Understanding the pathology of MPN: The value of in vivo and in vitro models

Among chronic Philadelphia-negative myeloproliferative neoplasms (MPN), polycythemia vera (PV) is considered the most common, often characterized by excessive formation of red blood cells (RBC), increased levels of hematocrit and hemoglobin, erythroid and megakaryocyte proliferation, and subnormal serum erythropoietin levels. Different in vitro and in vivo models have been designed to elucidate the pathophysiology of PV, to identify potential biomarkers, and to test therapeutic targets for the development of novel drugs. Niccolò Bartalucci, Paola Guglielmelli, and MPN Hub Steering Committee member Alessandro M. Vanunucchi have published a review on preclinical models used in PV in Expert Opinion on Therapeutic Targets; here we present the key points.¹

Background

The main characteristic of PV is increased RBC mass, leading to blood hyperviscosity, and other associated symptoms including thrombosis, tendency to bleeding, hypertension, and headache. Patients with PV may experience pruritus, however, the mechanism behind this is under investigation. Splenomegaly is common among patients (30–40% of patients), and thrombotic events are a leading cause of morbidity and mortality, affecting up to ~40% of patients. Leukocytosis may be associated with thrombosis and disease evolution as well as shorter survival. Within 10–15 years of diagnosis, disease evolution to a myelofibrotic phase or a blastic phase may occur in up to 30% and 20% of patients with PV, respectively.

The JAK2^V617F mutation is found in most patients with PV (> 95%), with distinct levels of variant allele frequency (VAF) varying based on individual characteristics including age and blood count levels, and it has been found to contribute to inflammation. JAK2 exon 12 mutation, identified in 3% of patients, is characterized by preferential erythroid proliferation, and diagnosed at a younger age than those with JAK2^V617F mutation. Additional mutations, such as ASXL1, IDH2, and SFRS2, may result in a poorer prognosis associated with shorter overall survival (OS), progression-free survival (PFS), and leukemia-free survival.

In vitro cell models

Different in vitro cell models have been used, including human erythro-leukemia cell lines harboring JAK2^V617F mutation and stem cells from patients with PV, which have the ability to grow endogenous erythroid colonies (EEC).

EEC is defined as formation of erythroid colony-forming units (BFU-E and CFU-E) independently of physiological signals. A potential target pathway for the treatment of PV was tested in in vitro EEC models targeting murine double minute 2 (MDM2), a negative regulator of p53. Idasanutlin, a MDM2 inhibitor, was shown to significantly reduce the growth of CFU-GM and BFU-E colonies.

Various models, including HEL cells genetically edited by CRISPR/Cas9 KO plasmid, have shown that activity of dual-specificity phosphatase 1 (DUSP1) allows JAK2^V617F-mutated progenitors to preserve themselves from DNA damage accumulation. Therefore, DUSP1 may be a potential therapeutic target in JAK2^V617F-mutated progenitors, inhibiting their survival in an inflammatory microenvironment where DUSP1 is overexpressed. Another pathway tested in HEL cell lines was phosphatidylinositol 3-Kinase (PI3K), which was later confirmed as a valid therapeutic target in in vivo models.

Therapeutics tested in HEL cell lines are summarized in Table 1.

Table 1. Therapeutic agents tested using HEL cell lines¹

In vivo mouse models

To date, different in vivo models have been used to investigate the role of mutations in the development of MPN, or the impact of additional factors in MPN phenotype modulation. We will look at the best described models used in PV, namely bone marrow transplant (BMT), transgenic, and knock-in JAK2^V617F models.

BMT models

Retrovirally transduced cells expressing JAK2 mutation were transplanted into BALB/c and C57B1/6 mouse strains and allowed, for the first time, to reproduce the phenotype of PV with splenomegaly, erythrocytosis, extramedullary hemopoiesis, and cytokine-independent growth of progenitor cells, which was also observed in secondary recipients. Interestingly, these models did not induce thrombocytosis and there was eventually a progression to myelofibrosis. Shortcomings of these models are the nonphysiologic locus of transgene integration and expression ratio of WT versus V617F alleles, indicating that other genetic changes may contribute to the development of the disease.

Nevertheless, fedratinib was successfully tested in these models and showed efficacy, with reduced hematocrit levels, dose-dependent reductions in spleen size, reduction of fibrosis, decrease in JAK2^V617F allelic burden, and better survival.

Another agent tested in a BMT model of BALB/c mice was momelotinib, which was associated with lower levels of white blood cells (WBC) and hematocrit as well as reduced BM fibrosis. Inflammatory cytokine levels were also improved.

In a different BMT model, hematopoietic progenitors created with a luciferase (Luc) JAK2^V617F expressing vector were transplanted in mice and resulted in a PV phenotype characterized by erythrocytosis and splenomegaly. These mice were treated with a JAK2 JH1-domain binder MRLB-11055, reducing the bioluminescent signal by 20-fold during the treatment.

