Simulating botulinum neurotoxin with constant pH molecular dynamics in Generalized Born implicit solvent

https://doi.org/10.1016/j.cpc.2007.02.095Get rights and content

Abstract

A new method was proposed by Mongan et al. for constant pH molecular dynamics simulation and was implemented in AMBER 8 package. Protonation states are modeled with different charge sets, and titrating residues are sampled from a Boltzmann distribution of protonation states. The simulation periodically adopts Monte Carlo sampling based on Generalized Born (GB) derived energies. However, when this approach was applied to a bio-toxin, Botulinum Neurotoxin Type A (BoNT/A) at pH 4.4, 4.7, 5.0, 6.8 and 7.2, the pKa predictions yielded by the method were inconsistent with the experimental values. The systems being simulated were divergent. Furthermore, the system behaviors in a very weak acidic solution (pH 6.8) and in a very weak basic solution (pH 7.2) were significantly different from the neutral case (pH 7.0). Hence, we speculate this method may require further study for modeling large biomolecule.

Introduction

Botulinum neurotoxins (BoNTs) are among the most potent toxins to human beings. There are seven different serotypes (A-G) of BoNT [1]. They were believed causative agents of botulism, a potentially fatal condition of neuromuscular paralysis and researched by many groups [2], [3], [4], [5], [6]. Currently no effective chemical antidote is available against botulism principally because of the lack of knowledge of the molecular structures and the mechanism of toxicity [1].

In the whole three-step toxic action, translocation domain across membranes attracts great attention since there are interesting phenomena in channel forming and enormous size transport. An acidic environment is believed to induce conformational changes in the translocation domain. Unfortunately, there was no experimental approach available for the range pH 4.0 to 5.0 [3]. For this reason, the property of BoNT/A in this range is of greater interests and an efficient constant pH MD modeling technique is highly desirable.

Section snippets

Constant pH MD

It has been long and well known that protein structures are strongly dependent on solvent pH [7], [8]. On the other hand, it is known that a structural change also affects the pKa values of titratable residues. Therefore, there is a tight coupling between protein conformation and protonation state [8]. Modeling such correlation requires a reliable constant pH molecular dynamics method and such a method was indeed proposed and implemented in AMBER 8 package by Mongan et al. [9]. This method

Numerical experiment

Our MD experiments were performed by using AMBER 8. The ff99 force field was employed. A refined GB model (igb=5) was used for solvation. Salt concentration was set at 0.2 M. The cutoff for nonbonded interactions and computation of effective Born radii was 30 Å. The time step was 2 fs. Sander [9] was used as MD engine and ptraj [9] was used to analyze the trajectories and calculate the RMSD values in each case.

Results

A valid constant pH MD simulation method should yield pKa predictions consistent with experimental values. It can rapidly converge to those predictions and maintain a stable trajectory. Furthermore, the method should be computationally efficient [8].

Analysis and conclusions

To explain these results, the following reasons are considered.

(1) Sufficiently conformational sampling is infeasible for the system of this size.

The instantaneous pKa is strongly dependent on conformation, so it is reasonable that they would have different protonation state population if two simulations sample conformation space differently. The random error due to incomplete conformational sampling may produce larger effects than those caused by a small change in pH [8]. Of all 1277 residues

Acknowledgement

We thank Brookhaven National Laboratory (BNL) for their support and assistance.

References (11)

There are more references available in the full text version of this article.

Cited by (0)

View full text