Interleukin-3 Induces the Phosphorylation of a Distinct Fraction of Bcl-2*

Bcl-2-related proteins (i.e. Bcl-2 and Bax) regulate the effector stage of apoptosis and can modulate the entry of quiescent cells into the cell cycle. Phosphorylation of Bcl-2 is presumed to modify its apoptosis-inhibitory function. By utilizing an interleukin-3 (IL-3)-dependent hematopoietic cell line, we examined the structural requirements of Bcl-2 phosphorylation and the correlation of this post-translational modification with its function. In the presence of IL-3, constitutively expressed Bcl-2 was phosphorylated on serine residue(s), and phosphorylated Bcl-2 lost its capacity to heterodimerize with Bax. Whereas the majority of Bcl-2 resided in mitochondria, phosphorylation only affected a minor pool of total Bcl-2 that selectively partitioned into a soluble fraction. Cytosolic targeting of Bcl-2 greatly increased its ratio of phosphorylation. Bcl-2 phosphorylation was reduced during IL-3 deprivation, and its phosphorylation was also delayed after transient cytokine deprivation. This pattern of phosphorylation temporally correlated with the accelerated exit and delayed reentry of Bcl-2-expressing cells into the cell cycle upon transient IL-3 deprivation and subsequent cytokine restimulation. Thus, IL-3-induced phosphorylation of a distinct pool of Bcl-2 may contribute to the inactivation of its antiproliferative function.

The process of apoptosis is executed by a distinct genetic pathway that is apparently shared by all multicellular organisms. The family of Bcl-2-related proteins constitutes a class of apoptosis-regulatory gene products that act at the effector stage of apoptosis. In vertebrates, two functional classes of Bcl-2-related proteins exist that share highly conserved Bcl-2 homology (BH) 1 1, BH2, BH3, and BH4 domains: (a) antiapoptotic members, including Bcl-2, that inhibit cell death; and (b) proapoptotic members, including Bax, that accelerate apoptosis (reviewed in Refs. [1][2][3]. Several gene ablation studies confirm that in vertebrates, the balance between death-promoting and death-repressing members of the Bcl-2 family does indeed contribute a critical checkpoint that determines a cell's susceptibility to an apoptotic stimulus (reviewed in Ref. 2).
Beside regulating apoptosis, a number of recent studies demonstrate that Bcl-2-related proteins can also modulate the entry of quiescent cells into the cell cycle (reviewed in Ref. 2). For instance, when constitutively expressed in lymphocytes, Bcl-2 significantly delays the response of both B and T cells to a variety of mitogenic stimuli (4 -6), whereas Bax exerts an opposite effect (7). The molecular mechanism by which Bcl-2 alters cell cycle entry is poorly understood, but mutagenesis experiments demonstrated that the apoptosis-inhibitory and antiproliferative effects of Bcl-2 can be genetically separated (8). However, mutations in the BH1 or BH2 domains of Bcl-2 (5) or removal of its membrane insertion domain (9) interferes with Bcl-2's ability to affect cell cycle entry, arguing that intactness of its presumed hydrophobic pocket (10,11) and membrane integration are required for its cell cycle-inhibitory function.
A major question in relation to controlling the function of Bcl-2 family members is their regulation by extracellular signals. Several studies demonstrated the phosphorylation of Bcl-2 in a variety of cell lines (12)(13)(14)(15)(16)(17)(18). Among these, cytokineinduced survival of an IL-3-dependent myeloid cell line temporally correlated with serine phosphorylation of Bcl-2 (12), suggesting growth factor-initiated modulation of its function. However, the role of this post-translational modification on regulating Bcl-2's function is controversial. For instance, whereas Bcl-2 phosphorylation correlated with Taxol-induced apoptosis in some cell lines (14,15,17), in other cells types, phosphorylation of Bcl-2 was suggested to correlate with its antiapoptotic function (12,16,18).
