JAK2 and genomic instability in the myeloproliferative neoplasms: a case of the chicken or the egg?

Abstract

The myeloproliferative neoplasms (MPNs) are a particularly useful model for studying mutation accumulation in neoplastic and the mechanisms of the molecular cells, understanding underlying defects our current This review summarizes acquisition. present their in patients with an MPN, and the effects of mutations targeting Janus kinase 2 (JAK2)-mediated intracellular signaling on DNA damage, and on the elimination of mutation-bearing cells by programmed cell death. Moreover, we discuss findings that suggest that the acquisition of disease-initiating mutations in hematopoietic stem cells of some MPN patients may be the consequence of an inherent genomic instability that was not previously appreciated.

You jest about what you suppose to be a triviality, in asking whether the hen came first from an egg or the egg from a hen, but the point should be regarded as one of importance, one worthy of discussion, and careful discussion at that.

- Ambrosius Theodosius Macrobius (AD395-423), philosopher

Disease progression and metastatic growth in human malignancies have long been thought to involve the Darwinian selection of mutant sub-clones, such that cells with the best evolutionary fitness will over time come to dominate the neoplastic population (1, 2). This selection process is facilitated by an increase in genomic instability that results in the accumulation of additional mutations. Some of these, the so-called “driver mutations”, are directly responsible for any competitive advantage, whereas “passenger mutations” contribute nothing to the oncogenic potential of affected cells. Next-generation sequencing studies are just beginning to reveal the identity of the genetic changes that separate the predominant cells within a primary tumor from their metastatic progeny (3-6). In contrast, little is known about the genomic changes driving the earliest stages of tumor evolution. Does this process involve a linear stepwise acquisition of mutations, as suggested by Nowell (1)? Does the malignant cell-of-origin instead give rise to a diverse set of sub-clones that continue to evolve in parallel, a scenario apparent in cases of acute lymphoblastic leukemia (ALL) (7, 8)? Or, in some instances, is a single catastrophic event that generates dozens of mutations sufficient to produce a malignancy (9)?

A particularly useful model for examining the pattern of mutation accumulation in neoplastic cells, and the mechanisms underlying their acquisition, are the myeloproliferative neoplasms (MPNs). These are a group of related hematologic disorders that includes polycythemia vera (PV), essential thrombocythemia (ET), primary myelofibrosis (MF) and chronic myeloid leukemia (CML), and several more rare entities, such as systemic mastocytosis and chronic neutrophilic leukemia (10, 11). The MPNs are characterized by preferential expansion of one or more morphologically normal myeloid cell types. Several aspects of MPN biology make them particularly useful in the investigation of the molecular mechanisms underlying clonal evolution. In particular, with appropriate disease management, the life expectancy of patients with ET or PV is near normal (12), permitting repeat biological sampling over many years. Sample taking can also be achieved with ease; a small aliquot of peripheral blood is sufficient to provide cells for most analyses. Finally, the ability to culture circulating hematopoietic progenitors in vitro facilitates the genetic analysis of mutations at the single-cell level, since all cells within a resulting colony are derived from the initial progenitor cell that was seeded, and can be readily retrieved from the methylcellulose in which they were propagated for subsequent analysis. Many meaningful insights into MPN biology have been obtained using this strategy (13-17).

Comprehensive reviews relating to the genomic instability associated with CML have been published (18-20), but the genomic instability associated with the Philadelphia chromosome (Ph)-negative MPNs, ET, PV and MF, has not been discussed in detail. Here, we summarize the current understanding of the molecular defects underlying these disorders, and the effects of mutations targeting Janus kinase 2 (JAK2)-mediated intracellular signaling on DNA damage and repair and on the elimination of mutation-bearing cells by apoptosis. Moreover, we discuss recent findings that suggest that acquisition of the disease-initiating mutation in MPN patients may itself result from an inherent genomic instability that had previously not been appreciated.

The molecular genetics underlying development of a human MPN

The MPNs have long been considered to arise from mutations acquired within a hematopoietic stem cell (HSC). This concept was initially based upon studies of the isoform expression patterns of polypeptides encoded by the X chromosome-encoded glucose-6-phosphate dehydrogenase locus (G6PD) in informative female patients (21, 22), which demonstrated that cells of the malignant clone all had the same X-inactivation pattern, and were likely to arise from a common HSC. Recent analyses, following the identification of most of the molecular lesions associated with the MPNs, confirm that these mutations can be detected in primitive hematopoietic stem or progenitor cells obtained from patients (23-26).

