Cloning and functional characterization of the smooth muscle ether-a-go-go-related gene K+ channel. Potential role of a conserved amino acid substitution in the S4 region.

The human ether-a-go-go-related gene (HERG) product forms the pore-forming subunit of the delayed rectifier K(+) channel in the heart. Unlike the cardiac isoform, the erg K(+) channels in native smooth muscle demonstrate gating properties consistent with a role in maintaining resting potential. We have cloned the smooth muscle isoform of HERG, denoted as erg1-sm, from human and rabbit colon. erg1-sm is truncated by 101 amino acids in the C terminus due to a single nucleotide deletion in the 14th exon. Sequence alignment against HERG showed a substitution of alanine for valine in the S4 domain. When expressed in Xenopus oocytes, erg1-sm currents had much faster activation and deactivation kinetics compared with HERG. Step depolarization positive to -20 mV consistently produced a transient outward component. The threshold for activation of erg1-sm was -60 mV and steady-state conductance was approximately 10-fold greater than HERG near the resting potential of smooth muscle. Site-directed mutagenesis of alanine to valine in the S4 region of erg1-sm converted many of the properties to that of the cardiac HERG, including shifts in the voltage dependence of activation and slowing of deactivation. These studies define the functional role of a novel isoform of the ether-a-go-go-related gene K(+) channel in smooth muscle.

The human ether-a-go-go-related gene (HERG) product forms the pore-forming subunit of the delayed rectifier K ؉ channel in the heart. Unlike the cardiac isoform, the erg K ؉ channels in native smooth muscle demonstrate gating properties consistent with a role in maintaining resting potential. We have cloned the smooth muscle isoform of HERG, denoted as erg1-sm, from human and rabbit colon. erg1-sm is truncated by 101 amino acids in the C terminus due to a single nucleotide deletion in the 14th exon. Sequence alignment against HERG showed a substitution of alanine for valine in the S4 domain. When expressed in Xenopus oocytes, erg1-sm currents had much faster activation and deactivation kinetics compared with HERG.
Step depolarization positive to ؊20 mV consistently produced a transient outward component. The threshold for activation of erg1-sm was ؊60 mV and steady-state conductance was ϳ10-fold greater than HERG near the resting potential of smooth muscle. Site-directed mutagenesis of alanine to valine in the S4 region of erg1-sm converted many of the properties to that of the cardiac HERG, including shifts in the voltage dependence of activation and slowing of deactivation. These studies define the functional role of a novel isoform of the ethera-go-go-related gene K ؉ channel in smooth muscle.
The human ether-a-go-go related gene (HERG) 1 encodes for a K ϩ channel that is essential for normal repolarization of the cardiac action potential. HERG was originally cloned from the human hippocampal cDNA library by homology to the Drosophila K ϩ channel gene, eag (1), and according to the latest International Union of Pharmacology (IUPHAR) nomenclature has been termed as Kv 11.1. It is strongly expressed in the mammalian heart, and inherited mutations in this gene cause one form of long Q-T syndrome, LQT2 (2)(3)(4). HERG forms the pore-forming subunit of the rapidly activating delayed rectifier K ϩ current, IK r , in native cardiac myocytes (5,6), and heter-ologous expression of the cardiac HERG channel in Xenopus oocytes and mammalian cells (7,8) has demonstrated the inwardly rectifying properties of this current. HERG channels slowly activate on depolarization and demonstrate fast inactivation at positive potentials, resulting in small outward currents at potentials positive to 0 mV; hence, inward rectification (9,10). On repolarization close to resting potentials, large outward tail currents occur that slowly deactivate. These features of HERG channel dictate its role in repolarization of the cardiac action potential and frequency-dependent modulation of the action potential duration (11).
Recent studies also suggest that an inwardly rectifying K ϩ conductance that is active near the resting membrane potential of gastrointestinal smooth muscle cells (12,13), pituitary cells (14 -16), carotid body (15)(16)(17), and microglia (18) has properties similar to that of HERG channels. In these cells HERG conductance has been directly demonstrated in single cells, and in esophageal and stomach smooth muscle the presence of a "window" current within the range of the resting potential and depolarization and contraction by HERG channel blockers of whole tissue segments strongly suggest a role for ether-a-gogo-related gene K ϩ conductance in maintaining resting membrane potential (12). Moreover, transcripts for erg1 and immunohistochemical labeling have been shown in gastrointestinal smooth muscle cells (13). However, it is not clear how the kinetics of the cardiac isoform of the HERG channels can maintain the resting membrane potential or repolarization of spikelike action potentials of gastrointestinal smooth muscle cells. For instance, 1) the threshold for activation of the heterologously expressed cardiac isoform of the HERG current and its native form, IK r , is positive to Ϫ 50 mV (5,19), whereas the resting potential of gastrointestinal smooth muscle generally lies around Ϫ60 mV, 2) unlike the native cardiac myocytes, where large tail currents occur upon repolarization (20,21), smooth muscle cells do not demonstrate large tail currents upon repolarization, and 3) it is not clear how recovery from inactivation of the cardiac HERG can generate sufficient outward current during repolarization of shorter duration of action potential trains that occur from a plateau potential of Ϫ30 mV in gastrointestinal smooth muscle.
In this report, we describe the cloning and characterization of the smooth muscle isoform of erg channels, which we have denoted as erg1-sm, from rabbit and human colon. These studies demonstrate that erg1-sm is a truncated isoform that lacks 101 amino acids in the C-terminal region. We also show that there is a unique substitution of a conserved amino acid in the S4 voltage sensor region between the cardiac and smooth muscle isoforms that confers significant hyperpolarizing shift in the voltage dependence of activation for erg1-sm and results in steady-state conductance within the potential range for maintaining the resting membrane potential and demonstrates ki-netics that are consistent with this isoform being able to generate repolarizing current during smooth muscle action potentials.

