Molecular identification and functional roles of a Ca(2+)-activated K+ channel in human and mouse hearts.

The repolarization phase of cardiac action potential is prone to aberrant excitation that is common in cardiac patients. Here, we demonstrate that this phase is markedly sensitive to Ca2+ because of the surprising existence of a Ca2+-activated K+ currents in cardiac cells. The current was revealed using recording conditions that preserved endogenous Ca2+ buffers. The Ca2+-activated K+ current is expressed differentially in atria compared with ventricles. Because of the significant contribution of the current toward membrane repolarization in cardiac myocytes, alterations of the current magnitude precipitate abnormal action potential profiles. We confirmed the presence of a small conductance Ca2+-activated K+ channel subtype (SK2) in human and mouse cardiac myocytes using Western blot analysis and reverse transcription-polymerase chain reaction and have cloned SK2 channels from human atria, mouse atria, and ventricles. Because of the marked differential expression of SK2 channels in the heart, specific ligands for Ca2+-activated K+ currents may offer a unique therapeutic opportunity to modify atrial cells without interfering with ventricular myocytes.

In humans, delineation of the outward currents that confer the late repolarization phase of the cardiac AP is crucial for our understanding of the etiology of arrhythmias. We provide a novel report that demonstrates that the repolarization phase of cardiac AP shows marked sensitivity toward apamin, an exclusive ligand for a small conductance Ca 2ϩ -activated K ϩ channel (2). Ca 2ϩ -activated K ϩ channels (K Ca ) are present in most neurons and mediate the afterhyperpolarizations following AP (3,4). However, functional significance of K Ca in the heart has not previously been documented. K Ca channels can be divided into three main subfamilies (3,(5)(6)(7). These include the large-conductance Ca 2ϩ -and voltage-activated K ϩ channels (BK), the intermediate-conductance K Ca channels (IK), and the smallconductance K Ca channels (SK), which are sensitive to apamin and scyllatoxin. Among the SK channels, they are encoded by at least three genes, SK1, SK2, SK3 (4, 6), with differential sensitivity toward apamin. SK2 is highly sensitive to apamin, with a half-blocking concentration (IC 50 ) of 60 pmol/liter, whereas SK1 channels are not affected by 100 nmol/liter apamin (2). SK3 channels are intermediate.
Here, we report for the first time, the presence of I K,Ca (Ca 2ϩ -activated K ϩ current) in cardiac myocytes that plays a crucial role in cardiac AP profile. Using a combination of electrophysiological recordings and biochemical and molecular biological techniques, we have identified the presence of SK2 isoform in the heart. Importantly, there is differential expression of SK2 channel with more abundant SK2 channel in the atria compared with the ventricles. Thus, specific ligands for I K,Ca may offer a unique therapeutic opportunity to directly modify the atrial cells without interfering with ventricular myocytes.

MATERIALS AND METHODS
The human study protocol was approved by the University of California, Davis, Institutional Review Board. All animal care and procedures were approved by the University of California, Davis, Institutional Animal Care and Use Committee.
Electrophysiologic Recordings-Twelve-week-old male CD-1 mice were purchased from Charles River Laboratories, Inc. (Wilmington, MA). All chemicals were purchased from Sigma unless stated otherwise. Single mouse atrial and ventricular and human atrial myocytes were isolated as described previously (8,9). Briefly, for human atrial myocytes, atrial appendages were collected in ice cold cardioplegic solution and transferred to the laboratory. Tissue was rinsed in a dissection buffer containing (mM/liter): NaCl, 140; KCl, 5.4; MgCl 2 , 1.2; HEPES, 5; glucose, 5; and 2,3-butanedione monoxime, 30; pH 7.0. The tissue was cut into small chunks (Ͻ1 mm 3 ) and washed with oxygenated buffered solution three times. Chunks were transferred to a 50-ml Falcon tube containing 10 ml of Ca 2ϩ -free buffer with a composition similar to that of dissection buffer, supplemented with 0.2% bovine serum albumin, collagenase (Worthington, type II (225 units/mg), 0.64 mg/ml), and protease (Sigma, type XXIV, 0.4 mg/ml) but without 2,3-butanedione monoxime. The tube was placed in a water bath (32-33°C) mounted over a magnetic stirrer (about 60 rpm) for 45 min. During a period of 45 min, fresh enzyme buffer was replaced at an interval of 10 FIG. 1. Evidence for I K,Ca in human atrial and mouse atrial and