Study of Apoptosis-Related Responses of Leukemic Blast Cells to in Vitro

Leukemia (2000) 14, 1266–1275  2000 Macmillan Publishers Ltd All rights reserved 0887-6924/00 $15.00 www.nature.com/leu Study of apoptosis-related responses of leukemic blast cells to in vitro anthracycline treatment M-A Belaud-Rotureau1,2, F Durrieu1, G Labroille2, F Lacombe1,3, O Fitoussi4, P Agape4, G Marit2,4, J Reiffers3,4 and F Belloc1,2

1Laboratoire d’He´matologie, Hoˆpital du Haut-Le´veˆque, Pessac; 2Laboratoire Universitaire d’He´matologie, Universite´ Victor Segalen, Bordeaux; 3CNRS-UMR 5540, Universite´ Victor Segalen, Bordeaux; and 4Service des Maladies du Sang, Hoˆpital du Haut-Le´veˆque, Pessac, France

Anthracyclines trigger an apoptotic cell death but their molecu- The increase in Fas on tumor cells after low-dose chemo- lar targets are not totally explored. We investigated the apop- therapy may be used as a therapeutic target. Because a syn- totic response of blast cells and lymphocytes from medullary samples of 31 de novo acute leukemia. Mononuclear cells were ergy between Fas-ligand and cytotoxic drugs was observed in treated in vitro by therapeutic concentrations of either dauno- vitro, the Fas increase could enhance tumor cell elimination rubicin (DNR) or idarubicin (IDA) for 1 h, washed and cultured by the immune cells.17 These observations supported the for 18 h. A multivariate analysis using flow cytometry and a rationale for a new therapeutic strategy which could combine CD45 gating on lymphocytes and blast cells was performed. chemotherapy and immunotherapy for cancer treatment.14,18 DNR and IDA induced a Fas enhancement on both leukemic For many years, the anthracyclines have been used in leu- and normal cells. In blast cells the DEVDases were activated kemia treatment and it is known that anthracyclines induce and the caspase 3 was cleaved in relation to phosphatidyl 19–21 serine exposure, showing a caspase-dependent pathway in an apoptotic cell death in a wide range of cultured cells. anthracycline-induced apoptosis. Apoptotic percentages were Therapeutic daunorubicin (DNR) concentrations are about 1– always higher for blast cells than for lymphocytes, confirming 2 µM.22 At these doses, DNR induces apoptosis in U937 and that anthracycline toxicity mainly affected tumor cells. More- HL60 leukemic cell lines.23,24 The Fas apoptotic pathway over, drug-induced apoptosis was not related to spontaneous could at least partially play a role,12–16 involving both death apoptosis, suggesting that variations in response intensities 17,25–27 were due to individual variations of sensitivity rather than to receptor/ligand accumulation on the cell surface and programmed life span time. The apoptotic response of P- activation of the death effectors such as caspases in the cyto- glycoprotein-expressing blast cells was not significant, giving plasm.28 Expression of the mdr-1 gene leads to the multidrug biological argument for the poor prognosis of multidrug resistance phenotype, which is one of the major cause of fail- resistance leukemia. Finally, Fas induction and anthracycline- ure in cancer therapy.29,30 This gene encodes for a 170 kDa induced apoptosis on blast cells were significantly higher when ATP binding glycoprotein (PGP) which acts as a pump to a complete remission was achieved, thus shedding light on 31 potential new prognostic factors in acute leukemia. Leukemia exclude therapeutic agents from the cell, thus decreasing the (2000) 14, 1266–1275. intracellular concentration of the drug and the apoptotic cell Keywords: anthracycline; apoptosis; leukemia; Fas; caspase response.32 However, the mechanisms mediating the anthracycline-induced killing of leukemic blast cells are not defini- tively known and a possible correlation between the apoptotic Introduction potential of blast cells and the clinical outcome of patients has never been explored. Fas (APO-1, CD95) is a tumor necrosis factor family membrane In the present work, we investigated the apoptotic response protein which triggers programmed cell death when engaged of in vitro anthracycline-treated leukemic blast cells and nor- by anti-Fas antibody or Fas-ligand.1 The death pathway mal lymphocytes isolated from medullary samples of 31 initiated by Fas activation involves a series of death-induced patients suffering from de novo acute leukemia (AL). Some molecules.1 Fas-associating protein with death domain (FADD) mechanisms of cellular drug resistance (PGP expression, or MORT-1 is recruited to Fas upon its engagement.2,3 FADD apoptotic deficiency) and others of cell death (Fas expression, then binds FADD-like ICE (FLICE) or MORT-1-associated CED- DEVDase activation, caspase 3 activation, phosphatidyl serine 3 homologue.4,5 The association with the Fas death-inducing exposure) were assayed. A possible correlation between these signaling complex activates FLICE,6 followed by activation of parameters and the clinical outcome of patients was then the caspase proteolytic cascade and apoptosis.7 Fas is found on examined. We clearly show that both anthracyclines (DNR immunity cells (lymphocytes, NK cells, monocytes) but also on and idarubicin (IDA)) induce apoptosis in leukemic blast cells numerous tumor cells such as tumor cell lines or leukemic blast and lymphocytes. This death was accompanied by a Fas cells.8,9 Its expression is decreased during tumor cell enhance- expression enhancement on both lymphocytes and blast cells. ment or during the cell acquisition of the drug resistance Caspase 3 was cleaved and consequently activated and the phenotype.10,11 caspase 3 cleavage was correlated with phosphatidyl serine Several cytotoxic drugs (cisplatin, doxorubicin, bleomycin, exposure. When PGP was expressed on the cell membrane, mitomycin, methotrexate) as well as ionizing radiation can the apoptotic potential of blast cells was strongly decreased, sensitize many tumor cells to Fas-induced apoptosis.12–16 This giving biological argument for the poor prognosis of PGP- sensitization is mediated by a transcriptional and a post- positive acute leukemia. The apoptotic response intensity of transcriptional regulation of the fas gene, which leads to blast cells was significantly associated with the clinical out- receptor accumulation on the drug-treated membrane cells. come of patients, thus shedding light on potential new prognostic factors in acute leukemia.

