生物谷报道:今天出版的Nature报道了在原核生物中氯通道的功能,氯通道在真核中是一个大的超家族,功能很多,但在原核生物中一直缺乏研究手段,对其功能并不清楚,Nature这篇报道认为它与原核生物Secondary active transport 有关.
Nature 427, 803 - 807 (26 February 2004); doi:10.1038/nature02314
Secondary active transport mediated by a prokaryotic homologue of ClC Cl- channels
ALESSIO ACCARDI AND CHRISTOPHER MILLER
Department of Biochemistry, Howard Hughes Medical Institute, Brandeis University, Waltham, Massachusetts 02454, USA
Correspondence and requests for materials should be addressed to C.M. (cmiller@brandeis.edu).
We have recently overcome technical barriers preventing electrophysiological recording of ClC-ec1 and have described conductance properties of this purified protein incorporated into planar lipid bilayer membranes1. As expected for a ClC channel, this homologue passes Cl--selective current. By ion channel standards, the single-molecule turnover is minuscule (104–105 ions s-1) but macroscopic current arising from hundreds to thousands of ClC-ec1 molecules may readily be observed. The current shows no voltage-dependent gating, but is instead maximally activated by low pH, in accord with its suggested biological function2 as a proton-activated electrical 'leak' for Cl-.
Cl- selectivity is imperfect, however, as seen from the 'reversal potential', that is, the voltage needed to null the current in a transmembrane KCl gradient. Currents recorded under salt-gradient conditions (300 mM KCl in the internal solution and 45 mM KCl in the external solution, symmetrical pH 3) in response to a family of voltage pulses are shown in Fig. 1a, along with current–voltage (I–V) curves derived from these recordings (Fig. 1c). The I–V curve is displaced along the voltage axis in the positive direction, indicating preferential permeation by Cl-, but the reversal potential, 30 mV, is substantially below the Nernst equilibrium potential expected for perfect Cl- selectivity, 45 mV (arrow). This fact means that some other ion in the system must carry current along with Cl-.
Figure 1 Proton dependence of ClC-ec1 currents. Full legend
High resolution image and legend (55k)
The missing ion
Which of the other ions present accounts for the sub-nernstian reversal potential? We set out to identify this competing ion in our minimal system, where the only ions present, besides Cl-, are K+, H+, OH- and the buffers glutamate and histidine. K+ and the buffers were previously ruled out by ion-substitution experiments1. Nor can OH- explain the discrepancy, as at the low pH of these experiments this anion is present at 10-11 M, far too low a concentration to permeate at rates on the order of turnover (105 s-1) given the constraints of diffusion limitation. We are thus left with H+ as the only possible candidate. If Cl- and H+ diffuse through a pore, each under the influence of its own electrochemical gradient, then the reversal potentials obtained above (6.7-fold Cl- gradient, symmetrical pH 3) require that H+ be 50–100-fold more permeant than Cl-. Because all previously studied ClC channels are specific for small anions, this conclusion is alarming. It is also immediately testable. If pH were to be raised symmetrically, the interfering permeation by H+ should be suppressed, and the reversal potential in a Cl- gradient should approach the ideal value. But we find that the I–V curve at pH 5 reverses at precisely the same value as at pH 3 across the entire range of Cl- concentrations (Figs 1b, c and 2a). A similar result was found at pH 7 (ref. 1). This is a paradoxical situation, as all ions in the system are ruled out as electrodiffusive interlopers, and yet the sub-nernstian reversal potentials demand that an ion besides Cl- carries current.
Figure 2 H+ and Cl- dependence of reversal potential. Full legend
High resolution image and legend (56k)
This other ion is in fact H+, as the experiments of Fig. 1d–g show. With symmetrical 300 mM Cl-, a transmembrane pH gradient (pH 3in/5ex) produces a 35–40-mV shift in the I–V curve in the direction of the H+ equilibrium potential (117 mV depending on the sidedness of the gradient). Under these symmetrical-salt conditions, the current observed at zero voltage cannot represent diffusive movement of Cl-. In terms of pore-mediated ion permeation, these results amplify the paradox above. H+ concentration, when varied symmetrically, has no effect on reversal potential, but when varied asymmetrically has a profound influence.
H+-Cl- countertransport
This conundrum vanishes if we abandon the assumption, natural to make for a Cl- channel homologue3, that Cl- permeates electrodiffusively through a water-filled pore. If instead we imagine that Cl- crosses the membrane in stoichiometric exchange for H+, the phenomena above are not only explained, but are required. We propose that ClC-ec1 mediates the Cl--H+ exchange reaction
where n and m are stoichiometric coefficients and subscripts refer to the two sides of the membrane, internal and external. With Cl- and H+ movements obligatorily coupled, the reversal potential is the voltage at which the reaction is at thermodynamic equilibrium, and may thus be derived without reference to any details of the exchange mechanism
where the stoichiometric ratio r = m/n, and the Nernst potential of each ion, EX, is
where zX is ionic valence, R is universal gas constant, T is temperature, F is Faraday's constant and where brackets denote ionic activity.
