Flirting in Little Space: The ER/Mitochondria Ca2+ Liaison

See allHide authors and affiliations

Science's STKE  13 Jan 2004:
Vol. 2004, Issue 215, pp. re1
DOI: 10.1126/stke.2152004re1


Mitochondria have long been known to accumulate Ca2+; the apparent inconsistency between the low affinity of mitochondrial Ca2+ uptake mechanisms, the low concentration of global Ca2+ signals observed in cytoplasm, and the efficiency in intact cells of mitochondrial Ca2+ uptake led to the formulation of the "hotspot hypothesis." This hypothesis proposes that mitochondria preferentially accumulate Ca2+ at microdomains of elevated Ca2+ concentration ([Ca2+]) that exist near endoplasmic reticulum (ER) Ca2+ release sites and other Ca2+ channels. Physiological Ca2+ signals may affect mitochondrial function--both by stimulating key metabolic enzymes and, under some conditions, by promoting apoptosis. Mitochondria in turn may affect both Ca2+ release from the ER and capacitative Ca2+ entry across the plasma membrane, thereby shaping the size and duration of the intracellular Ca2+ signal. Interactions between mitochondria and the ER are critically dependent on the spatial localization of mitochondria within the cell. The molecular mechanisms that define the organization of mitochondria with regard to the ER and other Ca2+ sources, and the extent to which mitochondrial function varies among different cell types, are open questions whose answers remain to be determined.


Calcium ions (Ca2+) probably represent the most ubiquitous signaling pathway in all cells (1). Changes in Ca2+ concentration mediate most of the short-term events that define changes in cell function. Mitochondria, however, represent a fundamental constant presence in the cell, responsible for maintaining cellular energy supply and guardians of the protein machinery that initiates programmed cell death (2-4). The reciprocal effects of Ca2+ on mitochondria and of mitochondria on the Ca2+ signal have fascinated biochemists, physiologists, and cell biologists for many years, but we now seem to be reaching an experimental phase where we entertain the conceit that at last we really begin to understand these fundamental processes.

Ca2+ uptake by isolated mitochondria was first described in the late 1950s [for review, see (5)], and it was soon demonstrated not only that the organelles can accumulate massive amounts of the cation, but also that this phenomenon could be fueled by either oxidation of respiratory substrates or by adenosine 5′-triphosphate (ATP) hydrolysis. The mechanism of such uptake remained mysterious until the chemiosmotic hypothesis provided the theoretical background to understand its energetic basis (6). The uptake of Ca2+ by mitochondria is, in fact, not mediated by pumps or exchangers, but by a "uniporter" (7), possibly a gated channel, although the molecular identity and nature of the uniporter remains unknown. The uniporter provides a pathway for the accumulation of Ca2+ into the mitochondrial matrix, driven by an electrochemical potential gradient across the inner mitochondrial membrane, usually estimated at ~200 mV negative to the cytosol, and generated either by the respiratory chain or ATP hydrolysis.

If Ca2+ accumulation were governed solely by thermodynamic parameters, at equilibrium, according to the Nernst equation, the Ca2+ concentration in the matrix should be about one million times the concentration outside the mitochondria, a value much higher than experimentally measured. A first hint to understanding how steady-state accumulation of Ca2+ in mitochondria could be prevented from reaching thermodynamic equilibrium came from the observation that addition of selective inhibitors of the uniporter such as Ruthenium Red or La3+ to mitochondria loaded with Ca2+ induces a slow and complete release of the accumulated cation (8, 9). It was then demonstrated that such Ca2+ efflux is catalyzed by antiporters that drive Ca2+ out of the mitochondrial matrix in exchange with either Na+ or H+ (7). Two types of exchangers were functionally characterized in the 1970s, the Na+/Ca2+ and the H+/Ca2+ exchangers; the former are particularly abundant in excitable tissues [for review, see (5)]. Although the molecular nature of the antiporters is still unknown, good evidence exists that the Na+/Ca2+ antiporter is electrogenic(3 Na+ exchanged for 1 Ca2+), similar to the well-characterized plasma membrane Na+/Ca2+ antiporter, whereas the stoichiometry of the H+/Ca2+ antiporters has not been investigated in any detail [for review, see (10, 11)].

The presence of membrane potential-driven uptake and of concentration-dependent release mechanisms ensures that a steady state is reached when the rate of uptake equals that of release (12, 13). The existence of this futile Ca2+ cycle across the inner mitochondrial membrane is the basis for the concept of the "mitochondrial Ca2+ set point," meaning the steady-state extramitochondrial Ca2+ concentration ([Ca2+]) that is maintained by energized mitochondria (7). The Ca2+ cycle would be energetically quite expensive (about 12 Ca2+ ions are transported per molecule of O2 consumed), except that the Ca2+ affinity of the uniporter is quite low. Thus, when extramitochondrial Ca2+ is in the physiological range (below 1 μM), the rate of Ca2+ cycling becomes negligible. Any displacement from these steady-state conditions, such as an increase in ambient Ca2+, leads to a transient accumulation of the cation as the rate of influx overwhelms the slower rates of the efflux pathways [for recent reviews, see (5, 10, 11, 14-16)].

