Location of the initiation site of calcium transients and sparks in rabbit heart Purkinje cells

Abstract

1. The distribution and localization of Ca²⁺ transients and Ca²⁺ sparks in isolated adult rabbit Purkinje cells were examined using confocal microscopy and the Ca²⁺ indicator fluo-3.

2. When cells were field stimulated in 2.0 mm Ca²⁺ buffer, a transverse confocal line scan (500 Hz) showed that the fluorescence intensity was greatest at the cell periphery during the onset of the Ca²⁺ transient ([Ca²⁺]_i). In contrast, the [Ca²⁺]_i of ventricular cells showed a more uniform pattern of activation across the cell. Staining with di-8-ANEPPS revealed that Purkinje cells lack t-tubules, whereas ventricular cells have an extensive t-tubular system.

3. When we superfused both cell types with a buffer containing 5 mm Ca²⁺-1 μm isoproterenol (isoprenaline) they produced Ca²⁺ sparks spontaneously. Ca²⁺ sparks occurred only at the periphery of Purkinje cells but occurred throughout ventricular cells. Sparks in both cell types could be completely abolished by addition of the SR inhibitor thapsigargin (500 nm). Brief exposure to nifedipine (10 μm) did not reduce the number of spontaneous sparks.

4. Immunofluorescence staining of Purkinje cells with anti-ryanodine antibody revealed that ryanodine receptors (RyRs) are present at both peripheral and central locations.

5. Computer simulations of experiments in which the calcium transient was evoked by voltage clamp depolarizations suggested that the increase in calcium observed in the centre of the cell could be explained by simple buffered diffusion of calcium. These computations suggested that the RyRs deep within the cell do not contribute significantly to the calcium transient.

6. These results provide the first detailed, spatially resolved data describing Ca²⁺ transients and Ca²⁺ sparks in rabbit cardiac Purkinje cells. Both types of events are initiated only at subsarcolemmal SR Ca²⁺ release sites suggesting that in Purkinje cells, Ca²⁺ sparks only originate where the sarcolemma and sarcoplasmic reticulum form junctions. The role of the centrally located RyRs remains unclear. It is possible that because of the lack of t-tubules these RyRs do not experience a sufficiently large Ca²⁺ trigger during excitation-contraction (E-C) coupling to become active.

Spontaneous local increases in intracellular [Ca²⁺]_i (calcium sparks) have been identified in cardiac, smooth and skeletal muscle (reviewed by Cheng et al. 1996b). A number of authors have suggested that Ca²⁺ transients are due the spatio-temporal summation of many Ca²⁺ sparks (Cannell et al. 1994; Cheng et al. 1995; Tanaka et al. 1998). These Ca²⁺ sparks are the result of release of Ca²⁺ from the SR and are believed to be the elementary events underlying E-C coupling in cardiac muscle (Cheng et al. 1993; Bridge et al. 1999). Shacklock et al. (1995) have shown that location of sparks coincides with the location of t-tubules in rat ventricular myocytes. Since L-type Ca²⁺ channels and SR Ca²⁺ release are co-localized at the t-tubule dyad junction, it has been suggested that a local accumulation of [Ca²⁺]_i near one or a cluster of SR Ca²⁺ release channels gates their opening (Cannell et al. 1994; Cheng et al. 1996a). However, it is unclear whether Ca²⁺ channels or junctions with the sarcolemma are absolutely necessary for the spontaneous occurrence of sparks.

In contrast to the information available on sparks and Ca²⁺ transients in ventricular muscle, relatively little is known about the spatio-temporal characteristics of calcium signalling in other cardiac cell types. In canine cardiac Purkinje fibres, measurements of the Ca²⁺ transient by aequorin demonstrated that it was composed of two components: a rapid initial increase in [Ca²⁺] followed by a slower secondary rise (Wier & Isenberg, 1982). Since ventricular cells have a monotonic increase in [Ca²⁺] during the calcium transient, this difference suggests that mechanisms underlying the Ca²⁺ transient may differ between the two cell types. Using laser scanning confocal microscopy we have obtained new data describing the localization and distribution of Ca²⁺ sparks and Ca²⁺ transients in rabbit heart Purkinje cells. Our results suggest that Ca²⁺ sparks only occur at the periphery of the cell in association with junctions between the SR and sarcolemma. Antibody staining revealed that ryanodine receptors (RyRs) are arranged in regularly spaced bands (corresponding approximately to the Z-line) throughout the cell. However, we only observed sparks at the cell periphery and computer simulations suggested that little, if any, Ca²⁺ release takes place from the central RyRs. This suggests that sparks are only produced by the gating of RyRs located at junctions between the SR and sarcolemma in Purkinje cells. It follows that Ca²⁺ transients in Purkinje fibres normally originate at the cell periphery by the recruitment of peripheral sparks and simple diffusion of Ca²⁺ leads to the Ca²⁺ transient deeper inside the cell. Some preliminary results have been presented as an abstract (Spitzer et al. 1997; Cordeiro et al. 1999).

