Measuring Local Gradients of Intra-Mitochondrial [Ca] in Cardiac Myocytes During SR Ca Release

Abstract

Rationale

Mitochondrial [Ca²⁺] ([Ca²⁺]_mito) regulates mitochondrial energy production, provides transient Ca²⁺buffering under stress and can be involved in cell death. Mitochondria are near the sarcoplasmic reticulum (SR) in cardiac myocytes and evidence for crosstalk exists. However, quantitative measurements of [Ca²⁺]_mito are limited and spatial [Ca²⁺]_mito gradients have not been directly measured.

Objective

To directly measure local [Ca²⁺]_mito during normal SR Ca release in intact myocytes, and evaluate potential subsarcomeric spatial [Ca²⁺]_mito gradients.

Methods and Results

We used in-situ calibration of the mitochondrially targeted inverse pericam indicator Mitycam and directly measured [Ca²⁺]_mito during SR Ca²⁺ release in intact rabbit ventricular myocytes by confocal microscopy. During steady state pacing Δ[Ca²⁺]_mito amplitude was 29 ± 3 nM, rising rapidly (similar to cytosolic [Ca²⁺]_i) but declining much more slowly. Taking advantage of the structural periodicity of cardiac sarcomeres, we found that [Ca²⁺]_mito near SR Ca²⁺ release sites (Z-lines) vs. mid sarcomere (M-line) reached a higher peak amplitude (37 ± 4 vs. 26 ± 4 nM, respectively P < 0.05) which occurred earlier in time. This difference was attributed to ends of mitochondria being physically closer to SR Ca²⁺ release sites, because the mitochondrial Ca²⁺ uniporter was homogeneously distributed and elevated [Ca²⁺] applied laterally did not produce longitudinal [Ca²⁺]_mito gradients.

Conclusions

We developed methods to measure spatiotemporal [Ca²⁺]_mito gradients quantitatively during excitation-contraction coupling. The amplitude and kinetics of [Ca²⁺]_mito transients differ significantly from those in the cytosol and are higher and faster near the Z- vs. M-line. This approach will help clarify SR-mitochondrial Ca²⁺ signaling.

INTRODUCTION

Sarcoplasmic reticulum (SR) and mitochondria are both essential for cardiac myocyte function. The SR is the site for Ca²⁺ storage and release that drives contraction during each heartbeat. Cytosolic Ca²⁺ influences many other targets, including ion channels and transporters, signaling cascades, gene transcription and mitochondrial ATP production.^1,2 Ca²⁺ pumping by the SR Ca-ATPase (SERCA) and out of the cell via Na⁺/Ca²⁺ exchange allow diastolic relaxation and cardiac refilling between beats. Excitation-contraction coupling (ECC) consumes much ATP, which is generated mainly in mitochondria by oxidative phosphorylation, regulated in part via Ca²⁺-dependent dehydrogenases.^2–4 Hence, mitochondrial Ca²⁺ participates in “excitation-metabolism coupling” (EMC) critical to ATP availability. EMC may be facilitated by structural interactions in cardiomycytes where mitochondria are surrounded by an SR network.^4–6 At least one end of most mitochondria is in close proximity to SR Ca²⁺ release sites, and physical transorganellar tethers connecting SR/ER and mitochondria have been identified.⁷ The low affinity mitochondrial Ca²⁺ uniporter (MCU; K_0.5 ~10–20 µM Ca²⁺)⁸ could also preferentially take up Ca²⁺ at the high local [Ca²⁺]_i expected near the release sites. In other words, local Ca²⁺ signal transmission may efficiently alter cellular energy supply to match demand, and consequently also shape cytosolic Ca²⁺ dynamics.^9–11 The concept of mitochondrial-SR Ca²⁺ microdomains is timely,^9,10,12 however, whether local spatial and temporal gradients of intra-mitochondrial free [Ca²⁺] ([Ca²⁺]_mito) occur near SR Ca release sites and what the absolute kinetics of [Ca²⁺]_mito signals are in adult ventricular myocytes has not been well defined. Moreover, it is controversial whether [Ca²⁺]_mito signals are dominated by rapid changes parallel to the cytosolic Ca²⁺ transients or more slow, integrating responses during ECC and what the functional implications this has in cardiac myocytes.^13–15 Furthermore, under pathological condition, mitochondrial–SR crosstalk also contributes to programmed cardiac myocyte death via opening Ca²⁺-sensitive mitochondrial permeability transition pore (MPTP) which permits unrestricted proton flow out of the mitochondrial matrix that interrupts the respiratory electron transport chain. This allows the release of cytochome c into the cytosol, and then initiates necrotic cell death or apoptosis.^16,17 However, limitations exist in the measurement of mitochondrial [Ca²⁺] in intact myocytes. Moreover, subsarcomeric [Ca²⁺]_mito microdomains have not been assessed functionally in intact adult cardiac myocytes. Here we detect subsarcomeric spatiotemporal gradients of [Ca²⁺]_mito using a genetically encoded and mitochondrially targeted Ca²⁺ sensor, Mitycam.¹⁸

