Oncogene (2005) 24, 7031–7042
& 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc
Arsenic trioxide induces regulated, death receptor-independent cell death through a Bcl-2-controlled pathway Christian Scholz1, Antje Richter1, Mario Lehmann1, Klaus Schulze-Osthoff2, Bernd Do¨rken1 and Peter T Daniel*,1,3 1
Department of Hematology, Oncology and Tumor Immunology, University Medical Center Charite´, Campus Buch, Humboldt University, Berlin, Germany; 2Institute of Molecular Medicine, Heinrich Heine University, Du¨sseldorf, Germany; 3Clinical and Molecular Oncology, Max Delbru¨ck Center for Molecular Medicine, 13125 Berlin, Germany
Arsenic trioxide (As2O3, arsenite) efficiently kills cells from various hematologic malignancies and has successfully been employed especially for the treatment of acute promyelocytic leukemia. There and in lymphoid cells, we demonstrated that As2O3 induces cell death in a caspase-2- and -9-independent fashion. Here, we address a potential role of death receptor signaling through the FADD/caspase-8 death-inducing signaling complex in As2O3-induced cell death. In detail, we demonstrate that As2O3 induces cell death independently of caspase-8 or FADD and cannot be blocked by disruption of CD95/Fas receptor ligand interaction. Unlike in death receptor ligation-induced apoptosis, As2O3-induced cell death was not blocked by the broad-spectrum caspase inhibitor z-VAD-fmk or the caspase-8-specific inhibitor z-IETDfmk. Nevertheless, As2O3-induced cell death occurred in a regulated manner and was abrogated upon Bcl-2 overexpression. In contrast, As2O3-induced cell demise was neither blocked by the caspase-9 inhibitor z-LEHD-fmk nor substantially inhibited through the expression of a dominant negative caspase-9 mutant. Altogether our data demonstrate that As2O3-induced cell death occurs independently of the extrinsic death receptor pathway of apoptosis. Cell death proceeds entirely via an intrinsic, Bcl-2-controlled mitochondrial pathway that does, however, not rely on caspase-9. Oncogene (2005) 24, 7031–7042. doi:10.1038/sj.onc.1208868; published online 11 July 2005 Keywords: arsenic trioxide; apoptosis; FADD; death receptor; CD95/Fas
caspase-8;
Introduction Arsenic trioxide (As2O3) has sucessfully been employed to treat patients with newly diagnosed and relapsed acute promyelocytic leukemia (APL) (Shen et al., 1997; *Correspondence: PT Daniel, Clinical and Molecular Oncology, Charite´ University Medical Center, Campus Virchow Klinikum, Augustenburger Platz 1, 13353 Berlin, Germany; E-mail:
[email protected] Received 23 February 2005; revised 16 May 2005; accepted 16 May 2005; published online 11 July 2005
Soignet et al., 1998; Niu et al., 1999). Mechanisms of As2O3-induced cell death, however, have only recently started to unravel. Initially, the central mechanism initiating As2O3-induced cell death was presumed to be degradation and targeting of the promyelocytic leukemia gene product-retinoic acid receptor alpha (PML-RARa) fusion protein, which is typically expressed in APL cells. However, recent data demonstrated that PML-RARanegative cell lines and cells from PML knockout mice are sensitive to As2O3 (Akao et al., 1998; Wang et al., 1998; Zhang et al., 1998; Zhu et al., 1999), thus questioning the importance of PML in cell demise mediated by As2O3. In general, cell death occurs either through necrosis or apoptosis. Apoptotic cell death is either mediated through death receptors, the mitochondrial or the endoplasmatic reticulum pathway. The mitochondrial or intrinsic death pathway is induced via growth-factor deprivation or chemotherapeutic drugs and results in the release of cytochrome c into the cytosol, which is either accompanied or followed by the loss of the mitochondrial transmembrane potential (DCm). Subsequently, cytochrome c oligomerizes with apoptotic protease activating factor-1 (APAF-1) in a strictly adenosine 50 -triphosphate (ATP)-dependent manner. The resulting multimeric complex, termed the apoptosome, facilitates the recruitment and activation of caspase-9 which subsequently triggers effector caspases, for example, caspase-3 or caspase-7 (Daniel et al., 2003). Such caspase activation is enhanced by the IAP-antagonists Smac/Diablo and Omi/HtrA2, which are also released from the mitochondria during apoptosis. In contrast, Bcl-2 and Bcl-xL, two antiapoptotic members of the Bcl-2 protein family, efficiently inhibit the mitochondrial or intrinsic pathway of apoptosis (Daniel et al., 2003). The death receptor pathway of apoptosis, in turn, is initiated through the ligation of a death receptor, for example, CD95 (Fas, APO-1) and the successive assembly of the death-inducing signaling complex (DISC) (Peter and Krammer, 2003). In case of CD95 or the TRAIL receptor, the DISC consists of the cytoplasmic region of the receptor, the adaptor protein Fas-associating protein with a death domain (FADD) and pro-caspase-8 (FLICE) or pro-caspase-10. DISC assembly mediates the activation of caspase-8, thereby triggering downstream effector caspases, such as cas-
As2O3-induced cell death C Scholz et al
7032
pase-3, resulting in DNA fragmentation (Kischkel et al., 1995, 2000). We recently demonstrated that As2O3mediated cell death functions in the absence of caspase2 and caspase-9, but is efficiently blocked by Bcl-xL, thus demonstrating the importance of mitochondria for As2O3-induced cell demise (Scholz et al., 2005). Intrinsic and extrinsic pathways of apoptosis are, however, connected through caspase-8-mediated processing and activation of the BH3-only protein Bid. Truncated Bid (tBid) facilitates actvation of Bak/Bax, proapoptotic members of the Bcl-2 family and has important implications for the sensitization of cancer cells to DNAdamaging anticancer therapies (Daniel et al., 2001; von Haefen et al., 2004; Rudner et al., 2005). Recent data suggested a key role for caspase-8 in As2O3-mediated cell death (Kitamura et al., 2000). Furthermore, the participation of the mitochondrial apoptosis pathway was implied in arsenic-induced cell demise (Akao et al., 1998; Wang et al., 1998; Larochette et al., 1999; Hossain et al., 2000; Perkins et al., 2000). This is supported by our recent finding that As2O3 induces a regulated form of necrosislike cell death through a Bcl-xL inhibitable pathway. In the present study, we therefore investigated the functional role of the death receptor pathway of apoptosis in As2O3-induced cell death. We show that As2O3-induced cell death is accomplished in the absence of the initiator caspase-8 or the FADD adaptor molecule. Our findings establish that As2O3 induces cell death in a CD95 death receptor, caspase-8 and FADD independent fashion.
