YbdG in Escherichia coli is a threshold-setting mechanosensitive channel with MscM activity

Abstract

We describe a mechanosensitive (MS) channel that has mechanosensitive channel of miniconductance (MscM) activity, and displays unique properties with respect to gating. Mechanosensitive channels respond to membrane tension, are ubiquitous from bacteria to man, and exhibit a great diversity in structure and function. These channels protect Bacteria and Archaea against hypoosmotic shock and are critical determinants of shape in chloroplasts. Given the dominant roles played in bacteria by the mechanosensitive channel of small conductance (MscS) and the mechanosensitive channel of large conductance (MscL), the role of the multiple MS channel homologs observed in most organisms remains obscure. Here we demonstrate that a MscS homolog, YbdG, extends the range of hypoosmotic shock that Escherichia coli cells can survive, but its expression level is insufficient to protect against severe shocks. Overexpression of the YbdG protein provides complete protection. Transcription and translation of the ybdG gene are enhanced by osmotic stress consistent with a role for the protein in survival of hypoosmotic shock. Measurement of the conductance of the native channel by standard patch clamp methods was not possible. However, a fully functional YbdG mutant channel, V229A, exhibits a conductance in membrane patches consistent with MscM activity. We find that MscM activities arise from more than one gene product because ybdG deletion mutants still exhibit an occasional MscM-like conductance. We propose that ybdG encodes a low-abundance MscM-type MS channel, which in cells relieves low levels of membrane tension, obviating the need to activate the major MS channels, MscS and MscL.

Bacterial mechanosensitive (MS) channels play a critical role in the retention of cell integrity during hypoosmotic shock (1). Rapid transfer of cells from a high osmolarity environment to one of low osmolarity causes influx of water into the cytoplasm down the osmotic gradient. This influx produces increased tension in the membrane bilayer that gates MS channels, leading to transient formation of pores in the cell membrane (2–4). Solute efflux from the cytoplasm through these pores reduces the rate of water influx, thus averting cell lysis (1). The channels close immediately after the stress is relieved, guaranteeing a short duration for the open pore. Martinac and colleagues discovered that hypoosmotic shock could be mimicked by application of slight pressure (0.1–0.3 atm) across an isolated bacterial membrane patch, leading to increased transmembrane electrical current (5). Subsequently, multiple MS conductances, activated at different pressure thresholds, were identified after solubilization of membrane fractions and reconstitution of protein into artificial lipid bilayers (6, 7).

Gene cloning and knockout strategies allowed a more controlled analysis of MS channel types and resulted in the identification and categorization of three major Escherichia coli MS channels: mechanosensitive channel of large conductance (MscL) (∼3 nS conductance), mechanosensitive channel of small conductance (MscS) (∼1.25 nS), and potassium-dependent mechanosensitive channel (MscK) (∼0.875 nS), with MscS and MscK belonging to the same class (1, 8, 9). Members of these families can be found in Bacteria, many Archaea, some plants, fungi, and oomycetes (10–12) but are best studied in E. coli (13–15). Another channel activity, mechanosensitive channel of miniconductance (MscM), is occasionally observed in patch-clamp recordings (6), but the corresponding gene is not known. This channel has a low conductance (∼0.375 nS) but exhibits long-lived open states in response to increased tension.

The well-characterized E. coli MscS protein (286 aa) has been crystallized, leading to closed (wild type; Fig. S1A) and open (A106V mutant) conformations being solved (13, 16, 17). Each subunit of the homoheptameric protein contains three transmembrane (TM) helices. The pore-forming sequence, TM3a, has a conserved pattern of Ala and Gly residues (Fig. S1B) at the interface between TM3a helices (10, 16–18). Mutations that alter the Ala-Gly packing modify the gating characteristics of the channel (18). This amino acid pattern is not highly conserved across the whole MscS family, which suggests that the multiple MscS homologs observed in many organisms might encode channels with unique gating properties. E. coli has four homologs additional to MscS and MscK: YjeP (1107 aa), YbiO (741 aa), YbdG (415 aa), and YnaI (343 aa) (Fig. S2). We report the characterization of the YbdG homolog and conclude that it is a component of the MscM channel activity.

