Characterization of the transcription unit of mouse Kv1.4, a voltage-gated potassium channel gene.

The mouse voltage-gated K+ channel gene, Kv1.4, is expressed in brain and heart as approximately 4.5- and approximately 3.5-kilobase (kb) transcripts. Both mRNAs begin at a common site 1338 bp upstream of the initiation codon, contain 3477 and 4411 nucleotides, respectively, and are encoded by two exons; exon 1 contains 0.5 kb of the 5'-noncoding region (NCR), while exon 2 encodes the remaining 0.8 kb of the 5'-NCR, the entire coding region (2 kb), and all of the 3'-NCR. The 3.5-kb transcript terminates at a polyadenylation signal 177 bp 3' of the stop codon, while the 4.5-kb mRNA utilizes a signal 94 bp farther downstream. Although the proteins generated from either transcript are identical, the two mRNAs are functionally different, the 3.5-kb transcript producing approximately 4-5-fold larger currents when expressed in Xenopus oocytes compared to the 4. 5-kb mRNA. The decreased expression of the longer transcript is due to the presence of five ATTTA repeats in the 3'-NCR which inhibit translation; such motifs have also been reported to destabilize the messages of many other genes and might therefore shorten the life of the 4.5-kb transcript during its natural expression. The Kv1.4 basal promoter is GC-rich, contains three SP1 repeats (CCGCCC, -65 to -35), lacks canonical TATAAA and GGCAATCT motifs, and has no apparent tissue specificity. One region enhances activity of this promoter. Thus, transcriptional and post-transcriptional regulation of mKv1.4, coupled with selective usage of the two alternate Kv1.4 mRNAs, may modulate the levels of functional Kv1.4 channels.

The Kv1.4 gene, located on mouse chromosome 2 and human chromosome 11p14.1 (3), encodes a rapidly inactivating, 4-aminopyridine-sensitive K ϩ channel with phenotypic properties resembling the neuronal after-hyperpolarization-inducing current (6), and the 4-aminopyridine-sensitive component of the cardiac I to current (7)(8)(9). Two distinct Kv1.4 transcripts, ϳ3.5 and ϳ4.5 kb in length, are present in rat heart and brain (4,10,13). Here we report the characterization of the entire transcriptional unit of mouse Kv1.4 (mKv1.4) which may represent the first such analysis for any member of the extended voltage-gated ion channel gene family.

Northern Blots
A multiple-tissue mouse mRNA blot obtained from Clontech (Palo Alto, CA) was probed with different regions of the 11-kb mKv1.4 genomic clone (stringency ϭ 0.5 ϫ SSC, 0.1% SDS and 60°C, exposure ϭ 3-7 days). Blots were stripped of probe by boiling in 0.1% SDS and cooling to room temperature, and reused after ensuring that no residual probe remained.

pLuciferase Vectors and Luciferase Assays
Promega pLuciferase Constructs-Three luciferase-containing constructs were used in this study. The control vector contains the SV40 enhancer and promoter elements upstream of the luciferase gene. The enhancer vector contains (from 5Ј to 3Ј) the SV40 enhancer, a multiple cloning site for introducing the putative promoter-containing element, and the luciferase reporter gene. The basic vector includes only the luciferase gene with an upstream multiple cloning site for introducing the putative promoter fragment.
Deletion Constructs-Several 5Ј-flanking fragments from the mKv1.4 genomic clone were cloned into the enhancer or basic vectors using appropriate restriction enzymes, and their ability to induce luciferase activity was determined. The integrity of all constructs was confirmed by restriction mapping and/or by sequencing.
