L-type Ca²⁺ current as the predominant pathway of Ca²⁺ entry during I_Na activation in β-stimulated cardiac myocytes

Abstract

1. In the present study Ca²⁺ entry via different voltage-dependent membrane channels was examined with a fluorescent Ca²⁺ indicator before and after β-adrenergic stimulation.

2. To clearly distinguish between Ca²⁺ influx and Ca²⁺ release from the sarcoplasmic reticulum the Ca²⁺ store was blocked with 0.1 μm thapsigargin and 10 μm ryanodine. Omitting Na⁺ from the pipette filling solution minimized Ca²⁺ entry via Na⁺-Ca²⁺ exchange.

3. Individual guinea-pig ventricular myocytes were voltage clamped in the whole-cell configuration of the patch-clamp technique and different membrane currents were activated using specific voltage protocols. The intracellular Ca²⁺ concentration was simultaneously recorded with a laser-scanning confocal microscope using fluo-3 as a Ca²⁺ indicator.

4. Ca²⁺ entry pathways were discriminated using pharmacological blockers under control conditions and during β-adrenergic stimulation with 1 μm isoproterenol (isoprenaline) in the bathing solution or 100 μm cAMP in the patch-clamp pipette.

5. Isoproterenol or cAMP potentiated the Ca²⁺ influx signals recorded during L-type Ca²⁺ current activation but, more interestingly, also during Na⁺ current (I_Na) activation. The Ca²⁺ influx signal arising from L-type Ca²⁺ current activation was usually blocked by 50 μm Cd²⁺. However, the Ca²⁺ influx signal elicited by the Na⁺ current activation protocol was only curtailed to 56.4 ± 28.2% by 100 μm Ni²⁺ but was reduced to 17.9 ± 15.1% by 50 μm Cd²⁺ and consistently eliminated by 5 mm Ni²⁺.

6. The pronounced Cd²⁺ and moderate Ni²⁺ sensitivity of the Ca²⁺ influx signals suggested that the predominant source of Ca²⁺ influx during the Na⁺ current activation – before and during β-adrenergic stimulation – was a spurious activation of the L-type Ca²⁺ current, presumably due to voltage escape during Na⁺ current activation.

7. Calculations based on the relationship between Ca²⁺ current and fluorescence change revealed that, on average, we could reliably detect rapid Ca²⁺ concentration changes as small as 5.4 ± 0.7 nm. Thus, we can estimate an upper limit for the Ca²⁺ permeability of the phosphorylated TTX-sensitive Na⁺ channels which is less than 0.04:1 for Ca²⁺ ions flowing through Na⁺ channels via the proposed ‘slip-mode’ Ca²⁺ conductance. Therefore the slip-mode Ca²⁺ conductance of Na⁺ channels does not contribute noticeably to the Ca²⁺ signals observed in our experiments.

Excitation-contraction coupling (EC coupling) and Ca²⁺ signalling in cardiac muscle are thought to be mediated by Ca²⁺ influx via the L-type Ca²⁺ current which subsequently triggers a much larger Ca²⁺ release from the sarcoplasmic reticulum (SR) by Ca²⁺-induced Ca²⁺ release (CICR) (Fabiato, 1985; Bers, 1991). In contrast, skeletal muscle EC coupling is believed not to require an influx of Ca²⁺. Instead, a direct mechanical connection links the voltage sensors with the SR Ca²⁺ release channels to control Ca²⁺ release by an allosteric interaction (Schneider & Chandler, 1973; Rios et al. 1992). Recently, several additional pathways that may trigger Ca²⁺ release from the SR of cardiac muscle have been proposed. These include Ca²⁺ entry via the voltage-sensitive Na⁺-Ca²⁺ exchange during the depolarization (Levi et al. 1994), possibly enhanced by subsarcolemmal Na⁺ accumulation during I_Na activation (Leblanc & Hume, 1990; Lipp & Niggli, 1994). In addition, Ca²⁺ influx via the T-type Ca²⁺ current has been reported to trigger slow CICR (Sipido et al. 1998).

Even more trigger pathways have been noted in cardiac muscle after β-adrenergic stimulation. One of these mechanisms has been attributed to Ca²⁺ influx via TTX-sensitive Na⁺ channels occurring after phosphorylation (‘slip-mode Ca²⁺ conductance’; Santana et al. 1998; Cruz et al. 1999). By analogy with skeletal muscle, the existence of a purely voltage-activated SR Ca²⁺ release mechanism operating in the same voltage range has also been proposed (Hobai et al. 1997b; Ferrier et al. 1998). However, after β-adrenergic stimulation the CICR mechanism may operate under a high gain regime where a tiny influx of Ca²⁺ could be sufficient to trigger Ca²⁺ release. It is very difficult to rule out a small amount of Ca²⁺ entry via, for example, incompletely blocked Ca²⁺ channels (Trafford & Eisner, 1998). Therefore, we decided to apply a fluorescent indicator technique to examine Ca²⁺ influx via voltage-dependent membrane channels activated during the voltage-clamp protocols used in this study, both before and during β-stimulation. To analyse Ca²⁺ influx without contamination by Ca²⁺ release from the SR, CICR was suppressed with ryanodine and SR Ca²⁺ uptake was blocked with thapsigargin, except in a few initial control experiments. Ca²⁺ influx via the Na⁺-Ca²⁺ exchange was minimized by omitting Na⁺ from the pipette filling solution because this pathway has been investigated in separate studies (Lipp & Niggli, 1994; Niggli & Lipp, 1996). Preliminary findings on the analysed Ca²⁺ influx pathways have been communicated to the Biophysical Society in abstract form (DelPrincipe et al. 1999).

