Mitochondria Suppress Local Feedback Activation of Inositol 1,4,5-Trisphosphate Receptors by Ca2+ *

The concerted action of inositol 1,4,5-trisphosphate (IP3) and Ca2+ on the IP3 receptor Ca2+ release channel (IP3R) is a fundamental step in the generation of cytosolic Ca2+ oscillations and waves, which underlie Ca2+ signaling in many cells. Mitochondria appear in close association with regions of endoplasmic reticulum (ER) enriched in IP3R and are particularly responsive to IP3-induced increases of cytosolic Ca2+([Ca2+] c ). To determine whether feedback regulation of the IP3R by released Ca2+ is modulated by mitochondrial Ca2+ uptake, the interactions between ER and mitochondrial Ca2+ pools were examined by fluorescence imaging of compartmentalized Ca2+ indicators in permeabilized hepatocytes. IP3 decreased luminal ER Ca2+ ([Ca2+]ER), and this was paralleled by an increase in mitochondrial matrix Ca2+([Ca2+] m ) and activation of Ca2+-sensitive mitochondrial metabolism. Remarkably, the decrease in [Ca2+]ER evoked by submaximal IP3 was enhanced when mitochondrial Ca2+ uptake was blocked with ruthenium red or uncoupler. Moreover, subcellular regions that were relatively deficient in mitochondria demonstrated greater sensitivity to IP3 than regions of the cell with a high density of mitochondria. These data demonstrate that Ca2+ uptake by the mitochondria suppresses the local positive feedback effects of Ca2+ on the IP3R, giving rise to subcellular heterogeneity in IP3 sensitivity and IP3R excitability. Thus, mitochondria can play an important role in setting the threshold for activation and establishing the subcellular pattern of IP3-dependent [Ca2+] c signaling.

The mobilization of intracellular Ca 2ϩ stores in response to receptor-stimulated formation of inositol 1,4,5-trisphosphate (IP 3 ) 1 is dependent on IP 3 receptor Ca 2ϩ channels (IP 3 R) in the endoplasmic reticulum (ER) (1)(2)(3)(4). Both activation and deactivation of the IP 3 R is regulated by cytosolic [Ca 2ϩ ] ([Ca 2ϩ ] c ) (5)(6)(7)(8)(9), and this feedback control of IP 3 R function by released Ca 2ϩ gives rise to the complex spatio-temporal organization of IP 3 -induced Ca 2ϩ release. Because the regulation of [Ca 2ϩ ] c involves a number of other Ca 2ϩ transport mechanisms (reviewed in Ref. 10), Ca 2ϩ feedback on IP 3 R may be modulated by other organelles that transport Ca 2ϩ .
Mitochondria are well known to participate in intracellular Ca 2ϩ homeostasis, although mitochondrial Ca 2ϩ uptake is relatively insensitive to submicromolar increases of [Ca 2ϩ ] c (reviewed in Refs. 10 and 11). Rizzuto, Pozzan, and co-workers (12)(13)(14) have demonstrated that IP 3 (15). The high efficiency of Ca 2ϩ signal transmission between the ER and mitochondria is likely to be established by a privileged or local transfer of Ca 2ϩ from ER release sites to the mitochondrial Ca 2ϩ uptake pathway (12)(13)(14)(15). Close associations of ER and mitochondrial membranes (14,16) and clustering of IP 3 R in ER membranes facing mitochondria (17)(18)(19) are consistent with such local Ca 2ϩ signaling. Although it is also becoming apparent that mitochondria modulate cytosolic Ca 2ϩ signaling (20 -25), it is not clear whether mitochondria can exert a local control over the feedback effects of IP 3 -induced Ca 2ϩ release on the IP 3 R itself.
In the present study we demonstrate that Ca 2ϩ uptake by the mitochondria suppresses the positive feedback effects of Ca 2ϩ on the IP 3 R in permeabilized hepatocytes. Moreover, our data demonstrate that the mitochondrial modulation of IP 3induced Ca 2ϩ release is limited to those elements of the ER Ca 2ϩ stores in proximity with the mitochondria, giving rise to subcellular heterogeneity in IP 3 sensitivity and IP 3 R excitability. These properties allow the mitochondria to play a key role in orchestrating the subcellular pattern of [Ca 2ϩ ] c signaling.

