Polycythemia vera (PV), one of the myeloproliferative neoplasms (MPNs), is characterized by the predominant expansion of morphologically normal erythroid cells. Most, if not all, patients with PV have acquired mutations in Janus kinase 2 (JAK2) [1], a cytoplasmic tyrosine kinase that is constitutively associated with cytokine receptors that lack intrinsic tyrosine kinase activity. The majority of patients have a JAK2V617F mutation, the result of a single base change within JAK2 exon 14 [2-5]. This substitution occurs in the pseudokinase (or JH2) domain, results in constitutive JAK2 activation in vitro and in vivo, and recapitulates many aspects of PV biology in several different transgenic mouse model systems [6-10]. JAK2V617F is also closely associated with the erythropoietin-hypersensitive expansion of erythroid progenitor cells that is a hallmark feature of the MPNs [11,12].

JAK2V617F-negative PV patients are instead positive for one of a series of JAK2 mutations that are collectively referred to as the JAK2 exon 12 mutations by virtue of their genomic location [13,14]. There have been at least 37 exon 12-encoded mutations identified to date; these all affect a stretch of 12 amino acids located immediately adjacent to JH2 domain. Despite having many functional similarities in vitro, the JAK2V617F and JAK2 exon 12 mutations are associated with distinct hematologic phenotypes in vivo. Whereas patients with JAK2V617Fpositive PV characteristically have tri-lineage involvement, JAK2 exon 12 mutation-positive patients tend to present with an isolated erythrocytosis [13,14]; most individuals have normal leukocyte and platelet counts, although approximately a fifth have elevated counts of either lineage (defined here as greater than 10×10⁶/L and 400×10⁶/L, respectively). Bone marrow samples from JAK2 exon 12 mutationpositive patients typically lack the prominent clusters of large, bizarrelooking megakaryocytes that characterize JAK2V617F-positive PV, although small clusters of two to four megakaryocytes, some of which may appear atypical, can be identified [15,16]. However, the effects on megakaryopoiesis associated with the presence of a JAK2 exon 12 mutation have not been examined. Here, in vitro assays as well as patient samples and haematologic data were used to determine the effects of a JAK2 exon 12 mutation on the viability and proliferation of megakaryocytic cells. Interestingly, evidence is uncovered that suggests that the megakaryocytic atypical observed in a proportion of mutationpositive patients is not cell-autonomous; that is, it is driven by other hematopoietic or non-hematopoietic elements within the bone marrow environment.

Mutation screening in platelets

Granulocyte and platelet cell populations were prepared from individual MPN patients on the same day. First, platelet-rich plasma was isolated from fresh whole blood samples by three cycles of mild centrifugation (150 g for 20 minutes). Platelets were then pelleted by centrifugation at 1500 g for 10 minutes. After the removal of platelet-rich plasma, blood samples were further fractionated using a Histopaque-1077 density gradient (Sigma-Aldrich; St Louis, MO). The mononuclear cell population was discarded, and red cells removed by lysis with ammonium chloride. Cell purities were confirmed using a Woodley ABC automated cell counter; only those samples with > 95% purity were used for subsequent analysis. Granulocyte and platelet RNAs were prepared immediately using Triazol reagent (Sigma-Aldrich) according to the manufacturer’s recommendations. RNA samples were then used as the template for first-strand DNA synthesis using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Invitrogen; Grand Island, NY). Mutation screening was performed using primers complementary to sequences in JAK2 exons 11 and 13 (5’- CAACCTCAGTGGGACAAAGAA-3’ and 5’-TGTTTCATGCAGTTGACCGTA-3’, respectively). The resulting 343 base pair PCR product was directly sequenced.