METHODS

Isolation of Purkinje cells

Purkinje cells were prepared from rabbit hearts using techniques described previously (Scamps & Carmeliet, 1989) with some modifications (Cordeiro et al. 1998). Male New Zealand white rabbits (Ellis and Pritchard, America Fork, UT, USA) were housed according to the Guide for the Care and Use of Laboratory Animals (NIH publication No. 85-23, revised 1985). Animals (2.0-3.0 kg) were anaesthetized with sodium pentobarbital (50 mg kg⁻¹) administered through the ear vein. The hearts were rapidly removed and perfused retrogradely through the aorta with nominally Ca²⁺-free Tyrode solution of the following composition (mm): NaCl 126, KCl 5.4, MgCl₂ 5.0, NaH₂PO₄ 1.0, glucose 22, Hepes 24, taurine 20, creatine 5, sodium pyruvate 5. This solution was equilibrated with 100% O₂ and the pH was adjusted to 7.4 with NaOH. After 6-8 min of this nominally Ca²⁺-free perfusion, the heart was perfused for approximately 18-20 min with the solution described above supplemented with collagenase (273 U ml⁻¹, Worthington type II), protease (5.2 U ml⁻¹, Sigma type XIV) and 0.1 mm CaCl₂. Following this enzyme digestion, the heart was perfused with enzyme-free Tyrode solution containing 0.1 mm Ca²⁺ for about 5 min. The heart was then placed in a dissecting dish containing 0.1 mm Ca²⁺ Tyrode solution. Purkinje fibres from both ventricles were dissected out and placed in a small dish containing fresh enzyme solution (described above). Dissociation of individual cells from the isolated fibres was aided by agitation of the enzyme solution with a stream of 100% O₂. The temperature was maintained at 36°C throughout. Periodically, enzyme solution containing Purkinje cells in suspension was removed and added to a modified storage solution (described below). Myocytes were stored for up to 6 h in storage solution and were used on the day of isolation. Fresh enzyme solution was added to the undigested Purkinje fibres to maintain a volume of approximately 2 ml. Digestion of the Purkinje fibres into populations of individual myocytes required between 15 and 60 min under these conditions.

Isolation of ventricular cells

Ventricular cells were also isolated using the above methods. After the Purkinje strands had been dissected from both ventricles, the left ventricle was removed, minced and gently swirled in 0.1 mm Ca²⁺ Tyrode solution. The cell suspension containing ventricular cells was then placed in storage solution (described below) and kept at room temperature (21-22°C) until use.

Fluorescence imaging

The acetoxymethylester (AM) form of fluo-3 was used in this study. A 100 μm stock solution of fluo-3 AM was prepared by adding 1 mg of fluo-3 AM (dissolved in 0.86 ml DMSO) to 9 ml of fetal bovine serum (containing 225 μl of 20% F-127 pluronic in DMSO). The fluo-3 AM stock solution was divided into 200 μl aliquots and stored at -70° C. When needed, 1.5 ml of cell suspension containing either Purkinje or ventricular cells were loaded with fluo-3 AM for 30 min as described previously (Cheng et al. 1993).

All experiments were performed with a Bio-Rad 1024 laser-scanning confocal microscope attached to an inverted microscope (Nikon Diaphot). Fluorescence signals were collected with a x 100 oil-immersion objective lens (NA = 1.3). Fluo-3-loaded myocytes were placed in a perfusion chamber mounted on the stage of the microscope. The perfusion chamber could be rotated, allowing orientation of the myocytes either perpendicular or parallel to the scan line direction. Cells loaded with dye were excited at 488 nm using a krypton-argon laser and fluorescence emission detected via a 520 nm band-pass filter and photomultiplier tube. All confocal images were acquired with the Laser Sharp acquisition software program from Bio-Rad and recorded on a microcomputer and Jaz disk (Iomega) for later analysis. Images acquired with Laser Sharp software were analysed with custom programs written in IDL (Research Systems, Boulder, CO, USA).

The myocytes were repetitively field stimulated (4 ms duration via a pair of platinum wires) at 0.5 Hz to ensure a constant SR load. All experiments were performed at 36°C.

Staining with di-8-ANEPPS

In some experiments, myocytes were stained with the membrane-selective fluorescent dye di-8-ANEPPS. Cells were exposed to 5 μm of the dye (dissolved in DMSO + 20% pluronic) for 5 min. After washout, cells were excited at 488 nm and the emitted signal was recorded at 520 nm.

Immunofluorescence labelling

Freshly isolated myocytes from Purkinje strands and ventricular muscle were fixed by adding a 1 ml aliquot of myocytes in storage solution to 1 ml of 4% paraformaldehyde solution. The myocytes were fixed onto a glass slide, were washed briefly with phosphate-buffered saline (PBS) and permeabilized with Triton-X 100 (0.5% solution in PBS) for 15 min at room temperature. After permeabilization, the fixed myocytes were washed with PBS and blocked with 10% goat serum for 30 min. Myocytes were then incubated with monoclonal anti-ryanodine receptor antibody (mouse anti-RyR2 antibody; Affinity Bioreagents, Inc.). The antibody was diluted in 10% goat serum at a concentration of 0.5-1.0 μg ml⁻¹ and myocytes were incubated with the antibody overnight at 4°C. The primary antibody was then rinsed with PBS and the secondary antibody was then added to the fixed myocytes for 4 h. As a control, the secondary antibody alone was added to fixed myocytes. The secondary antibody was goat anti-mouse Texas Red dissolved in PBS.

Ryanodine receptor antibody conjugated to Texas Red was visualized by confocal microscopy using 568 nm excitation wavelength with emission detected at 598 nm.

