JAK2 stimulates homologous recombination and genetic instability: potential implication in the heterogeneity of myeloproliferative disorders

Myeloproliferative disorders (MPDs) are clonal hematopoietic malignancies in which one or several myeloid lineages (ie, granulocytic, erythroid, and megakaryocytic) are abnormally amplified. This amplification is thought to result from the deregulation of hematopoietic stem cells with a downstream selective proliferation advantage in late myeloid differentiation. MPDs are classified by the World Health Organization into 3 categories: (1) chronic myeloid leukemia (CML); (2) classical MPD, including polycythemia vera (PV), essential thrombocythemia (ET), and primitive myelofibrosis (PMF); and (3) unclassified and rare MPDs comprising chronic neutrophilic leukemia (CNL), hypereosinophilic syndrome/chronic eosinophilic leukemia (HEL/CEL), and other unclassifiable myeloproliferation.

MPDs frequently involve the deregulation of a tyrosine kinase because of the acquisition of a monogenetic abnormality in a hematopoietic stem cell, the archetype of which is BCR/ABL in CML. Recently, several groups have identified a recurrent acquired mutation in the tyrosine Janus kinase 2 (JAK2) gene in most cases of PV patients and in approximately half of ET and PMF patients.^1-5  JAK2 is a tyrosine kinase that becomes activated downstream of several cytokine receptors after ligand binding, such as erythropoietin (EPO). JAK2 activation leads to activation of many signaling pathways, including the MAPK, PI3K, and STATs pathways. The V617F point mutation lies in the autoinhibitory JH2 domain of JAK2, and in vitro studies have demonstrated that the mutation leads to JAK2 autophosphorylation and to the constitutive activation of downstream pathways.^2-4 

Although ET, PV, and PMF share some common phenotypic features,^6-8  it is surprising that a single point mutation gives rise to several disorders.⁹  One possible explanation is the JAK2^V617F “dosage” hypothesis, which postulates that the JAK2 mutation could be the starting point of the 3 pathologies, and the occurrence of other genetic events, that may modify JAK2 kinase activity, could explain the heterogeneity among the classical MPD. This hypothesis is based on mouse models in which low levels of JAK2^V617F induce an ET phenotype, whereas higher expression leads to a PV progressing to myelofibrosis.^10,11  In addition, JAK2^V617F burden in granulocytes and bone marrow samples is markedly low in ET patients compared with PV and PMF patients.¹²  This is related to the fact that the JAK2 mutation is monoallelic in ET patients, whereas the mutation is bi-allelic in most PV patients (at least in some progenitors) because of a loss of heterozygosity after a mitotic recombination.^3,13,14  Progression of PV to PMF is accompanied by the expansion of the biallelic JAK2 mutated clone.¹⁵ 

Several mechanisms have been proposed that may contribute to the loss of heterozygosity and/or the acquisition of new genetic abnormalities. Among these, they are DNA-double-strand break repair mechanisms, including the homologous recombination (HR) and nonhomologous end-joining (NHEJ) mechanisms. Because both excessive and defective HR and NHEJ can lead to genome instability, a very precise and acute regulation of these mechanisms is essential to maintain the necessary equilibrium between stability and diversification of the genome. Indeed, HR can become deleterious, and an uncontrolled HR excess promotes genetic instability¹⁶  and diseases.¹⁷  For instance, HR with crossing over between dispersed repetitive sequences can lead to a variety of genome rearrangements.¹⁶  Gene conversion (a product of HR) between homolog allele can lead to loss of heterozygosity,¹⁸  whereas gene conversion with a pseudogene, which generally bears stop mutations, can inactivate a functional allele.¹⁹  Moreover, the accumulation of aberrant abortive recombination events can also be toxic.²⁰  Similar to excessive HR, defects in HR can also result in genetic instability.^21-23  Cells expressing a dominant negative form of RAD51, a key protein involved in HR,²⁴  produce more tumors when injected into nude mice.²¹  The level of RAD51 appears to be an important parameter for inducing gene instability. Decreased RAD51 protein levels have been reported in breast carcinomas²⁵ ; in contrast, increased levels of RAD51 were detected in other cancer cells^26-28  and, especially, in cells expressing BCR/ABL or other fusion tyrosine kinase implicated in hematologic diseases.²⁹  Alternatively, deregulation of NHEJ may also lead to genome rearrangement, including translocation, deletion, and insertion.³⁰ 

In this work, we hypothesized that JAK2^V617F may deregulate HR and/or NHEJ mechanisms, which could in turn be responsible for both the loss of heterozygosity of JAK2 and the acquisition of additional genetic events. Thus, deregulation of such mechanisms could explain not only the heterogeneity of MPD but also the evolution of MPD to acute leukemia.

