The endoplasmic reticulum gateway to apoptosis by Bcl-X_L modulation of the InsP₃R

Abstract

Members of the Bcl-2 protein family modulate outer mitochondrial membrane permeability to control apoptosis^1,2. However, these proteins also localize to the endoplasmic reticulum (ER), the functional significance of which is controversial^{3, 4}. Here we provide evidence that anti-apoptotic Bcl-2 proteins regulate the inositol 1,4,5-trisphosphate receptor (InsP₃R) ER Ca²⁺ release channel resulting in increased cellular apoptotic resistance and enhanced mitochondrial bioenergetics. Anti-apoptotic Bcl-X_L interacts with the carboxyl terminus of the InsP₃R and sensitizes single InsP₃R channels in ER membranes to low [InsP₃], enhancing Ca²⁺ and InsP₃-dependent regulation of channel activity in vitro and in vivo, reducing ER Ca²⁺ content and stimulating mitochondrial energetics. The pro-apoptotic proteins Bax and tBid antagonize this effect by blocking the biochemical interaction of Bcl-X_L with the InsP₃R. These data support a novel model in which Bcl-X_L is a direct effector of the InsP₃R, increasing its sensitivity to InsP₃ and enabling ER Ca²⁺ release to be more sensitively coupled to extracellular signals. As a consequence, cells are protected against apoptosis by a more sensitive and dynamic coupling of ER to mitochondria through Ca²⁺-dependent signal transduction that enhances cellular bioenergetics and preserves survival.

A central feature of molecular models of apoptosis is the control of outer mitochondrial membrane permeability by Bcl-2-related proteins. The pro-apoptotic Bcl-2-related proteins Bax and Bak are required to initiate cytochrome c release from mitochondria in response to diverse apoptotic stimuli^1,5. Anti-apoptotic properties of Bcl-2 and Bcl-X_L have been attributed to their ability to antagonize Bax/Bak by forming heterodimers that prevent their oligomerization and apoptosis initiation⁶. Pro- and anti-apoptotic Bcl-2 proteins also localize to the ER^3,7, and it is now recognized that the ER has an important role in regulating apoptosis^8,9. The ER is thought to contribute to apoptosis through its role as the principle Ca²⁺ storage organelle in cells^8–11. At physiological levels, Ca²⁺ released from the ER during cell activation is taken up by mitochondria to stimulate oxidative phosphorylation and enhance ATP production¹². However, sustained and complete release of ER Ca²⁺ can initiate Ca²⁺-dependent forms of apoptosis by activating calpain processing of caspase 12 (ref. ¹³) or by triggering the opening of the mitochondrial permeability transition pore¹⁴. It has been suggested that high levels of ER Ca²⁺ may sensitize cells to these pathways of apoptosis by providing a higher quantity of released Ca²⁺ (ref. ¹¹).

Recent evidence suggests that Bcl-2-related proteins can localize to the ER and the nuclear envelope³ to regulate permeabilities of these membranes as well as those of mitochondria⁷. However, the effects of this localization have not led to a consistent set of observations that link the functions of these proteins, ER Ca²⁺ dynamics and apoptosis control⁴. Genetic studies have suggested that pro-apoptotic Bax and Bak can promote Ca²⁺ release from the ER and initiate a caspase 12-associated form of apoptosis⁷, whereas anti-apoptotic Bcl-2 overexpression also promotes Ca²⁺ release by causing an unregulated ER Ca²⁺ leak¹⁵. Physiological studies are also conflicting regarding the effects of Bcl-2 proteins on Ca²⁺ leakage^4,11 and InsP₃R activity¹⁶. The nature of Bcl-2-associated ER regulation and the effect that this regulation has on apoptosis are therefore uncertain, prompting us to utilize distinct approaches to investigate the ability of the anti-apoptotic Bcl-2 family member Bcl-X_L to interact with the InsP₃R and regulate Ca²⁺ release, ER [Ca²⁺] ([Ca²⁺]_ER) and apoptosis.

In in vitro pull-down assays, full-length human Bcl-X_L interacted with all three mammalian InsP₃R isoforms (Fig. 1a). Immunoprecipitation of Bcl-X_L coprecipitated type 3 InsP₃R from COS-7 cell lysates (Fig. 1a), suggesting that the endogenous proteins interact in vivo. Bcl-2 also binds to InsP₃R (see Supplementary Information, Fig. S1), consistent with previous findings¹⁶. The interacting region in the InsP₃R was localized to the C-terminal region of the channel (Fig. 1c).

Figure 1.

