Cross-kingdom auxiliary subunit modulation of a voltage-gated sodium channel

Voltage-gated, sodium ion–selective channels (NaV) generate electrical signals contributing to the upstroke of the action potential in animals. NaVs are also found in bacteria and are members of a larger family of tetrameric voltage-gated channels that includes CaVs, KVs, and NaVs. Prokaryotic NaVs likely emerged from a homotetrameric Ca2+-selective voltage-gated progenerator, and later developed Na+ selectivity independently. The NaV signaling complex in eukaryotes contains auxiliary proteins, termed beta (β) subunits, which are potent modulators of the expression profiles and voltage-gated properties of the NaV pore, but it is unknown whether they can functionally interact with prokaryotic NaV channels. Herein, we report that the eukaryotic NaVβ1-subunit isoform interacts with and enhances the surface expression as well as the voltage-dependent gating properties of the bacterial NaV, NaChBac in Xenopus oocytes. A phylogenetic analysis of the β-subunit gene family proteins confirms that these proteins appeared roughly 420 million years ago and that they have no clear homologues in bacterial phyla. However, a comparison between eukaryotic and bacterial NaV structures highlighted the presence of a conserved fold, which could support interactions with the β-subunit. Our electrophysiological, biochemical, structural, and bioinformatics results suggests that the prerequisites for β-subunit regulation are an evolutionarily stable and intrinsic property of some voltage-gated channels.

These hetero-tetrameric channels are sensitive to changes in transmembrane potential (voltage) and support the complex electrical signaling paradigms that underlie behavior, sensation, contraction, and mobility (1)(2)(3). To date, 10 sodium ion selective ␣-subunits have been classified in humans. The four domain primary structure of eukaryotic voltage-gated calcium selective channels (Ca V ) and Na V s is the result of two rounds of gene duplication that occurred in protists in a presumptive Ca V (4). Na V s subsequently evolved from Ca V in apoikozoa, a phylum comprising animals and the closely related choanoflagellates (5). In tandem, the emergence of the sodium selectivity and fast (submillisecond) voltage-dependent gating served as functional prerequisites for complex action potential firing and neuronal complexity (6). Furthermore, five isoforms of sodium channel ␤-subunits have been identified, namely, ␤1, ␤2, ␤3, ␤4, and splice variant ␤1b (7)(8)(9)(10)(11)(12). These 30 -36-kDa proteins are comprised of an extracellular amino terminus immunoglobulin-like domain, followed by an extracellular juxtamembrane region, a single transmembrane segment, and an intracellular carboxyl terminus; except for the splice variant ␤1b, which has a non-conserved carboxyl terminus and lacks a transmembrane segment (9). Related Na V ␤-subunits genes are expressed in cardiovascular and nervous tissues where they are part of the V-set immunoglobulin superfamily of cell adhesion molecules that exert non-conduction-related functions such as facilitating cell adhesion and cell migration (7,(12)(13)(14). Na V ␤subunits recruit ␣-subunits to the surface membrane in conjunction with the extracellular matrix and adhesion molecules ankyrin and contactin (13,(15)(16)(17)(18)(19)(20). Such interactions mediate sodium channel and K V 7 clustering to the axon initial segment and nodes of Ranvier that mediate the saltatory conduction in neurons (21)(22)(23)(24). Na V modulation is also seen in a wide range of ␣and Na V ␤-subunit combinations, where a prominent effect of the ␤-subunit is enhancement of current density by promoting trafficking of the ␣-subunit to the plasma membrane and altering ␣-subunit voltage-dependent gating (7,8,11,17,19,(25)(26)(27)(28)(29)(30)(31)(32). The extracellular domains of ␤-subunits interact with the pore-forming subunit via covalent (␤2/4) and non-covalent (␤1/3)-interactions (33)(34)(35)(36)(37)(38)(39). Modulation of channel voltage-dependent gating is achieved through the effects of glycosylation that functions through a charge shielding mechanism (40 -42).
Inherited mutations in Na V ␤-subunit genes can result in a variety of human disorders, including severe epilepsy (43)(44)(45)(46), cardiac arrhythmia (47,48), and are associated with cancer (49) and neuropsychiatric disorders (50,51). The underlying stoichiometry, putative interacting surfaces, and mechanisms of ␤-subunit modulation of Na V s remains poorly understood. Recently, Na V ␤-subunits have been shown to play a role in neuronal excitability through novel interactions with K V s, suggesting Na V ␤-subunits may influence the function of channel-types outside of the eukaryotic Na V gene family (52)(53)(54).
