Oncogene (2007) 26, 4179–4188 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ONCOGENOMICS Identification and functional signature of regulated by structurally different ABL inhibitors

K Nunoda1, T Tauchi1, T Takaku1, S Okabe1, D Akahane1, G Sashida1, JH Ohyashiki2 and K Ohyashiki1

1First Department of Internal Medicine, Tokyo Medical University, Shinjuku-ku, Tokyo, Japan and 2Intractable Disease Research Center, Tokyo Medical University, Shinjuku-ku, Tokyo, Japan

Dasatinib is an ATP-competitive, multi-targeted SRC and Introduction ABL kinase inhibitor that can bind BCR-ABL in both the active and inactive conformations. From a clinical SRC family regulate multiple cellular events standpoint, dasatinib is particularly attractive because it such as proliferation, differentiation, survival, cytoske- has been shown to induce hematologic and cytogenetic letal organization, adhesion and migration as a con- responses in imatinib-resistant chronic myeloid leukemia sequence of their ability to couple with many diverse patients. The fact because the combination of imatinib and classes of cellular receptors and many distinct cellular dasatinib shows the additive/synergistic growth inhibition targets (Thomas and Brugge, 1997). SRC kinases are on wild-type p210 BCR-ABL-expressing cells, we rea- involved in BCR-ABL-mediated transformation and soned that these ABL kinase inhibitors might induce the have been implicated in imatinib resistance (Donato different molecular pathways. To address this question, we et al., 2003; Dai et al., 2004). Multiple domains of BCR- used DNA microarrays to identify genes whose transcrip- ABL interact with and activate SRC kinases indepen- tion was altered by imatinib and dasatinib. K562 cells dently of BCR-ABL kinase activity (Warmuth et al., were cultured with imatinib or dasatinib for 16 h, and 1997; Stanglmaier et al., 2003). The studies with expression data were obtained from three independent dominant-negative mutants suggest that SRC kinases microarray hybridizations. Almost all of the imatinib- and may contribute to the proliferation and survival of mye- dasatinib-responsive genes appeared to be similarly loid cell lines expressing BCR-ABL in vitro (Lionberger increased or decreased in K562 cells; however, small et al., 2000). HCK and LYN are expressed and activated subsets of genes were identified as selectively altered in the acute phase of chronic myeloid leukemia (CML) expression by either imatinib or dasatinib. The distinct patients, and their upregulation correlates with disease genes that are selectively modulated by dasatinib are progression and resistance in patients with imatinib -dependent kinase 2 (CDK2) and CDK8, which had (Hofmann et al., 2002; Donato et al., 2003; Dai et al., a maximal reduction of o5-fold in microarray screen. To 2004; Ptasznik et al., 2004). Therefore, dual SRC and assess the functional importance of dasatinib regulated ABL kinase inhibitors are attractive for the treatment of genes, we used RNA interference to determine whether imatinib-resistant CML. reduction of CDK2 and CDK8 affected the growth Dasatinib is an orally available multitargeted SRC inhibition. K562 and TF-1BCR-ABL cells, pretreated and ABL kinase inhibitor with two-log increased with CDK2 or CDK8 small interfering RNA, showed potency relative to imatinib (Shah et al., 2004). additive growth inhibition with imatinib, but not with Mutations in BCR-ABL that favor the adoption of an dasatinib. These findings demonstrate that the additive/ active, imatinib-resistant conformation are effectively synergistic growth inhibition by imatinib and dasatinib targeted by dasatinib, as shown in cell lines expressing may be mediated in part by CDK2 and CDK8. 14 imatinib-resistant mutants (Shah et al., 2004; Burgess Oncogene (2007) 26, 4179–4188; doi:10.1038/sj.onc.1210179; et al., 2005; O’Hare et al., 2005). Dasatinib prolonged published online 8 January 2007 survival of mice with BCR-ABL-driven disease, and ONCOGENOMICS inhibited proliferation of BCR-ABL-positive bone Keywords: BCR-ABL; tyrosine kinase inhibitor; imatinib; marrow progenitor cells from patients with imatinib- dasatinib sensitive or imatinib-resistant CML (Shah et al., 2004). A recent saturation mutagenesis screening of BCR-ABL also found that the spectrum of mutations that would allow for dasatinib resistance is reduced compared with that of imatinib, including L248R, Q252H, Correspondence: Dr T Tauchi, First Department of Internal Medicine, E255K, V299L, T315I/A and F317V (Burgess et al., Tokyo Medical University, 6-7-1 Nishishinjuku, Shinjuku-ku, Tokyo 2005). Molecular docking studies also showed that 160-0023, Japan. E-mail: [email protected] dasatinib is likely to bind the inactive form of BCR- Received 17 February 2006; revised 26 October 2006; accepted 27 October ABL, although requiring a lower conformational 2006; published online 8 January 2007 stringency, with the ability of binding more intermediate Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4180 conformations than imatinib (Gambacorti-Passerini imatinib would attenuate the downstream molecules of et al., 2005). From a clinical standpoint, dasatinib is BCR-ABL (Figure 1b–d). As compared with treatment particularly hopeful because it has been shown to induce with either agent alone, co-treatment with imatinib hematologic and cytogenetic responses in imatinib- (0.5 mM) and dasatinib (5 nM) or imatinib (1.0 mM) and resistant CML patients (Talpaz et al., 2006). The fact dasatinib (10 nM) for 24 h caused more inhibition of that because the combination of imatinib and dasatinib BCR-ABL autophosphorylation (Figure 1b). Combined shows additive/synergistic growth inhibition on wild- treatment with imatinib and dasatinib also caused more type p210 BCR-ABL-expressing cells, we reasoned that attenuation of the levels of p-Akt, phosphor-nuclear these ABL kinase inhibitors might induce different factor kappa B (NF-kB) and c-Myc, which are down- molecular pathways. To address this question, we used stream of BCR-ABL (Figure 1c and d). DNA microarrays to identify genes whose transcription was altered by imatinib and dasatinib. The distinct genes Dasatinib and imatinib-induced profiles that are selectively modulated by dasatinib are cyclin- K562 cells were cultured with dasatinib or imatinib for dependent kinase 2 (CDK2) and CDK8. To assess the 16 h, and gene expression data from three independent functional importance of dasatinib-regulated genes, we microarray hybridizations were analysed (GPL 2523) used RNA interference to determine whether reduction (Figure 2a and b). The microarray experiment data were of CDK2 and CDK8 affected the growth inhibition. deposited in GEO (GSE 2810). Although the data in K562 and TF-1BCR-ABL cells, pretreated with CDK2 this result demonstrated genuine differential expression, or CDK8 small interfering RNA (siRNA), showed most of these genes showed changes that were relatively additive growth inhibition with imatinib but not with small in magnitude. We, therefore, restricted to genes dasatinib. These findings demonstrate that the additive that showed at least a 1.5-fold change. Almost all of the growth inhibition by imatinib and dasatinib may be dasatinib or imatinib responsible genes appeared to be mediated by CDK2 and CDK8. similarly changed; however, small subsets of genes were identified as selectively altered in expression by either dasatinib or imatinib (Figures 3 and 4). In 16 h, 155 of the 667 unique dasatinib genes and 144 of the 667 Results unique imatinib genes were regulated >1.5-fold. This finding suggests that these two structurally diverse BCR- Analysis of combined drug effects ABL tyrosine kinase inhibitors initially regulate highly First, we investigated the combined use of dasatinib overlapping common target genes and pathway. The with the anti-leukemic agents imatinib, daunorubicin common genes whose expression was affected by (DNR), AraC and VP16 in vitro. Previous study dasatinib and imatinib were categorized into different demonstrated that imatinib showed additive or syner- functional groups based on their biological function, gistic effects in combination with some leukemic agents and genes in the cell proliferation and (Nakajima et al., 2003). We therefore determined categories were examined in greater detail (Figures 3 whether dasatinib could increase the effects of some and 4). anti-leukemic agents, including imatinib, in CML blastic crisis cell line K562. The concentration of anti-leukemic agents is plotted against percentage inhibition of Cell proliferation-related genes proliferation, and each anti-leukemic agent alone is Dasatinib and imatinib affected the expression of several compared with each anti-leukemic agent in combination cyclin-dependent kinases (CDK2, CDK4, CDC5R, with 0.2 nM of dasatinib (Figure 1a). Regression lines CDK6 and CDK8), cell division cycle genes (CDC2L5, generated by Microsoft Excel represent the best-fit CDC7, CDC25A and CDC25B) and cyclines (CCNA2, relationship between drug concentration and the per- CCNC, CCND2, CCND3, CCNE1, CCNE2 and centage inhibition of proliferation. R2 values are the CCNH) (Figure 3a and b). These regulators are known square of the correlation coefficient. When imatinib was to be involved in G1/S and G2/M transition, and their combined with 0.2 nM of dasatinib, the curve showed a decreased gene expression levels and thus reduced substantial shift downward, consistent with increased activities are essential for cell cycle arrest at early stage. antiproliferative activity of the drug combination Dasatinib and imatinib also affected the expression of (Figure 1a). Similarly, with the combination of DNR mitotic inhibitors (CDKN1A, CDKN1B, CDKN1C, or VP-16 and dasatinib, the curve also shifted down CDHN2C and CDK2AP1) (Figure 3b). Obviously, (Figure 1a). When AraC was combined with dasatinib, downregulation/inactivation of such a large number of the two curves crossed at 500 nM of AraC (Figure 1a). cell cycle-related genes may abolish the cell cycle There was no additive effect of the combination of AraC machinery, probably a prerequisite for apoptosis in and dasatinib. Although dasatinib and imatinib bind to dasatinib- or imatinib-treated CML cells. The distinct overlapping sites within the BCR-ABL kinase domain, genes that are selectively modulated by dasatinib are the clinically available concentrations of imatinib may CDK2 and CDK8, which had a maximal reduction not exert an interfering effect by restricting the of o5-fold in microarray screen. CDK2 and CDK8 dasatinib’s access to its . appeared to be candidates involved in the additive/ Next, we determined whether inhibition of BCR- synergistic growth inhibition by dasatinib and imatinib. ABL-tyrosine kinase by co-treatment with dasatinib and Downregulation of these genes can be largely the result

