TGF-β is a potent inducer of apoptosis in many Burkitt’s lymphoma (BL) cell lines. In this study, we characterize this apoptotic process in the EBV-negative BL41 cell line. Induction of apoptosis was detected as early as 8 h after TGF-β treatment, as assayed by TUNEL and poly(ADP-ribose) polymerase cleavage. FACS analysis demonstrates that this proceeds predominately from the G1, but also from the G2/M phases of the cell cycle. We observed no early detectable changes in the steady-state levels of Bcl-2 and several of its family members after TGF-β treatment. We detected cleavage of caspases 2, 3, 7, 8, and 9 into their active subunits. Consistent with the involvement of these enzymes in TGF-β-mediated apoptosis, the broad spectrum caspase inhibitor benzyloxycarbonyl-Val-Ala-Asp(Ome)-flouromethylketone (ZVAD-fmk) blocked TGF-β-induced apoptosis and revealed a G1 arrest in treated cells. Use of specific caspase inhibitors revealed that the induction of apoptosis is caspase 8 dependent, but caspase 3 independent. Activation of caspase 8 has been shown to be a critical event in death receptor-mediated apoptosis. However, TGF-β treatment of BL41 cells was found not to affect the cell surface expression of Fas, TNF-R1, DR3, DR4, or DR5, or the steady-state expression levels of Fas ligand, TNF-R1, DR3, DR4, and DR5. Furthermore, blocking experiments indicated that TGF-β-mediated apoptosis is not dependent on Fas ligand, TNF-α, tumor necrosis-like apoptosis-inducing ligand, or TNF-like weak inducer of apoptosis signaling. Therefore, it appears that TGF-β induces apoptosis in BL cell lines via caspase 8 in a death receptor-independent fashion.
Apoptosis, or programmed cell death, is a vital cell suicide mechanism in the development and homeostasis of multicellular organisms. Apoptosis is characterized by several morphological and biochemical changes, which include membrane blebbing, chromatin condensation, fragmentation into membrane-enclosed vesicles, degradation of chromosomal DNA, and cleavage of a specific subset of proteins. These cleavage events are mediated by the so-called executioners of apoptosis that comprise a family of cysteine proteases called the caspases (1). There are at least 13 mammalian caspases, which have been subdivided into initiator and effector caspases. The caspases are normally present in cells as proenzymes, which can be activated by the formation of complexes and autocatalytic cleavage. They can also act as substrates for each other (1). The paradigm of caspase activation has come from the study of the interaction of the death receptor Fas (also called Apo-1 or CD95) and its ligand (FasL3/CD95L) (2). This interaction plays a crucial role in deletion of activated lymphocytes at the termination of an immune response and in the elimination of cancer and virally infected cells by cytotoxic lymphocytes. The FasL/Fas interaction results in the formation of a death-promoting complex, which results in the activation of caspase 8. This then triggers the activation of a caspase cascade, resulting in the activation of the effector caspases 3 and 7 (3). This sequence of events is also believed to be important in apoptosis induced by other death receptor/ligand interactions, including those of DR4 and DR5/Apo2L, TNF-R/TNF, and DR3/Apo3L (4). Nondeath receptor-mediated apoptosis is also believed to result in sequential caspase activation triggered by the initiator caspase 9 culminating in activation of caspases 3 and 7 (3). The activation of caspase 9 is mediated by the ATP-dependent complex formation with Apaf-1, which itself is triggered by the release of cytochrome c from mitochondria (5, 6). The Bcl-2 family of proteins is likely to be involved in controlling cytochrome c release in an as yet poorly defined fashion (7).
TGF-β1 is the prototypical member of a large family of pleiotropic cytokines that exert their effects on a wide variety of cell types both during development and in the adult organism (8, 9). The resulting effects of TGF-β1 are cell type and environment dependent and include both positive and negative effects on cell growth, differentiation, and matrix organization and biogenesis (8, 9).
It is becoming increasingly clear that TGF-β1 is an important immunomodulatory cytokine and has effects on and is produced by most types of cell in the immune system (10). Many previous studies have focused on the effect of TGF-β1 on T cells and how it modulates their growth and survival and Th cell subset development (10). The effects of TGF-β1 on human B cells have been less well characterized. TGF-β1 pretreatment of primary B cells has been demonstrated to block activation signal-induced proliferation (11, 12, 13, 14) and to inhibit Ig secretion (15, 16, 17), and may promote class switching to an IgA phenotype (18). It has also been suggested that TGF-β1 can induce apoptosis in human primary B lymphocytes (19, 20, 21) and BL cell lines (20, 22, 23, 24, 25, 26). The induction of apoptosis by TGF-β1 has also been observed in myeloid leukemia cells (27), gastric carcinoma cells (28), primary endometrial cells (29), primary hepatocytes and hepatoma cells (30), human cervical carcinoma cell lines (31), and human lung epithelial cell lines (32).
We sought to characterize TGF-β1-mediated apoptosis in BL cell lines. Concomitant with our studies, others reported that this process involves activation of caspase 3 (24, 25) and may involve down-regulation of the Bcl-2 family member Bcl-XL (24) and cleavage of the retinoblastoma protein (pRb). In this study, we show that these aforementioned events are late in TGF-β1-mediated apoptosis and are preceded by activation of caspases 2, 7, 8, and 9 in a caspase 8-dependent, but caspase 3-independent manner. Furthermore, we also demonstrate that death is Fas, TNF-R1, DR3, DR4, and DR5 death receptor independent.