Interaction of Bcl-X_L with InsP₃R. (a) Bcl-X_L binds to full-length types 1, 2 and 3 InsP₃R. Lysates from DT40-InsP₃R-KO cells stably expressing rat type 1 InsP₃R and from COS-7 cells that endogenously express type 2 and type 3 InsP₃R, were incubated with GST–Bcl-X_L, and bound InsP₃R was detected with isoform-specific antibodies (top three panels). Bottom panel: co-immunoprecipitation of endogenous Bcl-X_L and type 3 InsP₃R from COS-7 cells. (b) Domain structure of full-length InsP₃R and of its C-terminal region. (c) Bcl-X_L binds within the C terminus of InsP₃R. GST–Bcl-X_L failed to pull-down the V5-tagged 1–600 type 1 InsP₃R fragment expressed in COS-7 cells (top panel). Expression of 1–600 InsP₃R was verified by a western blot of cell lysates with V5-specific antibody (lane 3). Rat type 1 InsP₃R lacking the first 600 residues (Δ1–600 InsP₃R) expressed in COS-7 cells was successfully pulled-down along with endogenous InsP₃R-1 (middle panel). GST–Bcl-X_L effectively binds to the C-terminal 2512–2750 residues of type 1 InsP₃R (bottom panel). All western blots depicted are representative of three independent experiments.

The functional consequences of the interaction of Bcl-X_L with the InsP₃R were explored by recording single InsP₃R channels in native ER membranes, by patch-clamp electrophysiology of outer membranes of isolated nuclei¹⁷ (Fig. 2a). Robust InsP₃R channel activity (open probability, P_o ~0.7) was observed with pipette solutions facing the cytoplasmic aspect of the channel containing 10 μM InsP₃ and 1 μM Ca²⁺, conditions optimal for channel activation (Ionescu, L., Mak, D.-O.D., C.W. & K.F., unpublished observations) (Fig. 2b). Channels were observed in 80% of patches, with the mean number of active channels per patch (N_A) ~1.8 (Fig. 2c). When [InsP₃] was reduced to 10 nM, channel activity was substantially decreased (P_o ~0.02) (Fig. 2b), consistent with the [InsP₃] dependence of channel gating¹⁷, and N_A was only ~0.1 (Fig. 2c). Addition to the pipette solution of purified recombinant human Bcl-X_L (rBcl-X_L; 1 μM), with biological activity confirmed in cytochrome c release assays (see Supplementary Information, Fig. S1), did not affect the P_o or N_A of channels activated by 10 μM InsP₃ (data not shown). In contrast, rBcl-X_L markedly increased the number of channels (N_A ~1.0) and gating (P_o ~0.6) activated by 10 nM InsP₃, to levels comparable to those observed with saturating [InsP₃] (Fig. 2b, c). The product of P_o and N_A, a measure of total InsP₃R-mediated ion flux across the ER membrane, was increased by over two orders of magnitude in the presence of Bcl-X_L (Fig. 2c).

Figure 2.

Effects of Bcl-X_L on InsP₃R single-channel activity. (a) Isolated nuclear preparation from Sf9 cells, which endogenously express only type 1 InsP₃R. Patch pipette approaching an isolated nucleus with an intact cell visible directly above; scale bar, 15 μM. (b) InsP₃R channel activity. Typical InsP₃R single-channel current recordings in the presence of saturating (10 μM) or low (10 nM) InsP₃; in the absence or presence of recombinant Bcl-X_L (rBcl-X_L; 1 μM); or with 10 nM InsP₃ in cells transiently transfected with Bcl-X_L. Channel activity was not evoked by rBcl-X_L (1 μM) alone. Pipette [Ca²⁺] was 1 μM, optimal for channel activity¹⁷; arrow indicates zero current level. (c) Summary of the effects of Bcl-X_L on InsP₃R channel activity. In pipettes containing 10 nM InsP₃, the open probability (P_o) increased from 0.022 ± 0.001 (n = 2) to 0.61 ± 0.09 (n = 10) with addition of 1 μM rBcl-X_L (n = number of patches used in P_o determination). Similarly, when Bcl-X_L was overexpressed, P_o increased to 0.42 ± 0.05 (n = 11). Bcl-X_L also increased the number of activated channels (N_A). In 10 nM InsP₃, N_A was increased from 0.10 ± 0.07 (n = 20) under control conditions to 1.00 ± 0.16 (n = 54) and 1.18 ± 0.17 (n = 61) in the presence of recombinant or expressed Bcl-X_L, respectively. The total ER ion flux as indicated by the product N_AP_o was increased by Bcl-X_L (note log scale). Asterisks indicate P < 0.001, unpaired t-test. Both His- and Flag-tagged rBcl-X_L, generated using distinct purification protocols, were equally effective, whereas His-tagged NCS-1, which does not interact with InsP₃R (ref. ³²), had no effect either alone or in combination with 10 nM InsP₃ (data not shown). (d) Dependence of InsP₃R channel activity on [Ca²⁺] at the cytoplasmic face of the channel. Effect of [Ca²⁺]_i on P_o, N_A and N_AP_o in the presence of 10 μM InsP₃ (black inverted triangles), 10 nM InsP₃ (green circles), 10 nM InsP₃ + 1 μM rBcl-X_L (blue diamonds) and 10 nM InsP₃ + Bcl-X_L expression (red squares).