Oncogene Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4181 a Imatinib DNR b ABL 0 0 lP : DNR only Imatinib only 20 20 R2 = 0.9068 R2 = 0.3741 40 40

60 2 60 2 5nM Dasatinib R = 0.7343 R = 0.9343 10nM lmatinib with 0.2nM of % Inhibition % Inhibition with 0.2nM of 80 dasatinib 80 dasatinib

00.10.20.30.40.5 0204060 Imatinib (µM) DNR (nM) Mlmatinib Mlmatinib Mlmatinib Mlmatinib AraC VP-16 1.0 0.5 1.0 10nM Dasatinib 5nM Dasatinib Control 0 0 0.5 VP-16 only 2 BCR-ABL 20 20 R = 0.9625 AraC only 40 2 40 R = 0.3734 Image 60 60 Quantification 100 66 58 71 70 26 16 R2 = 0.4726 (%)

% Inhibition 2 % Inhibition 80 with 0.2nM of R = 0.4726 80 with 0.2nM of BLOT : P. Tyr dasatinib dasatinib 0 200 400 600 0 200 400 600 BCR-ABL AraC (nM) VP-16 (nM) BLOT : ABL c d 5nM Dasatinib 10nM Dasatinib Mlmatinib Mlmatinib Mlmatinib Mlmatinib 10nM Dasatinib 5nM Dasatinib 1.0 10nM Dasatinib 0.5 1.0 5nM Dasatinib 0.5 Control

Phospho-NF- B Mlmatinib Mlmatinib Mlmatinib Mlmatinib Image Quantification 100 71 5263 53 25 18 % 0.5 1.0 1.0 10nM Dasatinib 5nM Dasatinib 0.5 Control BLOT : Phospho-NFkB Phospho-Akt NF- B Image Quantification 100 94 72 92 43 36 33 % BLOT : NF- B BLOT : Phospho-Akt c-Myc Akt Image Quantification 100 78 49 58 54 27 16 % BLOT : Akt BLOT : c-Myc

Figure 1 Analysis of combined drug effects. (a) K562 cells were suspended to a final concentration of 1 Â 105 cells/ml in fresh medium, and incubated with anti-leukemic agents alone or in combination with 0.2 nM of dasatinib at 371C for 72 h. The number of cells in each well was counted by flow cytometry. Regression lines generated by Microsoft Excel represent the best-fit relationship between drug concentration and the percentage inhibition of proliferation. R2 values are the square of the correlation coefficient. Similar results were obtained in each of three separate experiments. (b) K562 cells were cultured with indicated concentrations of dasatinib or imatinib for 24 h, and cell lysates were immunoprecipitated with ant-ABL mAb. The immunoprecipitates were immunoblotted with anti- phosphotyrosine mAb (PY20) or anti-ABL Ab. (c) K562 cells were cultured with indicated concentrations of dasatinib or imatinib for 24 h, and the cell lysates were immunoblotted with anti-phospho-Akt or anti-Akt Ab. (d) K562 cells were cultured with indicated concentrations of dasatinib or imatinib for 24 h, and the cell lysates were immunoblotted with anti-phospho-NF-kB, anti-NF-kBor anti-c-Myc Ab. of chain reactions of transcriptional inactivation. For Apoptosis-related genes instance, CCD2 is known to be directly regulated by The intrinsic apoptotic pathway is important for MYC at transcriptional sites, and its downregulation is dasatinib- or imatinib-mediated apoptosis. Both ABL logically linked to downregulated MYC (Figure 3a). kinase inhibitors modulated the expression of many genes that play a key role in this pathway, such as BAX, BAK1, BID, BCL2, BCL6, BCL9, MCL1, CASP1, Oncogenic signals genes CASP2, CASP3, CASP6, CASP7 CASP8, CASP9 and Inactivation of BCR-ABL tyrosine kinase by dasatinib CASP10 (Figure 4a). Moreover, genes that regulate or imatinib blocks the transcription factors, and thus the induction or prolongation of this pathway, includ- represses expression of target genes such as those ing NF-kB-pathway genes (NF-kB1 and NF-kB1A), encoding molecules, as highlighted were also transcriptionally reduced (Figure 4b). by members of STAT pathway (STAT1, STAT3, GADD45B, which induces cell cycle arrest at G2/M, STAT4, STAT5A, STAT5B and STAT6) (Figure 3c). was also reduced by dasatinib and imatinib (Figure 4b).

Oncogene Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4182

Figure 2 Hierarchical cluster analysis of gene expression profiles induced by dasatinib or imatinib in K562 cells. (a and b) K562 cells were cultured with 100 nM of dasatinib or 5 mM of imatinib for 16 h, and gene expression data from three-independent microarray hybridizations were analysed (GPL 2523). The microarray experiment data were deposited in GEO (GSE 2810). The scale in this figure shows the level of expression, where red indicates increased gene expression, green indicates decreased expression, and the intensity of color correlated to the magnitude changes. Black indicates no change.

Furthermore, genes coding Validation of expression profiles of selected genes (TNF) family receptors and ligands (TNF, TNFAIP3, To verify the changes of CDKs expression, which were TNFRSR5 and TNFRSF12) were activated by dasatinib identified as selectively altered by either dasatinib or or imatinib (Figure 4b). imatinib, we performed the immunoblot analysis focusing on CDK2, CDK3, CDK4, CDK6, CDK8 and CDK9 (Figure 5a). K562 cells were cultured with indicated DNA-damage repair genes concentrations of either dasatinib or imatinib for 48 h, BCR-ABL-transformed cells appeared to be better and the cell lysates were immunoblotted with the equipped to survive genotoxic damage because of indicated antibodies (Figure 5a). expressions of enhanced ability to repair DNA lesions, prolonged CDK2 and CDK8 were selectively reduced by dasatinib activation of the G/2M checkpoint to provide more time treatment (Figure 5a). Quantitative real-time PCR for repair, and inhibited apoptosis mechanisms. Dasa- analysis of these CDKs confirmed that protein expression tinib and imatinib reduced the expression of several changes induced by dasatinib and imatinib correlated DNA-damage repair pathway genes, such as RAD50, with changes in gene expression (data not shown). We RAD51L3, RECQL, XRCC1, XRCC3 and XRCC4 also determined the protein expression of CDK2 and (Figure 4c). Diverse modulation of these DNA-damage CDK8 in TF-1-BCR-ABL cells after dasatinib or repair genes by both ABL kinase inhibitors may imatinib treatment (Figure 5b). Expressions of CDK2 participate in the reduced possibility of therapeutic drug and CDK8 was mainly reduced by dasatinib-treatment resistance. (Figure 5b). As caspase-3 activity might affect the