Figure 1 Bcl-2 expression inhibits CD95-mediated cell death in Jurkat cells. Jurkat cells stably overexpressing Bcl-2 and the corresponding mock-transfectants were cultured for 48 h with medium alone, or with a plastic-bound anti-CD95/Fas antibody (anti-APO-1, 1 mg/ml). Cell death was determined at a single cell level by staining cells with annexin-V-FITC and counterstaining with PI. Dead cells were annexin-V-FITC and/or PI positive. Mean values7s.d. are shown
Results As2O3-induced cell death occurs via the mitochondrial pathway Data from myeloid leukemia- or Burkitt-like lymphomaderived cell lines as well as from cell-free systems indicate an involvement of mitochondria in As2O3-induced cell death (Larochette et al., 1999; Perkins et al., 2000). Here, we employ mutants of the acute lymphoblastic leukemia cell line Jurkat to address the issue of death receptor signaling in As2O3-induced cell death. To corroborate the role of mitochondria in As2O3-mediated cell demise in our current setting, we employed Jurkat cells overexpressing Bcl-2. These Jurkat Bcl-2 cells or mock-transfectants (mock) were cultured for 48 h in the presence or absence of plastic immobilized, agonistic anti-CD95 antibody APO-1 (IgG3, 1 mg/ml) (Dhein et al., 1992). CD95/Fas ligation induced apoptosis in Jurkat mock-transfectants, while Jurkat cells overexpressing Bcl-2 were protected from CD95-mediated apoptosis (Figure 1). Thus, Jurkat cells are ‘type II’ cells that regulate death receptor-induced apoptosis through the intrinsic pathway (Scaffidi et al., 1998), although this dichometry into type I and II cells depends on the strength of the death ligand signal (Rudner et al., 2005). Next, Jurkat transfectants were cultured for 24 or 48 h with increasing concentrations of As2O3 (Figure 2). Cell death was quantified by staining cells with annexinOncogene
V-FITC and propidium iodide (PI). Total cell death includes cells that are either annexin-V-FITC or PI positive (Figure 2a). In contrast, positivity for annexinV-FITC in the absence of PI staining (Figure 2b) is characteristic for apoptosis, while positivity for PI (Figure 2c), that is, membrane damage, indicates a necrotic type of cell death (Vermes et al., 1995). In line with our previous report (Scholz et al., 2005), As2O3 induced a substantial amount necrosis-like death in mock-transfected Jurkat cells as evidenced by the large proportion of PI-positive cells (Figure 2c, open circles). Bcl-2 overexpression efficiently blocked apoptosis- and necrosis-like cell death in response to treatment with As2O3 (Figure 2a–c, closed circles). Apart from assessing cell death by detecting annexin/PI, we measured cells with hypodiploid DNA, as an alternative method of quantifying apoptotic cells (Figure 2d). Similar to the data shown in Figure 2b, As2O3 induced a moderate amount of cells with hypodiploid DNA in mocktransfectants (Figure 2d, opern circles), while Bcl-2overexpressing Jurkat cells were effectively protected (Figure 2d, closed circles) from As2O3-induced apoptosis. Thus, As2O3 induces mainly necrotic cell death in Jurkat cells, which are type II cells. In addition, our data further support the critical role of mitochondria as gatekeepers in As2O3-induced cell death.
As2O3-induced cell death C Scholz et al
7033
As2O3-mediated cell death does not depend on caspase-8 Type II cells depend on a functional mitochondrium for the execution of programmed cell death. Nevertheless, the extrinsic pathway of apoptosis is initiated via the formation of a DISC, where the ligation of a death receptor, for example, CD95/Fas or TRAIL facilitates the oligomerization with the adaptor protein FADD. Subsequent recruitment of the initiator caspase-8 concludes the assembly of the DISC and results in activation of caspase-8. Activated caspase-8 triggers downstream effector caspase-3 and -7, which mediate cleavage of a broad range of substrate proteins (Fischer et al., 2003) and initiate DNA fragmentation and cell death. Previous data from our group (Wieder et al., 2001; von Haefen et al., 2003) demonstrated that caspase-8 can furthermore be processed ‘retrogradely’ via caspase-3 and may thereby mediate a caspase-3/8driven mitochondrial amplification loop (Fischer et al., 2003; von Haefen et al., 2003). Although recent data indicate that caspase-8 processing is dispensable for caspase-8 activation, the role of caspase-8 activation outside of the DISC as executioner caspase in the intrinsic cell death pathway is under controversial discussion (Boatright et al., 2003; Chandra et al., 2004; Sohn et al., 2005). Taking this into account, we next wanted to test whether caspase-8 activity was necessary for As2O3-mediated cell death. Accordingly, we utilized a caspase-8-deficient Jurkat cell line (caspase-8/). As anticipated, a 48 h culture of both cell lines with the plastic-bound, agonistic anti-CD95 antibody APO-1 (Dhein et al., 1992) (IgG3, 1 mg/ml) resulted in cell death of the parental Jurkat A3 cell line, while Jurkat caspase-8/ cells were refractory to CD95-mediated cell demise (Figure 3). In contrast, treatment of Jurkat A3 (open circles) and Jurkat caspase-8/ cells (closed circles) with increasing concentrations of As2O3 for 24 or 48 h (Figure 4) induced similar percentages of total cell death (Figure 4a), annexin-V-FITC single-positive cells (Figure 4b), PI-positive cells (Figure 4c), or cells with hypodiploid DNA (Figure 4d). Caspase-8/ Jurkat cells even appeared to be slightly more susceptible to apoptosis-like cell death as determined by measuring percentages of annexin-V-FITC single-positive cells (Figure 4b). To assess whether As2O3 mediates more pronounced apoptosis-like cell death at earlier time points, we exposed caspase-8 wild-type Jurkat A3 cells for 3, 6, and 12 h to As2O3. While cells remained largely unaffected after a 3 or 6 h treatment, there was a
