Gain-of-function mutations in interleukin-7 receptor-α (IL7R) in childhood acute lymphoblastic leukemias

We screened DNA derived from 133 bone marrow samples of BCP-ALLs, including 62 DS-ALLs, with aberrant expression of CRLF2 and additional 153 BCP-ALL not expressing CRLF2 including 21 DS-ALLs. Nine leukemias with IL7R mutations were identified. The rate of mutations in the CRLF2 group (8/133; 6%) was significantly higher than the rest of BCP-ALLs (1/153; 0.6%; P = 0.01, Fisher’s exact test; Table I).

Two types of mutations in IL7R were identified (Fig. 1, A and B, and Table I). Replacement of serine with cysteine at position 185 at the extracellular domain in four patients and complex in-frame insertions and deletions resulted in the addition of 3–7 aa at the transmembrane domain in 5 patients. Whereas the inserted amino acids varied from patient to patient, cysteine was always included (Fig. 1 C).

All mutations were heterozygous and somatic, as they were absent in remission bone marrows. The mutated mRNA was expressed (Fig. 1 D and Fig. S1). JAK2 mutations were present in three samples, whereas no mutations in JAK1 were identified. Four patients relapsed. Examination of DNA derived from matched diagnostic and relapse samples revealed clonal diversification. The same IL7R mutation was present in both diagnosis and relapse in patient M61. IL7R mutation was present only at relapse in patient M122, whereas CRLF2 expression was already noted at diagnosis, suggesting that IL7R mutation was a progression event. Conversely, in two patients, M90 and M112 CRLF2 abnormalities and IL7R mutations were present at diagnosis and not in relapse, suggesting that relapse arose from a different subclone. These findings are consistent with the recent studies on the marked clonal diversity of B- and T-ALLs (Rothman et al., 2005; Anderson et al., 2011; Clappier et al., 2011).

We next tested whether the somatic mutations in IL-7R are gain-of-function mutations that cooperate with CRLF2 to form a constitutively active TSLP receptor. We used the IL-3–dependent mouse pro–B cells BaF3 (not expressing endogenous TSLP receptor; Fig. S2) to generate cell lines that express either WT or mutated IL7R alone or together with CRLF2. We expressed IL-7R S185C and the c.819 Ins12 (CCCCCGTGCCTA) 243 insertion (Ins) PPCL (herein “InsPPCL”) representing the two types of mutations. All proteins were expressed at the cell membrane as demonstrated by flow cytometry (Fig. 2 A). 1 wk after IL-3 withdrawal, only cells transduced with CRLF2 and mutated IL-7R survived (Fig. 2, B and C). Closer examination revealed two populations of surviving BaF3 cells transduced with CRLF2 and IL-7R InsPPLC, suggesting that IL-7R InsPPLC provided survival advantage by itself. Indeed, growth assays of BaF3 cells transduced with the IL-7R InsPPLC construct in the absence of CRLF2 (Fig. 2 B) revealed robust cytokine-independent growth. In contrast, BaF3 cells expressing IL-7R S185C required CRLF2 co-expression for survival. Cells expressing only CRLF2, IL-7R S185C, or WT CRLF2 and WT IL-7R did not grow in the absence of cytokines.

Biochemical analysis of proteins extracted from BaF3 cell was consistent with the growth assays (Fig. 2 C). Both Stat5 and RPS6 were phosphorylated in the absence of cytokine in BaF3 cells transduced with the IL-7R InsPPCL alone, but not in cells expressing IL-7R S185C or WT IL-7R. Co-expression of CRLF2 with each of these mutated IL-7R proteins, but not with the WT IL-7R caused constitutive phosphorylation of Stat5 and RPS6. Together, the functional and biochemical assays demonstrate that the two types of somatic mutations in IL7R are activating mutations causing cytokine independent growth of mouse pro–B cells and constitutive activation of STAT and mTOR pathways. In accordance with the functional studies, the single patient in whom IL7R was mutated in the absence of CRLF2 expression (Table I, patient M61) had the insertion type of mutation with the CRLF2-independent activation phenotype.

To see if CRLF2 and mutated IL-7R form a functional TSLP receptor, BaF3 cells expressing CRLF2 and either WT or mutated IL-7R were starved from IL-3 and treated with 100 ng/ml TSLP for 25 min. As noted in Fig. 2 C, cells expressing the mutated receptor exhibited Stat5 and RPS6 phosphorylation in the absence of cytokine. Yet all cells responded to TSLP with a marked increase in Stat5 and RPS6 phosphorylation (Fig. 3 A). Thus, CRLF2 and mutated IL-7R form a functional TSLP receptor. To test if the presence of mutated IL-7R sensitizes BaF3 cells to TSLP, we treated BaF3 transduced with CRLF2 and either WT IL-7R or IL-7R S185C with increasing doses of TSLP in the absence of IL-3 (Fig. 3 B). Cells expressing CRLF2–IL-7R S185C grew in the absence of cytokines, but also demonstrated higher sensitivity to TSLP. As little as 0.1 ng/ml TSLP doubled the growth rate, but had no effect on BaF3 cells transduced with CRLF2–IL-7R WT. Similarly, cells transduced with IL-7R InsPPCL also responded to TSLP, in addition to marked self-activation of this allele (Fig. S3).

