Simulation of a dizinc metallo protein

User discussions
,
1

GROMACS version: 2024
GROMACS modification: Yes/No
Dear users,

I’d like to know if anyone has some experience simulating dizinc metalloproteins with the CHARMM force field, specifically metallo-beta-lactamases with one of the zincs coordinated by 3 histidines and the other 2 histidines and an aspartate.

I researched the literature but I didn’t find anyone simulating such a system using GROMACS either, usually, these simulations take place either with QM/MM, using bonded parameters in AMBER, or using the cationic dummy method (which can’t be directly transferred to my case anyway because of my geometry).

Approaching the problem with the default parameters for CHARMM36m doesn’t work, as the zinc ions are too close in this pocket, and the localized +2 charge ends up being a poor representation of the system: the ions quickly diffuse away from the pocket as soon as the restraints are lifted after a (very long) equilibration.

I would deeply appreciate some suggestions on how to tackle this problem. The way I see it right now, the most obvious way would be to use amber tools to prepare bonded parameters for Amber and convert the topology to GROMACS afterward.

Thank you very much in advance!

2

If you don’t need to study the dissociation of the ions, I’d go ahead with ZAFF (there’s an extended version available now).

One could also try to convert the Amber ZAFF files into Gromacs .ff files with this gromologist workflow, I haven’t really tried it before though.

3

Thank you very much for the advice, Milosz, I’ll give this EZAFF a try!

4

For anyone who might be interested/troubled by this in the future, I’ve managed to successfully simulate the former dizinc system using the parameters developed by Macchiagodena et al. in her two papers (https://pubs.acs.org/doi/10.1021/acsomega.0c01337?ref=recommended and https://pubs.acs.org/doi/10.1021/acs.jcim.9b00407). I had to implement a hydroxide ion bridging the two zincs, with parameters described as in Marion et al. (https://pubs.acs.org/doi/10.1021/acschembio.6b00847). I found all of this thanks to this excellent and thorough benchmarking by Melse et al. (https://onlinelibrary.wiley.com/doi/full/10.1002/jcc.27052).