The 3.5 ångström X−ray structure of the human connexin26 gap junction channel is unlikely that of a fully open channel
© Zonta et al; licensee BioMed Central Ltd. 2013
Received: 20 November 2012
Accepted: 12 February 2013
Published: 27 February 2013
The permeability of gap junction channels to metabolites, and not simply to small inorganic ions, is likely to play an important role in development, physiology as well as in etiology of several diseases. Here, we combined dual patch clamp and fluorescence imaging techniques with molecular dynamics (MD) simulations to investigate the permeation of calcein, a relatively large fluorescent tracer (MW 622 Da) through homomeric gap junction channels formed by wild type human connexin26 (hCx26wt) protomers. Our experimental data indicate that the unitary flux of calcein driven by a 125 μM concentration difference is Jpore = 226 molecule/s per channel. In the light of Eyring transition state theory adapted for the liquid phase, this value corresponds to an energy barrier of ~20 k B T (where k B is the Boltzmann constant and T is absolute temperature). The barrier predicted by our MD simulations, based on the 3.5 Å X–ray structural model of the hCx26wt gap junction channel, is ~45 k B T. The main contributions to the energetics of calcein permeation originated from the interaction between the permeating molecule and the charged aminoacids lining the channel pore. Assigning a fake zero total charge to the calcein molecule yielded a value for the barrier height compatible with the experimental data. These results can be accounted for by two different (although not mutually exclusive) hypotheses: (1) the X–ray model of the hCx26wt gap junction channel is not representative of a fully open state; (2) post translational modifications affecting the hCx26wt protein in our expression system differed from the modifications undergone by the proteins in the conditions used to obtain the crystal structure. Hypothesis (1) is compatible with data indicating that, only 10% or less of the channels forming a gap junction plaque are in the open state, and therefore the averaging procedure intrinsic in the generation of the crystal structure data more closely reflects that of a closed channel. Hypothesis (2) is compatible with recent mass spectrometry data and implies that the charge of several amino acid side chains may have been altered, thus modifying substantially the permeation properties of the channels in living cells.
KeywordsConnexin26 Calcein Dual patch clamp Fluorescence imaging Umbrella sampling Potential of mean force Transition rate Model
Gap junction channels mediate communication between adjacent cells by allowing the passage of a variety of cytoplasmic molecules. They are formed by the head−to−head docking of two connexin protein hexamers, known as hemichannels or connexons, located in two adjacent cells . Several studies showed that the permeation of cytoplasmic molecules through gap junction channels is fundamental in development and physiology, but also in the etiology of several diseases . Second messengers, amino acids, nucleotides, glucose and its metabolites can permeate through at least some types of gap junction channels [3, 4]. However, current understanding of the permeation properties and mechanisms is largely incomplete. Indeed, the unitary permeability of homomeric gap junction channels do not correlate well with ionic conductance and with presumptive pore sizes. The problem is exacerbated by the fact that gap junction channels in living cells can be formed by different connexin isoforms. Furthermore, most of the permeation properties of gap junction channels can, in principle, be dynamically regulated in response to external stimuli, such as voltage, pH or ionic concentrations [3, 4].
The structure of connexin proteins and their assemblies was largely unknown until the publication of a model based on high resolution (3.5 Å) X–ray data of a hCx26wt channel . The X–ray model permits to tackle issues left unresolved by previous models based on lower resolution data [6–9] such as the correct position of transmembrane helixes and the structure of extracellular regions. It also enables the study, by use of computational techniques, of ion permeation pathways [10, 11] and the prediction of unknown structures (wild type human connexin30, hCx30wt) .
With its 226 amino acids, hCx26wt is one of the smallest member of the connexin family. Mutations of GJB2, the gene encoding hCx26wt, are implicated in both syndromic and nonsyndromic deafness . The X–ray data indicate that hCx26wt comprises four transmembrane helixes (TM1, TM2, TM3 and TM4), which are connected by two extracellular loops (E1, E2) and one cytoplasm loop (CL) . When assembled in hexamers, hCx26wt subunits create an aqueous pore in the plasma membrane, whose walls are formed by TM1 and TM2, plus the N–terminus (NT) that folds inside the pore at the cytoplasmic mouth of the channel. The mouth is created by the CL and part of the NT and hosts several positively charged residues. On the extracellular side, instead, hCx26wt presents an accumulation of negatively charged residues .
