JAK2^V617F expression in mice amplifies early hematopoietic cells and gives them a competitive advantage that is hampered by IFNα

The acquired gain-of-function V617F mutation in the Janus Kinase 2 (JAK2^V617F) is the major mutation involved in BCR/ABL-negative classical myeloproliferative neoplasms (MPNs).^1-4  It is present in 95% of polycythemia vera (PV) patients and ∼50% to 60% of patients suffering from essential thrombocythemia (ET) or primary myelofibrosis. Retrovirally induced^5-7  and transgenic (TG) animals⁸  have demonstrated that the sole JAK2^V617F mutation is able to generate these diseases and that the level of expression of JAK2^V617F is crucial in determining MPN diversity.⁸  Several knock-in (KI) mouse models using the murine gene have been described and all demonstrate that endogenous heterozygous expression of mJAK2^V617F leads to a PV-like disease, usually followed by secondary myelofibrosis,^9-11  whereas homozygosity increases the severity of the PV-like phenotype.⁹  In contrast, a mild ET-like disease was reported in a KI model using hJAK2^V617F.¹² 

Besides JAK2^V617F, several other acquired mutations have been identified in BCR/ABL-negative MPN, although with a frequency often <10%. These mutations target several pathways/complexes: JAK/STAT signaling directly responsible for MPN phenotype, DNA methylation/demethylation, and PRC2 and RNA splicing.¹³  Functional analyses of several of these additional mutations have questioned the precise contribution of JAK2^V617F to MPN disease initiation, as deletion of Tet2^14,15  or Dnmt3a¹⁶  in mice cause hematopoietic stem cell (HSC) amplification and increased self-renewal. On the contrary, the contribution of JAK2^V617F to HSC selective amplification has become a matter of controversy.¹⁷  Accordingly, analysis of human PV patients has suggested that JAK2^V617F amplified terminal stages of hematopoietic cell differentiation without amplifying the early stages, as opposed to Tet2 mutations.^18,19  Therefore, the ability of JAK2^V617F-positive HSCs to compete with normal HSCs is largely questioned, and actual data suggest that other mutations, occurring before or after JAK2^V617F, are instead responsible for clonal emergence of JAK2^V617F-positive MPN.

To reevaluate the role of JAK2^V617F in HSC amplification, we developed a new conditional JAK2^V617F KI model and analyzed early stages of differentiation. We found that endogenous expression of JAK2^V617F greatly amplified multipotent progenitor cells and gave a proliferative advantage in a competitive setting. We also showed that interferon-α (IFNα), a drug promoting cycling of dormant cells^20,21  and clinically effective in PV patients,²²  prevented disease and JAK2^V617F-positive disease-initiating cell development. Overall, these results show that JAK2^V617F provides a strong competitive advantage to HSCs that may account by itself for MPN emergence and that is lost upon IFNα treatment.

Animal experiments were approved by the Institut Gustave Roussy review board, protocol no. 2012-061. The method to generate and genotype the conditional flexed JAK2 (JAK2^{FLEX/+ or L2/+}) KI mice, backcrossed >15 times with C57Bl/6 mice, is described in supplemental Data and Figure 1 (available on the Blood Web site). To express the mutation, KI mice were crossed with CmvCre TG mice to validate the KI construct and with VavCre TG mice^23,24  for the rest of the study. We termed these heterozygous recombined JAK2^V617F/WT mice (JAK2^VF/+), with a wild-type (WT) and a mutated Jak2 allele, as JAK2^V617F KI mice.

For bone marrow (BM) transplantation, 3 × 10⁶ C57/Bl6/6J BM cells were engrafted into lethally irradiated (9.5 Gy) congenic recipient mice. For competitive BM transplantation, lethally irradiated Ly5.1 WT recipient mice were injected with BM cells from Ly5.1+2 WT donor mice mixed with BM cells from Ly5.2 VavCre/JAK2^VF/+ mice. Murine IFNα (Miltenyi Biotech; 10⁷ UI/mL) was subcutaneously injected every day at a dose of 3 × 10⁴ IU in 0.2 mL phosphate-buffered saline/mouse (1.5 × 10⁶ IU/kg). Control vehicle mice were injected with phosphate-buffered saline.

