Molecular pathophysiology of Philadelphia-negative myeloproliferative disorders: beyond JAK2 and MPL mutations

The classic Philadelphia-chromosome negative chronic myeloproliferative disorders (MPDs), recently renamed as myeloproliferative neoplasms, that include polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (PMF), had been largely neglected among hematologic neoplasms until early 2005 when the first recurrent molecular abnormality was described, consisting of a G>T point mutation at nt1849 in JAK2 and resulting in a valine to phenylalanine substitution at residue 617 (V617F).1–4 Then, the discovery of mutations in MPL, represented by a W>L or W>K shift at codon 515,5 and of variable molecular abnormalities (point mutation, insertion, deletion) in JAK2 exon 126 was also reported. Almost all patients with PV have a somatic genetic defect in JAK2 that is represented by the V617F allele in 90–95% of cases and by abnormalities in exon 12 in roughly 2%, while they are spared by MPL mutations; on the other hand, only 60% of patients with ET or PMF harbor the JAK2V617F mutation and 3–7% exhibit the MPLW515L/K mutation. While other infrequent mutations in MPL can also occur,7 exon 12 abnormalities have not yet been reported in PMF or ET patients. The presence of any of these molecular abnormalities, that point to a clonal myeloproliferation, stands as a major diagnostic criterion in the revised classification of myeloid neoplasms of the World Health Organization.8 They are gain-of-function abnormalities that conferred growth-factor independence to cells transduced with mutant allele and induced a myeloproliferative disease when expressed in murine transplant models.9

It was unexpected to find a single mutated allele associated with more than one disease, notwithstanding the fact that the different MPDs are strictly related to each other and show substantial phenotypic mimicry. There are possible explanations for this. One is that the unique clinical phenotypes mirror the stem/progenitor cell level at which the mutational event occurred. However, this does not seem to be the case, since the V617F allele has been found in myelo-lymphoid progenitors in patients with PV or PMF, as well as in flow cytometry purified populations of hematopoietic stem cells and committed progenitors. Similar observations have been reported for MPLW515L/K mutations.10 In only a minority of ET cases a restriction of V617F allele to megakaryocytic lineage was described, although it was unclear whether a very low burden of mutated granulocytes might have meant the abnormal genotype went undetected. Another possibility is that different diseases are caused by variable dosage of mutant V617F allele (information about MPLW515L/K mutation in this regard is scarse) that in turn influences the level of activation of the JAK/STAT signaling pathway. Indeed, there are substantial differences in the median burden of V617F allele in peripheral blood granulocytes, with the highest level being found in PV and the lowest in ET patients.11,12 Furthermore, homozygosity for the JAK2V617F mutation, that originates from mitotic recombination of the short arm of chromosome 9, is present in approximately 30% of PV or PMF patients as opposed to 2–4% of ET. Variable proportions of wild-type, heterozygous and homozygous progenitors are present in most patients with PV, while homozygous progenitors are reported as being rare in ET; mutated erythroid progenitors are more sensitive to erythropoietin than normal ones, and most erythropoietin independent erythroid colonies (EEC) are made up of homozygous progenitors. Conceivably, duplication of mutant allele is expected to result in a higher level of JAK2/STAT activation than in cells harboring one mutant and one wild-type allele, possibly because of the loss of competition between normal and mutated allele and/or impaired interaction of mutant JAK2 with cellular regulators such as the suppressor of cytokine signaling-3 (SOCS3).13 A correlation between mutant allele burden and disease phenotype is indirectly supported also by findings in animal models. The adoptive transfer of the V617F allele caused a rapid appearance of polycythemia, leukocytosis and splenomegaly in mice, pointing to a PV-like disease that eventually recapitulated the evolution to a post-PV myelofibrosis with time. The fact that thrombocytosis was not observed in these mice was ascribed to the exceedingly high levels of mutant JAK2 mRNA that were produced in hematopoietic cells with the retroviral infection. This hypothesis has recently received experimental support thanks to the successful development of conditional transgenic mice that expressed variable levels of JAK2V617F,14 in which the phenotype closely reflected the ratio between normal and mutated JAK2 mRNA. Likewise, the V617F allele burden correlates with hematologic characteristics and clinical endpoints in MPD patients. However, owing to the wide distribution of individual values, it seems quite unlikely that load of mutant V617F allele alone represents the key mechanism at the basis of MPD pleiotropy, although it is likely that changes in V617F allele burden correlate with disease progression.12

