Apoptotic human cells inhibit migration of granulocytes via release of lactoferrin (original) (raw)
Apoptotic cells actively produce factor(s) that inhibit neutrophil chemotaxis. To address whether apoptotic cells influence migratory activity of neutrophils, we carried out a series of in vitro Boyden-type chemotaxis assays to investigate neutrophil migration toward Burkitt lymphoma (BL) cells. We initially employed BL as a model tissue as these tumor cell populations display high levels of apoptosis, a property that is retained constitutively in the tumor-derived cell lines. As at all sites of apoptosis, there is marked infiltration of macrophages that engulf the apoptotic cells, giving rise to the typical “starry sky” histological appearance of this tumor. As shown in Figure 1A (left), while macrophages were abundant in histological sections of BL, neutrophils were absent. We assessed the effects of BL cells on the migratory activity of neutrophils in vitro by adding neutrophils to the top compartment of a Transwell filter and inducing them to migrate toward the lower chamber containing BL cells in the presence of the powerful neutrophil chemoattractant formyl-methionyl-leucyl-phenylalanine (fMLP) (Figure 1B). As shown in Figure 1C, neutrophil migration was significantly inhibited in BL cells in a concentration-dependent manner. We observed a similar effect irrespective of the chemoattractant used (inhibition of neutrophil migration induced by C5a, IL-8, and leukotriene B4 [LTB4]; data not shown, but see below). We carried out subsequent chemotaxis assays using BL-conditioned medium obtained over a 7-hour time course and found that BL cells actively released an inhibitory factor(s) (Figure 1D). The release of the inhibitory factor appeared to be linked to the levels of apoptosis in the BL cell populations, since the inhibitory activity was significantly lower in further chemotaxis assays using BL-conditioned medium derived from cells overexpressing the apoptosis inhibitor Bcl-2, as compared with that of parental cells (Figure 1E).
Apoptotic cells release factor(s) that inhibit neutrophil migration. (A) Immunohistochemical analysis of neutrophils in BL (left) and spleen (positive control; right) sections. Inset images represent isotype control. (B) Representative images of stained Transwell filters. (C) Neutrophil chemotaxis toward increasing concentrations of BL cells was assessed in the presence of fMLP (100 nM). n = 3; *P < 0.05 vs. time 0. (D) BL cell–conditioned media obtained at the indicated time points were used to analyze fMLP-induced neutrophil chemotaxis. n = 3; *P < 0.05 vs. fMLP. (E) Neutrophil chemotaxis toward fMLP was analyzed in the presence of control or Bcl-2–transfected BL2 cells obtained following a 0- and 5-hour incubation at 37°C. n = 3; *P < 0.05 vs. BL2 0 h; NS vs. BL2/Bcl-2 0 h. Apoptosis levels were assessed by flow cytometry following staining with annexin V/propidium iodide. Error bars indicate SEM. Original magnification; ×400 (A; A, insets; B). hpf, high-power field.
Biochemical characterization of neutrophil migration–inhibitory factor(s). In an attempt to gain further insight into the biochemical nature of the factor(s) that BL cells secrete in order to exclude neutrophils from their environment, we initially estimated the molecular weight range of the inhibitory factor(s) by using filters with molecular weight cutoff points of approximately 3, 10, 30, 50, and 100 kDa. BL-conditioned media obtained after 24-hour incubation were fractionated, and each fraction was examined in vitro using the neutrophil chemotaxis assay described above. The results revealed that fractions containing molecules of less than 50 kDa failed to display any inhibitory effect on neutrophil migration (Figure 2A). However, the use of 100 kDa filters revealed that both fractions (>100 kDa and <100 kDa) displayed an inhibitory effect on neutrophil migration, indicating that at least 1 factor has a molecular weight that ranges between 50 and 100 kDa. The presence of inhibitory activity in the filtrate of the 100 kDa cutoff membrane is likely to result from (a) imprecise molecular weight cutoff of molecules in the 50–100 kDa range, (b) complex formation through multimerization of the 50–100 kDa factor or through interaction with other molecules, or (c) the existence of a distinct inhibitory activity of greater than 100 kDa. It should be noted that the selected isolation approach is skewed in favor of proteins and that additional low-molecular-weight (for example, lipid) mediators of neutrophil migration inhibition would not be identified by these procedures.
