DEATH AND ANTI-DEATH: TUMOUR RESISTANCE TO APOPTOSIS

DEATH AND ANTI-DEATH: TUMOUR RESISTANCE TO APOPTOSIS Frederik H. Igney and Peter H. Krammer Every cell in a multicellular organism has the potential to die by apoptosis, but tumour cells often have faulty apoptotic pathways. These defects not only increase tumour mass, but also render the tumour resistant to therapy. So, what are the molecular mechanisms of tumour resistance to apoptosis and how can we use this knowledge to resensitize tumour cells to cancer therapy? Tumor Immunology Program, German Cancer Research Center (DKFZ), Heidelberg, Germany. Correspondence to P.H.K. e-mail: P.Krammer @dkfz-heidelberg.de DOI: 10.1038/nrc776 The life and death of cells must be balanced if tissue homeostasis is to be maintained too much growth and too little death can lead to a severe disturbance that might, ultimately, result in cancer 1,2. Cells have an intrinsic mechanism of self destruction called programmed cell death or apoptosis. In multicellular organisms, many of the mechanisms that control tissue homeostasis are linked to apoptosis. Defects in the apoptosis-inducing pathways can eventually lead to expansion of a population of neoplastic cells. Resistance to apoptosis can also augment the escape of tumour cells from surveillance by the immune system. Moreover, because chemotherapy and irradiation act primarily by inducing apoptosis, defects in the apoptotic pathway can make cancer cells resistant to therapy. So, resistance to apoptosis constitutes an important clinical problem. How do tumour cells resist apoptosis? Understanding these mechanisms at the molecular level provides deeper insight into carcinogenesis, influences therapeutic strategy and might, ultimately, lead to new therapeutic approaches based on modulation of apoptosis sensitivity. Apoptosis signalling Initiation of apoptosis. In principle, there are two alternative pathways that initiate apoptosis: one is mediated by death receptors on the cell surface sometimes referred to as the extrinsic pathway ; the other is mediated by mitochondria referred to as the instrinsic pathway (FIG. 1; reviewed in REFS 3,4). In both pathways, cysteine aspartyl-specific proteases (caspases) are activated that cleave cellular substrates, and this leads to the biochemical and morphological changes that are characteristic of apoptosis. Death receptors are members of the tumour-necrosis factor (TNF) receptor superfamily and comprise a subfamily that is characterized by an intracellular domain the death domain (FIG. 2, reviewed in REFS 3,4). Decoy receptors are closely related to the death receptors but lack a functional death domain 5,6. Death receptors are activated by their natural ligands, the TNF family. When ligands bind to their respective death receptors such as CD95, TRAIL-R1 (TNF-related apoptosis-inducing ligand-r1) or TRAIL-R2 the death domains attract the intracellular adaptor protein FADD (Fas-associated death domain protein, also known as MORT1), which, in turn, recruits the inactive proforms of certain members of the caspase protease family. The caspases that are recruited to this death-inducing signalling complex (DISC) caspase-8 and caspase-10 function as initiator caspases (FIG. 3). At the DISC, procaspase-8 and procaspase-10 are cleaved and yield active initiator caspases, although the mechanism of initiation for caspase-10 is not yet clear 7 9. In some cells known as type I cells the amount of active caspase-8 formed at the DISC is sufficient to initiate apoptosis directly, but in type II cells, the amount is too small and mitochondria are used as amplifiers of the apoptotic signal (FIG. 1) 10. Activation of mitochondria is mediated by the BCL2 family member BID (BOX 1). BID is cleaved by active caspase-8 and translocates to the mitochondria. Death at the mitochondrial level is initiated by stimulation yet to be fully defined of the mitochondrial membrane, and proceeds via release of cytochrome c and other apoptogenic factors from the intermembraneous space of mitochondria (FIG. 4; BOX 1; reviewed in REFS 11,12). In the cytosol, cytochrome NATURE REVIEWS CANCER VOLUME 2 APRIL 2002 277

Summary Apoptosis is a multi-step, multi-pathway cell-death programme that is inherent in every cell of the body. In cancer, the apoptosis:cell-division ratio is altered, which results in a net gain of malignant tissue. Apoptosis can be initiated either through the death-receptor or the mitochondrial pathway. Caspases that cleave cellular substrates leading to characteristic biochemical and morphological changes are activated in both pathways. The apoptotic process is tightly controlled by various proteins. There are also other caspase-independent types of cell death. Many physiological growth-control mechanisms that govern cell proliferation and tissue homeostasis are linked to apoptosis. Therefore, resistance of tumour cells to apoptosis might be an essential feature of cancer development. Immune cells (T cells and natural killer cells) can kill tumour cells using the granule exocytosis pathway or the death-receptor pathway. Apoptosis resistance of tumour cells might lead to escape from immunosurveillance and might influence the efficacy of immunotherapy. Cancer treatment by chemotherapy and γ-irradiation kills target cells primarily by inducing apoptosis. Therefore, modulation of the key elements of apoptosis signalling directly influences therapy-induced tumour-cell death. Tumour cells can acquire resistance to apoptosis by the expression of anti-apoptotic proteins or by the downregulation or mutation of pro-apoptotic proteins. Alterations of the p53 pathway also influence the sensitivity of tumour cells to apoptosis. Moreover, most tumours are independent of survival signals because they have upregulated the phosphatidylinositol 3-kinase (PI3K)/AKT pathway. The present aim of research in this area is to understand further the molecular mechanisms of tumour resistance and sensitivity, and to use this insight to resensitize tumour cells to apoptosis and, accordingly, to tumour therapy. activation 15 17. These pathways share some, but not all, the characteristics of apoptotic classical pathways. Therefore, they cannot be readily classified as apoptosis or necrosis and have been called necrotic-like or apoptotic-like cell death or paraptosis. Regulation of apoptosis. The apoptotic self-destruction machinery is tightly controlled. Various proteins regulate the apoptotic process at different levels (FIGS 3 and 4). FLIPs (FADD-like interleukin-1 β-converting enzyme-like protease (FLICE/caspase-8)-inhibitory proteins) interfere with the initiation of apoptosis directly at the level of death receptors 18. Two splice variants a long form (FLIP L ) and a short form (FLIP S ) have been identified in human cells. Both forms share structural homology with procaspase-8, but lack its catalytic site. This structure allows them Death-receptor pathway Death ligand Death receptor Mitochondrial pathway Apoptotic stimulus c forms a complex with APAF1 (apoptotic protease activating factor-1), ATP and the inactive initiator caspase procaspase-9. Within this complex known as the apoptosome caspase-9 is activated (FIG. 4). Cleaved BID LAMINS A group of intermediatefilament proteins that form the fibrous network (nuclear lamina) on the inner surface of the nuclear envelope. Execution of apoptosis. Once the initiator caspases are activated, they cleave and activate executioner caspases, mainly caspase-3, caspase-6 and caspase-7.the active executioner caspases then cleave each other and, in this way, an amplifying proteolytic cascade of caspase activation is started 13. Eventually, the active