InsP₃R channel activity in nuclei isolated from cells transiently transfected with Bcl-X_L showed enhanced N_A and P_o in 10 nM InsP₃ in the absence of added recombinant protein (Fig. 2b, c), demonstrating an in vivo physical and functional membrane-delimited interaction between the InsP₃R and Bcl-X_L that was sufficiently strong to survive the nuclear isolation protocol. Bcl-X_L-enhanced channel activities required InsP₃ binding because they were not observed in the absence of added InsP₃ (Fig. 2b), and the competitive InsP₃ inhibitor heparin ablated the effects (data not shown). Gating of InsP₃-liganded InsP₃R channels is biphasically regulated by cytoplasmic [Ca²⁺] ([Ca²⁺]_i) (ref. ¹⁷ and Ionescu et al. unpublished observations). Enhanced channel recruitment and gating conferred by either recombinant or expressed Bcl-X_L was manifested between 100 nM and 1.8 μM Ca²⁺ (Fig. 2d). Thus, Bcl-X_L-activated InsP₃R channels retained biphasic [Ca²⁺]_i regulation. Taken together, these data indicate that Bcl-X_L binding to the C terminus of the InsP₃R allosterically increases the sensitivity of the channel to very low levels of InsP₃ that may exist in resting or minimally stimulated cells¹⁸ at physiological [Ca²⁺]. This suggests that the in vivo interaction of InsP₃R with Bcl-X_L does not induce an unregulated Ca²⁺ leak but rather increases the dynamic sensitivity of a highly regulated Ca²⁺ permeability of the ER membrane.

Purified recombinant Bax, a multi-domain family member, and tBid, a BH3-only member, each disrupted binding of InsP₃R to GST–Bcl-X_L (Fig. 3a) and abolished the functional effects of expressed Bcl-X_L (Fig. 3b, c). In contrast, neither protein affected gating in the absence of Bcl-X_L (see Supplementary Information, Fig. S2). It has been reported that Bcl-X_L and Bax can form ion channels in artificial membrane systems^19,20. However, novel conductances were never observed when either Bcl-X_L, Bax, or tBid was present in the patch pipette solution, or when Bcl-X_L was overexpressed, indicating that under the present experimental conditions these proteins do not form ion channels in native ER membranes.

Figure 3.

tBid and Bax antagonize the effects of Bcl-X_L on InsP₃R channel activity. (a) tBid and Bax inhibit binding of Bcl-X_L to InsP₃R. COS-7 cell lysates were incubated with GST–Bcl-X_L in the absence or presence of recombinant tBid (2–200 nM; upper panel) or recombinant Bax (200 nM; lower panel), and bound InsP₃R was detected with a type 3 antibody. Recombinant neuronal calcium sensor-1 (NCS-1) was used to control for total protein concentration in both experiments. Data are representative of three independent experiments. (b) tBid and Bax inhibit the electrophysiological effects of Bcl-X_L. Typical current traces showing the effects of 100 nM recombinant tBid or Bax in the presence of 10 nM InsP₃ in nuclei isolated from cells transiently transfected with Bcl-X_L. Pipette [Ca²⁺] was 1 μM. (c) Addition of 100 nM tBid or Bax decreased P_o from 0.28 ± 0.06 (n = 5) to 0.07 ± 0.03 (n = 3) and 0.09 ± 0.05 (n = 3), respectively. Similarly, N_A was reduced from 0.74 ± 0.14 (n = 38) to 0.26 ± 0.10 (n = 33) in the presence of tBid and to 0.09 ± 0.05 (n = 33) in the presence of Bax. The product N_AP_o was also reduced. Asterisks indicate P < 0.001, unpaired t-test.

