A novel K+ channel beta-subunit (hKv beta 1.3) is produced via alternative mRNA splicing.

Voltage-gated K+ channels can form multimeric complexes with accessory β-subunits. We report here a novel K+ channel β-subunit cloned from human heart, hKvβ1.3, that has 74-83% overall identity with previously cloned β-subunits. Comparison of hKvβ1.3 with the previously cloned hKvβ3 and rKvβ1 proteins indicates that the carboxyl-terminal 328 amino acids are identical, while unique variable length amino termini exist. Analysis of human β-subunit cDNA and genomic nucleotide sequences confirm that these three β-subunits are alternatively spliced from a common β-subunit gene. Co-expression of hKvβ1.3 in Xenopus oocytes with the delayed rectifier hKv1.5 indicated that hKvβ1.3 has unique functional effects. This novel β-subunit induced a time-dependent inactivation during membrane voltage steps to positive potentials, induced a 13-mV hyperpolarizing shift in the activation curve, and slowed deactivation (τ = 13 ± 0.5 ms versus 35 ± 1.7 ms at −40 mV). Most notably, hKvβ1.3 converted the Kv1.5 outwardly rectifying current voltage relationship to one showing strong inward rectification. These data suggest that Kv channel current diversity may arise from association with alternatively spliced Kv β-subunits. A simplified nomenclature for the K+ channel β-subunit subfamilies is suggested.

Voltage-gated K ϩ channels (Kv) 1 are important regulators of membrane action potentials as well as many other cellular functions including maintenance of the resting membrane potential, regulating neuron firing, and secretion (1)(2)(3). Most tissues contain multiple channel types belonging to one or more Kv gene subfamilies (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). However, assigning specific K ϩ channel clones to a native current often is difficult since most heterologously expressed Kv channels display either a fast inactivating or a delayed rectifier type current, often with similar pharmacology. While possible factors contributing to this diversity may include Kv channel glycosylation, phosphorylation, and heterotetrameric ␣-subunit formation within a gene subfamily (16 -19), recent studies have shown that ␤-subunits associate with and functionally alter Kv channel clones in heterologous systems (20 -26).
At present, four Kv ␤-subunits have been reported. Three distinct K ϩ channel ␤-subunits were cloned from rat brain (21,27). The rat Kv␤1 subunit confers rapid A-type inactivation on the Kv1.1 delayed rectifier channel, while the rat Kv␤2 isoform does not alter K ϩ channel current phenotypes in the Xenopus oocyte expression system (21). Heinemann and co-workers (27) have reported a third ␤-subunit from rat brain originally named Kv␤3 that shares 68% identity with rKv␤1 and induces partial inactivation in channels of the Kv1 family. The fourth distinct ␤-subunit clone, also termed Kv␤3, was isolated from human and ferret heart (23,24,26). Identity of this Kv␤3 to previously cloned ␤-subunits is greatest in the carboxyl-terminal region with complete identity of hKv␤3 and rKv␤1 in the carboxyl 329 amino acids and 85% identity to rKv␤2. However, the first 79 amino acids of hKv␤3 share only ϳ25% identity with rKv␤1 and do not align with rKv␤2. Fast inactivation of Kv1.4 was accelerated when expressed with hKv␤3 (23)(24)(25) and fast inactivation and a 20-mV hyperpolarizing shift in the activation curve was conferred on the delayed rectifier Kv1.5 (23,26). Human Kv␤3 has no functional effect on the Kv1.1, Kv1.2, and Kv2.1 channel clones (23), although these channels have been postulated to associate with accessory subunits (28,29).
The complete amino acid identity between rKv␤1 and hKv␤3 in the carboxyl terminus suggests that Kv channel ␤-subunit isoforms are encoded by a single ␤-subunit gene. Alternative splicing was suggested previously (21,26,30) since the point of divergence between rKv␤1 and hKv␤3 cDNA contained a potential splice junction and the nucleotide sequence identity in the carboxyl terminus was Ͼ90%. Calcium channel ␤-subunits are encoded by four different genes with alternative splicing of the ␤1 and ␤2 genes giving rise to multiple ␤-subunits within these subfamilies (31,32). Cell-specific alternative splicing of Kv ␤-subunits may be one mechanism responsible for the diversity of Kv channel current among cell types.