EXPERIMENTAL PROCEDURES
Hepatocytes plated on polylysine-coated coverslips were maintained in primary culture for 18 -24 h (15, 26). Cytosolic [Ca 2ϩ ] waves in fura2-loaded intact hepatocytes were measured essentially as described previously (15,27). The cells were stimulated with vasopressin (2-20 nM) prior to and after addition of mitochondrial inhibitors or solvent in sequential runs, and the rate of wave propagation was determined in each condition (15,27). For permeabilized cell experiments, cells were loaded with fluorescent dyes (obtained from Molecular Probes or Teflabs) by incubation for 30 -60 min at 37°C in medium composed of 121 mM NaCl, 5 mM NaHCO 3 , 10 mM Na-HEPES, 4.7 mM KCl, 1.2 mM KH 2 PO 4 , 1.2 mM Mg 2 SO 4 , 2 mM CaCl 2 , 10 mM glucose, and 2% bovine serum albumin, pH 7.4, essentially as described previously (9,15,26). Dye concentrations were: 150 nM MitoTracker Green, 2 M rhod2/AM, 5 M fura2FF/AM, and 5 M fluo3FF/AM. We have shown previously that compartmentalization of rhod2 occurs in the mitochondria (15) and fura2FF is trapped in the ER (26) of hepatocytes using this loading protocol. Dye-loaded cells were washed with Ca 2ϩ -free buffer and then permeabilized by incubation for 6 min with 15 g/ml digitonin in intracellular medium (ICM) composed of 120 mM KCl, 10 mM NaCl, 1 * This work was supported by Grants DK38422 (to A. P. T.) and DK51526 (to G. H.) from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  ] m using dual emission confocal imaging of permeabilized hepatocytes loaded with MitoTracker Green and rhod2 (show in green and red, respectively). These images are overlaid in the right two panels to show the coincidence of the labeled organelles (Overlay) and the decrease in the rhod2 signal following 5 min of treatment with 5 g/ml uncoupler 1799 and 5 g/ml oligomycin (ϩUncoupler). B, dual emission confocal images of CaGreenC18 and MitoTracker red (left two panels) and overlay of these images in permeabilized hepatocytes. C, overlay image prior to IP 3 addition taken from a dual emission confocal image series using CaGreenC18 (green) and rhod2 (red) to obtain simultaneous measurements of Experiments were carried out with at least three different cell preparations. Traces represent single cell responses unless indicated otherwise. Data are presented as the means Ϯ S.E. Significance of differences from the relevant controls was calculated by Student's t test.

RESULTS AND DISCUSSION
In previous studies we have demonstrated that global application of IP 3 to permeabilized hepatocytes results in oscillatory release and reuptake of [Ca 2ϩ ] ER and that this reproduces the basic mechanism of [Ca 2ϩ ] c oscillations in intact cells treated with hormones (26). We have used a similar approach to examine the interactions between mitochondrial and ER Ca 2ϩ stores. [Ca 2ϩ ] m was monitored with compartmentalized rhod2 (15,28). Double labeling with the vital mitochondrial dye MitoTracker Green (29) demonstrated that rhod2 fluorescence was completely coincident with the mitochondria in permeabilized hepatocytes (Fig. 1A). Moreover, the [Ca 2ϩ ] m decrease elicited by uncoupler was manifest in a reduction of rhod2 fluorescence for all of the intracellular structures that were double labeled with MitoTracker (compare overlay panels iii and iv of Fig. 1A indicators (fura2FF and fluo3FF) as described previously (26). As an alternative approach, Ca 2ϩ at the cytosolic face of intracellular membranes ([Ca 2ϩ ] memb ) was measured with the lipophilic indicator CaGreenC18 (30). CaGreenC18 labeled membranes throughout the cell, apart from the nuclear matrix, whereas MitoTracker Red fluorescence was predominantly perinuclear, consistent with the subcellular location of mitochon-dria in these cells (Fig. 1B). This differential distribution is also shown in the overlaid CaGreenC18 and rhod2 dual label images of Fig. 1C. A similar global distribution of the ER Ca 2ϩ stores was observed with fluo3FF or fura2FF, and the perinuclear organization of the mitochondria was also demonstrated based on pyridine nucleotide fluorescence and the alkaline pH of the mitochondrial matrix (see Fig. 3).
