Cloning and Functional Expression of Two Families of β-Subunits of the Large Conductance Calcium-activated K+ Channel*

We report here a characterization of two families of calcium-activated K+ channel β-subunits, β2 and β3, which are encoded by distinct genes that map to 3q26.2–27. A single β2 family member and four alternatively spliced variants of β3 were investigated. These subunits have predicted molecular masses of 27.1–31.6 kDa, share ∼30–44% amino acid identity with β1, and exhibit distinct but overlapping expression patterns. Coexpression of the β2 or β3a–c subunits with a BK α-subunit altered the functional properties of the current expressed by the α-subunit alone. The β2 subunit rapidly and completely inactivated the current and shifted the voltage dependence for activation to more polarized membrane potentials. In contrast, coexpression of the β3a–c subunits resulted in only partial inactivation of the current, and the β3b subunit conferred an apparent inward rectification. Furthermore, unlike the β1 and β2 subunits, none of the β3 subunits increased channel sensitivity to calcium or voltage. The tissue-specific expression of these β-subunits may allow for the assembly of a large number of distinct BK channels in vivo, contributing to the functional diversity of native BK currents.


From the Merck Research Laboratories, West Point, Pennsylvania 19486
We report here a characterization of two families of calcium-activated K ؉ channel ␤-subunits, ␤2 and ␤3, which are encoded by distinct genes that map to 3q26.2-27. A single ␤2 family member and four alternatively spliced variants of ␤3 were investigated. These subunits have predicted molecular masses of 27.1-31.6 kDa, share ϳ30 -44% amino acid identity with ␤1, and exhibit distinct but overlapping expression patterns. Coexpression of the ␤2 or ␤3ac subunits with a BK ␣-subunit altered the functional properties of the current expressed by the ␣-subunit alone. The ␤2 subunit rapidly and completely inactivated the current and shifted the voltage dependence for activation to more polarized membrane potentials. In contrast, coexpression of the ␤3ac subunits resulted in only partial inactivation of the current, and the ␤3b subunit conferred an apparent inward rectification. Furthermore, unlike the ␤1 and ␤2 subunits, none of the ␤3 subunits increased channel sensitivity to calcium or voltage. The tissue-specific expression of these ␤-subunits may allow for the assembly of a large number of distinct BK channels in vivo, contributing to the functional diversity of native BK currents.
Calcium-activated K ϩ channels (K Ca ) 1 modulate cellular electrical excitability. These channels are gated by both cytoplasmic calcium and membrane potential and, therefore, provide feedback mechanisms to modulate Ca 2ϩ influx. Activation of K Ca channels hyperpolarizes cells, and the way in which this hyperpolarization regulates Ca 2ϩ entry is dependent upon the nature of the influx pathway. For example, entry through voltage-gated calcium channels (e.g. in myocytes) may be decreased, due to voltage-dependent deactivation of the calcium channels (1)(2)(3). However, influx through voltage-independent channels (e.g. in endothelial cells) may be enhanced due to an increase in the driving force for Ca 2ϩ (4,5). Thus, the regulatory roles of K Ca channels are context-dependent and may vary with cell type. K Ca currents have been recorded from a variety of tissues and have traditionally been classified into broad categories based on single channel conductance (e.g. large, intermediate, or small conductance). However, in addition to these differences in unitary current amplitude, distinct K Ca currents may also vary in terms of their calcium-and voltage dependence, kinetics, or pharmacologic properties (3,6,7). Native K Ca channels, therefore, comprise a large and functionally diverse family. In recent years, the structural basis of this diversity has been investigated. Significant progress in this regard has been made with large conductance (BK) K Ca channels in particular, some of which have been both biochemically purified and cloned (8 -12). BK channels have been purified from tracheal and aortic smooth muscles and demonstrated to be composed of two distinct types of subunits in those tissues: a large poreforming ␣-subunit and a smaller modulatory ␤-subunit (13). To date, only a single gene encoding a BK ␣-subunit (KCNMA1) has been found, although multiple variants are likely to be produced by alternative splicing (12). Two distinct genes encoding BK ␤-subunits have been identified: KCNMB1, which encodes the ␤1 subunit originally isolated from airway smooth muscle (11) has been localized to human chromosome 5q34 (14), and a recently isolated homologue encoding the ␤2 subunit (15,16). Although functional BK channels can be expressed from ␣-subunits alone, coassembly with ␤-subunits can alter the biophysical and pharmacologic properties of the channel (15)(16)(17)(18)(19). However, the properties of some native BK currents are not well reproduced by combinations of currently known ␣and ␤-subunits, suggesting the possibility that novel subunits of these channels may still exist.
We identified two families of BK ␤-subunits. The first, which, to date, contains only a single member (␤2), is identical to that recently identified in a lung carcinoid EST library (15,16). The second, the ␤3 family, comprises four distinct subunits (␤3a-d) that arise by alternative splicing of a single gene. Coexpression of the ␤2 or ␤3a, -b, or-c subunits with a BK ␣-subunit alters the functional properties of the current from that of the ␣-subunit expressed alone. However, unlike the ␤2 subunit, which both inactivates the channel and increases its calcium and voltage sensitivity, the ␤3 subunits do not increase the calcium or voltage sensitivity of the current. The differential expression of these novel ␤-subunits may underlie part of the large functional diversity observed in native BK currents.

