Induction of a TRAIL Mediated Suicide Program by Interferon Alpha in Primary E€Usion Lymphoma

Induction of a TRAIL mediated suicide program by interferon alpha in primary eusion lymphoma

Ngoc L Toomey1,4, Vadim V Deyev2,4, Charles Wood3, Lawrence H Boise2, Duncan Scott1, Lei Hua Liu1, Lisa Cabral1, Eckhard R Podack2, Glen N Barber2 and William J Harrington Jr*,1

1Department of Medicine University of Miami School of Medicine, Miami, Florida, FL 33136, USA; 2Department of Microbiology and Immunology, University of Miami School of Medicine, Miami, Florida, FL 33136, USA; 3School of Biological Sciences, University of Nebraska, Lincoln, Nebraska, NE 68588, USA

Gammaherpes viruses are often detected in lymphomas Virus (EBV) or Human Herpes Virus Type 8 (HHV-8) arising in immunocompromised patients. We have found have been isolated from lymphomas found in im- that Azidothymidine (AZT) alone induces apoptosis in munosuppressed organ transplant recipients, children Epstein Barr Virus (EBV) positive Burkitt's lymphoma with hereditary immunode®ciencies and patients with (BL) cells but requires interferon alpha (IFN-a) to induce acquired immunode®ciency (AIDS) (Swinnen, 1999; apoptosis in Human Herpes Virus Type 8 (HHV-8) Goldsby and Carroll, 1998; Knowles, 1999). Many of positive Primary Eusion Lymphomas (PEL). Our these tumors can be categorized into distinct subtypes analysis of a series of AIDS lymphomas revealed that based on a variety of morphologic and molecular IFN-a selectively induced very high levels of the Death criteria. For example, AIDS associated large cell Receptor (DR) tumor necrosis factor-related apoptosis- diuse or immunoblastic lymphomas (DLCL, IBL) inducing ligand (TRAIL) in HHV-8 positive PEL lines are often EBV positive while AIDS associated Burkitt's and primary tumor cells whereas little or no induction lymphomas (BL) less frequently contain EBV (Gaidano was observed in primary EBV+ AIDS lymphomas and et al., 1994). A recently de®ned subtype, AIDS related EBV7Burkitt's lines. AZT and IFN-a mediated HHV-8 associated Primary Eusion Lymphoma (PEL) apoptosis in PEL was blocked by stable overexpression (Nador et al., 1996), usually occurs in the setting of of dominant negative Fas Associated Death Domain severe immunode®ciency of advanced HIV infection. (FADD), decoy receptor 2 (DcR2), soluble TRAIL PELs dier from most AIDS lymphomas by their receptor fusion proteins (DR-4 and DR-5) and thymi- absence of a discernable primary tumor mass, infre- dine. Trimeric TRAIL (in place of IFN-a) similarly quent expression of B lymphocyte dierentiation synergized with AZT to induce apoptosis in HHV-8 antigens, and lack of c-myc gene rearrangement positive PEL cells. This is the ®rst demonstration that (Mullaney et al., 2000; Gaidano et al., 2000; Demario IFN-a induces functional TRAIL in a malignancy that and Liebowitz, 1998). In general, immunode®ciency can be exploited to eect a suicide program. This novel related herpesvirus associated lymphomas are aggres- antiviral approach to Primary Eusion lymphomas is sive and poorly responsive to conventional chemother- targeted and may represent a highly eective and apy (Swinnen, 2000; Levine, 2000). relatively non-toxic therapy. Oncogene (2001) 20, Although AZT was originally developed as an anti- 7029 ± 7040. cancer agent, this thymidine nucleoside analog has demonstrated relatively little activity in solid tumors Keywords: human Herpes Virus Type 8; Epstein Barr (Findenig et al., 1996). Interest in AZT was revived virus; TRAIL; apoptosis; lymphoma; FADD only when it was found to inhibit HIV reverse transcriptase (De Clercq, 1992). To exert this antiviral activity, AZT must be phosphorylated by cellular Introduction thymidine kinase (TK) (Arner et al., 1992). Our initial studies indicated that there were two distinctly dierent Gammaherpes viruses are frequently associated with pro-apoptotic eects of AZT and Interferon alpha lymphoproliferative disease in immunocompromised (IFN-a) in primary lymphoma cell lines derived from individuals (Okano and Gross, 2000). Epstein-Barr AIDS patients. EBV+ BL cells underwent apoptosis in the presence of AZT alone while HHV-8+ PELs required the addition of IFN-a to undergo signi®cant *Correspondence: WJ Harrington Jr, University of Miami School of programmed cell death (Lee et al., 1999). Medicine/Sylvester Comprehensive Cancer Center, Room 3400 (D8- Interferons have multiple activities involved in host 4), 1475 NW 12th Avenue, Miami, Florida, FL 33136, USA; defense including anti-proliferative and antiviral eects. E-mail: [email protected] It is known that the interferons, which are potently 4These authors contributed equally to this work Received 14 May 2001; revised 17 July 2001; accepted 2 August upregulated by viruses and double stranded RNA 2001 (dsRNA), can synthesize eectors of apoptosis such as Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7030 2'5' oligoadenylate synthetase (2' ±5'A) which activates These data demonstrate that IFN-a upregulates RNAseL and degrades viral mRNA (Player and TRAIL in Primary Eusion Lymphomas while AZT Torrence, 1998). Interferon also induces synthesis of sensitizes these cells to TRAIL mediated apoptosis dsRNA activated protein kinase (PKR), which phos- resulting in the activation of a suicide program in phorylates the translation initiator eIF-2-a and inhibits this cancer. The unique tumorcidal eect of AZT protein synthesis in the cell (Zamanian-Daryoush et al., and IFN-a may have important applications for 2000). Recently, interferons were also found to induce therapy of herpesvirus associated lymphoproliferative expression of the pro-apoptotic protein tumor necrosis disease. factor-related apoptosis-inducing ligand (TRAIL) in human dendritic cells and T lymphocytes which is capable of inducing apoptosis in TRAIL receptor Results expressing targets (Fanger et al., 1999; Grith et al., 1999; Balachandran et al., 2000). AZT and IFN-a act synergistically to induce apoptosis in IFN also potentiates the adaptive immune response HHV-8+ PEL cells through upregulation of major histocompatibility complexes (MHC) and activation of cytotoxic T cells We have demonstrated that AZT and IFN-a have (CTLs) and natural killer cells (NKs) (Mingari et al., clinical activity against malignant herpesvirus asso- 2000). CTLs can recognize and kill cells that present ciated lymphomas (Lee et al., 1999). To further foreign peptides, in association with MHC, through the investigate the apoptotic eects, we studied the perforin/granzyme pathway (Barry et al., 2000). HHV-8+ PEL lines, BC-3 and BCBL-1, as well as Another important mechanism of immune eector the EBV+ AIDS BL primary lines, BL-7 and BL-5. clearance of virally infected or cancerous cells is by All lines were cultured for 48 h in the presence of transmission of an apoptotic signal through ligand medium, AZT (10 mg/ml) or IFN-a (1000 u/ml) binding to members of the tumor necrosis (TNF) alone, or AZT and IFN-a together. As shown in family of receptors (Peter et al., 1997). The binding of Figure 1a, AZT alone induced marked apoptosis in ligands such as Fas-L or TRAIL to extracellular the EBV+Burkitt's lymphoma lines (BL-7, BL-5) receptor domains (APO-1 (Fas/CD95), DR-5) results while IFN-a did not induce apoptosis nor add to in recruitment of components of the death domain the cytopathic eect of AZT. In contrast, HHV-8+ containing the adaptor molecule Fas associated death PEL cells (BC-3, BCBL-1) only underwent marked domain (FADD) and subsequent activation of FLICE apoptosis in the presence of both AZT and IFN-a (caspase 8) by autoproteolysis. Following recruitment (Figure 1a). Similar experiments revealed that other and cleavage of procaspase 8, downstream caspases are HHV-8+ PEL lines, BC-1 and BC-5, also required activated, including caspase 3, resulting in apoptosis. both AZT and IFN-a to undergo signi®cant TRAIL mediated signaling and apoptosis is blocked apoptosis (data not shown). Therefore, a general upon its binding to decoy receptors (DcR1, DcR2) property of HHV-8+ PEL cells was that AZT and which do not recruit FADD (Walczak and Krammer, IFN-a induced substantial apoptosis together but not 2000; Ashkenazi and Dixit, 1999; Bodmer et al., 2000). separately. This cytotoxic eect was seen only in The role of DR-4 in apoptosis is unclear since its virus infected lymphomas as both agents had no expression in FADD7/7 ®broblasts still results in cell eect on herpes virus negative BL cells (Ramos) death (Yeh et al., 1998). Signaling through other (Figure 1b). Further experiments on another herpes- members of the TNF receptor family also may activate virus negative lymphoma cell line (BJAB) demon- non-apoptotic processes such as in¯ammatory re- strated that AZT, with or without IFN-a, induced sponses and lymphoid organogenesis (Magnusson and little or no cytotoxic activity (data not shown). Vaux, 1999). To further investigate the mechanism of AZT and IFN-a induces TRAIL in PEL cells IFN-a mediated apoptosis, we studied a series of lymphoma cell lines and primary lymphoma cells It has recently been shown that IFN-a mediates the derived from AIDS patients. We report here that expression of several pro-apoptotic genes and can HHV-8+ PEL cell lines and primary tumor cells from induce cell death through a FADD dependent a PEL patient express high levels of TRAIL when mechanism (Balachandran et al., 2000). Since IFN-a cultured with IFN-a. Despite the induction of TRAIL potentiated apoptosis in PEL, we investigated its in PEL cells by IFN-a, apoptosis was only potentiated eect on the regulation of pro-apoptotic factors in upon the addition of AZT. In PEL cells, AZT and IFN-a sensitive and resistant B cell lymphomas. IFN-a mediated apoptosis was blocked by expression Accordingly, BC-3 and BL-7 cells were treated for of dominant negative FADD (FADD DN), decoy 8 h with AZT (10 mg/ml), IFN-a (1000 m/ml) or the receptor 2 (DcR2), soluble TRAIL receptor fusion chemotherapeutic agent etoposide (10 mg/ml). Total proteins and by the addition of thymidine. In contrast RNA was extracted and analysed for pro-apoptotic to HHV-8+ PEL, EBV+ AIDS BL lines and EBV7 gene expression by ribonuclease protection assay BL lines did not express signi®cant amounts of (RPA). RPA analysis revealed that IFN-a induced TRAIL, nor undergo increased apoptosis in response markedly higher levels of TRAIL mRNA in the IFN- to IFN-a. a sensitive PEL cells (BC-3) compared to the resistant