The functional characteristics of Ca2+ handling by isolated mitochondria were fairly well established by the end of the 1970s, and at that time, most researchers were convinced that these organelles comprised the key intracellular Ca2+ stores in living cells. The scenario changed dramatically at the beginning of the 1980s, when it was not only discovered that the total Ca2+ content of mitochondria in situ was negligibly low (17), but also that the Ca2+ mobilization from internal compartments elicited by receptor activation involved another cellular organelle, the endoplasmic reticulum (ER) (18). It is perhaps salutary to recall that the seminal paper (18) stressed this point; it was titled "Release of Ca2+ from a nonmitochondrial intracellular store in pancreatic acinar cells by inositol 1,4,5-trisphosphate" (IP3), which emphasized the prior perception of the importance of mitochondria. The implication of the ER in Ca2+ mobilization, coupled with the demonstration of the low affinity of the mitochondrial uniporter for Ca2+ (an apparent Kd of 20 to 30 μM under conditions thought to mimic the cytoplasm), convinced the majority of specialists that these organelles had little to do with physiological Ca2+ handling. These observations suggested that mitochondria would show only a very low rate of Ca2+ uptake at concentrations thought to occur not only in resting cells, but also in activated ones. The role of mitochondrial Ca2+ accumulation under conditions of massive cytosolic Ca2+ overload--in severely injured cells or highly specialized cells such as osteoclasts--was recognized during this time. Mitochondrial accumulation of Ca2+ under pathophysiological conditions of severe Ca2+ overload has been extensively reviewed and will not be discussed in the present article.

The demonstrations, at the beginning of the last decade, that mitochondria can rapidly accumulate Ca2+ under physiological conditions in living cells (19, 20), and that Ca2+ accumulation modulates mitochondrial metabolic efficiency (21, 22), affects calcium signaling (23-25), and can be a key factor in the activation of programmed cell death have revitalized interest in this process, far beyond the community of bioenergeticists or of Ca2+ physiologists. [For reviews, see, among others (4, 26-30).]

The "Hotspot Hypothesis"

Rationale and basic observations

Although several papers published during the 1980s indicated that mitochondria could accumulate a substantial proportion of the Ca2+ flooding the cytosol during activation of either plasma membrane or ER Ca2+ channels (31-33), these findings were either neglected or considered to be a peculiarity of specific experimental conditions. At best, the evidence for mitochondrial Ca2+ accumulation in situ was indirect. Thus, several studies demonstrated that mitochondrial depolarization by uncouplers after stimulation caused the intracellular release of sequestered Ca2+, which implied that mitochondria must have accumulated Ca2+ during stimulation (31). Other studies explored the calcium-dependent activation of mitochondrial metabolism implicit in the Ca2+ dependence of TCA (tricarboxylic acid) cycle enzymes demonstrated by McCormack et al. [ (34) and see below], and demonstrated Ca2+-dependent changes in mitochondrial redox state following the stimulation of several types of excitable cell (35, 36). Indeed, together, these studies showed that in intact cells, mitochondria must be accumulating Ca2+ when cytoplasmic Ca2+ concentration ([Ca2+]c) is high, but these data were nevertheless perceived by many as conflicting with an established literature on the isolated organelles. The first direct measurements of mitochondrial Ca2+ uptake in situ were made by selectively targeting the Ca2+-sensitive photoprotein aequorin to the mitochondrial matrix (19, 20). These studies showed that mitochondria in situ are much more efficient at taking up Ca2+ than predicted from their apparently low Ca2+ affinity. The first obvious explanation for these observations is that some cytosolic factors that increased effective mitochondrial Ca2+ affinity might be lost during organelle isolation. Indeed several factors that were able to slightly increase the apparent Ca2+ affinity of isolated mitochondria had been described, in particular polyamines such as spermine, but none of these agents could increase the Ca2+ affinity to levels comparable to those implied by the observations in situ (37-39).

The key experiment that gave rise to the Ca2+ "hotspot" hypothesis was published in 1993 (20). This experiment showed that the rate of mitochondrial Ca2+ accumulation elicited by adding IP3 to permeabilized cells was not only similar to that observed in intact cells stimulated through activation of G protein-coupled receptors, but comparable to that caused by exposing isolated mitochondria to about 10 μM Ca2+, a concentration 5 to 10 times the average value found in the cytosol upon maximal agonist stimulation. Because IP3 itself had no effect on isolated mitochondria, this experiment (repeated later in many cell types and with different protocols) suggested that release of Ca2+ through the IP3-gated ER channel created a microenvironment of Ca2+ concentration close to the mitochondria that was much higher than that measured in the bulk cytosol and high enough to activate the low affinity Ca2+ uniporter. The concept that the [Ca2+] close to the mouth of a Ca2+ channel reaches concentrations that exceed that in the bulk cytosol by over an order of magnitude was already well established--it had both been discussed at a theoretical level (40, 41) and addressed experimentally (42-44) for plasma membrane channels.

