FBXW7 Regulates Glucocorticoid Response in T-Cell Acute Lymphoblastic Leukaemia by Targeting the Glucocorticoid Receptor for Degradation

ORIGINAL ARTICLE FBXW7 regulates glucocorticoid response in T-cell acute lymphoblastic leukaemia by targeting the glucocorticoid receptor for degradation

A Malyukova1, S Brown1, R Papa1, R O’Brien1, J Giles1, TN Trahair2, L Dalla Pozza3, R Sutton1, T Liu1, M Haber1, MD Norris1, RB Lock1, O Sangfelt4,5 and GM Marshall1,2

Loss of function mutation in FBXW7, an E3 ubiquitin ligase, is associated with good prognosis and early glucocorticoid treatment response in childhood T-cell acute lymphoblastic leukemia (T-ALL) by unknown mechanisms. Here, we show that FBXW7 targets the glucocorticoid receptor a (GRa) for ubiquitylation and proteasomal degradation in a manner dependent on glycogen synthase kinase 3 b-mediated phsophorylation. FBXW7 inactivation caused elevated GRa levels, and enhanced the transcriptional response to glucocorticoids. There was significant enhancement of GR transcriptional responses in FBXW7-deficient cell lines and primary T-ALL samples, in particular, for those pro-apoptotic regulatory proteins, BIM and PUMA. Reduced FBXW7 expression or function promoted glucocorticoid sensitivity, but not sensitivity to other chemotherapeutic agents used in T-ALL. Moreover, this was a general feature of different cancer cell types. Taken together, our work defines GRa as a novel FBXW7 substrate and demonstrates that favorable patient prognosis in T-ALL is associated with FBXW7 mutations due to enhanced GRa levels and steroid sensitivity. These findings suggest that inactivation of FBXW7, a putative tumor suppressor protein, may create a synthetic lethal state in the presence of specific anticancer therapies.

Leukemia (2013) 27, 1053–1062; doi:10.1038/leu.2012.361 Keywords: FBXW7; T-ALL; glucocorticoid receptor; ubiquitylation

INTRODUCTION isoforms exist as a result of alternative splicing and the use of 15 The F-box protein FBXW7 forms an SCF (Skp1/Cul1/F-box) type of multiple promoters. Upon binding to GCs, the predominant E3 ubiquitin ligase complex that targets various proteins for isoform, glucocorticoid receptor a (GRa), dissociates from heat- ubiquitylation and proteasomal degradation. SCF FBXW7 regulates shock proteins in the cytoplasm and translocates into the nucleus, cell cycle progression and differentiation, and depending on the where it interacts directly with GC response elements, resulting in 16 cell type, its inactivation results in deregulated expression of transactivation or transrepression of specific target genes. oncoproteins c-Myc, cyclin E, c-Jun, Mcl1 and Notch1.1 Cumulative Here, we report that FBXW7 mediates ubiquitylation and evidence supports a critical function for FBXW7 in the development proteasomal degradation of GRa in a glycogen synthase kinase of T-cell acute lymphoblastic leukemia (T-ALL), where it is frequently 3 (GSK3) b phosphorylation-dependent manner. Inactivation of inactivated by mutation.2–9 Surprisingly, the presence of FBXW7 FBXW7 enhances GRa stability and activity, promoting the mutations in T-ALL defines a subgroup of patients characterized by transcription of GR-responsive target genes, including the good response to induction chemotherapy with glucocorticoids activation of pro-apoptotic genes. Altogether, our work defines (GCs) and a more favorable outcome.2–9 GRa as a novel FBXW7 substrate and further demonstrates that The ability of GCs to specifically induce apoptosis in malignant inactivation of FBXW7 leads to elevated sensitivity to the lymphoid cells has led to their inclusion in almost all chemother- cytostatic effects of GCs in T-ALL. apy protocols for lymphocytic malignancies. GC response, after the first week of treatment with prednisolone has remained one of MATERIALS AND METHODS the strongest markers of prognosis and is used for treatment stratification in childhood ALL.10,11 GC resistance is a well- Cell culture, transfections and treatments documented feature of relapse.12,13 Tumor-derived cell lines were cultured according to the guidelines at the GCs mediate antiproliferative and pro-apoptotic effects in American Type Culture Collection. The following cell lines were used in this study: HEK293, HeLa, HCT116, MiaPaCa, Peer, Molt4, K562. HCT116 FBXW7- lymphocytes through activation of the glucocorticoid receptor WT and FBXW7 knockout cells was a gift from B Vogelstein, and were (GR),14 a ubiquitously expressed nuclear transcription factor that 14 grown in McCoy’s medium. Media was supplemented with 10% (v/v) fetal regulates the transcription of GC responsive genes. Several GR bovine serum, 2 mM L-glutamine, 100 U per ml penicillin, and 100 g per ml

