1: %\documentclass[aps,preprint,groupedaddress,showpacs]{revtex4}
2: %\documentclass[aps,preprint,superscriptaddress]{revtex4}
3: \documentclass[aps,twocolumn,groupedaddress,showpacs]{revtex4}
4: %\documentclass[aps,twocolumn,superscriptaddress]{revtex4}
5: \usepackage{amsmath}
6: \usepackage{amssymb}
7: \usepackage{amsfonts}
8: \usepackage[dvips]{graphicx}
9: \usepackage[]{caption}
10: \usepackage[]{epsfig}
11: \bibliographystyle{apsrev}
12:
13: \begin{document}
14: %\bibliographystyle{apsrev}
15: %\bibliographystyle{prsty}
16:
17: \title{Electric field-controlled water permeation coupled to ion transport through a nanopore}
18:
19: \author{J. Dzubiella}
20: \email[e-mail address: ] {jd319@cam.ac.uk}
21: \affiliation{University Chemical Laboratory,
22: Lensfield Road,
23: Cambridge CB2 1EW,
24: United Kingdom}
25: \author{R. J. Allen}
26: \affiliation{University Chemical Laboratory,
27: Lensfield Road,
28: Cambridge CB2 1EW,
29: United Kingdom}
30:
31: \author{J.-P. Hansen}
32: \affiliation{University Chemical Laboratory,
33: Lensfield Road,
34: Cambridge CB2 1EW,
35: United Kingdom}
36:
37: \date{\today}
38:
39: \begin{abstract}
40: We report Molecular Dynamics (MD) simulations of a generic
41: hydrophobic nanopore connecting two reservoirs which are initially at
42: different Na${\bf{^+}}$ concentrations, as in a biological cell. The nanopore is
43: impermeable to water under equilibrium conditions, but the strong electric field
44: caused by the ionic concentration gradient drives water molecules
45: in. The density and structure of water in the
46: pore are highly field dependent. In a typical simulation run, we
47: observe a succession of cation passages through the pore,
48: characterized by approximately bulk mobility. These ion passages
49: reduce the electric field, until the pore empties of water and closes
50: to further ion transport, thus providing a possible mechanism for
51: biological ion channel gating.
52: \end{abstract}
53:
54: \pacs{61.20.Ja, 68.08.Bc, 87.16.Ac, 87.16.Uv}
55:
56: \maketitle
57:
58: Water and ion permeation of nanopores is a key issue for
59: biological membrane-spanning ion channels and aquaporins, as well as for
60: materials like zeolites, gels and carbon nanotubes. Recent simulations
61: report intermittent filling of hydrophobic nanopores by water under
62: equilibrium conditions \cite{hummer:nature,beckstein:jpc,
63: allen:prl}. However, imbalances in ion
64: concentrations in the inside and outside of cell membranes create
65: strong electric fields \cite{hille}. Experiments on water
66: near interfaces show that strong fields can induce considerable
67: electrostriction of water \cite{toney:nature}. The nonequilibrium
68: behavior of confined water and ions in strong fields should therefore
69: be very important for ion permeation and ion channel function. In
70: particular, biological ion channels ``open'' and ``close'' to ion
71: transport in response to changes in the electric field across the
72: membrane. This behavior, know as {\it voltage gating}, is crucial to
73: their function, but its mechanism is so far not well understood
74: \cite{hille}. Related approaches \cite{kuyucak:2001} to ion transport include
75: the application of a uniform external electric field \cite{crozier:prl,chung:2002}, more
76: specific models of particular proteins
77: \cite{berneche:nature,tieleman:2001,roux:1994,lopez:2002}, Brownian
78: Dynamics \cite{chung:2002} and
79: continuum theories \cite{nonner:1998}. Here we present the results of MD simulations in which a
80: strong electric field across the pore is explicitly created by an ionic
81: charge imbalance, as in a cell. We follow the relaxation of this
82: nonequilibrium system to equilibrium.
