Cytosolic and mitochondrial Ca²⁺ signals in patch clamped mammalian ventricular myocytes

Abstract

1. Ventricular myocytes isolated from ferret or cat were loaded with the acetoxymethyl ester form of indo-1 (indo-1 AM) such that ∼75 % of cellular indo-1 was mitochondrial. The intramitochondrial indo-1 concentration was 0.5–2 mm.

2. Myocytes were also voltage clamped (membrane capacitance, C_m = 100 pF) and a typical wash-out time constant of cytosolic indo-1 by a patch pipette was found to be ∼300 s. Depolarizations to +110 mV produced graded and progressive cellular Ca²⁺ load via Na⁺-Ca²⁺ exchange.

3. During these relatively slow Ca²⁺ transients, cell contraction (ΔL) paralleled fluorescence ratio signals (R) such that ΔL could be used as a bioassay of cytosolic [Ca²⁺] ([Ca²⁺]_c), where [Ca²⁺]_CL is the inferred signal which is delayed by ∼200 ms from true [Ca²⁺]_c.

4. In myocytes without Mn²⁺ quench, the kinetics of the total cellular indo-1 signal, ΔR (including cytosolic and mitochondrial components), match ΔL during stimulations at low basal [Ca²⁺]_i. However, after progressive Ca²⁺ loading, ΔR kinetics deviate from ΔL dramatically. The deviation can be completely blocked by a potent mitochondrial Ca²⁺ uniport blocker, Ru360.

5. When cytosolic indo-1 is quenched by Mn²⁺, initial moderate stimulation triggers contractions (ΔL), but no change in indo-1 signal, indicating both the absence of cytosolic Ca²⁺-sensitive indo-1 and unchanged mitochondrial [Ca²⁺] (Δ[Ca²⁺]_m). Subsequent stronger stimulation evoked larger ΔL and also ΔR. The threshold [Ca²⁺]_c for mitochondrial Ca²⁺ uptake was 300–500 nM, similar to that without Mn²⁺ quench.

6. At high Ca²⁺ loads where Δ[Ca²⁺]_m is detected, the time course of [Ca²⁺]_m was different from that of [Ca²⁺]_c. Peak [Ca²⁺]_m after stimulation has an ∼1 s latency with respect to [Ca²⁺]_c, and [Ca²⁺]_m decline is extremely slow.

7. Upon a Ca²⁺ influx which increased [Ca²⁺]_c by 0.4 μm and [Ca²⁺]_m by 0.2 μm, total mitochondrial Ca²⁺ uptake was ∼13 μmol (l mitochondria)⁻¹.

8. With Mn²⁺ quench of cytosolic indo-1, there was no mitochondrial uptake of Mn²⁺ until the point at which mitochondrial Ca²⁺ uptake became apparent. However, after mitochondrial Ca²⁺ uptake starts, mitochondria continually take up Mn²⁺ even during relaxation, when [Ca²⁺]_c is low.

9. It is concluded that mitochondria in intact myocytes do not take up detectable amounts of Ca²⁺ during individual contractions, unless resting [Ca²⁺]_c exceeds 300–500 nM. At high cell Ca²⁺ loads and [Ca²⁺]_c, mitochondrial Ca²⁺ transients occur during the twitch, but with much slower kinetics than those of [Ca²⁺]_c.

Ca²⁺ is a ubiquitous second message for cell functions, such as excitation-contraction coupling in myocytes (Bers, 1991), transmitter secretion in neuron and endocrine cells (Augustine & Neher, 1992) and gene expression (Hardingham, Chawla, Johnson & Bading, 1997). In cardiac myocytes, excitation-contraction coupling is controlled by cytosolic Ca²⁺ ions, which are in turn affected by endogenous passive buffers (Berlin, Bassani & Bers, 1994) and by several Ca²⁺ influx and efflux mechanisms including Ca²⁺ channels, sarcoplasmic reticulum Ca²⁺ pump, Na⁺-Ca²⁺ exchanger, sarcolemmal Ca²⁺-ATPase and mitochondrial Ca²⁺ uptake (Bers, 1991).

It has been known since the 1950s that isolated mitochondria in vitro can take up Ca²⁺ (for reviews, see Crompton, 1990; McCormack & Denton, 1993; Gunter, Gunter, Sheu & Gavin, 1994; Hansford, 1994). Subsequently it has been found that mitochondrial Ca²⁺ uptake in intact cells may be different from that in vitro (Hansford, 1994). More recently, novel techniques have been developed to measure mitochondrial Ca²⁺ uptake in intact cells (Thayer & Miller, 1990; Rizzuto, Simpson, Brini & Pozzan, 1992; Friel & Tsien, 1994; Hajnoczky, Robb-Gaspers, Seitz & Thomas, 1995; White & Reynolds, 1995; Herrington, Park, Babcock & Hille, 1996; Jou, Peng & Sheu, 1996) including heart cells (Miyata, Silverman, Sollott, Lakatta, Stern & Hansford, 1991; Wendt-Gallitelli & Isenberg, 1991; Bassani, Bassani & Bers, 1992, 1993, 1994a,b; Chacon, Ohata, Harper, Trollinger, Herman & Lemasters, 1996). Although there are considerable differences among these studies, all found mitochondrial Ca²⁺ uptake under some experimental conditions.

Cellular ATP production is mainly by mitochondria. At least three mitochondrial enzymes, which might regulate the rate of ATP production, are dependent on [Ca²⁺] in the physiological range. Thus, in addition to cytosolic Ca²⁺ buffering, it has been suggested that the physiological function of mitochondrial Ca²⁺ uptake is as a key regulator of ATP production (Crompton, 1990; Duchen, 1992; McCormack & Denton, 1993; Gunter et al. 1994; Hansford, 1994; Brandes & Bers, 1997). Cellular Ca²⁺ overload can also cause a sudden mitochondrial permeability transition, and release reactive oxygen species, which are toxic and can cause cell injury or death (Gunter et al. 1994). Thus, the mechanisms and regulation of mitochondrial Ca²⁺ uptake are important for both physiology and pathology in all kinds of cells.

In a study of rat ventricular myocytes by Miyata et al. (1991), mitochondrial free Ca²⁺ concentration, [Ca²⁺]_m, was directly measured using the Ca²⁺-sensitive fluorescent probe indo-1, after quenching of cytosolic indo-1 by addition of extracellular Mn²⁺. They found a significant increase in [Ca²⁺]_m upon physiological stimulation, implicating a physiological role of mitochondria in Ca²⁺ sequestration. Although these results provide much useful information about mitochondrial Ca²⁺ uptake in intact heart cells, there are at least three important issues remaining. The first issue concerns the use of Mn²⁺ to quench cytosolic indo-1 (i.e. intracellular Mn²⁺ may affect mitochondrial Ca²⁺ uptake; Gavin, Gunter & Gunter, 1990; Hansford, 1994, Chacon et al. 1996). Thus, it is important to examine mitochondrial Ca²⁺ uptake under Mn²⁺-free conditions. The second issue is that Miyata et al. (1991) found the build-up of [Ca²⁺]_m required many repeated field stimulations. Because the increase in [Ca²⁺]_m during each twitch was too small to be resolved, it is not clear yet what the rate of mitochondrial Ca²⁺ uptake in the intact cell is. The third issue is the temporal relationship between cytosolic [Ca²⁺] ([Ca²⁺]_c) and [Ca²⁺]_m recorded at the same time in the same cell. Thus, the main aims of the present study are to address all of these three issues.

In previous work from our laboratory, Bassani et al. (1992, 1993, 1994a) used cytosolic indo-1 and pharmacological dissections to infer that at a [Ca²⁺]_c of 1 μm, mitochondrial Ca²⁺ uptake from the cytosolic Ca²⁺ was at a rate of ∼1 μmol (l cytosol)⁻¹ s⁻¹ (or 2 μm s⁻¹ referred to mitochondrial volume). Indeed, they found that mitochondria are responsible for less than 1–2 % of Ca²⁺ removal from the cytoplasm during relaxation of a normal twitch (Bassani et al. 1994a). Although these data are useful, all mitochondrial transport was estimated indirectly from cytosolic indo-1 signals. In the present work we have developed a method to heavily and preferentially load mitochondria with indo-1 AM. This allows direct measurements of fluorescence signals from mitochondrial indo-1. We took advantage of the strong Na⁺-Ca⁺ exchanger to bring Ca⁺ into the cell to varying levels under voltage clamp control. Under these conditions we were able to detect substantial Ca²⁺ uptake by mitochondria, but only after progressive loading of cells with Ca²⁺ (and elevation of resting [Ca²⁺]_c). The increase in [Ca²⁺]_m could even be detected in the presence of small contamination by cytosolic indo-1. To simultaneously monitor pure [Ca²⁺]_m and [Ca²⁺]_c, Mn²⁺ was used to quench cytosolic indo-1. We also used cell contraction as a bioassay for [Ca²⁺]_c and refer to it as [Ca²⁺]_CL to distinguish it from [Ca²⁺]_c derived from the indo-1 signal. We found peak [Ca²⁺]_c preceded peak [Ca²⁺]_m by ∼1 s, and the kinetics of [Ca²⁺]_m were substantially different from [Ca²⁺]_c.

METHODS

Cells and solutions

Single ventricular myocytes were prepared, as described elsewhere, from hearts of either adult ferret (Bassani et al. 1994b) or adult cat (Wu, Vereecke, Carmeliet & Lipsius, 1991). Briefly, animals were anaesthetized with sodium pentobarbitone (50–75 mg kg⁻¹, i.p.) and hearts were removed and perfused with collagenase and pronase. The isolated cells were kept in Dulbecco's modified Eagle's medium (Gibco, for ferret cells) or standard Tyrode solution (for cat cells) prior to experimental use the same day. Before use cells were re-plated in the experimental chamber containing standard Tyrode solution. Single chromaffin cells were prepared from calf adrenal glands (obtained from a local abbatoir) as described by Artalejo, Elhamdani & Palfrey (1996).

The standard external Tyrode solution contained (mm): 140 NaCl, 6 KCl, 1 MgCl₂; 2 CaCl₂, 10 glucose, 10 Hepes; pH 7.4. In some experiments, 0.7 mm MnCl₂ was added to the standard Tyrode solution to quench the cytosolic indo-1. The bath solution for measurement of minimal indo-1 fluorescence ratio (R_min) was the same as standard Tyrode solution, except 2 mm Ca²⁺ was replaced with 1 mm EGTA to yield a Ca²⁺-free (0 Ca²⁺) solution. For maximal fluorescence ratio (R_max) the bath solution contained (mm): 110 KCl, 1 MgCl₂, 30 BDM (2,3-butanedione monoxime), 10 glucose, 0.1 EGTA, 10 Hepes; pH 7.4. Digitonin (10 μm) and 3 mm CaCl₂ were added during R_max test. BDM was added to inhibit myocyte hypercontracture at high [Ca²⁺].

The standard, internal, pipette solution contained (mm): 80 caesium methanesulphonate, 40 CsCl, 6 NaCl, 0.1 EGTA, 1 MgCl₂, 10 Hepes, 0.3 GTP, 10 ATP; pH 7.2. In some experiments, 1 mm indo-1 salt, 10 mm EGTA and/or 0.2–2 μm Ru360 were added to the internal solution. All experiments were done at room temperature (22°C).

Isolation of ferret heart mitochondria and measurement of Ca²⁺ uptake

Animals were anaesthetized with 30 mg pentobarbitone (kg body weight)⁻¹ (i.p.). Hearts were excised, rinsed in ice-cold saline, and the atria and aorta were trimmed off. The ventricles were immersed in isolation medium containing 180 mm KCl, 10 mm EGTA and 0.5 % bovine serum albumin, pH 7.4. Mitochondria were isolated according to ‘procedure B’ described by Matlib, Vaghy, Rouslin & Schwatz (1984). Protein concentration was determined by the Lowry method. The rate of Ca²⁺ uptake into isolated mitochondria was measured spectrophotometrically using the differential absorption of arsenazo III (Sigma; 675 nm vs. 685 nm) in an Aminco dual-beam and dual-wavelength spectrophotometer as described previously (Vaghy, Johnson, Matlib, Wang & Schwartz, 1982). The assay was carried out at 37°C in 3 ml medium containing (mm): 120 KCl, 10 Mops-KOH buffer (pH 7.2), 2 potassium phosphate buffer (pH 7.2), 5 potassium malate, 5 potassium pyruvate, 0.05 arsenazo III; with 1 mg mitochondrial protein. The uptake of Ca²⁺ was started with the injection of 16.67 μm CaCl₂ into the assay cuvette, which was continuously stirred. Ru360 was added 1 min before the addition of CaCl₂.

Preparation of Ru360

Ru360 was prepared according to a procedure similar to that of Ying, Emerson, Clarke & Sanadi (1991). Details of the chemical structure and specificity of action on Ca²⁺ uptake into mitochondria will be published elsewhere (Matlib et al. 1998).

Electrophysiology and data acquisition

Standard whole-cell recording techniques and an Axopatch-1B amplifier (Axon Instruments) were used for all electrophysiological recordings. Pipette-cell access resistance (R_acc) was routinely monitored by manual compensation. Except for the double whole-cell clamp experiments and indo-1 wash-out experiments (see Fig. 2), we did not try to control the access resistance, as long as it was below 20 MΩ. In the experiments of Fig. 2, to quantify the time constant of indo-1 wash-out or wash-in, we used larger pipettes (∼2 MΩ) and maintained R_acc constant and below 5 MΩ during prolonged whole-cell dialysis.