We report here the cloning and characterization of a cDNA from human heart that encodes a unique K ϩ channel ␤-subunit designated hKv␤1.3. The hKv␤1.3 subunit uniquely alters the functional properties of hKv1.5, converting it from a delayed rectifier to a channel with rapid, but partial, inactivation. In addition, this current activates at lower voltages, rectifies at depolarized potentials, and has slowed deactivation. Nucleotide sequence comparison of cDNA and genomic DNA encoding human Kv␤1.3, Kv␤3, and Kv␤1 indicate that these subunits are encoded by a common ␤-subunit gene, here designated the Kv␤1 subfamily gene. We suggest that the nomenclature be changed to reflect that the Kv␤1 subunit family members are generated through alternative mRNA splicing (Table I).

EXPERIMENTAL PROCEDURES
Isolation and Characterization of Kv␤1.3-PCR-generated cDNA fragments corresponding to nucleotides 435 to 1089 of rKv␤2.1 were used to screen 3.5 ϫ 10 5 amplified recombinants from a gt10 cardiomyopathic human heart ventricular cDNA library using previously described conditions (33). The primary screening yielded a partial clone (8-82, ϳ3.0 kb) which was subcloned into pBluescript (KSϩ) via NotI and sequenced using double-stranded templates and appropriate oligonucleotide primers (Sequenase 2.0, United States Biochemical Corp.). This clone lacked an in-frame stop codon 5Ј to the first ATG, suggesting it did not represent a full-length coding sequence. Repeated efforts to isolate additional 5Ј sequence from a cDNA library were unsuccessful. To clone the 5Ј end of 8-82, PCR-generated 260-nucleotide fragments unique to the 5Ј end of 8-82 were used to screen 4.2 ϫ 10 5 amplified recombinants from a EMBL-3 human genomic library (Clontech). The primary screening yielded one clone that was isolated using Wizard Magic Lambda Preps per the manufacturer's instructions (Promega). This ϳ14-kb clone was digested by SacI/EcoRI, electrophoresed, transferred to nitrocellulose, and hybridized at high stringency (4). A ϳ4-kb fragment that hybridized to the 260-nucleotide probe was subcloned into pGEM and sequenced. This genomic fragment contained the 26 coding nucleotides missing from the 5Ј end of 8-82 and contained inframe stop codons 5Ј to the ATG. The clone was assembled by linearizing clone 8-82 in pBluescript (KS) with BglII and ligating in a 43-bp fragment containing the 5Ј end of 8-82. The completed clone was verified by sequencing and referred to as hKv␤1.3.
Genomic Isolation of hKv␤1.2 and cDNA Isolation of hKv␤1.1-PCRgenerated cDNA fragments of Kv␤1.2 (nucleotides Ϫ74 to 348) were used to screen 4.2 ϫ 10 5 amplified recombinants from a EMBL-3 human genomic library (Clontech). The primary screening yielded an ϳ13-kb clone that was isolated as described above. This clone was digested with Sau3AI, and fragments of multiple sizes were ligated into the BamHI site of pGEM. A plasmid containing a 1-kb genomic fragment positive for Kv␤1.2 was selected by colony hybridization and sequenced using appropriate oligonucleotide primers.
Isolation of hKv␤1.1 cDNA was completed by generating PCR fragments corresponding to nucleotides 1-216 of rKv␤1 and screening 2.8 ϫ 10 5 unamplified recombinants from a newly constructed ZAPII (Stratagene) cDNA library made from human cerebral cortex mRNA (Clontech). A 4-kb clone was isolated, and plaque-purified clones were recovered by in vivo excision yielding hKv␤1.1 in pBluescript (SKϪ). Nucleotide sequence in various regions was determined as described above.
Electrophysiological Recording and Data Analysis-Templates for in vitro cRNA synthesis were prepared by isolating a XbaI/AccI fragment of hKv␤1.3 (nucleotides Ϫ20 to 1520) from pBluescript (KS), blunting the DNA ends with Klenow and ligating into the SmaI site of the modified pSP64T vector (34). This construct was linearized with EcoRI prior to cRNA synthesis. The hKv1.5 cRNA template was prepared as described previously (35). Human Kv␤1.3 and hKv1.5 cRNAs were synthesized using the SP6 mMessage mMachine TM kit (Ambion) according to the manufacturer's instructions.
Defolliculated Xenopus oocytes were prepared as described previously (19) and injected with approximately 40 nl (4 -20 ng) of in vitro transcribed cRNA. These dilutions resulted in peak currents of 1-10 A. Electrophysiological recordings have been described in detail previously (19,26). Oocytes were bathed in an extracellular solution containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM Hepes (pH 7.5 with NaOH). Membrane currents were recorded using a twomicroelectrode voltage clamp amplifier from Warner Instruments (New Haven, CT). Values are expressed as mean Ϯ S.E. unless indicated otherwise. All experiments were performed at room temperature.