While [Ca²⁺]_mito transients are known to occur during SR Ca release and reuptake in cardiac myocytes,^14,18–20 calibrations and exclusion of contamination by cytosolic signals are challenging for [Ca²⁺]_mito measurements in nanomolar values. Understanding how local and global Ca²⁺ transients influence [Ca²⁺]_mito is controversial¹⁵ and quantitative data is essential to deepen our understanding of these issues. We provide here quantitative estimates of [Ca²⁺]_mito transients in intact ventricular myocytes, and demonstrate detectable subsarcomeric spatial [Ca²⁺]_mito gradients during normal Ca transients in adult ventricular myocytes for the first time. The methods we use to detect the spatial [Ca²⁺]_mito established here could be useful approach to investigate mitochondrial Ca²⁺ handling in many conditions.

METHODS

Additional information is in the Online Supplement at http://circres.ahajournals.org.

Myocyte isolation & viral transfection

All protocols involving animals were performed in accordance with the Guide for the Care and Use of Laboratory Animals and approved by the University of California, Davis Institutional Animal Care and Use Committee. Adult rabbit ventricular cardiomyocytes were isolated from New Zealand White rabbits by standard enzymatic dissociation as described previously.²¹ Freshly isolated cells were plated on laminin-coated glass cover slips for 45 min before the transfection. Adenoviral-deiated Mitycam gene transduction was carried out at a multiplicity of infection (MOI) of 500 virus particles per cell (vp/cell).¹⁸ Infected cells were cultured in serum-free PC-1 medium (Lonza) for 36 hours, with 1 final replacement of fresh medium 1 hour before experiments. Lower MOI or shorter duration in culture often provided poorer signal resolution.

Fluorescence microscopy

Mitycam fluorescence was measured with excitation at 488nm (emission between 530 ± 15 nm) for 2D imaging (Zeiss, LSM5 Pascal) and line scan imaging (Bio-rad Radiance). Mitochondria were localized by 1 µM MitoTracker Red (Invitrogen Ltd) using 543 nm excitation. For some experiments, cells were field-stimulated and Mitycam and Di-8 ANEPPS (488 nm excitation; >600 nm emission) were recorded by line scan imaging. Image-J software was used for image analysis. Cytosolic Ca²⁺ transients were detected in myocytes loaded with Fluo-4 (Invitrogen).

Chemicals and solutions

For permeabilized myocyte experiments a highly Ca²⁺-buffered, Na⁺-free internal solution was used containing (in mM): EGTA 5, HEPES 20, K-aspartate 100, KCl 40, MgCl₂ 1, maleic acid 2, glutamic acid 2, pyruvic acid 5, KH₂PO₄ 0.5, pH 8.0 adjusted with Trizma base. To control [Ca²⁺]_i, 100 mM CaCl₂ solution (Thermo) was added as calculated using the program MaxChelator (http://www.stanford.edu/~cpatton/maxc.html). For intact myocyte experiments, cells were superfused with normal Tyrode’s (NT) solution containing (in mM) 140 NaCl, 4 KCl, 1 MgCl₂, 1.8 CaCl₂, 10 glucose, and 5 HEPES, pH 7.4. To detect intra-mitochondrial distribution of MCU, anti-MCU antibody (Sigma-Aldrich) was used at 1:500 dilution. The secondary antibody carried FITC derivative (Alexa Flour 488; Molecular Probes) and was used at a 1:1000 dilution.

