Comparative Study of the Gating Motif and C-type Inactivation in Prokaryotic Voltage-gated Sodium Channels*

Prokaryotic voltage-gated sodium channels (NaVs) are homotetramers and are thought to inactivate through a single mechanism, named C-type inactivation. Here we report the voltage dependence and inactivation rate of the NaChBac channel from Bacillus halodurans, the first identified prokaryotic NaV, as well as of three new homologues cloned from Bacillus licheniformis (NaVBacL), Shewanella putrefaciens (NaVSheP), and Roseobacter denitrificans (NaVRosD). We found that, although activated by a lower membrane potential, NaVBacL inactivates as slowly as NaChBac. NaVSheP and NaVRosD inactivate faster than NaChBac. Mutational analysis of helix S6 showed that residues corresponding to the “glycine hinge” and “PXP motif” in voltage-gated potassium channels are not obligatory for channel gating in these prokaryotic NaVs, but mutations in the regions changed the inactivation rates. Mutation of the region corresponding to the glycine hinge in NaVBacL (A214G), NaVSheP (A216G), and NaChBac (G219A) accelerated inactivation in these channels, whereas mutation of glycine to alanine in the lower part of helix S6 in NaChBac (G229A), NaVBacL (G224A), and NaVRosD (G217A) reduced the inactivation rate. These results imply that activation gating in prokaryotic NaVs does not require gating motifs and that the residues of helix S6 affect C-type inactivation rates in these channels.

iary subunits to form the native channel (2,3). The only structural information on Na V s available to date is a density map of the Na V from the electric organ of the electric eel determined by cryoelectron microscopy (4). Due to its limited resolution of 19 Å, the density map did not provide insights into the gating or sodium selectivity.
The first prokaryotic Na V , NaChBac, was cloned from Bacillus halodurans (5). Subsequently, three more prokaryotic sodium channels were cloned and characterized (6,7). All studied prokaryotic Na V s form homotetramers with a structure thought to be similar to that of some potassium channels with known structures (8 -10). Furthermore, because the proteins could be expressed in large amounts in Escherichia coli and purified by metal chelate affinity chromatography (5,7,11), they are promising candidates for high resolution structure determination and structure-function analyses.
The physiological role of prokaryotic Na V s may be related to pH homeostasis, motility, and chemotaxis (6,12). Searching bacterial genomic data bases, we found 26 sequences of putative NaChBac homologues from bacteria living in various environments. We were able to clone the putative Na V genes from three of these bacteria, Bacillus licheniformis, Shewanella putrefaciens, and Roseobacter denitrificans, and named them Na V BacL, Na V SheP, and Na V RosD, respectively. These homologues showed unique channel properties optimized for the specific environments in which the bacteria live.
We found an unexpected feature in the primary structure of Na V SheP, namely a lack of glycine residues in helix S6. Crystal structures of potassium channels revealed that glycine and proline residues in helix S6 that are part of the "glycine hinge" and "PXP motif," respectively, create kinks in the open state (13)(14)(15). It has therefore been thought that these residues are critical for activation gating in tetrameric cation channels (15)(16)(17)(18)(19). Although Na V SheP contains no glycine residues in helix S6, it still shows typical inward currents. To better understand their role, we mutated the glycine residues in helix S6 of Na V BacL, Na V RosD, and NaChBac to alanine. All mutants retained inward sodium currents. These results suggest that the previously proposed gating motifs are not obligatory for activation of prokaryotic Na V s.
