CBL Mutation-Related Patterns of Phosphorylation and Sensitivity to Tyrosine Kinase Inhibitors

ORIGINAL ARTICLE CBL mutation-related patterns of phosphorylation and sensitivity to tyrosine kinase inhibitors

H Makishima1, Y Sugimoto1, H Szpurka1, MJ Clemente1,KPNg1, H Muramatsu2, C O’Keefe1, Y Saunthararajah1 and JP Maciejewski1

Recurrent homozygous CBL-inactivating mutations in myeloid malignancies decrease ubiquitin ligase activity that inactivates SRC family kinases (SFK) and receptor tyrosine kinases (RTK). However, the most important SFK and RTK affected by these mutations, and hence, the most important therapeutic targets, have not been clearly characterized. We compared SFK and RTK pathway activity and inhibitors in acute myeloid leukemia cell lines containing homozygous R420Q mutation (GDM-1), heterozygous deletion (MOLM13) and wild-type (WT) CBL (THP1, U937). As expected with CBL loss, GDM-1 displayed high KIT expression and granulocyte-macrophage colony-stimulating factor (GM-CSF) hypersensitivity. Ectopic expression of WT CBL decreased GDM-1 proliferation but not cell lines with WT CBL. GDM-1, but not the other cell lines, was highly sensitive to growth inhibition by dasatinib (dual SFK and RTK inhibitor, LD50 50 nM); there was less or no selective inhibition of GDM-1 growth by sunitinib (RTK inhibitor), imatinib (ABL, KIT inhibitor), or PP2 (SFK inhibitor). Phosphoprotein analysis identiﬁed phosphorylation targets uniquely inhibited by dasatinib treatment of GDM-1, including a number of proteins in the KIT and GM-CSF receptor pathways (for example, KIT Tyr721, STAT3 Tyr705). In conclusion, the promiscuous effects of CBL loss on SFK and RTK signaling appear to be best targeted by dual SFK and RTK inhibition.

Leukemia (2012) 26, 1547--1554; doi:10.1038/leu.2012.7 Keywords: CBL; UPD11q; GDM-1; dasatinib

INTRODUCTION The poor prognosis associated with CBL mutations necessitates Acquired somatic copy-neutral loss of heterozygosity (LOH), also new treatment approaches. However, to date the impact of referred as to uniparental disomy (UPD), is frequently found in specific therapies currently available to patients with CBL mutant myelodysplastic syndrome (MDS), MDS/myeloproliferative neo- leukemia has not been explored. Identification of pathogenic plasms (MDS/MPN) and secondary acute myeloid leukemia (AML) pathways resulting from a blockade of the ubiquitination activity and may point towards genes harboring homozygous muta- of CBL may provide clues as to possible molecular targets, tions.1,2 UPD11q23.3 is present in 7% of patients with chronic including SFK and RTK. Our experiments were designed to clarify myelomonocytic leukemia (CMML) and 11% of cases of AML the pathogenesis of CBL mutations in myeloid malignancies, using derived from this condition.3--6 This recurrent lesion has been 2 CBL mutant cell lines, in particular the GDM-1 cell line with a shown to coincide with homozygous CBL mutations, mostly homozygous R420Q CBL mutation, as models for primary CBL located in ring finger (RFD) or linker domain of the gene.4--6 mutant neoplasms with RFD mutations. CBL is a member of the E3 ubiquitin ligase CBL family, which poly- or mono-ubiquitinates activated SRC family kinases (SFK) MATERIALS AND METHODS and various receptor tyrosine kinases (RTK).7--9 Inactivation of ubiquitination activity through mutations of the RFD may lead to Patients enhanced and/or prolonged receptor signaling, which can Informed consent was obtained according to protocols approved by the contribute to the clinical phenotype of patients with CBL institutional boards of Cleveland Clinic, Johns Hopkins University and UCLA mutations. In contrast to patients with UPD11q, those who have Medical Center. We enrolled 493 patients with various myeloid malig- LOH due to del11q only rarely had hemizygous CBL mutations and nancies (179 MDS, 162 MDS/MPN, 55 MPN, 97 primary AML). Diagnosis was 10 only 4/442 (0.9%) patients with heterozygous mutation of CBL assigned according to WHO classification criteria. were identified, suggestive of the tumor suppressor function of CBL.6 CBL mutant cases are associated with monocytosis, LOH detection in myeloid cell lines monocytoid blasts and aberrant KIT expression. Serial studies In order to identify LOH without copy number loss, including UPD, uniparental showed acquisition of CBL mutations during malignant evolution. trisomy (UPT) and uniparental tetrasomy, 30 cell lines with a myeloid CBL mutations were shown to be an independent adverse factor phenotype (Supplementary Table 1) from the SNP Array Based LOH and 6 for overall survival (hazard ratio 2.2 (95%CI; 1.2--4)) (P ¼ 0.013). Copy Number Analysis database from the Wellcome Trust Sanger Institute Patients with CBL mutations were frequently treated with intense (http://www.sanger.ac.uk/) were checked.Weidentifiedthetotalcopynumber chemotherapy or stem cell transplantation,6 suggesting that the and minor allelic copy number in each cell line and identified UPD, UPT aggressive biology of CBL mutant leukemia prompted the and uniparental tetrasomy when the minor allele was absent and the total initiation of more aggressive therapies. copy number was two (UPD), three (UPT) or four (uniparental tetrasomy).