Oncogene Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4183 a Cell Proliferation-Related Genes-1 a Apoptosis-Related Genes-1

CDK2 BAX Dasatinib Dasatinib CDK4 BAK1 Imatinib Imatinib CDK5R1 BID CDK6 BCL2 CDK8 BCL6 CDK9 BCL9 CDC2L5 MCL1 CDC7 CASP1 CDC25A CASP2 CDC25B CASP3 CDC25C CASP6 MYC CASP7 E2F3 CASP8 E2F4 CASP9 E2F5 CASP10 -6 -4 -20 2 4 -6 -4 -2 02 4 Normalized ratio (log2) Normalized ratio (log2) Proliferation-Related Genes-2 b Apoptosis-Related Genes-2 CCNA2 Dasatinib NFKB1 CCNC Dasatinib Imatinib NFKB2 CCND2 Imatinib NFKBIA CCND3 IKBKB CCNE1 IKBKE CCNE2 IKBKG CCNH GADD45B CDKN1A CDKN1B TNF CDKN1C TNFAIP3 CDKN2C TNFRSF5 CDK2AP1 TNFRSF12 DEDD -6 -4 -20 2 4 -6 -4 -2 02 4 6 Normalized ratio (log2) Normalized ratio (log2) c Oncogenic Signals Genes c DNA Damage Repair Genes STAT1 Dasatinib STAT3 RAD50 Imatinib Dasatinib STAT4 RAD51 Imatinib STAT5A RECQL STAT5B XRCC1 STAT6 XRCC2 -6 -4 -20 2 4 6 XRCC3 XRCC4 Normalized ratio (log2) -6 -4 -2 02 4 6 Normalized ratio (log2) Figure 3 Dasatinib- and imatinib-regulated cell proliferation genes and STAT family genes. (a) Genes encode members of cyclin-dependent kinases, cell division cycle genes, c-myc and E2F Figure 4 Dasatinib- and imatinib-regulated apoptosis genes and family. (b) Genes encode members of , mitotic inhibitors. DNA-damage repair genes. (a and b) Genes encode members of (c) Genes encode members of STAT family. apoptotic . (c) Genes encode members of NF-kB pathways and the death receptor pathway. reduction of CDK2 or CDK8, therefore, we determined to determine whether reduction in CDK2 and CDK8 the time-dependent changes of CDK2/CDK8 expression affect the proliferation. K562 and TF-1BCR-ABL and caspase-3 and poly(ADP-ribose)polymerase (PARP) cells were transfected with control siRNA or CDK2 activation in K562 cells (Figure 5c and d). K562 cells siRNA or CDK8 siRNA; then the CDK2 and CDK8 were cultured with either 100 nM of dasatinib or 10 mM of expression was analysed by immunoblotting after 48 h imatinib for the indicated time, and the cell lysates were (Figure 6a). The siRNA to CDK2 or CDK8 specifically immunoblotted with indicated antibodies (Figure 5c and reduced CDK2 or CDK8 expression (Figure 6a). At d). The expression of CDK2 or CDK8 proteins was 48 h after transfection, K562 and TF-1BCR-ABL cells reduced after 24 h of dasatinib treatment (Figure 5c); were treated with the indicated concentration of however, the expression of CDK2 or CDK8 proteins was dasatinib or imatinib for 48 h, and viable cells were nearly unchanged after imatinib treatment (Figure 5d). counted (Figure 6b and c). Increasing doses of imatinib, As activation of caspase-3 or PARP was observed 12– in the presence of CDK2 siRNA or CDK8 siRNA, 24 h after dasatinib or imatinib treatment (Figure 5c and shifted the dose–response curve substantially down- d), it is likely that caspase-3 activity may not be required ward, consistent with increased antiproliferative activity for the reduction of CDK2/CDK8 protein level. (Figure 6b and c). When dasatinib was treated in the presence of CDK2 siRNA or CDK8 siRNA, the two curves crossed at 200 nM of dasatinib (Figure 6b and c). siRNA-mediated knock down of CDK2 and CDK8 in These results demonstrated that transcriptional repres- K562 cells and TF-1BCR-ABL cells sion of CDK2 or CDK8 can play an important role in To assess the functional importance of dasatinib- and the combined growth inhibitory effects of dasatinib and imatinib-regulated gene, we used RNA interference imatinib.

Oncogene Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4184 Discussion and imatinib largely activated common apoptotic path- ways, although there were differences in the activities Imatinib and dasatinib are clinically active ABL tyrosine of the two compounds. Dasatinib is capable of bind- kinase inhibitors that show the additive/synergistic ing BCR-ABL with less stringent conformational growth inhibition on BCR-ABL-expressing cells requirements with respect to imatinib (Gambacorti- (Figure 1a). Earlier studies demonstrated that dasatinib Passerini et al., 2005). Although both compounds inhibit

a Dasatinib nM Imatinib M Dasatinib nM Imatinib M b 010 50 100 1 2 5 10 010 50 100 1 2 5 10 CDK2 CDK3 TF-1 BCR-ABL (%) 100 73 56 36 99 102 97 98 Image Quantification 100 8278 82 100 93 100 99 (%) BLOT : CDK2 BLOT : CDK3 Dasatinib nM Inatinib M 0 10 50 100 1 2 5 10 CDK4 CDK2 100 51 33 32 86 85 46 37 (%) CDK8 100 91 57 34 96 95 71 69 (%) BLOT : CDK4 (%) 100 56 35 16 96 97 75 66 Image BLOT : CDK2 Quantification BLOT : CDK8 CDK6 CDK8 100 53 44 38 94 99 62 51 (%) 100 87 39 17 91 90 83 77 BLOT : CDK6 (%) BLOT : CDK8

