DNA damage-induced apoptosis - Nature

Oncogene (2004) 23, 2797–2808

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $25.00 www.nature.com/onc

DNA damage-induced apoptosis Chris J Norbury1 and Boris Zhivotovsky*,2 1

Sir William Dunn School of Pathology, University of Oxford, South Parks Road, Oxford OX1 3RE, UK; 2Institute of Environmental Medicine, Karolinska Institutet, Box 210, Nobels va¨g. 13, SE-171 77 Stockholm, Sweden

Unicellular organisms respond to the presence of DNA lesions by activating cell cycle checkpoint and repair mechanisms, while multicellular animals have acquired the further option of eliminating damaged cells by triggering apoptosis. Defects in DNA damage-induced apoptosis contribute to tumorigenesis and to the resistance of cancer cells to a variety of therapeutic agents. The intranuclear mechanisms that signal apoptosis after DNA damage overlap with those that initiate cell cycle arrest and DNA repair, and the early events in these pathways are highly conserved. In addition, multiple independent routes have recently been traced by which nuclear DNA damage can be signalled to the mitochondria, tipping the balance in favour of cell death rather than repair and survival. Here, we review current knowledge of nuclear DNA damage signalling, giving particular attention to interactions between these nuclear events and apoptotic processes in other intracellular compartments. Oncogene (2004) 23, 2797–2808. doi:10.1038/sj.onc.1207532 Keywords: DNA damage; DNA repair; apoptosis; mechanisms

Introduction The integrity of genomic DNA is constantly under threat, even in perfectly healthy cells. DNA damage can result from the action of endogenous reactive oxygen species, or from stochastic errors in replication or recombination, as well as from environmental and therapeutic genotoxins. While unicellular organisms respond to the presence of DNA lesions by activating cell cycle checkpoint and repair mechanisms, multicellular animals have the additional possibility of eliminating the damaged cells by triggering their death. Induction of apoptosis has been recognized as a possible outcome of DNA damage for more than 20 years (Wyllie et al., 1980). The earlier literature in this area is dominated by studies of mammalian systems, but it has since become apparent that DNA damage can also induce apoptosis in diverse experimental model organisms, most notably Caenorhabditis elegans and Droso*Correspondence: B Zhivotovsky; E-mail: [email protected]

phila melanogaster. The combination of data from these genetically amenable models with those from mammalian and other vertebrate species has revealed conservation of several key molecular mechanisms, as well as species-specific variations on these themes. Most of the morphological criteria that were first used to distinguish apoptosis from necrosis relate to the nucleus (Kerr et al., 1972). Degradation of chromosomal DNA into oligonucleosome length fragments in irradiated lymphoid tissues was reported as early as 1976 (Skalka et al., 1976). This observation was first linked to endonuclease activation in 1980 (Wyllie, 1980) and has since been used as a biochemical marker of apoptosis. Transcriptional activation of gene expression is also a common, though not universal, feature of apoptosis. Unsurprisingly, the interest of many researchers in the field was therefore initially focused on understanding the nuclear involvement in apoptosis signalling. Questions about the primacy of nuclear events were raised when NUC-1, a protein essential for apoptotic DNA degradation in C. elegans, was shown to act downstream of the key apoptotic regulators CED-3 and CED-4 (Ellis and Horvitz, 1986). In some cases, DNA degradation required prior engulfment of apoptotic cells by phagocytes, again indicating that nuclease activation is a late event. In addition, upon treatment with staurosporine or antibodies against Fas (Apo-1/CD95), enucleate cells (cytoplasts) underwent morphological changes characteristic of apoptosis, which could be inhibited by overexpression of the antiapoptotic protein Bcl-2 (Jacobson et al., 1994; Schulze-Osthoff et al., 1994). A variety of strands of evidence subsequently indicated that caspases (cysteine-aspartate proteases) play central roles in apoptotic signalling and execution (Thornberry, 1999 and references therein). The shift of focus from the nucleus to the cytoplasm became even more pronounced when, in the mid-1990s, Wang and co-workers demonstrated that the release of cytochrome c from mitochondria into the cytosol resulted in caspase activation and the execution of apoptosis (Liu et al., 1996). This, along with related key observations, led to an explosion in research focusing on mitochondrial regulation of apoptosis (for a review, see Cory and Adams, 2002). It is clear that any coherent view of the apoptotic response to DNA damage must take into account events in the nucleus and the cytoplasm, as well

DNA damage and apoptosis CJ Norbury and B Zhivotovsky

2798

as the regulatory traffic within and between these compartments.

Sensitivity and resistance to DNA damage-induced apoptosis Animal cells differ markedly in their propensity to commit to apoptotic death following DNA damage. Thymocytes, for example, are exquisitely primed to engage apoptosis in response to DNA damage as well as a variety of other triggers. By contrast, primary fibroblasts appear to undergo DNA damage-induced apoptosis (DDIA) rarely, if at all – a feature of these cells familiar to anyone who has irradiated them for use as ‘feeder’ layers in cell culture systems. These differing apoptotic thresholds presumably reflect differing basal patterns of expression of pro- and anti-apoptotic regulators, and generally relate to the importance of apoptosis during the normal developmental programme of the cells in question. The relative radiosensitivity of thymocytes, for instance, can be related to their propensity for apoptosis during negative selection in vivo. Cells less prone to triggering apoptosis can use alternative strategies to avoid the propagation of mutations that might otherwise threaten the health of the whole organism.

Consequences of cell survival after DNA damage Loss of apoptosis undoubtedly plays a major role in tumorigenesis (Igney and Krammer, 2002). Studies in this area have frequently addressed the significance of loss of p53-dependent apoptosis in response to oncogene activation or hypoxia, but loss of DDIA probably also contributes significantly to tumour development. The issue is complicated by the fact that many of the mediators of DDIA have additional functions in cell cycle regulation and/or DNA repair, which in turn can have dramatic influences on genome stability. For example, loss of function of the XPB or XPD helicase, as seen in xeroderma pigmentosum patients, results in defective DDIA as well as defective nucleotide excision repair and a hugely elevated frequency of UV-induced skin tumours (Wang et al., 1996). Similarly, the ATM protein kinase is required for a variety of cell cycle checkpoint responses to DNA double-strand breaks (DSB), but roles for this protein in DNA repair and DDIA have also been identified, as discussed below. It is therefore difficult to determine the extent to which the excess tumour incidence seen in ataxia telangiectasia (AT) (ATM/) patients is attributable specifically to apoptotic defects. Even in the absence of faithful DNA repair, cell survival following DNA damage is not necessarily associated with an increased risk of tumorigenesis. The lack of DDIA in primary fibroblasts, for example, is balanced by the capacity of these cells to maintain long-term p53-dependent cell cycle arrest in G1 or G2 Oncogene

Figure 1 Life versus death decisions following DNA damage. Common damage detection and signal transduction mechanisms are shared between pathways governing DNA repair, cell cycle checkpoint activation and apoptosis. The consequence of any given level of DNA damage is cell type-specific. By allowing the survival and proliferation of cells with damaged DNA, failure of these processes contributes to tumorigenesis and resistance to cancer therapies

(Di Leonardo et al., 1994; Baus et al., 2003). In this way, the threat posed by proliferation of mutated cells is effectively eliminated without recourse to apoptosis. On the other hand, it is important that the apoptotic programme, once initiated, is completed (Figure 1). Transient activation of the caspase-activated DNase (CAD) followed by cell recovery might even contribute to the initiation of site-specific chromosomal translocations during tumorigenesis (Vaughan et al., 2002). Perhaps for this reason amplification mechanisms, notably those involving release of mitochondrial proteins following permeabilization of the outer mitochondrial membrane, drive the complete destruction of the cell following commitment to DDIA, as discussed below. Loss of apoptotic potential probably also makes a significant contribution to the resistance of cancer cells to therapies that induce DNA damage (Lowe et al., 1993a). It would be a gross oversimplification to consider cancer cells simply as apoptosis-defective versions of their normal counterparts. There is no doubt, however, that loss of function of proapoptotic regulators such as p53 and/or overexpression of antiapoptotic proteins such as Bcl-2 can tilt the balance in favour of tumour cell survival following DNA damagebased therapy.