Related voltage-gated sodium selective channels have been identified in prokaryotes where they are thought to play diverse roles in pH regulation, flagella movement, and oxidative phosphorylation (55)(56)(57)(58). Among these bacterial sodium channels, NaChBac was first isolated from an alkaliphilic bacterium Bacillus halodurans and has been shown to be sodium selective and gated by voltage (59). Structural studies of bacterial Na V s (bNa V s) indicate they are surprisingly similar to eukaryotic voltage-gated cation channels, with a "domain-swapped" architecture of four voltage-sensing domains, which surround a central, gated pore (60). Consistent with this observation, bNa V s are phylogenetically related to eukaryotic Na V s (eNa V s) and Ca V , but represent a distinct evolutionary lineage. Specifically, bNa V s are the result of an early differentiation from bacterial K V s that likely occurred before the emergence of eukaryotes, whereas eNa V s appeared more recently. Notably, bNa V s are characterized by a homotetrameric quaternary structure, in contrast to the pseudotetrameric single polypeptide eNa V .
The functional and structural relationships between prokaryotic and eukaryotic channel types are not fully known. For instance, whereas eNa V s and bNa V s are both sodium selective, the mechanisms are markedly different (61,62). On the other hand, bNa V s show a pharmacological profile similar to that of the eNa V family, an observation that suggests that the channels of these two phyla might share structural similarities that support mechanisms of activation and inactivation (63)(64)(65). Despite potential differences, bNa V s have been the subject of extensive functional and structural studies that have provided key insights into mechanisms of cation recognition, drug binding, voltage-sensing, and channel gating (63, 66 -75).
Here, we report that NaChBac is modulated by the mammalian voltage-gated sodium channel auxiliary ␤1-subunit in Xenopus laevis oocytes. ␤-Subunit modulation of NaChBac was isoform-specific, ␤1 strongly enhanced current amplitudes by affecting cell-surface expression and the voltage dependence of NaChBac gating, whereas ␤2 marginally decreased cell-surface expression of the channel resulting in modest sodium current amplitudes. Furthermore, in HEK293T cells, ␤1 and ␤3 but not ␤2, co-purified a co-expressed NaChBac indicating that the functional modulation of NaChBac is the product of the formation of a novel protein complex comprised of a bacterial sodium selective pore and a eukaryotic ␤-subunit. The reciprocal purification of NaChBaC and co-purification of ␤1 and ␤3, but not ␤2 indicates the specificity of this interaction. An evolutionary analysis of the ␤-subunit V-set immunoglobulin family pinpoints their emergence in osteichthyes, roughly 420 million years ago, and failed to identify related protein families in bacteria. Modeling analysis of the interaction regions between ␤1 and NaChBac based on the cryo-EM interaction between elec-tric eel ␤1 and Na V 1.4 indicate a conserved interaction of polar residues in the protein-protein interface between the two ␣-subunits. In total, the data suggests that the gene family that comprises the sodium channel auxiliary ␤-subunits may act as broad regulators of electrical signaling through the interaction and modulation of multiple channel types. This functional plurality may arise from the strict evolutionary conservation of specific structural elements found in the voltage-gated ion channels.

Results
Eukaryotic Na V ␤-subunits are known to modulate eukaryotic Na V channels (3,76), and recently this functionality has been extended to the potassium channel isoforms K V 1, K V 4.2/ 4.3, and K V 7 (52)(53)(54). To test the possibility that eukaryotic ␤-subunits may also modulate bNa V s, cRNA encoding NaCh-Bac was co-injected with eukaryotic Na V ␤1 and ␤2 in X. laevis oocytes, an expression system that lacks endogenous ␤-subunit expression, unlike a majority of mammalian cell lines (3,(77)(78)(79)(80)(81). Co-expression of NaChBac with ␤1 in Xenopus oocytes significantly increased sodium current amplitudes over NaChBac alone and this trend was sustained over a 50-h time course (Fig. 1, A-C). NaChBac injected alone resulted in currents roughly 36 h post-injection, whereas NaChBac/␤1-mediated currents were significantly above background 12 h after injection (Fig. 1A). Saturation of the functional expression was not observed, but NaChBac/␤1-mediated currents exceeded the level of reliable voltage clamp after 48 h (Fig. 1A). In contrast, the ␤2 effect on NaChBac-mediated currents was marginal ( Fig. 1, A-C) and small sodium currents were measurable only 36 h post-injection and did not increase over longer incubation time (Fig. 1A).