Correspondence: F Belloc, De´partement de Cytome´trie en Flux, Patients and methods Laboratoire d’He´matologie, Hoˆpital Haut-Le´veˆque, 33604 Pessac Cedex, France; Fax: 33 556 55 68 09 The study included 30 patients with previously untreated de Received 25 November 1999; accepted 24 February 2000 novo AL who were treated in our institution between Nov- Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1267 ember 1998 and August 1999. Another patient who relapsed (Gibco-BRL), 100 units/ml penicillin (Gibco-BRL), 50 µg/ml 1 year after the end of chemotherapy was included. One streptomycin (Gibco-BRL). Finally, the cell concentration was patient was not followed up for his clinical evolution because adjusted to 1 × 106 cells/ml. he was treated in another institution. All the patients gave informed consent. Their disease was classified according to the recommendations of the French– Cell incubation with daunorubicin or idarubicin American–British (FAB) committee.33 Twenty-four acute myeloid leukemia (AML) and seven acute non-myeloid leuke- The cell suspension was distributed in three tubes. One was mia (ANML) were found. According to the type of AL, the incubated without anthracyclines (NT) for 1 h at 37°C, the induction chemotherapy varied but all patients received an second was incubated with 1 µM daunorubicin (DNR, RPR- anthracycline, either DNR or IDA. On September 1999, 15 Bellon, Neuilly, France) for 1 h at 37°C and the third was patients had survived in complete remission (SCR) and 10 incubated with 0.2 µM idarubicin (IDA, Pharmacia, Saint patients died after treatment failure (DTF). The remaining were Quentin en Yvelines, France) for 1 h at 37°C. Then, BM-MNC not classified because no post-mortem biopsies were done or were washed, resuspended in CM and cultured for 18 h in ° because there was not enough distance from the first chemo- humidified 95% O2 and 5% CO2 atmosphere at 37 C. therapy induction. Complete remission was defined as the presence of fewer than 5% blast cells in a cellular BM smear, and the absence of circulating blast cells and extramedullary Cellular response to anthracycline exposure leukemic cell infiltration. The clinical and hematological characteristics are summarized in Table 1. Flow cytometry (FCM) analysis: Fas and PGP expression, apoptosis percentage, and caspase 3 activation were assayed by FCM with an EPICS XL cytometer (Coulter, Margency, BM collection and preparation France). The CD45 differential expression allowed a selective gating of lymphocytes and blast cells34 and the cell response Before induction chemotherapy, BM aspirates were collected to anthracycline exposure was thus assessed simultaneously for diagnosis and 1 ml was sampled on sodium heparinate in these two cellular populations. tubes (Roche, France). Bone marrow mononuclear cells (BM- After centrifugation (5 min, 400 g), 5 × 105 cells from NT, MNC) were isolated by centrifuging half diluted BM over Lym- DNR and IDA suspensions were resuspended in 100 µl phos- phoprep (Nyegeaard, Oslo, Norway) for 40 min at 400 g. BM- phate buffer saline (PBS, Eurobio, France) supplemented with MNC were washed in RPMI 1640 (Gibco-BRL, Eragny, 10% FCS (PBS-FCS) and labeled with 5 µl of PC5-conjugated France) and resuspended in culture medium (CM) consisting anti-CD45 antibodies (Coulter-Immunotech), 5 µl of FITC- of RPMI 1640 medium supplemented with 10% fetal calf conjugated anti-Fas antibodies (clone UB-2, Immunotech, serum (FCS) and 1 mML-glutamine (Gibco-BRL), 10 mM Hepes France), and 5 µl of PE-conjugated anti-PGP antibodies (clone UIC2, Immunotech, France). Negative controls were estab- lished with corresponding FITC or PE-conjugated-isotypic Table 1 Clinical and hematological characteristics of patients irrelevant IgG and PC5-conjugated anti-CD45 antibodies. Erythrocytes were lyzed with multi-Q-Prep (Coulter- Total SCR DTF Immunotech) according to the manufacturer’s instructions and analysis was performed by FCM. Sex The mean fluorescence intensity (MFI) was measured in Male 19 9 6 Female 12 6 4 lymphocyte and blast cell populations of NT, DNR and IDA samples and the specific mean fluorescence intensity (MFIsp) FAB relating Fas or PGP expression was calculated by subtracting AML Unclassified 3 0 3 negative control values. Fas or PGP expressions were con- M0 2 2 0 sidered as positive when the MFIsp was found positive. Such M1 7 3 2 an interpretation was allowed by gating on the homo- M2 4 2 1 geneous population. M3 1 0 1 BM-MNC (5 × 105) from NT, DNR and IDA suspensions M4 2 0 2 were incubated in 100 µl culture medium containing 5 µlof M5 4 1 1 M6 1 1 0 PC5-conjugated anti-CD45 antibodies (Coulter-Immunotech) for 20 min. The FITC-annexin V kit (Coulter-Immunotech) was ANML used after modification of the manufacturer’s specifications: L1 4 3 0 binding buffer (500 µl) containing 5 µl FITC-annexin V L2 2 2 0 L3 1 1 0 (Coulter-Immunotech) was added and further incubated for 20 min on ice. The addition of propidium iodide which is usually Age (years) proposed to stain permeabilized dead cells was omitted Mean ± s.d. 54 ± 18 47 ± 20 65 ± 9 Median 58 52 68 because the cells were further fixed, permeabilized and Range 17–79 17–75 51–79 stained with a PE-conjugated antibody whose fluorescence spectrum overlaps the propidium iodide spectrum. The cells Leukocytes (G/I) Mean ± s.d. 32 ± 81.4 26.1 ± 67.5 41.4 ± 104.2 were then pelleted, resuspended in 1 ml Permeafix (Ortho Median 3.6 3.5 3.3 Diagnostic Systems, Roissy, France) and incubated for 40 min Range 1.1–366.0 1.1–290 1.1–366.0 at 20°C. The suspension was then centrifuged and the pellet was washed twice with washing buffer (0.2 mM EDTA, 5% SCR, survived in complete remission; DTF, died after treatment FCS in phosphate buffer saline (PBS)). Labeling was performed failure. by adding to the cells 100 µl of washing buffer containing 5 µl

