Coexpression of the KCNA3B gene product with Kv1.5 leads to a novel A-type potassium channel.

Shaker-related voltage-gated potassium (Kv) channels may be heterooligomers consisting of membrane-integral alpha-subunits associated with auxiliary cytoplasmic beta-subunits. In this study we have cloned the human Kvbeta3.1 subunit and the corresponding KCNA3B gene. Identification of sequence-tagged sites in the gene mapped KCNA3B to band p13.1 of human chromosome 17. Comparison of the KCNA1B, KCNA2B, and KCNA3B gene structures showed that the three Kvbeta genes have very disparate lengths varying from >/=350 kb (KCNA1B) to approximately 7 kb (KCNA3B). Yet, the exon patterns of the three genes, which code for the seven known mammalian Kvbeta subunits, are very similar. The Kvbeta1 and Kvbeta2 splice variants are generated by alternative use of 5'-exons. Mouse Kvbeta4, a potential splice variant of Kvbeta3, is a read-through product where the open reading frame starts within the sequence intervening between Kvbeta3 exons 7 and 8. The human KCNA3B sequence does not contain a mouse Kvbeta4-like open reading frame. Human Kvbeta3 mRNA is specifically expressed in the brain, where it is predominantly detected in the cerebellum. The heterologous coexpression of human Kv1.5 and Kvbeta3.1 subunits in Chinese hamster ovary cells yielded a novel Kv channel mediating very fast inactivating (A-type) outward currents upon depolarization. Thus, the expression of Kvbeta3.1 subunits potentially extends the possibilities to express diverse A-type Kv channels in the human brain.

tured herring testis DNA at 42°C with a rat Kv␤3.1 cDNA probe (nucleotides 658 -1198; accession no. X76723) labeled by random priming (29) with [␣-32 P]dCTP. Filters were washed in 1ϫ SET, 0.1% SDS at 65°C. Exposure to x-ray films (Kodak BioMax MS) was overnight with an amplifying screen (Cronex Hi ϩ , DuPont), at Ϫ70°C. Eight positive phages were isolated with hKv␤3 cDNA inserts varying in length between 0.9 and 1.9 kb. cDNA inserts were cloned into the EcoRI site of pBluescript KS Ϫ (Stratagene). Both strands were sequenced using the dideoxy chain termination method (30) and a T7 sequencing kit from Amersham Pharmacia Biotech. Genetics Computer Group Inc. (GCG, Madison, WI) software was used for sequence analysis. The sequence of the open reading frame (ORF) has been submitted to the National Center for Biotechnology Information (NCBI) data bank (accession no. AF016411).
Isolation and Sequence Analysis of Genomic DNA KCNA2B Gene-hKv␤2-specific probes were generated using specific PCR fragments of the coding and noncoding region. The KCNA2Bspecific primer sequences are available on request (31). A human genomic library in EMBL3 SP6/T7 (CLONTECH) was hybridized with [␣-32 P]dCTP-labeled hKv␤2 probes as described (see above and Ref. 23). Twenty-five positive phages were isolated with KCNA2B DNA inserts varying in length between 14 and 20 kb. Restriction site mapping of phage inserts was carried out using standard protocols. Subcloned restriction fragments containing exon sequences were sequenced for genomic and cDNA sequence alignments. The 5Ј-termini of phage inserts 3,18,21,24,26,62,69, and the 3Ј-termini of phage inserts 18,21,23,24,62,64, and 69 were used as probes to construct the contiguous KCNA2B map shown in Fig. 1. A detailed map is available on request (31). Genomic and cDNA sequence alignments were done with GCG software.