In light of the emerging dual role of Bcl-2 as both an apoptosis-and cell cycle-inhibitory protein, we considered whether phosphorylation selectively alters one function of the molecule in certain cell types. To examine the structural requirements for Bcl-2 phosphorylation and the temporal correlation of Bcl-2 phosphorylation status and its function, we have utilized an IL-3-dependent lymphoid cell line, FL5.12, in which Bcl-2 blocks apoptosis after growth factor withdrawal (19). Of note, IL-3 promotes cell growth by simultaneously stimulating cell proliferation and suppressing apoptosis. This effect is achieved by receptor engagement-initiated activation of several distinct signal transduction pathways that regulate effector components of these two major cell fates (20,21).
Here we show that in the presence of IL-3, a fraction of constitutively expressed Bcl-2 was phosphorylated on serine residue(s), and this phosphorylated pool of Bcl-2 lost its capacity to heterodimerize with Bax. Whereas the majority of Bcl-2 resided in mitochondria, phosphorylation involved a minor pool of total Bcl-2 that selectively partitioned into a soluble fraction. Cytosolic targeting of Bcl-2 by deletion of its membrane insertion domain greatly increased its ratio of phosphorylation. The reduced phosphorylation of Bcl-2 upon IL-3 deprivation and its delayed rate of phosphorylation upon cytokine restimulation temporally correlated with the accelerated exit and delayed reentry of Bcl-2-expressing cells into the cell cycle. Thus, above a threshold level of Bcl-2 expression, IL-3-induced phosphoryl-ation of a distinct pool of Bcl-2 may represent a selective inactivation mechanism of its antiproliferative function.

EXPERIMENTAL PROCEDURES
Cell Culture and Apoptosis Induction-The IL-3-dependent murine cell line FL5.12, a lymphoid progenitor clone, and all its derivatives were maintained in Iscove's modified Dulbecco's medium supplemented with 10% fetal calf serum (Life Technologies, Inc.) and 10% WEHI-3B conditional medium as a source of IL-3 (19). To induce apoptosis, FL5.12 cells were washed three times in serum-free medium to remove the growth factor and cultured in the absence or presence of 25 IU/ml recombinant murine IL-3 (Genzyme).
Subcellular Fractionation-Cells were metabolically labeled as described above and lysed in hypotonic buffer (42.5 mM KCl, 5 mM MgCl 2 , and 10 mM HEPES, pH 7.4) by passaging them four times through a 30-gauge needle. Isotonicity was reestablished by adding an equal volume of hypertonic buffer (242.5 mM KCl, 5 mM MgCl 2 , and 10 mM HEPES, pH 7.4). Nuclei and unlysed cells were pelleted twice at 200 ϫ g for 10 min. The supernatant was centrifuged at 10,000 ϫ g for 10 min to collect the heavy membrane pellet. That supernatant was centrifuged at 100,000 ϫ g for 60 min, and the final supernatant was collected as the soluble fraction, and the pellet was collected as the light membrane fraction. The heavy membrane pellet was washed twice in H medium (0.25 mM mannitol, 0.075 M sucrose, 1 mM EGTA, 5 mM HEPES, pH 7.4, and 0.1% fatty acid-free bovine serum albumin). Both heavy and light membrane pellets were lysed in RIPA buffer and, together with the soluble fraction, were immunoprecipitated with mAb 6C8 as described above.
Western Blotting and Immunostaining-For immunoblots, proteins were electrotransferred at 4°C on polyvinylidine difluoride membranes (Millipore). Filters were blocked for 2 h with 3% non-fat milk in PBS. All additional immunostaining steps were performed in PBS with 0.05% Tween-20 (PBS-T) at room temperature. Filters were incubated with primary antibody and species-specific biotinylated secondary mAb (1: 300) for 2 h. Immunoblots were reacted with horseradish peroxidasestreptavidin (1:1000; Pierce) for 1 h. Filters were washed in PBS-T four times for 5 min between each step and developed with diazobenzidine (Bio-Rad) enhanced with nickel chloride (0.03%).
