Transposon Mutagenesis Reveals Cooperation of ETS Family Transcription Factors with Signaling Pathways in Erythro-Megakaryocytic Leukemia

Transposon mutagenesis reveals cooperation of ETS family transcription factors with signaling pathways in erythro-megakaryocytic leukemia

Jian Zhong Tanga, Catherine L. Carmichaela,b, Wei Shia,c, Donald Metcalfa,b,1, Ashley P. Nga,b, Craig D. Hylanda, Nancy A. Jenkinsd, Neal G. Copelandd, Viive M. Howelle, Zhizhuang Joe Zhaof, Gordon K. Smytha,g, Benjamin T. Kilea,b, and Warren S. Alexandera,b,1

aThe Walter and Eliza Hall Institute of Medical Research, Parkville, VIC 3052, Australia; Departments of bMedical Biology, cComputing and Information Systems, and gMathematics and Statistics, University of Melbourne, Parkville, VIC 3010, Australia; dMethodist Hospital Research Institute, Houston, TX 77030; eKolling Institute of Medical Research, University of Sydney, St. Leonards, NSW 2065, Australia; and fDepartment of Pathology, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73104

Contributed by Donald Metcalf, March 4, 2013 (sent for review January 30, 2013) To define genetic lesions driving leukemia, we targeted cre- SB transposase allele (3) with mice expressing cre recombinase dependent Sleeping Beauty (SB) transposon mutagenesis to the from the hematopoietic-selective vav 1 oncogene (vav1) promoter blood-forming system using a hematopoietic-selective vav1on- (4). Leukemias of diverse cellular origins developed in the ensuing cogene (vav1) promoter. Leukemias of diverse lineages ensued, offspring, most commonly erythroleukemia and lymphoid leuke- most commonly lymphoid leukemia and erythroleukemia. The mia. In the presence of a transgene expressing Janus kinase 2 inclusion of a transgenic allele of Janus kinase 2 (JAK2)V617F (JAK2)V617F (5), an active form of JAK2 associated with hema- resulted in acceleration of transposon-driven disease and strong tological disease (6), accelerated transposon-driven disease ensued selection for erythroleukemic pathology with transformation of with selection for erythroleukemic pathology characterized by bipotential erythro-megakaryocytic cells. The genes encoding the transformation of bipotential erythro-megakaryocytic cells. The E-twenty-six (ETS) transcription factors Ets related gene (Erg) and genes encoding the E-twenty-six (ETS) transcription factors Ets Ets1 were the most common sites for transposon insertion in related gene (Erg) and Ets1 were common sites for transposon MEDICAL SCIENCES SB-induced JAK2V617F-positive erythroleukemias, present in insertion (CIS) in the JAK2V617F-positive leukemias. We validated 87.5% and 65%, respectively, of independent leukemias examined. the role of Erg by reproducing the erythroleukemic pathology in The role of activated Erg was validated by reproducing erythroleu- mice transplanted with hematopoietic cells expressing translocated kemic pathology in mice transplanted with fetal liver cells express- in liposarcoma (TLS)-ERG,anactivatedformofERG found in ing translocated in liposarcoma (TLS)-ERG, an activated form of human leukemia (7). SB mutagenesis in TLS-ERG–induced leu- ERG found in human leukemia. Via application of SB mutagenesis kemia identified Jak2 as a CIS, together with other loci expressing to TLS-ERG–induced erythroid transformation, we identified mul- regulators of signal transduction, suggesting that a key mechanism tiple loci as likely collaborators with activation of Erg. Jak2 was in erythroleukemia may be the collaboration of lesions disturbing fi – identi ed as a common transposon insertion site in TLS-ERG in- erythroid maturation, most notably in genes of the ETS family, with duced disease, strongly validating the cooperation between mutations that reduce dependence on exogenous signals. JAK2V617F and transposon insertion at the Erg locus in the JAK2V617F-positive leukemias. Moreover, loci expressing other Results regulators of signal transduction pathways were conspicuous Mice homozygous for both a transposon array and a cre-inducible – among the common transposon insertion sites in TLS-ERG SB transposase allele (T2/Onc2/T2/Onc2;RosaSBLSL/RosaSBLSL) driven leukemia, suggesting that a key mechanism in erythro- (3) were mated with mice hemizygous for independent transgenes leukemia may be the collaboration of lesions disturbing ery- allowing expression of cre (4) and JAK2V617F (5) from the vav1 throid maturation, most notably in genes of the ETS family, promoter (vav-cre/+;JAK2V617F/+; Fig. S1A). Mice of each of the with mutations that reduce dependence on exogenous signals. four resultant genotypes [transposon-active, with or without the JAK2V617F transgene (T2Onc2/+;SBLSL/+;vav-cre/+;JAK2V617F/+ he study of cancer-causing genes using insertional muta- and T2Onc2/+;SBLSL/+;vav-cre/+,referredtoasJAK2SBvav and Tgenesis in the mouse (1, 2) provides a powerful complement SBvav), as well as transposon-inactive controls (T2Onc2/+;SBLSL/+; to human cancer genomics for functional