Synchrotron radiation circular dichroism spectroscopy-defined structure of the C-terminal domain of NaChBac and its role in channel assembly

Abstract

Extramembranous domains play important roles in the structure and function of membrane proteins, contributing to protein stability, forming association domains, and binding ancillary subunits and ligands. However, these domains are generally flexible, making them difficult or unsuitable targets for obtaining high-resolution X-ray and NMR structural information. In this study we show that the highly sensitive method of synchrotron radiation circular dichroism (SRCD) spectroscopy can be used as a powerful tool to investigate the structure of the extramembranous C-terminal domain (CTD) of the prokaryotic voltage-gated sodium channel (Na_V) from Bacillus halodurans, NaChBac. Sequence analyses predict its CTD will consist of an unordered region followed by an α-helix, which has a propensity to form a multimeric coiled-coil motif, and which could form an association domain in the homotetrameric NaChBac channel. By creating a number of shortened constructs we have shown experimentally that the CTD does indeed contain a stretch of ∼20 α-helical residues preceded by a nonhelical region adjacent to the final transmembrane segment and that the efficiency of assembly of channels in the membrane progressively decreases as the CTD residues are removed. Analyses of the CTDs of 32 putative prokaryotic Na_V sequences suggest that a CTD helical bundle is a structural feature conserved throughout the bacterial sodium channel family.

Sodium, calcium, and potassium channels form a family of integral membrane proteins that selectively regulate ion conductance across an otherwise impermeable lipid membrane. In eukaryotic voltage-gated sodium and calcium channels the functional unit is formed by assembly of four homologous domains of a single polypeptide chain. Each domain is composed of six transmembrane (TM) helices, with the four N-terminal helices forming the voltage-sensing subdomain and the two C-terminal helices comprising the pore subdomain. In contrast, in both eukaryotic and prokaryotic potassium channels the functional unit is a homo-tetramer formed from identical monomers that each correspond to one of the sodium or calcium channel domains (1). The crystal structures of a number of potassium channels have been determined (2–4), which have confirmed the tetrameric nature of the channels and details of their transmembrane (TM) regions. The structures of the extramembranous C-terminal domains of most of these channels, however, were not defined in the crystal structures, either because they had been removed after assembly (KcsA) (2) in order to improve crystallization, or because they were disordered in the crystal (KvAP and MlotiK1) (3, 4).

The extramembranous termini of potassium channels appear to play important roles in channel assembly, primarily as tetramerization domains. Examples include the N-terminal T1 domain of Shaker channels (5) and C-terminal domains (CTDs) of the ether-a-go-go (EAG) (6) and the inwardly rectifying (K_IR) channels (7). In the KcsA potassium channel the CTD has been shown to be critical for expression, tetramerization, and stability of the closed channel (8, 9) as well as having a role in pH sensing (10, 11). The structure of the CTD of KcsA has recently been defined crystallographically (12). It forms a right-handed four-helix coiled-coil bundle of approximately 40 residues in length that projects 70 Å beyond the membrane interface, not interacting significantly with the TM regions of the protein, and is stabilized by a hydrophobic core in addition to a network of hydrogen bonds and salt-bridges between adjacent helices. The distal section of this CTD adopts a leucine zipper configuration.

Although potassium channels have been well characterized, much less is known about the structural features of voltage-gated sodium channels (Na_Vs). However, in 2001 Ren et al. (13) isolated NaChBac, a simple prokaryotic sodium channel from Bacillus halodurans that, like potassium channels, but unlike eukaryotic sodium channels, is composed of a single 6TM domain. Subsequently, many more close homologues from both gram-positive and gram-negative sources have been identified (14–16). In NaChBac, it has been shown that, as predicted, four monomers assemble to form an active channel (17). This has prompted the question as to whether a specific assembly domain exists in this channel. The ∼40 residues of the CTD of NaChBac at the end of the S6 TM helix represent a candidate region for such a domain, even though they exhibit negligible sequence identity (∼10% overall) with the CTD of KcsA. The aim of the present study was to determine if the CTD of NaChBac influences expression of this channel and its assembly into the cell membrane fraction and to elucidate the nature of the CTD. To achieve the latter, we utilized the emerging technique of synchrotron radiation circular dichroism (SRCD) spectroscopy (18), which provides the sensitivity and accuracy needed to examine the structures of a series of closely related truncated constructs of the channel. As a result, this study has also exemplified a new range of possible applications of SRCD spectroscopy in structural biology.

Results

Expression and Assembly.