CDK9 actin Cleaved PARP BLOT : actin 100 86 50 41 91 94 66 63 (%) BLOT : CDK9 BLOT : PARP

c Dasatinib d lmatinib 026122448 hrs 026122448 hrs CDK2 CDK2 100 90 78 77 82 80 (%) 100 82 83 85 45 38 (%) BLOT : CDK2 BLOT : CDK2 CDK8 CDK8 100 98 106 98 84 85 (%) 100 105 102 41 35 14 (%) BLOT : CDK8 BLOT : CDK8 cleaved cleaved caspase-3 caspase-3 BLOT : cleaved-caspase-3 BLOT : cleaved caspase-3 cleaved PARP cleaved PARP

BLOT : PARP BLOT : PARP Figure 5 Validation of microarray data of cyclin-dependent kinase genes. (a) K562 cells were cultured with indicated concentrations of either dasatinib or imatinib for 48 h, and the cell lysates were immunoblotted with the indicated antibodies. Quantitative analysis of CDK2 and CDK8 expression was carried out by analysing hyper-ECL films of immunoblotting by using a densitometer. (b) TF- 1BCR-ABL cells were cultured with indicated concentrations of dasatinib or imatinib for 48 h, and the cell lysates were immunoblotted with anti-CDK2 or anti-CDK8 or anti-PARP Ab. (c and d) K562 cells were cultured with either 100 nM of dasatinib (c)or10mM of imatinib (d) for the indicated time, and the cell lysates were immunoblotted with anti-CDK2, anti-CDK8 or anti-cleaved caspase-3, or anti-PARP Ab.

Figure 6 Knock down of CDK2 and CDK8 expression affected the proliferation in dasatinib- or imatinib-treated K562 cells and TF- 1BCR-ABL cells. (a) K562 cells and TF-1BCR-ABL cells transfected with 1.25 mM of control (GFP), or CDK2 siRNA or CDK8 siRNA were analysed 48 h after transfection for CDK2 or CDK8 expression by immunoblotting. (b and c) At 48 h after transfection, K562 (b) and TF-1BCR-ABL cells (c) were treated with incubated concentrations of dasatinib or imatinib for 48 h; viable cells were counted by using a Vi-cell XR automated cell viability analyzer (Beckman Coulter). The mean number of viable cells at different concentrations of drug was normalized to the mean number of viable cells in the no-drug sample. Similar results were obtained in each of three independent experiments.

Oncogene Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4185 a K562 TF-1 BCR-ABL Control CDK2 siRNA CDK8 siRNA Control CDK8 siRNA Control CDK2 siRNA CDK8 siRNA Control CDK2 siRNA CDK8 siRNA CDK2 siRNA CDK2 CDK8 CDK2 CDK8 (%) 100 32 98 100 102 45 100 33 82 100 106 21 BLOT : CDK2 CDK8 CDK2 CDK8

actin actin actin actin

BLOT : actin actin actin actin b K562 100 lmatinib 100 lmatinib lmatinib+CDK2siRNA lmatinib+CDK8 siRNA 80 80

60 2 60 2 % R =0.8183 % R =0.8183 40 40

20 20 R2 = 0.8548 R2 = 0.764 0 0 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 lmatinib ( M) lmatinib ( M)

100 Dasatinib 100 Dasatinib Dasatinib+CDK2 siRNA Dasatinib+CDK8 siRNA 80 80 R2 = 0.5389 R2 = 0.5389 60 60 % R2 = 0.8085 % R2 = 0.7034 40 40

20 20

0 0 0200300100 0200300100 Dasatinib (nM) Dasatinib (nM) c TF-1BCR-ABL

100 lmatinib 100 lmatinib lmatinib+CDK2 siRNA lmatinib+CDK8 siRNA 80 80 R2 = 0.7819 R2 = 0.7819 60 60 % % 40 40 R2 = 0.9071 R2 = 0.9924 20 20

0 0 0 0.5 1.0 1.5 2.0 2.5 0 0.5 1.0 1.5 2.0 2.5 lmatinib ( M) lmatinib ( M)

100 Dasatinib 100 Dasatinib Dasatinib+CDK2 siRNA Dasatinib+CDK8 siRNA 80 80 R2 = 0.5086 R2 = 0.5086 60 60 % R2 = 0.964 % 40 40 R2 = 0.7468 20 20

0 0 0200300100 0200300100 Dasatinib (nM) Dasatinib (nM)

