The several effects of calcium on cardiac electrical

Transcription

1 Brief Communication 411 Reconstitution and Characterization of a Calcium-Activated Channel From Heart Joseph A. Hill Jr., Roberto Coronado, and Harold C. Strauss This paper is the first description of a calcium-activated nonspecific cation channel in adult ventricular muscle. We report gating kinetics and ionic selectivity data from experiments performed at the single channel level. Calcium activation is described by a gating model wherein two ions are involved in the reaction. Channel gating exhibited marked voltage dependence. Ionic selectivity experiments indicated that the channel is cation-selective but unable to discriminate between Na + and K +. We discuss evidence that this channel mediates transient inward current in ventricular tissue; thus, the channel may be involved in afterdepolarization-induced cardiac arrhythmias. (Circulation Research 1988;62: ) The several effects of calcium on cardiac electrical properties are the subject of intense interest at present. For example, an oscillatory inward current exists in calcium-overloaded cardiac muscle following depolarizing voltage-clamp steps. 1 " 3 At present, the molecular mechanism that underlies this transient inward current (LJ remains unidentified. A substantial body of literature supports the hypothesis that this current is calcium-induced 1-34 and involved in the genesis of afterdepolarizations and aftercontractions. 1-3 In addition, it is known that afterdepolarizations can trigger extra beats in isolated cardiac preparations 5-6 that sometimes develop into runs of repetitive activity. Thus, 1^ may be an important pathological mechanism underlying the clinical events accompanying calcium overload. Channel opening gives rise to an elementary current pulse, the size and duration of which can be observed directly by using single-channel recording techniques. One such technique, planar bilayer reconstitution, involves the incorporation of native channel protein into an artificial membrane with simultaneous measurement of transmembrane current at high gain. Using this technique, we describe a calcium-activated channel from cardiac ventricular sarcolemma that would mediate inward current under physiological conditions. As such, these data support the hypothesis that channel-mediated current plays a role in! and in From the Departments of Medicine and Pharmacology, Duke University Medical Center, Durham, North Carolina, and the Department of Physiology and Molecular Biophysics (R.C.), Baylor College of Medicine, Houston, Texas. Supported by National Heart, Lung, and Blood Institute grants 19216and 17670, and National Institute of General Medical Science grants 5 T32 GM and Address for correspondence: Joseph A. Hill Jr., MD, PhD, Division of Cardiology, Box 3845, Duke University Medical Center, Durham, NC Received March 26, 1987; accepted August 7, the development of afterdepolarizations and triggered arrhythmias. Materials and Methods Sarcolemmal vesicles from adult canine left ventricle were prepared according to the method of Jones. 7 Lipid bilayers were prepared by painting a 1:1 (wt/wt) phospholipid mixture of phosphatidylethanolamine and phosphatidylserine (Avanti Polar Lipids, Birmingham, Alabama) dissolved in decane (20 mg lipid/ml decane) across a 300-/xm diameter aperture separating two aqueous chambers (cis, trans). Typical bilayers exhibited pf capacitance. Resistances were always greater than 100 Gil. The ventricular vesicles were incorporated into the artificial membrane by adding them to the cis chamber under fusion conditions. 8 All aqueous solutions were buffered with 10 mm histidine (ph 7.1). Bath calcium activity was calculated in CaEGTA solutions using an effective stability constant of 4.31X10 6 M" 1 (0.1 M ionic strength, ph 7.1, 25 C). 9 All experiments were performed at room temperature (25-26 C) and under steady-state conditions. The cis chamber was connected to a voltage source while the trans chamber was held at virtual ground by a current-to-voltage converter circuit. Analog data were low-pass filtered (usually 3 db at 100 Hz, eight-pole Bessel) and stored on FM tape. Later, the signal was sampled and stored on magnetic disk for digital analysis. Linear regression was performed according to the method of least squares. Abbreviations used include: [X], concentration of species X; KOAc, potassium acetate; EGTA, ethylene glycol-bis03-amino-ethylether)a^,a f,a f ',A f '-tetraacetic acid; pca, -log([ca]). Results Gating Kinetics A representative continuous record of single channel current is illustrated in Figure 1A. As can be seen,

