T-Cell Receptor Diversity Prevents T-Cell Lymphoma Development

ORIGINAL ARTICLE T-cell receptor diversity prevents T-cell lymphoma development

S Newrzela1, N Al-Ghaili2,6, T Heinrich1,6, M Petkova1, S Hartmann1, B Rengstl1, A Kumar2, H-M Ja¨ck3, S Gerdes4, I Roeder4, M-L Hansmann1 and D von Laer5

Mature T-cell lymphomas (MTCLs) have an extremely poor prognosis and are much less frequent than immature T-cell leukemias. This suggests that malignant outgrowth of mature T lymphocytes is well controlled. Indeed, in a previous study we found that mature T cells are resistant to transformation with known T-cell oncogenes. Here, however, we observed that T-cell receptor (TCR) mono-/oligoclonal mature T cells from TCR transgenic (tg) mice (OT-I, P14) expressing the oncogenes NPM/ALK or DTrkA readily developed MTCLs in T-cell-deficient recipients. Analysis of cell surface markers largely ruled out that TCR tg lymphomas were derived from T-cell precursors. Furthermore, cotransplanted non-modified TCR polyclonal T cells suppressed malignant outgrowth of oncogene expressing TCR tg T lymphocytes. A dominant role of an anti-leukemic immune response or Tregs in the control of MTCLs seems unlikely as naı¨ve T cells derived from oncogene expressing stem cells, which should be tolerant to leukemic antigens, as well as purified CD4 and CD8 were resistant to transformation. However, our results are in line with a model in which homeostatic mechanisms that stabilize the diversity of the normal T-cell repertoire, for example, clonal competition, also control the outgrowth of potentially malignant T-cell clones. This study introduces a new innate mechanism of lymphoma control.

Leukemia (2012) 26, 2499–2507; doi:10.1038/leu.2012.142 Keywords: clonal competition in T-cell homeostasis; retroviral gene therapy; mature T-cell lymphoma; OT-I and P14 TCR transgenic mice

INTRODUCTION associated with T-cell lymphoma in man (TCL1, LMO2, DTrkA)do Mature T-cell lymphomas (MTCLs) are infrequent, but their not cause transformation, when selectively expressed in mature 7 incidence is rising and distinct nodal, leukemic, cutaneous and murine TCR polyclonal T-cell populations. extra-nodal forms have just recently been defined as coherent clinico-pathological and in part molecular entities. The prognosis is generally dismal and the development of novel therapeutic MATERIALS AND METHODS options is difficult as little is known about the pathogenesis and Ethics statement the mechanisms that control MTCL development.1 The experiments were performed in compliance with the local animal In MTCLs, just like in any other tumor (B-cell lymphoma, even experimentation guidelines. Animal experiments were approved by solid tumors), clonal homeostasis is obviously disturbed. Although, the regional council (Regierungspra¨sidium) Darmstadt, Hessen, Germany. mechanisms that stabilize the T-cell repertoire have been described, it is completely unclear why homeostatic control fails Mice and malignant clones can expand and gain prevalence in MTCLs. Six- to 8-week-old C57BL/6J.Ly5.1 (CD45.1 þ ), TCR tg OT-I (CD45.2 þ ) and The currently widely accepted clonal competition model for Rag-1-deficient (CD45.2 þ ) mice were obtained from Charles River normal T-cell homeostasis postulates that T-cell clones compete laboratories (Sulzfeld, Germany) and Jackson laboratory (Bar Harbor, ME, for interaction with antigen presenting cells (APCs). This interac- USA). P14 TCR tg mice were kindly provided by the laboratory of Professor tion is assumed to be essential for T-cell survival and proliferation.2 Dr Ulrich Kalinke (Institute for Experimental Infection Research, TWINCORE, According to this model, clonal abundances are regulated on a Centre for Experimental and Clinical Infection Research, Hannover, per-clone basis, and the affinity of a particular T-cell receptor (TCR) Germany). The TCR tg mice were leaky with 5–10% T cells that did not express the transgenic TCR in OT-1 and P14 mice. Animals were bred and to major histocompatibility complex/self-peptide (MHC/SP) maintained under specific pathogen-free (SPF)-like conditions in the complexes presented on APCs crucially affects the steady-state animal facilities of the Georg-Speyer-Haus. Cages were individually 3–5 abundance of the corresponding T-cell clone. The hetero- ventilated. Symptomatic/leukemic or healthy animals (donors) were killed geneity of the T-cell repertoire is thus preserved through the after anesthesia by cervical dislocation and examined for pathological heterogeneity of the MHC/SP complexes (T-cell niches) presented abnormalities, including histology, morphology, white blood cell (WBC) in the host.2,6 Here, we experimentally tested the hypothesis that counts and flow cytometry. clonal competition not only stabilizes the normal T-cell repertoire, but also suppresses malignant clonal outgrowth. This hypothesis Retroviral vectors/cloning was derived from our previous finding that (proto-)oncogenes SF91-DTrkA and the control vector MP91-eGFP were previously known to cause T-cell malignancies in transgenic mice and/or described.7,8 The human fusion oncogene NPM/ALK9 was purchased as

