Mitochondria support inositol 1,4,5-trisphosphate-mediated Ca2+ waves in cultured oligodendrocytes.

We have examined the spatial and temporal nature of Ca2+ signals activated via the phosphoinositide pathway in oligodendrocytes and the cellular specializations underlying oligodendrocyte Ca2+ response characteristics. Cultured cortical oligodendrocytes were incubated with fluo 3 or fura 2, and digital video fluorescence microscopy was used to study the effect of methacholine on [Ca2+]i. Single peaks, oscillations, and steady-state plateau [Ca2+]i elevations were evoked by increasing agonist concentration. The peaks and oscillations were found to be Ca2+ wave fronts, which propagate via distinct amplification regions in the cell where the kinetics of Ca2+ release (amplitude and rate of rise of response) are elevated. Staining with 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolecarbocyanine iodide (JC-1) and 3,3′-dihexyloxacarbocyanine iodide revealed that mitochondria are found in groups of three or more in oligodendrocyte processes and that the groups are distributed with considerable distance separating them. Cross-correlation analysis showed a high degree of correlation between sites where mitochondria are present and peaks in the amplitude and rate of rise of the Ca2+ response. Intramitochondrial Ca2+ concentration, measured using rhod 2, increased upon treatment with methacholine. Methacholine also evoked a rapid change in mitochondrial membrane potential as measured by the J-aggregate fluorescence of JC-1. Pretreatment with the mitochondrial inhibitors carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (1 μM, 2 min) or antimycin (2 μg/ml, 2 min) altered the methacholine-evoked Ca2+ response in most cells studied, responses being either markedly potentiated or inhibited. The results of this study demonstrate that stimulation of phosphoinositide-coupled muscarinic acetylcholinoceptors activates propagating Ca2+ wave fronts in oligodendrocytes and that the characteristics of these waves are dependent on mitochondrial location and function.

Cultured oligodendrocytes express a variety of receptors coupled to the mobilization of Ca 2ϩ from inositol trisphosphate receptor (InsP 3 R) 1 -expressing intracellular stores. These in-clude M 1 muscarinic cholinoceptors (1)(2)(3), which are also thought to be expressed on oligodendrocytes in vivo (4). While InsP 3 -evoked Ca 2ϩ responses are known to be present in oligodendrocytes (2,5), the spatial and temporal nature of these has not been characterized. In a wide variety of cell types, including several other types of glial cells, stimulation of InsP 3 -coupled receptors results in activation of Ca 2ϩ waves and oscillations (6 -10). In astrocytes, Ca 2ϩ waves are initiated in several distinct regions, which then propagate throughout the cell via multiple amplification sites, at which the amplitude of the wave and the rate of rise of the response are markedly elevated compared to surrounding regions (9,10). 2 The molecular specializations underlying wave initiation and propagation sites in glial and other cell types remain poorly defined. Recent reports have, however, implicated a role for mitochondria as well as endoplasmic reticulum (ER) in the regulation of cytosolic Ca 2ϩ signals and the propagation of Ca 2ϩ waves (11)(12)(13)(14)(15). Endoplasmic reticulum and mitochondria are found in close apposition in several cell types (16,17). Ca 2ϩ released via InsP 3 Rs is transferred into mitochondria more effectively than [Ca 2ϩ ] i increases evoked by other means (11,18), consistent with a close functional relationship between InsP 3 Rs and mitochondrial Ca 2ϩ uptake. Also, Ca 2ϩ waves are increased by up-regulation of mitochondrial function in Xenopus oocytes (14). Coordinate activity of mitochondria and ER (14,15) could enable differences in Ca 2ϩ buffering and Ca 2ϩ regulation in subcellular microdomains.
The present study was undertaken to determine whether Ca 2ϩ signals activated via the phosphoinositide pathway in oligodendrocytes would take the form of Ca 2ϩ waves with characteristic initiation sites and propagation sites. Further, we sought to determine the nature of the cellular specializations underlying oligodendrocyte Ca 2ϩ signals. For this, we analyzed the pattern of distribution of mitochondria in cells previously stimulated to evoke Ca 2ϩ responses and also examined the effect of altering mitochondrial activity on Ca 2ϩ signaling. Our findings indicate that oligodendrocytes respond to methacholine (MCh)-evoked InsP 3 generation with Ca 2ϩ waves. These waves include sites of high local Ca 2ϩ release kinetics in the same locations as for caffeine-evoked waves in the same cells, consistent with their corresponding to distinct cellular specializations. Mitochondria were found to be present at these wave amplification sites, apparently in intimate association with the ER, but not elsewhere along oligodendrocyte processes. MCh also evokes changes in mitochondrial Ca 2ϩ and membrane potential. Inhibition of mitochondrial activity was found to abolish or in some cases potentiate the ability of MCh to evoke a cytosolic Ca 2ϩ response. Thus, both mitochondrial location and function appear to be crucial to the generation and characteristics of Ca 2ϩ waves in oligodendrocytes. * 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 U.S.C. Section 1734 solely to indicate this fact.
