Proximal Events in 7,12-Dimethylbenz[a]anthracene-Induced, Stromal Cell-Dependent Bone Marrow B Cell Apoptosis: Stromal Cell–B Cell Communication and Apoptosis Signaling

Polycyclic aromatic hydrocarbons (PAHs), such as benzo[a]pyrene (B[a]P) and benz[a]anthracene (BA), are immunosuppressive in rodent models (1–6), and they alter immune responses in humans (7–10). These and other PAHs are capable of binding and activating the aryl hydrocarbon receptor (AhR) and are produced by a number of industrial processes and by the incomplete combustion of carbon-containing compounds including fossil fuels. Human exposure to environmentally ubiquitous PAHs, such as B[a]P and BA, regularly occurs through ingestion of contaminated food and inhalation of vehicle exhaust or cigarette smoke (11, 12).

7,12-Dimethylbenz[a]anthracene (DMBA) is a highly toxic, methylated derivative of BA that has been used extensively in model systems designed to elucidate mechanisms of PAH-mediated immunotoxicity and carcinogenicity (13). In rodent models, both B[a]P and DMBA induce a reduction in bone marrow cellularity resulting largely from a massive loss of B cells (5, 6, 14, 15). In an in vitro bone marrow stromal cell/B cell coculture system, DMBA induces death of pro- and pre-B cells through apoptosis signaling pathways that resemble those activated during immature B cell clonal deletion (15–19). Interestingly, exposure of B cells alone to PAHs does not induce apoptosis (20). B cell death requires contact with AhR- and cytochrome P450 1B1 (CYP1B1)-expressing stromal cells (21, 22).

Bone marrow stromal cells, multipotent cells that are capable of differentiating into either osteoblasts or adipocytes, are responsible for maintaining the milieu of cytokines and adhesion/interaction/matrix molecules that support B lymphopoiesis (23). In fact, deletion of key molecules that support stromal cell–B cell interactions suppresses B lymphopoiesis (e.g., stromal derived factor-1) (24, 25). Direct contact between stromal cells and B cells not only suppresses spontaneous pre-B cell apoptosis but also suppresses cytokine- and glucocorticoid-induced pre-B cell apoptosis (26–28). Conversely, stromal cell–B cell contact is required for initiation of DMBA-induced pro/pre-B cell apoptosis because treatment of B cells alone with DMBA or with conditioned medium from DMBA-treated stromal cells is not sufficient to induce B cell death (21, 29).

The nature of the death signal transferred from the stromal cells is unknown. However, we have demonstrated that AhR expression in the stromal cells is required, that metabolism of DMBA likely precedes transfer of the death signal to stromal cell-adherent B cells, and that DMBA-induced apoptosis does not result from the production of death receptor ligands by stromal cells or by activation of known death receptors on B cells (17, 20, 30). From these results, we have hypothesized that the death signal that is transferred between stromal cells and B cells is highly labile, likely a DMBA metabolite, and that it initiates an intrinsic apoptotic pathway.

The mechanism by which the death signal is transferred from the stromal cells to the bone marrow B cells is harder to postulate. Communication among cells in the immune system largely occurs through production/secretion of cytokines and exchange of membranes (trogocytosis). Cytokines are an unlikely mediator in this case, because we have previously shown that several common death-inducing cytokines and their receptors (e.g., TNF-α, TNF-β, lymphotoxin-β, TNFRs (TNFR1 and TNFR2), Fas, and death receptor 6) are not involved in DMBA-induced death (30). In the absence of evidence for death receptor or cytokine involvement, we hypothesized that a trogocytosis-like mechanism may be at work. Trogocytosis (rapid, contact-dependent, intercellular membrane patch exchange) typically facilitates transfer of Ag following formation of an immunological synapse (31). However, it can occur in Ag-independent cell–cell interactions as well (32).

Analysis of the interactions between stromal cells and B cells during DMBA-induced death signaling may reveal important information on the nature of communication between these bone marrow cell subsets in the presence or absence of environmental chemicals. The studies described in this paper were designed to examine the hypothesis that an intrinsic apoptotic pathway is activated by a DMBA metabolite transferred to the B cell via membrane exchange with a stromal cell capable of metabolizing DMBA. Accordingly, we defined the intracellular DMBA/stromal cell-induced B cell death pathway, showing that it is dependent upon cytochrome c release and apoptosome formation and that it is p53-dependent. We determined that a terminal DMBA metabolite, DMBA-3,4-dihydrodiol-1,2-epoxide (DMBA-DE), is sufficient to induce apoptosis in B cells in the absence of stromal cells and showed that p53 also is critical for DMBA-DE–induced apoptosis. These data are consistent with transfer of a death-inducing epoxide from the stromal cell to the B cell and the subsequent initiation of the intrinsic apoptosis pathway. Finally, we provide evidence that membrane transfer occurs between bone marrow stromal cells and B cells, but not T cells, and suggest that this is the mechanism of transfer of the otherwise labile, death-inducing DMBA metabolite.