identification and char- JAK2V617F/+ and T2Onc2/+;SBLSL/+,referredtoasJAK2and acterization of the genetic lesions driving tumor development. SBLSL)] were weaned in numbers expected from normal Mendelian Transposons are DNA elements with the unique capacity to change segregation of alleles (JAK2SBvav:SBvav:JAK2:SBLSL = 79:69:70:75). their genomic position, usually via expression of a transposase, an To examine the efficiency of activation of SB transposase, we enzyme that catalyzes excision of the transposon from the genome analyzed expression of GFP, which is present before excision but and facilitates its reintegration elsewhere. Although most mam- deleted upon cre-mediated recombination of the conditional al- mals lack active endogenous transposons, the Sleeping Beauty (SB) lele (3). Loss of GFP positivity was observed in all of six SBvav system has been developed recently to adapt the well-studied fish Tc/mariner transposon to allow insertional mutagenesis in the mouse (2). Temporally controlled or tissue-specific SB-mediated Author contributions: J.Z.T., C.L.C., W.S., D.M., A.P.N., G.K.S., B.T.K., and W.S.A. designed genetic screens have been designed by imposing cre-recombinase research; J.Z.T., C.L.C., W.S., D.M., A.P.N., C.D.H., G.K.S., and W.S.A. performed research; control over the expression of the transposase (2). Moreover, by N.A.J., N.G.C., V.M.H., and Z.J.Z. contributed new reagents/analytic tools; J.Z.T., C.L.C., W.S., D.M., A.P.N., C.D.H., G.K.S., B.T.K., and W.S.A. analyzed data; and J.Z.T., C.L.C., and W.S.A. activating transposition in mice carrying a predisposing germ-line wrote the paper. genetic lesion, insertional mutations that cooperate with that par- The authors declare no conflict of interest. fi ticular lesion in cancer pathogenesis can be isolated speci cally. 1To whom correspondence may be addressed. E-mail: [email protected] or alexandw@ To define genetic lesions driving leukemia, we targeted SB trans- wehi.edu.au. poson mutagenesis to the blood-forming system by intercrossing This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. mice transgenic for both a transposon array and a cre-inducible 1073/pnas.1304234110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1304234110 PNAS Early Edition | 1of6 Downloaded by guest on October 2, 2021 mice analyzed. In two of these mice, recombination was evident in displayed erythroleukemic pathology, occasionally coincident with effectively all hematopoietic cells examined whereas, in the oth- lymphoid disease (Fig. 1B) ers, efficiency varied from 30% to 80%. Although interanimal In SBvav mice, the lymphoid leukemias typically involved the variation was observed, in any individual mouse, the efficiency of bone marrow, spleen, thymus, and lymph nodes (Fig. S2), with or recombination was consistent among hematopoietic cells of dif- without a significant elevation in circulating white blood cells. ferent lineages and maturation stages (Fig. S1B). The disease displayed a consistent histological appearance, with affected organs containing tightly packed blast cells interspersed JAK2V617F Accelerates SB-Induced Erythroleukemia. As expected, with phagocytic macrophages and often the appearance of free SBLSL mice displayed no significant illness and, whereas the aggregates of leukemic cells within the peritoneal cavity fatty presence of the JAK2V617F transgene has been shown to cause tissue. Flow cytometric analysis confirmed the lymphoid origin of increased numbers of circulating blood cells (5), these mice also disease and established that the leukemic cells were T-lymphoid, remained healthy (Fig. 1A). In contrast, activation of SB trans- either CD4/CD8 double negative or single positive. position in SBvav mice resulted in disease in all mice within 12 mo, In both SBvav and JAK2SBvav mice, the erythroleukemic pa- with 50% affected within 100 d. Consistent with activation of thology included consistently high peripheral blood nucleated cell transposition in multiple hematopoietic lineages, histopathologi- counts dominated by nucleated erythroid cells, which were often cal analysis revealed examples of lymphoid, myeloid, megakaryo- accompanied by thrombocytopenia. The mice typically presented cytic, and erythroid disease in SBvav mice although the majority of with a grossly enlarged spleen dominated by a population of im- mice developed lymphoid malignancy, erythroleukemic pathology, mature nucleated erythroid cells and little surviving normal spleen or both (Fig. 1B). Strikingly, the presence of the JAK2V617F architecture. The bone marrow was usually infiltrated by similar transgene resulted in significantly accelerated disease: all but 1 of erythroid cells, as was the liver, with hepatomegaly resulting from 63 JAK2SBvav mice succumbed to disease within 70 d, and most the uniform distribution of these cells throughout the organ. Thickening of the lung alveolar walls by infiltrating nucleated erythroid cells was also commonly observed (Fig. 1C and Fig. S3).