METHODS

Cell preparation

Experiments were performed on single ventricular myocytes isolated from the guinea-pig (Cavia porcellus). Adult animals were killed by cervical dislocation and the hearts rapidly removed and retrogradely perfused on a Langendorff perfusion system at 37°C. The perfusion solution contained (mm): NaCl 135, KCl 5.4, MgCl₂ 1, NaH₂PO₄ 0.33, Hepes 5, glucose 11, pH adjusted to 7.3 with NaOH. After 5 min, collagenase B (Boehringer Mannheim, Rotkreuz, Switzerland) and protease type XIV (Sigma, Buchs, Switzerland) were added to final concentrations of 0.2 mg ml⁻¹ and 0.04 mg ml⁻¹, respectively, and the perfusion continued for another 4–6 min. Subsequently, the ventricles were minced and placed in perfusion solution containing 200 μm CaCl₂ on a rocking table to allow for dissociation of the tissue.

Cells were taken from the supernatant, transferred into a recording chamber with a coverslip floor and mounted onto the stage of an inverted microscope (Diaphot TMD, Nikon, Küsnacht, Switzerland). Ca²⁺ resistant cells readily adhered to the uncoated coverslip and were constantly superfused (1–2 ml min⁻¹) with extracellular solution containing (mm): NaCl 140, KCl 5, MgCl₂ 1, CaCl₂ 1, Hepes 10, glucose 10, pH adjusted to 7.4 with NaOH. Most experiments were carried out at room temperature (22°C), but some control experiments were performed at 37°C.

Electrophysiological recordings

Recording electrodes were pulled from filamented borosilicate glass capillaries (GC150F, Clark Electromedical Instruments, Pangbourne, UK) on a horizontal puller (DMZ, Zeitz Instrumente, Augsburg, Germany) and filled with intracellular solution containing (mm): caesium aspartate 120, TEA-Cl 20, Hepes 10, MgATP 5, MgCl₂ 1, fluo-3 0.1, pH adjusted to 7.2 with CsOH. Sodium was omitted to minimize Ca²⁺ entry via reverse-mode operation of the Na⁺-Ca²⁺ exchanger. Typical pipette resistances were around 2 MΩ. Cells were voltage clamped in the whole-cell configuration and held at −70 mV without correction for junction potentials using an Axopatch 200 amplifier (Axon Instruments, Foster City, CA, USA).

Na⁺ inward currents were activated by holding the voltage at −90 mV for 1.5 s before stepping to −40 mV for 50 ms. In the presence of 5 mm Ni²⁺ we observed a shift of the Na⁺ current activation curve of about +20 mV which was due to surface charge screening effects (McLaughlin, 1989). Therefore, to activate a Na⁺ current of similar amplitude in the presence of 5 mm Ni²⁺ the voltage was stepped to −20 mV. L-type Ca²⁺ currents were activated by holding the voltage at −40 mV for 1.5 s before stepping to +10 mV for 200 ms. During the Ca²⁺ release experiments, the cells were paced with a series of depolarizing voltage pulses (from −70 to 0 mV for 200 ms at 1 Hz) before each sweep to activate L-type Ca²⁺ currents and refill the SR.

Series resistance was compensated to about 60–75% with the built-in compensation circuit of the amplifier. Current recordings showing poor voltage control due to changes in series resistance were not used for analysis. No on-line leak subtraction was performed. For calculations, where the pure Ca²⁺ current was to be integrated, the current recorded in the presence of 50 μm Cd²⁺ was subtracted off-line. Currents were low-pass filtered at 5 or 10 kHz and digitized at 10 kHz using an A/D converter and the LabView acquisition software (National Instruments, Ennetbaden, Switzerland). Data were stored on hard disk for later analysis with the IgorPro software (WaveMetrics, Lake Oswego, OR, USA).

Confocal Ca²⁺ measurements

Cells were viewed with a × 63 oil-immersion objective lens (Neofluar, NA = 1.25, Zeiss, Oberkochen, Germany) and loaded with fluo-3 through the recording pipette. Fluo-3 was excited with the 488 nm line of an argon laser (model 5000, Ion Laser Technology, Salt Lake City, USA) at 150 μW intensity on the cell. The fluorescence was detected at 540 ± 15 nm with the photomultiplier tube (PMT) of a laser-scanning confocal system (MRC 1000, Bio-Rad, Glattbrugg, Switzerland) operated in the line-scan mode. The scan speed was set to 2 or 6 ms per line. The 512 lines recorded in one frame thus corresponded to 1.024 or 3.072 s, respectively. When long current trains were recorded, the speed was set to 13.28 ms per line. Each line-scan image contained a signal from a red light diode which was synchronized with the voltage protocol and recorded with the second PMT of the confocal system (> 600 nm).

Fluorescence images were processed using a customized version of the public domain NIH Image program (developed at the US National Institutes of Health and available on the Internet at http://rsb.info.nih.gov/nih-image/). Mean Ca²⁺ concentration profiles were extracted from fluorescence images and calculated with the IgorPro software using an established self-ratio calibration procedure (Cheng et al. 1993). We assumed a resting Ca²⁺ concentration of 100 nm at the beginning of each experiment and a K_d for fluo-3 of 500 nm.

Materials

All chemicals used were of reagent grade and dissolved in distilled water of cell culture grade. Thapsigargin and tetrodotoxin (TTX) were purchased from Alomone Labs (Jerusalem, Israel), ryanodine from Calbiochem (La Jolla, CA, USA), isoproterenol ([-]-N-isopropyl-l-noradrenaline hydrochloride) from Fluka (Buchs, Switzerland), cAMP (adenosine 3′,5′-monophosphate, free acid) from Sigma (Buchs, Switzerland) and fluo-3 (pentapotassium salt) from TefLabs (Austin, TX, USA).

