Role of ATP-conductive anion channel in ATP release from neonatal rat cardiomyocytes in ischaemic or hypoxic conditions

Abstract

It is known that the level of ATP in the interstitial spaces within the heart during ischaemia or hypoxia is elevated due to its release from a number of cell types, including cardiomyocytes. However, the mechanism by which ATP is released from these myocytes is not known. In this study, we examined a possible involvement of the ATP-conductive maxi-anion channel in ATP release from neonatal rat cardiomyocytes in primary culture upon ischaemic, hypoxic or hypotonic stimulation. Using a luciferin–luciferase assay, it was found that ATP was released into the bulk solution when the cells were subjected to chemical ischaemia, hypoxia or hypotonic stress. The swelling-induced ATP release was inhibited by the carboxylate-and stilbene-derivative anion channel blockers, arachidonic acid and Gd³⁺, but not by glibenclamide. The local concentration of ATP released near the cell surface of a single cardiomyocyte, measured by a biosensor technique, was found to exceed the micromolar level. Patch-clamp studies showed that ischaemia, hypoxia or hypotonic stimulation induced the activation of single-channel events with a large unitary conductance (∼390 pS). The channel was selective to anions and showed significant permeability to ATP⁴⁻ (P_ATP/P_Cl ∼ 0.1) and MgATP²⁻ (P_ATP/P_Cl ∼ 0.16). The channel activity exhibited pharmacological properties essentially identical to those of ATP release. These results indicate that neonatal rat cardiomyocytes respond to ischaemia, hypoxia or hypotonic stimulation with ATP release via maxi-anion channels.

It is known that ATP is released into the interstitial spaces within the heart in response to a variety of physiological and pathophysiological stimuli (Vassort, 2001) including hypoxia (Paddle & Burnstock, 1974; Forrester & Williams, 1977; Clemens & Forrester, 1981; Vial et al. 1987; Borst & Schrader, 1991), ischaemia or ischaemic pre-conditioning (Borst & Schrader, 1991; Kuzmin et al. 1998, 2000; Ninomiya et al. 2002), mechanical stretching (Uozumi et al. 1998) and stimulation by catecholamines (Darius et al. 1987; Vial et al. 1987; Borst & Schrader, 1991; Hall et al. 1995). Forrester & Williams (1977) showed that brief hypoxia induces ATP release from isolated rat ventricular myocytes. In the whole heart, ATP from other sources such as endothelial cells may also contribute to the ATP release (Pearson & Gordon, 1985; Vassort, 2001). However, the mechanism by which ATP is transported across the cell membrane is not understood.

Recently, our studies have demonstrated that a maxi-anion channel with a single-channel conductance of around 400 pS, called the volume- and voltage-dependent ATP-conductive large-conductance anion channel (VDACL) (Sabirov et al. 2001; Sabirov & Okada, 2004), serves as a pathway for ATP release from C127 mammary cells under hypotonic conditions (Sabirov et al. 2001; Dutta et al. 2002) and from kidney macula densa cells under an NaCl load (Bell et al. 2003). In a previous study by another group (Coulombe & Coraboeuf, 1992), a similar maxi-anion channel activity was observed in newborn rat cardiomyocytes exposed to a hypotonic solution. The question therefore arose as to whether the cardiac maxi-anion channel is conductive to ATP and whether it is activated by hypotonic stimulation, and possibly by ischaemic or hypoxic stimulation. In the present study, we addressed this question using neonatal rat cardiomyocytes in primary culture, which have long been studied as a model for cellular ischaemia/hypoxia (van der Laarse et al. 1979; Tanaka et al. 1994; Long et al. 1997; Mackay & Mochly Rosen, 1999; Schaffer et al. 2000; Adachi et al. 2001).

Methods

Cells

The experimental protocol was approved in advance by the Ethics Review Committee for Animal Experimentation of the National Institute for Physiological Sciences. Neonatal rat cardiomyocytes were prepared from ventricles isolated from 2- to 3-day-old Wistar rats (supplied from Japan SLC, Shizuoka Laboratory Animal Center) after decapitation, according to the method previously described (Simpson & Savion, 1982). Cells were cultured in M199 medium supplemented with 10% newborn bovine serum in 35 mm culture dishes for 1 h. Non-adherent cells (mostly cardiomyocytes) were plated on collagen-coated glass coverslips and then cultured for 2 days. Patch-clamp experiments and ATP release measurements were performed when the cell density reached 4 × 10³ cm⁻² and 6 × 10⁴ cm⁻², respectively.

PC12 cells were obtained from Riken Cell Bank (Tsukuba, Japan), cultured in DMEM supplemented with 10% FCS, and used for patch-clamp experiments without the induction of neuronal differentiation.

Luciferin–luciferase ATP assay

The extracellular ATP concentration was measured by a luciferin–luciferase assay (ATP Luminescence Kit; AF-2L1, DKK-TOA, Tokyo, Japan), as previously described (Hazama et al. 1999, 2000; Sabirov et al. 2001), with slight modification. Briefly, neonatal rat cardiomyocytes cultured on coverslips were superfused with control solution in a perfusion chamber, at a rate of 1.5 ml min⁻¹ (to minimize the effect of shear stress on ATP release), at room temperature. The superfusate was collected every minute for measurements of released ATP. After a steady-state level of ATP was attained in control conditions, control solution was replaced with a given test solution. In order to minimize the effect of salt concentration changes on the luciferin–luciferase reaction (Boudreault & Grygorczyk, 2002), the tonicity of the solution was changed by adding or removing mannitol. An aliquot (500 μl) of perfusate was mixed with 50 μl of a luciferin–luciferase assay mixture for luminometric ATP measurements. Since Gd³⁺ has been reported to interfere with the luciferase reaction (Boudreault & Grygorczyk, 2002), we supplemented the luciferin-luciferase assay mixture with 600 μmol l⁻¹ of EDTA when the sample perfusate contained Gd³⁺. Other drugs employed in the present study had no significant effect on the luciferin–luciferase reaction.