KCNA3B Gene-The EMBL3 SP6/T7 genomic DNA library was hybridized with a [␣-32 P]dCTP-labeled hKv␤3 probe (nucleotides 1-1215). Hybridization conditions were as above. One positive phage was isolated containing the complete KCNA3B gene. The insert DNA (11.85 kb) was subcloned and completely sequenced using the dideoxy chain termination method (31). The sequence was analyzed employing GCG software and the electronic PCR program of the NCBI data bank. 2

Chromosomal Localization
DNA panels from human/rodent hybrid somatic cell lines were purchased from BIOS Laboratories (Scotlab GmbH, Wiesloch, Germany). PCR and chromosomal localization studies were performed using primers B3-107 (ACCCAAAGTGAGGGGATATG) and B3-208 (CTGCTTT-TACCCTCCATAGG). The primers amplify a 620-bp product from human genomic DNA, corresponding to nucleotides 8327-8947 of the KCNA3B gene. The PCRs were carried out with 100 ng of genomic DNA as template, 20 pM of each primer and 1.5 units of Taq polymerase (Life Technologies, Inc.) in buffer containing 10 mM Tris-HCl (pH 8.3), an 80 M concentration of each dNTP, and 1 mM MgCl 2 . Amplification was carried out for 35 cycles (each cycle consisting of 94°C for 1 min; 60°C for 30 s; 72°C for 30 s).

Northern Blot Analysis
Northern blots of human brain poly(A ϩ ) mRNA were obtained from CLONTECH. The blots were hybridized according to the manufactur-erЈs protocol with a [␣-32 P]dCTP-labeled hKv␤3.1 cDNA probe (nucleotides 1-550 or 912-1162). Control hybridizations were done with an actin probe supplied by the manufacturer. After three washes with 0.5ϫ SET, 0.1% SDS at 65°C, membranes were autoradiographed. Exposure to x-ray films (Kodak BioMax MS) was for 72 h with an amplifying screen (Cronex Hi ϩ , DuPont) at Ϫ70°C.

Channel Expression and Functional Characterization
Both hKv␣1.5 and hKv␤3.1 cDNAs were cloned into the eucaryotic expression vector pcDNA3 (Invitrogen). Recombinant Kv␣1.5 cDNA (10 ng/l) alone or together with Kv␤3.1 cDNA (200 ng/l) was microinjected into CHO cells with green fluorescent protein cDNA (5 ng/l) as a reporter gene to allow the identification of injected cells during electrophysiological experiments. A 20-fold excess of ␤-subunit cDNA was used to favor high expression, which should allow maximal ␣␤ interaction. Currents activated by depolarizing voltage pulses in the outsideout patch configuration were obtained 12 h after DNA injection. Patch pipettes were pulled from borosilicate capillaries using a DMZ puller (Zeitz, Augsburg, Germany) and had bath resistances between 3 and 4 M⍀ when filled with an intracellular solution containing 115 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 11 mM EGTA, 10 mM HEPES, 2 mM Na 2 ATP, and 2 mM glutathione, pH 7.2, KOH. The bath solution contained 135 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 5 mM HEPES, 10 mM glucose, 20 mM sucrose, pH 7.4, NaOH. Currents were recorded with an EPC9 patch clamp amplifier (HEKA Elektronik, Landbrecht, Germany), and the program package PULSE ϩ PULSEFIT (HEKA Elektronik) was used for data acquisition and analysis. Activation and inactivation kinetics were fitted simultaneously with a Hodgkin-Huxley-related formalism (PULSEFIT) as described previously (32). Voltagedependent processes were fitted with Boltzmann functions of the form G/G max ϭ (1/1 ϩ exp((V m Ϫ V1 ⁄2 )/k)) n . For a first power Boltzmann function (n ϭ 1), which was used to fit steady-state inactivation, V1 ⁄2 is the potential of half-maximal inactivation. For a fourth power Boltzmann function (n ϭ 4), which was used to fit channel activation, V1 ⁄2 is the potential at which each of the four subunits is activated half of the time. The latter V1 ⁄2 value yielded an estimate of the activation threshold (0.5 4 ϭ 6.25% of the maximal activation). The steepness of voltage dependences is described by the slope factor k. Data are given as mean Ϯ S.E. All experiments were performed at room temperature.

RESULTS
Cloning of Human Voltage-gated Potassium Channel Kv␤3.1 cDNA-To isolate human Kv␤3.1 cDNA, we prepared a 32 Plabeled rat Kv␤3 cDNA probe for screening a human brain cDNA library (see "Experimental Procedures"). Eight phages were isolated, and the inserts were subcloned and sequenced. One insert of 1857 bp contained the complete Kv␤3.1 ORF. The derived human Kv␤3.1 protein sequence has a length of 404 amino acids (Fig. 1).
