Ca2+ uptake and release from endoplasmic reticulum (ER) and mitochondrial Ca2+ stores play important physiological and pathological roles, and these processes are shaped by interactions that depend on the structural intimacy between these organelles. Here we investigate the morphological and functional relationships between mitochondria, ER, and the sites of intracellular Ca2+ release inXenopus laevis oocytes by combining confocal imaging of local Ca2+ release events (“Ca2+ puffs”) with mitochondrial localization visualized using vital dyes and subcellularly targeted fluorescent proteins. Mitochondria and ER are localized in cortical bands ∼6–8 μm wide, with the mitochondria arranged as densely packed “islands” interconnected by discrete strands. The ER is concentrated more superficially than mitochondria, and the mean separation between Ca2+ puff sites and mitochondria is ∼2.3 μm. However, a subpopulation of Ca2+ puff sites is intimately associated with mitochondria (∼28% within <600 nm), a greater number than expected if Ca2+ puff sites were randomly distributed. Ca2+release sites close to mitochondria exhibit lower Ca2+ puff activity than Ca2+ puff sites in regions with lower mitochondrial density. Furthermore, Ca2+ puff sites in close association with mitochondria rarely serve as the sites for Ca2+ wave initiation. We conclude that mitochondria play important roles in regulating local ER excitability, Ca2+wave initiation, and, thereby, spatial patterning of global Ca2+ signals.
- endoplasmic reticulum
- calcium release
functional interactions between Ca2+ stores within the endoplasmic reticulum (ER) and mitochondria play important roles in determining the kinetics and, thereby, the physiological action of cellular Ca2+ signals (for reviews see Refs.10, 16, 37, and39). Ca2+ release from inositol 1,4,5-trisphosphate (IP3) or ryanodine-sensitive ER Ca2+ stores results in an increased intramitochondrial Ca2+ concentration (8, 18, 36, 38, 40, 47), which upregulates mitochondrial metabolism and ATP production (18, 27, 36, 41). Reciprocally, mitochondrial Ca2+ sequestration influences the spatiotemporal profile of cytosolic Ca2+ signals generated by Ca2+release and Ca2+ entry pathways (12, 20, 22, 45,49). In particular, mitochondrial Ca2+ uptake slows the rise and accelerates the decline of cytosolic Ca2+transients, while delayed Ca2+ extrusion from mitochondria slows the return of cytoplasmic Ca2+ concentration to its resting level (12, 19). Because the gating of IP3 receptors and ryanodine receptors is Ca2+sensitive, perturbation of local Ca2+ levels by adjacent mitochondria is likely to influence the probability of their activation (17) as well as their interactions through Ca2+-induced Ca2+ release to establish a propagating Ca2+ wave. Furthermore, in situations where mitochondria are exposed to high levels of cytosolic Ca2+, mitochondria can amplify IP3-evoked Ca2+release through concurrent cycles of mitochondrial depolarization and Ca2+ release triggered by Ca2+-stimulated gating of the permeability transition pore (21). Finally, the level of mitochondrial ATP production may influence the activity of Ca2+-ATPases that refill intracellular Ca2+stores as well as the local sensitivity of intracellular Ca2+ release channels (23, 25).
As a result of this functional interrelationship between mitochondria and ER Ca2+ stores, attention has focused on the structural basis of these interactions. Given the presumed low affinity of mitochondrial Ca2+ uptake mechanisms (14, 37) and the demonstration of large increases in mitochondrial matrix Ca2+ concentration on cell stimulation (16,37-40), ER Ca2+ release sites and mitochondria have been hypothesized to be closely apposed. Indeed, evidence from many cell types suggests that a portion of the cellular mitochondrial network is intimately associated with ER membranes. For example, electron micrographs in smooth muscle (31), cardiac muscle (43), neurons (42), oocytes (48), and hepatic tissue (44) show that ER and mitochondria membranes can be as close as 10–300 nm. Most elegantly, by employing ER- and mitochondria-targeted GFP fusion constructs with high-resolution fluorescence microscopy, Rizzuto et al. (40) imaged the three-dimensional morphological relationship between these organelles. Their results in HeLa cells show that ER and mitochondria form a finely intertwined network, with ∼5–20% of the surface of the mitochondrial network in close apposition (<60 nm) to the ER (40, 41). These morphological estimates are consistent with functional measurements employing Ca2+ buffers to interfere with the transmission of Ca2+ signals between the ER and mitochondria, which points to an average separation in the 100-nm range (6,47).
However, because only a small part of the ER network appears to be in close apposition to the mitochondria, it is important to determine whether active Ca2+ release channels are clustered within such regions. High-resolution confocal Ca2+ imaging (33, 46, 51) and electrophysiological analysis (24) have shown that IP3 receptors are not homogenously distributed throughout the ER but, rather, are clustered to form discrete Ca2+ release sites. These sites generate local, “elementary” Ca2+ signals during stimulation with submaximal IP3 concentrations (33, 46,51). Therefore, the following questions arise: 1) What proportion of active Ca2+ release sites are in intimate association with mitochondria? 2) What might the functional consequences be for the generation of local and global Ca2+ signals?
