A modified calcium retention capacity assay clarifies the roles of extra- and intracellular calcium pools in mitochondrial permeability transition pore opening

Calcium homeostasis is essential for cell survival and is precisely controlled by several cellular actors such as the sarco/endoplasmic reticulum and mitochondria. Upon stress induction, Ca2+ released from sarco/endoplasmic reticulum stores and from extracellular Ca2+ pools accumulates in the cytosol and in the mitochondria. This induces Ca2+ overload and ultimately the opening of the mitochondrial permeability transition pore (mPTP), promoting cell death. Currently, it is unclear whether intracellular Ca2+ stores are sufficient to promote the mPTP opening. Ca2+ retention capacity (CRC) corresponds to the maximal Ca2+ uptake by the mitochondria before mPTP opening. In this study, using permeabilized cardiomyocytes isolated from adult mice, we modified the standard CRC assay by specifically inducing reticular Ca2+ release to investigate the respective contributions of reticular Ca2+ and extracellular Ca2+ to mPTP opening in normoxic conditions or after anoxia–reoxygenation. Our experiments revealed that Ca2+ released from the sarco/endoplasmic reticulum is not sufficient to trigger mPTP opening and corresponds to ∼50% of the total Ca2+ levels required to open the mPTP. We also studied mPTP opening after anoxia–reoxygenation in the presence or absence of extracellular Ca2+. In both conditions, Ca2+ leakage from internal stores could not trigger mPTP opening by itself but significantly decreased the CRC. Our findings highlight how a modified CRC assay enables the investigation of the role of reticular and extracellular Ca2+ pools in the regulation of the mPTP. We propose that this method may be useful for screening molecules of interest implicated in mPTP regulation.

and the mitochondria occurs in microdomains known as mitochondria-associated membranes, which have been reported to control specific cell functions (reviewed in Ref. 2) such as mitochondrial bioenergetics, lipid metabolism, and cell fate. Disruption of mitochondria-associated membranes has been associated to different pathologies including hypoxia-reoxygenation (3)(4)(5). Ischemia or hypoxia induces a drop in ATP content, which in turn decreases ATP-dependent Ca 2ϩ pumps in both sarcolemma (plasma membrane Ca 2ϩ ATPase) and sarco/endoplasmic reticulum (SERCA), 3 leading to an increase in cytosolic (6) and consequently mitochondrial (7) Ca 2ϩ concentration. This Ca 2ϩ overload and enhancing factors like reactive oxygen species (8), partial mitochondrial membrane depolarization (9), pH restoration at reperfusion (10,11), gangliosides (12), P i (13), and outer mitochondrial membrane components (14) induce mitochondrial swelling and the opening of mPTP and ultimately promote cell death (1,15,16).
Although mitochondrial Ca 2ϩ overload has been well-identified as being a main contributor to the latter phenomena, the origin of this Ca 2ϩ is still unclear. Little is known about whether the internal Ca 2ϩ stores alone are sufficient to promote mPTP opening in cardiomyocytes or whether extracellular Ca 2ϩ entry is also required. Growing evidences have involved sarcolemmal channels and transporters such as the connexin and pannexin hemichannels (17) or the Na ϩ /Ca 2ϩ exchanger NCX (18). Their inhibition reduced Ca 2ϩ overload and decreased myocardial injury, suggesting the involvement of extracellular Ca 2ϩ . However, none of these studies have evaluated the respective participation of this extracellular Ca 2ϩ and the internal Ca 2ϩ stores to the mitochondria Ca 2ϩ overload process. Answering this question may help rationalizing therapeutic strategies relying on Ca 2ϩ fluxes modulation.
This maximum ability of Ca 2ϩ uptake by mitochondria before mPTP opening and mitochondrial swelling is defined as Ca 2ϩ retention capacity (CRC) (reviewed in Ref. 19). This feature of mitochondria may vary according to the cell type, as well as the physiological (e.g. oxidative stress and pH) or pathologi-cal cellular state. CRC measurement is experimentally performed on isolated mitochondrial preparations or on cellular models and is conventionally realized with the addition of exogenous Ca 2ϩ enabling the simple quantification of the Ca 2ϩ amount required to open mPTP. Therefore, the role of reticular Ca 2ϩ stores (the major Ca 2ϩ store besides the mitochondria) in the mitochondrial Ca 2ϩ overload was not studied.
Although several factors are implicated in mPTP opening, numerous studies have shown evidence of the crucial implication of Ca 2ϩ overload in mPTP opening especially during ischemia reperfusion (20,21). In this work we focused particularly on Ca 2ϩ contribution. However, we ensure that the other factors contributing to mPTP opening are still present in our experimental condition by using a cellular environment. We investigated concomitantly the role of reticular and extracellular Ca 2ϩ in the regulation of mPTP on a model of isolated cardiomyocytes. We modified the standard CRC method by combining the effect of extracellular Ca 2ϩ (depicted by the addition of Ca 2ϩ pulses) and reticular Ca 2ϩ mobilization by stimulating ryanodine receptors using caffeine or ryanodine. We validated our experimental model by showing that, as expected, both pharmacological treatment of cardiomyocytes with cyclosporine A (CsA), an inhibitor of cyclophilin D (CypD), or the genetic ablation of CypD in cyclophilin D knockout (CypD-KO) cardiomyocytes required a greater amount of Ca 2ϩ in the mitochondria to open mPTP. In addition, we demonstrated that in both normoxia and anoxia-reoxygenation (AR) conditions, reticular Ca 2ϩ stores alone were not sufficient to trigger mPTP opening and that extracellular Ca 2ϩ import (Ca 2ϩ pulses addition) was required to open mPTP. Finally, we proposed an estimation of these Ca 2ϩ sources contributions to mPTP opening.