RNA Extraction and RT-PCR-Adult male rabbits (New Zealand
White) were anesthetized with sodium pentobarbital, and the colons were excised. Rabbits were then sacrificed by pentobarbital overdose in accordance with the Institutional Animal Care and Use Committee of the University of Oklahoma Health Science Center. Circular muscle strips were carefully dissected from the underlying mucosa and quickfrozen in liquid nitrogen followed by homogenization. Total RNA was isolated using the S.N.A.P. total RNA isolation kit (Invitrogen) as per the manufacturer's instructions. In separate experiments, rabbit colon was embedded in OCT, and cross-sections were prepared in a cryostat. Cross-sections of the colon were placed on glass slides, fixed in alcohol, dehydrated, dipped in xylene, and placed on the microscope of the Arcturus PixCell II system (Arcturus Engineering Inc.). Single smooth muscle cells were selected by laser capture with a 7.5-m spot size. Approximately 40 -80 cells were collected for RNA isolation. The caps were placed in a 0.5-ml RNase-free Eppendorf tube and solubilized in lysis buffer (S.N.A.P. kit). RNA was isolated according to the manufacturer's instructions. Human colon, mouse small intestine, and mouse heart total RNA was purchased from Clontech (Palo Alto, CA). All RNA samples were quantified by spectrophotometer and run on denaturing RNA gels to check the quality and integrity of total RNA. First-strand cDNA was synthesized using 50 units of Superscript 11 RT (Invitrogen) at 42°C for 50 min in the presence of 3 g of total RNA, 1 g of oligo-dT primers, and 10 mM dNTPs in a total 20-l reaction. Three 5-l aliquots of reverse-transcribed reactions were used for performing PCR. Appropriate controls were carried out using water or omitting the enzyme.
Primer Design-The following PCR primers were used. Numbers in parenthesis are GenBank TM accession numbers. For erg K ϩ channels, the primers were designed to amplify the entire coding region from the published sequence for Merg1 (NM-013569). The sense primer was (nt 317-336) 5Ј-CGGATCCATGGGCTCAGGATGCCGGTG-3Ј). The antisense primer (nt 3862-3883) (5Ј-CGATATCCCAAGGAGAGCGGTCAG-GTAAT-3Ј) was designed 45 nucleotides downstream of the stop codon in the 3Ј-untranslated region. BamH1 and EcoRV restriction sites were added to the forward and reverse primers, respectively (shown in italics), to facilitate subcloning between the BamH1 and SmaI site of pSP64 vector. ␤-Actin (V01217) sense (2383-2402) and antisense (3071-3091) primers were designed to span an intron in addition to two exons. Thus contamination by genomic DNA would be detected as a band of 708 bp rather than the expected size of ϳ500 bp.
PCR and Cloning of erg Gene-1 g of cDNA made from total RNA of rabbit circular muscle strips and human colon, 2.5 units of Expand Long Template Taq DNA polymerase (Roche Molecular Biochemicals), 10 mM dNTPs, 10 M BamH1 forward, and EcoRV reverse primers in a total reaction volume of 50 l were used for PCR amplification of the erg gene. Controls were run in the absence of template DNA. PCR was performed in Robocycler (Stratagene, La Jolla, CA) under the following conditions: 1 cycle at 94°C for 3 min followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 65°C for 30 s, and extension at 68°C for 3.5 min for the first 10 cycles. Thereafter, the extension time was increased by 20s/cycle, followed by final extension at 68°C for 7 min. 10 l of RT-PCR products were separated on 1% agarose/1ϫ Tris-acetate-EDTA gel, and the DNA bands were visualized by ethidium bromide staining. The full length of the erg PCR product was gel-purified using Wizard PCR prep purification system (Promega, Madison, WI), digested with BamH1 and EcoRV, and ligated between the BamH1 and SmaI site of pSP64 poly(A) vector (Promega) using T4 DNA ligase (Promega). The ligated product was transformed in DH5␣ Escherichia coli cells. Colonies were screened for positive clones by plasmid DNA isolation using Qiagen plasmid miniprep kit (Qiagen, Valencia, CA). The orientation and fidelity of all the resulting plasmid DNA was confirmed by running it on 1% agarose gel after restriction digestion (EcoR1 and XhoI). Three independent clones were sequenced in their entirety using the dye-termination method on an ABI Prism (Foster City, CA) automated sequencer. Possible sequence changes due to expand long template PCR enzyme polymix-catalyzed replication errors were examined by comparison of sequences from a minimum of three independent PCR reactions. Nucleotide sequence common to at least two clones were considered correct due to the low probability of proofreader-introduced identical alterations at the same nucleotide position. For expression studies, the plasmid DNA in pSP64 was linearized with EcoR1 and transcribed into cRNA by mMessage mMachine kit (Ambion, Austin, TX) following the manufacturer's instructions. All cRNA samples were quantified by UV spectroscopy, and size and integrity was checked on denaturing RNA gels using appropriate RNA markers. HERG in pSP64 was kindly provided by Dr. Peter Spector (University of Oklahoma).
Mutagenesis-To introduce single amino acid substitutions in the S4 region of erg, the QuikChange site-directed mutagensis kit (Stratagene) was used according to the manufacturer's protocol. Primers were designed such that the nucleotide substituted was located in the middle of the forward and reverse primers. For creating S4 HERG mutant, the forward primer was 5Ј-GCGGCTGGTGCGCGCGGCGCGGAAGCT-3Ј, and the reverse primer was 5Ј-AGCTTCCGCGCCGCGCGCACCAGC-CGC-3Ј (NM_000238). For creating S4 erg1-sm mutant, the forward primer was 5Ј-CTGCGGCTGGTGCGCGTGGCTCGGAAGCTGG-3Ј, and reverse primer was 5Ј-CCAGCTTCCGAGCCACGCGCACCAGCC-GCAG-3Ј (AF439342). These primers were used to amplify the mutant from wt parental erg. Parental DNA was digested after PCR with DpnI and transformed into XL Blue supercompetent cells. Colonies were screened for the mutant plasmid and confirmed by DNA sequencing. HERG and erg1-sm S4 mutant plasmid DNA were linearized with EcoR1, and cRNA was synthesized in vitro as described above.
