Depletion of Jak2V617F myeloproliferative neoplasm-propagating stem cells by interferon-α in a murine model of polycythemia vera

JAK2V617F is the most common molecular alteration in the BCR-ABL negative myeloproliferative neoplasms (MPNs).^1-4  To definitively cure JAK2V617F-mediated MPN in humans, it will be necessary to eradicate all JAK2V617F mutant hematopoietic stem cells (HSCs) that solely possess the capacity to self-renew and therefore maintain the disease over time. Interferon-α (IFNα) is an effective therapy currently used in MPN patients and, importantly, it appears to be more effective than JAK2 kinase inhibitors, which inhibit the key molecular target in MPNs, at achieving molecular remissions in MPNs.^5-8  Despite years of observational clinical data, the mechanism by which IFNα induces complete molecular remission (CMR) in MPN patients remains unknown. In this study, we use a conditional Jak2V617F/+E2ACre+ (hereafter Jak2^VF) knockin murine model, in which we previously characterized the MPN-initiating stem cell population,⁹  to investigate the effects of IFNα on Jak2^VF MPN stem cells in vivo.

In MPN patients, the JAK2V617F mutation is detectable in the most primitive HSCs in the bone marrow¹⁰  and in all mature cell lineages.^11,12  JAK2V617F is also found in long-term culture initiating cells, and JAK2V617F mutant SCID repopulating cells are multi-potent and skewed toward myeloid differentiation,¹³  indicating that JAK2V617F is present in functionally competent long-term HSCs (LT-HSCs). Studies in retro-viral murine models demonstrate that JAK2V617F alone is sufficient to confer an MPN disease phenotype^14-17  and using a conditional Jak2V617F knockin model, we previously demonstrated that MPN-propagating cells are contained exclusively in the LT-HSC compartment.¹⁸  All of these lines of evidence indicate that, analogous to chronic myelogenous leukemia (CML),¹⁹  JAK2V617F-mediated MPN is maintained by a reservoir of disease-propagating stem cells that represent the ultimate therapeutic target for a definitive cure of the disease.

IFNα has a long history of efficacy in the treatment of hematological malignancies. Early studies demonstrated robust improvement in blood counts in response to IFNα treatment in patients with polycythemia vera (PV) and essential thrombocythemia.^20,21  More recent clinical trials have demonstrated that in addition to normalizing blood counts in the majority of PV and essential thrombocythemia patients treated, IFNα reduces JAK2V617F allelic burden and, in a significant proportion (15%), renders the JAK2V617F mutant clone undetectable by sensitive molecular assays.^5,6,22,23  The development of long-acting pegylated IFNα combined with the modest results shown by JAK kinase inhibitors in reducing JAK2V617F mutant allele burden in patients with myelofibrosis^8,24  has renewed interest in the use of IFNα for the treatment of MPN.²⁵  Until recently, it was thought that IFNα therapy acted primarily through immunomodulatory or antiproliferative effects.²⁶  However, 2 studies published in 2009 employed murine models to demonstrate that IFNα can directly activate the cell cycle in quiescent, LT-HSC populations.^27,28  These data suggest a novel mechanism by which IFNα may target primitive JAK2V617F MPN stem cell populations, leading to long-term disease eradication. In support of this hypothesis, 2 recent randomized clinical trials examined the addition of pegylated IFNα to imatinib therapy in patients with chronic-phase CML and observed a significantly higher rate of molecular response when compared with patients receiving imatinib alone,^29,30  suggesting that IFNα targets CML-maintaining stem cells and depletes them over time.

Using a chimeric bone marrow transplant (BMT) model generated with Jak2^VF and wild-type (WT) HSCs, we assessed the effects of IFNα on Jak2^VF disease-propagating stem cells in vivo.

Jak2^V617F/+E2Acre+ (hereafter Jak2^VF) and PCR genotyping primers were previously described.⁹  Jak2^VF mice were backcrossed and maintained on C57Bl/6 background (minimum 7-8 generations). B6.Ifnar1 knockout mice were generated by Paul Hertzog.³¹  CD45.1 Ptprc^a, CD45.2 C57Bl6/J, and C57Bl6/JxPtprc^a.F1 mice were obtained from Animal Resources Centre, Australia and Taconic, NY. All mice were maintained in pathogen-free facilities at the Queensland Institute of Medical Research and Children’s Hospital Boston, MA. All mouse experiments were approved by institutional ethics committees Queensland Institute of Medical Research protocol A11605M and Children’s Hospital Boston 10-04-1601R.

