Cardiac Purkinje fibers play an important role in cardiac arrhythmias, but no information is available about ionic currents in human cardiac Purkinje cells (PCs). PCs and midmyocardial ventricular myocytes (VMs) were isolated from explanted human hearts. K+ currents were evaluated at 37°C with whole cell patch clamp. PCs had clear inward rectifier K+current (I K1), with a density not significantly different from VMs between −110 and −20 mV. A Cs+-sensitive, time-dependent hyperpolarization-activated current was measurable negative to −60 mV. Transient outward current (I to) density was smaller, but end pulse sustained current (I sus) was larger, in PCs vs. VMs. I to recovery was substantially slower in PCs, leading to strong frequency dependence. Unlike VMI to, which was unaffected by 10 mM tetraethylammonium, Purkinje I to was strongly inhibited by tetraethylammonium, and PurkinjeI to was 10-fold more sensitive to 4-aminopyridine than VM. PC I sus was also reduced strongly by 10 mM tetraethylammonium. In conclusion, human PCs demonstrate a prominent I K1, a time-dependent hyperpolarization-activated current, and an I towith pharmacological sensitivity and recovery kinetics different from those in the atrium or ventricle and compatible with a different molecular basis.
- ion currents
- cardiac Purkinje cells
- potassium channel blockers
cardiac purkinje fibers play a key role in conduction and arrhythmogenesis. As the prime cellular component of the cardiac conducting system, they are critical for assuring appropriate timing and sequence of ventricular contraction. They appear to play a particularly important role as a generator of early afterdepolarizations and an initiator of transmural reentry in torsades de pointes arrhythmias associated with long QT syndromes (2, 3, 10, 23). In addition, there is evidence for significant participation of Purkinje cells (PCs) in ventricular tachyarrhythmias due to delayed afterdepolarizations (35), intraventricular reentry (26), and ventricular fibrillation (7).
K+ currents are a key determinant of cellular repolarization and, consequently, of the occurrence of cardiac arrhythmias. Recent work has shown that the properties of K+ currents in canine PCs differ from those in ventricular myocytes (14). Furthermore, the properties and molecular basis of specific K+ currents in human cardiac cells may be chamber specific and distinct from those of corresponding regions in hearts of other animals (11, 19, 41, 43). We were unable to identify any voltage-clamp studies of human cardiac PCs in the literature. The objectives of the present study were to isolate PCs from free-running false tendons of human hearts explanted at the time of cardiac transplantation and to characterize a variety of ionic currents by voltage-clamp recording.
MATERIALS AND METHODS
Free-running false tendons and left ventricular free wall midmyocardium were obtained from nine explanted failing human hearts removed at the time of cardiac transplantation. Midmyocardial myocytes were used because they share some repolarization and phase 0 upstroke properties with Purkinje tissue (32). Table1 lists the clinical characteristics of the patient population. Procedures for obtaining the tissues were approved by the Research Ethics Committee of the Montreal Heart Institute.
Free-running endocardial false tendons were excised quickly from both ventricles into modified Eagle's minimum essential medium (pH 7.0 with HEPES-NaOH; GIBCO-BRL), followed by incubation with the same solution containing collagenase (1,500 U/ml; Worthington type II) and 1% bovine serum albumin (Sigma). False tendons were distinguished from trabeculae in that they were pale in color and thinner and were removed from either ventricle by cutting with fine scissors. The fibers were agitated by continuous bubbling with 100% O2 in a 37°C shaker bath for 4–6 h. After the endothelial sheath had been digested, revealing single cells and/or cell columns under light microscope, the digested fibers were washed twice with the high-K+ storage solution and were incubated for an additional 10 min. Individual cells were dispersed by gentle hand pipetting, harvested by centrifugation, and kept in a high-K+ storage solution. Human ventricular myocytes were isolated with the use of coronary artery perfusion methods described previously in detail (19). Both PCs and ventricular myocytes were Ca2+ tolerant and were studied concurrently within 18 h of isolation. PCs were characterized by a typical spindle-shaped, narrower, and longer morphology with much less prominent cross-striations (Fig. 1) compared with ventricular myocytes, from which they looked clearly different under light microscopy.