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7031 EBV+ BL cells (BL-5) (Figure 2a, lanes 2 and 4 etoposide, at doses sucient to kill these cells, had compared to lanes 7 and 9). In contrast, AZT and no eect on TRAIL expression in either type of

Figure 1 (a) AZT and IFN-a synergize to induce apoptosis in HHV-8+ PEL (BC-3, BCBL-1) while EBV+ BL cells (BL-5, BL-7) undergo apoptosis with AZT alone. BC-3 and BCBL-1 (HHV-8+ PEL's) and BL-5 and BL-7 (EBV+ BL's) were treated in medium (Panels A), AZT 10 mg/ml (Panels B), IFN-a 1000 u/ml (Panels C), or AZT 10 mg/ml plus IFN-a 1000 u/ml (Panels D) for 48 h. Cells were then analysed for apoptosis by PI annexin staining and FACs analysis. Upper and lower right quadrants are the apoptotic populations. (b) AZT and IFN-a are synergistically cytotoxic in HHV-8+ PEL but not EBV7 BL. PEL cells BCBL-1 and BC-3, EBV+ BL cells BL-5 and BL-7, and EBV7 Ramos cells were cultured for 48 h in medium (control), IFN-a 1000 u/ml, AZT 10 mg/ml, and IFN-a 1000 u/ml plus AZT 10 mg/ml. Cell viability was then determined by Trypan blue exclusion. Bars represent per cent viable cells

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7032 a lymphoma (Figure 2a, lanes 3 and 5 and lanes 8 and 10). We then investigated the eect of IFN-a on a larger panel of lymphoma lines including HHV-8+ PELs, EBV+ BLs (derived from HIV positive patients) and EBV7 BL lines. We found that IFN- a signi®cantly induced TRAIL mRNA only in the HHV-8+ PEL lines and primary tumor cells (BCBL- LN) derived from a patient with PEL, while little or no induction was noted either in EBV positive or negative BL lines. IFN-a also had no eect on the expression of two death receptors, DR-4 and DR-5, b which are known to bind to TRAIL (Figure 2b). Further RPA experiments demonstrated that neither AZT nor IFN-a aected mRNA or protein surface expression levels of DR-4 or DR-5 in PEL cells (data not shown). FACS analysis demonstrated that IFN-a induced TRAIL protein surface expression in PEL cells but not in EBV+ BL cells (Figure 2c, panels 3 and 4 versus panels 1 and 2). Therefore, the ability of IFN-a to potentiate AZT mediated apoptosis in PEL correlated with its ability to induce the pro- apoptotic ligand TRAIL in these tumors.