The caspase-8–specific Ab was from Axxora (San Diego, CA). The cytochrome c-specific Ab and phenotyping Abs were from BD Biosciences (Palo Alto, CA). Abs specific for cleaved caspase-3 and -9 and cleaved lamin were from Cell Signaling Technology (Beverly, MA). The p53-specific Ab was from Santa Cruz Biotechnology (Santa Cruz, CA). Plasmocin was from InvivoGen (San Diego, CA). JC-1 was from Molecular Probes (Eugene, OR). Murine rIL-7 was from Research Diagnostics (Flanders, NJ). Caspase inhibitors were from R&D Systems (Minneapolis, MN). DMBA, propidium iodide (PI), Protease Inhibitor Cocktail for Mammalian Cells, and the β-actin–specific Ab were from Sigma Chemical Company (St. Louis, MO). All of the other reagents were from Thermo Fisher Scientific (Suanee, GA).

Stromal cell-dependent, CD43⁺ BU-11 cells expressing rearranged cytoplasmic Ig H chains (pro/pre-B cells) (15) were cocultured on cloned BMS2 bone marrow-derived stromal cells (33) (kindly provided by Dr. P. Kincade, Oklahoma Medical Research Foundation, Oklahoma City, OK). Stocks of BU-11 cells were maintained on BMS2 cell monolayers in an equal mixture of DMEM and RPMI 1640 medium with 5% bovine growth serum (BGS; Thermo Fisher Scientific, formerly Hyclone), plasmocin, l-glutamine, and 2-ME. All of the cultures were maintained at 37°C in a humidified, 7.5% CO₂ atmosphere. Cell cultures were determined to be mycoplasma negative by PCR (Mycoplasma Detection Kit; American Type Culture Collection, Manassas, VA).

All of the animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at Boston University. Primary bone marrow pro-B cell cultures were prepared from wild-type C57BL/6 and B6.129S2-Trp53^tm1Tyj/J or Apaf-1^fog/J homozygous and heterozygous mice (The Jackson Laboratory, Bar Harbor, ME) essentially as described (34). Bone marrow was flushed from the femurs of 4- to 8-wk-old mice. RBCs were lysed by incubation in 0.17 M NH₄Cl, 10 mM KHCO₃, and 1 mM EDTA at 37°C for 5 min. The remaining cells were cultured for 5–7 d in primary B cell medium (RPMI 1640 containing 10% FBS, penicillin/streptomycin, l-glutamine, 2-ME, and 16 ng/ml murine rIL-7). This procedure results in a B cell culture in which at least 95% of the cells express CD43 and B220.

B and T lymphocytes were isolated from spleens of C57BL/6 mice by dissociation and use of SpinSep Mouse B Cell Enrichment and T Cell Enrichment Kits, respectively (StemCell Technologies, Vancouver, British Columbia, Canada). To assess purity, aliquots of cells were stained with anti–B220-FITC or FITC-conjugated rat IgG_2a for B cells or anti–CD3-FITC or FITC-conjugated rat IgG_2b for T cells, fixed in 0.4% paraformaldehyde, and analyzed on a FACScan flow cytometer (BD Biosciences).

For coculture experiments, BMS2 cells (2 × 10⁴/cm²) were cultured for 24 h in DMEM containing 5% BGS to form a monolayer that was ∼75% confluent. BU-11 or primary pro-B cells were added in RPMI 1640 containing 5% BGS (or 10% FBS and 16 ng/ml rIL-7 for primary pro-B cells) at a final concentration of 2 × 10⁵ cells/ml and allowed to associate with the stromal cells overnight. For some experiments, B cells and BMS2 cells were separated by a 3-μm pore size Transwell insert (Corning, Corning, NY). Stromal cell/B cell cocultures were treated with vehicle (DMSO, 0.1% final concentration) or DMBA (1–10 μM; Sigma) for 2–16 h. Vehicle, the pan-caspase inhibitor Z-Val-Ala-Asp(OMe)-fluoromethylketone, the caspase-3 inhibitor Z-Asp(OMe)-Glu(OMe)-Val-Asp(OMe)-fluoromethylketone, or the caspase-6 inhibitor Z-Val-Glu(OMe)-Ile-Asp(OMe)-fluoromethylketone, or the caspse-9 inhibito Z-Leu-Glu(OMe)-His-Asp(OMe)-fluoromethylketone (15–30 μM) were added to cocultures 30 min prior to vehicle or DMBA treatment.

For single culture experiments, BU-11 cells (2 × 10⁵ cells/ml) were cultured for 24 h in RPMI 1640 containing 5% BGS and 16 ng/ml rIL-7 prior to treatment with vehicle, etoposide (0.1 μg/ml; Sigma), or anti–DMBA-DE (0.001–1 μM; National Cancer Institute Chemical Carcinogen Reference Standard Repository).

B cells were harvested and washed once in cold PBS. For analysis of cleavage of caspases or their substrates, cytoplasmic extracts were prepared as described previously (35). For analysis of cytochrome c release, cytoplasmic fractions from digitonin-permeabilized cells were prepared as described previously (36). Protein concentrations were determined by the Bradford method.

