BAX and BAK Regulation of Endoplasmic Reticulum Ca2+: A Control Point for Apoptosis

See allHide authors and affiliations

Science  04 Apr 2003:
Vol. 300, Issue 5616, pp. 135-139
DOI: 10.1126/science.1081208


BAX and BAK are “multidomain” proapoptotic proteins that initiate mitochondrial dysfunction but also localize to the endoplasmic reticulum (ER). Mouse embryonic fibroblasts deficient for BAX and BAK (DKO cells) were found to have a reduced resting concentration of calcium in the ER ([Ca2+]er) that results in decreased uptake of Ca2+ by mitochondria after Ca2+ release from the ER. Expression of SERCA (sarcoplasmic-endoplasmic reticulum Ca2+ adenosine triphosphatase) corrected [Ca2+]er and mitochondrial Ca2+ uptake in DKO cells, restoring apoptotic death in response to agents that release Ca2+ from intracellular stores (such as arachidonic acid, C2-ceramide, and oxidative stress). In contrast, targeting of BAX to mitochondria selectively restored apoptosis to “BH3-only” signals. A third set of stimuli, including many intrinsic signals, required both ER-released Ca2+ and the presence of mitochondrial BAX or BAK to fully restore apoptosis. Thus, BAX and BAK operate in both the ER and mitochondria as an essential gateway for selected apoptotic signals.

Intracellular organelles are key participants in apoptosis (1). Mitochondria are the best documented of these, but the ER—where members of the BCL-2 family of proteins also localize—has been implicated (2). The proapoptotic BCL-2 family members can be subdivided into “multidomain” and “BH3-only” classes (3). The multidomain proapoptotic members BAX and BAK are necessary for apoptosis in response to a diverse array of intrinsic death signals and extrinsic death receptor signals in type II cells, which require a mitochondrial amplification loop (4). BH3-only proteins reside upstream in the pathway and either directly or indirectly activate BAX and BAK, inducing their intramembranous homo- oligomerization, which in turn results in the permeabilization of the outer mitochondrial membrane (5, 6). Released intermembrane-space proteins include cytochrome c, which complexes with Apaf-1 and caspase-9 to form a postmitochondrial apoptosome that amplifies effector caspase activation (7).

The mitochondria and ER are interconnected both physically and physiologically, affecting mitochondrial metabolism and complex cellular processes (8, 9). Mitochondria, the main source of cellular adenosine triphosphate, also modulate and synchronize Ca2+ signaling. Stimuli that generate inositol 1,4,5-trisphosphate (IP3) cause release of Ca2+from the ER, which is rapidly taken up by closely juxtaposed mitochondria (10). Ca2+ has long been recognized as a participant in apoptotic pathways (11). Ca2+ is also a prominent modulator of mitochondrial permeability transition (PT) controlled by the PT pore (PTP) (12). The PT has been implicated as a mechanism of both apoptotic and necrotic cell death after selected stimuli (2,13). Overexpression of BCL-2 protects cells from death induced by thapsigargin, an irreversible inhibitor of the SERCA pump responsible for uptake of Ca2+ from the cytosol into the ER lumen (14). Cells overexpressing BCL-2 display reduced ER Ca2+ concentration and decreased capacitative Ca2+ entry (15, 16). Here, we pursued a loss-of-function approach to assess whether BAX and BAK influence Ca2+ dynamics.

Thapsigargin causes the passive release of Ca2+ from ER stores and an increase in cytosolic Ca2+([Ca2+]i) that we measured with the dye Fura-2. The increase in [Ca2+]i in DKO cells [mouse embryo fibroblasts (MEFs) deficient for BAX and BAK] was smaller than that in wild-type cells (Fig. 1A). Primary as well as SV40-immortalized DKO MEFs displayed identical defects. Consequently, the multiply repeated assays in this study were performed on cell lines after confirmation of the same effect in primary cells. Moreover, MEFs singly deficient for BAX or BAK displayed similar but intermediate amounts of Ca2+ release, suggesting a comparable influence of each gene. The transient increase in [Ca2+]ielicited by thapsigargin in cells exposed to no extracellular Ca2+ was also significantly reduced in DKO cells relative to wild-type cells (Fig. 1B). The extent of capacitative Ca2+ entry assessed by addition of extracellular Ca2+ was similar in wild-type and DKO cells (Fig. 1B). Histamine caused IP3-mediated release of Ca2+from ER pools and resulted in a significantly lower peak [Ca2+]i in DKO cells (Fig. 1C). Thus, the rise in [Ca2+]i elicited by discharge of intracellular Ca2+ stores is diminished in cells that lack BAX and BAK.