Loss-of-function mutations in IL-7R cause severe combined immunodeficiency characterized by the complete absence of T lymphocytes and the presence of B and NK cells (Puel et al., 1998). Both B- and T-ALLs express IL-7R and have been reported to respond to IL-7 (Touw et al., 1990). We have therefore hypothesized that gain-of-function mutations in IL-7R may also be detected in T-ALL. Accordingly, we screened 295 diagnostic childhood T-ALL samples treated on prospective BFM protocols (Kox et al., 2010) for IL7R mutations and identified 30 somatic mutations (10.5%). All the mutations were at the transmembrane domain encoded by exon 6, and all but one were in-frame insertions and deletions (Table II); 27 of the 30 cases included an insertion of cysteine. The patients with mutations were younger and tended to have higher white blood cells counts at diagnosis (Table S1). There was no association between the presence of the mutations and response to therapy, and a similar proportion (10%) relapsed. Notch mutations, which hallmark ∼50% of T-ALLs (Clappier et al., 2010; Kox et al., 2010), were detected in 64% of the IL7R-mutated leukemias compared with 48% of the T-ALL without IL7R mutations (P = 0.1, χ² test).

We and others (Hertzberg et al., 2010; Yoda et al., 2010) have recently reported an activating F232C mutation in CRLF2 that introduces cysteine in the juxtamembrane domain. All but three of the mutations observed in IL-7R included the addition of cysteine. To study whether the gain-of-function of the mutated IL-7R could be attributed to the addition of cysteine, we replaced the cysteines in the S815C and in the IL-7R InsPPCL alleles with glycine. These newly mutated receptors were expressed at the cell surface of transduced BaF3 cells (Fig. 4 A). Elimination of the cysteine abrogated the cytokine-independent growth (Fig. 4 B) and the constitutive phosphorylation of Stat5 (Fig. 4 C). Response to TSLP was not altered, confirming that “cysteine-lacking” mutated IL-7R formed a functional TSLP receptor with CRLF2 (Fig. 4 C). These results suggest that the presence of cysteine is critical for the gain-of-function phenotype both in the context of insertion of additional amino acids and as a point mutation.

The experimental insertion of cysteine into the transmembrane domain of the erythropoietin receptor (Constantinescu et al., 2001; Lu et al., 2006) activated the receptor by causing ligand-independent receptor dimerization. Indeed, we observed marked homodimerization of the IL-7R InsPPCL in protein analysis under nonreducing conditions (Fig. 4 D). Interestingly, CRLF2 was not present in these dimers, which is consistent with the observation that this mutant protein does not require CRLF2 for its activation and with the presence of these mutations in T-ALLs in which CRLF2 is not highly expressed. Our current and recent observations (Chapiro et al., 2010; Hertzberg et al., 2010; Yoda et al., 2010) of mutations in IL-7R and CRLF2, respectively, suggest that the addition of cysteine to the juxtamembranous domain is a general mechanism for mutational activation of type I cytokine receptors in leukemias.

These discoveries are a prime example of the interconnection of development and leukemia (Izraeli, 2004). Although loss-of-function mutations in IL-7R perturb lymphoid development, the novel gain-of-function mutations described here are associated with both BCP- and T-ALLs. Activating mutations of IL-7R are present in leukemias of the same lineage, for which it is developmentally required (T cells), or in B cell precursor leukemias that aberrantly express a T cell and monocytic receptor (CRLF2). The broad significance of this pathway in leukemia is further suggested by the aberrant deregulation of CRLF2–JAK–STAT pathway in BCP-ALL (Bercovich et al., 2008; Constantinescu et al., 2008; Kearney et al., 2009; Mullighan et al., 2009; Russell et al., 2009; Hertzberg et al., 2010; Yoda et al., 2010) and by these newly described IL7R mutations and the previously reported JAK1 activating mutations in T-ALL (Flex et al., 2008; Hornakova et al., 2010). Similar to mutational activations of other growth factor receptors in cancer (Gazdar, 2009), our observations have potential implications for targeted therapy.

DNA and RNA were derived from leukemic bone marrow of children with ALL enrolled in a prospective BFM (Berlin Frankfurt Munster) and AIEOP (Associazione Italiana Ematologia Oncologia Pediatrica Group) ALL 2000 study (Conter et al., 2010). The T-ALL and the DS-ALL cohorts were previously described (Hertzberg et al., 2010; Kox et al., 2010). All specimens were collected with informed consent and approval of local and national ethic committees. Samples were anonymized for the study. The study was approved by the Israeli Health Ministry Ethic committee, approval # 920070771.

