Mitochondrial Ca2+-Handling in Fast Skeletal Muscle Fibers from Wild Type and Calsequestrin-Null Mice

Mitochondrial calcium handling and its relation with calcium released from sarcoplasmic reticulum (SR) in muscle tissue are subject of lively debate. In this study we aimed to clarify how the SR determines mitochondrial calcium handling using dCASQ-null mice which lack both isoforms of the major Ca2+-binding protein inside SR, calsequestrin. Mitochondrial free Ca2+-concentration ([Ca2+]mito) was determined by means of a genetically targeted ratiometric FRET-based probe. Electron microscopy revealed a highly significant increase in intermyofibrillar mitochondria (+55%) and augmented coupling (+12%) between Ca2+ release units of the SR and mitochondria in dCASQ-null vs. WT fibers. Significant differences in the baseline [Ca2+]mito were observed between quiescent WT and dCASQ-null fibers, but not in the resting cytosolic Ca2+ concentration. The rise in [Ca2+]mito during electrical stimulation occurred in 20−30 ms, while the decline during and after stimulation was governed by 4 rate constants of approximately 40, 1.6, 0.2 and 0.03 s−1. Accordingly, frequency-dependent increase in [Ca2+]mito occurred during sustained contractions. In dCASQ-null fibers the increases in [Ca2+]mito were less pronounced than in WT fibers and even lower when extracellular calcium was removed. The amplitude and duration of [Ca2+]mito transients were increased by inhibition of mitochondrial Na+/Ca2+ exchanger (mNCX). These results provide direct evidence for fast Ca2+ accumulation inside the mitochondria, involvement of the mNCX in mitochondrial Ca2+-handling and a dependence of mitochondrial Ca2+-handling on intracellular (SR) and external Ca2+ stores in fast skeletal muscle fibers. dCASQ-null mice represent a model for malignant hyperthermia. The differences in structure and in mitochondrial function observed relative to WT may represent compensatory mechanisms for the disease-related reduction of calcium storage capacity of the SR and/or SR Ca2+-leakage.


Introduction
In all excitable and non-excitable cells, mitochondria take up and release Ca 2+ in a well regulated manner. This Ca 2+ transport is crucial for the regulation of mitochondrial activity [1] and is involved in the cascade of events leading to apoptosis [2,3]. It is well established that the activity of a number of key metabolic enzymes and the F0-F1 ATPase (ATP synthase) itself is controlled by the Ca 2+ concentration inside the mitochondrial matrix ([Ca 2+ ] mito ) [4,1]. This regulation is of special (patho)physiological interest in muscle tissue, where the high energy cost of contractile activity requires large (even 100 fold) and fast variations of aerobic ATP production [5,6].
Ca 2+ transport through the mitochondrial membranes has been the object of several studies in the last ten years, which have demonstrated that mitochondrial free Ca 2+ concentration undergoes rapid variations in parallel with the cytosolic Ca 2+ transients [7210]. The low affinity mitochondrial Ca 2+ uniporter (MCU) which has recently been identified as a 40-kDa protein located in the inner mitochondrial membrane [11,12] and the mitochondrial Na + /Ca 2+ exchanger (mNCX), e.g. [13] are considered as major pathways for mitochondrial Ca 2+ uptake and extrusion, respectively.
Evidence has been presented that microdomains are present near the Ryanodine receptor (RyR) where the local Ca 2+ concentration is high during the initial phase of the SR Ca 2+release process, see for a review [14,15]. Accordingly, the close proximity of mitochondria to the sarcoplasmic reticulum (SR), strengthened by the presence of small (10 nm) electron-dense linkages, or tethers [10,16], would facilitate exchange of Ca 2+ between the mitochondria and the SR during contraction and relaxation in skeletal muscle.
Our previous studies [17219] have shown that genetic ablation of calsequestrin has a very important effect on intracellular calcium homeostasis detectable when either the fast isoform CASQ1 or both CASQ1 and CASQ2 (the slow isoform) are removed. In those conditions Ca 2+ available inside the SR becomes insufficient for sustained contractile activity and muscle fibers become dependent on extracellular Ca 2+ which might enter the cells via store-operated Ca 2+ channels (SOCE) [17)]. Indeed, increased SOCE has been reported in skeletal muscle with reduced calsequestrin-1 expression [20]. In addition, the ablation of calsequestrin has been shown to be accompanied with an increase in mitochonchondrial volume [21] which might represent an additional compensatory mechanism. Therefore, the aim of this study was to determine whether and to which extent mitochondrial Ca 2+ handling is affected by a depletion of the SR calcium store and by the lowering of extracellular Ca 2+ concentration.
The intramitochondrial free Ca 2+ concentration [Ca 2+ ] mito was studied in adult skeletal muscle fibers using a ratiometric FRETbased probe (4mtD3cpv) which is specifically targeted to the mitochondrial matrix [22]. A comparison was made between the free mitochondrial Ca 2+ concentration in dCASQ-null (lacking both CASQ isoforms) and wild type fibers both at rest and during contractile activity induced by electrical stimulation. In addition we used electron microscopy to study the differences in mitochondrial density and in the local arrangement of the mitochondria and the SR between the dCASQ-null and wild type fibers.
The mitochondrial free Ca 2+ concentration was found to be higher at rest but lower during contraction in dCASQ-null fibers compared to wild type fibers. Detailed analysis of the kinetics of the changes in mitochondrial free Ca 2+ concentration during contraction revealed significant differences between dCASQ-null and WT fibers and significant changes therein upon removal of extracellular Ca 2+ . This indicates that mitochondrial Ca 2+handling in fast skeletal muscle fibers is sensitive to the calcium available in both intracellular and extracellular Ca 2+ -stores. The importance of these findings for altered muscle function in malignant hyperthermia will be discussed.

