New Ether-à-go-go K+ Channel Family Members Localized in Human Telencephalon*

A cDNA encoding a novel voltage-gated K+ channel protein was isolated from human brain. This protein, termed BEC1, is 46% identical to rat elk in theether-à-go-go K+ channel family. TheBEC1 gene maps to the 12q13 region of the human genome. Northern blot analysis indicates that BEC1 is exclusively expressed in human brain, where the expression is concentrated in the telencephalic areas such as the cerebral cortex, amygdala, hippocampus, and striatum. By in situ hybridization, BEC1 is detected in the CA1–CA3 pyramidal cell layers and the dentate gyrus granule cell layers of the hippocampus. Specific signals are also found in neocortical neurons. Transfection of mammalian L929 and Chinese hamster ovary cells with BEC1 cDNA induces a voltage-gated outward current with a fast inactivation component. This current is insensitive to tetraethylammonium and quinidine. Additionally, a second related gene BEC2 was isolated from human brain. BEC2 is also brain-specific, located in the neocortex and the striatum, and functional as a channel gene. Phylogenetic analysis indicates that BEC1 andBEC2 constitute a subfamily, together with elk, in the ether-à-go-go family. The two genes may be involved in cellular excitability of restricted neurons in the human central nervous system.

Kv4. Their protein structures are characterized by six transmembrane regions (S1-S6) with a voltage-sensing S4 region and an ion-conducting pore region located between S5 and S6. K ϩ channels in the eag family also have similar structural features. However, overall sequence similarity between eag and Shaker-type K ϩ channels is very low. Members of the eag family are related to cyclic nucleotide-gated cation channels, hyperpolarization-activated cation channels and plant hyperpolarization-activated K ϩ channels, rather than Shaker-type channels (6). Indeed, a common feature of the C terminus of eag-type channels is a putative cyclic nucleotide-binding (CNB) domain, a characteristic of such ion channels.
The eag family consists of eag, eag-related gene (erg), and elk (6 -8). In Drosophila, genetic mutations of eag or erg induces a hyperexcitable phenotype (9 -11). Human erg maps to LQT2, the locus of inherited long-QT syndrome, an abnormality of cardiac rhythm involving the repolarization of the action potential (12). This gene is expressed not only in the heart but also in the brain and parasympathetic ganglia (8). In a dorsal root ganglionic cell line, pharmacological blocking of erg currents causes the disappearance of spike-frequency adaptation of firing (13). This finding suggests that mammalian erg also contributes to the regulation of neuronal excitability. Consequently, we attempted to identify other novel erg-related genes expressed in mammalian brain. As a result, a new gene encoding a voltage-gated K ϩ channel was isolated, which is more closely related to elk than erg and is exclusively expressed in the human telencephalon. This study reports the molecular cloning, distribution, and channel activity of this gene, BEC1. In addition, the identification of another novel gene closely related to BEC1 is described, indicating the existence of a new subfamily in the eag K ϩ channel family.

EXPERIMENTAL PROCEDURES
Molecular Cloning of Human BEC1-A BLAST search of the expressed sequence tag (EST) data base of GenBank, using human erg as a query amino acid sequence, retrieved two sequences with the accession number R35526 (387 bases) and M79045 (231 bases). To identify the 5Ј and 3Ј ends of the cDNA corresponding to each EST, 5Ј and 3Ј rapid amplification of cDNA ends (RACE) were performed using the Human Brain Marathon-Ready cDNA (CLONTECH) and primers derived from each EST sequence. Amplified fragments were directly cloned into the plasmid pCR2.1 (Invitrogen). Sequencing the RACE products revealed that the two EST were part of a cDNA. Determined sequences were assembled into a large contig (3610 bases) with an open-reading frame encoding 1083 amino acids, which we called human BEC1.
