Luminespib

Combined effects of novel heat shock protein 90 inhibitor NVP-AUY922 and nilotinib in a random mutagenesis screen

To overcome imatinib resistance, more potent ABL tyrosine kinase inhibitors (TKIs), such as nilotinib and dasatinib have been developed, with demonstrable pre- clinical activity against most imatinib-resistant BCR– ABL kinase domain mutations, with the exception of T315I. However, imatinib-resistant patients already harboring mutations have a higher likelihood of develop- ing further mutations under the selective pressure of potent ABL TKIs. NVP-AUY922 (Novartis) is a novel 4,5-diaryloxazole adenosine triphosphate-binding site heat shock protein 90 (HSP90) inhibitor, which has been shown to inhibit the chaperone function of HSP90 and deplete the levels of HSP90 client protein including BCR–ABL. In this study, we investigated the combined effects of AUY922 and nilotinib on random mutagenesis for BCR– ABL mutation (Blood, 109; 5011, 2007). Compared with single agents, combination with AUY922 and nilotinib was more effective at reducing the outgrowth of resistant cell clones. No outgrowth was observed in the presence of 2 lM of nilotinib and 20 nM of AUY922. The observed data from the isobologram indicated the synergistic effect of simultaneous exposure to AUY922 and nilotinib even in BaF3 cells expressing BCR–ABL mutants including T315I. In vivo studies also demonstrated that the combination of AUY922 and nilotinib prolonged the survival of mice transplanted with mixture of BaF3 cells expressing wild-type BCR–ABL and mutant forms. Taken together, this study shows that the combination of AUY922 and nilotinib exhibits a desirable therapeutic index that can reduce the in vivo growth of mutant forms of BCR–ABL-expressing cells.

Keywords: BCR–ABL; tyrosine kinase inhibitor; nilotinib; HSP90; T315I

Introduction

Resistance to the ABL tyrosine kinase inhibitor (TKI), imatinib, in Ph-positive leukemia is often caused by selection of mutations in BCR–ABL kinase domain altering residues that are directly or indirectly critical for imatinib binding (O’Hare et al., 2007). To overcome imatinib resistance, more potent ABL TKIs, such as nilotinib and dasatinib have been developed, with demonstrable preclinical activity against most imati- nib-resistant BCR–ABL kinase domain mutations, with the exception of T315I (Shah et al., 2004; Weisberg et al., 2005). The T315I is the single most frequent mutation that outgrows and leads to relapse during nilotinib and imatinib-treatment (Jabbour et al., 2008). Other mutations, however, are found to have emerged at the time of relapse in ABL TKI-resistant patients (Branford et al., 2009). F359V and the P-loop mutants Y253H and E255K/V are associated with relapse to nilotinib, and F317A/L and V299L are found in dasatinib-resistant patients (Branford et al., 2009). Imatinib-resistant patients already harboring mutations have a higher likelihood of developing further mutations under the selective pressure of ABL TKIs (Garg et al., 2009). The challenge for development of an effective Ph- positive leukemia therapy is therefore to develop an alternative treatment strategy that does not rely solely on kinase domain inhibition but rather results in degradation of the offending BCR–ABL protein regard- less of its mutation status.

NVP-AUY922 is a novel 4,5-diaryloxazole adenosine triphosphate-binding site heat shock protein 90 (HSP90) inhibitor, which has been shown to inhibit the chaper- one function of HSP90 and deplete the levels of HSP90 client protein (for example, ErbB2, Akt, Raf and Bcr– Abl) (Brough et al., 2008; Eccles et al., 2008; Stuhmer et al., 2008). Combining AUY922 with ABL kinase inhibitors may provide several advantages, such as enhanced efficacy and reducing the potential emergence of new resistant mutations. In this study, we performed a comprehensive drug combination experiment using a broader range of concentrations for AUY922 and nilotinib or imatinib. Compared with single agents, combination with AUY922 and nilotinib was more effective at reducing the outgrowth of resistant cell clones. At the highest concentration of nilotinib, the mutation spectrum narrowed to T315I and E344V by direct sequencing, whereas, at intermediate concentra- tion of AUY922, the resistant clone was recovered by wild-type (WT) BCR–ABL only. The observed data from the isobologram indicated the synergistic effect of simultaneous exposure to AUY922 and nilotinib even in BaF3 cells expressing BCR–ABL mutants including T315I. In vivo studies also demonstrated that the com- bination of AUY922 and nilotinib prolonged the survival of mice transplanted with mixture of BaF3 cells expressing WT BCR–ABL and mutant forms. Taken together, this study shows that the combination of AUY922 and nilotinib exhibits a desirable therapeutic index that can reduce the in vivo growth of mutant forms of BCR–ABL-expressing cells.