JAK2SBvav Leukemia Is Transplantable. Suspensions of bone marrow and/or spleen cells from five erythroleukemic JAK2SBvav mice were injected into sublethally irradiated recipient mice. All recipients succumbed to disease 13–31 d after transplantation (Table S1), and, in all cases examined, leukemia was evident, with the same characteristics as that observed in primary mice: thrombocytopenia and high nucleated blood cell counts, as well as splenomegaly and hepatomegaly resulting from infiltration of immature nucleated erythroid cells. In methylcellulose cultures, spleen cells from primary JAK2SBvav leukemias failed to generate colonies in the absence of cytokine stimulation and produced small numbers of colonies in the presence of stem cell factor (SCF) and/or IL-3. Larger numbers of colonies developed in cultures containing erythropoietin (EPO), suggesting that the leukemic cells were primarily dependent on EPO, and the combination of EPO with SCF and/or IL-3 yielded maximal colony numbers (Table S2). A similar result was observed in cultures of secondary leukemias: whereas in three of five cultures some colonies developed in the absence of exogenous cytokine, the majority of clonogenic cells remained cytokine dependent (Table S2). The colonies observed were large and compact (Fig. 2A) and, when picked and stained, were remarkably observed to contain nucleated erythroblasts and acetylcholinesterase-staining megakaryocytic cells (Fig. 2A). The presence of both erythroblasts and megakaryocytic cells was confirmed by flow cytometric analysis of cells from individual colonies. All cells in each colony were + − CD71 Ter119 /lo, and a proportion also expressed CD41 (Fig. 2B). Individual colonies contained varying numbers of megakaryocytic cells, from just a few percent up to the majority of cells (Fig. 2C), but this observation was a consistent feature of almost all leukemias examined, suggesting that the transformed cells in JAK2SBvav leukemias were in fact bipotential. Thus, although the megakaryocytic component of the leukemias was not readily evident in most histological sections, based on these observations, Fig. 1. (A) Survival of SBLSL (n = 23), JAK2 (n = 27), SBvav (n = 48), and we classified the disease as erythro-megakaryocytic leukemia. JAK2SBvav (n = 63) mice. Time indicates age at intervention due to illness. P < vav vav 0.0001 for comparison of JAK2SB mice with all other genotypes and SB Common Transposon Insertion Sites in JAK2SBvav Leukemia. The trans- mice with SBLSL controls. (B) Spectrum of leukemias developing in SBvav and vav vav poson integration sites in spleen cells from 40 JAK2SB mice with JAK2SB mice. The sector size in the pie chart represents the proportion of erythro-megakaryocytic leukemia (7 with concurrent lymphoid mice that developed leukemia of the indicated cellular origin based on histopathological examination. In some mice, two distinct leukemia types leukemia) were isolated via linker-mediated PCR incorporating were evident. (C) Histological presentation of erythroleukemic pathology in barcoded primers and identified by pyrosequencing. CIS were representative JAK2SBvav mice showing infiltration of bone marrow and identified using bioinformatic strategies outlined in Materials and spleen with nucleated erythroid cells. Methods. From the 40 samples, a total of 247,502 sequence reads

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1304234110 Tang et al. Downloaded by guest on October 2, 2021 however, the lymphoid component was derived from uninfected − (GFP ) cells and likely reflected co-occurrence of TLS-ERG– induced erythroid disease and radiation-induced lymphoid leukemia. The leukemia in recipients of TLS-ERG–transduced fetal liver cells was notably similar to that observed in JAK2SBvav mice, with splenomegaly caused by excess immature nucleated erythroid cells, infiltration of the bone marrow, liver, and lungs by immature − erythroid cells, and an abnormal population of CD71 Ter119 /lo cells in the bone marrow, spleen, and blood of the affected mice. Expression of significant levels of CD41 was also observed in two of seven leukemias examined. To explore the multistep nature of leukemogenesis specifically involving activation of Erg, fetal liver cells carrying a transposon array and constitutively expressing SB transposase were infected with the TLS-ERG retrovirus. Consistent with cooperation between Erg activation and alleles affected by transposon insertion, disease developed significantly more rapidly in recipients of TLS- ERG–transduced SB fetal liver compared with recipients of TLS- ERG–transduced wild-type cells, or control GFP-transduced SB Fig. 2. (A) Morphology of colonies in methylcellulose cultures of spleen cells cells (Fig. 4B). Indicative of a particular role for Erg in driving from a representative JAK2SBvav mouse (Top). Cytocentrifuge preparations erythroid transformation, whereas control GFP-transduced SB from a single colony stained with megakaryocyte-specific acetylcholinester- cells yielded leukemias of diverse origins in recipient mice (six ase (AChE) plus Luxol fast blue/hematoxylin revealing characteristic red- lymphoid, three erythroid, one lymphoid/erythroid, and two stained megakaryocytes (Middle) or May Grunwald Giemsa showing ery- myeloid), all but one recipient of SB TLS-ERG cells developed throid cells (Bottom). (B) Flow cytometric profiles of cells pooled from + − eythroleukemic pathology indistinguishable from that observed JAK2SBvav colonies showing costaining of CD71 Ter119 /lo with CD41 in a proportion of cells. (C) Frequency of megakaryocytic cells in consecutively in recipients of TLS-ERG-transduced wild-type cells (Fig. S4). picked colonies from cultures of JAK2SBvav spleen cells. Individual colonies

+ MEDICAL SCIENCES contained varying numbers of AChE megakaryocytes, from a few percent vav up to the majority of cells, but this observation was a consistent feature in 12 Table 1. Common transposon insertion sites in JAK2SB of the 13 mice examined. erythroleukemic mice Gene ID Gene symbol* Chromosome Number†

were obtained, of which 112,888 (46%) were successfully mapped 13876 Erg chr16 35 to the mouse genome. 23871 Ets1 chr9 26 vav Twenty-three CIS were identified in the JAK2SB leukemias 330267 Thsd7a chr6 19 (Table 1). The most frequently identified were loci encoding 328365 Zmiz1 chr14 17 members of the ETS family of transcription factors: Erg and Ets1, 12661 Chl1 chr6 15 which were identified in 35 (87.5%) and 26 (65%) leukemias, 270035 Letm2 chr8 10 respectively. In 22 (55%) samples, transposon integration at both 93742 Pard3 chr8 8 Erg and Ets1 was evident. Analysis of the integration sites at the 18046 Nfyc chr4 6 Erg locus revealed a uniform transposon orientation and clus- 67299 Dock7 chr4 5 tering of insertion sites within introns 3 and 4 (Fig. 3A), consistent 29871 Scmh1 chr4 5 with expression of a transposon-Erg fusion transcript regulated by 434394 Gm5614 chr9 5 the T2/Onc LTR/splice donor elements (2) and reminiscent of 72555 Shisa9 chr16 5 the mode of ERG activation caused by intragenic translocations 208292 Zfp871 chr17 5 in prostate cancer and leukemia (7, 8). All of the transposon 94212 Pag1 chr3 4 integrations within the Ets1 gene clustered upstream of exon 1 or 21646 Tcte2 chr17 4 within the first intron and were uniformly oriented in a manner 22324 Vav1 chr17 4 indicative of transposon-activated expression of a full-length tran- 71458 Bcor chrX 4 script or a fusion transcript excluding the first exon (Fig. 3A). Of 209224 Enox2 chrX 4 note, in several individual leukemias, multiple transposon in- 331461 Il1rapl1 chrX 4 tegration sites were identified within Erg or Ets1, and this feature 83453 Chrdl1 chrX 4 may reflect ongoing transposition during leukemia development. 77031 Slc9a8 chr2 3 Loci previously implicated in hematopoietic malignancies were 236852 Magea10 chrX 3 also prominent, including Vav1, zinc finger, MIZ-type containing 1 20603 Sms chrX 3 (Zmiz1), and BCL6 interacting corepressor (Bcor)(Discussion). *Chl1, cell adhesion molecule with homology to L1CAM; Chrdl1, chordin-like 1; Dock7, dedicator of cytokinesis 7; Enox2, ecto-NOX disulfide-thiol ex- Identification of Cooperating Lesions in ERG-Driven Leukemia. To changer 2; Gm5614, predicted gene 5614; Il1rapl1, IL-1 receptor accessory validate the leukemia-causing potential of CIS identified in protein-like 1; Letm2, leucine zipper-EF-hand containing transmembrane JAK2SBvav mice, wild-type fetal liver cells were infected with protein 2; Magea10, melanoma antigen family A, 10; Nfyc, nuclear transcrip- a retrovirus expressing GFP plus TLS-ERG, a leukemia-derived tion factor-Y gamma; Pag1, phosphoprotein associated with glycosphingo- product of intragenic translocation of ERG, chosen to mimic the lipid microdomains 1; Pard3, partitioning defective 3 homolog; Scmh1, sex activation of Erg via intragenic transposon insertion, and then comb on midleg homolog 1; Shisa9, shisa homolog 9; Slc9a8, solute carrier family 9 member 8; Sms, spermine synthase; Tcet2, t-complex–associated transplanted into myeloablated recipient mice (Fig. 4A). Twenty- fi testis expressed 2; Thsd7a, thrombospondin type I domain containing 7A; ve of 27 transplanted mice developed disease (Fig. 4B), and, of 24 Zfp871, zinc finger protein 871. † mice examined, 22 displayed erythroleukemic pathology. Three of Number of samples from independent erythroleukemic mice with CIS at these mice also had evidence of a concurrent lymphoid leukemia; indicated locus, from a total of 40 analyzed.

Tang et al. PNAS Early Edition | 3of6 Downloaded by guest on October 2, 2021 locus has been studied extensively in prostate cancer, in which translocation of ERG to the TMPRSS2 locus and expression of a TMPRSS2-ERG fusion oncoprotein occurs in the majority of cases (8), ERG is also known to be translocated to the fused in sarcoma/translocated in liposarcoma (FUS/TLS) locus in rare cases of acute myeloid leukemia (AML). Moreover, increased expression of ERG is associated with poor prognosis in cytogenetically normal AML and T cell acute lymphoblastic leukemia (T-ALL) (7), and the ERG gene resides in the critical region of chromosome 21 that is associated with myeloproliferation and leukemia in Down syndrome (14, 15). Ets-1 is also important for normal hematopoiesis, being involved in the development