Finally, to verify that human BEC1 cDNA including the entire openreading frame could be cloned from an independent source, human poly(A) ϩ RNA (CLONTECH) was used for reverse transcriptase-polymerase chain reaction (RT-PCR). Reverse transcriptase reaction was primed with the random hexamer. Primers for PCR were designed from * 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) AB022696, AB022697, AB022698, and AB022699.
Molecular Cloning of Rat BEC1 and BEC2-To isolate rat homolog cDNAs of BEC1, RT-PCR was performed using primers derived from two nucleotide sequences of BEC1 conserved with elk, 5Ј-ACc TTC CTG GAC ACC ATC GC-3Ј and 5Ј-CCa AAc ACC ACc GCg TGC AT-3Ј, in which lowercase letters are indicated mismatches with elk. These sequences correspond to amino acid residues 13-19 (TFLDTIA) and 492-498 (MHAVVFG) of human BEC1, which are shared only with elk among the eag family (Fig. 1). Rat brain poly(A) ϩ RNA was prepared by guanidine thiocyanate extraction followed by oligo(dT)-cellulose chromatography (15), and converted into random-primed cDNA. The RT-PCR amplified two fragments of 1.5 and 1.4 kb, which encode polypeptides with 97% and 59% identities to human BEC1, respectively. Because the N and C termini were missing, 5Ј-and 3Ј-RACE were performed on the basis of each determined sequence using the Rat Brain Marathon-Ready cDNA (CLONTECH). Sequences of the 1.5-kb fragment and its 5Ј-and 3Ј-RACE products were assembled into a contig (3715 bases) with an open reading frame of 1087 amino acids, while sequences of the 1.4-kb fragment and its RACE were merged a contig (3736 bases) with an open reading frame of 1017 amino acids. The 1087-amino acid sequence is extremely similar (95%) to human BEC1, indicating that this is the rat counterpart of BEC1. Another sequence shares 48% sequence similarity with BEC1. We estimated that the latter gene was a homolog to BEC1, which we named BEC2.
Molecular Cloning of Human BEC2-BEC1 sequences are markedly conserved between the human and the rat. Nucleotide identity is 89% between both open reading frames of human and rat BEC1. If human BEC2 cDNA is also extremely similar to the rat one, human BEC2 cDNA will be amplified by RT-PCR using some primer sets derived from rat sequence. Thus, to identify human BEC2, RT-PCR with human poly(A) ϩ RNA was performed using primers designed from the sequences around the initiation and stop codon of rat BEC2 cDNA, 5Ј-GCC ATG CCG GTC ATG AAG G-3Ј and 5Ј-GCC AGG GTC AGT GGA ATG TG-3Ј. A 3.1-kb fragment was specifically amplified by the RT-PCR (35 cycles of 98°C, 15 s and 68°C, 3 min) with TaKaRa LA Taq (Takara Shuzo, Japan), directly cloned into pCR2.1 and sequenced. The sequence has an open reading frame encoding 1017 amino acids with 89% identity to rat BEC2, suggesting that this cDNA encodes human BEC2. Both 5Ј-and 3Ј-cDNA ends were determined by RACE. The results revealed a contig (3920 bases) of human BEC2 cDNA and the presence of only a silent mismatch (T17C) in the reverse primer. To construct an expression vector for human BEC2, the open reading frame of human BEC2 cDNA in pCR2.1 was subcloned into pME18S. This vector was named pME-E2.
Nucleotide Sequence Determination and Analysis-Both strands of each fragment were sequenced using ABI PRISM DNA sequencing reagents and an ABI 377 DNA sequencer (Perkin-Elmer). Multiple sequence alignment and phylogenetic analysis were performed using the Clustal W program, version 1.7 (16), and neighbor-joining method of the Phylip program, version 3.572c (17), respectively.