1Department of Translational Hematology and Oncology Research, Taussig Cancer Institute, Cleveland Clinic, Cleveland, OH, USA and 2Department of Pediatrics, Nagoya University Graduate School of Medicine, Nagoya, Japan. Correspondence: Professor JP Maciejewski, Department of Translational Hematology and Oncology Research, Taussig Cancer Institute/R40, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, OH 44195, USA. E-mail: [email protected] Received 19 April 2011; revised 16 December 2011; accepted 9 January 2012; accepted article preview online 13 January 2012; advance online publication, 7 February 2012 Sensitivity of homozygous CBL mutants to tyrosine kinase inhibitors H Makishima et al 1548 CBL sequencing and amplification refractory mutation system Lipofectamine 2000 (Invitrogen). Cells were cultured for additional 48 h (ARMS)-PCR and medium with viral particles were harvested and concentrated. THP-1 DNA was extracted either from fresh bone marrow, peripheral blood, or and GDM-1 cell lines were infected with CBL shRNA (h) lentiviral particles cell lines. To screen DNA for mutations in CBL (exons 8--9), CBLB (exons (Santa Cruz Biotechnology, Paso Robles, CA, USA) in the presence of 9--10) and CBLC (exons 7--8), direct genomic DNA or cDNA sequencing was polybrene (Sigma-Aldrich, St Louis, MO, USA). The stable CBL-expressing performed as previously described.6 and knocking down bulk cultures were selected using puromycin (1 ug/ml For CBL R420Q mutation detection, RNA was extracted from cell lines by for GDM-1 and 4 ug/ml for THP-1). TRIzol (Invitrogen, Carlsbad, CA, USA) and allele-specific RT-PCR was performed. 100 ng of cDNA was amplified in 35-cycle PCR reaction at an Proliferation kinetics annealing temperature of 58 1C. The status of the CBL R420Q mutations Cell proliferation was measured using CellTiter AQueous Non-Radioactive were determined by a DNA tetra-primer ARMS assay. The primer Cell Proliferation Assay (Promega, San Leandro, CA, USA) or the cell viability sequences were: 50-AAGACCATATCAAAGTGACCCAGGAA-30,50-GAAGG analyzer Vi-CELL XR (Beckman Coulter). Briefly, 1000 cells were incubated TCAGGGCTGTCCTTTCTGACA-30,50-CGATGGGTTCAGTACCTTTAATTTCA-30 in 100 ml RPMI/cell for 72 h. 100 ml of (3-(4,5-dimethylthiazol-2-yl)-5-(3- and 50-ATCATCAGCTCGTTCATCATCATCATC-30 (G/G genotype: 323 bp and carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetra-zolium, inner salt; MTS) 216 bp bands; A/A: 323 bp and 163 bp bands). For sequencing purposes, was added to each well, incubated for an additional 1 h at 37 1C and the cDNA samples were amplified using the outer primers. absorbance at 490 nm recorded with a Victor plate reader (PerkinElmer, Waltham, MA, USA). Alternatively, cell lines were washed twice and Flow cytometry/immunohistochemistry resuspended in RPMI 1640 with FCS alone or supplemented with the Bone marrow aspirates were stained using allophycocyanin-labeled murine indicated growth factors or with inhibitors at 2 105 cells/ml. Viable cells anti-c-Kit (CD117, Beckman Coulter, Fullerton, CA, USA), fluorescein were determined by trypan blue exclusion and were counted after 72 h isothiocyanate-labeled murine anti-granulocyte macrophage colony-sti- culture. mulating factor (GM-CSF) receptor (CD116, BD Biosciences, Franklin Lakes, NJ, USA), allophycocyanin-labeled murine anti-Flt3 (CD135, BD Bio- Phospho-specific protein microarray analysis sciences), and anti-CD34 PE-Cy7 (Beckman Coulter). Using side scatter vs For detection of 228 site-specific tyrosine phosphorylation profiles with six FL5 on a Beckman Coulter FC500, gates were set on CD34-positive cells replicates each, 100 ug of cell lysate was labeled with biotin and hybridized and CD117 expression on the cell surface was evaluated. The cell surface on Tyrosine Phosphorylation ProArray (Full Moon BioSystems, Sunnyvale, expression of three RTKs (KIT, GMCSF receptor and FLT3) were also tested CA, USA) using Antibody Array Assay Kit (Full Moon BioSystems). The by flow cytometry. conjugated labeled protein was detected using Cy3-streptavidin. Fluores- cence intensity was scanned by GenPix 4000B (Molecular Devices, RQ-PCR for CBL expression Sunnyvale, CA, USA) using GenePix Pro software (Molecular Devices). For For the measurement of CBL RNA expression, TaqMan PCR was performed the data analysis, background signals and negative control value were (Applied Biosystems, Foster City, CA, USA). Primers and probes were removed from all measurements. The results from the six replicate samples purchased from Applied Biosystems gene expression assays products (CBL are averaged, including actin value. After divided by the averaged actin assay ID: Hs00231981_m1). The probe was labeled at its 50 termini with value in each experiment, the reduction rate of phosphorylation by the FAM. Minor groove binder was attached to the nonfluorescent quencher at treatment of tyrosine kinase inhibitors was computed as follows: Reduction the 30 terminal. Each reaction contained 10 ng of cDNA and TaqMan rate (%) ¼ (Averaged value with treatment -- Averaged value without Universal PCR Master Mix. Real-time PCR and subsequent analysis were treatment) / Averaged value without treatment 100. performed with the ABI Prism 7500 Fast Sequence Detection System using default conditions. RESULTS UPD11q and CBL mutations in myeloid cell lines Cell culture We selected 30 cell lines derived from non-lymphoid leukemias, THP-1 was kindly provided by Professor S P Whitman. U937 and GDM-1 including erythroid, megakaryocytic, monocytic leukemias from a were purchased from ATCC (Manassas, VA, USA). NKM-1 cells were kindly collection of 127 hematopoietic and lymphoid cell lines in the provided by Dr Akihiro Abe. MOLM13 was purchased from DSMZ Wellcome Trust Sanger Institute database (http://www.sanger. (Braunschweig, Germany). Cells were cultured in RPMI-1640 (Invitrogen) ac.uk/cgi-bin/genetics/CGP/cghviewer/CghViewer.cgi; Supplementary supplemented with 10% FBS at 37 1C and 5% CO and when appropriate in 2 Table 1). Using single-nucleotide polymorphism (SNP) array data, the presence of human stem cell factor (SCF), recombinant human we detected 150 regions of copy number neutral LOH (Supple- thrombopoietin (TPO), Flt3 ligand and macrophage colony-stimulating mentary Figure 1); all chromosomes were affected, with chromo- factor (M-CSF) (PeproTech, Rocky Hill, NJ, USA). The cells were also cultured some 9 most frequently affected (22/30; 73% p-arm; 33%, q-arm; in the presence of granulocyte colony-stimulating factor (G-CSF), GM-CSF 40%)). We identified three instances of LOH without copy number and erythropoietin (Amgen, Thousand Oaks, CA, USA), dasatinib (Bristol- loss located on chromosome 11q (2 UPD and 1 UPT), including Myers Squibb, New York, NY, USA) and imatinib, sunitinib, the SRC family 11q13.1--q23.3 (UPD) in ML-2, 11q13.4-qter (UPD) in NKM-1 and inhibitor PP-2, rapamycin, Ly294002 and U0126 (LC Laboratories, Woburn, 11q12.3-qter (UPT) in GDM-1 (Figure 1a), of which 2 (NKM-1 and MA, USA). GDM-1) spanned the CBL locus (ch11:118,582,200-118,684,069) In certain circumstances hematopoietic colony cultures were performed. (Figure 1a). When we sequenced CBL in these cell lines, we Bone marrow cells were plated in methylcellulose (Stem Cell Technologies, identified a R420Q missense mutation in GDM-1 (Figures 1b--d). Vancouver, Canada) in the presence of 20 ng/ml GM-CSF and 2 U/ml This recurrent mutation has been reported to affect ubiquitination erythropoietin. Total BM cells were plated at a density of 1 105 cells/1 ml activity and to be associated with cytokine-independent growth.11 methylcellulose in 35-mm dishes. Colonies on duplicate plates were ARMS-PCR was designed for the R420Q mutation; as expected, we counted and the average number of CFU-GM was then calculated. detected only mutated allele, whereas the wild-type (WT) allele was absent (Figures 1b--d). Lentiviral transduction Sequencing of the NKM-1 and ML-2 cell lines did not reveal the CBL cDNA (accession number NM_005188) was cloned into pLVX-dsRED- presence of CBL mutations. As reported previously, a CBL gene Monomer-C1 expression vector (Clontech, Mountain View, CA, USA). splice mutation (deletion of exon 8-intron 8 boundary) was found Viral particles were prepared as follows: 293 cells were transfected in MOLM-13.12 By RT-PCR-based sequencing, CBL WT expression with a mixture of pCMV-dR8.2/pCMV-VSVG/pLVX-dsRED-Monomer-C1 and was observed as well as the mutant, with a shorter transcript.