Leukemia Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1268 of polyclonal PE-conjugated anti-active caspase 3 antibodies PGP more frequently found on lymphocytes than on (Pharmingen-Becton Dickinson, Le Pont de Claix, France). blast cells The samples were gently stirred for 1 h, and washed in washing buffer. A negative control sample incubated with PE- PGP expression is able to confer multidrug resistance pheno- conjugated IgG was run in parallel. Then, FCM analysis was type36 and so is likely to modify the cellular response to performed at 525 nm for annexin binding, 575 nm for cleaved anthracyclines. As was done for Fas expression, PGP caspase 3 labeling and 675 nm for CD45 expression. The expression was assessed by FCM on blast and lymphocytes FITC-annexin V binding cells were considered as apoptotic cells as described in Patients and methods. Interestingly, 29% and the negative cells as viable. From the same analysis, posi- of our AML were positive for PGP expression on lymphocytes tive and negative cell populations for the active caspase 3 although only 13% expressed PGP on blast cells (Figure 1e). were individualized. Thus, studying PGP expression on total MNC could lead to overestimating the percentage of PGP-positive AL. Therefore, a specific analysis of blast cells, as individualized by low CD45 expression, appeared necessary. Isotypic control (Figure Fluorimetric analysis: After CD45 labeling as described 1b, right) and PGP-labeled (Figure 1c, right) histograms were above, blast cells were sorted from NT, DNR, and IDA almost directly superimposed and only a small population was samples with an ELITE cell sorter (Coulter). The cell flow was positive after PGP labeling, so only a part of blast and lympho- set to analyze 2000 events/s, and cells in the windows of inter- cyte cell populations expressed PGP in some of the AL. Bivari- est were sorted in sterile tubes containing 0.5 ml PBS. Sorted ate histogram analysis (Figure 1b and c, middle) showed that cells (5 × 105) were washed in PBS and processed for PGP-positive cells expressed Fas as well as PGP-negative DEVDase activity as described below. cells. DEVDase activity was evaluated as previously described.24,35 Briefly, 3 × 105 cells were washed in PBS, then suspended in 50 µl of a pH 7.4 permeabilizing buffer contain- IDA induces Fas expression on leukemic cells ing 10 mM HEPES, 5 mM dithiothreitol, 0.02% Saponine, 1 mM PMSF, 10 µg/ml Pepstatin A, 10 µg/ml Leupeptin, 72 µM The variation in Fas expression on anthracycline-treated cells fluorogenic substrate Ac-DEVD aminomethylcoumarin (UBI, was measured by FCM. The results showed that IDA, but not Euromedex, Souffelweyersheim, France). Cells were incu- DNR, led to a significant increase in Fas expression on blast bated for 10 min at 37°C and then centrifuged for 3 min at cells of total studied AL (Figure 2a) and on PGP negative blast 14000 r.p.m. The supernatant was diluted in 1 ml dH O. Flu- 2 cells of AML (Figure 2b) and ANML (Figure 2c). Fas was not orescence was measured using a Jobin and Yvon spectrofluo- induced when PGP was expressed (Figure 2b), suggesting that rometer (λ = 380 nm, λ = 480 nm). A blank was perfor- exc em Fas expression may be related to the intracellular anthracyc- med without addition of cells in the substrate buffer and its line accumulation. Induction of the expression of Fas on value was subtracted from all the measurements. A control anthracycline-treated tumor cells could sensitize these cells to without any drug treatment was processed for each experi- apoptosis.17 As caspase activation is considered to be a com- ment. The relative enzyme activity was expressed as the ratio mon feature in apoptosis,37 it was interesting to verify if the of treated sample activity to untreated control activity or as effect of anthracyclines was accompanied by activation of fluorescence units per 105 sorted blast cells. caspases. DEVDase activity was therefore measured on sorted blast cells.