of both lymphoid and myeloid cell lineages (16). Aberrant expression of ETS1 is implicated in solid tumor development and tumor angiogenesis (17). In hematological malignancy, rare cases of translocation involving ETS1 have been reported (18, 19), and ETS1 may mediate differentiation arrest in T-ALL (20). In addition, the Ets-1 homolog, v-ets, contributes to E26 avian acute leukemia virus-induced myelo-erythroid leukemia in chickens (21). Analysis of transposon insertion sites within the Erg locus in JAK2SBvav leukemias revealed, in all cases, intragenic, sense- oriented transposon integration, consistent with activation of Erg via expression of a transposon-Erg fusion transcript, reminiscent of that observed in the ERG intragenic translocations that char- acterize prostate cancer and AML (7, 8). To validate the role of Erg vav Fig. 3. (A and B) Arrowheads indicate the sites of transposon integration transposon-mediated activation of in JAK2SB leukemia, within the genes encoding the ETS family proteins Erg and Ets1 in JAK2SBvav we demonstrated that TLS-ERG, a human leukemia-derived fu- leukemias (A) and within Jak2 in TLS-ERG–induced leukemias (B), with the sion gene mimicking intragenic insertion within Erg, induced direction indicating transposon orientation. Exons are numbered and shown erythroleukemic pathology in recipients of transduced fetal liver as raised boxes. (C) CIS were examined for common cellular processes or cells. These data complement our previous study in which trans- pathways using a combination of GO term analysis and literature searching. duction of fetal liver cells with full-length Erg was also found to induce erythro-megakaryocytic leukemia (22). Of interest, un- − like full-length Erg, TLS-ERG did not induce development of A minority of SB TLS-ERG recipients developed GFP lymphoid or erythroid leukemias, indicative of SB-only–driven disease. Twenty-five CIS encompassing 28 genes were identified by analysis of 13 recipients of TLS-ERG–transduced SB cells, all of which showed erythroleukemic pathology (Table 2). The Jak2 locus was identified as a CIS in six independent samples (Fig. 3B), strongly validating the cooperation between JAK2V617F and transposon insertion at the Erg locus in JAK2SBvav leukemia. To explore pathways cooperating with Erg in erythroid transformation, CIS that co-occurred with Erg in JAK2SBvav mice or were identified in TLS-ERG–driven leukemias were examined for common cellular processes or pathways using a combination of Gene Ontology (GO) term analysis and literature searching. Transcription factors and epigenetic regulators of gene expression were present, as were genes involved in stem cell pluripotency and cell migration and adhesion; however, particular enrichment was observed for growth factors, receptors, and regulators of signal transduction (Fig. 3C and Discussion). Discussion The molecular changes associated with human acute erythroleukemia (AEL) are poorly characterized (9, 10). The leukemias in JAK2SBvav mice provided an opportunity to define molecular lesions driving erythroid transformation via identification of transposon integration sites. Of note, the two most frequently observed CIS in JAK2SBvav erythroid leukemias were loci encoding members of the ETS family of transcription factors Erg and Ets1, which were the targets for transposons in 87.5% and 65% Fig. 4. (A) SB transposon mutagenesis in TLS-ERG–driven leukemia. Mice ho- of the independent JAK2SBvav leukemias analyzed. ETS proteins mozygous for a T2/Onc transposon array were mated with mice constitutively are a family of over 20 helix–loop–helix domain transcription expressing SB transposase (RosaSB/RosaSB), and fetal liver cells were infected with retroviruses expressing TLS-ERG-GFP or control GFP and transplanted into factors with diverse biological roles, including important functions wild-type recipient mice. (B) Survival curves for recipients of WT-GFP (n = 14), in hematopoiesis and leukemia. In normal hematopoiesis, Erg is SB-GFP (n = 17), WT-TLS-ERG (n = 27), and SB-TLS-ERG (n = 34) fetal liver cells. required for hematopoietic stem cell self-renewal during times of Time indicates age at intervention due to illness. P < 0.0001 for comparison of high hematopoietic stem cell cycling (11–13). Whereas the ERG survival of SB-TLS-ERG mice with that of SB-GFP or WT-TLS-ERG cohorts.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1304234110 Tang et al. Downloaded by guest on October 2, 2021 Table 2. Common transposon insertion sites in TLS-ERG-induced although not identified as a CIS in JAK2SBvav leukemias, is acti- erythroleukemic mice vated via retroviral insertional mutagenesis in friend murine leu- † – Gene ID Gene symbol* Chromosome Number kemia virus (F-MuLV) induced erythroleukemias, also blocking erythroid differentiation (26). In over half the JAK2SBvav leuke- 12914 Crebbp chr16 6 mias, transposon insertion into both Erg and Ets1 was evident, 328572 Ep300 chr15 6 suggesting that co-occurrence of activation of these ETS family 16452 Jak2 chr19 6 proteins may serve to reinforce the maturation block that con- 26364 Cd97‡ chr8 5 tributes to erythroid transformation. In addition to Erg and Ets1, ‡ 68278 Ddx39 chr8 5 CISinJAK2SBvav leukemias were identified at several addi- ‡ 320795 Pkn1 chr8 5 tional loci with known links to hematological malignancies, in- 77044 Arid2 chr15 5 cluding Zmiz1, a gene fusion partner of Abl oncogene 1 (ABL1)(27), 225432 Rbm27 chr18 5 Bcor, a transcriptional corepressor mutated in AML (28), and Vav1, 19271 Ptprj chr2 4 commonly deregulated in lymphoid malignancy (29), suggesting that 14056 Ezh2 chr6 4 these genes might also contribute to erythroleukemia development. 