Northern Blot and RT-PCR Analysis of mRNA-Northern blot analysis of mRNA was performed using Multiple Tissue Northern blots (CLONTECH), on which about 2 g of poly(A) ϩ RNA/lane had been immobilized. BEC1 distribution was determined using a randomprimed 32 P-labeled probe corresponding to amino acids 701-984 of BEC1. The membranes were hybridized overnight at 42°C, in 50% formamide, 5ϫ SSPE, 10ϫ Denhardt's solution, 2% SDS, and 100 g/ml sheared, denatured salmon sperm DNA, and then finally washed twice for 30 min in 0.1ϫ SSC, 0.1% SDS at 55°C. The blots were exposed to x-ray film (Hyperfilm-MP; Amersham Pharmacia Biotech) with two intensifying screens at Ϫ80°C for 5 days. To determine BEC2 distribution, the membranes were hybridized as described above with a random-primed 32 P-labeled probe corresponding to amino acids 747-965 of BEC2, finally washed twice for 30 min in 0.1ϫ SSC, 0.1% SDS at 60°C, and exposed to x-ray films for 10 days. Distribution of human erg was determined as a control gene using a probe corresponding to amino acids 1050 -1159. For RT-PCR analysis of human BEC2, the primers 5Ј-TCC GGC TCG CTT GAG GTG CT-3Ј and 5Ј-CCA GTG GGG GAA TGA GAA GC-3Ј were used. This primer set amplifies a 655-bp fragment corresponding to amino acids 598 -815 of BEC2. Random-primed cDNAs were synthesized from each human tissue poly(A) ϩ RNA (1 g, CLONTECH). An aliquot (1/100 volume) of each cDNA was amplified by 33 cycles of PCR (94°C, 15 s; 60°C, 15 s; 72°C, 1 min) with AmpliTaq (Perkin-Elmer). Ten-microliter aliquots of amplified products were separated by electrophoresis on a 1.5% agarose gel and stained by ethidium bromide. The plasmid (0.1 ng) carrying BEC1 or BEC2 cDNA was used as a control template. The efficiency of cDNA synthesis was estimated by RT-PCR (23 cycles) of the glyceraldehyde-3-phosphate dehydrogenase cDNA (18).
In Situ Hybridization-In situ hybridization was performed essentially as described (19) using rat brain sections and a digoxigeninlabeled single strand RNA probe corresponding to amino acids 840 -1013 of rat BEC1. The brains of adult Sprague-Dawley rats were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at 7 m thickness. Before hybridization, the sections were dewaxed, fixed again in 4% paraformaldehyde for 15 min, treated with proteinase K (5 g/ml), 0.2 N HCl, 0.25% acetic anhydride in 0.1 M triethanolamine, dehydrated with ethanol, and dried. The RNA probe was transcribed with T7 RNA polymerase using a DIG RNA labeling kit (Roche Molecular Biochemicals), from a partial fragment of BEC1 cDNA with a T7 promoter sequence incorporated by PCR. Hybridization was performed overnight at 45°C with the digoxigenin-labeled antisense RNA probe (approximately 0.5 g/ml) in 50% formamide, 10% dextran sulfate, 10ϫ Denhardt's solution, 600 mM NaCl, and 250 g/ml E. coli transfer RNA. After hybridization, sections were treated with ribonuclease A (5 g/ml) at 37°C for 30 min, and washed twice at 45°C with 2ϫ SSC and 0.2ϫ SSC for 20 min each. The hybridized digoxigenin-labeled probe was detected with a DIG nucleic acid detection kit (Roche Molecular Biochemicals). Hybridization with the sense probe was performed under identical conditions served as a negative control.
Electrophysiological Studies-L929 and CHO cells were transfected with BEC1 cDNA for electrophysiological studies as described (20,21). L929 cells were co-transfected with the BEC1 expression vector pME-E1 described previously and the green fluorescent protein expression vector phGFP S65T (CLONTECH) using the modified calcium phosphate precipitation method. Transfected cells were determined by observing green fluorescent protein fluorescence with an epifluorescence microscope. For stable transfection, BEC1 cDNA was subcloned into pEF-BOS(dhfr), a derivative incorporating the dihydrofolate reductase gene of the mammalian expression plasmid pEF-BOS (22). CHO cells deficient in dihydrofolate reductase gene were transfected with this plasmid using LipofectAMINE (Life Technologies, Inc.), and cultured in medium without nucleotides. BEC1 expression was amplified with methotrexate (Sigma) at 0.1-3 M.
