Voltage-dependent porin-like ion channels in the archaeon Haloferax volcanii.

Membrane vesicles isolated from the cell envelope of the archaebacterium Haloferax volcanii were either reconstituted in giant liposomes and examined by the patch-clamp technique or were fused into planar lipid bilayers. In addition, cell envelope proteins were solubilized by detergent and reconstituted in azolectin liposomes, which were then fused into planar lipid bilayers. Independently of the technique used the predominant channel activity encountered exhibited the following characteristics. Channels were open at all voltages in the range approximately −120 to +120 mV and exhibited frequent fast transitions to closed levels of different amplitudes. At voltages greater than 120 mV the channels tended to close in a manner characterized by large, slow transitions of variable amplitudes. The tendency to close at high membrane potentials was much stronger at one polarity. The channel gating pattern was complex exhibiting a range of subconductances of 10-300 picosiemens in symmetric 100 mM KCl. These electrophysiological characteristics are comparable with those of bacterial and mitochondrial porins, suggesting that the archaeal channels may belong to the general class of porin channels. Some channels showed preference for K+, whereas the others preferred Cl−, suggesting the existence of at least two types of porin-like channels in H. volcanii.

Membrane vesicles isolated from the cell envelope of the archaebacterium Haloferax volcanii were either reconstituted in giant liposomes and examined by the patch-clamp technique or were fused into planar lipid bilayers. In addition, cell envelope proteins were solubilized by detergent and reconstituted in azolectin liposomes, which were then fused into planar lipid bilayers. Independently of the technique used the predominant channel activity encountered exhibited the following characteristics. Channels were open at all voltages in the range approximately ؊120 to ؉120 mV and exhibited frequent fast transitions to closed levels of different amplitudes. At voltages greater than 120 mV the channels tended to close in a manner characterized by large, slow transitions of variable amplitudes. The tendency to close at high membrane potentials was much stronger at one polarity. The channel gating pattern was complex exhibiting a range of subconductances of 10 -300 picosiemens in symmetric 100 mM KCl. These electrophysiological characteristics are comparable with those of bacterial and mitochondrial porins, suggesting that the archaeal channels may belong to the general class of porin channels. Some channels showed preference for K ؉ , whereas the others preferred Cl ؊ , suggesting the existence of at least two types of porin-like channels in H. volcanii.
The phylogenetic tree is composed of three domains: Eucarya (eukaryotes), Bacteria (eubacteria), and Archaea (archaebacteria) (1). Archaebacteria comprises several different families of cells adapted to extreme environmental conditions (e.g. temperature, pH, salt concentration). Although eubacteria and archaebacteria are both prokaryotes, it is now generally accepted that archaebacteria are not closer, phylogenetically, either to eubacteria or to eukaryotes (2). Therefore, archaebacteria appear to constitute an intermediary domain between eubacteria and eukaryotes.
Ion channels have been mostly studied in eukaryotic cells (3) and have also been documented in eubacteria. In particular bacterial porins and mechanosensitive ion channels are well described. Porins, large water-filled pores across the outer membrane of Gram-negative bacteria (reviewed in Ref. 4), have also been found in the cell wall of certain Gram-positive bacteria such as mycobacteria (5)(6)(7). More recently, different mechanosensitive ion channels have been described in Gram-negative and Gram-positive bacteria (8 -17). These channels, most likely localized in the plasma membrane, are implicated in osmoregulation (13).
In contrast, the presence of ion channels in archaebacteria has not been documented. A search for ion channels in Archaea thus seems to be of interest not only for the understanding of the phylogeny of ion channels, but also for a better knowledge of the physiology of these unique organisms. For our studies we chose Haloferax volcanii (formerly Halobacterium volcanii), a moderate halophilic archaeon amenable to genetic studies.
We report here the existence of ion channels in the cell envelope of H. volcanii, that with regard to electrophysiological characteristics resemble bacterial porins. This finding indicates the importance of the porin family in the phylogeny of ion channels. Furthermore, it questions our current knowledge of the organization of the cell envelope of archaebacteria.
Isolation of the Cell Envelope-One liter of cells was grown up to an A 600 of 1 and harvested by centrifugation and the pellet resuspended in 10 ml of H 2 0. Addition of EDTA (50 mM final concentration) resulted in lysis of most of the cells. To ensure total lysis, the suspension, supplemented with DNase (20 g/ml) and MgCl 2 (5 mM), was passed through a French press (8000 p.s.i.). The resulting suspension was centrifuged at 10,000 ϫ g to eliminate cell debris. The supernatant was centrifuged for 1 h at 45,000 rpm, using a Ti45 Beckman rotor. The pellet was then resuspended in 10 mM Hepes-NaOH, pH 7.4, plus 5% ethylene glycol. Protein content was measured by the method of Lowry et al. (18). 50-l aliquots (at 2-5 mg of protein/ml) were stored at Ϫ80°C for further use.
