The oncogenic bcr-abl tyrosine kinase resulting from the Philadelphia chromosome (Ph) signals for leukemogenesis of CML. Imatinib mesylate, a specific inhibitor of c-abl and bcr-abl, induces apoptosis of CML cells and has created a revolutionary success in CML treatment.1 Imatinib down-regulates various antiapoptotic signals including bcl-XL, Akt kinase and NFκB in CML cells.2 Nevertheless, the proapoptotic mechanism elicited by imatinib has not been well-defined. p53 generally generates an important proapoptotic effect on cells under genotoxic and non-genotoxic stress. However, the p53 level in the imatinib-treated CML cells was reduced and failed to accumulate in response to DNA damage. This is explained by the fact that constitutive phosphorylation of p53 at serine 20 is inhibited by imatinib.3 Besides, CML cells with wild type p53 are more resistant to imatinib than those lacking p53.3 Therefore, we rationalized the existence of an imatinib-induced p53-independent proapoptotic mechanism.
p73 is a homologue of the p53 tumor suppressor and induces apoptosis by different mechanisms.4, 5 Unlike p53, there are several different isomers of p73 resulting from alternative promoter utilization and splicing. TAp73s are isomers that contain an N-terminal transactivation domain that is lacking in the ΔNp73 isomers. TAp73 transactivates p53 target genes and induces apoptosis, whereas the ΔN isomers are dominant inhibitors of the p53-responsive gene expression. Interestingly, the proapoptotic functions of TAp73 and p53 cannot be substituted equally by each other. Under some p53-nonfunctional circumstances, apoptosis of tumor cells may be induced by ectopic p73 but not by ectopic p53.6 Knowing that p73 is one of the substrates of p38 MAPK, which can be activated by imatinib in BCR-ABL+ cells,7–9 it raises the possibility that the p38 MAPK-activated p73 axis is involved in the imatinib-induced proapoptotic machinery.
Pml is a tumor suppressor, which counteracts some proliferative oncogenic signals.10–14 Pml is provoked by ras to induce senescence in primary cells usually with the activation of p53.11–14 With activated ras, PML-NBs in Ph+ B- acute lymphoblastic leukemia (ALL) were increased both in number and size.15 Since ras is activated in both CML and Ph+ B-ALL cells,15, 16 it is intriguing to know whether response of pml/PML-NBs to bcr-abl/ras is the same in CML as in Ph+ ALL. Furthermore, pml was recently found in other cellular context to modulate p73 by recruiting phosphorylated p73 to PML-NBs where p73 is stabilized by prolyl isomerization and CBP/p300-mediated acetylation.17 As such, we addressed the roles of p38 MAPK, pml and p73 in a probable imatinib-induced p53-independent proapoptotic mechanism.
Interestingly, the functions of both pml and TAp73 are subcellular compartment-restricted.18, 19 Therefore, we studied their topographical interaction as well, especially at NBs, in relation to apoptosis induction.
Imatinib (Novartis Pharmaceuticals) was stored at −20°C as a 10 mM stock solution in dimethyl sulfoxide (DMSO). Imatinib mesylate and interferon-α-2a (IFNα) (Roche Pharmaceuticals) at the final concentrations of 0.5 μM and 1 × 105 unit/ml, respectively, led to 50% inhibition of the 3-day growth of K562 cells (data not shown). The caspase inhibitor, Boc-Asp(OMe) Fluoromethyl Ketone (BOC) (Sigma, St. Louis, MO) at the final concentration of 200 μM was able to completely block the cleavage of poly (ADP-ribose) polymerase in K562 cells (data not shown). SB203580 (SB), 4-(4-fluorophenyl)-2-(4-methylsulfinylphenyl)-5-(4-pyridyl)-1H-imidazole, (Sigma, St. Louis, MO) was used as a p38 MAPK inhibitor at 10 μM, the minimal concentration titrated to suppress phosphorylation of TAp73. The antibodies used were: mouse anti-phosphotyrosine, 4G10, monoclonal (Upstate, Lake Placid, NY) for Tyr-phosphorylated TAp73; rabbit anti-p38 MAPK and anti-phospho-p38 MAPK (Thr180/Tyr182) polyclonal (Cell Signaling, Beverly, MA); rabbit anti-p73, H-79, polyclonal, directing at the N-terminal amino acids 1-80 of p73 (Santa Cruz Biotechnology, Santa Cruz, CA); mouse anti-pml, PG-M3 and anti-bax, B-9, monoclonal (Santa Cruz); and rabbit anti-phospho-chk2 (Thr68) polyclonal (Cell Signaling). Alkaline phosphatase-conjugated anti-rabbit or anti-mouse immunoglobin G (IgG) (Promega, Madison, WI) was used as secondary antibodies to detect primary antibody and enabled color-development with CDP-star chemiluminescent reagents (Bio-Rad, Hercules, CA).
Cell lines and primary CML bone marrow cells
BCR-ABL+ K562, Meg-01CML cell lines and BCR-ABL− HL60 acute myeloid leukemia cell line were obtained from the American Type Culture Collection (Rockville, MD) and were grown in RPMI 1640 medium supplemented with fetal bovine serum (FBS). All the 3 cell lines harbor p53mutant genes. In K562 and HL60 cell lines, normal p53 gene is lacking in both alleles; and in Meg-01 cell line, there is a triplet deletion in one p53 allele.20, 21 In agreement with the findings documented, we confirmed that normal p53 was not detectable by Western analysis in 562 cells. In Meg-01 cells, since the mutant p53 was indistinguishable from the wild type by resolution in SDS-PAGE, the p53 band in Western blot may contain both mutant and wild type p53. Notwithstanding, p53 did not accumulate more on exposure to imatinib (Supporting Information Fig. S1). Using reverse transcription polymerase chain reaction (RT-PCR)-subcloning-sequencing, we examined the mutation hot spot (exon 4 to exon 11) of p53 genes in bone marrow cells obtained from 6 untreated CML patients after informed consents were obtained. The mononuclear bone marrow cells were isolated by density gradient before being processed for confocal laser scanning microscopy. Three of the 6 patients were in chronic phase, and the other 3 in accelerated phase. Mutant p53 genes were found in 2 patients who were in accelerated phase while no p53 gene mutation was detected in cells from the other 4 patients (Supporting Information Fig. S2).
Interference with TAP73 and PML RNAs by siRNAs
The siRNA duplexes were designed using Quiagen (Hilden, Germany) siRNA design tool according to standard Tuschl-based guidelines. Chemically synthesized siRNA-duplexes with 2′-deoxythymidines instead of uridine residues in the 3′ overhangs to enhance nuclease resistance and the non-silencing mock siRNA (negative control) were purchased from Quiagen (Cat # 1022076). The sequence of siRNA duplexes synthesized were r(GGCAAUAAUCUCUCGCAGU)d(TT) and r(ACUGCGA GAUUAUUGCC)d(TT) for targeting the AAGGCAAUAAUCU CUCGCAGU (nt 740–760) motif of human TP73α mRNA (GenBank accession number AY040827); r(UGCCAGGCGGAAGC CAAGU)d(TT) and r(ACUUGGCUUCCGCCUGGCA)d(TT) for the AATGCCAGGCGGAAGCCAAGT (nt 317–337) motif of PML mRNA (GenBank accession number X63131). The TP73α mRNA-targeting siRNA may also interfere with TP73β and ΔNp73 mRNAs. However, the TAp73 siRNA should interfere with TAP73 (TP73α and TP73β) mRNAs much more than with ΔNp73 mRNA, for the copy number of TAP73 mRNA is 46.4-fold more than that of ΔNP73 mRNA in K562 cells.22 Interference with PML or TAP73 RNA was conducted by electroporating 2 × 106 K562 cells with 3 μg of duplex siRNA using Nucleofector™ (Amaxa Biosystems, Cologne, Germany) with program T16 according to the manufacturer's instruction. Cells were then incubated in RPMI1640 containing 20% FBS for 48 hr with unhealthy cells removed by Ficoll-Hypaque gradient. Cells were then treated with imatinib in RPMI medium with 5% FBS for an indicated time before analysis.
Apoptotic fractions after various treatments were analyzed on a FACScan (Becton-Dickinson, San Jose, CA) for the annexin V+/ propidium iodide (PI)- fractions of cells stained with fluorecein-conjugated annexin V and PI (Roche Diagnostics, Indianapolis, IN).
Western blot analyses
After 24 hr treatment with imatinib and/or IFNα, cells were lysed with 2x sodium dodecyl sulfate (SDS) buffer supplemented with protease inhibitors and subjected to a SDS- polyacrylamide gel electrophoresis (PAGE). For cells of which pml or TAp73 to be knocked down, electroporation with siRNA was conducted 48hr before incubation with imatinib and/or IFNα.
Following treatment with imatinib for 24 hr, K562 cells were washed, and cell lysates were prepared by incubation for 30 min on ice in lysis buffer (50 mM Tris-Cl [pH8.0], 120 mM NaCl, 0.5% [v/v] NP40, 5 μg/ml leupeptin, 10 μg/ml aprotinin, 50 μg/ml PMSF, 0.2 mM sodium orthovanadate, 100 mM NaF) followed by drawing and pushing for 10 times through a 22-gauge needle. The resulting lysates were centrifuged at 15,000g, 4°C, to remove cellular debris. For immunoprecipitation reactions, each cell lysate was mixed gently with anti-pml antibody (PG-M3, Santa Cruz) or anti-p73 antibody (H-79, Santa Cruz) and incubated on ice for 1 hr. Protein G beads after washing with the lysis buffer were added to the protein/antibody mixture. The lysate/bead mixtures were incubated on a rotator overnight at 4°C. The immunoprecipitates were then separated from unbound protein by centrifuging briefly at 8,000 rpm and washed trice. Immunoprecipitates were eluted from the beads by boiling in 1x SDS buffer before loading for Western blot analyses.
Co-localization of TAp73 and pml at NB visualized by laser scanning confocal microscopy
Cells after treatment were fixed and doubly immunofluorescence-stained for pml and TAp73 with mouse anti-pml (1:25) (PG-M3, Santa Cruz) and goat anti-TAp73 (1:25) (S-20, Santa Cruz) antibodies, coupled with donkey fluorescein-conjugated anti-mouse and TRITC-conjugated anti-goat antibodies (Jackson ImmunoResearch, West Grove, PA), respectively. The cells doubly labeled were then visualized using a Leica TCS-SP2 confocal scanning microscope equipped with an acousto-optical tunable filter and a Plan-Apochromat oil-immersion objective with a 63 ×/1.32 numeric aperture (Leica Microsystems, Bensheim, Germany). Images were acquired with Leica TCSNT software. The topographic co-localization of TAp73 dots to PML-NBs was visible as merged yellow dots.
Imatinib induces apoptosis of CML cells whereas knockdown of pml or TAp73, inhibition of p38 MAPK or inhibition of caspase abolishes the induced apoptosis
The imatinib-induced apoptosis of K562 accumulated through 48 hr. Imatinib also induced apoptosis of BCR-ABL+ Meg-01 CML cells but not of BCR-ABL- HL60 cells (Fig. 1a). Of K562 cells the apoptosis was significantly induced by imatinib while abolished by the p38 MAPK inhibitor (SB) or the caspase inhibitor (BOC) (Fig. 1a), or by interference with PML or TAP73 RNA (Fig. 1b). Interference with PML and TAP73 RNAs resulted in about 70% knockdown of pml and TAp73 as estimated by Western blot analysis (Figs. 3a and 3b).
Imatinib induces apoptosis of CML cells with concurrent increased phosphorylation of p38 MAPK, TAp73 and chk2 and increased expression of pml and bax
Imatinib induced the phosphorylation of p38 MAPK, chk2 and TAp73, as well as the expression of pml, TAp73, and bax in K562 and Meg-01 cells, but not in BCR-ABL- HL60 cells (Figs. 2a and 2b). Pml was presented with several bands on immunoblot due to the presence of several isoforms and various phosphorylated and sumoylated forms.23 The representative pml band shown in Figures 2a, 2b and 3a has the apparent molecular weight between 100 kD and 72 kD, which is most prominent by being probed with the PG-M3 antibody; meanwhile, the probing is most obviously blocked by the synthetic peptide with the epitope sequence (residues 37–51, GenBank accession number S50913.1) (Supporting Information Fig. S3). The TAp73 bands shown in Figures 2 and 3 represent those of p73α significantly knocked down by silencing with siRNA. Interference with PML RNA abolished the imatinib-induced phosphorylation of TAp73 and expression of TAp73 and bax (Fig. 3a, compared with those in Fig. 2a), but left unaffected the phosphorylation of p38 MAPK and chk2 in K562 cells (Fig. 3a). Interference with TAP73 RNA also abolished the expression of TAp73 and bax (Fig. 3b, compared with those in Fig. 2a), but left unaffected the phosphorylation of p38 MAPK and chk2 in K562 cells (Fig. 3b). Notably, treatment with p38 MAPK inhibitor (SB) led to abolishment of the imatinib-induced phosphorylation of p38 MAPK and TAp73 (Fig. 3c).
IFNα induces apoptosis of CML cells
IFNα also significantly induced apoptosis of K562 cells and significantly enhanced the imatinib-induced apoptosis of K562 cells (Fig. 1a). The apoptosis was abolished by the p38 MAPK inhibitor (SB) (Fig. 1c). The apoptosis-induction by IFNα was also concurrent with increased phosphorylation of p38 MAPK, TAp73 and chk2, and with increased expression of pml and bax (Fig. 2a).
Co-localization of TAp73 and pml at NBs in CML cells is induced by imatinib, enhanced by IFNα, unaffected by caspase inhibitor, and abolished by p38 MAPK inhibitor
Co-localization of TAp73 to PML-NBs in K562, Meg-01, HL-60, and primary CML bone marrow cells were examined at time points of 4, 12, 24, 36, and 48 hr after exposure to imatinib, and/or IFNα, BOC or SB. In K562, Meg-01, and the 3-patients' primary CML cells, imatinib induced a significant increase in TAp73/PML-NB co-localization dots, both in number and in size. The nuclear TAp73 dots, PML-NBs, and their co-localization increased as early as 4 hr after exposure to imatinib and/or IFNα, and achieved the maximum after a 12 hr exposure when cells still looked not apoptotic (Figs. 4 and 5). The co-localization sustained through 36 hr after exposure to imatinib and then decreased with cells looking increasingly apoptotic. The co-localization was regarded as significantly increased when the number of recognizable co-localization dots increased significantly after treatment and over 80% of the countable TAp73 dots co-localizing with PML-NBs. Significant increase of imatinib-induced TAp73/ PML-NB co-localization was seen in K562 cells (Fig. 4a), Meg-01 cells (Fig. 5b), and primary CML cells (Figs. 5c and 5d) from 3 out of 6 patients. Two (Patients1 and 2,) of the 3 patients were in accelerated phase and harbored p53mutant genes (Supporting Information Fig. S2); and the other was in chronic phase without p53 gene mutation. IFNα alone induced the TAp73/PML-NB co-localization in K562 cells and in combination with imatinib induced additively the co-localization (Fig. 4b). In BCR-ABL-/p53mutant HL60 cells, exposure to imatinib, however, did not increase the TAp73/PML-NB co-localization up to 36 hr, although a small quantity of background PML-NBs, nuclear TAp73 dots and their co-localization can be detected (Fig. 5a).
Imatinib induces co-immunoprecipitation of TAp73 and pml
Along with the co-localization, TAp73 was co-immunoprecipitable with anti-pml antibody from solubilized cellular extracts of imatinib-treated (0.5 μM, 12 hr) K562 cells when analyzed by Western blot following resolution in a 7% SDS-PAGE system (Fig. 4f). Also, pml was co-immunoprecipitated with anti-TAp73 antibody (Fig. 4g).