Figure 1

Reduced intracellular Ca2+([Ca2+]i) responses in cells deficient in BAX and BAK. (A) Representative traces of [Ca2+]i in wild type (wt) and DKO MEFs treated with thapsigargin (Tg, 200 nM) in the presence of extracellular Ca2+ as measured by Fura-2. Average peak [Ca2+]i (from 9 independent experiments) was wt, 741 ± 86 nM; DKO, 294 ± 75 nM (P < 0.01; Student's t test). (B) [Ca2+]i in wt and DKO MEFs incubated in Ca2+-free medium. Where indicated, Tg (200 nM) and Ca2+ (800 μM free extracellular Ca2+ final concentration) were added. Average peak [Ca2+]i was wt, 415 ± 60 nM; DKO, 135 ± 10 nM (n = 9 experiments; P< 0.01) and capacitative [Ca2+]i was wt, 272 ± 69; DKO, 216 ± 42 nM. (C) [Ca2+]i in wt or DKO MEFs treated with histamine (Hist, 100 μM) in the presence of extracellular Ca2+. Average peak [Ca2+]i was wt, 482 ± 32 nM; DKO, 174 ± 14 nM (n = 5 experiments; P < 0.01). (D to F) [Ca2+]i after exposure to 40 μM C2-ceramide (D), 60 μM ArA (E), or 1 mM H2O2 (F) in wt and DKO MEFs.

The lipid mediators C2-ceramide and arachidonic acid (ArA) are proposed to initiate apoptotic death by a Ca2+-controlled process that induces the mitochondrial PT and can be inhibited by cyclosporin A (CsA) (17–19). These second messengers participate in death pathways initiated by cell surface receptors, DNA damage, and chemotherapy. C2-ceramide (40 μM) (Fig. 1D) and ArA (60 μM) (Fig. 1E) caused release of Ca2+ in wild-type cells, which was markedly blunted in DKO cells. Ca2+ release was also observed in the absence of extracellular Ca2+, which suggests that these stimuli mobilized intracellular stores. We used H2O2 as a representative oxidant; at 1 mM, H2O2 induces cell death that has apoptotic morphology, requires BAX or BAK, and can be inhibited by overexpressed BCL-2 (20). Again, the DKO cells proved defective in mobilization of Ca2+ (Fig. 1F).

Immunolocalization studies have suggested that endogenous BAX and BAK localize to the ER as well as to mitochondria (21, 22). We used a combination of immunoelectron microscopy, confocal imaging, and subcellular fractionation and found that ∼10% of BAX and ∼15% of BAK reside at the ER in MEFs (fig. S1). ER-targeted aequorin (erAEQ), a calibrated Ca2+ photoprotein reporter with highly selective subcellular localization, was used to measure the concentration of [Ca2+]er (15). The [Ca2+]er was first depleted to <20 μM and then refilled by readdition of Ca2+ to the perfusate (15). The steady-state [Ca2+]erreached by DKO cells was significantly lower than that in wild-type cells (Figs. 2A and3A). Ratiometric imaging with Mag-Fura-2 confirmed the reduced steady-state [Ca2+]erin DKO cells (fig. S2). Furthermore, histamine stimulation of wild-type cells caused a rapid decrease in [Ca2+]erfollowed by a slower, prolonged loss. In contrast, the already reduced [Ca2+]er of DKO cells proved much less responsive (Fig. 2B), consistent with the significantly lower peak [Ca2+]i elicited by histamine (Fig. 1C).