The Xenopus laevis oocyte provides an excellent system in which to investigate these questions. First, confocal imaging techniques have resolved a microscopic architecture of local IP3-dependent Ca2+ signals, “Ca2+puffs” (33), which represent the activity of several IP3 receptors clustered within a functional Ca2+ release unit (24, 33, 46, 51). These signals arise at fixed positions in the ER over long periods of time (46), facilitating their mapping relative to the morphological architecture of the cell. Second, functional interactions between mitochondrial activity and global Ca2+ signals have been demonstrated in the Xenopus oocyte, where intracellular injection of oxidizable substrates increased the amplitude, velocity, and period of repetitive IP3-evoked Ca2+ waves (22) and disrupted their spatial patterning (11).
In this study, we exploit these advantages of the Xenopusoocyte as a model system to map Ca2+ puff sites in relation to mitochondria. By correlating these morphological data with functional analyses of Ca2+ puff behavior, we show that Ca2+ release sites located close to mitochondria exhibit a lower frequency of Ca2+ puffs than sites in regions where mitochondria are scarce. Ca2+ puff sites close to mitochondria rarely trigger Ca2+ waves; instead, waves are initiated at sites with low surrounding mitochondrial density. These results suggest that mitochondrial Ca2+ handling plays an important role in determining the local excitability of the ER as well as the global cellular architecture of IP3-evoked Ca2+ release.
MATERIALS AND METHODS
Preparation and Use of Xenopus Oocytes
Experiments were performed using defolliculated X. laevis oocytes, isolated as described previously (26,46). Female frogs were anesthetized by immersion in a 0.15% solution of MS-222 (3-aminobenzoic acid ethyl ester) for 20 min, and several ovarian lobes were removed by surgery. Epithelial cell layers were removed from the oocytes manually using watchmakers' forceps and then briefly (30 min) exposed to collagenase (0.5 mg/ml; type I, Sigma) for removal of follicular cells. Defolliculated oocytes were cultured at 16°C in Barth's solution [88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO3, 0.83 mM MgSO4, 0.33 mM Ca(NO3)2, 0.41 mM CaCl2, 10 mM HEPES, 550 mg/l sodium pyruvate, 0.05 mg/ml gentamicin]. For Ca2+ imaging studies, oocytes were microinjected 30 min before recording with Oregon green 488-1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid-1 (OG-1, dissociation constant for Ca2+ of ∼170 nM) or fura red (dissociation constant for Ca2+ of ∼140 nM) to final intracellular concentrations of ∼40 and ∼200 μM, respectively, together with caged IP3 (5 μM final concentration). Confocal Ca2+ imaging was performed using oocytes incubated in Barth's solution at room temperature from albino and pigmented donors.
For expression studies, a Drummond microinjector was used to inject plasmid cDNA (∼2 μg in 5 nl) encoding an ER-targeted yellow fluorescent protein (ER-EYFP) and a mitochondrially targeted red fluorescent protein (mito-dsRED) into the nucleus of pigmented X. laevis oocytes. Injected oocytes were separated into 96-well plates and incubated in Barth's solution, with repeated changes of solution at least every 12 h. Control oocytes from the same donor frogs were injected with the same volume of intracellular solution alone (in mM: 140 KCl, 10 HEPES, 3 MgCl2, 1 EGTA, 0.5 CaCl2). Typically, the distributions of ER-EYFP [wavelength (λ) > 530 ± 20 nm] and mito-dsRED (λ > 620 nm) were imaged ≥48 h after injection, and the intensity of fluorescence expression in the control cells was <1% of that in positively expressing oocytes.
For labeling of mitochondria, oocytes were incubated in 100 nM MitoTracker Green (MTG) or 500 nM MitoTracker Red (MTR) for 30–60 min on a rotator at room temperature and then washed for 30 min with two changes in Barth's solution to remove extracellular dye. For 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolecarbocyanine iodide (JC-1) staining, oocytes were incubated in 10 μg/ml JC-1 for up to 30 min. For C26H27ClN2O7[tetramethyl rhodamine ethyl ester (TMRE)] staining, background staining of oocyte lipid was noticeably high and was best minimized by incubating for short periods of time (<10 min) with much higher concentrations of dye (up to 1 μM) than routinely used for staining mitochondria in mammalian cells. Emitted fluorescence was then monitored at λ > 560 nm for TMRE and by dual emission (λ = 530 ± 20 nm and λ > 560 nm) for JC-1. For MTG, OG-1, and ER-EYFP, emitted fluorescence was monitored at λ = 530 ± 20 nm. For MTR, fura red, and mito-dsRED fluorescence, fluorescence emission was collected with a 620-nm long-pass filter.