AZT and IFN-a mediated apoptosis in PEL is FADD dependent TRAIL has been reported to induce apoptosis by binding its cognate receptors. Furthermore, TRAIL mediated apoptosis has recently been shown to be dependent upon signaling through the death adaptor protein FADD (Bodmer et al., 2000). To investigate the mechanism of IFN-a induced apoptosis we ®rst con®rmed that HHV-8+ PEL cells (BCBL-1) did express TRAIL receptors on their surface (DR-4 was more highly expressed than DR-5) (Figure 3a). BC-3 cells also expressed similar levels of DR-4 and DR-5 c (data not shown). To study whether apoptosis induced by AZT and IFN-a in PEL involved a FADD dependent mechanism, we transfected the BCBL-1 line with a dominant negative construct of FADD that lacks a death eector domain (Bala-

AZT 10 mg/ml. Lane 4=BC-3 in IFN-a 1000 u/ml and AZT 10 mg/ml. Lane 5=BC-3 in etoposide 10 mg/ml. Lane 6=BL-5 in medium. Lane 7=BL-5 in IFN-a 1000 u/ml. Lane 8=BL-5 in AZT 10 mg/ml. Lane 9=BL-5 in IFN-a 1000 u/ml and AZT 10 mg/ml. Lane 10=BL-5 in etoposide 10 mg/ml. (b) IFN-a induces a high level of expression of TRAIL mRNA in HHV- 8+ PEL but not EBV+ BL and EBV7 BL. PEL lines and primary tumor isolates (BCBL-1, BC-3, BC-1, BCBL-LN, BC-2, BC-5), EBV+ BL (P3HR-1), EBV+AIDS BL lines (BL-8, SM-1, BL-5, BL-7), and EBV7 BL lines (Ramos, BJAB) were treated with media (C) or IFN-a 1000 u/ml (I) for 8 h. mRNA was extracted and assayed by ribonuclease protection assay for expression of DR-4, DR-5 and TRAIL. GAPDH expression was used as an internal control. (c) IFN-a induces TRAIL surface expression in HHV-8+ PEL but not EBV+ BL. EBV+ BL: BL- Figure 2 (a) TRAIL mRNA is induced by IFN-a but not by 5 and BL-7 cells (Panels 1 and 2) and HHV-8+ PEL cells: BCBL- AZT or Etoposide in HHV-8+ PEL. PEL cells (BC-3, Lanes 1 ± 1 and BC-3 (Panels 3 and 4) were treated for 24 h with IFN-a 5) and EBV+cells (BL-5, Lanes 6 ± 10) were treated in the 1000 u/ml or media and assayed for TRAIL expression by FACs following manner for 8 h and examined for TRAIL mRNA analysis. Shaded area equals isotype control (mouse IgG1). expression by ribonuclease protection assay. Lane 1=BC-3 in Dotted line are cells treated with media and solid line are cells medium. Lane 2=BC-3 in IFN-a 1000 u/ml. Lane 3=BC-3 in treated with IFN-a

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7033 chandran et al., 2000). After G418 selection, we FADD protein (BCBL FADD DN) (Figure 3b). derived several stably transfected subclones which BCBL FADD DN cells were completely resistant to expressed a dominant negative mutation of the AZT and IFN-a mediated apoptosis but retained

Figure 3 (a) HHV-8+ PEL cells express TRAIL receptors. PEL cells (BCBL-1) were grown in IMDM media enriched with 10% FCS and surface expression of DR-5 (dotted line) and DR-4 (solid line) was determined by FACs analysis. Shaded area is mouse isotype control (IgG1). (b) Transfection of BCBL-1 cells with FADD-DN. BCBL-1 cells were transfected with FADD-DN by electroporation. Subclones were derived after G418 selection. Subclones (BCBL-2A4, BCBL-3A3, BCBL-3B2, BCBL-3C2, and BCBL-1A2), BCBL-Neo (transfected with neo-resistance gene only), and BCBL-1 parental cells were assayed for expression of FADD and FADD-DN by Western blot. (c) BCBL-1 FADD-DN cells are resistant to AZT and IFN-a but sensitive to VP-16. BCBL-1, BCBL-1 Neo transfectants and BCBL-1 FADD-DN clones (BCBL-2A4, BCBL-3C2, BCBL-1A2) were treated for 48 h with medium, AZT 10 mg/ml, IFN-a 1000 u/ml, AZT 10 mg/ml plus IFN-a 1000 u/ml, or VP-16 10 mg/ml (for 24 h) and cell death was measured by Trypan blue exclusion. The ®gure is representative of 52 experiments. (d) BCBL-1 FADD-DN cells express TRAIL in response to IFN-a. BCBL-1A2 cells were treated with medium (C) or IFN-a (I) for 8 h and caspase 8, DR-4 and DR-5 expression levels were detected by RPA

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7034 their sensitivity to etoposide (Figure 3c). To protein (data not shown). BCBL-1 FADD DN cells determine that resistant cells were not generated by also expressed similar levels of DR-4 and DR-5 and the selection process, clones were also characterized TRAIL mRNA when treated with IFN-a (Figure for expression of anti-apoptotic proteins, Bcl-2 and 3d). Therefore, our data indicates that apoptosis Bcl-x and found to express similar low levels induced by AZT and IFN-a in HHV-8+ PEL is a (compared to the parental BCBL-1 cells) of each FADD dependent process.

