Genomic Organization, Chromosomal Localization, Tissue Distribution, and Biophysical Characterization of a Novel MammalianShaker-related Voltage-gated Potassium Channel, Kv1.7*

We report the isolation of a novel mouse voltage-gated Shaker-related K+ channel gene,Kv1.7 (Kcna7/KCNA7). Unlike other known Kv1 family genes that have intronless coding regions, the protein-coding region of Kv1.7 is interrupted by a 1.9-kilobase pair intron. The Kv1.7 gene and the related Kv3.3(Kcnc3/KCNC3) gene map to mouse chromosome 7 and human chromosome 19q13.3, a region that has been suggested to contain a diabetic susceptibility locus. The mouse Kv1.7 channel is voltage-dependent and rapidly inactivating, exhibits cumulative inactivation, and has a single channel conductance of 21 pS. It is potently blocked by noxiustoxin and stichodactylatoxin, and is insensitive to tetraethylammonium, kaliotoxin, and charybdotoxin. Northern blot analysis reveals ∼3-kilobase pair Kv1.7transcripts in mouse heart and skeletal muscle. In situhybridization demonstrates the presence of Kv1.7 in mouse pancreatic islet cells. Kv1.7 was also isolated from mouse brain and hamster insulinoma cells by polymerase chain reaction.

Ion channels that exhibit a variety of gating patterns and ion selectivity are critical elements that transduce signals in diverse cell types (1). Voltage-gated potassium-selective (Kv) 1 channels represent the largest family within this class of proteins (2), and perform many vital functions in both electrically excitable and nonexcitable cells. Following initiation of an ac-tion potential in nerve and muscle cells, Kv channels play the important role of repolarizing the cell membrane (1). Kv channels can also modulate hormone secretion, for example insulin release from pancreatic islet cells (3)(4)(5)(6), and regulate calcium signaling during mitogen-stimulated activation of lymphocytes (7).
Kv channels in mammalian cells are encoded by an extended family of at least nineteen genes (2). The largest subfamily, Kv1, is related to the fly Shaker gene and contains six members, Kv1.1-Kv1.6 (2). The Shaker gene has 21 exons, which can be alternatively spliced to generate at least five functionally distinct transcripts (8,9). In contrast, the protein-coding regions of each of the six known mammalian Kv1 genes and the three known Xenopus homologues are contained in a single exon (2,10), precluding alternative splicing as a means of generating functionally different proteins. The evolutionary significance of this pattern of organization remains a puzzle.
Here we report the identification of a novel mammalian gene, Kv1.7 (Kcna7/KCNA7), that has a genomic organization distinct from the other members of the vertebrate Kv1 subfamily. We have defined the chromosomal location of this gene in the mouse and human genome, determined its tissue distribution, and characterized the biophysical and pharmacological properties of the cloned channel.
Mapping Mouse and Human Chromosomal Locations of Kv1.7 and Kv3.3-Interspecific backcross progeny were generated by mating (C57BL/6J ϫ Mus spretus)F 1 females and (C57BL/6J) males, and a total of 205 N 2 mice were used to map the two mouse genes, mKv1.7/Kcna7 and mKv3.3/Kcnc3, as described previously (11,(17)(18)(19)(20). The probe for mKv1.7 was the entire 6.4-kb EcoRI fragment shown in Fig. 1, and that for mouse Kv3.3 was a 4-kb genomic HindIII fragment containing the entire 3Ј-exon (11). Although 155 mice were analyzed for all markers and are shown in this segregation analysis, up to 188 mice were typed for some pairs of markers. Recombination frequencies were calculated as described (11,(17)(18)(19)(20) using the computer program SPRETUS MADNESS. Gene order was determined by minimizing the number of recombination events required to account for the allele distribution patterns. Fluorescent in situ hybridization of cosmids to metaphase human chromosomes was carried out as described previously (21,22).
Northern Blot Assays-A Northern blot of poly(A) ϩ RNA from mouse heart, brain, spleen, lung, liver, skeletal muscle, kidney, and testis (CLONTECH Inc., Palo Alto, CA) was probed with the mouse Kv1.7specific 3Ј-noncoding region sequence. The PstI/SacI Kv1.7 3Ј-noncoding region was labeled by the random primer method (Boehringer Mannheim Random Primed DNA labeling kit). The RNA blot was hybridized at 60°C for 18 h, washed at a final stringency of 0.2 ϫ SSC and 0.1% SDS at 60°C for 1 h, and exposed to x-ray film at Ϫ80°C with an intensifying screen for 3 days.
