Characterization of Murine JAK2^V617F-Positive Myeloproliferative Disease

Introduction

The myeloproliferative disorders (MPD) are a heterogeneous group of disorders characterized by expansion of one or more of the myeloid lineages. Classically, MPDs have included chronic myelogenous leukemia (CML), polycythemia vera (PV), essential thrombocythemia (ET), and idiopathic myelofibrosis (IMF; ref. 1). More recently, the definition has been broadened to include a variety of additional entities, frequently referred to as “atypical MPD,” among them are chronic myelomonocytic leukemia (CMML) and chronic neutrophilic leukemia (CNL; ref. 2). Constitutively active tyrosine kinases are thought to play a central role in the pathogenesis of MPD. Examples include BCR-ABL in CML ( 3) and the FIP1L1-platelet-derived growth factor-α fusion protein in the hypereosinophilic syndrome ( 4). Recently, a mutation (V617F) in the pseudokinase (JH2) domain of the Janus-activated kinase (JAK) 2 tyrosine kinase has been identified in 65% to 97% of patients with PV, 35% to 95% of patients with IMF, and 23% to 43% of patients with ET ( 5– 8). JAK2^V617F has also been observed at lower frequencies in patients with CMML, CNL, systemic mastocytosis, and AML-FAB M7 ( 9– 11). Based on structural considerations ( 12), it is thought that the V617F mutation disrupts the autoinhibitory function of the JH2 domain, leading to constitutive activation of the kinase ( 5). JAK2 is involved in signaling downstream of many cytokine receptors, including the erythropoietin (Epo) receptor (EpoR), thrombopoietin receptor, and granulocyte macrophage colony-stimulating factor (GM-CSF) receptors. Consistent with this function, JAK2^−/− embryos die from profound anemia around days 11 to 13 of embryogenesis ( 13, 14). The molecular basis of the phenotypic diversity of JAK2^V617F-positive leukemia is unknown. PV cells are characterized by hypersensitivity to several cytokines, including Epo, stem cell factor (Scf), and interleukin (IL)-3 ( 15, 16). In line with this, expression of JAK2^V617F in growth factor–dependent cell lines, such as BaF/3 cells, reduces their cytokine dependence ( 17). Recently, two groups reported that mice transplanted with JAK2^V617F-infected bone marrow develop a PV-like disease ( 18, 19). Here, we provide a detailed characterization of JAK2^V617F-positive murine MPD.

Materials and Methods

Expression vectors. The murine JAK2 cDNA (kindly provided by James Ihle, St. Jude Children's Research Hospital, Memphis, TN) was cloned into pBluescript II SK vector (Stratagene, La Jolla, CA). The JAK2^V617F mutation was generated by using site-directed mutagenesis (GeneTailor Site-Directed Mutagenesis System, Invitrogen Carlsbad, CA). Following confirmation of the mutation by full-length DNA sequencing, the JAK2^V617F and wild-type (WT) cDNA were cloned into the retroviral vector MSCV-IRES-GFP yielding plasmids MSCV-IRES-JAK2-GFP or MSCV-IRES-JAK2^V617F-GFP.

Cell culture. BaF/3 cells [American Type Culture Collection (ATCC), Manassas, VA] were grown in RPMI 1640 (Life Technologies, Carlsbad, CA) containing 10% fetal bovine serum (FBS; HyClone, Logan, UT) and 15% WEHI-3B (ATCC)-conditioned medium as a source of murine IL-3 (WEHI-CM) at 37°C and 5% CO₂. Cells were transduced by electroporation (300 mV, 20 ms, Bio-Rad Gene Pulser Xcell, Hercules, CA) with either MSCV-IRES-GFP (MIG)-JAK2 or MIG-JAK2^V617F vector. After 48 hours, cells expressing the constructs were selected by fluorescence-activated cell sorting (FACS). To assess for cytokine hypersensitivity, BaF/3 cells expressing WT and V617F mutant JAK2 were washed with PBS and cultured for 6 days in RPMI 1640 containing 10% FBS in the presence of a WEHI-CM gradient, with parental BaF/3 cells as a control. For the cell proliferation assay, cells were incubated for 3 hours in CellTiter 96 Solution (Promega, Madison, WI) and analyzed using a microplate spectrophotometer (Bio-Rad). For Western blot, BaF/3 cells were washed with PBS and resuspended in RPMI 1640 containing 0.5% bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, MO) and incubated for 12 hours. Following incubation, cells were stimulated with ±18% FBS and 0.1, 1, 5, and 10 ng/mL IL-3 (PeproTech, Inc., Rocky Hill, NJ) for 30 minutes. Cells were then washed with ice-cold PBS and 8 × 10⁵ cells were lysed, loaded on a 4% to 15% Criterion Tris-HCl gel (Bio-Rad), transferred to polyvinylidene fluoride membrane (Millipore, Bedford, MA), and blotted with indicated antibodies (Supplementary Materials and Methods).

Retroviral infection. For generating retrovirus, Bosc23 cells (ATCC) were grown in DMEM (Life Technologies) containing 10% FBS. Cells were transfected with either MSCV-IRES-GFP (empty vector control), MIG-JAK2, or MIG-JAK2^V617F vector using Fugene 6 transfection reagent (Roche, Indianapolis, IN). The next day, cells were incubated in fresh medium for an additional 24 hours. To estimate viral titers, NIH3T3 cells (ATCC) were incubated for 48 hours with graded amounts of viral supernatant added to DMEM containing 10% FBS and 8 μg/mL polybrene (hexadimethrine bromide, Sigma-Aldrich). Cells were trypsinized (0.05% trypsin-EDTA, Life Technologies), washed in PBS (Life Technologies), and analyzed for green fluorescent protein (GFP) expression by flow cytometry (FACSAria, BD Biosciences, San Diego, CA). Titers were estimated by plotting the proportion of GFP-positive cells versus the proportion of retroviral supernatant.