Leukemia (2012) 1547 --1554 & 2012 Macmillan Publishers Limited Sensitivity of homozygous CBL mutants to tyrosine kinase inhibitors H Makishima et al 1549 CBL (11q23.3)

ML-2 NKM-1 GDM-1

CBL mutation RING finger domain WT R420Q WT R420Q

323

216

163

CBL 4H EF SH2 LZ/ L RF PPP UBA Y Y S S S S S Y Y Y Y 0 906 368 371 619 623 639 642 667674 700 731 774 Chromosome 11q23.3 384 404 420 No mutation in 118,582,200-118,684,069 JAK2, FLT3, HRAS, KIT, KRAS, MYC, NF1, NRAS, WT1 RING finger (RF) domain Figure 1. UPD 11q lesions in myeloid cell lines and detection of CBL mutation in GDM-1 cell line. (a) The distribution of UPD on chromosome 11 among myeloid cell lines. In two out of three cell lines, CBL was located within the region of UPD11q; these cell lines had a monocytic phenotype. (b) DNA sequencing revealed a CBL R420Q homozygous mutation in GDM1. (c) Allele-specific RT-PCR confirmed the expression of CBL R420Q (163bp) and lack of WT CBL mRNA (216bp) in GDM-1. (d) CBL R420Q mutation is located in the RING finger (RF) domain (black arrow head), a highly conserved region essential for ubiquitination and degradation of RTK and SFK. The major domains of CBL are depicted: four-helix bundle (4H), EF-hand-calcium-binding (EF), SRC homology 2 (SH2), linker sequence (L), RING finger (RF), proline-rich (PPP) and leucine zipper (LZ)/ubiquitin-associated (UBA). Tyrosine (Y) and serine (S) residues, represented by red circles, are phosphorylated by tyrosine kinases. The lower right panel (blue) lists other oncogenes and proto-oncogenes sequenced; no mutations were found in any of the genes in GDM-1.

When we tested 493 samples from primary myeloid malignancies, the most effective in augmenting proliferation of GDM-1 (2.7-fold) we found homozygous R420 mutations in 5 cases of monocytoid followed by TPO (1.6-fold) as compared with controls (Supple- AML (29% of all CBL mutant cases), including three missense and mentary Figure 2C). Consistent with this finding, we also detected two frame shift mutations (Supplementary Table 2). All mutant high expression of the GM-CSF receptor on the surface of GDM-1 cases showed a strong surface expression of KIT, suggesting (Supplementary Figure 2D). When induction of WT CBL in GDM-1 defective receptor disposal due to a lack of receptor ubiquitina- on the response to growth factors was examined, the hyperpro- tion. In agreement with the findings in primary cells, CBL mutant liferative response to GM-CSF was decreased as compared with a GDM-1 (derived from a patient with secondary AML of a control vector infection (Figure 2d). A similar effect was observed monocytic phenotype) also showed high KIT expression (Supple- in the presence of TPO (Figure 2e). mentary Figure 2A, B). Effect of dasatinib treatment on GDM-1 Impact of overexpression of WT CBL on GDM-1 We have compared the differential effects of various RTK and SFK We used GDM-1 as a model to study agents with potential activity inhibitors, including sunitinib, dasatinib, imatinib and PP2, on the in CBL mutant leukemias. THP-1, a monocytic cell line without a proliferation of GDM-1, 2 control cell lines (WT CBL) and CBL mutation, was used as a WT control. After transduction of WT heterozygous CBL mutant MOLM-13 cells. We also used several CBL in GDM-1 and THP-1, ectopic expression of CBL cDNA was 20 inhibitors operating downstream of RTK and SFK, including times higher than control vector experiments in both cell lines rapamycin, Ly294002 and U0126 for mTOR, PI3K and MAPK, (Figure 2a). By ARMS-PCR we detected only the WT allele in THP-1, respectively. The selectivity of inhibitory effects was calculated as and both the WT and R420Q mutant alleles in GDM-1 (Figure 2b). a ratio between GDM-1/the other cell lines. GDM-1 showed Overexpression of WT CBL suppressed the growth of CBL mutant increased sensitivity to the SFK inhibitors dasatinib (LD50 50 nM vs GDM-1, whereas it had no effect on THP-1 (Figure 2c). Although 41000 nM) as compared with the other cell lines (Figure 3a). ShRNA knockdown suppressed expression of WT and R420Q Sunitinib, rapamycin, LY294002 and U0126 demonstrated almost mutant CBL to o60% in GDM-1 and THP-1, respectively (Figures equal responses between GDM-1 and the other cell lines and 2a and b), no significant impact on growth of either the mutant or blocked growth of all cell lines irrespective of the CBL function WT cell lines was found (Figure 2c). (Figure 3a). No differential sensitivity was found in MOLM-13 cells heterozygous for CBL mutation. Interaction of growth factors and CBL homozygous mutations To evaluate RTK hypersensitivity in homozygous CBL mutant cells, TK inhibitors and tyrosine phosphorylation levels we studied the effect of growth factors on GDM-1. We cultured Using a phosphoarray, we analyzed the tyrosine phosphorylation cells with and without 10 ng/ml of SCF, TPO, FLT3L, G-CSF, M-CSF profiles of GDM-1 with or without TKI inhibitors and assessed and GM-CSF for 3 days (Supplementary Figure 2C). GM-CSF was which kinases were inhibited in association with the suppressive