Animals
Experiments were carried out on 4 months old dCASQ-null and wild type (WT) mice. Double (d)CASQ-null mice [23] were generated by crossing CASQ1-null mice [18] with CASQ2-null mice [24]. Wild type C57BL/6J mice, same strain as dCASQ-null mice, were purchased from Charles River Laboratories (Wilmington, MA, USA). The use of the animals and the experimental protocol was approved by the Ethical Committee and by the animal welfare coordinator of the University of Padua. Mice were maintained in an accredited animal house and examined daily. Transfection 4mtD3cpv cameleon targeted to the mitochondrial matrix via a targeting sequence, derived from subunit VIII of human cytochrome c oxidase (COX), and replicated 4 times [25] in pcDNA3 (kindly donated by R.Y. Tsien (University of California, San Diego, CA, USA)) was used in transfection experiments. The structural and functional characterization of 4mtD3cpv is reported by Palmer et al. [22]. FDB muscles were transfected in vivo as described previously [26].

Isolation and culture of adult muscle fibers
Enzymatic isolation of single skeletal muscle fibers from FDB muscle was performed using collagenase treatment [23]. Isolated cells were seeded on laminin coated coverslips and left to adhere and to stabilize overnight in tissue culture medium at 36uC. The next morning the coverslips were mounted in a measuring chamber containing imaging buffer (composition in mM: NaCl 125, KCl 5, CaCl 2 1, MgSO 4 1, KH 2 PO 4 1, glucose 5, HEPES 20, pH 7.2) and equipped with platinum field electrodes. The zero external calcium solution used in part of the experiments had an identical composition except that CaCl 2 was omitted and 50 mM EGTA was added to scavenge any remaining Ca 2+ -ions. Fibers were incubated for 1 hour, prior to the measurements. As in our previous experiments we did not add MgCl 2 to replace CaCl 2 . Temperature during the measurements was kept at 25226uC, in order to slow the kinetics of Ca 2+ release and reuptake into the mitochondria, and to facilitate comparison with previous studies [17,18,21].

Immunocytochemistry
Fibers were fixed with 4% formalin in PBS for 30 minutes, permeabilized with 1% Triton and blocked in 10% goat serum for 1 hour to avoid non-specific signal. Fibers were incubated overnight at 4uC with primary antibodies specific for mitochondrial outer membrane protein Tom20 (FL-145, SantaCruz Biotechnology, Santa Cruz, USA) and ryanodine receptor (R129, Sigma-Aldrich, St. Louis, MO, USA) and with appropriate fluorescent secondary antibodies for 2 h at room temperature. The fibers were viewed with a confocal microscope (Leica SP2, Leica Microsystems, Rijswijk, Netherlands).

Electron microscopy (EM)
Electron-micrographs were obtained from FDB muscles as described previously [17,21]. Intermyofibrillar and subsarcolemmal mitochondrial density and their position relative to the CRUs were determined from 10 micrographs for each fiber (at 11,000X) of non-overlapping regions that were randomly collected from longitudinal sections. 10 fibers were analyzed in each animal (2 WT and 2 dCASQ-null mice) in a blinded fashion. Mitochondria and CRU/mitochondria couples were marked and counted in each micrograph and the area of the image was determined. Mitochondrial volume was determined using the well-established stereology point-counting techniques [27,28] in EM images taken at 14,000X of magnification after superimposing an orthogonal array of dots at a spacing of 0.20 mm to the electron micrographs. Subsarcolemmal mitochondria were counted in a layer underneath the sarcolemma with a thickness of 1 mm.

Measurements of the intramitochondrial free Ca 2+concentration
Measurements of the intramitochondrial free Ca 2+ -concentration ([Ca 2+ ] mito ) were carried out using an inverted fluorescence microscope (Eclipse-Ti, Nikon Instruments, Amsterdam, Netherlands) at 20X magnification equipped with the perfect focus system (Nikon Instruments) and a cooled CCD camera (C9100-13, Hamamatsu), essentially as described previously [17]. A relatively high excitation intensity of the YFP and CFP fluorophores and an image size of 176112 pixels per channel was required to achieve high time resolution (8 ms) in the measurements. The effect of bleaching of the Ca 2+ -indicator was studied in fibers stimulated at 0.1 Hz and occasionally at 0.033 Hz. At such low frequency of stimulation the recovery in the YFP/CFP ratio after the contraction is almost complete before the next stimulus starts and fibers, in addition, do not fatigue during the measurements. The results and the method to correct for bleaching are shown in the Supporting Information (Fig. S1 in File S1).
In a separate set of experiments, the time resolution of the recordings was reduced to 900 ms by increasing the integration time of the CCD camera and reducing the illumination intensity.
In this way bleaching of the probe was negligible and the duration of the recordings could be increased to assess very slow changes in Ca 2+ inside the mitochondria. In these latter experiments the effect of the mitochondrial Na + /Ca 2+ exchanger blocker (CGP37157, 1 mM; Tocris, Bristol, UK) was assessed.

