A Superfamily of Voltage-gated Sodium Channels in Bacteria*

NaChBac, a six-α-helical transmembrane-spanning protein cloned from Bacillus halodurans, is the first functionally characterized bacterial voltage-gated Na+-selective channel (Ren, D., Navarro, B., Xu, H., Yue, L., Shi, Q., and Clapham, D. E. (2001) Science 294, 2372-2375). As a highly expressing ion channel protein, NaChBac is an ideal candidate for high resolution structural determination and structure-function studies. The biological role of NaChBac, however, is still unknown. In this report, another 11 structurally related bacterial proteins are described. Two of these functionally expressed as voltage-dependent Na+ channels (NaVPZ from Paracoccus zeaxanthinifaciens and NaVSP from Silicibacter pomeroyi). NaVPZ and NaVSP share ∼40% amino acid sequence identity with NaChBac. When expressed in mammalian cell lines, both NaVPZ and NaVSP were Na+-selective and voltage-dependent. However, their kinetics and voltage dependence differ significantly. These single six-α-helical transmembrane-spanning subunits constitute a widely distributed superfamily (NaVBac) of channels in bacteria, implying a fundamental prokaryotic function. The degree of sequence homology (22-54%) is optimal for future comparisons of NaVBac structure and function of similarity and dissimilarity among NaVBac proteins. Thus, the NaVBac superfamily is fertile ground for crystallographic, electrophysiological, and microbiological studies.

Na ϩ channel (5,6), but the conductance has not been directly measured. In alkaliphilic bacteria, the prokaryotic ion channel responsible has not been identified. A bacterial 6-␣-helical transmembrane (6TM) channel subunit NaChBac was expressed in CHO cells as a functional voltage-gated Na ϩ channel (7), but its role in bacteria is still being elucidated.
The pore-forming subunits (␣ 1 ) of mammalian Na V and Ca V are composed of four similar repeats of 6TM domains (8,9), probably arising by gene duplication of a single 6TM gene (1,10). The first bacterial voltage-gated Na ϩ channel (NaChBac) functionally expressed in mammalian cells was cloned from Bacillus halodurans (7). It contains a single 6TM domain of 274 amino acids but almost certainly forms a tetramer (7). NaChBac voltagedependent activation and inactivation kinetics are 10 -100 times slower than that of Na V (7). NaChBac inactivation may result from pore inactivation (C-type inactivation), since it does not contain an obvious cytoplasmic inactivation gate (7).
Because single 6TM Na ϩ -selective ion channels do not appear to be present in vertebrates, the 24 TM structure of Na V s may have arisen under evolutionary selective pressure. It has been proposed that highly Na ϩ -selective Na V channels require pore asymmetry, which is achieved by the concatenated 4 ϫ 6TM structure (11,12). However, tetramers of identical 6TM NaChBac are equally Na ϩ -selective as Na V s (7). Therefore, it seems more likely that the concatenated 4 ϫ 6TM structure of vertebrate Na V s provided an evolutionary advantage by increasing the speed of activation, inactivation, or recovery from inactivation. To begin to address these questions, we searched for other bacterial ion channel subunits that might reveal the range of the gating speeds provided by the simplest tetrameric Na ϩ -selective channels. This information will also lay the groundwork for understanding their function in prokaryotes.
After mixing 1 M each of 2F, 3F, 1R, and 2R as a template, the first PCR was conducted using 1F and 3R as primers (10 M each). Similarly, 4R, 5F, 5R, and 6F were mixed, and PCR was conducted using 4F and 6R as primers. These PCR products were excised from agarose gels. Purified fragments were mixed and used as templates for further PCR, using 1F and 6R as primers. The amplified fragment was digested with NcoI and XhoI and purified from an agarose gel. The PCR product was then cloned into pTrcHis2B (Invitrogen). This synthesized DNA contained additional methionine and valine codons prior to the first methionine as required by the cloning strategy. We also intentionally introduced 10 silent mutations (45G3 T, 321G3 T, 330G3 T, 564C3 G, 606C3 G, 624A3 C, 775A3 C, 777G3 T, and 783G3 T) into the synthetic DNA. An additional silent mutation (495T3 C) occurred during the PCR process. Finally, this plasmid DNA was digested with SalI and XhoI and then self-ligated after the linker sequence was deleted. The resultant expression clone was used to transform Escherichia coli BL-21 (Stratagene).
