Modulation of voltage-gated K⁺ channels by the sodium channel β1 subunit

Abstract

Voltage-gated sodium (Na_V) and potassium (K_V) channels are critical components of neuronal action potential generation and propagation. Here, we report that Na_Vβ1 encoded by SCN1b, an integral subunit of Na_V channels, coassembles with and modulates the biophysical properties of K_V1 and K_V7 channels, but not K_V3 channels, in an isoform-specific manner. Distinct domains of Na_Vβ1 are involved in modulation of the different K_V channels. Studies with channel chimeras demonstrate that Na_Vβ1-mediated changes in activation kinetics and voltage dependence of activation require interaction of Na_Vβ1 with the channel’s voltage-sensing domain, whereas changes in inactivation and deactivation require interaction with the channel’s pore domain. A molecular model based on docking studies shows Na_Vβ1 lying in the crevice between the voltage-sensing and pore domains of K_V channels, making significant contacts with the S1 and S5 segments. Cross-modulation of Na_V and K_V channels by Na_Vβ1 may promote diversity and flexibility in the overall control of cellular excitability and signaling.

Sequential opening and closing of central nervous system (CNS) voltage-gated sodium (Na_V) and potassium (K_V) channels mediate the depolarization phase (1) and repolarization phase (2, 3), respectively, of neuronal action potentials. Although intrinsic domains within the Na_V α subunits underlie voltage-dependent gating properties and sodium-specific permeation, five Na_Vβ subunits (β1, β1B, β2, β3, β4) assemble with and modulate inactivation kinetics and the voltage dependence of activation and inactivation of Na_V α subunits (1, 4). In addition, these Na_V β subunits function in cell adhesion and contribute to neuronal migration, pathfinding, and fasciculation (4–8). Given their ubiquitous roles and distribution in the CNS, Na_V β subunits play a major role in fine-tuning of action potential generation, propagation, and frequency. Subtle changes to their function have resulted in a range of detrimental neurological diseases in humans such as genetic epilepsy with febrile seizures plus (GEFS+), Dravet syndrome (severe myoclonic epilepsy of infancy), temporal lobe epilepsy, febrile seizures, and decreased responsiveness to the anticonvulsant drugs carbamazepine and phenytoin (7–12).

K_V channels are important contributors to action potential repolarization. K_V channels in mammals are encoded by 40 genes grouped into 12 subfamilies (K_V1–K_V12). Na_Vβ1, which was previously thought to be specific for Na_V channels, was recently shown to coassemble with and modulate the properties of the K_V4.x subfamily of channels (13–15). In the rodent heart, Na_Vβ1 coprecipitates with K_V4.3, and in heterologous expression systems it increases the amplitude of the K_V4.3 current and speeds up activation (14, 15). In the rodent brain, Na_Vβ1 coprecipitates with K_V4.2, and in heterologous expression systems it enhances surface expression of the channel and increases current amplitude (13). Genetic knockout of Na_Vβ1 in mice prolongs action potential firing in pyramidal neurons through its action on K_V4.2 (13).

Here, we examined whether the Na_Vβ1–K_V4.x interaction was specific for this family of K_V channels, or whether Na_Vβ1 could interact and modulate the functions of other families of K_V channels in mammals. We selected three families of K_V channels (K_V1, K_V3, and K_V7) for these studies. We report that Na_Vβ1 modulates the function of K_V1.1, K_V1.2, K_V1.3, K_V1.6, and K_V7.2 channels, but not K_V3.1 channels, in an isoform-specific manner. Through the use of chimeric and mutational strategies we identified regions in Na_Vβ1 and K_V channels required for channel modulation, and docking simulations suggest a molecular model of the interaction of the two proteins.

Results

Na_Vβ1 Interacts with and Modulates K_V1.x Channels.