Results

AUY922 in combination with nilotinib or imatinib completely suppresses outgrowth of resistant clones by random mutagenesis screen

To assess whether combinations of ABL kinase inhibi- tors offer an advantage over AUY922 alone, we carried out a random mutagenesis screens with combinations of AUY922 and imatinib or nilotinib (Ray et al., 2007). This approach uses a DNA repair-deficient Escherichia coli strain to produce random mutagenesis of a BCR– ABL retroviral plasmid, infection of BaF3 cells and selection for BaF3 clones conferring varying degree of drug resistance (Ray et al., 2007). Profiling of AUY922 as a single agent revealed a concentration-dependent reduction of colonies (Figure 1). Sequencing revealed only WT BCR–ABL in AUY922-resistant clones, with the lowest concentration of AUY922 and 500 nM of imatinib (Figure 1). Sequencing of nilotinib-resistant clones with the highest concentration (2 mM) revealed T315I and E344V (Figure 1). We then performed a comprehensive drug combination using a broader range of concentrations of AUY922 and nilotinib or imatinib. Compared with single agent, combinations with AUY922 and nilotinib or AUY922 and imatinib were more effective at reducing the outgrowth of resistant clones. No outgrowth was observed in the presence of 2 mM of nilotinib and 20 nM of AUY922 (Figure 1).

Combined effects of AUY922 and nilotinib in mutant forms of BCR–ABL-expressing cells

We used the isobologram method to determine whether the combined effects of AUY922 and nilotinib are additive or synergistic in mutant forms of BCR–ABL- expressing BaF3 cells. Figure 2a showed the dose- response curve for AUY922 and nilotinib in WT-p210 BCR–ABL-expressing BaF3 cells. The isobologram was generated on the dose-response curve. The observed data from the isobologram indicated the synergistic effect on simultaneous exposure to AUY922 and nilotinib in WT BCR–ABL-expressing BaF3 cells (Figure 2a). M351T is a sensitive mutation to nilotinib, on the other hand, E255K is an insensitive mutation to nilotinib. The isobolograms show the combination with AUY922 and nilotinib has the synergistic effect on both mutation (Figures 2b and c). In constant, T315I BaF3 cells were resistant to nilotinib up to levels as high as 2 mM, however, the treatment with AUY922 and nilotinib showed the synergistic effect in T315I BCR– ABL BaF3 cells (Figure 2d). Next, we determined the colony growth of WT BCR–ABL BaF3 cells and WT BCR–ABL-expressing primary chronic myeloid leuke- mia (CML) mononuclear cells (Figure 3a). Co-treatment with AUY922 and nilotinib caused significantly more inhibition of colony growth than treatment of either agent alone in WT BCR–ABL BaF3 cells and primary CML cells (Figure 3a). Further, we examined the colony growth of T315I BaF3 cells and T315I-expressing primary cells (Figure 3b). Co-treatment with AUY922 and nilotinib caused significantly more inhibition of colony growth than treatment of either agent alone in T315I BaF3 cells and T315I-expressing primary leuke- mia cells (Figure 3b). WT BCR–ABL BaF3 cells and T315I BaF3 cells were cultured with the indicated concentrations of AUY922 and nilotinib for 72 h, after which the percentage of apoptotic cells was determined by annexin-V (Figures 3c and d). When 10 nM of AUY922 was combined with nilotinib in WT BCR– ABL BaF3 cells, the increase in apoptotic cells was virtually complete for nilotinib concentrations higher than 1 mM (Figure 3c). Treatment with 1 mM of nilotinib had no effect on T315I BaF3 cells, however, co- treatment of AUY922 and nilotinib also enhanced the induction of apoptosis in T315I BaF3 cells (Figure 3d). Together, these findings indicate that combination of minimally toxic concentrations of AUY922 and niloti- nib is effective in inducing apoptosis in both WT BCR– ABL-expressing cells and T315I BCR–ABL-expressing cells.