The hotspot hypothesis led to the following predictions: (i) There should be close apposition (in the 100-nm range) between ER and mitochondria and in particular between the IP3-gated channels and mitochondria. (ii) Some heterogeneity in the Ca2+ uptake rate should exist, at least transiently, among different mitochondria in the same cells. (iii) Mitochondria in the vicinity of Ca2+ channels should take up Ca2+ rapidly independently of the molecular nature of the channel, because hotspots should occur at the mouth of any Ca2+ channel, provided that the mitochondria are close enough. And (iv), the activity of Ca2+ channels might be modulated by mitochondrial Ca2+ uptake because the activity of many Ca2+ channels is modulated by local [Ca2+].

Morphological evidence

Close appositions between the ER (or sarcoplasmic reticulum, SR) and mitochondria have long been known to exist and have been observed by electron microscopy (EM) in fixed samples of several cell types. These regions have long been considered to constitute sites of phospholipid exchange between the two organelles. Although such close appositions between ER and mitochondria may, at least in part, represent fixation artifacts, an EM picture taken from quickly frozen samples, which prevents most artifacts of chemical fixation, reveals that such close appositions are not only visible, but are actually more frequent than in traditionally fixed samples (Fig. 1). By using a highly sensitive image deconvolution analysis in living HeLa cells expressing two differently colored green fluorescent proteins (GFPs) within the lumena of ER and mitochondria, it was demonstrated that a significant fraction of the organelles lie, in the living cell, within 100 nm of each other (45) (Fig. 2). Moreover, enriched IP3 receptor immunoreactivity in ER regions apposed to mitochondria has been demonstrated by immunocytochemistry in different cells, for example, in oligodendrocytes (46, 47).

Fig. 1.

Transmission electron micrograph of a PC12 cell. The field shows a PC12 cell processed by quick freezing-freeze drying (96) to preserve intracellular structures from fragmentation and extraction. Notice that most mitochondria (asterisks), looking like dense, elongated sausages, are located in the proximity of, or in apparent contact with, single or even double thin, long, and arborized ER cisternae (ER); N, nucleus. Magnification, ×25,000. [Picture kindly provided by Prof. J. Meldolesi (DIBIT, Milan, Italy)].

Fig. 2.

Contacts between mitochondria and the ER. (A) Three-dimensional rendering of mitochondria (red) and ER (green) in a living cell, labeled with mitotracker red and erGFP, respectively. (B) Display of the mitochondrial image from (A); mitochondrial subregions colocalizing with the ER are shown in red. (C) Series of different planes (through the z axis) of the mitochondrial image.

In most cells, the density of IP3 receptors at the ER membrane is so low that their identification at the electron microscopic level (and thus the demonstration that they are enriched in the regions of ER-mitochondria contact) is not feasible at present. However, in Purkinje neurons, IP3 receptors are so abundant that they are recognizable as electron-dense protrusions on the surface of the ER and stacks of ER cisternae rich in IP3 receptors are often found close to mitochondria (48, 49). Finally the proximity of mitochondria to Ca2+ release channels is firmly established in striated muscle, in particular in cardiac muscle, where ryanodine receptors can be identified as the feet structures of the SR membrane (50). Mitochondria can be as close as 5 nm from triads in rat cardiac cells, thus providing the structural basis for the tight coupling between Ca2+ mobilization from the SR and Ca2+ uptake by mitochondria in this tissue [(50) and see below].

Research on the morphological organization of mitochondria with respect to the plasma membrane, on the other hand, is much less extensive. Most experiments concerning the interaction between plasma membrane Ca2+ channels and mitochondria have been carried out in neurons. In a series of elegant experiments on rapidly frozen tissue samples, Pivovarova and co-workers (51, 52) not only demonstrated the efficiency of mitochondria in the cell body or dendrites at accumulating Ca2+ in response to depolarizing stimuli, but they also showed that mitochondrial Ca2+ accumulation depends on proximity to the plasma membrane. Again, this observation is consistent with a diffusion model in which a radial gradient of Ca2+ (highest closest to the plasma membrane) is established during opening of plasma membrane Ca2+ channels.

Although the morphological evidence supporting mitochondrial proximity to Ca2+ channels--and thus the first prediction of the hotspot hypothesis--appears reasonably well established, a major question concerns the localization of the Ca2+ uniporter on the mitochondrial inner membrane. In fact, given that the molecular identity of the uniporter (or any other component of the mitochondrial Ca2+ handling machinery) is unknown, there is no direct way of determining whether the mitochondrial uniporter is strategically enriched in regions close to the Ca2+ channels of the ER. Theoretically, three scenarios for uniporter distribution can be envisioned: (i) The uniporters are homogeneously distributed on the inner mitochondrial membrane. (ii) The uniporters are enriched or even exclusively localized in the regions of contact between the mitochondria and ER. (iii) The uniporters are somehow excluded from areas of close contact.