Total proteins (15–50 μg) were resolved on 12 or 15% gels, transferred to a 0.2-μm nitrocellulose membrane, and incubated with primary Ab. Primary Abs included polyclonal rabbit anti-cleaved caspase-3 (9661), polyclonal rat anti–caspase-8 (ALX-804-447), polyclonal rabbit anti–caspase-9 (9504), polyclonal rabbit anti-p53 (SC-6243), monoclonal mouse anti-cleaved lamin A (2036), and polyclonal rabbit anti-cytochrome c Ab (S2050). Immunoreactive bands were detected using HRP-conjugated secondary Abs (Bio-Rad, Hercules, CA) followed by ECL. To control for equal protein loading, blots were reprobed with a β-actin–specific Ab (A5441) and analyzed as above.

Mitochondrial membrane potential was analyzed by JC-1 and analyzed by flow cytometry as described (37). The percentage of cells with low mitochondrial membrane potential (ΔΨ_m^low) was determined to be those having an increased green fluorescence with or without a loss of red fluorescence. DNA fragmentation was analyzed by PI staining and flow cytometry as described previously (30, 35). The percentage of cells undergoing apoptosis was determined to be those having a weaker PI fluorescence than cells in the G₀/G₁ phase of the cell cycle. Caspase activity in cytoplasmic extracts was determined using p-nitroaniline-conjugated substrates and spectrometry, according to the manufacturer’s instructions (ApoAlert; Clontech, Mountain View, CA).

BMS2 cells were plated in DMEM containing 5% BGS on half of a glass-bottom 35-mm petri dish with a 14-mm microwell with a number 0 cover glass below the microwell (MatTek, Ashland, MA) using a removable barrier placed down the center of the microwell. After incubation for 24 h, the barrier was removed. Cultures were treated with 10 μM DMBA for 30 min, washed with PBS to remove free DMBA, incubated for 1 h, and then washed again with PBS. BU-11 cells were added in RPMI 1640 containing 5% BGS and centrifuged onto the BMS2 layer for 10 min at 1500 rpm. Excess nonadhering BU-11 cells were removed by gentle washing, fresh medium was added, and cultures were immediately analyzed by two-photon confocal microscopy for DMBA/metabolite fluorescence.

BMS2 cells were plated in six-well plates in DMEM containing 5% BGS and allowed to adhere overnight. Cells were treated with vehicle or DMBA (10 μM) for 30 min. Medium was removed, cells were washed with PBS, and new medium was added. Cells were incubated for an additional 5 h followed by medium removal, washing, and replacement. BU-11 cells, in RPMI 1640 containing 5% BGS, then were added directly to the BMS2 cells or to 3-μm pore size Transwell inserts over the BMS2 cells. BU-11 cells were harvested after 18 h and analyzed immediately for DMBA/metabolite fluorescence on a MoFlo flow cytometer (DakoCytomation, Carpinteria, CA). Only cells in the live B cell gate were analyzed.

Vybrant DiO (Invitrogen, Carlsbad, CA) was used essentially as directed by the manufacturer. BMS2 cells were plated in 12-well plates in DMEM containing 5% BGS and allowed to adhere overnight. Cells were stained with 5 μl/ml Vybrant DiO in culture medium for 20 min at 37°C followed by extensive washing. Medium (DMEM containing 5% BGS) was replaced, and BU-11 cells or primary pro-B, splenic B, or splenic T cells from wild-type C57BL/6 mice were added in RPMI 1640 containing 5% BGS. For some experiments, BU-11 cells were added to 3-μm pore size Transwell inserts over the stained BMS2 cells. Lymphocytes were harvested after 16 h, washed, and fixed in 0.4% paraformaldehyde. In addition, naive BU-11 cells were resuspended in 16 h supernatants from wells of DiO-stained BMS2 cells and incubated at room temperature for 20 min before washing and fixing. Samples were analyzed by flow cytometery. Only cells in the live lymphocyte gate were analyzed.

Statistical analyses were performed with Statview (SAS Institute, Cary, NC). Data are presented as means ± SEs. At least three experiments were performed in each BU-11 protocol. Experiments with primary B and T cells were performed with cells from a minimum of three independently prepared and maintained pools of bone marrow cells or three individual spleens. One-factor ANOVAs were used to analyze the data, with the Dunnett or Tukey-Kramer multiple-comparisons test to determine significant differences.

Previous studies demonstrated that, even though caspase-8 is activated in the DMBA/stromal cell-induced apoptosis pathway, production of death-inducing cytokines and triggering of an extrinsic, death receptor-mediated apoptosis pathway is not the mechanism of DMBA-induced apoptosis (30). These results suggest that DMBA/stromal cell-induced B cell apoptosis is initiated through the intrinsic pathway and are consistent with the hypothesis that a DMBA metabolite is the death signal transferred from the stromal cell to the B cell. Therefore, mediators of an intrinsic pathway were evaluated.

