Release of Thioredoxin via the Mechanosensitive Channel MscL during Osmotic Downshock of Escherichia coli Cells*

Escherichia coli cells possess several mechanosensitive ion channels but only MscL, the channel with the highest conductance, which is activated at the highest membrane tension, has been cloned. We investigated the putative involvement of MscL in the effluxes caused by osmotic downshock. Osmotic shock caused the release of potassium glutamate, trehalose, and glycine betaine from wild type cells and cells lacking MscL. There was no difference between the two strains, but the extreme rapidity of the efflux process, as shown herein for glycine betaine, suggests that it is channel-mediated. Osmotic downshock also induces the release of some cytosolic proteins from EDTA-treated cells. We investigated the release of thioredoxin. This protein was totally released from wild type cells but was retained by MscL− cells. Release was restored by expression of the gene coding for MscL. Thus MscL is not necessary for the excretion of osmoprotectants, but it does open in vivoduring shock and catalyzes the efflux of thioredoxin and possibly other small cytosolic proteins. It follows that the other mechanosensitive channels, which are known to be activated at lower tension, must also open during osmotic shock. Their opening and that of MscL could account for the rapid release of osmolytes.

Mechanosensitive (MS) 1 ion channels are channels gated by mechanical forces exerted on cell membranes. They are present in a large variety of cells, in animals, plants, and bacteria. In bacteria, these channels have been documented by patch-clamp experiments in Escherichia coli, Streptococcus faecalis, and Bacillus subtilis (reviewed in Sukharev et al. (1)). A characteristic of bacterial MS channels (which distinguishes them from their counterparts in animal cells) is their multiplicity and high conductance. High conductance MS channels are also present in Archaea (2). In E. coli, on the basis of conductance and kinetics, three families of MS channels were distinguished: MscM (M for mini), MscS (S for small), and MscL (L for large). These channels are activated at different thresholds of applied pressure; the higher the conductance of a channel, the higher the pressure needed to trigger its opening (3).
The MscL channel has been purified and its gene has been cloned (4). Expression of the mscL gene is necessary and sufficient for activity of the MS channel with the highest conductance (4 -6). The mscL gene codes for a 15-kDa protein with two transmembrane domains (6). Two-dimensional crystals of the channel have been obtained, indicating that it is a homohexamer (7). The protein is located in the plasma membrane (6,8). Homologues of mscL have been identified in Gram-negative and Gram-positive bacteria (9). The genes coding for the other MS channels have not been identified. Complete genome sequencing of organisms such as E. coli and B. subtlis shows that these genes probably have sequences different from that of mscL, for which no paralogue has been found.
The function of the MS channels in bacteria is still a matter of speculation. It has been known for a long time that hypoosmotic shock causes the release of solutes from E. coli, without lysis of the cells (10). The localization of high conductance MS channels in the plasma membrane of Gram-positive and Gramnegative bacteria (11)(12)(13) led to the suggestion that shockinduced effluxes are mediated by these channels (14). More specifically, it was proposed that the physiological role of these channels is to catalyze the efflux of osmolytes or osmoprotectants upon osmotic shock (14,15). Cultivated in high osmolarity media, bacteria such as E. coli are able to synthesize or accumulate high concentrations of osmoprotectants and potassium which counterbalance the outside osmolarity (reviewed in Czonka and Epstein (16)). Upon a shift to a low osmolarity medium, excretion of compatible solutes is required to restore normal turgor. The efflux of potassium and osmoprotectants, triggered by osmotic downshock, has indeed been reported in E. coli and Salmonella typhimurium (17)(18)(19)(20), Lactobacillus plantarum (21), Corynebacterium glutamicum (22), and Listeria monocytogenes (23).