Materials and Methods
Cell culture and TGF-β1 treatment
The BL41, BL40, and MUTU-I BL cell lines and the IB4 lymphoblastoid cell line were cultured in RPMI 1640 medium supplemented with 10% (v/v) heat-inactivated Serum Supreme (BioWhittaker, Wolkingham, U.K.), 2 mM l-glutamine (Life Technologies, Paisley, U.K.), and 100 U of penicillin and streptomycin (Life Technologies) per ml. Cells were maintained at 37°C in a 10% CO2 humidified incubator.
Human rTGF-β1 (R&D Systems, Minneapolis, MN) was rehydrated in a 4 mM HCl, 1 mg/ml BSA solution at a concentration of 2 μg/ml and used at a final concentration of 5 ng/ml in all experiments. Control cultures were treated with the appropriate equivalent volume of the TGF-β1 rehydration buffer. For experimental analysis, cells were diluted to a concentration of 3 × 105/ml 24 h before manipulation.
Abs and reagents
Sheep anti-mouse Ig conjugated to HRP (Amersham, Little Chalfont, U.K.), goat anti-rabbit HRP Ig (Dako A/S, Glostrup, Denmark), rabbit anti-mouse IgG conjugated to FITC (PharMingen, San Diego, CA), rabbit anti-goat IgG conjugated to FITC, goat anti-rabbit IgG conjugated to FITC (Dako A/S), anti-poly(ADP-ribose) polymerase (PARP) polyclonal Ab (Boehringer Mannheim, Mannheim, Germany), anti-caspase 2, 3, 7, and 9 (PharMingen), anti-caspase 8 (33), anti-active caspase 3, 7, and 9 (NEB, Beverly, MA), anti-Fas (staining, 33451A; PharMingen), anti-trinitrophenyl (PharMingen), anti-FasL (Transduction Laboratories, Lexington, KY), anti-Fas (killing, CH11; PharMingen), anti-Fas (blocking, ZB4; PharMingen), anti-TNF-R1 FITC, anti-TNF-R1, (blocking; R&D Systems), anti-CD3 FITC (Dako A/S), anti-DR3 (ImmunoKontact, Wiesbaden, Germany), anti-DR4, anti-DR5 (R&D Systems), anti-pRb (PharMingen), anti-Bcl-2 (Dako A/S), anti-Bcl-XL, anti-Mcl-1, anti-Bax, anti-Bak (Santa Cruz Biotechnology, Santa Cruz, CA), anti-Bad (NEB), and anti-Bag-1 (34) were all used as recommended by the suppliers. The pan-specific caspase inhibitor benzyloxycarbonyl-Val-Ala-Asp(Ome)-fluoromethylketone (ZVAD-fmk) and the negative control inhibitor benzyloxycarbonyl-Phe-Ala-fluoromethylketone (ZFA-fmk) were purchased from Enzyme Systems Products (Livermore, CA). The caspase 2 inhibitor benzyloxycarbonyl-Val-Asp(OMe)-Val-Ala-Asp(OMe)-fluoromethylketone (ZVDVAD-fmk), caspase 3 inhibitor DEVD-CHO, caspase 8 inhibitor benzyloxycarbonyl-Ile-Glu(OMe)-Thr-Asp(OMe)-fluoromethylketone (ZIETD-fmk), and caspase 9 inhibitor benzyloxycarbonyl-Leu-Glu(OMe)-His-Asp(OMe)-fmk (ZLEHD-fmk) were purchased from Calbiochem (Nottingham, U.K.). Stock solutions (500×, 100 mM) of ZVAD-fmk, ZVDVAD-fmk, DEVD-CHO, ZIETD-fmk, ZLEHD-fmk, and ZFA-fmk were prepared in DMSO (BDH, Poole, U.K.) and stored at −20°C. Soluble human rTRAIL and enhancer were purchased from Alexis (Nottingham, U.K.). Stock solutions (400×) of TRAIL (100 μg/ml) and enhancer (500 μg/ml) were prepared in PBS. Human rTNF-α (PharMingen) was used at a final concentration of 10 ng/ml. Affinity-purified goat anti-human IgM (μ-chain specific; Sigma, Poole, U.K.) was resuspended in 0.135 M sodium chloride at a concentration of 1 mg/ml (100×). Human rTRAIL R2/Fc chimera (DR5) was purchased from R&D Systems, and human rDR3/Fc chimera was a kind gift of C. Roff (R&D Systems).
Protein content estimation and Western blotting
Cells were lysed in RIPA lysis buffer (50 mM Tris, pH 8, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with 1 mM PMSF (Sigma) and Complete protease inhibitor cocktail (Boehringer Mannheim). Protein concentration was estimated spectrophotometrically at 750 nm in a Lambda Bio UV/Vis spectrometer (Perkin-Elmer, Norwalk, CT) using the Bio-Rad detergent-compatible assay, exactly as described by the manufacturers (Bio-Rad, Hemel Hempsted, U.K.). Protein was diluted to a concentration of 2 mg/ml and further diluted in an equal volume of 2× SDS protein sample buffer (60 mM Tris, pH 6.8, 2% (w/v) SDS, 20% (v/v) glycerol, 2% (v/v) 2-ME, and bromophenol blue), and 100 μg was loaded onto 7.5% or 10% SDS-PAGE gels. The Western blotting process was conducted as previously described (35), and proteins were visualized by ECL chemiluminescence (Amersham), as described by the manufacturer. Autoradiograms were then scanned and processed using a UMAX PowerLook III scanner and Adobe Photoshop software (Adobe Systems, Mountain View, CA).