We examined the ability of Na_Vβ1 to modulate activation kinetics, voltage dependence of activation and deactivation of K_V1.x channels expressed in mammalian cells or in Xenopus oocytes. In Xenopus oocytes, Na_Vβ1 accelerated K_V1.2 activation, shifted the voltage dependence of activation in the hyperpolarized direction and sped up the fast component of deactivation (τ_fast at −60 mV) (Fig. 1 A–C). In mammalian cells, K_V1.2 exhibited use-dependent activation (i.e., successive depolarizing pulses progressively increase the amplitude of the K_V1.2 current and speed up activation), a unique property of this channel (16). Na_Vβ1 significantly accelerated K_V1.2 activation (τ_fast at +40 mV) at the first pulse and to a lesser extent at the ninth pulse (Fig. 1D). Coprecipitation experiments demonstrated that Na_Vβ1 (dual-tagged at the C terminus with V5 and His_6×) and mouse K_V1.2 (FLAG-tagged at the C terminus) coassembled when heterologously expressed in mammalian cells (Fig. 1E). Furthermore, immunostaining experiments on normal mouse brain showed K_V1.2 colocalization with Na_Vβ1 in the axon initial segment of neurons in the cerebral cortex (Fig. 1F) but not in nodes of Ranvier (Fig. S1A).

Fig. 1.

Activation kinetics at +40 mV (A, gray hatched area in pulse protocol represents first 30 ms analyzed), voltage dependence of activation (B) (mean ± SEM), and deactivation (C, gray hatched area in pulse protocol represents last 10 ms analyzed) of K_V1.2 channels expressed in Xenopus oocytes in the presence or absence of Na_Vβ1 (n ≥ 7). Only currents of comparable amplitude between channel-alone, channel plus Na_Vβ1, were selected for analysis. (D) K_V1.2 stably expressed in L929 fibroblasts exhibits use-dependent activation when repetitive depolarizing pulses at +40mV are administered at 1-s intervals. Average normalized current traces show Na_Vβ1 speeds up activation of K_V1.2 at the first pulse and to a lesser extent at the ninth pulse (K_V1.2, n = 6; K_V1.2 + Na_Vβ1, n = 9 cells). (E) Western blot of coprecipitation experiment in transfected HEK cells showing that K_V1.2 and Na_Vβ1 coassemble. (F) Mouse cerebral cortex immunostained for K_V1.2 (green; Left) and Na_Vβ1 (red; Center), showing colocalization in the axon initial segment in the merged image (Right).

K_V1.1 and K_V1.3 had a different staining pattern and did not colocalize with Na_Vβ1 in normal mouse brain (Fig. S1 B and C). However, these channels may colocalize with Na_Vβ1 in demyelinating diseases where their distribution is altered (17–19). We therefore tested the effect of Na_Vβ1 on K_V1.1, K_V1.3, and K_V1.6 channels (Table S1). Na_Vβ1 shifted the voltage dependence of activation of K_V1.1 by ∼4 mV in the hyperpolarized direction and it slowed deactivation (τ_fast) of the channel without affecting its activation kinetics (Fig. 2 A–C). Na_Vβ1 significantly accelerated activation of K_V1.3 in Xenopus oocytes (Fig. 2D) and in mammalian cells (Fig. S2F) but had no effect on K_V1.3’s voltage dependence of activation or deactivation kinetics (Fig. 2 E and F). Na_Vβ1 significantly reduced cumulative inactivation of K_V1.3 (Fig. S2 A–C), a unique property of K_V1.3 where successive depolarizing pulses cause a progressive diminution in the amplitude of the K_V1.3 current as channels accumulate in the C-type inactivated state (18). His-tagged Na_Vβ1 complexes with and accelerates activation of K_V1.3 in mammalian cells and Xenopus oocytes (Fig. S2 D–F). Na_Vβ1 slowed K_V1.6’s activation, shifted its voltage dependence of activation in the depolarized direction, and significantly increased the amplitude of the K_V1.6 tail current, but had no effect on the channel’s deactivation kinetics (Fig. 2 G–I). Endogenous K_V channels in pheochromocytoma (PC12) cells, which include K_V1.x channels (20, 21), were also modulated by Na_Vβ1 (Fig. S3). Thus, Na_Vβ1’s modulatory effects on K_V1 channels vary in an isoform-specific manner.

Fig. 2.

Activation kinetics (A, D, G, and J), voltage dependence of activation (B, E, H, and K) (mean ± SEM), and deactivation kinetics (C, F, I, and L) of K_V1.1, K_V1.3, K_V1.6, and K_V3.1 expressed alone (black) or with Na_Vβ1 (red) in Xenopus oocytes (K_V1.1, K_V1.3, and K_V1.6) or mammalian cells (K_V3.1) (n ≥ 5). (M) Western blot of coimmunoprecipitation experiment in transfected HEK cells showing that K_V7.2 and Na_Vβ1 coprecipitate. (N) Na_Vβ1 slows activation of the K_V7.2 current in CHO cells at moderate depolarizing potentials (n = 5). (O) Relative instantaneous current of K_V7.2 shows a significant difference in the 1-s prepulse potential in the presence of Na_Vβ1 (n = 9).

