MD simulation run 2 of TARP gamma-2

Molecular dynamics simulations of pig TARP γ-2 were set up using an AlphaFold 3 model for UniProt ID A0A4X1U4N8. This run belongs to a set of 3 runs from a γ-2 simulation system that was prepared directly from the AlphaFold model without any modification. It is run 2 of 3 and has a duration of 100 ns.

The CHARMM-GUI Membrane Builder was used to prepare the simulation systems, including trimming the C-termini at residue 227, neutralising the truncated C-terminus with N-methylamide cap (C-terminal patch CT3), orienting and positioning the protein in a bilayer with PPM, building a 75 Å by 75 Å POPC bilayer around the protein using the pseudoatom replacement method, adding water and 0.15 M Na+ and Cl- ions, performing initial energy minimisation in CHARMM, and preparing GROMACS input files. 

3 MD simulations were run from the system prepared with CHARMM-GUI using GROMACS 2021.5 with the CHARMM36 all-atom force field, including the CHARMM36m update for proteins and the CHARMM TIP3P water model and standard parameters, such as particle mesh Ewald electrostatics and force switching for van der Waals with a cutoff of 1.2 nm and switching from 1.0 to 1.2 nm. 

The simulation protocol in GROMACS was the standard one created by CHARMM-GUI, run on a computer cluster using modified versions of the provided README file, including slurm submission parameters, and a modified version of the provided step7_production.mdp file with a larger number of time steps, corresponding to 100 ns. 

Energy minimisation was run for 5000 steepest descent steps with position and dihedral restraints on the protein and lipid to relax the solvent around them. Harmonic position restraints were used for protein backbone and side-chain heavy atoms with force constants of 4000 and 2000 kJ·mol⁻¹·nm⁻² were in all directions, respectively. Lipids were treated with planar position restraints were used in the z-direction with force constants of 1000 kJ·mol⁻¹·nm⁻² to maintain lipid head groups in the planes of the membrane leaflets and harmonic dihedral restraints with force constants of 1000 kJ·mol⁻¹·nm⁻² to maintain correct orientation of phosphate groups and lipid tails relative to the core choline head group. These are close to the previously published values of 10.0, 5.0 and 2.5 kcal·mol⁻¹·Å⁻², followed by several rounds of equilibration of different duration, gradually reducing the restraints as previously described [ref 2]. All steps after energy minimisation used the Bussi-Donadio-Parrinello stochastic velocity rescaling thermostat with three coupling groups corresponding to protein, lipid and solvent atoms and a time constant of 0.1 ps and a target temperature of 303.15 K. The first two equilibration steps were in the NVT (constant number of atoms, volume and temperature) ensemble, only equilibrating the temperature. This was followed by four rounds of NPT (constant number of atoms, pressure and temperature) equilibration, adding semi-isotropic pressure equilibration with the Bernetti-Bussi stochastic cell rescaling barostat with a time constant of 5.0 ps and compressibility of 4.5 x 10-5 bar-1, equilibrating the system to a pressure of 1.0 bar, which was also maintained for all steps after NVT equilibration. Finally, the restraints were removed and production MD simulations were run in the NPT ensemble with the same thermostat and barostat with the same parameters. The first three equilibration steps used 1-fs time steps, after which 2-fs time steps where used. Bonds containing hydrogen were constrained with LINCS for all steps, including minimisation.