DT40 cells lacking the Ca²⁺-binding protein annexin 5 are resistant to Ca²⁺-dependent apoptosis

Abstract

Annexins are widely expressed Ca²⁺-dependent phospholipid-binding proteins with poorly understood physiological roles. Proposed functions include Ca²⁺ channel activity and vesicle trafficking, but neither have been proven in vivo. Here we used targeted gene disruption to generate B-lymphocytes lacking annexin 5 (Anx5) expression and show that this results in reduced susceptibility to a range of apoptotic stimuli. By comparison B-lymphocytes lacking annexin 2 (Anx2) showed no such resistance, providing evidence that this effect is specific to loss of Anx5. The defect in the ANX5^−/− cells occurs early in the apoptotic program before nuclear condensation, caspase 3 activation, and cell shrinkage, but downstream of an initial Ca²⁺ influx. Only UVA/B irradiation induced similar levels of apoptosis in wild-type and ANX5^−/− cells. Unexpectedly, ANX5^−/− cells permeabilized in vitro also failed to release mitochondrial cytochrome C, suggesting a possible mechanism for their resistance to apoptosis. These findings demonstrate a role for Anx5 in determining the susceptibility of B-lymphocytes to apoptosis.

Members of the annexin (Anx) family of Ca²⁺- and phospholipid-binding proteins have been found in all eukaryotes examined to date except Saccharomyces cerevisiae. The vertebrate annexin gene family includes at least 12 family members that by cladistic analysis can be sorted into clusters, such as annexins 1, 2, and 3, and annexins 5 and 6 (1, 2). Annexins are evolutionarily ancient and have undergone extensive diversification during vertebrate evolution, suggesting continuing important cellular roles, although their functions are not clearly defined. Both Anx2 and Anx5 are widely expressed, although Anx5 expression is particularly prominent in skeletal, cardiac, and smooth muscle, Leydig cells, endothelia, chondrocytes, and some neurons (3–8). Like many annexins, Anx2 and Anx5 are predominantly intracellular, existing both as soluble cytosolic proteins and also in Ca²⁺-dependent and Ca²⁺-independent association with a range of intracellular membranes (9–12), including the plasma membrane and early endosomes (13, 14). However, only Anx5 has been associated with late endosomes and mitochondria (15).

There is now considerable evidence that the Ca²⁺-independent binding of Anx2 to early endosomes is mediated through an interaction that requires part of the Anx2 N terminus and membrane cholesterol (11, 12), and that this association may be important in vesicle fusion (13) and maintenance of endosomal structures (16). Anx2 and Anx5 both exhibit Ca²⁺-channel activity in vitro (17), which could be of significance in matrix vesicle Ca²⁺-uptake during bone mineralization in vivo (18–20). The Ca²⁺-channel activity of annexins has been attributed to a central hydrophilic pore identified in the crystal structure (21–23), although the lack of membrane-spanning domains raises questions as to whether annexins function as Ca²⁺ channels in cells. An alternative membrane-embedded form for members of the annexin family was reported using synthetic lipid bilayers and purified proteins (24). Membrane insertion by annexins was shown to depend on acidification, although mild peroxidation at physiological pH had a similar effect on Anx5 (25). This study also showed that Anx5 is essential for a peroxide-mediated Ca²⁺ influx in B-cells, adding weight to the idea that Anx5 either functions as a Ca²⁺ channel or at least acts in a signaling pathway on which Ca²⁺ influx depends.

Fluorescently labeled purified Anx5 has also been developed as a biochemical tool for the detection of apoptotic cells (26); like other annexins, it binds in a Ca²⁺-dependent manner to negatively charged phospholipids (27) that are exposed on the cell surface during apoptosis. This diagnostic application of exogenous extracellular Anx5 is, however, unlikely to reflect the in vivo role of endogenous Anx5, a mostly intracellular protein. To gain insight into the function of Anx5, we generated B-cells in which Anx5 expression is ablated through targeted gene disruption. We now report that cells lacking Anx5 are resistant to a range of agonists that induce apoptosis via a Ca²⁺-dependent pathway. In contrast, Ca²⁺-independent apoptosis induced by UV light occurs normally in Anx5^−/− cells. We show that the requirement for Anx5 is downstream of an initial Ca²⁺ influx, but upstream of early events such as cell shrinkage and caspase 3 activation. Our observations suggest a role for Anx5 in the control of apoptosis upstream of both mitochondrial membrane disruption and caspase activation.

