Cyclosporin A Inhibits Inositol 1,4,5-Trisphosphate-dependent Ca2+ Signals by Enhancing Ca2+ Uptake into the Endoplasmic Reticulum and Mitochondria*

Cytosolic Ca2+([Ca2+] i ) oscillations may be generated by the inositol 1,4,5-trisphosphate receptor (IP3R) driven through cycles of activation/inactivation by local Ca2+feedback. Consequently, modulation of the local Ca2+gradients influences IP3R excitability as well as the duration and amplitude of the [Ca2+] i oscillations. In the present work, we demonstrate that the immunosuppressant cyclosporin A (CSA) reduces the frequency of IP3-dependent [Ca2+] i oscillations in intact hepatocytes, apparently by altering the local Ca2+ gradients. Permeabilized cell experiments demonstrated that CSA lowers the apparent IP3 sensitivity for Ca2+ release from intracellular stores. These effects on IP3-dependent [Ca2+] i signals could not be attributed to changes in calcineurin activity, altered ryanodine receptor function, or impaired Ca2+fluxes across the plasma membrane. However, CSA enhanced the removal of cytosolic Ca2+ by sarco-endoplasmic reticulum Ca2+-ATPase (SERCA), lowering basal and inter-spike [Ca2+] i . In addition, CSA stimulated a stable rise in the mitochondrial membrane potential (ΔΨm), presumably by inhibiting the mitochondrial permeability transition pore, and this was associated with increased Ca2+ uptake and retention by the mitochondria during a rise in [Ca2+] i . We suggest that CSA suppresses local Ca2+ feedback by enhancing mitochondrial and endoplasmic reticulum Ca2+ uptake, these actions of CSA underlie the lower IP3 sensitivity found in permeabilized cells and the impaired IP3-dependent [Ca2+] i signals in intact cells. Thus, CSA binding proteins (cyclophilins) appear to fine tune agonist-induced [Ca2+] i signals, which, in turn, may adjust the output of downstream Ca2+-sensitive pathways.

Immunosuppressants exert their activity by binding to immunophilins, an evolutionary conserved, but structurally heterogeneous family of proteins that shares a common enzymatic activity and pharmacological profile (1-3). All immunophilins described to date possess cis-trans-peptidylprolyl isomerase or rotamase activity, which has been implicated in the folding, assembly, and trafficking of target proteins in vivo (1,2). The cis-trans-peptidylprolyl isomerase activity is inhibited by low concentrations of immunosuppressants. Based upon binding criteria, immunophilins are divided into two classes: (a) the cyclophilin family that selectively binds cyclosporin A (CSA) 1 and (b) the FK-506 binding proteins (FKBP), which bind to FK-506, its analogues, and rapamycin. Although the precise functions of immunophilins and their downstream targets remain to be determined, they have frequently been implicated in regulating a variety of Ca 2ϩ -dependent pathways. The best characterized therapeutic action of CSA or FK-506 is the suppression of interleukin-2 gene transcription during antigeninduced T-lymphocyte activation. Specifically, CSA⅐cyclophilin or FK-506⅐FKBP complexes are targeted to and inhibit the catalytic activity of the Ca 2ϩ /calmodulin-dependent protein phosphatase, calcineurin (4), thereby blocking the translocation of nuclear factor of activated T cells (3,5).
In isolated mitochondria, CSA binds to cyclophilin D (CyP-D), which is believed to be a component of the mitochondrial permeability transition pore (PTP). Displacing CyP-D from its binding site favors the closed state of the pore (6 -8). Recent evidence suggests that the PTP may be involved in Ca 2ϩ signaling (9,10), as well as necrotic and apoptotic cell death (11)(12)(13)(14)(15). Operating in a low conductance mode, the PTP is also proposed to furnish mitochondria with a fast Ca 2ϩ -efflux mechanism (i.e. mitochondrial calcium-induced calcium-release) that, in turn, amplifies inositol 1,4,5-trisphosphate (IP 3 )-dependent cytosolic calcium ([Ca 2ϩ ] i ) signals (9,10).
Immunophilins may also modulate the Ca 2ϩ release channels of the internal stores directly. FKBP12 forms tight complexes with both ryanodine receptors (RyRs) and IP 3 receptors (IP 3 Rs) that are perturbed by FK-506 or rapamycin (16 -19). Dissociation of the FKBP from the channels results in increased Ca 2ϩ fluxes in response to caffeine or IP 3 (16,18). This enhanced Ca 2ϩ release can either be explained by destabiliza-tion of the channel following FKBP12 dissociation (3,16) or a change in the phosphorylation state, because FKBP12 also anchors calcineurin to the channel (17). Moreover, two ubiquitously expressed members of the cyclophilin family, s-cyclophilin and cyclophilin A, colocalize with or bind to the Ca 2ϩ storage protein of the ER, calreticulin (20,21). Therefore, this raises the possibility that cyclophilins may modulate Ca 2ϩ storage properties of the ER and, indirectly, Ca 2ϩ flux through the RyR and/or IP 3 R. Taken together, there is a wealth of evidence that immunophilins can potentially activate or inhibit Ca 2ϩ -dependent signal transduction by modifying the activity of diverse cellular targets.
Periodic oscillations or spikes in [Ca 2ϩ ] i are utilized by a wide range of extracellular stimuli (22)(23)(24) and are decoded by a host of downstream sensors (25)(26)(27)(28). In non-excitable tissues, extracellular agonists mediate increases in [Ca 2ϩ ] i through the formation of the second messenger IP 3 and activation of intracellular Ca 2ϩ -release channels (22)(23)(24)29), which may in turn stimulate the influx of external Ca 2ϩ to sustain the agonist signal and refill the internal stores (30). In hepatocytes, the agonist dose, which presumably determines the intracellular [IP 3 ], sets the frequency of [Ca 2ϩ ] i spikes (i.e. frequency modulation). However, the positive feedback effect of [Ca 2ϩ ] i on IP 3 R generates the rapid-rising phase of the [Ca 2ϩ ] i spike, which is essentially independent of the agonist dose. This Ca 2ϩ -positive feedback is also thought to underlie the propagation of regenerative intracellular Ca 2ϩ waves (24). We have previously suggested that the minimal prerequisites for generating oscillatory [Ca 2ϩ ] i signals are the concordant actions of Ca 2ϩ and IP 3 on IP 3 R function (31). Because Ca 2ϩ is involved in both the activation and inactivation of the IP 3 R (22)(23)(24), other Ca 2ϩ transport mechanisms could exert profound effects on the dynamics and frequency of [Ca 2ϩ ] i signals by altering local Ca 2ϩ gradients. Indeed, mitochondrial Ca 2ϩ uptake can modulate IP 3 sensitivity by suppressing the local, positive feedback effects of Ca 2ϩ on the IP 3 R in hepatocytes (32).