1Children’s Cancer Institute Australia for Medical Research, Lowy Cancer Research Centre, University of New South Wales, Sydney, NSW, Australia; 2Centre for Children’s Cancer and Blood Disorders, Sydney Children’s Hospital, Sydney, Australia; 3Oncology Unit, Children’s Hospital at Westmead, Sydney, Australia; 4Department of Cell and Molecular Biology, Karolinska Institute, Stockholm, Sweden and 5Department of Oncology/Pathology, Cancercentrum Karolinska, Karolinska Institute, Stockholm, Sweden. Correspondence: Dr A Malyukova, Children’s Cancer Institute Australia for Medical Research, Lowy Cancer Research Centre, University of New South Wales, High St, Randwick, 2031, Australia. E-mail: [email protected] Or Professor GM Marshall, Centre for Children’s Cancer and Blood Disorders, Sydney Children’s Hospital, High St, Randwick, 2031, Sydney, Australia. E-mail: [email protected] Received 14 November 2012; accepted 30 November 2012; accepted article preview online 11 December 2012; advance online publication, 8 January 2013 FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1054 streptomycin. For all experiments described in this study, xenograft cells Labeling and hybridization to Affymetrix Human Gene 1.0 Arrays was were retrieved from cryostorage and resuspended in QBSF-60 medium performed by the Ramaciotti Center for Gene Function Analysis (University (Quality Biological, Gaithersburg, MD, USA) supplemented with Flt-3ligand of New South Wales, Sydney, Australia) according to the manufacturer’s (20 ng/ml; Amgen, Thousand Oaks, CA, USA), penicillin (100 units/ml), protocol. Hybridization was performed in duplicate. Raw cell files were streptomycin (100 Ag/ml) and L-glutamine (2 mmol/l; QBSF-60/F). Because processed for background correction and quantile normalization using the xenograft cells require high cell density and exhibit prolonged population robust multiarray averaging method and GenePattern software; Broad doubling times, the two most informative time points were used in the Institute, Cambridge, MA, USA (http://pwbc.garvan.unsw.edu.au/gp/pages/ cycloheximide (CHX) chase experiments. Transient plasmid transfections index.jsf) with ‘expression file creator’ module to generate the gene were performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA), expression data set (.gct file). according to the manufacturer’s protocols. Human small interfering RNA Differential expression analysis was performed using GenePattern (siRNA) oligonucleotides directed against FBXW7 have been described software with the ‘ComparativeMarkerSelection’ module, using a number previously.17 siRNA transfections were carried out using the Amaxa of permutations of 0 to calculate asymptotic P-values, and two-sided t-test nucleofector II (Lonza, Switzerland) as recommended by the to calculate P-values. The obtained data matrix file was imported into manufacturer (Lonza). pLentiLox3.7 lentiviral vectors expressing small ‘ComparativeMarkerSelectionViewer’ module for further analysis. Genes hairpin RNAs (shRNAs) against FBXW7 or a scrambled control, were significantly upregulated were generated by filtering using a false transfected into 293T cells along with the vectors of VSVG, RSV-REV and discovery rate p0.1 and P-value o0.01 and fold change 41.5. pMDLg/pRPE, to generate shFBXW7 lentiviruses. Peer and K562 cells were Gene set enrichment analysis was performed as described using the transduced with lentiviruses with a multiplicity of infection of 10, and gene sets from the molecular signatures database. We ranked genes by a incubated in roswell park memorial institute medium-1640 media decreasing moderated t statistic, created a pre-ranked list and used this as containing polybrene 8 mg/ml. Transduced cells were sorted based on input for running Gene set enrichment analysis in pre-ranked mode with GFP expression using a cell sorter (BD FacsAria; BD, Franklin Lakes, NJ, 1,000 permutations. We compared the moderated t statistic to the ‘C5_all’ USA). The Alamar Blue assay was used to determine cell viability as collection of curated gene sets from molecular signatures database 1.0. described.18 Treatments with doxorubicin, methotrexate, vincristine, The following parameters were used: probe set collapse ¼ true; permuta- dexamethasone (Pfizer, Groton, CT, USA), MG-132 (Enzo Life Sciences, tion: gene set, number of permutations ¼ 1,000. Gene set size: 15ono500. Exeter, UK) and BIO (Tocris Bioscience, Bristol, UK) were carried out as Gene sets were compared using the normalized enrichment score and indicated. For protein stability experiments, cells were treated with 100 mg deemed significant at 0.05 with a false discovery rate less than 0.25. per ml CHX (Enzo Life Sciences) as indicated. Flag–FBXW7 and HA-tagged ubiquitin constructs were described previously.17 Myc-DDK-GRa plasmid was purchased from OriGene RESULTS (Rockville, MD, USA). Flag–FBXW7-R465H, -R479Q, -R505C and Myc-DDK- Increased FBXW7 expression conferred sensitivity to GC GRa S404A mutants were generated with the QuikChange XL site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA, USA) according to To address the role of FBXW7 inactivation in conferring sensitivity the manufacturer’s instructions. to apoptosis, we analyzed the responses of human T-ALL cell lines with wild-type FBXW7 to increased doses of chemotherapeutic agents used in the induction phase of treatment of standard risk Western blotting, immunoprecipitation and in vivo ubiquitylation ALL: doxorubicin, methotrexate, vincristine and GCs. Molt4 and assay Peer T-ALL cell lines, with wild-type FBXW7,7 were transfected with The following primary antibodies were used: anti-HA (Y11), anti-MYC tag siRNA duplexes targeting a common region shared by the all (9E10) (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-Actin, anti- FBXW7 isoforms or scrambled control siRNAs. Silencing of FBXW7 Flag (M2) (Sigma-Aldrich, Sydney, Australia), anti-tubulin, anti-FBXW7 was confirmed by RT-PCR analysis in each cell line (Supplementary (ab12292-100) (Abcam, Cambridge, UK); anti-Mcl1 (Y37) (Cell Signaling, Figure S1A). No significant change in viability was observed in Danvers, MA, USA); anti-GR (BD biosciences, Franklin Lakes, NJ, USA); anti- Bim (AB17003, anti-Puma (AB4510) (Merck Millipore, Carlsbad, MA, USA). FBXW7-silenced cells following the treatment with vincristine, For immunoprecipitation studies, cells were lysed in NP-40 buffer with methotrexate or doxorubicin (Figure 1a). Although both Molt4 and 150 mM NaCl and protease inhibitors (Roche, Basel, Switzerland). For in vivo Peer T-ALL cell lines display GC resistance, we could show that at ubiquitylation assays, cells were transfected with the indicated plasmids the IC20 there was a 0.7 log lower GC concentration (P ¼ 0.0014) for 36 h following dexamethasone and proteasome inhibitor, MG-132 when FBXW7 expression was reduced. The only GC sensitive T-ALL treatment then lysed under denaturing conditions. Two mg of protein was cell line described (CEM-C7-14)19 contained point mutation in immunoprecipitated with specified antibodies. After immunoprecipitation, substrate binding region of FBXW7,7 therefore, was excluded from the samples were washed and analyzed by western blot using anti-HA the analysis. Importantly, GC sensitive solid cancer cell lines from antibody. cervical (HeLa) and pancreatic (MiaPaCa) cancers transfected with FBXW7-specific siRNA duplexes also exhibited increased sensitivity Microarray, quantitative real-time-PCR (qRT-PCR) and reporter to dexamethasone (Figure 1b). assays To improve the efficiency of FBXW7 gene silencing in leukemic Total RNA from siRNA/shRNA-transduced was extracted using RNeasy kit cells, lentiviral vectors expressing FBXW7-shRNA were generated, (Qiagen, Germantown, CA, USA). Complementary DNA was synthesized and the effect of dexamethasone on cell viability was examined in using the SuperScript III First-Strand Synthesis System (Invitrogen, CA, USA) leukemia cell lines, K562 and Peer, transduced with the shFBXW7 according to the manufacturer’s instructions. qRT-PCR was performed or scrambled control viruses. Knockdown of FBXW7 using the in triplicate and analyzed on an ABI 7,900 RT-PCR system. Glyceraldehyde shFBXW7 lentiviral vector was more effective than the transiently 3-phosphate dehydrogenase was used as a reference (TaqMan glycer- transfected siFBXW7 oligos (Supplementary Figure S1B). Reduced aldehyde 3-phosphate dehydrogenase control reagents, Applied Biosys- FBXW7 expression in K562 and Peer cells produced a small, but tems, Foster City, CA, USA). PCR primer sequences are available from the significantly enhanced GC sensitivity (Figure 1c and authors upon request. Comparative CT values were normalized against glyceraldehyde 3-phosphate dehydrogenase mRNA according to the Supplementary Figure S2). Similar results were found when we manufacturer’s protocol (ABI PRISM User Bulletin no. 2). assessed the ability of dexamethasone-treated FBXW7-deficient The luciferase reporter assay was performed in 293T cells transduced K562 cells to grow as colonies in semisolid media, as compared with pGL4.36-MMTV (Murine Mammary Tumor Virus)-luciferase plasmid with the control (Figure 1d). (Promega, Madison, WI, USA). 293T cells were stably transfected with pGL4.36-MMTV-luciferase reporter and used for further analysis. The Renilla FBXW7 regulates GRa protein stability luciferase vector pRL-TK (Promega) was used for normalization. Luciferase activities were quantified in a luminometer (VICTOR3, PerkinElmer, To test if the enhanced response to dexamethasone correlates Waltham, MA, USA) using the Dual-Luciferase Reporter Assay System with the expression level of the glucocorticoid receptor, steady- according to the manufacturer’s protocol (Promega). Fold change was state protein level of glucocorticoid receptor was examined in averaged from three separate experiments performed in triplicate. FBXW7-silenced cells following dexamethasone treatment. As

