Bcr-Abl reduces endoplasmic reticulum releasable calcium levels by a Bcl-2–independent mechanism and inhibits calcium-dependent apoptotic signaling

Chronic myeloid leukemia (CML) is a clonal disorder of the hematopoietic system resulting in the excessive accumulation of immature and mature myeloid cells. The hallmark of CML is the presence of the Philadelphia (Ph) chromosome, which is present in more than 95% of cases and harbors the chimeric gene BCR-ABL. The resultant tyrosine kinase, Bcr-Abl, constitutively activates several signal transduction pathways^1,2  and is believed to be the critical determinant in the pathogenesis of CML.

CML is a biphasic disease, starting as an indolent chronic phase followed by a much more aggressive and drug-resistant phase called blast crisis. A specific inhibitor of Bcr-Abl tyrosine kinase activity, imatinib (Gleevec, Glivec) has recently been introduced for the treatment of CML.^3,4  Impressive responses have been seen with patients in chronic phase; however, resistance is common in advanced CML. Reported resistance mechanisms are primarily mutations and amplification of the BCR-ABL gene.^5-8 

A number of groups have suggested that the reason blast crisis (BC) CML is multidrug resistant is due to the antiapoptotic properties of the Bcr-Abl oncoprotein. It has already been shown that BCR-ABL can confer drug resistance when it is transfected into other leukemic cell lines.⁹  Several clinical and molecular studies have suggested that the expression levels of Bcr-Abl may play a major role in resistance to apoptosis.^10-13  Down-regulation of Bcr-Abl transcript has been shown to increase drug sensitivity in cells derived from CML.¹⁴  Expression of Bcr-Abl at different levels in hematopoietic progenitor cells found that all cells became growth factor independent, but only the high Bcr-Abl–expressing cells were resistant to apoptosis induced by cytotoxic drugs.¹⁰  This is analogous to the 32D transfected cell lines that we have characterized.^12,15,16 

It has recently been established that the endoplasmic reticulum (ER) can play a critical role in the regulation of apoptosis. The ER can activate apoptosis independently of or upstream of the mitochondria.^17,18  ER Ca²⁺ release leads to rapid calcium uptake and accumulation in mitochondria (ER/mitochondria coupling).^19-23  Mitochondrial calcium overload can promote permeability transition pore (PTP) opening and cytochrome c release. Specific inhibition of ER-to-mitochondrion Ca²⁺ transport attenuates downstream biochemical events associated with apoptosis. Proapoptotic and antiapoptotic Bcl-2 family members located in the ER have also been shown to modulate ER calcium homeostasis, thereby influencing susceptibility to apoptosis.^24-30 

The influence of the Bcr-Abl oncoprotein on ER-initiated apoptotic signaling pathways has never been investigated. Here, we show that Bcr-Abl–expressing cells exhibit a decreased amount of free releasable calcium in the ER as well as a weaker capacitative calcium entry (CCE) response relative to parental 32D cells. This results in attenuation of the ER/mitochondria-coupling process and mitochondrial calcium overload. This effect is independent of Bcl-2 expression. This is the first evidence that Bcr-Abl expression affects ER homeostasis and calcium-dependent apoptotic signaling. This ability will undoubtedly contribute to the antiapoptotic repertoire of the Bcr-Abl oncogene.

Etoposide (VP-16), thapsigargin (TG), propidium iodide (PI), streptolysin O, Hoechst 33258, and ruthenium red (RuRed) were obtained from Sigma-Aldrich (Dublin, Ireland). RuRed was taken up by cells on extended incubation, and therefore no permeabilization was undertaken. Calcium-Green-1 am (CaGreen-1), MitoTracker Green fm (MitoTracker Green), and rhodamine-2 am (Rhod-2) were from Molecular Probes (Leiden, The Netherlands). Caspase inhibitor zVAD.fmk was from Bachem (St Helens, United Kingdom) and calpain inhibitor III (MDL 28170) from Calbiochem (Darmstadt, Germany).

Bcr-Abl–transformed 32D cells (obtained from American Type Culture Collection, Manassas, VA) were cloned and cultured as described previously.¹⁶  K562, BV173, Jurkat, and HL60 cells were grown in RPMI 1640 with 10% FCS, 2 mM l-glutamine, 1% penicillin and streptomycin (P/S).