The cell physiological implications of the Bcl-X_L–InsP₃R interaction were evaluated using chicken DT40 B-cell lines expressing either Bcl-X_L or vector alone (Fig. 4a). Stable expression of Bcl-X_L in wild-type DT40 cells (DT40-WT), which express all three InsP₃R isoforms²¹, reduced the Ca²⁺ content of the ER (Ca²⁺_ER) by 25% (Fig. 4b), consistent with some previous observations^4,11. In contrast, stable (Fig. 4b and see Supplementary Information, Fig. S3c) or transient (see Supplementary Information, Fig. S3a, b) Bcl-X_L expression did not affect the Ca²⁺_ER in a DT40 cell line that was genetically deficient in all InsP₃R isoforms (DT40-InsP₃R-KO)²¹. Thus, Bcl-X_L-associated reduction of Ca²⁺_ER depends critically on expression of the InsP₃R, consistent with the hypothesis that Bcl-X_L-mediated enhanced InsP₃R ligand sensitivity is the molecular basis of the Bcl-2-enhanced ER Ca²⁺ permeability observed previously^22–24.

Figure 4.

Interaction of Bcl-X_L with InsP₃R is essential for Bcl-X_L effects on ER Ca²⁺ regulation and inhibition of apoptosis. (a) The empty vectors pIRES2-DsRed2 or pBcl-X_L-IRES2-DsRed2 were stably expressed in DT40-WT and DT40-InsP₃R-KO cells. Expression levels of types 1 and 3 InsP₃R, Bcl-X_L and OxPhos complex IV (COXIV; subunit 1) were examined by western blot. Expression of the mitochondrial complex IV protein was unchanged in the Bcl-X_L-expressing clones. Depicted blots are representative of three independent experiments. (b) Effects of Bcl-X_L expression on the Ca²⁺ content of the ER (Ca²⁺_ER). Typical records depicting change in cytoplasmic [Ca²⁺] ([Ca²⁺]_i) in response to application of 1 μM thapsigargin (TG) in DT40-WT and DT40-InsP₃R-KO cells stably transfected with either Bcl-X_L or vector alone. Ca²⁺_ER was indirectly estimated by single-cell imaging of the [Ca²⁺]_i responses to acute inhibition by thapsigargin of ER Ca²⁺ uptake. Each trace represents mean ± s.e.m. of at least six cells within the image field. Bar graph summarizes the effects of thapsigargin; data represent mean ± s.e.m. for at least 30 cells in multiple trials. Asterisk indicates P < 0.05, ANOVA. Mean values of resting [Ca²⁺]_i ranged from 80 to 100 nM, with no significant differences among duplicate clones of the four cell lines (data not shown). (c) [Ca²⁺]_i transients in response to 5 μg ml⁻¹ anti-BCR antibody (anti-IgM) in DT40-WT cells stably transfected with either Bcl-X_L or vector alone. Summary data represent the peak amplitude (mean ± s.e.m.) for at least 30 cells in multiple trials. (d) Cell viability after treatment with 20 μg ml⁻¹ anti-BCR antibody (anti-IgM) (time 0) of DT40-WT (solid symbols) and DT40-InsP₃R-KO (open symbols) cells stably expressing Bcl-X_L (red) or vector alone (same clones as in b). Similar results were obtained using independent clones (see Supplementary Information, Fig. S3).

Next we examined whether lowered Ca²⁺_ER as a consequence of Bcl-X_L expression represents a fundamental control point in the regulation of apoptosis, because it reduces the amount of Ca²⁺ released through the InsP₃R by pro-apoptotic signals. DT40 pre-B cells undergo apoptosis following immunoglobulin receptor crosslinking in a process that recapitulates negative selection²⁵. Bcl-X_L blocks B-cell receptor (BCR)-mediated apoptosis²⁶. Therefore, we examined antigen-receptor-mediated apoptosis in DT40-WT and DT40-InsP₃R-KO cells stably expressing either Bcl-X_L or empty vector. An anti-BCR antibody (anti-IgM) elicited comparable [Ca²⁺]_i transients in Bcl-X_L-expressing and control wild-type cells (Fig. 4c), whereas no [Ca²⁺]_i transients were observed in the InsP₃R-KO cells²¹ (data not shown). DT40-InsP₃R-KO cells were resistant to apoptosis compared with DT40-WT cells during the initial 24 h (Fig. 4d), consistent with previous observations²¹. However, the resistance was transient, with apoptosis at later times being independent of InsP₃R expression. Bcl-X_L inhibited apoptosis in both wild-type and InsP₃R-KO cells, as expected, because Bcl-X_L has ER-independent (mitochondrial) mechanisms of action. We reasoned that if Bcl-X_L was anti-apoptotic because it lowered Ca²⁺_ER and therefore reduced the amount of Ca²⁺ that could be conveyed from the ER to mitochondria following InsP₃R activation^9,11, then InsP₃R-KO cells expressing Bcl-X_L should have maximal protection against apoptosis because they lack the mechanism (InsP₃R) to convey Ca²⁺ from the ER to mitochondria. Surprisingly, the anti-apoptotic action of Bcl-X_L was significantly greater in DT40-WT cells compared with InsP₃R-KO cells (Fig. 4d and see Supplementary Information, Fig. S3c). These data demonstrate that the presence of InsP₃R is required for Bcl-X_L to be fully efficacious as an anti-apoptotic mediator.