Bone marrow harvest, infection, and transplantation. BALB/c (Charles River Laboratories) donor mice (6-9 weeks old) were primed by retro-orbital injection of 3 mg 5-fluorouracil (5-FU; American Pharmaceutical Partners, Schaumburg, IL). After 5 days, bone marrow was harvested from mice by flushing the femur and tibia with medium. Cells were cultured at 5 × 10⁵ per mL with viral supernatant (Supplementary Materials and Methods). After 24 hours, the medium was replaced with fresh retroviral “cocktail” and cells were incubated for additional 24 hours. Next, cells were analyzed for GFP expression by FACS. Viable cells (5 × 10⁵) were injected retro-orbitally into lethally irradiated (2 × 450 cGy) syngeneic recipients (MSCV-GFP-empty vector control, n = 6; MSCV-IRES-JAK2-GFP, n = 12; and MSCV-IRES-JAK2^V617F-GFP, n = 13). Blood counts were determined using a Vet ABC animal blood counter (Heska, Fort Collins, CO). On day 80 after bone marrow transplantation, four empty vector control mice, four JAK2 mice, and five JAK2^V617F mice were euthanized for detailed analysis. For secondary transplant, 1 × 10⁶ bone marrow cells were injected retro-orbitally into sublethally irradiated recipients (1 × 450 cGy).

Pathologic examination. Blood was collected from anesthetized mice by lethal inferior vena cava bleeding and incubated at 5°C overnight (Supplementary Materials and Methods). Serum was harvested by centrifugation at 4,600 rpm for 10 minutes. Organs were dissected and WBC from spleen and bone marrow were harvested by passing through a 70-μm nylon cell strainer (BD Biosciences, Bedford, MA) followed by red cell lysis. For colony-forming assays, 1 × 10⁵ leukocytes from the spleen of mice were plated with 1 mL medium in methycellulose, without cytokines (Methocult M3234), with cytokines without Epo (Methocult GF M3534), and with cytokines plus Epo (Methocult GF M3434, StemCell, Vancouver, British Columbia, Canada) and incubated at 37°C and 5% CO₂ for 20 days.

Histologic and immunohistochemical analysis of murine tissue. Samples from liver, spleen, heart, humerus, brain, and lung were fixed in 3.5% paraformaldehyde (Protocol, Kalamazoo, MI) and embedded in paraffin. Brains were preserved by equilibration in increasing concentrations of sucrose (10%, 20%, and 30%). Sections of 5 μm were stained with H&E. Decalcified sections of humerus were stained with H&E and reticulin stain (American Master*Tech Scientific, Inc. Lodi, CA). Antibodies were used as indicated in Supplementary Materials and Methods.

Flow cytometric analysis. White cells from bone marrow and spleen were resuspended in staining buffer (BD Biosciences). Aliquots of 1 × 10⁶ cells were stained for 20 minutes with conjugated monoclonal antibodies. Antibodies were used as described in Supplementary Materials and Methods. Cells were washed in staining buffer and analyzed by FACS. For FACS analysis of intracellular JAK2 phosphorylation, peripheral blood of one JAK2^V617F mouse and three normal BALB/c mice were collected. WBC from normal BALB/c mice were pooled together. Cells were split into three aliquots and resuspended in RPMI 1640 (Life Technologies) containing 0.5% BSA and incubated for 15 hours. Following incubation, one aliquot was stimulated with 10 ng/mL IL-3 plus 10 ng/mL granulocyte colony-stimulating factor (G-CSF; StemCell) and second aliquot was stimulated with 10 ng/mL IL3, 10 ng/mL G-CSF, and 10 μmol/L AG-490 (Calbiochem) for 60 minutes and third aliquot was left unstimulated. Cells (1 × 10⁵) were resuspended in 50 μL staining buffer (1% BSA and 0.1% sodium azide in PBS) and permeabilized using Fix&Perm cell permeabilization kit (Caltag Laboratories, Burlingame, CA and An der Grub, Vienna, Austria). Antibodies were used as indicated in Supplementary Materials and Methods. The granulocyte population was selected by forward and side scatter (Supplementary Fig. S5).

Southern blot analysis. Genomic DNA was isolated from leukocytes extracted from the spleen of mice using DNeasy Tissue kit (Qiagen, Valencia, CA). For Southern blot analysis, 10 to 15 μg genomic DNA was digested with BglII, resolved on a 1.5% agarose gel, and transferred to Hybond-N⁺ membrane. The blots were hybridized with a ³²P-labeled GFP probe and exposed to autoradiography film.

Quantification of serum cytokine levels. For measurement of Epo levels, the Quantikine mouse Epo kit (R&D Systems, Minneapolis, MN) was used according to the manufacturer's instructions. Serum levels of IL-2, IL-6, tumor necrosis factor (TNF)-α, GM-CSF, and G-CSF were measured using a bead-based immunoflourescence assay (Luminex, Inc. Austin TX), using Millipore multiscreen 96-well filter plates. Cytokine analysis kits (LINCOplex kits) were purchased from Linco Research, Inc. (St. Charles, MO). Assays were run in duplicate according to the manufacturer's instructions. A four-variable regression formula was used to calculate the sample concentrations from the standard curves.

