Modulation of [Ca2+]i Signaling Dynamics and Metabolism by Perinuclear Mitochondria in Mouse Parotid Acinar Cells*

Parotid acinar cells exhibit rapid cytosolic calcium signals ([Ca2+]i) that initiate in the apical region but rapidly become global in nature. These characteristic [Ca2+]i signals are important for effective fluid secretion, which critically depends on a synchronized activation of spatially separated ion fluxes. Apically restricted [Ca2+]i signals were never observed in parotid acinar cells. This is in marked contrast to the related pancreatic acinar cells, where the distribution of mitochondria has been suggested to contribute to restricting [Ca2+]i signals to the apical region. Therefore, the aim of this study was to determine the mitochondrial distribution and the role of mitochondrial Ca2+ uptake in shaping the spatial and temporal properties of [Ca2+]i signaling in parotid acinar cells. Confocal imaging of cells stained with MitoTracker dyes (MitoTracker Green FM or MitoTracker CMXRos) and SYTO dyes (SYTO-16 and SYTO-61) revealed that a majority of mitochondria is localized around the nucleus. Carbachol (CCh) and caged inositol 1,4,5-trisphosphate-evoked [Ca2+]i signals were delayed as they propagated through the nucleus. This delay in the CCh-evoked nuclear [Ca2+]i signal was abolished by inhibition of mitochondrial Ca2+ uptake with ruthenium red and Ru360. Likewise, simultaneous measurement of [Ca2+]i with mitochondrial [Ca2+] ([Ca2+]m), using fura-2 and rhod-FF, respectively, revealed that mitochondrial Ca2+ uptake was also inhibited by ruthenium red and Ru360. Finally, at concentrations of agonist that evoke [Ca2+]i oscillations, mitochondrial Ca2+ uptake, and a nuclear [Ca2+] delay, CCh also evoked a substantial increase in NADH autofluorescence. This autofluorescence exhibited a predominant perinuclear localization that was also sensitive to mitochondrial inhibitors. These data provide evidence that perinuclear mitochondria and mitochondrial Ca2+ uptake may differentially shape nuclear [Ca2+] signals but more importantly drive mitochondrial metabolism to generate ATP close to the nucleus. These effects may profoundly affect a variety of nuclear processes in parotid acinar cells while facilitating efficient fluid secretion.

Regulation of both the spatial and temporal properties of intracellular Ca 2ϩ ([Ca 2ϩ ] i ) 1 signals is known to underlie the specificity of stimulus-response coupling in a variety of cell types (1,2). For example, gene transcription in T cells can be differentially regulated by the frequency of [Ca 2ϩ ] i oscillations (3), and in vascular smooth muscle cells global Ca 2ϩ signals control contraction, whereas localized Ca 2ϩ release (Ca 2ϩ sparks) causes vasodilatation (4). These examples highlight the importance of understanding the mechanisms that underlie the "shaping" of [Ca 2ϩ ] i signals and how ultimately different patterns of [Ca 2ϩ ] i signals can differentially activate physiological end points.
By comparing the spatial and temporal kinetics of [Ca 2ϩ ] i signaling in the morphologically and functionally related exocrine acinar cells of the pancreas and salivary glands, we have demonstrated previously (5) how different [Ca 2ϩ ] i signals may be tuned to evoke specific physiological responses in these cells. The major function of pancreatic acinar cells is the exocytosis of zymogen granules that can be activated by apically confined [Ca 2ϩ ] i signals (6), suggested to be the major physiological [Ca 2ϩ ] i signal evoked by threshold agonist concentrations (7). These apically confined [Ca 2ϩ ] i signals are in part because of the distribution of mitochondria, which form a belt around the apically located zymogen granules (8,9). These perigranular mitochondria likely serve two main functions. First, mitochondrial Ca 2ϩ uptake provides a buffer barrier thereby preventing the spread of [Ca 2ϩ ] i waves to the basal part of the cell (8 -10). Second, the Ca 2ϩ taken up by mitochondria drives ATP production locally for the energy-consuming exocytotic process (11). The major function of parotid acinar cells is fluid secretion, and this is most effectively activated by rapid global [Ca 2ϩ ] i signals. We proposed that rapid global [Ca 2ϩ ] i signals facilitate the almost simultaneous activation of Ca 2ϩ -dependent Cl Ϫ channels on the apical membrane and Ca 2ϩ -dependent K ϩ channels on the basolateral membrane (5). The latter maintain the membrane potential at hyperpolarizing potentials and the driving force for Cl Ϫ efflux, thereby ensuring efficient ion and water movement (12). Interestingly, even at threshold stimulation parotid acinar cells exhibit rapid global [Ca 2ϩ ] i signals, and apically confined [Ca 2ϩ ] i signals were never observed (5). Given the many similarities between parotid and pancreatic acinar cells, the differences in the spatial and temporal patterns of [Ca 2ϩ ] i signaling between these two cell types are difficult to reconcile if they have a similar distribution of mitochondria. Therefore, the aim of the present study was to determine the mitochondrial distribution and the role of mitochondrial Ca 2ϩ uptake in shaping [Ca 2ϩ ] i signals in parotid acinar cells. The study revealed for the first time that there was an absence of any definitive perigranular mitochondria in parotid cells compared with that observed in pancreatic acinar cells. The specific mitochondrial distribution in parotid cells therefore likely facilitates the rapid, global [Ca 2ϩ ] i signals observed. In contrast, the mitochondria are localized mainly to a region surrounding the nucleus in parotid acinar cells. In addition, mitochondrial Ca 2ϩ uptake in this region differentially shapes nuclear [Ca 2ϩ ] signals and enhances metabolism and thus ATP generation close to the nucleus. This may serve to differentially modulate a variety of processes within the nucleus.