Oncogene Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4186 BCR-ABL tyrosine kinase, whether these two structu- S phase requires sequential activation of CDK2 and rally diverse compounds mediate similar or disparate CDK4 (Sherr and Roberts, 1999). The role of CDK4 is gene expression changes has not been addressed. In the well established; however, CDK2 may be regulated present study, we used microarray gene expression differently in somatic cells and in cells (Tetsu and profiles to identify the genes commonly and selectively McCormick, 2003). In BCR-ABL-transformed cells, regulated by dasatinib and imatinib. cyclin D activates CDK4/CDK6, which inturn transac- On the basis of clustering, the transcriptional re- tivates cyclin E by releasing E2F. Cyclin E activates sponses induced by dasatinib and imatinib were CDK2, and BCR-ABL causes relocation of p27 to the remarkably similar (Figure 2a). Some degree of overlap cytoplasm, further stimulating CDK2 (Jiang et al., was expected because both compounds target BCR- 2000). Therefore, CDK2 seems to have a critical role ABL kinase. Compatible with this phenotype, some in cell cycle progression in BCR-ABL-transformed proapoptotic genes were upregulated, a number of cells. Downregulation of CDK2 or CDK8 by siRNA antiapoptotic genes were downregulated by both ABL increased antiproliferative activity in imatinib-treated kinase inhibitors, and a number of genes within the cells, but not in dasatinib-treated cells (Figure 6a–c). same molecular pathway were coordinately regulated Our results clearly demonstrate that diverse regulation (Figures 3 and 4). These include BAX, BCL-2, MCL-1, of CDKs by ABL kinase inhibitors may contribute at CASP2 and CASP10 (Figure 4a). Although some of the least in part to additive growth inhibition in these proapoptotic genes were repressed (i.e., BID, CASP3 compounds. The pleiotropic molecular sequel of SRC/ and CASP6), the overall response was one that would ABL inhibition does not allow us to conclude which provide a strong proapoptotic signal (Figure 4a and b). pathways are most important for anti-tumor effect, but Moreover, repression of the NF-kB pathway is also potentially offer a major therapeutic advantage, namely associated with apoptotic stimulus (Figure 4b). Further- the simultaneous targeting of different proliferative/ more, the transcription of TNF superfamily genes and antiapoptotic pathways in tumor cells (Supplementary genes involved in death-receptor signaling (DEDD) was Figure). altered by dasatinib or imatinib (Figure 4b). One of the objectives of this study of the molecular In addition to inducing apoptosis, ABL kinase profile of dasatinib-treated BCR-ABL-transformed cells inhibitors can suppress cell cycle progression (Gesbert was to establish a framework for designing of combina- et al., 2000; Parada et al., 2001). The transcriptional tion therapies with conventional anti-leukemia agents. response was consistent with growth suppression, and a This study identified the enhanced antiproliferative number of genes within key growth-regulatory pathways activity of combinations with dasatinib with anti- were coordinately regulated (Figure 3a and b). For leukemia agents, such as imatinib, daunorubicine and example, decreased expression of MYC, E2F3, E2F4 VP-16 (Figure 1a). The ability of dasatinib to suppress and E2F5 is consistent with suppression of the MYC genes involved in cell proliferation-related genes (i.e., pathway (Figure 3a). The number of CDKs and CDK CDK2, CDK6, CDK8, CDC7, CCNA2, CCNC, CCNE1, inhibitors, whose expression is highly coordinated to CCNE2 and MYC) and antiapoptotic genes (BCL2, regulate appropriate cell cycle progression, was aber- MCL1 and NFkB) superior to imatinib constitute a rantly expressed in dasatinib- or imatinib-treated cells molecular basis for the chemo-sensitizing effect of SRC/ (Figure 3b). CDK2 and CDK8 are selectively modulated ABL kinase inhibition. by dasatinib, which had a maximal reduction of o5-fold This study provides comparative gene expression in microarray screen (Figure 3a). We have shown that profiling analysis of the effect of dasatinib and imatinib, dasatinib and imatinib and the combination of these two compounds that are clinically active for BCR-ABL- compounds differently suppress the expression of c-Myc positive leukemia. Both agents commonly target a protein (Figure 1d). Recently, Samanta et al. (2006) number of important molecular pathways that regulate demonstrated that signal transduction by BCR-ABL/ cell growth and survival. A single proapoptotic or anti- Jak2 network results in of Akt, which proliferative pathway may not be critical for the leads to the stabilization of c-Myc and activates NF-kB therapeutic effects of ABL kinase inhibitors. The present to cause elevation of c-Myc transcripts. As compared findings have important implications for the clinical use with treatment with either agent alone, co-treatment of dasatinib and imatinib as an anti-leukemia agent, with dasatinib and imatinib caused more attenuation of either alone or in combination with other agents. the levels of phospho-Akt, phospho-NFkB and c-Myc (Figure 1c and d). As CDK2 is a c-Myc transcriptional target (Prathapam et al., 2006), downregulation of a large number of cell Materials and methods cycle-related genes, including CDK2 and CDK8, could be largely the result of chain reactions of transcriptional Antibodies and reagents inactivation by these ABL tyrosine kinase inhibitors. Anti-CDK2 Ab (D-12), anti-CDK3 Ab (Y-20), anti-CDK4 Ab (H-22), anti-CDK6 Ab (C-21), anti-CDK8 Ab (D-9), anti- Earlier study (Riley et al., 2001) demonstrated that CDK9Ab (C-20), anti-actin Ab (C-2) and anti-ABL Ab (24- v-SRC induces the expression of CDK2 in Rat-1 cells. 11) were purchased from Santa Cruz Biotechnology, Inc. Therefore, inactivation of c-SRC and SRC-family (Santa Cruz, CA, USA). Antiphosphotyrosine mAb (PY20) was kinase by dasatinib may decrease the expression of purchased from Becton Dickinson and Company (Franklin CDK2. Progression through the cell cycle from G1 to Lakes, NJ, USA). Anti-Akt Ab, anti-phospho-Akt (Ser473)

