Identification of Mechanosensitive Ion Channels in the Cytoplasmic Membrane of Corynebacterium glutamicum

Abstract

Patch-clamp experiments performed on membrane fragments of Corynebacterium glutamicum fused into giant liposomes revealed the presence of two different stretch-activated conductances, 600 to 700 pS and 1,200 to 1,400 pS in 0.1 M KCl, that exhibited the same characteristics in terms of kinetics, ion selectivity, and voltage dependence.

Mechanosensitive (MS) ion channels are channels activated or inactivated by mechanical forces (i.e., stretch exerted on cell membranes). They have been documented electrophysiologically in animals, plants, fungi, and bacteria (14, 16, 21) and recently in archaea (12). Stretch-activated channels have been detected by patch-clamp studies in gram-negative (Escherichia coli) and gram-positive bacteria (Streptococcus faecalis and Bacillus subtilis) (reviewed in reference 24). In E. coli, according to their conductance, gating kinetics, ion selectivity, and response to stretch, three different channel activities have been characterized: MscM, MscS, and MscL (5). The MscL, the channel with the highest conductance, has been purified and its gene has been cloned (23). MscL is present in a large number of bacteria (15).

Since their discovery in bacteria, it has been proposed that MS channels play a role in sensing and responding to changes in osmotic environments. Hypoosmotic stress leads to a massive influx of water into bacterial cells, followed by a rapid release of low-molecular-mass molecules, in particular osmolytes, from the cytoplasm. The hypoosmotic-induced efflux of osmolytes was observed in gram-negative (1, 10, 11, 22) and gram-positive bacteria (8, 20). The efflux pathways are not known, but the high-conductance bacterial MS channels are obvious candidates (4). The hypoosmotically induced efflux of compatible solutes in Corynebacterium glutamicum is well characterized (20). Study of the efflux process has suggested that it is mediated by channels which are tightly regulated on the level of activity. In this study we analyze the MS channels of C. glutamicum by the patch-clamp technique.

In contrast to previous studies performed on other gram-positive bacteria (25, 26), it was not possible to obtain protoplasts of C. glutamicum that were suitable for patch-clamp. Initial experiments in which crude membrane fragments were fused into giant liposomes also failed because no tight and stable seal could be generated on these structures. This was attributed to the pore-forming activity of the cell wall. Indeed, porin-like channels are present in C. glutamicum, presumably in a second membrane located in the cell wall (17). We therefore purified the plasma membrane by separation on a sucrose gradient according to Niederweis et al. (17).

Cells of C. glutamicum ATCC 13032 (wild type) were grown aerobically at 30°C in brain heart infusion medium (Difco). The cells were harvested, resuspended in membrane buffer (21.13 mM NaH₂PO₄, 28.76 mM Na₂HPO₄, 1 mM MgSO₄ [pH 7]) and passed three times through a French pressure cell at 20,000 lb/in². Unbroken cells and debris were removed by low-speed centrifugation, and the membranes were collected by centrifugation of the supernatant at 136,000 × g (rotor TLA 100.4, Beckman) for 30 min. The pellet was resuspended in 2 ml of membrane buffer and applied to a sucrose-step gradient of 30 (3 ml), 40 (4 ml), and 70% (3 ml) according to Niederweis et al. (17). The gradient was centrifuged at 110,000 × g for 16 h (rotor SW41Ti, Beckman). With this treatment, four fractions were obtained, which were analyzed for protein content (13) and assayed for NADH oxidase activity (18) as a marker of the plasma membrane. The second fraction, sedimenting with the 40% sucrose layer, possessed a higher activity and was kept for further use. Membrane fragments of this fraction were fused with liposomes of azolectin (from soybean, type IV-S; Sigma) at a lipid-to-protein ratio of 20, by a cycle of dehydration-rehydration to produce giant proteoliposomes amenable to patch-clamp recording as previously described (3).

High-resistance seals could be formed on these proteoliposomes. We found channel activity in every stable patch when suction, i.e., membrane stretch, was applied to the pipette interior with a syringe. Without suction no channel activity could be observed. Most frequently, two different unit conductances could be observed under symmetric conditions (100 mM KCl buffer), a smaller one between 600 and 700 pS (n = 15) and a larger one between 1,200 and 1,400 pS (n = 17).

Figure 1A gives an example of a recording of a patch containing smaller conductance channels. As soon as suction was applied, up to 12 channels opened one after another and closed at the same time when pressure was released. The channels had very slow kinetics; no fast flickering was observed. Under constant pressure we observed openings that lasted up to 1 min without flickering or closures in between (data not shown).

FIG. 1.

Patch-clamp recordings performed on giant proteoliposomes reconstituted with C. glutamicum plasma membrane fragments. Unitary currents were recorded in the excised mode (upper traces) with a Biologic RK-300 patch-clamp amplifier with a 10 GΩ feedback resistance. Records were filtered at 300 Hz through a four-pole Bessel low-pass filter and digitized at 833 Hz. Suction in the pipette (lower traces) was applied by syringe and monitored with a piezo-electric pressure transducer. Bath medium, 10 mM HEPES-KOH (pH 7.4)–100 mM KCl; pipette medium, similar to bath medium with, in addition, 2 mM CaCl₂–5 mM MgCl₂. Holding potential was −10 mV for all shown recordings (potential of bath solution, assigning zero potential level to the pipette). (A) Activation of the 600- to 700-pS conductance channels. Application of suction in the pipette induced the opening of up to 12 identical channels, which closed upon release of suction. (B) Activation of the 1,200- to 1,400-pS conductance channel. Application of suction induced the activation of three similar channels. (C) Substate of the 1,200- to 1,400-pS conductance channel. The same patch as in panel B is shown. Application of suction induced the opening of two similar channels. The incomplete closure of one channel (asterisk at left) was later followed by the closure (asterisk at right) of the corresponding substate (350 pS).