Standard Tyrode solution for cell isolation and patch-clamp studies contained (in mM) 136 NaCl, 5.4 KCl, 1.0 MgCl2, 1.0 CaCl2, 0.33 NaH2PO4, 5.0 HEPES, and 10 dextrose, pH 7.4 (adjusted with NaOH). High-K+ storage solution contained (in mM) 20 KCl, 10 KH2PO4, 10 dextrose, 70 glutamic acid, 10 β-hydroxybutyric acid, 10 taurine, 10 EGTA, and 0.1% albumin, pH 7.4 (adjusted with KOH). The pipette solution contained (in mM) 110 K+-aspartate, 20 KCl, 1 MgCl2, 5 Mg2ATP, 10 HEPES, 5 phosphocreatine, 0.1 GTP, and 5 EGTA, pH 7.2 (adjusted with KOH).
Atropine (1 μM) was included in the extracellular solution to eliminate any basal activity of acetylcholine-activated current (16) and 4-aminopyridine (4-AP)-dependent K+currents (30). CdCl2 (200 μM) was added to block L-type Ca2+ current (I Ca) and Ca2+-activated Cl− current. T-type Ca2+ current was suppressed by holding at −50 mV and in one protocol in which a more negative holding potential (HP) (−80 mV) was used by pulsing to a voltage (+50 mV) at which T-typeI Ca in PCs is minimal (30). Contamination by Na+ current was prevented with an HP of −50 mV or, when a more negative HP was necessary, with Tris · Cl substitution for extracellular NaCl.
Data Acquisition and Analysis
The whole cell patch-clamp technique was used to record currents at 37°C as previously described in detail (11, 41). The capacitance of PCs averaged 199 ± 17 pF (n = 20) and the capacitance of midmyocardial myocytes was 237 ± 23 pF (n = 18). Compensated series resistances and capacitive time constants averaged 2.7 ± 0.2 MΩ and 0.40 ± 0.02 ms for PCs and 1.7 ± 0.4 MΩ and 0.40 ± 0.05 ms for ventricular myocytes, respectively. Maximum mean voltage drop across the series resistance was 6 mV. Junction potentials between bath and pipette solutions averaged 10 mV and were not corrected. Drug effects were studied at steady state, after 5 min of exposure at each concentration.
Group data are presented as means ± SE. Nonlinear curve fitting was performed with Clampfit in pCLAMP 6. Student's t-tests were used for statistical comparisons. A two-tailed P< 0.05 indicated statistical significance.
The inward rectifier K+ current (I K1) was recorded as current sensitive to 1 mM Ba2+. Figure 2, A and B, shows typical recordings obtained from a PC and a ventricular myocyte, respectively, on voltage steps from −40 mV before and after the addition of 1 mM Ba2+ to the extracellular solution. In these cells, virtually all of the current elicited on voltage steps from −40 mV was suppressed by 1 mM Ba2+. Ba2+-sensitive currents were obtained by digital subtraction of recordings before and after Ba2+, as illustrated in Fig. 2, A andB. Figure 2 C shows means ± SE ofI K1 density in five PCs and nine ventricular myocytes as a function of test potential. Currents in both cell types reversed at approximately −70 mV, which, when corrected for the junction potential, provide an estimated reversal potential of approximately −80 mV. I K1 in both cell types showed strong inward rectification with a small but distinct outward component between approximately −70 and −20 mV, as shown on an expanded scale in Fig. 2 D. No statistically significant differences in I K1 density were observed over the voltage range between −110 and −20 mV; however,I K1 density was significantly greater in ventricular myocytes than in PCs at −120 mV.
In 5 of 11 PCs subjected to hyperpolarizing steps, clear time-dependent currents were seen on hyperpolarization. Figure3 A shows time-dependent hyperpolarization-activated current (I H) elicited on 3,910-ms voltage steps from an HP of −50 mV. Such currents were strongly suppressed by 2 mM Cs+, as shown in Fig.3 B. Figure 3 C shows Cs+-sensitive current obtained by digital subtraction. Figure 3 D shows mean I H density-voltage relations in five PCs.I H was relatively large (e.g., mean density −4.5 ± 1.0 pA/pF at −120 mV). Measurable time-dependent inward currents were detectable at voltages as positive as −60 mV. Current activation kinetics were biexponential, with mean time constants (τ) in five PCs as shown in Fig. 3 E. The proportion of activation in the faster phase decreased progressively at more positive voltages (Fig. 3 F). Current activation became slower at more positive potentials, both because of significant voltage-dependent slowing of the rapid phase time constant (τ fast) and a decreasing proportion of fast phase activation at more positive voltages. No time-dependent activating current was observed on hyperpolarization of ventricular myocytes (12 cells from 4 hearts).