Oncogene Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4187 Ab, anti-NF-kB p65 Ab and anti-phospho-NF-kB p65 USA) as raw data. The scanned data were normalized, verified (Ser536) Ab were purchased from Cell Signaling (Beverly, and analysed using the Genomic Profiler software (Mitsui MA, USA). Anti-c-Myc Ab was purchased from NOVUS Knowledge Industry, Tokyo, Japan) as described previously (Littleton, CO, USA). Dasatinib was kindly provided by (Ohyashiki et al., 2005; Takaku et al., 2005; Zhang et al., Bristol-Myers Squibb (New York, NY, USA). DNR, cytosine 2006). First, the value was adjusted by subtraction of arabinoside (Ara C) and etoposide (VP-16) were obtained background fluorescence of an equivalent area. Analysis was from Sigma (St Louis, MO, USA). carried out by taking the median signal of the probe value for each transcription set, and the 75% rank for the total Cells and cell culture hybridization was calculated. Microarray data obtained from K562 cells were obtained from the American Type Culture three independent experiments were then verified in a single Collection (Rockville, MD, USA). TF-1BCR-ABL cells were file. The normalized log data of fluorescence ratio (Cy5–Cy3) described previously (Komatsu et al., 2003). These cell lines which was quantified for each gene to reflect the relative were cultured in RPMI1640 (Life Technology Inc., Rockville, abundance of gene in each experimental sample compared with MD, USA), supplemented with 10% FCS (Hyclone Labora- reference sample, were deposited with GEO. For statistical tories, Logan, UT, USA). analysis of host gene expression, we also utilized a GeneSifter (VizXLabs, Seattle, WA, USA). Analysis of variance, and Analysis of combined drug effects Student’s t-test were performed using GeneSifter. Values of K562 cells were suspended to a final concentration of Po0.05 were considered to indicate a statistically significant 1 Â 105 cells/ml in fresh medium, plated in 24-well dishes and difference, and the Benjamini–Hochberg algorithm was used incubated with anti-leukemic agents alone or in combination for estimation of false discovery rates. with 0.2 nM of dasatinib at 371C for 72 h. The anti-leukemic agents used were imatinib, DNR, AraC or VP16. The number Immunoblotting and immunoprecipitation of cells in each well was counted by flow cytometry, and the Immunoblotting and immunoprecipitation were performed cell numbers were normalized by dividing the number of cells as described previously (Tauchi et al., 1994). Quantitative in the absence of anti-leukemic agents or with dasatinib alone. analysis of CDK2 and CDK8 expression was carried out by The data were plotted as the concentration of anti-leukemic analysing hyper-enhanced chemiluminescence (ECL) films of agents against the percentage inhibition of proliferation. To immunoblotting by using a densitometer (Bio-Rad, Hercules, determine the relationship between the percentage inhibition CA, USA). of proliferation and drug concentration, a best-fit regression line was generated by Microsoft Excel. Small interfering RNA experiments Oligonucleotide DNA microarray hybridization siRNA oligonucleotides for CDK2 and CDK8 were purchased We have designed two types of pathway focusing on low- from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA), density oligonucleotide microarrays (Novusgene Inc., Tokyo, and resuspended in RNase-free H2Oat20mM. siRNA Japan), which contain 667 selected genes related to cell growth, (1.25 mM) was added to prechilled 0.4 cm-gap electroporation cell cycle, apoptosis, transcription, DNA repair and cell stress cuvettes (Bio-Rad). K562 (5 Â 106) or TF-1BCR-ABL cells responses (GEO accession number: GPL2523). This DNA chip (5 Â 106) were washed twice in serum-free media and resus- was made up of tetra-plicate spots of 60-mer highly specific pended to 5 Â 106 cells per 250 ml of cold, serum-free RPMI oligonucleotide probes. For DNA microarray analysis of 1640. Cells were added to the cuvettes, mixed, and further genes, we used 1 mg of mRNA from K562 cells treated with mixed for 5 min on ice. Cells were then pulsed once at 250 mV, dasatinib or imatinib for 16 h. Hybridization was carried out 960 mF and 200 O by using a Bio-Rad electroporator. At 48 h automatically using Gene TAC Hybridization (Genomic after electroporation, protein knock down was determined by Solution, Ann Arbor, MI, USA) according to the supplier’s immunoblotting, cells were treated with the indicated con- instructions. The conditions of hybridization were 551C for centration of dasatinib or imatinib for 48 h, and viable cells 2h,501C for 2 h and 461C for 2 h, then 421C for 12 h. were counted by using a Vi-cell XR automated cell viability analyzer (Beckman Coulter, Fullerton, CA, USA). The mean Data analysis and statistic validation number of viable cells at different concentrations of drug was The hybridization signals were scanned by GenePix 4000B normalized to the mean number of viable cells in the no-drug Microarray Scanner (Axon Instruments, Union City, CA, sample.

References

Burgess MR, Skaggs BJ, Shah NP, Lee FY, Sawyers CL. overexpression in chronic myelogenous leukemia cells (2005). Comparative analysis of two clinically active BCR- selected for resistance to STI571. Blood 101: 690–698. ABL kinase inhibitors reveals the role of conformation- Gambacorti-Passerini C, Gasser M, Ahmed S, Assouline S, specific binding in resistance. Proc Natl Acad Sci USA 102: Scapozza L. (2005). Abl inhibitor BMS354825 binding mode 3395–3400. in Abelson kinase revealed by molecular docking studies. Dai Y, Rahmani M, Corey SJ, Dent P, Grant S. (2004). A Leukemia 19: 1267–1269. BCR/ABL-independent, Lyn-dependent form of imatinib Gesbert F, Sellers WR, Signoretti S, Loda M, Griffin GD. mesylate (STI-571) resistance is associated with altered (2000). BCR/ABL regulates expression of the cyclin- expression of Bcl-2. J Biol Chem 279: 34227–34239. dependent kinase inhibitor p27Kip1 through the phospha- Donato NJ, Wu JY, Stapley J, Gallick G, Lin H, Arlinghaus R tidylinositol 3-kinase/AKT pathway. J Biol Chem 275: et al. (2003). BCR-ABL independent and LYN kinase 39223–39230.

