Suppressors and activators of JAK-STAT signaling atdiagnosis and relapse of acute lymphoblasticleukemia in Down syndromeOmer Schwartzmana,b,c,d,1, Angela Maria Savinoa,b,1, Michael Gomberte, Chiara Palmif, Gunnar Cariog,Martin Schrappeg, Cornelia Eckerth, Arend von Stackelbergh, Jin-Yan Huangi, Michal Hameiri-Grossmanb,j,Smadar Avigadb,j, Geertruy te Kronniek, Ifat Gerona,b, Yehudit Birgera,b, Avigail Reina,b, Giulia Zarfatia,b, Ute Fischere,Zohar Mukamelc,d, Martin Stanullal, Andrea Biondif, Giovanni Cazzanigaf, Amedeo Veterem,n, Bridget K. Wagnerm,n,Zhu Cheni,2, Sai-Juan Cheni, Amos Tanayc,d, Arndt Borkhardte,2, and Shai Izraelia,b,2

aThe Gene Development and Environment Pediatric Research Institute, Pediatric Hemato-Oncology, Edmond and Lily Safra Children’s Hospital, ShebaMedical Center, Tel Hashomer, Ramat Gan 52621, Israel; bDepartment of Human Molecular Genetics and Biochemistry, Sackler Medical School, Tel AvivUniversity, Tel Aviv 69978, Israel; cDepartment of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot 76100, Israel;dDepartment of Biological Regulation, Weizmann Institute of Science, Rehovot 76100, Israel; eDepartment of Pediatric Hemato-Oncology and Immunology,Medical Faculty, Heinrich-Heine-University, Dusseldorf 40225, Germany; fCentro Ricerca M. Tettamanti, Clinica Pediatrica, Università di Milano Bicocca,Fondazione MBBM/Ospedale San Gerardo, 20900 Monza, Italy; gDepartment of Pediatrics, Medical University of Schleswig Holstein, 24105 Kiel, Germany;hPediatric Oncology/Hematology, Charité–Universitätsmedizin, 13353 Berlin, Germany; iState Key Laboratory of Medical Genomics, Shanghai Institute ofHematology, Rui Jin Hospital, Jiao Tong University School of Medicine, Shanghai 200025, China; jPediatric Hematology Oncology, Schneider Children’sMedical Center of Israel, Petah Tikva 49202, Israel; kDepartment of Women’s and Children’s Health, University of Padova, 35131 Padova, Italy; lPediatricHematology and Oncology, Hannover Medical School, 30625 Hannover, Germany; mCenter for the Science of Therapeutics, Proteomics Platform, Medicaland Population Genetics Program, Broad Institute, Cambridge, MA 02142; and nHoward Hughes Medical Institute, Broad Institute, Cambridge, MA 02142

Contributed by Zhu Chen, March 28, 2017 (sent for review February 16, 2017; reviewed by Sinisa Dovat and Veronika Sexl)

Children with Down syndrome (DS) are prone to development ofhigh-risk B-cell precursor ALL (DS-ALL), which differs geneticallyfrom most sporadic pediatric ALLs. Increased expression of cytokinereceptor-like factor 2 (CRLF2), the receptor to thymic stromal lym-phopoietin (TSLP), characterizes about half of DS-ALLs and also asubgroup of sporadic “Philadelphia-like” ALLs. To understand thepathogenesis of relapsed DS-ALL, we performed integrative geno-mic analysis of 25 matched diagnosis-remission and -relapse DS-ALLs. We found that the CRLF2 rearrangements are early eventsduring DS-ALL evolution and generally stable between diagnosesand relapse. Secondary activating signaling events in the JAK-STAT/RAS pathway were ubiquitous but highly redundant between di-agnosis and relapse, suggesting that signaling is essential but thatno specific mutations are “relapse driving.” We further found thatactivated JAK2 may be naturally suppressed in 25% of CRLF2pos DS-ALLs by loss-of-function aberrations in USP9X, a deubiquitinase pre-viously shown to stabilize the activated phosphorylated JAK2. In-terrogation of large ALL genomic databases extended our findingsup to 25% of CRLF2pos, Philadelphia-like ALLs. Pharmacological orgenetic inhibition of USP9X, as well as treatment with low-doseruxolitinib, enhanced the survival of pre-B ALL cells overexpressingmutated JAK2. Thus, somehow counterintuitive, we found that sup-pression of JAK-STAT “hypersignaling” may be beneficial to leuke-mic B-cell precursors. This finding and the reduction of JAK mutatedclones at relapse suggest that the therapeutic effect of JAK specificinhibitors may be limited. Rather, combined signaling inhibitors ordirect targeting of the TSLP receptor may be a useful therapeuticstrategy for DS-ALL.

acute lymphoblastic leukemia | Down syndrome | CRLF2 | JAK-STATsignaling | USP9X

Children with Down syndrome (DS) are at a markedly in-creased risk for B-cell precursor acute lymphoblastic leuke-

mia (BCP-ALL) (1). The poor survival of DS-ALL comparedwith ALL in children without DS (“sporadic” ALL) is related toincreased treatment toxicity and to increased incidence of re-lapse (2). Thus, better therapy is needed for these patients.Previous studies by our group and others revealed differences

between the genetics of DS-ALLs and of sporadic ALLs (3–6).The typical cytogenetic subgroups, ETV6-RUNX1 and hyper-diploid ALLs, are less common in DS-ALLs. Acquired somatic

activation of the thymic stromal lymphopoietin (TSLP) pathwayare present at diagnosis in about half of DS-ALLs. Aberrant ex-pression of cytokine receptor-like factor 2 (CRLF2) in these leu-kemias is caused by chromosomal rearrangements consisting eitherof a microdeletion on chromosome X, juxtaposing the promoter ofP2RY8 with the coding region of CRLF2, or by a translocation ofCRLF2 into the Ig heavy chain (IgH) locus. CRLF2 hetero-dimerizes with IL7 receptor-α (IL7R) to form the receptor toTSLP (reviewed in ref. 7). TSLP receptors signal by activation ofthe JAK-STAT pathway. Interestingly, in the majority of DS-ALLs,

Significance

Children with Down syndrome are at increased risk for B-cellacute lymphoblastic leukemia (DS-ALL), often expressing cytokinereceptor-like factor 2 (CRLF2). Here we studied matched diagnosisand relapse DS-ALLs to understand the pathogenesis of re-lapse. We confirm that enhanced JAK-STAT signaling frequently“drives” CRLF2pos DS-ALL at diagnosis, but discovered that cloneswith JAK mutations are unstable, suggesting that they alsoendowed the transformed cells with vulnerabilities. We findUSP9X loss in up to 25% of CRLF2pos ALLs, and demonstrate thatits ablation decreases the toxic effect of JAK2 hypersignaling.Thus, in CRLF2pos ALLs JAK-STAT signaling is often buffered byloss of USP9X. These results have therapeutic implications be-cause they suggest that ALL cells can tolerate a limited range ofJAK-STAT signaling.

Author contributions: M. Schrappe, C.E., G.t.K., M. Stanulla, A. Biondi, G. Cazzaniga,A. Borkhardt, and S.I. designed research; O.S., A.M.S., M.G., J.-Y.H., M.H.-G., S.A., I.G., Y.B.,A.R., G.Z., U.F., and A.V. performed research; C.P., G. Cario, M. Schrappe, C.E., A.v.S., G.t.K.,Z.M., M. Stanulla, A. Biondi, G. Cazzaniga, A.V., B.K.W., Z.C., S.-J.C., and A.T. contributed newreagents/analytic tools; O.S., A.M.S., M.G., J.-Y.H., I.G., Y.B., Z.C., S.-J.C., A.T., A. Borkhardt,and S.I. analyzed data; and O.S., A.M.S., and S.I. wrote the paper.

