Letters to the Editor 1200 1 2 1 3 SRo¨ttgers , M Gombert , A Teigler-Schlegel , K Busch , 2 Bullrich F, Morris SW, Hummel M, Pileri S, Stein H, Croce CM. 4 5 1 2 U Gamerdinger , R Slany , J Harbott and A Borkhardt Nucleophosmin (NPM) gene rearrangements in Ki-1-positive 1Department of Pediatric Hematology and Oncology, lymphomas. Cancer Res 1994; 54: 2873–2877. Oncogenetic Laboratory, University Hospital, Justus Liebig 3 Soda M, Choi YL, Enomoto M, Takada S, Yamashita Y, Ishikawa S University Giessen, Giessen, Germany; et al. Identification of the transforming EML4-ALK fusion gene in 2Clinic of Pediatric Hematology, Oncology and Clinical non-small-cell lung cancer. Nature 2007; 448: 561–566. Immunology, Heinrich-Heine University Duesseldorf, 4 Jazii FR, Najafi Z, Malekzadeh R, Conrads TP, Ziaee AA, Duesseldorf, Germany; Abnet C et al. Identification of squamous cell carcinoma asso- 3Hessian State Office of Criminal InvestigationFWiesbaden, ciated proteins by proteomics and loss of beta tropomyosin Wiesbaden, Germany; expression in esophageal cancer. World J Gastroenterol 2006; 12: 4Department of Pathology, University Hospital, Giessen, 7104–7112. Germany and 5 Chiarle R, Voena C, Ambrogio C, Piva R, Inghirami G. The 5 anaplastic lymphoma kinase in the pathogenesis of cancer. Nat Rev Department of Genetics, Friedrich-Alexander University Cancer 2008; 8: 11–23. Erlangen, Erlangen, Germany 6 Trumper L, Pfreundschuh M, Bonin FV, Daus H. Detection of the E-mail: [email protected] t(2;5)-associated NPM/ALK fusion cDNA in peripheral blood cells of healthy individuals. Br J Haematol 1998; 103: 1138–1144. 7 Voena C, Conte C, Ambrogio C, Boeri EE, Boccalatte F, Mohammed S References et al. The tyrosine phosphatase Shp2 interacts with NPM-ALK and regulates anaplastic lymphoma cell growth and migration. Cancer Res 1 Ma Z, Hill DA, Collins MH, Morris SW, Sumegi J, Zhou M et al. 2007; 67: 4278–4286. Fusion of ALK to the Ran-binding protein 2 (RANBP2) gene in 8 Li R, Morris SW. Development of anaplastic lymphoma kinase (ALK) inflammatory myofibroblastic tumor. Genes Chromosomes Cancer small-molecule inhibitors for cancer therapy. Med Res Rev 2008; 2003; 37: 98–105. 28: 372–412.

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

Haploinsufficiency of the IKZF1 (IKAROS) tumor suppressor gene cooperates with BCR-ABL in a transgenic model of acute lymphoblastic leukemia