Protein Analysis-In vitro translation of smooth muscle erg and HERG plasmid DNA was performed using TNT coupled reticulocyte Lysate system (Promega) in combination with the Transcend ™ nonradioactive translation detection system for incorporation of biotinylated lysine. 2 g of smooth muscle erg and HERG plasmid DNA was translated following the manufacturer's instructions. The SP6 luciferase plasmid DNA was used as a control. The in vitro translated protein products were subjected to electrophoresis on 7.5% SDS-PAGE, transferred on to nitrocellulose membranes, and blocked by incubation in 15 ml of TBS-T (TBS with 0.5% Tween 20) for 1 h with gentle shaking. The membrane was incubated in streptavidin-HRP conjugate (1:5000) for 1 h, and washed in 15 ml of TBS-T 3 times followed by several washes with deionized water. Streptavidin-HRP that binds incorporated biotinylated lysine in translated proteins was detected by chemiluminescence according to the manufacturer's instructions.
Immunoblots were performed to detect the presence of the HERG C terminus using HERG polyclonal antibody directed against the C-terminal peptide sequence corresponding to amino acids 1106 -1159 of HERG (Alomone Laboratories, Jerusalem, Israel). Proteins from plasmid DNA were translated in a similar fashion as above, except for the omission of biotinylated lysine. These proteins were subjected to electrophoresis on 7.5% SDS-PAGE. Separated proteins were transferred to nitrocellulose membranes and blocked in 10 ml of 5% milk in TBS-T (TBS with 0.1% Tween 20) for 1 h and subsequently probed with the primary antibody directed against HERG C-terminal at a concentration of 1:200 in TBS for 1.5 h at room temperature. The membranes were washed twice with TBS-T and incubated for 1 h with HRP-linked secondary antibody (Santa Cruz Biotechnology). After the final washing steps, the blots were visualized by enhanced chemiluminescence. A similar procedure was used for detection with the anti-N-terminal HERG antibody (1:1000) (kindly provided by Dr. Jeanne Nerbonne).
Isolation of Membrane Protein from Mice Colonic Tissues-Colons were excised from four mice, and muscle strips were dissected under a microscope. Muscle strips were flash-frozen in liquid N 2 and homogenized in 5 ml of TE buffer at pH 7.4. All buffer solutions contained the following protease inhibitors: 1 mM iodoacetamide, 1 mM phenanthroline, 7.9 M aprotinin, 1 mM benzamidine, 1.4 M pepstatin, 0.5 mM Pefabloc, 0.1 M leupeptine, and a protease inhibitor mixture tablet (Roche Molecular Biochemicals). The homogenized samples were centrifuged at 1000 ϫ g for 10 min. This procedure was repeated twice to remove all nucleic acid, and debris and supernatant from both spins were collected and centrifuged at 40,000 ϫ g for 10 min. The pellet was resuspended in TE buffer containing 0.6 M KI, incubated on ice for 30 min, and centrifuged at 40,000 ϫ g for 10 min. The final pellet was solubilized in TE (2% Triton X-100) on ice for 1 h and centrifuged at 17,400 ϫ g for 30 min. Membrane protein concentrations were measured and subjected to electrophoresis on SDS-PAGE. The membrane proteins were probed with anti-N terminus HERG antibody (1:1000) and anti-C terminus HERG antibody (1:200) (Alomone).
Electrophysiological Recordings-Female Xenopus laevis (Nasco, Fort Atkinson, WI) were anesthetized by a 20 -30-min exposure to 3-aminobenzoic acid ethyl ester (tricaine 1.5 g/liter). All protocols were approved by the Institutional Animal Care and Use Committee of the University of Oklahoma Health Science Center. Ovarian lobes were removed through a small incision in the abdominal wall and washed in a Ca 2ϩ -free OR-2 solution containing 82.5 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 5 mM HEPES (pH adjusted with NaOH). Stage IV and V Xenopus oocytes were defolliculated by treatment with 1 mg/ml collagenase (Type 1A; Sigma) in the OR-2 solution for 1.5 h. Oocytes were incubated at 18°C in a modified Barth's solution with antibiotics containing 96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 2.5 mM HEPES (free acid), 2.5 mM HEPES (sodium salt), 0.1 g/liter streptomycin sulfate, 0.27 g/liter pyruvic acid, and 0.5 g/liter gentamycin sulfate. On the following day, oocytes were injected with the appropriate cRNA (1.1-1.6 g/l) using a Drummond Nanoject II automatic nanoliter injector (Drummond Scientific Co., Broomall, PA). Currents were recorded by two-electrode voltage clamp using the Geneclamp 500B amplifier (Axon Instrument, Foster City, CA) 2-5 days after cRNA injection. All recordings were carried out at room temperature (20 -23°C) in Barth's solution without antibiotics.
The pClamp software (Axon Instruments) was used for generation of voltage-clamp protocols and data acquisition. Currents were evoked by 2-s depolarizing voltage steps from Ϫ70 mV to ϩ50 mV in 10-mV increments followed by a repolarizing step to Ϫ70 mV. The amplitude of HERG and erg1-sm currents was measured at the end of depolarizing steps then normalized and plotted against the voltage to obtain the I-V relationship. Steady-state activation curve was determined by normalizing the peak of tail current (I) at Ϫ70 mV, plotted against the test potential (V t ), and fitted to a Boltzmann function , where I max is the maximum amplitude of tail current.
