Triggering the TCR Developmental Checkpoint Activates a Therapeutically Targetable Tumor Suppressive Pathway in T-cell Leukemia

Developmental checkpoints in stem/progenitor cells are critical to their determination, commitment, and differentiation into distinct lineages. Cancer cells often retain expression of lineage-specific checkpoint proteins, but their potential impact in cancer remains elusive (1). T lymphocytes mature in the thymus following a highly orchestrated developmental process that entails the successive rearrangements and expression of T-cell receptor (TCR) genes. TCRδ rearrangements first occur at the CD5⁺, CD1a⁻, CD4/8 double-negative (DN) stage, followed by concurrent TCRγ and TCRβ rearrangements coinciding with CD1a expression. Although productive TCRγ and TCRδ rearrangements will determine the assembly of a TCRγδ, a complete productive TCRβ gene rearrangement will first allow surface expression of a pre-TCR complex formed by the assembly of the TCRβ chain with a pre-Tα (pTα) invariant chain. Pre-TCR surface expression is a critical checkpoint for developing αβ T-cell precursors to expand and mature into CD4/CD8 double-positive (DP) cortical thymocytes and for the initiation of Vα–Jα rearrangements. These rearrangements continue until a TCRα chain is formed that can associate with the already formed TCRβ chain to assemble a complete mature TCRαβ at the cell surface (2, 3). Cell surface expression of the TCR is a critical developmental checkpoint in establishing MHC restriction and in shaping the immunologic repertoire. Low-affinity recognition of self-peptide/major histocompatibility complexes (self-pMHC) presented by thymic epithelial cells to the TCR of CD4⁺CD8⁺ DP cortical thymocytes transduces positive selection signals and commitment to either the CD4 or CD8 lineages. DP thymocytes not receiving these signals die by lack of stimulation, whereas those that recognize self-pMHC with high affinity undergo TCR-mediated apoptosis and negative selection (4, 5).

T-cell acute lymphoblastic leukemia (T-ALL) results from the leukemic transformation of thymic cell precursors and their arrest at specific stages of differentiation (6–8). Despite recent and extensive insights into the molecular and cellular mechanisms responsible for T-ALL onset and progression, survival rates remain around 50% and 70% in adult and pediatric cases, respectively, calling for the search for novel therapeutic options. In T-ALL, leukemic transformation of maturing thymocytes is caused by a multistep pathogenesis involving numerous genetic abnormalities that drive normal T cells into uncontrolled cell growth and clonal expansion. T-ALL is classified into subgroups based upon exclusive gene rearrangement and/or expression of a limited set of transcription factors including TAL1, LMO1/2, TLX1/3, associated with the specific arrest at distinct stages of T-cell differentiation (8–10). Across these molecular subclasses, a number of other recurrent genetic alterations are found, including inactivation of the CDKN2A tumor suppressor locus, NOTCH1 pathway activating mutations, and many others (reviewed in ref. 11).

In this study, we hypothesized that tissue homeostatic regulators may be amenable to reactivation in tumor cells and demonstrate that experimentally induced TCR signaling in T-ALL induces cell death and a molecular program similar to negative selection of thymic T-cell progenitors. Importantly, switching on this TCR-induced cell death program dominates the multiple antiapoptotic and proproliferative mechanisms controlled by the oncogenes and altered tumor suppressors of T-ALL. These data provide a strong rationale for targeted therapy based upon anti-CD3 treatment in TCR-expressing T-ALL.