In the present work, we have investigated the effects of CSA on IP 3 -dependent [Ca 2ϩ ] i oscillations in hepatocytes. We demonstrate that CSA or its non-immunosuppressive analogue, N-methylvaline-cyclosporin (MeVal-CS), inhibits the frequency of IP 3 -dependent [Ca 2ϩ ] i oscillations by simultaneously activating ER and mitochondrial Ca 2ϩ uptake. We provide evidence that this activation suppresses local, positive Ca 2ϩ feedback on the IP 3 R and lowers IP 3 R sensitivity. Thus, cyclophilins may play an essential role in determining the shape and frequency of IP 3 -dependent [Ca 2ϩ ] i signals and, ultimately, the activity of downstream Ca 2ϩ -sensitive targets by regulating cellular Ca 2ϩ transport mechanisms.

Materials-Cyclosporin
A and MeVal-CS were generous gifts from Novartis (Basil, Switzerland) and FK-506 from Fujisawa, Inc. Stock solutions were prepared in Me 2 SO and diluted at least a 1000-fold in all experiments.
Cell Isolation and Culture-Hepatocytes were isolated by a two-step collagenase perfusion of livers from male Sprague-Dawley rats fed ad libitum as described previously (33). Isolated hepatocytes were stored on ice until assayed or maintained in primary culture for 1.5-3 h in Williams' E medium supplemented with 10% (v/v) fetal calf serum (complete WEM) as described previously (28). Hepatocytes used in confocal imaging experiments were cultured overnight in complete WEM containing insulin (14 nM). Prior to use, cells were incubated 30 -40 min in a KR-HEPES buffer composed of (in millimolar): 121 NaCl, 25 Na-HEPES, 5 NaHCO 3 , 4.7 KCl, 1.2 KH 2 PO 4 , 1.2 MgSO 4 , 2 CaCl 2 , 10 glucose, 0.1 sulfobromophthalein, and 0.25% (w/v) essentially fatty acid-free bovine serum albumin (fraction V, Sigma Chemical Co.), pH 7.4, at 37°C. Where indicated in the figure legends, L-glutamate (5 mM) and pyruvate (1 mM) were also present during the incubation period.
Single Cell [Ca 2ϩ ] i Measurements-Ca 2ϩ imaging experiments were preformed essentially as described previously (33,34). Briefly, hepatocytes attached to collagen (5 g/cm 3 )-coated glass coverslips were loaded with fura-2 by incubation with 5 M fura-2/AM for 15-30 min in a KR-HEPES buffer supplemented with 0.2% (w/v) Pluronic acid F-127. Fura-2-loaded hepatocytes were washed twice with KR-HEPES without dye or Pluronic acid and then transferred to a thermostatically regulated microscope chamber (37°C). Fura-2 fluorescence images (excitation 340 and 380 nm, emission 420 -600 nm) were acquired at 1-to 3-s intervals with a cooled charged-coupled device camera under computer control (33,34). Calibration of fura-2 in terms of [Ca 2ϩ ] i was calculated from the 340/380 nm ratio after correcting for autofluorescence (35). Autofluorescence values were obtained at the end of each experiment by permeabilizing the cells with digitonin in an intracellular-like media containing a Ca 2ϩ /EGTA buffer and Mg⅐ATP (35). The fura-2 calibration parameters were determined in vitro using a K d value of 224 nM.
Single Cell Measurements of ⌬⌿-Primary cultured hepatocytes were incubated with tetramethylrhodamine ethyl ester (TMREE, 5 nM) for 1-2 h in complete WEM then washed into a KR-HEPES buffer containing 5 nM TMREE. TMREE fluorescence images were collected using 548-nm excitation and 600-nm-long bandpass emission filter. The logarithm of TMREE fluorescence changes was normalized to the fluorescence intensity values obtained after collapsing the mitochondrial electrochemical gradient with 5 M FCCP plus 5 g/ml oligomycin.
Laser Scanning Confocal Microscopy-Simultaneous measurement of Mito Tracker Green and rhod-2 fluorescence were performed as described previously (33,36). Briefly, overnight-cultured hepatocytes were incubated with 100 nM Mito Tracker Green for 1-2 h in complete WEM supplemented with 0.05% (w/v) Pluronic acid F-127. Cells were washed extensively and then loaded with rhod-2/AM (10 M) in KR-HEPES buffer containing Pluronic acid F-127 (0.1% w/v) for 10 min at room temperature. Confocal images were acquired using a Bio-Rad MRC 600 laser scanning confocal microscope.
Mitochondrial Ca 2ϩ Uptake in Permeabilized Hepatocyte-Hepatocytes (5 mg/ml cell protein) were washed once in a Ca 2ϩ -free intracellular-like buffer (IB) composed of (in millimolar): 135 KCl, 15 NaCl, 1.2 MgSO 4 , 1.2 KH 2 PO 4 , 10 HEPES, 10 MES, 5 glutamate, 1 pyruvate, 10 units/ml aprotinin, and 1 g/ml each of leupeptin, antipain, and pepstatin A (pH 7.2 at 37°C), and then stored on ice. Prior to use, cells were briefly centrifuged, suspended in prewarmed IB, and then incubated for 5 min in the presence of indicated drugs or vehicle. Incubations were carried out in a thermostatically regulated cuvette (37°C) with continuous stirring. The cells were permeabilized with digitonin (25 g/ml), and the medium was further supplemented with 5 mM phosphocreatine, 5 units/ml creatine phosphokinase, 2 mM ATP⅐K, 2 M thapsigargin, and 10 M BTC-free acid (Molecular Probes). BTC fluorescence (excitation 405 and 470 nm, emission 540) was monitored using a dualwavelength excitation spectrofluorimeter. Calibration of BTC fluorescence ratio values was achieved by determining the minimum and maximum 405/470 ratios with EGTA/Tris and excess CaCl 2 , respectively, at the end of each run (35). Medium-free Ca 2ϩ concentration ([Ca 2ϩ ] ex ) was calculated assuming a K d of 7 M for BTC.
Mitochondrial Ca 2ϩ Content in Intact Cells-Hepatocytes (5 mg/ml cell protein) were suspended in oxygenated, Ca 2ϩ -containing KR-HEPES buffer and incubated for 15 min in a shaking water bath (37°C). Where indicated in the figure legend, vasopressin or other drugs were included during the preincubation period. Following the preincubation, cell suspensions were rapidly centrifuged (60 s) and then suspended in a stirred cuvette containing Ca 2ϩ -free KR-HEPES buffer plus 200 M BAPTA-free acid and 10 M fura-2-free acid. Rotenone (5 M), antimycin A (5 M), and oligomycin (5 g/ml) were added to block mitochondrial respiration (not shown) followed by ionomycin (20 M) to release compartmentalized Ca 2ϩ . In some experiments, FCCP (5 M) plus oligomycin (5 g/ml) was included during the preincubation period to prevent mitochondrial Ca 2ϩ accumulation. Fura-2 fluorescence was monitored by alternating excitation at 340 and 380 nm, and the emitted fluorescence was collected at 510 nm. The [Ca 2ϩ ] ex was calculated from the fura-2 ratio values using a K d of 224 nM. The total ionomycinreleasable Ca 2ϩ pool (buffered and free) was assessed to be predominately mitochondrial and calibrated by small additions of standardized CaCl 2 to the cuvette. Mitochondrial Ca 2ϩ content was calculated assuming hepatocytes contain 0.24 mg of mitochondrial protein/mg of cell protein (37). Matrix Ca 2ϩ determinations were performed in triplicate from two or three different cell preparations.