2 412 Circulation Research Vol 62, No 2, February rooms urn 56 ms 22 ms 9 ms 29 ms 500 TO 200 ms 50 ms 100 ms FIGURE 1. Single channel gating kinetics. Panel A, Continuous analog recording of single channel current, Vhold= + 10 mv,cis[kcl] = 73mM,tnns[KCl]=50mM, [Ca! +]<]00nM, open = up. In the record from which the illustration was taken, mean open time was 93 msec. Panel B, Cumulative probability distributions for open and closed states calculated from a record with 1,590 openings. Data were recorded at 40 mv in same solutions as in Panel A. Probability distributions were calculated according to the following protocol: current amplitude histograms were calculated, and means of approximately normally distributed current samples were identified. Transition discriminators for channel opening and closing were assigned midway between the two means. The sum of two exponential functions is superimposed on each histogram with time constants (ms) as depicted. gating occurs in bursts of activity separated by relatively long periods of silence. The open state (displayed up) exhibits flickery kinetics as it is interrupted by numerous short-lived closures. In all cases, distributions of open channel duration or closed state duration were fitted satisfactorily only by the sum of two exponential components (Panel B). Single channel open probability is modulated by cis calcium activity (Figure 2). Representative recordings selected from data points depicted in Panel B are shown in Panel A. Open probability ranges from 8% at 0.2 /xm Ca 2+ in the cis chamber to 99% at 10 mm Ca 2+ ; this effect is readily reversible upon the addition of EGTA. Popen, a function of cis calcium concentration defined as cumulative open time divided by the sum of cumulative open and closed times [0/(0 + C)], varied approximately between unity and 0 over the pca range of 6 to 7 (Panel B). This observation fits well with the experimentally observed values of intracellular calcium in ventricular myocytes under physiological conditions (approximately 100 nm diastolic Ca 2+ ) 10 and in states of calcium overload (1 pim and above)." 13 We have, however, observed shifts in the open state probability between different experiments (cf., Figure 2A and Figure 3) as have other investigators 14 ' 5 working with calcium-activated potassium channels. The sigmoidal curve superimposed on the data in Figure 2B is that predicted by a model wherein two calcium ions bind to the channel. A fit to the data was obtained by plotting 1/Popen versus [Ca 2+ ]~ 2 (Panel C). The slope of the fitted line was 3.1 X 10"" M 2 and is a measure of the equilibrium dissociation constant, K d. In all experiments (n = 9), the calcium-sensitive site was oriented toward the cis chamber. Methfessel and Boheim 14 have proposed an activation/blockade model to describe the gating kinetics observed in calcium-activated channels from skeletal muscle: Ca 2 Ca 2 R R-Ca< > R*-Ca R*-Ca,(l) where R is the channel receptor, and the asterisk signifies the open state. This linear scheme is a hybrid of two models, viz., an activation model in which a calcium ion binds to the closed channel allowing it to open (R <---> R-Ca <--> R*-Ca) and a "blockade" model in which a second calcium ion stabilizes the open configuration (R Ca < > R*-Ca< > R*-Ca 2 ). Scheme 1 allowed us to make several predictions for experimentally observable variables. First, the model predicts that except under limiting conditions (very high or very low [Ca 2+ ]), two exponential components will be required to describe the open and closed state probability distributions (Figure IB). Second, the model predicts that the ratio of mean open time to mean closed time (equilibrium constant, K) will be a function of calcium concentration raised to an exponent reflecting the number of ions that bind to the channel. A best-fit line obtained when these data are plotted as log(k) versus log([ca 2+ ]) had a slope of 2.5 (data not shown). Third, scheme 1 predicts mat under limiting conditions of calcium activity, mean open time is a linear function of [Ca 2+ ] and that mean closed time is