Hemoglobin, mean corpuscular volume, hematocrit, red blood cell (RBC), platelet, and white blood cell (WBC) counts were determined using an automated counter (MS9; Schloessing Melet, France) on blood collected from the retro-orbital plexus in citrated tubes. BM cells were removed by flushing both femurs. Spleens were weighed and single cell suspensions were prepared. For histopathology analysis, spleens were fixed in formaldehyde and sections (4.5 µm) were stained with hematoxylin/eosin, periodic acid Schiff, and Giemsa for cytology analysis. Reticulin fibers were revealed by silver staining according to the Gordon Sweet method. Flow cytometry of blood, BM, and spleen was used as described in supplemental Data.

Progenitor cell assays were carried out in 1 mL methylcellulose “MethoCult 32/34” (Stemcell Technologies) without stimulus or maximally stimulated by 10 ng/mL IL3, IL6, TPO, 100 ng/mL SCF, and 2 IU/mL EPO. Cultures in duplicate or quadruplicate were scored after 2 days for colony forming unit-erythroid (CFU-E) assays and 8 days for burst forming unit-erythroid (BFU-E) and colony forming unit granulocyte-macrophage (CFU-GM) assays. Total progenitor cell number was calculated assuming that one femur represents 6% of the total marrow and from the total number of cells isolated from the spleen.

BM or spleen cells were stained for LSK and SLAM isolation. For apoptosis studies, cells were stained with Annexin V-PB antibody (Biolegend). For cell cycle status, cells were fixed, permeabilized, and then labeled with Ki-67-FITC (Becton Dickinson) and Hoechst 33342 (Invitrogen). Flow cytometry was performed using an LSR II analyzer (Becton Dickinson).

Results are presented as mean ± SEM and data were analyzed with the 2-tailed Student t test.

We generated conditional JAK2^V617F KI mice using the FLEX switch strategy²⁵  (Figure 1; supplemental Data). We crossed the JAK2^FLEX/+ KI mice with TG mice expressing the Cre recombinase under the control of the Vav promoter.^26,27 VavCre TG mice targets in vivo Cre recombination in HSCs²⁷  and some endothelial cells.²⁴  Allele specific reverse-transcription polymerase chain reaction in BM revealed an equal amount of JAK2^WT and JAK2^V617F mRNA, demonstrating that all BM cells were recombined with a normal transcriptional activity of the mutated locus (supplemental Figure 1). These conditional VavCre/JAK2^VF/+ animals were born with severe thrombocytosis and leukocytosis (Figure 2). Surprisingly, the hemoglobin and hematocrit levels remained normal during the first month of life. RBC number was slightly in excess in 2-week-old mice, but severe microcytosis precluded any effect on hematocrit (Figure 2). At 2 to 3 months of age, hematocrit values were elevated and reached a plateau for 5 months. Reticulocyte counts were elevated (17 ± 2%, controls 2.8 ± 0.4%, n = 13). Platelet numbers rapidly increased and reached a plateau for 6 months. In aged mice, WBCs reached high values, but platelet counts along with hemoglobin levels decreased to subnormal levels, as observed during myelofibrosis development in other models (Figure 2).^5,28 

Hematopoietic tissues were abnormal. BM cellularity was reduced (60% of control) (Figure 3A) with an increase in the percentage of granulocytes (from 41 ± 2% in WT littermates to 70 ± 3%; P = .02) and a decrease in both mature erythroblasts (from 30 ± 4% in WT to 8 ± 1%; P = .03), as previously reported in JAK2^V617F mouse models,^5,10  and B cells (from 20 ± 2% in WT to 10 ± 3%). Splenomegaly was mild at 6 weeks (3-fold) but severe at 12 weeks (8-fold) (Figure 3A) and was due to an increase in erythroblasts (from 7 ± 5% in WT to 14 ± 2%) and myeloid cells (from 2 ± 1% in WT to 18 ± 4%), with a decreased percentage of B cells (from 51 ± 3% in WT to 27 ± 4%). In aged mice (>6 months), we observed high-grade fibrosis in BM and spleen and extra-medullary hematopoiesis in liver (supplemental Figure 2).

These data show that heterozygous endogenous JAK2^V617F expression in hematopoietic cells leads to hyperplasia of mature and maturing erythroid, granulocytic, and megakaryocytic cells in blood and hematopoietic tissues that models human PV with secondary myelofibrosis.