Figure 2. Time course of whole-cell dialysis of indo-1 via patch pipette.

A, loading and unloading of indo-1 salt by double whole-cell dialysis in a cat ventricular myocyte. The cell was first loaded with indo-1 salt by patch pipette no. 1 (containing 1 mm indo-1). After partial loading, the first pipette was removed. Then the preloaded indo-1 was washed out by a second ruptured patch pipette, no. 2 (containing no indo-1). This experiment provides quantitative data for the time course of whole-cell dialysis in ventricular myocytes. See text for details. B, complete wash-out of cytosolic indo-1 by whole-cell dialysis in another cat cell. This cell (Indo-1 AM, ○) was preloaded by the standard indo-1 AM method. To assay the cytosolic fraction of indo-1, the whole cell was dialysed with a pipette containing no indo-1. After ≈2000 s, only 25 % of the dye was washed out. According to A, cytosolic indo-1 should already be completely washed out at this time. Breaking the cell membrane by patch pipette does not reduce the dye further (±, last two data points), indicating that all cytosolic indo-1 had been washed out. Thus the cytosol contains only 25 % of the total indo-1, while the rest (75 %) is in cell organelles. The cell in A is displayed again in B (Indo-1 salt, •) together with the indo-1 AM cell. Note that time in B is normalized by eqn (11a), which allows the cytosolic wash-out times to be comparable for the two cells.

The holding potential was always -40 mV. Two kinds of depolarization pulses were used. To evoke rapid Ca²⁺ influx through voltage-gated Ca²⁺ channels, 200–400 ms pulses to +10 mV were used. Inactivation of Ca²⁺ channels limits Ca²⁺ influx with longer pulses. In addition, Ca²⁺ current run-down under whole-cell dialysis in ventricular cells makes it difficult to maintain Ca²⁺ influx during 20–60 min experiments. To overcome these limitations of Ca²⁺ channels, most experiments here use depolarization pulses to +110 mV, which allows Ca²⁺ entry through the Na⁺-Ca²⁺ exchanger in ventricular cells (Bers, 1991). The advantage of this type of stimulation is that there is no apparent inactivation or run-down of the Na⁺-Ca²⁺ exchanger under our experimental conditions. Thus, we can control the Ca²⁺ inflow simply by pulse duration in the range of 200–5000 ms.

Signals for whole-cell current, cell contraction (see below) and two indo-1 fluorescence signals were simultaneously recorded by a data acquisition system pCLAMP 6 (Axon Instruments) using an IBM-PC compatible computer (sampling rate, 125–500 Hz). Fluorescence and contraction signals were further low-pass filtered by binomial smoothing with Igor software (smooth factor, 10). Analysis was done on a Macintosh computer PowerBook 520c using IgorPro 2.04 software (WaveMetrics, Oregon, USA).

Indo-1 fluorescence microscopy

Fluorescence recording was similar to that previously described (Bassani et al. 1992), with an oil-immersion objective (Nikon Fluo 40/1.3). Light from a UV illuminator (75 W xenon, Oriel) passed through an excitation filter (360 ± 5 nm bandwidth) and was reflected into the objective by a 380 nm dichroic mirror (Chroma, Brattleboro, VT, USA). The emitted fluorescence was split by another dichroic mirror (440 nm) into two bandpass filters (400 ± 12 and 500 ± 12 nm). Optimizing the optical system gave a total indo-1 detection efficiency of about 36 % (not including light losses at the lens and coverslip).

To reduce the rate of indo-1 photobleaching, the UV light is normally blocked by a shutter under pCLAMP 6 software control. This allows exposure of the cell to UV light only during data acquisition. Photobleaching of cellular indo-1 is less than 10 % of initial indo-1 fluorescence after 500 s, which is the maximum exposure time during these experiments.

Procedure for indo-1 AM loading

Cells were loaded with indo-1 either by a patch pipette containing indo-1 salt, or by incubation with Tyrode solution containing 5 μm indo-1 acetoxymethyl ester (indo-1 AM; Molecular Probes). The procedure for indo-1 AM loading is designed to load indo-1 AM preferentially and heavily into mitochondria. Before indo-1 AM loading, cells were plated on a laminin-coated chamber with a coverslip glass bottom. Then, 500 μl Tyrode solution containing 5 μm indo-1 AM was incubated with the cells for 40–75 min at 37°C. This condition loads the cell with a final intracellular indo-1 concentration of ∼0.5–1 mm (estimated by comparison with steady-state fluorescence signal in cells loaded by patch pipette containing 1 mm indo-1 salt). We found that at 37°C indo-1 loading speed was > 5 times faster than that at 23°C. Higher incubation temperature (37°C) may also help in preferentially loading organelles such as mitochodria (Malgaroli, Milani, Meldolesi & Pozzan, 1987). In addition to temperature, the following factors were also important for indo-1 AM loading. (1) The cell density strongly affects indo-1 AM loading. Low cell density helps indo-1 loading. (2) Longer loading time gives a higher loading level. (3) Addition of Pluronic F-127 (Molecular Probes; 1: 10 dilution in the 1 mm indo-1 AM stock) was necessary for stable and heavy indo-1 AM loading. Indo-1 AM loading solution examined under the microscope showed that most of the fluorescence comes from small (1–2 μm diameter) fluorescence spheres in the absence of Pluronic F-127. However, in Pluronic-containing indo-1 AM stock, fluorescence was evenly distributed.

Isosbestic indo-1 signals

When [Ca²⁺] is increased, indo-1 fluorescence emission is increased at 400 nm and decreased at 500 nm. Theoretically there is no Ca²⁺ sensitivity at ∼440 nm for indo-1 emission. A Ca²⁺-insensitive, isosbestic signal (F_c) is very useful because it is directly proportional to the concentration of the intracellular dye. However, in most cases the two fluorescence signals were both sensitive to Ca²⁺. For fura-2, Zhou & Neher (1993a) introduced a method to extract an F_c signal from the two Ca²⁺-sensitive fluorescence signals. Here we extend this method to indo-1 with the following formula:

(1)

where F₄₀₀ and F₅₀₀ are indo-1 emission signals at 400 nm and 500 nm wavelengths, respectively, and α is the ‘isocoefficient’ for the indo-1 set-up. The value of α is sensitive to the optical system including the Ca²⁺ dye. However, for a given set-up with fixed optical hardware conditions, α is constant for different intracellular indo-1 concentrations, for different cells of same cell types, and even for different cell types. Figure 1 shows how α is determined in a ventricular myocyte. Changes in F₄₀₀ and F₅₀₀ are shown in Fig. 1A, and the slope of their relationship in Fig. 1B is equivalent to -α. Figure 1A also shows that the calculated F_c signal is Ca²⁺ insensitive. The α value in Fig. 1 (1.0) was obtained with different settings than for most experiments reported here (where α = 0.23). As noted by Zhou & Neher (1993a), the α value may differ for different dye lot numbers, and α may also change due to slight changes of hardware properties. For these reasons the α value was checked frequently.

Figure 1. Ca²⁺-dependent and isosbestic fluorescence signals.

A, depolarization-induced indo-1 signals of F₄₀₀ and F₅₀₀ in a cat ventricular myocyte. Note both indo-1 signals are sensitive to Ca²⁺ influx: F₄₀₀ increases in amplitude, while F₅₀₀ decreases. The Ca²⁺-insensitive signal, F_c = F₄₀₀+αF₅₀₀, is a modified indo-1 signal from the two original indo-1 signals. F_c, F₄₀₀ and F₅₀₀ are in bead units (BU; see Methods). The isocoefficient, α, is determined in B. B, plot of F₄₀₀vs. F₅₀₀ during depolarization (marked by dashed lines in A) gives a linear regression line with slope -α = -1.0, where α is the ‘isosbestic coefficient’ of eqn (1) for our indo-1 system. C, one Ca²⁺-sensitive signal, F_d = F₅₀₀ - F₄₀₀. This has a high Ca²⁺ sensitivity and is useful because ΔF_d is proportional to ΔCa²⁺-indo-1. D, second Ca²⁺-sensitive signal, R = F₅₀₀/F₄₀₀. R is the traditional ratio signal. All data are from the same cell and same pair of depolarization pulses.

Fluorescence beads (Fluorescence B/B beads, 4.5 μm; Polyscience, Warrington, PA, USA, Cat. No. 18340, Lot 400103) were used as a standard indo-1 fluorescence F_c unit, referred to as the bead unit (BU; Zhou & Neher, 1993b). The fluorescence amplitude of one standard bead is 1 BU, which is quite stable for at least 6 months when kept in the dark at 4°C.

Ca²⁺-sensitive indo-1 signals

We used two different Ca²⁺-sensitive indo-1 signals here. The first is the Ca²⁺-sensitive difference signal, F_d:

(2)

Since F₅₀₀ decreases whilst F₄₀₀ increases when indo-1 binds Ca²⁺, the sensitivity of F_d is the sum of its two components and is more sensitive than either F₄₀₀ or F₅₀₀. However, according to noise theory, the noise level is the same for F₄₀₀, F₅₀₀ and F_d. Thus, the signal-to-noise ratio of F_d is larger than that of F₄₀₀ or F₅₀₀. Nevertheless, the high sensitivity of F_d and its proportionality to Ca²⁺-bound indo-1 concentration makes it particularly useful in the estimation of Ca²⁺ influx into mitochondria (see below). The second Ca²⁺-sensitive indo-1 signal is the traditional fluorescence ratio (R = F₄₀₀/F₅₀₀). This signal is most convenient for calibration of free [Ca²⁺] (see below).

Correction of [Ca²⁺]_c in the presence of compartmentalized indo-1

In indo-1 AM-loaded cells, indo-1 is present in both the cytosol and also in some cellular organelles. Thus, compartmentalized indo-1 can introduce potentially severe artifacts in [Ca²⁺]_c measurement. However, if the compartment does not take up Ca²⁺, as in the case of cells treated with the mitochondrial Ca²⁺ uniport blocker Ru360, the contribution of this compartment can, in principle, be subtracted from the total fluorescence signal at each wavelength (as an additional background signal). We used the method described by Zhou & Neher (1993a) where the modified fluorescence ratio is R_e = (F₄₀₀ - F_400b)/(F₅₀₀ - F_500b) where the subscript b refers to the value in the compartment. This can be rewritten in terms of directly measured values as (see Zhou & Neher, 1993a):

(3)

(4)

(5)

where R is the usual indo-1 fluorescence ratio, r is the fraction of total cellular indo-1 in the compartment, F_c and α are as defined in eqn (1), R_b is the R value of the compartment, and F_c,b is the F_c value of the compartment. In practice, the unknown compartmental parameter, R_b, is assumed to be the same as R before cell stimulation (i.e. at rest R_b = R). The initial value of r can be estimated from average cells by either dialysis of indo-1 out of the cytosol (or sarcolemmal permeabilization) or by quenching of cytosolic indo-1 by Mn²⁺ (see Results). It can also be determined in situ by rupturing the cell membrane at the end of the experiment (Zhou & Neher, 1993a):

(6)

where R_max is the maximum R value when all cellular indo-1 is at very high [Ca²⁺] during calibrations. R_s is the measured R at very high [Ca²⁺]_c when the cell membrane is ruptured by breaking the pipette seal and mitochondrial uptake is blocked by Ru360 (leading to irreversible contraction). It is assumed that the difference between R_s and R_max is solely due to compartmentalized indo-1 which was not exposed to high [Ca²⁺]. The r value gradually increases during recording because whole-cell dialysis washes out part of the cytosolic indo-1 before the end of the experiment. To account for this effect, r is corrected according to the wash-out rate of indo-1 (see Fig. 2).

Calibration of intracellular [Ca²⁺]

The intracellular [Ca²⁺] is:

(7)

where R_min, R_max, K_d and β are calibration/system constants which are determined in the next paragraph. This [Ca²⁺]_i is based on all of the intracellular indo-1 and represents a mixed signal from the cytosol and compartments (e.g. mitochondria). When there is no compartmental indo-1, cytosolic free [Ca²⁺], or [Ca²⁺]_c, is equal to [Ca²⁺]_i. When there is compartmental indo-1, but the compartmental uptake is prevented, [Ca²⁺]_c can be estimated by:

(8)

where R_e is defined by eqn (3). Note that eqn (8) is only valid when there are no Ca²⁺ changes in the compartment.

To calculate [Ca²⁺]_i, one needs to know the four calibration constants of eqn (7). These can be measured by in situ calibration experiments in permeabilized cells with known [Ca²⁺] in the bath (Bassani, Bassani & Bers, 1995). Cells for calibration of R_min, R_max and α were heavily preloaded with indo-1 AM. Briefly, to measure R_min, indo-1 AM-loaded cells were incubated in 0 Ca²⁺ solution (see ‘Cells and solutions’). R_min was measured after resting [Ca²⁺]_i fell to a minimum value (> 20 min). R_min was 0.32 ± 0.02 (n = 6). In some experiments, 1 μm carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) was added together with 10 μm digitonin to assess R_min in mitochondria. We found that R_min values were nearly identical for cells with or without FCCP or digitonin treatment. To determine R_max from cytosol plus mitochondria, cells were first incubated in BDM solution (see ‘Cells and solutions’). R was measured after adding 10 μm digitonin and 4 mm CaCl₂ to the bath. R_max was 3.6 ± 0.2 (n = 5). The R_min and R_max from in vivo calibration were found to be the same for both cat and ferret ventricular myocytes. These calibration values were used for most of the experiments of the present work (where α = 0.23). Where α = 1.0, we found R_min = 0.68 and R_max = 6.31. A K_d of 0.844 μm for indo-1 was taken from previous intracellular estimates (Hove-Madsen & Bers, 1992; Bassani et al. 1995).