Results

Predicted YbdG Structure Differs from MscS.

YbdG differs from MscS in three respects. First, YbdG has a large membrane domain, which can be organized into five TM spans with the fifth TM helix bearing the closest sequence similarity to the MscS pore-lining helix (Fig. S1). Second, the YbdG carboxyl-terminal cytoplasmic domain bears an ∼50-amino-acid insertion that appears at, or close to, the junction between the upper β and the αβ domains in the MscS structure (Fig. S1 A and C). This insertion is moderately conserved among YbdG homologs (Fig. S3) but does not appear at a significant frequency in other proteins in the nonredundant database. Pairwise alignment of full-length YbdG with MscS using BLAST (19) terminated at the start of the insert. If the ∼50-residue insertion was removed in silico, the alignment continued to the carboxy terminus (BLAST score 2 × E⁻⁷; Table S1). Third, the predicted pore sequence is not highly conserved relative to MscS except at the upper region surrounding G101 (MscS numbering) (Fig. S1B). These differences, coupled with our previous demonstration that MscS and MscL are the dominant MS channels in E. coli (1), raised three questions. Does the ybdG gene encode a functional protein? Does the protein product retain a physiological role? And finally, does ybdG encode the MscM channel?

YbdG Expression Is Dependent on Growth Phase and Osmolarity.

The ybdG gene lies 108 bp downstream of nfnB (also called nfsB), the structural gene for dihydropteridine reductase (20). Immediately 3′ (81 bp) to ybdG is the pheP gene and its terminator lies proximal to the predicted 3′ end of the ybdG ORF (Fig. S4). There are no obvious transcriptional regulator binding sites 5′ to ybdG (Discussion), but nfnB is known to be regulated by MarA (21, 22). There is no classical transcription terminator between nfnB and ybdG. However, the ybdG gene has a putative promoter (a −10 sequence centered ∼30 nucleotides 5′ to the ATG) and a potential Shine–Dalgarno sequence (centered ∼8 nucleotides 5′ to ATG), indicating that the protein could be expressed independently of nfnB. We performed quantitative real-time PCR (qRT-PCR) to determine if ybdG expression occurs and to investigate coregulation of ybdG and nfnB expression. Expression was normalized against rpoB levels, which are expected to remain relatively constant (23). Both nfnB and ybdG were transcribed during exponential phase growth in LB medium, but only ybdG mRNA levels increased as cells entered stationary phase (Fig. 1A). Expression of mscS and mscL has been shown to be stimulated by growth at high osmolarity in an RpoS-dependent manner (24). Growth into exponential phase in high osmolarity similarly stimulated ybdG mRNA synthesis. In contrast to mscS and mscL genes, elimination of RpoS increased ybdG mRNA abundance. Expression of nfnB was not significantly changed by an rpoS mutation. The effect of rpoS mutation was most marked at high osmolarity where it resulted in a further 10-fold increase in pools of ybdG mRNA (Fig. 1A). These data can be rationalized by the known stabilization of RpoS protein at high osmolarity and in stationary phase (25). Previously, superinduction of genes in RpoS null mutants has been associated with relief of the competition between σ⁷⁰ and σ^S (26).

Fig. 1.