Purified HSC-enriched populations (lineage^negKit^highSca1⁺, LKS+) were isolated from Jak2^VF (CD45.2) or WT (CD45.1) mice aged 6-12 weeks. These LKS+ populations were mixed at an ∼1:2 ratio (Jak2^VF:WT) and injected into lethally irradiated C57Bl6/JxPtprc^a.F1 (CD45.1/2 double positive) recipient mice. For secondary transplants, 0.5-2 × 10⁶ bone marrow cells were mixed with 1 × 10⁶ competitor bone marrow cells (CD45.1/2 double positive) and injected into lethally irradiated recipient mice (CD45.1/2 double positive). Hematocrit (HCT) and blood parameters were measured 4 weeks after transplantation. Chimerism was determined 4 and 16 weeks after injection. For Ifnar1^−/−Jak2^VF transplants, we used unfractionated Jak2^VF or Ifnar1^−/−Jak2^VF bone marrow (post red cell lysis) that was normalized to an input of 2 × 10⁴ LKS+ cell content per recipient. Each recipient received an equivalent amount of CD45.1 helper bone marrow in an approximate cell ratio of Jak2^VF:WT or Ifnar1^−/−Jak2^VF:WT of 2:1. Recipient mice received 11 Gy total body irradiation in equal divided doses, at least 3 hours apart within 24 hours of transplantation. All transplanted cells were injected by lateral tail vein. Mice were maintained on antibiotic (Baytril) water for 2 weeks after transplantation.

Blood was collected into EDTA-coated containers and was analyzed on a Hemavet 950 analyzer (Drew Scientific).

Recombinant murine IFNα-4 was purchased (catalog no. 12110; PBL Interferon Source, Piscataway, NJ). A total of 10 000 units (correlated with the published specific activity of each batch) IFNα was diluted in phosphate buffered saline (PBS) containing 0.1% bovine serum albumin and administered by the subcutaneous route daily for 4 weeks or alternate daily for 2 weeks.²⁷  For the long-term (8-10 weeks) experiments, 20 000 units IFNα (Miltenyi Biotech, Teterow, Germany) was administered by subcutaneous route daily (Monday-Friday) for 8-10 weeks. Vehicle control mice received 0.1% bovine serum albumin in PBS injections.

Bone marrow, spleen, or peripheral blood were collected and prepared for staining by red blood cell lysis (BD Pharmlyse, BD Biosciences) and homogenization through a 70-μm filter. Erythroid precursor cell stainings were not pretreated with red cell lysis. All samples were analyzed by flow cytometry using FACS LSR Fortessa or LSR II (BD Biosciences). All staining steps were performed in ice-cold PBS containing 2% fetal bovine serum. Postacquisition analyses of data were performed with FlowJo software V9.2.3 (Treestar, CA). For peripheral blood chimerism studies, the following antibodies were used (clone in parentheses): CD45.1 (A20) and CD45.2 (104). For erythroid precursor cells, the following antibodies were used: CD71 (RI7217) and Ter119 (Ter-119). For stem cell and progenitor analysis, the following antibodies were used: lineage cocktail containing CD3ε (145-2C11), CD5 (53-7.3), Ter-119 (TER-119), Gr-1 (RB6-8C5), Mac-1 (M1/70), and B220 (30-F11); Kit (2B8), Sca-1 (D7), CD150 (TC15-12F12.2), CD48 (HM48-1), CD135 (A2F10), CD34 (Ram34), and CD16/32 (93) in addition to CD45.1 (A20) and CD45.2 (104). All antibodies were from Biolegend except CD45.1 Alexa 700 and CD45.2 Horizon V500 (BD Biosciences) and CD16/32 (eBioscience). For dead cell discrimination, Sytox blue (Invitrogen) was used. For cell cycle staining, cells were stained with the lineage cocktail. Samples were depleted of lineage-positive cells by biotin-binder Dynabeads (Invitrogen) and stained with secondary antibodies as above. Samples were then fixed and permeabilized (Fix & Perm, GAS-004; Invitrogen) as per the manufacturer’s instructions and stained with anti-Ki-67 (B56; BD Bioscience) during the permeabilization stage. Finally, cells were stained with Hoechst 33342 (Invitrogen) at a final concentration of 20 μg/mL in PBS containing 2% fetal bovine serum. For apoptosis analysis, the Vybrant FAM Poly Caspases Assay Kit (catalog no. V25117; Molecular Probes, Invitrogen) was used per the manufacturer’s instructions. Cells were first stained with lineage and stem cell antibodies and then with the cell permeant FAM-VAD-FMK poly caspases reagent, which detects active intra-cellular caspases, is fluorescent, and can be quantified using flow cytometry. The gating strategy used to determine LT-HSC frequency and chimerism is shown in supplemental Figure 1. In all cases, percentage chimerism was defined as the proportion of Jak2^VF cells as a percentage of total donor cells, ie, (%CD45.2 Jak2^VF)/(%CD45.2 Jak2^VF + %CD45.1 WT) × 100%. Total engraftment (Figure 5) was measured as the total percentage of CD45.2 cells in the blood of secondary transplant recipients either 4 or 16 weeks after transplantation.