Figure 2 Constitutive upregulation of Bcl-2 prevents As2O3mediated cell death. Jurkat cells either mock-transfected or cells stably overexpressing Bcl-2 were incubated for 24 and 48 h with increasing concentrations of As2O3, as indicated. Cell death was measured at the single cell level by labeling cells with annexinV-FITC and counterstaining with PI. (a) Total cell death (annexinV-FITC- and/or PI-positive cells), (b) apoptosis (annexin-VFITC-positive/PI-negative cells), (c) necrosis (PI-positive cells). (d) Alternatively, cells with genomic DNA fragmentation (sub-G1 DNA content) were identified by flow cytometric measurement of cellular DNA content. Mean values7s.d. are shown Oncogene
As2O3-induced cell death C Scholz et al
7034
Figure 3 CD95-mediated cell death is inhibited in caspase-8/ Jurkat cells. Jurkat cells deficient for caspase-8 or the parental cell line Jurkat A3 were cultured for 48 h with medium alone, or with a plastic-bound anti-CD95/Fas antibody (anti-APO-1, 1 mg/ml). Cell death was determined by staining cells with annexin-V-FITC and counterstaining with PI. Dead cells were annexin-V-FITC- and/or PI-positive. Mean values7s.d. are shown
moderate percentage of annexin-V-positive PI-negative Jurkat A3 cells after 12 h of incubation (Figure 5b). In contrast, no substantial percentage of PI þ Jurkat cells was detected at any of these earlier time points (Figure 5c). This indicates that a limited amount of up to 20% of the Jurkat cells dies early via an apoptotic mechanism. To further assess the functional role of caspase activation in As2O3-induced cell death, Jurkat A3 cells were treated for 48 h with either medium, the plasticbound anti-CD95 antibody (APO-1, 1 mg/ml), or As2O3 (6 mM) in the presence or absence of the broad-spectrum caspase inhibitor z-VAD-fmk (10 mM) (Figure 6a) or the caspase-8 inhibitor z-IETD-fmk (10 mM) (Figure 6b). To this end, z-VAD-fmk (Figure 6a) or z-IETD-fmk (Figure 6b) decreased anti-CD95-mediated cell death from 68.370.5 to 15.370.7 or to 38.970.6%, respectively. In contrast, As2O3-induced cell demise was only moderately reduced from 45.270.9 to 30.570.8% in the presence of z-VAD-fmk (Figure 6a) and not reduced in the presence of z-IETD-fmk (Figure 6b). This demonstrates that As2O3-induced cell death functions largely independent of caspase activity in general and of
Figure 4 Caspase-8 deficiency falls short of inhibiting As2O3mediated cell death. Jurkat cells deficient for caspase-8 or the parental cell line Jurkat A3 were treated for 24 and 48 h with increasing concentrations of As2O3, as depicted. Cell death was measured at the single cell level by labeling cells with annexin-VFITC and counterstaining with PI. (a) Total cell death, (b) apoptosis, (c) necrosis-like cell death, (d) cells with genomic DNA fragmentation. Mean values7s.d. are shown Oncogene
As2O3-induced cell death C Scholz et al
7035
Figure 5 Analysis of apoptotic and necrosis-like cell death induced by As2O3 at early time points. Jurkat A3 cells were treated for 3, 6, or 12 h with increasing concentrations of As2O3, as depicted. Cell death was measured at the single cell level by labeling cells with annexin-V-FITC and counterstaining with PI. (a) Total cell death, (b) apoptosis, (c) necrosis-like cell death. Mean values7s.d. are shown
Figure 6 Caspase inhibition does not block As2O3-mediated cell death. Jurkat cells (Jurkat A3) were cultured for 48 h in medium alone, with a plastic-bound anti-CD95/Fas antibody (anti-APO-1, 1 mg/ml) or with anti-CD95 antibody either in the presence of 10 mM z-VAD-fmk (a) or 10 mM z-IETD-fmk (b). Alternatively, Jurkat A3 cells were treated with 6 mM As2O3 alone or with 6 mM As2O3 either in the presence of 10 mM z-VAD-fmk (a) or 10 mM z-IETD-fmk (b). Cell death was determined at a single cell level by staining cells with annexin-V-FITC and counterstaining with PI. Dead cells were annexin-V-FITC and/or PI positive. Mean values7s.d. are shown
caspase-8 activity in particular. To test whether a higher concentration of z-VAD-fmk had a more pronounced blocking effect, we cultured Jurkat A3 cells with 4 mM As2O3 in the presence or absence of 20 or 40 mM z-VADfmk for 24 (Figure 7a) or 48 h (Figure 7b). As a control, cells were treated with the cytostatic drug taxol in the presence or absence of z-VAD-fmk. While taxol-induced cell death was substantially inhibited by either dosage of z-VAD-fmk, we only observed a moderate inhibition of As2O3-mediated cell death after a 24 h culture and hardly any decrease after a 48 h incubation with As2O3 in the presence of 20 or 40 mM z-VAD-fmk (Figure 7). Thus, a significant share of As2O3-triggered cell death proceeds via a caspase-independent mechanism as caspase-inhibition appears to only slightly diminish cell death of Jurkat cells. This is supported by our earlier data in caspase-2 and -9 knockout cells that remained sensitive upon exposure to As2O3 (Scholz et al., 2005). As2O3-mediated cell death is not affected by the absence of the adaptor protein FADD So far, we demonstrated that caspase-8 expression or activation is not required for As2O3-induced cell death. To further dissect the role of the death-inducing signaling complex during As2O3-mediated cell demise, we employed Jurkat cells deficient for the adaptor
Figure 7 Dose–response and time course for z-VAD-fmk inhibition of As2O3- and paclitaxel (taxol)-mediated cell death. Jurkat cells (Jurkat A3) were cultured for 24 (a) or 48 h (b) in medium alone. Alternatively, cells were treated with 0.01 mg/ml taxol or 4 mM As2O3 in the absence or presence of 20 or 40 mM z-VAD-fmk. Cell death was determined at a single cell level by staining cells with annexin-V-FITC and counterstaining with PI. Dead cells were annexin-V-FITC and/or PI positive. Mean values7s.d. are shown Oncogene
As2O3-induced cell death C Scholz et al
7036