Little was known about the molecular pathogenesis of the Ph-negative MPNs until seven years ago, when the presence of a one-nucleotide change within JAK2 exon 14 was identified in the majority of patients (13, 27-29). As a consequence, the valine residue located at position 617 of JAK2 is substituted with a phenylalanine (“V617F”). This change is neither a single nucleotide polymorphism (SNP) nor an inherited mutation, since it can be detected in patient granulocytes but not matched T-lymphocytes, which are not part of the malignant clone in the chronic phase of these disorders. The JAK2V617F mutant allele is present in 50-60% of ET and MF patients and in 95% of PV patients, but occurs only rarely in other myeloid disorders (30-32). Theoretical models of JAK2 structure predict that the V617F substitution induces a conformational shift that alleviates repressive interactions between its JH1 and JH2 (or kinase and pseudokinase) domains (33). In support of this hypothesis, the JAK2V617F mutation is present in cells of the erythropoietin (EPO)-independent erythroid colonies that are a hallmark feature of these disorders (13, 14). JAK2V617F expression also confers factor independence to cytokine-dependent cell-lines (27-29), and results in constitutive JAK2 phosphorylation and consequent activation of the STAT, AKT and ERK signaling pathways. JAK2V617F expression in retroviral transduction/transplantation models and in transgenic mice recapitulates most aspects of human ET and/or PV (27, 34, 35), strongly arguing that it alone is sufficient to produce an MPN phenotype.

Ph-negative MPN patients lacking the JAK2V617F mutation may nonetheless bear mutations that result in deregulated JAK2 signaling. The vast majority of JAK2V617F-negative PV patients have alternative JAK2 mutations (36-39), which consist of a wide variety of deletions, insertions and/or amino acid substitutions that target residues located adjacent to the JH2 domain (40). These “exon 12” mutations constitutively activate JAK2-mediated signaling in vitro and in vivo (36). About 10% of JAK2V617F-negative ET and MF patients instead carry activating mutations in MPL, the receptor for thrombopoietin (TPO), which utilizes JAK2 for intracellular signaling (41-43). These predominantly target residue 515, which is part of a juxta-membrane amphipathic motif that acts to maintain MPL in a constitutively inactive configuration (44); others mutations target residue 505, located in the trans-membrane domain, resulting in TPO-independent receptor dimerization (45). Occasional JAK2V617F-negative MPN patients have mutations in SH2B3 (also called LNK; (46)), which negatively regulates JAK2 signaling by recruiting phosphatases to the activated JAK2. Loss-of-function LNK mutations therefore augment and sustain JAK2 signaling. A third of ET and MF patients do not have mutations in JAK2, MPL or SH2B3; the molecular pathogenesis of these cases remains unclear. The observation in pre-clinical studies of ATP-competitive JAK2 inhibitors that these agents inhibit the in vitro viability of cells from JAK2V617F-positive and JAK2V617F-negative patients suggests that the unidentified mutations associated with ET and MF likely occur in genes involved in the regulation of JAK/STAT signaling (47, 48). Candidate gene sequencing has thus far excluded JAK1, JAK3, TYK2, STAT5A or STAT5B as frequent targets for mutation (31).

Other acquired mutations associated with an MPN: a link to epigenetic regulation

In the last three years, inactivating mutations in the TET2, ASXL1, DNMT3A or EZH2 genes have been detected in a subset of patients with an MPN (49-52). Strikingly, these mutations may be detected in patients with other myeloid malignancies, including myelodysplasia (MDS), chronic myelomonocytic leukemia (CMML) or de novo acute myeloid leukemia (AML) (53-56). TET2, ASXL1, DNMT3A and EZH2 are all involved in the epigenetic regulation of gene transcription, although the role of these proteins in MPN disease initiation or progression has not been fully elucidated. Conditional knockout mouse strains have nonetheless provided intriguing insights into the consequences of absent DNMT3A or TET2 expression (57-59). In particular, Dnmt3a-null, Tet2-null and Tet2-haploinsufficient HSCs have enhanced self-renewal activity in vivo, suggesting that inactivating TET2 or DNMT3A mutations acquired by patients provide affected cells of the malignant clone with a competitive advantage over those without the mutation. In contrast, Asxl1-deficient murine HSCs are unable to outcompete their wild-type equivalents in competitive reconstitution assays (60); additional studies will be required to elucidate the role of this protein in normal and malignant hematopoiesis.