& 2012 Macmillan Publishers Limited Leukemia (2012) 1547 --1554 Sensitivity of homozygous CBL mutants to tyrosine kinase inhibitors H Makishima et al 1550 100 CBL

323 10 216 163 (Log) 1

Relative expression of 0.1 Marker shRNA CBL shRNA shRNA Control shRNA CBL CBL expression vector expression vector Control shRNA Control shRNA dsRed control vector dsRed control vector CBL CBL expression vector expression vector

dsRed control vector dsRed control vector THP-1 GDM-1 CBL CBL THP-1 GDM-1

1.2 1.0 0.8 0.6 0.4 0.2 Relative cell viability 0.0 shRNA shRNA CBL CBL Control shRNA Control shRNA expression vector expression vector dsRed control vector dsRed control vector CBL CBL THP-1 GDM-1

100 80

80 60

60 40 40 20 Cell number (X10E4)

Cell number (X10E4) 20

0 0 GM-CSF (10ng/ml) - + - + TPO (10ng/ml) - + - + Control vector + + - - Control vector + + - - CBL expression vector - - ++CBL expression vector - - + + Figure 2. Over expression and knockdown of CBL and growth factor sensitivity in GDM-1 with over expression of WT CBL. (a) CBL was overexpressed (expression vector) and knocked down (shRNA) in THP-1 and GDM-1 via lentiviral transduction. mRNA copy number of CBL and GAPDH was measured by real-time quantitative PCR in triplicate. Relative CBL expression was calculated as CBL copy number divided by GAPDH copy number. In each control experiment using an empty plasmid, the value of expression was defined as 1. The data represent mean±s.e.m. (b) Allelic status of residue 420 of CBL was confirmed by allele-specific PCR. Amplification bands of 216bp and 163bp indicate WT and mutated allele, respectively. (c) Cell viability was evaluated by MTS assay in over expression and knockdown experiments in the cell lines with WT (THP-1) and mutated CBL (GDM-1). In each control experiment using an empty plasmid, the value of expression was defined as 1. The data represent means±s.e.m. *Indicates a statistically significant difference (Po0.05). GDM-1 was cultured with GM-CSF (d)or TPO (e) for 72 h and viable cells were counted by trypan blue exclusion with overexpression of CBL or with empty vector transduction as control. Assays were performed in triplicate and the data represent means±s.e.m.

effects of dasatinib in culture. Without inhibitors, 168 out of 228 phosphorylation of 44/288 tyrosine sites. As an example, ICAM1 sites were positive for phosphorylation and the highest value was Tyr512, IGF1R Tyr1161 and STAT3 Tyr705 phosphorylation levels observed for KIT Tyr721 (data not shown). In FLT3, two tyrosine were diminished (Figure 4a). Phosphorylation levels were sites (Tyr 969 and 842), and in SRC, three tyrosines (Tyr 216, 418 decreased at 56 sites in sunitinib-treated GDM-1 (Figure 4b). and 529) were phosphorylated. Dasatinib (10 nM) reduced the A comparison of the inhibitory spectra of dasatinib and sunitinib

Leukemia (2012) 1547 --1554 & 2012 Macmillan Publishers Limited Sensitivity of homozygous CBL mutants to tyrosine kinase inhibitors H Makishima et al 1551

1.0 1.0 1.0 0.8 0.8 0.8 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 Dasatinib Imatinib Sunitinib 0 0 0 -4 -3 01010 0.01 0.1 0 0.01 0.1 1 10 0 1.25 2.5 5 12.5

1.0 1.0 1.0 0.8 0.8 0.8 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 PP2 Rapamycin Ly294002 0 0 0 0 1 5 10 20 0 0.1 1 10 1000 0.1 1 10 100

1.0 GDM-1 0.8 THP1 Viability of monocytic cell lines U937 0.6 (relative to untreated control) MOLM13 0.4 0.2 Concentration U0126 (uM) 0 0 1 5 10 20

0.8

0.6

0.4

0.2

Relative number of CFU-GM 0 Dasatinib −+ −+ −+ −+

CBL mutation CBL wild type Figure 3. Tyrosine kinase inhibitor sensitivity in monocytic leukemia cell lines and dasatinib effect on bone marrow samples. (a) Four monocytic cell lines were cultured for 72 h with seven inhibitors of proteins involved in receptor tyrosine kinase and SRC family kinase pathways. Viability was evaluated by MTS assay in triplicate, and calculated as divided by the result in untreated control. (b) Bone marrow mononuclear cells from an AML patient with the CBL R420Q mutation and control patients without CBL mutation were cultured in 10% FBS RPMI for 24 h with and without dasatinib (10 nM). CFU-GM was counted after 10-day-semisolid culture in duplicate plates. Relative colony numbers were calculated by division with those in the experiment without dasatinib treatment.