In the first case, only a fraction of the total Ca2+-accumulating capacity of the mitochondria will be exposed to microdomains with the highest [Ca2+]. In this case, the measured average rate of mitochondrial Ca2+ accumulation in situ should be lower than the maximal rate of Ca2+ uptake. In the second case, the rate of Ca2+ accumulation at saturating [Ca2+], and that obtained under physiological conditions, should be close or identical, because mitochondrial Ca2+ uptake at the hotspots will represent a large fractional component of total mitochondrial Ca2+ accumulation. The third possibility would simply disprove the hotspot hypothesis. A direct test of these different possibilities is impossible at present, but indirect calculations suggest that a large proportion of the mitochondrial Ca2+ uniporters are engaged during physiological stimuli--in other words, these calculations support the second scenario and, thus, the hotspot hypothesis (15).

A few additional points worth mentioning have emerged from the work of several laboratories. First, although numerous ER-mitochondria and ER-plasma membrane close interactions are morphologically evident in all cell types analyzed, some mitochondria appear isolated or relatively far from potential Ca2+ sources. By performing single-cell experiments with either the fluorescent Ca2+ indicator dye Rhod-2 or genetically engineered, targeted fluorescent Ca2+ probes, it has been possible to demonstrate that there are nonresponding or poorly responding mitochondrial structures (53-56). Finally, substantial heterogeneity in the rate and peak of Ca2+ uptake among different mitochondria has been observed (see below).

Several groups have tried to estimate the [Ca2+] reached at the hotspots, making the assumption that all Ca2+ uniporters are exposed to microdomains of high Ca2+. Their calculated values agree fairly well (within a factor of 2) and vary between 20 μM in RBL (rat basophilic leukemia) cells (57) to around 40 μM in adrenal medullary cells (58).

The kinetic properties of mitochondrial Ca2+ homeostasis: A challenge to the hotspot hypothesis?

The hotspot hypothesis has been widely accepted by the scientific community and has rarely been experimentally challenged. However, two experimental observations could argue against it. The first is that several groups have reported a lag time between the cytosolic [Ca2+] rise and the mitochondrial Ca2+ upstroke (53, 55, 59, 60). Indeed, microdomains of high Ca2+ should be generated as soon as channels open and, if anything, should decrease with time because of diffusion. Moreover, several seconds are needed to reach the mitochondrial [Ca2+] peak; in other words, mitochondrial Ca2+ uptake continues throughout the duration of the "bulk" cytosolic response, including the initial part of the declining phase (53, 55, 59-61).

With regard to the mitochondrial lag, several issues have to be taken into account. The first is simply technical and relates to intramitochondrial heterogeneity. Indeed, if Ca2+ uptake starts at highly localized sites, and Ca2+ then diffuses intraluminally, a whole-cell (or low-resolution) analysis of [Ca2+]m (the [Ca2+] in the mitochondrial lumen) would detect a lag, that (at least in part) depends on the time needed for [Ca2+] to rise in a substantial part of the mitochondrial network. In addition, the existence of mitochondria located far from Ca2+ hotspots inevitably leads to an overestimation of the lag time. In support of this possibility, a detailed kinetic analysis of mitochondrial Ca2+ uptake carried out on individual cells revealed that in HeLa cells there is substantial heterogeneity among individual (or groups of) mitochondria (53, 55). With a GFP-based Ca2+ indicator (53), it was shown that the lag between the cytoplasmic and mitochondrial rise times in Ca2+ was faster in some mitochondria than the temporal resolution of the measuring system (less than 160 ms) whereas in others, the lag could be greater than 500 ms.

However, other evidence suggests that a true delay may occur between the two compartments and points to a molecular mechanism. Rapizzi et al. (60) showed that the average mitochondrial delay in HeLa cells stimulated with an IP3-generating agonist, histamine, is substantially reduced (from 0.4 to 0.2 s) when the outer mitochondrial membrane channel VDAC (the voltage-dependent anion channel) is overexpressed, which suggests that the channel repertoire of the outer mitochondrial membrane at the contact between ER and mitochondria is important for allowing the high [Ca2+] microdomain to reach the environment of the low-affinity uniporter. Given the lack of molecular information, other mechanisms cannot be excluded: for instance, the existence of an allosteric Ca2+ binding site on the uniporter that retards the activation of the uptake by the uniporter (55). As to the duration of the mitochondrial uptake phase (longer than the cytoplasmic peak [Ca2+], an observation confirmed by numerous laboratories), a comparison of the effects of stimulating a rapidly desensitizing transfected metabotropic glutamate receptor with the endogenous histamine response of HeLa cells showed that sustained release through the IP3 receptor is required for generating a large mitochondrial response (62).