executioner caspases cleave cellular substrates the death substrates which leads to characteristic biochemical and morphological changes 13. Cleavage of nuclear LAMINS is involved in chromatin condensation and nuclear shrinkage. Cleavage of the inhibitor of the DNase CAD (caspaseactivated deoxyribonuclease, DFF40), ICAD (also known as DNA fragmentation factor, 45 kda; DFF45), causes the release of the endonuclease, which travels to the nucleus to fragment DNA. Cleavage of cytoskeletal proteins such as actin, plectin, Rho kinase 1 (ROCK1) and gelsolin leads to cell fragmentation, blebbing and the formation of apoptotic bodies. After exposure of eat me signals (for example, exposure of phosphatidylserine and changes in surface sugars), the remains of the dying cell are engulfed by phagocytes 14. Besides these prototypic caspase-dependent apoptosis pathways, there are also molecularly less-well-defined cell-death pathways that do not require caspase Caspase-8 (Caspase-10) BID Executioner caspases Death substrates Apoptosome Caspase-9 Apoptosis Figure 1 The two main apoptotic signalling pathways. Apoptosis can be initiated by two alternative pathways: either through death receptors on the cell surface (extrinsic pathway) or through mitochondria (intrinsic pathway). In both pathways, induction of apoptosis leads to activation of an initiator caspase: caspase-8 and possibly caspase-10 for the extrinsic pathway; and caspase-9, which is activated at the apoptosome, for the intrinsic pathway. The initiator caspases then activate executioner caspases. Active executioner caspases cleave the death substrates, which eventually results in apoptosis. There is crosstalk between these two pathways. For example, cleavage of the BCL2-family member BID by caspase-8 activates the mitochondrial pathway after apoptosis induction through death receptors, and can be used to amplify the apoptotic signal. 278 APRIL 2002 VOLUME 2 www.nature.com/reviews/cancer

OPGL RANKL TRANCE TRAIL APO-2L TWEAK APO-3L CD95L TNF-α LTα/ LTβ LIGHT?? RANK OPG DcR1 TRAIL-R3 LIT TRID TRAIL-R4 DcR2 TRUNDD TRAIL-R1 DR4 APO-2 TRAIL-R2 DR5 KILLER TRICK2 DR6 DR3 APO-3 WSL TRAMP LARD APO-1 FAS CD95 TNF-R1 TNF-R2 TR6 DcR3 Figure 2 Death receptors and ligands. Ligands are shown at the top, receptors at the bottom. Death receptors and death ligands are grouped in a box. DcR3 (decoy receptor 3) acts as a decoy receptor for CD95L (dotted line). The other molecules outside the box can bind to death receptors or ligands as indicated, but have not been shown to transmit an apoptotic signal. The death domain is shown as a pink box. The question marks indicate the unknown ligand for DR6 (death receptor 6), and that the interaction between TWEAK (TNF-related protein with weak ability to induce cell death) and DR3 (death receptor 3) is not fully established. APO-2L, APO-2 ligand; DcR1, decoy receptor 1; DR4, death receptor 4; LARD, lymphocyte-associated receptor of death; LIGHT, a cytokine that is homologous to lymphotoxins, exhibits inducible expression and competes with herpes simplex virus glycoprotein D for herpesvirus entry mediator, a receptor expressed by T cells; LIT, lymphocyte inhibitor of TRAIL; LTα,/LTβ, lymphotoxin α/ lymphotoxin β; OPG, osteoprotegerin; OPGL, osteoprotegerin ligand; RANK, receptor activator of NF-κB; RANKL, receptor activator of NF-κB ligand; TNF-α, tumour-necrosis factor-α; TNFR1, tumour-necrosis factor receptor 1; TNFR2, tumournecrosis factor receptor 2; TRAIL, TNF-related apoptosis-inducing ligand; TRANCE, tumour-necrosis-factor-related activationinduced cytokine; TRICK2, TRAIL receptor inducer of cell killing 2; TRID, TRAIL receptor without an intracellular domain; TRUNDD, TRAIL receptor with a truncated death domain. to bind to the DISC, thereby inhibiting the processing and activation of the initiator caspase-8. The members of the BCL2 family, which regulate apoptosis at the mitochondrial level, are an important class of regulatory proteins (FIG. 4; BOX 1; reviewed in REFS 11,12). They can be divided into anti-apoptotic and proapoptotic proteins according to their function (BOX 1). BCL2 family proteins influence the permeability of the mitochondrial membrane. The IAPs (inhibitor of apoptosis proteins; reviewed in REF. 19) constitute a third class of regulatory proteins. IAPs bind to and inhibit caspases. They might also function as ubiquitin ligases, promoting the degradation of the caspases that they bind. IAPs are characterized by a domain termed the baculoviral IAP repeat (BIR). Nine IAP family members including XIAP (hilp, MIHA, ILP-1), ciap1 (MIHB, HIAP-2), ciap2 (HIAP-1, MIHC, API2), NAIP, ML- IAP, ILP2, livin (KIAP), apollon and survivin have been identified in human cells. However, not all BIRcontaining proteins have been shown to suppress apoptosis, and some of them might also have functions other than caspase inhibition. IAPs are inhibited by a protein named SMAC/DIABLO (second mitochondria-derived activator of caspase/direct IAP binding protein with low pi) 20,21, which is released from mitochondria along with cytochrome c during apoptosis and promotes caspase activation by binding to, and inhibiting, IAPs. Apoptosis induction in tumour cells Physiological growth control and apoptosis. In cells and tissues of multicellular organisms, potent physiological mechanisms govern cell proliferation and homeostasis. Many of these growth-control mechanisms are linked to apoptosis: excessive proliferation or growth at inappropriate sites induces apoptosis in the affected cells. Tumours can proliferate beyond these constraints, which limit growth in normal tissue. Therefore, resistance of tumour cells to apoptosis is an essential feature of cancer development. This assumption is confirmed by the finding that deregulated proliferation alone is not sufficient for tumour formation, but leads to cell death (reviewed in REF. 22): overexpression of growth-promoting oncogenes such as c-myc, E1A or E2F1 sensitizes cells to apoptosis 23. Besides the expression of proteins that promote cell proliferation, tumour progression NATURE REVIEWS CANCER VOLUME 2 APRIL 2002 279

DISC INTEGRINS A large family of heterodimeric transmembrane proteins that promote adhesion of cells to the extracellular matrix or to other cells. ADAPTIVE IMMUNE SYSTEM Adaptive immunity also known as specific or acquired immunity is mediated by antigen-specific lymphocytes and antibodies; it is highly antigen specific and includes the development of immunological memory. INNATE IMMUNE SYSTEM The innate immune system includes phagocytes, natural killer cells, the complement system and other non-specific components. It protects against infections using mechanisms that exist before infection, providing a rapid response to microbes that is essentially the same regardless of the type of infection. CD95L FADD Procaspase-8 (Procaspase-10) Caspase-8 (Caspase-10) Caspase-3, -6, -7 (executioner caspases) Death substrates Apoptosis CD95 FLIP L,S BID scd95, (DcR3) Figure 3 Apoptosis signalling through death receptors. Binding of death ligands (CD95L is used here as an example) to their receptor leads to the formation of the death-inducing signalling complex (DISC). In the DISC, the initiator procaspase-8 is recruited by FADD (FAS-associated death domain protein) and is activated by autocatalytic cleavage. Death-receptor-mediated apoptosis can be inhibited at several levels by anti-apoptotic proteins: CD95L can be prevented from binding to CD95 by soluble decoy receptors, such as soluble CD95 (scd95) or DcR3 (decoy receptor 3). FLICE-inhibitory proteins (FLIPs) bind to the DISC and prevent the activation of caspase-8; and inhibitors of apoptosis proteins (IAPs) bind to and inhibit caspases. FLIP L and FLIP S refer to long and short forms of FLIP, respectively. requires the expression of anti-apoptotic proteins or the inactivation of essential pro-apoptotic proteins 24 26. The molecular connections between cell cycle and cell death are not entirely clear, but the p53 pathway seems to be involved (FIG. 5) 26. Proliferative signals induce ARF, the product encoded by an alternative reading frame within the CDKN2A tumoursuppressor gene locus, which also encodes the cyclindependent kinase inhibitor INK4A. ARF interacts with the ubiquitin ligase MDM2, and prevents it from binding p53 and targeting it for destruction in the proteasome. Upregulation of p53 leads to cell-cycle arrest and apoptosis. The relationship between proliferation and cell death might also reflect the fact that cells require survival signals. Lack of these signals triggers apoptosis a phenomenon called death by neglect. Survival signals include growth factors, cytokines, hormones and other stimuli, such as signals given by adhesion molecules. In general, survival signals are mediated by means of the phosphatidylinositol 3-kinase (PI3K)/AKT pathway (BOX 2) 27,28. Depending on the stimulus, further mechanisms must be present that deliver anti-apoptotic survival signals. Anoikis is a special case of death by neglect and is triggered by inadequate or inappropriate cell matrix contacts (reviewed in REF. 29). Binding of INTEGRINS to the extracellular matrix conveys survival signals by activating the PI3K/AKT pathway 30,31. Anoikis involves the proapoptotic BCL2 family proteins BIM and BMF 32,33.In healthy cells, these proteins bind to the cytoskeleton, but after the cell has detached from the extracellular matrix, BIM and BMF are released and interact with the anti-apoptotic protein BCL2. Resistance to anoikis might facilitate metastasis by allowing cells to survive following detachment from the matrix in their tissue of origin and travelling to distant sites. Normal diploid cells have a limited replicative potential, and this is another means by which excessive proliferation is controlled. After progressing through 60 70 divisions, cells cease to proliferate a state called senescence and die. The finite number of divisions is determined by the length of the telomeres at the chromosome ends, which shorten during each cell cycle (reviewed in REF. 34). Once a critically short length is reached, the sensors for DNA damage are triggered and induce cell-cycle arrest or apoptosis 35,36. Again, p53 seems to be important for this response to telomere erosion but, although p53 deficiency temporarily rescues cells from apoptosis, telomere loss ultimately results in a genetic catastrophy, triggering p53- independent apoptosis 37. In tumour cells, telomeres are stabilized by expression of telomerase or a poorly characterized mechanism that is known as alternate lengthening of telomeres (ALT). Apoptosis induction by the immune system. If cells manage to circumvent the built-in constraints to unlimited proliferation, the organism has to rely on the immune system as a watch-dog against tumour initiation a concept called immunosurveillance. The main effector cells against tumours are cytotoxic T cells of the ADAPTIVE IMMUNE SYSTEM and natural killer (NK) cells of the INNATE IMMUNE SYSTEM. T cells and NK cells use two main mechanisms to kill tumour cells: the granule exocytosis pathway and the CD95L pathway (reviewed in REF. 38). In the calcium-dependent granule exocytosis pathway, lymphocytes secrete a membrane permeability protein called perforin and proteolytic enzymes known as granzymes from cytotoxic granules towards the target cell. In the presence of calcium, perforin polymerizes and initiates ill-defined changes in the target-cell membrane that allow granzymes to pass into the cell 39. Granzymes are neutral serine proteases that can activate caspases in the target cell 40,41. In addition, granzyme B might directly cleave the BCL2 family member BID to activate the mitochondrial death pathway 42. In the CD95L pathway, which is calcium independent, the lymphocyte exhibits the death ligand CD95L on the cell surface and triggers apoptosis through the CD95 receptor on the target cell 43,44. Resistance of tumour cells to these effector mechanisms not only leads to escape of the tumours from immunosurveillance, but might also markedly influence the efficacy of immunotherapy. 280 APRIL 2002 VOLUME 2 www.nature.com/reviews/cancer

Box 1 Mitochondria and the BCL2 family Death initiated at the mitochondrial level is regulated by the members of the BCL2 family (reviewed in REFS 11,12). BCL2 family members can be divided into antiapoptotic (BCL2, BCL-X L, BCL-w, MCL1, A1/BFL1, BOO/DIVA, NR-13) and proapoptotic proteins (BAX, BAK, BOK/MTD, BCL-X S,BID,BAD,BIK/NBK, BLK, HRK/DP5, BIM/BOD, NIP3, NIX, NOXA, PUMA, BMF). Most anti-apoptotic members contain the BCL2 homology (BH) domains 1, 2 and 4, whereas the BH3 domain seems to be crucial for apoptosis induction. The pro-apoptotic members can be subdivided into the BAX subfamily (BAX, BAK, BOK) and the BH3-only proteins (for example, BID, BAD and BIM) 178. Although the impact of BCL2 family members on apoptosis is well known, the biochemical mechanism of their function is not entirely clear. After activation by an apoptotic stimulus, mitochondria release cytochrome c, AIF (apoptosis inducing factor) and other apoptogenic factors from the intermembrane space to the cytosol. Concomitantly, the mitochondrial transmembrane potential ( Ψ m ) drops. According to one model, mitochondrial membrane permeabilization involves the permeability transition pore complex (PTPC), a multiprotein complex that consists of the adenine nucleotide translocator (ANT) of the inner membrane, the voltage-dependent anion channel of the outer membrane and various other proteins 11. BCL2 proteins might interact with the PTPC and regulate its permeability. According to another model, BH3-only proteins serve as death sensors in the cytosol or cytoskeleton. Following a death signal, they interact with members of the BAX subfamily. After this interaction, BAX proteins undergo a conformational change, insert into the mitochondrial membrane, oligomerize and form protein-permeable channels. Anti-apoptotic BCL2 proteins inhibit the conformational change or the oligomerization of BAX and BAK 12. The localization of the pro-apoptotic BCL2 family member BAD is regulated by phosphorylation. Only non-phosphorylated BAD is capable of antagonizing antiapoptotic BCL2 or BCL-X L on the mitochondrial membrane. BAD phosphorylation results in its redistribution to the cytosol and its sequestration by 14-3-3 proteins. and in the death receptor pathway such as CD95, TRAIL-R1 and TRAIL-R2 (REFS 53 55).Moreover,transcriptionally independent activities of p53 mediate some of its pro-apoptotic effects, including protein protein interactions, direct effects in the mitochondria and relocalization of death receptors to the cell surface (REFS 56 58). Another stress pathway that is activated in response to chemotherapy is the stress-activated protein kinase (SAPK, also known as JUN-N-terminal kinase or JNK) pathway 45. SAPKs, which are members of the mitogenactivated protein kinase family, can regulate the activity of AP-1 transcription factors. Known pro-apoptotic target genes for AP-1 are CD95L and TNF-α. Moreover, oxidative stress triggered by the production of reactive oxygen intermediates and glutathione depletion can also induce CD95L expression 59. The best-defined mechanism by which therapyinduced cellular stress eventually leads to the death of tumour cells particularly liver tumour cells Survival signal PI3K AKT BAD BID BAX, BIM BCL-2 BCL-X L ANTIMETABOLITES Antimetabolites (for example, methotrexate) block specific metabolic pathways by competitive binding to the substrate-binding site of enzymes that are involved in metabolism. TOPOISOMERASES A class of enzymes that control the number and topology of supercoils in DNA and that are important for DNA replication. SUMOYLATION A post-translational modification that consists of covalent attachment of the small ubiquitin-like molecule, SUMO-1 (also known as sentrin, PIC1). Sumoylation can change the ability of the modified protein to interact with other proteins and can interfere with its proteasomal degradation. Therapeutic induction of apoptosis. Cancer treatment by chemotherapy and γ-irradiation kills target cells primarily by the induction of apoptosis. However, few tumours are sensitive to these therapies, and the development of resistance to therapy is an important clinical problem. Patients who have a tumour relapse usually present with tumours that are more resistant to therapy than the primary tumour. Failure to activate the apoptotic programme represents an important mode of drug resistance in tumour cells. Anticancer drugs are classified as DNA-damaging agents, ANTIMETABOLITES, mitotic inhibitors, nucleotide analogues or inhibitors of TOPOISOMERASES. Treatment with these agents or with γ-irradiation causes cellular stress and finally cell death 45,46. A key element in stressinduced apoptosis is p53 (FIG. 5) (reviewed in REFS 47,48). Rapid induction of p53 function is achieved in response to most forms of stress through post-translational mechanisms. p53 can be stabilized and activated through the inactivation of MDM2, either by ARF, as discussed above, or by direct phosphorylation of MDM2. In addition, many post-translational modifications of p53 have been shown to enhance its transcriptional activity in response to stress, including phosphorylation, SUMOYLATION and acetylation. The transcriptional activity of p53 is important for its proapoptotic function. p53 can induce the expression of proteins involved in the mitochondrial pathway such as BAX, NOXA, PUMA and p53aip1 (REFS 49 52) Apoptosome: APAF1, Cyt c, Procaspase-9 Caspase-9 Caspase-3, -6, -7 (executioner caspases) Death substrates Apoptosis SMAC/DIABLO IAP Figure 4 Apoptosis signalling through mitochondria. Chemotherapy, irradiation and other stimuli can initiate apoptosis through the mitochondrial (intrinsic) pathway. Proapoptotic BCL2 family proteins for example, BAX, BID, BAD and BIM are important mediators of these signals. Activation of mitochondria leads to the release of cytochrome c (Cyt c) into the cytosol, where it binds apoptotic protease activating factor 1 (APAF1) to form the apoptosome. At the apoptosome, the initiator caspase-9 is activated. Apoptosis through mitochondria can be inhibited on different levels by anti-apoptotic proteins, including the anti-apoptotic BCL2 family members BCL2 and BCL-X L and inhibitors of apoptosis proteins (IAPs), which are regulated by SMAC/DIABLO (second mitochondria-derived activator of caspase/direct IAP binding protein with low pi). Another way is through survival signals, such as growth factors and cytokines, that activate the phosphatidylinositol 3-kinase (PI3K) pathway. PI3K activates AKT, which phosphorylates and inactivates the pro-apoptotic BCL2-family member BAD. NATURE REVIEWS CANCER VOLUME 2 APRIL 2002 281

Oncogene Chemotherapy, irradiation ARF ARF MDM2 p53 Ubiquitylation BAX, NOXA, CD95, TRAIL-R1 MDM2 Proteasome Nucleus Figure 5 p53 and apoptosis in tumours. p53 is a key element in apoptosis induction in tumour cells. p53 is inhibited by MDM2, a ubiquitin ligase that targets p53 for destruction by the proteasome. MDM2 is inactivated by binding to ARF. Cellular stress, including that induced by chemotherapy or irradiation, activates p53 either directly, by inhibition of MDM2, or indirectly by activation of ARF. ARF can also be induced by proliferative oncogenes such as RAS. Active p53 transactivates pro-apoptotic genes including BAX, NOXA, CD95 and TRAIL-R1 to promote apoptosis. TRAIL-R1, tumour-necrosis-factor-related apoptosis-inducing ligand receptor 1. involves the CD95 system 60,61. Chemotherapeutic drugs (for example, the nucleotide analogue 5-fluoruracil, 5-FU) induce CD95 by a transcriptionally regulated, p53-dependent mechanism 53,62,63. They also engage the SAPK/JNK pathway, which eventually leads to upregulation of CD95L 64. Upregulation of CD95 and CD95L then allows the cells to either commit suicide or kill neighbouring cells. Clearly, this is not the only pathway of chemotherapy-induced cell death 65 68. Many drugs seem to initiate the mitochondrial pathway directly. Moreover, cell death might not even require caspase activation. It is questionable whether a single predominant effector pathway of chemotherapy can be identified at all. Probably, the pathway engaged depends on the stress stimulus, the cell type, the tumour environment and many other factors. However, because chemotherapy and irradiation exert their effects primarily by apoptosis induction, it is conceivable that modulation of the key elements of apoptosis signalling directly influences therapy-induced tumour-cell death. Resistance mechanisms Expression of anti-apoptotic proteins. Tumour cells can acquire resistance to apoptosis by various mechanisms that interfere at different levels of apoptosis signalling (summarized in TABLE 1). One mechanism is the overexpression of anti-apoptotic genes. A common feature of follicular B-cell lymphoma is the chromosomal translocation t(14;18), which couples the BCL2 gene to the immunoglobulin heavy chain locus, leading to enhanced BCL2 expression 69 71. BCL2 cooperates with the oncoprotein c-myc or, in acute promyelocytic leukaemia, the promyelocytic leukaemia retinoic-acid-receptor-α (PML RARα) fusion protein, thereby contributing to tumorigenesis 72 74. Some studies have shown a correlation between high levels of BCL2 expression and the severity of malignancy of human tumours 75 77.Moreover,it has been shown in in vitro and in vivo models that BCL2 expression confers resistance to many kinds of chemotherapeutic drugs and irradiation 75,78 80.In some types of tumours, a high level of BCL2 expression is associated with a poor response to chemotherapy and seems to be predictive of shorter, disease-free survival 76,77,81. The tumour-associated viruses Epstein Barr virus (EBV) and human herpesvirus 8 (HHV8 or Kaposi s sarcoma-associated herpesvirus) encode proteins that are homologues of BCL2. Both proteins BHRF1 from EBV and KSbcl-2 (vbcl-2) from HHV8 have an anti-apoptotic function and enhance survival of the infected cells 82 84. In this way, they might contribute to tumour formation after virus infection, and to resistance of these tumours to therapy. In addition, other anti-apoptotic BCL2 family members also seem to be involved in resistance of tumours to apoptosis. For example, BCL-X L can confer resistance to multiple apoptosis-inducing pathways in cell lines and seems to be upregulated by a constitutively active mutant epidermal growth factor receptor (EGFR) in vitro 80,85 88. MCL1 (myeloid cell leukaemia sequence 1) can also render cell lines resistant to chemotherapy 89,90. In some leukaemia patients, MCL1 expression was increased at the time of relapse, which indicates that some anticancer drugs might select for leukaemia cells that have elevated MCL1 levels. Human melanomas and a murine B-cell lymphoma cell line were shown to express high levels of FLIP, which interferes with apoptosis induction at the level of the death receptors 18,91,92. Moreover, in EBV-positive Burkitt s lymphoma cell lines, an increased FLIP:caspase-8 ratio was correlated with resistance to CD95- mediated apoptosis 93. Viral analogues of FLIP, called viral FLIPs (v-flips), are encoded by some tumorigenic viruses, including HHV8 (REFS 94 96). In cells that are latently infected with HHV8, v-flip is expressed at low levels, but its expression is increased in advanced Kaposi s sarcomas or on serum withdrawal from lymphoma cells in culture 97. Therefore, v-flips might contribute to the persistence and oncogenicity of v-flipencoding viruses. Although FLIP expression prevents apoptosis induction through death receptors, it does 282 APRIL 2002 VOLUME 2 www.nature.com/reviews/cancer

Box 2 Survival signalling through PI3K/AKT Survival signals include growth factors such as epidermal growth factor (EGF) and platelet-derived growth factor (PDGF), cytokines such as the interleukins IL-2 and IL-3 and some hormones such as insulin. In general, they activate phosphatidylinositol 3-kinase (PI3K) 27,28. Active PI3K generates the 3- phosphorylated lipid phosphatidylinositol-3,4,5-trisphosphate (PtdIns(3,4,5)P 3 ). This leads to recruitment of the kinases PDK1 (PtdInsP 3 -dependent kinase 1), PDK2 and AKT (also known as protein kinase B or PKB) to the plasma membrane. In the complex formed, PDK1 and PDK2 activate AKT by phosphorylation. Active AKT interferes with the apoptotic machinery. It phosphorylates, and thus inhibits, the proapoptotic BCL2 family protein BAD 179. In addition, it influences gene expression by inactivating the forkhead family transcription factors AFX and FKHRL1 which can induce pro-apoptotic genes such as CD95L and can also activate the transcription factor nuclear factor κb (NF-κB), leading to expression of anti-apoptotic genes. Survival signalling by AKT is counteracted by the tumour suppressor PTEN, a lipid phosphatase that antagonizes the action of PI3K by removing the 3-phosphate from PtdIns(3,4,5)P 3. not inhibit cell death induced by perforin/granzyme, chemotherapeutic drugs or γ-irradiation 98. Nevertheless, it mediates the immune escape of tumours in mouse models 99,100. Tumours with high expression levels of FLIP were shown to escape from T- cell-mediated immunity in vivo, despite the presence of the perforin/granzyme pathway 99, so tumour cells with elevated FLIP levels seem to have a selective advantage. FLIP overexpression also prevents rejection of tumours by perforin-deficient NK cells 101. Another mechanism by which tumours interfere with death-receptor-mediated apoptosis might be the expression of soluble receptors that act as decoys for death ligands. To date, two distinct soluble receptors soluble CD95 (scd95) and decoy receptor 3 (DcR3) have been shown to competitively inhibit CD95 signalling. scd95 is expressed in various malignancies, and elevated levels can be found in the sera of cancer patients. High scd95 serum levels were associated with poor prognosis in melanoma patients 102 105. DcR3 binds to CD95L and the TNF family member LIGHT (a cytokine that is homologous to lymphotoxins, exhibits inducible expression and competes with herpes simplex virus (HSV) glycoprotein D for herpesvirus entry mediator (HVEM), a receptor expressed by T cells) and inhibits CD95L-induced apoptosis. It is genetically amplified in several lung and colon carcinomas and is overexpressed in several adenocarcinomas, glioma cell lines and glioblastomas 5,6,106. Ectopic expression of DcR3 in a rat glioma model resulted in decreased immune-cell infiltration, which indicates that DcR3 is involved in immune evasion of malignant glioma 106. Expression of the IAP-family protein survivin is highly tumour specific 107. It is found in most human tumours but not in normal adult tissues 108.In neuroblastoma, expression correlates with a more aggressive and unfavourable disease 109. But although survivin has a BIR domain, it is not clear whether it directly acts as an apoptosis inhibitor, for example by binding to caspase-9 or interacting with SMAC/DIABLO. Survivin might also be necessary for completion of the cell cycle. Nevertheless, overexpression of survivin counteracts apoptosis in some settings 107 : in transgenic mice that express survivin in the skin, its anti-apoptotic function was more prominent than its role in cell division. Survivin inhibited UVB-induced apoptosis in vitro and in vivo, whereas it did not affect CD95- induced cell death 110. Expression of a non-phosphorylatable mutant of survivin induces cytochrome c release and cell death. In xenograft tumour models, this mutant suppressed tumour growth and reduced intraperitoneal tumour dissemination 111,112. Another IAP family member, ciap2, is affected by the translocation t(11;18)(q21;q21) that is found in about 50% of marginal cell lymphomas of the mucosa-associated lymphoid tissue (MALT) 113. This indicates a role for ciap2 in the development of MALT lymphoma. ML-IAP is expressed at high levels in melanoma cell lines, but not in primary melanocytes. Melanoma cell lines that express ML-IAP are significantly more resistant to drug-induced apoptosis than those that do not express ML-IAP 114. Finally, tumour cells resist killing by cytotoxic lymphocytes not only by blocking the death-receptor pathway, but also by interfering with the perforin/granzyme pathway. Expression of the serine protease inhibitor PI-9/SPI-6, which inhibits granzyme B, results in the resistance of tumour cells to cytotoxic lymphocytes, leading to immune escape 115. Inactivation of pro-apoptotic genes. Besides overexpression of anti-apoptotic genes, tumours can acquire apoptosis resistance by downregulating or mutating pro-apoptotic molecules. In certain types of cancer, the pro-apoptotic BCL2 family member BAX is mutated 116 118. Frameshift mutations that lead to loss of expression, and mutations in the BH domains that result in loss of functions, are common. Tumour cell lines with frameshift mutations are more resistant to apoptosis. Reduced BAX expression is associated with a poor response rate to chemotherapy and shorter survival in some situations 119.Several studies in mice have confirmed the function of Bax as a tumour suppressor. In a transgenic mouse tumour, Bax expression is induced by p53, resulting in slow tumour growth and a high percentage of apoptotic cells. In Bax-deficient mice, however, tumour growth is accelerated and apoptosis decreases, indicating that Bax is required for a full p53-mediated response 120.In a different study, induction of Bax expression in an inducible cell line restored sensitivity to apoptosis and significantly reduced tumour growth in severe combined immunodeficient (SCID) mice 121. Moreover, others showed that inactivation of wildtype Bax confers a strong advantage during clonal evolution of the tumour. Injection of clones with either wild-type or mutant Bax into nude mice led to outgrowth of tumours that did not express Bax in both situations 122. NATURE REVIEWS CANCER VOLUME 2 APRIL 2002 283

Table 1 Mechanisms of tumour resistance to apoptosis Molecules References Expression of anti-apoptotic molecules BCL2 family members: BCL2 75 80,82 84 BCL-X L 80,85 88 MCL1 89,90 FLIP 18,91 101 Soluble receptors for death ligands: Soluble CD95 102 105 DcR3 5,6,106 IAPs: Survivin 107 112 ciap2 113 ML-IAP 114 PI-9/SPI-6 111 Downregulation and mutation of pro-apoptotic genes BAX 116 122 APAF1 123 Caspase-8 124 Death receptors: CD95 125 133 TRAIL-R1 134 135 TRAIL-R2 134,136,137 XAF1 138 Alterations of the p53 pathway p53 47,139 142 INK4A/ARF 48,143 ASPP 139 Alterations of the PI3K/AKT pathway PI3K 145 147 PTEN 143 145 AKT 146 Further mechanisms Expression of transporters: MDR1/P-glycoprotein 154,155 MRP 153 Alterations of NF-κB activity 157,158 Extracellular matrix 159 APAF1, apoptotic protease activating factor-1; ASPP, apoptosis stimulating protein of p53; DcR3, decoy receptor 3; FLIP, FAS-associated DD protein-like interleukin-1β-converting enzyme-like protease-inhibitory protein; IAP, inhibitor of apoptosis proteins; MCL1, myeloid cell leukaemia sequence 1; MDR1, multi-drug resistance protein-1; ML-IAP, melanoma IAP; NF-κB, nuclear factor κb; PI3K, phosphatidylinositol 3-kinase; PI-9/SPI-6, protease inhibitor 9/serine protease inhibitor 6; R, receptor; TRAIL, tumour-necrosis factor-related apoptosis-inducing ligand; XAF1, XIAP-associated factor 1. DOXORUBICIN A chemotherapeutic drug that induces DNA strand breaks, which initiate apoptosis. Metastatic melanomas have found another way to escape mitochondria-dependent apoptosis. These tumours often do not express APAF1, which forms an integral part of the apoptosome 123, and the APAF1 locus shows a high rate of allelic loss. The remaining allele is transcriptionally inactivated by gene methylation. APAF1-negative melanomas fail to respond to chemotherapy a situation that is commonly found in this type of tumour. A similar strategy has been reported for neuroblastomas in which the N-MYC oncogene has been amplified 124. In these tumours, the gene for the initiator caspase-8 is frequently inactivated by gene deletion or methylation. Caspase-8-deficient neuroblastoma cells are resistant to death-receptor- and DOXORUBICIN-mediated apoptosis. Moreover, death receptors are downregulated or inactivated in many tumours. The expression of the death receptor CD95 is reduced in some tumour cells for example, in hepatocellular carcinomas, neoplastic colon epithelium, melanomas and other tumours 125 128 compared with their normal counterparts. Loss of CD95, probably by downregulation of transcription, might contribute to chemoresistance and immune evasion. Oncogenic RAS seems to downregulate CD95 (REF. 129), and in hepatocellular carcinomas loss of CD95 expression is accompanied by p53 aberrations 128. Several CD95 gene mutations have been reported in primary samples of myeloma and T-cell leukaemia 130 132. The mutations include point mutations in the cytoplasmic death domain of CD95 and a deletion that leads to a truncated form of the death receptor. These mutated forms of CD95 might interfere in a dominant-negative way with apoptosis induction by CD95. In families with germ-line CD95 mutations, which usually result in autoimmune lymphoproliferative syndrome (ALPS), the risk of developing lymphomas is increased 133. Deletions and mutations of the death receptors TRAIL-R1 and TRAIL-R2 have also been observed in tumours 134 137. The frequent deletion of the chromosomal region 8p21-22 in head and neck cancer and in non-small-cell lung cancers affects the TRAIL-R2 gene. Mutations have been found in the ectodomain or the death domain of TRAIL-R1 or TRAIL-R2. Further mutations result in truncated forms of these TRAIL receptors or other anti-apoptotic forms. Finally, reduced expression of the pro-apoptotic protein XAF1 (XIAP-associated factor 1) has been observed in various cancer cell lines 138. XAF1 binds to XIAP and antagonizes its anti-apoptotic function at the level of the caspases. Alterations of the p53 pathway. As p53 has a central function in apoptosis induction, alterations of the p53 pathway influence the sensitivity of tumours to apoptosis 47,139 (FIG. 5). Tumours that are deficient in Trp53 (the gene that encodes p53 in mice) in immunocompromised mice and cell lineages from transgenic mice that express mutant Trp53 showed a poor response to γ-irradiation or chemotherapy 140. Specific mutations in TP53 (the gene that encodes p53 in humans) have been linked to primary resistance to doxorubicin treatment and early relapse in patients with breast cancer 141. In cancer cell lines, the specific disruption of the TP53 gene conferred resistance to 5-FU, but greater sensitivity to adriamycin or radiation in vitro 142. Mutations of CDKN2A, which encodes ARF (as well as INK4A), are almost as widespread in tumours as are TP53 mutations 48. Lymphomas from Trp53-knockout mice and from Cdkn2a-knockout mice are highly invasive, display apoptotic defects and are markedly resistant to chemotherapy in vitro and in vivo 143. In about 70% of breast cancers, wild-type TP53 is expressed but fails to suppress tumour growth 144. This might be explained by a lack of the ASPP (apoptosis stimulating protein of p53) family of proteins. ASPP proteins interact with p53 and specifically enhance the DNA-binding and transactivation function of p53 on 284 APRIL 2002 VOLUME 2 www.nature.com/reviews/cancer

STAT3 A member of the STAT (signal transducer and activator of transcription) family of transcription factors. STATs are activated through phosphorylation by Janus kinases and have an important role in cytokine receptor signalling. RNA INTERFERENCE (RNAi). Use of double-stranded RNA to target specific mrnas for degradation, resulting in sequence-specific posttranscriptional gene silencing. the promoters of proapoptotic genes in vivo. In this way, they stimulate apoptosis induction by p53 and do not affect proliferation. ASPP expression is frequently downregulated in breast carcinomas that express wild-type TP53, resulting in p53 unresponsiveness. Altered survival signalling. Most tumours are independent of the survival signals that protect normal cells from death by neglect. This is achieved by alterations in the PI3K/AKT pathway (BOX 2). Oncogenes such as RAS or BCR ABL can increase PI3K activity 145.The catalytic subunit of PI3K has been shown to be amplified in ovarian cancer 146,147. PTEN, the cellular antagonist of PI3K, is frequently deleted in advanced tumours, and a significant rate of PTEN mutations can be found in various cancer types 148 150. Moreover, AKT, the serine/threonine kinase that mediates survival signals, is overexpressed in several malignancies 151.All of these alterations lead to a constitutively active survival signalling pathway that enhances the insensitivity of tumour cells to apoptosis induction. Further mechanisms. Resistance to chemotherapy can also be attributed to the presence of a molecular transporter that actively expels chemotherapeutic drugs from the tumour cells 152. The two transporters that are commonly found to confer chemoresistance in cancer are the MDR1 gene products P-glycoprotein and MRP (multidrug resistance-associated protein) 153.P-glycoprotein protects cells not only from chemotherapyinduced apoptosis, but also from other caspase-dependent death stimuli such as CD95L, TNF and UV irradiation. However, it does not confer resistance to the perforin/granzyme pathway 154,155. An important factor influencing apoptosis of tumour cells is the transcription factor nuclear factor κb (NF-κB) 45,156. Normally, NF-κB remains sequestered in an inactive state by the cytoplasmic inhibitor of NF-κB (IκB) proteins. However, a variety of external stimuli including cytokines, pathogens, stress and chemotherapeutic agents can lead to activation of NF-κB by phosphorylation, ubiquitylation, and the subsequent degradation of IκB. The DNA-binding subunits of NF-κB migrate into the nucleus and activate expression of target genes. Depending on the stimulus and the cellular context, NF-κB can activate pro-apoptotic genes, such as those encoding CD95, CD95L and TRAIL receptors, and anti-apoptotic genes, such as those encoding IAPs and BCL-X L. Genes encoding NF-κB or IκB proteins are amplified or translocated in human cancer 157.In Hodgkin s disease cells, constitutive activity of NF-κB has been observed 158. The extracellular matrix might also contribute to drug resistance in vivo 159. Small-cell lung cancer is surrounded by an extensive stroma of extracellular matrix, and adhesion of the cancer cells to the extracellular matrix suppresses chemotherapy-induced apoptosis through integrin signalling. Furthermore, in myeloma, constitutive activation of STAT3 signalling upregulates BCL-X L and so confers resistance to apoptosis 160. Strategies to overcome apoptosis resistance In the past couple of years, numerous mechanisms of apoptosis induction in tumour cells have been discovered, and we now know many of the strategies that tumours use to acquire apoptosis resistance. The exciting challenge today is to translate this knowledge into clinical applications (reviewed in REF. 161). Apoptosis research has provided several new cytotoxic agents that might circumvent the pathways that are blocked in tumours. The death ligand TRAIL has attracted a great deal of attention in this regard because most cancer cell lines are sensitive to TRAIL, whereas normal cells are resistant 162.In SCID mice, treatment of tumour xenografts with TRAIL suppressed tumour growth, with no adverse effects. Chemotherapy or irradiation could sensitize resistant cells to TRAIL-induced apoptosis in vitro and in vivo 163 165. Initial success has also been obtained with the death ligand TNF. It selectively kills malignant cell lines, especially in combination with interferon, and can cause tumour regression in vivo 166,167. However, owing to its pleiotropic effects in humans, clinical trials in cancer patients have so far been disappointing 168. Another study showed that even small, synthetic, pro-apoptotic peptides might suffice to kill tumour cells selectively 169. These peptides contained two domains: a tumour blood vessel homing motif and an apoptosis-inducing sequence. In mice, the peptides displayed anticancer activity, but no apparent toxicity. A future strategy to exploit our insight into tumour resistance mechanisms is to downregulate anti-apoptotic molecules. Infusion of a BCL2 antisense oligonucleotide into tumour-bearing mice markedly reduced tumour growth 170, and the preliminary results of clinical studies seem promising 171.A further anti-apoptotic BCL2 family protein, BCL-X L, has also been downregulated by an antisense approach in human cell lines 172. Moreover, targeting of the IAP protein survivin has been shown to restore sensitivity to apoptosis in mice 111,112. Besides antisense technology, it might be possible to use RNA INTERFERENCE to downregulate the expression of anti-apoptotic genes in the future 173. A different concept is to influence specific signalling pathways that are involved in apoptosis. Recent studies provide evidence that small molecules can restore the activity of mutated p53 and activate apoptosis in tumour cells 174. Adenoviral transfer of the wild-type Trp53 gene into tumour cells with mutated p53 induced apoptosis and suppressed tumour growth in nude mice 175,176. In addition, inhibition of the transcription factor NF-κB sensitized chemoresistant tumours, resulting in tumour regression 177. Despite these efforts, before a successful therapy that restores normal sensitivity to apoptotic pathways in tumour cells can be developed, a number of general problems still have to be solved. In principle, tumours evolve as a clone. However, a macroscopic tumour is heterogeneous and it is likely that different cells within the tumour have also acquired different mechanisms of apoptosis resistance. Moreover, multiple resistance mechanisms might have developed in a NATURE REVIEWS CANCER VOLUME 2 APRIL 2002 285

single tumour cell. Therefore, a combination of various therapeutic strategies will probably be the most effective treatment. A further general problem is selectively modulating the sensitivity of tumour cells without affecting other normal cells. The phenomenon of apoptosis was first described by Carl Vogt in 1842 (REF. 180). From the discovery of its relevance for cancer, it has not taken us long to unravel the principles and intricacies of apoptosis signalling on a molecular level. Now, apoptosis research stands at the doorstep of the oncology clinic, and we are looking forward to new therapeutic strategies based on the modulation of apoptosis that will eventually be used routinely for the benefit of cancer patients. 1. Hanahan, D. & Weinberg, R. A. The hallmarks of cancer. Cell 100, 57 70 (2000). 2. Krammer, P. H., Galle, P. R., Moller, P. & Debatin, K. M. CD95(APO-1/Fas)-mediated apoptosis in normal and malignant liver, colon, and hematopoietic cells. Adv. Cancer Res. 75, 251 273 (1998). 3. Krammer, P. H. CD95(APO-1/Fas)-mediated apoptosis: live and let die. Adv. Immunol. 71, 163 210 4. Schmitz, I., Kirchhoff, S. & Krammer, P. H. Regulation of death receptor-mediated apoptosis pathways. Int. J. Biochem. Cell Biol. 32, 1123 1136 (2000). 5. Pitti, R. M. et al. Genomic amplification of a decoy receptor for Fas ligand in lung and colon cancer. Nature 396, 699 703 (1998). Identification of the decoy receptor DcR3 that inhibits CD95L-induced apoptosis. DcR3 is amplified in tumours. 6. Yu, K. Y. et al. A newly identified member of tumor necrosis factor receptor superfamily (TR6) suppresses LIGHT-mediated apoptosis. J. Biol. Chem. 274, 13733 13736 7. Sprick, M. R. et al. FADD/MORT1 and caspase-8 are recruited to TRAIL receptors 1 and 2 and are essential for apoptosis mediated by TRAIL receptor 2. Immunity 12, 599 609 (2000). 8. Kischkel, F. C. et al. Cytotoxicity-dependent APO-1 (Fas/CD95)-associated proteins form a death-inducing signaling complex (DISC) with the receptor. EMBO J. 14, 5579 5588 (1995). The first description of the death-inducing signalling complex (DISC) in death receptor-mediated apoptosis. 9. Kischkel, F. C. et al. Death receptor recruitment of endogenous caspase-10 and apoptosis initiation in the absence of caspase-8. J. Biol. Chem. 276, 46639 46646 (2001). 10. Scaffidi, C. et al. Two CD95 (APO-1/Fas) signaling pathways. EMBO J. 17, 1675 1687 (1998). Clarifies the role of mitochondria in death-receptormediated apoptosis. 11. Zamzami, N. & Kroemer, G. The mitochondrion in apoptosis: how Pandora s box opens. Nature Rev. Mol. Cell Biol. 2, 67 71 (2001). 12. Martinou, J. C. & Green, D. R. Breaking the mitochondrial barrier. Nature Rev. Mol. Cell Biol. 2, 63 67 (2001). 13. Rathmell, J. C. & Thompson, C. B. The central effectors of cell death in the immune system. Annu. Rev. Immunol. 17, 781 828 14. Savill, J. & Fadok, V. Corpse clearance defines the meaning of cell death. Nature 407, 784 788 (2000). 15. Borner, C. & Monney, L. Apoptosis without caspases: an inefficient molecular guillotine? Cell Death Differ. 6, 497 507 16. Xiang, J., Chao, D. T. & Korsmeyer, S. J. BAX-induced cell death may not require interleukin 1 β-converting enzyme-like proteases. Proc. Natl Acad. Sci. USA 93, 14559 14563 (1996). 17. Sperandio, S., de Belle, I. & Bredesen, D. E. An alternative, nonapoptotic form of programmed cell death. Proc. Natl Acad. Sci. USA 97, 14376 14381 (2000). 18. Krueger, A., Baumann, S., Krammer, P. H. & Kirchhoff, S. FLICE-inhibitory proteins: regulators of death receptormediated apoptosis. Mol. Cell. Biol. 21, 8247 8254 (2001). 19. Deveraux, Q. L. & Reed, J. C. IAP family proteins suppressors of apoptosis. Genes Dev. 13, 239 252 20. Verhagen, A. M. et al. Identification of DIABLO, a mammalian protein that promotes apoptosis by binding to and antagonizing IAP proteins. Cell 102, 43 53 (2000). 21. Du, C., Fang, M., Li, Y., Li, L. & Wang, X. Smac, a mitochondrial protein that promotes cytochrome c- dependent caspase activation by eliminating IAP inhibition. Cell 102, 33 42 (2000). 