The prokaryotic sodium channel NaChBac shows strong sequence homology to 6TM bacterial and eukaryotic potassium channels; in addition to the TM voltage sensor and pore regions, it also contains a CTD, consisting of ∼40 residues that extend beyond the TM domain. In order to identify the role (if any) these residues play in channel assembly and structure, a series of truncated channel constructs was produced, in which residues were removed consecutively from the C terminus (Fig. 1A). Western blots (Fig. 2 A and B) of the total cell lysates and the membrane fractions, respectively, were used to monitor the total amount of NaChBac produced and the amount of protein that assembles into the membrane as the CTD is truncated.

Fig. 1.

(A) Predicted secondary structure of NaChBac CTD based on its primary sequence using five algorithms. Code: (H) helix, (E) sheet, (N) nonhelical (random coil), or (-) not assigned. Predicted disordered regions are indicated by *. The coiled-coil prediction was done using sliding windows of either 14 or 21 residues, designated Lupas-14 and Lupas-21, respectively. The sites of truncation of the various constructs are indicated by vertical black arrows above the sequence, and the sites of the point mutations are indicated below the sequence by °. At the bottom of the figure is a hydrophobicity plot of the same region (residues 206 to 274), generated using the DNASTAR suite of programs. A window of 19 (solid line) was used for identifying TM segments, and a window of 9 (filled in shaded regions) was used for identifying leucine zipper features. (B) Homology model for the NaChBac pore (blue) and CTD (cyan) regions, starting at the N terminus of the S5 helix, with arrows showing the approximate positions of the constructs examined by SRCD spectroscopy. The green arrows indicate the incrementally removed residues were helical, and the red arrows indicate they were nonhelical.

Fig. 2.

Western blot analyses of the expression of the wild-type NaChBac and CTD truncation constructs. The same number of E. coli cells (as determined by OD₆₀₀) were used in each case for (A) whole cell extracts or (B) isolated membranes. Constructs are numbered according to the position of the stop codon along the NaChBac sequence. For example, in construct 231Δ the C-terminal residue is 230.

Firstly, as a positive control, total protein expression was shown to be virtually abolished when a stop codon was introduced at position 231 (construct designation 231Δ), a mutation designed to disrupt the S6 helix since the final ca. four residues of this helix are also removed in addition to ones in the CTD. Next, residues between 274 (full-length) and 254 (i.e., constructs 262Δ and 254Δ) were removed; in these cases the total amount of protein produced was essentially indistinguishable from the wild type. Even when the residues from 247 to 238 were progressively removed, there was only a very modest reduction in the total level of expression (Fig. 2B). When the membrane fractions were examined, the amounts of protein for constructs 262Δ, 254Δ, 247Δ, and 243Δ were found to be comparable to those in the total lysate; however, there was a progressive reduction in the amount of protein detected as residues from 242 to 235 were deleted.

Reductions in the total levels of protein may be a result of altered RNA stability, decreased protein synthesis, or altered protein stability. Because the 240Δ, 239Δ and 238Δ constructs appeared to be produced in roughly similar amounts as the wild-type channel when the total cell lysate was assayed, this suggests that neither altered RNA stability nor decreased protein synthesis are key factors in the reduction of the amount of protein in the membrane-associated fractions. In addition, the reduction in intensity of the Western blot bands of the shorter constructs is unlikely to be a result of altered accessibility of the His-tag used for binding of the antibody because the tag is located at the N terminus, distal to the C terminus; indeed removal of C terminus might even be expected to increase accessibility and hence staining. The reduced levels of assembled proteins detected in the membrane fraction are therefore likely to arise because removal of the CTD leads to a protein that is either unstable or one that is misfolded and located in the inclusion bodies. The latter is the most reasonable explanation because these proteins would be detectable in the whole cell lysate but not in the membrane fraction. This is similar to the pattern of expression observed for the KcsA potassium channel in which removal of the C-terminal-most 36 residues led to a limited reduction in expression of the channel, while removal of the next five residues from its extramembranous domain significantly reduced expression (8, 9).

These expression studies thus indicate that the CTD is important for expression of properly targeted and assembled NaChBac.

Structure.

Sequence-based secondary structure predictions of the S6 helix and the CTD of NaChBac using five algorithms (Fig. 1A) were used to produce a consensus secondary structure for NaChBac, and the DISOPRED algorithm was used to indicate regions with high propensities for forming natively disordered structures (Fig. 1A). Together they indicate that both the putative S6 TM region (as expected) and the last ca. 22 residues of the C terminus have a high propensity for forming helical structures, while the connecting 14 residues between S6 and the terminal 22 residues are predicted to be nonhelical (Fig. 1B).