The hKv␤3.1 sequence shares 92.3% identity with the previously published rat Kv␤3 protein sequence (17). This contrasts with an identity of 74.6% to the human Kv␤2.1 and 72.4% to the human Kv␤1.1 protein sequence, respectively. Alignment of hKv␤1.1, hKv␤2.1, and hKv␤3.1 sequences showed that the amino termini are highly divergent (Fig. 1). The remaining carboxyl-terminal Kv␤ sequences are conserved and share a sequence identity of 85%. Accordingly, several consensus sequences for the possible phosphorylation by protein kinases have been conserved between the different hKv␤ protein sequences ( Fig. 1). Remarkably, only one of the 10 consensus sites is a protein kinase A phosphorylation site.
Similar Exon/Intron Structure for Kv␤1, Kv␤2, and Kv␤3 Genes-Previously, we have cloned the human Kv␤1 (KCNA1B) gene (23). A comparison of the genomic DNA sequence with human Kv␤1 cDNA showed that the KCNA1B gene has an exceptionally large size of Ͼ380 kb. The Kv␤1 ORFs are composed of 14 exons, and the first exon(s) (1.1, 1.2, and 1.3 in Fig. 2A) encode the alternative Kv␤1 amino termini. Now, we have cloned the human Kv␤2 (KCNA2B) and Kv␤3 (KCNA3B) genes. A 32 P-labeled human hKv␤2.1 cDNA (21) was used to probe a human genomic -phage library. In order to construct a physical map of 91-kb DNA containing the KCNA2B gene, 25 positive partially overlapping phages were isolated (Fig. 2B). A comparison of the genomic sequence with the hKv␤2.1 cDNA revealed 15 exons (spread out over ϳ70-kb DNA. The last exon of the KCNA2B ORF contained an additional 1800 bp of 3Ј-untranslated sequence. The size of the intervening introns varied between 236 bp and 33 kb. All exon/intron borders followed the GT/AG rule (33). In contrast to Kv␤1 amino termini, the corresponding hKv␤2 amino terminus is encoded in two exons (1a and 1b in Fig. 2A). Skipping of exon 1b may give rise to the hKv␤2.2 splice variant, which has been discovered recently in human lens epithelium (accession no. AF044253).
To characterize the KCNA3B gene we probed a human genomic -library with 32 P-labeled hKv␤3.1 cDNA. One phage was isolated, containing an 11.8-kb insert. It encoded apparently the complete KCNA3B gene (Fig. 2B). In comparison, the KCNA3B gene is approximately 50 times smaller than the KCNA1B gene and approximately 10 times smaller than the KCNA2B gene. Alignment of the genomic KCNA3B DNA and hKv␤3.1 cDNA sequences showed that the KCNA3B gene comprised 14 exons (Fig. 1), comparable in size and structure with those of the KCNA1B and KCNA2B genes ( Fig. 2A). All exon/ intron borders followed the GT/AG rule (33). The last exon contained a 3Ј-untranslated sequence with two consensus polyadenylation signals (AATAAA) 752 and 760 bp downstream of the ORF stop codon, respectively. The results demonstrate that the different hKv␤ genes contain very similar exon patterns (Figs. 1 and 2). Thus, the greatly varying size of the Kv␤ genes, ranging from Ͼ380 kb (KCNA1B) to ϳ7 kb (KCNA3B), is largely due to differences in the sizes of the intervening introns; the KCNA1B gene contains some introns of Ն50kb, whereas introns of the KCNA3B gene have sizes of Յ1.4 kb (Fig. 2B).
