Electrophysiological Characterization of an Alternatively Processed ERG K+ Channel in Mouse and Human Hearts
Abstract Mutants of HERG, the human form of ERG (the ether-a-go-go–related K+ channel gene), are responsible for some forms of the long-QT syndrome, an abnormality of cardiac repolarization. HERG was cloned from brain and has properties similar but not identical to the rapidly activating component of the native cardiac K+ channel current (IKr). We identified in the mouse an alternatively processed form of ERG (MERG B) that is expressed abundantly in heart but only in trace amounts in brain. MERG B has a unique 36–amino acid NH2-terminal domain that is strongly basic and considerably shorter than the 376–amino acid NH2-terminal domain of HERG. When expressed in Xenopus oocytes, the kinetics of activation and deactivation of the MERG B current were best fit by a biexponential function, with the fast components dominant over the slow components. The fast component of activation had a mean τ value of 163±16 ms at −20 mV and 8±4 ms at +20 mV (n=4). The fast component of deactivation had a mean τ value of 145±29 ms at −20 mV and 12±4 ms at −90 mV (n=4). The MERG B current was blocked by the selective IKr blocker, dofetilide, with an IC50 of 54 nmol/L. In addition, we isolated HERG B, the human homologue of MERG B, which has electrophysiological characteristics qualitatively similar to those of MERG B. We have identified ERG B, an alternatively processed isoform of the ERG gene, expressed selectively in heart and with electrophysiological characteristics similar to those of native cardiac IKr.
Long-QT syndrome is generally manifested as torsade de pointes ventricular tachycardia and/or sudden cardiac death in humans. One form of congenital long-QT syndrome is associated with mutations in ERG.1 2 3 Human ERG (HERG) encodes a delayed rectifying K+ channel that is thought to be important in repolarization of the cardiac action potential.2 In addition, Smith et al4 speculated that the kinetics of inactivation and reactivation of the HERG current may be important in the physiological suppression of arrhythmias. HERG has electrophysiological properties similar but not identical to IKr.2 5 The time constants of activation and deactivation of HERG are 4 to 10 times slower than native IKr as expressed in guinea pig and mouse cardiac myocytes.5 Since HERG was originally cloned from human brain,6 7 we reasoned that cardiac myocytes might express an isoform of HERG with similarity to cardiac IKr. The present study shows the presence of an alternatively processed ERG mRNA in mouse and human hearts termed MERG B and HERG B, respectively. The MERG B current has kinetics of activation and deactivation similar to those of native IKr and is blocked by the selective IKr channel blocker, dofetilide, with an IC50 of 54 nmol/L.
Materials and Methods
Cloning of MERG B
A fragment of the mouse heart ERG message encoding the pore and cyclic nucleotide binding domains was isolated by PCR using degenerate oligonucleotides [5′-CTCACCAGIGTIGGITT(C/T)GG-3′and 5′-GAGAA(C/T)TCIGG(G/A)TACATITC-3′] based on the known HERG sequence. This fragment was labeled with 32P and used to screen an AT-1 cDNA library. AT-1 is a propagated atrial tumor derived from transgenic mice.8 The cDNA library was synthesized by reverse transcription of AT-1 mRNA with Superscript II (BRL) and random primers. The cDNAs were ligated into Lambda Zap Express (Stratagene). Clones that hybridized to the MERG probe were purified and excised in the plasmid pBK-CMV, restriction-mapped, and sequenced.
