Oncogene (2000) 19, 4669 ± 4684 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc The c-Myc-interacting adaptor Bin1 activates a caspase-independent cell death program

Katherine Elliott1, Kai Ge1,3, Wei Du1,2 and George C Prendergast*,1,2

1The Wistar Institute, Philadelphia, PA, USA; 2Glenolden Laboratory, DuPont Pharmaceuticals Company, Glenolden, PA, USA

Cell death processes are progressively inactivated during Introduction malignant development, in part by loss of tumor suppressors that can promote cell death. The Bin1 Cell suicide programs are crucial to development and encodes a nucleocytosolic adaptor protein with tumor homeostasis. A major role of these programs is to suppressor properties, initially identi®ed through its stanch inappropriate cell proliferation that can lead to ability to interact with and inhibit malignant transforma- cancer. Indeed, loss of the capacity for programmed tion by c-Myc and other oncogenes. Bin1 is frequently cell death (PCD) is a hallmark of the malignant cell. missing or functionally inactivated in breast and prostate Loss of this capacity is not due to inactivation of the cancers and in melanoma. In this study, we show that machinery responsible for the major form of PCD, Bin1 engages a caspase-independent cell death process apoptosis, which is minimally comprised of cytochrome similar to type II apoptosis, characterized by cell c, apoptosis promoting factors (Apafs), and caspases shrinkage, substratum detachment, vacuolated cyto- (Reed et al., 1998). Instead, it appears that malignant plasm, and DNA degradation. Cell death induction was cells suppress or eliminate signals needed to commit to relieved by mutation of the BAR domain, a putative PCD and/or to activate the apoptotic machinery. e€ector domain, or by a missplicing event that occurs in How cells commit to die and how malignant cells melanoma and inactivates suppressor activity. Cells in all sidestep this decision are questions of great interest in phases of the cell cycle were susceptible to death and p53 the areas of programmed cell death and cancer and Rb were dispensable. Notably, Bin1 did not activate research. Caspase regulation is important, but there is caspases and the broad spectrum caspase inhibitor emerging evidence that caspase-independent processes ZVAD.fmk did not block cell death. Consistent with may also have important roles. Death receptors the lack of caspase involvement, dying cells lacked directly activate caspases and cancer cells neutralize nucleosomal DNA cleavage and nuclear lamina degrada- these routes by multiple strategies (Ashkenazi and tion. Moreover, neither Bcl-2 or dominant inhibition of Dixit, 1998). Mitochondria indirectly regulate caspases, the Fas pathway had any e€ect. In previous work, we by controlling the release of cytochrome c release and showed that Bin1 could not suppress cell transformation thereby the status of Apaf, which controls the by SV40 large T antigen. Consistent with this ®nding, we activation of caspase-9 (Green and Reed, 1998). This observed that T antigen suppressed the death program route is blunted in cancer cells primarily by alterations engaged by Bin1. This observation was interesting in in the level of Bcl-2 family , which control light of emerging evidence that T antigen has roles in cell cytochrome c release (Chao and Korsmeyer, 1998; immortalization and human cell transformation beyond Reed et al., 1998; Thompson and Vander Heiden, Rb and p53 inactivation. In support of a link to c-Myc- 1999), but probably at other levels as well (Ding et al., induced death processes, AEBSF, a serine protease 1998; Fearnhead et al., 1997). Caspase activation is inhibitor that inhibits apoptosis by c-Myc, potently clearly sucient for death commitment but whether it suppressed DNA degradation by Bin1. Our ®ndings is necessary is much less clear. A viable hypothesis is suggest that the tumor suppressor activity of Bin1 that caspase-independent processes participate in re¯ects engagement of a unique cell death program. commitment and that caspase activation seals a PCD We propose that loss of Bin1 may promote malignancy decision made by the cell (Amarante-Mendes et al., by blunting death penalties associated with oncogene 1998; McCarthy et al., 1997; Thompson and Vander activation. Oncogene (2000) 19, 4669 ± 4684. Heiden, 1999). If so, then caspase-independent pro- cesses may be disrupted in cancer cells like caspase Keywords: transformation; apoptosis; cancer; amphi- activation pathways. physin-like protein/amphl One area of investigation into cancer cell death mechanisms centers on how c-Myc stimulates PCD and why it does not do so in malignancy (Prendergast, 1999). Oncogenic activation of c-Myc promotes the development of many clinically signi®cant cancers, such as those of the breast, colon, lung, and prostate (Cole, 1986; Garte, 1993; Jenkins et al., 1997). c-Myc activation usually occurs at later stages in carcinoma in humans and is usually a poor prognostic marker. *Correspondence: G C Prendergast, Glenolden Laboratory, DuPont However, in premalignant cells c-Myc is a robust Pharmaceuticals Co., Glenolden, PA, USA, 3Current address: The Rockefeller University, New York, NY, USA stimulator of PCD. Therefore, to exploit the growth- Received 13 March 2000; revised 2 May 2000; accepted 10 May promoting aspects of c-Myc, malignant cells must 2000 evolve strategies to escape the death penalty associated Bin1 induces caspase-independent PCD K Elliott et al 4670 with its activation. Mechanistic investigations in isoforms of Bin1 include sequences that are required to ®broblast and lymphocyte models have de®ned central interact with clathrin and the endocytosis regulatory roles for the tumor suppressors p53 and p19ARF, which complex AP-2 (Ramjaun and McPherson, 1998). Third, are frequently inactivated in human cancer (Sherr, muscle-speci®c isoform, which localizes to the nucleus 1998). Whether these mediate PCD by c-Myc or and binds c-Myc, is required for myoblasts to with- sensitize cells to its action is unclear. Nevertheless, it is draw from the cell cycle and to terminally di€erentiate evident that p53 inactivation does not compromize the (Mao et al., 1999; Wechsler-Reya et al., 1998). Lastly, ability of c-Myc to drive PCD in epithelial cells isoforms that localize to the nucleus and bind c-Myc (Prendergast, 1999; Sakamuro et al., 1995), indicating exhibit tumor suppressor properties which are inacti- that p53-independent mechanisms are also important. vated or missing in malignant melanoma, breast Interestingly, careful investigations have revealed that cancer, and prostate cancer (Ge et al., 1999, 2000a,b). caspase inhibition or Bcl-2 overexpression does not In contrast, and brain isoforms of Bin1 abolish the ability of c-Myc to commit cells to undergo lack suppressor activity. Indeed, one way by which PCD. Caspase inactivation eliminates nuclear pheno- Bin1 is functionally inactivated in cancer is by types characteristic of apoptosis and slows the kinetics missplicing of one of its brain-speci®c exons (Ge et of cell death, but it does not abolish all apoptotic al., 1999). Thus, Bin1 has two functions, one of which phenotypes nor ultimate cellular demise (Amarante- is linked to nuclear processes that in¯uence cell fate. In Mendes et al., 1998; Cecconi et al., 1998; McCarthy et this study, we provide evidence that the tumor al., 1997; Soengas et al., 1999; Yoshida et al., 1998). suppressor properties of Bin1 are related to induction Similarly, Bcl-2 proteins signi®cantly delay but do not of cell death and that Bin1 participates in a caspase- abolish PCD commitment induced by c-Myc either in independent process similar to type II apoptosis. vitro or in vivo (McCarthy et al., 1997; Trudel et al., 1997; Tsuneoka and Mekada, 2000). Thus, while c-Myc promotes PCD by activating caspases (Kangas et al., Results 1998), and Bcl-2 cooperates with c-Myc to promote malignancy by delaying this process, c-Myc apparently Bin1 activates a programmed cell death process in also a€ects caspase-independent processes that in¯u- malignant cells ence death commitment. Inactivation of such processes may be important in epithelial cells where c-Myc We have shown that reintroduction of Bin1 into activation occurs. This may be especially true at stages human cancer cell lines that lack endogenous expres- when malignant cell division is slow and inecient sion leads to loss of proliferative capacity and cellular modes of cell death which do not involve caspases may demise (Elliott et al., 1999; Ge et al., 1999, 2000a; be e€ective at stanching tumor outgrowth. Thus, Sakamuro et al., 1996). To investigate this phenomen- inactivation of caspase-independent processes may on in more detail, we used a set of recombinant contribute to tumorigenesis by helping cells escape adenoviral vectors that eciently deliver Bin1 to death penalties associated with activation of c-Myc or human cells. One vector was constitutive and used other oncogenes (Prendergast, 1999). the cytomegalovirus (CMV) early promoter to drive In this study, we investigated a role in PCD for Bin1 Bin1 expression (Ad-Bin1). A second vector was (Bridging INtegrator-1), one of an emerging set of c- inducible and allowed Bin1 expression to be controlled Myc-interacting adaptor proteins that are candidates by coexpression of Cre recombinase. In this vector, for regulating c-Myc or mediating its diverse actions in Bin1 cDNA was inserted downstream of a CMV cells (Sakamuro and Prendergast, 1999). Bin1 function promoter and a stu€er cassette ¯anked by loxP sites is complex and varied by tissue-speci®c splicing. It was (Ad-MABin1). A matched control virus was also identi®ed initially through its ability to interact with constructed for each system in which a loss-of-function and inhibit malignant transformation by c-Myc Bin1 gene was expressed (see below). Recombinant (Sakamuro et al., 1996). Subsequent investigations viruses were generated in human 293 cells by standard established that there are two ubiquitous slice isoforms methods. DNA sequencing from ampli®ed viral DNA of Bin1 and several other splice forms that are con®rmed the expected structure of each transgene restricted in expression to muscle or brain (Butler et (data not shown). For control experiments, we also al., 1997; Ramjaun et al., 1997; Tsutsui et al., 1997; prepared adenoviruses expressing LacZ (Ad-LacZ) or Wechsler-Reya et al., 1997b, 1998). Bin1 polypeptides Cre recombinase (Ad-cre) in 293 cells. The host cell are related in their terminal domains to amphiphysin, a line used for infection was HepG2 hepatoma cells, a neuronal protein involved in synaptic vesicle endocy- functionally null cell which lacks endogenous Bin1 tosis, and brain isoforms which are most similar have expression and is susceptible to Bin1-mediated growth been termed alternately amphiphysin II or amphiphy- suppression (Elliott et al., 1999; Sakamuro et al., 1996). sin-like (amphl). However, amphiphysin was named for Infections employing Ad-lacZ indicated that a its biochemical properties, rather than its function, and multiplicity of infection (m.o.i.) of 50 ± 100 was outside the brain Bin1 has functions that are not required to infect 490% of HepG2 cells exposed to amphiphysin-like. First, although the brain isoforms of virus in vitro (data not shown). HepG2 cells infected Bin1 are cytosolic, the ubiquitous Bin1 isoforms with Ad-Bin1 at 10, 50 or 100 m.o.i. exhibited localize to the nucleus as well as to the cytosol (Kadlec increasing levels of Bin1 expression as documented by and Pendergast, 1997; Wechsler-Reya et al., 1997a). Western analysis of cell extracts prepared and analysed Consistent with their nuclear localization, these iso- 48 h after infection (see Figure 1a). Bin1 was detectable forms functionally interact with the nuclear tyrosine within 12 h of infection and reached a maximum by kinase c-Abl as well as with c-Myc (Elliott et al., 1999; 48 h (see Figure 1b). When infected under similar Kadlec and Pendergast, 1997). Second, only brain conditions, the level of ectopic Bin1 in IMR90 diploid