1Senckenberg Institute of Pathology, Goethe-University Hospital, Frankfurt am Main, Germany; 2Georg-Speyer-Haus, Institute for Biomedical Research, Frankfurt am Main, Germany; 3Department of Internal Medicine III, Nikolaus-Fiebiger-Center, University of Erlangen-Nu¨rnberg, Erlangen, Germany; 4Institute for Medical Informatics and Biometry, Medical Faculty Carl Gustav Carus, Dresden University of Technology, Dresden, Germany and 5Institute of Virology, Innsbruck Medical University, Innsbruck, Austria. Correspondence: Professor D von Laer, Institute of Virology, Innsbruck Medical University, Fritz-Pregl-Strasse 3, Innsbruck 6020, Austria. E-mail: [email protected] 6These authors contributed equally to this work. Received 3 February 2012; revised 10 May 2012; accepted 11 May 2012; accepted article preview online 30 May 2012; advance online publication, 22 June 2012 Homeostatic control of T-cell lymphoma S Newrzela et al 2500 synthetized cDNA from the company GENEART (Regensburg, Germany) conjugated CD45.1 (A20), (PerCP-Cy5.5)-conjugated CD45.2 (104). For and cloned into the gammaretroviral vector MP91-eGFP 5-prime of an detection of the pre-TCR alpha chain, we used unconjugated anti-pre-TCR internal ribosome entry site from the encephalomyocarditis virus and the alpha chain (2F5), a secondary biotinylated anti-mouse IgG1 and gene for the enhanced green fluorescence protein (EGFP). Streptavidin-APC (BD), as a third-step reagent. To prevent non-specific binding to Fc receptors, samples were incubated with mouse seroblock FcR (Serotec, Du¨sseldorf, Germany) or CD16/CD32 MoAbs (2,4G2) (BD). Retroviral vector production After staining, blood, bone marrow and spleen samples were treated with Vector supernatants were produced in DMEM (Bio Whittaker, Rockville, MD, Cal-lysis solution (Invitrogen) to remove red blood cells. USA) supplemented with 10% fetal calf serum (PAN BIOTECH GmbH, Analyses were performed on a FACScalibur or FACS Canto II using the Aidenbach, Germany), 2% L-glutamine (Bio Whittaker), 1% of Penicillin/ CellquestPro or FACSDiva software (BD). All cell counts were performed Streptomycin (Pen/Strep) (PAN BIOTECH GmbH). Ecotropic supernatant on a CASY Cell Counter (Scha¨rfe System, Reutlingen, Germany). was produced in a split genome approach by calcium–phosphate- mediated transient transfection of HEK 293T human embryonic kidney producer cells. After 24, 48 and 60 h supernatant was collected, filtered Intracellular staining and phospho protein analysis (45 mm) and stored at 4 1C. All supernatants were pooled and titrated on Terminal deoxynucleotidyl transferase (TdT) expression and phosphoryla- the embryonic murine fibroblast SC-1 cell line. tion status of signal transducer and activator of transcription 3 (Stat3-p) were determined via intracellular staining by using BD Phosflow reagents, 6 À following the manufacturer’s instructions (BD). In all, 1–2 Â 10 cells were Retroviral transduction and transplantation of Lin HSCs/HPCs lysed and fixed in a single step using BD Phosflow Lyse/Fix Buffer for or mature T cells 10 min at 37 1C. Cells were permeabilized in BD Phosflow Perm Buffer III for Bone marrow cells were flushed from tibias and femurs of C57BL/6J.Ly5.1 30 min on ice. Afterwards cells were washed twice in BD Pharmingen Stain donor mice. Enrichment of Lin À cells was carried out with the Mouse Buffer and stained with either PE Mouse anti-phospho-Stat3 (4/P-STAT3) Lineage Cell Depletion Kit (Miltenyi Biotec, Bergisch-Gladbach, Germany) (BD), anti-Mouse TdT PE (19-3) (eBioscience) or with isotype control following the manufacturer’s protocol. Lin À Cells were cultivated in Stem antibodies: PE Mouse IgG2a, k (G155-178) (BD) and PE Mouse IgG2b, Span SFEM medium (STEMCELL Technologies, Vancouver, BC, Canada), k (27-35) (BD). supplemented with 10 ng/ml murine SCF, 20 ng/ml murine TPO, 20 ng/ml murine IGF-2 and 10 ng/ml human FGF-1 (Tebu-bio, Offenbach, Germany). Twenty-four-well non-tissue culture plates (BD, San Diego, CA, USA) were Western blotting coated with 50 mg/ml retronectin (Takara Bio Inc., Tokyo, Japan), according For western blot expression studies of Stat3 and phosphorylated Stat3 we to the manufacturer’s protocol and preloaded with retroviral supernatants prepared cell lysates from 4-day anti-CD3/CD28 stimulated T cells (from (loading of retronectin coated plates: 1 ml of vector supernatant per well; 8-week-old OT-I donors), NPM/ALK- and DTrkA-induced tumors (each 1000 g,41C, 30 min). Lin À cells were cultured on particle-covered plates 5 Â 106 cells). As primary antibodies we used rabbit anti-phospho-Stat3 and transduction was performed only once to ensure single-cell copies in (Tyr705) in a 1:1000 dilution (Cell Signaling, Beverly, MA, USA), rabbit anti- the transduced cells. Five days after isolation and transduction, 5 Â 105 Stat3 (C-20) in a 1:1000 dilution (Santa Cruz Biotechnology, Santa Cruz, CA, bone marrow cells were injected intravenously into Rag-1-deficient USA) and as a loading control we used monoclonal mouse anti-b-Actin recipients. Before transplantation (4–6 h) mice were sublethally irradiated antibody in a 1:40 000 dilution (Sigma, Munich, Germany). Western with 5 Gy using a BIOBEAM 2000 Cs-137 chloride gamma irradiator (Eckert blotting was performed according to the manufacturer’s instructions; the & Ziegler, Berlin, Germany). primary antibodies were detected with horseradish peroxidate-conjugated Expansion and transduction of stimulated T cells was as previously secondary antibodies: goat anti-rabbit HRP in a 1:2000 (Dako, described.7 Five days after isolation and after two rounds of transduction, Glostrup, Denmark), rabbit anti-mouse HRP in a 1:2000 dilution (Dako). 5 Â 106 transduced mature T cells were injected intravenously into Rag-1- deficient recipients. For competition experiments 5 Â 106 transduced TCR tg OT-I or P14 T cells were mixed with 5 Â 106 stimulated, untransduced Histopathological analysis wild-type (WT) C57BL/6 (Ly5.1) T cells before transplantation. Obviously sick mice, presenting scrubby fur, a crooked body appearance and a complete loss of activity and/or with a high WBC were killed for necropsy and examined for pathological abnormalities. After securing Serial transplantation of different DTrkA-expressing cell single-cell suspensions of infiltrated organs for FACS analysis, lymph nodes, populations spleen, thymus, liver, kidney, lung, heart, brain and bone marrow were Six weeks after the transplantation HSC/HPC-DTrkA recipients were killed. fixed in 10% Zinc-Formal-Fixx (Thermo Shandon, Pittsburgh, PA, USA). In all, 1 Â 106 total bone marrow, 1 Â 106 total thymocytes and 5 Â 106 CD3 Sections were prepared and stained with hematoxilin and eosin. For negatively selected, naı¨ve mature T cells from the lymph nodes (purity histological examination a light microscope (Eclipse E1000m Nikon, Tokyo, 490% CD3 þ ) were serially transplanted from primary recipients Japan) with Â 4, Â 20, Â 40 and Â 200 lenses with a digital camera (DXM into secondary Rag-1 À / À animals. Negative selection of T cells was 1200, Nikon) was used. performed using the Dynal Mouse T-Cell Negative Isolation Kit (Invitrogen, Carlsbad, CA, USA). Retroviral insertion site analysis and statistical analyses Methods for retroviral integration site (RIS) analysis and -associated Flow cytometry analysis and WBC counts of leukemia/lymphoma statistics can be found in the Supplement. samples Immediately before killing and necropsy, blood was obtained from symptomatic animals and a hemogram was prepared to determine the RESULTS WBC count using a Scil vet animal blood counter (Scil animal care company GmbH, Viernheim, Germany). Resistance of naı¨ve polyclonal as well as purified CD8 and CD4 Flow cytometry analyses were carried out on blood and single-cell T cells to transformation by overexpression of DTrkA suspensions of thymus, spleen, lymph nodes and bone marrow. The In a previous study, we found that upon transduction with following anti-mouse monoclonal antibodies (MoAbs) were used for gammaretroviral vectors expressing a T-cell oncogene mature staining: rat anti-mouse MoAbs (Invitrogen); phycoerythrin (PE)-conjugated T cells, in contrast to hematopoietic stem/progenitor cells CD8a (CT-CD8a), PE Cy5.5 (PE-Cy5.5)-conjugated CD4 (RM4-5), PE Cy5.5 (HSCs/HPCs), do not develop leukemia/lymphoma.7 However, in (PE-Cy5.5)-conjugated CD19 (6D5), PE Cy5 (PE-Cy5)-conjugated CD3 this previous study we investigated the transformability of (145-2C11) (BD); (PE)-conjugated CD2 (RM2-5), (PE-Cy5)-conjugated CD5 memory but not of naı¨ve T cells. The reason was that the T cells (53-7.3), (PE)-conjugated CD25 (3C7), (PE)-conjugated CD90.2 (30-H12), (PE)-conjugated Sca-1 (E13-161.7), Allophycocyanin (APC)-conjugated had to be activated to become susceptible to gammaretroviral CD117 (2B8) (eBioscience, San Diego, CA, USA); (PE-Cy5)-conjugated vectors and thereby acquired a memory phenotype. CD278/ICOS (15F9), hamster anti-mouse MoAbs (eBioscience); (PE)- To study whether naı¨ve T cells are also resistant to transforma- conjugated CD30 (mCD30.1) and (PE)-conjugated hamster IgG isotype tion, HSCs/HPCs from three C57BL/6 Ly5.1 mice were transduced control (530-6) (Invitrogen); mouse anti-mouse MoAbs (BD); (PE)- with the gammaretroviral vector SF91-DTrkA expressing an