Statistics. The SPSS statistics package was used for all statistical analyses. Continuous variables were compared by pairwise t test for two independent samples.

Results

JAK2^V617F induces IL-3 hypersensitivity in BaF/3 cells. We expressed murine JAK2^V617F and WT JAK2 (JAK2^WT) in the murine IL-3-dependent cell line BaF/3. The cells were cultured for 6 days in medium containing graded concentrations of WEHI-CM. Cells expressing JAK2^V617F showed significantly increased proliferation ( Fig. 1A ) and viability ( Fig. 1B) in low concentrations of IL-3 compared with cells expressing JAK2^WT or parental BaF/3 cells. No difference was observed in the absence of IL-3, indicating that JAK2^V617F induces cytokine hypersensitivity but not complete cytokine independence, consistent with the observation that primary PV progenitor cells are hypersensitive to cytokines but not cytokine independent ( 15). We examined the tyrosine phosphorylation of JAK2 by immunoblot analysis after serum starvation for 12 hours and stimulation with increasing IL-3 concentrations for 30 minutes ( Fig. 1C). BaF/3 cells expressing JAK2^V617F showed only a slight increase of JAK2 phosphorylation compared with parental BaF/3 cells. A more pronounced difference was detected in the phosphorylation of the JAK2 substrate signal transducer and activator of transcription 5 (STAT5) as well as Akt, a downstream effector of EpoR signaling ( 20). Consistent with the lack of cytokine-independent growth, we observed no JAK2 autophosphorylation in BaF/3 cells expressing JAK2^V617F in the absence of cytokine stimulation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

JAK2^V617F induces cytokine hypersensitivity in BaF/3 cells. Cells stably expressing JAK2^V617F (black solid line, triangles), JAK2^WT (gray dashed line, squares), and parental BaF/3 cells control (gray solid line, circles). Points, mean of triplicate results; bars, SD. A, total number of viable cells as measured by 3-(4-5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt (MTS) assay; B, viability was determined by propidium iodide exclusion. C, Western blot of parental BaF/3 and BaF/3-expressing JAK2^V617F cells.

JAK2^V617F induces trilineage MPD in mice. To examine the oncogenic potential of JAK2^V617F in vivo, we infected bone marrow from 5-FU-treated male BALB/c mice with MSCV-IRES-GFP (MIG), MIG-JAK2^WT, or MIG-JAK2^V617F. FACS analysis for GFP expression of infected bone marrow after two rounds of infection showed a low rate of GFP-positive cells for all three constructs (∼1%; data not shown). Equal numbers of viable cells were injected into lethally irradiated syngeneic recipients. Recipients of marrow infected with JAK2^WT and empty vector maintained normal blood counts during the entire observation period (166 days). In contrast, recipients of JAK2^V617F-infected marrow exhibited erythrocytosis ∼26 days after transplantation ( Fig. 2A ; Supplementary Fig. S1A). Erythrocytosis was defined as hematocrit >58% and hemoglobin >18 g/dL. All blood variables are based on reference range peripheral blood variables of healthy untransplanted BALB/c mice. Maximum hematocrit values varied between 57% and 89%, with a median of 83%, and maximum hemoglobin levels varied between 18 g/dL and 28 g/dL, with a median of 24 g/dL. Mice with erythrocytosis developed skin plethora ( Fig. 2A). Leukocytosis (defined as white cell counts >18,000/μL) developed in 12 of 13 mice. In contrast to red cell variables, maximal white cell counts were quite variable, ranging from normal to 120,000 per μL, with a predominance of granulocytes (Supplementary Fig. S1B). In two JAK2^V617F mice, slightly elevated platelet counts were observed over a period of 4 weeks, but median platelet counts in the cohort as a whole were not different from controls ( Fig. 2A). Thrombocytosis was defined as platelets >850,000/μL. Thus, most JAK2^V617F mice developed a disease that closely resembled PV. However, none of the mice died from their disease and did not exhibit significant symptoms, such as weight loss ( Table 1 ).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

JAK2^V617F induces erythrocytosis and leukocytosis in BALB/c mice. A, peripheral blood counts of empty vector (blue circles), JAK2^WT (green squares), and JAK2^V617F (red triangles) mice over 80-day time period. Points, mean of triplicate results; bars, SD. Gray patterned box, the reference ranges of healthy untransplanted male BALB/c mice. Top, left, hematocrit; top, right, WBC; bottom, left, platelets; bottom, right, skin plethora in a JAK2^V617F mouse but not JAK2^WT mouse. B, peripheral blood smear of an empty vector, JAK2^WT, and JAK2^V617F mouse. Top, H&E (HE); bottom, reticulin stain (RE). Original magnifications, ×400 and ×630.