Evidence of a possible ␤-subunit interaction with NaChBac, either direct or via a complex, was obtained by affinity purification. ␤1, ␤2, and ␤3 were isolated from protein lysates by a C-terminal His 6 tag using a Ni-NTA column. HEK293T cells transfected with NaChBac/␤1 and NaChBac/␤3 showed association via co-pulldown of NaChBac. This was observed by immunoblotting for a NaChBac N-terminal HA probe that was detected in the elution fraction upon pulldown of the ␤-subunit, which were identified via the V5 C-terminal tag on the ␤-subunits. Ni-NTA pulldown of NaChBac/␤2 showed pulldown of the ␤-subunit, observed via a C-terminal HA tag, however, NaChBac was not detected in the elution for NaChBac/␤2 or NaChBac only ( Fig. 2A). This result was replicated with NaChBac pulldown using Strep-Tactin-Sepharose resin against a N-terminally Twin-Strep-tagged NaChBac, whereby both ␤1 and ␤3 co-purified with NaChBac, however, ␤2 was not observed in the NaChBac purification elution (Fig. 2B). This reciprocal purification suggests either a direct interaction or the composition of a complex that is tightly associated with both NaChBac and the ␤1/3-subunits. However, the presence of an endogenous ␤1-subunit expressed in HEK293T cells precluded electrophysiological studies of NaChBac and co-expressed Na V ␤-subunits. Furthermore, cell lines that lack ␤1, such as CHL or COS cells, may contain ␤1b or ␤3, the latter being a post-transcriptionally regulated soluble ␤-subunit that is capable of regulating Na V s (82).

The eukaryotic Na V ␤1-subunit modulates NaChBac
To determine the possible mechanism of current enhancement by the ␤1-subunit we assayed the plasma membrane appearance of NaChBac by biotinylation of cell-surface proteins in the presence and absence of ␤1 or ␤2 in Xenopus oocytes. NaChBac surface expression was detectable biochemically at the plasma membrane after 24 h post-injection and the total (surface ϩ cytoplasmic) expression was comparable in all three conditions (Fig. 1D). However, the biotinylated (plasma membrane) fraction indicates co-expression of ␤1 promoted the surface expression of NaChBac, whereas ␤2 co-expression resulted in a reduction of surface-located channels. Densitometric analysis of the biotinylation data reveals that ␤1 increases NaChBac surface expression by ϳ450%, whereas the ␤2-subunit reduced expression by ϳ30% (Fig. 1D).
Given that the data demonstrated that ␤1 was able to form a stable protein complex with NaChBac and increase the cellsurface channel density, we next assessed possible effects on voltage-gated properties of NaChBac expressed in Xenopus oocytes. First, a variable length depolarizing pulse was used to minimize the effect of channel inactivation on the tail current.
These results suggest a vestigial conservation between the prokaryotes and eukaryotes ␣/␤-subunit structural interaction interface. This hypothesis raises several fundamental questions: do bacterial counterparts of ␤-subunits exist? If so, when did ␤1/3 and ␤2 begin to functionally diverge? To answer these questions, we performed a bioinformatics analysis aimed at tracing the evolutionary history of these proteins. Whereby, we scoured the comprehensive Uniprot database to identify genes homologous to human ␤1-subunits in other genomes.
This genome-wide scan identified an extensive set of diverse sequences including, as expected, the genes encoding ␤2, ␤3, promotes an increase in current density. Co-injection of ␤1-subunit increases the current of NaChBac, whereas ␤2 mildly reduces the current amplitude as compare with NaChBac alone. B, current-voltage relationship 42 h post-injection. C, peak current (at Ϫ20 mV) of NaChBac with ␤-subunits at the indicated time points. D, surface-expressed NaChBac channels assessed by channel biotinylation at 24 h post-injection. Left, total expression of NaChBac in the supernatant fraction is not altered by co-expression of the ␤1 or ␤2 subunit. Anti-tubulin serves as a loading control. Right panel indicates the surface expression of NaChBac is increased by ␤1 but not ␤2 (left to right), see "Experimental procedures" for experimental details. Absence of tubulin in the biotinylated fraction ensures biotinylation of surface proteins only.