3 Hill et al Cardiac Calcium-Activated Channel (TIIDCT 10EI2) 1 / [CaT LOG [Co] FIGURE 2. Popen is a function o/cis calcium concentration. Panel A, Representative records are shown from data points in Panel B. Popen [defined as cumulative open time divided by the sum of cumulative open and closed times, 0/(0 + C)] is as listed; symmetrical [KOAc] = 100 mm, cis [Co 2 *] = 10 mm, 1.6 (JLM, 0.4 ym, 0.25 \xm, and0.2 fxm, trans [Cd*]<100 nm (open = up). Panel B, Popen is plotted versus logdca 2 *]). All recordings were made at 40 mv holding potential. The continuous curve is the transformed equation from the analysis in Panel C. Panel C, 1/Popen versus IllCa 2 *] 2. Coefficient of correlation frj of the fitted line is Panel D, Open and closed state durations are functions of cis [Ca 2 *]., log mean open time versus logdca 2 *]). The slope of the best fit line is 0.93, x = 0.93; D, log mean closed time versus logdca 1 *]). The slope of the best fit line is -1.5, r = a linear function of l/[ca 2+ ] (Figure 2D). In all cases, our observations were consistent with scheme 1. Channel gating was strongly voltage-dependent (Figure 3). In this experiment, cis and trans calcium concentrations are symmetrical and represent base line calcium for solutions in our laboratory in which neither calcium nor chelator is added. We have subsequently measured this calcium concentration as less than 100 nm. Single channel open probability varied from approximately 0 to 1 over the voltage range - 60 mv to 0 mv under these conditions. The sigmoidal activation curve depicted is a Boltzmann relation obtained as a fit to the logit transformation ln[(l - Popen)/ Popen] versus voltage. The slope of the fitted line is interpreted to indicate that 2.5 gating charges move from trans to cis upon channel opening. The v intercept obtained by this method indicates that -2.2 kcal/mol "nonelectrical" free energy are required for channel opening. That this nonelectrical component is less than 0 confirms that the open state is favored at 0 mv. Ionic Selectivity Figure 4 depicts single channel current as a function of holding voltage. Each data point represents the mean of at least three (and usually four to seven) amplitude measurements. Single channel current was ohmic over the voltage range - 80 to + 20 mv. In the presence of a KC1 activity gradient across the bilayer of 1.4, the interpolated reversal potential was 7 mv, which agrees well with the Nernst potential for K + (- 9 mv) (Panel A). In the presence of increasingly asymmetric KC1 solutions, die single channel current-voltage relation shifts toward more negative potentials. In each case, the experimentally determined reversal potential agreed with the Nernst potential for a K + electrode within 3 mv. Thus, we conclude that the channel selects strongly for cations over anions. We have also measured the reversal potential under biionic conditions for Na + andk + (Figure4B). Afitted line is superimposed on the data that has an x intercept of +2 mv and a slope of 120 ps. By assuming ionic independence, we may invoke the Goldman-Hodgkin- Katz equation for biionic conditions to calculate relative ionic permeabil ities. In so doing, it is clear that these data are consistent with a channel that is completely nonselective between Na + and K + ions. This observation suggests that the channel conducts current that reverses near 0 mv under physiological

4 414 Circulation Research Vol 62, No 2, February VOLTAGE/mV FIGURE 3. Popen is a function of holding potential. Symmetrical [KOAc] = 100 mm, symmetrical [Ca'+]<100 nm. The sigmoid curve is a Boltzmann relation calculated as described in the text. conditions. We interpret the conductance (120 ps) to represent a first approximation (neglecting temperature effects) to the channel's conductance in a physiological environment. At present, we are unable to quantify relative calcium permeability. However, the experiment shown in Figure 2 allowed us to estimate relative Ca 2+ permeability with respect to K + at less than 0.2. Single channel current increased less than 0.7 pa (5 %) at 40 mv over the pca range of 2 to 7 in the cis chamber; from this, we conclude that the channel exhibited relatively little calcium permeability. -JO 0 VOLTAGE /HV?0 0 JO VOLT»CE/oV FIGURE 4. Steady-state current-voltage (I-V) relations, [Ca! *]<]00 nm. Panel A, I-V curves recorded under the following conditions (cis [KCl]// trans [KCl], ratio of activities): mm, 1.4 (*), mm, 1.8 (*) mm, 5.4 (O). In every case, standard error for each data point was less than 0.2 pa (Panels A and B); Panel B, Biionic potential experiment performed under the following conditions: cis 700 mm NaOAc, trans 700 mm KOAc. Discussion Calcium-Activated Channels in Other Tissues This report represents the first description of single calcium-activated channels from working adult myocardium. Colquhoun et al 16 have described a calciumactivated channel from cultured neonatal rat heart that exhibited some of these same characteristics except that the single channel conductance was markedly lower, viz., ps in the presence of physiological