Fetal bovine serum (FBS), murine recombinant thrombopoietin (TPO), and interleukin (IL)–6 were from StemCell Technologies (Vancouver, BC). Liquid cell culture media, including Iscove modified Dulbecco medium and Dulbecco modified Eagle medium (DMEM), were from Invitrogen (Paisley, United Kingdom). Human recombinant EPO, stem cell factor (SCF), fetal liver tyrosine kinase 3 (FLT3-L), and IL-3 were generous gifts from Amgen (Thousand Oaks, CA) and recombinant TPO from Kirin (Tokyo, Japan). Recombinant human IL-6 was provided by S. Burstein (University of Oklahoma, Oklahoma City, OK). Restriction enzymes were purchased from MBI Fermentas (Hanover, MD).

The HR-EGFP/3′EGFP plasmid for HR measurements was constructed by insertion of the puromycin resistance gene, one acceptor gene cassette (HR-EGFP), one spacer cassette (Hygromycin B), and one donor gene (3′EGFP) in series into multiple cloning site of the retroviral vector p5NM.³¹  The acceptor gene substrate was designed such that the I-SceI recognition sequence replaced 4 bp (HR-EGFP) and the donor gene 3′EGFP was designed such that the ATG was mutated.

Murine and human JAK2^V617F and wild-type JAK2 (JAK2wt) pMEGIX retroviral vectors were previously described.³²  The KS-TEL-JAK2 plasmid was a gift from Dr V. Lacronique (Paris, France). These 3 plasmids were subcloned into the retroviral pREX-CD4 plasmid kindly provided by Dr S. Constantinescu (Ludwig Institute for Cancer Research, Brussels, Belgium).³³  Retroviral particles were produced by transfection of 293EBNA cells with 3 different plasmids: pgag-pol (packaging plasmid), pVSV-G (coding for the VSV-G protein envelope), and pREX containing JAK2 constructs, in the presence of lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions.

The I-SceI gene was inserted into an HIV-derived lentiviral vector (pRRL sin PGK WPRE; Généthon, Evry, France). Lentivirus particles were produced by transfection of 293T cells with 3 different plasmids: pCMV (packaging plasmid), pMD2G (coding for the VSV-G protein envelope), and pRRL/PGK-I-Sce1 in the presence of lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions. Viral stocks were stored at −80°C.

The murine, pro B Ba/F3-EPOR cell line and derivative cell lines were cultured in DMEM medium (Invitrogen) complemented with 10% FBS (Stem Cell Technologies) and in the presence of 1 U/mL EPO. Parental, JAK2wt- or JAK2^V617F–expressing Ba/F3-EPOR cells were obtained by infecting cells with human or murine JAK2^V617F and wild-type JAK2 (JAK2wt) pMEGIX retroviral vectors.³²  The Ba/F3-HR2 cell line was obtained by electroporation of Ba/F3-EPOR cells with the HR-EGFP/3′EGFP plasmid and through a selection of clones grown in methylcellulose in the presence of EPO and puromycin (1 μg/mL). Intrachromosomal integration was verified by polymerase chain reaction (PCR) relying on the following oligonucleotides, as previously described³¹ : PCR-2-1 (5′-TACACAAATCGCCCGCAGAAGC-3′) and PCR-2-2 (5′-CTGTCTTTAACAAATTGGACTAATCG-3′). PCR conditions were 5 minutes at 94°C, then 35 cycles at 92°C for 60 seconds, 60°C for 60 seconds, and 72°C for 120 seconds, ending with a final extension step of 72°C for 7 minutes.

For the Ba/F3-HR2 cell line, cells were plated at 4 × 10⁴ cells per well in 96-well plates and infected with pREX-CD4 constructs containing the murine mutated or wt JAK2. For CD34⁺ cells isolated from patients and controls, cells were immunopurified and amplified in medium containing SCF/IL-3/EPO for 5 days. Then, cells were plated at 4 × 10⁴ cells per well in 96-well plates and infected with the HR substrate. Five days after infection, cells were washed and fixed in phosphate-buffered saline (PBS)/2% paraformaldehyde for 15 minutes at room temperature, and GFP⁺ cells were detected by flow cytometric analysis using a FACSort (Becton Dickinson, Franklin Lakes, NJ).