MPN disease progression in vivo: a pattern of mutation accumulation over time

The first evidence that abnormalities in DNA repair plays an important role in the progression of the Ph chromosome-negative MPNs was provided by the description of acquired uniparental disomy of chromosome 9p in a relatively high proportion (~30%) of PV patients (61). With the discovery of the JAK2V617F mutation several years later, it soon became apparent that most JAK2V617F-positive PV patients were homozygous for this mutation: when the genotype of individual hematopoietic colonies was determined, all those patients with PV were found to have a subset of cells with biallelic JAK2V617F mutations (14). As this homozygosity could be detected in erythroid and granulocytic cells from the same patient, it is likely that acquisition of the second JAK2V617F mutation occurred in a JAK2V617F-positive early myeloid progenitor, if not in an HSC. In contrast, only wildtype and JAK2V617F-heterozygous cells were detected in patients with ET, although JAK2V617F-homozygous cells were noted in two such patients that had subsequently experienced polycythemic transformation (14). Collectively, these data support the proposal that JAK2V617F-positive ET and PV form a disease continuum (62), with duplication of the mutant JAK2 allele driving disease transformation from thrombocythemia to polycythemia. Murine models of JAK2V617F-positive MPNs, in which the copy number (35) or the expression level (63) of the JAK2V617F transgene was shown to dictate the phenotype observed, provide experimental support for this “dosage hypothesis”.

On the basis of these and other studies, a simple model for disease initiation and progression in the Ph-negative MPNs might be proposed (Figure 1). Acquisition of a single JAK2V617F mutation in an HSC leads to the development of a phenotype consistent with a diagnosis of ET. Duplication of the mutant JAK2 allele results in a polycythemic phenotype in the vast majority of cases. Rare JAK2V617F-positive PV patients may have monoallelic JAK2 mutations, where the compound mutant allele (for example, the V615LV617F allele described in a later section) is a stronger gain-of-function mutation than JAK2V617F alone. Heterozygous mutations within JAK2 exon 12 are also sufficient to generate a polycythemic phenotype, whereas patients with an MPL mutation present with ET. Whilst homozygous JAK2 exon 12 or MPL mutations occur in rare instances (42, 64), there is no indication that mutation homozygosity profoundly changes the disease phenotype. Acquisition of additional mutations would initiate transformation from ET or PV to MF (Figure 1). Inactivating mutations that target ASXL1, TET2, EZH2 or DNMT3A are unlikely to underlie this process, as they can be readily detected in MPN patients that have not developed myelofibrosis (49-51). Several groups are currently using whole genome or exome sequence analysis to identify the mutations associated with myelofibrotic transformation.

Figure 1. Mutation accumulation in the MPNs.

A simple model for disease initiation and progression to myelofibrosis in the Ph-negative MPNs can be proposed. Patients diagnosed with essential thrombocythemia (ET) may have either a single acquired JAK2V617F (VF) allele, an MPLS505N, MPLW515L or MPLW515K mutation (MPL), or one or more mutation(s) affecting unknown targets (collectively referred to here as an X). Duplication of the JAK2V617F allele or the acquisition of additional mutations within the affected JAK2 allele (VF*) would instead produce a polycythemic phenotype, as would the acquisition of a mono-allelic JAK2 exon 12 (ex12) mutation. The acquisition over time of one or more “driver” mutations (Y), the identities of which are currently uncertain, would initiate the transformation to myelofibrosis.