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4671

Figure 1 Recombinant Bin1 adenoviruses. Western analysis of whole cell lysates prepared from cells infected with the adenoviral vectors indicated was performed. The anti-Bin1 monoclonal antibodies 99D were used as the primary antibodies in panels a, b, and d; antibodies 99D plus 99I were used in panel c (Wechsler-Reya et al., 1997a). (a) Expression from the constitutive vector Ad-Bin1. HepG2 cells lacking endogenous Bin1 were incubated with viruses at the multiplicity of infection (m.o.i.) indicated and extracts were prepared 48 h later. (b) Ad-Bin1 time course. Extracts were prepared from HepG2 cells harvested at the times indicated. (c) Comparative levels of expression in IMR90 diploid ®broblasts or HepG2 cells. Extracts were prepared 48 h after infection at the indicated m.o.i. IMR90 express endogenous Bin1 (Sakamuro et al., 1996) but at a level that is undetectable on this exposure of the blot, which illustrates similar steady-state levels of Bin1 in IMR90 or HepG2 cells infected with the same amount of Ad-Bin1. (d) Cre-inducible expression from Ad-MABin1. HepG2 cell lines stably expressing P1 bacteriophage Cre site-speci®c recombinase (HepG2/cre) or containing only vector sequences (HepG2/CMV) were incubated with the virus indicated (m.o.i.=100) and extracts were prepared 48 h later. The transgene in Ad-MABin1 is located downstream of a loxP site-¯anked stu€er sequence such that expression occurs only after Cre-induced recombination (see Materials and methods). Ad-vect is a control virus that contain no transgene. (e) Expression of the loss-of-function deletion mutant Bin1DBAR-C. Western analysis was performed using extracts isolated from cells infected with 100 m.o.i. Ad-Bin1 or Ad-Bin1DBAR-C, which lacks anti-transforming and tumor suppressor properties (Elliott et al., 1999b)

®broblasts used in control experiments was comparable appearance of this polypeptide did not a€ect the loss- to that seen in HepG2 cells (see Figure 1c). Indirect cell of-function of Bin1DBAR-C, which was employed as a immuno¯uorescence experiments indicated that the negative control for cell death induction (see below). In high level of expression driven by Ad-Bin1 was summary, these experiments validated the adenoviral correlated with localization of Bin1 throughout the vectors for constitutive or inducible Bin1 expression in cell. To document Bin1 expression from the Cre- human cells. inducible vector Ad-MABin1, we infected HepG2 cell HepG2 cells expressing Bin1 displayed striking lines that stably expressed Cre recombinase (HepG2/ morphological changes. Cells assumed a rounded, cre) or that contained only vector sequences (HepG2/ shrunken morphology and exhibited deformations of CMV). Bin1 was detected in HepG2/cre cells infected the plasma membrane before completely detaching with 50 m.o.i. of Ad-MABin1 (see Figure 1d). In from substratum (see Figure 2a). The changes seen contrast, Bin1 was not detected after infection of were consistent with induction of PCD like that HepG2/CMV cells, where the stu€er upstream of Bin1 observed previously (Ge et al., 1999, 2000a). Cells could not be removed, nor was it detected in either cell infected with Ad-lacZ at similar or higher m.o.i. did line infected with control viruses that lacked an insert not display similar morphologies. No signs of cellular (Ad-vect). A constitutive adenoviral vector was also demise were noted following infections of normal constructed for a loss-of-function mutant of Bin1, human diploid IMR90 ®broblasts with Ad-Bin1 (see termed Bin1DBAR-C, that lacks a.a. 125 ± 206 within Figure 2a). IMR90 cells showed some enlargement but the N-terminal BAR domain which is crucial for ¯ow cytometry indicated no signs of cell cycle arrest antineoplastic activity (Elliott et al., 1999). Extracts (data not shown), consistent with previous observa- derived from cells infected with Ad-Bin1DBAR-C tions in rodent embryo ®broblasts, Rat1 ®broblasts, or exhibited a polypeptide with the expected apparent mouse C2C12 myoblasts (Sakamuro et al., 1996; mobility of *50 kD (see Figure 1e). A second smaller Wechsler-Reya et al., 1998; unpublished observations). polypeptide was also detected at lower levels relative to Thus, Bin1 was not grossly toxic to cells, consistent the Bin1DBAR-C protein on Western blots. The with previous observations (Ge et al., 1999, 2000a;

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4672

Figure 2 Bin1 induces demise of malignant cells. (a) Morphology of HepG2 hepatoma cells or IMR90 diploid ®broblasts following infection with Ad-LacZ or Ad-Bin1. Cells were incubated with 100 m.o.i. virus indicated and photographed 48 h later using phase- contrast optics (magni®cation=1006). (b) Morphology of cells infected with inducible Ad-MABin1 virus. HepG2 cells stably expressing Cre recombinase or vector only were incubated with 100 m.o.i. virus indicated and processed as above. (c) Viable cell count. Cells were infected with the viruses indicated, as in Figure 1, and harvested by trypsinization 48 h later. The proportion of viable cells in the population was determined by trypan blue exclusion. (d) Cell death is associated with DNA degradation and occurs throughout the cell cycle. Cells were incubated with the indicated m.o.i. of Ad-LacZ or Ad-Bin1 and processed 48 h later for TUNEL reaction, propidium iodide staining, and ¯ow cytometry. In each graph, the X-axis corresponds to relative PI staining and the Y-axis to the log of the FITC signal re¯ecting relative TUNEL positivity. Time courses indicated a correlation between the kinetics of the cytopathic e€ect and DNA degradation. (e) Cell death elicited by Bin1 occurs throughout the cell cycle. Cells were incubated with 50 m.o.i. Ad-LacZ or Ad-Bin1 and processed after the period indicated for TUNEL reaction, propidium iodide staining, and ¯ow cytometry. In each graph, the X-axis corresponds to relative PI staining and the Y-axis to the log of the FITC signal re¯ecting relative TUNEL positivity

