[an error occurred while processing this directive]
[an error occurred while processing this directive]

Valinomycin and Potassium Transport

T R Forester
W. Smith

Daresbury Laboratory, Warrington, WA4 4AD

J H R Clarke
Dept of Chemistry, UMIST, Manchester M60

(a) In Vacuuo Studies

The selective transport of cations across biological membranes is a process of fundamental importance in cell biology. It is by this process that the alkali metal ion concentration gradients between the two sides of a membrane, which are strongly associated with oxidation processes within the cell, are created, maintained or destroyed. This transport is facilitated by the presence of small amounts of antibiotics, and broadly two modes of action have been identified. Antibiotics such as gramicidin form channels which bridge the membrane and through which ions diffuse. Valinomycin (VM) is however an example of a large class of carrier antibiotics and in this case three distinct processes appear to be involved. (i) Complexation between the cation and a membrane bound antibiotic, (ii) transport of the complex through the membrane by diffusion or electrophoresis, and (iii) cation release at the membrane interface. Concentration gradients are thus ameliorated by the capture and release of cations on the two sides of the membrane.

Our interest is with carrier antibiotics and in particular we wish to address the issue of the kinetics and mechanism of the ion capture at the membrane interface. VM is an important example of an ion-translocating carrier antibiotic since it is highly selective towards potassium transport and has been widely studied using a number of experimental techniques. All of these carrier antibiotics act by forming chelates with the alkali metal ions. For VM this chelation is associated with substantial change in the conformation of the ligand, from a half-open ring structure in polar solvents to a bracelet-like structure for the potassium complex.

Using an adapted AMBER potential derived for valinomycin [1], molecular dynamics simulations of the capture of both hydrated and unhydrated potassium ions by valinomycin have been performed [2]. The DL_POLY parallel macromolecular simulation package being used for all simulations [3]. There are no explicit solvent interactions although a stochastic bath is used to simulate thermal equilibrium at ambient temperature. An ``open ring'' conformer was used, consistent with common interpretations of experimental data. The distant attraction of the cation to the biopolymer is dominated by the dipole-charge interaction and as the cation approaches conformational changes are induced in the biopolymer to enhance its dipole moment. These changes involve both amide and ester carbonyl groups aligning towards the approaching cation. Initial coordination is via amide carbonyls although this is eventually overtaken by ester carbonyl coordination on a timescale of about 30 ps. When cation water of hydration is included in the simulations the time scale of the capture process is lengthened by approximately three orders of magnitude. The presence of the hydrated cation is sufficient to induce the conformational change from the twisted bracelet form of valinomycin to the open ring structure. The valinomycin rapidly changes shape in the early part of the capture but displacement of water molecules by (mainly) ester carbonyls is a slow process. One water molecule remained firmly attached to the complex after 18 ns. This raises the question as to whether water is transported with the complex through membranes in vivo.

Figure 1: The valinomycin-potassium complex, clearly showing the formation of the `tennis ball seam'.

(b) Solution Studies

Valinomycin is a relatively simple biological molecule with considerable conformational flexibility. In solid state complexes with alkali metals VM folds around the ion with the ring adopting a ``tennis ball seam'' arrangement. NMR, IR and CD measurements indicate the structure in solution (e.g. chloroform, methanol, and cell vesicles) is similar to the solid state although the K-VM complex is unstable in water. The uncomplexed molecule can be crystalised into a number of conformers including the ``twisted bracelet'' (TB) conformation from octane and a ``twisted propeller'' (TP) conformation from DMSO. In solution the conformation of VM is sensitive to polarity of the solvent. One set of conformers seem to predominate in non polar solvent while another set predominate in more polar environments such as methanol-water mixtures. Whatever the solvent no one predominant conformational state exists at room temperature nor are spectroscopic measurements (UV, CD, IR and NMR) of VM consistent with any of the solid state conformers. Clearly, an understanding of the solvent induced conformations of VM (and K-VM) is important for probing the ability of this molecule to facilitate ion transport across aqueous - membrane interfaces. Even though the relaxation times between the various VM conformers in solution is relatively short (1-10 nanoseconds) elucidation of the mean structures in solution is still a considerabe challenge for molecular simulation. As a prelude to expilicit simulation of antibiotic action at the interface we report Molecular Dynamics (MD) simulations of VM and the K-VM complex in the environments expected either side of the interface viz. in non polar solution and in water.