ventricular myocytes. A, Examples of APs recorded using the perforated patch clamp technique from an isolated human atrial myocyte in control (dotted line) and after 50 pmol/liter apamin (solid line). The right panel shows bar graphs for the summary data of APD 50 , APD 90 (*, p Ͻ 0.05, n ϭ 5). Control data are shown in open bars, and data after drug application are shown in closed bar. B and C, mouse atrial (left) and ventricular (right) myocytes. AP traces shown were obtained in control (dotted lines) and after application (solid lines) of apamin (50 pmol/liter, B) and DQ (1 mol/liter, C). Bar graphs in the inset show summary data of APD 50 , APD 90 (*, p Ͻ 0.05, n ϭ 8 for each group). Control data are shown as open bars, and data after drug application are shown as closed bar. or 15 min. After 45 min of enzyme treatment, the supernatant was aspirated and discarded. Fresh enzyme solution (collagenase only, 0.48 mg/ml) was added for an additional 10 min. From this step, the supernatant was collected and centrifuged for 2 min at 300 rpm (about 18 g) after every additional 10 min. The resulting supernatant was then discarded and the myocytes were stored using a high-K ϩ solution containing (mmol/liter) potassium glutamate, 120; KCl, 25; MgCl 2 , 1, EGTA, 0.1; glucose, 10; and HEPES, 10, pH 7.4, with KOH at room temperature. Cells were used for electrophysiologic recording within 7-8 h after isolation. This isolation procedure yields ϳ80% of Ca 2ϩtolerant ventricular myocytes with clear striations.
A voltage-ramp protocol was used to elicit whole-cell current to simulate AP to further illustrate the dependence of the SK current on intracellular Ca 2ϩ . The experiments were carried out using the perforated-patch technique to preserve intracellular milieu using external solution as in Fig. 1 during AP recordings in the absence of external K ϩ , Ca 2ϩ , or Na ϩ channel blockers (Fig. 2E). The pulse protocol used is shown in the inset.
For assessment of dose-response curve to apamin, the solid lines in Fig. 2F represent the least squares fit to the data points using the function: I/I 0 ϭ 1/{1ϩ([blocker]/IC 50 ) n }, where I and I 0 is the current in the presence and absence of apamin, respectively, and [blocker] represents the concentration of apamin in pmol/liter. Numbers in parentheses represent the number of cells for each group.
For single-channel current recordings, excised inside-out patches were used to allow application of Ca 2ϩ or Ca 2ϩ chelator to the internal side of the channels. The bath solution contained (in mmol/liter): potassium glutamate, 120; KCl, 20; MgCl 2 , 1; CaCl 2 , 1; glucose, 10; and HEPES, 10; pH 7.4 with KOH. The pipette solution contained (in mmol/liter): KCl, 5; NMG, 125; tetraethyl ammonium chloride, 20; 4-AP, 5; MgCl 2 , 1; CaCl 2 , 2; HEPES, 10; glucose, 10; pH 7.4 with HCl with a calculated Nernst potential for K ϩ ion (E K ) of Ϫ83 mV. Amplitude histograms at a given test potential were generated and fitted to a single Gaussian distribution to obtain the mean unitary currents. Leaksubtracted current records were idealized using a half-height criterion (15). Idealized records were used to construct the open probability (P o ). The single channel activity could not be recorded from all the inside-out patches. The probability of successful recording of the SK channel activity was estimated to be ϳ15% (7/45 patches).
Where appropriate, pooled data are presented as means Ϯ S.E. Statistical comparisons were performed using the Student's t test with p Ͻ 0.05 considered significant.
Western Blot Analysis-Immunoblots were performed as described previously (12). Protein samples were prepared from atria and ventricles of mice, cats and rats. Human heart lysate was obtained from Abcam (Cambridge, UK). Human atrial tissue was obtained during cardiac aortocoronary bypass graft surgery. The following primary antibodies were used: 1) anti-SK2 (Alomone Laboratories), a polyclonal antibody raised in rabbit against a purified peptide corresponding to amino acid residues 542-559 of rat SK2 located in COOH terminus (6); 2) monoclonal mouse anti-rabbit glyceraldehyde-3-phosphate dehydrogenase (GAPDH, Research Diagnostics, Inc., Flanders, NJ); and 3) anti-cyclophilin antibody (Upstate Biotechnology) were used as internal control for loading. Following incubation in the primary antibody, membranes were probed with peroxidase-conjugated secondary antibodies (Calbiochem). The protein bands were detected by chemiluminescence (Amersham Biosciences). The signal obtained from atrial and ventricular tissues were compared. Quantification of the signals was performed by densitometry (ImageQuant).