Validation of the experimental model
As shown in Fig. 1 (A and B), Ca 2ϩ transfers were properly recorded with both probes in isolated cardiomyocyte preparation. Ca 2ϩ Green-5N probe reported Ca 2ϩ uptake by the mitochondria after each Ca 2ϩ pulse, followed by a spontaneous cytoplasmic release indicating mPTP opening (Fig. 1A). Conversely, Ca 2ϩ uptake by mitochondria was reported by the increase in Rhod2-AM fluorescence after each Ca 2ϩ pulse ( Fig.   Figure 1. CRC in adult cardiomyocytes. A and B, CRC was performed in the presence of Ca 2ϩ Green-5N probe to detect extramitochondrial Ca 2ϩ (A) or in the presence of Rhod2-AM probe to detect intramitochondrial free Ca 2ϩ (B). C, cardiomyocytes were incubated for 900 s in CRC medium before CRC measurement. D, cardiomyocytes were incubated in CRC medium and recorded for 2000 s in the presence only of Ca 2ϩ Green-5N to assess cellular integrity and stability. For all experiments, fluorescence (F) is expressed in A.U. Ca 2ϩ was added every 120 s by increments of 20 nmol/injection. E, confocal microscopy images were acquired after 30 min incubation of cardiomyocytes with 5 M Rhod2-AM in the presence or absence of 40 M digitonin. F, confocal microscopy images were acquired after 30 min of incubation of cardiomyocytes with 5 M Rhod2-AM and 0.2 M MitoTracker Green. 1024 ϫ 1024 pixels images were acquired with an average of four scanning lines. Displayed graphs and images are representative of three independent experiments. 1B), proving that exogenous Ca 2ϩ added during the assay was indeed taken up by the mitochondria. When cardiomyocytes were subjected to ϳ900 s of incubation before Ca 2ϩ pulses (Fig.  1C), CRC was still similar to basal condition ( Fig. 1A) (i.e. 10 Ca 2ϩ pulses were needed to open mPTP). In addition, cardiac mitochondria in situ and permeabilized cellular integrity was preserved during at least 2000 s of incubation in the presence of Ca 2ϩ Green-5N probe (Fig. 1D). Fluorescence was stable throughout the experiment reflecting the absence of intracellular Ca 2ϩ leak and the absence of fluorescence bleaching. We also used Rhod2-AM staining to confirm that the mitochondrial network was not affected in permeabilized cardiomyocytes (40 M digitonin) (Fig. 1E) and that Rhod2-AM colocalized with MitoTracker Green, proving a strict mitochondrial localization of Rhod2-AM staining in our experimental conditions (Fig. 1F).

External and reticular Ca 2؉ stores in mPTP opening
Another control experiment for our method validation was performed on isolated mitochondria (250 g of protein). Standard CRC was performed in basal conditions ( Fig. 2A), and in the presence of 5 mM caffeine (Fig. 2B) or 10 nM ryanodine (Fig. 2C). CRC values were similar in all experiments (Fig. 2D), indicating that, at the used concentrations, these ryanodine receptors agonists had no direct mitochondrial targets.