Figure 2

Reduced [Ca2+]er in DKO cells, resulting in reduced Ca2+ uptake by DKO mitochondria. (A and B) Representative recording of [Ca2+]er measured with erAEQ. Cells in which Ca2+ was depleted from ER were perfused as indicated with high-Ca2+ buffer (2 mM Ca2+) (A) and with 100 μM histamine (B). (C and D) Quantitative change in Rhod2 fluorescence (mean of six experiments) over mitochondrial regions of MEFs in response to 200 nM thapsigargin (C) or 100 μM histamine (D). (E) Representative [Ca2+]m in response to histamine (100 μM) in MEFs, measured with mtAEQ. (F) Mitochondrial Ca2+ uptake in wt, DKO, and DKO-SERCA MEFs permeabilized with 0.001% digitonin 30 s before perfusion with a buffered Ca2+ solution (final free Ca2+ = 14.5 μM). Ruthenium red (RuR, 100 μM), a blocker of the mitochondrial Ca2+ uniporter, was added with digitonin.

Figure 3

Distinct effects of correcting mitochondria or ER in DKO MEFs. (A) wt, DKO, and DKO MEFs transfected with wt BAX, mtBAX, or SERCA were cotransfected with erAEQ and [Ca2+]er was measured. Data are means ± SE of 14 different experiments (steady-state [Ca2+]er for wt, 527 ± 64 μM; DKO, 289 ± 41 μM). Asterisks indicate P < 0.001 versus the DKO sample (paired t test). (B) Cells were transfected with human CD19 alone (vector) or with vector expressing tBID; apoptosis (mean ± SE of three experiments) of CD19-positive cells was determined after 18 hours. (C) Cells noted above and DKO cells corrected with both mtBAX and SERCA were treated with 1 μM staurosporine (STS), 1 μM etoposide, or brefeldin A (BFA, 1 μg/ml) for 48 hours, and viability (mean ± SE of three experiments) was determined. (D) Viability (mean ± SE of five experiments) of wt, DKO, DKO-mtBAX, and DKO-SERCA MEFs after treatment with 40 μM C2-ceramide, 60 μM ArA, or 1 mM H2O2.

The uptake of Ca2+ by mitochondria modulates [Ca2+]i (23). Changes in mitochondrial Ca2+ concentration ([Ca2+]m) were assessed in single cells by real-time fluorescence microscopy using the fluorescent Ca2+ indicator Rhod2 (24) (movie S1). Thapsigargin-induced release of ER Ca2+ stores was followed by a transient increase in [Ca2+]m in wild-type cells that was substantially blunted in DKO cells (Fig. 2C). The peak [Ca2+]m was also blunted in DKO cells treated with histamine (Fig. 2D), consistent with the impaired release of ER Ca2+ by DKO cells in response to IP3. Measurement of [Ca2+]m with targeted aequorin (mtAEQ) confirmed the lack of a substantial increase in [Ca2+]m in DKO cells (Fig. 2E). To determine whether these differences reflect defects intrinsic to mitochondrial Ca2+ buffering, we permeabilized and perfused DKO and wild-type cells with buffers of fixed [Ca2+] and found that the kinetics and extent of increased Rhod2 fluorescence were superimposable and completely sensitive to ruthenium red, an inhibitor of the mitochondrial Ca2+ uniporter (Fig. 2F). Thus, the diminished levels of [Ca2+]m in DKO cells are not caused by an intrinsic defect in mitochondrial Ca2+handling, but rather result from the lower [Ca2+]i to which the mitochondria are exposed.