To investigate the role of TCRαβ activation by a defined antigen in T-ALL, we stably expressed the negatively selecting Marilyn TCR-HY (Vα1.1Vβ6) specific for the HY male antigen (DBY) in the TCR-negative ALL-SIL T-ALL cell line. In vitro coculture of ALL-SIL/TCR-HY cells with splenocytes pulsed with increasing doses of DBY peptide, but not the unrelated OVA peptide, induced leukemic cell death (Fig. 1A and quantified in Fig. 1B). DBY peptide did not induce apoptosis of parental TCR-negative ALL-SIL cells nor of ALL-SIL/TCR-HY cells cultured in the absence of MHC-restricted antigen-presenting syngeneic splenocytes (Fig. 1A). These data demonstrate that TCR activation by the cognate peptide–MHC complex results in T-ALL cell death. To determine the effect of persistent TCR stimulation on T-cell leukemogenesis, we generated double transgenic mice (hereafter referred to as TEL-JAK2/TCR-HY mice) expressing the TEL–JAK2 fusion oncogene (12) and the Marilyn TCR-HY (13) in lymphoid progenitors. As compared with TEL-JAK2/TCR-HY double transgenic females, which developed rapid and fully penetrant T-ALL, the majority of TEL-JAK2/TCR-HY males either failed to become leukemic or developed leukemia with markedly delayed latency (Fig. 2A). Flow cytometry analysis showed that of seven diseased male mice analyzed, two developed B-cell lymphoma/leukemia whereas five developed T-cell leukemia (Fig. 2B). Strikingly, in the five T-ALL cases, transgenic TCR-Vβ6 cell surface expression was low or absent, and leukemic cells often expressed either no TCR or an endogenous TCR (Fig. 2C), suggesting that in these mice, expression of TCR-HY, but not of non–HY-specific TCR complexes, was counterselected during leukemogenesis. We conclude that self-antigen TCR stimulation suppresses thymocyte malignant transformation, thus delaying leukemogenesis in male mice.

To investigate whether delayed leukemogenesis in TEL-JAK2/TCR-HY male mice could be attributed to direct antigenic stimulation of TCR-expressing leukemic cells, female TEL-JAK2/TCR-HY CD45.1⁺ T-ALL cells were transplanted to either female (HY-negative) or male (HY-positive) syngeneic CD45.2⁺ recipient mice. As compared with female recipients, which rapidly developed fatal T-ALL, most recipient males failed to develop leukemia and survived longer (Fig. 2D). In contrast to male mice sacrificed simultaneously, female recipients presented high infiltration of CD45.1⁺ leukemic cells in spleen, lymph nodes, and bone marrow (Fig. 2E and F, and data not shown). The resistance of male mice to TEL-JAK2/TCR-HY–induced leukemogenesis depended on HY antigen MHC presentation but not upon an adaptive immune response, because T-ALL developed efficiently following transplantation of TEL-JAK2/TCR-HY leukemic cells in male allogeneic immunodeficient recipient mice (Fig. 2G), but not in male syngeneic immunodeficient recipients (Fig. 2H). These results show that the TCR antileukemic effect depends upon its stimulation by the cognate peptide–MHC complex.

We next investigated whether anti-CD3 treatment could mimic TCR signaling to induce leukemic cell apoptosis. Unlike control ALL-SIL cells, TCR-HY–expressing cells underwent massive apoptosis when treated with anti-CD3/anti-CD28 or anti-CD3 alone (data not shown), as revealed by increased levels of cell-surface Annexin V, increased cleavage of caspase-3 and -7, and reduced cell expansion (Fig. 3A–D). Importantly, anti-CD3/anti-CD28 treatment did not affect cell-cycle progression per se (Fig. 3E and data not shown). To investigate the signaling properties of the mouse TCR-HY complex in human leukemic ALL-SIL cells, we first demonstrated that CD3 cross-linking induced TCR clustering at the ALL-SIL cell surface (Fig. 3F). Next, phosphokinase-array analysis of anti-CD3/anti-CD28–stimulated ALL-SIL/TCR-HY and normal human DP thymocytes showed increased phosphorylation of overlapping TCR signaling components, including the key downstream second messengers ERK, JNK, and AKT (Fig. 3G). Anti-CD3–induced phosphorylation of these proteins, and not of the cytokine signaling STAT3 protein, was confirmed by flow cytometry (Fig. 3H and data not shown). Of note, inhibition of TCR signaling by cyclosporine or the MEK kinase inhibitor PD184352 impaired anti-CD3–induced apoptosis and expression of the CD69 and CD25 T-cell activation markers (Fig. 3I and J). Together, these results indicate that the TCR/CD3 signaling module is functional in T-ALL cells, and that its activation leads to apoptosis. Importantly, similar results were observed in primary T-ALL (Supplementary Fig. S1A–S1C).