CRLF2 levels were analyzed by the TaqMan Gene Expression Assay Hs00913509_s1 (Applied Biosystems), as we have previously reported (Cario et al., 2010; Hertzberg et al., 2010). The presence of the fusion transcript P2RY8-CRLF2 was analyzed by RT-PCR, as we have previously reported (Cario et al., 2010; Hertzberg et al., 2010), using primers designed in the first exon of P2RY8 (5′-GGACAGATGGAACTGGAAGG-3′) and the third exon of CRLF2 (5′-GTCCCATTCCTGATGGAGAA-3′). PCR product was ∼511 bp. Immunoglobulin heavy chain locus (IGHα) translocations were identified by fluorescence in situ hybridization (FISH) on interphase nuclei using the Vysis LSI IGHα Dual Color Break-Apart Rearrangement Probe (Abbott Molecular). The nuclei were counterstained with DAPI (4′, 6-diamino-2-phenylindol). Results were recorded using a fluorescence microscope (DMRB; Leica) fitted with a 100×/1.30 oil objective, charge-coupled device camera, and digital imaging software from MetaSystems (Isis FISH imaging system).

For the purpose of this study, the CRLF2-positive group consisted of BCP-ALL specimens that were either positive for the P2RY8-CRLF2 fusion or to the IGHα CRLF2 translocation or had CRLF2 RNA expression levels >20 fold of the median value of CRLF2 expression of 555 unselected BCP-ALLs (Cario et al., 2010). The CRLF2-negative group consisted of BCP-ALL specimens negative for the genomic rearrangements and with expression level at the median and below. JAK2 and CRLF2 mutations were identified as previously described (Hertzberg et al., 2010)

We used intronic primers of human IL7R sequence (available from GenBank/EMBL/DDBJ under accession no. NM_002185.2; Table S2) to amplify exons 1–8 of the gene with PCR. Fast start Taq DNA polymerase (Roche) was used with the following thermal cycling conditions: 1 cycle of 94°C for 5 min, followed by 5 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C 30 s, followed by 30 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C 30 s, followed by 1 cycle of 72°C for 7 min. We then analyzed the fragments by denaturing high-performance liquid chromatography as described before (Bercovich et al., 2008; Hertzberg et al., 2010).

Fragments with abnormal chromatography patterns were sequenced directly by the primer used for PCR, or cloned into TA vector pCR2.1 (Invitrogen). DNA from plasmid clones was extracted by AccuPrep Spin Miniprep kit (Bioneer) and sequenced.

The human IL7R cloned into the MSCViresGFP retrovirus (encoding GFP) was used as a template for the generation of mutations by site-directed mutagenesis (QuikChange II XL; Stratagene). The plasmid sequences were verified by sequencing.

Parental BaF3 cells and those expressing human CRLF2 receptor (BaF3/CRLF2; Hertzberg et al., 2010) were transduced with the retroviruses containing WT or mutant IL7R. Transduced cells were sorted by flow cytometer 2–4 d later. PE anti–human CRLF2 (322806; BioLegend) and Alexa Fluor 647 anti–human CD127 (IL-7R, 317606; BioLegend) antibodies were used. Cytokine withdrawal assays were performed as previously described (Hertzberg et al., 2010)

For Western blot analyses, 10% polyacrylamide linear gels were used. BaF3 cells were harvested after 5 h of IL-3 deprivation, and then stimulated with either 10 ng/ml IL-3 or 100 ng/ml TSLP for 25 min. The antibodies used were anti–phospho-STAT5 Tyr 694 (Epitomics), anti-STAT5, anti-S6 ribosomal protein and anti-phospho-S6 ribosomal protein Ser 235/6 (Cell Signaling Technology).

For Western blot analyses of disulfide-bonded dimmers, 8% polyacrylamide linear gels were used. For nonreducing conditions, proteins were boiled in SDS buffer without 2-mercaptoethanol. Antibodies used were anti–human IL-7R (R&D Systems) and anti–human CRLF2 (BioLegend).

Fig. S1 shows expression of mutated IL7R. Fig. S2 shows FACS analysis of parental BaF3 cells in the presence of IL-3. Fig. S3 shows CRLF2 and IL7R Ins PPCL respond to TSLP. Table S1 shows the clinical characteristics of T-ALL patients with IL7R mutations. Table S2 shows primers used for genomic amplification of IL7R exons.

We thank M. Happich, B. Fedders, Drs. S. Korem and C. Eckert for their assistance. The IL7R vector was a gift from D. Weinstock.

This study was funded by grants from the Israel Science Foundation Legacy program, Waxman foundation and Children for Leukaemia UK (S. Izraeli), Israel Trade Ministry (D. Bercovich), German Ministry for Education and Research (BMBF) in the context of the NGFN Leukemia Net and the Tour der Hoffnung (A. Kulozik), Deutsche Krebshilfe, and the Madeleine-Schickedanz-Kinder-krebsstiftung (G. Cario and M. Schrappe), EU NMP4-SL-2009-228971 (GteK), Associazione Italiana per la Ricerca sul Cancro (AIRC), Fondazione Cariplo, and Ministero dell’Istruzione dell’Università e della Ricerca (MIUR; A. Biondi and G. Cazzaniga).

This work was performed in partial fulfillment of the requirements for a PhD degree of Chen Shochat, Noa Tal and Ithamar Ganmore, Sackler Faculty of Medicine, Tel Aviv University, Israel. This is a study of the international BFM study group.

The authors have no competing financial interests.