Ba/F3-HR2 cells were infected with pREX-CD4 constructs containing JAK2 and infected with I-SceI virus. Noninfected cells served as controls. GFP⁻ cells were then sorted by a FACSDIVA to exclude HR events, and PCR was performed on genomic DNA with the oligonucleotides PCR-1-1 and PCR-1-2 as previously described.³¹  PCR products were then digested with I-SceI or with XhoI. Band intensities of uncleaved 2.3-kb PCR fragments were quantified and corrected for background and loading. Error-prone NHEJ was estimated by calculating the difference between the corrected band intensities for I-SceI and for the digestion control with XhoI.

Whole-cell extract (WCE) preparation and in vitro NHEJ assay were performed using a procedure adapted from Baumann.^34,35  Briefly, after washing in 1× PBS, cells were lysed through 3 freeze/thaw cycles in LB buffer (25 mM Tris, pH 7.5, 333 mM KCl, 1.3 mM ethylenediaminetetraacetic acid (EDTA), 4 mM dithiothreitol (DTT), protease, and phosphatase inhibitors). Lysates were incubated for 20 minutes at 4°C and cleared by centrifugation. Supernatants were dialyzed against 1× E buffer (20 mM of Tris, pH 8.0, 20% glycerol, 0.1 M K(OAc), 0.5 mM EDTA, 1 mM DTT). WCEs were kept frozen (−80°C) until use. For NHEJ assay, 15 μg WCE was incubated (10 μL reaction) with 25 ng linear DNA (EcoRI digested pEGFPN2) or 50 ng linear DNA (EcoRI digested pBluescript) in 1× LigB (250 mM Tris, pH 8.0, 300 mM K(OAc), 2.5 mM Mg(OAc)2, 5 mM adenosine triphosphate, 5 mM DTT, 0.5 mg/mL bovine serum albumin (BSA), and 1 μg/mL IP6 for 2 hours at 37°C. Reactions were then treated with 1 μL RNase (1 mg/mL) for 5 minutes at room temperature and with 2 μL 5× deproteination solution (10 mg/mL Proteinase K, 2.5% sodium dodecyl sulfate, 50 mM EDTA, 100 mM Tris, pH 7.5) for 30 minutes at 55°C. After migration of the samples in 0.7% agarose, the gels were stained with SYBR-Green (30 minutes), and fluorescence was detected via a FluorImager.

Cells were washed in PBS and lysed in denaturing loading dye buffer for 20 minutes at 4°C. Samples were boiled for 5 minutes and subjected to Western blot analysis using anti-JAK2, anti [pY⁷⁰⁵]STAT5 antibodies (Cell Signaling Technology, Danvers, MA), anti-RAD51 (Santa Cruz Biotechnology, Santa Cruz, CA) and anti–β-actin antibodies. Antibodies were visualized using the enhanced chemiluminescence detection kit (GE Healthcare, Chalfont St Giles, United Kingdom).

RAD51 foci were analyzed as described.³⁶  Briefly, after cytocentrifugation, smears were fixed in absolute methanol for 15 minutes at −20°C and then rinsed in ice-cold acetone for a few seconds. After 3 washes with PBS, the slides were incubated with blocking solution (PBS with 5% BSA and 0.1% Tween 20) for 30 minutes and then incubated with a rabbit anti-RAD51 antibody (1/100; Oncogene, VWR, Fontenay sous Bois, France) in a humidified incubator for 2 hours. The slides were then washed in PBS another 3 times for 10 minutes each and incubated for 1 hour with Alexa 488-conjugated anti–rabbit IgG appropriately diluted with PBS. After 3 additional washes with PBS, the preparations were counterstained with 4,6 diamidino-2-phenylindole for 1 minute. The slides were mounted with Fluoromount-G mounting medium. The number of RAD51 foci per cell was measured, and cells with more than 5 foci were considered positive. For centrosome analysis, cells were fixed and permeabilized during 15 minutes in cold methanol followed by 10 seconds in acetone. Centrosomes were counted in metaphases using an anti–gamma-tubulin antibody (Sigma-Aldrich Laborchemikalien, Seelze, Germany). Cells were examined under a Nikon E600 fluorescent microscope (Nikon, Tokyo, Japan) equipped with a Fluor 60×/1.40 numeric aperture or a Fluor 100×/0.05-1.3 oil objective. The photographs were obtained with a CoolSnap digital camera and software (RS Photometrics, Tucson, AZ).