Statistics

Pooled data are represented as the mean ± SEM. Statistical comparisons were made using unpaired or paired Student’s t tests where applicable (P < 0.05 was considered significant).

RESULTS

Mitycam targets to Mitochondria and provides [Ca²⁺]_mito signals

Figure 1A shows confocal images of a typical adult rabbit cardiomyocyte expressing Mitycam. Mitycam fluorescence (488 nm excitation) shows specific mitochondrial expression, overlaping completely with the mitochondrial marker MitoTracker Red (543 nm excitation). Mitycam was expressed in virtually all myocytes, as indicated by the low magnification images in Figure 1B, and again was well matched with MitoTracker Red imaging. Moreover, quantitative pixel by pixel analysis of many cells showed strong correlation between Mitycam (green) and MitoTracker Red signals (Fig 1B, bottom right). The overlap coefficient of Mitycam and MitoTracker was 0.843 ± 0.007 (r =∑(ch1_i)(ch2_i)/(∑(ch1_i)² × ∑(ch2_i)²)^1/2), where 0 implies no colocalization and 1 implies perfect colocalization). Importantly, upon myocyte permeabilization by saponin in an intracellular solution (with physiological [Ca²⁺] and [Na⁺]), there was no significant change of Mitycam fluorescence (Fig 1C). This indicates that virtually all Mitycam is targeted inside mitochondria, with no appreciable cytosolic Mitycam (which would have been lost upon permeabilization). Thus, Mitycam can provide truly mitochondrial-specific signals.

Figure 1. Mitycam in adult rabbit ventricular myocytes.

(A) Mitycam, MitoTracker Red and merged signals exhibit a mitochondrial pattern (scale bar 8 µm), especially apparent in enlarged insets. (B) Lower magnification image of myocyte (scale bar, 100 µm). Scatter 2D plots of pixel intensities in red (MitoTracker) and green (Mityam) channels (right-bottom). Intensity thresholds were automatically determined by excluding dark pixels via algorithm in image analysis software, ZEN (Zeiss). Region 1 and 2 pixels represent signal in channel 1 or 2 only, respectively; region 3 represents colocalized pixels. (C) Mitycam signal is unaltered by saponin-permeabilization (typical images (left), average fluorescence before and after at (right); n=10; scale bar, 8 µm).

Cytoplasmic Ca²⁺ and mitochondrial Ca²⁺ transients were recorded at 0.2 Hz with and without the MCU inhibitor 1 µM Ru360 (Fig 2), a concentration that does not alter SR Ca uptake or release, Ca current, Na/Ca exchange, Ca transients, contraction or myofilament Ca sensitivity.²² Figure 2A shows that neither the amplitude nor kinetics of [Ca²⁺]_i transients were altered significantly by Ru360. In contrast, the [Ca²⁺]_mito signal was virtually abolished by Ru360 (Fig 2B). This confirms that the Mitycam signal is due to [Ca²⁺]_mito changes and is prevented by MCU block. The time to peak of [Ca²⁺]_mito and [Ca²⁺]_cyto was not significantly different, but the decay time constant (τ) of [Ca²⁺]_mito was much slower (5 s) than that for [Ca²⁺]_cyto (Fig 2C).

Figure 2. Kinetics of [Ca²⁺]_mito in adult rabbit ventricular myocytes.

(A) Kinetics of [Ca²⁺]_cyto with and without 1 µM Ru360 during 0.2 Hz stimulation with 1.8 mM Ca (i) and mean Δ[Ca] and time to peak (ii). (B) [Ca²⁺]_mito transient and mean Δ[Ca²⁺]_mito with and without 1 µM Ru360. (C) Time to peak and decline tau of [Ca²⁺]_mito and [Ca²⁺]_i. (D) Influence of pacing frequency (as indicated above traces) on [Ca²⁺]_mito signal with (right) and without 100 nM isoproterenol (left). (E) Amplitude and kinetics of [Ca²⁺]_mito at different frequencies (±ISO). (n = 6, *P<0.05, ***P< 0.001).