a c

Figure 4 (a) Activation of caspases 3 and 8 by AZT and IFN-a in PEL cells is blocked by the caspase inhibitor DEVD and expression of FADD DN. PEL cells; BC-3 (top panel), BCBL-1 (middle panel) and BCBL-1A2 (FADD-DN cells) (lower panel) were cultured for 48 h in media (C), IFN-a 1000 u/ml (I), AZT 10 mg/ml (A), AZT 10 mg/ml plus IFN-a 1000 u/ml (A+I), AZT 10 mg/ml and IFN-a 1000 u/ml plus the caspase inhibitor DEVD 10 mM (A+I+D) and assayed by Western blot for caspase 3 and 8. The caspase 8 antibody used only detects the uncleaved pro-caspase form. (b) Soluble TRAIL receptors inhibit AZT and IFN-a mediated cytotoxicity in PEL cells. BCBL-1 (HHV-8+ PEL cells) were treated in triplicate with the following conditions: medium (Control); AZT 10 mg/ml+IFN-a 1000 u/ml; soluble DR-4 2 mg/ml+soluble DR-5 5 ng/ml; pre-treatment for 8 h with soluble DR- 42mg/ml and soluble DR-5 5 ng/ml followed by treatment for 48 h with AZT 10 mg/ml+IFN-a 1000 u/ml; pre-treatment for 8 h with soluble DR-4 2 mg/ml+soluble DR-5 5 ng/ml+ZB4 1 mg/ml followed by treatment for 48 h with AZT 10 mg/ml and IFN-a 1000 u/ml; TRAIL 100 ng/ml; pre-treatment for 8 h with soluble DR-4 2 mg/ml+soluble DR-5 5 ng/ml followed by treatment for 48 h with TRAIL 100 ng/ml. Cytotoxicity was measured after 48 h by Trypan blue exclusion. (c) DcR2 expression in BCBL-1 blocks apoptosis induced by AZT and IFN-a. Top panel: Flow cytometry pro®les of DcR2 expression in transfected BCBL-1 cells (bold line) versus wild type (thin line). Dotted line is isotype control. Bottom panel: Wild type and BCBL-1 cells were treated with IFN-a 1000 u/ml, AZT 10 mg/ml or IFN-a 1000 u/ml plus AZT 10 mg/ml and viability was measured by Trypan blue exclusion

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7035 combined with AZT in PEL cells. We found that in AZT and IFN-a induces cleavage of caspases 3 and 8 in PEL cells, TRAIL and AZT caused a synergistic pro- PEL which is inhibited by soluble DR-4, DR-5 or DcR2 apoptotic eect similar to that caused by IFN-a and Apoptosis mediated through death receptor ligand AZT (Figure 5a, panel E compared to panel C). pathways involves the recruitment of proteins forming Similar to AZT and IFN-a, AZT and TRAIL were not the death inducing signaling complex (DISC) resulting cytotoxic to FADD DN BCBL-1 cells (Figure 5b). in cleavage (activation) of procaspase 8 (FLICE) to its These data indicate that AZT enhances TRAIL active form and subsequent activation of downstream mediated apoptosis in HHV-8+ PEL cells. pro-caspases (Kischkel et al., 1995; Medema et al., 1997). We therefore treated HHV-8+ PEL cells (BCBL-1 and BC-3) and BCBL-DN FADD transfectants (BCBL-1A2) with AZT, IFN-a, or AZT plus IFN-a and assayed for procaspase 8 cleavage. Treat- a ment with either agent alone resulted in no cleavage of procaspase 8. However in BC-3 and BCBL-1 cells, AZT and IFN-a together activated procaspase 8 and procaspase 3 but did not in BCBL-1A2 (FADD DN) cells (Figure 4a, lane 4). We did note some cleavage of procaspase 3 in AZT treated BCBL-1 and BC-3 PEL cells which indicates that AZT alone does cause some degree of apoptosis in PEL cells (lane 3). Treatment with the caspase inhibitor DEVD abrogated AZT and IFN-a mediated cleavage of both procaspases in PEL cells (lane 5). Caspase inhibition of AZT and IFN-a mediated apoptosis with IETD, a speci®c caspase 8 inhibitor, was not possible since the inhibitor itself was toxic to PEL cells (data not shown). We then b investigated whether inhibition of TRAIL with soluble TRAIL receptors DR-4 and DR-5 could block AZT and IFN-a mediated apoptosis. These experiments were also performed with ZB4, a blocking anti-Fas antibody. As demonstrated in Figure 4b, soluble DR-4 and DR-5 together inhibited AZT and IFN-a mediated apoptosis, albeit incompletely, while the addition of ZB4 had little eect on AZT and IFN-a apoptosis. In contrast, soluble receptors DR-4, DR-5 and ZB4 had no inhibitory eect on AZT mediated apoptosis in EBV+ BL cells (data not shown). Control experiments demonstrated that soluble DR-4 and DR-5 almost completely inhibited TRAIL mediated cell death (at a highly cytotoxic concentration of TRAIL [100 ng/ml]). To further con®rm that AZT and IFN-a mediated apoptosis occurs principally through TRAIL signaling, we transfected BCBL-1 cells with the decoy receptor for TRAIL (DcR2). Overexpression of DcR2 markedly abrogated the IFN-a component of AZT and IFN-a mediated apoptosis (Figure 4c).

Figure 5 (a) TRAIL substitutes for IFN-a to potentiate AZT enhances TRAIL mediated apoptosis in PEL apoptosis in AZT treated PEL cells. BCBL-1 cells were treated under the following conditions for 48 h and apoptosis was Although IFN-a induced TRAIL expression in PEL, measured by Annexin/PI staining. Panel A=BCBL-1 cells in apoptosis was potentiated by the addition of AZT. To medium. Panel B=BCBL-1 cells in AZT 10 mg/ml. Panel determine whether TRAIL could substitute for IFN-a C=BCBL-1 cells in AZT 10 mg/ml and IFN-a 1000 u/ml. Panel and similarly synergize with AZT to induce apoptosis D=BCBL-1 cells treated with soluble TRAIL 10 ng/ml. Panel E=BCBL-1 cells treated with AZT 10 mg/ml plus soluble TRAIL in PEL cells, we ®rst performed dose response 10 ng/ml. The per cent of apoptotic cells in each panel is shown in experiments and found that titrating soluble trimeric Panel F, (b) AZT potentiates TRAIL mediated cytotoxicity in TRAIL, to concentrations of 10 ± 30 ng/ml for 48 h, BCBL-1. BCBL-1 and BCBL-1 FADD-DN (BCBL-1A2) were induced little apoptosis above the baseline seen in treated in media, AZT 10 mg/ml, AZT 10 mg/ml plus IFN-a untreated PEL cells (BC-3, BCBL-1) (data not shown). 1000 u/ml, TRAIL 10 ng/ml, or AZT 10 mg/ml plus TRAIL 10 ng/ml for 48 h and per cent viability was determined by We then substituted soluble trimeric TRAIL for IFN-a Trypan blue exclusion. Bars represent per cent viable cells. The to determine whether it also induced apoptosis when data is representative of at least two experiments

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7036 a AZT and IFN-a mediated apoptosis does not occur with other antiviral nucleosides and is inhibited by thymidine HHV-8 and EBV encode viral thymidine kinases (TKS) which have been shown to be capable of phosphorylating AZT (a thymidine nucleoside analog) (Gustafson et al., 2000). Since only herpesvirus associated lymphomas were sensitive to AZT, we investigated whether this cytotoxic eect occurred with other antiviral nucleoside analogues. When combined with IFN-a,a cytidine analogue, ddC, and a thymidine analog, d4T, were completely inactive in AZT and IFN-a sensitive PEL cells; BCBL-1 and BC-3 (Figure 6a). We therefore reasoned that in herpesvirus lymphomas, if phosphorylation of AZT was necessary for apoptosis then thymidine, which competes with AZT as a substrate for thymidine kinase, should abrogate the apoptotic eect. As expected, thymidine in a dose dependent fashion, blocked apoptosis in both EBV+ BL cells b treated with AZT and in HHV-8+ PEL cells treated with AZT and IFN-a (Figure 6b). These data suggest that the anti-tumor eect of AZT in both EBV+ BL and PEL may require phosphorylation by a herpesvirus encoded thymidine kinase with substrate speci®city for AZT. These results were further supported by the fact that herpesvirus negative cells BJAB and Ramos were unaected by AZT (data not shown). These data may also explain why AZT and IFN-a are inactive in herpesvirus negative lymphomas.