Figure 1 AUY922 in combination with nilotinib or imatinib completely suppresses outgrowth of resistant clones. The approach uses a DNA repair-deficient Escherichia coli strain to produce random mutagenesis of a BCR–ABL retroviral plasmid, infection of BaF3 cells, and selection for BaF3 clones conferring varying degree of drug resistance using methods previously described (Azam et al., 2003). BaF3 cells expressing the random mutagenesis of BCR–ABL were kindly provided by Dr James D Griffin (Dana-Farber Cancer Institute) (Ray et al., 2007). BaF3 cells expressing the random mutagenesis of a BCR–ABL were cultured with graded concentrations of AUY922 alone and in combination with imatinib or nilotinib. Bars represent the percentage of wells from which drug-resistant clones were recovered. Similar results were obtained in three independent experiments.

Figure 2 Combined effects of AUY922 and nilotinib in mutant forms of BCR–ABL-expressing cells. The theoretical basis of isobologram method and the procedure for making isobolograms have previously been described in detail (Kano et al., 2001). The dose-response curves and generated isobologram for AUY922 and nilotinib were shown in Figure a (WT p210 BCR–ABL), b (M351T BCR–ABL), c (E255K BCR–ABL) and d (T315I BCR–ABL). The observed data from the isobologram indicated the synergistic effect of simultaneous exposure to AUY922 and nilotinib even in BaF3 cells expressing BCR–ABL mutants including T315I. Similar results were obtained in three independent experiments.

Figure 3 Co-treatment of AUY922 and nilotinib suppresses colony formation of in WT BCR–ABL and T315I BCR–ABL-expressing cells, and enhances the induction of apoptosis. (a) WT BCR–ABL BaF3 cells and WT BCR–ABL-expressing primary CML mononuclear cells were grown in methylcellulose containing the indicated concentrations of AUY922 and nilotinib. Colony counts were assessed on each individual sample at least twice, and results are presented as average±s.d. for colonies counted from triplicate plates under each condition. WT BCR–ABL BaF3 cells; nilotinib 50 nM: 49.1±3.5%, AUY922 15 nM: 73.5±5.5%, AUY922 30 nM: 48.4±1.4%, AUY922 15 nM + nilotinib 50 nM: 9.4±0.3%, AUY922 30 nM + nilotinib 50 nM: 3.5±0.1%, respectively. *Po0.01 compared with nilotinib 50 nM-treatment. WT BCR–ABL-expressing primary CML mononuclear cells; nilotinib 50 nM: 64.3±1.9%, AUY922 15 nM: 73.3±1.3%, AUY922 30 nM: 53.2±3.1%, AUY922 15 nM + nilotinib 50 nM: 16.1±1.8%, AUY922 30 nM + nilotinib 50 nM: 1.8±0.1%, respectively. **Po0.01 compared with nilotinib 50 nM-treatment. (b) T315I BaF3 cells and T315I-expressing primary cells were grown in methylcellulose containing the indicated concentrations of AUY922 and nilotinib. Colony counts were assessed on each individual sample at least twice, and results are presented as average±s.d. for colonies counted from triplicate plates under each condition. T3151 BaF3 cells; nilotinib 1000 nM: 99.1±0.8%, AUY922 10 nM: 81.5±1.9%, AUY922 20 nM: 37.2±0.8%, AUY922 10 nM + nilotinib 1000 nM: 40.2±0.9%, AUY922 20 nM + nilotinib 1000 nM: 3.8±0.2%, respectively. *Po0.01 compared with nilotinib 1000 nM-treatment. T315I-expressing primary cells; nilotinib 1000 nM: 99.2±0.1%, AUY922 10 nM: 73.5±1.3%, AUY922 20 nM: 20.1±0.9%, AUY922 10 nM + nilotinib 1000 nM: 46.9±1.4%, AUY922 20 nM + nilotinib 1000 nM: 2.1±0.4%, respectively.

