Retrovirus-Mediated Gene Transfer of NPM-ALK Causes Lymphoid Malignancy in Mice

KARYOTYPIC ANALYSIS has identified a t(2; 5) (p23; q35) chromosomal translocation in approximately 5% to 10% of all non-Hodgkin's lymphomas (NHL), with the majority of cases being of the anaplastic large-cell type (anaplastic large-cell lymphoma [ALCL]) that express the CD30/Ki-1 antigen on their cell surface.1-6 We and others recently cloned the t(2; 5),7,8 showing it to involve the gene loci encoding nucleophosmin (NPM) on chromosome 5 and anaplastic lymphoma kinase (ALK) on chromosome 2. NPM is a highly conserved, ubiquitously expressed RNA-binding nucleolar phosphoprotein that includes among its functions the shuttling of ribonucleoproteins between the nucleolus and the cytoplasm to the ribosomes,9-13 whereas ALK is a novel orphan receptor tyrosine kinase with expression normally restricted to neural tissues that is most closely related to leukocyte tyrosine kinase.14,15 The fusion of these two genes leads to the constitutive activation of the ALK catalytic function and an unregulated mitogenic signal,8 16 although the exact mechanism(s) by which NPM-ALK contributes to the development of lymphoma remains to be determined.

Clinically, t(2; 5)-positive large-cell lymphomas are associated with frequent extranodal involvement including the skin, bone, soft tissues, gastrointestinal tract, and lung.5,6,17-19 These tumors typically behave as aggressive, intermediate- to high-grade NHL, with most patients having advanced stage disease at diagnosis. Although the vast majority of cases with a t(2; 5) rearrangement possess anaplastic large-cell histology, this alteration also occurs in the diffuse mixed-cell, diffuse large-cell, and immunoblastic types of NHL.19-21 The expression of the CD30/Ki-1 antigen by NHL does not strictly correlate with the presence of NPM-ALK, with at most two-thirds of CD30/Ki-1–positive ALCL also being NPM-ALK–positive and occasional CD30/Ki-1–negative ALCL being positive for the fusion gene.19-21 Although NPM-ALK is occasionally found in lymphomas of B-cell lineage, most lymphomas harboring this fusion protein express T-lymphoid markers or have a null phenotype.5,6,21 ALCL often responds well to treatment, and some studies suggest that ALCL/Ki-1–positive lymphomas bearing the t(2; 5) may have a better overall survival than other large-cell subtypes; nevertheless, 20% to 30% of these patients eventually succumb to their disease despite aggressive therapeutic intervention.17-20 22-27 

To date, the oncogenic properties of NPM-ALK have been shown only by in vitro transformation assays using immortalized rodent fibroblast cell lines such as NIH-3T3 or Fr3T3.8,16 The expression of NPM-ALK in these cell lines induces focus formation and anchorage-independent growth in soft agar; in addition, NIH-3T3 or Fr3T3 cells expressing NPM-ALK form tumors in nude mice. Analysis of various molecularly engineered NPM-ALK mutants using these in vitro transformation assays has shown that an intact NPM segment is absolutely required for NPM-ALK–mediated oncogenesis, because the NPM portion of the fusion protein provides an oligomerization motif that mediates NPM-ALK homodimerization, leading to intermolecular cross-phosphorylation and constitutive activation of the ALK phosphotransferase activity.8,16,28 The biologic effect of this activation process is the production of an unregulated proliferative signal mediated by the interaction of NPM-ALK with mitogenic substrate proteins that transmit growth responses to the nucleus upon their phosphorylation.28 These data provide strong evidence for a causative role of NPM-ALK in tumor formation, although proof of lymphoma induction in an animal model is still lacking.

To assess the transforming potential of NPM-ALK in an animal model, we infected murine bone marrow cells using a retroviral construct, pSRαMSVtkneo–NPM-ALK, that contains the human NPM-ALK cDNA.7 Lethally irradiated recipient mice transplanted with this infected marrow developed lymphoid malignancies within 4 to 6 months that involved the mesenteric lymph nodes, with metastases to the lungs, kidneys, liver, spleen, and the paraspinal area. The tumors expressed NPM-ALK, and injection of tumor cells into sublethally irradiated secondary recipients also led to tumor formation. These results support a causative role for NPM-ALK in the generation of human large-cell lymphomas bearing the t(2; 5) translocation.