Statistical analysis Anthracyclines activate caspase 3 and induce The paired Student’s t-test and the Chi-square test were used DEVDase activity in leukemic blast cells to analyze the data. Blast cells were isolated by flow cell sorting and the DEVDase activity was measured by fluorimetry in anthracycline-treated and control samples as described in Patients and methods. Results Both DNR and IDA led to a significant activation of DEVDase in AL blast cells (Figure 3a). The increase in blast cells Fas expressed by blast cells and lymphocytes in all DEVDase values after anthracycline treatment was nearly two- the AL fold compared to the control (Figure 3a) and the DEVDases were similarly activated after both DNR and IDA treatment. Because the Fas apoptotic pathway is involved in anthracyc- Interestingly, when PGP was expressed on AML blast cells, no line-induced apoptosis,17 Fas expression was assessed on the DEVDase activation occurred (Figure 3b). As for Fas membranes of blast and lymphocyte cells. Anthracycline- expression, the DEVDases activation may be related to the treated and control cells were labeled with an anti-CD45 and intracellular anthracycline accumulation. an anti-Fas antibody as described in the Patients and methods Because caspase 3 is the main effector of the caspase family section and analyzed by FCM after blast cell or lymphocyte and belongs to the DEVDase family, the specific activation of CD45 gating (Figure 1a). Almost all the AL were found positive this enzyme was assayed by FCM after anthracycline treat- for Fas expression on both blast and lymphocyte cells (Figure ment. During apoptosis, the enzyme is activated by the cleav- 1d). Moreover, whatever the gating on blast or lymphocyte age of the procaspase 3 zymogen. This activation was evalu- cells, histogram analysis after anti-Fas labeling (Figure 1c, left) ated by FCM both in lymphocytes and blast cells after CD45 was clearly right-shifted in comparison with the isotypic con- gating using a specific anti-cleaved-caspase 3 antibody as trol (Figure 1b, left), so the whole blast cell and lymphocyte described in Patients and methods (Figure 3c). Anthracycline populations expressed Fas. (DNR and IDA) treatment triggered caspase 3 activation in the

Leukemia Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1269

Figure 1 Percentage of AL or AML expressing Fas or PGP on lymphocytes and blast cells. MNC were labeled with a PC5-anti-CD45 antibody, a FITC-anti-Fas, and a PE-anti-PGP antibody. FCM analysis was performed. Bivariate CD45-SSc, Fas-PGP and the Fas- or PGP-fluorescence histograms are shown. (a) Gating based upon the differential CD45 expression allowed the simultaneous study of Fas and PGP on lymphocytes and blast cells. CD45 fluorescence is on x axis and side scatter (SSc) on the y axis. Examples of histograms gated on blast cells are represented respectively for isotypic controls (b) and positive labeling (c). The FITC (Fas-related) fluorescence histograms are on the left, the PE (PGP-related) histograms on the right and the bivariate (FITC/PE) histograms in the middle of the Figure. The mean fluorescence intensity of isotypic labeling was subtracted from the corresponding mean fluorescence intensity after Fas or PGP labeling and constituted the specific mean fluorescence intensity (MFIsp). A cell population was considered to express Fas or PGP when its MFIsp was positive (b, c). The frequencies (in percentages) of acute leukemias (AL) expressing Fas (d) or PGP (e) on their blast cells and/or on their lymphocytes are shown.

Leukemia Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1270 as described in Patients and methods. DNR and IDA induced apoptosis both in lymphocytes and in blast cells of total AL (Figure 4b). When PGP was expressed on AML blast cells, no signiﬁcant apoptosis was measured beyond spontaneous apoptosis (Figure 4c). This observation reinforces the idea that anthracycline-induced apoptosis is related to intracellular drug accumulation and gave supplementary evidence for the deleterious role of PGP in AL. As noted for caspase 3 activation, there were fewer apoptotic lymphocytes than blast cells in both untreated and anthracycline-treated samples (data not shown). This again suggested that lymphocytes could be more resistant to cyto- static agents and have a longer in vitro life span than blast cells. Both DNR and IDA induced similar percentages of apoptotic cells. Interestingly, caspase 3 cleavage and annexin V binding were strongly related in blast cells and in lymphocytes (Figure 4d). Moreover, bivariate analysis showed that activated caspase 3 positive cells were also positive for annexin V binding (Figure 4a). These data strongly support the hypothesis that anthracyclines induce apoptosis through the caspase pathway. Moreover, drug-induced apoptosis was not related to spontaneous apoptosis in either blast cells or lymphocytes (data not shown) suggesting that anthracyclines induce apoptosis by a speciﬁc mechanism, independently of the cell life span.