EXPERIMENTAL PROCEDURES
Identification and Cloning of cDNAs Encoding BK ␤-Subunits-Sequence encoding the ␤1 subunit (U61537) was used to search the GenBank data base for homologues using the BLASTN and TBLASTN algorithms of the GCG software package (Wisconsin Genetics Group). This search identified an EST (AA904191) that, when completely sequenced, was demonstrated to encode a full-length K Ca ␤-subunit, ␤2. A fragment of this cDNA containing the coding region and 105 bp of the 3Ј-UTR was amplified by PCR using gene-specific oligonucleo-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF204159 -AF204162.
‡ To whom correspondence should be addressed: Merck Research Laboratories, WP26-265, West Point, PA 19486. E-mail: victor_uebele@ merck.com. 1 The abbreviations used are: K Ca , calcium-activated K ϩ channel; UTR, untranslated region; RT, reverse transcription; PCR, polymerase chain reaction; RACE, rapid amplification of cDNA ends; EST, expressed sequence tag; bp, base pair(s); FISH, fluorescence in situ hybridization; contig, group of overlapping clones; BAC, bacterial artificial chromosome; LOD, log of the ratio of odds; BK, large conductance calcium-activated potassium channel. tide primers, cloned into pMVplϩ (a modified version of pSP64T (20) containing an expanded polylinker), and confirmed by complete sequencing of both strands.
The open reading frame and deduced amino acid sequence of ␤2 were then used to query the GenBank data base again. Iterative searches identified several human ESTs (AA195381, AA236930, AA236968, AA279911, AA761761, and AA934876) encoding partial sequences of novel putative BK ␤-subunits (␤3). Commercially available cDNAs encoding these ESTs (AA195381, AA279911, and AA761761) were purchased and sequenced, which demonstrated that none encoded a fulllength protein. To isolate the entire coding regions, 5Ј-RACE (21) was performed using Marathon-Ready spleen cDNA (CLONTECH) and nested gene-specific antisense primers, both of which were derived from the 3Ј-UTR of the putative ␤3. Specifically amplified products were identified in two ways: 1) by cloning and sequencing the major PCR products identified by agarose gel electrophoresis and 2) by cloning the products of the entire unfractionated amplification reaction and probing resultant colonies with a ␤3-specific probe (nucleotides 84 -449 of AA195381). The amplification reactions were performed several times on two different lots of spleen cDNA. RACE products obtained in this manner were sequenced on both strands and shown to encode four distinct members of a novel family of ␤-subunits, ␤3ad. For functional expression, the coding regions of the ␤3 cDNAs were amplified by PCR using sense primers that removed the 5Ј-UTR and improved the translation initiation sequence. The amplified fragments were cloned into pMVplϩ and confirmed by sequencing both strands.
Characterization of the BK ␤3 Gene-Sequences encoding ␤3ad were used to search the high through put genomic sequence data base, resulting in the identification of several ␤3 gene fragments (AA195511, AA279608, AC007823, AQ093921, AQ096353, AQ590923, and AQ673110). Sequence comparisons showed that AC007823 contains the alternatively spliced exons and the beginning of the conserved domain of the ␤3 cDNAs. The other genomic sequences each contain different amounts of the conserved core. To obtain the complete coding sequence, an arrayed human genomic DNA library was screened by PCR (Genome Systems) for the ␤3 gene using two sets of primers, one pair annealing in the ␤3b 5Ј-UTR and the other in the conserved core domain. Four BACs, each encoding part of the ␤3 gene, were identified and analyzed by PCR for exons encoding the splice variant-specific and conserved domains. These amplifications also enabled approximation of intron/exon boundaries and sizes within the gene. Appropriate regions of each clone were then sequenced to confirm and refine the results.
Chromosomal Mapping-The chromosomal locations of the ␤2, ␤3, and GCF2 genes were mapped by radiation hybrid analysis, which was carried out using DNAs isolated from the Stanford G3 (22) and Gene-Bridge4 (23) radiation hybrid panels (Research Genetics). The presence or absence of the human gene in each of the DNA samples was determined by amplification of a fragment of that gene by PCR. Control experiments demonstrated no amplification of homologous hamster genes by any of the primer pairs used. To distinguish between amplification of the ␤3b and GCF2 genes, a BsaBI restriction fragment length polymorphism was utilized. This enzyme distinguishes ␤3b from GCF2 by virtue of a BsaBI site unique to the ␤3b gene. Thus, only fragments cleaved by BsaBI were scored as positive for the presence of the ␤3b exon. Similarly, a complementary PstI restriction fragment length polymorphism, unique to the GCF2 gene, was used to score fragments as GCF2-positive. Scoring data were transferred electronically to the Whitehead Institute/MIT Center for Genome Research or the Stanford Human Genome Center for analysis with the LOD score cutoff set to 15.
The ␤3 gene was also mapped by fluorescence in situ hybridization (FISH) in experiments performed by Genome Systems. Briefly, BAC DNA was isolated from clones B716 (␤3c and conserved core), B766 (␤3a-c unique exons), and B767 (␤3b-core) and labeled with digoxigenin-dUTP by nick translation. Labeled probe was combined with a biotin-conjugated probe specific for centromeric sequences of chromosome 3 and hybridized to normal metaphase chromosomes derived from phytohemagglutinin-stimulated peripheral blood lymphocytes in a solution containing 50% formamide, 10% dextran sulfate, and 2% SSC. Probe detection was accomplished by incubating the slides with fluoresceinated antidigoxigenin antibodies and Texas Red avidin, followed by counterstaining with 4Ј,6-diamino-2-phenylindole (Molecular Probes).