Reverse Transcription-Polymerase Chain Reaction (RT-PCR)-Total RNA was prepared from human atria obtained during cardiac aortocoronary bypass graft surgery as well as from mouse brain and heart using TRIzol reagent (Invitrogen). cDNA was synthesized from total RNA samples by oligo(dT)-primed reverse transcription (Superscript II RNase H-Reverse Transcriptase, Invitrogen). cDNA was then subjected to PCR amplification using HotStarTaq DNA Polymerase (Qiagen). Full-length SK2 coding sequences were obtained from human atria and mouse atria and ventricles using the following primers: 1) for human atria, 5Ј-ATGAGCAGCTGCAGGTACAA-3Ј (forward) and 5Ј-CTAGC-TACTCTCTGATGAGG-3Ј (reverse) according to the published sequence of human myometrial SK2 (accession number AF397175) and 2) for mouse atria and ventricles, 5Ј-ATGAGCAGCTGCAGGTACAA-3Ј (forward) and 5Ј-CTAGCTACTCTCTGATGAAG-3Ј (reverse) according to the published sequence of mouse colonic myocyte SK2 (accession number AF357240). The primers used to amplify GAPDH from human tissue were 5Ј-CGTGTCAGTGGTGGACCTGACCTG-3Ј (forward) and 5Ј-CAAAGGTGGAGGAGTGGGT-3Ј (reverse). The absence of genomic contamination in the RNA samples was confirmed with reverse transcription-negative controls (no RT) for each experiment. Amplified products were cloned into pCR2.1-TOPO and pCRII-TOPO plasmid vector (Invitrogen) and analyzed by DNA sequencing of both forward and reverse strands.
Immunofluorescence Confocal Microscopy-Immunofluorescence labeling was performed as described previously (16). The cells were treated with anti-SK2 antibody (1:20 dilution in 10% goat serum, 0.1% Triton X-100 in phosphate-buffered saline overnight at 4°C). Immunofluorescence labeling for confocal microscopy were done by treatment with fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (Calbiochem, 1:30 dilution). Immunofluorescence-labeled samples were examined using a Pascal Zeiss confocal laser scanning microscopy. Control experiments performed by preincubation of the primary antibody with the respective antigenic peptide (1:1) did not show positive staining under the same experimental conditions used. Identical settings were used for all the specimens.
The nucleotide sequences of the mouse and human cardiac SK2 channels have been submitted to GenBank TM , and the accession numbers are AF533008 and AY258141, respectively. The following are the accession numbers of the published full-length SK2 coding sequences: mouse colon (accession number AF357240), rat brain (6) (accession number U69882), mouse cochlea (accession number AY123778), human myometrium (accession number AF397175), and human leukemic Jurkat T cells (17) (accession number AF239613).

Existence of I K,Ca in Human Atrial and Mouse Atrial and
Ventricular Myocytes-To directly examine the functional role of I K,Ca , perforated-patch techniques were used to record AP from human atrial and mouse atrial and ventricular myocytes. Fig. 1 A shows examples of AP recorded from an isolated human atrial myocyte in control and after application of a low concentration of a SK channel-specific blocker, apamin (50 pmol/liter). A significant prolongation of the terminal portion of the AP was observed after the toxin. The effect could be reversed upon wash out. Summary data for action potential duration at 50 and 90% repolarization (APD 50 , APD 90 ) is shown in the right panel in control and after application of 50 pmol/liter apamin. Fig. 1, B and C, show examples of mouse atrial AP (left panels) and ventricular myocytes (right panels) in control and after application of apamin (50 pmol/liter, Fig. 1B) and dequalinium chloride (DQ, 1 mol/liter, Fig. 1C), another known blocker of SK channels. Mouse atrial APs were exquisitely sensitive to apamin at picomolar concentration with significant prolongation of the APs. The effects were much more moderate FIG. 2. Whole-cell current recordings. A, examples of whole-cell currents recorded from mouse atrial myocytes using voltage steps from a holding potential of Ϫ50 mV. The current was recorded in control and after apamin (50 pmol/liter) and [Ca 2ϩ ] i was 500 nmol/liter. The voltage protocol used is shown above the current traces. The apamin-sensitive current traces were obtained using digital subtraction and are shown in the lower panel. B, current density-voltage relation of the apamin-sensitive current using a concentration of 50 pmol/liter and [Ca 2ϩ ] i of 500 nmol/liter in ventricular cells (right panels). Bar graphs in the inset show summary data for APD 50 , APD 90 in control, and after application of apamin or DQ (*, p Ͻ 0.05). Close examination of the data shows significant differences in the AP duration in the mouse atrial and ventricular myocytes. This is most likely due to a smaller K ϩ current density in atrial cells compared with ventricular cells, which has been described previously (18,19).