For electrophysiological studies of BEC2, L929 cells were co-transfected with pME-E2 and phGFP S65T as described previously.

Identification of BEC1 and BEC2
Genes-Two erg-related sequences, the accession numbers R35526 and M79045, were identified in the EST data base of GenBank using a BLAST search with an amino acid sequence of human erg (6). R35526 (387 bases) has two regions encoding amino acid sequences with 45% and 52% identities to 42-and 23-amino acid portions around the S1 and S2 regions of human erg, respectively (smallest sum probability p ϭ 1.1 ϫ 10 Ϫ8 ). M79045 (231 bases) encodes a sequence with 64% identity to a 28-amino acid portion of the S5 region (p ϭ 0.0032). As described under "Experimental Procedures," a series of RACE studies using primers derived from each EST sequences caused to identification a cDNA (3610 bases) with an open-reading frame encoding 1083 amino acids (Fig. 1), showing 28 -46% identity to K ϩ channels of the eag family. The greatest similarity (46% identity) is to rat elk, which has been identified only recently (23). Rat eag and human erg share 30% and 33% sequence similarity with BEC1, respectively. We named this new protein the brain-specific eag-like channel 1, BEC1. Additionally, rat counterpart of the BEC1 gene was identified in rat brain mRNA, which encodes 1087 amino acids with 95% identity to human BEC1.
Another related gene, termed BEC2, was identified in rat or human brain mRNA (see "Experimental Procedures"). Both human and rat BEC2 consist of 1017 amino aids with 48% identity to BEC1. Amino acid sequences of BEC2 proteins are highly conserved between human and rat (89% identity), although the C-terminal part of BEC2 with about 400 amino acids is rather divergent, compared with BEC1.
Primary Structure of BEC1 and BEC2-The sequence alignment of BEC1 and BEC2 is shown in Fig. 1. BEC1 and BEC2 contain a hydrophobic core corresponding to the six transmembrane regions, S1-S6, and the pore region of voltage-gated K ϩ channels. The hydrophobic core is highly conserved between BEC1 (residues 227-508) and BEC2 (residues 229 -482) with 70% identities. This sequence similarity is comparable to that among members in a given subfamily of all the known K ϩ channel superfamily genes. The putative voltage-sensing S4 and ion-conducting pore regions are also shared by BEC1 and BEC2. In the S4 region, only three amino acids are different between BEC1 and BEC2. The S4 regions of both BEC1 and BEC2 contain five positively charged residues found at every third position. Additionally, other two positively and one negatively charged residues are found at common positions. The pore regions of BEC1 and BEC2 have a GFG triplet, a common motif of eag-type K ϩ channels (6), and contain four distinct amino acids from each other. Potential N-glycosylation sites were found in the hydrophilic segments between the S5 and pore region; BEC1 contains three sites and BEC2 one site. An additional site was identified between the S3 and S4 region of BEC2. In the C-terminal region, there is a sequence with significant similarity to the CNB domain of cyclic nucleotide-binding proteins such as cyclic nucleotide-gated cation channels and hyperpolarization-activated cation channels. Although the CNB domains of BEC1 and BEC2 are homologous to eag-type K ϩ channels, they are markedly similar to each other (57% identity) among the cyclic nucleotide-binding proteins. These results indicate that BEC1 and BEC2 may be members of a new subfamily of eag-type K ϩ channels. Also, phylogenetic analysis suggests that these novel genes have a common ancestor with eag-type channels and represent an additional branch in the eag family (Fig. 2).