Tissue Distribution Analysis-First strand cDNAs, prepared from human heart, brain, placenta, lung, liver, skeletal muscle, kidney, pancreas, thymus, prostate, testes, ovary, small intestine, colon, and peripheral blood leukocyte mRNAs, were purchased from CLONTECH and used as templates for PCR amplification reactions using subunitand splice variant-specific primer pairs. To enable differentiation of fragments amplified from cDNA and contaminating genomic DNA (or incompletely spliced RNA), primer pairs were designed to span at least one intron. To improve specificity, ␤3 sense primers were designed to anneal in the splice variant-specific domains, which prevented amplification of cDNAs derived from chromosome 22 transcripts. Additionally, antisense primers were designed to anneal within the core domain, which prevented amplification of GCF2 fragments. PCR reactions were assembled according to the Advantage cDNA Polymerase Mix protocol (CLONTECH) and cycled 23 times for 30 s at 94°C and 4 min at 68°C. Positive controls for cDNA integrity were performed with primers derived from ␤-actin (the actin primers did not span an intron and, therefore, amplify fragments from both cDNA and genomic DNA). Products were identified by Southern analysis of blots that were probed with randomly primed, 32 P-labeled cDNAs specific for ␤2 (nucleotides 268 -1080), ␤3a (nucleotides 70 -384), ␤3b (nucleotides 463-797), or the core region common to all the ␤3 splice variants (nucleotides 1158 -1450 of ␤3c). Hybridization was carried out overnight at 42°C in 0.25 M NaPO 4 , 0.5 M NaCl, 1.0 mM EDTA, 7% SDS, and 1% bovine serum albumin. Blots were washed twice in 5ϫ SSC, 0.1% SDS at 42°C for 30 min and twice in 1ϫ SSC, 0.1% SDS at 42°C for 30 min each and then exposed to x-ray film. Positives were scored as tissues exhibiting a specifically amplified cDNA fragment of the expected size that also hybridized to the cognate probe. Expression analysis was repeated once using a different lot of template cDNAs with consistent results.
In Situ Hybridization and Immunohistochemistry-In situ hybridization experiments were performed using an oligonucleotide probe derived from the sequence unique to ␤3c. The antisense probe corresponded to nucleotides 907-857, and the control sense probe to nucleotides 825-875 of the ␤3c sequence. Probes were labeled at their 3Ј ends using the DIG oligonucleotide tailing kit (Roche Molecular Biochemicals) with biotin-16-dUTP (Roche) substituted for digoxigenin-dUTP. Formalin fixed human pancreas specimens (National Disease Research Interchange) were processed to paraffin, sectioned at 8 m, and mounted on Superfrost plus slides (Fisher). In situ hybridization was carried out as described previously using 2 pmol of labeled probe/ml of hybridization buffer (24). The hybridization signal was amplified using the TSA Direct Red FISH tyramide reagent (NEN Life Science Products) according to manufacturer's directions.
Immunohistochemistry was performed sequentially following in situ hybridization. Sections demonstrating optimal ␤3c mRNA signal were incubated with either guinea pig anti-human insulin sera (Dako) or rabbit anti-human glucagon sera (Dako) for 1 h at room temperature. Bound antibody was detected with fluorescein isothiocyanate-conjugated donkey anti-guinea pig IgG (Jackson Immunoresearch; 15 g/ml in phosphate-buffered saline) or donkey anti-rabbit IgG (Jackson Immunoresearch; 15 g/ml in phosphate-buffered saline), respectively. Sections were counterstained with 4Ј,6-diamino-2-phenylindole, and images were obtained and processed using a Nikon E1000 microscope, Micromax CCD camera (Princeton Instruments), and Metamorph imaging program (Universal Imaging).
Electrophysiology-The cDNA encoding the BK ␣-subunit was a kind gift from Ligia Toro (identical in sequence to U11058 with one exception: this clone contains the conservative R1112K mutation) (25). Plasmids encoding channel subunit cDNAs were linearized with appropriate restriction enzymes and cRNA synthesized by standard procedures (26,27). cRNAs were injected into Xenopus oocytes using 1.5 ng of ␣-subunit RNA/oocyte Ϯ ␤-subunit RNA at equimolar concentration, 5-fold, or 10-fold molar excess. The molar ratio of the ␤/␣ RNAs in coinjection experiments was varied from 1 to 10 in attempts to maximize stoichiometric assembly of the two subunits. Preliminary comparisons of the magnitudes of functional effects induced by ␤-subunits demonstrated saturation of effects at Ͻ5-fold molar excess of ␤ RNA. Therefore, all subsequent studies were done with ␤-subunit RNA in 10-fold molar excess over ␣. Oocytes were maintained at 18°C in ND-96 (28), and macroscopic K Ca currents were recorded from inside-out patches 3-14 days following injection. Recordings were performed in symmetrical potassium; the standard pipette and bath solutions contained 116 mM potassium gluconate, 4 mM potassium chloride, and 10 mM HEPES, pH 7.2. CaCl 2 was added to the bath solution to yield a final concentration of free ionized calcium of 30 M, taking the stability constant for calcium gluconate (15.9 M Ϫ1 ) into account (29). Currents were recorded using an EPC-7 amplifier (HEKA) and pClamp6.0 software (Axon Instruments). Most current records were filtered at 2 kHz using an 8-pole Bessel filter (Frequency Devices) and sampled at 5 kHz, but some records, included in group data, were filtered at 1 kHz and sampled at 3 kHz.