The glycine mutations had no effect on activation but affected inactivation of the channels. Generally, voltage-gated cation channels are immediately inactivated after activation. Tetrameric cation channels have mainly two inactivation mechanisms, N-type and C-type inactivation (20 -25 or fast inactivation is thought to be mediated by the interdomain linkers of Na V s or the N termini of voltage-gated potassium channels, a mechanism described by the "ball and chain" model (20 -22). Although C-type inactivation is common in tetrameric cation channels, its molecular mechanism remains unclear. It has been proposed that C-type inactivation may be related to a collapse of the selectivity filter (23)(24)(25). Prokaryotic Na V s lack an obvious cytoplasmic N-type inactivation peptide, and inactivation may thus only occur through the C-type inactivation mechanism (25,26). Furthermore, comparative studies and mutational analysis of these new Na V homologues provide evidence that the rate of C-type inactivation is affected by the residues of helix S6.  16 S rRNA sequence analysis, were obtained from the National Institute of Technology and Evaluation. DNAs of NaChBac homologues were directly cloned by PCR from colonies of bacteria using PrimeSTAR GXL DNA Polymerase (Takara Bio.). The following primers were used for cloning: 5Ј-CGCGGA-TCCATGAACACTCATCAAAATCGA-3Ј and 5Ј-CGCGGA-TCCATCTTTCTTTTCATTTTGGGT-3Ј for B. licheniformis; 5Ј-CGCGGATCCATGAGTACATCTTTACTT-3Ј and 5Ј-CGCGGATCCGTCAGAACCAATGTTTCC-3Ј for S. putrefaciens; 5Ј-CGCGGATCCATGGGTGTAAGAGAGCAA-3Ј and 5Ј-CGCGGATCCGGACCGGGGCCTGTCTTT-3Ј for R. denitrificans; 5Ј-CGCGGATCCATGAATTTGCGCGCCA-GAACA-3Ј and 5Ј-CGCGGATCCCTTGCCCGCCTGATCC-AGTCG-3Ј for L. vestfoldensis; 5Ј-CGCGGATCCATGACAG-TGCGGGAAATGGTA-3Ј and 5Ј-CGCGGATCCGCCGCG-CGTCTCGTCGCCCGC-3Ј for R. sphaeroides; and 5Ј-CGCG-GATCCATGAGCGACGCGTCGGCC-3Ј and 5Ј-CGCGGA-TCCCTCGGAATTGTCCGCCCG-3Ј for O. alexandrii.

Cloning of NaChBacs Homologues and Site-directed
PCR products were cloned into a modified pET21b plasmid (Novagen) using BamHI restriction sites. These cloned DNAs contained additional methionine, glycine, and serine codons FIGURE 1. Phylogenetic tree of bacterial NaChBac homologues. A phylogenetic tree shows the bacterial species that express NaChBac homologues and their GenBank TM accession numbers. The program ClustalW was used to calculate a multiple sequence alignment based on conserved sequences of the NaChBac homologues. The phylogenetic tree was generated using the program PROTDIST, part of the PHYLIP package (Phylogeny Inference Package; available on the World Wide Web). The branch lengths are proportional to the sequence divergence, with the scale bar corresponding to 0.1 substitution per amino acid site. Red underlined, homologues that were newly cloned and functionally characterized in this study; gray underlined, homologues that could not be cloned in this study; blue, homologues that were functionally characterized previously; gray, homologues that were previously cloned but did not show detectable channel activity.
before the first methionine and a thrombin cleavage site before the C-terminal His tag. For expression in mammalian cells, three NaChBac homologues (Na V BacL from B. licheniformis, Na V SheP from S. putrefaciens, and Na V RosD from R. denitrificans) were subcloned into the pCI vector (Promega) using the NheI and KpnI sites. Site-directed mutagenesis was accomplished by PCR of the full-length pCI plasmid containing the Na V gene using high fidelity Pfu DNA polymerase (method by Stratagene). All clones were confirmed by DNA sequencing.
Expression and Purification of Recombinant Prokaryotic Na V s-Proteins were expressed in E. coli BL21 ⌬star TM (DE3) (Invitrogen). 100 ml of LB medium (Invitrogen) containing 100 g/ml ampicillin (Wako) was inoculated with 10 l of glycerol stock and incubated overnight at 37°C. 12.5 ml of the saturated culture medium was then used to inoculate 1.25 liters of LB medium containing 40 g/ml ampicillin, and cells were grown at 37°C to an A 600 of 0.8. Cells were induced with 0.5 mM isopropyl 1-thio-␤-Dgalactopyranoside (Wako) and grown for 16 h at 18°C. Cells were suspended in TBS buffer (20 mM Tris-HCl, pH 7.4, 300 mM NaCl) and lysed using a French press (SLM AMINCO) at 12,000 p.s.i. Membranes were collected by centrifugation (100,000 ϫ g, 1 h, 4°C) and solubilized by homogenization in TBS buffer containing 40 mM CYMAL-5 (Anatrace). After centrifugation (40,000 ϫ g, 30 min, 4°C), the supernatant was loaded onto a HIS-Select cobalt affinity gel column (Sigma). The column was washed with 10 mM imidazole in TBS buffer containing 8 mM CYMAL-5, and the protein was eluted with 300 mM imidazole. Purified proteins were resolved on 10 -20% SDS-PAGE gradient gels (Wako) and stained with Coomassie Blue. The Bench-Mark TM prestained protein ladder (Invitrogen) was used as molecular weight markers.