Solutions

All solutions were made with Milli-Q grade water. The modified storage solution had the following composition (mm): potassium glutamate 100, potassium aspartate 10, KCl 25, KH₂PO₄ 10, MgSO₄ 2, taurine 20, creatine 5, EGTA 0.5, glucose 20, Hepes 5, BSA 0.2%. Myocytes were superfused with a Hepes buffer of the following composition (mm): NaCl 126, KCl 5.4, MgCl₂ 1.0, CaCl₂ 2.0, Hepes 24, glucose 11. pH was adjusted to 7.4 with NaOH. The anion exchange blocker probenecid (0.3 mm) was also added to the Hepes buffer to slow the loss of fluo-3 dye from the cells. In preliminary experiments, sparks were not always observable in the Hepes buffer containing 2 mm Ca²⁺. Therefore, the Ca²⁺ concentration in the Hepes buffer was elevated to 5 mm and 1 μm isoproterenol was added and in these conditions frequent sparks were detected throughout the experiments.

Electrophysiological recordings

In some experiments, voltage clamp recordings of the L-type Ca²⁺ current were made from Purkinje cells. Patch pipettes were fabricated from borosilicate glass capillaries (1.65 mm o.d., AM Systems, Everett, WA, USA). The pipettes were pulled using a gravity puller (DKI model 700C) and the tips were fire polished. The internal pipette solution had the following composition (mm): KCl 113, glucose 5.5, K₂ATP 5, MgCl₂ 0.5, Hepes 10, NaCl 15. The pH was adjusted to 7.2 with KOH. Pipette resistance ranged from 1 to 3 MΩ when filled with the internal solution. Myocytes were superfused with the Hepes buffer (described above) and 2 mm 4-AP was added to inhibit transient outward current (Cordeiro et al. 1998). L-type Ca²⁺ current was recorded with an Axopatch 200B amplifier (Axon Instruments Inc., Foster City, CA, USA) and series resistance errors were reduced by 65-70% with electronic compensation. The voltage clamp protocol for the measurement of I_Ca consisted of three pre-pulses followed by a step to +10 mV for 150 ms from a post-conditioning potential of -40 mV. Membrane currents were recorded and analysed with a microcomputer running pCLAMP 7.0 software (Axon Instruments).

Drugs

Nifedipine and probenecid were purchased from Sigma Chemicals (St Louis, MO, USA). Fluo-3 AM, di-8-ANEPPS and Texas Red were purchased from Molecular Probes (Eugene, OR, USA). Thapsigargin and ryanodine were obtained from Calbiochem (La Jolla, CA, USA). Anti-ryanodine receptor antibody was purchased from Affinity Bioreagents (Golden, CO, USA).

Mathematical modelling

To quantitatively examine the movements of Ca²⁺ within Purkinje cells, a mathematical model of Ca²⁺ diffusion and binding was constructed. The Purkinje cell was modelled as a right cylinder with calcium binding by ATP, fluo-3 and troponin (Tn). The diameter of the cylinder was set to the diameter of the cell observed in confocal imaging experiments. The cylindrical geometry allowed the cell to be divided into 100 concentric compartments to allow equations describing the diffusion of Ca²⁺, Mg²⁺, ATP and fluo-3 to be examined. The method of solution of the diffusion/reaction equations was identical to that described previously (Cannell & Allen, 1984). The chemical reactions modelled were:

with rate constants as given by Baylor & Hollingworh (1998) adjusted to 36°C by assuming a Q₁₀ of 2.0. The diffusion coefficient of Ca²⁺ and fluo-3 was 3.0 x 10⁻⁶ cm² s⁻¹ and 0.25 x 10⁻⁶ cm² s⁻¹, respectively. The concentrations of the various buffers has not (to our knowledge) been measured in Purkinje fibres so the computer model was fitted to experimental data by adjustment of the [Tn], [ATP] and [fluo-3] (see Results). The calcium current recorded in voltage clamp experiments was used to drive calcium changes in the outermost compartment. Ca²⁺ then diffused from this compartment by simple diffusion. A simple first order differential equation provided transsarcolemmal Ca²⁺ extrusion. The rate of extrusion was defined by the rate of decline of Ca²⁺ once the voltage clamp pulse was over. The variable parameters were then adjusted to achieve the best possible fit to data obtained in the presence of ryanodine.