Ca 2ϩ Dependence of the Apamin-sensitive Currents-To further document the presence of I K,Ca in cardiac myocytes, wholecell current was recorded using voltage steps from a holding potential of Ϫ55 mV to inactivate the transient outward K ϩ currents, which were known to be present in mouse atrial and ventricular myocytes (18,19) (Fig. 2). Na ϩ and Ca 2ϩ ions were eliminated from the external solution. The interior of the cells was dialyzed using pipette solutions containing known concentrations of Ca 2ϩ as described under "Materials and Methods." Using the above recording condition, a time-independent inward rectifier K ϩ current (I K1 ) could be recorded ( Fig. 2A). In addition, a component of both inward and outward current could be blocked by application of apamin (50 pmol/liter). This inhibition could be reversed upon washout. The apamin-sensitive traces were obtained using digital subtraction. The current density-voltage relation of the apamin-sensitive component is shown in B showing inward rectification similar to SK current previously reported for rat brain SK channels (20). In addition, we demonstrated the Ca 2ϩ dependence of the apamin-sensitive current (50 pmol/liter) using [Ca 2ϩ ] i of 10 and 1000 nmol/liter showing a significant difference in the apamin-sensitive current density (presented as percentage of the total sustained current) at the two different [Ca 2ϩ ] i . Importantly, no appreciable apamin-sensitive component could be demonstrated when [Ca 2ϩ ] i was lowered to or beyond 0.1 nmol/liter (Fig. 2C).
To further document the presence of I K,Ca in single isolated human atrial myocytes, we used a two-pulse voltage clamp protocol to record I K,Ca while minimizing contamination from Ca 2ϩ current (Fig. 2D). We first depolarized the cells from a holding potential of Ϫ55 to 0 mV for 10 ms to activate Ca 2ϩ current and to initiate the release of Ca 2ϩ from the sarcoplasmic reticulum. The cells were then stepped to ϩ40 mV to increase the driving force for K ϩ while decreasing the driving force for Ca 2ϩ current. The outward current identified was found to be exquisitely sensitive to apamin. Traces shown are in control, after application of apamin (50 pmol/liter) and after wash out. Composition of the external and internal solutions used are different from Fig. 2, A-C, and are indicated under "Materials and Methods." Calculated E K in this recording condition was Ϫ76 mV.
The contribution of SK current should parallel the variations in internal Ca 2ϩ concentration during AP. To further illustrate the dependence of the SK current on intracellular Ca 2ϩ , we performed experiments using voltage-ramp protocol to simulate AP. The experiments were carried out using perforatedpatch technique to preserve intracellular milieu using external solution as in Fig. 1 during AP recordings in the absence of external K ϩ , Ca 2ϩ , or Na ϩ channel blockers (Fig. 2E). A holding potential of Ϫ50 mV was used to inactivate the inward Na ϩ current and a large component of the 4-aminopyridine-sensitive transient outward K ϩ current (I to ), which is known to be present in mouse cardiac myocytes, without significant inactivation of the L-type Ca 2ϩ channels. We observed a component of the outward current, which was Ca 2ϩ -dependent and could be blocked by application of low concentration of apamin (100 pmol/liter) (Fig. 2E). This inhibition could be reversed upon washout. Moreover, the same component was found to be sensitive to DQ. In contrast to the current density-voltage relation in Fig. 2B showing inward rectification, the current elicited by the voltage-ramp protocol was mainly outward likely mirroring changes in the intracellular Ca 2ϩ concentration, which is voltagedependent (21).
We compared the dose-response curves of the outward current to inhibition by apamin from mouse ventricular and atrial myocytes (Fig. 2F). I K,Ca from atrial myocytes showed a marked sensitivity toward SK channel blocker, apamin, with an IC 50 of 42.3 pmol/liter. In contrast, the IC 50 obtained from the ventricular myocytes was 95.9 pmol/liter (Fig. 2F).