molecule FADD (FADD/) and compared them to the parental cell line Jurkat A3. Notably, functional FADD is required for the activation of caspase-8, the initiator casapase that is triggered by DISC formation. As expected from the literature, cell culture with the plastic-bound, agonistic anti-CD95 antibody (clone 2R2, IgG3, 5 mg/ml) resulted in substantial cell death of the parental Jurkat A3 cell line, while FADDdeficient Jurkat cells were unaffected by co-culture with the anti-CD95 antibody (Figure 8). Next, cells were treated with increasing concentrations of As2O3 for 24 and 48 h and cell death was assessed by staining cells with annexin-V-FITC and PI. The FADD deficiency only moderately decreased the amount of overall cell death, that is, the sum of PI-positive and annexin-VFITC-positive/PI-negative cells (Figure 9a) or PIpositive, necrotic cells (Figure 9c), while binding of annexin-V-FITC as a sign of apoptosis was hardly affected in FADD/ cells (Figure 9b). To corroborate our data in a different system, we employed a cell line derived from a Burkitt-like lymphoma (BJAB) that had been transfected with a dominant negative mutant of FADD (FADD-DN). Again, BJAB FADD-DN or mock-transfected controls were treated with increasing concentrations of As2O3 and cell death was measured by staining with annexin-V-FITC or PI after 24 and 48 h of culture. In line with our previous results, As2O3 induced similar percentages of cell death in both cell lines (Figure 10). Notably, total cell death was only moderately decreased in BJAB FADD-DN as compared
Figure 8 CD95-mediated cell death is inhibited in FADDdeficient Jurkat cells. Jurkat cells deficient for FADD (FADD/ ) or the parental cell line Jurkat A3 were cultured for 48 h with medium alone, or with a plastic-bound anti-CD95/Fas antibody (clone 2R2, 5 mg/ml). Cell death was determined by staining cells with annexin-V-FITC and counterstaining with PI. Dead cells were annexin-V-FITC and/or PI positive. Mean values7s.d. are shown Oncogene
to BJAB mock (Figure 10a). Therefore, FADD appears to be largely dispensible for As2O3-induced cell demise. As2O3-induced cell death cannot be inhibited through blockade of the death receptor CD95/Fas Recent data demonstrated that a sublethal dosage of As2O3 increases CD95/Fas expression on Hela cells, suggesting that As2O3 might facilitate CD95-mediated cell death (Woo et al., 2004). To address this and the
Figure 9 FADD-deficient Jurkat cells are susceptiple to As2O3induced cell death. Jurkat cells deficient for FADD (FADD/) or the parental cell line Jurkat A3 were cultured for 24 and 48 h with increasing concentrations of As2O3, as shown. Cell death was assessed at the single cell level by staining cells with annexin-VFITC and counterstaining with PI. (a) Total cell death, (b) apoptosis, (c) necrosis-like cell death. Mean values7s.d. are shown
As2O3-induced cell death C Scholz et al
7037
Figure 11 CD95 blockade does not inhibit cell death induced by As2O3. Jurkat cells (Jurkat A3) were cultured for 48 h in medium alone, with a plastic-bound anti-CD95/Fas antibody (clone 2R2, 5 mg/ml) either alone or in the presence of a soluble, blocking antiCD95 antibody (clone SM1/23, 5 mg/ml), or were treated with 6 mM As2O3 either alone or in the presence of a soluble, blocking antiCD95 antibody (clone SM1/23, 5 mg/ml). Cell death was determined at a single cell level by staining cells with annexin-V-FITC and counterstaining with PI. Dead cells were annexin-V-FITC and/ or PI positive. Mean values7s.d. are shown
Figure 10 Burkitt-like lymphoma cells (BJAB) transfected with a dominant negative FADD mutant stay As2O3-sensitive. BJAB cells either mock-transfected (BJAB mock) or cells stably overexpressing a dominant negative FADD mutant were treated for 48 h with increasing concentrations of As2O3 as indicated. Cell death was measured at the single cell level by labeling cells with annexin-VFITC and counterstaining with PI. (a) Total cell death, (b) apoptosis, (c) necrosis-like cell death. Mean values7s.d. are shown
minor impact of FADD on As2O3-induced apoptosis, we performed functional experiments where we blocked CD95/Fas receptor/ligand interaction. To test this, Jurkat A3 cells were cultured with As2O3 (6 mM) in the presence or absence of a soluble, antiCD95 antibody (IgG2b, clone SM1/23, 5 mg/ml), which was reported to block the induction of apoptosis mediated by anti-CD95 ligation (Klo¨pfer et al., 2004). To validate the effect of the blocking anti-CD95
antibody (clone SM1/23) in our system, we cultured Jurkat A3 cells with plastic-bound, agonistic anti-CD95 antibody (clone 2R2, IgG3, 5 mg/ml) in the absence or presence of the soluble blocking antibody (clone SM1/23) for 48 h. Blocking the CD95/Fas receptor reduced total cell death induced by the agonistic anti-CD95 antibody from 37.971.1 to 8.470.9% (Figure 11). In contrast, total cell death induced by As2O3 amounted to 65.171.7% in the absence compared to 69.073.5% in the presence of the soluble, blocking anti-CD95 antibody (clone SM1/23). Thus, apoptosis triggered through the ligation of the CD95/Fas receptor is not required for As2O3-mediated cell death. As2O3 induces substantial release of cytochrome c into the cytosol but mediates less pronounced activation of caspase-9 To corroborate the notion that As2O3 facilitates cell death independent from caspases-8 and -9, that is, the initiator caspases in the extrinsic and the intrinsic apoptosis pathway, we treated Jurkat mock- or Bcl-2transfected cells for 24 h with As2O3 (4 mM) or epirubicin Oncogene
As2O3-induced cell death C Scholz et al
7038
(2 mg/ml). Epirubicin has been shown to induce cell death through the mitochondrial pathway of apoptosis in a Bax-dependent manner (Hemmati et al., 2002). In the mock-transfected Jurkat cells, epirubicin induced release of cytochrome c into the cytosol and processing of caspase-9 and -3, while no substantial Bid or caspase8 cleavage were observed (Figure 12). As expected, overexpression of Bcl-2 completey blocked cytochrome c release and caspase-9 cleavage in epirubicin-treated Jurkat Bcl-2 cells. Interestingly and in correspondence with recently published data from our group (Scholz et al., 2005), treatment of Jurkat mock cells with As2O3 induced a comparable amount of cytosolic cytochrome c but less pronounced caspase-9 cleavage (Figure 12). Similar to epirubicin-treated cells, no substantial Bid or caspase-8 cleavage was observed. Interestingly, cleavage of caspase-3 was more pronounced following treatment with As2O3 as compared to epirubicin. It is of note, however, that treatment of Jurkat mock cells for 24 h with As2O3 at 4 mM induced substantially more cell death than culture with epirubicin, that is, 52.5% as compared to 13.7% as epirubicin-induced cell death follows a far slower time course and reaches a maximum at 72 h (data not shown). As expected, Jurkat Bcl-2 cells demonstrated neither cytochrome c release nor caspase-9 or -3 processing upon culture with As2O3 (4 mg/ml). Thus, As2O3 induces a regulated, Bcl-2 inhibitable type of necrosis-like cell death which does not rely on