Cell Culture-Oligodendrocytes were cultured from 2-day-old rat pups as described previously (19). After 8 days in vitro, the flasks were vigorously shaken overnight, and the supernatant was repeatedly plated onto plastic dishes. Nonadherent cells were then replated onto glass coverslips coated with polyornithine in DME-N1 containing 0.5% fetal bovine serum and maintained in 10% CO 2 , 90% air. After 24 h, culture medium was replaced with DME-N1 containing 2% fetal bovine serum, and cells were used 4 -8 days after plating on coverslips. More than 90% of these cells were positively labeled with a monoclonal antibody against galactocerebroside, a marker for oligodendrocytes (20).
[Ca 2ϩ ] i and Membrane Potential Measurements-Cells grown on coverslips were incubated with 5 M fura 2-AM for 20 min at room temperature as described previously (9,21). The perfusion chamber was positioned on the stage of an inverted microscope, with fluorescence images at 340 and 380 nm excitation wavelengths being acquired at 510 nm emission wavelength through a microchannel plate intensifier with a CCD camera (9). All cells in a field were analyzed individually. For analysis of Ca 2ϩ waves in single cells, cells were incubated with 5 M fluo 3-AM for 20 min, and fluorescence images were acquired with excitation at 495 nm (emission at 525 nm) (9). Images were digitized and averaged in a Trapix 55/4256 image processor. Nonzero pixels within each slice were averaged and plotted as normalized fluorescence intensities (⌬F/F) against time or against distance along the cell. Cells were divided for analysis into 0.83-m-wide regions sequentially along the longitudinal axis. For measurement of Ca 2ϩ within mitochondria, cells were incubated with 5 M rhod 2-AM for 30 min and returned to culture medium for 24 h at 37°C, which results in removal of most cytosolic rhod 2 while the mitochondria remain stained (15). Images were acquired with excitation at 525 nm and emission at 610 nm. High resolution imaging enabled delimiting of individual mitochondria for analysis. For measurement of mitochondrial membrane potential, cells were incubated with 10 g/ml JC-1 for 10 min, and red (J-aggregate) fluorescence was measured (33,34). Rhod 2 and JC-1 fluorescence signals were not calibrated to Ca 2ϩ concentration or membrane potential and represent qualitative measures.
Staining and Analysis of Endoplasmic Reticulum and Mitochondria Distribution-Cells were incubated in DiOC 6 (3) (0.5 g/ml) in phosphate-buffered saline for 1 min, washed, and examined under the fluorescence microscope. For JC-1 staining, cells were incubated with 10 g/ml JC-1 for 10 min in phosphate-buffered saline, washed once in phosphate-buffered saline, and viewed using either fluorescein isothiocyanate or Cy3 filter sets as appropriate. For high resolution analysis of cells stained with these fluorescent organellar dyes, we used an exhaustive photon reassignment procedure developed jointly by University of Massachusetts and Scanalytics Inc. (22). In this method, digital images from a standard wide field microscope are restored using an algorithm that removes out of focus plane fluorescence (22). 2 The axial resolution of our microscope under our measurement conditions was 0.75 m. Ca 2ϩ waves were evoked in a fluo 3-loaded cell using MCh and, after recovery, the cell was incubated with JC-1 or DiOC 6 (3) while still on the microscope stage, and the fluorescence of this dye was measured using appropriate optical filters. Ca 2ϩ wave kinetics and mitochondrial distribution patterns were then compared. Fluo 3 fluorescence was negligible (Ͻ1% of total fluorescence for both JC-1 and DiOC 6 (3)) in the DiOC 6 (3) or JC-1 channels due to the relatively bright fluorescence of these two dyes.