Detection of ATP release by a biosensor technique

The biosensor method originally established by Hazama et al. (1998) was employed to measure the local concentration of ATP released from a single cardiomyocyte. Whole-cell currents were recorded from a PC12 cell, which expresses P2X receptor channels, at a holding potential of −50 mV, before and after positioning it very closely to a cardiomyocyte. The current responses were observed upon changing the bath solution to a hypotonic, ischaemic or hypoxic solution. For calibration experiments, ATP was applied to a PC12 cell without positioning it near a cardiomyocyte, through a micropipette filled with an ATP-containing bath solution. The ATP-induced currents were detectable at concentrations greater than 1 μmol l⁻¹ and the half-maximal concentration was 30.4 ± 2.9 μmol l⁻¹, as previously reported (Hazama et al. 1998, 1999; Bell et al. 2003).

Patch-clamp recordings in cardiomyocytes

Patch electrodes, fabricated from borosilicate glass using a micropipette puller (P-97, Sutter Instruments), had a tip resistance of about 2 MΩ for whole-cell current measurements and 2–5 MΩ for macro-patch and single-channel recordings when filled with pipette solution. For whole-cell recordings, the access resistance did not exceed 5 MΩ and was always compensated for (by 70–80%). Membrane currents were measured with an Axopatch 200 A patch-clamp amplifier coupled to a DigiData 1322 A interface (Axon Instruments). Currents were filtered at 1 kHz and sampled at 2–5 kHz. Data acquisition and analysis were done using pCLAMP 8.1 (Axon Instruments) and WinASCD software (kindly provided by Dr G. Droogmans, KU Leuven, Belgium). Whenever the bath chloride concentration was changed, a salt bridge containing 3 mol l⁻¹ KCl in 2% agarose was used to minimize bath electrode potential variations. Liquid junction potentials were calculated using pCLAMP 8.1 algorithms and corrected when necessary.

Solutions and chemicals

The standard Ringer solution contained (mm): 135 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 5 Na-Hepes, 6 Hepes, and 5 glucose (pH 7.4, 290 mosmol (kg H₂O)⁻¹). To prevent cell contraction, we omitted Ca²⁺ ions from this solution in most patch-clamp experiments. Hypotonic bath solution was made by reducing the concentration of NaCl in this nominally Ca²⁺-free Ringer solution to 100 mm (215 mosmol (kg H₂O)⁻¹). The isotonic bath solution for cell-attached experiments was made from the hypotonic solution by adding an appropriate amount of mannitol. In inside-out and outside-out experiments we used standard Ringer solution in the bath. For selectivity measurements, both NaCl and KCl in the Ringer solution were replaced with NMDG-Cl, or NaCl was replaced with sodium glutamate.

The pipette solution used for outside-out experiments in cardiomyocytes as well as for whole-cell recordings in biosensor PC12 cells contained (mm): 125 CsCl, 2 CaCl₂, 1 MgCl₂, 5 Hepes (pH 7.4 adjusted with CsOH), and 10 EGTA (pCa 7.6; 275 mosmol (kg H₂O)⁻¹). The pipette solution for inside-out experiments was standard Ringer solution. In some experiments, CsCl was replaced with TEA-Cl in the pipette solution. For ATP⁴⁻ current measurements, a 100 mm Na₂ATP solution with the pH adjusted to 7.4 with NaOH was used as the bath solution. For MgATP²⁻ current measurements, a 100 mm Na₂ATP solution (without pH adjustment) was combined with Mg(OH)₂ powder to yield 100 mm Na₂MgATP solution. The pH of this solution was close to neutral; when necessary, the pH was adjusted to 7.4 with NaOH. All ATP-containing solutions were kept on ice and warmed to room temperature immediately before each experiment. For whole-cell current measurements in cardiomyocytes, the isotonic pipette and bath solution contained (mm): 100 NMDG-Cl, 1 MgCl₂, 6 Hepes, 5 Na-Hepes, 5 glucose (pH 7.4, 290 mosmol (kg H₂O)⁻¹ adjusted with mannitol). Hypotonic bath solution was made by omitting mannitol in the isotonic solution.

For chemical ischaemia experiments, 10 mm 2-deoxy-d-glucose (2DG) and 5 mm NaCN were added to control solution devoid of glucose (pH was adjusted to 7.4). For hypoxia experiments, bath Ringer solution was continuously bubbled with 100% argon gas for 1 h before and during experiments. The oxygen concentration (P_O2) measured using an oxygen sensor (LICOX A3R, Mielkendorf, Germany) was 102.9 ± 1.2 mmHg (n = 4) for standard Ringer solution exposed to air in the experimental perfusion chamber. In hypoxic conditions, the oxygen concentration in the perfusate solution decreased, with a half-time of 3.8 ± 0.7 min, to stable values of 3.7 ± 0.5 and 30.2 ± 2.5 mmHg (n = 4) at the inlet and outlet, respectively, of the perfusion chamber. The oxygen concentration readily recovered upon returning to the normal solution.

GdCl₃ was stored as a 50 mm stock solution in water and added directly to Ringer solution immediately before each experiment. 5-Nitro-2-(phenylpropylamino)-benzoate (NPPB), glibenclamide, 4-acetamino-4′-isothiocyanostilbene (SITS), arachidonic acid, indomethacin, nordihydroguaiaretic acid (NDGA), clotrimazole and octanol were purchased from Sigma-Aldrich and were added to Ringer solution immediately before use from stock solutions in DMSO. DMSO did not have any effect, when added alone at a concentration less than 0.1%.