Figure 12 As2O3 facilitates substantial cytochrome c release but comparatively little caspase-9 cleavage. Jurkat cells either mocktransfected or cells stably overexpressing Bcl-2 were incubated for 24 h with medium alone, 4 mM As2O3, or 2 mg/ml epirubicin. Western blot analysis is shown for Bcl-2, cytosolic cytochrome c, mitochondrial cytochrome c, caspase-9, Bid, and caspase-8 processing. The active subunits of caspase-9, -3, and -8 are indicated. The b-Actin control is shown Oncogene
caspase-8 or and involves only moderate caspase-9 activation. The contribution of caspase-3 in this setting remains to be established but should be marginal regarding the limited inhibition of cell death by zVAD-fmk. As2O3 mediates cell death regardless of caspase-9 inhibition To further assess the role of caspase-9 during As2O3induced cell death, we cultured Jurkat A3 cells with 4 mM As2O3 in the absence or presence of 20 or 40 mM of the caspase-9-specific inhibitor z-LEHD-fmk. As a control, cells were treated with 0.01 mg/ml of the cytostatic agent paclitaxel (taxol). While we observed a moderate inhibition of taxol-induced cell death at 24 and 48 h, z-LEHD-fmk had no inhibitory effect on As2O3-mediated cell demise (Figure 13). Finally, we employed Jurkat cells transfected with a dominant negative caspase-9 mutant (caspase-9 DN) that interferes with activation of endogenous caspase-9 (Weinmann et al., 2004). As a control, mock-transfected Jurkat cells transfected were employed. In line with the experiments employing the z-LEHD-fmk caspase-9 inhibitor, As2O3induced cell death was only slightly reduced in Jurkat caspase-9 DN cells as compared to Jurkat mock cells (Figure 14). Therefore, a functional caspase-9 does not seem to be a prerequisite of As2O3-induced cell death. Independence from caspase-9 is also supported by published data obtained in murine caspase-9/ fibro-
Figure 13 Caspase-9 inhibition by z-LEHD-fmk does not block As2O3-mediated cell death. Jurkat cells (Jurkat A3) were cultured for 24 (a) or 48 h (b) in medium alone. Alternatively, cells were treated with 0.01 mg/ml taxol or 4 mM As2O3 in the absence or presence of 20 or 40 mM z-LEHD-fmk. Cell death was determined at a single cell level by staining cells with annexin-V-FITC and counterstaining with PI. Total cell death was determined as apoptotic plus necrotic death. Dead cells were annexin-V-FITC/ PI þ (apoptotic) or PI þ (necrotic). Mean values7s.d. are shown
As2O3-induced cell death C Scholz et al
7039
Figure 14 Jurkat cells transfected with a dominant negative capase-9 cysteinyl mutant stay As2O3-sensitive. Jurkat cells either mock-transfected (mock) or cells stably overexpressing a dominant negative FADD mutant were treated for 24 or 48 h with increasing concentrations of As2O3 as indicated. Cell death was measured at the single cell level by labeling cells with annexin-V-FITC and counterstaining with PI. (a) Total cell death, (b) apoptosis, (c) necrosis-like cell death. Mean values7s.d. are shown
blasts that were fully sensitive to As2O3-induced cell death (Scholz et al., 2005).
Discussion Owing to the impressive effects of As2O3 in newly diagnosed and relapsed acute promyelocytic leukaemia
(APL), mechanisms of As2O3-mediated cell death have recently come under increasing scrutiny. In the present report, we show that As2O3 induces cell death independently of the death receptor pathway and the main constituents of the DISC, that is, CD95, caspase-8, and FADD. Corroborating this finding, As2O3-mediated cell death was only partially inhibited by the broad caspase inhibitor z-VAD-fmk and not at all by the caspase-8 inhibitor z-IETD-fmk. Likewise, blockade of the death receptor CD95 failed to reduce As2O3-induced cell demise. In line with this, As2O3-triggered cell death did not facilitate caspase-8- or Bid-cleavage. While the extrinsic death pathway appears dispensable for As2O3triggered cell death, overexpression of Bcl-2, an antiapoptotic protein that inhibits mitochondrial damage and cytochrome c release in apoptosis, successfully prevented As2O3-induced cell death. In contrast, experiments with the caspase-9-specific inhibitor z-LEHD-fmk and with caspase-9 DN Jurkat cells demonstrated that a functional caspase-9 is not a prerequisite for As2O3mediated cell death. Moreover, z-VAD-fmk only delayed induction of cell death by As2O3, indicating that caspases may contribute but are not decisive for As2O3-induced cell death. This is well in accordance with our findings in other systems showing independence from caspase-2 and -9. This emphasizes a key role for mitochondria during As2O3-mediated cell demise, but suggests that execution of cell death occurs through an apoptosome-independent pathway which does not require a functional caspase-9. As2O3 has been used as a medicinal compound for centuries and has experienced a recent revival for the treatment of APL. The impressive effects on APL initially fostered the notion that As2O3-mediated cell death depends on the availability of the PML-RARa fusion protein, which is typical for APL. However, a number of recent publications have reported As2O3mediated cell death in PML-RARa- and PML-negative cell lines (Akao et al., 1998; Wang et al., 1998; Zhang et al., 1998; Zhu et al., 1999). Recent reports indicated that As2O3 induces upregulation of the death receptor CD95/Fas and its ligand CD95L/FasL (Zhu et al., 2003; Woo et al., 2004). Inhibition of the CD95/Fas death pathway with a blocking antibody to CD95/Fas was, furthermore, reported to decrease As2O3-induced cell death by 20–30% (Zhu et al., 2003). This led the authors to suggest that the ligation of CD95 by CD95L plays a major role in As2O3-mediated cell death (Zhu et al., 2003). This is in sharp contrast to our system where we performed a detailed study on the impact of DISC components in As2O3-induced cell death. Here, treatment of Jurkat A3 cells induced similar amounts of cell death in the absence or presence of a CD95-blocking antibody. When analysing components of the death receptor apoptosis pathway, we show that As2O3induced cell death was largely unaffected in caspase8/ cells, was hardly blocked by the caspase-8-specific inhibitor z-IETD-fmk, and did not result in substantial caspase-8 cleavage as assessed by Western blot. Furthermore, As2O3-mediated cell death was largely undisturbed in FADD-deficient cells and cells transOncogene