IP 3 -induced Ca 2ϩ Fluxes and Ca 2ϩ Content of Internal Stores-Measurement of IP 3 -induced Ca 2ϩ release in permeabilized hepatocyte suspensions was performed essentially as described previously (38). Hepatocytes (5 mg/ml cell protein) were washed extensively in a Ca 2ϩ -free Chelex-treated intracellular-like media (ICM) composed of (in millimo-lar): 130 KCl, 10 NaCl, 1.0 KH 2 PO 4 , 20 HEPES, 20 Tris, 5 glutamate, 1 pyruvate, 10 units/ml aprotinin, and 1 g/ml each of leupeptin, antipain, and pepstatin A (pH 7.2 at 37°C), then stored on ice. Prior to the experiment, cells were washed into prewarmed ICM supplemented with phosphocreatine (5 mM) and fura-2-free acid (5 M) then incubated for 5 min in a stirred thermostatted cuvette. When indicated in the figure legends, CSA (5 M) or Me 2 SO vehicle (0.1%, v/v) was included during the preincubation period. Cells were permeabilized with digitonin (40 g/ml) in the presence of 5 units/ml creatine phosphokinase for 2 min followed by addition of 2 mM Mg⅐ATP to initiate Ca 2ϩ uptake. In some experiments, mitochondrial Ca 2ϩ uptake was inhibited by adding rotenone (5 M) plus oligomycin (5 g/ml) prior to permeabilization. fura-2 fluorescence changes were monitored as described in the previous section. IP 3 -induced Ca 2ϩ release was normalized to the size of the ionomycin-releasable Ca 2ϩ pool. In parallel experiments, the Ca 2ϩ content of the ionomycin-sensitive Ca 2ϩ store was determined. After cell permeabilization, rotenone, oligomycin, FCCP (0.5 M), and Mg⅐ATP were added and the cell suspension incubated for 15 min. FCCP was included to help facilitate the release of stored mitochondrial matrix Ca 2ϩ . The size of the ionomycin-releasable Ca 2ϩ store was calibrated by small additions of standardized CaCl 2 to the cuvette. Under these conditions, the size of the thapsigargin-sensitive Ca 2ϩ pool was 90 -95% of the ionomycin-sensitive Ca 2ϩ pool.
Protein Determination-Protein content was determined by alkaline biuret assay using the methods described by the manufacture (Sigma).
Statistical Analysis-Unless otherwise stated in the text, all experiments were repeated with at least three separate hepatocyte preparations. Data are presented as means Ϯ S.E. for the number of separate hepatocyte preparations shown in parentheses. Significance of differences from the relevant controls was calculated by Student's t test.

CSA Inhibits [Ca 2ϩ ] i Oscillations in Single Hepatocytes-
Stimulation of rat hepatocytes with submaximal concentrations of IP 3 -forming agonists induces a series of baseline-separated [Ca 2ϩ ] i oscillations or spikes. The frequency of the [Ca 2ϩ ] i oscillations is dependent upon the agonist concentration and remains constant for a given dose over the time course of the experiment (34). The traces in Fig. 1, A-C, and Fig Note that CSA resulted in an immediate, but small, decrease in inter-spike [Ca 2ϩ ] i (Figs. 1 and 2) that persisted throughout the time course of the experiment (20 -30 min). This effect also occurred in the absence of agonist (see Fig. 4A). In naive cells, CSA decreased basal [Ca 2ϩ ] i by 30 Ϯ 1.0 nM (mean Ϯ S.E.; n ϭ 250 cells, five cell preparations). These data suggest that CSA may promote Ca 2ϩ clearance from the cytosol, which could be one of the underlying mechanisms suppressing the frequency of [Ca 2ϩ ] i spikes. The effect of CSA on basal [Ca 2ϩ ] i was investigated in length and discussed in more detail below.
Another possible mechanism of action for the CSA⅐cyclophilin complex is the inhibition of calcineurin (3,5), which is known to increase PKC-dependent phosphorylation of IP 3 Rs in isolated cerebellar microsomes (17). The CSA derivative, MeVal-CS, binds cyclophilins with a similar efficacy to CSA (39), but the complex cannot inhibit calcineurin activity (40 -42). The addition of MeVal-CS (5 M) also inhibited the frequency of PEinduced [Ca 2ϩ ] i oscillations (Fig. 1C) in a concentration-dependent manner (data not shown) with a potency similar to CSA. Furthermore, a 10-to 15-min preincubation with a PKC inhibitor, bisindolylmaleimide-1 (0.1-1.0 M), did not block the inhibitory effects of CSA on PE-induced [Ca 2ϩ ] i responses (data not shown). Taken together, these data suggest that CSAinduced inhibition of IP 3 -dependent Ca 2ϩ signals is not mediated by either PKC or calcineurin.
In some cell types, IP 3 -induced Ca 2ϩ release has been proposed to trigger calcium-induced calcium-release via RyRs. Specific [ 3 H] ryanodine binding to isolated hepatic microsomes

FIG. 1. Cyclosporin A and its derivative, N-methylvaline-cyclosporin (MeVal-CS) inhibit IP 3 -dependent [Ca 2؉ ] i oscillations in hepatocytes.
In A-C, fura-2/AM-loaded primary cultured hepatocytes were stimulated with submaximal phenylephrine (PE) concentrations (0. has been reported, and RyR inhibitors appear to inhibit IP 3dependent Ca 2ϩ signaling in isolated intact hepatocytes (43). To rule out a direct CSA effect on RyRs, hepatocytes were preincubated for 10 -15 min with 100 M ryanodine prior to PE stimulation. In these experiments, CSA still inhibited IP 3 -dependent [Ca 2ϩ ] i responses suggesting that RyRs are not involved in the drug's action (data not shown).
CSA Alters the Kinetics of IP 3 -dependent [Ca 2ϩ ] i Spikes-Despite the lower basal [Ca 2ϩ ] i , peak amplitude and time to peak were the same before and after CSA addition (Fig. 2, A and C), although there was a small, but significant, increase in the rate of Ca 2ϩ rise (p Ͻ 0.001; Fig. 2D). The most obvious effect of CSA on the kinetics of the individual [Ca 2ϩ ] i spikes was to slow the rate of decline (p Ͻ 0.001), resulting in a prolongation of the Ca 2ϩ transient (Fig. 2, B and E). To investigate the mechanism for these effects on Ca 2ϩ removal from the cytosol, we examined the effects of CSA on Ca 2ϩ transport across the plasma membrane, as well as Ca 2ϩ uptake into internal stores.
CSA Does Not Affect Ca 2ϩ Transport Across the Plasma Membrane-Maximal VP stimulation rapidly releases Ca 2ϩ from IP 3 -sensitive stores and in the presence of extracellular Ca 2ϩ results in a sustained elevation in [Ca 2ϩ ] i (Fig. 3A). This sustained phase of [Ca 2ϩ ] i increase is dependent upon Ca 2ϩ influx and is set by the balance between Ca 2ϩ influx and efflux across the plasma membrane (cf. Fig. 3D). In Fig. 3A, BAPTA was added during the sustained rise in [Ca 2ϩ ] i to abruptly inhibit Ca 2ϩ influx. The initial rate of Ca 2ϩ clearance from the cytosol, which mostly reflects Ca 2ϩ efflux mechanisms, was unaltered by CSA pretreatment (Fig. 3A, gray trace). An alternative method to assess efflux rates is to measure the appearance of Ca 2ϩ in the external buffer after mobilizing internal stores. In cell suspension experiments, CSA pretreatment did not effect the rate of Ca 2ϩ efflux elicited by the addition of maximal concentrations of either VP or thapsigargin (Tg) (not shown).
