Inositol 1,4,5-Trisphosphate Directs Ca²⁺ Flow between Mitochondria and the Endoplasmic/Sarcoplasmic Reticulum: A Role in Regulating Cardiac Autonomic Ca²⁺ Spiking

Abstract

The signaling role of the Ca²⁺ releaser inositol 1,4,5-trisphosphate (IP₃) has been associated with diverse cell functions. Yet, the physiological significance of IP₃ in tissues that feature a ryanodine-sensitive sarcoplasmic reticulum has remained elusive. IP₃ generated by photolysis of caged IP₃ or by purinergic activation of phospholipase Cγ slowed down or abolished autonomic Ca²⁺ spiking in neonatal rat cardiomyocytes. Microinjection of heparin, blocking dominant-negative fusion protein, or anti-phospholipase Cγ antibody prevented the IP₃-mediated purinergic effect. IP₃ triggered a ryanodine- and caffeine-insensitive Ca²⁺ release restricted to the perinuclear region. In cells loaded with Rhod2 or expressing a mitochondria-targeted cameleon and TMRM to monitor mitochondrial Ca²⁺ and potential, IP₃ induced transient Ca²⁺ loading and depolarization of the organelles. These mitochondrial changes were associated with Ca²⁺ depletion of the sarcoplasmic reticulum and preceded the arrest of cellular Ca²⁺ spiking. Thus, IP₃ acting within a restricted cellular region regulates the dynamic of calcium flow between mitochondria and the endoplasmic/sarcoplasmic reticulum. We have thus uncovered a novel role for IP₃ in excitable cells, the regulation of cardiac autonomic activity.

INTRODUCTION

The signaling role of inositol 1,4,5-trisphosphate (IP₃) via intracellular Ca²⁺ mobilization has been established in many cell types and associated with secretion, neurotransmission, fertilization, cell motility, gene expression, or cell death (Berridge, 1993; Clapham, 1995). In the heart, IP₃ is generated by a plethora of neurohumoral agonists. These include acetylcholine, endothelin, catecholamines, or prostaglandins (Brown et al., 1985; Hilal-Dandan et al., 1992; Adams et al., 1998) that activate a Gq-dependent phospholipase Cβ (PLCβ) or purines or angiotensin II that stimulate a tyrosine kinase–dependent PLCγ (Puceat and Vassort, 1996; Goutsouliak and Rabkin, 1997). Yet, the role of IP₃ in the heart has remained elusive. In cardiac cells, rhythmic changes in membrane potential and associated Ca²⁺-induced Ca²⁺ release (CICR) from the sarcoplasmic reticulum (SR) are believed responsible for repetitive Ca²⁺ transients. Therefore, it remains an enigma, why would cardiac cells use, in addition to ryanodine receptors (RyR) of the SR, IP₃ receptors (IP₃R) to also regulate Ca²⁺ homeostasis.

Although IP₃ has been implicated in cardiac arrhythmias (Jacobsen et al., 1996), heart failure (Marks, 1997), and graft rejection (Felzen et al., 1997), the significance of IP₃ in cardiac function has remained unresolved, in part, because of confounding effects of abundant RyRs. We here used neonatal rat cardiomyocytes in culture, which feature an immature SR and express a low density of RyRs (Fitzgerald et al., 1994). Such cells exhibit spontaneous, rhythmic action potentials (Jongsma et al., 1983) resulting in repetitive Ca²⁺ oscillations and spontaneous autonomic activity. Development of automaticity in these cells depends primarily on the expression of two ion channels, a hyperpolarization-activated I_f current, with the same range of potential-dependent activation and Cs⁺ sensitivity as that found in pacemaker cells of the sinoatrial node, and a transient voltage-dependent Ca²⁺ (I_CaT) current (Gomez et al., 1994; Fares et al., 1998). Moreover, the establishment of intercalated disks, gap junctions, and T-tubes (Moses and Kasten, 1979) synchronizes the electrical activity of connected cells. With such cytoarchitectural and electrophysiological properties, neonatal rat cardiomyocytes in culture represent a unique cellular model in which to study the regulation of cardiac autonomic activity.

Here, we show that IP₃ generated by intracellular photorelease of a caged precursor or by purinergic stimulation of cells, and acting on IP₃Rs, triggers a spatially restricted Ca²⁺ release from a ryanodine-insensitive intracellular store. Moreover, IP₃ triggers mitochondrial Ca²⁺ uptake and depolarization associated with SR Ca²⁺ depletion that slows down or arrests autonomic Ca²⁺ spiking. Thus, our results provide the first evidence that in cardiac cells, IP₃-induced Ca²⁺ release from a ryanodine-insensitive pool modulates cardiac autonomic activity via mitochondrial signaling.

MATERIALS AND METHODS

Cell Isolation and Culture

Cardiomyocytes were isolated from 2- to 3-d neonatal rats (Puceat et al., 1994) and kept in culture for 5 d. Purkinje cells were isolated from rabbit hearts as described previously (Scamps and Carmeliet, 1989).

Cell Transfection

Cardiomyocytes were transfected with a mitochondria-targeted cameleon using Effectene (Qiagen, Hilden, Germany) according to the manufacturer's protocol. Experiments were performed 2–3 d later to ensure that mitochondrial targeting was completed (Miyawaki et al., 1997).

Immunoprecipitation of Proteins and Western Blotting

Whole-cell lysates were subjected to immunoprecipitation (Puceat and Vassort, 1996). Samples were run in 5% SDS PAGE and electrophoretically transferred to a nitrocellulose filter. Blots were treated as described previously (Puceat and Vassort, 1996) and probed with the antibody recognizing the protein of interest and a secondary peroxidase-conjugated antibody. Proteins were revealed using ECL reagent. Bands on films were quantified by an imaging system (SCION NIH IMAGE software; Bethesda, MD). The anti-IP₃ receptor type I antiserum was raised against the 14 C-terminal amino acid residues (GHPPHMNVNPQQPA), and the anti-IP₃ receptor type II antisera were raised against the 13 C-terminal amino acids (SNTPHVNHHMPPH [Parys et al., 1995]) or against the sequence of 16 C-terminal amino acids (FLGSNTPHENHHMPPH).

Immunostaining

Cells were fixed with 3% paraformaldehyde, immunostained with an affinity-purified anti-IP₃R antibody directed against the N-terminal domain of the IP₃RI (AA 337–349, QDASRSRLRNAQE), the anti-IP₃RII antiserum, an anti-calreticulin, or monoclonal anti-RyR2 antibodies and a secondary fluorescein-conjugated antibody, and imaged by confocal microscopy (Puceat et al., 1995).