RESULTS

Relationship between cell structure and Ca²⁺ transients

The fluorescent indicator di-8-ANEPPS was used to label surface membranes in rabbit Purkinje myocytes (Fig. 1A) and ventricular cells (Fig. 1B). The di-8-ANEPPS stain was only apparent at the periphery of the Purkinje cell showing that Purkinje cells lack t-tubules whereas the banding pattern in ventricular cells was indicative of a well developed t-tubular system. Since the t-tubular system permits the rapid spread of the calcium trigger signal to the interior of the cell (Cheng et al. 1994), the lack of t-tubular system in the Purkinje cell should result in a fundamentally different distribution of Ca²⁺ during excitation. As expected, transverse (x,t) line scan images from Purkinje cells (Fig. 2, top left) and ventricular cells (Fig. 2, top right) revealed very different spatial patterns of Ca²⁺-dependent fluorescence associated with electrical stimulation. In the ventricular cell, fluorescence increased more or less synchronously across the cell (Cheng et al. 1994; Cannell et al. 1994) whereas in the Purkinje cell, the fluorescence trace showed a U-shaped profile with the fluorescence increase at the edge preceding that in the centre. The time course of fluorescence change at the edge and centre of the two types of cells is shown in the lower panels of Fig. 2. Although the fluorescence change at the centre and edge of the ventricular cell was essentially identical (see Cheng et al. 1994), in the Purkinje cell, F/F_o reached ≈2.0 at the surface of the cell and 1.75 at the centre. Similar results were observed in every cell examined, with peak F/F_o being 2.28 ± 0.16 at 16.5 ± 1.4 ms at the edge and 1.69 ± 0.063 at 107.1 ± 15.8 ms in the centre (n = 8). These results show that in Purkinje cells, the rise of the Ca²⁺ transient was slower and of lower magnitude in the centre of the cell than at the surface. A similar pattern of activation has been observed in atrial cells, a preparation that also lacks t-tubules (Huser et al. 1996). The U-shaped fluorescence pattern is consistent with the idea that Ca²⁺ is released just beneath the surface of the cell and then moves into the interior by diffusion. The more uniform rise seen in the ventricular cell is consistent with the presence of t-tubules which, by providing contacts between the surface membrane and the SR deep inside the cell, allow synchronous activation (Cheng et al. 1994).

Figure 1. Sarcolemmal staining of Purkinje and ventricular myocytes.

Purkinje cell (A) and a ventricular cell (B) stained with the membrane dye di-8-ANEPPS.

Figure 2. Representative transverse (x,t) line scan recorded from a rabbit Purkinje cell (left panel) and ventricular myocyte (right panel).

The myocytes were field stimulated at a cycle length of 2 s. These images were obtained by calculating the ratio (F/F_o) of the single fluorescence (F) and dividing it by the background fluorescence (F_o) measured before field stimulation. Lower left panel, time course of F/F_o at the centre and edge of a Purkinje cell. Lower right panel, time course of F/F_o at the centre and edge of a ventricular cell.

Movement of Ca²⁺ from the periphery to the centre of the cell

The SR is believed to be distributed throughout the cells of Purkinje fibres (Sommer & Johnson, 1968; Jorgensen et al. 1993) and is essential for the ‘calcium-induced calcium release’ (CICR) mechanism (Fabiato, 1985). It follows that the rise in calcium in the cell interior might be due to both diffusion and CICR. The CICR mechanism is known to reside in the Ca²⁺-dependent gating of the SR calcium release channel (or ryanodine receptor (RyR)). We therefore performed immunocytochemistry using antibodies specific for the cardiac isoform of the RyR (RyR2). Cells were labelled with primary antibodies raised against RyR2 and visualized with secondary antibody conjugated to Texas Red. Ventricular cells showed prominent banding at the level of the t-tubules indicating that ryanodine receptors were located at sites adjacent to the t-tubules (Fig. 3B). Interestingly, the Purkinje cells also exhibited bands in the interior of the cell (Fig. 3a). Although the bands do not appear to be as well organized and intense as in ventricular cells, the result suggests that ryanodine receptors are located in the cell interior. In the absence of primary antibodies, application of the secondary antibody did not produce any intense regions of labelling (data not shown). Given that the primary antibody is highly specific for RyR2 (Lai et al. 1992; Jorgensen et al. 1993; Junker et al. 1994) we conclude that the RyR2 necessary for CICR is present in the interior of the cell.

Figure 3. Immunohistochemical localization of ryanodine receptors in Purkinje (A) and ventricular cells(B).

The ventricular cell shows prominent banding at the level of the t-tubules suggesting that ryanodine receptors were located at sites adjacent to the t-tubules (B). The Purkinje cell also exhibited bands of intense staining in the interior of the cell (A). This suggests that ryanodine receptors are located in the cell interior. The spacing between the bands of intensity measured in 4 Purkinje and 4 ventricular cells was found to be 1.95 and 1.98 μm, respectively.

Having established the possibility for CICR in Purkinje cells, we compared Ca²⁺ transients that were evoked by field stimulation in the absence and presence of ryanodine. Line scan images were recorded and the fluorescence from these images was averaged (Fig. 4A and B). The line scan images displayed the same general patterns under control conditions (top panel) and after exposure to ryanodine (200 nm) for 15 min (bottom panel). Exposure to ryanodine reduced the amplitude of the Ca²⁺ transient both at the surface and the interior of the cell. Since treatment with ryanodine should have inhibited any CICR in the interior of the cell, this result indicates that regenerative calcium release is not necessarily the major determinant of the amplitude and time course of the Ca²⁺ transient in the interior of the cell. However, full interpretation of this result requires a detailed understanding of the interaction between fixed and mobile calcium buffers within the cell as well as accounting for changes in the time course of the Ca²⁺ transient at the surface of the cell.

Figure 4. Ratio images (F/F_o) of transverse (x,t) line scan plots recorded from a Purkinje cell under control conditions (A, 2 mm Ca²⁺, left panel) and after 15 min exposure to 200 nm ryanodine (B, left panel).