The last calibration constant in eqn (7) is β, which is calculated from the following (Zhou & Neher, 1993a):

(9)

where R_min and R_max are indo-1 constants determined by the in vivo calibrations and α is defined by eqn (1). According to eqn (9), β was 6.96 when α = 0.23, and 4.35 when α = 1.0.

Mn²⁺ quench of cytosolic indo-1

Cytosolic indo-1 can be selectively quenched by adding micromolar concentrations of Mn²⁺ to the bath (Miyata et al. 1991; Blatter, 1995) or the patch pipette (Tse, Tse & Hille, 1994), without appreciable quenching of the compartmentalized indo-1. To measure a pure mitochondrial Ca²⁺ signal, 0.4–0.7 mm MnCl₂ was added to the extracellular Tyrode solution to quench cytosolic indo-1. In most cases, we found that 0.7 mm Mn²⁺ was needed to quench cytosolic indo-1 completely (see Results). This [Mn²⁺] is higher than previously reported data in rat ventricular myocytes, where 0.1 mm extracellular Mn²⁺ was sufficient to quench cytosolic indo-1. The two possible reasons for the higher concentration of Mn²⁺ required in the present study are the much heavier indo-1 load and the presence of a patch pipette which acts as a sink for slow Mn²⁺ entry.

Measurement of cellular [indo-1]

In the absence of Mn²⁺ quench, total indo-1 concentration, [indo]_t, can be estimated by the Ca²⁺-insensitive fluorescence signal, F_c, which is proportional to the total number of intracellular indo-1 molecules. To convert F_c to [indo], F_c values in indo-1 AM-loaded cells are compared with F_c values in test cells loaded to a steady state using a patch pipette containing known concentrations of indo-1 salt. Thus, the [indo] of an indo-1 AM-loaded cell is:

(10)

where V and F_c are the cell volume and isosbestic indo-1 signal of the test cell, respectively, V_s and F_c,s are corresponding values of the standard calibration cell, and [indo]_s is the concentration of indo-1 salt in the patch pipette for the calibration cell. k₁ is the relative fraction of indo-1-accessible cell volume vs. the calibration cell. That is, since in the calibration cell indo-1 salt only has access to the non-mitochondrial space, which is ∼70 % of cell volume in ventricular myocytes (Bers, 1991), but to 100 % of the cell in indo-1 AM-loaded cells, then k₁ = 0.7/1 = 0.7. Since cell volume is proportional to its membrane capacitance (C_m; Satoh, Delbridge, Blatter & Bers, 1996), V_s/V can be replaced with the corresponding C_m ratio. The value for k₂ is 1.5, reflecting 50 % of indo-1 binding to cytosolic proteins (Blatter & Wier, 1990; Hove-Madsen & Bers, 1992). For a typical cell, C_m = 80 pF if [indo]_s = 0.5 mm and F_c,s is ∼1 BU. Thus, if an indo-1 AM-loaded cell has a membrane capacitance of 100 pF and F_c = 1.12 BU, then the total intracellular [indo] is 0.49 mm:

Since F_c is proportional to the [indo], the decline in F_c during whole-cell dialysis or Mn²⁺ quench can easily be converted to indo-1 concentrations using eqn (10) and:

(11)

where [indo](t) and F_c(t) are the values at time t, and [indo](0) and F_c(0) are initial values.

In the presence of Mn²⁺ in the bath, the non-quenched intra-mitochondrial indo-1 concentration [indo]_m can also be calculated using a variation of eqn (10) where k₁ also reflects the 30 % of cell volume which is mitochondria, or k₁ = 0.7/0.3 = 2.3.

In the case of whole-cell dialysis of indo-1, the wash-out time constant is dependent on cell volume and access resistance, R_acc. To correct for cell to cell differences in R_acc and cell size, the recording time can be normalized by (Pusch & Neher, 1988; Zhou & Neher, 1993a):

(11a)

where t is the original recording time, and t_n is the normalized t for a ventricular cell. The cell size is reflected by the corresponding C_m (in pF). R_acc is in megaohms, and 300 is the value for R_accC_m for the standard cell in Fig. 2 (i.e. for R_acc = 3 MΩ, C_m = 100 pF and t_n = t).

Strategies for assay of mitochondrial [Ca²⁺]_m

Real time mitochondrial Ca²⁺ uptake was detected by two different methods. The first method is based on comparing two Ca²⁺-sensitive signals of the cell (contraction and fluorescence of indo-1, as R) which are measured simultaneously during depolarization. The principle of this approach is that cell shortening reflects a purely cytosolic Ca²⁺ signal (see below), while the indo-1 signal reports a mixed Ca²⁺ signal from both cytosol and mitochondrial compartments (at least in the absence of Mn²⁺ or Ru360). Thus, any major kinetic difference between cytosolic [Ca²⁺]_c and [Ca²⁺]_m can be detected by superimposition of the two signals. The advantage of this method is that the cells are in nearly physiological solutions without exposure to Mn²⁺, which could potentially influence mitochondrial Ca²⁺ uptake (Gunter et al. 1994; Hansford, 1994). The disadvantage is that it is not particularly quantitative.

The second method is based on the selective quenching of cytosolic indo-1 by Mn²⁺ (Miyata et al. 1991). In this case, the non-quenched indo-1 (present only in mitochondria) provides a pure mitochondrial Ca²⁺ signal (as [Ca²⁺]_m, using eqn (7)). Furthermore, a pure cytosolic [Ca²⁺] signal can be estimated using cell contraction as a ‘biosensor’ (see below). The advantage of this method is that the relationship between the [Ca²⁺]_c and [Ca²⁺]_m can be evaluated more quantitatively. The disadvantage of this method is the presence of Mn²⁺. However, the two methods are expected to produce complementary results and, therefore, both are used in the present work.

Assay of mitochondrial Ca²⁺ influx

Ca²⁺ influx into intracellular organelles (such as mitochondria) cannot be readily measured by membrane currents using voltage clamp. However, a new optical method using fura-2 has been used to measure transmembrane Ca²⁺ influx (independent of membrane current measurement; Zhou & Neher, 1993b). The method is based on the fact that F_d of eqn (2) is proportional to the total number of indo-1 molecules bound to Ca²⁺. Upon Ca²⁺ influx, the total amount of Ca²⁺ influx (Q_Ca) is proportional to the change in F_d:

(12)

(13)

where Q_CaB is the fraction of Ca²⁺ influx bound by indo-1, and f_max is a calibration constant which converts the optical signal, ΔF_d, into Ca²⁺ flux under the condition that indo-1 binds 100 % of Ca²⁺ influx (i.e. δ = 1). Because the endogenous Ca²⁺ buffers in cytosol or in mitochondria are finite, δ increases to 1 if [indo] is very high compared with endogenous buffers. The Ca²⁺ buffering capacity of indo-1, κ_B, is (Zhou & Neher, 1993a):

(14)

where K_d is the dissociation constant of indo-1. The total Ca²⁺ binding capacity, κ (bound: free) is:

(15)

where κ_S is the endogenous Ca²⁺ binding capacity. Combining eqns (12)-(15) gives:

(16)

(17)

Equation (17) allows the estimation of Ca²⁺ influx, Q_Ca, at any [indo]. Thus, eqn (17) is an extended form of the Ca²⁺ influx equation given by Zhou & Neher (1993b), which applies only when κ_B≥κ_S. In ventricular myocytes, the cytosolic K_d of indo-1 is ∼3 times larger than K_d of fura-2 used by Zhou & Neher (1993a) in chromaffin cells, and cytosolic Ca²⁺ buffering, κ_S, is ∼2.5 times larger in myocytes than in chromaffin cells (Zhou & Neher, 1993a; Hove-Madsen & Bers, 1993; Berlin et al. 1994). Thus, it is prudent to use eqn (17) for Q_Ca measurement in myocytes. To estimate mitochondrial Q_Ca by ΔF_d, we assumed that the two unknown parameters in eqn (17), K_d and κ_S, are the same in both cytosol and mitochondria. Thus, K_d = 0.84 μm and κ_S = 100 (Berlin et al. 1994). It is seen from eqns (14)-(17) that if κ_B >> κ_S, eqn (17) becomes Q_Ca = ΔF_d/f_max, which is independent of both K_d and κ_S. Thus, larger [indo] makes eqn (17) less dependent on assumptions of K_d and κ_S. In the present work, [indo]_m is in the range 0.5–2 mm. When the cell is not progressively stimulated, [Ca²⁺]_m is < 0.1 μm (see Results). Under these conditions, according to eqn (16), indo-1 will bind 90 % of Ca²⁺ influx, if the mitochondrial [indo]_m is 1 mm.

The calibration constant f_max in eqn (17) can be determined by a special whole-cell patch clamp experiment in chromaffin cells, in which Ca²⁺ influx is purely from voltage-gated Ca²⁺ current, so that Q_Ca can be measured by membrane current, assuming K_d of indo-1 is the same in both chromaffin cells and myocytes. With 2 mm indo-1 in the pipette, during whole-cell dialysis, depolarization pulses are applied to evoke Ca²⁺ influx. The ratio ΔF_d/Q_Ca is measured at each pulse. ΔF_d/Q_Ca increases within 60 s after ‘break-in’ of the patch, and then becomes constant (because after κ_B >> κ_S is reached, indo-1 will bind virtually 100 % of entering Ca²⁺, or δ = 1). The saturated ΔF_d/Q_Ca value is 0.005 BU pC⁻¹, which is f_max for chromaffin cells. We did not use ventricular myocytes to estimate f_max, because sarcoplasmic reticulum (SR) Ca²⁺ flux and Na⁺-Ca²⁺ exchange may complicate Q_Ca measured by Ca²⁺ current.

Since cardiac myocytes are large cells (100–150 μm long) and the fluorescence sensitivity is not completely uniform in the microscope field, f_max was also adjusted for this effect. The fluorescence bead signal is 25 % lower when moved 30 μm from the central point of the microscope. While the middles of cells are placed in the central point of the microscope field they are 10 times longer than the chromaffin cell diameter. Thus the f_max value from chromaffin cells is corrected to f_max = 0.0038 BU pC⁻¹ when used to calculate mitochondrial Ca²⁺ influx in ventricular myocytes.

Cytosolic [Ca²⁺] measured by cell length ([Ca²⁺]_CL)

Myocyte contraction was measured using a video-edge-detection system (Crescent Electronics, Sandy, UT, USA), and is shown as a shortening in length (ΔL). Contraction depends on [Ca²⁺]_c, and the Ca²⁺ transients here activated by Na⁺-Ca²⁺ exchange are very slow. We have taken advantage of the steady-state [Ca²⁺]vs.ΔL relationship as determined by Bassani et al. (1995) to use ΔL as a bioassay for [Ca²⁺]_c. The maximum cell length was measured after a long period of rest, and shortening (ΔL) is reported as a percentage of this maximum cell length (L₀). The dependence of ΔL on [Ca²⁺] can be well described by a Hill equation with a Hill coefficient of 2 (ΔL = ΔL_max/{1 + (K_d,c/[Ca²⁺])²}). The [Ca²⁺] (μm) inferred from the ΔL signal ([Ca²⁺]_CL) is then:

(18)

where [Ca²⁺]_CL is [Ca²⁺]_c estimated from cell contraction, K_d,c is the [Ca²⁺] at half-maximal contraction (0.8 μm) and ΔL_max is the maximum extent of cell shortening at very high [Ca²⁺] (ΔL_max = 37 % of resting cell length). Equation (18) assumes that before any stimulation, when the cell has its maximum length (L = L₀, ΔL = 0%), resting [Ca²⁺]_c = 0.07 μm.

Total cytosolic [Ca²⁺] ([Ca²⁺]_c,t) can be estimated from [Ca²⁺]_c (Hove-Madsen & Bers, 1993):

(19)

where B_max is the total of endogenous Ca²⁺-binding sites (232 μmol (l cytosol)⁻¹) and K_d,c,y is the apparent lumped dissociation constant (455 nM).

RESULTS

Fraction of indo-1 loaded into mitochondria

Strong mitochondrial fluorescence signals were obtained from cells plated at low density (24 000 ml⁻¹) with 5 μm indo-1 AM for 1 h at 37°C, followed by ∼30 min in indo-1- free normal Tyrode solution at room temperature (23°C). Under these loading conditions a total intracellular indo-1 concentration of 0.5–1 mm was attained (based on cells loaded by dialysis with known concentrations of indo-1 salt, eqn (10)).