EXPERIMENTAL PROCEDURES
Isolation of Parotid Acinar Cells-Single cells or small clusters of parotid acinar cells were isolated from freshly dissected parotid glands from wild type Swiss Black mice by sequential digestion with (single cells) or without (cell clusters) trypsin followed by collagenase as described previously (5,13). Following isolation, cells were resuspended in 1% bovine serum albumin containing basal modified Eagle's media supplemented with 2 mM glutamine, penicillin/streptomycin and incubated at 37°C and gassed with 5% CO 2 and 95% O 2 until ready for use. Prior to experimentation, aliquots of cell suspensions were loaded with the appropriate fluorescent dyes in a HEPES-buffered physiological saline solution (HEPES/PSS) containing the following (in mM): 5.5 glucose, 137 NaCl, 0.56 MgCl 2 , 4.7 KCl, 1 Na 2 HPO 4 , 10 HEPES, pH 7.4, 1.2 CaCl 2 , after which they were resuspended in HEPES-PSS and kept at 4°C until ready for use. For measurement of [Ca 2ϩ ] i , cells were loaded with 2 M fura-2/AM for 30 min at room temperature. For the simultaneous measurement of cytosolic and mitochondrial [Ca 2ϩ ], cells that had been pre-loaded with fura-2 were then loaded with 10 M dihydro-rhod-FF/AM for 10 min at 4°C. Dihydro-rhod-FF was formed by reacting 10 l of 1 mg/ml NaBH 4 with 40 l of 1 mM rhod-FF stock solution. In some experiments (to confirm mitochondrial Ca 2ϩ uptake) cells were simultaneously loaded with 5 M rhod-FF and 0.5 M Mito-Tracker Green FM for 10 min at 4°C. Pharmacological inhibition of mitochondrial Ca 2ϩ uptake was achieved by pre-incubating cells with 30 M ruthenium red (RuRed) (Sigma) (14) or 10 M Ru360 (Calbiochem) (15) for 30 min at room temperature. Stock solutions of Ru360 (1 mM) were made up in deoxygenated water immediately prior to use.
Digital Imaging of Cytosolic and Mitochondrial [Ca 2ϩ ]-Cells were allowed to adhere to a glass coverslip and continually perfused with HEPES-PSS using a gravity-fed perfusion system and automatic valves for rapid switching of solutions (Warner Instruments Inc., Hamden, CT). All [Ca 2ϩ ] imaging experiments were performed using an inverted epifluorescence Nikon microscope with a 40ϫ oil immersion objective lens (numerical aperture, 1.3) and a digital imaging system consisting of a TILL polychrome IV monochromator illumination system, a Sensicam 12-bit charged coupled device camera, and TILL VisION acquisition and analysis software as described previously (16). To achieve maximum spatial and temporal resolution of CCh-evoked [Ca 2ϩ ] i signals in the apical, nuclear, and non-nuclear basal regions, fluorescent images were captured with no binning at rates of 12-16 Hz, depending on image size, with 5-ms exposure times for both 340 and 380 nm images. 340/380 ratio images were calculated offline. For the simultaneous measurement of cytosolic and mitochondrial [Ca 2ϩ ], fura-2-and rhod-FF-loaded cells were excited alternately with light at 340 (10 ms), 380 (10 ms), and 550 nm (50 ms). Fura-2-and rhod-FF-emitted fluorescence was collected sequentially through a fura-2-rhodamine dual emission dichroic beamsplitter with dual emission bandpass filter (510 Ϯ 15 and 600 Ϯ 25 nm) (Chroma Technology Corp., Rockingham, VT). Sequential images were acquired at a rate of 1 Hz, and changes in mitochondrial [Ca 2ϩ ] were represented by ⌬F/F 0 ratio; where ⌬F/F 0 ϭ 100((F Ϫ F 0 )/F 0 ), F is the recorded fluorescence (in levels of gray), and F 0 is the average starting fluorescence recorded over the initial 10 frames of the image sequence. In some experiments mitochondrial Ca 2ϩ uptake was confirmed by dual loading cells with MitoTracker Green FM and rhodFF. These cells were excited with light at 488 Ϯ 15 (10 ms) and 550 Ϯ 25 nm (50 ms), and the emitted fluorescence from both dyes was collected through a fluorescein isothiocyanate/TRITC dual emission di-chroic beamsplitter and bandpass filter (520 Ϯ 15 and 600 Ϯ 25 nm, Chroma). All experiments were performed at room temperature.
