Characteristics of Brain Kv1 Channels Tailored to Mimic Native Counterparts by Tandem Linkage of α Subunits

Most neuronal Kv1 channels contain Kv1.1, Kv1.2 α, and Kvβ2.1 subunits, yet the influences of their stoichiometries on properties of the (α)4(β)4variants remain undefined. cDNAs were engineered to contain 0, 1, 2, or 4 copies of Kv1.1 with the requisite number of Kv1.2 and co-expressed in mammalian cells with Kvβ2.1 to achieve “native-like” hetero-oligomers. The monomeric (Kv1.1 or 1.2), dimeric (Kv1.1–1.2 or 1.2–1.2), and tetrameric (Kv1.1-(1.2)3) constructs produced proteins ofM r ∼62,000, 120,000, and 240,000, which assembled into (α)4(β)4 complexes. Each α cRNA yielded a distinct K+ current in oocytes, with voltage dependence of activation being shifted negatively as the Kv1.1 content in tetramers was increased. Channels containing 1, 2, or 4 copies of Kv1.1 were blocked by dendrotoxin k (DTX)k with similarly high potencies, whereas Kv(1.2)4 proved nonsusceptible. Accordingly, Kv1.2/β2.1 expressed in baby hamster kidney cells failed to bind DTXk; in contrast, oligomers containing only one Kv1.1 subunit in a tetramer exhibited high affinity, with additional copies causing modest increases. Thus, one Kv1.1 subunit largely confers high affinity for DTXk, whereas channel electrophysiological properties are tailored by the content of Kv1.1 relative to Kv1.2. This notable advance could explain the diversity of symptoms of human episodic ataxia I, which is often accompanied by myokymia, due to mutated Kv1.1 being assembled in different combinations with wild-type and Kv1.2.

) constructs produced proteins of M r ϳ62,000, 120,000, and 240,000, which assembled into (␣) 4 (␤) 4 complexes. Each ␣ cRNA yielded a distinct K ؉ current in oocytes, with voltage dependence of activation being shifted negatively as the Kv1.1 content in tetramers was increased. Channels containing 1, 2, or 4 copies of Kv1.1 were blocked by dendrotoxin k (DTX) k with similarly high potencies, whereas Kv(1.2) 4 proved nonsusceptible. Accordingly, Kv1.2/␤2.1 expressed in baby hamster kidney cells failed to bind DTX k ; in contrast, oligomers containing only one Kv1.1 subunit in a tetramer exhibited high affinity, with additional copies causing modest increases. Thus, one Kv1.1 subunit largely confers high affinity for DTX k , whereas channel electrophysiological properties are tailored by the content of Kv1.1 relative to Kv1.2. This notable advance could explain the diversity of symptoms of human episodic ataxia I, which is often accompanied by myokymia, due to mutated Kv1.1 being assembled in different combinations with wild-type and Kv1.2.
Authentic Kv1 channel proteins were first identified (11) using ␣ and k/dendrotoxin (DTX) 1 ; ␣DTX is less discriminating and inhibits Kv1.2 Ͼ Kv1.1 Ͼ Kv1.6 currents, whereas DTX k is more stringent and specifically blocks Kv1.1 channels (12). Sequential immunoprecipitation, using subunit-specific antibodies, and affinity chromatography on immobilized ␣DTX and/or DTX k (13)(14)(15)(16) (14,15,17,18). These subunits are co-localized in the juxta-paranodal region of the nodes of Ranvier, as well as in the axons and terminals of cerebellar basket cells of rat brain (19). Thus far, heterologous expression of Kv1 ␣ and ␤ subunits has failed to mimic the characteristics of neuronal K ϩ currents, highlighting the need to reproduce native K ϩ channels. Additionally, certain neurological conditions are associated with changes in Kv1 channel multimers. For example, an increase in the number of ␣DTX binding sites occurs in demyelinated brain plaques from patients with multiple sclerosis (20), whereas the content of ␣DTX and DTX k acceptors is decreased in hippocampus from aging patients or those with Alzheimer's disease (21). Furthermore, several mutations in Kv1.1 are associated with human disorders (e.g. episodic ataxia I and myokymia) (1,22,23), whereas mutations in certain other genes can distort expression and localization of Kv1.1 and 1.2, thereby inducing abnormal phenotypes (e.g. mouse strains Trembler and Shiverer) (24).