Apoptotic potential of blast cells higher for SCR patients

The apoptotic potential of blast cells was compared with the patients clinical evolution. Fas induction on blast cells was significantly higher for SCR patients compared to DTF patients (Figure 5a). Moreover, a statistical analysis using the chi- square test showed that a Fas-induction higher than 50% on Figure 2 Fas expression on leukemic blast cells after anthracycline blast cells was more frequently observed for SCR patients (ξ2 exposure. MNC were treated by either DNR (1 µM) or IDA (0.2 µM) , 0.05). Similarly, blast cell apoptosis was greater for SCR or non-treated (NT) and labeled for CD45, Fas and PGP as described patients (Figure 5b). This difference was significant for the in Patients and methods. FCM analysis was performed. For each sam- Student’s t-test only after IDA treatment. An IDA-induced- ple, the Fas expression on blast cells was calculated as the MFIsp and the mean Fas expression was plotted for the AL (a), for the AML apoptosis higher than 20% was more frequently found in SCR ξ2 , expressing or not the PGP on blast cells (b), and for the ANML (c). patients ( 0.05). Finally, no correlation was found The s.d. are indicated. * to n*: P , 5.10−2 to P , 5.10−(n+1) for the between leukocytosis on the day of diagnosis and the apop- paired Student’s t-test. totic potential of blast cells (data not shown). Leukocytosis reflecting a ratio between apoptosis and proliferation thus lymphocytes and blast cells of 10 AL (Figure 3d). The percent- appeared independent of the potentiality of blast cells to age of positive cleaved caspase 3 lymphocytes was signifi- undergo apoptosis under anthracycline treatment. cantly lower than the percentage of blast cells in control samples (P , 5.10−4, Figure 3d), suggesting that the life span was higher for lymphocytes. Moreover, this effect was strictly Discussion reproducible for all anthracycline treated samples, thus suggesting a greater specificity of anthracycline for tumor cells. The use of anthracyclines heralded a new era in the treatment The last event of the ‘roads to ruin’38 is apoptosis by itself. of leukemia.39 The fact that anthracyclines had a pro-apop- We previously described that caspase activation is an early totic effect20 opened up new research fields to investigate the event of apoptosis, occurring before morphological changes.24 molecular mechanisms triggering tumor cell death. Better Consequently, apoptosis following caspase activation in lym- understanding of the anthracycline apoptotic effect may lead phocytes and blast cells was quantified after anthracycline to new therapeutic agents and to better biological support for treatment in lymphocytes and blast cells by the binding of drug evaluation. FITC-annexin V. In this work, we studied the in vitro apoptotic response of lymphocytes and blast cells from medullary samples of 31 patients suffering from de novo AL. MNC were exposed for 1 Anthracyclines trigger caspase 3-mediated apoptosis h to anthracyclines reproducing the in vivo pharmacokinetics in normal and leukemic cells of these drugs.22,40 Because anthracyclines activate the Fas apoptotic pathway on cell lines,12–16 Fas expression was The percentage of apoptotic lymphocytes and blast cells was assessed on lymphocytes and blast cells. We found that Fas quantified by FCM after annexin V labeling and CD45 gating was expressed on both tumor blast cells and lymphocyte cells.

Leukemia Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1271

Figure 3 DEVDases and caspase 3 activation in anthracycline-treated cells. MNC were exposed to anthracyclines as previously described. (a, b) After blast cell sorting based upon their low CD45 expression, DEVDase activity was measured as described in Patients and methods using a ﬂuorogenic substrate and spectroﬂuorimetry. The mean relative DEVDase activation (activity of the DNR or IDA treated sample/activity of the untreated control) was plotted for the AL (a), and for the AML expressing or not the PGP on blast cells (b) and the s.d. are indicated. The cleavage of caspase 3, which led to the activation of this enzyme, was assayed by FCM after MNC labeling with a PE-anti-cleaved-caspase 3 antibody as described in Patients and methods (c). Neg indicates the cells negative for active caspase 3 (uncleaved form) and Pos the cells positive for the active caspase 3 (cleaved form). (d) The mean percentage of blast and lymphocytes cells containing active caspase 3 was plotted for the 10 AL studied and the s.d. were indicated.

Leukemia Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1272

Figure 4 Anthracycline-induced caspase-dependent apoptosis in leukemic blast cells. Bone marrow mononuclear cells were either untreated or treated with DNR or IDA. The cells were then labeled with PC5-CD45, FITC-annexin V, PE-anti-cleaved caspase 3 as described in Patients and methods. Blast cells and lymphocytes were gated as described in Figure 1 on the basis of their CD45 expression. (a) Flow cytometric analysis of untreated and anthracycline-treated blast cells after annexin V and anti-activated caspase 3 double labeling. The corresponding monovariate analysis are shown on the extreme left and extreme right parts of the Figure. The mean percentages of apoptotic (annexin V positive) blast cells were plotted for the total AL (b). The AML were divided in AML with either PGP expressing or PGP-negative blast cells (c). The mean and the s.d. of N patients are shown. (d) For 10 AL, the percentage of apoptotic (annexin V-positive) lymphocytes or blast cells and the percentage of activated-caspase 3-positive lymphocytes or blast cells were measured in anthracycline-treated and control samples as described in (a). The apoptotic values (x axis) and the activated-caspase 3 values (y axis) were plotted and the equations and correlation coefﬁcients obtained from linear regressions were indicated.

This conﬁrms on clinical samples previous work that showed The expression of Fas on all leukemic blast cells theoreti- a CD95 expression on various tumor cell lines.9 The use of cally would enhance the apoptotic response. Indeed, it has FCM allowed the detection of Fas in different cellular popu- been shown that cytotoxic agents such as cisplatin, doxorub- lations based on a differential CD45 expression,34 thus icin, and camptothecin sensitize tumor HT29 cells to Fas- reﬂecting the true expression on each cell type and avoiding mediated apoptosis.17 This sensitization is based upon an discrepancies due to contamination by other cell populations. upregulation of the fas gene, which leads to an overexpression