Cell viability was assessed by PI exclusion. Cells (0.5 × 10⁶) were resuspended in PBS and 50 μg/mL PI prior to flow cytometric analysis (FACScan; Becton Dickinson, Oxford, United Kingdom).

Cells were fixed with 3% para-formaldehyde for 15 minutes on ice, washed, and transferred to ice-cold 70% ethanol. For the terminal deoxynucleotidyl transferase–mediated dUTP nick-end labeling (TUNEL) assay, the 3′OH DNA ends were labeled with FITC-conjugated dUTP (Roche, Mannheim, Germany) by terminal deoxynucleotidyl transferase (TdT; Promega, Southampton, United Kingdom).

Cytosolic-free Ca²⁺ was assessed with the Ca²⁺-sensitive cell permeable dye CaGreen-1 and mitochondrial Ca²⁺ using Rhod-2. Cells (0.5 × 10⁶) were resuspended in 1 mL RPMI and loaded with either 1 μM CaGreen-1 or Rhod-2 for 15 minutes at room temperature. Labeled cells were washed twice with calcium-free Krebs-Henseleit buffer (120 mM NaCl, 4.8 mM KCl, 1.2 mM MgSO₄, 1.2 mM KH₂PO₄, 25 mM NaHCO₃, 5 mM HEPES, pH 7.2, 10 mM glucose). Analysis was performed using flow cytometry (Becton Dickinson) on FL-l for CaGreen-1 and FL-2 for Rhod-2, respectively.

Cells were resuspended in calcium-free medium and loaded with 1 μM CaGreen-1 as described (see “Measurement of intracellular calcium”). Fluorescence was quantitated by flow cytometry on the FL-1 channel as a time function. TG at a concentration of 2 μM was added after recording the basal cytosolic Ca²⁺ concentration (steady state). The ER calcium content was determined as the difference between the base level of CaGreen-1 fluorescence and the level after TG addition using the statistical analysis performed by CellQuest software (Becton Dickinson) and shown as arbitrary fluorescence units. CCE was determined by the subsequent addition of 5 mM CaCl₂ to the medium. Three regions corresponding to the steady-state cytosolic calcium level (R1), cytosolic calcium level after ER depletion by TG treatment (R2), and after capacitative calcium uptake (R3) and gated events were shown as histograms. The ER releasable calcium content and CCE were calculated as ΔF/F₀ = 100((F–F₀)/F₀), where F₀ is the median fluorescence of R1 (steady state) and F either the median fluorescence of R2 (ER content) or R3 (CCE). %ΔF/F₀ ratios from all experiments are presented as mean ± SD (mean %ΔF/F₀).

C2 and C4 cells were transfected with siRNA (230 pmol) corresponding to a 21-nucleotide (nt) region from Bcl-2 (M16506, nt 64-86) in the presence of streptolysin O (100 U). A nonspecific siRNA was used as a negative control.³¹  ER calcium content was measured 18 to 24 hours after transfection and cells were cytospun for immunofluorescence analysis.

siRNA-transfected cells were fixed and then permeabilized using 0.05% saponin, followed by incubation with anti–Bcl-2 (N-19) antibody (Santa Cruz Biotechnology, Santa Cruz, CA) at 1:500 and AlexaFluor-488 anti–rabbit IgG antibody at 1:200 (Invitrogen, Molecular Probes, Leiden, the Netherlands). Nuclei were counterstained with Hoechst 33258. Analysis was carried out using an Olympus BX51 fluorescence microscope equipped with an FITC filter (Olympus, Hamburg, Germany).