Preparation of Giant Proteoliposomes-Cell envelopes were mixed with azolectin (from soybean, type II-S, Sigma) liposomes at a lipid to protein ratio (w/w) of 50 and fused into giant proteoliposomes by dehydration-rehydration as described previously (11). Rehydration was performed in 100 mM KCl, 10 mM Hepes-KOH, pH 7.4.
Solubilization and Reconstitution of Membrane Proteins into Proteoliposomes-Membrane proteins from cell envelopes were solubilized using octyl ␤-D-glucoside; 100 l of the cell envelope suspension was mixed with 900 l of the solubilization buffer (10 mM Hepes-KOH, pH 7.4, 300 mM KCl, 1 mM dithiothreitol, 100 mM octyl ␤-D-glucoside) and incubated for 20 min. The suspension was then centrifuged at 90,000 rpm for 40 min using a TL100 Beckman ultracentrifuge. The supernatant was added to a 1-ml suspension of asolectin liposomes (1 mg of lipids/ml) in 10 mM Hepes-KOH, pH 7.4, 300 mM KCl and incubated for 20 min before the addition of Bio-Beads SM-2 (Bio-Rad) (160 mg/ml). The suspension was gently agitated for 5 h, the Bio-Beads were discarded, and the suspension was centrifuged for 30 min at 90,000 rpm, using a TL100 Beckman ultracentrifuge. The pellet was then resuspended in 200 l of 300 mM KCl, 10 mM Hepes-KOH, pH 7.4.
Electrophysiological Recording-Giant proteoliposomes were examined for channel activity using the standard patch-clamp method (19). After gigaohm-seal formation on a giant liposome, the patch was excised, and unitary currents were recorded using a Biologic (Claix, France) RK-300 patch-clamp amplifier with a 10-gigaohm feedback resistance. The membrane potential refers to the potential in the bath minus the potential in the pipette.
Planar lipid bilayers were formed across a 250-m diameter hole by presenting a bubble of asolectin lipids dissolved in n-decane (30 mg/ml) in front of the hole. After membrane formation, the mechanical and electrical stability of the membrane was monitored for 10 min before addition of crude cell envelopes or proteoliposomes to the cis compartment (8 g protein/ml final concentration in both cases). Fusion to the planar bilayer was induced by imposing a salt gradient between the two chambers (500 mM KCl in the cis compartment versus 100 mM KCl in the trans compartment). The trans compartment was held at virtual ground potential and the membrane current amplified using a homemade device. All solutions were filtered using 0.22-m Millipore filter.
Electrical recordings from patch-clamp or planar bilayer experiments were stored on a digital audio tape (Biologic DTR 1200 DAT recorder). Records were subsequently filtered at 1 kHz (Ϫ3 db point) through a 4-pole Bessel low-pass filter and digitized off-line at 2 kHz on a personal computer. Data were plotted by a Hewlett-Packard Laserjet printer.

RESULTS
Several attempts were made to record channel activities from native membranes of giant protoplasts of H. volcanii. Despite a concentrated effort to make the surface of the giant protoplasts amenable to patch-clamp recording, we were unable to form gigaohm-seal on their surface. Therefore, we decided to record ion channel activities by either reconstituting H. volcanii cell envelope in artificial liposomes or into planar bilayers.
The archaeal membrane vesicles were fused into giant liposomes by dehydration and rehydration (10,11), and the giant liposomes were then examined by the patch-clamp technique.
In parallel experiments the membrane vesicles were fused into planar lipid bilayers. In addition, cell envelope membrane proteins were solubilized using octyl ␤-D-glucoside, reconstituted in liposomes that were then fused into planar lipid bilayers. Independently of the technique used the predominant channel activity encountered exhibited the characteristics described below.
Channels were mostly open at all voltages in the range Ϫ120 mV to ϩ120 mV and exhibited frequent fast transitions to closed levels of different amplitudes (data not shown). At voltages greater than 120 mV the channels tended to close in a manner characterized by large, slow (seconds) transitions ( Figs. 1 and 2). The greater the magnitude of the membrane potential, the shorter the open-time duration following the voltage step. Sporadic fast closures were often superimposed on the slow kinetic closures. For some patches the channels also entered a sustained fast gating mode (millisecond) as shown in Fig. 1.
The tendency to close at high membrane potentials was greater at one polarity (positive potential for bilayer experiments and negative potential for patch-clamp experiments), indicating both a preferential insertion of the channels in the lipid bilayer and an asymmetry in the voltage dependence.
Figs. 1 and 2 show recordings from the same bilayer at positive and negative voltages. Upon application of ϩ180 mV across the bilayer the channels reached a closed inactivated state ( Fig. 1). At negative potential, full closure of the channels could not be reached, and the time course of closure was much slower (Fig. 2). In most cases the inactivated state was reversible and the channels reopened by returning to 0 mV for several seconds (Fig. 1). However, in a few instances, the channels appeared to be locked in the inactivated state, and reopening could not be obtained at any membrane potential.