To test whether anti-CD3 stimulation of TCR-expressing T-ALL diagnostic samples also induced apoptosis, primary human T-ALL cells (see Supplementary Table S1 for the immunophenotypic and genetic features of the diagnostic samples used) belonging to the major molecular oncogenic T-ALL subclasses were treated with anti-CD3/anti-CD28. Massive T-ALL apoptosis and TCR signaling activation were observed in TCR-expressing T-ALL but not in TCR-negative cases, regardless of their molecular oncogenic subtype (Fig. 4A–D). We next analyzed the transcriptional profiles of TCR-expressing ALL-SIL cells either unstimulated or after TCR stimulation and found in the ALL-SIL TCR-induced transcriptional profile a significant enrichment for the TCR-induced negative selection signature of normal thymic progenitors (ref. 14; Fig. 4E and F). This shows that T-ALL apoptosis induced by TCR stimulation is associated with signaling cues resembling those of TCR-induced negative selection in normal T-cell progenitors.

To assess the therapeutic potential of TCR stimulation by anti-CD3 treatment in vivo, we first inoculated female syngeneic mice with transgenic TEL-JAK2/TCR-HY leukemic cells and the following day initiated treatment with either an anti-CD3ϵ monoclonal antibody (mAb) known to induce negative selection in normal thymic progenitors (15) or an isotypic control antibody for 5 consecutive days. Strikingly, both peripheral blood leukemia and fatal disease were impaired in anti-CD3–treated mice, whereas all mice treated with control antibody rapidly succumbed to leukemia (Fig. 5A and B). Importantly, treatment with anti-CD3 mAb was also effective against TEL-JAK2 T-ALL cells expressing endogenous surface TCRαβ (Fig. 5C and D). Of note, anti-CD3 treatment only marginally prolonged the leukemia-free survival of mice infused with CD3ϵ^−/− TEL-JAK2 T-ALL cells (Fig. 5D). This shows that the extended survival seen in anti-CD3–treated mice inoculated with CD3ϵ^+/+ TEL-JAK2 T-ALL (Fig. 5D) critically depends upon engagement of the TCR/CD3 complex expressed by leukemic cells. To ensure that the results were not restricted to this particular model of activated tyrosine kinase–induced T-ALL, we extended these analyses to Ikzf1/Ikaros-mutant–induced T-ALL. Although Ikaros is rarely found mutated in T-ALL, recent evidence points to a suppressive function of Ikaros in NOTCH-mutated T-ALL cells (16). Treatment with anti-CD3 mAb inhibited leukemia expansion in this model (Fig. 5E), which translated into increased survival of anti-CD3–treated mice (Fig. 5F).

Finally and most importantly, we investigated whether anti-CD3 treatment also displayed antileukemic activity against human CD3-positive T-ALL diagnostic cases. Mice were xenotransplanted with cells from seven distinct human T-ALL cases, five of them expressing a TCR at the cell surface, one (UPNT374) showing a heterogeneous TCR expression pattern, and the last one (UPNT525) being negative for TCR cell-surface expression. Mice were treated 24 hours later with either anti-CD3 mAb or an isotypic control antibody. For all TCR⁺ T-ALL samples, anti-CD3 treatment showed strong antileukemic effects, as shown by delayed dissemination in blood, bone marrow, and other organs (Fig. 6A–D), an effect that translated into enhanced recipient mouse survival (Fig. 6E and Supplementary Fig. S2). In contrast, leukemic mice inoculated with TCR-negative T-ALL cells were insensitive to anti-CD3 treatment (Fig. 6B and Supplementary Fig. S2). Mice infused with UPNT374 cells only transiently responded to anti-CD3 (Fig. 6B and Supplementary Fig. S2). Interestingly, leukemic cells recovered from anti-CD3–treated mice showed low or absence of TCR cell-surface expression in contrast to the leukemic cells that expanded in control IgG-treated mice (Fig. 6F). This confirms the specific depletion of TCR-positive leukemic cells in response to TCR activation. We next investigated whether anti-CD3 treatment could also result in antileukemic effects in a curative setting. First, NOD/SCID/IL2Rgc^−/− (NSG) mice were inoculated with the luciferase-expressing human ALL-SIL/TCR-HY cells and then treated with either a control or an anti-CD3 mAb when mice presented leukemia engraftment in bone marrow (day 22 after inoculation, data not shown). Comparison of the two groups by live imaging showed a specific antileukemic effect of anti-CD3 treatment (Fig. 7A, top, and 7B). To investigate whether anti-CD3 antileukemic effects were dependent upon TCR signaling, we sought to silence the expression of LAT, a critical adaptor in TCR signaling in thymocytes and mature T cells (17). shRNAs to LAT resulted in LAT expression knockdown in ALL-SIL/TCR-HY cells (Supplementary Fig. S3A) and had no effect on TCR cell-surface expression (Supplementary Fig. S3B). LAT knockdown impaired anti-CD3–induced in vitro apoptosis (Supplementary Fig. S3C) and antileukemic effects in vivo (Fig. 7A and B and Supplementary Fig. S3D and S3E). As expected, LAT expression knockdown was maintained in leukemic cells retrieved from control and anti-CD3 mAb-treated mice (Supplementary Fig. S3F). In a second experiment, mice xenotransplanted with cells from T-ALL UPNT419 were treated with the anti-CD3 or control mAb when leukemic cell burden reached 10% to 20% of bone marrow cells. Leukemia expansion was strongly delayed in anti-CD3–treated mice, in contrast to the rapid expansion of leukemic cells in control mice (Fig. 7C). Despite the reemergence of peripheral blood leukemic cells upon cessation of treatment, survival of anti-CD3–treated mice was significantly prolonged (Fig. 7D). Thus, antibody-mediated TCR engagement results in striking antileukemic effects in xenotransplanted human T-ALLs in both preventive and curative settings.