Mutation screening in megakaryocytes

Deparaffinization of archived bone marrow trephine biopsy sections was performed by sequential 15 minute washes with xylene, followed by sample rehydration with decreasing concentrations of ethanol (95%, 85%, and 70%). Immediately after deparaffinization, slides were stained with hematoxylin and eosin using standard methodologies. To further facilitate their identification, megakaryocytes were highlighted by immuno-histochemical staining with an antibody against factor VIII (MS-722-P0, LabVision; Fremont, CA). Microdissection was performed using a P.A.L.M. laser capture microscope (P.A.L.M. Microlaser Technologies AG; Bernried, Germany). Fourteen atypical megakaryocytes were removed for genotype analysis. Genomic DNA was extracted from individual laser-captured cells using the QIAamp DNA micro kit (Qiagen; Valenica, CA) according to the manufacturer’s instructions. Purified DNA was first amplified using primers that flank JAK2 exon 12 (forward: 5’-CTCCTCTTTGGAGCAATTCA-3’; reverse: 5’-GAGAACTTGGGAGTTGCGATA-3’), and then with nested primers (forward: 5’-AATGGTGTTTCTGATGTACC-3’; reverse: 5’-AGACA-GTAATGAGTATCTAATGAC-3’), as described previously (17). The resulting 126 base pair amplicon was directly sequenced.

Retroviral production and 32D cl3 transduction

32D cl3 cells were maintained in RPMI-1640 containing 10% fetal calf serum (FCS) and 1 ng/mL murine interleukin-3 (IL3). The Jak2 mutations were introduced into an MSCV-Jak2-iresGFP vector by site-directed mutagenesis [17], transfected into 293-T kidney epithelial cells, and culture supernatants containing recombinant retroviral particles were produced as described previously [14,18]. 32D cl3 cells were first incubated with MSCV-PGKneo-EpoR or MSCV-PGKneo-TpoR retroviral supernatent, and then expanded in geneticin for a week. 32D/EpoR and 32D/TpoR cells were then each transduced with retroviruses encoding either wildtype or mutant Jak2, and the resulting green fluorescent protein (GFP)-positive populations purified by fluorescence-activated cell sorting (FACS) four days later [14,18]. Assays of cytokine-independence were performed essentially as previously described for BaF3 cells [14,18].

One possible explanation for the normal or near-normal platelet counts at presentation seen in most patients with a JAK2 exon 12 mutation could be that these confer a slight proliferative advantage to affected megakaryocytes or their precursors that only manifests over a long period. Those patients with elevated platelet counts at diagnosis would therefore have a more advanced disease than those with normal platelet counts. To address this possibility, two JAK2 exon 12 mutationpositive PV patients, for whom cell counts over a long period of time were available, were identified: Patient A had been treated for a decade, and Patient B for just over eight years (Figure 1). At diagnosis, each patient had a hemoglobin (Hb) concentration sufficient to meet the WHO 2008 major diagnostic criteria [19]: that is, 201g/L and 174 g/L, respectively. Hb levels decreased relatively rapidly after venesection was initiated. Over the treatment period, leukocyte numbers remained elevated in Patient B, whereas they rose gradually over time in Patient A; a molecular explanation for this increase was not readily apparent. However, platelet numbers remained static in both cases, not rising above a threshold of 400×10⁶/L. Platelet counts did not increase appreciably in a third JAK2 exon 12 mutation-positive patient over a contiguous 15-year period; however, the data from this patient is not included in Figure 1, as his platelet counts for the first 40 months following diagnosis were not available. Taken together, these clinical data suggest that, in contrast to JAK2V617F, JAK2 exon 12 mutations may not confer a proliferative advantage to affected megakaryocytic cells.

Figure 1: Platelet counts of patients with a JAK2 exon 12 mutation do not increase significantly over time.