Herein, ␣ subunits were tandem-linked to recreate subtypes with predefined stoichiometries for Kv1.1 and 1.2 and to quantitatively relate subunit composition to channel properties. Functional K ϩ channels containing different numbers of Kv1.1 co-assembled with Kv1.2 in the presence of Kv␤2.1 were generated using Semliki Forest virus (SFV), yielding oligomers resembling those prevalent in neurons. Electrophysiological recording of their respective K ϩ currents in oocytes and analysis of the binding of 125 I-labeled DTX k and ␣DTX to transfected mammalian cells revealed that varying proportions of Kv1.1 and 1.2 could subtly influence the biophysical and pharmacological properties of expressed channels. Such systematic profiling should allow identification of K ϩ channel counterparts in neurons and their altered phenotypes in diseases.
To prepare a Kv1.2 N -1.1 C chimera, the N-terminal region of Kv1.1 (487 bp) was replaced with the equivalent domain of Kv1.2 (475 bp). p␤ut2pA Kv1.1 and 1.2 plasmids were digested with HindIII (outside the coding region) and SacI (487 or 475 bp downstream of Kv1.1 and 1.2, respectively). The isolated larger (ϳ4500 bp from p␤ut2pA Kv1.1) and smaller (ϳ475 bp from p␤ut2pA Kv1.2) fragments were ligated to yield p␤ut2pA Kv1.2 N -1.1 C . Every construct made above was verified by restriction digestion and dideoxy DNA sequencing. For expression in the SFV system, all Kv1 p␤ut2pA constructs were subcloned into pSFV1 vector, employing HindIII and BglII sites to cut out the cDNA fragments that were blunt-ended before ligation with pSFV1 vector (10). cRNAs for each Kv1 construct (in p␤ut2pA and pSFV1) and pS-FVH1 (plasmid encoding viral packaging proteins) were prepared as described previously (10).
Electrophysiological and Biochemical Analysis of Recombinant K ϩ Channels-Oocytes were isolated from mature Xenopus laevis females (Xenopus I, Blades) and injected with Kv1.1 or 1.2 cRNA, as described by Main et al. (25). K ϩ currents were recorded after 72 h, using a two-microelectrode voltage clamp amplifier (TEC-03; NPI) as described previously (26) during voltage steps to ϩ20 mV from a holding potential of Ϫ80 mV. DTX k was applied by superfusion (at a rate of 5 ml/min), and its concentration increased cumulatively. Addition of tetraethylammonium chloride (0.1-100 mM) was made with substitution of NaCl to maintain ionic strength. A membrane fraction from the injected and noninjected oocytes was analyzed by SDS-PAGE.