Currents were elicited in the presence of 30 M bath Ca 2ϩ using a voltage command consisting of a holding potential of Ϫ80 mV, followed by a 200-ms prepulse to Ϫ160 mV, and finally 500-ms step depolariza-tions from Ϫ80 to ϩ80 mV in 10-mV increments. Tail currents were recorded for 120 ms after returning to the Ϫ80 mV holding potential. The hyperpolarizing prepulse was needed to remove steady state inactivation at the holding potential when an inactivating ␤-subunit was coexpressed. Peak currents were measured, transformed to macroscopic conductances, and plotted as a function of test potential to assess changes in the voltage dependence of activation at 30 M Ca 2ϩ . Where appropriate, Boltzmann equations were fit to these data to estimate midpoints of activation. Maximal inactivation parameters, fractional non-inactivating current, and inactivation rates were measured from current traces acquired in 30 M Ca 2ϩ at ϩ80 mV. These values were used to calculate saturation of ␤-subunit effects after coinjection of ␤and ␣-subunit cRNAs at increasing ratios. Inactivation rates were determined from single exponential fits to the data. Fractional noninactivating current was calculated as steady-state current/peak current, and fractional inactivating current was estimated as peak current minus steady-state current divided by peak current. Values are presented as mean Ϯ S.E.
Generation of Antiserum and Immunoprecipitation-Rabbit antipeptide antisera was raised (Genosys) against the amino-terminal 15 amino acids of ␤3b, and initially characterized by immunoprecitpitation of in vitro translated ␤-subunits. Proteins were translated in a rabbit reticulocyte lysate (Promega), in the presence of translation grade [ 35 S]Met (NEN Life Science Products) and canine microsomal membranes (Promega), according to the manufacturers' recommended protocol. Immunoprecipitations were then carried out as described previously (30) using 1% Triton X-100 to solubilize the proteins and a 1/100 dilution of either preserum or antiserum. Proteins were separated by SDS-polyacrylamide gel electrophoresis and visualized by fluorography.
To investigate the synthesis of ␤-subunits in Xenopus oocytes, cells were injected with cRNAs encoding ␤3b or ␤3d and incubated overnight at room temperature in 1 ml ND-96 supplemented with 0.5 mCi of ICN Tran 35 S-label (1175 Ci/mmol) to metabolically label newly synthesized proteins. Solubilization and immunoprecipitation (from batches of 24 -30 oocytes) was carried out as described previously (31), using 1% Triton X-100 as the detergent and a 1/100 dilution of preserum or antiserum. Immunoprecipitated proteins were fractionated by SDSpolyacrylamide gel electrophoresis and visualized by fluorography.

RESULTS
Identification and Cloning of Five BK ␤-Subunits-A combination of data base searching and classical molecular biological techniques resulted in the identification of two families of ␤-subunits, ␤2 and ␤3. The initial data base searches were carried out using the amino acid and nucleotide sequences of the BK ␤-subunit purified from airway and vascular smooth muscle (␤1) (11) as the query and identified an EST encoding a full-length homologue (␤2). Subsequent searches with the ␤2 sequence then identified several ESTs encoding parts of another putative ␤-subunit (␤3). None of these latter cDNAs encoded a full-length protein, so 5Ј-RACE was performed to complete the coding sequences. Because many of these ESTs were derived from a tonsil preparation enriched for B cells, cDNA from spleen (an available tissue that is also rich in B cells) was used as the template. A novel family of 4 ␤-subunits (␤3a-d) was cloned from the products of that reaction. The PCR reactions were repeated several times using two different spleen cDNA preparations with consistent results.
Although differing in primary sequence, the overall structure of the ␤2 and ␤3 subunits is similar to that of ␤1, containing two hydrophobic, putative transmembrane domains (Fig.  1). The deduced amino acid sequence of the ␤2 subunit is 235 residues in length, predicting a 27.1-kDa protein that shares 44% pairwise amino acid identity with ␤1. The ␤3 family consists of four related subunits (Fig. 1), ranging in length from 257 to 279 amino acids (29.1-31.6 kDa) that are approximately 32% and 41% identical to ␤1 and ␤2, respectively. The ␤3ad subunits vary only in their cytoplasmic, amino-terminal sequence and share 256 carboxyl-terminal amino acids. A single nucleotide polymorphism was identified within this conserved domain of the ␤3 family. The nucleotide (at position 1350 of the ␤3c cDNA sequence) was found to be either G or A (11 G and 9 A in 20 independent cDNAs), resulting in either a Ser or an Asn in the encoded protein (position 161 in the ␤3c amino acid sequence), and this single nucleotide polymorphism was found in cDNAs encoding each of the ␤3 splice variants.