Reviewers: S.D., Pennsylvania State University; and V.S., Veterinary University of Vienna.

The authors declare no conflict of interest.

Data deposition: Genomic data included in this study were deposited in the EuropeanGenome-phenome Archive (accession no. EGAS00001002410).1O.S. and A.M.S. contributed equally to this work.2To whom correspondence may be addressed. Email: zchen@stn.sh.cn, Arndt.Borkhardt@med.uni-duesseldorf.de, or sizraeli@sheba.health.gov.il.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1702489114/-/DCSupplemental.

E4030–E4039 | PNAS | Published online May 1, 2017 www.pnas.org/cgi/doi/10.1073/pnas.1702489114

additional activating mutations in this pathway are found. Theseadditional mutations include either activating mutations of thereceptors, CRLF2 or IL7R (3, 8), or in the downstream signalingcomponents, most commonly JAK2 (9), but also JAK1 and evenRAS (10). Together, these findings suggest that the CRLF2-JAK-STAT pathway is a major “driver” of DS-ALL.The discovery of the TSLP pathway activation in DS-ALLs

proved to be of general significance to sporadic ALLs in childrenand adults without DS. Indeed, about two-thirds of kinase-drivenALLs, commonly called “Philadelphia like” (Ph-like) ALLs,demonstrate activation of this pathway similarly to DS-ALL (11,12). The prognosis of these leukemias is worse (11–19) and aclinical trial incorporating ruxolitinib into the chemotherapybackbone for newly diagnosed patients has been recently opened(NCT02723994).However, two recent studies from the same laboratory question

the role of CRLF2 in driving therapeutic resistance and relapse ofALL (20, 21). These authors reported that in about a third of thepatients, the major CRLF2-positive (CRLF2pos) clone at diagnosisis lost at relapse. Moreover, in many patients CRLF2 aberrationswere subclonal. If independently confirmed, it would suggest thatCRLF2 (and the downstream signaling activating mutations) arenot important for relapse.The goal of the research presented herein was to decipher the

pathogenesis of relapse of DS-ALL by integrative genomicanalysis of matched leukemia samples from diagnosis and re-lapse. We envisioned that such analysis could lead to develop-ment of targeted relapse-preventing approaches in both patientswith DS and in sporadic CRLF2pos Ph-like ALLs. We report thesurprising finding of an intricate balance between lesions pro-moting and blocking JAK-STAT signaling in these leukemias.These findings have potential implications for targeted ther-apy for prevention of relapse in Ph-like JAK-STAT mutatedBCP-ALLs.

ResultsTo further understand the genetic make-up of DS-ALL, we per-formed whole-exome sequencing (WES) of diagnosis (DX) andmatched remission bone-marrow samples of 31 patients with DS-ALL selected for the absence of the typical cytogenetic abnor-malities of childhood ALL (with, in retrospect, the exception ofone patient DSALLB4, who was positive for ETV6-RUNX1)(Dataset S1). We further sequenced matched relapse samples(R1) from 25 of these patients including consecutive relapses (R2)in five patients. WES was complemented by barcoded (22) deep-targeted sequencing and, for selected samples, SNP array, andpaired-end RNA-seq for confirmation of copy number changesand detection of chimeric transcripts, respectively (Dataset S2).All patients were enrolled on therapeutic Berlin-Frankfurt-Munster–Associazione Italiana Ematologia Oncologia Pediatrica(BFM-AIEOP) protocols with informed consent and appropri-ate approvals of ethical committees. None of the patients wastreated with JAK inhibitors.The median number of detected mutations was significantly

greater in the R1 samples (23.5, range 10–227), compared withDX samples (14, range 3–58; P < 5 × 10−4, Wilcoxon rank sumtest). Moreover, in three relapse samples (Fig. 1) (DSALLB1_R1,DSALLG1_R1, and DSALLG11_R2), we detected 170 mutationsor more (range 170–1,225). In agreement with published literatureof BCP-ALL (23), we observed a mutational signature (24) withpredominance of C→T transitions in CpG context [39% of singlenucleotide variants (SNVs)] (Fig. S1A). However, at relapse theproportion of C→G transversions at the CCG context was higherthan at diagnosis (Fig. S1A) (1.4% at diagnosis, 5.3% at relapse,Fisher’s exact test, q < 0.05). We observed a similar effect in anindependent set (25) of matched diagnosis-relapse pediatricALLs (Fig. S1B). This finding might reflect a contribution of a

secondary mutagenic process to the observed increased muta-tional burden at relapse.

The Protein-Coding Mutational Spectrum of DS-ALL.One of the majorfeatures of DS-ALL is a striking enrichment of CRLF2 genomicrearrangements compared with sporadic ALLs (3). We classified17 patients (54%) positive for CRLF2 aberrations (Fig. 1, mainpanel). There were four patients with the IGH–CRLF2 translocationand 13 patients with P2RY8–CRLF2 rearrangement. We coulddetermine the allelic frequencies of the P2RY8–CRLF2 rearrange-ments for most the samples (Fig. S2). Both IGH–CRLF2 andP2RY8–CRLF2 were stable between diagnosis and relapse in 11 ofthe 12 paired patients carrying these lesions (Fig. 1), raising thepossibility that in DS-ALLs CRLF2 aberration represents an earlyevent during leukemia development and may also be importantfor relapse.We identified lesions affecting known driver genes in most of

our samples (Fig. 1). We grouped recurrently mutated genes orputative driver genes by functional classification. In 46 (75%)samples from 25 (80%) patients, we detected activating mutationsin genes whose protein products are involved in signaling, in-cluding receptors (CRLF2, IL7R, FLT3) or downstream effectorenzymes (JAK1/2, KRAS, and NRAS). Among the other affectedpathways that were frequently mutated are genes involved inB-lymphoid development and differentiation, transcription factorsand chromatin remodeling proteins, general tumor-suppressorgenes, and members of the mismatch repair (MMR) pathway.As expected, in all but one sample with mutations in JAK-STATgenes (CRLF2, IL7R, and JAK1/2), CRLF2 was rearranged. In-terestingly, the CRLF2neg DS-ALLs displayed a similar pattern ofmutations that was reported for hyperdiploid ALLs (26), namelyassociation between mutations in RAS signaling pathway and inchromatin remodeling genes, such as CREBBPs (Fig. 1).In nine patients, we observed deletions targeting the histone

gene cluster-1 (HIST1) on chromosome 6p22-21 (Fig. S3). Thesedeletions were generally stable between diagnosis and relapse(Fig. 2A), and were not associated with CRLF2 status. This findingis consistent with previous genomic studies in DS-ALL (27, 28),and to our knowledge, has not been reported in other cancers,including leukemias.In 16 leukemia samples for which RNA material was available,

we performed paired-end RNA-seq, aiming to identify chimerictranscripts of putative drivers (Fig. 1 and Dataset S2). In patientDSALLB3, both in diagnosis and in relapse samples, we observedoverexpression of the noncoding RNA MIR100HG, as well aschimeric transcripts mapped to IGH and MIR100HG (Fig. S4).RNA-seq data also demonstrated that the aberrant transcriptionstarted upstream to the locus of MIR125B1 (Fig. S4), a knownoncogenic microRNA (29).

Temporal Association of Somatic Mutations Between Diagnosis andRelapse. A series of driver events are probably necessary for acell to become leukemic. However, having acquired this set oflesions might not be sufficient for (or even impair the ability of)the leukemic blasts to propagate relapse after chemotherapytreatment. For example, a driver mutation that is found in bothtime points could be ubiquitously necessary for proliferation orsurvival of the leukemia cells. However, it is also possible that thismutation might be important at diagnosis, but redundant in therelapse, and its persistence during relapse merely represents thecommon cell-of-origin. Conversely, time-point–restricted eventsmight allude to alterations that confer selective advantage onlyunder certain circumstances, for example after treatment withchemotherapy. We thus selected likely pathogenic somatic eventsthat were recurrent in at least two paired patients and classifiedthem to either diagnosis-restricted, relapse-restricted, or sharedbetween time points (Fig. 2A).