The eukaryotic Na V ␤1-subunit modulates NaChBac and ␤4 isoforms in different organisms and provided some insights into their evolutionary history. First and foremost, all identified ␤-subunit genes are from eukaryotes and, in particular, metazoa. Second, the branch of the dendrogram containing the ␤-subunits highlights an interesting hierarchy of sequence similarities: ␤1 and ␤3 are closer to each other than to the pair formed by ␤2 and ␤4. Thus, it appears that the functional diver-gence between ␤1 and ␤2 predates the gene duplication event that gave rise to ␤3 and ␤4. We therefore decided to further investigate the close evolutionary relationship between ␤1 and ␤3 by inferring their phylogenetic tree (Fig. 4). We found that ␤1 and ␤3 have a relatively recent evolutionary history: these genes are found in euteleostomi (bony vertebrates) but not in chondrichthyes (cartilaginous fishes). This observation indi- . NaChBac and ␤-subunit reciprocal purification. A, blots depict ␤1 and ␤3 co-purified with NaChBac Strep-Tactin purification. Shown are lysate and elution fractions from Strep-Tactin-Sepharose pulldown of N-terminally Twin-Strep Tag NaChBac (left) and expression and co-purification of ␤-subunits with the NaChBac pulldown (right three sections) from HEK293T cells. B, blots depict NaChBac co-purified with ␤1 and ␤3. Shown are Ni-NTA pulldown fractions: flow through, washes, and elution of His tag containing ␤-subunits from HEK293T cells transfected with NaChBac alone or NaChBAc ϩ ␤1, ϩ ␤2 or ϩ␤3, see "Experimental procedures" for details.
The eukaryotic Na V ␤1-subunit modulates NaChBac cates that the last common ancestor between ␤1 and ␤3 occurred around 420 million years ago, and the absence of ␤2 in chondrichthyes further confirms that ␤-subunits were not present before the emergence of euteleostomi and therefore unlikely be present in prokaryotes. Furthermore, this possibility extends to the greater family of V-set proteins (pfam.xfam.org/ family/PF07686.15#tabviewϭtab7) containing the immunoglobulin domain, which is not identified in prokaryotes and is sufficient for ␤-subunit modulatory effects (35,36,83).
Overall, the evolutionary history of ␤-subunits and, in particular, the lack of a recognizable prokaryotic homologue suggests that the major structural elements of Na V s involved in the ␣/␤ interactions are conserved across kingdoms. Indeed, a conserved structural platform for docking ␤-subunits would explain the interaction of a human auxiliary subunit with a voltage-gated bacterial channel. To examine this possibility, we analyzed the recently determined structure of the complex between ␤1 and Na V 1.4 from electric eel obtained via cryoelectron microscopy (37). This structural information revealed that, besides the interactions between the immunoglobulin domain of ␤1 and the extracellular loops of Na V 1.4, a large set of residue-residue contacts are present between the transmembrane regions of the two proteins. In particular, the transmembrane helix of ␤1 shares an extended protein-protein interface with the membrane boundary S0 and transmembrane S2 of Na V 1.4 D III . Importantly, this "bidentate" interaction, involving  Table 1. C, steady-state inactivation (SSI) (protocol shown in the inset) shows ␤1 induces a statistically significant hyperpolarizing shift in the SSI relationship. SSI protocol is shown in the inset; 3-s prepulse (PP) at various potentials is followed by 37.5 ms of test pulse (TP) at Ϫ20 mV. Oocytes are held at holding potential of Ϫ120 mV throughout the experiments. D, co-expression of ␤1-subunit accelerates activation kinetics of NaChBac. Inset: normalized activation curves of NaChBac only (filled circle), NaChBac co-expressing ␤1 (open circle), NaChBac with ␤2 (open square). Scale bar represents 20 ms. E, ␤1-subunit speeds the inactivation kinetics of NaChBac. Inset: representative traces of WT NaChBac only, NaChBac with ␤1 and NaChBac with ␤2 (left to right) with single exponential fit. Scale bar denotes 500 ms.