The voltage dependence of steady-state inactivation (rectification factor) was obtained using protocol as previously described (5,23). Channels were fully activated by voltage steps to ϩ30 mV (5 s) followed by test potentials from Ϫ140 mV to ϩ40 mV to obtain the currentvoltage relationship of activated channels. Slope conductance was determined from the negative arm of the I-V curve, and the rectification factor was calculated using the formula R ϭ I/G (V m Ϫ E K ), where R is the rectification factor, and E K was Ϫ95 mV. Data were fit by the Boltzmann relationship.
To obtain the time constants of activation, long duration depolarizing pulses (15 s) were applied. Selected raw traces from Ϫ40 mV to Ϫ10 mV for HERG current and from Ϫ50 mV to Ϫ20 mV for Smerg currents were fitted with double-exponential function to obtain slow and fast time constants. Time constants for inactivation were determined by using three-pulse protocol similar to that previously described (24).
To assess the rate of deactivation, a two-pulse protocol was used. A prepulse of ϩ30 mV (5s) was applied to fully activate and inactivate currents following by a range of pulses from Ϫ140 mV to ϩ40 mV with 20-mV increments. Deactivation time constants were obtained by a double-exponential fit.

RESULTS
To determine the physiological properties of the smooth muscle isoform of erg, we cloned and characterized the full-length cDNA from circular smooth muscle of the rabbit colon using RT-PCR. Fig. 1A shows the presence of the full-length transcript for erg1 (ϳ3.5 kilobases) from the rabbit colon, human colon, and mouse heart. A similar size band was also identified in mouse small intestine (data not shown). In addition to the presence of erg1, the alternatively spliced isoform, ergB, was also detected in the human colon (Fig. 1B). To further confirm that erg1 transcript is present in smooth muscle cells, about 40 -80 cells were selected by laser capture microdissection. Multicell PCR of captured cells showed the presence of erg1 without contamination from either neurons or interstitial cells of Cajal (Fig. 1C). Although no bands were seen for neuronal or interstitial cells of Cajal in selected smooth muscle or in circular muscle strips, the fidelity of these primers was confirmed using whole colon tissue (data not shown).
The full-length transcripts for erg1 from the rabbit and human colon were sequenced using overlapping primers. In an initial analysis, these sequences were aligned against the HERG (NM_000238) and found to be ϳ87% identical at the nucleotide level and ϳ95% at the protein level (deposited in GenBank TM , rabbit colon accession number AF439342; human colon accession number AY130462). Within the transmembrane spanning region (S1 to S6), the sequences were identical except for the presence of a conserved amino acid substitution, alanine for valine at position 537, in the S4 region of the smooth muscle isoforms from human and rabbit erg (Fig. 2). This position corresponds to 535 for HERG due to two additional amino acids in the N terminus of the smooth muscle isoforms. The single amino acid substitution within the S4 region was also identified in the rabbit stomach smooth muscle and, as shown below, confers important kinetic properties that define its role in maintaining smooth muscle-resting potential and repolarization of the action potential. We have denoted the smooth muscle isoforms as erg1-sm.
The open reading frame of the smooth muscle isoforms (erg1sm) consisted of 1058 amino acids as compared with 1159 amino acids for HERG, resulting in a truncated C terminus of 101 amino acids. This truncation was due to a frameshift resulting from a single-base deletion of G, corresponding to the position 3159 in HERG producing a premature stop codon in the 14 exon (Fig. 3) (25,26) in both human and rabbit colon. However, a similar deletion was not observed in cDNA from the rabbit atrial tissue, which was sequenced by RT-PCR using the same primers as for the smooth muscle.
To confirm that the smooth muscle isoform results in a truncated protein, the plasmid DNA for erg1-sm and HERG were translated in vitro and probed with streptavidin-HRP for detection of the proteins. As shown in Fig. 4A, the protein bands for HERG were obtained around the expected size of 127 kDa. The erg1-sm band was about 10 kDa smaller in size as would be expected for a 101-amino acid truncated protein.
However, a faint band around 127 kDa was also observed in both erg1-sm as well as in control (luciferase DNA) lanes, indicating an additional nonspecific protein band may be present around this molecular weight. The presence of HERG protein was, therefore, further examined by immunoblots carried out using anti-HERG antibody that was raised against the C terminus. Fig. 4B shows that this antibody recognized the presence of HERG but did not show bands from erg1-sm or control proteins. The anti-N-terminal HERG antibody was raised against residues 174 -188 of HERG, which is almost identical to that in erg1-sm. Fig. 4C shows that immunoblot with anti-N-terminal antibody revealed a band of 127 kDa for HERG and a slightly lower band for erg1-sm (ϳ120 kDa), consistent with the hypothesis that erg1-sm is a truncated isoform of HERG. To determine the expression of erg1-sm in native colonic tissues, immunoblots of isolated membrane proteins were carried out using the anti-C-terminal and anti-Nterminal HERG antibodies. Fig. 4, D and E, show that the major band revealed by anti-C-terminal antibody was around 95 kDa, which is expected for erg1b isoform. Larger molecular weight bands corresponding to glycosylated forms of erg1 (27) were not detected in colonic smooth muscle. The anti-N terminus antibody revealed a band ϳ120 kDa that most likely represents erg1-sm. Taken together these data suggest that erg1-sm is truncated and that previous findings of immunohistochemical localization of erg1 in single smooth muscle cells using the anti-C terminus antibody (12) represent erg1b expression.
Properties of erg1-sm Channels-To compare the functional properties of the smooth muscle isoform to the cardiac isoform of HERG, currents were measured from Xenopus oocytes expressing erg1-sm or HERG. Preliminary studies suggested that expression of erg1-sm currents were significantly lower than that for HERG and required at least 3-4 days for full development of currents when approximately equal amounts of RNA were injected. To obtain currents of comparable amplitude, the volume of RNA injection was doubled for erg1-sm.