Figure 2A shows loading and wash-out of indo-1 salt via patch pipettes in a single ventricular myocyte. The cell was first loaded with indo-1 via ruptured patch for 2 min using a pipette containing 1 mm indo-1 salt. At this point the Ca²⁺-independent fluorescence signal (F_c) reached a value of ∼1 BU. Loading was then terminated by removal of the first patch pipette. This cell (and 5 of 12 other attempts) survived this step. The cell was then patch clamped with a second pipette containing the same filling solution, but without any indo-1. There was a small decline in F_c between the removal of the first pipette and patching with the second. This probably reflects redistribution of indo-1 in the cell, because the system is more sensitive to fluorescence in the centre of the field (and cell) than near the edges. Upon patch rupture with the second, indo-1-free pipette, the indicator washed out of the cell with an initial time constant of 200 s (for R_acc = 2 MΩ). This indo-1 wash-out slowed when there was an increase in series resistance (4–7 MΩ), but was largely restored when R_s was returned to 2 MΩ by brief application of suction to the pipette. This 200–300 s time constant is about 10 times longer than required for adrenal chromaffin cells (C_m = 7 pF and cell volume = 1.7 pl), consistent with the ventricular myocytes being 10 times larger (80 pF, or ∼15 pl). While 100 % of the cytosolic indo-1 could be washed out of the ventricular myocyte, the process took ∼40 min. The experiments in Fig. 2A were done with 10 mm EGTA in the patch pipette to prolong cell viability. In cells in which cytosolic Ca²⁺ transients were occurring, this recording lifetime is generally shorter. For these reasons, the very long dialysis time required makes it impractical to completely remove cytosolic indo-1 prior to experiments to assess mitochondrial signals uncomplicated by cytosolic fluorescence. Thus, in the experiments described below using indo-1 AM-loaded cells during dialysis, it can be expected that fluorescence signals emanate from both cytosolic and mitochondrial indo-1.

The experiment in Fig. 2B shows assessment of intramitochondrial indo-1 loading using F_c. The cell was preloaded with indo-1 AM via the standard conditions above. After membrane break-in with an indo-1-free pipette, cytosolic indo-1 was washed out with a time course similar to that in Fig. 2A (reproduced as •), but only 25 % of the indo-1 appears to dialyse out of the cell after indo-1 AM loading. No further reduction in F_c was observed after the cell membrane was damaged by aggressive movement of the patch pipette (last two points in the trace, □). In eight cells, the average fraction of indo-1 which was inaccessible to the dialysing patch pipette was 75 % (range, 65–85 %). This suggests that 75 % of the indo-1 was compartmentalized and most of that is likely to be in mitochondria because it is sensitive to mitochondrial Ca²⁺ transport blockers (see below). Since mitochondria occupy ∼30 % of ventricular myocyte volume (Barth, Stammler, Speiser & Schaper, 1992), a total cellular [indo] of 0.5–1 mm corresponds to cytosolic and mitochondrial indo-1 concentrations of 0.17–0.35 mm and 1.2–2.5 mm, respectively.

Contraction and fluorescence signals induced by depolarization

Due to run-down of Ca²⁺ currents in long (30–50 min) experiments in ruptured patch, and a limited Ca²⁺ entry at each depolarization resulting from channel inactivation, we used strong depolarizations to +110 mV to produce Ca²⁺ influx via Na⁺-Ca²⁺ exchange (see Fig. 3). Ca²⁺ entry via Na⁺-Ca²⁺ exchange does not run down and the duration of the pulse can be varied to alter the amount of Ca²⁺ which enters the cell at each pulse. Indeed, under our conditions a 400 ms pulse to +110 mV brings in a similar amount of Ca²⁺ via Na⁺-Ca²⁺ exchange as does a 400 ms pulse to +10 mV via the Ca²⁺ current (before substantial Ca²⁺ current run-down). Note that in Fig. 3[Ca²⁺]_i continues to rise during the depolarization, such that the pulse duration can be used to vary peak [Ca²⁺]_i. Thus, a 2 s pulse to +110 mV can bring in up to 5 times as much Ca²⁺ as a 2 s pulse to +10 mV. It may be noted that the theoretical reversal potential for Na⁺-Ca²⁺ exchange is ∼0 mV under our experimental conditions. Also, +110 mV is close to the theoretical Ca²⁺ reversal potential, so little or no Ca²⁺ current is expected during these pulses. The inward holding current at -40 mV develops slowly upon cellular dialysis and can be blocked by extracellular tetraethylammonium (TEA). This current might be partly due to inward current through potassium channels (since there is no K⁺ inside but 6 mm K⁺ outside). While the large outward current during the pulse to +110 mV undoubtedly includes a substantial outward Na⁺-Ca²⁺ exchange current, we did not attempt to quantify that current flux explicitly here.

Figure 3. Ca²⁺ influx evoked by strong membrane depolarization.

A cat ventricular myocyte was under whole-cell voltage clamp. Membrane potential (V_m) pulses induced cell shortening (ΔL), which was measured simultaneously with membrane current (I_m) and intracellular Ca²⁺ concentration [Ca²⁺]_i. [Ca²⁺]_i was calculated from the indo-1 signal using eqn (7).

Figure 3 shows a typical example of two sequential 2.4 s pulses to +110 mV in a feline ventricular myocyte. Both the [Ca²⁺]_i signal and cell contraction develop gradually during the depolarization pulses. The slow contraction (ΔL) and Ca²⁺ transient during pulses to +110 mV has another advantage for the present study. That is, contraction can be used to assess the purely cytosolic Ca²⁺ signal, whereas the fluorescence signal comes from indo-1 present in both the cytosol and the mitochondria. The slow rise and decline of [Ca²⁺]_i limits the impact of the kinetic discrepancy between [Ca²⁺]_i and contraction during normal excitation-contraction coupling (where peak [Ca²⁺]_i occurs in < 100 ms). It should also be noted that the 170–350 μm cytosolic indo-1 concentration may slow the time course of the cytosolic Ca²⁺ signal.

Figure 4 shows normalized and superimposed ratios of Ca²⁺ signals (F₄₀₀/F₅₀₀) and contraction traces from two cat cells. Figure 4A shows a 200 ms depolarization to +10 mV where the Ca²⁺ transient is initiated primarily by a Ca²⁺ current and may include SR Ca²⁺ release. Figure 4B shows a 1.6 s pulse to +110 mV and is typical of the type used for most of the experiments reported here. While both traces show similar kinetics between indo-1 and contraction signals, it can be seen that the signals in Fig. 4B are closer in time course, particularly during the declining phase. The major difference is that the indo-1 signal appears to lead the ΔL signal by ∼200 ms (see Fig. 12A). Several factors may contribute to the delay between the rise in [Ca²⁺]_i and ΔL (e.g. the chemical steps between Ca²⁺ binding to troponin C and cross-bridge power stroke, inertial and/or viscous components to cell shortening). Based on experiments described below (Figs 6 and 8), we believe that the indo-1 signals in Fig. 4 are due almost exclusively to cytosolic changes in [Ca²⁺]. This suggests that we can use cell contraction as a bioassay for cytosolic [Ca²⁺] under conditions where mitochondrial indo-1 fluorescence is changing.

Figure 4. Kinetic similarity of cytosolic [Ca²⁺] and cell shortening in ventricular myocytes.

A, contraction (ΔL) and F₄₀₀/F₅₀₀ signals evoked by Ca²⁺ influx via voltage-gated Ca²⁺ channels. A 200 ms depolarization from -40 to +10 mV was applied. B, ΔL and F₄₀₀/F₅₀₀ signals evoked by Ca²⁺ influx via the Na⁺-Ca²⁺ exchanger. A 1.6 s depolarization from -40 to +110 mV was applied. A and B are from two different cat cells, and ΔL and R signals are scaled vertically to allow superimposition. See text for details.

Figure 12. Delay of [Ca²⁺]_m compared with [Ca²⁺]_c.

A, delay of cell contraction (ΔL) after depolarization-induced changes in cytosolic calcium concentration, [Ca²⁺]_c (expanded version of part of Fig. 8). Note the oscillating peaks of [Ca²⁺]_c, recorded by indo-1 fluorescence, precedes that of ΔL. The latency of ΔL is about 200 ms. Since Ru360 blocks mitochondrial Ca²⁺ uptake, F₄₀₀/F₅₀₀ reports only cytosolic Ca²⁺. There was no Mn²⁺ in the bath but 0.2 μm Ru360 in the pipette to block mitochondrial Ca²⁺ uptake. B, delay of mitochondrial [Ca²⁺]_m after depolarization-induced cell contraction in another ferret cell without Ru360. Note the depolarization-induced [Ca²⁺]_m is preceded by cell contraction (shown as percentage of resting cell length, RCL). The latency of [Ca²⁺]_mvs. contraction was 0.8 s. Addition of the 200 ms delay between [Ca²⁺]_c and contraction (A) makes the total delay of [Ca²⁺]_mvs.[Ca²⁺]_c about 1 s. Mn²⁺ (0.7 mm) was in the bath to quench cytosolic indo-1.

Figure 6. Mitochondrial Ca²⁺ uptake in a ventricular myocyte without Mn²⁺ quench.

The cell was preloaded with indo-1 AM, so the total indo-1 signal has contributions from both cytosolic and mitochondrial indo-1. Intracellular [Ca²⁺] is estimated by both indo-1 R ([Ca²⁺]_i or F₄₀₀/F₅₀₀) and by cell contraction ([Ca²⁺]_CL or ΔL). A-C, Ca²⁺ signals upon progressive depolarization pulses no. 21, no. 27 and no. 37, respectively. To assess mitochondrial Ca²⁺ uptake in the upper traces, normalized F₄₀₀/F₅₀₀ is scaled and superimposed on ΔL. The difference between F₄₀₀/F₅₀₀ and ΔL (as marked by the arrows during the decay phase) is interpreted as the presence of mitochondrial Ca²⁺ uptake (see text for details). All data are from the same cat cell.

Figure 8. Block of mitochondrial Ca²⁺ uptake by Ru360.

Experimental conditions in this cell are similar to that in Fig. 6C, except the specific mitochondrial uniporter antagonist Ru360 (1 μm) is included in the patch pipette, and thus dialysed into the cell. In contrast to Fig. 6, the differences between normalized ΔL and F₄₀₀/F₅₀₀ are completely blocked by Ru360 even after progressive Ca²⁺ load (n = 4). The Corrected [Ca²⁺]_i (bottom) is cytosolic [Ca²⁺] estimated by indo-1 fluorescence after subtracting mitochondrial indo-1 fluorescence (eqns (3)-(8), see text for details).

Contraction as a biosensor for [Ca²⁺]_c

For the slow contractions used here, [Ca²⁺]_c can be ‘back’ calculated from the change in cell length (ΔL) using the modified Hill equation (eqn (18)). Resting cell length (L₀) is taken as the cell length after a long period of rest before progressive loading pulses. The maximum degree of contraction (ΔL_max) was determined experimentally as the shortest stable contracture at the end of a series of very long depolarizations to +110 mV for > 5 s, but prior to hypercontracture associated with cell death. This maximal degree of contracture (ΔL_max = 37 ± 2 % of resting cell length, n = 15) was the same for cat and ferret cells. The relationship between ΔL and [Ca²⁺]_c used here (Fig. 5A) was very similar to the steady-state relationship determined by Bassani et al. (1995) in a study of indo-1 calibration in ventricular myocytes. Figure 5B shows an example of [Ca²⁺]_c calculated from the ΔL trace for two different depolarizing pulses. During the break in the record there were five additional pulses accounting for the decrease in resting cell length and increase in diastolic [Ca²⁺]_c. In this cell cytosolic indo-1 was quenched with Mn²⁺ (see below), so there is no cytosolic Ca²⁺ signal from indo-1. Thus, cell contraction is the only indicator of [Ca²⁺]_c.

Figure 5. Pure cytosolic free [Ca²⁺] estimated by contraction signal.

A, model of the Ca²⁺ dependence of cell contraction in ventricular myocytes, based on Bassani et al. (1995), with slight modification in parameters (see text and eqn (18)). This model was used in B for ‘back’ calculation of cytosolic [Ca²⁺]_c. B, the ferret cell was stimulated by repetitive depolarization pulses to +110 mV, two of which are shown here. Not shown are five pulses during the 380 s break, which increased the resting Ca²⁺ level. The cytosolic Ca²⁺ is calculated from the contraction signal, ΔL, using eqn (18) as in A. The subscript ‘CL’ replaces ‘c’ for [Ca²⁺]_c estimations based on cell length ([Ca²⁺]_CL). The cytosolic indo-1 in this cell was quenched by 0.7 mm Mn²⁺ in the bath.

Mixed indo-1 signals and mitochondrial Ca²⁺ uptake in the absence of Mn²⁺

A first series of experiments was done in the absence of Mn²⁺ quench, when the indo-1 fluorescence signal comes from both cytosol and mitochondria. In this case we used the ΔL signal to assess the [Ca²⁺]_c and compared the kinetics with those of the overall mixed indo-1 fluorescence signal. The three sets of traces in Fig. 6 represent different times during a long series of progressive Ca²⁺ loading pulses. The first pair of pulses were of 1.6 s duration (Fig. 6A). At this stage diastolic [Ca²⁺]_c was less than 400 nM and the pulses produced normalized fluorescence signals (F₄₀₀/F₅₀₀) with kinetics which were superimposable with the ΔL signal (except for the usual delay of the rising phase). The global indo-1 signal ([Ca²⁺]_i) predicts a much smaller Ca²⁺ transient than [Ca²⁺]_c derived from the ΔL signal. We infer that the cytoplasmic Ca²⁺ transient must be greatly damped by either an unchanging low mitochondrial [Ca²⁺]_i or one that is very much smaller in amplitude (but with the same kinetics).