Synchrotron radiation circular dichroism spectroscopy (Fig. 3) was used to experimentally determine the secondary structures of the various constructs and compare them with the predicted helical contents for these constructs (Table 1). The use of SRCD, a newly emerging tool for biophysical characterization, has been essential for these studies due to its high signal-to-noise levels and hence high sensitivity to very small structural changes (19, 20). All of the SRCD spectra of the constructs are very similar in shape, indicative of proteins with high helix contents, but differences in their spectra are reflective of their resulting from proteins with subtly different helical contents. The experimental error bars (one standard deviation) seen in the inset to Fig. 3 show that the high precision of the measurements makes it possible to distinguish with significance the spectra of these very closely related structures; furthermore, their reproducibility was demonstrated by comparisons of measurements on different samples of the same construct at two different beamlines (Soleil in France and ISA in Denmark), where the differences are ≤ the error bars for repeated scans of a single preparation (data not shown). The increased accuracy in secondary structure determinations using SRCD, when the low wavelength vacuum ultraviolet data (not measurable with conventional CD instruments) is included [it has been shown to produce correlation coefficients for helical components between crystal and SRCD-derived structures of 0.956 (21)], means that in this study it has been possible to identify secondary structural differences on the order of 1% (Table 1) associated with the very small differences in protein constructs.

Fig. 3.

Synchrotron radiation circular dichroism (SRCD) spectra of NaChBac protein constructs in the detergent Cymal-5. Comparison of the SRCD spectra of NaChBac (solid line) and CTD-truncated channels 254Δ (dotted line) and 239Δ (dashed line). The spectra of the 262Δ and 247Δ constructs were intermediate and not shown for clarity. Inset: expansion of the wavelength region between 185 nm and 200 nm showing error bars (at one standard deviation in the measurement).

Table 1.

Predicted and experimentally determined secondary structures of the different CTD constructs of NaChBac

Construct	No. residues removed	No. α-helical residues predicted missing	Helical secondary structure determined from SRCD spectra (%)	No.α-helical residues calculated missing from SRCD spectra
wild-type (274)	0	0	66.4	0
262Δ	13	11	64.8	12
254Δ	21	20	63.3	22
247Δ	28	21	64.8	22
239Δ	36	21	66.6	23

Calculations indicate (Table 1) that when the 36 amino acids between positions 274 and 239 were removed, the total number of α-helical residues lost was ∼22, a strong correspondence with the predicted value; furthermore essentially all of the α-helical residues lost specifically involved the C-terminal-most residues between 274 and 254 (Table 1). The SRCD studies of constructs missing the additional residues between 247 and 239 showed essentially no further loss of helical residues, also in accord with the sequence-based secondary structure predictions. This then suggests the distal C-terminal residues are helical, but that the residues closest to the TM segments are nonhelical (Fig. 1B).

Hydropathy plots of the CTD of NaChBac indicate the sequence corresponding to the ∼22 amino acid C-terminal helical region has a typical repeat pattern of hydrophobic and hydrophilic residues found in coiled-coil structures (Fig. 1A). Coiled-coil proteins have a characteristic seven residue repeat of (a.b.c.d.e.f.g)_n, with helix interactions driven by hydrophobic interactions of the residues at the a and d positions (22). In soluble coiled-coil proteins leucines occupy ca. 80% of all d positions and these structures are often referred to as leucine zippers (23). For the stretch of α-helical residues located at the C terminus of NaChBac, all three d positions are occupied by leucine residues, whereas the a positions are occupied by either leucine, isoleucine, or valine. Residues 249–274 plotted as a helical wheel (Fig. S1A) show a single face of the wheel is hydrophobic, with smaller or charged/polar amino acids on the remaining surfaces. When four helical wheels are placed into a parallel tetrameric arrangement (Fig. S1B), the resultant structure has a central hydrophobic core (Fig. S1C) and could possess intersubunit salt bridges formed between neighboring helices, e.g., K253-E255 and R260-D262, which could also contribute to the overall stability of the coiled-coil structure. The observation (12) of a similar type of structural element in the CTD of the full-length KcsA channel (albeit without the linking nonhelical region attaching it to S6) sets a precedent for such an ion channel tetramerization domain composed of a coiled-coil four-helix bundle.