Peripheral blood samples were collected from PV and PMF patients. For PV, diagnoses were made according to the modified Polycythemia Vera Study Group criteria,³⁷  and the Italian criteria for PMF.³⁸  Only JAK2V617-positive patients were selected after analyzing mutational status using fluorescent competitive probes for quantitative real-time PCR on an ABI 7500 (Applied Biosystems, Foster City, CA), as reported previously.^4,32  Control peripheral blood samples were collected from patients undergoing cytapheresis. The study was approved by the Local Research Ethics Committee from the Hôtel-Dieu and the Henri Mondor hospitals, and informed consent was obtained from each patient in accordance with the Declaration of Helsinki.

Mononuclear cells were separated over a Ficoll density gradient, and CD34⁺ cells were purified by a double-positive magnetic cell sorting system (AutoMACS; Miltenyi Biotec, Auburn, CA), according to the manufacturer's recommendations. CD34⁺ cells were amplified for 5 days in Iscove modified Dulbecco medium with penicillin/streptomycin/glutamine, alpha-thioglycerol, BSA, a mixture of sonicated lipids, and insulin-transferrin in the presence of 5 recombinant human cytokines (25 ng/mL SCF, 100 U/mL IL-3, 10 ng/mL IL-6, 10 ng/mL FLT3-L, and 10 ng/mL TPO).

Cells were incubated with colchicine (50 μg/mL) for 2 hours, collected, and incubated in KCl (12.5 mM) for 20 minutes at 37°C. Cells were fixed overnight in methanol/acetic acid (3/1 vol/vol), washed 3 times in methanol/acetic acid, spread on cold microscope slides, and stained with 3% Giemsa in 1.5% phosphate buffer, pH 6.7. Metaphase-stage cells were analyzed by microscopy.

A total of 3.5 × 10⁵ cells were plated in a 24-well plate containing 4 μg/mL bromodeoxyuridine. After 2 cell divisions in the presence of bromodeoxyuridine (30 hours), colcemid (Sigma-Aldrich Laborchemikalien) was added at a final concentration of 0.5 μg/mL for 2 hours. Cells were then washed with PBS, trypsinized, collected, and centrifuged. Cells in pellet were dispersed in 10 mL of 0.075 M of KCl and incubated for 15 minutes at 37°C. Before centrifugation, 100 μL of the fixative (methanol-acetic acid, 3:1) was added. The pellet was suspended in the fixative and incubated overnight at 4°C. Cells were then centrifuged, washed in a fresh fixative, and dropped onto wet slides. The slides were covered for 20 minutes with 10 μg/mL of Hoechst 33 258 (Sigma-Aldrich Laborchemikalien) dissolved in water and subsequently incubated in a solution of (2×) sodium saline citrate twice, pH 6.3, for 2 hours under ultraviolet light. The slides were plunged into sodium saline citrate twice, pH 6.3, at 60°C for 15 minutes, and stained with 1.5% Giemsa for 3 to 5 minutes.

A fluctuation analysis was used to estimate spontaneous mutagenesis. For each cell line analyzed, several independent cultures were plated and cultured to confluence. Cells were then counted, and one part was used for plating efficiency estimation. The remaining cells were plated in methylcellulose with EPO, and the mutant colonies were selected by either 2 mM ouabain or 20 μM 6-TG. The number of ouabain- or 6-TG-resistant clones allowed us to calculate the mutagenesis frequency. The rate of recombination or mutagenesis per cell per generation was calculated using the fluctuation analysis of Luria and Delbrück.^39,40 

The Ba/F3 cell line expressing the EPO receptor was electroporated with the HR vector (HR-EGFP/3′EGFP; Figure 1A),³¹  and puromycin-resistant clones were isolated in methylcellulose. Several clones were amplified, genomic DNA was extracted, and PCR was performed to verify vector integration (Figure 1B). This intrachromosomal HR vector monitors all kinds of HR events, including HR events associated with and without crossing over and single-strand annealing events, by measuring the frequency of GFP⁺ cells. To study the impact of JAK2 activation on HR, Ba/F3-HR2 cells were infected with JAK2wt, JAK2^V617F, TEL/JAK2, or the empty vector. Overexpression of different forms of JAK2 and ectopic expression of TEL/JAK2 were verified by Western blotting using an anti-JAK2 antibody (Figure 1C). Then, we checked JAK2 activity by measuring STAT5 phosphorylation by Western blotting using an anti[pY⁷⁰⁵]STAT5 antibody. In Ba/F3-HR2 cells, EPO stimulated a transient phosphorylation of STAT5, whereas phosphorylation was more pronounced and prolonged in JAK2wt overexpressing cells. In contrast, a spontaneous STAT5 phosphorylation and sustained activation were observed in JAK2^V617F-overexpressing cells (Figure 1D). Moreover, in the presence of EPO, JAK2^V617F and, to a lesser extent, JAK2wt induced a 20- and 5-fold increase in spontaneous HR, respectively, compared with Ba/F3-HR2 parental cells (Figure 1E). As previously described, TEL/JAK2 induced an increase in HR.⁴¹  Importantly, when we infected cells with a lentiviral vector encoding the rare-cutting meganuclease I-SceI, which targets a unique double-strand break (DSB) in the HR vector,⁴²  induced HR increased significantly in JAK2^V617F and TEL/JAK2 overexpressing cells compared with Ba/F3-HR2 parental cells (Figure 1F).

