N-terminally cleaved Bcl-xL mediates ischemia-induced neuronal death.

Transient global ischemia in rats induces delayed death of hippocampal CA1 neurons. Early events include caspase activation, cleavage of anti-death Bcl-2 family proteins and large mitochondrial channel activity. However, whether these events have a causal role in ischemia-induced neuronal death is unclear. We found that the Bcl-2 and Bcl-x(L) inhibitor ABT-737, which enhances death of tumor cells, protected rats against neuronal death in a clinically relevant model of brain ischemia. Bcl-x(L) is prominently expressed in adult neurons and can be cleaved by caspases to generate a pro-death fragment, ΔN-Bcl-x(L). We found that ABT-737 administered before or after ischemia inhibited ΔN-Bcl-x(L)-induced mitochondrial channel activity and neuronal death. To establish a causal role for ΔN-Bcl-x(L), we generated knock-in mice expressing a caspase-resistant form of Bcl-x(L). The knock-in mice exhibited markedly reduced mitochondrial channel activity and reduced vulnerability to ischemia-induced neuronal death. These findings suggest that truncated Bcl-x(L) could be a potentially important therapeutic target in ischemic brain injury.

Transient global or forebrain ischemia, arising as a consequence of cardiac arrest or cardiac surgery in humans or induced experimentally in animals, causes selective, delayed death of hippocampal CA1 pyramidal neurons and cognitive deficits at 3-7 days after insult[1-3]. Early events are disruption of the functional integrity of the outer mitochondrial membrane by the formation of large ion channels, mitochondrial release of cytochrome c, and activation of caspases[4-6]. Although it is clear that caspases are engaged in response to global ischemia, a full understanding of the role of caspase substrates in global ischemia-induced death is still unknown [7].

Bcl-2 family proteins regulate apoptotic cell death by controlling the permeability of the outer mitochondrial membrane. The prevailing view is that anti-apoptotic Bcl-2 family members such as BcL-xL prevent homo-oligomerization of pro-apoptotic family members Bax and Bak in the outer mitochondrial membrane, thereby preventing the release of cytochrome c to promote caspase activation and apoptotic cell death. Bcl-2/Bcl-xL act by directly inhibiting Bax/Bak and by inhibiting the activators of Bax/Bak[8, 9]. The cancer chemotherapeutic agent, ABT-737 mimics the BH3 domain of pro-apoptotic family proteins and binds Bcl-xL, Bcl-2, and Bcl-w with high affinity to inhibit their anti-apoptotic activity [10-13]

Bcl-xL not only influences neuronal survival, but also modulates neuronal activity under physiological conditions. Introduction of recombinant Bcl-xL protein into the presynaptic terminal of the squid giant axon potentiates transmitter release and vesicle recycling following intense synaptic activity [14], and injection of the Bcl-2/Bcl-xL inhibitor ABT-737 slows recovery of synaptic responses. In contrast, in hypoxic conditions, ABT-737 increases synaptic transmission, preventing hypoxia-induced synaptic rundown at the squid giant synapse[15]. Moreover, caspase activation is critical to induction of long term depression at Schaffer collateral to CA1 synapses [16]. This raises the unexpected possibility that Bcl-xL and/or other targets of ABT-737 can have opposing effects on synaptic strength, depending on whether the synapse is hypoxic. One way in which this could occur is through proteolytic cleavage of BcL-xL to generate its pro-death fragment ΔN-Bcl-xL[17].

Bcl-2 family proteins are substrates for caspases and other proteases; cleavage generally elicits pro-death activitiy[1819, 20]. Application of recombinant ΔN-Bcl-xL, the C-terminal cleavage product of Bcl-xL, activates large conductance channel activity[21] that mimics channel activity in mitochondria of postischemic neurons[621]. The present study provides evidence for a causal role of caspase-cleaved Bcl-xL in formation of mitochondrial channel activity. We show that treatment of animals with the Bcl-xL inhibitor ABT-737 prior to or after induction of global ischemia in vivo markedly inhibits ischemia-induced formation of large channel activity in mitochondria and neuronal death. Mutation of the caspase cleavage sites in Bcl-xL in mice attenuates ischemia-induced neuronal death. Our findings indicate that ABT-737 prevents cleavage of Bcl-xL and inhibits the activity of cleaved Bcl-xL, thereby affording protection against ischemia-induced neuronal death.