This secondary active transport mechanism makes several predictions that are inconsistent with pore electrodiffusion. First, it demands that the reversal potential must vary logarithmically with the concentration ratio of each ion separately, and that the slope of this variation must always be sub-nernstian. Figure 2 illustrates this behaviour with three data sets, two of them varying Cl- gradient at fixed symmetrical pH (3 or 5), the third varying pH gradient at fixed Cl- (300 mM); the data adhere well to equation (2), with a global fit determining the stoichiometric ratio, r, as the single adjustable parameter. For the fit shown in Fig. 2a, b, r = 0.43; with individual data sets we find this parameter to fall in the range 0.41–0.66, consistent with a stoichiometry of 2 Cl- ions per H+ (r = 0.5). The slopes of the reversal potential plots are 38 and 18 mV per decade for Cl- and H+, respectively, well below the ideal Nernst slope of 58 mV per decade, whereas the sum of these slopes, 56 mV, is within error identical to the Nernst value, as equation (2) requires. Figure 2a, b also illustrates the futility of forcing an electrodiffusion mechanism, in which Cl- and H+ permeate individually through a pore, to agree with the experimental results. Further analysis (not shown) confirms that this inadequacy is a general property of dissipative electrodiffusion models. Also in contrast to pore diffusion, equation (2) predicts that the reversal potential at a fixed Cl- gradient should remain invariant when pH is changed symmetrically, as observed above (Figs 1c and 2a).
Finally, the H+-Cl- exchange mechanism predicts that the reversal potential measured in the presence of both Cl- and pH gradients should be the simple sum of those measured with each individual gradient alone (that is, with a Cl- gradient and symmetrical pH, or vice versa). To test this prediction, we measured reversal potentials with Cl- and pH gradients simultaneously present (Fig. 2c). These values (filled circles) compare well with the sum of the experimental values (filled triangles) measured in the individual gradients of Fig. 2a, b. The figure also presents the reversal potentials theoretically expected for the countertransport mechanism with 2 Cl-/1 H+ stoichiometry (r = 0.5, equation (2), bold arrows) and for the electrodiffusion model (thin arrows) with H+/Cl- permeability ratio determined from the global fit (Fig. 2a, b). In each case, the reversal potentials in the combined gradients are much closer to the exchanger than to the channel predictions, which in some cases get even the sign wrong.
Direct observation of secondary active transport
Despite the solid ground afforded by the thermodynamic considerations above, the idea that ClC-ec1 is not an ion channel is startling enough to call for further experimental confirmation. If this protein operates as a Cl--H+ exchanger, then a pre-established gradient of either ion should drive active transport of its exchange partner. We therefore carried out classical secondary active transport experiments4-6 in which a Cl- gradient was established across liposome membranes and H+ transport was followed by pH changes in the external solution. In the experiment of Fig. 3a (top trace), ClC-ec1-reconstituted liposomes loaded with 350 mM Cl- at pH 4.5 were suspended in 3 mM Cl- at pH 4.8; under these conditions no transport can occur, because the exchanger, being highly electrogenic (three charges moved per turnover), polarizes the membrane. The reaction is initiated by adding the K+ ionophore valinomycin to allow counterion movement and to hold the liposome voltage at a slightly positive value (5–10 mV). Now, H+ enters the vesicles against both pH and electrical gradients as Cl- exits, reaching steady state in about a minute. Addition of the weak-acid proton uncoupler carbonyl cyanide p-(trifluoro-methoxy)phenyl hydrazone (FCCP) reverses the H+ uptake. It is difficult to quantify precisely the electrochemical gradient against which H+ moves, but an estimate based on the intraliposomal volume and buffer capacity suggests that a steady-state pH gradient of at least 1.5 units is sustained by the transporter. This conservative estimate neglects the additional electrical gradient against which both protons and Cl- move. It is worth emphasizing that an ion channel permeable to Cl- and H+ would produce H+ efflux under these conditions rather than the influx observed. Additional experiments (data not shown) establish that with an inwardly oriented Cl- gradient, H+ is extruded against a pH gradient in a FCCP-sensitive manner, and that no H+ uptake occurs if ClC-ec1 is omitted from the reconstitution mix.
