CHARMM TIP3P vs standard TIP3P

GROMACS version: 2021
GROMACS modification: No
Force Field: CHARMM C36M

I’m simulating a peptide that has random structure in solution, but folds into an alpha helix in contact with negatively charged membranes. Using the CHARMM C36M forcefield and the CHARMM TIP3P water model, I simulated the peptide for 3 or 4 µS and started the simulation with the peptide in alpha helix. While the helix was much more stable in contact with the anionic membrane than in solution, with most of it being stable for a little under 1.3 µS, after this mark a portion of the helix unfolds and after 2µS a huge portion of the helix unfolds in all simulations, and something arround . While this shows that the equilibrium is shifted towards alpha-helix, the value of ~35% of alpha-helix is quite lower than the more than 70-80% measure experimentally by Circular Dichroism and even from previous resulrs of simulations using the AMBER 99SB-ILDN-NMR forcefield which also give about 72% of helix content which is 2 times higher than my results.

Even read a paper in which they show that the CHARMM modified TIP3P supresses protein folding. But in the CHARMM C36M paper they say the CHARMM TIP3P water is superior to standard TIP3P.

So I was wondering if the combination of CHARMM TIP3P water with the CHARMM C36M (which is recomended both at the C36M paper and when choosing the water molecule on pdb2gmx) could also lead to a less realistic helix-coil equilibrium and explain the helix unfolding on my simulations.