Discussion

Herpesvirus associated lymphomas in immunocompro- Figure 6 (a) Antivirals nucleosides ddC and d4T do not mised patients are usually dicult to treat, as synergize with IFN-a to induce apoptosis in PEL. BCBL-1 and BC-3 cells were cultured for 48 h in media (control), IFN-a conventional chemotherapy is poorly tolerated and 1000 u/ml, AZT 10 mg/ml, AZT 10 mg/ml and IFN-a 1000 u/ml, causes further immunosuppression. Biologic or immu- ddC 10 mg/ml, ddC 10 mg/ml and IFN-a 1000 u/ml, d4T 10 mg/ nomodulatory cancer therapies such as interferons ml, or d4T 10 mg/ml and IFN-a 1000 u/ml. Cell viability was initially held great promise. However, clear-cut clinical determined by Trypan blue exclusion. Bars represent per cent viable cells. (b) Inhibition of AZT or AZT+IFN-a induced ecacy has been restricted to a limited number of apoptosis by Thymidine in BL-7 (EBV+ BL) and BCBL-1 diseases, most notably viral associated chronic active (HHV-8+ PEL). BL-7 cells (Top panel) were cultured for 48 h in hepatitis (CAH) (Zavaglia et al., 2000). Recent data medium (control), AZT 10 mg/ml, Thymidine 10 mg/ml, AZT indicate that interferons appear to sensitize cells to 10 mg/ml plus Thymidine 10 mg/ml, Thymidine 100 mg/ml or AZT FADD dependent apoptosis. Evidence for this comes 10 mg/ml plus Thymidine 100 mg/ml. BCBL-1 cells (lower panel) were cultured for 48 h in medium (control), AZT 10 mg/ml plus from work which demonstrates that inhibitors of IFN-a 1000 u/ml, Thymidine 10 mg/ml, AZT 10 mg/ml plus IFN-a caspase 8, dominant negative mutation of FADD, or 1000 u/ml plus Thymidine 10 mg/ml, Thymidine 100 mg/ml, AZT the absence of FADD, block IFN mediated apoptosis 10 mg/ml plus IFN-a 1000 u/ml plus Thymidine 100 mg/ml (Balachandran, et al., 1998). Recently, understanding of these pro-apoptotic mechanisms has been furthered by the ®nding that IFN-a induces TRAIL expression in hepatocytes (Jo et al., 2000). However, the therapeutic monocytes and dendritic cells (Fanger et al., 1999; index of TRAIL might be improved if speci®c killing of Grith et al., 1999). Therefore, an important biologic virally infected tumor cells were enhanced by the property of interferons may be to directly or indirectly addition of antiviral nucleosides. Alternatively, IFN-a activate DR ligands such as TRAIL. Our study may induce tolerable, clinically active levels of TRAIL demonstrates that IFN-a mediated induction of when administered systemically. This eect might be TRAIL in malignancies (HHV-8+ PEL) can be restricted to certain virus associated tumors or disease exploited to activate a suicidal or fratricidal tumor processes since our data demonstrates that TRAIL is cell death. markedly induced by IFN-a speci®cally in HHV-8 Recent enthusiasm for the use of death receptor infected lymphomas. Whether this eect is unique to ligands such as TRAIL as a cancer therapy has been HHV-8+lymphomas will require analysis of a greater tempered by evidence of their toxicity in human number of tumors. This marked pro-apoptotic eect