Leukemia (2013) 1053 – 1062 & 2013 Macmillan Publishers Limited FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1055

Figure 1. Effect of silencing of FBXW7 on the viability of the cells treated with different chemotherapeutic agents. (a) Dose–response curves of Peer and Molt4 cell lines treated with doxorubicin, methotrexate, vincristine or dexamethasone. Cells were transiently transduced with siRNA against FBXW7 using AMAXA nucleofection. Following seeding in complete growth medium, cells were exposed to increasing concentrations of different agents, as indicated. Cell viability was assessed after 48 h of treatment by Alamar Blue assay. Each point is the mean±s.e.m. of three independent determinations. (b) Viability of HeLa (cervical carcinoma) and MiaPaCa (pancreatic carcinoma) cell lines expressing siRNA duplexes against FBXW7 treated with dexamethasone (10 mM). Cell viability was assessed after 48 h of dexamethasone treatment by Alamar Blue assay. (c) Viability of Peer cells transduced with two different shRNAs against FBXW7 following dexamethasone (10 mM) treatment. Cell viability was determined by Alamar Blue 72 h after exposure to the dexamethasone (10 mM). Each point is the mean±s.e.m. of three independent determinations. (d) Dexamethasone treatment suppressed colony formation in FBXW7-silenced K562 cells. Equal numbers of cells were plated in triplicates in a culture medium with or without dexamethasone 10 mM. Twelve days later resulted colonies were counted. Each point is the mean±s.e.m. of three independent experiments.

& 2013 Macmillan Publishers Limited Leukemia (2013) 1053 – 1062 FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1056

Figure 2. FBXW7 deficiency associates with decreased degradation and increased GR stability. (a)GRa protein level in FBXW7-silenced HeLa and HEK293 cells. FBXW7 was first silenced by means of siRNA transfection for 48 h, GR was induced by treatment with 10 mM dexamethasone for 1 h. Actin protein level is shown as a control for loading. (b) Cycloheximide (CHX)-chase experiment assessing GRa stability in Peer and K562 cells expressing shFBXW7 or sh control. GRa was first induced by treating cells with 10 mM dexamethasone for 1 h and subsequently chased with 100 mg/ml CHX for the indicated times. Tubulin protein level is shown as a control for loading. Quantitation was done using Quantity One software (BioRad). (c)GRa levels in HCT 116 FBXW7-knockout cells. Myc-DDK-tagged GRa and Flag À tagged FBXW7 were overexpressed for 48 h, GRa was induced by dexamethasone, and then chased with CHX for the indicated times. Actin protein level is shown as a control for loading. (d) Immunoblot analysis of the GRa protein stability in T-ALL xenografts with FBXW7 R479Q mutation (no.27) or WT FBXW7 (no.30, 16, 31). Xenograft cells were retrieved from cryostorage, exposed to dexamethasone (10 mM, 1 h), and subjected to CHX-chase analysis. Tubulin protein level is shown as a control for loading.

shown on Figure 2a, silencing of FBXW7 markedly increased GRa mutations affect the stability of GR protein in primary human protein level in HEK 293 and HeLa cells in a manner similar to the leukemic cells, we used a panel of previously described T-ALL proteasomal inhibitor MG-132, suggesting that FBXW7 regulates xenografts20–22 and performed CHX-chase experiments of GRa turnover of GRa protein. To specifically address whether silencing protein following dexamethasone treatment. As shown in of FBXW7 increased GRa protein levels due to reduced GRa Figure 2d, the T-ALL xenograft with an FBXW7 mutation (ALL-27) protein degradation, we next performed CHX-chase experiments showed a marked increase in GR stability, as compared with the in leukemic cell lines with wild-type FBXW7, Peer, following xenografts with WT-FBXW7 (ALL-30, 16, 31). dexamethasone treatment. Indeed, silencing of FBXW7 using an As FBXW7 binds to its E3 ubiquitin ligase substrates through FBXW7 shRNA (shFBXW7), as compared with shRNA scramble interaction between a b-propeller structure formed by its WD40 control (sh control), significantly increased GRa protein half-life repeats and a CDC4 phosphodegron motif in the substrate,23,24 we from 3.8 to 5.2 h (Figure 2b). Moreover, FBXW7 silencing also next investigated whether FBXW7 could directly interact with GRa increased GRa protein half-life in another leukemia cell line, K562 in vivo. We performed FBXW7-GRa coimmunoprecipitation assays cells from 7.5 to 9.8 h (Figure 2b). To further evaluate the effect of in HEK293 cells transfected with an expression vector encoding FBXW7 on GR stability, we analyzed GRa degradation in colon WT-FBXW7 or an F-box deletion mutant of FBXW7 (FBXW7-DF), carcinoma HCT116 FBXW7-knockout cells reconstituted with WT- which is unable to interact with the SCF complex, but is still FBXW7. As shown in Figure 2c, expression of WT-FBXW7 reduced capable of binding to substrates. FBXW7 was immuno-purified GRa protein half-life, compared with cells expressing empty from cells treated with 10 mM dexamethasone, and GRa interaction vector. Similar data were obtained in HEK293 cells (Supplementary was examined by immunoblot analysis. As shown in Figure 3a, Figure S3C). To determine whether T-ALL-derived FBXW7 GRa copurified with WT-FBXW7 and FBXW7-DF. Reciprocal

Leukemia (2013) 1053 – 1062 & 2013 Macmillan Publishers Limited FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1057