Cell Cycle Analysis-Cell cycle analysis was performed as described previously (25). Briefly, 1 ϫ 10 6 cells were washed in PBS, and the pellet was gently resuspended in 1 ml of hypotonic fluorochrome solution (3.4 mM sodium citrate, pH 7.8, 100 g/ml propidium iodide, 180 units/ml RNase A, 0.1% Triton X-100, and 30 mg/ml polyethylene glycol) and incubated for 20 min at 37°C. The cell suspension was then supplemented with 1 ml of hypertonic solution (356 mM NaCl, 100 g/ml propidium iodide, 0.1% Triton X-100, and 30 mg/ml polyethylene glycol) and stored at least for 6 h at 4°C before analysis. Propidium iodide fluorescence of individual nuclei were measured and analyzed using a FACScan flow cytometer (Becton Dickinson).
Our studies utilized the IL-3-dependent early lymphoid progenitor murine cell line FL5.12, whose viability and proliferation is maintained by a minimum of 25 IU/ml recombinant murine IL-3 (data not shown). In the absence of IL-3, FL5.12 dies by apoptosis, but overexpression of Bcl-2 significantly extends its survival without maintaining its proliferation (19). The Bcl-2 expression level in these FL5.12-Bcl-2 clones (26,27) was approximately the same as that seen in a cell line established from a patient with t(14;18)-bearing follicular B-cell lymphoma or in pre-B cells (19), thus representing physiologically relevant protein levels. Of note, FL5.12 cells do express a significant amount of endogenous Bax as well as a low amount of endogenous Bcl-2 that is insufficient to provide protection against IL-3 deprivation-induced apoptosis (24).
The Native Conformation of Bcl-2 Is Altered by Its Phosphorylation-In FL5.12 cells, the antiapoptotic Bcl-2 molecule resides predominantly in the mitochondria, whereas the majority of proapoptotic Bax is located in the cytosol in monomeric form (28). Upon apoptosis induction, cytosolic Bax undergoes a change in its conformation and translocates to mitochondria (28 -30). Here it can heterodimerize with Bcl-2 (28, 31), presumably involving interactions between their BH1, BH2, and BH3 domains (10,11,32). In the absence of proapoptotic signals, nonionic detergents, such as Nonidet P-40, are able to induce this conformational change of Bax and permit Bcl-2/Bax dimerization in cell lysates (33,34).
To test whether the native conformation of Bcl-2 is altered by its phosphorylation, nonionic detergent-induced dimerization between Bcl-2 and Bax was examined using co-immunoprecipitation experiments on Nonidet P-40 lysates of FL5.12-Bcl-2 cells. To compare the relative efficiency of co-immunoprecipitations with human Bcl-2-specific mAb 6C8 and murine Baxspecific mAb 4D2, [ 35 S]methionine labeling was performed. Parallel co-immunoprecipitations on Nonidet P-40 lysates of FL5.12-Bcl-2 cells revealed a reduced amount of Bcl-2 in anti-Bax 4D2 immunoprecipitates compared with the total amount of Bcl-2 precipitated through the human Bcl-2-specific 6C8 mAb ( Fig. 2A).
To examine anti-Bcl-2 and anti-Bax mAb immunoprecipitates of [ 32 P]orthophosphoric acid-labeled Nonidet P-40 cell lysates, gels were transferred onto polyvinylidine difluoride membranes and immunostained with human Bcl-2-specific mAb 124. This immunostaining demonstrated human Bcl-2 expression in FL5.12-Bcl-2 cells in a ratio comparable to that seen in [ 35 S]methionine-labeled samples (data not shown). Autoradiography of the same membrane revealed the expected ϳ25-kDa phosphorylated Bcl-2 in the anti-Bcl-2 immunoprecipitates of [ 32 P]orthophosphoric acid-labeled FL5.12-Bcl-2 lysates. In contrast, no corresponding protein could be immunoprecipitated with the anti-Bax 4D2 mAb. The faint ϳ25-kDa protein band visible in this lane is likely to represent the [ 32 P]orthophosphoric acid-contaminated light chain of the anti-Bax mAb utilized in the experiment (Fig. 2B). Thus, Nonidet P-40 is unable to induce dimerization between phosphorylated Bcl-2 and Bax, suggesting a conformational change of Bcl-2 after its phosphorylation.