Leukemia (2010) 24, 1200–1204; doi:10.1038/leu.2010.63; instrumental in BCR-ABL-induced B-ALL progression and poor published online 15 April 2010 clinical outcome.4 To investigate whether impaired IKAROS function can facili- tate BCR-ABL-induced leukemogenesis, we crossed BCR-ABL transgenic mice that exclusively develop B-ALL5 with mice IKAROS is a chromatin-remodeling and sequence-specific carrying an hypomorphic IKZF1 allele (IKL).6 Previous studies encoded by the IKZF1 gene that is involved have shown that the IKL allele, generated by the knock-in of Lac in a number of developmental processes in hematopoiesis. In Z into IKZF1 exon 2, produces 5–10% of wt levels of IKAROS particular, IKAROS is essential for the specification of the proteins that lack amino acids encoded by exon 2. Importantly, lymphoid lineages from myeloid/lymphoid multipotent progeni- these mutant proteins were as efficient as wt IKAROS at rescuing tors, is involved in the differentiation of pro-B cells into pre-B abnormal phenotypes from homozygous mutant mice, suggest- cells and, together with the related IKAROS family member ing that the IKL allele is mainly, if not exclusively, a loss of Aiolos, downregulates pre-BCR function (for a review, see Nutt function allele.7,8 Two cohorts of BCR-ABL þ compound and Kee1). Over the past 10 years, reverse transcription–PCR mice were generated that carried either two alleles of wt IKZF1 studies have reported the expression of short IKAROS isoforms (BCR-ABL þ /0;IKþ / þ ), or one wt and one mutant IKZF1 allele in acute lymphoblastic leukemia (ALL) patients at diagnosis, (BCR-ABL þ /0;IKL/ þ ), which were compared for leukemia with higher prevalence in BCR-ABL-associated B-ALL.2 These development. Nontransgenic IK þ / þ and IKL/ þ littermates were short isoforms encode IKAROS proteins that include only two included as controls. In line with previous data,5 BCR-ABL þ /0; out of the four central zinc fingers responsible for their specific IK þ / þ mice developed B-ALL with a median latency of 14,7 binding to DNA while keeping intact the C-terminal zinc fingers weeks (Figure 1a). Strikingly, leukemia onset was considerably involved in their oligomerization with DNA-binding-competent accelerated in BCR-ABL þ /0;IKL/ þ mice (median latency of isoforms of IKAROS and other IKAROS family members. 5.7 weeks; Figure 1a). None of the control groups developed Thus, these short isoforms can function as dominant-negative disease during a 6-month observation period (Figure 1a; data (DN) mutants of wild-type (wt) IKAROS and IKAROS family not shown). Histologically, the leukemias arising from both the members. Recently, comprehensive genome-wide SNP array BCR-ABL þ /0;IKþ / þ and BCR-ABL þ /0;IKL/ þ cohorts were analyses of DNA copy number alterations in ALL showed the similar, with recurrent leukemic cell infiltration of bone marrow, frequent monoallelic deletion of all or part of IKZF1, the latter spleen, lymph nodes, blood and nonhematopoietic tissues situation often linked to the expression of DN IKAROS isoforms including liver and brain (data not shown). Flow cytometry in 475% of pediatric and 490% of adult BCR-ABL B-ALL and analyses identified large leukemic blasts in both cohorts that lymphoid blast crisis chronic myeloid leukemia (CML) patients.3 were B220 þ sIgMÀ with a variable proportion of CD43 þ cells These data suggest that impaired function of IKAROS itself, and in individual leukemias (Figures 1b and c; median value 20% possibly of other members of the IKAROS family, could be and range 1À85% in individual BCR-ABL þ /0;IKþ / þ tumors;

Leukemia Letters to the Editor 1201 100 BCR-ABL; IKL/+ 90 +/+ +/+ BCR-ABL; IK 80 BCR ABL; IK 75 L/+ 70 BCR-ABL; IK IKL/+ 60 50 50 40 Survival (%) 25 30

% B220+ CD43+ 20 0 10 0 102030405060 0 Age of mice (weeks) BCR ABL; IK+/+ BCR-ABL; IKL/+

BCR-ABL; IK+/+ BCR-ABL; IKL/+ BCR-ABL; IK+/+ BCR-ABL; IKL/+

1000 10000 1000 10000 L/L 800 IK Prog. 1318 1396 1414 1443 1455 1466 1469 1000 83.8% 800 1000 77.1% 600 600 100 100 IK1

B220 SSC-H 400 B220 400 IK1* IK2 200 10 10 IK2* 69% 200 62.4% 0 1 0 1 0200 400 600 800 1000 1 10 100 1000 10000 0 200 400 600 800 1000 110 100 1000 10000 FSC-H IgMFSC-H IgM