In Fig. 3B (PMF), thus inhibiting mitochondrial Ca 2ϩ uptake. Oligomycin (Oligo) was included in these experiments to limit FCCP-stimulated ATP hydrolysis (33). Collapsing mitochondrial PMF re-leased matrix Ca 2ϩ stores and transiently raised [Ca 2ϩ ] i (Fig.  4C). Subsequent CSA addition still stimulated a robust decrease in [Ca 2ϩ ] i (Fig. 4C), indicating that CSA does not stim- ulate Ca 2ϩ uptake into the mitochondria under basal conditions. This is not surprising given the fact that the mitochondrial uniporter has a low affinity for Ca 2ϩ (i.e. K 0.5 ϭ 5-10 M; see Refs. 44,45). By contrast, inhibiting SERCA activity with Tg completely blocked the effect of CSA on basal [Ca 2ϩ ] i (Fig. 4D). Furthermore, when MeVal-CS was substituted for CSA, qualitatively similar results were obtained for each experimental protocol shown in Fig. 4, indicating that these effects were independent of calcineurin (not shown). These data suggest that the decrease in basal [Ca 2ϩ ] i is mediated by Ca 2ϩ uptake into internal stores via SERCA.
Permeabilized hepatocytes suspensions were used to further investigate the effects of CSA on Ca 2ϩ uptake into internal stores in the presence of inhibitors to block mitochondrial Ca 2ϩ transport. Fig. 5A shows that, in the presence of endogenous ATP, CSA-treated hepatocytes consistently buffered mediumfree Ca 2ϩ concentrations ([Ca 2ϩ ] ex ) to lower values when compared with Me 2 SO-treated cells. Addition of exogenous ATP and Ca 2ϩ (arrow) resulted in a prompt increase of [Ca 2ϩ ] ex , followed by Ca 2ϩ uptake into Tg-sensitive stores. The CSAtreated cells consistently showed an enhanced rate of Ca 2ϩ uptake and buffered the medium to lower Ca 2ϩ levels (Fig. 5A). Under these conditions, the [Ca 2ϩ ] ex was 175 Ϯ 14 nM and 133 Ϯ 12 nM in Me 2 SO-pretreated and CSA-pretreated hepatocyte suspensions, respectively, after ATP-dependent Ca 2ϩ uptake (mean Ϯ S.E.; n ϭ 3). Once steady-state Ca 2ϩ uptake reached completion, ionomycin was used to check the final store size. Despite enhancing the rate of Ca 2ϩ uptake, the ionomycin-sensitive Ca 2ϩ pool was not significantly increased in the presence of CSA. The final steady-state Ca 2ϩ pool was 2.75 Ϯ 0.1 and 2.88 Ϯ 0.1 nmol/mg of protein in the absence and presence of CSA, respectively (p Ͼ 0.1; n ϭ 3).
In parallel experiments, Tg was added to assess the passive Ca 2ϩ leak rate from internal stores (Fig. 5B). CSA pretreatment did not affect the rate of Ca 2ϩ leak (gray trace); [Ca 2ϩ ] ex increased 5.8 Ϯ 1.1 nM/s and 4.4 Ϯ 0.6 nM/s in Me 2 SO-and CSA-treated cells, respectively (p Ͼ 0.1; mean Ϯ S.E., n ϭ 3). Under both conditions, the size of the Tg-sensitive Ca 2ϩ pool was 90 -95% of the ionomycin-releasable pool, suggesting that CSA does not significantly stimulate Ca 2ϩ uptake into Tginsensitive stores. Taken together, these data suggest that CSA simulates Tg-sensitive Ca 2ϩ pumps to enhance the rate of Ca 2ϩ clearance from the cytosol and lower the final set point for basal [Ca 2ϩ ] i . Furthermore, these effects are independent of calcineurin and point to a role for an ER-associated cyclophilin that regulates SERCA activity.
Effect of CSA on IP 3  The most striking effect of CSA treatment is the decrease in the size of the IP 3 -sensitive Ca 2ϩ pool. The blunted Ca 2ϩ fluxes cannot be explained by differences in pool size, because the ionomycin-releasable Ca 2ϩ pool is not significantly increased in CSA-treated cells (see text). Moreover, CSA did not modify the initial rate of Ca 2ϩ release at maximal IP 3 concentrations, the fura-2 ratio increased 0.94 Ϯ 0.05 unit/s in Me 2 SO-treated cells verses 0.89 Ϯ 0.05 unit/s in CSA-treated cells (p Ͼ 0.2; n ϭ 3). The lack of effect on initial rates of IP 3 -induced Ca 2ϩ release, coupled with the increased rate of [Ca 2ϩ ] i rise in the intact cells (Fig. 2D), suggest that CSA did not directly inhibit Ca 2ϩ flux through the IP 3 R.
CSA has been reported to increase the size of the mitochondrial Ca 2ϩ pool in intact cells (47,48). An enhanced rate of mitochondrial Ca 2ϩ uptake could buffer IP 3 -induced [Ca 2ϩ ] ex responses. To test this possibility, hepatocyte suspensions were treated with mitochondrial inhibitors prior to digitonin permeabilization. Blocking mitochondrial Ca 2ϩ uptake resulted in a less pronounced rightward shift in the IP 3 concentration response curve but did not eliminate the effects of CSA on the IP 3 -sensitive Ca 2ϩ store size (Fig. 6B). The EC 50 values were 128 and 148 nM in the absence and presence of CSA, respectively. These data suggest that the mitochondria are not in- FIG. 5. Effect of cyclosporin A on ATP-dependent Ca 2؉ uptake into internal stores. Intact hepatocytes suspensions were pretreated 5 min with CSA (5 M, gray traces) or Me 2 SO (black traces) prior to permeabilization. Cells were digitonin-permeabilized in a Chelex-treated ICM containing mitochondrial inhibitors and fura-2-free acid (see "Experimental Procedures"). The addition of exogenous ATP plus Ca 2ϩ rapidly elevates medium Ca 2ϩ levels followed by ATP-dependent Ca 2ϩ uptake into internal stores (A). In parallel experiments, thapsigargin (Tg; 2 M) was added after ATP-dependent Ca 2ϩ uptake to assess the passive Ca 2ϩ leak rate from internal stores (B). Data shown are typical [Ca 2ϩ ] ex responses from three to four different cell preparations. volved in suppressing the magnitude of IP 3 -induced Ca 2ϩ fluxes but may partially mediate the reduction in IP 3 R sensitivity.