Assignment of the BEC1 Gene in Human Genome-Human EST R35526, which corresponds to BEC1, is a partial sequence of the clone 37299 in the human infant brain cDNA library 1NIB arrayed by IMAGE consortium (24). This clone has been termed DRES61 and assigned to the 12q13 region of the human genome using fluorescence in situ hybridization analysis by Banfi et al. (25). The complete sequence of DRES61 was recently submitted to GenBank (accession number U69184), and consists of 1088 bases, except a sequence of the adapter used in constructing the library. We performed sequence alignment of human BEC1 and DRES61. The alignment revealed that the 3Ј-end sequence of DRES61 with 879 bases was identical to the 5Ј-end sequence of human BEC1 upstream of codon 282, except a silent mismatch in codon 10 (CCT to CCG) of BEC1. An unique chromosomal region has been identified by fluorescence in situ hybridization with DRES61 cDNA, indicating that BEC1 is derived from an identical gene to DRES61. Taken together, the BEC1 locus is located in the 12q13 region.
Tissue Distribution of BEC1 and BEC2-Northern blot analysis revealed the presence of a 4-kb BEC1 transcript in human brain poly(A) ϩ RNA (Fig. 3A). The size of the mRNA concurs with the length of our identified cDNA. No signals were detected in other human tissues including the heart, placenta, liver, lung, skeletal muscle, kidney, pancreas, spleen, thymus, prostate, testis, ovary, small intestine, colon, and peripheral blood leukocytes. Within human brain, 4-kb transcripts of BEC1 were detected in cortical structures such as the cerebral cortex, amygdala, and hippocampus, and the striatal regions including the putamen and caudate nucleus (Fig. 3B). In these brain regions, an additional 6-kb transcript was identified, but it possessed weak signals. Expression of only the 6-kb transcript was found in human cerebellum, but the expression level was very low. BEC1 transcripts were not detectable in the spinal cord or in the corpus callosum, which contains primarily axons and glia. Thus, the expression of the BEC1 transcripts is largely restricted to the telencephalon in human tissues. The expression pattern of BEC1 contrasts with that of human erg, which displays a more ubiquitous distribution in the brain (Fig. 3B).
Tissue distribution of BEC2 was also elucidated by Northern blot analysis of human poly(A) ϩ RNA. Multiple signals were detected only in the brain, with 4.4, 7.5, and ϳ10 kb (Fig. 3A). We have identified BEC2 cDNA with 3920 bases, as described under "Experimental Procedures." Considering an addition of a poly(A) tail, the 4.4-kb transcript probably corresponds to the identified cDNA. Since the signals were weak, tissue distribution of BEC2 was further determined by RT-PCR analysis. A   FIG. 1. Sequence alignment of BEC1 and BEC2. Conserved residues among more than three sequences are enclosed in solid boxes. The six transmembrane regions (S1-S6), the pore region, and the CNB domain are indicated by overlines. Positively and negatively charged amino acids in the S4 region are marked by plus and minus signs, respectively. The potential N-glycosylation sites (circled) are also indicated. hBEC1, human BEC1; rBEC1, rat BEC1; hBEC2, human BEC2; rBEC2, rat BEC1; elk, Drosophila elk (GenBank U04246); HERG, human erg (U04270); r-eag, rat eag (Z34264).
655-bp fragment corresponding to BEC2 was amplified only from the brain mRNA (Fig. 4). This fragment was generated when the plasmid carrying BEC2 cDNA was used as a template, but not BEC1 cDNA. These results suggests that BEC2 is exclusively expressed in the brain, as well as BEC1. Northern blot analysis of the brain regions indicated the BEC2 expression restricted to the telencephalon, similar to BEC1 (Fig.  3B). All size BEC2 transcripts were predominantly detected in the striatal regions such as the putamen and caudate nucleus. In addition, hybridization signals for BEC2 were detected in the cerebral cortex and hippocampus. The expression pattern of BEC2 is nearly parallel to that of BEC1, although the expression levels appear to be different in each region.