AUY922 induces degradation of WT and mutant forms of BCR–ABL proteins

Previous studies have shown that the HSP90 inhibitors geldanamycine and 17-AAG disrupt HSP90 function and induce BCR–ABL degradation (Gorre et al., 2002). To determine whether AUY922 can similarly cause the degradation of BCR–ABL proteins, WT, T315I, E255K or M351T BCR–ABL-expressing BaF3 cells were exposed to varying concentrations of AUY922 for 24 h (Figures 4a and b). Immunoblot analysis revealed that AUY922 caused BCR–ABL protein levels to decrease significantly in WT BCR–ABL-expressing BaF3 cells at a dose of 50 nM, as expected (Figure 4a). Mutant forms of BCR–ABL proteins were also degraded at a lower concentration of AUY922 (Figures 4a and b). These results suggest that AUY922 may have greater potency against mutant forms of BCR–ABL proteins compared with WT.

The mechanism of the synergism between AUY922 and nilotinib in T315I BCR–ABL BaF3 cells

We next conducted the experiments to further evaluate the mechanism of the synergism between AUY922 and nilotinib in T315I BCR–ABL BaF3 cells. Besides the ABL kinases, the receptor tyrosine kinase DDR1 and the oxidoreductase NQO2 are target molecules for nilotinib (Bantscheff et al., 2007; Rix et al., 2007). T315I BaF3 cells were cultured with the indicated concentrations of nilotinib for 24 h, the cell lysates were immunoprecipitated with anti-DDR1 antibody (Ab) and then immunoblotted with anti-phosphotyrosine mAb (PY20) or anti-DDR1 Ab (Figure 5a). Higher concentrations of nilotinib abolished DDR1 autopho- sphorylation (Figure 5a). To assess the functional importance of DDR1 and NQO2, we used RNA interference to determine whether reduction in DDR1 and NQO2 affect the proliferation of T315I BaF3 cells after the treatment of AUY922. T315I BaF3 cells were transfected with control small interfering RNA (siRNA) or DDR1 siRNA or NQO2 siRNA; then the DDR1 and NQO2 expression was analyzed by immunoblotting after 48 h (Figure 5b). At 48 h after transfection, T315I BCR–ABL BaF3 cells were treated with indicated concentration of AUY922 for 48 h, and viable cells were counted (Figure 5c). In the presence of DDR1 siRNA, T315I BCR–ABL BaF3 cells increased anti- proliferative activity with AUY922 (at 5 or 10 nM) (Figure 5c). When AUY922 was treated in the presence of NQO2 siRNA, antiproliferative activity of T315I BCR–ABL was not observed (Figure 5c). These results showed that inhibition of DDR1 can have an important role in the synergism between AUY922 and nilotinib.

Figure 4 AUY922 induces degradation of WT and mutant forms of BCR–ABL proteins. (a) WT or T315I BCR–ABL-expressing BaF3 cells were exposed to varying concentrations of AUY922 for 24 h. The cell lysates were immunobloted with anti-ABL Ab or anti- HSP70 Ab. (b) E255K, or M351T BCR–ABL-expressing BaF3 cells were exposed to varying concentrations of AUY922 for 24 h. The cell lysates were immunobloted with anti-ABL Ab or anti-HSP70 Ab.

Further, we examined the phosphorylation of T315I BCR–ABL after treatment of AUY922 and nilotinib. T315I BCR–ABL BaF3 cells were cultured with indicated concentrations of AUY922 and nilotinib for 24 h. The cell lysates were immunoblotted with anti-phosho-ABL Ab or anti-ABL Ab (Figure 5d). Co-treatment with AUY922 and nilotinib partially decreased auto-phosphorylation of T315I BCR–ABL (Figure 5d).