Leukemia Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1273 expression on non-tumor cells appears unclear. It could rep- resent an activation phenotype, reflecting a cellular pro- gression toward particular functions. For example, some authors have shown that PGP in leukocytes could function to transport cytokines, cytotoxic effector molecules, or inflam- matory mediators.47,48 Our FCM histograms analysis showed that PGP was mostly expressed on a fraction of lymphocytes or blast cells. This suggests that the drug sensitivity of PGP- positive AL could be decreased without abolition of the whole cellular response. Overall, the efficiency of anthracycline treatment is reflected by the ability of tumor cells to undergo the program of apoptosis.19–21,23,49 Caspase 3 is currently described as a common effector of apoptosis required in many models of cell death.7 Caspase 3 is characterized by the DEVDase activity it shared with other members of the caspase family. During the apoptotic process, caspase 3, the main effector of the caspase family, is activated by cleavage.37 We have previously shown that in U937, HL60 and mononuclear cells from leukemic patients, DEVDase activity is evidenced after anthracycline treatment.24 In the present work, we reinforce this observation, showing such an activation in sorted blast cells. DEVDases activation was not significant when PGP was expressed, thus highlighting a potential link with intracellular anthracycline accumulation. Using a specific antibody against activated caspase 3, FCM and multivariate analysis, we dem- onstrate that DNR and IDA induce cleavage and consequently the activation of caspase 3 in both lymphocytes and blast Figure 5 Clinical evolution of patients in relation with the in vitro apoptotic potential of blast cells. Survival in complete remission (SCR) cells. Therefore, the response in normal and tumor cells is and death after therapeutic failure (DTF) were compared to the in vitro similar. Spontaneous apoptosis was lower in lymphocytes blast cell response after anthracycline exposure. MNC were treated than in blast cells and the rate of caspase 3 activation was by either DNR or IDA, and the Fas expression and the percentage of significantly higher in blast cells than in lymphocytes, con- apoptosis were evaluated by FCM on blast cells as described above. firming that anthracycline toxicity mainly affected tumor cells. A control without anthracyclines was processed for each experiment. The rate of increase in Fas expression on blast cells (a) and the per- Furthermore, in both blast cells and lymphocytes, caspase 3 centage of apoptotic blast cells (b) were plotted for SCR and DTF activation was highly correlated with phosphatidyl serine groups and the s.d. are indicated. exposure, indicating that anthracyclines induced apoptosis through the caspase pathway. When PGP was expressed, both IDA and DNR induced apoptosis to a lesser extent, thus pro- of the receptor at the cell surface. Since we found such an viding biological argument for the poor prognosis of PGP- increase on leukemic blast cells after IDA exposure, this positive AL. reinforces the idea that IDA, which exhibit a greater cytotoxic- The clinical evolution of patients was compared to the bio- ity than DNR,41 could be more active on leukemic cells. Fas induction was also found on lymphocytes after IDA and DNR logical data of our study. First, Fas-induction on anthracyc- treatment (data not shown), suggesting the activation of similar line-exposed blast cells was much higher in SCR patients as mechanisms in both normal and tumor cells after chemo- compared with DTF patients. A chi-square analysis showed therapy. Interestingly, the induction of Fas was not significant that a Fas-induction higher than 50% of the basal expression when PGP was expressed on blast cells suggesting that Fas was found significantly more frequently for SCR patients. expression might be related to intracellular anthracycline Moreover, anthracycline-induced apoptosis was also greater accumulation. for SCR patients and an IDA-induced apoptosis higher than PGP, which functions as a pump to expel drugs from the 20% of the blast cells was significantly more frequent for SCR cells, was able to modify the cellular response to drug patients. These observations shed light on the better biological exposure. Moreover, it has been shown that PGP expression response of blast cells in the case of SCR and may constitute on leukemic blast cells at diagnosis or in relapse is strongly original short-term prognosis factors. correlated to therapeutic failure.42 In our study, PGP was Finally, our results show that anthracycline could increase expressed on lymphocytes in 29% of the AML, while it was susceptibility to Fas-induced apoptosis in both leukemic blast found on blast cells only in 13%. A recent study using multi- cells and normal lymphocytes. Anthracycline-induced variate analysis on blast cells from 352 patients found a com- apoptosis was accompanied by a total DEVDase activation parable incidence.43 Conversely, older studies on whole MNC and was characterized by caspase 3 cleavage. Susceptibility found elevated frequencies in the expression of PGP, nearly to chemotherapy was higher in leukemic cells than in lympho- equal to 50%.36,42 Taken together, these data underline the cytes and the deleterious role of PGP expression was biologi- interest of FCM analysis after CD45 gating for the exact deter- cally confirmed in vitro. Interestingly, the biological chemo- mination of the percentage of positive PGP-positive AL. therapy response of blast cells seems to be associated with Evidence for PGP expression on lymphocytes has already patient outcome. Further studies including more patients are been demonstrated44–46 and we found lymphocytes expressing now underway to confirm these observations, in order to the PGP in 29% of the AML. The biological relevance of PGP affirm these new prognostic factors in acute leukemia.