Together these predictions and experimental observations suggest the CTD of NaChBac begins at the end of TM6 with a nonhelical region that links the TM pore with a C-terminal coiled-coil helical bundle.

Tetramerization.

Previous studies on the removal of the CTD from KcsA indicated that the protein migrates as a tetramer (as does the full-length KcsA) on SDS gels, suggesting that once the tetramer is assembled the channel is extremely stable in vitro, to both detergent and heat-induced dissociation (8, 9, 11). Unlike KcsA, the full-length NaChBac tetramer is not resistant to dissociation by SDS but instead runs as a monomer on SDS denaturing gels (Fig. 2); therefore, in order to determine the yield of oligomers in the membrane fraction, we monitored the gel-filtration (GF) profile of the isolated channels solubilized in the nondissociating detergent Cymal-5 (Fig. 4).

Fig. 4.

Gel-filtration characterization of wild-type NaChBac (solid line) and the 239Δ construct (dashed line) in the detergent Cymal-5. Both constructs run with molecular masses corresponding to tetramers, with the 239Δ mutant running slightly slower than the wild type due to the deletion of 36 residues in each of its monomers.

The GF data indicated that wild-type channel (plus the His-tag) had a peak elution volume corresponding to a calculated size of ∼210 kDa (protein plus detergent), as we had found earlier (17). This indicates that most of the wild-type channel is tetrameric, with no monomeric protein detected; a small population was observed to exist as aggregates, eluting at the void volume (Fig. 4). The shortest construct, 239Δ, eluted at a peak volume corresponding to a calculated molecular weight of ∼190 kDa. A shift in the peak elution volume for this construct is in agreement with it being smaller in size than the wild-type channel and consistent with the fact that 36 residues in each of its monomers (total ∼16 kDa) had been removed. However, the total yield of tetrameric channel isolated for this construct (despite starting with the same amount of membranes) was reduced by > 80% with respect to wild type, consistent with the relative amounts seen on the Western blots. Constructs 247Δ, 254Δ and 262Δ also eluted in comparably diminished yields but with retention volumes intermediate between that of the wild-type channel and the 239Δ construct (not shown for clarity).

Hence, it appears that while the yield of membrane-associated channels decreased with removal of the CTD, all channels isolated from the membrane fraction are tetrameric and remain so even upon solubilization. This suggests that the CTD influences the assembly but not necessarily the stability of the quaternary structure.

Homologues.

BLAST searches identified 32 putative prokaryotic Na_V sequences, a number of which had not previously been identified as members of this family. Sequence alignments of their CTDs (Fig. S2) suggest that all of these homologues possess a stretch of residues equivalent to at least four turns of a helix comprising at least two heptad repeats. This high level of conservation of the motif indicates that an intersubunit coiled-coil helical bundle may be an important structural feature in the bacterial sodium channel family.

Discussion

In the C-terminal truncated constructs, the levels of channel protein produced by the cell (as determined from whole cell lysate) did not appear to be significantly altered, while the amount of channel protein located in the membrane fraction was apparently reduced as the length of the CTD decreased. Once the CTD was fully deleted, no channels could be detected in the membrane fraction.

SRCD spectroscopy has shown that approximately 22 amino acids at the C-terminal end are helical in nature, but that there is also a novel nonhelical section between the end of the S6 helix and the helical C-terminal residues. Sequence predictions suggest the 22 helical residues may form an intersubunit coiled-coil motif. The CTD may contribute to the assembly of the tetramer. However, once a tetramer is formed, other stabilizing interactions such as van der Waals, hydrogen-bonding, electrostatic, and aromatic-aromatic interactions elsewhere in the molecule may also contribute to the stability of the tetrameric channel (24, 25). The nonhelical region linking S6 and the putative helical bundle appears to be particularly important for the assembly process. It contains a large number of negatively charged residues and a single positively charged residue (K237) (Fig. 1A). These polar residues, located in proximity to the TM-cytoplasmic interface, may form interactions with residues on the S5 helix or S4-S5 linker that are necessary for the correct folding of the channel. Alanine scanning mutagenesis studies (Fig. S3) has revealed that a number of the residues in this region, especially K237, are highly sensitive to mutation, with single changes greatly reducing or eliminating assembly of full-length channels in the membrane. Such a highly charged region may also be a common feature in the other prokaryotic sodium channels, because most of the homologue sequences display a high proportion of negative residues and a single positive residue at roughly the same positions with respect to the water-lipid interface in this region of the CTD (Fig. S2). We speculate that the role of the lysine residue, which when mutated essentially abolishes assembly in the membrane, may either be to form specific interactions with the head groups of the lipid molecules (26) or an intersubunit salt bridge or may simply be required to satisfy the positive-inside rule (27) for membrane protein insertion.