salt solutions. In addition, they observed voltageindependent gating and calcium sensitivity in the 1-10 (xm range. We speculate such discrepancies may be due to species and tissue development differences. One report has been published on single channel measurements of a calcium-activated potassium-selective channel in bovine cardiac Purkinje fibers." Calciumactivated channels have also been described from skeletal" and cardiac" sarcoplasmic reticulum. The sarcoplasmic reticulum channels, however, differ from the one described here in several respects: 1) they select for divalent cations over monovalents by a factor of 10, 1S 2) are weakly voltage-dependent," 3) display open lifetime kinetics that are independent of calcium concentration, 20 4) have a calcium binding stoichiometry of 0.3 (R. Coronado, personal communication) instead of 2 (Figure 2C), and 5) are inhibited by cis calcium concentrations above 1 mm 21 unlike activity as shown in Figure 2A. Most of the biophysical measurements we report here are similar to those reported by others studying calcium-activated channels. 22 Specifically, channel gating displays bursting kinetics, a characteristic feature of calcium-activated potassium channels from skeletal muscle as well as calcium-activated nonspecific cation channels from cultured neonatal rat hearts." Indeed, it was the distinctive appearance of the open state that led us to investigate the possibility of calcium-dependent gating. Second, large conductance is characteristic of calcium-activated channels, 22 with a few notable exceptions, such as the calcium-activated channel of neonatal rat heart (see above). 16 Marked voltage dependence of gating and an agonist-receptor stoichiometry of 2 : 1 are features of calcium-activated channels in skeletal muscle In fact, the binding of two agonist ions for channel activation is in keeping with the general properties of many ligand-gated channels, e.g., calcium-activated channels, I5J3 acetylcholine receptor channels, 24 and glycine and y-aminobutyric acid receptors. 23 Finally, the calcium dependence of mean open time and mean closed time is similar to that observed in calcium-activated potassium channels from cultured skeletal muscle studied with the patch-clamp technique 14 or reconstituted into bilayers. 23 Significance We were able to reconstitute this channel into bilayers sporadically; this suggests that the channel may be present in the preparation in relatively small numbers. Further, the fact that the cardiac channel has never been observed with the patch-clamp technique

5 Hill et al Cardiac Calcium-Activated Channel 415 suggests that this channel may be localized to clefts in the sarcolemmal membrane as in skeletal muscle. However, in order to elucidate further its physiological role, it will be necessary to demonstrate its function directly using either patch clamping or high-affinity ligand binding techniques. In planning such experiments, it is a useful exercise to estimate channel density. We have estimated the channel's conductance in a physiological environment at 120 ps (see above). In addition, we have assumed 100 na of transient inward current in a Purkinje fiber strand of radius 100 pirn and length 1,000 pm (see Cannell and Lederer, 3 Figure 1). Finally, true membrane surface area was taken to be 10 times the apparent surface area of a smooth envelope of membrane. 26 Macroscopic current I is given by I = nip, where n is number of channels, i is single channel current, and p is channel opening probability (taken to be 1 at 1 /i.m calcium and 50 mv, Figure 1 A). Further, i = vy, where v is electrochemical driving force (taken to be 50 mv), and y is single channel conductance. With these assumptions, we have estimated the channel density to be 1 per 200 /urn 2. Thus, we predict that if this channel mediates 1^, then it will be sparsely distributed in the cell membrane; Cannell and Lederer 3 have presented evidence that the transient inward current in sheep Purkinje fibers is dependent on channel current (although Na-Ca exchange-mediated current or current from other electrogenic mechanisms may also be involved). The putative transient inward channel must exhibit calciumdependent gating over a physiological range of intracellular calcium, conduct inward current at diastolic membrane potentials, and select for cations over anions. Our measurements of calcium sensitivity, P Ni : P K approaching 1, and chloride impermeability are consistent with these predictions. Further, the steep voltage dependence we observed may explain the characteristic curvature in the macroscopic currentvoltage relation. 