Signals linking DNA damage to cell cycle arrest, DNA repair or apoptosis Any signalling pathway responding to DNA damage must comprise one or more sensors, coupled via


2799

transducers to effector components (Figure 1). From an evolutionary standpoint, it appears that the requirement for DNA damage sensors predated the emergence of multicellularity by billions of years. DNA damage responses such as homologous recombinational repair were clearly in place before the emergence of the eukaryotes. Cell cycle checkpoint controls that are found in all modern eukaryotes must also have arisen early in eukaryotic evolution. It is debatable whether unicellular organisms could derive any selective advantage from the acquisition of a mechanism for DDIA; indeed, these organisms lack apoptotic effectors such as caspases that are found in multicellular animals. Nonetheless, molecular mechanisms that mediate DNA damage-induced cell cycle arrest and DNA repair in unicellular species have come to be involved additionally in the activation of apoptosis in metazoans. The identification of additional mechanisms has helped to explain how nuclear DNA damage signals can be coupled separately to the cytoplasmic events of apoptosis, in particular the release of apoptotic regulators from mitochondria. All eukaryotic cells studied to date employ PI3kinase-related protein kinases of the ATM/ATR family to trigger a variety of DNA damage responses (Shiloh, 2001). Members of this family were first identified through genetic screens in yeasts, where the protein kinase Mec1/Rad3 performs key roles in cell cycle checkpoint responses to DNA damage. Subsequent cloning of ATM, the gene defective in the human genetic disease AT, showed that it encodes a protein of the same family. Cells from AT patients (ATM/) display a variety of cell cycle checkpoint defects after exposure to agents that induce DNA DSBs. ATR, identified through its similarity to other members of the family, also performs important cell cycle checkpoint functions. Current models place the ATM/ATR kinases at or close to the sites of primary DNA damage, with ATM responding to DSB and ATR specific for partially single-stranded lesions coated with replication protein A. AT patients suffer from extreme tissue sensitivity to ionizing radiation. As well as defective cell cycle arrest on exposure to agents that induce DNA DSB, cells lacking ATM function exhibit a corresponding hypersensitivity to such agents, as measured by cell survival assays. It is therefore perhaps surprising that such cells have also been shown to be defective in DDIA, either in culture or in vivo (Duchaud et al., 1996; Westphal et al., 1997). Indeed the ataxia that is one of the hallmarks of AT has been attributed to a failure to eliminate by apoptosis cells that have sustained DNA damage during neural development (Lee et al., 2001). Clearly, in this case at least, radiosensitivity in vivo cannot be correlated directly with the extent of DDIA. Extensive genetic and biochemical data have identified further conserved components acting in concert with ATM/ATR and the corresponding kinases in unicellular eukaryotes. In yeasts, these include a replication factor C (RF-C)-related protein, which is thought to load a PCNA-related heterotrimeric sliding clamp onto damaged DNA. Each of these proteins is

required for transduction of the DNA damage signal to downstream protein kinases essential for the execution of cell cycle checkpoint arrest (Norbury and Hickson, 2001). All of these components are conserved from yeasts to human cells, and for the sake of simplicity here we will use the human nomenclature for the RF-C related protein (RAD17), the sliding clamp (RAD1, RAD9 and HUS1, otherwise known as the 9-1-1 complex) and the downstream kinases (CHK1 and CHK2). Activation of the latter is brought about by their direct phosphorylation by the ATM/ATR protein kinases (Matsuoka et al., 1998; Ahn et al., 2000), but loading of the 9-1-1 complex is independent of ATM/ ATR function (Roos-Mattjus et al., 2002). In the yeast models, loss of function of any one of these proteins results in cell cycle checkpoint defects, and some components of this checkpoint pathway have additional functions in, for example, stabilizing stalled replication forks. While these DNA damage signal transducers have retained their cell cycle checkpoint roles in animal cells, like ATM/ATR they have also acquired important roles in DDIA. CHK2-deficient mice, for example, exhibit reduced radiation-induced apoptosis in neurons, thymocytes and splenocytes (Takai et al., 2002). Further targets of ATM and ATR have evolved more recently and lack direct equivalents in the yeast models, yet are important for the efficient transduction of DNA damage signals to downstream effectors. These targets include H2AX, a variant histone specifically and rapidly phosphorylated within its C-terminal tail by both ATM and ATR in the nuclear microenvironment adjacent to DNA damage (Fernandez-Capetillo et al., 2002). Phospho-H2AX is thought to be involved in the concentration of repair and signalling proteins near DNA lesions. One such protein is BRCA1, which is also phosphorylated by ATM in response to DNA DSB (Cortez et al., 1999). Under these circumstances BRCA1, best known as a tumour suppressor in familial breast cancer, forms nuclear foci that contain a variety of proteins involved in mismatch repair, homologous recombination and checkpoint signalling (Wang et al., 2000). BRCA1 is required for a specific subset of ATM/ ATR-dependent phosphorylation events, including phosphorylation of p53 (Foray et al., 2003). 53BP1, yet another ATM target, interacts with phospho-H2AX, CHK2, p53 and BRCA1, and is required for CHK2 activation, BRCA1 phosphorylation, p53 stabilization and BRCA1 focus formation after DNA damage (Fernandez-Capetillo et al., 2002; Wang et al., 2002). While the details of these various interactions are complex and incompletely understood, many of these components clearly have the capacity to influence the efficiency with which proapoptotic signals are transduced to p53, as well as governing the integrity of cell cycle checkpoints and repair pathways. Indeed, mice deficient for H2AX exhibit defects in the efficiency and/ or fidelity of DNA DSB repair which, in the absence of p53 function, lead to massive genetic instability and a dramatically increased cancer incidence (Bassing et al., 2003; Celeste et al., 2003). Furthermore, the kinetics of H2AX phosphorylation suggest that this chromatin Oncogene


2800

modification could be induced by apoptotic DNA fragmentation itself (Rogakou et al., 2000). Extension of the ATM/ATR paradigm might suggest the existence of a battery of damage sensors that could recognize all possible DNA lesions, either directly or indirectly after lesion processing. The recent discovery of a chromatin-derived signal linking nuclear DNA damage to mitochondria via the linker histone H1.2 suggests an alternative and more parsimonious model for damage signalling in DDIA (Konishi et al., 2003). This histone H1 isoform was identified through biochemical fractionation of rat thymus cytosol as a factor mediating cytochrome c release from mitochondria following X-irradiation, or exposure to other agents that induce DNA DSB. H1.2, but not other H1 isoforms, in recombinant form was able to trigger this apoptosis event in vitro, and endogenous H1.2 was released from the nucleus, along with all other H1 isoforms, following DNA damage in vivo. Downregulation of H1.2 using small interfering (si) RNA or gene deletion led to increased cell viability over a wide range of radiation doses. Although the mechanism of H1.2induced cytochrome c release is unclear, this release is correlated with activation of Bak and is dependent on multidomain proapoptotic Bcl-2 family members. This appealing model removes the requirement to postulate multiple damage sensors, since a variety of primary lesions could be sensed indirectly following the release of H1.2 into the cytoplasm. p53-mediated apoptosis in response to DNA damage The key tumour suppressor p53 was the first determinant of susceptibility to DDIA to be identified (Clarke et al., 1993; Lowe et al., 1993b). In vertebrate cells p53 also has well-characterized roles in cell cycle checkpoints, providing another example of the overlap between the pathways controlling alternative DNA damage responses. p53 is functionally conserved in simpler multicellular animals, but is apparently absent from unicellular species. Although not required for developmentally regulated apoptosis, and hence not identified in early screens for cell death genes, the p53 homologue in C. elegans (cep-1) has since been shown to be required for DDIA, as well as for chromosome segregation in the germ line (Derry et al., 2001; Schumacher et al., 2001). In Drosophila, p53 is required specifically for DDIA, but not for cell cycle checkpoint activation (Sogame et al., 2003). Together, these data suggest that the primordial role of p53 was probably more closely related to apoptosis than to cell cycle arrest. A detailed account of the role and regulation of p53 in apoptosis is beyond the scope of this paper, but excellent, more extensive reviews are available elsewhere (Shen and White, 2001; Vousden and Lu, 2002). There is a broad consensus that the primary physiological role of p53 in DDIA is to act as a transcriptional activator of genes encoding apoptotic effectors. The DNA-binding specificity of p53 is conserved from C. elegans to human cells (Schumacher Oncogene