Here, we measured the unitary permeability of homomeric gap junction channels formed by hCx26wt to calcein, a widely used inorganic fluorescent tracer. We paralleled the experimental work with MD simulations  based on the 3.5 Å X–ray structure of hCx26wt . Term of comparison between experiments and simulation is the transition rate of calcein through the channel, i.e. the number of calcein molecule that are able to traverse the channel per unit time. Our results indicate that the 3.5 Å X–ray structure of hCx26wt is unlikely that of fully open channel and suggest that permeation properties of the channel may be significantly affected by post–translational modification of critical residues lining the pore.
Results and discussion
Experimental determination of calcein transition rate
MD analysis of the permeation process
The calcein transition rate k c , estimated from the PMF profile as described in the Methods, is twelve orders of magnitude smaller than kExp, meaning that no calcein molecule would ever traverse a homomeric hCx26wt gap junction channel with the structure predicted by the 3.5 Å X–ray data .
where A is the section area of the channel, and z0 a reference position. Function U S (z) for the hCx26wt connexon is plotted in Figure 2A (bottom panel). Not only is the maximum value of U S (z) significantly smaller than W c , but U S (z) is also qualitatively different from the PMF profile. Based on this analysis we conclude that the entropic contribution to the PMF due to the variable radius of the pore is not the dominant factor and the channel is not closed from a purely entropic point of view.
In this paper we measured the unitary flux of calcein through hCx26wt gap junction channels, and compared the experimentally determined value to that predicted by MD simulations based on the 3.5 Å X–ray structural data . Term of comparison is the unitary transition rate, i.e. the number of calcein molecules that are able to transit trough a single channel per unit time. Simulations were performed with two different charge states for the calcein molecule. In the first case calcein had all the carboxyl groups deprotonated, as expected at physiological pH. In the other case, calcein was protonated and set to zero total charge. Our simulations indicate that a calcein molecule with a presumptive physiological charge is unable to traverse the channel due to the large energy barrier it faces (45.2 k B T). In contrast, the predicted transition rate for a calcein molecule with zero charge is compatible with the experimentally determined value.
Based on this analysis we conclude that the structural model of the hCx26wt channel derived from the 3.5 Å X–ray data  is not permeable to calcein (even after MD relaxation) and the blockade is essentially electrostatic. Our conclusion is in contrast with the proposal of Maeda et al.  that the model represents an open channel. This proposal was based on the facts that: (i) unlike the M34A mutant channel structure , there are no obvious obstructions along the pore of the hCx26wt channel; (ii) the crystallization conditions adopted by Maeda et al. are compatible with the formation of channels in the open state (neutral pH without aminosulphonate buffer or any divalent ions). The discrepancy highlighted in the present work can be explained as follows: (1) there is no way to guarantee that the open channel structure was preserved during the partial dehydration and crystallization procedures; (2) in a gap junction plaque, only 10% or less of the channels are in an open state [22–24].
The following considerations lend further support to this conclusion. Structure relaxation during our MD simulations (carried out in a realistic environment) resulted in a widening of the pore, particularly at the cytoplasmic mouth of the channel . Even this wider pore is impermeable to charged calcein. Moreover, as mentioned in the results, the charges of several residues facing the pore may be altered by post translation modifications, which may differ in the mammalian (HeLa) cells used in our experiments with calcein and in the insect cells used by Maeda et al. in their crystallization study . A recent study showed that modification of the charge of these residues is sufficient to recover the correct current−voltage (I−V) relationship and ionic conductance, with the main role played by Met1 . Among these, Met1, Glu42 and Glu47 are crucially located in the region of steepest PMF increase. Our results indicate that charged Met1 and Lys41 play a crucial role in hindering calcein permeation. Note that Lys41 is not a candidate for acetylation based on Ref. . Furthermore, acetylation of this residue would reduce significantly the diameter of the pore at its narrowest point, thus we consider it unlikely. In this scenario, gamma–carboxilation of Glu42 can be a fundamental determinant of channel permeability. Indeed, the carboxylated side chain of Glu42 is well poised for interacting with the amino group of Lys41. This interaction could stabilize the side chain of Lys41 (which instead appears rather mobile in our simulations) reducing the electrostatic potential felt by permeant molecules in the narrowest part of the channel, altogether favoring their transit. Further simulations are required to explore the influence of post translation modifications on the PMF of the channel not only for small inorganic ions but also for large permeant molecules.
Dual whole cell patch–clamp recordings and fluorescence imaging
where c1 is calcein concentration in cell 1 and c2 < c1 is concentration in cell 2. Npore was derived by dividing the total junctional conductance gj back extrapolated between WC1 at WC2 (Figure 1B) by the previously determined single channel conductance γ =115 pS .