We had identified elevated expression of ER stress markers glucose-related protein (Grp78) and CAAT/enhancer-binding protein homologous protein (CHOP) in both high (C4) and low (C2) Bcr-Abl–expressing cells relative to parental 32D cells (Supplemental Figure S1, available on the Blood website by clicking the Supplemental Figures link at the top of the online article; refer also to the cell model described in our previous publications^12,15,16,31 ). To determine whether this was also associated with altered calcium homeostasis, we measured the relative amounts of steady-state ER releasable calcium in 32D and Bcr-Abl–expressing cells (Figure 1). The ER calcium content was assessed by quantifying TG-induced ER-calcium release. Addition of 2 μM TG to the cells in a Ca²⁺-free medium led to an immediate emptying of ER Ca²⁺ stores, observed as an increase in the cytosolic calcium level. ER calcium store depletion is also known to be a direct signal for CCE, which replenishes empty stores. This response is proportional to the amount of calcium released from the ER. We therefore investigated the CCE by subsequently adding 5 mM CaCl₂ to the medium. The further increase in the cytosolic calcium dye fluorescence indicated the level of calcium influx across the plasma membrane into the cytosol. The cytosolic calcium level was measured by flow cytometric analysis of cells loaded with a cytosolic calcium-sensitive fluorescent probe, CaGreen-1, and the level of fluorescence measured on the FL-1 channel was shown as a time function (Figure 1A). We found that in Bcr-Abl–expressing cells the amplitude of TG-induced Ca²⁺ release was markedly reduced compared with 32D cells. Accordingly, the CCE was much stronger in 32D cells, where the amount of calcium released from the ER was higher. Three regions corresponding to the steady-state cytosolic calcium level (R1), the calcium level after ER store depletion (R2), and after CCE to the cytosol (R3) were marked and shown as histograms (Figure 1B). Statistical analysis of data from triplicate experiments indicated that there is a 50% decrease in ER releasable calcium in C2 and C4 cells, followed by a similar decrease in CCE in these cells (Figure 1C).

To demonstrate that this reduced ER calcium content was a consequence of Bcr-Abl activity, we treated C4 and C2 cells with imatinib, a specific inhibitor of Bcr-Abl tyrosine kinase activity. We have previously shown that a concentration of 10 μM reduces cellular phosphotyrosine in C4 cells to basal (32D) levels.¹⁵  Treatment for a short period (4 hours) does not affect viability as demonstrated by lack of PI uptake (not shown). Imatinib treatment resulted in an increase in ER releasable calcium and stronger CCE in both cell lines, indicating that Bcr-Abl activity is required for their reduction in ER releasable calcium. Importantly, in 32D cells, imatinib did not alter the releasable calcium content or CCE, indicating that the drug is specifically influencing Bcr-Abl (Figure 1C).

To further investigate this association between Bcr-Abl expression and low ER releasable calcium, we examined the ER calcium content in 2 Bcr-Abl⁺ and 2 Bcr-Abl^– human hematopoietic cell lines. As shown in Figure 1D, the Bcr-Abl–expressing cell lines K562 and BV173 had comparable levels of releasable calcium in the ER to the 2 Bcr-Abl–expressing mouse cell lines, C2 and C4. The ER of Jurkat and HL60 cells contained higher amounts of releasable calcium. The CCE was also stronger in the Bcr-Abl^– cell lines than in K562 and BV173 cells. These data indicate that low releasable amounts of calcium in the ER and a weak CCE is not only a feature of C2 and C4 cells, but also of 2 Bcr-Abl⁺ human cell lines. We were, however, unable to elevate the ER calcium levels of K562 and BV173 to those of HL60 and Jurkat cells with imatinib. Although these cell lines are not directly comparable, due to different origins and lineages, these data suggest that, whereas 32D cells rely solely on the presence of Bcr-Abl to lower their ER calcium levels, the established human cell lines may have acquired additional survival mechanisms that contribute to the reduction in ER calcium. The attainment of low ER calcium may be a particularly important feature for cells transformed by Bcr-Abl.

Decreased levels of releasable calcium in the ER are associated with cell survival (for a review, see Szabadkai et al³² ) and are analogous to the reported effects of Bcl-2, which reduces ER releasable calcium and confers drug resistance. We therefore examined whether Bcl-2 may be acting downstream of Bcr-Abl and be responsible for the reduced releasable calcium. Bcl-2 transcript was down-regulated by siRNA expression in Bcr-Abl–expressing cell lines, and down-regulation of protein expression was confirmed by fluorescence microscopy (Figure 2A-B). The depletion of Bcl-2 had no effect on ER releasable calcium or CCE (Figure 2A-B).

We have previously reported that high levels of Bcr-Abl (as in C4 cells) can confer drug resistance.¹⁶  Inhibition of Bcr-Abl activity using imatinib can reverse the phenotype indicating that Bcr-Abl activity is associated with drug resistance and this phenotype is dose dependent. Cells expressing lower amounts of Bcr-Abl (C2 cells) are IL-3 independent, but their drug sensitivity is analogous to parental cells (Keeshan et al¹⁶  and subsequent data sections).