Co-treatment of AUY922 and nilotinib prolong the survival in mice model of BCR–ABL mutant-induced leukemia

We investigate the in vivo efficacy of AUY922 and nilotinib (Figure 6). Twelve-week-old nude mice were injected with 5 × 105 cells of mixture of BaF3 cells expressing WT BCR–ABL and mutant forms of BCR– ABL (M244V, G250E, Q252H, Y253F, T315A, T315I, F317L, F317V, M351T and H396P). At 24-h injection of the leukemia cells, these mice were treated with either vehicle or AUY922 (50 mg/kg intraperitoneal (i.p.); two times per week) or nilotinib (30 mg/kg; q.d.) or AUY922 et al., 2008; Stuhmer et al., 2008). In addition, optimization of pharmacokinetic properties led to robust therapeutic responses in a wide variety of human tumor xenografts tightly linked to high intratumor concentrations of compound and phamacodynamic response (Eccles et al., 2008). The promising preclinical data obtained with AUY922 supported the initiation of clinical phase I trials in patients with solid tumors.

Figure 5 The mechanism of the synergism between AUY922 and nilotinib in T315I BCR–ABL BaF3 cells. (a) T315I BaF3 cells were cultured with the indicated concentrations of nilotinib for 24 h, the cell lysates were immunoprecipitated with anti-DDR1 Ab and then immunoblotted with anti-phosphotyrosine mAb or anti-DDR1 Ab. (b) T315I BaF3 cells were transfected with control siRNA or DDR1 siRNA or NQO2 siRNA; then the DDR1 and NQO2 expression was analyzed by immunoblotting after 48 h. (c) At 48 h after transfection, T315I BCR–ABL BaF3 cells were treated with indicated concentration of AUY922 for 48 h, and viable cells were counted by using Vi-cell XR automated cell viability analyzer (Beckman Coulter). The mean number of viable cells at different concentration of drug was normalized to the mean number of viable cells in the no-drug samples. *Po0.01 compared with contro siRNA-treatment. Similar results were obtained in each of three independent experiments. (d) T315I BCR–ABL BaF3 cells were cultured with indicated concentrations of AUY922 and nilotinib for 24 h. The cell lysates were immunoblotted with anti-phosho-ABL Ab or anti-ABL Ab.

In this study, we investigated a comprehensive drug combination experiment using a broader range of concentrations for AUY922 and nilotinib or imatinib by saturation mutagenesis screen (Figure 1). AUY922 controlled the outgrowth of WT BCR–ABL, and mutated forms of BCR–ABL associated with imatinib or nilotinib resistance (Figure 1). When AUY922 was (50 mg/kg i.p.; two times per week) + nilotinib (30 mg/ kg; q.d.). The vehicle or nilotinib-treated mice died of a condition resembling acute leukemia by 28 days; the combination of AUY922 + nilotinib-treated mice sur- vived more than 60 days, significantly improved the survival (P = 0.009) compared with nilotinib-treated mice (Figure 6).

Figure 6 Co-treatment of AUY922 and nilotinib prolong the survival in mice model of BCR–ABL mutant-induced leukemia. Nude mice were injected with 5 × 105 cells of mixture of BaF3 cells expressing WT BCR–ABL and mutant forms of BCR–ABL (M244V, G250E, Q252H, Y253F, T315A, T315I, F317L, F317V,
M351T and H396P). At 24-h injection of the leukemia cells, these mice were treated with either vehicle or AUY922 (50 mg/kg i.p.; two times per week) or nilotinib (30 mg/kg; q.d.) or AUY922 (50 mg/kg i.p.; two times per week) + nilotinib (30 mg/kg; q.d.).

Discussion

Second-generation TKIs have demonstrated increased inhibitory potency against BCR–ABL tyrosine kinase and have shown efficacy in treating patients with number of the BCR–ABL kinase domain mutations that develop on imatinib (Kantarjian et al., 2006, 2007, 2009; Ottmann et al., 2007). Despite the significant clinical activity demonstrated in clinical trials, a number of patients do not show durable response (Garg et al., 2009). One reason for the lack of durable response could be explained by the emergence of new kinase domain mutations as patients are exposed to sequential TKIs (Branford et al., 2009; Garg et al., 2009). The challenge for development of an effective Ph-positive leukemia therapy is therefore to develop an alternative treatment strategy that does not rely solely on kinase domain inhibition but rather results in degradation of the offending BCR–ABL protein regardless of its mutation status.