Current T-ALL therapies involve complex, often toxic chemotherapeutic regimens. Although T-ALL outcome has improved with current therapies, survival rates remain only around 50% and 70% at 5 years in adult and pediatric T-ALL, respectively (18, 19). The genetic bases of T-ALL progression and maintenance are well characterized but have not translated so far into targeted therapies (11). There is thus unmet need for new treatments to offer therapeutic options for refractory disease and to prevent relapse. We previously observed that expression of a TCR transgene in a TCR-negative T-ALL cell line can induce cell death, provided these cells are subjected to differentiation-inducing cues (20). We report here that chronic/strong TCR signaling causes massive apoptosis of primary T-ALLs that express an endogenous TCR and shows potent tumor-suppressive function in vivo. These findings call for the incorporation of TCR-directed therapies in current treatment regimens of CD3-expressing T-ALL. Muromonab-CD3 (OKT3) was approved by the FDA in 1985 for therapy of acute, glucocorticoid-resistant rejection of allogeneic renal, heart, and liver transplants (21) and was in fact the first mAb introduced in the clinic. Since then, a number of other monoclonal antibodies to CD3 were developed (22) that may prove superior to OKT3 in T-ALL treatment. Of note, encouraging response to OKT3 therapy was reported in an adult patient with an aggressive and chemotherapy-resistant T-ALL, but the basis of this response was not studied (23).

A major drawback of current chemotherapeutic regimens in T-ALL is the frequent resistance to treatment and relapse. Leukemia initiating cells (LIC) from residual disease are thought to be responsible for relapsing cases (24, 25). In addition, resistance of T-ALL to chemotherapy is in part linked to the recurrent genetic abnormalities selected during disease progression, e.g., inactivation of the PTEN tumor suppressor gene and the resulting activation of PI3K/AKT signaling (26, 27). Whether administration of anti-CD3 therapy during the remission phase or its association with conventional chemotherapy regimens could target LICs and/or bypass molecular mechanisms of primitive resistance represent promising directions to be explored in future prospective studies.