We next studied the effect of JAK2^V617F on early stages of differentiation. We first analyzed progenitor cell numbers from VavCre/JAK2^VF/+ animals using clonogenic assays. In the marrow, the frequencies of all progenitor cells except CFU-E increased (supplemental Figure 3). Taking into account the decrease in BM cellularity (Figure 3A), the total numbers of myeloid progenitors (CFU-GM) and late erythroid progenitors (CFU-E) did not differ from littermates, but early erythroid progenitor cells (BFU-E) were 2-fold increased (Figure 3B). In the spleen, frequencies of all progenitor cells were increased (supplemental Figure 3) and taking into account splenomegaly, the myeloid progenitor cell compartment was largely amplified, particularly CFU-E (124-fold) (Figure 3B), with 56% of them being Epo-independent. Finally, cumulative numbers of CFU-GM, BFU-E, and CFU-E in marrow plus spleen were increased by 1.2-, 3-, and 15-fold, respectively (Figure 3B).

We then analyzed by flow cytometry Lin-, LSK, and LSK/CD48-/CD150+ cells, so-called SLAM cells, which are highly enriched (1 of 2.2)²⁹  in long-term reconstituting activity (LT-HSC) (supplemental Figure 4). In 3-month-old mice, JAK2^V617F KI animals demonstrated in marrow no change in the frequency of Lin- cells but an increase in LSK (×2.3) and SLAM (×2) cells (Figure 3C). In spleen, the percentages of Lin-, LSK, and SLAM cells increased 1.5-, 1.7-, and 3.7-fold, respectively. We noticed a progressive increase with age. A normal frequency was observed in 1-month-old mice, but the percentages of BM and spleen LSK and SLAM cells were increased in 6-month-old mice more than in 3-month-old mice. Overall, we observed a 3-, 4-, and 5-fold increase in the total numbers (BM plus spleen) of Lin- cells in 1-, 3-, and 6-month-old mice, respectively. In contrast, there was no increase in the total numbers of LSK and SLAM cells in 1-month-old mice but a 3- or 7-fold and a 4- or 6-fold increase in the total numbers of LSK and SLAM cells in 3- or 6-month-old mice, respectively (Figure 3D). One possible pitfall could be the upregulation of Sca-I by JAK2^V617F, but SLAM cells, analyzed regardless of Sca-I, were similarly increased. Related to c-Kit+/Lin- progenitor cells (LK), the total numbers of MEP, GMP, and CMP augmented in 3- and 6-month-old mice, mainly due to the spleen content with marked and early increase in MEP (supplemental Figures 4 and 5). Overall, the differences between KI and WT mice in the pools of LSK, GMP, CMP, and particularly SLAM cells increased with the age, whereas these differences in MEP and Lin- pools seemed precociously established. These results show that endogenous JAK2^V617F progressively amplifies the earliest stages of differentiation with the age while late stages are already amplified in young mice.

We next investigated if JAK2^V617F expression confers a proliferative advantage to LT-HSC using competitive grafts in CD45.1 WT lethally irradiated recipient animals transplanted with 30% CD45.2 KI and 70% CD45.1+2 WT BM cells collected from 3-month-old donor mice. As a WT control, we used 30% CD45.2 WT and 70% CD45.1+2 WT grafts and as a control for full disease engraftment, we used 100% CD45.2 KI grafts.

The disease was transmitted to all recipients grafted with 30% KI graft (supplemental Figure 6). Blood KI myeloid cells (CD45.2/CD11b/Gr-1) increased from ∼40% to ∼90% (similarly to mice transplanted with 100% KI cells) from 4 to 34 weeks posttransplantation (Figure 4A). In contrast, the percentages of blood KI lymphoid cells (CD45.2/B220/CD3) only reached ∼50% at 34 weeks posttransplant (Figure 4A-B). Mice were killed 34 weeks posttransplant to analyze chimerism in Lin-, LSK, and SLAM cells. Strikingly, the percentages of BM Lin-, LSK, and SLAM cells from KI origin in mice grafted with 30% KI cells reached similarly high percentages (95 ± 1%, 96 ± 2%, and 99 ± 1%, respectively) as those (99 ± 1%, 98 ± 2%, and 99 ± 1%, respectively) found in mice transplanted with 100% KI BM cells (Figure 4B). The percentages of spleen KI Lin-, LSK, and SLAM cells (65 ± 2%, 76 ± 1%, and 90 ± 8%, respectively) were slightly lower (Figure 4B). The extent of splenomegaly and the reduction in marrow cellularity were similar in mice transplanted with 30% and 100% KI cells, showing that disease severity was comparable in both groups (Figure 4B).