A previous study has reported that apoptotic stimuli can be classified into 3 categories depending on the apoptotic pathways they engage—calcium/ER, BH3/mitochondria, or both.²²  VP-16 was reported to engage both the ER and mitochondrial pathways. To investigate whether the lower steady-state ER calcium content observed in Bcr-Abl–expressing cells affects calcium changes associated with drug-induced ER/mitochondria coupling, levels of cytosolic and mitochondrial calcium were assessed following VP-16 treatment. Cells were treated with VP-16 at 7.5 μg/mL and the percentage of cell death was quantified by measuring PI uptake by flow cytometry.

To exclude the possibility that changes in internal calcium levels were a consequence of late apoptotic-cell membrane permeabilization, we analyzed cells after 14 hours of VP-16 treatment. At this time, the percentage of PI⁺ cells did not exceed 15%. Cells were loaded with the cytosolic Ca²⁺ probe CaGreen-1 or the mitochondrial probe Rhod-2 and analyzed by flow cytometry; representative histograms are shown in Figures 3A and 3B, respectively. In 32D cells loaded with CaGreen-1, a population with a higher cytosolic calcium level was apparent (38.8%; n = 3; 34.4% ± 5.5%) in contrast to C2 and C4 cells, where the increase in cytosolic calcium was much smaller, almost undetectable (7.1% for C2 cells, n = 3; 5.4% ± 2% and 9.1% for C4 cells, n = 3; 5.1% ± 3.6%; Figure 3A). After staining with Rhod-2, the level of mitochondrial calcium following VP-16 treatment was also analyzed (Figure 3B). Exposure of 32D cells to VP-16 resulted in a significant increase in the mitochondrial Ca²⁺ concentration and 37.1% (n = 3; 34.8% ± 3.5%) of cells possessed mitochondria with elevated calcium. In C2 cells, mitochondrial, calcium-dependent changes were weaker (16.8%; n = 3; 17.5% ± 3.4%) and almost undetectable in C4 cells (4.6%). A strongly elevated mitochondrial calcium level is a known inducer of PTP opening, releasing proapoptotic factors from mitochondria to the cytosol. Because we detected a pronounced mitochondrial calcium overload in 32D cells, this suggested that a different pathway of cell-death initiation might occur in these cells. We then analyzed DNA nicking by TUNEL assay as an early indicator of apoptosis 14 hours after inducing cell death (Figure 3C). We found that 48.1% of 32D cells (n = 3; 39.8% ± 9.7%) had nicked DNA, whereas only 17.4% of C2 cells (n = 3; 15.8% ± 2.8%) were considered as TUNEL⁺. This experiment suggested that 32D cells may be able to activate an early calcium-dependent apoptotic response caused by substantial ER calcium release, whereas this is diminished in both the high and low Bcr-Abl–expressing cell lines.

Because mitochondrial Ca²⁺ elevation is a critical parameter in assessing a potential role for ER/mitochondrial coupling and apoptosis, we undertook a number of approaches to verify the specificity of the Rhod-2 probe for mitochondrial calcium (colocalization with MitoTracker Green, and use of a mitochondrial uncoupling agent to dissipate fluorescence; Figure S2) and also to verify the quantitative differences in mitochondrial Ca²⁺ accumulation identified by flow cytometry. TILL photonic imaging (Till Photonics, Gräfelfing, Germany) and quantification of calcium content in individual cells indicated that 32D cells exhibit significantly higher (almost 3-fold higher) mitochondrial calcium following drug treatment compared with Bcr-Abl–expressing cells (Figure S2B).

In addition, flow cytometric analysis indicated that 32D cells exhibit a rise in cytosolic calcium following drug treatment and that this is not a significant feature of Bcr-Abl–expressing cells (Figure 3), again possibly due to their lower releasable ER calcium. Comparison of calcium-dependent calpain activity in 32D and C2 cells provided further biochemical evidence for higher cytosolic calcium in 32D cells following drug treatment (Figure S3).

We have already shown that both C2 and C4 cells have lower releasable calcium in the ER; however, C2 cells still show minor alterations in cytosolic or mitochondrial calcium, which become much more significant as the duration of drug treatment is increased. Such calcium elevations are never detected in the high Bcr-Abl–expressing C4 cells. It is therefore possible that these calcium changes are a secondary, caspase-dependent event in C2 cells because they are drug sensitive and are therefore initiating at least one apoptotic pathway, whereas the C4 cells are drug resistant. We already know that the classic (caspase-dependent) pathway is defective in C4 cells (reported in our previous publication¹² ) but is active in C2 and 32D cells. We therefore examined whether the mitochondrial calcium accumulation in 32D and C2 cells is caspase dependent.