Effects of the reticular Ca 2؉ release on mPTP opening
To study the effect of reticular Ca 2ϩ on mPTP opening, CRC was evaluated either by adding exogenous Ca 2ϩ pulses alone (Fig. 3A, black panel) or by using 5 mM caffeine or 10 nM ryanodine followed by exogenous Ca 2ϩ pulse addition (Fig. 3A, gray and blue panels, respectively). In basal conditions (Fig. 3, B and C, black plots), 10 pulses (20 nmol Ca 2ϩ /pulse) of exogenous Ca 2ϩ were needed to open mPTP. The addition of caffeine (Fig.  3B, gray plot) or ryanodine (Fig. 3C, blue plot) did not induce fluorescence elevation, suggesting that mitochondria take up the amount of Ca 2ϩ released by caffeine or ryanodine stimulation (Fig. 3, B and C, insets). This was confirmed when cardiomyocytes preincubated with 1 M FCCP were stimulated with 5 mM caffeine: an increase in extramitochondrial fluorescence was observed (Fig. 3D), representing the caffeine-induced reticular Ca 2ϩ release that cannot be taken up by the mitochondria. The same observation was made after stimulation with 5 mM caffeine in the presence of 1 M RU360, a specific mitochondrial Ca 2ϩ intake inhibitor (Fig. 3E).
No massive Ca 2ϩ depletion from mitochondria was observed after a 900-s caffeine treatment, indicating that the amount of Ca 2ϩ released from reticulum stores was not sufficient to open mPTP. Five additional pulses of 20 nmol of Ca 2ϩ were required to induce mPTP opening (Fig. 3, B and C). The same experiments performed in Rhod2-AM-loaded cardiomyocytes showed an increase in fluorescence after stimulation with 5 mM caffeine (Fig. 3F, gray plot and inset) or 10 nM ryanodine (Fig. 3F, blue plot and inset). Overall, these experiments show that the caffeine-induced Ca 2ϩ is quickly and efficiently taken up by mitochondria with a limited Ca 2ϩ leak toward cytosol.
Maximal release of reticular Ca 2ϩ was triggered by three repetitive stimulations with 5 mM caffeine. However, the amount of Ca 2ϩ released was still insufficient, and additional pulses of 20 nmol Ca 2ϩ were necessary to open mPTP (data not shown).

External and reticular Ca 2؉ stores in mPTP opening
mPTP opening in the basal condition (Fig. 3G). This suggests that the reticular Ca 2ϩ stores of cardiomyocytes cannot induce mPTP opening by itself.

Involvement of extracellular Ca 2؉ in mPTP opening
To confirm our previous results, we assessed whether preloading cardiac mitochondria with 50% of the Ca 2ϩ amount necessary to open mPTP could allow the reticular Ca 2ϩ content to drive mPTP opening (Fig. 4A). In three independent experiments, the addition of 5 mM caffeine (Fig. 4B) or 10 nM ryanodine (Fig. 4C) induced a strong and sustained increase in fluorescence intensity indicating a massive mitochondrial Ca 2ϩ leak. The reticular Ca 2ϩ release induced by either caffeine or ryanodine was sufficient to trigger mPTP opening when cardiomyocytes were loaded with 50% of the Ca 2ϩ amount necessary to open mPTP. Both caffeine and ryanodine induced a massive reticular Ca 2ϩ release after mitochondria Ca 2ϩ preloading, suggesting an absence of or a limited induction of the Ca 2ϩ -induced Ca 2ϩ -release (CICR) mechanism. This was confirmed by the addition of one 20-nmol Ca 2ϩ pulse after the massive fluorescence increase induced by caffeine and ryanodine. This Ca 2ϩ pulse was not taken up by the mitochondria, ensuring that the mPTP was indeed opened. As a control of the potential effect of the experiment duration on mPTP opening, we preloaded cardiomyocytes with the same Ca 2ϩ amount and recorded fluorescence without the addition of caffeine or ryanodine. In this condition, mPTP did not open, even after 2000 s of incubation time (Fig. 4D), demonstrating that mPTP opening observed in Fig. 4 (B and C) was triggered by caffeine-or ryanodine-induced Ca 2ϩ release. In the presence of 110 Ϯ 10-nmol Ca 2ϩ load, no additional external Ca 2ϩ was needed to open mPTP after internal Ca 2ϩ stimulation with caffeine or Figure 3. Reticular Ca 2؉ contribution to mPTP opening. A, CRC measurement was performed using Ca 2ϩ Green-5N probe and conditions. B and C, Ca 2ϩ pulses added in basal condition (black plots) were compared with Ca 2ϩ pulses added after stimulation with 5 mM caffeine (B, gray plot) or 10 nM ryanodine (C, blue plot) in adult cardiomyocytes. D and E, as a control, extramitochondrial Ca 2ϩ fluorescence was measured in cardiomyocytes preincubated with 1 M FCCP (D) or in the presence of 1 M RU360 (E) after reticular Ca 2ϩ release stimulation with 5 mM caffeine. Mitochondrial free Ca 2ϩ was also stained using Rhod2-AM after 5 mM caffeine or 10 nM ryanodine stimulation (F, gray plot and blue plot, respectively). At 900 s, a series of 20 nmol of Ca 2ϩ pulses were necessary to induce mPTP opening. In B and C, insets represent Ca 2ϩ Green fluorescence variation from 0 to 500 s and reflect extramitochondrial Ca 2ϩ after ryanodine receptor stimulation. In F, the inset represents Rhod2-AM fluorescence variation from 0 to 500 s and reflects mitochondrial free Ca 2ϩ uptake after ryanodine receptor stimulation. G, Ca 2ϩ amount added to open mPTP in basal condition or after reticular Ca 2ϩ stimulation was quantified and presented as dot plots. Fluorescence (F) is expressed in A.U. Ca 2ϩ was added every 120 s by increments of 20 nmol/injection. The displayed graphs are representative of three or four independent experiments. The values in G are presented as means Ϯ S.D. nmol Ca 2ϩ /mg protein of four independent experiments (Kruskal-Wallis; H ϭ 7.69, p ϭ 0.0081), followed by Dunn's post-test. a, p Ͻ 0.05 versus basal.