JAK2V617F expression is associated with increased DNA damage in vitro and in vivo

The pattern of mutation accumulation over time observed in MPN patients raises the possibility that these individuals may have defective DNA damage repair; several in vitro and in vivo studies support this hypothesis. JAK2V617F expression was found to appreciably enhance the frequency of spontaneous homologous recombination (HR) events in Ba/F3 pro-B cells, with a four-fold increase in incidence compared to that in cells engineered to express wildtype JAK2 (65). This was accompanied by an increased level of nuclear RAD51, a component of the DNA repair foci that form after the induction of double-strand breaks (66, 67). Additional indicators of genomic instability, including increased sister chromatid exchange, increased spontaneous mutation at the HPRT locus, changes in DNA ploidy, centromere abnormalities and increased DNA repair by single-strand annealing, are apparent in JAK2V617F-expressing Ba/F3 cells (65, 68). In contrast, JAK2V617F had little effect on non-homologous end joining (NHEJ) when assayed in cell-free extracts or intact cells (65). Changes in RAD51 distribution and HR frequency similar to that observed in vitro have also been noted in hematopoietic cells obtained from patients with JAK2V617F-positive PV or MF (65).

In a JAK2V617F-heterozygous conditional knock-in mouse, in which animals develop an ET-like disease soon after JAK2V617F expression is initiated, the HSC compartment appears compromised: six months after the induction of mutant JAK2 expression, mice had a two-fold reduction in the absolute number of lineage-negative, Sca1-positive, c-Kit-positive (LSK) cells compared to controls (69). Competitive reconstitution assays also indicated that these LSK cells were impaired in their long-term repopulating ability. To explain this phenomenon, investigators evaluated the level of DSBs in the total bone marrow population, as well as in the purified LSK fraction, from control and JAK2V617F knock-in animals. The intensity of γ-H2AX staining was significantly elevated in both cell populations obtained from the latter group of mice. Despite an accumulation of DNA damage, the proportion of JAK2V617F-positive LSK cells undergoing apoptosis was lower. However, in another JAK2V617F knock-in mouse model (70), quantitative or qualitative differences between the LSK compartments of JAK2V617F-heterozygous and wildtype littermates were not noted. This discrepancy might simply be attributable to subtle strain differences between the two models, but may also reflect the age of the animals analyzed in the later study (which was not specified). Interestingly, Li and colleagues detected differences in LSK frequency six months after JAK2V617F expression was induced, but these were not apparent five months earlier, suggesting any alteration in stem cell function might be subtle and only become apparent over time. Additional studies will be required to fully resolve this issue.

The molecular basis for the increase in HR events observed in JAK2V617F-expressing cells remains uncertain. A likely possibility is the STAT activation-dependent generation of reactive oxygen species (ROS), which can lead to DSB formation. Endogenous ROS levels are increased in many myeloid malignancies (71), including CML, although levels have not yet been evaluated in cells expressing JAK2V617F. Another possibility involves a non-canonical JAK signaling pathway revealed by studies of modifiers of Drosophila JAK activity (72). It has been long recognized that the JAK proteins are not located solely within the cytoplasm (73, 74), although little was known about their function(s) in the nucleus. Activated Hopscotch (Hop), the single JAK protein in flies, was recently found to disrupt gene silencing by displacing heterochromatin protein-1 (Hp1; (72)). This in turn induces the aberrant expression of otherwise silenced genes, several of which are likely to contribute to a malignant phenotype, since loss-of-function Hp1 mutants enhance the leukemic potential of the Hop mutants (72). Nuclear JAK2 is present in mammalian hematopoietic cell-lines and primary CD34⁺ progenitors, but is absent from mature blood cells (75). In mammalian nuclei, JAK2 directly phosphorylates histone H3 on tyrosine-41, abolishing the ability of HP1α to bind to this residue, thereby altering the chromatin structure surrounding the promoters of genes, including those of LMO2, NANOG and MYC (76-78). In addition to repressing heterochromatic gene transcription, HP1α has important roles in telomere structure and function, chromosome segregation during mitosis, the cellular response to DSBs, and the maintenance of chromatin in a state that minimizes the rates of HR and spontaneous gene conversion (79-82). Displacement of HP1α from regions of heterochromatin by activated JAK2 might therefore contribute to the effects associated with JAK2V617F expression in vitro.

JAK2V617F expression is likely to induce a “mutator phenotype” (83), with cells acquiring many additional genetic lesions, a small proportion of which may confer an in vivo advantage. Targets might include L3MBTL1, located within the region of chromosome 20q that is deleted in a small proportion of patients (84). L3MBTL1 itself plays an important role in DNA replication and genome stability: L3MBTL1-depleted cells have multiple DSBs and an increased number of repair foci that contain γH2AX, and also activate their ATM/ATR-dependent DNA damage response pathway (85). Loss of a L3MBTL1 allele might therefore enhance genomic instability in JAK2V617F-positive cells, accelerating the rate at which additional mutations occur.