Figure 3. FBXW7 interacts with GRa and promotes its ubiquitylation and degradation. (a) HEK293 cells were transfected with Flag À tagged FBXW7, FBXW7DF or an empty vector (EV) as indicated. Thirty-six hours later, endogenous GRa was induced by treatment with 10 mM dexamethasone for 1 h. Flag À FBXW7 was subsequently precipitated with Flag-antibody/protein G beads. (b) in vivo ubiquitylation of GRa in HEK293 cells. Cells were cotransfected with Myc-DDK-tagged GR, HA-tagged ubiquitin and Flag À tagged FBXW7 or EV control. Thirty-six hours later endogenous GRa was induced by treatment with 10 mM dexamethasone for 6 h. Before harvesting, cells were treated with 30 mmol/l MG-132 to preserve polyubiquitin chains of GRa, and GRa was subsequently precipitated with myc-antibody/protein G beads. Polyubiquitylated GRa conjugates were detected by western blot analysis with anti-HA antibodies as previously described.7 (c) in vivo ubiquitylation of GRa in HCT116 FBXW7-negative cells. Cells were cotransfected with Myc-DDK-tagged GRa, HA-tagged ubiquitin, and either Flag–tagged FBXW7 or different Flag À tagged arginine mutant versions of FBXW7 expression plasmids, or EV. Before harvesting, cells were treated with 30 mmol/l MG-132 to preserve polyubiquitin chains of GRa, and GRa was subsequently precipitated as described above. coimmunoprecipitation of ectopically expressed GRa and FBXW7 surrounding GRa S404 resembled a highly conserved CDC4 confirmed that these proteins interact in vivo (Supplementary phosphodegron consensus motif (Figure 4a). To investigate Figure S4A). To directly test whether GRa is an ubiquitylation whether FBXW7 promoted GRa ubiquitylation and degradation target of FBXW7, we performed in vivo ubiquitylation assays by in a GSK3b phosphorylation-dependent manner, HEK293 cells co-transfecting HEK293 cells with expression vectors for FBXW7, were treated with the pharmacological GSK3b inhibitor, BIO, GRa and HA-ubiquitin. Expression of FBXW7 induced a significant following transient transfection with GRa, HA-ubiquitin and WT- increase in ubiquitin À conjugated GRa in vivo (Figure 3b). To FBXW7 isoforms a or g. Different isoforms of FBXW7 were used, as determine whether the FBXW7 arginine mutants commonly there is evidence that substrates that lack a second phospho- identified in primary T-ALL specimens7 interfere with the group within the phospho-degron might require different FBXW7 formation of polyubiquitin chains on GRa, we performed in vivo isoforms.27 Inhibition of GSK3b activity with BIO markedly ubiquitylation assays in HCT116 FBXW7-knockout cells transfected impaired polyubiquitylation of GRa by FBXW7 (Figure 4b). We with either a WT-FBXW7 or with specific T-ALL-derived FBXW7 next created a GRa mutant, in which the serine residue at position mutants R456H, R479Q and R505C. All FBXW7 mutants were 404 was replaced by alanine (S404A). As shown in Figure 4c, the severely deficient in polyubiquitylating GRa as compared with WT- S404A GRa mutant was not efficiently ubiquitylated by FBXW7. FBXW7 (Figure 3c). These results, which were also recapitulated in Moreover, in CHX-chase experiments, we found that GRa protein HEK293 cells (Supplementary Figure S4B), supported the conclu- half-life was markedly increased in cells transfected with the GRa sion that T-ALL-derived FBXW7 mutations are not able to promote S404A mutant, compared with cells transfected with the WT GRa GRa ubiquitylation. (Figure 4d).

Phosphorylated GRa is degraded by FBXW7 Reduced FBXW7 expression enhances GC transcriptional response Ubiquitylation and proteasomal degradation of FBXW7 E3 As GC-induced cytostatic cellular effects depend on the transcrip- ubiquitin ligase substrates commonly requires prior phosphoryla- tional activation or repression of GR-responsive target genes,28,29 tion of the CDC4 phosphodegron motif on serine or threonine we next examined the inﬂuence of the FBXW7 inactivation on GR residues by GSK3.1,25 GSK3b has previously been reported to transcriptional activity. Transient transfection of HEK293 cells with phosphorylate GRa on serine 404 (S404), which correlated with a a MMTV luciferase reporter construct containing a GR-responsive decrease in GR protein levels.26 The amino acid sequence promoter,30 and shRNA against FBXW7 caused a signiﬁcant

& 2013 Macmillan Publishers Limited Leukemia (2013) 1053 – 1062 FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1058

Figure 4. FBXW7 promotes ubiquitylation of GRa in a GSK3b-mediated, phosphorylation-dependent manner. (a)GRa amino acid sequence surrounding serine-404 GSK3b phospho-site and the putative CDC4 phosphodegron degron are indicated (Upper panel). GSK3b phosphorylation site found in the GR is evolutionarily conserved. Amino acid sequence alignment of human GRa with chimpanzee, macaque, dog, cow, sheep and mouse (Lower panel). (b) FBXW7a and g isoforms catalyze polyubiquitylation of GRa in vivo in a GSK3b- dependent manner. HEK293 cells were co-transfected with Myc-DDK-GRa, HA-tagged ubiquitin, and Flag–tagged isoform speciﬁc FBXW7. Cells were treated with the GSK3b inhibitor, BIO, for 18 h. Before harvesting, cells were treated with 30 mmol/l MG-132, a proteasome inhibitor, to preserve polyubiquitin chains of GRa, and GRa was subsequently precipitated with antibodies/protein G beads. (c) in vivo ubiquitylation of GRa in HCT116 FBXW7-knockout cells. Cells were co-transfected with Myc-DDK-tagged GRa or phospho-deﬁcient S404A mutant of GRa, HA- tagged ubiquitin, and Flag–tagged FBXW7 or EV control. (d) CHX-chase analysis of GRa stability in HCT116 FBXW7 negative cells. Myc-DDK- tagged GRa or S404A mutant of GRa and Flag À tagged FBXW7 were overexpressed for 48 h, GRa was induced by dexamethasone, and then chased with CHX for the indicated times. Tubulin protein level is shown as a control for loading.

induction in GRa activity after dexamethasone treatment 8 h, and gene expression was profiled using Affymetrix HuGene1 (Figure 5a). Consistent with this finding, activation of the MMTV array. From 28 869 genes represented on the array, 327 were reporter was repressed in cells transfected with WT-FBXW7, while altered (X1.5 fold) upon dexamethasone treatment of Peer cells transfection of T-ALL-derived mutant FBXW7 did not significantly transduced with the control virus, with 25% being upregulated alter MMTV reporter activity, compared with empty vector control. and 75% downregulated (Figure 5b). FBXW7 depletion resulted in Thus, FBXW7 acted as a negative regulator of GRa and thereby considerable changes in gene expression following dexametha- modulated the response to GC treatment. sone treatment. We identified 645 genes with X1.5 fold change in To gain a better understanding of the molecular basis for the expression in Peer cells transduced with shFBXW7 at 8 h enhanced cytostatic effect of GC in FBXW7-depleted cells, we postdexamethasone treatment. Among the 645 genes, 37% were performed gene expression analyses of Peer T-cells transduced upregulated and 63% were downregulated (log fold change with shFBXW7 or control viruses. Peer cells transduced with X1.5, Pp0.01) (Figure 5b). Among the genes that were sig- shFBXW7 or control viruses were treated with dexamethasone for nificantly induced by dexamethasone alone, and further induced

Leukemia (2013) 1053 – 1062 & 2013 Macmillan Publishers Limited FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1059