Osmolality of all solutions was measured using a freezing-point depression osmometer (OM802, Vogel, Kevelaer, Germany).

Data analysis

Single-channel amplitudes were measured by manually placing a cursor at the open and closed channel levels. Mean patch currents were measured at the beginning (first 25–30 ms) of current traces in order to minimize the contribution of voltage-dependent current inactivation.

Permeability ratios for different monovalent anions (X) were calculated from the Goldman-Hodgkin-Katz (GHK) equation:

(1)

where ΔE_rev is the reversal potential in the presence of a given test anion at a concentration of [X]_i in the inside-out mode, [Cl]_o is the Cl⁻ concentration in the pipette (standard Ringer solution), and [Cl]_i is the Cl⁻ concentration in low Cl⁻ bath solutions containing different test anions. P_Cl and P_X are the permeability coefficients of Cl⁻ and the test anion, respectively.

The permeability ratio for ATP to Cl⁻ (P_A/P_Cl) was calculated from the GHK equation:

(2)

where α=F/RT; Z_A and Z_Cl are the valences of ATP and Cl⁻, respectively; [Cl]_o and [Cl]_i are the Cl⁻ concentrations in the pipette and in the bath, respectively; [A]_i is the ATP concentration in the bath; and E_rev is the reversal potential. When no Cl⁻ is present in the bath, and therefore [Cl]_i= 0, the equation simplifies to the one used by Cantiello et al. (1997). The calculated value for valence of ATP at pH 7.4 was Z_A=−3.9 in the case of free nucleotide (referred to as ATP⁴⁻ in the following text) and Z_A=−1.89 in the case of complex with Mg²⁺ ion (referred to as MgATP²⁻ in the following text). All the constants used for charge calculations were taken from Sigel (1987).

Data were analysed in Origin 6 and OriginPro 7.0 (OriginLab Corp., Northampton, MA, USA). Pooled data are given as means ±s.e.m. of observations (n). Statistical differences of the data were evaluated by ANOVA and paired or unpaired Student's t test where appropriate and considered significant at P < 0.05.

In all figures, the membrane potential (V_m) is indicated according to the following convention:V_m=V_p (the pipette potential) for whole-cell and outside-out experiments, and V_m=−V_p for inside-out experiments. For on-cell records, applied voltages represent −V_p values.

All experiments were performed at room temperature (20–24°C).

Results

Hypotonic, ischaemic or hypoxic stimulation induces ATP release from neonatal cardiomyocytes

The basal ATP concentration of isotonic perfusate from around (7.5 ± 1.0) × 10⁴ cardiac cells was measured to be 22.7 ± 4.2 pm (n = 13) by a luciferin–luciferase assay. The perfusate ATP concentration dramatically increased upon switching the superfusing solution (while maintaining a constant rate of fluid flow) from isotonic to hypotonic solution (74% osmolality), as shown in Fig. 1Aa. After reaching a transient peak with a lag period of 2.5 ± 0.2 min, the ATP concentration rapidly dropped down to a sustained level of around three times of the isotonic level. After restoring perfusate tonicity to a normal level, the bulk ATP concentration returned to the basal level.

Figure 1. ATP release from neonatal rat cardiomyocytes induced by hypotonic (A), ischaemic (B) and hypoxic (C) stress.

The bulk ATP concentration was measured by luciferin–luciferase assay and was plotted directly (in Ab) or after normalizing by the mean basal concentration (in Aa, B and C). A, time course (a) and pharmacological profile (b) of swelling-induced ATP release under hypotonic (74% osmolality) conditions. NPPB (100 μm), SITS (100 μm), arachidonic acid (20 μm), Gd³⁺ (50 μm) and glibenclamide (200 μm) were added to hypotonic solution and their effects observed. B, time course of ATP release induced by chemical ischaemia (10 mm 2DG and 5 mm NaCN). C, time course of ATP release induced by hypoxia (perfusion solution bubbled with 100% argon gas). Each data point represents the mean ±s.e.m.*Significantly different from the control value (in Ab) or the basal level (in Aa, B and C) at P < 0.05.

As shown in Fig. 1Ab, conventional anion channel blockers (NPPB, SITS) potently suppressed ATP release, whereas glibenclamide was ineffective. Arachidonic acid prominently blocked ATP release when a cocktail of oxygenase inhibitors (NDGA + indomethacin + clotrimazole, 20 μm each) was simultaneously added. The suppressive effect of arachidonic acid alone was significant (around 50% suppression) but less prominent (data not shown, n = 4), suggesting that oxygenase-mediated arachidonate metabolism took place in the intact cells. The cocktail itself did not have any effect on ATP release. Gd³⁺ also exhibited a marked blocking effect on ATP release. In contrast, octanol, which is a known blocker of gap junction hemichannels, failed to affect swelling-induced ATP release from cardiomyocytes at the concentration of 1 mm (data not shown, n = 8).

When the cells were subjected to ischaemia by applying an isotonic glucose-free solution containing a glycolysis inhibitor, 2DG (10 mm), and a blocker of mitochondrial oxidative phosphorylation, NaCN (5 mm), the perfusate ATP started to increase with a lag period of 2.8 ± 0.5 min and thereafter exhibited oscillatory increases in each experiment, though the peak level was not as prominent as that observed upon a hypotonic challenge. The averaged ATP level of five experiments also exhibited oscillatory responses to chemical ischaemia, as shown in Fig. 1B. After the cells were superfused with normal glucose-containing Ringer solution, the concentration of ATP was partially restored to its initial level.

The ATP release in response to perfusion with hypoxic solution was also oscillatory and more prominent than that of chemical ischaemia, as shown in Fig. 1C. The average time required to reach the first peak of ATP release was 5.3 ± 0.2 min. Restoration of perfusate P_O2 to a normal level led to a recovery toward the basal level of the perfusate ATP concentration.