JAK2V617F and BCR-ABL both inhibit apoptosis induced in response to DNA damage

A feature of cellular transformation is the prevention of DNA damage-induced apoptosis by inappropriate regulation of cell survival pathways (2). One such pathway involves modification of the anti-apoptotic factor, Bcl-x_L. In normal cells, the rapid conversion of asparagine residues within Bcl-x_L into isoaspartic acid occurs in response to DNA damage, disrupting its ability to sequester pro-apoptotic factors, and promoting cell death (86). This non-enzymatic conversion occurs through enhanced expression of NHE-1, a Na⁺/H⁺ ion exchange protein, thereby elevating the intra-cellular pH and facilitating Bcl-x_L deamidation (87). However, intracellular NHE-1 levels do not increase in granulocytes from patients with Ph-positive CML or JAK2-mutated PV after genotoxic stress or γ-irradiation; there is no change in the levels of deamidated Bcl-x_L, and programmed cell death is not apparent (88). Enforced alkalinization or NHE-1 over-expression reverses this phenomenon, restoring to normal the levels of Bcl-x_L deamidation and apoptosis. Inhibition of this pathway was found to be neither a general feature of malignancy nor a property specific to all constitutively-activate tyrosine kinases: several hematopoietic cell-lines expressing mutant tyrosine kinases responded to induced DNA damage by increasing their intracellular pH and levels of deamidated Bcl-x_L, whereas HEL (JAK2V617F-positive) and K562 (BCR/ABL-positive) cells did not (88). The kinase activity of JAK2 or ABL appears to be required to inhibit Bcl-x_L deamidation, since treatment of granulocytes from PV patients with JAK2 inhibitors, or CML blast cells with imatinib mesylate, reversed the block on DNA damage-induced apoptosis. Investigators also found that, in the blast cells of a patient carrying an E255V BCR/ABL kinase domain mutation (which renders cells insensitive to imatinib), enforced alkalinization or NHE-1 expression readily overcame this drug resistance.

The finding that in vitro expression of BCR/ABL or JAK2V617F increases the incidence of DNA damage, and that primary cells expressing either protein subsequently fail to undergo DNA damage-induced apoptosis, together suggest that these MPN-associated mutant proteins provide a two-pronged advantage to affected cells (Figure 2): over time, subsets of daughter cells will be generated that carry additional mutations, some of which will provide an in vivo advantage over others, facilitating disease progression.

Figure 2. MPN-associated mutant kinases block normal induction of apoptosis after DNA damage.

In a healthy cell (depicted on the left), the presence of DSBs induces increased NHE-1 expression, by a mechanism that remains to be elucidated. The resulting ion flux via this antiport protein increases the intracellular pH, leading to the deamidation of Bcl-x_L and programmed cell death. In contrast, BCR/ABL or JAK2V617F expression in a mutated cell (depicted on the right) results in nuclear translocation of activated JAK2, phosphorylation of histone H3 on tyrosine-41, and the displacement of HP1α from heterochromatin. This, combined with the generation of reactive oxygen species (ROS) by activated JAK2, is likely to result in an increase in number of DSBs, and inhibition of apoptosis by the prevention of NHE-1 up-regulation, noted in vitro and in vivo. As a consequence of these two alterations, MPN cells would accumulate more mutations, some of which may provide an in vivo proliferative advantage.

Cases of co-existent mutations suggest HSCs in MPN patients are predisposed to mutation

It is becoming increasingly apparent from the study of JAK2V617F-positive patients that not only are cells bearing this mutation more likely to acquire additional mutations as the result of genomic instability associated with deregulated JAK2, but also that there may be inherent genomic instability in the HSCs of a proportion of the individuals that eventually acquire a JAK2V617F mutation. Several of the more important findings on this intriguing possibility are discussed in the following sections.