Phosphorylated Bcl-2 Resides in a Soluble Subcellular Pool-In FL5.12 cells, the majority of Bcl-2 resides in the mitochondria as an integral membrane protein (19,28). A possible consequence of a phosphorylation-induced change in Bcl-2's conformation is a modification of the molecule's intracellular targeting.
To explore this idea, subcellular fractionations of FL5.12-Bcl-2 cells were performed as described previously (19,28). As an internal control, we used a C-terminally truncated human Bcl-2 mutant (Bcl-2 ⌬TM), in which the last 22 amino acids of the molecule that comprises its membrane insertion domain were removed (27,35). Of note, deletion of the membrane insertion domain of Bcl-2 shifts its localization from a membrane-bound form to the cytosol, as shown by both subcellular fractionation (27,36,37) and immunofluorescent localization (36,38,39) studies. Briefly, cells were labeled with [ 35 S]methionine or [ 32 P]orthophosphoric acid in the presence of IL-3 for 6 h and, after the removal of nuclear fractions and unlysed cells, were subfractionated to heavy membrane, light membrane, and soluble cytosolic fractions by differential ultracentrifugation. Heavy membrane fractions enriched in mitochondria and the fraction containing light membranes enriched in endoplasmic reticulum were solubilized with RIPA lysis buffer. These lysates and the remaining soluble S100 fraction representing the cytosol were immunoprecipitated with anti-Bcl-2 6C8 mAb, and the subcellular localization patterns of [ 35 S]methionine-and [ 32 P]orthophosphoric acid-labeled Bcl-2 were compared.
The majority of [ 35 S]methionine-labeled Bcl-2 localized to the heavy membrane fraction, with only minor amounts observed in the light membrane and S100 fractions, whereas Bcl-2 ⌬TM localized to the S100 fraction (Fig. 3A, left panels), as described previously (27,36,37). However, phosphorylated Bcl-2 and Bcl-2 ⌬TM were both found almost exclusively in the S100 fraction (Fig. 3A, right panels). We conclude that whereas the majority of Bcl-2 is localized to the heavy membrane fraction in FL5.12-Bcl-2 cells, the membrane integration of phosphorylated Bcl-2 is compromised.
Cytosolic Targeting Substantially Enhances Bcl-2 Phosphorylation-Because phosphorylated Bcl-2 localized to the S100 cytosolic fraction, we next asked whether targeting Bcl-2 to the cytosol by removal of its membrane insertion domain enhances its rate of phosphorylation. To this end, FL5.12-Bcl-2, FL5.12-Bcl-2 mI-3, and FL5.12-Bcl-2 ⌬TM cells were labeled with [ 35 S]methionine or [ 32 P]orthophosphoric acid in the presence of IL-3 for 6 h, lysed in RIPA lysis buffer, and immunoprecipitated with the human Bcl-2-specific 6C8 mAb. Immunoprecipitates of [ 35 S]methionine-labeled cell lysates revealed a comparable amount of Bcl-2 expression in all three cell lines (Fig. 3B,  top panel). However, Bcl-2 ⌬TM proved ϳ50ϫ more phosphorylated than the mostly membrane-bound Bcl-2 and Bcl-2 mI-3 proteins (Fig. 3B, bottom panel). Thus, targeting Bcl-2 into the cytosol by preventing its membrane association substantially Bcl-2 Delays IL-3-induced Cell Proliferation in FL5.12 Cells-Constitutively expressed Bcl-2 prevents the apoptosis of FL5.12 cells upon IL-3 withdrawal (19,24). As only a minor S100 fraction of Bcl-2 is phosphorylated in the presence of IL-3, the physiologic role of this post-translational modification may not directly relate to the molecule's apoptosis-inhibitory function in these cells. Besides inhibiting apoptosis, several studies demonstrated that Bcl-2 can also delay the entry of cells into the cell cycle (reviewed in Ref. 2). Thus, we wished to examine the potential relationship between Bcl-2's phosphorylation status and its antiproliferative effect in these cells.