Direct Demonstration of the Existence of I K,Ca Using Singlechannel Recordings-Single-channel currents were recorded using excised inside-out patches (Fig. 3, A-D). A shows singlechannel openings in the presence of 1 mmol/liter Ca 2ϩ in the bath solution. Single-channel activities were rapidly abolished after addition of 10 mmol/liter EGTA (a Ca 2ϩ chelator) into the bath solution. We have previously documented the presence of a Ca 2ϩ -activated Cl Ϫ current (I Cl,Ca ) in mouse ventricular myocytes (22). However, the single channel activities identified here were insensitive to niflumic acid, in contrast to the previously identified I Cl,Ca (22) (Fig. 3B). The single-channel current-voltage relation showed a single-channel conductance of ϳ3 picosiemens with a reversal potential of Ϫ84 mV, close to the calculated Nernst potential for K ϩ (E K ) of Ϫ83 mV, suggesting that the channel is a K ϩ -selective channel (Fig. 3C). Furthermore, the channel was completely voltage-independent (D). It has previously been documented that several features distinguish members of the SK family from their closest phenotypic neighbors, the BK channels, and these include the low single-channel conductance of the SK channels and the absence of voltage dependence and their sensitivities toward apamin (3). Our biophysical data agree well with the pharmacological data obtained using whole-cell recordings and suggest that the channel identified in the heart represents a small-conductance I K,Ca (SK family) and most likely the SK2 isoform as a result of its high sensitivity toward apamin.
Molecular Correlates of I K,Ca and Quantitative Immunoblots-To identify the channel proteins, we performed immunoblots using antibodies generated against different types of K Ca channels. We demonstrated the presence of a protein, which reacts to a specific antibody against SK2 isoform with an expected size of 60 kDa in mouse, cat, and human myocytes, respectively (Fig. 4, A-C). Whole human heart lysate (mainly ventricular tissue) as well as human atria obtained during cardiac aortocoronary bypass graft surgery were used for the experiments. Consistent with the functional data, the SK2 isoform is differentially expressed at the protein level in human (n ϭ 5 cells). C, Ca 2ϩ dependence of the apamin-sensitive currents elicited at a test potential of ϩ60 mV (n ϭ 5 cells at each concentration) using [Ca 2ϩ ] i of 10 and 1000 nmol/liter. *, p Ͻ 0.05. D, examples of current traces obtained using a two-pulse voltage clamp protocol to record I K,Ca from single isolated human atrial myocytes while minimizing contamination from Ca 2ϩ current. Cells were first depolarized from a holding potential of Ϫ55 to 0 mV for 10 ms and then stepped to ϩ40 mV. Traces shown are in control, after application of apamin (50 pmol/liter) and after wash out. E, examples of whole-cell currents recorded from mouse atrial myocytes using a voltage-ramp protocol from ϩ40 to Ϫ100 mV with a slope of Ϫ180 V/s. The current was recorded using the perforated patch technique and the same external solution as AP recordings without Na ϩ , Ca 2ϩ , or K ϩ channel blockers. An outward current can be appreciated at base line. A portion of the outward current can be blocked using apamin (100 pmol/liter) and can be readily reversed upon washout. The same component was found to be sensitive to DQ (1 mol/liter). The sequence of the experiment as shown in the figure was control, apamin, washout, then DQ. F, the percentage of inhibition of the total outward current measured at 0 mV was assessed in control and after application of apamin at different apamin concentrations using the voltage-ramp protocol as described in the legend to E (see also "Materials and Methods"). as well as mouse atria versus ventricles (Fig. 4). Cyclophylin or GAPDH antibodies were used in these experiments as internal control for amount of total protein loaded.
Immunodetection of SK2 Channels in Dissociated Human and Mouse Atrial and Ventricular Myocytes-To directly rule out the possibility that the immunoreactivity observed in the Western blots may reflect the presence of SK channel in other cell types in the cardiac homogenate, e.g. vascular smooth muscle cells, we performed a immunofluorescence study using single isolated mouse atiral and ventricular myocytes and human atrial myocytes. Immunofluorescence confocal microscopy data showed specific reactivity to SK2 antibodies in isolated atrial and ventricular myocytes. Fig. 4, E-G, show specific labeling with SK2 antibody in isolated mouse atrial, mouse ventricular, and human atrial myocytes, respectively. Preincubation of the primary antibodies with the antigenic peptide in control experiments eliminated the positive labeling further confirming that the labeling seen in E-G was epitope-specific (Fig. 4H). In addition, the mouse atrial myocytes were more intensely labeled compared with ventricular myocytes.