(1) The JAK2 locus in MPN patients is more prone to secondary mutations than might be expected

Genotyping of granulocyte DNA for the JAK2V617F mutation and a SNP present in JAK2 intron 14 revealed that this mutation had been acquired on more than one occasion in three of the 109 informative MPN cases studied (15), a frequency significantly higher than might be expected given the incidence of JAK2V617F acquisition in the general population. A similar analysis was performed using patients grouped by diagnosis (89); the JAK2V617F mutation was found to have occurred on multiple occasions in patients with ET, but not PV, indicating that there may be a disorder-specific pattern of JAK2V617F mutation. In either study, genotyping was not performed at the single-cell level (that is, using individual hematopoietic colonies), so it is unclear whether the JAK2 mutations were acquired in separate HSCs, resulting in independent JAK2V617F-heterozygous clones, or within a single HSC, producing a JAK2V617F-homozygous sub-clone, which is consistent with a diagnosis of PV. JAK2 point mutations may nevertheless be acquired on multiple occasions in patients with PV. We have identified a patient who acquired a second JAK2 mutation 5 years after presenting with JAK2V617F-positive PV (LMS, manuscript in preparation), producing a compound JAK2V615LV617F mutant allele that allowed affected cells to out-compete wildtype and JAK2V617F-positive cells over time. Colony analysis revealed a mutation pattern consistent with at least three separate genetic events: acquisition of the JAK2V617F mutation in a normal HSC, addition of a JAK2V615L mutation to the JAK2V617F allele, and the generation of JAK2V615LV617F-homozygous cells by HR.

Collectively, these studies suggest a fundamental difference in the nature of secondary JAK2 mutations that occur in JAK2V617F-positive MPN patients. One group is defined by mutations that occur in a JAK2V617F-heterozygous cell, including HR, duplication of the entire mutated chromosome 9 (with or without loss of the wildtype equivalent), or acquisition of additional base substitutions in the mutant JAK2 allele. Each event would likely be due to the mutagenic potential of JAK2V617F, result in a polycythemic phenotype, and may provide cells with an in vivo proliferative advantage. The second group is characterized by the acquisition of JAK2V617F mutations by separate HSCs, and implies the presence of a preceding mutation or natural genetic variant that predisposes an individual to acquiring a JAK2 mutation during their lifetime.

(2) Co-incident JAK2 and MPL mutations occur in distinct disease clones

Included in this second group would be those rare PV or ET JAK2V617F-positive patients that respectively carry co-incident JAK2 exon 12 or MPL mutations. Genotyping of hematopoietic colonies from these double mutation-positive patients has shown that the JAK2V617F mutation always existed in a clone distinct from that carrying a JAK2 exon 12 or an MPLW515L mutation (43, 90, 91). The mutation pattern observed could result from separate mutations within separate cells of a pre-existing mutant clone, or from separate events occurring in independent HSCs (Figure 3). Proof for the latter possibility was obtained by the expression analysis of genes that undergo random X-chromosome inactivation in individual progenitors collected from informative females (91): for example, in one patient, cells bearing the JAK2V617F mutation expressed the T allele of SNP rs11549009, located in the coding region of the X-inactivated IDS gene, whereas cells bearing an MPLW515L mutation expressed the C allele. The low likelihood of these mutational events happening in separate HSCs by chance alone suggests that an inherited allele plays a role in the subsequent acquisition of mutations.

Figure 3. Tracking the cell-of-origin in cases with co-incident MPN mutations.

The pattern of co-existent JAK2 and MPL mutations apparent in some MPN cases could arise in several ways. In Model A, JAK2V617F-positive and MPLW515L-positive colony-forming cells would result from secondary mutations acquired separately in cells of a common founder clone. As the likelihood of co-incident mutations occurring by chance is significantly less than the observed frequency, the shared founder clone would have to carry a mutation in unknown gene(s), Z, that increases the likelihood of subsequent mutation or promotes the survival of mutant JAK2- or MPL-positive cells. In Model B, the JAK2V617F and MPLW515L mutations occur separately in independent hematopoietic stem cells. Experimental proof for the latter possibility is obtained by evaluating in informative female patients expression from loci undergoing X-inactivation. In Model A, with allele C on the imprinted X chromosome (shaded in black), all mutation-positive cells express the “T” transcript. An identical pattern would occur in half of those cases in which Model B was operative: by chance, the JAK2 and MPL mutations were acquired in unrelated stem cells that both had an active T allele (“1st possibility”). However, in the remaining 50% (“2nd possibility”), one mutation would occur in a stem cell expressing “T”, the other in a stem cell expressing “C”.