Cell Culture, Transfection of Constructs, and Luciferase Assays-C 2 C 12 myoblast cells were grown in minimum essential medium (Cell Grow Mediatech, Washington, D. C.) containing 4.5 g/liter glucose, 10% fetal bovine serum (Summit Biotechnology, Greeley, CO), 0.1 IU/ml penicillin, and 0.1 mg/ml streptomycin (ICN Biomedicals Inc., Costa Mesa, CA). NIH-3T3 mouse fibroblasts were maintained under almost similar conditions, except for a lowered glucose concentration (1 g/liter) in the minimum essential medium. One day prior to transfection, 5 ϫ 10 5 cells/dish were plated into 100-mm Petri dishes, and fresh medium was added 1 h prior to transfection. Cells in each plate were transfected with 20 g of DNA (isolated using the Qiagen Maxi-prep, Qiagen Inc., Chatsworth, CA) for 16 h by the calcium phosphate method and placed into fresh medium. Seventy-two hours post-transfection, the cells were washed with 1 ϫ phosphate-buffered saline, harvested, and lysed for 20 min in 200 l of Promega lysis reagent (Promega, Madison, WI), the cell lysates were pelleted, and the supernatants were frozen until assayed using the Luciferase Assay Kit (Promega) in a Monolight 2010 Luminometer (Analytical Luminescence Laboratory, San Diego, CA).

Generation of Constructs and Analysis of mKv1.4 Expression in Xenopus Oocytes
Expression Constructs-The Kv1.4 coding region was cloned into the pTM-1 vector (3). The Kv1.3/Kv1.4 chimeric construct was created as follows: the 3Ј-NCR BglII/EcoRI fragment of mKv1.4 (containing five ATTTA motifs) was inserted downstream of the coding region of the reporter molecule, Kv1.3, containing an antibody-recognizable epitope (gene 10) at its N terminus. This construct was linearized at either a SalI site (upstream of the ATTTAs) or an EcoRI site (downstream of all five ATTTAs), and cRNA was prepared. In a second chimeric construct, an oligonucleotide (Chemgenes, Needham, MA) corresponding to nt 395-447 of the 3Ј-NCR of mKv1.4 (ATTTAs 1-3) was linked to the 3Ј-end of Kv1.3. In a third construct, an oligonucleotide containing AGTGAs in place of the three ATTTA motifs (bp 395-447 of the 3Ј-NCR) was attached to the 3Ј end of Kv1.3. The latter two constructs were linearized at the EcoRI site (downstream of the three ATTTAs or three AGTGAs), and cRNA was prepared. The cRNAs from all these constructs are identical in their Kv1.3-derived sequences (5Ј-NCR, coding region, and proximal 0.2 kb of the 3Ј-NCR), and differ only in the lengths of the Kv1.4-derived 3Ј-NCRs which they contain. cRNA from each construct was injected into Xenopus oocytes and evaluated as described (3,11).
Analysis of cRNA Isolated from Oocytes-cRNA for each of these constructs was synthesized and radiolabeled with [ 32 P]CTP. Unincorporated nucleotides were removed by centrifugation through a Sephadex C-50 chromatography column (Boehringer Mannheim) followed by ethanol precipitation, and the cRNA concentration was adjusted to 1 mg/ml. Forty-six ng of cRNA were injected into each oocyte, and total RNA was extracted from 7 oocytes for each data point, either at time 0 (immediately after injection) or after 48 h. Oocytes were lysed in 1.2 ml of cell lysis solution (Promega), RNA-extracted, and run on a 1.5% agarose gel, autoradiographed to quantitate the levels of extracted RNA (X-omat film, Eastman Kodak, Rochester, NY), scanned using a Hewlitt-Packard Scanjet IIcx with a transparency adaptor, and analyzed on a Macintosh computer using the public domain NIH Image program. 2 Isolation of Protein from Oocytes and Western Blot Analysis-For each experiment, cRNA was injected into six oocytes; 48 h later, the oocytes were washed in fresh, ice-cold oocyte Ringer solution (96 mM NaCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM Hepes, pH 7.4, 290 -320 mosM), and completely lysed in ice-cold buffer (7.5 mM Na 2 HPO 4 , pH 7.4, 1 mM EDTA) containing a mixture of protease inhibitors (1 mg/ml leupeptin, 20 mg/ml phenylmethylsulfonyl fluoride, and 1 mg/ml pepstatin A (Sig-ma). Oocyte yolk was removed by centrifugation (400 ϫ g, 5 min at 4°C), the membranes were pelleted (12,000 ϫ g, 30 min, 4°C), washed once in lysis buffer, and resuspended in Laemmli loading buffer (12). The membrane-associated proteins were electrophoresed in a 10% SDSpolyacrylamide gel (12), electroblotted onto a polyvinylidene difluoride membrane (Millipore Corp.) in Towbin buffer (192 mM glycine, 25 mM Tris base in 20% methanol), and blocked with 5% powdered milk. Western blot analysis was then performed using the mouse anti-gene 10 antibody (T-7 Tag, Novagen, Madison, WI; 1:4000 dilution) and the horseradish peroxidase-labeled goat-anti-mouse second antibody (Caltag, San Francisco, CA, 1:2000 dilution) as immunoprobes; the bands were visualized by the Enhanced Chemiluminescent System (Amersham) and exposed to Hyperfilm (Amersham, Buckinghamshire, UK) for 1-5 min. Autoradiographs were scanned and analyzed as described.