Confocal imaging was performed using custom-built slow-scan (32) and video-rate confocal microscopes (3). The slow-scan system was used in line-scan (x-t), frame-scan (x-y), and axial-scan (x-z) modes. Fluorescent signals in line-scan mode were monitored along a fixed line formed by repeated scans of a focused laser spot from an argon ion laser (λ = 488 nm), with emitted fluorescence monitored through confocal pin holes using avalanche diode photon counting modules. For frame scanning (x-y), the scan line was driven in the lateral plane using a motor linked to the confocal scan head mount, and for axial scans (x-z), a motorized focus unit was used to drive the scan line in the vertical plane (32).
The video-rate system employed a resonant scanner coupled to an Olympus IX50 microscope to image a 65-μm2 region (450 × 450 pixels; 30 Hz) focused at various depths into the oocyte (3). Excitation was by a 488-nm laser line, and fluorescence emission was detected by a photomultiplier through a confocal aperture providing an optical section of ∼1 μm. The lateral resolution of this system is ∼0.3 μm (3). This system was also used in video-rate axial (x-z) mode by turning off the y-scan mirror and, instead, focusing the objective lens at a rate of ∼30 Hz using a piezoelectric translator (3).
For Ca2+ imaging experiments on both microscopes, flashes of ultraviolet light (340–400 nm) from a mercury arc lamp were used to photorelease IP3 from a caged precursor over a wide field area (∼150 μm diameter) centered around the laser scan. Ca2+ images are displayed as ratios (ΔF/F0), depicting the change in fluorescence at a pixel during the response (ΔF) relative to the resting fluorescence at the same pixel before stimulation (F0).
3T3 fibroblasts were grown (37°C, 5% CO2) in Dulbecco's modified Eagle's medium containing l-glutamine (4 mM), sodium bicarbonate (1.5 g/l), glucose (4.5 g/l), and 10% bovine calf serum. Cells were grown on poly-d-lysine-coated petri dishes, stained with 50 nM MTG for 20 min, and subsequently imaged by confocal microscopy at room temperature.
MTG (C34H28Cl5N3O), MTR (C32H33ClN2O), JC-1, OG-1, fura red, caged IP3 (caged myo-IP3), and TMRE were obtained from Molecular Probes (Eugene, OR), nocodazole from Calbiochem (La Jolla, CA), ER-EYFP and mito-dsRED from Clontech, and 3T3 fibroblasts from American Type Culture Collection (CCL-92). All other reagents were obtained from Sigma Chemical (St. Louis, MO).
To investigate mitochondrial organization in theXenopus oocyte, we first established the basic morphology and distribution of mitochondria using confocal microscopy and then proceeded to correlate mitochondrial localization with the spatial distribution and functional properties of IP3-evoked Ca2+ release sites in the ER.
Staining of Mitochondria in the X. laevis Oocyte
We examined the suitability of a variety of fluorescent dyes (Figs. 1 and2), as well as subcellularly targeted fluorescent protein variants (Fig. 3), to selectively stain mitochondria in the Xenopus oocyte. With the vital dyes, the selectivity of mitochondrial staining was consistently better with MTG than with MTR; both of these dyes exhibited far less background staining than JC-1 or TMRE, which accumulated within lipid vesicles. The basis of this difference is unknown but may relate to the dependence on mitochondrial membrane potential for dye accumulation. TMRE and JC-1 are potentiometric stains, while MitoTracker staining, especially with MTG, appears to be less sensitive to membrane potential (35). We also used mito-dsRED to visualize mitochondrial organization (Fig. 3). Although the specificity of staining was presumably improved, two factors limited the utility of mito-dsRED as a tool for studying mitochondrial morphology. First, the time for maturation of red fluorescence (1) was considerably longer (maximal fluorescence ∼100 h after injection) than that for visualization of ER-EYFP fluorescence (<30 h) and well within a time frame in which oocyte viability is decreasing after nuclear cDNA injection. Second, in accordance with the tendency of dsRED to aggregate (1), mitochondrial organization appeared less well defined than that observed with any of the vital stains.
Organization of Mitochondria Within the X. laevis Oocyte Cortex
Oocytes stained with MTG were imaged using a line-scan confocal microscope (32) operated in axial (x-z) or lateral (x-y) scanning modes. Low-magnification axial scans into oocytes from albino frogs revealed a cortical band of fluorescence that extended several micrometers into the animal and vegetal hemisphere of all oocytes examined (Fig. 1 A). In higher-magnification axial scans from albino oocytes (50-μm laser scan line, Fig. 1, B and C), the peak fluorescence was consistently brighter in the vegetal than in the animal hemisphere (vegetal peak fluorescence was 148 ± 11.0% of that in the animal pole, n = 30 cells, 3 donors), although the width of the mitochondrial band was similar in the vegetal [6.6 ± 0.4 μm, full width at half-maximal intensity (FWHM)] and animal hemispheres (6.2 ± 0.4 μm). However, these estimates of organelle distribution are probably limited by decreases in optical resolution with increasing depth into the turbid cytoplasm of the oocyte. Similar results were obtained in albino oocytes stained with MTR (data not shown).