(3) JAK2V617F-positive patients may be positive for other MPN-associated mutations

There is a growing body of evidence that JAK2V617F-positive ET, PV or MF patients have an increased risk of developing BCR/ABL-positive CML. In several such cases, a Ph-negative MPN co-existent with CML became apparent only after imatinib-induced remission of the BCR/ABL-positive population; retrospective molecular analyses showed that the JAK2V617F allele was already present at the time of CML diagnosis, and was clearly not therapy-related (92-94). Two models could explain the disease history in these instances: the patient had an undiagnosed MPN that either preceded the acquisition of a t(9;22) translocation in a HSC that was part of the malignant clone, or developed CML separately as the result of transformation of a normal HSC. Unfortunately, samples that would have ruled out one of these possibilities were unavailable.

Rare examples of co-existent MF and systemic mastocytosis have also been reported (91, 95). Analysis of erythroid colonies from one patient revealed that the KITD816V mutation that is associated with systemic mastocytosis (96) and the MF-associated JAK2V617F mutation were mutually exclusive, supporting the contention that HSCs in these individuals were more prone to acquiring mutations than members of the general population.

(4) Leukemia in JAK2V617F-positive MPN patients often arises in a non-mutated HSC

Surprisingly, the mutant JAK2 allele is not present in the leukemic blast cells of about half of all JAK2V617F-positive MPN patients that develop a secondary AML (97, 98). SNP analysis of several matched MPN (pre-transformation) and AML (post-transformation) samples has shown that the apparent loss of the JAK2V617F allele during leukemic transformation is not the result of HR or gene conversion in a JAK2V617F-heterozygous cell (99). Instead, the JAK2V617F mutation and the mutation(s) driving leukemogenesis are likely to arise in separate HSCs, with the latter occurring in an HSC with a wildtype JAK2 genotype in half of cases.

(5) “Secondary” mutations seen in MPN patients can occur prior to any JAK2 mutation

Colony genotyping studies have shown that at least several of the mutations present in MPN patients that were assumed to occur secondary to the JAK2V617F mutation, including deletion of chromosome 20q (del20q) and TET2 inactivation, may in fact arise first, consistent with their being the “pre-JAK2” mutation postulated by Skoda and colleagues (28). In one study, del(20q) was found to occur in almost all hematopoietic cells from some patients, but that the mutant JAK2 allele was present in only a subset of cells (100). However, in other cases, del(20q) clearly can occur in a sub-population of cells that do not carry the JAK2V617F mutation (91) or are acquired secondary to the JAK2V617F mutation (16). Similarly, TET2 mutations do not follow a strict temporal pattern of acquisition, with the TET2 mutation arising first in almost half of the JAK2V617F-positive, TET2-mutated patients analyzed (17, 49). These data support the notion that there may underlying genomic instability in the HSCs of a subset of individuals that acquire a JAK2V617F mutation, although it cannot be ruled out that haploinsufficiency for TET2 or genes located on chromosome 20q renders HSCs more susceptible to subsequent acquisition of a JAK2 mutation.

A common JAK2 haplotype predisposes to the acquisition of mutations in JAK2 or MPL

A population-based study of almost 12000 Swedish patients with a non-familial MPN and 25000 of their first-degree relatives demonstrated that family members have a five-fold to seven-fold increased risk of developing ET, PV or MF, and a two-fold increased risk of CML (101). Genetic analyses recently provided a partial explanation for this intriguing finding: a relatively frequent JAK2 haplotype predisposes an individual to acquiring a JAK2 mutation (15, 102-104), accounting for approximately half of the risk identified in the Swedish cohort. Compared to healthy controls, the “46/1” JAK2 haplotype is significantly over-represented in individuals positive for a JAK2V617F, JAK2 exon 12 or MPL mutation. However, this JAK2 haplotype is not over-represented in patients with CML (105), suggesting that other genomic loci influence the initiation of this disorder in the first-degree relatives of MPN patients.