In addition to species of low molecular weight (ions, metabolites, and osmoprotectants), some cytoplasmic proteins including elongation factor Tu (24), thioredoxin (25), and DnaK (26) are also excreted from E. coli upon osmotic downshock (27). The release of thioredoxin and DnaK occurs under conditions in which other cytoplasmic proteins are totally retained (25,26); how these cytoplasmic proteins can be excreted in the absence of cell lysis constitutes a riddle.
To examine the possible involvement of MscL in these processes, we studied the efflux of various molecules triggered by osmotic downshock in the wild type and mscL Ϫ strains of E. coli.

EXPERIMENTAL PROCEDURES
Materials-E. coli thioredoxin reductase and thioredoxin were supplied by IMCO (Stockholm, Sweden). The thioredoxin antibody was obtained from Sigma. [methyl-14 C]Choline chloride was obtained from Amersham (Buckinghamshire, United Kingdom).
Bacterial Strains and Growth Conditions-All experiments were performed using the parental E. coli AW 405 strain (28) and various derivatives, E. coli AW405 (referred to here as the wild type strain), AW405 carrying a chromosomal insertion in the mscL gene (referred to here as the MscL Ϫ strain), and a MscL Ϫ RecA Ϫ strain carrying the mscL expression plasmid p5-2-2 (referred to here as the restored strain) (4). These strains were kindly donated by Dr. P. Blount. Cells were grown aerobically at 37°C in M9 minimal medium containing 20 g/ml thiamine, 40 g/ml L-threonine, 40 g/ml L-leucine, 40 g/ml L-histidine, 10 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. g/ml streptomycin, and 22 mM glucose as the carbon source with, in addition, 50 g/ml ampicillin for the restored strain. Expression of mscL in this strain was induced with 2 mM isopropyl-1-thio-␤-D-galactopyranoside. In some experiments (trehalose assays, detection of MalE) glucose was replaced with 44 mM glycerol or 20 mM maltose. Culture at high osmolarity was performed by adding 500 mM NaCl (final concentration) to the culture.
Shock-induced Release of Potassium-K ϩ efflux was studied with a valinomycin K ϩ -selective electrode, connected to a microcomputer, which was used to monitor the appearance of K ϩ in the external medium, as described previously (29). Cells were harvested by centrifugation and suspended at A 650 ϭ 50 (25 mg dry weight/ml) in 10 mM Hepes-NaOH buffer, pH 7.5, 200 mM NaCl, 1 mM KCl, and 0.4% glucose. The resulting suspension was diluted 25-fold in the shock medium (Hepes-NaOH buffer, pH 7.5, NaCl ranging from 0 to 80 mM).
Shock-induced Release of Endogenous Glutamate-Growing cells (A 650 between 0.1 and 0.2) were subjected to an osmotic upshock by addition of NaCl (500 mM final concentration). Osmotic downshock was achieved by a 5-fold dilution with the same medium but devoid of NaCl. At various time intervals, 10-ml samples of the cell suspension were withdrawn and centrifuged. The pellet was suspended in 1 ml of M9 medium, and the cells were disrupted with a vibro-disrupter (MM2000 type, Retsch, Germany). Glutamate content was determined enzymatically, as described by Beutler (30), using a Boehringer kit assay. Cell growth was monitored in parallel by absorbance measurement.
Shock-induced Release of Endogenous Trehalose-Growing cells were subjected to an osmotic upshock followed by an osmotic downshock as described above. At various time intervals, 8-ml samples of the cell suspension were withdrawn and centrifuged. The pellets were treated with ice-cold 10% trichloroacetic acid, and the trehalose content was measured using the phenol method, as described previously (31). Cell growth was monitored in parallel by absorbance measurement.