In order to detect ATP release from cardiomyocytes at the single-cell level, we employed a PC12-cell biosensor technique. As shown in Fig. 2Aa and Ab, a series of inward current spikes was recorded from a PC12 cell positioned close to a cardiomyocyte, upon exposure to a hypotonic solution after a lag period of 4.7 ± 1.1 min (n = 10). In contrast, no inward current spikes were observed for a PC12 cell alone, exposed to a hypotonic solution but not positioned near a cardiomyocyte (Fig. 2Ac). Inward current responses were suppressed when the hypotonic solution was supplemented with a P2-receptor blocker (suramin, 300 μm) or an ATP-hydrolysing enzyme (apyrase, 0.1 mg ml⁻¹), as shown in the traces in Fig. 2Aa and Ab. Figure 2Ad summarizes the mean peak current density before and during application of suramin or apyrase. When suramin or apyrase was applied, starting before and lasting into the hypotonic challenge, the hypotonicity-induced current response was completely abolished (data not shown, n = 3 and 5, respectively).

Figure 2. ATP release from single cardiomyocytes under hypotonic (A), ischaemic (B) or hypoxic (C) conditions, detected at the cell surface by the biosensor technique.

a–c, representative current traces. d, diagrams summarizing mean peak current responses. Currents were recorded from a PC12 cell placed close to a cardiomyocyte, before and during (horizontal bar) application of hypotonic (A), ischaemic (B) or hypoxic (C) stress. Addition of suramin (300 μm) or apyrase (0.1 mg ml⁻¹) is indicated. In Ac, currents were recorded from a PC12 cell without positioning it near a cardiomyocyte. In Bc, currents were recorded from a PC12 cell positioned near a cardiomyocyte in control solution without applying any stress. *Significantly different from the control value recorded in the absence of suramin and apyrase at P < 0.05.

Using the same biosensor technique, inward current responses were also recorded 4.3 ± 0.5 min (n = 8) after application of chemical ischaemia solution (Fig. 2Ba and Bb) but were never detected when ischaemic stimulation was not applied (Fig. 2Bc). Again, the current response was eliminated by addition of suramin or apyrase (Fig. 2Ba and Bb), as summarized in Fig. 2Bd. Pre-treatment with suramin or apyrase inhibited the occurrence of chemical ischaemia-induced current responses (data not shown, n = 5 or 6).

Similar spiky current responses were recorded under hypoxic conditions after a lag period of 6.5 ± 2.0 min (n = 6) (Fig. 2Ca and Cd). However, hypoxia-induced responses were never observed when suramin or apyrase was present in the hypoxic solution (Fig. 2Cb to Cd).

These results indicate that single cardiomyocytes respond to hypotonic, ischaemic or hypoxic stimulation with ATP release. Using a calibration curve the local ATP concentration could be assessed from the mean peak current density of the P2X receptor-based biosensor response (Fig. 2Aa, Ba and Cd). The ATP concentrations thus estimated are 22, 27 and 10 μm at the surface membrane of cardiomyocytes subjected to osmotic swelling, chemical ischaemia and hypoxia, respectively.

Hypotonic, ischaemic or hypoxic stimulation induces activation of maxi-anion channels

In the cell-attached (on-cell) configuration, no single-channel events were observed from cardiomyocytes perfused with isotonic solution. After excision of the patch membrane from cardiomyocytes bathed in isotonic solution, however, single-channel events with a large amplitude were observed. As shown in Fig. 3A, the macro-patch current in excised inside-out mode exhibited voltage-dependent inactivation at both positive and negative potentials greater than ±20 mV. The inactivation time course became more rapid at larger potentials. Figure 3B shows representative single-channel events recorded at ±50 mV from non-macro inside-out patches. Time-dependent inactivation of large conductance unitary events could be seen clearly (Fig. 3Ba). Sub-state events of approximately half-amplitude were sometimes seen between the full-amplitude events. In some cases (10 of 50 patches tested), stable half-amplitude events were also observed (Fig. 3Bb). The unitary current–voltage (I–V) relationship of full-amplitude events was linear with a unitary conductance of 394 ± 6 pS and reversed at around 0 mV, as shown in Fig. 3C (open circles). Neither the shape of the unitary I–V relationship nor the reversal potential altered when monovalent cations in the bath were replaced with NMDG⁺ (filled circles) or those in the pipette solution with TEA⁺ (filled squares). In contrast, replacement of Cl⁻ with glutamate⁻ in the bath solution shifted the reversal potential to a value of −33.8 ± 1.5 mV (Fig. 3D). These results indicate that the full-amplitude channel is anion selective with a permeability ratio of glutamate⁻ to Cl⁻ of 0.20 ± 0.02. The half-amplitude events exhibited a linear I–V relationship with a unitary conductance of 193 ± 3 pS, anion selectivity and a permeability ratio P_glutamate/P_Cl of 0.22 ± 0.03 (n = 3–10).

Figure 3. Biophysical properties of maxi-anion channel currents activated after excision of patch membranes from neonatal rat cardiomyocytes.

A, voltage-dependent inactivation of currents recorded from a macro-patch. Current responses were recorded upon application of step pulses (protocol shown in the top panel). B, voltage-dependent inactivation of single-channel currents exhibiting full-amplitude (a) and half-amplitude (b) events occurring in response to step pulses (protocol shown in the top panel) recorded from ordinary (non-macro) inside-out patches. C, lack of dependence on cations of the single-channel I–V relationship for full-amplitude events in inside-out patches. Currents were recorded in control conditions (open circles: standard Ringer solution in the bath and pipette), after substitution of monovalent cations in the pipette solution with TEA (filled squares) or after substitution of monovalent cations in the bath solution with NMDG (filled circles). Each symbol represents the mean value with s.e.m. (bar). D, anion selectivity of the channel, as seen in the single-channel I–V relationship for full-amplitude events in inside-out patches. Currents were recorded in control conditions (open circles) or after substitution of Cl⁻ in the bath solution with glutamate (filled circles). Each symbol represents the mean value with s.e.m. (vertical bar).