Materials and Methods

Cell Culture, Transfection, and Selection.

The DT40 chick preB cell-line was a gift from Jean-Marie Buerstedde, Basel Institute for Immunology, Basel, Switzerland. Cells were cultured in DMEM containing 10% FCS, 1% chicken serum (CS), penicillin (42 units/ml), streptomycin (42 mg/ml), and glutamine (1.7 mM) at 40°C in humidified incubators with 5% CO₂, at densities between 5 and 100 × 10⁴ cells per ml. Where required, DT40 conditioned medium was harvested when cells were in log-phase growth at 80 × 10⁴ per ml. All media and supplements were from GIBCO. For transfection, 10⁷ cells in logarithmic phase growth were washed once in PBS, and resuspended in 400 μl of ice-cold PBS with 25 μg of linearized DNA (in PBS) in a prechilled 0.4-cm electroporation cuvette. After 10 min on ice, samples were electroporated (950 V, 25 μF, and ∞ Ω), left on ice for a further 5 min, and diluted into 15 ml of warm media containing 20% conditioned medium. For selection of stable clones, 0.5 ml per well of 1× transfection mix and a 4× dilution (in 20% conditioned medium) were plated into the central 24 wells of 48-well, flat-bottomed tissue culture plates (outer wells were filled with PBS). After 24 h, 0.5 ml of medium containing 2× selection drugs was added to each well. Plates were wrapped in damp tissue paper with a wick to a reservoir containing a dilute solution of copper sulfate (to maintain sterility), to ensure 100% humidity and reduce loss by evaporation during incubation. After 10–14 days of undisturbed incubation, loosely clumped colonies of about 2 mm in diameter were pipetted in 100-μl volume into 5 ml of 20% conditioned medium containing selection drugs for a further 2 days undisturbed incubation before further culture and screening.

Generation of Targeting Constructs.

DT40 genomic DNA was extracted by a salting out procedure (28). PCR was performed according to the manufacturer's instructions with the Expand Long Template PCR system (Boehringer Mannheim). Neomycin (Neo), puromycin (Puro), and HisD resistance cassettes containing the β-actin promoter (1.3 kb) upstream of the relevant selection marker, followed by the thymidine kinase polyA sequence, were gifts from Jean-Marie Buerstedde. For the basic ANX2 construct, first-round PCRs with primers CAII-1 (5′-TTAAGGCTTACTCAAACTTTGATGCTGAC-3′) and CAII-2 (5′-ATTGCTGCGGTTTAAACAGGATGTTGATGATGGTAACTTC-3′) or CAII-3 (5′-TCATCAACATCCTGTTTAAACCGCAGCAATGAACAGAG-3′) and CAII-4 (5′-ATGGTGTCTTCAGCAAGCCCAAGATCACT-3′), respectively, were performed on genomic DNA. The 3.7-kb XbaI fragment of the strand-overlap PCR with primers CAII-1 and CAII-4, using a mixture of 1 in 250 dilutions of these first-round reactions as template was then cloned into the XbaI site of pBluescript SK (Stratagene). The basic ANX5 construct was generated by directional cloning of the XmnI/EcoRI 2.0-kb fragment of a genomic PCR product produced with primers CAV-3 (5′-TCCAGCCCCGTTTAAACCCGGCGAACCGCGGAG-3′) and CAV-4 (5′-TTCGTGGCCTTGCGAAGGGCTTCTGCATCA-3′) into SmaI/EcoRI cut pBluescript SK, followed by cloning of the 1.3-kb EcoRI fragment of a genomic PCR product produced with primers CAV-1 (5′-GTTTGAATTCCCTTTGGCTGTGCTCACACC-3′) and CAV-2 (5′-CGGTTCGCCGGGTTTAAACGGGGCTGGAGTCGAATCCGAG-3′) that had been cloned into the EcoRI site of the pTAg vector (Invitrogen). To generate the final targeting constructs, the HincII/EcoRV excised 2.3-kb Neo cassette, or the XhoI/EcoRV 2.4-kb Puro cassette were then blunt-end cloned into the resulting PmeI sites inserted into exons 4 or 2, encoded in the ANX2 CAII-2/CAII-3 primer overlap or the ANX5 CAV-2 primer, respectively. The Neo and Puro constructs were linearized for transfection with XbaI or BamHI for ANX2 and ANX5, respectively.