Shock-induced Release of Radiolabeled Glycine Betaine-[N-methyl- 14 C]Glycine betaine was synthesized enzymatically by oxidation of [N-methyl-14 C]choline (specific activity, 55 Ci/mol) and was purified as described previously (32). Cells grown at high osmolarity were harvested by centrifugation and suspended at A 650 ϭ 0.5 in 10 mM Hepes-KOH buffer, pH 7.2, 600 mM NaCl, containing 0.4% (w/v) glucose. Transport was initiated by the addition of [N-methyl-14 C]glycine betaine (1 mM final concentration). At given time intervals, 100-l samples were withdrawn, filtered through Millipore HA filters, and washed with 5 ml of buffer. The radioactivity of the filters was counted. Osmotic downshock was performed as described in the figure legends.
To increase the time resolution of the filtration experiments, a rapid filtration assay was used. An aliquot (100 l) of the suspension of cells that had accumulated betaine at a high NaCl concentration was deposited onto the filter, and vacuum was applied. Under these conditions the bacteria were adsorbed to the filter and retained their glycine betaine for at least 15 s, as shown by control experiments in which the filters were washed with 5 ml of iso-osmotic solution, after adsorption of the cells. Various volumes (from 5 ml to 800 l) of shock buffer were rapidly applied to the filters onto which the bacteria were adsorbed. The filters were then washed with 5 ml of buffer containing 600 mM NaCl. The permeation rate of the shock buffer through the filter, when vacuum was applied, was determined independently and found to be 4 ml/s. Thus, the cells loaded with glycine betaine and adsorbed onto the filters were in contact with the shock solution for times varying from 1.25 s to 200 ms. A volume of 800 l of the shock buffer was the smallest that wetted all the bacteria deposited on the filter. Also, for smaller volumes, the time required to deliver the solution manually became limiting.
Shock-induced Release of Thioredoxin-Cells were suspended at A 650 ϭ 5 in a plasmolysis buffer containing 10 mM Tris-HCl, pH 7.6, 20% (w/v) sucrose, in the absence or presence of 2.5 mM EDTA, and incubated for 10 min at room temperature. The iso-osmotic dilution or the hypo-osmotic shock were performed by diluting 200 l of the cell suspension 5-fold with suspension buffer or distilled water, respectively. The cells were then separated from the medium by centrifugation in a bench centrifuge. The pellet was suspended in 100 l of electrophoresis buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 0.2% bromphenol blue, 10% glycerol, 1% mercaptoethanol), and boiled for 4 min. A 4-l aliquot was then subjected to SDS-polyacrylamide gel electrophoresis in a 15% acrylamide gel. A 200-l aliquot of the supernatant was incubated overnight with 20% trichloroacetic acid and centrifuged in a bench centrifuge. The precipitated proteins were suspended in 20 l of electrophoresis buffer and boiled for 4 min. A 4-l aliquot was subjected to electrophoresis. For determination of total thioredoxin cell content, 1 ml of cells at A 650 ϭ 1 were centrifuged, and the pellet was suspended in 100 l of electrophoresis buffer. A 4-l sample was loaded onto the gel and subjected to electrophoresis. Western blotting was performed as described elsewhere (33). Thioredoxin antibody and goat anti-rabbit IgG conjugated with alkaline phosphatase were used for immunodetection. Release of the periplasmic protein MalE was detected with MalE antibodies provided by Dr. G. Richarme. Control experiments with pure commercial thioredoxin, in the presence of gadolinium, were performed to ensure that this ion does not modify immunodetection. Quantification of the intensity of protein bands was performed by scan densitometry.
Thioredoxin was determined enzymatically with thioredoxin reductase as described previously (34). Osmotic downshock was performed as described above. To determine the total thioredoxin cell content, cells were disrupted by two passages through a French press cell. The resulting suspension was centrifuged for 30 min at 90,000 rpm in a TL-100 ultracentrifuge (Beckman Instruments), and the supernatant was kept for thioredoxin determination. All measurements were made in triplicate.

