Fas-induced B Cell Apoptosis Requires an Increase in Free Cytosolic Magnesium as an Early Event*

Ligation of the Fas molecule expressed on the surface of a cell initiates multiple signaling pathways that result in the apoptotic death of that cell. We have examined Mg2+ mobilization as well as Ca2+mobilization in B cells undergoing Fas-initiated apoptosis. Our results indicate that cytosolic levels of free (non-complexed) Mg2+([Mg2+] i ) and Ca2+([Ca2+] i ) increase in cells undergoing apoptosis. Furthermore, the percentages of cells mobilizing Mg2+, fragmenting DNA, or externalizing phosphatidylserine (PS) increase in parallel as the concentration of anti-Fas monoclonal antibody is raised. Kinetic analysis suggests that Mg2+ mobilization is an early event in apoptosis, clearly preceding DNA fragmentation and probably occurring prior to externalization of PS as well. The source of Mg2+ that produces the increases in [Mg2+] i is intracellular and most likely is the mitochondria. Extended pretreatment of B cells with carbonyl cyanidem-chlorophenylhydrazone, an inhibitor of mitochondrial oxidative phosphorylation, produces proportional decreases in the percentage of cells mobilizing Mg2+, fragmenting DNA, and externalizing PS in response to anti-Fas monoclonal antibody treatment. These observations are consistent with the hypothesis that elevated [Mg2+] i is required for apoptosis. Furthermore, we propose that the increases in [Mg2+] i function not only as cofactors for Mg2+-dependent endonucleases, but also to facilitate the release of cytochromec from the mitochondria, which drives many of the post-mitochondrial, caspase-mediated events in apoptotic cells.

Signal transduction by cell-surface receptors is critical for the regulation of cell growth and differentiation, but may serve a variety of other functions, including induction of programmed cell death or apoptosis. The cell-surface receptor Fas (CD95), which belongs to the tumor necrosis factor/nerve growth factor receptor family, is the most extensively studied of the so-called death receptors (1,2). Cross-linking Fas molecules on the cell surface either by Fas ligand or by anti-Fas monoclonal antibody (mAb) 1 results in the apoptotic death of the cell. There have been a number of studies of the phenotypic changes that occur early in apoptosis. Among those events, the disruption of the mitochondrial transmembrane potential (⌬ m ) (3,4) and the aberrant exposure of phosphatidylserine (PS) have been studied extensively (5,6). It has also been reported that a sustained increase in intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) can trigger lethal processes (7) and that increased intracellular Mg 2ϩ concentration ([Mg 2ϩ ] i ) promotes apoptosis in rat hepatocytes (8). Most recently, Eskes et al. (9) demonstrated that the release of cytochrome c from mitochondria (induced by overexpression of the apoptosis-promoting molecule Bax) is independent of ⌬ m , but dependent on Mg 2ϩ . Although Mg 2ϩ has often been considered a messenger in signal transduction because of its ability to regulate a variety of cellular processes, changes in [Mg 2ϩ ] i following induction of apoptosis via receptor ligation in lymphocytes have not been demonstrated.
The susceptibility of primary B cells to Fas-mediated apoptosis is dependent on whether the cell has received signals in the absence of antigen through CD40, another member of the tumor necrosis factor receptor superfamily. Resting B cells express very low levels of Fas and are not prone to apoptosis induced by this receptor (10,11). Previous investigators (12)(13)(14)(15) have shown that engagement of CD40 can induce high levels of Fas expression and Fas sensitivity in primary B cells. In this study, we have used Fas-positive, CD40-primed primary B cells or the M12.C3 B cell lymphoma, which also expresses high levels of Fas, to explore the role of [Mg 2ϩ ] i in Fas-induced apoptosis. Our results demonstrate that B cells undergoing apoptosis have elevated [Mg 2ϩ ] i and that this increase in free cytosolic Mg 2ϩ appears to serve as a "second messenger" for downstream events in apoptosis. over protein A-Sepharose (Sigma). RPMI 1640 complete medium (Cellgro, Herndon, VA) was supplemented with 10% heat-inactivated fetal bovine serum, 2 mM L-glutamine, 10 mM HEPES, 1 mM sodium pyruvate, 40 g/ml gentamicin, 100 units/ml penicillin G, 100 g/ml streptomycin, and 50 M 2-mercaptoethanol. Hanks' balanced salt solution (HBSS) and modified HBSS lacking Mg 2ϩ and Ca 2ϩ were obtained from Sigma.