Channels were expressed in BHK cells and analyzed in the native state by gel filtration, or by SDS-PAGE and Western blotting as described previously (10). Saturable binding of 125 I-labeled ␣DTX and DTX k to intact cells expressing the various constructs was measured in triplicate under established conditions (10) by rapid filtration through GF/F glass microfiber filters that had been presoaked in 0.5% (v/v) polyethyleneimine. The radioactivity associated with the washed filters was quantified by ␥-radiation counting; data presented (ϮS.E.) were analyzed using the Graph Pad software (Prism 3.0) based on a one-site model. 2) 2 , respectively, whereas the third construct would yield a channel consisting of one Kv1.1 subunit and three Kv1.2 subunits. Biochemical evidence was obtained for formation of the expected oligomers in oocytes, to validate the data subsequently obtained from electrophysiological analysis of the K ϩ currents. Injection into oocytes of cRNAs for the p␤utpA Kv1 constructs followed by Western blotting demonstrated that Kv1.1 and Kv1.2 cRNAs were translated into single subunits of M r ϳ60,000 -64,000, whereas the dimers and tetramer gave proteins with M r ϳ120,000 and M r 240,000, respectively ( Fig. 2A). A two-electrode voltage clamp was used to establish the effects of varying the ratio of Kv1.1 and Kv1.2 in expressed tetramers. Because Kv␤2.1 has no appreciable effect on activation of Kv1.1 and Kv1.2 channels (26), it was omitted. Oocytes injected with equivalent amounts of cRNAs encoding each of the constructs yielded 1-10 A K ϩ currents after 72 h, establishing that the expressed proteins were inserted into the plasmalemma as functional channels.  (Table I) (threshold ϭ Ϫ45 to Ϫ40 mV) and halfmaximal activation voltage (V1 ⁄2 ϭ Ϫ15.1 and Ϫ16.97 mV); the slope (k) values of 10.8 and 8.3, respectively, confirmed the near-identical activation kinetics. Therefore, the linker does not exert a significant influence. Kv1.1 elicited a fast-activating, slow-inactivating K ϩ current that had a more negative activation threshold (Ϫ60 to Ϫ50 mV) than Kv1.2 (Fig. 2, B and C) and V1 ⁄2 ϭ Ϫ30.8 mV; these values for the homomeric channels are sufficiently different to allow them to be distinguished reliably. The heterodimer Kv1.1-1.2 cRNA yielded a current that had properties distinct from either parent (Fig. 2, B and C); interestingly, this showed a greater resemblance to Kv1.1 than Kv1.2 current (Table I), with an activation threshold between Ϫ55 and Ϫ50 mV and a V1 ⁄2 ϭ Ϫ26.5 mV. Despite this, the Kv1.1-1.2 K ϩ current proved much less susceptible to blockade by external tetraethylammonium (IC 50 ϭ 100 mM) than Kv1.1 (IC 50 Ͻ 1 mM). A delayed rectifying K ϩ current was observed with Kv1.1-(1.2) 3 (Fig. 2B), activating at a threshold of Ϫ55 to Ϫ50 mV (Table I); also, its V1 ⁄2 fell between that for the channels made from Kv1.  (10), the N-terminal part was replaced with an analogous moiety of Kv1.2 that regulates the efficiency of assembly (see "Introduction"). This construct gave increased surface expression in BHK cells (ϳ2-fold relative to the unmodified Kv1.1), yielding a subunit of the expected M r ϳ60,000 -62,000 on immunoblotting (Fig. 3A). The dimer was twice this size and was recognized by both anti-Kv1.1 and anti-Kv1.2 for better resolution and detected as described in Fig. 1A using the antibodies specified. D, another sample was detergent solubilized and chromatographed (10). Fractions were dot-blotted onto polyvinylidene difluoride membrane, with visualization as described in Fig. 1A using anti-Kv1.1 and anti-Kv1.2 IgGs; blots of the peak fractions are shown (inset). The column was calibrated with thyroglobulin, ferritin, catalase, and ␤-amylase; the arrow indicates the common elution position of the K ϩ channels.  