As the frequency of stimulation increases the diastolic [Ca²⁺]_mito increases progressively, but the amplitude of [Ca²⁺]_mito transients decreases, which leads to progressive fusion of [Ca²⁺]_mito transients, consistent with the slow kinetics of [Ca²⁺]_mito decline (Fig 2D). For this cell, the signal (1-F/F_o) is saturated (F_min) at 0.624, indicating that Mitycam is far from saturation at 0.5 Hz. Furthermore, to stress the myocyte we measured [Ca²⁺]_mito transients at different pacing frequencies in the absence and presence of the β-adrenergic agonist isoproterenol (ISO). Figure 2D–E shows that with ISO there is an increase in the amplitude of [Ca²⁺]_mito transients during individual beats (0.1Hz: 0.044 ± 0.006 vs. 0.063 ± 0.004; 0.2Hz: 0.031± 0.0028 vs.0.045± 0.003, p < 0.05) and a higher steady state [Ca²⁺]_mito at higher frequency. The time to peak [Ca²⁺]_mito was not altered by ISO, but [Ca²⁺]_mito decay was faster with ISO (Fig 2E). To focus on larger amplitude [Ca²⁺]_mito signals, further studies focused on control conditions at 0.2 Hz.

Calibration of [Ca²⁺]_mito signals during Ca transients

To calibrate [Ca²⁺]_mito signals in myocytes, we first assessed the affinity of Mitycam for Ca²⁺ in cardiac mitochondria in situ. Saponin-permeabilized myocytes expressing Mitycam were pretreated with 5 µM thapsigargin to block SR Ca uptake and release and then equilibrated with internal solution of different [Ca²⁺] containing 5 µM ionomycin (Ca²⁺ ionophore) for 20–30 min. The solution included 5 µM FCCP and 1 µM oligomycin to dissipate mitochondrial membrane potential and was at pH=8 (to mimic mitochondrial pH). The in situ K_d was 197 ±11 nM (Fig 3A).

Figure 3. Calibration and [Ca²⁺]_mito transients in cardiomyocytes.

(A) In situ Mitycam calibration in myocytes (n=8). (B) Average [Ca²⁺]_mito transients, diastolic [Ca²⁺]_mito and amplitude of transient during 0.2 Hz stimulation.

After measuring beat-to-beat [Ca²⁺]_mito transients we assessed F_max and F_min in the same myocyte (maximum and minimum fluorescence in low and high [Ca²⁺], respectively). First cells were saponin-permeabilized and the same type of calibration solutions as above were used for F_max and F_min conditions. [Ca²⁺]_mito was calculated using the relationship F = (F_min-F_max)/ (1+(K_d/[Ca²⁺]))+F_max). Mean diastolic [Ca²⁺]_mito was 146 ± 9 nM and a mean transient amplitude of 29±3 nM (Fig 3B).

[Ca²⁺]_mito gradients along the sarcomere during [Ca²⁺]_i transients

We took advantage of the periodic sarcomeric structure in ventricular myocytes. Mitochondria in cardiac myocytes are oriented in longitudinal rows (between myofibrils; Fig 4A) that are perpendicular to the transverse tubules (T-tubules, where Ca²⁺ is released from the junctional SR)² and the sarcomeric Z-line, which both create a physical boundary to mitochondria. T-tubule membranes are identified by transverse Di-8-ANEPPS striations (Fig 4A) that exist at Z-lines. Thus, the Mitycam signal from the ends of mitochondria nearest Z-lines are close to SR Ca release sites. Using longitudinal line scans (in the direction shown in Fig 4A inset) we pooled Mitycam signals from within 0.5 µm of the center of the T-tubule, and from 6–8 junctions in series, and refer to that as the Z-line signal (near SR Ca²⁺ release sites). The signal from the remainder of the sarcomere (~1 µm centered on the M-line) we call the M-line signal, coming from sites which are furthest from SR Ca release sites. This signal averaging allows us to assess whether there are detectable spatial [Ca²⁺]_mito gradients in confocal line-scan mode. Control experiments (Fig 4B) were done in saponin-permeabilized myocytes (pretreated with 5 µM thapsigargin) exposed to a rapid global [Ca²⁺]_i elevation from 50 nM to 2 µM with solution flow from the lateral side. No detectable [Ca²⁺]_mito gradient was detected.

