Novel activating mutations lacking cysteine in type I cytokine receptors in acute lymphoblastic leukemia

Interleukin-7 receptor α (IL7R) dimerizes with cytokine receptor-like factor 2 (CRLF2) to form the receptor for thymic stromal lymphopoietin and with interleukin-2 receptor γ to form the receptor for IL-7.^1,2  We and others described acquired activating mutations in acute lymphoblastic leukemias (ALL) that insert cysteines into the juxtamembrane domains of IL7R or CRLF2 causing ligand independent dimerization via disulfide bonds.^3-5  The creation of disulfide bonds is critical for the activation of the receptors because elimination of the cysteines abrogated the cytokine independent growth.^4,6,7 

Here we report and analyze a novel class of noncysteine mutations in cytokine type I receptors in ALL leading to ligand independent activation.

All specimens were collected with an informed consent and the approval of ethics committees.⁴  Samples were anonymized for the study. The study was approved by the Israeli Health Ministry Ethic committee, approval # 920070771.

Mutation detection and analysis using appropriate primers (supplemental Table 1; see supplemental Data available at the Blood Web site). was performed as described.^4,8  The human IL7R cloned into the MSCV-IRES-GFP and the human CRLF2 cloned into pMX-Puro retroviral vectors were used as templates for the generation of mutations by site-directed mutagenesis (QuikChange; Stratagene).

BaF3 cells were transduced with retroviruses and were sorted by flow cytometry (FACSAria; BD Biosciences) 2 to 4 days later using appropriate antibodies. BaF3 growth assays and immunoprecipitation and western analyses were performed as described previously.⁹ 

Mutated and wild-type (WT) cDNAs of human IL7R, CRLF2, and interleukin-2 receptor γ were inserted in pcDNA3.1/Zeo vector upstream of either the hGluc1 or hGluc2 fragments of Gaussia princeps luciferase¹⁰  and were cotransfected into HEK293 cells (see supplemental Methods). Signal intensities were read on an Infinite 200 reader (TECAN).

Six-week-old female Balb/c mice were injected intravenously with 1*10⁶ BaF3 cells transduced with either WT or mutated CRLF2 or IL7R. Survival of the mice was monitored, and tissues from the sick mice were subjected to flow cytometry analysis (Galios; Coulter Inc.).

During our previously reported screens of childhood ALLs^4,8  we identified mutations in IL7R and in CRLF2 that did not introduce cysteines (Figure 1A). All mutations were heterozygous, somatic, and the mutated mRNA was expressed (Figure 1B). Examination of published data showed additional similar mutations in IL7R (supplemental Table 2). These mutations cluster within a transmembranous region internal to the area afflicted by the cysteine mutations (Figure 1A; supplemental Figure 1). To test if the somatic mutations in IL7R and CRLF2 are gain-of-function mutations, BaF3 lines expressing either WT or mutated IL7R or CRLF2 were created. All proteins were expressed at the cell membrane (supplemental Figure 2). Mutated IL7R^insEKV, IL7R^V253G, and CRLF2^{ins EIM} enabled cytokine-independent growth of BaF3 cells. CRLF2^insEIM required the formation of the thymic stromal lymphopoietin receptor by coexpression of IL7R^WT (Figure 1C,D). One construct IL7R^insGEA did not transform BaF3 (Figure 1C). The transforming activity was confirmed in vivo in syngeneic Balb/c mice that developed fatal leukemia after intravenous injections of BaF3 cells expressing IL7R^insEKV, IL7R^V253G, or CRLF2^insEIMIL7R^WT (Figure 1E), manifested by infiltration of the blood (Figure 1F), bone marrow, and spleen (Figure 1H-I) with the transformed cells. Similarly, subcutaneous injection caused lymphoma (Figure 1G). Thus noncysteine mutational activation of IL7R and CRLF2 is leukemogenic.

Biochemical analysis was consistent with the growth assays. Stat5 (Figure 2A), Jak1, Jak2, and Stat3 (Figure 2B) were phosphorylated in the absence of cytokine in BaF3 cells transduced with the IL7R^insPPCL (positive control that constitutively activates the JAK-STAT pathway),⁴  IL7R^V253G, and IL7R^insEKV but not in cells expressing IL7R^WT or IL7R^insGEA. In IL7R^insEKV we found higher expression of Jak2, which correlated with the higher activation capacity of this mutant in BaF3 cytokine withdrawal assays. Similarly, coexpression of IL7R^WT with CRLF2^insEIM, but not the expression of CRLF2^insEIM by itself, caused constitutive phosphorylation of Stat5 and Jak2 (Figure 2C,D) consistent with the known binding of Jak2 to CRLF2¹¹ .