Cells-AKR/J mice were purchased from the Jackson Laboratory (Bar Harbor, ME) and were used at 6 -10 weeks of age. Splenic B cells were prepared from murine splenocytes after the depletion of T cells by a mixture of anti-CD4 (GK 1.5), anti-CD8 (HO ϭ 2.2), and anti-Thy 1.2 (T24) mAbs with baby rabbit complement as described previously (17). The isolated primary B cells (5 ϫ 10 6 /ml) were incubated with anti-CD40 antibody (1C10) at 8 g/ml of RPMI 1640 complete medium in six-well tissue culture plates (5.5 ml/well) at 37°C and 5% CO 2 for 3 days (13). The CD40-stimulated B cells were washed with phosphatebuffered saline, and the viable cells were obtained by sedimentation of dead cells with Lympholyte-M. The isolated activated B cells had high Fas expression as determined by staining with fluorescein isothiocyanate-labeled anti-Fas mAb Jo2. The cell line used in this study is the B cell lymphoma M12.C3, which is maintained in RPMI 1640 complete medium and also has high Fas expression. Both types of cells were treated with varying concentrations of anti-Fas mAb in RPMI 1640 complete medium at 37°C and 5% CO 2 .
Determination of Mg 2ϩ and Ca 2ϩ Mobilization-Untreated and anti-Fas mAb-treated cells were washed with HBSS, loaded with 20 M mag-indo-1 AM or indo-1 AM in HBSS (pH 7.0) containing 10 mM HEPES for 30 min at 37°C, diluted 2-fold with prewarmed HBSS (pH 7.4) containing 5% fetal bovine serum and 10 mM HEPES; and incubated for an additional 30 min as described previously (18). The fluorochrome-loaded cells were then washed and resuspended in HBSS (pH 7.2) containing 5% fetal bovine serum and 10 mM HEPES for flow cytometric analysis. In the experiments in which the role of extracellular Ca 2ϩ and Mg 2ϩ was explored, the cells were washed, resuspended, and analyzed in modified HBSS containing only the appropriate divalent cation(s). An Ortho 50H cytometer with a 5-watt argon laser set for 364 nm excitation at 50 milliwatts was used. Fluorescence emission data were collected through a 390/490-nm band-pass filter using Phoenix Flow Acquisition software. The ratio of emissions for 390/490 nm was determined, and a cytogram of [Mg 2ϩ ] i or [Ca 2ϩ ] i was constructed by reference to standard curves of 390/490 nm emission ratio versus cation concentration (19). Readily mobilized pools of Mg 2ϩ and Ca 2ϩ in the cells (10 6 /ml/sample) were measured by treating the mag-indo-1and indo-1-loaded cells with 1 M ionomycin. Flow cytometric analysis was begun before addition of ionomycin to obtain base-line values. The cell flow was halted for addition of ionomycin, resulting in a gap in the profiles. The data were analyzed using Phoenix Flow System MTIME-NEW software, and mean [Mg 2ϩ ] i and [Ca 2ϩ ] i were calculated as functions of time. In addition, the percentage of cells mobilizing Mg 2ϩ after ionomycin treatment was determined for each time point, and maximum values were calculated.