3 We have not tried to measure Ca 2+ current under the same conditions used by these investigators (replacing trans monovalent cations with Ca 2+ ); we do report, however, absence of substantial calcium permeability in the presence of monovalent cations. This observation is consistent with studies of calcium-activated channels from neuroblastoma. 27 For these reasons, we propose that this channel may contribute to transient inward current in canine ventricular muscle and may be involved in the development of triggered arrhythmias in calcium-overloaded tissue. Acknowledgments We gratefully acknowledge Dr. Jonathan Lederer for his critical reading of the manuscript. We thank Joseph C. Greenfield Jr. for continued support and S. Webb and B. Hill for editorial assistance. References 1. Kass RS, Lederer WJ, Tsien RW, Weingart R: Role of calcium ions in transient inward currents and aftercon tract ions induced by strophanthidin in cardiac purkinje fibres. J Physiol 1978^281: Kass RS, Tsien RW, Weingart R: Ionic basis of transient inward current induced by strophanthidin in cardiac purkinje fibres. J Physiol 1978;281: Cannell MB, Lederer WJ: The arrhythmogenic L, current in the absence of electrogenic sodium-calcium exchange in sheep cardiac purkinje fibres. J Physiol 1986;374: Eisner DA, Vaughan-Jones RD: Do calcium-activated potassium channels exist in the heart? Cell Calcium 1983;4: Ferrier GR: Digitalis arrhythmias: Role of oscillatory afterpotentials. Prog Cardiovasc Dis 1977; 19: Rosen MR, Reder RF: Does triggered activity have a role in the genesis of cardiac arrhythmias? AnnlntM ed 1981;94: Jones LR: Methods in Enzymology. (in press) 8. Miller C: Voltage-gated cation conductance channel from fragmented sarcoplasmic reticulum. Steady state electrical properties. J Membr Biol 1978;40: Martell AE, Smith RM: Critical Stability Constants. New York, Plenum Press, Wier WG, Cannell MB, Berlin JR, Marban E, Lederer WJ: Cellular and subcellular heterogeneity of [Ca 2+ ], in single heart cells revealed by fura-2. Science 1987;235: Cannell MB, Wier WG, Berlin JR, Marban E, Lederer WJ: Free intracellular calcium in normal and calcium-overloaded rat heart cells: Digital imaging fluorescent microscopy using fura-2 (abstract). Biophys J 1986;49:466a 12. Yue DT, Marban E, Wier WG: Relationship between force and intracellular [Ca 1+ ] in tetanized mammalian heart muscle. / Gen Physiol 1986;87: Valdeolmillos M, Eisner DA: The effects of ryanodine on calcium-overloaded sheep cardiac purkinje fibers. Circ Res 1985^6:452^ Methfessel C, Boheim G: The gating of single calciumdependent potassium channels is described by an activation/ blockade mechanism. Biophys Struct Mech 1982;9: Barrett JN, Magleby KL, Pallotta BL: Properties of single calcium-activated potassium channels in cultured rat muscle. J Physiol 1982;331: Colquhoun D, Neher E, Reuter H, Stevens CF: Inward current channels activated by intracellular Ca in cultured cardiac cells. Nature 1981 ;294: Callewaert G, Vereecke J, Carmeliet E: Existence of a calcium-dependent potassium channel in the membrane of cow cardiac purkinje cells. Pflugers Arch 1986;406: Smith JS, Coronado R, Meissner G: Sarcoplasmic reticulum contains adenine nucleotide-activated calcium channels. Nature 1985 ;316:446-^ Rousseau E, Smith JS, Henderson JS, Meissner G: Single channel and "Ca 2+ flux measurements of the cardiac sarcoplasmic reticulum calcium channel. Biophys J 1986;50: Smith JS, Coronado R, Meissner G: Single channel measurements of the calcium release channel from skeletal muscle sarcoplasmic reticulum: Activation by Ca 2+ and ATP and modulation by Mg^+. J Gen Physiol 1986;88: Meissner G, Darling E.EvelethJ: Kinetics of rapid Ca 2+ release by sarcoplasmic reticulum. Effects of Ca 2+, Mg 2+, and adenine nucleotides. Biochemistry 1986;25: Schwarz W, Passow H: Ca 2+ -activated K* channels in erythrocytes and excitable eel Is. Annu Rev Physiol 1983;45: Moczydlowski E, Latorre R: Gating kinetics of Ca 2+ -activated K + channels from rat muscle incorporated into planar lipid bilayers. J Gen Physiol 1983;82: Changeux J-P, Devillers-Thiery A, Chemouilli P: Acetylcholine receptor An allosteric protein. Science 1984;225: Sakmann B, Hamill OP, Borman J: Activation of chloride channels by putative inhibitory transmitters in spinal cord neurons (abstract). Pflugers Arch 1982;392:R Mobley BA, Page E: The surface area of sheep cardiac purkinje fibres. J Physiol 1972^220: Yellen G: Single Ca 2+ -activated non-selective cation channels in neuroblastoma. Nature 1982;296: KEY WORDS reconstitution calcium-activated channel triggered arrhythmia