KCNA3B Located on Chromosome 17p13.1-The genomic KCNA3B sequence information was used to choose a primer pair 5Ј and 3Ј to exon 8. The primer pair specifically amplified a 620-bp KCNA3B DNA fragment in PCRs using human genomic DNA as template but not in PCRs with mouse or hamster genomic DNA as template. In PCRs with a DNA panel of human/mouse/hamster hybrid cell lines, the 620-bp KCNA3B DNA fragment was only amplified with template DNA derived from hybrid DNA cell lines containing human chromosome 17 (not shown). A search in the NCBI dbSTS data bank revealed the two identical sequence-tagged sites (STS) SHGC 31899 and dbSTS 38881 (Fig. 2B) 50 bp upstream of the first polyadenylation site of the KCNA3B sequence. They have been mapped to band p13.1 on the short arm of human chromosome 17. Thus, the KCNA3B gene was localized between the microsatellite markers D17S786 and D17S960 (Ref. 34 ; Fig. 3).
The microsatellite markers are located near the genes for the postsynaptic density protein 95 (34) and retinal guanylate cy-clase (35). Mutations in the retinal guanylate cyclase gene cause a form of Leber congenital amaurosis (36) and an autosomally inherited dominant form of progressive cone dystrophy (CORD6; Ref. 37). Two-point analyses in a five-generation Swedish family carrying the CORD5 syndrome demonstrated linkage to the markers D17S786 and D17S960 with maximum logarithmic odds ratio scores of 6.5 and 4.4, respectively (39). Since these markers flanked the KCNA3B gene, we analyzed DNA obtained from Swedish families with CORD5-affected members. The results did not reveal mutations in the KCNA3B exon sequences. Apparently, CORD5 is not linked to the KCNA3B gene.
Tissue-specific Expression of hKv␤3.1 mRNA-The distribution of Kv␤3.1 mRNA in human tissue and in various areas of human brain was investigated in Northern blot analysis. With a Kv␤3.1 cDNA-specific probe, we detected two mRNAs of 2.8 and 4.4 kb in brain but not in heart, lung, liver, kidney, pancreas, placenta, and skeletal muscle (Fig. 4A). The exon structure of the KCNA3B gene predicts a ϳ4.4-kb-long transcript. The 2.8-kb transcript may correspond to an as yet undetected Kv␤3 splice variant comparable with mouse Kv␤4 (Ref. 14; see "Discussion").
The expression of hKv␤3.1 mRNA in human brain was analyzed in more detail using Northern blots containing mRNA derived from several brain areas. We obtained a strong hybridization signal for Kv␤3.1 mRNA with poly(A ϩ ) RNA of human cerebellum, weaker signals with the ones of cortex, occipital lobe, frontal lobe, and temporal lobe (Fig. 4B). No significant hybridization signals were obtained with poly(A ϩ ) RNA of medulla and spinal cord (Fig. 4B)  of Kv␤3.1 in human and rat brain (17) except for thalamic nuclei.
Functional Expression of hKv␤3.1-We have previously shown that hKv␤1.1 subunits confer rapid inactivation to hKv1.5 channels when coexpressed in mammalian tissue culture cells (23). In the present study, we tested a possible interaction between hKv1.5 and hKv␤3.1 subunits. The injection of hKv1.5 cDNA alone in CHO cells led to the expression of channels mediating typical outward currents with fast activation in response to depolarizing voltage pulses (Ref. 40; Fig.  5A). Kv1.5-mediated currents inactivated slowly, resulting in 70.9 Ϯ 1.6% (n ϭ 6) of the peak current at the end of a 1-s pulse to ϩ60 mV. Inactivation kinetics were best described by two exponentials with time constants 1 ϭ 70.0 Ϯ 14.0 ms (28.1 Ϯ 4.3% of the total decay) and 2 ϭ 1.5 Ϯ 0.3 s (n ϭ 5).