Data analysis
The correction for bleaching and the subsequent analysis of the recordings was performed using a computer program written in MatLab (MatWorks, Natick, MA, USA). A third order polynomial fit to the initial and final 5-s periods of the recordings was calculated and subsequently subtracted from the data points to correct for the change in baseline as a result of bleaching. The intercept at t = 0 s of the polynomial fit to the baseline of the first fiber recorded in each dish was used to estimate the basal level of the free mitochondrial Ca 2+ -concentration in quiescent fibers. Additional information on the correction for bleaching of the probe is given in the Supporting Information (Fig. S2 in File S1) The time averaged steady-state responses during a train of stimuli at 1 Hz were used for the analysis of the early rapid alterations in free [Ca 2+ ] mito . The rise time (t 10-90% ) of the increase in free [Ca 2+ ] mito from 10 to 90% of the peak amplitude was calculated from the interpolated recording after correction for bleaching. The decline in free [Ca 2+ ] mito during 1 Hz stimulation was fitted with a double exponential: y(t) = a 0 +a 1 exp (2k 1 ?t) + a 2 exp (2k 2 ?t). The value of this equation at t = 1 second was used as the minimum value of the averaged recording in the determination of the rise time in order to minimize the effect of noise. The amplitude a 1 was calculated by extrapolating the double exponential to t = 0 seconds, the moment at which the maximum value was reached. The decline in free [Ca 2+ ] mito after the train of stimuli was fitted to a single exponential: y(t) = a 3 ?exp (2k 3 ?t), in which the parameter a 3 reflected the decrease in free [Ca 2+ ] mito after the train of stimuli as well as the (quasi steadystate) increase in free [Ca 2+ ] mito during the train of stimuli.
To resolve very slow components of the decline in free [Ca 2+ ] mito , we extended the recording of the decay phase after a train of 5 Hz stimulation to 10 minutes. In these recordings the integration time of the camera was increased to 900 ms and the illumination intensity was reduced accordingly in order to reduce bleaching of the probe. The final decline in free [Ca 2+ ] mito in these recordings was fitted to a single exponential: y(t) = a 4 ?exp (2k 4 ?t).

Calibration of the cameleon (4mtD3cpv)
Calibration of the Ca 2+ sensor was performed as describes previously [29]. The equation, [Ca 2+ ] = b?K' d ?[(R-R min )/(R max -R)] 1/n was used to estimate the free [Ca 2+ ] mito in quiescent fibers as well as the levels reached at the end of stimulation, using a K'd of the 4mtD3cpv sensor of 0.76 mM and n value of 0.74 [22]. Details on the estimation of Rmin and Rmax are given in File S1. A value of b of 6 was adopted to account for the difference between the in vitro and in situ K'd value [30].
Cytosolic Ca 2+ determination with Fura-2 AM Fibers were loaded with 5 mM Fura-2 acetoxymethyl ester (Life Technologies, Bleiswijk, Netherlands) in incubation buffer as described previously [18]. After a minimum of 30 minutes, Ca 2+ signals were recorded using a dual-beam excitation fluorescence photometry setup (IonOptix Corp. Milton, MA, USA) at 25-26uC. Calibration of the Fura-2 signals and calculation of the free cytosolic Ca 2+ concentration were as reported previously [17].

Statistical analysis
Data are expressed as mean 6 SEM. Comparisons between groups were made using two-way ANOVA (Type x Ca 2+ or Ca 2+ x Stimulation Frequency) or three-way ANOVA (Type x Ca 2+ x Stimulation Frequency (repeated measures), respectively) followed by Bonferroni post hoc test, if appropriate, to assess differences between WT vs. dCASQ-null, 1 mM vs. 0 external Ca 2+ , at the various stimulation frequencies used. P,0.05 was considered statistically significant.

Results
Intracellular distribution of the probe in enzymatically dissociated FDB muscle fibers 4mtD3cpv transfected fibers stained with an antibody specific for the mitochondrial protein Tom20 and an antibody specific for the ryanodine receptor (RyR) showed a striated sarcomeric pattern ( Fig. 1). Colocalization of 4mtD3cpv and the mitochondria was confirmed in the merged image and in the superimposed intensity profiles ( Fig. 1B and D). Antibody staining for RyR and 4mtD3cpv showed a ''sandwich'' pattern ( Fig. 1A and C) as expected on the basis of the position of the Ca 2+ release units relative to the mitochondria, resolved by electron microscopy (EM, see Fig. 2).

Structural changes in mitochondria and their coupling with Ca 2+ release units
The EM images shown in Fig. 2 illustrate that myofibrillar morphology in dCASQ-null fibers was well maintained. However, quantitative analysis revealed that intermyofibrillar mitochondrial density was increased in dCASQ-null (n = 195 micrographs) relative to WT fibers (n = 194) from 45.062.6 to 69.762.1 per 100 mm 2 i.e. by 55610%. The percentage of mitochondria coupled to Ca 2+ release units (CRUs) was already high in WT fibers (70.460.8%) and showed a further significant increase (+12.060.2% P,0.0001) up to 78.960.7% in dCASQ-null. In addition, the width of the terminal cisternae of the SR in dCASQnull fibers was reduced (Fig. 2B, D, dotted line) relative to WT, c.f. [18,21]. The intermyofibrillar mitochondrial volume fraction was increased from 4.360.2% in WT (n = 183) to 7.260.2% in dCASQ-null fibers (n = 192), i.e. by 6766%. This increase corresponds well to the increase in mitochondrial density, indicating that intermyofibrillar mitochondrial shape and surface to volume ratio in WT and dCASQ-null are similar.
The subsarcolemmal mitochondrial volume was rather similar in WT and dCASQ-null and amounted to 5.860.5% in WT (n = 120) and 4.760.5% in dCASQ-null (n = 157). Subsarcolemmal mitochondria were counted in a layer of 1 mm in width and the typical radius of the fibers studied was 20 mm. The subsarcolemmal volume fraction thus represents 1-(19/20) 2 or 10% of the total volume. This indicates that the intermyofibrillar mitochondrial volume fractions and the differences therein between WT and dCASQ-null also apply to the overall mitochondrial volume fractions in WT and dCASQ-null fibers.  and dCASQ-null-Ca (n = 15) fibers, respectively. ANOVA indicated a highly significant increase in free [Ca 2+ ] mito in dCASQnull fibers relative to WT fibers (P,0.001), whereas no significant differences were observed between the measurements in the presence and absence of external Ca 2+ . The average values of the pooled data amounted to 1.8360.02 in WT fibers (n = 96) and to 1.9460.03 in dCASQ-null fibers (n = 54).