In Situ Hybridization-cRNA probes labeled with ␣-35 S-labeled UTP (1300 Ci/mmol) were alkaline-denatured to an average size of 100 nucleotides and used for in situ hybridization on pancreatic frozen sections (6 -10 m thick) from db/db mice. Briefly, sections were hybridized overnight at 42°C, RNase treated, washed five times in 0.5 ϫ SSC at 65°C, coated with Ilford K5 photographic emulsion, and exposed at 4°C for varying times. After development, the sections were counterstained with hematoxylin and eosin Y and examined with a Leitz Aristoplan microscope equipped with reflected polarized light to visualize silver grains in dark field. The probes used for hybridization were as follows: insulin, 1.6-kb human insulin gene including the 5Ј-and 3Ј-flanking sequences (ATCC no. 57399); hKv3.4, 1.9-kb cDNA fragment spanning S1 through the 3Ј end of the coding region (0.6 kb), and 1.3 kb of the 3Ј-noncoding region; mKv1.7, 540-bp PstI/SacI fragment containing 29 bp of C-terminal coding sequence and 511 bp of 3Јnoncoding sequence.
Electrophysiological Studies-To make a mKv1.7 expression construct we amplified a 588-bp fragment from mouse brain cDNA spanning the region encoded by the two Kv1.7 exons using reverse transcriptase PCR (5Ј-primer, 5Ј-TCTCCGTACTCGTCATCCTGG-3Ј; 3Јprimer, 5Ј-AAATGGGTGTCCACCCGGTC-3Ј). Exon 1 (850-bp BspHI/ BinI fragment), a 283-bp BinI/BglII fragment of our 588-bp PCR product, and exon 2 (747-bp BglII/HindIII), were ligated together at BinI and BglII sites to generate "fragment I" (1880 bp). Fragment I was blunt-ended at the 5Ј end and cloned into the pBluescript vector at SmaI/HindIII sites. To remove the 5Ј-NCR from fragment I, and for the purpose of cloning this fragment into the pBSTA expression vector, we introduced a unique BamHI site just upstream of the initiator methionine using PCR: 5Ј-primer, 5Ј-ACAAAAGCTTCATATGACTACAAG-GAAAGCT-3Ј; and 3Ј-primer: 5Ј-AAGCGCAACCCGGCCACG-3Ј. The resulting PCR product (corresponding to the first 233 nucleotides of the coding region) was spliced to fragment I at the NcoI site, and the 1870-nucleotide fragment was ligated to the pBSTA vector. mKv1.7 cRNA was transcribed in vitro (Ambion Kit, Austin, TX) and diluted in a 0.1-0.5% fluorescein-dextran (M r 10,000, Molecular Probes, Eugene, OR) in 100 mM KCl. Rat basophilic leukemic (RBL) cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Hyclone, Logan, UT) and glutamine, and were plated onto glass coverslips one day prior to use for electrophysiological experiments. RBL cells were injected with cRNA using pre-pulled injection capillaries (Femtotip) in combination with an Eppendorf microinjection system (micromanipulator 5171 and transjector 5242; Madison, WI) as described previously (23). Four to six hours later, fluorescent cRNA-injected cells were evaluated electrophysiologically. All membrane currents were recorded at room temperature (22-26°C) with a LIST EPC-7 amplifier (Heka Elektronik, Germany). Series resistance compensation was employed if the current exceeded 2 nA, and the command input was controlled by a PDP 11/73 computer via a digital-to-analog converter. Capacitative and leak currents were subtracted using a P/8 procedure and the holding potential in all experiments was Ϫ80 mV. When membrane currents exceeded 2 nA 80% series resistance compensation was used.