RESULTS
Efflux of Potassium and Glutamate-The first response of E. coli cells to an osmotic upshock is an uptake of potassium (35) which is released when the external osmolarity is lowered (14,17,19). The uptake and release of potassium by E. coli cells, in response to changes in the external NaCl concentration, were determined by monitoring the external potassium concentration with a valinomycin electrode (29). Cells, resuspended in a 200 mM NaCl medium were diluted 25-fold in media containing 0 to 80 mM NaCl. Upon shock, the cells lost 50 -85% of their potassium content. Whatever the intensity of the osmotic downshock, no difference was observed between wild type and MscL Ϫ strains (not shown).
In parallel with the uptake of potassium, cells synthesize and accumulate glutamate (31,36). In our experiments, cells (wild type, MscL Ϫ , and restored) growing in minimal medium were subjected to a large osmotic upshock by addition of NaCl (500 mM final concentration). This resulted in the synthesis of glutamate, the level of which reached some 300 nmol/mg of dry weight within 10 min before declining, probably as trehalose was synthesized. A 5-fold dilution with NaCl-free medium resulted in a monophasic release of 80% of the internal glutamate from wild type, MscL Ϫ strains, and from the restored strain (not shown). No clear differences between the three strains were observed in three separate experiments.
Efflux of Trehalose-In a second phase of adaptation to high osmolarity media, E. coli synthesizes trehalose, which accumulates in the cytoplasm and partially replaces potassium and glutamate (31,37). Cells (wild type, MscL Ϫ , and restored) growing in minimal medium were shocked by addition of NaCl (500 mM final concentration). This resulted in the synthesis of trehalose, the level of which reached some 300 nmol/mg of dry weight within 2 h (Fig. 1). A 5-fold dilution with NaCl-free medium caused similar releases of trehalose from the wild type (Fig. 1A), MscL Ϫ (Fig. 1B) and restored strains (not shown). Interestingly, the release of trehalose was clearly biphasic with a rapid and a slow component in each of three similar experiments.
Efflux of Glycine Betaine-E. coli cells subjected to an osmotic shock accumulate glycine betaine if it is present in the external medium (38). Glycine betaine then replaces potassium glutamate or trehalose (31). In the following experiments, cells were grown in minimal medium in the presence of 500 mM NaCl, washed, and suspended at room temperature in Hepes-KOH buffer, pH 7.5, containing 600 mM NaCl. Addition of 1 mM radioactive glycine betaine to the medium resulted in the uptake of more than 1 mol/mg of dry weight (Fig. 2). Given the cell volume, previously measured under similar osmotic conditions and in the presence of glycine betaine (39), this corresponds to a concentration higher than 1 M. A 5-fold dilution of the cell suspension with NaCl-free medium resulted in the total release of glycine betaine, which was then accumulated back to a much lower level. If, immediately after the hypo-osmotic shock and release of glycine betaine, the initial high osmolarity of the medium was restored by addition of concentrated NaCl, the internal concentration of glycine betaine returned to a level similar to that present initially. This shows that the massive efflux of glycine triggered by osmotic downshock occurred without damage to the cells. We studied the release of betaine from wild-type and MscL Ϫ strains as a function of the intensity of the shock. Cells that had previously accumulated glycine betaine in a buffer containing 10 mM Hepes-KOH, pH 7.4, 600 mM NaCl, were shocked in media of decreasing concentrations of NaCl. Efflux of glycine betaine occurred only above a threshold corresponding to a change of about 200 mM in external NaCl concentration. The extent of betaine release paralleled that of the shock intensity, and no significant difference between the two strains could be observed, regardless of intensity of the shock, in three separate experiments (Fig. 3).
Gadolinium is the only known inhibitor of bacterial MS channels (14), and it is unfortunately relatively nonspecific. It could not be used in the trehalose experiments because it precipitates with phosphate in culture media. We examined its possible inhibitory effect on the release of glycine betaine. However, in the absence of shock, when the cells were simply diluted in an iso-osmotic medium containing 1 mM gadolinium, 30% of the internal glycine betaine was released as compared with the control. The reason for this effect is unclear.