When we coexpressed hKv␤3.1 with hKv1.5, rapidly inactivating potassium currents were elicited upon depolarization (Fig. 5A). With equimolar injection of Kv␣ and Kv␤ subunit cDNA (10 ng/l of each), we obtained currents with rapid but largely incomplete decay (data not shown). Apparently, in these experiments the steady-state current component represented a distinct population of noninactivating channels rather than equilibrium conditions of a single rapidly inactivating channel population. The current decay at ϩ60 mV was virtually complete (2.3 Ϯ 0.3% of the peak current at the end of a 1-s pulse) in 25 out of 29 patches when the amount of injected Kv␤ cDNA was increased 20-fold (Fig. 5A). The inactivation kinetics of hKv1.5 channels in combination with hKv␤3.1 subunits were best described by two exponentials with 1 ϭ 5.8 Ϯ 0.4 ms (87.6 Ϯ 2.1% of the total decay) and 2 ϭ 32.2 Ϯ 4.2 ms (n ϭ 25). The steady-state inactivation due to the presence of hKv␤3.1 showed a steep voltage dependence with V1 ⁄2 ϭ Ϫ29.0 Ϯ 1.1 mV and k ϭ 2.7 Ϯ 0.2 mV (n ϭ 5; Fig. 5B). The recovery from inactivation measured at Ϫ80 mV showed a fast component with 1 ϭ 0.8 Ϯ 0.2 s, accounting for 48.9 Ϯ 11.7% of the complete recovery followed by a slower component with 2 ϭ 4.6 Ϯ 0.9 s (n ϭ 4; Fig. 5C).
Many investigators have described a hyperpolarizing shift in the voltage dependence of channel activation, when Kv␣ subunits were coexpressed with Kv␤ subunits (24 -26, 41, 42). We tested the voltage dependence of hKv1.5 channel activation using a pulse protocol with a holding potential of Ϫ80 mV and test potentials between Ϫ50 and ϩ100 mV. For the expression of hKv1.5 alone, a fourth power Boltzmann-function (see "Experimental Procedures") was fit to the conductance-voltage relationship ( Fig. 5D; open circles)  V1 ⁄2 ϭ Ϫ31.7 Ϯ 2.5 mV and a slope factor k ϭ 28.3 Ϯ 1.5 mV (n ϭ 5) if conductance voltage plots were derived from the measured peak amplitudes (Fig. 5D, filled triangles). However, the current peaks may not reliably reflect steady-state channel activation at a given membrane potential, because rapid inactivation occurs already during the activation phase, leading to an underestimation of the maximal activation. Therefore, we used simulated I max values based on a model with no inactivation (Hodgkin-Huxley-related formalism with h ϭ 4 and m ϭ 0) to create conductance-voltage plots (Fig. 5D, inset and filled circles). The values obtained with Boltzmann analysis were V1 ⁄2 ϭ Ϫ31.9 Ϯ 2.1 mV and k ϭ 20.0 Ϯ 2.3 mV (n ϭ 5), suggesting no significant alteration in the voltage dependence of hKv1.5 channel activation by the presence of hKv␤3.1 subunits (p ϭ 0.8518; two-tailed t test). DISCUSSION This paper presents a comparative analysis of the structure of human genes encoding Kv␤ subunits and the functional expression of a new human Kv␤ subunit.
Kv␤ Gene Structures-The Kv␤ genes map to different chromosomes. KCNA1B is located on chromosome 3q25.1 (23), KCNA2B on 1p36 (43), and KCNA3B on 17p13.1 (this paper). Despite their different chromosomal locations, the KCNAB genes have a similar exon/intron pattern. This pattern reflects the conserved structure of Kv␤ proteins. They have differing amino termini, varying in length from 26 to 91 amino acids and a highly conserved carboxyl terminus of ϳ325 amino acids. The variant Kv␤1.1, -1.2, -1.3, and -3.1 amino termini are encoded each in a distinct exon. Thus, the different Kv␤1 subunits are produced by alternative splicing of exons 1. By contrast, the alternative Kv␤2 amino termini may be generated by an exonskipping mechanism, leaving out exon 1b (Fig. 2A). The 5Јterminal exons are spliced to a Kv␤ "core" region. It invariably consists of 13 exons that are highly homologous among the three genes. In contrast to the exon pattern, the KCNA1B, KCNA2B, and KCNA3B genes contain intervening sequences, which vary markedly in length. Accordingly, the KCNA1B gene is ϳ50 times larger and the KCNA2B gene ϳ10 times larger than the KCNA3B gene. The similarity of the exon/intron patterns, despite the greatly varying sizes of the different genes, suggests that the three human Kv␤ genes were derived from an ancestral precursor gene. The data indicate that unlike Kv␣ subunit genes (44,45) Kv␤ subunit genes are not clustered in the human genome. Most likely, Kv␤ subunit diversity has evolved from chromosomal rearrangements rather than from local gene duplications.