The most likely explanation for the apparent smaller IP 3sensitive Ca 2ϩ pool in CSA-treated cells is the enhanced rate of Ca 2ϩ clearance by SERCA, which is expected to buffer the [Ca 2ϩ ] ex responses. Moreover, the findings described above are consistent with both SERCA activity and the mitochondria acting synergistically to inhibit IP 3 -induced Ca 2ϩ release in CSA-treated hepatocytes. This effect could be mediated by suppression of the positive feedback effects of Ca 2ϩ on the IP 3 R. To determine the role of the mitochondria in this process, we investigated the effects of CSA on mitochondrial Ca 2ϩ transport mechanisms.
Effect of CSA on Mitochondrial Transport Mechanisms-Hepatocytes were maintained in primary culture overnight then co-loaded with Mito Tracker Green, to visualize the mitochondria, and the Ca 2ϩ -sensitive indicator, rhod/2-AM to measure mitochondrial matrix-free Ca 2ϩ concentrations ([Ca 2ϩ ] m ) (33,36). Cells were simulated with maximal VP concentrations and [Ca 2ϩ ] m responses monitored by confocal microscopy. As reported previously (33,36), VP stimulation rapidly increased rhod-2 fluorescence in individual mitochondria (Fig. 7A). Analysis of rhod-2 fluorescence changes (500 -1000 mitochondria per cell) revealed that CSA pretreatment increased the rate of rise in [Ca 2ϩ ] m after VP stimulation from 8.3 Ϯ 2.5%/s to 21.7 Ϯ 5.8%/s (p Ͻ 0.05; n ϭ 2-4 cells). It is worth re-emphasizing that CSA did not alter the rate nor the magnitude of VP-induced [Ca 2ϩ ] i responses (Fig. 3A), indicating that rhod-2 is reporting a differential effect on mitochondrial Ca 2ϩ uptake in situ. No significant differences in the magnitude of [Ca 2ϩ ] m responses could be detected with rhod-2; fluorescence intensities increased 210 Ϯ 34% and 240 Ϯ 7% above basal values in the absence and present of CSA, respectively (p Ͼ 0.5). However, this may reflect saturation of rhod-2.
The total accumulation of mitochondrial matrix Ca 2ϩ during a rise in [Ca 2ϩ ] i was measured in intact hepatocyte suspensions (49) pretreated with maximal VP (Fig. 7B) or VP plus Tg (Fig. 7C). Cells were quickly washed into a BAPTA-containing medium plus fura-2-free acid, and the compartmentalized Ca 2ϩ was released by ionomycin (Iono). CSA significantly increased the amount of Ca 2ϩ accumulated by the mitochondria during VP stimulation (Fig. 7B). Under these conditions, the ionomycin-releasable Ca 2ϩ pool predominately originated from the mitochondria, because the same effect was also observed in cells treated with VP plus Tg (Fig. 7C). Moreover, greater than 90% of the ionomycin-releasable Ca 2ϩ pool was blocked by including mitochondrial inhibitors during the preincubation period (Fig. 7C, FCCP). In the presence of VP alone, CSA increased mitochondrial Ca 2ϩ content from 3.9 Ϯ 0.7 to 12.6 Ϯ 1.9 nmol/mg of mitochondrial protein (p Ͻ 0.05; mean Ϯ S.E., n ϭ 3). In VP-plus Tg-treated hepatocytes, matrix Ca 2ϩ content was 13.6 Ϯ 0.8 and 21.2 Ϯ 1.3 nmol/mg of mitochondrial protein in the absence and presence of CSA, respectively (p Ͻ 0.05; mean Ϯ S.E., n ϭ 3).
CSA is known to decrease the open probability of the mitochondrial PTP (6), which in some cell types leads to an increase in mitochondrial membrane potential (⌬⌿ m ) (50,51). Because the initial rate of mitochondrial Ca 2ϩ uptake is proportional to the magnitude of ⌬⌿ m (52), we explored the possibility that CSA mediates its effect through changes in ⌬⌿ m . We have shown previously that real-time measurement of ⌬⌿ m can be achieved in intact cells using the fluorescent indicator, tetramethylrhodamine ethyl ester (TMREE) (33,36). Addition of CSA to naive cells caused a stable increase in the mean TM-REE fluorescence, indicating a rise in ⌬⌿ m corresponding to a 1-2% increase in the logarithm of the fluorescence ratio (Fig.  7D). This observation is consistent with a constitutively active mitochondrial PTP or megachannel operating in a low conductance mode (i.e. flickering). For comparison, hepatocytes were subsequently treated with K ϩ /H ϩ ionophore, nigericin (Nig, Fig. 7D), that slowly abolishes the mitochondrial proton gradient converting the free energy into ⌬⌿ m (36) and, thus, hyperpolarizes the mitochondrial inner membrane.
To further investigate the effects of CSA on mitochondrial Ca 2ϩ transport without the interference of plasma membrane Ca 2ϩ transporters, we returned to the permeabilized hepatocyte preparation. In these experiments, Tg was added to preclude any CSA-dependent effects on the SERCA pumps and a low affinity Ca 2ϩ indicator dye (BTC; K d ϭ 7 M) was employed to measure [Ca 2ϩ ] ex . A CaCl 2 pulse (30 nmol/mg of cell protein) to pre-energized mitochondria initiated Ca 2ϩ uptake (Fig. 8A) which was indicated by a decrease in [Ca 2ϩ ] ex following the Ca 2ϩ pulse. A 5-min preincubation with CSA (5 M) prior to FIG. 6. Concentration response curves for IP 3 -induced Ca 2؉ release. Hepatocyte suspensions were washed extensively to removed extracellular Ca 2ϩ then suspended in Chelex-treated ICM supplemented with 5 M cyclosporin A (CSA, f) or Me 2 SO (0.1% v/v, q) and incubated 5 min in a stirred cuvette. Cells were then digitonin-permeabilized in the presence of an ATP-regenerating system plus glutamate and pyruvate as described under "Experimental Procedures." Fura-2-free acid was used to measure IP 3 -induced Ca 2ϩ fluxes, which are normalized to the magnitude of the ionomycin-sensitive Ca 2ϩ store (% Iono). A, effect of CSA on the IP 3 concentration response curve in the presence of respiring, coupled mitochondria. B, hepatocyte suspensions were treated with rotenone and oligomycin prior to permeabilization to block mitochondrial Ca 2ϩ uptake. The data are the means Ϯ S.E. (n ϭ 3-7 hepatocyte preparations). The data points were fit to the Hill equation.