The Protein-coding Region of mKv1.7 Contains an Intron Unlike Its Vertebrate Homologues
A restriction map of a 6.4-kb EcoRI DNA fragment containing the entire mouse Kv1.7 coding region is shown in Fig. 1. The coding region is contained in two exons separated by a 1.9-kb intron. The upstream exon encodes the entire N terminus, S1, and part of the S1-S2 loop. The downstream exon contains the region extending from the S1-S2 loop to the C-terminal end of the protein. The intron-exon splice sites were determined by comparing the genomic sequence with cDNA sequences obtained from the hamster insulinoma cell line, HIT-T1S, and from mouse brain (Fig. 1).
The complete coding sequence of the mKv1.7 is shown in Fig.  2. The mKv1.7 sequence is identical in the N terminus from bp 52 to 996 with the mouse EST sequence AA021711. Betsholtz et al. (24) amplified a short segment of Kv1.7 cDNA spanning the S5/S6 region from mouse (MK-6), rat (RK-6), and hamster (HaK-6) insulin-producing cells using PCR. Our sequence is identical to their MK-6 sequence, except for four nucleotide changes.
The deduced mKv1.7 protein consists of 532 amino acids and contains six putative membrane-spanning domains, S1-S6 (Fig. 2). The hydrophobic core of this protein shares considerable sequence similarity with other Shaker family channels, while the intracellular N and C termini and the external loops between S1/S2 and S3/S4 show little conservation. The protein contains conserved sites for post-translational modifications as indicated in Fig. 2. As do all other Shaker-related channels, mKv1.7 has a potential tyrosine kinase phosphorylation site (RPSFDAVLY) in its N-terminal region (2); the proline-rich stretch within the N terminus (see residues 29 -42) may be a binding site for SH3 domains of tyrosine kinases. Two protein kinase C consensus sites (Ser/Thr-X-Arg/Lys) are present in the cytoplasmic loop between S4 and S5 of mKv1.7; at least one FIG. 1. Genomic organization of mouse Kv1.7. Top, the mKv1.7 coding sequence is indicated by the two stippled boxes, and the six bars within these regions indicate the putative membrane-spanning domains S1 through S6. Restriction sites are indicated as follows: B, BglII; E, EcoRI; P, PstI; S, SacI. Bottom, comparison of the genomic sequence of mKv1.7 with that of mouse and hamster (haKv1.7) cDNAs shows the splice donor and acceptor sites which form the boundaries of the single intervening sequence. of these sites is present in all members of the Kv1 family (2). mKv1.7, like Kv1.6, lacks an N-glycosylation site in the extracellular loop linking the S1 and S2 transmembrane segments; this consensus sequence is conserved in all other Kv1 family genes. Fig. 3 shows a phylogenetic tree of the entire Shaker family of genes based on parsimony analysis of a nucleotide sequence alignment (generated from the amino acid sequence alignment) using the program PAUP (25). Our analysis places mKv1.7 within the Shaker family of genes. The mKv1.7 gene does not cluster with any of the known genes, and its position within the tree is not firmly established.
The interval on mouse chromosome 7 containing mKv1.7/ Kcna7 and mKv3.3/Kcnc3 shares a region of homology with human chromosomes 19q13 and 11p15, and the human homologues of these K ϩ channel genes may therefore be expected to reside on one of these chromosomes. Confirming this predic-tion, we mapped both genes to human 19q13.3-13.4 using fluorescent in situ hybridization. The idiogram of human chromosome 19 shown in Fig. 4B, and a more detailed map shown in Fig. 4C, indicate that hKv1.7/KCNA7 is located ϳ1.3 mb centromeric of hKv3.3/KCNC3. The genes for both muscle glycogen synthase (GYS1) and the histidine-rich calcium protein (HRC) also map to this region; Kv1.7/KCNA7 lies ϳ25 kb telomeric to GYS1 and ϳ200 kb centromeric to HRC (Fig. 4C).  The six putative membrane-spanning domains (S1 through S6) and pore-forming region (P) are indicated. The potential sites for phosphorylation by tyrosine kinase (TY-K) and protein kinase C (PKC) are shown. The position of the single intron between S1 and S2 is indicated by an arrow, and the critical residue for tetraethylammonium block (Ala-441, within the P region) is shown in bold and underlined.
Interestingly, a putative diabetes susceptibility gene has been suggested to be present at 19q13.3 (26,27), especially in Finnish families with associated hypertension and difficulties in insulin-stimulated glucose storage. This region has also been suggested to contain a modifier locus for cystic fibrosis (28).