Cellular Localization of BEC1 within the Brain-In situ hybridization was performed using rat brain sections to determine the cellular localization of BEC1 within the brain. Hybridization signals were prominently found in the hippocampus, when only the antisense probe specific to BEC1 was used ( Fig. 5A and B). Specific signals were also detectable in the cerebral cortex. In the hippocampus, in situ hybridization showed that BEC1 transcripts were concentrated in the pyramidal cell body layers of the CA1 and CA3 field and in the granule cell layers of the dentate gyrus. In the cerebral cortex, BEC1 signals were widely present from layer II to layer VI. Specific signals were detected in cell bodies of neurons with typical pyramidal shapes in the cerebral cortex (Fig. 5,  C and D).
Electrophysiological Characteristics of BEC1-To characterize electrophysiological properties of BEC1 using the whole-cell voltage-clamp method, L929 cells were transiently transfected with the BEC1 expression vector. The cells were clamped at a holding potential of Ϫ90 mV and were depolarized to voltages between Ϫ60 and 100 mV. Depolarizing steps induced an outward current in BEC1-transfected cells (Fig. 6A, lower traces). When the voltage was stepped to potentials above 20 mV, the outward current was rapidly inactivated and relaxed to a sustained plateau. The peak current amplitude induced by depolarization to 100 mV was 0.14 -7.3 nA (n ϭ 41). Such responses were not observed in control cells (Fig. 6A, upper traces). Fig.  6B shows averaged current-voltage curves for the peak current within 40 ms and the current at the end of the 200-ms voltage pulse, which correspond to transient and steady-state currents, respectively. The transient current amplitude increased from Ϫ60 to 100 mV voltage-dependently. In contrast, the steadystate current amplitude increased up to ϳ20 mV and then decreased with further depolarization.
For tail current analysis of BEC1, transfectants were depolarized to 80 mV from a holding potential of Ϫ70 mV, and then were repolarized to voltages between Ϫ20 and Ϫ120 mV (Fig.  6C). The BEC1 current was activated and inactivated by the 200-ms-long depolarizing steps. Following repolarization allowed recovery from inactivation and induced a tail current. The tail current reversed at approximately Ϫ80 mV in the bath solution containing 5.4 mM K ϩ . Considering the Nernst potential (Ϫ87 mV, 25°C) for K ϩ , this result supports that BEC1 is a member of the K ϩ channel family.
We examined the effect of two classical K ϩ channel blockers, TEA and quinidine, on the BEC1 channel. Kv2.1 was used as a control K ϩ channel. TEA (10 mM) reduced current amplitude of the Kv2.1 channel but did not affect that of the BEC1 channel in response to depolarization pulses (data not shown). In general, K ϩ channels of the eag family are less sensitive to TEA, compared with the Shaker family (26,27). Our results indicate that BEC1 also has the common feature. Conversely, sensitivities to quinidine are heterogeneous in the eag family. Quinidine potently inhibits the erg current with an IC 50 value of 0.9 M (28), whereas the sensitivity of eag current is low (27). BEC1 was insensitive to 10 M quinidine, which inhibited the channel activity of Kv2.1 (data not shown).
Stable BEC1 transfectants were constructed using dihydrofolate reductase gene-deficient CHO cells and the gene amplification induced by methotrexate. The 5C1-5 clone was selected as a transfectant expressing high amount of BEC1. This cell produced an outward current in response to depolarization FIG. 6. Electrophysiological studies of BEC1. A, voltage-dependent outward currents in L929 cells untransfected (control; upper traces) and transfected with BEC1 cDNA (lower traces). Cells were held at Ϫ90 mV, depolarized to voltages between Ϫ60 and 100 mV, and returned to Ϫ120 mV. B, the current-voltage relationships of the peak current within 40 ms (circle) or the current at the end of the 200-ms voltage pulse (square). The current amplitude at each potential is normalized to amplitude of the peak current recorded at 100 mV. Each point represents mean Ϯ S.E. of 23-25 cells. C, tail current of the BEC1 channel. Cells were depolarized to 80 mV to activate and inactivate BEC1 channels, and then was repolarized to voltages between Ϫ20 and Ϫ120 mV to give a tail current. Holding potential was Ϫ70 mV. steps (data not shown), just as L929 cells transiently transfected with BEC1 cDNA. The parent CHO cells showed no responses. These results suggest that the identified current is caused by BEC1.