RAD51 is a key protein involved in HR, whose expression is deregulated by an ectopic expression of BCR/ABL and TEL/JAK2.^29,41  Surprisingly, we did not find any significant changes in RAD51 protein levels in Ba/F3 cells expressing human or murine JAK2 constructs (Figure 2A and data not shown).

After a genotoxic stress, RAD51 assembles into nuclear DNA repair foci.^36,43  Thus, we counted the number of cells with RAD51 foci (Figure 2C) and the number of foci per cell (Figure 2D) after immunostaining with an anti-RAD51 antibody. As a positive control, we treated Ba/F3-EPOR cells with mitomycin C (MMC) for 3 hours and found a 5-fold increase in the number of cells with RAD51 foci. As a confirmation of our observations that HR was increased by JAK2 activation, we observed in unchallenged cells a significant increase both in the number of cells harboring nuclear RAD51 foci and in the number of foci/cell in Ba/F3-JAK2^V617F and, to a lesser extent, in Ba/F3-JAK2wt compared with parental cells (Figure 2B-D). Because the amount of RAD51 is known to vary during cell cycle, we verified whether the Ba/F3-JAK2^V617F cell cycle was not altered and found no difference compare with the control cell line (Figure 2E). Altogether, the present results show that activation of JAK2 greatly stimulates HR in the Ba/F3-EPOR cell line.

We then wanted to determine the impact of JAK2 activation on NHEJ because both NHEJ and HR can lead to genome rearrangement. First, we analyzed in vitro end-joining of 2 linearized plasmid DNA using cell-free extracts^34,35  (Figure 3A,B). This DNA-end ligation assay, which results in the formation of DNA concatemers, requires a functional NHEJ apparatus, as demonstrated by the absence of DNA oligomers when using extracts from a DNA Ligase IV deficient cell line (N114) vs parental lymphoid cell line (Nalm6). When using extracts from Ba/F3-EPOR, Ba/F3-JAK2wt, or Ba/F3-JAK2^V617F cell lines, we found no significant changes in DNA end-joining between the 3 Ba/F3 cell lines (Figure 3A,B). We then investigated the NHEJ fidelity by the intrachromosomal substrate (HR-EGFP/3′EGFP). Ba/F3-HR2 cells were infected or not with I-SceI virus, and EGFP⁻ cells were sorted to exclude HR events. In the remaining cells, error-free religation may have reconstituted the I-SceI site after cleavage, whereas error-prone NHEJ may have caused loss of the I-SceI site. Therefore, we performed PCR reactions as indicated in Figure 3C and cleaved the resulting fragments with I-SceI and as a control digest with XhoI. Error-prone NHEJ was estimated by quantification of the intensities of the uncleaved 2.3-kb bands for I-SceI and XhoI digests (corrected for background), subsequent calculation of the fraction of the 2.3-kb fragment in total DNA (loading), and comparison of specific I-SceI vs control XhoI digest values (Figure 3D). The data from 3 experiments did not indicate significant differences between the 3 cell lines regarding specific I-SceI site alteration, which is indicative of error-prone NHEJ. Altogether, these results indicate that JAK2 activation has no impact on both NHEJ efficiency and fidelity.