Figure 4. Spatial [Ca²⁺]_mito signals at Z- and M-line regions.

(A) Localization of T-tubule/Z-line (Di-Anepps, middle) and Mitycam. Merged image (with enlarged region) shows how Z- and M-line Mitycam signals are obtained by separating signals spatially. (B) Direct lateral application of 2 µM Ca²⁺ internal solution to permeabilized myocyte (pretreated with 5 µM thapsigargin and 40 µM cytochalasin D) and line scan image (right).

Intact myocytes with functioning SR during normal Ca²⁺ transients exhibited differences in the kinetics and amplitude of the [Ca²⁺]_mito transient between Z- and M-line (Fig 5). The [Ca²⁺]_mito amplitude was significantly higher at the Z-line vs. the M-line (37 ± 4 nM vs 26 ± 5 nM, P < 0.05), and peaked earlier at the Z-line site (0.24 ± 0.05 vs. 0.57 ± 0.17 s, P <0.05). The time constant of [Ca²⁺]_mito decline was similar between M-line and Z-line regions (τ ~5 s), suggesting that the slow [Ca²⁺]_mito decline is less influenced by location and that [Ca²⁺]_mito gradients dissipate during [Ca²⁺]_mito decline. During rapid caffeine application (10 mM) there was still a detectably higher peak [Ca²⁺]_mito near Z-lines (driven by RyR-mediated Ca²⁺ release from SR) vs. M-line sites (Online Figure I; 62 ± 2 vs 49 ± 2 nM, P < 0.05). The higher amplitude is consistent with a higher fractional SR Ca release with caffeine and less competition by SR Ca uptake.

Figure 5. [Ca²⁺]_mito transients at Z-line and M-line.

Averaged [Ca²⁺]_mito transients (A), diastolic and amplitude (B), time to peak (C) and decay time constant (τ; D). (E) Isochronal dependence of [Ca²⁺]_mito on distance from Z-line at 50 ms before peak in the mean Z-line signal (horizontal line and large symbol indicate mean values for 0–0.5 and 0.5–1 µm as in A). (n = 6, *P < 0.05).

Because the average [Ca²⁺]_mito within 0.5 µm of the Z-line may underestimate the maximal [Ca²⁺]_mito in the region closest to the Z-line, we also examined the spatial profile of [Ca²⁺]_mito as [Ca²⁺]_Mito approaches its peak. Figure 5E shows isochronic [Ca²⁺]_mito signals 50 ms before the peak of the Z-line signal. Within 0.1 µm [Ca²⁺]_mito is significantly higher than the mean Z-line peak value from Fig 5A (horizontal line in Fig 5E). The [Ca²⁺]_mito declines with increasing distance from Z-line regions (Fig 5E). Thus, during SR Ca²⁺ release, [Ca²⁺]_mito may reach as high as 226 ± 13 nM at the regions nearest to Z-line, although the average signal from M-line regions is approximately 152 nM.

It is possible that the higher and faster [Ca²⁺]_mito rise near the Z-line could also be due to a higher localization of MCU channels near Z-line vs. M-line. Indeed, MCU channel density at the inner mitochondrial membrane was estimated to be 10–40 µm⁻², and inhomogeneous distribution could also cause spatial [Ca²⁺]_mito gradients.²³ To test for differential sarcomeric distribution of MCU channels, we used a MCU-specific antibody for immunolocalization. The longitudinal sarcomeric distribution of MCU-related fluorescence was very similar to that of Mito tracker (Fig 6A) and Mitycam (Online Fig II). This suggests relatively uniform MCU distribution over the mitochondrion and is consistent with the lack of [Ca²⁺]_mito gradients seen upon global (vs. local) [Ca]_i elevation (Fig 4B).

Figure 6. Sarcomeric expression of MCU.