Ribonuclease Protection Assays
Total RNA was extracted from mouse brain using Promega's total RNA isolation kit (Promega). This RNA (10 g) was hybridized to an antisense cRNA probe derived from the mKv1.4 genomic clone. The 1.4-kb SacI/BamHI fragment was linearized at MspI (580 bp), and a [ 32 P]dCTP (ICN, Costa Mesa, CA) radiolabeled antisense cRNA probe was generated using T7 RNA polymerase. The total RNA and probe were hybridized overnight at 47°C in hybridization buffer (4 M NaCl, 400 mM PIPES, 10 mM EDTA, and 80% formamide). The reaction was digested with RNase A and T1 (Sigma) for 1 h at room temperature. The digestion reaction was stopped with 10% SDS, followed by phenol/ chloroform extraction and ethanol precipitation. The precipitated pellet was resuspended in 5 l of water and 4 l of loading buffer, heated at 85°C, and then loaded onto a 6% denaturing polyacrylamide gel. The gel was dried and autoradiography was performed overnight (Ϫ70°C, intensifying screen). In the negative control, yeast tRNA was substituted for the mouse brain RNA. A separate lane contained the fulllength undigested probe.

Primer Extension Experiments to Define the Transcription Start Site
A synthetic oligonucleotide, 5Ј-GACAGCAGCGATCACTTG-3Ј, representing the sequence 53-70 nt downstream of the PstI site in the genomic region thought to be near the 5Ј-end of mKv1.4 (3), was annealed to 3 mg of total mouse brain RNA at 55°C for 3 min and placed on ice. To this was added [ 32 P]dCTP (25 mCi in 1 ml, 3000 Ci/mmol) and a mixture consisting of unlabeled deoxynucleotides (1 mM each dATP, dGTP, dTTP, Perkin-Elmer Corp., Norwalk, CT), reverse transcriptase (20 units,Promega), and RNase inhibitor (20 units, Boehringer Mannheim) in a final reaction volume of 50 l, and the mixture was incubated for 10 min at 24°C, followed by 1 h at 40°C. The product was concentrated, mixed with acrylamide gel loading dye, and loaded onto a 6% acrylamide sequencing gel (6 ml/lane) adjacent to a sequencing reaction of the genomic clone primed with the identical oligonucleotide (20) (Sequenase, U. S. Biochemical Corp.). These were then electrophoresed and autoradiographed.

Tissue Distribution of Kv1.4 in the Mouse-A Northern blot
probed with a mKv1.4-specific 3Ј-NCR fragment revealed hybridizing ϳ3.5and ϳ4.5-kb bands in mouse brain and heart poly(A) ϩ mRNA, a weakly hybridizing ϳ2.5-kb band in skeletal muscle, and additional faint signals in both lung and muscle; the probe did not hybridize detectably to mRNA from spleen, liver, kidney, or testes (Fig. 1). These results are consistent with previous reports describing the tissue distribution of the rat and human Kv1.4 homologues (4,5,10,13).
Mapping the Mouse Kv1.4 Transcripts-DNA fragments derived from six regions of the genomic clone ( Fig. 2A) were used to probe poly(A) ϩ mRNA from mouse brain (Fig. 2B). The 4.5-kb band was found to hybridize with probes 1-5, while the 3.5-kb band was only detected by probes 1-3. Probe 6 hybridized to neither transcript, although it did hybridize to the Kv1.4 genomic clone used as a positive control (not shown); reprobing this lane with probe 4 revealed the expected 4.5-kb hybridizing band, indicating that the lack of hybridization to probe 6 was not due to the absence of mRNA in this lane. These hybridization patterns show that both of the mouse Kv1.4 transcripts share the 5Ј exon, the entire coding region, and at least 0.2 kb of the 3Ј-NCR (probe 3), while the 4.5-kb mRNA contains an additional ϳ900 nt in the 3Ј-NCR, the region spanned by probes 4 and 5.