T cells mature in the thymus following a highly orchestrated process controlled by both cell-intrinsic (e.g., transcription factors) and cell-extrinsic (e.g., stroma-derived cytokines/chemokines) molecular cues (28, 29). Cell-surface TCRαβ expression in DP thymocytes allows recognition of specific self-MHC/peptide to transduce a positive selection signal and maturation into single-positive thymocytes. DP thymocytes not receiving this signal die through lack of stimulation, whereas those whose TCR binds too strongly to self-MHC/peptide undergo activation-induced apoptosis and negative selection (5, 30). In both situations, TCR binding to self-pMHC is the triggering event, but how TCR engagement leads to such divergent outcomes (survival and proliferation versus death) remains unclear (31). Remarkably, these two contrasting processes are driven by a TCR signaling machinery of qualitatively similar composition (30). Available evidence indicates that the difference lies in the molecular interpretation of signals of different strength, which may rely on compartmentalization of key signaling players (32), and in the induction of divergent, complex transcriptional responses (14, 33). Indeed, it has been shown that a small increase in ligand affinity for the TCR leads to a marked change in the subcellular localization (plasma membrane for negative selecting ligands vs. Golgi complex for positively selecting ones) of essential adaptors of the RAS signaling pathway. This compartmentalization induces the conversion of a small change in analogue input (affinity for ligand) into a digital output (positive vs. negative selection; refs. 32, 34). Our results show that the antileukemic activity of anti-CD3 mAb depends upon TCR signaling cues, as it was impaired upon LAT silencing in T-ALL cells. This also suggests that, in the present experimental setting, antibody dependent cell-mediated cytotoxicity (ADCC)–like responses of the host play a minor role in anti-CD3 antileukemic properties. It is likely that in patients with T-ALL eligible for anti-CD3 therapy (about half of pediatric and 30% of adult cases), ADCC-like responses will play a more important role than in immunosuppressed mice and will add therapeutic value to the intrinsic antileukemic effects of anti-CD3. In addition, in a clinical context, TCR downregulation and/or selection of a TCR-negative subclone, as observed here in a xenotransplanted T-ALL under pressure of anti-CD3 treatment, could constitute a potential escape mechanism to treatment. These findings need to be explored in clinical trials. More extensive studies on the composition and function of TCR-induced signalosomes formed in T-ALL stimulated by anti-CD3 and the identification of critical components in T-ALL of the transcriptomic signature akin to that of thymic negative selection (14, 35, 36) should provide information on the molecular pathways involved in anti-CD3–induced apoptosis in leukemic cells. These pathways could in turn constitute new pharmaceutical targets to treat T-ALL.

Although signaling from the B-cell receptor (BCR) and pre-BCR was shown to promote B-cell malignancies (37, 38), the role of TCR signaling in T-ALL has thus far remained controversial. Studies in T-ALL mouse models indicated a pro-oncogenic role for TCR signaling (39–42). However, these data were based on transgenic TCR systems, which do not reflect the expression levels and receptor diversity found among patients with T-ALL where TCR expression is heterogeneous or absent (6). Moreover, gene inactivation studies in other mouse T-cell leukemia models have shown that TCR expression is not essential for T-ALL development (43, 44). The current work demonstrates that anti-CD3–mediated activation of endogenous TCR in human T-ALL has an antileukemic function. T-ALL often arises from immature T-cell precursors before the stage of negative selection (6, 42). Consequently, contingent expression of a non–negatively selecting TCR will render T-ALL cells sensitive to TCR-activating apoptotic signals that mimic negative selection, as demonstrated here by anti-CD3 treatment in vitro and in vivo. However, during T-ALL development, it is possible that selective events enable cells to escape post-malignancy negative selection, such as loss of TCR surface expression, as found in a subset of patients with T-ALL (6) and as demonstrated here in TEL-JAK2/TCR-HY double transgenic male mice. Such leukemias could be amenable to other therapies targeting downstream TCR signaling effectors.

The dual role of developmental molecular pathways in organogenesis and tumorigenesis is increasingly recognized, the modulation of which may provide potential therapeutic opportunities (1). In line with this, experimental restoration of cell-surface expression of Igα and Igβ in Ph⁺ B-ALL was also shown to result in cell death (45). In our study, we found that reactivation in TCR-positive T-ALL blasts of the lineage-specific checkpoint control normally set by TCR signaling during T-cell development displays antitumoral functions. Importantly, despite the multiple and complex oncogenic mechanisms driving T-ALL, which include antiapoptotic (46) and proproliferative (47) signaling cues, this TCR-dependent checkpoint remains switchable to induce massive tumor cell apoptosis. Thus, reactivation of similar lineage-specific developmental checkpoints in malignancies originating from other lineages and tissues could provide a novel class of therapeutic targets in cancer.