digitonin permeabilization significantly enhanced the initial rate of mitochondrial Ca 2ϩ uptake (Fig. 8, A and B; p Ͻ 0.05), consistent with the rhod-2 measurements in intact hepatocytes (Fig. 7A). Ruthenium red (1 M) was added after Ca 2ϩ uptake to block the mitochondrial uniporter, thus permitting the measurement of Ca 2ϩ efflux rates (Fig. 8C). Addition of ruthe- FIG. 7. Cyclosporin A enhances mitochondrial Ca 2؉ uptake and increases ⌬⌿ m in intact hepatocytes. A, rhod-2 fluorescence changes during maximal vasopressin (VP, 100 nM) stimulation. Dual-wavelength confocal microscopy was used to simultaneously monitor Mito Tracker Green fluorescence, to visualize mitochondrial structure, and rhod-2 to measure [Ca 2ϩ ] m responses (33,36). CSA (5 M, gray trace) was present 5 min prior to data acquisition. The data represent the mean rhod-2 fluorescence from all the mitochondria within the cell. B and C, effect of CSA on the total mitochondrial Ca 2ϩ content. Intact hepatocytes were suspended in Ca 2ϩ -containing KR-HEPES buffer supplemented with (B) VP (100 nM) or (C) VP plus thapsigargin (2 M, Tg) and gently shaken for 15 min in the presence or absence of CSA (5 M). Cell suspensions were then washed into Ca 2ϩ -free KR-HEPES buffer supplemented with BAPTA and fura-2-free acid followed by ionomycin (Iono) to release compartmentalized Ca 2ϩ . In some experiments, FCCP plus oligomycin was included during the preincubation period to prevent mitochondrial Ca 2ϩ accumulation (C, FCCP). D, typical mean mitochondrial membrane potential (⌬⌿ m ) response in TMREE-loaded hepatocytes. Cells were challenged with CSA (5 M) followed by the K ϩ /H ϩ ionophore, nigericin (Nig, 1 M). The logarithm of TMREE fluorescence intensity changes were normalized as described under "Experimental Procedures." nium red prior to the Ca 2ϩ pulse completely blocked all Ca 2ϩ uptake (not shown). CSA dramatically suppressed the rate of Ca 2ϩ egress (p Ͻ 0.05; Fig. 8, C and D), which could account for the large increase in mitochondrial Ca 2ϩ content measured in the intact hepatocytes (Fig. 7, B and C). As expected, FK-506 did not affect mitochondrial Ca 2ϩ fluxes (Fig. 8), consistent with intact cell data and the absence of FKBP immunophilins in the mitochondrial matrix (2). Taken together, these results suggest that the higher ⌬⌿ m in the presence of CSA can increase the driving force for mitochondrial Ca 2ϩ uptake. Moreover, transient mitochondrial PTP openings may also function as a route for Ca 2ϩ efflux, and CSA appears to inhibit this mitochondrial Ca 2ϩ release pathway, as well as enhancing Ca 2ϩ uptake by increasing ⌬⌿ m .
Effect of CSA on IP 3 -dependent Ca 2ϩ Signaling in the Absence of Mitochondrial Ca 2ϩ Transport-CSA-dependent stimulation of either Tg-sensitive Ca 2ϩ pumps or mitochondrial Ca 2ϩ uptake could interfere with local positive Ca 2ϩ feedback on the IP 3 R. To evaluate the role of the mitochondria in this process, we compared the effects of CSA on IP 3 -dependent [Ca 2ϩ ] i signals in the presence and absence of mitochondrial inhibitors (i.e. FCCP, rotenone, oligomycin; UNC) in intact hepatocytes. As shown previously, ⌬⌿ m depolarization results in a transient increase in [Ca 2ϩ ] i as matrix Ca 2ϩ is released (Fig. 9A). Under these conditions, cellular ATP does not appear to be limiting for [Ca 2ϩ ] i homeostatic mechanisms, because both basal and inter-spike [Ca 2ϩ ] i levels remained stable after the initial FCCP-induced [Ca 2ϩ ] i response. Importantly, the mitochondrial inhibitors did not prevent IP 3 -dependent [Ca 2ϩ ] i oscillations (Fig. 9A); stimulation with submaximal PE concentrations resulted in repetitive baseline-separated [Ca 2ϩ ] i oscillations in 70 -80% of the cells, which was not significantly different from control. Moreover, the magnitude of the [Ca 2ϩ ] i spike remained the same, [Ca 2ϩ ] i peak height was 466 Ϯ 43 nM and 497 Ϯ 26 nM in the absence and presence of mitochondrial toxins, respectively (mean Ϯ S.E.; n ϭ 100 -125 cells, nine cell preparations). In the presence of mitochondrial inhibitors, the inhibitory effects of CSA on PE-induced [Ca 2ϩ ] i responses was significantly reduced (Fig. 9B, ϩCSA). CSA inhibited the frequency of PE-induced [Ca 2ϩ ] i oscillations by 62 Ϯ 2% under physiological conditions, but by only 38 Ϯ 2% after blocking mitochondrial Ca 2ϩ uptake (p Ͻ 0.001). Thus, blocking mitochondrial Ca 2ϩ uptake reduced the effects of CSA on both IP 3 -dependent [Ca 2ϩ ] i oscillations in intact cells (Fig. 9B) and IP 3 -induced Ca 2ϩ release in permeabilized hepatocytes (Fig.  6B), indicating that mitochondria contribute to the decrease in IP 3 R sensitivity.
The rates of rise and fall of individual [Ca 2ϩ ] i spikes were significantly increased in the presence of mitochondrial inhibitors (p Ͻ 0.001; compare Fig. 2, D-E, with Fig. 9, D-E). However, CSA did not significantly affect the kinetics of the [Ca 2ϩ ] i spikes under these conditions (Fig. 9, C-E). Thus, the disruption of mitochondrial Ca 2ϩ transport eliminated the effects of CSA on the shape of the [Ca 2ϩ ] i spike. Half-peak width was 10.6 Ϯ 0.4 and 11.7 Ϯ 0.5 s before and after CSA, respectively (p Ͼ 0.05). These data suggest that the CSA-mediated decrease in the declining phase of the [Ca 2ϩ ] i spike, observed in the absence of mitochondrial inhibitors (Fig. 2E), could be explained by an increase in total mitochondrial Ca 2ϩ content (Fig.  7B) coupled with slower mitochondrial Ca 2ϩ egress (Fig. 8D).

DISCUSSION
Regulation of Ca 2ϩ Transport Mechanisms-The ubiquitous tissue expression, broad subcellular distribution, and multiple protein targets situate the cyclophilin family at a critical juncture to regulate a diverse array of cellular activities (1-3, 5). Our study has identified two distinct effects of CSA on Ca 2ϩ transport pathways in hepatocytes: Tg-sensitive Ca 2ϩ pumps and mitochondrial Ca 2ϩ uptake and retention. MeVal-CS has similar effects on Ca 2ϩ transport mechanisms, indicating that calcineurin is probably not involved. Thus, it would appear that specific cyclophilin isozymes directly regulate these cellular Ca 2ϩ transport mechanisms.
In intact hepatocytes, CSA or MeVal-CS stimulated a Tgsensitive decrease in baseline [Ca 2ϩ ] i under resting conditions or during agonist stimulation (Figs. 1-4). This effect on free Ca 2ϩ levels could be fully reconstituted in a permeabilized cell system, pointing to an ER-associated cyclophilin as a possible mediator. Several mechanisms could contribute to lower [Ca 2ϩ ] ex in permeabilized hepatocytes: (a) decreased Ca 2ϩ leak from internal stores (b) increased luminal Ca 2ϩ -buffering capacity and/or (c) a change in the kinetics properties of the Ca 2ϩ pumps. Because CSA does not affect the size of the ionomycinsensitive Ca 2ϩ pool (see text) nor the passive Ca 2ϩ leak rate from internal stores (Fig. 5B), the first two options do not appear to be major contributors. However, we cannot rule out the loss of cytosolic factors or partial disruption of ER integrity during permeabilization affecting our results. Nevertheless, the simplest explanation is a change in the kinetic properties of the Ca 2ϩ pump; an enhanced rate of Ca 2ϩ uptake and lower final set point for [Ca 2ϩ ] ex suggest an increased affinity for Ca 2ϩ .