A key feature of the intrinsic apoptotic pathway is mitochondrial outer membrane permeabilization (MOMP). To examine mitochondrial integrity, the nontransformed pro/pre-B cell line BU-11 was treated in coculture with BMS2 cell monolayers with vehicle (DMSO, 0.1% final concentration) or DMBA (1 μM). Cytosolic proteins recovered from digitonin-permeabilized BU-11 cells were analyzed for cytochrome c by immunoblotting. Cytochrome c release into the cytosol was evident 4–6 h after DMBA treatment (Fig. 1A), indicating that the B cell mitochondria were permeabilized.

Loss of mitochondrial membrane potential (ΔΨ_m) often, but not always, occurs in apoptosis mediated by the intrinsic pathway (38, 39). To determine if mitochondrial membrane potential is altered during DMBA-induced apoptosis, BU-11/BMS2 cell cocultures were treated with vehicle or DMBA, and mitochondrial membrane potential was analyzed in BU-11 cells by JC-1 staining and flow cytometry. BU-11 cells did not undergo significant mitochondrial membrane potential loss at any time point tested (Fig. 1B). Similar results were seen using 3,3′-dihexyloxacarbocyanine iodide (data not shown). Because it is unusual for cells to undergo mitochondria-dependent apoptosis without membrane potential loss, there was a possibility that the JC-1 staining was not sensitive enough to detect changes in mitochondrial membrane potential in BU-11 cells. However, etoposide (0.1 μg/ml), a compound known to induce mitochondrial membrane potential loss (40, 41), induced significant mitochondrial membrane potential loss within 4 h of treatment that continued to increase through 6 h (Fig. 1B, inset). These results support the conclusion that MOMP occurs without causing a loss of mitochondrial membrane potential in DMBA/stromal cell-induced B cell apoptosis.

On the basis of the facts that DMBA-induced activation of caspase-8 is caspase-dependent (30) and that cytochrome c release from the mitochondria occurs early in this pathway, it seemed likely that MOMP takes place before the caspase cascade is activated. To test this hypothesis, BU-11/BMS2 cell cocultures were treated with vehicle or the pan-caspase inhibitor VAD-FMK (30 μM) prior to treatment with vehicle or DMBA. BU-11 cells then were harvested, and cytoplasmic fractions were analyzed for cytochrome c release from the mitochondria by immunoblotting. Cytochrome c was released from the mitochondria of B cells in DMBA-treated cocultures, and this release was not blocked by pretreatment with VAD-FMK (Fig. 1C). Therefore, MOMP appears to precede activation of the caspase cascade in the DMBA/stromal cell-induced apoptosis pathway.

Having determined that mitochondrial permeabilization is likely caspase-independent, the contribution of caspases downstream of cytochrome c release was investigated. BU-11/BMS2 cell cocultures were treated with vehicle or DMBA, and BU-11 cells were analyzed for caspase-9 activation by immunoblotting for cleaved, active caspase-9 and for caspase-9 catalytic activity. An increase in the formation of active caspase-9 fragments in BU-11 cells occurred 4–6 h after DMBA treatment of the coculture (Fig. 2A). Concomitantly, an increase in caspase-9-like activity was detected 6 h after DMBA treatment and reached statistical significance at 8 h (Fig. 2B).

To demonstrate a requirement for caspase-9 activation in the DMBA-induced death pathway, two approaches were taken. First, experiments with Z-Leu-Glu(OMe)-His-Asp(OMe)-fluoromethylketone, a caspase-9 inhibitor, showed that DMBA/stromal cell-induced B cell death was at least partially sensitive to inhibition of caspase-9 (data not shown). Second, primary pro-B cells with mutant apoptosis peptidase activating factor-1 (Apaf-1) were tested for sensitivity to DMBA/stromal-induced apoptosis. Apaf-1 is an integral component of the apoptosome and is essential for maximal activation of caspase-9 (42). Primary pro-B cells were generated from the bone marrow of mice hetero- or homozygous for the Apaf-1^fog mutation. Pro-B/BMS2 cell cocultures were treated with vehicle or DMBA, and pro-B cells were analyzed for apoptosis by hypotonic PI staining and flow cytometry. DMBA induced a significant increase in apoptosis of pro-B cells from mice heterozygous for the Apaf-1^fog mutation, but not of pro-B cells from mice homozygous for the Apaf-1^fog mutation (Fig. 2C). These results indicate that MOMP initiates assembly of the apoptosome, which results in caspase-9 activation and is required for DMBA/stromal cell-induced B cell apoptosis.