In all these experiments, the release of glycine betaine was completed within the 3-4 s required to dilute and filter the cell suspension. We attempted to increase the time resolution of our experiments by performing the rapid filtration assay described under "Experimental Procedures." We found that the complete release of glycine betaine took less than 200 ms, the limit of resolution for this assay. No difference was observed between the strains for this type of assay, whatever the intensity of the shock. Efflux of Thioredoxin-Some small cytosolic proteins are released by osmotic shock from Tris-EDTA treated E. coli cells (24 -27). Our results confirm this surprising finding for thioredoxin. Cells were resuspended in a plasmolysis medium and subjected to an osmotic downshock by a 5-fold dilution in water as described previously (25), and release of thioredoxin was monitored by immunodetection. The Tris-EDTA treatment that disrupts the outer membrane was found essential; no release of thioredoxin was observed in the absence of treatment (Fig. 4A).
FIG. 2. Shock-induced release of accumulated glycine betaine from wild type cells. Cells grown in M9 medium containing 500 mM NaCl were suspended at room temperature in 10 mM Hepes-KOH buffer, pH 7.2, 600 mM NaCl containing 0.4% (w/v) glucose. Radiolabeled glycine betaine (1 mM final concentration) was added at zero time to initiate glycine betaine uptake, and samples were withdrawn and filtered at intervals. After 150 min the suspension was divided into three fractions. The first served as a control (f), the second (OE) and the third (q) were diluted 5-fold in the same medium devoid of NaCl and glycine betaine. Immediately after the osmotic downshock, the initial high osmolarity of the medium was restored in the third fraction (q) by addition of concentrated NaCl without further addition of glycine betaine. At intervals, samples were withdrawn and filtered, and the filters were counted for radioactivity.

FIG. 3. Release of accumulated glycine betaine as a function of shock intensity in wild type (f) and
MscL ؊ (q) strains. Cells that had accumulated glycine betaine in a buffer containing 600 mM NaCl were subjected to osmotic shock by a 5-fold dilution of the suspension with shock media of various NaCl concentrations (0 -600 mM). The diluted suspension was filtered, and the filters were counted for radioactivity. The glycine betaine content retained in the cells after the shock is plotted against the NaCl concentration of the shock medium.

FIG. 1. Shock-induced release of endogenous trehalose in wild type (A) and MscL Ϫ (B) strains.
Cells growing in M9 medium, at 37°C, were first subjected to a hyperosmotic shock (first arrow) by addition of NaCl (500 mM final concentration), to induce the synthesis of trehalose. Trehalose efflux was triggered by a hypo-osmotic shock (second arrow) obtained by diluting the cell culture 5-fold in the growth medium. Ⅺ, before osmotic upshock; f, after osmotic upshock; OE, after osmotic downshock.