TEL-JAK2 mice were bred with Marilyn transgenic, RAG2-deficient or CD3ϵ-deficient mice (all maintained on the C57BL/6 background), as described previously (43). Swiss Nude and NSG mice were purchased from Charles River Laboratories. IK^L/L mice, carrying two alleles of the hypomorphic IK^L allele (48), were a generous gift from Dr. P. Kastner (IGBMC, Strasbourg, France). Mice were maintained under specific pathogen-free conditions in the animal facilities of the Institut Curie (Orsay, France) or University of Algarve (Faro, Portugal). NSG mice were maintained under constant antibiotic treatment (Baytril 0.01% in drinking water). All experimental procedures were performed in strict accordance with the recommendations of the European Commission (Directive 2010/63/UE), French National Committee (87/848), and Portuguese authorities (Decreto-Lei n°113/2013) for the care and use of laboratory animals. Mice were euthanized when terminally ill, as evidenced by either severe dyspnea or weakness caused by leukemic dissemination in the thymus or vital organs (bone marrow, lung, and liver), respectively. For mouse leukemia transplantation assays, 0.5–2 × 10⁶ leukemic cells collected from diseased female TEL-JAK2/TCR-HY, TEL-JAK2, TEL-JAK2/CD3ϵ^−/−, or IK^L/L mice were intravenously injected in the tail vein of recipient 8-to-12-week-old C57BL/6 or NSG mice (for the IK^L/L tumor) of the indicated gender and genotype and regularly monitored through peripheral blood detection of leukemic cells (CD45.1⁺CD45.2⁺TCR-Vβ6⁺ TEL-JAK2/TCR-HY cells or CD4⁺CD8⁺ TEL-JAK2 cells). Anti-CD3ϵ (145-2C11) mAb or control ChromPure Syrian Hamster IgG (Jackson Immunoresearch) was diluted in sterile PBS and intravenously administered on a regimen of 20 μg/mouse/day for 5 consecutive days for the TEL-JAK2 tumors and on a regimen of 50 μg/mouse/day for 2 periods of 5 consecutive days separated by a 2-day pause for the IK^L/L leukemia. For human leukemia xenotransplantation assays, 1 × 10⁶ fresh leukemic cells obtained from primary NSG mice engrafted with patient-derived cells from T-ALL cases M106, UPNT419, UPNT420, M149, UPNT610, UPNT374, and UPNT525 were intravenously injected in 2-month-old NSG mice. These mice were injected intravenously with either anti-CD3ϵ OKT3 mAb (BioXCell; 40 μg/mouse/day treatment for 2 rounds of 5 consecutive days separated by a 2-day interval) or the isotype-matched (IgG2a) C1.18.4 mAb (BioXCell), both diluted in PBS. In the curative setting, mice were treated as above when leukemic cells reached 1% to 4% of blood nucleated cells. Mice were monitored weekly by flow cytometry for leukemic load (FSC^hi, hCD7⁺, hCD45⁺ cells) in peripheral blood. Statistical analyses and survival curves were calculated using Prism 5 (GraphPad). Kaplan–Meier survival curves were compared using the log-rank test. Histologic analyses were performed on paraffin-embedded sections as detailed in the Supplementary Methods.

Immunophenotypic and oncogenic features of patients with T-ALL were identified as described (6, 27, 49, 50) and detailed in Supplementary Table 1. Fresh or thawed primary T-ALL samples (peripheral blood or bone marrow) obtained at diagnosis from adult and pediatric patients were used, as well as T-ALL cells xenografted in NSG mice. T-ALLs were all included within the GRAALL-2005 study, and informed consent was obtained from all patients at trial entry (trial registration ID: NCT00327678). Studies were conducted in accordance with the Declaration of Helsinki and approved by local and multicenter research ethical committees. All samples used contained ≥80% blasts. CD4⁺CD8⁺ (DP) human thymocytes were obtained and processed as described in the Supplementary Methods. The ALL-SIL cell line (DSMZ; ACC511) was grown in vitro and transduced as described in the Supplementary Methods. The ALL-SIL parental line and all derivatives were authenticated by short tandem repeat DNA profiling on June 10, 2016.

Splenocytes were obtained from C57BL/6 female mice, subsequently irradiated (2000 rad) and preincubated with antigenic peptides at 37°C for 1 hour (peptide pulsing). Peptides used were DBY (NAGFNSNRANSSRSS), the male antigen epitope specifically recognized by Marilyn transgenic TCR-HY, at 10, 1, 0.1, 0.01, 0.001, and 0.0001 μmol/L, and OVA (ISQAVHAAHAEINEAGR), a nonspecific peptide, at 10 μmol/L. Splenocytes were then plated out in 48-well plates, without removal of peptides. ALL-SIL/TCR-HY cells were added to the coculture system at a 1:100 ratio. Control conditions were performed in parallel with non-pulsed splenocytes, DBY without splenocytes, and ALL-SIL without transgenic TCR cocultured on DBY-pulsed splenocytes. Apoptosis was analyzed by flow cytometry at day 3 of culture, as described in the Supplementary Methods.

Cell suspensions prepared from mouse lymphoid organs, bone marrow, or blood were stained with fluorochrome-labeled antibodies and analyzed on a FACSCalibur cytometer (BD Biosciences), as described (43). The list of antibodies used is described in the Supplementary Methods. Data were analyzed using Diva, CellQuest (BD Biosciences) and FlowJo (TreeStar) softwares.

Short-term stimulation was performed on cells cultured in serum-free medium for at least 15 minutes at 37°C prior to stimulation. Murine anti-hCD3 mAb (OKT3, Biolegend; 20 μg/mL) was added to the cell culture on ice for 10 minutes and mAbs were then cross-linked by the addition of goat–anti-mouse antibody to a final concentration of 50 μg/mL for 15 minutes. Stimulation (0–10 minutes) was triggered by warming cells at 37°C and terminated either by cooling cells in cold PBS or fixing cells for phosphokinase array and intracellular phosphoprotein (Phosflow) assays, respectively. For prolonged stimulation, a bead-based assay was used in which cell lines were cultured in 96- or 24-well plates and exposed to 4.5-μm diameter superparamagnetic beads covalently coupled to anti-CD3 and anti-CD28 antibodies (Dynabeads Human T-Activator CD3/CD28, Invitrogen; duration of stimulation from 1 to 12 days). Prolonged stimulation of primary T-ALL cells was performed in a coculture assay with confluent OP9-DL1 in an α-MEM media supplemented with 20% FBS (Hyclone; ThermoFisherScientific), 50 μg/mL streptomycin, and 50 IU penicillin and recombinant human cytokines hFLT3L (5 ng/mL), hIL7 (2 ng/mL), and hSCF (10 ng/mL; Miltenyi). Similarly, CD3/CD28-coated beads were added to the coculture assay following the manufacturer's instructions.

Stimulation of cells was terminated by fixation with prewarmed formaldehyde (Cytofix Buffer; BD Biosciences) at 37°C for 10 minutes. After cooling on ice for 1 minute, cells were pelleted at 4°C and permeabilized in 1 mL of ice-cold methanol (Perm Buffer III; BD Biosciences). After 30 minutes, cells were washed 3 times with 3 mL of PBS and resuspended in 50 μL PBS for intracellular staining with phospho-ERK1/2-Alexa647 (pT202/pY204), phospho-STAT3-Alexa488 (pY705; BD Biosciences) or phospho-AKT-Alexa647 (pS473; Cell Signaling Technology) and incubated for 20 minutes at room temperature. After washing, cells were resuspended in PBS for flow cytometric analysis on FACSCanto II (BD). Data were processed using Cytobank software.

Sample processing and analysis are described in Supplementary Methods.

ALL-SIL/TCR-HY cells were stained with TCR Vβ6-PE mAb (BD Biosciences) for 30 minutes at 4°C and DAPI for 5 minutes at room temperature prior to stimulation and split into two samples; one sample remained unstimulated whereas the other one was stimulated by styrene beads (Polyscience) precoated with human anti-CD3ϵ antibody (OKT3; Biolegend) for 10 minutes at 37°C. Cells were fixed with 1% formaldehyde. Staining was analyzed using ImageStream X mkII (Amnis Merck-Millipore). Data analysis was performed using the IDEAS image analysis software (Amnis).

ALL-SIL, ALL-SIL/TCR-HY cells, human DP thymocytes, and CD3-negative (UPNT525) and CD3-positive (M149) primary T-ALLs either were left unstimulated or were stimulated with CD3/CD28 mAbs for 5 minutes. Each stimulated condition was performed in replicate. Cell lysates and Proteome Profiler Human phosphokinase array (R&D Systems) were performed according to the manufacturer's protocol. Chemiluminescence was detected by ChemiDoc XRS+ (Bio-Rad). The average signal (pixel density) of duplicate spots representing each phosphorylated kinase protein was normalized with the average signal of reference positive duplicate spots of each membrane using Image Lab software (Bio-Rad). Signal ratios for selected phosphoproteins were displayed in a heat map using Treeview software. Cyclosporine A (Novartis) and PD184352 (Selleckchem) were used at 1 μmol/L. Effects were determined by analysis of apoptosis (Annexin V/PI assay) and expression of CD25 and CD69 activation markers.