Flash Photolysis of Caged InsP 3 -For measurement of [Ca 2ϩ ] i during flash photolysis of caged InsP 3 , single parotid acinar cells were loaded with 75 M Oregon Green 488 BAPTA 2 (OGB2), which was introduced to the cell interior by means of diffusion through a patch pipette similar to that described previously (5). The standard intracellular pipette solution contained the following (in mM): 140 CsCl, 10 HEPES-Tris, 1.5 MgCl 2 , 3 MgATP, 1 N-(hydroxyethyl)ethylenediaminetriacetic acid, 0.075 OGB2, and 0.003 D-myo-InsP 3 P-4(5)-1-(2-nitrophenyl)-ethyl ester (caged InsP 3 ), pH 7.3. After acquiring the whole cell configuration, 4 min were allowed for equilibration of dye with the cytosol, after which cells were excited at 488 Ϯ 15 nm, and emitted fluorescence was collected. Flash photolysis of caged InsP 3 was achieved using a pulsed xenon arc lamp (TILL Photonics, Eugene, OR) to provide a 0.5-ms flash of UV light (360 Ϯ 7.5 nm). Sequential images were acquired at a rate of 1 Hz, and changes in [Ca 2ϩ ] i following flash photolysis of caged InsP 3 were represented by ⌬F/F 0 ratio of OGB2 fluorescence.
Confocal Imaging of Mitochondria and Nuclei-To assess the distribution of mitochondria and nuclei, clusters of parotid acinar cells were simultaneously loaded with either 50 nM MitoTracker Red CMXRos (Molecular Probes, OR) and 1 M SYTO-16 (Molecular Probes), respectively, for 2 min at room temperature or 0.5 M MitoTracker Green FM (Molecular Probes) and 1 M SYTO-61 (Molecular Probes), respectively, for 10 min at 37°C. Images were acquired using a Nikon C1 visible light laser scanning confocal microscope with a 40ϫ (1.3 NA) and 100ϫ (1.4 NA) oil immersion objective either simultaneously or sequentially for each dye. MitoTracker Green FM and SYTO-16 were excited by the 488-nm argon laser line, and emission was collected through a 543-nm bandpass filter and photomultiplier tube. MitoTracker Red CMXRos and SYTO-61 were excited by the 543-and 633-nm lines, respectively, from He-Ne lasers, and emission was collected through a 585 Ϯ 30-and a 670 Ϯ 20-nm bandpass filter and photomultiplier tube, respectively. Image stacks in the Z dimension with 0.5-0.8-m step size were acquired using an automated focus. Volume-rendered three-dimensional reconstructed images were then calculated offline by the Nikon EZ-C1 analysis software using a maximum projection algorithm.
Digital Imaging of NADH Autofluorescence-By using the conventional epifluorescence imaging system described above, unloaded cells were excited with light at 350 nm with an exposure of 100 ms, and NADH autofluorescence was collected through a 400-nm dichroic and 450 Ϯ 45-nm bandpass filter. Sequential images were acquired at a rate of 1 Hz, and changes in NADH autofluorescence were represented by ⌬F/F 0 ratio.
Data Analysis and Experimental Design-Due to the nature of most experiments, an unpaired experimental design was applied (unless otherwise stated in the text), whereby statistical significance was determined between control cells and cells pre-treated with mitochondrial inhibitors using an unpaired t test or Mann-Whitney test. For any parameter analyzed from several cells in a particular experiment, an average value was determined. These values were in turn averaged to give the values expressed in the text as means Ϯ S.E.

CCh-and InsP 3 -evoked Ca 2ϩ Signals Are Significantly Slower in the Perinuclear Region of Parotid Acinar Cells-We
have shown previously (5), by using flash photolysis of caged InsP 3 , that Ca 2ϩ release initiates in the apical region but very rapidly becomes global in nature even at threshold concentrations of InsP 3 . Most important, the apically confined [Ca 2ϩ ] i signals typical of pancreatic acinar cells were never observed in parotid acinar cells (5). In these studies, however, it was also noted that in ϳ35% of cells (8 of 23 cells), there was a region in the basal part of the cell where the increase in [Ca 2ϩ ] i was smaller and exhibited slower kinetics than other regions of the cell (Fig. 1A). We hypothesized that this region corresponded to the nucleus and that this anomalous [Ca 2ϩ ] i change would therefore only be observed where the focal plane imaged passed through the nucleus.