The eukaryotic Na V ␤1-subunit modulates NaChBac simultaneous contact with the voltage sensor domain and the extracellular part of the pore, dictates a precise arrangement between these two parts of the channel. Utilizing this structure as a template for the ␤1-␣ binding domain, we set out to examine these interaction elements in NaChBac. However, because an experimentally derived structural model of NaChBac is not available, we considered the recently determined structure of the full-length model of Na V Ms (84). This channel has been extensively electrophysiology characterized and, as NaChBac, has been shown to possess a pharmacological profile very similar to that of eNa V s (64). We performed a structural superposition between Na V 1.4 and Na V Ms using MISTRAL (85), an algorithm that finds the optimal rotations and translation to minimize the root mean square deviation between corresponding C ␣ atoms in two proteins. In this case, MISTRAL identifies the two transmembrane regions as structural homologues and returns a remarkably small root mean square deviation value (2.3 Å). This is indicated by the closely superimposed pore domains (Fig. 5, A and B), divergence is mainly from the voltage sensors, with the exception of D III . Thus, it appears that only D III , the domain involved in the ␣/␤ interaction, has conserved the "ancestral" arrangement with respect to the pore domain.
Having ascertained that the structure of a bNa V can provide an interaction interface for the ␤-subunits similar to that of eNa V s, we focused our attention on the specific networks of residue-residue interactions between the two proteins. To this end we considered a homology model of NaChBac based on the structural template of Na V Ab and subject to multi-s molecular dynamics simulations (75,86). Despite being a theoretical prediction, this structural model has been repeatedly experimentally validated (87,88). The superposition of NaChBac and Na V 1.4 reveals a structural match as good as the one observed for Na V Ms. Moreover, a peculiar pattern emerges in the side chain chemical identity at the protein-protein interface: there are two clusters of polar residues, one in the middle of the membrane and the other more displaced toward the extracellular compartment that are present in both NaChBac and Na V 1.4 that engage in directional interaction with hydrophilic residues on ␤1 (Fig. 5C). We therefore examined the evolutionary pattern of the polar transmembrane helices as a potential basis for a NaChBac preferential interaction with ␤1/3 over ␤2/4.
To address these issues, we analyzed comprehensive multiple sequence alignments of eNa V s and bNa V s containing thousands of genes from all sequenced organisms (4). Surprisingly, the residues of S2 involved in the interaction with ␤1 are the least conserved, not only in bNa V s, but also in eNa V s (Fig. 6A). The absence of a statistical signal of sequence conservation suggests that the ␣/␤ interaction might not be absent in many organisms. An "intermittent" presence of a protein-protein interaction, whereby mutation of one or two residues switches "on" or "off" an interaction in different organisms, is not uncommon in evolution and might be at work also in the case of ␤-subunits (89). However, despite the poor conservation of the contacting residues, a separate analysis of the sequences encoding for D III reveals a prevalence of hydrophilic residues at the positions where the polar interactions take place (Fig. 6B). This conservation is, to some extent, present also in bacterial channels but not, for instance, in eukaryotic D II .

Discussion
Eukaryotic Na V ␤-subunits are multifunctional transmembrane proteins that can have numerous modulatory effects on the expression and function of the pore-forming Na V ␣-subunit. In addition to the well known effects on Na V biophysical characteristics by ␤-subunits, they play important roles in cellular maintenance as adhesion molecules that interact with cytoskeletal ankyrin and contactin proteins, interactions that can also enhance the functional expression of sodium channels at the cell surface (21)(22)(23)(24). In the present study, the data shows that the eukaryotic Na V ␤1-subunit significantly increases the peak current and functional properties of the bacterial channel NaChBac in X. laevis oocytes. By contrast, Na V ␤2 modestly influenced the number of functional NaChBac channels at the plasma membrane, with no significant effects on channel function. The results surprisingly mirror ␤-subunits isoform-specific surface enhancement of cardiac Na V s by the ␤1-subunit, but not by ␤2 (90). ␤1 co-expression induced hyperpolarizing shifts in the voltage dependence of both activation and inactivation of NaChBac. Additionally, this functional regulation was Table 1 eNaV␤-subunits affect the gating kinetics of NaChBac Time constants were obtained from single exponential fits of activation and inactivation kinetics at various voltages. Values in parentheses represent sample replicates used to calculate the mean Ϯ S.E. Statistical significance was set at p Ͻ0.05 and marked with asterisks.

Table 2 Effects of eNa V ␤-subunits on NaChBac activation and inactivation
Conductance-voltage relationship and steady-state inactivation protocol were fit with a Boltzmann function. Mean values of V 1/2 and the slope (dx) of the fit are depicted in mv with number of samples used to calculate noted in brackets. Statistical significance was set at p Ͻ0.05 and marked with asterisks.