Transient Outward Current
Voltage- and time-dependent properties.
Figure 4, A and B, shows typical recordings of transient outward current (I to) from a PC and a ventricular myocyte. Because of slower inactivation in PCs, longer pulses (300 ms) were used for PCs to permit steady-state inactivation (note the difference in time scale for Fig.4, A vs. B).I to activated and inactivated rapidly in both cell types, but the sustained “pedestal” component was larger in PCs. Inactivation was best fitted by biexponential functions. Figure4 C shows mean inactivation time constants in eight PCs and seven ventricular myocytes. Both kinetic components were slower in PCs than in ventricular muscle. Figure 4 D shows current density-voltage relationships for I to in eight PCs and seven ventricular myocytes. I to density (measured from peak current to end-pulse steady state) was significantly smaller in PCs between +20 and +60 mV. However, sustained current (I sus; measured from the end-pulse level to the 0-current level) density in the same cells was significantly larger between −10 and +60 mV in PCs (Fig. 4 E). Figure4 F shows mean data for I to activation and inactivation voltage dependence based on recordings in five PCs and five ventricular myocytes. Inactivation voltage dependence was determined with the use of 1,000-ms conditioning pulses followed by 200-ms test pulses to +50 mV. Activation voltage dependence was determined from the relation IV =I max(GV /G max)(V− V rev), where IV andI max are currents at voltage (V) and maximum current (at the most positive test potential), respectively;GV and G max are conductances at voltage and maximum conductance, respectively; andV rev is reversal potential. Data were fitted by Boltzmann relations, as illustrated in Fig. 4. The half-maximal inactivation voltage averaged −27 ± 2 mV in PCs and -21 ± 2 mV in ventricular myocytes (P = not significant). The inactivation slope factor in PCs was −13 ± 2 mV, significantly larger than in ventricular myocytes (−5 ± 1 mV, P < 0.01). Half-maximal activation occurred at +20 ± 9 and +17 ± 6 mV in PCs and ventricular myocytes, respectively (P = not significant), and the activation slope factor averaged 13 ± 5 and 14 ± 5 mV in PCs and ventricular myocytes (P = not significant), respectively.
An analysis of I to recovery kinetics is shown in Fig. 5. Figure 5 A showsI to recordings during paired pulses from a PC and a ventricular myocyte during 100-ms conditioning pulses (P1) from a HP of −80 mV to +50 mV, followed by currents during test pulses (P2) with the same duration and voltage at varying P1-P2 intervals (paired pulse protocol delivered at 0.1 Hz in ventricular myocytes and 0.066 Hz in PCs). PC I to recovery was extremely slow, whereas I to recovery was much faster in the ventricular myocyte. Figure 5, B and C, shows means ± SE data for the time dependence ofI to reactivation in PCs (n = 4) and ventricular myocytes (n = 6), respectively. The ratio of I to during P2 to that in P1 (I 2/I 1ratio) was plotted as a function of the P1-P2interval and fit with a biexponential function. Ventricular muscleI to had a larger portion (69 ± 5%) of reactivation in the rapid phase with a time constant (τ 1) of 12 ± 1 ms, whereas PCI to had a smaller rapid phase (38 ± 2% of total, P < 0.01) with a time constant of 31 ± 7 ms. The majority of Purkinje I to recovery (62 ± 1%) proceeded very slowly, with a slow phase time constant (τ 2) of 1,575 ± 189 ms (n = 4). Ventricular I to slow phase τ 2 averaged 217 ± 55 ms (n = 6, P < 0.001 vs. PC). Correspondingly, Purkinje I to was much more frequency dependent than ventricular I to (Fig.5 D).
Sensitivity to K+ channel blockers.