In pigmented oocytes, relatively little mitochondrial staining could be detected in the animal hemisphere. However, it is likely that the heavy pigmentation in the animal pole attenuates fluorescence emission from the mitochondria located there, and not that the asymmetric distribution of mitochondria is more pronounced in pigmented oocytes. Consistent with this, treatment of pigmented oocytes with the microtubule depolymerizing drug nocodazole (10 μM, 6–10 h), which disorganizes the pigment distribution across the animal pole, revealed mitochondrial structures that appeared brighter and more radially extended than those in control cells (data not shown).
Lateral (x-y) images obtained at various focal depths within the vegetal hemisphere of pigmented and albino oocytes revealed a similar morphology of mitochondrial organization (Fig. 1 D). At the depth of peak mitochondrial fluorescence (0 in Fig.1 D, ∼5–8 μm inward from the highly convoluted oocyte plasma membrane), MTG staining revealed large “island” masses comprising many densely packed intertwining strands (Fig.2 A). Individual masses appeared to be interconnected by one or more individual strands, thereby establishing an extensive mitochondrial network. In more superficial sections (labeled in positive micrometer increments), individual mitochondria extended upward from the densely packed islands, ramifying radially outward toward the plasma membrane. In deeper cross sections, the mitochondrial islands extended downward into the oocyte beyond the limit of our resolution (∼15 μm into the oocyte, owing to the optical turbidity of the lipid-filled cytoplasm).
Several lines of evidence suggest that MTG faithfully reports mitochondrial morphology in Xenopus oocytes. First, the fluorescence-staining pattern in the oocyte is similar to the morphology reported by other cell-permeant mitochondrial dyes. Figure 2shows the similar staining pattern at the level of the mitochondrial islands in oocytes incubated with MTG (Fig. 2 A), MTR (Fig.2 B), TMRE (Fig. 2 C), and JC-1 (Fig.2 D). Second, the staining pattern of individual mitochondria with MTG closely resembles that observed in mammalian cell types; e.g., 3T3 fibroblast cells (Fig. 2 F, inset). The width of individual mitochondrial strands measured in the oocyte (FWHM ∼ 0.44 ± 0.08 μm; Fig. 2 E) is consistent with measurements of mitochondrial size from mammalian cells (FWHM ∼ 0.3 μm; Fig. 2 F).
Morphological Relationship of ER and Mitochondria
To visualize the distribution of ER relative to mitochondria, we expressed ER-EYFP together with mito-dsRED by injecting appropriate cDNA constructs into the nuclei of pigmented oocytes. Oocytes expressing both green and red fluorescence were selected (n = 20 cells), and the morphology of both organelle networks was visualized by confocal imaging with dual-wavelength emission. In axial (x-z) scans in the vegetal hemisphere, ER-EYFP fluorescence was distributed in a band 8.2 ± 0.5 μm wide (Fig. 3 A; FWHM). The red mitochondrial fluorescence was similarly localized to a band 7.9 ± 0.3 μm wide, with the peak mitochondrial intensity located, on average, 3.3 ± 1.4 μm deeper within the oocyte than the peak ER fluorescence (Fig.3 A). This axial separation of ER and mitochondria was further confirmed in oocytes (n = 6 cells) expressing ER-EYFP alone after staining with MTR (mitochondrial peak fluorescence was 2.5 ± 0.4 μm deeper within the oocyte than the peak ER fluorescence intensity).
Figure 3, B and C, illustrates lateral (x-y) sections showing dual-fluorescence overlays taken at the level of peak ER-EYFP fluorescence (green arrow, Fig.3 A) and at the level of peak mito-dsRED fluorescence (red arrow, Fig. 3 A). At the depth into the cell where the anastomosing ER network was most dense (Fig. 3 B), mitochondrial tubules were sparsely distributed and predominantly orientated as individual strands projecting outward from the deeper mitochondrial islands. In deeper sections (Fig. 3 C), ER and mitochondrial fluorescence were more closely associated, with large areas devoid of either organelle that were presumably filled by lipid vesicles in the ooplasm.
Mitochondria in Xenopus Oocytes Are Relatively Immobile
To map the relative locations of mitochondria and Ca2+release sites in the oocyte, we first needed to establish that both of these structures remain immobile throughout the time required (<3 min) for imaging. Using line-scan (46) and video-rate (26) confocal microscopy, we previously showed that Ca2+ puffs can be reproducibly mapped to discrete locations in the cytoplasm, with the same sites repeatedly responding from the same fixed positions over long periods (≤20 min).