The eukaryotic Na V ␤1-subunit modulates NaChBac supported via a biochemical pulldown assay in mammalian HEK293T cells whereby heterologously expressed NaChBac co-affinity purified with Na V ␤1 and Na V ␤3 but not with Na V ␤2 the reciprocal purification was also observed whereby Na V ␤1 and Na V ␤3 co-purified with NaChBac, whereas Na V ␤2 did not. Furthermore, co-expression of Na V s with the ␤1-subunit has been shown to increase the peak of brain sodium current and to shift the voltage dependence of inactivation (34). It is not known for eNa V s if the same underlying interactions and mechanisms are responsible for ␤-subunit-induced surface membrane enhancement and tuning of voltage gating. Recent cryo-EM evidence shows a direct interaction of the ␤1 extracellular domain with the Na V voltage-sensing domain, the ␤1 interacting domain is highly conserved with the ␤3-subunit and we theorize this to be the interface of not only the ␤1 but also the ␤3-subunit interaction with NaChBac observed in our copurification conditions (37). NaChBac lacks any cysteine residues to form a disulfide bond with the ␤2/4 extracellular domain that is necessary in ␤-subunit covalent interaction with Na V s and localization to nodes of Ranvier (91). Na V s have a presumed 1:1 ratio with the ␤-subunit yet this is believed to interact with a distinct voltage-sensing domain, whereas NaChBac like K V s have multiple identical voltage-sensing domains, making it unclear if the stoichiometry of the NaChBac-␤subunit complex is conserved (92). The amenable nature of bNa V structural studies makes it conceivable that a structure of a bNa V -␤-subunit complex is feasible and such studies could shed light on the general mechanism(s) of ␤-subunit regulation.
The current density arising from the inward flux of sodium ions is related to the total number of channels in the surface membrane, the amount of time that they spend in the open conformation (open probability), and the amount of sodium ions that flow through the open channel (unitary conductance). However, up-regulation of the mammalian sodium current by ␤1 co-expression in both native settings and in heterologous expression systems can be simply explained by enhanced surface expression (25,47), similar to the effect here by ␤1 on NaChBac. Yet, for eNa V s or bNa V s the data cannot distinguish between enhanced forward trafficking, enhanced surface sta-  Each domain is rendered using a different color: D I is shown in purple, D II in cyan, D III in blue, and D IV in light brown. Na V Ms and the ␤1-subunit bound to Na V 1.4 are shown in white and red, respectively. Note that, whereas the voltage sensor from D III is perfectly superimposed, the same domain from D I or D II (purple and cyan, respectively) does not show the same degree of structural similarity. C, structural alignment between S0 and S2 of Na V 1.4 D III (blue) and NaChBac (white); the ␤1-subunit bound to Na V 1.4 is shown in red. Side chains engaged in residue-residue contacts across the protein-protein interface are shown as sticks colored according either the non-polar (white) or polar (green) character of their side chains.
The eukaryotic Na V ␤1-subunit modulates NaChBac bility, or diminished endocytic and/or lysosomal retrieval. Regardless, the enhanced surface expression of NaChBac by co-expression with ␤1 may have practical applications in the study of other bNa V s that have shown poor expression profiles in heterologous systems.
The extent to which eNa V s can be modulated by ␤-subunits can depend strongly on the expression system. One possible cause of this functional variability may be the widespread endogenous expression of Na V ␤-subunits in mammalian cell lines (3,15,93,94). Unlike HEK293T cells, X. laevis oocytes do not have a known ␤-subunit thus motivating their use for the present study (93,94). Interestingly, NaChBac demonstrates robust expression in HEK293T cells in our hands but these currents were not obviously modulated by either ␤1 or ␤2 expression, possibly because of a competing interaction with the expressed channel such protein being expressed in HEK293T and not oocytes may be an endogenous ␤-subunit (not shown). Thus it is possible that the previous reports of expressed Na V s (95) or NaChBac currents in HEK293T or tSA 201 cells (61,96,97), may have been, in fact, channels associated with endogenous ␤-subunits.