First, the cell cycle status of FL5.12-Neo R and two FL5.12-Bcl-2 clones was tested after transient IL-3 deprivation and subsequent IL-3 restimulation, in a manner similar to that described previously (40). Before IL-3 deprivation, the apparent rate of cell proliferation and the ratio of cells in S phase ϩ G 2 /M phase were essentially identical in all clones (Fig. 4A,  0hr), although the proportion of Bcl-2-expressing cells in S phase was somewhat lower than that in FL5.12-Neo R cells (Fig.  4B, 0hr). Upon transient IL-3 withdrawal, Bcl-2-expressing clones started to accumulate in the G 1 phase of the cell cycle faster than Neo R cells, and the ratio of Bcl-2-expressing cells in G 2 /M phase also increased. Because control FL5.12-Neo R cells remain fully viable for only 12 h after cytokine deprivation (24), IL-3 was added back to all clones at this time. 12 h after the readdition of IL-3, a lower proportion of FL5.12-Bcl-2 cells was in the S phase ϩ G 2 /M phase of the cell cycle compared with FL5.12-Neo R cells (Fig. 4A, ϩ 12hr). At this time, FL5.12-Neo R cells demonstrated a synchronous entry of cells into S phase, which was delayed in FL5.12-Bcl-2 cells (Fig. 4B, ϩ 12hr). By 24 h after the readdition of IL-3, this transient difference had disappeared (Fig. 4, A and B). These data demonstrate that similarly to that seen before (40), Bcl-2 is able to provoke a temporary refractoriness to IL-3-stimulated cell proliferation in FL5.12 cells.
Phosphorylation of Bcl-2 Is Dependent on the Presence of Interleukin-3-To determine the temporal correlation of Bcl-2's phosphorylation status and its antiproliferative function, FL5.12-Bcl-2 and FL5.12-Neo R cells were metabolically labeled with either [ 32 P]orthophosphoric acid or [ 35 S]methionine in the presence or absence of recombinant murine IL-3 and solubilized with RIPA lysis buffer at distinct time points thereafter.
When lysates of FL5.12-Neo R cells were immunoprecipitated with the endogenous murine Bcl-2-specific mAb 3F11, no phosphorylation of endogenous Bcl-2 was detected during the first 8 h of [ 32 P]orthophosphoric acid labeling (Fig. 5A, top panel) in either the presence (ϩ) or absence (Ϫ) of IL-3. Similarly, lysates of FL5.12-Neo R (or FL5.12-Bcl-2) cells did not demonstrate any phosphorylation of Bax when precipitated with the anti-Bax mAb within the same time frame (Fig. 5A, middle  panel). In contrast, efficient phosphorylation of constitutively expressed Bcl-2 was detected in the presence of IL-3 when it was immunoprecipitated with the human Bcl-2-specific 6C8 mAb (Fig. 5A, bottom panel, ϩ). However, in the absence of IL-3 during metabolic labeling, a significant reduction in the amount of phosphorylated human Bcl-2 was observed (Fig. 5A,  bottom panel, Ϫ).
To ascertain that the differences seen in the phosphorylation of overexpressed Bcl-2 in the presence or absence of IL-3 were not due to differences in the rate of its de novo protein synthesis, [ 35 S]methionine labeling was performed within the same time course. Of note, the amount of endogenous murine Bcl-2 is about 10% compared with the amount of overexpressed human Bcl-2 (19), whereas endogenous Bax levels are comparable to that of overexpressed Bcl-2 (24). Both endogenous Bcl-2 and Bax incorporated a detectable amount of radiolabeled methionine at the same rate in the presence or absence of IL-3 (Fig.  5B, top and middle panels). Similarly, metabolic labeling demonstrated identical amount of newly synthesized constitutively expressed Bcl-2 in the presence or absence of the cytokine (Fig.  5B, bottom panel). We conclude that above a threshold level of protein expression, a distinct pool of Bcl-2 is phosphorylated in the presence of IL-3, but upon cytokine deprivation, Bcl-2 is either dephosphorylated or its phosphorylation is inefficient.