Mass Spectrometry-Mass spectrometry was carried out with a MALDI-TOF instrument (Voyager-DE, Applied Biosystems), using ␣-cyanohydroxycinnamic acid (Sigma) as matrix. Protein samples were mixed with the same volume of 90% acetonitrile (Wako) and 0.2% trifluoroacetic acid (Sigma), loaded onto a sample plate, dried, and washed with deionized water. Finally, 1% ␣-cyanohydroxycinnamic acid dissolved in 50% acetonitrile and 0.1% trifluoroacetic acid was FIGURE 2. Primary structures and expression of NaChBac homologues. A, alignment of the deduced amino acid sequences of Na V BacL, Na V SheP, Na V RosD, and NaChBac. The six putative transmembrane domains (S1-S6) are underlined in light gray, and the putative pore region is underlined in dark gray. Conserved arginines in helix S4 implicated in voltage sensing and residues constituting the putative sodium selectivity filter are shown as white letters in black boxes. The arrowheads indicate cleavage sites in Na V SheP identified by MALDI mass spectrometry. B, Coomassie-stained SDS-PAGE of Na V BacL, Na V SheP, and Na V RosD purified by Co 2ϩ affinity chromatography. C, MALDI-TOF peptide mass spectrum of purified Na V SheP. The x axis represents the massto-charge ratio (m/z), and the y axis represents relative abundance. The arrow indicates the full-length NaChBac homologues, and the arrowheads indicate fragments of Na V SheP. loaded onto the dried samples. Mass spectra were collected using the delayed extraction mode.
Mammalian Electrophysiology-Chinese hamster ovary (CHO)-K1 or human embryonic kidney (HEK) 293 cells were grown in Dulbecco's modified Eagle's medium (Sigma) complemented with 10% fetal bovine serum (BioWhittaker), 100 units/ml penicillin (Invitrogen), and 100 g/ml streptomycin (Invitrogen) at 37°C under 5% CO 2 . Cells were transfected with Na V and pEGFP DNA using the calcium phosphate transfection kit (Invitrogen) and plated onto coverslips. Currents were recorded 12 h (Na V BacL and Na V SheP) or 36 h (Na V RosD and NaChBac) after transfection. The pipette solution contained 105 mM CsF, 35 mM NaCl, 10 mM EGTA, and 10 mM HEPES (pH 7.4). The bath solution contained 150 mM NaCl, 1.5 mM CaCl 2 , 1 mM MgCl 2 , 2 mM KCl, 10 mM HEPES (pH 7.4), and 10 mM glucose. For measurements of the voltage dependence of Na V BacL and Na V SheP, 150 mM NaCl in the bath solution was replaced by 75 mM NaCl and 75 mM N-methyl-D-glucamine. All experiments were conducted at 25 Ϯ 2°C. Nifedipine (Sigma) was dissolved in DMSO (Wako). The final concentration of the solvent was less than 1%. Whole-cell patch clamp recordings of wild-type NaChBac confirmed that 1% DMSO did not affect the channel activity. All chemicals were dissolved in water. All other compounds were applied to the cultured cells using a perfusion system. For pharmacological measurements, cells were perfused with bath solution containing the inhibitors for 5 min prior to recording currents. All results are presented as mean Ϯ S.E. A simple comparison between Na V BacL wild type and G201A was made by Student's t test. To assess statistical differences in time constants between wild type and mutant proteins, multiple comparisons were made by one-way analysis of variance followed by the Tukey-Kramer multiple comparisons test.

RESULTS
We used the full NaChBac sequence for a standard BLASTP search against the GenBank TM data base of prokaryotic genomic sequences. The BLASTP search yielded 34 sequences that contained both a "TLESW" sequence, the putative selectivity filter of prokaryotic sodium channels (5-7), and positively charged amino acids at every third position in helix S4, the hallmark of voltage sensors (3, 5) (Fig. 1). The channel activity of four of the identified homologues (NaChBac as NP242367, Na V BP as AAR21291, Na V PZ as CAD24429, and Na V SP as AY392137) had already been established (5-7), whereas four other homologues (NP 693313 of Oceanobacillus iheyensis HTE831, ZP 00067595 of Microbulbifer degradans 2-40, Ay392141 of Colwellia psychrerythraea 34H, and ZP 00043768 of Magnetococcus sp. MC-1) showed no channel activity in previous studies (7). Of the bacteria containing the remaining 26 homologues, six of them were commercially available: B. licheniformis, isolated from soil and plant material (27), and L. vestfoldensis (28), R. sphaeroides (29), S. putrefaciens (30), R. denitrificans (31), and Oceanicaulis sp. (32) isolated from sea water (28 -33). Using the colony-direct PCR method, we were able to clone the putative Na V genes from three of these six bacteria, namely from B. licheniformis, S. putrefaciens, and R. denitrificans.