Previous studies have identified members of the cyclophilin family as calreticulin binding proteins (21). In addition to its role as a Ca 2ϩ storage protein, calreticulin is also a lectin-like molecular chaperone and participates in the correct folding of newly synthesized glycoproteins (53). Recently, it has been proposed that the chaperon domain of calreticulin might also specifically interact with the mature form of SERCA 2b to modulate IP 3 -mediated Ca 2ϩ release in Xenopus oocytes (54,55). SERCA 2b is the only isoform expressed in hepatocytes (56) and differs from other SERCA sub-types, containing an additional transmembrane domain that localizes a putative N-glycosylation site in the C-terminal to the ER lumen (53). In oocytes, overexpression of calreticulin inhibits the repetitive [Ca 2ϩ ] i spikes induced by IP 3 microinjection, apparently by modulating the functional conformation and, thus, Ca 2ϩ transport activity of SERCA 2b (54,55). In this model, dissociation of calreticulin from SERCA 2b restores full enzymatic activity, whereas calreticulin binding reduces Ca 2ϩ transport. It is an intriguing possibility that the CSA⅐cyclophilin complex may directly inhibit calreticulin binding to the Ca 2ϩ pump. Alternatively, calreticulin and cyclophilin may form a complex with SERCA 2b, which is disrupted by CSA, a mechanism similar to displacing CyP-D from the mitochondrial PTP (6 -8).
CSA is known to inhibit the PTP in isolated mitochondria and suppresses necrotic cell death associated with mitochondrial Ca 2ϩ overload in a variety of cells types (6,12,13). The CSA-sensitive mitochondrial target is presumably CyP-D; CSA binding disassociates CyP-D from the PTP complex (7,8) decreasing the open probability of the channel. Classic mitochondrial studies have put forward the concept that PTP opening has irreversible deleterious effects on mitochondrial function. This includes mitochondrial depolarization, release of matrix ions, and the loss mitochondrial metabolites into the cytosol (6). However, more recent evidence suggests that the PTP operates in two distinct conducting states, high conductance involved in the irreversible opening of the pore, and a low conductance mode, more selective to protons and Ca 2ϩ (9,10). In hepatocytes, CSA promoted a stable increase in ⌬⌿ m (Fig. 7D) consistent with the presence of a constitutively active but low conducting state of the PTP that partially dissipates the mitochondrial electrochemical gradient. CSA has also been shown to increase basal ⌬⌿ m in several other cell types using similar fluorescence imaging techniques (50,51). Moreover, a spontaneous and transient CSA-sensitive increase in membrane permeability has been reported in single isolated cardiac mitochondria and mitochondria in intact hepatocytes (57,58). Taken together, these data strongly suggest that CSA-sensitive "pores" or channels undergo cyclical opening and closing to contribute to the basal proton leak rate across the mitochondrial inner membrane.
CSA enhanced mitochondrial Ca 2ϩ uptake in both permeabilized and intact hepatocytes while inhibiting Ca 2ϩ egress (Figs. 7 and 8). Ca 2ϩ transport across the mitochondrial inner membrane is regulated by components of the mitochondrial PMF. The large negative ⌬⌿ m drives mitochondrial Ca 2ϩ uptake through the Ca 2ϩ uniporter, whereas both ⌬⌿ m and the mitochondrial proton gradient can contribute to Ca 2ϩ efflux (59). In isolated rat mitochondria, the initial rate of Ca 2ϩ uptake increases in proportion to the magnitude of ⌬⌿ m (52). Thus, the CSA-induced rise in ⌬⌿ m is the most likely parameter mediating the enhanced rate of mitochondrial Ca 2ϩ uptake. In liver, mitochondrial Ca 2ϩ efflux is predominately mediated by a Na ϩ -independent mechanism, presumably H ϩ /Ca 2ϩ exchange, because the Na ϩ /Ca 2ϩ exchanger is suppressed by physiological concentrations of Mg 2ϩ in this tissue (59). Both mitochondrial Ca 2ϩ efflux mechanisms have been proposed to be electrogenic (60,61) and thus, Ca 2ϩ egress should be stimulated by a rise in ⌬⌿ m and not suppressed as we observed in this investigation (Fig. 8). Thus, our data suggests two possible alternatives: (a) the low conducting mitochondrial PTP is the major Ca 2ϩ efflux pathway in hepatocytes or (b) the CSA⅐cyclophilin complex directly inhibits mitochondrial Ca 2ϩ efflux pathways.
Effect of CSA on IP 3 -dependent [Ca 2ϩ ] i Spikes-CSA and MeVal-CS both inhibited the frequency of PE-induced [Ca 2ϩ ] i oscillations in a concentration-dependent manner (Fig. 1). In the case of CSA, we show a small, but reproducible, decrease in IP 3 sensitivity in permeabilized hepatocytes (Fig. 6). The rightward shift in the IP 3 concentration response curve was sensitive to inhibitors that eliminated mitochondrial Ca 2ϩ uptake (Fig. 6B). Moreover, the same inhibitor mixture decreased the effects of CSA on IP 3 -dependent Ca 2ϩ signals in intact hepatocytes (Fig. 9B), indicating that Ca 2ϩ uptake by the mitochondria plays a role in shifting IP 3 R sensitivity both in vitro and in situ. We have previously reported that the close association between ER Ca 2ϩ release channels and mitochondrial Ca 2ϩ uptake sites allows the mitochondria to influence IP 3 R excitability by suppressing the local Ca 2ϩ gradients surrounding the mouth of the channel (32). The observation that CSA manifests a profound effect on the rate of mitochondrial Ca 2ϩ uptake in both permeabilized and intact hepatocytes (Figs. 7 and 8) is consistent with the modulation of IP 3 R function through such local Ca 2ϩ gradients.
The mitochondria are not the sole participants in this complex mechanism; elimination of mitochondrial Ca 2ϩ uptake did not completely reverse the effects of CSA (Figs. 6B and 9B). Although we cannot absolutely exclude a direct effect of CSA on IP 3 R function, the simplest explanation is that the Tg-sensitive Ca 2ϩ pumps, together with mitochondria, act synergistically to reset the threshold for initiating a [Ca 2ϩ ] i spike. Because Ca 2ϩ is involved in both the activation and inactivation of the IP 3 R (22-24), Ca 2ϩ transport mechanisms would be excepted to influence IP 3 R function. Indeed, both SERCA activity and mitochondrial Ca 2ϩ uptake have been shown to exert a negative influence on oscillatory [Ca 2ϩ ] i signals by suppressing Ca 2ϩ spike initiation (32,62,63), which is consistent with our interpretation.
CSA prolonged the decay phase of the [Ca 2ϩ ] i spike (Fig. 2E). We hypothesized that this may reflect slow Ca 2ϩ release from the mitochondria, because CSA enhances matrix Ca 2ϩ accumulation and retention during agonist stimulation (Figs. 7 and 8).