There were six pulses between Fig. 6A and B, and the pulse duration for Fig. 6B was 2.4 s. As diastolic and peak [Ca²⁺]_c increased, a difference in the kinetics of decline of the fluorescence F₄₀₀/F₅₀₀ signal became apparent. That is, relaxation still occured promptly upon repolarization, but there was a long tail of slowed F₄₀₀/F₅₀₀ decline. At this stage the diastolic [Ca²⁺]_c was increased to ∼400 nM. The slower decline in F₄₀₀/F₅₀₀ could reflect an increase in mitochondrial [Ca²⁺] during the declining phase of the cytosolic [Ca²⁺]. This effect was even more dramatic after an additional eight pulses of 2.4 s, and further elevated diastolic [Ca²⁺]_i to > 400 nM (Fig. 6C). In this case the global [Ca²⁺]_i increased slowly while [Ca²⁺]_c was declining, as evidenced by the cell relaxation. Similar results were observed in all nine cells studied with this protocol. These results suggest that mitochondrial Ca²⁺ uptake may be slow compared with [Ca²⁺]_c decline. In addition, since most of the cellular indo-1 is in the mitochondria, the global [Ca²⁺]_i signal should be very sensitive to mitochondrial [Ca²⁺] changes. This is consistent with changes in [Ca²⁺]_ivs.[Ca²⁺]_CL: [Ca²⁺]_i < [Ca²⁺]_CL in Fig. 6A and B, but [Ca²⁺]_i > [Ca²⁺]_CL in Fig. 6C. To confirm that the slowed decline of global [Ca²⁺]_i during relaxation (or even [Ca²⁺]_i increase) is really due to mitochondrial Ca²⁺ uptake, we performed additional experiments using the mitochondrial Ca²⁺ uniport inhibitor Ru360 in the pipette.

Mitochondrial Ca²⁺ uniport inhibitor Ru360

The action of Ru360 on the rate of Ca²⁺ uptake into isolated mitochondria was tested to determine its effectiveness in vitro. The rate of Ca²⁺ uptake into isolated mitochondria was inhibited by Ru360 in a concentration-dependent manner (Fig. 7). The half-maximum inhibition of Ca²⁺ uptake rate (IC₅₀) based on the curve in Fig. 7 was 0.19 nM. This indicates that Ru360 is > 30 times more potent than Ruthenium Red, a well-known mitochondrial Ca²⁺ uptake inhibitor. The IC₅₀ and data correspond to a 99 % inhibition at 20 nM Ru360. These data indicate that Ru360 is an effective and potent inhibitor of Ca²⁺ uptake into mitochondria.

Figure 7. Inhibition of the rate of Ca²⁺ uptake into mitochondria in vitro.

The maximum rate of Ca²⁺ uptake into mitochondria upon addition of Ca²⁺ is plotted vs.[Ru360] in the assay medium. The maximum rate of Ca²⁺ uptake without Ru360 was 122.7 ± 8.6 nmol Ca²⁺ min⁻¹ (mg protein)⁻¹. Each data point represents mean ± s.e.m. of four experiments with two mitochondrial preparations. The curve is a simple Michaelis-Menten fit: V = V_max(1 - 1/(1 + IC₅₀/[Ru360])), where V is the rate of mitochondrial Ca²⁺ uptake and IC₅₀ is the [Ru360] required for 50 % inhibition (0.193 nM or 0.32 ng (mg mitochondrial protein)⁻¹; MW, 550.5).

Figure 8 shows an experiment like that in Fig. 6, but where 1 μm Ru360 was included in the patch pipette. At this stage, where diastolic [Ca²⁺]_i is very high (e.g. like Fig. 6C) the cell showed spontaneous oscillations characteristic of Ca²⁺ overload. Even at this extremely high cellular Ca²⁺ load with [Ca²⁺]_i oscillations, the contraction and [Ca²⁺]_i signals are still well matched kinetically. This confirms that the kinetic disparity in Fig. 6C is due to Ca²⁺ uptake into mitochondria and can be completely suppressed by Ru360. The [Ca²⁺]_c oscillations probably reflect SR Ca²⁺ release and reuptake because Ru360 (up to 1 μm) does not affect either SR Ca²⁺ uptake or SR Ca²⁺ release in permeabilized cells (Matlib et al. 1998) in contrast to high concentrations of Ruthenium Red.

Since there is no mitochondrial Ca²⁺ uptake in Fig. 8, the change in indo-1 fluorescence comes exclusively from the cytosol. In this case we can subtract the constant mitochondrial fluorescence at both wavelengths to obtain a ‘corrected [Ca²⁺]_i’ signal (see eqn (8), Methods). The raw [Ca²⁺]_i in Fig. 8 is much lower than either the corrected [Ca²⁺]_i or [Ca²⁺]_CL. This is probably because intramitochondrial [Ca²⁺] is very low in the presence of intracellular Ru360. Ideally the corrected [Ca²⁺]_i and [Ca²⁺]_CL would give the same signal via independent calibrations. While the kinetics are the same, the moderate differences probably reflect inaccuracies in the constants or forms of the equations for these Ca²⁺ concentrations (eqns (7), (8), or (18)).

Mitochondrial [Ca²⁺] measured with Mn²⁺ quench of cytosolic indo-1

Selective quenching of cytosolic indo-1 by Mn²⁺ allows detection of a relatively pure mitochondrial [Ca²⁺]_m in cells loaded with indo-1 AM (Miyata et al. 1991). Figure 9A shows the time course of Mn²⁺ quench of purely cytosolic indo-1 (which had been loaded as the salt via patch pipette). Addition of 0.7 mm Mn²⁺ to the bath quenched 92 % of the indo-1 fluorescence with a time constant of 240 s. The remaining 8 % of indo-1 fluorescence is lost when the membrane is ruptured. This 8 % of F_c is also not sensitive to increasing Mn²⁺ concentration and probably reflects residual fluorescence of Mn²⁺-bound cytosolic indo-1. This is consistent with in vitro experiments where Mn²⁺ quenches only 95 % of indo-1 fluorescence (not shown). This small fraction of unquenched fluorescence is very minor for the present work because 75 % of intracellular indo-1 is mitochondrial. However, in cells with either less mitochondrial indo-1 loading or a smaller fraction of cell volume occupied by mitochondria, > 90 % of the dye can be cytosolic (Zhou & Neher, 1993a). In this case up to 50 % of the residual signal could be Mn²⁺-bound indo-1 fluorescence, which could cause serious artifacts.

Figure 9. Time course of Mn²⁺ quench of cytosolic indo-1.

A, the cell was loaded by a patch pipette containing 1 mm indo-1 for 8 min whole-cell dialysis. After withdrawing the patch pipette, 0.7 mm Mn²⁺ was added to the bath. The time course of cytosolic indo-1 quenched by Mn²⁺ is monitored by the normalized F_c signal. After about 15 min, F_c becomes stable. Additional 0.3 mm Mn²⁺ does not reduce F_c further (not shown), indicating all of the cytosolic indo-1 is quenched by Mn²⁺. However, breaking the cell membrane by a patch pipette reduces F_c to the original level before loading indo-1. B, the cell was preloaded with indo-1 AM. Addition of 0.7 mm Mn²⁺ reduces F_c by 29 % after > 20 min. According to A, cytosolic indo-1 is completely quenched by Mn²⁺ at this time. This suggests that Mn²⁺-resistant indo-1 (71 % of total indo-1) is not located in the cytosol, but in cell organelles.

Figure 9B shows the time course of Mn²⁺ quench in a typical cell loaded with the AM form of indo-1. After 10 min of Mn²⁺ quench, F_c is reduced by only 29 %, indicating that ∼71 % of the total cellular indo-1 is in mitochondria. These experiments (n = 6) are consistent with the average of 75 % mitochondrial indo-1 loading from Fig. 2, and confirm that most of the cellular indo-1 with our standard indo-1 AM-loading method is in mitochondria.

This situation increases the sensitivity of our mitochondrial Ca²⁺ uptake measurement for several reasons. First, our method loads mitochondria to an [indo] of 1–2 mm, which gives a fluorescence 50–100 times larger than the autofluorescence. Second, the high mitochondrial [indo] also gives a better signal-to-noise ratio, partly because most of the Ca²⁺ which enters the mitochondria will be bound by indo-1 (and produce signal). The heavy loading also allows long-term recording. Third, the whole-cell patch clamp gradually dialyses out much of the cytosolic indo-1. Mn²⁺ quench of the remaining cytosolic indo-1 also eliminates interference by [Ca²⁺]_c, yielding a pure mitochondrial signal. Finally, patch clamp allows larger Ca²⁺ influx via Na⁺-Ca²⁺ exchange during a single depolarization, helping to make mitochondrial Ca²⁺ uptake detectable within the pulse.

Figure 10 shows pure mitochondrial Ca²⁺ signals during single depolarizations in a ferret myocyte. Figure 10A shows an early 400 ms pulse from -40 to +110 mV. In contrast to the moderate contraction signal (ΔL) and inferred cytosolic Ca²⁺ signal ([Ca²⁺]_CL), there is no increase in [Ca²⁺]_m, indicating the absence of mitochondrial Ca²⁺ uptake at this [Ca²⁺]_c level.

Figure 10. Mitochondrial Ca²⁺ uptake during Mn²⁺ quench of cytosolic indo-1 in a ferret cell.

To measure both pure cytosolic and pure mitochondrial Ca²⁺, the cytosolic indo-1 was quenched by 0.7 mm Mn²⁺ in the bath. A-C, signals evoked by three individual depolarization pulses to +110 mV. Pulse number represents the corresponding number of depolarization pulses applied to this cell. [Ca²⁺]_CL and [Ca²⁺]_m are estimated by contraction and indo-1 fluorescence, respectively. D, contraction signals from five individual pulses superimposed and labelled with the pulse number. E and F, [Ca²⁺]_CL and [Ca²⁺]_m from the five pulses shown in D.

Figure 10B shows that after seven more pulses, a subsequent 0.8 s depolarization induced a large contraction, a large [Ca²⁺]_CL and a small but detectable [Ca²⁺]_m signal. Note the increased resting ΔL and [Ca²⁺]_CL. After fifteen additional pulses (Fig. 10C) a similar ΔL and Δ[Ca²⁺]_CL are evoked, but from substantially elevated resting levels. The pulse in Fig. 10C produces a substantial increase in [Ca²⁺]_m. Figure 10D-F shows five representative and superimposed depolarization-induced contractions (D), Δ[Ca²⁺]_CL (E) and Δ[Ca²⁺]_m (F). These panels emphasize that very little change in [Ca²⁺]_m is observed at physiological levels of resting [Ca²⁺]_c and cell length. It may be noted that a typical twitch contraction in intact ventricular myocytes is from a resting value of 0 to a peak of ∼10 %. Substantial phasic [Ca²⁺]_m changes during the pulse are only apparent when the resting cell is substantially contracted and [Ca²⁺]_c is greatly elevated (> 400 nM). As phasic Δ[Ca²⁺]_m becomes detectable there is also a slow progressive rise in resting [Ca²⁺]_m, consistent with the very slow decline in [Ca²⁺]_m, compared with [Ca²⁺]_c (Fig. 10Evs. F). Thus, mitochondrial Ca²⁺ appears to accumulate with progressive pulses once uptake begins.

These results are entirely consistent with results from the Mn²⁺-free data of Fig. 6. Comparing Figs 6 and 10 it is clear that (with or without Mn²⁺ quench), mitochondria start to take up Ca²⁺ at similar basal [Ca²⁺]_c levels (0.3–0.5 μm), and mitochondrial Ca²⁺ uptake is larger at high [Ca²⁺]_c levels. This may suggest that 0.7 mm Mn²⁺ in the bath does not affect mitochondrial Ca²⁺ uptake significantly. It should be noted that cytosolic Mn²⁺ may be mostly bound to indo-1, such that free cytosolic [Mn²⁺] might be quite low in our conditions. Indeed, contraction amplitude is not different in the Mn²⁺-quenched cell (which could be depressed if cytosolic free [Mn²⁺] were high). As it happens, extracellular Mn²⁺ does not block Ca²⁺ entry via Na⁺-Ca²⁺ exchange very strongly, but does block Ca²⁺ current. This would make it difficult to do these same experiments using Ca²⁺ current as the source of Ca²⁺ influx rather than Na⁺-Ca²⁺ exchange.

We have used Mn²⁺ and Ru360 to block separately cytosolic and mitochondrial indo-1 signals, respectively (Figs 10 and 7). Figure 11 shows that combining Mn²⁺ and Ru360 can completely block all of the indo-1 signal. In Fig. 11A, cytosolic indo-1 has been quenched with 0.7 mm Mn²⁺ in the bath. F_c indicates the level of Ca²⁺-insensitive indo-1 in the mitochondria, which is about 0.6 mm (see Methods). Upon progressively strong depolarization, indo-1 fluorescence reported increases in [Ca²⁺]_m, and contraction reported large Δ[Ca²⁺]_CL and elevated resting [Ca²⁺]_c. Figure 11B shows a similar experiment, except that 1 μm Ru360 was included in the patch pipette. Despite the higher concentration of indo-1 in this cell (F_c implies about 2 mm mitochondrial indo-1), very strong depolarization did not increase [Ca²⁺]_m (or indo-1 signal) at all. This was the case even though extremely strong contractions indicated large and oscillatory changes in [Ca²⁺]_c. These experiments (n = 3) further confirm that mitochondrial Ca²⁺ uptake can be fully blocked by cytosolic Ru360. The small and transient increase in [Ca²⁺]_m is probably secondary to displacement of cytosolic Mn²⁺ from indo-1 by the very large cytosolic Ca²⁺ transient (Fig. 11B inset, see below).