Western Blotting.

Ten million cells were washed in PBS and lysed in 1× SDS/PAGE sample buffer. A volume equivalent to 10⁶ cells was loaded per track on a 10% polyacrylamide gel for electrophoresis and blotting (29). A rabbit α-chick Anx2 polyclonal antibody (gift from Brian Genge, University of South Carolina, Columbia) and a rabbit α-chick Anx5 polyclonal antibody (gift from John Walker, University of Leeds, Leeds, U.K.) were used as primary antibodies, both at 1:1,000. Goat α-rabbit IgG conjugated to alkaline phosphatase (AP) or horseradish peroxidase (HRP; Promega) were used at 1:10,000 as secondary antibodies and were developed with Western Blue substrate (Promega) or by enhanced chemiluminescence (ECL), respectively. For cytochrome C blots, samples equivalent to 4 × 10⁶ cells were removed from the experimental chamber (see later) and separated by centrifugation at 10,000 × g for 5 min at 4°C into mitochondria-containing pellet and cytosolic supernatant. A quarter of each sample equivalent to 10⁶ cells was run per track on a 12% polyacrylamide gel. Anti-cytochrome C monoclonal (PharMingen clone 6H2.B4) and anti-Cox IV monoclonal (Molecular Probes) antibodies were used at 1:1,000 and 1:10,000, respectively, with α-mouse IgG HRP as a secondary at 1:10,000. The mitochondrial protein Cox IV was used as a control to verify equal protein loadings, and to demonstrate the absence of mitochondria from supernatants (data not shown; ref. 30).

Induction of Apoptosis and Fluorescence-Activated Cell Sorter (FACS) Analysis of Nuclei.

Experiments were performed in U-shaped 96-well plates (Camlab) in 200 μl at 5 × 10⁵ cells per ml of full medium. Chemicals were purchased from Calbiochem. UVA/B treatment was for the stated time through the base of the wells on a LKB 2011 Macrovue transilluminator with a peak output at 302 nm of 8 J/m². Inhibitors of apoptosis were titrated on chelerythrine- or staurosporine-treated and untreated cells over 7 h to identify the most inhibitory concentration that produced no apoptosis on its own. Samples were pelleted and nuclei were stained with propidium iodide and analyzed by FACS (31). All samples were in duplicate or triplicate; error bars in Figs. 2 A and B and 3 A–D represent the standard deviation from the mean.

Figure 2.

Loss of Anx5 inhibits apoptosis. (A) Percentage of apoptotic nuclei in wild-type and ANX5^−/− cells 7 h after treatment with 20 s UVA/B irradiation (UVA/B), 4 μM chelerythrine (Chel), 100 nM staurosporine (Stauro), 10 μM 1-pyrrolidine-carbodithioic acid (Pyr), and 1 mM hydrogen peroxide (H₂O₂). (B) Percentage of apoptotic nuclei 7 h after treatment with 20 s UVA/B irradiation or 100 nM staurosporine (Stauro) in wild-type ANX2^−/− cells and two independent ANX5^−/− clones. Caspase 3 activation time course in wild-type and ANX5^−/− cells after induction of apoptosis with 100 nM staurosporine (C) or 2 μM chelerythrine (D). Caspase activities in lysates from untreated cells were subtracted at each time point. These values never exceeded 20% of those obtained in treated cells. (E) Cell size (FSC-H) of wild-type and ANX5^−/− Calcein-positive, propidium iodide-negative live intact cells 0, 2, or 4 h after treatment with 200 nM staurosporine. Percentage gated as shrunk indicated on each panel. Results are representative of three sets of experiments.

Caspase 3 Assay.

One million cells in 2 ml of full medium were treated as above in duplicate, lysed in a 100-μl volume and 20 μl was assayed using the CaspACE Assay System (Promega). Because cell extracts were dilute, accurate protein assays were not possible, and samples were standardized by cell number. Results, given as pMol AMC released per min, were background subtracted. Error bars in Fig. 2 C and D represent the standard deviation from the mean.

Cell Shrinkage FACS Analysis.

Cells (5 × 10⁵) were treated as above in 1 ml of full medium. 30 min before the end of the treatment, Calcein-AM (Molecular Probes) was added to a final concentration of 500 nM. Propidium iodide (PI) was then added to 1 μg/ml, and the samples were analyzed by FACS. Intact live cells were identified and gated as Calcein positive and PI negative, and their size was displayed on a forward scatter (FSC-H) histogram.