Leukemia (2012) 2499 – 2507 & 2012 Macmillan Publishers Limited Homeostatic control of T-cell lymphoma S Newrzela et al 2501 oncogenic, constitutively active form of TrkA8 or with MP91-eGFP of this experiment shows that regulatory T cells are not required for as a control vector (Supplementary Figure S1) and then the suppression of malignant outgrowth, as oncogene transduced transplanted into five sublethally irradiated Rag-1 À / À Ly5.2 mice polyclonal CD8 T cells alone did not develop any malignancy. Finally, (Figure 1a). In our previous study,7 DTrkA was the most potent as observed in our previous study,7 the transgenes were expressed oncogene, with which we had shown resistance of memory T cells stably throughout the observation period in the polyclonal T-cell to development of MTCLs. The Ly5 marker allowed differentiation population without the development of MTCLs, which is not of donor (Ly5.1) and recipient cells (Ly5.2). Six weeks after expected if an immune response specifically eliminated potentially transplantation, before DTrkA-associated lymphoma developed, malignant T-cell clones. bone marrow, thymocytes and CD3 þ T cells, purified by negative selection (purity 490% CD3 þ ) from the lymph nodes, were TCR diversity suppresses development of MTCLs harvested from each mouse. At this time point over 90% T cells To address the clonal competition hypothesis, we used T cells have a naı¨ve phenotype (CD44low, data not shown). The average from TCR transgenic (tg) OT-I10 and P14(ref. 11) donor mice. No DTrkA expression in these populations was 12%, 32% and 25%, interclonal competition is expected in these TCR tg T-cell respectively (Supplementary Figure S2a). Each population populations and each TCR tg T cell is subjected to the same was serially transplanted into Rag-1 À / À recipients (Figure 1a). niche regulation. The clonal competition hypothesis predicts that All recipients were repopulated with DTrkA-expressing cells at a control of clonal outgrowth due to niche dependency will fail in level between 14 and 25% in the periphery (data not shown). HSC/ this situation, and premalignant cells can indeed expand and gain HPC and thymocyte transplanted mice of the DTrkA-group prevalence. succumbed to lymphoma within 1–5 months (Figure 1b). The For these studies, we switched to the T-cell-specific oncogene malignancies were all EGFP positive and displayed an immature ALK expressed as the NPM/ALK fusion protein. ALK was chosen T-cell phenotype, as previously described.7 However, the mice because preliminary experiments had shown that ALK-induced transplanted with DTrkA-expressing naı¨ve T cells remained malignancies develop more rapidly than DTrkA-induced lympho- healthy with normal leukocyte levels and no macroscopic signs mas. Furthermore, ALK is well proven to be associated with of lymphoma throughout a follow-up of 9 months (Figure1b) with anaplastic large cell lymphoma (ALCL) in humans.9,12,13 stable expression levels of DTrkA (Supplementary Figure S2b). Mature T cells were collected from the spleen and the lymph Thus, in our experimental setup naı¨ve peripheral T cells were also nodes of healthy C57BL/6 Ly 5.1 (WT) mice, as well as from TCR tg resistant to transformation by overexpression of DTrkA. OT-I mice. Cells were transduced ex vivo with a gammaretroviral In addition, the results argue against the hypothesis that a vector coexpressing NPM/ALK and EGFP (MP91-NPM/ALK) or specific anti-leukemic immune response significantly contributes with a control vector expressing EGFP only (MP91-eGFP, to the control of T-cell lymphoma in this setting with the Supplementary Figure S1). The transduction efficacies were investigated oncogene. Naı¨ve T cells derived from transplanted 18% and 8% for NPM/ALK expressing WT T cells and OT-I TCR tg DTrkA expressing HSCs/HPCs should be tolerant to DTrkA or T cells, respectively (Supplementary Figure S3a). Gene-modified potential T-cell lymphoma-associated antigens and therefore cells were transplanted into T-cell-deficient Rag-1 À / À recipients should be unable to exert a specific anti-leukemic immune (5 Â 106 per mouse, Figure 2a). response. Furthermore, the fact that serially transplanted naı¨ve As predicted by the clonal control hypothesis, all mice T cells as well as the transplanted memory T cells in our previous transplanted with OT-I TCR tg T cells (n ¼ 10) that expressed study7 stably expressed the oncogene over months also argues NPM/ALK succumbed to lymphoma within 30–80 days post against a lymphoma-specific immune response in these settings. transplantation (Figure 2b). Only two of the mice transplanted To further test the anti-leukemic immune response hypothesis, with NPM/ALK expressing WT T cells (n ¼ 6) developed lymphoma, we transplanted purified CD4 or CD8 cells, either expressing EGFP however, after a considerably longer latency period (161, 229 or DTrkA (n ¼ 5 per group). During the follow-up of 300 days, none days). All animals had enlarged lymph nodes and splenomegaly. of the animals developed malignancies, although stable transgene The histology resembled human mature peripheral nodal T-cell expression levels were maintained (data not shown). The lymphoma (Figure 4). Mice transplanted with cells expressing the development of a potent de novo anti-leukemic immune response control vector (n ¼ 10) remained healthy throughout the whole within the CD4 or CD8 population alone appears unlikely, thus observation period (data not shown). again arguing against a significant role of an anti-leukemic The clonal competition hypothesis also predicts that polyclonal immune response in lymphoma control. In addition, the outcome T cells will suppress the development of lymphoma from