Figure 11. Blockade of indo-1 signals by Mn²⁺ and Ru360.

A, Mn²⁺ alone does not block indo-1 signals from mitochondria in a ferret ventricular myocyte. Mn²⁺ (0.7 mm) was in the bath to quench cytosolic indo-1, but no Ru360 was in the pipette. Thus the indo-1 signal indicates [Ca²⁺]_m exclusively. Before the two pulses shown here, the cell had been stimulated to a high level of Ca²⁺ load (note resting [Ca²⁺]_CL). The Ca²⁺-insensitive fluorescence, F_c, which is directly proportional to mitochondrial [indo] ([indo]_m) is given by eqn (11). B, similar experiment to A in another cell, but with 1 μm Ru360 in the patch pipette (as well as Mn²⁺ quench). Despite even stronger Ca²⁺-loading pulses, as seen by [Ca²⁺]_CL, there is almost no mitochondrial indo-1 signal. This suggests that cytosolic and mitochondrial indo-1 signals are completed blocked by Mn²⁺ and Ru360. The small and transient changes in [Ca²⁺]_m (inset) may be due to slight displacement of cytosolic indo-1-bound Mn²⁺ by Ca²⁺ (see text). Note F_c in B is larger than in A, indicating [indo]_m is larger in the cat cell (B). See text for details.

Latency of [Ca²⁺]_m with respect to [Ca²⁺]_c

The combined [Ca²⁺]_c and [Ca²⁺]_m measurements allow us to assess the temporal latency between these signals. Figure 12A shows the temporal relationship between cytosolic [Ca²⁺]_c and ΔL in a ventricular cell. These traces are expanded versions of part of the right-hand pulse from Fig. 8. There was 0.2 μm Ru360 in the pipette to block mitochondrial Ca²⁺ uptake, and thus the changes of indo-1 signal reflect purely cytosolic Ca²⁺ binding of indo-1 (see Fig. 8). It can be seen that the contraction signal tracks the cytosolic indo-1 signal, [Ca²⁺]_c, except that there is a 200 ms latency for [Ca²⁺]_CL calculated from contraction with respect to that based on indo-1 fluorescence ([Ca²⁺]_c). Because Ca²⁺ binding with indo-1 is much faster than the contraction delay, 200 ms is the intrinsic latency of [Ca²⁺]_CL. The mean latency of ΔL vs. F₄₀₀/F₅₀₀ is 222 ± 62 ms (n = 10). Thus, to compensate for this latency where both signals are not available, [Ca²⁺]_CL should be shifted ∼200 ms earlier in time when comparing with [Ca²⁺]_m timing.

Figure 12B shows a direct comparison of contraction (ΔL) with [Ca²⁺]_m (estimated from the pure mitochondrial indo-1 signal after Mn²⁺ quench without Ru360). The lower panel shows an enlargement of the rising phases and includes the inferred [Ca²⁺]_CL time course (estimated from the ΔL) and offset by 200 ms. It can be seen that the peak contraction (ΔL) precedes the indo-1-based peak [Ca²⁺]_m by ∼800 ms. With the 200 ms delay for the rising phase of contraction vs.[Ca²⁺]_c, the overall delay in time to peak of [Ca²⁺]_m from that of [Ca²⁺]_c is ∼1.0 s.

Ca²⁺ influx into mitochondria

When cytosolic indo-1 is quenched with Mn²⁺, the Ca²⁺-sensitive mitochondrial indo-1 F_d signal can be used to measure the amount of Ca²⁺ influx into mitochondria during the pulse (see Methods and eqn (17)). Figure 13 shows an experiment in which mitochondrial Ca²⁺ uptake (Q_Ca,m) is measured in this way. In addition, [Ca²⁺]_m based on the indo-1 fluorescence ratio and [Ca²⁺]_CL based on ΔL are shown. It may be noted that this recording is at a time when the progressive Ca²⁺ loading of the cell has already elevated resting [Ca²⁺]_c to 800 nM. The intramitochondrial indo-1 concentration, [indo]_m, was 0.6 mm in this cell (using eqn (10)). During each of the 1.2 s pulses shown in Fig. 13, cytosolic [Ca²⁺]_CL increased by 0.4 μm, mitochondrial [Ca²⁺]_m increased by 0.2 μm, and the mitochondria in this cell took up ∼20 pC (corresponding to 13 μmol (l mitochondria)⁻¹, or 0.104 fmol) of Ca²⁺. The maximum rate of mitochondrial Ca²⁺ uptake during the [Ca²⁺]_c spike is about 20 pC s⁻¹ in this cell. In cellular volume units this corresponds to a Ca²⁺ flux of ∼6 μmol (l cytosolic volume)⁻¹ s⁻¹.

Figure 13. Real time Ca²⁺ influx into mitochondria.

Depolarization to +110 mV was applied to a ferret cell with Mn²⁺ quench of cytosolic indo-1. [Ca²⁺]_m and [Ca²⁺]_CL are estimated by ΔL and mitochondrial indo-1 fluorescence using eqns (7) and (18), respectively. The net mitochondrial Ca²⁺ influx, Q_Ca,m, was measured by the Ca²⁺-sensitive indo-1 signal, ΔF_d (see eqn (17)). Scale for Q_Ca,m is given in both Ca²⁺ charge (30 pC) and total mitochondrial Ca²⁺ concentration (17 μm), assuming a total mitochondrial volume of 9 pl for this cell.

Figure 14 shows the progression of a typical experiment as a function of the gradual rise in resting [Ca²⁺]_c during twenty-two pairs of depolarizing pulses in a cell after Mn²⁺ quench. As resting [Ca²⁺]_CL rises in Fig. 14A, resting [Ca²⁺]_m declines to 100–150 nM and does not increase much until [Ca²⁺]_CL is above 400 nM. Figure 14B shows Δ[Ca²⁺]_CL and Δ[Ca²⁺]_m per pulse as a function of resting [Ca²⁺]_CL. Figure 14C is similar, except the free [Ca²⁺] is replaced by total [Ca²⁺] ([Ca²⁺]_t) using eqns (17) and (19). At 200 nM [Ca²⁺]_CL, increasing Ca²⁺ influx (Δ[Ca²⁺]_CL) at two individual pulses to levels comparable with those attained at greater [Ca²⁺]_CL (> 400 nM) failed to produce detectable Δ[Ca²⁺]_m. This indicates that it is not only the Δ[Ca²⁺]_c amplitude that causes mitochondrial Ca²⁺ uptake, but also a threshold level of resting [Ca²⁺]_c. Here the threshold is at about 400 nM [Ca²⁺]_CL where mitochondrial Ca²⁺ uptake starts to become detectable. There also seems to be some sort of facilitation of mitochondrial Ca²⁺ uptake. That is, for a given sarcolemmal Ca²⁺ influx, or Δ[Ca²⁺]_CL, mitochondrial Ca²⁺ uptake is greater at higher basal [Ca²⁺]_c (or higher [Ca²⁺]_m) than at lower basal [Ca²⁺]_CL (or [Ca²⁺]_m). Similar results were seen in three other cells.

Figure 14. Ca²⁺ dependence of mitochondrial Ca²⁺ uptake during progressive Ca²⁺ loading.

A, steady-state relationship of resting [Ca²⁺]_CL and [Ca²⁺]_m, measured just prior to test pulses. B, resting [Ca²⁺]_CLvs. depolarization-induced changes in cytosolic (Δ[Ca²⁺]_CL) and mitochondrial (Δ[Ca²⁺]_m) Ca²⁺ concentration. C, resting [Ca²⁺]_CLvs. depolarization-induced total (free + bound) changes in cytosolic (Δ[Ca²⁺]_c,t) and mitochondrial (Δ[Ca²⁺]_m,t) Ca²⁺ concentration. Pulses to +110 mV ranged from 0.2 s to 1.2 s in duration in this ferret myocyte. Cytosolic indo-1 was quenched with 0.7 mm Mn²⁺ in the bath, and the total experimental time was 35 min.

Transient relief of cytosolic Mn²⁺ quench of indo-1 with strong [Ca²⁺]_CL pulses

In Fig. 11B, with Mn²⁺ quench there is a small, but apparent, transient increase in the [Ca²⁺]_m signal during each of the very strong depolarizations, despite the presence of Ru360. This is not due to a contraction-induced artifact since [Ca²⁺]_m is based on the fluorescence ratio. As shown in Fig. 11B inset, the depolarization induced a small artifactual ΔF_c (because of the non-uniform optical effect in the microscope field), which is kinetically identical to the normalized ΔL. The rise in ΔF_c also corresponds to the apparent rise in [Ca²⁺]_m, but [Ca²⁺]_m declines rapidly with oscillations which are consistent with [Ca²⁺]_CL changes. Thus, the apparent Δ[Ca²⁺]_m is not due to a contraction artifact, but could be reflecting a very small cytosolic signal. The transient [Ca²⁺]_m is also not due to unblocking of the Ca²⁺ uniporter by Ru360 because the Δ[Ca²⁺]_m kinetics are too fast (see Fig. 10). Also, the Ru360 concentration is ∼1000 times higher than the K_i for block. Furthermore, the basal [Ca²⁺]_m (< 100 nM) is lower than the threshold for mitochondrial uptake (see Fig. 14).

We suggest that the very small transient rise in apparent [Ca²⁺]_m in Fig. 11B is due to Ca²⁺ influx-induced transient Mn²⁺ unbinding (relieving quench) and Ca²⁺ binding of cytosolic indo-1. This interpretation seems reasonable because we have used the minimum extracellular [Mn²⁺] to stably quench cytosolic indo-1. Thus, cytosolic [Mn²⁺], while unknown, is probably only a little higher than that required to bind all cytosolic indo-1 in competition with Ca²⁺ at resting [Ca²⁺]_c levels. When local[Ca²⁺]_c rises to high levels it is reasonable to expect Ca²⁺ to compete and displace a small amount of Mn²⁺ from indo-1 (e.g. near Na⁺-Ca²⁺ exchange sites), thus relieving some of the Mn²⁺ quench and producing some cytosolic Ca²⁺-indo-1 fluorescence signal. The transient unbinding of Mn²⁺-indo-1 would increase the total non-quenched [indo] in the cell and produce a transient F_c component that is not paralleled by contraction, which is indeed observed (data not shown). This interpretation is also consistent with our observation that when we remove Mn²⁺ completely from the bath, there is some recovery of cytosolic Ca²⁺ signals (n = 5). Thus, the small apparent rise in [Ca²⁺]_m in Fig. 11B is probably artifactual, but given its small size it is unlikely to alter any of our conclusions with respect to mitochondrial Ca²⁺ transport.

Mn²⁺ uptake by mitochondria

Manganese can permeate the plasma membrane and quench cytosolic indo-1 when present in the bath (Miyata et al. 1991; Blatter, 1995; this work). This raises the question of whether Mn²⁺ can also enter mitochondria after it enters the cytosol. Figure 9 shows that Mn²⁺ can quench virtually all of the cytosolic indo-1 within 10 min, while the majority of mitochondrial indo-1 remains unquenched even after > 30 min. However, this result is from non-stimulated cells.

After Mn²⁺ has quenched the cytosolic indo-1, changes in F_c can be used as a Ca²⁺-insensitive monitor for Mn²⁺ quench of mitochondrial indo-1 as well. In isolated mitochondria, it is known that Mn²⁺ can compete with Ca²⁺ for entry into mitochondria (Gavin et al. 1990). In our intact cells, we assume that nearly all Mn²⁺ entry into mitochondria is bound by indo-1 due to the high mitochondrial [indo]. Thus the amount of the Mn²⁺ uptake can be estimated by the quench-induced decline in the F_c signal. Figure 15C shows the time course of F_c in a cell. At each data point the cell was stimulated by a pair of depolarization pulses (similar to Fig. 2A; pulse duration, 0.8–1.2 s). For the first 1200 s, F_c did not decrease even during repetitive stimulation of the cell. During this period, when mitochondrial Mn²⁺-free indo-1 concentration was about 1.0 mm (estimated by eqn (11)), there was no mitochondrial Ca²⁺ uptake either. The absence of mitochondrial uptake of Ca²⁺ and Mn²⁺ during this time is evident in Fig. 15A (recorded at the time of the first arrow in Fig. 15C). In Fig. 15A there is neither a detectable decrease in F_c nor an increase in [Ca²⁺]_m with the stimulation. However, when the mitochondria start to take up Ca²⁺, after the first arrow in Fig. 15C, then F_c starts to decrease. This is evident from the progressive decrease in F_c and increase in [Ca²⁺]_m upon stimulation. In Fig. 15B (taken at the second arrow in Fig. 15C) stimulation causes F_c to decrease by 0.01 BU, while [Ca²⁺]_m increases by 140 nM. The decrease in F_c during the stimulation corresponds to 10 % of the unquenched indo-1, which is bound by Ca²⁺ during the stimulation. This suggests that, during the stimulation 10 times more Ca²⁺ than Mn²⁺ enters mitochondria. On the other hand, during the time between two stimuli, while there is extrusion of Ca²⁺ from mitochondria (evident from the slow [Ca²⁺]_m decline), net Mn²⁺ influx into mitochondria continues (evident from the F_c decline in Fig. 15B and C). This is entirely consistent with the hypothesis that Mn²⁺ is a Ca²⁺ substitute for mitochondrial uptake. These experiments (n = 4) suggest the following properties of mitochondria in intact cells exposed to sub-millimolar levels of Mn²⁺. (1) Mn²⁺ uptake is dependent on the Ca²⁺ uptake process. Before [Ca²⁺]_c (or [Ca²⁺]_m) reach the apparent threshold of Ca²⁺ uptake, there is no Mn²⁺ uptake. (2) Ca²⁺ uptake during a strong (1–2 s) depolarization is much larger than Mn²⁺ uptake under our experimental conditions. Thus, during a single pulse the artifact introduced by Mn²⁺ quench is negligible. (3) After the start of Ca²⁺ and Mn²⁺ uptake, mitochondrial Mn²⁺ uptake continues, even during the diastolic period when mitochondrial Ca²⁺ efflux is also occurring.