Table I. antibodies (Fig. 3, AϪC), confirming the presence of both ␣ subunits. Kv1.1-(1.2) 3 construct gave a protein of the expected molecular weight (240,000) that was also reactive with anti-Kv1.1 and anti-Kv1.2 antibodies (Fig. 3, B and C). Thus, the dimer and tetramer cRNAs were correctly and fully translated, without any detectable proteolytic breakdown products; importantly, the channels were correctly assembled and inserted into the plasmalemma and functional (Table II; detailed later). Direct evidence for the formation of ␣/␤ subunit oligomers was provided by the oligomeric sizes of the channels extracted from BHK cells in nondenaturing detergent obtained from gel filtration on Superose 6HR (Fig. 3D); the similar elution position for the monomer, dimer, and tetramer expressed with Kv␤2.1 gave an apparent molecular weight for the oligomer-detergent complex of ϳ515,000 (Fig. 3D, inset). Because this value is very similar to the size for Kv(1.2) 4 (␤2.1) 4 (10), it can be concluded that all the Kv1 constructs produced proteins that assembled into tetramers containing four ␣ and four ␤ subunits, as observed for neuronal K ϩ channel complexes (27 (Table II); accordingly, the K ϩ current generated in oocytes was insensitive to blockade by 100 nM DTX k (Table I). In contrast, Kv(1.1) 4 displayed high affinity for 125 I-DTX k (Fig. 4A; Table II), consistent with its K ϩ current in oocytes being inhibited by low concentrations of DTX k (Table I). An avid interaction was reaffirmed for the latter channel by the K i values of DTX k competing for the binding of 125 I-DTX k (Fig. 4G) and 125 I-␣DTX (Fig. 4I). Comparison of the binding of both toxins to Kv(1.1) 4 showed that 125 I-␣DTX displayed a 12-fold lower affinity than 125 I-DTX k (Table II). On the other hand, 125 I-␣DTX exhibited a ϳ5-fold higher affinity for Kv(1.2) 4 than Kv(1.1) 4 (Table II), in agreement with ␣DTX blocking Kv1.2 K ϩ current with greater potency than Kv1.1 or Kv1.6 (5).

Voltage Activation of K ϩ Channel Subtypes Prominent in Neurons Is Varied by Predetermining Their
One Kv1.  (Table II). Likewise, the modest changes in the K i values (Fig. 4G) for DTX k antagonizing 125 I-DTX k binding to the latter oligomers relative to that for Kv(1.1) 4 support the above deduction. These binding data were corroborated by the DTX k sensitivities of the channels when expressed in oocytes; the IC 50 values of the Kv(1.1-1.2) 2 and Kv1.1-(1.2) 3 K ϩ currents are only 2-and 9-fold lower than that for Kv(1.1) 4 (Table I). Thus, based on both toxin binding and functional blockade, one copy of Kv1.1 is adequate to bestow near maximal affinity for DTX k and susceptibility to inhibition (see "Discussion").
Moreover, behavior of the channels in binding 125 I-␣DTX gave a similar trend with no major difference in the K D or K i values upon increasing the number of copies of Kv1.2 in the (␣) 4 (␤) 4 multimers (Table II). Likewise, the similar K i values for ␣DTX antagonizing 125 I-DTX k binding to Kv(1.1-1.2) 2 and Kv1.1-(1.2) 3 (Fig. 4H) revealed that two Kv1.2 subunits are sufficient for binding ␣DTX with high affinity. Because ␣DTX has a lower affinity for Kv(1.1) 4 than DTX k , it proved significantly less potent in antagonizing 125 I-DTX k binding, but this was increased substantially (14 -34-fold) when Kv1.2 subunits were introduced to the channels (Fig. 4H). As expected, DTX k was less effective in displacing 125 I-␣DTX from Kv1.2-containing multimers than Kv(1.1) 4 (Fig. 4I).
Finally, it is noteworthy that not only can the two toxins discriminate channel subtypes, but the B max values for ␣DTX were 2-3-fold higher than for DTX k in the same batch of BHK cells (Table II; see "Discussion"). Kv1.1-, Kv1.2-, and Kv␤2.1containing oligomers predominate in brain (13,17); Kv(1.2) 4 is also present, but Kv1.1 always occurs in association with other members (14,15). Due to their abundance and functional importance (see "Introduction"), we profiled the characteristics of those with several different proportions of Kv1.1 and Kv1.2 to encompass the subtypes found in neurons. All the linked Kv1 subunits were expressed in both amphibian and mammalian cells as single, intact proteins without premature translation or degradation products. Their functionality was documented by the K ϩ currents recorded after expression in oocytes, with characteristics matching those expected. When co-expressed with Kv␤2.1, each channel co-assembled into (␣) 4 (␤) 4 complexes and was inserted correctly into the plasma membrane of BHK cells, as established from measurement of high affinity binding of DTX k and ␣DTX, which require ␣ subunits to be assembled into tetramers (28).