Alignment of the deduced amino acid sequences of BK ␤-subunits (Fig. 1) demonstrates that sequence conservation between families is strongest within the two putative transmembrane domains and, to a lesser extent, in the connecting FIG. 1. Alignment of the deduced amino acid sequences. The deduced amino acid sequences of the ␤1 (U61537), ␤2, and ␤3ad subunits were aligned using the PILEUP program of the GCG software package. The amino-terminal sequences, unique to each of the four ␤3 splice variants, are individually shown for this family. The conserved core of the ␤3 splice variants is depicted as only a single sequence labeled ␤3 (dashes in the top part of this figure indicate amino acid identity within this domain). Sequence conservation between families is denoted by colored letters with identical amino acids in red and conservative substitutions in blue. The consensus sequence was generated by comparing the ␤1, ␤2, and ␤3 (core) sequences. Within the consensus, uppercase indicates amino acid identity at that position and lowercase indicates identity or conservative substitution between any 2 of the 3 sequences. Notable sequence features are indicated: the two boxed regions indicate positions of the hydrophobic, putative transmembrane domains; q, N-linked glycosylation sites; Ⅺ, alternative splice site in ␤3 variants; ‚, single nucleotide polymorphism in ␤3 variants; E, protein kinase A sites; , tyrosine kinase sites; ‹, leucine zipper start point. extracellular loop. The unique amino-terminal sequences of the ␤2 and ␤3 variants are longer than that of ␤1 and contain consensus sequences for phosphorylation by several different protein kinases, including some that are found in all of the subunits (e.g. protein kinase A sites) and some that are restricted to particular family members (e.g. tyrosine kinase sites in ␤3). Each of the ␤-subunits contains potential N-linked glycosylation sites within the extracellular loop, but the number is subunit-dependent, ranging from 1 to 3. The carboxylterminal domain of the ␤3 subunit is also longer than that found in either ␤1 or ␤2, and contains a leucine zipper motif that starts in the second transmembrane domain.
Other Related Sequences-Data base searches with the nucleotide sequences encoding ␤2 and ␤3ad resulted in the identification of two additional DNAs that are homologous to different regions of the ␤3 gene. The first, homologous to the ␤3b domain, is a cDNA (U69609 and AJ223075) encoding a transcription factor, GCF2 (32,33). Although transcribed from an independent gene (see mapping data below), this cDNA is ϳ95% identical to exon 1b and parts of the 2 flanking introns of the ␤3 gene ( Fig. 2A). However, since exon 1b encodes only the 5Ј-UTR and the initiating methionine of the ␤3b subunit, there is no homology between GCF2 and any of the BK ␤-subunits at the amino acid level. The second sequence, a fragment of chromosome 22q11.2 (AP000365 and AP000547), is homologous to the conserved core region of the ␤3 variants. Sequence analysis revealed Ͼ95% nucleotide identity between this region of chromosome 22 and exons 3, 4, and the intervening intron of the ␤3 gene (Fig. 2). However, there is no further homology to any of the other introns or exons of the ␤3 gene in the additional Ͼ100 kilobase pairs of chromosome 22 sequence available in AP000365 or AP000547.
Alternative Splicing Generates the BK ␤3 Family-Analysis of the structure of the gene encoding ␤3ad demonstrated that this family of subunits arises from alternative splicing of a single gene. A search of the high throughput genomic sequence data base revealed a single entry (AC007823) that contains the unique (5Ј) sequences of the ␤3a, ␤3b, ␤3c, and ␤3d cDNAs. All of these sequences, as well as the first 191 bp of the conserved core domain, are present on a 37-kilobase pair fragment of human genomic DNA in the following order: ␤3a-␤3b-␤3c/dcore(1-191) (Fig. 2). The sequences unique to the ␤3a and ␤3b subunits are present on distinct exons (1a and 1b), whereas the unique ␤3c and ␤3d sequences arise by differential splicing of a third exon (1c/1d). The first 191 bp of the ␤3 conserved core domain are encoded on another exon (2) that is contiguous with exon 1c.
Since no additional ␤3 genomic sequence was yet present in the data base, a human genomic DNA library was screened to isolate the remaining part of the gene. Four BACs were isolated in two screens, and each was analyzed by a combination of PCR and direct sequencing to determine which parts of the ␤3 gene they encode. In this way, a contig of two overlapping BACs (B766 and B767) was demonstrated to contain all of the coding sequence of the ␤3 gene. Detailed characterization of B767 demonstrated that two additional exons (3 and 4) encode the ␤3 core sequence (and some 3Ј-UTR) that was not present in the genomic data base (Fig. 2). Thus, the ␤3 subunits arise by alternative splicing of a gene containing 6 exons, 3 of which (1a-c/d) encode sequence unique to each of the splice variants. The other 3 exons (2-4) encode the carboxyl-terminal sequence common to all members of this family (Fig. 2).