Cytosolic Ca²⁺ Measurements.

Ten million cells were loaded for 45 min at 40°C with 1.25 μM fura-2/AM and 0.0025% Pluronic F-127 (both from Calbiochem, combined 2,000× stock in DMSO) in 4 ml of growth medium. Washes and final resuspension were in supplemented ice-cold Hanks' balanced salt solution (HBSS; 1.26 mM CaCl₂/0.81 mM MgSO₄/5.4 mM KCl/0.44 mM KH₂PO₄/136.9 mM NaCl/0.34 mM Na₂HPO₄/4.2 mM NaHCO₃/5.5 mM d-glucose/10 mM Hepes). Cells were washed three times at 4°C and resuspended in ice-cold HBSS to a density of 10⁶ cells per ml and kept on ice in 2.5-ml aliquots. Samples were then prewarmed to 40°C for 4 min in a water bath. Fluorescence of the stirred cell suspension was measured ratiometrically by emission at 510 nm and excitation at 340 nm and 380 nm, respectively, using an LS 50 B fluorimeter (Perkin–Elmer). Background drift was corrected by subtraction of typical linear drift traces obtained in separate experiments.

Permeabilized Cell Mitochondrial Assay.

Cells (to a final density 10⁷ cells per ml) were washed once in ice cold HBSS and added to intracellular KCl buffer (120 mM KCl/20 mM Hepes/3 mM K₂HPO₄/10 mM NaHCO₃/10 mM dl-3-hydroxybutyrate, treated with Chelex resin from Sigma, pH 7.4, MgCl₂ added to 0.5 mM) at 40°C containing 10 nM Calcium green 5N dye (CG5N, Molecular Probes). The stirred cell suspensions were permeabilized by 4 μM digitonin at t = 0 s, a concentration insufficient to damage the mitochondrial membranes (32), and reached equilibrium after approximately 200 s. Mitochondrial swelling was assessed by light scatter (ex 550 nM : em 566 nM), while extra-mitochondrial free Ca²⁺ was estimated at the same time fluorimetrically with CG5N (ex 506 nM : em 532 nM), using an LS 50 B fluorimeter.

Results and Discussion

Targeted Disruption of the Annexin 2 and 5 Genes in DT40 Cells.

To investigate the functions of Anx2 and Anx5 in vivo, we used the DT40 chicken preB cell line, which expresses high levels of recombinases and is thus amenable to targeted gene disruption (33). Annexins 1, 2, 5, and 6 were shown by Western blotting to be expressed in the DT40 cell line (data not shown). Using available chicken cDNA sequence data and known intron/exon boundary sites from other species (34, 35), homologous targeting constructs were generated by genomic PCR, encoding disruptions in coding exons 4 or 2 for ANX2 and ANX5, respectively (Fig. 1 A and B). First-round Neo-resistant clones were selected and screened by genomic PCR (not shown). Independent first-round targeted clones were transfected with the respective Puro-encoding targeting constructs. After Puro selection, second-round clones targeted at both alleles were identified by genomic PCR (not shown) and Western blotting as being negative for Anx2 or Anx5 expression (Fig. 1 C and D). Loss of Anx5 was also confirmed by Southern blotting (data not shown). At each stage of selection, ratios of targeted to random integration greater than 50% were achieved, as previously described with this cell line (33). During each subsequent experiment, results were confirmed in at least two ANX5^−/− clones derived from independent first-round clones to ensure that any phenotype was due to loss of Anx5, and not to a secondary mutation carried through the selection process. Furthermore, in this study ANX2^−/− clones are used as a negative control to show that observed phenotypes are specific to Anx5 and are neither a simple “annexin” effect, nor caused by the selection process or expression of the Neo or Puro selection markers.

Figure 1.

Gene targeting at the chick ANX2 and ANX5 loci. Structure of the targeting constructs, and wild-type, Neo-disrupted, and Puro-disrupted alleles for chick annexin 2 (A) and annexin 5 (B). Open boxes represent exons, heavy lines represent regions in the targeting constructs, and thick arrows show the location and orientation of the selection cassettes. ATG indicates the start methionine codon for Anx5. Restriction sites: Xb, XbaI; E, EcoRI. Western blots of whole-cell lysates of wild-type (+/+) and two independent doubly targeted clones (−/−) for Anx2 developed with Western Blue (C), and Anx5 developed with enhanced chemiluminescence (D). Bands were sized according to GIBCO prestained molecular weight markers (not shown). Background bands of higher molecular weight demonstrate equivalent sample loadings.