Regulation of YbdG transcription and translation. (A) The ybdG gene is transcribed independently of nfnB. Transcription levels of ybdG (bars with light shading), nfnB (bars with dark shading), and rpoB (internal control) were determined by quantitative real-time PCR. Frag1 or MJF372 (Frag1 rpoS::Tn10) cells were grown in LB medium with (+) or without (−) an additional 0.5 M NaCl, to the respective growth phases. Total RNA fractions were isolated and reverse transcribed into cDNA. Transcript levels were normalized to the internal control gene rpoB. All transcript levels are relative to sample Frag1 grown to exponential phase without additional NaCl. Error bars indicate SD of three independent experiments. (B) Growth phase, salt concentration, and RpoS affect YbdG protein levels. Cultures of Frag1 (+) and MJF372 (Frag1 rpoS::Tn10) (−) were grown in LB (Upper) or LB + 0.5 M NaCl (Lower) media and cells were harvested at the indicated growth phase. Membrane fractions were isolated and 25 μg of total membrane protein was separated on SDS/PAGE. Western blots were probed with antibodies specific for YbdG followed by incubation with anti-rabbit-HRP antibodies. Blots were developed by the standard ECL method.

Protein abundance during different growth conditions was investigated using peptide-specific antisera to YbdG (Fig. 1B). In the presence of 0.5 M NaCl, protein expression matched the pattern of mRNA production (Fig. 1 A and B), confirming the role of high osmolarity in YbdG abundance in the cell. In the absence of salt, protein expression declined as the cells entered stationary phase (Fig. 1B) and this pattern was also observed in an RpoS mutant despite the increased mRNA levels. Surprisingly, in cells grown at low osmolarity there was no significant correlation between mRNA production and YbdG protein production. We infer from these data that ybdG mRNA is subject to translational control that is overcome when the cells are exposed to hyperosmotic stress.

Oligomeric Structure of YbdG.

MscS is a homoheptamer (16). With significant variations in pore-lining sequences between MS channel homologs, the potential exists for alternative oligomeric structures. Using five independent preparations of YbdG, a mass range of 546 ± 25 kDa was obtained by blue native gel analysis, which after correction for dye binding (27) indicated a mass for the channel complex at 303 ± 14 kDa, consistent with six to seven 45-kDa subunits per channel (Fig. S5A). Similar analyses with MscS (n = 6) gave a stoichiometry of 6.6 ± 0.4.

We have previously demonstrated the efficacy of cross-linking to estimate the oligomeric state of MscS channel complexes (28). A single native Cys residue (Cys102) is at the periplasmic end of the putative TM1 helix of YbdG (Fig. S1). A single band at ∼38 kDa was observed on nonreducing SDS/PAGE, indicating that Cys102 does not spontaneously cross-link subunits. However, a significant proportion of YbdG is driven into cross-linked dimers by oxidation with copper phenanthroline (Cu-Phen) reagent (Fig. 2). Introduction of a Cys residue into the carboxyl-terminal domain (YbdG A388C) at a position similar to S267C in MscS that allows detection of heptamers (28) did not significantly increase oligomer formation in the presence of Cu-Phen (Fig. 2). In contrast, introduction of a Cys residue into the predicted periplasmic vestibule (S164C or S158C) led to the formation of higher oligomers up to and including a heptamer (Fig. 2). Together, the blue native and cross-linking data support the formation of heptamer by YbdG.

Fig. 2.

The YbdG protein assembles as a homoheptamer in the membrane. Membrane fractions of MJF612 cells expressing the indicated mutant were isolated and cysteine cross-linking was performed using copper phenanthroline (Cu-phen) (28). A total of 75 μg of membrane proteins was incubated with 167 μM phenanthroline (+) or the solvent ethanol (−) and 13 μg was separated on SDS/PAGE under nonreducing (A) or reducing (B) conditions. Western blots were probed with Penta-His antibodies and developed using the standard ECL method.

YbdG Contributes to Cell Viability During Hypoosmotic Shock.