Biochemical characterization of the inhibitory factor. Conditioned media from BL2 cells cultured for 24 hours were size fractionated using filters with 50 kDa (A) molecular weight cutoff sizes. Unfiltered medium was included as control. *P < 0.001 compared with the corresponding positive control. Error bars indicate SEM. Ion exchange analysis included the use of Q Sepharose beads (positively charged) in order to distinguish positively and negatively charged molecules in the <100 kDa fraction (B) of the BL medium. Unbound molecules (Q1 fraction) were collected, whereas bound molecules were eluted from the beads (Q2 fraction). Neutrophil migration toward these fractions in the presence of fMLP (100 nM) was assessed. Q1 and Q2 fractions (unbound and eluant fraction) of serum-free medium (no BL) were included as control. †P < 0.05 compared with the corresponding control. Error bars indicate SEM. (C) Chemotaxis assay of neutrophils toward BL-conditioned medium that was heat inactivated (90°C for 10 minutes). (D) MALDI-TOF mass spectrum for the tryptic digest of the peptide band identified as lactoferrin.
To investigate further the biochemical properties of the retentate and filtrate of the 100 kDa cutoff membrane, we first determined the charge (pI value) of the migration-inhibitory activity by means of an ion exchange analysis of BL-conditioned media. Using positively charged ion exchange beads (Q Sepharose), the 100 kDa membrane retentate and filtrate were separated into positively charged (Q-supernatant) and negatively charged (Q-eluant) fractions. As shown in Figure 2B, the filtrate of the 100 kDa membrane (<100 kDa fraction) contained migration-inhibitory activity in both the supernatant (positive charge) and the eluant (negative charge) of the Q beads. By contrast, analyses of the retentate (>100 kDa fraction) revealed that only the negatively charged eluant displayed significant activity in inhibiting neutrophil migration. These results indicate that at least 2 moieties with neutrophil migration–inhibitory activity were present in BL cell–conditioned medium: one of 50–100 kDa with positive pI and a second of 100 kDa or more and negatively charged.
We subsequently determined whether the neutrophil migration–inhibitory activity was heat labile. As shown in Figure 2C, heat inactivation completely abrogated all chemotaxis inhibitory activity in BL cell–conditioned medium, suggesting that the inhibitory factor(s) were most likely protein in nature. We then analyzed the proteins released from BL cells in viable and apoptotic states by protein fingerprinting. Polypeptide bands of greater than 50 kDa were excised from a 10% SDS polyacrylamide gel. Tryptic peptides were gel extracted, and matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) mass spectrometric analysis was carried out (Figure 2D). Also, given the crude biochemical characteristics described above, we undertook a candidate approach based on the proteins released from apoptotic BL cells. We identified the factor released by BL cells that prevented neutrophil chemotaxis as lactoferrin.
Lactoferrin specifically inhibits neutrophil chemotaxis toward a range of chemoattractants. Lactoferrin is a glycoprotein of approximately 75–80 kDa that belongs to the transferrin family of proteins due to its iron-binding properties. It is a well-characterized component of neutrophil secondary granules, lacrimal fluid, colostrum, saliva, and mucosal secretions, in which it confers antibacterial activity. We observed that addition of anti-lactoferrin antibody to BL-conditioned medium neutralized its neutrophil migration–inhibitory activity using either polyclonal or monoclonal antibodies (Figure 3A and Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI36226DS1). Similar results were observed with supernatants obtained from the mammary carcinoma line MCF-7, indicating that the neutrophil migration–inhibitory activity is not restricted to BL cell–derived lactoferrin (Figure 3B). Furthermore, lactoferrin purified from human milk displayed dose-dependent inhibitory activity toward neutrophil migration in response to fMLP (Figure 3E) and also inhibited migration toward C5a, IL-8, and LTB4 to similar levels (Figure 4A). The neutrophil migration–inhibitory effect was also displayed by lactoferrin purified from human neutrophils (Figure 4D). It should be noted that both types of purified lactoferrin used in this study were free of endotoxin contamination, and the observed inhibitory effect did not appear to be due to any