Cells were incubated with thapsigargin and ryanodine for 30 min after the dissociation process to block the SR Ca²⁺ pump and the ryanodine receptor. Thapsigargin was dissolved as 1 mm stock in ethanol and used at 0.1 μm. Ryanodine was dissolved as 1 mm stock in distilled water and used at a concentration of 10 μm. Isoproterenol was freshly prepared as a 10 mm stock in an aqueous 1 mm L-ascorbic acid solution before each experiment and added at a concentration of 1 μm to the extracellular solution. Cyclic AMP was freshly prepared as a 10 mm aqueous stock and added to the pipette solution at a concentration of 100 μm. Drugs were delivered to the cells by means of a gravity-driven rapid superfusion system placed in the vicinity (∼200 μm) of the cell selected for an experiment.

RESULTS

Potentiation of Ca²⁺ signals by β-adrenergic stimulation

In most experiments we used isoproterenol as an agonist for the β-adrenergic receptors. The β-adrenergic receptors represent the starting point of the signalling cascade leading to elevated levels of cAMP and subsequent activation of protein kinases (for reviews, see Tsien, 1977; Reuter, 1983). The main advantage of isoproterenol over cAMP is its convenient and rapid extracellular administration, which allowed us to record data under control and test conditions in one and the same cell. In addition, by comparing test data with control values we could check for the presence of the complete signalling cascade. The concentration of isoproterenol used, 1 μm, is about 100-fold the EC₅₀ (Katsube et al. 1996; Calaghan et al. 1998) and was chosen to activate a large fraction of the β-receptors. To assess the effects of isoproterenol on the Ca²⁺ current, the Na⁺ current and Ca²⁺ release from the SR, we applied a specific voltage protocol based on the activation and inactivation curves of each current. L-type Ca²⁺ currents were activated by depolarizing from a holding potential of −40 mV to +10 mV for 200 ms, whereas Na⁺ currents were activated by depolarizing the cell from −90 mV to −40 mV for 50 ms (Weidmann, 1955). In previous experiments we had confirmed that under our experimental conditions we activated maximal currents and that the currents were separable with these voltage protocols. This could be verified from the amplitudes and from the distinct time courses of the two currents.

Figure 1 shows representative results obtained from two cells. In the absence of ryanodine and thapsigargin activation of L-type Ca²⁺ currents produced large Ca²⁺ signals as a result of Ca²⁺ influx and CICR from the SR (Fig. 1A, left panel). The administration of 1 μm isoproterenol to the same cell resulted in a ∼2-fold potentiation of both the L-type Ca²⁺ current and the Ca²⁺ transient (Fig. 1A, right panel). In most cells, activation of the Na⁺ current produced a clear Ca²⁺ signal which was markedly smaller than that produced by activation of the Ca²⁺ current, even before β-adrenergic stimulation (Fig. 1B, left panel; note change in scale). In some cells, no Ca²⁺ signal could be detected at all during a Na⁺ current voltage-clamp protocol. However, after the application of 1 μm isoproterenol the Ca²⁺ signal either increased in amplitude (Fig. 1B, right) or became visible in the cells where it was not detectable in the control solution. The Ca²⁺ signals activated by I_Na had a faster onset than the Ca²⁺ signals activated by I_Ca, typically peaking within 6 ms. Occasionally, the inactivation of I_Na appeared to be slowed down and formed a shoulder in the inactivation phase of the current (Fig. 1, expanded current traces shown in red; see below for discussion). In general, the amplitude of the recorded I_Na was quite variable among different cells under both control conditions and β-adrenergic stimulation.

Figure 1. Simultaneous recording of inward current and Ca²⁺ release.

A, activation of the L-type Ca²⁺ current (200 ms) elicited a substantial Ca²⁺ release signal. Both the current and the Ca²⁺ signal were markedly enhanced by administration of 1 μm isoproterenol in the bathing solution. B, in an analogous experiment, activation of the Na⁺ current (50 ms) in a different cell showed smaller Ca²⁺ signal amplitudes but similar β-adrenergic potentiation. Traces show from top to bottom: voltage protocol (mV), current record, line-scan image, mean Ca²⁺ concentration profile, same current record expanded 10 times (red traces, not temporally aligned to the other signals).

Potentiation of Ca²⁺ influx by β-adrenergic stimulation

As mentioned above, the positive feedback of CICR can vary considerably under different SR Ca²⁺ loads (Han et al. 1994; Spencer & Berlin, 1995; Bassani et al. 1995). The gain can reach levels where the influx of a minute amount of Ca²⁺ could trigger some Ca²⁺ release from the SR by Ca²⁺-induced Ca²⁺ release, which then becomes self-sustaining. Particularly in the presence of high levels of cAMP, the SR Ca²⁺ pump is disinhibited due to phosphorylation of phospholamban, whereby Ca²⁺ uptake into the SR is increased and Ca²⁺ release is enhanced (Tsien & Weingart, 1976; Calaghan et al. 1998). An increased SR Ca²⁺ load presumably renders the SR Ca²⁺ release channels more sensitive to trigger Ca²⁺ (Györke & Györke, 1998). To obtain direct information about the various Ca²⁺ influx pathways and how they might be modulated by β-adrenergic stimulation we used pharmacological tools that allowed us to reduce the complexity inherent in cardiac Ca²⁺ signalling. For this purpose, we suppressed the SR Ca²⁺ release and uptake with 10 μm ryanodine and 0.1 μm thapsigargin. Thus, after the initial control experiments shown above, we recorded Ca²⁺ signals arising exclusively from Ca²⁺ influx via the plasmalemma and we relied on the confocal microscope system for the detection of Ca²⁺ influx signals without any amplification by the SR. Residual SR activity would manifest itself in the form of Ca²⁺ sparks or Ca²⁺ waves in the line-scan images. Figure 2 shows examples of Ca²⁺ influx signals elicited by I_Ca and I_Na in a cell with blocked SR function. The amplitudes of the Ca²⁺ influx transients were severalfold smaller than in cells not treated with ryanodine and thapsigargin, confirming the absence of a Ca²⁺ release component (note the absence of sparks). Nevertheless, a potentiation of the Ca²⁺ influx transient by isoproterenol was still evident in all cells analysed with both voltage-clamp protocols (i.e. during both I_Ca and I_Na).