To identify suitable concentrations for additional selection drugs for generating further stable transfectants, wild-type, ANX5^−/−, and ANX2^−/− DT40 cells were treated with a range of concentrations of l-histidinol and hygromycin. Unexpectedly, the minimum lethal concentration of l-histidinol was 0.25–0.5 mg/ml for wild-type and ANX2^−/− cells, but was 1 mg/ml for ANX5^−/− cells. Furthermore, for the slower acting drug hygromycin, 0.25 mg/ml killed wild-type and ANX2^−/− cells, whereas ANX5^−/− cells were resistant even at the highest concentrations tested (1.5 mg/ml). Thus, loss of Anx5 leads to increased resistance to l-histidinol and hygromycin.

Loss of Anx5 Confers Resistance to Apoptosis.

Because hygromycin is thought to kill cells by apoptosis (36), these empirical observations suggested that disruption of the ANX5 gene could lead to resistance to apoptosis. A range of faster acting and better characterized apoptotic agents was thus tested on these cells. The level of apoptosis was taken as the percentage over background levels of condensed nuclei present after 7 h of treatment, as determined by flow cytometric analysis of propidium iodide-stained nuclei. Doses of apoptotic agents were titrated so as to cause 20–60% apoptosis in wild-type cells over 7 h. Within these concentration ranges, ANX5^−/− cells showed almost complete resistance to chelerythrine, staurosporine, 1-pyrrolidine-carbodithioic acid, and hydrogen peroxide (H₂O₂), but not to UVA/B irradiation (Fig. 2A). Similar experiments conducted over 48 h showed that resistance to apoptosis in the ANX5^−/− cells was sustained rather than merely delayed (see the supporting information, which is published on the PNAS web site, www.pnas.org). Resistance to agents such as staurosporine, but not to UVA/B, was shared by independent ANX5^−/− clones, but not ANX2^−/− cells (Fig. 2B). ANX5^−/− clones also showed reduced caspase 3 activation, an earlier marker of apoptosis than nuclear condensation (37), compared with wild-type cells in response to staurosporine (Fig. 2C) or chelerythrine (Fig. 2D). We next investigated cell shrinkage, because this has recently been demonstrated to be one of the earliest detectable events in the apoptotic program (38). Fig. 2E shows that even this very early event, detectable after 2 h, and clearly in advance of significant caspase 3 activation (Fig. 2C, and also confirmed by flow cytometry at an individual cell level, using PhiPhiLux-G1D1 following Alexis Corporation's instructions; data not shown), fails to occur in ANX5^−/− cells. Collectively, these results show that loss of Anx5 reduces the susceptibility of DT40 cells to a specific subset of apoptotic agonists, possibly through functional modulation of an early step in the apoptotic program.

We next sought to define a common element that might account for resistance to the various drugs tested, without increasing resistance to UVA/B. Resistance could be due to altered expression of other proteins involved in the apoptotic pathway. The expression of such proteins for which chick cross-reactive antibodies were available (Transduction Labs Apoptosis Sampler Kits) was thus examined by Western blotting. Expression of Bcl-2, Bad, Bax, Bcl-x, Fas-L, and CAS were all unaffected by loss of Anx5; for several other proteins such as Fas and Mcl-1 expression levels may have been below the level of detection by enhanced chemiluminescence (see the supporting information for full details). These observations argue for a direct role for Anx5 in drug-induced apoptosis, but not in UV-induced apoptosis.

Anx5 and Ca²⁺ in the Apoptotic Program.

