Sharon L. McKenna, Cork Cancer Research Centre, Biosciences Institute, University College Cork, Ireland. E-mail: email@example.com
Summary. Chronic myeloid leukaemia invariably progresses from a drug-sensitive to a drug-resistant, aggressive acute leukaemia. The mechanisms responsible for this are unknown, although loss of p53 has been reported in ≈ 25% of cases. Elevated expression of Bcr-Abl is also associated with disease progression. We have shown that cells expressing high levels of Bcr-Abl also express elevated levels of p53 and the cell cycle inhibitor, p21WAF-1. Despite this, cells continue to cycle and are drug resistant. As p21WAF-1 inhibitory activity is associated with nuclear localization, we investigated its localization in Bcr-Abl-expressing cells, and found that it is predominantly cytoplasmic. We have also shown that it associates physically with the serine/threonine kinase AKT, but this association and the cytosolic location of p21WAF-1 are phosphinositide-3-kinase (PI3K) independent. Cytosolic p21WAF-1 has been reported to have a prosurvival role in other transformed cells. In Bcr-Abl-expressing cells, p21WAF-1 rapidly diminishes as the cells are sensitized to apoptosis, using the inhibitor STI571. It is possible therefore that p21WAF-1 could also have a positive, prosurvival role in these cells. This study suggests that, by retaining p21WAF-1 in a cytosolic location, Bcr-Abl can evade the cell cycle arrest normally induced by nuclear p21WAF-1 and therefore also enable the cells to negate an important feature of a tumour suppressor response.
The Bcr-Abl chimaeric oncoprotein exhibits constitutive tyrosine kinase activity and is believed to be the critical determinant in the pathogenesis of chronic myeloid leukaemia (CML). CML is a clonal disorder of the haematopoietic system, resulting in the excessive accumulation of immature and mature myeloid cells. It initially presents as a chronic disease that can be successfully controlled with relatively mild therapeutic agents. After a period of ≈ 3–5 years (in the absence of stem cell transplantation), this is then followed by a much more aggressive and highly drug-resistant stage referred to as blast crisis (BC). Despite considerable efforts, the main reasons why BC-CML is so difficult to treat remain elusive.
Disease transition to BC has been associated with p53 mutations in ≈ 25% of cases (Ahuja et al, 1991). Another common secondary abnormality is elevation of Bcr-Abl expression. We have shown previously that elevated Bcr-Abl expression is sufficient to confer a drug-resistant phenotype in myeloid cells (Keeshan et al, 2001). In contrast to cells expressing low levels of Bcr-Abl, high Bcr-Abl-expressing cells fail to undergo cell cycle arrest and apoptosis in the presence of drugs, despite constitutive elevation of the tumour suppressor gene p53 and concomitant elevation of the cell cycle inhibitor p21WAF-1. This ability of Bcr-Abl to evade the cell cycle inhibitory functions of p21WAF-1 may provide a key survival/proliferative advantage to these cells. The mechanism by which high Bcr-Abl expression negates p21WAF-1 inhibitory activity is unknown.
Progression through the cell cycle is regulated by several cyclin-dependent kinases (CDKs) with activities that are regulated by cyclin-dependent kinase inhibitors (CKIs) (Sherr & Roberts, 1999). The transition from G1 to S phase is initiated by the expression of D-type cyclins and their assembly into kinase complexes with CDK4 and CDK6. Subsequently, the retinoblastoma protein (pRb) is maximally phosphorylated involving cyclinE-CDK2, allowing the release of E2F and S phase entry. The kinase inhibitor protein (CIP/KIP) family of cell cycle inhibitors includes p21WAF-1/CIP-1 (El-Deiry et al, 1993; Harper et al, 1993), p27KIP-1 (Polyak et al, 1994; Toyoshima & Hunter, 1994) and p57KIP-2. (Lee et al, 1995; Matsuoka et al, 1995). They are known to be broad-based cyclin inhibitors; however, recent evidence suggests that they may be more potent inhibitors of cyclins E and A, and can also be positive regulators of cyclin D (LaBaer et al, 1997; Cheng et al, 1999; Sherr & Roberts, 1999). p21WAF-1 also induces cell cycle arrest by binding to proliferating cell nuclear antigen (PCNA). It is generally accepted that p53-dependent cell cycle arrest occurs as a result of p21WAF-1-mediated inhibition of CDKs and PCNA (El-Deiry et al, 1994), allowing either repair of DNA damage or apoptosis.
Recent studies have shown that p21WAF-1 can also have an antiapoptotic function in the cytosol, by binding to apoptosis signal regulating kinase 1 (ASK1) and inhibiting the activation of the stress-activated protein kinase SAPK/JNK (Asada et al, 1999). A decrease in or loss of p21WAF-1 expression in cells that would normally undergo cell cycle arrest instead leads to apoptosis (Gorospe et al, 1997). This seemingly antiapoptotic function of p21WAF-1 is further supported by studies demonstrating the requirement of p21WAF-1 in cell survival (Poluha et al, 1996; Asada et al, 1998). The AKT signalling pathway has been reported to be responsible for p21WAF-1-mediated survival (Fujio et al, 1999; Lawlor & Rotwein, 2000). Therefore, it is apparent that the activity of p21WAF-1 is dependent upon its protein interactions at specific stages of the cell cycle, and its inhibitory activity seems to rely specifically upon its interaction with nuclear targets.
The cytoplasmic retention of p21WAF-1 has been shown to be important for the growth and survival of HER-2/neu-overexpressing cancer cells. These cells express elevated p21WAF-1, which is sequestered in the cytosol because of its physical association with the serine/threonine kinase AKT (Zhou et al, 2001). This study demonstrated that phosphorylation of p21WAF-1 on threonine 145 by AKT triggered its cytoplasmic sequestration. However, in another study, the phosphorylation of this site by AKT was found to promote endothelial cell proliferation by abrogating PCNA binding to p21WAF-1 (which interferes with DNA replication) rather than its cytoplasmic sequestration (Rossig et al, 2001).
Bcr-Abl can activate several signalling pathways, which are normally associated with receptor tyrosine kinases (Di Bacco et al, 2000). Bcr-Abl has also been reported to promote cell cycle entry directly by the activation of CDK2 (Cortez et al, 1997) and downregulation of the CKI p27KIP-1 (Jonuleit et al, 2000). The phosphinositide-3-kinase (PI3K)/phosphate and tensin homologue deleted from the chromosome 10 (PTEN)/AKT pathway has been shown to be a critical regulator of survival and cell cycle progression in Bcr-Abl-expressing cells (Skorski et al, 1997). The antiapoptotic and prosurvival role of AKT has been reported in several studies to be mediated through the phosphorylation of apoptotic regulators including Bad, caspase 9, glycogen synthase kinase 3 (GSK3), p70S6K and members of the forkhead transcription family (Krasilnikov, 2000). However, the downstream targets of AKT in Bcr-Abl-expressing cells are not well defined, as both Bad-dependent and -independent pathways have been identified (Neshat et al, 2000).
In this study, we have shown an association between Bcr-Abl expression levels and p21WAF-1 expression. We have investigated the localization of p21WAF-1 in Bcr-Abl-expressing cells and the role of the PI3K/AKT pathway. In drug-sensitive, low Bcr-Abl-expressing cells, p21WAF-1 is expressed only in response to drug treatment. It is predominantly nuclear, and the cells arrest and undergo apoptosis. In contrast, in drug-resistant high Bcr-Abl-expressing cells, p21WAF-1 is constitutively expressed and is predominantly cytoplasmic. p21WAF-1 is associated with AKT in these cells, but this association and cellular localization are PI3K independent. These data suggest that, by retaining p21WAF-1 in a cytosolic location, high Bcr-Abl-expressing cells could negate a cell cycle inhibitory pathway that is central to the p53 tumour suppressor response, and may therefore provide these cells with a significant survival advantage.
Materials and methods
Cell culture/reagents. The 32D cell line was maintained in Roswell Park Memorial Institute (RPMI) 1640, 10% fetal calf serum (FCS), 1% penicillin–streptomycin (P/S) (Gibco BRL, Paisley, UK) and 10% Walter and Eliza Hall Institute (WEHI) conditioned media (WEHI-CM) as a source of interleukin (IL)-3. Bcr-Abl-transformed 32D cells (32Db2a2) were cloned by serial dilution in media containing 15% FCS and 1·7 µg/ml puromycin (Sigma-Aldrich, Dublin, Ireland). Five clones (C2, C4, C5, C7 and C8) were selected for assessment of Bcr-Abl expression. Two representative clones expressing low and high levels of Bcr-Abl (C2 and C4 respectively) were selected and chosen for further study. The clones were maintained in RPMI 1640, 10% FCS, 1% P/S and 0·2 µg/ml puromycin. 32D cells were transfected with pBABE Puro empty vector by electroporation, selected in 1·7 µg/ml puromycin and maintained in 10% FCS, 10% WEHI-CM, 1% P/S and 0·2 µg/ml puromycin. LY294002 (Calbiochem, Darmstadt, Germany), etoposide (VP16, Sigma-Aldrich) and STI571 were kindly provided by Novartis, Basel, Switzerland,
Assessment of cell viability and number. Cell viability was assessed by propidium iodide (PI) exclusion on a FACScan (Becton Dickinson, Oxford, UK) flow cytometer at 590 nm (forward scatter, FL-2). Cells were incubated with 50 µg/ml PI at room temperature (RT) before analysis. The criteria for cell death as measured by flow cytometry were based on changes in light scattering properties of dead cells resulting from cell shrinkage and increased granularity and also permeability to PI. The trypan blue exclusion assay was used to determine cell viability and cell number using a Neubauer haemocytometer.
Measurement of phosphatidyl serine (PS) exposure. The exposure of PS on the extracellular surface of the plasma membrane was monitored by the binding of annexin V–fluorescein isothiocyanate (FITC), according to the manufacturer's instructions (IQProducts, LabRon Ltd, Dublin, Ireland). Briefly, 5 × 105/ml cells were resuspended in calcium buffer (10 mmol/l Hepes, 2·5 mmol/l CaCl2, 140 mmol/l NaCl) and incubated with annexin V–FITC for 5 min at RT in the dark. Cells were incubated with 50 µg/ml PI at RT before analysis. Fluorescence resulting from FITC and PI was measured at 530 nm (FL1) and 590 nm (FL2), respectively, and analysed using cellquest software on a FACScan flow cytometer (Becton Dickinson) using an excitation of 488 nm.
Cell lysis and immunoblotting. The antibodies used were: anti-c-Abl (ab-3), anti-p21WAF-1 (ab-4), anti-poly (ADP-ribose) polymerase (PARP; Calbiochem, Darmstadt, Germany), anti-cyclin D and anti-PTEN (Upstate Biotechnology, Lake Placid, NY, USA), anti-AKT and anti-phospho-AKT (ser-473) (Cell Signalling Technology, Hertfordshire, UK) and anti-phosphotyrosine (PY20). Cells were lysed in modified radioimmunoprecipitation (RIPA) buffer [50 mmol/l Tris-HCl, pH 7·4, 1% Igepal, 0·25% sodium deoxycholate, 150 mmol/l NaCl, 1 mmol/l EDTA, 1 mmol/l Na3VO4, 1 mmol/l NaF, 1 µg/ml antipain, 1 µg/ml aprotinin, 1 µg/ml chymostatin, 0·1 µg/ml leupeptin, 1 µg/ml pepstatin and 100 µmol/l phenylmethylsulphonyl fluoride (PMSF)]. Equivalent amounts of protein were resolved using sodium dodecyl sulphate = polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membrane (Schleicher and Schuell, Dassel, Germany). All secondary antibodies were peroxidase conjugated, and proteins were detected using enhanced chemiluminescence (ECL; Amersham, Little Chalfont, UK).
(Co)immunoprecipitations. Untreated and treated cells (5 × 106/sample) were lysed in 500 µl of lysis buffer (20 mmol/l Tris, 50 mmol/l NaCl, 1% Igepal, 50 mmol/l NaF, 1 µmol/l NaVO4, 1 mmol/l PMSF, 1 µg/ml antipain, 1 µg/ml aprotinin, 1 µg/ml chymostatin, 0·1 µg/ml leupeptin, 1 µg/ml pepstatin). AKT and p21WAF-1 were immunoprecipitated using anti-AKT (Cell Signalling Technology) and anti-WAF-1 (ab-4; Calbiochem) and rotated overnight at 4°C, followed by anti-rabbit or anti-mouse IgG whole-molecule agarose (Sigma-Aldrich) for 1·5 h at 4°C. The beads were pulse centrifuged (12 000 r.p.m. for 5 s) at all times. Control immunoprecipitations were performed to check non-specific binding of beads (no primary added). Western blotting was carried out as described.
Subcellular fractionation. Nuclear and cytoplasmic fractions were obtained using the NE-PERTM nuclear and cytoplasmic extraction kit (Pierce, Medical Supply Co., Dublin, Ireland). Briefly, untreated and treated cells (12 × 106/sample) were lysed using cytoplasmic extraction reagent (CER) I and II, centrifuged at 16 000 g to obtain the nuclear pellet. The supernatant contained the cytosolic fraction. The pellet was washed five times in Tris-buffered saline (TBS) and lysed using the nuclear extraction reagent (NER) for 40 min on ice. The sample was centrifuged at 16 000 g, and the resultant supernatant was the nuclear fraction. After Bio-Rad protein determination, equal quantities of nuclear and cytosolic fractions were resolved using SDS-PAGE, transferred to nitrocellulose membrane and immunoblotted. Nuclear and cytosolic fractions were verified using antibodies detecting PARP and paxillin respectively.
Cell cycle analysis by bromodeoxyuridine (BrdU) incorporation. Cells (2 × 106/sample) were incubated in 30 μmol/l LY294002 for 24 h and then pulse-labelled with 10 µmol/l BrdU (Sigma-Aldrich) for 4 h at 37°C. BrdU was washed out, and cells were resuspended in 1 ml of PBS/5 mmol/l EDTA and fixed by slowly adding ice-cold 100% ethanol to give a final concentration of 70% for 20 min on ice. Cells were centrifuged and resuspended in 0·5 ml of washing buffer [0·5% Tween-20 in PBS, 0·5% bovine serum albumin (BSA)] and 0·5 ml of 2 N HCl for 30 min at RT to partially denature the DNA. Cells were washed again and resuspended in borax buffer (0·1 mol/l sodium tetraborate-10-hydrate) for 5 min at RT. Cells were incubated with monoclonal anti-BrdU antibody (1:500 dilution in washing buffer; Sigma-Aldrich) for 1 h at 4°C. Incorporated BrdU was detected after incubation of cells with goat anti-mouse-FITC conjugated antibody (1:50, Sigma-Aldrich) for 30 min in the dark at 4°C and counterstaining with PI (20 μg/ml PI and 200 μg/ml DNase-free RNase in buffer containing 3·4 mmol/l trisodium citrate, 9·65 mmol/l NaCl and 0·03% Igepal) for 30 min at 4°C in the dark. Fluorescence resulting from FITC and PI was measured at 530 nm (FL1) and 590 nm (FL2), respectively, and analysed using cellquest software on a FACScan flow cytometer (Becton Dickinson) using an excitation of 488 nm. Single cell populations were gated on FL2-W (width) and FL2-A (area) to exclude fluorescence from cell doublets/triplicates.
Statistical analysis. Data sets were compared using the Student's t-test, and differences were considered significant when P < 0·05.
Bcr-Abl expression levels correlate with p21WAF-1 and cyclin D expression
In our previous study, we reported that cells expressing high levels of Bcr-Abl were drug resistant and actively cycling despite p53 and p21WAF-1 elevation (Keeshan et al, 2001). We have shown that p21WAF-1 is only expressed in parental cell lines after drug treatment, whereas it is constitutively elevated in Bcr-Abl-expressing cells (e.g. 32D Bcr-Abl clone 4) (Fig 1A; Keeshan et al, 2001). We have also noted that p21WAF-1 expression is more strongly elevated in high Bcr-Abl-expressing cells (e.g. clones 4 and 5) (Fig 1A and B). Despite the expression of this cell cycle inhibitor, the cell lines continue to proliferate. In this study, we have examined further the relationship between Bcr-Abl and p21WAF-1 expression in clones C2 (low Bcr-Abl expression) and C4 (high Bcr-Abl expression). We have also demonstrated previously that the drug resistance of C4 cells can be reversed using the specific Bcr-Abl inhibitor STI571, and complete inhibition of Bcr-Abl kills all clones in the absence of IL-3 (as with the parental cells 32D), indicating that they have not acquired any additional survival mechanisms as a consequence of clonal variation (Keeshan et al, 2001, 2002).
D-type cyclins have been shown to co-operate with ABL oncogenes (Afar et al, 1995). p21WAF-1 has also been reported to stabilize cyclin D and to promote the assembly of cyclin D/CDK complexes (LaBaer et al, 1997; Cheng et al, 1999) Therefore, we also assessed the levels of cyclin D1 in 32D parental cells, C2, C4 and vector only (32DPuro) cells. 32D and 32DPuro cells did not express detectable levels of cyclin D1; however, when Bcr-Abl was expressed at low levels, cyclin D1 was expressed concomitantly. Indeed, high levels of Bcr-Abl expression led to a further increase in cyclin D1 expression levels (Fig 1C). These results indicate that the level of p21WAF-1 and cyclin D1 is dependent upon the level of Bcr-Abl expression.
P21WAF-1 is located in the cytoplasm of C4 cells
The cell cycle inhibitory function of p21WAF-1 is closely associated with its nuclear localization (El-Deiry et al, 1994; Ritt et al, 2000). As C4 cells (high Bcr-Abl) express high levels of p21WAF-1 and continue to proliferate in culture (Keeshan et al, 2001), we investigated the subcellular localization of p21WAF-1. Immunoblot analysis of cytoplasmic and nuclear fractions from C2 and C4 cells show that p21WAF-1 is predominantly located in the cytoplasm of C4 cells (Fig 2A).
We then investigated localization of p21WAF-1 in the various cell lines after cytotoxic drug treatment. 32D, C2 and C4 cells were treated with etoposide, and nuclear and cytoplasmic extractions were performed at 4 and 8 h (Fig 2B). At both time points, etoposide-induced p21WAF-1 was located predominantly in the nucleus in 32D and C2 cells, with some cytosolic protein also evident. These cells arrested and underwent apoptosis as described previously (Keeshan et al, 2001). Etoposide treatment of the resistant C4 cells increased the level of nuclear p21WAF-1 without significantly altering the cytoplasmic expression. These results demonstrate that p21WAF-1, which was elevated due to high Bcr-Abl expression, localized to the cytoplasm, and this was distinct from the nuclear p21WAF-1, which was induced by a cytotoxic stimulus.
Inhibition of PI3K induces cell cycle arrest in Bcr-Abl-positive and -negative cells
The PI3K/AKT pathway is required for the transformation of haematopoietic cells by Bcr-Abl (Skorski et al, 1997), and has been reported to play a role in the regulation of p21WAF-1 expression and localization (Mitsuuchi et al, 2000).Indeed, we have also found that an important negative regulator of this pathway, PTEN, is downregulated in C4 high Bcr-Abl-expressing cells (Fig 3A). PTEN is a tyrosine and serine/threonine phosphatase with lipid phosphatase activity (Vazquez & Sellers, 2000). It has been demonstrated to dephosphorylate phosphatidylinositol(3,4,5)triphosphate (PIP-3), a product of PI3K (Maehama & Dixon, 1998), and thus functions in the negative regulation of AKT. We therefore assessed the activation status of this pathway in 32D, C2 and C4 cells. PI3K was activated at a basal level in all these cell lines in the presence of serum-derived growth factors. However, even in growth factor-depleted culture conditions, high Bcr-Abl-expressing cells (C4) still showed considerable AKT phosphorylation. PI3K phosphorylates AKT on serine 473 and using an antibody directed against AKT-phosphoserine-473, we showed that, although the level of AKT was unaltered in 32D, C2 and C4 cells, there was an increase in the level of AKT-phosphoserine-473 in C4 cells, which was most evident after serum depletion (Fig 3B). The C4 cells may be predisposed to increased AKT activity by both downregulation of PTEN and Bcr-Abl activation of the PI3K pathway. An examination of the subcellular distribution of AKT indicated that it was also a predominantly cytoplasmic protein (Fig 3C).
Cytoplasmic p21WAF-1 associates with AKT in cells expressing elevated Bcr-Abl and cannot be altered by inhibition of PI3K
It has been demonstrated recently that AKT can physically associate with p21WAF-1 and mediate its cytosolic sequestration (Zhou et al, 2001). To investigate whether AKT and p21WAF-1 were associating in C4 cells, we carried out co-immunoprecipitation experiments. When p21WAF-1 was immunoprecipitated from C4 cells, AKT was found to be associated with it, as assessed by Western blotting (Fig 4A).
We then assessed the importance of PI3K activity for the proliferation of these cells, and examined whether this pathway influences p21WAF-1/AKT association and the cytoplasmic localization of p21WAF-1. The inhibitor LY294002 was used specifically to inhibit PI3K activity. The concentration of inhibitor required to reduce phosphorylation of AKT was determined initially. To obtain a strong positive control for AKT phosphorylation, 32D cells were stimulated with IL-3 after a period of starvation (which elevates phosphorylated AKT). The cells were then cultured in the presence and absence of LY294002. Figure 4B shows that, in the presence of 20 µmol/l LY294002, AKT phosphorylation was completely abolished in 32D, C2 and C4 cells. In addition, cell proliferation was completely blocked in C2 and C4 cells in the presence of 20 µmol/l and 30 µmol/l LY294002, without any induction of cell death as assessed by Trypan blue exclusion (Fig 4C). The affect was similar in 32D cells although at 48 h there was a slight increase in cell number. We therefore assessed the cell cycle profiles after pulse labelling with BrdU in 32D, C2 and C4 cells in the presence of 30 µmol/l LY294002 for 24 h. As Fig 4D shows, incubation with LY294002 significantly reduced the percentage of DNA-synthesizing cells (S phase) as assessed by BrdU incorporation (P < 0·01). G1 arrest was more pronounced in Bcr-Abl-expressing cells, suggesting that they are slightly more dependent upon PI3K signalling than the 32D cell line. The inhibition of PI3K lead to a G1 cell cycle arrest in Bcr-Abl-positive cells, irrespective of the level of Bcr-Abl expression.
To determine whether the association of p21WAF-1 and AKT was dependent on PI3K activity, we performed immunoprecipitations in the presence and absence of 30 µmol/l LY249002. Despite cell cycle arrest, no alteration between AKT and p21WAF-1 association was evident after PI3K inhibition (Fig 5A). We also investigated the effect of PI3K inhibition on p21WAF-1 localization in C4 cells. Cytoplasmic and nuclear extractions were performed in the presence of 30 µmol/l LY249002. At the indicated time points, we consistently detected p21WAF-1 in the cytoplasmic fractions, without alteration in the level of expression of p21WAF-1 (Fig 5B). These data demonstrate that, in C4 cells, AKT and cytoplasmic p21WAF-1 physically associate and that this association and the cytoplasmic location of p21WAF-1 is not dependent on the activity of PI3K.
Cytoplasmic p21WAF-1 expression decreases after inhibition of Bcr-Abl kinase activity, and increased susceptibility to apoptosis
As cytoplasmic p21WAF-1 has also been associated with resistance to apoptosis, we assessed whether the expression levels and/or distribution of this protein would be altered with a Bcr-Abl inhibitor (STI571) at concentrations known to reduce the drug resistance of C4 cells. We have shown previously that cells with a high expression of Bcr-Abl are sensitized to apoptosis using concentrations as low as 0·5 µmol/l STI571 (Keeshan et al, 2001). Using concentrations of STI571 up to 50 µmol/l for 4 h, Bcr-Abl tyrosine kinase activity was reduced (as assessed by Western blotting of phosphotyrosine) to a level at which the cells started to die with the inhibitor alone (assessed by annexin V/PI staining) (Fig 6A and B). When the kinase activity of Bcr-Abl was reduced using 10 µmol/l STI571, the cells were not apoptotic, as shown by staining with the early apoptotic marker annexin V (Martin et al, 1995), but this was more than enough STI571 to sensitize the cells to apoptosis. We therefore performed cytosolic and nuclear extractions to examine whether moderate Bcr-Abl inhibition could facilitate the migration of p21WAF-1 into the nucleus. Although a marginal increase in nuclear p21WAF-1 was evident at 20 µmol/l (Fig 6C), expression levels were also diminished within 4 h and, therefore, significant translocation cannot be established before protein levels are depleted. The separation of expression and distribution may only be possible upon further dissection of the responsible signalling pathways.
This data, however, further confirms the correlation between Bcr-Abl activity in C4 cells and the levels of cytosolic p21WAF-1 as, when Bcr-Abl activity was diminished, a similar reduction in the levels of cytosolic p21WAF-1 was observed. In addition, decreased p21WAF-1 expression also correlated with increased susceptibility to apoptosis in C4 high Bcr-Abl-expressing cells.
The data presented in this study suggests a novel Bcr-Abl prosurvival mechanism. p21WAF-1 is elevated in high Bcr-Abl-expressing cells; however, this does not lead to cell cycle arrest but, rather, cells continue to proliferate and are resistant to apoptosis. Our data suggest that this may be related to the predominantly cytosolic location of p21WAF-1.
We have shown that increased Bcr-Abl expression levels correlate with increased cytosolic p21WAF-1 expression. In addition to the inhibitory activity associated with nuclear p21WAF-1, this protein is also known to have a complex relationship with cyclin D, which ultimately promotes proliferation. Expression of p21WAF-1 can promote cyclin D expression and vice versa (Hiyama et al, 1997). 32D cells do not express cyclin D, and therefore it is not required for proliferation. However, cyclin D has been reported to co-operate with Bcr-Abl in transformation (Krasilnikov, 2000). We found that high Bcr-Abl-expressing cells (C4) express relatively high levels of cyclin D. It is possible therefore that the elevated p21WAF-1 in these cells may play a positive role in the co-operation with cyclin D.
The best-known tumour suppressor response is the p53 response to DNA damage. This involves stabilization of p53 and upregulation of nuclear p21 leading to cell cycle arrest (El-Deiry, 1998). When 32D and C2 cells are treated with drugs, expression of p53 and p21WAF-1 is associated with arrest followed by apoptosis (Keeshan et al, 2001). We have shown a high nuclear expression of p21WAF-1 in this case. C4 cells can also elevate nuclear p21WAF-1 in response to drugs, but the cytoplasmic p21WAF-1 is retained, and this does not lead to arrest and apoptosis. The tumour suppressor p53 is also reported to have a less well-defined role in ‘sensing’ oncogene deregulation (Lowe, 1999). In the presence of high levels of Bcr-Abl, p53 and p21WAF-1 are elevated, but cells continue to cycle (Keeshan et al, 2001). This suggests that an additional signalling mechanism in these cells facilitates the cytoplasmic retention of p21WAF-1. Although we have reported an elevated p53 level in our high Bcr-Abl-expressing cell line (Keeshan et al, 2001), we do not know whether this is the main mechanism leading to the upregulation of p21WAF-1. Other transcription factors can also transcriptionally upregulate p21WAF-1, including E2F and several signal transducers and activators of transcription (STATs) (Gartel & Tyner, 1999). Bcr-Abl can also activate STATs; therefore these may play a role in the elevation of p21WAF-1. Regardless of the mechanism by which p21WAF-1 was elevated, it appeared not to have a significant inhibitory function in these cells.
Bcr-Abl-expressing cells are known to stimulate the PI3K pathway strongly (Skorski et al, 1997). Moreover, we have shown that high Bcr-Abl-expressing cells downregulate the PTEN phosphatase, which negatively regulates this pathway. Cells lacking the Pten gene have been reported to increase AKT activity, which is associated with advanced entry into S phase (Sun et al, 1999) and resistance to apoptosis (Stambolic et al, 1998). Furthermore, PTEN inactivation has been associated with haematological malignancies, including leukaemias (Dahia et al, 1999). Recent studies have suggested that p21WAF-1 localization and/or activity may be regulated by its phosphorylation by AKT (Rossig et al, 2001; Zhou et al, 2001). In our study, we found that cytoplasmic p21WAF-1 was indeed associated with phosphorylated AKT. However, although inhibition of the PI3K pathway dephosphorylated AKT, it did not eliminate binding to p21WAF-1 or influence its cytoplasmic location.
However, inhibition of the PI3K pathway did induce cell cycle arrest, which was evidently not associated with nuclear translocation of p21WAF-1. This arrest was particularly prominent in the Bcr-Abl-expressing cells. PI3K is known to promote cyclin D levels by phosphorylating and inhibiting the negative regulator GSK-3β (Diehl et al, 1998). Expression of the inhibitor p27KIP-1 can also be suppressed by PI3K in Bcr-Abl-expressing cells (Gesbert et al, 2000). Thus, there are a number of mechanisms by which PI3K inhibition may lead to cell cycle arrest.
Although cytoplasmic retention of p21WAF-1 appears to be PI3K independent, we cannot assume that it is also AKT independent, as there is evidence for the activation of AKT by other kinases (Kandel & Hay, 1999). Indeed, tyrosine phosphorylation of AKT by the tyrosine kinase, Src, both in vivo and in vitro, has been reported to contribute to its biological function (Chen et al, 2001). It is also possible that the cytoplasmic retention of p21WAF-1 is facilitated by other kinases, as p21WAF-1 is also a substrate for protein kinase C (PKC) and protein kinase A (PKA) (Scott et al, 2000; Rossig et al, 2001).
It is notable that the v-abl oncogene also elevates p21WAF-1 and cyclin D. However, p21WAF-1 is predominantly nuclear and the cells cannot proliferate (Khanna et al, 2001). When additional growth/survival factor (IL-3) is added, p21WAF-1 is modified (by a post-translational mechanism), and the cells can enter the cell cycle. In this case, p21WAF-1 is still found in the nucleus. Therefore, in these cells, a combination of oncogenic and normal growth factor signalling is required to negate the effects of a cell cycle inhibitor. It is possible that high Bcr-Abl signalling stimulates growth factor-associated signalling pathways that are not activated by low levels of Bcr-Abl, enabling p21WAF-1 cytosolic location and abrogation of the p21WAF-1 inhibitory activity.
Recently, there is more direct evidence for the antiapoptotic role of p21WAF-1 and the importance of its intracellular location in carrying out that role. Apoptotic binding partners of p21WAF-1 include ASK1 (Asada et al, 1999) and pro-caspase 3 (Suzuki et al, 1998). The caspase family of proteins are key regulators of apoptotic cell death, and cytoplasmic p21WAF-1 can interact with pro-caspase 3 to prevent Fas-mediated apoptosis (Suzuki et al, 1998; Steinman & Johnson, 2000). In addition, antisense constructs of p21WAF-1 have proved very effective in sensitizing leukaemic cells to drug-induced apoptosis (Wang et al, 1999; Sato et al, 2000). Although we have not established a p21WAF-1 antiapoptotic role in high Bcr-Abl-expressing cells, we have noted that decreased levels of this protein correlate with an increased susceptibility to apoptosis. We have shown previously that the kinase activity of Bcr-Abl can be reduced using STI571 at concentrations that do not kill the cells, but sensitize them to apoptosis (Keeshan et al, 2001). In this study, we have also demonstrated that p21WAF-1 expression levels are significantly reduced within 4 h in such non-apoptotic drug-sensitive cells. When STI571 concentrations were increased further to a concentration that will kill the cells, p21WAF-1 expression was severely diminished. It remains possible therefore that the increased cytoplasmic p21WAF-1 expression does contribute to the antiapoptotic features of the high Bcr-Abl-expressing cells.
In conclusion, we have identified a novel target of Bcr-Abl kinase signalling. p21WAF-1 expression is elevated in high Bcr-Abl-expressing cells. This may be a consequence of an attempted p53 tumour suppressor response or signalling originating from Bcr-Abl itself. Either way, p21WAF-1 is not effective as a cell cycle inhibitor as cells continue to proliferate. p21WAF-1 is predominantly cytosolic and is associated with the serine/threonine kinase AKT. However, this association and cytosolic location are PI3K independent. It remains possible that cytosolic p21WAF-1 contributes to the antiapoptotic phenotype of the Bcr-Abl-expressing cells. These data highlight the importance of p21WAF-1 subcellular localization for its biological activity, and provide a further insight into the Bcr-Abl-mediated survival mechanisms, which may contribute to aggressive CML.
This work was funded by grants from the Irish Cancer Society. We are also grateful to the Children's Leukaemia Research Project, the Health Research Board, Ireland, Cork Cancer Research Centre and Enterprise Ireland for financial support. We are grateful to Dr Ken Mills for cell lines and constructs. We also thank Dr Ruaidhri Carmody and Dr Susanne Vejda for helpful discussions and critical reading of the manuscript.