| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biophys J, September 2002, p. 1235-1236, Vol. 83, No. 3
Department of Chemistry, Brandeis University, Waltham, Massachusetts 02454-9110 USA
Selective ion channels share a common feature.
They exhibit the apparently contradictory properties of high turnover
and high selectivity. The structures of four such membrane proteins
have led to an explosion of computational papers designed to illustrate in detail how various architectural features can couple at the molecular level and resolve the conundrum. A stringent macroscopic test
of any channel theory is its ability to reproduce experimentally observed current-voltage-concentration (I-V-C) profiles. Ideally this
would be done microscopically, via applied field molecular dynamics
(MD) (Crozier et al., 2001 In this issue of Biophysical Journal Edwards et al.
(2002) The electrostatic model treats the gramicidin channel (~24
Å long, ~4 Å diameter) as a distinct dielectric
phase, surrounded by a low dielectric milieu (protein and membrane),
sandwiched between two high To provide functional estimates of the unknown dielectric constants,
Edwards et al, (2002) The results should be cautionary for electrophysiologists and
theoreticians alike. Current-voltage profiles alone provide too little
independent information to seriously constrain a channel's ionic
energy profile. Does the success of electrostatics in providing a
framework for interpreting potassium channel conductance guarantee that
it will be equally reliable when applied to the chloride channel? Quite
possibly not, since potassium channels are effectively blocked by their
own permeant ions; conductance requires relief of this block via ionic
acceleration through the channel's wide inner pore, a region large
enough to be treated by continuum electrostatics (Chung et al.,
2002
ARTICLE
). At present this is not
feasible and mesoscopic approaches are employed, either based on free
energy profiles determined from MD simulations or by treating ionic
conduction as electrodiffusion through a viscous, continuum dielectric.
The lure of the electrodiffusive approach is its physical transparency and its computational simplicity. Its major pitfall, neglecting non-uniformity and non-locality, has a long history (Warshel and Russell, 1984; Komyshev, 1985). Nonetheless, if
the narrow water-filled transmembrane conduit typical of a channel
protein is well approximated as a uniform dielectric phase, coupling
the structural data with electrostatic arguments determines the ionic
translocational potential profile, from which I-V-C relationships are
efficiently computed (Nonner et al., 1999).
subject continuum electrostatics to trial by ordeal.
Using the 30-residue gramicidin A dimer as their exemplar, they
determine whether this approximation can adequately account for three
observables: the I-V profiles, the I-C profiles, and the binding site
locations. For this exceptionally well-characterized (both structurally
and electrophysiologically) and very narrow channel the conclusion is
unambiguous. Continuum electrostatics fails, at least in its simplest
form; the single-file waters in this long, narrow pore are not
realistically represented as a uniform dielectric phase, regardless of
the choice of 
. The authors then go further, using Brownian dynamics
(BD) to deconvolute the structural and electrophysiological observations and determine an ionic free energy profile for potassium translocation through the pore, in the process demonstrating that I-V
data alone are inadequate to critically test phenomenological permeation models. The resultant profile is contrasted with the results
of MD simulations; these, while qualitatively similar, are
quantitatively vastly different.
aqueous regions. Electrical forces
acting on ion(s) in the pore are determined by solving Poisson's
equation, taking into account the peptide charge distribution
(determined from the channel structure and molecular force field charge
parameters) and the reaction field due to dielectric variability; if
the pore's effective is chosen different from that of bulk water
there is also a self-energy term. These forces depend upon the s
chosen to describe the pore and the surrounding peptide. The ion's
electrical energy determines the position of the binding site. With an
estimate of the ion's channel diffusion coefficient,
Deff, BD provides the tool for monitoring ion
movement and computing I-V-C profiles.
first consider the electrical
energy. An acceptable potential profile must have significant energy
wells near the channel entrances (to create the binding sites) but
cannot have large internal barriers (lest translocation be forbidden). No plausible choices are satisfactory, regardless of the permittivity ascribed to the peptide and membrane regions. With a large pore the
wells are too shallow to create preferential occupancy near the channel
mouths; with a small pore the internal barrier to translocation is
enormous, essentially forbidding ion passage. But it gets worse. With a
pore large enough to permit ion passage, I-V profiles can be
reproduced with a reasonable choice of Deff. However, the current cannot be made to saturate at high C. Furthermore, monovalent cation occupancy is not limited to regions near the channel
mouth; the ion is essentially uniformly distributed throughout the
channel. However, some electrostatic predictions do work. Models that
account for I-V data predict that anion entry and divalent cation
passage through the channel are both forbidden. But even here there is
a bitter element. Divalent cations are known to block gramicidin;
continuum electrostatics forbids their binding. Continuum
electrostatics has successfully described electrical behavior in the
potassium channel pore. What is it about gramicidin that makes it
resistant to similar modeling? The obvious culprit is its exceptionally
long single-file domain where water molecules, surrounded by the
deformable gramicidin backbone embedded in a non-permittive domain, act
as electrical transducers, transmitting electric fields in ways that
have no continuum analog. In contrast, the filter of the potassium
channel pore is a bit more than half as long; that of the chloride
channel is probably even shorter (Dutzler et al., 2002).
In both, the filters abut much wider aqueous intra-peptide regions that
can significantly influence the pore's electrical behavior. But there
is something else that may make gramicidin exceptionally hard to model.
The electrical data are so extensive and the structural data so highly
resolved that theoretical predictions are left with fairly little
"wiggle room."
). In chloride channels conduction and gating are
intimately coupled (Richard and Miller, 1990;
Pusch et al., 1995), suggesting a transductive process
involving structural changes in the filter, something unlikely to be
susceptible to continuum electrostatic description. What about the
simulation of ionic free energy profiles? The present state of the art
implies internal barriers in gramicidin that are 4-5 times those
inferred from experiments; this suggests that theoretical studies of
potassium channel energetics may well be more uncertain than one would
like. Until a computational model quantitatively accounts for barium and sodium block in potassium channels, it may be advisable to take
simulational predictions with a sizeable pinch of salt.
| |
ACKNOWLEDGMENTS |
|---|
This work was supported by National Institutes of Health grant GM 28643.
| |
FOOTNOTES |
|---|
Address reprint requests to Peter C. Jordan, Department of Chemistry, MS 015, Brandeis University, P. O. Box 549110, Waltham, MA 02454-9110. Tel.: 781-736-2540; Fax: 781-736-2516; E-mail: jordan{at}brandeis.edu.
Submitted April 30, 2002 and accepted for publication May 8, 2002
| |
REFERENCES |
|---|
Biophys J, September 2002, p. 1235-1236, Vol. 83, No. 3
© 2002 by the Biophysical Society 0006-3495/02/09/1235/02 $2.00
This article has been cited by other articles:
 |
|
 |
|
  T. Bastug and S. Kuyucak Energetics of Ion Permeation, Rejection, Binding, and Block in Gramicidin A from Free Energy Simulations Biophys. J., June 1, 2006; 90(11): 3941 - 3950. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