83:
84: Our generic model ion channel consists of a cylindrical hydrophobic pore of
85: length $L_{\rm p}= 16$\AA~ and radius $R_{\rm p}=5$ to $7$\AA, through a membrane
86: slab which separates two reservoirs containing water and Na$^+$ and Cl$^-$
87: ions, as shown in Fig. 1. One reservoir has initial
88: concentrations $c_{\rm {Na^{+}}}\approx 0.9$M (12 cations) and $c_{\rm
89: Cl^{-}}\approx 0.6$M (8 anions), while for the second reservoir, $c_{\rm
90: Na^{+}}\approx 0.3$M (4 cations) and $c_{\rm Cl^{-}}\approx 0.6$M (8
91: anions). This imbalance of charge generates an average electric field
92: of $0.37$V/\AA~ across the membrane. These ion concentrations and
93: electric field are typically five times larger than under ''normal``
94: physiological conditions, but they could be achieved in the course of a
95: rare, large fluctuation at the pore entrance. The enhanced
96: ion concentrations in the initial state were chosen to improve the
97: signal-to-noise ratio in the simulation and to allow the detection of
98: novel transport mechanisms.
99: The chosen pore dimensions are
100: comparable to those of the selectivity filter of a K$^{+}$ channel \cite{zhou:nature,berneche:nature}.
101:
102:
103:
104: The simulation cell contains two pores in sequence along the z-axis,
105: one of which is shown in Fig.1. The reservoir to the right of this
106: pore thus forms the left reservoir for the other pore. Due to
107: periodic boundary conditions, the right reservoir of the latter
108: is also the left reservoir of the first. In this arrangement, the
109: simplest to allow the use of full three-dimensional periodic
110: boundaries, the flows of ions in response to the concentration
111: gradient will be anti-parallel in the two channels. The relaxation
112: towards equilibrium, where the two reservoirs are individually
113: electroneutral, will thus involve an indirect coupling between the two
114: pores.
115:
116: The water molecules are modelled by the SPC/E potential
117: \cite{berendsen:jpc} which consists
118: of an O atom, carrying an electric charge
119: $q=-0.8476$e, and two H atoms with $q=0.4238$e. The O-atoms on
120: different water molecules interact via a Lennard-Jones (LJ) potential
121: with parameters $\epsilon=0.6502$kJmol$^{-1}$ and $\sigma=3.169$\AA~. The
122: model is rigid,
123: with OH bond length $1$\AA~ and HOH angle $109.5^\circ$. The Na$^+$
124: ions have LJ parameters $\epsilon=0.3592$kJmol$^{-1}$, $\sigma=2.73$\AA~ and $q=+$e
125: and for the Cl$^-$, $\epsilon=0.1686$kJmol$^{-1}$, $\sigma=4.86$\AA~ and
126: $q=-$e \cite{spohr:1999}.
127:
128: Ions and water O-atoms interact with the confining pore and membrane
129: surfaces by a potential of the LJ form
130: $V=\epsilon'[(\sigma'/r)^{12}-(\sigma'/r)^{6}]$, where $r$ is the
131: orthogonal distance from the nearest confining surface. The potential
132: parameters are $\epsilon'=1.0211$kJmol$^{-1}~$ and
133: $\sigma'=0.83$\AA~. If $R_{\rm p}$ is the geometric radius of the
134: cylindrical pore, one may conveniently define an effective radius $R$
135: by the radial distance from the cylinder axis at which the interaction
136: energy of a water O-atom with the confining surface is $k_{B}T$,
137: leading at room temperature to $R\approx R_{\rm p}-2$\AA~; similarly the
138: effective length of the pore is $L\approx L_{\rm p}+4$\AA~. Ion-water,
139: ion-ion, water-surface and ion-surface cross terms are defined using
140: the usual Lorentz-Berthelot combining rules.
141: Polarizability of
142: the membrane \cite{allen:jpcm} and of the water molecules and ions is
143: neglected.