Our early work showed that MscS and MscL are the dominant channels providing protection against hypoosmotic shock (1). We extended this analysis to mutants lacking YbdG. A ybdG::apr replacement mutant was constructed (Fig. S4) and transferred to strains MJF429 (MscK⁻, MscS⁻) and MJF465 (MscK⁻, MscS⁻, MscL⁻) for analysis (Table S2 and Fig. S5B). As reported previously, MJF465, which retains YbdG, suffers a substantial loss of viability when subjected to a rapid osmotic shock ≥0.25 M NaCl, but cells are able to survive smaller osmotic shifts (Fig. 3A). A quadruple MS channel null strain MJF612 (MJF465 ybdG::apr) exhibits a lowered threshold at which cell death occurs (0.15 M for MJF612 compared with 0.25 M NaCl for MJF465; Fig. 3A). Cells that retain YbdG (MJF465) exhibit 3-fold better survival than MJF612. Thus native levels of expression of YbdG confer protection against mild hypoosmotic shock. Protection could be enhanced by overexpression of YbdG (see Materials and Methods for plasmid construction). Basal expression did not offer protection from a 0.5-M hypoosmotic shock (Fig. 3B). When YbdG expression was increased (0.3 mM IPTG, 30 min), 100% survival was restored (Fig. 3B). Thus, YbdG possesses the capacity to protect cells against hypoosmotic shock, a characteristic expected of a functional MS channel.

Fig. 3.

YbdG is a functional MS channel, essential for survival of low levels of osmotic shock. (A) Strains Frag1 (MscK⁺, MscS⁺, MscL⁺, YbdG⁺) (solid bars), MJF465 (MscK⁻, MscS⁻, MscL⁻, YbdG⁺) (bars with light shading), and MJF612 (MscK⁻, MscS⁻, MscL⁻, YbdG⁻) (bars with dark shading) were grown in McIlvaine's medium supplemented with 0.5 M NaCl, to midexponential phase. Cultures were diluted 20-fold into McIlvaine's medium supplemented with various NaCl concentrations, creating the indicated osmotic shock. Cells were recovered overnight on solid media containing the same level of additional NaCl and the percentage of survival was determined. *Significance after Student's t test comparing MJF612 survival to MJF465 (P = 0.025). Error bars indicate SD of at least three independent experiments. (B) Overexpression of YbdG confers protection against a 0.5-M NaCl hypoosmotic shock. MJF612 cells either alone (con) or expressing WT YbdG or mutant V229A YbdG were adapted in LB + 0.5 M NaCl and then diluted 20-fold into LB medium, creating an osmotic shock. Plasmid expression was either basal (solid bars) or induced with 0.3 mM IPTG for 30 min (shaded bars). Percentage survival was calculated and the mean of at least three independent experiments is shown; error bars indicate SD.

Electrophysiological Characterization of YbdG Channels.

Previously we noted that mutant cells lacking MscL, MscS, and MscK can display 350–400 pS MscM activities in patch-clamp assays (1, 29) but openings of this size were infrequent (∼5% occurrence) and unpredictable. To determine whether YbdG is a component of the MscM channel activity, strains MJF429 (YbdG⁺) and MJF611 (ΔybdG), in which MscL acts as a reference for the presence of MS channels in each patch (30), were compared. No obvious differences were detected between membrane patches of these two strains (n = 15 for each strain) (Fig. 4). MscM-sized openings were observed only in occasional patches in the presence or absence of chromosomal ybdG. Moreover, overexpression of wild-type YbdG in either MJF429 or MJF611 did not lead to new or more abundant channel openings (n = 15 for each strain).

Fig. 4.

Channel activity of YbdG mutant V229A. Strain MJF429 (MscL⁺, MscS⁻, MscK⁻, YbdG⁺) alone or strain MJF611 (MscL⁺, MscS⁻, MscK⁻, YbdG⁻) expressing the YbdG V229A mutant was analyzed by patch clamp after protoplast formation. Excised membrane patches were clamped at 20 mV holding potential. A representative trace is depicted for each strain, exhibiting channel openings comparable to MscM activity (1). The data table summarizes the number of patches recorded that exhibited MscM-like conductances from various strains in the absence (−) or presence of YbdG wild-type (WT) or mutant V229A protein. YbdG V229A expression in MJF611 led to a significant increase in the number of patches observed to contain MscM activity.