Another explanation for the lack of an apparent platelet phenotype may be that the JAK2 exon 12mutant proteins are unable to interact with and/or mediate intracellular signalling through the TpoR. To investigate this possibility in vitro, the Jak2K539L mutation was expressed in the 32D cl3 murine myeloid cell-line, and tested for its ability to confer IL3-independent proliferation in the presence of the thrombopoietin or erythropoietin receptors (TpoR and EpoR, respectively). This mutant Jak2 was previously shown to enable cytokine-independence to murine lymphoid cells (BaF3) expressing EpoR [14]. In the absence of exogenous IL3, 32D/TpoR cells expressing Jak2K539L proliferated at a comparable rate to Jak2K539L-positive 32D/EpoR cells, and to either cell-line expressing Jak2V617F (Figure 2A). Cells that expressed wild type Jak2 failed to proliferate significantly, as shown previously [3-5,14]. To alleviate any concerns that the relatively abundant level of Jak2 in these cell-lines (indicated by the level of co-expressed GFP) was saturating, 32D/EpoR and 32D/TpoR cells that express lower Jak2 levels were engineered, using less retroviral supernatent to transduce cells and selecting those with low GFP expression. Again, 32D/TpoR cells expressing Jak2K539L or Jak2V617F continued to proliferate robustly after cytokine withdrawal (Figure 2B). Interestingly, the reduction in Jak2 expression levels had a more profound effect in 32D/EpoR cells than in 32D/TpoR cells, with net proliferation rates being lower in the former. Taken together, our in vitro studies suggest that Jak2K539L is able to associate with the TpoR in vivo, and that the normal platelet counts in patients with a JAK2 exon 12 mutation might not reflect an absence of JAK2-mediated signalling in megakaryocytic cells.

Figure 2: Jak2K539L expression in 32D cells co-expressing the thrombopoietin receptor enables proliferation in the absence of exogenous cytokine.

A third potential explanation for the lack of thrombocytosis in JAK2 exon 12 mutation-positive patients may be that the mutant protein has deleterious effects on megakaryocyte viability or differentiation, or on pro-platelet formation. To address this possibility, the level of mutant JAK2 transcript was analyzed in the granulocyte and platelet cell populations present within single individuals. Peripheral blood samples were taken from three JAK2 exon 12 mutation-positive PV patients and a single JAK2V617F-positive PV patient, and these populations were purified and analyzed immediately; the results obtained are depicted in figure 3. Irrespective of the nature of the JAK2 exon 12 mutation, a significant proportion of platelets in each individual expressed the mutant JAK2 transcript. Whilst Sanger sequencing may not be generally considered to be an accurate method for measuring mutation burdens, comparison to the level of mutant transcript in granulocytes from the same patient strongly suggests that platelet viability is not significantly impacted by the presence of a JAK2 exon 12 mutation.

Figure 3: Detection of mutant JAK2 transcripts in the platelets of three patients suggests that this mutation does not significantly compromise megakaryocytic viability. Peripheral blood samples were taken from three JAK2 exon 12 mutation-positive PV patients (JAK2F537-K539delinsL, JAK2H538QK539L or JAK2E543- D544del) and a single JAK2V617F-positive PV patient, granulocytes and platelets were simultaneously purified, and the level of mutant JAK2 transcript assessed in these populations by direct sequencing. The locations of base substitutions (marked by arrowheads) or deletions (marked by diamonds) are provided to highlight the relevant mutation within JAK2 exon 12.

The JAK2 exon 12 mutation status of bone marrow-derived megakaryocytes was also assessed. Examination of trephine sections taken from five patients with a JAK2 exon 12 mutation revealed clusters of two to four megakaryocytes, many of which had atypical nuclear morphology, could be identified in one case (Figure 4A). Individual megakaryocytes were purified by laser capture microdissection, and their JAK2 mutation status determined. Surprisingly, the genotyping of ten individual megakaryocytes from three separate clusters revealed that only four of these cells were positive for the JAK2N542-E543del mutation; typical DNA sequences are shown in figure 4B.

Figure 4: Megakaryocytes with hyperlobulated nuclei may be present in patients with a JAK2 exon 12 mutation, yet not be positive for the mutant allele. (A) Clusters of two or three megakaryocytes, many of which had atypical nuclear morphology, were identified in the bone marrow sample of a patient with a JAK2N542-E543del mutation. (B) Individual megakaryocytes were purified by laser capture microdissection, and genotyped. Only a proportion of the cells within a single cluster were positive for the JAK2 exon 12 mutation. The location of the six deleted bases (marked by diamonds) is provided to highlight the relevant mutation within JAK2 exon 12.