Cross-correlation Analysis-The spatial patterns of Ca 2ϩ responses and of mitochondrial distribution were compared in the same cells using a cross-correlation function derived from the fast Fourier transform of the data sets as a quantitative test for similarity (23, 24). The mean values of normalized fluorescence intensities were subtracted out, and the resulting zero mean waves were embedded in surrounding zeros. The windowed data were Fourier transformed via a fast Fourier transform algorithm, using standard functions in Mathematica (Wolfram Research, Inc., Champaign, IL). Cross-spectra were then formed as a product of one wave with the complex conjugate transform of a second wave. The cross-correlation function was produced by inverse Fourier transformation of the cross-spectrum. Changing the order of the functions produced the expected mirror image cross-correlation functions. Performing these operations on a single wave produces the power density spectrum and the autocorrelation function. Data are presented as mean Ϯ S.D.

InsP 3 -mediated Ca 2ϩ Responses in Oligodendrocytes-Ex-
periments were performed to examine the spatial and temporal characteristics of Ca 2ϩ responses evoked in oligodendrocytes by MCh, an agonist of muscarinic receptors. Single peaks, oscillations, and steady-state plateau [Ca 2ϩ ] i elevations were evoked by increasing agonist concentration (Fig. 1, a and b; n ϭ 240 cells). 73.3% of cells examined responded to 1 mM MCh. In addition, in individual cells, a graded increase in the magnitude of the peak was observed (Fig. 1a), unlike the all or none responses to MCh previously reported in several other cell types (25,26). The EC 50 of the peak response amplitude was 1.5 Ϯ 0.5 M, whereas the EC 50 of the plateau phase was 33.6 Ϯ 26.2 M; however, these values were not significantly different. Addition of 100 M MCh in nominally Ca 2ϩ -free conditions ([Ca 2ϩ ] o Ͻ 5 M) evoked an initial peak response of the same magnitude as in normal Ca 2ϩ , but whereas in normal Ca 2ϩ a plateau phase or oscillations were evoked, in Ca 2ϩ -free medium no significant plateau phase (Fig. 1c), or only one or two small oscillations (not shown), was found (n ϭ 10). Oligodendrocytes consecutively stimulated three times with MCh in normal extracellular solution ([Ca 2ϩ ] o ϭ 1.5 mM, at Ͼ15-min intervals, displayed responses of consistent magnitude. Pretreatment with the store-depleting agent thapsigargin abolished responses to MCh (n ϭ 34; not shown). These results suggest that the initial peak responses are due to InsP 3 -mediated Ca 2ϩ mobilization from stores, whereas the sustained oscillations and the plateau phase apparently require Ca 2ϩ entry across the plasma membrane.
To investigate the spatial nature of the Ca 2ϩ response evoked by InsP 3 generation, we examined the MCh response in fluo 3-loaded oligodendrocytes at higher time resolution (1 image every 66 ms), and sectioning of the cell into successive 0.83-m-wide segments along the cell axis (see Ref. 9). Stimulation of an oligodendrocyte with 100 M MCh resulted in the activation of propagating Ca 2ϩ wave fronts from several initiation sites (Fig. 2a). Although oligodendrocyte processes are approximately cylindrical (27), the magnitude and rate of rise of the Ca 2ϩ response were found to be non-uniform, with some regions of the processes displaying markedly higher amplitude responses or faster rate of rise of response than in surrounding regions (Fig. 2b). The regions of large amplitude and regions with steeply rising responses corresponded closely to each other (see later). These appear highly analogous to the amplification regions previously described for InsP 3 -mediated responses in type 1 astrocytes (9), the elevated Ca 2ϩ kinetics of which probably contribute to the sustenance and propagation of the wave (10). A plot of the time to 50% peak against cell length clearly shows the wave initiation sites (Fig. 2c, asterisks) which are located near but not colocalized with sites where the higher kinetics in Ca 2ϩ release are measured (compare Fig. 2b and Fig. 2c). The specialized regions of wave initiation and amplification varied in diameter from 1 to 5 m.