When the isotonic bath solution was replaced with hypotonic solution, even in the giga-sealed cell-attached configuration, we could consistently observe single-channel events with a large amplitude after a lag period of 5.7 ± 2.1 min (n = 11), as shown in Fig. 4Aa. Together with full-amplitude events, half-amplitude events could also be observed in 4 of 11 patches tested (data not shown). The mean amplitudes of full-amplitude events were 11.7 ± 0.8 pA (n = 8) and −5.2 ± 0.5 pA (n = 10) at the holding potential (−Vp) of +25 and −25 mV, respectively (Fig. 4Ab). The channels underwent apparent inactivation when subjected to larger depolarizing or hyperpolarizing potentials (data not shown). The unitary I–V relationship for full-amplitude events exhibited slightly outward rectification (Fig. 4Ac), presumably due to a lower Cl⁻ concentration within the cell. The mean slope conductances were 359 ± 12 and 247 ± 20 pS at positive and negative potentials, respectively.

Figure 4. Maxi-anion channel activity induced by hypotonic (A), ischaemic (B) and hypoxic (C) stress in neonatal rat cardiomyocytes.

a, representative mean patch currents during application of alternating pulses from 0 to ±25 mV (every 10 s) in a cell-attached patch before and during (horizontal bar) exposure to hypotonic, ischaemic or hypoxic solution. b, representative current traces of full-amplitude events occurring in response to step pulses of ±25 mV (protocol shown at the top of traces) in cell-attached or inside-out patches. Half-amplitude events are also shown in C (b, middle panel). c, unitary I–V relationships for full-amplitude events and half-amplitude events recorded from cell-attached (open circles and filled triangles, respectively) and inside-out (filled circles) patches. Each symbol represents the mean amplitude of unitary currents with s.e.m. (vertical bar).

Similar single-channel events of large unitary amplitude could be recorded from cell-attached patches on cardiomyocytes which were subjected to chemical ischaemia (Fig. 4B). The channel activity appeared after a lag period of 8.4 ± 2.9 min (n = 15) (Fig. 4Ba). Most channel events had mean unitary amplitudes of 10.3 ± 0.7 pA (n = 19) and −6.8 ± 0.5 pA (n = 22) at +25 and −25 mV, respectively (Fig. 4Bb). Together with such full-amplitude events, half-amplitude events could also be seen in 5 of 13 patches (data not shown). At larger positive and negative potentials, the unitary currents also showed inactivation kinetics (data not shown). The unitary I–V relationship for full-amplitude events was slightly outwardly rectifying with a mean slope conductance of 390 ± 16 pS at positive potentials and of 312 ± 21 pS at negative potentials (Fig. 4Bc, open circles). After excision of the patch membrane in ischaemic solution, the unitary I–V relationship became linear with a slope conductance of 393 ± 4 pS (Fig. 4Bc, filled circles).

Hypoxic stress also induced the activation of large-conductance channel activity in neonatal rat cardiomyocytes after a lag period of 8.1 ± 1.8 min (n = 10), as shown in Fig. 4Ca). The full-amplitude currents were 8.5 ± 0.6 pA (n = 5) at +25 mV and −5.7 ± 0.6 pA (n = 5) at −25 mV (Fig. 4Cb, top trace). Under hypoxic conditions, in contrast to in hypotonic and chemically ischaemic conditions, most (9 of 13) patches on cardiomyocytes exhibited stable half-amplitude channel events. The unitary currents of half-amplitude events were 4.1 ± 0.3 pA (n = 16) at +25 mV and −2.9 ± 0.2 pA (n = 13) at −25 mV (Fig. 4Cb, middle trace). The unitary I–V relationship for cell-attached channel currents exhibited slight outward rectification (Fig. 4Cc). The mean slope conductances of full-amplitude events (open circles) and half-amplitude events (filled triangles) were 311 ± 37 and 153 ± 11 pS at positive potentials, respectively. At negative potentials, they were 240 ± 30 and 142 ± 15 pS (n = 3–7). After excision of the patch membrane, however, the full-amplitude events became predominant (Fig. 4Cb, bottom trace). The unitary I–V relationship for inside-out channel events was linear with a mean slope conductance of 341±8 pS (Fig. 4Cc, filled circles).

Maxi-anion channels serve as a pathway for ATP release

Since similar maxi-anion channels in C127 cells have been shown to exhibit significant ATP conductivity (Sabirov et al. 2001), we next tested the ATP conductivity of the maxi-anion channel in cardiomyocytes. When all anions were replaced with ATP⁴⁻ or MgATP²⁻ and currents recorded from voltage-clamped, excised patch membranes containing several channels, large outward currents (carried by Cl⁻ from the pipette solution) as well as small inward currents (presumably carried by ATP⁴⁻ or MgATP²⁻ from the bath solution) were consistently observed. This was seen in the case of both ramp pulses (Fig. 5Aa and Ba) and step pulses (Fig. 5Ab and Bb). The small inward ATP currents were found to be inhibited by arachidonic acid (data not shown, n = 3), similar to our previous observation in C127 cells (Dutta et al. 2002). With the replacement of anions by ATP⁴⁻ or MgATP²⁻, the reversal potential shifted from 0 mV to −14.8 ± 1.7 mV or −30.9 ± 1.6 mV (n = 3–5), giving a P_ATP/P_Cl value of 0.12 ± 0.02 and P_MgATP/P_Cl of 0.16 ± 0.01. When Na⁺ in the pipette was replaced with TEA⁺, and ATP⁴⁻ or MgATP²⁻ was present in the bath, similar inward currents with a reversal potential of −16.9 ± 2.1 or −29.9 ± 1.5 mV (n = 5) could be observed (Fig. 5Ab and Bb: bottom traces) giving a P_ATP/P_Cl value of 0.10 ± 0.02 and P_MgATP/P_Cl of 0.17 ± 0.01. This excludes the possibility that the small inward currents were carried by Na⁺. We therefore conclude that the maxi-anion channel identified in cardiomyocytes conducts both ATP⁴⁻ and MgATP²⁻.