Figure 1.The figure describes a working model of genetic events and other mechanisms possibly involved in the pathophysiology of myeloproliferative disorders. A hematopoietic pluripotent stem cell (HSC) is the target of JAK2V617F mutation (other abnormalities in JAK2 exon 12 or MPL may have similar effects; for clarity, the JAK2V617F mutation only is presented as a model, and the mutated allele is indicated by a red dot). However, a pre-JAK2 mutated HSC cell (green) might also exist as the result of a still unknown molecular lesion, indicated by a single blue dot; this cell might subsequently acquire the JAK2V617F mutation, although also the inverse process could happen (dashed arrow). A JAK2V617F heterozygous cell may then undergo a mitotic recombination process at 9p leading to JAK2V617F homozygosity, which is likely associated with growth advantage. As a consequence, the clonal population at the basis of a myeloproliferative disorder may be comprised of a variable proportion of cells harboring isolated or multiple molecular defects (please note that the existence of some of the cells depicted has not been formally proven). JAK2V617F mutation is probably necessary for PV to occur, while other still undiscovered genetic abnormalities may be at the basis of JAK2 un-mutated ET or PMF. Finally, normal HSC persist in many patients for length of time and in variable amount. All these cell-heritable and irreversible genetic events may be complemented by the effects of post-genetic regulatory mechanisms, including miRNAs or epigenetic modifications, and/or of host genetic modifiers, such as gender, SNPs and possibly others, including also non-genetic modifiers (ie, ferritin iron availability). The phenotypic pleiotropy of a myeloproliferative disorder would then be the result of a combination of somatic mutation(s), inherited genetic variability, post-genetic regulation and host modifiers.

There is evidence to support the possibility that disease alleles other than JAK2V617F or MPLW5151L/K are involved in the pathogenesis of MPDs,15 in addition to the fact that almost 40% of ET or PMF patients still lack a molecular marker. First, in females with JAK2V617F-negative ET substantiated proof of the existence of clonal hematopoiesis has been provided through the analysis of X-chromosome inactivation pattern (X-CIP).16,17 Furthermore, the percentage of granulocytes and platelets with JAK2V617F was often lower than expected by the analysis of X-CIP. In addition, in two JAK2V617F-mutated patients with a del(20q), discrepancies were found in the size of del(20)q and JAK2V617F-mutated clones, leading to the suggestion that V617F allele had been acquired by a pre-existing del(20q) clone.18 Similarly, some patients contemporarily harbor JAK2V617F and MPLW515L/K mutation,19, 20 and clonal dominance by the MPL mutated cells has been demonstrated.21 Second, presence of the V617F allele is robustly associated with the growth of EEC, that is a hallmark of PV and MPDs in general. However, EEC wild-type for JAK2 have been detected at low frequency in PV patients22 which would indicate the JAK2 mutation is a secondary, not a disease-initiating, genetic event. A similar interpretation could be advocated also to explain why exon 12 abnormalities are almost exclusively associated with an erythroid phenotype in the setting of a clear myeloproliferative disorder. Third, the incidence of MPDs among relatives is greater than expected in the normal population, but neither the diagnosis of MPD nor the distinct clinical phenotype encountered in these families were invariably associated with the presence of JAK2 or MPL mutation, suggesting that the JAK2V617F mutation might occur as a secondary genetic event on the background of an inherited genetic predisposition.23 Finally, blast cells of acute myeloid leukemia that developed in patients with a preceding JAK2V617F-positive MPD were often JAK2V617F-negative, indicating that they might derive from the transformation of a pre-JAK2V617F mutated hematopoietic stem cell that was originally at the basis of the MPD itself.24 However, that these JAK2V617F negative blasts represented the de-novo transformation of a residual normal progenitor cannot be excluded on the basis of available data. It is more unlikely that the JAK2 un-mutated blasts originated from a single cell heterozygous for the V617F mutation that underwent a reverse mitotic recombination process leading to an un-mutated JAK2 genotype. On the contrary, the possibility that the JAK2V617F and MPL mutation alone might be sufficient for causing an MPD is compellingly supported by the results of animal models, including the recently described transgenic mice that developed phenotypes closely resembling human ET and PV and presented MF-like symptoms as they aged.25

There is also the non-mutually exclusive possibility that, on the grounds of the JAK2 or MPL mutation and/or of any other pre-existing or additional molecular abnormality, individual-related characteristics contribute to the phenotypic pleiotropy of the MPDs. This was initially suggested by results in animal models. Depending on the genetic background of mice transplanted with JAK2V617F transfected cells, the characteristics of their myeloproliferative disease were somewhat different. In fact, in addition to polycythemia, the JAK2V617F allele induced marked leukocytosis and marrow reticulin fibrosis in Balb/c mice, while leukocytosis was minimal and fibrosis was limited to the spleen in case of C56Bl/6 strain.26 Compelling support to the existence of genetic modifiers in humans has been recently provided by Pardanani et al.,27 who genotyped a number of single nucleotide polymorphism (SNP) loci in JAK2 and EPO-R of patients with MPDs. They found that 3 JAK2 SNPs were inversely associated with PV or ET, but not PMF, while 3 additional SNPs in JAK2 and one in EPO-R were significantly associated with PV only.27 Due to the design of the study, these should not be strictly considered as genetic abnormalities leading or predisposing to the disease, but rather as genetic markers of potential interest due to their significant association with the disease, similar to polygenic predisposition to many common diseases. Gender is another genetic-based host-modifier that may have relevance,28 exemplified by the fact that PV is more common in men while ET is definitely more frequent in women. An example of a non-genetic modifier might be the availability of iron stores. It was by taking into an account these possible variables that an intriguing hypothesis of a biological continuum of ET and PV was formulated.29