10000 10000

1000 90.4% 9.18% 1000 81.5% 18.1% Pax5

100 100

B220

B220 10 10 β-Actin

1 1 110 100 1000 10000 1 10 100 1000 10000 CD43 CD43

Figure 1 Impaired IKAROS function accelerates B-ALL onset in BCR-ABL transgenic mice. (a) Kaplan–Meyer survival curves for BCR-ABL þ /0; IK þ / þ , BCR-ABL þ /0;IKL/ þ and IKL/ þ mice. Mice harboring a BCR-ABL construct encoding P190BCR-ABL under control of the metallothionein promoter5 were obtained from the Jackson Laboratory (Charles River Laboratories, L’Arbresle, France), re-derived and backcrossed for eight generations on C57B/6 mice. These mice were crossed to IKL/ þ mice,6 maintained on C57B/6 background and the resulting cohorts of BCR-ABL þ /0; IK þ / þ (n ¼ 101; black line) and BCR-ABL þ /0;IKL/ þ (n ¼ 86; gray line) followed for leukemia development. Cohorts of control IKL/ þ (n ¼ 10; hatched line) and nontransgenic littermates (not shown) were also analyzed. Mice were killed when moribund (mobility weakness or paralysis, lusterless fur, respiratory distress). Survival curves were plotted using the Prism 4 software (Graph Pad Software Inc., San Diego, CA, USA). The nnn P-value (o0.0001 ) for the BCR-ABL þ /0;IKþ / þ and BCR-ABL þ /0;IKL/ þ cohorts was calculated using the log-rank test. (b) Cell-surface analysis of the leukemic cells obtained from the bone marrow of representative BCR-ABL þ /0;IKþ / þ and BCR-ABL þ /0;IKL/ þ cases. Leukemic blasts were gated as FSC/SSC large cells and analyzed for B220, sIgM and CD43 expression, as indicated. The antibodies used were from BD Biosciences (Le Pont De Claix, France): anti-IgM-FITC (R6-60.2), anti-B220-PE-Cy5 (RA3-6B2), anti-CD43-PE (S7). (c) Expression of CD43 in a series of individual leukemias obtained from BCR-ABL þ /0;IKþ / þ and BCR-ABL þ /0;IKL/ þ mice. (d) Western blot analyses of IKAROS, Pax5 and b-actin expression in normal B-cell progenitors (Prog), sorted as B220 þ IgMÀ cells from the bone marrow cells of a pool of 2-month-old control C57B/6 mice and in sorted leukemic blasts (large B220 þ IgMÀ cells) obtained from the bone marrow of BCR-ABL þ /0;IKþ / þ and BCR-ABL þ /0;IKL/ þ mice. The migration positions of the DNA-binding-competent IK1 and IK2 proteins are shown as black arrowheads. An extract obtained from spleen cells from IKL/L mice identify the IK1/2n products (open arrowheads) encoded by the mutant IKL allele.6 IKAROS proteins were identified using an antibody specific of all IKAROS isoforms (E-20; Santa Cruz Biotechnologies, Santa Cruz, CA, USA) and Pax5 was identified using a rat monoclonal antibody to amino acids 154–284 of mouse Pax5 (clone 1H9; a generous gift from Dr M Busslinger, IMP, , ). Western blots analyses were normalized using an antibody specific to b-actin (clone AC-15; Sigma, St Louis, MO, USA).