Accumulation of mitochondrial Ca 2+ during contractions induced by electrical stimulation
Representative examples of the recordings obtained during contractions elicited by trains of electrical stimulation at 1, 5, 20 and 60 Hz in fibers from WT and dCASQ-null animals are shown in Fig. 3. An example of the increase in mitochondrial calcium concentration upon a single electrical pulse (a twitch response) is shown in Fig. S3 in File S1. The general patterns in the recordings in WT and dCASQ-null fibers are rather similar. Periodic increases in YFP/CFP ratio, and thus in free [Ca 2+ ] mito were observed during repetitive electrical stimulation at 1 Hz. The rising and decay phases of the mitochondrial Ca 2+ transients were highly asymmetric and this resulted in incomplete recovery of free [Ca 2+ ] mito during the contractions even at low stimulation frequency. During the first contractions, the transients were accompanied by a gradual increase relative to the baseline, reflecting an accumulation of Ca 2+ inside the mitochondrial matrix, reaching a dynamic steady state that was maintained during the remainder of the train of stimuli. At the stimulation rate of 5 Hz mitochondrial Ca 2+ accumulated more rapidly and [Ca 2+ ] mito reached a higher final level than in those at 1 Hz, while during the recordings at 20 and 60 Hz the rates and the levels attained were only slightly higher than those at 5 Hz. Whereas in WT fibers the accumulation of calcium in mitochondria continued during the stimulation period even when cytosolic calcium concentration was stable and constant, in dCASQ-null fibers, the initial fast rise in [Ca 2+ ] mito was followed by a slow increase or, in many instances, by a clear interruption of calcium accumulation, see for example the 60 Hz recording shown in Fig. S4 in File S1. This behavior may be linked to the decline in cytosolic free [Ca 2+ ] and in contraction strength observed previously (17,21) in CASQ1-null and dCASQ-null fibers and suggests that the decline in cytosolic free [Ca 2+ ] limited further increase in [Ca 2+ ] mito in dCASQ-null fibers.
The overall mitochondrial Ca 2+ accumulation during the train of stimuli was quantified by measuring the drop in the YFP/CFP ratio at the end of the train of stimuli (DR), after correction for bleaching of the probe as described in the Supporting Information (Fig. S2 in File S1). The averaged data obtained in WT and dCASQ-null fibers are shown in Fig. 4. The increases in free [Ca 2+ ] mito in dCASQ-null fibers were significantly smaller than in WT fibers (P,0.01) and somewhat blunted in the absence of extracellular Ca 2+ .

Kinetics of mitochondrial calcium accumulation
Information on the kinetics of the rising phase of the mitochondrial calcium transient was obtained in twitches or the first contractions during stimulation trains at 0.1 and 1 Hz (Fig. 5A). These contractions were all initiated in rested fibers. No significant differences were observed between groups and the average rise time from 10% to 90% of the increase in [Ca 2+ ] mito (t 10-90% ) for the pooled data amounted to 27.861.6 ms (n = 179).
The analysis of averaged responses at 1 Hz allowed us to estimate t 10-90% during repetitive 1 Hz stimulation, when the transient responses had reached a steady state. The average t 10-90% from all groups amounted to 18.560.4 ms (n = 94). This value is significantly smaller (233%) than the value obtained during the initial response in rested fibers.
There were no significant differences between WT and dCASQ-null but small, highly significant differences (P,0.01, ANOVA) were observed in the presence and absence of external Ca 2+ (Fig. 5B). Posthoc analysis indicated a significant prolongation of rise time in dCASQ-null fibers from 16.760.8 ms in the presence (n = 22) to 21.562.3 ms in the absence (n = 4) of external Ca 2+ (P,0.05).