Several reports have suggested that differential [Ca 2ϩ ] signals in the nucleus can be attributed to nuclear dye artifacts that are particularly prevalent with single wavelength fluorescent dyes such as OGB2 (17). This is because the affinity of such dyes for Ca 2ϩ may be affected by the different protein, ionic, pH, and osmotic environment in the nucleus compared with the cytosol. In an attempt to minimize potential dye artifacts and mimic a more physiological stimulus, intact parotid acinar cells were loaded with the ratiometric dye, fura-2, and stimulated with 300 nM CCh to track the spatiotemporal [Ca 2ϩ ] i signals in a similar manner to Fig. 1A. Fig. 1, B and C, shows a bright field image, a SYTO-16-fluorescent image, pseudocolor-enhanced fluorescent ratio images ( Fig. 1, Bi, panels A-H and Ci, panels A-H), and the corresponding kinetic profiles of CCh-evoked [Ca 2ϩ ] i signals (Fig. 1, Bii and Cii) in the apical (blue), non-nuclear basal (green), and the nuclear region (red). The location of the nucleus was confirmed by staining cells with 1 M SYTO-16 at the end of each experiment. Because [Ca 2ϩ ] signals rapidly propagate from the apical region to the basal region of parotid acinar cells (5), boxes for analysis of the nuclear and non-nuclear basal region were strategically placed equidistant from the apical region to reveal any delay in the nucleus. Although these experiments were not  Table I). performed on a confocal microscope, care was taken to ensure that the focal plane was as close to the center of the nucleus as possible to avoid contamination from out-of-focus light coming from dye above and below the nucleus. Cells were discarded from analysis if the nucleus was out of focus. Fig. 1B shows an initial [Ca 2ϩ ] i signal evoked by 300 nM CCh in the apical, basal, and nucleus of a typical control untreated cell. These data show that the CCh-evoked [Ca 2ϩ ] i signal is significantly delayed in the nucleus compared with a corresponding non-nuclear basal region, thereby supporting preliminary data using flash photolysis of caged InsP 3 (Fig. 1A). The time to half-maximum response was 0.19 Ϯ 0.07 s slower in the basal region compared with the apical region, whereas the nuclear response was 0.47 Ϯ 0.08 s slower than the apical region ( Fig. 1B and Table I). This nuclear [Ca 2ϩ ] i delay was quantified by subtracting the time between the half-maximum increase in the apical and non-nuclear basal region (A-B) from the time between the half-maximum increase in the apical and nuclear basal region (A-N) in every cell (Table I). On average, the nuclear Ca 2ϩ delay was found to be 0.28 ؎ 0.1 s. The observed differences in the kinetics of CCh-evoked [Ca 2ϩ ] i signals in the nucleus versus the non-nuclear basal region could be due to differential fura-2 compartmentalization into intracellular organelles. One might predict that this would be more pronounced in the non-nuclear basal region compared with the nucleus due to the presence of endoplasmic reticulum. The average fluorescence signal from this region would therefore be affected by the "saturated" fura-2 signal emitted from the endoplasmic reticulum. To test this, a series of pilot experiments were performed whereby fura-2-loaded cells were permeabilized by perfusing with a "cytosol-like" solution containing 0.4 IU/ml streptolysin-O (BD Biosciences), identical to that described previously (13). Permeabilization caused a complete loss of cytosolic dye, as monitored by the fura-2 isosbestic 360-nm fluorescence signal (data not shown). Any residual fluorescence would most likely be due to dye trapped within the intracellular organelles. Under these conditions it was found that streptolysin-O treatment decreased fluorescence from 1166 Ϯ 110 gray levels above background to 70 Ϯ 12 gray levels above background (6 experiments, in which Ͼ4 cells were imaged). This was not significantly different from autofluorescence of non-loaded cells treated identically (fluorescence decreased from 63 Ϯ 7 to 49 Ϯ 8 gray levels above background; 6 experiments, in which Ͼ4 cells were imaged), confirming that fura-2 compartmentalization into intracellular organelles is unlikely to contribute to the nuclear [Ca 2ϩ ] delay.
The  (18,19), a derivative Ru360 is a more specific and potent inhibitor of mitochondrial Ca 2ϩ uptake (14,15). At the concentrations and pre-incubation times tested, both RuRed and Ru360 have been shown previously (20) to inhibit significantly mitochondrial Ca 2ϩ uptake in intact cells.