32D and C2 cells were pretreated with 50 μM zVAD, a potent inhibitor of multiple caspases. Cells were then treated with VP-16 for 18 hours and loaded with Rhod-2 and the mitochondrial calcium level was analyzed by flow cytometry. For this experiment the 18-hour time point was chosen to achieve appreciable Ca²⁺ accumulation in C2 cells (which is a slightly later event than in 32D), and examine whether or not this can be inhibited with zVAD. We found that 35% of 32D cells were positive for PI without zVAD pretreatment and 40% with pretreatment. C2 cells were 40% PI⁺ without zVAD and 14% with treatment (not shown; refer to combined data in Figure 4B). As Figure 4A shows, after 18 hours of VP-16 treatment, a population with strongly elevated mitochondrial calcium level was present in 32D cells and, to a lesser extent, in C2 cells. A total of 57% of 32D cells (n = 3; 52.6% ± 4.6%) possessed mitochondria overloaded with calcium. The caspase inhibitor had no effect on the mitochondrial uptake in 32D cells (61.6%), but significantly prevented the mitochondrial calcium overload observed in C2 cells from 42.8% (n = 3; 40.6% ± 5.2%) to 17.1% (n = 3; 15.1% ± 2.9%). These results confirm that the mitochondrial calcium uptake observed in 32D cells occurs independently of caspase activation, in contrast to C2 cells, where it is a secondary event and a consequence of caspase-dependent, downstream events. These data suggest that 32D cells may be able to activate a caspase-independent, calcium-dependent apoptotic pathway, in contrast to the Bcr-Abl–expressing cell lines C2 and C4 where this pathway is blocked.

Data using protease inhibitors also indicated that 32D cells could not be rescued from cell death by caspase inhibition, whereas C2 cells are protected (Figure 4B). Calpain inhibitors are also slightly protective for 32D cells, whereas they have no effect on C2 cells. These data further suggest 32D cells possess a distinct pathway for apoptosis initiation, which can still cause apoptosis when caspases are inhibited. Calpain activity may partially contribute to the death process because inhibitors marginally reduce cell death. In contrast, C2 cells depend on a pathway in which caspase initiation is a critical feature because they can be rescued by zVAD. C4 cells (high Bcr-Abl expression) do not initiate either pathway and are drug resistant.

To investigate directly if the ER/mitochondrial calcium-coupling process is an important proapoptotic event in VP-16–treated 32D cells, we studied the effect of an inhibitor of the mitochondrial Ca²⁺ uniporter RuRed on the cytosolic and mitochondrial calcium levels and sensitivity to apoptosis. Because this uniporter is the primary channel for the uptake of calcium into mitochondria, its inhibition would be expected to induce a rise in the cytosolic calcium levels and a decrease in mitochondrial calcium levels. Representative histograms obtained by flow cytometry are presented in Figure 5. We initially confirmed that RuRed at 30 μM for 18 hours had no toxic effects, as determined by PI viability assay (not shown). Because we already know that caspases are activated during VP-16–induced apoptosis, we preincubated all cells for 1 hour with 30 μM RuRed in the presence or absence of 50 μM zVAD to protect them from caspase-dependent effects, followed by VP-16 treatment. RuRed strongly influenced the cytosolic and mitochondrial calcium changes observed in 32D cells after 16 hours of VP-16 treatment (Figure 5A). In VP-16–treated 32D cells the amount of cells with elevated mitochondrial calcium decreased from 48.8% (n = 6; 45.9% ± 7.1%) to 24.9% (n = 3; 29.2% ± 4.4%) and 20.9% (n = 5; 26% ± 3.9%) after pretreatment with 30 μM RuRed and RuRed plus zVAD, respectively. In RuRed-treated cells calcium uptake by the mitochondria was inhibited, and therefore also resulted in an elevated cytosolic calcium level in comparison to that observed in cells treated with VP-16 alone. The percentage of cells characterized by an increase of cytosolic level above normal values rose from 38.4% (n = 3; 39.6% ± 4.6%) to 65% (n = 3; 55.3% ± 9.5%) after pretreatment with 30 μM RuRed and to 68.2% (n = 4: 56% ± 6.4%). In C2 cells after 16 hours of VP-16 treatment, major changes in cytosolic or mitochondrial calcium were not visible and RuRed did not influence calcium levels (Figure 5B).