lactoferrin-associated molecules such as LPS. Furthermore, inhibition of lactoferrin expression by shRNA in BL cells provides additional support for the specificity of the observed lactoferrin effect. Thus, we found that neutrophil chemotaxis toward supernatants obtained from BL2 cells transfected with shRNA vectors targeted against lactoferrin was higher compared with supernatants from BL control or mock-transfected cells (Figures 3, C and D). Lactoferrin exerted no toxic effects on neutrophils, as assessed by annexin V/propidium iodide staining of control and lactoferrin-treated neutrophils (>98% cell viability). These results suggested that lactoferrin binds to neutrophils and inhibits their ability to undergo chemotaxis. To exclude the possibility that the observed inhibitory activity of lactoferrin was due instead to its ability to bind and functionally neutralize the chemoattractants, additional chemotaxis assays were performed using chemoattractants (fMLP, C5a, IL-8) that were preabsorbed with lactoferrin. To achieve this, we preincubated chemoattractants with lactoferrin. Subsequently, anti-lactoferrin antibody was used to remove the lactoferrin with the aid of magnetic beads. As shown in Figure 4B, no difference in neutrophil chemotactic activity was observed between the control and lactoferrin-absorbed chemoattractants, which excludes the possibility that lactoferrin binds to, and alters the activity of, the chemoattractants. Further supporting our conclusion that lactoferrin exerts its inhibitory effects through binding to neutrophils, we also observed that purified lactoferrin can directly associate with neutrophils (Figure 4C). In addition, Scatchard binding analysis of 125I-labeled apolactoferrin indicated, in accordance with earlier findings (26, 27), that lactoferrin bound to neutrophils via 2 classes of receptor that differ in affinity and number of binding sites per cell. We determined the higher-affinity receptors to be expressed at a density of 9,100 ± 2,500 binding sites per cell, with an affinity of 350 ± 65 nM, and the lower-affinity receptors to be expressed at a density of 2.5 × 106 ± 0.7 × 106 per cell, with an affinity of 20 ± 10 μM (Supplemental Figure 2).
Lactoferrin specifically inhibits neutrophil chemotaxis. Neutrophil chemotaxis in the presence of human anti-lactoferrin (anti-LTF) polyclonal antibody (gray) or isotype control (black) using conditioned media from BL (A) and MCF7 (B) cells (A: n = 3, *P < 0.05 vs. isotype control, NS vs. fMLP anti-lactoferrin control; B; n = 3, †P < 0.001 vs. fMLP/isotype; NS vs. fMLP/anti-LTF. (C) RT-PCR analysis to assess lactoferrin expression in BL cells stably expressing LTF shRNA (LTF) cells and mock-transfected (mm) cells induced to become apoptotic (1 μM staurosporine; 37°C). (D) Chemotaxis assay to determine neutrophil migration toward supernatants obtained from control, LTF shRNA, and mock-transfected BL cells (n = 5; §P < 0.05 compared with mm shRNA control; **P < 0.05 compared with fMLP; NS compared with fMLP control). (E) Dose-response analysis of purified human lactoferrin. n = 3; ¶P < 0.05 vs. 0 g/ml purified LTF + fMLP. Error bars indicate SEM.
Neutrophil chemotaxis toward lactoferrin is irrespective of the chemoattractant used and its iron saturation status. (A) Neutrophil chemotaxis toward different chemoattractants. n = 3; *P < 0.05. (B) Neutrophil chemotaxis toward chemoattractants (control) or chemoattractants that were incubated with lactoferrin (10 μg/ml) followed by the addition of isotype or anti-lactoferrin monoclonal antibody (10 μg/ml). Antibodies were removed using magnetic IgG beads. n = 3; *P < 0.05, NS compared with chemoattractant control. (C) Immunoblot analysis of lysates of neutrophils incubated with or without biotinylated lactoferrin (10 μg/ml) at 37°C for 1 hour. (D) Neutrophil chemotaxis toward lactoferrin (10 μg/ml) purified from human neutrophils or human milk. **P < 0.001 vs. fMLP. C5a-induced monocyte (E) or macrophage (F) chemotaxis. (G) Neutrophil migration in the presence of lactoferrin (10 μg/ml) in the top or bottom compartment of the Transwell insert (n = 3; NS vs. corresponding +LTF controls). (H) Chemotaxis assay to determine neutrophil migration toward purified recombinant iron-depleted (Apo-), partially iron-saturated, and fully iron-saturated (Holo-) recombinant lactoferrin (10 μg/ml). Milk-purified lactoferrin and partially iron-saturated transferrin (TF; 10 μg/ml) were added as control. n = 4; †P < 0.001 compared with fMLP control. Error bars indicate SEM.
