[AMBER] CpHMD – Effect of explicit ions when calculating the pKa of model compounds

From: Eric Lang <eric.lang.bristol.ac.uk>
Date: Mon, 30 Jan 2017 12:33:37 +0000


I have noticed in the original paper on CpHMD in explicit solvent (
http://pubs.acs.org/doi/abs/10.1021/ct401042b) that the pKa of the model
compounds were calculated with the model compounds solvated in explicit
water but in the absence of explicit ions. However, when switching to the
GB potential in the explicit solvent CpHMD workflow, an ionic concentration
of 0.1 M is used. Therefore, the reference energies are calculated assuming
a 0.1 M ionic concentration but the dynamics of the model compounds are
performed without ions.

In contrast, in the hen egg white lysozyme example of the paper, Na+ and
Cl- ions are added to neutralise the system and have a 0.1 M concentration
of NaCl. This is also done in Jason’s tutorial on Explicit solvent MD.

I was therefore wondering if there was a reason for not adding ions to the
water box when calculating the pKa of the model compounds as it would make
sense to do it to remain consistent with the GB calculations.

I ran a quick 1 ns CpHMD test of the Cys model compound (ACE-CYS-NME)
following the same protocol as reported in the paper.

If I do not add any explicit ions, as per the paper, I end up after at pH
=8.5 with the expected pKa/protonated fraction:

Solvent pH is 8.500

CYS 2 : Offset 0.007 Pred 8.507 Frac Prot 0.504 Transitions 3403

Average total molecular protonation: 0.504

However, if I add 1 Na+ and 1 Cl- to approximate a 0.1 M NaCl concentration
and run the same CpHMD, then I obtain the following:

Solvent pH is 8.500

CYS 2 : Offset -0.187 Pred 8.313 Frac Prot 0.394 Transitions 3923

Average total molecular protonation: 0.394

If I increase the simulation time to 3 ns this result is confirmed:

Solvent pH is 8.500

CYS 2 : Offset -0.171 Pred 8.329 Frac Prot 0.403 Transitions 11383

Average total molecular protonation: 0.403

When I look at the trajectory, the Na+ ion interacts with the S- of Cys, so
it is not due (or only due) to a change in dynamics due to the presence of
the ions but because of electrostatic interactions between the deprotonated
Cys state and the sodium ion.

I understand that this is a simple test and more work would be required to
indeed acuratly capture the effect of adding explicit ions to the model
compounds simulations, having for example a larger solvent box would enable
to better approximate the 0.1 M NaCl concentration, and running pH-replica
exchange would give more accurate results. In addition, I didn’t try any
other titrable residues at this stage either.

Nonetheless, I am a bit worried about these results and the effect of not
having explicit ions when the reference energies where calculated. Overall
is this not going to lead to a small error in the calculated pKa due to the
fact that the model compounds were simulated in the absence of explicit
ions? When running the simulations of a protein of interest with explicit
ions, the same ions-charged residues interaction will occur. But perhaps I
am missing something important there, so I would be grateful to have your
opinion on this matter.

Many thanks in advance,


Eric Lang
BrisSynBio Postdoctoral Research Associate Modelling
Centre for Computational Chemistry
School of Chemistry - University of Bristol
Bristol BS8 1TS - United Kingdom
AMBER mailing list
Received on Mon Jan 30 2017 - 05:00:02 PST
Custom Search