Microinjection, Confocal Microscopy, and Cell Ca²⁺ Imaging

Microinjection of neonatal cells was performed as described (Puceat et al., 1998). The concentrations of Fluo3, Ca²⁺ Green, caged IP₃, and Ca²⁺-saturated caged EGTA in the pipette were 2.5, 2.5, 2, and 5 mM, respectively. Noninjected cells were loaded with 3 μM Fluo3 AM for 20 min at room temperature. Cells were transferred to the stage of an epifluorescence microscope (Zeiss, Thornwood, NY) and superfused with a medium containing (in mM): HEPES 20, NaCl 117, KCl 5.7, NaH₂PO₄ 1.2, NaHCO₃ 4.4, MgCl₂ 1.7, and CaCl₂ 1.8. In Na⁺ and Ca²⁺-free solution, 1 mM EGTA was added, LiCl replaced NaCl, and NaH₂PO₄ and NaHCO₃ were omitted. Cardiomyocytes were imaged with a Zeiss LSM-410 or LSM-510 laser-scanning microscope (Thornwood, NY) using the 488-nm line of an argon/krypton laser. Fluo3 or Ca²⁺ Green emission fluorescence was recorded through a dichroic mirror (cutoff of 510 nm) and a long-pass emission filter (cutoff of 520 nm) as described (Jaconi et al., 1997). Caged compounds were photoreleased by simultaneously scanning the field of interest with the 363-nm line of an argon/UV laser. The scan duration was determined using a Uniblitz shutter (Vincent Associates, New York, NY) placed in the UV path. The power of the UV laser beam was set to 200–250 μW, as measured at the aperture of the 40× objective. Exposure of cardiomyocytes to several UV laser scans (for 0.1 or 1 s) in the absence of a caged compound did not affect [Ca²⁺]_i, indicating that UV light does not damage cells under this condition. For localized and spatially restricted IP₃ and Ca²⁺ uncaging, a Zeiss LSM-510 microscope was used. Uncaging was performed in a region of “bleaching” drawn freehand around the nucleus of a cell. Uncaging was performed by bleaching (UV scanning) the area during one scan (100 ms) of concomitant argon/krypton and UV lasers. Experiments were performed at 20 ± 2°C. Sequences of digitized images were background-subtracted and analyzed using the ANALYZE software (Mayo Foundation, Rochester, MN). In experiments designed to look at localized Ca²⁺ events, the first image of the series (Fo) was first subtracted from the following. Then each image of the series was divided by the first image (ΔF/Fo). This normalization of images allows one to take into account the local inhomogeneities of Fluo3. Caged EGTA saturated with Ca²⁺ (Molecular Probes, Eugene, OR) was used to release Ca²⁺ in a locally restricted perinuclear area.

Microspectrofluorimetry and Imaging of Cell Ca²⁺ and Membrane Potential

A cell-imaging system was used to record fluorescence from Fluo3-injected cells. The field was illuminated at 485 ± 22 nm with a xenon lamp. Images were recorded at 530 nm using a charge-coupled device (CCD) camera (Hamamatsu, Bridgewater, NJ) and digitized on-line by computer (Argus software; Hamamatsu). Experiments were performed at 35 ± 2°C in cardiomyocytes microinjected using an Eppendorf (Hamburg, Germany) transjector. The intrapipette concentrations were as follows: Fluo3 2.5 mM; heparin 5 mg/ml; anti-PLCγ antibody (Roche et al., 1996), affinity-purified IgG, or anti-yes·6 antibody at 500 μg/ml in KCl 150 mM; EGTA 0.025 mM; EDTA 0.1 mM; and piperazine-N,N′-bis(2-ethanesulfonic acid) 1 mM (pH 7.2). Wild-type (wt) and mutated (mt) glutathione-S-transferase (GST) fusion proteins composed of the two SH2 domains of PLCγ (N- and C-terminal) were prepared as described (Carroll et al., 1997). In the mutated protein, a lysine was replaced by an arginine in the conserved FLVR sequence of the SH2 domains, eliminating binding of the target tyrosine kinase. Fusion proteins were separately injected together with Fluo3 into cardiomyocytes using an intrapipette concentration of 2 mg/ml. IP₃-5-phosphatase was injected at an intrapipette activity of 10 μmol·min⁻¹·ml⁻¹ of injection buffer supplemented with 2.5 mM MgCl₂. In experiments requiring rapid acquisition, we used a photomultiplier tube (Nikon, Garden City, NY) coupled to a microscope. The fluorescence signal was digitized and sampled on a computer using Axotape software (Axon Instruments, Foster City, CA).

Cell membrane potential was measured in cells microinjected with JPW1114 (3 mg/ml in the pipette) (Antic and Zecevic, 1995). Cells were microinjected and fluorescence was measured with a photomultiplier tube at 640 ± 20 nm using a DM 580 dichroic mirror (Nikon) and an excitation wavelength at 514 ± 10 nm. When both membrane potential and [Ca²⁺] were measured, both JPW1114 (3 mg/ml) and Fluo3 (2.5 mM) were injected. The sets of a dichroic mirror and filters mounted on a slider under the objectives of a Nikon microscope were manually switched to allow alternative monitoring of JPW1114 or Fluo3 fluorescence. The JPW1114 fluorescence signal acquired by the computer was filtered, by averaging adjacent data points using the Origin microcal software (Microcal, Northampton, MA). The signal was calibrated by the addition of external KCl to a cell, and the membrane potential was calculated in accord with the Nernst equation (Puceat et al., 1991).

Mitochondrial Ca²⁺ and Membrane Potential Measurements

Cells were loaded for 30 min at room temperature with 3 μM Rhod2 AM. The field was illuminated at 514 ± 10 nm (Rhod2) with a xenon lamp. Images were recorded using a 100× objective at 580 nm (Rhod2) using a CCD camera (Hamamatsu) and digitized on-line using a computer (Argus software; Hamamatsu) or a 63× objective using the LSM-510 confocal microscope (optical section of 0.8 μm). Only cells in which Rhod2 was compartmentalized into mitochondrial clusters (as determined by mitotracker staining; Molecular Probes, Portland, OR) were used in the experiments. Regions of interest (ROIs) were set into the center of bright clusters of mitochondria for recording. The position of these ROIs was monitored during the experiment using an off-line analysis (line scan mode using ANALYZE software), and cells in which the mitochondria moved because of their intrinsic motion or of cell contraction were discarded. Cells were also loaded with TMRM (3 μM) and Fluo3 (3 μM) for 20 min and then washed. Cells were imaged by confocal microscopy using the 488-nm line of an argon laser and a cutoff dichroic mirror of 560 nm. Fluorescence was recorded using two photomultipliers and emission filters of 522DF35 for Fluo3 and 605DF32 for TMRM (Nikon). A focal plane of 0.8 μm was selected in these experiments. Experiments were analyzed using the ANALYZE software (Mayo Foundation). In experiments using a mitochondria-targeted cameleon, cells were illuminated at 430 nm, and fluorescence was recorded with a CCD camera at 480 ± 30 and 535 ± 25 nm. Images were acquired every 3 s. The ratio of images at 535 nm/480 nm was calculated off-line after background subtraction, by the software Metamorph (Universal Imaging, West Chester, PA), and used as an index of mitochondrial Ca²⁺changes. Only cells in which the cameleon was fully targeted into mitochondria were used.