et al., 2001), but the identities of the key target genes activated in p53-dependent DDIA vary markedly among species, and even among different human cell types. In Drosophila, these targets include the apoptotic regulators reaper and sickle (Sogame et al., 2003), which encode negative regulators of inhibitor of apoptosis (IAP) proteins. No homologous genes have been identified among p53 targets in human cells, though human p53 can effectively neutralize IAP proteins indirectly through triggering release of apoptotic effectors from mitochondria. To this end, human p53 directly activates transcription of several genes encoding members of the Bcl-2 family. This diverse family includes multidomain pro- and anti-apoptotic proteins as well as a variety of ‘BH3-only’ proteins that together regulate mitochondrial permeability (Cory and Adams, 2002). The p53 targets considered most important in this respect are the proapoptotic Bax and the BH3-only proteins Noxa and PUMA (Miyashita and Reed, 1995; Oda et al., 2000; Nakano and Vousden, 2001). In mammalian cells, p53 also activated transcription of Fas following DNA damage, while Fas ligand expression was induced independently of p53 (Owen-Schaub et al., 1995; Muller et al., 1998). These data raise the surprising notion that nuclear DNA damage signals can be transduced via cell surface receptors. This mechanism renders cells with damaged DNA more susceptible to Fas ligand-induced apoptosis and cytotoxic T lymphocyte-mediated killing (Chakraborty et al., 2003). Nonetheless, p53-mediated expression of FAS or any other transcriptional target of p53 is insufficient on its own to induce DDIA (Reinke and Lozano, 1997). Aside from transcriptional activation, several other roles for p53 have been identified that could contribute to its ability to induce DDIA. The sequence-specific DNA binding activity of p53 can mediate transcriptional repression, and p53-mediated downregulation of transcription of genes including the IAP-family protein member survivin has been reported (Zhou et al., 2002). In addition p53 itself has been shown to bind DNA strand breaks (Bakalkin et al., 1994), suggesting a possible role in DNA damage detection, or to translocate to mitochondria specifically during p53-dependent apoptosis (Marchenko et al., 2000; Mihara et al., 2003). Once at the mitochondrial outer membrane, p53 appears to antagonize the antiapoptotic Bcl-2 and Bcl-XL proteins directly by binding to them. Intriguingly, tumour-derived mutant forms of p53 that had previously been shown to be defective in sequence-specific DNA binding were also defective in binding to Bcl-2 and Bcl-XL. These findings are consistent with early demonstrations that p53-mediated DDIA does not necessarily involve new protein synthesis (Caelles et al., 1994), and suggest that the central importance of p53 transcriptional activation in DDIA should probably be re-evaluated. Functional p53 was also found to be required for the specific deamidation of two asparagine residues in Bcl-XL, neutralizing its ability to block the action of proapoptotic proteins such as Bax (Deverman et al., 2002), but it is currently unclear whether this also reflects a direct p53-Bcl-XL interaction. Interestingly, as


2801

was mentioned above, the release of histone H1.2 and other histone H1 isoforms from the nucleus after DNA damage was also p53-dependent (Konishi et al., 2003). It remains to be seen if the H1.2-releasing function of p53 requires its activity as a transcription factor, or one of the alternative activities outlined above, though H1.2 release was found to be upstream of caspase activation (Konishi et al., 2003). Activation of p53 by ATM/ATR/CHK2 The first hint of a connection between the ATM/ATRdependent pathways and p53 predated the identification of ATM itself, and derived from the observation that cells from AT patients showed delayed accumulation of p53 protein after exposure to ionizing radiation (Lu and Lane, 1993). The biochemical basis for this connection is thought to involve ATM-dependent multisite phosphorylation of p53 (Caspari, 2000; Saito et al., 2002), and of MDM2, preventing both MDM2-mediated inhibition of p53 DNA binding and its ubiquitin-dependent proteolysis (Khosravi et al., 1999). Downstream from ATM/ ATR, CHK2 can in turn phosphorylate p53 at Ser20, again potentially blocking the MDM2–p53 interaction and modulating the DNA-binding activity of p53 (Caspari, 2000). The relative contributions of these various phosphorylation events to physiological DDIA have been difficult to assess, but valuable clues once again come from model systems. In Drosophila, expression of a dominant-negative mutant form of Chk2 inhibited p53-mediated DDIA (Peters et al., 2002). Evidence from Chk2 knockout mice further supports the idea that Chk2 is important for regulation of p53 transcriptional activity, as well as stability (Takai et al., 2002). In addition to the negative regulation provided by MDM2, the activity of p53 is limited by association with iASPP (inhibitory member of the ASPP family), which is conserved from C. elegans to human cells (Bergamaschi et al., 2003). The steady-state level of this p53 inhibitor may set the threshold level of damage required for DDIA, since overexpression of iASPP conferred relative resistance to UV- and cisplatin-induced apoptosis. Further connections between checkpoint proteins and apoptosis The conserved DNA damage-signalling pathway downstream from ATM/ATR has been exploited in several further ways during the evolution of apoptotic regulation (Figure 2). The induction of p53-independent DDIA has been attributed to CHK2 acting through the nuclear PML (promyelocytic leukaemia) protein (Yang et al., 2002), although the downstream effectors of this pathway have yet to be described. CHK2 has also been found to phosphorylate and activate the E2F-1 transcription factor specifically after DNA damage (Stevens et al., 2003). E2F-1 activity determines the level of expression of a variety of apoptotic effectors in

Figure 2 An overview of DNA damage signalling in apoptosis. Early damage sensing in the nucleus involves the ATM and ATR protein kinases, the RAD1-RAD9-HUS1 (9-1-1) complex, and their downstream effector CHK2. Note that these components are also required for cell cycle checkpoint and DNA repair responses (not shown). Once CHK2 is activated, the signalling processes can be grouped into p53-dependent (left) and p53-dependent events (right). The activities of p53 include transcriptional activation of the genes encoding Bax, Noxa, PUMA and Fas, as well as direct effects on mitochondrial permeabilization and mediating the release of histone H1.2 from the nucleus. The proapoptotic effects of Bax are antagonized by the release of nuclear Ku70 into the cytoplasm. The p53-independent pathways include CHK2mediated signalling to PML, and redirection of E2F-1 towards proapoptotic transcriptional target genes, including those encoding p73 and procaspases. ATM also phosphorylates c-Abl, which promotes the neutralization of the antiapoptotic Bcl-2 and Bcl-XL by RAD9. Caspase-2 and Nurr77 transduce p53-independent damage signals from the nucleus to mitochondria in less welldefined ways. For further details, see text

Drosophila and human cells (Nahle et al., 2002; Zhou and Steller, 2003), and after DNA damage the specificity of E2F-1 was found to shift in favour of selected targets including the gene encoding the p53-related protein p73 (Pediconi et al., 2003). The reduced DDIA seen in cells lacking CHK2 therefore reflects defective p53-dependent and p53-independent death pathways. Components of the 9-1-1 complex also serve to couple DNA damage signalling to apoptotic effectors. In C. elegans the RAD1 homologue MRT-2 was found to be required for DDIA in addition to the core apoptotic regulators CED-4 and CED-3 (Gartner et al., 2000). In the same system, HUS1 was relocalized within the nucleus after DNA damage, and was required for the CEP-1(p53)-induced transcriptional induction of egl-1, which encodes a proapoptotic BH3-only protein (Hofmann et al., 2002). Damage signals relayed independently by the ATM/ATR and 9-1-1 pathways therefore appear to be integrated via CHK2 at the level of p53 activation. More direct involvement of 9-1-1 proteins in regulation of mitochondrial permeability may also be important in DDIA. Intriguingly, the RAD9 component of this complex contains a BH3-like domain that is conserved from yeast to human cells (Komatsu et al., 2000a), indicating that this structural motif is more ancient than the Bcl-2 family itself. Human RAD9, which was found by coimmunoprecipitation to interact Oncogene


2802

with the antiapoptotic Bcl-2 and Bcl-XL proteins, but not with related proapoptotic proteins, induced apoptosis when overexpressed (Komatsu et al., 2000b). It is currently unclear whether RAD9 involved in such interactions would also participate in 9-1-1 complex formation and, if so, how the nuclear 9-1-1 might communicate with Bcl-2 family members, presumably at the mitochondria. RAD9 was identified independently as a significant target of the c-ABL tyrosine kinase, with which it is also apparently stably associated (Yoshida et al., 2002a). The c-Abl-mediated phosphorylation of RAD9 occurred within the BH3 domain and promoted the association between RAD9 and Bcl-XL, in response to DNA damage. Since c-Abl activation had previously been shown to be dependent on ATM (Shafman et al., 1997), these findings establish a further pathway linking DNA damage sensing to regulation of Bcl-2 family members, and hence mitochondrial permeability. DNA-dependent protein kinase A further member of the ATM/ATR family, DNAdependent protein kinase (DNA-PK), has no obvious counterpart in unicellular eukaryotes but appears to contribute to apoptotic regulation in mammals. DNAPK is a heterotrimeric nuclear enzyme consisting of a 470 kDa catalytic subunit (DNA-PKcs) and the regulatory DNA binding subunits, Ku70 and Ku80 (Durocher and Jackson, 2001). Cells deficient in DNA-PKcs, Ku70 or Ku80 are hypersensitive to ionizing radiation and defective in V(D)J recombination, suggesting a primary role for this kinase in DSB repair. Binding of the Ku heterodimer to DSB protects the DNA ends from degradation, and serves to recruit the catalytic subunit and additional factors essential for successful end joining. A role for DNA-PK in p53-mediated death has been suggested (Woo et al., 2002), although the nature of this role is currently unclear. Ku70 was found to accumulate following ionizing irradiation, and binding of nuclear clusterin, also known as TRPM-2, SGR-2 or XIP8, to Ku70 correlated with dramatically reduced cell growth, colony-forming ability and increased cell death (Yang et al., 2000). On the other hand, recent studies have identified associations between cell death and loss of nuclear Ku70 (Song et al., 2003) and/or cytosolic accumulation of this protein (Um et al., 2003). In certain situations, Ku70 is even constitutively cytosolic (Fewell and Kuff, 1996). Thus, it seems that Ku70 has a dual function in DSB repair and apoptotic regulation. Cytosolic Ku70 can bind to Bax and neutralize its proapoptotic function (Sawada et al., 2003). Overexpression of Ku70-suppressed DDIA and Bax-induced apoptosis; conversely, Bax-mediated apoptosis was enhanced by downregulation of Ku70. The ability of Ku70 to protect cells from apoptotic stimuli that do not induce DNA DSB probably reflects this cytosolic interaction with Bax. Conformational changes of Bax and/or Bak are essential for apoptotic permeabilization of mitochondria (Wei et al., 2001). However, Ku70 appears specifically to control Bax-mediated killing. If the Oncogene

apoptotic signal is sufficiently strong, Ku70 is released from the complex with Bax. As a result, the aminoterminal region of Bax is exposed, allowing the protein to form a large, transient complex, the so-called ‘baxosome’ that includes Bax itself, BH3-only proteins, membrane lipid, cardiolipin, and possibly other unidentified proteins. Bax then undergoes conformational changes that allow it to be inserted into the outer mitochondrial membrane and to oligomerize, resulting in the release of mitochondrial apoptotic factors to the cytosol. The concomitant upregulation of Ku70 and Bax after DNA breakage could favour the formation of BaxKu70 complexes, which in turn could be central to the decision of cells to repair the damage and survive, or to die. However, it is still unclear whether there are any links between upregulation of Ku70 and other p53activated proteins such as Noxa and PUMA. Interestingly, downstream from the mitochondrial apoptotic events, activated caspase-3 cleaves another protein from DNA-PK complex, namely DNA-PKcs, thus destroying a key component of the cellular repair machinery. DAXX: a nuclear regulator of p53 and other apoptotic effectors DAXX, initially identified as a protein associated with the death domain of Fas (Fas Death Domain-Associated protein XX), promotes Fas-induced cell death (Yang et al., 1997). However, several lines of evidence suggest that this activity does not involve a direct association between the two proteins. Most importantly, DAXX was found to be exclusively nuclear and, upon induction of apoptosis by Fas, DAXX did not translocate out of the nuclei (Torii et al., 1999). Moreover, DAXX was localized to subnuclear structures known as PODs (PML-Oncogenic Domains), which are nuclear bodies primarily defined by their constituent proteins, most notably PML itself. PML-deficient mice display resistance to apoptosis induced by a variety of stimuli, including DNA damage, while PML overexpression can induce cell death (Wang et al., 1998). In this light, it is interesting that UV-induced apoptosis, which requires activation of the protein kinase JNK, was blocked by a dominant-negative form of DAXX (Wu et al., 2002). DAXX directly binds to and activates the apoptosis signal-regulating kinase 1 (ASK1), a mitogen-activated protein kinase kinase kinase that activates JNK (Chang et al., 1998). It was suggested that ASK1 might translocate DAXX to the cytoplasm, and that binding to DAXX could induce a conformational change within ASK1 to activate its kinase activity. However, DAXX was also reported to promote ASK1-mediated apoptosis in a kinase- and caspase-independent manner (Charette et al., 2001). The antiapoptotic protein HSP27 may inhibit cytosolic translocation of DAXX and block the DAXX-ASK1 interaction (Charette et al., 2000). In addition, certain mutant forms of p53 may promote cell survival by inhibiting the DAXX-ASK1 pathway (Ohiro et al., 2003). Yeast two-hybrid studies indicated that both p73a and its relative p53 can also associate with


2803

DAXX (Kim et al., 2003). This interaction involves the oligomerization domain of p73a or p53 and the Cterminal domain of DAXX, which is also the region that binds ASK1 and PML. Interestingly, transient coexpression of DAXX resulted in strong inhibition of p73and p53-mediated transcriptional activation of p53responsive promoters. It is therefore possible that DAXX modulates cell cycle arrest and apoptosis through regulation of the transcriptional activity of p53 and its family members. Is DAXX normally pro- or anti-apoptotic? Targeted deletion of Daxx in the mouse results in severe developmental abnormalities in knockout embryos, as a result of unscheduled apoptosis during early embryogenesis (Michaelson et al., 1999). Similarly, Daxx/ embryonic stem cells and siRNA-treated DAXX depleted cells are characterized by elevated levels of apoptosis (Michaelson et al., 1999; Michaelson and Leder, 2003). These and other data suggest that DAXX acts as antiapoptotic regulator, probably via suppression of proapoptotic gene expression. DAXX silencing might sensitize cells to multiple apoptotic pathways, including DDIA (Chen and Chen, 2003). Nuclear FADD: a further regulator of the response to DNA damage? Fas-associated death domain protein (FADD) is an adaptor protein that bridges death receptors with initiator caspases. FADD was at first considered to be confined to the cytoplasm, since most studies focused on its interaction with plasma membrane proteins. However, several groups recently demonstrated that FADD is mainly nuclear (Gomez-Angelats and Cidlowski, 2003; Screaton et al., 2003; Sheikh and Huang, 2003). It was suggested that sequestration of FADD in the nucleus might ensure that the death receptor pathway of apoptosis is not constitutively activated. However, it seems that FADD may have a distinct role in the nucleus, where it interacts with the methyl-CpG binding domain protein 4 (MBD4), which excises thymine from GT mismatches in methylated regions of chromatin (Screaton et al., 2003). The MBD4-interacting mismatch repair factor MLH1 was also found in a complex with FADD. Consistent with the idea that MBD4 can signal to an apoptotic effector, MBD4 regulated DDIA. Interestingly, another death domain-containing protein LRDD, also known as PIDD, interacts with FADD (Lin et al., 2000). LRDD is transcriptionally induced in a p53-dependent manner following exposure to DNA damaging agents. Together these data suggest that the nuclear localization of FADD and its interaction with a genome surveillance/DNA repair protein might be involved in regulation of DDIA. Nur77: a p53-independent response to DNA damage The immediate early gene Nur77 (also known as NGF1B and TR3), which encodes an orphan nuclear receptor,

is rapidly induced by various stress stimuli, including DNA damage. It was initially suggested that DNA binding and transactivation by Nurr77 might be required for its proapoptotic effect. More recent data showed that even a form of Nurr77 lacking DNAbinding ability could potently induce p53-independent apoptosis in some tumour cells. Although normally nuclear, in response to etoposide and 5-fluorouracil Nur77 was translocated to the cytosol where it was implicated in the release of mitochondrial cytochrome c (Li et al., 2000; Wilson et al., 2003). It is unclear whether Nur77 directly targets mitochondria or acts through interaction with Bax and relocalization of Bax from the cytosol to mitochondria. Nuclear Nur77 was previously implicated in cell proliferation, so the cytoplasmic relocalization of Nur77 might play a key role in coordinating growth and apoptotic pathways through the cessation of the transcriptional program mediated by this protein after its translocation from the nucleus. Role of caspase-2 in DNA damage responses Following DNA damage, mitochondrial release of cytochrome c and the subsequent activation of procaspase-9 are essential for activation of downstream apoptotic effectors. In addition to the contributions made by histone H1.2 release and other p53-mediated pathways reviewed above, a parallel line of investigation has led to the notion that the link between nuclear DNA damage responses and mitochondrial events also involves caspase activity. A low dose of etoposide, an inhibitor of nuclear topoisomerase II that induces DNA strand breaks, resulted in the release of a heat-labile factor(s) that, in turn, interacted with mitochondria to elicit cytochrome c release (Robertson et al., 2000). This activity was inhibited by the general caspase inhibitor, zVAD-fmk, and caspase-2 was the earliest caspase activated in treated cells. This and subsequent studies showed that, in response to DNA damage, activation of caspase-2 is indeed required before mitochondrial permeabilization and apoptosis can take place (Guo et al., 2002; Lassus et al., 2002; Paroni et al., 2002; Robertson et al., 2002). Subcellular fractionation studies revealed that, although procaspase-2 is present in several intracellular compartments, including the Golgi, cytosol and nucleus, it is the only procaspase present constitutively in the nucleus (Zhivotovsky et al., 1999; Mancini et al., 2000). Nuclear localization of caspase-2 is strictly dependent on the presence of the prodomain (Colussi et al., 1998). Caspase-2 has been shown to associate via its prodomain with RAIDD, a death adaptor molecule that is thought to be involved in death receptormediated apoptosis through an association with the adaptor protein RIP and TRADD. Endogenous RAIDD is mostly localized to the cytoplasm, but in cells ectopically expressing caspase-2, a fraction of RAIDD is recruited to the nucleus (Shearwin-Whyatt et al., 2000). More recent observations revealed that caspase-2 is spontaneously recruited to a large protein complex and that this recruitment is sufficient to Oncogene


2804

mediate caspase-2 activation. No RAIDD was detected in these high molecular weight fractions, ruling out the involvement of RAIDD in caspase-2 complex formation (Read et al., 2002). Using substrate-binding assays the authors provide clear evidence that caspase-2 activation might occur without processing of the precursor molecule. However, oligomerization is important step for caspase-2 activation. Yeast two-hybrid system experiments with dominant-negative caspase-2 as bait revealed the presence of a novel proapoptotic molecule named proapoptotic caspase adaptor protein (PACAP) (Bonfoco et al., 2001; Lassus et al., 2002). In cell extracts and in cotransfection experiments, this protein bound to caspase-2 and -9 but not to caspase-3, -4, -7 or -8. Transient overexpression of PACAP-triggered apoptosis, which was prevented by inhibition of caspase activity. However, it is unclear whether this new adaptor protein is important for caspase-2 complex formation. A further study suggested that cyclin D3 is involved in activation of caspase-2 (Mendelsohn et al., 2002). Expression of cyclin D3 and caspase-2 in human cells potentiated apoptosis compared with expression of caspase-2 alone. Moreover, overexpression of cyclin D3 increased the amount of cleaved caspase-2. It was suggested that interaction with cyclin D3 may stabilize caspase-2. This interaction may be just one way in which the baseline sensitivity of proliferating cells to apoptotic stimuli is increased, relative to that of the equivalent quiescent populations. A second such mechanism involves the direct transcriptional activation of procaspase genes by E2F-1 on cell cycle re-entry (Nahle et al., 2002). Following activation in response to DNA damage, caspase-2 is involved in the release of apoptotic effectors from mitochondria. Cells stably expressing antisense procaspase-2, or the corresponding transiently expressed siRNA were refractory to cytochrome c release and as well as downstream events, such as procaspase-9 and -3 activation, loss of phosphatidylserine asymmetry in the plasma membrane, and DNA fragmentation, induced by DNA damage (Lassus et al., 2002; Robertson et al., 2002). Expression of a caspase-2 cDNA variant resistant to siRNA restored the ability of cells to undergo apoptosis. How does caspase-2 act in this sequence of events? During early apoptosis nuclear caspase-2 may trigger mitochondrial dysfunction without relocalizing into the cytoplasm (Paroni et al., 2002). This view is based on the observation that release of cytochrome c occurred in the absence of obvious changes in nuclear/ cytoplasmic transport, while caspase-2 was able to diffuse out of the nucleus only when the size limit for passive diffusion through nuclear pores had increased later in apoptosis. In any case, it is clear that caspase-2 activity is required for translocation of Bax to the mitochondria, as well as for release of mitochondrial cytochrome c and Smac. Two groups demonstrated that caspase-2 can induce release of these proteins from mitochondria directly (Guo et al., 2002; Robertson et al., 2002), or through cleavage of the proapoptotic protein Bid, which moves to mitochondria and facilitates cytochrome c release (Guo et al., 2002). These data Oncogene

suggest that caspase-2 is an apical caspase in the proteolytic cascade initiated by DNA damage and that it plays a key role in transduction of the DNA damage signal to the mitochondria. It seems that the role of caspase-2 in apoptosis is cell type-specific. Thus, neurons from caspase-2 knockout mice are more prone to apoptosis than neurons from wild-type mice, whereas caspase-2-deficient oocytes or lymphoblasts from the same animals are resistant to apoptosis induced by chemotherapeutic drugs (Bergeron et al., 1998). Perhaps the upstream role assigned in some cells to caspase-2 is performed by another caspase in neurons. It is clear that many questions in this area have yet to be resolved. Mitochondrial regulation of DNA damage-induced apoptosis Release of mitochondrial endonuclease G/CPS-6 and the oxidoreductase apoptosis-inducing factor (AIF/ WAH-1) is important for chromatin fragmentation during the execution phase of apoptosis, and is a feature of the process conserved from C. elegans, where it occurs downstream from caspase activation, to human cells, where its relation to caspase activation is less clearcut (Susin et al., 1999; Parrish et al., 2001; Arnoult et al., 2003). Release of mitochondrial components into the cytosol and nucleus has also assumed a more general role in the amplification of apoptotic signals, including those activated by DNA damage, in vertebrates (Waterhouse et al., 2002). The best known of these components is cytochrome c, which on permeabilization of the outer mitochondrial membrane binds the apoptotic protease activating factor APAF-1 and promotes caspase activation. Smac/Diablo is also released on mitochondrial permeabilization and blocks the action of IAP proteins that would otherwise inhibit caspases. Cytochrome c and Smac/Diablo release can both be caspase dependent following DNA damage, indicating that these are primarily amplification mechanisms, rather than routes by which caspase activation is initially brought about (Adrain et al., 2001; Robertson et al., 2002). Does mitochondrial DNA damage play a role in cell death? Ionizing radiation and some anticancer drugs induce damage both in genomic and in mitochondrial DNA (mtDNA). Although the role of nuclear DNA damage in initiation of cell death has been discussed extensively, the involvement of mtDNA is this process is controversial. Damage to mtDNA, if not repaired, could lead to disruption of the electron transport chain and the production of reactive oxygen species (ROS). For some years, it has been known that mitochondrial and nuclear DNA can be affected differently by DNA damaging drugs. Detailed investigation of DNA integrity in mitochondria suggested that DNA fragmentation during apoptosis is usually specifically nuclear (Murgia


2805

et al., 1992). Furthermore, when human leukaemia cell lines were exposed to conditions which resulted in necrosis, mtDNA was damaged at approximately the same rate as nuclear DNA, but in apoptosis mtDNA was not degraded (Tepper and Studzinski, 1993). At this time, it was believed that DNA degradation occurred early after the delivery of the ‘lethal hit’ and before intracellular organelles and the plasma membrane were damaged (Bursch et al., 1990). It is now clear that DNA fragmentation is a relatively late event, so these data did not address the question of whether mtDNA and its damage might have a role to play in early apoptotic signalling. Not surprisingly, the human mtDNA depletion syndrome is lethal. However, cells without mtDNA can be maintained in long-term culture if they are provided with substrates for the glycolytic production of ATP. These so-called r0 cells undergo apoptosis in response to staurosporine in the same way as the parental cells (Jacobson et al., 1993). Even though these cells are defective for respiration, their apoptosis was characterized by cytochrome c release, caspase-3 activation and loss of mitochondrial membrane potential (Jiang et al., 1999). Moreover, Bcl-2 overexpression was able to protect both parental and r0 cells from apoptosis. In contrast to staurosporine-induced apoptosis, death of leukaemia cells in response to tumour necrosis factor (TNF) required mtDNA (Higuchi et al., 1997). However, in some cell lines TNF-a was able to induce apoptosis despite the absence of mtDNA (Marchetti et al., 1996), although with much slower kinetics than in the parental cells. A significant difference in sensitivity of hepatocellular carcinoma cells to TNF-related apoptosis-inducing ligand (TRAIL) was found on comparison of a r0 and parental cell line (Kim et al., 2002). After TRAIL treatment, translocation of Bax, subsequent cleavage of Bid, release of cytochrome c and dissipation of mitochondrial transmembrane potential were all seen in the parental line but not in r0 cells. However, parental and r0 cells did not show differential susceptibility to agonistic anti-Fas antibody, TNF-a or staurosporine. To what extent are the effects of mtDNA explicable in terms of ROS production? Apoptotic r0 cells, unlike apoptotic parental cells, did not exhibit a change in the redox potential of glutathione (Jiang et al., 1999). This suggests that a change in redox potential or an increase in ROS generation does not play an essential role in apoptotic induction following treatment with staurosporine. Several observations revealed that oxidative stress induces damage in both nuclear and mtDNA and cells lacking mtDNA exhibit a marked resistance to cell death (Yoneda et al., 1995). Attempts to determine whether loss of mtDNA and disturbance in respiratory chain function result in apoptosis in vivo were also undertaken (Wang et al., 2001). Using embryos with homozygous disruption of the gene encoding the mitochondrial transcription factor Tfam, as well as tissue-specific Tfam knockout animals with severe respiratory chain deficiency in the heart, the authors showed massive apoptosis in both experimental systems.

These data provide in vivo evidence that respiratory chain deficiency predisposes cells to apoptosis. Interestingly, Tfam is essential not only for mitochondrial gene expression, but also for maintenance and repair of mtDNA. It was shown that Tfam preferentially recognizes cisplatin-damaged and oxidized DNA (Yoshida et al., 2002b). More detailed investigation revealed that binding of Tfam to cisplatin-modified DNA was significantly enhanced by p53, with which Tfam was found to interact physically (Yoshida et al., 2003). Thus, although the role of mtDNA in cell death is contentious, it is clear that its absence or impaired function can profoundly influence the rate of apoptosis. Mutations in the mtDNA leading to mitochondrial dysfunction have been reported in a variety of cancers. Might these mutations influence cellular responses to cancer therapy? In comparison with their r þ parents, r0 cells were extremely resistant to adriamycin, photodynamic therapy or ionizing radiation, as judged by clonogenic survival assays (Singh et al., 1999; Tang et al., 1999). Importantly, resistance to adriamycin, which induces ROS formation as well as acting as a topoisomerase II inhibitor, was not due to the lack of drug uptake, or to altered cell cycle responses. Thus, mitochondrial function is a determinant of cellular sensitivity to important cancer therapeutic agents. Back to the nucleus: DNA damage-induced histone modifications in apoptosis The data reviewed above concerning the involvement of histones H1.2 and H2AX in DNA damage signalling were by no means the first suggestion of a connection between histones and apoptosis. Compared with native chromatin, apoptotic oligonucleosomal fragments contained less histone H1 (Borisova et al., 1984; Bell et al., 1990). Interestingly, a rapid increase in poly(ADPribose) polymerase (PARP) activity was observed upon induction of apoptosis by radiation, followed by the poly(ADP-ribosyl)ation of histone H1 (Manome et al., 1993). Inhibition of PARP activity by nicotinamide or similar compounds inhibited DNA fragmentation. This histone modification increases the accessibility of chromatin to nucleases and probably promotes DNA fragmentation during apoptosis. On the other hand, histone H1.2 did not undergo any obvious posttranslational modifications following DNA damage, and the nuclear and cytosolic forms of H1.2 were indistinguishable (Konishi et al., 2003). Moreover, no modifications in H1.2 were required for mitochondrial cytochrome c release in vitro. In certain experimental systems, induction of apoptosis was accompanied by histone H1 phosphorylation (Lee et al., 1999) and activation of CDC2/CDK1 (Meikrantz et al., 1994), events normally associated with entry into mitosis. These data suggest that apoptotic cells could be entering premature, and therefore catastrophic, mitosis in response to DNA damage. However, CDC2/CDK1 activation was not required for apoptosis in thymocytes (Norbury et al., 1994), and Oncogene


2806

inactivation of CDC2/CDK1 even enhanced DDIA under some circumstances (Ongkeko et al., 1995). Broadly speaking, similar data have been obtained for histone H3, which is phosphorylated during mitosis and premature chromosome condensation, but not during apoptosis. Histone H2AX phosphorylation, as noted above, plays a role in DNA damage signalling, but has no known direct connection with the chromatin changes that accompany apoptosis. Perhaps the most clearcut case for direct involvement of histone phosphorylation in DDIA concerns mammalian histone H2B, which is phosphorylated at Ser14 in response to UV irradiation or treatment with the topoisomerase II inhibitor etoposide, which induces DNA strand breaks. This phosphorylation is mediated by mammalian sterile twenty (Mst1) kinase (Cheung et al., 2003), activation of which is dependent upon its cleavage by caspase-3. Histone H2B phosphorylation occurred immediately prior to DNA laddering and, like H1 poly(ADP-ribosyl)ation, may be involved in triggering this process.

Conclusions The wealth of recent literature has brought this field to a stage where it is now possible to begin to explain cellular sensitivity to DDIA in terms of a cellular apoptotic ‘rheostat’. This comprises elements of the DNA damage detection machinery, including proteins that also signal cell cycle checkpoint arrest, signal transducers including CHK2, p53 and E2F-1, as well as the downstream caspases and their regulators (Figure 2). The setting of the rheostat is determined in part by the abundance and

basal activities of these various components, in part by their upregulation during cell proliferation and in part by the in-built amplification loops that operate via Fas and the release of mitochondrial regulators of apoptosis. Multiple independent routes have now been traced by which nuclear DNA damage can be signalled to the mitochondria. These encompass p53-dependent transcriptional activation, less well-characterized direct mitochondrial roles for p53, histone H1.2 release, Nurr77 and caspase-2 activation. A degree of coordination between DNA repair and cellular commitment to apoptosis may be achieved through the release of Ku70, conventionally considered to be a nuclear DNA repair protein, into the cytoplasm. Conversely, DAXX and FADD, initially identified through their association with the integral plasma membrane death receptor Fas, now appear to play important nuclear roles in DDIA. It is also apparent that damage to the mitochondrial genome, as well as nuclear DNA damage, has the capacity to influence the extent of subsequent apoptosis. Each of these events is potentially of crucial importance both in tumour development and in the response of cancer cells to therapies based on the induction of DNA damage. Acknowledgements We apologize to those authors whose work could not be cited directly due to space limitations. The work in our laboratories was supported by grants from Cancer Research UK, the Medical Research Council and the Association for International Cancer Research (to CJN), and from the Swedish (3829B02-07XBC) and Stockholm (03:173) Cancer Societies, and EC grant (QLK3-CT-2002-01956) (to BZ).

References Adrain C, Creagh EM and Martin SJ. (2001). EMBO J., 20, 6627–6636. Ahn JY, Schwarz JK, Piwnica-Worms H and Canman CE. (2000). Cancer Res., 60, 5934–5936. Arnoult D, Gaume B, Karbowski M, Sharpe JC, Cecconi F and Youle RJ. (2003). EMBO J., 22, 4385–4399. Bakalkin G, Yakovleva T, Selivanova G, Magnusson KP, Szekely L, Kiseleva E, Klein G, Terenius L and Wiman KG. (1994). Proc. Natl. Acad. Sci. USA, 91, 413–417. Bassing CH, Suh H, Ferguson DO, Chua KF, Manis J, Eckersdorff M, Gleason M, Bronson R, Lee C and Alt FW. (2003). Cell, 114, 359–370. Baus F, Gire V, Fisher D, Piette J and Dulic V. (2003). EMBO J., 22, 3992–4002. Bell DA, Morrison B and VandenBygaart P. (1990). J. Clin. Invest., 85, 1487–1496. Bergamaschi D, Samuels Y, O’Neil NJ, Trigiante G, Crook T, Hsieh JK, O’Connor DJ, Zhong S, Campargue I, Tomlinson ML, Kuwabara PE and Lu X. (2003). Nat. Genet., 33, 162–167. Bergeron L, Perez GI, Macdonald G, Shi L, Sun Y, Jurisicova A, Varmuza S, Latham KE, Flaws JA, Salter JC, Hara H, Moskowitz MA, Li E, Greenberg A, Tilly JL and Yuan J. (1998). Genes Dev., 12, 1304–1314. Oncogene

Bonfoco E, Li E, Kolbinger F and Cooper NR. (2001). J. Biol. Chem., 276, 29242–29250. Borisova EA, Zhivotovsky BD and Khanson KP. (1984). Radiobiologiya, 24, 607–611 (in Russian). Bursch W, Kleine L and Tenniswood M. (1990). Biochem. Cell Biol., 68, 1071–1074. Caelles C, Helmberg A and Karin M. (1994). Nature, 370, 220–223. Caspari T. (2000). Curr. Biol., 10, R315–R317. Celeste A, Difilippantonio S, Difilippantonio MJ, FernandezCapetillo O, Pilch DR, Sedelnikova OA, Eckhaus M, Ried T, Bonner WM and Nussenzweig A. (2003). Cell, 114, 371–383. Chakraborty M, Abrams SI, Camphausen K, Liu K, Scott T, Coleman CN and Hodge JW. (2003). J. Immunol., 170, 6338–6347. Chang HY, Nishitoh H, Yang X, Ichijo H and Baltimore D. (1998). Science, 281, 1860–1863. Charette SJ, Lambert H and Landry J. (2001). J. Biol. Chem., 276, 36071–36074. Charette SJ, Lavoie JN, Lambert H and Landry J. (2000). Mol. Cell. Biol., 20, 7602–7612. Chen LY and Chen JD. (2003). Mol. Cell. Biol., 23, 7108–7121.


2807 Cheung WL, Ajiro K, Samejima K, Kloc M, Cheung P, Mizzen CA, Beeser A, Etkin LD, Chernoff J, Earnshaw WC and Allis CD. (2003). Cell, 113, 507–517. Clarke AR, Purdie CA, Harrison DJ, Morris RG, Bird CC, Hooper ML and Wyllie AH. (1993). Nature, 362, 849–852. Colussi PA, Harvey NL and Kumar S. (1998). J. Biol. Chem., 273, 24535–24542. Cortez D, Wang Y, Qin J and Elledge SJ. (1999). Science, 286, 1162–1166. Cory S and Adams JM. (2002). Nat. Rev. Cancer, 2, 647–656. Derry WB, Putzke AP and Rothman JH. (2001). Science, 294, 591–595. Deverman BE, Cook BL, Manson SR, Niederhoff RA, Langer EM, Rosova I, Kulans LA, Fu X, Weinberg JS, Heinecke JW, Roth KA and Weintraub SJ. (2002). Cell, 111, 51–62. Di Leonardo A, Linke SP, Clarkin K and Wahl GM. (1994). Genes Dev., 8, 2540–2551. Duchaud E, Ridet A, Stoppa-Lyonnet D, Janin N, Moustacchi E and Rosselli F. (1996). Cancer Res., 56, 1400–1404. Durocher D and Jackson SP. (2001). Curr. Opin. Cell Biol., 13, 225–231. Ellis HM and Horvitz HR. (1986). Cell, 44, 817–829. Fernandez-Capetillo O, Chen HT, Celeste A, Ward I, Romanienko PJ, Morales JC, Naka K, Xia Z, CameriniOtero RD, Motoyama N, Carpenter PB, Bonner WM, Chen J and Nussenzweig A. (2002). Nat. Cell Biol., 4, 993–997. Fewell JW and Kuff EL. (1996). J. Cell Sci., 109, 1937–1946. Foray N, Marot D, Gabriel A, Randrianarison V, Carr AM, Perricaudet M, Ashworth A and Jeggo P. (2003). EMBO J., 22, 2860–2871. Gartner A, Milstein S, Ahmed S, Hodgkin J and Hengartner MO. (2000). Mol. Cell, 5, 435–443. Gomez-Angelats M and Cidlowski JA. (2003). Cell Death Differ., 10, 791–797. Guo Y, Srinivasula SM, Druilhe A, Fernandes-Alnemri T and Alnemri ES. (2002). J. Biol. Chem., 277, 13430–13437. Higuchi M, Aggarwal BB and Yeh ET. (1997). J. Clin. Invest., 99, 1751–1758. Hofmann ER, Milstein S, Boulton SJ, Ye M, Hofmann JJ, Stergiou L, Gartner A, Vidal M and Hengartner MO. (2002). Curr. Biol., 12, 1908–1918. Igney FH and Krammer PH. (2002). Nat. Rev. Cancer, 2, 277–288. Jacobson MD, Burne JF, King MP, Miyashita T, Reed JC and Raff MC. (1993). Nature, 361, 365–369. Jacobson MD, Burne JF and Raff MC. (1994). EMBO J., 13, 1899–1910. Jiang S, Cai J, Wallace DC and Jones DP. (1999). J. Biol. Chem., 274, 29905–29911. Kerr JF, Wyllie AH and Currie AR. (1972). Br. J. Cancer, 26, 239–257. Khosravi R, Maya R, Gottlieb T, Oren M, Shiloh Y and Shkedy D. (1999). Proc. Natl. Acad. Sci. USA, 96, 14973–14977. Kim EJ, Park JS and Um SJ. (2003). Nucleic Acids Res., 31, 5356–5367. Kim JY, Kim YH, Chang I, Kim S, Pak YK, Oh BH, Yagita H, Jung YK, Oh YJ and Lee MS. (2002). Oncogene, 21, 3139–3148. Komatsu K, Hopkins KM, Lieberman HB and Wang H. (2000a). FEBS Lett., 481, 122–126. Komatsu K, Miyashita T, Hang H, Hopkins KM, Zheng W, Cuddeback S, Yamada M, Lieberman HB and Wang HG. (2000b). Nat. Cell Biol., 2, 1–6.

Konishi A, Shimizu S, Hirota J, Takao T, Fan Y, Matsuoka Y, Zhang L, Yoneda Y, Fujii Y, Skoultchi AI and Tsujimoto Y. (2003). Cell, 114, 673–688. Lassus P, Opitz-Araya X and Lazebnik Y. (2002). Science, 297, 1352–1354. Lee E, Nakatsuma A, Hiraoka R, Ishikawa E, Enomoto R and Yamauchi A. (1999). IUBMB Life, 48, 79–83. Lee Y, Chong MJ and McKinnon PJ. (2001). J. Neurosci., 21, 6687–6693. Li H, Kolluri SK, Gu J, Dawson MI, Cao X, Hobbs PD, Lin B, Chen G, Lu J, Lin F, Xie Z, Fontana JA, Reed JC and Zhang X. (2000). Science, 289, 1159–1164. Lin Y, Ma W and Benchimol S. (2000). Nat. Genet., 26, 122–127. Liu X, Kim CN, Yang J, Jemmerson R and Wang X. (1996). Cell, 86, 147–157. Lowe SW, Ruley HE, Jacks T and Housman DE. (1993a). Cell, 74, 957–967. Lowe SW, Schmitt EM, Smith SW, Osborne BA and Jacks T. (1993b). Nature, 362, 847–849. Lu X and Lane DP. (1993). Cell, 75, 765–778. Mancini M, Machamer CE, Roy S, Nicholson DW, Thornberry NA, Casciola-Rosen LA and Rosen A. (2000). J. Cell Biol., 149, 603–612. Manome Y, Datta R, Taneja N, Shafman T, Bump E, Hass R, Weichselbaum R and Kufe D. (1993). Biochemistry, 32, 10607–10613. Marchenko ND, Zaika A and Moll UM. (2000). J. Biol. Chem., 275, 16202–16212. Marchetti P, Susin SA, Decaudin D, Gamen S, Castedo M, Hirsch T, Zamzami N, Naval J, Senik A and Kroemer G. (1996). Cancer Res., 56, 2033–2038. Matsuoka S, Huang M and Elledge SJ. (1998). Science, 282, 1893–1897. Meikrantz W, Gisselbrecht S, Tam SW and Schlegel R. (1994). Proc. Natl. Acad. Sci. USA, 91, 3754–3758. Mendelsohn AR, Hamer JD, Wang ZB and Brent R. (2002). Proc. Natl. Acad. Sci. USA, 99, 6871–6876. Michaelson JS, Bader D, Kuo F, Kozak C and Leder P. (1999). Genes Dev., 13, 1918–1923. Michaelson JS and Leder P. (2003). J. Cell Sci., 116, 345–352. Mihara M, Erster S, Zaika A, Petrenko O, Chittenden T, Pancoska P and Moll UM. (2003). Mol. Cell, 11, 577–590. Miyashita T and Reed JC. (1995). Cell, 80, 293–299. Muller M, Wilder S, Bannasch D, Israeli D, Lehlbach K, LiWeber M, Friedman SL, Galle PR, Stremmel W, Oren M and Krammer PH. (1998). J. Exp. Med., 188, 2033–2045. Murgia M, Pizzo P, Sandona D, Zanovello P, Rizzuto R and Di Virgilio F. (1992). J. Biol. Chem., 267, 10939–10941. Nahle Z, Polakoff J, Davuluri RV, McCurrach ME, Jacobson MD, Narita M, Zhang MQ, Lazebnik Y, Bar-Sagi D and Lowe SW. (2002). Nat. Cell Biol., 4, 859–864. Nakano K and Vousden KH. (2001). Mol. Cell, 7, 683–694. Norbury C, MacFarlane M, Fearnhead H and Cohen GM. (1994). Biochem. Biophys. Res. Commun., 202, 1400–1406. Norbury CJ and Hickson ID. (2001). Annu. Rev. Pharmacol. Toxicol., 41, 367–401. Oda E, Ohki R, Murasawa H, Nemoto J, Shibue T, Yamashita T, Tokino T, Taniguchi T and Tanaka N. (2000). Science, 288, 1053–1058. Ohiro Y, Usheva A, Kobayashi S, Duffy SL, Nantz R, Gius D and Horikoshi N. (2003). Mol. Cell. Biol., 23, 322–334. Ongkeko W, Ferguson DJ, Harris AL and Norbury C. (1995). J. Cell Sci., 108, 2897–2904. Owen-Schaub LB, Zhang W, Cusack JC, Angelo LS, Santee SM, Fujiwara T, Roth JA, Deisseroth AB, Zhang WW, Oncogene