External and reticular Ca 2؉ stores in mPTP opening
ryanodine (Fig. 4E). These results are consistent with the previous observation from Fig. 3 and confirm our estimation that caffeine/ryanodine-induced reticular Ca 2ϩ release represented ϳ50% of the Ca 2ϩ amount necessary to open mPTP.
One of the most studied cardioprotective strategy within the last decade has relied on either the pharmacological inhibition or the genetic ablation of cyclophilin D, which, under Ca 2ϩ stimulation, is known to enhance mPTP opening. Assuming that the reticular Ca 2ϩ content is stable, we wondered whether CypD inhibition/suppression would also shift the amount of extracellular Ca 2ϩ required to open mPTP in our experimental model. We first determined the amount of Ca 2ϩ necessary to open mPTP after CypD inhibition/suppression and then loaded cardiomyocytes with 50% of this amount prior to inducing reticular Ca 2ϩ release with 10 nM ryanodine (Fig. 5A). When CsA was added to the preparation, an additional 63.3 Ϯ 14.5 nmol Ca 2ϩ /mg protein were necessary to induce mPTP opening after ryanodine stimulation (Fig. 5B, blue plot). If rotenone, a complex I inhibitor, was added to CsA, the amount of Ca 2ϩ required to open mPTP was raised to 133.7 Ϯ 27.0 nmol Ca 2ϩ /mg protein (Fig. 5C, blue plot). Similar results were observed in CypD-KO adult cardiomyocytes (Fig. 5D, blue plot), in which the addition of 177.7 Ϯ 15.3 nmol Ca 2ϩ /mg protein was necessary to induce mPTP opening. Compared with the control condition tested in Fig. 4C (where no addi-tional Ca 2ϩ was needed to open mPTP after stimulation of reticular Ca 2ϩ release), the inhibition/suppression of CypD in our model significantly increase mitochondria capacity to uptake Ca 2ϩ (Fig. 5E).