Leukemia Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1274 Acknowledgements programmed cell death (apoptosis) in tumour cells. Biochem Phar- macol 1993; 46: 375–382. 21 Zaleskis G, Berleth E, Verstovsek S, Ehrke MJ, Mihich E. Doxorub- This work was supported in part by grants from the Ligue icin-induced DNA degradation in murine thymocytes. Mol Pharm- Nationale contre le Cancer and from the Association pour la acol 1994; 46: 901–908. Recherche sur le Cancer. 22 Gewirtz DA. A critical evaluation of the mechanisms of action proposed for the antitumor effects of the anthracycline antibiotics adriamycin and daunorubicin. Biochem Pharmacol 1999; 57: 727–741. References 23 Vial JP, Belloc F, Dumain P, Besnard S, Lacombe F, Boisseau MR, Reiffers J, Bernard P. Study of the apoptosis induced in vitro by 1 Nagata S. Apoptosis by death factor. Cell 1997; 88: 355–365. antitumoral drugs on leukemic cells. Leukemia Res 1997; 21: 2 Chinnaiyan AM, O’Rourke K, Tewari M, Dixit VM. FADD, a novel 163–172. death domain-containing protein, interacts with the death domain 24 Durrieu F, Belloc F, Lacoste L, Dumain P, Chabrol J, Dachary- of Fas and initiates apoptosis. Cell 1995; 81: 505–512. Prigent J, Morjani H, Boisseau MR, Reiffers J, Bernard P, Lacombe 3 Boldin MP, Varfolomeev EE, Pancer Z, Mett IL, Camonis JH, Wal- F. Caspase activation is an early event in anthracycline-induced lach D. A novel protein that interacts with the death domain of apoptosis and allows detection of apoptotic cells before they are Fas/APO1 contains a sequence motif related to the death domain. ingested by phagocytes. Exp Cell Res 1998; 240: 165–175. J Biol Chem 1995; 270: 7795–7798. 25 Friesen C, Fulda S, Debatin KM. Deficient activation of the CD95 4 Boldin MP, Goncharov TM, Goltsev YV, Wallach D. Involvement (APO-l/Fas) system in drug-resistant cells. Leukemia 1997; 11: of MACH, a novel MORT1/FADD-interacting protease, in 1833–1841. Fas/APO-1- and TNF receptor-induced cell death. Cell 1996; 85: 26 Friesen C, Herr I, Krammer PH, Debatin KM. Involvement of the 803–815. CD95 (APO-1/Fas) receptor/ligand system in drug-induced 5 Muzzio M, Chinnaiyan AM, Kischkel FC, O’Rourke K, Shevch- apoptosis in leukemia cells. Nature Med 1996; 2: 574–577. enko A, Ni J, Scaffidi C, Bretz JD, Zhang M, Gentz R, Mann M, 27 Mu¨ller M, Strand S, Hug H, Heinemann EM, Walczak H, Hofmann Krammer PH, Peter ME, Dixit VM. FLICE, a novel FADD-homolo- WJ, Stremmel W, Krammer PH, Galle PR. Drug-induced apoptosis gous ICE/CED-3-like protease, is recruited to the CD95 (Fas/APO- in hepatoma cells is mediated by the CD95 (APO-1/Fas) 1) death-inducing signaling complex. Cell 1996; 85: 817–827. receptor/ligand system and involves activation of wild-type p53. 6 Medema JP, Scaffidi C, Kischkel FC, Shevchenko A, Mann M, J Clin Invest 1997; 99: 403–413. Krammer PH, Peter ME. FLICE is activated by association with the 28 Droin N, Dubrez L, Eymin B, Renvoize´ C, Breard J, Dimanche- Cd95 death-inducing signaling complex (DISC). EMBO J 1997; 16: Boitrel MT, Solary E. Upregulation CASP genes in human tumor 2794–2804. cells undergoing etoposide-induced apoptosis. Oncogene 1998; 7 Thornberry NA, Lazebnik Y. Caspases: enemies within. Science 16: 2385–2394. 1998; 281: 1312–1316. 29 Yu DK. The contribution of P-glycoprotein to pharmacokinetic 8 Solary E, Micheau O, Dimanche-Boitrel MT, Martin F. Role of the drug–drug interactions. J Clin Pharmacol 1999; 39: 1203–1211. Fas/Fas-L system in the immune response to tumors and the resist- 30 Praet M, Stryckmans P, Ruysschaert JM. Cellular uptake, and trans- ance to cytotoxic drugs. Bull Cancer 1998; 85: 685–694. 9 Dirks W, Scho¨ne S, Uphoff C, Quentmeier H, Pradella S, Drexler port kinetics of anthracyclines in human sensitive and multidrug- HG. Expression and function of CD95 (Fas/APO-1) in leukemia– resistant K562 cells. Biochem Pharmacol 1996; 51: 1341–1348. lymphoma tumour lines. Br J Haematol 1997; 96: 584–593. 31 Gottesman MM, Pastan I. Biochemistry of multidrug resistance 10 Keane MM, Ettenberg SA, Lowrey GA, Russell EK, Lipkowirz S. mediated by the multidrug transporter. Ann Rev Biochem 1993; Fas expression and function in normal and malignant breast cell 62: 385–425. lines. Cancer Res 1996; 56: 4791–4798. 32 Frankfurt OS, Seckinger D, Sugarbaker EV. Pleiotropic drug resist- 11 Landowski TH, Gleason-Guzman MC, Dalton WS. Selection for ance and survival advantage in leukemia cells with diminished drug resistance to Fas-mediated apoptosis. Blood 1997; 89: apoptotic response. Int J Cancer 1994; 59: 217–224. 1854–1861. 33 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DAG, 12 Eischen CM, Kottke TJ, Martins LM, Basi GS, Tung JS, Earnshaw Gralnick HR, Sultan C. Proposal for the classification of the acute WC, Leibson PJ, Kaufmann SH. Comparison of apoptosis in wild- leukaemias. Br J Haematol 1976; 33: 451–458. type and Fas-resistant cells: chemotherapy-induced apoptosis is 34 Lacombe F, Durrieu F, Briais A, Dumain P, Belloc F, Bascans E, not dependent on Fas/Fas-ligand interactions. Blood 1997; 90: Reiffers J, Boisseau MR, Bernard P. Flow cytometry CD45 gating 935–943. for immunophenotyping of acute myeloid leukemia. Leukemia 13 Mizutani Y, Okada Y, Yoshida O, Fukumoto M, Bonavida B. 1997; 11: 1878–1886. Doxorubicin sensitizes human bladder carcinoma cells to Fas- 35 Belaud-Rotureau MA, Lacombe F, Durrieu F, Vial JP, Lacoste C, mediated cytotoxicity. Cancer 1997; 79: 1180–1189. Bernard P, Belloc F. Ceramide-induced apoptosis occurs indepen- 14 Morimoto H, Yonehara S, Bonavida B. Overcoming tumor dently of caspases and is decreased by leupeptin. Cell Death Differ necrosis factor and drug resistance of human cell lines by combi- 1999; 6: 788–795. nation treatment with anti-Fas antibody and drugs or toxins. Can- 36 Kuwazuru Y, Yoshimura A, Hanada S, Utsunomiya A, Makino T, cer Res 1993; 53: 2591–2596. Ishibashi K, Kodama M, Iwahashi M, Arima T, Akiyama S. 15 Nakamura S, Takeshima M, Nakamura Y, Ohtake S, Matsuda T. Expression of the multidrug transporter, P-glycoprotein, in acute Induction of apoptosis in HL60 leukemic cells by anticancer drugs leukemia cells and correlation to clinical drug resistance. Cancer in combination with anti-Fas monoclonal antibody. Anticancer 1990; 66: 868–873. Res 1997; 17: 173–179. 37 Martin SJ, Green DR. Protease activation during apoptosis: death 16 Uslu R, Jewett A, Bonavida B. Sensitization of human ovarian by a thousand cuts? Cell 1995; 82: 349–352. tumor cells by subcytotoxic CDDP to anti-Fas antibody-mediated 38 Green DR. Apoptotic pathways: the roads to ruin. Cell 1998; 94: cytotoxicity and apoptosis. Gynecol Oncol 1996; 62: 232–291. 695–698. 17 Micheau O, Solary E, Hammann A, Martin F, Dimanche-Boitrel 39 Zubrod CG. Treatment of the acute leukemias. Cancer Res 1967; MT. Sensitization of cancer cells treated with cytotoxic drugs to 27: 2557–2560. Fas-mediated cytotoxicity. J Natl Cancer Inst 1997; 89: 783–789. 40 Robert J, Gianni L. Pharmacokinetics and metabolism of anthracy- 18 Kimura H, Yamaguchi Y. A phase III randomized study of interleu- clines. Cancer Surv 1993; 17: 219–251. kin-2 lymphokine-activated killer cell immunotherapy combined 41 Borchmann P, Hubel K, Schnell R, Engert A. Idarubicin: a brief with chemotherapy or radiotherapy after curative or noncurative overview on pharmacology and clinical use. Int J Clin Pharmacol resection of primary lung carcinoma. Cancer 1997; 80: 42–49. Ther 1997; 35: 80–83. 19 Ling YH, Priebe W, Perez-Soler R. Apoptosis induced by anthra- 42 Campos L, Guyotat D, Archimbaud E, Calmard-Oriol P, Tsuruo cycline antibiotics in P388 parent and multidrug-resistant cells. T, Troncy J, Treille D, Fiere D. Clinical significance of multidrug Cancer Res 1993; 53: 1845–1852. resistance P-glycoprotein expression on acute nonlymphoblastic 20 Skladanowski A, Konopa J. Adriamycin and daunomycin induce leukemia cells at diagnosis. Blood 1992; 79: 473–476.