Kinetics of mitochondrial [Ca 2+ ] decay
To obtain more information on the kinetics of the intial phase of the decay in mitochondrial [Ca 2+ ], we studied the averaged steady state recordings obtained during 1 Hz stimulation. At this stimulation rate, cytosolic calcium transients and contractile responses are well separated by periods at resting calcium concentration, while the free [Ca 2+ ] mito remained well above the resting initial values (see for example Fig S4A in File S1).
Averaging of the steady state responses as well as pooling the results obtained in different fibers from the same experimental group allowed us resolve changes in the sub-millisecond range. A typical example is shown in Fig. 6A. The decline in the YFP/CFP ratio and hence [Ca 2+ ] mito during the 1 Hz stimulation period was fitted to a double exponential as indicated in Methods. The decay in free [Ca 2+ ] mito appeared to be dominated by two exponentials with a very fast rate constant (k 1 ) of about 40 s 21 and a slower rate (k 2 ) of about 1.5-2 s 21 .
Recovery of the free [Ca 2+ ] mito after the end of a train of stimuli was dominated by a single exponential with a rate constant (k 3 ) of approximately 0.20-0.25 s 21 (see Fig. 6B). The recordings at 20 and 60 Hz revealed an initially fast recovery with relatively small amplitude which resembled the fast decay during the 1 Hz stimulation period. However, the main component of the decline after the train of stimuli did not depend on stimulation frequency as can be appreciated also comparing the panels of Fig. 3. Accordingly, the results obtained at different stimulation frequencies were pooled. Small (,10%) but significant (P,0.05) differences were observed in k 3 between WT and dCASQ-null, indicating that recovery of free [Ca 2+ ] mito at the end of a train of stimulation in dCASQ-null fibers was faster than in WT fibers (see Table 1).
To quantitate possible very slow components of the [Ca 2+ ] decay process we extended the recording of the decay phase after a train of 5 Hz stimulation to 10 minutes. In these recordings the illumination intensity was reduced accordingly in order to reduce bleaching of the probe. The time resolution of the recording amounted to 900 ms. The results obtained in 2 WT and 2 dCASQ-null fibers indicated the presence of an additional, very slow component. The relative amplitude (a 4 ) and the rate constant (k 4 ) of this component amounted to 1264% and 0.06060.005 s 21 and 962% and 0.01460.010 s 21 , in WT + Ca and dCASQ-null + Ca fibers, respectively. In these experiments, the effects of CGP37157 a specific inhibitor of the mitochondrial Na + /Ca 2+ exchanger were also tested. It can be seen in Fig. 7 that addition of 1 mM CGP37157 resulted in a more abrupt rise in [Ca 2+ ] mito and an increase in the final level reached, during the 10 s period of 5 Hz stimulation.
An overview of the data obtained with 1 Hz stimulation in WT and dCASQ-null fibers in the presence or absence of extracellular Ca 2+ is shown in Table 1. In general, the parameter values observed in dCASQ-null -Ca group showed more variability than the responses in the other groups, most likely because of larger variability in Ca 2+ content of intracellular stores in dCASQ-null fibers in the absence of extracellular Ca 2+ . It can be seen that the amplitude of the slow (k 3 <0.25 s 21 ) component (a 3 ) in WT + Ca fibers was larger than the amplitude of the fast (a 2 ) and the very fast component (a 1 ). Highly significant differences were observed in a 3 as well as in the rate constant of the fast component (k 2 ). The differences observed in a 3 were both type-(WT vs. dCASQ-null; P,0.001) as well as Ca 2+ -dependent (1 mM vs. 0 mM external Ca 2+ ; P,0.01). This latter observation is in line with the data reported in Fig. 4 and suggests the mitochondrial Ca 2+ concentration was reduced even in WT fibers in the absence of external Ca 2+ . The differences in k 2 were Ca 2+ -dependent (P,0.01) but not type-dependent. In WT-Ca fibers, the amplitude of a 3 was significantly smaller than in WT + Ca fibers (P,0.05). Post-hoc analysis revealed that the rate constant k 2 in the absence of extracellular Ca 2+ was significantly reduced (P,0.05). The typedependent difference in the rate constant k 3 which is significant when the values obtained at all stimulation rates are pooled together (see above), is visible as a trend but did not reach statistical significance in the results using 1 Hz stimulation only.
In summary, our results imply that overall mitochondrial Ca 2+ decay can be described by (at least) 4 different exponentials (very fast, fast, slow and very slow) with clearly distinct rate constants (k 1 , k 2 , k 3 and k 4 ) of approximately 40, 1.5, 0.25 and 0.03 s 21 . The amplitudes and rates of several of these components turned out to differ between dCASQ-null and WT and to depend on the presence or absence of external Ca 2+ .

Estimation of the free [Ca 2+ ] mito
The actual values of the free [Ca 2+ ] mito were derived from the equation: [Ca 2+ ] =b?K' d ?[(R-R min )/(R max -R)] 1/n [22], using average R max and R min values of 5.83 and 1.50, respectively, an in vitro K' d of the 4mtD3cpv sensor of 0.76 mM, an n value of 0.74 [22] and a value for b of 6 [30]. The values obtained in the presence of extracellular Ca 2+ in quiescent fibers amounted to 0.16 mM (WT) and 0.23 mM (dCASQ-null) and to a maximum free [Ca 2+ ] mito at the end of the train of stimuli at 20 or 60 Hz of 1.10 mM (WT) and 0.73 mM (dCASQ-null).

Cytosolic Ca 2+ transients during and after repetitive stimulation
To assess the relation between free cytosolic [Ca 2+ ] (or [Ca 2+ ] cyto ) and the free [Ca 2+ ] mito , we determined free [Ca 2+ ] cyto with the fluorescent indicator Fura-2. In quiescent fibers, no significant differences were observed in the free [Ca 2+ ] cyto between WT and dCASQ-null fibers both in the absence and presence of extracellular Ca 2+ . The free [Ca 2+ ] cyto of the pooled data amounted to 10162 nM (n = 90).
The recovery of the cytosolic Ca 2+ transient was virtually complete at 1 Hz stimulation frequency ( Fig. 8A and Fig. S4 in File S1), but at 5 and 60 Hz (Fig. 8B and Fig. S4 in File S1), an increase in the baseline was observed during stimulation. In WT fibers immersed in 1 mM Ca 2 , in agreement with the results of previous studies, e.g. [31], the peak values of the Ca 2+ transients  during repetitive stimulations increased with stimulation frequency (Fig. 8C). In the absence of extracellular Ca 2+ the increase in peak values during the repetitive stimulations was very similar (P = 0.47). In dCASQ-null fibers, even in the presence of external Ca 2+ , a marked decline followed the initial peak in Ca 2+ concentration reached during stimulation (Fig. 8B and Fig. S4 in File S1). In the absence of extracellular Ca 2+ , the amplitude of the Ca 2+ transients decayed rapidly with subsequent stimulation at 1 Hz and hardly any increase in [Ca 2+ ] cyto was observed during repetitive stimulations at 5, 20 and 60 Hz, pointing to a rapid depletion of the SR [17]. In those fibers the amplitude of the Fura-2 transient was quite variable and on average much smaller than in the presence of extracellular Ca 2+ .