Purified Jak2^VF and WT HSC (CD150⁺LKS⁺) populations were isolated from IFNα-treated chimeric recipient mice or vehicle-treated controls (4 weeks of IFNα, n = 4 in each group). RNA was extracted using the Picopure RNA extraction kit according to the manufacturer’s instructions (Invitrogen). The samples were amplified with the Illumina Total Prepamp kit (Invitrogen) and hybridized on Illumina MouseRef-8 gene expression arrays (V2_0_R0_11278551_A). The data were analyzed with GenePattern online analysis software including quantile normalization.³²  Gene set enrichment analysis (GSEA) v2.0 was used for analysis against datasets of molecular signature datasets (www.broadinstitute.org/msigdb).³³  The microarray dataset reported in this article has been deposited in NCBI’s Gene Expression Omnibus and is accessible through GEO Series accession number GSE44961 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE44961).

Data were analyzed and figures were generated with Prism for Mac OsX, V5.0 (Graphpad Software Inc.). For comparison between 2 groups, a Student 2-tailed t test was performed. Results given are mean ± standard deviation unless otherwise specified.

In this study, we used the Jak2^VF knockin model to evaluate the effects of IFNα on Jak2^VF mutant and WT hematopoietic stem and progenitor cells (HSPCs) in vivo. Because hematopoiesis in patients with MPN is comprised of a mix of both clonal MPN cells bearing somatic genetic mutations and residual polyclonal hematopoietic cells,³⁴  we first undertook to accurately model this heterogeneity in vivo. To do this, we generated a chimeric model of Jak2^VF-induced MPN by isolating HSC-enriched LKS⁺ populations from CD45.2 Jak2^VF mice and CD45.1 Ptprc^a control mice. LKS+ were mixed at a 2:1 ratio (WT:Jak2^VF) and injected into lethally irradiated B6xPtprc^a.F1 (expressing CD45.1 and CD45.2) (Figure 1A). After engraftment (at least 4 weeks), mice were treated with IFNα 10 000 U/d for 4 weeks or 10 000 U on alternate days for 2 weeks. To examine the effects of longer term IFNα, chimeric mice received 20 000 U/d on weekdays for 8 weeks.

To determine the effects of murine IFNα on Jak2^VF MPN in vivo, we treated chimeric recipient mice with IFNα 10 000 U/d for 28 days. IFNα treatment was associated with a modest, but significant reduction in HCT, white cell count (WCC), red cell count (RCC), and hemoglobin in Jak2^VF chimeric recipients. (Figure 1B-E). There was no significant change in platelet count or total bone marrow cellularity (supplemental Figure 2A-B). IFNα treatment was well tolerated at this dose as evidenced by the absence of a significant reduction in mouse body weight (supplemental Figure 2C). Consistent with human clinical data indicating that molecular responses require prolonged IFNα treatment to be achieved, there was no significant reduction in the percentage of Jak2^VF peripheral blood chimerism following this 28-day IFNα treatment schedule in chimeric recipient mice (supplemental Figure 2D). Prolonged treatment with IFNα for 8 weeks (5 d/weeks, 20 000 U/d) demonstrated sustained and progressive normalization of HCT, WCC, RCC, and hemoglobin in Jak2^VF chimeric recipients (Figure 1F-I). Together, these data confirm that recombinant murine IFNα treatment induces hematological responses in Jak2^VF-mediated MPN.