Ca²⁺ is known to be required during apoptosis in response to a range of stimuli, and is able to cause apoptosis itself under conditions of Ca²⁺-overload, although its mechanism of action is unclear (37). Anx5 is a Ca²⁺-binding protein and shows Ca²⁺-channel activity in vitro (17), and recently the cardioprotective benzothiazepine derivative K201, proposed to have Anx5 as its intracellular target (39), has been shown to inhibit the apoptotic cellular damage that accompanies Ca²⁺-overload in the heart (40). The role of Ca²⁺ in Anx5-dependent apoptosis in our system was thus investigated. Cells were treated with the intra- and extracellular Ca²⁺-chelators BAPTA-AM (1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetate–acetoxymethyl ester) or EGTA, respectively, or the Ca²⁺-channel blockers SKF96365, nicardipine, or diltiazem, from the point of induction of apoptosis. All of these agents, except diltiazem, were found to significantly inhibit staurosporine-induced apoptosis (Fig. 3A) in control DT40 cells. However, as with loss of Anx5, none of these agents had any inhibitory effect on UV-induced apoptosis (Fig. 3B). The time course of the requirement for intracellular Ca²⁺ during apoptosis was also investigated by BAPTA-AM treatment at progressively later points after induction of apoptosis. For staurosporine, BAPTA-AM treatment was still partially effective even at 120 min after induction (Fig. 3C), whereas for chelerythrine, BAPTA-AM treatment had very little effect unless applied either before or within the first 30 min of induction of apoptosis (Fig. 3D). The time course of inhibition obtained with nicardipine on 1-pyrrolidine-carbodithioic acid-treated cells (data not shown) was similar to that observed with BAPTA-AM on chelerythrine-treated cells. Consistent with these data, BAPTA-AM treatment also blocked cell shrinkage, one of the earliest markers of apoptosis (Fig. 3E).

Figure 3.

Ca²⁺ requirement of drug-induced apoptosis. Percentage of apoptotic nuclei 7 h after treatment with 200 nM staurosporine (A) or 20 s UVA/B irradiation (B) in wild-type DT40 cells. Samples were treated in addition with 3 μM BAPTA-AM, 2.5 mM EGTA, 25 μM SKF96365 (SKF), 10 μM nicardipine (Nic), 30 μM diltiazem (Dilt), or with no blocker (Control). (C and D) Percentage of apoptotic nuclei 7 h after treatment with 100 nM staurosporine (C) or 4 μM chelerythrine (D). Three micromoles BAPTA-AM was added at the indicated time points in minutes before or after induction of apoptosis. No BAPTA-AM was added to control samples. The Ca²⁺ responses to staurosporine of control and BAPTA-AM loaded wild-type cells are shown in the supporting information. Fluorescence ratio is plotted against time (0–600 s), and each trace shows an average of ten separate experiments. (E) Cell size (FSC-H) of wild-type Calcein-positive, propidium iodide-negative live intact cells 3 h after treatment with 200 nM staurosporine with or without 3 μM BAPTA-AM. The percentage gated as shrunk indicated on each panel. Results are representative of three sets of experiments. (F) Average population of intracellular free Ca²⁺ concentrations measured fluorimetrically with fura-2 in wild-type and ANX5^−/− cells in the presence of extracellular Ca²⁺. Staurosporine (200 nM) was added at 100 s.

Intracellular Ca²⁺ is thus required at an early stage in the induction of apoptosis by a range of agents, except UVA/B illumination, in control DT40 cells. The requirement for Ca²⁺ as a possible apoptotic initiator therefore mirrors the requirement for Anx5, and is suggestive of a functional link between the two. Given the proposed role for Anx5 as a Ca²⁺ channel, and that several apoptotic agents such as staurosporine have been shown to cause rises in cytosolic free Ca²⁺ (41), could there be a defect in these signals in ANX5^−/− cells? To address this question changes in cytosolic free Ca²⁺ in response to staurosporine were measured fluorimetrically in wild-type and ANX5^−/− cells loaded with fura-2/AM. A small but sustained rise in cytosolic free Ca²⁺ was observed over an 8-min time course in both control and ANX5^−/− cells (Fig. 3F), ruling out a possible role for Anx5 in the mediation of this signal. Chelerythrine-induced Ca²⁺ rises could not be analyzed in this way because of interference of chelerythrine with the fura-2 signal. Thus if this Ca²⁺ rise is required for the induction of apoptosis, the defect in ANX5^−/− cells lies downstream of this signal, but upstream of cell shrinkage.

Mitochondrial Retention of Cytochrome C in ANX5^−/− Cells.