While no release of thioredoxin was observed when Tris-EDTAtreated cells were simply diluted in an iso-osmotic medium, the release was total when the cells were osmotically shocked (Fig.  4B). Release was totally blocked by 1 mM gadolinium in the shock medium (Fig. 4B). In the MscL Ϫ strain, release was severely impaired and most of the thioredoxin remained in the cells (Fig. 4C). Significantly, in the restored strain, the efflux of thioredoxin was similar to that found for the wild type strain (Fig. 4C). A trivial explanation for these results could be that, for unknown reasons, the outer membrane in MscL Ϫ is more resistant to the Tris-EDTA treatment; thioredoxin would then simply remain trapped in the periplasm. Fig. 4D shows that such is not the case. The release, triggered by osmotic shock, of the periplasmic maltose binding protein MalE was similar in both wild type and MscL Ϫ cells (n ϭ 2). The impairment of thioredoxin release in MscL Ϫ was observed in six separate experiments. However, it was never complete. Moreover, it showed some variability; the amount of thioredoxin released by MscL Ϫ cells ranged from 19 to 40% of the total cell content versus 99 -100% in wild type cells (e.g. compare Fig. 4, C and  D). Finally, we studied the same phenomenon using a different, quantitative, assay in which thioredoxin was enzymatically determined with thioredoxin reductase (34). In all three strains, the cell thioredoxin content was found to be about 60 pmol/mg of dry weight in good agreement with a previous determination (25). In three separate experiments, a 90 -100% release of thioredoxin was observed from wild type EDTAtreated cells upon osmotic shock. In contrast, MscL Ϫ EDTAtreated cells released only 16 -20% of their total thioredoxin. In the mutant expressing the mscL gene, 82-90% of thioredoxin was released upon shock. DISCUSSION The study of osmoregulation in bacteria had initially focused on the accumulation of potassium and osmoprotectants following osmotic upshock. More recent reports, dealing with different bacterial species, have shown that osmoprotectants are released upon osmotic downshock (18 -23). As shown here for glycine betaine, efflux occurs without lysis of the cells, and the cells are able to accumulate glycine betaine again immediately after the shock if necessary. This efflux is thus clearly an important physiological process, probably widespread in bacteria, which allows an immediate release of membrane tension. The efflux pathways have not been identified, but they clearly differ from the transport systems used for influx. In L. plantarum a fast and a slow efflux of glycine betaine were reported (21). No slow phase of glycine betaine efflux was observed in our experiments, but it may have been masked by reuptake. It is noteworthy that we observed a biphasic release of trehalose, which is not transported back into the cells. It was proposed that the slow component of glycine betaine efflux in L. plantarum is carrier-mediated. A similar system may exist for trehalose in E. coli. A fast component of efflux was observed for all osmoprotectants. It is highly probable that it is channel-mediated. This conclusion is dictated by the extreme rapidity of the phenomenon. In this study we attempted to increase the time resolution. The efflux rate of glycine betaine was found to be higher than 300,000 nmol/min/mg of dry weight, a figure which is not compatible with a carrier-mediated process. The high conductance MS channels present in the cell membrane of bacteria are therefore obvious candidates for the fast efflux pathway (14,15). Nevertheless, we have shown here that deletion of mscL, the only known gene that codes for a bacterial MS channel (4), has no detectable effect on the release of K ϩ , glutamate, trehalose, and glycine betaine. However, as MscM and MscS channels are present (3), this does not rule out the involvement of MS channels in the process.
A few cytoplasmic proteins including thioredoxin, translation elongation factor Tu, and DnaK are known to be released by osmotic shock (24 -26). Release of these cytoplasmic proteins has been documented in Tris-EDTA-treated cells. When this FIG. 4. Shock-induced release of thioredoxin. A, thioredoxin is not released by cells that have not been treated by EDTA. Wild type cells were suspended in 10 mM Tris, pH 7.6, 20% (w/v) sucrose, and diluted 5-fold with the same buffer (iso-osmotic dilution) or distilled water (osmotic shock). The suspension was centrifuged, and the presence of thioredoxin in the pellet and in the supernatant, after trichloroacetic acid precipitation, was revealed by immunoblotting. 1, thioredoxin cell content before dilution or shock; 2, thioredoxin in the pellet (P) and supernatant (S), after iso-osmotic dilution; 3, thioredoxin in the pellet and supernatant after osmotic shock; 4, thioredoxin in the pellet and supernatant after osmotic shock in the presence of 1 mM gadolinium. B, release of thioredoxin after EDTA treatment. Wild type cells were suspended in 10 mM Tris, pH 7.6, 2.5 mM EDTA, 20% (w/v) sucrose and diluted 5-fold with the same buffer or distilled water. 