To determine whether the migration-inhibitory effects of lactoferrin were specific to neutrophils among professional phagocytes, we analyzed its effects on monocyte and macrophage migration in vitro. As shown in Figure 4, E and F, C5a-induced chemotaxis of mononuclear phagocytes was unimpaired by lactoferrin. We further assessed whether lactoferrin acted by inhibiting neutrophil migration or promoting neutrophil repulsion. In chemotaxis assays, in which we added lactoferrin to the upper chamber along with neutrophils, we observed inhibition of neutrophil migration toward fMLP and control medium (Figure 4G), suggesting that lactoferrin exerts a direct effect on neutrophils by inhibiting their migratory ability and not by forcing them to migrate in all directions away from the chemoattractant.
Neutrophil migration–inhibitory effects of lactoferrin are not related to its iron-binding properties and are demonstrable in vitro and in vivo. Iron and iron-associated molecules have been previously shown to play an important role in many immunomodulatory functions. Indeed, suppression of IL-1 release by monocytes is observed by purified iron-saturated lactoferrin, whereas an inhibition of GM colony-stimulating activity production by monocytes and macrophages correlated with the iron saturation status of lactoferrin (28–30). Therefore, we further examined whether differences in the iron saturation profile of lactoferrin affect its ability to inhibit neutrophil migration. Chemotaxis assays to determine neutrophil migration toward iron-depleted (apo-lactoferrin), partially iron-saturated, or fully iron-saturated (holo-lactoferrin) lactoferrin revealed that the level of iron saturation was not responsible for the observed inhibition in neutrophil migration (Figure 4H). Also, as lactoferrin belongs to the transferrin family of proteins sharing 74% sequence homology with transferrin (both of them are 80 kDa cationic iron-binding glycoproteins), we reasoned that, if the underlying neutrophil migration–inhibitory mechanism of lactoferrin was rooted in its ability to chelate iron, transferrin might show similar effects on neutrophil migration. To explore this possibility, we performed chemotaxis assays in which neutrophils were induced to migrate toward fMLP in the presence of partially iron-saturated transferrin. Our results showed that transferrin, unlike lactoferrin, had no effect on neutrophil chemotaxis, further supporting the conclusion that the observed neutrophil migration–inhibitory effect is lactoferrin specific and is unlikely to require iron-chelating activity (Figure 4H).
Having established the inhibitory effects of lactoferrin on neutrophil chemotaxis in vitro, we then used a murine peritonitis model to assess the effect of lactoferrin on leukocyte recruitment in vivo. Lactoferrin and transferrin were tested for their ability to affect thioglycollate-induced leukocyte recruitment to the peritoneal cavity. As shown in Figure 5, A and B, thioglycollate caused rapid recruitment of leukocytes compared with vehicle alone, and the recruited leukocytes were predominantly neutrophils (88%). In the presence of lactoferrin, the total number of neutrophils recruited to the peritoneal cavity was reduced by 52% compared with control, whereas transferrin had no effect. Lactoferrin reduced specifically the proportion and number of neutrophils migrating into the cavity but did not affect recruitment of other types of leukocytes in response to thioglycollate (Figure 5C). These results demonstrate that, similar to its effect on neutrophil chemoattraction in vitro, lactoferrin is a potent inhibitor of neutrophil migration in vivo.
Lactoferrin inhibits neutrophil migration in vivo. Total cell (A) or neutrophil number (Gr-1+; B) obtained from peritoneal lavage. *P < 0.05 vs. transferrin; †P < 0.05 vs. thioglycollate (TG) control, **P < 0.01 vs. transferrin control. Error bars indicate SEM. (C) Characteristic cytospin images. Original magnification, ×400, top; ×200, bottom.
Impairment of neutrophil activation profile by lactoferrin. Neutrophil migration involves activation, adhesion, and extravasation processes that are accompanied by gross changes in cell morphology: whereas nonactivated neutrophils are rounded, activated neutrophils acquire a polarized morphology with spreading and adhesion to the available substratum (15). In order to initially assess the effects of lactoferrin on neutrophil activation, we performed time-lapse video microscopy of neutrophils and recorded directly their activation morphology, cell spreading, and locomotion. During a 1-hour time course, lactoferrin-pretreated neutrophil populations stimulated with fMLP displayed a greater proportion of nonadherent cells as well as cells presenting a rounded, nonactivated morphology as compared with neutrophils treated with fMLP alone (Figure 6, A and B). These quantitative differences between lactoferrin-treated and untreated neutrophils stimulated with fMLP were reflected in the locomotion of the cells around the substratum, with lactoferrin-treated cells displaying markedly reduced movement.