An alternative hypothesis to a defect in cytosolic Ca²⁺ signaling could be that mitochondrial responses to these signals, such as cytochrome C release, leading to initiation of the apoptotic program are altered in ANX5^−/− cells. To investigate this we developed a permeabilized cell assay based on published techniques (42–44) to analyze the Ca²⁺-induced mitochondrial permeability transition (MPT) and associated cytochrome C release. Fig. 4A shows that wild-type cell mitochondria respond to a bolus of Ca²⁺ by an initial uptake seen as a transient drop from a peak value. This is followed by rapid decrease in light scatter indicating swelling (42) and re-release of Ca²⁺, showing that the mitochondria have undergone the MPT. Cyclosporin A, a known blocker of the MPT, did indeed abolish this Ca²⁺-induced swelling (Fig. 4B). Instead, under these conditions the extramitochondrial free Ca²⁺ returned to near basal levels as the bolus of Ca²⁺ was almost completely taken up, and was accompanied by a concomitant increase in light scatter most likely due to the formation of Ca²⁺ crystal complexes in the inner mitochondrial space (Fig. 4B; ref. 42).

Figure 4.

Susceptibility of mitochondria in permeabilized cells to Ca²⁺-induced pore transition. (A–C) Black traces indicate mitochondrial swelling, whereas blue traces indicate extramitochondrial free Ca²⁺. A 100 μM bolus of Ca²⁺ was added at 300 s (arrow) in each experiment, giving an instantaneous Ca²⁺ rise. (A) Wild-type DT40 cells. (B) Wild-type cells with 5 μM Cyclosporin A (CsA) present from t = 0 s. (C) ANX5^−/− cells. (D) Samples taken from experiments A–C and similar experiments at the given time points were separated into mitochondria-containing pellets and western blotted for cytochrome C and Cox IV. (E) As in D, but cytosolic supernatants are also shown (S/N) and Cox IV is not shown. Data are representative of more than two experiments in each case.

Using this method we were then able to show that in fact ANX5^−/− cells responded in the same way as wild-type cells with uptake and rapid re-release of Ca²⁺ accompanied by swelling (Fig. 4C), indicating that the Ca²⁺-induced MPT is not affected by loss of Anx5. However, we also investigated the release of cytochrome C from mitochondria by removing aliquots at various time points during these experiments for fractionation and Western blotting. Unexpectedly, we found that cytochrome C release, indicated by disappearance of the immunoreactive protein band from the mitochondrial pellets, occurred in wild-type cells independently of the MPT (Fig. 4D Upper); cytochrome C was lost from the mitochondria 15 min after starting the experiment both in the presence and absence of Ca²⁺ (added at 5 min). In addition, we found that ANX5^−/− cells failed to release cytochrome C under both these conditions (Fig. 4D Lower). For both wild-type and ANX5^−/− cells, Cox IV was retained in the pellets, demonstrating the presence and integrity of mitochondria. Fig. 4E shows that the cytochrome C released from the mitochondrial pellets is indeed found in the cytosolic supernatant after 15 min for wild-type cells whether or not the MPT has been induced by Ca²⁺. Furthermore, inhibition of the MPT with cyclosporin A failed to inhibit the release of cytochrome C, whereas no release was observed in ANX5^−/− cells (Fig. 4E). These results confirm that under these experimental conditions cytochrome C release is independent of the MPT and occurs constitutively in wild-type cells but not ANX5^−/− cells, regardless of the level of free Ca²⁺.

In summary, these results suggest that Anx5 and intracellular Ca²⁺ are both involved early in the apoptotic program induced by a range of drugs, but not by UVA/B illumination. ANX2^−/− cells were not resistant to apoptotic stimuli, and indeed DT40 lines disrupted for the tyrosine kinases lyn, syk, or Btk, or the PLCγ2 or IP3R genes involved in Ca²⁺ signaling, also show no resistance to the Ca²⁺-dependent apoptotic stimulus, Calphostin C (45), again indicating that the phenotype reported here is a specific effect. Our results indicate that the requirement for Anx5 and intracellular Ca²⁺ appears to be downstream of changes in cytosolic free Ca²⁺, but upstream of cell shrinkage, caspase 3 activation, and nuclear condensation. Moreover, although the Ca²⁺-induced MPT may occur normally in ANX5^−/− cells, cytochrome C release can occur independently of the MPT in these cells, as has been previously reported for other cell types (30, 42). Thus it may be that cytochrome C release induced in an MPT-independent manner, although potentially still Ca²⁺-dependently, is defective in ANX5^−/− cells and leads to the observed defects in apoptosis. In conclusion, our data show that Anx5 has a role in determining apoptotic susceptibility in B cells, possibly by functioning as a Ca²⁺ mediator in the early phase of apoptosis.

Abbreviations

Anx

annexin

Neo

neomycin

Puro

puromycin

MPT

mitochondrial permeability transition

BAPTA-AM

1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetate–acetoxymethyl ester