215436 Slc35e3 chr10 4 To identify other loci contributing to erythroid transforma- 328365 Zmiz1 chr14 4 tion and to specifically explore pathways that cooperate with 213988 Tnrc6b chr15 4 activation of Erg in disease, the CIS identified in JAK2SBvav mice 56375 B4galt4 chr16 4 were supplemented with CIS in transposon-accelerated TLS- 11856 Arhgap6 chrX 4 ERG-induced leukemia. Whereas loci encoding sequence-specific 54598 Calcrl chr2 3 DNA binding transcription factors were not identified as CIS in 13418 Dnajc1 chr2 3 TLS-ERG leukemias, several epigenetic regulators of gene ex- 17329 Cxcl9 chr5 3 pression emerged, including CREB binding protein (Crebbp), 70178 Fam108c chr7 3 E1A binding protein p300 (Ep300), and enhancer of zeste 330814 Lphn1 chr8 3 homolog 2 (Ezh2), which have previously been linked to hemato- 234788 Slc38a8 chr8 3 logical malignancies (30, 31). Significantly, Jak2 was identified 11804 Aplp2 chr9 3 as a CIS in TLS-ERG–driven disease, providing strong valida- 217593 Slc25a21 chr12 3 tion for cooperation between deregulation of Erg and Jak2 in 66143 Eef1e1 chr13 3 erythroid transformation. Except for two loci, the CIS identified 26419 Mapk8 chr14 3 in JAK2SBvav and TLS-ERG–induced leukemias were mutually MEDICAL SCIENCES 18753 Prkcd chr14 3 exclusive, presumably reflecting the need for different leukemo- 353283 Eras§ chrX 3 genic secondary lesions in erythroid cells already expressing 15185 Hdac6§ chrX 3 deregulated Jak2 or Erg. Strikingly, classification of CIS in TLS- ERG–driven leukemias into specific pathways or functional classes *Aplp2, amyloid beta (A4) precursor-like protein 2; Arhgap6, Rho GTPase activating protein 6; Arid2, AT rich interactive domain 2; B4galt4, UDP-Gal: revealed a particular enrichment for loci expressing regulators of betaGlcNAc beta 1,4-galactosyltransferase, polypeptide 4; Calcrl, calcitonin cytokine/growth factor signal transduction pathways, with Jak2 receptor-like; Cd97, CD97 antigen; Cxcl9, chemokine (C-X-C motif) ligand 9; being just one of several such regulators identified, also observed to vav Ddx39, DEAD box polypeptide 39; Dnajc1, DnaJ homolog C1; Eef1e1, eukaryotic alesserextentinJAK2SB leukemias. Thus, accumulation of translation elongation factor 1 epsilon 1; Eras, ES cell-expressed Ras; Fam108c, multiple lesions in signaling pathways may contribute to the evo- family with sequence similarity 108; Hdac6, histone deacetylase 6; Lphn1, lution of disease. Indeed, analysis of primary JAK2SBvav leukemias latrophilin 1; Mapk8, mitogen-activated protein kinase 8; Pkn1, protein revealed few if any EPO-independent leukemic clones, but these kinase N1; Prkcd, protein kinase C, delta; Ptprj, protein tyrosine phospha- clones began to emerge in the leukemias developing in transplant tase receptor J; Rbm27, RNA binding motif protein 27; Slc25a21, solute recipients. Together, these results support a model in which ac- carrier family 25 member 21; Slc35e3, solute carrier family 35, member E3; Slc38a8, solute carrier family 38, member 8; Tnrc6b, trinucleotide repeat cumulation of lesions inhibiting erythroid maturation, most nota- containing 6b. bly in ETS family members, in combination with mutations that † Number of samples from independent erythroleukemic mice with CIS at reduce dependence on exogenous cytokine/growth factor signal- indicated locus, from a total of 13 analyzed. ing, commonly collaborate to drive erythroid transformation in ‡Associated with the same CIS on chromosome 8. mice. Consistent with this hypothesis, activation of the EPO re- §Associated with the same CIS on chromosome X. ceptor by spleen focus forming virus (SFFV) glycoprotein gp55 is accompanied by insertional activation of the Ets transcription factor SFFV proviral integration oncogene 1 (Spi1) in SFFV- lymphoid leukemia, and erythro-megakaryocytic leukemia cells vav induced erythroleukemia, and mutations in the gene encoding from JAK2SB mice were found to grow more aggressively in the SCF receptor, c-Kit, are commonly found in erythroleukemias culture than their full-length Erg counterparts. These observa- driven by transgenic expression of Spi1 (26). tions suggest subtle differences in the activity of full-length The rarity of AEL has contributed to the relative paucity of versus truncated Erg alleles. vav information on molecular changes associated with human dis- The TLS-ERG–induced leukemias, like those in JAK2SB ease. Aneuploidy and complex karyotype, particularly hypodip- mice with Erg and/or Ets1 insertions, were characterized by accu- loidy, are common in AEL and are associated with poor mulation of immature erythroid cells, a phenotype also observed in