JAK2^V617F induces extramedullary hematopoiesis. The peripheral blood smears of JAK2^WT and empty vector mice were morphologically normal. In contrast, JAK2^V617F mice showed prominent abnormalities in the red cells, including nucleated red cells, anisocytosis, spherocytes, and reticulocytosis ( Fig. 2B). White cells were predominantly mature and morphologically normal neutrophils. Randomly selected JAK2^V617F, JAK2^WT, and empty vector mice were sacrificed for detailed examination at day 80 posttransplantation, whereas observation of the remaining mice continued (latest follow-up, day 166). Necropsy revealed mild hepatomegaly and significant splenomegaly in all recipients of JAK2^V617F-infected marrow ( Table 1). All other organs appeared macroscopically normal. Histopathology of the spleen showed normal white pulp with minimal extramedullary hematopoiesis in JAK2^WT and empty vector mice. Some megakaryocytes were present but were morphologically unremarkable ( Fig. 3A ). In contrast, the white pulp of JAK2^V617F mice showed lymphoid hyperplasia and extramedullary hematopoiesis. The red pulp was expanded, with extensive extramedullary hematopoiesis and effacement of splenic architecture. Extramedullary erythropoiesis and myelopoiesis were normal, except for the presence of dysplastic megakaryocytes with increased size and large, lobular nuclei and apoptotic features, such as condensed pyknotic nuclei ( Fig. 3A). The livers of JAK2^V617F mice also displayed infiltration by hematopoietic cells with occasional atypical megakaryocytes, although not to the degree observed in the spleens ( Fig. 3B). Livers of JAK2^WT and empty vector mice showed no abnormalities ( Fig. 3B). The bone marrow histology of mice transplanted with JAK2^WT or empty vector-infected cells was normal with trilineage hematopoiesis, a myeloid to erythroid ratio of 1:1 to 2:1 and morphologically normal megakaryocytes ( Fig. 3C). In contrast, JAK2^V617F mutant mice showed hypercellular marrow with right-shifted granulocytic hyperplasia and megakaryocytic hyperplasia ( Fig. 3C). Megakaryocytes appeared in clusters, most prominently around sinuses and again exhibited dysplastic features, such as increased cell and nuclear size and abnormal nuclear lobation, and apoptotic features. Unexpectedly, erythroid cells were overall markedly reduced. A reticulin fiber stain showed no increase in reticulin fibers in JAK2^WT and empty vector control mice but moderate to severe reticulin fibrosis in JAK2^V617F mice ( Fig. 3D). Altogether, the histologic findings are consistent with a trilineage MPD resembling PV. However, the finding of reduced erythropoiesis in the marrow suggests that on day 80 the disease had evolved to a stage comparable with the ‘spent phase’ of PV with postpolycythemic myelofibrosis ( 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Histopathology of mouse tissues. Representative histologic sections from spleen (A), liver (B), and bone marrow (BM; C and D) from an empty vector, JAK2^WT, and JAK2^V617F mouse, respectively. H&E and reticulin fiber stain. Original magnifications, ×100, ×200, ×630, and ×1,000.

Thromboembolic events are common in patients with PV and ET ( 21). We therefore evaluated sections of heart and brain for signs of embolic infarctions or thrombosis. Microscopically, these tissues appeared normal, without obvious differences between JAK2^V617F mice and controls (data not shown). This is consistent with the lack of clinical signs and symptoms in JAK2^V617F mice.

Constitutive JAK2 phosphorylation in murine leukocytes expressing JAK2^V617F. To test if the mutant JAK2^V617F kinase in murine leukocytes is constitutively active, we analyzed by FACS peripheral blood granulocytes for phosphorylated JAK2. Peripheral blood leukocytes of a JAK2^V617F mouse and normal control BALB/c mice were cultured in cytokine-free medium for 15 hours. Subsequently, aliquots of cells were stimulated in the presence and absence of IL-3 and G-CSF and/or 10 μmol/L AG490, a JAK2 inhibitor ( 5). The JAK2^V617F mouse was leukemic with a white cell count of 39,000 cells/μL and 41% GFP-positive peripheral WBC (data not shown). FACS analysis showed pronounced JAK2 phosphorylation in cytokine unstimulated GFP-positive compared with GFP-negative peripheral blood granulocytes of the JAK2^V617F mouse and compared with granulocytes of normal BALB/c mice ( Fig. 4A ). After cytokine stimulation for 60 minutes, the GFP-positive peripheral blood granulocytes of the JAK2^V617F mouse showed a further increase in JAK2 phosphorylation ( Fig. 4B, top). Treatment with AG490 reduced phosphorylated JAK2 levels below baseline, consistent with constitutive JAK2 phosphorylation in the leukemic cells ( Fig. 4B, bottom).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

JAK2^V617F-induced constitutive JAK2 phosphorylation in peripheral granulocytes. A, cells cultured without cytokines of a JAK2^V617F mouse (top) and normal control BALB/c mice (red, bottom). Phosphorylated JAK2 was analyzed in GFP-positive cells (green) versus GFP-negative cells (blue). B, GFP-positive cells of a JAK2^V617F mouse cultured without cytokines (top, green; bottom, blue) and then stimulated or not for 60 minutes with IL-3 and G-CSF (top, red; bottom, green) or stimulated in the presence of 10 μmol/L AG-490 (bottom, red).

JAK2^V617F-induced MPD is oligoclonal. DNA extracted from the spleen of four JAK2^V617F mice and one JAK2^WT mouse was probed with GFP to determine the numbers of retroviral integrations/clones (Supplementary Fig. S2). Two JAK2^V617F mice showed one band (lanes 2 and 4) and two mice showed two bands (lane 3; data not shown), consistent with a monoclonal or biclonal leukemic population. No bands were distinguishable in JAK2^WT mice (lane 1). The low number of retroviral integration sites is in line with the low infection rate of transplanted bone marrow.