The presence of association domains may be common structural features for tetrameric members of the family of voltage-gated ion channels. Our sequence analyses suggest that similar structural features may be present in all (or most) prokaryotic sodium channels. Previous studies with K_IR and EAG potassium channels demonstrated the necessity of a tetramerization domain for channel expression (5–7, 9). In some cases the domain is located at the N terminus (5), and in some cases at the C terminus, of the TM regions (6, 7). In KcsA, the CTD is also ∼40 residues in length but is virtually entirely helical in the closed-state structure (12) and shares only ∼10% sequence identity with the NaChBac. Hence it appears that it is the presence of a domain that can aid association, rather than the details of the domain structure, that is the common feature among channels in this family. Tetramerization domains presumably aid channel assembly by recruiting nascent monomeric chains to an oligomeric ensemble, allowing physical convergence of the TM regions. Structural rearrangements of the TM domains may then be required to form the functional product once these regions are brought into close proximity.

In summary, we have used the highly sensitive method of SRCD in conjunction with bioinformatics predictions to propose a structure for the CTD of NaChBac at the residue level, consisting of a four-helix coiled-coil bundle with a leucine zipper configuration preceded by a highly charged nonhelical region adjacent to the S6 helix. Further, we suggest that the CTD plays an important role in channel expression and assembly and may be a common familial feature in all prokaryotic sodium channels.

This study has also demonstrated the utility of SRCD spectroscopy for characterizing proteins. In order to produce the level of detail described, high levels of precision, reproducibility, and accuracy not ordinarily achievable with conventional CD spectroscopy, but possible with SRCD, were required. The sodium channel has been ideal for exemplification of the new application because its CTD domain is dominated by helical features, which are very accurately determined by SRCD spectroscopy, and because its CTD is essentially a separate domain, so removals of individual amino acids do not alter the bulk of the protein remaining. This type of application of SRCD adds a new dimension to our ability to characterize protein structural domains.

Materials and Methods

Materials.

The cDNA for NaChBac was generously provided by David Clapham of Harvard Medical School. N-dodecyl-β-D maltopyranoside and 5-cyclohexyl-1-pentyl-β-D maltoside (Cymal-5) were obtained from Anatrace. All other reagents were purchased from Sigma, unless otherwise stated.

C-terminal truncated NaChBac constructs were generated using the QuikChange protocol from Stratagene to introduce stop codons into the full-length NaChBac cDNA (corresponding to 274 amino acids) that had previously been cloned into expression vector pET15b (17). Constructs are numbered according to the position of the stop codon along the NaChBac sequence. For example, construct 231Δ has the sequence of DNA encoding the amino acid at 231 replaced by a stop codon so that the C-terminal residue is now 230. All mutations were confirmed by sequencing on both strands.

Membrane Preparations and Western Blots.

Western blots were performed on either whole cell lysate or on isolated membrane preparations from cells. Single colonies of Escherichia coli C41(DE3) transformants harboring the pET15b plasmid with the NaChBac genes were grown in Luria broth supplemented with ampicillin (100 μg/ml) at 37 °C. Upon reaching an OD₆₀₀ of ca. 0.7, cultures were induced with 0.3 mM isopropyl-β-D thiogalactopyranoside at 37 °C for 1 h. An aliquot of cells equivalent to an OD₆₀₀ of 2.0 was then pelletted at 6,000  × g for 5 min. For samples of whole cell lysate the pellet was resuspended in NuPAGE lithium dodecyl sulfate (LDS) sample buffer. Alternatively, to isolate cell membranes, the cell pellet was resuspended in 150 mM NaCl, 20 mM Tris·HCl, pH 7.8 (TBS) and lysed by sonication. Lysed cells were spun at 6,000  × g for 5 min and the supernatant spun at 100,000  × g for a further 20 min. Membrane pellets were resuspended and solubilized by aspiration in TBS supplemented with NuPAGE LDS sample buffer.

Proteins from whole cell lysates or membrane fractions were separated on a precast denaturing gel (NuPAGE 4–12% gradient Bis-Tris gel w/MOPS) and transferred to a polyvinyl difluoride membrane. Monoclonal anti-polyHis-alkaline phosphatase antibody was used to identify the His-tagged proteins and immunoblotting signals were developed by BCIP/NBT FAST tablets.