RESULTS AND DISCUSSION
Cloning and Sequence Analysis-While one potential factor underlying functional diversity of the Kv channels in both brain and heart has been attributed to heteromultimeric formation of various ␣-subunits (16 -19), other possibilities include association of one or more function-altering ␤-subunits. ␤-Subunits have been shown to modulate inactivation kinetics, voltage dependence, and current amplitudes of voltage-gated K ϩ , Na ϩ , and Ca 2ϩ channels (20, 21, 24 -26, 32, 36 -41), and the voltage and calcium sensitivity of the Ca 2ϩ -activated K ϩ channels (42). In order to understand the relationship between Kv cardiac clones and native currents, it is necessary to identify the possible ␣and ␤-subunit interactions.
Screening an amplified human heart cDNA library at low stringency with a PCR-generated cDNA probe corresponding to nucleotides 435-1089 of rKv␤2.1 yielded two partial cDNA clones as determined by nucleotide sequencing. The deduced amino acid sequence of one of these clones (h␤2-1) was found to be nearly identical with rKv␤2.1 and likely represents the human homologue of this previously cloned rat subunit (21). The other cDNA (h␤8-82) was most similar to rKv␤1.1 but exhibited little amino acid identity within the postulated amino terminus and lacked a likely translation start site. Additional screens of cDNA libraries did not yield a full-length clone. Screening a human genomic library produced the 26 nucleotides missing from the 5Ј end and an obvious translation start site. The initiating methionine was assigned because it represents the first in-frame ATG positioned 3Ј of termination codons in an open reading frame encoding a 419-amino acid protein (47 kDa). Hydropathy analysis did not reveal a hydrophobic domain, suggesting that similar to other ␤-subunits, hKv␤1.3 is likely a cytoplasmic protein. An amino acid se- To determine whether hKv␤1.3 and hKv␤1.2 represent splice variants of the same gene, 3Ј-untranslated regions were compared. Alignments of this 1800-base pair region showed 99% nucleotide identity between clones isolated from separate individuals, suggesting that hKv␤1.3 and hKv␤1.2 represent splice variants with minor allelic differences. To confirm that Kv␤1.1 is also derived from this gene, Kv␤1.1 was cloned from a human cerebral cortex cDNA library. Three different regions from the 3Ј-untranslated region corresponding to ϳ250, ϳ750, and ϳ1500 bp 3Ј of the translation stop codon were sequenced and showed complete identity with Kv␤1.2 and Kv␤1.3 in this 3Ј-untranslated region suggesting that all three ␤-subunits are encoded by this gene. To confirm the splice junction, we attempted to clone the entire Kv␤1 gene from a human genomic library by screening with a probe corresponding to the carboxyl-terminal 328 amino acids of Kv␤1.1. Although several clones were 10 -20 kb in length, a full-length gene was not isolated based on the finding that no single clone hybridized to either hKv␤1.2 or hKv␤1.3 amino-terminal specific probes. Likewise, when the unique amino termini of hKv␤1.3 and hKv␤1.2 were used to isolate additional genomic clones, these clones did not hybridize to the carboxyl-terminal probe. The complete gene encoding the hKv␤1 subfamily likely exceeds 40 kb. Fig. 2 illustrates the genomic sequences of Kv␤1.2 and Kv␤1.3 surrounding the predicted splice site. Both Kv␤1.2 and Kv␤1.3 genomic sequences correspond to their respective cDNA in the region marked exon. Both genomic clones contain a consensus sequence for donor/acceptor splice sites as shown by the underlined sequence (43). These data provide further evidence that at least three ␤-subunits result from alternative splicing. Therefore, differential regulation of Kv ␤-subunit expression and alternative splicing are likely to be two mechanisms regulating Kv channel diversity. Further in situ analysis and antibody-based immunohistochemical localization of the Kv ␤-subunits will further our understanding of Kv channel ␣and ␤-subunit association.