The ER is an important control point for apoptotic responses to diverse stimuli^8,9. Our results provide a molecular mechanism that links the ER, Ca²⁺ and Bcl-2 family proteins to apoptosis, involving the ubiquitous InsP₃R ER Ca²⁺ release channel as a target of a direct effector function of anti-apoptotic Bcl-X_L. This function of Bcl-X_L is antagonized by proapoptotic family members, which conforms to a proposed rheostat model in which the balance of pro- to anti-apoptotic Bcl-2 members regulates apoptosis¹⁰, but differs from the prevailing concept that anti-apoptotic Bcl-2 proteins prevent apoptosis solely by antagonizing pro-apoptotic Bcl-2 family members^2,6. Whereas this paradigm may account for the competing effects of pro- and anti-apoptotic members at mitochondria, the present results suggest that the converse is true at the ER. Here, the pro-apoptotic family member Bax can reverse the function of Bcl-X_L without affecting ER permeability on its own.

Previous studies that observed reduced Ca²⁺_ER in response to Bcl-2 expression or Bax/Bak deficiency concluded that the underlying mechanism is associated with an enhanced ER Ca²⁺ leak^15,22–24. The current results establish the molecular identity of the enhanced leak to be the InsP₃R. It was proposed that phosphorylation status of the type 1 InsP₃R linked Bcl-2 expression to Ca²⁺_ER (ref. ¹⁵). However, phosphorylation cannot account for the rapid effects of Bcl-X_L on InsP₃R gating observed here, as the experiments were performed with solutions that lacked Mg²⁺ — necessary for phosphorylation reactions — and treatment of nuclei with alkaline phosphatase did not affect the ability of Bcl-X_L to activate channel gating (C.W. & K.F., unpublished observations). The effect of Bcl-X_L expression in lowering Ca²⁺_ER has not been consistently observed ^4,11. The current observations demonstrate that this effect of Bcl-X_L is InsP₃R-dependent, with Bcl-X_L-stimulated InsP₃R channel activity remaining exquisitely regulated by its ligands InsP₃ and Ca²⁺. Thus, cell-type differences in basal concentrations of these ligands and InsP₃R expression can contribute to the magnitude of the effect of Bcl-2 expression on Ca²⁺_ER, which might reconcile discrepant published observations.

Previously it was believed that reducing InsP₃-mediated ER Ca²⁺ release would protect against Ca²⁺-induced mitochondrial permeability transition^11,14. This model may possibly account for the early transient protection observed in the control InsP₃R-KO cells (Fig. 4d), but it is insufficient to account for the InsP₃R-dependent anti-apoptotic effects of Bcl-X_L observed in DT40 cells. If the sole function of Bcl-X_L at the ER was to lower [Ca²⁺]_ER, as a mechanism to protect mitochondria from toxic effects of released Ca²⁺, then protection afforded by Bcl-X_L should be maximal in the InsP₃R-KO cells, where Bcl-X_L can exert its mitochondrial effects^1,2 and the ER is irrelevant because the absence of the release channels prevents delivery of Ca²⁺ to mitochondria. However, InsP₃R expression conferred greater Bcl-X_L-mediated protection. Thus, despite the correlation between reduced [Ca²⁺]_ER and apoptosis resistance observed here and previously¹¹, the InsP₃R provides an anti-apoptotic signal that cannot be accounted for by reduced mitochondrial Ca²⁺ delivery through its effect on [Ca²⁺]_ER.