prognosis. Specific genetic changes, however, are poorly defined, the leukemias driven by full-length Erg (22). Expression of TLS- with fms-related tyrosine kinase 3 (FLT3) mutations in a small ERG in human cord blood progenitor cells has been shown to subset of AEL patients, very rare translocations involving JAK2, result in a block in erythroid differentiation (23) whereas expres- runt-related transcription factor 1 (RUNX1) mutations in two of sion of ERG in K562 cells induces down-regulation of erythroid a small sample of AELs, and P53 mutations among the few markers (15), and overexpression of Ets-1 blocks erythroid dif- molecular lesions defined (9, 10). Improved diagnosis, prognosis, ferentiation in human and mouse hematopoietic progenitor cells and treatment of AEL would benefit from an improved un- (24, 25). Together, these observations suggest that transposon derstanding of the genetic lesions underlying disease pathogen- insertion activates the Erg and Ets1 loci and contributes to eryth- esis. The suite of genes commonly targeted for transposon- roleukemia by interfering with erythroid differentiation. In this mediated mutagenesis in the murine erythroleukemias described regard, it is noteworthy that the ETS family member Fli-1, here provide a resource for informing future studies examining

Tang et al. PNAS Early Edition | 5of6 Downloaded by guest on October 2, 2021 genetic changes in human erythroleukemia, as well as a refined list cells from spleen or bone marrow of each leukemia-bearing donor, with or of candidates for systematically dissecting the functional contribu- without 5.5 Gy γ-irradiation. tion of genetic lesions to erythroleukemia in model systems. DNA Sequencing and Identification of Transposon Integration Sites. Genomic Materials and Methods DNA was extracted from the spleens of leukemic mice using the DNeasy Blood and Tissue Kit (QIAGEN). Enzymatic digestion of DNA and linker-mediated Mice. Mice homozygous for both a T2/Onc transposon array and a cre- PCR enrichment of transposon junctions were performed following published inducible (3) or constitutive (32) SB transposase allele and mice hemizygous protocols (35–37), with modifications as described in SI Materials and Methods. for transgenes allowing expression of cre (4) and JAK2V617F (5) from the vav1 Identification of common transposon insertion was based on published meth- promoter have been published previously. Animal experiments were ap- ods (38) as described in SI Materials and Methods. proved by the Walter and Eliza Hall Institute Animal Ethics Committee. For infection of fetal liver cells, virus supernatant was produced in 293T cells as ACKNOWLEDGMENTS. We thank Janelle Lochland, Jason Corbin, Sheree described previously (33) and cells were injected i.v. into mice receiving a γ Brown, Melanie Howell, Lauren Wilkins, and Keti Stoev for skilled assistance single dose of 5.5 Gy -irradiation. and Jelle ten Hoeve (Netherlands Cancer Institute) for making available the unpublished CIMPL software package. This work was supported by Program Hematological and Histopathological Analysis. Peripheral blood from the or- Grants 1016647 and 490037; fellowships (to W.S.A. and G.K.S.); Independent bital plexus was analyzed by an ADVIA 120 blood analyzer (Bayer). Clonal Research Institutes Infrastructure Support Scheme Grant 361646 from the agar cultures were performed as described (34). Methylcellulose cultures Australian National Health and Medical Research Council; the Carden Fellow- were incubated for up to 14 d at 37 °C in a mixture of 5% (vol/vol) CO in air ship (to D.M.) of the Cancer Council, Victoria; the Australian Cancer Research 2 Fund; Victorian State Government Operational Infrastructure Support; Cancer in Methocult M3234 (Stem Cell Technologies) supplemented with 100 ng/mL Institute New South Wales and Northern Translational Cancer Research Net- SCF, 10 ng/mL IL-3, and 4 IU/mL EPO. For histopathological studies, organs work Fellowships (to V.M.H.); Cure Cancer Australia Foundation and Cancer fi fi were xed in 10% (vol/vol) buffered formalin and paraf n embedded. Council New South Wales project grants; and a Cure Cancer Australia/Leukae- Sections were prepared and stained with hematoxylin and eosin. In trans- mia Foundation Australia Post Doctoral Fellowship and Lions Fellowship, Can- plantation assays, C57BL/6 recipients were injected via the tail vein with 107 cer Council of Victoria (to A.P.N.).