Na V PZ was cloned into a modified pTracer-CMV2 vector (Invitrogen) containing enhanced green fluorescent protein for expression in mammalian cells. Briefly, DNA was amplified by PCR from the E. coli expression clone using the following primers: 5Ј-AATGGATCCATGAG-CCTGCGCGCGCGC-3Ј (containing a BamHI site) and 5Ј-ATTGAATT-CTCAGACACGCCCACGGCCGCC-3Ј (containing an EcoRI site). The PCR product was then cloned into the modified pTracer-CMV2 plasmid (Invitrogen) between the single restriction sites for BamHI and EcoRI. All clones were confirmed by DNA sequencing.
Cloning of Na V SP-The NaChBac DNA sequence (NC_002570) was used for a TBLASTX search against the Microbial Genomic data base at NCBI. A sample of Silicibacter pomeroyi was obtained from the American Type Culture Collection (catalog no. 700808). S. pomeroyi genomic DNA was collected by standard procedures (13). The genomic sequence homologous to NaChBac was identified, and the preliminary sequence data was obtained from The Institute for Genomic Research (TIGR) site on the World Wide Web at www.tigr.org. Na V SP was cloned into a pTrcHis2A (Invitrogen) for expression in E. coli. Briefly, DNA was amplified by PCR from a S. pomeroyi genomic DNA using the following primers: 5Ј-AATCCATGGTAATGCAAAGAA-TGCAGGCCTTT-3Ј (containing an NcoI site) and 5Ј-ATTCTCGAGAG-AACCGCGTGGCACCAGCTTTTTGGTTTCACCAAG-3Ј (containing a thrombin recognition site and an XhoI site). The PCR product was cloned into the pTrcHis2A plasmid using NcoI and XhoI sites. This cloned DNA contains additional methionine and valine codons prior to the first methionine as required by the cloning strategy. The resulting expression clone was used to transform E. coli BL-21.
Na V SP was cloned into a modified pTracer-CMV2 vector containing enhanced green fluorescent protein for expression in mammalian cells. Briefly, DNA was amplified by PCR from a S. pomeroyi genomic DNA using the following primers: 5Ј-AATGGATCCATGCAAAGAATGCAG-GCCTTT-3Ј (containing a BamHI site) and 5Ј-ATTGAATTCTCACTTT-TTGGTTTCACCAAG-3Ј (containing an EcoRI site). The PCR product was cloned into the modified pTracer-CMV2 plasmid between the single restriction sites for BamHI and EcoRI. All clones were confirmed by DNA sequencing. Information related to the cloning of other NaChBac homologs is provided as supplementary data.
Expression and Purification of Recombinant Proteins-100 -300 ml of LB medium containing ampicillin (50 g/ml) was inoculated from glycerol stocks and grown overnight at 30°C. 40 ml of culture medium was inoculated into 2 liters of Terrific Broth medium (Invitrogen) containing ampicillin and grown at 37°C to A 600 ϭ 1.2. Cells were induced with 1 mM 1-␤-D-thiogalactopyranoside and grown at 37°C for 3h. Cells were then suspended in PBS buffer (pH 8.0) containing protease inhibitors (Protease Inhibitor Mixture; Sigma) and lysed by sonication. The carboxyl-terminal histidine-tagged protein was extracted by homogenization and solubilization in 15 mM n-undecyl-␤-D-thiomaltopyranoside (Anatrace). Following centrifugation, the supernatant was loaded onto a Talon Co 2ϩ affinity column (Clontech). Resin was washed with 20 mM imidazole, and the protein was then eluted in the presence of 400 mM imidazole. Purified protein was resolved by 4 -12% SDS-PAGE (Invitrogen) and stained with Coomassie Blue. Molecular weight marker was purchased from Invitrogen (BenchMark TM Prestained Protein Ladder).