Amounts of the major Ca2+-regulatory ER proteins—including IP3 receptors 1 and 3 and their regulatory phosphatase calcineurin, SERCA2 and SERCA3 pumps, calreticulin, and GRP78 and GRP94—were unchanged in DKO cells (fig. S3A). Reexpression of BAX increased [Ca2+]erto nearly that in wild-type cells (Fig. 3A), and DKO-BAX cells exhibited a corresponding increase in apoptosis. BAX selectively targeted to mitochondria (mtBAX) had no effect on [Ca2+]er (Fig. 3A). Similar attempts to exclusively target BAX or BAK to the ER were all lethal to cells. Overexpression of SERCA targeted to the ER of DKO cells (fig. S3B) significantly increased [Ca2+]er (Fig. 3A) and restored the mitochondrial uptake of Ca2+ in response to histamine (Fig. 2, D and E). The Mag-Fura-2 indicator confirmed the correction of [Ca2+]er in DKO-SERCA cells (fig. S2). Transient as well as stable overexpression of SERCA restored [Ca2+]er, favoring a direct correction of the ER Ca2+ defect in DKO cells rather than an adaptive response during selection.

DKO cells selectively reconstituted with SERCA or mtBAX enabled us to test a range of apoptotic stimuli with respect to whether they are principally controlled by [Ca2+]eror by BAX or BAK at mitochondria. mtBAX restored apoptosis of DKO cells to essentially wild-type levels in response to the BH3-only protein tBID, whereas DKO-SERCA cells remained resistant (Fig. 3B). DKO cells were completely resistant to C2-ceramide or ArA even in combination with histamine (Fig. 3D) (fig. S4). However, expression of SERCA restored death of DKO cells in response to C2-ceramide or ArA, whereas expression of mtBAX did not (Fig. 3D). H2O2 (1 mM) is an oxidant that stimulates discharge of intracellular Ca2+ stores (Fig. 1F) and induces death of wild-type but not DKO cells (Fig. 3D). Death after treatment with H2O2 was restored in DKO-SERCA but not DKO-mtBAX cells (Fig. 3D).

When cells were exposed to a pulse of 2 μM ionomycin in the presence of extracellular Ca2+, the ionophore caused an increase in [Ca2+]i, primarily due to Ca2+ influx. Mitochondrial Rhod2 fluorescence changes were equivalent in DKO and wild-type cells in response to ionomycin (fig. S5). Under these conditions, ionomycin did not cause mitochondrial dysfunction as assessed by transmembrane potential, Δψm. A pulse of ionomycin restored killing of DKO cells by C2-ceramide, ArA, or H2O2, whereas histamine, which fails to increase [Ca2+]m, did not restore death (fig. S4). We measured Δψm by real-time imaging of tetramethylrhodamine methyl ester (TMRM) fluorescence intensity. Histamine-triggered, IP3-mediated Ca2+ mobilization augmented the capacity of H2O2, ArA, or C2-ceramide to decrease Δψm in wild-type cells (Fig. 4A) (fig. S6 and movies S2 to S4). In contrast, histamine-treated DKO cells maintained Δψm(Fig. 4A). The correction of ER Ca2+ stores and mitochondrial Ca2+ uptake in DKO-SERCA cells restored the response to H2O2, including the loss of Δψm and its acceleration by histamine (Fig. 4A) (movies S2 to S4). CsA, which inhibits mitochondrial PT, prevented mitochondrial depolarization (Fig. 4A) (fig. S6) and prevented cell death (fig. S4). Real-time imaging of calcein release from the mitochondria confirmed the presence of a Ca2+-dependent mitochondrial PT.

Figure 4

Correction of Ca2+ restores apoptotic death. (A) Left panel: quantitation of mitochondrial TMRM fluorescence in wt MEFs (mean ± SE of four experiments) pulsed with 100 μM histamine (red circles) where indicated and then treated with 1 mM H2O2. Cells were also pretreated with 0.5 μM CsA where indicated before H2O2 (open squares). Arrow indicates the addition of 5 μM CCCP (carbonyl cyanidem-chlorophenyl-hydrazone). Right panel: quantitation of mitochondrial TMRM fluorescence in DKO (squares and circles) or DKO-SERCA (triangles) MEFs. Cells were briefly treated with 100 μM histamine where indicated and then treated with 1 mM H2O2. (B) Subcellular cytochrome c distribution (left panels) and caspase activation (right panels) in wt (left) and DKO (right) cells. Cells were treated with 2 μM ionomycin or 60 μM ArA; where indicated, cells were first pulsed for 5 min with 100 μM histamine or 2 μM ionomycin. After the indicated time, cells were fixed and stained for cytochrome c (red) and Mn2+–superoxide dismutase (green) (left panels) or for activation of effector caspases using fmk-VAD–fluorescein isothiocyanate (green, right panels). Scale bars, 20 μm.