median value 28% and range 8–60% in BCR-ABL þ /0;IKL/ þ Western blot analysis failed to detect any change in Pax5 expres- tumors), indicating a block at the pro-B to pre-B stage of sion in leukemic cells obtained from our BCR-ABL transgenic development. cohorts irrespective of their IKAROS status, as compared with Western blot analyses showed that sorted BCR-ABL þ /0; normal B-cell progenitors (Figure 1d). We conclude from these IK þ / þ leukemic cells expressed the DNA-binding-competent experiments that loss of a single wt IKZF1 allele is sufficient to IK1 and IK2 IKAROS isoforms at levels indistinguishable from cooperate strongly with BCR-ABL in B-cell leukemogenesis. that of normal B-cell progenitors, with no evidence for expres- PCR analysis of the VH-to-DJH immunoglobulin heavy chain sion of short IKAROS isoforms (Figure 1d; Supplementary (IgH) locus rearrangement using 50 primers specific for distal Figure 1). This shows that the natural course of leukemia (VHJ558) or proximal (VHGam3.8, VHQ52, VH7183) VH gene progression in BCR-ABL þ /0;IKþ / þ mice is not associated with families (see Figure 2a, for a scheme) and a 30 primer located impaired expression of IKAROS or with induction of the downstream of the JH4 segment detected dominant clonal expression of DN IKAROS isoforms. BCR-ABL þ /0;IKL/ þ rearrangements in BCR-ABL þ /0;IKþ / þ leukemias (Figure 2b). leukemic cells showed the expected 50% decrease in expression In contrast, the fast arising BCR-ABL þ /0;IKL/ þ leukemias in IK1/2 and the expression at low levels of the N-terminally showed a high degree of oligoclonality/polyclonality as all shortened IK1/2n polypeptides encoded by the hypomorphic tumors analyzed recurrently showed the rearrangement of 6 IKAROS allele, but no expression of DN IKAROS isoforms. This several of the VH family members investigated, each rearranged indicates that the experimentally imposed impaired function of to all possible JH 1–4 segments (Figure 2b). No rearrangement of þ / þ L/ þ þ IKZF1 does not favor the inactivation of the remaining wt allele the Vk locus was observed in either IK or IK BCR-ABL during leukemia progression. Hemizygous deletion of PAX5 is leukemias, consistent with their similar differentiation arrest observed in about 50% of BCR-ABL-positive B-ALL patients.3 at the pro-B/pre-B differentiation stage. These data suggest that

Leukemia Letters to the Editor 1202 J558 Gam3.8 Q52 7183 D J with complete or partial duplication of chromosomes 9, 10, 12 IgH or 18 (tumors 1370, 1271, 1330, 1309, 1399) (Figure 3; distal proximal Supplementary Table 1; Supplementary Figure 2). In contrast, BCR-ABL þ /0;IKL/ þ leukemic cells clearly showed fewer chromosomal abnormalities. A single tumor (1419) showed duplication of both chromosomes 14 and 17, with no additional þ /0 L/ þ BCR-ABL; BCR-ABL; numerical abnormality. The other BCR-ABL ;IK leukemias IK+/+ IKL/+ showed the acquisition of either a full (two of six tumors), or most often (four of six tumors) a partial gain of one of these chromosomes (Supplementary Table 1). Furthermore, except for - 1443 1466 1316 1396 1414 1455 1469 B220+ Splenic cells DP Thymocytes tumor 1432 (which showed a partial duplication of chromo- þ þ 1 some 12; Supplementary Table 1), the other BCR-ABL /0;IKL/ 2 * tumors showed no numerical abnormalities in chromosomes 9, 3 10, 12 and 18, or in other chromosomes. No recurrent focal VHJ558 gains or losses of genetic material were observed in any of these