Cell culture. NIH-3T3 mouse fibroblasts (kindly provided by Dr Martine Roussel, Department of Tumor Cell Biology, St Jude Children's Research Hospital, Memphis, TN) and 293T cells29 were grown in Dulbecco's modified Eagle's medium (DMEM; Mediatech, Herndon, VA) supplemented with 10% fetal calf serum (FCS; Bio Whittaker, Walkersville, MD), 100 U/mL penicillin and 100 μg/mL streptomycin (P/S; GIBCO, Grand Island, NY), and 2 mmol/L L-glutamine (GIBCO). SUP-M2, a human lymphoma cell line containing the t(2; 5),30 was grown in RPMI 1640 (Bio Whittaker) supplemented with 20% FCS, P/S, and L-glutamine.

Preparation of retroviral construct pSRαMSVtkneo–NPM-ALK. Clone 7-8 contains the full-length NPM-ALK coding sequence and 5′ and 3′ untranslated (UT) sequences inserted into the EcoRI site of pBluescript SK(+) (Stratagene, La Jolla, CA).7 To prepare an NPM-ALK retroviral construct, the 5′ UT and 3′ UT ends of clone 7-8 were trimmed using conventional polymerase chain reaction (PCR) methods before cloning into the HindIII site of the ecotropic retroviral vector pSRαMSVtkneo.31 

Production of retroviral stocks. Helper-free retrovirus was generated by transfecting 293T cells with 20 μg each of ψ2 packaging DNA³¹ (kindly provided by Dr Owen Witte, Howard Hughes Medical Institute, University of California, Los Angeles, CA) and DNA from empty pSRαMSVtkneo or the pSRαMSVtkneo–NPM-ALK construct using the calcium phosphate precipitation method.32 Beginning 30 hours after transfection, retroviral stocks were collected at 6- to 12-hour intervals for the next 36 hours. Harvested medium was filtered through 0.45-μm filters (Gelman Sciences, Ann Arbor, MI) and stored at −80°C until use. Viral titration with NIH-3T3 cells in G418-containing media (Geneticin; GIBCO), using viral supernatant containing 8 μg/mL of polybrene (Sigma, St Louis, MO), consistently yielded relatively low titers of approximately 10³ infectious particles per milliliter of supernatant for empty pSRαMSVtkneo and ≤10² infectious particles per milliliter of supernatant for the pSRαMSVtkneo–NPM-ALK construct.

Animals. BALB/cByJ mice were purchased at 4 to 6 weeks of age from the Jackson Laboratories (Bar Harbor, ME) and kept in standard microisolator cages in the animal facility at St Jude Children's Research Hospital. All animal experiments were performed in accordance with institutional guidelines approved by the Animal Care Committee of St Jude Children's Research Hospital.

Transduction of murine bone marrow cells. Bone marrow was harvested from femurs and tibiae of 10-week-old female BALB/cByJ mice 2 days after the intraperitoneal administration of 150 mg/kg 5-fluorouracil (SoloPak Laboratories, Elk Grove Village, IL). Cells were prestimulated for 48 hours in untreated suspension culture dishes (100-mm, catalogue no. 25070-100; Corning Glass Works, Corning, NY) at a concentration of 5 × 10⁵ cells/mL in DMEM supplemented with 20% FCS, P/S, L-glutamine, 100 ng/mL recombinant rat stem cell factor (rrSCF ), and 100 ng/mL recombinant human interleukin-6 (rhIL-6) (both cytokines from Amgen, Thousand Oaks, CA). For viral infection, non–tissue-culture–treated bacterial dishes (100-mm, catalogue no. 08-757-12; Falcon, Lincoln Park, NJ) were used after coating with 8 μg/cm2 fibronectin fragment CH-296³³ (Takara Shuzo, Otsu, Japan) dissolved in 5 mL phosphate-buffered saline (PBS; GIBCO) for 2 hours at room temperature. After blocking with 2% bovine serum albumin (BSA; Fraction V; Boehringer Mannheim, Indianapolis, IN) for 30 minutes at room temperature, the dishes were washed once with Hank's Balanced Salt Solution (HBSS) supplemented with 2.5% (vol/vol) 1 mol/L HEPES (both from GIBCO). Five million prestimulated donor cells were incubated on dishes coated with fibronectin fragment CH-296 with 10 mL of virus supernatant supplemented with 100 ng/mL rrSCF and 100 ng/mL rhIL-6. Polybrene (Sigma) was added at a concentration of 0.8 μg/mL. Ten milliliters of fresh virus-containing supernatant including growth factors was added after 2 and 22 hours, with the incubation repeated for an additional 2 and 22 hours (48 hours of total incubation). Thereafter, nonadherent cells were decanted, and adherent hematopoietic cells were collected from the cultures using cell dissociation buffer (GIBCO; catalogue no. 13151-014) according to the manufacturer's instructions. The adherent cells were added to the nonadherent fraction, washed twice, and suspended in approximately 1 mL HBSS/HEPES. One million cells were injected retro-orbitally into BALB/cByJ male recipient mice 24 hours after they had received a lethal dose of total body irradiation (450 rads twice within a 4-hour interval). The mice were caged in microisolator cages, received chlorinated water and sterile food, and were observed for the development of disease.