We determined that YbdG protein is present in membrane samples derived from cephalexin-treated cells, the immediate precursor of giant protoplasts used for patch clamp (5). YbdG accumulated at levels similar to MscS protein, which is readily assayed (Fig. S5C). Thus, failure to detect abundant single channels was not due to lack of YbdG protein, but must reflect specific requirements for gating that are lacking in membrane patches. To isolate mutant YbdG channels with altered gating frequencies, we randomly mutagenized the ybdG gene and screened for mutant channels that suppressed the potassium transport deficiency of a strain deleted for the major K⁺ transport systems. This selection operates on the basis that the channel gates frequently enough to open the nonselective MS pore and allow K⁺ ions to move into the cell down the membrane potential; through frequent cycles of transient openings, followed by rapid closure, the cell can acquire the K⁺ needed for growth (31). Strain MJF622 is deficient for both potassium uptake and YbdG protein (Table S2) and requires ≥30 mM K⁺ for rapid growth. Gating mutants were identified by growth of the transformants on medium containing only 5 mM K⁺. Among several mutants, V229A was chosen for analysis because it proved functional, protecting cells from hypoosmotic-induced death when overexpressed (Fig. 3B), and it provided strong complementation (Fig. S6), suggesting an increased frequency of gating. This mutation location in YbdG is also sufficiently distant from the pore that the channel conductance should not be affected. In patch clamp, YbdG V229A exhibited a high frequency of MscM-like channel activity when expressed in MJF611 (ΔybdG): 9/23 patches exhibited one to two long open-dwell openings of 350–400 pS that are indistinguishable from those of MscM (Fig. 4). However, no increased channel activity was observed when the same mutant was expressed in MJF429 (YbdG⁺). These data are consistent with competition for a limited supply of a protein, metabolite, or lipid required for activity or with heteromeric channels being formed that are inhibited from gating by integration of wild-type and mutant subunits into the same complex.

Discussion

In this study we demonstrate that one of the previously uncharacterized homologs of MscS in E. coli, YbdG, is a functional and physiologically relevant MS channel that exhibits conductance properties similar to MscM. Overexpression of this channel gives excellent protection against hypoosmotic shock, equivalent to that provided by MscS or MscL. Because the contribution of chromosomally expressed YbdG to protection is subtle, it is clear that the normal abundance of this channel protein is too low to afford the high levels of protection seen with MscS and MscL. This lack of channel abundance occurs despite increased transcription and translation in response to high osmolarity. Expression of YbdG is inhibited by RpoS in contrast to the RpoS-dependent expression of MscS, MscL, and YbiO in E. coli (24, 32). Thus expression at high osmolarity is the outcome of two conflicting effects: stimulation by an unknown signal reflecting the increased osmolarity of the environment and the increased stability of RpoS observed at high osmolarity. Analysis of the ybdG promoter region did not identify any major transcriptional regulatory sites. RegulonDB (33) identifies RpoE (σE) and CrcB as potential regulators. However, the published data suggest that the RpoE effect is, at best, weak (34) and ybdG is not mentioned in the most recent complete analysis (35). The CrcB effect is mediated by modulation of DNA topology caused by overexpression of this protein, rather than by direct binding to the promoter region (36).