Finally, in intact neonatal cardiac myocytes, mitochondrial [Ca2+] oscillates in synchrony with cytoplasmic Ca2+ (63). Given that the [Ca2+]c transients have a rise time of 10 to 20 ms and a duration of less than 500 ms, it must be concluded that heart mitochondria can respond to such brief signals. The ability of mitochondria to respond to very rapid Ca2+ changes is not limited to heart cells. Heterogeneous Ca2+ accumulation in mitochondria at physiologically activated neuromuscular synapses has been demonstrated by EM in quickly frozen samples (64). Whether these differences depend on a different morphology or molecular composition of the contacts between ER and mitochondria awaits the elucidation of the molecular nature of these signaling microdomains and their key components (including the uniporter, the exchangers, and the docking proteins).

Overall, no experimental evidence contradicts the notion that mitochondria tend to accumulate Ca2+ most effectively at [Ca2+] microdomains at the mouth of a Ca2+ channel. Indeed, no other hypothesis can explain a key experimental observation [confirmed by numerous laboratories, including our own and those of Thomas, Hajnóczky, and Bootman, to name a few; see, for example (20, 21, 55, 57)]: A cytosolic Ca2+ rise achieved by maneuvers that cause a slower generalized increase in Ca2+, such as passive release of Ca2+ from the ER induced by blockers of the sarco-endoplasmic reticulum Ca2+ ATPases or perfusion of permeabilized cells with Ca2+ buffered media, produces substantially less mitochondrial Ca2+ accumulation than does a Ca2+ rise of the same amplitude produced through opening of Ca2+ channels. Additional strong evidence in support of the hotspot model is the relative insensitivity of the mitochondrial Ca2+ uptake rate, at least in some cell types, to cytoplasmic Ca2+ buffering. In cardiac cells, for example, concentrations of the Ca2+ chelator EGTA sufficient to practically abolish the cytosolic Ca2+ transient in response to caffeine are much less effective at inhibiting Ca2+ increases within mitochondria (65), which suggests that the distance is so small that Ca2+ diffuses from ER release channels into the mitochondria more rapidly than it can be buffered by EGTA.

In considering the mechanisms that lead to the heterogeneity in Ca2+ responses among different mitochondria, an additional factor needs to be taken into account: heterogeneity in mitochondrial membrane potential, as recently suggested by Collins et al. (56). It is also possible that, together with local hotspots, the speed of the cytosolic Ca2+ upstroke (much higher in the case of a physiological Ca2+ release), also plays a major role in the mitochondrial response, as might occur, for instance, if mitochondria undergo some form of rapid adaptation. In this case, an experiment with protocols that mimic the speed of the Ca2+ increase upon physiological stimulation (such as eliciting rapid Ca2+ release from a photosensitive Ca2+ chelator) would be very informative, but to our knowledge, such an experiment has not yet been carried out.

Finally, we wish to emphasize that the hotspot hypothesis should not be overinterpreted to imply that mitochondrial Ca2+ accumulation occurs only when the organelles are exposed to microdomains of high Ca2+ generated in the proximity of activated channels. This is clearly not the case, as seen from both the theoretical considerations mentioned above and experimental observations. Mitochondria do accumulate Ca2+ in response to modest increases in cytoplasmic Ca2+ (55), and in some cell types, this process is functionally relevant. For example, in luteal cells, submicromolar slow cytosolic Ca2+ increases have been shown to elicit very slow and small increases in mitochondrial [Ca2+] that were nevertheless sufficient to enhance mitochondrial pyridine nucleotide formation and steroid hormone synthesis (66).

Feeding back to the ER: A role for ER/mitochondria microdomains in shaping cytoplasmic Ca2+ signals

The rediscovery of the process of mitochondrial Ca2+ uptake has been paralleled by the appreciation of its role in regulating diverse cellular functions. Moreover, appreciation of its functional significance has greatly ignited interest in this physiological process. We will briefly review the main observations, paying closer attention to the events that occur at contact sites between the ER and mitochondria.

Within the mitochondrial matrix, two radically different effects can be triggered by a [Ca2+] rise. The first, as originally discovered by Denton, McCormack, and Hansford in the 1960s, is the activation of three key metabolic enzymes (the pyruvate, α-ketoglutarate and isocitrate dehydrogenases). Pyruvate dehydrogenase is activated through a Ca2+-dependent dephosphorylation step, whereas the other two enzymes are activated through direct binding of Ca2+ to the enzyme complex (34, 67). The transmission of a Ca2+ signal from the cytoplasm to the mitochondria thus activates the energy powerhouse of the cell, matching aerobic metabolism with energy demand. This elegant and simple process works most efficiently because most energy-dependent processes (such as contraction and secretion) are themselves signaled by a [Ca2+]c signal. Direct measurement of mitochondrial ATP levels, by using a targeted chimera of the ATP-sensitive photoprotein luciferase, demonstrated that the Ca2+ signal within mitochondria drives enhanced ATP production, an effect that lasts longer than the Ca2+ signal itself, which highlights a novel form of cellular "metabolic memory" (68).