(A) MCU immunofluorescence, MitoTracker Red fluorescence, merged image and plot profiles (normalized to Z-line minimum). (B) Scheme of SR Ca²⁺ release and mitochondrial Ca²⁺ uptake.

DISCUSSION

Cardiac mitochondrial Ca²⁺ uptake is important in maintaining cellular ATP, protecting myocytes from transient Ca²⁺ overload and in mediating cell death pathways.^1,2,16,,24 However, the amplitude, kinetics and time-dependent integration of mitochondrial Ca²⁺ uptake during ECC in adult ventricular myocytes are controversial.^13,15 It seem clear now that [Ca²⁺]_mito transients can occur with each cytosolic Ca²⁺ transient and that diastolic [Ca²⁺]_mito increases progressively at increased frequency or cellular Ca²⁺ loading.^14,18,20 However, obtaining calibrated [Ca²⁺]_mito signals has been difficult because of potential contamination by cytosolic fluorescent indicator (for membrane permeable esters), complications of using Mn²⁺ (to quench cytosolic Ca²⁺ indicator) and for mitochondrial targeted aequorin calibrations are highly non-linear and require correction for indicator consumption. Here, we take advantage of work using carefully calibrated [Ca²⁺]_mito measurements in permeabilized myocytes (using fura-2 and rhod-2)²⁰ and the genetically encoded mitochondrially targeted Ca²⁺ sensor Mitycam¹⁸ to assess [Ca²⁺]_mito in intact adult rabbit ventricular myocytes during ECC.

We calibrated Mitycam in situ in cardiomyocyte mitochondria, obtaining a K_d value (~200 nM) similar to that in Hela cells¹⁸ and comparable to that of organic Ca²⁺ indicators used to measure [Ca²⁺]_mito transients in myocytes.¹⁴ We used F_max and F_min from each myocyte to directly infer [Ca²⁺]_mito values. Diastolic [Ca²⁺]_mito was ~150 nM, with a ~30 nM Δ[Ca²⁺]_mito during individual Ca²⁺ transients at 0.2 Hz pacing frequency. This Δ[Ca²⁺]_mito amplitude is larger than in our previous study²⁰ in permeabilized cardiomyocytes during spontaneous SR Ca²⁺ release waves with internal [Ca²⁺] of 150 nM. Differences could be due to the intracellular buffer, SR Ca²⁺ content, lower frequency and unsynchronized nature of Ca²⁺ waves in permeabilized myocytes. The rapid intra-mitochondrial Ca²⁺ buffering power in these intact myocytes is unknown, but a value of 33–100 (Δbound/Δfree; similar to the ~100 in cytosol) would imply a total mitochondrial Ca²⁺ uptake during a twitch of 0.5–1.6 µmol/l cytosol. That is consistent with quantitative analysis of Ca²⁺ transport rates during normal Ca²⁺ transients in rabbit ventricular myocytes by SR Ca-ATPase, Na/Ca exchanger, mitochondrial uptake and plasma membrane Ca-ATPase (54, 21, 0.8 and 0.8 µmol/l cytosol, respectively).^25,26 Thus mitochondrial uptake is ~1% of the total Ca²⁺ removed from the cytosol at each beat. That also agrees with the lack of significant impact of Ru360 on the amplitude and kinetics of [Ca²⁺]_i transients here (Fig 2A). If mitochondrial uptake rate and buffering are upregulated at higher cellular Ca²⁺ loads, the impact could be increased.^14,27

The time to peak [Ca²⁺]_mito (236 ± 48 ms) and slow kinetics of [Ca²⁺]_mito decline (τ~5 s) are consistent with our previous data,^20,28 but much faster [Ca²⁺]_mito declines have also been reported,^14,18 especially with stronger Ca²⁺ loading conditions (isoproterenol and/or elevated extracellular [Ca²⁺]). We think that our slow rates of [Ca²⁺]_mito decline are consistent with the known slow Ca²⁺ extrusion rate via mitochondrial Na/Ca exchanger, but we suspect that faster reported rates might also involve increased intramitochondrial Ca²⁺ buffering at higher Ca²⁺ loads.^27,29