Cell surface Fas, TNF-R1, DR3, DR4, and DR5 expression was measured in an immunofluorescence flow cytometry assay. A total of 1 × 106 cells was collected by centrifugation and washed once in 0.1% FCS/PBS solution. Cells were resuspended in 100 μl of appropriate primary Ab (1 μg of anti-FAS mAb, anti-TNF-R1 FITC, anti-DR3 rabbit polyclonal, anti-DR4 goat polyclonal, anti-DR5 rabbit polyclonal, or anti-trinitrophenyl mAb or anti-CD3 FITC isotype-matched controls) and incubated on ice for 40 min. Cells were then washed twice and resuspended in 100 μl of secondary Ab (1 μg of rabbit anti-mouse conjugated to FITC, rabbit anti-goat conjugated to FITC, goat anti-rabbit conjugated to FITC). After a further incubation for 30 min on ice, cells were washed twice and resuspended in 500 μl of PBS. Staining was then measured by flow cytometry on a FACSort flow cytometer using the CellQuest analysis program (Becton Dickinson, Mountain View, CA).
Cell cycle analysis.
Cell cycle analysis was performed by flow cytometry. Cells were harvested by centrifugation, washed in ice-cold PBS, and fixed in 80% ethanol that had been prechilled to −20°C. Fixed cells were stored at 4°C for up to 1 wk. They were then repelleted and resuspended at a concentration of ∼1 × 106/ml in PBS containing 18 μg/ml propidium iodide (PI; Sigma) and 8 μg/ml RNase A (Sigma; PI solution). After incubation in the dark for at least 1 h, cell cycle profile analysis was performed on 10,000–20,000 cells on a FACSort flow cytometer using the Cellquest analysis program (Becton Dickinson).
5-Bromo-2′-deoxyuridine (BrdU) labeling.
Cells were incubated with BrdU (Sigma) at a concentration of 10 μM for 1 h at each time point. Cells were then collected by centrifugation at 1300 rpm for 5 min, washed twice in 2 ml of 1% BSA in PBS, and resuspended in 500 μl of ice-cold 70% ethanol on ice for 30 min before storage at −20°C or direct analysis. Cells were washed in PBS before thorough resuspension in 750 μl 2 N HCl containing 0.5% v/v Triton X-100 for 30 min at room temperature (RT) to denature the labeled, dsDNA. Acid was neutralized by resuspending cells in 750 μl of 0.1 M sodium tetraborate, pH 8.5, and incubation at RT for 5 min. Cells were centrifuged and resuspended in 20 μl of FITC-conjugated anti-bromodeoxyuridine Ab (Becton Dickinson), which was then further diluted with 380 μl of 1% BSA/0.5% Tween-20/PBS. After incubation in the dark, at RT for 30 min, cells were washed twice in 0.5% Tween-20/PBS and resuspended in 500 μl PI solution. FACS analysis was then performed.
Cells were analyzed using the FITC in situ cell death detection kit (Boehringer Mannheim). Briefly, 2 × 106 cells were harvested at each time point and resuspended in 200 μl of PBS. An equal volume of freshly made 2% formaldehyde/PBS solution was added, and cells were fixed for 30 min at RT with agitation. Cells were washed twice in PBS, and stored in 500 μl 80% ethanol until analysis. After washing in PBS, cells were permeabilized with 100 μl of 0.1% v/v Triton X-100 in 0.1% sodium citrate for 5 min on ice. Cells were then incubated at 37°C for 90 min in a ratio of enzyme/FITC label solution, according to manufacturer’s instructions, before a final wash in PBS and resuspension in PI solution. FACS analysis was then performed.
The 3-(4, 5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium inner salt (MTS) cell cytotoxicity/proliferation assays were performed using the CellTiter 96 aqueous one-solution cell proliferation assay (Promega, Southhampton, U.K.), as described by the manufacturers. Briefly, 100 μl of cells was plated into a 96-well microtiter plate and incubated at 37°C in a 10% CO2 humidified incubator at least 2 h before assaying. A total of 20 μl of MTS reagent was added to each well, and plates were incubated at 37°C. OD 492 nm was measured in a 96-well microtiter plate reader (Anthos Labtec, Durham, NC) upon color conversion.
