Homologous recombination of wild-type JAK2, a novel early step in the development of myeloproliferative neoplasm

Mathias Vilaine, Damla Olcaydu, Ashot Harutyunyan, Jonathan Bergeman, Mourad Tiab, Jean-François Ramée, Jian-Min Chen, Robert Kralovics and Sylvie Hermouet

To the editor:

Transformation of hematopoietic cells depends on the acquisition of genetic events leading to cytokine independence, typically associated with acquisition of an autocrine cytokine loop or/and increased expression or/and mutation of JAK genes.1 Rearrangement of the JAK2 gene, which presumably alters JAK2 transcription, is reported in hematopoietic cells.2 Murine models of myeloproliferative neoplasms (MPN) demonstrated that the polycythemia vera (PV) phenotype requires the combination of high expression and activation of Jak2.3 Indeed, expression of both wild-type (WT) and mutant JAK2 transcripts can be high in PV.4 PV is characterized by a high frequency of the JAK2 46/1 (GGCC) haplotype (represented in Figure 1A) predisposing to the JAK2V617F mutation.5,6 The JAK2V617F mutation facilitates the acquisition of homozygous status for the JAK2V617F by mitotic homologous recombination (HR) occurring between the JAK2WT and JAK2V617F alleles, resulting in chromosome 9p uniparental disomy (9pUPD).7,8 Here we report 2 cases where high JAK2 mRNA expression was associated with a novel early step in MPN development, HR preceding JAK2 mutation.

Figure 1

Analysis of the JAK2 gene in patients Na1061 and Na1253 and proposition of a new pathogenic model for polycythemia vera. (A) Representation of the 46/1 haplotype. The 46/1 haplotype is an approximately 280 Kb-long region of chromosome 9p that includes the entire JAK2, INSL6 and INSL4 genes. (B) Schematic representation of the JAK2 gene. JAK2 exons are represented by black boxes. JAK2 SNP rs10429491 (in exon 6), rs7034539 (in intron 18) and rs2230724 (in exon 19) positions are indicated with black bars. (C) Analysis by direct sequencing of JAK2 SNPs and JAK2V617F allelic ratios in gDNA of granulocytes and CD3+ lymphocytes (used as a control, healthy cells) of PV patients Na1061 and Na1253 (see primers in supplemental Table 3 and supplemental Figure 1). Black arrows indicate the different SNPs and JAK2V617F. Both patients were heterozygous for SNPs in CD3+ lymphocytes yet had SNP rs12343867 C-allele ratios in granulocyte gDNA (80% and 100%) compatible with homologous recombination of JAK2. (D) Detailed view of the JAK2 region. Results of the distortion of SNP allelic differences showed HR of part of JAK2 (exons 6-25) for Na1061 and of the whole 46/1 haplotype for Na1253. Regions of pre-JAK2V617F homologous recombination, not readily visible unless one looks for them, are indicated by double black arrows. (E) Karyoview of chromosomal aberrations. Bars depict the physical position and size of the aberration (purple, homologous recombination; blue, uniparental disomy events). Black arrows indicate the chronology of events, as deduced from the rs12343867 and V617F allelic ratios. For both patients the distortion of SNP allelic differences because of homologous recombination was higher at the telomeric end than in the centromeric region of chromosome 9p indicating 2 distinct partial 9pUPDs for Na1061 and 1 partial 9pUPD for Na1253. For both patients, SNP allelic distortion revealed pre-JAK2 homologous recombination (in purple). (F) Main and new pathogenic models for polycythemia vera and other MPN. The current model states that MPN patients carry or acquire a predisposition to MPN and mutation in the JAK2 gene; the JAK2 GGCC haplotype is one such genetic predisposition. In other patients, another genetic abnormality, congenital or acquired, presumably in a myeloid progenitor, is responsible for clonality, growth advantage and eventually, acquisition of JAK2 mutation -V617F being the most frequent - and MPN phenotype. Because high JAK2V617F loads are usually acquired via 9pUPD and most frequent in PV, acquisition of the PV phenotype is assumed to result from 9pUPD facilitated by JAK2V617F. Both JAK2 mutation and 9pUPD may occur more than once, leading to the development of one or several JAK2V617F-homozygous subclone(s). Disease phenotype and evolution, and occurrence of 9pUPD, may vary depending on parallel genetic events (eg, TET2 mutations) and the type of JAK2 mutation (eg, high mutant loads and 9pUPD are rare in patients with JAK2 exon 12 mutations). The new model adds an early step to the conventional model, stating that subsets of patients carrying the JAK2 GGCC haplotype may be predisposed to homologous recombination (HR) of JAK2 associated with growth advantage, followed or not by mutation in the JAK2 gene on the recombined allele and high JAK2 mRNA expression. Early JAK2 HR is compatible with all of the later steps leading to MPN according to the conventional pathogenic model: JAK2 mutation, 9pUPD, acquisition of parallel events in genes other than JAK2. The new model allows that a non-identified genetic event may facilitate JAK2 recombination and subsequent genetic alterations eventually leading to PV phenotype.