The myocytes were field stimulated at a cycle length of 2 s. Time course of F/F_o at the centre and edge of the Purkinje cell in the absence (A, right panel) and presence (B, right panel) of ryanodine. The rate constants for diffusion of Ca²⁺ from the edge to the centre were calculated by fitting the curve with a single exponential. The rate constant for this rise was found to be 68 ms⁻¹ in the absence of ryanodine and 77 ms⁻¹ after application.

Mathematical modelling of calcium changes in Purkinje cells

A computer model of the calcium movements and fluo-3 fluorescence signals in Purkinje fibres was constructed (see Methods). Since there appears to be little information in the literature regarding Ca²⁺ movements in Purkinje cells we used voltage clamp techniques to measure the time course of the Ca²⁺ current in the presence of ryanodine and then fitted the model to the resulting data (Fig. 5). Since the time course of calcium influx was known and CICR was blocked, with suitable adjustment of parameters, the computer model should be able to describe the passive processes that determine the time course of calcium movements across the cell. An optimal fit was obtained with the [ATP] set to 5 mm, [fluo-3] to 60 μm and [TNC] to 50 μm. The former two parameters acted as validating benchmarks for the model as they should be closely related to the solution of the patch pipette-filling solution that was used to voltage clamp the cell. The close agreement between these parameters and the composition of patch pipette-filling solution suggested that we obtained a reasonable first order description of the calcium buffering properties as well as calcium transport by passive processes. Next, the model was re-run with a calcium influx at the surface scaled by a factor of 1.5 which resulted in the peripheral fluorescence transient closely matching the experimental observations obtained in the absence of ryanodine. With this modification, it is notable that the model also reproduced the peak amplitude of the Ca²⁺ transient in the centre of the cell also (Fig. 5). The ability of the model to reproduce the amplitude of the central Ca²⁺ changes when the SR was not inhibited by ryanodine suggests that the observed changes in Ca²⁺ could be explained by simple buffered diffusion alone. In other words, this result implies that it is not necessary to invoke any compartment except the cell surface to explain the observed Ca²⁺ changes. In addition, since simple scaling of the calcium influx at the cell edge enabled the model to reproduce the peripheral Ca²⁺ transient also, the model results suggest that CICR contributed about 50% of the Ca²⁺ influx underlying the Ca²⁺ transient at the cell surface and that the time course of this calcium release was very similar to that of the calcium current. Although the model did not reproduce the time course of the central Ca²⁺ transient exactly, we ascribe this shortcoming to uncertainties in the exact kinetic properties of the Ca²⁺ buffers as well as the possibility that the cell was ellipsoidal (rather than circular) in cross section. This deficiency does not affect the conclusion that the increased amplitude of the central Ca²⁺ transient observed in the absence of ryanodine could be explained by simple buffered diffusion alone.

Figure 5. Comparison of experimentally determined and computed calcium transients in Purkinje cells.

Top, representative recordings of I_Ca measured at +10 mV from a Purkinje cell. I_Ca measured at +10 mV was used to determine the time course of Ca²⁺ movement. A and B, graphs showing time course of Ca²⁺ movement and fluo-3 fluorescence signals obtained from a Purkinje cell in the absence and presence of ryanodine. Ryanodine decreased the fluorescence intensity but did not change the time course of Ca²⁺ movement. C and D, time course of Ca²⁺ movement and fluo-3 fluorescence signals obtained by mathematical modelling (see Results). No CICR mechanism was included in the formulation and all Ca²⁺ movement was by simple diffusion. The time course of Ca²⁺ movement in the model resembles the experimental data suggesting that the amplitude of the central transient is the result of simple diffusion from the cell periphery.

Localization of Ca²⁺ sparks in Purkinje cells

As rabbit Purkinje cells lacked t-tubules (Fig. 1a), it seemed likely that dyadic junctions between the sarcolemma and the SR should occur only at the periphery of the cell. Since Ca²⁺ sparks (Cheng et al. 1993) are believed to arise from a cluster of RyRs in such dyadic junctions (Cannell et al. 1994; Bridge et al. 1999), we expected to find sparks near the cell periphery. On the other hand, if sparks did not depend on clusters of RyRs in restricted spaces it might be possible to observe them in the cell interior since RyRs are also present there. In initial experiments, we observed that the frequency of Ca²⁺ sparks was very low in quiescent Purkinje cells under control conditions. Since increasing external Ca²⁺ increases the frequency of spontaneous spark occurrence (Cheng et al. 1996b), we superfused Purkinje cells with a buffer containing 5 mm Ca²⁺-1 μm isoproterenol and evoked [Ca²⁺]_i transients by field stimulation (Fig. 6). After exposure to this solution, the intensity of the Ca²⁺ transient (Fig. 6B) markedly increased at the cell periphery with a more modest increase at the cell centre compared with control (Fig. 6A). The maximum F/F_o was 4.12 ± 0.26 at the edge and 2.27 ± 0.11 at the centre of the Purkinje cells (n = 5). The distribution of Ca²⁺ during these [Ca²⁺]_i transients is easier to appreciate in the 3-dimensional reconstructions of the fluorescence signal (Fig. 6C and D). Images obtained from cells treated with isoproterenol revealed spark-like release events only at the cell periphery after repolarization.

Figure 6. Calcium sparks associated with evoked calcium transients in a Purkinje myocyte.