Effect of reticular Ca 2؉ mobilization on mPTP opening after AR
Altogether, our results emphasized the major contribution of both internal Ca 2ϩ stores and extracellular Ca 2ϩ to mitochondrial/cellular fate in normoxic condition. We next assessed whether AR would modify this dynamic.
Cardiomyocytes were subjected to 30 min of anoxia followed by a 15-min reoxygenation period (in the presence or absence of 1 mM extracellular Ca 2ϩ ) prior to CRC measurement. As a control (sham groups), cardiomyocytes were incubated for 45 min in normoxic condition also in the presence or absence of 1 mM extracellular Ca 2ϩ (Fig. 6A).
AR protocol was determined based on our previous work and on a recent study from Panel et al. (22), to guarantee a sufficient amount of viable cardiomyocytes after AR and to achieve a valuable measurement of mPTP opening. The amount of Ca 2ϩ needed to open mPTP was determined (as in Fig. 3) either using only Ca 2ϩ pulses addition (sham groups and AR1 groups) or External and reticular Ca 2؉ stores in mPTP opening using 5 mM caffeine (to induce reticular Ca 2ϩ release) followed by Ca 2ϩ pulse addition (AR2 groups) (Fig. 6A).
As expected, the amount of Ca 2ϩ added to open mPTP decreased after AR protocol. AR1 with or without 1 mM Ca 2ϩ compared with respective sham (Fig. 6B): 108 Ϯ 19 versus 210 Ϯ 35 nmol Ca 2ϩ /mg protein (p Ͻ 0.0001; with 1 mM extracellular Ca 2ϩ ) and 193 Ϯ 22 versus 290 Ϯ 40 nmol Ca 2ϩ /mg protein (p Ͻ 0.0001; without 1 mM extracellular Ca 2ϩ ). In AR2, our results showed that even after AR protocol, either with or without 1 mM extracellular Ca 2ϩ , reticular Ca 2ϩ release after caffeine stimulation was not sufficient to open mPTP. Additional Ca 2ϩ pulses were added to trigger mPTP opening.
Moreover, in the presence of 1 mM extracellular Ca 2ϩ , CRC values significantly decreased when compared with the respective condition in the absence of 1 mM extracellular Ca 2ϩ (sham, AR1, and AR2 with 1 mM extracellular Ca 2ϩ versus sham, AR1, and AR2 without 1 mM extracellular Ca 2ϩ , respectively; p Ͻ 0.0001). Strikingly, in none of these conditions were intracellular Ca 2ϩ stores sufficient to trigger mPTP opening, even though the presence of extracellular Ca 2ϩ during AR facilitates mPTP opening. In all these conditions, the addition of exogenous Ca 2ϩ pulses was needed to induce mPTP opening. Interest-ingly, other biophysical features can be evaluated using our measurement method, which is described below.
The mobilized internal Ca 2ϩ stores-These stores can be estimated by measuring the difference between CRC value in sham without extracellular Ca 2ϩ and the CRC value in AR2 without extracellular Ca 2ϩ ( Table 1). The mobilized internal Ca 2ϩ stores in our model would be ϳ187 Ϯ 38 nmol Ca 2ϩ /mg protein.
The remaining internal Ca 2ϩ stores after AR-The measured CRC values without 1 mM extracellular Ca 2ϩ were significantly different in AR1 compared with AR2 (p Ͻ 0.005). Therefore, the remaining internal Ca 2ϩ stores after AR can be estimated by subtracting CRC value in AR1 from CRC value in AR2 without 1 mM extracellular Ca 2ϩ and were equivalent to 80 Ϯ 39 nmol Ca 2ϩ /mg protein ( Table 1).
The proportion of internal Ca 2ϩ stores leak during AR-Knowing both the total internal Ca 2ϩ amount (187 Ϯ 38 nmol Ca 2ϩ /mg protein) and the remaining internal Ca 2ϩ after AR (80 Ϯ 39 nmol Ca 2ϩ /mg protein), we can estimate the internal Ca 2ϩ that leaked during AR to 57 Ϯ 23% ( Table 1).
The extracellular Ca 2ϩ amount that entered the cell during AR-The difference between the CRC values of AR2 in the absence and the presence of 1 mM extracellular Ca 2ϩ represents

External and reticular Ca 2؉ stores in mPTP opening
an estimation of the amount of extracellular Ca 2ϩ that entered the cells during AR and was equal to 48 Ϯ 14 nmol Ca 2ϩ /mg protein ( Table 1). The 48 Ϯ 14 nmol Ca 2ϩ /mg protein represented approximately half of the internal Ca 2ϩ stores that leaked to the mitochondria (187 Ϫ 80 ϭ 107 nmol Ca 2ϩ /mg protein). Overall, we can conclude that in our experimental conditions, two-thirds of the Ca 2ϩ loaded to the mitochondria during AR may come from internal Ca 2ϩ stores (mainly from the reticulum), whereas approximately one-third seems to be of extracellular origin.