Intracellular stores were loaded with Ca 2ϩ by incubating the permeabilized cells in the presence of ATP without Ca 2ϩ buffers, essentially as described previously (26). Addition of maximal IP 3 to cells loaded with CaGreenC18 and rhod2 resulted in rapid Ca 2ϩ release that was detected as an increase in [Ca 2ϩ ] memb and a simultaneous increase in [Ca 2ϩ ] m (Fig. 1D). Nevertheless, the CaGreenC18 and rhod2 signals responded differently to uncoupler, which selectively reduced [Ca 2ϩ ] m (not shown). The IP 3 -induced decrease in [Ca 2ϩ ] ER could be monitored directly with luminal fura2FF (26), and this was also accompanied by a rapid increase in [Ca 2ϩ ] m measured simultaneously with rhod2 ( Fig. 1E). At the maximal levels of IP 3 used in Fig. 1E, the ER remained depleted of Ca 2ϩ , but [Ca 2ϩ ] m declined after the peak, as reported previously for [Ca 2ϩ ] m in intact cells stimulated with a maximal dose of hormone (15). Addition of heparin to block the IP 3 receptor allowed recovery of [Ca 2ϩ ] ER with essentially no effect on [Ca 2ϩ ] m , whereas addition of uncoupler to collapse the mitochondrial membrane potential caused [Ca 2ϩ ] m to decrease without affecting [Ca 2ϩ ] ER (Fig. 1E). The residual ER Ca 2ϩ could be released with ionophore. Several intramitochondrial dehydrogenases are activated by elevated [Ca 2ϩ ] m (31), and this activation can be monitored fluorometrically through changes in pyridine nucleotide redox state in intact hepatocytes (15,32). Fig. 1F shows that the IP 3 -induced Ca 2ϩ release led to an increase in NAD(P)H fluorescence in permeabilized hepatocytes, reflecting the Ca 2ϩ -dependent dehydrogenase activation. Taken together, the data of Fig. 1 demonstrate that mitochondrial Ca 2ϩ uptake and the consequent regulation of intramitochondrial metabolism is coupled to IP 3 -induced Ca 2ϩ release from the ER in permeabilized hepatocytes. Because the Ca 2ϩ released by IP 3 plays a key role in both positive and negative feedback regulation of the IP 3 receptor Ca 2ϩ channel (5-9), we used this system to investigate whether mitochondrial Ca 2ϩ uptake modulates IP 3 -induced Ca 2ϩ release.
The [Ca 2ϩ ] ER decrease elicited by submaximal and maximal IP 3 was measured under the conditions described above, where the mitochondria were able to take up part of the released Ca 2ϩ , and compared with conditions where mitochondrial Ca 2ϩ , uptake was blocked with ruthenium red or uncoupler (Fig. 2). These inhibitors affect neither the steady state [Ca 2ϩ ] c nor the amount of Ca 2ϩ released by IP 3 in liver microsomes, suggesting that they have no direct effect on Ca 2ϩ release from ER in hepatocytes (33,34). Despite the fact that the mitochondrial blockers removed a sink for the released Ca 2ϩ , the extent of ER Ca 2ϩ release at submaximal IP 3 was actually increased in the presence of ruthenium red or uncoupler ( Fig. 2A). Under the experimental conditions used in Fig. 2A, the Ca 2ϩ release response to 100 nM IP 3 was increased from 8.6 Ϯ 1.8% under control conditions to 12.1 Ϯ 1.6% in the presence of ruthenium red (p Ͻ 0.01, n ϭ 4). This did not reflect a change in the size of the releasable ER Ca 2ϩ store, because there was no significant difference in the extent of Ca 2ϩ release in response to maximal IP 3 ( Fig. 2A; 25.3 Ϯ 3.5 and 25.1 Ϯ 2.8% in the absence and presence of ruthenium red, respectively; n ϭ 4).
We hypothesized that the paradoxical increased efficacy of submaximal IP 3 to release Ca 2ϩ when the mitochondria are no longer available to act as a sink for this released Ca 2ϩ reflects the feedback effects of Ca 2ϩ on the IP 3 R. To examine this possibility, we repeated the experiments of Fig. 2A (Fig. 2B). Alternatively, this may be explained by a decrease in IP 3 sensitivity due to the pharmacological effect of BAPTA (35). More importantly, the potentiation by mitochondrial inhibitors at submaximal IP 3 was completely eliminated when the Ca-BAPTA buffer was included. Thus, mitochondrial Ca 2ϩ uptake in the immediate vicinity of the IP 3 -activated Ca 2ϩ release sites can suppress the positive feedback effects of released Ca 2ϩ that would otherwise facilitate activation of neighboring IP 3 Rs.