Under some conditions, however, a Ca2+ signal within the mitochondria may trigger cell death. Work from various laboratories has revealed that excessive mitochondrial Ca2+ accumulation or the coincidence of a mitochondrial Ca2+ signal with proapoptotic signals or other pathophysiological conditions (such as oxidative stress) can profoundly alter organelle structure and function (69, 70). As a consequence, proteins normally retained in the organelle [such as cytochrome c (71)--an important component of the respiratory chain--and more recently discovered proteins, including apoptosis-inducing factor (72) and Smac/Diablo (73)] are released into the cytoplasm, where they activate effector caspases and drive cells to apoptotic death. In relation to this effect, the antiapoptotic oncogene Bcl-2 is apparently able to reduce steady-state [Ca2+] in the ER (and thus dampen the proapoptotic Ca2+ signal) (74, 75). Conversely, knockout of the proapoptotic genes Bax and Bak results in a drastic reduction in ER Ca2+, as well as resistance to apoptosis (76). Finally, mitochondrial Ca2+ uptake is a prerequisite for glutamate neurotoxicity, as it is the determinant factor in N-methyl-D-aspartate receptor (NMDAR)-triggered neuronal cell death (77). For brevity, we refer to recent reviews (14, 78-80) for a detailed discussion of Ca2+ signaling in apoptosis, a topic that has recently received much attention.

On the cytosolic side, mitochondrial Ca2+ uptake exerts two different effects. In the first, the spatial clustering of mitochondria in a defined portion of the cell, as seen in some cell types, represents a physiological "fixed spatial buffer" that can limit the propagation of cytoplasmic Ca2+ waves. Such a mechanism was clearly shown to operate in pancreatic acinar cells by Petersen and co-workers. In these cells, the Ca2+ response to low-dose agonist stimulation is restricted to the apical pole (where it causes granule secretion) by the action of a mitochondrial "firewall" located between the apical and basolateral portions of the cell (24). Similarly, in neurons, resident mitochondria have been shown to buffer [Ca2+] increases in defined cellular regions, for instance, in the presynaptic motor neuron terminal (81). Here, not only do mitochondria buffer rapid increases in intracellular Ca2+, but the slow release of Ca2+ from mitochondria through the mitochondrial Na+/Ca2+ exchanger maintains a relatively high cytosolic Ca2+ for a period following intense stimulation, providing the Ca2+ platform on which the phenomenon of posttetanic potentiation rests (81, 82). Wang et al. showed that, in neurons, redistribution of mitochondria caused by cytoskeleton disruption enhanced the Ca2+ increase seen after the opening of NMDARs, which confirmed the importance of the spatial distribution of these organelles in their capacity to buffer Ca2+ during stimulation (83).

The second mechanism by which mitochondrial Ca2+ uptake affects global Ca2+ signaling relates to events occurring in the ER/mitochondria microdomain and will be discussed in greater detail. These events depend on the Ca2+ sensitivity of the Ca2+ channels of the ER (allowing, for different Ca2+ concentrations and channel isoforms, positive or negative modulation of the Ca2+ release process) and the capacity of mitochondria to remove Ca2+ from the microdomain at the mouth of the channel. Indeed, the presence of nearby mitochondria actively accumulating Ca2+ may control the amplitude and the kinetics of ER Ca2+ release. The first demonstration of such an effect was obtained by Lechleiter and co-workers in Xenopus oocytes, in which the mitochondrial energization state (and thus the capacity to accumulate Ca2+) influenced the spatio-temporal pattern of propagating Ca2+ waves induced by IP3 (23). This effect has been confirmed in mammals, but the situation appears distinctly different in various cell models. In HeLa cells, mitochondrial Ca2+ uptake and rerelease of Ca2+ appears to serve to refill the ER through a very local privileged pathway (54). Similarly to HeLa cells, in permeabilized hepatocytes the release of ER [Ca2+] evoked by submaximal IP3 was enhanced when mitochondrial Ca2+ uptake was blocked (84). Moreover, ER in subcellular regions that were relatively deficient in mitochondria showed greater sensitivity to IP3 than regions of the cell with a high density of mitochondria. These data were interpreted as evidence that Ca2+ uptake by mitochondria suppresses the local positive feedback effects of Ca2+ on the IP3 receptor, giving rise to subcellular heterogeneity in IP3 sensitivity and IP3 receptor excitability. A similar picture emerged in astrocytes, in which inhibition of mitochondrial Ca2+ uptake almost doubled the rate of propagation of the calcium wave across the cell (85), which suggests that mitochondria, by removing local Ca2+ from the vicinity of the IP3 receptor, prevented the Ca2+-induced sensitization of the receptor required to maintain the excitability of the cytosol. A totally different picture, however, was observed in a hamster kidney cell line in which inhibition of mitochondrial Ca2+ uptake resulted in reduction of ER Ca2+ release (86). In the latter case, the interpretation is that mitochondria, by buffering Ca2+ at the mouth of the Ca2+ channel, prevent the well-known inhibition of the IP3 channel by high Ca2+.