Phosphatidylserine Expression-PS expression was determined using an ApoAlert annexin V apoptosis kit. The untreated and anti-Fas mAb-treated cells were washed with phosphate-buffered saline, and the washed cells (5 ϫ 10 5 ) were resuspended in 200 l of 1ϫ binding buffer. After adding 5 l each of fluorescein-labeled annexin V (0.5 g/ml final concentration) and propidium iodide (PI; 1 g/ml final concentration), the treated cells were incubated at room temperature for 15 min in the dark; the cells were then analyzed on a FACSCalibur flow cytometer (Becton Dickinson Advanced Cellular Biology, San Jose, CA). Two-color analysis allowed the resolution of viable (PS Ϫ PI Ϫ ), early apoptotic (PS ϩ PI Ϫ ), and late apoptotic and necrotic (PS ϩ PI ϩ ) populations.
DNA Fragmentation-Cells (5 ϫ 10 5 ) were incubated in 200 l of a solution containing PI (5 g/ml), saponin (0.3%), RNase (50 g/ml), and EDTA (5 mM) for 30 min at room temperature (20). Apoptotic cells were identified on a FACSCalibur flow cytometer as a distinct hypodiploid population that showed a diminished staining relative to the G 0 /G 1 population of the normal viable cells.  (Fig. 1, A and B, respectively; 0ϳ2 min). The mean value of [Mg 2ϩ ] i in the untreated cells was 500 M (Fig. 1E, dashed line at 0ϳ2 min), in good agreement with the range of values reported for hepatocytes (21), vascular smooth muscle cells (22), and cardiac muscle cells (23). The mean value for [Ca 2ϩ ] i was 100 nM (Fig. 1F, dashed line at 0ϳ2 min), in agreement with the published value of 90 nM for [Ca 2ϩ ] i in murine splenic B cells (19). Incubating M12.C3 cells for 3 h with anti-Fas mAb (0.5 g/ml) produced two distinguishable subpopulations with regard to [Mg 2ϩ ] i . In the anti-Fas mAb-treated cells, one population had a base-line [Mg 2ϩ ] i of 500 M (Fig. 1C, arrowheads), and a second population had a base-line [Mg 2ϩ ] i of 900 M (Fig. 1C, 0ϳ2 min). The existence of these two populations is reflected in the mean base-line [Mg 2ϩ ] i of 700 M for the treated cells (Fig. 1E, solid line, 0ϳ2 min). Intracellular Ca 2ϩ was monitored in parallel by flow cytometric analysis of indo-1-loaded cells. Fas cross-linking also gave rise to two separate populations, a well defined major one with a [Ca 2ϩ ] i of 100 nM and a broad "tail" with Ca 2ϩ i levels ranging up to 1000 nM or higher ( After the base-line values for [Mg 2ϩ ] i and [Ca 2ϩ ] i were established, the cells were treated in the flow cytometer with 1 M ionomycin and were monitored for several minutes longer (ϳ2-10 min) to determine if there are pool(s) of divalent cations that can be mobilized by ionophore treatment. In the untreated cells, ionomycin produced the classical Ca 2ϩ mobilization pattern with an initial sharp rise followed by a gradual decrease to a plateau value somewhat above base-line levels ( Fig. 1, B and F, dashed line). Untreated cells did not have a pool of Mg 2ϩ that could be mobilized by this same dose of ionomycin and showed an insignificant increase in the mean [Mg 2ϩ ] i (Fig. 1E, dashed line) as well as a transient response by no more than 22% of the cells in the population (Fig. 1G, dashed line). Ionomycin treatment of cells that had been incubated with anti-Fas mAb revealed significant differences from the control cells. The vast majority of the high [Mg 2ϩ ] i cells clearly had a pool of Mg 2ϩ that could be further mobilized by ionomycin treatment (Fig. 1, C and G, solid line), which is reflected in the gradual rise in the mean [Mg 2ϩ ] i , reaching maximum values by 8 -10 min after ionomycin treatment (Fig. 1E, solid line). The low [Mg 2ϩ ] i population (Fig. 1C, arrowheads) did not further mobilize Mg 2ϩ in response to ionomycin. Ionomycin treatment also revealed that the cells that had been incubated with anti-Fas mAb fell into two populations with regard to Ca 2ϩ mobilization. Although both subpopulations responded to ionomycin treatment, one continued to show the "classic" mobilization profile of the untreated cells, whereas the other reached and maintained extremely high Ca 2ϩ i levels (Fig. 1D). The behavior of these populations is reflected in the increased mean [Ca 2ϩ ] i response (Fig. 1F, solid line). Because both subpopulations mobilized Ca 2ϩ , it was not possible to use the flow cytometer software to calculate a percent response for Ca 2ϩ in order to discriminate between them.