4 leukemias. The fact that only partial duplication of the chromosomes recurrently duplicated in IK þ / þ tumors is 1 L/ þ 2 observed in IK tumors is consistent with the oligoclonal * þ /0 L/ þ 3 nature of BCR-ABL ;IK leukemias. The quantitative and VH7183 qualitative reductions of chromosomal numerical abnormalities * in IKAROS-deficient versus BCR-ABL þ /0;IKþ / þ leukemias 4 indicate that impaired IKAROS function facilitates leukemia 1 progression and accelerates tumor onset by bypassing the 2 requirement to select for and accumulate specific additional 3 pro-oncogenic genetic hits that cooperate with BCR-ABL in V Q52 H B-cell leukemogenesis. Therapies targeting the tyrosine kinase domain of BCR-ABL, 4 for example, imatinib, are the first-line treatment of CML 1 2 patients and show remarkable success in the chronic phase of 3 the disease. These treatments do not induce lasting remission for VHGam3.8 patients in blast crisis and for patients with BCR-ABL-associated * B-ALL. Imatinib administered in combination with chemo- 4 therapy during the induction or consolidation phase is asso- 1 ciated with improved clinical responses, but relapse and drug 2 resistance are common (for review, see Gruber et al.9). Drugs κ κ targeting either signaling molecules downstream of BCR-ABL, or V -J 4 molecular pathways that synergize with BCR-ABL in leukemia 5 maintenance, will likely offer additional therapeutic options. The mouse model described here will be an important tool to Cμ identify and dissect the role of molecular pathways deregulated *Unspecific background bands by impaired IKAROS function in BCR-ABL-induced B-ALL. Figure 2 PCR analysis of IgH locus VDJ rearrangements in BCR-ABL leukemias. (a) Scheme of the organization of the VH gene family clusters in the mouse IgH locus, identifying VH family genes used in Conflict of interest this study that are distal (J558) or proximal (Gam3.8, Q52, 7183) to the diversity (DH) and joining (JH) segments. (b) PCR analysis of the VH-to- þ /0 þ / þ The authors declare no conflict of interest. DJH,Vk-to-Jk and Cm rearrangements in sorted BCR-ABL ;IK and BCR-ABL þ /0;IKL/ þ leukemic blasts. Sorted B220 þ IgM þ splenic B cells and CD4 þ CD8 þ thymocytes were used as positive and negative controls, respectively. The primers used and PCR conditions Acknowledgements were as described previously.10 We thank Meinrad Busslinger for discussions and the gift of anti- PAX5 antibody; Zofia Maciorowski, Annick Viguier and Niraja Shah for help with FACS analyses; Philippe Hupe´, Philippe La BCR-ABL-induced leukemia progression in IK þ / þ mice requires Rosa and Ste´phane Liva for VAMP analyses of CGH array data; the selection of additional oncogenic events that ultimately Josiane Ropers for assistance in mouse husbandry; and Maryvonne result in the emergence of a single or few dominant leukemic Williame for help with mouse genotyping. This work was clone(s). Impaired IKAROS function in BCR-ABL þ /0;IKL/ þ mice supported by funds from Centre National de la Recherche alleviates this requirement, resulting in the emergence of dozens Scientifique (CNRS), Institut National de la Sante´ et de la of leukemic clones to generate oligoclonal/polyclonal tumors. Recherche Me´dicale (INSERM), Institut Curie, Institut National In line with this notion, comparative genomic hybridization du Cancer (INCA, Grant no. PLBIO08-092), Ligue Nationale (CGH) array analysis showed that leukemia progression contre le Cancer (JG and SC/PK are Equipe labelise´e Ligue contre in BCR-ABL þ /0;IKþ / þ mice is associated with recurrent le Cancer). SM was supported by a post-doctoral fellowship from genomic abnormalities, in particular the acquisition of one the Ligue Contre le Cancer and CV by pre-doctoral fellowships additional copy of chromosomes 14 and 17 (two additional from the Ministe`re de l’Education Nationale and Association copies of chromosome 14 for tumor 1309), often associated contre le Cancer (ARC). Research in CC laboratory was supported

Leukemia Letters to the Editor 1203 BCR-ABL; IKL/+ BCR-ABL; IK+/+

4.5 4.5 3.0 3.0 1422 1.5 1330 1.5 0.0 0.0 -1.5 -1.5 -3.0 -3.0 1 2 3 4 5 6 7 8 9 10 1112 13 14 1516171819 X Y 1 2345 6 7 891011 1213 14 1516171819 XY

4.5 4.5 3.0 3.0 1397 1.5 1394 1.5 0.0 0.0 -1.5 -1.5 -3.0 -3.0 1 2345678 910111312 14 1516171819 X Y 1 2 3 4 5 6 7 8 9 10 1112 13 14 1516171819 X Y

4.5 4.5 3.0 3.0 1384 1.5 1370 1.5 0.0 0.0 -1.5 -1.5 -3.0 -3.0 1 2 3 4567 8910 1112 13 14 151617 1819 X Y 1 2 3 4 5 6 7 8 9 10 1112 13 14 1516171819 X Y