Figure 5. Effects of FBXW7 on GR transcriptional activation. (a) HEK293-pGL4.36 cells stably expressing an MMTV promoter were transduced with different shRNAs against FBXW7 or transiently transfected with Flag À tagged FBXW7 or different arginine mutants of FBXW7 in expression plasmids, and, a Renilla luciferase internal control plasmid. Forty-eight hours later GR was induced by 10 mM dexamethasone and luciferase activity was measured using the Dual-Luciferase Reporter Assay System. (b, a) Venn diagram derived from microarray analysis of gene expression in Peer cells treated with 10 mM dexamethasone, for 8 h. The diagram shows the number of up- and downregulated genes (Pp0.01, FC 4±1.5) in the Peer cells transduced with shFBXW7 (shFBXW7/Dex) or control virus (Dex) following dexamethasone treatment. (c), Biological processes represented by the top Gene Set Enrichment Analysis Gene Ontology (C5_all) gene sets. Bars indicate the highest normalized enrichment score achieved for gene sets for Peer cell treated with dexamethasone and transduced with shFBXW7 or control. Lines represent False Discovery Rate for each gene set. (d) Downregulation of GC target genes in Peer cells in response to FBXW7 depletion by shRNA. FBXW7-silenced Peer cells were treated with dexamethasone for 8 h, total RNAs were extracted and used for qRT-PCR with gene- specific primers for the indicated GR target genes. Expression of PUMA was not significantly increased. Error bars represent s.d. from triplicate experiments normalized to b-actin. (e) qRT-PCR analysis documenting expression changes of selected GC target genes in T-ALL xenografts treated with dexamethasone for 8 h. FBXW7 mutant xenografts (ALL-8 and ALL-27) and FBXW7-WT xenografts (ALL-16, 30, 33, 34) were grouped for this analysis. Values represent a mean taken from multiple xenografts. Error bars represent s.d. from triplicate experiments normalized to b-actin. in FBXW7-depleted cells was the dexamethasone-responsive gene, RNA synthesis, rRNA processing (DNA directed RNA polymerase II TSC22D3 (GILZ), involved in the control of T-lymphocyte activation holoenzyme, ribonucleoprotein complex, structural constituent and apoptosis, the thioredoxin À interacting protein (TXNIP), of ribosome, ribosome, spliceosome, negative regulation of which is a GC-regulated primary response gene involved in transcription factor activity, RNA processing) and regulation of mediating GC-induced apoptosis, and nuclear factor of kappa light apoptosis (Figure 5c) at the P-value threshold of 0.001 and default polypeptide gene enhancer in B-cells inhibitor a (NFKBIA), involved false discovery rate cut off within gene set enrichment analysis of in the inhibition of the NFkB signaling pathway (Table 1). 0.25. However, at this cut off only 63 Gene Ontology classes were Dexamethasone treatment of FBXW7-silenced cells also resulted enriched in dexamethasone-treated cells. Significantly enriched in upregulation of several GR target genes with important Gene Ontology classes identified from this analysis were very functions in apoptosis control, such us BIM and PUMA (Table 1). similar for both gene sets, and were generally associated with Gene Set Enrichment Analysis31 was used to identify Gene lower false discovery rates and P-values in shFBXW7 cells treated Ontology classes specifically enriched in FBXW7-silenced cells with dexamethasone, compared with cells treated with dexa- treated with dexamethasone. A total of 383 Gene Ontology classes methasone alone (Figure 5c). from the molecular signatures database C5 collection were A subset of the differentially expressed GC target genes enriched in the dexamethasone-treated shFBXW7 cells including involved in apoptosis (BIM, PUMA, TXNIP and NFKBIA) was further metabolism (mitochondrion, mitochondrion organization and verified by qRT-PCR in shFBXW7-silenced Peer cells (Figure 5d). To biogenesis, organelle envelope, mitochondrial transport, further validate our microarray data, we compared mean envelope, glucose metabolic process, mitochondrial envelope), expression level of these GC target genes in four WT-FBXW7 and

& 2013 Macmillan Publishers Limited Leukemia (2013) 1053 – 1062 FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1060 Table 1. Representative genes differentially expressed in shFBXW7 Peer cells treated with dexamethasone and sh control Peer cells treated with dexamethasone, compared with sh control Peer cells treated with vehicle control

ProbeSetID Gene symbol FC shcontrol/Dex FC shFbxw7/Dex

upregulated 8174361 ‘TSC22D3: TSC22 domain family, member 3’ (GILZ) 2.5 2.5 7904726 TXNIP: thioredoxin interacting protein 2 2.5 7978644 ‘NFKBIA: nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha’ 1.6 2 8130556 ‘SOD2: superoxide dismutase 2, mitochondrial’ 1.5 1.8 8044375 BCL2L11: BCL2-like 11 (apoptosis facilitator) (BIM) * 1.2 8037872 BBC3: BCL2 binding component 3 (PUMA) * 1.5

downregulated 8100758 UGT2B7: UDP glucuronosyltransferase 2 family, polypeptide B7 À 1.5 À 2.5 7904414 ‘HSD3B1: hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 1’ À 1.7 À 2 7982028 ‘SNORD115-11: small nucleolar RNA, C/D box 115-11’ À 2 À 2.5 7982024 ‘SNORD115-12: small nucleolar RNA, C/D box 115-12’ À 2 À 2.5 8095680 IL8: interleukin 8 À 1.8 À 1.5 8094130 USP17: ubiquitin speciﬁc peptidase 17 À 1.5 À 2 8160394 ‘IFNA17: interferon, alpha 17’ À 1.9 À 1.9 Asterisk denotes no signiﬁcant changes. FC represents log fold change.

two mutant human T-ALL xenografts treated with dexamethasone for 8 h (Figure 5e). Expression of BIM and PUMA were signiﬁcantly upregulated in the group of FBXW7 mutant xenografts, compared with WT-FBXW7 xenografts (Figure 5e).