22. Evan, G. I. & Vousden, K. H. Proliferation, cell cycle and apoptosis in cancer. Nature 411, 342 348 (2001). 23. Evan, G. I. et al. Induction of apoptosis in fibroblasts by c- myc protein. Cell 69, 119 128 (1992). 24. Bissonnette, R. P., Echeverri, F., Mahboubi, A. & Green, D. R. Apoptotic cell death induced by c-myc is inhibited by bcl-2. Nature 359, 552 554 (1992). References 24 and 69 73 showed that anti-apoptotic proteins are important oncogenes by reporting the oncogenic potential of BCL2. 25. Harrington, E. A., Bennett, M. R., Fanidi, A. & Evan, G. I. c-myc-induced apoptosis in fibroblasts is inhibited by specific cytokines. EMBO J. 13, 3286 3295 (1994). 26. Sabbatini, P., Lin, J., Levine, A. J. & White, E. Essential role for p53-mediated transcription in E1A-induced apoptosis. Genes Dev. 9, 2184 2192 (1995). 27. Stambolic, V., Mak, T. W. & Woodgett, J. R. Modulation of cellular apoptotic potential: contributions to oncogenesis. Oncogene 18, 6094 6103 28. Datta, S. R., Brunet, A. & Greenberg, M. E. Cellular survival: a play in three Akts. Genes Dev. 13, 2905 2927 29. Frisch, S. M. & Screaton, R. A. Anoikis mechanisms. Curr. Opin. Cell Biol. 13, 555 562 (2001). 30. Frisch, S. M. & Francis, H. Disruption of epithelial cell-matrix interactions induces apoptosis. J. Cell Biol. 124, 619 626 (1994). 31. Khwaja, A., Rodriguez-Viciana, P., Wennstrom, S., Warne, P. H. & Downward, J. Matrix adhesion and Ras transformation both activate a phosphoinositide 3-OH kinase and protein kinase B/Akt cellular survival pathway. EMBO J. 16, 2783 2793 (1997). 32. Puthalakath, H., Huang, D. C., O Reilly, L. A., King, S. M. & Strasser, A. The proapoptotic activity of the Bcl-2 family member Bim is regulated by interaction with the dynein motor complex. Mol. Cell 3, 287 296 33. Puthalakath, H. et al. Bmf: a proapoptotic BH3-only protein regulated by interaction with the myosin V actin motor complex, activated by anoikis. Science 293, 1829 1832 (2001). 34. Oulton, R. & Harrington, L. Telomeres, telomerase, and cancer: life on the edge of genomic stability. Curr. Opin. Oncol. 12, 74 81 (2000). 35. Hahn, W. C. et al. Inhibition of telomerase limits the growth of human cancer cells. Nature Med. 5, 1164 1170 36. Zhang, X., Mar, V., Zhou, W., Harrington, L. & Robinson, M. O. Telomere shortening and apoptosis in telomerase-inhibited human tumor cells. Genes Dev. 13, 2388 2399 37. Chin, L. et al. p53 deficiency rescues the adverse effects of telomere loss and cooperates with telomere dysfunction to accelerate carcinogenesis. Cell 97, 527 538 38. Shresta, S., Pham, C. T., Thomas, D. A., Graubert, T. A. & Ley, T. J. How do cytotoxic lymphocytes kill their targets? Curr. Opin. Immunol. 10, 581 587 (1998). 39. Kagi, D. et al. Cytotoxicity mediated by T cells and natural killer cells is greatly impaired in perforin-deficient mice. Nature 369, 31 37 (1994). 40. Heusel, J. W., Wesselschmidt, R. L., Shresta, S., Russell, J. H. & Ley, T. J. Cytotoxic lymphocytes require granzyme B for the rapid induction of DNA fragmentation and apoptosis in allogeneic target cells. Cell 76, 977 987 (1994). 41. Medema, J. P. et al. Cleavage of FLICE (caspase-8) by granzyme B during cytotoxic T lymphocyte-induced apoptosis. Eur. J. Immunol. 27, 3492 3498 (1997). 42. Heibein, J. A. et al. Granzyme B-mediated cytochrome c release is regulated by the Bcl-2 family members Bid and Bax. J. Exp. Med. 192, 1391 1402 (2000). 43. Rouvier, E., Luciani, M. F. & Golstein, P. Fas involvement in Ca 2+ -independent T cell-mediated cytotoxicity. J. Exp. Med. 177, 195 200 (1993). 44. Li, J. H. et al. The regulation of CD95 ligand expression and function in CTL. J. Immunol. 161, 3943 3949 (1998). 45. Herr, I. & Debatin, K. M. Cellular stress response and apoptosis in cancer therapy. Blood 98, 2603 2614 (2001). 46. Stahnke, K., Fulda, S., Friesen, C., Strauss, G. & Debatin, K. M. Activation of apoptosis pathways in peripheral blood lymphocytes by in vivo chemotherapy. Blood 98, 3066 3073 (2001). 47. Ryan, K. M., Phillips, A. C. & Vousden, K. H. Regulation and function of the p53 tumor suppressor protein. Curr. Opin. Cell Biol. 13, 332 337 (2001). 48. Sherr, C. J. The INK4A/ARF network in tumour suppression. Nature Rev. Mol. Cell Biol. 2, 731 737 (2001). 49. Miyashita, T. & Reed, J. C. Tumor suppressor p53 is a direct transcriptional activator of the human bax gene. Cell 80, 293 299 (1995). 50. Oda, E. et al. Noxa, a BH3-only member of the Bcl-2 family and candidate mediator of p53-induced apoptosis. Science 288, 1053 1058 (2000). 51. Nakano, K. & Vousden, K. H. PUMA, a novel proapoptotic gene, is induced by p53. Mol. Cell 7, 683 694 (2001). 52. Oda, K. et al. p53aip1, a potential mediator of p53- dependent apoptosis, and its regulation by Ser-46- phosphorylated p53. Cell 102, 849 862 (2000). 53. Muller, M. et al. p53 activates the CD95 (APO-1/Fas) gene in response to DNA damage by anticancer drugs. J. Exp. Med. 188, 2033 2045 (1998). 54. Guan, B., Yue, P., Clayman, G. L. & Sun, S. Y. Evidence that the death receptor DR4 is a DNA damage-inducible, p53- regulated gene. J. Cell Physiol. 188, 98 105 (2001). 55. Wu, G. S. et al. KILLER/DR5 is a DNA damage-inducible p53-regulated death receptor gene. Nature Genet. 17, 141 143 (1997). 56. Caelles, C., Helmberg, A. & Karin, M. p53-dependent apoptosis in the absence of transcriptional activation of p53- target genes. Nature 370, 220 223 (1994). 57. Soengas, M. S. et al. Apaf-1 and caspase-9 in p53- dependent apoptosis and tumor inhibition. Science 284, 156 159 58. Bennett, M. et al. Cell surface trafficking of Fas: a rapid mechanism of p53-mediated apoptosis. Science 282, 290 293 (1998). 59. Hug, H. et al. Reactive oxygen intermediates are involved in the induction of CD95 ligand mrna expression by cytostatic drugs in hepatoma cells. J. Biol. Chem. 272, 28191 28193 (1997). 60. Friesen, C., Herr, I., Krammer, P. H. & Debatin, K. M. Involvement of the CD95 (APO-1/FAS) receptor/ligand system in drug- induced apoptosis in leukemia cells. Nature Med. 2, 574 577 (1996). Provides evidence that the CD95 system is involved in drug-induced apoptosis and so identifies one particular mechanism by which apoptosis is caused by anticancer drugs at the molecular level. 61. Fulda, S. et al. Activation of the CD95 (APO-1/Fas) pathway in drug- and γ-irradiation-induced apoptosis of brain tumor cells. Cell Death Differ. 5, 884 893 (1998). 62. Muller, M. et al. Drug-induced apoptosis in hepatoma cells is mediated by the CD95 (APO-1/Fas) receptor/ligand system and involves activation of wild-type p53. J. Clin. Invest. 99, 403 413 (1997). 63. Muller, M., Scaffidi, C. A., Galle, P. R., Stremmel, W. & Krammer, P. H. The role of p53 and the CD95 (APO-1/Fas) death system in chemotherapy-induced apoptosis. Eur. Cytokine Netw. 9, 685 686 (1998). 64. Eichhorst, S. T. et al. A novel AP-1 element in the CD95 ligand promoter is required for induction of apoptosis in hepatocellular carcinoma cells upon treatment with anticancer drugs. Mol. Cell. Biol. 20, 7826 7837 (2000). 65. Newton, K. & Strasser, A. Ionizing radiation and chemotherapeutic drugs induce apoptosis in lymphocytes in the absence of Fas or FADD/MORT1 signaling. Implications for cancer therapy. J. Exp. Med. 191, 195 200 (2000). 66. Yeh, W. C. et al. FADD: essential for embryo development and signaling from some, but not all, inducers of apoptosis. Science 279, 1954 1958 (1998). 67. Wesselborg, S., Engels, I. H., Rossmann, E., Los, M. & Schulze-Osthoff, K. Anticancer drugs induce caspase- 8/FLICE activation and apoptosis in the absence of CD95 receptor/ligand interaction. Blood 93, 3053 3063 286 APRIL 2002 VOLUME 2 www.nature.com/reviews/cancer