Upon depolarization from a holding potential of Ϫ70 mV, erg1-sm K ϩ channels activated at potentials positive to Ϫ 60 mV. As shown in Fig. 5, the magnitude of this current increased up to Ϫ20 mV, and thereafter, larger depolarizations resulted in a progressive decrease in the amplitude, as expected for the inwardly rectifying properties of erg channels. At potentials positive to Ϫ 20 mV, transient currents were elicited whose rate of activation and the amplitude increased with increasing step depolarization (inset, Fig. 5A). Upon repolarization to the holding potential of Ϫ70 mV, tail currents with fast deactivation kinetics were obtained. The currents from erg1-sm were abolished by the class III anti-arrhythmic blocker, E-4031 (10 M) (Fig. 5B). In comparison, the currents from HERG-injected oocytes demonstrated much slower activation at negative potentials, with peak currents obtained at 0 mV. Unlike erg1-sm, currents from HERG showed no fast transient components at positive potentials (inset, Fig. 5A,  bottom panel). Upon repolarization to Ϫ70 mV, tail currents from HERG had much larger amplitudes and deactivated slowly compared with erg1-sm. The current-voltage relationship measured at the end of the test pulse showed a bell-shaped response with erg1-sm demonstrating a leftward shift of 20 mV compared with HERG (Fig. 5C).
The voltage dependence of activation was measured as the relative amplitude of tail currents and plotted as a function of test potential. This isochronal activation curve had a V 1/2 of Ϫ36 Ϯ 0.7 mV with a slope factor 10 for erg1-sm and a more positive V 1/2 potential of Ϫ16 Ϯ 0.4 mV (slope factor 9.5) for HERG (Fig. 6A). The voltage dependence of the steady state inactivation (the rectification factor) was measured using protocols as described under "Materials and Methods." The V 1/2 for erg1-sm was Ϫ26 Ϯ 2.5 mV with a slope factor of 17 mV, whereas that for HERG was Ϫ48 Ϯ 1 mV with a slope factor of 24 (Fig. 6B). The rightward shift meant that erg1-sm channels had less steady-state inactivation, particularly around the resting membrane potential of smooth muscle (between Ϫ70 mV and Ϫ50 mV). To determine the voltage dependence of the steady-state conductance, the fraction of channels activated at each potential was multiplied by the fraction of channels inactivated at that potential. The resulting window current showed a significant amount of steady-state current for erg1-sm channels between Ϫ60 and Ϫ50 mV (Fig. 6C). Within this potential range, steady-state current was almost 10-fold greater for erg1-sm than HERG, consistent with the findings in native smooth muscle cells of erg-like currents maintaining the resting membrane potential (12,13).
Both activation and inactivation rates were much faster for erg1-sm than HERG (Fig. 7). Activation was fit by a doubleexponential, and at almost all voltages (between Ϫ50 mV and Ϫ20 mV) time constants for activation for erg1-sm were almost 3 times smaller than for HERG currents (Fig. 7B). The time constant for inactivation was measured using a three-pulse protocol. The oocytes were depolarized to ϩ40 mV for 5 s followed by a short pulse for 15 ms to Ϫ140 mV to remove inactivation, and re-inactivation was induced at various test potentials from Ϫ60 mV to ϩ40 mV. The resulting currents were fit by a single exponential and demonstrated voltage dependence. As shown in Fig. 7C, the time constant for inactivation of erg1-sm was significantly smaller than that for HERG at all potentials. The rate of activation was measured by "envelope" of tail currents as shown in the voltage protocol of Fig.  7D. The rate of activation was also faster for erg1-sm than HERG.
Upon repolarization to Ϫ70 mV, tail currents of much smaller amplitude were recorded consistently in erg1-sm-injected oocytes compared with HERG. HERG channels transition into an open state on repolarization before slowly closing, leading to large amplitude of tail currents. Although similar transition was evident in erg1-sm at potentials positive to Ϫ 20 mV where inactivation occurs, the amplitude of the tail currents was similar to that of the amplitude of the transient outward current at that potential. The significantly smaller amplitudes of the tail currents are also consistent with findings from native cells, where tail currents due to K ϩ channel activation are quite small and do not show the typical hook observed for ventricular myocytes (19). The most likely explanation for the smaller amplitude of tail currents is the faster rate of deactivation in erg1-sm. Fig. 8A shows tail currents obtained upon repolarization to Ϫ60 mV from a test potential of ϩ30 mV. Deactivation was significantly faster for erg1-sm than HERG between Ϫ80 mV and Ϫ40 mV (Fig. 8B).
To assess whether the biophysical characteristics of erg1-sm currents are consistent with a physiological role in the repolarization of smooth muscle action potential, a voltage protocol was employed that mimicked the electrical activity of gastrointestinal smooth muscle cells. Contractions of gastrointestinal smooth muscle are elicited by multiple action potentials that are superimposed on an underlying rhythmic slow wave activity (28). The upstroke of the action potentials is mediated by activation of L-type Ca 2ϩ channels that occur during the plateau phase of the slow wave. The duration of the action potentials can also vary during the train. Activation of outward currents from erg1-sm and HERG-injected oocytes were examined using ramp depolarizations that varied in the inter-spike interval. Ramp depolarization was applied from Ϫ70 mV to Ϫ30 mV (100 ms) to mimic the upstroke of the slow wave that was then followed by the first spike by ramping from Ϫ30 mV to ϩ20 mV in 100 ms and repolarizing back to Ϫ40 mV (50 ms). The voltage was then slowly ramped back to Ϫ30 mV in 150 ms to mimic the inter-spike interval, which was followed by a second spike as shown in the upper traces of Fig. 9 (left-hand  panel). The pattern of currents obtained from erg1-sm and HERG-injected oocytes with this protocol is illustrated in Fig. 9 (left-hand panel). In both, depolarization from Ϫ70 mV to Ϫ30 mV produced little outward currents; however, in the erg1-sminjected oocytes (middle panel) a transient outward current was evoked during the first spike that partly inactivated at the peak of depolarization. Upon repolarization to Ϫ40 mV, currents slowly activated and maintained a plateau during the inter-spike interval; thereafter, further depolarization to ϩ20 mV (the second spike) resulted in relaxation of the outward currents as the channels reverted to an inactivated state. Repolarization of the second spike and return to the holding potential resulted in significant outward currents similar to the tail currents obtained during step repolarization. The same voltage protocol evoked a different current profile from HERG-injected oocytes (Fig. 9, bottom panel). During the first spike, there was no transient component, and the currents activated slowly until the end of the inter-spike interval. Thereafter, currents decreased during the upstroke of the second spike due to inactivation, and large tail currents were observed upon repolarization similar to the erg1-sm-injected oocytes.