AUY922 is a most potent resorcinylic isoxazole amide HSP90 inhibitor, which binds to the adenosine triphosphate-binding pocket of HSP90 (Brough et al., 2008). AUY922 has excellent cellular potency against a panel of tumor cell lines (Brough et al., 2008; Eccles included with imatinib or nilotinib, outgrowth of resistant subclones were significantly reduced (Figure 1). Although further pharmacokinetic analysis of AUY922 will be necessary, it is remarkable that even the lowest dose of AUY922 tested in combination with clinically relevant concentrations of nilotinib completely sup- pressed the emergence of resistant clone.

The T315I BCR–ABL mutation is emerging as a common mechanism of failure to second line ABL TKIs. Thus, even in advanced phase of CML, BCR– ABL remains the critical therapeutic target. So far, reports of successful salvage therapy for CML patients who acquire the T315I BCR–ABL mutation are limited to small clinical trials. The isobologram analysis indicated the synergistic effect of simultaneous exposure to AUY922 and nilotinib even in BaF3 cells expressing BCR–ABL mutants including T315I (Figure 2). Further, combination of AUY922 and nilotinib is also effective in inducing apoptosis in both WT BCR–ABL- expressing cells and T315I BCR–ABL-expressing cells (Figures 3c and d). These results indicate that co- treatment with AUY922 sensitizes T315I BCR–ABL- expressing cells to clinically achievable trough levels of nilotinib. However, the structural basis for how co- treatment with AUY922 leads to enhanced activity of nilotinib against the gatekeeper T315I BCR–ABL mutant is entirely unclear. It is possible that nilotinib may collaborate with AUY922 in significantly inhibiting the off-target tyrosine kinase besides BCR–ABL. DDR1 is thought to transducer signals to NFkB pathway (Matsuyama et al., 2004). Higher concentrations of nilotinib abolished DDR1 phosphorylation (Figure 5a). Further, down regulation of DDR1 by siRNA increased antiproliferative activity in AUY922-treated T3151 BCR–ABL BaF3 cells (Figure 5c). Our results clearly show that DDR1 may contribute at least in part to the synergism between AUY922 and nilotinib. Alternative possibility is that AUY922-mediated inhibition of HSP90 chaperon function for BCR–ABL affects the conforma- tional change of BCR–ABL that allows higher concen- trations of nilotinib to interact with T315I BCR–ABL. Co-treatment with AUY922 and nilotinib partially decreased autophosphorylation of T315I BCR–ABL (Figure 5d). In this regard, it is interesting to test whether AUY922 might influence the conformational dynamics of the adenosine triphosphate-binding site of the ‘gatekeeper’ mutant, T315I BCR–ABL. Further studies are required to resolve these mechanisms.

The simultaneous use of AUY922 and nilotinib in chronic phase CML patients might prevent the devel- opment of nilotinib-resistant clones and inhibit growth of highly proliferative leukemia cells through inhibition of kinase activity, thereby providing a rationale for combination strategy. In a survival mouse model using BaF3 cells expressing WT BCR–ABL and mutant forms of BCR–ABL (M244V, G250E, Q252H, Y253F, T315A, T315I, F317L, F317V, M351T and H396P), co-treatments with AUY922 and nilotinib significantly improved the survival (P = 0.009) (Figure 6). The results from these studies suggest that combined use of AUY922 and nilotinib would be a viable strategy for preventing emergence of resistant clones in clinic.

In summary, our preclinical results indicate that AUY922 has potential as an important option for controlling resistance in CML. The combined results of cell-based and in vivo studies suggest that AUY922 exhibits sufficient activity against mutants form of BCR–ABL to warrant consideration for combined use with ABL TKIs. Although several HSP90 inhibitors have now entered clinical evaluation, it is expected that through new formulations of AUY922, orally adminis- trable, it will be more favorably modulate the schedule for CML patients.