Figure 1. Transformation resistance of TCR polyclonal, naı¨ve mature T cells. (a) Lin À HSCs/HPCs were isolated from five WT C57BL/6 (Ly5.1) mice, stimulated for 3 days and transduced with SF91-DTrkA or MP91-eGFP as a control. Gene-modified cells were transplanted into five sublethally irradiated (5 Gy) Rag-1-deficient recipients. Six weeks after transplantation mice were killed. HSCs/HPCs from total bone marrow, thymocytes and CD3 negatively selected naı¨ve mature T cells from the lymph nodes were serially transplanted into secondary Rag-1 À / À recipients. Secondary recipients transplanted with bone marrow cells were sublethaly irradiated (5 Gy) to enable engraftment of transferred cells. Gene marking of serially transplanted populations can be found in Supplementary Figure S2a. (b) Survival of Rag-1-deficient recipients transplanted with one of the three different cell populations. HSC/HPC and thymocyte transplanted animals succumbed to immature T-cell leukemias within 1–5 months. Mice transplanted with DTrkA-expressing naı¨ve T cells remained healthy for an observation period of 9 months.

& 2012 Macmillan Publishers Limited Leukemia (2012) 2499 – 2507 Homeostatic control of T-cell lymphoma S Newrzela et al 2502

Figure 2. Transgenic (tg) OT-I T cells expressing NPM/ALK develop lymphoma whereas TCR diversity suppresses development of malignancy. (a) TCR polyclonal WT C57BL/6 (Ly5.1) or TCR tg OT-I T cells were stimulated, transduced with NPM/ALK and transplanted into Rag-1-deficient recipients. Transduction efficacies can be found in Supplementary Figure S3a. Furthermore, a group of animals (n ¼ 6) was transplanted with a mixture of 5 Â 106 NPM/ALK-transduced tg OT-I T cells mixed with 5 Â 106 untransduced WT C57BL/6 (Ly5.1) competitor cells. (b) Survival of animals transplanted with NPM/ALK-transduced T cells. All recipients of NPM/ALK TCR tg OT-I T cells developed mature T-cell lymphoma after short latencies (30–80 days). Only two of the six animals transplanted with NPM/ALK transduced TCR polyclonal WT T cells developed mature T-cell lymphoma with a delayed onset. Furthermore, all animals transplanted with mixtures of oncogene-transduced, TCR oligoclonal and non- modified, TCR polyclonal WT competitor T cells (n ¼ 6) survived without any indication of lymphoma. P-value determined by a Mantel–Cox rank sum test.

Figure 3. The transformability of TCR oligoclonal T-cell populations is not restricted to NPM/ALK and OT-I. (a) TCR oligoclonal tg OT-I or tg P14 T cells were transduced with the oncogenes DTrkA, NPM/ALK or with EGFP and transplanted into Rag-1-deficient recipients. Transduction efficacies can be found in Supplementary Figure S3b. Furthermore, a group of animals (n ¼ 6) was transplanted with a mixture of 5 Â 106 DTrkA-transduced TCR tg OT-I T cells and 5 Â 106 untransduced WT C57BL/6 (Ly5.1) competitor cells. (b, c) All recipients of DTrkA- or NPM/ALK- transduced TCR oligoclonal tg OT-I or tg P14 T cells developed mature T-cell lymphoma after characteristic latencies, whereas control vector transduced animals remained healthy during the experiment. Only one of the animals transplanted with mixtures of oncogene-transduced, TCR tg OT-I and non-modified, TCR polyclonal WT competitor T cells (1/6) developed lymphoma after more than six months. P-value determined by a Mantel–Cox rank sum test.

oncogene expressing TCR tg T cells. To test this prediction, the hypothesis that polyclonal T cells suppress the development T cells from OT-I TCR tg mice were puriﬁed and transduced with of MTCLs. In addition, this result demonstrates that T cells are not MP91-NPM/ALK. Cells (5 Â 106 per mouse) were transplanted into per se resistant to transformation. Furthermore, it shows that Rag-1 À / À mice, alone or together with 5 Â 106 WT TCR polyclonal even if a T cell expressing a tg TCR was more susceptible competitor T cells. As expected, all animals transplanted with to transformation, it is still accessible to the control within a NPM/ALK-transduced OT-I cells developed MTCLs. In contrast, polyclonal T-cell population. Taken together, the ﬁndings are in none of the six NPM/ALK-expressing animals cotransplanted with line with the hypothesis that oncogene expressing TCR tg T-cell polyclonal competitor T cells developed lymphoma within the populations develop lymphoma because they lack external 300-day follow-up (Figure 2). This observation further corroborates homeostatic control exerted by polyclonal T cells.

Leukemia (2012) 2499 – 2507 & 2012 Macmillan Publishers Limited Homeostatic control of T-cell lymphoma S Newrzela et al 2503 Transformability of TCR oligoclonal T-cell populations—a more oligoclonal T-cell populations to develop T-cell lymphomas is general phenomenon neither specific for NPM/ALK nor for OT-I TCR tg T cells, but a more To determine whether the observed control of malignant general phenomenon. outgrowth is a more general phenomenon, we tested whether other oncogenes can transform TCR tg T cells and whether other TCR tg T cells can be transformed. OT-I as well as P14 TCR tg T cells Phenotype analyses of induced tumors were transduced with a DTrkA-expressing vector or with MP91- The phenotypes of all lymphomas were analyzed in detail (Table 1 eGFP and transplanted into Rag-1 À / À mice. Additionally, animals and Supplementary Table S1). As frequently observed for were transplanted with P14 TCR tg T cells transduced with the lymphomas in humans, marker expression was not stringent and NPM/ALK oncogene (n ¼ 6). Again, a group of Rag-1 À / À mice was often markers from different lymphocyte lineages were coex- transplanted with 5 Â 106 DTrkA-expressing OT-I T cells alone pressed, while several typical T-lineage markers were missing (n ¼ 20) or together with 5 Â 106 WT TCR polyclonal competitor (CD3, TCR). Nevertheless, NPM/ALK-induced lymphomas could be T cells (n ¼ 6, Figure 3a). Transduction efficacies of the transplants classified as T-cell lymphomas as all expressed at least one pan-T- are shown in Supplementary Figure S3b. cell marker. For the P14-NPM/ALK group, all tumors expressed at As expected, none of the control mice transplanted with OT-I or least two pan-T-cell markers and the coreceptor CD4 (Table 1). P14 transgenic T cells expressing only EGFP developed lymphoma. Interestingly, 8 of 10 OT-I-NPM/ALK and 6 out of 6 P14-NPM/ALK In contrast, all animals transplanted with DTrkA- and NPM/ALK- tumors expressed inducible costimulator (ICOS), a marker highly expressing TCR tg T cells (OT-I or P14) developed lymphoma, specific for T lymphocytes (Supplementary Table S1). Out of the 21 however, for DTrkA with a longer latency than for NPM/ALK lymphomas that developed from DTrkA-expressing lymphocytes, (Figures 3b and c). The mice showed a macroscopic and 16 expressed at least one pan-T-cell marker. Four DTrkA-induced histological phenotype typical of a peripheral nodal lymphoma lymphomas expressed CD19, a B-cell marker. All three DTrkA- (Figure 4). For the OT-I/DTrkA plus competitor group, only one of transformed P14 lymphomas analyzed expressed ICOS as a T-cell- the six animals developed lymphoma with a delay of more than specific antigen, one of the three coexpressed CD19. None of the 100 days relative to the group transplanted with OT-I TCR tg T cells CD19-expressing lymphomas analyzed were found to have a expressing DTrkA. Thus, the enhanced susceptibility of TCR recombined B-cell receptor, which excluded a mature B-cell origin

Figure 4. Histopathological analyses of induced T-cell lymphomas. Animals showed enlarged thymi (a–c) and splenomegaly (a–d), as well as enlarged intraabdominal (a, b and d) or mesenteric (c) lymph nodes. Hematoxilin and eosin (H&E)-stained tumors demonstrated lymphoblastic infiltrates in the spleen and lymph node, induced by DTrkA (e, f and i, j). Anaplastic blast infiltrates in the spleen and lymph node induced by NPM/ALK were reminiscent of anaplastic large cell lymphoma (g, h and k, l). (m, n) Lymphoblastic infiltrate induced by DTrkA in the portal tracts and liver sinusoids. Anaplastic blast infiltrate induced by NPM/ALK, mainly restricted to portal tracts of the liver (o, p). (e–p) H&E stainings, Â 200.

& 2012 Macmillan Publishers Limited Leukemia (2012) 2499 – 2507 Homeostatic control of T-cell lymphoma S Newrzela et al 2504 Table 1. Phenotypes and WBC count of induced lymphomas

Cell type/oncogene GFP þ DN SP CD19 CD30 At least At least two WBC4 one pan-T-cell pan-T-cell 15 000/ml marker markersa

OT-I-NPM/ALK (n ¼ 10) 8/10 10/10 0/10 0/10 10/10 10/10 7/10 6/10 OT-I-DTrkA (n ¼ 21) 13/21 20/21 1/21 4/21 0/21 16/21 8/21 9/17 P14-DTrkA (n ¼ 4) 1/3 3/3 0/3 1/3 0/3 1/3 0/3 0/3 P14-NPM/ALK (n ¼ 6) 6/6 0/6 6/6 0/6 6/6 6/6 6/6 2/6 Abbreviations: DN, double negative; GFP, green ﬂuorescent protein; SP, single positive for T-cell markers CD4 or CD8; WBC, white blood cell. n ¼ Number of animals for each cell type/oncogene. aPan-T-cell markers: CD2, CD3, CD5, CD90.2, ICOS.

Figure 5. NPM/ALK-induced tumors resemble human ALCL. Representative FACS and western blot analyses of NPM-ALK- and DTrkA-induced tumors for CD30 expression (a) and for the phosphorylation status of Stat3 (b, c). Compared with the isotype control (shaded histograms) and DTrkA-induced tumors (solid lines), NPM/ALK-induced tumors (dashed lines) highly expressed CD30 (a) and phosphorylated Stat3 (Stat3-p) (b, c). (c) Unphosphorylated Stat3 and b-actin levels were determined as controls for equal loading.

(data not shown). The expression of ICOS on the majority of the rearrangement analysis to determine the differentiation status of lymphomas suggests that these lymphomas may have developed the lymphomas. We therefore analyzed immature T-cell markers, from follicular T helper cells.14 that is, CD117, TdT, pre-TCR alpha chain, in the T-cell transplants We could initiate cell cultures from spleen and lymph node- and lymphomas. We were not able to detect immature T-cell derived lymphoma cells of diseased NPM/ALK animals. Further- markers in our transplants (WT or TCR tg), in which 495% of cells more, established cell lines were injected intravenously into expressed a TCR (data not shown). In addition, no cells expressing secondary recipients, which developed histologically and immu- the immature T-cell markers analyzed were detected in our tumor nophenotypically identical malignancies as presented in the samples (Supplementary Table S1). primary tumors (data not shown). However, in accordance with an ‘ALCL-like’ phenotype of NPM/ To further exclude immature T-cell transformation in our ALK-induced tumors in transgenic mice,12 CD30 was highly experimental setup, we transduced TCR tg OT-I T cells with expressed and the signal transducer and activator of the proto-oncogene LMO2 and transplanted these cells into transcription 3 (Stat-3p) was phosphorylated in all NPM/ALK Rag-1 À / À animals. As LMO2 transforms immature T cells15 and is tumors analyzed (Figure 5). characterized by a long latency (9–12 months), we observed transplanted animals for over 360 days. None of the animals developed leukemia/lymphoma, although the proto-oncogene Retroviral integration analysis of induced MTCLs was stably expressed in the recipients (data not shown). In this study, we used gammaretroviral vectors with a strong LTR enhancer. To determine the potential contribution of insertional mutagenesis to lymphoma development, we analyzed the Induced MTCLs were negative for immature T-cell markers and distribution of retroviral integration sites (RISs). Despite the fact, NPM/ALK-induced tumors resembled an ‘ALCL-like’ phenotype that there is a substantial proportion of tumors with a lost As we were using TCR tg donor animals (OT-I and P14) on a Rag- transgene expression, we detected RISs in all analyzed tumors. 1 þ / þ background, we checked the rearrangement status of the Downregulation of the transgene could possibly reflect that in intrinsic TCR-beta-chain on the genomic level in the tumors. these cases tumor sustenance became independent from Unfortunately, we were not able to detect any rearranged gene oncogene addiction after accumulation of secondary mutations segments for the intrinsic TCR chain (data not shown), most likely during tumorigenesis. A total of 131 RISs were identified in 31 because of the expression of the tg TCR during maturation in the tumors (a detailed list of all RIS surrounding genes, in a ±250 kb thymus. In all analyzed tumors we were able to detect the window can be found in Supplementary Table S2, with a transgene for the OT-I receptor on the genomic level. A further preference for gene-dense regions (Po0.001) and the ±10 kb genomic analysis of the alpha chain seems rather difficult, as to windows around transcription start sites (Po0.001 Supplementary our knowledge, there is currently no reliable protocol available. Figure S4), as previously described for gammaretroviral vec- Thus, we could not rely on the prevalent method of TCR tors.16,17 Furthermore, a twofold overrepresentation (Po0.001) of

Leukemia (2012) 2499 – 2507 & 2012 Macmillan Publishers Limited Homeostatic control of T-cell lymphoma S Newrzela et al 2505 RTCGD genes18 was observed among the genes in the ±250 kb We also observed a loss in T-cell marker expression in the window of the RISs (Supplementary Table S3). We found two RISs MTCLs induced in this study (Table 1, Supplementary Table S1). in the vicinity of the RTCGD gene Dnalc4 with a distance of only Marker loss is a common feature in NPM/ALK-induced ALCL.25,26 1536 bp in two independent lymphomas fulfilling the classical Moreover, in a previous study, we also observed a downregulation definition of a common insertion site of first order.19 The of mature T-cell markers after immortalization of a mature T-cell probability to observe a RIS pair within such a window among clone.27 To further exclude an immature T-cell contamination in 131 uniformly distributed RISs would be 0.01. Additionally, six RIS our experimental setting, we transduced OT-I T cells with the pairs had a distance of o320 kb to each other, among them two immature T-cell proto-oncogene LMO2. Just like in our previous pairs with a RTCGD gene in their close proximity (LMO2, PIM1) study,7 LMO2 overexpression in mature T cells did not result in (Supplementary Figure S5). The common insertion site near LMO2 malignancy development. This is in line with the general opinion is of special interest as insertional mutagenesis of a gammare- that the transforming capacity of an oncogene is highly troviral vector leading to activation of LMO2 was previously found dependent on the analyzed cell type. to lead to T-cell leukemia in children treated for X-linked severe Our experimental system is highly relevant for evaluating the combined immunodeficiency with a stem cell gene therapy safety of T cell-based gene therapies and clearly supports regimen.20,21 Gene ontology (GO) analysis of RIS-flanking genes the assumption that therapeutic transfer of polyclonal T cells revealed several significantly overrepresented GO classes (Mus transduced with integrating vectors is generally safe. However, musculus genome as reference set). Although the analyzed dataset T cells transduced with a recombinant TCR can expand after is small, we could observe differences in GO classes between reinfusion in the presence of the cognate antigen and thus finally malignant clones from DTrkA- and NPM/ALK-induced tumors dominate the T-cell repertoire.28,29 Subsequently, in such a (Supplementary Figures S6a–c). Taken together, these findings ‘TCR-oligoclonal situation’ our data would predict that pre- suggest that in addition to the chosen oncogenes, most likely leukemic clones are controlled less efficiently. Further studies insertional mutagenesis contributed to the transformation of TCR need to be performed to test the safety of TCR-associated tg T cells in this study. adoptive immunotherapy regimens, that is, the question of how much ‘TCR-polyclonality’ is required to suppress malignant outgrowth should be addressed. DISCUSSION Another possible explanation for suppression of MTCLs in our The by far lower incidence of MTCLs relative to B-cell lymphomas is experimental setting could be that WT T cells outcompete thought to be due to additional steps of error prone B-cell receptor transduced TCR tg T cells during engraftment. However, for gene rearrangements and affinity maturation during B-cell differ- DTrkA-transduced TCR tg T cells transplanted in combination entiation. Our study suggests that, in addition, the stronger clonal or without WT T cells, levels of gene modification in the blood of cross-control of T cells relative to B cells could explain why B-cell recipient animals was initially very similar (data not shown). homeostasis is more readily perturbed than T-cell homeostasis. Clearly, our performed experiments do not completely rule out Indeed, the presented in vivo data are reminiscent of this ‘clonal an immune response in addition to other mechanisms, but to our competition model’ of T-cell homeostasis. When applied to the knowledge a specific immune response cannot be mounted in the maximal perturbation of T-cell homeostasis, the MTCLs, this model absence of T-cell help (CD8 purified T cells were also resistant to would predict that T-cell competition for stimulatory MHC/SP transformation). Subsequently, in our mouse model, a specific niches also controls the outgrowth of potentially malignant T-cell anti-leukemic immune response seems not to have a pivotal role clones. In this study, several predictions of this hypothesis were in MTCLs control. Also a dominant role of NK cells is unlikely in our tested and confirmed: (1) in contrast to polyclonal T cells, TCR tg T setting, as these are not present in the cotransplanted purified WT cells were highly susceptible to transformation with the oncogenes T cells but were present in the recipient Rag-1 À / À mice for the DTrkA and NPM/ALK; (2) addition of polyclonal T cells suppressed monoclonal, as well as for the polyclonal transplants. the malignant outgrowth of oncogene expressing TCR tg T cells; In general, the efficacy of clonal cross-control of (pre-)malignant (3) purified TCR polyclonal CD4 and CD8 were resistant to clones is expected to be dependent on the mode of action and transformation in vivo; (4) naı¨ve polyclonal T cells were resistant the ‘strength’ of the oncogene involved. Here, individual to transformation by DTrkA, while thymocytes and HSCs/HPCs were predictions are difficult as little is known about the mechanisms readily transformed; (5) the induced MTCLs expressed several involved in clonal cross-control in normal T-cell homeostasis. markers of mature T cells (CD2, CD4, CD5, CD25, ICOS) and Further studies will be required that compare homing capacity, the phenotype of NPM/ALK-induced MTCLs closely resembled gene expression and signaling pathways ex vivo in monoclonal human ALCL. T cells expressing an oncogene in the presence and absence of The currently best-available mouse models for MTCLs are based cotransplanted WT T cells. on the CD4-Cre inducible expression of T-cell oncogenes12,22 or The clonal cross-control model is also in accordance with two inactivation of T-cell-specific tumor suppressors.23 One major recently published studies, in which MTCLs developed in TCR tg drawback of these models is that oncogene expression is initiated mice. Kelly et al.30 reported the development of thymic T-cell at the CD4 þ CD8 þ stage of T-cell development. As this is highly lymphoblastic lymphomas in TCR tg mice, in which Stat5a or unlikely to be the target cell of transformation in mature T-cell Stat5b is overexpressed within the lymphoid compartment. neoplasms, we decided to directly modify mature T cells isolated The rate of lymphoma induction was markedly enhanced from the secondary lymphoid organs spleen and lymph nodes. by immunization or by the introduction of TCR transgenes. Indeed, the phenotype of the induced tumors was similar to In a second study, Tsatsanis et al.31 crossed the TCR transgene 2C, that of human MTCLs. NPM/ALK-induced lymphomas resembled which recognizes class I MHC presented peptides, into the human ALCL much more closely than previous transplan- Tpl2 À / À genetic background. Parental mouse strain (Tlp2 À / À tation mouse models for NPM/ALK.13,24 As typical for ALCL, the and 2C transgenic mice) showed no signs of T-cell malignancy. expression of specific T-cell markers was lost (Table 1, Surprisingly, all TCR2Ctg/tg/Tpl2 À / À mice developed T-cell Supplementary Table S1), while other markers such as CD30, lymphomas with a latency of 4–6 months. The tumor cells ICOS and phosphorylated Stat3 were highly expressed in the used were consistently TCR2C þ CD8 þ CD4 À suggesting that they were transplantation model. In contrast, these markers are rarely found derived from chronically stimulated mature tg T cells. in transgenic models of NPM/ALK. The proposed model can easily The clonal competition model would explain several typical be established and the results obtained are thus expected to be features of human T-cell lymphomas. T cells are long-lived and of potential interest for modeling human MTCLs. experience massive perturbations of the repertoire with clonal