MOMP can be amplified by caspase-dependent positive feedback loops. Indeed, caspase-8 and Bid are cleaved to their active forms in DMBA/stromal cell-induced B cell apoptosis (30). Bid is a BH3-only, direct-activator, proapoptotic Bcl-2 family member that is cleaved and activated by a number of caspases leading to its translocation to the mitochondria where it stimulates Bax/Bak pore formation (43). Bid can be cleaved directly by caspase-2, caspase-3, and caspase-8 and also may be activated through a caspase-3 to caspase-6 to caspase-8 to Bid loop (44–49). The likely intermediate in the activation of caspase-8 by a caspase-3 feedback mechanism, caspase-6, was activated in B cells following treatment of stromal cell/B cell cocultures with DMBA (Supplemental Fig. 1A, 1B). Furthermore, although caspase-6 and caspase-8 activation was largely suppressed by caspase inhibitors, activation of caspase-3 was not (Supplemental Fig. 1C). Thus, the caspase-9 intrinsic pathway likely is amplified by activation of caspases that can cleave and activate Bid, further stimulating MOMP.

One of the most common activators of the intrinsic pathway is p53, a tumor suppressor protein that can trigger mitochondrial permeabilization through transcriptional as well as nontranscriptional mechanisms (50). p53 is a likely candidate for initiation of MOMP in DMBA/stromal cell-induced apoptosis in developing B cells because p53 plays a role in DMBA-induced bone marrow toxicity and immunosuppression in vivo (51, 52), and p53 protein levels are increased in the nuclei of BU-11 cells at least at a late time point in this system (19).

To begin to determine how p53 could be acting in DMBA/stromal cell-induced apoptosis, p53 protein expression was examined at early time points. BU-11/BMS2 cell cocultures were treated with vehicle or DMBA, and BU-11 whole-cell lysates were analyzed for p53 expression by immunoblotting. An increase in p53 protein was seen beginning 2–4 h after DMBA treatment and continuing through 10 h (Fig. 3A). An apparent decrease in p53 electrophoretic migration is likely due to posttranslational modifications, because p53 is known to be modified following cellular stress (53). The early time at which p53 protein levels were increased is consistent with p53 acting upstream of MOMP.

To evaluate the requirement for functional p53 in DMBA-induced apoptosis, primary pro-B cells were generated from wild-type C57BL/6 mice (controls) and from congenic mice homozygous for a mutation in p53 (Trp53^tm1Tyj). Primary pro-B/BMS2 cell cocultures were treated with vehicle or DMBA, and pro-B cells were analyzed for apoptosis by hypotonic PI staining and flow cytometry and by immunoblotting for cleaved, active caspase-3. Both wild-type and p53 mutant pro-B cells underwent significant apoptosis in response to DMBA treatment as measured by PI staining; however, mutant pro-B cells underwent significantly less apoptosis than wild-type pro-B cells at 16 h (Fig. 3B). An increase in cleaved caspase-3 was detected in extracts from wild-type pro-B cells 4 h after DMBA treatment (Fig. 3C). In contrast, primary p53 mutant pro-B cells exhibited a low background level of caspase-3 cleavage, which did not increase over time in response to DMBA (Fig. 3C). These results indicate that p53 is essential for optimal DMBA/stromal cell-induced B cell apoptosis.

Results gathered to date, including those described above, indicate that DMBA induces bone marrow stromal cells to produce a death signal that is transferred to developing B cells, initiating a p53-dependent intrinsic apoptotic pathway. However, the nature of the stromal cell-derived apoptotic signal that activates the apoptotic pathway in the B cells and the mechanism of its delivery to the B cells remained unclear. A reactive metabolite of DMBA appeared likely to be the apoptotic signal in this system because DMBA metabolites are known to form DNA adducts (54), which would be predicted to lead to p53 upregulation and mitochondrial permeabilization. In addition, metabolism is known to be important in DMBA/stromal cell-induced apoptosis of developing B cells (17, 22).

To assess the ability of a terminal reactive DMBA metabolite to induce developing B cell apoptosis directly, bypassing the need for stromal cells, BU-11 cells cultured in the absence of BMS2 cells were treated with vehicle or a terminal reactive DMBA metabolite, DMBA-DE (1 μM), and apoptosis was assessed by hypotonic PI staining and flow cytometry. A significant increase in apoptosis was observed with DMBA-DE treatment at all of the time points tested, including the relatively early time point of 4 h (Fig. 4A). Note that this is considerably earlier than the point at which BU-11 cell apoptosis is seen in DMBA-treated cocultures (i.e., 10 h) (30). The ability of this terminal reactive metabolite to rapidly induce apoptosis in BU-11 cells while bypassing the need for stromal cells is consistent with the hypothesis that stromal cells generate and deliver this putative mediator of apoptosis induction to adjacent developing B cells.

