AMBER Archive (2009)

Subject: Re: [AMBER] Using GAFF with a covalently bound ligand or label

From: FyD (
Date: Fri Jan 16 2009 - 03:13:43 CST

Dear Ken,

> I would like to simulate proteins to which a covalently bound label (i.e.
> cross-linker, spin label, fluorescent probe) has been attached at the SG of
> a cysteine. I am thinking that the best approach is to use a standard AMBER
> FF for the protein and the GAFF for label. My questions are: 1) does this
> make sense? and 2) if so, how do I treat the SG to label connection? I
> assume I will have to add a few FF parameters to deal with this, but can
> this be done across two FFs with different atom type nomenclature?

You first need to create a new molecular fragment starting from a
whole molecule. R.E.D. has been
designed to do such a task. You will find a detailed tutorial about
charge derivation & force field library building for AMBER FF @
R.E.D. generates at the end a force field library in the Tripos mol2
file format that you can load in LEaP.

- Usually, such a case is solved using "intra-molecular charge
constraint(s)" (intra-mcc) during the fitting step on a group of atoms
belonging to the whole molecule. If you download the R.E.D.-III.1
tools distribution, you will get (besides the programs) a MiniHowTo
with input files for R.E.D. In "4-Fragment-1", you have an example of
use of 2 "intra-molecular charge constraints" for a dipeptide. You
could follow a similar strategy for your ligand linked to the
Cysteine. You need to find a connecting group R where you apply this

     Cys-S .... R-ligand --> Cys-S-ligand

     total charge of "Cys-S" = 0
     total charge of "R-ligand" = total charge of "ligand"
     total charge of "R" = 0 using an intra-mcc

- Now in the AMBER force field topology database (FFTopDB), you have
two Cysteine residues: Cys (cysteine) & Cyx (cystine): Cys of for the
thiol form and Cyx for the disulfide bride version. This is NOT your
case! Thus, you might decide to derive charge values for a Cys residue
bound to your ligand (or better for a series of your ligands) as well.
I wonder if this would not be a better choice... For that, you start
from a dipeptide (ACE-CYA-NME) for your sulfide residue "CYK" and you
use an "inter-molecular charge constraint" (inter-mcc set to a value
of zero for the chemical groups R1 & R2 ; see below) during the
fitting step between the dipeptide & your ligand (besides the two
required intra-mcc for the ACE and NME chemical groups):

   CYA-S-R1 R2-ligand --> CYA-ligand (only R1 & R2 are removed !)
   NME (S belongs to CYA)

     total charge of "ACE" = total charge of "NME" = 0 using two intra-mcc
     total charge of "CYA-S-ligand" = integer
           because total charge of "R1 + R2" = 0 using 1 inter-mcc

You will find examples of inter-mcc in the MiniHowto in the
R.E.D.-III.1 distribution as well.

Do not hesitate to ask if you need more information. This is not a
straightforward task if you do want to do it rigorously. But for sure,
you will get what you want using R.E.D.

I hope this helps, regards, Francois

AMBER mailing list