We further investigated whether the JAK2-induced hyper-recombination phenotype could be associated with genetic instability. Indeed, previous reports have shown that HR mutants for RAD51, XRRC2, or XRCC3 lead to centrosome and ploidy defects.^21,44,45  For this reason, we measured the numbers of centrosomes in mitosis of the different cell lines by immunofluorescence using an anti–γ-tubulin antibody (Figure 4A). The Ba/F3-EPOR cells exhibited a low number of mitoses (3.2%) where more than 2 centrosomes could be seen. However, a higher occurrence of centrosome abnormalities was recorded in the Ba/F3-JAK2^V617F cell lines (12.4% of mitosis; P < .001; Figure 4B). To a lesser degree, centrosome abnormalities were found in 6.9% of mitoses in the Ba/F3-JAK2wt cell lines, a percentage slightly elevated compared with the parental cell line (P = .082). These results show that JAK2 activation induces an increase in the number of centrosomes during mitosis. Because centrosome defects should lead to aneuploidy, we measured cell ploidy. As shown in Figure 4C, a significantly increased number of aneuploid cells was seen for Ba/F3-JAK2wt and Ba/F3-JAK2^V617F cells compared with the parental cell line as indicated by different distribution (P < .01). Indeed, we observed both an increase in the number of metaphases with less than 39 chromosomes (corresponding to a loss of chromosome) as well as of metaphases with more than 39 chromosomes (corresponding to a gain of chromosome; Figure 4D). Thus, aneuploidy corresponded to both gain and loss of chromosomes and was consistent with the increased centrosome abnormalities detected after JAK2 activation. In addition, we also observed a significant increase in metaphase corresponding to a tetraploid content in Ba/F3-JAK2^V617F cells (16.4%) and Ba/F3-JAK2wt (6.1%) vs control Ba/F3 (2.8%).

We also evaluated the genomic instability by measuring the sister chromatid exchange (SCE), which is the result of HR associated with crossing over mechanism. An increase (56%) in spontaneous SCE in Ba/F3-JAK2^V617F metaphases was observed compared with Ba/F3 cells and Ba/F3-JAK2wt (Figure 4E). Because SCE involves HR, this result also fits with a JAK2-induced hyperrecombination phenotype. Altogether, these results suggest that JAK2 activation may produce genetic instability.

These data show that JAK2 activation deregulates the HR pathway. Consequently, one prediction would be that mutagenesis should be increased in cells expressing ectopic JAK2. For that reason, we measured spontaneous mutagenesis at 2 different loci: the Na⁺/K⁺-ATPase membrane gene and the HPRT gene that leads to ouabain resistance and 6-thioguanine (6-TG) resistance, respectively, in mutant cells. Whereas Na⁺/K⁺-ATPase membrane mutants reflect only point mutations, HPRT mutants monitor point mutations, deletions, and insertions.⁴⁶  We first measured spontaneous mutagenesis in the HPRT locus. Spontaneous mutagenesis was calculated by fluctuation analysis using the Luria and Delbrück⁴⁰  or the Capizzi and Jameson assays.³⁹  In Ba/F3-EPOR cells, JAK2^V617F led to a 100-fold increase in the spontaneous rate of mutagenesis per cell per generation compared with control or Ba/F3-JAK2wt cells (Table 1). We then measured spontaneous mutagenesis in the Na⁺/K⁺-ATPase membrane locus. At this locus, JAK2^V617F induced a 10-fold increase in the spontaneous rate of mutagenesis per cell per generation compared with control or Ba/F3-JAK2wt cells (Table 1). Altogether, these findings show that JAK2^V617F generates a mutator phenotype that can be potentially associated with punctual mutations, deletion, and/or insertion.

Ba/F3 parental or Ba/F3 cells expressing either JAK2wt or JAK2^V617F were treated for one hour with genotoxic drugs, and viable cells were counted by trypan blue exclusion test 48 hours later. First, the cell lines were treated with the DNA cross-linking drug MMC because HR is stimulated in BCR/ABL-expressing cells exhibiting resistance to MMC, on one hand, and sensitivity to this drug is a hallmark of a defective HR pathway, on the other hand.²⁹  Expression of JAK2^V617F in Ba/F3 cells resulted in a 4- to 5-fold increased resistance to MMC (concentration that inhibits response by 50% [IC₅₀] = 3.0 μM) compared with parental Ba/F3 (IC₅₀ = 0.7 μM) or Ba/F3-JAK2wt (IC₅₀ = 0.6 μM) cells, respectively (Figure 5A). Second, we treated the cell lines with bleomycin because this agent generates DNA double-strand breaks that can be repaired by the HR mechanism.^47,48  The expression of JAK2^V617F in Ba/F3 cells resulted in a 13-fold increased resistance to bleomycin (IC₅₀ = 9.0 μM) compared with parental Ba/F3 cells (IC₅₀ = 0.7 μM; Figure 5B). Moreover, JAK2wt induced a resistance to bleomycin but to a lesser extent than JAK2^V617F (IC₅₀ = 4.6 μM). Importantly, under these treatment conditions (1-hour exposure), neither typical morphologic features of apoptosis nor caspase-3 cleavage was found in contrast to the situation after continuous exposure with these agents (Figure 5C,D). Altogether, these results show that a strong activation of JAK2 induces a resistance against genotoxic agents because of increased repair rather than decreased apoptosis.