Effect of lactoferrin on neutrophil polarization morphology and spreading. (A) Time-lapse video microscopy frames of control or lactoferrin-pretreated neutrophils (10 μg/ml; 40 minutes at 37°C) stimulated with 1 μM fMLP over a 1-hour incubation time course. Original magnification, ×400. (B) Quantification of neutrophils (nonpolarized) counted from 5 different fields; *P < 0.05, **P < 0.01 vs. corresponding +LTF control. Error bars indicate SEM. (C) Representative plot (of 3 independent experiments) showing measurement of [Ca2+]i levels in neutrophils incubated in the presence or absence of lactoferrin (10 μg/ml; 30 minutes at 37°C) followed by stimulation with 10 nM fMLP.
Changes in cell morphology following stimulation with fMLP or other neutrophil agonists are characterized by a rapid increase in intracellular cytoplasmic calcium levels through mobilization of calcium from ER stores and activation of calcium influx channels of the plasma membrane, mediated by the inositol triphosphate (IP3) and diacylglycerol/PLC (DAG/PLC) pathways (31–33). In order to determine whether the observed cell shape alterations following lactoferrin treatment are related to changes in intracellular calcium concentrations ([Ca2+]i), we measured the levels of [Ca2+]i in control and lactoferrin-treated cells in response to fMLP stimulation (1 nM or 10 nM). No changes were observed in the fMLP-mediated [Ca2+]i response between control and lactoferrin-treated neutrophils, suggesting that lactoferrin acts downstream or independently of the mechanisms involved in intracellular calcium flux (Figure 6C).
As lactoferrin was shown to prevent neutrophil migration, we next explored whether it could also affect the neutrophil activation state. To this end, we chose to measure the expression of 2 known neutrophil activation–associated markers, CD62L (L-selectin) and CD11b, using 2-color flow cytometry. Upon activation, CD62L is cleaved from the neutrophil surface, whereas CD11b expression is upregulated following translocation from cytoplasmic granules to the cell membrane. Freshly isolated neutrophils were pretreated with lactoferrin and then exposed to the activation stimuli fMLP, TNF-α, and PMA. As shown in Figure 7, A–D, we found that, in lactoferrin-treated neutrophils compared with control cells, CD62L expression was significantly higher, whereas CD11b levels were lower. These effects were common to all activation stimuli used. Transferrin-treated neutrophils were also included but showed no significant differences compared with control cells. It is noteworthy that the lactoferrin effect was also evident when PMA, a specific PKC activator, was used as an agonist, indicating that lactoferrin acts downstream of PKC and not on pathways involved in PKC activation and Ca2+i responses, such as the IP3 and DAG/PLC pathways. This finding prompted us to investigate putative downstream targets of PKC involved in the late signaling cascades following neutrophil activation that also regulate cell motility and actin reorganization. Such cascades involve the activation of MAP family kinases (34), and we therefore examined the phosphorylation of p44/42 (ERK1 and ERK2) MAPKs. Whereas in untreated neutrophils, ERK1 and ERK2 were phosphorylated following fMLP stimulation (100 nM), lower levels of phosphorylated ERK1/2 were observed in neutrophils that had been pretreated with lactoferrin prior to stimulation with fMLP (Figure 7E). Collectively, these data suggest that lactoferrin has a clear impact on neutrophil activation, including impairment of neutrophil degranulation, inhibition of expression of β2 integrins, and reduction of activation of intracellular kinases, with profound effects on cell migration and motility.
Effect of lactoferrin on neutrophil activation status. The expression of CD62L (A and B) and CD11b (C and D) was assessed in fMLP- (100 nM), TNF-α– (1 ng ml), or PMA-stimulated (100 nM) neutrophils (30 minutes at 37°C) that were preincubated (40 minutes at 37°C) in the presence or absence of lactoferrin (10 μg/ml). Representative flow cytometry overlays of CD62L (A) and CD11b (C) expression in control (gray) and stimulated neutrophils (lactoferrin-treated: red; untreated: blue). n = 3; *P < 0.05, **P < 0.01. Error bars indicate SEM. (E) Western blot analysis to determine levels of ERK1/2 phosphorylation. Neutrophils were incubated with lactoferrin (10 μg/ml; 40 minutes at 37°C), followed by stimulation with fMLP (100 nM) for the indicated times. Membrane was stripped and reprobed for total ERK2. Results are representative of 3 independent experiments.