The differences between erg1-sm and HERG became more apparent when the duration of the spikes was decreased ( Fig. 9; right-hand panel). With shorter durations, the outward currents from erg1-sm were evident during the first spike and regenerated with the second spike, whereas they were absent in HERG. These data illustrate the contribution of the erg1-sm channels to multispike potentials, particularly with respect to shorter action potential durations that can be frequently obtained in native gastrointestinal smooth muscle tissues.  6. Voltage dependence of activation, inactivation, and steady-state conductance. A, voltage dependence of activation was measured by normalizing the peak amplitude of tail currents relative to the largest current obtained upon repolarization to Ϫ70 mV from test potentials of Ϫ70 to ϩ50 mV. Data were fit by the Boltzmann relationship. The V 1/2 for erg1-sm was Ϫ36 Ϯ 0.7 mV (slope factor 10) (n ϭ 8) and Ϫ16 Ϯ 0.4 mV (slope factor 9.5) for HERG (n ϭ 9). B, the voltage dependence of rapid inactivation (rectification factor) was measured as described under "Materials and Methods." The data were fit by the Boltzmann equation. erg1-sm currents had V 1/2 values of Ϫ26 Ϯ 2.5 mV with a more steep slope factor (17 mV) compared with a V 1/2 of Ϫ48 Ϯ 1 mV and a slope factor of 24 for HERG (n ϭ 7). Note that the values obtained for HERG are similar to those described previously (5). C, calculated steady-state conductance of erg1-sm and HERG. The steady-state conductance was measured by multiplying the fraction of channels activated by inactivated channels at each potential. The steady-state conductance was significantly larger for erg1-sm, between Ϫ60 mV and Ϫ50 mV, near the resting membrane potential of smooth muscle.
Role of Amino Acid Substitution in the S4 Region-Sequence alignment of the rabbit colonic smooth muscle transcript showed a substitution in the S4 region of an alanine (Ala-537) in erg1-sm to valine (Val-537) in HERG (Fig. 2). Sequence by RT-PCR of transcripts from other smooth muscle tissues of the S4 region identified the same substitution of alanine for valine in rabbit stomach and human colon. In comparison across species, the mouse heart, canine heart, and rabbit heart all FIG. 8. Deactivation kinetics of erg1-sm. A, superimposed raw traces demonstrating deactivation of erg1-sm and HERG currents at Ϫ60 mV from a prepulse of ϩ30 mV (5 s). The currents from erg1-sm were scaled by a factor of 1.6 to obtain similar peak currents. Right panel, voltage protocol to assess deactivation kinetics. Deactivating currents were fit with double-exponential function. B, rates of deactivation were significantly faster in erg1-sm than HERG at potentials between Ϫ80 mV to Ϫ40 mV (n ϭ 8).  7). C, time constant of inactivation. Time constants were obtained by single exponential fit to currents obtained during the test potential from Ϫ60 mV to ϩ40 mV using the three-pulse protocol illustrated in the inset (n ϭ 8). Rate of inactivation in erg1-sm was significantly faster at each potential. D, the rate of activation for erg1-sm and HERG, measured using an envelope of tail currents. Oocytes were depolarized to ϩ40 mV for variable duration ranging from 25 to 250 ms to introduce complete activation and inactivation followed by repolarization step to Ϫ140 mV (as shown in the inset). Peak amplitudes of tail currents at Ϫ140 mV were measured, normalized, and plotted against pulse duration (n ϭ 10). contain valine at this position (Fig. 10A). Fig. 10B shows the sequence chromatogram, demonstrating that this substitution occurred as a result of a single nucleotide change of C to T at position 1610 in smooth muscle. To determine whether this uncharged conserved amino acid substitution in smooth muscle results in the changes in the biophysical properties of the channel, the alanine in erg1-sm was mutated to valine by site-directed mutagenesis. Similarly, valine (535) was also mutated to alanine in HERG. All mutations were confirmed by DNA sequencing, transcribed to cRNA, and expressed in oocytes.