To confirm the hyper-recombination phenotype observed in the JAK2^V617F-expressing cell line, we measured nuclear RAD51 foci in cells from JAK2^V617F-positive patients (at least heterozygous for JAK2^V617F in granulocytes at the diagnosis). For this purpose, CD34⁺ cells were immunopurified from the blood of 5 PV patients and 6 MF (5 PMF and 1 post-PV MF) patients. CD34⁺ cells were then amplified for 5 days with 5 cytokines (SCF/IL-3/FLT3-L/IL-6/TPO), and the number of cells presenting RAD51 foci was counted for each patient (Figure 6A). CD34⁺ cells from G-CSF-mobilized healthy donors were used as a control. We observed that PV and MF cells displayed a high number of cells with RAD51 foci compared with mobilized blood (Figure 6B). We next investigated the HR activity in CD34⁺ cells from the blood of 3 PV patients: 2 PMF patients and 3 mobilized donors. CD34⁺ cells were amplified for 5 days with 3 cytokines (SCF/IL-3/EPO) and infected with the retrovirus containing the HR substrate. Importantly, no difference in the proliferation was found in CD34⁺ cells from healthy donors versus patients under these culture conditions (Figure 6C). In these experiments and as measured by the percentage of GFP⁺ cells, we observed that cells derived from PV and MF patients displayed a higher HR activity compared with mobilized blood (Figure 6D). Taken together, our data indicate that JAK2^V617F also induced a hyperrecombination phenotype in MPD patients.

In the present study, we investigated the effect of the JAK2^V617F mutation on HR to understand whether this could lead to genetic instability and explain disease progression. The point mutation lies in exon 14 of the JAK2 gene and results in a valine to phenylalanine substitution at position 617 (V617F) in the pseudo-kinase domain of JAK2. This mutation induces a spontaneous activation of the tyrosine kinase, which seems to require a homodimeric type I cytokine receptor. However, the presence of such a receptor does not appear totally indispensable,^4,32,49  but a receptor such as the erythropoietin receptor (EPO-R) facilitates the transforming effects of JAK2^V617F.^49,50  Our results show, for the first time, that activation of an overexpressed JAK2wt in the presence of EPO resulted in an increase in both HR activity and RAD51 nuclear foci formation in the Ba/F3-EPOR cell line. Overexpression of JAK2^V617F in Ba/F3-EPOR cells had an even more potent effect, probably because of its stronger activity compared with JAK2wt (Figure 1D). Importantly, the concomitant increase in spontaneous HR activity and RAD51 nuclear foci formation was also observed in CD34⁺-derived cells from JAK2^V617F-positive PV and PMF patients, indicating a hyper-recombination phenotype. These results are consistent with previous data showing that strong JAK2 activation by TEL/JAK2 fusion tyrosine kinase stimulated HR,⁴¹  a result confirmed in the present work. Furthermore, our findings also fit with the BCR/ABL-induced hyper-recombination phenotype observed in CML²⁹  and, therefore, strengthen the assumption that numerous kinases are capable of interfering with HR.

Although the increase in HR measured intrachromosomally correlated with RAD51 focus formation, we were unable to detect change in RAD51 protein levels by JAK2 activation in contrast to CML.²⁹  However, these differences in RAD51 levels associated with BCR/ABL vs JAK2 activation may be related to differences in the cell cycle distribution because low levels of RAD51 are found in G₀/G₁ followed by an increase during S and G₂/M.⁵¹  Consistent with the stable RAD51 expression, we did not observe marked cell cycle changes in cells expressing either JAK2wt or JAK2^V617F compared with control cells.