Discussion
The use of a highly specific ratiometric probe indicates that electrical stimulation of a muscle fiber is followed by a mitochondrial Ca 2+ transient with a rise time of 20-30 ms and a decay phase consisting of (at least) 4 different components, with distinct rate constants. The comparison between WT fibers and dCASQ-null fibers revealed significant differences in mitochondrial calcium concentration at rest and during contraction and in the kinetics of both the rising and the falling phases of the mitochondrial Ca 2+ transient. These results suggest that the mitochondrial calcium handling is affected by the amount of calcium stored in the SR. In addition, our results indicate that mitochondrial Ca 2+ handling is sensitive to Ca 2+ entering into the fibers from the extracellular space. Our results also confirm and extend previous observations on alterations in mitochondrial volume and in the coupling of the CRU's to the mitochondria in dCASQ-null mice, a model of malignant hyperthermia, and suggest that these alterations may be adaptive.

Free mitochondrial [Ca 2+ ] in quiescent muscle fibers
In quiescent fibers free [Ca 2+ ] mito in dCASQ-null fibers is significantly higher than in WT fibers, as indicated by the difference in the basal YFP/CFP ratios. Our estimates of the free [Ca 2+ ] mito are 0.16 mM in WT and 0.23 mM in dCASQ-null fibers. These values are very similar to previous estimates of 0.2 mM in WT, see e.g. [32]. It should be noted, however, that our estimates of the free [Ca 2+ ] mito are rather crude because of uncertainties in the (in situ) values of K' d and b of the probe used in our experiments.
Evidence obtained in skeletal and cardiac muscle cells suggests a free/bound Ca 2+ proportion inside mitochondria of 1/100 [32,33], thus implying a total concentration of Ca 2+ stored inside the mitochondria at rest of 16 mM in WT fibers and 20 mM in dCASQ-null fibers. The intermyofibrillar mitochondrial volume fraction was increased from 4.360.1% in WT to 7.260.2% in   dCASQ-null fibers, i.e. by 6766%. The intermyofibrillar mitochondria represent a major fraction of the whole population which includes subsarcolemmal and perinuclear mitochondria [34]. Thus, our estimate of the overall mitochondrial volume fraction in WT fibers is less than half of the value of 9.9% obtained from fluorescence measurements in murine FDB fibers [35]. The reason for this difference is unclear but it can be noted that different techniques were used to visualize the mitochondria. Assuming an overall mitochondrial volume fraction of 4.3% in WT and 7.2% in dCASQ-null fibers, the mitochondrial Ca 2+ content expressed per fiber volume would correspond to 0.7 mmol/L in WT fibers and to 1.7 mmol/L in dCASQ-null fibers. This increase in mitochondrial Ca 2+ concentration at rest in dCASQ-null relative to WT could be related to the increase of Ca 2+ leakage from the SR lacking calsequestrin [36].

Free mitochondrial [Ca 2+ ] in contracting muscle fibers
Our results indicate that the mitochondrial Ca 2+ transient is highly asymmetric with a short rising phase (rise time: 20-30 ms) and a much longer decay phase which can be described by (at least) 4 different exponentials, with clearly distinct rate constants of 40, 1.5, 0.25 and 0.03 s 21 . These results add quantitative evidence to earlier findings that the free mitochondrial Ca 2+ concentration increases when SR Ca 2+ release is triggered by action potentials [8] and leads to accumulation of Ca 2+ inside mitochondria during sustained muscle activity [8,35].
Our estimations yield to maximal levels of free Ca 2+ concentration reached during repetitive stimulation of 1.10 and 0.73 mM, in WT and dCASQ-null fibers, respectively. Taking the difference in mitochondrial volume fraction into account (see above), the total amount of Ca 2+ inside the mitochondrial expressed per cell volume during contraction would correspond to 4.71 and 5.27 mmol/L, in WT and in dCASQ-null fibers, respectively. Assuming that the total calcium release ranges from 200-300 mmol/L after a single action potential to 400-500 mmol/L after repeated stimulation [37] mitochondrial Ca 2+ -uptake corre-  sponds to approximately 1% of the total amount of Ca 2+ released into the cytosol. A similar value has been calculated in cardiomyocytes [33], while determinations in skeletal muscle fibers range from very low values [35,38] to higher values as the recent estimate of 10-18%, derived under voltage clamp conditions in skeletal muscle fibers [7]. This implies that the contribution of mitochondrial Ca 2+ -uptake in fast muscle fibers would be rather small but not negligible, as was concluded previously [39]. However, the small fraction taken up by the mitochondrial could still be highly relevant for the regulation of key metabolic enzymes and the F0-F1 ATPase.
Kinetics of the rising phase of the mitochondrial Ca 2+ transient No differences were observed in the kinetics of the rising phase of the mitochondrial Ca 2+ transient between WT and dCASQnull. The average rise time of the increase in [Ca 2+ ] mito (t 10-90% ) amounted to 27.861.6 ms during the first contraction in a rested fiber and to 18.560.4 ms during steady state contractions at 1 Hz. Since the rise time is influenced by the intrinsic response time of the probe, the actual rise time of mitochondrial Ca 2+ -uptake might even be faster. The agreement between values obtained in WT and dCASQ-null fibers indicates that it is not significantly influenced by a reduction in the SR storage capacity, induced by ablation of both CASQ isoforms, possibly because the local cytosolic calcium concentrations during the initial contrations in WT and dCASQ-null are similar.
Since the cytosolic Ca 2+ transients during the initial contraction and the steady state contractions at 1 Hz are very similar (see Fig. S4 in File S1), the difference in the rise time in rested fibers and during steady state at 1 Hz is in fact opposite to what would be expected in a situation where substantial Ca 2+ -accumulation in the mitochondria would have taken place. However, the speed of mitochondrial Ca 2+ -uptake process could be increased as a result of a rise the local Ca 2+ -concentration in the microdomains between CRUs and mitochondria during repetitive stimulation at 1 Hz. In order to be consistent with Ca 2+ -uptake through the low affinity MCU, this local Ca 2+ -concentration would need to be increased considerably above the cytosolic free Ca 2+ -concentration at rest (,100 nM). Evidence suggests that the Ca 2+ uniporter can be sensitized by the locally elevated [Ca 2+ ], see e.g. [3] and estimates of local [Ca 2+ ] in cardiac muscle cells, up to 10 mM [40], are also consistent with this possibility.
A small but significant prolongation of the rising phase of mitochondrial Ca 2+ transient during 1 Hz stimulation was observed in dCASQ-null in the absence of external Ca 2+ . Under these conditions the intracellular stores are almost exhausted and unable to support a prolonged contractile activity [17]. It is likely therefore that the prolongation is caused by limited Ca 2+ availability in the cytosol rather than a direct specific effect of extracellular Ca 2+ on mitochondrial Ca 2+ -uptake kinetics per se.