Figure 5. ATP conductivity of maxi-anion channels, shown in recordings from inside-out patches excised from neonatal rat cardiomyocytes.

Aa, representative ramp I–V records from a macro-patch exposed to standard Ringer solution and one exposed to 100 mm Na₄ATP solution. The pipette solution was standard Ringer. The amplitude of outward currents recorded from inside-out patches became smaller when ATP⁴⁻ was present in the bath, suggesting a blocking of Cl⁻ currents by ATP⁴⁻, as reported in C127 cells (Sabirov et al. 2001). Ab, one and two representative traces of single-channel currents recorded at +50 and −50 mV, respectively, when 100 mm Na₄ATP was present in the bath. The pipette solution was standard Ringer solution for the two upper traces and contained TEA-Cl for the bottom trace. The dashed lines indicate the zero-current level. The mean single-channel amplitude of outward Cl⁻ current at +50 mV was 12.3 ± 0.3 pA (n = 8), and that of inward ATP⁴⁻ current at −50 mV was −1.9 ± 0.1 pA (n = 8, Ringer solution in pipette) and −1.8 ± 0.1 pA (n = 10, TEA in pipette). Ba, representative ramp I–V records from a macro-patch exposed to standard Ringer solution and one exposed to 100 mm Na₂MgATP solution. The pipette solution was TEA-Cl. Bb, one of each representative traces of single-channel currents recorded at +50 and −50 mV, respectively, when 100 mm Na₂MgATP was present in the bath. The dashed lines indicate the zero-current level. The mean single-channel amplitude of outward Cl⁻ current at +50 mV was 11.7 ± 0.8 pA (n = 6), and that of inward MgATP²⁻ current at −50 mV was −2.3 ± 0.3 pA (n = 7, TEA in pipette).

A blocker of gap junction hemichannels, octanol (1 mm), never affected the single-channel activity of maxi-anion channels observed after excision (data not shown, n = 5). In contrast, as shown in Fig. 6, a stilbene-derivative Cl⁻ channel blocker, SITS (100 μm), caused profound flickery block of inside-out single-channel events activated after excision. Also, a carboxylate analogue Cl⁻ channel blocker, NPPB (100 μm), prominently decreased the single-channel amplitude in symmetrical Cl⁻ conditions. Arachidonic acid (20 μm) completely abolished the channel activity detected in the inside-out mode. Although bath application of Gd³⁺ (50 μm) failed to block the channel activity in inside-out patches (data not shown, n = 4), it rapidly blocked the channel activity in outside-out patches (Fig. 6A). In both the inside-out and outside-out modes, the maxi-anion channel was insensitive to glibenclamide (200 μm), a potent inhibitor of cystic fibrosis transmembrane conductance regulator (CFTR) and VSOR (volume-sensitive outwardly rectifying) Cl⁻ channels as well as of ATP-sensitive K⁺ channels. Effects of these drugs on the mean macro-patch currents activated after excision are summarized in Fig. 6B. The pharmacological profile observed for maxi-anion channels in inside-out patches excised after hypoxia-induced activation was essentially the same (Fig. 6C). The pharmacological properties described above are identical to those of the channel in C127 cells (Sabirov et al. 2001; Dutta et al. 2002). They are also essentially the same as those of ATP release (Fig. 1Ab), supporting the hypothesis that the channel serves as a pathway for ATP release from cardiomyocytes.

Figure 6. Pharmacological profile of the maxi-anion channel in neonatal rat cardiomyocytes.

A, representative single-channel current traces recorded during application of step pulses (protocol shown at the top of traces) from excised inside-out and outside-out (only for Gd³⁺) patches in the absence (control) or presence of drugs. The drugs used were SITS (100 μm), NPPB (100 μm), arachidonic acid (20 μm), Gd³⁺ (50 μm) and glibenclamide (200 μm). B and C, effects of drugs on mean currents for excised macro-patches, where maxi-anion channels were activated by excision (B) or by hypoxic stress before excision (C). Currents were recorded at +25 mV (open columns) and −25 mV (filled columns). Data are normalized to the mean current measured before application of drugs and after correction for the background current. Each column represents the mean ±s.e.m. (vertical bar). *P < 0.02 versus control.

In the whole-cell recording mode, NMDG-based bath and pipette solutions were employed in order to effectively exclude cationic components. With ATP-containing pipette solution, a VSOR chloride current was found to be prevailing upon hypotonic stimulation. We therefore used ATP-free pipette solution and supplemented bath solution with 300 μm phloretin, a blocker relatively selective to VSOR Cl⁻ channels (Fan et al. 2001). In these conditions, activation of non-rectifying current with voltage- and time-dependent inactivation characteristic to maxi-anion channels was observed with a lag period of 4.8 ± 1.5 min (n = 12) after a hypotonic challenge (Fig. 7A and Ca). The peak current density was 84 ± 12 pA pF⁻¹ (n = 12) at +25 mV and −80 ± 12 pA pF⁻¹ (n = 12) at −25 mV. The average number of maxi-anion channels expressed in a single cardiomyocyte, estimated by the ratio of peak swelling-induced whole-cell current to the single-channel amplitude, was 59 ± 13 (n = 12) ranging from 17 to 167 channels per cell. Chemical ischaemia also activated similar whole-cell currents after a lag period of 3.9 ± 0.9 min (n = 14) with a peak current density of 93 ± 22 pA pF⁻¹ (n = 14) at +25 mV and −95 ± 22 pA pF⁻¹ (n = 14) at −25 mV. Whole-cell currents activated both by osmotic swelling and by chemical ischaemia had linear current-to-voltage relationships with a reversal potential of around 3 mV (Fig. 7B), which is close to the equilibrium potential of Cl⁻ in these conditions. Effective inhibitors of ATP release (Gd³⁺ (50 μm), NPPB (100 μm) and arachidonic acid (20 μm)), significantly decreased the whole-cell currents activated in both hypotonic and ischaemic conditions (Fig. 7D).