Western blot analysis. Frozen tissue samples were crushed, then lysed in 1× Laemmli's sample buffer, and boiled before analysis by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) on 7.5% polyacrylamide gels. The proteins were then transferred to a polyvinylidene difluoride (PVDF ) membrane filter (Immobilon-P; Millipore, Bedford, MA) using a semidry blotting system (SemiPhor Transfer Unit; Hoefer Scientific Instruments, San Francisco, CA). The PVDF membrane was incubated for 30 minutes at room temperature in blocking buffer (rinsing buffer consisting of 10 mmol/L Tris, pH 7.2, 150 mmol/L NaCl, with the addition of 3% nonfat dry milk), followed by 2 hours of incubation at room temperature in blocking buffer containing the anti-ALK 11 polyclonal antibody diluted 1:500. Preparation of the anti-ALK 11 rabbit polyclonal antibody, directed against residues 419-520 of NPM-ALK, has been described.15 Five 4-minute room temperature washes with rinsing buffer were performed after incubation with antibody, and the membrane was incubated for 1 hour in blocking buffer containing horseradish peroxidase (HRP)-conjugated rat antirabbit IgG diluted 1:2,000 as a secondary antibody at room temperature and then washed seven times (4 minutes each) at room temperature. Immunoreactive proteins were detected by enhanced chemiluminescence (ECL kit; Amersham, Arlington Heights, IL).

Genotypic analysis of tumors. Genomic DNA was extracted from snap-frozen tumor specimens and digested with EcoRI for Southern analysis with the J_H and T-cell receptor β (TCRβ) probes (both generously provided by Dr Michael Hsiao, Department of Neurosurgery, Stanford University Medical Center, Stanford, CA) as well as the murine Alk cDNA probe, and with HindIII for analysis with the J_κ probe (kindly provided by Dr Christopher Coleclough, Department of Immunology, St Jude Children's Research Hospital, Memphis, TN). Ten micrograms of DNA was run on 1% agarose gels, blotted to nylon filters (Duralon-UV; Stratagene), and then hybridized with radioactive probes ³²P-labeled using the random priming method.34 Posthybridization washes were performed twice with 0.1× SSC/0.1% SDS at 50°C for 30 minutes. The probes for Southern hybridization included a mouse 2.2-kb BamHI/EcoRI J_H genomic fragment,35 a mouse 1-kb HindIII/Xba I J_κ genomic fragment,36 a mouse 0.73-kb Pst I TCR C_βl cDNA fragment,37 and a 187-bp mouse Alk cDNA fragment representing the exon encoding the juxtamembrane segment of the receptor.15 The murine Alk cDNA hybridization was performed to assess proviral integration sites, taking advantage of the presence of a single EcoRI restriction endonuclease site within the pSRαMSVtkneo–NPM-ALK construct (see Fig 1), together with unique sites in the host genomic DNA, to produce variously sized fragments indicative of a specific clone of cells.

Immunocytochemical staining. Lymphomatous tumors as well as normal mouse spleen, Peyer's patches, liver, and kidney were embedded in paraffin or snap frozen in liquid nitrogen. Cryostat sections (6 mm) were cut, fixed in acetone, air-dried, and stored at −20°C.38 

Rabbit antibodies raised against peptides of the CD5, CD8, and CD79a antigens were obtained from one of the investigators' laboratories.39 DAKO-CD3 (anti–T-cell), HRP-conjugated rabbit anti-fluorescein isothiocyanate (FITC) Ig (P0404) and HRP-conjugated goat antirabbit-Ig (P0447) were obtained from DAKO A/S (Glostrup, Denmark). FITC-conjugated goat antibodies against mouse Ig subtypes and HRP-conjugated rabbit antimouse λ light chain were obtained from Southern Biotechnologies Associates (Birmingham, AL). Hamster antimouse CD30 antibody was obtained from PharMingen (San Diego, CA), whereas HRP-conjugated goat antihamster Ig was obtained from Caltag (Burlingame, CA).

Immunoperoxidase staining was performed using a two-stage technique.40 Briefly, cryostat tissue sections were incubated at room temperature for 30 minutes with one of the rabbit antibodies recognizing ALK, CD3, CD5, CD8, or CD79a. After washing in PBS, the sections were then incubated with HRP-conjugated goat antirabbit Ig for a further 30 minutes. To visualize the CD30 antigen, sections were incubated with the hamster antimouse CD30 reagent, washed, and then incubated with the HRP-goat antihamster Ig. After a final washing, the peroxidase reaction was developed using diaminobenzidine tetrahydrochloride (DAB; Sigma) and hydrogen peroxide (0.01% vol/vol). The slides were counterstained with hematoxylin before mounting.

For isotyping the tumor, cryostat sections were incubated with one of the following FITC-conjugated goat antibodies recognizing mouse IgM, IgG1, IgG2a, IgG2b, IgG3, and a reagent against both heavy and light mouse Igs. After 30 minutes, the sections were washed and then incubated in HRP-conjugated rabbit anti-FITC Ig. To establish which light chain the tumor cells contained, cryostat sections were incubated for 30 minutes with HRP-conjugated rabbit antimouse λ antibody. The peroxidase reaction for all of these slides was developed as described above.

Transfer of tumor cells to secondary recipients. After a single sublethal dose of radiation (450 rads), secondary mice received 4 × 10⁶ cells from various tissues of a mouse affected with lymphoma. This material was injected retro-orbitally in a volume of 200 μL. The mice were then maintained on normal diets in microisolator cages and observed for disease.

NPM-ALK causes lymphoid malignancy in mice. Two separate but identical sets of experiments with infectious retroviral stocks were performed in this study, the first using a total of 8 and the second a total of 3 male mice transplanted with bone marrow from female donors at 10 weeks of age. In the initial transplantation, 4 mice, designated GT1 through GT4 (gene transfer mice 1 through 4), received stocks produced using the retroviral construct, pSRαMSVtkneo–NPM-ALK, that contains the human NPM-ALK cDNA (Fig 1). Another 4 mice, designated C1 through C4 (control mice 1 through 4), served as a control group, receiving empty pSRαMSVtkneo-infected marrow. All 4 animals in the control group remained alive and well up to 11 months after transplantation. However, 1 control animal did spontaneously develop a large-cell lymphoma 11 months after transplantation at 13 months of age. This was most likely caused not only by advanced age, but also by irradiation, both of which greatly increase the spontaneous incidence of lymphomas in BALB/c mice.41 42 Two of the four mice (GT2 and GT4) transplanted with marrow containing the NPM-ALK construct developed lymphoid malignancies at 14 and 15 weeks postprocedure, respectively, that involved the mesenteric lymph nodes, with metastases to the lungs, kidneys, liver, and spleen (Table 1 and Fig 2A). An additional mouse (GT1) was found dead at 13 weeks posttransplantation by animal care personnel but was destroyed before necropsy could be performed. A remaining animal (GT3) was killed at 14 weeks after transplantation because of severe cachexia (weight loss from ∼30 gm to 17 gm in the preceding 8 weeks), but there was no evidence of malignancy seen on either gross or histologic examination. Detailed gross and histologic examination of mice GT2 and GT4 showed no other abnormalities, including the lack of hepatosplenomegaly and the absence of significant blast cell infiltration of the bone marrow. Although analysis of the peripheral blood showed anemia with a hematocrit level of 21.4% for animal GT4, the white blood cell number and differential were normal, as was the platelet count. A recently performed second transplantation of pSRαMSVtkneo–NPM-ALK–infected bone marrow cells has resulted in the development of lymphoma with identical histology, immunophenotype, and pattern of involvement in 2 of the 3 transplanted animals (GT5 and GT6; Table 1).

NPM-ALK–induced lymphomas are composed of immunoblast-like B cells. Histologically, the tumors were composed of a relatively uniform population of large cells of immunoblastic morphology with basophilic cytoplasm, centrally placed nuclei, and distinct nucleoli (Fig 2B). Areas of the immunoblastic lymphoma also showed plasmacytoid features with eccentric nuclei and peripherally marginated, clockface-like chromatin. A polyclonal antibody against ALK stained both the cytoplasm and nucleoli of the lymphoma cells (Fig 2C), and immunoblotting of tumor protein extracts confirmed expression of the 75-kD NPM-ALK protein (Fig 3). Immunophenotypic analysis of the tumors showed that the tumor cells expressed IgM heavy chain (Fig 2E) and κ light chain, but were CD30⁻. All of the tumor cells were also CD3⁻ (Fig 2D), CD5⁻, CD8⁻, and CD79a⁻ (data not shown). However, normal mouse lymphocytes expressing T-cell antigens were observed admixed among the tumor cells (Fig 2D). Genotyping by Southern blot analysis of EcoRI- and HindIII-digested DNA from various tumor tissues was performed to detect rearrangements of the Ig heavy- and κ light-chain and T-cell receptor loci and confirmed the B-cell origin of the lymphomas (Fig 4). Specifically, a murine J_H probe showed rearrangements in the Ig heavy-chain locus; in addition to an approximately 6.6-kb germline band observed in DNA from the liver of a healthy control mouse (Fig 4A, lane 1), rearranged EcoRI fragments were detected in tumor samples from both animals GT2 (Fig 4A, lanes 2, 3, and 4) and GT4 (Fig 4A, lanes 5 and 6). Furthermore, a murine J_κ probe showed rearrangements in HindIII digests of tumor DNA from animals GT2 and GT4 in the Ig κ light-chain locus, with the germline band at approximately 2.9 kb (Fig 4B). By contrast, no rearrangement of the T-cell receptor β chain locus was evident (Fig 4C). Thus, both immunophenotypic and genotypic analysis were consistent with a B-cell origin for the lymphomas induced by NPM-ALK in our tumor model.

Clonality of NPM-ALK–associated lymphomas. Southern blot analysis for proviral integration sites in EcoRI-digested genomic DNA from the tumor cells of mice GT2 and GT4 using a murine Alk cDNA fragment as a probe showed that, in addition to a germline Alk band of approximately 23 kb, several retroviral insertions were detected for both animals (Fig 4D). Consistent with our J_H and J_κ probe analysis of tumor cells from mouse GT2 (Fig 4A and B), proviral integration analysis showed identical hybridization patterns in the DNAs from tumors found in the lungs, kidneys, and mesenteric nodes, indicating that each arose from a single malignant clone that had undergone metastatic spread. By contrast, Southern hybridization of DNAs prepared from a mesenteric node tumor and the spleen of mouse GT4 with the Alk probe suggested the presence of two distinct malignant clones in this animal. In the mesenteric tumor, roughly 14-kb and much less prominent approximately 6.5-kb fragments were observed (Fig 4D, lane 5). The markedly unequal intensity of these bands suggested the presence of two separate clones within the nodal tumor, a hypothesis supported by the presence of a small population of tumor cells metastatic to the GT4 spleen that contained only the approximately 6.5-kb proviral integration fragment (Fig 4D, lane 6). Thus, the lymphomas induced by NPM-ALK were either monoclonal or oligoclonal in nature.

Serial transplantation of NPM-ALK–induced lymphoma into secondary recipients. Cells derived from the mesenteric nodal and subcapsular kidney tumors, the spleen, and the bone marrow of mouse GT4 were injected into the retro-orbital space of sublethally irradiated secondary recipients to determine the transplantability of NPM-ALK–induced lymphomas. Ten of these 13 secondary recipient mice developed tumors within a period ranging from 5 to 38 weeks. Consistent with molecular evidence (Fig 4D, lane 6) for the presence of a small number of lymphomatous cells in the spleen of GT4 (despite the absence of tumor based on routine histologic examination), all 5 animals receiving splenic cells developed lymphoma (Table 2). In 2 such animals, tumors developed at retro-orbital injection sites, whereas the other 3 mice suffered hind limb paralysis that resulted from spinal cord compression due to lymphoma growth within the intradural and epidural spaces. In keeping with their origin from the injected tumor cells, the lymphomas observed in the secondary recipient animals were histologically (large cells of immunoblastic morphology) and immunophenotypically (anti-ALK and IgM heavy-chain, κ light-chain–positive; negative for CD3, CD5, CD8, CD79a, and CD30) identical to the primary tumors from mouse GT4 (data not shown).

In this study, retroviral transfer of the NPM-ALK gene to mice induced lymphoid tumors that were similar to human large-cell lymphomas. Two central findings support a causative role for NPM-ALK in the pathogenesis of the murine B-cell lymphomas that we observed. First, the NPM-ALK protein was expressed in all analyzable tumors. Second, the tumors contained integrated proviral sequences and could be easily transmitted to secondary recipients. Other explanations for the development of tumors, ie, insertional mutagenesis by replication-competent helper virus or activation of endogenous retroviruses by the marrow transfer procedure itself, are very rare; in addition, not only are these mechanisms of tumorigenesis typically associated with a longer latency period, but they are unlikely to cause consecutive cases of phenotypically identical B-cell lymphoma in two series of transplantation experiments in multiple animals.

Whereas in humans, large-cell anaplastic lymphoma typically affects the lymph nodes, skin, lung, soft tissue, bone, and the gastrointestinal tract,5,17-19,23,25-27 mice with NPM-ALK–induced lymphoma lacked involvement of the skin and bones, but showed infiltration of the kidneys and paraspinal area. Most t(2; 5)-positive ALCL are CD30/Ki-1–positive, express T-lymphocyte–related antigens, and show rearrangements of the TCRβ gene. In contrast, the lymphoid tumors induced by retroviral gene transfer of NPM-ALK in mice were CD30/Ki-1–negative and lacked T-cell–associated antigens. Instead, the neoplastic cells showed rearrangements in their Ig heavy-chain and κ light-chain genes. The presence of IgM heavy-chain and κ light-chain Igs in the tumor cells was confirmed by immunocytochemical studies. The transformation by NPM-ALK of B-lymphoid cells in our lymphoma model confirms the oncogenic ability of the chimeric protein in this lineage and lends support to reports of the presence of the t(2; 5) in occasional B-cell NHL.19-21,43,44 Interestingly, like our NPM-ALK–induced mouse lymphomas, most of the translocation-positive B-lineage NHLs described in the literature have been reported to be CD30/Ki-1–negative.21,44 The lack of CD30/Ki-1 antigen expression by our mouse lymphomas casts doubt on the importance in lymphomagenesis of the reported putative physical interaction between NPM-ALK and the CD30/Ki-1 antigen.45 Thus, CD30/Ki-1 expression may not be required for oncogenesis, but rather could be of importance in modulating tumor growth and aggressiveness; this interpretation would be consistent with the previously noted growth-inhibitory effects of the CD30 ligand on ALCL cells.46 

The lymphomatous tumors that we observed could be easily transmitted to secondary recipients, consistent with their malignant nature. Transplantability, an indication of the immortality of the transplanted cells, is also a characteristic of E2A-PBX1–induced acute myeloid leukemia (AML)47 and vFMS-induced malignancies,48 but is not found in some chronic myelogenous leukemia (CML)-like diseases induced by retroviral gene transfer of P210^BCR-ABL.49,50 Despite their ease of transplantability, several attempts to establish NPM-ALK–induced lymphoma cell lines in continuous culture from the mouse tumors were unsuccessful, suggesting that expression of the chimeric protein did not result in “complete” malignant transformation, the cells apparently still being dependent on B-cell stimulatory lymphokines for growth. Other observations from our experiments also suggest that NPM-ALK expression alone may not be sufficient to produce lymphoma by itself. For example, the relatively long latency period (∼4 to 6 months) and the development of either monoclonal or pauciclonal tumors (without polyclonal proliferations) suggest that an additional mutation(s) might be required to allow outgrowth of a tumorigenic clone(s), although the latter finding could simply be due to productive infection of only a few transplanted stem cells because of the low pSRαMSVtkneo–NPM-ALK viral titers used in our experiments. The identity of the gene or genes that might cooperate with NPM-ALK in malignant transformation is unknown; however, one candidate could be cMYC, which is altered in up to one third of ALCL.51 