Figure 3 Secondary active transport. Full legend
High resolution image and legend (73k)
An exchange mechanism requires reciprocity in transport behaviour; just as a Cl- gradient drives active H+ transport, so should a pH gradient drive uphill Cl- flux. An example of such a reciprocal experiment is shown in Fig. 3b. Here (top trace), an inwardly directed proton gradient is imposed across the liposome membrane, and uphill Cl- efflux, initiated by valinomycin, is followed by the de-quenching of a Cl--sensitive fluorometric dye trapped within the liposomes. This increase in fluorescence is reversed by collapsing the proton gradient with FCCP. A rough calculation based on the approximate doubling of fluorescence implies that the H+/Cl- exchange establishes at least a fivefold Cl- gradient. No such responses were observed in protein-free liposomes.
Abolition of proton coupling by E148A mutation
In ClC-ec1 a conserved glutamate residue, E148, lies close to a bound Cl- ion, occluding the extracellular side of the transport pathway7, 8. Mutating this glutamate to Ala or Gln maximally activates Cl- fluxes in reconstituted liposomes, completely abolishes their pH sensitivity, and renders the currents nearly ideally Cl--selective1. These results imply that E148 is specifically required for coupling of protons to Cl- permeation. Accordingly, we reconstituted ClC-ec1(E148A) into liposomes and found the protein inactive in both of the secondary active-transport assays of Fig. 3 (lower traces). These liposomes are still fully functional in Cl--Cl- exchange1 and net electrogenic downhill Cl- movement (data not shown); the mutation's effect is to eliminate H+ coupling to Cl- flux. This conclusion is further supported by the electrical behaviour of the E148A mutant in planar bilayers. In marked contrast with wild-type currents (Fig. 1), the I–V curve of the mutant protein (Fig. 4) is completely unresponsive to pH manipulation, with the reversal potential in symmetrical Cl- remaining at zero with a 4-unit pH gradient across the bilayer. As this mutation apparently retains the low unitary transport rate of wild-type protein1, we suggest that uncoupled Cl- translocation through E148A, although now dissipative, is nonetheless mediated by a cycle of conformational changes, not by pore diffusion. This suggestion must remain tentative, however, as macroscopic electrophysiological properties do not readily distinguish a simple carrier mechanism from a channel.
Figure 4 E148A mutation abolishes H+ dependence of reversal potential. Full legend
High resolution image and legend (43k)
Conclusions and implications
These experiments taken individually and in aggregate establish that this bacterial ClC homologue is not an ion channel as we had originally imagined3, but instead acts as a H+-Cl- exchange transporter with a likely stoichiometric ratio of 2 Cl-/H+. This conclusion forces a reinterpretation of the activation of Cl- currents by low pH1, 2 in terms of proton transport rather than proton-dependent channel gating. These results should not be taken to imply that eukaryotic ClC channels are also exchangers; indeed, ClC-0, ClC-1 and ClC-2 are unambiguously Cl--selective channels9-11, which, while displaying proton-dependent gating, show no indications of H+ permeability. The results do raise questions, however, as to the status of other homologues such as ClC-6 and ClC-7, which have thus far been assumed, without proof, to be channels.
Our conclusions are harmonious with the recent X-ray structures of wild-type and mutant ClC-ec1 determined by Dutzler and colleagues7, 8. In particular, none of these shows a transmembrane pore, which would be required for a channel but forbidden in any well-coupled exchanger. Figure 5 represents a cut-away view of a single subunit showing the central Cl- ion (green) and the occluded escape-ways leading to the internal and external solutions. E148 (red) forms a steric barrier separating the Cl- ion from the external solution, whereas S107 (yellow) and Y445 (blue) obstruct the internal pathway. Because E148 is strongly implicated in linking H+ and Cl- transport, these two ions appear to share, at least in part, a common translocation pathway.
Figure 5 Occlusion of the central Cl- ion in ClC-ec1. A cut-away view transecting the central Cl--coordination region of a single ClC-ec1 subunit is shown in surface representation. Full legend
High resolution image and legend (83k)
For ion channel proteins, a single high-resolution structure—that of the ion-occupied pore—may suffice for illuminating the mechanistic underpinnings of electrodiffusive ion permeation12. But this is not the case for transporters, as substrate permeation in these proteins arises from a cycle of conformational changes. We suppose that the multiple high-resolution structures of ClC-ec1 (refs 7, 8) include two of the conformations involved in substrate translocation. In the wild-type protein, the buried Cl- ion is reminiscent of occluded states of P-type ATPase ion pumps13, 14. In the E148Q mutant, which presumably models a protonated glutamate, this Cl- ion becomes exposed to the extracellular solution as the side chain rotates away from the anion to open an external 'gate'8. Is this a protonated outward-facing conformation of the exchange cycle? We cannot answer this question now, nor can we propose a detailed mechanism for substrate transport in terms of the structure; we suggest, however, that the conformational changes involved in transport are relatively minor, as the unitary turnover rate of 105 s-1, although low for a channel, is high for a transporter. Such a picture would differ from that of the MFS transporter family, for which the countertransporters OxlT15 and GlpT16 and the co-transporter LacY17 all show wide aqueous cavities that are thought to alternate in exposure to the two sides of the membrane via large polypeptide backbone rearrangements.