144:
145: The total simulation cell including both of the channels is of
146: dimensions $l_{x}=l_{y}=23.5\pm 0.3$\AA~ and
147: $l_{z}=112.9\pm1.5$\AA~ and contains 1374
148: water molecules, 16 Na$^+$ and 16 Cl$^-$ ions. Molecular Dynamics
149: simulations were carried out with the DLPOLY2 package \cite{dlpoly},
150: using the Verlet algorithm \cite{frenkelsmit,allen} with a timestep of 2fs. The
151: pressure was maintained at $P=1$bar and the temperature at $T=300$K using
152: a Berendsen barostat and thermostat \cite{berendsen:jcp}. Electrostatic
153: interactions were calculated using the particle-mesh Ewald
154: method \cite{essmann:jcp}.
155:
156:
157: \begin{figure}[htb]
158: \begin{center}
159: \epsfig{file=fig1.ps, width=8cm, angle=0}
160: \caption{Simulation snapshots. Molecular configurations (a) before a cation
161: (dark grey spheres) permeates the channel and (b), 10 ps later, while it goes
162: through. Anions are shown as light grey spheres. Only half of the
163: periodically repeated simulation cell is shown.}
164: \label{profiles}
165: \end{center}
166: \end{figure}
167:
168: \begin{figure}[htb]
169: \begin{center}
170: \epsfig{file=fig2.ps, width=8cm, angle=0}
171: \caption{Electric field in and around the pore. (a), Magnitude
172: and direction of the electric field, depicted by vectors on a
173: two dimensional rectangular grid in the $x$-$z$-plane. The contour
174: of the membrane pore is sketched as black solid line. (b),
175: Averaged $z$-component $E_{z}(z)$ of the electric field inside the
176: channel.}
177: \label{profiles}
178: \end{center}
179: \end{figure}
180:
181:
182:
183: \begin{figure}[htb]
184: \begin{center}
185: \epsfig{file=fig3.ps, width=6cm, angle=-90}
186: \caption{Occupancy and structure of water inside the pore. The number of
187: water molecules inside the channel $n_{\rm H_{2}O}$ is plotted as a
188: function of time $t$. The
189: shades of grey code the average magnitude of the electric field inside the
190: pore: from black to light grey: $E \approx 0.37 V/$\AA, $E \approx 0.25 V/$\AA,
191: $E \approx 0.15V/$\AA~, $E \approx 0.10V/$\AA. The inset shows the
192: corresponding radial density profiles
193: of the water molecules inside the channel averaged over periods of
194: constant field. $\rho_{0}$ is the bulk density of water in
195: equilibrium. $n_{\rm bulk}$ (eq. (2)) is the expected number of water
196: molecules inside the pore if bulk density is assumed.}
197: \label{profiles}
198: \end{center}
199: \end{figure}
200:
201:
202: \begin{figure}[htb]
203: \begin{center}
204: \epsfig{file=fig4.ps, width=6cm, angle=-90}
205: \caption{Cation trajectories inside the pore and in its vicinity. The $z$-coordinate of
206: the cations is plotted versus time for three successive permeations
207: in one typical simulation run. The channel is located between
208: $z=-8{\rm\AA}$ and $z=8{\rm\AA}$ marked by the two long dashes
209: lines. $t=0$ defines the time at which the ions are located at the
210: entrance of the channel. The shades of grey code the average magnitude of
211: electric field experienced by the ions, as in Fig.3}
212: \label{profiles}
213: \end{center}
214: \end{figure}
215: Water permeation of the pore is strongly affected by the electric
216: field. The effective channel radius chosen for
217: most of the simulations ($R=3$\AA)
218: is such that under equilibrium conditions (i.e. with equal numbers of
219: anions and cations on both sides and hence no electric field), the channel is
220: empty of water and ions \cite{hille}. However, the ionic charge
221: imbalance across the the membrane causes the pore to fill
222: spontaneously with water. The electric field
223: throughout the system was monitored by measuring the electrostatic
224: force on phantom test particles on a three dimensional
225: grid \cite{crozier:prl}. Fig.~2a shows the average local electric
226: field around one pore before the first
227: ion moves through a channel. It is nearly zero in the reservoirs.
