AMBER Archive (2008)

Subject: AMBER: Problems of MM/MM-GBSA calculations

From: Sergey Samsonov (sergeys_at_biotec.tu-dresden.de)
Date: Thu Nov 27 2008 - 04:53:53 CST

Hi AMBERs,

recently we've been calculating MM-GBSA energy of binding of ligands
(dipeptides) in the enzyme catalytic center, where Zn2+ coordination is
carried out by ligand carboxyl group and, therefore, Zn2+ was considered
as a part of the receptor. However, MM-GBSA results did not correlate
with the experimental data, which we explained by the fact that 1. There
is quite an important influence of water molecules (MD was done with
explicit solvent and TIP3PBOX) building hydrogen bonding network with
Zn2+ ion and mediating interactions between a ligand and the receptor 2.
To many charges in comparably small the binding site. So in the next
step we took explicitly water molecules (with the command closest)
around the ligand and calculated simply MM energies. The results
correlated nicely with the experimentally observed trends. As far as we
used standard amino acids, this approach worked. However, when we tried
to use methyl-carboxyl group at the N-terminus of the dipeptide,
parametrized by the standard RESP procedures with the use of the already
existing corresponding atom types from the force field (frcmod03,
parm99.dat), the energies, obtained by MM, had really strange values.
For example, it was obvious from the visual inspection of the MD
simulation that the methyl-carboxyl group stabilized the interaction by
VDW contacts and also the introduction of the group contributed
favorably to electrostatic interactions (additional hydrogen bonds and
salt bridges were formed). Nevertheless, electrostatic component of MM
in binding energy was 200 kcal/mol less favorable than for not
methyl-carboxylated variant of the ligand. The attempt to parametrize
the whole amino-acid with the methyl-carboxylated group as a new residue
led to the same results. Previously we had successful experience with
RESP charge calculations and libraries creation for non-standard amino
acid residues as well as their use for MM-PBSA/MM-GBSA calculations.

Do you think we could do something principally wrong or the system could
be under certain limitations for the method because of the comparably
small size of ligand and quite a charged environment in the binding site?

We would appreciate very much any suggestion.

Sergey Samsonov

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo_at_scripps.edu