Schwartzman et al. PNAS | Published online May 1, 2017 | E4031

MED

ICALSC

IENCE

SPN

ASPL

US

In agreement with a recent report (30), we found relapse-restricted mutations in MSH6 (n = 4), MLH1 (n = 2), andTP53 (n = 3). In addition, four of five leukemia samples withmutations in CREBBP (all CRLF2neg), and four of six sampleswith SETD2 mutations were relapses (Fig. 2B). MSH6 andMLH1 are both members of the MMR pathway and three of thefour samples with MSH6 or MLH1 mutations, demonstrated amarkedly increased mutational burden (≥170 SNVs) (Fig. 2C).The single case (DSALLG15_R1) with MSH6 mutation (p.T915A) and no increase in the mutational burden carried anamino acid alteration predicted as benign by PolyPhen2 (31). In-terestingly, the two cases with MLH1 mutation presented anidentical alteration, a splicing mutation (X264_splice; rs267607789)

observed in patients with Lynch syndrome (32). In addition, thesecond relapse of patient DSALLG11 displayed an exceptionallyhigh number of SNVs (1225), possibly because of a synergistic ef-fect of mutations both in MSH6 and SETD2 (33). Although recentreports described enrichment of NT5C2 mutations in up to 19% ofrelapsed ALLs (34, 35), none were detected in our cohort.Aside from the CRLF2 rearrangements, we found alterations

that were predominantly shared between diagnosis and relapsein IKZF1 (12 of 14), HIST1 (7 of 9), ETV6 (7 of 10), RB1 (5 of5), USP9X (3 of 3), and EBF1 (2 of 2). Contrary to our expec-tations, JAK mutations, highly enriched in DS-ALL, were oftenlost at relapse: four of seven of JAK2 mutations and two of fourJAK1 mutations were restricted to diagnosis (Fig. 2A). This finding

Fig. 1. General distribution of somatic events in 31 patients with DS-ALL. The main panel lists recurrent SNV/indels and copy number variants (CNVs, rep-resented by arrows). The rows correspond to the genes; the columns correspond to leukemia samples. Only SNVs/indels with VAF ≥ 5% are shown. Noticepaired samples (i.e., diagnosis and relapse) are adjacent. The bars to the right indicate the number of patients harboring each alteration. The bars at thebottom indicate the total number of exonic somatic SNVs/indels in each sample (log2 scale).

E4032 | www.pnas.org/cgi/doi/10.1073/pnas.1702489114 Schwartzman et al.

prompted us to examine in more depth the dynamics of ALLsubclones with signaling mutations.

The Dynamics of Signaling Mutations in DS-ALL. As been observed byus and others, the majority (12 of 17, 70%) of CRLF2pos ALLs hadan activating mutation in components of the JAK-STAT signalingpathway. We used the improved sensitivity (SI Materials andMethods) achieved by targeted sequencing to delineate the dy-namics between diagnosis and relapse of major and rare mutatedalleles of signaling genes. In the eight paired patients where mu-tations in JAK1/2 were present at diagnosis, the dominant clonewith the mutation was either undetected after therapy or, in asingle case (DSALLG8), was diminished significantly [variant allele

frequency (VAF) 0.8%] (Fig. 3A). Conversely, in the three patientswith JAK2 mutations at relapse, we detected the surviving clone atdiagnosis as a minor clone (VAF 1–5%). An exception was patientDSALLB9, in whom JAK1 was present in the first relapse as asubclone (VAF 5.2%) and subsequently disappeared at the secondrelapse (Fig. 3A). IL7R and CRLF2 SNVs and insertion/deletions(indels), which activate the receptors by mediating ligand in-dependent dimerization (3, 8, 36), are also common in CRLF2pos

ALLs, and in our cohort we found two patients with the IL7Rmutation (p.S185C) and two patients with the CRLF2 muta-tion (p.F232C) (Fig. 1). These mutations were shared betweendiagnosis and relapse in two patients (DSALLG11, DSALLB5), orrelapse-restricted in one patient (DSALLB1). Thus, the data suggest

A

B C

Fig. 2. Temporal association of somatic mutations between diagnosis and relapse. (A) The bar plot reports the total number of paired patients (n = 25) withsomatic variation in each of the loci listed. Variations were classified as shared regardless whether the identical or different genomic alterations in the samegene were present in both time points. (B) The genomic distribution of protein-changing mutations identified in our cohort in MSH6, MLH1, TP53, SETD2, andCREBBP. (C) Boxplot showing the mutational burden of each sample grouped by MLH1/MSH6 mutated (n = 4) or wild-type (n = 61).

Schwartzman et al. PNAS | Published online May 1, 2017 | E4033

MED

ICALSC

IENCE

SPN

ASPL

US

that mutations in the JAK-STAT genes are secondary events andthat clones harboring these mutations are mostly unstable.RAS mutations displayed different dynamics between diagnosis

and relapse. Overall, we spotted mutations in KRAS and NRAS in11 paired patients, clones harboring mutations in these genes weredetected in 6 (24%) patients at diagnosis, in 9 (36%) patientsR1 relapse, and in 4 (80%) patients at R2 relapse. Interestingly, allsix paired patients with RAS mutations at diagnosis wereCRLF2neg and in all but one patient, KRAS/NRAS were alsomutated at relapse. Moreover, in all patients with clones harboringRAS mutations that were present at both diagnosis and relapse,the relapse emerged from the same clone that was identified in thediagnosis sample (Fig. 3A). In four CRLF2pos patients and oneCRLF2neg patient, either the primary (R1) or the second relapse(R2) emerged from RAS-mutated clones that were undetectablein the preceding time point. This relative durability of clones withRAS mutation could be explained by increased resistance tochemotherapy, as has been suggested by Oshima et al. (25).Overall, we observed mutations in components of JAK-STAT

or RAS pathways in 49 (75%) sequenced exomes from 25 (80%)DS-ALL patients (including both 21 paired and 4 diagnosis-onlysamples). However, a heterogeneous clonal population, with re-spect to the mutated signaling effectors, often presented at thesame leukemia sample. Because leukemogenesis is assumed to be

a process of gradual accumulation of mutations, we asked whetherselection of clones with specific signaling oncogenes might reflectthe background biology of the tumor. To this end, we designed athree-level graph (Fig. 3B), where the vertices correspond tomutated signaling genes, the edges correspond to co-occurrence ina single patient, and the levels reflect the time point (i.e., DX, R1,R2). Concretely, intralevel edges signify coexistence at the samesample, whereas cross-level edges reflect for each gene (vertex) inlevel x, what were the observed mutations in the succeeding levels(time points) for each patient. We placed cross-level edges be-tween the same gene regardless whether the relapse emerged fromthe same clone or from a different clone harboring mutation inthat same gene. This graphic representation allows us to evaluatethe general landscape of signaling in DS-ALL. It is clear that thereare samples in which subclones with mutations in different sig-naling genes coexists (Fig. 3B). In addition, there are cases whereleukemias present with JAK mutations at diagnosis can switch to aRAS mutation, but we have never observed the opposite, sug-gesting the dominance of RAS signaling.