A, ratio images of transverse line scans from a Purkinje cell recorded under control conditions (2 mm external Ca²⁺, left side) and after application of 5 mm Ca²⁺-1 μm isoproterenol (B). Cells were field stimulated at a cycle length of 2 s. After application of 5 mm Ca²⁺-1 μm isoproterenol, the relative fluorescence intensity (F/F_o) of the Ca²⁺ transient was greater compared with control conditions. C and D, a 3-dimensional reconstructon of A and B, respectively (filtered with a 3rd order boxcar filter). This provides an impression of the distribution of Ca²⁺ during transients evoked in the presence and absence of isoproterenol. Spark-like events are apparent in both the 2- and 3-dimensional images obtained for isoproterenol-treated cells.

To localize the distribution of Ca²⁺ sparks in Purkinje cells, we performed x,y scans on resting Purkinje cells exposed to a 5 mm Ca²⁺-1 μm isoproterenol solution. A representative montage of images (Fig. 7) obtained from a single Purkinje cell shows that sparks are only observed at the cell periphery. Furthermore, the amplitude (F/F_o) of these sparks was between 1.5 and 2.5 which is similar to values obtained in studies on ventricular cells (see Cheng et al. 1996b). In 132 sweeps recorded from four Purkinje cells, Ca²⁺ sparks were observed in 64 sweeps and no sparks were observed after exposure to thapsigargin (500 nm). These results confirm the involvement of the SR in the genesis of these calcium sparks, again similar to results obtained from ventricular cells.

Figure 7. Montage of x,y line scan diagrams showing the distribution of Ca²⁺ sparks in a quiescent Purkinje cell.

Each panel represents one sweep of the cell. Cells were exposed to 5 mm Ca²⁺-1 μm isoproterenol and were not field stimulated. The first image (uppermost panel) represents 30 signal-averaged images. In the ratio images, the average fluorescence intensity from 30 x,y sweeeps was determined (F_o, shown at the top of the figure). The fluorescence intensity from each panel (F) was then divided by the fluorescence intensity of the averaged image (F_o). The intensity (F/F_o) of 8 representative sweeps are shown in this figure. This montage clearly shows that Ca²⁺ sparks occurred at the periphery of the Purkinje cell and no central Ca²⁺ release was observed.

Because of the relatively slow rate of scanning (about 300 ms per frame), the x,y scans in Fig. 7 do not give detailed information concerning the magnitude and time course of Ca²⁺ sparks. To examine spontaneous spark formation in greater detail, we performed transverse x,t scans under the same conditions as shown in Fig. 7. The sparks in the Purkinje cell only occurred in close proximity to the surface membrane (Fig. 8C and D). In contrast and as reported by others (Cheng et al. 1995; Tanaka et al. 1998), sparks in ventricular cells are observed throughout the cell. The time course of decay of the sparks in both cell types appeared to be similar. These results suggest that evoked Ca²⁺ transients and Ca²⁺ sparks originate at the periphery of Purkinje cells, presumably where the sarcolemma and SR form junctions.

Figure 8. x,t line scan images (F/F_o) recorded from a ventricular cell (A and B) and a Purkinje cell (C and D) displaying individual sparks.

Cells were exposed to 5 mm Ca²⁺-1 μm proterenol and spontaneous sparks recorded. The spark in the Purkinje cell (C and D) occurred in close proximity to the cell membrane and Ca²⁺ diffused to more internal sites over time. The spark in the ventricular cell was not limited to the periphery of the cell. The magnitude and the time course of decay of the sparks in both cell types appeared to be similar. B and D show a 3-dimensional reconstruction of A and C, respectively.

Since the Purkinje cells were quiescent and were being superfused with a buffer containing 5.4 mm external K⁺, it seemed likely that the cells were well polarized. Therefore it was unlikely that Ca²⁺ influx through L-type Ca²⁺ channels was responsible for initiation of these spontaneous Ca²⁺ sparks. We tried to exclude further the possibility that Ca²⁺ influx through L-type Ca²⁺ channels initiated spontaneous Ca²⁺ sparks by measuring the frequency of spark production in Purkinje myocytes before and after addition of 10 μm nifedipine, an inhibitor of L-type Ca²⁺ channels. Quiescent cells were scanned in the x,y mode once per second. Under control conditions, Ca²⁺ sparks were observed in 27/60 sweeps from three Purkinje cells but only observed at peripheral sites. Myocytes were then exposed to 10 μm nifedipine. As the nifedipine was applied, cells were briefly field stimulated at a cycle length of 2 s to promote blockade of the L-type Ca²⁺ channels (Sanguinetti & Kass, 1984). Application of nifedipine did not decrease the frequency of sparks (23/60 sweeps) after 2 min of exposure to the drug. These results suggest that spontaneous Ca²⁺ sparks in resting Purkinje cells are not initiated by influx of Ca²⁺ via L-type Ca²⁺ channels.