Two models have been proposed to explain this phenomenon: the “hypermutability” model, which suggests local mutations occur more frequently in JAK2 alleles with the 46/1 haplotype; and the “fertile ground” model, which states that the mutation rate is not dependent on haplotype, but that mutations occurring on the 46/1 haplotype provide some selective advantage over mutations occurring on other JAK2 haplotypes. Currently, the published data are such that neither model can be ruled out. Genome-wide association studies independently identified that rs10758669, a JAK2 SNP that tags the 46/1 haplotype, is being associated with Crohn’s disease (106); as this is a non-malignant disorder of the gastrointestinal tract, it would appear that the hypermutability model might not be valid, although experimental evidence has suggested that this SNP enhances susceptibility to DNA damage (107). Furthermore, the 46/1 haplotype is over-represented in ET patients with an MPL mutation (104); it is currently unclear how sequence variation on chromosome 9 increases the mutability of a locus located on another chromosome. To date, changes in JAK2 mRNA levels in granulocytes or coding sequence variants other than SNPs that lend support to the “fertile ground” hypothesis have not been linked to the 46/1 haplotype. However, healthy individuals with a 46/1 JAK2 allele do have reduced numbers of myeloid progenitors, supporting the contention that there is a functional difference between JAK2 haplotypes. One possibility that remains to be explored is that the non-mutated 46/1 allele is transcribed or translated in HSCs at higher levels than JAK2 alleles with other haplotypes. This may predispose HSCs to acquiring mutations, including those in JAK2 and MPL, since increased rates of mutation are observed in vitro following over-expression of wild-type JAK2 (65).

Concluding comments

Meaningful insights into the pattern of mutation accumulation in patients with a Ph-negative MPN have been recently provided by genotyping individual hematopoietic colonies. In many instances, these studies revealed a linear stepwise acquisition of mutations over time, involving two types of genomic target: those that result in constitutive JAK2 signaling (including JAK2, MPL and SH2B3), and those encoding epigenetic regulators (such as TET2, ASXL1, DNMT3A or EZH2). These mutations do not, however, follow a strict temporal pattern of acquisition; it may be the combination of mutations, rather than the order in which they are acquired, that is the important factor in disease progression. In a small proportion of MPN cases, the malignant cell-of-origin can instead give rise to a limited number of distinct sub-clones that continue to evolve in parallel. Clonal evolution in either patient group is likely to be assisted by the increased levels of DNA damage observed in cells carrying mutations that enhance JAK2-mediated signaling, and by the suppression of pathways that normally eliminate cells with compromised genomic integrity. Currently, it is unclear which came first: the chicken (genomic instability as a direct consequence of constitutive JAK2-mediated signaling) or the egg (an inherent predisposition to acquiring MPN-associated mutations, including JAK2V67F). Further insights into evolution in the MPNs are anticipated as next-generation sequencing methodologies identify the complete repertoire of genetic lesions underlying these disorders.

Colony genotyping studies have also revealed that acquisition of an MPN-initiating mutation in some instances may be the consequence of an inherent genomic instability within the patient’s HSCs. This intriguing possibility is supported by both epidemiological data and the identification of a common JAK2 haplotype that elevates the risk of an individual acquiring a JAK2 or MPL mutation. Any such genomic instability in hematopoietic cells of MPN patients has important implications for those treated with pharmacologic agents that inhibit JAK2 activity: although these drugs may alleviate the constitutional symptoms associated with myelofibrosis and significantly reduce the proportion of JAK2V617F-positive blood cells (108, 109), they may nonetheless fail to normalize the elevated risk of leukemic transformation seen in MPN patients.

Elucidation of the impact of constitutively active JAK2 signaling on genome instability and tumor evolution in the MPNs is likely to shed light on the mechanisms behind this process in other malignancies. JAK2 activating mutations are not only found in patients with an MPN, but also occur in a significant proportion of children with Down syndrome-associated or high-risk ALL (110), and in a number of solid tumors, including cancers of the lung and breast (3, 111). Moreover, JAK2 activation occurs in a variety of malignancies that are not known to carry JAK2 mutations; its inhibition significantly reduces constitutive STAT3 activation and suppresses the growth of breast, ovarian and prostate cancer xenografts in mice (112). Commercially available JAK2 inhibitors may therefore reduce genome instability, diminishing the metastatic potential of these tumor types, and improving patient outcome.