USP9X Is Disrupted in CRLF2pos ALLs. As mentioned above, we foundin our cohort several genes with mutations shared between diagnosisand relapse. One of these genes was USP9X, which was mutated in4 of the 17 patients with CRLF2 rearrangements (Fig. 1). In three of

A

B

Fig. 3. The dynamics of signaling mutations in DS-ALL. (A) VAF of JAK1, JAK2, NRAS, and KRAS mutated alleles at each time point. All 17 patients withmutations in at least one of the genes are shown. Patients are sorted by CRLF2 status at diagnosis. DX, diagnosis; R1, relapse 1; R2, relapse 2. (B) Graphicalrepresentation of somatic alterations in signaling genes along the different time points of relapsed DS-ALL. The vertices represent the mutated genes; edgesbetween vertices indicate a patient with a mutation in both genes. Horizontal edges indicate that the connected genes were mutated in the same sample,and vertical edges connect genes mutated in the same patient at different time point. The thicker edges indicate events observed in more than one patient.The size of the vertices illustrates the number of patients with mutations in the gene. Only mutations with VAF ≥ 5% are included.

E4034 | www.pnas.org/cgi/doi/10.1073/pnas.1702489114 Schwartzman et al.

the patients, we found focal chromosomal deletions that involvedthe USP9X locus (Fig. 4A), whereas in the fourth patient wedetected a frameshift mutation (p.F1115Lfs) (Fig. 4B, in red).In patient DSALLG14, the deletion extended from exon 32 ofUSP9X to exon 2 of its nearest downstream gene, DDX3X.This deletion could result in a chimeric transcript USP9X–

DDX3X, a fusion that was recently reported in CRLF2pos

ALLs with a similar interstitial deletion (28). However, thenature of USP9X deletions in the two other patients was dif-ferent. In case DSALLA2, the deletion was of 5.3 MB, andstretched until the downstream gene ZNF674, whereas in pa-tient DSALLB7, an upstream deletion of 200 kb that extendedto USP9X gene body resulted in reduced USP9X expression atboth diagnosis and relapse (Fig. S5B). The observations thatthe VAF of USP9X in the four cases corresponded to themajor clone persisting at relapse (Fig. 4C and Fig. S5F) aresuggestive of an early event in leukemia genesis.USP9X is a deubquitinating enzyme that has been suggested

to have a protumorigenic role by preventing the ubiquitin-mediated degradation of prosurvival and growth proteins, suchas MCL2 and ERG (37–40). Furthermore, increased expressionof USP9X was proposed to mediate resistance to glucocorticoidsin ALL (40). Therefore, we were surprised to identify loss-of-

function mutations in USP9X. Importantly, even the predictedUSP9X–DDX3X in-frame fusion protein loses the deubuiqui-tination (UCH) domain of USP9X. To further validate thesefindings, we explored publicly available data. Additional mutationsidentified in USP9X in several studies (25, 41–44) are depicted inFig. 4B. Many of these mutations are frameshift or missensemutations within the UCH domain, suggesting loss-of-function.Similar to our findings, the USP9X-mutated BCP-ALLs de-scribed by Oshima et al. (25) carried JAK2 and JAK3 activatingmutations (Fig. S5E) (and hence, are likely to be CRLF2pos).We further analyzed RNA-seq data from the pediatric cancer(PeCAN) genomic database of childhood cancer (45). We com-pared USP9X expression to CRLF2 expression in Ph-like ALL(Fig. 4D). We found that a subset of 7 samples of the 28 CRLF2pos

ALLs (all classified as Ph-like) demonstrated markedly diminishedexpression of USP9X, whereas USP9X expression was preservedin all CRLF2neg samples (P < 10−4, Fisher’s exact test). Addi-tionally, four samples in the PeCAN database were annotated withthe USP9X–DDX3X chimeric transcript; of these, three wereamong the CRLF2 rearranged group. Taken together, these datasuggest that loss of USP9X by genomic aberrations is a recurrentevent in ALL and may be in up to 25% of CRLF2pos ALLs.

A

C

D

B

Fig. 4. USP9X genomic aberrations. (A) Log2R values (Log2 ratio sample/normal) in the USP9X genomic region in the three DS-ALL patients with deletions inthe region. Data shown are derived from a high-resolution SNP array (DSALLB7 and DSALLG14), or from exome coverage data (DSALLA2; indicated with “§”).(B) Coding mutations found in USP9X in independent published cohorts of B-ALLs (25, 41–44). The mutation found in patient DSALLG11 is indicated in red.(C) Scatter plots of the VAF of mutations in diagnosis and in relapse in three patients with USP9X genomic lesions (Und, undetected). (D) CRLF2 and USP9XRNA-seq expression in Ph-like ALLs downloaded from the PeCan database (45). Samples classified as either carrying CRLF2 fusions (n = 28) or negative for thefusions (n = 96). Samples with the known USP9X–DDX3X fusion transcript are indicated with darker color.

Schwartzman et al. PNAS | Published online May 1, 2017 | E4035

MED

ICALSC

IENCE

SPN

ASPL

US

Buffering of JAK-STAT Signaling by Loss-of-Function of USP9X.USP9Xwas recently shown to positively regulate JAK-STAT signaling bybinding to JAK2, thereby enhancing JAK phosphorylation viaremoval of a competing ubiquitin group (46). Consequently, phar-macological inhibition or genetic ablation of USP9X was shownto reduce JAK signaling. Loss of USP9X in CRLF2pos ALLs,thought to be driven by enhanced JAK-STAT signaling, is there-fore counterintuitive. Recent studies have suggested that B-lineagelymphocytes have limited capacity to tolerate signaling. For ex-ample, up-regulation of phosphatases was reported to be impor-tant for lymphoid transformation by BCR-ABL (47). Similarly,experimental ablation of JAK2 was shown to accelerate leukemictransformation by BCR-ABL (48). We thus hypothesized thatloss of USP9X promote survival of CRLF2pos cells by restrictingJAK signaling.To test this hypothesis, we simultaneously overexpressed CRLF2

and JAK2R683G in a BCP-ALL cell line (49). Three days aftertransduction, 96% of the cells expressed surface CRLF2 (Fig.S6A). The cells were then treated with increasing amountsUSP9X specific inhibitor BRD0476 (46) in the presence of thehuman cytokine TSLP (10 ng/mL). After 2 wk of treatment, mostof the cells did not express CRLF2, demonstrating a selectiveadvantage for the nontransduced cells. This observation reflectsthe toxicity of hypersignaling by mutated JAK2. Treatment withUSP9X inhibitor increased the fraction of CRLF2pos cells in a

dose-dependent manner (Fig. 5A), but did not alter the fractionof cells transduced with the control vector (Fig. S6B). Treatmentwith the inhibitor resulted in a decrease in STAT5 phosphorylation(Fig. 5B) (median fluorescence intensity fold-decrease 1.5 ± 0.2).This response was similar to the decrease in STAT1 phosphoryla-tion observed in other biological systems treated with the sameagent (46). BRD0476 did not alter STAT3 phosphorylation, sup-porting a specific effect of the inhibitor on STAT5 (Fig. S6C).Although BRD0476 was shown to be highly specific for USP9X,

we wished to further rule out nonspecific drug effects by directgenetic disruption of USP9X. Using CRISPR/Cas9, we targetedUSP9X at the UCH domain and generated cells lacking USP9Xexpression (Fig. 5C and Fig. S6D). Similar to the pharmacologicalstudies, the genetic elimination of USP9X rescued the growth ofcells expressing CRLF2 and JAK2R683G (Fig. 5D).Because the observed effect of USP9X inhibition might not be

directly mediated through attenuation of JAK signaling, we askedwhether mild pharmacological inhibition of JAK would produce asimilar phenotype. We treated CRLF2/JAK2R683G-transduced cellswith increasing doses of ruxolitinib, an inhibitor of JAK2 and JAK1.At low concentrations (0.25–1 μM), ruxolitinib provided growthadvantage to the transduced cells in the culture compared withvehicle (DMSO 0.25%), consistent with the results obtained withBRD0476 (Fig. 5E). Ruxolitinib exerted the expected cytotoxiceffect on the transduced cells only at higher concentrations. No-tably, the molecular effect of reduction in STAT5 phosphorylation(with no effect on STAT3 phosphorylation) was achieved even atlow concentration of ruxolitinib (Fig. 5F). Together, these resultssuggest that loss of USP9X in CRLF2pos ALL cells may promotesurvival by buffering the intensity of JAK signaling.Because USP9X interacts with both JAK1 and JAK2 (Fig.

