact with the 4-phosphate group of the PIP2 headgroup. This interaction has a well depth of 2.4 kcal/mol. The strength of this interaction is guided by experimental data to account for high binding affinity of ENTH for the PIP2 headgroup. […]

Pipeline specifications

Software tools CHARMM, NAMD, VMD, MPEx
Applications Protein structure analysis, Membrane protein analysis
Diseases Neural Tube Defects