These results show that endogenous JAK2^V617F expression confers a selective advantage to LT-HSC and myeloid cell progenies.

Our 30% KI graft initiated from 3-month-old mice should correspond to a 46% KI graft at the stem cell level due to the previously shown 2-fold BM SLAM amplification. Therefore, to know the minimum graft necessary for disease emergence, we developed competitive grafts from young donor mice, including different KI/WT BM cell percentages corresponding to similar KI/WT SLAM cell percentages and known KI SLAM numbers (Figure 5A). Results showed that 30% KI BM containing ≥50 KI SLAM cells always generated a long-term disease (29 weeks posttransplant) with polycythemia, leukocytosis, thrombocytosis, splenomegaly, and low BM cellularity, although long-term thrombocytosis was not always observed (Figure 5A-B). In contrast, only 3 of 5 recipients grafted with 10% KI BM or 17 KI SLAM cells developed the disease. Experiments including 10 and 15 mice confirmed that 10% KI BM grafts generated only long-term disease (15 months) in ∼50% recipients (5/10 and 7/15). These results suggest that in competition assays, ∼30 JAK2^V617F SLAM cells are sufficient to generate MPN with complete penetrance. Interestingly, the percentages of blood lymphocytes from KI origin 30 weeks posttransplant were similar or lower to the percentages of KI SLAM cells initially transplanted, suggesting no or very delayed KI lymphoid cell amplification (Figure 5C). In contrast, as previously shown, the percentages of KI blood myeloid cells, BM/spleen Lin-, LSK, and SLAM cells were higher than the initial KI SLAM cell input, showing overgrowth of these cells in recipients (Figure 5B-C).

These results show that endogenous JAK2^V617F expression generates MPN in mice from a low number of HSCs and confers no selective advantage or bias to lymphoid cells.

Previous results in KI mice have shown that a JAK2 inhibitor could not eradicate disease-initiating cells.¹¹  Therefore, we investigated if another drug, IFNα, could hamper this early cell reservoir for MPN propagation. IFNα seemed a good candidate, because it induces HSC exhaustion^20,21  and molecular response in patients suffering from PV through still unclear mechanisms.^22,30 

IFNα treatment was initiated in 2 groups of CD45.1 WT recipient mice 7 weeks after the transplantation of 30% KI BM cells collected from 1- or 3-month-old donor mice in order to generate low or high disease burden, as previously shown. In both groups of mice, IFNα treatment induced a rapid suppression of leukocytosis without leukopenia (Figure 6A). All cell types dropped, but mainly B cells and the percentage of T cells, particularly CD3+/LY6C+, and myeloid cells increased (supplemental Figure 7). Platelet levels also dropped quickly after IFNα treatment but reached a plateau lower than control (Figure 6A). Hematocrit, hemoglobin, and RBC counts normalized early or late after treatment according to low or high disease severity (Figure 6A). Microcytosis was corrected after polycythemia (supplemental Figure 8), suggesting correction of iron deficiency.¹⁰  The percentage of myeloid cells from KI origin (CD45.2/CD11b/Gr-1) in vehicle-treated mice increased with time, highlighting disease progression. IFNα cancelled out this progression, and myeloid chimerism at the end of treatment was reduced to ∼0% or 27 ± 10% vs 53 ± 3% or 85 ± 6% in vehicle-treated mice in the first or second group of mice, respectively (Figure 6A). Chimerism was also strongly reduced in BM lymphoid, myeloid progenitor, and SLAM cells but not in blood and spleen lymphocytes (Figure 6B; supplemental Figure 9). Splenomegaly was slightly reduced in the severe disease group (323 ± 4 mg vs 485 ± 20 mg). In the low severity group, splenomegaly was strongly reduced (176 ± 3 mg vs 580 ± 27 mg; P = .001) with normalization of its architecture. BM cellularity was increased. Myeloid and MKC hyperplasia were suppressed in BM and spleen (supplemental Figure 10).

These results show that IFNα hampers JAK2^V617F cell proliferation with normalization of most blood hematological values, except thrombocytopenia, according to disease severity.