S7A), it is likely that the observed effects of USP9X loss would notbe restricted merely to JAK2 signaling (46). Activating mutationsin IL7RA are associated with CRLF2 rearrangements (Fig. 1) andwere shown to signal through JAK1 (8, 36). We therefore trans-duced 018z cells with CRLF2/IL7RINS, a Pro-Pro-Cys-Leu insertionin the transmembrane domain that promotes homodimerizationand activation of the receptor (8). Pharmacological and geneticinhibition of USP9X (Fig. S7 B–D) and pharmacological inhibi-tion JAK1 (Fig. S7E) resulted in increased survival of transducedcells, similar to our observations following transduction with CRLF2/JAK2R683G (Fig. 5).

DiscussionThe association between DS and ALL is intriguing. The geneticmakeup of these BCP leukemias differs from the most commoncytogenetic variants of childhood ALL. Therapeutically, DS-ALLspose a significant challenge. Increased toxicity of chemotherapy iscoupled with an intrinsic increased risk for relapse to negativelyimpact the survival of children with DS-ALL (2). By focusing onDS-ALL unique leukemias and on matched diagnosis and relapsesamples, our goal was to decipher the pathogenesis of relapse ofDS-ALLs aiming at improved therapy. We discovered that thegenomic landscape differs significantly between the two majorsubtypes of DS-ALLs. CRLF2neg DS-ALLs are characterized byenhanced RAS signaling coupled by mutations in chromatinremodeling genes, in particular CREBBP. This genetic architec-ture is similar to hyperdiploid ALL, in which the cooperationbetween these two genetic lesions is thought to drive relapse (26,50). In contrast CRLF2pos DS-ALLs are characterized by highdynamics of proliferative signaling. Enhanced JAK-STAT signal-ing mediated by the increased expression of CRLF2, the receptorto TSLP, is an early event during the evolution of DS-ALL, as isevident by its high allelic frequency and its stability between di-agnosis and relapse. However, it also endows the leukemic cellswith vulnerabilities, as evident by the presence of mutations inUSP9X that attenuate JAK-STAT signaling, by the subclonalnature of downstream JAKmutations, and by decreased prevalence

A B

C D

E F

Fig. 5. Silencing of USP9X enhances leukemia cell survival by buffering JAKsignaling. (A) The effect of increasing doses of the USP9X inhibitorBRD0476 on viability of 018z BCP-ALL cells transduced with CRLF2/JAK2R683G.Shown is flow cytometry analysis of the fraction of viable transduced cellsexpressing surface CRLF2 (CRLF2+) 2 wk after treatment (*P< 0.05).(B) Phosphoflow cytometry analysis of STAT5 phosphorylation (pSTAT5)30 min after treatment with BRD0476 (10 μM). (C) Immunoblot for USP9X inwild-type and CRISPR/Cas9-knockout 018z leukemia cells. (D) The effect ofUSP9X knockout on the relative growth of CRLF2/JAK2R683G transduced 018zcells after 2 wk in culture. Shown is flow cytometry analysis of the fraction ofviable transduced cells expressing surface CRLF2 (CRLF2+) (*P < 0.05). (E) Theeffect of the JAK inhibitor ruxolitinib on proliferation of 018z cells trans-duced with CRLF2/JAK2R683G one week following treatment. (F) Phosphoflow cytometry analysis of STAT5 phosphorylation (pSTAT5) after treatmentwith ruxolitinib (0.25 μM).

E4036 | www.pnas.org/cgi/doi/10.1073/pnas.1702489114 Schwartzman et al.

of JAK-activating mutation in relapse. These results have implica-tions to Ph-like ALLs in patients without DS and for the incorpo-ration of JAK inhibitors for treatment of ALL.Analysis of our cohort and publicly available databases revealed

recurrent loss-of-function mutations in USP9X. This finding wassurprising because USP9X is perceived mainly as an oncogene insolid tumors and in BCP-ALL and B-cell lymphoblastic lympho-mas (37, 40, 51, 52). Because USP9X is a deubiquitinase, its effecton tumor growth should largely be dependent on its substrates,which probably explains its ascribed role as tumor suppressor inpancreatic cancer (53). In a recent report describing 17 childrenwith germ-line mutations in USP9X, one child developed ALL(54); however, this single case is not enough to define USP9X asan ALL tumor suppressor.USP9X was also reported to enhance JAK2 signaling, as well as

signaling of immune receptors (46, 55, 56). Thus, the loss ofUSP9X activity in up to 25% of CRLF2pos ALLs, a subtype ofleukemia believed to be driven by enhanced JAK-STAT signaling,is intriguing. We show here that USP9X loss may promote thesurvival of leukemic cells by restricting the intensity of JAK2 andJAK1 signaling. Importantly, low-dose ruxolitinib, a JAK2/JAK1 inhibitor currently in clinical trials for CRLF2pos ALLs, alsoincreased the survival of leukemic cells transduced with CRLF2-and JAK2R683G, the most common JAK2 mutation in ALL (9).This observation suggests the need for high-dose ruxolitinib inthese trials.Our discoveries are consistent with recent findings that B cells,

including leukemic B-cell precursors, can tolerate a relativelynarrow range of signaling, possibly reflecting the evolutionaryimportant autoimmunity checkpoint (57). Although signaling isrequired for their proliferation, enhanced signaling could causetheir demise. The increased lymphoid sensitivity to hypersignalingmay also explain the intriguing observation that the V617F strong-activating JAK2 mutation, characterizing myeloproliferative neo-plasms, is never found in ALL (9). The buffering effect of loss ofUSP9X in CRLF2pos ALLs is, therefore, similar to the bufferingeffect of increased expression of phosphatases observed in BCR-ABL ALL (47). Interestingly, a high expression of the phosphatasePTPRC (CD45) was also recently reported to be associated withP2RY8–CRLF2 ALLs (58).Elimination of USP9X may promote ALL through other

mechanisms yet to be discovered. The very early occurrence ofUSP9X loss-of-function mutations and their presence in othersubtypes of ALL [including T-ALL (25)] are consistent with ad-ditional tumor-suppressive mechanisms. In fact, it is also possiblethat the selection of activating JAK-STAT mutations may haveoccurred to compensate for the loss of USP9X. Hematopoieticspecific elimination of Usp9x in mice results in differentiationblock of both T and B cells (56). Thus, mutations in USP9X maytherefore participate in the leukemogenesis process by arrestingB-cell differentiation similar to the mutations in other B-cell–differentiation transcription factors. Clearly, additional studiesare needed to clarify the leukemogenic role of USP9X loss-of-function mutations.A major goal of our study was to decipher the mechanisms of

relapse of DS-ALL. Relapse reflects two cooperating processes:the primary leukemogenic process driving the growth and survivalof leukemic cells and the selection pressure imposed by frontlinetherapy. Similar to other studies of relapsed ALL (21, 25, 26, 44,50, 59), we observed increased RAS activation in relapse. Indeed,it has been proposed that RAS may mediate resistance to che-motherapy. This hypothesis is questioned by inspection of theclonal dynamics between diagnosis and relapse. Both in our cohortand in published studies (21, 25, 60) the expanded clones at re-lapse often carry a different RAS mutation than clones detected indiagnosis. This phenomenon is consistent with a general role ofsignaling in enhancing growth and proliferation of leukemic cellsand not with resistance to chemotherapy.