Mutation of alanine to valine in the S4 mutant of erg1-sm (A537V) converted many of the biophysical properties similar to that for HERG (Fig. 10C). The A537V conversion resulted in loss of the transient component and produced large outward tails upon repolarization, similar to those obtained with HERG. The current-voltage relationship was also shifted to more positive potentials with a peak at Ϫ10 mV. In contrast, mutation of V535A in HERG resulted in transient currents at potentials positive to Ϫ 20 mV, with peak currents shifted to Ϫ20 mV compared with 0 mV in wt-HERG (Fig. 10D). Moreover, tail currents were also smaller to peak currents, comparable with that of the wt-erg1-sm with fast deactivation kinetics. Fig. 11A shows the changes in the voltage dependence of activation, which was shifted from a V 1/2 of Ϫ36 Ϯ 0.7 mV in wt-erg1-sm to Ϫ23 Ϯ 0.8 mV in the erg1-sm S4 mutant. In contrast, the S4 mutant of HERG had a mid-point of steady-state activation of Ϫ66 Ϯ 12 mV compared with Ϫ16 mV for the wild type. The rate of activation was also shifted to the left for the V535A mutant of HERG and rightward for A537V mutant for erg1-sm from its corresponding wild type (Fig. 11B). The activation time constants were slower in the S4 V537A mutant of erg1-sm and enhanced in the S4 mutant of HERG (not shown). Interestingly, the time constant for inactivation was not significantly changed by the S4 mutations. Thus, inactivation in V535A in HERG remained slower than that in A537V in erg1-sm (Fig.  11C) similar to those obtained with the corresponding wild types. However, the rate of deactivation was markedly slower in the erg1-sm S4 mutant compared with the HERG S4 mutant (Fig. 11D). DISCUSSION The findings of the present study demonstrate the functional properties of the smooth muscle isoform of the ether-a-go-gorelated gene K ϩ channel and show the differences compared with the cardiac isoform. At least three members of the ethera-go-go-related gene family, namely erg1, erg2, and erg3, have been previously identified. erg1 is present in several tissues including strong expression in the heart (2, 29) and brain (1). erg2 and erg3 are neuronal-specific isoforms (23). A weak signal for the presence of erg1 in smooth muscle was also previously reported in the small intestine (30) and stomach (22). Subsequently, patch-clamp studies in single smooth muscle cells from the opossum esophagus (12), rat stomach (13), and murine portal vein (31) have demonstrated the presence of HERG-like K ϩ channels in native cells. HERG channel blockers, including the gastrointestinal prokinetic agent, cisapride, were found to depolarize smooth muscle cells, and the presence of a window current around the typical resting potential of Ϫ50 mV strongly indicates that these channels were responsible for maintaining the resting membrane potential. In the present study, we show that the cloned erg K ϩ channel from gastrointestinal smooth muscle shares almost 95% homology at the amino acid level with the cardiac isoform up to the truncation at the C-terminal end. However, a critical substitution of an alanine for valine within the S4 region is present in several gastrointestinal smooth muscle cells, and this substitution confers a hyperpolarizing shift in the activation threshold and a significantly larger steady-state conductance around the rest- ing membrane potential. Moreover, differences in the rates of activation, recovery from inactivation, and deactivation compared with the cardiac HERG define the mechanisms for the smooth muscle isoform in the repolarization of multiple spikelike action potentials that are typically observed in native gastrointestinal smooth muscle cells.
The smooth muscle isoforms contained a premature stop codon that occurs in the 14th exon, resulting in a truncated C terminus. erg1 consists of 15 exons and at least 2 alternatively spliced variants of HERG1 have been identified in the heart, HERGB and HERG uso . HERGB and its mouse counterpart, MergB, appear to contain a unique N terminus (22,32) that on co-injection with HERG1 results in a current that approximates native cardiac IK r . However, recent studies by Pond et al. (29) suggest that HERGB protein may not be expressed in the heart. HERG uso is alternatively spliced at exon 9 and leads to a truncated isoform (33). However, expression of HERG uso in mammalian cells does not appear to generate a functional channel. erg1-sm on the other hand appears to encode for a functional protein that is truncated at the C terminus. The in vitro translation of erg1-sm resulted in a protein band that was ϳ10 kDa smaller compared with HERG, as would be expected for a 101-amino acid truncation. Furthermore, immunoblots using an anti-HERG antibody that is raised against the Cterminal amino acids 1106 -1159 of HERG while recognizing the HERG protein failed to detect erg1-sm. This is in contrast to the immunohistochemical localization of HERG in single smooth muscle cells using a similar antibody (12,13). A likely explanation for this is that smooth muscle cells also appear to express the ergB isoform that would, thus, result in immunolocalization of the C terminus ( Fig. 1) (22,32). Further confirmation that erg1-sm is truncated in native tissues was obtained by using the anti-N-terminal HERG antibody. This antibody recognized the in vitro translated protein as well as a 120-kDa protein in colonic membrane. Further studies are, however, necessary to define whether the ergB isoform from smooth muscle makes functional channels alone or in combination with erg1-sm.
Although erg1-sm currents showed similar inward rectification as the HERG channel, notable differences were the presence of a large transient component at potentials positive to Ϫ 20 mV, a lower threshold for activation, a larger steady-state conductance around the resting membrane potential, and faster activation, inactivation, and deactivation kinetics. These differences are consistent with the ability of the smooth muscle isoform to provide significant outward currents during smooth muscle action potential, particularly from a plateau phase of the gastrointestinal slow wave. The transient component is only occasionally seen in cardiac HERG (34,35) and mainly at potentials positive to ϩ20 mV. In single channel recordings, Kiehn et al. (34) show that HERG channels undergo direct closed to inactivated states, and the absence of a transient FIG. 10. Biophysical characteristics of S4 mutants. A, comparison of the amino acid sequences of the S4 segment from heart and smooth muscles. The GenBank TM accession number for the heart sequences is presented in parenthesis. Sequences between heart and smooth muscle are identical except for the substitution of alanine for valine. B, sequence chromatogram of part of the S4 region from human colon and HERG, demonstrating nucleotide change resulting in coding of alanine. C, raw traces of erg currents corresponding to S4 mutant of erg1-sm (A537V in erg1-sm (upper panel)) and S4 mutant of HERG (V535A in HERG (lower panel)). Insets above each panel show that early transient components at a faster time scale were present in the S4 mutant of HERG but not in the corresponding erg1-sm mutant. Currents were elicited by the same protocol as in Fig. 5A. D, current-voltage relationship for S4 mutants of erg1-sm (closed squares, n ϭ 11) and HERG (open circles, n ϭ 6). component allows for the delayed repolarization of the cardiac action potential. The upstroke of the action potential in smooth muscle cells is principally mediated by L-type Ca 2ϩ channels, whose threshold lies around Ϫ30 mV, at the plateau of the slow wave. The action potentials are also of shorter duration, and as demonstrated by using voltage protocols that mimic the gastrointestinal smooth muscle action potential, erg1-sm currents can activate during the first action potential spike. HERG on the other hand requires significant longer duration to fully activate and contribute to this type of action potential, as would be expected with slower activation kinetics.