To check if IFNα targeted the disease-initiating stem cells, we carried out secondary transplantation from 3 × 10⁶ BM cells collected from IFNα- or vehicle-treated mice. As expected, mice transplanted with the vehicle-treated BM cells developed MPN with high hematocrit and WBC values (Figure 6C; supplemental Figure 11). In contrast, mice transplanted with the IFNα-treated BM cells developed normal hematocrit, platelet, and WBC values. We finally checked whether JAK2^V617F-positive cells were still present in transplanted mice. From the IFNα-treated mice displaying the low severity disease, no cell from KI origin (CD45.2) was detected in the secondary recipients in contrast to vehicle-treated mice (supplemental Figure 11). Because we used CD45.2 recipients for IFNa-treated mice displaying the severe disease, we developed an allele-specific PCR assay to determine JAK2^V617F allele burden in WBC. JAK2^V617F was detected in WBC up to 7 weeks posttransplant and thereafter became undetectable (Figure 6C). In contrast, the level of JAK2^V617F-WBC remained stable in mice transplanted with vehicle-treated BM cells from week 7 to 20 of survey.

These results show that long-term treatment by IFNα is able to alleviate JAK2^V617F-derived MPN by suppressing long-term disease-initiating cells.

IFNα is known to induce antiproliferative and proapoptotic activities. To know if the effect of IFNα on JAK2^V617F disease-initiating cells was mediated through cell cycle or apoptosis, we measured these parameters in LSK/SLAM cells from KI and WT littermate mice treated 3 days with IFNα (3 × 10⁴IU) or vehicle. First, the results showed that LSK/SLAM cells from KI mice were more proliferative and less apoptotic than those from WT mice (Figure 7A-B; supplemental Figures 12 and 13). These properties may explain their amplification and proliferative advantage. Second, IFNα strongly induced the proliferation of WT LSK/SLAM, which, as previously reported,²⁰  had an effect on KI LSK cell cycle but none on KI SLAM cell cycle. Overall, IFNα treatment induced WT SLAM cells to become as proliferative as KI SLAM cells. We noticed that IFNα increased Sca-I expression, as reported for IFNγ,³¹  and could bias our results, but we obtained similar results without including Sca-1 (that is dispensable) in SLAM cell characterization. Third, LSK and SLAM cells were clearly less apoptotic in KI than in WT mice. However, IFNα markedly induced apoptosis of spleen, but not BM, KI LK, and LSK/SLAM cells, and KI SLAM became as apoptotic as WT SLAM (Figure 7B; supplemental Figure 13).

These results suggest that IFNα eradicates JAK2^V617F disease-initiating cells in long-term treatment by increasing WT HSC cell cycle and KI HSC apoptosis, suppressing overall the proliferative advantage of KI over WT HSC.

In this study, we show that endogenous expression of JAK2^V617F in mouse HSCs induces a progressive amplification of early hematopoietic cells and provides to HSCs a strong competitive advantage over normal ones that could be explained by their high proliferative and low apoptotic status. Furthermore, long-term IFNα treatment suppresses disease progression by targeting JAK2^V617F-expressing disease-initiating cells. Overall, this study suggests that JAK2^V617F is capable of driving MPN emergence by itself from a low number of HSCs and that IFNα hampers this ability.

We previously described a constitutive JAK2^V617F KI.¹⁰  In this report, a conditional KI with restricted expression in hematopoietic and some endothelial cells²⁴  also generated a 2-stage MPN-mimicking human PV evolving into secondary myelofibrosis but with lower/more delayed splenomegaly/leukocytosis and polycythemia than our constitutive KI. This may be due to the late JAK2^V617F gene expression controlled by the Vav promoter during fetal but not embryo life.³²  All JAK2^V617F Retroviral,^5-7  TG,⁸  and KI^9-12  mouse models have described the occurrence of MPN, emphasizing the responsibility of the mutation in this disorder. However, the phenotype of these models often differs. Studies have shown that the mutated/WT JAK2 ratio is crucial in determining disease diversity through various levels of RBC and platelet counts.^5-8  However, it is not yet well understood why heterozygous expression of mouse JAK2^V617F in all KI models induces a PV-like disease, one without thrombocytosis,¹¹  whereas heterozygosity is usually associated with ET in humans. An ET-like disease is observed only in the heterozygous KI using the human gene.¹² 