A full-length MERG B cDNA was synthesized from clones C1-MERGB and C8-MERGA using a common Xho I restriction site. In order to demonstrate that this clone truly represents the endogenous MERG B message, we carried out a full-length RT-PCR of MERG B from mouse atrium RNA (Fig 1b⇓). An antisense oligonucleotide primer (5′-TCTGCACAGGTCT-3′) that hybridizes to the 3′-untranslated region of MERG was designed. This primer functioned in the reverse transcription reaction at 40°C but not in the PCR reaction with a hybridization temperature of 65°C. Reverse transcription was carried out with 1 μg of mouse atrial RNA using Superscript II reverse transcriptase (GIBCO-BRL). For the PCR step, one oligonucleotide (ERGB52) was derived from 5′ MERG B–specific sequences (see Fig 2⇓); the second oligonucleotide (MERGC32, 5′-CACTAACTGCCTGGATCTGA-3′) spanned the translation stop codon (TAG) of the common ERG sequence. PCR was carried out for 36 cycles at 94°C for 1 minute, at 65°C for 1 minute, and at 72°C for 3 minutes at a magnesium concentration of 2 mmol/L using Tsg DNA polymerase (Sangon). The resulting RT-PCR product was electrophoresed on a 1% agarose-TAE gel and isolated on silica with the Ultra Clean 15 kit (Mo Bio). The product was directly sequenced with the Thermo-Sequenase cycle sequencing kit (Amersham).
In order to determine the genomic basis of MERG B expression, we made a 32P-labeled riboprobe (Promega) from the first 355 nucleotides of MERG B and screened a mouse genomic library in Lambda DASH II (kindly provided by Dr Derek Rancourt). Thirty-two of 5×105 plaques were positive, and of these, three were purified. Restriction fragments of the lambda clones were subcloned into p-Bluescript KS+ (Stratagene) and sequenced with Sequenase (USB).
Cloning of HERG B
In order to clone the human homologue of MERG B (HERG B), we designed oligonucleotides (ERGB52 and ERGC3, Fig 2⇑) on the basis of sequences from HERG and MERG B. These oligonucleotides were used in a PCR reaction with reverse-transcribed mRNA from human atrium. The resulting PCR product was cloned into the Sma I site of the pBluescript KS+ vector (Stratagene) and sequenced with Sequenase (USB). A full-length HERG B clone was synthesized from the HERG B PCR product and HERG A (kindly provided by Mark Keating) using a common Pst I restriction site. Note that the NH2-terminal seven amino acids of HERG B are encoded from the oligonucleotide ERGB52 that was derived from the mouse cDNA sequence.
RNA was prepared by the guanidinium isothiocyanate/cesium chloride centrifugation method.9 Total RNA (1 μg) and 2×107 molecules of the cRNA internal standard (determined by UV absorption) were reverse-transcribed in 16 μL using Superscript II reverse transcriptase (GIBCO-BRL) and random primers. Three 5-μL aliquots of each reverse-transcribed reaction were made for PCR. Oligonucleotide pairs for PCR are illustrated in Fig 2⇑ and consist of the following: for MERG A, MERGA51 and ERGC3; for MERG B, MERGB51 and ERGC3; and for cytoplasmic β-actin, 5′-GTTCCGATGCCCCGAGGCTCT-3′ and 5′-GCATTTGCGGTGCACGATGGA-3′.
Each PCR cycle was at 94°C for 1 minute, 56°C for 1 minute, and 72°C for 2 minutes and 30 seconds. Actin samples contained 2.0 mmol/L MgCl2 and were run for 19 cycles. MERG A and MERG B samples contained 1.1 mmol/L MgCl2 and were run for 22 cycles. The cycle numbers were optimized by plotting cycle number versus band density for a single starting concentration of ventricular RNA. Each sample contained 1 μCi of [32P]dCTP for detection by autoradiography. The samples were electrophoresed on 5% polyacrylamide-TBE gels before exposure on Kodak XAR-5 film. To control for variability in the efficiency of reverse transcription and subsequent amplification, an internal standard was created. The internal standard was constructed by first inserting a 262-bp fragment of DNA into the Eag 1 restriction site of MERG B (Figs 1a⇑ and 2⇑) to create ΔMERG B. The 262-bp fragment is an Sma I–Sca I restriction fragment from the mouse Kv1.1 gene. The GC composition of the fragment changes the overall GC content from 61% in MERG B to 60% in ΔMERG B. Standard curves of both the MERG B and ΔMERG B cDNAs did not indicate that this fragment significantly altered amplification during PCR reactions. In order to allow use of this internal standard for MERG A–specific RT-PCR reactions, we synthesized a composite oligonucleotide (5′-AGCGGACCCACAATGTCAGCATCTCCAGCCT-3′) containing both the MERGA51 and MERGB51 sequences. The ΔMERGB construct was used as a PCR template for the composite oligonucleotide and the common oligonucleotide ERGC3. The PCR product ΔMERG AB was ligated into pBluescript KS+, and the cRNA internal standard was synthesized with T7 RNA polymerase.