Mammalian Electrophysiology-Na V PZ and Na V SP as well as other NaChBac homologues were subcloned into an enhanced green fluorescence protein-containing pTracer-CMV2 vector (Invitrogen) for expression into CHO-K1 and HEK293T cells. CHO-K1 and HEK293T cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum at 37°C under 5% CO 2 . DNA was transfected using LipofectAMINE TM 2000 (Invitrogen) and plated onto coverslips, and recordings were made after 12 h (Na V PZ) or 48h (Na V SP), respectively. Unless otherwise stated, the pipette solution contained 147 mM Cs ϩ , 120 mM methane sulfonate, 8 mM NaCl, 10 mM EGTA, 2 mM Mg-ATP, and 20 mM HEPES (pH 7.4). Bath solution contained 140 mM NaCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 5 mM KCl, 20 mM HEPES (pH 7.4), and 10 mM glucose. All experiments were conducted at 22 Ϯ 2°C. Unless otherwise indicated, all chemicals were dissolved in water. Nifedipine (dissolved in Me 2 SO) was purchased from Sigma. As reported previously (7), unknown agents, presumably leached from the perfusion tubing, caused fast inactivation, and these perfusion systems were subsequently avoided.
We cloned all 11 sequences (see "Experimental Procedures") and studied them by expression in mammalian cell lines. As shown below, we were able to measure currents produced by two NaChBac homologs, Na V PZ ( Fig. 2A) from P. zeaxanthinifaciens (a zeaxanthin-producing marine bacteria (19)) and Na V SP ( Fig. 2A) from S. pomeroyi (a dimethylsulfoniopropionate-degrading marine bacteria (18)). Isolation and sequencing of the gene encoding Na V PZ revealed an open reading frame of 262 amino acids with a predicted molecular size of 29 kDa. Similarly, the Na V SP gene encoded an open reading frame of 258 amino acids with a predicted molecular mass of 29 kDa. Na V PZ and Na V SP share 39% identity (60% similarity) and 39% identity (59% similarity) with NaChBac, respectively (Fig. 2B). Notably, Na V PZ is 65% identical (77% similar) to Na V SP. Upon electrophoresis, both Na V PZ and Na V SP proteins migrated as a single band (ϳ31 kDa; Fig. 2C), almost identical to the predicted molecular sizes of the His-tagged constructs.
CHO-K1 or HEK293T cell lines were transfected with NaCh-Bac homologs (in pTracer), and whole-cell currents were recorded 12-48 h after transfection (see "Experimental Proce-dures"). Among 11 NaChBac homologs, only two (Na V PZ and Na V SP) produced detectable currents. Similar current are not present in nontransfected or mock-transfected cells (data not shown). Na V PZ-transfected cells exhibited large (up to 10,000 pA; 10 nA) voltage-activated inward currents (Fig. 3, A-C). Na V PZ-mediated current (I Na V PZ ) activated with a time constant ( activation ) of 21.5 Ϯ 1.3 ms at ϩ10 mV (n ϭ 19), significantly slower than both mammalian Na V channels ( activation Ͻ 2 ms) and I NaChBac ( activation Ͻ 13 ms). Inactivation of I Na V PZ was slow ( inactivation ϭ 102 Ϯ 4.2 ms at ϩ10 mV, n ϭ 19) compared with the typically fast inactivating Na V currents ( inactivation Ͻ 10 ms) but faster than I NaChBac ( inactivation Ͼ 160 ms).