To test for a mitochondrial origin, we inhibited matrix Ca 2ϩ transport in intact hepatocytes by depolarizing ⌬⌿ m with mitochondrial inhibitors prior to agonist stimulation. Depolarizing the mitochondria significantly increased both the rising and falling phases of the [Ca 2ϩ ] i spike (compare Figs. 2 and 9). We have previously shown that blocking mitochondrial Ca 2ϩ uptake increases the efficacy of submaximal IP 3 to release Ca 2ϩ from internal stores (32), which could account for the stimulation of the rising phase of the [Ca 2ϩ ] i spike. Moreover, the faster rate of decline is presumably due to the absence of mitochondrial Ca 2ϩ being released back to the cytosol. Although some ATP depletion occurs over the time course of these experiments (33), it does not appear to impair [Ca 2ϩ ] i homeostasis (Fig. 9A). Indeed, the decay rate of the [Ca 2ϩ ] i spike increased, not decreased as would be expected if ATP-dependent Ca 2ϩ transport was inhibited. Consequently, the effects of mitochondrial toxins on the kinetic properties of the [Ca 2ϩ ] i spike appear to be specific to blocking mitochondrial Ca 2ϩ transport and not due to a drop in cellular ATP levels.
Importantly, the effects of CSA on the kinetic properties of the [Ca 2ϩ ] i spike were eliminated in the presence of mitochondrial inhibitors (Fig. 9). Under physiological conditions, CSA presumably prolongs the [Ca 2ϩ ] i transient by increasing mitochondrial Ca 2ϩ accumulation during the rising phase of the [Ca 2ϩ ] i spike, coupled with a dramatic reduction in mitochondrial Ca 2ϩ egress. By contrast, the CSA-induced increase in the rate of rise of the [Ca 2ϩ ] i spike (Fig. 2D) is probably not due to its action on mitochondria, because inhibition of mitochondrial Ca 2ϩ uptake with uncoupler also enhanced the rate of [Ca 2ϩ ] i rise (Fig. 9D). The effect of CSA on the rate of [Ca 2ϩ ] i increase could be explained by its effect on the SERCA pump, either by increasing the pool of ER Ca 2ϩ for release, or by increasing the proportion of IP 3 Rs in the basal state and available to participate in the Ca 2ϩ release process.
Finally, the inhibition of agonist-induced [Ca 2ϩ ] i responses was not limited to the cyclophilin family. Preliminary experiments revealed that FK-506, ascomycin, and rapamycin have similar effects on the frequency on hepatic Ca 2ϩ signals, although at a reduced efficacy compared with CSA. The relationship between the FKBP family of proteins and IP 3 -dependent [Ca 2ϩ ] i responses is still under investigation. To date, none of the FKBP binding drugs have been shown to effect plasma membrane Ca 2ϩ fluxes, whereas each drug potently stimulates a Tg-sensitive decrease in basal [Ca 2ϩ ] i . Moreover, these drugs do not modify mitochondrial Ca 2ϩ transport in permeabilized cell suspensions (Fig. 7), which is consistent with a lack of FKBPs in the mitochondrial matrix (2). Thus, it would appear that cyclophilins and FKBPs can both regulate specific Ca 2ϩ transport pathways, thereby influencing IP 3 -dependent [Ca 2ϩ ] i responses.
Mitochondria and Ca 2ϩ Signaling-Mitochondria are well known participants in the regulation of [Ca 2ϩ ] i homeostasis, capable of modulating cytosolic Ca 2ϩ signals (9, 63-67) and exerting local control over IP 3 R excitability (32). In many cells, the ER envelops the mitochondria placing the mitochondrial Ca 2ϩ uptake sites in juxtaposition to the IP 3 Rs (68) or to the RyRs (69). Thus, mitochondria have privileged access to the highly localized Ca 2ϩ gradients generated during ligand-induced ER Ca 2ϩ mobilization (28,67,70,71). This strategic location facilitates the activation of mitochondrial Ca 2ϩ uptake and, thus, Ca 2ϩ -dependent regulation of metabolic processes in the mitochondrial matrix (33,36). Moreover, in some cell types, these local Ca 2ϩ gradients may also serve as a trigging mechanism to open the mitochondrial PTP, resulting in mitochondrial calcium-induced calcium-release and amplification of IP 3dependent [Ca 2ϩ ] i spikes (9).
Our data are clearly consistent with the mitochondria playing an important role in setting the frequency and modulating the kinetic properties of IP 3 -dependent [Ca 2ϩ ] i oscillations in hepatocytes. However, it is worth emphasizing several points regarding the role of the mitochondria in IP 3 -dependent [Ca 2ϩ ] i signaling. First, functional mitochondria are not essential components required to generate oscillatory Ca 2ϩ signals in hepatocytes (Fig. 9), consistent with our previous suggestion that only IP 3 and Ca 2ϩ are needed to produce periodic opening and closing of the IP 3 R (31). Second, in contrast to previous reports, blocking mitochondrial Ca 2ϩ uptake does not affect the magnitude of the [Ca 2ϩ ] i spike (reviewed in Ref. 14), indicating that other Ca 2ϩ transport mechanisms or cellular Ca 2ϩ buffers regulate peak height. Nevertheless, mitochondrial Ca 2ϩ uptake can alter the kinetics properties of the [Ca 2ϩ ] i spike (Fig. 9) and modulate the propagation rate of intracellular Ca 2ϩ waves (32). Finally, our data are not consistent with localized Ca 2ϩ gradients triggering the opening of the mitochondrial PTP. In hepatocytes, total mitochondrial PMF increases during IP 3 -dependent Ca 2ϩ mobilization (33,36) inconsistent with the opening of a large pore in the inner mitochondrial membrane. Perhaps, the sluggish rate of mitochondrial Ca 2ϩ uptake, time-to-peak ϳ14 s (33), is insufficient to "trigger" the PTP into the low conductance mode required for ⌬⌿ m depolarization and subsequent rapid matrix Ca 2ϩ release. Nevertheless, the mitochondrial CSA-sensitive Ca 2ϩ release pathway should still have profound effects on the activation of cytosolic Ca 2ϩ -sensitive targets, such as CaM kinase II (72), by controlling the duration of the [Ca 2ϩ ] i spike.
In summary, the properties of hepatic mitochondria devise a potentially complex feedback mechanism to modulate oscillatory Ca 2ϩ signals. The close association between ER and mitochondrial membranes (68) ensures efficient transfer of Ca 2ϩ between the organelles during IP 3 -dependent Ca 2ϩ mobilization. The initial [Ca 2ϩ ] i spike would activate Ca 2ϩ -dependent metabolic pathways in the mitochondrial matrix stimulating the respiratory chain (28,33,36) and increasing the driving force for mitochondrial Ca 2ϩ uptake which, in turn, suppresses the initiation of subsequent [Ca 2ϩ ] i spikes. Thus, the mitochondria may set the upper limit for the frequency of IP 3 -dependent [Ca 2ϩ ] i oscillations. A feedback mechanism, including the mitochondria balances the need for stimulating ATP production without exposing the mitochondria or the cytosol to excessive [Ca 2ϩ ] i signals, beyond what is required to maximally stimulate metabolism.