It could be speculated that a positive or negative effect of mitochondria on local Ca2+ buffering might be a function of the IP3 receptor isoform expressed in a particular cell type. Under conditions where inhibition of the IP3 receptor by local Ca2+ was the predominant effect, mitochondrial uptake would favor Ca2+ release; however, under conditions where local Ca2+ had a positive effect, mitochondrial uptake would inhibit Ca2+ release. In addition, such an effect would depend on the efficiency of mitochondrial Ca2+ uptake. This involves many factors, including the thermodynamic driving force, the molecular repertoire of channels and other structural components of the ER/mitochondria contacts, and the three-dimensional structure of the organelle. With regard to the last named, recent work has highlighted the continuity of the mitochondrial network, showing that it is controlled by the activity of proteins causing mitochondrial fusion, such as mitofusin, or fission, such as dynamin-related protein 1 (87). This indicates that various modulatory mechanisms exist, many of which still await molecular clarification.

Mitochondrial Ca2+ uptake plays a key role in the activity of neighboring plasma membrane Ca2+ channels. The groups of Lewis and Parekh have shown that, in T lymphocytes, Ca2+ uptake by energized mitochondria relieves Ca2+-dependent inhibition of CRAC (Ca2+ release-activated Ca2+) channels, in other words, the influx pathway triggered by the depletion of Ca2+ from intracellular stores (88, 89) . This notion may explain in part the high Ca2+ buffering (mimicking mitochondrial activity) required to observe ICRAC (the Ca2+ release-activated Ca2+ current) in many experimental conditions. If one considers that mitochondrial Ca2+ uptake also modulates the kinetics and amplitude of ER Ca2+ release, it becomes apparent that mitochondrial Ca2+ handling in T lymphocytes plays a fundamental physiological role, ultimately facilitating Ca2+-dependent long-term gene activation triggered by antigen binding [see (90) for extended coverage of this topic].

In chromaffin cells, Montero and co-workers showed a close functional interaction between voltage-gated channels of the plasma membrane, ryanodine receptors, and mitochondria in determining the amplitude of the [Ca2+] rise in a specialized secretory microdomain, and thus the net efficiency of agonist- or depolarization-induced secretion (58). A role similar to that proposed for the modulation of capacitative Ca2+ influx through plasma membrane channels in T lymphocytes, has been proposed by Nicholls and co-workers for the modulation of glutamate- and voltage-operated Ca2+ channels by mitochondrial Ca2+ uptake in cerebellar granule cells (91), in which mitochondrial control of the subplasmalemmal Ca2+ concentration apparently serves to regulate Ca2+ influx through NMDA receptors as these are inhibited by a rise in local [Ca2+].

Open Questions

ER mitochondria, stochastic versus specific localization.

As discussed above, overwhelming evidence confirms the idea that the measured high rate of Ca2+ accumulation by mitochondria in situ largely depends on the proximity of mitochondria to the channels through which Ca2+ penetrates into the cytosol. A key, but still unanswered, question is whether this proximity is random and occurs only because some of these very abundant organelles, which are present throughout the cytosol, happen to be close to these channels, or whether there is a specific mechanism to ensure that mitochondria are located close to sites of Ca2+ influx and release. In highly structured cells, such as cardiac and skeletal muscle, the disposition of mitochondria with respect to the SR is highly ordered, and the organelles have very little mobility. The localization of mitochondria in pancreatic acinar cells, where a "firewall" of organelles appear to be strategically located immediately adjacent to ER Ca2+ release sites, is less well organized, but still remarkably reproducible, as mentioned above (24). In many other cells that are characterized by a high and efficient rate of mitochondrial Ca2+ accumulation in response to different agents, the distribution of mitochondria with respect to the ER does not display obvious morphological features.

Hints that the organelles that are exposed to high Ca2+ microdomains remain stably associated with Ca2+ channels came from the research of Montero et al. in bovine chromaffin cells (58), and the problem has been specifically addressed recently by Filippin et al. (53). Though indirect, these data strongly support the notion that despite the continuous reshaping of mitochondrial morphology with time, the organelles that consistently accumulate the most Ca2+ are in stable associations with the Ca2+ release sites. Such stable association suggests that yet-unknown molecular entities keep the two organelles bound to each other at selected regions.

Much less is known about the relations between plasma membrane Ca2+ channels and mitochondria. As mentioned above, functional evidence has pointed to the importance of mitochondrial Ca2+ uptake for the activity of store-activated Ca2+ currents (88-90). In addition, in HeLa cells, Filippin et al. showed that the mitochondria that respond best to Ca2+ mobilization from the ER are distinct from those most sensitive to capacitative Ca2+ influx through the plasma membrane and, again, this subpopulation of mitochondria apparently remains stably associated with this pathway (53).

To what concentrations can Ca2+ in the mitochondrial matrix rise?