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4673 Sakamuro et al., 1996; Wechsler-Reya et al., 1998). propidium iodide staining, TUNEL assay, and ¯ow Induction of the cre-inducible virus Ad-MABin1 cytometry 48 h later. A dose-dependent increase in the produced identical e€ects. In HepG2/CMV cells, the proportion of TUNEL-positive cells was observed after control virus Ad-vect had little morphological e€ect, Ad-Bin1 infection (see Figure 2d). TUNEL positivity whereas uninduced Ad-MABin1 had a slight e€ect on was detected in cells in all phases of the cell cycle as cell shape (see Figure 2b). The e€ect of the latter virus indicating by propidium iodide staining. An increase in might re¯ect low leaky expression seen in other cell cells with sub-G1 phase DNA occurred in the TUNEL- types infected by Ad-MABin1 (Ge et al., 1999). positive population with similar kinetics. A time-course However, the uninduced luciferase vector Ad-MA19 experiment con®rmed observations that the appearance produced similar e€ects and we observed no increase in of TUNEL-positive cells and the accumulation of cells detached cells using either vector (data not shown), with sub-G1 phase DNA began 24 ± 36 h after arguing for a nonspeci®c e€ect of the Ad-MA vector in infection, corresponding to a time approximately 12 ± HepG2 cells. In any case, there was a profound 24 h after Bin1 expression (see Figure 2e). This di€erence seen in HepG2/cre cells, where Ad-vect had experiment also highlighted the ®nding that apoptotic no discernible e€ect but Ad-MABin1 elicited a cells (as de®ned by TUNEL positivity) emerged from dramatic increase in detached cells (see Figure 2b). all phases of the cell cycle. Taken together, these results Taken together, these results con®rmed that the suggested that Bin1 induced a programmed cell death cytopathic e€ect elicited by the constitutive Ad-Bin1 (PCD) process. Similar experiments performed with the vector was not due to nonspeci®c toxicity of the vector inducible vector con®rmed these observations and system. con®rmed that they were dependent on Bin1 expression To explicitly assess cell viability in infected cultures, rather than a nonspeci®c e€ect of the adenoviral vector cells incubated with various m.o.i. of Ad-Bin1 or Ad- (data not shown). The features of DNA degradation lacZ were harvested 48 h after infection and the seen in ¯ow cytometry experiments paralleled the proportion of viable cells was determined by trypan kinetics of morphological features of apoptosis and blue exclusion. Cells infected with Ad-lacZ at all m.o.i. loss of viability. In multiple trials using each vector were 490% viable whereas cells infected with Ad-Bin1 system, no reproducible e€ects of Bin1 on the exhibited a linear relationship between m.o.i. and loss distribution of cells in the cell cycle were noted, of viability (see Figure 2c). Similar results were consistent with the observation that TUNEL-positive obtained using Ad-MABin1 (data not shown). Thus, cells emerged from all phases. Taken together, the the cytopathic phenotype elicited by Bin1 was results suggested that Bin1 engaged a PCD process that associated with cell death. could be initiated at any point in the cell cycle, similar We next examined whether cells induced to die by to c-Myc and other oncogenes (Evan et al., 1992). Bin1 exhibited any di€erences in endocytosis, because brain-speci®c splice isoforms of Bin1 that localize to BAR domain is crucial for Bin1 to induce cell death the cytosol in di€erentiated neurons have been implicated in this process, like the related adaptor The Bin1 BAR domain is crucial to inhibit transforma- protein amphiphysin (David et al., 1996; Ramjaun et tion of rodent ®broblasts by c-Myc and to suppress the al., 1997; Wigge et al., 1997). Ubiquitous splice proliferation of HepG2 cells (Elliott et al., 1999b). To isoforms of Bin1 that localize to the nucleus lack determine whether this domain was also crucial for brain-speci®c exons required for interaction with death activity, we infected cells with Ad-Bin1DBAR-C, clathrin-coated endocytotic vesicles and AP2 (Ramjaun which expresses the loss-of-function mutant Bin1D and McPherson, 1998; P. deCamilli, personal commu- BAR-C (Elliott et al., 1999). Ad-Bin1DBAR-C did nication). However, while this structural di€erence not elicit the cell detachment and cytopathic phenotype suggested strongly that endocytotic roles were probably produced by Ad-Bin1 (see Figure 3a). Similarly, ¯ow brain-speci®c, we wished to explicitly rule out a role for cytometric analysis of cells harvested 48 h after viral endocytosis in cell death induction by the nonneuronal infection showed that Ad-Bin1DBAR-C did not elicit Bin1 isoform used in these experiments. Fluid-phase or DNA degradation, even at elevated multiplicities of receptor-mediated endocytosis was monitored by infection (see Figure 3b). These observations provided comparing the rate of uptake of horseradish peroxidase additional evidence that the cytotoxic e€ects of Bin1 in or ¯uorescein-conjugated transferrin, respectively (Bar- HepG2 cells were not due to some nonspeci®c cause, bieri et al., 1998; Benmerah et al., 1998). No di€erences since overexpression of a loss-of-function mutation were apparent in the level of uptake of either reagent in that eliminated tumor suppressor properties also cells infected with Ad-Bin1 or Ad-LacZ (data not abolished PCD activity. We concluded that Bin1 shown). These observations supported the expectation engaged a PCD process that could explain the basis that the endocytotic function of the Bin1 gene is for its tumor suppressor properties. regulated by alternate splicing and that splice isoforms that localize to the nucleus do not have these roles p53 and Rb are dispensable for PCD by Bin1 (Kadlec and Prendergast, 1997; Sakamuro et al., 1996; Wechsler-Reya et al., 1997a, 1998). We concluded that p53 has a central role in many types of PCD but it is the ability of Bin1 to induce cell death was unrelated to mutated in many cancers. The retinoblastoma protein e€ects on endocytosis. (Rb) has been reported to be antiapoptotic in many To begin to de®ne the basis for cell death, we systems but in some cases it also has proapoptotic roles analysed HepG2 cells by ¯ow cytometry following (Bowen et al., 1998; Day et al., 1997). The central terminal transferase-catalyzed dUTP labeling of nicked importance of p53 and Rb in PCD control prompted DNA ends (TUNEL assay). Cells were infected with us to determine whether these genes were needed for increasing m.o.i. of adenoviral vector and harvested for Bin1 action. HepG2 cells have wild-type p53 and Rb

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4674

Figure 3 BAR deletion abolishes cell death by Bin1. (a) Lack of cytopathic e€ect. HepG2 cells were infected with 100 m.o.i. Ad- LacZ or Ad-Bin1 and photographed 48 h later (magni®cation=1006). (b) Flow cytometry. Cells were infected with the m.o.i. virus indicated and harvested and processed 48 h after infection for ¯ow cytometry

genes, so we examined the e€ect of Ad-Bin1 on nucleoplasm (see Figure 5a). However, cells maintained another tumor cell line, SAOS-2 osteosarcoma, which considerable genomic DNA integrity and there was no has homozygous deletions in both genes. SAOS-2 cells evidence of nuclear lamina breakdown. This phenotype were infected with 100 m.o.i. Ad-LacZ or Ad-Bin1 was di€erent from that seen in classical or type I virus and expression was con®rmed by b-galactosidase apoptosis, which is associated with a distinct nuclear staining or Western blotting, respectively (data not condensation phenotype, and suggestive of type II shown). Parallel dishes of cells were examined for the apoptosis in which cytosolic features predominate. The appearance of morphological features of apoptosis or contrast in nuclear morphology was highlighted by harvested and processed for TUNEL labeling and ¯ow comparison to the nuclear phenotype elicited by the cytometry. Similar to its e€ects on HepG2, Ad-Bin1 protein kinase inhibitor staurosporine, which induces caused cell rounding, plasma membrane deformations, type I apoptosis (see Figure 5a, right panel). Transient and substratum detachment (Figure 4a). Flow cyto- transfection of HepG2 cells with Bin1 plasmid vectors metry showed an increase in the number of TUNEL- elicited the same nuclear phenotype as Ad-Bin1, ruling positive cells with sub-G1 phase DNA following Ad- out a nonspeci®c adenovirus vector artifact (data not Bin1 infection. Positive cells emerged from all phases shown). To determine whether Bin1 caused nucleoso- of the cell cycle and kinetics were similar to those mal DNA cleavage (`DNA laddering'), genomic DNA observed in HepG2 cells (see Figure 4b). These results was isolated from cells infected with adenoviral vectors were consistent with evidence that p53 is dispensable or treated with staurosporine and fractionated by for PCD by c-Myc in epithelial cells (Sakamuro et al., agarose gel electrophoresis. Genomic DNA degrada- 1995; Trudel et al., 1997) and with ®ndings in breast tion was observed in cells infected by Ad-Bin1 but not cancer and melanoma cell lines that p53 status was not by Ad-lacZ (see Figure 5b, left panel). This result was correlated with Bin1 susceptibility (Ge et al., 1999, consistent with the ability of Bin1 to induce positive 2000a). Consistent with the SAOS-2 susceptibility, the TUNEL reaction. However, Bin1 did not induce results of extensive RNAse protection and Western nucleosomal DNA cleavage, even though this feature analyses of HepG2 cells expressing Bin1 did not reveal could be induced in HepG2 cells by staurosporine (see any di€erences in the expression of a variety of cell Figure 5b, right panel). cycle and apoptosis regulators, including p53, Rb The caspase-3 activated nuclease DFF/CAD is family members, cell cycle-dependent kinase inhibitors primarily responsible for nucleosomal DNA degrada- (p16INK4, p14ARF, p21WAF1, p27KIP1, or tion and chromatin collapse, which is a characteristic p57KIP2), or Bcl-2 family genes (Bcl-2, Bcl-XL, Mcl- of classical or type I apoptosis (Enari et al., 1998; Liu 1, Bik, Bax, or Bak). We concluded that Bin1 acted et al., 1997; Woo et al., 1998). The absence of these independently of the p53 and Rb pathways. features suggested that Bin1 did not activate caspase-3. To con®rm this likelihood, caspase-3-like activities were assayed in extracts from Ad-Bin1-infected cells Bin1 does not activate caspases by measuring cleavage of the ¯uorescent substrate Ac- The type of PCD process elicited by Bin1 was further DEVD-pNA (see Figure 5c). Cleavage of Ac-DEVD- characterized by examination of nuclear phenotypes pNA by caspase-3-like activity was monitored by associated with caspase activation. HepG2 cells in- absorbance at 405 nM measured at various times after fected with Ad-Bin1 but not Ad-LacZ displayed signs addition to extracts. Exogenous recombinant caspase-3 of nuclear deformation and chromatin margination at was used as a positive control for the assay. As the nuclear periphery and at focal sites in the expected, staurosporine induced caspase-3-like activity

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4675 In contrast, staurosporine-treated cells were shrunken and exhibited cytosolic vacuolation, chromatin con- densation, and nuclear degeneration (plasma mem- brane blebs were also observed but were not so dramatic in this cell system). Ad-Bin1-infected cells were similarly shrunken and heavily vacuolated. Margination of chromatin at the nuclear periphery was evident. However, there was no nuclear degenera- tion and limited chromatin condensation by compar- ison to staurosporine-treated cells. Cell nuclei were shrunken relative to Ad-LacZ-infected cells but not so severely as staurosporine-treated cells. Strikingly, the nuclear lamina in Ad-Bin1-infected cells remained essentially intact. This feature supported the lack of caspase involvement because lamins are subjected to caspase-mediated proteolysis during PCD (Lazebnik et al., 1993; Rao et al., 1996). In addition, lamin cleavage is separable from chromatin collapse (Lazebnik et al., 1995) but is important for complete nuclear degenera- tion (Rao et al., 1996), so the more limited nuclear degeneration in Ad-Bin1-infected cells was also con- sistent with a caspase-independent process. We ob- served no `exploded' cells or ¯occulent densities in organelles or the cytosol that would signal necrosis. Trials in which ZVAD.fmk was added con®rmed that Figure 4 p53 and Rb are dispensable for programmed cell death the PCD phenotype induced by Bin1 was not a€ected by Bin1. SAOS-2 osteosarcoma cells, which have homozygous by caspase inhibition. Addition of 100 mM ZVAD.fmk deletions of p53 and Rb, were infected with 100 m.o.i. Ad-LacZ reduced cell volume and induced the appearance of or Ad-Bin1. (a) Morphology. Cells were photographed 48 h after focal densities in the nucleus, but these features were a infection (magni®cation=1006). (b) Flow cytometry. Cells were harvested 48 h after infection and processed for TUNEL assay nonspeci®c artifact of ZVAD.fmk, because the same and propidium iodide staining. Representative results from features were also apparent in Ad-LacZ-infected cells. analysis by ¯ow cytometry are shown. The X-axis corresponds Interestingly, ZVAD.fmk did not block the Bin1 to relative PI staining and the Y-axis to the log of the FITC phenotype, in which cytosolic vacuolization predomi- signal re¯ecting relative TUNEL positivity nated. In contrast, ZVAD.fmk dramatically a€ected the phenotype of staurosporine-treated cells, reversing nuclear degradation to a large extent. We concluded in cell extracts. This induction con®rmed that HepG2 that Bin1 engaged a PCD process that was caspase- cells expressed pro-caspase-3-like enzymes that were independent yet associated with limited chromatin competent for activation by apoptotic stimuli, in degradation and cytosolic features of PCD. support of evidence of a predominent role for caspase-3 in classical apoptotic responses in HepG2 Cell death by Bin1 is not blocked by Bcl-2 or cells (Suzuki et al., 1998). In contrast, no signi®cant Fas pathway inhibition cleavage of Ac-DEVD-pNA occurred in extracts prepared from cells infected with Ad-Bin1 or Ad-LacZ. To further delineate the death process activated by In Western blotting experiments, we also did not detect Bin1 we investigated links to two classical PCD proteolytic cleavage of either pro-caspase-3 or the regulatory pathways. Bcl-2 proteins inhibit caspase caspase-3 substrate PARP (data not shown). activation through their ability to in¯uence mitochon- One possibility was that Bin1 activated a caspase drial physiology, whereas death receptors directly other than caspase-3. To assess this possibility, we activate caspases (Ashkenazi and Dixit, 1998; Thomp- tested whether Bin1-induced cell death was suppressed son and Vander Heiden, 1999). HepG2 cell lines by ZVAD.fmk, a broad spectrum inhibitor of caspases. overexpressing Bcl-2 or a dominant inhibitor of In these experiments, the relative number of cells FADD, which blocks Fas signals (Muzio et al., displaying sub-G1 phase DNA by ¯ow cytometry was 1996), were tested for response to Ad-Bin1. One used as a measurement of PCD. As expected, would predict the response would be una€ected by ZVAD.fmk signi®cantly blocked staurosporine-induced Bcl-2 or Fas pathway disruption if Bin1 acted in a cell death. In contrast, ZVAD.fmk did not a€ect PCD caspase-independent manner. Cells overexpressing Bcl- by Bin1 even when added at high concentrations (see 2 were resistant to staurosporine-induced apoptosis Figure 5d). Taken together, these results argued that relative to vector control cells (data not shown). In the death process activated by Bin1 was caspase- contrast, there was no di€erence in the susceptibility independent. of cells overexpressing Bcl-2 or the FADD dominant To con®rm this conclusion and gain additional inhibitor to death induction by Bin1 (see Figure 7a). insight into this death process we examined cells by On a di€erent line of work, we had observed that electron microscopy. These experiments revealed cyto- recombinant Bin1 baculoviruses increased the kinetics solic features consistent with a programmatic death of cell death in Sf9 cells, and that Bcl-2 could not process such as type II apoptosis (see Figure 6). Ad- inhibit this e€ect (see Figure 7b), despite the fact that lacZ-infected cells showed no signs of cytopathology. Bcl-2 inhibits baculovirus-induced cell death in this

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4676

Figure 5 Lack of caspase activation or requirement. (a) Hoechst nuclear stain. HepG2 cells seeded on cover slips were infected with 100 m.o.i. Ad-LacZ or Ad-Bin1 and 48 h later ®xed and stained with Hoechst dye, mounted, and examined by immuno¯uorescence microscopy. A separate culture was treated with 0.5 mM staurosporine as a positive control for apoptotic morphology. Cells were photographed at 4006 magni®cation. (b) Nucleosomal DNA degradation. HepG2 cells were infected with viruses or treated with staurosporine as above. Hirt DNA (left panel) or total genomic DNA (right panel) was prepared and analysed by agarose gel electrophoresis. (c) Caspase-3 assay. HepG2 cells were infected with m.o.i. 100 Ad-LacZ or Ad-Bin1 or treated with 0.5 mM staurosporine and harvested 36 h later. Extracts were prepared and assayed for the presence of caspase-3-like activity using the colormetric substrate Ac-DEVD-pNA. Reactions were monitored for production of cleavage product at 405 nm at the times indicated. Recombinant caspase-3 was used as a positive control. (d) Caspase inhibition does not block Bin1-induced cell death. HepG2 cells were infected with 100 m.o.i. Ad-Bin1 or treated with 0.5 mM staurosporine in the presence or absence of the caspase inhibitor ZVAD.fmk at the concentration indicated. Cells were harvested 36 h later and processed for ¯ow cytometry. The graph shows the relative proportion of the cell population undergoing cell death, using the appearance of sub-G1 phase DNA as an indicator

system (Alnemri et al., 1992). In support of these cell transformation beyond inactivation of these tumor results, we also obtained a set of negative results from suppressors (Conzen and Cole, 1995; Hahn et al., 1999; experiments aimed at determining whether Bin1 Powell et al., 1999). The e€ects of T antigen were caused cytochrome c release or altered mitochondrial examined using WI-38 diploid ®broblasts and a WI-38 membrane potential (data not shown). In summary, derivative that expresses T antigen (WI-38/T cells). we concluded that Bin1 acted via non-classical Like HepG2, WI-38 cells are functionally null for Bin1, mechanisms that were independent of the Bcl-2 and in this case due to a missplicing event which causes Fas pathways, two chief regulators of caspases in loss-of-function identical to that which occurs in Bin1 cells. in melanoma (Ge et al., 1999). Western analysis con®rmed RT ± PCR results (Wechsler-Reya et al., 1997b) establishing missplicing of brain-speci®c exon SV40 T antigen inhibits induction of cell death by Bin1 12A in WI-38 cells and also in the WI-38/T derivative. In previous work, we found that Bin1 inhibited Ras co- The 12A isoform exhibited reduced mobility on SDS transformation of primary rodent ®broblasts by c-Myc gels relative to wild-type Bin1 (see Figure 8a, top and adenovirus E1A but not by SV40 large T antigen panel) and was also recognized by a monoclonal (Elliott et al., 1999; Sakamuro et al., 1996). Since Bin1 antibody that is speci®c for exon 12A-encoded residues activated a PCD process in malignantly transformed (Ge et al., 1999) (see Figure 8a, bottom panel). human cells we hypothesized that T antigen might T antigen blocked susceptibility to cell death by suppress these e€ects. This hypothesis was not Bin1 (see Figure 8b,c). Loss of viability occurred only invalidated by the ®nding that p53 and Rb were if Ad-MABin1 was induced by coinfection with Ad- dispensable for Bin1-induced death, because T antigen cre virus. Death correlated with the appearance of has additional role(s) in immortalization and human rounded, detached cells similar to the cytopathic seen

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4677

Figure 6 Death phenotype. HepG2 cells were infected with m.o.i. 200 Ad-LacZ or Ad-Bin1 or treated with 0.5 mM staurosporine as above, stained with osmium tetroxide, and processed for electron microscopy using standard methods. Where indicated, cells were treated with 100 mM ZVAD.fmk. The bars under each panel represent 2 mm

Figure 7 PCD by Bin1 is not inhibited by Bcl-2 or Fas pathway inhibition. (a) Bcl-2 or Fas pathway inhibition does not block Bin1-induced cell death. HepG2 cells overexpressing Bcl-2 or a dominant inhibitory mutant of the Fas-interacting adaptor protein FADD (Muzio et al., 1996) were infected with 100 m.o.i. Ad-vect or Ad-MABin1 plus 100 m.o.i. Ad-cre and viability was measured by trypan blue exclusion 48 h later. Inset: Western analysis showing expression of Bcl-2 and FADD dominant negative (FADD DN) proteins relative to vector controls for each expression construct. (b) Increased kinetics of insect cell death elicited by recombinant Bin1 baculovirus are una€ected by Bcl-2. Sf9 cells were infected as described (Alnemri et al., 1992; Elliott et al., 1999b) and the proportion of viable cells in the culture were determined at various times post-infection. Inset: Western analysis of cell extracts processed at 24 h after infection demonstrating expression of Bin1 and Bcl-2 in HepG2 and other malignant cell lines (Ge et al., concluded that SV40 T antigen suppressed PCD by 1999, 2000a). This response con®rmed that the Bin1. missplicing and inactivation of Bin1 in WI-38 was functionally meaningful. Similar results were obtained A serine protease implicated in PCD by c-Myc inhibits using derivatives of HepG2 that were engineered to DNA degradation by Bin1 express T antigen gave (data not shown). To con®rm that endogenous misspliced isoform in WI-38 was The serine protease inhibitor AEBSF (4-(2-aminoethyl) truly a loss-of-function alteration, we infected cells benzenesulfonyl ¯uoride) has been reported to inhibit with an adenoviral vector expressing this isoform (Ad- c-Myc-induced death in Rat1 ®broblasts (Kagaya et MABin1-10+12A) (Ge et al., 1999). No loss of al., 1997). This report also indicated that AEBSF does viability or cell detachment was observed. We not block apoptosis induced by Fas activation, the

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4678

Figure 8 SV40 T antigen blocks cell death induction by Bin1. (a) Western analysis demonstrates missplicing of exon 12A in Bin1 in WI-38 diploid ®broblasts. Previous results from RT ± PCR experiments demonstrated the presence of exon 12A in Bin1 messages expressed in WI-38 cells (Wechsler-Reya et al., 1997b). Top panel: extracts prepared from C2C12 myoblasts (positive control), WI- 38 cells, and WI-38 cells transformed by SV40 T antigen (WI-38/T cells; also known as VA13 cells) were analysed with anti-Bin1 mAb 99D (Wechsler-Reya et al., 1997a). The arrow indicates a slower mobility band in WI-38 and WI-38/T consistent with the presence of aberrantly spliced isoform. Bottom panel: extracts prepared from the human melanoma cell line WM793 (positive control), WI-38 cells, and WI-38/T cells were analysed with anti-12A mAb (Ge et al., 1999). The arrow indicates a polypeptide including exon 12-derived residues in WI-38 and WI-38/T that has identical mobility to the aberrant Bin1-10+12A splice isoform expressed in human melanoma (Ge et al., 1999). (b) WI-38 cells are susceptible to Bin1 PCD, which is suppressed by SV40 T antigen. Cells were infected with 100 m.o.i. adenoviral vector indicated plus 100 m.o.i. Ad-cre and viable cells were determined by trypan blue exclusion 48 h later. (c) Cytopathic e€ect of Bin1 in WI-38 cells and its inhibition by SV40 T antigen. Cells were processed as above and photographed 48 h after viral infection

cytotoxic T cell granule protein granzyme B, or a inactive analog AEBSA (4-(2-aminoethyl)benzenesulfo- variety of cytotoxic drugs, nor did other kinds of serine namide). Flow cytometery was used to monitor the protease inhibitors a€ect death by c-Myc. Thus, appearance of sub-G1 phase DNA as a measurement AEBSF was a relatively speci®c inhibitor of a feature of PCD. In contrast to ZVAD.fmk, AEBSF suppressed of the death process(es) induced by c-Myc. We Bin1-induced DNA degradation to the background reasoned that if there was any overlap in the levels seen in cells infected with Ad-LacZ (see Figure mechanisms used by Bin1 and c-Myc, then AEBSF 9). This suppression re¯ected inhibition of a serine might a€ect the death process activated by Bin1. To protease(s), rather than some nonspeci®c e€ect, test this notion, HepG2 cells were infected with Ad- because the inactive compound AEBSA was ine€ective. LacZ or Ad-Bin1, or treated with staurosporine, and Consistent with previous ®ndings that AEBSF is not a then left untreated or treated with AEBSF or its general inhibitor of apoptosis (Kagaya et al., 1997),

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4679 dent processes may be important because they are evolutionarily ancient: `classic' apoptotic regulators such as CED-4, Bak, and Bax elicit death in yeast, with vacuolation phenotypes similar to those seen in metazoan cells, despite the fact that yeast lacks caspases (Ink et al., 1997; James et al., 1997; Jurgensmeier et al., 1997; Xiang et al., 1996). Such programs may be integrated yet distinct from apopto- sis, which apparently evolved later in metazoan cells.

Independence from mitochondrial processes and caspases We found that PCD by Bin1 was not subject to suppression by Bcl-2 or to inhibition of the Fas Figure 9 Implication of a serine protease(s) in programmed cell pathway and that cellular demise was not associated death by Bin1. HepG2 cells were infected with 100 m.o.i. Ad-Bin1 or Ad-LacZ or treated with 0.5 mM staurosporine and left with mitochondrial alterations. These observations untreated or treated additionally with 0.4 mM AEBSF or its were internally consistent given that the death process inactive analog AEBSA. Cells were processed for ¯ow cytometry was caspase-independent. However, they were unex- 24 h later. The graph presents the relative proportion of the cell pected given that Bin1 interacts with c-Myc, which population exhibiting sub-G1 phase DNA. The mean and standard error of three trials is shown activates classical apoptosis by eliciting cytochrome c release and subsequent caspase activation (Juin et al., 1999). Two important issues impact consideration of this under the same conditions where AEBSF suppressed apparent conundrum. First, it seems likely that c-Myc the action of Bin1 it did not suppress DNA acts in a complex manner to induce death. Cytochrome degradation induced by staurosporine (the e€ects of c release has been shown to be an important which were actually accentuated by AEBSF treatment). component of the process activated by c-Myc, but this Thus, AEBSF speci®cally inhibited DNA degradation event is apparently insucient (Juin et al., 1999). In by Bin1. We concluded Bin1 acted in part through addition, as noted above, careful investigations have activation of a serine protease, similar to c-Myc, in shown that neither Bcl-2 nor caspase inhibitors block support of the conclusion that Bin1 and c-Myc shared the death commitment decision induced by c-Myc, but certain PCD mechanisms. rather the phenotype and kinetics of the execution step (McCarthy et al., 1997). Interestingly, caspase-indepen- dent suicides that occur retain the cytosolic but not the Discussion nuclear features of classical apoptosis (McCarthy et al., 1997), not unlike the death phenotype induced by Bin1. Bin1 activated a PCD process in malignant cells It is important to draw a distinction between the mechanisms which a€ect death kinetics and those This study de®nes a function for the BAR family which a€ect death commitment decisions, because in adaptor protein Bin1 in a caspase-independent PCD slow growing tumors where c-Myc is overexpressed process that limits neoplastic transformation. Bin1 (such as many carcinomas) the e€ects of Bcl-2 or other function is complex and regulated by alternate splicing. antiapoptotic signals may be insucient to provide Splice isoforms that interact with c-Myc have tumor adequate escape from the death penalty that is suppressor properties and are missing or inactivated in associated with c-Myc overexpression. If this is the breast carcinoma, metastatic prostate cancer, and case, there would be a selection for loss of caspase- malignant melanoma (Ge et al., 1999, 2000a,b; independent processes that impact execution and that Sakamuro et al., 1996). The results of this study are separate from the mitochondrial and death receptor suggest that the antineoplastic e€ects of these Bin1 axes. isoforms in malignant cells are a consequence of their A second issue is the possibility that caspase- ability to activate PCD. Caspase-independent death independent steps are required for apoptotic outcomes processes where cytosolic features predominate without by c-Myc or other proapoptotic stimuli. A precedent accompanying classical nuclear phenotypes have been for this concern is raised by studies of the tumor termed type II apoptosis. The mechanisms underlying suppressor Pml, which activates caspase-independent this type of PCD are poorly understood. Two other death in tumor cells but which is also necessary for cancer suppression genes, Pml and CD47, have been apoptosis induced by a wide variety of death stimuli reported to activate suicide programs with similar (Quignon et al., 1998; Wang et al., 1998). It is features. Pml is a nuclear dot-associated protein that is interesting that the death processes used by c-Myc or disrupted in acute promyelocytic leukemia. Expression Bin1 are each susceptible to inhibition at some level by of wild-type Pml in malignant cells engages a death AEBSF, a serine protease inhibitor which does not program with features similar to that engaged by Bin1 generally a€ect apoptosis (Kagaya et al., 1997). Serine (Quignon et al., 1998). CD47 is a receptor for the death proteases and caspases appear to act in various antiangiogenic extracellular matrix protein thrombos- cells in di€erent but integrated hierarchies (Kagaya et pondin. CD47 ligation engages a death program on al., 1997; Wright et al., 1997), so the fact caspases are chronic lymphocytic leukemia cells (Mateo et al., di€erentially activated by c-Myc and Bin1 supports the 1999). It is unclear whether these programs share any notion of more than one death signal emerging from c- common aspect. Nevertheless, such caspase-indepen- Myc. Results from our laboratory using dominant

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4680 inhibitory and antisense approaches support the has similar properties (unpublished observations). In possibility that Bin1 may be necessary for apoptosis contrast, isoforms that include brain exon 12A did not by c-Myc (unpublished observations). However, due to induce death, consistent with previous results (Ge et the unusual caspase-independent nature of the Bin1 al., 1999). Thus, we believe that the death induction death phenotype in malignant cells, we are addressing function of Bin1 is a unique feature of isoforms that this question further using cells targeted for Bin1 gene can localize to the nucleus and interact with c-Myc. deletion. In summary, despite the key role of apoptotic Bin1 isoforms form heterodimers in the brain (Wigge et escape in malignant development, it is notable that al., 1997). Therefore, it is tempting to speculate that caspases and other apoptosome components may be the endocytosis connection in neurons re¯ects a inactivated in cancer cells less frequently than caspase- specialized link in those cells between survival and independent functions such as Pml or Bin1 (Ge et al., the achievement of a synaptically active state, which 1999, 2000a,b; Mu et al., 1994; Sakamuro et al., 1996; would be associated with neurotransmitter release and Zhang et al., 2000). Our ®ndings suggest that Bin1 hence endocytotic membrane tracking. participates in some caspase-independent process which From a functional standpoint, although ubiquitous can in¯uence cell death commitment in malignant cells. Bin1 isoforms are not amphiphysin-like, the high conservation between amphiphysin and the ubiquitous Bin1 isoforms nonetheless suggest a functional connec- Susceptibility to inhibition by SV40 T antigen tion to membrane dynamics at some level (perhaps We found that the death process activated by Bin1 was related to internal vesicle dynamics). The BAR domain susceptible to suppression by SV40 large T antigen. comprises the major part of this conservation. This The ability of this tumor virus protein to block the domain is crucial for death activity but what it does is death process supported the notion that its inactivation currently obscure. A membrane connection is intri- is relevant to cancer development. This result was also guing in light of emerging interest in possible consistent with earlier ®ndings that transformation of connections between autophagy and apoptosis rodent ®broblasts by T antigen is not subject to (Thompson and Vander Heiden, 1999). However, we suppression by Bin1 (Sakamuro et al., 1996). This have not observed any death inhibitory e€ects of 3- observation does not con¯ict with the ®nding that Bin1 methyladenine, a classical inhibitor of autophagy acts independently of p53 and Rb, because there is (unpublished observations). Nevertheless, there is evidence that T antigen acts beyond inactivating p53 certainly a precedent for a nucleocytosolic adaptor and Rb in immortalization and human cell transforma- protein that has dual functions in transcriptional tion (Conzen and Cole, 1995; Hahn et al., 1999; Powell regulation and membrane dynamics. CtBP/BARS is et al., 1999). For example, T antigen cooperates with an adaptor protein initially identi®ed through its ability activated Ras and the telomerase catalytic subunit to interact with and inhibit the tumorigenic activity of TERT to cause malignant transformation of human adenovirus E1A (Schaeper et al., 1995). CtBP/BARS cells, and one can not complement the e€ects of T binds to cell cycle and transcriptional regulatory antigen by coexpression of the human papilloma virus complexes (Meloni et al., 1999; Sewalt et al., 1999; E6 and E7 proteins (Hahn et al., 1999), which Sollerbrant et al., 1996) but it also controls membrane inactivate p53 and Rb, respectively. In this context, it vesiculation in the Golgi complex (Weigert et al., is interesting to note that before the discovery of 1999). Parallels between CtBP and Bin1 are intriguing TERT the rare successes achieved in immortalizing given their connections to E1A and c-Myc, which are human cells were achieved usually with T antigen or biologically distant cousins. A complex nature for SV40 DNA. Interestingly, Bin1 is inactivated by Bin1, like that exhibited by CtBP/BARS, would not be missplicing in diploid WI-38 cells in the same manner out of step with the complex and still largely obscure as in malignant melanoma (Ge et al., 1999) and this nature of c-Myc, which has a highly integrated cell event makes them susceptible to killing by Bin1, unlike regulatory function. IMR diploid ®broblasts. Thus, Bin1 activation may be We did not detect any changes in the expression of relevant to WI-38 biology, perhaps a€ecting some several target genes linked to apoptosis by c-Myc, process also targeted by T antigen. including ornithine decarboxylase, CDC25A, or Fas ligand (Galaktionov et al., 1996; Hueber et al., 1997; Packham and Cleveland, 1994), but no target gene Bin1 function identi®ed to date has been assigned an unambiguous Bin1 has two functions that are varied by alternate role in apoptosis (Dang, 1999; Evan and Littlewood, splicing. The most signi®cant alterations in Bin1 occur 1998). In transient assays, Bin1 can suppress c-Myc in brain. Brain isoforms localize to clathrin-coated transactivation and repress transcription when tethered vesicles and promote synaptic vesicle endocytosis, like to DNA (Elliott et al., 1999). It is unclear whether amphiphysin, by recruiting enzymes that modify lipids these activities are physiologically germane; if they are and alter membrane structure (Wigge and McMahon, epiphenomenon of Myc-Bin1 interaction transient 1998). In contrast, the ubiquitous Bin1 isoforms lack assays, then Bin1 may have a signaling role that is sequences needed for targeting to clathrin-coated unaliated with transactivation. The latter possibility vesicles (Ramjaun and McPherson, 1998) and instead can be entertained since not all biological actions of c- exhibit nuclear localization (Elliott et al., 1999; Myc can be ascribed strictly to gene regulation (Gusse Wechsler-Reya et al., 1997a, 1998). The c-Myc- et al., 1989; Lemaitre et al., 1995; Li et al., 1994; interacting isoform analysed in this study did not Prendergast and Cole, 1989; Yang et al., 1991) and a€ect endocytosis under conditions where it activated because Bin1 more resembles a signaling protein than a cell death. Preliminary examination of Bin1-10, a transcription factor. Based on existing data, we have closely related c-Myc-interacting isoform, suggest it suggested that PCD by c-Myc involves distinct

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4681 `priming' and `triggering' steps, the former of which is and Graham, 1995) was replaced with the 99fE cDNA. The associated with gene regulation but the latter of which resulting plasmid was cotransfected into 293 cells as above to is not (Prendergast, 1999). Further investigations are obtain the recombinant virus by homologous recombination. required to unravel the physiological relationship In this virus, the Bin1 cDNA is conditionally expressed under between Bin1 and the proapoptotic and transcriptional the control of the cytomegalovirus (CMV) promoter, from which it is separated by an loxP-¯anked stu€er sequence. properties of c-Myc. Bin1 expression is suppressed in the unrearranged virus because the stu€er region includes stop codons in all three Bin1, cell death, and cancer reading frames. The analogous Ad-MA19 virus, which expresses luciferase, was a gift of FL Graham. In cells Cancer is characterized by dysfunctional adhesion and expressing P1 bacteriophage Cre site-speci®c recombinase, the cell survival signaling. Our ®ndings suggest that loss of intervening stu€er region is removed by Cre-mediated Bin1 may eliminate one mechanism which can limit the excision, leading to expression of the transgene. Ad-vect is consequences of inappropriate activation of c-Myc or a control adenovirus kindly provided by JM Wilson which is other oncogenes. Suicide mechanisms are progressively similar in structure to the others except that it contains no eliminated during neoplastic progression (Williams, transgene (Davis and Wilson, 1996). 1991), and in invasive breast cancers and metastatic prostate cancers where Bin1 losses occur frequently Cell culture (Ge et al., 2000a,b) there is strong evidence that loss of HepG2, SAOS-2, and 293 cell lines were maintained in cell suicide capacity corresponds with malignant Dulbecco's modi®ed Eagle's media (DMEM) supplemented conversion (Kyprianou et al., 1990, 1991; McDonnell with 10% fetal bovine serum (Gibco) and 50 U/ml penicillin et al., 1992). However, malignant conversion is and streptomycin (Fisher). For adenovirus infections, HepG2 associated generally with altered adhesive capabilities or SAOS-2 cells were plated at 56105 cells per 6 cm culture that facilitate invasion and metastasis. Resistance to dish and allowed to recover overnight. Virus was added to anoikis (adhesion deprival-induced cell suicide) is likely the cells at the indicated m.o.i. in a volume of 1 ml of to be a crucial feature of the pathophysiology of DMEM supplemented with 2% fetal bovine serum and epithelial malignancy. Interestingly, cell death by many allowed to adhere for 2 ± 3 h at 378C. Cells were then washed oncogenes including c-Myc is suppressed by integrin and fed with DMEM 10% FBS. Where appropriate, the signaling (Crouch et al., 1996), and Bin1 has been indicated concentrations of the broad spectrum caspase inhibitor ZVAD.fmk (Enzyme Systems Products) was added reported to interact with certain a-integrins (Wixler et to cells at the time of infection and maintained until al., 1999). The interaction between Abl kinase and Bin1 harvesting. Thapsigarin and staurosporine (Calbiochem) were in cells (Kadlec and Pendergast, 1997) may be used as chemical inducers of apoptosis at concentrations of 2 consistent with an integrin connection, given evidence and 0.5 mM, respectively, and as positive controls produced that Abl can become activated at focal adhesions where similar results. For cell viability determinations, cells were integrins are localized (Lewis et al., 1996; Taagepera et harvested 48 h postinfection by trypsinization, washed in al., 1998). Therefore, further investigation of the links PBS, stained with trypan blue and counted using a between Bin1-dependent cell death and integrin signal- hemacytometer. Where indicated, the serine protease in- ing may provide insights into the signi®cance of Bin1 hibitor 4-(2-aminoethyl)-benzenesulfonyl ¯uoride (AEBSF) losses in cancer as well as to mechanisms that govern (Sigma) or its inactive analog 4-(2-aminoethyl)benzenesulfo- namide (AEBSA) (Aldrich), which substitutes an amino cell death induction by c-Myc and other oncogenes. group for the crucial ¯uoride moiety in AEBSF. Both inhibitors were added to cells at infection and maintained at a ®nal concentration of 0.4 mM, a concentration which is nontoxic but sucient to suppress c-Myc-mediated apoptosis Materials and methods (Kagaya et al., 1997). For electron microscopy, HepG2 cells were infected with Ad-LacZ or Ad-Bin1 (m.o.i.=100) or Recombinant adenoviruses treated with 0.5 mM staurosporine and where indicated Constitutive and inducible adenoviral vectors were developed treated with 100 mM ZVAD.fmk. Adherent and ¯oating cells by standard methods. The constitutive virus Ad-Bin1 were harvested 36 h after adenoviral infection or 24 ± 30 h contained Bin1 cDNA driven by a cytomegalovirus (CMV) after staurosporine treatment and processed for osmium enhancer-promoter in place of the E1 region of the virus. tetroxide staining and electron microscopy using standard Brie¯y, the full-length Bin1 cDNA 99fE, that includes exons methods. Baculovirus experiments were performed in the 1 ± 11 and 13 ± 16 but not brain-speci®c exons 12A ± 12D insect cell line Sf9 using full-length recombinant viruses (Sakamuro et al., 1996; Wechsler-Reya et al., 1997b), was expressing Bin1, Bcl-2 or no insert, essentially as described subcloned into pAdCMVlink-1. 293 cells were cotransfected (Alnemri et al., 1992; Elliott et al., 1999b). Viable cell counts with this derivative plus ClaI-digested dl7001 adenovirus were determined at the indicated times after virus infection by DNA to obtain the recombinant virus by homologous the trypan blue exclusion method. recombination as described (Davis and Wilson, 1996). Similar viral vector constructions used Bin1-10+12A isoform, an aberrant tumor-speci®c and loss-of-function isoform (Ge et Western analysis al., 1999), or Bin1DBAR-C, a loss-of-function deletion Cell lysates was prepared and analysed by standard protocols mutant lacking a.a. 126 ± 207 (Elliott et al., 1999b). Plaque- (Harlow and Lane, 1988). Brie¯y, lysates were fractionated puri®ed clones were identi®ed by Southern and Western by SDS ± PAGE, gels were electrophoretically transferred to analysis and the DNA sequence of the insert was determined ECL membrane (Amersham) or Immobilon-P (Millipore). to verify wild-type status. A variation of up to several hours Blots were blocked in 3% skim milk and probed with anti- was noted in the kinetics of death induction by di€erent Bin1 monoclonal antibodies 99D, 99I or a12A, or anti-Bcl-2 preparations of Ad-Bin1. The inducible virus Ad-MABin1 antibody #124 (DAKO). 99D and 99I recognize epitopes in was constructed similarly in a cre-loxP adenoviral system the c-Myc binding domain encoded by exon 13 whereas a12A kindly provided by FL Graham (Anton and Graham, 1995). recognizes an epitope encoded by brain-speci®c exon 12A Brie¯y, the luciferase cDNA in the vector pMA19 (Anton (Wechsler-Reya et al., 1997a; Ge et al., 1999). Antibodies

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4682 were diluted 1 : 50 in PBS with 2.5% skim milk and 0.1% Nucleosomal DNA assay Tween-20 and incubated with the membrane 12 h at 48C. Genomic DNA was prepared by a modi®ed Hirt method Blots were washed and incubated 1 h in the same bu€er with (Debbas and White, 1993; Hirt, 1967) or by the following secondary goat anti-mouse antibodies conjugated to horse- method, as indicated. Cells were gently resuspended in 250 ml radish peroxidase (BMB) and developed using a chemilumi- TBE bu€er (90 mM Tris-borate/2 mM EDTA) containing nescence kit using the protocol suggested by the vendor 0.4% NP40 and 5 mg RNAse A and incubated 30 min at (Pierce). 378C. Proteinase K and SDS were added to 100 mg/ml and 1% (v/v), respectively, and samples were incubated overnight Immunofluorescence at 558C. Genomic DNA was extracted with phenol-chloro- form, precipitated with ethanol, and fractionated by agarose Cells were seeded on coverslips and the next day infected gel electrophoresis. with recombinant adenoviruses or treated with drugs. After the periods indicated cells were washed twice with PBS, stained 5 min with 5 mg/ml Hoechst 33323 (Sigma) dissolved Caspase activity assay in PBS, and analysed immediately by immuno¯uorescence. Lysates prepared from equivalent numbers of cells were used For endocytosis assays, cells seeded on coverslips were to assess caspase-3-like activities, employing the Quantizyme infected with recombinant adenoviruses and 20 h later Assay System as recommended by the vendor (Biomol processed for ¯uid-phase uptake or receptor-mediated Research Laboratories). Brie¯y, duplicate plates were endocytosis as described (Barbieri et al., 1998; Benmerah et harvested 48 h post-infection, one for cell counts and the al., 1998). Brie¯y, cells were incubated 30 min in serum-free other for cell lysates. 56107 cells/ml were extracted in the cell DMEM and then treated with 100 n ¯uorescein-conjugated M lysis bu€er provided by the vendor and 10 ml of extract was transferrin (Sigma) (receptor-mediated) or 2 mg/ml horse- used in reactions to monitor cleavage of the substrate Ac- radish peroxidase (Sigma) in 0.2% BSA (¯uid-phase). After DEVD-pNA. Reactions were examined at various times after incubation for 15 min, cells were cooled to 48C, washed 26 addition of the substrate by spectroscopy at 405 n to with ice-cold PBS, and ®xed in 3.7% paraformaldehyde. Cells M monitor release of the product. Data were analysed with were mounted for analysis by confocal ¯uorescence micro- DeltasoftII Software (Biometallics Inc.). scopy or processed with horseradish peroxidase substrate before microscopic analysis.

Flow cytometry Acknowledgments Cells were harvested by trypsinization and ®xed in 1% For providing cDNA clones we thank D Andrews, D paraformaldehyde and then 80% ethanol. TUNEL assay Beach, S Frisch, S Nagata, R Press and D Quelle. Support was performed by incubating samples 1 h at 378C in TdT from the Wistar Flow Cytometry Core Facility and the reaction bu€er (Boehringer Mannheim) and then staining Wistar Cancer Core Grant CA10815 is acknowledged. We 30 min at room temperature with ¯uoresceinated streptavi- thank T Halazonetis, A Oli€, E Routhier and A Muller for din (DuPont). Cells were then incubated 30 min at room critical comments. This work was funded by grants from temperature with propidium iodide. Alternatively, to analyse the US Army Breast Cancer Research Program (DAMD17- only DNA degradation, cells were ®xed in 70% ethanol and 96-1-6324) and the American Cancer Society (CN-160). K permeabilized in PBS/0.2% Tween-20 followed by propi- Elliott was supported by an NIH training grant. K Ge was dium iodide staining as described (Sakamuro et al., 1997). the recipient of a fellowship award from the Adler Flow cytometry was performed on a EPIC/XL cell analyser Foundation. GC Prendergast is a Pew Scholar in the (Coulter). Biomedical Sciences.

References

Alnemri ES, Robertson NM, Fernandes TF, Croce CM and Crouch DH, Fincham VJ and Frame MC. (1996). Oncogene, Litwack G. (1992). Proc. Natl. Acad. Sci. USA, 89, 7295 ± 12, 2689 ± 2696. 7299. Dang CV. (1999). Mol. Cell. Biol., 19, 1 ± 11. Amarante-Mendes GP, Finucane DM, Martin SJ, Cotter David C, McPherson PS, Mundigl O and de Camilli P. TG, Salvesen GS and Green DR. (1998). Cell Death (1996). Proc. Natl. Acad. Sci. USA, 93, 331 ± 335. Di€er., 5, 298 ± 306. Davis AR and Wilson JM. (1996). Current Protocols in Anton M and Graham FL. (1995). J. Virol., 69, 4600 ± 4606. Human Genetics. John Wiley & Sons, Inc.: New York, pp Ashkenazi A and Dixit VM. (1998). Science, 281, 1305 ± 12.4.1 ± 18. 1308. DayML,FosterRG,DayKC,ZhaoX,HumphreyP, Barbieri MA, Kohn AD, Roth RA and Stahl PD. (1998). J. Swanson P, Postigo AA, Zhang SH and Dean DC. (1997). Biol. Chem., 273, 19367 ± 19370. J. Biol. Chem., 272, 8125 ± 8128. Benmerah A, Lamaze C, Begue B, Schmid SL, Dautry- Debbas M and White E. (1993). Genes Dev., 7, 546 ± 554. Varsat A and Cerf-Bensussan N. (1998). J. Cell Biol., 140, Ding HF, McGill G, Rowan S, Schmaltz C, Shimamura A 1055 ± 1062. and Fisher DE. (1998). J. Biol. Chem., 273, 28378 ± 28383. Bowen C, Spiegel S and Gelmann EP. (1998). Cancer Res., Elliott K, Sakamuro D, Basu A, Du W, Wunner W, Staller P, 58, 3275 ± 3281. Gaubatz S, Zhang H, Prochownik E, Eilers M and Butler MH, David C, Ochoa G-C, Freyberg Z, Daniell L, Prendergast GC. (1999). Oncogene, 18, 3564 ± 3573. Grabs D, Cremona O and De Camilli P. (1997). J. Cell Enari M, Sakahira H, Yokoyama H, Okawa K, Iwamatsu A Biol., 137, 1355 ± 1367. and Nagata S. (1998). Nature, 391, 45 ± 50. Cecconi F, Alvarez-Bolado G, Meyer BI, Roth KA and Evan G and Littlewood T. (1998). Science, 281, 1317 ± 1322. Gruss P. (1998). Cell, 94, 727 ± 737. Evan GI, Wyllie AH, Gilbert CS, Littlewood TD, Land H, Chao DT and Korsmeyer SJ. (1998). Ann. Rev. Immunol., 16, Brooks M, Waters CM, Penn LZ and Hancock DC. 395 ± 419. (1992). Cell, 69, 119 ± 128. Cole MD. (1986). Ann. Rev. Genet., 20, 361 ± 384. Fearnhead HO, McCurrach ME, O'Neill J, Zhang K, Lowe Conzen SD and Cole CN. (1995). Oncogene, 11, 2295 ± 2302. SW and Lazebnik YA. (1997). Genes Dev., 11, 1266 ± 1276.

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4683 Galaktionov K, Chen X and Beach D. (1996). Nature, 382, McDonnell TJ, Troncoso P, Brisbay SM, Logothetis C, 511 ± 517. Chung LW, Hsieh JT, Tu SM and Campbell ML. (1992). Garte SJ. (1993). Crit. Rev. Oncog., 4, 435 ± 449. Cancer Res., 52, 6940 ± 6944. Ge K, DuHadaway J, Du W, Herlyn M, Rodeck U and Meloni AR, Smith EJ and Nevins JR. (1999). Proc. Natl. Prendergast GC. (1999). Proc. Natl. Acad. Sci. USA, 96, Acad.Sci.USA,96, 9574 ± 9579. 9689 ± 9694. Mu ZM, Chin KV, Liu JH, Lozano G and Chang KS. (1994). Ge K, DuHadaway J, Sakamuro D, Wechsler-Reya R, Mol. Cell. Biol., 14, 6858 ± 6867. Reynolds C and Prendergast GC. (2000a). Int. J. Cancer, Muzio M, Chinnaiyan AM, Kischkel FC, O'Rourke K, 85, 376 ± 383. Shevchenko A, Ni J, Scadi C, Bretz JD, Zhang M, Gentz Ge K, Minhas F, DuHadaway J, Mao N-C, Wilson D, R, Mann M, Krammer PH, Peter ME and Dixit VM. Sakamuro D, Buccafusca R, Nelson P, Malkowicz SB, (1996). Cell, 85, 817 ± 827. Tomaszewski JT and Prendergast GC. (2000b). Int. J. Packham G and Cleveland JL. (1994). Mol. Cell. Biol., 14, Cancer, 86, 155 ± 161. 5741 ± 5747. Green DR and Reed JC. (1998). Science, 281, 1309 ± 1312. Powell AJ, Darmon AJ, Gonos ES, Lam EW, Peden KW and Gusse M, Ghysdael J, Evan G, Soussi T and Mechali M. Jat PS. (1999). Oncogene, 18, 7343 ± 7350. (1989). Mol. Cell. Biol., 9, 5395 ± 5403. Prendergast GC. (1999). Oncogene, 18, 2966 ± 2986. Hahn WC, Counter CM, Lundberg AS, Beijersbergen RL, Prendergast GC and Cole MD. (1989). Mol. Cell. Biol., 9, Brooks MW and Weinberg RA. (1999). Nature, 400, 464 ± 124 ± 134. 468. Quignon F, De Bels F, Koken M, Feunteun J, Ameisen JC Harlow E and Lane D. (1998). Antibodies: A Laboratory and de The H. (1998). Nat. Genet., 20, 259 ± 265. Manual. Cold Spring Harbor Laboratory Press, Cold Ramjaun AR and McPherson PS. (1998). J. Neurochem., 70, Spring Harbor NY. 2369 ± 2376. Hirt B. (1967). J. Mol. Biol., 26, 365 ± 369. Ramjaun AR, Micheva KD, Bouchelet I and McPherson PS. Hueber AO, Zornig M, Lyon D, Suda T, Nagata S and Evan (1997). J. Biol. Chem., 272, 16700 ± 16706. GI. (1997). Science, 278, 1305 ± 1309. Rao L, Perez D and White E. (1996). J. Cell Biol., 135, 1441 ± InkB,ZornigM,BaumB,HajibagheriN,JamesC, 1455. Chittenden T and Evan G. (1997). Mol. Cell. Biol., 17, Reed JC, Jurgensmeier JM and Matsuyama S. (1998). 2468 ± 2474. Biochim. Biophys. Acta, 1366, 127 ± 137. James C, Gschmeissner S, Fraser A and Evan GI. (1997). Sakamuro D and Prendergast GC. (1999). Oncogene, 18, Curr. Biol., 7, 246 ± 252. 2942 ± 2953. Jenkins RB, Qian J, Lieber MM and Bostwick DG. (1997). Sakamuro D, Elliott K, Wechsler-Reya R and Prendergast Cancer Res., 57, 524 ± 531. GC. (1996). Nature Genet., 14, 69 ± 77. Juin P, Hueber AO, Littlewood T and Evan G. (1999). Genes Sakamuro D, Eviner V, Elliott K, Showe L, White E and Dev., 13, 1367 ± 1381. Prendergast GC. (1995). Oncogene, 11, 2411 ± 2418. Jurgensmeier JM, Krajewski S, Armstrong RC, Wilson GM, Sakamuro D, Sabbatini P, White E and Prendergast GC. Oltersdorf T, Fritz LC, Reed JC and Ottilie S. (1997). Mol. (1997). Oncogene, 15, 887 ± 898. Cell. Biol., 18, 325 ± 339. Schaeper U, Boyd JM, Verma S, Uhlmann E, Subramanian Kadlec L and Pendergast A-M. (1997). Proc. Natl. Acad. Sci. T and Chinnadurai G. (1995). Proc. Natl. Acad. Sci. USA, USA, 94, 12390 ± 12395. 92, 10467 ± 10471. Kagaya S, Kitanaka C, Noguchi K, Mochizuki T, Sugiyama Sewalt RG, Gunster MJ, van der Vlag J, Satijn DP and Otte A, Asai A, Yasuhara N, Eguchi Y, Tsujimoto Y and AP. (1999). Mol. Cell. Biol., 19, 777 ± 787. Kuchino Y. (1997). Mol. Cell. Biol., 17, 6736 ± 6745. Sherr CJ. (1998). Genes Dev., 12, 2984 ± 2991. Kangas A, Nicholson DW and Hottla E. (1998). Oncogene, Soengas MS, Alarcon RM, Yoshida H, Giaccia AJ, Hakem 16, 387 ± 398. R, Mak TW and Lowe SW. (1999). Science, 284, 156 ± 159. Kyprianou N, English HF, Davidson NE and Isaacs JT. Sollerbrant K, Chinnadurai G and Svensson C. (1996). Nuc. (1991). Cancer Res., 51, 162 ± 166. Acids Res., 24, 2578 ± 2584. Kyprianou N, English HF and Isaacs JT. (1990). Cancer Suzuki A, Tsutomi Y, Akahane K, Araki T and Miura M. Res., 50, 3748 ± 3753. (1998). Oncogene, 17, 931 ± 939. Lazebnik YA, Cole S, Cooke CA, Nelson WG and Earnshaw Taagepera S, McDonald D, Loeb JE, Whitaker LL, McElroy WC. (1993). J. Cell Biol., 123, 7 ± 22. AK, Wang JY and Hope TJ. (1998). Proc. Natl. Acad. Sci. Lazebnik YA, Takahashi A, Moir RD, Goldman RD, USA, 95, 7457 ± 7462. Poirier GG, Kaufmann SH and Earnshaw WC. (1995). Thompson CB and Vander Heiden MG. (1999). Nat. Cell Proc. Natl. Acad. Sci. USA, 92, 9042 ± 9046. Biol., 1, E209 ± E216. Lemaitre JM, Bocquet S, Buckle R and Mechali M. (1995). Trudel M, Lanoix J, Barisoni L, Blouin MJ, Desforges M, Mol. Cell Biol., 15, 5054 ± 5062. L'Italien C and D'Agati V. (1997). J. Exp. Med., 186, Lewis JM, Baskaran R, Taagepera S, Schwartz MA and 1873 ± 1884. Wang JY. (1996). Proc. Natl. Acad. Sci. USA, 93, 15174 ± Tsuneoka M and Mekada E. (2000). Oncogene, 19, 115 ± 123. 15179. TsutsuiK,MaedaY,TsutsuiK,SekiSandTokunagaA. Li L, Nerlov C, Prendergast G, MacGregor D and Zi€ EB. (1997). Biochem. Biophys. Res. Comm., 236, 178 ± 183. (1994). EMBO J., 13, 4070 ± 4079. Wang ZG, Ruggero D, Ronchetti S, Zhong S, Gaboli M, Liu X, Zou H, Slaughter C and Wang X. (1997). Cell, 89, Rivi R and Pandol® PP. (1998). Nat. Genet., 20, 266 ± 272. 175 ± 184. Wechsler-Reya R, Elliott K, Herlyn M and Prendergast GC. Mao NC, Steingrimsson EJD, Ruiz J, Wasserman W, (1997a). Cancer Res., 57, 3258 ± 3263. Copeland NG, Jenkins NA and Prendergast GC. (1999). Wechsler-Reya R, Elliott K and Prendergast GC. (1998). Genomics, 56, 51 ± 58. Mol. Cell. Biol., 18, 566 ± 575. Mateo V, Lagneaux L, Bron D, Biron G, Armant M, Wechsler-Reya R, Sakamuro D, Zhang J, Duhadaway J and Delespesse G and Sarfati M. (1999). Nat. Med., 5, 1277 ± Prendergast GC. (1997b). J. Biol. Chem., 272, 31453 ± 1284. 31458. McCarthy NJ, Whyte MKB, Gilberg CS and Evan GI. (1997). J. Cell Biol., 136, 215 ± 227.

Oncogene Bin1 induces caspase-independent PCD K Elliott et al 4684 Weigert R, Silletta MG, Spanol S, Turacchio G, Cericola C, Wright SC, Schellenberger U, Wang H, Kinder DH, Talhouk Colanzi A, Senatore S, Mancini R, Polishchuk EV, JW and Larrick JW. (1997). J. Exp. Med., 186, 1107 ± Salmona M, Facchiano F, Burger KN, Mironov A, Luini 1117. A and Corda D. (1999). Nature, 402, 429 ± 433. Xiang J, Chao D and Korsmeyer S. (1996). Proc. Natl. Acad. Wigge P, Kohler K, Vallis Y, Doyle CA, Owen D, Hunt SP Sci. USA, 93, 14559 ± 14563. and McMahon HT. (1997). Mol. Biol. Cell, 8, 2003 ± 2015. Yang B-S, Geddes TJ, Pogulis RJ, de Crombrugghe B and Wigge P and McMahon HT. (1998). Trends Neurosci., 21, Freytag SO. (1991). Mol. Cell. Biol., 11, 2291 ± 2295. 339 ± 344. Yoshida H, Kong YY, Yoshida R, Elia AJ, Hakem A, Williams GT. (1991). Cell, 65, 1097 ± 1098. Hakem R, Penninger JM and Mak TW. (1998). Cell, 94, Wixler V, Laplantine E, Geerts D, Sonnenberg A, Petersohn 739 ± 750. D, Eckes B, Paulsson M and Aumailley M. (1999). FEBS ZhangP,ChinW,ChowLTC,ChanASK,YimAPC,Leung Lett., 445, 351 ± 355. S-F, Mok TSK, Chang K-S, Johnson PJ and Chan JYH. Woo M, Hakem R, Soengas MS, Duncan GS, Shahinian A, (2000). Int. J. Cancer, 85, 599 ± 605. Kagi D, Hakem A, McCurrach M, Khoo W, Kaufman SA, Senaldi G, Howard T, Lowe SW and Mak TW. (1998). Genes Dev., 12, 806 ± 819.

Oncogene