The subclonal nature and the complex dynamics of CRLF2pos

leukemia cells carrying either JAK-STAT or RAS suggest thatinhibition of a single signaling pathway will select for leukemicclones driven by signaling of the other pathway. Our genomic data(Fig. 1) also suggest that signaling by either JAK-STAT or RAS isessential for leukemia growth. Therefore, we suggest that JAKinhibitors should be combined with MEK inhibitors for treatmentof DS-ALL and CRLF2pos Ph-like ALLs.Mutations selected by chemotherapy, either from preexisting

subclones or by the enhanced genomic instability, are likely to beenriched in relapse specimens. The relapse-enriched mutationsfound in our DS-ALL cohort are similar to those found in otherALL subtypes (21, 25, 26, 30, 44, 50, 59). Intriguingly, however, wehave found no NT5C2 mutations in our cohort. Previous studiesidentified NT5C2 mutations in 8–45% of the relapses of BCP-ALLs (34, 35, 44). Mutations in NT5C2 were associated withearly relapses, whereas in our cohort the median time to relapsewas relatively long, 1,071 d (range 375–2,226 d). NT5C2 mutationsare thought to provide resistance to 6-mercaptopurine (6MP), akey drug in the treatment of ALL. Instead, we identified relapse-specific mutations in MMR genes in 15% of our patients. TheMMR pathway is important for repair of damage caused by drugs,such as 6MP, and changes in MMR gene expression have beenreported in association with resistance to 6MP (61). It will beinteresting to see if different mutations conferring resistance to6MP are mutually exclusive in relapse ALL.Strikingly, most of relapsed ALLs, in our and in published

studies, are not characterized by a major clone carrying a relapse-specific chemotherapy-resistant mutation. This finding impliesthat relapse may be often facilitated by one of the early foundingmutations that initiated the primary leukemia. This hypothesis issupported by the strong prognostic significance of certain muta-tions found already as a major clone at diagnosis. Two such goodexamples, IKZF1 aberrations and CRLF2 rearrangements thatoften co-occur in the same leukemias (14, 17, 18, 21, 28, 62),predict worse prognosis and, as clearly shown in our data, areusually shared between diagnosis and relapse samples.A recent experimental study, modeling diagnosis and relapse

ALL, suggests that relapse initiates from dormant leukemic cellsthat survive by adhering to protective bone-marrow stroma cells(63). According to this model, initial chemotherapy eliminates thebulk of leukemic cells (e.g., JAK and RAS mutated cells). Theslow dividing cells are eliminated by the continuous maintenancetherapy (6MP and methotrexate). In relapsing patients a few re-sidual cells remain dormant on stroma. Creation of full-blownleukemia depends on proliferative signaling (e.g., RAS).IKZF1 mutations have been shown to promote adherence to

bone-marrow stromal cells and therefore, through this mecha-nism, may promote dormancy during complete remission and re-lapse (64, 65). Hence, targeting IKZF1 by, for example, caseinkinase II (66) or FAK inhibitors (65), may be useful in targetingthis dormancy stage. Similarly, we speculate that leukemic cellscarrying CRLF2 rearrangements—and, hence, expressing theTSLP receptor—may survive chemotherapy during complete re-mission by interacting with TSLP-producing cells in the bone-marrow niche. The experimental examination of this hypothesisis challenging, because mouse TSLP is not reactive with the hu-man TSLP receptor (67). The hypothesis suggests that targetingTSLP [for example, by the novel anti-TSLP antibodies (68)] or theTSLP receptor [by for example, CRLF2 antibodies or CAR-T cells(69, 70)] may be a useful additional therapy—better than JAKinhibition—to prevent relapse in patients with DS and sporadicCRLF2pos ALL.

Materials and MethodsDNA and RNA of bone-marrow–derived DS-ALL blasts were obtained frompatients after informed consent and approval of the ethics committees of allparticipating institutions in the AIEOP-BFM protocol, in accordance with the

Schwartzman et al. PNAS | Published online May 1, 2017 | E4037

MED

ICALSC

IENCE

SPN

ASPL

US

Declaration of Helsinki. Our study was also approved by the ethic committeeof Israeli Health Ministry and Sheba Medical Center approval #4841–2346. Asrequired by this approval, samples were anonymized. Samples were includedbased on the availability of DNA from matched diagnosis-remission (n = 31)and -relapse (n = 26). The clinical details are listed in Dataset S1. More de-tailed materials and methods are in the SI Materials and Methods. SeeDataset S3 for the high-confidence variant list and Table S1 for 018z cell lineshort tandem repeat (STR) information.

Genomic Analysis. Whole-exome sequencing was performed using SureSelectHuman All Exons+UTRs V5 kit (Agilent Technologies) and paired-end se-quencing on was done on HiSeq2500 platform (Illumina). Somatic variantswere detected using MuTect2 (71). Copy number variations (CNV) wereinferred from WES coverage data using EXCAVATOR (72), or using CytoscanHD (Affymetrix). DNA targeted resequencing was done with a UMI-taggedTruSeq custom amplicon low input kit (TSCA-LI, Illumina) and processed usingcustom scripts. RNA-seq libraries were prepared using in-house protocols andpaired-end sequencing on a NextSeq500 platform (Illumina). Analysis of chimerictranscript was done with FusionCatcher (biorxiv.org/lookup/doi/10.1101/011650).A minimal threshold of 5% VAF was used for the SNVs and indels presented inthe main-text figures. An exception was Fig. 3A, in which no VAF threshold wasused, to include mutations in rare clones that becamemore dominant at relapse.

In Vitro Studies. All in-vitro studies were performed on ALL 018z BCP-ALL cells(49) transduced with lentiviruses. Flow cytometry analysis was done onGallios platform and Kaluza software (Beckman-Coulter) with the follow-ing reagents: 7AAD, anti-TSLPR (CRLF2; clone 1D3 1B4 PE-conjugated;Biolegend), LIVE/DEAD Fixable staining (ThermoFisher Scientific), pSTAT5

antibody (APC-conjugated, E-Bioscience, ThermoFisher Scientific), Countingbeads (Countbright, ThermoFisher Scientific). Small molecule inhibitors usedwere BRD0476 (46) and ruxolitinib (Sellek Chem). Targeting USP9X usingCRISPR/Cas9 was done per a previously published protocol (73). Western blotanalysis was done with the following antibodies: anti-human USP9X (BethylLaboratories) and anti–β-actin (Sigma Aldrich).

Data Availability. Genomic data included in this study were deposited in theEuropean Genome-phenome Archive (accession no. EGAS00001002410). Thestudy protocol’s patient consent did not include permission to make the datapublically available. Researchers wishing to gain access to the raw data mustcontact the corresponding authors directly.

ACKNOWLEDGMENTS. We thank Inna Muller for excellent technical support,and members of the A.T. and S.I. research groups for fruitful discussions andadvice. These studies were funded by an Israel Science Foundation Legacygrant complemented by an Israel Science Foundation-Israel National Centerfor Personalized Medicine grant (to S.I.); an Israel Science Foundation jointgrant with China (to S.I. and S.-J.C.); European Union European Research Area-NET TRANCALL (S.I., G. Cario, M. Stanulla, G.t.K., G. Cazzaniga, and C.E.); theIsrael Cancer Research Fund (S.I.); the Waxman Foundation (S.I.); the WilliamLawrence and Blanche Hughes foundation (S.I.); the German Israel Foundation(S.I. and C.E.); the Israel Absorption Ministry (I.G.); the Israel Cancer Association(S.I.); Children with Cancer UK (S.I.); the Dotan Center for Hematological Ma-lignancies in Tel Aviv University (S.I.); the Dora and Giorgio Shapiro Chair forhematological malignancies, Tel Aviv University (S.I.); Fondazione Italiana perla Ricerca Sul Cancro (A.M.S.); the German Consortium of Translational CancerResearch (A. Borkhardt). This research partially fulfils the requirements for aPhD of Tel Aviv University for O.S., A.R., and I.G.