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7037 between the thymidine analog AZT and IFN-a was EBV+ BL, although quite sensitive to AZT alone, not reproducible by substituting a cytidine analog, did not express signi®cant amounts of TRAIL in 3TC, or another thymidine analog, d4T. A likely response to IFN-a. IFN-a also did not increase the reason for this is that the HHV-8 encoded thymidine apoptotic eect of AZT in EBV+ BL. There are kinase speci®cally phosphorylates AZT. HHV-8 and several potential reasons for the disparate eects of EBV encoded TKs have been shown to be capable of IFN-a in PEL and EBV+ BL. IFN-a signaling that phosphorylating AZT (Gustafson et al., 2000), how- activates DR ligands or other components of IFN ever, these (TKs) are expressed during the viral lytic signaling may be defective in these EBV+ BL lines. cycle, although low level expression has been reported Alternatively, other pro-apoptotic eects of AZT such in HHV-8+ PEL lines (Cannon et al., 1999). Whether as mitochondrial damage may be much more pro- AZT itself may induce lytic herpesvirus genes or is nounced and predominant in EBV+ BL cells. It is phosphorylated by low level expression of viral TK is possible that AZT alone might activate other DR presently unknown. ligand pathways in EBV+ BL. It is interesting to note AZT and IFN-a mediated apoptosis in PEL was that like PEL cells, EBV+ cells were also killed by a blocked in cells expressing dominant negative FADD thymidine analog (AZT) and not by other antiviral or DcR2 and soluble TRAIL receptors inhibited AZT nucleosides and that this eect was also inhibited by and IFN-a mediated cytotoxicity in PEL cells. This thymidine. We have recently noted that high dose AZT demonstrates that AZT and IFN-a mediated apoptosis has marked clinical activity in AIDS related EBV+ in PEL involves signaling through FADD and TRAIL/ primary central nervous system lymphoma (Raez et al., DR interaction. In contrast to what has been reported 1999). We recently observed a marked clinical im- in etoposide treated tumors (Gibson et al., 2000), we provement in a PEL patient with lymphomatous found no evidence that AZT, at the doses used, induced meningitis treated with parenteral AZT and IFN-a TRAIL receptors DR-4 and DR-5 (data not shown). after having failed conventional chemotherapy. We also found that the eect of IFN-a in PEL cells can The ability to speci®cally induce TRAIL mediated be reproduced with soluble TRAIL and apoptosis apoptosis in PEL may prove quite clinically relevant. potentiated by AZT. This indicates that in PEL, AZT Laboratory and clinical studies should further de®ne may promote signaling through death receptors rather the anti-tumor mechanism of these agents and their than by enhancing their (DR-4 and DR-5) expression. therapeutic potential. A similar phenomenon has been reported in cells transfected with Herpes Simplex Virus (HSV) TK. Apoptosis in these cells was shown to be associated with recruitment of components of the DISC mediated Materials and methods by phosphorylation of the pro-drug GCV (Beltinger, et al., 1999). In our study, the viral kinase was present in Cell lines each tumor cell, therefore transfection was unnecessary. PEL cell lines, BCBL-1, BC-2, BC-3, and BL-1 were obtained Phosphorylated AZT may enhance death receptor from the American Type Culture Collection (ATCC, mediated apoptosis via this mechanism which is Manassas, VA, USA). BC-5 was donated by Dr Ethel accentuated by IFN-a mediated induction of TRAIL. Cesarman of Cornell University. BCBL-1 and BC-3 are Another possibility is that AZT may block NF-kappaB infected with only HHV-8, while BC-1, BC-2 and BC-5 are which has been shown to be activated by death co-infected with HHV-8 and EBV. The primary PEL isolate receptors (Chaudhary et al., 1997; Yang et al., 2000). BCBL-LN was obtained from a diagnostic paracentesis This may result in an unfettered apoptotic eect. It is performed on an AIDS patient with lymphomatous ascites. possible that AZT may induce mitochondrial damage The lymphoma did not express B cell surface antigens (CD19, in HHV-8+ PEL which alone could cause a mild CD20) but its lineage was B cell as de®ned by gene apoptotic eect and recruitment of FADD and rearrangement studies. Analysis of tumor DNA by PCR was strongly positive for HHV-8 and EBV (Arvanitakis et al., procaspase 8. Upon addition of IFN-a (or soluble 1996). BL-5, BL-7, BL-8, and SM-1 are primary cell lines TRAIL) apoptosis is accentuated through activation of derived from AIDS patients with EBV+ AIDS related BL a DR/Ligand signal. This would explain the detection and carry the typical t8 : 14 c-myc translocation. P3HR-1, of caspase 3 cleavage in PEL cells treated only with BJAB and Ramos are established BL lines and were obtained AZT. It is also possible that AZT alone may activate a from the ATCC. P3HR-1 is EBV+ and BJAB and Ramos Type II apoptotic pathway while AZT and IFN-a are EBV7. together activate a Type I pathway response. The co- existence of Type I and II pathways has recently been Apoptosis analyses demonstrated in Jurkat cells treated with tumor necrosis factor alpha (Johnson et al., 2000). HHV-8 Apoptosis was determined by annexin V-FITC/propidium iodide (P.I.) ¯ow analysis. 5.06105 cells were grown in 10 ml also encodes a putative oncogene, vIRF, which inhibits of IMDM (GIBCO ± BRL) supplemented with 10% heat IFN signaling as well as a viral inhibitor of FADD inactivated fetal bovine serum (FBS) in the presence of dependent apoptosis, vFLIP (Gao et al., 1997; Sarid et media, 10 mg/ml AZT, ddC, or d4T, 1000 u/ml IFN-a (or al., 1999). It is unknown whether the combination of 10 ng/ml soluble TRAIL) (provided by Immunex Corpora- AZT and IFN-a aect expression of these virally tion, Seattle, WA, USA), 10 mg/ml AZT (or ddC or d4T) and encoded anti-apoptotic proteins. 1000 u/ml IFN-a (or 10 ng/ml soluble TRAIL), for 48 h in

