Description of a knock-in mouse model of JAK2V617F MPN emerging from a minority of mutated hematopoietic stem cells

Myeloproliferative neoplasms (MPN) associated with JAK2V617F mutation have been modeled through different means, among others, knock-in (KI) models in which JAK2V617F is expressed under the control of its own endogenous promoter. This can be achieved using the Cre-Lox technology based on a tissue-specific expression of the Cre-recombinase. However, in previous KI models (Tie2-Cre, Vav-Cre, E2A-Cre),^1-3  the JAK2^V617F mutation has been introduced in nearly all hematopoietic stem cells (HSC), in contrast to essential thrombocythemia (ET) and polycythemia vera (PV) patients, who only carry a minority of mutated HSC.⁴  Because JAK2V617F in HSC has a low fitness to give rise to a disease,^5-8  assessing the mechanisms ruling JAK2V617F clonal dominance is crucial but requires a mouse model that mimics the HSC compartment of ET and PV patients. Here, we observed that the PF4iCre system allows expression of JAK2V617F in a minority of HSC leading to a model in which a small subset of JAK2V617F HSC induces a full-blown MPN phenotype.

PF4iCre;JAK2^V617F/WT mice were generated by crossing JAK2V617F KI mice⁹  with PF4iCre transgenic mice.¹⁰ PF4iCre; mTmG;JAK2^V617F/WT mice were then generated by crossing PF4iCre;JAK2^V617F/WT with ROSA^mTmG mice.¹¹ PdgfbiCre-ERT2;JAK2^V617F/WT were previously described.¹²  This study was conducted in accordance with both Bordeaux University Institutional Committee guidelines (committee CEEA50) and those in force in the European Community for experimental animal use (L358-86/609/EEC).

All methods are detailed as outlined in the supplemental Methods, available on the Blood Web site.

Because different groups reported recombination in a few subsets of HSC in the PF4iCre mouse model,^10,13,14  we developed a conditional JAK2V617F-driven MPN mouse model by crossing conditional heterozygous JAK2^V617F/WT KI mice⁹  with PF4iCre mice. As expected, we observed an early thrombocytosis in PF4iCre;JAK2^V617F/WT mice mimicking an ET (Figure 1A). A few weeks after, the hematocrit and the leukocyte counts increased, indicative of an evolution toward a PV-like disease (Figure 1A). PF4iCre;JAK2^V617F/WT mice developed a splenomegaly (Figure 1B) with red pulp hyperplasia compatible with a PV phenotype (Figure 1C). There was no evidence of fibrosis, consistent with the absence of cytopenia. The clonal origin of this MPN phenotype was demonstrated by the expression of JAK2V617F in all myeloid lineages, with a variant allele frequency of ∼10% in erythroid cells, 9% in granulocytes, and 50% in megakaryocytes (MK) (Figure 1D-E). This corresponds to a frequency of JAK2 mutated cells of 20%, 18%, and 100%, respectively. The mutation remained undetectable in control mice (Figure 1D-E).

To precisely measure the size of the JAK2V617F clone, we crossed PF4iCre;JAK2^V617F/WT mice with reporter mT/mG mice.¹¹ PF4iCre;mT/mG;JAK2^V617F/WT mice developed a PV-like phenotype with similar kinetics as PF4iCre;JAK2^V617F/WT mice, but with a more limited thrombocytosis (supplemental Figure 1). This difference previously reported^1,15,16  highlights the effect of the genetic background on the phenotypic expression of an oncogenic mutation.^17,18  As expected, GFP expression was observed in nearly all platelets of control and JAK2-mutated mice (Figure 2A). However, GFP was also detectable in other myeloid lineages at a higher level in PF4iCre;mT/mG;JAK2^V617F/WT than in control mice (16.6 ± 3.6% vs 0.52 ± 0.09% for Ter119⁺ cells, P < .001, 16.0 ± 2.6% vs 6.9 ± 1.6% for neutrophils, P = .0025, Figure 2A; supplemental Figure 2A). These results illustrate the proliferative advantage conferred by the JAK2V617F mutation to myeloid cells. GFP⁺ cells were also detectable in a small lymphoid population, with a similar level between PF4iCre;mT/mG;JAK2^WT/WT and PF4iCre;mT/mG;JAK2^V617F/WT mice (supplemental Figure 2B). As expected, in the bone marrow, the majority of MK expressed GFP in JAK2WT (59.2 ± 13.2%) and JAK2V617F mice (70.1 ± 6.1%). In JAK2WT mice, only a minority of erythroid (8.5 ± 1.3%) and granulocytic precursors (1.2 ± 0.2%) were GFP⁺, whereas in JAK2V617F mice we observed a higher frequency of GFP⁺ cells in erythroid (29.0 ± 8.3%, P = .05) and granulocytic (5.5 ± 1.2%) compartments (Figure 2B; supplemental Figure 2C). The low proportion of Ter119⁺GFP⁺ in the blood is probably linked to the absence of nucleus in mature erythrocytes. Indeed, the absence of new protein synthesis in erythrocytes is probably responsible for the disappearance of the protein from the membrane. This is consistent with the absence of detection of tdTomato at the surface of Ter119⁺ cells as previously described.¹⁴  Ter119⁺GFP⁺ cells in PF4iCre;mT/mG;JAK2^V617F/WT mice probably represent the reticulocytes in which messenger RNA are still present, allowing a residual production of GFP.