Induction of apoptosis upregulates lactoferrin expression and release in diverse cell types. Pursuing our initial hypothesis, which was strengthened by early observations that inhibition of neutrophil migration by BL cells appeared to be correlated with BL cell apoptosis (Figure 1E), we assessed lactoferrin expression following induction of apoptosis in a panel of cells of diverse lineages. By transcriptional analysis using RT-PCR, we found that lactoferrin was expressed, as reported previously (35), by MCF-7 mammary epithelial cells in their viable state but not by Jurkat, BL2, or A549 cells. Upon apoptosis induction, lactoferrin expression was upregulated in MCF-7 cells and expressed de novo in Jurkat, BL2, and A549 cells (Figure 8). More specifically, lactoferrin was transcribed de novo early after induction of apoptosis in A549 cells by either 100 nM etoposide or 1 μM staurosporine (Figure 8B). Levels of lactoferrin induced by etoposide were reduced in cells treated in the presence of the broad-spectrum caspase inhibitor zVAD-fmk, which prevented apoptosis induction (Figure 8C). The link between lactoferrin expression and apoptosis induction was further supported by the effects of the apoptosis inhibitor Bcl-2. BL cells expressing exogenous Bcl-2 that provided protection from apoptosis expressed lower levels of lactoferrin upon exposure to staurosporine than did their parental counterparts (Figure 8, A and D). Not only was apoptosis-related lactoferrin expression demonstrated at the transcriptional level, lactoferrin protein was also recovered from supernatants of cells undergoing apoptosis (Figure 8D). Treatment of A549 cells with brefeldin A, which interferes with intracellular transport of newly synthesized proteins, resulted in inhibition of apoptosis-induced lactoferrin release, providing further evidence for de novo synthesis and secretion of lactoferrin by cells undergoing apoptosis (Figure 8E). An analogous effect was also evident from supernatants obtained from primary lymphocytes induced to become apoptotic in the presence of 1 μM staurosporine. Finally, immunoblotting analyses and chemotaxis assays using supernatants of BL cells undergoing primary necrosis further revealed that the release of lactoferrin is not linked to necrosis but that lactoferrin is expressed and actively released from cells as a consequence of activation of their apoptosis program (Figure 8F and Supplemental Figure 3).
Induction of apoptosis upregulates lactoferrin expression and release. (A) RT-PCR analysis in cell lines stimulated to undergo apoptosis (A) and unstimulated controls (V). MCF7 cells transfected with caspase-3 (25.4% apoptosis; 100 μM etoposide, 20 hours), Jurkat (18.4% apoptosis; 1 μM staurosporine, 3 hours), BL2 (12.46% apoptosis), and BL2/Bcl-2 (7.42% apoptosis; 1 μM staurosporine, 1 hour). The lanes were run on the same gel but, where indicated by the vertical lines, were noncontiguous. (B) Lactoferrin expression in A549 cells at defined time points (hours) following stimulation with 100 μM etoposide or 1 μM staurosporine. (C) Addition of pan-caspase inhibitor zVAD-fmk (100 μg/ml) for 12 hours in order to prevent etoposide-induced apoptosis in A549 cells. (D) Immunoblot analysis of cell supernatants from: BL2 and primary lymphocytes in the presence (+) or absence (–) of staurosporine (1 μM) in serum-free conditions for 1 hour. A549 cells were stimulated with (+) or without (–) 100 μM etoposide for 5 hours. All samples were run on the same gel. Noncontiguous samples of A549 cells and lymphocytes (Lymph) are indicated by the vertical lines. (E) A549 cells were induced to become apoptotic (100 μM etoposide; 20 hours) in the presence or absence of brefeldin A (1 μg/ml), a protein release inhibitor. (F) Immunoblot analysis of cell supernatants from control BL2 cells (1 × 106/ml) induced to undergo apoptosis (1 μM staurosporine, 1 hour) or primary necrosis (56°C, 1 hour) in serum-free conditions. St, staurosporine; con; control; Etop, etoposide.