& 2012 Macmillan Publishers Limited Leukemia (2012) 2499 – 2507 Homeostatic control of T-cell lymphoma S Newrzela et al 2506 expansion and retraction during antigen encounters followed by 9 Morris SW, Kirstein MN, Valentine MB, Dittmer KG, Shapiro DN, Saltman DL et al. long-term persistence of the evolved memory cell clones. Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-Hodgkin’s Surprisingly, however, in this highly versatile system, the T-cell lymphoma. Science 1994; 263: 1281–1284. repertoire remains fairly stable up to an age of around 65 years 10 Hogquist KA, Jameson SC, Heath WR, Howard JL, Bevan MJ, Carbone FR. T cell and outgrowth of malignant T-cell clones is a rare event.32 Mature receptor antagonist peptides induce positive selection. Cell 1994; 76: 17–27. T-cell malignancies are much less frequent than their immature 11 Pircher H, Burki K, Lang R, Hengartner H, Zinkernagel RM. Tolerance induction in counterparts. For instance, although HTLV-I transforms mature T double specific T-cell receptor transgenic mice varies with antigen. Nature 1989; cells readily in vitro, only a small fraction of infected patients 342: 559–561. 12 Chiarle R, Gong JZ, Guasparri I, Pesci A, Cai J, Liu J et al. NPM-ALK transgenic mice develop leukemia after long-term persistence of the virus in 33 spontaneously develop T-cell lymphomas and plasma cell tumors. Blood 2003; mature T cells. In addition, MTCLs mostly develop in tissues 101: 1919–1927. such as skin, lymph node or mucosa, and locally perturbed T-cell 13 Kuefer MU, Look AT, Pulford K, Behm FG, Pattengale PK, Mason DY et al. 34 repertoires have been found to precede malignancy. Retrovirus-mediated gene transfer of NPM-ALK causes lymphoid malignancy in Autoimmune diseases and chronic infections involving chronic mice. Blood 1997; 90: 2901–2910. T-cell stimulations and dominance of specific T-cell clones 14 de Leval L, Rickman DS, Thielen C, Reynies A, Huang YL, Delsol G et al. The gene are often associated with MTCLs.35,36 For example, patients with expression profile of nodal peripheral T-cell lymphoma demonstrates a molecular complicated celiac disease (refractory celiac disease), bearing a link between angioimmunoblastic T-cell lymphoma (AITL) and follicular helper monoclonal T-cell infiltrate, are described to be especially prone to T (TFH) cells. Blood 2007; 109: 4952–4963. 37 15 Larson RC, Osada H, Larson TA, Lavenir I, Rabbitts TH. The oncogenic LIM protein enteropathy-associated T-cell lymphoma. Rbtn2 causes thymic developmental aberrations that precede malignancy in Taken together, we propose a model in which inter- and intra- transgenic mice. Oncogene 1995; 11: 853–862. clonal competition among T cells does not only stabilize the 16 Cattoglio C, Maruggi G, Bartholomae C, Malani N, Pellin D, Cocchiarella F et al. normal T-cell repertoire, but also suppresses malignant clonal High-definition mapping of retroviral integration sites defines the fate of outgrowth. This would be a completely new mechanism of natural allogeneic T cells after donor lymphocyte infusion. PLoS One 2010; 5: e15688. resistance to a malignancy and could explain why MTCLs are so 17 Hematti P, Hong BK, Ferguson C, Adler R, Hanawa H, Sellers S et al. Distinct rare. Further studies on the molecular effector mechanisms of genomic integration of MLV and SIV vectors in primate hematopoietic stem and clonal cross-control may fuel the development of innovate progenitor cells. PLoS Biol 2004; 2: e423. treatment strategies that improve the dismal prognosis of MTCLs. 18 Akagi K, Suzuki T, Stephens RM, Jenkins NA, Copeland NG. RTCGD: retroviral tagged cancer gene database. Nucleic Acids Res 2004; 32 (Database issue): D523–D527. 19 Suzuki T, Shen H, Akagi K, Morse HC, Malley JD, Naiman DQ et al. New genes involved in cancer identified by retroviral tagging. Nat Genet 2002; 32: 166–174. CONFLICT OF INTEREST 20 Hacein-Bey-Abina S, von Kalle C, Schmidt M, Le Deist F, Wulffraat N, The authors declare no conflict of interest. McIntyre E et al. A serious adverse event after successful gene therapy for X-linked severe combined immunodeficiency. N Engl J Med 2003; 348: 255–256. 21 Hacein-Bey-Abina S, Von Kalle C, Schmidt M, McCormack MP, Wulffraat N, Leboulch P et al. LMO2-associated clonal T cell proliferation in two patients after ACKNOWLEDGEMENTS gene therapy for SCID-X1. Science 2003; 302: 415–419. NG was supported by a scholarship of Merck Serono within the Graduiertenkolleg 22 Pechloff K, Holch J, Ferch U, Schweneker M, Brunner K, Kremer M et al. The fusion GRK1172 of the Deutsche Forschungs-gemeinschaft. Furthermore, the study was kinase ITK-SYK mimics a T cell receptor signal and drives oncogenesis in funded by the Deutsche Forschungs-gemeinschaft (LA1135/9-1) within the SPP1230. conditional mouse models of peripheral T cell lymphoma. J Exp Med 2010; 207: We thank Ekaterini Hadzoglou and Robin Wistinghausen for their excellent technical 1031–1044. assistance. We also thank Felix Hermann, Jo¨rg Kirberg and Ralf Ku¨ ppers for the highly 23 Wang X, Werneck MB, Wilson BG, Kim HJ, Kluk MJ, Thom CS et al. TCR-dependent valuable discussions of the presented data. transformation of mature memory phenotype T cells in mice. J Clin Invest 2011; 121: 3834–3845. 24 Miething C, Grundler R, Fend F, Hoepfl J, Mugler C, von Schilling C et al. The oncogenic fusion protein nucleophosmin-anaplastic lymphoma kinase (NPM-ALK) AUTHOR CONTRIBUTIONS induces two distinct malignant phenotypes in a murine retroviral transplantation SN designed experiments, performed research, analyzed data and wrote the model. Oncogene 2003; 22: 4642–4647. paper. TH performed research, analyzed data and edited the paper. NAG, MP, 25 Bonzheim I, Geissinger E, Roth S, Zettl A, Marx A, Rosenwald A et al. Anaplastic BR and AK performed research. HMJ performed Ig-rearrangement analysis, SH large cell lymphomas lack the expression of T-cell receptor molecules or and MLH performed histological analyses, SG and IR performed the statistical molecules of proximal T-cell receptor signaling. Blood 2004; 104: 3358–3360. 26 Geissinger E, Sadler P, Roth S, Grieb T, Puppe B, Muller N et al. Disturbed analyses. DVL designed experiments, analyzed data and wrote the paper. expression of the T-cell receptor/CD3 complex and associated signaling molecules in CD30 þ T-cell lymphoproliferations. Haematologica 2010; 95: 1697–1704. 27 Newrzela S, Cornils K, Heinrich T, Schlager J, Yi JH, Lysenko O et al. REFERENCES Retroviral insertional mutagenesis can contribute to immortalization of mature T 1 Dunleavy K, Piekarz RL, Zain J, Janik JE, Wilson WH, O’Connor OA et al. lymphocytes. Mol Med 2011; 17: 1223–1232. New strategies in peripheral T-cell lymphoma: understanding tumor biology and 28 Johnson LA, Morgan RA, Dudley ME, Cassard L, Yang JC, Hughes MS et al. Gene developing novel therapies. Clin Cancer Res 2010; 16: 5608–5617. therapy with human and mouse T-cell receptors mediates cancer regression and 2 Min B, Paul WE. Endogenous proliferation: burst-like CD4 T cell proliferation in targets normal tissues expressing cognate antigen. Blood 2009; 114: 535–546. lymphopenic settings. Semin Immunol 2005; 17: 201–207. 29 Morgan RA, Dudley ME, Wunderlich JR, Hughes MS, Yang JC, Sherry RM et al. 3 Kirberg J, Berns A, von Boehmer H. Peripheral T cell survival requires continual Cancer regression in patients after transfer of genetically engineered ligation of the T cell receptor to major histocompatibility complex-encoded lymphocytes. Science 2006; 314: 126–129. molecules. J Exp Med 1997; 186: 1269–1275. 30 Kelly JA, Spolski R, Kovanen PE, Suzuki T, Bollenbacher J, Pise-Masison CA et al. 4 Nesic D, Vukmanovic S. MHC class I is required for peripheral accumulation of Stat5 synergizes with T cell receptor/antigen stimulation in the development of CD8 þ thymic emigrants. J Immunol 1998; 160: 3705–3712. lymphoblastic lymphoma. J Exp Med 2003; 198: 79–89. 5 Surh CD, Sprent J. Regulation of mature T cell homeostasis. Semin Immunol 2005; 31 Tsatsanis C, Vaporidi K, Zacharioudaki V, Androulidaki A, Sykulev Y, Margioris AN 17: 183–191. et al. Tpl2 and ERK transduce antiproliferative T cell receptor signals and inhibit 6 Troy AE, Shen H. Cutting edge: homeostatic proliferation of peripheral transformation of chronically stimulated T cells. Proc Natl Acad Sci USA 2008; 105: T lymphocytes is regulated by clonal competition. J Immunol 2003; 170: 672–676. 2987–2992. 7 Newrzela S, Cornils K, Li Z, Baum C, Brugman MH, Hartmann M et al. Resistance of 32 Schwab R, Szabo P, Manavalan JS, Weksler ME, Posnett DN, Pannetier C et al. mature T cells to oncogene transformation. Blood 2008; 112: 2278–2286. Expanded CD4 þ and CD8 þ T cell clones in elderly humans. J Immunol 1997; 8 Meyer J, Rhein M, Schiedlmeier B, Kustikova O, Rudolph C, Kamino K et al. 158: 4493–4499. Remarkable leukemogenic potency and quality of a constitutively active 33 Tokudome S, Tokunaga O, Shimamoto Y, Miyamoto Y, Sumida I, neurotrophin receptor, deltaTrkA. Leukemia 2007; 21: 2171–2180. Kikuchi M et al. Incidence of adult T-cell leukemia/lymphoma among

Leukemia (2012) 2499 – 2507 & 2012 Macmillan Publishers Limited Homeostatic control of T-cell lymphoma S Newrzela et al 2507 human T-lymphotropic virus type I carriers in Saga, Japan. Cancer Res 1989; 49: antigens in skin lesions of a patient with Sezary syndrome. J Am Acad Dermatol 226–228. 2005; 53: 897–900. 34 Humme D, Lukowsky A, Steinhoff M, Beyer M, Walden P, Sterry W et al. 36 Smedby KE, Baecklund E, Askling J. Malignant lymphomas in autoimmunity and Dominance of nonmalignant T-cell clones and distortion of the TCR repertoire in inﬂammation: a review of risks, risk factors, and lymphoma characteristics. Cancer the peripheral blood of patients with cutaneous CD30 þ lymphoproliferative Epidemiol Biomarkers Prev 2006; 15: 2069–2077. disorders. J Invest Dermatol 2009; 129: 89–98. 37 Carbonnel F, Grollet-Bioul L, Brouet JC, Teilhac MF, Cosnes J, Angonin R et al. Are 35 Knol AC, Guilloux Y, Quereux G, Marques-Briand S, Pandolﬁno MC, complicated forms of celiac disease cryptic T-cell lymphomas? Blood 1998; 92: Khammari A et al. CD8( þ ) T lymphocytes reactive against Epstein-Barr virus 3879–3886.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

& 2012 Macmillan Publishers Limited Leukemia (2012) 2499 – 2507