Sequencing of the Na V BacL gene from B. licheniformis revealed an open reading frame of 276 amino acids with a predicted molecular size of 31.8 kDa, and sequencing of the Na V RosD gene from R. denitrificans showed an open reading frame of 259 amino acids (29.2 kDa). Both sequences were identical with the deposited sequences. Our sequence of the Na V SheP gene from S. putrefaciens, encoding an open reading frame of 295 amino acids (33.1 kDa), differed, however, from the sequence deposited in the Gen-Bank TM data base in several places (T31A, A54T, V62L, and N88S) ( Fig. 2A). Na V BacL, Na V SheP, and Na V RosD share with NaChBac sequence identities (sequence similarities) of 49% (57%), 32% (44%), and 37% (48%), respectively, and hydropathy plots predict a sixtransmembrane architecture for all three proteins ( Fig. 2A).
C-terminally His-tagged constructs of all three Na V s expressed well in E. coli and could be purified by metal chelate affinity chromatography (Fig. 2B). Na V BacL and Na V RosD migrated as a single band on SDS-polyacrylamide gels (ϳ34 and ϳ25 kDa, respectively), whereas Na V SheP migrated mainly as a ϳ35 kDa band but also showed additional smaller bands. The smaller bands could also be detected on Western blots probed with anti-His antibodies (data not shown). Mass spectra confirmed that all main bands represented the fulllength proteins and identified the molecular masses of the smaller bands of Na V SheP as 22,662 and 23,053 Da (Fig. 2C). The additional bands thus represent C-terminal fragments of Na V SheP that were generated by cleavage of the carboxyl group of residues Arg 110 , the first arginine of helix S4, and Ala 107 ( Fig. 2A).
We transfected mammalian cell lines (CHO-K1 or HEK 293) with the genes encoding Na V BacL, Na V SheP, and Na V RosD and measured the resulting currents (Fig. 3). Although no currents were observed in non-transfected or mock-transfected cells (data not shown), currents could be recorded from cells transfected with the Na V genes, and all three channels required at least 1 s for 90% recovery from inactivation. Cells transfected with each of the three Na V s exhibited large voltage-activated inward currents of up to 10 nA (Fig. 3, A-C), peaking at Ϫ40, Ϫ50, and Ϫ10 mV for Na V BacL, Na V SheP, and Na V RosD, respectively (Fig. 3, D-F). The reversal potential of Na V RosD-mediated current (I NaVRosD ) was FIGURE 4. Voltage-dependent activation and inactivation of Na V BacL, Na V SheP, and Na V RosD. A, I NaVBacL deactivation tail currents. After prepulses of varying depolarization (from Ϫ120 to ϩ30 mV, increments of ϩ10 mV), tail currents were measured at Ϫ120 mV. B, I NaVBacL steady-state inactivation currents. After a 2-s prepulse, the channels inactivated to a steady-state level and were reactivated by a second depolarizing pulse (Ϫ20 mV). C, I NaVBacL -normalized activation curve (open circle, n ϭ 9; ϮS.E.) and steady-state inactivation curve (closed circle, n ϭ 9; ϮS.E.). D, I NaVSheP deactivation tail currents. After prepulses of varying depolarization (from Ϫ120 to ϩ30 mV, increments of ϩ10 mV), tail currents were measured at Ϫ120 mV. E, I NaVSheP steady-state inactivation currents. After a 500-ms prepulse, the channels inactivated to a steady-state level and were reactivated by a second depolarizing pulse (Ϫ20 mV). F, I NaVSheP -normalized activation curve (open circle, n ϭ 6; ϮS.E.) and steady-state inactivation curve (closed circle, n ϭ 6; ϮS.E.). G, I NaVRosD deactivation tail currents. After prepulses of varying depolarization (from Ϫ100 to ϩ40 mV, increments of ϩ10 mV), tail currents were measured at Ϫ120 mV. H, I NaVRosD steady-state inactivation currents. After a 500-ms prepulse, the channels inactivated to a steadystate level and were reactivated by a second depolarizing pulse (ϩ20 mV). I, I NaVRosD -normalized activation curve (open circle, n ϭ 6; ϮS.E.) and steady-state inactivation curve (closed circle, n ϭ 6; ϮS.E.). The voltages with half-inactivation/activation and the slope factors are indicated as V1 ⁄2 (mV) and (mV/e-fold), respectively in C, F, and I. The intersweep interval was 15 s in A, B, D, E, G, and H. Comparative Study of Prokaryotic Na V s FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 JOURNAL OF BIOLOGICAL CHEMISTRY 3689 ϩ42.7 mV in a bath solution containing 150 mM Na ϩ . The inward currents produced by Na V BacL and Na V SheP in a bath solution containing 150 mM Na ϩ were, however, peaking at a membrane potential that was too low to clamp the membrane potentials at low voltage. To measure the currents produced by Na V BacL and Na V SheP, we therefore reduced the NaCl concentration in the bath solution from 150 to 75 mM and added instead 75 mM N-methyl-D-glucamine. The reversal potential of I NaVBacL and I NaVSheP could then be determined as ϩ28.3 and ϩ33.0 mV, respectively. Ion substitution (Ca 2ϩ , K ϩ , and N-methyl-D-glucamine replacement) confirmed that all three proteins form Na ϩ -selective channels and that they are sensitive to high concentrations of nifedipine (30 M) (data not shown). Their pharmacological sensitivity closely resembles that of L-type Ca V s, as already reported for the previously characterized homologues (5-7).
