Regulation of cell death by the Abl tyrosine kinase - Nature

Oncogene (2000) 19, 5643 ± 5650 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc

Regulation of cell death by the Abl tyrosine kinase Jean YJ Wang*,1 1

Department of Biology and the Cancer Center, University of California, San Diego, La Jolla, California, CA 92093-0322, USA

The c-abl proto-oncogene encodes a protein tyrosine kinase that is distributed in the nucleus and the cytoplasm of proliferating cells. In the nucleus, c-Abl activity is negatively regulated by the retinoblastoma protein (RB) and positively regulated by DNA damage signals. Activation of the c-Abl kinase by DNA damage requires the function of ATM, which regulates cell cycle checkpoint, DNA repair and apoptosis in response to DNA damage. Cells lacking c-Abl can activate cell cycle checkpoints and DNA repair, but show defects in apoptosis. The apoptosis defect of c-Abl de®cient cells is correlated with a defect in the induction and activation of p73, which is a functional homologue of the p53 tumor suppressor protein and has pro-apoptotic activity. The inhibition of c-Abl by RB is consistent with RB's ability to block apoptosis; while the activation of c-Abl by ATM is consistent with ATM's ability to activate cell death. The oncogenic Bcr-Abl tyrosine kinase is a potent inhibitor of apoptosis, and it is retained exclusively in the cytoplasm of transformed cells. Interestingly, when Bcr-Abl is trapped inside of the nucleus through a combined disruption of its cytoplasmic retention and its nuclear export, this oncogenic Abl kinase induces apoptosis. Taken together, the current results support a role for the nuclear c-Abl tyrosine kinase in the regulation of apoptosis. Whether the cytoplasmic c-Abl kinase can actively inhibit apoptosis remains to be determined; however, a deliberate retention of c-Abl in the cytoplasm could potentially contribute to the attenuation of apoptosis response. Oncogene (2000) 19, 5643 ± 5650. Keywords: Bcr-Abl; retinoblastoma tumor suppressor; ATM; p53; p73 Introduction The Abl protein contains the characteristic protein kinase domain and the Src homology domains SH3, SH2 (for a recent review, see Wang, 2000b). Interestingly, however, the cellular Abl kinase (c-Abl) has several unique features that distinguish it from the Src tyrosine kinases. First, c-Abl can shuttle between the cytoplasmic and the nuclear compartments (Taagepera et al., 1998) because it contains three nuclear localization signals (Wen et al., 1996) and a nuclear export signal (Taagepera et al., 1998). Second, c-Abl can interact with the cytoskeleton because it contains binding sites for F-actin (McWhirter and Wang, 1993; Van Etten et al., 1994) and G-actin (Van Etten et al.,

*Correspondence: JYJ Wang, Bonner Hall 3326, Department of Biology, 0322, UCSD 9500 Gilman Drive, La Jolla, CA 92093-0322, USA

1994). Third, c-Abl is associated with the chromatin (Kipreos and Wang, 1992) because it contains three HMG-like boxes (HLBs) that preferentially bind to A/T duplexes and distorted DNA structures (DavidCordonnier et al., 1998; Miao and Wang, 1996). The c-Abl tyrosine kinase is essential to the proper development of a mouse, because c-abl-knockout mice exhibit embryonic and neonatal lethality which can be rescued by the transgenic expression of c-abl cDNA (Hardin et al., 1996; Tybulewicz et al., 1991). A wide variety of low-penetrant phenotypes have been observed with abl-null mice (Li et al., 2000; Schwartzberg et al., 1991; Tybulewicz et al., 1991); yet, the underlying mechanisms for these developmental defects are unknown. A second Abl-related-gene, Arg, has also been knocked-out (Koleske et al., 1998). Arg-de®cient mice are born normal and do not exhibit overt developmental defects. Because the abl/arg double knockout mice show a more severe embryonic lethality, Arg must be there to sustain the development of ablde®cient mice. However, Arg must not be able to replace all of Abl's functions, because abl-de®cient mice have developmental defects that are not observed with arg-de®cient mice (Koleske et al., 1998). The nuclear localization sequences and the DNA binding HLBs of c-Abl are not well conserved in the Arg protein. It is therefore possible that the nuclear function of c-Abl cannot be substituted by Arg; hence, the abl-de®cient mice show defects that are not shared by the arg-de®cient mice. The Abl tyrosine kinase has oncogenic potential, which is activated in the Bcr-Abl hybrid protein of chronic myelogenous leukemia (CML). The human cabl gene, on chromosome 9, is fused with the bcr gene, on chromosome 22, through chromosomal translocation, resulting in the formation of Bcr-Abl hybrid protein in CML (Sawyers, 1999). A coiled-coil domain in Bcr forms homodimers and homotetramers and it is this oligomerization that activates the Abl tyrosine kinase in Bcr-Abl (McWhirter et al., 1993). Similarly, the Tel-Abl hybrid protein is activated by the coiledcoil of Tel in a few rare variant of CML (Golub et al., 1996). A Tel-Arg hybrid protein has recently been identi®ed, suggesting that Arg kinase shares the oncogenic potential of the Abl kinase (Iijima et al., 2000). The oncogenic Bcr-Abl kinase can activate a number of cytoplasmic signal transduction pathways that are normally controlled by receptor tyrosine kinases (Sawyers, 1999). Therefore, Bcr-Abl can abrogate the dependence on interleukin-3 in hematopoietic cell lines (McWhirter and Wang, 1993), and the dependence on adhesion in ®broblasts (Renshaw et al., 1995). The transforming activity of Bcr-Abl is not simply due to an increase in the Abl kinase activity, because overproduction of c-Abl does not transform cells.