RNase Protection Assay
We subcloned a 269-bp Nco I–Eag 1 restriction fragment of MERG B into the pGex vector.10 The fragment includes 110 nucleotides of the MERG B–specific sequence and 159 nucleotides of the common MERG sequence (see Fig 1a⇑). The construct was linearized at the HindIII site of the pGex vector, and a 32P-labeled antisense RNA was synthesized from the T7 promoter of the pGex vector. The antisense RNA includes 66 nucleotides of vector sequence. Using the RPA II kit (Ambion), the labeled antisense probe was hybridized with 20 μg of total tissue RNA. The nonhybridized probe was digested with RNases A and T1. The hybridization product was precipitated, resuspended in a denaturing buffer, and run on a 6% polyacrylamide gel containing 8 mol/L urea and Tris/borate/EDTA buffer. Autoradiography was carried out on Kodak XAR film.
To study the electrophysiological properties of MERG B, it was subcloned into a Xenopus oocyte expression vector (pGex)10 and transcribed in vitro from the SP6 promoter. The cRNA transcript (≈4 ng) was injected into Xenopus oocytes, and the expressed currents were examined after 3 to 8 days using two-microelectrode voltage-clamp techniques. The oocytes were perfused with modified frog Ringer’s solution at room temperature (21°C to 22°C) containing (mmol/L) NaCl 114, KCl 2.5, MgCl2 1, CaCl2 1.8, and HEPES 10, pH 7.2–adjusted with NaOH. Niflumic acid was included (0.15 mmol/L) to block chloride currents. Glass microelectrodes were filled with 3 mol/L KCl with tip resistances of 0.5 to 2 MΩ. Oocytes were clamped with a Geneclamp 500 amplifier, and voltage clamp protocols were generated with pClamp software (Axon Instruments), a 486 personal computer, and a Digidata 1200 interface board (Axon Instruments). Currents were sampled at a rate of 2 kHz and were filtered at 1 kHz using a four-pole Bessel filter. The oocyte membrane was held at −80 mV between pulses.
All averaged and normalized data were presented as mean±SE unless otherwise indicated. Statistical significance among groups were determined using one-way ANOVA. A value of P<.05 was considered significantly different.
Cloning of MERG B
To clone the mouse homologue of HERG from cardiac cells, we used a PCR-derived fragment of MERG to screen a cDNA library constructed from atrial tumor cell (AT-1) mRNA.8 Of the 5 ×105 plaques screened, there were 37 positives, 10 of which were purified and excised in the plasmid pBK-CMV. There were six unique clones out of 10 (the library had been amplified once before screening). Three of the clones contained only the sequence that was COOH-terminal to the putative S1 transmembrane domain. Of the remaining three clones, the one with the largest insert (C8-MERGA) encoded 1138 amino acids and exhibited 95% amino acid sequence identity with HERG (Figs 1 to 3⇑⇑⇓). Only one substitution was present in the 400 amino acids between the S2-transmembrane and cyclic nucleotide binding domains. Hereafter, MERG A is defined as the mouse transcript that contains the 5′ end–specific sequences found in clone C8-MERGA, and the homologous HERG transcript is referred to as HERG A.