Leukemia Anthracycline-induced apoptosis in leukemia M-A Belaud-Rotureau et al 1275 43 Leith CP, Kopecky KJ, Chen IM, Eijdems L, Slovak ML, McConnell protein expression and function in circulating blood cells from TS, Head DR, Weick J, Grever MR, Appelbaum FR, Willman CL. normal volunteers. Blood 1994; 83: 2451–2458. Frequency and clinical signiﬁcance of the expression of the multi- 47 Drach J, Gsur A, Hamilton G, Zhao S, Angerler J, Fiegl M, Zojer drug resistance proteins MDR1/P-glycoprotein, MRP1, and LRP in N, Raderer M, Haberl I, Andreeff M, Huber H. Involvement of P- acute myeloid leukemia. A Southwest Oncology Group study. glycoprotein in the transmembrane transport of interleukin-2 (IL- Blood 1999; 94: 1086–1099. 2), IL-4, and interferon-gamma in normal human T lymphocytes. 44 Coon JS, Wang Y, Bines SD, Markham PN, Chong ASF, Gebel Blood 1996; 88: 1747–1754. HM. Multidrug resistance activity in human lymphocytes. Hum 48 Raymond M, Gros P, Whiteway M, Thomas DY. Functional comp- Immunol 1991; 32: 134–140. lementation of yeast ste6 by mammalian multidrug resistance mdr 45 Drach D, Zhao S, Drach J, Mahadevia R, Gattringer C, Huber H, gene. Science 1992; 256: 232–234. Andreeff M. Subpopulations of normal peripheral blood and bone 49 Durrieu F, Belaud-Rotureau MA, Lacombe F, Dumain P, Reiffers marrow cells express a functional multidrug resistance phenotype. J, Boisseau MR, Bernard P, Belloc F. Synthesis of Bcl-2 in response Blood 1992; 80: 2729–2734. to anthracycline treatment may contribute to an apoptosis-resistant 46 Klimeckj WT, Furscher BW, Grogan TM, Dalton WS. P-glyco- phenotype in leukemic cell lines. Cytometry 1999; 36: 140–149.

Leukemia