Cloning of the Full-length Human and Mouse Heart SK2 cDNA Using RT-PCR Amplification-We further probed for the existence of SK2 channel in the human and mouse cardiac myocytes using RT-PCR and previously published primers. Fig.  5 shows representative RT-PCR amplified products of fulllength coding regions from mouse ventricles and atria (Fig. 5A) and human atria (Fig. 5B). Mouse brain was used as a positive control (lane 4 in panel A). Fig. 5C shows RT-PCR amplified product from the corresponding total RNA sample from human atria using primers specific for human GAPDH as positive control. The SK2 cDNA from human atria and mouse atria and ventricles were 1740 and 1722 bp long, respectively. The sequences obtained have been deposited in GeneBank TM (accession numbers are AY258141 and AF533008, respectively). The deduced SK2 protein sequences contain the functional domains that have been described previously in the SK channels. These domains include transmembrane segments S1-S6 and the pore region, characteristic of the K ϩ channel family of proteins, as well as the calmodulin binding domain (CaMBD). Fig. 5D shows the best alignment (ClustalW 1.8) of SK2 channels human and mouse cardiac myocytes in comparison with SK2 from human myometrium (accession number AF397175), human Examples of immunoblot with positive labeling of protein bands using an antibody specific to SK2 clone from rat brain. Each lane was loaded with 25 g of protein from mouse (A), cat (B), and whole human heart lysate (mostly ventricular tissue) and human atrial tissue (C). A and V refer to atria and ventricles, respectively. The expected size of the SK2 protein is 60 kDa. Negative controls were performed with antibody preincubated with the blocking peptide and labeled as Ϫve control. Lower bands are cyclophilin (expected size of 18 kDa). Similar data were obtained from rat heart (data not shown). D, summary data for the relative SK2 protein level normalized to the cyclophilin level obtained from mouse atria compared with the ventricles (n ϭ 7 for each group, AU ϭ arbitary unit, p Ͻ 0.05). E and F, confocal photomicrographs showing subcellular distribution of SK2 channels in mouse atrial (E), mouse ventricular (F), and human atrial (G) myocytes. Immunofluorescence labeling was done by treatment with secondary antibodies (fluorescein isothiocyanate-conjugated goat anti-rabbit antibody). In all cases, the specificity of labeling was confirmed by elimination of immunoreactivity after preincubation of the primary antibody with the respective antigenic peptide (1:1) (G). The scale bar is 20 m.

FIG. 5. Identification of the SK channel isoform.
A, representative agarose gels of RT-PCR amplified products for the full-length SK2 coding sequence from total RNA from mouse ventricles (lane 2), atria (lane 3), and mouse brain (lane 4). Lane 1 is HindIII DNA marker. Lane 5 is a negative control (PCR amplified without RT to make certain that there are no genomic contamination of the RNA samples). Similar data were obtained using rat atria and ventricles. B, representative agarose gels of RT-PCR amplified products from total RNA from two different samples of human atria (lane 2) and mouse brain (lane 3) using primers specific for human SK2 for the full-length coding sequence. Lanes 1 and 4 are HindIII DNA marker and a negative control, respectively. C, representative agarose gels of RT-PCR amplified products from total RNA from two different samples of human atria (lanes 1 and 2) using primers specific for GAPDH as positive control. Lanes 3 and 4 are a negative control and 50 -10,000-bp ladder, respectively. D, amino acid sequences alignment (ClustalW) of full-length SK2 channels from human and mouse cardiac myocytes in comparison with SK2 from human myometrium, leukemic Jurkat T cell (17), and rat brain (6) (H.s. ϭ human; M.m. ϭ mouse; R.n. ϭ rat). Short dashed lines represent gaps introduced to optimize the alignment. The six predicted transmembrane domains and the pore region leukemic Jurkat T cells (accession number AF239613) (17), and rat brain (accession number U69882) (6). The amino acid sequence of SK2 channel from mouse cardiac myocyte was identical to that of mouse cochlea (accession number AY123778) and mouse colon SK2 channel (accession number AF357240, not shown). In addition, the SK2 channels obtained from mouse atria were identical to those from ventricles (a total of eight clones analyzed from a total of four separate PCR reactions). The amino acid sequence of human cardiac SK2 shows 99% homology to SK2 channels from human myometrium and human leukemic Jurkat T cells. Sequence divergence is clustered almost exclusively to the cytoplasmic amino end and the carboxyl end after the calmodulin binding domain moiety of the channel. However, within these regions, there are also subdomains that are completely conserved among all the sequences. DISCUSSION I K,Ca confers the membrane repolarization and firing in a variety of cells. However, their functional existence in the heart cells, until now, remained unknown. We reported that I K,Ca is present in cardiac myocytes and is expressed differentially in atrial compared with ventricular myocytes. The current produces a marked effect on membrane repolarization in atrial myocytes. We have identified the channel protein as a small conductance channel subtype (SK2) and have cloned heart SK2 channel from mouse and human. This report represents the first study of the function of SK channels in cardiomyocytes.