RESULTS

ABT-737 attenuates ischemia-induced neuronal death in rats

Transient global ischemia in rats, induced by the 4-vessel occlusion model (10 min), followed by reperfusion, mimics global ischemia arising in human brain following cardiac arrest[5, 6, 22]. Although the entire brain becomes hypoxic, neurons throughout the brain depolarize, ATP is depleted and a massive rise in intracellular Ca2+ occurs, global ischemia elicits highly selective, delayed death primarily of hippocampal CA1 pyramidal neurons. We first sought to determine whether anti-apoptotic Bcl-xL, protein, which is abundantly expressed in adult hippocampal neurons[23], protects neurons from ischemia-induced death. Surprisingly, pretreatment of animals with ABT-737 (1 μM final, a concentration sufficient to inhibit Bcl-xL in vivo,[10] injected unilaterally into the right ventricle stereotactically at 1 h prior to transient global ischemia), significantly inhibited neuronal death in the hippocampal CA1, assessed by the number of neurons that are positive for Fluoro-Jade (FJ), a marker of degenerating neurons (Fig. 1a, b) and by neuronal counts of toluidine blue-stained sections at 6 d postischemia relative to that of vehicle-injected animals subjected to ischemia (Supplementary Fig. 1a, b). To test ABT-737 in a more clinically relevant scenario, ABT-737 was administered at 15 min or 1 h following reperfusion. Under these conditions, ABT-737 afforded robust neuroprotection against ischemia-induced neuronal death in the hippocampal CA1 (Fig. 1a, b). In addition, ABT-737 attenuated activated caspase-3 as late as 5 d after ischemia (Supplementary Fig. 2). This is a novel and unexpected finding, given the extensive evidence that caspase-3 activity is increased as early as 1-3 h after ischemia[5, 21, 24, 25] and peaks at approximately 24 h after ischemia[5, 24] and that ultrastructural studies[26] provide compelling evidence that postischemic neurons exhibit morphological features of necrotic injury. These findings, together with our previous findings, demonstrate that ABT-737 suppresses a pro-death activity over an extensive time period.

Figure 1

Treatment with the Bcl-xL inhibitor ABT-737 protects against ischemia-induced neuronal death in the CA1

(a) High magnification images taken at 5 d after ischemia of Fluoro-Jade-labeled brain sections at the level of the dorsal hippocampus from animals treated with vehicle or ABT-737 (icv, 125 μM) at three times (1 h before ischemia (−1 h), 15 min after ischemia (+15 m), and 1 h after ischemia (+1 h)) and subjected to sham operation or global ischemia. (b) Histograms ± S.E.M (in all Figs.) showing number of degenerating neurons per region of interest within the hippocampal CA1 pyramidal cell layer. n = 4 sections per animal; number of animals per treatment group as indicated on bars, ttest *, P < 0.05. **, P < 0.01. ***, P < 0.001. “Veh”, vehicle, “ABT”, ABT-737.

Pretreatment of rats with ABT-737 attenuates ischemia-induced mitochondrial channel activity

Transient global ischemia in vivo induces large channel openings in the outer mitochondrial membrane of whole mitochondria isolated from brain[6, 21]. We next examined whether ABT-737 attenuates large channel activity in mitochondria by performing patch clamp recordings of whole mitochondria isolated from the hippocampal CA1 at 1 h after ischemia or sham surgery. ABT-737 (administered at 1 h prior to ischemia) had little or no effect on sham-operated animals, but markedly attenuated the appearance of ischemia-induced large (>760 pS) and intermediate (180-760 pS) conductance mitochondrial channel activity, and increased the closed-time and prevalence of small (<180 pS) channel activity (Fig. 2a-d). ABT-737 reduced the peak conductance of channel openings assessed by current-voltage analysis (Fig. 2e) and peak conductance at a single voltage (ABT-737-treated mitochondria, 118 ± 24 pS, n = 5 mitochondria, 3 animals; vehicle treated mitochondria, 922 ± 76 pS. n = 7 mitochondria, 3 animals, P < 0.0001). Thus, ABT-737 attenuates ischemia-induced induction of mitochondrial channel activity.

Figure 2

The Bcl-xL inhibitor ABT-737 blocks ischemia-induced mitochondrial channel formation

(a,b) Sample recordings from mitochondria isolated from the hippocampus at 1 h after ischemia. Animals were pretreated with ABT-737 (icv, 125 μM) or vehicle 1 prior to induction of global ischemia. Organelle-attached patches were recorded at Vh = −80 mV. (c,d) Histograms showing channel activity for recordings like those illustrated in (a,b). Channel activity was classified as follows: closed, small: < 180 pS, intermediate: 180 to 760 pS, and large: > 760 pS. For sham-operated vehicle-treated n=5 mitochondria, 35 traces (10 s per trace); ABT-737-treated, n=8 mitochondria, 51 traces. For ischemia, vehicle-treated, n = 10 mitochondria, 68 traces; ABT-737, n = 9 mitochondria, 62 traces. 6 animals were used to pool mitochondria for each condition. (e) Current–voltage relations for recordings from mitochondria as in a, b. Vehicle, n = 7 mitochondria from 3 animals; ABT-737, n = 4 mitochondria from 3 animals. Pretreatment of animals with ABT-737 prior to ischemia decreased the slope of the I-V relationship, indicative of decreased conductance through the mitochondrial outer membrane. ttest **, P < 0.001. ***, P < 0.0001. “Intermed”, Intermediate.