We evaluated the voltage-dependent activation of Na V BacL by measuring deactivation tail currents (Fig. 4A). A Boltzmann fit of the averaged activation curve yielded an activating potential of 50% activation (V1 ⁄2 ) of Ϫ63.7 Ϯ 1.8 mV (n ϭ 9) and a slope factor () of 13.5 Ϯ 1.7 mV/e-fold change in current (Fig. 4C). The steady-state inactivation of the channel was determined by sequential depolarization to test voltages, followed by voltage clamp to the peak of activation at Ϫ20 mV (Fig. 4B). Steady-state inactivation was a steep function of voltage, with 50% inactivation at Ϫ83.2 Ϯ 0.7 mV (n ϭ 9) and ϭ 5.8 Ϯ 0.6 mV/e-fold change (Fig. 4C). Although Na V BacL had similar time constants of activation and inactivation as NaChBac (5), Na V BacL activated and inactivated at voltages about 40 mV lower than those of NaChBac (Table 1).
Boltzmann fits of their activation curves yielded V1 ⁄ 2 of Ϫ80.0 Ϯ 0.9 mV and of 7.4 Ϯ 0.8 mV/e-fold change for Na V SheP (n ϭ 6; Fig. 4, D and F) and V1 ⁄ 2 of Ϫ25.9 Ϯ 1.2 mV and of 9.9 Ϯ 1.1 mV/e-fold change for Na V RosD (n ϭ 6; Fig. 4, G and I). The steady-state inactivation of Na V SheP and Na V RosD were dependent on the voltage with half-inactivations at Ϫ116.5 Ϯ 0.9 mV ( ϭ 9.5 Ϯ 0.9 mV/e-fold) (n ϭ 6; Fig. 4, E and F) and Ϫ78.7 Ϯ 1.6 mV ( ϭ 10.7 Ϯ 1.6 mV/e-fold) (n ϭ 6; Fig.  4, H and I), respectively. Na V SheP activated and inactivated at the lowest membrane potential of all NaChBac homologues studied to date. The voltage dependences of the three prokaryotic Na V s analyzed in this study are summarized in Table 1.
Previous studies indicated that a glycine residue in the middle of helix S6, named the "glycine hinge," is critical for gating of tetrameric cation channels (15)(16)(17). A glycine residue corresponding to the glycine hinge is present in NaChBac (Gly 219 ) but not in Na V BacL (Ala 214 ), Na V SheP (Ala 216 ), and Na V RosD (Thr 207 ) (red arrow in Fig. 5A). Other glycine residues in Na V BacL (Gly 224 ), Na V RosD (Gly 217 ), and also NaChBac (Gly 229 ) are found in the lower part of helix S6. This part is near the cytoplasmic side and corresponds to the position of the PXP motif, a Pro-X-Pro or glycine motif that is important for gating in hERG (human ether-a-go-go related gene) and K V channels (18,19) (green arrow in Fig. 5A) and creates a kink in helix S6 of K V 1.2 (Fig. 5C). By contrast, Na V SheP contains no glycine residues in helix S6 (Fig. 5A).