What is the nature of the anti-apoptotic signal that is conferred by InsP₃R expression? We reasoned that Ca²⁺ released from the ER can enhance mitochondrial function by stimulating the citric-acid cycle and oxidative phosphorylation with consequent production of ATP and other important metabolic intermediates¹². We speculated that maintenance of cellular metabolic homeostasis by Ca²⁺ activation of mitochondrial metabolism would enable cells to better withstand apoptotic insults. In such a model, Bcl-X_L interaction with the InsP₃R generates [Ca²⁺]_i signals that enhance cellular bioenergetics that afford apoptosis resistance. In single DT40-WT cells perfused with serum-containing medium, infrequent, spontaneous, transient [Ca²⁺]_i spikes were observed in fewer than 4% of cells (Fig. 5a, b). In contrast, spontaneous [Ca²⁺]_i spikes were observed in over 25% of Bcl-X_L-expressing DT40-WT cells and these spikes were consistently of higher frequency (Fig. 5a, b). Similarly, [Ca²⁺]_i oscillations triggered in response to a low concentration of anti-IgM were observed in more Bcl-X_L-expressing wild-type cells than in control wild-type cells (see Supplementary Information, Fig. S4), reminiscent of observed Bcl-2 enhancement of ligand-induced [Ca²⁺]_i oscillations²³. These spikes represent transient release from ER through InsP₃R because they were absent in DT40-InsP₃R-KO cells (Fig. 5a). Thus, Bcl-X_L enhances low-level InsP₃R-mediated [Ca²⁺]_i signalling under resting cellular conditions and during low-level agonist stimulation.

Figure 5.

The Bcl-X_L–InsP₃R interaction modulates [Ca²⁺]_i signalling and mitochondrial NADH levels. (a) Spontaneous [Ca²⁺]_i oscillations in three representative DT40-WT cells stably expressing vector alone (left) or Bcl-X_L (middle), and DT40-InsP₃R-KO cells expressing Bcl-X_L (right).(b) The difference in frequency and number of oscillating cells (mean ± s.e.m.) between vector only and Bcl-X_L-expressing cells. (c) NAD(P)H fluorescence measurements from DT40-WT and DT40-InsP₃R-KO cells expressing Bcl-X_L or vector in response to 5 μg ml⁻¹ anti-BCR antibody (anti-IgM) and FCCP (2 μM). The average (± s. e.m.) resting NAD(P)H fluorescence and the change in NAD(P)H fluorescence in response to anti-IgM stimulation for four independent experiments are plotted in d. Asterisk indicates P < 0.001, ANOVA. Similar results were obtained using independent clones (see Supplementary Information, Fig. S4).

Periodic Ca²⁺ release from the ER can elevate mitochondrial matrix [Ca²⁺] that stimulates Kreb's cycle dehydrogenases, elevating mitochondrial [NADH]^12,27. Bcl-X_L elevated NAD(P)H levels in both wild-type and InsP₃R-KO cells, as observed in other cell types²⁸, but the enhancement was greater in the cells expressing InsP₃R (Fig. 5c, d). In wild-type cells, the elevated signal was in part due to increased mitochondrial NADH because the FCCP-induced fluorescence decrease from basal levels — although not different from those observed in InsP₃R-KO cells — was greater in the Bcl-X_L-expressing cells (P < 0.05; n = 7), a result that cannot be accounted for by increased numbers of mitochondria (Fig. 4a). Thus, spontaneous Bcl-X_L -dependent InsP₃R-mediated [Ca²⁺]_i signals are correlated with enhanced NAD(P)H levels, consistent with these [Ca²⁺]_i signals stimulating mitochondrial dehydrogenases²⁷ under resting conditions. Although anti-IgM induced similar [Ca²⁺]_i signals in wild-type cells in either the presence or absence of Bcl-X_L (Fig. 4b), NAD(P)H fluorescence was unchanged in the Bcl-X_L-transfected wild-type cells (Fig. 5c,d), whereas it was enhanced in the control wild-type cells (Fig. 5c) to levels closer to those achieved in the Bcl-X_L-transfected cells (Fig. 5c), with kinetics correlated with the BCR-induced InsP₃R-mediated [Ca²⁺]_i signal (Fig. 4b). These results suggest either that mitochondrial dehydrogenases in the Bcl-X_L-transfected wild-type cells were already maximally stimulated under resting conditions, or that Ca²⁺-stimulated NADH production was well compensated by increased oxidation with higher flux through the electron transport chain in these cells. In InsP₃R-KO cells, anti-IgM elicited neither a [Ca²⁺]_i response (data not shown) nor a stimulation of NAD(P)H fluorescence (Fig. 5c, d), demonstrating that the stimulation observed in wild-type cells was mediated by ER Ca²⁺ release. Similar results were obtained in different cell clones (see Supplementary Information, Fig. S4c). Thus, these results suggest that Bcl-X_L enables mitochondria to cope with increased metabolic demand with high efficiency that preserves mitochondrial redox status, dependent upon the InsP₃R.