An evolutionary analysis of the ␤-subunits family reveals that these auxiliary subunits are part of a large family comprising several myelin-associated proteins. Moreover, the phylogenetic tree of these genes reveals that ␤1/␤3 and ␤2/␤4 form two distinct groups. Strikingly, a recent cryo-EM structure of the complex between Na V 1.4 and ␤1 shows that the part of the channel involved in the protein-protein interaction is structurally superimposable to the corresponding region in NaChBac, even though the individual amino acids involved in the ␣/␤ interaction are not conserved across evolution. In total, the electrophysiological, biochemical, structural, and bioinformatics data suggests that the structural prerequisites for ␤-subunit regula-tion are an evolutionarily stable, intrinsic property of some voltage-gated channels, including bNa V s, that were present before the emergence of the auxiliary proteins.

Two-electrode voltage clamp
Voltage-clamped sodium currents were recorded with two microelectrodes using an OC-725C voltage clamp (Warner, Hamden, CT) in a standard Ringers solution (in mM): 116 NaCl, 2 KCl, 1 MgCl 2 , 0.5 CaCl 2 , 5 HEPES, pH 7.4. Current recordings were performed at 20 -22°C. Glass microelectrodes had resistances of 0.1-1 megaohm and were backfilled with 3 M KCl. The holding potential was Ϫ120 mV in all cases to minimize current rundown. GV relationships were derived by plotting the isochronal tail current amplitudes (the current amplitude measured after stepping from the test potential back to a holding potential of Ϫ120 mV) as a function of the depolarizing pulse potential. Depolarizing pulses were delivered as envelope protocols to minimize the contribution of inactivation in GV analysis. Displayed current traces show the full active voltage range of GVs. All data are mean Ϯ S.E. The eukaryotic Na V ␤1-subunit modulates NaChBac Biochemical detection of NaChBac surface expression Biotinylation was performed using 63 oocytes per experimental condition. All biotinylation procedures were performed as previously described (98). Briefly, anti-HA high affinity antibody (Roche Applied Science, 11867423001), directed against an HA peptide on the N terminus of NaChBac was used because the C-terminal HA tag on NaChBac increased the current amplitude and abolished modulatory effects of ␤1 and ␤2. Mouse monoclonal antibody directed against ␣-tubulin (T6199) was purchased from Sigma. Both anti-rat (SAB3700526) and anti-mouse (SAB3700993) horseradish peroxidase-conjugated secondary antibodies were obtained from Sigma. HAtagged NaChBac was detected by incubating the membrane with 1:1,000 diluted anti-HA at 4°C overnight and 1:10,000 of anti-rat secondary antibody incubation for 1 h at room temperature. The nitrocellulose membrane was stripped with mild stripping buffer, as described in the Abcam stripping for reprobing protocol. For tubulin detection in each sample, membrane was incubated with anti-tubulin (1:1,000) for 3 h at room temperature, followed by a 1-h room temperature incubation with anti-mouse (1:5,000). ImageJ software (NIH) was used to quantify band densities.
His tag purification-Lysate was diluted in half with PBS and imidazole to a final 10 mM imidazole concentration before being loaded onto a 250-l PerfectPRo Ni-NTA-agarose column (5Prime, 2900510). The column was washed 4 times with 1 ml of PBS ϩ 0.1% Triton X-100 and 20 mM imidazole and eluted with 1 ml of PBS ϩ 0.1% Triton X-100 with 500 mM imidazole and concentrated 50-fold with an Amicon Ultra 10-kDa concentrator (UFC501024). Fractions were loaded onto a 9% polyacrylamide gel (1% total lysate, 1% pulldown fraction or 3% elution for ␤-blots or 12% elution for NaChBac blots) and run out at 50 V for 30 min, 150 V for 1 h. Protein was transferred to polyvinylidene difluoride (GVS NA, 1212637) at 100 V for 1 h.

Bioinformatics analysis of ␤-subunits
A multiple sequence alignment was obtained using first HMMER (100) to scan the Uniprot database (RP35 proteome), using phmmer tool to collect homologous sequences of human ␤1 and ␤2 sequences. These were then accurately aligned with MAFFT and redundant sequences with pairwise identity higher than 95% were removed to obtain a first general alignment (99). From the derived phylogenetic tree the branches containing SwissProt-annotated ␤1 and ␤3 sequences were extracted and the relative sequence realigned with MAFFT to obtain a more refined tree. All the trees were generated using FastTree (101); the statistical significance of each branch was estimated using the bootstrap approach (a posterior probability higher than 0.7 is considered significant).