Biophysical and Pharmacological Characterization of Kv1.7 Channels
We carried out a detailed characterization of mKv1.7 channels expressed in RBL cells which express no native Kv currents (29,30). The mKv1.7 gene expressed in Xenopus oocytes produced currents (data not shown) similar to those obtained in RBL cells (Fig. 5).
Voltage Dependence-Patch clamp studies were performed in the whole-cell mode. Fig. 5A shows a family of outward currents elicited by a 200 ms depolarizing pulse from a holding potential of Ϫ80 mV in RBL cells injected with mKv1.7 cRNA; no outward currents were detected in control cells (data not shown). The currents activate rapidly, and from the conductance-voltage curve shown in Fig. 5B we determined that the 1/2 activation potential (V 1/2 ) is Ϫ20 mV.
Inactivation and Deactivation-Inactivation of mKv1.7 channels was rapid following a 200 ms depolarizing pulse from Ϫ80 to 40 mV (Fig. 5A). The rate of inactivation, measured by fitting the data to a single exponential function, yielded a time constant ( h ) of 14 Ϯ 2.1 ms (S.E., n ϭ 7). As shown in Fig. 5C, the current became progressively smaller following repeated depolarizing pulses at 1-s intervals. This phenomenon, termed "cumulative inactivation," is due to the accumulation of channels in the inactivated state which are then unavailable for opening. The related channels, Kv1.3 (7) and Kv1.4 (31), also display this behavior.
The kinetics of channel closing (deactivation) was determined by first opening the channels with a 15 ms conditioning depolarizing pulse, and then forcing the channels to close by repolarizing to different potentials (Fig. 5D). The time constant, tail , of the resulting "tail" current was 5.1 and 5.3 ms at Ϫ60 mV in two cells.
Single-channel Conductance-We measured single-channel currents in an outside-out patch by applying 450-ms voltage ramps from Ϫ90 to 80 mV every second (Fig. 5E). Single channel events were seen at potentials more positive than ϳϪ15 mV. The single-channel conductance of 21 pS was estimated from the slope of the current recorded during an opening (Fig. 5E).

Expression of mKv1.7 Transcripts in Different Tissues
Northern blot assays using a mKv1.7-specific probe revealed strongly hybridizing 3-kb bands in heart and skeletal muscle; faint bands of similar size were visible in liver and lung (together with larger 7-8-kb bands), but none were seen in spleen, kidney, testis, or brain (Fig. 6) We were able to isolate mKv1.7 transcripts from mouse brain by PCR (see Fig. 1). mKv1.7 is also expressed in placenta, since the mouse EST AA021711 was derived from this tissue.
PCR analysis demonstrated the presence of haKv1.7 mRNAs in hamster insulinoma cells (Fig. 1). We verified the presence of mKv1.7 mRNA in pancreatic islet cells obtained from 9 -16week-old diabetic db/db mice by in situ hybridization (Fig. 7C) using a mKv1.7-specific antisense probe (12)(13)(14); mKv1.7 mRNA was also present in islets from normal db/ϩ mice (data not shown). Scattered acinar cells outside the islets also showed mKv1.7 hybridization (Fig. 7C). In contrast, mKv3.4 mRNA was found in acinar cells surrounding islets, but not in islets, of both db/db (Fig. 7B) and db/ϩ mice (data not shown). As a control, insulin mRNA was detected in both normal and diabetic islets, but not in acinar cells (Fig. 7A). A Kv1.5-specific probe did not show appreciable hybridization to islets (data not shown), despite a report of Kv1.5 cDNA having been cloned from human insulinoma cells (33). DISCUSSION Unlike all other known mammalian Shaker-related genes (Kv1.1-Kv1.6) that have intronless coding regions (2,9), the protein-coding region of mKv1.7 is interrupted by a single 1.9-kb intron. The fly Shaker gene also contains an intron in the S1-S2 loop, raising the possibility that the intron in Kv1.7 may be ancient, predating the divergence of flies and mammals. Both the mouse Kv.1.7 and the fly Shaker intron are placed between codons, i.e. they are "phase 0" introns. While this is consistent with their having a common origin it may also be fortuitous, since there are only three possible "phases." Although we favor the idea that Kv introns were lost in the vertebrate lineage before their expansion by gene duplication (in which case the Kv1.7 intron would represent a more recent insertion), the evolutionary history of this complex gene family remains to be elucidated.