Figure 7. Whole-cell currents activated in hypotonic and ischaemic conditions in neonatal rat cardiomyocytes.

A, representative record during application of alternating pulses from 0 to ±25 mV (every 10 s). Horizontal bar indicates the time of application of hypotonic solution containing 300 μm phloretin. B, current–voltage relationships measured at the beginning of the pulses in the hypotonic (filled circles) and ischaemic (open circles) conditions. The whole-cell currents were normalized by the cell capacitance (n = 5–6). C, current responses to step pulses from −50 to +50 in 10 mV increments, hypotonic (a) and ischaemic (b) conditions, respectively. D, effects of drugs on peak whole-cell currents activated by hypotonic (a) or ischaemic (b) stress. Currents were recorded at +25 mV (open columns) and −25 mV (filled columns). Data are normalized to the mean current measured before application of drugs and after correction for the background current. Each column represents the mean ±s.e.m. (vertical bar). *P < 0.05 versus control.

These results are in good agreement with the hypothesis that maxi-anion channels serve as a pathway for ATP release from cardiomyocytes.

Discussion

The normal basal levels of plasma and interstitial ATP are very low, for example less than 20–40 nm in the human venous plasma (Forrester, 1972) and in the cardiac interstitial space (Kuzmin et al. 1998). However, the local concentration of extracellular ATP is known to often exceed micromolar levels due to ATP release associated with local trauma, vascular injury and platelet aggregation (Ugurbil & Holmsen, 1981; Born & Kratzer, 1984). In the heart, it is well known that ATP is released into the interstitial space during electrical stimulation (Abood et al. 1962), application of cardiotonic agents (Darius et al. 1987; Vial et al. 1987; Borst & Schrader, 1991; Katsuragi et al. 1993; Hall et al. 1995), mechanical stretch (Ninomiya et al. 2002), increased blood flow (Darius et al. 1987; Vials & Burnstock, 1996), hypoxia (Paddle & Burnstock, 1974; Forrester & Williams, 1977; Clemens & Forrester, 1981; Vial et al. 1987; Borst & Schrader, 1991) and ischaemia or ischaemic pre-conditioning (Borst & Schrader, 1991; Kuzmin et al. 1998, 2000; Ninomiya et al. 2002). Since ATP-selective purinergic receptors have affinities as high as 100–1000 nm (Dubyak & El-Moatassim, 1993), released ATP may serve as an extracellular signal. In the heart, many ionotropic P2X receptor and metabotropic P2Y receptor subtypes are expressed (Vassort, 2001), and a variety of effects on the heart mediated by extracellular ATP have been reported (Burnstock & Kennedy, 1986; Pelleg et al. 1990).

There are a number of possible sources of cardiac ATP release, including purinergic nerves innervating the heart (Burnstock, 1972), cardiac vascular endothelial cells (Sparks & Bardenheuer, 1986) and cardiomyocytes themselves (Forrester & Williams, 1977). In the present study, neonatal rat cardiomyocytes in primary culture were demonstrated, by a luciferin–luciferase assay, to respond to ischaemic/hypoxic stress as well as to a hypotonic challenge with massive release of ATP. Although a large part of the released ATP is undoubtedly degraded by ecto-nucleotidases and ecto-ATPases, the concentration of ATP released from a single cardiomyocyte was found to reach over 10 μm at the cell surface by a P2X receptor-based biosensor technique.

Since most ATP molecules exist in anionic forms at physiological pH, it is possible that non-lytic and non-exocytotic ATP release is mediated by some type of anion channel. The present study showed that swelling-induced ATP release from neonatal rat ventricular myocytes is actually sensitive to the anion channel blockers, SITS and NPPB. So far three types of anion channels have been reported to be involved in ATP release: the cAMP/PKA-activated CFTR Cl⁻ channel in a variety of epithelial cell types (Schwiebert, 1999), the volume-sensitive outwardly rectifying (VSOR) Cl⁻ channel in endothelial cells (Hisadome et al. 2002) and maxi-anion channels (Sabirov & Okada, 2004) in mammary C127 cells (Sabirov et al. 2001; Dutta et al. 2002), and kidney macula densa cells (Bell et al. 2003). Ventricular cardiomyocytes are known to express all three types of anion channel (Coulombe & Coraboeuf, 1992; Tseng, 1992; Horowitz et al. 1993). Lader et al. (2000) reported that neonatal rat cardiomyocytes possess a cAMP-activated, glibenclamide-sensitive ATP-conductive pathway associated with CFTR. In the present study, however, swelling-induced ATP release from neonatal rat ventricular myocytes was found to be insensitive to glibenclamide, which is a potent blocker of cardiac CFTR (Tominaga et al. 1995) and VSOR Cl⁻ channels (Liu et al. 1998). In contrast, cardiac ATP release was sensitive to Gd³⁺ and arachidonic acid, which are the most effective blockers of maxi-anion channels in C127 cells (Sabirov et al. 2001; Dutta et al. 2002).