Indeed--how is mitochondrial Ca2+ buffered? This is a matter of great debate, because of the widely varying results obtained with different probes and calibration procedures used by different groups. The two extremes are represented by the values obtained by measuring mitochondrial [Ca2+] with the fluorescent Ca2+ indicator Rhod-2 and its derivatives and the bioluminescent protein aequorin. With Rhod-2, the reported peak values of mitochondrial Ca2+ vary between 1 and 3 μM (55, 59, 92). Similar values were also calculated with a low-affinity Rhod-2 variant, Rhod-5N (93). In contrast, with estimates obtained using targeted recombinant aequorin, the reported peak values range between 5 and 500 μM, depending on the cell type and the stimulus used (20, 58, 62). Because of their intrinsic characteristics, calibration of Rhod-2 is dominated by the average lower values, whereas aequorin calibration is biased toward the highest responding organelle population. The other probes available are those based on engineered GFPs: cameleons, camgaroos, and pericams. Using cameleons with different Ca2+ affinities targeted to the mitochondria, Arnaudeau et al. concluded that in HeLa cells a subpopulation of around 30% of the mitochondria can reach Ca2+ values higher than 100 μM (54). Using pericams, Filippin et al. concluded that the average value for a peak Ca2+ is about 10 μM with subpopulations of the organelles reaching values above 50 μM (53).

Clearly, it is difficult to draw any conclusion from such divergent values. Our biased opinion is that the highest values are most likely to be closest to reality, at least in the subpopulations of most responsive organelles. However, one could also argue that this would represent a huge Ca2+ flux, which would carry a large current through the uniporter and so would be expected to cause a massive mitochondrial depolarization. In most cases, measurements of mitochondrial membrane potential show only small changes in response to physiological Ca2+ stimuli (94), whereas very large changes in potential are seen under more pathological conditions (95). The relatively slow responses of the indicators of membrane potential, the heterogeneity of mitochondrial Ca2+ uptake, and the rapid increase in mitochondrial respiration in response to depolarization that will rapidly restore the potential may provide some explanation for this apparent contradiction.

The ratio between free and bound Ca2+ within mitochondria is a matter of similar debate. Under resting conditions, both the total and the free Ca2+ content within the mitochondrial matrix seem to be similar to those in the cytosol, with a total/free ratio of around 100. Quantitative data on the increase of total Ca2+ after stimulation are relatively scarce and, most important, reveal a significant heterogeneity among different mitochondria. Thus, for example, in hippocampal neurons stimulated by triggering afferent fibers, the maximum increase in total Ca2+ can be over 100 times the basal value, with a mean increase of about 10-fold [for 1 s of tetanic stimulation (51, 52)]. To our knowledge, direct comparison between total Ca2+ (by EM techniques) and free Ca2+ (by Ca2+ probes) has not yet been carried out in the same preparation.

It must be stressed that two types of Ca2+ buffering mechanism probably coexist in mitochondria, one based on classical Ca2+ binding sites to proteins or phospholipids and one depending on Ca2+ phosphate precipitation. In fact, given the alkaline pH of the matrix (around 8), it is likely that Ca2+ phosphate precipitates when the [Ca2+]m exceeds a few tens of micromolar concentration. Indeed, electron-dense granules containing both phosphate and Ca2+ in the classical 3/2 ratio of hydroxyapatite are not only readily apparent in heavily damaged cells, but, most important, also in quickly frozen samples of neurons after brief periods of stimulation (51, 52). The observation that Ca2+ phosphate precipitates can be formed under physiologically relevant conditions, at least in neurons, further supports the notion that, in some mitochondria, the free Ca2+ must indeed exceed tens of micromolar concentration. It must be stressed that in addition to the possibility of buffering Ca2+ by forming precipitates of Ca2+ phosphate, several intramitochondrial proteins are known to specifically bind the cation, but all of them with relatively high affinity and low capacity.


Over the last 10 years or so, a picture has emerged that is startlingly different from anything we might previously have expected, revealing details of a remarkably evolved intimate interaction between mitochondria and Ca2+ signaling mechanisms. From a relatively crude general understanding that physiological Ca2+ signals might indeed have an impact on mitochondrial function and that mitochondria may in turn shape Ca2+ signals, we now are beginning to see that this interaction is tightly and subtly organized and coordinated on a subcellular basis, so that spatial and temporal details of the Ca2+ signal may be defined by the spatial organization and possibly specialization of mitochondrial populations within cells. A schematic view of the various processes described above is presented in Fig. 3. These observations are just in their infancy, and, as our technology becomes more sophisticated, further details of cellular specialization are bound to emerge. These observations also spur us on to new efforts to understand the molecular mechanisms that define the organization of mitochondria and their relation to Ca2+ stores and to identify the extent to which mitochondrial function may be specialized in relation to the specific needs of different cells. One wonders what the next 10 years will bring to this emerging field.

Fig. 3.

Scheme of the interrelations between cytoplasmic Ca2+ signals and the mitochondria. cADPr, cyclic ADP ribose receptor; CICR, calcium-induced calcium release channel; DAG, diacylglycerol; DHPR, dihydropyridine receptor; GPCR, G protein-coupled receptor; IP3, inositol triphosphate; PLC-β, phospholipase C-β; PM Ca2+ channel, plasma membrane calcium channel; PMCA, plasma membrane channel pump; RyR, ryanodine receptor; SERCA, sarcoplasmic reticulum.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
View Abstract

Stay Connected to Science Signaling

Navigate This Article