Protein Purifications and Gel-Filtration Measurements.

Protein samples were purified from E. coli C41(DE3) cells expressing the NaChBac genes, as described previously (28), with the following modification; protein was eluted in 0.3% Cymal-5, 150 mM NaCl, 20 mM sodium phosphate, pH 7.8 containing 300 mM imidazole and concentrated in an Amicon concentrator (either a 100 KDa or 10 KDa cut-off). The sample was spun at 100,000  × g at 4 °C for 20 min and loaded onto a size-exclusion GF column (Superdex 200 10/300; GE Healthcare) equilibrated with 50 mM NaCl, 20 mM sodium phosphate, pH 7.8 containing 0.3% Cymal-5 at 4 °C and eluted at a rate of 0.3 ml/ min. The GF column had been calibrated with thyroglobulin, ferritin, amylase, adolase, alcohol dehydrogenase, covalbumin, ovalbumin, carbonic anhydrase, and cytochrome. The fractions corresponding to the expected tetrameric channel molecular mass were then run on a precast denaturing gel (NuPAGE 4–12% gradient Bis-Tris gel w/MOPS).

Synchrotron Radiation Circular Dichroism Spectroscopy.

SRCD spectra were collected on the DISCO beamline at the Soleil Synchrotron, France, and repeated on the CD1 beamline at the ISA Synchrotron, Denmark. For each protein, the extinction coefficient at 280 nm was calculated from the amino acid sequence using the Expasy ProtParam website (29) and the concentration determined from the A₂₈₀ measured on a Nanodrop 1000 UV spectrophotometer in triplicate immediately before the SRCD spectrum was acquired. The values thus determined are very reproducible (variation of less than +/-1%) and have been shown to be highly correlated with quantitative amino acid determinations of protein concentration (30). Three SRCD spectra were collected for each protein sample [∼1 mg/ml in a 0.054 mm pathlength Suprasil demountable cell (Hellma)] at 4 °C over the wavelength range from 260 nm to 175 nm, with a 1 nm step size and a 1 s dwell time. Three baseline spectra (which consisted of the GF column flow-through) were averaged and subtracted from the averaged sample spectrum, and the variability (one standard deviation) in the experimental measurements calculated at each wavelength. The spectra were smoothed with a Savitsky-Golay filter, and scaled to delta epsilon values using mean residue weights of 114.4, 114.5, 114.4, 115.0, 114.9, respectively, for the wild-type, 262Δ, 254Δ, 247Δ, and 239Δ constructs. Data processing was carried out using the CDTools software (31). Data were analyzed with the DichroWeb analysis server (32). The reported values are the average results from the CONTINLL, SELCON3, and CDDSTR (33) algorithms calculated using the SP175 reference dataset (21).

Sequence Analysis and Molecular Modeling.

The NaChBac sequence was used in a BLASTP search of the nonredundant protein database (34) to identify 32 putative prokaryotic Na_Vs based upon inclusion of the characteristic positively charged residues in the S4 voltage-sensing helix followed by a TxExW sequence in the filter region (in NaChBac this is TLESW) (13, 14).

The secondary structure of the NaChBac CTD from the N-terminal end of the S6 helix onward was predicted using PSIPRED (35), JPred3 using PSIBLAST and HMMER2 profiles (36), the HNN Hierarchical Neural Network method (37); PHD, (38) and APSSP2 (39). The DISOPRED algorithm (part of PSIPRED) was used to indicate regions with high propensities for forming natively disordered structures. The program Coils (version 2.2) (40) was used for the prediction of coiled-coil domains. Sequence alignments of the 32 putative prokaryotic Na_Vs were carried out using the ClustalW algorithm (41) and the sequence corresponding to the CTD immediately carboxyl to the S6 pore helix was manually aligned for maximal hydrophobicity at positions a and d of the heptad sequence.

An homology model of NaChBac was generated as described in O’Reilly et al. (28), except using the full-length KcsA crystal structure (12) as template (PDB code 3EFF). Residues 207–237 and 251–274 of the NaChBac sequence were aligned, respectively, with residues 87–117 and 136–159 of KcsA to generate the S6 and CTD helical sections. The 14-residue nonhelical section of the NaChBac CTD was modeled using the loop function of the SYBYL Biopolymer module (Version 8.0, Tripos Inc., St. Louis). Fig. 1B was produced using the PyMOL molecular graphics system (DeLano Scientific, San Carlos, CA).