The principal known function of MS channels is protection against hypoosmotic shock (1). Removal of YbdG from a strain lacking MscK, MscS, and MscL lowers the threshold of hypoosmotic stress at which death occurs (Fig. 3). The implication of these data is that YbdG should gate at a lower pressure than MscK, MscS, or MscL. However, in patch clamp no correlation could be found between YbdG expression and the frequency of MscM-containing patches. Even overexpression of YbdG did not significantly increase the observation of MscM-like openings. The lack of correlation between physiological and electrophysiological assays has been noted previously for mutants of MscS and MscL channels (18, 37, 38). The basis for these observations is unknown. Recent work has reported changes in lipid composition in cells grown at high osmolarity (39). Because MS channels should be particularly sensitive to their lipid environment (40–42), the inability to observe the activity of wild-type YbdG in patch clamp may reflect differences between lipid compositions of the cells used in the two assays. Physiological assays are conducted on cells grown at high osmolarity whereas cells for patch clamp are grown at low osmolarity. Incubating cells in cephalexin and high NaCl halts growth and does not generate filamentous cells that are essential for protoplast formation, precluding tests of “high-salt-treated” protoplasts in patch clamp. The screen resulting in isolation of the V229A mutant was conducted at low osmolarity and thus the selection was specifically for channels that could gate under these conditions.

In the earliest experiments measuring E. coli MS channels using cytoplasmic membranes fused with phosphatidylcholine liposomes (5–7, 43), multiple activities were observed that could be differentiated by their kinetics. Individual patches contained single types of activities rather than the mixtures of channels routinely measured in cells, suggesting that the channels are clustered in the mixed membranes of the reconstituted system (6). These data can be interpreted to suggest that some channel activities require either specific lipids for their activity or the dilution of inhibitory lipids to allow gating. Microarray data indicate transcription of the MscS homologs under a range of conditions, suggesting that the proteins, like YbdG, are expressed (http://genexpdb.ou.edu/). Therefore the failure to detect frequent channel activities of the other MscS homologs by electrophysiology indicates a further subtlety in their regulation that may involve specific lipid requirements for activity. The ease with which MscS, MscL, and MscK are observed may be due to a combination of their abundance and an inherent lack of constraint in their lipid interactions.

A remarkable observation is that the YbdG mutant V229A activity is apparently suppressed by coexpression with the wild-type subunit. Possible explanations for this phenomenon include formation of heteromeric channels, competition for limiting amounts of either a lipid or an activating molecule, and/or competition during assembly into the membrane. Expression from the chromosome should be substantially lower than from the plasmid. Purification of mutant YbdGH₆ after expression in both ∆ybdG and YbdG⁺ cells did not reveal any significant change in protein abundance when wild-type subunits were present, but we have been unable to detect heteromeric complexes. Thus, the available data point to complex regulation of YbdG channel activity by either an unknown protein or specific lipids. The V229A channel exhibits activity similar to MscM, namely a conductance of 350–400 pS, sustained open dwell times, and rarity. Only small numbers of V229A channels were recorded in any one patch despite control experiments that indicated an abundance of protein (Fig. S5C). It appears that the mutation, although facilitating gating, does not completely compensate for the missing factors required to gate YbdG in membrane patches.

Genome analysis across the microbial and plant kingdoms has revealed the presence of multiple MscS homologs, but only rarely multiple MscL proteins (10–12). Thus, it seems plausible that MscS homologs have evolved a wide range of sequences that enable them to gate in response to different tensions and/or lipid and environmental contexts. This multiplicity would allow a diversity of channel characters that introduces subtlety to the hypoosmotic shock response. In contrast, MscL may not be able to be engineered to possess such subtlety and has evolved as the “channel of last resort,” opening when the increase in turgor is too rapid to be countered by the multiple members of the MscS family.

Materials and Methods

Strains and Media.

All strains used are derivatives of E. coli K12 and are listed in Table S2. JM109 and JM110 were used for transformation of newly generated mutants. Frag1 derivative strains were grown at 37 °C in either LB medium (10 g tryptone, 5 g yeast extract, 5 g NaCl per liter) or McIlvaine’s buffer, pH 7.0 [220 mOsmol; 8.58 g Na₂HPO₄, 0.87 g K₂HPO₄, 1.34 g citric acid, 1.0 g NH₄SO₄, 0.001 g thiamine, 0.1 g MgSO₄·7H₂O, and 0.002 g (NH₄)₂SO₄·FeSO₄·6H₂O per liter]. Strain DY330 was grown in LB at 32 °C. All strains were stored at 4 °C on agar plates containing selective antibiotic, where appropriate. Overnight cultures contained 25 μg/mL ampicillin, where appropriate, to ensure maintenance of a plasmid. Solid medium contained 14 g/L agar.