IL-3-induced Bcl-2 Phosphorylation
Correlates with Reentry of FL5.12 Cells into the Cell Cycle-To further examine the temporal correlation of Bcl-2's phosphorylation status and its cell cycle inhibitory function, we determined the phosphorylation status of Bcl-2 after transient IL-3 deprivation and subsequent IL-3 restimulation. To this end, FL5.12-Bcl-2 cells were first transiently deprived of IL-3 for 12 h. These cells were then metabolically labeled with [ 32 P]orthophosphoric acid or [ 35 S]methionine in the presence of recombinant IL-3 for 3, 6, and 12 h (Fig. 6, IL-3 Depr.). As controls, identical metabolic labeling was performed on FL5.12-Bcl-2 cells that were incubated in the continuous presence of IL-3 (Fig. 6, Control). Samples were lysed in RIPA lysis buffer and immunoprecipitated with the human Bcl-2-specific 6C8 mAb at the indicated time points.
Compared with control cells, the amount of phosphorylated Bcl-2 was reduced in transiently IL-3-deprived cells at all time points tested. This difference was most pronounced at 3 h after the initiation of metabolic labeling. However, at 6 h, but not at 12 h, the phosphorylation level of Bcl-2 was still weaker in the transiently IL-3-deprived cells than in control cells (Fig. 6, top  panel). To ascertain that the differences seen in Bcl-2's level of phosphorylation were not due to differences in the rate of its de novo protein synthesis, [ 35 S]methionine labeling was performed within the same time course. As shown in Fig. 6, bottom panel, both transiently IL-3-deprived and control cells incorporated radiolabeled methionine at similar rates at all time points. Thus, after transient cytokine withdrawal, the delayed rate of IL-3-stimulated phosphorylation of Bcl-2 ( Fig. 6) correlated to a certain degree (3 and 6 h) with the reduced sensitivity of FL5.12-Bcl-2 cells to IL-3-stimulated cell proliferation (Fig. 4). DISCUSSION In vertebrates, death-promoting and death-repressing members of the Bcl-2-related proteins are important regulators of the effector stage of apoptosis. The regulation of their biologic activity is poorly understood, but differential subcellular targeting of these molecules is clearly involved. For instance, Bcl-2 is an integral membrane protein that resides in mitochondria, endoplasmic reticulum, and nuclear membranes (19,41,42). In contrast, the majority of proapoptotic Bax is located in the cytosol in monomeric inactive form (28,29). Upon apoptosis induction, cytosolic Bax translocates from the cytosol to the mitochondria, where it displays its apoptosis-inducing function (28 -30), which perhaps involves its heptamerization (43). Alternatively, mitochondrial Bax can heterodimerize with Bcl-2 (28,31), an association that is likely to involve interactions between their BH1, BH2, and BH3 domains. NMR and x-ray crystallography structure of Bcl-x L monomer (10) and of a Bcl-x L -Bak BH3 peptide complex (32) revealed both hydrophobic and electrostatic interactions between the BH3 amphipathic ␣-helix and the Bcl-x L hydrophobic pocket formed by its BH1, BH2 and BH3 domains (32). Selective BH1 mutations that abolish Bcl-2's heterodimerization capacity with Bax in a yeast two-hybrid assay (22) can also reduce its antiapoptotic function in mammalian cells (26,44), further underlying the functional significance of this interaction.
Another type of functional regulation of Bcl-2 family members involves their post-translational modification by phosphorylation. For instance, phosphorylation of proapoptotic Bad (45) on two serine residues in response to IL-3 stimulation promotes its cytosolic targeting and association with the 14-3-3 family of proteins (46). Similarly, a number of studies demonstrated the phosphorylation of Bcl-2 in a variety of cell lines, but the functional consequence of this post-translational modification remains unclear. In some studies, chemotherapeuticinduced apoptosis correlated with concomitant phosphorylation of Bcl-2 (14,15,17,47,48), whereas in others phosphorylation of Bcl-2 correlated with its antiapoptotic func- tion (12,16,18). Among these, cytokine-induced survival of an IL-3-dependent myeloid cell line temporally correlated with serine phosphorylation of Bcl-2 (12), suggesting growth factorinitiated modulation of its function.