TGF-β1-induced apoptosis is rapid and proceeds from the G1 and G2/M phases of the cell cycle
We (26) and others (20, 22, 23, 24, 25) have previously shown that TGF-β1 can induce apoptosis in the BL41, Ramos, and L3055 EBV-negative BL cell lines. We have also recently shown that this is the case for the majority of EBV-negative and Group I BL cell lines (26). To date, our study and other studies have shown that apoptosis is evident following 24 h of TGF-β1 treatment and that the cells that undergo apoptosis in response to this cytokine are present in the G1 phase of the cell cycle (20). We sought to further characterize the TGF-β1-mediated apoptotic process and focused our studies on the BL41 and MUTU-I BL cell lines. Recent work from many labs has demonstrated that the cysteine proteases of the IL-1-converting enzyme/cell death abnormal 3 family (the caspases) play a critical role in the apoptotic process (1). Activation of caspase 3 is markedly increased in many cells undergoing apoptosis, and the cleavage of one of its substrates, PARP, has been used as a useful indicator of its activity (36). PARP cleavage analysis clearly demonstrated that apoptosis is detectable in BL41 cells following 8 h of TGF-β1 treatment (Fig. 1). Similar results were also obtained in the MUTU-I cell line (data not shown). This rapid induction of apoptosis was also confirmed by TUNEL and PI staining and FACS analysis. An increase in TUNEL-positive cells and cells with a sub-G1 DNA content (a characteristic of cells undergoing apoptosis) (37, 38) was clearly detectable after 8 h of TGF-β1 treatment (Fig. 2, A and C). Interestingly, the sub-G1 DNA content measure of cells was found to be an underestimate of the cells that were undergoing apoptosis, as measured by TUNEL staining over the TGF-β1 treatment time course (after 48 h of treatment, 60.7% of cells are TUNEL positive, whereas only 23.2% of cells have a sub-G1 DNA content; Fig. 2, A and C). Counterstaining the TUNEL-stained cells with PI enabled us to determine the location of the apoptosing cells in the cell cycle. These were found to be predominantly in the sub-G1 and G1 phases of the cell cycle early in the time course, but appeared to be present in all phases of the cell cycle after 24–48 h of TGF-β1 treatment (Fig. 2,A). Labeling with BrdU and counterstaining with PI demonstrated that the TUNEL-positive cells with an apparent S phase DNA content were in fact derived from cells in the G2/M phase of the cell cycle (Fig. 2,B). BrdU-positive cells (i.e., S phase cells) were not found with a sub-G1 DNA content (Fig. 2,B), and BrdU-negative cells with a >2 N but <4 N appeared over the time course as a tail of cells coming from the 4 N-containing population (Fig. 2 B). Thus, it appears that TGF-β1 can induce apoptosis in BL41 cells rapidly and from both the G1 and G2/M phases of the cell cycle, but probably not during DNA synthesis.
TGF-β1-mediated apoptosis does not correlate with changes in the steady-state level of Bcl-2 family proteins
The Bcl-2 family of proteins has been shown to be important in regulating apoptosis induced by several noxious agents (7). This family is comprised of both proapoptotic and antiapoptotic members, and it has also been recently shown that cleavage of some antiapoptotic Bcl-2 members may result in proapoptotic proteolytic fragments (7). We measured the steady-state levels of the antiapoptotic proteins Bcl-2, Bcl-XL, Bag-1, and Mcl-1 by Western blotting analysis. We detected no changes in expression of Bcl-2 or Mcl-1 over the TGF-β1 treatment time course (Fig. 3). We observed a slight down-regulation of Bcl-XL relative to untreated control samples and the appearance of another form of Bag-1 after 24 h of TGF-β1 treatment (Fig. 3). These findings are in agreement with a recent report (24), which showed that Bcl-XL is down-regulated in the EBV-negative BL cell line Ramos after 24 h of TGF-β treatment. We did not detect any cleaved products of these proteins. Similarly, we observed no changes in the steady-state levels of the proapoptotic proteins Bax and Bak and only a modest down-regulation of Bad after 24–48 h of TGF-β1 treatment (Fig. 3). Thus, TGF-β1-mediated apoptosis in these cells is not initiated by, or associated with, gross changes in the steady-state levels of these proteins.
TGF-β1-mediated apoptosis involves caspase activation
The caspase family of cysteine proteases has been shown to be the critical executioners of apoptosis (1). Recent studies in hepatocytes (39, 40, 41) and prostate carcinoma cell lines (42) have shown that these enzymes are involved in TGF-β1-mediated apoptosis.
Caspases are normally present in cells as inactive zymogens. Active caspases are derived from the proteolytic processing and self-association of two of these procaspase zymogens. This is a process that results in the clipping of a prodomain (1). This property of caspases conveniently allows us to measure the activation of these enzymes by assaying for the appearance of the cleaved, active subunits. Using a panel of Abs, we sought to determine which of these enzymes are involved in TGF-β1-mediated apoptosis. Western blotting analysis clearly demonstrated that following TGF-β1 treatment, caspases 2, 3, 7, 8, and 9 are all cleaved and therefore activated in response to this cytokine (Fig. 4 A). We could detect the appearance of the active p33 form of caspase 2, and after prolonged exposure of the autoradiogramme, the p32 and p19 active forms of caspase 7 and the p37 active form of caspase 9 following 8 h of TGF-β1 treatment. These cleavage products were first detected coincident with the onset of apoptosis. The cleavage of caspase 8 into its p43, p40, and p18 active forms was detected following 12–24 h of TGF-β1 treatment. We could only detect the active p20 and p19 cleavage products of caspase 3 after 24 h of TGF-β1 treatment consistent with other findings (24, 25). The necessity for long exposures of autoradiogrammes indicated that a small percentage of the total zymogen pool of caspases 3, 8, and 9 is cleaved into their active subunits. We did observe a decrease in the steady-state levels of caspase 2 and caspase 7 zymogens following prolonged TGF-β1 treatment, indicating that a large proportion of these procaspases is cleaved during TGF-β1-mediated apoptosis.
Many previous studies of apoptosis have attempted to order caspase activation cascades by Western blotting for the appearance of active caspase subunits (1). We sought to further clarify the order of caspase activation following TGF-β1 treatment by employing another panel of Abs raised against the active subunits of caspases 3, 7, and 9. Western blotting analysis confirmed that TGF-β1 treatment results in the early activation of caspases 7 and 9, followed by a delayed activation of caspase 3, which is detected after 16 h of induction (Fig. 4,B). We have shown above that TGF-β1 induces apoptosis after 8 h of treatment, as measured by PARP cleavage and TUNEL analysis (Figs. 1 and 2). Caspase 3 has been shown to be very efficient in the cleavage of PARP (36), but in this system, the early cleavage of PARP appears to precede caspase 3 activation (Figs. 1 and 4). The other major effector caspases are caspases 7 and 6 (1), and caspase 7 has been shown to cleave PARP in vitro (43). We were unable to detect caspase 6 expression in BL41 cells (data not shown); however, Western blotting experiments described above detected the appearance of the processed forms of caspase 7 coincident with PARP cleavage (Figs. 1 and 4). Thus, it appears that caspase 7 may be responsible for the early PARP cleavage observed following TGF-β1 treatment of BL41 cells.
TGF-β1-mediated apoptosis is caspase 8 dependent, but caspase 3 independent
Although it can be informative, the ordering of caspase cascades by Western blotting for the appearance of active subunits relies upon the Abs affinities. Therefore, we sought to determine which caspases are important in TGF-β1-induced apoptosis by employing specific caspase inhibitors. Pretreatment of the BL41 cell line with the broad spectrum caspase inhibitor ZVAD-fmk blocked TGF-β1-mediated apoptosis, as measured by the appearance of cells with a sub-G1 DNA content (Fig. 5,A). Treatment with this inhibitor revealed a TGF-β1-mediated G1 arrest, with 86.8% of the cells containing a G1 DNA content compared with 63% of the untreated cells (Fig. 5,A). These effects were not visible in the control population treated with the ZFA-fmk peptide (Fig. 5,A). Therefore, we conclude that TGF-β1-mediated apoptosis is caspase dependent, but TGF-β1-mediated growth arrest does not require caspase activity. Western blotting analysis confirmed these findings, showing that in the presence of ZVAD-fmk, TGF-β1-mediated activation of caspases 2, 3 7, 8, and 9 was blocked (Fig. 5 B). Treatment with ZVAD-fmk was also found to inhibit spontaneous apoptosis of the BL41 culture, as the subG1 population in untreated cells was 6.3% compared with 1.1% in the ZVAD-fmk- and TGF-β1-treated cells.
Similar results were also obtained when cells were pretreated with the caspase 8-specific inhibitor ZIETD-fmk. Treatment with this inhibitor blocked the TGF-β1-mediated increase in the sub-G1 population (21.7% sub-G1 compared with 9% sub-G1; Fig. 5,A) and similarly revealed a TGF-β1-mediated G1 arrest, with 75.5% of the cells containing a 2 N DNA content in the presence of TGF-β1 and ZIETD-fmk. The residual apoptotic population detectable in the presence of the caspase 8 inhibitor probably indicates that TGF-β1-induced, but not spontaneous, apoptosis is caspase 8 dependent. The failure to detect caspase 8 activation by Western blotting after 8 h of TGF-β1 treatment (Fig. 4) is probably due to the lack of sufficient Ab sensitivity. The inhibitor experiments (Fig. 5) clearly show that caspase 8 activation is required for activation of the other caspases.
Pretreatment with the caspase 2 (ZVDVAD-fmk) or caspase 9 (ZLEHD-fmk) inhibitors revealed partial effects on TGF-β1-induced apoptosis (Fig. 5, A and B). Sub-G1 populations decreased by ∼4%, and G1 populations correspondingly increased by a similar amount. Western blotting analysis indicated that the caspase 2 inhibitor partially blocked caspase 2 and caspase 9 activation, but had no effect on caspase 8 activation (Fig. 5,B). Likewise, the caspase 9 inhibitor blocked caspase 9 activation, but had no effect on caspase 8 activation (Fig. 5,B). These results indicate that caspase 8 activation is upstream of caspase 2 and caspase 9 activation. Treatment with either the caspase 2 or caspase 9 inhibitors did not block activation of caspase 3 or caspase 7, but prevented their complete cleavage. Larger active caspase 3 and caspase 7 polypeptides were detected by Western blotting following treatment with TGF-β1 in the presence of these inhibitors, indicating that complete activation of caspases 3 and 7 may require caspase 2 and 9 activation (Fig. 5 B).
Pretreatment with the caspase 3 inhibitor DEVD-CHO did not block TGF-β1-mediated apoptosis (Fig. 5,A) or activation of caspases 2, 8, or 9. However, it did block activation of caspases 3 and 7 (Fig. 5 B). These data indicate that TGF-β1-induced apoptosis, although it involves activation of caspases 3 and 7, is not dependent on these effector caspases. Another as yet identified effector caspase must be involved in TGF-β1-mediated apoptosis.
TGF-β1-mediated apoptosis is FAS independent
A defining characteristic of BL is that the c-myc gene is deregulated due to translocation to an Ig locus (44). Recent studies in fibroblasts have shown that deregulated c-myc expression can drive apoptosis in response to growth factor deprivation (45, 46) and that this process is dependent on Fas/FasL interaction and signaling (47). Fas/FasL interaction stimulates apoptosis by direct activation of caspase 8, which autoprocesses itself following recruitment and aggregation by the death domain-containing protein FADD to the Fas/FasL complex (2). Studies with antisense oligonucleotides have shown that apoptosis driven by ionomycin treatment of BL cell lines is dependent on c-myc (48).
Our findings that TGF-β1 activates caspase 8 are consistent with the model that this process may be dependent on c-myc and c-myc-dependent sensitization to Fas/FasL-mediated apoptosis. Immunofluorescence and subsequent Facs analysis using a Fas mAb revealed that the BL41 cell line expresses a very low level of cell surface Fas compared with the positive control lymphoblastoid cell line IB4 (Fig. 6,A). This is consistent with the findings of others (49, 50) who have shown that BL41 has a low level of cell surface Fas compared with lymphoblastoid cell lines. Pretreatment of BL41 cells with TGF-β1 for 8 h (a time period long enough to initiate apoptosis) did not up-regulate detectable cell surface Fas expression on BL41 cells (Fig. 6,A). Western blotting analysis revealed that FasL expression is readily detectable in BL41 cells (Fig. 6,B). However, TGF-β1 treatment was not found to alter the steady-state levels of FasL (Fig. 6,B). Exposing Fas-sensitive cells to the agonistic mouse anti-Fas mAb CH-11 can readily trigger activation of the Fas death pathway. Treatment of the IB4 cell line with CH-11 (1 μg/ml) for 24 h resulted in ∼30% of the cells undergoing apoptosis, as measured by determining the number of cells with sub-G1 DNA content visualized by FACS analysis (Fig. 6,C). In contrast, 24 h of CH-11 treatment did not induce apoptosis above background levels in BL41 cells (Fig. 6,C). Pretreatment with TGF-β1 24 h before the addition of CH-11 did not sensitize BL41 cells to Fas activation, as the level of TGF-β1-induced apoptosis was unaffected by CH-11 addition (Fig. 6,C). C-myc-induced apoptosis in serum-starved fibroblasts can be inhibited by the mAb ZB4, which inhibits the Fas/FasL interaction (47). Addition of ZB4 (1 μg/ml) 1 h before the addition of CH-11 to the IB4 cell line efficiently blocked Fas-mediated apoptosis (Fig. 6,D). However, pretreatment of BL41 cells with ZB4 had no effect on TGF-β1-mediated apoptosis (Fig. 6 D). These results all indicate that TGF-β1-mediated apoptosis in BL41 cells does not depend on Fas/FasL interaction.
TGF-β1-induced apoptosis is TNF-α/TNF-R1 independent
Recent studies have indicated that TNF-α, through interaction with its receptor TNF-R1, can induce apoptosis in a caspase 8-dependent fashion (51). Therefore, it was possible that TGF-β1-mediated apoptosis could proceed via the TNF-α/TNF-R1 pathway. FACS analysis revealed that BL41 cells expressed readily detectable levels of TNF-R1 on their cell surface (mean fluorescence 29.8 compared with control CD3 FITC-treated cells, which had a mean fluorescence of 3.3; Fig. 7,A). Treatment with TGF-β1 was found to slightly down-regulate surface TNF-R1 expression. Similarly, Western blotting analysis indicated that TGF-β1 did not significantly affect the steady-state levels of TNF-R1 expression (Fig. 7 B).
It has been demonstrated recently that TNF-α can block anti-IgM-induced apoptosis in the Ramos Burkitt’s lymphoma cell line (52). We found that treatment of BL41 cells with 10 μg/ml of anti-IgM for 48 h could readily induce apoptosis in BL41 cells, as measured by the MTS cell proliferation assay (Fig. 7,C). Treatment with TNF-α (10 ng/ml) had no effect on BL41 proliferation, but pretreatment with TNF-α before anti-IgM addition significantly inhibited anti-IgM-mediated apoptosis. Treatment with the TNF-α-blocking Ab did not affect BL41 cell proliferation or anti-IgM-mediated apoptosis, but efficiently blocked TNF-α-mediated protection of anti-IgM-induced apoptosis (Fig. 7,C). In contrast, TNF-α was found to have no significant protective effect against TGF-β1-mediated apoptosis, and TNF-α-blocking Abs did not affect TGF-β1-mediated apoptosis (Fig. 7 D). Thus, it appears that TGF-β1-mediated apoptosis is independent of TNF-α and TNF-R1.
TGF-β1-mediated apoptosis is DR3, DR4, DR5, TRAIL, and TNF-like weak inducer of apoptosis independent
Recent studies have identified a family of TNF/Fas death receptors and their ligands (2, 4). Apo2L (TRAIL) is closely related to FasL and can induce apoptosis in tumor cells via interaction with its receptors DR4 or DR5, a process that results in caspase activation (2). Apo3L (TNF-like weak inducer of apoptosis) is closely related to TNF and signals through its receptor, DR3, to induce apoptosis like Fas and TNF via activation of caspase 8 (2, 4). As we observed caspase 8 activation in BL41 cells in response to TGF-β1 treatment, we sought to determine whether TGF-β1-mediated apoptosis was dependent on signaling through these death receptors. We found that rApo2L (250 ng/ml) could potently induce apoptosis in the BL40 cell line, as measured by the MTS cell proliferation assay (Fig. 8,A) and FACS analysis (data not shown). Chinnaiyan and coworkers (53) have demonstrated that recombinant nonsignaling death receptor DR5 can block TRAIL-mediated apoptosis when added exogenously to cells (53). Similarly, we found that TRAIL-induced apoptosis in BL40 cells could be inhibited by soluble rDR5/Fc fusion protein (250 ng/ml), but not soluble rDR3/Fc fusion protein (250 ng/ml) (Fig. 8,A). Pretreatment of BL41 cells for 1 h before the addition of TGF-β1 with soluble DR5/Fc or soluble DR3/Fc protein did not block TGF-β1-mediated apoptosis (Fig. 8,B). To evaluate whether TGF-β1-mediated apoptosis could proceed through ligand-independent up-regulation of these death receptors, we performed cell surface staining and Western blotting analysis employing DR3-, DR4-, and DR5-specific Abs. The BL40 cell line was found to express a low but readily detectable level of cell surface DR3, DR4, and DR5 receptors (Fig. 8,C). These polypeptides were also readily detected in Western blotting experiments (Fig. 8,D). BL41 cells were found to express very low levels of surface DR3 and DR4 death receptors and almost background levels of DR5 receptors (Fig. 8,C). Treatment with TGF-β1 for 8 h slightly decreased surface expression of DR3 and DR4. Consistent with these findings, we found that TGF-β1 treatment did not significantly affect the steady-state levels of DR3 or DR4 over a 48-h time course of treatment, and that we could not detect DR5 expression in BL41 cells (Fig. 8,D). The data presented in Figs. 6, 7, and 8 indicate that TGF-β1-mediated apoptosis in BL41 cells is death receptor independent.
TGF-β1-mediated apoptosis results in the cleavage of pRb following activation of caspase 3
It has previously been observed that pRb can be cleaved by caspase 3 at the consensus site DEAD/G located in the C terminus (54, 55), and that this event may be critical in TNF-α, but not Fas-induced apoptosis (54). Schrantz et al. (25) have also recently suggested that pRb cleavage may be important in TGF-β1-mediated apoptosis in BL41 cells. Western blotting analysis with a pRb-specific mAb revealed the appearance of a faster migrating form of pRb following 24 h of TGF-β1 treatment and a general reduction in the steady-state levels of all forms of pRb by 48 h of TGF-β1 treatment (Fig. 9). These data are completely consistent with previous findings (25). However, we did not observe the appearance of the faster migrating form of pRb before 24 h of TGF-β1 treatment, indicating that this is a relatively late event in TGF-β1-mediated apoptosis and occurs after the activation of caspases 2, 7, 8, and 9, and after the induction of PARP cleavage, TUNEL positivity, and the accumulation of cells with a sub-G1 DNA content.
This study aimed at characterizing TGF-β1-mediated apoptosis in Burkitt’s lymphoma cell lines. We demonstrated that TGF-β1 induces apoptosis within 8 h of treatment and that this apoptosis proceeds from the G1 and G2/M phases of the cell cycle. Cells in S phase appear to be refractory to TGF-β1-mediated apoptosis.
The caspase family of proteases has been shown to be key executioners in the apoptotic process in many cell types in response to diverse apoptotic stimuli (1). It has previously been suggested that caspases are involved in TGF-β1-mediated apoptosis in rat and human hepatocytes and hepatocarcinoma cell lines (39, 40, 41) and in human prostate cancer cell lines (42). These studies demonstrated that pretreatment of cells with the pan-specific caspase inhibitor ZVAD-fmk blocked TGF-β1-mediated apoptosis. We also observed that ZVAD-fmk could block TGF-β1-mediated apoptosis in BL cell lines. Normally, caspases exist in cells as inactive zymogens that are activated by a cleavage event. This property enables researchers to observe caspase activation by assaying for the appearance of the cleaved active subunits of caspase enzymes. We used this technique to demonstrate that caspase 3 is activated only after 16 h of TGF-β1 treatment. This result confirms the recent findings of Saltzman et al. (24) and Sanchez et al. (25). We have extended these findings by demonstrating for the first time that TGF-β1 can initiate the activation of caspases 7, 8, and 9. Furthermore, by employing specific caspase inhibitors, we demonstrated that TGF-β1-induced apoptosis is caspase 8 dependent, but caspase 3 independent. Blocking activation of caspase 3 also blocked the activation of caspase 7, but it did not alter the level of apoptosis, which suggests that another effector caspase other than caspases 3 and 7 must be activated in response to TGF-β1 treatment. By careful analysis of caspase activation in the presence of these specific inhibitors, it is possible to build a model of a caspase network that is activated in response to TGF-β1 (Fig. 10).
Activation of caspase 8 is considered to be the initial caspase cleavage event in cells responding to the apoptotic-inducing effects of FasL, TNF, and TRAIL (2, 4). However, we found that TGF-β1-mediated apoptosis is independent of the death receptors Fas, TNF-R1, DR3, DR4, and DR5, and therefore appears to activate caspase 8 in a death receptor-independent fashion. In other cell systems, it has been shown that TGF-β1 can interact with the Fas/FasL pathway. Dybedal et al. (56) suggested that TGF-β1 can block Fas-mediated apoptosis in murine bone progenitor cells, but not in thymocytes. Similarly, TGF-β1 has been demonstrated to block FasL expression induced by IFN-γ treatment of keratinocytes (57) and to block FasL expression and subsequent activation-induced cell death in T cells (58). It has also been shown recently that TGF-β1 can enhance Fas-mediated apoptosis in human glioma cells (59). These effects are clearly cell-type dependent. We did not observe any effect on the Fas/FasL pathway and believe it not to be involved in TGF-β1-mediated apoptosis in BL. Mistletoe lectin treatment of the Jurkat and BJAB cell lines has recently been shown to result in death receptor-independent activation of caspase 8 (60); however, to our knowledge, this is the first report of a mammalian cytokine inducing death receptor-independent activation of caspase 8.
Recent work of others has shown that apoptosis induced in BL cell lines by Ag receptor ligation is dependent on early activation of caspase 2 (61), implying that in these cells caspase 2 may well act as an initiator caspase. The caspase cascade activated by Ag receptor ligation does not involve activation of caspase 8 (61). Anti-IgM-mediated apoptosis can also be blocked by TNF-α (51 ; data shown here). TGF-β1-mediated apoptosis involves activation of caspase 2, but is caspase 8 dependent and is not inhibitable by TNF-α, and therefore clearly activates a separate death program.
It is interesting to note that TGF-β1-mediated apoptosis pathways may be different in human BL cells and murine B cell lines. Brown and coworkers (62, 63) have shown that in the WeHI-231 cell line, TGF-β1-induced apoptosis does not involve caspase 3 and is not blocked by ZVAD-fmk. TGF-β1-induced apoptosis in the BL41 cell line involves activation of caspase 3, but is not dependent upon its activation and is clearly blocked by ZVAD-fmk.
It has been reported that TGF-β1 can down-regulate Bcl-XL in the Ramos BL cell line (24). It is currently believed that the Bcl-2 family of proteins is involved in controlling cytochrome c release from mitochondria (7). We also observed a down-regulation of Bcl-XL expression following 24 h of TGF-β1 treatment of BL41 cells, but we did not detect any changes in the steady-state level of Bcl-2 family members early in TGF-β1-mediated apoptosis. It has yet to be determined whether changes in cellular localization of these proteins take place during the TGF-β1 response.
A fast migrating form of pRb could be detected after 24 h of TGF-β1 treatment. This form of pRb has been postulated to represent ΔpRB and correspond to the truncated form of pRb produced by caspase 3-mediated cleavage (25). This does indeed seem likely because we observed the appearance of this form shortly after the appearance of active caspase 3, and Schrantz et al. (25) demonstrated that the appearance of this form can be blocked by blocking caspase 3-like activity. As we now demonstrate that this is a late event in TGF-β1-mediated apoptosis and occurs after activation of caspases 2, 7, 8, and 9, it remains unclear how important this cleavage is in determining the outcome of TGF-β1 treatment, as it is clearly not required for the acquisition of an apoptotic phenotype.
A defining feature of BL is a c-myc translocation to a region in the Ig enhancer region, which results in deregulated c-myc expression (44). Recently, Evan and coworkers (46, 47) have shown that c-myc can induce apoptosis in growth factor-deprived fibroblasts. In this system, c-myc-induced apoptosis is dependent on Fas/FasL interaction and signaling. It has been suggested that apoptosis in BL cell lines in response to ionomycin is c-myc dependent (48). If c-myc is involved in TGF-β1-mediated apoptosis in BL, it is clearly Fas/FasL interaction independent. Recently, it was shown that c-myc can stimulate cytochrome c release from mitochondria in fibroblasts (64). If TGF-β1-mediated apoptosis is dependent on c-myc in BL, then again this is clearly different from the case in WEHI-231 cells because c-myc can actually block TGF-β1-mediated apoptosis in these cells (65). TGF-β1-mediated down-regulation of c-myc appears to be necessary for apoptosis in this cell line (65). We have also observed that TGF-β1 down-regulates c-myc expression in BL41 cells, but that this does not correlate with the initiation of apoptosis and probably represents proteolytic cleavage of the protein late in the death pathway (G.J.I. and M.J.A., unpublished observations).
BL cell lines have a phenotype resembling that of germinal center B cells (66). Follicular dendritic cells, which are present in germinal center regions, produce bioactive TGF-β1 (10). Recently, it was demonstrated that dendritic cells can promote class switching of naive CD40-triggered human B cells to an IgA phenotype in the presence of IL-10, and that this process is partially dependent on endogenous TGF-β1 (67). It is has also been reported that TGF-β1 may cause apoptosis in human B cells isolated from tonsil, preparations that contain germinal center B cells (20). In addition, TGF-β1 may inhibit Ag-induced rescue of germinal center B cells (68). Therefore, it is tempting to speculate that TGF-β1 may play a role in both the development and the survival of germinal center B cells, and so the study of BL cell lines may provide a useful model to investigate these phenomena.
We gratefully acknowledge gifts of reagents from C. Roff (R&D Systems), C. Scaffidi, P. Krammer, and M. Peter (Deutsches Krebsforschungszentrum (DKFZ), Heidelberg, Germany), and M. Brimmel and G. Packham (Ludwig Institute for Cancer Research, London, U.K.). We also thank A. Mouzakiti and G. Packham (Ludwig Institute for Cancer Research) for advice during the preparation of this manuscript.
This work was supported by The Wellcome Trust Project Grants 047383 and 050096 to M.J.A.
Abbreviations used in this paper: FasL, Fas ligand; BL, Burkitt’s lymphoma; BrdU, 5-bromo-2′-deoxyuridine; MTS, 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt; PARP, poly(ADP-ribose) polymerase; PI, propidium iodide; pRb, retinoblastoma protein; RT, room temperature; TRAIL, tumor necrosis-like apoptosis-inducing ligand; ZFA-fmk, benzyloxycarbonyl-Phe-Ala-flouromethylketone; ZIETD-fmk, benzyloxycarbonyl-Ile-Glu(OMe)-Thr-Asp(OMe)-fluoromethylketone;ZLEHD-fmk,benzyloxycarbonyl-Leu-Glu(OMe)-His-Asp(OMe)-fmk; ZVAD-fmk, benzyloxycarbonyl-Val-Ala-Asp(Ome)-flouromethylketone;ZVDVAD-fmk,benzyloxycarbonyl-Val-Asp(OMe)-Val-Ala-Asp(OMe)-fluoromethylketone.