allowed us to discern phosphorylation targets uniquely associated In contrast, the reduction effect of colony counts was less with inhibitory effects of dasatinib in GDM-1 (Figure 3a). We pronounced (28--71%) in the patients with WT CBL (Figure 3b). detected the phosphorylation of 26 proteins (29 tyrosine sites), which are reduced specifically by dasatinib treatment, including not only RTK (KIT, FGFR1 and M-CSFR) but also other phosphory- DISCUSSION lated proteins (SRC, STAT1, STAT3 and P73) (Figure 4c). Some of Based on the observation that most of the mutations in CBL found these proteins are associated with the downstream cascades of in patients were homozygous, we identified cell lines with UPD11q activated KIT and GM-CSF receptor. and sequenced CBL in a targeted fashion. As a result, we found a CBL RFD R420Q homozygous mutation with LOH with trisomy 11q and concurrent UPT in the GDM-1 cell line derived from a patient Growth suppression by dasatinib in a patient sample with a with secondary AML. Because R420 was one of the most homozygous CBL RFD mutation commonly affected residues in patients,3 --6,11 GDM-1 cells are an To confirm the effects of dasatinib in primary cells, we also appropriate model of myeloid malignancies with homozygous CBL performed semisolid cultures of bone marrow derived from an mutation, whereas MOLM-13 has retained a copy of WT CBL, AML patient with R420Q mutation with and without dasatinib which may be protective and the interactions with other specific (10 nM) and compared the effects to cultures of control bone mutations and a single mutated CBL allele may make this cell line marrow from three AML patients without CBL mutation. We unique and not representative. Conversely, one would expect that observed a 95% reduction of colony counts in the patient sample. MOLM-13 would show similar behavior as GDM-1 transduced with

& 2012 Macmillan Publishers Limited Leukemia (2012) 1547 --1554 Sensitivity of homozygous CBL mutants to tyrosine kinase inhibitors H Makishima et al 1552 a 100 90 80 70 60 50 40 30

Reduction rate (%) 20 10 0 p73 (Tyr99) KIT (Tyr936) KIT Src (Tyr529) AurB (Tyr12) FAK (Tyr407)FAK LCK (Tyr504) LCK Pyk2 (Tyr402) Pyk2 (Tyr579) Met (Tyr1234)Met c-Abl (Tyr245) Dok1 (Tyr398) Dok1 (Tyr362) (Tyr362) Dok1 Dab1 (Tyr220) ALK (Tyr1507) c-Jun (Tyr170) HER2 (Tyr877) (Tyr877) HER2 WASP (Tyr290)WASP STAT6 (Tyr641)STAT6 STAT3 (Tyr705)STAT3 (Tyr701)STAT1 FGFR1 (Tyr154) (Tyr154) FGFR1 hospho-Tyr721) hospho-Tyr721) Paxillin (Tyr118) RAD52 (Tyr104) EGFR (Tyr1069) EGFR EGFR (Tyr1172) EGFR STAM2 (Tyr192) ortactin (Tyr421) EGFR (Tyr1110) EGFR ZAP-70 (Tyr292) ICAM-1 (Tyr512) lta/PKCD (Tyr64) 8 MAPK (Tyr182) in beta3 (Tyr773) (Tyr773) in beta3 2 (Tyr1221/1222) (a/b/g) (Tyr55/53) KC delta (Tyr313) delta KC IGF-1R (Tyr1161) IGF-1R M-CSFR (Tyr809) M-CSFR P r Catalase (Tyr385) /PKC mu (Tyr463) C e 3 n R P p HE KIT ( KIT Integ CD22/BL-CAM (Tyr807) PKD1 PKC d M-CSFR (Phospho-Tyr561)M-CSFR ACTIN Pa PDGF-R beta PDGF-R (Phospho-Tyr740)

b 100 90 80 70 60 50 40 30

Reduction rate (%) 20 10 0 ) ) ) ) ) ) ) ) r15 y r353 r349 r171 r753 r307 r580 y y y y y T y ( T T T T T T ( ( ( ( ( ( r749/681 y k2 T Bcr (Tyr177) Bcr y Shc LAT LAT ( BTK (Tyr222)BTK FAK (Tyr397) FAK ETK (Tyr566)ETK AXL (Tyr691) FAK (Tyr861)FAK LYN (Tyr507) LYN FER (Tyr402) FER CD5 (Tyr453) BLNK (Tyr96) BRK (Tyr447) BRK BLNK (Tyr84) Abl1 (Tyr412) Abl1 (Tyr204) Abl1 CrkII (Tyr221) CrkII CDK5 CDK5 Met (Tyr1356) Met (Tyr1234) Met Met (Tyr1003) Met P Ezrin FLT3 (Tyr969) Trk A Trk (Tyr701) B Trk (Tyr705) TFII-I (Tyr248) Trk A Trk (Tyr791) Vav2 (Tyr142) JAK2 (Tyr221) c-Jun (Tyr170) Gab2 (Tyr643) Gab2 Dab1 (Tyr232) FRS2 (Tyr436) ACK1 (Tyr284) CD19 (Tyr531) HER2 (Tyr877) HER2 K-beta (Tyr188) K-beta MAPK (Tyr322) MAPK nuclein (Tyr125) nuclein (Tyr136) nuclein AR2B (Tyr1472) AR2B GFR2 (Tyr1214) GFR2 -R beta (Tyr751) beta -R KC delta (Tyr52) delta KC EGFR (Tyr1197) EGFR KY KY PLCG2 adherin (Tyr731) adherin 5-alpha (Tyr607) 5-alpha kB-alpha (Tyr42) C alpha (Tyr657) 9 (Tyr153) spase PP2A-a /CD127 (Tyr449) /CD127 B-alpha (Tyr305) B-alpha 8 y y K I F E I D P S k 8 a C K R I V p3 P C NM IL7 VE- PDG MER/ PI3K p alpha-S alpha-S

c Dasatinib Sunitinib

Abl1 ACK1 ACTIN ALK alpha-Synuclein AXL AurB c-Abl Catalase FAK Bcr BLNK BRK BTK CD22/BL-CAM HER2 Caspase 9 CD19 CD5 KIT Cortactin Pyk2 CDK5 CrkII ETK Ezrin Dok1 FGFR1 ICAM-1 c-Jun FER FLT3 FRS2 Gab2 IkB-alpha IKK-beta IGF-1R PKC-delta IL7R/CD127 JAK2 LAT Integrin beta3 Dab1 LYN MER/SKY NMDAR2B LCK MCSFR PDGF-R beta PI3-kinase p85-alpha p73 Paxillin P38 MAPK PKC alpha PLCG2 PKD1/PKCmu EGFR EGFR PP2A-a Shc TFII-I RAD52 SRC STAM2 Met STAT1 STAT3 STAT6 Trk A Trk B Vav2 WASP ZAP-70 VE-Cadherin VEGFR2 BRK