Wild-type MEFs in response to ArA release cytochrome c and subsequently activate caspases, hallmarks of apoptosis that correlate with the time course of PT and loss of Δψm (Fig. 4B). A pulse of histamine enhanced the release of cytochrome c and caspase activation in wild-type cells in response to ArA, but DKO cells showed no such evidence of apoptosis (Fig. 4B). However, a pulse of ionomycin restored apoptosis to ArA in DKO cells (Fig. 4B). DKO cells reconstituted with mtBAX underwent death in response to classic apoptotic stimuli, including staurosporine and etoposide (Fig. 3C), that act through mitochondria. DKO-SERCA cells displayed limited susceptibility to staurosporine and etoposide, but doubly corrected DKO-mtBAX-SERCA cells displayed the most complete killing (Fig. 3C). Sensitivity to agents that mimic ER stress, including brefeldin A, was markedly restored in DKO cells reconstituted with mtBAX, but only DKO-mtBAX-SERCA cells exhibited death equivalent to that of wild-type cells (Fig. 3C). DKO and wild-type thymocytes were reconstituted to generate chimeric recombination-activating gene–1 (Rag-1)–deficient mice, and testing for T cell receptor (TCR)–induced apoptosis (25) confirmed the Ca2+ dependence of a physiologic death signal in another cell type (26).

We have shown that the apoptotic gateway proteins BAX and BAK are required to maintain homeostatic concentrations of Ca2+ in the ER, and that this is a critical component of cell death regulation by these proteins. Consequently, DKO cells have less intracellular Ca2+ to release upon stimulation by IP3, Ca2+-mobilizing death signals, or inhibition of SERCA. Although DKO mitochondria have no intrinsic Ca2+ handling defect, decreased release of ER Ca2+ accounts for diminished mitochondrial uptake of Ca2+. This proved a critical control point in the apoptotic fate of cells responding to Ca2+-dependent stimuli, including oxidative stress.

Transient overexpression of BAX results in the release of ER Ca2+, with a subsequent increase in mitochondrial Ca2+ and enhanced cytochrome c release (22). Overexpression of antiapoptotic BCL-2 also reduces steady-state [Ca2+]er and mitochondrial Ca2+uptake (15, 16). In total, this is consistent with a rheostat model in which the ratio of the amount of proapoptotic and antiapoptotic BCL2 members dictates susceptibility to death. However, overexpression of BCL-2 or BCL-XL results in decreased capacitative Ca2+ entry (15, 16), reduction in amounts of calreticulin and SERCA2 (27), and altered IP3 receptor levels (28), none of which were noted in DKO cells. These differences argue for some distinct actions of proapoptotic versus antiapoptotic BCL-2 members in controlling Ca2+ dynamics. Calpains (29) and caspase-12 (30) participate in certain Ca2+-dependent deaths but appear to require BAX and BAK.

Distinct categories of death stimuli require BAX and BAK at either mitochondria or ER control points (fig. S7). BAX and BAK at mitochondria are requisite for “BH3-only” proteins, whereas regulation of ER Ca2+ by BAX and BAK proved an obligate control point for lipid second messengers (17–19) and oxidative stimuli (31). A third category of stimuli is influenced by control points at both mitochondria and ER and includes intrinsic signals initiated by staurosporine, etoposide, and brefeldin-A. The coordinate roles for BAX and BAK at the ER and mitochondria illustrate the importance of Ca2+ dynamics between these organelles in apoptosis.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Movies S1 to S4

  • These authors contributed equally to this work.


View Abstract

Navigate This Article