1. Kool J, Berns A (2009) High-throughput insertional mutagenesis screens in mice to 21. Metz T, Graf T (1991) v-myb and v-ets transform chicken erythroid cells and cooperate identify oncogenic networks. Nat Rev Cancer 9(6):389–399. both in trans and in cis to induce distinct differentiation phenotypes. Genes Dev 5(3): 2. Copeland NG, Jenkins NA (2010) Harnessing transposons for cancer gene discovery. 369–380. – Nat Rev Cancer 10(10):696 706. 22. Carmichael CL, et al. (2012) Hematopoietic overexpression of the transcription factor fi 3. Dupuy AJ, et al. (2009) A modi ed sleeping beauty transposon system that can Erg induces lymphoid and erythro-megakaryocytic leukemia. Proc Natl Acad Sci USA be used to model a wide variety of human cancers in mice. Cancer Res 69(20): 109(38):15437–15442. 8150–8156. 23. Pereira DS, et al. (1998) Retroviral transduction of TLS-ERG initiates a leukemogenic 4. Croker BA, et al. (2004) SOCS3 is a critical physiological negative regulator of G-CSF program in normal human hematopoietic cells. Proc Natl Acad Sci USA 95(14): signaling and emergency granulopoiesis. Immunity 20(2):153–165. – 5. Xing S, et al. (2008) Transgenic expression of JAK2V617F causes myeloproliferative 8239 8244. disorders in mice. Blood 111(10):5109–5117. 24. Lulli V, et al. (2006) Overexpression of Ets-1 in human hematopoietic progenitor cells 6. Scott LM (2011) The JAK2 exon 12 mutations: A comprehensive review. Am J Hematol blocks erythroid and promotes megakaryocytic differentiation. Cell Death Differ 86(8):668–676. 13(7):1064–1074. 7. Martens JH (2011) Acute myeloid leukemia: A central role for the ETS factor ERG. Int J 25. Marziali G, et al. (2002) Role of Ets-1 in erythroid differentiation. Blood Cells Mol Dis Biochem Cell Biol 43(10):1413–1416. 29(3):553–561. 8. Clark JP, Cooper CS (2009) ETS gene fusions in prostate cancer. Nat Rev Urol 6(8): 26. Moreau-Gachelin F (2008) Multi-stage Friend murine erythroleukemia: molecular in- 429–439. sights into oncogenic cooperation. Retrovirology 5:99. 9. Santos FP, Bueso-Ramos CE, Ravandi F (2010) Acute erythroleukemia: Diagnosis and 27. De Braekeleer E, et al. (2011) ABL1 fusion genes in hematological malignancies: A – management. Expert Rev Hematol 3(6):705 718. review. Eur J Haematol 86(5):361–371. 10. Zuo Z, Polski JM, Kasyan A, Medeiros LJ (2010) Acute erythroid leukemia. Arch Pathol 28. Grossmann V, et al. (2011) Whole-exome sequencing identifies somatic mutations of Lab Med 134(9):1261–1270. BCOR in acute myeloid leukemia with normal karyotype. Blood 118(23):6153–6163. 11. Ng AP, et al. (2011) Erg is required for self-renewal of hematopoietic stem cells during 29. Oberley MJ, Wang DS, Yang DT (2012) Vav1 in hematologic neoplasms, a mini review. stress hematopoiesis in mice. Blood 118(9):2454–2461. Am J Blood Res 2(1):1–8. 12. Loughran SJ, et al. (2008) The transcription factor Erg is essential for definitive he- 30. Shih AH, Abdel-Wahab O, Patel JP, Levine RL (2012) The role of mutations in epige- matopoiesis and the function of adult hematopoietic stem cells. Nat Immunol 9(7): – 810–819. netic regulators in myeloid malignancies. Nat Rev Cancer 12(9):599 612. 13. Taoudi S, et al. (2011) ERG dependence distinguishes developmental control of 31. Shima Y, Kitabayashi I (2011) Deregulated transcription factors in leukemia. Int J hematopoietic stem cell maintenance from hematopoietic specification. Genes Dev Hematol 94(2):134–141. 25(3):251–262. 32. Dupuy AJ, Akagi K, Largaespada DA, Copeland NG, Jenkins NA (2005) Mammalian 14. Ng AP, et al. (2010) Trisomy of Erg is required for myeloproliferation in a mouse mutagenesis using a highly mobile somatic Sleeping Beauty transposon system. Na- model of Down syndrome. Blood 115(19):3966–3969. ture 436(7048):221–226. 15. Rainis L, et al. (2005) The proto-oncogene ERG in megakaryoblastic leukemias. Cancer 33. Glaser SP, et al. (2012) Anti-apoptotic Mcl-1 is essential for the development and Res 65(17):7596–7602. sustained growth of acute myeloid leukemia. Genes Dev 26(2):120–125. 16. Bories JC, et al. (1995) Increased T-cell apoptosis and terminal B-cell differentiation 34. Alexander WS, Roberts AW, Nicola NA, Li R, Metcalf D (1996) Deficiencies in pro- – induced by inactivation of the Ets-1 proto-oncogene. Nature 377(6550):635 638. genitor cells of multiple hematopoietic lineages and defective megakaryocytopoiesis 17. Hahne JC, et al. (2008) The transcription factor ETS-1: Its role in tumour development in mice lacking the thrombopoietic receptor c-Mpl. Blood 87(6):2162–2170. and strategies for its inhibition. Mini Rev Med Chem 8(11):1095–1105. 35. Keng VW, et al. (2009) A conditional transposon-based insertional mutagenesis 18. Goyns MH, Hann IM, Stewart J, Gegonne A, Birnie GD (1987) The c-ets-1 proto- screen for genes associated with mouse hepatocellular carcinoma. Nat Biotechnol oncogene is rearranged in some cases of acute lymphoblastic leukaemia. BrJCancer – 56(5):611–613. 27(3):264 274. 19. Rovigatti U, Watson DK, Yunis JJ (1986) Amplification and rearrangement of Hu- 36. Largaespada DA, Collier LS (2008) Transposon-mediated mutagenesis in somatic cells: fi – ets-1 in leukemia and lymphoma with involvement of 11q23. Science 232(4748): Identi cation of transposon-genomic DNA junctions. Methods Mol Biol 435:95 108. fi 398–400. 37. Starr TK, et al. (2009) A transposon-based genetic screen in mice identi es genes 20. Dadi S, et al. (2012) TLX homeodomain oncogenes mediate T cell maturation arrest in altered in colorectal cancer. Science 323(5922):1747–1750. T-ALL via interaction with ETS1 and suppression of TCRα gene expression. Cancer Cell 38. March HN, et al. (2011) Insertional mutagenesis identifies multiple networks of co- 21(4):563–576. operating genes driving intestinal tumorigenesis. Nat Genet 43(12):1202–1209.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1304234110 Tang et al. Downloaded by guest on October 2, 2021