Na_Vβ1 Modulates K_V7 Channels but Not K_V3 Channels.

The K_V3 family contains four members, K_V3.1–K_V3.4. K_V3.1 is a fast-activating/fast-deactivating channel that regulates spike frequency in several brain regions (22–24). When expressed heterologously in mammalian cells, Na_Vβ1 had no effect on activation kinetics, voltage dependence of activation or deactivation of K_V3.1 (Fig. 2 J–L).

The K_V7 family contains five members, K_V7.1–K_V7.5. K_V7.2 and K_V7.3 are the main molecular components of the neuronal M-current, a noninactivating, slowly deactivating, subthreshold current that regulates neuronal excitability (25). We examined whether Na_Vβ1 modulated K_V7.2 channels, which are found in the axon initial segment and at nodes of Ranvier (25). In mammalian cells, Na_Vβ1 coassembled with FLAG-tagged K_V7.2 (Fig. 2M), and it slowed the channel’s activation at moderate depolarization voltages (Fig. 2N, Fig. S4) and altered current measured at different 1-s prepulse potentials (Fig. 2O). These data together with those presented in Fig. 1 and in the literature (13–15) demonstrate that Na_Vβ1 modulates K_V1.1, K_V1.2, K_V1.3, K_V1.6, K_V4.2, K_V4.3, and K_V7.2, but not K_V3.1, each in a unique fashion.

Distinct Domains of Na_Vβ1 Are Involved in K_V Channel Modulation.

Na_Vβ1 and myelin zero (P₀), the major protein in peripheral nerve myelin (26), are both type-1 membrane proteins that contain an extracellular Ig fold, a single transmembrane segment, and an intracellular C-terminal region (27) (Fig. 3A). Na_Vβ1 sped up acceleration of K_V1.2 and K_V1.3, whereas P₀ slowed down K_V1.3 activation and had no effect on K_V1.2 activation (Fig. 3 B and E). Because Na_Vβ1 and P₀ have different effects on K_V1.2 and K_V1.3, we used two chimeras of Na_Vβ1 and P₀ to define the regions required for modulation of these channels (Fig. 3A). These chimeras were previously used to identify Na_Vβ1 domains required for Na_V channel modulation (28). The Na_Vβ1-P₀ chimera contains the extracellular Ig domain of Na_Vβ1 and the transmembrane and intracellular segments of P₀, whereas the P₀-Na_Vβ1 chimera contains the extracellular domain of P₀ and the transmembrane and intracellular segments of Na_Vβ1 (Fig. 3A).

Fig. 3.

Schematic representation of Na_Vβ1 (red), myelin P₀ protein (blue), and their chimeras (A). Inset, Extracellular domain of Na_Vβ1 showing the R85C and C121W epilepsy-causing mutations. (B and E) Effect on K_V1.2 and K_V1.3 activation kinetics by Na_Vβ1 versus P₀ (B and E), chimeras (C and F), and Na_Vβ1 mutants (D and G) studied in Xenopus oocytes. Current traces are averages of normalized, representative currents of comparable amplitudes and shown are the first 30-ms activating phase of the 200-ms traces.

The external Ig domain of Na_Vβ1 is solely responsible for acceleration of K_V1.3 activation because the Na_Vβ1-P₀ chimera (containing Na_Vβ1’s external domain) sped up K_V1.3 activation like Na_Vβ1, but the P₀-Na_Vβ1 chimera (containing Na_Vβ1’s transmembrane/intracellular domain) had no effect (Fig. 3B). In support, two GEFS+ mutations (C121W and R85C) that disrupt the external Ig domain abolished Na_Vβ1-mediated acceleration of K_V1.3 activation (Fig. 3C). The external domain of Na_Vβ1 was also responsible for reducing cumulative inactivation of K_V1.3 because the Na_Vβ1-P₀ but not the P₀-Na_Vβ1 chimera significantly reduced cumulative inactivation, and the two GEFS+ mutations abolished this modulation (Fig. S5 A and B).

Full-length Na_Vβ1 accelerated K_V1.2 activation but neither chimera recapitulated this modulation, indicating that the entire protein is required (Fig. 3 E–G). Two Na_Vβ1 GEFS+ mutants with a disrupted external Ig domain accelerated K_V1.2 activation like Na_Vβ1 (Fig. 3E), indicating that an intact external Ig domain is also not sufficient for this modulation.