What are the molecular mechanisms that allow NPM-ALK to induce lymphoma? A number of identified receptor tyrosine kinase substrates are known to interact with activated receptors through their SRC homology region 2 (SH2) or phosphotyrosine-binding (PTB) domains, binding to a specific phosphotyrosine residue contained within a distinct amino acid sequence of the receptor.52,53 To date, three such substrates (IRS-1, SHC, and GRB2) have been reported to associate with NPM-ALK in fibroblasts; however, both IRS-1 and SHC appear to be nonessential for fibroblast transformation, whereas the requirement for GRB2 remains unknown.8 16 The establishment of our mouse tumor model will now allow us to determine which known (or unknown) substrates are essential for oncogenic signaling in hematopoietic cells and to identify the exact motifs of the chimera that are critical for its lymphomagenic activity.

Other activated tyrosine kinases tested in similar mouse malignancy models have been shown to be nonspecific, producing not only lymphomas of B or T lineage but also myeloid or lymphoid leukemias, with differences in presentation between humans and mice.48,49,54-56 As an example, in contrast to the essentially invariant CML phenotype produced in humans by the p210^BCR/ABL protein generated by the t(9; 22),57,58 mice transplanted with bone marrow infected with a retrovirus encoding p210^BCR/ABL developed several different hematologic malignancies; in addition to a myeloproliferative syndrome closely resembling the chronic phase of human CML, macrophage tumors and acute lymphoblastic leukemias were observed.49,50,59 Similarly, v-abl induces only pre–B-cell leukemia in mice under natural conditions, but a spectrum of leukemias including myelomonocytic, granulocytic, and pre–B-cell types under gene transfer conditions.60 Furthermore, whereas the CSF-1 receptor is monocyte/macrophage lineage-specific in its action, the v-fms gene product acts promiscuously as a kinase and can induce clonal proliferation of various bone marrow progenitors, ultimately resulting in malignancies of multiple hematologic lineages.48 Most human ALCL are of the T-cell phenotype, and retroviral gene transfer of NPM-ALK initially was thought to produce T-cell tumors as well when analyzed by flow cytometry⁶¹; as described in this report, detailed studies of Ig heavy and light chain and TCRβ gene rearrangements together with tumor immunostaining data showed that NPM-ALK actually appears to selectively transform murine lymphoid cells of B lineage in our model. The Moloney murine sarcoma virus LTR present in the pSRαMSVtkneo retroviral vector has been shown to drive gene expression in most hematopoietic cell types^31,62; hence, the basis for this observed selectivity is unclear. For retroviral gene transfer of BCR-ABL it has been shown that the choice of internal promoter, retroviral regulatory sequences, culturing conditions, infection procedure, and genetic background of the grafted mice all may have an influence on the nature and kinetics of the induced hematologic disease.59,63 For NPM-ALK, the alteration of one or more of these variables in retroviral transduction studies, the use of specific promoters to produce conventional transgenic mice, or the generation of a knock-in animal model64 with expression driven by the endogenous Npm promoter will all be of interest to establish the complete spectrum of the chimera's transforming potential and in attempts to establish an exact murine model of CD30/Ki-1–positive ALCL.

The guidance of Dr Martine Roussel (Department of Tumor Cell Biology, St Jude Children's Research Hospital, Memphis, TN) in the production of infectious retroviral stocks is gratefully acknowledged. We very much appreciate the assistance of Dr James A. Allay, Dr Brian Sorrentino (both of the Department of Experimental Hematology, St Jude Children's Research Hospital), and Dr Michael Brown (Department of Cell and Gene Therapy, St Jude Children's Research Hospital) in the establishment of murine bone marrow transplantation. The authors thank Takara Shuzo (Otsu, Japan) for providing fibronectin fragment CH-296 and Amgen (Thousand Oaks, CA) for supplying rrSCF and rhIL-6 for these studies.