228: Inside the pore the field is
229: very strong ($\sim 0.37$V/\AA) and has a small inward radial component. The profile
230: of the $z$-component of the field $E_{z}$ is shown in Fig. 2b and is
231: constant inside the pore.
232:
233: During
234: the course of the simulation, a number of Na$^+$ ions move through the
235: pore. Each of these events changes the reservoir charge imbalance and
236: reduces the electric field in the pore. This has a dramatic effect on
237: the behavior of the water, as shown in Fig.~3 in which the number
238: $n_{\rm H_{2}O}$ of water molecules inside one pore is plotted as a
239: function of time for a typical simulation run.
240:
241: Initially, the water
242: in the pore undergoes strong electrostriction, comparable to
243: experimental observations \cite{toney:nature}, with an average density
244: \begin{equation}\label{mth1}
245: \rho\approx n_{\rm H_{2}O}/(\pi R^{2}L),
246: \end{equation}
247: twice as large as that of bulk water in
248: equilibrium. If we assume bulk density of water $\rho_{0}$ inside a
249: channel of
250: radius $R$ and length $L$ we expect an average number $n_{\rm bulk}$
251: of molecules inside the channel with
252: \begin{equation}\label{mth2}
253: n_{\rm bulk}=\rho_{0}(\pi R^{2}L).
254: \end{equation}
255: $n_{\rm bulk}$ is indicated with an arrow on the right-hand side in
256: Fig. 3.
257: At each ion crossing, the number $n_{\rm H_2O}$ of water
258: molecules inside the pore drops but is still larger than $n_{\rm
259: bulk}$. In this particular simulation run shown in Fig.3 three
260: cations went through this pore.
261: After the third ion crosses ($t\approx 2.25$ns), the
262: electric field is no longer strong enough to sustain channel filling
263: and the pore spontaneously empties of water, thereby becoming
264: impermeable to further ion transport. However the other pore in the
265: simulation cell remains filled and the final ion crossing eventually
266: occurs through this pore which after that also empties of
267: water. Finally equilibrium is restored to the system and both channels
268: are empty of water. On repeating the simulation 5 times we observe
269: that the closing of one pore after the third ion passed through it
270: occurs in all runs.
271:
272: The structure of water in the filled pore is strongly affected by the field, as
273: shown in the inset of Fig.~3.
274: Before the first ion crossing, water forms clear layers near
275: the pore wall and along the z-axis. The central layer disappears after
276: the first ion crossing and the outer layer becomes less
277: well-defined. After a further ion crossing, water is more-or-less
278: evenly distributed.
279:
280: Ion transport through the pore is found to occur essentially at
281: constant velocity. Fig. 1
282: shows snapshots from a typical simulation run just before (a) and
283: while (b) a sodium ion passes through the channel. Within the
284: reservoirs the anions and cations diffuse among the water
285: molecules. When a cation in the Na$^{+}$-rich reservoir comes
286: close to the channel entrance, it experiences the strong axial field
287: shown in Fig. 2 and is dragged into the channel. Analysis of $15$
288: simulation runs, with the same initial charge imbalance but
289: different initial configurations, shows that once the first ion enters
290: the channel, it moves with a constant velocity which is approximately
291: the same in all runs, and then reverts to diffusive motion at the
292: other end of the channel.
293:
294: Fig. 4 shows typical cation positions along
295: the z-axis, as a function of time, for the first, second and third ion
296: crossings, as shown in Fig. 3. The second ion also traverses the
297: channel at constant, somewhat reduced velocity, although it appears to
298: pause for approximately 10ps at the channel entrance, perhaps in order
299: to shed its bulk-like solvation shell. We observe that this ``pausing
300: time'' is rather widely distributed between simulation runs.