To assess whether the functional and biochemical evidence of receptor activation is associated with the oligomerization of IL7R and CRLF2 mutated proteins in cell membranes, we used the luciferase protein fragment complementation assay¹⁰  (supplemental Figure 3A). HEK293 cells were transduced with homodimeric or heterodimeric combinations of IL7R and CRLF2 fused to hGluc1 or hGluc2 (supplemental Figure 3B). After transduction, luciferase expression and receptor expression were analyzed in live cells (supplemental Figure 4). Dimerization was calculated by dividing luminescence by the mean fluorescence intensity of each receptor thereby enabling normalization of the luminescence signal for experimental variability resulting from transfection efficiency.

Consistent with the requirement of complementation for luciferase expression, none of the single-receptor hGluc1 or hGluc2 constructs that were individually transfected in HEK293 cells generated a signal (supplemental Figure 3C). Yet when the IL7R^WT homodimer or heterodimer was expressed, a signal was obtained reflecting a basal level of receptor dimerization, as recently reported.¹²  We then analyzed the influence of the somatic mutations of IL7R and CRLF2 on receptor dimerization. As shown in Figure 2E, IL7R^{ins EKV}, IL7R^V253G, and CRLF2^insEIM increased receptors dimerization (P < .05) whereas the IL7R^insGEA mutant that failed to provide cytokine-independent survival of BaF3 cells, or to cause Stat5 phosphorylation, also did not increase receptor dimerization (Figure 2E). Thinking that it may be a “passenger” mutation, we looked for additional activating mutations but found none (supplemental Table 3).

As CRLF2^insEIM alone increased receptor dimerization but did not induce downstream constitutive signaling or transformed BaF3 cells, we concluded that there is no absolute correlation between dimerization and activation.

We next proceeded to decipher the amino acids that are important for the highly activating IL7R^insEKV mutation. We designed 6 variants of the EKV mutation: AKV, EAV, EKA, EK, KV (including Del^AL like the natural mutation), and ins V (Figure 2F). BaF3 lines expressing each variant were created (supplemental Figure 5). The variants IL7R^insEAV and IL7R^insEKA enabled cytokine-independent growth of BaF3 cells, whereas IL7R^insAKV, IL7R^insEK, IL7R^insKV, and IL7R^insV did not transform BaF3 (Figure 2G). Biochemical analysis was consistent with the growth assays. Stat5 was phosphorylated in the absence of cytokine in BaF3 cells transduced with the IL7R^insEAV and IL7R^insEKA but much less in cells expressing IL7R^insAKV, IL7R^insEK, IL7R^insKV, and IL7R^insV (Figure 2H).

ConSurf¹³  calculations with IL7R showed that the TMD segment manifests a unique evolutionary conservation pattern with i/i+4 periodicity, equivalent to that of a perfect α-helix (supplemental Figure 6). The conserved positions would reside in the same helix face, providing an interface for IL7R homodimerization. All of the natural and experimental mutations can be interpreted assuming that activation requires IL7R dimerization with the cytosolic kinases in predefined orientation with respect to each other. The conserved face of the TMD helix in WT IL7R marks the orientation. Dimerization along a different TMD helix face would not activate because the kinases will not be facing each other. Molecular detailed interpretations of the mutations’ effect are provided in supplemental Figures 7 and 8 and supplemental Table 4.

Taken together, these data indicate that the functional and biochemical evidence for receptor activation by mutations lacking cysteine correlates with ligand independent dimerization. There are several examples of TMD mutations in receptors that cause constitutive activation without introducing cysteine: V664E in the Neu receptor tyrosine kinase^14-16  and the dimerization inducing mutations in MPL^17,18  that was identified in rare familial thrombocytosis patients.¹⁹  Our observations of noncysteine transmembrane mutations in type I cytokine receptors in patients with ALL demonstrate for the first time that such somatic mutations activate these receptors through ligand-independent dimerization and are leukemogenic.

This work was supported by Israel Science Foundation Legacy and iCORE programs, the Israel Cancer Research Foundation, Swiss Bridge Foundation, USA Israel Binational Science Foundation, Waxman Cancer Research Foundation, and William Lawrence and Blanche Hughes Foundation. C.S. and N.T. are PhD candidates at Tel Aviv University and this work is submitted in partial fulfillment of the requirement for a PhD.

Contribution: S.I. designed the study and wrote the paper; C.S. performed research, analyzed data, and wrote the paper; N.T., N.G., Y.B., and N.B.-T. performed research and analyzed data; V.G., S.N.C., M.R.M., J.-C.T., and D.B. provided reagents and scientific expertise; and O.R.B., G.C., A.E.K., A.B., and M.U.M. provided reagents and data.

Conflict of interest disclosure: The authors declare no competing financial interests.

Correspondence: Shai Izraeli, Sheba Medical Center, Tel Hashomer Ramat Gan, 52621, Israel; e-mail: shai.izraeli@sheba.health.gov.il.