If DMBA-DE induces developing B cell apoptosis in DMBA-treated BU-11/BMS2 cell cocultures, then it would be predicted that key hallmarks of DMBA-induced apoptosis, such as p53 induction, would be manifested after treatment of BU-11 cells with DMBA-DE. Furthermore, by bypassing the requirement for DMBA metabolism in stromal cells and the subsequent transfer of reactive metabolite to B cells, markers of apoptosis would be expected to occur at earlier time points than when treating stromal cell/B cell cocultures with parent compound (i.e., DMBA). To test these predictions, first BU-11 cells cultured in the absence of stromal cells were treated with vehicle or DMBA-DE and p53 expression was analyzed by immunoblotting. Expression of p53 protein increased within 30 min to 1 h of DMBA-DE treatment (Fig. 4B). Second, primary pro-B cells from wild-type C57BL/6 and congenic Trp53^tm1Tyj mice were treated with vehicle or DMBA-DE (0.01 μM) for 16 h, and the B cells were analyzed for apoptosis by hypotonic PI staining and flow cytometry. DMBA-DE induced significant apoptosis in primary pro-B cells from wild-type mice, but not in pro-B cells from p53 mutant mice (Fig. 4C). A similar resistance to apoptosis was seen with pro-B cells from Apaf-1^fog mutant mice (46 ± 12% in Apaf-1^WT/Mut versus 6 ± 6% in Apaf-1^Mut/Mut pro-B cells treated with 0.1–0.001 μM DMBA-DE, p < 0.05, ANOVA, Tukey-Kramer). The induction of apoptosis with DMBA-DE in a shorter time frame than when treating stromal cell/B cell cocultures with DMBA and the suppression of apoptosis in p53 and Apaf-1 mutant B cells are consistent with the hypothesis that DMBA-DE initiates the DMBA/stromal cell-induced apoptosis cascade in developing B cells.

To begin to determine the requirements for transfer of the putative effector metabolite from stromal to developing B cells, the requirement for cell–cell contact was evaluated. BU-11 cells either were plated on BMS2 cell monolayers or were cultured in the upper chambers of Transwell inserts with the BMS2 cells maintained in the lower chambers. The membranes of these Transwells allow free transfer of anything ≤3 μm in diameter between upper and lower chambers, which would include various types of vesicles, including exosomes (55). These cultures then were treated with vehicle or DMBA for 16 h, and apoptosis of BU-11 cells was analyzed by hypotonic PI staining. A significant percentage of the BU-11 cells underwent apoptosis when they were allowed to contact the BMS2 cells directly but not when they were cultured above the BMS2 cells in Transwells (Fig. 5A). This result implies that cell–cell contact is required for apoptosis. However, it is formally possible that a DMBA effector metabolite is carried from the BMS2 cells to the BU-11 cells within vesicles and that the failure to observe BU-11 cell apoptosis in the experiment described above reflected the inability of the vesicles to migrate up through the membrane to the BU-11 cells. Therefore, the experiment was repeated with BMS2 cells in the upper chamber and BU-11 cells in the lower chamber. The failure of DMBA to induce apoptosis in this protocol (Fig. 5A) further argues against the possibility of the death signal being delivered by a vesicle and is consistent with the hypothesis that BU-11/BMS2 cell contact is required for delivery of the death signal. Additional studies in which BU-11 cells were treated with DMBA-treated BMS2-derived vesicles similarly failed to implicate vesicles in the transfer of the apoptotic signal (data not shown).

Taking advantage of the fact that DMBA and its metabolites are weakly fluorescent, transfer of DMBA and/or its metabolites from BMS2 to BU-11 cells was investigated by two-photon microscopy and flow cytometry. It should be noted that DMBA and DMBA-DE are excited and emit fluorescence under identical conditions, so it was not possible to differentiate between DMBA and its metabolite in these experiments (although even if DMBA were to be transferred to the B cells it cannot induce apoptosis). BMS2 cells were grown on one half of a cover glass. The following day, the whole cover glass was treated with DMBA (10 μM) and excess DMBA/metabolites were washed off. BU-11 cells then were centrifuged onto the dish, and analysis of adherent BU-11 cells by two-photon microscopy was begun within 15 min. BU-11 cells that were in direct contact with BMS2 cells at the time of observation appeared to have a higher fluorescence than BU-11 cells clearly not in contact with BMS2 cells (Fig. 5B–D). Statistical analysis of the fluorescent intensity of BU-11 cells in contact with BMS2 cells as compared with the fluorescence of BU-11 cells not in contact with BMS2 cells showed a trend toward significance (p = 0.06, Student t test).

For a more rigorous quantification of transfer, BU-11 cells were analyzed for DMBA/metabolite uptake by flow cytometry. BMS2 cells were treated with vehicle or DMBA (10 μM) and were washed to remove excess DMBA/metabolites. BU-11 cells were added either to the stromal cell monolayer directly or to the upper chamber of Transwells over the BMS2 cells. BU-11 cells that were allowed to contact BMS2 cells directly exhibited a statistically significant increase in fluorescence consistent with DMBA and/or DMBA metabolite uptake (Fig. 5E, 5F). In contrast, there was no significant increase in the fluorescence of BU-11 cells that were separated from BMS2 cells by Transwell membranes (Fig. 5E, 5F). These results suggest that DMBA and/or its metabolites can be transferred from BMS2 cells to BU-11 cells in a contact-dependent manner.