Numerical simulation methods
hCx26wt connexon MD model
The fully atomistic model used for the hCx26wt hemichannel was developed in our previous work . Briefly, we completed the published structure , adding the atoms that were missing in the original structure, and then inserting the initial hCx26 connexon configuration in a hole opened in a pre–relaxed membrane bilayer of phospholipids (palmytol posphatidyl choline, POPC). The final membrane configuration comprises 493 phospholipids. The positive net charge of the hCx26 connexon was neutralized with 54 chloride ions; additional pairs of potassium and chloride ions were added to mimic a physiological ionic strength. The system was solvated with a total of 39189 water molecules.
The calcein molecule parameters required by our Molecular Dynamics simulations are not present in any standard library. We parameterized it as described below in two different protonation states: (i) standard charge, as reported by the manufacturer (Invitrogen, C481); (ii) completely protonated, i.e. zero total charge. The initial guess of calcein coordinates was obtained using the GlycoBioChem PRODRG2 server and the JME Molecular Editor provided on the server , from the molecular structure provided by the manufacturer of the calcein moiety used in the experiments (Invitrogen). After this step we refined the coordinates and obtained the parametrization for GAFF force field of the two different protonation state of calcein using the Antechamber package .
Evalutation of PMFs by use of the umbrella sampling technique
from the cytoplasmic to the extracellular side. Here z, our chosen reaction coordinate, is position along the pore axis, K pull = 2000 kJ mol−1 nm−2 is the stiffness of a harmonic spring one end of which moved with constant velocity v = 10 nm/ns (pull rate) along z while the center of mass of the calcein molecule was attached to the opposite spring end and also restrained to move along the pore axis.
where K umb = 1000 kJ mol−1 nm−2 is the elastic force constant and z i is the position of the i–th window center along z axis. The dynamics was initially followed for 1 ns for each window. After obtaining a preliminary PMF profile, we refined the spacing to 1 Å and extended the duration of the simulated dynamics in the region were the PMFs in Figure 5 are rapidly increasing, until we reached convergence at 3.5 ns (meaning that the PMF profiles did not change appreciably by lengthening the simulation). The final total number of windows was n = 41 for both charged and uncharged calcein. Overall, the simulation time used for evaluating the two PMF profiles was in excess of 200 ns. All MD simulations were performed with Gromacs 4.5 software  using the Amber03 force field, in the NTV ensemble . Temperature was kept constant at 300 K using the Berendsen thermostat . Particle Mesh Ewald summation  was used for the long–range electrostatic interactions, with a cut off of 1.0 nm for the direct interactions. The simulation time step was comprised between 1 and 2 fs.
Estimate of transition rate from PMF
which favors its exit from the channel in either direction. This process is several orders of magnitude more probable than the reverse one, i.e. climbing W(z) in the uphill direction. For this reason we computed the above integral over the interval [z0, z M ] considering that the rate–limiting step in the permeation process is determined by reaching coordinate z M , i.e. the apex of the free energy barrier.
Summary of computed quantities
W(z M )(k B T)
Final note: the estimates of the transition rates represent the number of transitions (per unit time) that occur under saturating conditions, i.e. when the wait time between successive transitions is null. To realize such conditions, the bulk calcein concentration in cell 1 must be such that (at least) one calcein molecule is present, at any given time, within the cytoplasmic vestibule of the channel. To assess whether saturation was achieved under our experimental conditions (with a calcein concentration in the patch pipette equal to 125 μm), we simulated the diffusion of calcein inside the cell as a Brownian random walk. The results of this independent set of simulations indicate that the number of calcein molecules diffusing from bulk cytoplasm to the vestibule of an individual hemichannel is 2×104 per second, suggesting that the zero wait state condition is a reasonable assumption.
Cyan Fluorescent Protein
E2: Extracelullar loop 1,2
Wild type human connexin 26
Wild type human connexin 30
Intra cellular solution
Ensemble with fixed number of particles, volume and temperature
Potential of mean force
Region of interest
Standard error of the mean
- TM1 to 4:
Transmembrane helix 1 to 4
We thank Giuseppe Zanotti (Dept. of Biological Chemistry, University of Padua) for useful discussions.
Supported by MIUR PRIN grant no. 2009CCZSES and Telethon grant GGP09137 to FM, and from a University of Padua grant to FZ (prot. GRIC101108).
Computer simulations were performed at the CINECA and CASPUR supercomputer centers.
Pictures of the connexin at molecular level have been obtained by use of Visual Molecular Dynamics (VMD) software.