Full-length Na_Vβ1 slowed K_V1.1 deactivation (Fig. 2C). This modulation requires the transmembrane and/or cytoplasmic region of Na_Vβ1 because the C121W Na_Vβ1 mutant was as effective as full-length Na_Vβ1 in slowing K_V1 deactivation (Fig. S5C). Collectively, our results demonstrate that distinct Na_Vβ1 domains are required for modulating different K_V channels, each in a channel isoform-specific manner.

Na_Vβ1 Interacts with the Voltage-Sensing and Pore Domains of K_V1 Channels.

We constructed chimeras of K_V1.1 and K_V1.3 to identify channel regions required for Na_Vβ1-mediated modulation. We chose this pair of channels because of the distinctive modulatory effects of Na_Vβ1. Na_Vβ1 sped up activation and decreased cumulative inactivation of K_V1.3, whereas it had no effect on K_V1.1’s activation and inactivation (Fig. 2, Table S1). Na_Vβ1 shifted the voltage dependence of activation of K_V1.1 in a hyperpolarizing direction and it slowed deactivation (τ_fast at −60 mV) of K_V1.1, but did not alter these properties in K_V1.3 (Fig. 2, Table S1). The K_V1.1–K_V1.3 chimera contains the voltage-sensing domain (VSD) of K_V1.1 (S1–S4) and the pore domain (PD) of K_V1.3 (S5–P–S6), whereas the K_V1.3–K_V1.1 chimera contains the VSD of K_V1.3 and the PD of K_V1.1 (Fig. S6). Both chimeras produced robust currents (Fig. 4 A and B). Cumulative inactivation, a unique property of the K_V1.3 PD, was exhibited by the K_V1.1–K_V1.3 chimera but not by the K_V1.3–K_V1.1 chimera (Fig. 4 G and H), indicating that both chimeras function as predicted.

Fig. 4.

Activation kinetics (A and B), voltage dependence of activation (D and E; mean ± SEM), cumulative inactivation (G and H), and deactivation of K_V1.1–K_V1.3 and the K_V1.3–K_V1.1 chimeras (J and K), expressed alone (black) or with Na_Vβ1 (red) in Xenopus oocytes. Schematic showing interaction between Na_Vβ1 and domains within the channel chimeras (C, F, I, and L). Current traces shown are averages of normalized currents of selective sample currents of comparable amplitudes. Side view (M) and view from the extracellular side of the membrane (N) of the model of the complex of K_V1.2 with Na_Vβ1. The channel monomers are colored in four different shades of blue, Na_Vβ1 in tan, and residues that are lipid exposed on the S1 and S5 segments and that differ in character to equivalent residues in other K_V channel families are displayed as green and orange spheres, respectively.

The VSD of K_V1.3 is required for Na_Vβ1-mediated acceleration of K_V1.3 activation because the K_V1.3–K_V1.1 chimera (containing K_V1.3’s VSD) exhibited faster activation in the presence of Na_Vβ1, whereas the activation of the K_V1.1–K_V1.3 chimera (containing K_V1.3’s PD) was unaffected by Na_Vβ1 (Fig. 4C, Table S1). The PD of K_V1.3 is required for Na_Vβ1-mediated reduction of K_V1.3 cumulative inactivation because Na_Vβ1 reduced cumulative inactivation of the K_V1.1–K_V1.3 chimera (containing K_V1.3’s PD), but not the K_V1.3–K_V1.1 chimera (containing K_V1.3’s VSD) (Fig. 4 G and H).

The VSD of K_V1.1 is required for the Na_Vβ1-mediated change in the voltage dependence of K_V1.1 activation because the K_V1.1–K_V1.3 chimera (which contains K_V1.1’s VSD), but not the K_V1.3–K_V1.1 chimera (which contains K_V1.1’s PD), exhibited a hyperpolarizing shift in the voltage dependence of activation in the presence of Na_Vβ1 (Fig. 4 D and E, Table S1). The PD of K_V1.1 is required for Na_Vβ1-mediated slowing of K_V1.1 deactivation because the K_V1.3–K_V1.1 chimera (K_V1.1’s PD), but not the K_V1.1–K_V1.3 chimera (K_V1.1’s VSD), showed slower deactivation in the presence of Na_Vβ1 (Fig. 4 J and K).