In the present study pre-incubation with either 30 M RuRed or 10 M Ru360 for 30 min prior to stimulation with CCh altered the kinetics of the CCh-evoked [Ca 2ϩ ] i increase ( Fig. 1C and Table I). Most important, the nuclear [Ca 2ϩ ] delay was significantly reduced from 0.28 Ϯ 0.1 s (n ϭ 6) to 0.02 Ϯ 0.01 s (n ϭ 4) by RuRed and to 0.03 Ϯ 0.03 s (n ϭ 6) by Ru360 (Fig. 1C and Table I). Further analysis showed that both RuRed and Ru360 significantly reduced the time to peak response in all regions of the cell and abolished any spatial differences in the time to peak values (Table I). Moreover, in control cells the magnitude of the CCh-evoked [Ca 2ϩ ] i response in the basal region (0.37 Ϯ 0.07 ratio units) and nucleus (0.39 Ϯ 0.08 ratio units) was significantly lower than the corresponding apical region (0.49 Ϯ 0.07 ratio units), although there was no significant difference between all three regions of RuRed-or Ru360treated cells (Table I). These data indicate that both the magnitude (in the basal and nuclear regions) and rate of Ca 2ϩ release (in all cellular regions) are markedly increased by inhibition of mitochondrial Ca 2ϩ uptake. These data are consistent with previous studies (21) suggesting that mitochondrial Ca 2ϩ uptake suppresses the local feedback activation of Ca 2ϩ release by buffering local [Ca 2ϩ ]. One might also predict, from these data, that mitochondrial Ca 2ϩ uptake shapes [Ca 2ϩ ] i signals in all regions of the cell consistent with a homogeneous distribution of mitochondria. However, upon closer examination, the effect of mitochondrial Ca 2ϩ uptake inhibitors was most striking in the nucleus. This is therefore consistent with perinuclear mitochondria providing a buffer barrier around the nucleus that impedes the propagation or diffusion of [Ca 2ϩ ] i signals to the nucleus. In addition, the effects of RuRed and Ru360 suggest that the nuclear [Ca 2ϩ ] delay is largely independent of any nuclear dye artifacts. Perhaps more importantly, because the affinity of most dyes is reportedly higher in the nucleus (17), the apparent magnitude of the nuclear [Ca 2ϩ ] signals may be significantly overestimated.
Parotid Acinar Cell Mitochondria Exhibit a Predominantly Perinuclear Distribution-To investigate whether this anomalous nuclear [Ca 2ϩ ] i change is due to a specific distribution of mitochondria, we used conventional epifluorescence imaging ( Fig. 2A) and confocal imaging in the Z dimension, to obtain three-dimensional reconstructed images (Fig. 2B) of parotid acinar cell clusters co-stained with mitochondrial and nuclear staining dyes. Fig. 2A shows a conventional epifluorescence image of cells loaded with MitoTracker Red (CMXRos), which accumulates in the mitochondrial matrix, and SYTO-16, which labels nuclei by binding to DNA. The mitochondria appear to be distributed as "ring-like" structures around the nucleus. However, due to the much higher signal from SYTO-16 compared with MitoTracker Red, it is difficult to simultaneously image both mitochondria and nuclei in the same cells and accord any three-dimensional association or colocalization of these two organelles. Therefore, we used volume-rendered three-dimensional confocal imaging of cells simultaneously loaded with MitoTracker Green FM and SYTO-61. By using this technique, the most striking observation was that there was a lack of any obvious perigranular mitochondria in parotid acinar cells compared with that observed in pancreatic acinar cells. In addition, although mitochondria appear in other regions of the cell, there was a clear population of mitochondria that had a perinuclear distribution. These mitochondria appear to form a basket-like structure or cavity within which the nucleus sits (Fig. 2B). Each cavity appears to be lined with mitochondria consisting of "finger-like" projections that wrap around the nucleus (see Z stack images, Fig. 2B Ru360 or RuRed is due to inhibition of mitochondrial Ca 2ϩ uptake, cells were loaded with both fura-2 and rhod-FF (or rhod-2) to monitor simultaneously cytosolic and mitochondrial [Ca 2ϩ ]. This produced a highly punctate distribution of dye reminiscent of the mitochondrial accumulation of MitoTracker dyes. However, due to the high affinity of rhod-2 for Ca 2ϩ (K d , ϳ195 nM (22)), any residual cytosolic rhod-2 may detect cytosolic [Ca 2ϩ ] changes, possibly masking mitochondrial changes. We therefore chose rhod-FF as this has a much lower affinity for Ca 2ϩ (K d , ϳ19 M (23)) and is unlikely to detect cytosolic [Ca 2ϩ ] changes in the face of small amounts of residual cytosolic dye.
