CHARMM Development Project
Posted By: Abhishek.2015 CHARMM parameters for Mn2+ - 05/01/20 05:55 PM
Dear all,
I want to run a MD simulation of Mn2+ bound protein but I could not find CHARMM parameters for Mn2+. I came to know that transition metals are highly problematic, as they are typically present as a complex with a different specific geometry depending on the oxidation state and available ligands.
In my system, Mn2+ in complex with Glu, Asp, His, and 2 water molecules. So how can I get Mn2+ CHARMM parameters for my system?

Many thanks for your kind attention.
Posted By: rmv Re: CHARMM parameters for Mn2+ - 05/03/20 04:44 PM
As you may have noticed, the answer to this question has not changed for the past 16 years or more.

How you approach this depends on the importance of the Mn++ binding site in the study being considered. If you only need it as a structural element, unrelated to the rest of the study, stabilizing the preferred binding geometry with restraints may be the best approach. For that, you should undertake a review of relevant crystal structures, esp. those with good resolution for the complexation site, to determine a reasonable geometry.
Posted By: Corey Taylor Re: CHARMM parameters for Mn2+ - 07/31/20 11:15 AM
I'd like to tack onto this conversation if nobody minds because my questions are probably in a similar vein. I have a couple of protein-actinide pockets (e.g. Transferrin/Cm) to play with. The composition of each pocket is a metal bound to two deprotonated Tyrosines and, depending on which pocket, nearby Aspartic Acid and Histidines and a couple of carbonate counter ions. So far I have DFT opts of each pocket and currently running QMMMs to obtain equilibrium angle and bond measurements.

The question concerns what I would need to run long classical simulations as I'm mostly interested in conformational changes that occur once the metal is complexed, the binding process I'm getting from the other calculations. With no other parameters than LJ and what's in toppar_ions_won.str, I can certainly set-up and run equilibration dynamics to get a reasonable geometry prior to a QMMM run.

So what would be required for longer-term dynamics with metals in this context? Something along the lines of the SOD LJ+NBFIX parameters from Rick's paper that ended up in toppar_water_ions.str is a start but I'd be missing bond and angle parameters for the salt-bridges between the deprotonated TYRs and the metal. So, this means equilibrium distance, angles and force and spring constants (e.g. Kb). The bonded items I guess I'd get from lsfit, not sure about the angle constant. Is this enough to kick off some production dynamics for some testing/validation or am I missing something else essential? Cheers, guys.
Posted By: rmv Re: CHARMM parameters for Mn2+ - 07/31/20 02:32 PM
For L-J terms, the initial metal ion parameters are based on the free energy of solvation, and non-bonded interactions with formal anions can require adjustment of radii and well depths. However, for bonded topologies, the L-J terms are not used for atoms connected by a bond or angle term, so you may not need to do a lot. Charges may be more of an issue, since the standard method for CHARMM charges uses a fairly low level of theory, and may not be suitable for Mn++ bonded to oxyanions.

I suggest looking through the heme stream to see how iron is handled. One consideration is the dihedral terms, which are typically autogenerated based on the bonded topology; for heme the autogeneration is suppressed, and dihedral terms are added explicitly.
Posted By: Corey Taylor Re: CHARMM parameters for Mn2+ - 07/31/20 06:28 PM
Yeah I saw the heme stream, was thinking I'd need to do something similar. Good tip on the dihedrals and good reminder about charges - cheers for that. I assume they're based on something HF/6-31G-ish? I've calculated AIM charges at a DFT/TZVP level for each of the complexes so they should be reliable, especially since the geometry is fairly predictable. I guess I can calculate, roughly, force constants from the calculated vib spectra for now. It won't be accurate, of course, but these are apparently guessed in the heme stream file for Fe so as long as it's reasonable, well tested, based on an analogous metal, etc.

So, to be clear, a way forward would be to come up with a new RESI comprised of, say, the metal + two deprotonated TYRs/whatever other residues form an ionic interaction + counter ion. For everything else non-bonded (e.g. waters) LJ parameters would come into play.

I do have a couple of other (hopefully) quick questions.

- use of dummy atoms in the dihedral terms. e.g.

X    CPB  CT3  X        0.0000  6     0.00 ! ALLOW HEM! Heme (6-liganded): substituents (KK 05/13/91)
X    FE   CM   X        0.0500  4     0.00 ! ALLOW HEM   ! Heme (6-liganded): ligands (KK 05/13/91)

I think I know what's going on. The purpose here is to ensure a dihedral term is created and it doesn't really matter what's connected to the respective atoms?

- Why would the LJ epsilon for Fe be zero?

FE     0.010000   0.000000     0.650000 ! ALLOW HEM  ! Heme (6-liganded): Iron atom (KK 05/13/91)

Appreciate your help as always.
Posted By: rmv Re: CHARMM parameters for Mn2+ - 07/31/20 10:53 PM
I would probably create a patch residue (PRES) that adds the Mn++, deletes hydrogens, adjusts the partial charges, adds bonds, angles, and possibly dihedrals. I recommend using a custom stream file such as the distributed toppar_all36_*.str files for any custom additions, so that it is easy to document the additions.

If you add explicit dihedrals, autogenerate should be disabled, and it would be best to not make any further changes to the PSF other than adding water or ions. More complex molecules require autogenerate, which you don't want.

The first dihedral term above suggests a way out; an X in the 1st and 4th position is a wildcard that allows any atom, with no impact since the force constant is zero. Adding similar parameter terms would allow autogenerate to operate, and not have any effect on the resulting potential.

I'm not sure why the well depth is zero for Fe.
Posted By: Corey Taylor Re: CHARMM parameters for Mn2+ - 08/01/20 08:27 AM
Yup, that makes more sense than making an entire new residue. I had actually setup a PRES/PATCH one of the TYRs just to setup the QMMM but of course it makes sense just to extend it with more parameters (once I have them...).

Well this turned out to be relatively simple. Cheers again, Rick.
Posted By: Corey Taylor Re: CHARMM parameters for Mn2+ - 08/01/20 08:45 AM
In the interests of transparency and for someone else reading this, my PRES looked like this:

*Additional PRESs for Tf
36  1


PRES TYD         -1.00 !
ATOM CZ   CA     -0.18 !
ATOM OH   OH1    -0.82 !
!ATOM HH   H      0.00 !


And corresponding PATCHs in the input file:

*QM/MM calculation of protein, water box


! Read topology and parameter files
stream toppar.str

!    Read in protein

open read card unit 10 name "prep/6ctc_bp1_prep_prot.pdb"
read sequence pdb resi unit 10
generate Tf setup warn first nter last cter

! Termini, disulfide bonds
patch NTER Tf 1 setup warn
patch CTER Tf 330 setup warn
patch disu Tf 7 Tf 46 setup warn
patch disu Tf 17 Tf 37 setup warn


open read card unit 10 name "prep/6ctc_bp1_prep_prot.pdb"
read coor pdb resi unit 10

!     Read in metal and counter ions

open read card unit 10 name "prep/6ctc_bp1_prep_FeCO3.pdb"
read sequence pdb unit 10
generate Tf2 setup warn

! Patch Fe-TYR bond
patch TYD  Tf 186 Tf2 331 setup warn

open read card unit 10 name "prep/6ctc_bp1_prep_FeCO3.pdb"
read coor pdb unit 10

© CHARMM forums