Pharmacology of SK Channels-The bee venom peptide toxin, apamin, is an 18-amino acid peptide with two internal disulfide bridges that hold the peptide in a tight tertiary conformation (23). The amino acids that mediate apamin sensitivity have been determined. Two residues, an aspartic acid and an asparagine, that reside on opposite sides of the deep pore are essential for apamin sensitivity (2). No other class of K ϩ channels is blocked by this drug, and among the cloned K ϩ channels, the residues that endow sensitivity are present at those positions only in SK2 and SK3 channels. These data suggest that apamin is a very specific blocker for SK channels and that the SK channels may be the sole class of apamin receptors in the body (2). These same residues also mediate differential SK channel sensitivity to the nicotinic acetylcholine receptor antagonist D-tubocurarine (2) and are likely to be the determinants for block by other selective SK blockers. Therefore, SK channels may be classified pharmacologically on the basis of their sensitivity to apamin. SK2 is highly sensitive to apamin, whereas SK1 channels are the least sensitive (2). SK3 channels have intermediate sensitivity. In contrast, the potent small molecule blockers show little differentiation between the channel subtypes. For example, the antiseptic compound DQ has been found to specifically block SK channels and block SK1 and SK2 with similar half-blocking concentrations (24). Our electrophysiologic data suggest that channels identified in the heart most likely represent the SK2 isoform and can be reversibly blocked by picomolar concentration of apamin. This was further supported by our biochemical and molecular biological data. We have obtained full-length coding sequence from mouse atria and ventricles and human atria. Indeed, the mouse SK2 channel showed 97 and 99% homology in the nucleotide sequence to previously published rat brain (6) and mouse colon SK2 (accession number AF357240) and was identical to that of mouse cochlear SK2 (accession number AY123778).
There have been some uncertainty with regards to the existence of I K,Ca channel in cardiomyocytes since some years ago. The presence of I K,Ca has been demonstrated in rabbit myocytes and the current was found to be more prominent in atria than in ventricle (25). On the other hand, one previous study has suggested that a higher dosage of apamin (50 nM) may have an irreversible effect on the delayed rectifier K ϩ current (I K,s ) in guinea pig ventricular myocytes (26). By contrast, the SK current identified in our present study could be reversibly blocked by as low as 50 pM apamin. Finally, one study using apamin binding assay has suggested the presence of high affinity apamin receptors in rabbit cardiac myocytes with biochemical properties in common with both an L-type Ca 2ϩ channel and an SK channel. Therefore, in our present study, we directly tested the effects of apamin on L-type Ca 2ϩ channel in mouse atrial myocytes. Apamin up to 1 mol/liter concentration had no appreciable effects on Ca 2ϩ current in mouse atrial myocytes (peak Ca 2ϩ current density of Ϫ1.47 Ϯ 0.34 versus Ϫ1.46 Ϯ 0.45 pA/pF at a test potential of 5 mV before and after application of 1 mol/liter apamin, n ϭ 5). Taken together, the published report (26) and our experimental data, it is suggestive that the high affinity apamin receptors previously identified may represent SK channel subunits associated with the Ca 2ϩ -channel complexes and confer apamin sensitivity in their study.
Contribution of SK Current in Comparison with Other Repolarizing Currents-Electrophysiological studies and molecular cloning techniques have documented the expression of multiple types of voltage-gated K ϩ channels in cardiac myocytes isolated from different species and from different regions of the heart (27,28). These currents have been classified into two broad categories: 1) rapidly activating and inactivating transient outward K ϩ currents, I to and 2) slowly or non-inactivating K ϩ currents, typically referred to as I K . In adult mouse atrial myocytes, at least three kinetically distinct components of Ca 2ϩ -independent depolarization-activated K ϩ currents have been documented: a fast, transient outward current, I to,f , a rapidly activating, slowly inactivating current, I K,s , and a noninactivating, steady-state current, I ss (18). It has been demonstrated that K v 4 ␣ subunits underlie mouse atrial and ventricular I to,f (29). In addition, ␣ subunits of both K v 1.5 and K v 2.1 contribute to the generation of I K,s in mouse atrial and ventricular myocytes (30,31) and that K v 2.1 also contributes to mouse atrial I ss (18).