Kinetics of the decay phase of the mitochondrial Ca 2+ transient
The decline in [Ca 2+ ] mito during and after stimulation trains is governed by 4 exponentials with rate constants of approximately 40, 1.5, 0.2 and 0.03 s 21 (at 26uC). The amplitude of the very fast component amounted to 25-30% of total Ca 2+ release both in WT as in dCASQ-null fibers ( Table 1). The occurrence of rapid uptake and release processes simultaneously may explain why previous attempts to determine mitochondrial Ca 2+ uptake, with lower time resolution, yielded contrasting results [41].
Selective blockade of the mNCX by CGP37157 (Fig. 6) enhanced the rise in free [Ca 2+ ] mito and increased the final level attained at the end of 10 s stimulation train. CGP37157 had no effect on the rise time of the increase in [Ca 2+ ] mito nor on the fast rate of the Ca 2+ -recovery process (results not shown). Assuming that the free [Ca 2+ ] mito is the result of a balance between Ca 2+ influx and efflux across mitochondrial inner membrane [42], these findings suggest that the mitochondrial Na + /Ca 2+ exchanger promotes Ca 2+ efflux already during the stimulation period. Since the effect was visible on a 10 s time scale we consider it likely that the mNCX may contribute to the intermediate (1.5 and 0.3 s 21 ) phases of the mitochondrial Ca 2+ -extrusion process.
Differences in mitochondrial structure and Ca 2+ handling between WT and dCASQ-null fibers Analysis of the EM-images indicated that in dCASQ-null fibers the mitochondrial volume was increased by 67%, the percentage of mitochondria coupled to the CRUs was increased by 12% and that the width of the terminal cisternae of the SR was reduced in comparison to WT. These findings and the increased [Ca 2+ ] mito at rest imply that the total amount of Ca 2+ stored in mitochondria is increased, as discussed above, and that the exchange of Ca 2+ between the SR and mitochondria could be facilitated by an improved coupling and a shortened Ca 2+ diffusion distance between the sites of release and the mitochondria [10]. Assuming that the increased Ca 2+ leakage from the SR results in an elevation of the Ca 2+ concentration near the mitochondria relative to the concentration in the bulk of the cytosol, the improved coupling with SR and the increased Ca 2+ leakage from SR could explain the increase in free [Ca 2+ ] mito in quiescent dCASQ-null fibers relative to WT fibers, whereas global free cytosolic Ca 2+ concentration remained the same. Moreover, it can be noted that the intermyofibrillar mitochondria are located close to the regions of the SR where SERCA is localized (see Fig. 2). Thus, leakage through SERCA in the absence of calsequestrin could play a specific role as previously suggested [36]. The increased calcium concentration in the mitochondrial matrix might play a role in increasing the metabolic activity. The greater activity of the mitochondrial metabolic machinery at rest might, in turn, lead to greater production of ATP and reactive oxygen species (ROS).
The steady levels of the free mitochondrial Ca 2+ concentration attained during the repetitive contractions were lower in dCASQnull than in WT. This reduction can be attributed to the depletion of SR and to the progressive decay of cytosolic Ca 2+ concentration during the contractions as a result of Ca 2+ -binding to parvalbumin (see Fig. 7 and Fig. S4 in File S1) and [17,19]. Interestingly, this reduction in Ca 2+ -accumulation is predominantly reflected during the decline phase in a reduction of the amplitude of the slow recovery process (a 3 ). Thus, the amplitude of this latter phase appears to be particularly sensitive to the reduced mitochondrial Ca 2+ -content of dCASQ-null fibers in comparison with WT fibers. This observation is compatible with the idea that the speed of this phase is determined at least partly by the Ca 2+ -off rate(s) of mitochondrial Ca 2+ buffer(s).