Figure 4. Reduction effect of tyrosine kinase inhibitors on protein phosphorylation in GDM-1. Tyrosine phosphorylation in GDM-1 with treatment of dasatinib (a) or sunitinib (b) was comprehensively investigated using an antibody microarray system. The results from the six replicate samples were averaged, including actin value. After division by averaged actin value in each experiment, the reduced rate of phosphorylation by the treatment of tyrosine kinase inhibitors was computed as follows: Reduction rate (%) ¼ (Averaged value with treatment -- Averaged value without treatment) / Averaged value without treatment 100. Dasatinib (10 nM) reduced the phosphorylation of 44/288 tyrosine sites. As an example, ICAM1 Tyr512, IGF1R Tyr1161 and STAT3 Tyr705 phosphorylation levels were diminished (a). Phosphorylation levels were decreased at 56 sites in sunitinib-treated GDM-1 (b). (c) Comparison of inhibitory spectrum of dasatinib and sunitinib resulted in the identiﬁcation of phophorylation targets uniquely associated with inhibitory effects of dasatinib in GDM-1. Phosphorylation of 26 proteins (29 tyrosine sites) was reduced speciﬁcally by dasatinib treatment.

Leukemia (2012) 1547 --1554 & 2012 Macmillan Publishers Limited Sensitivity of homozygous CBL mutants to tyrosine kinase inhibitors H Makishima et al 1553 aWTCBL. Our results demonstrate that the proliferation of GDM-1 In addition to autophosphorylation, activation of various RTKs depends on the absence of WT CBL, consistent with the results in phosphorylation of multiple proteins in their down- homozygous nature of mutations in this cell line. GDM-1 also stream cascades. CBL is critically associated with reducing showed hypersensitivity to growth factors, which could be phosphorylation of RTK or SFK through ubiquitination and diminished by ectopic expression of WT CBL. We have found that degradation.27 --30 Consequently, our hypothesis was simple: loss the most effective TKI in GDM-1 was dasatinib, which specifically of CBL function should result in more phosphorylation. Indeed, we reduced the phosphorylation levels of KIT, SRC, STAT1 and STAT3 observed tyrosine phosphorylation in 73% of target proteins we in this highly phosphorylated CBL mutant cell line. Dasatinib also examined in GDM-1 without any additional growth factor suppressed proliferation of bone marrow cells from an AML stimulation or TKI, the highest level of phosphorylation occurring patient with homozygous CBL RFD mutation more than that of in KIT. In general, mutations of RTK genes, including KIT and FLT3, control bone marrow with WT CBL. have been found in hematopoietic malignancies, producing a Recently, multiple groups found that CBL homozygous muta- permanent signal for cell proliferation by activation of the kinase tions are associated with UPD11q in the cases of juvenile domain in the absence of ligand. However, KIT and FLT3 are not myelomonocytic leukemia in which no mutations of the RAS mutated in GDM-1 but are highly phosphorylated, possibly based family, PTPN11 or NF1 were found.13 --16 One of the diagnostic on inappropriate activation by loss of a negative regulator (CBL criteria in juvenile myelomonocytic leukemia is GM-CSF hyper- RFD). Also, LCK, STAT1 and STAT3 positioned downstream of KIT sensitivity of malignant clone. In analogy to juvenile myelomono- were phosphorylated in GDM-1, suggesting that this cell line is a cytic leukemia, among various kinds of growth factors tested, strong model for studying the outcomes of KIT overactivation. proliferation of GDM-1 was highest with GM-CSF treatment; not Similarly, previously reported in vitro models showed that surprisingly, the GM-CSF receptor is expressed on the cell surface concomitant expression of CBL mutations and KIT induced of GDM-1. This hypersensitivity was suppressed by overexpression cytokine-independent proliferation, survival and clonogenic of WT CBL. Dasatinib specifically reduced phosphorylation of SRC, growth.20,21 STAT3 and STAM2, associated with signal transduction down- As CBL mutations are related to disease phenotypes with poor stream of GM-CSF receptor activation. These findings suggest that prognosis, new therapeutic approaches are required.4,6 RTK and CBL homozygous mutations might result in hypersensitivity of SFK inhibitors could suppress growth potential in CBL mutant mutant clone to GM-CSF, as there might be less ubiquitination or disease models. Unlike cell lines with WT CBL, GDM-1 showed an degradation of its receptor. increased sensitivity to a RTK and SFK inhibitor, dasatinib. A low A relatively high frequency of homozygosity of CBL RFD concentration of dasatinib was effective in suppressing the mutations encountered in CMML and CMML-derived AML also proliferation of GDM-1, but at an equivalent dose dasatinib did implies that the WT allele may protect against malignant not suppress cell lines with WT CBL. Inhibitors of mTOR, PI3K or evolution. Here, we showed that the WT allele reduces the MAPK demonstrated almost equal responses between CBL mutant proliferation and growth factor hypersensitivity in GDM-1 cells. In and WT cell lines. Based on the inhibition spectrum, our results animal models, WT CBL knockout mice show only mild myelo- indicate that activating SRC family pathways downstream of RTK proliferative potential.17 However, WT CBL knockout with mutant are most likely targets involved in the pathogenic process CBL knock-in developed a more aggressive myeloproliferative mediated by CBL mutations, whereas downregulation of other disease, which progresses to leukemia and involves hematopoietic signaling cascades including the mTOR, PI3K and MAPK pathways progenitors that exhibit augmented FLT3 signaling.5 Suppressing may not be sufficient for the reduction of pro-oncogenic effect of this pathway through crossbreeding with FLT3 ligand knockout mutant CBL. In an animal model with CBL mutation, one member mice prevents leukemia development.17 These and other findings of the SFK, Fyn, is expressed and binds to CBL and is inhibited by suggest that FLT3 may be one of the most significant targets of SRC inhibitors.21 In our experiments, phosphorylation of SRC and ubiquitination by CBL.4,11,18 --22 In addition to the GM-CSF receptor, LCK were also inhibited specifically by dasatinib therapy. In the we observed activation of FLT3 in GDM-1 cells; however, future, the same results should be reproduced in vivo, before such enhanced phosphorylation of other RTK and SFK was also therapeutic strategies are to human disorders. detected that was subsequently reduced by TKI. These results In this paper, we focused our study on the specific mutation of suggest that other RTKs, except for FLT3 and SFK, may also be very CBL R420Q, because R420 is functionally very important site for important for leukemia evolution. In our primary and cell line ubiquitination and degradation of RTK and most frequently models, the downstream pathway of receptors of GM-CSF and affected site functionally well known for loss of function.11,21 TPO contribute more than FLT3 signaling. Interestingly, in a CBL Although we examined the MOLM-13 cell line with a splice and CBLB double knockout mouse model, STAT5 activation was mutation, the spectrum of TKIs’ effectiveness was not the same as observed after GM-CSF, SCF and TPO stimulation.23 Of note is that that in a homozygous CBL R420Q mutation, perhaps because by in a recent report, a mutation of CSF1R was described in GDM-1.24 RT-PCR, both WT and CBL mRNA splice variants were detected in In this study, GDM-1 showed sensitivity to imatinib but in our MOLM-13. Functionally, the presence of WT CBL is likely similar to report dasatinib showed more suppressive effect than imatinib. In overexpression of the WT protein in a CBL homozygous mutation fact, dasatinib showed the most effective inhibition of this CBL background (GDM-1). Because WT CBL functions against TK mutant cell line among all drugs we screened, and reduced the activation and proliferation in GDM-1, it is reasonable that the phosphorylation of specific RTK and non RTK proteins, which may effect of dasatinib on MOLM-13 was not observed. However, the result in an increased proliferative potential in myeloid malig- effect of TKIs on other CBL mutations remains unclear and further nancies with CBL homozygous mutations. investigation is required. Consequently, it appears that CBL family genes behave as tumor The GDM-1 cell line is known to respond to imatinib and is suppressors, because a loss of function results in the reduction of positive for the CSF1R mutation.24 Chase et al.24 showed that ubiquitination and degradation of activated RTKs. However, CSF1R in GDM-1 is very strongly phosphorylated and that CSF1 mutated CBL also appears to act as an oncogene that induces neutralization only partially inhibited its proliferation. This result malignant evolution with a specific phenotype that includes suggests that not only the effect of the driver CSF1R mutation, but monocytosis and growth factor hypersensitivity. In agreement also other mechanisms of proliferation, for example, through with this theory, in mutant models CBL seems to act with dual activation of various TK might be involved as demonstrated by our properties: as a tumor suppressor gene and an oncogene, as the results. Moreover, we showed that dasatinib almost completely phenotype of knock-in models in CBL (/) mice showed a more inhibited GDM-1 proliferation with very low concentration. Of pervasive leukemic phenotype than the simple knockout.5,25,26 course, CSF1R is one of the main targets of dasatinib and the