Functional Expression of hKv␤1.3- Fig. 3 illustrates the effects of Kv␤1.3 on hKv1.5 currents. Current tracings were obtained during voltage clamp steps to depolarizing membrane potentials where outward current is activated and tail currents are measured during channel deactivation upon steps to Ϫ40 mV. Human Kv1.5 normally displays a modest degree of outward rectification and begins to activate at a membrane potential of about Ϫ30 mV (Fig. 3A) (44). In the presence of Kv␤1.3, FIG. 1. Comparison of hKv␤1.3, hKv␤1.2, rKv␤1.1, and rKv␤2 2. Comparison of hKv␤1.3 and hKv␤1.2 genomic sequences surrounding the proposed splice junction. Genomic sequences corresponding to the variable amino termini cDNA sequences of hKv␤1.3 and hKv␤1.2 are shown. Genomic and cDNA sequences match in the region marked exon and diverge at the exon/intron border. Consensus splice site sequences are indicated by the underlining (43) .   FIG. 3. Functional effects of hKv␤1.3 on hKv1.5. Whole cell potassium current was recorded from Xenopus oocytes expressing hKv1.5 in the absence (A) and presence (B) of hKv␤1.3, and each cell was normalized to peak current at ϩ50 mV (ϭ 1). The cells were voltageclamped at a holding potential of Ϫ80 mV for 30 s prior to a variable test potential (shown as inset voltage protocol) and then stepped to Ϫ40 mV to record deactivating tail currents. In A, the test step was 75 ms in duration which allowed steady state current levels to be attained at each potential. This duration was increased to 100 ms in the presence of the hKv␤1.3 subunit (B) to permit steady state to be achieved. Normalized tail currents of hKv1.5 in the presence (closed circles) and the absence (closed squares) of hKv␤1.3 are plotted as a function of test step potential (C). D represents the steady-state current-voltage relationship for Kv1.5 alone (open squares) and Kv1.5 coexpressed with hKv␤1.3 (open circles). Potassium current was measured at steady state (see open symbols in A and B) and plotted as a function of the test potential. In order to compare different cells, the current was normalized by dividing the current at each membrane potential by the value measured at 0 mV (ϭ 1). Note that hKv␤1.3 causes an apparent rectification and that the channels begin to open at more negative membrane potentials compared to Kv1.5 alone. Symbols represent between 5 and 10 observations and are plotted as the mean Ϯ 2 ϫ S.E. Kv1.5 current displays a time-dependent decay or partial inactivation at large depolarizing steps (Fig. 3B) that occurs only at membrane potentials greater than approximately 0 mV. Fig.  3B illustrates also the slower rate of channel deactivation seen in the presence of Kv␤1.3 relative to the Kv1.5 alone. These deactivating tail currents were best fit by one exponential with time constants of 13.2 Ϯ 0.5 ms for wild-type Kv1.5 and time constants of 34.9 Ϯ 1.7 ms in the presence of Kv␤1.3 at Ϫ40 mV (p Ͻ 0.05). Due to the effect of the hKv␤1.3 which decreases current at larger membrane potentials, the magnitude of the tails may be underestimated. Time constants for the apparent inactivation induced by Kv␤1.3 were 8.9 Ϯ 0.3 ms at ϩ50 mV and 9.2 Ϯ 0.45 ms at ϩ30 mV (n ϭ 12) (p Ͼ 0.1, not significant). Fits of the decay at less positive potentials were less reliable and were not done.
An additional effect observed during co-expression of Kv1.5 with Kv␤1.3 was that the threshold for Kv1.5 channel activation occurred at more negative potentials. The shift in the activation curve toward more negative potentials is illustrated in Fig. 3C where the amplitude of the tail currents is plotted as a function of the membrane potential. Since the driving force is constant during this measurement, the curve reflects the fraction of channels open at each membrane potential. The average midpoint of the activation curve for the hKv1.5 was Ϫ7.1 Ϯ 0.5 mV (n ϭ 6) whereas in the presence of the hKv␤1.3 it was Ϫ20 Ϯ 0.5 mV (n ϭ 6). Fig. 3D shows steady state outward current measured during depolarizing steps. Note that hKv1.5 current is observed at more negative potentials when hKv␤1.3 is present, even at potentials that do not show apparent inactivation (i.e. Ϫ20 mV). At membrane potentials greater than approximately 0 mV, hKv1.5 current in the presence of hKv␤1.3 is suppressed relative to the Kv1.5 alone, thereby converting this apparent outwardly rectifying current voltage relationship to one that shows inward rectification. Future detailed analysis of these interactions will elucidate the underlying effects of hKv␤1.3.

CONCLUSIONS
Discovery of the novel Kv␤1.3 subunit and that alternative mRNA splicing generates multiple function altering ␤-subunits further complicates determination of the relationship between cardiac clones and native currents. Future cell-specific localization and co-purification studies using Kv ␤-antibodies will enable us to understand both the pattern of Kv ␤-subunit isoform expression and the Kv channels with which these subunits associate. In addition, analysis of the mechanisms by which Kv␤1.3 alters the voltage sensitivity, inactivation, and rectification of Kv1.5 will advance our understanding of Kv channel function.