Control cells loaded with both fura-2 and rhod-FF showed that CCh evoked a substantial increase in rhod-FF fluorescence (ϳ7.98 Ϯ 0.98% ⌬F/F 0 , Fig. 3, A and C, mean data), which occurred with slower kinetics than the rapid cytosolic [Ca 2ϩ ] increase ( Fig. 3A; n ϭ 6, 24 cells). This suggests that the CCh-evoked increase in [Ca 2ϩ ] i promotes mitochondrial Ca 2ϩ uptake, as reported in other non-excitable cell types (22,24,25). Although the changes in rhod-FF fluorescence are uncalibrated signals, the low affinity of rhod-FF for Ca 2ϩ suggests that these changes likely represent changes in mitochondrial [Ca 2ϩ ] (several M) that are much larger than those observed in the cytosol (Յ1 M). This is not surprising because several studies (using both fluorescent dyes and targeted aequorin) have demonstrated that mitochondrial [Ca 2ϩ ] can reach such levels (25,26). It was also noted that the rate of decrease in rhod-FF fluorescence following the removal of CCh was much slower than the rate of increase (see Fig. 3A). This could be due to the slow activity of the mitochondrial Na ϩ /Ca 2ϩ -exchanger, the major pathway for mitochondrial Ca 2ϩ efflux, as has previously been reported in other cells (27). Following stimulation with CCh, treatment of cells with 0.5 M FCCP to depolarize the mitochondria caused the punctate rhod-FF fluorescence to decrease and become more diffuse (Fig. 3Ai, panel C) with a corresponding increase in [Ca 2ϩ ] i (Fig. 3Aii). This suggests that the dye accumulates in mitochondria and that the mitochondria take up Ca 2ϩ during stimulation with CCh (27).
Because stimulation of parotid acinar cells with CCh causes profound cell shrinkage, the slower rate of increase in rhod-FF fluorescence could be due to cell volume changes or mitochondrial volume changes that would be exacerbated using a single wavelength dye such as rhod-FF. To test for this, cells were simultaneously loaded with rhod-FF and MitoTracker Green FM. Both these dyes accumulate in mitochondria, but rhod-FF is Ca 2ϩ -sensitive, whereas MitoTracker Green FM is Ca 2ϩinsensitive. Changes in the ratio of rhod-FF versus Mito-Tracker Green FM fluorescence (550/488 ratio) therefore represent mitochondrial [Ca 2ϩ ] changes corrected for cell and/or mitochondrial volume changes or mitochondrial movement. Using this strategy, stimulation with CCh evoked changes in the 550/488 ratio with similar kinetics to the raw rhod-FF fluorescence signal (data not shown), confirming that that this was due to mitochondrial [Ca 2ϩ ] changes rather than changes in cell volume.
Pre-treatment of Cells with Ru360 or RuRed Prevent Mitochondrial Ca 2ϩ Uptake-The CCh-evoked increase in rhod-FF fluorescence (representative trace, Fig. 3A, and mean control %⌬F/F 0 , 7.98 Ϯ 0.98, Fig. 3C) was largely inhibited when cells were pre-incubated with 10 M Ru360 (Fig. 3, Bii and C; %⌬F/ F 0 , 1.07 Ϯ 0.47) or 30 M RuRed (%⌬F/F 0 , 1.40 Ϯ 0.50, Fig. 3C). In addition, the FCCP-evoked decrease in rhod-FF fluorescence and the corresponding increase in [Ca 2ϩ ] i was also largely abrogated (Fig. 3Bii). These data therefore provide evidence that the increase in rhod-FF fluorescence is due to mitochondrial Ca 2ϩ uptake and that the subsequent addition of FCCP promotes mitochondrial Ca 2ϩ efflux. Collectively these data provide convincing evidence that the CCh-evoked increase in [Ca 2ϩ ] i promotes perinuclear mitochondrial Ca 2ϩ uptake which in turn slows the increase in nuclear [Ca 2ϩ ].
CCh Increases NADH Autofluorescence-In addition to shaping nuclear Ca 2ϩ signals, perinuclear mitochondrial Ca 2ϩ uptake is likely to drive mitochondrial metabolism and the generation of ATP by activation of mitochondrial dehydrogenases (28,29). To test this, we measured NADH autofluorescence, a technique employed extensively to monitor mitochondrial metabolism (30,31). Fluorescent NADH is generated during the Krebs cycle from NAD ϩ (non-fluorescent) where it feeds into the electron transport chain to drive ATP synthesis. Moreover, mitochondrial Ca 2ϩ activates three dehydrogenases of the Krebs cycle (28,29); thus NADH autofluorescence represents a measure of Ca 2ϩ -dependent mitochondrial metabolism and presumably ATP generation. NADH autofluorescence was monitored by exciting parotid acinar cells with light at 350 nm and measuring the emitted light at 450 nm. In Fig. 4A cells were excited at 350 nm (for 1 s exposure) before and during treatment with 300 nM CCh or 0.5 M FCCP. Continuous NADH autofluorescence was not monitored using these exposure times due to profound photobleaching and possible photo-toxicity. However, this strategy clearly illustrates that NADH autofluorescence has a similar spatial distribution to perinuclear mitochondria. Reducing the exposure time to 100 ms and acquiring every second, NADH autofluorescence was continuously monitored during