RESULTS

IP₃ Induces Spatially Restricted Intracellular Ca²⁺ Release from a Caffeine- and Ryanodine-insensitive Pool

Cardiomyocytes, plated at low density, feature a steady resting membrane potential associated with an inward-rectifying I_K1 current (Maltsev et al., 1994). A brief UV laser scan, aimed at a portion of a quiescent cell microinjected with both the Ca²⁺-sensitive probe Fluo3 and caged IP_3, triggered the photorelease of IP_3. This, in turn, induced a fast increase in the intracellular Ca²⁺ concentration ([Ca²⁺]_i). Longer laser scans (illuminating cells for several consecutive frames) uncaged a larger amount of IP₃ and induced a larger Ca²⁺ transient (Figure 1A). IP₃-induced Ca²⁺ release was confined to the perinuclear area of cardiomyocytes and did not trigger regenerative waves through the cytosol, nor did it induce CICR from other cellular compartments as observed recently in neurons (Finch and Augustine, 1998) (Figure 1A). Photorelease of IP₃ in a localized region of interest set around the nucleus further revealed the limited diffusion pattern of Ca²⁺ released by IP₃. The Ca²⁺ signal was much smaller and delayed a few micrometers away from the site of IP₃ release. In contrast, Ca²⁺ photoreleased in the perinuclear area from Ca²⁺-saturated caged EGTA triggered Ca²⁺-induced CICR, even far beyond the site of Ca²⁺ release (Figure 1B).

Figure 1.

Localized changes of [Ca²⁺]_i in neonatal rat cardiomyocytes after photorelease of intracellular caged IP₃ or extracellular caged ATP. (A) Caged IP₃ was microinjected together with Fluo3 into cardiomyocytes, and IP₃ was photoreleased by simultaneously scanning the optical field of the microscope with a UV laser beam. The duration of the scans (pink lines) is indicated at the top of the images. Images were acquired every 200–300 ms, and line scan images were built by setting a line crossing the cell (white arrow). The two images on top of the line scan show the cell (normalized fluorescence) before (middle) and after (right) IP₃ release. The left cell image shows the Fluo3-loaded cell, after IP₃ release. The graph represents changes in Fluo3 fluorescence in a region of interest shown by the square. Results (graph and images) are expressed as changes in fluorescence divided by the resting fluorescence (ΔF/Fo) to normalize the Fluo3 signals with the basal fluorescence. Similar results were obtained in seven cells. (B) IP₃ (top) or Ca²⁺ (bottom) was photoreleased around the nucleus (pink area) from caged IP₃ or Ca²⁺-saturated caged EGTA microinjected together with Ca²⁺ Green into cardiac cells. Note that Ca²⁺ was released to the same extent as the one released by IP₃. The images were acquired using the LSM-510 confocal microscope (confocal section of 0.8 μm). The graphs show the intensity of the Ca²⁺ Green signal after normalization (ΔF/Fo) in different regions of interest (measurement areas). The panel is representative of at least seven experiments. (C) Cardiomyocytes were bathed in a Ca²⁺-free solution containing 1 mM caged ATP. ATP was photoreleased by a UV laser scan. Line scan images were built as described above. The two normalized images on top of the line scan show the cell before (middle) and after (right) ATP uncaging in the surrounding medium. The left cell image shows the Fluo3-loaded cell, after ATP release. The image is representative of at least five experiments. (D) A spontaneously beating cell was stopped by the addition of 10 mM caffeine (Caf) in the presence of 100 μM ryanodine (Ry) and 1 mM EGTA. Then after 1 min, IP₃ was photoreleased. The graph (ΔF/Fo) shows the release of Ca²⁺ induced by IP₃ in the region of interest delimited by the square; the line scan image was built as described above. Inset, the bar graph represents the changes in Fluo3 fluorescence induced by IP₃ (10-ms UV laser scan)_, measured in an ROI including the whole cell, under control conditions (1.8 mM extracellular Ca²⁺) or after Ca²⁺ depletion of the SR (by application of 10 mM caffeine) followed by superfusion of the cell with ryanodine and EGTA (n = 7). Yellow bars, 10 μm.

The purinergic agonist ATP, known to generate IP₃, also induced the release of intracellular Ca²⁺ when quickly photoreleased from its caged precursor in a Ca²⁺-free extracellular environment of quiescent cardiomyocytes. Similarly to IP₃-triggered Ca²⁺ release, the ATP-induced Ca²⁺ release was spatially localized to the perinuclear region without triggering Ca²⁺ oscillations in other cellular domains (Figure 1C). Ca²⁺ release was still observed with the same magnitude (Figure 1D, inset) when IP₃ or ATP was uncaged in or around cells bathed in Ca²⁺-free solution after a fast application of caffeine (10 mM) in the presence of ryanodine (100 μM). This experimental condition was used to deplete Ca²⁺ from the SR and to prevent Ca²⁺ store refilling (Figure 1D). Thus, IP₃ releases Ca²⁺ within a spatially restricted area from a ryanodine- and caffeine-insensitive pool.

IP₃ Arrests Spontaneous Ca²⁺ Spikes

Cardiomyocytes plated at high density exhibited spontaneous beating associated with Ca²⁺ firing after development of pacemaker currents (Gomez et al., 1994; Maltsev et al., 1994). Caged IP₃, microinjected in such cells and photoreleased by a UV laser scan, resulted in a dramatic slowing or transient arrest of Ca²⁺ oscillations at diastolic Ca²⁺ levels (Figure 2A). The effect of IP₃ was dependent on the duration of cell exposure to the uncaging UV light and, thus, to the amount and/or site of released IP₃. A local IP₃ photorelease around the nucleus also dramatically and transiently decreased the rate of Ca²⁺ spiking (Figure 2B). Superfusion of cells with ATP slowed the rate (in 45% of cells) (our unpublished results) or abolished (in 55% of cells) spontaneous Ca²⁺ oscillations (Figure 2C). A similar effect was observed when the related homologue UTP, an agonist selective for purinergic P2Y receptors, was used instead of ATP that binds both P2X and P2Y receptors (our unpublished results). Confocal imaging of a Ca²⁺-oscillating cell revealed that the purinergic agonist first induced transient intracellular Ca²⁺ release, mainly localized around the nucleus, which preceded the arrest of Ca²⁺ oscillations (Figure 2D).

Figure 2.