As2O3-induced cell death C Scholz et al
7040
fected with a dominant negative mutant of FADD. The limited decrease of As2O3-induced cell death, as observed in Figure 9 and 10, might be related to a putative role of FADD in the death-receptor-independent cleavage of initiator caspases such as caspase-8 (Chandra et al., 2004). In line with this, we recently observed that FADD deficiency slightly reduces the amount of taxol- or fludarabine-induced cell death in a death-receptor-independent fashion (Wieder et al., 2001; von Haefen et al., 2003; Klo¨pfer et al., 2004). Our data, therefore, demonstrate that neither CD95 ligation nor a functional FADD or caspase-8 is necessary for As2O3induced cell death. These findings contradict data where caspase-8 was attributed with a central role during As2O3-induced cell death (Kitamura et al., 2000). However, there, the authors solely reported the disappearance of a caspase-3 cleavage band and a partial reduction of the DCm decrease when adding the caspase8-specific inhibitor z-IETD-fmk at 50 mM to cells treated with As2O3. Such high dosages of peptide inhibitors may affect other caspases and even other protease systems. Furthermore, neither loss of DCm nor caspase-3 cleavage is a specific consequence of caspase-8 activation. Thus, an essential role for caspase-8 in arsenicinduced cell demise was excluded in our systems and remains highly questionable in others. It was in fact intriguing to observe in our data that As2O3-induced cell death was only partially delayed by the broad-spectrum caspase inhibitor z-VAD-fmk, suggesting that As2O3mediated cell death does not functionally depend on caspases. In line with this, the broad-spectrum caspase inhibitor z-VAD-fmk was shown to prevent As2O3mediated nuclear fragmentation, a key sign of caspasedependent apoptosis, while failing to stop DCm reduction and loss of viability (Larochette et al., 1999). This corroborates our finding that As2O3 induces cell death in the absence of functional caspases. In accordance with this, we previously established that As2O3 triggers cell death in the absence of caspase-2 and caspase-9 (Scholz et al., 2005). This in turn coincides with our present data, where As2O3 and the topoisomerase poison epirubicin induced similar amounts of cytosolic cytochrome c release but As2O3 mediated a higher percentage of cell death along with a lower amount of cleaved caspased-9, as compared to epirubicin. Furthermore, neither inhibition of caspase-9 nor usage of caspase-9 DN mutants substantially blocked As2O3mediated cell demise. To our notion, our results do not conflict with recently published data where As2O3 was reported to massively induce apoptosis and to facilitate caspase-8, -9, and -3 processing (Liu et al., 2003). Notably, the authors assessed cell death by staining cells with annexin-V-FITC only without counterstaining with PI. Thus, they failed to distinguish between cells with an apoptosis- and a necrosis-like phenotype and supposedly missed the often substantial proportion of necrotic, that is, annexin-V-FITC-negative, but nevertheless PI-positive cells. While our present data alongside with our previously reported results (Scholz et al., 2005) are compatible with the idea of a limited As2O3mediated caspase activation, we emphasize that As2O3 Oncogene
does not depend on functional initator caspases, that is, caspase-8, -2, or -9 for effective execution of cell death. In contrast, constitutive expression of Bcl-2 completely blocked As2O3-mediated cell death, mitochondrial permeability transition, and cytochrome c release. This is in line with data where constitutive expression of Bcl-2 in myelomonocytic cells inhibited As2O3-induced cell demise (Larochette et al., 1999). In conclusion, we show here that As2O3 triggers a regulated, but predominantly necrotic mode of cell death that occurs independently of CD95 ligation, caspase-8, and FADD expression and does not require a functional caspase-9. The complete blockade of As2O3-mediated cell death in cells with constitutive overexpression of Bcl-2 corroborates the prominent role of the mitochondrial pathway during As2O3-induced cell death.
Materials and methods Reagents As2O3 was purchased from Sigma-Aldrich (Taufkirchen, Germany). As2O3 (1 mg) was dissolved in 1 ml 1 M sodium hydroxide (NaOH) solution to generate a 5 mM stock solution, which was serially diluted in RPMI 1640 supplemented with 100 U/ml penicillin, 100 mg/ml streptomycin, 4 mM L-glutamine, and 10% fetal calf serum (GibcoBRL, Life Technologies, Karlsruhe, Germany), further depicted as complete medium. Cell lines Jurkat cells stably overexpressing Bcl-2 (Jurkat Bcl-2) and the corresponding mock-transfectants (Jurkat mock) have been described in detail before (Belka et al., 2000). The caspase-8/ Jurkat cell line (Jurkat A3/C-8/) and the parental (Jurkat A3) line have been introduced elsewhere (Engels et al., 2000). BJAB cells overexpressing a dominant-negative FADD mutant have been described recently (Klo¨pfer et al., 2004). Jurkat cells deficient for FADD or the parental cell line (Jurkat A3) were kindly provided by J Blenis (Harvard Medical School, Boston, MA, USA) and have been characterized before (Klo¨pfer et al., 2004). Jurkat cells transfected with a dominant negative mutant of caspase-9 (Jurkat caspase-9 DN) or the vector control cell line were provided by C Belka, University of Tu¨bingen (Belka et al., 2003; Jendrossek et al., 2003a, b; Klo¨pfer et al., 2004; Weinmann et al., 2004; Rudner et al., 2005). Prior to the experiments, cells were cultured overnight in complete medium, at a concentration of 4 105 to allow logarithmic growth. Subsequently, cells were recounted, seeded at a denisty of 1 105 cells/ml and cultured with medium alone, As2O3, agonistic anti-CD95/Fas antibody antiAPO-1 (IgG3) (Dhein et al., 1992), agonistic anti-APO-1/Fas clone 2R2 (IgG3) (Bender MedSystems, Vienna, Austria), or the anti-CD95/Fas-blocking antibody anti-APO-1/Fas clone SM1/23 (IgG2b) (Bender MedSystems, Vienna, Austria), at concentrations and time periods as indicated. Where stated, the irreversible, broad caspase inhibitor z-VAD-fmk, the caspase-8 inhibitor z-IETD-fmk, or the caspase-9 inhibitor z-LEHD-fmk (Calbiochem, Bad Soden, Germany) were employed at the respective concentrations 1 h before the addition of cytotoxic agents. For longer incubation periods, the inhibitors were replenished every 24 h. The maximal concentration of DMSO present when the caspase inhibitor