To examine whether the interaction between Na_Vβ1 and the K_V1.3's PD sterically hinders toxin access to the external channel vestibule, we measured the affinity of the channel for ShK-186, a K_V1.3-selective peptide inhibitor (29). ShK-186 blocked K_V1.3 in a dose-dependent manner and its potency did not change in the presence of Na_Vβ1 (Fig. S7). This result suggests that the Na_Vβ1–K_V1.3 interaction does not alter toxin access to the channel vestibule. Overall, these results demonstrate that Na_Vβ1 accelerates K_V1.3 activation by interacting with the K_V1.3's VSD, affects cumulative inactivation of K_V1.3 through an interaction with K_V1.3’s PD, induces a hyperpolarizing shift in the voltage dependence of activation of K_V1.1 via an interaction with the K_V1.1's VSD, and slows deactivation of K_V1.1 via an interaction with K_V1.1’s PD.

Molecular Modeling.

Docking of full-length Na_Vβ1 with the K_V1.2 homotetramer resulted in a model in which the transmembrane segment of Na_Vβ1 lies in the groove between the VSD and PD of K_V1.2, making significant contacts with the channel’s S1 and S5 segments (Fig. 4 M and N). An alignment of the S5 segments shows several key differences between Na_Vβ1-resistant K_V3.1 and Na_Vβ1-modulated channels (Fig. S8A). The transmembrane segment of Na_Vβ1 makes contact with many of the key S5 residues. Several highly conserved residues on Na_Vβ1 make contact with the channel. Leu13 and Leu17 of Na_Vβ1 (transmembrane numbering; Fig. S8B) form a pocket into which Phe25 of the K_V1.2-S5 segment binds. Trp20 of Na_Vβ1 packs against Gly18 of the S5 segment, and Glu24 of Na_Vβ1 forms hydrogen bonds with the uncapped N-terminal main-chain amides of the slide helix (Fig. S8A). Residues of the S1 segment predicted to interact with Na_Vβ1 include Ala3, Val7, Leu11, Val15, and Leu19 (S1 numbering; Fig. S8C). Val15’s side chain packs against the side chain of Tyr11 of the transmembrane segment of Na_Vβ1. In the S1 segment, K_V3 channels differ significantly from K_V1 channels at position 15 (V/T), from K_v4 channels at positions 3 (Y/A), 7 (G/L), 11 (A/L), and 15 (I/T), and from K_v7 channels at positions 3 (Y/A) and 15 (L/T). The Ig domain of Na_Vβ1 forms extensive interactions with the S1–S2, S5–P, and P–S6 extracellular loops of the channel (Fig. 4 M and N).

Discussion

The precision of neuronal action potential generation and propagation is controlled by the sequential activation and inactivation of Na_V and K_V channels. In the CNS, four Na_V α subunits (Na_V1.1–Na_V1.3, Na_V1.6) in tight complex with Na_Vβ1–Na_Vβ4 subunits underlie the depolarization phase of the action potential. Several subfamilies of K_V channels are important contributors to action potential repolarization (1–4). Na_Vβ1, although considered to be a subunit of Na_V channels, was recently shown to interact with and modulate the function of K_V4 channels in the heart and brain (13–15). Here, we demonstrate that Na_Vβ1 is a promiscuous protein that can also interact with and modulate the properties of K_V1 (K_V1.1, K_V1.2, K_V1.3, K_V1.6) and K_V7 (K_V7.2) channels. Of the channels tested, only K_V3.1 was not affected by Na_Vβ1. Modulation occurs in a channel isoform-specific manner. Through structure–function approaches, we identify regions of Na_Vβ1 and K_V channels required for Na_Vβ1-dependent modulation, and we provide a model of the Na_Vβ1 and K_V channel complex.

Na_Vβ1 coassembles with K_V1.2, speeds up its activation, shifts its voltage dependence of activation by 4 mV in the hyperpolarized direction, and slows its deactivation in mammalian cells and in Xenopus oocytes. Na_Vβ1 and K_V1.2 colocalize in the axon initial segment in the mouse cerebral cortex where the interaction between the two proteins may affect neuronal excitability. Na_Vβ1 shifts the voltage dependence of activation of K_V1.1 in the hyperpolarized direction, slows deactivation (τ_fast), and has no effect on activation kinetics. Na_Vβ1 accelerates activation of K_V1.3, abolishes cumulative inactivation (accelerates recovery from C-type inactivation) and has no effect on its voltage dependence of activation or deactivation kinetics. Na_Vβ1 slows K_V1.6 activation, shifts its voltage dependence of activation in a depolarized direction, and significantly increases the amplitude of the K_V1.6 tail current without affecting its deactivation kinetics. Another K_V channel found in the axon initial segment in the brain, K_V7.2, coassembles with Na_Vβ1 in mammalian cells and its activation is slowed at moderate depolarizing potentials. However, Na_Vβ1 does not alter activation or deactivation kinetics, or the voltage dependence of activation of the K_V3.1 channel.