301: The cation mobility $\mu_+$, defined by $\vec
302: v = \mu_{+} e \vec E$, can be calculated from the slopes of the
303: trajectories in Fig. 4, together with the measured electric fields, as in
304: Fig.2. The resulting values are $\mu_{+}\approx 4.5\times
305: 10^{11}{\rm\;s\,kg^{-1}}$ for the first ion, $\mu_{+}\approx 3.8\times
306: 10^{11}{\rm\;s\,kg^{-1}}$ for the second ion, and $\mu_{+}\approx 2.4\times
307: 10^{11}{\rm\;s\,kg^{-1}}$ for the third ion. These values are close to
308: the value of $\mu_{+}\approx2.3\times 10^{11}{\rm\;s\,kg^{-1}}$
309: obtained from the self-diffusion constant in the reservoir $D_+$,
310: using Einstein's relation $\mu_{+}=D_{+}/k_{B}T$, but seem to increase
311: with the magnitude of the electric field inside the pore. This
312: enhancement of the mobility correlates with the change of
313: structure of the water inside the channel, shown in the inset of Fig. 3.
314: The tetrahedral hydrogen bond network which water forms under
315: equilibrium conditions is disrupted inside the pore under high
316: electric fields.
317:
318: Simulations of wider ($R=5$\AA) pores and with different lengths give
319: qualitatively the same results. The critical electric
320: field for water permeation is, however, sensitive to the pore radius
321: and length. This suggests that voltage-dependent gating in ion
322: channels, if it were to occur by changes in water permeation of a
323: hydrophobic section of the pore \cite{allen:prl,hille}, might be strongly
324: dependent on channel geometry.
325:
326: The key finding which emerges from our simulations is the strong
327: correlation between water and ion behavior under non-equilibrium
328: conditions. Ionic charge imbalance across the membrane induces water
329: permeation of the hydrophobic pore and thus makes it permeable to
330: ions. This suggests that voltage gating of ion channels may be linked to the
331: coupling between water and ion permeation in pores far from
332: equilibrium. The structure and density of water in the pore is
333: dramatically
334: affected by the strong electric field. The passage of a cation
335: through the channel causes an abrupt jump of the electric
336: field, and an ensuing jump in the number of water molecules inside the
337: pore. Ion passage through the pore occurs at constant velocity and
338: with a mobility coefficient similar to that of the bulk solution at
339: equilibrium.
340:
341: \acknowledgments
342: The authors are grateful to
343: Jane Clarke and Michele Vendruscolo for a careful reading of the
344: manuscript. This work was supported in part by the EPSRC. R.~J.~A. is grateful
345: to Unilever for a Case award.
346: \begin{thebibliography}{10}
347: \expandafter\ifx\csname bibnamefont\endcsname\relax
348: \def\bibnamefont#1{#1}\fi
349: \expandafter\ifx\csname bibfnamefont\endcsname\relax
350: \def\bibfnamefont#1{#1}\fi
351: \expandafter\ifx\csname url\endcsname\relax
352: \def\url#1{\texttt{#1}}\fi
353: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
354: \expandafter\ifx\csname bibinfo\endcsname\relax \def\bibinfo#1#2{#2}\fi
355: \expandafter\ifx\csname eprint\endcsname\relax \def\eprint#1{#1}\fi
356:
357: \bibitem{hummer:nature}
358: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Hummer}},
359: \bibinfo{author}{\bibfnamefont{J.~C.} \bibnamefont{Rasaih}},
360: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~P.}
361: \bibnamefont{Nowortya}}, \bibinfo{journal}{Nature}
362: \textbf{\bibinfo{volume}{414}}, \bibinfo{pages}{188} (\bibinfo{year}{2001}).
363:
364: \bibitem{beckstein:jpc}
365: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Beckstein}},
366: \bibinfo{author}{\bibfnamefont{P.~C.} \bibnamefont{Biggin}},
367: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.~S.~P.}
368: \bibnamefont{Sansom}}, \bibinfo{journal}{J. Phys. Chem.}
369: \textbf{\bibinfo{volume}{105}}, \bibinfo{pages}{12902}
370: (\bibinfo{year}{2001}).
371:
372: \bibitem{allen:prl}
373: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Allen}},
374: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Melchionna}},
375: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.-P.}
376: \bibnamefont{Hansen}}, \bibinfo{journal}{Phys. Rev. Lett.}
377: \textbf{\bibinfo{volume}{89}}, \bibinfo{pages}{175502}
378: (\bibinfo{year}{2002}).