The effect of JAK2^V617F on the amplification of late mature and maturing myeloid cells is well established in mice^5-11  as in humans.^18,33  In contrast, the effect of JAK2^V617F on the amplification of early hematopoietic cells remains a matter of controversy.¹⁷  Our study shows that JAK2^V617F increases in 6-month-old mice the percentage of LSK/SLAM cells by 5-/2-fold and 6-fold in marrow and spleen, respectively. These phenotypically defined cell pools are the most HSC-enriched compartments.²⁹  Taking into account the BM and spleen cellularity, we measured a 7- or 5-fold expansion in the total numbers of LSK or SLAM cells in these mice, not including the contribution of the extra-medullary hematopoiesis evidenced in these animals. Interestingly, the pool of HSCs is quasi-normal at 1 month and expands only with age, while more mature cells amplify precociously. Our study also demonstrates the ability of JAK2^V617F-expressing HSCs to compete with normal HSCs in transplantation settings with JAK2^V617F-expressing blood granulocytes and tissue Lin-, LSK, and SLAM cells rapidly prevailing upon normal ones. Our results are compatible with 2 other KI models reporting an increase in BM LSKCD34-Flk2- cells³⁴  and a competitive advantage of JAK2^V617F-expressing HSCs recorded after 1 year of competitive transplantation.³⁵  In contrast, others reported a functional impairment of HSCs by JAK2^V617F,¹²  which could be related to the use of a human JAK2^V617F gene. Overall, murine models indicate that JAK2^V617F provides a clonal advantage to HSCs. Our study further shows that ∼30 JAK2^V617F SLAM cells are sufficient to generate MPN with complete penetrance, which further emphasizes the sole role of JAK2^V617F in this model and is compatible with a mathematic model suggesting that clonal dominance would need more than a single transformed cell to appear in a mouse lifetime.³⁶  We also demonstrated that KI SLAM cells are less apoptotic and more proliferative than normal SLAM cells. This result may explain the proliferative advantage and amplification of KI over WT HSCs. The use of JAK2 by cytokines involved in HSC regulation, such as TPO,^37,38  granulocyte colony stimulating factor,³⁹  or IFNγ,³¹  may explain this result. This study shows no or very delayed lymphoid cell amplification by JAK2^V617F, which emphasizes the mild role of JAK2 in lymphocyte signaling or suggests that amplified SLAM cells are myeloid biased.⁴⁰  Intriguingly, amplification of KI SLAM cells in vehicle-treated primary recipients did not result in the same amplification of KI donor cells in secondary hosts. This result suggests that the amplification of SLAM cells by JAK2^V617F does not correlate with HSC activity. Similar results were reported by Mullally et al,⁴¹  indicating that further experiments are needed to precisely determine the HSC properties of SLAM cells amplified by the mutation, with the caveat that high proliferative status may compromise their long-term self-renewal capacities.

In ET and PV patients, most studies agreed on the fact that the percentage of JAK2^V617F-positive cells is lower in progenitor cells than in differentiated stages. In contrast, an amplification of JAK2^V617F-positive cells was identified in progenitor and HSC of PMF or heterozygous PV patients,^18,33,42  that may present more evolved forms of disease. Therefore, as shown by this study, JAK2^V617F might give only a subtle advantage to HSCs that can be detected after several months in mice but only after several years in humans. As previously predicted,^18,19  amplification of heterozygous JAK2^V617F HSCs could also explain why heterozygosity leads to PV in mice but only to ET in humans unless additional molecular events, such as TET2 mutations, accelerate the upstream amplification of heterozygous cells. This hypothesis is corroborated by the association between an ET-like phenotype and the lack of early amplification in our young KI mice or KI mice using human JAK2^V617F.¹²  In humans, the presence of other mutations affecting TET2,¹⁹  DNMT3a,¹⁶  or ASXL1 might not only give a further opportunity to the clone to develop the disease but also modify the phenotype or accelerate disease evolution. However, the possibility remains that human JAK2^V617F is less functional, at least in a mouse context, than its mouse homolog and that these mutations become mandatory for disease emergence itself.