Trogocytosis, acquisition of membrane-anchored Ag by intercellular membrane exchange, is a mechanism commonly used by cells of the immune system, including B cells (56). If lipophilic metabolites of DMBA are transferred from BMS2 cells to BU-11 cells, then one mechanism for that transport may be exchange through the BMS2 cell plasma membrane. To test this hypothesis, BMS2 cells were labeled with the lipophilic plasma membrane dye Vybrant DiO before BU-11 cells were added either onto BMS2 cell monolayers directly or into Transwells above the BMS2 cell cultures. After 16 h, BU-11 cells then were analyzed for DiO uptake by flow cytometry. In addition, BU-11 cells maintained on unstained BMS2 cells were resuspended in supernatants from DiO-stained BMS2 cells for 20 min before analysis. A large and significant proportion of BU-11 cells took up DiO when they were allowed to contact the stained BMS2 cells directly (Fig. 6A). However, there was no significant uptake of DiO when the cells were separated by a permeable barrier or when BU-11 cells were exposed to supernatants from stained cells (Fig. 6A). If membrane transfer delivers the apoptotic metabolite, then it would be expected that membrane exchange would occur prior to overt apoptosis. To test this prediction, the kinetics of membrane transfer was determined. BMS2 cells were labeled with DiO as described above. BU-11 cells were added to the culture, harvested over time (0.5–2 h), and analyzed for DiO uptake by flow cytometry. Significant membrane exchange occurred within 2 h of contact (Fig. 6B), prior to significant changes in the earliest markers of apoptosis (cytochrome c release, Fig. 1; increased p53 expression, Fig. 3). These experiments demonstrate, for the first time, the transfer of plasma membrane components from bone marrow stromal cells, on which developing B cells depend for growth and differentiation signals, to associated bone marrow-derived B cells. The results also are consistent with the hypothesis that lipophilic metabolites, embedded in cell membranes, may be transferred from bone marrow stromal cells to developing B cells.

To determine if this membrane component transfer is unique to the BU-11 cell line, primary pro-B cells, splenic B cells, or splenic T cells were cocultured with BMS2 cells and assessed for dye uptake by flow cytometry. Primary pro-B cells, which undergo apoptosis when treated with DMBA in coculture with BMS2 cells in a fashion similar to BU-11 cells (30), also behaved similarly to BU-11 cells in this assay, taking up membrane from BMS2 cells (Fig. 6C, 6D). Interestingly, splenic B cells acquired BMS2 membrane components when they were allowed to contact BMS2 cells, similar to the developing B cells, whereas splenic T cells did not acquire appreciable levels of membrane dye (Fig. 6C, 6D). These results suggest that membrane uptake is not unique to the BU-11 cell line but rather that it is a property of B cells throughout development and that transfer of the DMBA-induced apoptotic signal from bone marrow stromal cells to developing B cells may occur through membrane uptake.

Previous studies have shown that DMBA induces significant immunosuppression (5) by reducing cellularity in the bone marrow (14, 15) as well as by suppressing spleen cell numbers and function (57, 58). The reduced cellularity in bone marrow is caused largely by a loss of B cells (20), and loss of these developing B cells could result in skewing of the B cell repertoire toward apoptosis-resistant, autoreactive clones and/or a diminution in the production of competent mature B cell populations. B cell death requires contact with AhR- and CYP1B1-expressing stromal cells (21, 22), suggesting that a metabolite of DMBA is the death factor transferred from the stromal cells to the B cells. In this study, we addressed the potential for a terminal DMBA metabolite to induce B cell death and investigated the mechanism of transfer of this molecule between its source, the stromal cells, and its target, the B cells.

To identify a key component of the DMBA/stromal cell apoptosis pathway that could be used as a point of reference in the study of the DMBA-DE apoptosis pathway, the DMBA/stromal cell pathway was examined. Several key observations support the conclusion that MOMP is the initiating event in DMBA/stromal cell-induced B cell apoptosis. 1) Cytochrome c release from the mitochondria was evident at early time points, 2) the general caspase-inhibitor, VAD-FMK, did not inhibit DMBA-induced cytochrome c release, indicating that cytochrome c release occurs upstream of caspase activation, 3) DMBA induces the formation of catalytically active caspase-9, and 4) DMBA-induced apoptosis was suppressed in Apaf-1^fog primary pro-B cells, cells that are unable to form apoptosomes required for caspase-9 activation (42, 59). Caspase-8 activation appears to occur as a caspase-dependent downstream event, potentially participating in a positive feedback loop (45, 47, 60).

Unlike many intrinsic pathway-mediated apoptotic programs, mitochondrial membrane potential was not lost in B cells after DMBA treatment of the coculture. This is uncommon but not without precedent (38, 39). In fact, apoptosis induced by a terminal metabolite of B[a]P in a lung cancer cell line also involves cytochrome c release without ΔΨ_m loss (61). Although the lack of ΔΨ_m loss in this system is somewhat unusual for an intrinsic apoptotic pathway, the presence or absence of ΔΨ_m loss may not be important to apoptotic pathways in general. The previously held view that ΔΨ_m loss mediates apoptosis has been challenged (62–65). Cells from mice lacking cyclophilin D, a protein essential for the mitochondrial permeability transition pore complex responsible for ΔΨ_m loss, do not undergo mitochondrial membrane potential loss but still undergo apoptosis in response to etoposide, x-ray irradiation, staurosporine, and TNF-α (62, 66–68). However, although loss of ΔΨ_m may not initiate apoptosis, it may enhance it by disruption of the mitochondrial christae, thereby enhancing cytochrome c release (65). In light of these data, we conclude that, although the lack of mitochondrial membrane potential loss in DMBA-induced developing B cell apoptosis is curious, it is functionally irrelevant.