379:
380: \bibitem{hille}
381: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Hille}},
382: \emph{\bibinfo{title}{Ionic Channels of Excitable Membranes}}
383: (\bibinfo{publisher}{Sinauer}, \bibinfo{year}{1992}).
384:
385: \bibitem{toney:nature}
386: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Toney}},
387: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{368}},
388: \bibinfo{pages}{444} (\bibinfo{year}{1994}).
389:
390: \bibitem{kuyucak:2001}
391: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Kuyucak}},
392: \bibinfo{author}{\bibfnamefont{O.~S.} \bibnamefont{Andersen}},
393: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{S.-H.} \bibnamefont{Chung}},
394: \bibinfo{journal}{Rep. Prog. Phys.} \textbf{\bibinfo{volume}{64}},
395: \bibinfo{pages}{1427} (\bibinfo{year}{2001}).
396:
397: \bibitem{crozier:prl}
398: \bibinfo{author}{\bibfnamefont{P.~S.} \bibnamefont{Crozier}},
399: \bibinfo{author}{\bibfnamefont{R.~L.} \bibnamefont{Rowley}},
400: \bibinfo{author}{\bibfnamefont{N.~B.} \bibnamefont{Holladay}},
401: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Henderson}},
402: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{D.~D.}
403: \bibnamefont{Busath}}, \bibinfo{journal}{Phys. Rev. Lett.}
404: \textbf{\bibinfo{volume}{86}}, \bibinfo{pages}{2467} (\bibinfo{year}{2001}).
405:
406: \bibitem{chung:2002}
407: \bibinfo{author}{\bibfnamefont{S.-H.} \bibnamefont{Chung}},
408: \bibinfo{author}{\bibfnamefont{T.~W.} \bibnamefont{Allen}}, \bibnamefont{and}
409: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Kuyucak}},
410: \bibinfo{journal}{Biophys. J.} \textbf{\bibinfo{volume}{83}},
411: \bibinfo{pages}{262} (\bibinfo{year}{2002}).
412:
413: \bibitem{berneche:nature}
414: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Berneche}} \bibnamefont{and}
415: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Roux}},
416: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{414}}, \bibinfo{pages}{73}
417: (\bibinfo{year}{2001}).
418:
419: \bibitem{tieleman:2001}
420: \bibinfo{author}{\bibfnamefont{D.~P.} \bibnamefont{Tieleman}},
421: \bibinfo{author}{\bibfnamefont{P.~C.} \bibnamefont{Biggin}},
422: \bibinfo{author}{\bibfnamefont{G.~R.} \bibnamefont{Smith}}, \bibnamefont{and}
423: \bibinfo{author}{\bibfnamefont{M.~S.~P.} \bibnamefont{Sansom}},
424: \bibinfo{journal}{Quart. Rev. Biophys.} \textbf{\bibinfo{volume}{34}},
425: \bibinfo{pages}{473} (\bibinfo{year}{2001}).
426:
427: \bibitem{roux:1994}
428: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Roux}} \bibnamefont{and}
429: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Karplus}},
430: \bibinfo{journal}{Ann. Rev. Biophys. Biomol. Str.}
431: \textbf{\bibinfo{volume}{23}}, \bibinfo{pages}{731} (\bibinfo{year}{1994}).
432:
433: \bibitem{lopez:2002}
434: \bibinfo{author}{\bibfnamefont{C.~F.} \bibnamefont{Lopez}},
435: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Montal}},
436: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Blasie}},
437: \bibinfo{author}{\bibfnamefont{M.~L.} \bibnamefont{Klein}}, \bibnamefont{and}
438: \bibinfo{author}{\bibfnamefont{P.~B.} \bibnamefont{Moore}},
439: \bibinfo{journal}{Biophys. J.} \textbf{\bibinfo{volume}{83}},
440: \bibinfo{pages}{1259} (\bibinfo{year}{2002}).