We found that ISO increased the [Ca²⁺]_mito transient amplitude, which may be driven by the larger SR Ca²⁺ content and Ca transients induced by β-adrenergic signaling, resulting in higher local [Ca²⁺]_i and promoting Ca²⁺ influx into mitochondria. ISO also accelerated [Ca²⁺]_mito decline, consistent with data from other groups.^14,18 Twitch [Ca²⁺]_i decline is accelerated by ISO because of PKA-dependent phosphorlamban phosphorylation and accelerated SR Ca-ATPase activity.²⁶ However, it is not clear that that effect would suffice to hasten [Ca²⁺]_mito decline because [Ca²⁺]_i is already near diastolic levels during most of the [Ca²⁺]_mito decay time. Since [Ca²⁺]_mito decline relies mainly on mitochondrial Na/Ca exchange (mNCX), slower decline could result from elevated [Na⁺]_i. However, PKA-dependent phospholemman phosphorylation and Na/K-ATPase stimulation limit the rise in [Na⁺]_i expected from other effects of ISO,³⁰ making that explanation unlikely. It is possible that ISO could result in activation of the mNCX, but such regulation has not been described.

We focused on a low stimulation frequency here for two reasons. First, this allows [Ca²⁺]_mito to largely recover between beats and attain steady state. Second, at higher frequency the amplitude of [Ca²⁺]_mito transients becomes smaller as [Ca²⁺]_mito accumulates. Moreover, as [Ca²⁺]_mito rises it may begin to approach saturation for Mitycam. While we did not approach that limit here, lower affinity mitochondrial Ca²⁺ sensors might be valuable to examine the full physiological range of [Ca²⁺]_mito at high Ca²⁺ loading conditions and under pathological conditions. These local [Ca²⁺]_mito gradients detected may produce local increases in mitochondrial dehydrogenase activity and ATP production, because the [Ca²⁺]_mito levels are in the range where these enzymes are Ca²⁺-sensitive.¹ In that sense the subcellular regions where Ca²⁺ transients are highest may have enhanced ATP production, matching supply and demand. We suggest that at more physiological heart rates and temperature that [Ca²⁺]_mito will be somewhat higher than values reported here, but also that the phasic [Ca²⁺]_mito signals and associated spatiotemporal [Ca²⁺]_mito gradients will be more limited. Thus, the true physiological impact of these [Ca²⁺]_mito gradients on cardiac energy balance will require further study.

A central aim here was to assess whether spatial [Ca²⁺]_mito gradients could be detected during cardiac Ca²⁺ transients. Ultrastructural evidence exists for proximity and even explicit tethering of mitochondria to the SR membrane, suggesting that diffusional distance from SR junctional couplings is 37–270 nm from the end of a nearby mitochondrion.^4–6,31,32 Indeed, in permeabilized cells SR Ca²⁺ release drives mitochondrial Ca²⁺ uptake that is less sensitive to cytosolic Ca²⁺ buffers than is global [Ca²⁺]_i suggesting some preferential local mitochondrial Ca²⁺ uptake.^31,33

The discernible spatiotemporal [Ca²⁺]_mito gradient between Z- and M-lines is consistent with the idea that the part of a mitochondrion nearest the SR Ca²⁺ release sites exhibits preferential Ca²⁺ uptake (Fig 6B) despite relatively uniform sarcomeric [Ca²⁺]_i during ECC (Online Fig III). Of course, much of the total mitochondrial surface (e.g. that near M-lines) is far from the SR Ca²⁺ release sites and is expected to sense a similar local [Ca²⁺]_i as that sensed by the myofilaments. These subsarcomeric [Ca²⁺]_mito gradients are near the limit of spatial resolution, and the true [Ca²⁺]_mito gradient seems to be larger and dissipates over ~0.5 µm (Fig 5E). We also used the same analytical methods to see whether similar cytosolic spatial gradients would be detectable along the sarcomere (using Fluo-4 AM as Ca²⁺ indicator, and Di-8ANNEPS for Z-line; Online Fig I). We could not readily detect such gradients. That is not particularly surprising because either isolated local release events (Ca sparks) or special conditions are required (combining EGTA with low affinity indicators) to detect local high [Ca²⁺]_i near release sites.^34,35 Likewise, gradients of free intra-SR [Ca²⁺] are not normally discernible during normal ECC, even using methods like those used here,³⁶ although such local gradients in both [Ca²⁺]_i and SR [Ca²⁺] are readily detected during isolated local release events (Ca sparks).