We identified two clones, C1-MERGB and C9-MERGB, which were identical over 1451 bp. Clone C1-MERGB was slightly longer than C9-MERGB, with an additional 10 bp at the 5′ end and an additional 109 bp at the 3′ end. These clones were identical to clone C8-MERGA over 1151 and 1042 nucleotides, respectively, in the putative transmembrane coding region but diverged from C8-MERGA in the amino-terminal coding region (Figs 1⇑ and 2⇑). Hereafter, the mouse transcript that contains the 5′ end–specific sequences found in the clones C1-MERGB and C9-MERGB is defined as MERG B. The unique 5′ end of C1-MERGB consists of 419 nucleotides. There is an in-frame ATG that is 118 nucleotides upstream from the point of divergence from the common sequence. This is the probable translation-initiation site of MERG B, since there is an in-frame stop codon (TAG) 180 nucleotides upstream, with no intervening in-frame ATGs. The 5′ end of the MERG B transcript therefore encodes a 36–amino acid domain without similarity to the 376–amino acid NH2-terminal domain of HERG A. This unique region of MERG B carries a strong net positive charge, since it contains eight basic compared with only two acidic residues.
Since C1-MERGB and C9-MERGB both lack COOH-terminal coding regions, a full-length MERG B cDNA was synthesized from C1-MERGB and C8-MERGA using a common Xho I restriction site (Fig 1⇑). In order to demonstrate that our synthetic MERG B clone truly represents the endogenous MERG B message, we carried out a full-length RT-PCR of MERG B from mouse atrium RNA (Fig 1b⇑). For the PCR step, one oligonucleotide (ERGB52) was derived from 5′ MERG B–specific sequences; the second oligonucleotide (MERGC32) spanned the translation stop codon (TAG) of the common ERG sequence. We obtained a PCR product of ≈2300 nucleotides, which was the same size as the product obtained by amplifying the synthetic MERG B cDNA. No product was obtained when the reverse transcriptase had been previously heat inactivated (Fig 1b⇑). The resulting mouse atrial RT-PCR product was directly sequenced and was found to be identical to MERG B.
Cloning of HERG B
To clone HERG B, mRNA from human atrium was reverse-transcribed into cDNA. This cDNA was amplified in a PCR reaction with an oligonucleotide pair based on MERG B and common ERG sequences (ERGB52 and ERGC3; see Fig 2⇑). We obtained an RT-PCR product of the appropriate size, which was cloned and then sequenced (Fig 2⇑). The 5′ end–specific region of the HERG B PCR product has four differences out of 98 nucleotides compared with MERG B but shares no similarity with HERG A in this region (Figs 1 to 3⇑⇑⇑). In contrast, the 3′-end common region of the HERG B PCR product is identical to HERG A over 237 nucleotides but differs from the common region of both MERG A and B at 28 of these 237 nucleotides. Similar to MERG B, the NH2-terminal region of HERG B carries a strong net positive charge, since it contains nine basic residues compared with only one acidic residue.
Characterization of the MERG Gene
The presence of a large region of absolute identity flanked by a region of great diversity is indicative of alternative mRNA processing. To establish the basis for alternative processing of MERG gene primary transcripts, we cloned the appropriate regions of mouse genomic DNA (Fig 3⇑). The NH2-terminal sequence of MERG A is encoded by at least four exons, designated A1 to A4. In addition, we identified the first exon (C1) that is common to MERG A and MERG B. This exon is located ≈3800 nucleotides downstream from exon A4. The intron 3′ to exon C1 is homologous to intron 1 of the partially mapped HERG gene.1 Exon B1, encoding the novel NH2-terminal sequence of MERG B, was located between exons A4 and C1 (Fig 3⇑).
The divergent 5′ end of MERG B may result from an alternative promoter and/or alternative mRNA splicing. Since the 5′ ends of clones C1-MERGB and C9-MERGB are both located in the genomic sequence between exons A4 and B1 (Fig 2a⇑), it is possible that transcription of MERG B is initiated from an independent promoter within this region. Further studies are required to establish the nature of the elements that regulate MERG B expression.