(A) IP₃ stops or slows Ca²⁺ spiking. Spontaneously beating cardiomyocytes injected with Fluo3 and caged IP₃ were bathed in Ca²⁺-containing solution. IP₃ was photoreleased by scanning the optical field with a UV laser beam (100 ms) that reached the cells at resting [Ca²⁺]_i (first UV exposure) or at the peak of the Ca²⁺ transient (second UV exposure). The panel is representative of six experiments. (B) Spontaneously beating cardiomyocytes injected with Fluo3 and caged IP₃ were bathed in Ca²⁺-containing solution. IP₃ was locally photoreleased around the nucleus by “bleaching” (100 ms) with a UV laser beam the uncaging area. The panel is representative of three experiments. The graphs in A and B areexpressed as ΔF/Fo and represent changes in Fluo3 fluorescence in the ROI (square). (C and D) A spontaneously Ca²⁺-spiking cell loaded with Fluo3 was superfused with 20 μM ATP (indicated by horizontal bar), and Fluo3 fluorescence was monitored with a photomultiplier (C) or by imaging the cell by confocal microscopy (optical section of 0.8 μm; D). The line scan (D, right) was built as described in Figure 1. The graph in D is expressed as ΔF/Fo and represents changes in Fluo3 fluorescence in the ROI (square). Bars, 10 μm.

In principle, an arrest in cardiac automatic activity may result from perturbations in cell excitability and thus in membrane ionic currents. This prompted us to monitor the effect of ATP on the spontaneous firing of action potentials in automatic cardiomyocytes injected with the potential sensitive probe JPW1114 (Antic and Zecevic, 1995). Action potentials recorded with this probe featured a shape and time course similar to the ones recorded using the current-clamp technique (Figure 3A, inset). The purinergic agonist slowed but never stopped spontaneous action potentials as expected from the lack of effect of IP₃ or ATP on Ca²⁺ or K⁺ currents in these cells (our unpublished results). No change in resting membrane potential was observed upon application of ATP (Figure 3A). We then alternatively recorded action potentials and cytosolic Ca²⁺ in the same single cell, microinjected with both JPW1114 and Fluo3. Although ATP stopped repetitive Ca²⁺ spiking, action potentials were still firing although at a lower frequency (Figure 3B, 52 ± 6% of control frequency; n = 8) as observed previously (Figure 3A). This excludes the possibility that a change in membrane potential that may result from a modulation of depolarizing ionic conductances such as the ATP-gated P2X channel (Vassort et al., 1994) could account for the purinergic effect on Ca²⁺ spiking. A direct action on intracellular Ca²⁺ homeostasis is thus more likely to account for the modulation in Ca²⁺ spiking.

Figure 3.

Spontaneous firing of action potentials in neonatal cardiomyocytes. (A and B) Cells were microinjected with JPW1114 (A) or with both JPW1114 and Fluo3 (B). ATP was superfused at a concentration of 20 μM, and changes in fluorescence were recorded by a photomultiplier, digitized, and sampled by a computer. Inset, typical action potentials recorded using JPW1114 or current clamp are shown. The figure is representative of at least six experiments.

Furthermore, we isolated Purkinje cells (Scamps and Carmeliet, 1989) of the cardiac conduction system rich in IP₃Rs (Gorza et al., 1993). Membrane depolarization–induced Ca²⁺ spiking was triggered by external pacing at 1 Hz. Application of ATP dramatically decreased Ca²⁺ transients or stopped Ca²⁺ spiking of these cells (n = 5 cells). This effect was no longer observed in Purkinje cells treated with neomycin or U79122, two inhibitors of PLCs (n = 5 cells) (our unpublished results).

IP₃-induced Ca²⁺ Release Is Required for ATP-induced Arrest of Ca²⁺ Spiking

ATP, unlike most other agonists, generates IP₃ predominantly, if not exclusively, via activation of the tyrosine kinase–regulated PLCγ (Puceat and Vassort, 1996). To disrupt the purinergic signal transduction pathway that leads to IP₃ generation, we microinjected neonatal cardiomyocytes with a blocking anti-PLCγ antibody raised against the SH2 domains of the lipase (Roche et al., 1996). Affinity-purified rabbit IgG or an antibody directed against the yes tyrosine kinase, not expressed in cardiomyocytes, was microinjected as a control antibody. In these cells, application of 20 μM ATP rapidly abolished spontaneous Ca²⁺ spiking as observed in noninjected cells (Figure 4A), indicating that microinjection of antibodies did not affect the purinergic response. In anti-PLCγ antibody–injected cells, however, the purine induced an increase in diastolic Ca²⁺ associated with a slight decrease in the amplitude of Ca²⁺ oscillations, but in 80% of these cells, Ca²⁺ oscillations were still observed, and their frequency was increased (12 out of 15 cells) (Figure 4B). In contrast to ATP, PGF_{2 α} generates IP₃ via activation of the Gq-coupled PLCβ (Adams et al., 1998), an isoform that lacks the SH2 domain. Like ATP, PGF_2α blocked the Ca²⁺ firing of cardiomyocytes, but this was not blocked by the anti-PLCγ antibody raised against the SH2 domain, demonstrating the specificity of the antibody (Figure 4C). A GST fusion protein, composed of the two SH2 domains of PLCγ (at the N- and C-terminals), acts as a “dominant-negative” protein (GSTSH2) (Carroll et al., 1997) when microinjected in cardiomyocytes and prevents PLCγ phosphorylation and thus its activation. The wild-type GSTSH2 prevented ATP from abolishing spontaneous Ca²⁺ spiking in 19 out of 23 microinjected cardiomyocytes, with ATP inducing an increase in diastolic Ca²⁺ in these cells (Figure 4D). In cells microinjected with the mutated GSTSH2 (Carroll et al., 1997), ATP stopped or significantly decreased both the amplitude and the frequency of oscillations (in 20 out of 22 cells) (Figure 4E). In heparin-injected cells in which IP₃ cannot bind its receptor, ATP increased diastolic Ca²⁺ (n = 9) but did not stop spontaneous Ca²⁺ spiking (Figure 4F). A lack of effect of ATP was also observed in cells microinjected with the IP₃-5-phosphatase that accelerates IP₃ hydrolysis or with an anti-Cst1 antibody that prevents activation of the tyrosine kinase–dependent pathway (Puceat et al., 1998) (our unpublished results).

Figure 4.

IP₃ generated by PLCγ is required for ATP-induced arrest of Ca²⁺ spiking. Fluo3 was coinjected in cardiomyocytes with an anti-yes antibody (A), an anti-SH2PLCγ antibody (B and C), a wild-type (wt GSTSH2; D) or mutated (mt GSTSH2; E) glutathione-S-transferase fusion protein, or heparin (F). The effects of ATP (20 μM) or prostaglandin F_2α (PGF_2α; 1 μM in C) on Ca²⁺ oscillations were tested on spontaneously beating cells. [Ca²⁺]_i changes were recorded using a CCD camera and the Argus-50 image processor (Hamamatsu). The horizontal bar below each recording indicates the time scale (1 min). The y-axes of the graphs represent the Fluo3 fluorescence ratio between each image of the series and the first image acquired at resting [Ca²⁺]_i as calculated by the Argus-50 software. Three to four images were acquired per second. The number of experiments performed in each experimental condition is indicated in the text. In this series of experiments, a block in Ca²⁺ oscillations was taken as a readout of the ATP effect because the limited time acquisition of the camera did not allow for an accurate estimation of the spiking rate.