In general the channel gating was complex, showing a range of conductances in between 10 and 300 picosiemens in symmetric 100 mM KCl. While this could be attributed in part to a mixture of different channels (see below), in some cases clear cooperative events were observed, indicating that the channels can gate at different conductance states. This is illustrated in Fig. 3, which shows recordings obtained by two successive steps to Ϫ160 mV on the same patch. The most frequent transition had a conductance of 80 picosiemens, but a clear cooperative 160-picosiemens transition was also observed. Furthermore, channel gating at a subconductance level of approximately 10 picosiemens was observed.
The selectivity of the channels was examined in bilayer experiments performed in asymmetric (500/100 mM KCl) and symmetric solution (500 mM KCl). I-V curves for unitary currents were difficult to obtain because of the different conductances encountered at each membrane potential. Instead, we chose to plot the total initial current through a given bilayer obtained during steps to various values of the membrane potential. Between steps the membrane potential was returned to 0 mV to ensure proper reopening of the channels. The current thus corresponded to that flowing through all the fully open channels present in a bilayer. In a series of nine different bilayer experiments, the following reversal potentials (in millivolts) were obtained: ϩ24, ϩ20, ϩ20, ϩ12, Ϫ4, Ϫ8, Ϫ14, Ϫ18, Ϫ23. The two extreme cases are shown in Fig. 4. These results suggest that at least two different types of porin-like channels with opposite selectivity are present in H. volcanii envelopes. DISCUSSION We report here for the first time the existence of ion channels found in the cell envelope of an archaebacterium. The electro-physiological characteristics of these channels, documented by two different techniques, can be summarized as follows: 1) the channels that have large conductances are maximally open at or around 0 mV and close at positive and negative voltages, 2) voltage dependence is asymmetric, 3) the channels exhibit both fast and slow kinetics, and 4) the channels exhibit several subconducting states. The same electrophysiological characteristics can be found in eubacterial porins as documented for OmpF and OmpC (20) and PhoE from Escherichia coli 1 and also in mitochondrial porins (reviewed in Ref. 21). This suggests that all these channels may belong to the same family of functionally and/or structurally related membrane pores. The observation that these types of channels are found in all domains of the tree of life indicates the importance of the porin family in the phylogeny of ion channels.
What is the localization of these channels? Due to their characteristics, their presence in the cytoplasmic membrane is unlikely. Although the channels would normally be closed at the high negative membrane potential characteristic of microbial cells, any depolarization would result in the sustained opening of these high conductance channels, probably leading to cell death. Therefore it is likely that these porin-like channels are located in the cell envelope outside the plasma membrane, as in eubacteria.
In eubacteria, porins are located in the outer membrane of Gram-negative bacteria where they allow the diffusion of small hydrophylic compounds through this protection barrier (4). Gram-positive bacteria were long supposed to lack porin channels. However, porin-like channels have been recently identified in the cell wall of Gram-positive bacteria such as Mycobacterium chelonae (5,6), Mycobacterium smegmatis (7), and Corynebacterium glutamicum (22). These bacteria are known to contain lipids in their cell wall in the form of mycolic acids, and it is possible that mycolic acids and other lipids form part of another bilayer (23), explaining the low permeability of the mycobacterial cell wall (24). Diffusion through this bilayer would be facilitated by porins.
Only one membrane, the plasma membrane, is documented in archaea such as Halobacterium halobium or H. volcanii. The main component of the cell wall is the S-layer formed by a hexagonal arrangement of a glycoprotein (25). A detailed structural study of the S-layer of H. volcanii (26) together with a a previous x-ray study of the envelope of H. halobium (27) led to a model in which the upper external part of the glycoprotein forms a dome-shaped region separated from the cell membrane by spacer elements (26). The protein is thought to be anchored in the plasma membrane by a hydrophobic stretch that is observed near the C terminus of both the H. halobium and H. volcanii glycoprotein sequence (28,29). The interspace between the plasma membrane and the external part of the S-layer has been considered as an aqueous space analogous to the periplasm of Gram-negative bacteria (26,27). However, if porins are present in the cell wall, but do not belong to the plasma membrane, they should be located in this interspace, which should then contain a lipid membrane. This membrane should represent in itself a permeability barrier through which movement of solutes would be facilitated by porins. Therefore, the discovery of porins in H. volcanii would suggest a more complex organization for the cell envelope of archaea than previously believed, possibly more similar to that of some Gram-positive eubacteria such as Mycobacteria (30).

FIG. 4. Selectivity of H. volcanii channels.
Membrane vesicles were fused into planar lipid bilayers, and recordings were performed in 500/100 mM KCl, 10 mM Hepes-KOH, pH 7.4, asymmetrical media (E) (cis versus trans) and 500 mM KCl symmetrical solution (q) after addition of concentrated KCl in the trans chamber. The total initial current through the bilayer obtained during steps to various potentials is plotted against the corresponding membrane potential. Between steps the membrane potential was returned to 0 mV to ensure proper reopening of the channels. Plots A and B are from two different bilayers. The reversal potentials were Ϫ23 and ϩ24 mV, respectively.