Oncogene Microarray analysis of BCR-ABL tyrosine kinase inhibitors K Nunoda et al 4188 Hofmann WK, de Vos S, Elashoff D, Gschaidmeier H, Ptasznik A, Nakata Y, Kalota A, Emerson SG, Gewirtz AM. Hoelzer D, Koeffler HP et al. (2002). Relation between (2004). Short interfering RNA (siRNA) targeting the Lyn resistance of Philadelphia--positive acute lym- kinase induces apoptosis in primary, and drug-resistant, phoblastic leukemia to the tyrosine kinase inhibitor STI571 BCR-ABL1(+) leukemia cells. Nat. Med 10: 1187–1189. and gene-expression profiles: a gene expression study. Riley D, Carragher NO, Frame MC, Wyke JA. (2001). The Lancet 359: 481–486. mechanism of cell cycle regulation by v-Src. Oncogene 20: Jiang Y, Zhao RC, Verfaillie CM. (2000). Abnormal 5941–5950. integrin-mediated regulation of chronic myelogenous Samanta AK, Lin H, Sun T, Kantarjian H, Arlinghaus RB. leukemia CD34+ cell proliferation: BCR/ABL up-regulates (2006). Janus kinase 2: a critical target in chronic the cyclin-dependent kinase inhibitor, p27Kip, which myelogenous leukemia. Cancer Res 66: 6468–6472. is relocated to the cell cytoplasm and incapable of regula- Shah NP, Tran C, Lee FY, Chen P, Norris D, Sawyers CL. ting cdk2 activity. Proc Natl Acad Sci USA 97: (2004). Overriding imatinib resistance with a novel ABL 10538–10543. kinase inhibitor. Science 305: 399–401. Komatsu N, Watanabe T, Uchida M, Mori M, Kirito K, Sherr CJ, Roberts JM. (1999). CDK inhibitors: positive and Kikuchi S et al. (2003). A member of Forkhead transcrip- negative regulators of G1-phase progression. Genes Dev 13: tion factor FKHRL1 is a downstream effector of STI571- 1501–1512. induced cell cycle arrest in BCR-ABL-expressing cells. J Biol Stanglmaier M, Warmuth M, Kleinlein I, Hallek M. (2003). The Chem 278: 6411–6419. interaction of the BCR-ABL tyrosine kinase Hck is mediated Lionberger JM, Wilson MB, Smithgall TE. (2000). Transfor- by multiple binding domains. Leukemia 17: 286–289. mation of myeloid leukemia cells to cytokine independence Takaku T, Ohyashiki JH, Zhang Y, Ohyashiki K. (2005). by BCR-ABL is suppressed by kinase-defective Hck. J Biol Estimating immunoregulatory gene networks in human Chem 275: 18581–18585. herpesvirus type 6-infected T cells. Biochem Biophys Res Nakajima A, Tauchi T, Sumi M, Bishop WR, Ohyashiki K. Commu 336: 469–477. (2003). Efficacy of SCH66336, a farnesyl Talpaz M, Shah NP, Kantarjian H, Donato N, Nicoll J, inhibitor, in conjunction with imatinib against BCR-ABL- Paquette R et al. (2006). Dasatinib in imatinib-resistant positive cells. Mol Cancer Ther 2: 219–224. Philadelphia chromosome-positive leukemias. N Engl J Med O’Hare T, Walters DK, Stoffregen EP, Jia T, Manley PW, 354: 2531–2541. Mestan J et al. (2005). In vitro activity of BCR-ABL Tauchi T, Feng GS, Shen R, Song HY, Donner D, Pawson T inhibitors AMN107 and BMS-354825 against clinically et al. (1994). SH2-containing phosphotyrosine phosphatase relevant imatinib-resistant Abl kinase domain mutants. Syp is a target of p210bcr-abl tyrosine kinase. J Biol Chem Cancer Res 65: 4500–4505. 269: 15381–15387. Ohyashiki JH, Takaku T, Ojima T, Abe K, Yamamoto K, Tetsu O, McCormick F. (2003). Proliferation of cancer cells Zhang Y et al. (2005). Transcriptional profiling of human despite CDK2 inhibition. Cancer Cell 3: 233–245. herpesvirus type B (HHV-6B) in an adult T cell leukemia cell Thomas SM, Brugge JS. (1997). Cellular functions regulated line as in vitro model for persistent infection. Biochem by Src family kinases. Annu Rev Cell Dev Biol 13: 513–609. Biophys Res Commun 329: 11–17. Warmuth M, Bergmann M, Priess A, Hauslmann K, Parada Y, Banerji L, Glassford J, Lea NC, Collada M, Emmerich B, Hallek M. (1997). The Src family kinase Hck Rivas C et al. (2001). BCR-ABL and interleukin 3 promote interacts with BCR-ABL by a kinase-independent mechan- haematopoietic cell proliferation and survival through ism and phosphorylates the Grb2-binding site of BCR. modulation of cyclin D2 and p27Kip1 expression. J Biol J Biol Chem 272: 33260–33270. Chem 276: 23572–23580. Zhang Y, Ohyashiki JH, Takaku T, Shimizu N, Ohyashiki K. Prathapam T, Tegen S, Oskarsson T, Trumpp A, Martin GS. (2006). Transcriptional profiling of Epstein–Barr virus (2006). Activated Src abrogates the Myc requirement for the (EBV) genes and host cellular genes in nasal NK/T-cell G0/G1 transition but not for the G1/S transition. Proc Natl lymphoma and chronic active EBV infection. Br J Cancer Acad Sci USA 103: 2695–2700. 94: 599–608.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene