Close Contacts with the Endoplasmic Reticulum as Determinants of Mitochondrial Ca2+ Responses

See allHide authors and affiliations

Science  12 Jun 1998:
Vol. 280, Issue 5370, pp. 1763-1766
DOI: 10.1126/science.280.5370.1763


The spatial relation between mitochondria and endoplasmic reticulum (ER) in living HeLa cells was analyzed at high resolution in three dimensions with two differently colored, specifically targeted green fluorescent proteins. Numerous close contacts were observed between these organelles, and mitochondria in situ formed a largely interconnected, dynamic network. A Ca2+-sensitive photoprotein targeted to the outer face of the inner mitochondrial membrane showed that, upon opening of the inositol 1,4,5-triphosphate (IP3)–gated channels of the ER, the mitochondrial surface was exposed to a higher concentration of Ca2+ than was the bulk cytosol. These results emphasize the importance of cell architecture and the distribution of organelles in regulation of Ca2+ signaling.

Upon physiological stimulation with IP3-generating agonists, mitochondria undergo an increase in the concentration of Ca2+ in the matrix ([Ca2+]m) (1), well in the range of the Ca2+ sensitivity of the matrix dehydrogenases (2). This process, besides playing a direct role in the control of organelle function, may contribute to the modulation of the cytosolic Ca2+ concentration ([Ca2+]c), by buffering [Ca2+]c (3) or influencing its spatiotemporal pattern (4). The accumulation of Ca2+ by mitochondria is rapid, despite the low affinity of their transport mechanisms (5). Because mitochondria might respond to microdomains of high [Ca2+] that were generated in their proximity by the opening of the IP3-gated channels (1), we conducted high-resolution imaging of mitochondria and of their relation with the intracellular Ca2+ store (the ER). We directly monitored the [Ca2+] sensed by the mitochondrial Ca2+uptake systems by using a targeted aequorin chimera.

The combined use of green fluorescent protein (GFP) chimeras with distinct spectral and targeting properties allows identification of two different subcellular structures in living cells (6). We expressed the S65T GFP mutant targeted to mitochondria [mtGFP(S65T)] (6) in HeLa cells (7) and used a high-speed imaging system that allows a three-dimensional (3D) fluorescence image of high resolution to be obtained from computationally deblurred optical sections (8). The 3D images, derived from image stacks taken at 30-s intervals with a 60× objective (pixel size 133 nm), revealed that mitochondria form a largely interconnected “tubular” network that undergoes continuous rearrangement (Fig.1A). Within 1 min of observation, both growth and retraction, as well as fusion to other portions of the network, were frequently observed (see arrow), indicating a high structural plasticity. In agreement with previous observations (9), the “mitochondrial network” was even more obvious when a portion of a mtGFP(S65T)-transfected cell was analyzed at higher resolution (Fig. 1B). The visual appearance of a connected network and the luminal continuity were confirmed by the rapid recovery of fluorescence after photobleaching of mtGFP in a portion of the network (Fig. 1C). Finally, to simultaneously visualize the mitochondria and the ER, we cotransfected in HeLa cells a mitochondrially targeted blue mutant of GFP, mtGFP(Y66H, Y145F) (6), and a chimera of GFP(S65T) targeted to the ER [erGFP(S65T)] (10) (Fig. 1D). Domains of close apposition were evident in Fig. 1D and in similar images. From these data, the surface of the mitochondrial network in close apposition to the ER was estimated to be ∼5 to 20% of total (11).

Figure 1

High-resolution 3D imaging of mitochondria and ER. (A) Time-lapse 3D imaging of mitochondrial structure in a HeLa cell transiently expressing mtGFP (each image was taken 30 s apart). Transfection, image acquisition (with a 60× objective), and processing were done as described (7, 8). (B) A 3D image of mitochondria, taken with a 100× objective; all other experimental conditions as in (A). (C) Recovery of mtGFP fluorescence after photobleaching; experimental conditions as in (A). The first and second image were taken immediately before and after photobleaching mtGFP fluorescence in a small area within the cell. The following three images were taken at 2-min intervals after and the final image 30 min after the photobleaching. (D) Combined 3D imaging of mitochondria and ER in a HeLa cell transiently expressing mtGFP(Y66H,Y145F) and erGFP(S65T). The two 3D images were processed as in (A) and superimposed. The mitochondrial and ER images are represented in red and green, respectively; the overlaps of the two images are white. On the bottom, a detail of the main image (80-nm pixel).