Distribution of MERG B
To study the tissue expression of MERG A and MERG B by RT-PCR, we synthesized two sets of oligonucleotide primers with one primer common to both sets (see Figs 1⇑ and 2⇑). Note that in both cases the primer sets span introns in the genomic DNA (see Fig 3⇑). An internal standard was synthesized that was recognized by both primer sets. The RT-PCR products derived from mouse heart RNA were of appropriate size and restriction-digest pattern for both primer sets. No products were detected when the reverse transcriptase had been previously heat-inactivated (not shown). The RT-PCR method demonstrated that MERG A is abundantly expressed in both the atrium and ventricle of the heart and in the brain, whereas MERG B is abundantly expressed only in the heart. Further evidence for the tissue distribution of MERG A and MERG B was provided by ribonuclease protection studies. A probe was designed that hybridized to 269 bases of the MERG B mRNA compared with 159 bases of the MERG A mRNA. The ribonuclease protection analysis confirms that MERG B is relatively heart specific, whereas MERG A is abundantly expressed in the heart and brain (Fig 4b⇓). It also indicated that trace amounts of MERG B are found in the brain, lung, and skeletal muscle and that trace amounts of MERG A are present in the lung. RT-PCR studies that are not shown also indicated trace amounts of MERG A and MERG B in the stomach. Both RT-PCR and ribonuclease protection also indicated that MERG B is more abundant in the atrium than ventricle, whereas, on average, MERG A is expressed in similar amounts in both tissues.
Electrophysiology of MERG B
To study the electrophysiological properties of MERG B, a transcript was synthesized in vitro and injected into Xenopus oocytes. Fig 5⇓ shows the MERG B current recordings before (Fig 5A⇓) and after (Fig 5B⇓) dofetilide (10 μmol/L). The stimulation protocol is shown in the inset. The dofetilide-resistant current (Fig 5B⇓) showed only a linear leak current resembling that seen in water-injected oocytes. The dofetilide-sensitive current is shown in Fig 5C⇓. Fig 5D⇓ shows the mean current-voltage relationships for drug-free, dofetilide-sensitive, and dofetilide-resistant currents (n=7). The drug-free and the dofetilide-sensitive currents have very similar current-voltage relationships, and both currents show the characteristics of inward rectification. Fig 5E⇓ shows the mean current-voltage relationships of the tail currents recorded in the drug-free condition and during treatment with 10 and 300 nmol/L dofetilide (n=5). The amplitude of the tail currents progressively increased with depolarization up to a potential of 10 mV but plateaued thereafter. The normalized peak tail currents were plotted as a function of test potential and fitted to the Boltzmann function. The half-activation voltage was −22±0.6 mV, and the slope factor was 10±1 mV (n=4) (figure not shown). Dofetilide at 10 nmol/L modestly but significantly reduced the amplitude of the tail currents at positive potentials. Dofetilide at 300 nmol/L substantially blocked the tail currents. Fig 5F⇓ shows the concentration dependence of block of the tail currents relative to baseline. The IC50 of dofetilide was 54 nmol/L. In subsequent figures, the MERG B current is presented as the dofetilide-sensitive current.
Since the time-dependent activating current of IKr and HERG A was increased by increases in [K+]o, we assessed whether a similar characteristic would be seen with MERG B. The effect of changes in [K+]o on the time-dependent activating current of MERG B is shown in Fig 6A⇓ and 6B⇓. Fig 6A⇓ shows the records at 1 mmol/L [K+]o; Fig 6B⇓ shows the records at 10 mmol/L. Mean current-voltage relationships are shown at [K+]o of 1, 2.5, and 10 mmol/L in Fig 6C⇓ (n=5). As has been reported for HERG A, increasing [K+]o increased the time-dependent activating current. To further examine the ion selectivity of the expressed MERG B channel, the reversal potential of tail currents was evaluated using the standard two-pulse protocol. The reversal potential was measured at five different levels of [K+]o: 1, 2.5, 10, 20, and 40 mmol/L. With increased [K+]o, there was a reciprocal reduction in [Na+]o. The reversal potentials versus log [K+]o data were fit to the Goldman-Hodgkin-Katz equation (Fig 5D⇑). The regression line for this plot had a terminal slope of ≈58 mV per decade, with a best-fit permeability ratio of [Na+]o to [K+]o of 0.026, indicating the K+ selectivity of this channel (panel D).