Mutant FBXW7 enhances the GC apoptotic signal via effects on BH3-only proteins Lastly, we examined the expression level of the pro-apoptotic proteins BIM and PUMA, and the antiapoptotic protein Mcl1, following dexamethasone treatment of K562 leukemia cells transduced with shFBXW7.InFBXW7-silenced K562 cells, with or without dexamethasone, we observed a marked increase in the levels of BIM and PUMA, but no change in Mcl1 protein level (Figure 6a). We also compared BIM, PUMA, and Mcl1 protein levels following dexamethasone treatment in a single FBXW7-mutant human T-ALL xenograft (ALL-27), to three xenografts with WT-FBXW7 (ALL-30, 16, 31). We found increased PUMA and BIM levels in the mutant FBXW7 T-ALL xenografts, compared with WT-FBXW7 xenografts (Figure 6b). Consistent with these data, we also observed increased BIM protein levels in dexamethasone- treated T-ALL cell lines with mutant FBXW7 (CEM, Molt4, Jurkat), compared with the Peer cell line with WT-FBXW7 (Figure 6c). These results demonstrate that FBXW7 regulates GRa protein turnover and support a causal relationship between FBXW7 mutation status and GC sensitivity in T-ALL, where GC treatment of FBXW7-deficient cells potentiates expression of GRa-responsive Figure 6. Effects of FBXW7 on pro-apoptotic proteins BIM and Puma. pro-apoptotic genes. (a) Mcl1, Bim and Puma levels in FBXW7-silenced K562 cells. K562 cells were transduced with shFBXW7 lentivirus; GR was induced by treatment with 10 mM dexamethasone for 8 h. Tubulin protein level is shown as a control for loading. (b) Immunoblot analysis of the Mcl1, DISSCUSSION Bim and Puma protein expression in T-ALL xenografts with mutant Inactivation of FBXW7 has been linked to tumor development, (no. 27) or WT-FBXW7 (no. 30, 16, 31). Xenograft cells were retrieved as a tumor suppressor function by targeting oncoproteins for from cryostorage and exposed to dexamethasone (10 mM, 8 h). proteasomal degradation.1 The phenotypic outcome associated Tubulin protein level is shown as a control for loading. (c) Immunoblot analysis of the Bim protein expression in FBXW7 with FBXW7 deficiency has also been reported to depend on 32–34 mutant (CEM, MOLT4, Jurkat) or WT (Peer) T-ALL cell lines. Actin cell type. However, inactivation of FBXW7 by the loss of protein level is shown as a control for loading. expression or mutation was associated with favorable prognosis in primary breast cancer,35 and with good prognosis and early GC treatment response in childhood T-ALL2–8 Here, we have shown that, while FBXW7 loss of function in T-ALL may contribute The GR is a critical target in chemotherapy protocols used to to the malignant phenotype before diagnosis, it then functions treat ALL, however, very little is known about the processes that as a treatment sensitizer in the synthetic environment of regulate GR stability and activity. Studies have confirmed that GC GC treatment. sensitivity could be restored through overexpression of GR in

Leukemia (2013) 1053 – 1062 & 2013 Macmillan Publishers Limited FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1061 GC-resistant ALL cells,36–38 and suggested that GC resistance may have greater overall therapeutic benefit by enhancing GC be linked to GR expression levels. However, other studies have sensitivity. shown that GC resistance in ALL can also occur downstream of ligand-induced translocation of GR.21,22 In the present study, we demonstrate that FBXW7 interacts with CONFLICT OF INTEREST GRa through a conserved, GSK3b phosphorylated degron in GRa, The authors declare no conflict of interest. promoting its ubiquitylation and proteasomal degradation. GR protein stability has been reported to be a crucial factor in determining its transcriptional activity.39 We found that ACKNOWLEDGEMENTS knockdown of FBXW7 in leukemic cells promoted the expression of several GC responsive pro-apoptotic target genes, and We thank Bert Vogelstein (Johns Hopkins University, Baltimore, MD) for kindly accordingly, enhanced GC sensitivity in GC resistant T-ALL cells. providing FBXW7-deficient HCT 116 cells and parental control cells. Children’s Cancer Institute Australia for Medical Research is affiliated with the Sydney Children’s Our microarray studies also correlate with previously published 40,41 Hospital and the University of New South Wales. This study was supported by grants results on GC-induced transcriptional response. Notably, from National Health and Medical Research Council of Australia (G.M., M.N., M.H.), inactivation of FBXW7 by shRNA resulted in enhanced induction Cancer Council New South Wales (G.M., M.N., M.H.), Steven Walter Children’s Cancer or repression of a substantial number of dexamethasone- Foundation (G.M.), Cancer Institute New South Wales (G.M., M.N., M.H.), and Leukemia regulated genes. Among the genes that were further induced in Foundation (G.M.), Swedish Society for Medical Research (A.M.), the Swedish Cancer shFBXW7/DEX T-ALL cells were the GC target genes: BCL211/BIM, Foundation, the Swedish Research Council (O.S.), the Swedish Children’s Cancer BBC3/PUMA, TXNIP, NFKBIA, TSC22D3/GILZ. The expression of these Foundation and Karolinska Institute Foundations (O.S.). genes was also induced in FBXW7-mutated T-ALL xenografts. Thus, these results suggest that FBXW7 inactivation in T-ALL reinforces the transcriptional response to GC, through a reduced rate of REFERENCES degradation of GR in T-ALL cells. 1 Tan Y, Sangfelt O, Spruck C. The Fbxw7/hCdc4 tumor suppressor in human cancer. Besides elevated levels of the GR protein, we also found Cancer Lett 2008; 271: 1–12. increased protein expression of the pro-apoptotic proteins BIM 2 Asnafi V, Buzyn A, Le Noir S, Baleydier F, Simon A, Beldjord K et al. NOTCH1/ and PUMA in FBXW7-deficient cells. Accumulating evidence FBXW7 mutation identifies a large subgroup with favorable outcome in adult suggests that dexamethasone-induced apoptosis in lymphocytes T-cell acute lymphoblastic leukemia (T-ALL): a Group for Research on Adult Acute is executed through the intrinsic pathway, which involves Lymphoblastic Leukemia (GRAALL) study. Blood 2009; 113: 3918–3924. 