Comparison between InsP 3 -and Caffeine-evoked Ca 2ϩ Waves-Oligodendrocytes express a single Ca 2ϩ pool sensitive to InsP 3 , caffeine, and cyclopiazonic acid (28). 3 We have previously demonstrated that cells of the oligodendrocyte lineage respond to caffeine with propagating Ca 2ϩ waves the local kinetics of which vary along the length of the cell (28). Here we compared, in the same cells, InsP 3 -mediated (MCh-induced) and caffeine-mediated waves. These experiments showed that the local amplitudes and rates of rise of the Ca 2ϩ signal were similar under each stimulation condition (Fig. 3a). In the cell shown in Fig. 3, at least four specialized regions of high amplitude (marked with an asterisk) are identifiable, which appear in similar locations in the cell for the two responses. An additional high amplitude region for the caffeine response (42-48 m) does not appear in the MCh response. Cross-correlation analysis of the caffeine peak amplitude versus MCh peak amplitude showed high correlation values in phase (Fig. 3b). In three separate experiments, the cross-correlation values were 0.70 Ϯ 0.05 (mean Ϯ S.D.). The pattern of cross-correlation is similar to the autocorrelation of the local peak amplitude data for the caffeine response (Fig. 3b). Comparison of the slopes between the two responses also yielded similar high crosscorrelation values ( Fig. 3b; 0.82 Ϯ 0.17; n ϭ 3). This result indicates that a local cellular specialization that supports high Ca 2ϩ release may be due not only to increased expression of InsP 3 R, as suggested by us previously (10) 2 but to a generalized specialization of the Ca 2ϩ release machinery, perhaps including Ca 2ϩ -induced Ca 2ϩ -release processes and specialization of other organelles (see later).
Mitochondrial Distribution in Oligodendrocytes-One possible cause of regions with higher local kinetics of Ca 2ϩ release in response to InsP 3 or caffeine would be mitochondrial regulation of Ca 2ϩ in cellular microdomains. Mitochondria are known to be able to take up considerable amounts of Ca 2ϩ in response to InsP 3 R channel activation (15,18,29). Because both InsP 3 Rs and ryanodine receptors display bell-shaped sensitivity to Ca 2ϩ (30), such local regulation could in theory modulate the level of Ca 2ϩ surrounding store release channels sufficiently to regulate gating kinetics of these receptor channels. The pattern of distribution of mitochondria and ER in oligodendrocytes was investigated using the lipophilic dyes JC-1 and DiOC 6 (3). DiOC 6 (3) is a potentiometric carbocyanine dye widely used to stain ER, particularly in cultured cells (31,32), 2 but which also intensely stains mitochondria. 2 JC-1, another carbocyanine dye, partitions into mitochondrial membranes and at low membrane potential emits at 527 nm when excited at 490 nm. At high membrane potentials, JC-1 forms J-aggregates and the emission wavelength shifts toward red (590 nm) (32,33), enabling the dye to function as a highly specific marker of both mitochondrial location and activity.
Typical staining patterns of these dyes in oligodendrocytes are displayed in Fig. 4. Both dyes indicate that mitochondria are found at highest density in the cell body region but are also present along oligodendrocyte processes. In thicker processes, groups of three or more mitochondria can be typically found in a convoluted, closely interrelated group (Fig. 4a), at a considerable distance from the next such group or individual mitochondrion; while in thin processes, single, long mitochondria are found with separations of several micrometers (Fig. 4, a  and b). This is similar to mitochondrial distribution described in cultured neurons (35); but in several other types of cells in culture, mitochondria are not found along processes (36). While bright DiOC 6 (3) staining marks groups of mitochondria along processes, less intense DiOC 6 (3) staining is diffusely distributed through the cell (Fig. 4a), indicating that ER is present throughout the network of oligodendrocyte processes. With JC-1, all mitochondria fluoresce green, while parts of most of them also were red, indicating variations in membrane potential (and corresponding level of activity) within single mitochondria, and between different mitochondria in a given cell (Fig. 4c) (33). The finding of single or groups of mitochondria along oligodendrocyte processes, filling most of the available space and apparently in close proximity to ER (Fig. 4), indicates that the potential exists for functional interactions between ER and mitochondria in these cells. Close physical apposition between mitochondria and ER is also found in retinal oligodendroblasts (37) and other cell types (16,17).