441:
442: \bibitem{nonner:1998}
443: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Nonner}} \bibnamefont{and}
444: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Eisenberg}},
445: \bibinfo{journal}{Biophys. J.} \textbf{\bibinfo{volume}{75}},
446: \bibinfo{pages}{1287} (\bibinfo{year}{1998}).
447:
448: \bibitem{zhou:nature}
449: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Zhou}},
450: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{Morais-Cabral}},
451: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Kaufman}}, \bibnamefont{and}
452: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{MacKinnon}},
453: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{414}}, \bibinfo{pages}{43}
454: (\bibinfo{year}{2001}).
455:
456: \bibitem{berendsen:jpc}
457: \bibinfo{author}{\bibfnamefont{H.~J.~C.} \bibnamefont{Berendsen}},
458: \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Grigera}},
459: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{T.~P.}
460: \bibnamefont{Straatsma}}, \bibinfo{journal}{J. Phys. Chem.}
461: \textbf{\bibinfo{volume}{91}}, \bibinfo{pages}{6269} (\bibinfo{year}{1987}).
462:
463: \bibitem{spohr:1999}
464: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Spohr}},
465: \bibinfo{journal}{Electrochim. Acta} \textbf{\bibinfo{volume}{44}},
466: \bibinfo{pages}{1697} (\bibinfo{year}{1999}).
467:
468: \bibitem{allen:jpcm}
469: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Allen}},
470: \bibinfo{author}{\bibfnamefont{J.-P.} \bibnamefont{Hansen}},
471: \bibnamefont{and}
472: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Melchionna}},
473: \bibinfo{journal}{J. Phys.: Condens. Matt.} \textbf{\bibinfo{volume}{15}},
474: \bibinfo{pages}{5297} (\bibinfo{year}{2003}).
475:
476: \bibitem{dlpoly}
477: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Smith}} \bibnamefont{and}
478: \bibinfo{author}{\bibfnamefont{T.~R.} \bibnamefont{Forester}}
479: (\bibinfo{year}{1999}), \bibinfo{note}{the DLPOLY\_2 User Manual}.
480:
481: \bibitem{frenkelsmit}
482: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Frenkel}} \bibnamefont{and}
483: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Smit}},
484: \emph{\bibinfo{title}{Understanding Molecular Simulation: From Algorithms to
485: Applications}} (\bibinfo{publisher}{Academic Press}, \bibinfo{year}{1996}).
486:
487: \bibitem{allen}
488: \bibinfo{author}{\bibfnamefont{M.~P.} \bibnamefont{Allen}} \bibnamefont{and}
489: \bibinfo{author}{\bibfnamefont{D.~J.} \bibnamefont{Tildesley}},
490: \emph{\bibinfo{title}{Computer Simulation of Liquids}}
491: (\bibinfo{publisher}{Clarendon Press}, \bibinfo{year}{1987}).
492:
493: \bibitem{berendsen:jcp}
494: \bibinfo{author}{\bibfnamefont{H.~J.~C.} \bibnamefont{Berendsen}},
495: \bibinfo{author}{\bibfnamefont{J.~P.~M.} \bibnamefont{Postma}},
496: \bibinfo{author}{\bibfnamefont{W.~F.~V.} \bibnamefont{Gunsteren}},
497: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{DiNola}}, \bibnamefont{and}
498: \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Haak}},
499: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{81}},
500: \bibinfo{pages}{3684} (\bibinfo{year}{1984}).
501:
502: \bibitem{essmann:jcp}
503: \bibinfo{author}{\bibfnamefont{U.}~\bibnamefont{Essmann}},
504: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Perera}},
505: \bibinfo{author}{\bibfnamefont{M.~L.} \bibnamefont{Berkowitz}},
506: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Darden}}, \bibnamefont{and}
507: \bibinfo{author}{\bibfnamefont{H.~L. L.~G.} \bibnamefont{Pedersen}},
508: \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{103}},
509: \bibinfo{pages}{8577} (\bibinfo{year}{1995}).
510:
511: \end{thebibliography}
512:
513: \end{document}
514:
515:
516:
517: