Growth inhibition by STI571 in combination with radiation in human chronic myelogenous leukemia K562 cells

c-Abl is a 145-kDa nonreceptor tyrosine kinase involved in a variety of cellular processes, including regulation of the cell cycle, actin organization, transduction of mitogenic signals, differentiation, and response to genotoxic stress (1, 2). Loss of c-Abl regulation may be oncogenic. Indeed, in pluripotent hematopoietic stem cells, the reciprocal t(9; 22)(q34; 11) chromosomal translocation generates the Philadelphia chromosome (3) and results in the fusion of the BCR and ABL genes. This translocation is the hallmark of chronic myelogenous leukemia (CML) and provided the first example of a specific genetic change associated with human cancer. BCR-ABL encodes a p210^Bcr-Abl chimeric protein in which the Abl tyrosine kinase activity is constitutively turned on by dimerization-induced, intermolecular autophosphorylation (2, 4).

Multiple signaling proteins have been shown to interact with Bcr-Abl through various functional domains and/or to become phosphorylated in Bcr-Abl-expressing cells (5). These include Ras, mitogen-activated protein kinase, phosphatidylinositol-3-kinase, Akt, Jnk, Src family kinases, and their respective downstream targets as well as transcription factors, such as the signal transducers and activators of transcription (STAT), nuclear factor-κB, and Myc (6–12). In addition, the activation of STAT5 and the phosphatidylinositol-3-kinase pathway by Bcr-Abl may mediate inhibition of apoptosis via up-regulation of Bcl-2 expression and phosphorylation of the proapoptotic protein Bad (13).

Whether Bcr-Abl affects on the DNA damage and repair capacity of cells is a matter of controversy. It has been proposed that Bcr-Abl induces resistance to DNA-nicking agents via protection from apoptosis, prolongation of the G₂ checkpoint, and stimulation of DNA repair (14, 15). Bcr-Abl has also been reported to promote the expression of the homologous recombination effector Rad51 (16). Other authors, however, did not reach the same conclusions (17). Alternative mechanisms, including the Bcr-Abl-dependent down-regulation of the expression of the nonhomologous end-joining kinase DNA-PKcs (18) and the homologous recombination effector BRCA1 (19), have been proposed. All these pathways are liable to be altered by STI571 (Imatinib, Glivec), a 2-phenylaminopyrimidine derivative that acts as a specific inhibitor of Bcr-Abl. STI571 actually competes with ATP for the binding site at the SH1 catalytic domain of Bcr-Abl (20, 21) and inhibits Bcr-Abl autophosphorylation, thus maintaining the kinase in an inactive conformation. STI571 has proven its efficiency in CML (22) and is nowadays the standard treatment in this disease.

Combination of STI571 and radiation has been attempted in non-Bcr-Abl cells. The results are controversial. Uemura et al. (23) found no evidence of a radiosensitizing effect in U937 leukemia cells. In contrast, Russell et al. (24) and Podtcheko et al. (25) reported that STI571 potentiates radiation-induced cell kill at high drug concentration in glioma and anaplastic thyroid cancer cells, respectively. The purpose of the present study was to assess the effect of STI571 on the radiation response in K562 CML cells expressing Bcr-Abl using growth inhibition, colony-forming ability, peroxidative metabolism, erythroid differentiation, apoptosis, senescence, and cell cycle progression as endpoints. The results show that in K562 cells STI571 cooperates with radiation to repress cell growth in a strictly additive way.

4-(4-Methylpiperazin-1-ylmethyl)-N-[4-methyl-3-[4-(3-pyridyl)pyrimidin-2-ylamino]phenyl] benzamide (STI571) was synthesized according to Zimmermann et al. (26, 27) with slight modifications. Briefly, commercial p-toluic acid methyl ester was brominated with N-bromosuccinimid. The bromomethylated compound was coupled with N-methylpiperazine and the resulting ester was directly condensed with 4-methyl-N-3(4-pyridin-3-yl-pyrimidin-2-yl)-benzene-1,3-diamine using trimethylaluminum in dichloromethane. Purification by flash chromatography yielded the pure compound in 85% yield. Purity of STI571 was confirmed by TLC and nuclear magnetic resonance analysis.

Aliquots of STI571 were dissolved in pure DMSO and stored as a 10 mmol/L stock solution at −20°C. Dilutions were made daily in growth medium. The final concentration of DMSO was ≤0.5% so as not to alter cell growth or radiation responses. All experiments were done in dim light to avoid the photodegradation of the drug.

The human CML, Bcr-Abl-expressing cell line K562 (ATCC CCL-243) was kindly provided by Dr. Philippe Rousselot (Service d'Immunologie Clinique, Hôpital Saint-Louis) and maintained in RPMI 1640 supplemented with Glutamax I, penicillin, streptomycin, and 10% FCS at 37°C with 5% CO₂. K562 cells express high levels of Bcr-Abl and are mutated for p53 (28).

Irradiation without or with concomitant exposure to STI571 was done at room temperature using a ¹³⁷Cs γ-ray, IBL-637 irradiator (CIS-Biointernational) at a dose rate of 1.15 Gy/min. Each measurement was done in duplicate or triplicate. The relative cell growth or colony count was fitted to the classic linear-quadratic equation,

where S is the growth or surviving fraction, D is the radiation dose, and α and β are adjustable variables characterizing radiosensitivity in the low (α) and high (β) dose ranges of radiation, respectively. Calculations were made through nonlinear least-squares regression, taking all data points into account, using Kaleidagraph software (Synergy Software).

Nuclear and cytoplasmic extracts were prepared each from 2 × 10⁷ cells with protease and phosphatase inhibitors as described (29). Total extracts were made using M-PER reagent (Pierce) with protease and phosphatase inhibitors (Sigma-Aldrich). Proteins were titrated by the Bradford method using the Bio-Rad protein assay. Cell extracts were boiled in Laemmli loading buffer and separated on 5% (DNA-PKcs), 7.5% (Bcr-Abl, STAT5, and Lyn), or 10% (Rad51 and CDK1) SDS-PAGE gel. Proteins were transferred to nitrocellulose membranes (Schleicher & Schuell) and blocked for 1 h in 5% bovine serum albumin in TBST at 37°C. Membranes were subsequently incubated with primary monoclonal antibodies overnight at 4°C in TBST buffer, washed for 1 h, and incubated with horseradish peroxidase–conjugated secondary antibodies (Jackson ImmunoResearch Laboratories) in TBST buffer and revealed with an enhanced chemiluminescence detection kit (GE Healthcare-Amersham Biosciences). Nitrocellulose membranes were rehybridized with anti-α-tubulin antibody (Sigma-Aldrich). For immunoprecipitates, calibration of protein loading was made by scanning SDS-PAGE gels after Coomassie blue staining. Densitometric analysis was done using QuantityOne software (Bio-Rad).

Because antibodies directed against CDK1 (Cdc2) cross-react with other CDKs in Western blots, the analysis of CDK1 expression and Tyr¹⁵ phosphorylation was carried out using CDK1 immunoprecipitates. Nuclear extracts were prepared in lysis buffer containing 50 mmol/L HEPES, 100 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L EGTA, and 0.5% NP40 (pH 7.0) plus protease and phosphatase inhibitors. Three hundred microliters of the extract (250 μg protein) were incubated with 3 μL monoclonal mouse anti-Cdc2 (POH1, Cell Signaling Technology) at 4°C overnight with gentle rocking. Protein G-Sepharose beads (20 μL of 50% bead slurry) were added to the lysate and incubated for 4 h at 4°C on a rotator. The immunoprecipitates were washed four times with lysis buffer. The pellets were resuspended in 20 μL of 2× SDS loading buffer. Immunoprecipitates were separated by 12% SDS-PAGE and detected by chemiluminescence using monoclonal or polyclonal antibodies to Cdc2 and p-Tyr¹⁵-Cdc2 (1:100 dilution; Cell Signaling Technology). GST-Cdc2 fusion protein and SK-N-MC total cell extracts treated with hydroxyurea (3 mmol/L, 24 h) were used as negative and positive controls, respectively.

Determination of apoptosis and necrosis was done using an Annexin V-FITC kit (Calbiochem). Briefly, 10⁶ cells were suspended in 1 mL ice-cold binding buffer [10 mmol/L HEPES, 150 mmol/L NaCl, 2.5 mmol/L CaCl₂, 1 mmol/L Mg Cl₂, 4% BSA (pH 7.4)]. Ten microliters of media binding buffer and 1.25 μL Annexin V-FITC antibody (Calbiochem) were added to 500 μL of the cell suspension for 15 min at room temperature in the dark. Cells were harvested by centrifugation and resuspended in 0.5 mL ice-cold binding buffer. Annexin V-FITC–labeled cells were analyzed immediately using a FACStar PLUS cytofluorometer (Becton-Dickinson Biosciences). Propidium iodide (10 μL; 30 μg/mL in PBS) was subsequently added to allow discrimination between apoptotic (propidium iodide–negative) and necrotic (propidium iodide–positive) cells among the Annexin V–positive cells, and both propidium iodide and FITC fluorescence was determined by fluorescence-activated cell sorting. Signal analysis was carried out using CellQuest Pro software (Becton-Dickinson Biosciences).

The hypodiploid (sub-G₁) fraction was measured using fluorescence-activated cell sorting analysis of propidium iodide–stained cells after overnight fixation with cold 70% ethanol. The sub-G₁ region was determined by a gate on the DNA content histogram excluding cell debris.

The expression of glycophorin A, a sialoglycoprotein present at the surface of human RBCs and erythroid precursors, was detected by direct immunofluorescence according to Kawano et al. (30). Briefly, 10⁶ K562 cells that had been exposed or not to STI571, radiation, or a combination of both were harvested, washed once with PBS supplemented with 2% FCS, and incubated in the dark (45 min, 4°C) with a monoclonal antibody directed against human glycophorin A following the manufacturer's instructions (PharMingen). After three washes with PBS containing 2% FCS, the cells were analyzed by flow cytometry. Mouse IgG antibody was used as isotype negative control.

See Supplementary Material.⁶

The effect of STI571 and/or radiation on the cell cycle distribution and progression of cells through S phase was monitored by dual-variable flow cytometry using a FACStar PLUS cytofluorometer as above. Pulse-chase labeling of K562 cells by bromodeoxyuridine (10-30 μmol/L, 10-20 min, 37°C) was done ahead of or following exposure to radiation and/or STI571 (see text and figure legends for details). Cells were finally harvested by centrifugation, washed once with ice-cold PBS, and fixed in 70% ice-cold ethanol. Treatment of fixed cells and data acquisition were carried out as described previously (31). Data analysis was done with CellQuest Pro software (Becton-Dickinson Biosciences).

STI571 induced a time- and concentration-dependent inhibition of cell proliferation (Fig. 1). After 72-h contact, the amount of drug that reduced cell growth to 50% of that in controls (IC₅₀) was 0.14 ± 0.01 μmol/L. This value is in close agreement with those determined by other authors using a similar assay (32, 33). Past 48-h incubation, a drop in the cell count was observed at the highest drug concentrations used, suggesting a cytotoxic effect. This effect correlated with erythroid differentiation, a phenomenon described previously by Jacquel et al. (34). Indeed, STI571 generated tiny, hemoglobin-positive cells that expressed the same level of glycophorin A as undifferentiated cells but incorporated 10-fold as much propidium iodide (Fig. 2) and have proven unable to resume growth after drug removal.

For determination of the effect of combined treatment on cell growth, K562 cells were first incubated with STI571 for 24 h before irradiation. The results are shown in Fig. 3A. The growth curves obtained in the absence of STI571 fitted the linear-quadratic model [Eq. (1); see Materials and Methods]. In combination with radiation, STI571 brought about suppression of the β variable together with a 2.1-fold increase of the α variable, suggesting radiosensitization in the low-dose range of radiation. Interestingly, radiation antagonized STI571-induced erythroid differentiation under similar conditions (Fig. 2).

For clonogenic assays, K562 cells were harvested at the end of treatment, freed from drug, and grown in semisolid medium. Reduction of the radiation susceptibility by a factor of 1.6 among drug survivors, specifically in the low-dose range of radiation was experienced in this assay (Fig. 3B).

In a third experiment, STI571 was introduced shortly after irradiation and was left in the medium for a maximum of 48 h. Under these conditions, we observed the same effect as in the first growth assay, that is, suppression of the quadratic component characteristic of radiation response together with a substantial increase of the α variable that was already apparent after 24 h contact with drug (Fig. 3C).

In summary, the comparison of growth and clonogenic experiments shows that STI571 is able to induce a cytotoxic effect likely to parallel erythroid differentiation but does not interfere substantially with the lethal effect of radiation, although a moderate radioprotecting effect was observed at low radiation dose in the clonogenic assay. At particular times past radiation exposure, however, the growth inhibitory effect of the drug resulted in suppression of the quadratic component (β) together with a substantial increase of the α variable characterizing response in the low-dose range of radiation.

Apoptosis and necrosis of K562 cells were assessed by flow cytometric analysis of the sub-G₁ DNA fragments and Annexin V binding in cells exposed to 4 Gy alone, 0.3 μmol/L STI571 alone, and a combination of both. The susceptibility to STI571-induced apoptosis was in good agreement with Jacquel et al. (34). From both methods, the interaction between radiation and drug appeared to be purely additive with respect to the level of apoptosis (Fig. 4). Based on the DNA content, the necrosis index was below 1% under these conditions. However, necrosis was more pronounced than apoptosis at high concentrations of STI (14% for 1 μmol/L STI571, 20% for 10 μmol/L STI571, 72-h contact) in agreement with Okada et al. (35).

To extend these observations, Bcr-Abl and STAT5 expression and phosphorylation were analyzed by Western blot. Indeed, the phosphorylated form of STAT5 reportedly imparts sustained resistance to DNA damage-induced apoptosis (7) via up-regulation of the transcription of the antiapoptotic Bcl-X_L protein (36). No significant change of Bcr-Abl and STAT5 expression was observed following radiation, STI571, or a combination of both. In contrast, STI571 abolished Bcr-Abl and STAT5 phosphorylation independently of whether cells were irradiated or not (Fig. 5).

Ionizing radiation induced senescence-like changes in K562 cells exposed to radiation. Indeed, after 5-day incubation, a substantial part of irradiated cells underwent phenotypic changes characteristic of senescence, associated with β-galactosidase staining. At the concentration used, STI571 alone did not induce such changes. In combination with radiation, STI571 overcame the effect of radiation and abolished radiation-induced formation of β-galactosidase-positive cells to near completion (see Supplementary Material).

The time dependence of cell cycle disruption by radiation (6 Gy), STI571 (0.3 or 1 μmol/L), and a combination of both was investigated by flow cytometry for up to 72 h following initial treatment. The results are shown in Fig. 6.

Radiation alone induced a depletion of the G₁- and S-phase compartments together with prolonged accumulation in the G₂ phase. The effect reached a maximum at ∼12 h (G₁, G₂) to 24 h (S) after irradiation. STI571 (0.3 μmol/L) alone brought about a transient G₁-phase accumulation and S-phase depletion peaking at 12 h of contact with drug followed by a progressive decrease of the G₁- and G₂-phase content and accumulation in S phase. These effects were more pronounced in the presence of 1 μmol/L drug. In particular, 1 μmol/L drug abrogated the radio-induced G₂ block and elicited prolonged accumulation in S phase paralleling G₁-phase depletion.

Abolition of the G₂-phase arrest by caffeine has long been known to slow the progress of DNA repair and result in a radiosensitizing effect. We used two methods to determine whether a similar event could be induced by STI571 in K562 cells. For this purpose, we analyzed (a) the pathway involved in the control of the G₂ transition through Tyr¹⁵ phosphorylation of CDK1 (Cdc2) and (b) the cell cycle progress in cells that had been labeled with bromodeoxyuridine in S phase before exposure to drug or radiation.

Control of the G₂-Phase Progression. Loss of Bcr-Abl phosphorylation is expected to reduce the activity of the tyrosine kinase Lyn through Tyr⁵⁰⁷ dephosphorylation (37). The phosphorylated form of Lyn has been involved in the control of Tyr¹⁵ phosphorylation of CDK1, thus controlling the transit through late G₂ and mitosis (38). Exposure to STI571 was found to induce a loss of Lyn phosphorylation (Fig. 5), consistent with the lack of a radio-induced G₂ block in the presence of STI571 (Fig. 6). This study was completed by analysis of CDK1 expression and phosphorylation using immunoprecipitates. As expected, radiation alone elicited a transient increase of the nuclear content and phosphorylation of CDK1, peaking at ∼16 h following radiation (Fig. 5). Preincubation (24 h) with 1 μmol/L STI571 prevented CDK1 phosphorylation, consistent with the flow cytometric analysis that showed a lack of accumulation of cells in G₂ under these conditions.

Control of the S-Phase Progression. From the data presented above, it may be inferred either that STI571 abrogates the radio-induced G₂ block through down-regulation of Lyn phosphorylation or that cells are blocked at an earlier stage.

To resolve this issue, the effect of STI571 and/or radiation on the relative movement of cells through S phase was analyzed in asynchronous K562 cells that had been pulse labeled with bromodeoxyuridine in S phase. Both STI571 and radiation slowed down the transit of cells through the S phase of the cell cycle. The combination of radiation and STI571 produced an additive effect (see Supplementary Material).

STI571 at high concentration has been reported to increase radiation susceptibility in glioblastoma cells both in vitro (24) and in mouse models (39, 40). This effect was assigned to disruption of an autocrine loop involved in phosphorylation (activation) of the platelet-derived growth factor receptor (39, 40) and was not observed in breast and colon cancer cells (40). At high enough concentrations to inhibit the activity of c-Abl, STI571 was able to potentiate radiation response in cells from anaplastic thyroid cancer (25). However, at concentrations achieving complete inhibition of Bcr-Abl but insufficient to depress c-Abl activity, STI571 did not alter the radiation response in normal bone marrow cells from BALB/c mice and in U937 human leukemia cells in a clonogenic assay in vitro (23). Here, we show that STI571 at cytostatic concentrations represses cell growth cooperatively with radiation in K562 CML cells but does not induce any significant change in radiation susceptibility, consistent with the fact that the expression of DNA-PKcs was not altered (see Supplementary Material).

STI571 at submicromolar concentration did not alter the expression of Bcr-Abl and STAT5 expression in K562 cells but abrogated the phosphorylation of Bcr-Abl-Tyr²⁴⁵ and STAT5-Tyr⁶⁹⁴, as also observed by others (34, 41) and independently of whether cells were irradiated or not. As the phosphorylated form of STAT5 imparts sustained resistance to DNA damage-induced apoptosis (7, 36), down-regulation of STAT5 phosphorylation would be expected to result in an enhanced susceptibility of K562 cells to apoptosis. However, although K562 cells showed time- and dose-dependent apoptosis in response to STI571 and radiation, as shown in Fig. 4, the drug did not potentiate radiation-induced apoptosis and the interaction between both modalities was strictly additive. The same result was obtained by Podtcheko et al. (25) using non-Bcr-Abl anaplastic thyroid cancer cells.

A substantial part of K562 cells exposed to radiation alone displayed phenotypic changes associated with the formation of β-galactosidase-positive cells (see Supplementary Material) typical of senescence-like, terminal growth arrest in nearly the same way as described for hydroxyurea treatment also in K562 cells (42). Surprisingly enough, STI571 acted antagonistically with radiation in this pathway. Symmetrically, in our hands, radiation antagonized drug-induced erythroid differentiation of K562 cells (Fig. 2). These results are at variance with those obtained at high drug concentration in the non-Bcr-Abl anaplastic thyroid cancer cells (25). One would tentatively propose to explain these differences that pathways of DNA damage response under control of the c-Abl tyrosine kinase are not operating in CML cells. Alternatively, inhibition of the catalytic activity of other tyrosine kinases, such as c-Kit or platelet-derived growth factor receptor, at concentrations 1 order of magnitude above those used in this study should also be taken into consideration.

Riordan et al. (43) and Jeong et al. (44) earlier reported that herbimycin A, an inhibitor of c-Abl and Bcr-Abl, increased the level of radiation-induced apoptosis in K562 cells. This effect was associated with the abrogation of the G₂ checkpoint (44). Actually, bypassing the G₂ checkpoint reduces the amount of DNA repair that can take place and leads to an increase of radiation susceptibility. To determine whether this effect occurred in K562 cells, we analyzed the effect of STI571 on the phosphorylation of Lyn, a member of the Src tyrosine kinase family. As a matter of fact, constitutive activation of Lyn through Tyr⁵⁰⁷ phosphorylation is driven by Bcr-Abl (45). Lyn can also be activated by radiation and other DNA-damaging agents. p-Tyr⁵⁰⁷-Lyn binds to and phosphorylates CDK1 on the Tyr¹⁵ residue, and p-Tyr¹⁵-CDK1 is necessary to block G₂-M progression (38). Although Lyn expression did not vary after radiation or prolonged exposure to STI571, Lyn phosphorylation slightly increased after radiation exposure and was abolished by STI571. Such inhibition of Lyn phosphorylation is sufficient to induce suppression of the radio-induced G₂ block. However, this explanation does not hold in the present case. Indeed, studies with pulse-chased bromodeoxyuridine-labeled cells showed that a short contact with STI571 lengthened S-phase progression (see Supplementary Material) but did not affect the duration of the G₂ block induced by radiation. On the other hand, prolonged exposure to 1 μmol/L STI571 resulted in the accumulation of cells in G₁ and in early S phase. This block was efficient enough to starve the G₂ compartment.

The ability of STI571 to induce transient accumulation in early S phase suggests that STI571 might be able to potentiate the cytotoxic effect of topoisomerase I and II poisons in c-Kit and platelet-derived growth factor receptor–expressing solid tumors, which are both targets for STI571. Indeed, STI571 has been reported to act synergistically with etoposide and topotecan in small cell lung cancer xenografts (46). Moreover, in c-Kit-positive small cell lung cancer cells, STI571 was found to up-regulate topoisomerase I activity and potentiate induced cell kill by camptothecin derivatives (47). Clearly, the possibility of selective sensitization to S-phase targeting drugs by STI571 in Bcr-Abl, platelet-derived growth factor receptor, or c-Kit-expressing tumors warrants further investigation.

We thank Dr. Philippe Rousselot for the generous gift of K562 cells, Drs. Eric Deutsch (Institut Gustave-Roussy), Janet Hall and Frédérique Mégnin-Chanet (Institut Curie) for helpful discussion, and Danièle Rouillard for the flow cytometric analysis of the cell cycle.