There are several similarities between the current profile of erg1-sm and the nervous system specific isoform, erg3. Shi et al. (23) show that erg3 produces a transient component at potentials of ϩ20 mV and has a more negative threshold for activation and a larger steady-state conductance around Ϫ50 mV than erg1. These properties allow for substantial contribution of these channels in the shorter duration action potentials in neurons. The transient component of erg3 occurs as a result of much slower rate of inactivation compared with erg1. However, in erg1-sm we found that the rate of inactivation was considerably faster than HERG. Moreover, the activation of the transient outward component appears to depend on the presence of an alanine in the S4 voltage sensor region, whereas erg3, like HERG, contains a valine in this position. erg3 is also only 57% identical to erg1, whereas erg1-sm is ϳ95% identical at the amino acid level up to the truncated C-terminal portion.
It is noteworthy that in several gastrointestinal smooth muscle tissues there is a substitution of an alanine for valine in the S4 region. The S4 segment of voltage-gated ion channels is highly conserved, consisting of repeating basic residues. The S4 segment is proposed to function as a voltage sensor with positively charged residues critical in voltage-dependent transitions for activation (36). Recent studies by Smith-Maxwell et al. (37,38) suggest that single conserved uncharged amino acid substitution within the S4 region may be important in cooperativity and voltage dependence of Shaker K ϩ channels. These authors demonstrated that substitution of leucine for isoleucine (that differ in the attachment of a methyl group) resulted in changes in the gating charge movement. This could be attributed to steric interactions within the S4 region. Mutation of an alanine to valine in erg1-sm resulted in loss of the transient component and shifted the voltage dependence of activation to more positive potentials. Moreover, there was an obvious increase in the amplitude of the tail currents as well as a decrease in the rate of deactivation. In contrast, mutation of valine in HERG to alanine resulted in a transient component, leftward shift in the threshold and voltage dependence of activation and a faster rate of deactivation. These findings are consistent with interactions between the amino and carboxyl termini with the S4 region. Deactivation of the HERG channels is strongly dependent on the initial N-terminal residues that constitute the PAS domain (39,40). Previous studies from several laboratories have confirmed that the interaction of the PAS domain with the channel core, including the S4 region (41) and/or the S4-S5 linker (42), determines the slow deactivation kinetics of the HERG channel. The PAS domain is identical in erg1-sm and HERG and suggests that the faster deactivation process may be related to the interaction with the S4 domain. Thus, mutation of the S4 alanine to valine resulted in significant slowing of the deactivation process, similar to that for wt-HERG. At present, it is not clear how single nucleotide substitution occurs within the S4 region of smooth muscle isoform. However, it is interesting to note that the C to T substitution is reminiscent of RNA editing. Further experiments to determine whether this may be due to specific RNA editing in smooth muscle cells will be required.
In addition to the initial PAS domain, the stretch of amino acids from 138 to 373 (just before the S1 region) can also affect channel activation. The deletion of this segment appears to FIG. 11. Kinetics of activation, inactivation, and deactivation for S4 mutants. A, voltage dependence of activation for S4 mutant of erg1-sm (A537V) (closed squares) and S4 mutant of HERG (V535A) (open circles). B, rate of activation for S4 mutants obtained by using envelope of tail voltage protocol as described in Fig. 7. C, time constants of inactivation for in the S4 mutants obtained in similar fashion to that for wild-type given in Fig. 7. D, raw data show slower deactivation in the S4 mutant of erg1-sm compared with that for S4 mutant of HERG. Deactivation currents were obtained at Ϫ60 mV after a prepulse of ϩ30 mV (5 s). accelerate channel activation and shifts voltage dependence to hyperpolarized potentials (43 (44). There are several differences in the amino acid sequence between erg1-sm and HERG within this region including the addition of two amino acids in the smooth muscle isoform that could account for both faster rates of activation and hyperpolarized shift in the voltage dependence. However, further studies with specific mutations in this domain are needed to confirm the role of the N-terminal domain.
Interestingly, inactivation was significantly faster in erg1-sm than HERG and, unlike activation or deactivation, was not altered by S4 mutation. This could be related to the specific interaction of the C and N termini in maintaining the inactivation gate. Aydar and Palmer (45) demonstrate that truncation of the C terminus of HERG resulted in faster deactivation kinetics and that double deletion mutants of N and C termini increased channel inactivation time constants. It is possible that in erg1-sm a truncated C terminus specifically interacts with the N terminus, resulting in faster inactivation. The C terminus is also known to be responsible for expression. Kupershmidt et al. (46) recently demonstrated that removal of the C-terminal 147 amino acids resulted in endoplasmic reticulum retention of the HERG channel. However in mutants in which the RXR motif is absent, there is robust expression of functional channels. Our findings that expression levels were reduced in erg1-sm-injected oocytes could be related to the presence of the RGR sequence. The fact that erg1-sm is expressed, albeit to a lesser extent, suggest that part of the forward trafficking domain and/or protein folding is intact. These findings are also consistent with lower expression levels in smooth muscle, where high K ϩ concentrations are required to identify HERG-like currents (12).
In conclusion, our studies demonstrate several features of the smooth muscle isoform of the ether-a-go-go-related gene K channel that may explain its role in the native tissues. (a) The negative threshold for activation and steady conductance around the resting potential is consistent with these channels maintaining the resting membrane potential, (b) the faster rate of activation and recovery from inactivation define the role of erg1-sm in repolarization of the action potential and of gastrointestinal slow waves, (c) the lower levels of expression of erg message in smooth muscle may be due to a truncated C terminus, and (d) substitution of a conserved amino acid within the S4 voltage-sensing region defines some of the biophysical differences between smooth muscle and cardiac ether-a-go-gorelated gene K ϩ channels. Further studies to determine the association of erg1-sm with accessory K ϩ channel subunits, such as MiRP1, that co-assemble with HERG (47) will be necessary to demonstrate the functional basis of erg currents in native smooth muscle tissues.