Role of Mitochondria in Propagation of Ca 2ϩ Waves-To investigate whether the distribution of mitochondria is related to characteristics of the Ca 2ϩ waves evoked by store release, we studied Ca 2ϩ waves induced by MCh (100 M) and compared the spatial kinetics with the mitochondrial distribution in the same cells. If mitochondrial function is fundamental to Ca 2ϩ signaling in cellular microdomains, the pattern of mitochondrial distribution will be expected to be well correlated with the kinetics of local Ca 2ϩ release. Ca 2ϩ waves were evoked by MCh in fluo 3-loaded cells and, following recovery, they were incubated with JC-1 or DiOC 6 (3) while still on the microscope stage. The fluorescence of this second dye was measured, and the resultant wave and organellar distribution patterns were then compared. Fig. 5 shows two representative cells, one stained with DiOC 6 (3) (Fig. 5, a and b) and another stained with JC-1 (Fig. 5, c and d) showing the fluorescence intensity profiles of the cyanine dyes and the local Ca 2ϩ release kinetics. Crosscorrelation analysis was used to compare the patterns of local amplitudes and the rate of rise of the signals with the patterns of JC-1 or DiOC 6 (3) fluorescence (see "Experimental Procedures"). Comparison of the patterns of DiOC 6 (3) staining intensity along the cellular process (distribution of mitochondria) with the patterns of both the amplitude and rate of rise of the Ca 2ϩ response showed high correlation values (Fig. 5b), in this cell ϳ2-3 m from phase. The local amplitude and the rate of rise (slope) of the Ca 2ϩ responses when compared with each other showed good correlation values in phase. Results from six separate experiments are summarized in Table I. JC-1 staining was discrete and appeared at all sites of the cell process where higher Ca 2ϩ release kinetics were measured (Fig. 5c). No peaks in the Ca 2ϩ response were found at sites along processes where mitochondria were absent, and mitochondria were not found at sites where response peaks were not observed. With JC-1, the green and red fluorescence intensity patterns of staining were essentially identical with each other (Table I), such that the cross-correlation appears essentially similar to autocorrelation. Comparison of the patterns of Ca 2ϩ wave kinetics (amplitude and rate of rise) also resulted in very high correlation values (Fig. 5d). The magnitude and the pattern of cross-correlation were similar between wave characteristics and either JC-1 green fluorescence or red J-aggregates ( Fig. 5d; Table I). This result may suggest that the location of functional mitochondria, rather than their local membrane potential at rest, may be most crucial in determining wave characteristics. Mitochondrial location was found to be invariant over Ͼ15-min periods either at rest or after stimulation with MCh (not shown).
The effects of agonist-induced Ca 2ϩ waves on mitochondrial activity were measured by monitoring intramitochondrial Ca 2ϩ concentration ([Ca 2ϩ ] m ) and mitochondrial membrane potential using the fluorescent dyes rhod 2 and JC-1, respectively. Rhod 2 can be selectively loaded into mitochondria, and used as an indicator of mitochondrial free [Ca 2ϩ ] m (see Ref. 15). High resolution imaging allowed measurement of fluorescence within individual mitochondria along oligodendrocyte processes. MCh (100 M) caused a rapid onset increase in mitochondrial rhod 2 fluorescence (43 of 66 mitochondria, 65% ) Fig. 6a)). These responses reflect uptake of Ca 2ϩ into mitochondria and appear to occur sufficiently rapidly to act as a fast local Ca 2ϩ buffering system. Such a mechanism could potentially play a role in modulation of Ca 2ϩ sensitivity of nearby InsP 3 Rs (see "Discussion"). Treatment with the protonophore FCCP (1 M), which collapses the proton gradient across mitochondrial membranes (12, 38), decreased rhod 2 fluorescence in mitochondria, typically within 2 min of treatment, indicating release of mitochondrial Ca 2ϩ (27 of 43 mitochondria, 63% (Fig. 6b)). Similarly, antimycin (2 g/ml), an inhibitor of electron transport through complex III (14,39), also decreased rhod 2 fluorescence (i.e. [Ca 2ϩ ] m ) in 24 of 30 mitochondria (80%, Fig. 6c). The antimycin-induced decrease in rhod 2 fluorescence began within 30 s of treatment and took up to 4 min to reach completion. To examine the effect of MCh on mitochondrial membrane potential, cells were incubated with JC-1, and the level of red J-aggregate formation was measured. In a proportion of mitochondria, MCh (100 M) evoked a rapid onset, transient increase in fluorescence (34 of 191 mitochondria, 18% (Fig. 6d)). In contrast, both FCCP (11 of 11, 100%) and antimycin (20 of 20, 100%) caused a slow decrease in J-aggregate fluorescence in oligodendrocyte mitochondria which reached a plateau in 2-4 min (Fig. 6, e and f), consistent with previous studies in other cell types (33,34).