To evaluate their role, we mutated the glycines in helix S6 of Na V BacL, Na V RosD, and NaChBac to alanine, a residue known to strongly stabilize ␣-helical secondary structure (35). All mutants retained the channel currents, but their time constants of inactivation changed significantly. On the other hand, their activation rates and also voltage dependences were less influenced by the mutations. Table 2 summarizes the activation and inactivation values of all analyzed mutant and wild-type channels as well as the statistical differences in inactivation between the wild-type channels and the various mutants. As a control, we made an alanine substitution of residue Gly 201 in Na V BacL, which is located in the pore-S6 loop (blue arrow in Fig. 5A). This mutation had no effect on the time constant of inactivation ( Table  2). With inactivation of 21.4 Ϯ 2.19 ms at ϩ10 mV, the NaChBac G219A mutant showed faster inactivation than wild-type NaCh-Bac (p Ͻ 0.01) (Fig. 5D). NaChBac G229A ( inactivation ϭ 230 Ϯ 33.8 ms at ϩ10 mV), Na V BacL G224A ( inactivation ϭ 270 Ϯ 53.3 ms at ϩ10 mV), and Na V RosD G217A ( inactivation ϭ 60.9 Ϯ 13.0 ms at ϩ20 mV) showed significantly slower inactivation than the corresponding wild-type channels (p Ͻ 0.01) (Fig. 5, D-F). Inactivation of the NaChBac G219A/G229A double mutant ( inactivation ϭ 156 Ϯ 16.2 ms at ϩ10 mV) showed no statistically relevant difference compared with the inactivation of the wildtype protein and was significantly slower than that of NaChBac G219A (Fig. 5D). All mutants lacking glycine residues in helix S6 still showed activation and inactivation. Na V BacL, Na V SheP, and Na V RosD do not have a glycine at the position that corresponds to the glycine hinge in helix S6. When the residues in this position were mutated to glycine, the mutants, with the exception of Na V RosD T207G (see below), also produced measurable inward currents (more than 1 nA) (Fig. 5, TABLE 1 V1 ⁄2 of inactivation and activation of prokaryotic Na V s V1 ⁄ 2 of inactivation is the membrane potential of 50% steady-state inactivation. V1 ⁄ 2 of activation is the potential of 50% activation. All results are expressed as mean Ϯ S.E. All experiments were performed at pH 7.4 except for the indicated experiment at pH 9.0.

V1 ⁄ 2 Note Inactivation Activation
Ϫ64 At pH 9.0 D-F). Furthermore, Na V BacL A214G ( inactivation ϭ 29.5 Ϯ 3.14 ms at ϩ10 mV) and Na V SheP A216G ( inactivation ϭ 5.35 Ϯ 0.72 ms at Ϫ10 mV) showed significantly faster inactivation than the respective wild-type channels (p Ͻ 0.01) (Fig. 5, E-G). Na V RosD T207G showed only a small and fast inactivated current (ϳ200 pA; inactivation Ͻ 2 ms). Because HEK293 or CHO-K1 cells express endogenous sodium channels that produce similar small currents (36), the current produced by the Na V RosD T207G mutant could not be distinguished from that produced by the endogenous sodium channels. Ala 226 of Na V SheP is the residue in the region in the lower part of helix S6 that cor-respondstothePXPmotifinvoltagedependent potassium channels (green arrow in Fig. 5A). The inactivation rate of Na V SheP A226G ( inactivation ϭ 11.6 Ϯ 0.50 ms at Ϫ10 mV) was slightly slower than that of the wild-type channel, but the difference is not statistically significant (Fig. 5F). Double mutation of two residues also retained the inward current of sodium ions. The inactivation rate of Na V BacL A214G/ G224A ( inactivation ϭ 54.1 Ϯ 4.51 ms at ϩ10 mV, n ϭ 9) was significantly faster than that of Na V BacL G224A (p Ͻ 0.01) and slower than that of Na V BacL A214G, although the difference was not statistically significant (Fig. 5E). Na V SheP A216G/A226G ( inactivation ϭ 7.61 Ϯ 0.65 ms at Ϫ10 mV) inactivated as fast as Na V SheP A216G. Na V RosDT207G/G217A( inactivation ϭ 20.93 Ϯ 3.90 ms at ϩ20 mV) produced a current less than 500 pA. This current could be distinguished from that of endogenous sodium channels and showed a faster inactivation than Na V RosD G217A (p Ͻ 0.01). In summary, mutations of the residues corresponding to the glycine hinge accelerated the inactivation rate, mutations of glycines in the lower part of helix S6 to alanine reduced the inactivation rate, and the A226G mutation in Na V SheP had no effect on the inactivation rate.