Examination of the 3Ј-NCR sequence reveals a consensus poly(A) ϩ signal (AATAAA) 177 bp downstream from the stop codon ( Figs. 2A and 3). The 3.5-kb transcript must terminate at or near this poly(A) ϩ signal, since it does not hybridize with probe 4 (Fig. 2B) which begins 40 bp farther downstream ( Figs.  2A and 3). The 4.5-kb transcript terminates at or close to a second poly(A) ϩ signal located 934 bp 3Ј of the first, since it shows no hybridization with probe 6 which begins 40 bp downstream of this signal ( Figs. 2A and 3). The significance of the A/T-rich motifs in the 3Ј-NCR (Fig. 3B) is discussed below.
To delineate more precisely the 5Ј end of the transcript, we used a 580-bp MspI/BamHI probe (Fig. 4A) in RNase protection assays. Mouse brain RNA (lane 3, Fig. 4B) protected a radiolabeled ϳ280-bp fragment from RNase digestion, while the probe did not survive RNase treatment when incubated with control yeast tRNA (lane 2); the full-length undigested 580-bp probe is visible in lane 1. The size of the protected fragment suggests that the transcript initiation site is likely to be at or very close to the PstI site indicated by the asterisk in Fig. 2A.
In primer extension experiments (Fig. 4C), a single band was visible. This band aligns with an adenosine in the nearest sequencing lane, which is 3 bp downstream from the PstI site (CTGCAGCCA) and 1337 bp upstream of the methionine initiation codon; we therefore designated this nucleotide nt ϩ1 of mKv1.4 (Fig. 5).
Thus, the two mKv1.4 mRNAs (4.5 and 3.5 kb) begin at the same site and appear to be identical except for the presence of an additional longer 3Ј-NCR sequence in the 4.5-kb transcript. The proteins generated by either mRNA are identical since they use the same coding region. The "3.5-kb" transcript contains 3476 nt (excluding any poly(A) ϩ tail) comprised of 1337 nt of 5Ј-NCR, 1962 nt of protein-coding region, and 170 nt of 3Ј-NCR. The "4.5-kb" transcript contains an additional 948 nt of 3Ј-NCR, yielding a total length of 4415 nt. The 2.5-kb transcript which is faintly visible in skeletal muscle (Fig. 1) remains to be characterized. If it includes the entire 1962-bp coding region, it must contain no more than ϳ0.5 kb of total NCR. Assuming it uses the first polyadenylation site (yielding 170 nt of 3Ј-NCR), it would include only ϳ370 nt of 5Ј-NCR.
The 3Ј-NCR Alters Functional Expression of the Longer Transcript-Do the two mKv1.4 mRNAs differ functionally? To address this question, we made a mKv1.4 expression construct containing the entire coding region which was linearized either at BglII or EcoRI sites (Fig. 6A); the BglII-linearized version ends 212 bp downstream of the stop codon and corresponds

FIG. 3. Sequence of the 3-NCR between the two polyadenylation signals.
A, polyadenylation signals and ATTTA motifs. Two consensus polyadenylation signals and five ATTTA motifs (see text) are indicated by asterisks and numbered 1-5. The bracketed area represents the sequence included in the synthetic oligonucleotide used in the experiments described in Fig. 7. The first BglII site, 40 bp downstream of the first polyadenylation signal, represents the end of probe 3 and the beginning of probe 4 ( Fig. 2A). Probe 5 spans the region between the EcoRI site and the downstream BglII, while probe 6 begins at this BglII site. B, conservation of ATTTA motifs. The mouse sequence shown was used in experiments described in Fig. 7. Note that the first and third ATTTAs are conserved in mouse, rat, and bovine Kv1.4, while the second motif is only found in mouse and rat. The fourth and fifth ATTTAs are also present in rat; however, the published sequences of bovine and human Kv1.4 do not extend into this region. closely to the 3.5-kb mRNA, while the EcoRI-linearized form terminates 892 bp farther 3Ј and represents the longer transcript. Both cRNAs produced typical rapidly inactivating K ϩ currents when expressed in Xenopus oocytes, although the shorter transcript produced ϳ4 -5-fold more mKv1.4 currents than the longer one (Fig. 6B). Clearly, the presence of the additional 3Ј-NCR sequence in the 4.5-kb mRNA reduces functional channel expression in oocytes (Fig. 6, B and C).