A recording of the activity of the larger conductance channels is given in Fig. 1B. With increasing suction, three identical channels opened and then closed when the pressure was released. All large channels that were observed displayed slow kinetics. In terms of their kinetics they were not distinguishable from the smaller conductance channels. Transitions lower than 600 pS were sometimes observed but they were always associated with the 1,200- to 1,400-pS conductance channel and they displayed all the characteristics of substates (Fig. 1C). With the smaller channel, no substate was observed.

In most of the cases (28 of 32), only one type of channel, e.g., the larger or the smaller conductance channel, was observed in a single patch. The larger conductance channel was observed together with the smaller channel twice. Most patches contained several similar channels, suggesting that they form clusters which segregate on dilution involved in reconstitution, as previously observed with E. coli (5). In two other cases, the larger channel was observed together with even higher stretch-activated conductances (2.6 nS and higher) which displayed the same kind of slow kinetics.

Activation of both types of channel occurred within a very narrow range of applied pressure as an all or none process: in most patches, the channels were either closed or permanently open. The minimum suction needed to activate both types of channels varied between 30 and 100 mm of Hg. The threshold of activation was variable not only between different patches but also when suction was repeatedly applied to one patch. Taking all this into account, it was not possible to distinguish between the two types of the channels on the basis of their pressure sensitivities.

The conductance and the selectivity of both types of MS channel in C. glutamicum were obtained by recording at different membrane potentials under symmetric and under asymmetric conditions. For both channels, the current-voltage relationship was linear (Fig. 2). Measurements under asymmetric conditions indicated for both types of channel a weak cationic selectivity. From the Goldman-Hodgkin-Katz equation, P_K⁺/ P_CI⁻ for the large and small conductance channels were calculated to be 1.75 and 1.8, respectively. In the range of potentials used for the current-voltage curves, none of the channels displayed clear voltage dependence.

FIG. 2.

Current-voltage plot of the MS channels. The sizes of the current steps are plotted against the imposed membrane voltage. Recordings were obtained from excised patches of giant liposomes reconstituted with membrane fragments from C. glutamicum either under symmetrical conditions (○; 0.1 M KCl_pip/0.1 M KCl_bath) or asymmetrical conditions (□; 0.1 M KCl_pip/0.3 M KCl_bath) (KCl_pip, KCl concentration in pipette medium; KCl_bath, KCl concentration in bath medium). (A) Current-voltage relationship of the smaller channel. Data are means of two patches. The slope of the curve under symmetrical conditions is 720 pS. Changing the KCl concentration in the bath to 300 mM shifted the reversal potential to −7 mV. (B) Current-voltage relationship of the larger channel. Data are means of two patches. Under symmetrical conditions the slope of the curve is 1,400 pS. Asymmetrical conditions shifted the reversal potential to −7.4 mV.

Gadolinium is a known inhibitor of MS channels (9) and in particular of bacterial MS channels (4). We found that both types of C. glutamicum channels could be reversibly inhibited by 100 μM Gd³⁺.

The MS channels of C. glutamicum show the characteristics of all the bacterial stretch-activated channels detected so far: a very high conductance, a weak selectivity, and the ability to be reconstituted in a lipid bilayer. In contrast with previous patch-clamp studies performed in bacteria, only two types of channel activities were detected in C. glutamicum and they showed very similar properties in terms of kinetics and selectivity. In fact they could only be distinguished by their conductance, which differed by a factor of 2. In gram-positive bacteria (B. subtilis and S. faecalis), the existence of a complex electrical activity was detected, with a multiplicity of stretch-activated conductances, ranging from 100 pS to 3 nS, with different kinetics (25, 26). Similarly, in E. coli three subfamilies of channels could be distinguished which differed by their conductance, selectivity, kinetics, and sensitivity to applied pressure (5).

Our observations can be explained in different ways. One possibility is that there is only one type of channel in C. glutamicum and that the larger conductance channel corresponds to two smaller channels opening cooperatively. Another possibility is that the two channels corresponds to different oligomers of the same subunit protein. The 2.6-nS stretch-activated conductance channel, which is rarely observed, might correspond to an unstable oligomer of higher degree. Finally, it cannot be excluded that these conductances correspond to different channels with different subunits. In E. coli, deletion of the mscL gene results in the suppression of the MscL channel but not of the other channels. This indicates that this bacterium possesses at least two genes coding for MS channel subunits and perhaps more. The genome of C. glutamicum is only 3.1 Mbp (2), compared to 4.6 Mbp for the E. coli genome, and C. glutamicum is known to possess a lower number of iso-enzymes. It is thus not surprising that C. glutamicum may be equipped with a minimum of MS channels.

Is one of the channels described here related to MscL? The conductance of the observed larger MS channel in C. glutamicum lies within the range of that of the MscL. But its cationic selectivity and slow opening kinetics make it more comparable to the smallest MS channel of E. coli, MscM (5). We tried to isolate the gene for the MS channel in C. glutamicum by PCR by using degenerate primers derived from highly conserved regions of the mscL gene (15), including the closely related Mycobacterium tuberculosis. We were not able, however, to detect a fragment that shows any sequence similarity to the mscL.

Regarding their physiological roles, a direct correlation between MS channel activity and hypoosmotic stress response in bacteria is still missing. So far all evidence is indirect (1, 4, 6, 7, 19, 25, 26). The molecular identification of the different bacterial channels should help clarify this issue.