Figure 2. Simultaneous recording of inward current and Ca²⁺ influx.

SR Ca²⁺ release was blocked by treating the cells with 0.1 μm thapsigargin and 10 μm ryanodine. Na⁺ current was activated in the presence of 1 μm isoproterenol. A, activation of L-type Ca²⁺ current elicited an detectable Ca²⁺ influx signal with reduced amplitude due to the blockade of the SR release (note the absence of sparks). A potentiating effect of isoproterenol on the current amplitude and the Ca²⁺ signal is evident. B, analogous experiment with activation of the Na⁺ current. Note the change in the scale and the small amplitude of the Ca²⁺ signals. Traces are arranged in the same way as in Fig. 1.

Identification of Ca²⁺ influx pathways during β-adrenergic stimulation

While an increase of the Ca²⁺ influx signal via L-type Ca²⁺ current during β-adrenergic stimulation can easily be explained by the known effect of L-type Ca²⁺ channel phosphorylation (Kameyama et al. 1985; Katsube et al. 1996), the reason for the changes observed during the I_Na activation protocol was not clear. Having a reproducible experimental protocol available to show the significant potentiating effect of β-stimulation on Ca²⁺ influx, we tried to identify the Ca²⁺ entry pathway that, first, was enhanced by β-stimulation and, second, was manifest during the voltage-clamp protocol designed to activate I_Na only. Considering the known Ca²⁺ entry pathways present in heart cells (Bers, 1991; Boyett et al. 1996), we used pharmacological tools to block L- and T-type Ca²⁺ channels and also the Na⁺-Ca²⁺ exchanger.

Initially, we applied various concentrations of Ni²⁺, an ion that is known to block T-type Ca²⁺ currents at micromolar concentrations and L-type Ca²⁺ currents (and the Na⁺-Ca²⁺ exchange current, I_Na-Ca) at millimolar concentrations (McDonald et al. 1994; Lee et al. 1999). The effect of Ni²⁺ on the Ca²⁺ influx signal observed during I_Na in our experiments was strongly dose dependent (see Fig. 3). The Ca²⁺ influx amplitude was reduced to 80.9 ± 19.4% of the control by 50 μm Ni²⁺ (n = 7). With 100 μm Ni²⁺ in the extracellular solution the Ca²⁺ influx signal amplitude was 56.4 ± 28.2% of the control (n = 8). Interestingly, the Ca²⁺ influx signal elicited by I_Na activation was only reduced to undetectable levels by a high concentration of Ni²⁺ (5 mm; n = 15 cells). Figure 3B shows the blocking effect of 5 mm Ni²⁺ on a substantial Ca²⁺ influx signal. Note that in the presence of 5 mm Ni²⁺ the activation curve for I_Na was shifted by +20 mV and thus we had to depolarize the cell to −20 mV to elicit a comparable current amplitude (Fig. 3B, right panel, top trace). Thus, it appeared that millimolar concentrations of Ni²⁺ were required to reliably suppress Ca²⁺ influx during the voltage-clamp protocol designed to elicit I_Na.

Figure 3. Pharmacological identification of the Ca²⁺ influx pathway.

All cells were treated with 0.1 μm thapsigargin and 10 μm ryanodine. Na⁺ current was activated in the presence of 1 μm isoproterenol. A, the administration of 100 μm Ni²⁺ failed to block the Ca²⁺ influx signal. However, a slight reduction of the Ca²⁺ signal amplitude is noticeable. B, the administration of 5 mm Ni²⁺ blocked the Ca²⁺ signal completely. C, the Ca²⁺ influx signal was eliminated by 50 μm Cd²⁺. Traces are arranged in the same way as in Fig. 1. D, normalized Ca²⁺ influx amplitudes versus different concentrations of Ni²⁺ and of 50 μm Cd²⁺. Data are plotted as means ±s.d. The inset shows a Hill function fitted to the data points of the Ni²⁺ block. The calculated IC₅₀ was 114 ± 33.2 μm (95% confidence); the Hill coefficient was 1.71. For comparison, the Hill function of the Ni²⁺ dose-response curves of the L-type current in guinea-pig ventricular myocytes (blue trace, taken from Hobai et al. 1998) and in the human heart T-type α1H subunit (T-type, red trace, taken from Lee et al. 1999) have been added.

For a quantitative analysis the averaged and normalized Ca²⁺ influx amplitudes were plotted against the Ni²⁺ concentration and fitted with a Hill function (Fig. 3D, inset). The function revealed an IC₅₀ of 114 ± 33.2 μm (confidence interval 95%) and a Hill coefficient of about 1.71. These values are approximately in the same range as the values obtained for Ni²⁺-induced block of the L-type Ca²⁺ current in guinea-pig ventricular myocytes stimulated with 100 μm cAMP (IC₅₀ of about 510 μm and a Hill coefficient of about 1.48: Hobai et al. 1998; see Fig. 3D, inset, blue line). For comparison, the Hill function of the human heart T-type α-subunit (α1H) is also plotted in the same graph (Fig. 3D, inset, red line). The dose-response data for this line were obtained from recombinant α1H subunits expressed in human embryonic kidney HEK-293 cells or Xenopus oocytes (Lee et al. 1999). In contrast to the L-type Ca²⁺ current, these data yielded an IC₅₀ of about 12 μm (HEK-293) or 5.7 μm (oocytes), with Hill coefficients of about 0.77. It is worth noting that we measured fluorescence changes caused by Ca²⁺ influx and not a membrane current. Thus, a direct comparison is not straightforward. Nevertheless, the IC₅₀ for Ni²⁺-induced block of Ca²⁺ influx that we obtained is comparable to that for L-type Ca²⁺ channels but 10- to 20-fold higher than that for the T-type Ca²⁺ channels.