Intracellular Localization and Properties of the IP₃-sensitive Intracellular Ca²⁺ Pool

Immunoprecipitation and Western blotting of IP₃Rs using isoform-specific antibodies revealed that neonatal rat cardiomyocytes express both IP₃R type I and II proteins migrating with an apparent molecular mass of 240 kDa, as expected for IP₃Rs (Mikoshiba et al., 1994). The specificity of the antibodies used was confirmed by competition using respective immunizing peptides (Figure 5A).

Figure 5.

IP₃RIs are localized in a caffeine- and ryanodine-insensitive Ca²⁺ store. (A) IP₃RIs and IP₃RIIs immunoprecipitated from cultured cardiomyocytes, cerebellum, or human embryonic kidney 293 (HEK293) cells using specific antisera. We used 1.5, 0.3, and 1 mg of cardiac whole-cell lysate, cerebellum homogenate, and HEK293 whole-cell lysate proteins, respectively, in this experiment. After immunoprecipitation (IP), SDS-PAGE, and Western blotting (WB), the blots were probed with the respective specific affinity-purified antibodies. The peptides used to raise the antibodies were used in competition assay. Two different anti-IP₃RII antisera were tested, giving the same results. No proteins were immunoprecipitated using the preimmune IP₃RI or IP₃RII antisera. The blot is representative of four experiments. (B) Immunofluorescence labeling and confocal imaging of neonatal rat cardiac cells using an affinity-purified anti-IP₃RI antibody (raised against the N-terminal AA 337–349; I), an anti-calreticulin antibody (II), green mitotracker (III), green mitotracker (green staining) together with an affinity-purified anti-IP₃RI antibody (red staining; IV), or an anti-RyR 2 receptor monoclonal antibody (Affinity Bioreagents, Golden, CO; V). IP₃RI and calreticulin stainings were revealed by a secondary FITC-conjugated anti-rabbit IgG, and RyR was revealed by a TRITC-conjugated anti-mouse IgG except in IV where a TRITC-conjugated anti-rabbit IgG was used to reveal IP₃Rs in green mitotracker-loaded cells. No staining was observed when cells were incubated with only the secondary antibody. A nuclear localization of calreticulin has been observed recently in different cell lines (Roderick et al., 1997). The labeling was observed by confocal microscopy using a Bio-Rad MRC1024 microscope (Richmond, CA) and a 60× magnification objective. The images show a 0.5-μm optical section using a zoom of 1 in III (left), 1.5 in I, II, and III (right), and 2 in IV. Bars, 10 μm.

Confocal micrographs of cells immunostained with a specific anti-IP₃RI antibody revealed labeling around the nucleus that projected as a reticular network toward the cell periphery (Figure 5B, I). This network was however less dense than that revealed by an anti-calreticulin antibody that stains both endoplasmic reticulum (ER) and SR (Figure 5B, II). Staining of cells preincubated with green mitotracker revealed a filamentous (Figure 5B, III, right image) or a more clustered (Figure 5B, III, left image) distribution of mitochondria. Dual staining of cells with both mitotracker and purified anti-IP3RI antibody showed a distribution of IP3Rs very close to the one of mitochondria, specifically around the nucleus (Figure 5B, IV). Immunocytochemistry failed to detect IP₃RII probably because of the low level of expression of this isoform in cardiomyocytes as assessed by immunoprecipitation (Figure 5A). Double immunostaining of cells with both the anti-IP₃RI and anti-RyR antibodies clearly showed that both receptors were not colocalized in cardiomyocytes (Figure 5B, V).

Mechanism of IP₃-induced Abolishment of Spontaneous Ca²⁺ Spiking

A major question is how IP₃-induced Ca²⁺ release may affect CICR originating from the SR in spontaneously Ca²⁺-spiking cells. Recent studies have shown in noncardiac cells a close interaction between the IP₃-sensitive ER and mitochondria (Rizzuto et al., 1993, 1998; Loew et al., 1994; Babcock et al., 1997). Furthermore, these organelles are known to take up Ca²⁺ released on a beat-to-beat basis by the SR in the heart (Chacon et al., 1996; Duchen et al., 1998; Ohata et al., 1998). Parallel experiments showed that Ca²⁺ sequestration in mitochondria using the inhibitor of the transient permeability pore cyclosporin A also slowed the rate and stopped cell Ca²⁺ spiking (Figure 6A). We thus designed experiments to test whether cardiac mitochondria could sequester both Ca²⁺ released by the IP₃-sensitive store and Ca²⁺ cycling from and into the SR, depleting the caffeine-sensitive store (Figure 1A). First, SR Ca²⁺ content was estimated by caffeine-triggered Ca²⁺ release. Consecutive applications of caffeine shortly after stopping Ca²⁺ spiking of cells with a Na⁺- and Ca²⁺-free medium that prevents the activity of the Na⁺/Ca²⁺ exchanger triggered a Ca²⁺ release of similar magnitude (Figure 6B). However, upon stopping Ca²⁺ spiking with extracellular ATP, the magnitude of the caffeine-induced Ca²⁺ release was significantly decreased (Figure 6C, bottom; n = 9). A similar effect was observed when PGF2_α was used instead of ATP to generate cytosolic IP₃ (n = 3 cells) (our unpublished results).

Figure 6.

Ca²⁺ sequestration into mitochondria mimics the IP₃ effect on the rate of spontaneous Ca²⁺ spiking. Purinergic stimulation of cardiomyocytes induces Ca²⁺ depletion of the sarcoplasmic reticulum and mitochondrial Ca²⁺ uptake. (A) A Fluo3-loaded cell was superfused with 50 ng/ml cyclosporin A (CPA). A similar effect of CPA was observed in four other cells. (B) A Fluo3-loaded cell was superfused with a Ca²⁺- and Na⁺-free solution to stop Ca²⁺ oscillations and to prevent the activity of the Na⁺/Ca²⁺ exchanger. Caffeine (10 mM) added to the Ca²⁺- and Na⁺-free solution was applied to the cell. After washing out caffeine, the same protocol was repeated and triggered Ca²⁺ release of a similar amplitude. Similar results were obtained in seven cells. (C) Top, ATP (20 μM) was applied to the cell. Caffeine was then added in Ca²⁺- and Na⁺-free solution. After washout, the same protocol was repeated in the absence of the purine. Bottom, similar results were observed in at least nine cells and gathered in the bar graph (Ca²⁺ content of the SR was estimated from the ratio of the magnitude of the Ca²⁺ peak triggered by caffeine to the magnitude of the averaged Ca²⁺ transients). **, significantly different (p ≤ 0.01).