To assess mitochondrial motility, we compared high-magnification (10 × 10-μm regions) time-lapse images of mitochondria stained with MTG in 3T3 fibroblasts with similar images from Xenopusoocytes (Fig. 4). Figure 4 shows two montage panels of successive images (3 s apart) of mitochondria in a 3T3 fibroblast (Fig. 4 A) and a Xenopus oocyte (Fig. 4 B). In the fibroblast, a large degree of mitochondrial movement was evident, with existing mitochondria moving around within the confocal plane and previously out-of-focus mitochondrial strands entering and traversing the section during the recording period. To better convey the extent of mobility, up to 180 individual frames from a 3-min time-lapse stack (1 s/frame) were projected as a maximal-intensity overlay on top of the original image of mitochondrial location. Colored regions represent the original mitochondrial position, whereas white areas reflect the extent of organellar travel during the recording period. The same algorithm was applied to records from 3T3 fibroblasts (Fig. 4 C) andXenopus oocytes (Fig. 4 D). Figure 4, Cand D, shows that whereas mitochondria are dynamic organelles in fibroblasts (even at room temperature), mitochondrial position in the oocyte is largely static. These data reporting a predominantly static location of mitochondria and puff sites over the time frame of our experiments facilitated our attempts to map their respective positions.
Localization of Ca2+ Puff Sites With Mitochondria by Confocal Microscopy
To correlate the localization of mitochondria with the sites of Ca2+ release from the ER, we used two different methods, each of which afforded specific advantages. First, we exploited the high spatial resolution [∼0.3-μm lateral resolution (32)] of the line-scan confocal microscope to map elementary Ca2+ release sites (Ca2+ puffs) relative to mitochondria. This afforded the highest-resolution measurements but has the drawback that information is restricted to a single spatial dimension (the laser-scan line), making it difficult to appreciate the three-dimensional relationship between sites of Ca2+ release and mitochondria. Therefore, as an additional method, we employed a video-rate confocal microscope used in lateral (x-y) and axial (x-z) scanning modes (3). Although the spatial and temporal resolution of the video-rate system was poorer than that of the line-scan system, the ability to image rapidly (30 Hz) in two spatial dimensions facilitated data measurement and reconstruction of the three-dimensional organization of Ca2+ puff sites in relation to mitochondrial morphology.
To allow simultaneous monitoring of the localization of mitochondria and Ca2+ puff sites, we used combinations of fluorescent dyes with well-separated fluorescence emission spectra. For the line-scanning experiments, we monitored green fluorescence from mitochondria stained with MTG, together with Ca2+-dependent changes in red fluorescence emission of fura red (note that fura red fluorescence decreases on binding Ca2+). Figure5 A shows fluorescence line-scan images from oocytes microinjected with caged IP3together with fura red. Whereas strong photolysis flashes evoked a Ca2+ wave shortly after the photolysis flash (Fig.5 A, top), weaker photolysis flashes (∼50–80% of the intensity that evoked a Ca2+ wave) evoked discrete, localized Ca2+ puffs at various positions along the laser-scan line (Fig. 5 A, bottom).
To image Ca2+ puff sites and mitochondria, anx-y image of mitochondrial distribution (MTG, λ = 530 ± 20 nm) was first obtained with the confocal section focused at a depth corresponding to the mitochondrial islands (∼5–8 μm into the oocyte; Fig. 5 B). The microscope was then switched to line-scan mode, and fura red fluorescence (λ > 620 nm) was monitored at the same depth along a scan line at a defined lateral (y) position on the mitochondrial x-y image. Analysis was restricted to bright, sharply resolved Ca2+release events, which presumably arise from Ca2+ puff sites located on, or very close to, the laser-scan line. Thex-coordinate of the center of the fluorescence signal from individual Ca2+ puffs (presumably corresponding to the site of Ca2+ release) was then transposed onto the mitochondrial image (Fig. 5 B), and the lateral distance from the edge of the nearest mitochondrial fluorescence was recorded (whether mitochondrial island or discrete mitochondrial strand). The resulting distribution from a total of 292 discrete Ca2+ puff sites is shown in Fig. 5 C (21 oocytes, 5 donor animals). This was compared with the distribution expected if Ca2+ puff sites were randomly located throughout the imaging field, without any preferred association with mitochondria (Fig. 5 C). The mean separation between Ca2+ puff sites and mitochondria was slightly smaller (2.33 μm) than expected from a random distribution (2.80 μm), owing to a larger proportion of Ca2+ puff sites closely juxtaposed to mitochondria (50.2% of Ca2+puff sites within <2 μm of a mitochondrion compared with only 40.0% expected from the control distribution).
Video-rate lateral scan (x-y) imaging.
For video-rate experiments, we monitored red fluorescence from mitochondria stained with MTR, together with Ca2+-dependent increases in green fluorescence emission of OG-1. Repeated photolysis flashes were used to evoke Ca2+ puffs at numerous sites throughout the imaging field during a 3-min recording period. The locations of Ca2+ puffs were then mapped onto a composite mitochondrial image formed as a maximum-intensity projection of confocal sections at the depth used for Ca2+ imaging and at 0.5 μm above and below the site (Fig.6 A). Measurements of the lateral separation between individual Ca2+ puff sites and the edge of the nearest mitochondrial fluorescence are shown in Fig.6 B, together with a control distribution derived from randomly selected points on the same composite image. The mean distance between Ca2+ puff sites and mitochondria was smaller than expected if Ca2+ puff sites were randomly distributed (2.27 vs. 2.96 μm, n = 355 puff sites, 16 oocytes, 4 donor animals), and a greater proportion of Ca2+ puff sites (50.3%) were located within <2 μm of the nearest mitochondrial mass than expected from a random distribution (37.6% of sites within <2 μm).
Data derived from the line-scan (Fig. 5) and frame-scan imaging techniques (Fig. 6) indicate that Ca2+ puff sites are located, on average, at a lateral distance of ∼2.3 μm from the closest neighboring mitochondrion but that a subpopulation of Ca2+ release sites is more closely localized with mitochondria than would be expected from a random distribution (Figs.5 C and 6 B).
Video-Rate Axial Imaging
The experiments described above provided information on the spacing between Ca2+ puff sites and mitochondria in a lateral plane (i.e., parallel to the “plane” of the cell membrane) but give no information regarding their relative radial spacing (i.e., whether Ca2+ puff sites lie “above” or “below” mitochondria). To address this question, we performed experiments in which Ca2+ puffs were captured by axial video-rate scans in oocytes stained with MTR (3). Experiments were done as described above, except, instead of scanning the laser line laterally in the y-dimension, a piezoelectric translator was used to focus the microscope objective at 30 Hz in the axial (z) dimension. Each video frame thus represents a “side-on” (x-z) view into the oocyte. Figure7 A shows a composite image of two Ca2+ puffs captured at times when their fluorescence intensities were maximal, superimposed on a image of mitochondrial distribution and resting OG-1 fluorescence in the same image field. The location of Ca2+ puffs was identified from the sites of initial fluorescence increase during the local Ca2+transients measured with OG-1, before this became appreciably blurred by diffusion of Ca2+ within the cytoplasm. Figure7 B shows measurements of radial (z) separation between the position of Ca2+ puff sites (144 sites, 5 oocytes) and the peak axial mitochondrial fluorescence. The majority of Ca2+ puff sites (∼70%) were localized in a band ∼2.5 μm wide, centered ∼2 μm superficial to the peak mitochondrial fluorescence.
Modulation of Ca2+ Puff Characteristics by Closely Apposed Mitochondria
The results presented above indicate that although the majority of Ca2+ puff sites are located a few micrometers from mitochondria, a subpopulation of deeper Ca2+ release sites may be more intimately apposed to mitochondria. We were therefore interested to determine whether the properties of the closely juxtaposed sites differed from the more-distant Ca2+release sites.
From video-rate imaging experiments such as those shown in Fig. 6, the properties of Ca2+ puffs at particular Ca2+release sites were measured and correlated with measurements of the surrounding density of mitochondria (Fig.8). All measurements of puff activity were made in oocytes at rest, without specifically energizing mitochondria with oxidizable substrates (22). Under these conditions, no differences were apparent between the magnitude of Ca2+ puffs generated at sites most closely juxtaposed to mitochondria (<1.25-μm separation) and the overall population of Ca2+ puff sites: (F/F0) = 1.62 ± 0.3 for peak fluorescence of Ca2+ puffs situated <1.25 μm from mitochondria and 1.56 ± 0.1 for a random selection of Ca2+ puff sites (n = 15 and 42 puff sites, respectively). However, a marked difference was apparent in Ca2+ puff frequency. Figure 8 A shows the frequencies of Ca2+ puffs generated at multiple sites during sustained photorelease of IP3 as a function of their localization with mitochondria. Proximity was quantified in terms of a “mitomass” parameter, which provides a measure of the density of mitochondria throughout a volume of ∼27 μm3 surrounding each Ca2+ puff site. Ca2+ puff sites surrounded by a higher density of mitochondria displayed progressively lower frequencies of Ca2+ puffs (Fig. 8 A). Furthermore, in each imaging frame, the Ca2+ puff site with the highest frequency of activity (average frequency = 3.5 puffs/min, n = 11 oocytes, 3 donor animals) was surrounded by a low mitochondrial density, whereas the Ca2+puff site most intimately associated with mitochondria displayed a much lower frequency of activity (Fig. 8 B; average frequency = 0.6 puffs/min).
Sites of Ca2+ Wave Initiation
We recently showed that a subset of Ca2+ puff sites repeatedly serves as the site of Ca2+ wave initiation during Ca2+ oscillations in the oocyte (26). Those Ca2+ puff sites where Ca2+ waves initiate, called “focal” Ca2+ puff sites, display a higher IP3 sensitivity and a higher frequency of Ca2+ puff activity than other sites (26). We were therefore interested to compare mitochondrial density associated with focal puff sites with a random sample of “nonfocal” Ca2+ puff sites in the same record. Focal Ca2+puff sites were identified in MTR-stained oocytes as the sites of Ca2+ wave initiation during repetitive Ca2+oscillations evoked by sustained photorelease of low concentrations of IP3. Measurements of the mitomass associated with these focal sites showed that sites of Ca2+ wave initiation were associated with a lower mitochondrial density than Ca2+ puff sites that did not initiate Ca2+ waves (Fig. 8 C).
Mitochondrial Morphology in the Xenopus Oocyte
There have been few previous studies documenting mitochondrial morphology in intact Xenopus oocytes commonly used for Ca2+ imaging (stages V and VI). Therefore, using confocal imaging, we began by examining the morphological relationship between mitochondria, ER, and sites of intracellular Ca2+ release within live X. laevis oocytes.
MTG and MTR staining revealed large islands of densely intertwined mitochondria, interconnected by discrete vermiform mitochondrial strands, establishing an extensive cortical network spreading a ∼6- to 8-μm-wide band around the entire oocyte cortex. The appearance of mitochondrial islands is reminiscent of the mitochondrial clustering shown to occur within peripheral corridors of yolk-free cytoplasm in oocyte electron micrographs (2). Denser mitochondrial staining was evident within the vegetal hemisphere than within the animal hemisphere of the oocyte (Fig. 1 C), possibly reflecting the asymmetric dispersal of the mitochondrial cloud of previtellogenic oocytes to the cortex of the vegetal pole (50). Although the mitochondrial network is highly dynamic, undergoing continuous rearrangement in many cell types (40), the mitochondrial network of unenergized oocytes was predominantly static (Fig. 4). Given the influence of cell cycle and metabolic state on mitochondrial morphology, the static nature of oocyte mitochondria is not unexpected; the oocyte is in G2arrest, and mitochondrial potentials are low at rest (∼93 mV) (22). Whatever the explanation, the lack of mitochondrial dynamics facilitated long-term mapping of mitochondria in relation to Ca2+ release sites.
Relationship of Ca2+ Puff Sites to Mitochondria
Three main conclusions can be drawn from the morphological and functional data presented in Figs. 5-7. First, the mean separations between Ca2+ release sites and their neighboring mitochondrial islands (∼2.3 μm) are appreciable compared with the spatial spread of Ca2+ during a Ca2+ puff (effective space constant for Ca2+diffusion of ∼2.8 μm). Most mitochondria are, therefore, located well away from the microdomain of high cytosolic Ca2+concentration surrounding puff sites and, as a consequence of the low affinity for mitochondrial Ca2+ uptake, are unlikely to buffer the Ca2+ released during Ca2+ puffs. Second, a major factor determining the separation between Ca2+ puff sites and mitochondria is that most Ca2+ puff sites are concentrated in a band located ∼2 μm more peripheral in the oocyte cortex than the location of the mitochondrial islands (Fig. 7). Finally, a proportion of Ca2+ puff sites (∼28%) are located in close juxtaposition with mitochondria (within the lateral resolution of the confocal microscope, <400 nm) (3, 32). These Ca2+ puff sites most likely represent the deeper sites in the cortical ER, which extends downward into the layer where mitochondrial density is highest.
Our results thus demonstrate a heterogeneous spacing between Ca2+ puff sites and mitochondria (Figs. 5 and 6), which is expected to establish a broad range of functional cross talk between IP3 receptors and mitochondria. If a comparable range of separation also occurs in other cell types, then such morphological heterogeneity could contribute to the diversity in agonist-evoked Ca2+ signals (5, 29, 36) and membrane potential responses observed from discrete mitochondrial regions (9).
Role of Mitochondria in the Spatial Patterning of Local Ca2+ Signals
Does mitochondrial proximity influence the behavior of local IP3 receptor clusters in the Xenopus oocyte? The frequency of Ca2+ puffs at sites with low mitochondrial density was relatively high (Fig. 8, A and B), and these sites also preferentially served as initiation foci for Ca2+ waves (Fig. 8 C). We recently reported that particular focal Ca2+ release sites tend to repeatedly initiate Ca2+ waves and that these sites are functionally distinguished by higher sensitivity to IP3 and greater Ca2+ puff frequency than are nonfocal sites (26). The present results indicate that mitochondrial proximity is likely to be an important factor underlying these differences, although the broad range of activity exhibited by Ca2+ puff sites in areas of sparse mitochondrial density suggests that other factors are also likely to be involved. The basis for these effects of mitochondria on Ca2+ puff activity is unclear but, speculatively, might include effects of local Ca2+ (34) or ATP gradients (25) on the Ca2+-induced Ca2+ release sensitivity of the Xenopus IP3 receptor. These results in theXenopus oocyte parallel those of Gordienko et al. (13) in smooth muscle myocytes, where mitochondria are sparse around Ca2+ spark sites with a high frequency of activity that serve as the focus of Ca2+ wave initiation. A low mitochondrial density surrounding sites of Ca2+ wave initiation has also been reported in endothelia (30) and pancreatic acinar cells (49), whereas in hepatocytes, heightened IP3 receptor sensitivity is evident in subcellular regions lacking mitochondria (17). However, in the processes of oligodendrocyte progenitors, local Ca2+signals develop into Ca2+ waves not in regions defined by low mitochondrial density, but only at sites where mitochondria are present (15, 45).
A second question concerns whether local IP3-dependent Ca2+ puffs stimulate mitochondrial Ca2+ uptake. In the oocyte, the separation between Ca2+ puff sites and mitochondria suggests that this is unlikely to be the case for most mitochondria, but a direct answer to this question necessitates simultaneous measurement of cytoplasmic and mitochondrial Ca2+ or mitochondrial potential. This is difficult given the nonspecificity of mitochondrial staining in the oocyte with TMRE or acetoxymethyl ester (AM)-loading protocols using Ca2+-indicator dyes such as rhod 2-AM (5, 7, 8,29) and dihydro-rhod 2-AM (18). The presumed ineffectiveness of discrete Ca2+ puffs in signaling to mitochondria certainly does not preclude the existence of preferential, and even graded, communication between the organelles. However, such cross talk probably occurs under stronger conditions of stimulation, as demonstrated during global Ca2+ oscillations in the oocyte (11, 22).
In other cell types where the geometry is different and mitochondria are more closely apposed to Ca2+ release sites, discrete elementary Ca2+ release signals can be sensed by mitochondria. For example, in rat cardiac myocytes, local Ca2+ release events resulting from activation of ryanodine receptors in sarcoplasmic reticulum Ca2+ stores (“Ca2+ sparks”) are sufficient to trigger transient depolarizations of mitochondrial membrane potential (9). In cardiac myocytes, ryanodine receptors are closer to mitochondria (∼40–270 nm) (43), and the majority of mitochondria share such intimate contact with the ER. Similarly, Ca2+sparks in smooth muscle appear competent in evoking transient mitochondrial depolarizations (7), where again both organelles are in close proximity (∼10–20 nm) (31). For IP3-evoked Ca2+ puffs, the situation is less clear; for example, in mast cells, it has been suggested that elementary Ca2+ signals may stimulate mitochondrial Ca2+ uptake on an individual basis (6), whereas in HeLa cells, Ca2+ puffs appear not to increase intramitochondrial Ca2+ concentration (5).
Role of Mitochondria in the Spatial Patterning of Global Ca2+ Signals
The emerging picture is that the relationship between ER and mitochondrial architecture is cell-type specific, with the respective morphology of both organelles molding the spatial patterning of Ca2+ release appropriate to cellular function. In the immature oocyte, the Ca2+ buffering capacity of the cortical mitochondrial band may serve to constrain Ca2+waves to the oocyte periphery, as observed during axial imaging (4). More speculatively, in mature oocytes, the same morphological arrangement would aid establishment of the fertilization envelope, rather than allow the fertilization Ca2+ wave to dissipate into the internal volume of the egg. This situation resembles, albeit on a much different scale, that seen in frog sympathetic neurons (28), where circumferential propagation of a Ca2+ wave on top of a mitochondrial band occurs in preference to radial Ca2+ elevations. Another example of a cell-specific use of a mitochondrial barrier to mold the spatial patterning of Ca2+ signals occurs in the pancreatic acinar cell, where a band of mitochondria envelop the apical pole of the cell, restricting localized Ca2+ signals to the region where the secretory granules are localized (49). When mitochondrial function is inhibited, local Ca2+oscillations are abolished and the agonist-evoked Ca2+signal pervades the entire cell (49). In all these cases, the role played by mitochondria in cellular Ca2+ dynamics appears tailored to cellular function, with mitochondria influencing the spatial patterning of Ca2+ release through their global capacity for Ca2+ buffering and through more localized effects on ER excitability.
This work was supported by National Institute of General Medical Sciences Grant GM-48071 and Wellcome Trust Fellowship 053102 to J. S. Marchant and National Institute of General Medical Sciences Minority Biomedical Researchers Program Studentship GM-55246-03 to V. Ramos.
Address for reprint requests and other correspondence: I. Parker, Dept. of Neurobiology and Behavior, University of California, Irvine, CA 92697-4550 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published February 6, 2002;10.1152/ajpcell.00446.2001
- Copyright © 2002 the American Physiological Society