On the basis of the morphological data, we expected that the microdomains of high [Ca2+] generated by the opening of the IP3-gated channels might be sensed by only a small portion of the mitochondrial surface. To verify this possibility, we constructed an aequorin chimera targeted to the mitochondrial intermembrane space (MIMS) (12). This chimera (designated mimsAEQ), when transiently expressed in HeLa cells, appeared properly sorted, as shown by the pattern of the immunocytochemical stain (13) (Fig.2A) and by results of dual-labeling experiments with the mitochondrial marker cytochrome c oxidase (14). The MIMS location of aequorin was confirmed by the characteristics of agonist-dependent [Ca2+] changes (15). Indeed, the peak [Ca2+] increase elicited by histamine, an IP3-generating agonist (Fig. 2B), was much smaller than that measured in the mitochondrial matrix with mtAEQ (Fig. 2C) and was unaffected by treatment with the uncoupler carbonylcyanide p -(trifluoromethoxy) phenylhydrazone (FCCP), which collapses the driving force for Ca2+ uptake in the matrix (16).

Figure 2

An aequorin chimera targeted to the mitochondrial intermembrane space (mimsAEQ) reveals domains of high [Ca2+] upon IP3-mediated Ca2+release from the ER. (A) Immunolocalization of mimsAEQ in transiently transfected HeLa cells. The effect of mitochondrial uncoupling on the histamine-dependent increases in (B) [Ca2+]mims and (C) [Ca2+]m. Both [Ca2+]mims and [Ca2+]m were monitored with mimsAEQ and mtAEQ, respectively. Red trace, control cells; blue trace, cells treated with 5 μM FCCP 1 min before addition of 100 μM histamine (Hist.). Traces [in (B) through (E), and in Fig. 3] are representative of at least five similar experiments. (D) Effect of repeated agonist stimulation on [Ca2+]c and [Ca2+]mims. Transfected HeLa cells (red, mimsAEQ; blue, cAEQ) were treated with 100 μM histamine (Hist.) and 100 μM ATP. (E) IP3-dependent [Ca2+]mims and [Ca2+]c changes in permeabilized cells. Transfected HeLa cells (red, mimsAEQ; blue, mGluR1/AEQ) were permeabilized with digitonin and perfused with IB, supplemented as specified (15). Where indicated (IP3), 5 μM IP3 was added.

The histamine-dependent [Ca2+]mims changes differed also from those of [Ca2+]c. In particular, the initial maximal [Ca2+]mims increase, which is mostly contributed by the release of Ca2+ from intracellular stores, exceeded that of [Ca2+]c (3.5 ± 0.2 compared with 2.5 ± 0.3 μM, n = 10), and then declined to similar concentrations (Fig. 2D). Because the outer mitochondrial membrane is freely permeable to ions, a possible explanation of this finding is that a small fraction of the photoprotein is transiently exposed to a local domain of saturating [Ca2+] and is completely discharged. Thus, although the increase in [Ca2+] in most of the MIMS is in fact similar to that of the cytosol, the maximal light emission of this aequorin fraction contributes to the total luminescence signal; hence, the calibrated [Ca2+]mims increase appears to be larger than that of [Ca2+]c. If this were the case, then, because of the irreversible photoprotein consumption in these domains, the difference in the apparent [Ca2+] of the two compartments would decrease during a subsequent agonist stimulation applied shortly after the first. Indeed, when the cells were exposed to another IP3-generating agonist, adenosine triphosphate (ATP), after the stimulation with histamine, the difference in the peak [Ca2+] increase of the cytoplasm and of the MIMS was nearly abolished (17). The discrepancy between the increases in [Ca2+]mims and [Ca2+]c (Fig. 2D) is not a calibration artifact due either to an intrinsic difference in the Ca2+affinity of the two chimeras or to local pH or pMg gradients. Indeed, using a membrane-bound cytosolic probe (mGluR1/AEQ) (18), we observed, in digitonin-permeabilized cells, that release of Ca2+ from the ER induced by the administration of IP3 caused a greater increase in [Ca2+] in the MIMS than in the bulk cytosol, whereas perfusion of a buffered Ca2+ solution increased the [Ca2+] of the two compartments to the same extent (Fig. 2E).