Cation replacement by N-methyl-D-glucamine NMDG (bath) resulted in complete removal of voltage-dependent I Na V PZ inward current (Fig. 3C). Similarly, no significant inward current was seen in isotonic [Ca 2ϩ ] o (monovalent cations replaced with 105 mM Ca 2ϩ ) (Fig. 3, B and C). I Na V PZ reversed at ϩ75 mV (Fig.  3C), close to the Nernst potential of Na ϩ under our recording conditions (E Na ϭ ϩ72 mV). These results, together with the large leftward shift of the reversal by external Na ϩ removal (N-methyl-D-glucamine ϩ and isotonic Ca 2ϩ solution substitution), suggested that Na V PZ, like NaChBac, is a Na ϩ -selective channel. We assumed that the outward currents in 0 mM [Na ϩ ] o (Fig. 3B) were carried by internal Na ϩ (8 mM [Na ϩ ] i ). Due to the very negative E rev (Fig. 3C), sizable outward currents were observed at most voltages tested (Fig. 3, B and C).
We evaluated the voltage-dependent activation of I Na V PZ by measuring deactivation tail currents (Fig. 4A). A Boltzmann fit of the averaged activation curve yielded a V 1/2 of Ϫ9.5 Ϯ 0.8 mV (n ϭ 9) and slope factor () of 10.7 Ϯ 0.7 mV per e-fold change in current (Fig. 4C). Steady-state inactivation of the channel was determined by sequential depolarization to test voltages followed by voltage clamp to the peak of activation at ϩ10 mV (Fig. 4B). Steady-state inactivation was a steep function of voltage, with 50% inactivation at Ϫ35 Ϯ 0.4 mV (n ϭ 10) and slope factor () of 6.3 Ϯ 0.3 mV/e-fold (Fig. 4C). We investigated the time course of I Na V PZ inactivation at Ϫ30 mV, where activation was minimal. The degree and speed of inactivation was strongly dependent on the duration of the inactivating prepulse (Ϫ30 mV; ϭ 2123 Ϯ 434 ms; n ϭ 6; Fig. 4, D and E). I Na V PZ recovered slowly with time constant, ϭ 839 Ϯ 90 ms (n ϭ 7, HP ϭ Ϫ90 mV, Fig. 4, F and G).
Na V SP-transfected cells also yielded voltage-activated inward currents (Fig. 5A), peaking at ϳϩ30 mV. The E rev of Na V SP-mediated current (I Na V SP ) was ϩ76 mV. Ion substitution (Ca 2ϩ replacement) experiments confirmed that Na V SP, like Na V PZ, was also a Na ϩ -selective channel (data not shown). I Na V SP activated and inactivated significantly faster than I Na V PZ and I NaChBac ( activation ϭ 3.4 Ϯ 0.3 ms at ϩ30 mV, n ϭ 17; inactivation ϭ 35 Ϯ 1.5 ms at ϩ30 mV, n ϭ 17) but still severalfold slower than Na V currents. The Boltzmann fit activation curve yielded a V of ϩ21 Ϯ 0.4 mV and of 11.8 Ϯ 0.4 mV/e-fold change (n ϭ 28; Fig. 5E). Steady state inactivation was strongly dependent on the voltage ( ϭ 10.3 Ϯ 0.5 mV/efold), with half-inactivation at Ϫ22 Ϯ 0.8 mV (n ϭ 11; Fig. 5E).

FIG. 2.
Putative primary structure of Na V PZ and Na V SP. A, deduced membrane topology of Na V PZ (left) and Na V SP (right). B, alignment of the deduced amino acid sequence of Na V PZ, Na V SP, and NaChBac. The putative 6TM domains (S1-S6) and pore region (P) are indicated. C, Coomassie-stained SDS-PAGE of Co 2ϩ affinity chromatography purification of Na V PZ and Na V SP, indicating that both are ϳ29-kDa monomers. Na V PZ and Na V SP were sensitive to high concentrations of nifedipine (30 M; data not shown).