Fig 7A⇓ shows representative examples of the activation time course fitted to monoexponential and biexponential functions. The stimulation protocol is shown in the inset. Activation was inadequately fit by a monoexponential function (inset, Fig 7A⇓) but was well fit to a biexponential function, with the fast component dominant over slow. The mean time constants of the fast components of activation had τ values of 163±16 ms at −20 mV and 8±4 ms at +20 mV (n=4). The mean time constants of the fast components of activation of MERG B current are plotted as a function of the test potentials in Fig 7C⇓.
To assess deactivation kinetics, a double-pulse protocol was used. The first pulse was introduced from a holding potential of −80 mV to a test potential of +20 mV for 1.75 s to fully activate the current. The first pulse was followed by a second pulse, which repolarized the cells to a variety of test potentials ranging from −30 to −90 mV for 10 seconds. Fig 7B⇑ shows representative examples of the deactivation time course fitted to monoexponential and biexponential functions. Deactivation was inadequately fit by a monoexponential function (inset, Fig 7B⇑) but was well fit to a biexponential function, with the fast components dominant over slow components. The mean time constants of the fast components of deactivation of MERG B current are plotted as a function of the test potentials in Fig 7C⇑. The mean time constants for the rapid component of deactivation ranged from 145±29 ms at −20 mV to 12±4 ms at −90 mV (n=4).
Electrophysiology of HERG B
There existed a possibility that the divergent properties of MERG B and HERG A currents resulted from species differences rather than from the structural difference in the NH2-terminal region. Therefore, we constructed a full-length HERG B cDNA. Representative electrophysiological characteristics of the MERG B and HERG B K+ channels are compared in Fig 8A⇓ and 8B⇓. The stimulation protocol is shown in the inset. Like the MERG B current, the HERG B current has characteristics of rapid activation and inward rectification. The activation process of the HERG B current is also best fit by a biexponential function, with fast constants dominant over slow. The mean fast component of activation had a τ value of 267±23 ms at −20 mV and 85±12 ms at +20 mV. Deactivation of the HERG B current was voltage dependent and best fit a biexponential function. The time constants for the rapid component of deactivation changed from 141±38 and 15±2 ms for return potentials of −20 and −90 mV, respectively. These values are quite similar to that seen with MERG B.
We have shown that alternative processing of the ERG gene in humans and mice leads to protein products with strikingly different NH2-terminal regions. The relatively short and highly basic NH2 terminals of HERG B and MERG B appear to confer distinct electrophysiological properties compared with those previously reported for HERG A. The ERG B current has activation and deactivation time constants that are rapid compared with those of IKr, whereas those of HERG A are slow compared with those of IKr.5 11 12 13 14 15 16 As indicated by our tissue distribution studies, it is possible that native IKr results from a mixture of ERG A and ERG B homomultimers or from heteromultimers. It seems probable that changes in the ratio of expression of ERG A to ERG B may alter the kinetics of the endogenous IKr.