Increasing the Number of Kv1.1 Subunits in a Tetramer Gave Commensurate Changes in the Voltage
Dependence of Activation of the K ϩ Currents-The tetramers containing varying ratios of Kv1.1 and Kv1.2 gave slowly inactivating, outward currents with distinct voltage dependences of activation that differed from either parent. The V1 ⁄2 values were slightly skewed toward that of Kv1.1, which may be due to the effect of Kv1.1 upon activation (29,30). However, the V1 ⁄2 values also clearly reflect the influence of both parental subunits. Establishing the presence of both Kv1.1 and Kv1.2 in this way in the expressed channels constructed by tandem linkage of subunits in the oligomers afforded exploitation of the predetermined stoichiometries for the subsequent toxin block and binding studies.
A Single Kv1.1 Subunit in Tetramers Containing Kv1.2 Creates High Affinity Functional Interaction with DTX k -Both saturable binding and competition analysis using intact BHK cells and inhibition of the K ϩ currents in oocytes confirmed that one Kv1.1 subunit in an oligomer is enough to confer a high affinity interaction with DTX k and blockade of the currents. This conclusion from measurements on defined popula-tions of native-like subtypes accords with a deduction from functional studies on biochemically uncharacterized channels, namely, that a single toxin-sensitive subunit can give K ϩ currents susceptibility to DTX k homologues (6). Mutagenesis of DTX k has shown that two domains, 3 10 helix and ␤-turn, that are 14 Å apart contribute to interaction with possibly two adjacent channel subunits (31). Notably, the 3 10 helix is essential for both high affinity binding of DTX k to Kv1.1 and precluding its interaction with other subunits; in contrast, the ␤-turn seems less important for recognition of Kv1.1 (31). Hence, it was suggested that the latter region could interact with a different subunit of the K ϩ channels in synaptic membranes; in the present study, this would be another Kv1.1 or Kv1.2. Thus, the slightly lower DTX k affinity for Kv(1.1-1.2) 2 and Kv1.1-(1.2) 3 compared with Kv(1.1) 4 may be attributed to a reduction in the number of sites for interaction with the 3 10 domains and/or a decrease in the affinity of ␤-turn for the adjacent subunit (i.e. Kv1.2 instead of Kv1.1). With two copies of Kv1.1 and Kv1.2, a significant increase in DTX k affinity was not observed, possibly because of its access being restricted due to the two Kv1.1 in some of the oligomer being positioned diagonally rather than adjacently (Fig. 1B).
Notably, a single high affinity site for DTX k was observed regardless of the number of Kv1.1 subunits, whereas synaptic membranes show a high affinity and low affinity binding site (31). The retention of high affinity for DTX k by Kv1.1 channels with up to three insensitive Kv1.2 subunits provides good evidence for the lower affinity site being due to multimerization with other subunits (e.g. Kv1.4, Kv1.3, or Kv1.6) known to be associated with Kv1.1 (13)(14)(15)19) that may result due to steric hindrance. Because Kv(1.1) 4 does not exist in normal human brain (14), and a majority of native K ϩ channels containing Kv1.2 also have Kv1.1 (50%) (13,17), the latter must represent the bulk of the higher affinity DTX k sites (31) and thus could explain the overlapping location of DTX k and ␣DTX acceptors in rat brain (32).