DISCUSSION
Expression of bacterial genes in systems where the protein can be studied (mammalian cells for patch clamp) is crucial to interpreting and extending static structural data through structure-function studies. Such functional expression is also important to understanding their native roles in bacteria. However, successful functional expression of bacterial proteins in mammalian cells is rare. Here, we identified 11 putative 6TM Na ϩ prokaryotic channel subunit genes and were able to functionally express 2 of the 11 in mammalian cells, where their electrophysiological properties could be studied. Both channels were Na ϩ -selective and activated by voltage. One conclusion based on comparison of bacterial channels and Na V s is that the major evolutionary pressure for gene duplication and concatenation of subunits was to increase the speed of channel gating. Given the high selectivity of presumed homomeric bacterial Na ϩ channels, the case for pore asymmetry as a means to increase Na ϩ selectivity seems a less likely scenario.
Little is known about the molecular determinants that control mammalian Na V activation rates. Mammalian Na V channels activate and inactivate within a few milliseconds (Ͻ10 ms), roughly 10 -100 times faster than NaChBac, the only bacterial voltage-gated channel functionally expressed up to now. Interestingly, Na V SP activation is ϳ4-fold faster than NaChBac, whereas Na V PZ activation is ϳ2 times slower than NaChBac, despite 77% sequence homology between Na V SP and Na V PZ. In Na V SP, Na V PZ, and NaChBac, the S4 domain and short S3-S4 linker are highly conserved, suggesting that the structural determinants for the kinetics differences are located elsewhere. Notably, there are several Na V SP-specific residues in the putative pore-forming domains (Gly 143 in S5, Ile 172 in the pore loop, Val 189 in the linker between the P loop and S6, and Met 202 in S6). These residues may contribute to the relatively fast activation kinetics of Na V SP.
Na ϩ channel inactivation mechanisms are better understood than those of activation. Interdomain linkers mediate fast inactivation in Na V s by "ball and chain" or N type inactivation (26), but these domains are obviously missing in tetramers of The cell was bathed in 2 mM Ca 2ϩ Tyrode's external solution (140 mM Na ϩ , 2 mM Ca 2ϩ , 5 mM K ϩ , 1 mM Mg 2ϩ , pipette; 147 mM Cs ϩ , 8 mM Na ϩ ; see "Experimental Procedures"). B, I Na V PZ in isotonic Ca 2ϩ (105 mM Ca 2ϩ , 0 mM Na ϩ ) external solution (ϩ20 mV, blue; ϩ60 mV, red). C, averaged (n ϭ 24) peak current-voltage (I/V) relation of Na V PZ in Tyrode's solution, normalized by cell capacitance (pF). E rev ϭ ϩ75 mV. Also shown are peak current I-V relations of I Na V PZ from a cell bathed in either isotonic 105 mM Ca 2ϩ (diamonds) or 150 mM N-methyl-D-glucamine ϩ (0 mM Ca 2ϩ (NMDG ϩ ), 0 mM Na ϩ ; triangles). E rev ϳϪ40 mV. V HP indicates holding potential, 15 s between pulses.
6TM bacterial channels. Additionally, the removal of segments within the N and C cytoplasmic domains of NaChBac (24) (unpublished data) does not substantially alter its inactivation rate. If the cytoplasmic domains do not participate in inactivation, we can then begin to look at other domains. Studies on 6TM HERG K ϩ channels indicated that the S5-P linker was crucial for its C-type inactivation, probably by providing allosteric coupling between its outer mouth and the voltage sensor (27). NaChBac inactivates with a time constant similar to Na V PZ, but the NaChBac S5-P linker (from Gln 167 to Ser 180 ) has low homology to Na V PZ. NaChBac and Na V PZ inactivates 5-fold more slowly than Na V SP. In the pore-S6 linker, Na V SP lacks the negatively charged glutamate present in both NaCh-Bac and Na V PZ. Future studies will focus on this and other sequence differences. However, the difficulty of obtaining functional expression of many mutants highlights the need for structural data. By exclusion of alternative mechanisms, we hypothesize that C-type inactivation, in which the Na ϩ pore is shut, is the more likely mechanism for Na V Bac channel inactivation.