Besides regulating apoptosis, Bcl-2-related proteins can also modulate the entry of quiescent cells into the cell cycle (reviewed in Ref. 2), suggesting a cell autonomous coordination between proliferation and cell death. In light of the emerging dual role of Bcl-2 as both an apoptosis-and cell cycle-inhibitory protein, we considered whether phosphorylation of Bcl-2 may differentially alter just one function of the molecule in certain cell types. To examine the temporal correlation of Bcl-2's phosphorylation status and its antiproliferative function, we have utilized an IL-3-dependent lymphoid cell line in which Bcl-2 blocks apoptosis after growth factor withdrawal.
The data presented in this paper suggest that IL-3-induced phosphorylation of Bcl-2 may temporally correlate better with abrogation of its cell cycle-inhibitory effect than with regulation of its apoptosis-inhibitory function. First, transient IL-3 deprivation resulted in a reduced level of Bcl-2 phosphorylation and, in time, correlated with the accelerated exit of Bcl-2expressing cells from the cell cycle (Figs. 4 and 5). Similarly, upon cytokine restimulation, Bcl-2-expressing clones exhibited a temporary refractoriness to IL-3-induced cell proliferation that correlated with the delayed rate of Bcl-2 phosphorylation (Figs. 4 and 6). Thus, in both cases, the phosphorylation status of Bcl-2 correlated with its antiproliferative effect. Also, in FL5.12 cells, Bcl-2 provides an extended protection from IL-3 deprivation-induced apoptosis (19). Taken together, these data suggest that IL-3-induced phosphorylation of Bcl-2 may contribute to the inactivation of its antiproliferative function rather than altering Bcl-2's antiapoptotic effect. To directly test this hypothesis, identification of IL-3-induced Bcl-2 phosphorylation sites and the creation and testing of phosphorylationdeficient Bcl-2 mutants will be needed.
Several considerations also suggest that in FL5.12 cells, IL-3-stimulated Bcl-2 kinase may not affect the majority of constitutively expressed Bcl-2. First, in FL5.12 cells in the presence of IL-3, Bcl-2 predominantly localized to the mitochondria-rich heavy membrane fraction, in accordance with that described previously (Refs. 19 and 28; Fig. 3A, left panel). However, almost all phosphorylated Bcl-2 resided in the soluble S100 fraction (Fig. 3A, right panel). This demonstrates that phosphorylation-induced conformational change of Bcl-2 results in the loss of its capacity for firm membrane integration. Second, Bcl-2 ⌬TM, a Bcl-2 mutant that is predominantly cytosolic due to deletion of its membrane insertion domain, was ϳ50ϫ more phosphorylated than membrane integrated Bcl-2 when both exhibited a similar overall protein expression level (Fig. 3B). This suggests that IL-3-stimulated Bcl-2 kinase is sufficiently active to phosphorylate perhaps all intracellular Bcl-2. Consequently, the majority of Bcl-2 is apparently shielded from this kinase activity and integrates to the mitochondria in the presence of IL-3.
But what shields Bcl-2 from this kinase effect? Two separate scenarios can be envisioned. First, IL-3-induced Bcl-2 kinase may require sufficient time to phosphorylate cytosolic Bcl-2. Thus, rapid mitochondrial integration of newly synthesized Bcl-2 may prevent such an effect. However, slower integration to alternative sites, such as endoplasmic reticulum membranes or selected mitochondrial subregions, may allow sufficient time for phosphorylation to take place that subsequently prevents Bcl-2's membrane integration (Fig. 7). Of note, in vitro targeting experiments demonstrated that efficient insertion of Bcl-2 into the mitochondrial outer membrane is mechanistically different from its comparatively low-affinity association with en-doplasmic reticulum that may not be dependent on its C-terminal membrane insertion domain (35). In the second scenario, Bcl-2 that is already inserted in the membrane may show a differential sensitivity to kinase activity according to its site of integration. Thus, a small pool of mitochondria-or endoplasmic reticulum-localized Bcl-2 with subtle differences in its protein conformation may be selectively accessible to IL-3-stimulated Bcl-2 kinase, whereas the majority of its mitochondrial counterpart is shielded from it. Expression of Bcl-2 mutants targeted selectively to the endoplasmic reticulum or mitochondrial subregions in FL5.12 cells will be needed to clarify this issue.