DISCUSSION
We cloned three new NaChBac homologues from B. licheniformis, S. putrefaciens, and R. denitrificans. The three channels differ in their kinetics and voltage dependences, which may reflect differences in the environment in which the respective bacteria live. The physiological role of the Na V homologue found in the alkaliphilic bacterium Bacillus pseudofirmus, Na V BP, has been well studied (6,12). Deletion of Na V BP resulted in impaired pH homeostasis and reversed pH chemotaxis (6). These results suggest that Na V BP functions in conjunction with the multiple resistance and pH adaptation Na ϩ /H ϩ antiporter in pH homeostasis, with the sodium-driven motor protein complex in motility (6), and with the methyl-accepting chemotaxis FIGURE 5. Electrophysiological analysis of the effects of mutations in helix S6. A, alignment of helix S6 and inner helix sequences of Na V BacL, Na V SheP, Na V RosD, Na V SP, Na V PZ, Na V BP, NaChBac, KcsA (potassium channel from Streptomyces lividans), K V 1.2 (potassium channel from Rattus norvegicus), and the four subdomains (I-IV) of Na V 1.2 (voltage-gated sodium channel from rat brain). The arrowheads indicate the positions of G201 in Na V BacL (blue), the glycine hinge (red), and the PXP motif (green); residues of these motifs are boxed in the corresponding colors. B and C, ribbon diagrams of KcsA (Protein Data Bank code 1r3j) (B) and K V 1.2 (Protein Data Bank code 2r9r) (C) with the residues and motifs described in A shown in the same colors. D, representative traces of currents produced by wild-type and mutant NaChBac channels activated by the following voltage protocol. After holding the potential at Ϫ120 mV, currents were measured at ϩ10 mV for 500 ms (wild type (WT) and G219A/G229A), 100 ms (G219A), and 1 s (G229A). E, representative traces of currents produced by wild-type and mutant Na V BacL channels activated by the following voltage protocol. After holding the potential at Ϫ160 mV, currents were measured at ϩ10 mV for 500 ms (wild type, A214G, and A214G/G224A) and 1 s (G224A). F, representative traces of currents produced by wild-type and mutant Na V SheP channels activated by the following voltage protocol. After holding the potential at Ϫ160 mV, currents were measured at Ϫ10 mV for 100 ms. G, representative traces of currents produced by wild-type and mutant Na V RosD channels activated by the following voltage protocol. After holding the potential at Ϫ140 mV, currents were measured at ϩ20 mV for 100 ms (wild type) and 300 ms (G217A and T207G/G217A). FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 proteins in chemotaxis (12). Multiple resistance and pH adaptation antiporters (37), sodium-driven motor protein complex (38), and methyl-accepting chemotaxis proteins (39) are found in many bacteria, and prokaryotic Na V s might thus play the same roles in other bacteria as Na V BP in B. pseudofirmus.

Comparative Study of Prokaryotic Na V s
Because bacteria live in environments that differ in pH, temperature, and ionic conditions, it seems reasonable that they express Na V s with channel properties that have adapted to their specific environments. First, activation and inactivation of prokaryotic Na V s cloned from marine bacteria (Na V SP, Na V SheP, and Na V RosD but not Na V PZ) were faster than those of homologues cloned from Bacillus species (NaChBac, Na V BP, and Na V BacL) ( Table 2). Marine bacteria live in an environment with a higher concentration of sodium ions than soil bacteria, such as Bacillus species. The fast inactivation of Na V s found in marine bacteria might thus be an adaptation that evolved to prevent the influx of excess sodium ions into the cytoplasm. Second, although Na V BacL and Na V BP, both expressed in Bacillus species isolated from soil, have similar kinetics of activation and inactivation, their voltage dependences differ from each other. At the neutral pH of 7.4, half of Na V BP is activated by a voltage of Ϫ35 mV and inactivated by a voltage of Ϫ57 mV (6) ( Table 1). These membrane potentials are 30 mV higher than those that activate and inactivate Na V BacL. At pH 9.0, the pH of the environment in which B. pseudofirmus lives, half of Na V BP is, however, activated by a voltage of Ϫ64 mV and inactivated by a voltage of Ϫ86 mV (6). Hence, at the pH of the environment in which the respective bacterium lives, the two channels have similar voltage dependences of activation and inactivation. This finding suggests that the channel characteristics of prokaryotic Na V s have been optimized for the specific environments of the bacteria.