2808 Krunzel E and Radinsky R. (1995). Mol. Cell. Biol., 15, 3032–3040. Paroni G, Henderson C, Schneider C and Brancolini C. (2002). J. Biol. Chem., 277, 15147–15161. Parrish J, Li L, Klotz K, Ledwich D, Wang X and Xue D. (2001). Nature, 412, 90–94. Pediconi N, Ianari A, Costanzo A, Belloni L, Gallo R, Cimino L, Porcellini A, Screpanti I, Balsano C, Alesse E, Gulino A and Levrero M. (2003). Nat. Cell Biol., 5, 552–558. Peters M, DeLuca C, Hirao A, Stambolic V, Potter J, Zhou L, Liepa J, Snow B, Arya S, Wong J, Bouchard D, Binari R, Manoukian AS and Mak TW. (2002). Proc. Natl. Acad. Sci. USA, 99, 11305–11310. Read SH, Baliga BC, Ekert PG, Vaux DL and Kumar S. (2002). J. Cell Biol., 159, 739–745. Reinke V and Lozano G. (1997). Oncogene, 15, 1527–1534. Robertson JD, Enoksson M, Suomela M, Zhivotovsky B and Orrenius S. (2002). J. Biol. Chem., 277, 29803–29809. Robertson JD, Gogvadze V, Zhivotovsky B and Orrenius S. (2000). J. Biol. Chem., 275, 32438–32443. Rogakou EP, Nieves-Neira W, Boon C, Pommier Y and Bonner WM. (2000). J. Biol. Chem., 275, 9390–9395. Roos-Mattjus P, Vroman BT, Burtelow MA, Rauen M, Eapen AK and Karnitz LM. (2002). J. Biol. Chem., 277, 43809–43812. Saito S, Goodarzi AA, Higashimoto Y, Noda Y, Lees-Miller SP, Appella E and Anderson CW. (2002). J. Biol. Chem., 277, 12491–12494. Sawada M, Sun W, Hayes P, Leskov K, Boothman DA and Matsuyama S. (2003). Nat. Cell Biol., 5, 320–329. Schulze-Osthoff K, Walczak H, Droge W and Krammer PH. (1994). J. Cell Biol., 127, 15–20. Schumacher B, Hofmann K, Boulton S and Gartner A. (2001). Curr. Biol., 11, 1722–1727. Screaton RA, Kiessling S, Sansom OJ, Millar CB, Maddison K, Bird A, Clarke AR and Frisch SM. (2003). Proc. Natl. Acad. Sci. USA, 100, 5211–5216. Shafman T, Khanna KK, Kedar P, Spring K, Kozlov S, Yen T, Hobson K, Gatei M, Zhang N, Watters D, Egerton M, Shiloh Y, Kharbanda S, Kufe D and Lavin MF. (1997). Nature, 387, 520–523. Shearwin-Whyatt LM, Harvey NL and Kumar S. (2000). Cell Death Differ., 7, 155–165. Sheikh SM and Huang Y. (2003). Cell Cycle, 2, 346–347. Shen Y and White E. (2001). Adv. Cancer Res., 82, 55–84. Shiloh Y. (2001). Curr. Opin. Genet. Dev., 11, 71–77. Singh KK, Russell J, Sigala B, Zhang Y, Williams J and Keshav KF. (1999). Oncogene, 18, 6641–6646. Skalka M, Matyasova J and Cejkova M. (1976). FEBS Lett., 72, 271–275. Sogame N, Kim M and Abrams JM. (2003). Proc. Natl. Acad. Sci. USA, 100, 4696–4701. Song JY, Lim JW, Kim H, Morio T and Kim KH. (2003). J. Biol. Chem., 278, 36676–36687. Stevens C, Smith L and La Thangue NB. (2003). Nat. Cell Biol., 5, 401–409. Susin SA, Lorenzo HK, Zamzami N, Marzo I, Snow BE, Brothers GM, Mangion J, Jacotot E, Costantini P, Loeffler M, Larochette N, Goodlett DR, Aebersold R, Siderovski DP, Penninger JM and Kroemer G. (1999). Nature, 397, 441–446. Takai H, Naka K, Okada Y, Watanabe M, Harada N, Saito S, Anderson CW, Appella E, Nakanishi M, Suzuki H,

Oncogene

Nagashima K, Sawa H, Ikeda K and Motoyama N. (2002). EMBO J., 21, 5195–5205. Tang JT, Yamazaki H, Inoue T, Koizumi M, Yoshida K and Ozeki S. (1999). Anticancer Res., 19, 4959–4964. Tepper CG and Studzinski GP. (1993). J. Cell Biochem., 52, 352–361. Thornberry NA. (1999). Cell Death Differ., 6, 1023–1027. Torii S, Egan DA, Evans RA and Reed JC. (1999). EMBO J., 18, 6037–6049. Um JH, Kang CD, Hwang BW, Ha MY, Hur JG, Kim DW, Chung BS and Kim SH. (2003). Leuk. Res., 27, 509–516. Vaughan AT, Betti CJ and Villalobos MJ. (2002). Apoptosis, 7, 173–177. Vousden KH and Lu X. (2002). Nat. Rev. Cancer, 2, 594–604. Wang B, Matsuoka S, Carpenter PB and Elledge SJ. (2002). Science, 298, 1435–1438. Wang J, Silva JP, Gustafsson CM, Rustin P and Larsson NG. (2001). Proc. Natl. Acad. Sci. USA, 98, 4038–4043. Wang XW, Vermeulen W, Coursen JD, Gibson M, Lupold SE, Forrester K, Xu G, Elmore L, Yeh H, Hoeijmakers JH and Harris CC. (1996). Genes Dev., 10, 1219–1232. Wang Y, Cortez D, Yazdi P, Neff N, Elledge SJ and Qin J. (2000). Genes Dev., 14, 927–939. Wang ZG, Ruggero D, Ronchetti S, Zhong S, Gaboli M, Rivi R and Pandolfi PP. (1998). Nat. Genet., 20, 266–272. Waterhouse NJ, Ricci JE and Green DR. (2002). Biochimie, 84, 113–121. Wei MC, Zong WX, Cheng EH, Lindsten T, Panoutsakopoulou V, Ross AJ, Roth KA, MacGregor GR, Thompson CB and Korsmeyer SJ. (2001). Science, 292, 727–730. Westphal CH, Rowan S, Schmaltz C, Elson A, Fisher DE and Leder P. (1997). Nat. Genet., 16, 397–401. Wilson AJ, Arango D, Mariadason JM, Heerdt BG and Augenlicht LH. (2003). Cancer Res., 63, 5401–5407. Woo RA, Jack MT, Xu Y, Burma S, Chen DJ and Lee PW. (2002). EMBO J., 21, 3000–3008. Wu S, Loke HN and Rehemtulla A. (2002). Neoplasia, 4, 486– 492. Wyllie AH. (1980). Nature, 284, 555–556. Wyllie AH, Kerr JF and Currie AR. (1980). Int. Rev. Cytol., 68, 251–306. Yang CR, Leskov K, Hosley-Eberlein K, Criswell T, Pink JJ, Kinsella TJ and Boothman DA. (2000). Proc. Natl. Acad. Sci. USA, 97, 5907–5912. Yang S, Kuo C, Bisi JE and Kim MK. (2002). Nat. Cell Biol., 4, 865–870. Yang X, Khosravi-Far R, Chang HY and Baltimore D. (1997). Cell, 89, 1067–1076. Yoneda M, Katsumata K, Hayakawa M, Tanaka M and Ozawa T. (1995). Biochem. Biophys. Res. Commun., 209, 723–729. Yoshida K, Komatsu K, Wang HG and Kufe D. (2002a). Mol. Cell. Biol., 22, 3292–3300. Yoshida Y, Izumi H, Ise T, Uramoto H, Torigoe T, Ishiguchi H, Murakami T, Tanabe M, Nakayama Y, Itoh H, Kasai H and Kohno K. (2002b). Biochem. Biophys. Res. Commun., 295, 945–951. Yoshida Y, Izumi H, Torigoe T, Ishiguchi H, Itoh H, Kang D and Kohno K. (2003). Cancer Res., 63, 3729–3734. Zhivotovsky B, Samali A, Gahm A and Orrenius S. (1999). Cell Death Differ., 6, 644–651. Zhou L and Steller H. (2003). Dev. Cell, 4, 599–605. Zhou M, Gu L, Li F, Zhu Y, Woods WG and Findley HW. (2002). J. Pharmacol. Exp. Ther., 303, 124–131.