JAK2^V617F induces expansion of hematopoietic stem/progenitor cells. To define more precisely the lineage and differentiation stages of the hematopoietic cells involved in the JAK2^V617F-induced MPD, we did multicolor FACS analysis of bone marrow and spleen cells and immunohistochemistry on tissue sections (Supplementary Figs. S3 and S4). In the bone marrow, the proportion of mature myeloid cells, characterized by expression of Gr-1 and CD11b, was consistently increased in JAK2^V617F mice compared with controls ( Table 1). As expected from light microscopy, there were few lymphoid cells (CD19⁺ or CD90⁺), with little difference between the groups. FACS of splenic cells from JAK2^V617F mice revealed >50% of cells expressing myeloid markers ( Table 1), consistent with myeloperoxidase positivity by immunohistochemistry (Supplementary Fig. S3C). The proportion of cells with a B-cell phenotype (CD19⁺) was increased in both JAK2^WT and JAK2^V617F mice compared with empty vector mice, consistent with the histologic finding of lymphoid hyperplasia in the white pulp and confirmed by immunohistochemistry using the B220 monoclonal antibody (Supplementary Fig. S3E). The proportion of bone marrow cells with a T-cell phenotype (CD90⁺) was comparable between the three groups, but slightly reduced in the spleen of JAK2^V617F mice, reflecting the relative increase in myeloid cells.

The fact that JAK2^V617F is found in PV, ET, and IMF suggests that the mutation may be acquired by a multipotent stem cell, with additional factors determining the disease phenotype. Consistent with this notion, recent reports have suggested that JAK2^V617F-positive ET is closely related to PV, with extrinsic and possibly genetic factors modulating the penetrance of the PV phenotype ( 22, 23).

The BALB/c mice used in our experiments express only low levels of Sca-1; therefore, identification of stem cells based on the combination of CD117 (KIT) and Sca-1 expression was not possible. However, using a combination of lineage negativity and CD117 positivity, a cell population can be identified that contains stem and progenitor cells ( 24, 25). JAK2^V617F mice had a significantly higher percentage of lineage-negative (Lin⁻)/CD117⁺ cells in the bone marrow and spleen compared with mice transplanted with marrow infected with vector control ( Fig. 5A, left ). Unexpectedly, the Lin⁻/CD117⁺ population was also expanded in the marrow of mice transplanted with JAK2^WT-infected marrow. A much more pronounced expansion of Lin⁻/CD117⁺ cells in the spleens of JAK2^V617F mice relative to empty vector and JAK2^WT was observed. The small proportion of stem and progenitor cells in the spleens of the JAK2^WT mice (∼0.2%) and the vector control mice (∼0.1%; Supplementary Fig. S4C) is consistent with the presence of minimal extramedullary hematopoiesis in spleens of normal mice ( 26). Overall, our data show that JAK2^V617F leads to a significant expansion of the stem and progenitor cell compartment in the spleen and the bone marrow. The relatively discrete expansion in the marrow may reflect the fact that at the time of the histology the disease had already evolved to ‘spent phase’.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

FACS analysis of mouse tissue from mice transplanted with marrow infected with empty vector (white), JAK2^WT (gray), and JAK2^V617F (black). Columns, mean; bars, SD. A, left, Lin⁻, CD117⁺ hematopoietic stem cell/progenitor population in bone marrow and spleen of mice; right, ratio JAK2^V617F mice versus MSCV empty vector mice of Lin⁻CD117⁺ hematopoietic stem cell/progenitor cells, methycellulose colonies without cytokines (−Cytokines), colonies with cytokines plus Epo (+Cytok. +Epo), and colonies with cytokines but no Epo of cells extracted from spleen (+Cytok. −Epo). B, left, Lin⁻, CD117⁺, Fcγ receptor^+/−, and CD34^+/− progenitor populations in bone marrow and spleen of mice; right, Lin⁻, CD117⁺, CD9⁺, Fcγ receptor^low, and CD41⁺ megakaryocyte progenitor population in bone marrow and spleen of mice. C, left, Lin⁻, CD117⁺, and CD71^+/− erythroid progenitors in bone marrow and spleen of mice; right, percentage of EpoR⁺ cells of Lin⁺, CD71^high/low, and Ter-119^high/low erythroblast cells. D, serum Epo, G-CSF, and TNF-α levels of JAK2^WT-infected mice (gray) and JAK2^V617F-infected mice (black).

To investigate whether the clonogenic activity of splenic cells was correlated with the relative proportion of stem and progenitor cells (Lin⁻/CD117⁺ cells) in the spleens of JAK2^V617F mice compared with controls, we assessed colony formation in semisolid medium. Equal numbers of leukocytes isolated from the spleens were plated in methylcellulose with and without cytokines (IL-3, IL-6, and Scf) and with and without Epo. After 20 days in culture, colonies were counted ( Table 1). No growth was observed in any of the groups in the absence of cytokines. In the presence of cytokines plus Epo, a 19-fold increase of colonies was observed in cultures from JAK2^V617F mice compared with empty vector mice, which is significantly higher than the 10-fold increase of Lin⁻/CD117⁺ cells in the JAK2^WT mice compared with empty vector mice, suggesting increased clonogenic activity per Lin⁻/CD117⁺ cells in optimal growth factor concentrations ( Fig. 5A, right). With cytokines but without Epo, a more pronounced difference in colony formation between JAK2^V617F mice and empty vector mice was observed, consistent with increased clonogenic potential of JAK2^V617F cells under suboptimal growth factor conditions.