Sensitivities to the K+ channel blockers tetraethylammonium (TEA) and 4-AP are characteristic properties of K+ currents (24, 25). Figure 6, A and B, shows recordings of I to from a representative PC and a ventricular myocyte obtained on depolarization to +50 mV before and after exposure to a series of 4-AP concentrations. Whereas 500 μM 4-AP almost completely suppressed I to in the PC, it caused <50% I to inhibition in the ventricular myocyte. Figure 6 Cshows mean concentration-response data, along with best fit Hill equations. The 50% blocking concentration (IC50) for 4-AP at +50 mV was 56 ± 7 μM for Purkinje I to(n = 4), significantly less than the IC50of 511 ± 110 μM (n = 5, P < 0.01) for ventricular I to. Hill coefficients averaged 1.9 ± 0.6 and 1.8 ± 0.6 (P = not significant), respectively.
Figure 6, D and E, illustrates the response of a ventricular myocyte and a PC to 10 mM TEA. I toin the ventricular myocyte was not appreciably different before compared with after TEA exposure (Fig. 6 D). In four ventricular myocytes, TEA did not significantly affectI to, with a density averaging 9.3 ± 1.3 and 9.1 ± 1.4 pA/pF (P not significant) at +50 mV, respectively, before and after TEA. In contrast, the same concentration of TEA strongly inhibited Purkinje I to, as illustrated by currents from one PC recorded before and after TEA (Fig.6 E). Figure 6 E, inset, shows meanI to in five PCs before and after TEA, which caused a 67 ± 9% (P < 0.05) current reduction.
As mentioned in Transient Outward Current, we found a substantially larger sustained end-pulse current afterI to inactivation in human PCs than in ventricular myocytes. The end-pulse current in PCs was inhibited by 4-AP, with an IC50 of 54 ± 15 μM at +50 mV. For ventricular myocytes, 4-AP had no significant effect on end-pulse current, with a mean 4.6 ± 2.1% reduction at 1 mM. The end-pulse sustained component in PCs was also sensitive to TEA, which decreasedI sus by 46 ± 5% (n = 5,P < 0.01) at 10 mM, in contrast to a 4.3 ± 1.7% decrease (P = not significant) in ventricular myocytes at the same concentration. In an attempt to dissociateI sus from I to, we recorded I sus in additional cells with the use of 100-ms prepulses from −50 to +50 mV 10 ms before test depolarizations to suppress I to, as previously used to study ultrarapid delayed rectifier current (I Kur) in human atrial myocytes (11, 19, 37), followed by a repolarizing step to −40 mV to observe tail currents. I sus either activated very rapidly or was instantaneous (activation kinetics could not be resolved) and showed slow inactivation (Fig. 7, A andC) with no tail currents. 4-AP (50 μM) decreased the current modestly by 33 ± 8% (P = not significant) in five cells (Fig.7 B). TEA (10 mM) reduced the current (Fig. 7 D) by 42 ± 9% (P < 0.05, n = 5).
We have succeeded in isolating human cardiac PCs, in which we have characterized a variety of ionic currents, includingI K1, I H,I to, and I sus. These currents were compared with corresponding currents in human ventricular myocytes with some potentially important differences observed, particularly for I to andI sus.
Comparison with Previous Studies of Corresponding Currents in PCs
Although the free-running Purkinje fiber false tendon preparation was widely used for classical voltage- clamp studies in multicellular preparations, much less work has been done in isolated PCs, largely because of the difficulty of isolating single PCs. Cordeiro et al. (6) found I K1 to be much smaller in rabbit PCs than in ventricular myocytes. In the present study, we found human PC I K1 to be substantial. PCI K1 density in our study was significantly smaller than that in ventricular myocytes at −120 mV, but was not significantly different from ventricular muscleI K1 density between −110 and 0 mV. Cordeiro et al. (6) found I to to be inhibited by 4-AP, but the concentration dependence of inhibition was not characterized. We (14) previously evaluated in detail the properties of canine PC I to. Like human PCI to in the present study, canine PCI to recovered more slowly than ventricularI to, with a large, slowly recovering component. Canine PC I to had time constants averaging 35 and 1,427 ms, quite similar to the values of 31 and 1,575 ms for human PC I to in the present study. The 4-AP sensitivities of canine PC and ventricular I towere also quite similar to those of human PC and ventricularI to in the present study (14). Additional findings in canine hearts similar to those in the present study were the TEA sensitivity of canine PC I toand the TEA insensitivity of canine ventricularI to (14).