The next tool we used was Cd²⁺ which should noticeably block the L-type Ca²⁺ current at 50 μm while minimally affecting the T-type Ca²⁺ current or the Na⁺-Ca²⁺ exchanger (Hobai et al. 1997a). Figure 3C shows the blocking effect of 50 μm Cd²⁺ on the Ca²⁺ influx signal. This low dose of Cd²⁺ reduced the Ca²⁺ influx signal observed during I_Na activation to an average of 17.9 ± 15.1% of control values (n = 10; Fig. 3D). In addition, the shoulder present during I_Na inactivation was removed by this inhibitor, suggesting that it was carried by a Cd²⁺-sensitive current, most likely the L-type Ca²⁺ current. The clear-cut effect of Cd²⁺, suggesting Ca²⁺ influx via L-type Ca²⁺ channels, does not imply that other sources of Ca²⁺ cannot contribute to a Ca²⁺ signal under different conditions, particularly when Na⁺-Ca²⁺ exchange is not minimized (Leblanc & Hume, 1990; Lipp & Niggli, 1994; Niggli & Lipp, 1996) or when the amplification by SR Ca²⁺ release is not suppressed (Santana et al. 1998).

We were also interested in examining whether the same results could be reproduced by elevating intracellular cAMP directly via the patch pipette instead of applying isoproterenol. In these experiments the concentration of cAMP in the pipette solution was 100 μm to provide maximal stimulation of cAMP-dependent protein kinase (Kameyama et al. 1985). In addition, the high concentration in the pipette was necessary to counteract the reduction in cAMP concentration caused by phosphodiesterase activity (Kameyama et al. 1985). To maximize putative Ca²⁺ entry via phosphorylated sodium channels (Santana et al. 1998), we also increased extracellular Ca²⁺ to 10 mm in some of these experiments. The results obtained in the presence of intracellular cAMP were essentially the same as those with isoproterenol added to the bath. Figure 4A shows a representative recording where the Ca²⁺ influx signal, in the presence of 100 μm cAMP in the pipette, was completely abolished by 5 mm Ni²⁺.

Figure 4. Ca²⁺ influx under elevated cAMP and blockade by TTX.

A, with 100 μm cAMP in the pipette filling solution, activation of I_Na elicited a Ca²⁺ influx signal which was blocked by 5 mm Ni²⁺. Note the increase in fluorescence in the right line-scan, which is due to elevated extracellular Ca²⁺ (10 mm). B, the Na⁺ current and the Ca²⁺ influx signal elicited in the presence of 1 μm isoproterenol were completely blocked by 20 μm TTX.

To provide evidence that the Ca²⁺ influx signal we observed was related to I_Na and not to the voltage change or to shifts of I_Ca activation and gating we blocked the Na⁺ channels with 10–20 μm TTX. Figure 4B shows that as I_Na disappeared after application of 20 μm TTX the Ca²⁺ influx signal also vanished, suggesting that activation of I_Na is indeed required to initiate the observed Ca²⁺ influx.

Ca²⁺ influx during trains of I_Na

Inspired by a recent report on Ca²⁺ influx via sodium channels (Santana et al. 1998), we adopted a strategy designed to favour Ca²⁺ entry through this proposed pathway using a vigorous voltage-clamp protocol. To obtain better control over the voltage we used lower resistance electrodes (∼1 MΩ) and attempted to elicit larger Na⁺ currents. In addition, we applied trains of I_Na in order to accumulate the presumably small Ca²⁺ influx signal which we otherwise might have missed with our detection system (see below).

Figure 5A documents an experiment where we applied a train of 50 depolarizing voltage pulses at 33 Hz in the presence of 1 μm isoproterenol. Care was taken to space the pulses appropriately to allow I_Na to recover from inactivation. This protocol led to a substantial current amplitude (Fig. 5A, expanded current shown in red). The Ca²⁺ influx signal increased progressively during the train reaching a plateau at the end (Fig. 5A, left). However, the Ca²⁺ influx signal was again completely abolished in the presence of 5 mm Ni²⁺ (Fig. 5A, right). Figure 5B shows a similar experiment that was performed with 100 μm cAMP in the pipette instead of administration of isoproterenol and with a train of 100 depolarizing pulses. Once more, the Ca²⁺ influx was abolished by 5 mm Ni²⁺. To test whether the proposed Ca²⁺ influx through TTX-sensitive Na⁺ channels was sensitive to temperature under these conditions, like, for instance, the proposed voltage-sensitive Ca²⁺ release (Hobai et al. 1997b; Ferrier et al. 1998), the same experiments were repeated at 37°C (n = 6). The results were identical to those observed at room temperature (data not shown).

Figure 5. Forced Ca²⁺ influx under extreme conditions.

SR release was blocked and the Na⁺ current was activated repetitively in the presence of 1 μm isoproterenol. A, a Ca²⁺ influx signal could be elicited by a train of 50 depolarizing pulses at 33 Hz. The Ca²⁺ influx signal was eliminated by 5 mm Ni²⁺. B, an analogous experiment was performed in the absence of isoproterenol but with 100 μm cAMP in the pipette filling solution. Sodium currents were activated repetitively 100 times at 33 Hz.

Sometimes a small and slowly progressive elevation of the resting Ca²⁺ concentration was detectable (see, for example, Figs 3B and 5B). However, this rise did not coincide temporally with the voltage-clamp pulses. This monotonic rise in [Ca²⁺]_i may be attributed to block of the Na⁺-Ca²⁺ exchanger by 5 mm Ni²⁺ which prevented any extrusion of Ca²⁺ entering by slow leakage through the plasmalemma.