To determine how the relative proportions of the earliest committed progenitor cells are affected by JAK2^V617F, we analyzed by FACS Lin⁻/CD117⁺ cells according to Fcγ receptor and CD34 expression, which allows separating the cell population into common myeloid progenitors (CMP), granulocyte-monocyte progenitors (GMP), and megakaryocyte-erythrocyte progenitors (MEP; ref. 27). The proportion of the three different progenitor populations (CMP, GMP, and MEP) to each other was similar to published data in C57BL mice, which express Sca-1 on their hematopoietic stem cells (Supplementary Fig. S4D; ref. 28).

In the bone marrow, both JAK2^WT and JAK2^V617F mice showed a relative increase of GMP, CMP, and MEP ( Fig. 5B, left), in line with the overall expansion of Lin⁻/CD117⁺ cells in the marrow. The spleens of JAK2^V617F mice contained significantly increased CMP, GMP, and MEP compared with the empty vector and JAK2^WT control groups. The relative proportions of GMP, CMP, and MEP were similar in the bone marrow and spleens of JAK2^V617F mice. Comparison with normal bone marrow suggests that JAK2^V617F expression does not seem to favor expansion of one particular progenitor cell population, as the relative expansion of GMP, CMP, and MEP is similar. This is in contrast to recent findings in PV patients with high white cell counts who show a significant increase of the CMP population in the peripheral blood ( 29).

The JAK2^V617F mutation causes expansion of late progenitors. As erythrocytosis is not commonly observed in murine MPD models, we determined Epo levels to exclude secondary erythrocytosis. JAK2^V617F mice had low Epo levels compared with JAK2^WT mice (95 pg/mL compared with 182 pg/mL; P = 0.017; Fig. 5D). To determine more precisely at which differentiation stage the expansion of the red cell compartment occurred, we assessed early erythroid progenitor (EEP) and late erythroid progenitor (LEP) by FACS. Gates were set to include Lin⁻/CD117⁺ cells, with EEP defined as CD71^low and LEP as CD71^high (Supplementary Fig. S4F; ref. 30). In contrast to vector controls, JAK2^V617F but also JAK2^WT mice showed a predominance of LEP over EEP in their bone marrow ( Fig. 5C, left). Similar results were seen in the spleens of JAK2^V617F mice. These data suggest that JAK2^V617F predominantly induces expansion of the LEP cell compartment.

We hypothesized that the preferential expansion of the erythrocyte compartment may be due to increased expression of the EpoR on the cells. Early erythroblast cells were gated as lineage-positive (Lin⁺), CD71^high, and Ter-119^med/high, and late erythroblast cells were gated as Lin⁺, CD71^low/negative, and Ter-119^high ( 30). For the late erythroblast population, EpoR was not different between the three groups of mice ( Fig. 5C, right). In contrast, we observed a reduction of EpoR expression on early erythroblasts of JAK2^V617F mice that reached borderline significance (P = 0.08).

We also analyzed the proportion of megakaryocyte progenitors, identified as Lin⁻/CD117⁺/CD9⁺/Fcγ receptor^low/CD41⁺ (Supplementary Fig. S4E; ref. 31). The bone marrow and spleens of JAK2^V617F mice showed a 3- and 15-fold relative increase of the megakaryocyte progenitor population compared with empty vector mice ( Fig. 5B, right). As seen earlier with LEP, the bone marrow of JAK2^WT mice also showed a relative increase of the megakaryocyte progenitor population compared with empty vector controls.

Increased levels of TNF-α and reduced levels of G-CSF in JAK2^V617F mice. Cells from patients with PV are hypersensitive to various cytokines in addition to Epo ( 15, 16). Serum levels of IL-2 and IL-6 were raised during progression of patients with ET and PV to myelofibrosis ( 32). In addition, it has been suggested recently that reduced responsiveness to death receptor ligands, such as TNF-related apoptosis-inducing ligand (TRAIL), may contribute to the proliferative advantage of PV erythroid progenitor cells over their normal counterparts ( 33). We hypothesized that in addition to decreased Epo serum levels other cytokines involved in regulation of hematopoiesis could be affected by JAK2^V617F expression. We found no differences in serum levels of IL-2, IL-6, and GM-CSF between JAK2^WT and JAK2^V617F mice (data not shown), whereas average serum levels of G-CSF were significantly lower (99 pg/mL compared with 343 pg/mL; P = 0.037) in JAK2^V617F mice compared with JAK2^WT mice ( Fig. 5D). In contrast, average serum levels of TNF-α were significantly higher in JAK2^V617F mice (15 pg/mL compared with 1 pg/mL; P < 0.001).

Long-term outcome of JAK2^V617F mice and secondary transplants. A cohort of six JAK2^V617F mice was followed beyond day 80 for up to 166 days. None of these mice showed clinical symptoms of disease. Peripheral blood counts were extremely variable (Supplementary Table S1). FACS analysis for GFP expression detected only low numbers of GFP-positive cells (4-7.7%) in three of six mice; two mice were negative for GFP and one mouse with a WBC of 114,000/μL had 50.6% of GFP-positive cells. These data are consistent with reestablishment of normal nonleukemic hematopoiesis in the peripheral blood of most of the mice.

We did secondary transplants into sublethally irradiated mice (Supplementary Table S1). Two of eight mice developed transient MPD characterized by high hematocrit (>58%) over 30 days, but subsequently hemoglobin levels normalized. Both mice had received marrow from the same donor mouse. All other mice have maintained normal peripheral blood counts up to 115 days after transplantation.