This observation suggests that megakaryocyte atypia may not be a cell-intrinsic phenomenon in MPN patients, but rather a response to other factors within the bone marrow microenvironment.

In this study, we sought to more fully understand the effects of a JAK2 exon 12 mutation on megakaryopoiesis. These mutations were found to be present within the platelet population of affected individuals at levels comparable to those in paired granulocyte samples, suggesting that they do not profoundly affect the viability of platelets or their precursors. Nevertheless, most mutation-positive patients present with platelet counts in the normal range [13,14,16], and numbers fail to rise substantially over time. Cytokine withdrawal assays also demonstrated that Jak2K539L is able to interact with, and mediate intracellular signalling through, the TpoR in vitro. It therefore remains unclear why patients with a JAK2 exon 12 mutation do not have the elevated platelet counts typically observed in patients with JAK2V617Fpositive PV. However, in performing these analyses, an interesting phenomenon was identified: only a subset of those megakaryocytes with an atypical nuclear structure were positive for a JAK2 exon 12 mutation. This observation suggests that cell types within the bone marrow environment of MPN patients may influence the phenotype of cells that lack a JAK2 mutation.

Several different mouse models of human hematologic malignancies suggest that changes within the bone marrow microenvironment may impact significantly upon myelopoiesis [20-22]. In one instance, deletion of the Dicer microRNA processing enzyme specifically within osteogenic cells resulted in myelodyslasia, with mice having anemia, leukopenia and thrombocytopenia, along with marked dyslasia in the granulocytic and megakaryocytic lineages [20]. The cell-extrinsic nature of this defect was confirmed when wildtype bone marrow cells were transplanted into Dicer-deficient recipients, recapitulating the myelodysplastic phenotype. Further insights into the underlying mechanism behind this phenotype were obtained when gene set enrichment analysis revealed significant downregulation of the Shwachman-Diamond-Bodian Syndrome gene (Sbds). In humans, inactivating SBDS mutations are associated with a bone marrow failure syndome that includes myelodysplastic changes [23]. Targeted Sbds deletion in murine osteoprogenitor cells also resulted in leukopenia and dysplastic changes in granulocytic and megakaryocytic cells [20].

Although more frequently associated with cell-intrinsic defects, myeloproliferation can also arise as a consequence of changes within the bone marrow environment. Modulation of the number or function of osteoblastic cells extrinsically influences hematopoiesis by altering the hematopoietic stem cell (HSC) niche [24-26]. In other instances, the cell type within the microenvironment that drives myeloproliferation is not entirely clear, but likely includes both stromal and endothelial elements and even other hematopoietic cells [21,27]. Individual proteins can also positively or negatively affect the number of HSCs present within an organism: the loss of angiopoietin, osteopontin or retinoic acid receptor-γ expression results in marked myeloproliferation in vivo [28-30], whereas over-expression of Jagged-1 or HoxB4 has a similar effect in vitro [31,32].

To date, there has been little evidence of a non-cell-autonomous effect in patients with an MPN. Of particular note, however, is a study in which immunocompromised mice were transplanted with human cord blood cells expressing TEL-JAK2 [22]. This fusion protein is the consequence of a translocation between chromosomes 9 and 12 that can be observed in rare leukemic patients [33,34], confers cytokine independence to BaF3 cells in vitro, and produces an acute leukemia in transgenic animals [35]. Mice xenotransplanted with TEL-JAK2- positive human cells rapidly developed myelofibrosis, with their bone marrow containing increased numbers of atypical megakaryocytes. These had hyperlobulated nuclei, were distributed in clusters and, surprisingly, were murine in origin. This finding demonstrates that the formation of megakaryocyte atypia may be a reactive event, and lends support to the hypothesis that the megakaryocytic hyperplasia seen in MPN patients is not cell-intrinsic. Further studies will be required to more clearly investigate this possibility.

Dr Wei Tong (Children’s Hospital of Philadelphia, PA) kindly provided the MSCV-PGKneo-TpoR and MSCV-PGKneo-EpoR retroviral vectors used here.