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7038 25 cm ¯asks. 2.06105 cells were removed from the ¯ask and cia Biotech, Uppsala, Sweden) and visualized using ECL washed twice with 5 ml of PBS, resuspended in 0.1 ml of stain (Amersham). binding buer (10 mM HEPES/NaOH, 140 mM NaCl, 2.5 m CaCl ) and stained with 5 ml of annexin V-FITC M 2 Flow cytometry for TRAIL, DR-5 and DcR2 (PharMingen, San Diego, CA, USA) and 2.5 ml of 0.5 mg/ml P.I. for 15 min. 0.4 ml of binding buer was added to the 1.06105 cells were grown in 10 ml of IMDM (GIBCO ± suspension. Cells were then analysed using a FACScanTM BRL) supplemented with 10% FBS in the presence of media ¯ow cytometer (Becton Dickinson, San JoseÂ , CA, USA). or 1000 u/ml IFN-a. Cells were harvested after 24 h, washed For caspase inhibitor studies, cells (1.06105 cells/ml, once with 1.0 ml of PBS containing 0.5% BSA, pelleted at 50 ml) were grown in IMDM (GIBCO ± BRL) supplemented 1000 g, and resuspended in 50 ml of PBS. 2.06105 cells in with 10% heat inactivated FBS in the presence of media, 20 ml volume was used for detection of surface TRAIL 10 mg/ml AZT, 1000 u/ml IFN-a,10mg/ml AZT and 1000 u/ protein. Nonspeci®c binding sites were blocked by the ml IFN-a,10mg/ml AZT and 1000 u/ml IFN-a with 10 mM addition of 20 ml of human IgG (4.1 mg/ml, Sigma Chemical of Ac-DEVD-CHO (PharMingen) for 8 h. An additional Co., St. Louis, MO, USA) and incubated on ice for 1 h. 10 mM of Ac-DEVD-CHO was then added to the Ac-DEVD- Two ml of mAb anti-TRAIL (2.0 mg/ml, Immunex M181) or CHO treated cells and incubated for an additional 24 h. Cells Isotype mouse IgG (2.0 mg/ml, Sigma Chemical Co.) were were then harvested, lysed, and subjected to Western blot then added to the suspension and incubated for another analysis in a 12% SDS ± PAGE. Caspase 3 and caspase 8 hour. After three washes with 1.0 ml of PBS containing 0.5% protein bands were detected using mAb (PharMingen). This BSA, cells were resuspended in 50 ml of sheep anti-mouse IgG caspase 8 antibody detects only the uncleaved procaspase FITC conjugated (1 : 50, Sigma Chemical Co.) and incubated form. in ice for 30 min. After three washes with 1.0 ml of PBS, cells were analysed using a FACScanTM ¯ow cytometer (Becton Dickinson). Cell viability assays For DR-5 surface expression, 1.06106 cells were washed To determine the eects of antiviral nucleosides and IFN-a three times in ice cold PBS. Cells were then incubated in on the dierent lines, 0.56105 cells were grown in 1.0 ml of 100 mg of human IgG (Sigma) in PBS in a total volume of IMDM supplemented with 10% heat inactivated FBS (in 50 ml for 60 min. The cells were washed once with 1.0 ml of triplicate) in the presence of media, 10 mg/ml AZT (3TC or 0.1% BSA in PBS and resuspended in 20 ml of 4.0 mg DR-5 d4T), 1000 u/ml IFN-a,10mg/ml AZT and 1000 u/ml IFN-a mAb (Immunex, M413) and 0.1% BSA in PBS and incubated for 48 h. Control experiments were also performed with on ice for 60 min. After three washes with 0.1% BSA in PBS, etoposide at a concentration of 10 mg/ml for 24 h. Cells were 50 ml of sheep anti-mouse-FITC labeled (1 : 100 dilution in harvested and viability was determined by Trypan blue 0.1% BSA in PBS; Sigma) was added to the cells and exclusion. To determine the eects of thymidine on AZT and incubated for 30 min. Cells were then washed two times with IFN-a mediated apoptosis, cells were cultured as described 1.0 ml of PBS, then resuspended in 0.5 ml of binding buer above for 48 h in the presence of 10 or 100 mg/ml thymidine and analysed using a FACScanTM ¯ow cytometer (Becton and viability determined by Trypan blue exclusion. To study Dickinson). the eect of inhibitory DR-4 and DR-5 fusion proteins on The expression level of DcR2 on the surface of transfected AZT and IFN-a mediated apoptosis in PEL, a triplicate set and wild type PEL cells was evaluated by ¯ow cytometry. 4 of 2.0610 BCBL-1 PEL cells were grown in 100 mL of 10% Cells were stained with 50 mg/ml of goat anti-DcR2 anity FBS IMDM media, in the presence of 5.0 ng/ml of puri®ed polyclonal antibody (R&D), followed by staining rhTRAIL-R2:Fc (Alexis Corporation, San Diego, CA, with 1 : 200 diluted donkey anti-goat FITC labeled antibody USA), 2.0 mg/ml of rhTRAIL-R1:Fc (Alexis Corporation) (Jackson Immunoresearch Laboratories). As a control, cells and/or 1.0 mg/ml of ZB4, a CD-95 blocker, (Immunotech, were stained with the second antibody only. To reduce non- Beckman/Coulter Corporation, Brea, CA, USA) for 8 h. speci®c binding, stainings were done in the excess of human Ten mg/ml AZT and 1000 u/ml IFN-a (or 10 ng/ml soluble IgG. TRAIL in place of IFN-a) were then added to the media for another 48 h. As a control assay, 100 ng/ml (a cytotoxic RNAse protection assay dose) of soluble TRAIL (Immunex) was used with and without soluble blockers. Cells were harvested and viability 1.06105 cells/ml were grown in 50 ml IMDM (GIBCO ± determined by Trypan exclusion. BRL) supplemented with 10% heat inactivated FBS in the presence of 10 mg/ml AZT, 1000 u/ml IFN-a,10mg/ml AZT and 1000 u/ml IFN-a,10mg/ml etoposide or medium for 8 h. Western blot analysis Cells were harvested and total RNA's extracted using a 1.06105 cells/ml were grown in 50 ml IMDM (GIBCO ± RNAeasy puri®cation kit (QIAGEN Genomics, Inc., Bothell, BRL) supplemented with 10% heat inactivated FBS in the WA, USA). Twenty mg of RNA's were used per reaction. presence of 10 mg/ml AZT, 1000 u/ml IFN-a,10mg/ml AZT RPAs were done using RiboQuant RPA (PharMingen) kit. and 1000 u/ml IFN-a,10mg/ml etoposide or medium for Assays were done according to manufacturer's instructions 48 h. Cells were harvested and lysed in lysis buer (50 mM using customized sets of probes synthesized by PharMingen. TRIS-HCl pH 7.5, 0.5% NP-40, 10% glycerol, 250 mM NaCl, and 5 mM EDTA). Forty mg of total protein per lane Stable DNA transfections was loaded and subjected to electrophoresis in a 15% acrylamide SDS ± PAGE. Proteins were transferred from the Cells were transfected with dominant negative FADD (19) in acrylamide gel to nitrocellulose membranes. Membranes were a 35 mm dish using NovaFECTOR cationic lipid (VennNova blocked with 5% non-fat dry milk, then probed with 1 mg/ml LLC, Pompano Beach, FL, USA). 1.06106 of BCBL-1 cells mAb anti-caspase-8, anti-caspase 3 (PharMingen), or mAb were grown to log phase in IMDM media supplemented with anti-FADD (Transduction Laboratory, Lexington, KY, 10% heat inactivated FBS. Cells were then washed twice with USA) complexed with anti-mouse-HRP (Amersham Pharma- 10 ml of serum-free IMDM media and centrifuged at 1000 g

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7039 prior to the addition of DNA/lipid complex. Four mgof limiting dilutions. Dominant negative FADD transfectants pcDNA3 (Invitrogen, Carlsbad, CA, USA) or pcDNA3- were con®rmed by Western blot using mouse mAb anti- FADD DN were complexed with 20 mg of lipid for 15 min at FADD (Transduction Laboratories). The expression of DcR2 room temperature in 1.0 ml of serum-free IMDM media. Cell transfectants was assessed by ¯ow cytometry. pellets were then resuspended with the 1.0 ml of DNA/lipid complex and incubated in 5% CO2 at 378C for 4 h. One ml of IMDM supplemented with 20% heat inactivated FBS was added to the dish and incubated for 16 h. The full-length TRAIL receptor 4 (decoy receptor 2), DcR2 was ampli®ed by RT ± PCR from poly-A of PHA-stimulated peripheral blood Acknowledgments lymphocytes and cloned into episomal expression vector The authors wish to thank Drs Parkash Gill (University of pBMG-Neo. The construct was introduced into BCBL cells Southern California) and Scott Kaufmann (Mayo clinic, by electroporation. Cells were then transferred to a 75 cm Rochester, MN) for their helpful suggestions. This work ¯ask and grown in 50 ml of IMDM supplemented with 10% was supported by grants CA82274 (WJ Harrington Jr and heat inactivated FBS and 500 mg/ml G418 for selection of LH Boise), CA77837 (WJ Harrington Jr and LH Boise), positive transfectants. Fresh medium was changed every 4 CA86431 (GN Barber), CA80228 and CA39201 (ER days for a duration of 4 weeks. Single clones were selected by Podack) from the National Institutes of Health.