Discussion
Mitochondria are involved in multiple cell processes such as the control of cytoplasmic Ca 2ϩ level because of their Ca 2ϩ uptake capacity. This participation is particularly important when the cellular environment is subjected to major stress such as ischemia-reperfusion during which the absence of oxygen induces the drop of ATP synthesis and consequently the inhi-bition of SERCA activity contributing thus to several physiological mishandling including a massive intracellular Ca 2ϩ overload (23). In such a situation, mPTP transient activity could act as a safety valve, which enables Ca 2ϩ leak with an increase rate of efflux. Ultimately, this mechanism triggers mitochondrial swelling, which has been associated with cell death. This finely tuned regulation, balanced by Ca 2ϩ and mediated by CypD, has been extensively studied (24 -26). However, in most of the studies, no distinctions were made regarding the origin of Ca 2ϩ (internal Ca 2ϩ stores and/or extracellular Ca 2ϩ ) inducing the opening of mPTP, more likely because there was no reliable method to investigate it.
The proposed method in this study is derived from classical CRC assay and proposes a simple mean to quantitatively estimate the respective contribution of internal Ca 2ϩ stores and extracellular Ca 2ϩ toward mPTP opening. CRC can be achieved by the addition of exogenous Ca 2ϩ pulses to isolated mitochondria (27) or permeabilized cells (28). However, this classical CRC method does not allow the study of reticular Ca 2ϩ input. In parallel to CRC experiment, most of the standard protocols use Ca 2ϩ ionophores and Ca 2ϩ addition to induce mPTP opening. These Ca 2ϩ ionophores can induce artifacts in cellular models including in isolated adult cardiomyocytes (22) and can also induce a release of internal Ca 2ϩ stores in a nonspecific manner, as compared with caffeine or ryanodine treatment, which specifically induces reticular Ca 2ϩ release.
Interestingly, our presented method relies on the analysis of mitochondria-reticulum interaction in situ. We used freshly isolated adult murine cardiomyocytes as a cellular model. Cardiomyocytes plasma membranes were permeabilized with 40 M digitonin. This chemical agent and the used concentration are both respectful of the intracellular membrane structure (29 -31). The intracellular microdomains and organelle interactions are thus preserved in this model, allowing the study of mitochondria-reticulum interaction within an integrated physiological-like cellular environment (31,32). This was also shown by confocal microscopy images showing no impact of 40 M digitonin on the mitochondrial network in our model (Fig.  1E). We also confirmed the reliability of our cellular model and our experimental conditions, as well as the absence of Ca 2ϩ Green-5N fluorescence bleaching throughout the duration of our experiments. In our permeabilized model, the low ATP level ensures a natural inhibition of SERCA activity and therefore excludes reticular Ca 2ϩ uptake (33). In addition, with this method, one can activate the reticular Ca 2ϩ release via ryanodine receptors with specific agonists such as caffeine or ryanodine (34,35) to estimate its involvement in mPTP opening. In isolated adult cardiomyocytes, we estimated that reticular Ca 2ϩ represents ϳ50% of the Ca 2ϩ amount necessary to open mPTP in basal conditions. This proportion may represent an indicator of physiological Ca 2ϩ steady state between the mitochondria and the reticulum. A second biophysical parameter that can be deduced with this method is the proportion of Ca 2ϩ content that leaks from the internal stores during AR: ϳ57% during a 15-min reoxygenation. Finally, the third biophysical parameter that can be estimated is the proportion of extracellular Ca 2ϩ entering the cells during AR, which represents ϳ50% of the

External and reticular Ca 2؉ stores in mPTP opening
internal Ca 2ϩ leaking during AR protocol. Considering these two last parameters, we can suggest that ϳ15-20% of the total Ca 2ϩ content inducing mPTP opening comes from the extracellular Ca 2ϩ pulses added during the CRC experiments measured after AR protocol. To our knowledge, this is the first demonstration that internal Ca 2ϩ leaks during AR require additional factors like external Ca 2ϩ to open mPTP.
Another advantage offered by this method is the quantification estimate of the relationship between extracellular Ca 2ϩ content and steady-state mitochondrial Ca 2ϩ content. It has always been assumed that Ca 2ϩ homeostasis translates extracellular Ca 2ϩ changes to shifts in Ca 2ϩ concentration in the cytosol and organelles. Although the quantification of this interdependence could be realized in live cells with fluorescent probes, the calibration of these probes in mitochondria is much harder than in the cytosol. With our method, we found that the CRC value in cardiomyocytes incubated for 45 min (sham groups) in the absence of extracellular Ca 2ϩ was increased by 80 nmol Ca 2ϩ /mg protein compared with an incubation in the presence of 1 mM extracellular Ca 2ϩ . This emphasizes the fact that extracellular Ca 2ϩ contribute to the mitochondrial Ca 2ϩ homeostasis. Future experiments will be performed as a dose-effect response to fully quantify and understand this interdependence.
One limitation of our method could be that a proportion of the Ca 2ϩ released by caffeine leaks in the medium instead of entering the mitochondria. However, this would mean that upon caffeine stimulation, the fluorescence of Ca 2ϩ Green-5N probe should rise. The dissociation constant of Ca 2ϩ Green-5N probe is ϳ14 M. In our experimental conditions, the caffeinemediated Ca 2ϩ released is ϳ70 M (187 nmol Ca 2ϩ /mg protein with 750 g of protein in 2 ml of CRC medium). It is thus very unlikely that the probe could not detect a reticular Ca 2ϩ leak, if there is any. Along the same line, it should be noted that the insets in Fig. 3 (B and C) showed no significant Ca 2ϩ variation in the cytosol, suggesting that it was rapidly taken up by the mitochondria, whereas a fluorescence increase was observed in the mitochondria as shown by a Rhod2-AM probe, which is sensitive to free Ca 2ϩ variation (insets Fig. 3F). This was also confirmed by the absence of reticular Ca 2ϩ uptake in the presence of the mitochondrial uncoupler FCCP (Fig. 3D) and mitochondrial Ca 2ϩ uptake inhibitor, RU360 (Fig. 3E). These results highlight the efficiency of mitochondriareticulum connection to channel Ca 2ϩ between both organelles in isolated adult cardiomyocytes.
Another limitation is that in our experimental model, CICR is very limited because of cellular permeabilization and very low ATP amount. However, in the situation of ischemiareperfusion injury, the absence of ATP production makes CICR unlikely to occur, suggesting the absence of CICR contribution to mPTP opening during ischemia-reperfusion (36) similarly to our experimental model.
Finally, pH restoration during reperfusion, one of the factors that could participate in enhancing mPTP opening (11,37), is probably underestimated in our experimental model because of the use of buffered medium in AR protocol. Further experiments are needed to assess any synergetic activation of mPTP with this factor.
In this work, we showed that the mitochondrial Ca 2ϩ level required to open mPTP exceeded the reticular Ca 2ϩ content in both sham and AR experimental conditions either when the protocol was performed in the presence or in the absence of 1 mM extracellular Ca 2ϩ . In all these conditions, additional Ca 2ϩ pulses were needed to induce mPTP opening. This work emphasizes the prior knowledge that both extracellular Ca 2ϩ and internal Ca 2ϩ stores are important triggers of mPTP opening, by discerning, estimating, and contextualizing their respective contribution. Our results highlight the involvement of extracellular Ca 2ϩ to mitochondrial Ca 2ϩ homeostasis but also suggest that the mPTP opening process requires additional mechanisms such as protein and membrane oxidation by reactive oxygen species (38) or pH restoration (10,11).
This original and simple approach could be applied to a large panel of cell types and cellular models to detect in situ reticulum-mitochondria Ca 2ϩ transfer dysfunction and would open new perspectives for the study and screening of pharmacological molecules for mitochondrial and reticular targets. This method could thus bring new and valuable insights into the physiopathological investigation of numerous diseases.
The present study was in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health Publication 85- 23, revised 1996), and all experiments were approved by the University of Lyon Ethics Committee (UCBL1 approval BH2007-07).