Together the above results suggest that the effects of Bcl-X_L at the ER are anti-apoptotic not only because they minimize the maximal delivery of Ca²⁺ to mitochondria as previously thought, but also because Bcl-X_L facilitates InsP₃R-mediated Ca²⁺ delivery to mitochondria that regulates mitochondrial bioenergetics. Such a model is consistent with a general scheme that links cellular resistance to apoptosis with the regulation of cellular metabolism^29,30. Furthermore, it is consistent with the observation that Bcl-X_L does not stimulate an unregulated leak — which would be energetically costly — but rather modulates the activity of an exquisitely regulated permeability. Nevertheless, because our results do not establish a causal link between enhanced bioenergetics and apoptosis susceptibility, other models are also possible. For example, it may be that the ER Ca²⁺ load per se, regulated by the interaction of Bcl-X_L with the InsP₃R, rather than Ca²⁺ release, directly determines apoptotic susceptibility. Bcl-X_L interacts with the InsP₃R at the ER membranes and increases its sensitivity to very low levels of InsP₃, resulting in increased channel gating and cytoplasmic [Ca²⁺]_i signals that enhance mitochondrial bioenergetics and increase apoptosis resistance. Anti-apoptotic Bcl-2 family members promote a more dynamic InsP₃R response to achieve more sensitive coupling of Ca²⁺ release to extracellular signals and to increase cell survival. We suggest that Bcl-X_L interaction with the InsP₃R Ca²⁺ release channel is a critical cellular control point that couples the ER to apoptotic pathways. The interaction between Bcl-X_L and InsP₃R may provide a novel therapeutic target, with implications for processes involving apoptosis including cancer, neurodegeneration and cardiovascular disease.

METHODS

Cell culture and transfection

Spodoptera frugiperda (Sf9) cells were maintained in suspension culture at 27 °C in serum-free Sf-900 II SFM media (GIBCO/BLR, Gaithersburg, MD). A baculovirus expression system was used to transiently express Bcl-X_L. COS-7 (Cercopithecus aethiops kidney) cells were grown in DMEM/high glucose medium containing 10% (v/v) fetal bovine serum (FBS) (GIBCO/BLR) in a humidified 95/5% air/CO₂ atmosphere. DT40 cells were maintained in suspension culture at 37 °C (95/5% air/CO₂) in RPMI 1640 media (GIBCO/BLR) supplemented with 10% (v/v) FBS, 1% chicken serum, 2 mM glutamine, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin. Full-length human Bcl-X_L cDNA was cloned into pIRES2-DsRed2 vector (Clontech, Palo Alto, CA). Cells were transfected with either empty vector pIRES2-DsRed2 or pBcl-X_L-IRES2-DsRed2 using a Nucleofector Device (Amaxa, Gaithersburg, MD). For selection of stable clones, transfected cells were cultured for 2 weeks in the presence of 2 mg ml⁻¹ G418. Transfected cells were identified on the basis of DsRed fluorescence and were further sub-cloned using fluorescence-activated cell sorting (Becton Dickinson, San Jose, CA) into individual wells of a 96-well plate. Bcl-X_L expression was confirmed by western blot and quantified using infrared imaging Odyssey (LI-COR).

Biochemistry

Human Bcl-X_L or Bcl-2 cDNA in pGEX6P-1 (Amersham Pharmacia, Piscataway, NJ) were expressed as GST-fusion proteins in Escherichia coli (BL-21; Stratagene, La Jolla, CA) and immobilized on glutathione–Sepharose 4B (Amersham Pharmacia). InsP₃R (type 1) fragments were cloned into pcDNA3.1 or pcDNA3.1/V5 vectors (Invitrogen, Carlsbad, CA). COS-7 cells were transfected (Lipofectamine 2000), and 48–60 h later washed twice with PBS and harvested into 1 ml PBS containing 2% glycerol, 0.05% Triton and protease inhibitor cocktail (Sigma, St Louis, MO). After brief sonication (10 s) and centrifugation, total protein concentration in the lysate was adjusted to 5 mg ml⁻¹ and incubated with GST-fusion protein (1 h, 4 °C). Beads were centrifuged, washed three times and prepared for western blot. Co-immunoprecipitation and western blot analysis were performed according to standard protocols.