To determine whether, in cardiomyocytes, mitochondria distributed in the close vicinity of IP₃Rs (Figure 5B I, III, and IV) sense Ca²⁺ released by IP₃, cells were first loaded with Rhod2, a mitochondrial Ca²⁺-sensitive probe. Purinergic stimulation of cells induced a rapid and transient mitochondrial Ca²⁺ uptake (Figure 7A, top left and bottom). These changes were not observed in cells in which ATP-induced IP₃ formation was inhibited by pretreatment for 30 min with 2 mM neomycin (n = 3 cells and 17 mitochondrial clusters), 10 μM U79122 (n = 4 cells and 22 mitochondrial clusters), or 20 μg/ml genistein (n = 3 cells and 21 mitochondrial clusters) or in cells microinjected with a blocking anti-PLCγ antibody (n = 13 cells and 65 mitochondrial clusters) (Figure 7A, top right). Nevertheless, the mitochondrial uncoupler FCCP, which collapses the mitochondrial potential and reverses the activity of the Ca²⁺ uniporter (Kroner, 1992), routinely released Ca²⁺ from the organelles as observed previously in many other cell types (Babcock et al., 1997; Ichas et al., 1997; Boitier et al., 1999; Hajnoczky et al., 1999). FCCP further prevented the ATP-induced mitochondrial Ca²⁺ loading (our unpublished results). Altogether, these data demonstrate the requirement of IP₃ in the purinergic effect on the rate of cell Ca²⁺ spiking. Localized uncaging of IP₃ around the nucleus more directly showed, in Ca²⁺ Green- and Rhod2-loaded cells, that mitochondria pump Ca²⁺ released by IP₃ (Figure 7B). In contrast, global Ca²⁺ release from the SR by caffeine did not trigger any significant increase in Rhod2 fluorescence, whereas ATP applied to the same cell did (Figure 7C).

Figure 7.

Purinergic stimulation of cardiomyocytes transiently increases mitochondrial Ca²⁺. (A) Rhod2-loaded cells were superfused with 20 μM ATP (top left). The increase in Rhod2 fluorescence indicates a concomitant rise of mitochondrial Ca²⁺, whereas a decrease in Rhod2 fluorescence indicates an efflux of Ca²⁺ from mitochondria. The recording is an average of the fluorescence of five mitochondrial clusters around the nuclei of two adjacent beating cells (ROIs delimited by white squares). This average was used to attenuate possible individual changes in the shape of mitochondria. Similar results were observed in 11 cells and expressed in the bar graph (bottom) as the mean ± SEM of ΔF/Fo measured in 43 mitochondrial clusters from 11 cells. The cell was injected with the anti-PLCγ antibody (500 μg/ml of injection buffer) and then loadedwith Rhod2 AM (top right). The recording is an average of the five ROIs delimited by white squares on the cell picture. Three to four images were acquired per second using a CCD camera. (B) Cells were microinjected with caged IP₃ and Ca²⁺ Green and loaded with Rhod2. Cells were imaged by confocal microcopy using a 63× objective. A scanning ROI was set in a small area of the cell that includes the nucleus to get a fast scanning of the UV laser to photorelease IP₃ around the nucleus (pink area). The line scan images were built as described previously (after fluorescence normalization, ΔF/Fo) and established as a line along the nucleus (red line that crosses several clusters of mitochondria). Original fluorescent colors were kept in these experiments. The red fluorescence is generated by Rhod2-loaded mitochondria, whereas the green fluorescence is generated by cytosolic Ca²⁺ Green. The pinhole of the microscope was set to get an optical slice of 2 μm. The yellow fluorescence is generated by an increase in both the mitochondrial and surrounding cytosolic Ca²⁺, around, above, and/or under the nucleus within the 2-μm confocal plane. This focal plane was selected to avoid artifactual changes in focus. Note that leakage or spatial redistribution of mitochondrial Rhod2 is unlikely because the effect of IP₃ on mitochondrial Ca²⁺ was fully reversible. The graph (bottom) represents the changes in Rhod2 and Ca²⁺ Green fluorescence in the ROI (green/red square) expressed as ΔF/Fo. The images on top of the line scan image show the cell before (middle) and after (right) IP₃ release. The figure is representative of seven separate experiments. (C) Caffeine was added to a Rhod2-loaded cell imaged by confocal microscopy (left). Mitochondrial Ca²⁺ was recorded in the absence or presence of 10 mM caffeine and 20 μM ATP in six different clusters of mitochondria (yellow-squared ROIs). The images show the cell before (control) and after ATP or caffeine application. The yellow lines (intermitochondrial cluster distance) were drawn to check the motion of mitochondria in the course of the experiments (using ANALYZE software). Although mitochondria may move within the cell, such a motion was not recorded within the short time course of our experiments. Cell shortening associated with Ca²⁺ spiking was also limited in the z-direction because of the strong attachment of cells on a laminin-coated coverslip. The graph (right) is an average of the change in Rhod2 fluorescence expressed as ΔF/Fo in these six ROIs. The figure is representative of seven separate experiments. (D) A cardiomyocyte expressing a mitochondria-targeted cameleon was stimulated by 20 μM ATP or treated with 10 mM caffeine (caf), 1 μM thapsigargin (thapsi) in the presence of 1 mM external Ca²⁺, or 1 μM ionomycin (iono) in the presence of 5 mM Ca²⁺. The color images were obtained after making the ratio of the image acquired at 535 nm to the image acquired at 480 nm in the absence or presence of ATP, caffeine, or thapsigargin. As a control for the targeting of the cameleon into mitochondria, a cell was challenged with 10 μM FCCP after superfusion with ionomycin and high Ca²⁺. The line scan images were obtained by off-line analysis using ANALYZE software as described in the previous figures; lines were set across clusters of mitochondria as indicated by white arrows. Similar data were obtained in at least six cells. Bars, 10 μm.

To investigate further the privileged transfer of Ca²⁺ from IP₃-sensitive pools to mitochondria, cardiomyocytes were transfected with a mitochondria-targeted cameleon (Miyawaki et al., 1997). Figure 7D shows that the distribution of the cameleon is comparable with the one of Rhod2 and mitotracker. Addition of ATP to a cameleon-expressing cell induced an increase in mitochondrial Ca²⁺ as observed previously using Rhod2. Caffeine slightly decreased mitochondrial Ca²⁺, whereas addition of thapsigargin, a Ca²⁺ ATPase inhibitor that further empties caffeine-insensitive pools, did not affect mitochondrial Ca²⁺ fluxes. Addition of 5 mM Ca²⁺ and ionomycin significantly increased mitochondrial Ca²⁺ (Figure 7D, top). FFCP quickly released Ca²⁺ from mitochondria, confirming the targeting of the cameleon into the organelles (Figure 7D, bottom).