Figure 15. Ca²⁺-induced Mn²⁺ uptake coincides with Ca²⁺ uptake.

The cytosolic indo-1 was quenched by 0.7 mm Mn²⁺ in bath. At the beginning of the recording the F_c value (≈0.55 BU) corresponded to an [indo]_m of ≈1.0 mm. A and B, pulses indicated by the first and second arrows in C, respectively. (Dashed lines indicate the time of stimulation.) Note the time course of F_c (or [indo]_m) decreases as Δ[Ca²⁺]_m goes up, suggesting that Mn²⁺ uptake is dependent on the same system that takes up Ca²⁺ into mitochondria. Note, in C, the Ca²⁺ uptake is shown in Δ[Ca²⁺]_m divided by pulse duration. The increase in Δ[Ca²⁺]_m s⁻¹ indicates a facilitation of Ca²⁺ uptake with increasing resting [Ca²⁺]_m.

DISCUSSION

We have addressed three open questions concerning mitochondrial Ca²⁺ in ventricular myocytes, finding that (1) mitochondria take up Ca²⁺ in intact cells in physiological solutions without Mn²⁺ (but only when [Ca²⁺]_c is elevated); that (2) the maximum rate of mitochondrial Ca²⁺ uptake is 6 μmol (l cytosol)⁻¹ s⁻¹; and that (3) the kinetics of [Ca²⁺]_m are significantly different from [Ca²⁺]_c. We have used preferential loading of mitochondria with indo-1, cytosolic dialysis, Mn²⁺ quench of cytosolic indo-1, a selective mitochondrial Ca²⁺ uniport blocker (Ru360) and cell contraction as a bioassay of [Ca²⁺]_c in a complementary fashion to address these issues.

Contraction as a bioassay of [Ca²⁺]_c

Muscle contraction has been used for many years as a semi-quantitative indicator of [Ca²⁺]_c. Indeed, the [Ca²⁺] dependence of isometric contractile force has been extensively characterized in both skinned and intact cardiac muscle (Yue, Marban & Wier, 1986; Bers, 1991; Gao, Backx, Azan-Backx & Marban, 1994). The steady-state [Ca²⁺]_c dependence of cell shortening in isolated myocytes has also been characterized (Spurgeon et al. 1990; Bassani et al. 1995). Here we have used this quantitative relationship to infer a purely cytosolic [Ca²⁺] signal (distinct from a fluorescence signal, which includes a mitochondrial component). During the rapid rising phase of the Ca²⁺ transient in physiological excitation-contraction (E-C) coupling, the [Ca²⁺]_c-length relationship is different from the steady state (Spurgeon et al. 1990). The slow Ca²⁺ transients used here minimize this difference during the rising phase (∼200 ms delay) and during [Ca²⁺]_c decline. Thus, the steady-state relationship should provide a good approximation for these data. Indeed, our results show good kinetic agreement for contractions and indo-1 signals where mitochondrial Ca²⁺ uptake does not occur (Figs 3, 4, 6A, 8 and 12A).

Ionic currents via Ca²⁺-activated K⁺ channels have also been used as biosensors of local [Ca²⁺] near the membrane in hair cells, smooth muscle and chromaffin cells (Stehno-Bittel & Sturek, 1992; Prakriya, Solaro & Lingle, 1996). The contractile proteins in our case provide information about mean cytosolic [Ca²⁺], which is appropriate for our purposes.

In addition to the kinetic constraints mentioned above, there are several limitations in using contraction as a bioassay for [Ca²⁺]_c. We assume that the [Ca²⁺]-contraction relationship does not change during the course of an experiment. While this may be a reasonable default, changes in intracellular conditions (e.g. pH, [ATP], [PO₄⁻] and [Mg²⁺]) which can occur during ischaemia are known to shift the myofilament Ca²⁺ sensitivity (e.g. see Bers, 1991). We have no reason to expect substantial changes in these parameters and the continual whole-cell dialysis should help to buffer any changes in these substrates after each stimulation. Nevertheless, we have not measured these parameters directly.

The equation used to translate ΔL to [Ca²⁺]_CL (eqn (18)) may also be imperfect, although it is consistent with direct steady-state measurements by Bassani et al. (1995) in a study of calibration of intracellular indo-1 in cardiac myocytes. The maximum reversible steady-state ΔL measured here (37 %) is typical of cardiac myocytes from many species, although this does not include the cellular hypercontracture associated with cell death. We used a lower apparent Ca²⁺ affinity (K_d,c = 0.8 μm) and Hill coefficient (2) than reported for isometric force in muscles (0.5–0.6 μm and 5–6, respectively; Yue et al. 1986; Gao et al. 1994). This might be reasonable because as the cell contracts, the steady-state myofilament Ca²⁺ sensitivity decreases (Kentish, ter Keurs, Ricciardi, Bucx & Noble, 1986). We also chose to include a basal [Ca²⁺]_CL (70 nM) at maximal resting cell length in eqn (18). This prevents predicted [Ca²⁺] from approaching zero when the cell relaxes completely (i.e. as ΔL approaches 0). The relationship in eqn (18) predicts that at 112 nM [Ca²⁺]_CL the cell will contract by 0.1 % of resting cell length. Figure 8 shows reasonable, if imperfect, agreement between two methods for estimating [Ca²⁺]_c when mitochondrial Ca²⁺ uptake was blocked ([Ca²⁺]_CL from ΔL and a corrected [Ca²⁺]_c after compensating for unchanging mitochondrial indo-1 fluorescence, eqn (8)). Despite some limitations, we feel that [Ca²⁺]_CL calculated from ΔL is a good approximation and a valuable approach to assess [Ca²⁺]_c in these experiments.

Time course of whole-cell dialysis in ventricular myocytes

We had hoped initially to be able to remove all cytosolic indo-1 by dialysis with the patch pipette, such that the remaining indo-1 was entirely mitochondrial. Despite extensive efforts, including large pipettes and small myocytes, this strategy appeared experimentally impractical in ventricular myocytes (since > 40 min of dialysis was required). This is why we used complementary Mn²⁺ quench experiments to effectively remove any cytosolic indo-1 signal. Although the time constant of Mn²⁺ quench is similar to that of whole-cell dialysis (10 min, Fig. 8), the cells are intact during the quench process. Thus, [Ca²⁺]_m can be measured immediately after formation of whole-cell patch clamp configuration.

The time course of whole-cell dialysis has been studied in detail in small spherical chromaffin cells, where a molecule like fura-2 can be completely washed out by a patch pipette within 3 min (Pusch & Neher, 1988; Zhou & Neher, 1993b). The larger and more complex geometry of ventricular myocytes may account for the difference. Our measurements of myocyte dialysis may allow a more general estimation of the time constant (τ) of wash-out (Pusch & Neher, 1988):

(20)

where M is molecular mass (in daltons) of the dialysed substance, R_acc is in megaohms, C_m is in picofarads (and is proportional to myocyte volume; Satoh et al. 1996), and 300 (in s) is the τ value for standard conditions (e.g. M = 840 for indo-1, R_acc = 3 MΩ and C_m = 80 pF, then τ = 300 s). In this context the 10–20 times slower dialysis of ventricular myocytes vs. chromaffin cells is explained by the 10–20 times larger cell size and different cell shape (cylindrical vs. spherical).

Mitochondrial Ca²⁺ uptake in intact ventricular myocytes

Figures 6 and 8 demonstrate that in the absence of Mn²⁺, mitochondria can take up Ca²⁺ in intact cells. However, Ca²⁺ uptake by mitochondria is apparent only after progressive Ca²⁺ load, with a [Ca²⁺]_c threshold of ∼300–500 nM, and is sensitive to the mitochondrial Ca²⁺ uniport blocker, Ru360. Mitochondrial Ca²⁺ uptake becomes greater when basal [Ca²⁺]_c is further elevated, and is characterized by a distinctly different time course to that of the cytosolic Ca²⁺ transient. For [Ca²⁺]_c below this threshold, the normalized contraction signal (ΔL) is identical to the indo-1 ratio signal (F₄₀₀/F₅₀₀). This suggests that there is no contribution from mitochondria in the normalized F₄₀₀/F₅₀₀, unless [Ca²⁺]_m is kinetically indistinguishable from [Ca²⁺]_c (as reported by Chacon et al. 1996). The latter possibility seems unlikely for several reasons. (1) We see a threshold for this kinetically distinct and clearly mitochondrial component at ∼300–500 nM [Ca²⁺]_c. (2) A very similar threshold and much slower [Ca²⁺]_m are seen with Mn²⁺ quench of cytosolic indo-1. (3) Only the slow component is sensitive to the mitochondrial Ca²⁺ uniport blocker, Ru360. (4) As dialysis proceeds, the rapid Ca²⁺-sensitive indo-1 signal, ΔF_d (see Fig. 1C), associated with contraction becomes smaller (i.e. the fast Ca²⁺ signal washes out as cytosolic indo-1 washes out; not shown). In addition, the rates of Ca²⁺ transport into and out of mitochondria required to match the cytosolic Ca²⁺ transient are inconsistent with other data (see below).

These results and conclusions are consistent with the seminal report by Miyata et al. (1991) in intact myocytes with cytosolic indo-1 quenched by Mn²⁺. They also found that mitochondria take up Ca²⁺, but they only saw slow progressive changes in [Ca²⁺]_m with high stimulation frequencies. We have also used Mn²⁺ to quench cytosolic indo-1 (in combination with dialysis) to obtain a relatively pure [Ca²⁺]_m signal (Figs 9-15). Again, at low basal [Ca²⁺]_c (< 300 nM) no Δ[Ca²⁺]_m was seen with depolarization-induced Ca²⁺ influx into the cell. When resting [Ca²⁺]_c reaches 300–500 nM, mitochondrial Ca²⁺ uptake starts. The Ca²⁺ uptake becomes stronger when basal [Ca²⁺]_c is higher. These phenomena, including the threshold for [Ca²⁺]_c in the presence of Mn²⁺, are essentially the same as in the absence of Mn²⁺, indicating that Mn²⁺ did not substantially alter the basic Ca²⁺ dependence of mitochondrial Ca²⁺ uptake.

The lack of effect of Mn²⁺ on mitochondria under our experimental conditions may be because free cytosolic [Mn²⁺] is probably very low (just enough to quench all cytosolic indo-1). In isolated mitochondria, both Ca²⁺ influx and efflux were found to be sensitive to intramitochondrial Mn²⁺. However, the effective intramitochondrial Mn²⁺ concentration was in the millimolar range (Gavin et al. 1990), which is far larger than the estimated nanomolar to micromolar range of Mn²⁺ in the present work. Because mitochondrial indo-1 is less than half-saturated by Mn²⁺ (even at the end of Fig. 15), the mitochondrial [Mn²⁺] must be below the 10 nM K_d for Mn²⁺ binding of indo-1 (Spurgeon et al. 1990). Thus, it is unlikely that either mitochondrial Ca²⁺ influx or Ca²⁺ efflux in this work are greatly altered by the presence of Mn²⁺, although some minor quantitative effect of Mn²⁺ cannot be ruled out.

Delay of [Ca²⁺]_m after [Ca²⁺]_c

Another major new finding of the present work is that even when mitochondrial Ca²⁺ uptake at a single pulse becomes detectable (Figs 10 and 12) the peak of [Ca²⁺]_m is delayed by 1 s with respect to the slow rising [Ca²⁺]_c with our protocol (Fig. 12B). In this case further increase in [Ca²⁺]_c is terminated by repolarization. After the peak, [Ca²⁺]_m decays very slowly, consistent with the 40 s time constant of mitochondrial Ca²⁺ extrusion inferred by Bassani et al. (1993). In some cases, when cellular Ca²⁺ load is high, [Ca²⁺]_m continues to rise even after the cell contraction and [Ca²⁺]_CL has completely declined (e.g. Fig. 6C).