3 Ben Abdelali R, Asnafi V, Leguay T, Boissel N, Buzyn A, Chevallier P et al. Pediatric- activation of pro-apoptotic factors, loss of mitochondrial trans- 14 inspired intensified therapy of adult T-ALL reveals the favorable outcome of membrane potential and subsequent activation of caspases. A NOTCH1/FBXW7 mutations, but not of low ERG/BAALC expression: a GRAALL number of recent studies have identified BIM and PUMA as crucial study. Blood 2011; 118: 5099–5107. 22,42–44 factors for initiating apoptosis induced by GR, but the 4 Breit S, Stanulla M, Flohr T, Schrappe M, Ludwig WD, Tolle G et al. Activating mechanism, by which GCs regulate BIM and PUMA expression in NOTCH1 mutations predict favorable early treatment response and long-term lymphoid cells is poorly understood. Mechanisms involving outcome in childhood precursor T-cell lymphoblastic leukemia. Blood 2006; 108: transcriptional activation of BIM and/or PUMA by GR in T-ALL 1151–1157. cell lines and primary thymocytes have been suggested,40,43,45 but 5 Clappier E, Collette S, Grardel N, Girard S, Suarez L, Brunie G et al. NOTCH1 and induction of apoptosis through interaction of GSK3b-BIM upon FBXW7 mutations have a favorable impact on early response to treatment, but 46 not on outcome, in children with T-cell acute lymphoblastic leukemia (T-ALL) dexamethasone treatment has also been proposed. Our data on treated on EORTC trials 58881 and 58951. Leukemia 2010; 24: 2023–2031. BIM upregulation in FBXW7-deficient T-ALL cell lines are consistent 6 Kox C, Zimmermann M, Stanulla M, Leible S, Schrappe M, Ludwig WD et al. The with the increased sensitivity to GC in these cells. The molecular favorable effect of activating NOTCH1 receptor mutations on long-term outcome mechanism behind upregulation of BIM in FBXW7-deficient cells in T-ALL patients treated on the ALL-BFM 2000 protocol can be separated from remains to be determined, but our data suggest that increased FBXW7 loss of function. Leukemia 2010; 24: 2005–2013. GRa activity results in enhanced BIM expression, although we 7 Malyukova A, Dohda T, von der Lehr N, Akhoondi S, Corcoran M, Heyman M et al. cannot exclude the possibility that upregulation of other FBXW7 The tumor suppressor gene hCDC4 is frequently mutated in human T-cell acute substrates like c-Myc or c-Jun may also result in increased BIM lymphoblastic leukemia with functional consequences for Notch signaling. Cancer and/or PUMA expression.47,48 Res 2007; 67: 5611–5616. 8 Park MJ, Taki T, Oda M, Watanabe T, Yumura-Yagi K, Kobayashi R et al. FBXW7 and We propose a model in normal cells, whereby phosphorylation NOTCH1 mutations in childhood T cell acute lymphoblastic leukaemia and T cell of GRa by GSK3b triggers the ubiquitylation of GRa and terminates non-Hodgkin lymphoma. Br J Haematol 2009; 145: 198–206. GC-dependent GRa signaling. Our study suggests that the 9 Zuurbier L, Homminga I, Calvert V, te Winkel ML, Buijs-Gladdines JG, Kooi C et al. functional inactivation of FBXW7 might specifically sensitize NOTCH1 and/or FBXW7 mutations predict for initial good prednisone response leukemic cells to apoptosis in response to dexamethasone, but not for improved outcome in pediatric T-cell acute lymphoblastic leukemia through increased GR protein stability and expression of GR patients treated on DCOG or COALL protocols. Leukemia Dec 2010; 24: target genes such as BIM and PUMA. Cells with mutant FBXW7 in 2014–2022. physiological conditions survive and evade apoptosis through the 10 Dordelmann M, Reiter A, Borkhardt A, Ludwig WD, Gotz N, Viehmann S et al. activation of Mcl1 and/or NFkB2.49–52 In our model, enhanced GR Prednisone response is the strongest predictor of treatment outcome in infant acute lymphoblastic leukemia. Blood 1999; 94: 1209–1217. protein stability and induction of BIM and PUMA in FBXW7- 11 Annino L, Vegna ML, Camera A, Specchia G, Visani G, Fioritoni G et al. Treatment deficient leukemic cells will predispose these cells to apoptosis, of adult acute lymphoblastic leukemia (ALL): long-term follow-up of the GIMEMA whereas Mcl1 protein expression was not altered. ALL 0288 randomized study. Blood 2002; 99: 863–871. To our knowledge, these findings are the first to show that 12 Kaspers GJ, Wijnands JJ, Hartmann R, Huismans L, Loonen AH, Stackelberg A et al. loss-of-function mutation of a protein, which acts as a tumor Immunophenotypic cell lineage and in vitro cellular drug resistance in childhood suppressor in other contexts can provide a synthetic lethal relapsed acute lymphoblastic leukaemia. Eur J Cancer 2005; 41: 1300–1303. environment in the presence of GCs. Our results provide 13 Pieters R, den Boer ML, Durian M, Janka G, Schmiegelow K, Kaspers GJ et al. important new insights into the complex interactions between Relation between age, immunophenotype and in vitro drug resistance in 395 putative tumor suppressors and the therapeutic environment children with acute lymphoblastic leukemia-implications for treatment of infants. Leukemia 1998; 12: 1344–1348. of cancer cells and suggest that therapeutic attempts to reactivate 14 Distelhorst CW. Recent insights into the mechanism of glucocorticosteroid- the mutant FBXW7 function may create treatment resistance induced apoptosis. Cell Death Differ 2002; 9: 6–19. in specific instances. Importantly, the results suggest that 15 Yudt MR, Cidlowski JA. The glucocorticoid receptor: coding a diversity of proteins blocking the function of wild-type FBXW7 in T-ALL cells may and responses through a single gene. Mol Endocrinol 2002; 16: 1719–1726.