ABT-737 directly attenuates ischemia-induced mitochondrial channel activity

To examine the impact of ABT-737 delivered to intact animals on mitochondrial channel activity, ABT-737 was applied in vitro to mitochondria isolated from hippocampal CA1 at 1 h after global ischemia. Application of ABT-737 via the patch pipette and bath perfusate markedly reduced ischemia-induced channel activity in isolated post-ischemic mitochondria (Fig. 3a, b). Moreover, ABT-737 reduced the peak conductance of channel openings, assessed by current-voltage analysis (Fig. 3c) and by determination of channel openings at a single voltage (ABT-737, 398 ± 62 pS, n = 8; vehicle, 966 ± 241 pS, n = 11, P < 0.05).

Figure 3

ABT-737 directly attenuates channel activity in post-ischemic mitochondria

(a) Representative sample recordings from mitochondria isolated from the CA1 of the hippocampus of experimental animals at 1 h after ischemia and treated in vitro in the absence or presence of ABT-737 (5 μM, applied via patch pipette and bath perfusate). Organelle-attached patches were recorded at Vh −100 mV. (b) Histograms showing closed, small, intermediate and large channel activity for mitochondrial recordings as in a. Channel activity was classified as described in the legend to Figure 2. Ischemia, n = 10 independent mitochondria recorded from 3 animals; ischemia + ABT-737, n = 9 independent mitochondria, recorded from 3 animals (c) Current–voltage relations for recordings from mitochondria as in a; ischemia, n = 11 mitochondria; ischemia + ABT, n = 7 mitochondria). ttest ***, P < 0.0001; **, P < 0.005; *, P < 0.05.

Because ABT-737 does not inhibit pro-apoptotic Bcl-2 family members Bax and Bak [10], these proteins are presumably not the targets of ABT-737 relevant to protection against ischemic injury. However, Bcl-xL, a known target of ABT-737[10] can be proteolytically-processed to release pro-apoptotic C-terminal cleavage fragments[17, 27]. We examined the impact of ischemia on the appearance of cleaved Bcl-xL in CA1. Under physiological conditions, ΔN61-Bcl-xL was present at low abundance (Fig. 4a, b), consistent with the possibility that caspases may have physiological, as well as a pathological, functions[16]. Ischemia increased ΔN61-Bcl-xL abundance, assessed by Westerns probed with an antibody specific for ΔN61-Bcl-xL (n = 5 independent experiments). Ischemia promoted formation of cleaved Bcl-xL evident at 1 h and at 24 h, but levels declined to near control values by 48 h, a time when histologically-detectable cell death is first apparent (Fig. 4a,b, Supplementary Fig. 3)[21]. In contrast, Bcl-2 and Bcl-w were unchanged (Supplementary Fig. 3).

Figure 4

Pretreatment of animals with ABT-737 attenuates ischemia-induced cleavage of Bcl-xL

(a) Western blot analysis of protein samples from the mitochondrial fraction of the hippocampus taken at 1 hr after surgery from animals pretreated with vehicle or ABT-737 (icv, 125 μM) and subjected to sham operation or global ischemia. Pretreatment of animals with ABT-737 markedly attenuated appearance of ΔN-Bcl-xL, the cleavage fragment of Bcl-xL. (b) Summary data showing Bcl-xL and ΔN-Bcl-xL abundance. The abundance of ΔN-Bcl-xL was normalized to that of full-length Bcl-xL by means of ImageJ software. Sham, n = 5; ischemia, n = 7; sham+ABT-737, n = 5; ischemia+ABT-737, n = 6 animals; ttest *, P < 0.05, **, P < 0.005. “Isch”, Ischemia. Western blot images illustrated in panel a have been cropped. Full-length blots are presented in Supplementary Fig. 11.