Abl kinase regulates apoptosis JYJ Wang

5644

There is much confusion in the current literature regarding the biological functions of the c-Abl tyrosine kinase. A contributing factor to this confusion is the diculty in determining which of the many functions of Bcr-Abl are relevant to the normal functions of cAbl. To avoid this confusion, I will focus this review on recent results that shed light on the function of the non-oncogenic c-Abl tyrosine kinase. I will then discuss the Bcr-Abl function in the context of the normal cAbl function. Cell death inducers activate c-Abl tyrosine kinase The early studies of the Abl tyrosine kinase were focused on the transforming functions of the oncogenic v-Abl (of the Abelson murine leukemia virus) and Bcr-Abl kinases. It was therefore rather surprising to ®nd that the normal c-Abl tyrosine kinase is involved in the activation of cell death. A number of reports have shown that the c-Abl tyrosine kinase activity can be stimulated by physiological and pharmacological agents that induce apoptosis, and these results are summarized in Table 1. DNA damage-inducers, such as ionizing radiation (IR), cisplatin, mitomycin C, and topoisomerase inhibitors (etoposide, doxorubicin, camptothecin) can activate cAbl tyrosine kinase (Table 1). Tumor necrosis factor (TNF), a physiological inducer of apoptosis, can also activate c-Abl tyrosine kinase (Table 1). Overproduction of c-Abl tyrosine kinase in Saos-2 cells, which do not express p53 or RB, can activate apoptosis (Theis and Roemer, 1998). In NIH3T3 cells, transient expression of c-Abl weakly activated apoptosis in about 10% of transfected cells (Cong and Go€, 1999). These results suggest that c-Abl has an active role in the induction of cell death. Preferential activation of nuclear c-Abl tyrosine kinase by DNA damage DNA damage inducers preferentially activate the nuclear c-Abl tyrosine kinase (Kharbanda et al., 1995b; Liu et al., 1996). It should be noted that cAbl can shuttle between the nuclear and cytoplasmic compartments (Taagepera et al., 1998), and therefore could conceivably transmit the DNA damage signal from the nucleus to the cytoplasm. In other words, the DNA damage-activated c-Abl tyrosine kinase may exit the nucleus and perform a function in the cytoplasm. The currently available immune complex kinase assay for c-Abl activity might not be sensitive enough to detect the nuclear export of a small fraction of the activated c-Abl kinase. In any event, the current data is Table 1 Apoptosis inducers activate the c-Abl tyrosine kinase Inducers

Cell type

Ref

Ionizing radiation

Thymocytes Carcinoma cells Fibroblasts Carcinoma cells Fibroblasts Leukemia Leukemia Leukemia

Baskaran et al., 1997 Yuan et al., 1997 Liu et al., 1996 Gong et al., 1999 Liu et al., 1996 Dan et al., 1999 Dan et al., 1999 Dan et al., 1999 Chen and Wang, unpublished

Cisplatin Mitomycin C Etoposide Camptothecin TNF

Oncogene

consistent with the idea that c-Abl is regulated by a nuclear signal transduction pathway that is activated by DNA damage and other apoptosis inducers. Abl-deficient cells showed apoptosis defect in response to DNA damage The study of abl-de®cient mouse and chick cells supports a role for c-Abl tyrosine kinase in DNA damage-induced apoptosis. Embryo ®broblasts derived from c-abl-de®cient mice are more resistant to cisplatin, they die slower and are less sensitive than their wild type counterparts (Gong et al., 1999). Re-introduction of a functional c-Abl in these abl-de®cient cells increased their sensitivity to cisplatin, demonstrating a direct contribution of this tyrosine kinase in cisplatininduced cytotoxicity (Gong et al., 1999). In another study using the avian DT10 cells, homozygous disruption of the chick c-abl gene was found to cause a reduction in apoptosis following exposure to IR (Takao et al., 2000). Despite a reduced apoptosis response, the abl-de®cient DT10 cells showed a similar sensitivity to IR as the wild type DT10 cells in clonogenic survival assays (Takao et al., 2000). It should be noted that clonogenic survival measures the ability of irradiated cells to recover from the DNA damage and undergo at least ®ve to six doublings, thus, it is more than a `survival' assay. Cells that undergo a permanent cell cycle arrest will not form colonies and will be scored as `non-survivors' in clonogenic assays. The observations that abl-de®cient DT10 cells exhibited reduced apoptosis but normal clonogenic survival are therefore not internally inconsistent. These observations suggest that, at least in DT10 cells, the c-Abl tyrosine kinase is not required for repair and recovery from IR-induced DNA damage (Takao et al., 2000). This observation also indicated that while abl-de®cient cells showed reduced apoptosis, their growth arrest response to IR was normal and could prevent the clonogenic survival of damaged cells. In fact, the abl-de®cient mouse embryo ®broblasts also showed normal G1, S and G2 checkpoint response to IR (Baskaran et al., 1997). Taken together, the current evidence supports a role for c-Abl tyrosine kinase in the activation of apoptosis by DNA damage. The previously proposed function of c-Abl in cell cycle checkpoint response (Yuan et al., 1996) or in DNA repair (Chen et al., 1999) is not supported by the phenotype of abl-de®cient cells. However, the possibility that c-Abl might contribute to the regulation of cell cycle checkpoints and DNA repair in cell types other than mouse embryo ®broblasts or the chick DT10 cells cannot be ruled out at this time. Negative role of RB in DNA damage-induced activation of c-Abl kinase The activation of nuclear c-Abl tyrosine kinase by DNA damage is observed in proliferating cells but not in growth arrested cells (PL Puri and JYJ Wang, unpublished). By treating synchronized cells with DNA damage inducers, it was possible to show that c-Abl kinase cannot be activated by DNA damage in G0/G1 cells, but c-Abl kinase activation is observed in cells that have committed to enter S phase (Liu et al., 1996).

Abl kinase regulates apoptosis JYJ Wang

5645

In G0/G1 cells, the nuclear c-Abl tyrosine kinase activity is suppressed by the retinoblastoma tumor suppressor protein (RB), which binds to the tyrosine kinase domain of c-Abl to block its activity (Welch and Wang, 1993, 1995). Phosphorylation of RB at G1/S disrupts the RB-c-Abl interaction, leading to the release of c-Abl in the nucleus of S phase cells (Welch and Wang, 1993, 1995). The observation that DNA damage does not activate c-Abl kinase in growth arrested cells or G0/G1 cells suggests that the inhibitory e€ect of RB is dominant and can prevent c-Abl from being activated by DNA damage signal. RB has an essential role in DNA damage-induced growth arrest The RB protein has been shown to play an essential role in DNA damage response. RB becomes activated, that is, maintained in the hypophosphorylated form, following DNA damage and the activation of RB can arrest cells in G1, S and G2 phases of the cell cycle (Harrington et al., 1998; Knudsen et al., 2000; Naderi and Wang, unpublished). RB plays an essential role in DNA damage-induced G1 checkpoint, because Rbde®cient mouse embryo ®broblasts cannot be arrested in G1 following DNA damage (Harrington et al., 1998; Knudsen et al., 2000). Interestingly, RB also plays an essential role in a DNA damage-induced intra-S phase checkpoint. Under conditions of high dose exposure to cisplatin, mitomycin C or etoposide, and possibly when the DNA damage becomes irreparable, RB can become dephosphorylated and activated in S phase cells to block DNA replication. Rb-de®cient cells do not have this intra-S phase checkpoint, and hence, continue to replicate DNA even after a high dose of DNA damage (Knudsen et al., 2000). Although Rb-de®cient cells are defective in G1 or the intra-S phase checkpoint response to DNA damage, their G2 checkpoint response in intact. However, Rb-de®cient cells cannot undergo the longterm G2 arrest, which follows the G2 checkpoint to arrest cell proliferation after heavy damage (Naderi and Wang, unpublished). Previous work has demonstrated that the G1 checkpoint and the long-term G2 arrest are both dependent on the functions of p53 and p21Cip1 (Bunz et al., 1998). The induction of p21Cip1 by p53 leads to an inhibition of the cyclin-dependent kinases and hence maintaining RB in the active, hypophosphorylated form (Brugarolas et al., 1999). The ®nding that the Rb-de®cient cells do not undergo G1 checkpoint or long-term G2 arrest suggests that RB is an essential e€ector for the p53-p21Cip1 regulated growth arrest pathway (Figure 1, left). The intra-S phase checkpoint mediated by RB can occur in p53- and p21Cip1-de®cient cells (Knudsen et al., 2000), hence, RB can be activated by alternative mechanisms to block DNA replication. Because Rbde®cient cells are defective in DNA damage induced G1 checkpoint, intra-S phase arrest, and long-term G2 arrest, RB plays an essential role in cell cycle arrest at all phases following DNA damage (Figure 1, left). Degradation of RB sensitizes cells to apoptosis The RB-mediated cell cycle arrest protects cells from apoptosis, because Rb-de®cient cells are more sensitive

Figure 1 DNA damage-activated signaling pathways that regulate growth arrest and apoptosis through ATM, p53, RB and c-Abl. The c-Abl tyrosine kinase plays a role in the regulation of apoptosis following DNA damage. DNA damage can induce either growth arrest or apoptosis. The p53 tumor suppressor protein and the ATM kinase can activate both biological responses. The retinoblastoma protein (RB) plays an essential role in DNA damage-induced growth arrest and this RBmediated growth arrest protects cells from apoptosis. In cells undergoing apoptosis, RB is cleaved by caspases and then degraded and the degradation of RB sensitizes cells to apoptosis (Tan and Wang, 1998). The activation of c-Abl tyrosine kinase by DNA damage requires the function of ATM and the inactivation of RB, either by RB phosphorylation or by RB degradation. Activated c-Abl can induce and activate p73, a homologue of the p53 protein with pro-apoptotic function. Degradation and inactivation of RB also activates E2F, which can promote apoptosis as well. It is possible that the activation of p53, p73 and E2F transcription factors can potentiate the activation of caspases, which ultimately perform the functions of apoptosis induction. Activation of the c-Abl/p73 pathway can also contribute to the induction of p21Cip1 and reinforces the p53 function in growth arrest. The balance and interplay between the components of these pathways are likely to determine the cell fate following DNA damage. The cleavage and degradation of RB is also triggered by TNF, and TNF can activate c-Abl (Dan et al., 1999) to induce p73 (Chen and Wang, unpublished). Thus, the apoptosis depicted in this scheme may also play a role in cell death response to signals other than DNA damage

to DNA damage induced cell death (Almasan et al., 1995; Knudsen et al., 2000). In Rb+/+ mouse embryo ®broblasts, a prolonged treatment with cisplatin can cause the dephosphorylation of RB in all phases of the cell cycle (Knudsen, et al., 2000). During the drug treatment and prior to the establishment of growth arrest, a fraction of the Rb+/+ MEFs underwent apoptosis. However, after the dephosphorylation of RB and the establishment of growth arrest, the Rb+/+ MEFs remained arrested in G1, S and G2 phase of the cell cycle for as long as 7 days after the removal of cisplatin. These arrested Rb+/+ MEFS did not undergo further apoptosis and survived for a long period of time after the removal of cisplatin (Knudsen et al., 2000). By contrast, the Rb7/7 MEFs could not undergo growth arrest and continued to die even after the Oncogene

Abl kinase regulates apoptosis JYJ Wang

5646

removal of drugs leading to a complete eradication of the culture (Knudsen et al., 2000). These observations support the notion that cells with activated RB are more resistant to damage-induced apoptosis. The negative role of RB in DNA damage-induced apoptosis is consistent with its ability to inhibit the c-Abl tyrosine kinase (Figure 1). Of course, RB also inhibits E2F, which has the ability to activate apoptosis as well (Figure 1). The RB protein is actively cleaved by caspases and degraded during apoptosis (Tan and Wang, 1998; Tan et al., 1997). The cleavage of RB is observed under conditions of genotoxic stress and in cells treated with physiological apoptosis inducers, such as TNF (Tan and Wang, 1998). In this regard, it is interesting to note that an inhibitor of caspases when added to cells can block the activation of c-Abl kinase by etoposide or TNF (Dan et al., 1999). One possible explanation for the requirement of caspase activity in the activation of c-Abl tyrosine kinase is that the cleavage of RB is required to remove the dominant inhibitory e€ect on cAbl kinase. This notion is consistent with the observation that the immediate activation of c-Abl kinase by DNA damage only occurs after cells commit to S phase, when RB is phosphorylated and inactivated. An RB protein with mutation at the caspase cleavage site and is resistant to cleavage by caspase has been shown to protect cells against apoptosis induced by TNF (Tan et al., 1997). If RB cleavage by caspases is required for the activation of c-Abl tyrosine kinase in apoptotic cells, the expression of this cleavage-resistant RB mutant should block c-Abl activation by death inducers. This has indeed been found to be the case (Chen and Wang, unpublished). Taken together, the accumulated data supports a pathway in which RB antagonizes apoptosis, in part, by blocking the activation of c-Abl tyrosine kinase (Figure 1). The inhibitory e€ect of RB on c-Abl kinase is therefore relevant to the regulation of apoptosis. Positive role of ATM in DNA damage-induced activation of c-Abl kinase The activation of nuclear c-Abl tyrosine kinase by ionizing radiation requires a functional ATM, which is a protein kinase encoded by the gene that is mutated in the human disorder Ataxia telangiectasia (Baskaran et al., 1997). The basal activity of c-Abl tyrosine kinase is similar in normal and ATM-de®cient cells. Exposure of normal cells to IR leads to a 3 ± 5-fold increase in the nuclear c-Abl kinase activity. However, this DNA damaged-induced increase in c-Abl activity does not occur in ATM-de®cient cells (Baskaran et al., 1997; Shafman et al., 1997). The activation of c-Abl kinase by IR can also be abolished by mutation of a single Serine residue (Ser465) in the tyrosine kinase domain of the Abl protein (Baskaran et al., 1997). Mutation of Ser465 to Ala does not a€ect the basal activity, but abolishes the IR-induced activation of c-Abl kinase (Baskaran et al., 1997). The basal activity of c-Abl is increased by mutating Ser465 to Glu, and this higher basal activity cannot be further stimulated by IR (Gong and Wang, unpublished). Co-expression of the ATM kinase with c-Abl leads to an increase in the c-

Oncogene

Abl tyrosine kinase activity, which is dependent on the presence of Ser465 (Baskaran et al., 1997). The Ser465 of c-Abl can be phosphorylated by ATM kinase and the DNA-dependent protein kinase in vitro (Baskaran et al., 1997). Taken together, these results show that ATM is a necessary activator of the c-Abl tyrosine kinase in DNA damage response. ATM may directly phosphorylate c-Abl at Ser465 or indirectly phosphorylate c-Abl through DNA-PK or other as yet unidenti®ed protein kinases. In any event, DNA damageinduced activation of c-Abl requires the function of ATM and the phosphorylation of Ser465 in the Abl tyrosine kinase domain (Figure 1). ATM contributes to the activation of apoptosis by DNA damage The ATM kinase activity is activated by 1.5 ± 2-fold within the ®rst hour after exposure of cells to ionizing radiation and decline to the basal level thereafter (Banin et al., 1998; Canman et al., 1998). The ATM function is important for IR-induced cell cycle checkpoints and DNA repair (Rotman and Shiloh, 1999; Wang, 2000a). Because ATM-de®cient cells are hypersensitive to IR and show increased chromosome breaks, ATM has a protective role in the IR response (Wang, 2000a). It should be noted that the low level and transient activation of ATM kinase is observed with cells that do not undergo apoptosis following IR. In addition to its role in cell cycle checkpoints and DNA repair, ATM can also contribute to IR-induced apoptosis, most notably in neurons (Chong et al., 2000). This has been demonstrated with ATM knockout mice, that the immature neurons in the developing central nervous system (CNS) of ATM-de®cient mice show defects in IR-induced apoptosis (Chong et al., 2000). This apoptosis defect of developing CNS is also observed with p53-de®cient mice (Chong et al., 2000), consistent with a functional interaction between ATM and p53 in DNA damage response. The activation of cAbl tyrosine kinase might also contribute to ATMdependent apoptosis. At present, there is no direct evidence supporting a role for c-Abl in the induction of apoptosis of developing CNS. In a separate study using cisplatin to damage DNA and trigger apoptosis, we have observed a delayed activation of the ATM kinase that is of a much higher magnitude and for a longer duration than that which is induced by IR (Gong and Wang, unpublished). We found that the ATM kinase can be activated by 10 ± 15-fold at 8 ± 12 h after treatment with cisplatin, and declined to basal level by 24 h of treatment. This delayed and high level of activation of ATM kinase correlates with the activation of caspase activity and the induction of apoptosis (Gong et al., unpublished). These preliminary observations suggest the possibility that the extent and the duration of ATM kinase activation might determine the biological response. A transient and low-level of ATM kinase activation is correlated with cell cycle checkpoint response, whereas a delayed and high-level of activation may trigger apoptosis. Mechanism of c-Abl action in apoptosis The mechanism by which c-Abl contributes to the activation of cell death is not completely understood.

Abl kinase regulates apoptosis JYJ Wang

The current evidence suggests several possible downstream e€ectors, and they are the transcription factors p73 and p53 and the stress-activated kinases, JNK and p38. The p53 tumor suppressor protein belongs to a family of related transcription factors, including p53, p63 and p73 (Levrero et al., 1999). Ectopic expression of either p53 or p73 can induce apoptosis (Jost et al., 1997; Levrero et al., 1999). The central role of p53 in DNA damage-induced apoptosis is well established. A role for p73 in the regulation of apoptosis has begun to emerge from recent studies. The connection between cAbl and p53 is tenuous, however, a functional interaction between c-Abl and p73 is supported by genetic evidence. The strongest evidence that c-Abl regulates p73 comes from the observation that cisplatin can induce the accumulation of p73 protein, and this requires c-Abl (Gong et al., 1999). Using primary mouse embryo ®broblasts, Gong et al. (1999) have shown that cisplatin can induce the accumulation of p73 protein in wild type and p53-de®cient cells, but the induction of p73 by cisplatin is not observed in c-Ablde®cient cells. By contrast, the induction of p53 protein by cisplatin can occur without a functional c-Abl (Gong et al., 1999). Thus, c-Abl has an essential role in the regulation of p73 protein by DNA damage (Figure 1, right). While c-Abl has been linked to the stabilization and activation of p53 in transient cotransfection experiments (Sionov et al., 1999), c-Abl is not essential to the induction and activation of p53 by DNA damage. The precise mechanism by which p73 activates apoptosis is not known. Because p73 can activate p53-regulated genes (Levrero et al., 1999), the simplest hypothesis would suggest that p73 and p53 activate apoptosis by the same mechanism. This idea is consistent with a recent report that describes the ability of a dominant negative variant of p73, the delta-N-p73 protein, to protect developing neurons from apoptosis induced by the withdrawal of nerve growth factor (Pozniak et al., 2000). This delta-N-p73 protein lacks the N-terminal transactivation domain of p73, but contains an intact DNA binding domain, the oligomerization domain and the C-terminal region of p73 (Yang et al., 2000). Ectopic expression of this delta-N-p73 protein can block p53-induced apoptosis and protect p737/7 neurons from death induced by the withdrawal of NGF (Pozniak et al., 2000). These observations suggest that the delta-N-p73 may bind to the promoters of p53-activated genes and thus block p53 from activating apoptosis. Taking this notion one step further, the induction of full-length p73 protein that contains the transactivation domain may over-ride the inhibitory e€ect of the delta-N-p73 to activate apoptosis. Recent studies using caspase-de®cient cells have shown that p53-induced apoptosis requires the function of caspase 9 (Soengas et al., 1999). It will be of interest to determine whether the induction of apoptosis by p73 also requires the function of caspase 9 (Figure 1). The c-Abl kinase can directly bind to and phosphorylate the p73 protein. The SH3 domain of c-Abl appears to interact with a proline-rich region in p73 (Agami et al., 1999), and this region is found in both the full-length and the delta-N-p73 proteins and is also common among the di€erent C-terminal spliced

variants of p73. The p73 protein appears to be phosphorylated at tyrosine-99 by c-Abl (Yuan et al., 1999); again, this tyrosine is present in both the fulllength and the delta-N-p73 protein and the di€erent Cterminal spliced variants. The c-Abl tyrosine kinase can increase the half-life of the p73 protein (Gong et al., 1999), and this is observed with at least four di€erent C-terminal spliced variants of the p73 proteins (alpha, beta, gamma, and delta) (Levrero et al., 1999). Whether c-Abl can stabilize the delta-N-p73 protein is currently not known. However, given that the deltaN-p73 contains the c-Abl interaction domain and the cAbl phosphorylation site, it is likely that the stability of delta-N-p73 may also be a€ected by c-Abl. If c-Abl could stabilize both the pro-apoptotic full-length p73 and the anti-apoptotic delta-N-p73, the e€ect of c-Abl on apoptosis would then be determined by the expression pattern of the p73 gene. In cells that predominantly express the delta-N-p73, for example, the sympathetic neurons in neonatal brains, activation of the c-Abl kinase may enhance the anti-apoptotic function of delta-N-p73 and thus promote survival. In cells that predominantly express the full-length p73, such as colon cancer cells and mouse embryo ®broblasts, activation of the c-Abl kinase will stimulate the pro-apoptotic function of p73 and thus promote cell death (Gong et al., 1999; Levrero et al., 1999). This raises the interesting possibility that the e€ect of c-Abl on apoptosis is highly dependent on the cell context, and one of the determinants may well be the alternative splicing of the p73 gene. The functional interaction between c-Abl and the stress activated kinases such as JNK and p38 has been described by several reports (Barila et al., 2000; Kharbanda et al., 1995a). However, the role of c-Abl in JNK-dependent apoptosis has not been well documented. JNK plays an essential role in UVinduced apoptosis, as demonstrated by the resistance of MEFs from JNK1/JNK2-de®cient mice to UVinduced cell death (Tournier et al., 2000). While UV induces p53, this DNA damage inducer does not activate the c-Abl tyrosine kinase (Liu et al., 1996), and does not induce the accumulation of p73 (Kaghad et al., 1997). Thus, c-Abl is not likely to participate in UV-induced cell death. An early report has suggested that IR-induced activation of JNK is dependent on the c-Abl function (Kharbanda et al., 1995b). However, subsequent studies have failed to support this conclusion (Dan et al., 1999; Liu et al., 1996). The functional interaction of c-Abl and the stress-activated kinases might be important in some cell types under some speci®c stress conditions. Identi®cation of such biological conditions will be required to clarify the role of cAbl in JNK-dependent cell death.

5647

Nuclear entrapment of oncogenic Bcr-Abl kinase induces apoptosis The recent ®ndings that c-Abl tyrosine kinase contributes to the activation of apoptosis appear to be at odds with the oncogenic activities of Bcr-Abl tyrosine kinase. The Abl tyrosine kinase is constitutively activated in Bcr-Abl, through the formation of homooligomers mediated by a coiled-coil domain at the Nterminus of Bcr (McWhirter et al., 1993). The Bcr-Abl kinase has been shown to activate a number of signal Oncogene

Abl kinase regulates apoptosis JYJ Wang

5648

Oncogene

transduction pathways that are normally controlled by receptor tyrosine kinases to regulate cell proliferation and apoptosis. Bcr-Abl can inhibit apoptosis through the activation of PI 3-kinase, Akt and by other mechanisms (Amarante-Mendes et al., 1998; McGahon et al., 1993, 1995; Skorski et al., 1995, 1997). The Bcr-Abl protein is localized exclusively to the cytoplasm of transformed cells (McWhirter and Wang, 1991, 1993), although it contains the nuclear localization sequences (NLS) (Wen et al., 1996) and the nuclear export sequence (NES) of Abl (Taagepera et al., 1998). The c-Abl protein is both cytoplasmic and nuclear, and shuttles between the two compartments (Taagepera et al., 1998). The nuclear export of c-Abl is inhibited by leptomycin B (LMB) (Taagepera et al., 1998) which inactivates the NES receptor known as CRM1/exportin-1 (Fornerod et al., 1997; Fukuda et al., 1997). As discussed above, the nuclear c-Abl kinase can be activated by DNA damage to induce the accumulation of p73 and activate apoptosis. The activation of Akt by the cytoplasmic Bcr-Abl also requires the kinase activity (Skorski et al., 1997). These observations led us to test whether the di€erent subcellular localization could contribute to the opposing e€ect of c-Abl and Bcr-Abl kinase on apoptosis. We have found that the Bcr-Abl tyrosine kinase is not imported into the nucleus, despite the fact that it contains three NLS (P Vigneri and JYJ Wang, 2000. Nature Medicine). The cytoplasmic localization of BcrAbl is mediated by retention mechanisms that involve the constitutive kinase activity and Bcr sequences (Vigneri and Wang, unpublished). Inhibition of the Bcr-Abl tyrosine kinase can partially alleviate the cytoplasmic retention leading to the nuclear import of a fraction of the Bcr-Abl protein. Because Bcr-Abl also contains the nuclear export signal (NES) of Abl, we did not detect the nuclear accumulation of kinasedefective Bcr-Abl protein unless nuclear export is also inhibited. Thus, a kinase defective mutant of Bcr-Abl can be trapped in the nucleus by treating cells with leptomycin B (LMB), which irreversibly inactivates Crm1/exportin (Kudo et al., 1999). The oncogenic BcrAbl can also be trapped in the nucleus, and this is achieved by using STI571 to inhibit the Bcr-Abl kinase and LMB to block the nuclear export. STI571 is a competitive inhibitor of the Abl kinase with a high degree of selectivity towards the ATP binding site in the Abl kinase domain (Buchdunger et al., 1996; Druker et al., 1996). We have found that the combined treatment with STI571 and LMB can lead to the nuclear accumulation of Bcr-Abl in ®broblastic or hematopoietic cells. The nuclear Bcr-Abl kinase can be re-activated either by the removal of STI571 or through the metabolic decay of STI571. Following reactivation of the Bcr-Abl kinase in the nucleus, we observed the activation of caspases, the appearance of phosphotidylserine on the cell surface and the laddering of chromosomal DNA, which are phenotypes of apoptotic cells (P Vigneri and JYJ Wang, 2000. Nature Medicine. Submitted). The combined treatment with STI571 and leptomycin B also causes the irreversible and complete killing of cells expressing Bcr-Abl, while either drug alone cannot achieve this e€ect. Our results suggest that the Bcr-Abl tyrosine kinase can be converted into a potent inducer of apoptosis by allowing it to function in the nucleus. Apoptosis

induced by the nuclear Bcr-Abl does not appear to be suppressed by the cytoplasmic Bcr-Abl, because nuclear entrapment of only a fraction of the total BcrAbl is sucient to kill cells. The ®nding that nuclear localized Bcr-Abl kinase can kill cells provides an explanation for the longstanding observation that the oncogenic Abl kinases (Bcr-Abl, Gag-v-Abl and DSH3c-Abl) are localized exclusively to the cytoplasm (Van Etten et al., 1989). Because nuclear Abl kinase can induce apoptosis, cell transformation must require cytoplasmic retention of an activated Abl kinase. Our results also suggest that Bcr-Abl would not be able to transform cells that cannot achieve an ecient cytoplasmic retention of this tyrosine kinase. That the Bcr-Abl kinase can induce apoptosis from the nucleus is consistent with the role of the nuclear c-Abl tyrosine kinase in DNA damage-induced cell death. Conclusions and future prospects The ability of the Abl tyrosine kinase to activate apoptosis is unique among the protein tyrosine kinases. Taken together, the current evidence suggests that activation of the nuclear c-Abl kinase can contribute to apoptosis by stabilizing and activating the p73 transcription factor. Because the p73 gene can generate a number of spliced variants with pro-apoptotic and anti-apoptotic functions, it remains to be determined whether c-Abl can also have pro-apoptotic and antiapoptotic functions depending on the mRNA expression pattern from the p73 gene. While p73 is a downstream e€ector for c-Abl in the regulation of apoptosis, p73 does not have to be the only e€ector. It will be of interest to identify other nuclear substrates of the c-Abl kinase. Since the constitutively active Bcr-Abl kinase is a strong inducer of apoptosis when it is trapped in the nucleus, it might be easier to use BcrAbl to ®nd the nuclear substrates that may contribute to c-Abl-regulated apoptosis response. The cytoplasmic Bcr-Abl tyrosine kinase is a strong inhibitor of apoptosis and this activity accounts for the resistance of CML cells to radiation and chemotherapy (Amarante-Mendes et al., 1998; McGahon et al., 1993, 1995). It is tempting to speculate that the cytoplasmic c-Abl tyrosine kinase might also have an anti-apoptotic function. The cytoplasmic c-Abl kinase is regulated by cell adhesion (Lewis et al., 1996), and has been shown to regulate cell motility (F Frasca, unpublished). At present, there is no evidence that c-Abl can activate any of the pathways, e.g., PI-3 kinase, Akt and Erk, that are activated by the cytoplasmic Bcr-Abl tyrosine kinase. Although the cytoplasmic c-Abl may not activate survival pathways, its retention in the cytoplasm may by default contribute to the attenuation of apoptosis. Two sets of observations are consistent with this notion. In the ®rst set of observations, we have found that c-Abl is localized exclusively in the cytoplasm of di€erentiated muscle cells because its nuclear import is inhibited (LL Whitaker and JYJ Wang, unpublished). It is interesting to note that terminally di€erentiated cells, such as muscle cells, are more resistant to DNA damage-induced apoptosis than proliferating cells. It will be of interest to determine whether the cytoplasmic retention of c-Abl contributes to this resistance phenotype in terminally di€erentiated cells. In the

Abl kinase regulates apoptosis JYJ Wang

second set of observations, we have shown that cell adhesion to extracellular matrix proteins can stimulate the nuclear export of c-Abl protein (Taagepera et al., 1998). In addition, we found that c-Abl is mostly cytoplasmic in cells that have grown into a dense monolayer and are contact-inhibited (S Taagepera and JYJ Wang, unpublished). Cell adhesion to extracellular matrix and cell ± cell contact, in general, can promote cell survival. Thus, the increased cytoplasmic localization of c-Abl tyrosine kinase under these conditions is consistent with a default mechanism to attenuate the apoptosis response. This review is focused on the regulation of cell death by c-Abl and thus emphasizes the nuclear c-Abl kinase function. This, however, should not be taken to suggest that c-Abl does not have a cytoplasmic function. On the contrary, the Drosophila Abl kinase, which does not appear to enter the nucleus, plays an important role in the regulation of cell motility (Lanier and

Gertler, 2000). The actin-binding function of Abl is conserved through evolution (Wang, 1993), and the mammalian c-Abl kinase has been shown to play a role in the ru‚ing response to PDGF (Plattner et al., 1999), and in the motile response to HGF (F Frasca, unpublished). Regulation of the cytoplasmic and nuclear distribution of c-Abl may in¯uence which of its many biological functions is of importance to a cell under a speci®ed set of physiological conditions. There is much work to be done before we can fully understand why nature designs the c-Abl tyrosine kinase to be this complex and with so many capabilities.

5649

Acknowledgments The author is supported by grants from the National Cancer Institute (CA43054 and CA58320).

References Agami R, Blandino G, Oren M and Shaul Y. (1999). Nature, 399, 809 ± 813. Almasan A, Yin YX, Kelly RE, Lee E, Bradley A, Li WW, Bertino JR and Wahl GM. (1995). Proc. Natl. Acad. Sci. USA, 92, 5436 ± 5440. Amarante-Mendes GP, Naekyung Kim C, Liu L, Huang Y, Perkins CL, Green DR and Bhalla K. (1998). Blood., 91, 1700 ± 1705. Banin S, Moyal L, Shieh S, Taya Y, Anderson C, Chessa L, Smorodinsky N, Prives C, Reiss Y, Shiloh Y and Ziv Y. (1998). Science, 281, 1674 ± 1677. Barila D, Mangano R, Gon¯oni S, Kretzschmar J, Moro M, Bohmann D and Superti-Furga G. (2000). EMBO J., 19, 273 ± 281. Baskaran R, Wood L, Whitaker L, Canman C, Morgan S, Xu Y, Barlow C, Baltimore D, Wynshaw-Boris A, Kastan M and Wang J. (1997). Nature, 387, 516 ± 519. Brugarolas J, Moberg K, Boyd SD, Taya Y, Jacks T and Lees JA. (1999). Proc. Natl. Acad. Sci. USA, 96, 1002 ± 1007. Buchdunger E, Zimmermann J, Mett H, Meyer T, Muller M, Druker BJ and Lydon NB. (1996). Cancer Res., 56, 100 ± 104. Bunz F, Dutriaux A, Lengauer C, Waldman T, Zhou S, Brown JP, Sedivy JM, Kinzler KW and Vogelstein B. (1998). Science, 282, 1497 ± 1501. Canman C, Lim D, Cimprich K, Taya Y, Tamai K, Sakaguchi K, Appella E, Kastan M and Siliciano J. (1998). Science, 281, 1677 ± 1679. Chen G, Yuan SSF, Liu W, Xu Y, Trujillo K, Song BW, Cong F, Go€ SP, Wu Y, Arlinghaus R, Baltimore D, Gasser PJ, Park MS, Sung P and Lee E. (1999). J. Biol. Chem., 274, 12748 ± 12752. Chong MJ, Murray MR, Gosink EC, Russell HRC, Srinivasan A, Kapsetaki M, Korsmeyer SJ and McKinnon PJ. (2000). Proc. Natl. Acad. Sci. USA, 97, 889 ± 894. Cong F and Go€ SP. (1999). Proc. Natl. Acad. Sci. USA, 96, 13819 ± 13824. Dan S, Naito M, Seimiya H, Kizaki A, Mashima T and Tsuruo T. (1999). Oncogene, 18, 1277 ± 1283. David-Cordonnier M-H, Hamdane M, Bailly C and D'Halluin J-C. (1998). Biochemistry, 37, 6065 ± 6076. Druker BJ, Tamura S, Buchdunger E, Ohno S, Segal GM, Fanning S, Zimmermann J and Lydon NB. (1996). Nat. Med., 2, 561 ± 566. Fornerod M, Ohno M, Yoshida M and Mattaj I. (1997). Cell, 90, 1051 ± 1060.

Fukuda M, Asano S, Nakamura T, Adachi M, Yoshida M, Yanagida M and Nishida E. (1997). Nature., 390, 308 ± 311. Golub T, Goga A, Barker G, Afar D, McLaughlin J, Bohlander S, Rowley J, Witte O and Gilliland D. (1996). Mol. Cell. Biol., 16, 4107 ± 4116. Gong JG, Costanzo A, Yang HQ, Melino G, Kaelin WG, Levrero M and Wang JYJ. (1999). Nature, 399, 806 ± 809. Hardin J, Boast S, Mendelsohn M, Santos Kdl and Go€ S. (1996). Oncogene, 12, 2669 ± 2677. Harrington EA, Bruce JL, Harlow E and Dyson N. (1998). Proc. Natl. Acad. Sci. USA, 95, 11945 ± 11950. Iijima Y, Ito T, Oikawa T, Eguchi M, Eguchi-Ishimae M, Kamada N, Kishi K, Asano S, Sakaki Y and Sato Y. (2000). Blood, 95, 2126 ± 2131. Jost C, Marin M and Kaelin W. (1997). Nature, 389, 191 ± 194. Kaghad M, Bonnet H, Yang A, Creancier L, Biscan J, Valent A, Minty A, Chalon P, Lelias J, Dumont X, Ferrara P, McKeon F and Caput D. (1997). Cell, 90, 809 ± 819. Kharbanda S, Pandey P, Ren R, Mayer B, Zon L and Kufe D. (1995a). J. Biol. Chem., 270, 30178 ± 30281. Kharbanda S, Ren R, Pandey P, Shafman TD, Feller SM, Weichselbaum RR and Kufe DW. (1995b). Nature, 376, 785 ± 788. Kipreos ET and Wang JYJ. (1992). Science, 256, 382 ± 385. Knudsen KE, Booth D, Naderi S, Sever-Chroneos Z, Fribourg AF, Hunton I, Feramisco JR, Wang JYJ and Knudsen ES. (2000). Mol. Cell. Biol., 20, in press. Koleske AJ, Gi€ord AM, Scott ML, Nee M, Bronson RT, Miczek KA and Baltimore D. (1998). Neuron, 21, 1259 ± 1272. Kudo N, Matsumori N, Taoka H, Fujiwara D, Schreiner EP, Wol€ B, Yoshida M and Horinouchi S. (1999). Proc. Natl. Acad. Sci. USA, 96, 9112 ± 9117. Lanier LM and Gertler FB. (2000). Curr. Opin. Neurobiol., 10, 80 ± 87. Levrero M, De Laurenzi V, Costanzo A, Gong J, Melino G and Wang JYJ. (1999). Cell. Death. Di€er., 6, 1146 ± 1153. Lewis J, Baskaran R, Taagepera S, Schwartz M and Wang J. (1996). Proc. Natl. Acad. Sci. USA, 93, 15174 ± 15179. Li BJ, Boast S, de los Santos K, Schieren I, Quiroz M, Teitelbaum SL, Tondravi MM and Go€ SP. (2000). Nature Genetics, 24, 304 ± 308. Liu Z, Baskaran R, Lea-Chou E, Wood L, Chen Y, Karin M and Wang J. (1996). Nature, 384, 273 ± 276.