To investigate the nature of the role played by mitochondria in Ca 2ϩ signaling in oligodendrocytes, the effect of modifying mitochondrial activity on MCh-evoked cytosolic Ca 2ϩ signals was studied. After initial stimulation with MCh, cells were treated with FCCP. Relatively brief (2 or 5 min) pretreatment with 1 M FCCP caused MCh responses in some cells to be abolished (Fig. 7a) but responses in other cells to be markedly potentiated (Fig. 7b), while others were unaffected (Table II). In the cell shown in Fig. 7b, the amplitude of the MCh-evoked peak is increased after FCCP pretreatment, and the peak is now followed by a second peak and then a steady-state plateau.  (3). The cell body is seen to contain numerous mitochondria, while fewer mitochondria are found along the process in clusters. The low level fluorescence in the rest of the process is weak staining of the ER. In finer processes, a single mitochondrion occasionally fills virtually all the cytosolic space. b, high power image of a region from a illustrates the convoluted clustering of mitochondria that occurs at several sites along the process. This reveals the complex entanglement of multiple mitochondria that makes up a group, compared to the simpler orientation of a single mitochondrion in a nearby thin process. c, oligodendrocyte stained with JC-1. The cell was incubated with 10 g/ml JC-1 and then washed and photographed using either a fluorescein isothiocyanate or Cy3 filter as appropriate. JC-1 staining in large processes also showed mitochondria distributed in groups as in a (not shown), but in thin processes long, single mitochondria are found which all fluoresce green and parts of which also fluoresce red due to J-aggregate formation in response to elevated mitochondrial membrane potential.
MCh-evoked Ca 2ϩ response (Table II). In most cells examined, a 2-min pretreatment with CCCP (5 M), a structural analog of FCCP, completely abolished MCh-induced [Ca 2ϩ ] i signals ( Table II). An alternative approach to modulation of mitochondrial function was treatment of the cells with antimycin. Incubation with antimycin (2 g/ml) for as little as 2 min inhibited Ca 2ϩ responses to MCh stimulation in most cells (Fig. 7c; Table II). Fig. 7c shows an example trace of a single oligodendrocyte which initially displayed a large oscillatory response to MCh (100 M). Antimycin (2 g/ml, 2 min) pretreatment completely abolished this response. Prolonged treatment with antimycin had no greater effect on Ca 2ϩ signals (Table II), while a lower concentration (0.2 g/ml, 15 min) also inhibited MCh-evoked responses to a marked extent (Table II). The cell was sliced into 0.8-m-wide segments and numbers indicate discrete peaks in both Ca 2ϩ wave amplitude and DiOC 6 (3) fluorescence. b, cross-correlation analysis of the patterns of Ca 2ϩ release kinetics and DiOC 6 (3) fluorescence. Comparison of the wave amplitude and rate of rise of response (E) showed high correlation consistent with these being representations of the same release phenomenon. The rate of rise of response (q) and the peak amplitudes (ç) when compared with DiOC 6 (3) fluorescence also showed high cross-correlation 2-3 m from phase. c, measurements of wave amplitude (⌬F/F (q) and both red (E) and green (ç) JC-1 fluorescence intensities plotted against distance along the process. Peaks of apparent correlation are numbered for both amplitude data and JC-1 fluorescence. Peaks in wave amplitude are found only where peaks of fluorescence occur, and no mitochondria are found in the absence of a peak wave front amplitude. Red JC-1 fluorescence displays particularly high discrimination of mitochondrial signal. d, cross-correlation analysis of patterns of JC-1 fluorescence and Ca 2ϩ release kinetics in an oligodendrocyte process. Comparison of wave amplitudes and JC-1 red (E) and green (q) and rate of rise of response and JC-1 red (Ç) and green (å) fluorescence. The red and green fluorescence profiles have high cross-correlation with wave characteristics, and also showed high crosscorrelation with each other (not shown).