Channel Activity of BEC2-The transient transfection of L929 cells and the whole-cell voltage-clamp method were used to identify channel activity of BEC2, as well as BEC1. As shown in Fig. 7, depolarization steps induced an outward current in BEC2-transfected cells, indicating that the BEC2 gene also encodes a channel protein. The inactivation process such as the BEC1 current was not observed in the BEC2 current. DISCUSSION This study describes new members of the voltage-gated K ϩ channel superfamily, BEC1 and BEC2, which are exclusively expressed in human brain. Although a number of voltage-gated K ϩ channel genes have been isolated from the mammalian brain, almost all of those genes are also expressed in other tissues such as the heart and skeletal muscle. Among the Shaker-type channel genes identified in the brain, Kv1. 2 (32,33). Expression of mammalian erg also is predominant in heart (Ref. 12; Fig. 3A). Exclusive expression in the brain is a remarkable feature of BEC1 and BEC2. In addition, both messages are highly concentrated in the telencephalon of the human brain, suggesting that this channel may contribute to excitability of restricted neurons in the central nervous system. BEC1 messages in the hippocampus are prominently detected in the CA1 and CA3 pyramidal neurons and the dentate gyrus granule neurons, which constitute the trisynaptic excitatory pathway, a neural circuit important to establish long term synaptic potentiation and depression (34). Given the contribution of voltage-gated K ϩ channels to learning and memory, it is of high interest to study involvement of BEC1 channel in this neural circuit.
BEC1 belongs to the eag family of voltage-gated K ϩ channels. Identification of BEC2, a gene closely related to BEC1, reveals the existence of a new subfamily of eag-type channels. Both BEC1 and BEC2 are homologous to elk, a Drosophila putative K ϩ channel gene. The recently identified rat homolog of elk, rat elk1 (23), is not identical but is the most closely related to BEC1 and BEC2. Intriguingly, we have already isolated a full-length cDNA of a third human gene, BEC3. 2 Rat elk1 seems the rat ortholog of BEC3, since this gene is 91% identical to BEC3. Thus, this subfamily consists of at least three mammalian homologs of elk.
BEC1 elicits an outward current with a fast inactivation component. Fast inactivating currents are also elicited by Shaker-type channels such as Kv1.4 and Kv4.2 (5). However, unlike these Shaker-type channels, the BEC1 current has not only the transient component but also a steady-state component showing the bell-shaped current-voltage relationship. The difference between voltage dependence of both components is a feature of the BEC1 channel. In contrast, BEC2 elicits a noninactivating outward current. Sequence similarity between BEC1 and BEC2 is high (70% identity) in the hydrophobic core region important to channel properties. Comparing two sequences may be considerable to identify amino acids determining the difference between these channel properties. Genetic mutations in several K ϩ channel genes have been identified as genes responsible for inherited disorders (12,(35)(36)(37). Thus, the possibility that BEC1 mutants associate with diseases should be considered. The BEC1 gene maps to the 12q13 region of the human genome. To date, several unsolved genetic disorders have been assigned to the 12q13 region, for example a late-onset form of familial Alzheimer disease (38) and nocturnal enuresis (39). Genotype analyses using polymorphic markers within the BEC1 gene are necessary to analyze whether BEC1 is genetically linked to such disorders. The nucleotide T at position 3 of codon 10 of BEC1 is replaced by G in DRES61, as described under "Results." This substitution is perhaps a single nucleotide polymorphism within BEC1.
In summary, the identification of BEC1 and BEC2 genes further extends to the diversity of the K ϩ channel multigene superfamily. We have demonstrated that both genes are localized in human telencephalon and encode channel proteins. Delineating their physiological roles may provide new insights into neuronal signal transduction and processing in human telencephalon.