4.5 4.5 3.0 3.0 1400 1.5 1271 1.5 0.0 0.0 -1.5 -1.5 -3.0 -3.0 1 234 5 6 7 8910 1112 13 14 15 16171819 XY 1 2345 6 7 891011 1213 14 151617 1819 XY

4.5 4.5 3.0 3.0 1419 1.5 1309 1.5 0.0 0.0 -1.5 -1.5 -3.0 -0.0 1 234 5 6 7 8 9 10 1112 13 1415 16 171819 X Y 1 2345 6 7 891011 1213 14 1516171819 XY

4.5 4.5 3.0 3.0 1408 1.5 1281 1.5 0.0 0.0 -1.5 -1.5 -3.0 -3.0 1 23456789 10 11 12 13 1415 16 171819 XY 1 2345 6 7 891011 1213 14 1516171819 XY

4.5 4.5 3.0 3.0 1432 1.5 1348 1.5 0.0 0.0 -1.5 -1.5 -3.0 -3.0 1 234 5 6 7 8 9 10 1112 13 14 1516 171819 X Y 1 2345 6 7 891011 1213 14 1516171819 XY

Figure 3 Array-based CGH ideograms of seven BCR-ABL þ /0;IKþ / þ and seven BCR-ABL þ /0;IKL/ þ leukemias. DNA was extracted from FACS- sorted BCR-ABL þ /0;IKþ / þ (tumors 1281, 1370, 1271, 1330, 1348, 1309, 1394) and BCR-ABL þ /0;IKL/ þ (tumors 1432, 1422, 1408, 1397, 1384, 1419, 1400) leukemias and processed for hybridization to 244K Whole Mouse Genome Chip (G4122A; Agilent Technologies, Massy, France), according to the manufacturer’s instructions. Arrays were scanned with an Agilent scanner and analyzed with the Agilent Feature Extraction software. After processing on the Genomic Workbench system 5.0 (Agilent Technologies), normalized array-CGH values were used to calculate the median ratio at each individual chromosome for each leukemia. Graphical representation of the data used the Visualization and Analysis of Array-CGH, transcriptome and other Molecular Profiles (VAMP; http://bioinfo-out.curie.fr/vamp/doc/).11–13 Chromosomes are listed in numerical order from left to right. Red squares above zero on the y axis represent genomic gains relative to the diploid control. The yellow lines indicate the absence of numerical alterations relative to the diploid control. X chromosome profiles should be discounted, as some leukemias were derived from female mice and the reference DNA was obtained from the liver of a wild-type C57BL/6 male mouse.

10 by FEDER (FIS PI080164), CSIC and Junta de Castilla y Leon CNRS UMR 7104, Institut de Ge´ne´tique et de Biologie (SA060A09 and proyecto Biomedicina). Mole´culaire et Cellulaire, Illkirch, France; 11Department of Cancer Biology, Institut de Ge´ne´tique et de 1,2,3,4,13 1,2,3,13 5 1,2,3 Biologie Mole´culaire et Cellulaire, Universite´ de Strasbourg, C Virely , S Moulin , C Cobaleda , C Lasgi , Strasbourg, France and A Alberdi6,7, J Soulier6,7, F Sigaux6,7, S Chan8,9,10,11, 12Faculte´ de Me´decine, Universite´ de Strasbourg, Strasbourg, P Kastner8,9,10,11,12 and J Ghysdael1,2,3 France 1Centre de Recherche, Institut Curie, E-mail: [email protected] 13 Centre Universitaire Bat 110, Orsay, France; These authors contributed equally to this work. 2CNRS UMR 3306, Centre Universitaire Bat 110, Orsay, France; 3INSERM U1005, Centre Universitaire Bat 110, Orsay, France; 4Ecole Doctorale B2T, Universite´ Paris–Diderot, Paris, France; 5 Area de Desarrollo y Diferenciacion, Centro de Biologia References Molecular ‘Severo Ochoa’, CSIC–Universidad Autonoma de Madrid, Madrid, Spain; 6 1 Nutt SL, Kee BL. The transcriptional regulation of B cell lineage INSERM U944, Institut Universitaire d’He´matologie, Hopital commitment. Immunity 2007; 26: 715–725. Saint Louis, Paris, France; 7 2 Olivero S, Maroc C, Beillard E, Gabert J, Nietfeld W, Chabannon C Institut Universitaire d’He´matologie, Universite´ Paris–Diderot, et al. Detection of different Ikaros isoforms in human leukaemias Paris, France; 8 using real-time quantitative polymerase chain reaction. Br J Department of Cancer Biology, Institut de Ge´ne´tique et de Haematol 2000; 110: 826–830. Biologie Mole´culaire et Cellulaire, Illkirch, France; 3 Mullighan CG, Miller CB, Radtke I, Phillips LA, Dalton J, Ma J et al. 9 INSERM U 964, Institut de Ge´ne´tique et de Biologie BCR-ABL1 lymphoblastic leukaemia is characterized by the Mole´culaire et Cellulaire, Illkirch, France; deletion of Ikaros. Nature 2008; 453: 110–114.

