¹Ca²⁺ Free Tyrodes solution contains (in mM): 140 NaCl, 4 KCl, 10 Hepes, 1 MgCl_2.6H₂O, 10 Glucose. Adjusted to pH 7.4 with 5 M NaOH.

²To make 1 mM di-hydroRhod-2 solution: Dissolve 1 μg vial of Rhod-2 with 22.25 μL DMSO. Add 22.25 μL of 20% Pluronic in DMSO. Once dissolved, add 10 μL of Na⁺ borohydride dissolved in 20 μL methanol. Wait until solution becomes colorless, add 5 μL to 1 mL of cell suspension for 5 μM working concentration.

³100 nM mitotracker deep red: Dissolve 50 μg vial of mitotracker with 90 μL DMSO (brings stock concentration to 1 mM). Dilute further to 0.1 mM using DMSO then load 1 mL cells with 1 μL for 100 nM working concentration.

Healthy, Wistar rats (male, 300–350 g) were anesthetized using 2% isoflurane in 100% O₂ as a carrier gas. Once unconscious, the rat was euthanized by decapitation, hearts excised at the aorta and placed in ice cold Ca²⁺ free Tyrode’s buffer (refer to section “For Cell Isolation”). Approval for this research was provided by the University of Auckland Animal Ethics Committee (AEC: 001929), in accordance with the Code of Ethical Conduct of The University of Auckland, and the New Zealand Animal Welfare Act 1999.

Within 30 s after dissection, hearts were cannulated by the aorta, secured with suture, and Langendorff-perfused with oxygenated Ca²⁺ free Tyrode’s buffer for 5 min at 37°C using a gravity fed system. During this time blood should be cleared from the coronary circulation and effluent should run clear demonstrating adequate perfusion of the coronary circulation. Perfusion was then switched to 0.2 mM Ca²⁺ Tyrode’s containing enzymes: 275 U/mL Type 2 Collagenase (Worthington Biochemical Corp, United States) and 1.8 U/mL Protease (Sigma Aldrich, United States). After 9–12 min of enzymatic digestion, ventricles were cut off and immersed in 0.15 mM Ca²⁺ Tyrode’s containing 0.1% BSA (Sigma Aldrich, United States) and 0.05% Trypsin inhibitor (Worthington Biochemical, United States). The ventricles were minced to yield isolated, quiescent myocytes and extracellular Ca²⁺ was gradually increased to 1.5 mM.

As previously mentioned, Rhod-2 is a positively charged fluorescent Ca²⁺ indicator which results in electrical potential-driven uptake into the mitochondria (Hajnóczky et al., 1995). The acetoxymethyl (AM) form of Ca²⁺ indicators can readily cross the sarcolemma and enter the cytosol, where it becomes de-esterified. However, we used a method that exploited the activity of the mitochondrial reactive oxidative species, in order to enhance mitochondrial Ca²⁺ signal specificity. This was done by reducing Rhod-2 to di-hydroRhod-2 (dhRhod-2) prior to introducing it into the cardiomyocytes (Bowser et al., 1998). The reduction of Rhod-2 to dhRhod-2 improves the selectivity for mitochondrial loading as the reduced dye does not display Ca²⁺ dependent fluorescence (Hajnóczky et al., 1995). DhRhod-2 then re-oxidizes to Rhod-2 in the mitochondria, resulting in mitochondrial-specific Ca²⁺ signals (refer to Supplementary Figure 1). To make a 1 mM stock of Rhod-2, one 50 μg vial of Rhod-2 indicator (Invitrogen, Thermo Fisher Scientific, Life Technologies NZ, catalog no. R-1244) was dissolved in 22.25 μL dimethyl sulphoxide anhydrous (DMSO) and 22.25 μL 20% Pluronic in DMSO (Invitrogen, Thermo Fisher Scientific, Life Technologies NZ, cat no. D12345 and P6867). The smallest possible amount of Na⁺ borohydride was dissolved in 20 μL methanol, then 10 μL was added as the reducing agent to the Rhod-2 vial. The indicator then transitioned from a pink color to a clear, colorless solution after 5–10 min. Cardiomyocytes were then loaded with 5 μM dhRhod-2 solution for 1 h, carried out at 37°C to further increase compartmentalization of the indicator into the mitochondria. To allow for de-esterification, cardiomyocytes were washed in 1.5 mM Ca²⁺ Tyrode’s for at least 30 min prior to imaging. Loading of Rhod-2 for cytosolic measurements was completed using the steps above but excluding the addition of Na⁺ borohydride in methanol. A group of myocytes were also loaded with a ratiometric cytosolic Ca²⁺ indicator Fura-2AM, to compare its kinetics to Rhod-2 cytosolic transients. A single 50 μg vial of Fura-2AM (Invitrogen, Thermo Fisher Scientific, Life Technologies NZ, catalog no. F1221) was diluted to a 1 mM stock solution with DMSO and 20% Pluronic. For a working concentration of 10 μM, 10 μL of 1 mM Fura-2AM solution was added to 1 mL of cells. Cells were subsequently loaded for 20 min at room temperature then washed with 1.5 mM Ca²⁺ Tyrode’s for 10 min to allow de-esterification.

In order to validate Rhod-2 localization in the mitochondria, cardiomyocytes were simultaneously loaded with 100 nM mitotracker deep red (refer to section “For Di-HydroRhod-2 Loading” subscript 3) and 5 μM Rhod-2 or dhRhod-2 (Thermo Fisher Scientific, Life Technologies NZ). A perspex cell bath was fixed to the stage of an inverted laser scanning confocal microscope (LSM 710, Zeiss, Oberkochen, Germany). Using a Zeiss 63x oil-immersion objective lens (NA 1.40), images with a 0.1 μm pixel resolution were captured using two lasers sequentially: 561 nm (Rhod-2) and 633 nm (Mitotracker) at 2% laser power. Images obtained for the validation experiments were a single-framed snapshot of the whole myocyte with 2 channels (dhRhod-2/Rhod-2 and Mitotracker), which were subsequently merged for analysis (see section “Data Analysis”). Furthermore, cardiomyocytes loaded with dhRhod-2 (and Rhod-2) were electrically stimulated (D100, Digitimer, Welwyn, United Kingdom) with 5 ms pulses at 1 Hz (room temperature) before (baseline) addition of 20 mM caffeine (Sigma Aldrich, cat. no C0750). The response to caffeine was also measured by delivering a 20 mM bolus to the cardiomyocyte in the absence of electrical stimulation. This technique was used to detect potential loading of dhRhod-2 into the SR. The response to caffeine was then compared between myocytes loaded with dhRhod-2 and those loaded with Rhod-2. Using a similar method, the response to 1 μM isoproterenol (Sigma Aldrich, cat. no I6379) was recorded by collecting the emitted fluorescence from the whole cell over 2–3 min (1 frame/s). Cardiomyocytes loaded with dhRhod-2 were electrically stimulated with 5 ms pulses at 1 Hz (room temperature) before (baseline) and during addition of isoproterenol. This method was employed to determine whether dhRhod-2 signals were enhanced during beta-adrenergic stimulation. The response to 1 μM isoproterenol was also tested in cells that were pre-treated with 50 μM MCU inhibitor Ru265 for 30–60 min. Further evidence against cytosolic contamination was obtained by imaging permeabilized cardiomyocytes loaded with dhRhod-2. For cell permeabilization, cardiomyocytes were loaded with dhRhod-2 (as stated in section “Loading of Rhod-2 and Di-HydroRhod-2”), then washed in intracellular buffer containing (in mM); 2 CaCl₂ (for 100 nM free [Ca²⁺]_i), 5 NaCl, 140 KCl, 1 MgCl.6H₂O, 5 Glucose, 10 Hepes, 15 butanedione monoxime and 5 EGTA. Myocytes were imaged as above, at baseline and after bolus application of 0.05% saponin, which was used for cell permeabilization.

Fluorometric measurements were made from cardiomyocytes loaded with di-hydroRhod-2 to determine beat-to-beat changes in mitochondrial Ca²⁺ fluxes. Loaded cardiomyocytes were transferred into a perspex cell bath, which was fixed to the stage of an inverted fluorescence microscope. Cells were then externally field stimulated between 0.1 and 1 Hz (room temperature) and continuously super-fused with 1.5 mM Ca²⁺ Tyrodes containing 1 μM isoproterenol and 150 μM spermine (Cayman Chemical United States, cat no. 136587-13-8). Spermine is an allosteric MCU agonist, which increases MCU affinity to Ca²⁺, enhancing its uptake into the mitochondria (Nicchitta and Williamson, 1984). Therefore, its addition with isoproterenol is useful in enhancing the total dhRhod-2 signal. Myocytes were imaged using a 20x objective lens, 0.75 NA, and illuminated with a 542 ± 10 nm excitation wavelength using an Optoscan monochromator and spectrofluorometric, PMT based system (Cairn, Faversham, United Kingdom). Recordings were acquired at a rate of 400 Hz and sampled in 10 ms intervals using Acquisition Engine software (Cairn, Faversham, United Kingdom). Emitted fluorescence was collected at 581 nm (± 10 nm) from the whole cell (approx. 100–150 × 10–30 μm collection window). The total emitted fluorescence was used as a measure of mitochondrial Ca²⁺. The mitochondrial Ca²⁺ fluxes were then compared with cytosolic Ca²⁺ transients recorded in Rhod-2-loaded myocytes. Cytosolic Ca²⁺ transients were also recorded in myocytes loaded with Fura-2. Using the same spectrofluorometric system, cardiomyocytes loaded with Fura-2 were illuminated with alternating 340 and 380 nm excitation wavelengths every 5 ms. Emitted 510 ± 15 nm fluorescence was acquired at 400 Hz using Acquisition Engine software (Cairn, Faversham, United Kingdom) from the whole cell.

Confocal data was acquired using Zen Blue software (Zeiss, Germany) and analyzed using FIJI (ImageJ analysis). The relationship in fluorescence intensity was measured by calculating the correlation index from a specific region of each cardiomyocyte co-labeled with Rhod-2/mitotracker and dhRhod-2/mitotracker. The correlation index, also known as Pearson’s coefficient, measures the overlap of pixels from two-channel images. The ImageJ Analysis Software Plugin “Just Another Co-localization Program” or “JACoP” (Bolte and Cordelieres, 2006), was used to determine the correlation index between Rhod-2/mitotracker and dhRhod-2/mitotracker. The mean correlation index between Rhod-2/mitotracker and dhRhod-2/mitotracker was then analyzed by unpaired, two-tailed t-tests to compare the overlap between the two indicators. The correlation index indicated the degree of mitochondrial localization, with a value of 1 representing mitochondrial compartmentalization and a value of 0 representing no compartmentalization (Bolte and Cordelieres, 2006). In addition, the intensity plots from myocytes exposed to pharmacological interventions were statistically analyzed by unpaired, two-tailed t-tests to compare the amplitudes at the peaks of dhRhod-2 fluorescence at baseline vs. during maximum drug response.

In separate experiments, fluorometric measurements of cytosolic and mitochondrial Rhod-2 fluorescence were carried out using an Optoscan monochromator (Cairn Research Ltd., Faversham, United Kingdom). Comparison of the following parameters were made from both the mitochondrial and cytosolic Ca²⁺ transients: ΔF/F0, where ΔF is the transient amplitude and F0 the baseline fluorescence, maximum rate of fluorescence rise, time to peak fluorescence following stimulation, and time constant of fluorescence decay. Ca²⁺ transients were acquired once steady state was obtained for all stimulation frequencies measured, and the parameters from an average of three sequential transients were analyzed. Mean data from fluorometric measurements of Rhod-2 cytosolic Ca²⁺ transients and dhRhod-2 mitochondrial Ca²⁺ transients were analyzed using two-way ANOVA for multiple comparisons between groups and stimulation frequencies. Statistical significance was indicated by a P-value of < 0.05 for all data sets. All mean data (including confocal) was statistically analyzed using Prism 9 Analysis software.

Fluorometric measurements were optimized and made following the protocols stated in section “Fluorometric Measurements in Live Myocytes” in order to: (i) determine the characteristics of the mitochondrial Ca²⁺ transient relative to the cytosolic Ca²⁺ transient and (ii) to record beat-to-beat mitochondrial Ca²⁺ fluxes at different stimulation frequencies. Figure 6A shows two single transients recorded from representative myocytes loaded with Rhod-2 (black) and dhRhod-2 (red). When comparing a Rhod-2 cytosolic Ca²⁺ transient superimposed with a dhRhod-2 mitochondrial Ca²⁺ transient, visually there are clear differences in both the transient amplitude and kinetics. Figure 6B shows the Rhod-2 transient resembles a cytosolic Ca²⁺ transient with similar characteristics to a transient recorded from a representative myocyte loaded with the commonly used cytosolic Ca²⁺ indicator, Fura-2 (green).

Loading myocytes with Rhod-2 resulted in widespread, non-localized fluorescence (Figures 2A,B). However, there was apparently some Rhod-2 mitochondrial compartmentalization, as several peaks visible in the intensity plot profile (Figure 2B) coincided with smaller amplitude peaks observed with the mitotracker label. This differed to the punctate mitochondrial pattern displayed in di-hydroRhod-2 (dhRhod-2) loaded myocytes (Figure 3A). DhRhod-2 substantially overlapped with the mitotracker indicator as shown in the intensity plot profile (Figure 3B). As mentioned in section “Data Analysis,” the correlation index between Rhod-2/mitotracker and dhRhod-2/mitotracker indicated the degree of mitochondrial localization. DhRhod-2/mitotracker had a correlation index closer to 1 (Figure 3C), which confirmed mitochondrial specific loading of dhRhod-2. This was opposite to Rhod-2/mitotracker, which had a correlation index closer to 0, indicating only partial mitochondrial localization but mostly spread throughout the cytosol. As previously stated, Rhod-2 is a low affinity Ca²⁺ indicator, and increased mitochondrial specificity is achieved when reducing Rhod-2 to dhRhod-2. The reduced dhRhod-2 form of the indicator is reactive to mitochondrial reactive oxidative species (Supplementary Figure 1), which allows it to be re-oxidized to Rhod-2 specifically in the mitochondria (Bowser et al., 1998). This means that any un-oxidized dhRhod-2 in the cytosol, cannot respond to changes in [Ca²⁺], making it a useful indicator for specifically measuring mitochondrial Ca²⁺. Trollinger et al. (1997) was the first group to show “cold loading” of Rhod-2 resulted in mitochondrial localization in adult rabbit cardiomyocytes. They also reported warm loading favors cytosolic dye retention, which can ultimately contaminate mitochondrial Ca²⁺ signals. This is contrary to the present study, where myocytes were loaded with dhRhod-2 at 37°C and imaged within 2 h. Our data showed strong mitochondrial localization whereby cytosolic contamination was negligible. This was also supported by our findings in permeabilized myocytes loaded with dhRhod-2 (Supplementary Figure 2), which still retain mitochondrial distribution in the absence of an intact cell membrane and subsequent removal of cytosolic contents.

Using confocal microscopy, we were able to confirm localized dhRhod-2 loading (Figure 3). We were also able to confirm gradual enhancement of mitochondrial Ca²⁺ uptake with β-adrenergic stimulation (Figure 5). Next, we: (i) determined the characteristics of the mitochondrial Ca²⁺ transient relative to the cytosolic Ca²⁺ transient, and (ii) recorded beat-to-beat mitochondrial Ca²⁺ fluxes at different stimulation frequencies. Fluorometric measurements were made from cardiac myocytes loaded with dhRhod-2 to determine beat-to-beat changes in mitochondrial Ca²⁺ fluxes. This could not be achieved using confocal microscopy as these events occur at a timescale beyond the temporal acquisition rate. Faster changes in signal can be obtained in a smaller region of the myocyte using line scans. However, line scanning only captures a single region of the myocyte, which is insufficient when measuring a small change in mitochondrial Ca²⁺ flux. Imaging of the whole myocyte can be done (as discussed in section “Live Cell Confocal Imaging”), but only with a slower acquisition rate. Any changes that might occur < 300 ms apart cannot be detected due to limited temporal resolution. Therefore, we have developed a technique for fluorometric measurements of [Ca²⁺]_mito transients, which can capture changes that occur on a faster time scale (i.e., between 50 and 300 ms). Using a spectrofluorometric system, we acquired rapid changes in mitochondrial Ca²⁺ on a beat-to-beat basis (see section “Fluorometric Measurements in Live Myocytes”). A window was fitted around the perimeter of a single cell, which allowed for [Ca²⁺]_mito measurements across the whole cell (i.e., a “global” change).

Figure 6A shows a single [Ca²⁺]_mito transient superimposed with a single [Ca²⁺]_cyto transient. The [Ca²⁺]_mito transient had a smaller amplitude relative to the [Ca²⁺]_cyto transient. Interestingly, the [Ca²⁺]_mito also showed a slower time to peak fluorescence. When considering the kinetics of a [Ca²⁺]_cyto transient, the time to peak fluorescence occurs approximately 30–60 ms after the stimulus (Shannon et al., 2000; Dibb et al., 2007). This was evident in Figure 6B, which shows a single Fura-2 (ratiometric) [Ca²⁺]_cyto transient superimposed with a single Rhod-2 (non-ratiometric) [Ca²⁺]_cyto transient. The kinetics between Rhod-2 and Fura-2 transients were almost identical, which suggested that Rhod-2 was a suitable measure of cytosolic [Ca²⁺] for comparison to dhRhod-2 [Ca²⁺]_mito transients. Figure 6A shows MCU uptake begins to rise once the [Ca²⁺]_cyto transient has peaked, then it begins to decline at approximately 70–80% of the [Ca²⁺]_cyto transient decay. This is evidence that MCU uptake begins when intracellular [Ca²⁺] is highest (i.e., at the peak of the [Ca²⁺]_cyto transient) and then begins to decline as cytosolic Ca²⁺ approaches diastolic levels. This supports original findings from Isenberg et al. (1993), who showed peak [Ca²⁺]_mito occurred 40 ms after the start of systole and a subsequent decline in [Ca²⁺]_mito 95 ms after the start of systole. Studies using adenoviral probes targeted to the mitochondria also reported similar findings. For example, Robert et al. (2001) found beat-to-beat oscillations in cultured neonatal rat myocytes, whereby [Ca²⁺]_cyto peaked at 50–100 ms and [Ca²⁺]_mito peaked 100–150 ms after. Furthermore, in cultured rat adult myocytes, Bell et al. (2006) also observed [Ca²⁺]_mito transients at a cytosolic [Ca²⁺] of 0.9 μM, which is equivalent to the peak [Ca²⁺]_cyto transient at systole. These data are in contrast to a study by Maack et al. (2006) showing peak [Ca²⁺]_mito preceded peak [Ca²⁺]_cyto in isolated guinea pig myocytes. However, Maack et al. (2006) made measurements at 37°C as opposed to the present study where experiments were performed at 23°C, but it is not clear how a difference in temperature could account for the difference in transient kinetics. Unlike Maack et al. (2006), we could not record simultaneous Ca²⁺ transient events. However, we were able to compare the kinetics of Rhod-2 [Ca²⁺]_mito to [Ca²⁺]_cyto transients in separate myocytes as we used the same indicator, which allowed direct comparison without variability in the fluorophore binding kinetics. Collectively, our findings provide further evidence that mitochondrial Ca²⁺ fluxes occur during EC coupling, and are therefore able to regulate mitochondrial metabolism on a beat-to-beat basis.