We used chimeras of Na_Vβ1 and P₀ to define regions in Na_Vβ1 responsible for K_V channel modulation. The Na_Vβ1 domains required for channel modulation varied in an isoform-specific manner. The external Ig domain of Na_Vβ1 is solely responsible for modulation of K_V1.3, whereas the entire Na_Vβ1 protein is required to modulate K_V1.2 and K_V1.1. Chimeras of K_V1.3 and K_V1.1 were used to identify channel regions required for Na_Vβ1-mediated modulation. The PD (S5–P–S6) of K_V1.3 is required for Na_Vβ1’s modulation of cumulative inactivation, whereas its VSD (S1–S4) is essential for Na_Vβ1-mediated acceleration of K_V1.3 activation. The PD of K_V1.1 is required for Na_Vβ1’s slowing of deactivation, whereas the VSD is essential for the Na_Vβ1-mediated hyperpolarizing shift in K_V1.1’s voltage dependence of activation.

The model of Na_Vβ1 docked with K_V1.2 permits an interpretation of the isoform-specific properties of the different channels at the level of sequence-specific interactions. All of the Na_Vβ1-sensitive K_V channels contain Leu at position 17 in place of a Phe in K_V3 channels (numbering based on Fig. S8A), an aromatic residue (Phe or Tyr) at position 22 in place of Ile, and an Ile at position 29 in place of Leu. The increased bulk and hydrophobic interactions of K_V3’s Phe17, and the decreased bulk of K_V3’s Ile22, and the loss of hydrophobic interactions between the side chain of Phe/Trp22 (in K_V1, K_V4, K_V7 channels) with Na_Vβ1, likely impacts the interaction between K_V3 and Na_Vβ1. Trp20 of Na_Vβ1 packs against Gly18 of the S5 segment; this residue is Leu in the K_V3 family. The increased bulk of the side chain of Leu likely impacts the interaction between the proteins. Phe25 of the S5 segment of K_V1.2 fits into a pocket formed by two Leu residues (Leu13, -17) of Na_Vβ1; the smaller Ala present in K_V3 channels may pack less well into this pocket and thereby affect binding affinity. The side chain of Val15 of the S1 segment of K_V1.2 packs against the side chain of Tyr11 of the transmembrane domain of Na_Vβ1. Substitution of the hydrophobic Val15 with Thr in K_V3 will impact the interaction. Therefore, K_V3.1’s resistance to Na_Vβ1 modulation may be because residues in its S1 and S5 hinder optimal interaction with Na_Vβ1.

In conclusion, our results provide a structural basis for a Na_Vβ1-mediated regulation of K_V1, K_V4, and K_V7 channels. By modulating CNS potassium currents, Na_Vβ1’s importance in channel gating-modulation is extended beyond its influence on Na_V currents. Our results provide a mechanism for coordinated control of two important classes of ion channels that underlie neuronal excitability.

Materials and Methods

Electrophysiology.

Whole-cell patch-clamp and two-electrode voltage-clamp techniques were used for analysis of mammalian cells and Xenopus oocytes, respectively. Details are provided in SI Materials and Methods.

Expression Plasmids, Immunohistochemistry, Coprecipitation, Western Blots, and Molecular Modeling.

Expression plasmids, immunohistochemistry, coprecipitation, Western blots, and molecular modeling are described in SI Materials and Methods.

Cell Culture and Transfections.

L929 cells stably expressing K_V1.2, K_V1.3, and K_V3.1, CHO cells, and PC12 cells were maintained in standard DMEM containing 10% (vol/vol) heat-inactivated FCS (Summit Biotechnology), 4 mM l-glutamine, 1 mM sodium pyruvate, and 500 μg/mL G418 (Calbiochem) as described in SI Materials and Methods. Transient transfections were carried out using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. After 24–30 h, transfection efficiency was assessed by fluorescence microscopy (Olympus).