Baculoviral generation and protein purification

Baculoviruses were generated by first sub-cloning full-length human Bcl-X_L into pVL1393 vector (BD PharMingen, Bedford, MA) with specific tags. pVL1393-flag-Bcl-X_L and pVL1393-his-Bcl-X_L were transfected into Sf9 cells and viral supernatants collected. Viral supernatants were used to infect Sf9 cells until recombinant Bcl-X_L expression was maximized. Flag-tagged Bcl-X_L was purified by anti-FlagM2-agarose affinity beads (Sigma) and eluted by incubation with buffer 0.1 μg ml⁻¹ Flag peptide. Bcl-X_L cDNA for His-tagged recombinant Bcl-X_L was cloned into pET-16b (Novagen, Madison, WI) and expressed in BL-21, and the resulting His-tagged Bcl-X_L was purified on a Ni-NTA column (Qiagen, Valencia, CA). The plasmid pGEX-4T-1-tBid-his (gift from D. Newmeyer) was transformed into BL-21 and the cell lysate applied to glutathione-Sepharose, digested with thrombin (Novagen; 4 U ml⁻¹, 25 °C, 5 h ) and purified using Ni-NTA chromatography. PTYB1-mBax was obtained by inserting mouse Bax cDNA into pTYB1 vector (New England Biolabs, Beverly, MA). E. coli ER2566 (New England Biolabs) bearing the plasmid pTYB1-mBax were cultured and the fusion protein of Bax–chitin-binding protein was purified using a chitin-affinity column (New England Biolabs). Full-length Bax protein was eluted by incubating with 50 mM DTT at 16 °C for 40 h. All purified recombinant proteins were dialysed in buffer with 20 mM HEPES pH 7.5 and 20% glycerol, and stored at –80 °C. Protein samples for patch-clamping were first dialysed against buffer containing 10 mM HEPES pH 7.3, 140 mM KCl and 10% glycerol, and concentrated using Microcon (Millipore, Bedford, MA) to concentrations higher than 50 μM.

Electrophysiology

Sf9 cells were washed twice with PBS and suspended in a nuclear isolation solution containing: 150 mM KCl, 250 mM sucrose, 1.5 mM β-mercapoethanol, 10 mM Tris-HCl, 0.05 mM phenylmethylsulphonyl fluoride, protease inhibitor cocktail (Complete, Roche Diagnostics, Indianapolis, IN), pH 7.5. Nuclei were isolated using a Dounce glass homogenizer and plated onto a 1-ml glass-bottomed dish containing standard bath solution: 140 mM KCl, 10 mM HEPES and 0.5 mM BAPTA (free [Ca²⁺] = 300 nM), pH 7.1. The pipette solution contained 140 mM KCl, 0.5 mM ATP, 10 mM HEPES, pH 7.1. The free [Ca²⁺] in all solutions was adjusted to the desired level by the addition of an appropriate Ca²⁺ chelator, as described previously¹⁷. Experiments were performed at room temperature. Data were acquired using an Axopatch-1D amplifier (Axon Instruments, Union City, CA) and single-channel analysis performed using QuB software (University of Buffalo).

Calcium measurement and apoptosis assay

DT40 cells were plated onto a glass-bottomed perfusion chamber mounted on the stage of an inverted microscope (Nikon Eclipse TE2000) and incubated with Fura-2 AM (Molecular Probes, Eugene, OR; 2 μM) for 30 min at room temperature in normal culture media. Cells were then continuously perfused with Hanks’ balanced salt solution (Sigma), containing 1.8 mM CaCl₂ and 0.8 mM MgCl₂, pH 7.4. In experiments in which calcium oscillations were measured, cells were perfused with complete culture media without phenol. Fura-2 was alternately excited at 340 and 380 nm, and the emitted fluorescence filtered at 510 nm was collected and recorded using a CCD-based imaging system running Ultraview software (Perkin-Elmer, Norwalk, CT). Thapsigargin (Sigma) was applied in Ca²⁺-free Hanks’ solution containing 0.5 mM LaCl₃ to block Ca²⁺ extrusion across the plasma membrane³¹. Dye calibration was achieved by applying experimentally determined constants to the standard equation: [Ca²⁺] = K_d β(R−R_min)/(R_max−R). Cell viability was determined after application of anti-BCR antibody (IgM; SouthernBiotech, Birmingham, AL) using DAPI staining (1 μg ml⁻¹) and assays were performed on an LSR flow cytometer (Beckton Dickinson).

NAD(P)H measurements

DT40 cells (10 × 10⁶ cells per ml) were suspended in Hanks’ balanced salt solution (Sigma). Autofluorescence of NAD(P)H was monitored at 350/460 nm (excitation/emission) using a multi-wavelength excitation, dual-wavelength emission fluorimeter (Delta RAM, PTI , Birmingham, NJ). Experiments were performed at 37 °C.

Analysis and statistics

Data were summarized as the mean ± s.e.m. and the statistical significance of differences between means was assessed using unpaired t-tests or analysis of variance (ANOVA) for repeated measures, using Fisher's protected least-significant difference (PLSD) test. Differences between means were accepted as statistically significant at the 95% level (P < 0.05).

BIND identifiers

Four BIND identifiers (www.bind.ca) are associated with this manuscript: 331015, 331016, 331017 and 331018.