treatment with CCh or FCCP (Fig. 4, B and C). The kinetic profile of the relative change in autofluorescence (%⌬F/F 0 , see Fig. 4C) and the corresponding fluorescent images (Fig. 4B) clearly demonstrate that CCh increases and FCCP decreases autofluo-rescence with similar kinetics to mitochondrial [Ca 2ϩ ] shown in Fig. 3Aii. Finally, the CCh-evoked increase in NADH autofluorescence was significantly inhibited by pre-incubation with either 10 M Ru360 (11.28 Ϯ 0.80 to 2.50 Ϯ 0.76, Fig. 4, C and E) or 30 M RuRed (11.28 Ϯ 0.80 to 2.73 Ϯ 0.53, Fig. 4E) and by transient treatment with 0.5 M FCCP (11.28 Ϯ 0.80 to 1.00 Ϯ 0.24, Fig. 4, D and E). These data collectively suggest that perinuclear mitochondrial Ca 2ϩ uptake directly drives mitochondrial metabolism and thus likely generates elevated levels of ATP close to the nucleus. DISCUSSION The primary function of parotid acinar cells is to secrete copious amounts of salivary fluid. This is achieved by the almost simultaneous activation of Ca 2ϩ -dependent Cl Ϫ channels on the apical membrane and Ca 2ϩ -dependent K ϩ channels on the basolateral membrane (5). The latter maintain the membrane potential at hyperpolarizing potentials and the driving force for Cl Ϫ efflux thereby ensuring efficient ion and water movement (12). Parotid acinar cells exhibit rapid, global [Ca 2ϩ ] i signals that are important for the almost simultaneous activation of these spatially separated ion fluxes (5). Apically confined [Ca 2 ] i signals, which are commonly observed in pancreatic acinar cells and are due, in part, to the localization of active perigranular mitochondria (8 -10), were never observed in parotid acinar cells (5). In the face of these observations the following important questions are raised. What is the mitochondrial distribution in parotid acinar cells? What function does mitochondrial Ca 2ϩ uptake serve? Data from the present study revealed an absence of perigranular mitochondria in parotid acinar cells. This likely facilitates the rapid, global [Ca 2ϩ ] i signals that are important for effective fluid secretion and helps to reconcile some of the spatial and temporal differences in [Ca 2ϩ ] i signaling observed between parotid and pancreatic acinar cells (5). Instead, it was found that a large proportion of mitochondria exhibited a perinuclear distribution in parotid acinar cells, and that this resulted in both CCh-and InsP-evoked [Ca 2ϩ ] i signals being significantly delayed in the nucleus of these cells.
Differential regulation of nuclear [Ca 2ϩ ] signaling or "dampened" nuclear [Ca 2ϩ ] signals have been observed in a variety of cells under different experimental conditions (32)(33)(34)(35). Several reports have suggested that these observations are due to modulation of the nuclear pore complex (36,37), differential nuclear Ca 2ϩ buffering (35,38), or even nuclear dye artifacts (17). Although we cannot completely rule out these effects, an important observation from the present study is that inhibition of mitochondrial Ca 2ϩ uptake completely abolishes the nuclear [Ca 2ϩ ] delay.
It has been demonstrated that pancreatic acinar cells contain a minor population of mitochondria that are also localized to the perinuclear region and that have been suggested to give rise to an observed differential nuclear [Ca 2ϩ ] signaling (10,32). In particular, it was shown that perinuclear mitochondria take up Ca 2ϩ specifically when Ca 2ϩ is uncaged locally in the nucleus (10). These elegant experiments suggested that perinuclear mitochondria could play a role in preventing [Ca 2ϩ ] i signals from invading the nucleus or confine [Ca 2ϩ ] signals to the nucleus that originates there (39). However, these studies did not specifically demonstrate functional perinuclear mitochondrial Ca 2ϩ uptake during agonist-evoked [Ca 2ϩ ] i signaling, but rather emphasized a greater functional role of perigranular mitochondria in shaping the [Ca 2ϩ ] i changes in pancreatic acinar cells (9,10).
Differential shaping of nuclear [Ca 2ϩ ] signaling by perinuclear mitochondrial Ca 2ϩ uptake has the potential to generate diverse transcriptional responses and thus profoundly affect cell function and cell fate. This results from the presence of many Ca 2ϩ -dependent effectors in the nucleus, such as the transcriptional repressor DREAM (40), Ca 2ϩ /CaM-dependent kinases (41), as well as CREB-dependent transcription (42). It is, however, important to note that the current data show that CCh-evoked [Ca 2ϩ ] i signals are on average only ϳ0.3 s slower in the nucleus than in the corresponding non-nuclear basal region. In addition, there was no significant difference in the magnitude of the observed nuclear and basal [Ca 2ϩ ] i response. Given this, it seems unlikely that any currently known Ca 2ϩdependent effectors could sufficiently decode such subtly different [Ca 2ϩ ] i signals.