Fas-mediated
The  Fig. 2A). This is also reflected in a dose-dependent increase in the mean [Mg 2ϩ ] i , which reached a plateau value of 700 M with 0.03 g/ml anti-Fas mAb (Fig.  2C, closed squares). Similarly, the percentage of cells in the high [Ca 2ϩ ] i state increased in a dose-dependent manner (Fig.  2B), and the base-line [Ca 2ϩ ] i increased from 100 to 170 nM (Fig. 2D, closed circles). These latter observations are consistent with the generally accepted model that elevation of [Ca 2ϩ ] i is required for apoptotic death. Ionomycin was used to determine the additional Ca 2ϩ and Mg 2ϩ mobilization potential at each dose of anti-Fas mAb (Fig. 2, C and D, open symbols). With ionomycin treatment, [Ca 2ϩ ] i increased ϳ100 nM regardless of the dose of anti-Fas mAb used. In contrast, ionomycin caused [Mg 2ϩ ] i to increase markedly only at anti-Fas mAb concentrations above 0.03 g/ml. We used the Mg 2ϩ response to ionomycin as a readout for cells in the high [Mg 2ϩ ] i state because these data (Figs. 1C and 2, A and C) suggest that the cells that are in the high [Mg 2ϩ ] i state are the same cells that further mobilize Mg 2ϩ in response to ionomycin treatment and because the available cytometric analysis software provides a direct calcu-lation of the percentage of cells mobilizing Mg 2ϩ in response to ionomycin.
Duplicate wells from the experiments shown in Fig. 2 (A-D) were examined for apoptotic cells using PI staining and flow cytometry. In saponin-permeabilized cells, PI uptake is proportional to DNA content, and apoptotic cells show diminished ("hypodiploid") staining below the G 0 /G 1 population of viable cells (20,24,25). The results of this analysis indicate that the percentage of cells mobilizing Mg 2ϩ correlates with the apoptotic (hypodiploid) cells (Fig. 2E).
Examination of Fas-mediated Apoptosis in Anti-CD40 Antibody-treated B Cells-Incubation of murine splenic B cells with CD40 ligand leads to enhanced Fas expression and increased sensitivity to Fas-mediated apoptosis, as seen in other systems (13,26). In a similar manner, we treated splenic B cells (which had been pretreated with anti-CD40 mAb for 72 h) with anti-Fas mAb for 2 h and then determined Mg 2ϩ mobilization, PS We also examined DNA fragmentation and PS externalization in the same experiment. PS is largely excluded from the external leaflet of the plasma membrane in viable cells, but becomes expressed on the external leaflet during apoptosis (5). Therefore, we used staining with annexin V, a naturally occurring PS-binding protein, to detect cells in early stages of apoptosis. Cells from parallel cultures used for Mg 2ϩ mobilization were doubly stained with fluorescein isothiocyanate-labeled annexin V and PI, without permeabilizing the cells (5). This technique allows the definition of three subpopulations: viable cells are PS Ϫ PI Ϫ , early apoptotic cells are PS ϩ PI Ϫ , and late apoptotic cells (and necrotic cells) are PS ϩ PI ϩ . In our experiments, we considered only the early apoptotic (PS ϩ PI Ϫ ) cells. The results (Fig. 3B) show that each of the responses reaches a plateau above 0.03 g/ml anti-Fas mAb and also indicate, for each dose of mAb, that the relative percentage of cells that mobilize Mg 2ϩ is same as the percentage of cells with increased PS externalization and DNA fragmentation. These results further demonstrate that Mg 2ϩ mobilization is a property of primary B cells undergoing apoptosis and not just an artifact associated with the use of the B cell lymphoma M12.C3.