All together, these observations support the idea that additional factors, possibly represented by yet unknown disease-causing allele(s) or by host modifying characteristics, or both in variable combinations, concur with JAK2 or MPL mutation to the pathophysiology of MPDs (Figure 1). Two papers published in this issue of Haematologica deal with the beyond JAK2 and MPL mutation perspective by two different approaches. With the aim of identifying novel disease-associated molecular markers, Ioana Borze and colleagues employed comparative genomic hybridization (CGH) to discover cytogenetic abnormalities not otherwise detectable with conventional cytogenetics.30 CGH is a method designed for the identification of genomic regions characterized by copy number alteration. More recently, array-based CGH platforms (aCGH) have been developed where arrays of oligonucleotides covering up to more than 200,000 genes and spaced 150 kb across the genome have replaced metaphase chromosomes as hybridization targets, resulting in higher resolution and facilitating direct mapping of copy number changes to relatively narrow genome regions. Ideally, CGH analysis represents the basis for subsequent molecular genetic strategies aimed at identifying altered genes involved in tumorigenesis. As a matter of fact, a number of regions gained or lost in solid and hematologic malignancies have been identified, which should result in variegated cellular mRNA expression level. Among the 36 patients (21 ET and 14 PV) evaluated by Borze et al. only one abnormal case was identified, which presented duplication at 1q and deletion at 13q in relatively large regions including approximately 1,100 and 300 genes respectively. Similar negative findings have been presented in a preliminary report on 20 patients with ET (Espinet B et al., 11 EHA meeting, 2006; abstract 098). On the contrary, the Mayo Clinic group recently reported that the percentage of patients showing copy number changes was 17% of 12 patients with ET, 32% of 22 patients with PV, and 52% of 27 PMF patients (49 ASH meeting, 2007; abstract n.1550). Although the number of patients included in each of these series was quite small and different aCGH platforms were employed, possible explanations for these discrepancies are not obvious, and this approach certainly deserves further investigation. However, a preliminary suggestion from these studies may be that chromosomal defects within the analytical resolution of aCGH approach are unlikely to be a common occurence in MPDs.

The second paper is from Hana Bruchova and colleagues, who performed a gene expression profiling of micro-RNAs (miRNAs) in granulocytes from PV patients.31 miRNAs constitute a large family of small non-coding RNAs functioning as regulators of post-transcriptional processes; they have been implicated in many physiological and pathological processes, including cancer. The miRNA signature observed by Bruchova et al. in PV granulocytes resulted clearly different from that of normal granulocytes, and allowed the identification of a number of miRNAs (at present, 40 out of a total of 326) whose abnormal expression was either correlated or independent from the JAK2V617F mutation. It is of interest, and probably not unexpected, that at least some of the abnormally regulated miRNAs found in PV had been detected in PMF granulocytes in another study,32 and that they were also detected in ET or PMF patients by Bruchova et al. Thus, a complex pattern of aberrant miRNA expression might contribute to the molecular complexity of MPDs through transcriptional regulation, although the mechanisms for the abnormal expression of those miRNAs remain completely unknown. As a matter of fact, the loss of expression of miR-15 and miR-16 in chronic lymphocytic leukemia is due to deletion of a region included in the 13q14,33 while no somatic molecular lesion of a miRNA gene has yet been reported in MPDs. It is still premature to conclude whether and how aberrant expression of miRNAs concurs with genetic lesions to modify disease presentation, and also if it is related to genetic abnormalities. One limitation of this kind of study is the difficulty of correlating deregulated expression of miRNAs with that of target genes, since there is a pleiotropy of putative target genes. Furthermore, the final regulatory activity of a given miRNA may be that of inhibiting transcription in most instances but also of facilitating it in other cases. Accordingly, the demonstration that expression level of some genes, selected by in silico analysis as possible targets of the deregulated miRNAs, was modulated in the granulocytes of PV patients does not represent a formal proof of a cause-effect relationship.31 In fact, clarification of the role of abnormally expressed miRNAs in PV and other MPDs will require manipulation of endogenous miRNA levels in hematopoietic stem cells and/or their descendants obtained from patients with MPDs; these experiments are technically demanding, and unlikely to be easily accomplished. Finally, it should also be underlined that in these studies31, 32 the number of miRNAs included in the analysis represents only a proportion of the total number of miRNAs that are currently known (>600 according to the v.11 release of miRBase at http://microrna.sanger.ac.uk/sequences), raising the suspicion that additional miRNA abnormalities could be found once more comprehensive array platforms are eventually used.

The discovery of the JAK2 and MPL mutations has represented a tremendous improvement in the diagnosis and characterization of the MPDs, and has opened new perspectives for the clarification of disease pathogenetic mechanisms and hopefully for targeted therapy. However, it was also soon realized that a simple model incorporating only JAK2 and MPL mutation is inadequate to explain the pathophysiology and pleiotropy of the MPDs, and in this perspective the two papers in this issue of Haematologica represent valuable contributions to help decipher this intriguing story.