Since Kv1.7 mRNA is expressed in the mouse heart, we searched the literature for native cardiac A-type Kv currents with properties resembling those of Kv1.7. The Kv1.7 homotetramer shares many properties with the rapidly inactivating transient outward (I to ) current in cardiac Purkinje fibers, but not the I to current in atrial and ventricular myocytes. Kv1.7 and the Purkinje I to currents activate at negative potentials (ϳϪ30 to Ϫ20 mV), inactivate rapidly ( h Ͻ 25 ms), exhibit cumulative inactivation, are blocked by micromolar concentrations of 4-aminopyridine, and are resistant to Ͼ20 mM tetra-  5. Kv1.7 currents. A, family of mKv1.7 currents. The holding potential was Ϫ80 mV and depolarizing pulses were applied every 30 s. The test potential was changed from Ϫ50 to 50 mV in 10-mV increments. B, peak K ϩ conductance-voltage relation for currents shown in A. The line through the points was fitted with the Boltzmann equation: g k (E) ϭ g k(max) /{1 ϩ exp[(E n Ϫ E)/k]}, with parameter values g k(max) ϭ 20 nS and k ϭ Ϫ8 mV. C, cumulative inactivation of Kv1.7 currents. Currents were elicited by a train of six depolarizing voltage steps (200-ms duration) to 40 mV once every second from a holding potential of Ϫ80 mV. The current amplitude decreases significantly during this train of pulses from the largest (first trace) to the smallest (last). D, kinetics of deactivation of Kv1.7 currents. Tail currents were elicited by voltage steps from Ϫ100 to Ϫ40 mV after a 15-ms depolarizing prepulse to 40 mV. The tail current-decay time constants, t , were measured by fitting single-exponential functions to the decay of the K ϩ current during repolarization. E, singlechannel currents of Kv1.7 in an outsideout patch. The broken line shows the slope conductance. ethylammonium (34 -36) (this study). In contrast, the I to current in atrial and ventricular muscle, a product of the Kv4.3 gene, does not exhibit cumulative inactivation (36). These studies suggest that at least part of the Purkinje fiber I to might be encoded by the Kv1.7 gene, although more extensive biophysical and pharmacological studies are required to confirm the link, and the presence of Kv1.7 mRNA and/or protein has yet to be demonstrated in these fibers. The abundant expression of Kv1.7 mRNA in mouse heart suggests that this channel is also likely to be present in ventricular and/or atrial muscle where it may co-assemble with other Kv1 family channels to form heterotetramers.
Recent studies suggest an important role for Kv channels in regulating islet cell function, specifically in repolarizing the membrane potential following each action potential during the glucose-induced bursting phase associated with insulin secretion (3)(4)(5)(6). Despite these interesting findings, the genes encoding Kv genes in ␤-cells have not been identified. Although the Kv1.5 gene was isolated from human insulinoma cells (33), we did not detect Kv1.5 mRNA in normal or diseased islets. We have, however, demonstrated the presence of Kv1.7 mRNA in these cells. Unlike the noninactivating Kv channels in pancreatic ␤-cells (3,4), the Kv1.7 homotetramer exhibits rapid Ctype inactivation. Since Kv1.7 mRNA is expressed in pancreatic islets, it is possible that heteromultimers composed of Kv1.7 and other Kv1 subunits constitute the native Kv channels in ␤-cells. Enhanced levels of such Kv channels would excessively hyperpolarize the membrane of ␤-cells and impair insulin secretion (5). The mapping of the Kv1.7 gene to human chromosome 19q13.3, a region thought to contain a diabetic susceptibility gene (26,27), also suggests that Kv1.7 might contribute to the pathogenesis of type II diabetes mellitus in some humans.
In conclusion, we have described a novel Kv1 family gene with a genomic organization distinct from all the other members of the family. The Kv1.7 channel produces a typical A-type current, and transcripts are expressed in the heart, skeletal muscle, brain, placenta, and pancreatic ␤-cells. This channel is biophysically and pharmacologically similar to the Purkinje fiber I to current, and the gene may contribute at least one subunit to heteromultimeric Kv channels in pancreatic ␤-cells.