We have demonstrated a novel imatinib-induced p53-independent and caspase-dependent proapoptotic mechanism in CML cells using p38 MAPK/pml/TAp73 axis, in which chk2 and bax are very likely involved. The proapoptotic mechanism is p53 independent because it can be conducted in K562 CML cells lacking p53 and in Meg-01 CML cells harboring p53mutant. The monoallelic p53mutant in Meg-01 cells may co-exist with wild type p53, but due to either the dominant negative effect of the mutant p53 or failure of the wild type to accumulate on exposure to imatinib, p53 is much less likely responsible for the imatinib effect.
The imatinib-induced apoptosis of K562 cells increasing from 24 hr through 48 hr after exposure to imatinib was preceded by subcellular TAp73/PML-NB co-localization, which appeared at 4 hr and achieved the maximum at 12 hr after exposure. Furthermore, abolishment of TAp73/PML-NB co-localization by p38 MAPK inhibitor, or knockdown of pml or TAp73 led to the subsequent imatinib-induced apoptosis. TAp73/PML-NB co-localization thereby heralds the imatinib-induced apoptosis of CML cells. Noticeably, the imatinib-induced TAp73/PML-NB co-localization was accompanied by TAp73/pml co-immunoprecipitation. This implies that TAp73 and pml not only co-localize topographically but also interact physically. Functional interaction of TAp73 and pml, either direct or indirect, is further evident by that knockdown of pml led to decrease of imatinib-induced phosphorylation of TAp73 and the subsequent apoptosis. It seems that phosphorylation of TAp73 is essential for co-localizing of TAp73 to PML-NB for inhibition of p38 MAPK led to both dephosphorylation of TAp73 and abolishment of TAp73/PML-NB co-localization. Therefore, both p38MAPK and pml are required for phosphorylation of TAp73 and TAp73/PML-NB co-localization in the imatinib-treated CML cells. This observation agrees with that in other cellular context.17 It was observed in other cellular context that p38 MAPK-mediated phosphorylation of p73 was required for TAp73 to be recruited onto PML-NBs whereby prolyl isomerization and acetylation of p73 occurred and resulted in the escape of p73 from degradation.17 Pml herein probably executes the imatinib-induced proapoptotic machinery partly by stabilizing TAp73. Nonetheless, there are other possible mechanisms by which TAp73 may be up-regulated, such as those mediated by chk1 or chk2.24
IFNα exerted a proapoptotic effect on K562 cells and increased additively the effect with imatinib. The IFNα-induced apoptosis was also abolished by SB, a p38 MAPK inhibitor. The proapoptotic signals transduced by IFNα via p38 MAPK activation is hereby approved by our observation, agrees with those reported previously, and coincides partly with those transduced by imatinib.25
p38 MAPK responses to a variety of cellular stressors.26 Imatinib herein induced activating phosphorylation of p38 MAPK in CML cells. Inhibition of p38 MAPK blocked not only phosphorylation of TAp73 but also that of p38 MAPK itself. The activity of p38 MAPK is controlled by dual phosphorylation of the Thr180-Gly181-Tyr182 motif within its activation loop/lip.26 The dual activating phosphorylation of p38 MAPK can unquestionably be achieved by upstream MAPK kinases (MAPKKs) and is SB-insensitive, but also can be achieved by autophosphorylation as in many cases and is intrinsic p38 MAPK activity-dependent and SB-sensitive.27 The reason comes from that SB is an inhibitor of p38 MAPK but not that of upstream MAPKKs.27, 28 That the imatinib-induced phosphorylation of p38 MAPK is SB-sensitive suggests authphosphrylation is included. The details of how p38 MAPK is activated by imatinib are unknown. Although imatinib is an inhibitor of c-abl tyrosine kinase, c-abl may still activate p38 MAPK to induce apoptosis of CML cells independently of its tyrosine kinase activity as reported elsewhere.29
Pml plays an essential pleiotropic role in multiple proapoptotic pathways, both p53-dependent and independent.14, 18 PML-NB as a subcellular domain to recruit apoptosis or senescence-inducing proteins may or may not be required for pml to induce apoptosis or senescence.11–14 In the model of imatinib-induced apoptosis of K562 cells, the association of TAp73/pml at NB with apoptosis suggests that NB integrity is required for the mechanism.
An imatinib-induced increase of TAp73/PML-NB co-localization was seen in the p53mutant K562 and Meg-01 CML cells and also in the primary p53mutant CML cells, but not in BCR-ABL- p53mutant HL60 cells. Therefore, the TAp73/ PML-NB co-localization, which heralded apoptosis, was consistently found in the imatinib-treated p53mutant CML cells. However, as found previously, the p53 level in the imatinib-treated CML cells was reduced and constitutive phosphorylation of p53 at serine 20 was inhibited by imatinib.3 Moreover, we found imatinib-induced TAp73/PML-NB co-localization in primary CML cells from one of the 4 patients with wild type p53. It is still possible that the imatinib-induced pml/TAp73-mediated proapoptotic mechanism may also play in CML cells with wild type p53. However, with the small number of cases examined, this notion remains to be verified.
Pml was provoked by ras to transduce proapoptotic and prosenescence signals in primary fibroblasts.11–14 In CML cells, bcr-abl constitutively activates ras.16 However, augmentation of pml/PML-NB formation remains required for imatinib to proceed a proapoptotic machinery in non-lymphoblastic CML (K562 and Meg-01) cells. Paradoxically, imatinib inhibited bcr-abl and probably ras, but increased pml/PML-NBs. In a different situation observed by Puccetti et al, bcr-abl/ras also induced pml/PML-NB in Ph+ ALL cells and inhibition of bcr-abl by imatinib decreased pml/PML-NB.15 Therefore, imatinib may affect differentially the response of pml/PML-NB in different BCR-ABL+ cell types. Inadequate response of pml/PML-NB to counteract bcr-abl/ras may contribute prosurvival advantage to CML cells. In BCR-ABL+ ALL cells, although pml/PML-NB provoked by bcr-abl/ras may counteract their prosurvival, other signals not inhibited by imatinib, such as those from activated SRC-family kinases, may contribute prosurvival advantage.30
As reported recently, pml was phosphorylated by chk2 (hCds1) and led to irradiation-induced apoptosis in a p53-independent manner.31, 32 Chk2 was also reported to control the induction of p73 in response to DNA damage, and interference with or augmentation of chk2 signaling strongly impacted p73 accumulation.24 Our data demonstrated that inhibition of bcr-abl signaling without direct DNA damage in imatinib-treated K562 cells could still lead to an increase of phosphorylated chk2, which was not affected by interference with PML or TAP73 RNAs. Possibly, phosphorylation of chk2 occurs upstream and participates in the activating phosphorylation of pml and/or TAp73 to induce apoptosis in K562 cells.
TAp73 was found to directly transactivate the expression of BH3 only protein PUMA and bax. The BH3 only protein PUMA translocates bax from cytosol to mitochondria membrane and facilitates its conformational activation.5, 33 Although the BH3 only protein PUMA was not examined here, expression of bax was induced in the imatinib and/or IFNα-treated K562 or Meg-01 cells and was abolished by knockdownTAp73. Besides, the imatinib and/or IFNα -induced apoptosis is caspase-dependent. Therefore, the imatinib-induced bax may be presumed to be activated and lead to activation of the relevant caspases as under other bax-activating circumstances.34 The PUMA/bax/caspases machinery may thus be implicated in the imatinib-induced and TAp73-mediated apoptosis.
In summary, we have demonstrated a new p53-independent proapoptotic mechanism triggered by imatinib in some CML cells. The mechanism functions via p38 MAPK/pml/TAp73 axis with pml/TAp73 interaction at NB and probably with chk2 and bax being involved. The mechanism may disclose new therapeutic targets.
The authors thank Dr. De-Ming Yang from Department of Medical Research and Education, Taipei Veterans General Hospital, for the technical support in confocal laser scanning microscopy.