References

Arner ES, Valentin A and Eriksson S. (1992). J. Biol. Chem., Grith TS, Wiley SR, Kubin MZ, Sedger LM, Maliszewski 267, 10968 ± 10975. CR and Fanger NA. (1999). J. Exp. Med., 189, 1343 ± Arvanitakis L, Mesri EA, Nador RG, Said JW, Asch AS, 1354. Knowles DM and Cesarman E. (1996). Blood, 88, 2648 ± Gustafson EA, Schinazi RF and Fingeroth JD. (2000). J. 2654. Virol., 74, 684 ± 692. Ashkenazi A and Dixit VM. (1999). Curr. Opin. Cell. Biol., Jo M, Kim TH, Seol DW, Esplen JE, Dorko K, Billiar TR 11, 255 ± 260. and Strom SC. (2000). Nat. Med., 6, 564 ± 567. Balachandran S, Kim CN, Yeh WC, Mak TW, Bhalla KN Johnson BW, Cepero E and Boise LH. (2000). J. Biol. Chem., and Barber GN. (1998). EMBO J., 17, 6888 ± 6902. 275, 31546 ± 31553. Balachandran S, Roberts PC, Kipperman T, Bhalla KN, Kischkel FC, Hellbardt S, Behrmann I, Germer M, Pawlita Compans RW, Archer DR and Barber GN. (2000). J. M, Krammer PH and Peter ME. (1995). EMBO J., 14, Virol., 74, 1513 ± 1523. 5579 ± 5588. Barry M, Heibein JA, Pinkoski MJ, Lee SF, Moyer RW, Knowles DM. (1999). Mod. Pathol., 12, 200 ± 217. Green DR and Bleackley RC. (2000). Mol. Cell Biol., 20, LeeRK,CaiJ-P,DeyevV,GillPS,CabralL,WoodC, 3781 ± 3794. AgarwalRP,XiaW,BoiseLH,PodackEandHarrington Beltinger C, Fulda S, Kammertoens T, Meyer E, Uckert W Jr WJ. (1999). Cancer Res., 59, 5514 ± 5520. and Debatin KM. (1999). Proc. Natl. Acad. Sci. USA, 96, Levine AM. (2000). Semin. Oncol., 27, 442 ± 453. 8699 ± 8704. Magnusson C and Vaux DL. (1999). Immunol. Cell Biol., 77, Bodmer J-L, Holler N, Reynard S, Vinciguerra P, Schneider 41 ± 46. P, Juo P, Blenis J and Tschopp J. (2000). Nat. Cell. Biol., 2, MedemaJP,ScadiC,KischkelFC,ShevchenkoA,Mann 241 ± 243. M, Krammer PH and Peter ME. (1997). EMBO J., 16, Cannon JS, Hamzeh F, Moore S, Nicholas J and Ambinder 2794 ± 2804. RF. (1999). J. Virol., 73, 4786 ± 4793. MingariMC,PonteM,VitaleC,BellomaRandMorettaL. Chaudhary PM, Eby M, Jasmin A, Bookwalter A, Murray J (2000). Hum. Immunol., 61, 44 ± 50. and Hood L. (1997). Immunity, 7, 821 ± 830. Mullaney BP, Ng VL, Herndier BG, McGrath MS and De Clercq E. (1992). AIDS Res. Hum. Retroviruses, 8, 119 ± Pallavicini MG. (2000). Arch. Pathol. Lab. Med., 124, 134. 824 ± 826. Demario MD and Liebowitz DN. (1998). Semin. Oncol., 25, Nador RG, Cesarman E, Chadburn A, Dawson DB, Ansari 492 ± 502. MQ, Sald J and Knowles DM. (1996). Blood, 88, 645 ± 656. Fanger NA, Maliszewski CR, Schooley K and Grith TS. Okano M and Gross TG. (2000). Am.J.Med.Sci.,319, 392 ± (1999). J. Exp. Med., 190, 1155 ± 1164. 396. Findenig G, Mader RM, Fritzer-Szekeres M, Steger GG, Peter ME, Heufelder AE and Hengartner MO. (1997). Proc. Jaeger W and Szekeres T. (1996). Oncol. Res., 8, 189 ± 196. Natl. Acad. Sci. USA, 94, 12736 ± 12737. Gaidano G, Capello D, Fassone L, Gloghini A, Cilia AM., Player MR and Torrence PF. (1998). Pharmacol. Ther., 78, Ariatti C, Buonaiuto D, Vivenza D, Gallicchio M, Avanzi 55 ± 113. GC, Prat M and Carbone A. (2000). J. Clin. Virol., 16, Raez L, Cabral L, Cai J-P, Landy H, Sfakianakis G, Byrne Jr 215 ± 224. GE, Hurley J, Scerpella E, Jayaweera D and Harrington Jr Gaidano G, Pastore C, Lanza C, Mazza U and Saglio G. WJ. (1999). AIDS Res. Hum. Retroviruses, 15, 713 ± 719. (1994). Ann. Hematol., 69, 281 ± 290. Sarid R, Wiezorek JS, Moore PS and Chang Y. (1999). J. Gao S-J, Bosho C, Jayachandra S, Weiss RA, Chang Y and Virol., 73, 1438 ± 1446. Moore PS. (1997). Oncogene, 15, 1979 ± 1985. Swinnen LJ. (2000). Ann. Oncol., 11 (Suppl. 1), 45 ± 48. Gibson SB, Oyer R, Spalding AC, Anderson SM and Swinnen LJ. (1999). Semin. Oncol., 26 (5 Suppl. 14), 21 ± 25. Johnson GL. (2000). Mol. Cell Biol., 20, 205 ± 212. Walczak H and Krammer PH. (2000). Exp. Cell. Res., 256, Goldsby RE and Carroll WL. (1998). J. Pediatr. Hematol. 58 ± 66. Oncol., 20, 282 ± 296.

Oncogene Interferon induces TRAIL in primary effusion lymphoma NL Toomey et al 7040 YangCH,MurtiA,PfeerSR,BasuL,KimJGandPfeer Zamanian-Daryoush M, Mogensen TH, DiDonato JA and LM. (2000). Proc Natl Acad Sci USA, 97, 13631 ± 13636. Williams BR. (2000). Mol. Cell Biol., 20, 1278 ± 1290. Yeh WC, Pompa JL, McCurrach ME, Shu HB, Elia AJ, Zavaglia C, Airoldi A and Pinzello G. (2000). J. Clin. Shahinian A, Ng M, Wakeham A, Khoo W, Mitchell K, Gastroenterol., 30, 234 ± 241. El-Deiry WS, Lowe SW, Goeddel DV and Mak TW. (1998). Science, 279, 1954 ± 1958.

Oncogene