Kv1.1 Channels Apparently Have More Sites for ␣DTX than DTX k -The B max for ␣DTX binding sites was always 2-3-fold higher than that for DTX k in the same batches of cells. This intriguing finding can possibly be explained by structural elements of DTX k , ␦DTX (96% homologous to DTX k ), and ␣DTX. Similar residues (e.g. Lys-3/6 and Lys-26) in the two domains (3 10 helix and ␤-turn), equivalent to those important for DTX k recognition of rat Kv1.1-containing channels (31,33), have also been identified from scanning mutagenesis and thermodynamic mutant cycle analysis in ␦DTX (34). Mutation of the ShaKv1.1 channel revealed that ␦DTX binds at some distance from the pore, involving Lys-3 and Arg-10 among other residues; these, with Lys-26, form a triangle whose vertices are 20 Å apart. This distance would allow interaction with adjacent subunits through Lys-3, Arg-10 (34), and/or Lys-26 (31). However, mutations in the Shaker chimeric channel did not yield evidence for a contribution of Lys-26 to the strong interaction, possibly because it binds residues not substituted that form part of the DTX acceptor site (35,36). On the other hand, residues at the N terminus of ␣DTX are most influential for Kv1 interaction (37) and are equivalent to the essential amino acids in the 3 10 helix of DTX k , although ␣DTX lacks such a secondary structural feature; this could underlie its less stringent specificity. Furthermore, the contributions of ␤-turn residues in ␣DTX are less pronounced than that for DTX k (33,37). Based on the collective findings, DTX k appears to interact with two Kv1 subunits, whereas ␣DTX requires predominantly one interactive domain; hence, twice as many ␣DTX molecules might be able to bind each tetramer, or perhaps not all the channels are folded perfectly and therefore cannot accommodate the stringent requirements for DTX k binding.
Functional Properties of the Recombinant Channels in Relation to Those of Neuronal K ϩ Currents in Health and Disease-Because Kv1.1 and 1.2 subunits exhibit distinct properties, it is not surprising that changing their ratios resulted in K ϩ currents with unique electrophysiological and pharmacological properties. The profiling of their characteristics ought to help molecular entities to be ascribed to the native K ϩ channels, a feat not feasible to date. Thus, properties of the recombinant channels were compared with the two types of DTX-sensitive, sustained K ϩ currents recorded in neuronal cells. Our data for Kv(1.1-1.2) 2 and Kv1.1-(1.2) 3 indicate that they resemble a DTX-sensitive, low-threshold current (I DS ) in various neurons that activates within Ϫ50 to Ϫ60 mV and prevents repetitive firing (38,39). Because Kv1.1 and Kv1.2 are the major subunits known to give sustained outward K ϩ currents that are highly sensitive to DTX, their combinations must be responsible for such current phenotypes. Gold et al. (40) have described a similar current (I Kit ) in rat sensory neurons that may relate to a DTX k -susceptible current found in the same preparation (41); as with the currents we observed, I Kit activates at low thresholds and is fully activated by ϩ20 mV. From the results herein, it seems that one Kv1.1 subunit in an oligomer with Kv1.2 would be sufficient to give such voltage sensitivity and DTX k susceptibility; moreover, the slowly inactivating nature of the currents excludes the presence of Kv1.4 and thereby implicates either Kv1.1/1.2 or Kv1.1/1.2/1.6 (15).
These properties derived from channels encompassing most combinations of Kv1.1/1.2 give new insights into the molecular basis of the symptoms seen in patients suffering from episodic ataxia I/myokymia (22,23). Because we demonstrated herein that these channels' biophysical parameters show gradual changes proportional to their content of Kv1.1 subunits, and a variety of human mutations are known to alter the properties of Kv1.1 homomers (23), the diversity of abnormalities in different cases supports the existence of several hetero-oligomeric combinations of mutated and wild-type Kv1.1, together with Kv1.2; note that Kv(1.1) 4 does not occur in human neurons (14). Such subtypes would exist in different locations, neurons, or compartments (e.g. nerve terminals, axons, and so forth) where each normally serves a pivotal role; thus, a spectrum of abnormalities in individual patients is likely to be due to different stoichiometries of Kv1.2, Kv1.1, and a variant that could be mutated at one of several residues (22,23). Importantly, the major advance accomplished herein will allow elucidation of the modified properties of all these channel forms constructed by tandem linking Kv1.1, its distinct mutants, and Kv1.2 in the various stoichiometries, followed by co-expression with Kv␤2.1. Although this strategy has already been found to be informative for dimers of normal and mutated Kv1.1 (42), creating the authentic ␣/␤ combinations should prove much more pertinent.