1. Hasle H (2001) Pattern of malignant disorders in individuals with Down’s syndrome.Lancet Oncol 2:429–436.

2. Buitenkamp TD, et al. (2014) Acute lymphoblastic leukemia in children with Downsyndrome: A retrospective analysis from the Ponte di Legno study group. Blood 123:70–77.

3. Hertzberg L, et al. (2010) Down syndrome acute lymphoblastic leukemia, a highlyheterogeneous disease in which aberrant expression of CRLF2 is associated withmutated JAK2: A report from the International BFM Study Group. Blood 115:1006–1017.

4. Izraeli S (2016) The acute lymphoblastic leukemia of Down syndrome—genetics andpathogenesis. Eur J Med Genet 59:158–161.

5. Mullighan CG, et al. (2009) Rearrangement of CRLF2 in B-progenitor- and Downsyndrome-associated acute lymphoblastic leukemia. Nat Genet 41:1243–1246.

6. Russell LJ, et al. (2009) Deregulated expression of cytokine receptor gene, CRLF2, isinvolved in lymphoid transformation in B-cell precursor acute lymphoblastic leuke-mia. Blood 114:2688–2698.

7. Tal N, Shochat C, Geron I, Bercovich D, Izraeli S (2014) Interleukin 7 and thymicstromal lymphopoietin: From immunity to leukemia. Cell Mol Life Sci 71:365–378.

8. Shochat C, et al. (2011) Gain-of-function mutations in interleukin-7 receptor-α (IL7R)in childhood acute lymphoblastic leukemias. J Exp Med 208:901–908.

9. Bercovich D, et al. (2008) Mutations of JAK2 in acute lymphoblastic leukaemias as-sociated with Down’s syndrome. Lancet 372:1484–1492.

10. Nikolaev SI, et al. (2014) Frequent cases of RAS-mutated Down syndrome acutelymphoblastic leukaemia lack JAK2 mutations. Nat Commun 5:4654.

11. Roberts KG, et al. (2014) Targetable kinase-activating lesions in Ph-like acute lym-phoblastic leukemia. N Engl J Med 371:1005–1015.

12. Izraeli S (2014) Beyond Philadelphia: ‘Ph-like’ B cell precursor acute lymphoblasticleukemias—Diagnostic challenges and therapeutic promises. Curr Opin Hematol 21:289–296.

13. Palmi C, et al. (2016) CRLF2 over-expression is a poor prognostic marker in childrenwith high risk T-cell acute lymphoblastic leukemia. Oncotarget 7:59260–59272.

14. Ge Z, et al. (2016) High CRLF2 expression associates with IKZF1 dysfunction in adult acutelymphoblastic leukemia without CRLF2 rearrangement. Oncotarget 7:49722–49732.

15. van der Veer A, et al. (2013) Independent prognostic value of BCR-ABL1-like signatureand IKZF1 deletion, but not high CRLF2 expression, in children with B-cell precursorALL. Blood 122:2622–2629.

16. Palmi C, et al. (2012) Poor prognosis for P2RY8-CRLF2 fusion but not for CRLF2 over-expression in children with intermediate risk B-cell precursor acute lymphoblasticleukemia. Leukemia 26:2245–2253.

17. Moorman AV, et al. (2012) IGH@ translocations, CRLF2 deregulation, and micro-deletions in adolescents and adults with acute lymphoblastic leukemia. J Clin Oncol30:3100–3108.

18. Chen IM, et al. (2012) Outcome modeling with CRLF2, IKZF1, JAK, and minimal re-sidual disease in pediatric acute lymphoblastic leukemia: A Children’s OncologyGroup study. Blood 119:3512–3522.

19. Cario G, et al. (2010) Presence of the P2RY8-CRLF2 rearrangement is associated with apoor prognosis in non-high-risk precursor B-cell acute lymphoblastic leukemia inchildren treated according to the ALL-BFM 2000 protocol. Blood 115:5393–5397.

20. Morak M, et al. (2012) Small sizes and indolent evolutionary dynamics challenge thepotential role of P2RY8-CRLF2-harboring clones as main relapse-driving force inchildhood ALL. Blood 120:5134–5142.

21. Vesely C, et al. (2017) Genomic and transcriptional landscape of P2RY8-CRLF2-positive

childhood acute lymphoblastic leukemia. Leukemia, 10.1038/leu.2016.365.22. Kivioja T, et al. (2011) Counting absolute numbers of molecules using unique mo-

lecular identifiers. Nat Methods 9:72–74.23. Paulsson K, et al. (2015) The genomic landscape of high hyperdiploid childhood acute

lymphoblastic leukemia. Nat Genet 47:672–676.24. Alexandrov LB, et al.; Australian Pancreatic Cancer Genome Initiative; ICGC Breast

Cancer Consortium; ICGC MMML-Seq Consortium; ICGC PedBrain (2013) Signatures of

mutational processes in human cancer. Nature 500:415–421.25. Oshima K, et al. (2016) Mutational landscape, clonal evolution patterns, and role of

RAS mutations in relapsed acute lymphoblastic leukemia. Proc Natl Acad Sci USA 113:

11306–11311.26. Malinowska-Ozdowy K, et al. (2015) KRAS and CREBBP mutations: A relapse-linked

malicious liaison in childhood high hyperdiploid acute lymphoblastic leukemia.

Leukemia 29:1656–1667.27. Loudin MG, et al. (2011) Genomic profiling in Down syndrome acute lymphoblastic

leukemia identifies histone gene deletions associated with altered methylation pro-

files. Leukemia 25:1555–1563.28. Russell LJ, et al. (2016) Characterisation of the genomic landscape of CRLF2-

rearranged acute lymphoblastic leukemia. Genes Chromosomes Cancer 56:363–372.29. Shaham L, Binder V, Gefen N, Borkhardt A, Izraeli S (2012) MiR-125 in normal and

malignant hematopoiesis. Leukemia 26:2011–2018.30. Mar BG, et al. (2014) Mutations in epigenetic regulators including SETD2 are gained

during relapse in paediatric acute lymphoblastic leukaemia. Nat Commun 5:3469.31. Adzhubei I, Jordan DM, Sunyaev SR (2013) Predicting functional effect of human

missense mutations using PolyPhen-2. Curr Protoc Hum Genet Chapter 7:Unit7 20.32. Auclair J, et al. (2006) Systematic mRNA analysis for the effect of MLH1 and

MSH2 missense and silent mutations on aberrant splicing. Hum Mutat 27:145–154.33. Li F, et al. (2013) The histone mark H3K36me3 regulates human DNA mismatch repair

through its interaction with MutSα. Cell 153:590–600.34. Tzoneva G, et al. (2013) Activating mutations in the NT5C2 nucleotidase gene drive

chemotherapy resistance in relapsed ALL. Nat Med 19:368–371.35. Meyer JA, et al. (2013) Relapse-specific mutations in NT5C2 in childhood acute lym-

phoblastic leukemia. Nat Genet 45:290–294.36. Shochat C, et al. (2014) Novel activating mutations lacking cysteine in type I cytokine

receptors in acute lymphoblastic leukemia. Blood 124:106–110.37. Schwickart M, et al. (2010) Deubiquitinase USP9X stabilizes MCL1 and promotes tu-

mour cell survival. Nature 463:103–107.38. Wang S, et al. (2014) Ablation of the oncogenic transcription factor ERG by deubi-

quitinase inhibition in prostate cancer. Proc Natl Acad Sci USA 111:4251–4256.39. Yang B, et al. (2016) Deubiquitinase USP9X deubiquitinates β-catenin and promotes

high grade glioma cell growth. Oncotarget 7:79515–79525.40. Zhou M, et al. (2015) Targeting of the deubiquitinase USP9X attenuates B-cell acute

lymphoblastic leukemia cell survival and overcomes glucocorticoid resistance. Biochem

Biophys Res Commun 459:333–339.41. Liu Y-F, et al. Genomic profiling of adult and pediatric B-cell acute lymphoblastic

leukemia. EBioMedicine 8:173–83.42. Yasuda T, et al. (2016) Recurrent DUX4 fusions in B cell acute lymphoblastic leukemia

of adolescents and young adults. Nat Genet 48:569–574.