Changes in mitochondrial membrane potential reflect Ca²⁺ movements across the membrane of the organelle (Duchen et al., 1998). In cells loaded with the mitochondrial potentiometric dye TMRM and imaged by confocal microscopy, ATP triggered a transient quenching of the fluorescence, whereas FCCP that leads to a complete dissipation of the mitochondrial potential induced sustained and more pronounced quenching of TMRM fluorescence (Figure 8A). Confocal imaging of cells coloaded with TMRM and Fluo3 showed that depolarization of the mitochondrial membrane occurred just before arrest of the Ca²⁺ oscillations induced by extracellular ATP (Figure 8B).

Figure 8.

Purinergic stimulation of cells induces a mitochondrial depolarization. Cells were loaded with TMRM (A) or with both TMRM and Fluo3 (B), and changes in the fluorescence of the probes were monitored in the absence or presence of 20 μM ATP using confocal microscopy. Images were acquired every 200–300 ms. Line scan images were built as described in Figure 1. Lines were set across clusters of TMRM-labeled mitochondria or across the cell (Fluo3). FCCP (5 μM) was used as a control to induce a complete dissipation of the mitochondrial potential. The figure is representative of five separate experiments. Bars, 10 μm.

DISCUSSION

IP₃ Affects Ca²⁺ Intracellular Homeostasis

We have uncovered that IP_3, acting within a restricted spatial range, modulates spontaneous autonomic activity of cardiomyocytes

Modulation of the rate of cell Ca²⁺spiking may arise either from a change in cell membrane excitability or from a change in the Ca²⁺ content of intracellular stores. Alteration in membrane ion channel conductances by IP₃ generated by the purinergic agonist was discarded because neither the resting membrane potential nor the action potential of cardiomyocytes was affected by purinergic stimulation (Figure 3A). This could be expected from the lack of effect of IP₃ on Ca²⁺ currents (Saeki et al., 1999) or other ionic currents including an iberiotoxin-sensitive Ca²⁺-activated K⁺ current or I_f current (our unpublished results). Furthermore, ATP-induced IP₃ generation stops Ca²⁺-spiking of paced Purkinje cells, which argues against an IP₃ effect on cell membrane excitability. However, when the IP₃ pathway is disrupted, ATP increases diastolic Ca²⁺ and speeds the rate of Ca²⁺ oscillations (i.e., anti-PLCγ antibody-, GSTSH2-, and heparin-injected cells; Figure 4). This observation might be attributed to an IP₃-independent effect of purinergic stimulation that occurs via P2X receptors (Vassort et al., 1994). This effect is masked when IP₃ is generated by ATP, which further suggests that the P2Y receptor is preferentially activated as reported in transfected cells (Murthy and Makhlouf, 1998). Thus, the effect of IP₃ on autonomic Ca²⁺ spiking can be attributed to a perturbation in intracellular Ca²⁺ homeostasis rather than to a direct change of cell excitability.

The IP₃-sensitive Ca²⁺ Pool: An ER Functionally Distinct from the SR

The main question addressed in this study was the origin of Ca²⁺ released by IP₃. Several findings point to the ER but not to the SR as the IP₃-sensitive Ca²⁺ store. First, spatially restricted IP₃-induced Ca²⁺ release is not sensitive to ryanodine or caffeine. Second, IP₃Rs, but not RyRs, feature a perinuclear distribution. Third, perinuclear release of Ca²⁺ triggers a CICR, whereas IP₃ induces a transient single spike (Figure 1B). Furthermore, the kinetics of IP₃-triggered Ca²⁺ release is slower than what would be expected from an SR Ca²⁺ release (Robert et al., 1998) but faster than what may be expected from the Golgi apparatus, a compartment shown recently to be responsive to IP₃ (Pinton et al., 1998; Lin et al., 1999). Finally, ATP or IP₃, but not caffeine, triggered a localized Ca²⁺ release that was sensed by mitochondria. This also excludes the SR as the IP₃-sensitive Ca²⁺ store. Neonatal rat cardiac cells in culture thus feature an IP₃-sensitive ER pharmacologically and functionally distinct from the SR, as suggested previously for smooth muscle cells (Golovina and Blaustein, 1997) and for cells of the cardiac conduction system (Gorza et al., 1993). Although IP3Rs and RyR do not colocalize (Figure 5), our findings cannot exclude the possibility that the ER and the SR may be two subcompartments of the same store, namely, an ER/SR network

Privileged Ca²⁺ Flow between Mitochondria and ER, a Major Component of the IP₃ Effect on Spontaneous Ca²⁺ Spiking

Despite the absence of IP₃Rs in the SR and the caffeine-insensitive IP₃-dependent Ca²⁺ release, purinergic stimulation slows or stops spontaneous Ca²⁺ spiking after a partial but significant Ca²⁺ depletion of the SR (Figure 6). Participation of mitochondria in the Ca²⁺ homeostasis network (Babcock et al., 1997) as a Ca²⁺ sink, which under specific conditions sequesters Ca²⁺ released by either ER or both ER and SR, may account for this observation.