& 2012 Macmillan Publishers Limited Leukemia (2012) 1547 --1554 Sensitivity of homozygous CBL mutants to tyrosine kinase inhibitors H Makishima et al 1554 CSF1R mutation is involved in proliferation of GDM-1 cell line. It is 11 Sargin B, Choudhary C, Crosetto N, Schmidt MH, Grundler R, Rensinghoff M et al. also clear that dasatinib is more potent than imatinib, because Flt3-dependent transformation by inactivating c-Cbl mutations in AML. Blood dasatinib can affect many other RTK (as well as CSF1R) and SFK as 2007; 110: 1004 --1012. we showed by experiments using several growth factors and 12 Caligiuri MA, Briesewitz R, Yu J, Wang L, Wei M, Arnoczky KJ et al. Novel c-CBL and phosphoprotein microarray. CBL-b ubiquitin ligase mutations in human acute myeloid leukemia. Blood 2007; In conclusion, homozygous CBL mutations result in hypersensi- 110: 1022 --1024. tivity to growth factors. Over expression of the WT CBL inhibited 13 Loh ML, Sakai DS, Flotho C, Kang M, Fliegauf M, Archambeault S et al. Mutations in the growth of a CBL mutant cell line, consistent with the CBL occur frequently in juvenile myelomonocytic leukemia. Blood 2009; 114: 1859 --1863. homozygous nature of CBL mutations identified in most patients 14 Muramatsu H, Makishima H, Jankowska AM, Cazzolli H, O’Keefe C, Yoshida N et al. with myeloid malignancies who harbor CBL mutations. Dasatinib is Mutations of an E3 ubiquitin ligase c-Cbl but not TET2 mutations are pathogenic the most effective TKI in a mutant CBL background and specifically in juvenile myelomonocytic leukemia. Blood 2010; 115: 1969 --1975. reduced the phosphorylation of RTK and SFK. Based on these 15 Shiba N, Kato M, Park MJ, Sanada M, Ito E, Fukushima K et al. CBL mutations in findings, further investigations in vivo are required for clinical juvenile myelomonocytic leukemia and pediatric myelodysplastic syndrome. decision making in cases with CBL mutation and UPD11q. Leukemia 2010; 24: 1090 --1092. 16 Matsuda K, Taira C, Sakashita K, Saito S, Tanaka-Yanagisawa M, Yanagisawa R et al. Long-term survival after nonintensive chemotherapy in some juvenile myelomo- CONFLICT OF INTEREST nocytic leukemia patients with CBL mutations, and the possible presence of healthy persons with the mutations. Blood 2010; 115: 5429 --5431. The authors declare no conflict of interest. 17 Rathinam C, Thien CB, Flavell RA, Langdon WY. Myeloid leukemia development in c-Cbl RING finger mutant mice is dependent on FLT3 signaling. Cancer Cell 2010; ACKNOWLEDGEMENTS 18: 341 --352. 18 Lavagna-Sevenier C, Marchetto S, Birnbaum D, Rosnet O. FLT3 signaling in This work was supported in part by RO1HL-082983, U54 RR019391 (JPM and MAS), hematopoietic cells involves CBL, SHC and an unknown P115 as prominent K24 HL-077522, DOD-MP048018 (MAM), and by a grant from AA&MDS International tyrosine-phosphorylated substrates. Leukemia 1998; 12: 301 --310. Foundation and Robert Duggan Charitable Fund (JPM). The THP-1 cell line was kindly 19 Hartman AD, Wilson-Weekes A, Suvannasankha A, Burgess GS, Phillips CA, Hincher provided by Professor S P Whitman, the Ohio State University. NKM-1 cells were KJ et al. Constitutive c-jun N-terminal kinase activity in acute myeloid leukemia kindly provided by Dr Akihiro Abe of the Department of Hematology, Nagoya derives from Flt3 and affects survival and proliferation. Exp Hematol 2006; 34: University, School of Medicine. Granulocyte colony-stimulating factor, granulocyte 1360 --1376. macrophage colony-stimulating factor and erythropoietin were obtained from 20 Fernandes MS, Reddy MM, Croteau NJ, Walz C, Weisbach H, Podar K et al. Novel Amgen. Dasatinib was obtained from Bristol-Myers Squibb. oncogenic mutations of CBL in human acute myeloid leukemia that activate growth and survival pathways depend on increased metabolism. J Biol Chem 2010; 285: 32596 --32605. AUTHOR CONTRIBUTIONS 21 Bandi SR, Brandts C, Rensinghoff M, Grundler R, Tickenbrock L, Kohler G et al. E3 H Makishima designed the study, collected patient data, performed experiments ligase-defective Cbl mutants lead to a generalized mastocytosis and myelopro- and wrote the manuscript; Y Sugimoto, HS, MJC and KPN performed experiments; liferative