ABT-737 inhibits ΔN-Bcl-xL–elicited channel activity of mammalian brain mitochondria

In squid giant synapse, injection of the Bcl-2/Bcl-xL inhibitor ABT-737 into the presynaptic terminal slows the recovery of synaptic responses after repetitive synaptic activity, but ameliorates hypoxia-induced synaptic rundown[15], raising the unexpected possibility that Bcl-xL and/or other targets of ABT-737 can produce opposing effects on synaptic strength. In response to injurious stimuli, endogenous Bcl-xL can be cleaved into two cleavage fragments, ΔN61-Bcl-xL or ΔN76-Bcl-xL, each of which elicits cell death in cultured cells[17, 27]. We tested the effects of recombinant ΔN61-Bcl-xL and ΔN76-Bcl-xL, introduced via the patch pipette, on the induction of channel activity under control (non-ischemic) conditions. Both proteins induced discrete intermediate- and large-conductance channel openings in response to a wide range of amplitudes (Fig. 5a,b, Supplementary Fig. 4a,b)[15, 21]. Although application of ABT-737 alone decreased the appearance of small conductance channel activity (Fig. 5a-c), application of ABT-737 together with recombinant cleaved Bcl-xL via the patch pipette prevented the appearance of large and intermediate conductance channel activity (Fig. 5a-c, Supplementary Fig. 4a,b) and decreased the ΔN61-Bcl-xL-elicited peak conductance (ΔN61-Bcl-xL, 911 ± 81 pS; ΔN61-Bcl-xL+ABT-737, 384 ± 58 pS, n = 25 of each, P < 0.01). Application of ABT-737 to mitochondria produced a greater inhibition of ΔN61-Bcl-xL-than ΔN76-Bcl-xL-elicited channel activity (Fig. 5a-c, Supplementary Fig. 4a,b). Whereas ΔN61-/76-Bcl-xL elicited channel activity similar to that reported for the pro-apoptotic Bcl-2 family protein Bax[21, 28, 29], ABT-737 does not bind or inhibit Bax[30]. Therefore, Bax is not a likely candidate for the ischemia-induced large conductance channel activity. For subsequent experiments, we focused on ΔN61-Bcl-xL (hereafter termed ΔN-Bcl-xL).

Figure 5

ABT-737 attenuates ΔN61-Bcl-xL-elicited channel activity and cytochrome c release

(a) Sample mitochondrial recordings from control hippocampus performed in the absence and presence of ABT-737. Organelle-attached patches recorded at Vh = +60 mV. (b) Sample mitochondrial recordings from control hippocampus performed in the absence and presence of ΔN61-Bcl-xL (30 μg/ml) or ΔN61-Bcl-xL+ABT-737 (5 μM, applied via the patch pipette). Organelle-attached patches recorded at Vh = +60 mV. (c) Histograms showing closed, small, intermediate and large channel activity for recordings like those illustrated in a and b. Control, 8 (X10sec) traces per mitochondrion, N=4 mitochondria; control + ABT-737, 30 traces n = 3 mitochondria; ΔN61-Bcl-xL, n = 25 traces, 5 mitochondria; ΔN61-Bcl-xL + ABT-737, n = 25 traces, 5 mitochondria. (d, e) Mitochondria were treated with recombinant ΔN61-Bcl-xL at concentrations as indicated. Mitochondrial pellet and supernatant were analyzed for cytochrome c, COX IV, SMAC and VDAC. The concentration of cytochrome c in the mitochondrial fraction was normalized to cytochrome c in the cytoplasmic fraction. (f) Summary data from d, e showing cytochrome c release from the mitochondria (n = 3-6 samples per treatment). (g, h) Mitochondria were treated with recombinant ΔN61-Bcl-xL (1 μM) in the presence or absence of 5 μM ABT-737 (n = 3-4 samples per treatment). Mitochondrial pellet and supernatant were analyzed for cytochrome c and VDAC. (i) Summary data from g, h showing that ABT-737 prevents cytochrome c released from the mitochondria. (n = 6-9; 2-4 samples per treatment group per experiment; 3 independent experiments). ttest *, P < 0.05; **, P < 0.01. Western blot images illustrated in panels d, e, g and h have been cropped. Full-length blots are presented in Supplementary Fig. 11.

Caspase cleavage of Bcl-2/Bcl-xL has been reported to promote release of cytochrome c from lipid vesicles and from mitochondria[31, 32]. Although recombinant ΔN61-Bcl-xL promoted only limited release of cytochrome c from hippocampal mitochondria, this effect was reversed by 1 μM ABT-737 (Fig. 5d-i, Supple. Fig. 5), suggesting that ABT-737 can inhibit the direct effects of ΔN61-Bcl-xL.

ΔN-Bcl-xL induces cell death in neurons and in cells lacking Bax and Bak

ΔN-Bcl-xL could act directly to promote neuronal death, or could act as an activator of Bax or Bak to promote death. To examine whether ΔN-Bcl-xL elicits cell death in neurons in a Bax- or Bak-independent manner, Bax/Bak double knockout mouse embryonic fibroblasts (MEFs) were transfected with ΔN-Bcl-xL. Transfected cells were marked with co-transfected eGFP and viability was assessed by MTS assay (see Methods) at 18 h after transfection. Under these conditions, ΔN61-Bcl-xL induced cell death in cells lacking Bax and Bak, relative to that of cells expressing eGFP alone (which itself induced little or no cell death). Application of ABT-737 to cells at the time of transfection markedly inhibited ΔN-Bcl-xL-elicited cell death (Fig. 6a). These findings indicate that ΔN-Bcl-xL is sufficient to trigger cell death, even in the absence of Bax and Bak. In contrast, experiments performed on single knockout MEFs lacking either Bax or Bak suggest that Bax and Bak can participate in ΔN-Bcl-xL-induced cell death, and that ABT-737 is ineffective at inhibiting this death (Supplementary Fig. 6), presumably due to the effects of Bax and/or Bak alone.

Figure 6

ΔN61-Bcl-xL induces cell death in hippocampal neurons and Bax−/− Bak−/− MEFs

(a) ABT-737 attenuates ΔN61-Bcl-xL-elicited cell death in double knock out MEFs. Double knockout MEFs were transfected with ΔN61-Bcl-xL plus eGFP or eGFP alone. Summary data show percent cell survival as assayed by MTS solution (see methods) in the absence and presence of 200 nM ABT-737 (Bax−/− Bak−/− MEFs: control, n = 9; ABT-737, n = 10; ΔN61-Bcl-xL, n = 10; ΔN61-Bcl-xL + ABT-737, n = 6; Results represent 5 independent experiments (b) Hippocampal neurons at DIV 14 expressing eGFP with ΔN61-Bcl-xL or eGFP alone (green, eGFP; yellow arrowheads an example of a transfected neuron) at 24 h after transfection were assayed for caspase-3-like activity (red). ΔN61-Bcl-xL, had increased cell death as compared to eGFP control. (c) ABT-737 attenuates ΔN61-Bcl-xL-elicited cell death in hippocampal neurons. Hippocampal neurons were transfected with eGFP and indicated constructs and maintained in the absence or presence of ABT-737 (1 μM, applied in the medium). At 4 d after transfection, DIV 18, the number of dead cells was detected by propidium iodide uptake and abnormal morphology and expressed as a percent of total transfected cells (eGFP, n = 22 coverslips; eGFP+ABT-737, n = 10 coverslips; eGFP+Bax, n = 10 coverslips; eGFP+Bax+ABT-737, n = 10 coverslips; eGFP+ΔN61-Bcl-xL, n = 16 coverslips; eGFP+ΔN61-Bcl-xL+ABT-737, n = 11 coverslips from 3 independent experiments; P < 0.001, ΔN61-Bcl-xL vs. ΔN61-Bcl-xL+ABT-737; P < 0.001, ΔN61-Bcl-xL vs. eGFP). ttest: ***, P < 0.001. **, P < 0.005; *, P < 0.05.

To verify that expression of ΔN-Bcl-xL can trigger death of neurons, we co-expressed ΔN61-Bcl-xL with eGFP to mark transfected primary hippocampal neurons (DIV 18, Fig. 6b). At 1-2 days after transfection, neurons expressing eGFP exhibited little or no cell death. In contrast, neurons expressing ΔN61-Bcl-xL exhibited substantial cell death, relative to the eGFP control, as monitored by caspase-3-like activity (44% ± 2, n = 30 cells expressing ΔN61-Bcl-xL+eGFP; 8% ± 2, cells expressing eGFP alone, n = 16 from three independent experiments, P<0.05). In a separate experiment, ABT-737 markedly attenuated death of neurons expressing ΔN61-Bcl-xL as assessed by cell morphology (see Methods) and propidium iodide uptake (a marker of degenerating neurons/cells) by eGFP-positive neurons (Fig. 6c). In contrast, neurons expressing eGFP alone exhibited little or no cell death. Neurons expressing Bax exhibited less cell death relative to that of neurons expressing ΔN61-Bcl-xL, and the small fraction of cells exhibiting Bax-elicited cell death were not protected by ABT-737.

Knock-in of cleavage-resistant Bcl-xL protects mice from ischemic injury

To determine whether protease cleavage of endogenous Bcl-xL contributes to cell death in CA1 neurons following ischemia-reperfusion, we constructed knock-in mice in which Bcl-xL harbors mutations at both caspase cleavage sites, D61A and D76A, rendering Bcl-xL resistant to cleavage by caspases (Fig. 7c; Supplementary Fig. 7). The mice were viable and fertile and exhibited no gross morphological changes in brain anatomy. We subjected wild type and homozygous Bcl-xL cleavage-resistant mice to either sham surgery or transient global ischemia induced by bilateral common artery occlusion (BCCO), followed by reperfusion (see Methods). Sham-operated (control) wild type and knock-in mice did not differ in the number of neurons in the hippocampal CA1 (Fig. 7a,b) and showed no signs of degeneration, as assessed by FJ staining (Supplementary Fig 8). In wild type mice subjected to global ischemia, we observed significant cell death in the CA1, as assessed by neuronal counts of the hippocampal CA1 in toluidine blue-stained sections at 6 days after surgery, relative to sham wild type controls. In contrast to wild type mice, cleavage-resistant Bcl-xL knock-in mice were resistant to ischemia-induced neuronal death in the hippocampal CA1 (Fig. 7a,b). Furthermore, knock-in mice were rescued from ischemia-induced neuronal degeneration, as assessed by FJ-staining at 5 d after ischemia (Supplementary Fig 8). To determine if blocking cleavage of Bcl-xL attenuated mitochondrial channel activity, we isolated mitochondria from sham and post-ischemic knock-in mice and wild type littermates, and performed patch clamp recordings of whole mitochondria. Non-ischemic wild type and KI mice failed to show significant differences in large and intermediate conductance channel activity. In contrast, while ischemia elicited a significant increase in channel activity in wild type mitochondria compared to non-ischemic wild type controls, mitochondria isolated from ischemic knock-in animals had significantly less channel activity than that of ischemic wild type mice (Fig. 7d, e), suggesting that cleavage of endogenous Bcl-xL was required for the increase in mitochondrial channel activity after ischemia in mouse brain.

Figure 7

Bcl-xL cleavage-resistant mice are protected against ischemia-induced neuronal death

(a) Toluidine blue-stained coronal brain sections at the level of the dorsal hippocampus at 6 d after in vivo ischemia from wild-type and homozygous Bcl-xL cleavage-resistant knockin mice. (b) Summary data of neuronal counts within the region of interest. Number of animals per treatment group is indicated on bars, 4 sections per animal. (c) Westerns probed with an anti-Bcl-xL antibody that detects both full-length and N-terminally truncated forms of Bcl-xL. Residual cleavage of Bcl-xL in slices of homozygous mice is presumably due to calpain-mediated proteolytic activity, which increases after ischemia (Yamashima et al., 1996). (d) Sample recordings from mitochondria isolated from the hippocampus of control (sham-operated) or ischemic wild-type and knock-in animals. Organelle-attached patches recorded at Vh = −100 mV. (e) Histograms showing closed, small, intermediate and large channel activity for recordings like those illustrated in d. Channel activity was classified as follows: closed, small: < 180 pS, intermediate: 180 to 760 pS, and large: > 760 pS., n = 5-14 10-s traces per condition. The number of mice was: 4 wild-type sham, 4 KI homozygous sham, 3 wild-type ischemic, 3 KI homozygous ischemic. For sham vs. ischemia: ttest *, P < 0.05; **, P < 0.01; for wild-type vs. knock-in: #, P < 0.05; ##, P < 0.01. Western blots illustrated in panel c have been cropped. Full length blot is presented in Supplementary Fig. 11.

To examine the effect of knock-in mice in a second model of neuronal death, we prepared hippocampal slice cultures from homozygous knock-in mice, heterozygous and wild type littermates and from wild type rats (DIV 9) and subjected slices to oxygen-glucose deprivation (OGD), a well-established in vitro model of global ischemia (45 min, followed by 2 days of reperfusion). Hippocampal slices from knock-in mice exhibited a marked reduction in OGD-induced neuronal death, compared with slices from heterozyous or wild type littermates, as assessed by propidium iodide uptake (Supplementary Fig. 9 a, b). OGD-induced formation of ΔN-Bcl-xL in hippocampal slices from wild type mice was increased compared to knock-in cleavage-resistant Bcl-xL mice (Fig. 7c). To examine the effect of ABT-737, OGD was performed in rat hippocampal slices. OGD induced cleavage of Bcl-xL to generate ΔN-Bcl-xL and elicited neuronal death (Supplementary Fig. 10 a, b). Application of ABT-737 (5 μM) to rat hippocampal slices attenuated OGD-induced neuronal death, assessed by propidium iodide uptake at 24-48 h after ischemia (Supplementary Fig. 10 b). Taken together, these data strongly suggest that cleavage of Bcl-xL contributes to the selective, delayed neurodegeneration associated with global ischemia and that truncated Bcl-xL contributes importantly to delayed ischemia-induced death of hippocampal neurons.