Splenomegaly and extramedullary hematopoiesis are cardinal features of JAK2V617F-induced MPN in patients and murine models of disease.^9,35-38  We found that IFNα treatment decreased extramedullary hematopoiesis as evidenced by a significant reduction in the total spleen WCC and splenic weight (Figure 2A-B). There was also a significant reduction in the percentage of mobilized progenitor cells (lineage^lowKit^high) in the spleens after IFNα treatment (supplemental Figure 3). To evaluate the effects of IFNα treatment on extramedullary erythropoiesis, we assessed erythrocyte differentiation using flow cytometry (Figure 2C for representative plots). Immature or early erythroid precursor cells (CD71^highTer119^mid-high) were expanded in Jak2^VF transplant mice as previously described in Jak2^VF primary mice.⁹  IFNα treatment resulted in a marked reduction in the total immature erythroid population of Jak2^VF chimeric mice (Figure 2E), whereas late erythropoiesis (CD71^low-midTer119^high) was relatively preserved (Figure 2F). We found increased apoptosis in erythroid precursors after IFNα treatment, and again this effect was most pronounced in immature erythroid precursors (Figure 2D,G). To determine if these effects were specific for the extramedullary erythropoiesis seen in Jak2^VF MPN, we treated nontransplanted primary Jak2^VF or WT mice with IFNα. We found a relative reduction in immature erythropoiesis in IFNα-treated Jak2^VF mice; however, no changes were observed in IFNα-treated WT mice (supplemental Figure 4). In aggregate, these findings demonstrate that IFNα reverses splenomegaly due to extramedullary hematopoiesis and that IFNα induces apoptosis in Jak2^VF erythroid precursor cells, with immature erythroid precursors (CD71^highTer119^mid-high) demonstrating enhanced sensitivity compared with mature erythroid precursors (CD71^low-midTer119^high).

Oncogenic Jak2^VF is sufficient to engender MPN, and the expression of Jak2^VF in LT-HSC is required for the full manifestation of disease in vivo.¹⁸  This identifies Jak2^VF LT-HSC as the exclusive reservoir of disease-initiating stem cells in vivo. IFNα has been shown to exhaust long-term WT HSC function in murine models^27,28  and the observation that molecular responses in MPN patients can be sustained even after discontinuation of IFNα therapy^23,25,39  suggests that IFNα may act by depleting Jak2^VF disease-propagating stem cells. To determine the effects of IFNα treatment on Jak2^VF LT-HSC, we determined the relative frequency of LT-HSC in chimeric mice treated with IFNα or vehicle by using flow cytometry (CD45.1 WT or CD45.2 Jak2^VF lineage^lowKit^highSca1⁺CD150⁺CD48⁻ cells; Figure 3A). Jak2^VF LT-HSCs were significantly depleted after 4 weeks of IFNα treatment (Figure 3B). Next, we determined the effects of IFNα treatment on myeloid-specified progenitor cell compartments using flow cytometry (Figure 3C). Surprisingly, IFNα treatment impaired megakaryocyte-erythroid progenitor (MEP) cell numbers in WT cells and increased granulocyte-macrophage progenitor (GMP) differentiation. In contrast, IFNα selectively increased MEP differentiation in Jak2^VF mice at the expense of GMP numbers (Figure 3D-E). The overall levels of apoptosis in LT-HSC populations were low and there was no significant induction of apoptosis in Jak2^VF compared with WT LT-HSC populations (supplemental Figure 5). These findings suggest that IFNα is not directly toxic to Jak2^VF stem cells but enforces a preexisting Jak2^VF-driven erythroid-lineage differentiation program within Jak2^VF HSCs.⁹ 

The maintenance of a quiescent LT-HSC population is essential for long-term self-renewal and this correlates with HSC function in vivo.^40,41  To test the hypothesis that IFNα depletes Jak2^VF LT-HSCs by activating the stem cell cycle and enforcing differentiation, we determined the cell cycle status of LT-HSC–enriched populations (lineage^lowKit^highSca1⁺CD150⁺). We used Hoechst 33342 and Ki-67 staining to determine the relative contribution of quiescent G0 cells (Ki-67^-H33342²ⁿ), activated G1 cells (Ki-67⁺H33342²ⁿ) and actively cycling cells (Ki-67⁺H33342⁴ⁿ) (Figure 4A). In vehicle-treated mice, there was a consistent reduction in G0 Jak2^VF LT-HSCs compared with WT cells. Consistent with previous reports, IFNα treatment caused activation of cell cycle in WT and Jak2^VF LT-HSCs.^27,28  Importantly, Jak2^VF LT-HSCs were more sensitive to IFNα treatment compared with WT LT-HSCs, as evidenced by significantly reduced quiescent, but more cycling LT-HSCs in Jak2^VF compared with WT cells after IFNα treatment (Figure 4B-C). These findings were confirmed with Hoechst 33342/ Pyronin Y staining on sorted LT-HSC–enriched populations (supplemental Figure 6). We found similar results in primary (nontransplanted) Jak2^VF mice (Figure 4F-G).

To determine the underlying molecular pathways that were differentially activated by IFNα treatment in Jak2^VF HSCs compared with WT HSCs, we performed gene expression profiling on purified HSC populations (lineage^lowKit^highSca1⁺CD150⁺). GSEA revealed significant enrichment of genes regulating cell cycle in IFNα-treated Jak2^VF HSCs compared with IFNα-treated WT HSCs (Figure 4D). In contrast, there was a relative enrichment of genes regulating quiescence in WT HSCs compared to Jak2^VF HSCs after IFNα treatment (Figure 4E). An enrichment of cell cycle genes in Jak2^VF LT-HSCs compared to WT LT-HSCs in vehicle-treated chimeric animals was also observed (supplemental Table 1). Finally, IFNα²⁷ and Stat-1 targets were nondifferentially enriched in IFNα-treated Jak2^VF and IFNα-treated WT HSCs compared with vehicle controls (data not shown). Together, these findings demonstrate that Jak2^VF LT-HSCs are preferentially sensitive to cell cycle activation by IFNα.

A defining characteristic of disease-initiating stem cells is the ability to transplant or recapitulate the disease in vivo.^9,42  To evaluate the effects of IFNα treatment on Jak2^VF MPN-initiating cells, we performed serial transplantation experiments. A total of 5 × 10⁵ bone marrow cells from vehicle- or IFNα-treated Jak2^VF chimeric mice were transplanted into lethally irradiated congenic recipient mice together with congenic helper bone marrow (1 × 10⁶ cells) and recipient animals were monitored for MPN development. Recipients of bone marrow from vehicle-treated Jak2^VF chimeric mice developed elevated HCT that was present by 4 weeks after transplantation; however, recipients of bone marrow from IFNα-treated Jak2^VF chimeric mice did not (Figure 5A). In comparison with vehicle-treated Jak2^VF chimeric donors, IFNα-treated Jak2^VF chimeric donors demonstrated significantly reduced Jak2^VF chimerism in secondary recipients, indicating that IFNα selectively targets Jak2^VF MPN-initiating cells (Figure 5B). Finally, there was significantly reduced total engraftment in recipient mice that received IFNα-treated bone marrow, with a selective reduction in the engraftment of IFNα-treated Jak2^VF bone marrow over IFNα-treated WT bone marrow (Figure 5C). This demonstrates preferential depletion of functional repopulating Jak2^VF LT-HSCs compared with WT LT-HSCs after IFNα. The effects on Jak2^VF chimerism and total engraftment were sustained up to 16 weeks after transplantation (Figure 5D-E), although persistent, long-term, low-level Jak2^VF engraftment was present in all recipient mice. The reduction in secondary transplantation of disease was confirmed at an increased cell dose (2 × 10⁶ donor bone marrow cells). For vehicle-treated chimeric donors, 4 of 4 recipients of the 2 × 10⁶ cell dose demonstrated elevated HCT 4 and 16 weeks after transplantation. In contrast, for IFNα-treated chimeric donors, 1 of 4 recipients of the 2 × 10⁶ cell dose demonstrated elevated HCT 4 weeks after transplantation and 3 of 4 recipients had elevated HCT 16 weeks after transplantation. These findings demonstrate that IFNα treatment depletes functional Jak2^VF disease-initiating stem cells and impairs the transmission of Jak2^VF MPN in vivo and suggest that in the absence of sustained treatment with IFNα, Jak2^VF MPN-propagating stem cells can persist and maintain disease.

To determine whether the effects on Jak2^VF LT-HSCs were cell autonomous and specific to type 1 interferon signaling, we established chimeric recipient mice from Jak2^VF or Jak2^VF IFNα receptor null (Ifnar1^−/−) donors.³¹  These chimeric recipients were ∼70% Jak2^VF CD45.2 and 30% WT CD45.1. Chimeric recipient mice were treated with 10 000 U/d IFNα for 4 weeks. Consistent with the previous data, IFNα treatment caused a significant reduction in the WCC and RCC in the peripheral blood of chimeric mice, whereas no changes were seen in Jak2^VFIfnar1^−/− chimeras (Figure 6A-B). Furthermore, extramedullary hematopoiesis, as evidenced by spleen size and spleen WCC, was reduced in Jak2^VF chimeras after IFNα treatment and these changes were not seen in Jak2^VFIfnar1^−/− chimeras (Figure 6C-D). Treatment with IFNα caused increased MEP and suppressed GMP differentiation in Jak2^VF cells; however, this was not observed in Jak2^VFIfnar1^−/− chimeras (Figure 6E-F). There was a reduction in total LT-HSC numbers in Jak2^VF chimeras after treatment with IFNα; however, there was no change in total LT-HSC numbers in Jak2^VFIfnar1^−/− (Figure 6G). Finally, treatment with IFNα led to a reduction in the G0 (quiescent) HSC population and an increase in the actively cycling HSC population in Jak2^VF cells. However, there was no change to cell cycle in Jak2^VFIfnar1^−/− chimeras after treatment with IFNα (Figure 6H-J). Together, these data demonstrate that the effects of IFNα on Jak2^VF HSCs are specific to type 1 interferon signaling and are cell autonomous. Moreover, these data confirm the effects of IFNα treatment in a system that is biased toward a higher Jak2^VF chimeric burden.

IFNα treatment achieves molecular remissions in MPN patients,^5,6,22  and in murine models, IFNα has been shown to exhaust long-term WT HSC function.^27,28  In aggregate, these results suggest that IFNα may eliminate the JAK2V617F mutant clone in MPN patients by preferentially targeting JAK2V617F MPN-propagating stem cells and depleting them over time. However, because the JAK2V617F allelic burden is typically quantified in peripheral blood granulocytes and because the majority of clonal expansion in MPN occurs in progenitor cells, it is plausible that patients that have no detectable JAK2V617F in granulocytes could still have a reservoir of JAK2V617F mutant HSC that persists. Indeed, JAK2V617F mutant colonies can be detected at a low level in some MPN patients in CMR.⁴³  Intriguingly, several groups have reported long-term molecular responses after discontinuation of IFNα therapy, suggesting that JAK2V617F mutant HSCs are being eradicated by IFNα,^23,25,39  although molecular relapse after cessation of IFNα has also been observed.⁴⁴  In this study, we use a conditional knockin murine model of Jak2V617F-induced MPN to investigate the effects of IFNα on Jak2^VF mutant and WT LT-HSC in vivo and to evaluate the means by which IFNα eradicates Jak2^VF mutant cells in MPN.

Given the clinical IFNα data as outlined and the experimental evidence demonstrating the effects of IFNα on HSC homeostasis,^27,28,45,46  we focused on the HSC compartment. We have not observed differences in LT-HSC numbers at steady state in primary Jak2^VF compared with WT mice.⁹  We find that Jak2^VF MPN stem cells are preferentially sensitive to IFNα therapy with depletion of Jak2^VF LT-HSC over time. We show that IFNα treatment promotes the exit from quiescence of Jak2^VF LT-HSC and Jak2^VF-induced terminal erythroid differentiation. Using flow cytometry and gene expression profiling performed on populations enriched for LT-HSC activity (CD150+LKS+), we show that Jak2^VF LT-HSC demonstrate enhanced cell cycle activation compared with WT LT-HSC, in response to treatment with IFNα. Using Ifnar1 knockout mice, we demonstrate that the effects of IFNα on cell cycle are cell autonomous and specifically mediated through type 1 interferon signaling. In vehicle-treated mice, we find a more activated cell cycle in Jak2^VF LT-HSCs compared with WT LT-HSCs, and consistent with this, GSEA indicated an enrichment of cell cycle genes in Jak2^VF LT-HSCs compared with WT LT-HSCs in vehicle-treated chimeric animals. We previously reported no difference in cell cycle between Jak2^VF and WT cells, but this earlier analysis was performed on less purified total LKS+ cells from nontransplanted primary mice.⁹  In aggregate, these results indicate that Jak2^VF LT-HSCs demonstrate enhanced cell cycle activity at baseline and preferential sensitivity to cell cycle activation in response to IFNα treatment.

The apparently paradoxical finding of enhanced LT-HSC cell cycling and enhanced self-renewal has been demonstrated in other MPN models. We previously reported a subtle self-renewal advantage for Jak2^VF LT-HSC,¹⁸  and in the present study, we find enhanced cell cycle activity in Jak2^VF LT-HSC compared with WT LT-HSC in vehicle-treated chimeric mice. Interestingly, Reynaud et al⁴⁷  recently reported, using a murine transgenic BCR-ABL model, that CML LT-HSC also demonstrate enhanced cycling that occurs together with a gain in LT-HSC self-renewal and similar findings have been reported in a transgenic JAK2V617F MPN model.⁴⁸  The exact mechanisms underlying these findings remain unclear; however, altered regulation of key self-renewal pathways (eg, Notch) may play a role.^47,49  Our data evaluating the effects of IFNα indicate that Jak2^VF LT-HSCs are preferentially sensitive to IFNα treatment by showing a more marked increase in cycling, which results in a preferential depletion in response to IFNα. This sensitivity may occur as a result of baseline cell cycle activation in Jak2^VF MPN stem cells. These data also suggest that increased cell cycle may not be the sole mechanism by which IFNα depletes Jak2^VF LT-HSCs, because vehicle-treated Jak2^VF LT-HSCs also have increased cell cycling but do not undergo depletion during long-term BMT assays. Additional studies aimed at identifying key cell cycle and self-renewal modulators differentially regulated at baseline and in response to IFNα treatment in Jak2^VF and WT LT-HSCs remain an ongoing focus of our work. Finally, we also note that the Jak2^VF LT-HSC competitive repopulation advantage we previously found became manifest with prolonged follow-up (>1 year)¹⁸  and our findings in the present study, where Jak2^VF LT-HSCs have no significant repopulating advantage at 8 weeks (Figure 3B), are consistent with this. In the present study, we find no significant difference in the engraftment of Jak2^VF vehicle-treated cells compared with WT vehicle-treated cells at 16 weeks in a secondary BMT assay (Figure 5E). It is not clear why a self-renewal advantage is not seen in this secondary BMT; however, it is possible that a change in competitive repopulation ability is too subtle to be observed after only 16 weeks posttransplantation. Alternatively, we cannot exclude that the late clonal advantage observed could occur as a result of the acquisition of secondary mutations.

Notably, depletion of Jak2^VF LT-HSC was observed only in IFNα-treated mice, and importantly, depletion of the MPN-propagating stem cell population was associated with reduced MPN transplantation efficiency in serially transplanted recipient mice. Serial transplantation is the most stringent functional assay to assess effects on the disease-initiating stem cell population, and these experiments indicate that sustained IFNα treatment targets MPN-propagating HSC in vivo. The presence of persistent low-level Jak2^VF chimerism after IFNα treatment in serially transplanted mice suggests that IFNα may not be sufficient to completely eradicate Jak2^VF LT-HSCs or that a significantly prolonged IFNα treatment course may be required for definitive disease eradication. We do not find any evidence to indicate that IFNα treatment in vivo has proapoptotic effects on LT-HSCs and the absence of a reduction in Ifnar1^−/−Jak2^VF HSCs in response to 4 weeks of continuous IFNα treatment argues strongly against an immune-mediated depletion of Jak2^VF cells. These results are also consistent with human clinical trials in MPN, in which at least 1 year of IFNα therapy was required for CMR to be achieved.^5,6  In aggregate, our results indicate that IFNα has differential effects on Jak2^VF and normal WT LT-HSCs in vivo and suggest that the durable molecular remissions induced by IFNα in patients with MPN occur as a result of MPN stem cell exhaustion through activated cell cycling and enforced differentiation.

The Jak2^VF model we use in this study closely recapitulates the features of human PV and we have exploited this to evaluate the effects of IFNα on Jak2^VF erythropoiesis in vivo. We find that IFNα treatment selectively expands the Jak2^VF MEP compartment in the bone marrow by driving Jak2^VF LT-HSCs out of quiescence and down a predetermined erythroid-lineage differentiation program. IFNα treatment induces apoptosis in Jak2^VF erythroid precursor cells, with immature Jak2^VF erythroid precursors (CD71^highTer119^mid-high) demonstrating enhanced sensitivity to IFNα inhibition compared with mature Jak2^VF erythroid precursors (CD71^mid-lowTer119^high cells). These effects account for the majority of the reduction in Jak2^VF-mediated splenomegaly that we find in response to IFNα treatment and are consistent with the known sensitivity of JAK2V617F erythroid colonies to IFNα treatment in vitro.⁵⁰  We find that the reduction in HCT is modest after 4 weeks of IFNα treatment but that the HCT is normalized after 8 weeks of IFNα treatment, a finding likely related to the prolonged life span of a mature RBC. In aggregate, these results indicate that IFNα exerts distinct effects on different Jak2^VF hematopoietic cellular compartments, activating Jak2^VF LT-HSCs to induce differentiation while inducing apoptosis in Jak2^VF erythroid precursor cells to normalize RBC counts.

Although IFNα clearly has therapeutic efficacy in MPN, it is far from an ideal treatment and remains an experimental MPN therapy at present.²⁵  One of the major challenges in the clinical use of IFNα is poor patient tolerability due to the multitude of side effects IFNα causes. Although these have improved with the use of lower dosing and pegylated forms of the drug, many patients remain intolerant of IFNα. Future studies aimed at identifying unique IFNα effector pathways in Jak2^VF MPN stem cells have the potential to facilitate the development of more specific, less toxic IFNα derivatives for the treatment of MPN. This work also provides support for the rational combination of IFNα with new agents with promising but incomplete activity in MPN, including the JAK2 inhibitors. Combinatorial IFNα/JAK2 inhibitor treatment may be therapeutically synergistic as a result of IFNα-mediated induction of cell cycle in quiescent Jak2^VF LT-HSCs coupled with the antiproliferative effects of JAK2 kinase inhibition in progenitor cells. Finally, it is important to note that only a minority of MPN patients achieves a CMR in response to IFNα treatment. Part of the heterogeneity in response may be related to the presence of other genetic abnormalities, in addition to JAK2V617F, in some patients. Emerging evidence suggests that TET2 and DNMT3A mutations may contribute to IFNα resistance in MPN.^43,51,52  Given the key roles of TET2^53,54  and DNMT3A⁵⁵  in HSC homeostasis, it will be important to investigate the mechanisms of this resistance as a means to further delineate the specific effects of IFNα on MPN-propagating stem cells.

In conclusion, we report the therapeutic efficacy of IFNα on Jak2^VF disease-initiating LT-HSCs in vivo, a finding consistent with that previously reported in abstract form by Marty et al.⁵⁶  We demonstrate that IFNα preferentially depletes Jak2^VF MPN-propagating stem cells by activating cell cycle and promoting Jak2^VF-driven erythroid-lineage differentiation. Through this work, we offer insights into the mechanism by which IFNα achieves CMR in MPN patients and provide a biological rationale in support of combinatorial IFNα/JAK2 inhibitor clinical trials in MPN patients.

The authors thank members of the Ebert, Armstrong, Hill, and Williams laboratories for technical assistance, helpful discussion, and insights. In particular, the authors acknowledge the expert advice from Fatima Al-Shahrour, Amit Sinha, and Glen Boyle with the gene expression data analyses. The authors thank Grace Chojnowski and Paula Hall for flow cytometry and sorting.

This work was supported by the MPN Foundation (B.L.E.) and the National Institutes of Health, National Cancer Institute (P01 CA108631 and P01 CA066996 to B.L.E.) and the National Institutes of Health, National Heart, Lung, and Blood Institute (K08 HL109734 to A.M.). G.R.H. is a NHMRC Australia Fellow and Queensland Health Senior Clinical Research Fellow. F.H.H. was supported by a Mildred-Scheel grant of the German Cancer Aid (DKH D/08/00661, Deutsche Krebshilfe). D.H. was supported by the German Cancer Foundation, Mildred-Scheel Fellowship. D.K. was supported by National Institute of Diabetes and Digestive and Kidney Diseases grant K01DK092300. A.M. has received support from the Jeanne D. Housman Fund for Research on Myeloproliferative Disorders and is an ASH Scholar recipient. S.W.L. has received funding from the Leukaemia Foundation of Australia, National Health and Medical Research Council project grant 1026594, J.J.Richards/ In Vitro Technologies Rhys Pengelly Fellowship in Leukaemia Research and private philanthropic donations.

Contribution: A.M. and S.W.L. provided the concept, designed experiments, performed experiments, analyzed data, and wrote the manuscript; C.B., L.P., F.H.H., A.P., T.V., R.A., D.H., L.J.B., C.P.K., and D.K. performed experiments and analyzed data; S.A.A., D.A.W., G.R.H., and B.L.E. designed experiments and provided supervision for the project. All authors provided critical review for the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Steven Lane, Queensland Institute of Medical Research, 300 Herston Rd, Herston, Brisbane, Australia, 4006; e-mail: steven.lane@qimr.edu.au; and Ann Mullally, 1 Blackfan Circle, Karp 5.007, Boston, MA 02115; e-mail: amullally@partners.org.