1, 2, and 3 are as indicated above. C, release of thioredoxin is impaired in MscL Ϫ . Wild type (WT), MscL Ϫ and restored (R) cells were EDTA-treated and subjected to an osmotic shock. After centrifugation, thioredoxin in the pellet (P) and in the supernatant (S) was revealed by immunoblotting. Quantification of protein bands by densitometry indicated that the release of thioredoxin from wild type, MscL Ϫ , and restored cells represented, respectively, 99, 34, and 88% of the total thioredoxin content of each strain. D, retention of thioredoxin in MscL Ϫ cells is not due to an inadequate treatment of the outer membrane for this strain. Wild type (WT) and MscL Ϫ cells were EDTA-treated, shocked, and centrifuged. The release of the periplasmic binding protein MalE to the supernatant (S), detected by immunoblotting, was similar, indicating that in both strains the Tris-EDTA treatment annihilated the outer membrane. In contrast, the release of thioredoxin from wild type and MscL Ϫ cells represented, respectively, 100 and 19% of the total thioredoxin content of each strain. treatment was omitted, thioredoxin was not excreted into the external medium (23) (this study). As the Tris-EDTA treatment disrupts the outer membrane, in intact cells, thioredoxin is in fact released during shock from the cytoplasm to the periplasm. Thioredoxin is a 12-kDa protein that participates in several thiol-dependent, cellular-reductive processes (40). The physiological significance of this transfer of thioredoxin into the periplasm is unclear.
The release of thioredoxin was severely impaired in cells lacking MscL, and was totally blocked by gadolinium, an inhibitor of bacterial MS channels (14), and in particular of MscL (5). This demonstrates clearly that most of the thioredoxin efflux occurs via MscL. Residual thioredoxin efflux may be carried by MscS. The finding that a protein can pass through a channel is unexpected and surprising. However, the conductance of MscL (1500 pS in 0.1 M KCl) is unusually high for an endogenous channel. The size of the open channel pore was recently probed experimentally by measuring blockade of MscL by large organic polymers (poly-L-lysines) of increasing sizes. The open pore diameter was estimated to be around 40 Å, consistent with direct calculation based on the channel conductance (41). The size of the thioredoxin molecule, determined by x-ray diffraction studies, is 25 ϫ 30 ϫ 35 Å (42). It is thus compatible with that of the pore. In this context, it is noteworthy that high conductance MS channels of B. subtilis, reconstituted in a planar lipid bilayer, have recently been found to be able to catalyze the transfer of double-stranded DNA (43).
Our results demonstrate for the first time that the opening of MscL does occur in vivo, during osmotic shock. This could not be inferred from previous patch-clamp studies. In these experiments, negative pressure is applied via the pipette to the patch, but the relevant parameter is the membrane tension which is related to pressure by Laplace's law, via the radius of curvature of the patch. The radius of curvature of the patch, which cannot be evaluated easily, differs from that of the cells. These considerations have precluded a simple comparison between the pressure applied in patch-clamp experiments and the intensity of osmotic shock felt by intact cells.
MscL is the channel with the highest conductance, which requires the highest tension to open (2). Therefore, if MscL opens during osmotic shock, so do the other MS channels, MscM and MscS. It is thus highly probable that rapid efflux of osmoprotectants occurs via these channels (and via MscL in the wild type). MscL is present in a the large number of bacteria (9). If it is not required for osmoprotectant efflux, at least when MscS and MscM are present, what is its physiological role? One possibility is that, being activated at the highest tension, it is an emergency device used as a last resort. The multiplicity of MS channels present in bacteria would reflect a redundancy dictated by the importance of osmoregulation for cell viability. Another possibility, which might account for the unusually high conductance of MscL, is that transfer of some proteins from the cytoplasm to the periplasm during osmotic shock, as documented herein for thioredoxin, has a physiological meaning which is not understood at present.
In conclusion, we have shown that MscL opens during osmotic downshock and is responsible for the release of thioredoxin, and possibly of other small proteins. A full understanding of the process taking place during osmotic downshock will require the identification of the various genes coding for MS channels in E. coli.