Our results contrast with a recent study by Chacon et al. (1996) using confocal microscopy to spatially distinguish mitochondrial from cytosolic Ca²⁺ signals in rabbit ventricular myocytes. They reported twitch transients in [Ca²⁺]_m which had the same kinetics as [Ca²⁺]_c. They did not use either Mn²⁺ quench or dialysis to minimize cytosolic signals, or any mitochondrial uniport blocker to confirm the source of the signals. It is possible that the rapid [Ca²⁺]_m transients they reported are contaminated with cytosolic fluorescence signals that are not adequately excluded by confocal microscopy. Although they did have high mitochondrial indicator loading compared with cytosol, even a small rapid cytosolic fluorescence transient might complicate the interpretation. Indeed, in our experiments even though most of the indo-1 was trapped in mitochondria (especially after partial dialysis), the small amount of cytosolic indo-1 can give a signal that could easily be misinterpreted as coming from mitochondria. These possibilities or alternatives were not addressed in the study of Chacon et al. (1996).

Comparison with other preparations

Our results and apparent threshold for mitochondrial Ca²⁺ uptake are in agreement with results with digitonin-permeabilized ventricular myocytes (Fry, Powell, Twist & Ward, 1984; Hove-Madsen & Bers, 1993). Those studies showed no evidence of mitochondrial Ca²⁺ uptake below 1 μm[Ca²⁺]_c. Indeed, measurements with isolated mitochondria (at physiological concentrations of Na⁺ and Mg²⁺) have also typically shown very low Ca²⁺ uptake below 0.5 μm medium [Ca²⁺], but with a sharp increase in Ca²⁺ uptake (and [Ca²⁺]_m) starting between 0.5 and 1.5 μm Ca²⁺ (Crompton, 1990; McCormack & Denton, 1993; Leisey, Grotyohann, Scott & Scaduto, 1993).

The finding of a [Ca²⁺]_c threshold of 300–500 nM for mitochondrial Ca²⁺ uptake in ventricular myocytes is also consistent with previous reports in sympathetic neurons (Thayer & Miller, 1990; Friel & Tsien, 1994), chromaffin cells (Herrington et al. 1996), and smooth muscle cells (McGeown, Drummond, McCarron & Fay, 1996). This general agreement may allow data from the present work in cardiac myocytes to be useful for other cell types. At the single cell level, the large cell size (100–200 pF) and the large fraction of cell volume occupied by mitochondria (30 %) gives large mitochondrial fluorescence signals. However, a disadvantage of ventricular myocytes is the long time required for cytosolic dialysis by patch pipette.

The kinetics of [Ca²⁺]_c changes in cardiac myocytes are normally fast compared with other cell types (although we have intentionally slowed them in the present studies). In other cell types, where the [Ca²⁺]_c changes are slower, the participation of mitochondria may differ. Using the Ca²⁺-sensitive photoprotein aequorin, Rizzuto and co-workers found [Ca²⁺]_m decline to be faster than that of [Ca²⁺]_c in both non-excitable cells and excitable cells (Rizzuto et al. 1992; Rutter, Theler, Murgia, Wollheim, Pozzan & Rizzuto, 1993). Using electron probe microanalysis (EPMA) of rapidly frozen ventricular myocytes, Isenberg, Han, Schiefer & Wendt-Gallitelli (1993) reported only a 20 ms delay in the rise of mitochondrial Ca²⁺ content compared with that of [Ca²⁺]_c. While EPMA can give a purely mitochondrial signal (in contrast to confocal microscopy used by Chacon et al. 1996), the space between inner and outer membranes is still sampled. They also used a very aggressive Ca²⁺-loading protocol, whereas Moravec & Bond (1992) found no mitochondrial Ca²⁺ changes using EPMA, even in the steady state with β-adrenergic agonists (seeing increases only upon removal of extracellular Na⁺). The results of Isenberg et al. (1993) also suggest that the amount of Ca²⁺ going into the mitochondria within 40 ms of depolarization is ten times higher than the total amount of cellular Ca²⁺ cycling during E-C coupling (Bers, 1991; Bassani et al. 1994a). Thus, there are some kinetic and quantitative issues which remain to be resolved in cardiac myocytes.

In both astrocytes and chromaffin cells, large cytosolic Ca²⁺ transients occur and precede mitochondrial Ca²⁺ transients by up to 10 s (Jou et al. 1996; Babcock, Herrington & Hille, 1996). Furthermore, Herrington et al. (1996) demonstrated that mitochondrial Ca²⁺ uptake in chromaffin cells plays a dominant role in the time course of [Ca²⁺]_c decline. Interestingly, [Ca²⁺]_c decline takes 20–40 s in chromaffin cells, which is similar to that in ventricular myocytes when the SR Ca²⁺-ATPase and Na⁺-Ca²⁺ exchange are disabled and mitochondria take up ∼50 % of the cytosolic Ca²⁺ (Bassani et al. 1992, 1994a). Thus, the more dominant role of mitochondria in the chromaffin cell may be primarily due to weaker competition by the SR/endoplasmic reticulum Ca²⁺-ATPase or Na⁺-Ca²⁺ exchange, which are both very strong in ventricular myocytes. These studies also showed slow decline of [Ca²⁺]_m in chromaffin cells and astrocytes (t_½ > 20 s; Jou et al. 1996; Babcock et al. 1996). This is consistent with our observations here and elsewhere for cardiac mitochondria (Bassani et al. 1993).

Quantification of Ca²⁺ influx into mitochondria

We have measured a maximum mitochondrial Ca²⁺ uptake rate after progressive Ca²⁺ loading at ∼10 μm s⁻¹ (Fig. 13). Since a large fraction of the intramitochondrial Ca²⁺ buffer is indo-1 (e.g. ∼0.6 mm in this cell), our estimate depends less on precise knowledge of the endogenous buffering. This maximal uptake rate at 1–2 μm[Ca²⁺]_c would correspond to transport out of the cytosol of ∼5 μmol (l cytosol)⁻¹ s⁻¹ (since 2/3 of the cell volume is cytosolic). This is 5–10 times larger than the rate in intact myocytes derived from the data of Bassani et al. (1994a) during a single cytosolic Ca²⁺ transient reaching a peak of 1 μm. In digitonin-permeabilized cells and isolated mitochondria, maximal Ca²⁺ uptake rates of 2–5 μm s⁻¹ at 1 μm[Ca²⁺]_c can be inferred from the data of Fry et al. (1984) and Crompton (1990). The slightly higher maximal Ca²⁺ uptake rate here (vs.Bassani et al. 1994b) may be a consequence of the preceding progressive cellular Ca²⁺ loading and non-linear dependence on [Ca²⁺]_c (note that diastolic [Ca²⁺]_c in Fig. 13 is nearly 1 μm). The 13 μm Ca²⁺ entering the mitochondria in 1.2 s can raise [Ca²⁺]_m by 200 nM (as in Fig. 13).

During a normal cardiac twitch [Ca²⁺]_c declines rapidly due to transport by the SR Ca²⁺-ATPase and Na⁺-Ca²⁺ exchange, such that much less total mitochondrial Ca²⁺ uptake can occur. During a twitch, if only 0.7 μm Ca²⁺ enters the mitochondria (based on Bassani et al. 1994a), it would raise [Ca²⁺]_m from a resting value of 100 nM by only 7 nM (or 21 nM without mitochondrial indo-1 buffering). This small change might be difficult to detect. This may be why Miyata et al. (1990) did not detect mitochondrial transients at single twitches, but did see gradual cumulative changes after many contractions upon changes in frequency. Indeed, these slow cumulative changes in [Ca²⁺]_m are consistent with net transport models (Crompton, 1990) and may have regulatory effects on mitochondrial energy production due to the Ca²⁺-dependent stimulation of several key mitochondrial enzymes involved in NADH and ATP generation (McCormack & Denton, 1993; Hansford, 1994). Indeed, Brandes & Bers (1997) recently demonstrated [Ca²⁺]_c-dependent stimulation of mitochondrial NADH production upon changes in work in intact ventricular muscle which has temporal delays consistent with slow mitochondrial Ca²⁺ uptake and extrusion.

The high concentration of indo-1 in the mitochondria here assures us that almost all Ca²⁺ influx into the mitochondria produces a fluorescence signal. Indeed, the resolution of [Ca²⁺]_m is technically the same as ΔF_d, indicating that any detectable Ca²⁺ influx reported by ΔF_d will also be seen as Δ[Ca²⁺]_m (although the absolute change in free [Ca²⁺]_m for a given Ca²⁺ influx is smaller with more mitochondrial indo-1). While maximizing the detectability of mitochondrial Ca²⁺ influx, this buffering effect of indo-1 may diminish the absolute rate of Ca²⁺ extrusion from the mitochondria. This may limit quantitative conclusions about Ca²⁺ efflux kinetics.

Thus, we saw no mitochondrial Ca²⁺ influx at a resting [Ca²⁺]_c of 100–200 nM, even with strong depolarizations bringing large amounts of Ca²⁺ in via Na⁺-Ca²⁺ exchange. It is conceivable that mitochondria could respond differently to Ca²⁺ influx via Ca²⁺ channels and SR Ca²⁺ release during physiological twitches. However, Miyata et al. (1991) could not detect any beat-to-beat changes in [Ca²⁺]_m during field stimulation, even with high frequency stimulation. This finding with Mn²⁺ quench of cytosolic indo-1 was also very recently confirmed using another approach to eliminate cytosolic indo-1 (Griffiths, Stern & Silverman, 1997).

The high rates of mitochondrial Ca²⁺ uptake measured here (e.g. > 10 μm s⁻¹) at high levels of cellular Ca²⁺ load and at a resting [Ca²⁺]_c of ∼1 μm are probably not physiological, but may occur under pathological conditions, such as Ca²⁺ overload. This may serve as protection for the cell from sustained high [Ca²⁺]_c and irreversible contracture. However, prolonged Ca²⁺ overload may result in dramatic mitochondrial Ca²⁺ uptake, which may lead to both permeability transition and cell death.

Intracellular Mn²⁺ in mitochondrial studies

Adding Mn²⁺ to the bath to quench selectively cytosolic indo-1 has been a useful strategy in studying [Ca²⁺]_m in cells (Miyata et al. 1991; Blatter, 1995; present work). Mn²⁺ probably enters the cell via voltage-gated Ca²⁺ channels (Masumiya, Tateyama & Ochi, 1997) or some other pathway. There may be two main reasons why this quench is relatively selective for cytosolic indo-1. First, the same transport mechanism may not be present or as active in the inner mitochondrial membrane. Second, the low free [Mn²⁺] in the cytosol may limit Mn²⁺ entry into mitochondria. In our case dialysis with the patch pipette may also be important in preventing progressive increases in cytosolic [Mn²⁺]. Thus, cytosolic Mn²⁺ may be in a quasi-steady state in our experiments. This may explain why studies without patch clamp dialysis have required lower extracellular [Mn²⁺] to accomplish Mn²⁺ quench of cytosolic indo-1. Since we can only just detect very small displacements of Mn²⁺ from cytosolic indo-1 at very high [Ca²⁺]_c (Fig. 11B, inset) we think that we are at an optimal cytosolic free [Mn²⁺]. That is, [Mn²⁺] is just high enough to quench cytosolic indo-1, but not high enough to produce major alterations in other cellular functions (perhaps 0.1 μm Mn²⁺). Notably, maximum contraction is not altered and a given depolarization protocol produces similar contractions in the presence and absence of Mn²⁺ quench. This allows us to measure pure [Ca²⁺]_m signals after Mn²⁺ quench of cytosolic indo-1.

The present results show that Mn²⁺ permeability of the mitochondrial membrane is dramatically increased at the same degree of cellular Ca²⁺ load required to reach the threshold for mitochondrial Ca²⁺ uptake (Fig. 15). It seems plausible that Ca²⁺ and Mn²⁺ use the same pathway (the uniporter) to enter mitochondria. However, the mitochondrial uptake rate for Mn²⁺ is much smaller than that for Ca²⁺ (∼10:1). No doubt, part of this difference is because free [Mn²⁺] is probably lower than [Ca²⁺]_c. In addition, during [Ca²⁺]_m decline, while Ca²⁺ is extruded from mitochondria, Mn²⁺ continues to be taken up. On the other hand, while Ca²⁺ uptake may continue after the rising phase of [Ca²⁺]_c (as evident in Fig. 6), in most cases [Ca²⁺]_m started to decline shortly after repolarization (Fig. 10), such that Ca²⁺ extrusion must be faster than Ca²⁺ influx. If Mn²⁺ is only poorly extruded from mitochondria (Gunter et al. 1994), this could explain the sustained Mn²⁺ uptake even during [Ca²⁺]_m decline.

Conclusions

From multiple types of experiments and analyses we conclude that beat-to-beat fluctuations in mitochondrial Ca²⁺ are likely to be extremely small, and are essentially undetectable, even with the sensitive methods used here (where most of that Ca²⁺ would bind to a sensitive fluorescent indicator). On the other hand, cumulative changes in [Ca²⁺]_m occur with recurrent stimulation. We suggest that such slow cumulative changes in [Ca²⁺]_m may have physiological importance in the regulation of energy metabolism. When the cell and/or mitochondria reach a threshold Ca²⁺ level (e.g. [Ca²⁺]_c, 400–500 nM) mitochondria begin to take up Ca²⁺ much more rapidly and changes in [Ca²⁺]_m can be readily detected even during individual large and slow contractions (and [Ca²⁺]_c transients). This rapid Ca²⁺ uptake mode at very high [Ca²⁺]_c is sensitive to the mitochondrial Ca²⁺ uniport blocker Ru360 and can also transport Mn²⁺. This rapid uptake is probably not functioning during physiological conditions and normal beat-to-beat regulation in heart.