Patients Na1061 and Na1253 presented with a high hematocrit, slightly elevated leukocyte counts, normal (Na1061) or elevated (Na1253) platelet counts, aquagenic pruritus, absence of splenomegaly, and presence of JAK2V617F (20.7% for Na1061, 30.0% for Na1253), and were diagnosed with PV (see supplemental Table 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). Sequencing and allele-specific qPCR analysis in granulocyte DNA of marker rs12343867 (C/T) in intron 14 of JAK2, characteristic of the 46/1 haplotype, revealed rs12343867 ratios sharply different from JAK2V617F ratios: 80% C-alleles for Na1061 and 100% T-alleles for Na1253 (Figure 1B-C). For both patients, CD3+ lymphocytes were unambiguously heterozygous for rs12343867 (Figure 1C). This indicated granulocyte acquisition of homozygosity for rs12343867 but not for the V617F mutation. In other words, the acquisition of homozygosity for rs12343867 must have preceded JAK2 mutation in these patients. This was confirmed by further analysis of JAK2 in granulocytes and CD3+ lymphocytes (Figure 1C), and of chromosome 9p using SNP arrays (Figure 1D). These studies showed that the DNA regions recombined involved JAK2 exons 6-25 for Na1061, and the complete 46/1 haplotype for Na1253. Moreover, SNP array studies revealed the presence of 1 subclone for Na1253 (28.24 Mb) or 2 subclones for Na1061 (5.7 and 24.54 Mb) with partial 9pUPD (supplemental Figures 2-3 and Figure 1E). Sequencing of the complete JAK2 cDNA excluded any mutation other than V617F.

These first cases of HR of JAK2WT led us to propose a new model for MPN: the 46/1 haplotype may predispose carriers to diverse alteration of JAK2 including early HR of wild-type JAK2, associated or not with mutation in JAK2 or other genes important for myelopoiesis, the V617F mutation facilitating additional HR involving the JAK2V617F-mutated allele, leading to 9pUPD and JAK2V617F homozygosity (Figure 1E-F). The new model allows that a nonidentified somatic genetic event may facilitate JAK2 recombination and subsequent genetic alterations eventually leading to PV phenotype (Figure 1F).

In the context of inherited gene mutations, meiotic HR can increase expression of the gene involved.9 In the case of JAK2, mitotic HR could result in a configuration that amplifies JAK2 expression and subsequently cell growth after activation of Jak2 by cytokine receptors. This is of importance because MPN progenitors produce Jak2-activating cytokines.10 For both patients, cDNA quantitative analysis revealed high JAK2 mRNA levels with > 96% JAK2V617F (see supplemental Table 2), implying an mRNA expression almost 100-fold higher for recombined alleles in V617F/V617F cells than for alleles in WT/WT cells. Finally, finding recurrent JAK2 recombination associated with high mRNA expression suggests that residual JAK2V617F disease may be best assessed in cDNA.


The online version of this article contains a data supplement.

Acknowledgments: The authors thank Dr Ariane Plet (Nantes, France), Dr Eric Lippert (Bordeaux, France), and Dr Richard Redon (Nantes, France) for reading the manuscript.

This study was performed thanks to grants from the Association pour la Recherche contre le Cancer (ARC) and the Comités Morbihan and Ille-et-Vilaine of the Ligue Nationale contre le Cancer to S.H. and the MPN Research Foundation to R.K. M.V. is recipient of a scholarship from the French Ministry of Research (2009-2012) and benefited from a scholarship for short term scientific missions (November 2010) from MPN & MPNr-EuroNet (COST Action BM0902). M.V., J.B., and S.H. are members of MPN & MPNr-EuroNet.

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

Contribution: S.H. designed the research, analyzed data, and wrote the paper; R.K. designed the research and analyzed data; M.V. performed research, analyzed data, and wrote the paper; D.O., A.H., and J.B. performed research and analyzed data; M.T. and J.-F.R. contributed patient samples and clinical data; and J.-M.C. contributed with scientific and technical advice and helped write the paper.

Correspondence: Sylvie Hermouet, Inserm U892, Institut de Recherche Thérapeutique, Université de Nantes, 8 quai Moncousu, 44007 Nantes cedex 1, France; e-mail: sylvie.hermouet{at}