Even though the diversity of the voltage-gated K ϩ channels in cardiac myocytes has been well documented, the identification and functional significance of K Ca in the heart has not been previously described. In general, to ensure stable recording conditions, whole-cell currents are recorded using relatively high concentration of Ca 2ϩ chelators, which would mask the Ca 2ϩ -activated currents. The contribution of SK current to the cardiac repolarization should parallel the variations in internal Ca 2ϩ concentration during AP and as such be voltage dependent (21). To illustrate the dependence of the SK current on intracellular Ca 2ϩ , we performed experiments using voltageramp protocol to simulate AP. The experiments were carried out using perforated-patch technique to preserve intracellular milieu and were performed in the absence of external K ϩ , Ca 2ϩ , or Na ϩ channel blockers. In this recording condition, there exists an outward component, which is sensitive to low concentration apamin and DQ, known blockers of SK currents. are overlined. The calmodulin binding domain (CaMBM) is located in the carboxyl terminus and is shaded in yellow. Residues that are identical among all of the clones are marked by an asterisk, while a colon marks the highly conservative and a period the weakly conservative substitutions, respectively, and no mark represents nonconservative substitution. Amino acid numbers for the full-length coding sequences are given on the right.
To directly illustrate the presence of an apamin-sensitive outward current in human atria, we used a two-pulse protocol as illustrated in Fig. 2D. In these experiments, 4-AP and E4031 were used to minimize the contribution from voltage-gated K ϩ channels. In addition to its effect on I to , 4-AP at 2 mM has previously been shown to also block the sustained outward currents in isolated human atrial myocyte (32). Therefore, the contribution of the SK current to the total repolarizing current in human atria cannot be directly assessed from these experiments. However, AP recordings performed in human as well as mouse atrial myocytes (Fig. 1) confirm the importance of the SK current in the cardiac repolarization in these tissues. In these experiments, the perforated-patch technique was used. In addition, the experiments were performed in the absence of Na ϩ , Ca 2ϩ , or K ϩ channel blockers. Even though the Ca 2ϩ -dependent apamin-sensitive current represents a modest component of total sustained current (Fig. 2C), we predict that the current remains active after the large transient outward K ϩ current (I to ) is inactivated. Therefore, the current plays an important role in the later part of the repolarization when the total inward and outward currents become small as shown with the AP recordings.
Physiological Significance-SK channels have been shown to play an important role in setting the tonic firing frequency of neurons (33). Their activation causes membrane hyperpolarization, which inhibits cell firing and limits the firing frequency of repetitive APs. The increase in intracellular Ca 2ϩ evoked by AP firing decays slowly, allowing SK channel activation to generate a long lasting hyperpolarization, termed the slow afterhyperpolarization. This spike-frequency adaptation protects the cell from the deleterious effects of continuous tetanic activity. The possible roles of K Ca channels in cardiac myocytes have not previously been documented. In contrast to the hyperpolarization effects of SK channels in neuron, in the heart, the current contributes markedly toward the late phase of the cardiac repolarization. The importance of the current is underpinned by the fact that the late phase of the cardiac AP is susceptible to aberrant excitation, e.g. early after depolarization and arrhythmias. It is conceivable that alteration in intracellular Ca 2ϩ under certain pathologic conditions may produce profound changes in the AP profiles via this Ca 2ϩ -dependent channel. For example, during the acute phase of paroxysmal atrial fibrillation, the rapid depolarization may increase intracellular Ca 2ϩ and potentiate the I K,Ca and thereby is responsible at least in part for acute electrical remodeling with shortened AP. This phenomenon of electrical remodeling has been well documented in several models of atrial fibrillation. Furthermore, the importance of these findings transcends its novelty. Because SK2 channel is differentially expressed in atria versus ventricles, the use of specific ligands for SK channel may offer a unique therapeutic opportunity to directly modify the atrial cells without interfering with ventricular myocytes.
Previous studies using cloned SK channels and SK1-SK2 dimer have demonstrated that SK subunits may form heteromeric channels, giving rise to structural and functional diversity (2). It is likely that cardiac myocytes may also express more than one SK channel subtypes and this may explain the higher IC 50 to apamin block in the ventricular cells compared with atrial myocytes. A lower density of SK2 channel in the ventricles alone would not suffice to explain the higher IC 50 to apamin. In particular, the SK2 channel expressed in atria was identical in sequence to that expressed in the ventricles. However, it remains to be shown in cardiac myocytes whether other SK channel isoforms exist and their specific roles in the heart.