Estimation of the detection limit for Ca²⁺ entry

As we relied on confocal Ca²⁺ fluorescence measurements to detect Ca²⁺ influx across the sarcolemma, we were interested in determining the detection limit of our system. This would enable us, based on theoretical assumptions, to estimate the minimal amount of Ca²⁺ entering the cell that can be detected. In addition, we could indirectly calculate the maximal permeability of TTX-sensitive Na⁺ channels to Ca²⁺ ions that would be consistent with our results. Since we were dealing with changes in Ca²⁺ concentration, the relevant parameter determining the detection limit was the signal-to-noise ratio of the fluorescence record. Therefore, we determined the noise in a typical line-scan image recorded at the beginning of an experiment. We assumed a resting Ca²⁺ concentration of 100 nm and a K_d of fluo-3 for Ca²⁺ of 500 nm, values on which we based our calibrations for all data presented. Statistical analysis of different mean Ca²⁺ concentration profiles yielded a mean noise level of 1.78 ± 0.24 nm (r.m.s. ±s.d., n = 10).

In a subsequent computer simulation, we added step increases of [Ca²⁺] to a Ca²⁺ concentration profile recorded from one particular cell and were thus mimicking the Ca²⁺ influx signal observed during I_Na activation (Fig. 6). The amplitudes of the steps corresponded to multiples of the standard deviation of the noise derived from the same record. We identified the [Ca²⁺] step which we could detect by eye and defined our detection limit to be 3 standard deviations of the noise. This turned out to be ∼8 nm in the fairly noisy record shown in Fig. 6, or ∼5.34 nm on average. It is well known that each calibration procedure relies on several assumptions. The K_d of fluo-3 in the cytoplasmic environment can differ considerably from the in vitro value (Harkins et al. 1993). In addition, in the presence of a low intracellular Na⁺ concentration the resting Ca²⁺ concentration could be lower than 100 nm in a quiescent guinea-pig myocyte, because a low [Na⁺]_i increases the driving force of the Na⁺-Ca²⁺ exchanger for Ca²⁺ removal. Hovever, calibrating with a K_d of 1000 nm instead of 500 nm would not change the Ca²⁺ concentrations appreciably (they would increase by factor of ∼1.2) whereas assuming a resting [Ca²⁺] of 50 nm would improve our detection limit by a factor of ∼2.4 (data not shown).

Figure 6. Estimation of the detection limit and relationship to a calculated permeability ratio.

A, in a cell with blocked SR CICR an L-type Ca²⁺ current and the resulting Ca²⁺ influx signal were recorded simultaneously under control conditions. B, the Cd²⁺-sensitive difference current was integrated and plotted against the rising phase of the Ca²⁺ signal. The fitted line relates the observed Ca²⁺ influx fluorescence amplitude to the corresponding Ca²⁺ current in this given cell and corresponds to a cytosolic buffer capacity of about 28 (note that the SR was blocked). C, an mean Ca²⁺ concentration profile was plotted from a line-scan image and the standard deviation of the noise was calculated. After 500 ms, a step increase in Ca²⁺ was simulated by adding a concentration jump corresponding to a multiple of the s.d. The first step that is clearly visible coincides with +3 s.d. and, for this particular cell, corresponds to a concentration change of ∼8 nm[Ca²⁺]_i. Taking the relationship illustrated in B we can estimate our mean detection limit for Ca²⁺ influx to be around 2 pC (dashed horizontal line in D). D, using the Goldmann-Hodgkin-Katz equation, the permeability ratio P_Ca/P_Na was calculated for a 50 nA (□, ▪) or 10 nA Na⁺ current (○, •) and for 10 mm (•, ▪) or 1 mm[Ca²⁺]_o (○, □), respectively.

Finally, we could relate the observed Ca²⁺ concentration change to Ca²⁺ influx using the relationship between the measured and numerically integrated Ca²⁺ current (Cd²⁺-sensitive difference) and the corresponding Ca²⁺ influx fluorescence signal (Fig. 6A). The linear relationship obtained by plotting the integral of the Ca²⁺ current versus the Ca²⁺ concentration allowed us to determine the minimal Ca²⁺ influx needed to produce a detectable Ca²⁺ signal, which on average was ∼2 pC (Fig. 6B). Based on this amount of charge it is possible to calculate the maximal permeability of the Na⁺ channel to Ca²⁺ which would be consistent with our data. The Goldmann-Hodgkin-Katz current equation (Hille, 1992) of the form:

was used to calculate the absolute permeability for Na⁺ at the ionic concentrations, membrane potential and measured current obtained under our experimental conditions. We introduced the ionic concentrations for Na⁺ ([S]_i = 1 mm, [S]_o = 140 mm), the charge (z_S =+1), and the driving force (E = E_m–E_S =−164 mV for these Na⁺ concentrations) and calculated the absolute permeability P_S for assumed current amplitudes. F stands for the Faraday constant, R for the universal gas constant and T for the absolute temperature. The index ‘S’ refers to the ionic species in question. By introducing assumed values for the absolute permeability of Ca²⁺ (P_S) into the same equation and taking the variables [S]_i, [S]_o, z_S and E for Ca²⁺ we derived a fractional current amplitude (I_S) carried by Ca²⁺ ions.