TABLE I Correlation between distribution of mitochondria and Ca 2ϩ wave propagation characteristics
Results of cross-correlation analysis of Ca 2ϩ wave patterns and mitochondrial staining with carbocyanine dyes. Data are presented as mean Ϯ S.D. High correlation is found between mitochondrial location and two parameters of high Ca 2ϩ release kinetics, peak (amplitude of Ca 2ϩ response) and slope (rate of rise of Ca 2ϩ response). Correlation occurs very close to phase, indicating that the locations of maxima in these kinetic parameters are very close to maxima in dye fluorescence, i.e. mitochondrial loci. The results of this study demonstrate that stimulation of a phosphoinositide-coupled muscarinic acetylcholinoceptor activates propagating Ca 2ϩ wave fronts in oligodendrocytes and that the characteristics of these waves are dependent on mitochondrial location and function. MCh activates a rapid onset Ca 2ϩ peak in a concentration-dependent manner in ϳ73% of oligodendrocytes tested. Plateau and sustained oscillations, but not initial peak, elevations were dependent on extracellular Ca 2ϩ , consistent with store-operated channel activation and the need for store refilling (8). Although a number of reports have shown that cells of the oligodendrocyte lineage in culture possess functional receptor systems for neurotransmitters and that their stimulation results in cellular Ca 2ϩ signals (2, 5, 40 -42), the spatiotemporal characteristics of these responses have not previously been examined. We show here that MChevoked Ca 2ϩ signals in oligodendrocytes are propagating Ca 2ϩ wave fronts. Such waves might enable oligodendrocytes to convey signaling information over relatively long distances in brain (43). A possible role for long distance signaling by astrocytes in brain function has already been hypothesized (6,44).
As in astrocytes (9, 10), propagation of Ca 2ϩ waves in oligodendrocytes occurs with significant differences in Ca 2ϩ release kinetics in certain regions of the cell. These regions of cellular specializations, where the local signal amplitude and rate of rise are higher, were highly correlated in both InsP 3 -and caffeine-evoked Ca 2ϩ waves. Thus, microdomains that may support wave propagation (10) can apparently do so via both InsP 3 Rs and ryanodine receptors, giving the oligodendrocyte a generalized "Ca 2ϩ fingerprint" for signal propagation. These sites appear to be distinct from sites of wave initiation, of which each cell can possess five or more throughout its arborization. The changes in local wave kinetics are not due to large changes in cell shape, since oligodendrocyte processes have been shown to be approximately uniformly cylindrical (27). Our results are consistent with the idea that oligodendrocytes possess a single functional Ca 2ϩ pool that is specialized in certain regions of the cell as wave amplification sites. While it is possible that InsP 3 Rs and ryanodine receptors are colocalized at these specialized sites, an alternative or complementary explanation would be that the major factor or factors governing this Ca 2ϩ fingerprint is related to cellular characteristics other than the density of ER receptor channels alone.
Fluorescent dyes targeted to either ER and mitochondria (DiOC 6 (3)) or mitochondria alone (JC-1) showed that in oligodendrocyte processes mitochondria are typically found in closely associated groups at cellular sites that closely correspond to the sites of high Ca 2ϩ wave kinetics. Mitochondria were always located at such sites and were not found elsewhere along the processes. This indicates that mitochondria are likely to be important for regulation of Ca 2ϩ response characteristics and that regulation of the spatial distribution of mitochondria could be one way in which spatially discrete signaling is achieved in glial cells. Mitochondrial location in oligodendrocyte processes did not vary over Ͼ15-min periods under resting or agonist-stimulated conditions (not shown). Evidence from this and previous (11, 16 -18, 36) studies show that a close spatial relationship exists between mitochondria and ER. One possible explanation of mitochondrial effects on InsP 3 R-mediated waves is that the proximity of mitochondria to the ER InsP 3 Rs alters gating kinetics of InsP 3 R channels. Mitochondria located close to InsP 3 Rs could buffer considerable quantities of Ca 2ϩ , thus maintaining a microdomain of Ca 2ϩ near the receptor that is at a favorable concentration for channel activation. When mitochondria are inhibited (or physiologically when mitochondrial Ca 2ϩ load is already high), the local cytosolic Ca 2ϩ concentration rises to a level at which Ca 2ϩ is inhibitory to InsP 3 R activation, such that even during InsP 3 generation the InsP 3 R is not activated. Thus, by altering the excitability of the medium, mitochondria could shape the characteristics of the signal that InsP 3 can evoke (14,29,45,46). This hypothesis is supported by the finding that MCh evokes a rapid elevation of [Ca 2ϩ ] m and a depolarization of mitochondrial membranes in oligodendrocytes. Rapid mitochondrial Ca 2ϩ uptake enables mitochondria to sequester Ca 2ϩ during the early stages of Ca 2ϩ oscillations (29). The ability of mitochondria to buffer Ca 2ϩ loads could enable them to decrease the magnitude of an agonist-evoked Ca 2ϩ peak by uptake of Ca 2ϩ from the cytoplasm or to modify the Ca 2ϩ -dependent sensitivity of the release channel.