We then investigated whether the protective effect of RuRed on mitochondrial overload increases the viability of 32D cells incubated with VP-16. Cell death/PI permeability was assessed at 3 time points to assess the influence of inhibitors on the kinetics of cell death (Figure 5C). These experiments indicated that RuRed significantly reduced cell death in 32D cells at 16 hours and, in fact, the kinetics of cell death then resembled C2 cells. A combination of RuRed plus zVAD further protected 32D cells, suggesting that the caspase pathway is also active as death proceeds. zVAD alone is, however, unable to protect 32D cells from apoptosis (as noted in Figure 3), demonstrating the predominance of a caspase-independent pathway. In contrast, in C2 cells RuRed is not protective, suggesting that ER/mitochondrial coupling is irrelevant, whereas zVAD is strongly protective (Figure 5D).

These data indicate that a calcium-dependent apoptotic pathway is the primary initiator of apoptosis in 32D cells. In contrast, in C2 cells, Bcr-Abl reduces the amount of releasable ER calcium and therefore this Ca²⁺/ER pathway does not play a role. In both cells, other caspase-dependent pathways can be activated, and the cells are drug sensitive. Together, these results demonstrate that calcium signals, which contribute to the apoptotic process activated in 32D cells by VP-16, are not activated in Bcr-Abl–expressing cells. Furthermore, our combined data suggest that the high-expressing cells (C4 cells) have the capacity to inhibit at least 2 apoptosis-initiating pathways (Ca²⁺/ER pathway and caspase dependent; Figure 6), and have therefore obtained drug resistance.

The data presented in this study suggest a novel function for the Bcr-Abl oncoprotein in the regulation of apoptotic pathways. We have demonstrated that a calcium-dependent, ER-initiated apoptotic pathway is inactive when Bcr-Abl is expressed and that this is likely to contribute to the prosurvival mechanisms associated with this oncogene.

We have detected significantly decreased ER calcium levels in both mouse and human Bcr-Abl–expressing cells and a weaker CCE response. This reduced CCE is analogous to the effect of other treatments that reduce ER Ca²⁺ levels, such as overexpression of Bcl-2 or prolonged culture in a low-calcium buffer.^24,28  It has been suggested that sufficient and probably sustained ER calcium release, supported by CCE, has to occur to induce a calcium overload and loss of mitochondrial membrane potential (MMP) in the mitochondria.²³  Because C2 and C4 cells released 2-fold less calcium from the ER, the corresponding CCE in these cells was also weaker. This reduction in both ER Ca²⁺ levels and CCE may severely compromise all calcium-dependent apoptotic signaling in Bcr-Abl–expressing cells. Imatinib treatment could restore ER calcium levels, indicating dependence on Bcr-Abl activity, whereas Bcl-2 down-regulation had no effect.

To assess the calcium-dependent apoptotic response, we treated cells with VP-16, a chemotherapeutic drug that has been reported to induce both classic BH3/mitochondria-dependent and ER calcium–dependent apoptotic pathways.²²  C4 cells displayed a highly drug-resistant phenotype, as previously described.¹⁶  Their ER calcium status was similar to C2 cells, but due to complete drug resistance (ie, more than one pathway blocked), they could not contribute further to the analysis. Subsequent studies compared 32D to C2 cells. After apoptosis induction, at early time points, when the percentage of cell death did not exceed 15%, cytosolic calcium changes were not observed in C2 cells and mitochondrial uptake was weak. In contrast, a strong elevation of cytosolic Ca²⁺ and mitochondrial Ca²⁺ uptake was evident in 32D cells.

Analysis of calpain activity in 32D and C2 cells further suggested that only 32D cells had sufficient cytosolic calcium for an early activation of these calcium-dependent proteases (Figure S3). In addition, differential effects of protease inhibitors indicated disparate death pathways in 32D and C2 cells. In 32D cells mitochondrial calcium uptake is an early event, part of a Ca²⁺-dependent signaling pathway, which is not inhibited by zVAD. In contrast, in Bcr-Abl–expressing C2 cells, it is a consequence of downstream events that are reversed by caspase inhibition. The source of the later caspase-dependent calcium rise in C2 cells may be (1) general loss of membrane integrity, to which caspases contribute, (2) caspase-dependent cleavage of a calcium channel in the plasma membrane,^33,34  or (3) caspase-mediated cleavage of Bap31 to Bap 20 leading to ER calcium release.³⁵ 