& 2013 Macmillan Publishers Limited Leukemia (2013) 1053 – 1062 FBXW7 regulates glucocorticoid response in T-ALL A Malyukova et al 1062 16 Greenstein S, Ghias K, Krett NL, Rosen ST. Mechanisms of glucocorticoid-mediated tumorigenesis mediated through negative regulation of c-Myc and Notch. apoptosis in hematological malignancies. Clinical cancer research: an official Oncogene; e-pub ahead of print, 4 June 2012; doi:10.1038/onc.2012.213. journal of the American Association for Cancer Research 2002; 8: 1681–1694. 35 Akhoondi S, Lindstrom L, Widschwendter M, Corcoran M, Bergh J, Spruck C et al. 17 van Drogen F, Sangfelt O, Malyukova A, Matskova L, Yeh E, Means AR et al. Inactivation of FBXW7/hCDC4-beta expression by promoter hypermethylation is Ubiquitylation of cyclin E requires the sequential function of SCF complexes associated with favorable prognosis in primary breast cancer. Breast Cancer Res containing distinct hCdc4 isoforms. Mol Cell 2006; 23: 37–48. 12: R105. 18 Marshall GM, Bell JL, Koach J, Tan O, Kim P, Malyukova A et al. TRIM16 acts as a 36 Reichardt HM, Umland T, Bauer A, Kretz O, Schutz G. Mice with an increased tumour suppressor by inhibitory effects on cytoplasmic vimentin and nuclear glucocorticoid receptor gene dosage show enhanced resistance to stress and E2F1 in neuroblastoma cells. Oncogene 2010; 29: 6172–6183. endotoxic shock. Mol Cell Biol 2000; 20: 9009–9017. 19 Bornhauser BC, Bonapace L, Lindholm D, Martinez R, Cario G, Schrappe M et al. 37 Pazirandeh A, Xue Y, Okret S, Jondal M. Glucocorticoid resistance in thymocytes Low-dose arsenic trioxide sensitizes glucocorticoid-resistant acute lymphoblastic from mice expressing a T cell receptor transgene. Biochem Biophys Res Commun leukemia cells to dexamethasone via an Akt-dependent pathway. Blood 2007; 2000; 276: 189–196. 110: 2084–2091. 38 Schmidt S, Irving JA, Minto L, Matheson E, Nicholson L, Ploner A et al. Gluco- 20 Liem NL, Papa RA, Milross CG, Schmid MA, Tajbakhsh M, Choi S et al. Char- corticoid resistance in two key models of acute lymphoblastic leukemia occurs at acterization of childhood acute lymphoblastic leukemia xenograft models for the the level of the glucocorticoid receptor. FASEB J 2006; 20: 2600–2602. preclinical evaluation of new therapies. Blood 2004; 103: 3905–3914. 39 Wallace AD, Cidlowski JA. Proteasome-mediated glucocorticoid receptor degra- 21 Bachmann PS, Gorman R, Mackenzie KL, Lutze-Mann L, Lock RB. Dexamethasone dation restricts transcriptional signaling by glucocorticoids. J Biol Chem 2001; 276: resistance in B-cell precursor childhood acute lymphoblastic leukemia occurs 42714–42721. downstream of ligand-induced nuclear translocation of the glucocorticoid 40 Miller AL, Komak S, Webb MS, Leiter EH, Thompson EB. Gene expression profiling receptor. Blood 2005; 105: 2519–2526. of leukemic cells and primary thymocytes predicts a signature for apoptotic 22 Bachmann PS, Gorman R, Papa RA, Bardell JE, Ford J, Kees UR et al. Divergent sensitivity to glucocorticoids. Cancer Cell Int 2007; 7:18. mechanisms of glucocorticoid resistance in experimental models of pediatric 41 Medh RD, Webb MS, Miller AL, Johnson BH, Fofanov Y, Li T et al. Gene expression acute lymphoblastic leukemia. Cancer Res 2007; 67: 4482–4490. profile of human lymphoid CEM cells sensitive and resistant to glucocorticoid- 23 Strohmaier H, Spruck CH, Kaiser P, Won K-A, Sangfelt O, Reed SI. Human F-box evoked apoptosis. Genomics 2003; 81: 543–555. protein hCdc4 targets cyclin E for proteolysis and is mutated in a breast cancer 42 Planey SL, Abrams MT, Robertson NM, Litwack G. Role of apical caspases and cell line. Nature 2001; 413: 316–322. glucocorticoid-regulated genes in glucocorticoid-induced apoptosis of pre-B 24 Koepp DM, Schaefer LK, Ye X, Keyomarsi K, Chu C, Harper JW et al. Phosphor- leukemic cells. Cancer Res 2003; 63: 172–178. ylation-dependent ubiquitination of cyclin E by the SCFFbw7 ubiquitin ligase. 43 Wang Z, Malone MH, He H, McColl KS, Distelhorst CW. Microarray analysis Science 2001; 294: 173–177. uncovers the induction of the proapoptotic BH3-only protein Bim in 25 Wei W, Jin J, Schlisio S, Harper JW, Kaelin Jr WG. The v-Jun point mutation allows multiple models of glucocorticoid-induced apoptosis. J Biol Chem 2003; 278: c-Jun to escape GSK3-dependent recognition and destruction by the Fbw7 ubi- 23861–23867. quitin ligase. Cancer Cell 2005; 8: 25–33. 44 Erlacher M, Michalak EM, Kelly PN, Labi V, Niederegger H, Coultas L et al. BH3-only 26 Galliher-Beckley AJ, Williams JG, Collins JB, Cidlowski JA. Glycogen synthase proteins Puma and Bim are rate-limiting for gamma-radiation- and glucocorti- kinase 3beta-mediated serine phosphorylation of the human glucocorticoid coid-induced apoptosis of lymphoid cells in vivo. Blood 2005; 106: 4131–4138. receptor redirects gene expression profiles. Mol Cell Biol 2008; 28: 7309–7322. 45 Zhao YN, Guo X, Ma ZG, Gu L, Ge J, Li Q. Pro-apoptotic protein BIM in apoptosis of 27 Welcker M, Clurman BE. Fbw7/hCDC4 dimerization regulates its substrate inter- glucocorticoid-sensitive and -resistant acute lymphoblastic leukemia CEM cells. actions. Cell Div 2007; 2:7. Med Oncol 2011; 28: 1609–1617. 28 Renner K, Ausserlechner MJ, Kofler R. A conceptual view on glucocorticoid- 46 Spokoini R, Kfir-Erenfeld S, Yefenof E, Sionov RV. Glycogen synthase kinase-3 plays lnduced apoptosis, cell cycle arrest and glucocorticoid resistance in lymphoblastic a central role in mediating glucocorticoid-induced apoptosis. Mol Endocrinol 2010; leukemia. Curr Mol Med 2003; 3: 707–717. 24: 1136–1150. 29 Frankfurt O, Rosen ST. Mechanisms of glucocorticoid-induced apoptosis in 47 Cazanave SC, Mott JL, Elmi NA, Bronk SF, Werneburg NW, Akazawa Y et al. JNK1- hematologic malignancies: updates. Curr Opin Oncol 2004; 16: 553–563. dependent PUMA expression contributes to hepatocyte lipoapoptosis. J Biol 30 Belikov S, Gelius B, Wrange O. Hormone-induced nucleosome positioning in the Chem 2009; 284: 26591–26602. MMTV promoter is reversible. EMBO J 2001; 20: 2802–2811. 48 Egle A, Harris AW, Bouillet P, Cory S. Bim is a suppressor of Myc-induced mouse B 31 Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA et al. cell leukemia. Proc Natl Acad Sci USA 2004; 101: 6164–6169. Gene set enrichment analysis: a knowledge-based approach for interpreting 49 Arabi A, Ullah K, Branca RM, Johansson J, Bandarra D, Haneklaus M et al. Pro- genome-wide expression profiles. Proc Natl Acad Sci USA 2005; 102: 15545–15550. teomic screen reveals Fbw7 as a modulator of the NF-kappaB pathway. Nat 32 Matsumoto A, Onoyama I, Sunabori T, Kageyama R, Okano H, Nakayama KI. Commun 2012; 3: 976. Fbxw7-dependent degradation of Notch is required for control of ‘‘stemness’’ and 50 Busino L, Millman SE, Scotto L, Kyratsous CA, Basrur V, O’Connor O et al. neuronal-glial differentiation in neural stem cells. J Biol Chem 2011; 286: Fbxw7alpha- and GSK3-mediated degradation of p100 is a pro-survival 13754–13764. mechanism in multiple myeloma. Nat Cell Biol Apr 14: 375–385. 33 Onoyama I, Tsunematsu R, Matsumoto A, Kimura T, de Alboran IM, Nakayama K 51 Inuzuka H, Shaik S, Onoyama I, Gao D, Tseng A, Maser RS et al. SCF(FBW7) reg- et al. Conditional inactivation of Fbxw7 impairs cell-cycle exit during T cell ulates cellular apoptosis by targeting MCL1 for ubiquitylation and destruction. differentiation and results in lymphomatogenesis. J Exp Med 2007; 204: Nature 471: 104–109. 2875–2888. 52 Wertz IE, Kusam S, Lam C, Okamoto T, Sandoval W, Anderson DJ et al. Sensitivity 34 Ishikawa Y, Hosogane M, Okuyama R, Aoyama S, Onoyama I, Nakayama KI et al. to antitubulin chemotherapeutics is regulated by MCL1 and FBW7. Nature 471: Opposing functions of Fbxw7 in keratinocyte growth, differentiation and skin 110–114.

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

Leukemia (2013) 1053 – 1062 & 2013 Macmillan Publishers Limited