DISCUSSION

Transient global or forebrain ischemia arising as a consequence of cardiac arrest or open heart surgery elicits selective, delayed death of hippocampal CA1 neurons and cognitive deficits[1-3, 33, 34]. Appearance of large channel activity in mitochondrial membranes and release of cytochrome c are hallmarks of the early post-ischemic period. Proteolytic cleavage of the anti-apoptotic protein, Bcl-xL, to generate ΔN-Bcl-xL is associated with the formation of large-conductance mitochondrial channels, and with release of cytochrome c in postischemic neurons[6, 21]. Here we show that expression of ΔN-Bcl-xL in hippocampal neurons and mouse embryonic fibroblasts lacking Bax and Bak elicits cell death. We further show the unexpected finding that pretreatment of animals with the Bcl-2/Bcl-xL inhibitor ABT-737, which elicits apoptosis in a wide array of tumor cells[10-13], when given before or after ischemia, markedly attenuates ischemia-induced cleavage of Bcl-xL, formation of large channel activity in the mitochondrial outer membrane and affords robust protection of CA1 neurons. Consistent with this, ischemia-induced mitochondrial channel activity and death of hippocampal neurons is attenuated in vivo and in organotypically cultured hippocampal slices from knock-in mice expressing a mutated form of Bcl-xL resistant to protease-dependent cleavage. These findings support a role for ΔN-Bcl-xL in the delayed cell death of hippocampal CA1 neurons and implicate ΔN-Bcl-xL as a putative target for therapeutic intervention in brain ischemic injury. Although not addressed by the present study, it remains to be seen whether ABT-737 administered many hours after the ischemic event would still afford protection. ABT-737 is up to now the most specific and selective small molecule inhibitor designed against Bcl-xL and thus provides important proof-of-principle for development of future therapeutic compounds. Other agents similar to ABT-737 have been designed that may cross the blood brain barrier, but their specificity and selectivity have not withstood the rigorous tests that have been applied to ABT-737. Nevertheless, these other drugs may have clinical relevance in brain ischemia paradigms in future studies.

ABT-737 protects hippocampal neurons from ischemia-induced cell death

A novel finding of the present study is that treatment of animals with ABT-737 prior to or after induction of global ischemia affords robust protection of hippocampal CA1 neurons in a clinically relevant model of global ischemia. These findings are somewhat unexpected in that ABT-737 is a BH3-mimetic which inhibits anti-apoptotic Bcl-2 family members by binding within the hydrophobic cleft typically occupied by pro-apoptotic, BH3-domain only proteins such as Bim and Bad[10]. ABT-737 exhibits high affinity for Bcl-xL, Bcl-2 and Bcl-w, with approximately 10-fold higher affinity for Bcl-xL than for Bcl-2[10]. ABT-737 promotes apoptosis in lymphoma, multiple myeloma and small-cell lung carcinoma lines, as well as primary patient-derived cancer cells, thereby effectively suppressing tumors[10-13]. ABT-737 (ABT-263) is presently in clinical trials as an anti-tumorigenic agent that promotes regression of solid tumors[10-13]. In tumor cells, the mechanism by which ABT-737 elicits cell death is well delineated: ABT-737 sequesters Bcl-xL away from pro-apoptotic BH3-only proteins. When unleashed, these players initiate the oligomerization of Bax and Bak, which permeabilize the outer mitochondrial membrane[8, 9]. Nevertheless, given that ABT-737 can bind both anti-apoptotic full-length Bcl-xL, as well proapoptotic ΔN-Bcl-xL, it would be difficult to infer from findings in squid (our previous work) or rats (present study) what the actual impact of ABT-737 would be in humans in a clinical context.

Findings in the present study are consistent with a model whereby ABT-737 also binds both anti-apoptotic and pro-apoptotic ΔN-Bcl-xL, thereby aborting ischemia-induced neuronal death by preventing cleavage of full length Bcl-xL (as in Fig. 4a, b) and by blocking the channel activity and death-inducing effects of cleaved Bcl-xL. Consistent with this, expression of ΔN-Bcl-xL, but not Bax, elicited cell death in hippocampal neurons and ΔN-Bcl-xL also produced death in mouse embryonic fibroblasts lacking Bax and Bak. Whereas ABT-737 attenuated ΔN-Bcl-xL-elicited cell death in hippocampal neurons and in bax double knockout MEFs (present study see Fig. 6a), ABT-737 had little or no effect on cell death in single knockout MEFs that contain Bax or Bak. Thus, Bax/Bak may also contribute to cell death observed in MEFs. However, in adult neurons, endogenous Bcl-xL abundance is still high at ages when Bax and Bak expression have declined[23], and, although genetic ablation of Bax protects against cardiac ischemia[35] and Bax inhibitors protect against brain ischemia in other models[29], Bax inhibitors do not protect adult mice from ischemic injury induced by MCAO[36] and endogenous and overexpressed Bax and Bak also can be profoundly protective in the brain depending on the developmental stage, death stimulus and brain subregion[37, 38]. Furthermore, we show that exogenously expressed Bax does not significantly increase death of cultured hippocampal neurons (see Fig. 6c), indicating that these neurons are not very sensitive to Bax. Nevertheless, cleaved Bcl-xL may stimulate both Bax/Bak-dependent and independent death.