Chromosomal Mapping-The chromosomal locations of the ␤2, ␤3, and GCF2 genes were mapped by radiation hybrid analysis. PCR reactions were carried out against DNAs isolated from G3 and G4 radiation hybrid panels using subunitand splice variant-specific primers. Due to the high degree of sequence homology, primers used to amplify the unique ␤3b domain also amplified a fragment of GCF2. However, ␤3b and GCF2 amplification products were distinguished by their differential susceptibility to digestion by either BsaBI (which specifically cleaves the ␤3b fragment) or PstI (which cleaves only the GCF2 fragment). Therefore, fragments amplified from the ␤2, ␤3ac/d, and GCF2 exons could each be identified unambiguously. These experiments mapped the ␤2 gene to human chromosome 3q26.2-27 by linkage to STS markers WI-5562 and D3S3511 (LOD Ͼ 19). Similarly, exons 1a, 1b, and 1c/d of the ␤3 gene were also linked to 3q26.2-27 by linkage to markers WI-5562, D3S3511, WI-3735, WI-7917, and SHGC-56395 (LOD Ͼ 15).
The ␤3 location deduced from the radiation hybrid experiments was independently confirmed in two ways. First, one of the STS markers to which the ␤3c exon was linked (WI-7917) was also found on two of the BACs (B717 and B767) isolated in screens of the human genomic DNA library for clones encoding ␤3. Second, three of the isolated BAC clones (B717, B766, and B767) were physically mapped by FISH. Hybridization of the BAC DNAs to metaphase chromosome spreads resulted in the specific labeling of chromosome 3 (Fig. 3). Measurements of 10 specifically labeled chromosomes 3 for each of the BAC clones demonstrated that the ␤3 gene is located at a position that is 84 Ϯ 2% of the distance from the centromere to the telomere of the long arm of chromosome 3, an area corresponding to 3q26.3-27.1. A total of 80 metaphase cells were analyzed for each BAC, with 68 -76 exhibiting specific labeling.
Thus, the mapping experiments demonstrate that the ␤2 and ␤3 genes are clustered together toward the terminus of 3q. In contrast, the GCF2 gene was mapped to chromosome 2 by linkage to D2S331 (LOD Ͼ 19), confirming that the GCF2 and ␤3 proteins are encoded by distinct genes. These data also demonstrate that the ␤3 gene is distinct from the homologous fragment located on chromosome 22q11.2.
Differential Expression of the ␤2 and ␤3ad Subunits-Since the high degree of nucleotide identity among 1) GCF2 and ␤3b, 2) ␤3c and ␤3d, and 3) ␤3 core and chromosome 22q11.2 precluded unambiguous Northern analysis for expression of those subunits, tissue distribution was studied by RT-PCR. Specificity was ensured in several ways. For example, primers were designed to span at least one intron, thereby allowing differentiation of products amplified from mRNA and contaminating genomic DNA (or incompletely spliced RNA). In addition, since the ␤3 core has no homology to GCF2 cDNA, the use of an antisense primer annealing in that domain allowed ␤3b-specific amplification. Similarly, sense primers were designed to anneal within the 5Ј unique regions of each of the ␤3 splice variants to prevent amplification of transcripts from chromosome 22. The ␤3c and ␤3d products were distinguished based on their 101-bp size difference, and the ␤3c distribution was independently confirmed using a sense primer that anneals in the region not present in ␤3d.
This RT-PCR analysis demonstrated that the BK ␤-subunits exhibit distinct patterns of expression (Fig. 4) and that the ␤2 and ␤3a subunits are more restricted in their tissue distribution than are the ␤3b-d variants. Although precise quantitation of expression is not possible using this technique, the data in Fig. 4 suggest that expression of ␤2 is strongest in kidney and pancreas with weaker expression (some requiring prolonged exposures of the blot) in ovary, testes, and small intestine. ␤3a is expressed in placenta, pancreas, kidney, and heart, whereas ␤3b-d are widely distributed.
The abundance of ␤3c mRNA in pancreas prompted further sublocalization studies. In situ hybridization analysis demonstrated colocalization of ␤3c with insulin, and not with glucagon, indicating that expression of this subunit in the pancreas is limited to ␤ cells (Fig. 5). Similar results were obtained using either an antisense oligonucleotide probe specific for ␤3c (data in Fig. 5) or a 600-nucleotide antisense riboprobe that did not distinguish between ␤3c and ␤3d (data not shown).
Functional Effects of the ␤2 and ␤3 Subunits-To examine the effects of these ␤-subunits on the function of K Ca channels, cRNA encoding a BK ␣-subunit (h-slo) (25) was injected into Xenopus oocytes with or without each ␤-subunit cRNA. Coinjection of the ␤2 transcript had significant effects on the biophysical properties of the current (Fig. 6). For example, whereas cells expressing the ␣-subunit alone exhibited noninactivating currents, coexpression of the ␤2 subunit resulted in rapid ( ϭ 51 Ϯ 6 ms at ϩ80 mV, 30 M Ca 2ϩ , n ϭ 7) and complete inactivation. Furthermore, the ␤2 subunit shifted the voltage dependence for activation of the channel by approximately Ϫ60 mV (Table I).