Discussion

The pathogenesis of MPD, with the exception of CML, had remained enigmatic for a long time. Clinical observations indicate that MPDs represent a spectrum of diseases. There are well-defined entities that can however overlap and may metamorphose from one into another. Recently, several groups reported that the majority of patients with PV and large proportions of patients with ET and IMF exhibit a point mutation of the JAK2 nonreceptor tyrosine kinase (JAK2^V617F; refs. 5– 8). The molecular basis of the heterogeneity of disease phenotypes seen in patients is unknown. Many constitutively active tyrosine kinases, such as BCR-ABL or FLT3 with internal tandem duplications, induce cytokine independence in factor-dependent hematopoietic cell lines. Thus, in a first experiment, we expressed JAK2^V617F in BaF/3 cells. Complete IL-3 withdrawal invariably led to cell death. However, over a range of IL-3 concentrations, JAK2^V617F-positive cells showed increased growth and viability compared with parental BaF/3 cells and cells expressing JAK2^WT. However, we did not observe complete cytokine independence, consistent with the lack of JAK2 phosphorylation in the absence of IL-3. This is in line with observations by Levine et al. ( 7) but in contrast to a study by James et al. ( 5) A possible explanation for the discrepancies is differences in the BaF/3 cell lines used in these experiments, which might for example differ in the level of IL-3 receptor expression. Lu et al. ( 34) have shown that, in BaF/3 cells expressing JAK2^V617F, high levels of a homodimeric type I cytokine receptor must be coexpressed to provide a scaffold for correct positioning of JAK2^V617F, optimal signal transduction, and cytokine independence. Apparently, the heterodimeric IL-3 cytokine receptor can fulfill this function only at high expression levels.

To investigate whether JAK2^V617F induces MPD in mice, we transplanted lethally irradiated BALB/c mice with bone marrow infected with JAK2^V617F, JAK2^WT, and empty vector. All mice transplanted with JAK^2V617F-infected marrow developed MPD, characterized by erythrocytosis, granulocytosis, skin plethora and, in 2 of 13 mice, transient mild thrombocytosis. Disease characteristics were very variable in individual mice. Thus, altogether, the murine disease closely resembles PV, but with considerable phenotypic variation, similar to human JAK2^V617F-positive MPD. The precise cause of the phenotypic variability in our model is unknown. One possible explanation is that the low infection rate of bone marrow cells leads to random effects (i.e., specific rare cell populations may not be targeted consistently). As mentioned, JAK2^V617F requires association with a homodimeric cytokine receptor to induce cytokine independence ( 34) and it is thus conceivable that in our model different target cells were infected, whose differential expression of critical cytokine receptors modulates the disease phenotype. Alternatively, additional mutations may have occurred. In contrast to BaF/3 cells transduced with JAK2^V617F, we were able to show JAK2 phosphorylation in serum and cytokine-starved peripheral blood leukocytes from a leukemic mouse. Cytokine stimulation increased phosphorylated JAK2 levels, whereas treatment with AG490, a JAK2 inhibitor, reduced phosphorylation. This is consistent with constitutive phosphorylation that can be further enhanced by cytokine stimulation.

A group of mice was sacrificed on day 80. Necropsy uniformly revealed splenomegaly and mild hepatomegaly, without macroscopic evidence for any other organ involvement. Light microscopy showed extramedullary hematopoiesis predominantly in the spleen and to a much lesser degree in the liver. A minor degree of splenic extramedullary hematopoiesis is physiologic ( 26), but interestingly, we noted a slight increase also in JAK2^WT mice. However, in contrast to JAK2^V617F mice, the splenic architecture remained intact and there were no atypical megakaryocytes. Bone marrow histology of JAK2^V617F mice showed a predominance of mature granulocytes and atypical megakaryocytes, with significant reticulin fibrosis but largely absent erythropoiesis. This morphology is reminiscent of the ‘spent phase’ of PV and consistent with this, median hematocrit levels had peaked around day 60 and a slight decline was evident at the time of the histology on day 80 ( Fig. 2A). At the same time, we observed a relative increase in Lin⁻/CD117⁺ bone marrow cells, which represent a stem and progenitor cell population. Further analysis of this population showed an expansion of CMP, GMP, and MEP that was largely proportionate compared with vector control mice. Surprisingly, findings were similar or even more pronounced in JAK2^WT mice, suggesting that enforced expression of WT JAK2 induces some degree of cytokine hypersensitivity. Kralovics et al. ( 8) also observed moderately increased JAK2 and STAT5 phosphorylation in BaF/3 cells expressing WT JAK2 compared with vector controls in response to IL-3. An alternative explanation is that the marrow findings at the time of the analysis reflect a relatively late stage of disease evolution and a more pronounced expansion would have been observed in JAK2^V617F mice at an earlier time point after transplantation. A very pronounced expansion of the stem/progenitor cell compartment was seen in the spleens of JAK2^V617F mice, which involved CMP, GMP, and MEP to a similar degree. Colony assays showed that these cells were not cytokine independent, but cytokine hypersensitive, in agreement with the observation that CD34⁺ progenitor cells from PV patients are not cytokine independent but cytokine hypersensitive ( 35). In contrast, Wernig et al. ( 18) and Lacout et al.( 19) reported cytokine-independent colony-forming unit-erythroid growth of spleen cells of JAK2^V617F. This discrepancy may be related to the fact that they analyzed cell clusters after 2 days in culture, whereas we analyzed overall colony formation after prolonged culture. We did not attempt to morphologically distinguish between colony types and it thus remains unclear whether JAK2^V617F induces a shift toward erythroid differentiation compared with normal progenitor cells, as was shown recently for human PV ( 29).