Primers.

All primers used in this study were manufactured by Sigma-Genosys and are listed in Table S3.

Bioinformatic Methods.

All programs used are available online. Protein sequence alignments were performed using either BLASTP (19) or MAFFT (44). Conservation analysis was performed using the WebLogo server (45). Protein hydrophobicity and mutant sequencing results were analyzed using the DNASTAR Lasergene program suite. All E. coli protein and DNA sequences were obtained from the Colibri Web Server (http://genolist.pasteur.fr/colibri). Reference was made to RegulonDB (33) and OU Gene expression databases (46).

Creation of ybdG Deletion Strains.

Deletion of the ybdG gene by homologous recombination was essentially as described (47), using the strategy depicted in Fig. S4 for primer design. See SI Materials and Methods for full details.

Cloning of the ybdG ORF.

The full ORF was amplified by standard PCR using primers YbdG-F and YbdG-R (Table S3). The obtained product was cloned into the pTrcHis2-TOPO vector (Invitrogen) and verified by sequencing. The cloned gene contained a number of nucleotide changes compared with the published sequence (Colibri Web Server); however, all were silent mutations leaving the protein sequence unaffected. Restriction sites NcoI and XhoI were introduced in frame at the start and stop of the gene, respectively, by site-directed mutagenesis using primers YbdGXhoI-1, YbdGXhoI-2, YbdGNcoI-1, and YbdGNcoI-2. To create a His₆-tagged protein, the plasmid pTrcMscSH₆ was used for subcloning (28). Successful creation of pTrcYbdGH₆ was checked by PCR amplification and restriction analysis and confirmed by sequencing.

Generation of YbdG Mutants.

Mutants were created either by QuikChange site-directed mutagenesis or by selection from XL1-generated mutated plasmids as specified in SI Materials and Methods.

Hypoosmotic Shock Assay.

To test channel function, assays were performed essentially as described previously (1, 28) with the minor modifications specified in SI Materials and Methods.

YbdG Antipeptide Antibody Purification.

The peptide sequence 276-FLDEDEMQRLNKAHL-290, in the C-terminal region of YbdG, was selected for the antibody target. Further details are given in SI Materials and Methods.

Membrane Preparations and Western Blot Analysis.

Protein expression and detection were carried out essentially as described previously (28) with the modifications outlined in SI Materials and Methods.

Cross-Linking of Cysteine Residues.

Cysteine cross-linking using copper phenanthroline was carried out on membrane samples as reported previously (28). Reactions were run on SDS gels and Western blot analysis was carried out as described above.

qRT-PCR Analysis.

Cells were grown to the desired optical density (exponential phase, OD₆₅₀ = 0.5; early stationary, OD₆₅₀ = 1; late stationary, OD₆₅₀ = 3), total RNA was isolated, and cDNA was obtained by reverse transcription and treated as described in SI Materials and Methods. Primers for qRT-PCR were designed within the ORF of nfnB and ybdG; rpoB was used as internal standard.

Electrophysiology.

Patch-clamp recordings were conducted as previously published (18), with amendments as in SI Materials and Methods.

Materials.

All media components were obtained from Oxoid and all general chemicals purchased from Sigma. Restriction enzymes and polymerases were from Roche or Promega and Novex precast SDS-polyacrylamide gels were from Invitrogen. All qRT-PCR materials and chemicals were from Roche. Kits for DNA or RNA isolation and purification were from Qiagen.