If ER/mitochondrial Ca²⁺ coupling is important for drug sensitivity, then direct inhibition of this pathway should also inhibit the zVAD-independent cell-death pathway in 32D cells. We therefore examined the effect of a mitochondrial Ca²⁺ uptake inhibitor, RuRed, on the apoptotic events induced by VP-16. RuRed has been previously shown to inhibit mitochondrial calcium uptake and protect from Ca²⁺-dependent cell death.²²  RuRed pretreatment protected 32D cells from mitochondrial calcium uptake and diverted calcium to the cytosol. This significantly reduced apoptosis in these cells (at 16 hours) until other proteolytic enzymes became active at 18 hours. Addition of zVAD inhibited the caspase-dependent pathway and reduced apoptosis at 18 hours compared to RuRed only. zVAD has less effect on 32D cells than on C2 cells because other enzymes such as calpains will also contribute to cell death as shown in Figure 4 and Figure S3. These experiments confirmed that ER/mitochondria coupling is a critical part of VP-16–induced apoptosis in 32D cells, whereas it does not play a role in Bcr-Abl–expressing cells.

To our knowledge this is the first report that has shown a link between the Bcr-Abl oncoprotein and ER Ca²⁺ homeostasis. The mechanism by which Bcr-Abl reduces ER Ca²⁺ levels is unknown. ER calcium levels are strongly influenced by Bcl-2 family members, although the mechanism by which this is achieved is unknown, and the literature still contains some conflicting data.³⁶  Studies have demonstrated that Bcl-2 overexpression reduces the steady-state ER Ca²⁺ concentration and limits mitochondrial calcium uptake.^37,38  Depletion of Bcl-2 in Bcr-Abl–expressing cells did not, however, increase ER calcium levels, suggesting that it is not responsible for the decrease in ER releasable calcium. It remains possible that other ER-localized Bcl-2 family members contribute to this effect. Bax and Bak promote Ca²⁺ uptake into the mitochondria,^21,39  and knockouts are resistant to a wide range of apoptotic stimuli, exhibit decreased ER calcium, and ineffective calcium delivery to the mitochondria.²² 

It is also possible that Bcr-Abl exerts a direct effect at the ER. This protein is a cytosolic tyrosine kinase; however, c-Abl has been reported to also be present in the ER⁴⁰  and to be associated with a proapoptotic ER protein Aph-2.⁴¹  It is conceivable, therefore, that a portion of Bcr-Abl is also present at the ER and exerts a direct influence on ER homeostasis. Bcr-Abl expression can, however, influence various signaling pathways, including Ras and phosphatidylinositol 3-kinase. Of interest in this regard is a study indicating that the Ras-related G protein (R-Ras) alters Ca²⁺ homeostasis by increasing the calcium leak across the ER membrane.⁴²  It is also possible that downstream signaling might influence existing channels that regulate calcium release, for example, the inositol-1,4,5-triphosphate receptor. In addition, ER calcium homeostasis can also be influenced by the cellular redox state and Bcr-Abl has been reported to increase ROS production.⁴³ 

We propose that the drug resistance observed during the progression of CML is a consequence of the acquisition of various antiapoptotic functions of Bcr-Abl (Figure 6). Low and high Bcr-Abl–expressing cells show impaired ER homeostasis and are unable to activate ER calcium-mediated apoptotic pathways. However, low Bcr-Abl–expressing (C2) cells can still activate a classic BH3/mitochondrial pathway leading to caspase activation. In high Bcr-Abl–expressing, drug-resistant cells (C4) the classic BH3/mitochondrial pathway is also inhibited, as indicated by several previous studies including our own.^12,44,45 

There is still uncertainty regarding the molecular mechanisms of CML progression to blast crisis and the nature of drug resistance mechanisms occurring at this stage. This study provides a new insight into the repertoire of survival mechanisms that can be used by Bcr-Abl. It is clear that if we intend to combat drug resistance with combination regimens, we need a better understanding of available apoptosis pathways in cancer cells and classes of drugs that can adequately target them.

We thank Dr Karen Keeshan for early contributions to this study and Dr John Mackrill for valuable advice. We also thank mgr Bogusz Kulawiak for the help with TILL photonics. We are grateful to Lisa O'Connor for help with the siRNA transfection experiments. Imatinib was kindly provided by Novartis (Basel, Switzerland).