Alternatively, the observed nuclear [Ca 2ϩ ] delay may simply be a consequence of the pronounced perinuclear mitochondrial Ca 2ϩ uptake and, rather than specifically shaping nuclear [Ca 2ϩ ] signals, perinuclear mitochondria may serve to generate ATP close to the nucleus for the plethora of ATP-dependent nuclear effectors. The present study clearly demonstrated that CCh substantially increases NADH autofluorescence and that this was sensitive to inhibitors of mitochondrial Ca 2ϩ uptake (RuRed and Ru360) or mitochondrial uncouplers (FCCP). Due to the slow kinetics of mitochondrial Ca 2ϩ uptake and Ca 2ϩ efflux (27), mitochondria have the capacity to decode oscillations in [Ca 2ϩ ] i into efficient metabolism and ATP production (24,30,43). For example, NADH autofluorescence has been shown to oscillate during CCK-evoked [Ca 2ϩ ] i oscillations in pancreatic acinar cells (30). Interestingly, as the frequency of [Ca 2ϩ ] i oscillations increases the mitochondrial NADH responses fuse into an elevated plateau. Similar results had been observed previously in hepatocytes (24). In the present study, CCh at concentrations that produce [Ca 2ϩ ] i oscillations evoked a substantial sustained increase in NADH, and oscillations in the NADH response were never observed. This mechanism is clearly designed to match increased energy demand to increased ATP supply, which appears to be highly efficient in parotid acinar cells particularly in the perinuclear region.
There are several important ATP-dependent effects in the nucleus, including regulation of kinases (41), chromatin remodeling (44), DNA replication (45), and cell cycle control (46). In addition, ATP is required for a variety of ion transporters and FIG. 4. CCh increases NADH autofluorescence indicative of enhanced mitochondrial metabolism and localized ATP generation. NADH autofluorescence was monitored in unloaded parotid acinar cells by exciting at 350 nm light and measuring the emitted light at 450 nm. A illustrates that NADH autofluorescence exhibits a similar perinuclear distribution to mitochondria (1-s exposure) before and after stimulation with CCh and FCCP. B, continuous monitoring of NADH autofluorescence (100-ms exposure every second) showing bright field and NADH fluorescent images (Bi) and corresponding kinetic profile of the relative (%⌬F/F 0 ) change in autofluorescence (Bii, n ϭ 9). This shows that CCh increases and FCCP decreases fluorescence suggesting that NADH autofluorescence was due to changes in mitochondrial metabolism. C-E, representative traces (C and D) and mean data (E) showing that the CCh-evoked increase in NADH autofluorescence was inhibited by pre-incubation of cells with 10 M Ru360 (C, n ϭ 7 and mean data in E) or 30 M ruthenium red (RuRed) (mean data in E, n ϭ 6) or by short term treatment with 0.5 M FCCP (D, n ϭ 11). E, summary of mean data (*, p Ͻ 0.05, Mann-Whitney unpaired test). channels on the nuclear envelope, such as Na ϩ /K ϩ -ATPases (47), P2X 7 receptors (48), and the nuclear envelope Ca 2ϩ -ATPase (49). However, one of the most important functions of perinuclear mitochondria is likely the modulation of nuclear translocation of macromolecules, such as CaM, kinases/phosphatases, transcription factors, as well as RNA, by the nuclear pore complex (NPC). There is now increasing evidence that such nuclear translocation is ATP-and/or Ca 2ϩ -dependent (50 -54). Specifically, ATP on the cytoplasmic face and Ca 2ϩ on the nucleoplasmic face both promote opening of the NPC and thus nuclear translocation (52,54). There is, however, some evidence that high [Ca 2ϩ ] i (Ͼ1 M) (55,56) and depletion of the nuclear envelope Ca 2ϩ store inhibits the NPC and nuclear translocation (54,57,58). Thus, local ATP generated close to the nuclear envelope could facilitate nuclear translocation either directly or by maintaining nuclear envelope Ca 2ϩ store refilling (39,57). In addition, more recently it was shown that nuclear translocation of the protein histone H1 in intact cardiomyocytes is critically dependent on functional mitochondria and enzymatic phosphotransfer by creatine kinase and/or adenylate kinase (59). Perinuclear mitochondrial Ca 2ϩ uptake, which drives metabolism and ATP generation close to the nucleus, therefore likely plays a pivotal role in nuclear translocation.
In summary, parotid acinar cells do not exhibit the striking perigranular mitochondrial distribution observed in pancreatic acinar cells (8 -10). In fact a large proportion of mitochondria are localized to the perinuclear region of these cells. The absence of a definitive perigranular mitochondrial belt in parotid acinar cells likely facilitates the rapid, global [Ca 2ϩ ] i signals that are important for the activation of effective fluid secretion in these cells. These agonist-evoked rapid global [Ca 2ϩ ] i signals stimulate Ca 2ϩ uptake into perinuclear mitochondria that may shape nuclear [Ca 2ϩ ] signals, but perhaps more importantly profoundly increase mitochondrial metabolism, thereby generating ATP close to the nucleus. These data collectively suggest that perinuclear mitochondria likely play an essential role in regulating a variety of nuclear functions while facilitating effective fluid secretion in parotid acinar cells.