disease. Blood 2009; 114: 4197 --4208. H Muramatsu collected patient data, CO and Y Saunthararajah edited the manuscript; 22 Reindl C, Quentmeier H, Petropoulos K, Greif PA, Benthaus T, Argiropoulos B et al. JPM designed the study, collected patient data and wrote the manuscript. CBL exon 8/9 mutants activate the FLT3 pathway and cluster in core binding factor/11q deletion acute myeloid leukemia/myelodysplastic syndrome subtypes. Clin Cancer Res 2009; 15: 2238 --2247. REFERENCES 23 Naramura M, Nandwani N, Gu H, Band V, Band H. Rapidly fatal myeloproliferative 1 Maciejewski JP, Mufti GJ. Whole genome scanning as a cytogenetic tool in disorders in mice with deletion of Casitas B-cell lymphoma (Cbl) and Cbl-b in hematologic malignancies. Blood 2008; 112: 965 --974. hematopoietic stem cells. Proc Natl Acad Sci USA 2010; 107: 16274 --16279. 2 O’Keefe C, McDevitt MA, Maciejewski JP. Copy neutral loss of heterozygosity: a 24 Chase A, Schultheis B, Kreil S, Baxter J, Hidalgo-Curtis C, Jones A et al. novel chromosomal lesion in myeloid malignancies. Blood 2010; 115: 2731 --2739. Imatinib sensitivity as a consequence of a CSF1R-Y571D mutation and 3 Dunbar AJ, Gondek LP, O’Keefe CL, Makishima H, Rataul MS, Szpurka H et al. 250K CSF1/CSF1R signaling abnormalities in the cell line GDM1. Leukemia 2009; 23: single nucleotide polymorphism array karyotyping identifies acquired uniparental 358 --364. disomy and homozygous mutations, including novel missense substitutions of 25 Ogawa S, Shih LY, Suzuki T, Otsu M, Nakauchi H, Koeffler HP et al. Deregulated c-Cbl, in myeloid malignancies. Cancer Res 2008; 68: 10349 --10357. intracellular signaling by mutated c-CBL in myeloid neoplasms. Clin Cancer Res 4 Grand FH, Hidalgo-Curtis CE, Ernst T, Zoi K, Zoi C, McGuire C et al. Frequent CBL 2010; 16: 3825 --3831. mutations associated with 11q acquired uniparental disomy in myeloproliferative 26 Ogawa S, Sanada M, Shih LY, Suzuki T, Otsu M, Nakauchi H et al. Gain-of-function - neoplasms. Blood 2009; 113: 6182 --6192. CBL c-CBL mutations associated with uniparental disomy of 11q in myeloid 5 Sanada M, Suzuki T, Shih LY, Otsu M, Kato M, Yamazaki S et al. Gain-of-function neoplasms. Cell Cycle 2010; 9: 1051 --1056. of mutated C-CBL tumour suppressor in myeloid neoplasms. Nature 2009; 460: 27 Donovan JA, Wange RL, Langdon WY, Samelson LE. The protein product of 904 --908. the c-cbl protooncogene is the 120-kDa tyrosine-phosphorylated protein in 6 Makishima H, Cazzolli H, Szpurka H, Dunbar A, Tiu R, Huh J et al. Mutations of e3 Jurkat cells activated via the T cell antigen receptor. J Biol Chem 1994; 269: ubiquitin ligase cbl family members constitute a novel common pathogenic 22921 --22924. lesion in myeloid malignancies. J Clin Oncol 2009; 27: 6109 --6116. 28 Andoniou CE, Thien CB, Langdon WY. Tumour induction by activated abl involves 7 Thien CB, Langdon WY. c-Cbl and Cbl-b ubiquitin ligases: substrate diversity and tyrosine phosphorylation of the product of the cbl oncogene. Embo J 1994; 13: the negative regulation of signalling responses. Biochem J 2005; 391: 153 --166. 4515 --4523. 8 Thien CB, Langdon WY. Cbl: many adaptations to regulate protein tyrosine 29 Marcilla A, Rivero-Lezcano OM, Agarwal A, Robbins KC. Identification of the major kinases. Nat Rev Mol Cell Biol 2001; 2: 294 --307. tyrosine kinase substrate in signaling complexes formed after engagement of Fc 9 Ryan PE, Davies GC, Nau MM, Lipkowitz S. Regulating the regulator: negative gamma receptors. J Biol Chem 1995; 270: 9115 --9120. regulation of Cbl ubiquitin ligases. Trends Biochem Sci 2006; 31:79--88. 30 Odai H, Sasaki K, Iwamatsu A, Hanazono Y, Tanaka T, Mitani K et al. The proto- 10 Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H et al. WHO oncogene product c-Cbl becomes tyrosine phosphorylated by stimulation with Classification of Tumors of Haematopoietic and Lymphoid Tissues, Fourth Edition. GM-CSF or Epo and constitutively binds to the SH3 domain of Grb2/Ash in human IARC Press: Lyon, 2008. hematopoietic cells. J Biol Chem 1995; 270: 10800 --10805.

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

Leukemia (2012) 1547 --1554 & 2012 Macmillan Publishers Limited