Identification and Functional Interrogation of Dormant Cancer Stem Cells in JAK2V617F-Driven MPNs

The transient cellular state of dormancy is proposed to mediate cancer stem cell therapy resistance and subsequent disease relapse. However, dormancy is challenging to interrogate experimentally, and has not been formally demonstrated to exist in hematological malignancies. We have developed a novel transgenic mouse model that combines: (i) inducible expression of the JAK2-V617F mutation (JAK2VF), which drives a Polycythemia Vera-like (PV) disease; and (ii) an inducible H2B-GFP fusion protein, which facilitates prospective identification and isolation of live dormant JAK2VF hematopoietic stem cells (dHSCs) based on their long-term retention of H2BGFP-labelled chromatin once the expression of the fusion protein is switched off. To interrogate if dHSCs exist in vivo, we performed label retention experiments after activation of the JAK2VF mutation. During the chase period, all mice developed PV, demonstrating disease progression (hematocrit (HCT) >75%; splenomegaly with spleen weight >0.5 g). Despite the presence of the pro-proliferative JAK2VF mutation, almost a quarter of mutant HSCs maintained a dormant state for 12 weeks in vivo (22.4±3.7%), and dHSCs could still be identified after 20 weeks of chase (1.9±0.5%). To assess if dormancy held any biological relevance in cancer stem cells, we prospectively isolated both dHSCs and label-diluted “active” HSCs (aHSCs) from the same mice, and interrogated disease-initiating capacity following secondary transplantation of 150 dHSCs or aHSCs into each recipient mouse. While none of the mice transplanted with aHSCs developed disease, 90% of mice receiving dHSCs developed a PV phenotype within 8 weeks post-transplantation. We next employed our model to assess the relative impact of clinically relevant therapies on dHSCs by treating PV mice for 4 consecutive weeks with either the JAK2 inhibitor Fedratinib (FED) or Peg-IFN-α (IFN-α) during the label chase period. Both FED and IFN-α resolved splenomegaly (0.4±0.04g, 0.17±0.02g, 0.22±0.05g for control (CON), FED and IFN-α-treated, p < 0.0001 for CON vs. FED, p<0.0001 for CON vs. IFN-α) and partially corrected HCT levels (78±3.9%, 77±5%, 60±10.3% for CON, FED and IFN-α-treated, p<0.0001 for CON vs IFN-α). However, FED had no impact on the frequency of disease-propagating dHSCs in the bone marrow, while IFN-α significantly reduced their frequency (29.2±3%, 28.3±7%, 13.6±4% for CON, FED and IFN-α), suggesting that dHSCs may not be addicted to JAK2VF signaling for survival. This is consistent with the clinical observation that FED and IFN-α both resolve hematological parameters but have differential effects on JAK2VF allele burden. Importantly, residual dHSCs enriched by flow cytometry post-IFN-α therapy were still capable of propagating disease in secondary recipient mice, while aHSCs were not, indicating it may be necessary to completely eradicate dHSCs with prolonged therapy to achieve full remission. We next wished to employ this model as a pre-clinical tool to assess the effect of novel therapeutic regimens on dHSCs. We first examined whether halving the dose of IFN-α could still result in dHSC depletion (Low dose (LD)-IFN-α), based on the rationale that such an approach might circumvent side effects associated with IFN-α therapy leading to therapy discontinuation. However, LD-IFN-α only achieved a modest correction of hematological parameters while failing to deplete dHSCs. We next tested a novel combination therapy of LD-IFN-α plus FED. This combination resulted in a correction of HCT (78.9±2.6%, 61.5±4.1%, p=0.04 for CON vs. FED + LD-IFN-α) and splenomegaly (0.41±0.05 g, 0.19±0.02 g, p<0.001 for CON vs. FED + LD-IFN-α), as well as a reduction in dHSCs similar to the standard higher IFN-α dosage (29.2±3%, 15.6±4.5%, 13.6±3.4% for CON, FED+LD-IFN-α and IFN-α). This suggests that LD-IFN-α renders dHSCs reliant on JAK2 signaling and thus susceptible to JAK2 inhibition.

In conclusion we have demonstrated the existence of a rare subpopulation of JAK2VF-mutant dHSCs that represents the PV disease propagating cell, which demonstrates differential susceptibility to JAK2 inhibition (resistant) and IFN-α (sensitive). As proof of concept of employing this model for pre-clinical evaluation of new therapeutic avenues, we have tested a novel combination therapy of FED plus LD-IFN-α which effectively depletes dHSCs while also correcting clinically-relevant hematological parameters.