Generation of cleaved Bcl-xL in ischemic injury

Bcl-xL can be cleaved by both calpain and caspases[17, 39]. Unlike calpain, which has less defined substrate cleavage sites, caspases cleave specifically after Asp residues, and we found that mutation of the two known caspase cleavage sites at Asp61 and Asp76 of endogenous Bcl-xL protects neurons significantly from ischemic injury. The specific caspases required to generate the pro-death ΔN-Bcl-xL fragment in neurons subjected to global ischemia in vivo or to oxygen glucose deprivation in organotypically-cultured hippocampal slices are not known. The responsible caspases could be activated by a pathway prior to mitochondrial involvement, or by mechanisms not involving mitochondrial permeabilization. These possibilities are consistent with the non-consensus caspase cleavage recognition sites in Bcl-xL. Caspase-cleaved Bcl-xL could also potentially lead to increased cleavage of full-length Bcl-xL if ΔN-Bcl-xL leads to mitochondrial permeabilization and subsequent amplification of caspase activity. Regardless of the specific caspases involved, these findings are consistent with previous observations that caspase inhibitors ameliorate global ischemia-induced neuronal death[24, 40, 41].

ABT-737 inhibits ΔN-Bcl-xL induced channel activity

Another finding of the current study is that ABT-737 can inhibit functional activity associated with truncated, pro-apoptotic forms of Bcl-xL, leading to protection of postischemic neurons. First, ABT-737 inhibits the large channel activity in the outer mitochondrial membrane elicited by recombinant ΔN61- and ΔN76-Bcl-xL applied via the patch pipette. Second, ABT-737 attenuates ΔN-Bcl-xL-elicited cell death in hippocampal neurons and mouse embryonic fibroblasts lacking Bax and Bak. These findings provide a mechanism by which ABT-737 affords neuroprotection: namely, ABT-737 directly binds and inhibits ΔN-Bcl-xL and thereby prevents its ability to permeabilize the outer mitochondrial membrane and promote death in hippocampal neurons. Whereas these studies address inhibition of ΔN-Bcl-xL function by ABT-737, they do not establish exactly how ABT-737 inhibits channel activity. We envision at least three possible mechanisms by which ABT-737 might function. First, ABT-737 could act directly on ΔN-Bcl-xL. This is supported by the finding that ABT-737 can inhibit the effects of recombinant, as well as overexpressed ΔN-Bcl-xL. Second, ABT-737 might act directly on full-length Bcl-xL to render it less sensitive to proteolytic digestion, thereby blocking the cleavage of Bcl-xL to ΔN-Bcl-xL. Third, ABT-737 could act indirectly to reduce caspase cleavage of full-length Bcl-xL to ΔN-Bcl-xL in a feed-forward loop by suppressing the ability of ΔN-Bcl-xL to stimulate mitochondrial permeabilization leading to amplification of caspase activity. The latter two possibilities are consistent with our immunoblot analysis of processed Bcl-xL in ischemic brain.

In our study we identify a specific role for the cleavage of Bcl-xL in ischemia-induced neuronal death. We further show evidence for a ΔN-Bcl-xL-dependent activation of a Bax-like apoptotic pathway. However, these observations do not necessarily implicate apoptotic cell death as defined morphologically. Whereas global ischemia-induced neuronal death in the hippocampal CA1 has been attributed to apoptosis defined as caspase-dependent death[42], analysis of the hippocampal CA1 by electron microscopy following global ischemic injury, and in other ischemia models reveals necrotic morphology[26,43]. In addition, we have recently reported that Bcl-xL increases mitochondrial energetic efficiency[44, 45]. Thus, loss of full-length Bcl-xL together with ΔN-Bcl-xL–induced mitochondrial damage could result in a plethora of cellular defects leading to a range of cell morphologies.

In summary, the present study extends our previous findings that ischemic insult triggers cleavage of full length Bcl-xL to generate its pro-apoptotic cleavage product ΔN-Bcl-xL in that we show that ΔN-Bcl-xL expressed in hippocampal neurons or in mouse embryonic fibroblasts lacking Bax and Bak elicits cell death. We further show the novel finding that ΔN-Bcl-xL is critical to neuronal death in a clinically relevant model of global ischemia. A key event is disruption of the functional integrity of the outer mitochondrial membrane, as evident by the increase in large channel activity in the early post-ischemic period. In addition, we show the novel finding that ABT-737, known to promote apoptosis and death of tumorigenic cells, affords robust protection of hippocampal neurons from global ischemia-induced neuronal death. Together these findings identify Bcl-xL as a putative therapeutic target for intervention in the neuronal injury and cognitive deficits associated with global ischemia.