Once identification of an intrinsic apoptosis pathway was established, we investigated potential key protein mediators responsible for initiating MOMP. The proapoptotic proteins Bax and Bak are responsible for pore formation leading to MOMP (69). Multiple signaling proteins may trigger the movement of Bax/Bak to the mitochondria to initiate MOMP, including p53 (70, 71), Bid and Bim (43), and JNK (72). We have shown that Bid is cleaved to its active form during DMBA-induced apoptosis (21). However, apoptosis is reduced only ∼20% in Bid knockout primary pro-B cells (data not shown), supporting the conclusion that it participates in an amplification loop rather than acts as an initiator of the death pathway. p53 activates apoptosis through two distinct mechanisms; 1) it engages Bcl-2 family members at the mitochondrial membrane resulting in neutralization of antiapoptotic proteins (e.g., Bcl-2) or activation of proapoptotic proteins (e.g., Bax), stimulating MOMP to occur and 2) it transcriptionally upregulates proapoptotic Bcl-2 family members (e.g., PUMA, Noxa, and Bax) (70, 71). Previous studies have shown that DMBA-induced bone marrow toxicity and immunosuppression are reduced in the presence of mutant p53 (Trp53^tm1Tyj) in vivo (51, 52). In this study, we demonstrate that p53 expression is upregulated in B cells exposed to DMBA in the coculture system and that pro-B cells with mutant p53 are largely resistant to DMBA/stromal cell-induced apoptosis. Thus, p53-mediated mitochondrial apoptosis is a likely pathway for DMBA/stromal cell-induced apoptosis.

Treatment of B cells alone with DMBA-DE results in apoptosis coincident with p53 induction, supporting the hypothesis that the stromal cell-derived death factor is indeed the terminal DMBA metabolite. Similarly, a terminal metabolite of B[a]P, B[a]P-7,8-dihydrodiol-9,10-epoxide, also can directly induce B cell death (data not shown). Consistent with this hypothesis are results from a previous study demonstrating that DMBA/stromal cell-induced apoptosis requires CYP1B1 and that DMBA-diol-epoxide adducts are formed in the B cells (22). Interestingly, DMBA-DE adducts also form in the stromal cells (22); however, DMBA is not toxic to these cells. Developing B cells are known to be “ultrasensitive” to apoptotic stimuli, and this may result from the fact that these cells express very low levels of the antiapoptotic protein Bcl-2 (73). Furthermore, DMBA-induced death is associated with the downregulation of NF-κB, a lymphocyte survival factor, another factor that may sensitize B cells to DMBA-induced death in particular (16).

We hypothesized that the highly labile and reactive nature of DMBA-DE would require a specialized delivery mechanism between stromal and B cells. Indeed, the stromal cell and B cell must be in direct contact for the death factor to be transferred from one cell to the other, resulting in B cell apoptosis. Trogocytosis (i.e., membrane transfer) is a common but underappreciated mechanism of information transfer between cells in the immune system (31). Thus, membrane transfer was considered as a possible mechanism for the delivery of otherwise labile DMBA-DE to the B cells. Consistent with this hypothesis, DMBA and/or its metabolites and membranes were transferred between stromal cells and B cells. Interestingly, bone marrow and splenic B cells, but not splenic T cells, efficiently incorporated membranes from bone marrow stromal cells. This may reflect the fact that B cells are capable of performing trogocytosis without the support of intracellular signaling or active processes, whereas T cell trogocytosis is an active, TCR signal-dependent process (56, 74). Typically [although not always (32)] trogocytosis is thought to be triggered by membrane Ag as a mechanism of Ag capture (56). Membrane transfer between stromal cells and bone marrow B cells may represent a more general mechanism of communication.

Results from these studies show, for the first time, that a terminal DMBA metabolite can initiate apoptosis directly in B cells and can initiate an apoptotic pathway that shares features of DMBA/stromal cell apoptosis, consistent with the conclusion that DMBA-DE is the stromal cell-derived death factor that induces B cell death in stromal cell/B cell cocultures and, presumably, in vivo. A trogocytosis-like mechanism is likely the way in which the stromal cells transfer such a highly labile and reactive molecule. In addition, these studies provide support for the novel hypothesis that DMBA, via its reactive metabolite, induces a p53-intrinsic apoptosis pathway. Further studies are required to determine the mechanisms of p53 activation and mitochondrial interactions in DMBA/stromal-induced B cell apoptosis.

Technical assistance was provided by the Boston University Medical Campus Flow Cytometry Core Facility.

Disclosures The authors have no financial conflicts of interest.