Callewaert et al. (4) studied pacemaker current in single sheep PCs and found that activation accelerated at more negative potentials. Properties of the I H we recorded in human PCs are similar to those described by Cellewaert et al. (4). Human PC I H had a rapid kinetic component that activated more rapidly and constituted a larger proportion of current activation at more negative potentials. We found human PC I H activation kinetics to be biexponential, whereas Callewaert (4) used a single time constant to characterize pacemaker current activation. The voltage dependence and time dependence of I H in human PCs would be consistent with a role in pacemaker function, although any such inferences must be very cautious in view of our inability to record action potentials from human PCs.
Comparison with Other Studies of K+Currents in the Human Heart
We were unable to identify previous studies of ionic currents in human cardiac PCs with which to compare our results. Several studies have evaluated human ventricular myocyte I to. Wettwer et al. (40) found I toactivation and inactivation voltage dependence in human ventricular muscle cells that was similar to what we observed. They obtained only a single recovery time constant (∼20 ms), similar to the rapid phase time constant in our ventricular myocytes. Näbauer et al. (22) observed regional differences in human ventricularI to, with I to being smaller and recovering more slowly in subendocardial compared with subepicardial tissue. Time constants of I torecovery were of the order of 10 and 900 ms for both tissue types, with the fast phase constituting 89% of recovery in subepicardium vs. 4% of subendocardium. Li et al. (18) characterizedI to in epicardial, midmyocardial, and endocardial regions of the human heart, reporting biexponential recovery with time constants averaging 12, 20, and 34 ms for fast phase recovery in epicardium, midmyocardium, and endocardium, respectively, and slow phase time constants of 229, 254, and 490 ms. Recovery was significantly slower in endocardium than in epicardium or midmyocardium, and midmyocardial values were of the same order as our observations for midmyocardial ventricular myocytes. Amos et al. (1) studied I to in human atrial and ventricular subepicardial cells. They noted a recovery time constant of 24 ms for ventricular I to. Neither atrial nor ventricular I to were affected by 20 mM TEA (1). Cerbai et al. (5) recorded hyperpolarization-activated current (I f) in ventricular myocytes from failing human hearts. They noted a density of ∼2 pA/pF at −120 mV, about half of the value forI H in human PCs in the present study.I f activation was found to be monoexponential, with a time constant that increased at more positive voltages, consistent with the voltage dependence of I Hactivation kinetics that we observed.
In the present study, we noted a larger I sus in human PCs compared with ventricular myocytes. Human atrial myocytes also have a larger I sus compared with ventricular myocytes (11). Several observations suggest that PC I sus has a different basis compared with atrial. Human atrial I sus shows tail currents, even at 37°C (12); we did not observeI sus tail currents in human PCs at either room temperature or 37°C. Human atrial I sus is insensitive to TEA (37), whereas PurkinjeI sus was substantially inhibited by TEA. In fact, the response of PC end pulse current to 4-AP (IC5054 ± 15 μM) and 10 mM TEA (46 ± 5% reduction) was quite similar to the effects on PC I to of 4-AP (IC50 56 ± 7 μM) and 10 mM TEA (67 ± 9%), suggesting that PC I sus is a non- or slowly inactivating component of PC I to rather than a distinct TEA-insensitive current like the I Kurthat underlies most of human atrial I sus(11, 19).
PCs play an important role in a variety of cardiac arrhythmia mechanisms (2, 3, 7, 10, 23, 26, 35). They seem to be particularly significant in the generation of torsades de pointes arrhythmias associated with the long QT syndrome (2, 3, 10,23), in which abnormalities in repolarization are central. It is therefore important to understand the properties of K+currents that govern PC repolarization. The present study is, to our knowledge, the first voltage-clamp study of K+ currents in human cardiac PCs. We previously showed that canine cardiac PCs haveI to properties different from those of ventricular and atrial I to (14). In the present study, we found that human PC I tohas very similar voltage-dependent, kinetic, and pharmacological sensitivity patterns to canine PC I to, with clear differences from human ventricular I to. Furthermore, the sensitivity of human PC I to to TEA contrasts with the TEA insensitivity to concentrations as high as 100 mM (9, 29, 33) of the α-subunits (Kv1.4, Kv4.2, and Kv4.3) believed to underlie mammalian atrial and ventricularI to (25, 33, 38). These results are compatible with a role for α-subunits other than Kv1.4 and Kv4.x in human PC I to, such as TEA-sensitive Kv3-related subunits (14), which have been found to be expressed at the mRNA (15) and protein (unpublished data) levels in canine PCs.