This current was numerically integrated to obtain a theoretical Ca²⁺ influx. The Ca²⁺ permeability was expressed as permeability ratio, P_Ca/P_Na, and plotted against the expected Ca²⁺ influx (Fig. 6D). The horizontal dashed line separates the detectable Ca²⁺ influx signals (above) from the undetectable signals (below). It represents the mean detection limit calculated from the mean noise (1.78 nm) using a K_d of 500 nm and a resting [Ca²⁺] of 100 nm. Whole-cell patch-clamp Na⁺ current measurements usually underestimate the real current amplitude because of the access resistance. For large Na⁺ currents (50 nA: Brown et al. 1981; Makielski et al. 1987) and high extracellular Ca²⁺ (10 mm) we should detect signals elicited by single Na⁺ currents corresponding to a permeability ratio of about 0.04:1. Furthermore, the detection for Ca²⁺ influx during trains consisting of 50 or 100 Na⁺ currents would be considerably more sensitive. Using the same noise analysis procedure we estimated that we could detect a slow elevation of Ca²⁺ concentration during a current train if the deviation from a control trace at the end of the train corresponded to ∼8 nm (or 3 s.d. of the noise). Therefore, the detection limit in a train would correspond to about 0.16 nm per single Na⁺ current or to a Ca²⁺/Na⁺ permeability ratio of 0.0012: 1.

DISCUSSION

The Ca²⁺ influx signals mainly result from activation of L-type Ca²⁺ channels

During voltage-clamp protocols designed to activate L-type Ca²⁺ currents our experimental results confirmed the expected changes in Ca²⁺ influx after β-adrenergic stimulation. Activation of β-receptors is known to induce a signalling cascade which ultimately leads to phosphorylation of various Ca²⁺ signalling proteins and membrane channels by protein kinase A (PKA), including the L-type Ca²⁺ channels (Kameyama et al. 1985). After PKA phosphorylation, L-type Ca²⁺ channels exhibit a higher open probability (Tsien et al. 1986). In contrast, the mechanism underlying the increase in the Ca²⁺ influx signals during voltage-clamp protocols tailored to specifically activate Na⁺ currents was less clear. Before reaching a conclusion, several experimental difficulties and possible pitfalls need to be considered. It is well known that Na⁺ currents have very large amplitudes in cardiac myocytes (Brown et al. 1981; Makielski et al. 1987). Consequently, it is notoriously difficult, if not impossible, to control the clamp voltage during the activation of the current with conventional amplifiers and patch-clamp recording pipettes (Makielski et al. 1987; Hüser et al. 1996). As a consequence, during I_Na activation the membrane potential briefly escapes away from the imposed clamp potential towards the Na⁺ reversal potential and hence towards the voltage range for the activation of L-type Ca²⁺ currents. In principle, L-type Ca²⁺ channels could be activated and carry Ca²⁺ influx during such voltage escape. Since the L-type Ca²⁺ channels are phosphorylated by cAMP-dependent protein kinase after β-stimulation, the Ca²⁺ influx signal generated during voltage escape may become larger after the application of isoproterenol. In addition, it has to be considered that isoproterenol not only increases the amplitude of I_Ca but also shifts its activation threshold towards more negative potentials (Katsube et al. 1996). This may lead to the activation of some L-type Ca²⁺ channels during the I_Na protocol (at −40 mV) after β-stimulation. The proposed voltage-activated Ca²⁺ release mechanism has also been reported to be facilitated by β-stimulation (Hobai et al. 1997b; Ferrier et al. 1998). However, in the present study any Ca²⁺ release from the SR was eliminated by ryanodine and thapsigargin and a contribution by SR Ca²⁺ release can thus be excluded. Other Ca²⁺ entry pathways which may possibly be involved are discussed below.

Considering all these pitfalls and after analysing the voltage sensitivity and the pharmacological profile of the Ca²⁺ influx signal observed in this study, we reached the conclusion that the major fraction of these Ca²⁺ influx signals was indeed generated by spurious activation of L-type Ca²⁺ channels during escape of the membrane potential from voltage clamp. β-Stimulation of these channels then led to a more pronounced manifestation of this effect, presumably by increasing the open probability and by shifting the activation of the L-type Ca²⁺ channels. Several lines of evidence suggested this conclusion. First, the analysis of the current traces (I_Na) revealed an apparent slowing of inactivation after exposure to isoproterenol. In some cases a distinct shoulder appeared during the late phase of the inactivation, indicating that another current had been activated (Fig. 3C). Interestingly, this shoulder, as well as the Ca²⁺ influx signal, turned out to be Cd²⁺ sensitive. Second, the sensitivity of the Ca²⁺ influx signal to 5 mm Ni²⁺ also suggested the participation of either the L- or T-type Ca²⁺ current (or both). However, the small or absent effect of low concentrations of Ni²⁺ (50–100 μm) indicates that T-type Ca²⁺ channels contribute very little, if at all, since they are known to exhibit a much higher Ni²⁺ sensitivity (Sipido et al. 1998; but see also Lee et al. 1999). In this case, the relatively small effect of low Ni²⁺ concentrations on the influx of Ca²⁺ could be explained by the known overlap of the dose-response curves for T- and L-type Ca²⁺ channels (i.e. even low Ni²⁺ concentrations have a minor blocking effect on L-type Ca²⁺ current). The almost complete block by 50 μm Cd²⁺, which is an indicator of the involvement of the L-type Ca²⁺ current (Hobai et al. 1997a), provides strong support for the view that most of the Ca²⁺ influx during the I_Na protocol was carried by L-type Ca²⁺ channels. As mentioned above, the depolarisation to −40 mV during the I_Na activation protocol could activate a small fraction of the L-type Ca²⁺ channels, particularly following the leftward shift in activation after β-stimulation. However, the disappearance of the Ca²⁺ influx signal upon administration of 20 μm TTX clearly suggests that the depolarization to −40 mV did not per se induce any detectable Ca²⁺ influx, not even after β-stimulation. This result can be explained if block of I_Na with TTX prevents voltage escape. The following experimental observation provides further support for this view. When the series resistance compensation of the voltage-clamp amplifier was impaired on purpose, the Ca²⁺ influx signals during I_Na activation clearly increased in amplitude (not shown). Thus, precise control of the membrane potential seems to prevent the activation of I_Ca.