Cardiomyocytes isolation
Adult mouse cardiomyocytes were freshly isolated using two different enzymatic digestions as previously described (39,40). Briefly, hearts from cervically dislocated male mice were

External and reticular Ca 2؉ stores in mPTP opening
quickly removed and retrogradely perfused with Krebs-Henseleit buffer. Ventricular cardiomyocytes were isolated using enzymatic digestion with 0.167 mg⅐ml Ϫ1 Liberase (Roche) and 0.14 mg⅐ml Ϫ1 trypsin 2.5% (Invitrogen), or with 2.4 mg⅐ml Ϫ1 collagenase type II (Gibco). Cellular protein concentration was measured using the Bradford method. Isolated cardiomyocytes viability was between 85 and 95%. The cells were used within 5 h after isolation.

Mitochondria isolation
Isolated mitochondria were used as part of the validation of the study model. After being removed, the hearts were quickly placed in an ice-cold isolation buffer (70 mM sucrose, 210 mM mannitol, and 10 mM EGTA, 50 mM Tris-HCl, pH 7.4). Myocardial tissue was finely minced and then homogenized in the same buffer. Mitochondria were isolated in accordance with our previous studies (39,41). Briefly, the homogenate was centrifuged at 1300 ϫ g for 3 min, and the supernatant was centrifuged at 10,000 ϫ g for 10 min. The mitochondrial pellet was then suspended in a cold buffer containing: 70 mM sucrose and 210 mM mannitol in 50 mM Tris-HCl, pH 7.4, and centrifuged at 10,000 ϫ g for 10 min. Mitochondrial protein concentration was measured using the Bradford method. Mitochondria were used within 3 h after isolation.