Class III antiarrhythmic drugs, such as dofetilide, are known to target IKr.13 14 15 16 17 MERG B was blocked by dofetilide, at a mean IC50 value of 54 nmol/L. This is similar to results indicating that HERG A is also blocked by dofetilide.13 14 15 16 17 This finding is not surprising in that the dofetilide binding site is likely located in the H5 pore region, which is common to the ERG A and ERG B isoforms.17
The discovery of the ERG B isoform has a number of other implications. First, previous studies have focused on mutants of HERG A as the defective gene product responsible for abnormalities of repolarization associated with the congenital long-QT syndrome. However, all known naturally occurring mutations of HERG occur in the region common to HERG A and HERG B1 2 3 and will therefore be manifested in both isoforms. A further important finding of the present study is that ERG B is a relatively cardiac-specific isoform and that it is a target for class III antiarrhythmic drugs. Therefore, it is plausible that drugs may be developed that interact with only the ERG B isoform and thereby produce cardiac-specific effects. In addition, the specificity of tissue distribution of ERG A and ERG B suggests that their expression is independently regulated. It may be possible to upregulate or downregulate these isoforms independently, thus altering the repolarization phenotype. This possibility will have implications for gene therapy.
Selected Abbreviations and Acronyms
|HERG, MERG||=||human and mouse ERG|
|I Kr||=||rapidly activating component of cardiac delayed rectifier K+ current|
|PCR||=||polymerase chain reaction|
This study was supported by the Alberta Heritage Foundation for Medical Research (Edmonton, Alberta, Canada), the Medical Research Council of Canada (Ottawa, Ontario, Canada), the Heart and Stroke Foundation of Canada (Ottawa, Ontario, Canada), and Pfizer Central Research (Sandwich, UK). We also acknowledge Robert Sheldon, Robert Clark, and Kelly Thorstad for their assistance with this work.
- Received January 13, 1997.
- Accepted August 1, 1997.
- © 1997 American Heart Association, Inc.
Sanguinetti MC, Curran ME, Spector PS, Keating MT. Spectrum of HERG K+ channel dysfunction in an inherited cardiac arrhythmia. Proc Natl Acad Sci U S A. 1996;93:2208-2212.
Sanguinetti MC, Jurkiewicz NK. Two components of delayed rectifier K+ current: differential sensitivity to black by class III antiarrhythmic agents. J Gen Physiol. 1990;96:195-215.
Warmke JW, Ganetzky B. A family of potassium channel genes related to EAG in Drosophila and mammals. Proc Natl Acad Sci U S A. 1994;91:3438-3442.
Trudeau MC, Warmke JW, Ganetsky B, Robertson GA. HERG, a human inward rectifier in the voltage gated potassium channel family. Science. 1995;269:92-95.
Field LJ. Atrial natriuretic factor-SV40 T antigen transgenes produce tumors and cardiac arrhythmias in mice. Science. 1988;239:1029-1033.
Hopkins WF, Allen ML, Houamed KN, Tempel BL. Properties of voltage-gated K+ currents expressed in Xenopus oocytes by mKv 1.1, mKv 1.2 and their heteromultimers as revealed by mutagenesis of the dendrotoxin-binding site in mKv1.1. Pflugers Arch. 1995;428:382-390.
Duff HJ, Feng Z-P, Sheldon RS. High- and low-affinity site for [3H]dofetilide binding to guinea pig myocytes. Circ Res. 1995;77:718-725.
Yang T, Wathen MS, Felipe A, Tamkun MM, Snyders DJ, Roden DM. K+ currents and K+ channel mRNA in cultured atrial cardiac myocytes (AT-1 cells). Circ Res. 1994;75:870-878.
Spector PS, Curran ME, Keating MT, Sanguinetti MC. Class III antiarrhythmic drugs block HERG, a human cardiac delayed rectifier K+ channel: open channel block by methanesulfonanilides. Circ Res. 1996;78:499-503.
Kiehn J, Lacerda AE, Brown AM. Molecular physiology and pharmacology of HERG: single-channel currents and block by dofetilide. Circulation. 1996;94:2572-2576.
Snyder DJ, Chaudhary A. High affinity open channel block by dofetilide of HERG expressed in a human cell line. Mol Pharmacol. 1996;49:949-955.
Wang L, Duff HJ. Identification and characterization of delayed rectifier K+ current in fetal mouse ventricular myocytes. Am J Physiol. 1996;270:H2088-H2093.