Another Kv␤ genomic structure has been reported for Drosophila Hk. The derived Hk protein sequence of 546 amino acids indicates that this Kv␤ subunit is considerably larger than the human Kv␤ subunits. Yet, the Hk ORF comprises six instead of 14 exons (12). Due to the lack of detailed sequence information, it is not clear how Hk exons relate to hKv␤ exons. Nevertheless, the nonconserved exon/intron structures between Drosophila and humans suggest that they have arisen in evolution after the separation of vertebrates and invertebrates. Interestingly, the exon/intron structures of Drosophila Shaker and human Kv␣ subunit genes (2, 3) are not similar either, despite the highly conserved protein sequences of Shaker-related Kv␣ subunits (2).
Kv␤3.1 mRNA Expression-The hKv␤3.1 cDNA probes hybridized to two mRNA species (4.4 and 2.8 kb in length) in the Northern blot experiments. The longer transcript is consistent with our genomic sequence analysis of the KCNA3B gene, which predicts a hKv␤3.1 mRNA length of 3868 nucleotides without the poly(A) tail. The shorter transcript may be a crosshybridizing mRNA species or an alternatively spliced KCNA3B transcript. This notion may be supported by a recent report describing a Kv␤4 subunit in mice (14). A comparison of the derived hKv␤3.1 and mouse Kv␤4 protein sequences revealed that both differ in their amino termini, whereas the carboxyl termini consisting of 225 amino acid residues are nearly identical, sharing an identity of 95%. The hKv␤3.1/mKv␤4 sequence identity starts exactly with the first amino acid of KCNA3B exon 8. Analysis of the mouse Kv␤3 gene showed that the mouse Kv␤4 amino terminus is encoded in the intron 7 sequence preceding mouse exon 8 (Fig. 6). This indicates that mKv␤4 is translated from a mouse Kv␤3 mRNA variant where intron 7 is not spliced out. The mouse and human intron 7 sequences have a similarity of 70%. However, an Alu element is inserted into human intron 7 DNA (Fig. 6). Therefore, the KCNA3B cannot code for a Kv␤3 variant that is analogous to mKv␤4.
Rapid Kv Channel Inactivation Conferred by Kv␤3.1-Kv␤1 and Kv␤2 subunits can assemble with Shaker-related Kv␣ subunits both in vitro (9,10,16) and in vivo (46). This assembly may enhance the surface expression of Kv1␣ subunits, leading to an increased number of Kv channels in the plasma membrane (15). In addition, Kv␤1 but not Kv␤2 subunits may affect the inactivating properties of Kv1␣ subunits (18 -20, 47). Kv␤1.1 contains an N-terminal inactivating domain, which confers rapid N-type inactivation to otherwise noninactivating channels of the Kv1 family (16). The Kv␤1.1 N-terminal inactivating domain is similar in structure and function to the ones of Shaker channels (16). They inactivate the channel by binding to a receptor at or near the inner pore entrance (28), resembling a tethered ball and chain type mechanism (27). The Kv␤3.1 amino terminus also has the characteristic features of N-terminal inactivating domains (17). Coexpression of rat Kv␤3 in Xenopus oocytes imposed rapid inactivation on Kv1.4⌬1-110 channels, and the sensitivity of this inactivation to intracellular redox potential and its kinetics were comparable with the inactivation conferred on Kv1.4⌬1-110 channels by Kv␤1.1 (17). However, it was not possible to detect in the Xenopus oocyte expression system functional assemblies of Kv␤3 with Kv1.1 or Kv1.5 channels (26). In the present study, we show that hKv␤3.1 subunits confer rapid inactivation to hKv1.5 channels upon coexpression in CHO cells. From a kinetic point of view, the Kv1.5/Kv␤3.1-mediated A-type channel shows the fastest inactivation of all cloned Kv channels that have been heterologously expressed so far. The time course and the completeness of inactivation as well as the parameters for both steady-state inactivation and recovery from inactivation obtained in our experiments are in excellent agreement with shaded. An asterisk marks the first ATG of the mouse Kv␤4 ORF (14). Identical nucleotides are indicated by dashes. Dots indicate gaps introduced for optimal sequence alignment. Alu repeat sequence is in italic letters.