Involvement of T-type Ca²⁺ channels

We expected that the I_Na activation protocol (depolarization from −90 mV to −40 mV) would also activate T-type Ca²⁺ currents which could have led to a noticeable Ca²⁺ influx. However, the weak sensitivity of the Ca²⁺ signal to 50 μm Ni²⁺, a concentration which was shown to be discriminative for cardiac T-type Ca²⁺ channels (Sipido et al. 1998; Lee et al. 1999), suggests that this pathway only contributed a small fraction of the detectable Ca²⁺ influx. It is also worth mentioning that the T-type Ca²⁺ current has been reported to be insensitive to β-stimulation in the guinea-pig heart (Tytgat et al. 1988; Balke et al. 1993; but see Vassort & Alvarez, 1994). Based on these findings and on the Cd²⁺ sensitivity discussed above, we believe that a significant participation of the T-type Ca²⁺ channels in the generation of a Ca²⁺ influx signal sensitive to β-stimulation is unlikely.

Contribution of the Na⁺-Ca²⁺ exchanger to the Ca²⁺ influx signal

In the present study, we were concerned with Ca²⁺ influx via voltage-dependent membrane channels and therefore tried to minimize the reverse mode Na⁺-Ca²⁺ exchange by omitting Na⁺ from the intracellular solution. It is clear, however, that this procedure cannot completely prevent some accumulation of Na⁺ close to the membrane during I_Na activation, which in turn could elicit a small Ca²⁺ influx via Na⁺-Ca²⁺ exchange (Lederer et al. 1990; Lipp & Niggli, 1994; Niggli & Lipp, 1996). In our experiments, this pathway would also be blocked by 5 mm Ni²⁺. Therefore this pathway is indistinguishable from spurious activation of I_Ca based on this observation alone. However, it has to be pointed out that the Ca²⁺ influx signal observed in the absence of Na⁺ in the pipette filling solution was suppressed by the administration of 50 μm Cd²⁺ or the L-type Ca²⁺ channel blocker nifedipine (10 μm; not shown). Cd²⁺ is known to block the Na⁺-Ca²⁺ exchange only at much higher concentrations (IC₅₀ 320 μm; Hobai et al. 1997a). Taken together, these findings strongly suggest that Ca²⁺ influx via the Na⁺-Ca²⁺ exchanger only played a small role when Na⁺ was omitted from the patch pipette solution.

Other sources for Ca²⁺ influx

Potential additional pathways for Ca²⁺ influx via Na⁺ channels have been reported recently in guinea-pig (Cole et al. 1997) and rat ventricular myocytes (Aggarwal et al. 1997). These voltage-activated currents were blocked by TTX and were insensitive to 50 μm Ni²⁺. However, no evidence was provided that Ca²⁺ influx really occurred, nor that Ca²⁺ release could be triggered. Moreover, the current presumably carried by Ca²⁺ ions in these experiments was only observed when Na⁺ was completely absent from the extracellular solution because ${\text{N}}_{\text{o}}^{+}$ blocked the current at micromolar concentrations. Since we used physiological Na⁺ concentrations in all experiments, the participation of these currents in generating a Ca²⁺ influx can be excluded.

The possibility of Ca²⁺ permeation through TTX-sensitive Na⁺ channels, even with physiological extracellular Na⁺ concentrations, was discussed several years ago (Sorbera & Morad, 1990; Johnson & Lemieux, 1991; Hume et al. 1991; Lederer et al. 1990). Recently, experimental support for this controversial pathway has been reported by Santana et al. (1998). They observed SR Ca²⁺ release triggered by Ca²⁺ apparently entering through TTX-sensitive Na⁺ channels, but only after β-adrenergic stimulation. In addition, the permeability ratio P_Ca/P_Na was shifted from almost 0: 1 to 1.25: 1. This newly described peculiarity of the Na⁺ channels was termed ‘slip-mode Ca²⁺ conductance’. A salient feature of this proposed mode of conductance is the insensitivity of the Ca²⁺ release to 5 mm Ni²⁺.

Since this TTX-sensitive β-stimulation-dependent Ca²⁺ influx signal had several characteristics resembling the Ca²⁺ influx signals observed in the present study, we also tried to detect Ca²⁺ influx via slip-mode conductance, taking advantage of the insensitivity of the influx to 5 mm Ni²⁺. This allowed us to supress all other known Ca²⁺ entry pathways with a single unspecific blocker. During such experiments, we anticipated that the Ca²⁺ fluorescence signal would be small since we had suppressed any amplification by CICR. The estimate of the detection limit of our confocal fluorescence measurement suggested that we should be able to detect Ca²⁺ influx through slip-mode conductance during a single I_Na as soon as the Ca²⁺/Na⁺ permeability ratio exceeded 0.04: 1. This estimate of the detection limit is very conservative because the real Na⁺ current is presumably larger than the measured current. Indeed, experiments carried out with low access resistance electrode systems had suggested peak currents of up to 140 nA in rat ventricular myocytes (Brown et al. 1981; Makielski et al. 1987). For trains of 100 I_Na activations, which rapidly accumulate the Ca²⁺ influx, the detection would be correspondingly more sensitive.

In conclusion, we believe that when Ca²⁺ entry via Na⁺-Ca²⁺ exchange is minimized, and in the absence of blockers for L-type Ca²⁺ channels, escape of the voltage clamp, leading to spurious activation of L-type Ca²⁺ channels, is the most important pathway for Ca²⁺ influx during I_Na activation protocols. Ca²⁺ entry via the slip-mode conductance of the Na⁺ channels did not participate in the generation of measureable Ca²⁺ influx under our experimental conditions. In principle, a much smaller Ca²⁺ influx than ∼8 nm into the diadic space could still be sufficient to trigger Ca²⁺ release from the SR, but such a small slip-mode conductance would then not be consistent with the Ca²⁺/Na⁺ permeability ratio determined from shifts of the reversal potential (Santana et al. 1998).