Our results here on local and calibrated [Ca²⁺]_mito place new explicit spatiotemporal constraints on models of Ca²⁺ uptake and extrusion from mitochondria. Detailed diffusional-flux models will be required to determine whether these [Ca²⁺]_mito gradients could be a simple consequence of the geometric position of mitochondria with respect to junctions, or whether more specialized communication is necessary. The approach described here will be valuable for clarifying many aspects of mitochondrial Ca²⁺ regulation.

Novelty and Significance.

What Is Known?

• Mitochondria are located close to cytosolic [Ca²⁺] ([Ca²⁺]_i) release sites (i.e. ryanodine receptors of the sarcoplasmic reticulum, SR), and crosstalk may facilitate the excitation-metabolism coupling.

• Mitochondria take up Ca²⁺ uptake via a low affinity Ca²⁺ uniporter (MCU). This requires a high local [Ca²⁺], which occurs near the release sites. However, intra- mitochondrial [Ca²⁺] gradients have not been measured and the kinetics of mitochondrial [Ca²⁺]_i rise during normal Ca transients are debated (large phasic Ca²⁺ transients vs. slow integrating changes).

• A genetically- encoded, Ca²⁺ sensor located in the mitochondria could measurr mitochondrial [Ca²⁺]_mito in intact cardiac myocytes.

What New Information Does This Article Contribute?

• In cardiomyocyte mitochondria in situ, the Ca²⁺ sensor Mitycam has a K_d value (~200 nM).

• The measurement of Mitycam signal provides quantitative estimates of [Ca²⁺]_mito transients in ventricular myocytes.

• During normal calcium transients in adult ventricular myocytes, subsarcomeric spatial [Ca²⁺]_mito gradients with ~50% larger amplitude reache an earlier peak near SR Ca²⁺ release sites (at the Z-line) than in the middle of the sarcomere (M-line).

• Measurements of spatial [Ca²⁺]_mito gradients using this approach could be useful for investigating mitochondrial Ca²⁺ handling in many conditions.

During excitation-contraction coupling (ECC) cardiac myocytes increases in [Ca²⁺]_mito can enhance ATP synthesis via Ca²⁺-dependent dehydrogenases. The highly organized juxtaposition of sarcolemma, SR and mitochondria provides a possibility for their crosstalk. However, the kinetics and amplitude of [Ca²⁺]_mito remain unknown and it is unclear whether spatiotemporal [Ca²⁺]_mito gradients exist upon SR Ca²⁺ release during ECC. Here we report quantitative estimates of [Ca²⁺]_mito transients in intact adult ventricular myocytes and measure subsarcomeric spatial [Ca²⁺]_mito gradients during normal Ca transients (larger and faster near the Z- vs. M-line). The amplitude and kinetics of [Ca²⁺]_mito transients rises quickly along with cytosolic [Ca²⁺], but are much smaller in amplitude and decay slowly, leading to slow progressive changes during repeated stimuli. This approach to measuring [Ca²⁺]_mito will be useful to further our understanding of how mitochondria handle Ca²⁺ in spatiotemporal detail and also how [Ca²⁺]_mito regulates mitochondrial function under various physiological and pathological conditions.

Non-standard Abbreviations

[Ca²⁺]

calcium concentration

[Ca²⁺]_i

cytosolic free Ca²⁺ concentration

[Ca²⁺]_mito

mitochondrial free Ca²⁺ concentration

Δ[Ca²⁺]_mito

concentration change of mitochondrial Ca²⁺

ECC

excitation-contraction coupling

EMC

excitation-metabolism coupling

ISO

isoproterenol

MCU

mitochondrial Ca²⁺ uniporter

mNCX

mitochondrial Na/Ca exchange

MOI

multiplicity of infection

MPTP

mitochondrial permeability transition pore

SERCA

SR Ca-ATPase

SR

sarcoplasmic reticulum

τ

time constant