Mitochondrial Ca²⁺ uptake is a low-affinity (Gunter et al., 1994; Csordas et al., 1999) but a fast process (Sparagna et al., 1995). Evidence supporting the participation of mitochondria in the transient arrest of Ca²⁺ spiking in cardiomyocytes emerges from a series of observations. First, an increase in Rhod2 fluorescence was observed when IP₃ was photoreleased around the nucleus or was generated by purinergic stimulation of cells (Figure 7). Ca²⁺ sequestration into mitochondria upon an IP₃ challenge was further demonstrated in cells expressing a mitochondria-targeted cameleon (Figure 7D). Second, depolarization of mitochondria clustered around the nucleus in the vicinity of the ER coincides with the arrest in cytosolic Ca²⁺ oscillations, both triggered by ATP (Figure 8). Finally, cyclosporin A, which elicits a Ca²⁺ load of mitochondria by preventing Ca²⁺ efflux from organelles, also stops cell autonomic Ca²⁺ spiking. Our findings obtained using three Ca²⁺ or potential-sensitive probes in both quiescent and spontaneously spiking cells strongly suggest that IP₃ is able to create microdomains of high Ca²⁺ concentration around neighboring mitochondria. This event triggers a Ca²⁺ load and a depolarization of the organelles because Ca²⁺ around the mitochondrial uniporter reached the concentration required to open it (Rizzuto et al., 1993, 1998). Ca²⁺ microdomains are more directly revealed in cardiac cells by the experiments using caffeine in Rhod2-loaded cells or in myocytes transfected with a mitochondria-targeted cameleon (Figure 7, C and D). Indeed, in contrast to focal release from the SR (Duchen et al., 1998), a global cytosolic Ca²⁺ increase after caffeine addition to the cells in normal Na⁺- and Ca²⁺-containing extracellular buffer was not sensed by mitochondria. However, the organelles took up Ca²⁺ locally released by ATP or IP₃ from the caffeine-insensitive ER. This is in line with previous observations in cardiac cells in which caffeine induced a mitochondrial Ca²⁺ load only in Na⁺- and Ca²⁺-free medium, a condition that prevents activity of the Na⁺/Ca²⁺ exchanger, the main Ca²⁺-extruding mechanism in these cells (Bassani et al., 1993). Ca²⁺ concentration can reach 100 μM close to the IP₃Rs (20-nm distance), a value 10-fold higher than the K_m of the mitochondrial Ca²⁺ uniporter, and can go down to 5–10 μM at a 200-nm distance as calculated for a voltage-gated channel (Neher, 1998). Thus, our findings suggest that mitochondria do not face the SR Ca²⁺-releasing sites and that Ca²⁺ microdomains are not generated around them. In contrast, ER Ca²⁺-releasing sites (i.e., IP₃Rs) are more likely to be in close connection with mitochondria (Figure 5B) in cardiomyocytes like in any other cell types because we also observed that IP₃ but not thapsigargin triggered a significant mitochondrial Ca²⁺ loading. In agreement with our observations in cardiomyocytes, thapsigargin or 2,5-di-(t-butyl)-1,4-hydroquinone (tBuBHQ) did not affect mitochondrial Ca²⁺ in RBL-2H3 (Csordas et al., 1999) or in the MH75 cell line (Rizzuto et al., 1994). However, BHQ and thapsigargin induced a Ca²⁺ uptake by mitochondria in chromaffin cells (Babcock et al., 1997) and in hepatocytes, respectively (Hajnoczky et al., 1995). This apparent discrepancy may be related to a different structure and/or spatial localization of the ER depending on the cell geometry. Together with the spatial distribution of IP₃Rs, primarily clustered, like mitochondria, around the nucleus, mitochondrial Ca²⁺ uptake is likely to account for the limited diffusion of Ca²⁺ locally released by IP₃ (Figure 1) and for the inability of IP₃ to trigger CICR from the SR. This further supports the diffusion-limiting role of mitochondria in IP₃-dependent Ca²⁺ signals as observed in pancreatic cells (Tinel et al., 1999). Ca²⁺ is then released from the organelles after depolarization of mitochondria (Sparagna et al., 1995; Massari, 1996; Ichas et al., 1997; Ichas and Mazat, 1998; Huser et al., 1998) (Figure 7). Depending on the Ca²⁺-loading state of mitochondria, their Ca²⁺ content can go below its initial value after IP₃-triggered Ca²⁺ uptake (Figure 7A) (Ichas et al., 1997). This release prevents mitochondrial Ca²⁺ overload and refills the SR to regenerate Ca²⁺ transients.

We thus propose that part of the Ca²⁺ released from the SR with each beat flows into mitochondria through the Ca²⁺ uniporter (Bassani et al., 1993; Duchen et al., 1998; Ohata et al., 1998; Zhou et al., 1998) as the later is switched on by Ca²⁺ microdomains generated by IP₃. Alternatively, mitochondria may sequester only Ca²⁺ released by IP₃ from the ER, and Ca²⁺ could flow from the SR into the ER through a privileged communication between both stores (Figure 9). This would explain why the IP_3-induced Ca²⁺ release is faster in Ca²⁺ spiking (Figure 2) than in quiescent cells (Figure 1). Under these circumstances, the SR would also be transiently Ca²⁺ depleted, resulting in a slowing or arrest in cell Ca²⁺ spiking. To discriminate between these mechanisms further, we emptied mitochondrial Ca²⁺ or prevented efflux by using mitochondrial uncouplers (FCCP together with oligomycin) or a blocker of the permeability transition pore (cyclosporin A), respectively. These drugs however induced a large cytosolic Ca²⁺ transient that irreversibly blocks Ca²⁺ spiking (FCCP) or, like IP₃, slows (cyclosporin A; Figure 6A) spontaneous Ca²⁺ spiking of cardiomyocytes. Despite the fact that this prevented us from further testing the effect of IP₃ or ATP on these cells, the latter observation further supports the hypothesis that mitochondrial Ca²⁺sequestration, whatever its origin, leads to a slowing of cytosolic Ca²⁺spiking. Thus, these findings also provide evidence of a major role of mitochondria in closely regulating SR Ca²⁺ cycling. Because the firing of an action potential was slowed by ATP, it further suggests that Ca²⁺-induced Ca²⁺ release closely modulates the generation of action potentials in these cardiomyocytes and in turn cellular automaticity, as it has been reported in pacemaker cells of the sinoatrial node (Ju and Allen, 1999).

Figure 9.

Diagram of the mechanism of IP₃-induced modulation of cytosolic Ca²⁺ oscillations: an integrated control of the filling state of the ER/SR and mitochondria. IP₃-induced Ca²⁺ release from the ER generates a high Ca²⁺ concentration in the vicinity of mitochondria. Thereby, mitochondria take up Ca²⁺ released by the IP₃-sensitive ER and trap part of the Ca²⁺cycling from and into the SR through the uniporter. Alternatively, the SR may be in privileged communication with and may provide the ER with Ca²⁺ to refill after mitochondrial Ca²⁺ sequestration. In both situations, the SR is partially Ca²⁺depleted, and in turn, cytosolic Ca²⁺ oscillations are slowed or stopped.

Conclusions

We report that IP₃ and mitochondria are key components in the neurohumoral regulation of autonomic activity of neonatal rat cardiac cells, a reliable model of cellular automaticity. We bring a novel function of IP₃, as a modulator of cardiac rhythmic and autonomic activity. Although several IP₃-generating neurohormones including α₁-adrenergic, muscarinic, and purinergic agonists (Rosen et al., 1988, 1990; Takikawa et al., 1990; Terzic et al., 1993) regulate cardiac rhythm, no direct proof of a critical role of IP₃ in gating cardiac rhythmic activity has been provided to date. The present findings show that in a beating cardiomyocyte, rhythmic Ca²⁺ oscillations are transiently blocked or slowed down while action potential firings are slowed down by an IP₃-induced spatially restricted Ca²⁺ release. Such a property of IP₃ may prove essential in preventing deleterious rhythmic accelerations (Hauswirth et al., 1968), which could disrupt synchronous activity of the myocardium and provides a mechanistic basis for the previously established antiarrhythmic property of ATP. Thus, in addition to the CICR that is essential for cardiac excitation–contraction coupling, IP₃-triggered Ca²⁺ release may provide a previously unrecognized component in the regulation of cardiac rhythm of particular significance under physiological and pathological conditions associated with upregulated IP₃Rs (Marks, 1997). Furthermore, our findings implicate both an IP₃-sensitive Ca²⁺ store and mitochondria in regulating a major cell function. Such a coordinated mechanism may apply to other cell functions in excitable or nonexcitable tissues. These include the modulation of intracellular Ca²⁺ waves, as reported recently in astrocytes (Boitier et al., 1999), the regulation of gene expression by Ca²⁺ spike frequency (Li et al., 1998), and the excitability of neurons that also feature distinct ER and SR.