The data of Fig. 2 were averaged over a number of cells in the imaging field. However, because mitochondria show a perinuclear distribution in individual hepatocytes, it might be expected that the modulation of IP 3 -induced Ca 2ϩ release would occur heterogeneously at the subcellular level. The confocal image of Fig. 3A (panel i) shows that the entire reticular network is labeled with compartmentalized fluo3FF in permeabilized hepatocytes. The decrease of [Ca 2ϩ ] ER in response to maximal IP 3 occurred homogeneously throughout each cell, apart from the nuclear matrix, as shown by the difference image of Fig. 3A (panel ii). By contrast, subsequent staining of the mitochondria with the pH-sensitive dye fluorescein diacetate revealed the more centralized mitochondrial distribution (Fig. 3A, panel iii). Thus, although the IP 3 -sensitive Ca 2ϩ store appears to be distributed throughout the hepatocyte, the modulation of IP 3 sensitivity by the mitochondria may occur predominately in the central domain of each cell. Evidence in support of this is shown in Fig. 3B, where the spatial pattern of [Ca 2ϩ ] ER decrease evoked by submaximal and maximal IP 3 is compared with the distribution of the mitochondria. Compartmentalized fura2FF was used to monitor [Ca 2ϩ ] ER (Fig. 3B,  panel i), and the mitochondria were localized functionally by their redox response to the mitochondrial substrate ␤-hydroxybutyrate (yellow overlay in Fig. 3B, panel ii). The functional mitochondria showed the same perinuclear distribution observed with other techniques in Figs. 1 and 3A. Addition of 100 nM IP 3 elicited a partial decrease of [Ca 2ϩ ] ER (purple overlays) in cells 1 and 2, and this response was larger in the peripheral regions than in the central domains where the mitochondria were located (compare panels ii and iii of Fig. 3B). By contrast, subsequent addition of maximal IP 3 elicited a larger decrease in [Ca 2ϩ ] ER in the mitochondria-rich domains of these cells (Fig. 3B, panel iv), which primarily reflects the prior depletion of peripheral [Ca 2ϩ ] ER by the submaximal IP 3 dose. Time courses of [Ca 2ϩ ] ER change in cells 1 and 2 are shown below the images of Fig. 3B (panels i-iv) for regions with high mitochondrial density (traces 1A and 1B) and for regions that were relatively deficient in mitochondria (traces 1B and 2B).
Similar differences in IP 3 sensitivity between regions with high and low mitochondrial density were observed in every cell in the imaging field, but because the responses were asynchronous they do not all show in the images. In addition, some cells gave [Ca 2ϩ ] ER oscillations and waves at submaximal IP 3 (26). For cell 3 of Fig. 3B, addition of 100 nM IP 3 did not cause an immediate Ca 2ϩ release. Instead, the [Ca 2ϩ ] ER decrease elicited in cell 2 propagated into cell 3 as a slow wave of Ca 2ϩ release (not shown). Significantly, the greatest magnitude and rate of [Ca 2ϩ ] ER decrease occurred in the distal part of cell 3, which was largely devoid of mitochondria. [Ca 2ϩ ] ER recovered in this oscillating cell and then after about 90 s in the continuing presence of 100 nM IP 3 there was a second wave of [Ca 2ϩ ] ER decrease that was intrinsic to cell 3. This intrinsic [Ca 2ϩ ] ER wave propagated from the mitochondrial-deficient region of the cell (Fig. 3B, panels v-vii). The suppression of IP 3 sensitivity in subcellular regions that were rich in mitochondria relative to other subcellular regions was observed in all experiments of the type shown in Fig. 3B.
To further evaluate the role mitochondrial Ca 2ϩ uptake in shaping the subcellular pattern of Ca 2ϩ release, the effects of mitochondrial inhibitors on the spatial distribution of [Ca 2ϩ ] ER decrease induced by IP 3 was examined in the permeabilized cells. Fig. 3C shows that the peripherial distribution of [Ca 2ϩ ] ER decrease observed during the first stimulation with submaximal IP 3 (Fig. 3C, panel i) was replaced by an essentially uniform response when the same cell was restimulated with the same dose of IP 3 in the presence of mitochondrial uncoupler. Consistent with the idea that mitochondrial Ca 2ϩ uptake suppresses IP 3 -mediated Ca 2ϩ mobilization in intact cells, the rate of propagation of global [Ca 2ϩ ] c waves evoked by the IP 3 -linked agonist vasopressin in intact hepatocytes was increased by 92 Ϯ 24% (n ϭ 7 cells, p Ͻ 0.005) when the cells were restimulated in the presence of uncoupler (1799ϩ oligomycin, 5 g/ml each). By contrast, oligomycin alone had no significant effect on vasopressin-induced [Ca 2ϩ ] c waves (27 Ϯ 18% of control, n ϭ 5).