Kinetics of Anti-Fas mAb-induced Mg 2ϩ Mobilization, DNA Fragmentation, and PS Externalization in B Cells-
We used kinetic analysis to determine whether Mg 2ϩ mobilization is an early or late event in apoptosis. In M12.C3 cells treated with 0.5 g/ml anti-Fas mAb, the mean base-line [Mg 2ϩ ] i increased with time, reaching a value of 1000 M after 1 h of incubation (Fig. 4A). At each time point (other than t ϭ 0), ionomycin caused a significant Mg 2ϩ mobilization in the high [Mg 2ϩ ] i population of cells that were committed to die by apoptosis as previously demonstrated. The kinetics of Mg 2ϩ mobilization versus DNA fragmentation were also measured in cells treated with 0.5 g/ml anti-Fas mAb (Fig. 4B). The percentage of cells showing Fas-potentiated, ionomycin-induced Mg 2ϩ mobilization was significantly higher than the percentage of those fragmenting DNA at 1-3 h after mAb treatment, but by 16 h, DNA fragmentation exceeded Mg 2ϩ mobilization. These results suggest that Mg 2ϩ mobilization precedes DNA fragmentation in B cells. We decided to examine loss of plasma membrane asymmetry on these cells because, whereas DNA fragmentation is considered to be a late stage apoptotic event in many systems, PS externalization occurs early in the process. Kinetic analysis of PS externalization versus Fas-potentiated, ionomycin-mediated Mg 2ϩ mobilization suggested that Mg 2ϩ mobilization preceded PS externalization (Fig. 4C), although the lag in PS response was not as striking as with DNA fragmentation (Fig.  4B). Because the percentage of cells in the PS ϩ PI Ϫ population never reached the same level of those mobilizing Mg 2ϩ , we checked to see if we were underestimating apoptosis by excluding the PS ϩ PI ϩ population. The PS ϩ PI ϩ population did not increase significantly over the 0-h time point except at the 16-h time point (data not shown), suggesting that the apparent decrease in PS ϩ PI Ϫ cells at that time point may be due to their conversion to PS ϩ PI ϩ , late apoptotic (or necrotic) cells. However, at the earlier time points (1-3 h), loss of PS ϩ PI Ϫ cells to the PS ϩ PI ϩ population is not an explanation for lower PS ϩ PI Ϫ numbers, and thus, PS externalization clearly lags behind Mg 2ϩ mobilization.

Intracellular Stores Are the Primary Source of [Mg 2ϩ ] i in B Cells
Undergoing Apoptosis-Calcium mobilization in B cells involves both release from intracellular stores (primarily endoplasmic reticulum) and the influx of extracellular Ca 2ϩ (see Fig. 1 and Ref. 27). Significantly less is known about the rela-tive roles of intracellular versus extracellular sources of Mg 2ϩ for Mg 2ϩ mobilization during apoptosis. We therefore examined the source of [Mg 2ϩ ] i in cells undergoing apoptosis. The experiments described above were repeated with M12.C3 cells in defined media lacking Mg 2ϩ , Ca 2ϩ , or both ions. The increase in the base-line [Mg 2ϩ ] i caused by anti-Fas mAb was not dependent on Mg 2ϩ in the extracellular milieu (Fig. 5A). Likewise, the percentage of cells possessing a readily mobilizable store of Mg 2ϩ was independent of extracellular Mg 2ϩ (Fig. 5B). In contrast, the production of a high [Mg 2ϩ ] i state and the generation of a pool of readily mobilized Mg 2ϩ were completely dependent on extracellular Ca 2ϩ (Fig. 5). These data support the concept that Ca 2ϩ mobilization is required for the induction of the apoptosis signaling pathway and place the Ca 2ϩ -dependent event(s) upstream of the release of Mg 2ϩ from intracellular stores.