Oncogene

Abl kinase regulates apoptosis JYJ Wang

5650

Oncogene

McGahon A, Bissonnette R, Schmitt M, Cotter KM, Green DR and Cotter TG. (1993). Blood, 83, 1179 ± 1187. McGahon AJ, Nishioka WK, Martin SJ, Mahboubi A, Cotter TG and Green DR. (1995). J. Biol. Chem., 270, 22625 ± 22631. McWhirter JR, Galasso DL and Wang JYJ. (1993). Mol. Cell. Biol., 13, 7587 ± 7595. McWhirter JR and Wang JYJ. (1991). Mol. Cell. Biol., 11, 1553 ± 1565. McWhirter JR and Wang JYJ. (1993). EMBO J., 12, 1533 ± 1546. Miao Y and Wang J. (1996). J. Biol. Chem., 271, 22823 ± 22830. Plattner R, Kadlec L, DeMali KA, Kazlauskas A and Pendergast AM. (1999). Genes Dev., 13, 2400 ± 2411. Pozniak CD, Radinovic S, Yang A, McKeon F, Kaplan DR and Miller FD. (2000). Science, 289, 304 ± 306. Renshaw MW, McWhirter JR and Wang JYJ. (1995). Mol. Cell. Biol., 15, 1286 ± 1293. Rotman G and Shiloh Y. (1999). Oncogene, 18, 6135 ± 6144. Sawyers CL. (1999). N. Engl. J. Med., 340, 1330 ± 1340. Review. No abstract available. Schwartzberg PL, Stall AM, Hardin JD, Bowdish KS, Humaran T, Boast S, Harbison ML, Robertson EJ and Go€ SP. (1991). Cell, 65, 1165 ± 1175. Shafman T, Khanna K, Kedar P, Spring K, Kozlov S, Yen T, Hobson K, Gatei M, Zhang N, Watters D, Egerton M, Shilo Y, Kharbanda S, Kufe D and Lavin M. (1997). Nature, 387, 520 ± 523. Sionov RV, Moallem E, Berger M, Kazaz A, Gerlitz O, BenNeriah Y, Oren M and Haupt Y. (1999). J. Biol. Chem., 274, 8371 ± 8374. Skorski T, Bellacosa A, Nieborowska-Skorska M, Majewski M, Martinez R, Choi JK, Trotta R, Wlodarski P, Perrotti D, Chan TO, Wasik MA, Tsichlis PN and Calabretta B. (1997). EMBO J., 16, 6151 ± 6161. Skorski T, Kanakaraj P, Nieborowska-Skorska M, Ratajczak MZ, Wen S-C, Zon G, Gewirtz AM, Perussia B and Calabretta B. (1995). Blood, 68, 726 ± 736. Soengas MS, Alarcon RM, Yoshida H, Giaccia AJ, Hakem R, Mak TW and Lowe SW. (1999). Science, 284, 156 ± 159. Taagepera S, McDonald D, Loeb JE, Whitaker LL, MacLeod AK, Wang JYJ and Hope TJ. (1998). Proc. Natl. Acad. Sci. USA, 95, 7457 ± 7462.

Takao N, Mori R, Kato H, Shinohara A and Yamamoto K. (2000). J. Biol. Chem., 275, 725 ± 728. Tan X and Wang JYJ. (1998). Trends Cell Biol., 8, 116 ± 120. Tan X-Q, Martin S, Green D and Wang J. (1997). J. Biol. Chem., 272, 9613 ± 9616. Theis S and Roemer K. (1998). Oncogene, 17, 557 ± 564. Tournier C, Hess P, Yang DD, Xu J, Turner TK, Nimnual A, Bar-Sagi D, Jones SN, Flavell RA and Davis RJ. (2000). Science, 288, 870 ± 874. Tybulewicz VLJ, Crawford CE, Jackson PK, Bronson RT and Mulligan RC. (1991). Cell, 65, 1153 ± 1163. Van Etten RA, Jackson P and Baltimore D. (1989). Cell, 58, 669 ± 678. Van Etten RA, Jackson PK, Baltimore D, Sanders MC, Matsudaira PT and Janmey PA. (1994). J. Cell Biol., 124, 325 ± 340. Wang JYJ. (1993). Curr. Opin. Genet. Dev., 3, 35 ± 43. Wang JYJ. (2000a). Nature, 405, 404 ± 405. Wang JYJ. (2000b). Signaling Networks and Cell Cycle Control. The Molecular Basis of Cancer and Other Diseases. In Cancer Drug Discovery and Development. Gutkind, J.S. (ed.). Humana Press Inc.: Totowa NJ, pp 303 ± 324. Welch PJ and Wang JYJ. (1993). Cell, 75, 779 ± 790. Welch PJ and Wang JYJ. (1995). Mol. Cell. Biol., 15, 5542 ± 5551. Wen S-T, Jackson PK and Van Etten RA. (1996). EMBO J., 15, 1583 ± 1595. Yang A, Walker N, Bronson R, Kaghad M, Oosterwegel M, Bonnin J, Vagner C, Bonnet H, Dikkes P, Sharpe A, McKeon F and Caput D. (2000). Nature, 404, 99 ± 103. Yuan Z-M, Huang Y, Whang Y, Sawyer C, Weichselbaum R, Kharbanda S and Kufe D. (1996). Nature, 382, 272 ± 274. Yuan Z-M, Huang Y, Ishiko T, Kharbanda S, Weichselbaum R and Kufe D. (1997). Proc. Natl. Acad. Sci. USA, 94, 1437 ± 1440. Yuan ZM, Shioya H, Ishiko T, Sun XG, Gu JJ, Huang YY, Lu H, Kharbanda S, Weichselbaum R and Kufe D. (1999). Nature, 399, 814 ± 817.