Taken together the findings described above demonstrate that the mitochondrial modulation of IP 3 -induced Ca 2ϩ release is limited to those elements of the ER Ca 2ϩ stores in proximity with the mitochondria. As a result, the distribution of mitochondria establishes spatial heterogeneity in IP 3 sensitivity, such that regions lacking mitochondria are most likely to respond first and/or with a greater amplitude of [Ca 2ϩ ] ER release. Thus, the major finding of the present study is that mitochondrial Ca 2ϩ uptake exerts strong control over local Ca 2ϩ feedback regulation of IP 3 receptors. This occurs because the mitochondria rapidly sequester a fraction of the released Ca 2ϩ , which presumably suppresses the positive feedback effects of this Ca 2ϩ on neighboring IP 3 receptors. Because this positive feedback is a key component of the mechanisms responsible for the initiation and propagation of [Ca 2ϩ ] c waves, the mitochondria can play a key role in orchestrating the subcellular pattern of [Ca 2ϩ ] c signaling.
Mitochondrial Ca 2ϩ uptake following IP 3 -induced Ca 2ϩ release appears to be driven by the relatively large rapid changes in [Ca 2ϩ ] c and the privileged access of the mitochondria to IP 3 R Ca 2ϩ release sites in closely apposed regions of the ER (10 -15 (15). These frequency-modulated [Ca 2ϩ ] m oscillations establish dynamic control of mitochondrial energy metabolism (15,32). Although mitochondrial Ca 2ϩ uptake clearly serves to transduce [Ca 2ϩ ] c signals from the cytosol to regulate Ca 2ϩ -dependent processes in the mitochondrial matrix (15,31), it is also becoming apparent that mitochondria modulate cytosolic Ca 2ϩ signaling (20 -25). The simplest way in which the mitochondrial Ca 2ϩ transport pathways can modify [Ca 2ϩ ] c signals is by acting as a slow buffer that accumulates Ca 2ϩ during rapid [Ca 2ϩ ] c increases and then returns the Ca 2ϩ as [Ca 2ϩ ] c declines. In this way the mitochondria can blunt and prolong a [Ca 2ϩ ] c transient, as occurs during depolarization-induced Ca 2ϩ influx in chromaffin cells (22). However, the present study demonstrates that mitochondria can also directly regulate the Ca 2ϩ release function of the IP 3 R in the ER by modulating the feedback effects of cytosolic Ca 2ϩ . This process could account for the observation that mitochondrial energization in Xenopus oocytes enhances the organization of IP 3 -activated [Ca 2ϩ ] c waves by decreasing frequency and increasing the am-plitude of Ca 2ϩ release (20). Specifically, mitochondrial suppression of the positive feedback effects of [Ca 2ϩ ] c should reduce the excitability of the system. This stabilization of the basal state would lower Ca 2ϩ wave frequency and ensure that a greater proportion of IP 3 Rs are in the resting state available to contribute to Ca 2ϩ release when the activation threshold is finally achieved at the Ca 2ϩ wave front. A different picture has emerged in oligodendrocytes, where mitochondria appear to be selectively localized at sites of Ca 2ϩ wave amplification (23,28). This could reflect a role for mitochondrial Ca 2ϩ -induced Ca 2ϩ release, whereby the accumulation of [Ca 2ϩ ] m elicits mitochondrial depolarization and consequent Ca 2ϩ release (24). However, the mechanism described in the present work could also operate in this system, but instead of suppressing positive feedback effects of [Ca 2ϩ ] c , the spatial and temporal properties of mitochondrial Ca 2ϩ uptake in the oligodendrocyte may act predominantly to suppress the negative feedback effects of [Ca 2ϩ ] c .
Overall, it appears that mitochondria can have a number of important effects on cytosolic Ca 2ϩ signaling. These effects are not limited to simple Ca 2ϩ buffering but include direct modulation of the feedback effects of [Ca 2ϩ ] c on its own release. In addition to shaping the temporal and spatial pattern of [Ca 2ϩ ] c transients, the suppression of IP 3 sensitivity by mitochondria may also play a role in stabilizing basal [Ca 2ϩ ] c . This function of the mitochondria in setting the threshold for [Ca 2ϩ ] c spikes, together with the effects on spatial organization and signal amplification can all contribute to enhance the fidelity of Ca 2ϩ signaling.