RNA extraction was performed for the following cells: ALL-SIL/TCRαβ-GFP cocultured on OP9-DL1 and ALL-SIL/TCRαβ-GFP without coculture. These samples were obtained as previously described (14) in duplicate, and RNA extraction was performed at an early time point before apoptosis onset (48 hours of coculture for ALL-SIL/TCRαβ-GFP). RNA hybridization was performed on Affymetrix U133 plus 2.0 microarrays. The statistical data analysis was performed with R version 2.9.0 using the “Affy” package from Bioconductor. The probe intensities were log₂ transformed and normalized using Robust Multiarray Average. Identification of differentially expressed genes was performed by Significance Analysis of Microarrays, using 500 permutations and a false discovery–rate threshold of 5%. Gene Set Enrichment Analysis (GSEA) was performed using the negative selection signature described by Baldwin and colleagues (14) as gene set. GSEA was run using signal-to-noise for the ranking gene metric and 1,000 permutations. All microarray data have been submitted to the Gene Expression Omnibus database under accession number GSE65496.

A total of 1 × 10⁶ control pLKO or pLKO-shLAT GFP-Luciferase sorted ALL-SIL/TCR-HY cells were intravenously injected into 2-month-old NSG mice (n = 10). Isofluorane-anesthesized mice were intraperitoneally injected with 2.5 mg of D-luciferin (CALIPER Life Sciences) and monitored on a weekly basis to detect luciferase activity using an IVIS Spectrum (Perkin-Elmer). When leukemic cells were detected (22 days after injection), mice were treated with either OKT3 or control mAb as described above.

No potential conflicts of interest were disclosed.

Conception and design: A. Trinquand, N.R. dos Santos, C. Tran Quang, N. Ifrah, E. Macintyre, J. Ghysdael, V. Asnafi

Development of methodology: A. Trinquand, C. Tran Quang, F. Rocchetti, B. Zaniboni, F.-L. Cosset, E. Verhoeyen, F. Pflumio, N. Ifrah, D.-A. Gross, J. Ghysdael, V. Asnafi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Trinquand, N.R. dos Santos, C. Tran Quang, C. Da Costa de Jesus, M. Tesio, M. Dussiot, F. Pflumio, H. Dombret, J. Ghysdael, V. Asnafi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Trinquand, N.R. dos Santos, C. Tran Quang, M. Belhocine, C. Da Costa de Jesus, L. Lhermitte, M. Dussiot, N. Ifrah, S. Spicuglia, O. Hermine, J. Ghysdael, V. Asnafi

Writing, review, and/or revision of the manuscript: A. Trinquand, N.R. dos Santos, C. Tran Quang, M. Belhocine, E. Verhoeyen, N. Ifrah, H. Dombret, L. Chatenoud, O. Hermine, E. Macintyre, J. Ghysdael, V. Asnafi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Trinquand, E. Verhoeyen, N. Ifrah, L. Chatenoud, E. Macintyre, J. Ghysdael

Study supervision: A. Trinquand, C. Tran Quang, N. Ifrah, E. Macintyre, J. Ghysdael, V. Asnafi

We thank E. Belloir and C. Alberti for assistance with mouse husbandry, Anne Reynaud for help with blood analyses, and Charlene Lasgi from the flow-cytometry platform of the Institut Curie.

A. Trinquand was supported by a grant from INCa (Institut National du Cancer: “Soutien à la Recherche Translationnelle 2012”). N.R. dos Santos was supported by an FCT Investigator contract (Fundação para a Ciência e Tecnologia) and institutional support (UID/BIM/04773/2013 CBMR). This work was supported in the Necker team by grants from the Institut National du Cancer (INCa) PLBIO13-218 and PLBIO15-094, “La Ligue Contre le Cancer” and the SFCE (Société Française des Cancers de l'Enfant), la Fédération Enfants et Santé, les associations l'Etoile de Martin, Hubert Gouin “Enfance et Cancer,” Les Bagouz's à Manon, Capucine, AREMIG et Thibault Briet, and la Fondation EDF; and in the Institut Curie team by funds from Institut Curie, CNRS, INSERM, INCa (PLBIO15-094), and Ligue Contre le Cancer (Comité de l'Essonne).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.