Materials and methods

Antibodies and reagents

Anti-ABL Ab (24-11), anti-HSP70 Ab, anti-DDR1 Ab and anti-NQO2 Ab were purchased from Santa Cruz Biotechno- logy, Inc. (Santa Cruz, CA, USA). Anti-phosphotyrosine mAb (PY20) was purchased from Becton Dickinson and Company (Franklin Lakes, NJ, USA). Anti-phospho-ABL Ab was obtained from Cell Signaling (Beverly, MA, USA). AUY922, nilotinib and imatinib were kindly provided by Novartis Pharma AG (Basel, Switzerland).

Cells and cell culture

The approach uses a DNA repair-deficient E. coli strain to produce random mutagenesis of a BCR–ABL retroviral plasmid, infection of BaF3 cells, and selection for BaF3 clones conferring varying degree of drug resistance. BaF3 cells expressing the random mutagenesis of BCR–ABL were kindly provided by Dr James D Griffin (Dana-Farber Cancer Institute, Boston, MA, USA) (Ray et al., 2007). BaF3 cells expressing WT BCR–ABL and mutant forms of BCR–ABL (M244V, G250E, Q252H, Y253F, T315A, T315I, F317L, F317V, M351T and H396P) were described previously (Deguchi et al., 2008). These cell lines were cultured in RPMI1640 (Life Technology, Inc., Carlsbad, CA, USA) supplemented with 10% fetal calf serum (Hyclone Labora- tories, Logan, UT, USA).

The isobologram method

The theoretical basis of isobologram method and the procedure for making isobolograms have previously been described in detail (Kano et al., 2001). Cells were suspended to a final concentration of 1 × 105 cells/ml in fresh medium, plated in 24-well dishes and incubated with AUY922 or imatinib or nilotinib or in combination at 37 1C for 72 h. The number of cells in each well was counted by flow cytometry, and the cell numbers were normalized by dividing the number of cells (Nunoda et al., 2007).

Apoptosis assay

The cells were treated with the indicated concentration of AUY922 and/or nilotinib for 48 h. Annexin V/propidium iodide apoptosis assay was performed according to the manufacturer’s protocol (Becton Dickinson and Company). The cells were gently mixed and immediately analyzed by flow cytometry.

Immunoblotting

Immunoblotting was performed as described previously (Tauchi et al., 1994).

siRNA experiments

siRNA oligonucleotides for murine DDR1 and NQO2 were purchased from Santa Cruz Biotechnology, Inc., and resus- pended in RNase-free H2O at 20 mM. siRNA (1.25 mM) was added to prechilled 0.4 cm-gap electroporation cuvetts (Bio- Rad, Hercules, CA, USA). T315I BaF3 cells (5 × 106) were washed twice in serum-free media and resuspended to 5 × 106 cells per 250 ml of cold, serum-free RPMI 1640. Cells were added to the cuvetts, mixed, and mixed on ice for 5 min. Cells were then pulsed once at 250 mV, 960 mF and 200 ohms by using a Bio-Rad electroporator. At 48 h after electroporation, protein knockdown was determined by immunoblotting, and cells were treated with the indicated concentration of dasatinib or imatinib for 48 h, viable cells were counted by using a Vi-cell XR automated cell viability analyzer (Beckman Coulter, Brea, CA, USA). The mean number of viable cells at varying concentrations of drug was normalized to the mean number of viable cells in the no-drug sample.

In vivo experiments

Twelve-week-old nude mice were injected with 5 × 105 cells of mixture of BaF3 cells expressing WT BCR–ABL and mutant forms of BCR–ABL (M244V, G250E, Q252H, Y253F, T315A, T315I, F317L, F317V, M351T and H396P). At 24-h injection of the leukemia cells, these mice were treated with either vehicle or AUY922 (50 mg/kg i.p.; two times per week) or nilotinib (30 mg/kg; q.d.) or AUY922 (50 mg/kg i.p.; two times per week) + nilotinib (30 mg/kg; q.d.). Mice were observed daily, and body weights as well as signs of stress Luminespib (for example, lethargy, ruffled coat or ataxia) were used to detect possible toxicities.