- Goodenough DA, Paul DL: Gap junctions. Cold Spring Harb Perspect Biol. 2009, 1: a002576-10.1101/cshperspect.a002576.PubMed CentralPubMedView ArticleGoogle Scholar
- Kar R, Batra N, Riquelme MA, Jiang JX: Biological role of connexin intercellular channels and hemichannels. Arch Biochem Biophys. 2012, 524: 2-15. 10.1016/j.abb.2012.03.008.PubMed CentralPubMedView ArticleGoogle Scholar
- Harris AL: Connexin channel permeability to cytoplasmic molecules. Prog Biophys Mol Biol. 2007, 94: 120-143. 10.1016/j.pbiomolbio.2007.03.011.PubMed CentralPubMedView ArticleGoogle Scholar
- Ek-Vitorin JF, Burt JM: Structural basis for the selective permeability of channels made of communicating junction proteins. Biochim Biophys Acta. 2013, 1828: 51-68. 10.1016/j.bbamem.2012.02.003.PubMed CentralPubMedView ArticleGoogle Scholar
- Maeda S, Nakagawa S, Suga M, Yamashita E, Oshima A, Fujiyoshi Y, Tsukihara T: Structure of the connexin 26 gap junction channel at 3.5 A resolution. Nature. 2009, 458: 597-602. 10.1038/nature07869.PubMedView ArticleGoogle Scholar
- Unger VM, Kumar NM, Gilula NB, Yeager M: Three-dimensional structure of a recombinant gap junction membrane channel. Science. 1999, 283: 1176-1180. 10.1126/science.283.5405.1176.PubMedView ArticleGoogle Scholar
- Fleishman SJ, Unger VM, Yeager M, Ben-Tal N: A Calpha model for the transmembrane alpha helices of gap junction intercellular channels. Mol Cell. 2004, 15: 879-888. 10.1016/j.molcel.2004.08.016.PubMedView ArticleGoogle Scholar
- Kovacs JA, Baker KA, Altenberg GA, Abagyan R, Yeager M: Molecular modeling and mutagenesis of gap junction channels. Prog Biophys Mol Biol. 2007, 94: 15-28. 10.1016/j.pbiomolbio.2007.03.013.PubMed CentralPubMedView ArticleGoogle Scholar
- Pantano S, Zonta F, Mammano F: A fully atomistic model of the Cx32 connexon. PLoS One. 2008, 3: e2614-10.1371/journal.pone.0002614.PubMed CentralPubMedView ArticleGoogle Scholar
- Kwon T, Harris AL, Rossi A, Bargiello TA: Molecular dynamics simulations of the Cx26 hemichannel: evaluation of structural models with Brownian dynamics. J Gen Physiol. 2011, 138: 475-493. 10.1085/jgp.201110679.PubMed CentralPubMedView ArticleGoogle Scholar
- Zonta F, Polles G, Zanotti G, Mammano F: Permeation pathway of homomeric connexin 26 and connexin 30 channels investigated by molecular dynamics. J Biomol Struct Dyn. 2012, 29: 985-998. 10.1080/073911012010525027.PubMed CentralPubMedView ArticleGoogle Scholar
- Scott CA, Kelsell DP: Key functions for gap junctions in skin and hearing. Biochem J. 2011, 438: 245-254. 10.1042/BJ20110278.PubMedView ArticleGoogle Scholar
- Hernandez VH, Bortolozzi M, Pertegato V, Beltramello M, Giarin M, Zaccolo M, Pantano S, Mammano F: Unitary permeability of gap junction channels to second messengers measured by FRET microscopy. Nat Methods. 2007, 4: 353-358.PubMedGoogle Scholar
- Bastianello S, Ciubotaru CD, Beltramello M, Mammano F: Dissecting key components of the Ca2+ homeostasis game by multi-functional fluorescence imaging. Three-Dimensional and Multidimensional Microscopy: Image Acquisition and Processing XI, Proc. SPIE 5324. 2004, 5: 265-274. 10.1117/12.553731.Google Scholar
- Chandler D: Statistical mechanics of isomerization dynamics in liquids and the transition state approximation. J Chem Phys. 1978, 68: 2959-2971. 10.1063/1.436049.View ArticleGoogle Scholar
- Roux B, Karplus M: Ion-transport in a gramicidin-like channel - dynamics and mobility. J Phys Chem. 1991, 95: 4856-4868. 10.1021/j100165a049.View ArticleGoogle Scholar
- Kirkwood JG: Statistical mechanics of fluid mixtures. J Chem Phys. 1935, 3: 300-314. 10.1063/1.1749657.View ArticleGoogle Scholar
- Torrie GM, Valleau JP: Nonphysical sampling distributions in Monte Carlo free-energy estimation: Umbrella sampling. J Comput Phys. 1977, 23: 187-199. 10.1016/0021-9991(77)90121-8.View ArticleGoogle Scholar