In addition to spatial distribution, evidence for a functional role for mitochondria in oligodendrocyte signaling comes from the fact that inhibitors of mitochondrial activity typically inhibited or abolished Ca 2ϩ responses. Brief treatment with FCCP revealed a bimodal effect of mitochondrial modulation, causing either an increase or a decrease in the amplitude of the response. The bimodal effect of FCCP pretreatment on Ca 2ϩ responses in oligodendrocytes appears consistent with a role for mitochondria in altering the Ca 2ϩ "setpoint" for InsP 3 Rs, thus either increasing or decreasing the sensitivity of the receptor to InsP 3 (8,30). FCCP was found to evoke decreases in [Ca 2ϩ ] m as measured by rhod 2 and increases in [Ca 2ϩ ] i as measured by fluo 3 over 2-3 min of incubation (Fig. 6b). No consistent relationship was evident between the magnitude of [Ca 2ϩ ] i elevation evoked by FCCP and its effect on Ca 2ϩ signals. The important modulatory effect of mitochondrial Ca 2ϩ release, however, would be expected to be very localized, i.e. effects on [Ca 2ϩ ] i around nearby InsP 3 Rs, rather than via bulk cytosolic [Ca 2ϩ ] i elevation, and as such may not be readily measured. Potentiation of responses was also found in a small percentage of cells treated with antimycin. However, some cells appeared relatively resistant to the effects of mitochondrial inhibitors, with ϳ25% of cells unaffected by antimycin and ϳ40% unaffected by 1 M FCCP, and after MCh stimulation only a proportion of mitochondria showed changes in membrane potential and [Ca 2ϩ ] m . Such differences may indicate either that mitochondria are more important for Ca 2ϩ signaling in some cells than in others or that the interaction between cytosolic and mitochondrial responses depends on several independent factors. Possible factors modulating the consequences of mitochondrial inhibition on cytosolic responses include: (i) the amount of Ca 2ϩ loaded into mitochondria in a given cell; (ii) the concentration of cytosolic Ca 2ϩ around InsP 3 Rs near the mitochondria; and (iii) the spatial relationship between the mitochondria and InsP 3 Rs. It remains possible that a primary role of mitochondria in modulating Ca 2ϩ signaling is simply to produce sufficient ATP to maintain Ca 2ϩ uptake into ER via the sarcoplasmic-endoplasmic reticulum Ca 2ϩ ATPase. Indeed, the effects of long term treatment with mitochondrial inhibitors on Ca 2ϩ responses may be due to a combination of modulation of release and ATP depletion. However, only brief periods (2 min) of FCCP, CCCP, or antimycin pretreatment were required to modify or abolish Ca 2ϩ responses in oligodendrocytes, a time period no greater than that required to inhibit mitochondrial function as measured by [Ca 2ϩ ] m or mitochondrial membrane potential.
Our results detail several novel functional consequences of colocalization of mitochondria and ER: (i) the present study demonstrates that active mitochondria/ER interactions appear to be crucial to Ca 2ϩ wave propagation, such that in the absence of functional mitochondrial Ca 2ϩ uptake, oligodendrocyte Ca 2ϩ responses are modified, resulting in enhanced release kinetics, or inhibition or even abolition of wave generation, depending on the particular cell examined; (ii) the location of mitochondria appears to underlie the spatial distribution of sites of enhanced Ca 2ϩ release, which are thought to be crucial to the maintenance of wave propagation (10); (iii) changes in both intramitochondrial Ca 2ϩ and mitochondrial membrane

Effects of mitochondrial inhibitors on MCh-evoked Ca 2ϩ responses in oligodendrocytes
Fura-2-loaded oligodendrocytes were challenged with MCh (100 M) first; after recovery, they were exposed to FCCP, CCCP, or antimycin for different times, and the cells challenged with MCh a second time. Data represent response to the second stimulation with MCh. n is total number of cells responding to MCh that were examined under each condition. potential occur upon agonist stimulation of oligodendrocytes. A variety of factors may govern Ca 2ϩ waves in glia, including regional heterogeneities in ER protein expression, 2 but clearly sites of mitochondrial expression play a major role in shaping Ca 2ϩ responses.