FIG. 5. Properties of hKv␣1.5 channels coexpressed with hKv␤3.1 in CHO cells. A, outward currents during voltage pulses of 1 s to ϩ60 mV from a holding potential of Ϫ80 mV recorded in outside-out patches excised from two different CHO cells. One cell was injected with hKv␣1.5 cDNA alone (slow and incomplete decay), and in the other cell hKv␣1.5 and hKv␤3.1 cDNAs were coinjected (rapid and virtually complete decay). Two current traces with equal peak amplitudes were chosen for illustrative purposes. B, voltage dependence of steady-state inactivation. Data points represent current amplitudes normalized to the maximal peak obtained at ϩ60 mV following a prepulse of 500-ms duration to potentials between Ϫ50 and ϩ10 mV in 5-mV steps from a holding potential of Ϫ80 mV. The curve fitted to the data points represents a first power Boltzmann function. Mean values for V1 ⁄2 and k as indicated. C, recovery from inactivation measured with a set of twin pulses to ϩ60 mV separated by intervals of different length at Ϫ80 mV. The line fitted to the data points represents a double-exponential function with the mean values for 1 and 2 as indicated. D, voltage-dependent activation of hKv1.5 channels in the absence (open circles) and presence of hKv␤3.1 (filled triangles). Conductancevoltage plots were obtained from the peak amplitudes measured during 200-ms pulses to potentials between Ϫ50 and ϩ100 mV in 10-mV increments from a holding potential of Ϫ80 mV. Conductance values were normalized to the value obtained at ϩ100 mV. Open circles represent conductance values obtained by using a calculated I max not accounting for inactivation. The inset illustrates fitting of a Hodgkin-Huxley-related formalism to a recorded A-type current trace (only 25% of acquired data points are shown for clarity). I max was with the obtained parameters for activation calculated by setting m ϭ 0. This eliminates inactivation from the model. Curves fitted to the conductance-voltage plots represent fourth power Boltzmann-functions. Mean values for the respective parameters were as follows: V1 ⁄2 ϭ Ϫ32.3 mV and k ϭ 21.0 (hKv1.5); V1 ⁄2 ϭ Ϫ31.7 mV and k ϭ 28.3 (hKv1.5 ϩ hKv␤3.1); V1 ⁄2 ϭ Ϫ31.9 mV and k ϭ 20.0 (from simulated I max values). Note that no apparent shift in activation threshold was observed irrespective of the method of analysis. previously published data for Kv1.5 channels coexpressed with an N(Kv␤3)-C(Kv␤2) chimeric subunit in Xenopus oocytes (26). These findings indicate comparable mechanisms for the Kv␤3.1 ball-mediated N-type inactivation observed in the CHO cell and the Xenopus oocyte expression system. The opposing results obtained for Kv1.4⌬1-110 (17) and Kv1.5 (26) suggest an intracellular environment in the Xenopus oocyte nonpermissive for the functional interaction between Kv␤3.1 and Kv1.5.
The binding of Kv␤1 and Kv␤2 subunits is highly specific for members of the Kv1-subfamily (9,10). This is consistent with immunoprecipitation experiments of Shamotienko et al. (48), who investigated the subunit composition of native Kv channels containing Kv␣ subunits of the Shaker-related family. Interestingly, mKv␤4, the splice variant of the Kv␤3 gene in mice, can be coimmunoprecipitated with both Kv1.5 and Kv2.2 subunits when coexpressed in Sf9 cells (14). However, coexpression with Kv␤3.1 in CHO cells does not confer rapid inactivation to Kv2.2 channels. 3 Therefore, future experiments should be focused on the question of which Kv␣ subunits are associated with Kv␤3.1 in native Kv channels.