E4038 | www.pnas.org/cgi/doi/10.1073/pnas.1702489114 Schwartzman et al.

43. Lilljebjörn H, et al. (2016) Identification of ETV6-RUNX1-like and DUX4-rearrangedsubtypes in paediatric B-cell precursor acute lymphoblastic leukaemia. Nat Commun7:11790.

44. Ma X, et al. (2015) Rise and fall of subclones from diagnosis to relapse in pediatricB-acute lymphoblastic leukaemia. Nat Commun 6:6604.

45. Zhou X, et al. (2016) Exploring genomic alteration in pediatric cancer using Pro-teinPaint. Nat Genet 48:4–6.

46. Chou DH, et al. (2015) Kinase-independent small-molecule inhibition of JAK-STATsignaling. J Am Chem Soc 137:7929–7934.

47. Shojaee S, et al. (2015) Erk negative feedback control enables pre-B cell trans-formation and represents a therapeutic target in acute lymphoblastic leukemia.Cancer Cell 28:114–128.

48. Grundschober E, et al. (2014) Acceleration of Bcr-Abl+ leukemia induced by deletionof JAK2. Leukemia 28:1918–1922.

49. Frishman-Levy L, et al. (2015) Central nervous system acute lymphoblastic leukemia:Role of natural killer cells. Blood 125:3420–3431.

50. Dixon ZA, et al. (2016) CREBBP knockdown enhances RAS/RAF/MEK/ERK signalling inRas pathway mutated acute lymphoblastic leukaemia but does not modulate che-motherapeutic response. Haematologica 102:736–745.

51. Engel K, et al. (2016) USP9X stabilizes XIAP to regulate mitotic cell death and che-moresistance in aggressive B-cell lymphoma. EMBO Mol Med 8:851–862.

52. Vucic D, Dixit VM, Wertz IE (2011) Ubiquitylation in apoptosis: A post-translationalmodification at the edge of life and death. Nat Rev Mol Cell Biol 12:439–452.

53. Pérez-Mancera PA, et al.; Australian Pancreatic Cancer Genome Initiative (2012) Thedeubiquitinase USP9X suppresses pancreatic ductal adenocarcinoma. Nature 486:266–270.

54. Reijnders MR, et al.; DDD Study (2016) De novo loss-of-function mutations in USP9Xcause a female-specific recognizable syndrome with developmental delay and con-genital malformations. Am J Hum Genet 98:373–381.

55. Naik E, Dixit VM (2016) Usp9X is required for lymphocyte activation and homeostasisthrough its control of ZAP70 ubiquitination and PKCβ kinase activity. J Immunol 196:3438–3451.

56. Naik E, et al. (2014) Regulation of proximal T cell receptor signaling and toleranceinduction by deubiquitinase Usp9X. J Exp Med 211:1947–1955.

57. Chen Z, et al. (2015) Signalling thresholds and negative B-cell selection in acutelymphoblastic leukaemia. Nature 521:357–361.

58. Cario G, et al. (2014) High CD45 surface expression determines relapse risk in childrenwith precursor B-cell and T-cell acute lymphoblastic leukemia treated according to theALL-BFM 2000 protocol. Haematologica 99:103–110.

59. Irving JA, et al. (2016) Integration of genetic and clinical risk factors improves prog-nostication in relapsed childhood B-cell precursor acute lymphoblastic leukemia.Blood 128:911–922.

60. Trentin L, et al. (2016) Deciphering KRAS and NRAS mutated clone dynamics in MLL-AF4 paediatric leukaemia by ultra deep sequencing analysis. Sci Rep 6:34449.

61. Cheok MH, et al. (2003) Treatment-specific changes in gene expression discriminate in

vivo drug response in human leukemia cells. Nat Genet 34:85–90.62. Collins-Underwood JR, Mullighan CG (2010) Genomic profiling of high-risk acute

lymphoblastic leukemia. Leukemia 24:1676–1685.63. Ebinger S, et al. (2016) Characterization of rare, dormant, and therapy-resistant cells

in acute lymphoblastic leukemia. Cancer Cell 30:849–862.64. Schwickert TA, et al. (2014) Stage-specific control of early B cell development by the

transcription factor Ikaros. Nat Immunol 15:283–293.65. Joshi I, et al. (2014) Loss of Ikaros DNA-binding function confers integrin-dependent

survival on pre-B cells and progression to acute lymphoblastic leukemia. Nat Immunol

15:294–304.66. Song C, et al. (2015) Targeting casein kinase II restores Ikaros tumor suppressor ac-

tivity and demonstrates therapeutic efficacy in high-risk leukemia. Blood 126:

1813–1822.67. Francis OL, et al. (2016) A novel xenograft model to study the role of TSLP-induced

CRLF2 signals in normal and malignant human B lymphopoiesis. Haematologica 101:

417–426.68. Gauvreau GM, et al. (2014) Effects of an anti-TSLP antibody on allergen-induced

asthmatic responses. N Engl J Med 370:2102–2110.69. Vetter T, et al. (2016) Blockade of thymic stromal lymphopoietin (TSLP) receptor in-

hibits TSLP-driven proliferation and signalling in lymphoblasts from a subset of

B-precursor ALL patients. Leuk Res 40:38–43.70. Qin H, et al. (2015) Eradication of B-ALL using chimeric antigen receptor-expressing

T cells targeting the TSLPR oncoprotein. Blood 126:629–639.71. Cibulskis K, et al. (2013) Sensitive detection of somatic point mutations in impure and

heterogeneous cancer samples. Nat Biotechnol 31:213–219.72. Magi A, et al. (2013) EXCAVATOR: Detecting copy number variants from whole-

exome sequencing data. Genome Biol 14:R120.73. Sanjana NE, Shalem O, Zhang F (2014) Improved vectors and genome-wide libraries

for CRISPR screening. Nat Methods 11:783–784.74. Van der Auwera GA, et al. (2013) From FastQ data to high confidence variant calls:

The Genome Analysis Toolkit best practices pipeline. Curr Protoc Bioinformatics 43:

11.10.1–11.10.33.75. Lek M, et al.; Exome Aggregation Consortium (2016) Analysis of protein-coding ge-

netic variation in 60,706 humans. Nature 536:285–291.76. Mermel CH, et al. (2011) GISTIC2.0 facilitates sensitive and confident localization of

the targets of focal somatic copy-number alteration in human cancers. Genome Biol

12:R41.77. Carter SL, et al. (2012) Absolute quantification of somatic DNA alterations in human

cancer. Nat Biotechnol 30:413–421.78. Tiscornia G, Singer O, Verma IM (2006) Production and purification of lentiviral

vectors. Nat Protoc 1:241–245.

Schwartzman et al. PNAS | Published online May 1, 2017 | E4039

MED

ICALSC

IENCE

SPN

ASPL

US