The diversity in the channel properties of prokaryotic Na V s can help us to obtain a better understanding of the molecular mechanisms that underlie ion channel function. The crystal structure of the K V 1.2/K V 2.1 chimera potassium channel suggested that the S4-S5 linker helix pushes down on the lower part of helix S6 and thus keeps the ion pore closed in the resting state (13). A single molecule study demonstrated that this region of KcsA is rotated in the open state (40) and thus confirmed that this region works as an activation gate. If the pore domain were folded tightly, such as seen in crystal structures of potassium channels, the role of the glycine hinge and the PXP motif in channel gating might be to provide sufficient flexibility to allow the lower part of helix S6 to bend and move for activation. On the other hand, prokaryotic Na V mutants lacking glycine and proline residues in helix S6 and wild-type Na V SheP still show activation and inactivation ( Fig. 5 and Table 2). Neither motif can thus be essential for gating in prokaryotic Na V s. Because bending of helix S6 would be difficult without such gating motifs, it is likely that in these mutants and in wild-type Na V SheP, the entire helix S6 has to incline to open the activation gate. Therefore, it is likely that the pore domain of prokaryotic Na V s is folded more loosely to increase the mobility of helix S6.
Inactivation of prokaryotic Na V s is thought to occur through the C-type inactivation mechanism seen in other tetrameric cation channels. C-type inactivation has been explained by conformational changes around the selectivity filter of the channel (23)(24)(25). These conformational changes, which cause the selectivity filter to collapse, act as an inactivation gate. Therefore, the stability of the selectivity filter could affect the rate of C-type inactivation. In the case of KcsA, it has been proposed that the residues adjacent to Gly 99 , which is the residue corresponding to the glycine hinge, interact with residues forming the bottom of the selectivity filter, thus stabilizing its conductive conformation (41). This motif of K v 1.2 also locates beside the selectivity filter (Fig. 5, B and C). Based on the similarity of the primary sequences, the corresponding region in prokaryotic Na V s would also be near their selectivity filters. All mutations in this region in Na V BacL (A214G), Na V SheP (A216G), Na V RosD (T207G), and also NaChBac (G219A) accelerated the inactivation rate. Remarkably, the T207G mutation reduced ion currents in Na V RosD T207G and Na V RosD T207G/G217A. These mutations would be expected to make the selectivity filter less stable and easier to collapse. As a result, the mutants would inactivate faster than the respective wild-type channels.
The lower part of helix S6 is thought to act as activation gate (13,40). C-type inactivation always occurs after activation and thus appears to be coupled to activation. The coupling between the activation and inactivation gates is thought to be mediated by residues adjacent to the glycine hinge in helix S6 and the bottom of the selectivity filter (41). Our glycine-to-alanine mutations in the lower part of helix S6 reduced the inactivation rate. This result suggests that the glycine-to-alanine mutations, which stabilize the lower part of helix S6, weaken the coupling between the activation and inactivation gates and thus delay inactivation. Na V SheP is the only channel in which the alanineto-glycine mutation (A226G) showed no effect. It thus appears that although this region is important for the coupling in other channels, it may not have the same importance for coupling in Na V SheP. This notion is supported by the fact that Na V SheP lacks glycines in helix S6 and may thus use a coupling mechanism that in its details differs from that of other prokaryotic Na V s. In a previous report, a G219P mutation in NaChBac eliminated C-type inactivation (16). However, this glycine residue is not obligatory for activation gating. Therefore, the G219P mutation might force helix S6 to adopt an unusual bent conformation, in which helix S6 would not be able to mediate the coupling between the activation and inactivation gates.
Double mutations in the region corresponding to the glycine hinge and the lower part of helix S6 independently affected the inactivation rate in Na V BacL, Na V RosD, and NaChBac. It is possible that other regions in helix S6 are also involved in C-type inactivation. Nevertheless, the mutational results at these two regions may serve as useful footholds for the study of C-type inactivation. All mutations in this study had stronger effects on the inactivation rate than on the activation rate. The driving force for activation is depolarization, whereas C-type inactivation is due to the collapse of the selectivity filter, which depends on thermal instability. It is thus not unexpected that C-type inactivation is more sensitive to mutations than channel activation. Although these results provide first clues to the coupling of activation and inactivation in voltage-gated cation channels, a deep understanding of this phenomenon will require further mutational studies as well as the analysis of atomic structures.
Without knowing the three-dimensional structure of Na V channels, it is impossible to understand the sodium selectivity of these channels and how small changes in their primary sequences result in different voltage dependences and gating kinetics. A high resolution structure is particularly crucial for understanding the structural basis of sodium selectivity.
Because the three Na V s characterized in this study could be expressed at high levels and purified in large amounts, they are promising candidates for structural studies and thus increase the chances of obtaining a high resolution crystal structure.
Prokaryotic Na V s are the simplest voltage-gated cation channels known to date. They are composed of fewer than 300 amino acids, and differences in their channel characteristics are due to small changes in their primary sequences. As functional analyses proceed and structures become available, prokaryotic Na V s have the potential not only to reveal the basic molecular mechanisms of voltage-gated cation channels but also to provide insight into the molecular evolution of these channels.