Three of the four ␤3 splice variants also altered the properties of the K Ca channel, but their effects differed from those induced by ␤2 and, in fact, also varied between splice variants. For example, coexpression of the ␤3a, -b, and -c subunits resulted in partial inactivation (Fig. 6) of the current. Thus, although the time constants for inactivation of ␤3a and ␤3c currents were similar to that induced by ␤2 ( ϭ 45 Ϯ 15 ms for ␤3a (n ϭ 6), and 60 Ϯ 6 ms for ␤3c (n ϭ 9)), and voltageindependent under the conditions tested (data not shown), the fractional inactivation during a 500-ms pulse to ϩ80 mV (in 30 M Ca 2ϩ ) was only 0.76 Ϯ 0.03 (n ϭ 6) and 0.61 Ϯ 0.03 (n ϭ 10) in cells coexpressing ␤3a or ␤3c, respectively, compared with 0.97 Ϯ 0.02 (n ϭ 7) in cells expressing ␤2. The ␤3b subunit conferred a small component of extremely fast inactivation that could be resolved in only a fraction of the patches ( ϭ 1.5 Ϯ 0.2 ms, n ϭ 3 out of 7; Fig. 6, inset). The ␤3d subunit did not induce detectable inactivation.
Although the measurable time-dependent inactivation conferred by ␤3b was small and rapid, a comparison of the ratio of steady state currents at ϩ80 mV to peak tail current at Ϫ80 mV revealed an apparent rectification consistent with very rapid inactivation that was established within 1-2 ms. In cells expressing the ␣-subunit alone, this ratio was 0.97 Ϯ 0.03 (n ϭ 11), demonstrating equivalent current magnitudes at ϩ80 and Ϫ80 mV, consistent with a linear, ohmic conductance. In contrast, in cells coexpressing ␣ and ␤3b, this ratio was 0.46 Ϯ 0.05 (n ϭ 7), indicating an apparent inward rectification conferred by the ␤3b subunit. This apparent rectification may be the result of an extremely rapid inactivation, as has been previously described for currents expressed from h-erg (34).
Tail current decay during hyperpolarizing steps was markedly slower in channels containing ␤3a subunits (Fig. 6). For FIG. 4. Tissue-specific expression of ␤2 and ␤3ad transcripts. Expression in different tissues was analyzed by RT-PCR using primers that distinguish the individual subunits (see "Experimental Procedures"). Products were detected by Southern analysis. The positive control cDNAs for the ␤3c/d amplification were mixed after completion for simplicity of presentation, but a single band for each control was observed at the appropriate size when the two samples were run separately. The spleen cDNA was the original template (from CLONTECH) used to clone each of the ␤3 variants and serves as an additional positive control. Sizes of the products are indicated in bp. example, two time constants were required to fit ␣ ϩ ␤3a tail currents at Ϫ80 mV (1.3 Ϯ 0.06 ms and 72 Ϯ 16 ms, n ϭ 6). In contrast, when ␣ was either expressed alone, or with other ␤3 subunits, tail currents could be fit with a single time constant that approximates the fast component observed with ␤3a: 1.3 Ϯ 0.18 ms for ␣ alone (n ϭ 11), 2.3 Ϯ 0.65 ms for ␣ ϩ ␤3b (n ϭ 7), 1.0 Ϯ 0.05 ms for ␣ ϩ ␤3c (n ϭ 9), and 1.1 Ϯ 0.22 ms for ␣ ϩ ␤3d (n ϭ 8). Tail currents in cells expressing ␤2-containing channels were too small for analysis.
The effects of the ␤3 splice variants on the Ca 2ϩ and voltage dependence of the channel were also distinct from that of ␤2. As noted above, at a given Ca 2ϩ concentration, the ␤2 subunit, like ␤1, induced a large shift to more polarized potentials in the voltage dependence for activation of the channel. None of the ␤3 splice variants, however, induced a similar phenotype (Table I). Coexpression of the ␤3b or ␤3d subunits caused no significant change in voltage dependence for activation (V 1 ⁄2 ϭ Ϫ2 Ϯ 4 mV, Ϫ11 Ϯ 9 mV, and Ϫ3 Ϯ 6 mV for ␣ alone, ␣ plus ␤3b, and ␣ plus ␤3d, respectively, at 30 M Ca 2ϩ ). Even more dissimilar to the ␤1 and ␤2 phenotype, coexpression of the ␤3a and ␤3c subunits resulted in shifts in channel activation to more depolarized potentials (V 1 ⁄2ϭ ϩ28 Ϯ 5 mV and ϩ15 Ϯ 0.3 mV for ␤3a and ␤3c, respectively, at 30 M Ca 2ϩ ). Thus, the biophysical properties of this K Ca channel ␣-subunit were modified in diverse ways by different ␤-subunits.
