Isoleucine

All MD simulations were performed with NAMD^{40 (link)} employing CHARMM-formatted parameter files^{41 (link)} for all force fields tested, which are provided in the Supporting Information. For all simulations, a temperature of 300 K and pressure of 1 atm were maintained with a Nose–Hoover Langevin piston barostat with a piston period of 100 fs and a piston dampening time scale of 50 fs and a Langevin thermostat with a damping coefficient of 1 ps⁻¹. Nonbonded cutoffs were employed at 11 Å with a smoothing function starting at 9 Å, with particle mesh Ewald used to treat long-range electrostatics. The systems were solvated in cubic water boxes with edge lengths ranging from 25 to 58 Å. Sodium and chloride ions were added to neutralize the charges in the system and provide approximately a 150 mM concentration of salt. A 2 fs time step was employed with the use of SHAKE and SETTLE.
Triplicate 205 ns simulations were run for an unblocked alanine pentapeptide (Ala₅) with and glycine tripeptide (Gly₃) with protonated C-termini with the first 5 ns discarded as equilibration. The remaining amino acids, with the exception of proline, were simulated for 205 ns as blocked dipeptides, again in triplicate with the first 5 ns discarded as equilibration. Values and error bars throughout the paper represent the mean and standard deviation of the calculated quantities from the triplicate runs. Ala₅ and Gly₃ simulations were run with each of the four weighting temperatures examined in this work, as well as the previous OPLS-AA and OPLS-AA/L force field. Dipeptide simulations were performed with OPLS-AA, OPLS-AA/L, and the new parameters optimized at 2000 K. As each system was studied for 600 ns with at least three different force fields, over 50 μs of validating simulations have been executed. In analyzing the molecular dynamics simulations for the short alanine and glycine peptides, the definitions of secondary structure, the three sets of Karplus parameters for calculating J couplings, and the experimental error values used to calculate χ² from Best et al.^{42 (link)} were employed. For the dipeptide simulations, only the first set of Karplus parameters, that of Hu and Bax,⁴³ was employed. χ₁ rotamer populations were determined by dividing the range of χ₁ values into three equal sized bins, corresponding to the p (+60°), t (180°) and m (−60°) conformers. Definitions of p, t, and m for valine, isoleucine, and threonine were adopted from the work of Dunbrak and co-workers^{27 (link)} and are depicted in Figure 1.
The proteins ubiquitin and GB3 were started from the PDB structures 1UBQ^{44 (link)} and 1P7E^{45 (link)} and gradually heated to 300 K over 400 ps before 205 ns simulations were run. Both the heating period and the first 5 ns were discarded as equilibration, and simulations were performed in triplicate for each protein. All other simulation parameters were identical to those used for the dipeptides. For calculation of backbone J couplings of the full protein, both the 1997 empirical Karplus parameters⁴³ used for the dipeptides and another empirical model developed from work with GB3^{46 (link)} are employed. Side chain J couplings were calculated for couplings to methyl side chains with the set of Karplus parameters developed by Vögeli et al.,^{46 (link)} while all other couplings employed Karplus parameters from Perez et al.^{48 (link)}