As2O3-induced cell death C Scholz et al
7041 z-VAD-fmk, z-IETD-fmk, or z-LEHD-fmk was used had no effect on cell growth or cell death. Measurement of cell death by assessing annexin-V-FITC and PI Cell death was determined by staining cells with annexin-VFITC and counterstaining with PI. Annexin-V-FITC binds to phosphatidylserine (PS) on the outer leaflet of the plasma membrane. PI is excluded by cells with intact membranes. PI positivity is therefore a sign of necrosis, whereas cells positive for annexin-V, but negative for PI are generally defined as apoptotic (Vermes et al., 1995). Briefly, to determine apoptosis, cells were washed twice with cold PBS and resuspended in binding buffer (10 mM N-[2-hydroxyethyl]piperazin-N0 -3[propansulfonicacid]/NaOH, pH 7.4, 140 mM NaCl, 2.5 mM CaCl2) at a concentration of 1 106 cells/ml. Next, 5 ml of annexin-VFITC (BD PharMingen, Heidelberg, Germany) and 10 ml PI (20 mg/ml, Sigma-Aldrich) were added. Analyses were performed using a FACScan (Becton Dickinson, Heidelberg, Germany) and CellQuest analysis software. Assessment of genomic DNA fragmentation by flow cytometric measurement of nuclear DNA content To measure genomic DNA fragmentation, nuclear DNA content was determined by flow cytometry as described (Daniel et al., 1998). Briefly, 105 cells were pelleted in a 96-well U-bottom plate. The pellet was then gently resuspended with 200 ml buffer (50 mg/ml PI, 0.1% sodium citrate, 0.1% Triton X-100) and incubated overnight in the dark at 41C. Genomic DNA content was determined using a FACScan flow cytometer (Becton Dickinson, Heidelberg, Germany) and CellQuest analysis software. Nuclei displaying a hypodiploid, sub-G1 DNA content were identified as apoptotic. Cell debris, characterized by low forward and side scatter values, was excluded from the analysis. Immunoblotting After treatment under the respective conditions, cells were washed twice with PBS and lysed in buffer containing 10 mM Tris/HCl pH 7.5, 300 mM NaCl, 1% Triton X-100, 2 mM MgCl2, 5 mM EDTA, 1 mM pepstatin, 1 mM leupeptin, and 0.1 mM phenylmethylsulfonyl fluoride. Protein concentration was determined using the bicinchoninic acid assay (Pierce, Rockford, IL, USA), and equal amounts of protein (usually
20 mg per lane) were separated by SDS–PAGE. Immunoblotting was performed as described (Wieder et al., 2001). Membranes (Schleicher & Schuell, Dassel, Germany) were swollen in CAPS-buffer (10 mM 3-[cyclohexylamino]propane1-sulfonic acid, pH 11, 10% MeOH), and blotting was performed at 1 mA/cm2 for 1 h in a transblot SD cell (BioRad, Munich, Germany). The membrane was blocked for 1 h in PBST (PBS, 0.05% Tween-20) containing 3% nonfat dry milk and incubated with the primary antibody for 1 h. As primary antibodies, either mouse monoclonal anti-human Bcl2 antibody (Novocastra, Newcastle, UK), mouse monoclonal anti-human-cytochrome c antibody (BD PharMingen, Heidelberg, Germany), goat polyclonal anti-human caspase-9 antibody (R&D Systems, Wiesbaden, Germany), goat polyclonal anti-human Bid antibody (Santa Cruz, Heidelberg, Germany), mouse monoclonal anti-human caspase-8 antibody (Cell Signaling Technology, Beverly, MA, USA), or polyclonal rabitt anti-human b-actin antibody (Sigma-Aldrich, Taufkirchen, Germany) were employed. After washing in PBST, the respective secondary goat-anti-mouse and rabbit-anti-goat antibodies conjugated to horseradish peroxidase (Southern Biotechnology Associates, Birmingham, AL, USA) were applied for 1 h. Finally, the membrane was washed in PBST, and protein bands were detected using the enhanced chemiluminescence system (Amersham Buchler, Braunschweig, Germany). To detect cytochrome c released from mitochondria into the cytosol, cells were pelleted by centrifugation at 300 g for 5 min and resuspended in lysis buffer containing 20 mM HEPES pH 7.4, 10 mM KCl, 2 mM MgCl2, 1 mM EDTA, 100 mM DTT, 0.1 mM PMSF, and 0.75 mg/ml digitonin. Following incubation for 30 min on ice, cell lysates were centrifuged at 13 000 g at 41C for 15 min. Supernatants containing cytosolic protein were processed as described above. Statistics All experiments were performed at least three times. Results are expressed as means7standard deviation (s.d.). Acknowledgements This work was supported by a grant SCHO 534/2-1 from the Deutsche Forschungsgemeinschaft and a grant DJCLS-R01/02 from the Deutsche Jose´ Carreras Stiftung.
References Akao Y, Mizoguchi H, Kojima S, Naoe T, Ohishi N and Yagi K. (1998). Br. J. Haematol., 102, 1055–1060. Belka C, Gruber C, Jendrossek V, Wesselborg S and Budach W. (2003). Oncogene, 22, 176–185. Belka C, Rudner J, Wesselborg S, Stepczynska A, Marini P, Lepple-Wienhues A, Faltin H, Bamberg M, Budach W and Schulze-Osthoff K. (2000). Oncogene, 19, 1181–1190. Boatright KM, Renatus M, Scott FL, Sperandio S, Shin H, Pedersen IM, Ricci JE, Edris WA, Sutherlin DP, Green DR and Salvesen GS. (2003). Mol. Cell, 11, 529–541. Chandra D, Choy G, Deng X, Bhatia B, Daniel P and Tang DG. (2004). Mol. Cell. Biol., 24, 6592–6607. Daniel PT, Kroidl A, Kopp J, Sturm I, Moldenhauer G, Do¨rken B and Pezzutto A. (1998). Blood, 92, 4750–4757. Daniel PT, Schulze-Osthoff K, Belka C and Gu¨ner D. (2003). Essays Biochem., 39, 73–88. Daniel PT, Wieder T, Sturm I and Schulze-Osthoff K. (2001). Leukemia, 15, 1022–1032.