Considering intracellular ATP and Cl⁻ concentrations are 2 and 20 mm, respectively, it can be estimated that a cardiac maxi-anion channel may transport, in the full open state, around 4 × 10⁵ MgATP²⁻ s⁻¹ and around 5 × 10⁵ ATP⁴⁻ s⁻¹ at −40 mV. In our experiments, the maximal concentration of ATP determined in 1.5 ml perfusate collected every 1 min from (7.5 ± 1.0) × 10⁴ cells attached on one coverglass was 550 ± 62 pm (Fig. 1Aa). Therefore, the measured rate of ATP release from cardiac myocytes was around 1.1 × 10⁵ molecules s⁻¹ cell⁻¹ in response to cell swelling. From current inactivation shown in Fig. 3A we can roughly estimate the open probability of maxi-anion channel to be about 0.11–0.2 at −40 mV. Therefore, we suggest that brief activation of only a few maxi-anion channels would be sufficient to provide the observed rate of ATP release. In contrast, the total number of maxi-anion channels expressed in a single cardiomyocyte (around 60) seems to be much larger, as revealed by whole-cell recordings.

To hypotonic, ischaemic or hypoxic stimulation, cardiomyocytes responded with ATP release and activation of whole-cell maxi-anion channel current with a similar time course. The lag time required to respond to these stimuli with ATP release detected by the luciferine–luciferase assay (2.5–5.3 min: Fig. 1) or by the biosensor technique (4.3–6.5 min: Fig. 2) was comparable to that for activation of whole-cell maxi-anion channel current (3.9–4.8 min: Fig. 7). However, the average lag time for on-cell activation of unitary maxi-anion channel current (5.7–8.4 min: Fig. 4) was longer compared with those for the above whole-cell events. This apparent discrepancy may be explained by the fact that the channels existing within the patch membrane are spatially separated from the rest of the plasma membrane which actually receives these stimuli, and therefore a longer time lag is necessary for their signals to reach the maxi-anion channel in the patch membrane. Also, there is a possibility that activation of maxi-anion channels might have been retarded by mechanical perturbation due to giga-seal attachment of a patch pipette on a cardiomyocyte. In the loose-patch on-cell configuration, in fact, the lag time (1.7 ± 0.3 min, n = 5) for swelling-induced activation of unitary maxi-anion channel current was much shorter than that in the giga-seal on-cell configuration (A.K. Dutta, R.Z. Sabirov and Y. Okada, unpublished observations). Such mechanical perturbation by giga-seal on-cell patch pipettes may also explain why the maxi-anion channel activation could be only partially reversible after removal of stimuli (Fig. 4C), whereas the ATP release responses detected by both luciferine–luciferase and biosensor assays were fully reversible (Figs 1 and 2).

The biophysical properties of the cardiac maxi-anion channel, such as the single-channel conductance, voltage-dependent inactivation and anion selectivity, were identical to those of VDACL channels in C127 cells. Although every individual pharmacological agent used in our experiments is not absolutely specific to maxi-anion channels, essential identity of the whole pharmacological profile between the ATP release and maxi-anion channels may provide evidence for ATP release via maxi-anion channels. In fact, in the present study, the pharmacological profile of cardiac maxi-anion channels was found to be essentially the same as that of ATP release from swollen cardiomyocytes. Moreover, cardiac maxi-anion channels were found to actually conduct both MgATP²⁻ and ATP⁴⁻. Not only a hypotonic challenge but also hypoxic or ischaemic stress were found to be effective stimuli for the activation of the cardiac maxi-anion channel and massive release of ATP from cardiomyocytes. From these results, we conclude that cardiac maxi-anion channels serve as a pathway for ATP release under hypotonic, ischaemic and hypoxic conditions. Since cardiac cell swelling is known to be induced during ischaemia or hypoxia (Tranum-Jensen et al. 1981; Steenbergen et al. 1985; Jennings et al. 1986), it seems likely that cell swelling underlies the mechanism by which maxi-anion channels are activated in response to hypotonic, hypoxic and ischaemic stress.

Previously it was reported that the maxi-anion channel is only transiently expressed in neonatal rat cardiomyocytes and could not be found in adult cells (Coulombe & Coraboeuf, 1992). Attaching patch pipettes with very fine tips on cardiomyocytes freshly isolated from adult rat hearts, however, we have recently succeeded in observing functional maxi-anion channels with properties similar to those of, though less frequently than in, neonatal cells (A.K. Dutta, R.Z. Sabirov and Y. Okada, unpublished observations). Thus, it seems possible that adult cardiomyocytes also release ATP via maxi-anion channels in response to ischaemic, hypoxic or osmotic stress. This possibility is currently under investigation in our laboratory.

In the present study, most patch-clamp experiments were performed in Ca²⁺-free solutions in order to prevent spontaneous cardiomyocyte contraction. Removing Ca²⁺ ions is known to activate connexin hemichannels; however, octanol, a known blocker of gap junction hemichannel, at the concentration of 1 mm, had no significant effect on maxi-anion channels in Ca²⁺-free conditions as well as on ATP release by hypotonic stress, which was normally measured in the presence of 2 mm Ca²⁺. Therefore, we can exclude the contribution of hemichannels to the ATP release and patch-clamp data.

In summary, we conclude that neonatal rat cardiomyocytes respond to ischaemia, hypoxia and osmotic swelling with ATP release via maxi-anion channels, because (1) hypotonic, hypoxic or ischaemic stress induces both ATP release and activation of maxi-anion channels in cardiomyocytes, (2) both cardiac ATP release and the maxi-anion channel activity share the same pharmacology, and (3) the cardiac maxi-anion channel showed significant conductivity to ATP⁴⁻ and MgATP²⁻.