Inhibition of Mg 2ϩ Mobilization and Apoptosis in B Cells by CCCP-Mitochondria are one of the major storage sites of intracellular Mg 2ϩ (28) and play a central role in intracellular signaling by releasing Mg 2ϩ in response to a variety of agents  (22,29). Therefore, we investigated Mg 2ϩ release from mitochondria during Fas-induced, ionomycin-potentiated Mg 2ϩ mobilization using CCCP. This agent uncouples oxidative phosphorylation in mitochondria (30), and short-term treatment of cells with CCCP has been reported to disrupt ⌬ m (4). We treated M12.C3 B lymphoma cells overnight with 20 M CCCP. Although the cells remained viable, they did show disrupted ⌬ m (as measured by decreased uptake of the cationic fluorochrome DiOC 6 (3); data not shown). Cells treated with CCCP for 16 h and with anti-Fas mAb for the last 3 h were examined for their ability to mobilize Mg 2ϩ and Ca 2ϩ in response to 1 M ionomycin. Pretreatment of the cells with CCCP did not alter their ability to mobilize Ca 2ϩ relative to controls not treated with CCCP (Fig. 6A, shaded bars). The slight (6%) decrease in ability to mobilize Ca 2ϩ between cells treated with anti-Fas mAb plus CCCP and those treated with anti-Fas mAb alone was not statistically significant. These results suggest that pretreatment of the cells with CCCP did not significantly alter their ability to flux Ca 2ϩ . Conversely, whereas pretreatment of the cells with CCCP did not alter the ability of the control cells to mobilize Mg 2ϩ , it did cause a marked (40%) and highly significant decrease in the ability of the anti-Fas mAb-treated cells to mobilize Mg 2ϩ .
In a subsequent experiment, parallel cultures were assayed for Mg 2ϩ mobilization, PS externalization, and DNA fragmentation (Fig. 6B). In the anti-Fas mAb-treated cells, pretreatment with CCCP caused a 37% inhibition of Mg 2ϩ mobilization relative to the untreated cells, a 41% inhibition of PS externalization, and a 45% inhibition of DNA fragmentation. In each case, the decrease in response was statistically significant, and the decreases in cells showing PS externalization and DNA fragmentation were proportional to the decrease in the number of cells mobilizing Mg 2ϩ . These data identify the mitochondria as a major source of intracellular Mg 2ϩ and, furthermore, are consistent with a role for Mg 2ϩ as a second messenger for apoptosis in B cells. The data also clearly separate the role of Mg 2ϩ from that of Ca 2ϩ since long-term treatment with CCCP does not significantly inhibit Ca 2ϩ mobilization in response to ionomycin treatment even while Mg 2ϩ mobilization and apoptosis are decreased. [Ca 2ϩ ] i rarely, if ever, achieved these levels in our experiments, it is unlikely that interference from [Ca 2ϩ ] i plays a significant role in our measurements of [Mg 2ϩ ] i .
The role of intracellular Ca 2ϩ in apoptosis is generally accepted (7,32), and our studies suggest a role for elevated [Ca 2ϩ ] i in B cell apoptosis as well. Multiple functions in apoptosis have been ascribed to increased [Ca 2ϩ ] i . These include activation of calpain, a caspase-like protease postulated to be one of the initiators of the apoptotic cascade (33,34); inactivation of the PS-specific flippase and activation of the phospholipid scramblase (35); and activation of Ca 2ϩ -dependent endonucleases (36 -38). Mitochondria serve as sinks for Ca 2ϩ released from the endoplasmic reticulum during intracellular Ca 2ϩ oscillations, and it recently has been postulated that sustained elevated [Ca 2ϩ ] i may cause the mitochondria to become overloaded with Ca 2ϩ , leading to downstream events that facilitate apoptosis (39). It should be noted, however, that although elevations in [Ca 2ϩ ] i may be necessary for B cell apoptosis, they are not sufficient, as proliferative signals such as those produced by ligation of the B cell receptor (surface immunoglobulin) also result in Ca 2ϩ mobilization.