I f is believed to play an important role in pacemaker function of automatic cardiac tissues, like Purkinje fibers (8, 31). We found a robust current compatible withI f in human cardiac PCs, consistent with PC function as cardiac escape pacemakers. We have referred to the current as I H rather than I f, because due to the limited number of cells available we could not characterize the current in biophysical detail. SubstantialI H was present at −70 mV, and measurable current was still present at −60 mV, compatible with the voltage range of pacemaker activity in PCs of other species. Shi et al. (31) have shown that rabbit Purkinje fibers are rich in mRNA corresponding to hyperpolarization-activated cation channel (HCN)1 and HCN4, which encode cyclic nucleotide-sensitive cation channels with distinctive properties that strongly resemble I f(21). Because expression studies of HCN channels have generally been performed at room temperature (21), it is difficult to compare directly the kinetics and voltage dependence of human PC I H components with those of cloned channels. Further work in this area is of potentially great interest.
Isolation of PCs from human cardiac false tendons was very difficult, with a very small number of Ca2+-tolerant PCs available from each isolation. Action potentials could be recorded only rarely and had an abnormal, triangularized morphology. We were unable to record delayed rectifier K+ currents, known to be particularly sensitive to isolation technique (42). Of note, PCs are notoriously difficult to isolate even from normal animal hearts, and several studies have been unable to record significantI K from PCs (6, 27) isolated from the hearts of animals in which I K blockers are known to prolong Purkinje fiber action potential duration (20,34). The currents we selected for study were large and had reproducible properties across cells. I to, in particular, is known to be resistant to cell isolation (42).
Hearts from which PCs were isolated were clearly diseased and patients were taking a variety of medications that could have affected the results. These are well-recognized limitations of virtually all voltage- clamp studies of human ventricular myocytes in the literature, which have almost invariably been obtained from explanted recipient heart tissue obtained at the time of heart transplantation. The only exceptions are studies from countries in which explanted donor hearts are used for valve transplantation and cardiac tissue may be available for electrophysiological study (36), and studies of small (∼5 mg) tissue cores obtained by myocardial biopsy (17,39). Myocardial biopsy specimens are not useful for PC isolation. Despite the fact that the PCs we obtained were from diseased hearts, we were able to record large currents with analyzable properties compatible with those of corresponding currents previously studied in animal models. PC ionic currents are remodeled by congestive heart failure (13), with downregulation of bothI K1 and I to densities but no change in their other properties. We were careful to compare currents in PCs with those of ventricular myocytes from the same hearts; because comparable downregulation occurs in ventricular myocytes (28) and PCs (13), the comparisons should have at least qualitative validity. Nevertheless, our results must be considered in the context of the potential role of cardiac disease and cell isolation effects.
We identified PCs on the basis of their characteristic appearance and the fact that they were isolated from free-running false tendons. Previous studies (4, 6, 27) have used similar approaches. Unfortunately, there is no independent marker that typifies PCs with absolute certainty, requiring us (as well as previous investigators) to rely on these criteria. The characteristic features of human PCI to we observed that are different from those of human ventricular I to in the present study and from mammalian atrial or ventricular I to in numerous previous studies make it unlikely that our PC population was contaminated by ventricular myocytes.
Unlike Cerbai et al. (5), we were unable to recordI f from ventricular cardiomyocytes of failing hearts. We are unable to account for this discrepancy in a direct way, although it may be related to differences in patient population and/or recording methods. We used Cd2+ to inhibitI Ca in studies of I to andI sus. Divalent cations like Cd2+ can produce shifts in current voltage dependence by neutralizing fixed negative surface charges, a phenomenon that must be kept in mind when considering our results.
We thank Evelyn Landry for expert technical assistance and France Thériault and Annie Laprade for secretarial help with the manuscript.
This study was funded, in part, by the Canadian Institutes of Health Research and the Quebec Heart Foundation. W. Han was supported by a Heart and Stroke Foundation of Canada studentship and G. Schram is the recipient of a Canada Institutes of Health Research/Aventis fellowship.
Address for reprint requests and other correspondence: S. Nattel, Research Center, Montreal Heart Institute, 5000 Belanger St. East, Montreal, Quebec H1T 1C8, Canada (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
August 15, 2002;10.1152/ajpheart.00389.2002
- Copyright © 2002 the American Physiological Society