JAK2 inhibitors are currently tested in MPN patients and ruxolitinib was approved for use in the treatment of myelofibrosis.⁴³  These drugs have essentially demonstrated palliative virtues with limited effects on the selective reduction of JAK2^V617F-positive cells in human as in mice.^11,44  In contrast, IFNα treatment has demonstrated hematological and molecular responses, including complete molecular responses in PV patients.²²  Therefore, we tested the efficacy of IFNα in competitive transplantation models that included normal hematopoiesis and 2 different levels of JAK2^V617F disease burden. IFNα rapidly reduced platelet and WBC counts and corrected erythrocytosis and splenomegaly according to disease severity. This is in agreement with previous studies in normal mice showing that IFNα impaired lymphopoiesis⁴⁵  as well as myelopoiesis,⁴⁶  inducing anemia, leukopenia, and thrombocytopenia.

Prolonged IFNα exposures (11-13 weeks) were used, because it seems an important factor for complete molecular responses in humans.²²  Development of thrombocytopenia was observed. It has been attributed in humans to inhibition of platelet production by megakaryocytes.⁴⁷  Most interestingly, the treatment targeted granulocytes but also SLAM JAK2^V617F cells that were considerably reduced at the end of treatment. Secondary transplantations using treated marrow showed that disease developed normally from vehicle-treated cells but not from IFNα-treated cells. This result indicated that IFNα treatment suppressed JAK2^V617F HSC amplification and proliferative advantage. This finding may give the opportunity to understand the mechanism of action of this drug that is largely unknown. Our results show that IFNα promoted cycling of normal stem cells, as previously reported,^20,21  but did not further increased proliferation of JAK2^V617F-SLAM cells. The reason could be that JAK2^V617F already maximally stimulates the JAK/STAT signaling pathway that the IFNα receptor uses to promote HSC proliferation. However, cycling induced through the IFNα receptor and through other receptors^31,37-39  using JAK2^V617F may have distinct/antagonist effects³¹  on HSC maintenance (self-renewal or differentiation). It is also conceivable that increased signaling may be deleterious for HSCs, leading to loss of function, apoptosis, and finally exhaustion. Indeed, IFNα also increased apoptosis of SLAM cells in the spleen, and this effect may be major in reducing splenomegaly and suppressing neoplastic proliferation in a location that mostly contributed to its amplification. Among other mechanisms possibly involved, IFNα may promote an antitumor immune response through the stimulation of central memory T cells and activated dendritic cells.^48,49  In humans, IFNα was shown to eradicate JAK2^V617F cells without affecting TET2-mutated cells,⁵⁰  suggesting a mechanism of action that is related to JAK2 signaling. Altogether, our results are in part similar to those recently obtained by Mullally et al⁴¹  in KI mice by showing that IFNα targets JAK2^V617F HSCs, but our results differ in that the mechanism of IFNα is not yet clearly related to an entry of JAK2^V617F HSCs in cell cycle and their subsequent exhaustion.

In conclusion, our study demonstrates that JAK2^V617F is self-sufficient for HSCs to drive MPN in mice. Additional mutations may not be necessary to promote clonal emergence of MPN. The fact that IFNα alleviates the disease in mice offers the opportunity to understand the mechanism of action of this drug.

The authors thank Sylvie Jacquot and Marie-Christine Birling from the Mouse Clinical Institute (ICS-MCI), Illkirch, France, for establishing the conditional KI mouse line; Dr Warren Alexander from the Walter Elisa Hall Institute, Melbourne, Australia, for the VavCre mice; the staff of the animal facilities of the IGR directed by Patrick Gonin, Philippe Rameau, and Yann Lecluse for cell sorting experiments; and Olivia Bawa for the histology and Françoise Wendling for critically reviewing the manuscript.

This work was supported by grants from Inserm, IGR, Institut National Contre le Cancer, and Cancéropôle Ile-de-France and by funding from La Ligue Nationale Contre le Cancer (labeled team 2009). S.H. had fellowships from the Ministère de la Recherche et de la Technologie and la Ligue Nationale Contre le Cancer. C.M. is a Fondation de France recipient.

Contribution: S.H., C.L., M.C., and C.M. performed research and analyzed data; J.-L.V. designed experiments, performed research, analyzed data, and wrote the paper; and W.V. and E.S. designed experiments.

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

Correspondence: Jean-Luc Villeval, Inserm U1009, 114 rue Edouard Vaillant, Institut Gustave Roussy, PR1, 94805 Villejuif Cedex, France; e-mail: villeval@igr.fr.