As demonstrated above, functional effects of the ␤2 and ␤3a, and ␤3c subunits are readily apparent when they are coexpressed with a BK ␣-subunit, while the effects of the ␤3b subunit are more subtle. To examine expression of ␤3b in oocytes, we have generated an antisera that is capable of immunoprecipitating each of the ␤3 splice variants from in vitro translations. Using this antiserum, we have been able to im-munoprecipitate ␤3b from oocytes injected with ␤3b cRNA (data not shown.) Thus, the small functional effects noted upon coexpression of the ␤3b subunit are supported by the synthesis of that protein in those cells. Despite successful in vitro translation and immunoprecipitation of the ␤3d subunit, we have neither been able to demonstrate functional effects nor immunoprecipitation of this subunit from oocytes injected with encoding cRNA (data not shown). DISCUSSION We report here the characterization of 2 families of BK ␤-subunits. The first, ␤2, contains only a single member and is identical to that recently described by two other groups (15,16). The second ␤3ad, consists of four related proteins, which arise by alternative splicing of a distinct gene. The sequence of one (␤3c) of these four splice variants was also recently reported (35). Although similar in general structure to the known ␤1 subunit, each of these novel ␤-subunits exhibits properties that distinguish it from the others. For example, the pattern of expression for each of these subunits is distinct, though several subunits are often present within a single tissue. In addition, the various channel subunits contain distinct sites for regulation by diverse kinases, raising the possibility of subunit-specific modulation by various second messengers in vivo. Finally, coexpression of these ␤-subunits with a BK ␣-subunit demonstrated that most alter the properties of the channel and do so in subtype-and splice variant-specific ways. For example, coexpression of the ␤1 subunit has dramatic effects on the voltage and Ca 2ϩ dependence of the channel (shifting V 1 ⁄2 to significantly more negative potentials at a given Ca 2ϩ concentration (18). Although similar to ␤1 in terms of its effects on voltage dependence, ␤2 also significantly alters the inactivation prop- erties of the channel, converting the current to one that rapidly and completely inactivates. The ␤3ac subunits also have distinct effects on the functional properties of BK currents and, in splice variant-specific ways, alter channel kinetics, voltage dependence, and rectification properties. During revision of this article, Brenner et al. (36) reported both the sequence of the ␤3b subunit and another of its subtle functional effects, a small increase in the activation kinetics of the current. Those results are consistent with the data reported here because the apparent alteration of activation kinetics could result from the presence of the fast inactivation process we describe here.
We detected no apparent ␤3d-dependent effects on the prop-erties of the BK current in these studies. Although formally possible, we think this subunit is not a cloning artifact since it was independently isolated several times and detected in several tissues. There are several other possible explanations for the lack of observed ␤3d functional effects. As discussed above, we have not yet been able to biochemically demonstrate synthesis of ␤3d in oocytes injected with its cRNA, despite successful in vitro translation and immunoprecipitation of the same lot of cRNA. We cannot be certain whether this negative result represents a failure of the antibodies to detect ␤3d synthesized in oocytes or whether ␤3d is inefficiently translated in oocytes. The lack of functional effects of ␤3d may, therefore, reflect a lack of expression of this subunit in Xenopus oocytes, inactivity of the protein in this particular expression system, or its inability to assemble with the particular splice variant of the BK ␣-subunit used in these studies, or it may have effects beyond those we have examined. It is also possible that some of these ␤-subunits may have effects on protein processing or on other types of K Ca channels (e.g. small or intermediate conductance channels) in lieu of or in addition to effects on BK channels, and further work will be required to distinguish these possibilities. This work has increased the total number of K Ca ␤-subunits currently known to eight, from five families: the original ␤1 subunit purified and cloned from smooth muscle (11), the ␤2 (15, 16) and ␤3ad variants reported here, and two additional subunits, Y21839 (36) and C06 (37) (␤4 and putative ␤5). Y21839 is another protein that shares 25-31% identity to human ␤1-␤3 and represents a novel family (␤4) of this rapidly expanding class of channel subunits. The Y21839 sequence was recently released in a patent data base (the GENSEQ data base), and functional characterization has recently been reported (36). CO6 is an avian homologue, most closely related to ␤1, that has functional properties similar to those reported for other ␤1 species variants (37,38). However, the low degree of sequence conservation between CO6 and ␤1 (47% pairwise amino acid identity with human ␤1 compared with ϳ80% identity between other known ␤1 species variants) suggests that CO6 might indeed represent a novel family of ␤-subunits, ␤5. Therefore, although only one gene encoding a BK ␣-subunit has been identified, the large number of genes encoding distinct ␤-subunits and their splice variants allows generation of functional diversity in vivo through coassembly of different ␣-subunit splice variants and different modulatory ␤-subunits. Because the functional channel is thought to be composed of four ␣ and up to four ␤-subunits (9), diversity may be further enhanced by coassembly of multiple, functionally distinct splice variants of these ␤-subunits into the same K Ca channel.
The mapping of the ␤2 and ␤3 genes to chromosome 3q26.2-27.1 makes them potential candidate genes in several diseases that also have been mapped to this region. These include cerebral cavernous malformations-3 (3q25. , myelodysplasia syndrome (3q26), Cornelia de Lange syndrome (3q26.3), and autosomal dominant optic atrophy (3q28 -29) (39 -43). Further work will be required to elucidate the role, if any, of these BK channel subunits in the etiology of these pathophysiologic states.
FIG. 6. Effects of ␤-subunit expression on the functional properties of a K Ca channel. Current traces from representative insideout patches expressing a BK ␣-subunit alone or with 10-fold molar excess of the indicated BK ␤-subunit. Currents shown were recorded in 30 M Ca 2ϩ and symmetrical 120 mM K ϩ and were evoked by 500-ms depolarizations from Ϫ80 to ϩ80 mV in 20 mV steps, following a 200-ms prepulse to Ϫ160 mV. Traces for ␣ ϩ ␤3b are also shown at a different time scale to demonstrate the inactivating component (inset). Patches were held at Ϫ80 mV for 10 s between pulses.

TABLE I
Midpoints for activation of currents in patches from cells coexpressing different combinations of ␣and ␤-subunits V 1/2 values were determined by Boltzmann fits to the data. n ϭ number of patches, p ϭ probability determined by t test, and asterisk (*) denotes significant difference between the two groups at 0.05 level.