Dhein J, Daniel PT, Trauth BC, Oehm A, Moller P and Krammer PH. (1992). J. Immunol., 149, 3166–3173. Engels IH, Stepczynska A, Stroh C, Lauber K, Berg C, Schwenzer R, Wajant H, Janicke RU, Porter AG, Belka C, Gregor M, Schulze-Osthoff K and Wesselborg S. (2000). Oncogene, 19, 4563–4573. Fischer U, Janicke RU and Schulze-Osthoff K. (2003). Cell Death Differ., 10, 76–100. Hemmati PG, Gillissen B, von Haefen C, Wendt J, Starck L, Gu¨ner D, Do¨rken B and Daniel PT. (2002). Oncogene, 21, 3149–3161. Hossain K, Akhand AA, Kato M, Du J, Takeda K, Wu J, Takeuchi K, Liu W, Suzuki H and Nakashima I. (2000). J. Immunol., 165, 4290–4297. Jendrossek V, Handrick R and Belka C. (2003a). FASEB J., 17, 1547–1549. Jendrossek V, Muller I, Eibl H and Belka C. (2003b). Oncogene, 22, 2621–2631. Oncogene
As2O3-induced cell death C Scholz et al
7042 Kischkel FC, Hellbardt S, Behrmann I, Germer M, Pawlita M, Krammer PH and Peter ME. (1995). EMBO J., 14, 5579–5588. Kischkel FC, Lawrence DA, Chuntharapai A, Schow P, Kim KJ and Ashkenazi A. (2000). Immunity, 12, 611–620. Kitamura K, Minami Y, Yamamoto K, Akao Y, Kiyoi H, Saito H and Naoe T. (2000). Leukemia, 14, 1743–1750. Klo¨pfer A, Hasenjager A, Belka C, Schulze-Osthoff K, Do¨rken B and Daniel PT. (2004). Oncogene, 23, 9408–9418. Larochette N, Decaudin D, Jacotot E, Brenner C, Marzo I, Susin SA, Zamzami N, Xie Z, Reed J and Kroemer G. (1999). Exp. Cell Res., 249, 413–421. Liu Q, Hilsenbeck S and Gazitt Y. (2003). Blood, 101, 4078–4087. Niu C, Yan H, Yu T, Sun HP, Liu JX, Li XS, Wu W, Zhang FQ, Chen Y, Zhou L, Li JM, Zeng XY, Yang RR, Yuan MM, Ren MY, Gu FY, Cao Q, Gu BW, Su XY, Chen GQ, Xiong SM, Zhang T, Waxman S, Wang ZY and Chen SJ. (1999). Blood, 94, 3315–3324. Perkins C, Kim CN, Fang G and Bhalla KN. (2000). Blood, 95, 1014–1022. Peter ME and Krammer PH. (2003). Cell Death Differ., 10, 26–35. Rudner J, Jendrossek V, Lauber K, Daniel PT, Wesselborg S and Belka C. (2005). Oncogene, 24, 130–140. Scaffidi C, Fulda S, Srinivasan A, Friesen C, Li F, Tomaselli KJ, Debatin KM, Krammer PH and Peter ME. (1998). EMBO J., 17, 1675–1687. Scholz C, Wieder T, Starck L, Essmann F, Schulze-Osthoff K, Do¨rken B and Daniel PT. (2005). Oncogene, 24, 1904–1913. Shen ZX, Chen GQ, Ni JH, Li XS, Xiong SM, Qiu QY, Zhu J, Tang W, Sun GL, Yang KQ, Chen Y, Zhou L, Fang ZW, Wang YT, Ma J, Zhang P, Zhang TD, Chen SJ, Chen Z and Wang ZY. (1997). Blood, 89, 3354–3360.
Oncogene
Sohn D, Schulze-Osthoff K and Janicke RU. (2005). J. Biol. Chem., 280, 5267–5273. Soignet SL, Maslak P, Wang ZG, Jhanwar S, Calleja E, Dardashti LJ, Corso D, DeBlasio A, Gabrilove J, Scheinberg DA, Pandolfi PP and Warrell Jr RP. (1998). N. Engl. J. Med., 339, 1341–1348. Vermes I, Haanen C, Steffens-Nakken H and Reutelingsperger C. (1995). J. Immunol. Methods, 184, 39–51. von Haefen C, Gillissen B, Hemmati PG, Wendt J, Gu¨ner D, Mrozek A, Belka C, Do¨rken B and Daniel PT. (2004). Oncogene, 23, 8320–8332. von Haefen C, Wieder T, Essmann F, Schulze-Osthoff K, Do¨rken B and Daniel PT. (2003). Oncogene, 22, 2236–2247. Wang ZG, Rivi R, Delva L, Konig A, Scheinberg DA, Gambacorti-Passerini C, Gabrilove JL, Warrell Jr RP and Pandolfi PP. (1998). Blood, 92, 1497–1504. Weinmann M, Jendrossek V, Handrick R, Gu¨ner D, Goecke B and Belka C. (2004). Oncogene, 23, 3757–3769. Wieder T, Essmann F, Prokop A, Schmelz K, Schulze-Osthoff K, Beyaert R, Do¨rken B and Daniel PT. (2001). Blood, 97, 1378–1387. Woo SH, Park IC, Park MJ, An S, Lee HC, Jin HO, Park SA, Cho H, Lee SJ, Gwak HS, Hong YJ, Hong SI and Rhee CH. (2004). Int. J. Cancer, 112, 596–606. Zhang W, Ohnishi K, Shigeno K, Fujisawa S, Naito K, Nakamura S, Takeshita K, Takeshita A and Ohno R. (1998). Leukemia, 12, 1383–1391. Zhu J, Okumura H, Ohtake S, Nakamura S and Nakao S. (2003). Oncol. Rep., 10, 705–709. Zhu XH, Shen YL, Jing YK, Cai X, Jia PM, Huang Y, Tang W, Shi GY, Sun YP, Dai J, Wang ZY, Chen SJ, Zhang TD, Waxman S, Chen Z and Chen GQ. (1999). J. Natl. Cancer Inst., 91, 772–778.