Besides, the mTmG reporter allowed us to demonstrate that the KI only occurs in the hematopoietic system but not in endothelial cells, as confirmed by the absence of detection of JAK2 mutation in sorted CD31⁺ cells (Figure 1D-E; supplemental Figure 1B).

We then explored the recombination events at early stages of hematopoiesis (Figure 2C; supplemental Figure 2D). At 12 weeks of age, in PF4iCre;mT/mG;JAK2^WT/WT mice, GFP⁺ cells were detectable mainly in MK progenitors (MkP, 15.1 ± 0.7%) and multipotent progenitors 2 (MPP2, 7.0 ± 1.2%, Figure 2C). In contrast, GFP⁺ cells were rare in other hematopoietic compartments, including short term–HSC (3.0 ± 0.7%) and SLAM-HSC (4.7 ± 1%). In PF4iCre;mT/mG;JAK2^V617F/WT mice, we detected higher level of GFP⁺ cells in the different hematopoietic populations albeit only significant for MkP and MPP2 cells (20.8 ± 2.6%, P = .007, and 13.8 ± 2.4, P = .002, respectively) suggesting that in PF4iCre;mT/mG;JAK2^V617F/WT mice, the JAK2 mutation is preferentially expressed in MPP2 and MkP, previously described as “megakaryocytes-primed populations.”^19,20  Of note, we observed a nearly twofold increase in the GFP⁺ SLAM-HSC (8.9 ± 1.5%) compared with control mice (Figure 2C; supplemental Figure 2D). Altogether, these results demonstrate that in PF4iCre;mT/mG;JAK2^V617F/WT mice, the recombination occurred not only in the megakaryocytic lineage but also in a small subset of HSC because of the leakiness of the PF4 promoter in HSC, as previously suggested.^13,14  The discrepancy with recently described PF4-Cre driven JAK2V617F KI model^15 ,16  could once again highlight the impact of the genetic background on the phenotypic expression of a mouse model. Moreover, the comparison of the proportion of recombined cells according to quantitative polymerase chain reaction and flow cytometry analyses points to a possible underestimation by the mTmG reporter, which could be more pronounced in the model of Woods et al compared with ours.

Finally, we studied the kinetics of expansion of hematopoietic cells when they express JAK2V617F. In blood cells (Figure 2A) as well as in bone marrow precursors (Figure 2B), the proportion of GFP⁺ cells remained stable in control mice from 12 to 24 weeks in all lineages. In contrast, we observed an increase of the frequency of Ter119⁺GFP⁺ and Ly6G⁺GFP⁺ cells in mutant mice, consistent with the PV phenotype observed in these animals. A progressive amplification of nearly all categories of immature progenitors was observed in PF4iCre;mT/mG;JAK2^V617F/WT mice (Figure 2C). We did not notice any significant amplification of HSC in PF4iCre;mT/mG;JAK2^V617F/WT, but a high variability among mice (Figure 2C; supplemental Figure 2D). This is consistent with other mouse models and the clonal architecture observed in ET and PV patients that suggested a minor impact of the mutation on the more immature compartments.^3,4 

In conclusion, in PF4iCre;mT/mG;JAK2^V617F/WT mice, the JAK2V617F mutation occurs in a few HSC and in a majority of MK leading to the development of a full-blown MPN. This KI model is the first to recapitulate the clonal architecture (few mutated HSC, and expanded population of mutated progenitors) observed in ET and PV (but not primary myelofibrosis) patients without using any deleterious irradiation or transplantation. Thus it represents a precious tool to study the mechanisms involved in disease initiation, clonal dominance and evolution to myelofibrosis.

The authors thank Jean Luc Villeval for the kind gift of JAK2^V617F KI mice and Christian Gachet for the PF4-iCre mice.

This project was supported by research grants from ANR-DFG JAKPOT (no. ANR-14-CE35-0022-02), INSERM, and The Fondation Bettencourt Schueller, the Aquitaine Region. A.V.G. was supported by la ligue nationale contre le cancer and a research starting grant from INSERM and V.G.L. by ANR-DFG JAKPOT. I.P. was funded by INCA-PLBIO 2017.

Contribution: C.J. designed the study and analyzed data; B.K., O.M., V.G.-L., and C.M. performed animal experimentation, flow cytometry, and analyzed data; O.M. and A.V.G. prepared the figures; O.M., A.V.G., C.J., and W.V. wrote the paper; C.J., I.P., and W.V. supervised the study; and all authors revised the manuscript.

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

Correspondence: Chloe James, INSERM UMR 1034, CHU Haut-Lévêque, 1 Av Magellan, 33600 Pessac, France; e-mail: chloe.james@inserm.fr.