- Zwanzig R: Diffusion past an entropy barrier. J Phys Chem. 1992, 96: 3926-3930. 10.1021/j100189a004.View ArticleGoogle Scholar
- Locke D, Bian S, Li H, Harris AL: Post-translational modifications of connexin26 revealed by mass spectrometry. Biochem J. 2009, 424: 385-398. 10.1042/BJ20091140.PubMed CentralPubMedView ArticleGoogle Scholar
- Oshima A, Tani K, Hiroaki Y, Fujiyoshi Y, Sosinsky GE: Three-dimensional structure of a human connexin26 gap junction channel reveals a plug in the vestibule. Proc Natl Acad Sci USA. 2007, 104: 10034-10039. 10.1073/pnas.0703704104.PubMed CentralPubMedView ArticleGoogle Scholar
- Bukauskas FF, Jordan K, Bukauskiene A, Bennett MV, Lampe PD, Laird DW, Verselis VK: Clustering of connexin 43-enhanced green fluorescent protein gap junction channels and functional coupling in living cells. Proc Natl Acad Sci USA. 2000, 97: 2556-2561. 10.1073/pnas.050588497.PubMed CentralPubMedView ArticleGoogle Scholar
- Palacios-Prado N, Sonntag S, Skeberdis VA, Willecke K, Bukauskas FF: Gating, permselectivity and pH-dependent modulation of channels formed by connexin57, a major connexin of horizontal cells in the mouse retina. J Physiol. 2009, 587: 3251-3269. 10.1113/jphysiol.2009.171496.PubMed CentralPubMedView ArticleGoogle Scholar
- Palacios-Prado N, Briggs SW, Skeberdis VA, Pranevicius M, Bennett MV, Bukauskas FF: pH-dependent modulation of voltage gating in connexin45 homotypic and connexin45/connexin43 heterotypic gap junctions. Proc Natl Acad Sci USA. 2010, 107: 9897-9902. 10.1073/pnas.1004552107.PubMed CentralPubMedView ArticleGoogle Scholar
- Schuttelkopf AW, van Aalten DM: PRODRG: a tool for high-throughput crystallography of protein-ligand complexes. Acta Crystallogr D Biol Crystallogr. 2004, 60: 1355-1363. 10.1107/S0907444904011679.PubMedView ArticleGoogle Scholar
- Case DA, Cheatham TE, Darden T, Gohlke H, Luo R, Merz KM, Onufriev A, Simmerling C, Wang B, Woods RJ: The Amber biomolecular simulation programs. J Comput Chem. 2005, 26: 1668-1688. 10.1002/jcc.20290.PubMed CentralPubMedView ArticleGoogle Scholar
- Roux B: The calculation of the potential of mean force using computer-simulations. Comput Phys Commun. 1995, 91: 275-282. 10.1016/0010-4655(95)00053-I.View ArticleGoogle Scholar
- Souaille M, Roux B: Extension to the weighted histogram analysis method: combining umbrella sampling with free energy calculations. Comput Phys Commun. 2001, 135: 40-57. 10.1016/S0010-4655(00)00215-0.View ArticleGoogle Scholar
- Hess B, Kutzner C, van der Spoel D, Lindahl E: GROMACS 4: Algorithms for highly efficient, load-balanced, and scalable molecular simulation. Journal of Chemical Theory and Computation. 2008, 4: 435-447. 10.1021/ct700301q.PubMedView ArticleGoogle Scholar
- Berendsen HJC, Postma JPM, van Gunsteren WF, Dinola A, Haak JR: Molecular dynamics with coupling to an external bath. J Chem Phys. 1984, 81: 3684-3690. 10.1063/1.448118.View ArticleGoogle Scholar
- Darden T, York D, Pedersen L: Particle mesh Ewald: An N · log(N) method for Ewald sums in large systems. J Chem Phys. 1993, 98: 10089-10093. 10.1063/1.464397.View ArticleGoogle Scholar
- Eyring H: The activated complex and the absolute rate of chemical reactions. Chem Rev. 1935, 17: 65-77. 10.1021/cr60056a006.View ArticleGoogle Scholar
- Ermak DL, McCammon JA: Brownian dynamics with hydrodynamic interactions. J Chem Phys. 1978, 69: 1352-1361. 10.1063/1.436761.View ArticleGoogle Scholar
- Brown EB, Wu ES, Zipfel W, Webb WW: Measurement of molecular diffusion in solution by multiphoton fluorescence photobleaching recovery. Biophys J. 1999, 77: 2837-2849. 10.1016/S0006-3495(99)77115-8.PubMed CentralPubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.