This region is 66% AU-rich and contains five AUUUA repeats that are conserved in rat and bovine homologues of Kv1.4 (Fig. 3B, 1 and 3). These motifs are also present in the 3Ј-NCRs of many other genes expressed in diverse mammalian cell types (14 -18). The presence of three or more such repeats, either by themselves or as part of the larger nonanucleotide motif (UUAUUUA(A/U)(A/U)), has been reported to reduce mRNA stability in almost every case examined (e.g. Refs. 14 -18). In view of this, it seems very likely that the 3Ј-NCR of Kv1.4 has a destabilizing effect on the 4.5-kb transcript in mammalian cells. In Xenopus oocytes, however, AUUUA-mediated mRNA degradation is minimal; this mechanism therefore cannot account for the decreased expression of the 4.5-kb transcript in oocytes.
The octanucleotide repeat, UUAUUUAU, has been reported to inhibit translation of diverse messages in oocytes and in in vitro translation systems (19 -23). While the 3Ј-NCR of mKv1.4 does not contain this octanucleotide motif (Fig. 5), we considered the possibility that the shorter AUUUA motif might possess translation inhibitory activity. To test this idea, we coupled a reporter gene (gene-10-containing Kv1.3) to mKv1.4 3Ј-NCR sequences containing 0 -5 ATTTA repeats (Fig. 7A). The presence of 5 AUUUAs decreased expression of the reporter protein 11-fold, while three AUUUAs caused 3.5-fold suppression (Fig. 7C); mutating these three AUUUAs to AGUGA completely relieved the suppression. The total amount of protein loaded in each of these lanes was roughly equal, as evidenced by the Coomassie stain of a 100-kDa endogenous oocyte protein (Fig. 7D).
The presence of the AUUUA repeats did not appreciably alter levels of Kv1.4 cRNA during the course of the experiment (Fig. 7B). The levels of [ 32 P]CTP-labeled reporter-gene cRNA remained unchanged even after 48 h, regardless of the presence or absence of AUUUAs (Fig. 7B), and densitometric analysis revealed no more than an 8% difference in the intensity of the bands. Thus, the observed AUUUA-mediated reduction in functional expression of the longer 4.5-kb transcript in oocytes must be mediated via suppression of translation. Using the reasoning of Kruys and co-workers (19 -23), we argue that the AUUUA-motifs in the Kv1.4 3Ј-NCR might also reduce translational efficiency of the 4.5-kb mRNA in mammalian cells.  An ϳ1000-bp region of sequence from the genomic clone, flanking the PstI site indicated by the asterisk in Fig. 2A, is shown. The underlined nucleotide at position ϩ1 indicates the transcript initiation site observed in primer extension experiments. Consensus sequences for SP-1 (CCGCCC), AP-2, and the fibroblast enhancer element are shown, as are four E box repeats and 2 MspI sites. The sequence extends through the end of exon 1. The upstream ATGs in the 5Ј-NCR sequence contained in exon 1 are italicized and underlined. An additional seven ATGs are present in the 5Ј-NCR sequence contained in exon 2 (3). Identifying the mKv1. 4 Promoter and Enhancer(s)-To identify the mKv1.4 promoter, 5Ј-flanking regions upstream of exon 1 were engineered either into the pLuciferase-enhancer or the basic vector (see "Materials and Methods"). These were transfected into two cell lines (C 2 C 12 , NIH-3T3), and luciferase activity was measured; of the two, only C 2 C 12 cells are known to express Kv1.4 (10).
Kv1.4 Promoter Activity Is Not Cell Line-specific-A 2.1-kb SacI/BamHI fragment was found to confer very high levels of promoter activity in both cell lines (Fig. 8). Deletion of an internal 464-bp PstI fragment eliminated promoter activity suggesting that the basal promoter is located within this region. Consistent with this notion, this PstI fragment, located just upstream of the transcription initiation site (Figs. 4 and 5), induced transcription of the luciferase gene, although at a substantially lower level than the entire SacI/BamHI fragment. These data indicate the presence of additional enhancer element(s) outside the 464-bp PstI fragment.