Molecular dynamics simulations of Valinomycin (VM) and its potassium complex in water [4] and in a Lennard Jones solvent have been studied using the DL_POLY parallel macromolecular simulation package [3]. In agreement with experimental evidence the structure of K-VM in non polar solution is similar to the solid state structure while the structure of uncomplexed VM is not. In water uncomplexed VM retains the Lac and HyV faces (which are lost in non polar solution) and shows some similarity with the solid state structure obtained by crystalisation from DMSO. However, also in agreement with spectroscopic data a dynamic equilibrium between a set of conformers is established in both solvents. Our model reproduces the experimental dipole moment of VM in non polar solution (3.6 D). In keeping with the experimental stability constant, we also observed the spontaneous decomplexation of K-VM in water. The face for potassium (de)complexation in methanol has been the subject of some debate in the literature. Our calculations in water support a mechanism involving ion passage through the ``HyV' face in preference to the Lac face. Water attack was observed through both faces.

(c) In Vivo Studies

Atomistic level simulations of the antibiotic valinomycin (VM) and its potassium complex at a water/membrane interface have been performed [5]. These studies demonstrate the practical feasibility of realistic large scale simulations of complex systems of biological importance and were achieved using the recently introduced parallel supercomputing facility at Edinburgh (UK) as part of the work of the HPCI Materials consortium. The parallel molecular dynamics package DL_POLY was used for all calculations. The simulations, involving almost 19,000 atomic sites, cover a total of over 500 ps, and are used to examine both VM at the interface and the K-VM decomplexation reaction.

Key findings are that uncomplexed VM acts as a surfactant with hydrophobic groups embedding in the membrane while the hydrophilic carbonyl groups hydrogen bond with water. Consequently the conformers VM adopts at the interface are quite distinct from those seen in the solid state or in bulk solution where most experimental measurements are made. The open ring conformer, long postulated to be involved in complexation and found to be important in in vacuo capture was not stable at the interface. Rather more compact and globular conformations are adopted. In contrast, the K-VM complex remains embedded in the membrane until the decomplexation process is initiated by water attack through the HyV face. What we observe in the simulations is precisely the ideal behaviour for an ion translocating molecule : the uncomplexed molecule concentrates at the interface where the complexation reaction occurs, but upon complex formation favours the interior of the hydrophobic phase where it may be transported by diffusion or electrophoresis. Neither our model membrane nor the simulation timescales available to us make it possible to explore the movement of the complex through the membrane interior. However this is not of particular interest as this process could be adequately studied by solving the Navier-Stokes equation without the need for a detailed atomistic simulation.

The decomplexation process at the interface show important differences from that observed in pure water. In particular since water attack at the interface is restricted to the HyV face only, decomplexation takes place more raipdly than in pure water. This points to the role of the interface itself in the function of carrier antibiotics such as valinomycin.

In potassium selective detection devices and in biomembranes, the driving force for the transport of potassium across the membrane interface is an electric field arising either from an applied potential or from an ion concentration gradient. Accordingly, we have also studied the system in the presence of electric field applied across the membrane.

Figure 2: Schematic diagram of the MD simulation cell, three dimension periodic boundary conditions turn this into a laminar system. (a) Initial membrane (wavey lines) and valinomycin (VM) positions, the remaining space is occupied by 0.4 molal KCl solution (b) initial packing of the VM and K-VM complexes viewed along the z-direction. The VM are at the top interface, the K-VM complexes (shaded) at the lower interface.


[1] T.R. Forester, W. Smith and J.H.R. Clarke, J. Phys. Chem. (1994) 98 9422

[2] T.R. Forester, W. Smith and J.H.R. Clarke, J. Phys. Chem. (1995) 99 14418

[3] T.R. Forester and W. Smith, The DL_POLY Molecular Simulation Package, J. Molec. Graphics 14 (1996) 136.

[4] T.R. Forester, W. Smith and J.H.R. Clarke, Biophys. Journal 71 (1996) 544.

[5] T.R. Forester, W. Smith and J.H.R. Clarke, J. Chem. Soc. Faraday Trans. 93 (1997) 613.

For more information about the work of the Computational Chemistry Group please contact Paul Sherwood p.sherwood@dl.ac.uk or Bill Smith w.smith@dl.ac.uk
back to top
 Quick links





Modelling Superoxide Dismutases
Crystal Polymorph Prediction
Modelling powders
Diffusion at Grain Boundaries
Paul Sherwoood
Bill Smith
Laurence Ellison
John Purton
C.W. Yong
Michael Seaton
Tom Keal
Rick Anderson
Sebastian Metz
Ilian Todorov