DISCUSSION

Ca²⁺ transients in Purkinje cells

When Ca²⁺ transients were activated by field stimulation, line scan images showed a U-shaped profile (Fig. 2) which indicated that Ca²⁺ release is initiated at the cell periphery and Ca²⁺ then moves towards the cell centre by diffusion. When we stained the cell with a monoclonal antibody specific for RyR2 we found that RyRs exist in rows perpendicular to the sarcolemma throughout the cell interior although the membrane stain di-8-ANEPPS indicated a complete lack of t-tubules. These results are consistent with the findings of others (e.g. Sommer & Johnson, 1968; Hatem et al. 1997). It follows that junctions between RyRs and the sarcolemma could only occur at the cell periphery. Current cardiac E-C coupling theories propose that evoked Ca²⁺ transients represent the summation of many Ca²⁺ sparks which originate at junctions between the sarcolemma and SR (Cannell et al. 1994; Shacklock et al. 1995; Tanaka et al. 1998), and consistent with this idea, Ca²⁺ sparks were also only observed at the cell periphery. Therefore, the structural data, location of calcium sparks and the peripheral activation of calcium transients are in accord. The requirement for junctional regions may be due to the need for Ca²⁺ influx via Ca²⁺ channels to be sufficiently concentrated (or localized in space) to achieve a sufficiently high concentration to activate one or more RyRs (Santana et al. 1996; Bridge et al. 1999). The restricted nature of the junctional space would also help increase local Ca²⁺ levels (Soeller & Cannell, 1999) to promote activation of junctional RyRs.

The spontaneously occurring Ca²⁺ sparks had a fluorescence intensity (F/F_o) of approximately 1.3-2.5 and could be abolished by thapsigargin, confirming the involvement of the SR in their genesis. As such, they resembled sparks observed in other heart cells (Cheng et al. 1996b; Satoh et al. 1997). However, it has been suggested that small calcium release events may occur from single RyRs or small clusters of RyRs separate from those associated with junctions (Lipp & Niggli, 1996). The central RyRs which were detected by immunohistochemistry could perform such a role, but computer modelling of the electrically evoked Ca²⁺ transient did not reveal a component of calcium release at the cell centre (see below). Consistent with this result, we did not detect any spontaneous release events away from the cell periphery. The lack of detectable spontaneous Ca²⁺ release cannot be explained by the distance of the central RyRs from the cell surface (and the lower [Ca²⁺] experienced by such RyRs) since the resting Ca²⁺ level was sufficient to produce spontaneous surface Ca²⁺ sparks in the presence of nifedipine. Three possible explanations for this result are: (1) RyRs in the centre of the cell are effectively ‘silent’ (and do not contribute significantly to the electrically evoked Ca²⁺ transient). (2) The junctional RyRs at the surface membrane are exposed to higher levels of Ca²⁺ at rest. Since the interior of the cell is in continuity with the interior surface of the cell, it is difficult to explain how the junctional RyRs could experience a higher resting level of Ca²⁺ unless the junctional region has a higher level of calcium entry (or lower level of calcium extrusion) at rest than the rest of the surface membrane. (3) Sparks must arise from clusters of RyRs which are only present at junctions between the surface membrane and the SR.

It is possible that the RyRs and the SR elements that contain them are targeted to the cell cytoskeleton associated with Z-lines, but since t-tubules do not invaginate the cell interior they are not assembled into clusters from which sparks may occur. Their inability to release calcium during the calcium transient may then be explained by the low amplitude of the calcium trigger arising from calcium diffusion from the cell surface to the periphery. The lack of developed clusters would explain the absence of detectable spontaneous calcium release in the cell centre (and thereby overcome the need to propose different local surface membrane properties) since the local calcium flux produced by a single RyR would not be amplified by its neighbours in the cluster (it is also possible that there may be a more limited SR calcium store in their location). In this scheme, the RyRs may be functional but are uncoupled from their normal function by the geometric organization associated with Purkinje cell microarchitecture. In this regard, we can consider that the central Purkinje RyRs may be similar to the ‘extra-junctional’ RyRs of ventricular muscle (Franzini-Armstrong et al. 1999). Although we favour this explanation, we cannot exclude the possibility that the RyRs in the cell centre may be ‘silent’.

In our study, we observed a monotonic rising phase of the Ca²⁺ transient, which has also been observed by others in rabbit Purkinje cells using epifluorescence (Sipido et al. 1993; Papp et al. 1995). In contrast, canine Purkinje cells exhibit a two-stage rise in the Ca²⁺ transient (Wier & Isenberg, 1982; Boyden et al. 2000). Furthermore, Ca²⁺ waves were frequently noted in that preparation (Boyden et al. 2000). We never observed Ca²⁺ waves in rabbit Purkinje cells, even during Ca²⁺ overload (Cordeiro et al. 2000). However, we have observed Ca²⁺ waves in canine Purkinje cells (unpublished observations). Therefore, the nature of the Ca²⁺ transient appears to be different in rabbit and canine Purkinje cells, even though both preparations lack t-tubules.

Ca²⁺ transients in atrial cells, which also lack t-tubules, are also initiated at the cell periphery (Huser et al. 1996). Huser et al. (1996) reported that in these cells, spontaneous Ca²⁺ sparks were rarely observed in the cell centre but commonly observed at the cell periphery. Evidence for evoked Ca²⁺ release in the centre of these cells was also obtained by these authors, but it is important to note that (1) the spatial pattern of Ca²⁺ movement across the atrial cells was much less uniform than observed in the Purkinje cells studied here and (2) the central Ca²⁺ transient could have the same amplitude as the peripheral transient. These observations suggest that the atrial cell relies on Ca²⁺ diffusion to trigger SR calcium release from internal sites (Huser et al. 1996) while the Purkinje cell has a very limited capacity to reinforce the inward propagation of Ca²⁺ via CICR. Given the relatively low amounts of contractile protein in Purkinje cells (Truex & Smythe, 1965) compared with atrial cells (Sommer & Johnson, 1979), these differences in the pattern of calcium release within the cell seem consistent with function. In addition, since the contractile protein may be concentrated at the periphery of the Purkinje cell in some cases (Truex & Smythe, 1965), calcium release from deeper SR (in such cells) would serve no obvious function in either electrical propagation or mechanical contraction.