Mitochondrial Ca 2+ -handling in the absence of extracellular Ca 2+
Our previous study [17] showed a rapid loss of contractile response of dCASQ-null fibers after incubating them in a solution without Ca 2+ , indicating that sarcolemmal Ca 2+ -influx possibly via SOCE occurred during stimulation and was required to support the activation of the contractile process in those fibers. Our present results indicated that the elevated free mitochondrial [Ca 2+ ] in quiescent dCASQ-null fibers was maintained in the absence of external Ca 2+ . We attribute this to enhanced coupling between the SR and the mitochondria because the cytosolic Ca 2+ concentration at rest: 1) did not differ between WT and dCASQ-null fibers and 2) was not significantly affected by external Ca 2+ removal.
Interestingly, the rate constant (k 2 ) of the fast Ca 2+ extrusion phase is reduced in dCASQ-null fibers when extracellular Ca 2+ is removed, back to WT values. This suggests that in addition to improved coupling between CRUs and mitochondria in dCASQnull fibers, the activity of the mNCX could be reduced in these fibers in the absence of extracellular Ca 2+ , e.g. by a rise in mitochondrial Na + concentration [43].

Conclusions and implications
This study represents the first detailed quantitative analysis of the kinetics of mitochondrial Ca 2+ handling in fast skeletal muscle fibers. It extends previous data on mitochondrial Ca 2+ handling in skeletal muscle fibers [8,10,35,44]. In addition, our results provide evidence of the dependence of [Ca 2+ ] mito on intracellular and external Ca 2+ stores and reveal the presence of multiple processes of free mitochondrial Ca 2+ buffering or extrusion operating at a time scale ranging from milliseconds to minutes. Our analysis also provides direct evidence for the involvement of the mitochondrial Na + /Ca 2+ exchanger (mNCX) in mitochondrial Ca 2+ extrusion process, relevant at the intermediate time scale (seconds or tens of seconds).
Rapid Ca 2+ uptake and release by the mitochondria play an important but still under-illuminated role in the regulation mitochondrial metabolic activity. Our study indicates that mitochondrial Ca 2+ -uptake and release are fast enough to allow the rapid adjustment of metabolic activity to increased demand for instance during exercise. Moreover, Ca 2+ uptake by the mitochondria is crucial in skeletal and cardiac dysfunction and in programmed cell death, e.g. [3,4].
Calsequestrin-null mice represent a model for malignant hyperthermia [21,45]. The number of mitochondria and the resting free Ca 2+ concentration inside the mitochondria were increased in calsequestrin-null muscle and the maximum free Ca 2+ levels reached during electrical stimulation were sensitive to the presence of extracellular Ca 2+ . These alterations in mitochondrial Ca 2+ handling may represent a compensatory response when SR Ca 2+ -content is reduced, e.g. by calsequestrin ablation and/or increased SR Ca 2+ leakage, but also may render the tissue more prone to the development of contractures.

Supporting Information
File S1 Supporting text and figures. Figure S1. In situ determination of R min (left) and R max (right) of the cameleon (4mtD3cpv). The 4mtD3cpv responses (YFP, red; CFP, blue; ratio YFP/CFP, green) are shown. Aliquots of ionomycin (Iono) were added up to a final concentration of 5 mM (left) and 10 mM (right); 50 mM N-benzyl-ptoluene sulphonamide (BTS) was added in order to reduce movement artifacts during the R max determination. The minimum YFP/CFP value reached (R min ) amounted to 1.5. Addition of 5 mM CaCl 2 caused an abrupt increase in the YFP/CFP ratio up to a maximum value of 6.3. Upon the addition of 5 mM CaCl 2 , the fiber started to contract despite of the presence of BTS (for 7 minutes), resulting in an increase in the CFP (and YFP) intensity and moved out of the focal plane of the microscope near the end of the recording (resulting in a decline in the YFP and CFP intensity). Figure S2. Correction for bleaching. Top: during a train of stimuli at 0.1 Hz, the amplitude of the increase in the YFP/CFP ratio declined as a result of bleaching of the probe. Bottom: a linear relation was observed between peak amplitude and the concurrent baseline value, yielding a straightforward way to correct the DR values for bleaching of the probe (see Results). Figure S3.
Recording of the change in [Ca 2+ ] mito during and after a single twitch. The baseline corrected 4mtD3cpv response (YFP, red; CFP, blue; ratio YFP/CFP, green) in a WT fiber electrically stimulated by a single pulse in the presence of 1 mM Ca 2+ . The YFP/CFP ratio shown was obtained by using a 10-points running average. The final part of the decay phase could be well fitted to a single exponential (bright green) with a rate constant of 0.21 s 21 . Figure S4. Comparison of the time course of the 4mtD3cpv responses and the Fura-2 responses. In each panel, the upper figure shows the 4mtD3cpv response and the lower figure shows the Fura-2 response in WT at 1, 5 and 60 Hz stimulation (A, B and C) and in dCASQ-null at 60 Hz (D). Note that the decline in the 4mtD3cpv transients at 1 and 5 Hz occurs more slowly than in the Fura-2 transients. As a result there isduring the stimulation period-a gradual rise in the baseline free Ca 2+ concentration inside the mitochondria, whereas the cytosolic Ca 2+ concentration in between stimuli remains rather constant (Panels A and B). The comparison between WT and dCASQ-null at 60 Hz illustrates that in dCASQ-null fibers the initial peak value is similar to WT but there is a gradual decline in the cytosolic free Ca 2+ concentration. At 60 Hz stimulation, the free [Ca 2+ ] mito after an early rapid increase accumulates gradually in WT but remains rather constant in dCASQ-null fibers (Panels C and D). (DOC)