Because deregulation in NHEJ can also lead to genome rearrangements³⁰  and because HR and NHEJ might be competitive mechanisms,⁵²  we also investigated the impact of JAK2 activation on the NHEJ mechanism but did not find any significant change either in the efficiency or in the fidelity of NHEJ. These results represent the first marked difference in the regulation of DNA repair between CML and JAK2^V617F-positive-MPD because BCR/ABL is able to greatly activate NHEJ.⁵³ 

Because excess in HR can promote genetic instability,^16,17  the hyperrecombination phenotype induced by JAK2^V617F may explain both the heterogeneity of MPD and disease progression. Moreover, because gene conversion between homologs is a product of HR and leads to loss of heterozygosity,^18,54  the JAK2^V617F-induced hyperrecombination phenotype might account for the loss of heterozygosity in 90% of PV patients.³  The work by Kralovics et al suggests that mitotic recombination is the major mechanism for loss of heterozygosity.³  Importantly, mitotic recombination requires HR and is facilitated by a hyperrecombination state. However, the exact mechanism by which JAK2^V617F controls the hyperrecombination remains to be investigated. One can speculate that, in analogy to BCR/ABL, this mechanism could be indirect through the activation of MAPK, PI3K, or STAT5 or via direct tyrosine phosphorylation of RAD51.²⁹ 

In vertebrate cells, dysfunction of HR can result in genome instability^21,22,44,45  and increased mutagenesis.⁵⁵  Defective centrosome duplication leads to multipolar mitotic cells resulting in aberrant chromosome segregation and aneuploidy, a hallmark of tumor cells. Interestingly, this study also revealed that activation of JAK2 generates centrosome abnormalities associated with chromosomal instability. This situation is highly consistent with JAK2 impact on HR, which leads to genetic instability as monitored by mutagenesis assays. Mutagenesis results further suggested that JAK2^V617F not only generates point mutations but also increases deletions and/or insertions based on the differences between the inducible mutation rates obtained in Na⁺/K⁺-ATPase and HPRT loci. It is worthwhile noting that only JAK2^V617F was able to stimulate mutagenesis, suggesting that a strong activation of the kinase is necessary to generate this mutator phenotype. Alternatively, JAK2^V617F may phosphorylate substrates other than JAK2wt. Thus, JAK2^V617F-induced chromosomal instability and mutagenesis may explain the numerous different cytogenetic abnormalities found in rare cases of ET, PV (∼10%), and more frequently in MF (20%-40%).^56-58 

As previously reported in BCR/ABL-expressing cells,²⁹  our study also provides evidence that JAK2^V617F induces resistance to genotoxic agents, such as MMC and bleomycin, which are considered to be good inducers of HR.⁴⁴  Importantly, we excluded the fact that JAK2 activation–induced resistance was the result of decreased apoptosis because no apoptosis was found under our conditions of treatment. The discrepancy found in the resistance against these 2 agents may result from the fact that they differentially damage DNA. MMC is a DNA crosslinker that may stimulate HR and nucleotide excision repair. Bleomycin works through free radicals and causes DNA strand breaks,⁴⁷  in which at least HR, NHEJ, or base excision repair could be implicated. It remains to be determined whether JAK2^V67F may also modify other DNA repair mechanisms such as nucleotide excision repair or base excision repair. Moreover, the differences in the resistance observed between JAK2wt and JAK2^V67F may come from their respective activities as shown in Figure 1D. One can hypothesize that this resistance could be the result of accelerated DNA repair of the genotoxic lesions by HR and/or to the mutator phenotype induced by JAK2^V617F. Our findings may have major implications for future strategies in predicting the individual response of post-MPD leukemia patients to chemotherapeutic treatments, depending on the type of drug as well as the dosage applied. In addition, these data may implicate that JAK2^V617F can induce resistance to JAK2 inhibitors via the mutator phenotype.

In conclusion, our data show that strong activation of JAK2^V617F stimulates HR, centrosome, and ploidy abnormalities and induces a mutator phenotype and a resistance against genotoxic agents. Taken together, this study suggests for the first time that the heterogeneity of MPD features and its evolution into secondary leukemia could be in part ascribed to JAK2^V617F-induced genetic instability.

This work was supported by grants from Ligue Nationale Contre le Cancer Equipe labellisée 2007, INCa (projets libres 2006) and Inserm. I.P. was funded by the Fondation pour la Recherche Médicale. M.N. was funded by a postdoctoral fellowship from Ligue Nationale Contre le Cancer.

Contribution: I.P. performed all the experiments, wrote the manuscript, and designed the study; M.N. assisted I.P. in all experiments; L.M. and J.P.d.V. did the nonhomologous end-joining efficiency experiments and contributed to the discussion; L.W. provide the HR vector and contributed to the discussion; S.G. and J.-L.V. collected the samples of PV and PMF patients; and W.V. designed the study, supervised the work, wrote the manuscript, and was responsible for the final draft of the manuscript.

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

Correspondence: William Vainchenker, INSERM U790, Institut Gustave Roussy, 39 Camille Desmoulins, 94 805 Villejuif, France; e-mail: verpre@igr.fr.