At contacts between the ER and mitochondria, microdomains of high [Ca2+] may be generated upon opening of the IP3-gated channels. These microdomains could allow the rapid uptake of a large amount of Ca2+ by mitochondria. The rapid diffusion of Ca2+ within the mitochondrial network (as revealed by the discharge of a major portion of mitochondrial matrix aequorin, mtAEQ) could allow the rapid tuning of mitochondrial metabolism to cell needs. On the cytosolic side, diffusion of Ca2+ would dissipate the microdomains, thus extending the Ca2+ signal to the bulk cytosol (and eliciting the cell response). The lower [Ca2+] would limit further accumulation into mitochondria, avoiding organelle overload, Ca2+ cycling, and collapse of the proton gradient.

On the basis of this model, we would predict that if a second release of Ca2+ from the ER is induced after the first, then because of the depletion of active aequorin in the mitochondrial regions closer to the ER, the apparent [Ca2+]m increase should be underestimated. However, if enough time elapses between two consecutive stimulations, unconsumed mtAEQ should diffuse intralumenally from other regions of the mitochondrial network, leading to a larger increase in light emission, and thus the calibrated [Ca2+]mincrease should recover its initial amplitude. We treated cytosolic aequorin (cytAEQ)– or mtAEQ-transfected cells with ATP first, and then 1.5 or 10 min later, with histamine (Fig.3). In the former case, the histamine-dependent [Ca2+]m increase was smaller than the increase caused by ATP and drastically less than that observed in cells in which the ATP stimulation was omitted (43 ± 3%). The amplitudes of the [Ca2+]m increases did not correlate with those of [Ca2+]c (for example, in the second stimulation with histamine, the [Ca2+]c increases were larger than those caused by ATP), but rather suggested that, during stimulation with a first agonist, mtAEQ was preferentially consumed at the “hotspots.” In fact, if the second histamine treatment was given after a 10-min delay, the increase in [Ca2+]m was larger, approaching the values measured when histamine was applied as first stimulus (84 ± 8%).

Figure 3

Time-dependent recovery of mtAEQ luminescence during a second agonist stimulation. The value of [Ca2+]m and, for comparison, the value of [Ca2+]c were monitored in cells transfected with mtAEQ or cAEQ. All conditions as in Fig.2. Where indicated, the cells were treated with 100 μM histamine (Hist.) or ATP.

The observation that mitochondria form in vivo a largely connected, continuous network has consequences for understanding physiological events, such as organelle biogenesis and mitochondrial energy conservation, and for clarifying pathophysiological events, such as the mechanisms that lead to defects in mtDNA. Close appositions between ER and mitochondria may represent the site where microdomains of high [Ca2+] are generated upon IP3-mediated Ca2+ release. Indeed, there is a good agreement between the area of the apposition sites and the area in which the increase in [Ca2+] saturated the binding of Ca2+ to aequorin (19). The microheterogeneity of the Ca2+ signal, and the spatial relation between ER and mitochondria, may thus be determinants of mitochondrial Ca2+ uptake, which influences organelle function (1) and may modulate the cytosolic Ca2+ signal (2, 3).

  • * To whom correspondence should be addressed. E-mail: rizzuto{at}

  • Deceased.


View Abstract

Navigate This Article