Ca 2؉ retention capacity assay
This study was mainly based on the measurement of CRC, an in vitro surrogate for the susceptibility of mPTP opening following a Ca 2ϩ overload. Briefly, the sample was placed in the CRC medium (150 mM sucrose, 50 mM KCl, 2 mM KH 2 PO 4 , 20 mM Tris-HCl, and 5 mM succinate-Tris) and under continuous stirring in the spectrofluorometer. After 120 s of stabilization time, 20 nmol of CaCl 2 (pulses) were added every 120 s. Modification of extramitochondrial Ca 2ϩ concentration was continuously recorded in the presence of a Ca 2ϩ fluorescent probe. After sufficient Ca 2ϩ loading, an increase in fluorescence intensity represents an elevation of extramitochondrial Ca 2ϩ concentration, indicating a massive release of Ca 2ϩ by mitochondria caused by mPTP opening. The amount of Ca 2ϩ necessary to trigger a massive Ca 2ϩ release, expressed as the CRC value, was used as an indicator of mPTP susceptibility to Ca 2ϩ overload.
CRC assay can be performed on isolated mitochondria or permeabilized cells and uses only exogenous Ca 2ϩ addition to achieve mitochondrial Ca 2ϩ overload. The purpose of the present study was to evaluate mitochondria-reticulum Ca 2ϩ cross-talk.
Consequently, (a) CRC was specifically performed on a model of permeabilized cardiomyocytes to benefit from the entire intracellular environment. (b) Intracellular Ca 2ϩ store release was triggered using caffeine or ryanodine addition. Another specification of the permeabilized cell context is the very low ATP amount. ATP is diluted in the assay medium and is not sufficient to induce SERCA activity (33), which guarantees no refilling of the sarco/endoplasmic stores and prevent CICR. (c) CRC assay was then performed as previously described. Ca 2ϩ fluorescence was measured using a spectrofluorometer (F-2500, Hitachi, High-tech) with two different Ca 2ϩ probes: Ca 2ϩ Green-5N and Rhod2-AM (Invitrogen). (d) Ca 2ϩ Green-5N probe (excitation, 506 nm; emission, 530 nm) detects extramitochondrial calcium. Cardiomyocytes (750 g of protein) were added in 2 ml of CRC medium. This medium was supplemented with 0.4 M Ca 2ϩ Green-5N and 40 M digitonin for cardiomyocytes permeabilization (42). (e) Rhod2-AM probe (excitation, 551 nm; emission, 582 nm) detects free intramitochondrial Ca 2ϩ and attests that Ca 2ϩ was actually taken up by the mitochondria. Cardiomyocytes (750 g of protein) were incubated with 5 M of Rhod2-AM for 30 min at room temperature and then washed for 30 min with a buffer containing 50 mM Tris-HCl, 70 mM saccharose, and 210 mM of mannitol, pH 7.4, to remove remaining Rhod2-AM, as well as Ca 2ϩ traces.

Confocal microscopy
Cardiomyocytes (750 g of protein) were incubated with 5 M of Rhod2-AM and 0.2 M MitoTracker Green under the same conditions described above. Images were taken on a confocal microscope Nikon A1r using an oil-immersion 40ϫ objective (N.A. 1.3). MitoTracker Green and Rhod2-AM were excited with 488 and 560-nm wavelength laser lines, respectively. The emitted light was filtered by 525 Ϯ 25 and 595 Ϯ 25 bandpass filters, respectively. 1024 ϫ 1024 pixels images were acquired with an average of four scanning lines.

Anoxia-reoxygenation protocol
The effect of depleting reticulum Ca 2ϩ stores on mPTP opening (as described above) was evaluated after AR in the presence of Ca 2ϩ Green-5N. Cardiomyocytes (750 g of protein) were subjected to 30 min of hypoxia followed by 15 min of reoxygenation in the presence or absence of 1 mM extracellular Ca 2ϩ (nonpermeabilized cardiomyocytes). Anoxia was achieved using CRC medium (with or without 1 mM Ca 2ϩ ) degassed for 5 min with nitrogen and supplemented with 0.5 mM dithionite. At the end of anoxia, the cells were washed and centrifuged at 20 ϫ g for 3 min and resuspended with 2 ml of fresh oxygenated CRC medium. Sham groups (cardiomyocytes placed for 45 min in normoxic conditions) were subjected to centrifugation/resuspension and used as controls for AR groups. At the end of reoxygenation time, CRC was measured following two different procedures: either using only Ca 2ϩ pulses addition (Sham groups and AR1 groups) or after the addition of 5 mM caffeine first (to release reticular Ca 2ϩ ) followed by the addition of Ca 2ϩ pulses to complete the measurement (AR2 groups).

Data processing and presentation
The animals were randomly distributed between groups. The data were analyzed using GraphPad Prism 6 (GraphPad software, San Diego, CA). Displayed graphs are representative of three independent experiments and presented as fluorescence arbitrary unit (A.U.) as a function of time in seconds.
Histograms and scatter plots are represented as means Ϯ S.D. nmol Ca 2ϩ /mg protein of three to five distinct experiments. Interactions and comparisons between groups were made using one-way ANOVA or two-way ANOVA followed by Tukey multicomparison post-test. Kruskal-Wallis followed by

External and reticular Ca 2؉ stores in mPTP opening
Dunn's multicomparison post-test were used as nonparametric tests in absence of Gaussian normality. The values for p Ͻ 0.05 were considered significant.