We observed a significant increase of megakaryocyte progenitors in JAK2^V617F mice, yet there was no or only mild and transient thrombocytosis. The reasons for this discrepancy are unknown, but it is obvious that additional factors must be modulating the disease phenotype, similar to patients with JAK2^V617F. For example, it has been suggested that the JAK2^V617F/JAK2^WT ratio may be critical ( 19). Thus, hematopoietic cells with a low JAK2^V617F/JAK2^WT ratio might show activation of different pathways compared with cells with a high JAK2^V617F/JAK2^WT ratio and predominant EpoR-like signaling. Additional mutations may also be involved.

FACS analysis further showed that the splenic expansion of erythropoiesis predominantly involved LEP cells. Terminal erythrocyte differentiation in PV patients occurs independently of Epo, and consistent with this, the LEP compartment may have expanded in spite of the low Epo levels in our mice ( 35). In this context, it is interesting that EpoR expression was relatively low on early erythroblasts of JAK2^V617F mice. This observation suggests that JAK2^V617F may affect the coordinated expression of EpoR during erythropoiesis. If this is relevant to human, PV remains to be established.

Several studies have shown that hematopoietic cells from PV patients are hypersensitive to various cytokines and serum levels of cytokines involved in regulation of hematopoiesis are altered ( 14, 15, 31). We therefore investigated serum levels of a panel of cytokines in addition to Epo in JAK2^V617F and JAK2^WT mice. We found lower serum levels of G-CSF in JAK2^V617F compared with JAK2^WT mice, suggesting negative counter-regulation. However, other cytokines, such as GM-CSF, IL-2, and IL-6, which have been reported to be altered in PV patients, were unchanged. Unexpectedly, we found that levels of TNF-α in JAK2^V617F mice were 3.5-fold higher than in the JAK2^WT mice. TNF-α plays an important regulatory role in hematopoiesis ( 36, 37). Epo has been shown to induce TNF-α in hematopoietic cells ( 37), which in turn regulates erythroid homeostasis by inhibiting the early stage of erythroid progenitors ( 36). On the other hand, erythroid progenitor cells isolated from PV patients are insensitive to TRAIL and FAS ligand, two other members of the TNF family ( 33), suggesting that the high TNF-α levels in the JAK2^V617F mice may suppress normal eythropoiesis while sparing JAK2^V617F-expressing cells. High TNF-α levels could also be involved in the bone marrow fibrosis of JAK2^V617F mutant mice, as it has been shown to stimulate fibroblasts in vitro ( 38). Thus, it is conceivable that JAK2^V617F exerts complex effects on hematopoiesis, including inhibitory effects on normal cells.

As thromboembolic events and hemorrhage are the major cause of morbidity in patients with PV and ET, we carefully examined sections of brains and hearts for evidence of infarction or hemorrhage but failed to identify histologic evidence for such events, in accord with the benign clinical phenotype of JAK2^V617F-positive MPD in BALB/c mice. The fact that the mice do not suffer from these complications of PV despite the rather tight correlation between hematocrit and thromboembolic risk in humans points to the involvement of additional human-specific factors, such as platelet and neutrophil activation, and endothelial damage ( 39, 40). Notably, Wernig et al. observed significant morbidity in their BALB/c mice transplanted with JAK2^V617F-infected marrow, although no details are given. Nonetheless it is apparent that no straightforward correlation exists between peripheral blood variables and clinical symptoms in the BALB/c model of JAK2^V617F-positive MPD.

PV is thought to originate in a hematopoietic stem cell but the disease in our model is not transplantable. Only two of the secondary transplanted mice with JAK2^V617F developed a transient erythrocytosis lasting for ∼30 days. Observation of the primary recipient mice for 166 days showed no or few GFP-positive peripheral blood leukocytes, with the exception of one mouse with persistently high white cell counts. Southern blot analysis on day 80 identified only a single clone in two of four mice and only two clones in the remainder. Thus, in contrast to other murine MPD models ( 41), the number of clones was very limited, probably reflecting the use of relatively low titer retrovirus for infection. Thus, the likelihood of infecting a long-term stem cell will be relatively low, which would be consistent with the transient nature of the MPD in most animals. Transient MPD may thus arise by expression of JAK2^V617F in a short-term stem cell or progenitor cells capable of only limited expansion. This implies that JAK2^V617F, similar to BCR-ABL, does not confer self-renewal capacity to hematopoietic cells and explains why the disease tends to 'burn out' in most mice ∼20 weeks after transplantation ( 42). Similar to our findings, a recent study showed that JAK2^V617F induced a transient two-stage PV-like disorder in C57BL/6J mice with abatement of the PV phenotype ∼4 months after transplantation ( 19). However, in this study, the mice progressed to end-stage myelofibrosis, with typical peripheral blood red cell morphology and anemia, although we have not observed this type of disease evolution.

In summary, in BALB/c mice, JAK2^V617F induces a PV-like MPD. We show that the expansion of the erythroid, granulocyte-monocyte, and megakaryocyte lineages at the stem cell and progenitor stage is proportionate, consistent with a trilineage MPD. Our data further suggest that the effects of JAK2^V617F on hematopoiesis may be complex, including effects on EpoR expression and regulatory cytokines, such as TNF-α. Last, our data suggest that JAK2^V617F does not confer self-renewal capacity. This implies that the establishment of a nontransient disease requires transformation of a hematopoietic stem cell. These findings will be important to direct investigations into the pathogenesis of the human disease.