Modelling the Purkinje cell Ca²⁺ transient

The simple computer model presented here was able to reproduce the major spatio-temporal features of the Purkinje cell Ca²⁺ transient. By fitting the parameters of the model to the Ca²⁺ transient observed in the presence of ryanodine with a measured Ca²⁺ influx (due to the Ca²⁺ current) we obtained an estimate of the buffering properties of the cell. By fixing the buffering parameters we showed that the increased [Ca²⁺] observed in the cell centre in the absence of ryanodine (i.e. when the SR was not inhibited) could be explained entirely by the increased Ca²⁺ levels that occur at the cell periphery in these conditions. From this result we concluded that there is little evidence for evoked SR calcium release at any site other than the cell surface. However, the model also showed that a much lower concentration of fixed binding sites (Tn) appeared to be present than in models of ventricular cells. The value we obtained for these sites (50 μm) is considerably less than that usually considered to be appropriate for mathematical models of ventricular cells. However, we do not consider this to be a deficiency of the present model, but rather that it reflects the lower density of contractile machinery present in these cells (see above).

The model also showed that the peripheral release of Ca²⁺ by the SR was a surprisingly small fraction of the Ca²⁺ influx due to the Ca²⁺ current. It is likely that the smaller than normal Ca²⁺ transient was the result of the conditions used to achieve voltage clamp recordings, but the internal dialysis of the cell by the voltage clamp pipette and possible alteration of SR Ca²⁺ load is unavoidable with this technique. This concern leads to the possibility that, under normal conditions, the increased SR Ca²⁺ content might permit CICR in the cell interior. When isoproterenol was applied there was a marked increase in the amplitude of the peripheral Ca²⁺ transient, confirming an increase in SR content (Fig. 6). However, under these conditions the U-shaped diffusion profile was not diminished suggesting that diffusion to the centre of the cell, which would be accelerated by internal SR release, was not very different from that observed in control conditions. In addition, the magnitude of the Ca²⁺ transient at the edge of the cell was increased by a factor of ≈5 while at the centre it increased by a factor of only ≈1.5. The computer model also predicted a similar disproportionate increase in Ca²⁺ at the cell periphery compared with the cell centre (due to local buffer saturation – not shown). It follows that even when the SR calcium content was augmented by isoproterenol, Ca²⁺ release by internal SR Ca²⁺ stores remained very small (or non-existent).

Local control of SR release in Purkinje cells

The spontaneous Ca²⁺ sparks in Purkinje cells (which only occurred at the cell periphery) were similar to those measured in other cell types (Cheng et al. 1996a). Others have shown that the primary site of Ca²⁺ release in ventricular muscle is in regions adjacent to the t-tubules (inferred to be junctional regions) (Shacklock et al. 1995). This similarity suggests that the existence of typical sparks is predicated on the existence of junctions between the cell surface membrane and the SR. The RyRs in the cell interior revealed by immunocytochemistry would therefore be exposed to a much lower [Ca²⁺] and probably would not be activated. However, the activation of L-type Ca²⁺ channels is not an absolute requirement for the activation of Ca²⁺ sparks since spontaneous Ca²⁺ sparks also occur at rest during L-type Ca²⁺ channel blockade (e.g. Cannell et al. 1994).

To reconcile the presence of central RyRs with a lack of any detectable Ca²⁺ release we suggest that such ‘silent’ RyRs occur at low local densities i.e. they are not assembled into clusters (see below). Although such isolated RyRs have been proposed to underlie microscopic Ca²⁺ release events called ‘calcium quarks’ (Lipp & Niggli, 1996), we did not detect any evoked or spontaneous Ca²⁺ release in the cell centre.

The lack of detectable Ca²⁺ release at internal sites supports the idea that (under normal circumstances) Ca²⁺ does not enter a common pool to stimulate regenerative release of Ca²⁺ from other RyRs. Stern (1992) has explained in some detail the difficulties associated with such ‘common pool models’ for achieving graded or controlled responses. Our results are consistent with the local control mechanism described by Stern (1992) which requires the presence of a Ca²⁺‘synapse’ to confine regeneration. The clustering of RyRs in junctional regions at the sarcolemma provides high gain that is stable and spatial separation of RyRs will limit the possibility of regeneration. In the case of Purkinje fibres, if the central RyRs were activated by the levels of Ca²⁺ that occur as a result of calcium movement from the periphery, they probably would also be activated at resting levels of Ca²⁺ as well. In this case, the RyRs would not be under the tight control of the surface membrane and they would represent an unstable system with a high probability for propagating spontaneous calcium release. The uncoupling of the central RyRs from E-C coupling would therefore help prevent uncontrolled regenerative release and the resulting electrical disturbances (e.g. transient depolarizations). Given the importance of the Purkinje system in the spread of cardiac activation, it is possible that this uncoupling is an important modification of E-C coupling to reduce the probability of abnormal activation.