Transgenic mouse models

Transgenic models, developed in 2008, confirmed the role of driver mutations and expression levels in modulating the MPN phenotype.

Models expressing mutated JAK2 from H2Kb or Vav promoters (V617F/wild-type (WT) allele ratio = 1) have shown a phenotype resembling essential thrombocythemia (ET), with increases of hematocrit and WBC and apparent thrombocytopenia. In contrast, another model with high JAK2^V617F expression developed a phenotype more closely resembling PV, with a distinct increase in all blood cell lineage counts. While JAK2^V617F mice generated by Vav-Cre mice demonstrated a V617F/wild-type (WT) allele ratio < 1, and presented with thrombocytosis and neutrophilia. However, when JAK2^V617F was transiently induced in these Vav-Cre mice to increase expression levels, a different phenotype with erythrocytosis, myelofibrosis, osteosclerosis, and splenomegaly occurred.

Knock-in V617F models

Preclinical experience investigating genome integration site has confirmed the importance of timing and intensity of oncogene expression for the development of specific disease phenotypes. A phenotype close to PV developed in several JAK2^V617F knock-in mice, where the oncogene was expressed from physiologic promoters, with erythrocytosis, leukocytosis, splenomegaly, reduced survival, and, in one model, additional tri-lineage hyperplasia and myelofibrosis. A notable exception was one model with a phenotype more related to ET, which could possibly be explained by the use of a human mutated JAK2 gene instead of the murine gene. Most of these knock-in models had comparable expression of the mutated and wild-type JAK2 allele, and the disease was transplantable.

JAK2^V617F homozygous and heterozygous mice generated by conditional transgene activation and a V617F/WT allele ratio of 1:2 presented a PV-like phenotype with increases in hemoglobin, hematocrit, WBC, RBC, and platelet levels, and an increase in spleen size. In addition, homozygous mice had marked thrombocytosis, greater reticulocyte count and Epo-independent colonies formation, and lower hemoglobin levels in comparison with heterozygous mice.

Various knock-in models were used to assess therapeutic efficacy of drug candidates:

• Vorinostat, a class I/II histone deacetylase (HDAC) inhibitor, demonstrated improvements in blood counts including WBC, RBC, hemoglobin, and hematocrit levels, with less effect on high platelet count. It also reduced spleen size and the number of cells expressing V617F.
• Fedratinib administration was associated with lower levels of blood counts and reductions in spleen size and erythroid precursor numbers.
• The class II JAK2 inhibitor CHZ868 demonstrated reduced JAK2^V617F allele burden as well as the frequency of erythroid, restored architecture of spleen, improved hyperplasia of BM megakaryocyte, and reduced megakaryocyte/erythrocyte progenitors.
• Interferon alpha improved blood count, microcytosis, and B and T cell numbers, and reduced myeloid precursors.

Three other models were designed to establish a correlation between JAK2 inhibition and disease severity, namely, PV, post-PV MF (PPV-MF), and post-ET MF (PET-MF). PV and PPV-MF models were established with transplanting cells from JAK2^WT and JAK2^V617F knock-in mice, and a PET-MF model was generated using TPO^high mice.

When fedratinib was administered in all three models, it was associated with

• reduced leukocytosis, thrombocytosis (specially in PET-MF model), and erythrocytosis in all models;
• reduced BM fibrosis in the PET-MF model;
• reduction in splenomegaly in the PPV-MF and PV models; and
• normalized reticulocyte count and improved erythroid hyperplasia in the PV and PPV-MF models.

Other mouse models

In a PV model designed to study JAK2 exon 12 mutation, increased phosphorylation of STAT3, Erk1/2, STAT5, and STAT1 was observed compared with WT mice. Survival was also reduced, and genes involved in iron metabolism were overexpressed.

JAK2^V617F mice with Ezh2 deletion exhibited reduced hematocrit levels and RBC production, increased platelet count, and more megakaryocytic colonies, indicating an association between Ezh2 deletion and progression to myelofibrosis.

BMT models of p53 and JAK2^V617F demonstrated that mice with a loss of p53 progressed to a leukemic state, providing a justification to therapeutically target p53 dysregulation.

Conclusion

The presence of JAK2^V617F mutation is a well-known feature of PV and was used in most models to elucidate the pathology and phenotype of PV, including hematologic progression. These models have also been employed to test novel agents to be used in the treatment of PV. However, given the complex nature of MPN, results obtained with models using JAK2^V617F-mutated cells should be interpreted with caution as other mutations can contribute to the PV phenotype. Patient-derived xenografts may be an option to facilitate the understanding of this complex disease, which may also allow investigation of the mechanisms behind relapse or refractoriness following targeted treatment.