Since DNA fragments may exhibit spurious promoter activity in the presence of powerful enhancers (e.g. SV40 enhancer), it was important to determine if the PstI fragment could induce transcription in the absence of the SV40 enhancer. For this purpose, the fragment was engineered into the basic vector (lacking an enhancer), and transcription of the luciferase gene was measured following transfection into NIH 3T3 cells. The PstI fragment clearly contains a fully functional promoter, since it is capable of initiating transcription of the luciferase gene without the help of an exogenous enhancer (Fig. 8, legend). This region, seen in Fig. 5, is 66% (G/C)-rich, lacks canonical TATAAA and GGCAATCT boxes (24), contains three SP1 sites (CCGCCC), multiple E-boxes (CANNTG) (Myod-1 binding sites), one each of the motifs for fibroblast enhancer (CCAAT) and AP2 (ATTTGC), and resembles the promoters of many housekeeping genes (25)(26)(27)(28). DISCUSSION We report the mapping of two mKv1.4 transcripts, 4.5 and 3.5 kb, that are expressed in brain and heart. These two are identical except in the 3Ј-NCR, with the longer transcript terminating at a polyadenylation signal 934 nt downstream from the signal utilized by the 3.5-kb form. Although the same coding region is utilized by both transcripts to generate identical mKv1.4 proteins, we have shown that the 3Ј-NCR regulates the levels of functional channels via AU-dependent regulation of translation efficiency and most likely also of mRNA stability. The presence of multiple ATGs lying upstream of the initiation methionine in the 5Ј-NCRs of both transcripts would be expected to decrease translational efficiency (Fig. 5). Thus, the unusually long and well conserved NCRs of Kv1.4 play important roles in modulating Kv1.4 channel expression.
The promoter for this gene has little if any tissue specificity, is located in a GC-island containing three SP1 binding sites (CCGCCC), and is modulated by at least one region containing enhancer activity. The high GC content of the Kv1.4 promoter might reduce its accessibility to RNA polymerase (e.g. Ref. 29). The mechanisms responsible for tissue-specific expression of this gene have yet to be determined. Positive and negative transcriptional regulators are known to be responsible for tissue-and developmental stage-specific expression of sodium channel genes (30 -32), and similar elements upstream of the SacI/BamHI fragment might be required for tissue-specific expression of mKv1.4. Alternatively, mKv1.4 gene expression might be inactivated by methylation in tissues other than the brain, heart, and skeletal muscle; this GC-rich region, in fact, contains multiple potential methylation sites (33)(34)(35). Future studies will provide answers to these questions and determine whether other K ϩ channels use similar mechanisms to control their expression.
The tight transcriptional and post-transcriptional regulation . Construct 4, this construct is identical with construct 3, except that the three ATTTAs are mutated to AGTGAs. B, radiolabeled RNA extracted from oocytes immediately (time 0) or 48 h following injection of 32 P-labeled cRNA. The RNA seen in the right two lanes is larger because of the additional 907 bp of 3Ј-NCR present in construct 2. C, Western blot of Kv1.3 protein extracted from oocytes 48 h after injection of cRNA. A 60-kDa band, corresponding to the predicted size for Kv1.3, is visible. D, control for protein loading. A Coomassie stain of the identical gel used in C, after electroblotting, shows a 100-kDa endogenous oocyte protein, and serves to demonstrate that similar amounts of protein were loaded in each lane. of mKv1.4 expression may reflect the need to maintain the number of functional channels at closely defined levels in mammalian tissues. A rapidly inactivating K ϩ current like mKv1. 4, which can open in the subthreshold range of excitation, would likely reduce the excitatory effects of depolarizing currents on neurons in a time-dependent fashion (6); a deficit of Kv1.4 current might therefore enhance neural excitability, while increased numbers of such channels may have hypoexcitatory effects. Alteration of transcriptional or post-transcriptional processes might be associated with pathological states, possibly contributing, for example, to the enhanced Kv1.4 mRNA expression seen in cardiac hypertrophy (36).