Na ϩ channels have been proposed to play a central role in Na ϩ -dependent flagellar mobility in some prokaryotes. Marine Vibrio species utilize their Na ϩ -driven polar flagella for swimming (5,6), and in the alkaliphilic Bacillus species, [Na ϩ ] o determines the activity of the flagellar motor (4,28). Under alkaliphilic conditions, the H ϩ -motive force is weak due to the high intracellular pH (pH 8 -9) of these bacteria (4,29). Therefore, a Na ϩ cycle driven by the Na ϩ channel may have evolved to power the flagellar motor (4, 28, 30 -32). Interestingly, bacteria whose flagellar motors are powered by Na ϩ (as opposed to H ϩ ) express NaChBac homologs. In bacteria, the Na ϩ /H ϩ exchanger prevents cytotoxic Na ϩ accumulation and also supports pH homeostasis at elevated pH (4,31,33). In low [Na ϩ ] o environments or in the absence of solutes to support Na ϩ uptake through Na ϩ -coupled solute transporters, the pH homeostasis function may rely on a Na ϩ channel (4,31,33). We propose that sustained voltage-gated Na ϩ channel opening is primarily responsible for this Na ϩ entry (33). It is possible that some mammalian Na ϩ channels play a role in Na ϩ or H ϩ homeostasis. Interestingly, the mammalian persistent and resurgent Na ϩ currents have similar kinetics to Na V Bac (34,35). These persistent Na ϩ currents may be mediated FIG. 4. I Na V PZ voltage-and time-dependent activation and inactivation. A, I Na V PZ deactivation tail currents. After prepulses of varying depolarization (from Ϫ40 to ϩ70 mV, increments ϭ ϩ10 mV), tail currents were measured upon return to the holding potential (V HP , Ϫ90 mV). B, I Na V PZ steady-state inactivation currents. After a 4-s prepulse, the currents inactivated to a steady-state level and were reactivated by a second 2-s depolarizing pulse. The intersweep interval was 16 s. C, normalized activation curve (n ϭ 9; ϮS.E.) and steady-state inactivation curve (n ϭ 10; ϮS.E.). D, time-dependent inactivation of I Na V PZ increased slowly at Ϫ30 mV. V HP ϭ Ϫ90 mV. The currents were elicited by depolarization to ϩ10 mV preceded by a prepulse (Ϫ30 mV) with a variable length of time (0 ms (black), 500 ms (magenta), 1500 ms (dark yellow), 4500 ms (blue), 9000 ms (red)). E, the average time course of I Na V PZ inactivation at Ϫ30 mV. The time constant of inactivation development is ϳ2 s (Ϫ30 mV). F, recovery from inactivation. The time interval between the test pulse (ϩ10 mV, 1000 s) and the inactivation pulse (ϩ10 mV, 1000 ms) was varied from 0.2 to 9.8 s. G, relative peak I Na V PZ recovered with an approximately exponential time course ( ϳ840 ms). The ratios between currents elicited by the two pulses (F) were used to construct the recovery curve (n ϭ 7; ϮS.E.). by subthreshold gating of fast Na V channels (36) or perhaps by Na V 1.8 and Na V 1.9.
NaChBac selectivity is converted from Na ϩ to Ca 2ϩ by replacing an amino acid adjacent to glutamatic acid in the putative pore domain by a negatively charged aspartate (from TLESWAS to TLEDWAS or TLDDWAD) (24). Interestingly, two bacterial strains (C. psychrerythraea and M. degradans) have a putative pore sequence (TFEDWTD) similar to that of the Ca 2ϩ -selective NaChBac mutant. We have not been able to functionally express these channel subunits in mammalian cells, but one possibility is that these proteins form heteromeric channels with other related subunits in the same species. . I Na V SP activated above Ϫ20 mV and peaked at ϩ30 mV. E rev ϭ ϩ76 mV. C, I Na V SP activation currents. Tail currents were measured as described in the legend to Fig. 4A. D, I Na V SP steady-state inactivation currents in response to a protocol similar to that of Fig. 4B. E, normalized activation curve (n ϭ 28; ϮS.E.) and steady-state inactivation curve (n ϭ 11; ϮS.E.).