Leukemia Letters to the Editor 1204 4 Mullighan CG, Su X, Zhang J, Radtke I, Phillips LA, Miller CB et al. 9 Gruber F, Mustjoki S, Porkka K. Impact of tyrosine kinase inhi- Deletion of IKZF1 and prognosis in acute lymphoblastic leukemia. bitors on patient outcomes in Philadelphia chromosome-positive N Engl J Med 2009; 360: 470–480. acute lymphoblastic leukaemia. Br J Haematol 2009; 145: 5 Heisterkamp N, Jenster G, ten Hoeve J, Zovich D, Pattengale PK, 581–597. Groffen J. Acute leukaemia in bcr/abl transgenic mice. Nature 10 Cobaleda C, Jochum W, Busslinger M. Conversion of mature B 1990; 344: 251–253. cells into T cells by dedifferentiation to uncommitted progenitors. 6 Kirstetter P, Thomas M, Dierich A, Kastner P, Chan S. Ikaros is Nature 2007; 449: 473–477. critical for B cell differentiation and function. Eur J Immunol 2002; 11 Liva S, Hupe´ P, Neuvial P, Brito I, Viara E, La Rosa P et al. 32: 720–730. CAPweb: a bioinformatics CGH array analysis platform. 7 Dumortier A, Jeannet R, Kirstetter P, Kleinmann E, Sellars M, Nucl Acids Res 2006; 34: W477–W481. dos Santos NR et al. Notch activation is an early and critical event 12 La Rosa P, Viara E, Hupe´ P, Pierron G, Liva S, Neuvial P et al. during T-Cell leukemogenesis in Ikaros-deficient mice. Mol Cell VAMP: visualization and analysis of array-CGH, transcriptome and Biol 2006; 26: 209–220. other molecular profiles. Bioinformatics 2006; 22: 2066–2073. 8 Sellars M, Reina-San-Martin B, Kastner P, Chan S. Ikaros controls 13 Hupe´ P, Stransky N, Thiery JP, Radvanyi F, Barrillot E. Analysis of isotype selection during immunoglobulin class switch recombina- array CGH data: from signal ratio to gain and loss of DNA regions. tion. J Exp Med 2009; 206: 1073–1087. Bioinformatics 2004; 20: 3413–3422.