Finally, we are impressed by the success of the bacterial ClC structure in predicting and rationalizing electrophysiological behaviour of eukaryotic ClC channels8, 18-21. This is, in retrospect, unexpected, as ClC-0 and ClC-1 are ion channels but ClC-ec1 is not. It is notable that mutations of the conserved glutamate produce constitutive activity and eliminate the normal pH-dependence of both ClC-0 and ClC-ec1. This functional congruence on mechanistically dissimilar backgrounds suggests that structurally subtle alterations accumulated through ClC evolution might precipitate a profound mechanistic switch. Unusual as it is, this situation is not unprecedented. The widespread family of ABC transporters22, whose members operate as ATP-driven pumps—inward for nutrients or outward for xenobiotics—offers the striking example of CFTR, in which the protein, although maintaining molecular hallmarks of the family, functions not as a solute pump but rather as an ATP-gated Cl- channel23. Moreover, a P-type ATPase ion pump is known to degrade to an ion channel under the influence of a natural toxin that opens the transport pathway to both sides of the membrane simultaneously24; an analogous 'slippage' may also operate in several neurotransmitter-reuptake transporters25-27. Finally, our results bring to mind the puzzling finding that ClC-0 gating exhibits tight coupling to Cl- permeation, as though this ion channel carries mechanistic remnants of a transporter conformational cycle28-30. These observations collectively suggest that in structural terms, transporters and channels may be separated by an exceedingly fine line.
Methods
Biochemical procedures and solutions Methods for expression, purification, reconstitution into liposomes, and electrophysiological recording of ClC-ec1 have been described in detail1. Protein was reconstituted at 5–25 µg mg-1 lipid for liposomes used in flux assays, and at 50–75 µg mg-1 for liposomes fused into planar lipid bilayers. For bilayer recording, liposomes were formed in 450 mM KCl, 25 mM citric acid, 25 mM phosphoric acid, adjusted to pH 7.0 with KOH. The standard high-salt (HS) solution was 290 mM KCl, 10 mM HCl, 5 mM histidine base, 5 mM glutamic acid, adjusted to the desired pH with KOH; KCl was changed in solutions using lower Cl- concentrations. Typically, currents were in the range 0.05–2 nA at 100 mV holding voltage, except for the E148A mutant, for which the currents tend to run down with time.
Coupled fluxes of Cl- and H+ Uphill movement of Cl- driven by a pH gradient was followed by the Cl--sensitive fluorophore 6-methoxy-N-(3-sulphopropyl)-quinolinium (SPQ)31. SPQ (300 µM) was added to a suspension of liposomes (20 mg ml-1) formed in 20 mM KCl, 20 mM citrate-phosphate, pH 7.0 (RH buffer); the liposomes were frozen at -80 °C and sonicated briefly to give unilamellar vesicles trapping the fluorophore. External SPQ was removed by spinning a 100-µl aliquot through a 1.5-ml Sephadex G-50 column equilibrated in RH buffer, which was immediately added to 1.9 ml of RH buffer in a stirred fluorimeter cuvette. Fluorescence was measured with excitation and emission at 317 and 445 nm, respectively. A pH gradient was set up across the liposome membranes by acidifying the external solution to pH 4.0 with citric acid, and after a stable baseline reading had been confirmed, Cl- efflux was initiated by adding valinomycin (1 µg ml-1); after steady-state fluorescence was reached, the H+ gradient was collapsed with FCCP (0.25 µg ml-1).
Uphill movement of H+ driven by a Cl- gradient was followed by recording the pH of a lightly buffered vesicle suspension32. Reconstituted vesicles loaded with high Cl- medium (350 mM KCl, 50 mM citrate, 20 mM phosphate, pH 7.0) were acidified to pH 4.5 with phosphoric acid, and were frozen, thawed and sonicated as above. External Cl- was lowered by spinning 100 µl aliquots through Sephadex G-50 columns equilibrated 3 mM KCl, 300 mM K2SO4, 2 mM glutamic acid, pH 4.8. The resulting spin-through (100 µl) was then diluted into 1.9 ml containing this low-Cl- solution in a stirred cell monitored by a recording pH meter. Proton uptake (indicated by an increase in external pH) was initiated by valinomycin as above, and the experiment was terminated by collapsing the pH gradient with FCCP.