In contrast to the well studied role of Ca 2ϩ in apoptosis, the role of Mg 2ϩ has been largely ignored. There is only a single report that elevating [Mg 2ϩ ] i in hepatocytes, using glycodeoxycholates (bile salts) to allow the ingress of extracellular Mg 2ϩ , leads to their death by apoptosis (8). Intracellular Mg 2ϩ is stored primarily in the microsomes (50% of total intracellular Mg 2ϩ ) and mitochondria (20% of total), with much smaller amounts in the nucleus and cytosol (28). About 98 -99% of total intracellular Mg 2ϩ (23) and 94% of cytosolic Mg 2ϩ (21) are complexed to nucleic acids, proteins, or membranes. Therefore, complex equilibria between compartments and between bound and free forms of Mg 2ϩ contribute to the concentration of free cytosolic Mg 2ϩ , i.e. [Mg 2ϩ ] i (28). The relatively slow release of Mg 2ϩ (as compared with Ca 2ϩ ) in the response to ionomycin treatment in our studies may reflect the need for Mg 2ϩ not only to cross the mitochondrial membrane, but also to dissociate from immobile ligands within the mitochondria.
Because our data with the mitochondrial oxidative phosphorylation inhibitor CCCP suggest that mitochondrial stores of Mg 2ϩ are critical for obtaining apoptosis and because our kinetic analysis suggests that Mg 2ϩ mobilization is an early event in apoptosis, we are struck by the parallel between disruption of ⌬ m and Mg 2ϩ mobilization. On one hand, the release of Mg 2ϩ from the mitochondria could be merely a consequence of the opening of mitochondrial pores as ⌬ m is reduced. On the other hand, there are at least two reports that, taken together, imply that Mg 2ϩ release from the mitochondria is independent of the loss of ⌬ m . First, the release of mitochondrial cytochrome c has been reported to occur prior to and to be independent of disruption of ⌬ m (40). Second, when isolated mitochondria are incubated with the pore-forming, pro-apoptotic molecule Bax, they incorporate Bax and release cytochrome c in a manner that is independent of ⌬ m , but "highly dependent on Mg 2ϩ ions" (9). Because of the importance of cytochrome c release in activating apoptotic protease activating factor-1 (Apaf-1) (41,42) and thereby driving post-mitochondrial events in cells undergoing apoptosis, we suggest that achieving high levels of [Mg 2ϩ ] i is central to the apoptotic process.
Although Mg 2ϩ is an important cofactor for a diverse set of functions in viable and proliferating cells (28), the Mg 2ϩ dependence of a number of endonucleases implicated in apoptosis (36,(43)(44)(45) suggests that elevated Mg 2ϩ levels are involved in apoptosis as well. The role played by Mg 2ϩ in PS externalization is less clear. One of the best candidates for the aminophos-pholipid translocase ("flippase") responsible for maintaining PS asymmetry in the plasma membrane of erythrocytes is a Mg 2ϩ -ATPase (46 -48). One might therefore assume that elevating [Mg 2ϩ ] i should decrease PS expression on the outer leaflet of the plasma membrane. However, although the Mg 2ϩ dependence of the flippase activity has been well documented, the effect of supernormal Mg 2ϩ concentrations is unknown. Thus, it may be possible that elevated [Mg 2ϩ ] i desensitizes the flippase, thus turning it off, or that Mg 2ϩ has no direct effect on flippase activity and that the regulation of PS externalization by Mg 2ϩ may be indirect, e.g. through the activation of caspases or other death-signaling molecules. Future experiments will establish the mechanism of, and the role played by, mitochondrial Mg 2ϩ in B cells undergoing Fas-initiated apoptosis.