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

WT1 protein is a transcriptional activator of the antiapoptotic bag3 gene

Leukemia (2010) 24, 1204–1206; doi:10.1038/leu.2010.68; WT1(+KTS), containing an extra three aminoacids (KTS) published online 22 April 2010 between the third and fourth zinc fingers; WT1(ÀKTS) appears to exert its effect mainly as a transcriptional factor, whereas WT1( þ KTS) is involved in RNA processing.1 To investigate the transcriptional modulation of bag3 promoter by WT1, we WT1 gene, originally identified as a tumor suppressor involved introduced a luciferase reporter plasmid containing the bag3 in the formation of Wilms’ tumor of the kidney, was promoter into HEK293 cells and analyzed luciferase levels in the subsequently described to have an oncogenic role in a variety presence of increasing amounts of WT1(ÀKTS) or WT1( þ KTS) of tumors from different origins, including leukemias.1 In expression plasmids. As shown in Figure 1b, the ‘transcriptional comparison to normal progenitor cells, it is overexpressed in isoform’, WT1(ÀKTS), enhanced bag3 promoter activity in a acute lymphoblastic and myeloblastic leukemia, and in the dose-dependent manner, whereas transfection of the ‘post- blast crisis phase of chronic myelogenous leukemia;1,2 high transcriptional isoform’, WT1( þ KTS), did not influence the levels of the protein are associated with a poor response transcriptional activity. to therapy.1,3 WT1 knockdown by antisense oligonucleotides or To assess whether forced expression of WT1 isoforms could RNA interference was shown to induce apoptosis; conversely, modulate endogenous bag3, we transfected HEK293 cells with its overexpression in myeloid leukemia cells protected against increasing concentrations of the expression vectors encoding cell death. Modulation of some members of the bcl-2 family has WT1(ÀKTS) or WT1( þ KTS) and evaluated bag3 mRNA and been associated with apoptosis inhibition by WT1.1,3 BAG3 protein levels by quantitative real-time-PCR and western Among proteins that regulate apoptosis in leukemia cells, blot analyses, respectively. We observed a progressive increase a role is assigned to BAG3, a member of the family of proteins in the levels of bag3 RNA and BAG3 protein in cells transfected that, through their BAG domain, interact with HSC70/HSP70 with increasing amount of WT1(ÀKTS), and not in those heat shock proteins. bag3 is constitutive in transfected with WT1( þ KTS) (Figures 1c–d), confirming the some tumor types, including leukemias; in these cells, BAG3 transcriptional activation of endogenous bag3 gene by WT1. protein has been shown to sustain cell survival and down- To provide a further argument for bag3 regulation by WT1, modulate cell apoptotic response to drugs, by either HSP70- we evaluated bag3 expression in K562 cells following dependent or -independent mechanisms.4–7 small interfering RNA (RNAi)-mediated knockdown of WT1. Here we report that WT1 induces bag3 gene expression. This As shown in Figure 1e, the levels of bag3 mRNA in WT1 knock- finding identifies a novel target of WT1 protein involved in down cells were decreased by about 40%; western blotting apoptosis regulation in leukemia cells. confirmed downmodulation of BAG3 protein. As using in silico analysis we had found two putative WT1 Altogether, these observations ascribe a role to WT1 in bag3 binding sites on the bag3 promoter, we decided to investigate gene regulation. whether WT1 is directly recruited onto the bag3 promoter using We next evaluated the effect of WT1 silencing on apoptosis. a chromatin immunoprecipitation assay (ChIP) in K562 cells, In K562 cells treated with the proapoptotic agent phenethyl which express significant amounts of endogenous WT1 and isothiocyanate (PEITC), WT1 silencing significantly enhanced BAG3. Chromatin was immunoprecipitated with anti-WT1 anti- apoptosis, as assessed by flow cytometry, with respect to control bodies. Subsequent PCR analysis, performed using oligonucleo- scrambled RNA (Figure 2a). Conversely, BAG3 protein appeared tides covering the putative WT1 binding sites, revealed that to exert an antiapoptotic effect against PEITC treatment, WT1 bound to the bag promoter sequences (350-bp band), since its overexpression obtained by cell transduction with a whereas rabbit IgG antibody controls did not (Figure 1a). WT1 bag3 cDNA-carrying adenovirus (AdhBAG3) downmodulated protein has two major isoforms, designated WT1(-KTS) and PEITC-induced apoptosis (Figure 2b). We therefore investigated

Leukemia